Control tools for flow
chemistry processing and their 
application to the synthesis of 
bromodomain inhibitors
A thesis presented by
Richard J. Ingham
for the degree of
Doctor of Philosophy
June 2014
Jesus College
Jesus Lane
Cambridge
CB5 8BL
Ley Group
Department of Chemistry
Lensfield Road
Cambridge
CB2 1EW

Declaration
This dissertation is the result of my own work and includes nothing that is the outcome of
work done in collaboration except where specifically included in the text and bibliography.
This thesis is not substantially the same as any that I have submitted for a degree, diploma
or other qualification at any other University. I further state that no part of my thesis has
already been or is being concurrently submitted for any such degree, diploma or other
qualification.
It does not exceed 60 000 words, including tables, footnotes, bibliography and appendices.
Richard Ingham, June 2014
i
ii
Acknowledgments
I am hugely grateful to Professor Steven Ley for welcoming me into his research group
and making it such a wonderful place to work. I cannot think of anywhere I would have
preferred to spend four years! Steve’s infectious enthusiasm for new ideas, support and
encouragement make the exciting research that takes place in the Ley group possible.
Professor Ian Baxendale was also instrumental during the first half of my time in the group,
and in founding the fantastic team ITC.
Thanks to Dr Elena Riva for her collaboration in the solid-supported synthesis project, and
to Dr Nikzad Nikbin for introducing Elena and I to the “way of the monolith.” Thanks to Dr
Claudio Battilocchio for his collaboration with the pyrazine and adamantanone projects (for
making the chemistry work, for his frequent displays of musical prowess, and for teaching
the ITC the more colourful aspects of the Italian language). Thanks to Daniel Fitzpatrick
for construction of the rotating filter prototype, and to Dr Éric S´liwin´ski for creating some
fantastic 3D models. Finally, thanks (again) to Nik, to Dr Lucie Guetzoyan, Charlotte
Sutherell, Michael Wolling, Julien Rossignol, the members of the SGC and everyone else
who took part in the bromodomain project.
Thank you to everyone who is in or has passed through the Ley group during my stay, for
creating such a special atmosphere, planning the parties, and creating all of the memories.
Particular thanks to Richard, Nik, Claudio, Eric, Duncan, James, Charlotte, Gonçalo, Daniel,
Lucy and Lucie for taking the time to read this thesis in its varying states of completion,
and for their helpful comments and suggestions.
The department would not function without the hard work of the technical staff: thanks
to Dr Richard Turner, Melvyn Oriss, Keith Parmenter and Matt Pond for keeping the labs
working and stocked with reagents and equipment. And thanks to Jude, Jacqui, Jo and Rose
for keeping the Ley group stocked with happy people!
Finally, a big thank you to all of my friends and family for always being around to provide
help and support, for making life both in Cambridge and out so wonderful, and without
whom I wouldn’t have made it this far.
iii
Summary
Flow chemistry and continuous processing techniques are now frequently used in synthetic
laboratories, taking advantage of the ability to contain reactive or hazardous intermediates
and to perform moderate scale-up processes for important compounds. However, only
a limited number of methods and tools for connecting flow synthesis steps into a single
protocol have been described, and as a result manual interventions are frequently required
between consecutive stages.
There are two main challenges to overcome. Work-up operations such as solvent extractions
and filtrations are invariably needed to ensure high purity of the intermediates. Solutions
for achieving this are well established within industrial facilities for continuous production,
but adapting such machinery for laboratory use is rarely straightforward. Secondly, the
combination of multiple steps tends to result in a more elaborate reactor configuration.
The control procedures required to achieve optimum performance may then be beyond
the capabilities of a single researcher. Computer control and remote monitoring can help
to make such experiments more practical; but commercially-available systems are often
highly specialised, and purpose-built at high cost for a particular system, and so are not
suitable for laboratory scientists to use routinely.
This work describes the development of software tools to enable rapid prototyping of control
systems that can integrate multiple instruments and devices (in Chapter 2). These are applied
to three multi-step synthesis projects, which also make use of enabling technologies such
as heterogeneous reagents and in-line work-up techniques so that material can be passed
directly from one stage to the next:
In Chapter 1, a series of analogues of a precursor to imatinib, a tyrosine kinase inhibitor
used for the treatment of chronic myeloid leukaemia, are prepared. A “catch-react-release”
technique for solid-phase synthesis is used, with computer-controlled operation of the
reactors.
In Chapter 3, a two-step procedure for the synthesis of piperazine-2-carboxamide, a valuable
3D building block, is developed. A computer control system enabled extended running and
the integration of several machines to perform optimisation experiments.
In Chapter 4, improvements to the continuous synthesis of 2-aminoadamantane-2-carboxylic
acid are discussed. This includes an integrated sequence of three reactions and three work-
up operations.
iv
The final chapter describes a project to evaluate the application of control techniques to a
medicinal chemistry project. New ligands for BRD9 and CECR2, proteins involved in the
recognition of acetylated histone proteins, are produced. A number of triazolopyridazine
compounds were synthesised and tested using a number of assay techniques, including a
frontal-affinity chromatography system under development within our group. Pleasingly,
the qualitative FAC data showed a good correlation with biological assessments made using
established assay techniques. Further work using the FAC method is ongoing.
Contents
Glossary and Abbreviations ix
Key to flow reactor symbols xiii
Introduction 1
Automation in Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Flow Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Automation in Flow Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1 Solid-supported flow synthesis of 2-aminopyrimidines 7
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.1 Synthesis of 2-aminopyrimidines . . . . . . . . . . . . . . . . . 8
1.1.2 Solid-supported reagents . . . . . . . . . . . . . . . . . . . . . . 9
1.1.3 Monolithic reagents . . . . . . . . . . . . . . . . . . . . . . . . 10
1.1.4 Proposed synthetic route to 2-aminopyrimidines . . . . . . . . . 13
1.2 Development of Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.1 Generation of the monolithic reagent . . . . . . . . . . . . . . . 14
1.2.2 Monolith characterisation . . . . . . . . . . . . . . . . . . . . . 15
1.2.3 Heterocycle Formation . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.4 Oxidation of monolith M2 . . . . . . . . . . . . . . . . . . . . . 18
1.2.5 Release of 2-aminopyrimidine product . . . . . . . . . . . . . . . 19
1.3 Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.3.1 Advantages of automation . . . . . . . . . . . . . . . . . . . . . 21
1.3.2 Solvent switching and stop-flow . . . . . . . . . . . . . . . . . . 24
1.3.3 Multi-step processing . . . . . . . . . . . . . . . . . . . . . . . . 25
1.4 Application to the preparation to derivatives of imatinib amine . . . . . . 27
1.4.1 Reported syntheses of imatinib . . . . . . . . . . . . . . . . . . . 27
1.4.2 Monolith-assisted route to imatinib . . . . . . . . . . . . . . . . 30
1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
v
vi Contents
2 New control and monitoring software for flow chemistry systems 33
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.1.1 Existing control software . . . . . . . . . . . . . . . . . . . . . . 35
2.1.2 Flow chemistry equipment . . . . . . . . . . . . . . . . . . . . . 38
2.1.3 Interaction with equipment . . . . . . . . . . . . . . . . . . . . . 38
2.1.4 Control computers . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.2 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2.1 Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.2.2 Data handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.2.3 Command sequences . . . . . . . . . . . . . . . . . . . . . . . . 49
2.2.4 Data manipulation . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.2.5 Pausing experiments . . . . . . . . . . . . . . . . . . . . . . . . 53
2.2.6 Error handling . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.2.7 Attached control systems . . . . . . . . . . . . . . . . . . . . . . 53
2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3 Automated two-step reactions with heterogeneous metal catalysis 57
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.2 Synthesis of primary amides by nitrile hydration . . . . . . . . . . . . . . 58
3.2.1 Hydration under flow conditions . . . . . . . . . . . . . . . . . . 60
3.3 Reaction control and monitoring . . . . . . . . . . . . . . . . . . . . . . 61
3.4 Automation of multiple experiments . . . . . . . . . . . . . . . . . . . . 63
3.5 Integration of two steps . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.5.1 Intermediate reservoir . . . . . . . . . . . . . . . . . . . . . . . 67
3.5.2 Combining synthesis and DoE . . . . . . . . . . . . . . . . . . . 70
3.5.3 Extended run . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid 75
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.1.1 Challenges of telescoped reactions . . . . . . . . . . . . . . . . . 76
4.2 Previous synthetic routes . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.2.1 Reported batch–mode syntheses . . . . . . . . . . . . . . . . . . 79
4.2.2 First generation flow–mode synthesis . . . . . . . . . . . . . . . 80
4.3 Grignard reaction and aqueous extraction . . . . . . . . . . . . . . . . . 82
4.3.1 Mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.3.2 Contactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.3.3 Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.3.4 Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.3.5 Control of the Grignard reaction . . . . . . . . . . . . . . . . . . 90
4.4 Solvent exchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.5 Ritter reaction and quenching . . . . . . . . . . . . . . . . . . . . . . . . 94
4.5.1 Intermediate reservoir . . . . . . . . . . . . . . . . . . . . . . . 94
4.5.2 Ritter reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Contents vii
4.5.3 Base quenching . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.5.4 Cyclisation reaction . . . . . . . . . . . . . . . . . . . . . . . . 100
4.6 Ozonolysis and Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.7.1 Further work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5 Development of inhibitors for bromodomain-containing proteins 105
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.1.1 The bromodomain . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.1.2 Frontal Affinity Chromatography . . . . . . . . . . . . . . . . . 109
5.2 Chemical probes for BRD9 . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.2.1 Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.2.2 Proposed synthesis . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.2.3 Synthesis of core . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.2.4 Synthesis of boronic ester coupling partners . . . . . . . . . . . . 124
5.2.5 Cross-coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.2.6 Carbamate formation . . . . . . . . . . . . . . . . . . . . . . . . 133
5.2.7 Alternative route . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.2.8 Biological assessment . . . . . . . . . . . . . . . . . . . . . . . 136
5.2.9 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.3 Chemical probes for CECR2 . . . . . . . . . . . . . . . . . . . . . . . . 140
5.3.1 “Reversed” sulfonamide boronic ester . . . . . . . . . . . . . . . 142
5.3.2 Miyaura borylation . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.3.3 Suzuki cross-coupling . . . . . . . . . . . . . . . . . . . . . . . 147
5.3.4 Biological analysis . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Conclusion 155
A Synchronous instrument controller 157
B Control scripts for monolith processing 163
Experimental 169
General Experimental Information . . . . . . . . . . . . . . . . . . . . . . . . 170
Solid-supported synthesis of 2-aminopyrimidines . . . . . . . . . . . . . . . . 172
Piperazine-2-carboxamide synthesis . . . . . . . . . . . . . . . . . . . . . . . 190
2-Aminoadamantane-2-carboxylic acid synthesis . . . . . . . . . . . . . . . . 194
Chemical Probes for BRD9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Chemical Probes for CECR2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Frontal Affinity Chromatography . . . . . . . . . . . . . . . . . . . . . . . . . 228
References and Notes 228
viii
Glossary and Abbreviations
 Inner diameter
1D One-dimensional
2D Two-dimensional
Ac Acetate
ACHC Azobiscyclohexylcarbonitrile (radical initiator)
Ar Unspecified aryl substituent
BEMP 2-tert-Butylimino-2-diethylamino-1,3-dimethylperhydro-1,3,2-diazaphosphorine
BPR Back pressure regulator
br. Broad peak (NMR / IR spectroscopy)
BRD Bromodomain
Bu Butyl
CV Column volumes (column chromatography)
conv. Conversion
COSY 2D 1H correlation spectroscopy (NMR spectroscopy)
CPME Cyclopentyl Methyl Ether
δ Chemical shift (NMR spectroscopy)
d Days, or doublet peak (NMR spectroscopy)
DCM Dichloromethane
DEPT Distortionless Enhancement by Polarization Transfer (NMR spectroscopy)
DIN Deutsches Institut fur Normung E.V. (German Institute for Standardisation)
DIPEA Diisopropylethyl amine
DMAP 4-Dimethylaminopyridine
DME Dimethoxyethane
ix
x Glossary and Abbreviations
DMF N,N-Dimethylformamide
DMF-DMA N,N-Dimethylformamide dimethylacetal
DMSO Dimethyl sulfoxide
DMSO-d6 Deuterated dimethyl sulfoxide
DoE Design of Experiments
DPE-Phos Bis[(2-diphenylphosphino)phenyl]methane
DPPA Diphenoxyphosphoryl azide
dppf 1,1′-Bis(diphenylphosphino)ferrocene
DtBPF 1,1′-Bis(di-tert-butylphosphino)ferrocene
DVB Divinylbenzene
EI Electron-Impact Ionisation (mass spectroscopy)
EMM Ethoxymethylene(malononitrile)
equiv. Equivalents
ESI Electrospray Ionisation (mass spectroscopy)
Et Ethyl
ETFE Ethylene tetrafluoroethylene (polymer)
FAC Frontal Affinity Chromatography
FEP Fluorinated Ethylene Propylene (polymer)
FT-IR Fourier Transform Infrared (spectroscopy)
HMBC Heteronuclear Multiple Bond Connectivity (NMR spectroscopy)
HMQC Heteronuclear Multiple Quantum Coherence (NMR spectroscopy)
HPLC High-Performance Liquid Chromatography
HRMS High-Resolution Mass Spectroscopy
I2C Inter-Integrated Circuit (serial data communication protocol)
IC50 Median inhibition concentration (concentration of a substrate that reduces a
protein’s activity by 50 %)
ICT Information and Communications Technology
ITC Innovative Technology Centre, University of Cambridge
ID Inner Diameter
IR Infrared
isol. Isolated
Glossary and Abbreviations xi
J Coupling constant (NMR spectroscopy)
Kd Dissociation constant
LC-MS Liquid Chromatography coupled Mass Spectroscopy
m Multiplet (NMR spectroscopy),
or medium absorbance (IR spectroscopy)
m meta (position around a phenyl ring)
M Moles per litre, or unspecified metal atom
M+ Molecular ion
m-CPBA meta-Chloroperoxybenzoic acid
Me Methyl
mol% Molar percentage
m.p. Melting point
Ms Methanesulfonyl
MW Microwave
m/z Mass to charge ratio (mass spectroscopy)
NMR Nuclear Magnetic Resonance (spectroscopy)
Nu Unspecified nucleophile
o ortho (position around a phenyl ring)
OD Outer Diameter
p para (position around a phenyl ring)
PC Personal Computer
PEEK Polyether ether ketone (polymer)
PEPPSI Pyridine-enhanced precatalyst preparation, stabilization and initiation
PFA Perfluoroether (class of polymer)
Ph Phenyl
ppm Parts per million
PTFE Polytetrafluoroethylene (polymer)
QP-BZA QuadraPure benzylamine
QP-DMA QuadraPure dimethylamine
QP-TU QuadraPure thiourea
quant. Quantitative
xii Glossary and Abbreviations
q Quartet peak (NMR spectroscopy)
R Unspecified alkyl substituent
RS-232 Recommended Standard 232 (serial data communication standard)
s Singlet peak (NMR spectroscopy),
or strong absorbance (IR spectroscopy)
SGC Structural Genomics Consortium, Oxford University
SMS Short Message Service
SS Stainless steel
t Triplet peak (NMR spectroscopy)
TFA Trifluoroacetic acid
tR Retention time in minutes (LC-MS)
TLC Thin Layer Chromatography
THF Tetrahydrofuran
USB Universal Serial Bus (serial data communication standard)
UV Ultraviolet (spectroscopy)
v/v volume / volume
w Weak absorbance (IR spectroscopy)
w/w weight / weight
Key to flow reactor symbols
100 psi
RR
Pump
Mixer
Spring-type
back-pressure
regulator
In-line detector
Glass column
Immobilised
reagent
Valve
Sample loop
Heated
ow coil
Monolithic
immobilised
reagent
xiii
xiv
Introduction
1
2 Introduction
Automation in Synthesis
Automation has had a significant impact on many different industries since the mid 20th
century. Robots of all sizes and specifications are installed in factories to perform tasks that
are routine, or too difficult or strenuous for a person to do unassisted, and allowing 24 hour
working with a smaller manpower requirement.
In a research laboratory, automation refers to the use of computers or micro-controllers
to interact with equipment to perform tasks that the researchers would otherwise have to
do manually. There are a number of tasks that already routinely benefit from automation.
Repetitive tasks such as injecting HPLC samples or making serial dilutions of samples for
biological assays can be performed by laboratory robots, though these are often highly
engineered for a particular application and very expensive. In-situ analytical equipment can
collect more data about experiments than may be possible with manual observations alone,
and digital data collection is often more accurate than manual observations, and can be run
indefinitely, without breaks. In general, the use of automation can produce better data, and
give researchers more time to do less mundane tasks.
In the context of synthetic chemistry, automation is typically associated with combinatorial
chemistry.1,2 This involves repetition of simple synthetic steps to produce huge numbers of
building blocks or compounds for biological analysis. The problem was that by creating
large numbers of compounds using only a small number of reactions, the size of the
chemical space that was explored turned out to be relatively limited, and highly dependent
on the availability of different starting materials. Combinatorial chemistry went through a
phase of popularity in the 1980s and 1990s before falling out of favour having not produced
significant results.3,4 However, modern concepts such as diversity-oriented synthesis5 may
offer new targets for automated synthesis platforms.
There are a number of potential benefits to automating the control of synthetic procedures.
The key is that the focus should be on improving the quality of individual reactions rather
than the quantity of reactions that are performed. Computers and robots can enable scientists
to accomplish more complex tasks more easily than if they were done manually, and by
delegating routine work to automated systems skilled chemists can be freed to do more
important tasks. There may also be an improvement in the reporting and reproducibility
of experiments: by consistent recording of the reaction parameters (such as the amount of
reagents used, the timing of their addition, temperatures and heating times), and by accurate
measurements and recording of data collected during an experiment, reactions in the lab of
Flow Chemistry 3
the future could be made more reproducible and the results even more useful.6 As a result,
the quality of the data reported in the literature would only improve.7,8
The use of automated platforms could provide other benefits as well. Disconnecting the
scientist from the manual aspects of a reaction means that hazardous materials could be used
more safely. Chemists could run reactions from a different room, or even from different
institutions or countries, providing global access to key facilities.
Flow Chemistry
Flow chemistry commonly refers to the continuous processing of chemical reagents within
small tubes or channels. In conventional batch-mode chemistry, many addition, mixing
and reacting operations for one reaction may be performed in the same vessel. By contrast,
in a flow chemistry system the reagents are delivered continuously by pumping devices;
multiple streams are mixed and then allowed to react in tube or column reactors which may
be heated, cooled or irradiated to provide the desired reaction conditions. Reaction times
are determined by the reactor volume and the flow rate, and reaction scale by the duration
of operation.
The history of flow chemistry is rooted in the continuous processing of petrochemicals.
The volume of material handled in processes such as cracking and refining are so large that
batch methods would be impractical and economic due to the number of manual operations
that would be needed. Early examples of continuous manufacturing of bulk chemicals
includes the processes for producing ammonia9 and sulfuric acid,10 where the economies
of scale again made continuous processing an ideal tool. The use of continuous methods
for fine chemicals manufacture originated as a means for process intensification,11 intended
to reduce the footprint and manpower requirements of a manufacturing plant, and also to
improve safety, energy efficiency and environmental-friendliness.12 However, the speciality
chemicals and pharmaceuticals sectors have been slow to adopt continuous manufacturing,
primarily as a result of having much smaller production volumes and a large amount of
sunken capital in batch reactor equipment. Work within academic laboratories may lead to
the development of new methods and processes that will encourage further uptake of flow
technology by enabling new or hazardous reactivity to be exploited under a continuous
regime.
4 Introduction
Flow reactors are often considered to have a number of advantages over batch-mode
reaction vessels. With appropriate static or dynamic mixers, the concentration profile can
be controlled predictably; if the mixing is very efficient then remarkable improvements in
reaction times can be observed for some transformations.13,14 In-line heterogeneous reagent
or catalyst cartridges (containing either inorganic or polymeric reagents) can perform
chemical transformations with little or no purification requirement.15 Addition of further
reagent streams can enable direct quenching or use of reactive intermediates, leading to
improved synthetic efficiency and consistency. The large surface area to volume ratio
of a flow reactor can allow precise temperature control, extreme conditions and rapid
temperature changes,14 as well as safe containment of exothemic events.16 Factors such
as mixing and heat transfer are critical parameters when scaling a reaction; in a flow
reactor these are independent of the amount of material that is to be processed. Combined
with a small reaction volume (1 – 20 mL) compared to standard batch apparatus (10 –
500 mL), continuous processing can deliver significant improvements in the safety of
hazardous transformations.16–19 For example, hydrogenation reactions are increasingly
being performed in much smaller, safer continuous flow reactors instead of in batch
hydrogenators that had to be housed in concrete bunkers to shield against catastrophic
explosions.20,21
There are of course some disadvantages that must be kept in mind when using flow
reactors.22 They are by their nature much more complex than a flask, and many of the
devices currently available are very susceptible to blockage or disablement by solids that
form during the reaction. Flow reactions operate most consistently at steady state, and
this is only attained after a period of time that is dependent on the reactor’s volume, so
a b c
Figure 1 Some of the commercially available flow chemistry systems. (a) Vapourtec
R2+/R4. (b) Uniqsis FlowSyn. (c) Syrris Asia.
Automation in Flow Chemistry 5
are often inappropriate for running large numbers of small-scale reactions. In this regard,
they are much less flexible than batch chemistry apparatus: frequently the reactor must
be reconfigured to perform different reactions, meaning that a significant time investment
is required for each new procedure. This is another reason why the uptake of continuous
manufacturing has been slow in the speciality chemicals and pharmaceuticals sectors.
Even so, there are a large number of applications for which continuous processing tech-
niques are highly effective. For this reason, flow synthesis has become increasingly popular
over the past decade.23
Whilst flow reactions can be performed using any appropriate pumping and heating
equipment, many researchers choose to employ commercial reactor platforms such as
the Vapourtec R2+/R4, Uniqsis FlowSyn, or Syrris Asia systems which are used within
our own laboratory (Figure 1). These systems contain pumps, switching valves and heaters
within a conveniently integrated unit, and provide software tools to run computer-controlled
experiments.
Automation in Flow Chemistry
Using these systems, our group and others have reported numerous applications of auto-
mated flow chemistry technology to combinatorial chemistry activities.23,24 By utilising
tools such as polymer-supported reagents and scavengers,25 the desired compounds can be
collected in very high purity with little or no purification.
Perhaps as a direct result of these early projects, the control software packages provided
with commercial flow chemistry platforms tend to be designed for exactly this kind of
a b
Figure 2 (a) Fluidic configuration using Vapourtec Flow Commander. (b) Choosing
experiments using Vapourtec Flow Commander.
6 Introduction
procedure. A fluidic configuration can be programmed in, and then long lists of identical
reactions using different starting materials can be run with the aid of a liquid handler
(Figure 2). In this regard, commercial flow chemistry equipment offers good tools for
starting automation of synthesis within the laboratory. The in-built computer connections
mean that the apparatus is ready to automate without significant additional engineering
effort.
If control and monitoring tools are applied to laboratory platforms then the process of
scaling reactions may become even easier. If a synthesis system can be left under computer
control then it can be run for a long time without active observation, enabling moderate
scale-up operations to be performed within the research laboratory itself.
Objectives
Until recently, programming skills have tended to be limited to specialists. However, with
the increasing pervasiveness of the internet connectivity, and concepts such as the “internet
of things”, there is much greater interest in the use of computers to influence the physical
world around us.26 In the United Kingdom, the school curriculum is being updated to
cultivate software creation as well as more traditional ICT skills.27 This is in part aided
by organisations such as the Raspberry Pi foundation,28 whose devices were designed to
provide affordable computer systems for everyone. This may result in a greater number of
scientists wanting to run experiments under computer control.
Flow chemistry can provide access to new reactive windows and easier scalability of
synthetic processes. However, most reported flow procedures consist of a single step,
and to our knowledge very few true multi-step flow syntheses have been described in the
literature.23 One reason for this is that the orchestration of the many devices required for an
integrated system is very difficult. Computer control and automation can provide the means
to do this, but is currently a relatively unexplored aspect of laboratory flow chemistry, since
commercial control system software does not tend to offer a level of flexibility appropriate
for discovery workflows. The objective of this project was to apply or develop in-line
processing and computer control tools to enable the construction of continuous multi-step
flow processes, and to demonstrate these within the research laboratory.
Chapter 1
Solid-supported flow synthesis of
2-aminopyrimidines
7
8 Solid-supported flow synthesis of 2-aminopyrimidines
1.1 Introduction*
1.1.1 Synthesis of 2-aminopyrimidines
One of the primary goals within the field of organic synthesis is to create cleaner and more
efficient methods to assemble functional molecules. 2-Aminopyrimidines are biologically
important building blocks, found in a wide range of compounds reported to possess anti-
emetic,30 anti-tumour,31–33 anti-malarial,34 pain relief35 and other biological properties
(Scheme 1.1a). Besides late-stage formation of the pyrimidine ring from a guanidine
moiety, these compounds are commonly prepared by cross-coupling methods31,32,36 or
by halogen substitution.30–32,35 However, there is an alternative method for accessing the
2-aminopyrimidine unit involving the displacement of an alkylthiol with a nucleophile such
as a secondary amine (Scheme 1.1b).34,37–39 This is potentially very powerful, because the
starting materials can be prepared easily and the reaction generally proceeds in good yield.
Unfortunately, this protocol has the disadvantage that alkylthiol by-products are released
during the reaction, which are malodorous and often toxic. We envisaged that this could be
overcome with a solid-supported methodology.
N
N NH2
NH2
Pr
Cl
N
N
H
N
N
HN
O
N
N
N
N
N
H
CF3
O
N
S
HN
O
N
a
1 2 3
N
N
H
MeS O
R3
N
R2
R1
N
N
H
O
R3NH
R2
R1
alcohol solvent
100 - 180 °C
O
RO O
R3
NH
NH2MeS
b
Scheme 1.1 (a) Reported bioactive compounds containing the 2-aminopyrimidine scaf-
fold: imatinib (1), a kinase inhibitor;33 2, an anti-malarial compound;34 3, a pain relief
compound.35 (b) Method for the preparation of 2-aminopyrimidines by nucleophilic dis-
placement of a thiol with an amine.37
* The work described in this chapter is reported in Org. Lett. (Reference 29).
1.1. Introduction 9
1.1.2 Solid-supported reagents
Thirty years after the discovery of Bakelite,40 this plastic found an application within chemi-
cal synthesis as an ion-exchange resin.41,42 Later, similar ion-exchange resins were used for
purification using “catch-and-release” methods.43 Following Merrifield’s seminal reports of
a solid-phase technique for peptide synthesis,44 a number of functionalised polymer resins
were reported and applied to chemical synthesis, including polymer supported phosphine,
halogenation reagents, redox reagents and protecting groups.45
The application of polymer-assisted solution phase synthesis provides a number of ad-
vantages over conventional solution-phase chemistry.15 If a reagent is covalently attached
to an insoluble polymer support, then it can be removed at the end of the reaction by a
Me
O
N2
P
O
OMe
OMe
N3
F
Ph
OH
SO3 NO
NMe2.CuI
N
H
S
NH2
NMe2 SO3H NH2
Ph
N
NN
F
100 psi
MeOH
0.2 mL/min
KOtBu
MeCN
0.2 mL/min
48 min
100 °C
70 °C
A21.CuI
A21
PS-Ts-TEMPO
A15 QP-BZA
Q
P-TU
2 equiv.
2 equiv.
1 equiv.
4
5
6
7
Scheme 1.2 A three-step reaction enabled by polymer-supported reagents.51 A sup-
ported TEMPO reagent effects an oxidation of alcohol 6 to an aldehyde. Upon heating, this
reacts with the Bestmann–Ohira reagent (4) to form an alkyne, which then undergoes an
alkyne–azide “click” reaction52 in the presence of copper (immobilised on an Amberlyst-21
resin). A sequence of scavengers (present in excess) then remove any unreacted materi-
als: the thiourea cartridge sequesters any copper which is leached from the A21 resin;
the benzyl amine cartridge removes any unreacted aldehyde; the sulfonic acid cartridge
removes the base and protonates any residues from the phosphoric reagent; finally, the
dimethyl amine cartridge removes any remaining acidic material to leave a pure stream of
the product.
10 Solid-supported flow synthesis of 2-aminopyrimidines
simple filtration. This can reduce the number of labour-intensive workup procedures such
as extractions and chromatographic purification that must be performed to obtain materials
of high purity.
Using these solid-supported reagents, the preparation of complex scaffolds using multi-
step pathways can be achieved46–48 with only a small number of intermediate purification
steps. A further advantage is that toxic, sensitive or valuable reagents can be confined on
the solid material25,46,49 creating a lower exposure hazard than that when working with
powders or volatile substances. It is also possible to recycle catalytic reagents easily as
they can be separated from the reaction mixture by a simple filtration. For these reasons,
polymer-assisted synthesis lends itself particularly well to automation50 to generate large
arrays of compounds.
Polymer-supported reagents are also well suited to flow chemistry processes, because they
can be confined within a cartridge so that the flow stream is not contaminated. They can be
used as reagents, catalysts, or scavengers which remove by-products to simplify multi-step
synthesis. For example, a three-step synthesis of triazole 7 from alcohol 6 and azide
5 reported by our group51 involved sequential reduction, Bestmann–Ohira reaction and
copper-catalysed alkyne–azide cyclisation reactions (Scheme 1.2). The excess unreacted
materials were removed by the scavenging columns to afford the product which was
collected by concentration and crystallisation.
However, the integration of polymer-supported reagents into flow chemistry processes can
present some difficulties, depending on the morphology of the polymeric structure.
1.1.3 Monolithic reagents
Since the description of Merrifield’s resin,44 by far the most common solid support systems
have been “gel-type” polystyrene resins which are prepared as small beads with a diameter
of 0.5 – 1 mm. These have a low degree of cross-linking (1 – 2 %), and as a result they tend
to swell and contract depending on the solvating power of a particular solvent (Figure 1.3).53
An alternative support method is the use of macroporous beads which have a higher degree
of cross-linking and thus greater rigidity, maintaining their structure across a wide range
of solvents. However, they still suffer from some of the other problems associated with
gel-phase beads. Heterogeneous flow processes involve five conceptual steps for mass
transfer between the two phases (Figure 1.4). High molecular-weight solutes diffuse slowly
1.1. Introduction 11
a b
Figure 1.3 In a “poor” solvent, which does not effectively solvate the polymer, the polymer
beads contract as the solvent is expelled, allowing the fluid stream to channel around them.
In a “good” solvent they swell up, a process that can block the flow stream. Whilst this
behaviour is not a problem for a batch-mode reaction, in a flow reactor these changes in
flow dynamic can impair the reaction.
into the stagnant solvent present in the pores of the beads. Although diffusion for low
molecular weight solutes is rapid, the large void volume (at least 27 % for spherical beads)
provides a low resistance flow path through which convective flow travels preferentially.54
1 2
3
4 5
Figure 1.4 Five steps for reactivity of solutes in a heterogeneous system. (1) External
mass transfer through convective flow. (2) Diffusion into stagnant solvent surrounding solid
and then into pores. (3) Chemical reaction. (4) Diffusion back into stream. (5) External
mass transfer.
Monoliths represent another class of immobilised reagent which has been developed to
overcome some of these problems. A polymer monolith is a uniformly polymerised porous
structure made by precipitation polymerisation of organic or inorganic monomers (as
opposed to a system of parallel channels such as found in a catalytic converter). Monoliths
were originally developed for protein chromatography,54 for which the slow diffusion was
the primary challenge. They were also found to be useful for flow chemistry, since the
porous polymer structure contains channels with diameters in a tight distribution around
400 nm, which are large enough to support convective flow54 without containing paths for
unimpaired channelling, and keep the flow in close proximity to the pores such that diffusion
between the external flow and the stagnant solution can occur rapidly.54 Since monolithic
reagents can be prepared in any shape, preparation within a glass column (Figure 1.5a)
results in a cylindrical monolith through which a flow stream can pass evenly.
12 Solid-supported flow synthesis of 2-aminopyrimidines
5 μm 10 μm
a b c
Figure 1.5 (a) Monolith prepared within a glass column. The plugs at either end are
replaced with fluidic connections for the subsequent steps. (b, c) Scanning-electron
micrographs showing the porous structure of a monolith.
Figures 1.5b-c show the structure of the monolithic polymer using scanning-electron
micrography. The visible structures constitute the microscale channels that allow convective
flow, as well as the nanoscale channels (<100 nm54), which allow reagent ingress only by
diffusion. The pore sizes can be tuned by altering the polymerisation mixture to create a
robust and reactive monolith.
Monoliths have found applications as ion-exchangers,55 chromatographic packings,56 and
polymeric reagents and catalysts, for example for Diels-Alder,57 polymerisation,58 cross-
coupling,59 or cyclopropanation60 reactions. Many of these involved continuous flow
systems due to the benefits of flow-through reactors. In an example of a synthesis involving
polymeric reagents, an azide-exchange reactor produced by treatment of a quarternary
ammonium monolith with sodium azide was used to generate acyl azides (8) from acyl chlo-
rides (9).61 These underwent a Curtius rearrangement to produce an isocyanate (10) product
which could be recovered in quantitative yield by removal of the solvent (Scheme 1.6).
Monolithic reactors can be produced in the laboratory without requiring any specialist
equipment. They are made by co-polymerisation of a monomer or mixture of monomers
with a cross-linker and a porogen, usually an alkyl alcohol that is also the solvent for the
monomer. As the polymer chain length grows, its solubility drops until it precipitates from
the solvent; the pattern of phase separation produces the porous structure. The composition
of the solvent system determines the point at which this phase separation occurs and thus
affects the porosity. A balance must be struck between the solubility of the components
and the desired porosity, because if the porosity is too low then the monolith will create a
large back-pressure when placed in a flow stream.
In summary, monolithic reagents are a powerful tool for the application of polymer-
1.1. Introduction 13
100 psi
MeCN
0.5 mL/min
20 min
120 °C
R
O
N3
R N C O
R
O
Cl
NEt3 N3
13 min
Azide monolith
Drying
agent
9
8
10
Scheme 1.6 Use of an azide exchange monolith to generate acyl azides, which were
then subjected to high temperature reactor coil to perform a Curtius rearrangement
reaction. The drying agent cartridge (sodium sulfate) ensured that any water present in
the acetonitrile solvent was removed to prevent undesired reactions with the isocyanate
product.
supported synthesis methods to flow processes. We proposed to create a monolithic
reagent to assist with the synthesis of compounds containing the 2-aminopyrimidine motif.
1.1.4 Proposed synthetic route to 2-aminopyrimidines
For the synthesis of 2-aminopyrimidines, we designed a route based on a monolith-
supported isothiourea. By attaching this reagent onto a solid support (Scheme 1.7),38,39
release of sulfurous by-products can be prevented, and the potential for contamination with
thiourea is reduced.
SH2N
NH2Cl
HS
N S
N
R
N
N
R
N
R'
R'R''-NH
isothiourea 2-thiopyrimidine sulfur retained
R''
Scheme 1.7 Proposed route to 2-aminopyrimidines involving a solid-supported sulfur
moiety.
14 Solid-supported flow synthesis of 2-aminopyrimidines
1.2 Development of Chemistry
1.2.1 Generation of the monolithic reagent
Previous work within our group62 had investigated the use of compound 11 as a monomer
for the preparation of a monolith which might be used for the synthesis of imidazoles and
pyrimidines using a solid-supported technique.38 However, generation of the monolith had
proven to be challenging, most likely because the polarity of 11 in its hydrochloride salt
form made it poorly soluble in dodecanol, an alcohol used as the porogen. Attempts to
solubilise the monomer with a methanol co-solvent led to an amorphous monolith, and
this was speculated to be the result of having too much solvent in the polymerisation
mixture. Experimentation by Dr Nikzad Nikbin, based on the use of a water/propanol
solvent system,63 resulted in a monolith with a much more favourable microstructure,
having a rigid structure and a back-pressure profile appropriate for use in a flow reactor
(Scheme 1.8).
This water/propanol solvent system is important because it allows the polymerisation of
charged monomers. This can be beneficial in cases such as this one, where the isothiourea
starting material was used in its (easily purified) hydrochloride salt form. Salts are generally
very difficult to polymerise using the traditional methods for making polymer beads, because
beads tend to be made in an aqueous/organic biphasic system,64 so amphiphilic monomers
such as 11 can give poor results. In fact, Masquelin and co-workers38 used the Merrifield
resin as a starting material in their synthesis of the same solid-supported isothiouronum salt,
rather than attempting to polymerise a functional monomer. However, direct polymerisation
of the salt is advantageous because this is likely to result in a solid reagent with consistent
functionality across the entire polymer; and if it is made as a cross-linked monolith, then its
SH2N
NH2Cl
ACHC
6.6:1 1-propanol/water
90 °C, 20 h
SH2N
NH2Cl
11 M1
Scheme 1.8 Preparation of functionalised monolith M1. Polymerisation conditions: 4-
(vinylbenzyl)isothiuronium chloride (11; 18.8 % w /w); divinyl benzene (DVB, cross-linker;
12.5 % w /w); 1-propanol (porogen; 57.5 % w /w); water (porogen; 10.9 % w /w); 1,1′-
azobis-(cyclohexane carbonitrile) (ACHC, initiator; 1 % w /w relative to 11 + DVB).
1.2. Development of Chemistry 15
morphological properties can be tuned during polymerisation.
The monomer (11) was easily prepared by a nucleophilic substitution of the chlorine moiety
of 4-vinylbenzyl chloride (12) with thiourea (Scheme 1.9). The presence of an imine
functionality was suggested by an absorbance at 1655 cm−1 in the infra-red spectrum, and
a 13C resonance at 169 ppm. In the 1H NMR spectrum, the two broad singlet resonances
at 9.4 and 9.5 ppm from the NH2 groups are in very similar chemical environments, as
expected for this structure. The iminium centre is more deshielded than would be expected
for a monoprotonated guanidine moiety,65 possibly due to the greater electronegativity of
the sulfur atom compared to an amino group.
Following the polymerisation, the new monolith was washed with either ethanol or THF to
remove the residual porogen and any unpolymerised material. Very little of the monomer
was observed in the efflux stream, indicating a good degree of polymerisation. Having
identified an efficient method for producing the functional monoliths, methodology for the
solid-supported synthesis of 2-aminopyrimidines could be explored.
SH2N
NH2Cl
Cl
H2N
S
NH2
EtOH
reflux, 6 h
13 12 11
Scheme 1.9 Preparation of the functionalised monomer 11 by condensation of thiourea
(13) with 4-vinylbenzyl chloride (12).
1.2.2 Monolith characterisation
One of the drawbacks of solid-phase synthesis is that analysis of the reaction intermediates
is much more challenging than for a solution phase synthesis. Conventional methods such
as chromatography or mass spectrometry can only be used for fluid samples, and so it
was only possible to use these to characterise the waste stream after a reaction. Elemental
microanalysis proved to be a useful technique for following the reactions when on the solid
support, particularly as many of the transformations involved a change in the number of
nitrogen atoms.
Although solid samples might not normally be considered for NMR analysis, solid state-
NMR by “magic angle” spinning makes use of a method for overcoming the significant
line broadening brought about by anisotropy within the sample. Because the positions and
16 Solid-supported flow synthesis of 2-aminopyrimidines
SH2N
NH2Cl NC
H2N N S
N(iPr)2NEtEtOH
EtO
CN
CN
25 °C, 30 min
83%1.01 eq11
14 15
Scheme 1.10 Solution-phase cyclisation of functionalised monomer with EMM.
orientations of molecules in a solid are fixed, there is no random tumbling which in the
solution phase averages out time-dependent interactions. However, by spinning a sample
at 54.7° relative to the magnetic field (the “magic angle”) the influence of anisotropic
interactions is reduced, providing greater signal resolution. The resulting spectral peaks
are often still quite broad, but in some cases the peak shapes can provide further structural
information about the sample.66
It can also be very useful to record the dispersion pattern generated by a solid packed bed
such as a monolith, to confirm that the desired porosity has been generated. This could be
achieved by injecting a pulse of a dye, a salt solution or a UV-active compound into the
column, and then measuring the concentration profile of the output stream using a visible
light, conductivity or UV detector.67
Monolith M1 was characterised by infra-red and 13C NMR spectrometry. The iminium
resonance at 171 ppm was conserved in the NMR spectrum; in the IR spectrum the absorp-
tion at 1647 cm−1, attributed to the imine C−N bond, was also still visible. These data
suggested that the key thiouronium structure was preserved during the polymerisation.
In some cases reactions were tested in the solution phase using unpolymerised monomers,
allowing traditional analytical techniques to be used, before attempting a solid-phase
reaction. However, solution- and solid-phase conditions are not always comparable; for
instance, more forcing conditions were often needed for the heterogeneous reactions.
1.2.3 Heterocycle Formation
Previous reports37,68,69 indicated that an isothiourea would combine with a bifunctional
compound such as ethoxymethylene(malononitrile) (14, EMM)38 to form a heterocyclic
product. Pleasingly, this was found to be the case; mixing 11 and 14 at room temperature
afforded pyrimidine 15 as a precipitate (Scheme 1.10). Hünig’s base (DIPEA) was selected
1.2. Development of Chemistry 17
for these reactions because, being a liquid, it was more amenable to this flow process than
the inorganic bases such as K2CO3 used in previous reports.
In a mixed-phase system, stirring a crushed monolith overnight with EMM and DIPEA
afforded a yellow solid after filtration and washing (Scheme 1.11).
SH2N
NH2Cl NC
H2N N S
N
(iPr)2NEt
EtOH
EtO
CN
CN
25 °C, 18 h
M1 14 M2
Scheme 1.11 Solid-phase cyclisation of functionalised monolith with EMM.
The formation of the desired pyrimidine M2 was supported by evidence from the elemental
analysis, showing 8.8 mmol/g of nitrogen and 2.9 mmol/g sulfur,† suggesting approximately
1.8 mmol/g of the pyrimidine, with 0.8 mmol/g of the isothiourea remaining unreacted. The
infrared spectrum showed characteristic signals, such as new absorptions in the regions
of 1250 and 1400 cm−1, as well as a strong absorption at 2217 cm−1, indicative of a C−−N
functional group. The disappearance of the thiouronium centre at 171 ppm could be
observed by solid-state NMR spectroscopy experiments. When performed using a flow
reactor (Scheme 1.12), the progress of this reaction could be observed by a change in colour
of the monoliths from white to yellow (Figure 1.13).
For further information about the functional loading of M1, a known quantity of EMM (14)
DIPEA 
(2.5 equiv.)
MeCN
0.1 mL/min
(2 equiv.) 1:1 DMF/MeCN
0.1 mL/min
CN
CN
EtO
100 psi
unreacted 
EMM, DIPEA
+
DIPEA.HCl
14
M1 M2
Scheme 1.12 Flow schematic for the transformation of M1 to M2. EMM (14) and DIPEA
are combined in a tee mixer and passed through the monolith at room temperature. The
solvent mixture was chosen for maximum solubility of EMM. Only unreacted starting
materials were observed by NMR spectroscopy of the reactor output.
† The sulfur proportion was inferred from the values measured for carbon, hydrogen and nitrogen, and was
consistent with a 1:1.17 ratio of monomer (11) to DVB.
18 Solid-supported flow synthesis of 2-aminopyrimidines
Figure 1.13 The monolithic reagents assumed a yellow colour during the cyclisation
reaction with EMM (14). The uneven distribution of this colour in the initial stages of the
reaction indicates some degree of channeling within the structure.
was injected into a monolith. The quantity of EMM remaining in the output stream was
measured, indicating that 3 mmol of 14 had reacted with the monolithic reagent (2.0 mmol/g
for a monolith weighing approximately 1.5 g). Elemental microanalysis of samples taken
from the resulting monolith M2 at the ends, from immediately beneath the edges, and from
the core reproducibly gave similar results, indicating a consistent reaction throughout the
monolith.
1.2.4 Oxidation of monolith M2
Unlike the methylthiol moiety,34,37 the benzyl sulfur linkage has been reported to be
unreactive to substitution without first performing an oxidation of the sulfur atom.39 meta-
Chloroperoxybenzoic acid (m-CPBA) is a oxidant which is highly soluble in organic
solvents and which has precedence for this oxidation (Scheme 1.14).38
NC
H2N N S
N
m-CPBA
CH2Cl2
25 °C, 18 h
NC
H2N N S
N
O O
15 16
Scheme 1.14 Solution-phase oxidation using m-CPBA.
In a mixed-phase system, oxidation of monolith M2 was performed at room temperature
using an excess of m-CPBA to encourage complete oxidation of all the sulfur atoms
(Scheme 1.15). The progress of the reaction could be observed in the infrared spectrum,
with new strong peaks at 1038 cm−1, 1127 cm−1 and 1325 cm−1 suggesting the presence of
1.2. Development of Chemistry 19
both sulfoxide and sulfone groups. This was puzzling given that an excess of the reagent
had been used. However, the exact composition and the effects of this could only be
calculated after the nucleophilic release step had been performed.
NC
H2N N S
N
m-CPBA
CH2Cl2
25 °C, 18 h
NC
H2N N S
N
O O
NC
H2N N S
N
O
M2 M3a M3b
Scheme 1.15 Oxidation of monolith M2.
This “safety-catch” activation represents an important advantage of the polymer-supported
technique, because reactions with other nucleophiles could be performed to access further
derivatives of the supported heterocyclic intermediate whilst retaining it on the solid
support. The use of latent chemical functionality in this way can provide an advantage by
reducing the number of steps needed to reach a complex target. For the synthesis of the
2-aminopyrimidines, no protecting group would be required for the sulfur linkage, so the
reactivity of the heterocycle could be exploited without impacting the step economy70 of
the overall process.
1.2.5 Release of 2-aminopyrimidine product
Following activation, the oxidised thioether could be displaced using an amine nucleophile
such as morpholine (17) to provide the desired 2-aminopyrimidine (18). Initial experimen-
tation revealed that long residence times were required for a complete reaction even at
100 ◦C. Therefore, we decided to investigate the use of a stop–flow technique that would
increase the residence time without requiring a lower flow rate than could be achieved
reliably using the pumps.
Once approximately one column volume worth of amine solution had been infused, the
flow was stopped for a period of time whilst the column was kept at an elevated temperature.
This cycle was repeated multiple times to achieve a prolonged residence time, whilst
maintaining a high local concentration of the nucleophilic reagent. Although there was
no flow of solution within the column, mixing would continue by diffusion allowing the
reaction to continue during these periods.
20 Solid-supported flow synthesis of 2-aminopyrimidines
SN
N
H2N
O
NN
N
H2N
1,4-dioxane
80 °C, 8 h
HN
O
NC NC
O
SN
N
H2N
O O
NC
+
M3a M3b
18
35 %
Scheme 1.16 Reaction of monolith M3 with morpholine to afford the heterocyclic product
18. The yield is calculated based on the measured loading of isothiourea monolith M1.
Following concentration of the output, the pyrimidine product 18 was obtained following
a single chromatographic purification. Therefore, a large proportion of the experimental
effort was spent operating the flow reactor, and very little reagent and intermediate handling
was necessary to obtain the product. We envisaged that an automated system could perform
many parts of this synthesis with minimal manual intervention.
1.3. Automation 21
1.3 Automation
1.3.1 Advantages of automation
The “catch–react–release” method described is essentially a serial batch procedure, and so
the throughput of material is restricted. The size of each batch is limited by the amount
of material loaded onto the monolith, and under the thermal polymerisation conditions
frequently employed there is currently an upper limit to the size of monolith that can be
prepared with a favourable morphology.61 Furthermore, the monoliths were not reusable
and to avoid contamination we used new columns for each reaction, so a large number of
them had to be prepared. Using the column heater modules for a Vapourtec R2+/R4 reactor
unit, we were able to produce batches of four monoliths at a time (Figure 1.17), which was
sufficient for the pace of work.
However, the protocol for generating the monoliths involved heating the polymerisation
mixture for 20 hours. When this procedure was performed manually, preparation of the
polymerisation mixture had to be started at about 3 pm to ensure that the monoliths would
be ready and the heater could be switched off early the next morning. By implementing a
timed shut-off, the preparation could be done at any point during the day, which was more
practical.
Of course, this simple control task could be achieved with an off-the-shelf plug timer, but a
more complete solution was required for the later synthesis steps to enable a programmed
sequence of operations to be carried out. In these cases, the software available for control of
a b c
Figure 1.17 (a) Some of the different sizes of glass column available. From left to right:
3 mm ID × 100 mm length, 6.6 mm ID × 50 mm length, 6.6 mm ID × 100 mm length,
10 mm ID × 100 mm length. (b) Heating jackets installed on a Vapourtec R4 heating unit.
(c) Monoliths being prepared in parallel.
22 Solid-supported flow synthesis of 2-aminopyrimidines
the R2+/R4 unit was also unsuitable, because it was not possible to add new logic beyond
the prescribed applications, and because in some cases we wanted to incorporate additional
devices to be controlled simultaneously. Therefore, custom computer software had to be
created to control these experiments; the 20 hour heating programme will serve as an
introductory example to explain how the solution was created and used.
Vapourtec kindly provided details of how their reactors can be remotely controlled. The
reactor connects to a PC using an RS-232 serial cable, and then the communication is
carried out using a text language (Table 1.18). Communication though a serial port in this
way is well within the capabilities of any modern programming language. The Python
programming language was chosen because it is quick and easy to read and write, and
commonly used for scientific software.71 A sequence to set up the reactor, wait for 20 hours
and then turn off the system is shown in Listing 1.19.
1.3. Automation 23
Table 1.18 Typical commands and responses for the Vapourtec R2+/R4 unit.
Action Command Response
Switch power on "" "OK"
Switch power off "" "OK"
Set Temperature "  0 100" "OK"
Listing 1.19 This programme sets all four heaters to 90 ◦C, runs for 20 hours and then
switches off the R4. Note that no data is collected during the experiment and there is no
facility for checking that the commands have been received. The \r\n (carriage-return
and new line) characters are used to terminate commands. Please note that the serial
commands have been redacted in the PDF version of this document.
# Python includes modules for serial communication and timing
import serial
import time
# Set up a connection to the R2
connection = serial.Serial(
"COM1",
baudrate = 19200,
timeout = 1
)
# Set all four heaters to 90 degrees
for i in range(4):
connection.write("  {:d} 90\r\n".format(i))
time.sleep(0.1)
# Turn on the heaters
connection.write("\r\n")
# Wait for 20 hours
time.sleep(20 * 60 * 60)
# Switch off the heaters
connection.write("\r\n")
24 Solid-supported flow synthesis of 2-aminopyrimidines
1.3.2 Solvent switching and stop-flow
A second application of automation was that when working with a monolithic reactor, it
is important to make gradual solvent switches, much like must be done when changing
the solvent within a chromatographic column. This is because a sudden change in solvent
composition can block the pores or cause the structure to collapse. Another application was
the stop–flow protocol which otherwise required many carefully timed manual operations.
Automation of the solvent switching protocol (Listing 1.20), and the stop-flow protocol
(Listing B.1 in Appendix B) created a much more efficient working regime, where the bulk
of these tasks could be managed by a computer.
Initially, we encountered a problem in some cases relating to the under-pressure shut off
feature which is hardwired into the Vapourtec machine. Once the reactor is running and
the pressure has built up to over 3 bar, if the pressure then drops below this level then the
pumps and heaters are automatically switched off after a period of time and the machine
reports an error.72 In a normal flow chemistry procedure this is very useful, because a
pressure drop may indicate a leak or a sudden rupture in one of the flow tubes. However,
when pumping a solution slowly through a monolith this system could be triggered during
Listing 1.20 Example of a programme to change the solvent mixture gradually from
100 % solvent A (orange channel) to 100 % solvent B (purple channel) in four increments.
This programme makes use of an abstraction object (vapourtec.R2R4) that handles the
RS-232 communication (see Appendix A).
import transport, vapourtec
R2 = vapourtec.R2R4(transport.serial("COM1"))
flow_rate = 200 # Total flow rate
steps = 4 # Number of increments to make
# Calculate change in flow rate for each step
step = flow_rate / steps
for i in range(steps):
change = (i + 1) * step
# Adjust the flow rate of both pumps
R2.flowrate("orange", flow_rate - change)
R2.flowrate("purple", change)
# Run each solvent composition for 20 min
R2.wait(mins = 20)
1.3. Automation 25
normal operation. Furthermore, if the structure of the monolith was not perfect then large
pressure fluctuations could sometimes occur, triggering the safety shut off. Most crucially,
it was also triggered when the pump was switched off but the heaters were still running,
during the “stop” phase of a stop–flow sequence. The pressure gradually fell to nothing
after the pumps were stopped, having the undesired effect that the heaters were switched
off as well.
To circumvent this problem, the programme performed frequent checks on the status of
the machine during each waiting period, as part of the R2.wait() function. If it had
stopped due to an under-pressure then the reactor was started again allowing the sequence
to continue uninterrupted.
1.3.3 Multi-step processing
The combination of solid-phase synthesis and automation means that multiple synthetic
steps can be programmed to take place in a sequence, in much the same way as a solid-phase
peptide synthesiser is used. The valves on the R2+/R4 reactor can be used to select reagents
and solvents for different phases of the synthesis.
EtOH
DIPEA / MeCN
MeCN
Reagent / MeCN
100 psi
orange
waste
purple
1 1
5 5
Figure 1.21 Reactor configuration for sequential washing and loading of four monoliths.
A multi-position valve was added to take advantage of the four heating manifolds available
(Figure 1.21). During stop-flow sequences the reagents could be loaded onto one monolith
whilst the others were heating, allowing multiple reactions to be performed simultaneously.
All four could be washed and then the cyclisation reaction performed on each in turn.
Listing B.2 (Appendix B) shows a programme to wash four new monoliths and then react
each in turn with a solution of a reagent.
A similar system could be used for performing the nucleophilic release step with four
different amines (Figure 1.22, Listing B.3 in Appendix B). In this case the valve was
26 Solid-supported flow synthesis of 2-aminopyrimidines
connected to four reservoirs containing the amines, and four collection vessels. A separate
wash line allowed the dead volumes to be cleaned out between switching the reagents. NMR
measurement of the collected materials confirmed that there was no cross-contamination
between the amine reagents.
Dioxane
Dioxane
m-CPBA / DCM
DCM
Amine 1
Amine 2
Amine 3
Amine 4
100 psi
100 psi
100 psi
100 psiorange
load
inject
waste
collect 1
collect 2
collect 3
collect 4purple
1
5
5
1
Figure 1.22 Reactor configuration for nucleophilic displacement from four monoliths with
four amines. The additional valve is used to enable thorough rinsing of the system to avoid
cross-contamination. Since the two multi-position valves were part of the same unit, Amine
1 was infused into column 1, Amine 2 into column 2, and so on.
Multiplexing reactions in this way improves the efficiency of a single piece of apparatus.
The complex switching schedule required makes the computer control an important addition.
After programming the appropriate sequence of operations, the chemist can be confident
that it will be carried out correctly without having to (for example) set lots of stopwatches
for himself, or worrying about getting one step wrong.
Having developed these automation methods, they were applied to the preparation of
analogues of a precursor to the kinase inhibitor imatinib.
1.4. Application to the preparation to derivatives of imatinib amine 27
1.4 Application to the preparation to derivatives of ima-
tinib amine
Imatinib (1) is a drug marketed by Novartis AG for the treatment of chronic myeloid
leukaemia and gastrointestinal stromal tumours.33,73–75 It is marketed in the USA (as
Gleevec) and in Europe, Australasia and Latin America (as Glivec), in both cases as its
mesylate salt. We envisaged that the polymer-supported methodology for the synthesis of
2-aminopyrimidines could be applied to the synthesis of compound 24 (Scheme 1.23), an
important precursor to imatinib.
1.4.1 Reported syntheses of imatinib
A number of routes for the synthesis of imatinib have been reported. Many have a number
of similarities to the initial discovery route patented by Zimmermann and co-workers.33,73
N
N
N
H
N
X
O
N
H
N
N
N
N
H
N
HN
O
N
N
Cl
NH2
H2N
H
N
NO2
N
O
NMe2
NO3
H2N
NO2
N
O
1. NH2CN,
    HCl, i-PrOH
    80 °C, 3 h
2. HNO3 
    60 °C
    81% (2 steps)
NaOH
DMF, H2O
reflux, 8 h
73% (2 steps)
X = NO2
X = NH2
MeO OMe
NMe2
neat
reflux, 3.5 h
Cl
+
N
N
N
H
N
HN
O
Cl
Raney Ni
N2H4.H2O
MeOH
35 °C, 5.5 h
81%
neat
130 °C, 1.5 h
96 %
1. K2CO3, THF
    20 °C, 1 h
2. H2O
    20 °C, 80 min
    
    98% (2 steps)
19
20
21b
22 23
24,
25,
26
1 27
Scheme 1.23 Szczepek’s route to imatinib,76 involving condenstion of enaminone 21b
with guanidine 23. This route is similar to the original discovery route.33,73
28 Solid-supported flow synthesis of 2-aminopyrimidines
The key 2-aminopyrimidine is constructed by a condensation reaction between an enone
(21b) and a guanidine (23) (Scheme 1.23).76–81 One patented route82 involved the Negishi
coupling of 3-pyridyl zinc bromide with 2,4-dichloropyrimidine which avoided the high-
temperature enone condensation.
The key amide bond is often formed by acylation of the aniline revealed by reduc-
tion of a nitro group such as that present in compound 24. A number of methods for
amide coupling have been reported, including aluminium trichloride mediated reactions
with carboxylic acids or esters (Scheme 1.24, Route A),77 as well as using common
peptide coupling reagents.83,84 The combination of an acyl chloride with an aniline
(Scheme 1.24, Route B)76,77,79,80,85 is a particularly high-yielding method. Construc-
tion of the 2-aminopyrimidine centre has also been reported using cross-coupling methods
such as by a Buchwald coupling (Scheme 1.25)36,77,86 or by Ullmann coupling.87
N
N
N
H
N
NH2
N
N
N
H
N
HN
O
N
N
HO
O
N
N
AlCl3 (2.5 equiv.)
PhMe / MeCN
40 °C, 8 h, 55%
Cl
O
N
N
Pyridine
50 °C, 4.5 h, 70%
Route A
Route B
25
28
29
1
Scheme 1.24 Formation of the amide bond from aniline 25.
Finally, the synthesis of imatinib has also been performed using some advanced synthe-
sis techniques. Our own group developed a flow synthesis procedure36 involving solid-
supported reagents and scavengers, and a catch-and-release method for the preparation of
the aryl bromide cross-coupling partner 31. This compound was obtained in high purity
with no manual purification. Leonetti and co-workers reported a solid-supported synthesis
of imatinib (Scheme 1.26),88 although this method required a double protection of the
guanidine moiety as well as a number of expensive or toxic reagents such as BEMP and
HgCl2 to force the reactions.
1.4. Application to the preparation to derivatives of imatinib amine 29
N
N
N
H
N
HN
O
N
N
N
N
N NH2 Br
HN
O
N
N
conditions
+
30 31 1
Scheme 1.25 Synthesis of imatinib by cross-coupling methods. Conditions: Pd2(dba)3 ·
CHCl3, NaOtBu, rac-BINAP, xylene, reflux, 5 h, 72 % after reverse phase HPLC;
77
BrettPhos precatalyst (2 mol %), BrettPhos (2 mol %), K2CO3, t-BuOH, 110
◦C, 6 h, 84 %;86
BrettPhos precatalyst (2.5 mol %), NaOt-Bu, 2:1 1,4-dioxane / t-BuOH, 150 ◦C, 30 min
(continuous flow), 69 % after column chromatography.36
O
H
O2N
NH2
Merrifield
Resin
O2N
NH
1. Ti(OiPr)4,
    NEt3, THF
    25 °C, 18 h
2. NaBH(OAc)3
    DCM, 
    25 °C, 4 h
    80% (2 steps)
H2N
N
O
N
N
N
H
N
O
N
N
AllocN
AllocHN
SMe
AllocN
AllocHN
HgCl2, NEt3, DMF
0 °C, 10 min
then 80 °C (MW), 5 min
+
N
H
N
O
N
N
NH
H2N
1. Enaminone
    PhNO2, BEMP
    120 °C (MW)
    50 min
2. TFA, CH2Cl2
    25 °C, 1 h
    90% (2 steps)
imatinib
Pd(PPh3)4
PhSiH3
CH2Cl2
25 °C, 1 h
98% 
(2 steps)
32
22
33
3435
36
(37)
Scheme 1.26 Solid-phase synthesis of imatinib. The steps were generally high-yielding
but some unattractive reagents were required.
30 Solid-supported flow synthesis of 2-aminopyrimidines
1.4.2 Monolith-assisted route to imatinib
A collection of enaminones (21) were produced by treatment of an acetylated aryl com-
pound with DMF-DMA (20), following Szczepek’s protocol.76 These were reacted with
monolith M1 using the flow procedure described earlier in this chapter to afford the different
thiopyrimidine monoliths M4. By heating the monolith to 80 ◦C during the heterocycle
formation reaction, this transformation proceeded to completion as indicated by elemental
analysis. From the oxidised monoliths M5, analogues 38 were prepared (Table 1.28).29
These compounds were isolated in yields from 29 – 72 % (based on a 4.2 mmol loading
of monolith M1) over the three steps with only a single chromatographic purification.
This represents a powerful technique for the preparation of a number of derivatives of a
compound of interest, without requiring a large number of manual purification operations.
In order to transform compound 24 into imatinib, we chose to use a coupling with an acyl
chloride.87 Before this could be carried out, it was necessary to perform a reduction of the
nitro group of 24. A number of different conditions have been reported to be effective for
this procedure, including Pd/C and H2,
73 SnCl2,
78 hydrazine and FeCl3,
87 hydrazine and
Raney Nickel,76 and sodium borohydride with CoCl2.
76 We chose to employ Pd/C catalysis
using an H-Cube reactor (Scheme 1.27), which performs heterogeneous catalysis using
hydrogen generated in-situ in an electrolytic cell. No toxic metals like SnCl2 are involved
and the use of a heterogeneous catalyst reduces the purification requirement.
The resulting aniline (25) was treated with an acyl chloride, and then an SN2 reaction
afforded imatinib (1) after purification by column chromatography.
N
N
N
H
N
HN
O
N
N
N
N
N
H
N
NO2
O
Cl
H2 (15 bar)
Pd/C 
CH2Cl2
H-Cube
40 °C
73%
HN
N
Cl
2.
NEt3 (1.15 equiv)
THF, 0 °C, 3 h, 57%
MeCN, 150 °C (MW)
2 h, 53%
N
N
N
H
N
NH2
1.
24 25 1
Scheme 1.27 Synthesis of imatinib (1) from precursor 24. Yields are unoptimised.
1.4. Application to the preparation to derivatives of imatinib amine 31
Table 1.28 Preparation of 2-aminopyrimidine derivatives 38.
SH2N
NH2Cl
SN
N
Ar
Ar
O
NMe2
DIPEA
EtOH, 80 °C
SN
N
Ar
m-CPBA
CH2Cl2
O ONR
1R2N
N
Ar
R1R2NH
1,4-dioxane
90−120 °C
a  Ar = 4-pyridinyl
b  Ar = 3-pyridinyl
c  Ar = C6H5
d  Ar = 4-(CF3)C6H4
M1
21
M4a–d
M5a–d24, 38
Entry Monolith Amine Product Product structure Yielda
1 M5a
NH
O 38a N
N
N
NO
72 %
2 M5a
NH2
OMe
38b N
NN
H N
OMe
29 %
3 M5a
NH2
38c N
N
N
H N
50 %
4 M5b
NH
O 38d
N
NN N
O
71 %
5 M5b
NH2
NO2
24
N
NN
H
NO2
N
48 %
6 M5c
NH
O 38e
N
NN
O
40 %
7 M5d
NH
O 38f
N
NN
O CF3
62 %
a Isolated yield over all three steps, after purification by flash column chromatography,
based on an assumed 4.2 mmol loading of monolith M1.
32 Solid-supported flow synthesis of 2-aminopyrimidines
1.5 Conclusion
In this chapter an automated, multi-step synthesis for a focused array of substituted 2-
aminopyrimidines was described. The method involved the generation of a polymer-
supported pyrimidine which was then released to afford an advanced intermediate. A
significant advantage of this procedure is that it avoids problems of low solubility which
are commonly encountered when working with heterocyclic compounds. Indeed, attempts
to directly polymerise intermediate 15 were unsuccessful primarily due to its low solubility
(Scheme 1.29).
S S
N
N
NC
H2N
N
N
NC
H2N
Polymerisation
Conditions
15 M2
Scheme 1.29 Attempts to prepare a monolith using cyclised monomer 15 were unsuc-
cessful, primarily due to its poor solubility.
Solid phase synthesis is a powerful technique, which can eliminate the need for manual
purification methods and thus is very useful for library synthesis. In this case, the oxidative
“safety-catch” could enable further derivatisation without the need for protecting groups.
However, solid-phase synthesis is strongly associated with combinatorial chemistry and so
has generally been overlooked since the latter has fallen out of fashion. Nevertheless, when
combined with programmed automation attractive synthetic pathways can be harnessed
without the need for specialised combinatorial equipment.
The introduction of automation techniques allowed a number of compounds to be rapidly
prepared by what would otherwise be a relatively labour-intensive process. The potential
applications of automation within synthesis are widespread, and in the next chapter a
more generalised system for controlling laboratory devices such as the R2+/R4 reactor is
described.
Chapter 2
New control and monitoring software
for flow chemistry systems
33
34 New control and monitoring software for flow chemistry systems
2.1 Introduction
As continuous processing techniques such as flow chemistry become more commonplace in
modern laboratories, there is an increasing demand to expand the capabilities of laboratory
devices. For instance, the ability to share information between different devices enables
seamless co-ordination of an experimental protocol to be achieved. We may wish for
the sensor reading from a spectroscopic detector to determine the flow rate of a pump,89
or to trigger a multi-position valve to collect samples at regular intervals throughout the
procedure (Figure 2.1). One particular challenge in this regard is communication with
different equipment made by different manufacturers.
heaterdetector
controlled 
pump
supported reagent fractionsdetector
Control
Figure 2.1 Heterogeneous reagents produce additional dispersion of a reagent slug
within the flow system. By connecting a detector and a pump, addition of a reagent
stream after the column could be performed at a varying flow rate to match the immediate
concentration of the substrate. The same control computer could also interact with other
devices, such as valves or fraction collectors, in an integrated fashion.
There have been a number of examples of integrated flow systems reported recently, such
as for multi-step synthesis,14,89,90 combined synthesis and purification,91 or new work-up
technologies,92 which confirms that this is an important aspect of laboratory flow chemistry.
Although manufacturers generally provide appropriate control software for use with their
particular devices, these are often limited to the applications that the manufacturer had
envisaged, and to other devices sold by the same manufacturer. For commercial reasons,
the control software is invariably provided in a closed format that cannot be extended to
implement control of additional hardware. Fortunately, most manufacturers are willing to
share control commands for their products; however, extensive integration work is often
necessary in order to connect different devices such that data can be passed between them.
In a manufacturing environment, process control systems tend to be custom designed
(involving a significant financial and time investment) for each individual setup; the cost
2.1. Introduction 35
and specific nature of the resulting system make this unsuitable for a research laboratory,
where a particular reactor configuration may only exist for a few days or weeks. Ideally, a
chemist should be able to define a sequence of operations to carry out a control protocol
for any new process with minimal up-front effort: it ought to be possible to prototype the
control system for an experiment in a few hours at most. In this way, more time can be
spent on the engineering and chemistry challenges inherent in the synthesis process, rather
than the logistics of control systems and communications interfaces.
Furthermore, the control software needs to be sufficiently flexible to enable different
pieces of equipment to be swapped in and out of the integrated system without having to
make significant changes to the automation protocol. Whilst an industrial process might
be designed for hardware that will be dedicated to that system, a process in a research
laboratory would have to use whatever devices are available. Flow reactor devices tend to
be modular and functionally interchangeable, so they ought to be interchangeable in the
control algorithm as well, allowing maximum reuse of programmes from one experiment
to another.
2.1.1 Existing control software
A large number of software projects can be identified that have been developed to enable
laboratory automation. The majority are aimed at running analytical devices, since these
are now invariably automated with data stored directly in electronic lab notebooks (ELN,93
such as the IDBS E-Workbook94 as used within our own research group95) or in laboratory
information management systems (LIMS)96 such as that sold by Thermo Scientific.97
A significant minority are designed for device automation. Clarity,98 developed between
the Evolutionary Biology and Systems Biology departments at Harvard University, is an
automation management programme written using the C Sharp (C#) programming language
to interface with laboratory robots, and includes a scheduler for executing sequences of
operations, in particular allowing each instrument to be involved in many such protocols
simultaneously. It also has an interface for starting and monitoring protocols, and an
alerting system for monitoring the computer that is running them. Importantly, it has the
ability to resume a protocol from a stored log file after a system crash. However, the control
protocols are stored simply as lists of method calls, allowing very limited decision making
abilities.
36 New control and monitoring software for flow chemistry systems
The aim of the project was to provide an alternative to expensive and proprietary commercial
integration and task-scheduling solutions such as those provided by Peak Analysis and
Automation Ltd.,99 PerkinElmer Inc.100 and infoteam Software AG.101 These systems are
generally not suitable for academic use as they tend to be sold on a basis where a consultant
from the company will design and implement the controlling software for the particular
process that needs to be automated. Furthermore, Clarity and all of the products produced by
these companies are primarily aimed at the biochemical sector. Clarity includes interfaces
for an incubator, a plate handler and a transfer station, and the case studies described for
the commercial products revolve around plate handling and reading.
The SiLA (Standardization in Lab Automation) consortium,102 a prominent laboratory
automation standards body, has published specifications for interfaces with different classes
of device.103 These specifications have been primarily developed to address the require-
ments for in vitro screening tasks, and this is confirmed by the operations defined in the
specifications (for example: Move Plate, Dispense Liquid Volume, Read Plate, and so on).
However, the idea of a specification is very important, because if similar devices have
the same interface, then the physical unit can be exchanged without requiring extensive
modifications to the control protocol.
Flow chemistry applications offer different challenges — most importantly, the control
of reactors based on the current state of other devices and instruments. Each process
will tend to have a dedicated set of physically and fluidically integrated apparatus for
the duration of its operation, so there is no need for the software to allow each device to
participate in multiple experiments. Rather, its focus should be on responding to incoming
information correctly and rapidly. Flow reactors are modular devices which are frequently
repurposed for many different experiments, so each control protocol should represent a
low time investment. This is not necessarily possible with the currently available software.
Therefore, a new open access and freely-available control solution would be a valuable
addition to the flow chemistry toolbox.
Some reported projects89,104,105 have designed automation protocols using LabVIEW,106
which is a visual programming tool and control platform developed by National Instruments
Corporation. Notably, LabVIEW has also been used to create the control software for
commercially available Uniqsis107 and Syrris108 devices. A similar tool is OpenAPC109
which offers a similar visual process development interface, for which new software
components can be created using the C++ programming language. Another popular
programming language is MATLAB,110 which works in a more conventional textual
2.1. Introduction 37
manner, and has been employed for some automation projects.111
A number of manufacturers provide LabVIEW or MATLAB interfaces for their instruments,
and so for simple tasks, the tools provided by these or other programming languages can
allow very rapid development. However, without a specification or standard protocol
scheduling system, the software for each new flow chemistry process must be created from
scratch. This has largely been the case with previous projects:89,92,112–114 for each new
project the researchers created a control programme to send the raw commands to the
particular machines integrated into the synthesis process. Frequently, the control system is
not published with the work; but even if it were, it may be useless without having exactly
the same set of equipment (with exactly matching firmware versions).
A more powerful solution would be to abstract each machine to its fundamental capabilities,
and work to a specification enabling similar reactors and devices to be exchanged easily;
each new protocol would be focused on its specific purpose rather than on the implementa-
tion of particular instruments. Furthermore, if common requirements such as recording all
of the data collected during a protocol and providing an interface to monitor this data in
real time are performed automatically, then researchers developing a new process will be
free to concentrate on other tasks.
Control platforms such as LabVIEW and MATLAB offer a promising ecosystem for
running lab automation, but the key disadvantage is the closed-source and proprietary
nature of their distribution and data formats. To overcome these disadvantages we selected
the Python programming language.115 Its open-source nature encourages collaboration,116
and enables technology to be transferred between institutions without the large initial set-up
costs typically involved with commercial packages. There are a number of libraries created
for Python which are specifically targeted at scientific and chemical applications.71,117 In
common with most programming languages, libraries for communication using a large
number of standard protocols are also available, enabling programmes written using Python
to communicate with laboratory instruments.
The goal of this project was to develop an automation platform that implemented the
communication protocols for standard devices, and which provided the ability to define
logic behind the commands that need to be sent, and to schedule them to occur at the correct
time.
38 New control and monitoring software for flow chemistry systems
2.1.2 Flow chemistry equipment
Basic flow chemistry equipment (see page 3) consists of pumps, mixers, switching valves,
heaters / coolers and sensors. Broadly speaking, any controllable reactor component will
have two modifiable parameters: power (on / off) and some target value (for example: tem-
perature or speed), in addition to readouts of current sensor values. Recent developments
in reactor technology such as electrochemical,118 photochemical119,120 and microwave121
reactors fall broadly within the same categories, but may of course require different pa-
rameters (such as voltage, wavelength or power instead of temperature). Sensors, such as
pressure sensors, pH sensors or thermocouples may not have any controllable parameters,
instead only providing readings.
As mentioned earlier, work has been done already towards defining standards for interfaces
for interaction with laboratory instruments: the SiLA specification103 defines interfaces for
syringe pumps, liquid pumps, valves, heating/cooling units, as well as generic devices with
analogue or digital parameters and outputs. These concepts could be adapted to create a
new interface which is initially aimed at comtrolling flow chemistry equipment but which
could later be made compatible with the SiLA (or other) standards (Table 2.2).
Table 2.2 Parameters for typical flow chemistry equipment.
Type Parametera Sensorb
HPLC Pump Target flow rate Actual flow rate
Syringe Pump Target volume Actual volume
Valve Position n/a
Heater / Cooler Target Actual value
Sensor n/a Sensor value
a Generally writeable. b Generally read-only parameters.
2.1.3 Interaction with equipment
There is a range of communication mechanisms for laboratory devices, some more common
than others. The simplest devices communicate with analogue voltage signals and electronic
switches (referred to as contact closures). To use these signals and switches we must
physically connect one wire from a device to the controller for each parameter of interest.
More recently developed devices use “combined” interfaces such as USB, Ethernet or
2.1. Introduction 39
start / stop
power
ow rate
pressure reading
pump head size
error messages
...
3V
% solvent B
0V
Figure 2.3 Analogue connectivity, such as with a hypothetical binary HPLC pump (left)
might include a switch to start or stop the pump, and a voltage output to read the solvent
composition. For each value to be read or controlled, (for example, the flow rate) an addi-
tional physical wire is required. Newer methods such as RS-232 allow many commands
and parameters to be conveyed using the same physical connection.
RS-232. These are digital interfaces that have larger speeds or bandwidths,* meaning that
only one wire is required for each device (Figure 2.3).
RS-232 is one of the more common interfaces for laboratory devices, and it works by
sending text-based messages along a serial cable. For example, to interrogate a valve
manufactured by Valco Instruments Inc.,122 a text command NP (Number of Ports) can be
sent, to which the valve responds with 10: the number of available ports. A large majority
of instruments support RS-232 communication, and there are many options for adaptors
which can transmit RS-232 messages within other protocols such as USB or Ethernet. In
these cases, the adaptor hardware is used along with software that creates a “virtual serial
port” on the control computer.
USB is the successor to RS-232, and has a much higher data rate (typically 4000 × faster).†
However, there is limitation in the length of the cables that can be used, and often USB
devices require their own software or drivers to perform binary communications, rather
than simply using text-based communication.
Ethernet is another transport mechanism, which uses a protocol called TCP/IP. It has a long
physical range, and is typically 1000 to 10 000 × faster than RS-232.‡ The advantage of
using Ethernet communication is that it is used extensively for computer networking, so
this form of communication can take advantage of existing building-wide cabling systems.
There are some other less commonly encountered protocols, including Modbus and I2C.
Modbus can operate over different transport mechanisms such as serial, USB or Ethernet;
instead of sending messages, it defines commands to interact with parameters (called
* How fast or how much data can be transmitted at the same time.
† 480 Mbps for USB 2.0, compared with a maximum of 115 kbps for typical RS-232 devices. To minimise
interference, few RS-232 devices operate over 19 kbps.
‡ Modern consumer ethernet equipment operates at 100 to 1000 Mbps.
40 New control and monitoring software for flow chemistry systems
registers) stored in the device. I2C is a protocol which operates across simple wires, and is
often used internally within instruments to control components such as switching valves.
It is also possible to use some of these protocols over a wireless transport layer such
as wireless Ethernet; this has the potential to reduce the amount of clutter in the fume
cabinet and laboratory. However, when using wireless communication, issues of connection
dropouts, interference and security must be considered, introducing significant unwanted
complexity to the system.
2.1.4 Control computers
Previous work in the laboratory has employed laptop computers outside the fumehood, or
desktop computers standing inside the fumehood. These take up space, can be damaged
easily by solvents or corrosive substances (requiring expensive replacements) and tend to
require a large number of wires to be routed around the fumehood. Of course, without
wireless connectivity many of the wires are necessary; nevertheless, the use of smaller and
lower-powered computers can reduce clutter and cost, save space and energy, and benefit
from increased portability. Additionally, the introduction of graphical interfaces that can
be accessed using a portable device such as a tablet computer, rather than being confined
to a screen adjacent to the equipment, can provide the chemist with greater flexibility and
freedom within the laboratory (Figure 2.4).123
instrument
instrument
instrument
portable
device
control
computer
Figure 2.4 The use of wired connections between the instruments and the computer
running the control programme ensures reliable communications throughout an experiment.
By allowing a user interface to be accessed on portable devices, the operator can work
with greater flexibility.
Fortunately, the control computer for these systems does not necessarily have to be a
standard laptop or desktop machine. Provided that software is written to suit them, there
are many compact systems now available (Figure 2.5). Platforms such as Arduino132 and
Microsoft .NET Gadgeteer133 contain a small, very low power, programmable microchip
2.1. Introduction 41
a b
c d
Figure 2.5 (a) Arduino Uno,128 (b) Microsoft .NET Gadgeteer (shown with attached
button module),129 (c) Raspberry Pi (model B),130 (d) Intel Galileo board.131
with an array of digital and analogue inputs and outputs, as well as plug-in modules for
interacting with motors, sensors, or other devices. Generally these have a small amount
of memory for storage, and can run a single looping programme. With their low power
consumption they are ideal for developing integrated controllers for new devices. A more
powerful device is likely to be needed for controlling multiple machines using easy-to-
write programs. Raspberry Pi28 is a small computer with an 800 MHz processor running a
Linux operating system, which can run software written in Python, as well as most other
Linux software. As ubiquitous computing becomes more commonplace, more miniaturised
computer systems are being made available, including the Intel Galileo134 and Edison135
systems, and the Phidgets Single Board Computer (SBC).136 The more flexible and powerful
a system is, generally the more expensive it will be both in terms of cost and energy usage:
small computers such as the Raspberry Pi provide the capability for performing advanced
applications whilst having only a small power requirement.
42 New control and monitoring software for flow chemistry systems
2.2 Implementation
As part of this project, a new software tool for computerised control of flow chemistry
platforms was developed, under the name Octopus.137 This has a similar structure to that
of the Clarity project mentioned earlier,98 but was designed specifically for use with flow
chemistry systems.
Each device is controlled by a virtual instrument, which is a software component that sends
commands and records all of the data produced by the device (Figure 2.6). The virtual
instruments are controlled by a scheduler, which executes a control protocol and instructs
the virtual instruments to perform the required actions at the correct times. The scheduler
and virtual instruments can run on a regular desktop computer, or a miniature computer
such as the Raspberry Pi.
Multiple protocols can be run independently or on different computers; they all commu-
nicate with a user interface server which delivers the experimental data in numerical or
graphical formats to client devices using a web site interface. These client devices can
be desktop computers or portable devices connected via a wireless network or over the
internet.
The remainder of this chapter discusses some of the key features of the design.
user
interface
protocol
scheduler alerts
virtual
instrument
data
storage
physical
instrument
virtual
instrument
data
storage
physical
instrument
additional
instruments
physical
instrument
client
client
client
Figure 2.6 The control software system is structured so that each instrument is controlled
by a virtual instrument within the software. Experimental protocols are run by a scheduler,
which can interface with any number of virtual instrument objects. The state of the
instruments and the experiment can be displayed on a web interface which can be
accessed by multiple users simultaneously; the user interface can be provided by the
same computer or a different computer to the one running the protocol. Events within the
protocol can trigger alerts such as emails or SMS messages to be sent to the operator.
2.2. Implementation 43
2.2.1 Communication
Concurrency and asynchronous operation
A fundamental requirement for the control software was for it to be able to communicate
with several different machines and devices simultaneously. Computers perform multi-
tasking by rapidly switching between running processes; one simple method for achieving
this in software is by using an event-based programming model. Under this system,
execution of different parts of the programme is triggered by external messages or events.
Commonly, new events wait in a queue and only one event is processed at once — ideally,
the part of the programme that processes an event should be very efficient so that new
events are dealt with as quickly as they arrive. The event-based model is considered
to be conceptually easier and more efficient to run than alternative methods for multi-
tasking,138 such as the creation of multiple threads (where the operating system manages
task-switching) or by programming for the utilisation of multiple processor cores. Since
event-driven programmes can be run on a single processor, and can have a relatively low
resource requirement, even quite complex programmes can be run on low-powered devices
such as the Raspberry Pi.
The capacity for a computer programme to perform multiple tasks simultaneously using the
same data is called concurrency.139 It turns out that concurrency represents quite a complex
problem and so it is important to consider how it will be achieved. In this case, concurrent
operation is beneficial because one of the slowest parts of the system is communication
with different devices. Typically, it can take around 200 000 times as long to access a file
from the hard drive (magnetic storage) as it does to read a value from memory (RAM;
solid-state storage). Even when using the fastest interfaces, communicating over a serial
cable or over a network is even slower than accessing data from the hard drive, by a
factor of up to a hundred.140 Because reading data stored in the RAM is so much faster, a
computer can carry out thousands of calculations or instructions in the time that it takes to
receive a response from an external device. A programme that waits for responses from
each communication before moving on to the next instruction will waste a lot of time
doing nothing, whereas a programme made to run in a concurrent manner can perform its
communications asynchronously§ and thus be more efficient by doing other things whilst
waiting for responses from remote devices.
§ Without having to wait for them to complete before moving on to the next instruction.
44 New control and monitoring software for flow chemistry systems
“OK”
“Start R1”
R1
“OK”
“Start R2”
R2
30 ms
30 ms “OK”
“Start R1”
R1
“OK”
“Start R2”
R2
30 ms
Synchronous communication Asynchronous communication
wait...
wait...
other instructions
message
response
Figure 2.7 Event-based concurrency enables asynchronous communication so that
other tasks can be done while waiting for responses from remote devices such as reactors
R1 and R2. (30 ms is a typical time for a message to be delivered to a remote device and
then for a response to be received).
For example, to start two devices simultaneously, messages must be sent along two different
cables with as little time separation as possible (Figure 2.7). Whilst it is possible to send
two simultaneous messages using a regular programme, one written with concurrency in
mind can achieve this and also listen for and handle the responses, essentially allowing a
two-way dialogue with each device. In an event-based system, messages are dispatched to
the remote devices asynchronously, and then the arrival of each response triggers an event
to which the programme responds, to process the incoming data.
Twisted141 is an open-source software package designed to enable the development of soft-
ware for running networking tasks using events, using the Python programming language.
This package was employed to implement the concurrent operation and communication
with flow chemistry instruments. Using the Twisted package as a basis, the virtual instru-
ments were developed. These define the commands that are to be sent to each type of
machine and how to interpret the responses.
Implementation of virtual instruments
Python is an object-oriented programming language, which means that it is built around the
concept of objects (software representations of physical things). Objects contain properties
(data) and define methods (actions that can be performed on the object or its data). With
this in mind, the interface to each connected machine was implemented as an object (a
virtual instrument), which has properties corresponding to the status, parameters, outputs
2.2. Implementation 45
and sensor readings from the physical instrument.¶
For safety reasons, it is important that the physical control panel on the machine is not
disabled (even though there is often the option to do so) so that if necessary it can be
controlled manually even while it is ostensibly under computer control. As a result, there
may be changes to the state of the physical machine which are not caused by commands
sent by the virtual instrument. Therefore, it needs to regularly request this information so
that its internal representation of the physical instrument is kept up-to-date. This enables
the control programme to make decisions based on the actual state of the machine.
To create the virtual instruments, a different class is defined for each type of machine that
is to be monitored or controlled. A class is a generator which can create an appropriate
object to correspond to a particular machine when given a connection through which to
communicate with it. The class is defined with all of the parameters appropriate for that
type of machine, and methods for requesting the current state of the physical instrument
and updating the object’s parameters accordingly (Figure 2.8).
Virtual Pump Object
Power on
Flow rate 100 µL/min
Pressure 7 bar
Figure 2.8 A pump object is defined with virtual representations of each property of
the pump; for example, pressure, flow rate and power. Some are read-only, such as the
pressure; others can be changed to control the device.
There is likely to be more than one set of commands being sent to each machine whilst
the programme is running (Figure 2.9). In the background, the virtual instrument will
be periodically updating its state to match the physical one; at the same time, there may
be commands defined by the control sequence for a particular experiment. The ability to
manage both foreground and background tasks in the same way is a key advantage of a
concurrent system.
¶ The virtual instruments perform a similar function to the R2R4 object which was used in Chapter 1 (see
Appendix A).
46 New control and monitoring software for flow chemistry systems
Co
nt
ro
l S
eq
ue
nc
e
data
commands
updates
Command
Command
virtual
Command
Command
virtual
physical
physical
Figure 2.9 The commands defined by the control sequence cause commands to be sent
to the physical instrument. At the same time the virtual instrument object requests updates
to keep its internal data up-to-date. The internal data is used for calculations and decision
making in the control sequence.
Command queues
Even though a concurrent programme can run multiple tasks in parallel, in general the
physical instruments can only process one command at once (this is true for almost all
of the machines encountered during this work). However, multiple commands intended
for one machine may be generated within a very short time, and so the programme has
to wait for each to have been sent and to have received a response before moving on to
the next one. Internally, each machine object contains a connection object to represent
the communications link with the physical device. The connection object keeps track of
M
ac
hi
ne
 O
bj
ec
t
Queue
Connection Object
response
send
Commands from
control sequence
command
handler
Parameter update 
commands
report 
result
command
update
parameters
command
update
parameters
command
physical
instrument
comm.
loop
Figure 2.10 The machine object contains a reference to its connection object, which
keeps a queue of waiting commands. When a command gets to the front of the queue,
the connection sends it to the machine and waits for a response. When the response has
been received an appropriate handler is invoked and then the next command is sent.
2.2. Implementation 47
the queue of commands waiting to be sent, and performs the appropriate action when a
response is received from each one (Figure 2.10).
If no response is received, this could indicate a temporary or permanent loss of the con-
nection to the machine. It is not possible to tell whether the message was received or not
so the best course of action is to send the message again. For this reason, most devices’
command interfaces are implemented in the form of a state to end up in, rather than an
action to perform. For example, a command should be switch valve to position 1 rather
than switch valve to the next position. The latter command would lead to different results
depending on how many times it is sent.
The continuous monitoring of the current state of each instrument gives the software
increased confidence that a machine is in the correct state at any point in the sequence.
Logging of all of this data throughout an experiment gives the scientist the ability to check
that the reaction had run correctly once it is complete.
2.2.2 Data handling
The data collected from the connected machines may be needed to observe the current
conditions or to measure trends over time. In order to look for trends, the data must be
stored for a period of time to allow calculations to be performed on data collected over a
particular time frame. As data are received by the virtual instruments, it is stored in two
places: in memory within the property associated with the relevant parameter or sensor,
and also to a log file on the hard disk.
Over a long experiment, the hard disk log files could then become extremely large, or even
fragmented if a maximum file size is enforced. This would make reading data from disk
a slow and relatively complex task, and so these log files are used only as an archive for
later analysis. All calculations involved in the control sequence are evaluated using the
in-memory data. Therefore, calculations can be assumed to be instantaneous with respect
to the time taken to communicate with an instrument.
However, the Python interpreter requires 16 bytes to store a typical numeric value, so a
machine with ten properties being updated ten times per second would use around 100 kB
per minute. To avoid a large memory requirement the amount of data stored in memory
must be limited.
48 New control and monitoring software for flow chemistry systems
D
at
a 
O
bj
ec
t
Low resolution
archive
High resolution
cache
Current 
value
Filter
New value
single value xed size
hard drive
archive
Figure 2.11 Illustration of data storage within a property object. As new data points are
added to the property object, it writes them all to a log on the hard drive, and retains
the most recent one for quick access to the current value. All of the data from the last
minute are stored for use by calculations that need high time resolution. Old data points
are removed from this storage as new ones arrive. A long-term archive of filtered data is
also kept so that the overall shape of the graph for any time period can be recalled.
For real-time graph presentation and for the purposes of most calculations, a time resolution
of even one second is probably far greater than necessary. If there are a few prominent
features in a particular data set, such as peaks or oscillations, then the precise detail of any
additional noise is generally less important. Therefore, it is possible to filter the data points
as they are stored to pick out just enough values to represent the overall shape of the data.
Each data object keeps a low-resolution archive of all data that has been collected (Fig-
ure 2.11). As new data points are added they are filtered to retain only values that represent
changes of more than 5 % of the current maximum spread. Key maxima and minima
 0
 5
 10
 15
 20
 20  25  30  35  40
Ab
sor
ba
nc
e
Time /min
Input
Filtered
Figure 2.12 A section of filtered data from the UV absorption during an HPLC run. The
red crosses indicate the key points that were picked out for storage. In this case, the
number of data points was reduced by a factor of 100.
2.2. Implementation 49
are also retained to ensure that a reasonably accurate representation of the data is stored
(Figure 2.12).
In summary, there is a balance between the requirements to respond to changes as soon
as they occur, for any calculations to be based on a full set of current data, and for the
programs to operate at as high a speed as possible (minimising any calculation or response
delays caused by latency between the computer and the instrument, or the software and the
hard drive). This is achieved by maintaining current and historic information about each
parameter in the computer’s RAM, so that calculations can be performed instantly without
waiting to receive new data from the instrument.
2.2.3 Command sequences
To run a sequence of commands, a list of steps to be carried out must be defined, similar to
the lines of Python code listed in Chapter 1. However, that code was synchronous: nothing
else could happen whilst the programme was waiting for one command to be completed.||
Instead, such a sequence can be implemented using the event-based system. Once each
command has finished executing,** it triggers an event that starts the next one. The time
taken for a step to complete and become finished depends on the action that it is carrying
out. Switching on a machine will only take as long as is takes the machine to respond,
whereas waiting for 20 hours will take (not surprisingly) 20 hours. Importantly, since
the whole system runs concurrently, multiple sequences can be executed simultaneously,
and sequences can have sections where certain parts run in parallel. Control of each
of these command sequences is performed as a separate task, alongside the background
communication between the virtual and physical instruments.
The sequence is constructed from individual instructions, implemented as Step objects,
which can be linked together. Each one can perform actions or encapsulate logic, or contain
child steps. The step objects keep track of their state, such as ready, running, finished or
error (Figure 2.13).
A control sequence then consists of a connected list of Step objects. By creating Step
implementations for various programming constructs (such as decision making and loops),
|| In Chapter 1, Python’s time.sleep() command was replaced with a function which checked the current
time to decide when to stop whilst monitoring the state of the machine.
** For instance, on receiving a response from the instrument.
50 New control and monitoring software for flow chemistry systems
St
ep
 O
bj
ec
t
State:
reset
ready running
paused
nished
error
Figure 2.13 Each step keeps track of its state; as soon as it becomes “finished” it triggers
an event that starts the following step.
arbitrarily complex sequences are made possible (Table 2.14). The fact that the system
is built using Python means that any of the large selection of libraries available for the
language — for chemical intelligence,142 statistical analysis143 and computer vision,144
among others — can be used for data processing or decision making.
Table 2.14 Some of the available step types for building sequences.
Command Description
set Set a variable or parameter on a machine.
wait Wait for a specified period of time.
wait_until Wait until a specified calculation evaluates to True.
loop_until Run a sequence repeatedly until a specified calculation
evaluates to True.
do_if Run one sequence if a specified calculation evaluates
to True, or another if it evaluates to False.
log Store a message in the log file.
call Execute a function written in Python.
Control scripts are written in Python, defining the machines to be used, the appropriate
connections and the sequence of operations to carry out. Once all of this is ready, the control
sequence is executed (Listing 2.15). The control sequence may make decisions based on
data collected from the machines — to do this, calculations (for example an expression
such as r.rate < 0) must be evaluated when their value is needed during execution, rather
than when they are defined.
2.2.4 Data manipulation
Ensuring that expressions are evaluated when they are required, rather than when they are
2.2. Implementation 51
Listing 2.15 A simple automated protocol using a syringe pump and a temperature
sensor, controlling an exothermic reaction. Each instrument is set up with an appropriate
serial connection, and then the logic is defined. Here the pump will inject reagent at
50 µl/min until the thermocouple registers a temperature rise over 42 ◦C. The pump is
then stopped until the temperature drops to below 38 ◦C. The loop continues until the
experiment is stopped using the user interface. For full documentation, refer to Reference
137.
# Import the necessary libraries
from octopus.runtime import *
from octopus.manufacturer import wpi, phidgets
from octopus.transport.basic import serial
from octopus.transport.phidgets import Phidget
# Set up the machines that will be used
pump = wpi.Aladdin(serial("/dev/ttyS0", baudrate = 9600))
temp_sensor = phidgets.TemperatureSensor(Phidget(12345), inputs = [{
"index": 0,
"type": phidgets.ThermocoupleType.K
}])
# Create a reference to the relevant sensor
temp = temp_sensor.thermocouples[0].temperature
# Define a simple sequence: loop until 1 mL has been added
main = loop_while(pump.dispensed < 1000, sequence(
wait_until(temp < 38),
set(pump.rate, 50), # Only pump while the temperature < 42 C
wait_until(temp > 42),
set(pump.rate, 0)
))
# Start the experiment
run(main)
defined, is an important consideration when making an asynchronous programme using an
imperative language such as Python.145 In Python, the actions of operators like +, * and <
can be redefined so that rather than returning the current value, they return an Expression
object which has the ability to compute the value when required. The expression object
acts like a “magic” variable that keeps itself up-to-date based on the components of the
contained expression (Listing 2.16).
Similarly to the properties of a virtual instrument (Section 2.2.1), an expression needs to
be able to provide trends of its value over a particular time period (because it needs to
behave exactly like a variable). However, there is no need for this data to be stored within
the expression object because it can easily be constructed from the component variables.
52 New control and monitoring software for flow chemistry systems
Listing 2.16 After defining an expression, for example temp < 38, the result can be
interrogated at any point, and will depend on the current value of the temp variable.
# Define an expression
>>> expr = (temp < 38)
# The value of the expression depends on the value of the variable
>>> temp
<Stream at 0x2e460d0: temp_sensor.thermocouples[0].temperature (float) = 20.0C>
>>> expr
<LtExpression at 0x2e32ff0: LtExpression_1 (bool) = True>
# Later on...
>>> temp
<Stream at 0x2e460d0: temp_sensor.thermocouples[0].temperature (float) = 41.7C>
>>> expr
<LtExpression at 0x2e32ff0: expr (bool) = False>
>>> expr.value
False
Whenever a trend is accessed on an expression, the trends from the component variables
over the same time period are obtained and the expression is evaluated at each time point to
return the result. This is achieved using the NumPy library146 which can perform efficient
arithmetic over arrays of data. For this to be achieved, the data of each trend has to be
interpolated so that there are data points at regular time intervals. The NumPy library has a
method that can map a data series between different x-values (reversing the data-filtering
process in order to run a calculation: Figure 2.17).
 0
 1
 2
 3
 4
 5
 0  2  4  6  8  10  12  14
Va
lue
Time
A
B
A + B
Figure 2.17 Arithmetic with interpolated values. A and B are data series (large crosses
indicate stored values). To add them, they are interpolated at regular intervals (small dots)
and the result is calculated at each point.
2.2. Implementation 53
2.2.5 Pausing experiments
Whilst running flow chemistry experiments, it can become necessary to suspend a running
sequence in order to make manual interventions, for example in order to fix a leak or clear
a blockage. Therefore, a key requirement is the ability to pause a sequence and then resume
it without having to start again from the beginning.
When instructed to pause, each step propagates this command to any child steps, so that
the whole sequence stops running. Crucially, sequences do not run any new steps until a
resume command has been issued, and steps that involve a waiting time pause their timers
so that the full remaining time must elapse after it is resumed again. The machines that are
also involved in an experiment are also sent a pause command, which generally means that
pumps and heaters are switched off, but data are still collected from the sensors.
2.2.6 Error handling
There are a number of errors that may occur during a running sequence:
• Errors in the communication with a machine.
• Loss of connection to a machine.
• Physical machine faults.
• Physical problems with the apparatus.
• Logic errors within a command protocol.
In some cases automatic recovery is possible; for instance, virtual instrument drivers
can include logic to repeat or discard corrupted messages. For more complex problems,
handling logic can be incorporated within the control programme, including recovery
attempts, notifications (for example via an on-screen message or an SMS notification), or
as a last resort, flushing the reactor with solvent and halting the experiment.
2.2.7 Attached control systems
A significant simplification in the implementation of complex control algorithms was
achieved by adding the ability to connect control loops to individual steps in a sequence.
54 New control and monitoring software for flow chemistry systems
For example, when attached to a sequence or wait command, a control loop starts and stops
(and pauses and resumes) with this step.
The control loops created initially were Bind, StateMonitor and PID. Bind updates a
variable based on the value of another, optionally transformed by a function, whenever
the master variable changes. StateMonitor executes one sequence and then another as
a specified expression transitions from evaluating from True to False and back. This
functionality is particularly useful for error handling. PID is a form of feedback controller
which adjusts one variable based on changes in another (see Chapter 4).147
The example in Listing 2.15 can alternatively be written using these control loops (List-
ing 2.18). The two aspects of the logic have been separated: the StateMonitor controls
the temperature-dependent reagent addition, while the wait_until keeps track of the
volume that has been added and terminates the StateMonitor at the appropriate time.
2.3 Conclusion
An extensible software package was developed to control diverse elements of laboratory
flow chemistry apparatus. The use of open-source tools should lead to sharing of control
programmes and interfaces for new machines. Furthermore, the relative simplicity of the
Python programming language may enable more researchers to create their own automated
systems with a minimum amount of training.
Whilst this software would not be appropriate for an engineering environment, where
commercial software packages with support contracts are favoured for their guaranteed
reliability, it is ideal for the research laboratory. Here, it is more important for a software
tool to be low cost and have good flexibility than to offer a robust monitoring capability.
A key aspect of this software is that the machine interfaces are abstracted, so that the
researcher does not have to consider any aspects of communication with the instruments,
and algorithms for common operations can be copied between protocols using similar
devices. Common tasks such as remote data observation and complete logging are done
automatically. By enabling researchers to spend less time on laborious and repetitive tasks,
more time can be spent focusing on the chemistry and engineering aspects of a project.
Importantly, control programmes can be easily adapted to employ similar equipment that
may communicate over a different protocol.
2.3. Conclusion 55
Listing 2.18 The same protocol using an attached control system. This is more powerful
than the example in Listing 2.15, because the sequence will terminate immediately once
1 mL has been dispensed, rather than having to wait for an iteration of the loop’s child
sequence to complete before the expression is evaluated again. Note: there are ways of
accomplishing this without the StateMonitor.
# Import the necessary libraries
from octopus.runtime import *
from octopus.manufacturer import wpi, phidgets
from octopus.transport.basic import serial
from octopus.transport.phidgets import Phidget
from octopus.sequence.control import StateMonitor
# Set up the machines that will be used
pump = wpi.Aladdin(serial("/dev/ttyS0", baudrate = 9600))
temp_sensor = phidgets.TemperatureSensor(Phidget(12345), inputs = [{
"index": 0,
"type": phidgets.ThermocoupleType.K
}])
# Create a reference to the relevant sensor
temp = temp_sensor.thermocouples[0].temperature
# Run until 1 mL has been added
wait_for_addition = wait_until(pump.dispensed > 1000)
# Add the control loop
wait_for_addition.dependents.add(StateMonitor(
tests = [temp < 42],
trigger_step = set(pump.rate, 0),
reset_step = sequence(
wait_until(temp < 38),
set(pump.rate, 50)
)
))
# Start the experiment
run(sequence(
wait_for_addition,
set(pump.rate, 0)
))
A key target for future development involves the creation of a graphical interface for
designing control sequences. A drag and drop interface would open up the technology to
non-programmers with a minimal learning curve.
56
Chapter 3
Automated two-step reactions with
heterogeneous metal catalysis
57
58 Automated two-step reactions with heterogeneous metal catalysis
3.1 Introduction*
There are two important ways in which software for the rapid prototyping of simple control
systems can augment flow chemistry projects. Traditionally, an experimental protocol
involving multiple devices might require several people to supervise it to ensure that the
required actions, such as starting the different reactors and switching the appropriate valves,
occur at the correct times. In contrast, a computerised control system can provide precise
control of the experimental timing and parameters, and can be designed to handle incoming
information and respond to undesired events in a predictable and consistent manner. (Also,
a computer can perform its task without having to contend with the external distractions
commonly present in a busy laboratory!)
The drawback is that an automated system is only as intelligent as its control programme
and therefore lacks the ability of a skilled scientist to respond to unforeseen occurrences.
Therefore, whilst it is in the scientist’s interest not to have to continually stay next to the
equipment, it is important that they have the ability to observe the state of the system from
wherever they are and to receive alerts or hear alarms if there is a problem that cannot be
rectified automatically.
In this chapter an application of the software control tools described in Chapter 2 is explored,
within the context of the development of hydrolysis techniques using a hydrous zirconia
catalyst.
3.2 Synthesis of primary amides by nitrile hydration
Primary amide moieties are present in a number of biologically-active compounds and
their building blocks. Amide formation is a very common transformation within phar-
maceutical synthesis,149,150 but the standard protocols for achieving this are hampered
by the widespread use of reagents that offer poor atom efficiency.151 One of the most
atom-economical and environmentally-friendly methods for the preparation of a primary
amide is by the hydration of a nitrile.152 Unfortunately, traditional methods can have low
functional group tolerance due to the strong acids or harsh conditions employed. Under
base-catalysed conditions over-hydrolysis is commonly observed since the formation of a
carboxylic acid is kinetically favourable.
* The work described in this chapter is reported in Beilstein J. Org. Chem. (Reference 148).
3.2. Synthesis of primary amides by nitrile hydration 59
Just as with solid-supported reagents (see Chapter 1), heterogeneous catalysis offers a
number of benefits, particularly in flow systems where the catalyst can be contained within
a cartridge.153 Purification is simplified because the catalyst will not contaminate the flow
stream, and it may be possible to use a smaller amount of the catalyst because it can be
recycled more easily. In cases where the activity of the catalyst can reduce during an
experiment, the conversion can be monitored and the cartridge replaced at an appropriate
time.
A number of homogeneous metal-catalysed systems for nitrile hydration have been re-
ported,152 and recent work within our own group has identified manganese dioxide as a
mild heterogeneous catalyst for this transformation.154 Following investigations into the
use of zirconium dioxide for Meerwein-Ponndorf-Verley reductions155 and Oppenauer
oxidations,156 we were interested to find that zirconia was also an effective catalyst for the
hydrolysis of heteroaromatic nitriles.
We had identified piperazine-2-carboxamide (39) as an important 3D building block157
and at the time of writing it was costly to purchase even a racemic mixture of this com-
pound.158 We envisaged that compound 39 could be obtained by a reduction of pyrazine-2-
carboxamide (40), which could itself be prepared by a catalytic hydration of pyrazine-2-
carbonitrile (41) (Scheme 3.1).
NH2
H
N
N
H
O
NH2
N
N
O
N
N
N
MOx M/C
H2O H2
41 40 39
Scheme 3.1 Two step approach to piperazine-2-carboxamide via metal oxide catalysed
hydrolysis followed by metal catalysed reduction.
60 Automated two-step reactions with heterogeneous metal catalysis
3.2.1 Hydration under flow conditions
Optimal reaction parameters for the hydration of nitrile 41 (Scheme 3.2) had previously been
determined by Dr Claudio Battilocchio. An in-line infrared spectrometer (Mettler-Toledo
FlowIR) was used to measure the conversion of the nitrile to the amide.
hydrous
zirconia
100 ºC
100 psi
R1
N
N
NH2
O
N
N CN
8:1 EtOH/H2O
0.1 mL/min FlowIR41 39
Scheme 3.2 Optimum reaction parameters for the zirconia hydration. Compound 42
is infused into a heated glass column containing zirconium hydroxide. An in-line IR
spectrometer is used to measure the degree of conversion.
A sensor with a silicon window (SiComp) was employed so that the nitrile stretching
frequency could be observed.† In a normal reaction, only absorptions from the product are
observed; characteristic signals are seen at 1685 cm−1 corresponding to the amide C−O
 0
 0.2
 0.4
 0.6
 0.8
 1
 1.2
 1.4
 1.6
 0  0.5  1  1.5  2  2.5  3  3.5  4
Ab
sor
ba
nc
e /
AU
Time /h
Peak at 1051 cm-1
Peak at 1685 cm-1
Figure 3.3 Infrared absorbance profile of the reactor output during a 3 hour process,
indicating the slug of material passing through the reactor column. The response reaches
a maximum after approximately one column volume (approx. 1 mL) with a small degree
of dispersion. In this case, absolute ethanol without additional water was used as the
solvent so the reaction stopped once the water intrinsically contained within the hydrous
zirconia was exhausted. This fall-off also occurred over approximately one column volume,
indicating a low retention of material by the heterogeneous bed.
† Frequencies in the range 1950 – 2250 cm−1 are normally masked by C−C stretching within the standard
diamond-windowed (DiComp) sensor.
3.3. Reaction control and monitoring 61
bond stretching frequency, and at 1051 cm−1 in the fingerprint region. These could be used
to observe the reacted materials exiting the zirconia column (Figure 3.3).
We envisaged that a protocol incorporating automated monitoring and control using data
from the IR spectrometer, could assist with the processing material over a long period of
time.
3.3 Reaction control and monitoring
In a manufacturing plant, the concept of process analytical technologies (PAT) refers to the
observation of key parameters by sensors and analytical devices installed into the reactor
equipment.159 In some cases a continuous process will have a greater number of more easily
observable parameters than a comparable batch process. In addition to chemical analyses
of the products there are numerous other properties that can be monitored to characterise
the system. A continuous process can be defined by parameters such as the flow rate of
the inputs, the temperature of the reactors, and the pressure of the system. Electronic
measurements can also be useful; for example, in some continuous processing plants the
voltage across a pump can be used as a proxy for the pressure and flow rate. An increased
voltage requirement might imply a pressure spike, or a lower voltage a pump failure.
Generally, if the reactor conditions stay within acceptable limits (for example, if the flow
rates and reactor temperature are consistent, and there are no sudden pressure changes)
then the output can be assumed to be of the desired quality. Laboratory processes could
potentially benefit from similar systems.
The devices depicted in Scheme 3.2 were connected to a computer (Figure 3.4a) and a
selection of parameters was monitored using a simple control script that performed two
functions. At the start of a run the column heater was powered on, and the reagent injected
as soon as the temperature was stable. During the run, the output valve was controlled
based on the PAT measurements reported by the Vapourtec R2+/R4 and the FlowIR.
The infra-red absorbance represented the conversion at the exit point of the reactor, so
readings outside the defined acceptable range were set to trigger immediate changes to the
valve. Problems in the pressure (indicating flow rate changes) or the temperature could
affect the conversion for one whole column volume and so this was set to stop collection
for a longer time (Figure 3.4c).
62 Automated two-step reactions with heterogeneous metal catalysis
EtOH / H2O hydrous
zirconia
100 oC
20 min
waste
FlowIR
N
N
NH2
O
N
N CN
ow rate, 
pressure
temperature
absorbance
collect / waste
Python
Raspberry Pi
100 psi
R2
V1
a b
41
40
c
Start
Wait for temperature stability
Switch to reagent
Switch reactor on
Set temperatures
Set ow rates
Wait for absorption to be
above a trigger value
Wait for absorption to be
below a trigger value
Switch to collect
Switch to waste
End
Wait until reagent is used up
(dened time)
Switch to solvent
Wash for two column volumes
Switch o reactor
If there are severe uctuations
in temperature or pressure
Switch to waste
Wait for parameters to 
become stable
Wait one column volume
Switch to collect
Figure 3.4 (a) Reactor and control setup for the continuous process. (b) Photograph of
the apparatus as used. (c) Control sequence: the solid lines connect the commands that
are to be executed in sequence.
3.4. Automation of multiple experiments 63
 0
 20
 40
 60
 80
 100
 120
 140
 160
 180
 200
 0  0.5  1  1.5  2  2.5  3  3.5  4
-2
 0
 2
 4
 6
 8
 10
 12
°C,
 µA
U
ba
r
Time /h
Absorbance /µAU
Heater Temp. /°C
System Pressure /bar
Collecting
Figure 3.5 Selected data collected over a three-hour reaction simulating a long process.
The infra-red absorbance shown is that at 1685 cm−1, which is indicative of the amide
bond in compound 40. The thick bar represents the time during which valve V1 was set to
direct the output to the collection vessel. The pressure drop at 1.5 h caused the control
system to direct the output to waste until one column volume (a pre-configured waiting
time) after the pump had been manually re-primed.
To test this behaviour, an air bubble was introduced into the reagent stream. This caused
the pump to stop, creating a loss of pressure (Figure 3.5, at 1.5 h). The pressure drop was
detected, triggering an interruption in the collection until one column volume after the
problem had been rectified by manually priming the pump. Finally, after injection of the
reagent had ended, collection stopped automatically by the same mechanism as soon as the
output signal fell below a threshold .
3.4 Automation of multiple experiments
With pyrazine-2-carboxamide (40) in hand, initial testing showed that reduction of this
compound to pyrazine-2-carboxamide (39) could be performed in an H-Cube hydrogenation
reactor.
When working with a new reaction it is common to investigate the effects of the different
variables. Interest is building in the concept of automatic optimisation.104,105,111,113,160
Parameters can be varied gradually using different mathematical methods such as simplex
64 Automated two-step reactions with heterogeneous metal catalysis
algorithms161,162 in order to approach a maximum or minimum of some metric (for example:
conversion, yield, maximum yield with lowest temperature and time, and so on). Initially,
we had hoped to apply such a methodology to these reactions, but unfortunately this turned
out to be impractical for two reasons. Firstly, each iteration was very slow (30 to 60 minutes)
due to the long stabilisation time required for the H-Cube to reach steady state, the slow
reaction rate and the large dead volume within the heterogeneous column. Secondly, many
of the available parameters were limited to a set of discrete values (for example, the column
temperature could only be set in units of 10 ◦C). In this situation, simple linear optimisation
methods are difficult to apply,163 and more advanced machine learning techniques would
have been required.
Instead, we decided to employ a Design of Experiments method164–166 to investigate
which of the available parameters were more important for determining the conversion and
selectivity of this reaction. A two-level factorial design with three parameters (temperature:
40 ◦C and 100 ◦C; H2 pressure: 20 bar and full hydrogen mode; and flow rate: 0.1 mL/min
and 0.2 mL/min) suggested 16 experiments: two repeats each of eight sets of conditions.
waste
samples
FlowIRH-Cube
100 psi 75 psi
V2
P2
N
N
NH2
O
EtOH / H2O
sa
m
pl
e 
/ w
as
te
o
w
 ra
te
te
m
pe
ra
tu
re
,
pr
es
su
re
s
ab
so
rb
an
ce
a
40
Start
Release hydrogen
For each condition...
Start pump at 0.2 mL/min
Set valve to waste
Set column temperature
Set hydrogen mode
Switch on hydrogen if o
Wait for temperature stability 
if temperature changed
Wait one column volume 
if ow rate changed
Set ow rate
Set valve to correct output
Set valve to waste
End
Wait one column volume
Wait 10 min (collection)
b
Figure 3.6 (a) Reactor setup for the optimisation process. A ten-position valve was
included to enable collection of nine different samples. (b) The control sequence ran over
an experimental subroutine for each of a pre-programmed list of conditions.
3.4. Automation of multiple experiments 65
This would have been laborious to carry out manually, but multiple similar experiments can
be performed easily using an automated protocol.
The control of a Knauer HPLC pump, the H-Cube reactor and a ten-port switching valve
(V2) (Figure 3.6) was combined into a single programme, allowing up to nine reactions to
be performed in sequence. In each iteration, the parameters of the reactors were adjusted
to the desired set of conditions, and then a sample was taken once steady state was
reached. Two major reduction products were observed: 1,4,5,6-tetrahydropyrazine-2-
carboxamide (43; Scheme 3.7) and fully-reduced pyrazine-2-carboxamide (39). The
carbonyl stretching frequencies of these two amides were not individually resolvable
using the FlowIR (this instrument reported absorption maxima at 1651 and 1695 cm−1
respectively). The secondary frequencies had only low intensities, so unfortunately infrared
spectroscopy was not adequate for analysis of the reaction mixture. Therefore, the samples
were subsequently analysed by NMR to quantify the results.
NH2
H
N
N
H
O
NH2
N
N
O
Pd/C
H2 NH2
H
N
N
H
O
Pd/C
H2
40 43 39
Scheme 3.7 Reduction of the pyrazine ring is likely to occur in at least two steps.
Intermediate 43 was also observed in the reaction output under some hydrogenation
conditions.
To quantify each sample, four parameters were calculated (Table 3.8): the degree of
conversion, based on the amount of compound 40 remaining; the amount of the desired
product 39 formed; and an estimated amount of undesired compounds, based on the
integration of prominent by-product peaks visible in the NMR spectra (a doublet at 3.8 ppm
and a triplet at 3.2 ppm, the first corresponding to compound 43 and the other to another,
unidentified compound) relative to the integration of compounds 39 and 40. As expected,
the hydrogen pressure had the most significant effect on the conversion. The temperature
and the flow rate had smaller effects, although the combination of higher temperatures and
lower flow rates gave a higher purity output because intermediates such as compound 43
are fully reduced. For maximum efficiency further hydrogenation procedures were carried
out at the lower flow rate of 0.1 mL/min and the higher temperature of 100 ◦C.
With both steps running smoothly, we turned our attention to the development of an
integrated synthetic process.
66 Automated two-step reactions with heterogeneous metal catalysis
Table
3.8
R
esults
for
16
experim
ents
over
the
eight
sets
of
conditions
specified
by
the
design
of
experim
ents
protocol.
H
2
pressure
had
the
m
ostsignificanteffecton
conversion
and
purity.
R
un
H
2
Pressure
/bar
Tem
perature
/ ◦C
Flow
rate
/m
L
/m
in
C
onversion
a
Product a
Side-product1
a
Side-product2
a
1
Full
100
0.1
1.00
1.00
0.00
0.00
2
20
bar
100
0.1
0.53
0.35
0.08
0.10
3
20
bar
40
0.2
0.85
0.08
0.07
0.70
4
Full
40
0.1
0.78
0.74
0.04
0.00
5
20
bar
40
0.2
0.52
0.34
0.18
0.00
6
20
bar
40
0.1
0.38
0.26
0.12
0.00
7
20
bar
100
0.2
0.37
0.34
0.03
0.00
8
Full
40
0.2
0.80
0.54
0.26
0.00
9
20
bar
40
0.1
0.46
0.42
0.04
0.00
10
Full
100
0.1
1.00
1.00
0.00
0.00
11
Full
40
0.1
0.73
0.59
0.14
0.00
12
Full
100
0.2
1.00
1.00
0.00
0.00
13
Full
100
0.2
1.00
1.00
0.00
0.00
14
20
bar
100
0.1
0.40
0.37
0.04
0.00
15
Full
40
0.2
0.75
0.52
0.23
0.00
16
20
bar
100
0.2
0.44
0.44
0.00
0.00
a
V
alues
w
ere
calculated
by
1H
N
M
R
spectroscopy
(400
M
H
z,M
eO
H
-d
4 ),and
are
based
on
the
relative
integration
ofthe
peaks
atδ
9.24
ppm
(starting
m
aterial),3.13
ppm
(product),3.79
ppm
(side-product1)and
3.18
ppm
(side-product2).
3.5. Integration of two steps 67
3.5 Integration of two steps
3.5.1 Intermediate reservoir
When running a batch of experiments such as those performed in the previous section,
the problem of resource management is commonly encountered. The amount of starting
material required must be calculated based on the anticipated number of experiments. This
is an even greater challenge when performing an automated optimisation procedure, as the
number of experiments is unlikely to be known in advance. If the starting material has
to be synthesised then production of a large excess of material may lead to wasted time
and materials. It may be more efficient to make the intermediate in situ and then use it
immediately for the automated step under each set of conditions.
However, this would normally require that the first and second steps are compatible, for
example by having the same flow rate. A make-up pump (Figure 3.9) could be introduced
to vary the flow rate in the second step — but this would limit the minimum flow rate for
the second step, as well as altering the concentration of the reagents.
Step 1
Step 2
make-up pump
waste
Figure 3.9 A second pump can only add material, which affects the concentration. The
lowest possible flow rate for step 2 is determined by the flow rate of step 1.
We envisaged that a simple camera-based liquid-level sensor could be used to create a
reservoir that would hold a small quantity of the intermediate solution. The start of the
following reaction could then be delayed until sufficient material had been collected in the
reservoir.
Following previous work using visual observation methods to measure a phase boundary
for continuous separation92,167 we anticipated that a similar method could be applied to
measure the level of any solvent within a reservoir. By adding an air-filled float made
from a standard disposable syringe168 the liquid level could be measured using a camera
(Figure 3.10). The camera was connected to a computer by USB and integrated into the
control system using SimpleCV,144 an open-source computer vision library for the Python
programming language.
68 Automated two-step reactions with heterogeneous metal catalysis
input output
webcam
Python
SimpleCV
a b
Figure 3.10 (a) A laboratory flask was converted to a reservoir by adding a coloured
float and an inlet tube for the pump. A camera observed the float to measure the liquid
level within the flask. (b) Photograph of the apparatus. A sheet of white plastic created
contrast between the float and the background, and ensured that no other green objects
that could disrupt the reading appeared in the field of view.
The programme collected an image from the camera at regular (1 s) intervals and the images
were interpreted by a computer-vision algorithm (Figure 3.11). By subtracting the red
channel from the green one, the float is highlighted and its position reported. The distance
from the bottom of the image represented the volume of liquid in the flask, which was used
as part of the control algorithm (see Section 3.5.2).
This apparatus has a number of advantages over traditional liquid-level detectors such as
ultrasonic, pneumatic or capacitance sensors, optical or air-bubbler detectors, or floats with
mechanically or magnetically actuated switches. Since the only part in contact with the
liquid is the float, which can be exchanged for different materials, there is less concern
about chemical compatibility because none of the electronic components are in contact
with the reactive solution. It is also a more universal method because the solvent does not
need to be conductive and standard laboratory equipment (flasks) can be easily retrofitted
with a float, rather than requiring the design of a bespoke vessel.
Furthermore, the introduction of a reservoir disconnects one step from the next, allowing
the reactors for different phases to be started and stopped independently. It also enables the
introduction of workups and solvent-switches between steps (see Chapter 4), improving the
flexibility of optimisation procedures.
3.5. Integration of two steps 69
Height 163
Area 93
Height 166
Area 584
G - R G - B
ih
g
fe
dcb
a
Figure 3.11 Computer-vision image processing. The image (a) is separated into red (b),
green (c) and blue (d) components, or channels. There is a much lower intensity of red
in the region of the float and this shows up as a dark area. Subtracting the red channel
from the green (e) highlights this area whilst filtering out the background (grey and white
areas have similar amounts of each colour). In contrast, the result of green – blue (f)
does not show any significant features. A threshold filter (g) picks out the most significant
areas, and then an erosion filter (h) removes smaller artefacts. The remaining blobs are
recognised using the computer vision tool and the vertical position of the centre of the
largest (by area) is measured as the height of the float. Note: the lower five images are
shown with inverted colours for clarity.
70 Automated two-step reactions with heterogeneous metal catalysis
3.5.2 Combining synthesis and DoE
Introduction of the reservoir and the liquid-level sensing logic into the original control
programme allowed the two steps to be combined (Figure 3.12, Figure 3.13). The maximum
solution required for a single hydrogenation experiment was approximately 6 mL, for which
the float height was measured in advance. An experiment was performed only when the
liquid level in the reservoir was above this threshold.
EtOH / H2O
0.2 mL/min hydrous
zirconia
100 oC
20 min
R2
V1
V2P2
P1
waste
samples
FlowIR
100 psi
reservoir
100 psi N
N
NH2
O
N
N CN
webcam
ow rate, 
pressure
temperature
absorbance
collect / waste
co
lle
ct
 / 
w
as
te
o
w
 ra
te
oat height
te
m
pe
ra
tu
re
,
pr
es
su
re
s
41
40
Figure 3.12 Reactor setup for the two-step DoE process. The intermediate was collected
in the reservoir, and once the liquid level had risen over a predetermined threshold a
hydrogenation experiment was performed.
Additional monitoring logic was required so that valve V1 would stop directing material
to the reservoir if the liquid level rose too high. This would also trigger an SMS notifica-
tion, since it would indicate a problem with pump P2 that was likely to require manual
intervention.
3.5.3 Extended run
This control programme was extended to perform a long synthetic run using the same
reactor configuration (Figure 3.12) by removing the multiple experiment logic and fixing the
3.5. Integration of two steps 71
el
se
St
ar
t
Po
w
er
 o
 
H
-C
ub
e 
pu
m
p
Po
w
er
 o
n 
H
-C
ub
e 
pu
m
p
Se
t 
ow
 ra
te
 to
 0
.2
 m
L/
m
in
Se
t h
ea
te
r t
o 
10
0 
°C
W
ai
t u
nt
il 
he
at
er
 >
 9
5 
°C
Se
t p
re
ss
ur
e 
lim
it 
to
 2
5 
ba
r
Sw
itc
h 
R2
 in
pu
t t
o 
re
ag
en
t
Sw
itc
h 
R2
 in
pu
t t
o 
so
lv
en
t
Sw
itc
h 
R2
 o
ut
pu
t t
o 
co
lle
ct
W
ai
t u
nt
il 
th
e 
liq
ui
d 
le
ve
l i
n 
th
e
re
se
rv
oi
r h
as
 ri
se
n 
ag
ai
n
Po
w
er
 o
 
H
-C
ub
e 
(k
ee
p 
hy
dr
og
en
)
W
ai
t f
or
 4
0 
m
in
(z
irc
on
ia
 c
ol
um
n 
an
d 
co
nn
ec
tin
g 
tu
be
 d
ea
d 
vo
lu
m
e)
A
le
rt
 c
he
m
is
t i
f a
ny
 o
f t
he
pu
m
ps
 lo
se
 p
re
ss
ur
e
W
ai
t u
nt
il 
th
e 
liq
ui
d 
le
ve
l h
as
be
en
 re
du
ce
d
W
ai
t u
nt
il 
IR
 s
ho
w
s 
co
nv
er
si
on
to
 in
te
rm
ed
ia
te
W
ai
t u
nt
il 
th
e 
re
se
rv
oi
r c
on
ta
in
s
en
ou
gh
 m
at
er
ia
l t
o 
be
gi
n
If 
IR
 re
sp
on
se
 o
f i
nt
er
m
ed
ia
te
fa
lls
 to
o 
lo
w
Sw
itc
h 
to
 w
as
te
 u
nt
il 
th
e 
IR
re
sp
on
se
 re
tu
rn
s 
to
 n
or
m
al
If 
co
lu
m
n 
te
m
pe
ra
tu
re
 fa
lls
be
lo
w
 9
5 
°C
 o
r p
um
p 
pr
es
su
re
u
ct
ua
te
s
Sw
itc
h 
to
 w
as
te
 u
nt
il 
on
e 
co
lu
m
n 
vo
lu
m
e 
af
te
r t
he
 s
ys
te
m
 
ha
s 
re
tu
rn
ed
 to
 n
or
m
al
W
ai
t f
or
 re
ag
en
t t
o 
be
 u
se
d 
up
(p
re
co
n
gu
re
d 
vo
lu
m
e)
or
 fo
r l
oo
p 
to
 
ni
sh
If 
th
er
e 
is
 s
u
ci
en
t m
at
er
ia
l i
n
th
e 
re
se
rv
oi
r t
o 
pe
rf
or
m
 o
ne
ru
n 
at
 th
e 
re
qu
ire
d 
o
w
 ra
te
Sw
itc
h 
va
lv
e 
to
 w
as
te
Sw
itc
h 
va
lv
e 
to
 w
as
te
If 
th
e 
liq
ui
d 
le
ve
l i
n 
th
e 
re
se
rv
oi
r
ris
es
 a
bo
ve
 a
 s
et
 th
re
sh
ol
d
En
d
Po
w
er
 o
 
R2
Sw
itc
h 
R2
 o
ut
pu
t t
o 
w
as
te
W
ai
t u
nt
il 
R2
 s
w
itc
he
s 
ba
ck
 
to
 s
ol
ve
nt
W
ai
t u
nt
il 
th
e 
st
or
ed
 h
yd
ro
ge
n 
ha
s 
be
en
 re
le
as
ed
Re
le
as
e 
hy
dr
og
en
Fo
r e
ac
h 
co
nd
iti
on
...
St
ar
t p
um
p 
at
 0
.2
 m
L/
m
in
Se
t v
al
ve
 to
 w
as
te
Se
t c
ol
um
n 
te
m
pe
ra
tu
re
Se
t h
yd
ro
ge
n 
m
od
e
Sw
itc
h 
on
 h
yd
ro
ge
n 
if 
o
W
ai
t f
or
 te
m
pe
ra
tu
re
 s
ta
bi
lit
y 
if 
te
m
pe
ra
tu
re
 c
ha
ng
ed
W
ai
t o
ne
 c
ol
um
n 
vo
lu
m
e 
if 
o
w
 ra
te
 c
ha
ng
ed
Se
t 
ow
 ra
te
Se
t v
al
ve
 to
 c
or
re
ct
 o
ut
pu
t
Se
t v
al
ve
 to
 w
as
te
W
ai
t o
ne
 c
ol
um
n 
vo
lu
m
e
W
ai
t 1
0 
m
in
 (c
ol
le
ct
io
n)
Fi
gu
re
3.
13
In
te
gr
at
io
n
of
co
nt
ro
ll
og
ic
fo
r
m
ul
tip
le
hy
dr
og
en
at
io
n
ex
pe
rim
en
ts
un
de
r
di
ffe
re
nt
co
nd
iti
on
s,
w
ith
ca
m
er
a-
co
nt
ro
lle
d
m
an
ag
em
en
to
ft
he
in
te
rm
ed
ia
te
pr
od
uc
ed
by
th
e
hy
dr
at
io
n
st
ep
.
Th
e
co
nt
ro
ls
ys
te
m
s
in
th
e
bo
x
ru
n
th
ro
ug
ho
ut
th
e
pr
oc
ed
ur
e.
72 Automated two-step reactions with heterogeneous metal catalysis
Start
Start pump at 0.1 mL/min
Stop pump
Set ow rate to 0.2 mL/min
Set heater to 100 ˚C
Wait until temperature > 95 ˚C
Set pressure limit to 25 bar
Set input to reagent
Set input to solvent
Power o reactor
Set valve to waste
Set valve to collect
Switch on column heater
Wait for pressure and
temperature to build up
Switch on hydrogen
Wait for hydrogen to be
released
Switch o H-Cube
Wait for 30 minutes
(H-Cube dead volume) Alert chemist if any of the
pumps lose pressure
Wait until IR shows good
conversion to intermediate
Wait until reservoir contains 
enough solution to continue
Wait until reservoir is almost
used up
If IR response of intermediate
falls too low
Switch to waste until the IR
response returns to normal
If column temperature falls
below 95 °C or pump pressure
uctuates
Wait for 2 hours, or until the
reservoir gets close to being full
Wait for 40 minutes 
(dead volume of zirconia 
column + connecting tube)
Wait until input switches 
back to solvent
Collect output
If H-Cube temperature falls
below 95 °C
Stop collecting output until
30 min after temperature 
returns to normal 
(H-Cube dead volume)
Stop collecting output
End
Switch to waste until one 
column volume after the system 
has returned to normal
Figure 3.14 Control sequence for the two-step continuous process.
hydrogenation parameters to the optimised values of 0.1 mL/min and 100 ◦C (Figure 3.14).
The flow rate of the hydrogenation was limited by the performance of the H-Cube reactor,
which would ideally be replaced with a higher-throughput equivalent. Nevertheless, this
procedure demonstrates how two steps with disparate flow rates can be joined in sequence.
In this case the hydration/hydrogenation sequence was run for 14 hours (Figure 3.15), with
the first step running until the intermediate reservoir was full, and then shutting down.
However, we envisage that this protocol could be extended to a longer sequence where the
first step would run for only half the time, to keep the reservoir topped up. The size of the
reservoir could be adjusted to minimise the amount of stored material and the amount of
material wasted during startup and shutdown of the first reactor.
3.6 Conclusion
The Python software developed in Chapter 2 was applied to a controlled two-step synthesis
of piperazine-2-carboxamide (39).148 Over 14 hours, 7.3 g (95 %, 99 % purity) of 39
was produced. We were able to integrate the devices from different manufacturers and
define time- or event-triggered transitions between different stages of the experiment, and
3.6. Conclusion 73
-20
 0
 20
 40
 60
 80
 100
 120
 0  2  4  6  8  10  12  14
Time /h
Float Height (arbitrary units)
IR Absorbance at 1670 cm-1 /µAU
H-Cube column temperature /°C
Vapourtec heater temperature /°C
Figure 3.15 Selected parameters as observed over a 15 hour reaction. The output from
the hydrolysis step is directed into the reservoir as soon as the infra-red absorption at
1670 cm−1 passes a pre-defined threshold. When enough of the intermediate has been
collected, the H-Cube is powered on and the hydrogenation is started. The hydrolysis is
stopped after the reservoir fills up, while the hydrogenation continues until the intermediate
has been used up.
responses to events such as pump failures. Data from each device were logged so that the
performance of the system could be evaluated after an experiment. The software provided
real-time feedback via its web interface so that the system could be left to carry out the
experiment unattended. Having established the utility of the software control for this
process, we investigated the control of more complex reactor configurations.
74
Chapter 4
An integrated flow synthesis of
2-aminoadamantane-2-carboxylic acid
75
76 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
4.1 Introduction
Flow chemistry has the potential to enable moderate scale-up operations within the research
laboratory. Greater efficiency can be achieved by creating integrated multi-step processes
which are designed such that material produced in one stage can be used directly in the next,
reducing the requirement for manual intervention between steps.169 Continuous processing
is one good way to carry out such procedures, because by making the appropriate fluidic
connections material can move directly from one reaction vessel to the next. However, the
continuous processing of a sequence of synthetic steps (often referred to as “telescoping”)
creates similar challenges to the design of one-pot reactions in batch mode.170
4.1.1 Challenges of telescoped reactions
Chemical compatibility and purification
In order to connect two chemical transformations in sequence, either as a one-pot batch or
a telescoped flow process, the output of the first reaction must be chemically compatible
with the second. This output is likely to contain spent reagents and by-products, which
must be removed unless they are compatible with the conditions of the next reaction. For
this reason, batch protocols nearly always include work-up or solvent extraction operations,
or purification of the desired intermediate by chromatography or crystallisation.
A particular challenge for a flow system is that the compatibility of the solvents between the
two steps must also be considered. In a batch method it is usually straightforward to change
the solvent between steps; in flow this is not trivial, because by its nature the materials must
generally be kept in solution.
Flow rates and dilution
The introduction of flow streams for quenching the reaction output or for reagent addition
will lead to an increase in the flow rate within the system, and dilution of the reagents, as
more solvent is added (Figure 4.1). The higher flow rate will mean that reactors with larger
volumes will be required for later steps, and this effect may make some multi-step processes
impractical unless some of the solvent can be removed to concentrate the reagents and
reduce the flow rate.
4.1. Introduction 77
0.5 mL/min 0.5 mL/min
1.0 mL/min
0.50 M 
1.5 mL/min
0.33 M 
2.0 mL/min
0.25 M 
Figure 4.1 Dilution of reagents and increase in flow rate as additional streams are added.
It is also important to start the pumps involved in the downstream processes at the correct
times, because otherwise the solvents and reagents for the later stages will be wasted whilst
material moves through the earlier steps (Figure 4.2). The addition of an in-line detector to
identify when a reaction plug is exiting from one reactor can allow the next reagent to be
matched (both temporally and stoichiometrically) so that an appropriate amount of material
can be used.
No need to start this pump
for the rst ~ 20 min
(depending on dispersion)
20 min
Figure 4.2 Reagents are wasted when downstream pumps are started before the reac-
tion plug has passed through previous stages.
Solutions
The use of solid-supported and heterogeneous reagents has produced a number of elegant
multi-step flow chemistry procedures,25,51,171,172 but unless these reagents are used catalyti-
cally then the limited quantity of reactive material in any cartridge can make this technique
unattractive. A solid reagent cartridge must be replaced when it is exhausted, whereas
liquid or gaseous consumables can be added and removed continuously making the process
easier to operate. Furthermore, the high cost of some polymer-supported reagents can be
prohibitive when large quantities are needed.
Continuous workup operations such as solvent extraction,173 filtration,174 distillation175
and evaporation176 are well established when working on a large (manufacturing) scale, but
the application of these techniques into a bench-top device presents a number of challenges.
These generally result from the formation and handling of solids and slurries within the
78 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
system.177–179 The reduced dimensions of the equipment mean that any solid particles that
form are much larger in comparison to the pumps and channels. In addition, precipitates
can create blockages much more rapidly, and the smaller pumping equipment is often more
susceptible to damage or disablement by even very fine suspensions.
Continuous solvent extraction is well-developed on the laboratory scale. There are a number
of reported methods utilising hydrophobic membranes,180–182 settling tanks,92 selectively
wetting channels183 or counter-current flow.184 With two of these devices in succession, it
can also be possible to achieve a solvent exchange by passing materials from an organic
phase into an aqueous phase, and then back into another organic stream.180
Our group has recently developed an in-line solvent evaporator based on nebulisation of a
liquid stream within a heated chamber.185 This may offer one solution to the removal of
volatile solvents or in-line concentration of streams to reduce the dilution effect.
Tools for continuous (multiplexed) chromatography are commercially available, though
expensive; and continuous crystallisation methods have been reported186,187 although the
resulting slurry must then be processed.
Software control can be used to time reagent addition to minimise waste,89 to ensure
that each step is run at the correct time, and to monitor for any problems that arise and
either make corrections or alert the operator. In this chapter the application of downstream
processing techniques and software control systems to multiple step chemistry involving
work-up operations is discussed, using the example of a recently reported synthesis of
2-aminoadamantane-2-carboxylic acid (44, Figure 4.3).188
H2N
O
OH
Figure 4.3 2-aminoadamantane-2-carboxylic acid (44)
Our group had originally become interested in the synthesis of this unnatural amino
acid during investigations into the synthesis of probes for neurotensin receptors 1 and
2.189 As part of a programme to synthesise large quantities of adamantane-containing
compounds,190,191 effective methods to prepare the key component 44 on a large scale were
developed.
4.2. Previous synthetic routes 79
4.2 Previous synthetic routes
4.2.1 Reported batch–mode syntheses
Previously-reported methods to access 2-aminoadamantane-2-carboxylic acid include a
Bucherer–Bergs reaction as employed by Nagasawa and co-workers192 (Scheme 4.4, con-
ventional conditions), involving heating with sodium cyanide to form a hydantoin (45) that
could be directly hydrolysed to the desired amino acid. However, our group was unable
to isolate this compound with reasonable purity either using the original conditions, or
modified microwave processing conditions (Scheme 4.4, microwave conditions), due to a
high degree of inorganic impurities.188
H2N
O
OHO
N
H
NH
O O
NaCN
(NH4)2CO3 NaOH
a b
46
45
44
Scheme 4.4 Bucherer–Bergs routes to amino acid 44. Conventional conditions: (a)
NaCN (2 equiv.), (NH4)2CO3 (1.5 equiv.), EtOH, 170
◦C, 170 psi, 3 h (99 %); (b) 1.25 N
NaOH, 195 ◦C, 250 psi (pressure vessel), 2 h (94 %). Microwave conditions: (a) NaCN (2.2
equiv.), (NH4)2CO3 (1.5 equiv.), 3:1 EtOH/H2O, 170
◦C, 150 psi (sealed vial), 2 h (94 %);
(b) 1.25 N NaOH, 175 ◦C, 3 h (71 %).
A second route attempted was based on a Strecker reaction as inspired by Commeyras193
and Edward.194 The nitrile product generated by treatment of the ketone with sodium
H2N
CO2H
O H2N
CN
HN
CN
Ph
O
HN
CO2H
Ph
O
NaCN, NH3,
NH4Cl
EtOH / H2O
(68 %)
HCl
THF / H2O
25 °C, 5 min
(93%)
HCl
AcOH / H2O
reflux, 30 h
(65%)
PhCOCl
K2O3
THF / H2O
25 °C, 1 h
(52%)
46 47 48
4944
Scheme 4.5 Edward’s Strecker route to amino acid 44.
80 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
cyanide was hydrated using a milder method than in Nagasawa’s route. Reaction of
nitrile 47 with benzoyl chloride gave an α–amide; under acidic conditions this compound
underwent hydrolysis to afford intermediate 49, which could be hydrolysed to give the
amino acid (Scheme 4.5). However, a significant drop in reaction yield and purity was
noticed when this reaction was performed at scale.188
4.2.2 First generation flow–mode synthesis
To overcome these problems, our group reported188 a flow synthesis, involving five separate
stages: an initial Grignard addition, followed by a Ritter reaction, cyclisation, ozonolysis
and hydrolysis (Scheme 4.6).
The Grignard addition of ethynylmagnesium bromide (50) to adamantanone was performed
in THF at 40 ◦C, and the product was subjected to an in-line aqueous workup to quench the
resulting magnesium salts. Under the continuous conditions the temperature of this exother-
mic process was controlled effectively. This was an important consideration because the
product 51 had been found to be liable to thermal decomposition at elevated temperatures.
Subsequently, compound 51 was subjected to strong acidic conditions in an acetonitrile
solvent to effect a Ritter reaction.17,195 The output was quenched by simultaneous addition
of ethanolic potassium hydroxide to the collection tank. After filtration, the resulting basic
solution could be used directly for the 5-exo-dig cyclisation by pumping though a coil
heated to 120 ◦C.
Cyclic intermediate 53 was subjected to ozonolysis using a biphasic gas / liquid system, and
the resulting ozonide quenched using a polymer-supported thiourea25 to provide azlactone
54. This could either be used directly as a coupling reagent196 or subjected to acidic
hydrolysis to afford the amino acid 44.
The scope of the chemistry for this flow synthesis had been thoroughly established during
the first-generation synthesis.188 However, a number of manual work-up operations were
required which prevented the steps from being combined into a single sequence. The aim
of this work was to use our knowledge of the process to integrate the chemistry steps to
produce a single sequence. We envisaged that recently developed in-line workup techniques,
along with appropriate automation and control, could be employed to achieve this.
The main obstacles that were identified are highlighted in Scheme 4.6. The Grignard
reagent 50 used in Step 1 was only commercially available in tetrahydrofuran (THF), which
4.2. Previous synthetic routes 81
Step 1: Grignard
Step 4: Ozonolysis
Step 3: Cyclisation
Step 5: Hydrolysis
Step 2: Ritter
1A
2A
3A
4A
3B
2B
4B
1B
1C
1D
Reagent is only available in THF. 
THF solvent is incompatible
with the Ritter step.
Aqueous work-up is required 
to quench and remove 
magnesium salts.
Acid must be quenched.
Produces large quantity
of insoluble salts.
Base must be
neutralised 
before the acid
hydrolysis in 
Step 5.
Stoichiometric
thiourea to 
decompose 
the ozonide.
40 min sonication
40 oC
0.5 M THF
NH4Cl (aq)0.20 mL/min
0.5 M THF
0.18 mL/min
40 psi
O
MgBr
HO
aqueous extraction
remove THF solvent
7 min
30 oC
MeCN / AcOH3 M KOH
40:1 EtOH / H2O
3 mL/min
1 mL/min
1 mL/min
HO
H2SO4,
AcOH, Ac2O50 min
120 oC
0.3 mL/min NH
O
100 psi
lter
remove base
O3
ozone generator
500 mL/min
(7–10% O3 ; 2–3 mmol/min)
Ar, O3
Ar
gas trap
ozonolysis reactor
O2
N O
N
H
NH2
S
QP-TU
CH2Cl2
4 mL/min
0.3 mL/min
18 min
150 oC
AcOH
0.4 mL/min
0.4 mL/min
HCl,
AcOH, H2O
N O
O
H2N OH
O 100 psi
50
46
51
51
52
53
54
44
Scheme 4.6 Overview of the first-generation flow route to compound 44.188 The main
obstacles to an integrated continuous sequence are highlighted. Isolated yields obtained:
Step 1, 106 g, 90 %. Step 2, 22 g, 91 %. Step 3, 8.2 g, 91 %. Step 4, 6.7 g, 95 %. Step 5,
6.3 g, 94 %.
82 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
is incompatible with the Ritter reaction in Step 2. Initial experimentation showed that
a solvent exchange from THF to acetonitrile was possible using an in-line evaporation
device,185 which we considered to be preferable to manually changing the solvent of the
Grignard reagent.
The Grignard reaction produces magnesium salt by-products that must be removed by an
aqueous extraction. Furthermore, each step requires different pH conditions: Step 1 in
base, Step 2 in acid, Step 3 in base and Step 5 in acid; so multiple quenching operations
are needed. Finally, the first generation synthesis employed a stoichiometric quantity of
solid-supported thiourea to decompose the ozonide intermediate, which we hoped to avoid.
We turned our attention first to a Grignard reaction and aqueous extraction sequence.
4.3 Grignard reaction and aqueous extraction
The output of the Grignard reaction consisted of salt 55 in THF, along with any unreacted
reagent 50. To perform an aqueous quench and extraction, this stream must be combined
with an aqueous stream, mixed and then separated (Scheme 4.7).
Organic
0.3 mL/min
phase separatorcontactor
HO
BrMgO
Aqueous
0.5 mL/min
aqueous waste
mixer55
51
Scheme 4.7 Liquid-liquid extraction involves combination of the aqueous and organic
streams at an appropriate mixer (see Section 4.3.4) followed by phase separation.
4.3.1 Mixing
The first-generation procedure (Scheme 4.6) involved a standard tee connector ( 0.5 mm,
PEEK) immersed in a sonication bath to prevent aggregation of the magnesium salts upon
mixing. Indeed, without the sonication bath the tee mixer blocked within 10 minutes. Use
4.3. Grignard reaction and aqueous extraction 83
a
mixing zone
biphasic streamorganic stream
aqueous stream
1/8” OD, 1/16” ID 
PTFE tubing
1/8” tee union
1/16” OD, 1 mm ID 
PFA tubing
1/8” OD, 2.4 mm ID 
ETFE tubing
1/8” OD, 1/16” ID
PTFE tubing
b c
Figure 4.8 (a) Schematic of the tube-in-tube mixer, constructed using a Swagelok tee
connector and Swagelok reducers. (b) Photograph of a tube-in-tube mixer. (c) A vibrating
micromotor that can be attached to the mixer to reduce precipitate aggregation.
of a wider-bore tee mixer (Omnifit,  1.5 mm, PTFE) worked for a longer period (30 – 60
minutes) before becoming blocked.
Introduction of a tube-in-tube mixer (Figures 4.8a, b), emerging into a wide-bore tube
(R2,  2.4 mm, ETFE) provided a more robust solution that could run for around an hour
before precipitates caused a catastrophic pressure increase. Precipitation occurred within
the inner tubing as traces of water ingressed against the flow. It was found that a more
compact alternative to an ultrasonic bath for preventing aggregation was the attachment of
a vibrating micromotor (Figure 4.8c) to the tubing just after the tee mixer; the vibration
displaced the precipitates, which were then dissolved by the aqueous stream. With this
device in place, the mixer was able to operate continuously for over 6 hours.
This mixer was configured so that the inner tubing protruded at least 5 cm into the wide
tubing. The aqueous flow around the inner tubing displaced solids adhering to the end of the
inner tubing. Slicing the inner tubing at an angle to create a tapered end and thus potentially
a larger contact area had no noticeable effect on the traces of precipitate remaining at this
point. Without the precipitation, the mixer was found to operate best when the outlet was
84 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
Figure 4.9 Orientation of the tube-in-tube mixer. Since the organic solvents in this
system (indicated by yellow shading) were less dense than water, the organic phase may
have risen towards the tee body if the mixer was oriented downwards. However, if the
mixer was oriented upwards then the precipitates could fall towards the tee body. An
inverted-U configuration was the most effective, because the organic phase did not move
backwards and the solids fell away from the mixer.
inclined upwards so that the organic phase would tend to rise away from the tee connector.
However, when precipitates were formed they tended to fall back towards the tee body,
disrupting the aqueous flow. A better arrangement involved an inverted U-bend in the
tubing so that any organic bubbles which moved backwards would travel no further than
the apex (Figure 4.9).
4.3.2 Contactors
When a biphasic system is formed, it is important to ensure that effective contact occurs
between the two phases, for complete extraction.183 Contactors effective in laboratory scale
continuous processing include static mixers, dynamic mixers, or slug flow in a capillary.
In a static mixer, the flow stream passes through a packed column which disrupts the flow.
In a dynamic mixer the flow stream moves past an agitating device such as a magnetic bead
contained within the flow tube or column.92,178 Biphasic slug flow creates a mixing effect
due to internal circulation within each slug.183,197
Without using wide-bore tubing, we did not have magnetic beads in an appropriate size to
create a dynamic mixing effect. Due to the solid particles present, use of a capillary was
not suitable, but we anticipated that circulation within the slugs as they passed though R2
would provide a good degree of mixing. Fortunately, we later found that a filtration device
was necessary which turned out to provide an additional static mixing effect.
4.3. Grignard reaction and aqueous extraction 85
4.3.3 Filtration
Whilst the majority of the salts formed during quenching of the Grignard reaction output
were removed by the aqueous stream, a very small quantity of insoluble salts were formed
and persisted within the flow stream. These created blockages in the phase separation and
evaporation steps, and so a filtration mechanism was included to capture them.
Addition of a 50 µm filter into the flow stream was ineffective because even a large filter
became blocked by particulates. Instead we envisaged the use of a filtration cartridge
with varying degrees of filter, which could accumulate the precipitates without becoming
blocked.
A number of potential filtration media were identified which could be loaded into a glass
column: cotton wool, glass wool, celite, sand, charcoal or glass beads (Figure 4.10). Cotton
wool became compacted and then blocked after approximately 4 hours. A glass wool plug
was more robust and collected the precipitates without becoming blocked over several runs
(up to 10 hours). Although it might have been possible to perform an occasional acidic
flush to remove the accumulated particulates, glass wool is hazardous to handle, proved to
be difficult to pack into the column, and the plug introduced a large dead volume into the
flow system.
We were concerned that sand might leach iron into the system, so instead we evaluated the
glass
wool
plug
frit small
glass beads
medium
glass beads
large
glass beads
charcoalcelite
Figure 4.10 Filter cartridge configurations. (1) A cartridge containing a frit, creating an
aqueous chamber through which the organic phase passed. The frit blocked rapidly. (2) A
glass wool plug caught larger particles to protect the frit, but there was still a large dead
volume. (3) Small glass beads were not stable under the flow conditions and migrated
downwards. (4) Celite did not migrate but had undesirable packing within the column. (5)
A charcoal powder bed provided effective filtration. The glass beads kept the charcoal
powder in place.
86 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
use of glass beads of a comparable size to the sand particles (53 – 106 µm). A three-layer
column was prepared, with a top layer of the small beads, an intermediate layer of medium
beads (0.5 – 0.7 mm) and a bottom layer of large beads (4 mm); under static conditions
each layer held the one above in place. However, when the slug-flow stream was applied,
the top layer seemed to fluidise, causing mixing of the differently-sized beads such that the
smallest ones migrated towards and out of the bottom of the column.
We anticipated that smaller particles were more likely to pack than migrate, and so different
materials were evaluated in place of the smallest glass beads. Celite powder198 remained in
place as desired, but seemed to aggregate upon wetting forming channels through which
the majority of the flow stream passed. Charcoal powder199 was much more effective, since
it was easy to load into the column, it stayed in place without packing, and it collected
the small particulates without blocking. The glass beads collected the larger particles and
prevented the charcoal from falling down the column. Although charcoal is more likely to
be chemically active than the other options, we did not expect this to be a problem since
the adamantane-containing compounds are not aromatic.
Pleasingly, this column also provided good mixing, and the biphasic stream emerging from
the top of the column was separated into very small slugs. 1H NMR analysis of the aqueous
phase after separation could not detect any of the product 51, and removal of the solvent
from the organic phase left no inorganic salts, indicating that the quenching and extraction
were effective.
4.3.4 Separation
We chose a gravity-based separation method,92 whereby the biphasic stream is directed
into a reservoir (S1) where the phases separate under gravity. A pump (P4) withdraws the
lower aqueous layer, allowing the organic layer to escape from the top (Figure 4.11). When
performed in an Omnifit glass column this can operate under pressure,200 and the lack of a
membrane means that it is robust to small quantities of solid formation.
Unfortunately, under continuous operation, the THF / water solvent mix proved to be highly
miscible and the phase separation was not effective: there was no visible meniscus and
a significant quantity of the product (51) passed into the aqueous stream. The use of a
concentrated salt solution instead of water did not create any significant improvement in
the separation, so alternative solvent mixtures for the Grignard sequence were investigated.
4.3. Grignard reaction and aqueous extraction 87
75 psi
Sat. NH4Cl (aq) / H2O(3:1 v/v)
0.5 mL/min
mixer
webcam
40 psi
250 psi
organic out 
aqueous out
(       also present in aqueous due to poor separation with THF solvent)
HO
P3
P4
M2
R2
S1
charcoal
lter
glass
beads
40 min
40 oC
FlowIR
0.5 M THF
0.30 mL/min
0.5 M THF
0.27 mL/min
40 psi
O
MgBr
BrMgO
P1
P2
R1
M1
a
50
46
55
51
51
15 mm Omnit
column xed end 
tting with O-ring.
PTFE; no frit.
Drilled and tapped 
to match Omnit
screw thread.
Upchurch Scientic
pressure-release valve
tee piece, bored through
to 1/8”.
21G needle inserted
into tubing, and
trimmed to protrude
into glass chamber.
1/16” OD, 0.5 mm ID
PTFE tubing.
Heavy phase
outlet
Combined inlet
b
Figure 4.11 (a) Grignard reaction and liquid-liquid extraction apparatus. For the solvent
extraction, the aqueous and organic streams are combined at tee mixer M2, and then pass
through a wide-bore tubing (R2) and the filtration cartridge into the phase separator (S1).
The high-pressure BPR on the lower exit prevents material being forced through pump P4.
The low-pressure BPR on the upper exit acts as a check valve to prevent back-flow. (b)
Construction of the tee fitting for the bottom port of the phase separation chamber.
88 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
Use of diethyl ether as a solvent for the stream of adamantanone (46) led to very good
separation, but low solubility of compound 51 in a THF / Et2O solvent mixture resulted in
increased precipitation in the reactor and aqueous mixer (Table 4.12). Addition of toluene
led to improved solvation of the product, and also a less compressible solvent system which
could be pumped more reliably. A 3 : 2 (v/v) mixture of toluene / Et2O was found to provide
a good balance of solvation, immiscibility with water when combined with the THF stream,
and sufficient volatility to allow the solvent-switch into acetonitrile.
Table 4.12 Solvent conditions for the adamantanone stream in the Grignard reaction.
Both THF and toluene are readily removed under the nebulising conditions.185
Solvent mixture Outcome
THF High miscibility with water.
Et2O Poor pump operation due to high compressibility.
Low solubility of the starting material.
Toluene Low solubility of reaction products.
Toluene / Et2O (1 : 1 v/v) Some precipitation of starting material.
Toluene / Et2O (3 : 2 v/v) Good solubility.
Although the use of Et2O is undesirable,
201 it was necessary when using this equipment
configuration because volatile solvents were required for a practical solvent exchange
operation.
Extractor control
As in the originally reported phase separation system,92 the separation tank was observed
using a camera and the position of the phase boundary measured using computer-vision
enabled monitoring of a plastic float167 (see Section 3.5.1). This enabled the system to
compensate for flow rate fluctuations: if the two aqueous pumps did not operate at exactly
the same flow rate (for example, in the event of a pump failure or mis-calibration) then
over time the phase boundary would drift up or down. By controlling the flow rate of the
pump that removes the lower phase based on the position of the interface, this was avoided
(Figure 4.13*).
* For the Python control protocol, refer to Reference 202.
4.3. Grignard reaction and aqueous extraction 89
Solvent Extractor Feedback Control
max P4 ow rate: 
V1 collecting: (P1 + P2 + P3) total rate
V1 waste: (P3 + P5) total rate
Run control systems
until end of experiment
Reservoir S1 oat height Aqueous out pump P4 rate
> 0  on
< 0  oOrganic stream ow rate Aqueous in pump P3 power
Aqueous out pump P4 power
Acetonitrile pump P6 power
Reservoir S1 oat 
too high or too low
Alert chemist (pump failure)
Figure 4.13 Control protocol for the solvent extractor. The flow rate of pump P4 is
adjusted based on the liquid level in S1. If the float moves out of bounds, an alert is
triggered. If the organic stream is flowing (either valve V4 is collecting or pump P5 is
powered on) then the three extractor pumps are powered on.
The original system used a proportional-gain controller to control the output pump.200
A similar algorithm was implemented for this system, but in this case incorporating a
predictive extrapolation to slow down the rate of change of the response as it approaches
the set point, to reduce the effect of the lag time between the flow rate adjustment and the
response which otherwise tended to lead to oscillations in the system. This is based on a
proportional-differential controller; the sample interval was set to 20 seconds to reduce
the effect of noise on the derivative term. A threshold was added so that the flow rate was
kept constant when the liquid level was near the desired point (Figure 4.14). The feedback
control was performed according to:
errort = heightt− target
di f f erential = errort− error(t−interval)
change = (0.25×|errort |)× (errort + interval×di f f erential)
response = responset + change
where target is the desired float height (in pixels as observed by the camera), heightt is the
float height at the current time t, and response is the new flow rate of the pump.
90 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
 200
 250
 300
 350
 400
 450
 0  0.2  0.4  0.6  0.8  1  1.2  1.4  1.6
0.2
0.4
0.6
0.8
1.0
Flo
at 
he
igh
t /p
ixe
ls
Flo
w r
ate
 /m
L/m
in
Time /h
Float height
Flow rate
Figure 4.14 Response of the pump rate to the float height during part of an experiment.
The liquid level is maintained within a small window (8 mm). The plateaus indicate when
the threshold filter was taking effect.
4.3.5 Control of the Grignard reaction
Control systems were also added to the Grignard reaction step (Scheme 4.15). The output
from the reactor was monitored by an IR spectrometer. The absorption at 1120 cm−1 was
indicative of the product, and by passing the output through valve V4 it could be directed to
waste until the product was observed. An additional pump (P5) was added to flush solvent
40 min
40 oC
FlowIR
solvent waste
0.30 mL/min
0.5 M THF
THF MeOH
0.28 mL/min
0.5 M 3:2 Toluene/Et2O
40 psi
100 psiO
MgBr
BrMgO
P1
P2
R1
M1
V4
V1
V2
V3
3:2 Toluene/Et2O
0.5 mL/min
P5
50
46
55
Scheme 4.15 Grignard reactor configuration with four valves and an additional pump P5
to enable flushing of the pumps with solvent, and flushing of mixer M1 to prevent blockages
by precipitation at times when the output of R1 was being directed to waste.
4.3. Grignard reaction and aqueous extraction 91
through valve V4 at 0.5 mL/min whenever the reactor output stream was being directed
to waste. This ensured that there was no standing solution within the mixer, preventing
precipitation and blockage.
The system was also adapted to reduce the incidence of problems with pump P1 caused by
the reagent stream. Some batches of ethynylmagnesium bromide contained small quantities
of particulates from water ingress, which could disable the pump. A 2 µm in-line filter was
installed before the pump in order to filter the solution prior to entering the pump, and a
rigorous drying protocol was used before pumping the organometallic reagent. Nevertheless,
pumping performance was occasionally degraded.
The control software monitored the pressure sensor readings on the pump outputs to identify
the characteristic oscillations caused by reduced effectiveness of either pump’s check valves.
If the oscillations persisted for over a minute, then the input streams were switched to the
solvents (valves V1 and V3) and the pump flushed at a higher flow rate (1 mL/min) for ten
seconds. If this did not rectify the problem, the solvent was switched to methanol (valve
V2) and the flushing process repeated before reverting to THF. If this was also unsuccessful
then the operator was alerted, otherwise normal operation was resumed.
Using these control systems reliable pumping was achieved for several hours at a time with
only minimal intervention. An overview of the control protocol is shown in Figure 4.16.
Grignard Reaction
Experiment Complete
Switch P1, P2 to solvent
Set pressure limit to 30 bar
Set ow rates for pumps P1 – 5
Power on FlowSyn
Wait for R1 temperature
Switch P1, P2 to reagent
Run control system until 
reagent is exhausted 
(pre-congured duration)
Wait for product IR peak to
decrease below threshold
Wait for product IR peak to
rise above threshold
Wait 20 min
Power o FlowSynStart collecting output
Stop collecting output
IR output below threshold
Monitor pump P1, P2 pressure
Valve V4 position
Pump P5 power
Switch valve V4 to waste
waste  on / collect  o
Figure 4.16 Control protocol for the Grignard reaction. Once the system is running, two
control loops are started that switch P5 on and off, and control the position of V4 based on
the IR sensor output.
92 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
4.4 Solvent exchange
The organic output from the separator was combined at mixer M3 ( 0.5 mm, PEEK) with
a stream of acetonitrile delivered by pump P6 and directed into the solvent evaporator.
The evaporation device developed within our group185 consists of a narrow-bore tube
( 0.1 mm) projecting into a glass chamber. A jet of gas, usually nitrogen, is emitted
from a slightly larger tube surrounding the first one (Figure 4.17). The gas causes the
liquid stream to break up into droplets, and their larger surface area promotes evaporation.
Solvents with high evaporation rates (DIN 53170 standard203) are readily removed by this
system, enabling a solvent exchange from a mixture of THF, diethyl ether and toluene into
acetonitrile, which was used as both the solvent and a reagent for the subsequent Ritter
reaction step.
MeCN
1.2 mL/min
THF / Et2O / Toluene
N22 L/min
100 psi
HO
organic out 
(MeCN)
vapours out
(condenser)
HO
a bP6 M3
P7
peristaltic
51
51
Figure 4.17 Solvent extractor apparatus. The organic stream from S1 is combined with
a stream of acetonitrile and directed into the evaporation chamber. A jet of N2 nebulises
this stream and vapourises the THF, Et2O and toluene solvents to leave a solution of 51 in
acetonitrile, which is removed by peristaltic pump P7.
The combined gases pass out from the top of the chamber, whilst the liquids are withdrawn
from the bottom by a peristaltic pump (P7). This pump also functions as a back-pressure
regulator, preventing the liquid from being forced out by the gas pressure before it has had
time to concentrate. A peristaltic pump also has the advantage that it can operate just as
effectively when there is no liquid feedstock present, unlike normal HPLC pumps.
Under optimised conditions (Table 4.18, entry 15) the THF content of the stream was
reduced to around 5 % which was found to be sufficiently low to prevent extensive poly-
merisation under the acidic Ritter reaction conditions. The incoming spray of acetonitrile
4.4. Solvent exchange 93
Ta
bl
e
4.
18
S
ol
ve
nt
ev
ap
or
at
io
n
co
nd
iti
on
s
to
m
ax
im
is
e
TH
F
re
m
ov
al
an
d
av
oi
d
so
lid
or
so
lv
en
ta
cc
um
ul
at
io
n
w
ith
in
th
e
co
lu
m
n.
Th
e
ou
tle
tp
um
p
(P
7)
w
as
tu
ne
d
su
ch
th
at
th
e
liq
ui
d
w
as
no
tw
ith
dr
aw
n
to
o
qu
ic
kl
y;
th
e
pr
es
en
ce
of
so
lv
en
ti
n
th
e
co
lu
m
n
pr
ev
en
te
d
th
e
de
po
si
tio
n
of
so
lid
s.
E
nt
ry
G
as
pr
es
su
re
E
va
po
ra
to
r
P6
flo
w
ra
te
P7
flo
w
ra
te
So
lid
L
iq
ui
d
M
eC
N
:T
H
F
:T
ol
ue
ne
:E
t 2
O
a
/b
ar
Te
m
p.
/◦ C
/m
L
m
in
−1
/m
L
m
in
−1
de
po
si
tio
n
pr
es
en
t
1
1.
0
23
1.
2
0.
6
L
ot
s
A
cc
um
ul
at
io
n
N
ot
de
te
rm
in
ed
2
1.
2
23
1.
2
1.
0
–
–
1
:0
.0
5
:0
.0
3
:0
3
1.
2
25
1.
2
1.
3
L
ot
s
–
1
:0
.1
1
:0
.0
7
:0
.0
1
4
1.
2
25
1.
2
1.
0
L
ot
s
Sl
ow
A
cc
um
.
1
:0
.0
9
:0
.0
7
:0
.0
1
5
1.
2
25
1.
2
1.
6
L
ot
s
Sl
ow
A
cc
um
.
1
:0
.1
0
:0
.0
7
:0
.0
1
6
1.
2
25
1.
2
1.
6
L
ot
s
A
cc
um
ul
at
io
n
1
:0
.0
8
:0
.0
7
:0
7
1.
4
23
1.
1
0.
7
So
m
e
–
1
:0
.0
8
:0
.0
7
:0
.0
1
8
1.
4
21
1.
1
0.
7
–
–
1
:0
.0
6
:0
.0
6
:0
9
1.
4
21
1.
1
0.
6
So
m
e
Pr
es
en
t
1
:0
.0
7
:0
.0
7
:0
10
1.
4
23
1.
2
1.
0
–
–
1
:0
.0
8
:0
.0
6
:0
.0
1
11
1.
4
23
1.
2
0.
8
–
Pr
es
en
t
1
:0
.0
7
:0
.0
7
:0
12
1.
4
23
1.
2
0.
6
So
m
e
Pr
es
en
t
1
:0
.0
5
:0
.0
6
:0
13
1.
4
21
1.
2
0.
8
–
Pr
es
en
t
1
:0
.0
8
:0
.0
7
:0
.0
1
14
1.
4
21
1.
2
0.
6
–
–
1
:0
.0
6
:0
.0
7
:0
15
1.
4
22
1.
2
0.
6
–
Pr
es
en
t
1
:0
.0
4
:0
.0
4
:0
a
D
et
er
m
in
ed
by
1 H
N
M
R
sp
ec
tr
os
co
pi
c
an
al
ys
is
.
94 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
washes any solids from the walls of the chamber, and the resulting solution is directed into
the next phase of the process.
This represents an important development for continuous processing, since combining
reactions with differing solvent requirements often poses a significant challenge to multi-
step procedures. The possibility to change the solvent of a reagent stream in a continuous
fashion will enable a larger spectrum of chemistries to be connected together in flow
processes.
4.5 Ritter reaction and quenching
4.5.1 Intermediate reservoir
The output of the solvent-switch operation was collected in a reservoir (see Section 3.5.1),
which in this case was made from a solid-phase extraction (SPE) cartridge (S2). These are
equipped with a Luer connection at the bottom from which the solution could be withdrawn
(Figure 4.19, left), and a 20 µm frit which provides protection for the subsequent pump
(P8). The use of this reservoir was necessary because the output from the peristaltic pump
(P7) did not have a constant flow rate, and more importantly this pump cannot generate the
pressure required to run a flow reaction at elevated temperature with a BPR. The reservoir
also evens out any variation in concentration created by the evaporator.
Figure 4.19 Three of the work-up devices, mounted such that the reservoir and the
separator could be observed by one camera. Left: Intermediate reservoir consisting of an
SPE cartridge with a 20 µm frit and plastic float. Centre: Liquid extraction settling tank; the
float indicates the phase interface. Right: Charcoal/glass bead filter cartridge.
4.5. Ritter reaction and quenching 95
Ritter Reaction PI Control
max 1.4 mL/min
min 0.15 mL/min
Wait for 20 min
Run control systems
until end of experiment
Switch pump P10 
to reagent
Wait until reservoir contains 
enough material to start
Reservoir S2 oat height Reagent pump P8 rate
o / onAcid pump P9 power Base pump P10 power
Reservoir S2 oat too low
Reservoir S2 oat too high
Reservoir S2 oat no longer 
too low
Pause PI control
Stop reagent pump P8
Wait 1h in case reservoir rells
Wait for reservoir to ll up
Resume PI control
Switch valve V5 to waste
Reservoir S2 oat level dropped Switch valve V5 to collect
Flush with solvent for 40 min
Stop pump P8
Figure 4.20 Control protocol for the Ritter reaction step and control loops for reservoir
S2. The flow rate of pump P8 is adjusted based on the liquid level; if this falls too low
then the system waits for one hour before flushing the system with solvent. Pump P10 is
switched on whenever acid is being injected.
Additionally, the reservoir could be used to start the following step only when sufficient
material was available, and also to stop it if the material ran out, preventing disablement of
pump P8 by air bubbles (Figure 4.20). This allowed the control of the different steps to be
kept separate, leading to a less complex control system.
4.5.2 Ritter reaction
The reagent stream was withdrawn from the reservoir and mixed with an acidic stream at
a tee mixer (M4,  0.5 mm, ETFE) before passing through a flow coil (R3, 10 mL, PFA)
which provided approximately 15 minutes residence time at 25 ◦C. The flow rate of the
reagent stream was controlled by a second feedback loop such that the quantity of reagent
in the reservoir remained relatively constant, to allow buffering of the variable flow rate of
pump P7. A valve (V5) was set to discard the output from the evaporator if the reservoir
was close to overflowing (Scheme 4.21).
The acid stream was initially set at 0.2 mL/min to provide an excess of acid for the reaction.
By using an acid-resistant pump and removing all PEEK components from the fluidic path,
we were able to use a stream of neat sulfuric acid at a lower flow rate of 0.05 mL/min,
rather than as a solution in acetic acid and acetic anhydride. This reduced the dilution and
96 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
N22 L/min
HO
vapours out
waste
peristaltic
P7
S2
M4
R3
P8
V5
webcam
0.15 - 1.0 mL/min
H2SO4
0.05 mL/min
1:2:1 H2SO4 / Ac2O / AcOH
0.2 mL/min
30 min
25 oC
P9
or
NHO
51
52
Scheme 4.21 The output of the evaporation chamber is directed into reservoir S2. The
reservoir is used to buffer the material passed into the Ritter reaction step. Valve V5
is triggered to switch to waste if the reservoir is in danger of overflowing. Pump P9 is
switched on only when pump P8 is running.
created an overall lower flow rate, and more importantly reduced the quantity of acid that
had to be quenched in the following stage.
4.5.3 Base quenching
The chemistry of the 5-exo-dig cyclisation reaction (Step 3) had been extensively explored
during the development of the first generation synthesis.188 An acid-catalysed cyclisation
of compound 52 had been reported204 but this reaction was very slow, requiring up to 6
hours and was thus unsuitable for a flow process. The base-catalysed reaction was much
more efficient, being complete in 40 – 60 minutes at 120 ◦C. The key consideration was
to avoid the introduction of significant quantities of water, because this had been found
to result in hydration of the residual acetonitrile, competing with the desired reaction and
reducing the overall rate of cyclisation.
Thus, a solution of potassium hydroxide in 40 : 1 ethanol / water (v/v)188 was initially used
to quench the acidic stream. Removal of the salts afforded a primarily ethanolic stream
which could be used directly in the cyclisation reaction. It was very important to remove
all of the solid material generated by the quenching reaction, since it had the potential to
damage the low-pressure reagent selection valves or block the check valves in the pump,
causing it to fail. Therefore, a number of strategies for removing the salts formed during
the quenching reaction were explored.
4.5. Ritter reaction and quenching 97
Filtration method
Initial attempts to perform this filtration continuously were promising, as combination of the
two streams within a second tube-in-tube tee mixer (M5) generated a slurry which passed
through a wide-bore residence tubing (R4;  2.4 mm, 30 cm, ETFE) and was dropped onto
a sintered glass filter funnel (Figure 4.22). In dilute cases this worked well, but at the
full concentration, precipitation was not complete by the time it passed though the filter,
resulting in large quantities of solid settling in the reservoir. The ethanolic slurry had a high
viscosity, and R4 eventually became blocked with solid. The slurry was too viscous for a
vibrating motor (such as that used for the previous extraction process) to keep it moving
effectively.
KOH (3 M)
40:1 EtOH / H2O
2-3 mL/min
MeCN, H2SO4 
0.4 - 0.7 mL/min
NHO
wide-bore
residence tubing
to reservoir (S3)
a b
M5
R4
P10
sand
glass 
sinter
funnel52
Figure 4.22 Removal of salts by filtration using a glass sinter funnel. A sand bed prevents
the sinter from becoming blocked by the salt.
In this case, further dilution of the stream was not practical because the combined flow rate
was already high in order to provide sufficient base to completely neutralise the acid. Since
a relatively long residence time (up to an hour) was required for the cyclisation step there
was a practical limit on the flow rate for this step without needing a reactor with a very
large volume.
Extraction method
With this restriction on the flow rate in mind, we considered an alternative system involving
a second extraction. It was possible to use a biphasic stream of Hünig’s Base (DIPEA) in
cyclopentyl methyl ether (CPME, a high boiling point solvent which has a low miscibility
98 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
with water) and aqueous KOH to neutralise the acid and remove the generated acetate
salts in the aqueous phase. Under these conditions, the resulting slurry after mixing had a
significantly lower viscosity compared to the case where ethanolic KOH was used. Aided
by a second vibrating micromotor, blockages at mixer M5 were avoided. The slurry was
directed into a commercially available phase separation device182 containing a replaceable
expanded macroporous PTFE membrane, which allows only the organic phase to pass
through (Figure 4.23). A 5 M aqueous solution of KOH could be used at a lower flow rate,
ensuring complete quenching of the acid stream whilst reducing the volume of solvent
required.
KOH (4 M)
H2O1 mL/min
DIPEA (0.5 M)
CPME
0.5 mL/min
MeCN, H2SO4
0.4 - 0.7 mL/min
NHO wide-boreresidence tubing
hydrophobic
membrane
separator
to reservoir (S3)
aqueous and
salts out
peristaltic
a b
M5
R4
P10
P11
P12
52
Figure 4.23 (a) Hydrophobic membrane separation apparatus. The acidic stream is
mixed with a biphasic base stream and then directed into the membrane separator. The
organic phase passes through the membrane and drops into the reservoir; the lower
aqueous phase is removed by peristaltic pump P12. (b) The separator consists of a
chamber and PTFE membrane. A custom glass chamber was made such that the liquid
level might be monitored by a camera and modulated with a pump such that none of the
organic phase was removed by the peristaltic pump.
The sulfate salts were not soluble in aqueous or organic solvents, but we envisaged that
a peristaltic pump could remove the remaining aqueous phase and sulfate salts from the
bottom of the separator. The organic solution passed through the membrane and dripped into
the reservoir for use in the following step. We were pleased to find that a good extraction
was achieved in this case, and that heating of compound 52 with DIPEA in CPME led to
the desired cyclisation.
Unfortunately, the peristaltic pump was not able to remove the solids from the phase
separator efficiently, and its inlet tube blocked rapidly (the peristaltic tubing did not
4.5. Ritter reaction and quenching 99
withstand immersion in CPME, so a fitting was required to connect a PTFE tubing which
would extend into the phase separator. The blockage occurred at this fitting). Therefore, we
turned our attention to further development of the filtration method to remove this salt.
Continuous filtration method
An effective neutralisation method was developed by using a KOH stream with a mixture
of ethanol and water as solvent. By adjusting the amount of water in this solution, the
viscosity of the resulting mixture was minimised to prevent blockages in the mixer, whilst
still retaining a high conversion in the subsequent step. The salt that was produced was
removed by a filter allowing the remaining solution to pass through into reservoir S3. We
envisaged that some of the water would be retained by the salt so that only a small amount
of the water would pass into the cyclisation step.
Using a glass sinter attached to a motor to allow it to rotate, a continuous filtration platform
was constructed (Scheme 4.24, Figure 4.25a).205 The slurry passing through R4 was dripped
onto the rotating disc; the rotation allowed the liquid to pass through before the residual
solid was scraped off. A replaceable paper filter caught the solid, whilst allowing the liquid
to percolate through. A hole in the centre of this filter allowed most of the solution that
passed through the sinter to drip straight through it. A final thin layer of cotton wool was
added to stop any large solids that were not caught by the paper filter. This system was
found to operate successfully for periods of up to six hours, with periodic replacement of
the filters.
KOH (4 M)
4:1 EtOH / H2O2 mL/min
MeCN, H2SO4
0.4 - 0.7 mL/min
NHO
wide-bore
residence tubing
to reservoir (S3)
cotton wool
replaceable lter
solids removed
rotating sinter
liquids pass 
through
M5 R4
P10
52
Scheme 4.24 Continuous filtration apparatus. The slurry stream was dropped onto a
sintered glass plate that was attached to a rotating motor. The liquid passed through the
sinter into the reservoir, and excess solids were pushed off by a fixed bar onto a paper
filter, from where any remaining solution could percolate through.
100 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
a b
Figure 4.25 (a) Continuous filtration apparatus. (b) Reservoir S3. The plastic surround
provides a blank background for the camera, and the dots drawn on the reservoir trigger
the camera’s autofocus on the optimal focal plane.
4.5.4 Cyclisation reaction
The 5-exo-dig cyclisation was performed by passing the basic stream collected in reservoir
S3 (Figure 4.25b) through a flow coil heated to 120 ◦C (Scheme 4.26). Another 20 µm frit
in the bottom of the reservoir protected the pump from any solids that managed to pass
through the main filtration device. The flow rate was controlled by another feedback loop
based on the volume present in the reservoir (Figure 4.27). Although this resulted in a
fluctuating flow rate for the subsequent step, a large 50 mL reactor coil (R5,  2.5 mm, SS)
provided sufficient residence volume for the flow rate to be varied between 1 to 3 mL/min
without affecting the degree of conversion of the reaction.
KOH, EtOH
NHO
S3
P13
webcam
1.0 – 3.0 mL/min
R5
25 min
(average)
120 oC
5 bar
gas-pressure
BPR
N O
N2
gas out
52
53
Scheme 4.26 Cyclisation step apparatus. Intermediate 52 was collected in reservoir S3
and then pumped through reactor R5 at 120 ◦C to effect the cyclisation. Feedback control
based on the liquid level of S3 attempted to match the flow rate of P13 with the rate of the
solution passing through the filter into S3.
4.6. Ozonolysis and Hydrolysis 101
Problems with blockages in the BPR were encountered when using a standard spring-based
BPR after R5. This was avoided by using a prototype gas pressure BPR developed within
our group.190,206 This device consists of a sealed vessel pressurised using a N2 supply
which can be connected at the end of a flow system. The product 53 was periodically
drained from the vessel using a manual valve at the bottom.
Cyclisation Reaction
PI Control
max 3 mL/min
min 1 mL/min
Set heaters to 120 °C
Wait for heaters to heat up
Set pump P13 to pump 
solvent at 0.1 mL/min
Set pump P13 to pump 
reagent at 1 mL/min
Run control system
until end of experiment
Wait until reservoir S3 contains 
enough material to start
Wait for reservoir S3 
to start to ll up
Reservoir S3 oat height
Reagent pump P13 rate
Figure 4.27 Control protocol for the cyclisation step. The reaction started when sufficient
material has been collected in reservoir S3, and then the flow rate of pump P13 was
controlled based on the liquid level.
Unfortunately, this device could not then be connected to feed into the next step. Further-
more, the base had to be removed before the final hydrolysis step to prevent salt formation
which might complicate the purification of the final compound. We were also concerned
that if this basic stream was mixed with a gaseous stream in the ozonolysis reaction, deposi-
tion of salts within the mixer could block the reactor. Therefore, the synthesis was broken
at this point to allow manual solvent evaporation and aqueous extraction to be performed.
4.6 Ozonolysis and Hydrolysis
To improve on the first generation flow synthesis, we envisaged that reported ozonolysis
conditions using an acetone/water solvent207 could be used to decompose the carbonyl
oxide intermediate in situ, avoiding the formation of an ozonide and thus avoiding the
requirement for the solid-supported thiourea reagent. Pleasingly, these conditions proved to
be very effective, generating a mixture for which the NMR spectra were compatible with
a 1 : 1 ratio of azlactone 54 and uncyclised 56,188 along with formaldehyde and hydrogen
peroxide by-products. The peroxide was removed by a bed of manganese dioxide,208
through which the liquid output of the ozonolysis reaction was allowed to percolate.
The output was directly combined with an acidic stream to perform the hydrolysis of both
compounds to the desired amino acid 44. The product was collected from the solvent
102 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
O3
ozone generator
O2 + O3
1:5:8 HCl, AcOH, H2O
0.25 mL/min
MnO2
5:1 acetone / H2O
2 mL/min
0.25 mL/min
ozonolysis reactor
(solvent evaporator)
18 min
150 oC
O2
N O
N O
O
H2N
O
OH
HCl NHO OH
O
100 psi +
P14
P15
P16
R6 R7
53
54 56
44
Scheme 4.28 Ozonolysis and hydrolysis sequence. A stream of compound 53 in acetone
/ water was combined with a gaseous stream of O3/O2 from the ozone generator. These
were mixed and allowed to react in R6, and then directed into column R7 from which the
gases could escape and a plug of MnO2 decomposed the hydrogen peroxide by-product.
The output was combined with an acidic stream and heated to effect hydrolysis before
being collected using a second solvent evaporator device.
stream by evaporation of the acids and solvents (Scheme 4.28).
The heated output of reactor R7 was passed directly into a solvent evaporator without
cooling. The stream at 150 ◦C was vapourised very efficiently, so all solvents and acids
were removed to leave a solid precipitate of compound 44.
4.7. Conclusion 103
4.7 Conclusion
The three chemistry steps and three work-up steps to prepare the cyclised compound 53
were combined into an integrated sequence which could operate semi-autonomously for
at around six hours to produce 9 g of 53 (92 %, 9.0 mmol/h). The integration of flow
chemistry steps represents a significant challenge, as when multiple devices are used
there are many parameters which must be managed simultaneously. There is an even
greater challenge when suspensions are generated or used, because key components of
flow chemistry apparatus that are currently standard are not suitable for handling solids
(in particular: most pumps, mixers and connections, and BPR assemblies). Technologies
adapted from those used for larger scale systems may help to enable such processes.
The combination of a tube-in-tube mixer and a vibrating micromotor was used to avoid
blockages whilst performing the quenching reactions, during which which precipitates were
formed. A prototype continuous filtration device was used to remove the salts formed during
a continuous quenching reaction, allowing two reactions with different pH conditions — a
Ritter reaction and a base-catalysed cyclisation — to be performed in succession.
A nebulising evaporator was used to perform an in-line solvent exchange, allowing the
Grignard and Ritter steps — with incompatible solvent systems — to be performed in
sequence. Finally, the ozonolysis reaction was improved by the use of an aqueous solvent
system and manganese dioxide catalysed decomposition of the hydrogen peroxide by-
product instead of a solid-supported thiourea.
Software control helped to make the management of multiple devices feasible. A camera-
enabled system controlled the rate at which the aqueous phase was withdrawn from the
separation tank, to compensate for flow rate fluctuations in the input streams. Similarly, the
application of camera-enabled reservoirs meant that the steps could be started automatically
as soon as the reagent solutions became available, and stopped if upstream problems caused
the input stream to stop. Feedback control loops were used to match the flow rates between
these steps so that they did not stop and start more frequently than necessary. The inclusion
of these reservoirs meant that the software control could be designed in a modular fashion,
with each step being mostly independent, resulting in a simpler control programme.
104 An integrated flow synthesis of 2-aminoadamantane-2-carboxylic acid
4.7.1 Further work
A number of potential improvements were identified, to advance this project or to enable
similar projects in the future. Further development of the control software could allow
the system could be left running unattended overnight. The key to this would be a safe
automatic shut-down in the case of a problem, in a manner that left the system in a state that
it could re-start immediately as soon as the problem was rectified. The second half of the
reaction sequence could also be fully automated by the inclusion of additional reservoirs to
connect the steps.
Improvements to the filtration device might result in a device which is less susceptible to
clogging by the solid material, and from which removal of the collected salts is easier. We
envisage that this might be a device based on a centrifugal principle, or a device including
a pressure differential across the filter, to force the liquid through and leave a dry solid
residue to be removed.
The addition of an internal liquid level sensor and a controllable valve at the bottom could
enable the gas pressure BPR to be fully automated, by periodically releasing the collected
material into a reservoir feeding another reactor. It might be possible to develop a self-
regulating solvent evaporator by the application of computer vision to measure the quantity
of solids and liquid inside the evaporation chamber, and adjust the temperature and flow
rates accordingly.
With further development of these devices, techniques and software, it is possible that in
the future integrated processes such as this one might be can be constructed and used as a
matter of course, rather than having to be developed as a dedicated project.
Chapter 5
Development of inhibitors for
bromodomain-containing proteins
105
106 Development of inhibitors for bromodomain-containing proteins
5.1 Introduction
A typical process for the discovery of a biologically-active molecule, whether for phar-
maceutical or for research purposes, often includes hit identification, lead optimisation
and then a structural optimisation stage based on the results of biological testing. “Hit”
compounds, which are compounds with a low but measurable activity against the target, are
often identified by either physical or virtual high-throughput screening (HTS) of a library
of compounds against a biological target. The efficiency of HTS methods has increas-
ingly been called into question209 and more recently, fragment-based discovery210–212 has
become more common. This entails the measurement of weak activities of small “frag-
ments”,* followed by the design of more active compounds by connecting these fragments
together to form larger hit molecules.
By whichever method they are identified, the compounds that are from the first stage will
give an idea of the structural features that may be important for binding, and thus demarcate
a region of “chemical space” for further investigation. Subsequently, chemical synthesis
of a number of similar compounds will allow the biological activity of these chemotypic
regions to be described in more detail, and provide information about how the structure
influences the activity: the structure–activity relationship (SAR). Repeated cycles of design
and experimentation may eventually provide a potent and selective lead compound.
There are a number of companies and academic research groups investigating how this
cycle (or parts of it) can be automated to speed up the development process, and to what
extent this can augment the traditional discovery pattern. The reports thus far have often
focused on increased automation: to achieve 24/7 working and produce greater numbers of
test compounds. Some have implemented machine learning or intelligent search protocols
to improve the automated compound design process so as to reduce the number of trials,213
in much the same way as optimisation protocols for individual reactions harness numerical
and linear programming methods for increased efficiency.160,164
The question of whether an automated process can contribute to a molecular design project
by performing intelligent ligand design with efficient use of the structure–activity data has
not yet been answered. A well-designed algorithm can make decisions in well-defined and
reproducible ways, and investigation into the most effective way to incorporate this into a
discovery programme is still ongoing.
* Compounds with fewer than 16 heavy atoms and a high proportion of hydrogen-bond donors and acceptors.
5.1. Introduction 107
The generation of realistic chemical complexity may well be beyond the scope of current
compound-design algorithms,214 particularly if a route for manual or automated synthesis
of each novel structure must be suggested as well. Fortunately, a great deal of research is
ongoing into these areas.215
Finally, given the huge challenge of predicting the activity of a certain ligand when com-
bined with a protein, the SAR method is a key technique for compound development. An
individual novel compound may have no detectable activity, but an array of 20 similar
compounds might contain one or two active molecules.216
Alongside work to further develop machine-assisted techniques for producing arrays of
functional compounds,24,50,112 there is also ongoing work within our group to apply in-
telligent automation to close the loop of lead optimisation by integrating chemistry and
biology.217 To understand the challenges involved in a typical discovery programme we
undertook a synthesis assignment in collaboration with the Structural Genomics Consortium
(SGC).
5.1.1 The bromodomain
A key research area in which the SGC is involved is in the study of human proteins,
especially those containing bromodomain (BRD) modules, and ligands which modify
or inhibit their function. Bromodomains constitute one part of a large family of protein
modules which assist in gene regulation by recognising or performing post-transcriptional
modifications of chromatins.
The bromodomain is a protein domain which recognises –N–acetylated lysine residues on
a histone protein.218 Histones form a scaffold that DNA winds around to form nucleosomes,
which condense to form a chromosome. One effect of acetylation is to alter the charge on
the histone; this is thought to weaken the association between the nucleosomes, allowing
the genes in that particular region of DNA to become more accessible to the transcription
factors that start the process of protein synthesis (Figure 5.1). Therefore, bromodomains
are a key instrument in gene regulation within organisms and thus an important area for
study, in particular in relation to diseases such as cancers.218
Over 40 proteins in the human body have been identified which contain bromodomains.218
Despite significant differences in the amino acid sequences of the BRD modules, they have
a conserved structure formed of four α–helices which surround a hydrophobic acetylated
108 Development of inhibitors for bromodomain-containing proteins
N
H
O
Transcription factor
Inaccessible gene
Acetylated histone
DNA
Histone Histone tail
Accessible gene
Figure 5.1 Histones are proteins around which DNA is wrapped within a chromosome.
Acetylation of the histone tails is an epigenetic modification which (amongst other effects)
is thought to weaken the association between adjacent histones, allowing greater access
of the genes to transcription factors, thus affecting gene expression.
lysine binding site.218 The role of BRDs is poorly understood, and hence there is a need for
selective ligands. Investigation by the SGC had bromosporine (57) as a potent inhibitor of
many BRDs.219 Figure 5.2 shows a co-crystal structure of bromosporine with the BRD4
bromodomain unit. The key interactions are those between ASN-140 and the northern
triazole and its pendant carbamate NH. The southern tolyl moiety displays a pi–pi interaction
with TRP-81, and the sulfonamide unit with ASP-88.
Two regions of this molecule were identified for further exploration: the heterocycle amine
and the sulfonamide. Depending on the observed activity of the analogous compounds, the
objective would be to develop either (a) a more potent pan-bromodomain inhibitor, or (b) a
more selective inhibitor for one or more BRDs.
a b
Figure 5.2 (a) Crystal structure and ligand interaction diagram of the pan-bromodomain
inhibitor bromosporine (57) bound into the BRD4A(1) bromodomain. (b) The key interaction
between compound 57 and BRD4A(1) is from the triazine nitrogen atoms to an asparagine
residue and a water molecule.
5.1. Introduction 109
5.1.2 Frontal Affinity Chromatography
Recently, the technique of Frontal Affinity Chromatography (FAC)220 on a miniaturised
scale has been developed.217 This method allows an assessment of the binding strength of a
test compound to be made using chromatographic equipment and an immobilised protein.
This technique was originally used for studying the interaction between carbohydrates and
lectins,220,221 but when used in combination with mass spectrometry can be applied to
a wider range of biological targets.222–225 Miniaturisation of the column containing the
immobilised protein reduces the amount of protein that is required: the high cost of some
proteins means that this is a very important consideration, and smaller columns offer the
ability to measure affinities with smaller quantities of ligand (which may also be highly
valuable).
Proteins can be immobilised using a number of methods, such as by covalent attachment223
or more commonly by a streptavidin–biotin H–bonded linkage.217 Since a protein can
be biotinylated at either the C or the N terminus, the latter is more likely to lead to a
site-selective immobilisation (i.e. where all of the protein molecules are attached from the
same position on the protein) so that the interaction of substrates with protein is consistent.
Biotin units can be attached to the terminus of a protein by expressing an avidin tag
and then performing a biotinylation;226 the protein can then be immobilised in a column
containing solid-supported streptavidin units (Figure 5.3). The interaction of biotin with
the streptavidin protein is the strongest known biological non-covalent interaction, having a
Kd of the order of 4×10−14 M;227 this strong attachment is crucial because it prevents the
protein from being washed off under a continuous solvent stream.
Biotinylation
Enzyme
Immobilisation
Supported
Streptavidin
Figure 5.3 A protein (blue) with an avidin tag expressed onto the protein terminus
is biotinylated (red) and immobilised by being captured by a solid-supported (green)
streptavidin unit (brown). The streptavidin complex has four units and thus four binding
sites, but due to the size of the protein they may not all be filled.
Under a FAC protocol, a relatively large quantity of a compound of interest is infused into
the protein-containing column. Whilst an inactive compound will pass through in one
column volume, an active compound will bind to the protein and be retained. The “front” of
110 Development of inhibitors for bromodomain-containing proteins
Co
nc
en
tr
at
io
n
Void Marker Compound
Volume
Δv
Immobilised Protein
No anity = No retention Anity = Retention
Figure 5.4 In a FAC analysis, the compound of interest is continuously infused into a
column containing an immobilised protein. Compounds are retained according to the
strength of their affinity for the enzyme. The breakthrough volume is (arbitrarily) chosen to
be the mid-point between the lowest absorption and the plateau, and is recorded relative
to that of a void marker that has no affinity for the protein.
the compound will emerge only after all of the binding sites have filled up (Figure 5.4). The
relative binding affinities of two compounds can be compared by measuring the respective
breakthrough volumes (which are proportional to the breakthrough times).
The FAC platform consists of an HPLC system running an isocratic buffered aqueous
solvent system, which is important for maintaining the stability of the immobilised protein.
Downstream of the column there are standard analytical detectors, such as a UV diode
array detector or a mass spectrometer (Figure 5.5).
By injecting a series of compounds of the same concentration, an order of affinity can be es-
Compact mass
spectrometer HPLC Pump
15 µL column
Make-up pump
Dynamic mixer
(removes pulsing)
Buer Solutions
UV/Vis Detector
Column Oven
Figure 5.5 The Frontal Affinity Chromatography apparatus consists of an HPLC system,
with a column (right) containing the immobilised enzyme. The pump delivers an isocratic
stream of an aqueous buffer, and 600 µL of the sample solution is infused (large compared
to the 15 µL volume of the column) into the FAC column that is maintained at 20 ◦C by
the column oven. The output is detected by a UV/Vis diode array detector and/or a mass
spectrometer.
5.1. Introduction 111
tablished. Alternatively, by injecting each compound as a series of different concentrations,
the binding coefficient Kd can be calculated from the breakthrough volumes. As part of this
project we hoped that the FAC technique could be used to discriminate between molecules
and give results which were consistent with mainstream assay techniques.
Processing FAC data If a solution of substrate A with concentration [A0] is continuously
infused into a column on which there are Bt immobilised protein units, then some of the
substrate molecules will bind to the proteins and be retained on the column. The number of
molecules of substrate removed from the stream corresponds to the retention volume:
Bocc = ∆V [A0] = (V −V0)[A0] (5.1)
where V is the substrate’s breakthrough volume, and V0 is that of a void marker. The ratio
of occupied (Bocc) to unoccupied (Bunocc) sites will depend on the binding affinity Kd . They
are related by the equilibrium:
A+Bunocc
k+1−−⇀↽−
k−1
ABocc (5.2)
and assuming that the activity quotient of the equilibrium is constant:
Kd =
k−1
k+1
=
[Bunocc][A]
[Bocc]
(5.3)
This assumption is reasonable in this case because the analysis is taking place in a buffered
aqueous solution, which has a high ionic potential.228 Since the protein is immobilised,
[B] =
B
Vcolumn
(5.4)
where Vcolumn is the volume of the column. Since Bt = Bocc+Bunocc,
Kd =
(Bt−Bocc)[A]
Bocc
(5.5)
=
Bt [A]
Bocc
− [A] (5.6)
112 Development of inhibitors for bromodomain-containing proteins
Therefore, using equation (5.1):
Kd =
Bt [A]
(V −V0)[A0] − [A] (5.7)
Under the conditions of constant infusion, [A] = [A0] and so
Kd +[A0] =
Bt
(V −V0) (5.8)
(V −V0) = BtKd +Bt
1
[A0]
(5.9)
This is analogous to a Michaelis-Menten system, and so a Lineweaver-Burk linearisation229
can be used to calculate Kd and Bt by plotting (V −V0) against 1[A0] .
5.2 Chemical probes for BRD9†
5.2.1 Literature
Previously reported approaches to the triazolopyridazine core such as seen in bromo-
sporine (57, Figure 5.6) have generally proceeded by generation of the triazole 60 from
a dichloropyridazine precursor 58, using for example acetyl hydrazide231,232 or in two
steps using hydrazine and then either triethyl methanetricarboxylate,233 acetic acid234–236
or acetic anhydride237,238 (Scheme 5.7). These reports suggest that after an SNAr reaction
N
N
N N
N
H
S
OO
H
NEtO
O
Figure 5.6 Bromosporine (57).
† The work described in this section is reported in MedChemComm (Reference 230).
5.2. Chemical probes for BRD9 113
onto one of the C−Cl positions, the other bond becomes deactivated and so only single
substitutions are observed.
N
N
Cl
Cl
R
N
N
HN
Cl
RN2H4.H2O
NH2
N
N
Cl
Rconditions
NN
AcNHNH2
58 59 60
Scheme 5.7 Reported preparations of the triazolopyridazine core, by either hydrazination
followed by cyclisation, or by direct cyclisation using a hydrazide. R = H, Me,239 NH2
235 or
CO2H.
236
A standard method for the preparation of the pyridazine starting material 58 is by treatment
of furan-2,5-dione (61) with hydrazine followed by chlorination of the pyridazine-3,6-diol
intermediate 62240 (Scheme 5.8). However, many 3,6-dichloropyridazine derivatives are
now commercially available.
O
O
O
N2H4.2HCl
NH
NH
O
O
N
N
OH
OH
N
N
Cl
Cl
POCl3
61 62a 62b 63
Scheme 5.8 Synthesis of 3,6-dichloropyridazine (63) from furan
Cross-coupling of triazolopyridazine 60 (R = H) is well represented in the literature, with
examples of Suzuki,241 Stille,242 Negishi243 and other metal-mediated reactions. However,
to the best of our knowledge, there have been no reports of cross-coupling reactions onto
substituted triazolopyridazines. Cross coupling of an analogous imidazopyridazine has been
established244 using palladium acetate and S-Phos (66; Scheme 5.9a), and more recently245
using Pd2(dba)3 and X-Phos (69; Scheme 5.9b). Finally, nucleophilic substitution at the
same position has been reported using secondary amines, although low yields were reported
(Scheme 5.9c).236
114 Development of inhibitors for bromodomain-containing proteins
N
N
N
Cl
H2N
N
N
N
Ar
H2N
ArBPin (1.5 equiv.)
Pd(OAc)2 (10 mol%)
S-Phos (20 mol%)
K3PO4 (2 equiv.)
5:2 n-butanol / water
100 °C, 48 h 23%
PCy3
OMeMeO
S-Phos
a
64 65 66
N
N
N
Cl
H
NR N
N
NH
NR
R = 6-(piperidin-1-yl)pyridin-2-yl
PhB(OH)2 (1.5 equiv.)
Pd2(dba)3 (10 mol%)
X-Phos (40 mol%)
K2CO3 (3 equiv.)
10:1 dioxane / water
100 °C, 4 h
58%
PCy3
i-Pri-Pr
i-Pr
X-Phos
b
67 68 69
N
N
NN
Cl
MeO2C N
N
NN
N
DMSO
130 °C (MW)
30 min
15%
Ph
H
N
MeO2C
Ph
c
70 71 72
Scheme 5.9 (a) Cross-coupling of imidazopyridazine 64 with an arylboronic ester using
Pd(II) and S-Phos.244 (b) Cross-coupling of imidazopyridazine 67 with phenylboronic acid
using Pd(0) and X-Phos.245 (c) SNAr reaction of triazolopyridazine 70 with a secondary
amine.236
SGC synthesis
The synthesis of 57 as reported by the SGC219 began with a sequence to form the tria-
zolopyridazine 76 from 3,4,6-trichloropyridazine (73). An SNAr reaction using ammonia
N
N
Cl
Cl
Cl
N
N
Cl
Cl
H2NNH3
EtOH
51%
N
N
HN
Cl
H2N
40%
N2H4.H2O
NH2
N
N
Cl
H2NAcOH
NN
82%
73 74 75 76
Scheme 5.10 Synthesis of precursor 76 starting from 3,4,6-trichloropyridazine (73).
5.2. Chemical probes for BRD9 115
N
N
Cl
H2N
NN
Boc2O
DMAP
THF
0 °C to r.t.
N
N
Cl
BocHN
NN Boronic acid___
Pd(PPh3)4
K2CO3
Dioxane / H2O
120 °C
N
NBocHN
NN
NO2
N
NH2N
NN
NO2
TFA, CH2Cl2
25 °C
EtOCOCl
NEt3
CH2Cl2
25 °C
N
N
H
N
NN
NO2
O
EtO
SnCl2.H2O
EtOH
reflux
N
N
H
N
NN
NH2
O
EtO
N
N
H
N
NN
N
H
O
EtO
MsCl, Pyridine
25 °C
78%87%
23% 100%
44%40%
S
OO NO2
B(OH)2
76 77
78
79
8081
82 57
Boronic acid
78
Scheme 5.11 Synthesis of bromosporine from precursor 76.
afforded 4-amino-3,6-dichloropyridazine (74), before hydrazination and cyclisation with
acetic acid (Scheme 5.10).
Attempts to perform cross-coupling reactions with 76 gave only traces of the desired
products.246 To overcome this problem, a protection strategy was adopted whereby the free
amine was carbamoylated. The cross-coupling reaction then proceeded in acceptable yield
to give an arylated product that was deprotected with trifluoroacetic acid (TFA). Following
deprotection, the desired carbamate moiety was installed (81) before the pendant amine was
revealed by a reduction with tin chloride. The methanesulfonyl group was then installed to
provide the desired compound 57 (Scheme 5.11).
116 Development of inhibitors for bromodomain-containing proteins
5.2.2 Proposed synthesis
In order to produce analogues of 57 a more efficient synthetic protocol than the SGC
protocol previously described was attempted. We hypothesised that electron-rich phos-
phine ligands for palladium such as S-Phos or X-Phos247 could make the cross-coupling
more efficient, perhaps without the need for a protecting group; and that use of the com-
pleted sulfonamide as cross-coupling partner would lead to a more convergent synthesis
(Scheme 5.12).
N
N
Cl
H2N
NN
Pd / Ligand N
NH2N
NN
RR
B
OO
76
83 84
Scheme 5.12 Proposed cross-coupling between 76 and a boronic ester.
In addition, the original starting material 3,4,6-trichloropyridazine (73, see Scheme 5.10)
was prohibitively expensive at the time, and so a route based on the more readily available
carboxylic acid derivative 85 was used. There were a number of different options to explore
for this synthesis, depending on the order in which the synthetic steps were to be carried
out (Scheme 5.13).
5.2.3 Synthesis of core
Carboxylic acid route (Route I)
Previous work by Dr Nikzad Nikbin248 had found that protecting the carboxylic acid group
of compound 85 as an ester was difficult and only possible using diazomethane, providing
the methyl ester in low yield. Other protection strategies had been considered, but each
possibility was considered to add considerable complexity to the reaction procedure.
Nevertheless, Route I was very attractive: retention of the pendant carboxyl group would
enable late-stage installation of the carbamate via a Curtius rearrangement (perhaps even
after the cross-coupling reaction had been carried out) for efficient derivatisation at this
position. Unfortunately, although compound 89 was detectable by LC-MS the subsequent
5.2. Chemical probes for BRD9 117
N
N
Cl
Cl
OCN
N
N
HN
Cl
H2N
NH2
N
N
Cl
H2N
NN
N
N
Cl
Cl
HO2C Curtius
N
N
Cl
Cl
H2N
N
N
Cl
Cl
H
NRO
O
ROH
N2H4
N
N
Cl
H
N
NN
Cyclisation
RO
O
N
N
HN
Cl
HO2C
NH2
N
N
Cl
HO2C
NN
N2H4
Cyclisation
N
N
Cl
OCN
NN
Curtius
Route I
Route II
Route III
Carbamate
Formation
N
N
HN
Cl
H
NRO
O
NH2
N2H4
Cyclisation
Hydrolysis
Carbamate
Formation
Route IV
Hydrolysis
Hydrolysis
85 86
R = t-Bu, 87 88
89 R = t-Bu, 90 91
92 R = t-Bu, 77 76
93
Scheme 5.13 Proposed routes to access key intermediate 76 from pyridazine 85.
118 Development of inhibitors for bromodomain-containing proteins
cyclisation procedure proved to be very challenging in the presence of a free carboxylic
acid, particularly with respect to isolation, and no product (92) could be isolated.
Carbamate and amine routes (Routes II and III)
Initial experimentation using microwave irradiation suggested that the Curtius rearrange-
ment to give 86 was an effective reaction; in situ hydrolysis (Route II) was not successful but
reaction with tert-butanol (Route III) to give the Boc product 87 worked well. Unfortunately,
the hydrazine derivative 90 was unreactive to the cyclisation conditions and so hydrolysis
of 87 to 88 had to be carried out before the cyclisation could be performed (Route IV).
Pleasingly, the cyclisation of compound 91 was straightforward, as both intermediate 91
and the product 76 could be isolated in reasonable yield by simple filtration.
Early attempts at cross-coupling reactions with compound 77 indicated that considerable
optimisation would be required, and that the carbamate group was unlikely to survive the
conditions. Thus, more material was to be prepared using Routes III and IV to provide
enough of the key intermediate 76 to test a number of cross-coupling conditions.
Azide formation and Curtius rearrangement in continuous flow
The Curtius rearrangement involves the thermal decomposition of an acyl azide to form an
isocyanate.249 Carboxylic acid starting materials 94 can be converted directly to acyl azides
95 using diphenyl phosphoryl azide (DPPA, 96) as the azide source (Scheme 5.14).250
R OH
O
R N
O
N N
R N
O
N N R
N C
Oheat
- N2
azide 
source
R
H
NNuH Nu
O
94 95 97 98
Scheme 5.14 The Curtius rearrangement transforms an acyl azide (95) to an isocyanate
(97). This can subsequently react with a nucleophile such as water or an alcohol to form
an amide (98).
There are a number of hazards associated with this reaction. DPPA is toxic and explosive,
nitrogen gas is released, and the isocyanate products are also toxic and very reactive. Most
of these problems can be avoided by performing the reaction in continuous flow:19 the small
volume of a flow reactor means that only a small quantity of the DPPA is heated at any one
5.2. Chemical probes for BRD9 119
time; a back-pressure regulator (BPR) can be used to contain the gas that is released; and
the isocyanate can be reacted in situ to avoid holding stocks of it. When an alcohol such as
tert-butanol is added to the reaction mixture this combines with the isocyanate intermediate
in situ to produce a carbamate.
In the flow reactor, a stream of DPPA was combined with a stream of the carboxylic acid,
triethylamine and tert-butanol at a tee mixer (M1;  0.5 mm, ETFE) and then passed into a
series of four heated flow coils (R1;  1 mm, PFA, total volume 40 mL). The efflux passed
through a BPR (100 psi) and then a collection valve V1.
Initial runs encountered problems related to precipitation of a gummy solid during the
reaction. This caused blockages in a number of places, notably in the heating coils, and at
the BPR and the collection valve. Optimisation of the solvent system and flow rates led to
Table 5.15 Development of conditions for the Curtius reactions in flow.
100 psi
120 °C
time waste
M1
R1/R2
V1
n
Toluene / MeCN
Toluene / MeCN
N
N
Cl
Cl
H
Nt-BuO
O
 (0.25 M)
NEt3 (1.0 M)
t-BuOH (1.5 M)
DPPA (0.375 M)
85
87
Entry Solvent ratio Scale Reactor Flow Rate Residence Yield
(Tol. / MeCN) /mmol volume /mL /mL/mina time /min
1248 100 : 0 2b — — 240 0.4 g (77 %)
2248 50 : 50 4 20 0.20 100 0.3 g (36 %)
3 0 : 100 2 10 0.10 100 0.1 g (15 %)
4248 70 : 30 6 20 0.17 120 0.5 g (32 %)
5 90 : 10 20 20 0.17 120 — c
6248 80 : 20 10 20 0.17 120 1.1 g (45 %)
7 80 : 20 40 30 0.17 180 3.2 g (30 %)d
8 70 : 30 20 20 0.17 120 2.1 g (40 %)
9 70 : 30 20e 20 0.17 120 2.6 g (50 %)
10 70 : 30 40e 50 0.36 140 — f
11 70 : 30 40e 50 0.36 140 8.5 g (39 %)g
a Flow rate of each stream.
b Performed in batch, in a sealed vial under microwave irradiation. DPPA was added in three
portions. c Rapid blockage in the BPR. d Blockage after 10 hours.
e Reactor R2 used; valve V1 not present. f Blockage after 17 hours. g Combined yield
from two 40 mmol batches.
120 Development of inhibitors for bromodomain-containing proteins
improvements in the performance of the flow system (Table 5.15, entries 2 – 8), but the
reactor would frequently become blocked after long periods of operation.
To minimise the risks of such blockages, the flow coils R1 were swapped for two larger-
bore alternatives R2 ( 2.5 mm, SS, total volume 50 mL) and valve V1 was removed
(Figure 5.16). The proportion of acetonitrile in the solvent was increased to 30 % which
improved solvation of the precipitates without compromising the yield (Table 5.15, entries
9 – 11). Although the BPR was a frequent location for blockages, it was essential for
preventing the solvent from boiling during the reaction and for containing the evolved N2
gas.
We wanted to run the reaction for long periods of time in order to obtain a large quantity of
100 psi
120 °C, 2.3 h
2 x 25 mL
steel reactor
T-mixer
21.5g 21.3g
M1 R2
computer
residual mass
pressure
temperature
remote
access
 (0.25 M)
NEt3 (1.0 M)
t-BuOH (1.5 M)
7:3 PhMe / MeCN
DPPA (0.375 M)
7:3 PhMe / MeCN
N
N
Cl
Cl
H
Nt-BuO
O
a
85
87
b
Figure 5.16 Curtius reaction with remote monitoring and attempted mass monitoring,
using wide-bore stainless steel heated flow coils. (a) Reactor schematic, showing data
connections to the control computer. (b) Photograph of system.
5.2. Chemical probes for BRD9 121
the carbamate intermediate 87. Under standard control, the increase in pressure created by
a blockage caused the reactor to stop immediately; often the contents of R2 — 40 mL of
reagents — would be lost. Immediate notification in the case of a blockage would mean that
the material already inside the reactor could be collected and cycled again later, conserving
the valuable carboxylic acid starting material 85.
Using the software described in Chapter 2 we created a control sequence to run this reaction
and allow live monitoring of the reactors. Alerts were set to be triggered by elevated
pressures or if the reactor stopped due to an over- or under-pressure. The ability of the
software to stop a procedure mid-sequence and then resume it from the same point enabled
the reactor to be cleaned out when blocked and for the contents to be recovered, before
re-starting the reaction.
Incorporation of balances in flow synthesis
After the reaction was nominally complete it was discovered that a significant quantity of
the solution of pyridazine 85 was remaining. Despite both pumps working as expected
when tested independently before the reaction, the observation of excess reagent suggested
that this pump was not delivering the correct flow rate during the reaction. Digital balances
were used to record the mass of liquid within the reservoirs (Figure 5.16).
The rate of change of the mass remaining in each reservoir could be calculated, and would
remain constant with a consistent flow rate through the pump. The control software would
also be able to make decisions based on the masses (Figure 5.17), such as switching the
pump input valves to solvent when either reagent was about to be used up to avoid infusing
air into the pump or wasting reagent. The software could also make real-time adjustment of
the flow rates to achieve a steady flow rate.
55.2g mass derivative
ow rate
controlled
Figure 5.17 The mass of the reagent was measured by a digital balance. The first
derivative of the measurement should be zero to indicate a constant flow rate. Deviations
in the first derivative could be fed back to the pump to adjust its flow rate.
122 Development of inhibitors for bromodomain-containing proteins
-60
-40
-20
 0
 20
 40
 60
 0  1  2  3  4  5  6  7  8
Ma
ss 
/g
Time /h
Balance 1
Balance 2
a
-6
-5
-4
-3
-2
-1
 0
 1
 0  1  2  3  4  5  6  7  8
M
as
s c
ha
ng
e 
/1
0-
3
Time /h
Balance 1
b
 0
 5
 10
 15
 20
 25
 30
 0  1  2  3  4  5  6  7  8
Pr
es
su
re
 /b
ar
Time /h
Pump 1
Pump 2
c
 0
 4
 8
 12
 16
 20
 24
 5.4  5.5  5.6  5.7  5.8
Pr
es
su
re
 /b
ar
Time /h
Pump 1
Pump 2
d
Figure 5.18 (a) Change in mass of two balances over the course of a reaction. The
difference in zero indicates that they were not equally tared at the start. The slow decrease
after the end of the experiment may suggest evaporation of the residual solvent. (b) The
derivative of the mass has too much noise to be useful. (c) There was a problem with the
pressure of pump B near the end of the reaction. (d) Close-up of pump B’s pressure loss.
Unfortunately, in this system the flow rate was too low to be reliably calculated in real-time
based on the change in mass. The available balances were only accurate to 0.1 g, and so
the differential of the reading produced was very noisy even after being passed through a
smoothing filter (Figure 5.18). The recorded data could be used to calculate an approximate
flow rate over the entire reaction, but were not adequate for real-time decision making.‡
Purification
The carbamate product 87 was found to be retained on silica giving poor recovery. It
was postulated that the reason for this poor recovery may be that the carbamate is lost,
leaving the basic amine, as it is known to be removed under the acidic conditions. However,
the same poor recovery was again observed even using a basic solvent system previously
known to inhibit carbamate deprotection. KP-NH amine-functionalised silica produced by
‡ Pleasingly, later testing with a more accurate balance confirmed that it was possible to visualise a mass
change with sufficient accuracy to be a useful measure at low flow rates.
5.2. Chemical probes for BRD9 123
Biotage AB251 was found to offer a good alternative. KP-NH is a proprietary functionalised
silica, described as acting like a mixture of primary and secondary amines, suitable for
chromatography of amine functionalised materials. This modified silica was found to be
effective for the purification of carbamate 87 from the diphenyl phosphate and other by-
products of the Curtius rearrangement. The Curtius reaction was performed up to 40 mmol
scale over 8 hours to provide 8 g of the product, in 30 – 50 % yield after chromatography,
which was sufficient for the following step.
The introduction of a batch chromatographic purification meant that it would be difficult
to combine this reaction with the following steps without development of an alternative
methodology to trap out the by-products. However, there is still a benefit to performing the
azide formation and Curtius rearrangement reactions in flow, from the reduced hazards of
handling the azide reagent on large scale.
Cyclisation
The desired amine 88 could be unmasked by removing the carbamate from 87 using
aqueous HCl. However, the high solubility of the product 88 in water meant that the
greatest isolated yield was achieved when this reaction was performed with an anhydrous
solution of HCl in dioxane. A precipitate of compound 88 hydrochloride slowly formed;
this was collected by filtration and then re-dissolved in NH3 / MeOH solution. A precipitate
of free amine 88 formed, which was collected and washed with a minimum quantity of
methanol (Scheme 5.19).
N
N
Cl
Cl
H2N
N
N
Cl
Cl
H
Nt-BuO
O
1. HCl/dioxane (30 equiv.)
    25 °C, 24 h
2. NH3, MeOH, 
    25 °C, 30 min
quant. (6.4 g)
87 88
Scheme 5.19 Deprotection of carbamate 87 using acid, and isolation by precipitation
from ammonia in methanol.
Hydrazine 91 was prepared by heating compound 88 with an excess of hydrazine hydrate.
The resulting precipitate provided compound 91 without further purification (Scheme 5.20).
Compound 91 was heated in excess glacial acetic acid to give product 76 as a precipitate,
which was collected by filtration and purified by recrystallization from water (Scheme 5.21).
124 Development of inhibitors for bromodomain-containing proteins
N
N
HN
Cl
H2N
N
N
Cl
Cl
H2N NH2NH2.H2O (30 equiv.)
H2O
100 °C (MW), 3 h
NH2
80% (combined)(4 × 1.5 g)
88 91
Scheme 5.20 Reaction of aminopyridazine 88 with hydrazine hydrate.
N
N
Cl
H2N
N
N
HN
Cl
H2N AcOH (30 equiv.)
120 °C (MW), 5 h
NH2 NN
50%
(after recrystallisation)
(4 × 1.0 g batches)
91 76
Scheme 5.21 Cyclisation of compound 91 with acetic acid to afford intermediate 76.
Development of a flow procedure for these steps was not attempted, because the starting
material was relatively valuable and many of the steps up to 76 involved precipitation
purification which has not yet been effectively developed for small-scale flow reactions.
Attempts to improve solvation of the intermediates by introduction of additional solvents
were found to result in lower conversions and increased difficulty of isolation. In summary,
all of the reactions from 87 to 76 gave acceptable yields at moderate scale using batch
methods, with yields limited mainly by the filtration processes.
5.2.4 Synthesis of boronic ester coupling partners
A small collection of boronic esters were prepared in order to make the desired analogues
of bromosporine (57). Boronic esters are reactive equivalents to boronic acids for Suzuki
couplings, which have the advantage of being crystalline and thus easier to purify and store
than the equivalent boronic acids.252 Boronic esters, like other boronic acid equivalents, are
converted to boronic acids in the basic conditions of the Suzuki coupling.252
Treatment of (3-amino-4-methyl)phenylboronic acid 99 with a slight excess of pinacol gave
complete conversion to the boronic ester (100). Following the removal of the solvent this
could be directly combined with a benzenesulfonyl chloride derivative (101) to form the
5.2. Chemical probes for BRD9 125
Table 5.22 Synthesis of boronic ester coupling partners 102.
B(OH)2
NH2
B
NH2
OO
B
N
H
OO
S
OO
Cl
Pinacol 
(1.1 equiv.)
MeCN
25 °C, 18 h
Pyridine (2 equiv.)
EtOH
25 °C, 18 h
Cl S
OO
Cl
99 100
101
102
101 Product Yielda
2-Cl 102a 68 %
3-Cl 102b 55 %
4-Cl 102c 77 %
3,4-diCl 102d 63 %
a Isolated yield after double recrystalli-
sation from ethanol.
coupling partners (102; Table 5.22). These were purified by recrystallization from ethanol
in reasonable yield over the two steps.
5.2.5 Cross-coupling
The Suzuki reaction
The carbon-carbon bond forming cross coupling reaction has been one of the most important
recent developments in synthetic chemistry.253,254 The ability to join molecules without
using traditional functional group methods created the opportunity to synthesise complex
targets more easily, leading to an explosion in the adoption of this strategy.255
Cross-coupling generally occurs between a molecule bearing an electron-poor carbon
atom covalently bonded to a halogen or an oxygen atom, and a second molecule with an
electron-rich carbon centre bonded to a metal. Of the numerous types of such partnerships
that are possible, the palladium catalysed Suzuki cross-coupling reaction254 between an
organo halide and an aryl boronic acid is one of the most reliable, having a large substrate
scope.255 Importantly, the key boronic acid intermediates have an improved safety profile
when compared to equivalent organotin reagents used in Stille coupling:256 boronic acids
also have a much lower environmental impact and toxicity.256 Furthermore, aryl boronic
126 Development of inhibitors for bromodomain-containing proteins
acids and their derivatives, such as boronic esters,252 boron-MIDA complexes257 and
trifluoroborate salts258 are thermally stable and chemically inert, enabling them to be safely
stored under standard laboratory conditions whilst being suitable for use in further chemical
processing.
Novel ligands for the Suzuki reaction have been developed and many are commercially
available pre-complexed to palladium. Many of these ligands are much less sensitive to air
and moisture than simpler phosphine reagents, thus reducing the requirement for glove-box
manipulation. Their better handling characteristics have led to the development of Suzuki
reaction methodologies that can be performed in air, and even in the presence of water.
The mechanism of the Suzuki reaction (Scheme 5.23)254 is very similar to that of other
metal-catalysed cross-coupling reactions,259 being facilitated by a transition metal moving
between two oxidation states. The electron-rich Pd(0) catalyst is relatively nucleophilic and
can react with an electron poor carbon such as that bonded to a halogen. In this oxidative
addition process, electrons move from the Pd to form a Pd−C bond. The resulting Pd(II)
complex is relatively electrophilic and liable to be attacked by a carbon nucleophile; in
LnPd(0)
Cl
Pd(II) Ln
Cl
Pd(II) Ln
R'O MOR'
R B(OH)2 R B(OH)2
OR'
MOR'
R'O B(OH)2
OR'
Pd(II) L(n-1)
R
R
MCl
oxidative
addition
reductive
elimination
base
organoborate
transmetallation
organoboronic 
acid
aryl
halide
metathesis
L   +
L
Scheme 5.23 Mechanism of the Suzuki cross-coupling reaction.254
5.2. Chemical probes for BRD9 127
this case, a carbon–boron bond. This transmetallation process leaves two carbon atoms
associated to the Pd; their close proximity encourages the formation of a C−C bond. This
is the reductive elimination part of the cycle, and the two electrons released by the bond
formation reduce the Pd centre to Pd(0), ready to participate in another oxidative addition
reaction.
The palladium ligand-bound reagents are often added in an oxidised Pd(II) form, which
must be converted into the active catalyst Pd(0) before being able to participate in the
oxidative addition reaction. This reduction reaction can be performed in situ by the base or
the boronic acid260 (the latter process uses up an equivalent of the boronic acid coupling
partner). Therefore, minimised catalyst loadings are preferred, being more efficient for cost,
for environmental reasons and to avoid increased consumption of the boronic acid.
Heteroaromatic cross-coupling
The coupling system for the desired reaction (Scheme 5.24) was expected to be challenging.
In any aryl ring system, aryl chlorides are the least reactive of the aryl halides towards
oxidative addition, because of the lower polarisation of the C−Cl bond compared to C−Br
or C−I bonds, which reduces the activity of the oxidative addition reaction.256 Free amine
groups compromise Suzuki reactions and are normally protected, because they are thought
to bind to the Pd(II) and shut down the catalytic cycle.261 Finally there is a regioselectivity
issue with a chloride functionality present on both of the coupling partners.
However, we envisaged that the electron-deficient chlorotriazolopyridazine (76) would be
active towards oxidative addition, and that this electron deficiency would provide sufficient
regioselectivity between it and the aryl chloride coupling partner (102). Dialkylbiarylphos-
phine ligands such as S-Phos and X-Phos262 are reported to work well in systems with
N
H
S
OO
Cl
N
N
NN
H2N
Cl
B
N
H
OO
S
OO
Cl
N
N
NN
H2N
Cross-coupling
76 102 103
Scheme 5.24 Proposed cross-coupling reaction.
128 Development of inhibitors for bromodomain-containing proteins
heteroaryl chlorides,247 and even those with free amine groups.261 The cyclohexyl groups
present in these ligands make the phosphine phosphorus electron rich, which has the effect
of making the Pd(0) more active to oxidative addition. They are reported to be sufficiently
bulky to promote reductive elimination, and also to increase the proportion of L1Pd species
(palladium atoms with only one ligand), which are more open to approach by the elec-
trophilic carbon species.262 Finally, the presence of ortho-substituents on the lower ring of
both of these ligands increases the stability and activity of the palladium by preventing the
formation of palladacycles through C−H insertion into the ligand.263
Initial experimentation248 failed to reproduce the cross coupling conditions reported219
for the synthesis of 57 (see Scheme 5.11). Investigation of the reaction components and
optimisation discovered that a Pd2(dba)3 / X-Phos combination gave positive results when
conducted in an n–butanol / water solvent system.248 Alcoholic solvents are thought to be
particularly effective in this reaction because the presence of alkoxide base accelerates the
transmetallation.264 Of the common bases used for Suzuki couplings,262 K3PO4 was found
to be the most promising. It was hypothesised that the increased conversion was due to the
greater basicity of PO4
3– compared to other bases such as CO3
2–.
As suggested earlier, the carbamate protecting group was found not to survive the Suzuki
conditions; in the presence of water at high temperature it is likely to be removed.265
However, some conversion was observed under these conditions and thus investigation of
this reaction was continued with the free amine (Table 5.25).
The temperature proved to be optimal at 100 ◦C; lower temperatures had been found to give
very low conversions, and higher temperatures led to decomposition of the starting material
by protodeboronation to compound 104 (entry 2). The conversion was observed to be highly
dependent on catalyst loading (entries 1, 3 and 4), although the loading should ideally be
kept as low as possible. It was found that the amount of boronic ester 102 had only a small
effect on the conversion; since the boronic ester was difficult to remove by chromatography,
the quantity was reduced to only a 10 % excess for future reactions. The addition of stronger
bases such as KOH or a phosphazene base (BEMP) gave no improvement to the conversion.
Deactivation of catalyst
Previous work had shown that 24 – 44 % of the cross-coupling product was obtained when
the amine substituent was not present in coupling partner 76.230 We postulated that the poor
5.2. Chemical probes for BRD9 129
Table 5.25 Investigation of the effect of temperature, reaction time and catalyst loading
on the Suzuki cross-coupling reaction.
N
H
S R
OO
Pd2(dba)3
X-Phos
K3PO4 (2 equiv.)
7:3 n-BuOH / H2O
MW
N
N
NN
H2N
Cl
B
N
H
OO
S R
OO
N
N
NN
H2N
N
H
S R
OO
76 102 103
104
Entry R Equiv. Equiv. Equiv. Temp. Time Product Conv. Conv.102 Pd2(dba)3 ligand /
◦C /h to 103a to 104b
1 2-ClC6H4 1.5 10 % 20 % 100 4 103b 60 % 25 %
2 2-ClC6H4 1.5 10 % 20 % 120 4 103b 65 % 45 %
3 2-ClC6H4 1.5 20 % 40 % 100 4 103b 35 % 20 %
4 2-ClC6H4 1.5 1 % 2 % 100 8 103b 80 % 30 %
5 4-ClC6H4 1.5 10 % 20 % 100 4 103d 60 % 30 %
6 4-ClC6H4 1.1 10 % 20 % 100 4 103d 65 % 30 %
7248 Me 1.5 50 %c 100 % 105 3 103a 100 %d n/ae
a Conversion measured by LC-MS; relative peak areas for compounds 103 and 76.
b Conversion measured by LC-MS; relative peak areas for compounds 104 and 102.
c 4 equiv. K3PO4 used.
d 10 % isolated by column chromatography on KP-NH silica.
e The majority of the remaining boronic ester was deborylated.
conversions seen in Table 5.25 were due to deactivation of the palladium catalyst by one of
the reaction components, which was supported by the formation of palladium precipitates
during the reaction. Indeed, when the reaction was performed using a stoichiometric
quantity of the palladium catalyst full conversion was observed, but the product was very
difficult to isolate (Table 5.25, entry 7).
Catalyst deactivation is likely to be caused by poisoning or sequestration, possibly by
the primary amine261 or by a bidentate interaction with the amine and the triazole. The
monophosphine ligand X-Phos is reported to give good results even with heteroaryl chlo-
rides which have free amine substituents,247 which would suggest that the X-Phos ligand
has a particularly strong interaction with the palladium centre.
It was noticed that a large amount of material was retained by the silica gel during chromato-
graphic purification, and this might indicate that polar complexes between the palladium
130 Development of inhibitors for bromodomain-containing proteins
and either the starting material or product were forming. However, attempts to break any
such complexes by stirring in acid conditions prior to chromatography were unsuccessful.
Alternative ligands
Therefore, despite the positive reports of using X-Phos in this kind of system,245 we
postulated that a bidentate ligand might be held more strongly to the metal centre and resist
catalyst deactivation. Since the substrates 76 and 102 are relatively low in steric bulk (they
do not have any substituents in an ortho position) we anticipated that a bidentate ligand
would not have a negative effect on the rate of oxidative addition.
The reaction conditions were optimised using the methyl-derivative boronic ester (102e)
which was not expected to give cross-coupling side reactions. During these investigations
(Table 5.27) only starting materials, product 103a and homocoupled product 105 were
observed.
Improved degassing of the solvent prior to the reaction appeared to give superior results
(Table 5.27, entries 2 and 4). Slightly increased conversion and lower homocoupling
was observed when using DPE-Phos (Figure 5.26) as a ligand (entry 3), and even further
improvement when using DtBPF (entry 5), which is likely to have higher activity due to
the greater electron density on the phosphorus atoms. The homocoupling product was very
difficult to remove from the reaction mixture, and so attempts were made to reduce its
formation. A more rapid reaction turnover may disfavour the competing homocoupling
process.
Reducing the catalyst loading compromised the conversion (entry 6), and increasing the
equivalence of boronic ester present had minimal effect (entry 7). However, increasing the
proportion of n-BuOH in the solvent mixture gave lower homocoupling (entry 8).
P
P
Fe
PhPh
PhPh
Pd(II)Cl2
PdCl2.dppf
P
P
Fe
t-But-Bu
t-But-Bu
Pd(II)Cl2
PdCl2.DtBPF
O
P
P
Pd(II)Cl2
Ph
Ph
Ph
Ph
PdCl2.DPEPhos
Figure 5.26 Palladium complexes with bidentate ligands employed in the Suzuki reaction.
5.2. Chemical probes for BRD9 131
Table 5.27 Investigation of the use of different bidentate ligands for the Suzuki cross-
coupling reaction.
N
H
S
OO
Catalyst
K3PO4 (4 equiv.)
x:y n-BuOH / H2O
100 °C, time
N
N
NN
H2N
Cl
B
N
H
OO
S
OO
N
N
NN
H2N
N
H
S
OO
H
N S
OO
76 102e 103a 105
Entry Equiv. Catalyst (equiv.) a Solv. Ratio Time Conversionb Homoc.b
102e (x:y) /h to 39 (105)
1 1.1 Pd/dppf (0.1) 7 : 3 4 20 % 100 %
2 1.1 Pd/dppf (0.1)c 7 : 3 4 60 % 70 %
3 1.1 Pd/DPE-Phos (0.1) 7 : 3 2 80 % 60 %
4 1.1 Pd/DPE-Phos (0.1)c 7 : 3 2 90 % 50 %
5 1.1 Pd/DtBPF (0.10)c 7 : 3 4 90 % (64 %)d 40 %
6 1.1 Pd/DtBPF (0.05)c 7 : 3 4 80 % 50 %
7 2 Pd/DtBPF (0.10)c 7 : 3 4 60 % 30 %
8 1.1 Pd/DtBPF (0.10)c 9 : 1 4 60 % 20 %
9 1.1 Pd/DtBPF (0.10)c 7 : 3 2 50 % 50 %
10 1.1 Pd/DtBPF (0.10)c 1 : 0 2 0 % n/a
11 1.1 Pd/DtBPF (0.10)c dioxane 2 0 % n/a
a Precatalysts: Pd/dppf = PdCl2 · dppf ·CH2Cl2; Pd/DPE-Phos = PdCl2 ·DPE−Phos;
Pd/DtBPF = PdCl2 ·DtBPF.
b Approximate consumption of compound 76 / generation of homocoupled compound
102; as determined using LC-MS peak areas.
c Freeze-dry degassing technique used (instead of standard nitrogen-bubbling technique).
d Isolated by preparative HPLC.
The optimisation studies indicated that, PdCl2 ·DtBPF was the most successful cata-
lyst/ligand combination, and that a reduction in the amount of water present was beneficial.
Although the presence of water is known to be important in Suzuki processes — it can be
involved in in the hydrolysis of the boronic ester and formation of the reactive boronate
species — DtBPF was reported to have been used in anhydrous dioxane, specifically in
order to avoid side-reactions caused by water.266 However, no reaction was observed when
anhydrous n-butanol (entry 10) or dioxane (entry 11) were used as solvents.
132 Development of inhibitors for bromodomain-containing proteins
Purification
The optimised cross-coupling conditions using PdCl2 ·DtBPF in n-butanol / water were
used to obtain compounds 103 in moderate yields. The original hypothesis that the Suzuki
cross-coupling could be performed selectively in the presence of aryl chlorides appeared to
be mostly correct, but although the desired product tended to be the major one there were
many by-products that made isolation very challenging.
Table 5.28 Suzuki cross-coupling reactions to provide intermediates 103.
N
H
S R
OO
PdCl2.DtBPF (0.1 equiv.)
K3PO4 (4 equiv.)
9:1 n-BuOH / H2O
100 °C, 4 h
N
N
NN
H2N
Cl
B
N
H
OO
S R
OO
N
N
NN
H2N
(1.5 equiv.)
76 102 103
Entry R Compound Yielda
1 Me 103a 64 %
2 (2-Cl)C6H4 103b 49 %
3 (3-Cl)C6H4 103c 19 %
4 (4-Cl)C6H4 103d 30 %
5 (3,4-Cl)C6H3 103e 13 %
a Isolated yield after preparative HPLC.
Column chromatography using KP-NH silica had previously been found to consistently
enable separation of compounds with a primary amine substituent when normal-phase
silica failed. However, although isolation of compounds 103 by column chromatography
was possible, it was found to be more efficient to separate the whole mixture using prepar-
ative HPLC, which provided material of high purity for the final step to be investigated.
Compound 103e was notably difficult to purify, as a result of the many minor by-products
present in the mixture. These appeared to be the products of unwanted secondary Suzuki
coupling reactions, as evidenced by 1H NMR and LC-MS.
5.2. Chemical probes for BRD9 133
5.2.6 Carbamate formation
Combination of the Suzuki products 103b–e with 2 – 4 equivalents of ethyl chloroformate
in the presence of DMAP gave good conversion to one major product, as observed by
LC-MS (Table 5.29). Nevertheless, further purification by preparative HPLC was required
to give sufficiently pure samples for NMR analysis.
Interestingly, when THF was used as the solvent, product 108 was obtained; whereas when
the reaction was performed in DCM, double addition of the carbamate moiety occurred to
provide compound 109.
Table 5.29 Installation of carbamate moieties onto compounds 103b–e. The desired
product (107), with the carbamate present only on the heterocyclic amine, was not ob-
served in any case.
Ethyl chloroformate____
DMAP
NEt3
Solvent
25 °CN
H
S
OO
N
N
NN
H2N
R
Product
N
H
S
OO
N
N
NNH
N
R
O
EtO
N S
OO
N
N
NNH
N
R
O
EtO
N S
OO
N
N
NN
H2N
R
OEtO OEtO
103
(106)
107 108 109
R 103 Solvent Equiv. base Equiv. 106 Product Mass (Yield)
2-Cl 103b THF 10 2 108b <1 mg a,b
3-Cl 103c THF 10 2 108c <1 mg a,b
4-Cl 103d THF 2 2 108d 11 mg (95 %)c
3,4-Cl 103e DCM 2 4 109e 2 mg (38 %)a
a Isolated yield following preparative HPLC.
b Full conversion was observed by LC-MS, but the reaction scale was too small to
record an accurate yield.
c Isolated by solvent extraction.
134 Development of inhibitors for bromodomain-containing proteins
This result was surprising: in particular, we had expected the sulfonamide NH to be
deactivated with respect to the aromatic amine, and indeed the relative chemical shifts of
these two environments in the NMR spectra of compound 103b–e suggested that this was
the case. However, NMR analysis of the products of this reaction (Figure 5.30 shows 103c
as an example) indicated that it was the sulfonamide that had in fact reacted first.
5.2.7 Alternative route
The alternative route, as reported for bromosporine (57),219 whereby the carbamate group is
installed whilst the aryl amine is masked as a nitro moiety, before reduction and sulfonamide
formation, turned out to be even less efficient (Scheme 5.31). The amine seemed to be
non-nucleophilic and almost completely deactivated, and only very low conversions were
observed upon treatment with ethyl chloroformate. The starting material had to be cycled
multiple times through this process in order to obtain useful quantities of the desired
carbamate.
NO2
NEt3 (10 equiv.)
DMAP (1 equiv.)
THF, 25 °C, 30 min
N
N
NNH
N
(1.5 equiv.)
NO2
N
N
NN
H2N
Cl
O
OEt
O
EtO
trace
80
106
81
Scheme 5.31 Carbamate formation from nitro analogue 80 was unsuccessful. The
pendant amine seemed to be very deactivated in the presence of the nitro group.
Experiments conducted using the methylsulfonamide derivative 103a suggested that carba-
mate formation took place on the sulfonamide nitrogen rather than on the heteroaryl amine,
even when multiple equivalents of a strong base such as sodium hydride were added to
attempt to activate this position.267
The carbamoylation provided four additional compounds which were suitable subjects for
assessing their performance using the Frontal Affinity Chromatography (FAC) and other
biological assay techniques.
5.2. Chemical probes for BRD9 135
0.
5
1.
0
1.
5
2.
0
2.
5
3.
0
3.
5
4.
0
4.
5
5.
0
5.
5
6.
0
6.
5
7.
0
7.
5
8.
0
8.
5
f1
 (p
pm
)
2.67
2.63
1.95
1.93
0.95
0.96
0.87
1.75
0.89
0.88
0.86
2.91
2.88
2.02
1.02
0.97
1.09
1.19
1.03
0.89
0.84
0.96
3.05
1.05
0.99
0.97
3.81
0.94
1H
1H
1H
 (H
et
A
r-
H
)
2H 2H
2H
 (r
ot
am
er
s)
B
N H
SO
O
O
O
C
l
NNN
N
H
2N
N H
SO
O
C
l
NNN
N
H
2N
N
SO
O
O
O
C
l
Tr
ac
e 
1
Tr
ac
e 
2
Tr
ac
e 
3
10
2b
10
3c
10
8c
Fi
gu
re
5.
30
1 H
N
M
R
sp
ec
tra
of
10
2b
(tr
ac
e
1)
,1
03
c
(tr
ac
e
2)
an
d
10
8c
(tr
ac
e
3)
sh
ow
in
g
ch
an
ge
s
in
th
e
ch
em
ic
al
sh
ift
s
of
th
e
N
H
pr
ot
on
s
th
at
su
gg
es
tt
ha
tt
he
ca
rb
am
at
e
m
oi
et
y
w
as
in
st
al
le
d
on
to
th
e
su
lfo
na
m
id
e
am
in
e.
136 Development of inhibitors for bromodomain-containing proteins
5.2.8 Biological assessment
Compounds 103, 108b–d and 109e were assessed for binding activity with bromodomain
proteins, including a FAC assay using BRD9.
The FAC column was prepared by filling a 15 µL guard cartridge with biotinylated BRD9
protein immobilised onto streptavidin-coated beads. Column performance was validated by
injecting samples of bromosporine (57) and analogue 110 at different concentrations.230
The quantity of BRD9 units available for binding was determined to be approximately
2 nmol, and the trend and magnitude of the affinity constants of 57 and 110 corresponded
well with reported IC50 data.
219
N
N
NNH
N
N
H
S
OO
EtO
O N
N
NNH
N
O
t-BuO
57 110
Figure 5.32 Compounds used for validating the BRD9 FAC column.
A disadvantage of a chromatographic method is that there is the possibility of binding
by non-specific interactions, such as those involving the streptavidin units or the polymer
support onto which the BRD9 is loaded. In particular, streptavidin exists as a tetramer, with
four biotin-binding positions per unit. The bulk of the BRD9 protein bound in one site is
likely to block the other sites from being occupied, so there are spare binding sites available
to interact with the test compounds.
To determine the effect of non-specific interactions, bromosporine (57) and the void marker
111 were run down a column loaded with streptavidin-coated beads. No significant dif-
ference in the retention times were observed between compounds 57 and 111, confirming
that the retention times observed during the initial validation were caused primarily by
interactions with BRD9.
Void compound 111 is not UV active, but can be be detected using a mass spectrometer.
The use of a mass spectrometer enables a wider range of compounds to be detected and
thus tested, and has greater sensitivity compared to a UV detector; however, it also imposes
some restrictions on the analysis method. In particular, a volatile buffer had to be used to
5.2. Chemical probes for BRD9 137
avoid filling the nebulisation chamber with salt deposits. Fortunately, the BRD9 column
was found to be stable and functional in the presence of a volatile ammonium acetate buffer.
HO2C
N
H
O N
N
NN
CF3 N
N
NN
CF3
OMe
N
N
NNH
N
N
H
S
OO
EtO
O
111
57 112 113
Figure 5.33 Additional compounds used during the FAC experiments.
Compounds 103b,c,e, 108b–d and 109e, along with bromosporine (57) and analogues
112268 and 113,268 were injected separately into the BRD9 FAC-MS system. By performing
single injections with each compound at the same concentration (Figure 5.35), a ranking of
affinities was constructed by observing the relative retention times within the functionalised
column. These results were compared with thermal shift assays performed against a number
of other proteins spread across the BRD phylogenetic tree (Table 5.34).230,246
Table 5.34 Tm shift data collected230 for compounds across a suite of bromodomain-
containing proteins.
Protein 57 112 113 103b 103e 108b 108c 108d 108e 109e
BRD9 (∆Tm,obs) +8.8 -0.6 +0.6 +2.4 +4.3 +4.4 n.d. +2.7 +1.7 +1.1
BRD1 (∆Tm,obs) +2.5 -1.6 +0.0 +2.3 +3.7 -1.1 +1.8 +2.0 +2.0 +1.5
BRD4A (∆Tm,obs) +6.6 -3.5 -1.3 +6.5 +6.7 +6.2 +5.6 +4.3 +1.8 +3.2
BRPF1A (∆Tm,obs) +0.6 n.d. n.d. +1.0 +5.4 n.d. n.d. +1.7 +1.8 -0.3
BRPF3 (∆Tm,obs) +2.1 n.d. n.d. +1.1 +3.1 n.d. n.d. +1.6 +0.4 +0.0
CECR2 (∆Tm,obs) +8.3 -7.4 -0.1 +3.0 +5.6 n.d. n.d. -10.9 +1.7 +1.6
CREBBP (∆Tm,obs) +3.8 -2.1 -0.9 +4.1 +4.3 +3.5 +4.7 +4.6 +2.0 +3.6
EP300A (∆Tm,obs) n.d. +1.3 +1.3 +3.5 +8.0 +4.2 +4.1 +4.3 +2.5 +3.7
TIF1 (∆Tm,obs) +0.9 -1.7 -0.5 +1.3 +1.8 n.d. +0.2 +1.2 +0.9 +0.5
∆Tm,obs = Difference in temperature of denaturation of the protein with and without the
ligand (at 10 µM), data in ◦C; n.d. = not determined.
138 Development of inhibitors for bromodomain-containing proteins
0.0
0.5
1.0
1.5
2.0
2.5
0
10
20
30
40
50
60
Ion Count /106
Tim
e /m
in
N N
H
2 N
N
N
NH
S O
O
C
l
C
l
N N
H
2 N
N
N
NH
S O
O
C
l
N N
H
2 N
N
N
NH
S O
O
C
l
N N N
N
N
S O
O
E
tO
O
C
l
H
2 N
<
<
<
N N
H
2 N
N
N
N
S O
O
C
l
E
tO
O
N
H
A
c
C
O
O
H
N N
H
2 N
N
NCl
A
B
C
D
E
F
G
A
B
C
D
E
F
G
111
76
108b
103b
103c
108d
103e
Figure
5.35
FA
C
-M
S
analysis
ofcom
pounds
76,103
and
108
using
a
B
R
D
9-functionalised
colum
n.C
om
pounds
112
and
113
eluted
betw
een
com
pounds
111
and
76,and
are
notshow
n.
B
rom
osporine
(57)
did
notelute
during
the
120
m
inute
run,due
to
its
strong
affinity
forB
R
D
9.
5.2. Chemical probes for BRD9 139
5.2.9 Evaluation
The test compounds were generally found to have reduced activity for BRD9. Interestingly,
compound 103e had increased selectivity for BRPF1A, CECR2 and EP300A, and com-
pounds 103b and 108b retained activity for BRD4 whilst losing their activity for BRD9
and CECR2. Most importantly, the observed trend in the FAC retention times of the test
compound and standard correlated very well with the measured Tm shift of the same com-
pounds with BRD9 (Table 5.36), which gave us additional confidence in the use of this
technique for rapid assessment of new compounds.
Table 5.36 Comparison of FAC results with Tm shift data.
Compound ∆Tm,obs (BRD9) FAC rankinga
57 +8.8 1
103e +4.3 2
108d +2.7 3
103c n.d. 4
103b +2.4 5
108b +4.4 6
108e +1.7 n.d.
109e +1.1 n.d.
113 +0.6 7
112 -0.6 8
a The FAC ranking was obtained by visual iden-
tification of the curve fronts of the FAC-MS
traces (Figure 5.35) and ordering them from
the latest to the earliest.
By tuning the column size it should be possible to analyse compounds with different ranges
of affinities within a reasonable time. The 15 µL column with 2 nmol binding capacity was
found to be able to distinguish between low-to-medium affinity substrates (equivalent to
∆Tm,obs between 0 and 5 ◦C), so it may be useful as a technique for fragment-based drug
discovery programmes.
140 Development of inhibitors for bromodomain-containing proteins
5.3 Chemical probes for CECR2
Analogous compounds prepared as part of the research programme were included in a
screening assay against a range of bromodomain proteins.246 Compounds with a “reversed”
sulfonamide moiety (compounds 115–118, Table 5.38) exhibited increased selectivity for
CECR2, another bromodomain protein (Figure 5.37).
The identification of these compounds as being selective for CECR2 was used to inspire
the design and synthesis of a small array of similar compounds. Importantly, the pendant
amine or carbamate chain appeared to be less important for CECR2 binding (compounds
115–119 have increased potency for CECR2 compared to BRD4A), and so we decided to
a b
c
Figure 5.37 (a) Crystal structure of bromosporine (57) bound into BRD4A(1). (b) Bromo-
sporine docked into CECR2, calculated using Schrödinger Glide.248 (c) Calculated binding
interactions between bromosporine (57) and CECR2. Bromosporine is calculated to make
similar interactions with CECR2 as with BRD4A(1), including between the northern triazole
and an asparagine residue, and between the pendant sulfonamide and an aspartic acid
residue.
5.3. Chemical probes for CECR2 141
Table 5.38 Structure-activity relationship data for a number of compounds as determined
by SGC experiments. A red background indicates strong binding, green indicates weak
binding.
Compound PotencyBRD9 CREBBP CECR2 BRD4A BRD2
57
N
N
NNH
N
N
H
S
OO
EtO
O
0.12 a 0.16 b 0.02 a 0.20 a 0.41 a
114
N
N
NNH
N
NO2
EtO
O
1.18 a 10000 b 0.04 a 35.9 a n.d.
115
N
N
NN
S N
O O
7.14 c 170.6 a 0.05 b 10000 a n.d.
116
N
N
NN
S N
O O
n.d. 10000 a 0.36 a 8.91 c n.d.
117
N
N
NN
S N
O O
22.9 c 10000 a 0.50 b 10000 a n.d.
118
N
N
NN
S N
O O
17.8 c 10000 a 15.9 b n.d. n.d.
119
N
N
NN
N
H
S
OO
13.6 c 10000 a 69.5 b 10000 a n.d.
Data are given as IC50, measured or estimated by:
a AlphaScreen;
b AlphaScreen (single concentration); c Tm shift (estimated).
142 Development of inhibitors for bromodomain-containing proteins
focus initially on deaminated analogues of 57. Since this amine group was thought to be
the main source of problems with the cross-coupling step, we anticipated that this strategy
might enable more rapid access to the compounds.
This targeted array would test how well the trend in the binding affinities of the deaminated
analogues correlated with that of the aminated analogues, providing further confirmation
of whether FAC is an effective technique for providing data on fragments as well as final
compounds.
5.3.1 “Reversed” sulfonamide boronic ester
Previous work had investigated chlorosulfonylation of aminotolueneboronic ester 114269,270
or tolueneboronic ester 114271,272 (Scheme 5.39).273 Unfortunately, these attempts to install
the sulfonyl group onto a phenyl boronic ester proved to be unfruitful, thus a different
strategy for their preparation was investigated whereby the boronic acid group was installed
onto a phenylsulfonamide.
BPin
X
BPin
S Cl
O O
BPin
S NR
1R2
O O
HNR1R21. Sulfonylation
2. Chlorination
114a; X = NH2
114b; X = H
115 116
Scheme 5.39 Unsuccessful chlorosulfonylation of phenylboronic esters.
An effective method was the Miyaura borylation274 of bromobenzylsulfonamides 118
prepared by the treatment of sulfonyl chlorides 117 with cyclic amines (Scheme 5.40).
S Cl
OO
HNR1R2
Br
S NR1R2
OO
Br
Borylation S NR1R2
OO
PinB
117 118 119
Scheme 5.40 Access to boronic esters 119 by Miyaura borylation of sulfonamides 118.
The sulfonamides 118 were generated smoothly by treatment of a sulfonyl chloride 117
with a secondary amine at room temperature. This reaction was found to have a good
solvent tolerance, with the highest solubility of the starting material 117 observed in CPME,
5.3. Chemical probes for CECR2 143
DME or n-butanol. Compounds 117 were treated with polymer-supported amine base (QP-
DMA) before addition of the amine to sequester any traces of hydrolysed sulfonyl chloride
which would be unreactive under these conditions. With this route in hand, borylation of
the sulfonamides 118 was investigated.
5.3.2 Miyaura borylation
In 1995, Miyaura and coworkers reported274 that tetra(alkoxo)diborons such as bis(pina-
colato)diborane 120 can act as a boron nucleophile for cross-coupling with organic elec-
trophiles in the presence of a base.
LnPd(0)
Pd(II)
R
Ln
Cl
Pd(II)
R
Ln
AcO
KOAc
Pd(II)
R
L(n-1)
PinB
BPin
KCl
oxidative
addition
reductive
elimination
base
transmetallation
aryl halide
metathesis
L
B B
O
O O
O
AcO B
O
O
Pd(II)Cl
R ClR
(PPh3)n-1
PPh2R
BPin
L = PPh3
+  L
120
121
Scheme 5.41 Mechanism of the Miyaura borylation reaction.274
The accepted mechanism (Scheme 5.41) is analogous to the Suzuki-Miyaura cross coupling
reaction,274 and includes the critical presence of a base such as potassium acetate to generate
an activated intermediate which undergoes transmetallation with the diboron. The use of
KOAc is postulated to disfavour homocoupling products; stronger bases such as K2CO3 or
K3PO4 lead to the formation of biaryls by further coupling of the transmetallated palladium
boronate intermediate with the aryl halide.
In cases where electron-poor or electron-rich aryl halides are used, slow transmetallation
can permit a competing pathway involving aryl exchange with the phosphine ligand to
144 Development of inhibitors for bromodomain-containing proteins
afford small quantities of a phenylboronate by-product (121).275
Borylation is reported to be accelerated by polar solvents such as DMSO or DMF, with
slower reactions in dioxane or toluene.274 Considering the difficulty of removing reaction
products from such polar solvents, alternative solvents were considered. CPME was
selected as the first solvent to be evaluated, because it had been found to be useful in the
previous step, a beneficial feature potentially allowing telescoping of steps. CPME is more
polar than other commonly used solvents such as dioxane (with a molecular dipole moment
of 1.27 D compared to 0.45 D for 1,4-dioxane276), and thus functionalisation using this
solvent was investigated first (Table 5.42).
Under the first set of conditions attempted, the reaction was found to proceed with good
conversion to the desired product (123a) (entry 1). However, the formation of an inorganic
Table 5.42 Investigation of ligand systems for Miyaura borylation.
S N
OO
O
B2Pin2
Catalyst
KOAc (3 equiv.)
Solvent
Temperature, Time
B
S N
OO
O
SN
O O
O
O O
2
Br
122a 123a 124
Entry Catalysta Equiv. Solvent (v/v) Temp. Time Conv.
b Homoc.b
B2Pin2 /◦C /h to 123a (124)
1 Pd/dppf 3 CPME 110 3 100 %c 0 %
2 Pd/XP 3 CPME 110 3 100 %c 0 %
3 Pd/XP 1.5 CPME 110 3 100 %c 0 %
4 Pd/XP 1.5 CPME 110 1 100 %c 0 %
5 Pd/XP 1.5 8 : 1 CPME / H2O 110 1 25 % 0 %
6 Pd/XP 1.5 n-BuOH 110 1 100 % 15 %
7 Pd/XP 1.5 9 : 1 n-BuOH / H2O 110 1 100 % 25 %
8 Pd/PEPPSI 1.1 n-BuOH 110 1 0 % –
9 Pd/PEPPSId 1.1 n-BuOH 110 1 100 % 65 %
10 Pd/XP 1.1 n-BuOH 150 1 95 %e 20 %
a Catalysts: Pd/XP = Pd2(dba)3 ·CHCl3 (2 mol%), X-Phos (4 mol%); Pd/dppf = dppf ·PdCl2 ·
CH2Cl2 (4 mol%); PD/PEPPSI = PEPPSI−IPr ·PdCl2 (3 mol%).
b As determined using LC-MS peak areas.
c Formation of precipitate observed.
d 5% catalyst loading.
e 12% isolated yield (column chromatography).
5.3. Chemical probes for CECR2 145
precipitate, likely to be from the palladium or the KOAc, was also observed. Similar results
were observed using X-Phos as a ligand (entry 2); in this case HPLC analysis suggested a
cleaner reaction. Reduction of the amount of bis(pinacolato)diborane (120) added to the
reaction had no significant effect on the conversion (entry 3). A shorter time of 1 hour was
equally effective (entry 4).
We wanted to investigate the possibility for performing the three-step borylation process
in continuous flow as this would enable the use of automation to generate an array of
coupling partners. For this reason, we investigated further conditions to attempt to reduce
the processing time and also avoid the generation of a precipitate, since this would be highly
undesirable within a heated flow reactor coil.
The precipitate observed in entries 1 – 4 was found to be water soluble; unfortunately,
inclusion of water in the solvent mixture to give a CPME / water ratio of 8/,:/,1 (v/v) did
not promote the reaction, possibly due to the low miscibility of these solvents leading to
sequestration of the base in the aqueous phase (entry 5). n-Butanol is more polar (1.66 D)
than CPME, and is often recommended for Suzuki-type reactions; this solvent was good
in both the sulfonamide formation and Suzuki steps, allowing the possibility for reaction
telescoping. However, in this case the use of n-butanol tended to lead to the generation of
the homocoupled product (124) (entries 6 – 7).
Oxygen is postulated to be involved in homocoupling processes.277 However, oxgyen was
unlikely to be present under the reaction conditions because the same degassing protocol
was followed as for the Suzuki reactions in Section 5.2.5, which had been found to remove
almost all of the oxygen. It was likely therefore that the homocoupling product was formed
due to a slow transmetallation step in the mechanism.
In a limited catalyst and ligand screen, PEPPSI-IPr (Figure 5.43) was included as it is
reported to be an effective ligand for borylation,278 and is in general a highly active
P
P
Fe
PhPh
PhPh
Pd(II)Cl2 Pd
NN
iPr
iPr iPr
iPr
N
ClCl
Cl
PdCl2.dppf PEPPSI-IPr
Figure 5.43 Palladium complexes employed in the Miyaura borylation reaction.
146 Development of inhibitors for bromodomain-containing proteins
ligand for Suzuki-type processes.279 N-Heterocyclic carbene ligands are in general stronger
electron donors than phosphine ligands;280 however, the use of PEPPSI-IPr showed no
advantage for this system (entries 8 – 9).
In a similar, related system, a limited screen of solvents and bases was conducted (Ta-
ble 5.42), using dppf (Figure 5.43) as a ligand because it had been observed to keep the
palladium complex to in solution throughout the reaction (Table 5.42, entry 1).
Using an n-BuOH / MeCN solvent system, the sulfonamide (125a) remained in solution.
Full conversion was obtained with low degree of homocoupling (Figure 5.43, entry 1);
however, the isolated yield of compound 126a was only 53 %. Considering the bases,
triethylamine has been reported as an effective base for the Miyaura borylation281 and
since it is a liquid there is no concern of solubility. However, the use of NEt3 led to a
reduced conversion as measured by LC-MS even though there was little effect on the
quantity of 126a isolated (entry 2). Furthermore, although the use of NEt3 enabled the use
of different solvent systems whilst maintaining a homogeneous reaction mixture, there was
no significant improvement in the isolated yield (entries 3 – 5).
Table 5.44 Investigation of Miyaura borylation parameters using PdCl2(dppf) ·CH2Cl2
catalyst complex.
Br
S N
OO
O
B2Pin2 (1.1 eqiv.)
PdCl2(dppf).CH2Cl2
Base (3 equiv.)
Solvent
Temperature, Time
B
S N
OO
O
SN
O O
O
O
O
2
125a 126a 127
Entry Base Solvent (v/v) Temp. Time Conv.a Homoc.
a Yieldb
/◦C /min (127) of 126a
1 KOAc 2:1 n-BuOH / MeCN 150 60 100 % 20 % 53 %
2 NEt3 2:1 n-BuOH / MeCN 150 60 50 % 0 % 54 %
3 NEt3 MeCN 150 60 35 % 0 % 50 %
4 NEt3 1:1 CPME / MeCN 150 60 40 % 0 % 58 %
5 NEt3 8:2 DME / H2O 150 60 100 % 25 % 53 %
6 KOAc 2:2:1 n-BuOH / MeCN / H2O 150 30 100 % 15 % 52 %
7 KOAc 1:1 EtOH / H2O 150 10 100 % 15 % 58 %
8 KOAc 1:1 MeCN / H2O 150 10 100 % 45 % 24 %
a Approximate consumption of 125a / generation of 127; peak areas measured by LC-MS.
b Isolated yield, following column chromatography.
5.3. Chemical probes for CECR2 147
The use of water as an additive to the n-BuOH / MeCN solvent system was evaluated as
it allowed better solvation of the inorganic base. The results indicate that there was no
detrimental effect on the conversion or yield, and in this case the reaction was complete
within 30 minutes as evidenced by LC-MS (entry 6).
Monophasic aqueous solvent systems were also investigated, because these may have led to
more favourable continuous flow conditions. Both ethanol / water and acetonitrile / water
solvent systems led to full consumption of the starting material 125a within 10 minutes
as observed by LC-MS, but without significant improvement in the isolated yield. The
homocoupling product (127) was difficult to separate from the desired product by column
chromatography, leading to the low isolated yields. Therefore, in the interests of time, we
decided to progress with a batch-mode synthesis method using the original KOAc / CPME
conditions, since these had been found to produce the smallest quantity of by-products.
5.3.3 Suzuki cross-coupling
Initial screening of reaction conditions using the procedure previously investigated for the
preparation of the BRD9 probes (see Section 5.2.5) indicated very high and clean conversion
as observed by LC-MS, but only a trace of compound 128a was isolated after purification
by preparative HPLC (Scheme 5.45). The poor recovery and isolation of compound 128a
indicated that purification was the most significant problem for these compounds.
Pd2(dba)3 (10 mol%)
X-Phos (20 mol%)
K3PO4 (4 equiv.)
4:1 n-BuOH / H2O
100 °C, 2 h
B
S N
OO
O
S N
OO
OO
O
N
N
NN
Cl
NN
N
N
(< 1 mg)
129 126a 128a
Scheme 5.45 Initial attempt at cross-coupling between 129 and boronic ester 126a.
The mass of the product recovered after preparative HPLC was too low for accurate
measurement.
One-pot borylation and cross-coupling reactions are precedented in the literature,282 so an
attempt at telescoping the clean Miyaura borylation reaction with the Suzuki cross-coupling
reaction was attempted. Pleasingly, this methodology allowed rapid access to an array
148 Development of inhibitors for bromodomain-containing proteins
Br
S NR
1R2
O O
N
N
NN
3 steps
S Cl
O O
4-substituted, 130
3-substituted, 131
4-substituted, 128
3-substituted, 132
Scheme 5.46 One-pot synthesis of compounds 128 and 132. Conditions: Step 1.
HNR1R2 (1.1 equiv.), 8 : 1 n-BuOH / water, 25 ◦C, 5 min. Step 2. B2Pin2 (1.1 equiv.),
Pd2(dba)3 (0.02 equiv.), X-Phos (0.04 equiv.), KOAc (3 equiv.), CPME, 110
◦C, 1 h. Step 3.
129 (1 equiv.), Pd2(dba)3 (0.2 equiv.), X-Phos (0.4 equiv.), K3PO4 (3 equiv.), 4 : 1 n-BuOH
/ water, 100 ◦C, 2 h.
of compounds; after each step the solvents were removed and new reagents were added
without purification of the intermediates and the final products were isolated by column
chromatography (Scheme 5.46, Table 5.47).
Table 5.47 Compounds synthesised for biological testing against CECR2.
Entry Boronic Ester R Compound Yielda
1 BPin
S R
O O
N
O
128a 2 %
2 N 128b 2 %
3 N 128c 4 %
4 N
N
128d 3 %
5
S R
O O
BPin N 132a 12 %
6 N 132b 4 %
7 N 132c 1 %
a Isolated yield (column chromatography) over three steps.
5.3. Chemical probes for CECR2 149
5.3.4 Biological analysis
The affinities of compounds 128a,c,d and 132a,b, as well as compounds 133248 and 134248
for BRD proteins were determined using four different methods. Firstly, using the FAC
method described in Section 5.2.8, solutions of each compound (12.5 µM in ethanol) were
injected into a column functionalised with CECR2 protein. All of the compounds were
UV-active, so the output was detected by UV spectroscopy, using DMSO as the co-solvent
and void marker. This assay method allowed an approximate ranking of their affinities to be
contructed (Figure 5.49). The results were found to be highly dependent on concentration;
compound 132b was very poorly soluble in the ethanol / DMSO system making assessment
of its binding ambiguous, with a large margin of error.
In a second FAC assay method, injections of each substrate at three different concentrations
(6.25, 12.5 and 25 µM) were performed, to obtain a trend of ∆v against concentration. To
calculate ∆v for each compound, a sigmoid curve was fitted to the “front” of the absorbance
signal pattern. These data were used to calculate a value for Kd for each compound by
performing a nonlinear curve fitting of ∆v against concentration (Figure 5.48).
 0
 100
 200
 300
 400
 500
 0  5  10  15  20  25  30
Ab
sor
ba
nc
e
Time /min
a
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
 0  5  10  15  20  25  30
(V
 - 
V 0
) [
A 0
] /
10
-9
 n
m
ol
[A0] /µM
Non-linear t
Points
b
Figure 5.48 (a) FAC curves for compound 128c at 6.25, 12.5 and 25 µM. (b) Nonlinear
regression of (V −V0)[A0] = (Bt [A])/(Kd +[A]) to obtain a value for Kd .
In a third (virtual) method, the structures were docked into a model of the CECR2 protein in
silico using Schrödinger Glide,283 to estimate their relative binding affinities. The affinities
that were calculated by Glide§ followed a similar trend to the FAC results, suggesting
that the same binding site is active in the protein in vitro and in the FAC column. The
involvement of water atoms, which are found to be conserved within the binding sites
of many bromodomain proteins, may affect compound binding.284 The effect of water
§ Glide generates numerical binding scores; more negative scores indicate stronger bonding.
150 Development of inhibitors for bromodomain-containing proteins

 



























	

	

N
N
NN
S
N
N
O
O
N
N
NN
S
N
O
O
O
B
r
S
N
O
O
NN N
NS
N
O
O
NN N
N
S
N
O
O
NN N
N
S
N
HN
E
tO
O
O
O
NN N
N
S
N
HN
E
tO
O
O
O
NN N
N
S
N
O
O
A
A
B
B
C
C
D
D
E
E
F
F
H
G
G
H
<
<
<
<
132b
128a
128d
118a
128c
132a
133
134
Figure
5.49
Injections
ofa
num
berofcom
pounds
ata
single
concentration
(12.5
µM
)to
the
C
E
C
R
2
FA
C
colum
n,show
ing
consistent
retention
for
each
substrate
over
atleastthree
experim
ents.
T
he
results
give
an
approxim
ate
ranking
ofthe
com
pounds
(top).
A
n
artefactis
visible
in
each
trace
at16.5
m
inutes;this
is
thoughtto
have
been
created
by
the
end
ofthe
600
µL
injection
loop.
5.3. Chemical probes for CECR2 151
N
N
NN
N
O O
S
N
N
N
N
S
N
O
O O
N
N
N
N
S
N
N
O O
N
N
N
N
S
N
O O
N
N
NN
S
N
O O
N
N
NN
S
N
O O
H
NEtO
O
N
N
NN
S
O O
H
NEtO
O
57
132b 128a
128d 128c
132d
134
Figure 5.50 Depiction of the test compounds docked into CECR2, showing the most
favourable binding orientations as calculated by Glide.
152 Development of inhibitors for bromodomain-containing proteins
molecules in compound binding may be used to explain the high binding results returned
for compounds 128a and 132b which were found to have a low affinity in the FAC assay.
In the fourth method, the compounds were analysed246 against a set of BRD proteins using
Tm shift measurements (Table 5.51).
Table 5.51 Tm shift data collected for compounds 128, 132, 133 and 134 across a suite
of bromodomain-containing proteins.246
Compound ∆Tm,obsBRD1 BRD4A BRD9 BRPF1B CECR2 CREBBP BRPF3
128a −0.92 0.17 2.64 −0.63 0.66 −1.34 −0.01
128c −0.27 1.30 2.35 0.35 0.48 0.48 −0.11
128d 0.14 1.20 1.40 0.30 0.61 −0.81 0.19
132a 0.89 4.79 1.15 0.78 0.74 0.79 0.43
132b 0.30 1.14 1.90 0.12 0.83 0.36 0.30
133248 0.82 7.39 6.44 2.57 8.08 2.21 0.91
134248 0.49 4.86 7.03 2.41 9.80 1.71 0.12
∆Tm,obs = Difference in temperature of denaturation of the protein with and without the ligand,
data in ◦C.
Compounds 133 and 134, which have a pendant carbamate moiety similar to bromosporine,
showed a greater affinity for the proteins. This indicates that the presence of this side-chain
significantly affects their affinity for CECR2. Furthermore, the assay methods show the
ability to distinguish between substrates that have weak binding affinities. This result is
very promising, since fragments and intermediates such as 131a could be tested at early
stages during a library synthesis providing additional data with which to direct a discovery
project.
5.4 Conclusion
An array of compounds was generated and tested against a set of bromodomain compounds
using both conventional and FAC biological assay methods. The data produced by the FAC
technique was found to correlate well with that produced by the conventional methods
(Table 5.52). With an appropriately-sized column, FAC is able to distinguish between
compounds with relatively low affinities for the protein under analysis, suggesting that it
could be an effective assay method for fragment-based discovery programmes.
5.4. Conclusion 153
Table 5.52 Comparison of data from different assay techniques for compounds 132, 128,
133 and 134. (A more negative Glide score indicates stronger calculated affinity).
Compound FAC rankinga FAC Kd /µM Glide score ∆Tm,obs (CECR2)
134 1 n.d. −8.3 9.80
133 2 54±6 n.d. 8.08
132b n.d. n.d. −7.6 0.83
132a 3 45±7 −7.8 0.74
128c 4 41±8 −7.4 0.48
128d 5 62±30 −6.4 0.61
128a 6 208±156 −7.3 0.66
a Determined by visual inspection of absorption curves (Figure 5.49) to identify
the breakthrough volume, then ordering from the latest to the earliest.
Since FAC is fundamentally a chromatographic technique it is possible that with further
development FAC analysis could be integrated into a flow synthesis platform to provide
rapid feedback on compounds as they are produced. Alternatively, with additional equip-
ment such as a column-switching device sometimes included in laboratory HPLC setup,
compounds could be tested against a suite of proteins within a matter of hours using a
bench-top system.
Flow synthesis was found to provide benefits in the early stages of this synthesis to enable
relatively large-scale reactions to be performed using hazardous reagents. Future develop-
ment of flow chemistry systems285 or alternative automated batch-synthesis platforms286
may lead to improvements in the later stages of a discovery programme where materials
are being handled in smaller quantities and under a greater variety of conditions.
154
Conclusion
The goal of this work was to investigate how automation could contribute to flow synthesis
programmes, and to develop tools to make computer-enabled processes easier to implement
so that the use of this technology might become more commonplace in the future.
“Solid-phase” synthesis is an important application of immobilised reagents, a key compo-
nent in the flow chemistry toolbox. Automation methods were found to enable the more
repetitive synthetic tasks to be performed under complete computer control so that long
tasks, or operations requiring complex sequences, could be carried out overnight and with
greater consistency between experiments.
The software packages currently supplied with commercially available flow chemistry
platforms were found to be incapable of the more complex operations that might be
required. The available laboratory automation software packages are primarily targeted
at biochemical assaying applications. A new control and monitoring system (Octopus)
was developed using the Python programming language, and the source code released
under an open-source license so that publications using this software might be more readily
reproducible.
This system was applied to a two-step hydration/reduction sequence to obtain piperazine-
2-carbonitrile, a valuable 3D building block. This involved the integration of a Vapourtec
R2+/R4 system, a Knauer HPLC pump, a FlowIR spectrometer, an H-Cube hydrogenation
reactor and a multi-position valve; this would have been much more challenging using pre-
vious automation technologies. Application of the system for running multiple experiments,
such as for a Design of Experiments procedure, was demonstrated.
Automation techniques were also found to be invaluable in the multi-step flow synthesis of
2-aminoadamantane-2-carboxylic acid, which required intermediate work-up operations
between different stages. We were able to implement multiple independent steps with
155
156 Conclusion
control of several different devices and operations, which would have been very difficult to
achieve manually.
Finally, flow chemistry and remote control and monitoring methods were applied where
appropriate to a discovery programme for inhibitors of bromodomain proteins. The Frontal-
Affinity Chromatography (FAC) technique was shown to provide results that were consistent
with conventional assay methods. Since FAC could conceivably be integrated into a flow
synthesis system it has the potential to form part of a machine-assisted, “closed-loop”
discovery programme where biological data are fed directly into an automated synthesis
platform for iterative development of biologically-active molecules.
In summary, automation technology has allowed more complex flow synthesis platforms
to be operated under precise control. A number of other research programmes within
our laboratory have started to use these automation systems. Future development of the
software control package should make this technology accessible to non-programmers.
Another important area for future development will be for the intelligent automation of
analysis and purification procedures, since the running of reactions is probably one of the
easier aspects of synthetic chemistry to automate! Nevertheless, routine automation of
experiments could remove some of the variation inherent in synthesis, and lead to much
more reproducible and better documented procedures.
Appendix A
Synchronous instrument controller
157
158 Synchronous instrument controller
Listing A.1 vapourtec.py Class for creating objects to control Vapourtec R2+/R4
machines using a Python script. Commands are synchronous, i.e. the programme freezes
until a response is received from the machine. Please note that the commands have been
redacted in the PDF version of this document.
import time
import math
import sys
class R2R4 (object):
transport = None
def __init__ (self, transport):
""" Initialise the controller with a transport object. """
transport.setTerminator("\r\n")
transport.open()
self.transport = transport
def _send (self, line):
return self.transport.send(line)
def _press (self, key):
""" Emulate a button press on the machine. """
return self._send(" " + str(key))
def status (self):
""" Get the current status.
returns: 0 = off, 1 = on, >1 = error """
return self._send("").split()[0]
def on (self):
""" Power on the machine. """
return self._send("")
def off (self):
""" Power off the machine. """
return self._send("")
def waste (self):
""" Direct output to waste. """
return self._press(8)
def collect (self):
""" Collect output. """
return self._press(9)
vapourtec.py 159
def temp (self, heater, target):
""" Set heater temperature.
heater = 0, 1, 2 or 3.
target = -1000 or 20 to 250 """
return self._send("  {:d} {:d}".format(heater, target))
def valve (self, channel, position):
""" Switch a valve.
channel = "orange", "purple" or "both"
position = "solvent", "reagent", "load" or "inject"
"""
positions = ["solvent", "reagent", "load", "inject"]
orange = [0, 1, 4, 5]
purple = [2, 3, 6, 7]
if channel == "orange":
return self._press(orange[positions.index(position)])
if channel == "purple":
return self._press(purple[positions.index(position)])
if channel == "both":
return (
self._press(orange[positions.index(position)]),
self._press(purple[positions.index(position)])
)
def flowrate (self, channel, rate):
""" Set the pump flowrate.
channel = "orange", "purple" or "both"
rate = flowrate in uL/min
"""
if channel == "both":
return (
self._send(" 0 {:d}".format(rate)),
self._send(" 1 {:d}".format(rate))
)
if channel == "orange":
channel = 0
if channel == "purple":
channel = 1
return self._send(" {:d} {:d}".format(channel, rate))
def pressures (self):
""" Read the current pressures """
return self._send("").split("&").pop().split(',')[1:]
# Continued...
160 Synchronous instrument controller
def wait (self, hours = 0, mins = 0, seconds = 0, keepalive = False):
""" Wait for specified time, printing out current pressure every minute.
If keepalive = True, reactor is switched back on if it stops """
# Caclulate the time to stop
end_time = time.time() + (hours * 60 * 60) + (mins * 60) + seconds
while 1:
now = time.time()
# Stop at the end time
if (now > end_time):
break;
left = math.ceil((end_time - now) / 60)
m_left = left % 60
h_left = left / 60
p = self._pressures()
# Print the current pressure readings
sys.stdout.write(
"\rWaiting: %dh %dm Pressures S:%d A:%d B:%d "
% (
h_left, m_left,
int(p[0]), int(p[1]), int(p[2])
)
)
sys.stdout.flush()
# If the R2 is not running, start it.
if keepalive:
for j in range(60):
if self.status() != 1:
self.on()
time.sleep(1) # Check once per second.
else:
time.sleep(60)
print "\n"
vici.py 161
Listing A.2 vici.py Class for creating objects to control Valco VICI microelectric
multiposition valves.
import time
import math
import sys
class MultiValve (object):
transport = None
positions = 0
def __init__ (self, transport):
transport.setTerminator("\r")
transport.open()
self.transport = transport
self.positions = self.num_positions()
def _send (self, line):
return self.transport.send(line)
def num_positions (self):
""" Read the current number of positions """
return int(self._send("NP").split('=')[1])
def position (self):
""" Read the current valve position """
return int(self._send("CP").split('=')[1])
def advance (self, number):
""" Move by [number] positions """
return self.move(self.position() + int(number))
def move (self, position, direction = 'fastest'):
""" Move to [position].
direction = "clockwise", "counterclockwise" or "fastest"
"""
if (direction in ('f', 'fastest')):
command = 'GO'
elif (direction in ('c', 'cw', 'clockwise')):
command = 'CW'
elif (direction in ('a', 'cc', 'counterclockwise')):
command = 'CC'
# Make sure position is between 1 and NP
position = ((int(position) - 1) % self.positions) + 1
# Trying to move to current position would give an error
if self.position() == position:
return 'OK'
# Continued...
162 Synchronous instrument controller
# Move, then get new pos immediately. (The valve will not respond to
# GO/CC/CW commands, so the CP request is needed so that transport.send
# doesn't wait forever, and is useful as CP only responds once the move
# has been completed.)
new_pos = int(self._send(command + str(position) + "\rCP").split('=')[1])
if new_pos == position:
return 'OK'
else:
return 'Error'
Listing A.3 transport.py Class for creating objects to communicate over a serial
connection. Messages are sent with a terminating sequence, and then the response is
read until the terminating sequence is received.
class serial (object):
ser = None
_terminator = None
_terminator_len = 0
def __init__ (self, port):
import serial
self.ser = serial.Serial()
self.ser.port = port
self.ser.baudrate = 19200
self.ser.timeout = 1
def setTerminator (self, terminator):
self._terminator = terminator
self._terminator_len = len(terminator)
def open (self):
self.ser.open()
def close (self):
self.ser.close()
def send (self, line):
buff = ""
self.socket.send(line + self._terminator)
while len(buff) < self._terminator_len \
or buff[-self._terminator_len:] != self._terminator:
buff = buff + self.ser.read()
return buff[:-self._terminator_len]
Appendix B
Control scripts for monolith
processing
163
164 Control scripts for monolith processing
Listing B.1 Example of a programme to perform a stop-flow sequence over a number of
cycles.
import transport, vapourtec
R2 = vapourtec.R2R4(transport.serial("COM1"))
flow_rate = 200 # Flow rate in uL/min
column_volume = 3000 # Column volume in uL
cycles = 4 # Number of stop-flow cycles to run
temp = 100 # Heating temperature
wait_mins = 120 # Heating time per cycle
# Set up reactor
R2.valve("orange", "reagent")
R2.temp(1, temp)
R2.on()
# Wait for the heater (~80 C/min)
time.sleep(temp / 80.)
for i in range(cycles):
print "Running cycle {:d}".format(i)
# Pump reagent in
R2.flowrate("orange", flow_rate)
R2.wait(mins = column_volume / float(flow_rate))
# Adjust the flow rate of both pumps
R2.flowrate("orange", 0)
R2.wait(mins = wait_mins, keepalive = True)
# Final solvent wash
print "Washing"
R2.valve("orange", "solvent")
R2.flowrate("orange", flow_rate)
R2.wait(mins = column_volume / float(flow_rate))
R2.off()
165
Listing B.2 Programme to wash four new monoliths and then perform a cyclisation
reaction on each one.
import transport, vapourtec, vici
R2 = vapourtec.R2R4(transport.serial("COM1"))
V = vici.MultiValve(transport.serial("COM2"))
# Function to run a solvent switch
def solvent_switch (old_channel, new_channel, flow_rate, steps):
for i in range(steps):
change = (i + 1) * flow_rate / steps
R2.flowrate(old_channel, flow_rate - change)
R2.flowrate(new_channel, change)
R2.wait(mins = 20)
# Set up reactor
R2.valve("both", "solvent")
R2.flowrate("purple", 0)
R2.flowrate("orange", 100)
for i in range(4):
R2.temp(i, 25)
R2.on()
# Run 4 columns
for i in range(4):
print "Column {:d}".format(i + 1)
V.move(i + 2)
print "Ethanol Wash"
R2.wait(hours = 2)
print "Solvent Switch to MeCN"
solvent_switch("orange", "purple", 200, 4)
print "Cyclisation"
R2.valve("both", "reagent")
R2.flowrate("both", 100)
R2.wait(hours = 5)
print "Rinse\n"
R2.temp(i, -1000)
R2.valve("both", "solvent")
R2.wait(mins = 20)
R2.flowrate("purple", 0)
R2.off()
166 Control scripts for monolith processing
Listing B.3 Programme to perform oxidative activation on four columns, and then run
interleaved stop/flow reactions using a different amine for each column. The load/inject
valve on the orange channel is used to isolate the multi-position valve during position
changes to avoid contaminating columns with the wrong reagents; careful washing of the
dead volume is required.
import transport, vapourtec, vici
R2 = vapourtec.R2R4(transport.serial("COM1"))
V = vici.MultiValve(transport.serial("COM2"))
cycles = 4 # Number of stop-flow cycles
column_volume = 3000 # Monolith dead volume
# Function to run a solvent switch
def solvent_switch (old_channel, new_channel, flow_rate, steps):
for i in range(steps):
change = (i + 1) * flow_rate / steps
R2.flowrate(old_channel, flow_rate - change)
R2.flowrate(new_channel, change)
R2.wait(mins = 20)
# Set up reactor - initially pump DCM
R2.valve("both", "solvent")
R2.flowrate("purple", 100)
R2.flowrate("orange", 0)
V.move(1)
for i in range(4):
R2.temp(i, 25)
R2.on()
# Continued...
167
# Run oxidation + solvent switch on 4 columns
for i in range(4):
print "Column {:d}".format(i + 1)
R2.valve("orange", "reagent")
R2.flowrate("purple", 0)
R2.flowrate("orange", 90)
R2.wait(mins = 2) # Flush out any dioxane
print "Pre-wash"
R2.valve("orange", "load")
V.move(i + 2)
R2.valve("orange", "inject")
print "Oxidation"
R2.wait(hours = 2)
print "Solvent Switch to Dioxane"
R2.valve("both", "solvent")
R2.flowrate("purple", 200)
R2.flowrate("orange", 0)
solvent_switch("purple", "orange", 200, 4)
print "Post-wash"
R2.valve("orange", "load")
V.move(1)
R2.valve("orange", "inject")
R2.flowrate("purple", 500)
R2.flowrate("orange", 0)
R2.wait(mins = 1)
# Wash through with Dioxane
R2.valve("purple", "reagent")
R2.valve("orange", "solvent")
R2.flowrate("both", 500)
R2.wait(mins = 2)
# Continued...
168 Control scripts for monolith processing
# Run stop-flow release cycles
for cycle in range(cycles + 1):
# Once for each column, per cycle...
for i in range(4):
print "Column {:d}".format(i + 1)
# Wash through to second valve with dioxane
R2.flowrate("orange", 500)
R2.flowrate("purple", 0)
R2.wait(mins = 1)
if cycle < cycles:
R2.temp(i, 100)
# Wash through to first valve with amine
R2.valve("orange", "load")
V.move(i + 2)
R2.flowrate("purple", 200)
R2.wait(mins = 2)
# Inject amine into column
R2.flowrate("orange", 0)
R2.flowrate("purple", 100)
R2.valve("orange", "inject")
R2.wait(mins = column_volume / 100.)
R2.flowrate("purple", 0)
# Rinse with dioxane after last cycle
else:
R2.temp(i, -1000)
R2.valve("orange", "load")
V.move(i + 2)
R2.flowrate("orange", 100)
R2.valve("orange", "inject")
R2.wait(mins = column_volume / 100.)
# Set up to wash through to second valve
# with dioxane (on next cycle)
R2.valve("orange", "load")
V.move(1)
R2.valve("orange", "inject")
R2.off()
Experimental
169
170 Experimental
General Experimental Information
Ethyl acetate, hexane, dichloromethane and toluene were obtained from Fisher Scientific
and distilled before use. Absolute ethanol was obtained from Sigma-Aldrich, and HPLC
grade methanol, isopropanol and acetonitrile were obtained from Fisher Scientific and used
without further purification. Unless otherwise specified, all other reagents and solvents
were obtained from commercial sources and used without further purification. All reactions
were performed under an atmosphere of air unless otherwise specified.
Silica gel chromatography was performed on a Biotage SP1 purification system with pre-
packed SiliCycle silica cartridges (4, 12, 25 or 40 g), or similar cartridges manually packed
with Biotage KP-NH silica. Analytical thin-layer chromatography (TLC) was performed
using pre-coated glass-backed plates (Merck Kieselgel 60 F254) and visualized under ultra-
violet light (short and long-wave) or using potassium permanganate (KMnO4). LC-MS
analysis was performed on an Agilent HP 1100 series chromatography (Mercury Luna 3u
C18 (2) column) attached to a Waters ZQ2000 mass spectrometer with ESCi ionisation
source in ESI mode. Elution was carried out at a flow rate of 0.6 mL/min using a reverse
phase gradient of MeCN and water containing 0.1 % formic acid:
Time /min % MeCN
0.0 5
1.0 5
4.0 95
5.0 95
7.0 5
8.0 5
1H NMR spectra were recorded in CDCl3, DMSO-d6, or MeOH-d4 on a Bruker Avance
DPX-400, DRX-500 Cryo or DRX-600 spectrometer with residual CHCl3, DMSO or
MeOH as the internal reference (δH = 7.26, 2.50 or 4.87 ppm, respectively). 1H resonances
are reported to the nearest 0.01 ppm. 13C NMR spectra were recorded in CDCl3, DMSO-d6
or MeOH-d4 on the same spectrometers with the central peak of CHCl3, DMSO or MeOH
as the internal reference (δC = 77.0, 39.5 or 49.1 ppm, respectively). 13C resonances are
reported to the nearest 0.01 ppm. DEPT-135, COSY, HMQC, and HMBC experiments were
used to aid structural determination and spectral assignment. The multiplicity of the 1H and
13C signals are indicated as: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet,
Experimental 171
br = broad, dd = doublet of doublets, etc. Coupling constants (J) are quoted in Hz and
reported to the nearest 0.1 Hz.
Infrared spectra were recorded on a Perkin-Elmer Spectrum One FT-IR spectrometer using
universal ATR sampling accessories. Letters in parentheses refer to relative absorbance
with respect to the most intense peak: w = weak (< 40 %), m = medium (40 – 75 %), s =
strong (> 75 %). Melting points were performed on a Stanford Research Systems MPA100
(OptiMelt) automated melting point system and are uncorrected. High resolution mass
spectrometry (HRMS) within ±5 ppm was carried out either on a Waters Micromass LCT
Premier spectrometer using time of flight with positive electrospray ionisation, or by Mr
Paul Skelton on a Bruker BioApex 47e FTICR spectrometer using positive ESI or EI at
70 eV, to within a tolerance of 5 ppm of the theoretically calculated value.
Elemental composition microanalysis was performed by the Microanalytical Laboratories
at the Department of Chemistry, University of Cambridge and results are reported to two
decimal places.
172 Experimental
Solid-supported synthesis of 2-aminopyrimidines
General Experimental Information
Monolithic reagents were formed and used in glass Omnifit columns, sealed for polymeri-
sation with custom made PTFE end pieces and heated using a Vapourtec R4 convection
heating system.
Magic Angle Spinning NMR spectra were recorded by Dr David Reid on a Bruker DRX-
400 spectrometer at room temperature using a Bruker HR-MAS probe at the Cambridge
University Chemical Laboratory.
Divinylbenzene (80 % technical grade) was obtained from Sigma-Aldrich.
Amino[(4-ethenylbenzyl)sulfanyl]methanaminium chloride (11)287
3
S1H2N
NH2 2'
3'
5'
6'
Cl
4-Vinylbenzyl chloride (60.0 g, 0.394 mmol) and thiourea (25.00 g, 0.329 mmol) in ethanol
(165 mL) were heated to reflux for 6 h, until all of the thiourea had been consumed as
observed by TLC (mobile phase: 1:1 EtOAc / hexane, v/v). The ethanol was removed
under reduced pressure and the resulting yellow solid was washed with ether (3×20 mL).
Drying in vacuo gave the title compound (67.1 g, 0.29 mmol, 90 %) as an off-white solid,
m.p. >250 ◦C (decomp.).
1H NMR (600 MHz, DMSO-d6): δ /ppm = 9.49 (br. s, 2 H, NH2), 9.37 (br. s, 2 H, NH2),
7.46 (d, J = 8.1 Hz, 2 H, H-2′), 7.39 (d, J = 8.1 Hz, 2 H, H-3′), 6.71 (dd, J = 10.9 Hz and
6.7 Hz, 1 H, H-5′), 5.83 (d, J = 6.7 Hz, 1 H, H-6′cis), 5.26 (d, J = 10.9 Hz, 1 H, H-6′trans),
4.52 (s, 2 H, H-3). 13C NMR (100 MHz, DMSO-d6): δ /ppm = 169.29 (C-1), 136.75 (C-4′),
136.06 (CH, C-5′), 134.93 (C-1′), 129.33 (CH, C-2′), 126.46 (CH, C-3′), 114.83 (CH2,
C-6′), 33.97 (CH2, C-3). IR (neat): ν /cm−1 = 3270 (w, br.), 3179 (w, br.), 3043 (m, br.),
2722 (w), 1655 (s), 1629 (w), 1508 (w), 1437 (m), 1405 (m), 1306 (w), 1256 (w), 1206 (w),
1164 (w), 1113 (w), 1093 (w), 1017 (w), 987 (m), 915 (m), 908 (m), 832 (m), 675 (s, br).
HRMS: m/z calc. for [C10H13S1N2]
+ ([M – Cl]+) 193.0794, found 193.0800, ∆ = 3.1 ppm.
Experimental 173
Thiouronium chloride monolith (M1)
SH2N
NH2Cl
The reaction scale specified below forms a monolith that fills a 7.845 mL glass Omnifit
column (10 mm ID × 100 mm length). Multiple monoliths, or monoliths of other volumes,
can be made from the same stock solution by multiplying the values below as appro-
priate. Divinylbenzene (DVB) (0.77 g, 5.9 mmol) was added to amino[(4-ethenylbenz-
yl)sulfanyl]methanaminium chloride (1.16 g, 5.0 mmol) suspended in 1-propanol (3.59 g,
81.5 mmol) and water (0.65 g, 36.1 mmol) and the mixture was heated to 50 ◦C. Once
complete dissolution had been achieved, 1,1′-azobis(cyclohexanecarbonitrile) (ACHC)
(19.4 mg, 0.08 mmol, 0.3 mol% with respect to the monomer + crosslinker) was added and
the mixture stirred for another 5 minutes without further heating. A 10 mm ID × 100 mm
length glass column was filled to a height of 7 cm with this solution, and both ends of the
column were sealed with custom-made PTFE end plugs. The column was heated at 90 ◦C
for 20 hours using a Vapourtec R4 convection heater, resulting in a rigid white monolith
(the start of precipitation polymerisation was observed after 30 minutes at temperature).
Following polymerisation, the monolith was allowed to cool to room temperature, and the
end seals were exchanged for standard tubing connectors. To remove the porogen and any
residual nonpolymeric material, the monolith was then heated to 60 ◦C and washed with
ethanol at 0.2 mL/min, without a back-pressure regulator, for 1 hour using a Vapourtec R2+
flow system. A 100 psi back-pressure regulator was added to the system and the monolith
was then flushed at 0.2 mL/min for another hour. The flow rate was then slowly increased
to 1.5 mL/min over the next hour, the monolith inverted and then the process repeated. The
monolith was finally flushed at 0.2 mL/min for 30 minutes while being allowed to cool to
room temperature to give the title compound (dry weight 1.6 g) as a white monolith.
13C NMR (100 MHz, solid state): δ /ppm = 171.38, 144.98, 128.23, 40.35. IR (neat):
ν /cm−1 = 2160 (w, br.), 1647 (m), 1438 (m). Microanalysis: found C 67.47%, H 6.71%,
N 7.34%, Cl 9.25% (Loading = 2.6 mmol/g, based on proportions of N and Cl).
174 Experimental
Linear pressure response of monolith M1:
 2
 4
 6
 8
 10
 12
 14
 16
0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
Pre
ssu
re 
(ba
r)
Flow rate (mL/min)
4-Amino-2-((4-vinylbenzyl)thio)pyrimidine-5-carbonitrile (15)
NC
H2N 4 N 2 S
2'
3'
5'
6'
N
6
To a mixture of amino[(4-ethenylbenzyl)sulfanyl]methanaminium chloride (11, 1.6 g,
7.0 mmol) and ethoxymethylene(malononitrile) (620 mg) in ethanol (40 mL) was added
(i−Pr)2NEt (1.3 mL) in one portion at 25 ◦C with stirring. The mixture was stirred for 1
hour, during which time a yellow precipitate formed. The solvent was removed, and the
residue partitioned between ethyl acetate (30 mL) and water (30 mL). The organic phase
was washed with water (2×30 mL), brine (2×30 mL), dried (MgSO4) and the solvent
removed to afford the title compound (1.6 g, 5.8 mmol, 83 %) as a yellow solid.
A sample was purified by column chromatography for analysis (eluent: ethyl acetate
/ hexane; 0 : 1 to 3 : 7 over 35 CV, 3 : 7 to 9 : 1 over 5 CV, then 9 : 1 for 20 CV), m.p.
163 – 164 ◦C.
1H NMR (400 MHz, CDCl3): δ /ppm = 8.34 (s, 1 H, H-6), 7.35 (br. s, 4 H, H-2′ and H-3′),
6.69 (dd, J = 17.6 Hz and 10.9 Hz, 1 H, H-6′trans), 5.73 (dd, J = 17.6 Hz and 0.9 Hz, 1 H,
H-6′cis), 5.60 (br. s, 2 H, NH2), 5.24 (dd, J = 10.9 Hz and 0.9 Hz, 1 H, H-5
′), 4.35 (s, 2 H,
CH2).
13C NMR (100 MHz, CDCl3): δ /ppm = 175.73, 161.55, 159.61 (CH), 136.84,
136.30 (CH), 129.18 (CH), 126.41 (CH), 114.94, 114.04 (CH2), 86.57, 35.22 (CH2). IR
(neat): ν /cm−1 = 3804 (w), 3751 (w), 3711 (w), 3630 (w), 3600 (w), 3399 (w), 3328 (w),
Experimental 175
3217 (w), 2383 (w), 2359 (w), 2343 (w), 2326 (w), 2310 (w), 2225 (w), 2126 (w), 2036
(w), 1908 (w), 1647 (s), 1587 (s), 1538 (s), 1511 (m), 1474 (m), 1432 (w), 1405 (w), 1381
(s), 1326 (w), 1236 (m), 1159 (m), 1100 (w), 1048 (w), 1018 (w), 986 (w), 971 (w), 951
(m), 899 (m), 844 (m), 773 (m), 754 (w), 733 (w), 710 (w), 682 (w). HRMS: m/z calc. for
[C14H13N4S]
+ ([M + H]+) 269.0861, found 269.0857, ∆ = −1.5 ppm.
Monolith M2
N S
N
NC
H2N
A solution of ethoxymethylene(malononitrile) (15 mmol, 3 equiv.) in DMF / MeCN (1 : 1,
v/v, 20 mL) and (i−Pr)2NEt (5.21 mL, 30 mmol, 6 equiv.) in MeCN (20 mL) were each
pumped at 0.1 mL/min, to combine in a tee mixer ( 0.5 mm, PEEK). The resulting stream
passed into the monolith M1, which was maintained at 25 ◦C. The system was pressurised
using a 100 psi back-pressure regulator placed in-line after the monolith. During the loading
process, the monolith gradually assumed a dark yellow colour. After loading, the system
was flushed for 1 hour with EtOH at 0.2 mL/min, whilst maintaining the temperature at
25 ◦C. The obtained monolith M2 was used directly in the following step.
IR (neat): ν /cm−1 = 3361 (w), 2919 (w), 2218 (w), 1618 (m), 1574 (s), 1535 (m), 1511
(m), 1485 (m), 1419 (w), 1363 (s), 1315 (w), 1221 (m), 1020 (w), 950 (w), 779 (m), 709
(m). Dry weight: 1.47 g. Microanalysis: found C 72.43%, H 5.82%, N 12.39%.
Oxidised monolith M3
N S
N
NC
H2N
O O
Monolith M2 was washed for 30 minutes with CH2Cl2 pumped at 0.5 mL/min. A solution
of m-CPBA (15 mmol, 3 equiv.) in CH2Cl2 (20 mL) was pumped, from a reservoir stored at
0 ◦C (ice/water bath), at 0.1 mL/min through the monolith, which was maintained at 25 ◦C.
176 Experimental
The monolith assumed a pale yellow colour after the oxidation reaction. At the end of the
reaction, the system was flushed with CH2Cl2 for 2 hours at room temperature with a flow
rate of 0.1 mL/min. The obtained monolith M3 was used directly in the following step.
IR (neat): ν /cm−1 = 2919 (w), 2231 (w), 2162 (w), 2023 (w), 2012 (w), 1955 (w), 1642
(m), 1576 (m), 1545 (m), 1511 (m), 1486 (w), 1441 (m), 1382 (m), 1325 (m), 1218 (m),
1168 (m), 1127 (s), 1038 (s), 1020 (m), 961 (w), 881 (w), 829 (m), 787 (m), 710 (s). Dry
weight: 1.63 g. Microanalysis: found C 67.32%, H 5.47%, N 8.27%.
4-Amino-2-morpholin-4-ylpyrimidine-5-carbonitrile (18)288
4
N
2
N
6
N
O
2'
3'
N
H2N
Monolith M3 was washed for 30 minutes with dioxane pumped at 0.5 mL/min. A solution
of morpholine (30 mmol, 6 equiv.) and (i−Pr)2NEt (5.21 mL, 30 mmol, 6 equiv.) in
dioxane (20 mL) was infused with 5 mL of amine solution, pumping at 0.1 mL/min. After
50 minutes, the pump was stopped and monolith was heated at a temperature of 80 ◦C.
After 2 hours, the reaction solution was flushed off with a new portion of the reaction
solution (flow rate 0.1 mL/min) and the output collected in a flask. The flushing operation
was repeated four times to afford 20 mL of reaction solution. After the last loading process,
the monolith was washed with EtOH at 0.2 mL/min for 30 minutes. The combined output
was concentrated under vacuum, and the resulting solid suspended in CH2Cl2 (50 mL)
and washed with water (3×50 mL). The organic phase was dried (MgSO4) and the
solvent removed under reduced pressure. The crude material was purified by flash column
chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to 9.8 : 0.2 over 40 CV) to afford the
title compound (297 mg, 1.45 mmol, 35 % over 3 steps) as a white solid, m.p. 227 – 229 ◦C
(dichloromethane / methanol) [lit: 229 – 231 ◦C (ethanol)289].
1H NMR (500 MHz, DMSO-d6): δ /ppm = 8.26 (s, 1 H, H-6), 7.28 (br. s, 2 H, NH2), 3.70
(t, J = 4.8 Hz, 4 H, H-3′), 3.59 (t, J = 4.8 Hz, 4 H, H-2′). 13C NMR (126 MHz, DMSO-d6):
δ /ppm = 162.87 (C-4), 162.03 (CH, C-6), 160.79 (C-2), 117.29 (CN), 78.66 (C-5), 65.99
(CH2, C-2
′), 43.76 (CH2, C-3′). IR (neat): ν /cm−1 = 3418 (w), 3335 (m), 3231 (m), 2960
(w), 2906 (w), 2852 (w), 2222 (m), 1643 (s), 1590 (s), 1540 (s), 1516 (s), 1477 (s), 1443
Experimental 177
(s), 1349 (s), 1311 (m), 1275 (s), 1251 (s), 1223 (m), 1193 (m), 1117 (m), 1103 (s), 1066
(m), 1036 (m), 970 (s), 918 (m), 905 (s), 851 (m), 838 (m), 793 (m), 784 (s), 756 (m), 732
(w), 698 (w). HRMS: m/z calc. for [C9H12N5O]
+ ([M + H]+) 206.1042, found 206.1051,
∆ = 4.4 ppm.
General Procedure A for the preparation of enaminones 21
A mixture of the ketone (50 mmol) and 1,1-dimethoxy-N,N-dimethylmethanamine (7.30 mL,
1.1 equiv.) was heated at 100 ◦C for 12 hours. The reaction mixture was concentrated
under vacuum and crystallized from diethyl ether. The crystals were collected by filtration,
washed with diethyl ether (2×20 mL) and dried under high vacuum to give crystalline
solid.
(E)-3-(Dimethylamino)-1-(pyridin-4-yl)prop-2-en-1-one (21a)290,291
2'
N
3'
2
O
3
N
Prepared according to General Procedure A, to give the title compound (7.2 g, 41 mmol,
82 %) as yellow crystals, m.p. 115 – 116 ◦C (diethyl ether) [lit: 115 – 117 ◦C (ethyl acetate
/ hexane)291].
1H NMR (400 MHz, CDCl3): δ /ppm = 8.65 (dd, J = 4.4 Hz and 1.7 Hz, 2 H, H-2′), 7.78
(d, J = 12.3 Hz, 1 H, H-3), 7.63 (dd, J = 4.4 Hz and 1.7 Hz, 2 H, H-3′), 5.60 (d, J = 12.3 Hz,
1 H, H-2), 3.13 (s, 3 H, Me), 2.90 (s, 3 H, Me). 13C NMR (100 MHz, CDCl3): δ /ppm =
186.56 (C-1), 155.21 (CH, C-3), 150.21 (CH, C-2′), 147.25 (C-4′), 121.22 (CH, C-3′),
91.75 (CH, C-2), 45.32 (CH3), 37.45 (CH3). IR (neat): ν /cm−1 = 3032 (w), 1637 (m),
1594 (w), 1562 (m), 1523 (m), 1485 (w), 1432 (m), 1408 (m), 1367 (m), 1320 (w), 1274
(m), 1247 (m), 1130 (m), 1070 (m), 1058 (m), 1013 (m), 992 (m), 904 (m), 848 (m), 774
(s), 698 (s), 660 (s). HRMS: m/z calc. for [C10H13N2O]
+ ([M + H]+) 177.1028, found
177.1032, ∆ = 2.3 ppm. Microanalysis: calc. (found) for C10H12N2O C 68.16% (68.03%),
H 6.86% (6.88%), N 15.90% (15.79%).
178 Experimental
(E)-3-(Dimethylamino)-1-(pyridin-3-yl)prop-2-en-1-one (21b)291
5'
6' N 2'
4'
2
O
3
N
Prepared according to General Procedure A, to give the title compound (6.6 g, 37 mmol,
75 %) as orange crystals, m.p. 78 – 79 ◦C (diethyl ether) [lit: 71 – 73 ◦C (ethyl acetate /
hexane)291].
1H NMR (400 MHz, CDCl3): δ /ppm = 9.03 (d, J = 2.2 Hz, 1 H, H-2′), 8.61 (dd, J = 4.8 Hz
and 1.8 Hz, 1 H, H-6′), 8.13 (dt, J = 7.9 Hz and 2.0 Hz, 1 H, H-4′), 7.78 (d, J = 12.3 Hz,
1 H, H-3), 7.30 (dd, J = 7.9 Hz and 4.8 Hz, 1 H, H-5′), 5.63 (d, J = 12.3 Hz, 1 H, H-2), 3.12
(s, 3 H, Me), 2.89 (s, 3 H, Me). 13C NMR (100 MHz, CDCl3): δ /ppm = 186.35 (C-1),
154.68 (CH, C-3), 151.47 (CH, C-2′), 148.93 (CH, C-6′), 135.67 (C-3′), 135.05 (CH, C-4′),
123.29 (CH, C-5′), 91.86 (CH, C-2), 45.27 (CH3), 37.43 (CH3). IR (neat): ν /cm−1 = 3372
(w), 3054 (w), 2918 (w), 2805 (w), 1637 (s), 1575 (s), 1533 (s), 1439 (m), 1411 (s), 1364
(s), 1328 (s), 1275 (s), 1249 (s), 1186 (m), 1125 (m), 1107 (m), 1065 (s), 1035 (m), 1023
(s), 1001 (m), 965 (m), 902 (s), 813 (m), 771 (s), 723 (s), 701 (s), 684 (s). HRMS: m/z calc.
for [C10H13N2O]
+ ([M + H]+) 177.1028, found 177.1028, ∆ = 0 ppm. Microanalysis: calc.
(found) for C10H12N2O C 68.16% (67.75%), H 6.86% (6.83%), N 15.90% (15.67%).
(E)-3-(Dimethylamino)-1-phenylprop-2-en-1-one (21c)291
3'
4'
2'
2
O
3
N
Prepared according to General Procedure A, to give the title compound (6.9 g, 39 mmol,
79 %) as yellow crystals, m.p. 90 – 91 ◦C (diethyl ether) [lit: 89 – 92 ◦C (diethyl ether)292].
1H NMR (400 MHz, CDCl3): δ /ppm = 7.91 – 7.83 (m, 2 H, H-2′), 7.76 (d, J = 12.3 Hz,
1 H, H-3), 7.46 – 7.32 (m, 3 H, H-4′ and H-3′), 5.68 (d, J = 12.3 Hz, 1 H, H-2), 3.06 (s, 3 H,
Me), 2.87 (s, 3 H, Me). 13C NMR (100 MHz, CDCl3): δ /ppm = 188.56 (C-1), 154.22 (CH,
C-3), 140.54 (C-1′), 130.84 (CH, C-4′), 128.10 (CH, C-2′), 127.46 (CH, C-3′), 92.18 (CH,
C-2), 44.98 (CH3), 37.20 (CH3). IR (neat): ν /cm−1 = 3024 (w), 2910 (w), 2808 (w), 1634
Experimental 179
(m), 1596 (w), 1582 (m), 1537 (s), 1483 (m), 1446 (w), 1429 (m), 1410 (m), 1363 (m),
1310 (m), 1273 (s), 1232 (m), 1205 (m), 1122 (m), 1079 (w), 1052 (s), 1025 (m), 1014 (m),
1000 (m), 925 (w), 899 (m), 807 (w), 754 (m), 739 (s), 696 (s), 659 (s). HRMS: m/z calc.
for [C11H14NO]
+ ([M + H]+) 176.1075, found 176.1067, ∆ = −4.5 ppm. Microanalysis:
calc. (found) for C11H13NO C 75.40% (75.33%), H 7.48% (7.49%), N 7.99% (7.98%).
(E)-3-(Dimethylamino)-1-(4-(trifluoromethyl)phenyl)prop-2-en-1-one
(21d)291
3'
2'
2
O
3
N
F3C
Prepared according to General Procedure A, to give the title compound (8.6 g, 35 mmol,
71 %) as yellow crystals, m.p. 96 – 97 ◦C (diethyl ether) [lit: 99 – 100 ◦C (ethyl acetate /
hexane)291].
1H NMR (400 MHz, CDCl3): δ /ppm = 7.95 (d, J = 8.0 Hz, 2 H, H-2′), 7.80 (d, J = 12.3 Hz,
1 H, H-3), 7.63 (d, J = 8.0 Hz, 2 H, H-3′), 5.65 (d, J = 12.3 Hz, 1 H, H-2), 3.13 (s, 3 H, Me),
2.90 (s, 3 H, Me). 13C NMR (100 MHz, CDCl3): δ /ppm = 187.02 (C-1), 154.80 (CH,
C-3), 143.69 (q, 5JC−F = 1.1 Hz, C-2′), 132.17 (q, 2JC−F = 32.3 Hz, C-4′), 127.70 (CH,
C-1′), 125.06 (q, 3JC−F = 3.8 Hz, CH, C-3′), 123.97 (q, 1JC−F = 273.3 Hz, CF3), 91.91
(CH, C-2), 45.12 (CH3), 37.29 (CH3). IR (neat): ν /cm−1 = 1638 (m), 1586 (m), 1547 (s),
1489 (w), 1427 (m), 1365 (w), 1319 (m), 1279 (m), 1237 (m), 1157 (s), 1106 (s), 1063
(s), 1011 (m), 982 (m), 960 (m), 902 (m), 856 (s), 843 (m), 819 (w), 777 (s), 764 (m),
750 (m), 695 (s), 680 (m). HRMS: m/z calc. for [C12H13NO
19F3]
+ ([M + H]+) 244.0949,
found 244.0956, ∆ = 2.9 ppm. Microanalysis: calc. (found) for C12H13NOF3 C 59.26%
(59.39%), H 4.97% (4.98%), N 5.76% (5.74%).
180 Experimental
General Procedure B for the preparation of monolith M4
N S
N
N
N S
N
N
N S
N
M4a M4b M4c
N S
N
CF3
N S
N
S
N S
N
OMe
M4d M4e M4f
Solutions (20 mL each) of enaminone (15 mmol, 3 equiv.) and (i−Pr)2NEt (5.21 mL,
30 mmol, 6 equiv.) in EtOH were each pumped at 0.1 mL/min, to combine in a tee mixer
( 0.5 mm, PEEK). The resulting reaction stream passed into the monolith M1, which was
maintained at 80 ◦C. The system was pressurised using a 100 psi back-pressure regulator
placed in-line after the monolith. During the loading process, the monolith gradually
assumed a pale yellow colour. After loading, the system was flushed for 1 hour with EtOH
at 0.2 mL/min, whilst maintaining the temperature at 80 ◦C. The obtained monoliths M4
were used directly in the following step.
For monolith M4a: IR (neat): ν /cm−1 = 3663 (m), 2988 – 2901 (s), 1406 – 1382 (s),
1250 – 1230 (s), 1066 (s), 892 – 879 (m). 13C NMR (100 MHz, solid state): δ /ppm =
144.22, 127.70, 39.82. Dry weight: 1.6 g. Microanalysis: found C 78.90%, H 6.2%, N
8.49%, Cl 0%.
Experimental 181
General Procedure C for the preparation of monolith M5
N S
N
N
O O
N S
N
N
O O
N S
N
O O
M5a M5b M5c
N S
N
CF3
O O
N S
N
S
O O
N S
N
OMe
O O
M5d M5e M5f
The monolith was washed for 30 minutes with CH2Cl2 pumped at 0.5 mL/min. A solution
of m-CPBA (15 mmol, 3 equiv.) in CH2Cl2 (20 mL) was pumped, from a reservoir stored
at 0 ◦C using an ice-water bath, at 0.1 mL/min through the monolith M4, which was
maintained at room temperature. The monolith retained its pale yellow colour after the
oxidation reaction. At the end of the reaction, the system was flushed with CH2Cl2 for 2
hours at room temperature with a flow rate of 0.1 mL/min. The obtained monoliths M5
were used directly in the following step.
The enaminone starting materials could be recovered by removing the solvent from the
collected waste stream, partitioning the concentrated mixture between ethyl acetate and
water, and then collecting and drying the organic phase.
For monolith M5a: IR (neat): ν /cm−1 = 3663 (m), 2988 – 2901 (s), 1715 (w), 1406 –
1381 (s), 1250 – 1229 (s), 1066 (s), 892 (m). Dry weight: 1.6 g. Microanalysis: found C
75.20%, H 5.7%, N 8.2%, Cl 0%.
182 Experimental
General Procedure D for the preparation of 2-amino pyrimidines
(38)
The monolith was washed for 30 minutes with dioxane pumped at 0.5 mL/min. A solution
of amine (30 mmol, 6 equiv.) and (i−Pr)2NEt (5.21 mmol, 30 mmol, 6 equiv.) in dioxane
(20 mL) was prepared. The monolith M5 was infused with 5 mL of amine solution, pumping
at 0.1 mL/min. After 50 minutes, the pump was stopped and monolith was heated at a
temperature of 90 ◦C (except for Entry 4: 120 ◦C). After 2 hours, the reaction solution
was flushed off with a new portion of the reaction solution (flow rate 0.1 mL/min) and the
output collected in a flask. The flushing operation was repeated four times to afford 20 mL
of reaction solution. After the last loading process, the monolith was washed with EtOH
at 0.2 mL/min for 30 minutes. The combined output was concentrated under vacuum, and
the resulting oil was suspended in CH2Cl2 (50 mL) and washed with water (3×50 mL).
The organic phase was dried (MgSO4) and the solvent removed under reduced pressure.
The crude material was purified by flash column chromatography to afford the desired
pyrimidine compound.
4-(4-(Pyridin-4-yl)pyrimidin-2-yl)morpholine (38a)
N
1 N
3
2
N
1'
2'
O
6
N
7
Prepared from M5a and morpholine using General Procedure D. Purification by flash
column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to 9.8 : 0.2 over 40 CV)
afforded the title compound (726 mg, 3 mmol, 73 % over 3 steps) as a pale yellow solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 8.74 (dd, J = 5.0 Hz and 1.6 Hz, 2 H, H-7), 8.46
(d, J = 5.0 Hz, 1 H, H-2), 7.90 (dd, J = 5.0 Hz and 1.6 Hz, 2 H, H-6), 7.00 (d, J = 5.0 Hz,
1 H, H-3), 3.91 – 3.89 (m, 4 H, H-2′), 3.81 – 3.79 (m, 4 H, H-1′). 13C NMR (126 MHz,
CDCl3): δ /ppm = 161.99 (C-1 or C-4), 161.81 (C-1 or C-4), 159.13 (CH, C-2), 150.27
(CH, C-7), 145.07 (C-5), 121.01 (CH, C-6), 106.18 (CH, C-3), 66.84 (CH3, C-2
′), 44.22
(CH3, C-1
′). IR (neat): ν /cm−1 = 2538 (w, br.), 2159 (s), 2031 (s), 1977 (s), 1583 (s), 1561
– 1543 (s, br.), 1487 (s), 1340 (s), 1248 (s), 985 – 962 (s), 800 (s). HRMS: m/z calc. for
[C13H15O1N4]
+ ([M + H]+) 243.1240, found 243.1232, ∆ = −3.5 ppm.
Experimental 183
N-(4-Methoxybenzyl)-4-(pyridin-4-yl)pyrimidin-2-amine (38b)
N
1 N
3
2
N
H
1'3'
4'
MeO N
7
6
Prepared from M5a and (4-methoxyphenyl)methanamine using General Procedure D.
Purification by flash column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to
9.7 : 0.3 over 40 CV) afforded the title compound (320 mg, 1.2 mmol, 29 % over 3 steps) as
a pale yellow solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 8.77 (d, J = 5.5 Hz, 2 H, H-7), 8.46 (d, J = 5.1 Hz,
1 H, H-2), 7.92 (d, J = 5.5 Hz, 2 H, H-6), 7.35 (d, J = 8.4 Hz, 2 H, H-4′), 7.07 (d, J = 5.1 Hz,
1 H, H-3), 6.92 (d, J = 8.4 Hz, 2 H, H-3′), 5.61 (br. s, 1 H, NH), 4.69 (d, J = 5.5 Hz, 2 H,
H-1′), 3.83 (s, 3 H, H-6′). 13C NMR (126 MHz, CDCl3): δ /ppm = 162.61 (C-1 or C-4),
162.46 (C-1 or C-4), 159.43 (CH, C-2), 159.09 (C-5′), 150.46 (CH, C-7), 145.04 (C-5),
131.15 (C-2′), 129.04 (CH, C-3′), 121.18 (CH, C-6), 114.18 (CH, C-4′), 107.04 (CH, C-3),
55.46 (CH3, C-6
′), 45.23 (CH2, C-1′). IR (neat): ν /cm−1 = 2161 (w, br), 1977 (w, br),
1570 (s), 1548 (s), 1510 (s), 1454 (s), 1346 (s), 1230 (s), 1175 (s), 1030 (s), 795 (s). HRMS:
m/z calc. for [C17H17O1N4]
+ ([M + H]+) 293.1402, found 293.1410, ∆ = 2.7 ppm.
N-Cyclohexyl-4-(pyridin-4-yl)pyrimidin-2-amine (38c)
N
1 N
3
2
N
H
1'
2'
3'
4'
N
7
6
Prepared from M5a and (4-methoxyphenyl)methanamine using General Procedure D.
Purification by flash column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to
9.8 : 0.2 over 40 CV) afforded the title compound (533 mg, 2.1 mmol, 51 % over 3 steps) as
a white solid.
1H NMR (400 MHz, CDCl3): δ /ppm = 8.74 (d, J = 4.5 Hz, 2 H, H-7), 8.40 (d, J = 5.1 Hz,
1 H, H-2), 7.86 (d, J = 4.5 Hz, 2 H, H-6), 6.96 (d, J = 5.1 Hz, 1 H, H-3), 5.18 (d, J = 8.0 Hz,
1 H, H-1′), 2.13 – 1.22 (m, 11 H, NH and cyclohexyl). 13C NMR (100 MHz, CDCl3):
184 Experimental
δ /ppm = 162.09 (C-1 or C-4), 159.31 (CH, C-2), 150.48 (CH, C-7), 144.87 (C-5), 120.90
(CH, C-6), 106.25 (CH, C-3), 49.80 (CH, C-1′), 33.13 (CH2, C-2′), 25.74 (CH2, C-4′),
24.86 (CH2, C-3
′), (C-1 or C-4 was not visible). IR (neat): ν /cm−1 = 2921 (w, br), 2852
(s), 1581 (s), 1548 (s), 1583 (s), 1528 (s), 1421 (s), 1368 (s), 796 (s). HRMS: m/z calc. for
[C15H19N4]
+ ([M + H]+) 255.1610, found 255.1611, ∆ = 0.4 ppm.
4-(4-(Pyridin-3-yl)pyrimidin-2-yl)morpholine (38d)
N
1 N
3
2
N
1'
2'
O
6
10
9
8
N
Prepared from M5b and morpholine using General Procedure D. Purification by flash
column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to 9.7 : 0.3 over 40 CV)
afforded the title compound (713 mg, 2.9 mmol, 70 % over 3 steps) as a white solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 9.24 (d, J = 2.3 Hz, 1 H, H-6), 8.69 (dd, J = 4.8 Hz
and 1.7 Hz, 1 H, H-8), 8.42 (d, J = 5.1 Hz, 1 H, H-2), 8.33 (ddd, J = 8.0 Hz and 2.3 Hz
and 1.7 Hz, 1 H, H-10), 7.45 – 7.38 (m, 1 H, H-9), 6.98 (d, J = 5.1 Hz, 1 H, H-3), 3.90 (dd,
J = 5.7 Hz and 4.0 Hz, 4 H, H-2′), 3.80 (dd, J = 5.7 Hz and 4.0 Hz, 4 H, H-1′). 13C NMR
(126 MHz, CDCl3): δ /ppm = 161.90 (C-1 or C-4), 161.86 (C-1 or C-4), 158.76 (CH, C-2),
150.95 (CH, C-6), 148.26 (CH, C-8), 134.58 (CH, C-10), 133.24 (C-5), 123.59 (CH, C-9),
105.90 (CH, C-3), 66.85 (CH2, C-2
′), 44.24 (CH2, C-1′). IR (neat): ν /cm−1 = 2868 (w, br),
1586, (s), 1565-1556 (s), 1444-1412 (s), 1346 (s), 1253 (s), 1110 (s), 984 (s), 962 (s), 790
(s), 800 (s). HRMS: m/z calc. for [C13H15O1N4]
+ ([M + H]+) 243.1246, found 243.1252,
∆ = 2.5 ppm.
Experimental 185
N-(2-Methyl-5-nitrophenyl)-4-(pyridin-3-yl)pyrimidin-2-amine
(24)293,294
N
1 N
3
2
N
H
2'4'
5'
7'
NO2
6
10
9
8
N
Prepared from M5b and 2-methyl-5-nitroaniline using General Procedure D. Purification by
flash column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to 9.7 : 0.3 over 40 CV)
afforded the title compound (619 mg, 2.01 mmol, 49 % over 3 steps) as a pale yellow solid.
1H NMR (400 MHz, DMSO-d6): δ /ppm = 9.31 (s, 1 H, H-6), 9.18 (s, 1 H, NH), 8.81
(s, 1 H, H-2′), 8.70 (d, J = 4.7 Hz, 1 H, H-8), 8.61 (d, J = 5.2 Hz, 1 H, H-2), 8.46 (d, J =
8.2 Hz, 1 H, H-10), 7.87 (dd, J = 8.2 Hz and 2.5 Hz, 1 H, H-4′), 7.59 – 7.45 (m, 3 H, H-3,
H-9 and H-5′), 2.42 (s, 3 H, H-7′). 13C NMR (100 MHz, DMSO-d6): δ /ppm = 161.56
(C-1), 160.25 (C-4), 159.60 (CH, C-2), 151.57 (CH, C-6), 148.13 (CH, C-8), 145.80 (C-3′),
138.85 (C-6′), 138.54 (C-1′), 134.24 (CH, C-10), 131.84 (C-5), 131.15 (CH, C-5′), 123.80
(CH, C-9), 117.82 (CH, C-4′), 117.33 (CH, C-2′), 108.85 (CH, C-3), 18.33 (CH3, C-7′). IR
(neat): ν /cm−1 = 3250 (w), 3099 (w), 3066 (w), 1602 (w), 1580 (m), 1557 (m), 1531 (s),
1515 (s), 1477 (m), 1456 (m), 1440 (m), 1420 (m), 1406 (s), 1345 (s), 1305 (m), 1284 (m),
1270 (m), 1248 (w), 1208 (w), 1196 (w), 1136 (w), 1108 (w), 1085 (w), 1027 (m), 991 (w),
961 (w), 930 (w), 882 (w), 830 (w), 817 (w), 810 (w), 791 (s), 754 (w), 736 (m), 698 (m),
686 (m). HRMS: m/z calc. for [C16H14N5O2]
+ ([M + H]+) 308.1148, found 308.1145, ∆ =
−1.0 ppm.
4-(4-Phenylpyrimidin-2-yl)morpholine (38e)
N
1 N
3
2
N
1'
2'
O 8
7
6
Prepared from M5c and morpholine using General Procedure D. Purification by flash
column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to 9.8 : 0.2 over 40 CV)
afforded the title compound (482 mg, 2.0 mmol, 48 % over 3 steps) as a pale yellow solid.
186 Experimental
1H NMR (500 MHz, CDCl3): δ /ppm = 8.38 (d, J = 5.1 Hz, 1 H, H-2), 8.05–8.01 (m, 2 H,
H-6), 7.48–7.45 (m, 3 H, H-7 and H-8), 6.97 (d, J = 5.1 Hz, 1 H, H-3), 3.92–3.88 (m, 4 H,
H-2′), 3.82–3.78 (m, 4 H, H-1′). 13C NMR (126 MHz, CDCl3): δ /ppm = 164.34 (C-1),
161.98 (C-4), 158.30 (CH, C-2), 137.53 (C-5), 130.52 (CH, C-8), 128.66 (CH, C-7), 126.98
(CH, C-6), 106.11 (CH, C-3), 66.92 (CH2, C-2
′), 44.28 (CH2, C-1′). IR (neat): ν /cm−1 =
2969 (w, br.), 2850 (s), 1588 (s), 1564 (s), 1547 (s), 1475 (s), 1441-1426 (m, br), 1340
(s), 1248 (s), 982 (s), 787 (s), 694 (s). HRMS: m/z calc. for [C14H16N3O]
+ ([M + H]+)
242.1293, found 242.1294, ∆ = 0.4 ppm.
4-(4-(4-(Trifluoromethyl)phenyl)pyrimidin-2-yl)morpholine (38f)
N
1 N
3
2
N
1'
2'
O
6
7
CF3
Prepared from M5d and morpholine using General Procedure D. Purification by flash
column chromatography (eluent: CH2Cl2 / MeOH from 10 : 0 to 9.9 : 0.1 over 40 CV)
afforded the title compound (803 mg, 2.6 mmol, 63 % over 3 steps) as a pale yellow solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 8.43 (d, J = 5.1 Hz, 1 H, H-2), 8.13 (d, J = 8.1 Hz,
2 H, H-7), 7.71 (d, J = 8.1 Hz, 2 H, H-6), 6.98 (d, J = 5.1 Hz, 1 H, H-3), 3.93 – 3.89 (m, 4 H,
H-2′), 3.85 – 3.79 (m, 4 H, H-1′). 13C NMR (126 MHz, CDCl3): δ /ppm = 163.04 (C-1),
162.12 (C-4), 158.94 (CH, C-2), 141.14 (C-6), 127.48 (C-5), 125.77 (q, 3JC−F = 3.8 Hz,
CH, C-7), 124.12 (q, 1JC−F = 272.6 Hz, C-9), 106.42 (CH, C-3), 67.03 (CH2, C-2′), 44.42
(CH2, C-1
′), (C-8 was not visible). IR (neat): ν /cm−1 = 2960 (w, br.), 2861 (s), 1592 (s),
1556 (s), 1490 (s), 1440 (s), 1325 (s), 1252 (s), 1164 (s), 1100 (s), 1068 (s), 983 (s), 852
(s), 800 (s). HRMS: m/z calc. for [C15H15N3OF3]
+ ([M + H]+) 310.1167, found 310.1166,
∆ = −0.3 ppm.
Experimental 187
6-Methyl-N1-(4-(pyridin-3-yl)pyrimidin-2-yl)benzene-1,3-diamine
(25)293
10
9
8
N
6
N
1
N2
3
H
N
2' 4'
5'
NH2
7'
A solution of compound 24 (105 mg, 0.34 mmol) in CH2Cl2 (15 mL, approx. 0.02 mM)
was prepared. A H-Cube hydrogenation reactor equipped with a 5 % Pd/C cartridge was
prepared as follows: The reactor and column were washed with CH2Cl2 at 1 mL/min for 5
minutes. For the reaction, the system pressure regulator was set to 20 bar and the hydrogen
pressure to 35 bar in “controlled” mode. The column temperature was set to 40 ◦C, and
the solution of 24 infused at 0.5 mL/min. The product was cycled through the reactor once
more, and then the collected material concentrated under reduced pressure to afford the
title compound (69 mg, 73 %) as a yellow oil.
1H NMR (400 MHz, CDCl3): δ /ppm = 9.26 (d, J = 2.2 Hz, 1 H, H-6), 8.71 (dd, J = 4.8 Hz
and 1.7 Hz, 1 H, H-8), 8.48 (d, J = 5.1 Hz, 1 H, H-2), 8.33 (dt, J = 8.0 Hz and 2.2 Hz, 1 H,
H-10), 7.59 (d, J = 2.4 Hz, 1 H, H-2′), 7.41 (dd, J = 8.0 Hz and 4.8 Hz, 1 H, H-9), 7.13
(d, J = 5.1 Hz, 1 H, H-3), 6.94 – 7.03 (m, 2 H, NH and H-5′), 6.41 (dd, J = 8.0 Hz and
2.4 Hz, 1 H, H-4′), 3.65 (br. s, 2 H, NH2), 2.24 (s, 3 H, H-7
′). 13C NMR (100 MHz, CDCl3):
δ /ppm = 162.53 (C-1), 160.70 (C-4), 158.99 (CH, C-2), 151.40 (CH, C-6), 148.54 (CH,
C-8), 145.12 (C-3′), 137.90 (C-1′), 134.43 (CH, C-10), 132.75 (C-5), 131.03 (CH, C-5′),
123.60 (CH, C-9), 118.16 (C-6′), 110.66 (CH, C-2′ or C-4′), 108.47 (CH, C-2′ or C-4′),
108.02 (CH, C-3), 17.21 (CH3, C-7
′). HRMS: m/z calc. for [C16H16N5]
+ ([M + H]+)
278.1406, found 278.1404, ∆ = −0.7 ppm.
188 Experimental
4-(Chloromethyl)-N-(4-methyl-3-((4-(pyridin-3-yl)pyrimidin-2-yl)
amino)phenyl)benzamide (27)295
10
9
8
N
6
N
1
N2
3
H
N
2'
5'
HN 8'
O
10'
11'
13'
Cl
7'
Compound 25 (69 mg, 0.25 mmol) was suspended in distilled THF (4 mL). Triethylamine
(0.07 mL, 1.15 equiv.) was added, and the mixture cooled to 0 ◦C and stirred for 10 minutes.
A solution of 4-(chloromethyl)benzoyl chloride (54 mg) in distilled THF (2 mL) was added
dropwise over 3 minutes. During the addition a darkening in colour of the solution was
observed. The mixture was stirred at 0 ◦C for 3 hours. After this time, LC-MS analysis
indicated complete conversion. Water (5 mL) was added dropwise over 5 minutes, and the
mixture allowed to warm to room temperature and then stirred for 18 hours, during which
time a yellow precipitate formed. Filtration afforded the title compound (61 mg, 57 %) as a
yellow crystalline solid.
1H NMR (500 MHz, DMSO-d6): δ /ppm = 10.22 (br. s, 1 H, NHamide), 9.26 (d, J = 2.1 Hz,
1 H, H-6), 8.97 (br. s, 1 H, NH), 8.67 (dd, J = 4.8 Hz and 1.7 Hz, 1 H, H-8), 8.50 (dd, J =
5.2 Hz and 1.7 Hz, 1 H, H-2), 8.47 (dq, J = 8.0 Hz and 2.2 Hz, 1 H, H-10), 8.08 (dt, J =
10.1 Hz and 2.9 Hz, 1 H, H-2′), 7.94 (d, J = 8.3 Hz, 2 H, H-10′), 7.57 (d, J = 8.3 Hz, 2 H,
H-11′), 7.51 (dd, J = 8.0 Hz and 4.8 Hz, 1 H, H-9), 7.43 – 7.50 (m, 1 H, H-4′), 7.42 (dd,
J = 5.2 Hz and 2.1 Hz, 1 H, H-3), 7.20 (d, J = 8.3 Hz, 1 H, H-5′), 4.83 (s, 2 H, H-13′), 2.21
(s, 3 H, H-7′). 13C NMR (126 MHz, DMSO-d6): δ /ppm = 164.89 (C-8′), 161.64 (C-4),
161.22 (C-1), 159.55 (CH, C-2), 151.42 (CH, C-8), 148.23 (CH, C-6), 141.00 (C-12′),
137.88 (C-6′), 137.04 (C-1′), 134.95 (C-9′), 134.54 (CH, C-10), 132.28 (C-5), 130.13 (CH,
C-5′), 128.84 (CH, C-10′), 128.07 (CH, C-11′), 123.88 (CH, C-9), 117.14 (CH, C-2′),
116.67 (CH, C-4′), 107.63 (CH, C-3), 45.52 (CH2, C-13′), 17.75 (CH3, C-7′). LC-MS: tR =
4.80 min, m/z 430.18 ([M + H]+). HRMS: m/z calc. for [C24H21N5O35Cl]
+ ([M + H]+)
430.1429, found 430.1415, ∆ = −3.2 ppm.
Experimental 189
N-(4-Methyl-3-((4-(pyridin-3-yl)pyrimidin-2-yl)amino)phenyl)-4-((4-
methylpiperazin-1-yl)methyl)benzamide (Imatinib, 1)33
10
9
8
N
6
N
1
N2
3
H
N
2' 4'
5'
HN 8'
O
10'
11'
13'
N 14'
15'N
16'
7'
Compound 27 (11.5 mg, 27 µmol), N-methylpiperazine (0.05 mL) and acetonitrile (2 mL)
were combined and heated in a sealed vial under microwave irradiation at 150 ◦C for 2 hours.
The solvent was removed and the resulting solid purified by flash column chromatography
to afford the title compound (7 mg, 53 %) as a yellow solid.
1H NMR (400 MHz, DMSO-d6): δ /ppm = 10.17 (s, 1 H, NHamide), 9.27 (d, J = 2.1 Hz,
1 H, H-6), 8.98 (s, 1 H, NH), 8.68 (dd, J = 4.8 Hz and 1.7 Hz, 1 H, H-8), 8.51 (d, J = 5.1 Hz,
1 H, H-2), 8.48 (dt, J = 7.9 Hz and 2.0 Hz, 1 H, H-10), 8.08 (d, J = 2.2 Hz, 1 H, H-2′), 7.90
(d, J = 8.2 Hz, 2 H, H-10′), 7.52 (dd, J = 7.9 Hz and 4.7 Hz, 1 H, H-9), 7.48 (dd, J = 8.2 Hz
and 2.2 Hz, 1 H, H-4′), 7.41 – 7.46 (m, 3 H, H-3 and H-11′), 7.20 (d, J = 8.5 Hz, 1 H, H-5′),
3.52 (s, 2 H, H-13′), 2.34 (m, 8 H, H-14′ and H-15′), 2.22 (s, 3 H, H-7′), 2.14 (s, 3 H, H-16′).
13C NMR (100 MHz, DMSO-d6): δ /ppm = 165.23, 161.57, 161.15, 159.46 (CH), 151.37
(CH), 148.18 (CH), 142.09, 137.76, 137.18, 134.40 (CH), 133.74, 132.19, 130.00 (CH),
128.60 (CH), 127.56 (CH), 123.77 (2 × CH), 117.17 (CH), 116.69 (CH), 107.49 (CH),
61.61 (CH2), 54.70 (CH2), 52.58 (CH2), 45.74 (CH3), 17.66 (CH3). LC-MS: tR = 4.26 min,
m/z 494.48 ([M + H]+). HRMS: m/z calc. for [C29H32N7O]
+ ([M + H]+) 494.2668, found
494.2690, ∆ = −4.5 ppm.
190 Experimental
Piperazine-2-carboxamide synthesis
General Experimental Information
Hydrous zirconia was kindly donated by MEL Chemicals (cod. XZO 631/01).
FlowIR spectrometers from Mettler Toledo, with silicon or diamond windows, were used
for in-line analyses.
The flow hydration reaction was performed using a Vapourtec R2+/R4 flow platform. A
Knauer K-120 HPLC pump was used for the hydrogenation step, in combination with a
ThalesNano H-Cube reactor. Unless otherwise specified, flow tubing is PFA  1 mm with
ETFE connections.
Pyrazine-2-carboxamide (40)296
100 ºC
100 psi
R2
N
N
NH2
O
N
N CN 1´
35
6
FlowIR
A solution of pyrazine-2-carbonitrile (41; 7.4 g, 70 mmol) in EtOH / H2O (0.6 M, 8 : 1 v/v)
was pumped at 0.2 mL/min through column reactor R2 (100 mm × 10 mm, 5 g hydrous
zirconia) heated at 100 ◦C (residence time approximately 20 minutes). Pressure was
provided by a 100 psi back-pressure regulator. Concentration of the reactor output afforded
the title compound (8.5 g, 69 mmol, >98 %) as a colourless solid, m.p. 189.4 – 189.8 ◦C
(ethanol) [lit. 190 ◦C (H2O)
296].
For the control programme, see:
https://gist.github.com/richardingham/0a58a291bad2e3b9009f
1H NMR (400 MHz, DMSO-d6): δ /ppm = 9.18 (d, J = 1.5 Hz, 1 H, H-3), 8.84 (d, J =
2.5 Hz, 1 H, H-6), 8.70 (dd, J = 2.5 Hz and 1.5 Hz, 1 H, H-5), 8.25 (br. s, 1 H, NH), 7.86
(br. s, 1 H, NH). 13C NMR (100 MHz, DMSO-d6): δ /ppm = 165.09 (C-1′), 147.38 (CH,
C-5), 145.08 (C-2), 143.63 (CH, C-6), 143.36 (CH, C-3). IR (neat): ν /cm−1 = 3410 (m),
3148 (m), 1704 (s), 1608 (m), 1579 (m), 1524 (m), 1478 (w), 1436 (m), 1376 (s), 1182 (m),
1164 (m), 1086 (m), 1053 (m), 1022 (s), 869 (m), 785 (s), 669 (s). LC-MS: tR = 0.28 min,
Experimental 191
m/z 124.19 ([M + H]+). HRMS: m/z calc. for [C5H6ON3]
+ ([M + H]+) 124.0505, found
124.0504, ∆ = −0.8 ppm. Microanalysis: calc. (found) for C5H5ON3 C 48.78% (48.69%),
H 4.09% (4.19%), N 34.13% (33.70%).
(R,S)-Piperazine-2-carboxamide (39)297
FlowIRH-Cube
100 psi 75 psi
N
N
NH2
O
N
H
N
H
NH2
O
1´
2
35
6
A solution of pyrazine-2-carboxamide (40; 0.62 mg, 5.0 mmol) in EtOH / H2O (0.6 M, 8 : 1
v/v) was pumped at 0.1 mL/min using a Knauer K-120 HPLC pump (through tubing 
0.5 mm, PTFE then SS) into the H-Cube reactor, which was equipped with a 10 % Pd/C
catalyst cartridge, heated to 100 ◦C. The reactor was held at pressure by a 100 psi and 75 psi
BPR in series (connected to the reactor by tubing  0.5 mm, PTFE). Concentration of the
reactor output afforded the title compound (0.61 g, 4.8 mmol, 95 %) as a colourless solid,
m.p. 143 – 145 ◦C (ethanol) [lit. 144 – 145 ◦C (ethanol)297].
1H NMR (400 MHz, MeOH-d4): δ /ppm = 3.35 (dd, J = 9.9 Hz and 3.3 Hz, 1 H, H-2),
3.09 (dd, J = 12.4 Hz and 3.4 Hz, 1 H, H-3cis), 2.96 (dt, J = 12.2 Hz and 2.8 Hz, 1 H, H-6),
2.90 – 2.77 (m, 1 H, H-5), 2.75 (dd, J = 12.4 Hz and 2.6 Hz, 1 H, H-6), 2.73 – 2.61 (m, 2 H,
H-3trans and H-5). 13C NMR (100 MHz, MeOH-d4): δ /ppm = 176.50 (C-1′), 59.61 (CH,
C-2), 49.69 (CH2, C-3), 46.19 (CH2, C-2), 45.64 (CH2, C-1). IR (neat): ν /cm−1 = 3333
(m), 3308 (m), 3194 (w), 2949 (m), 2904 (m), 2832 (m), 1674 (m), 1611 (m), 1488 (w),
1438 (m), 1409 (m), 1368 (w), 1355 (m), 1306 (s), 1245 (m), 1187 (w), 1137 (m), 1117
(s), 1071 (w), 1057 (m), 1003 (w), 960 (w), 909 (m), 823 (s), 723 (m), 666 (s). LC-MS:
tR = 0.26 min, m/z 130.14 ([M + H]+). HRMS: m/z calc. for [C5H12ON3]
+ ([M + H]+)
130.0975, found 130.0979, ∆ = 3.0 ppm. Microanalysis: calc. (found) for C5H11ON3 C
46.50% (46.49%), H 8.58% (8.50%), N 32.53% (32.30%).
192 Experimental
DoE Reduction Experiments
To the reactor setup above was appended a Valco VICI 10-position switching valve, such
that samples could be collected from the different experiments. See Figure 3.12, page 70.
Representative NMR spectra:
2.53.03.54.05.56.06.57.07.58.08.59.0
ppm
 
Experiment 15
 
Experiment 13
 
Experiment 7
For the control programme, see:
https://gist.github.com/richardingham/83401127622036c6afd0
Experimental 193
Telescoped flow synthesis
(See Figure 3.12, page 70).
A solution of pyrazine-2-carbonitrile 41 in EtOH / H2O (0.6 M, 8 : 1 v/v) was passed through
a zirconia column reactor as described above in the experimental section for compound 40.
The intermediate solution was used directly without purification in the second step. This
could be performed either by matching the flow rates of the two steps, or using a reservoir
arrangement as described in Section 3.5.1. The intermediate solution was delivered to the
H-Cube apparatus (flow rate 0.1 mL/min) using a Knauer K-120 pump as described in the
experimental section for compound 39. Over a total of 14 h of operation, 7.3 g of compound
39 was generated in 95 % yield and 99 % purity.
For control programme, see:
https://gist.github.com/richardingham/31f6f8efa47771c2ed02
194 Experimental
2-Aminoadamantane-2-carboxylic acid synthesis
General Experimental Information
THF, toluene and methanol were obtained from Fisher Scientific and distilled before use.
Diethyl ether was obtained from Sigma-Aldrich and distilled before use.
Integrated flow synthesis
Reagent Stream A A solution of 2-adamantanone (7.5 g, 50 mmol) in toluene / diethyl
ether (3 : 2 v/v, 100 mL), stored in an oven-dried 250 mL, one-necked, pear-shapedflask
equipped with a rubber septum and under a N2 atmosphere.
Reagent Stream B A solution of ethynyl magnesium bromide (0.5 M), in a Sigma-
Aldrich SureSeal bottle equipped with an inert gas inlet connected to a nitrogen manifold.
Solvent Stream A THF stored under a nitrogen atmosphere.
Solvent Stream B Toluene / diethyl ether (3 : 2 v/v) stored under a nitrogen atmosphere.
This can be switched to a stream of distilled methanol using valve V2.
Stage 1 Reagent Stream A was delivered at a flow rate of 0.18 mL/min (P2; Uniqsis
FlowSyn) and combined at tee mixer M1 (Uniqsis FlowSyn integrated tee mixer) with
Reagent Stream B delivered at a flow rate of 0.20 mL/min (P1; Uniqsis FlowSyn) and
then heated to 40 ◦C in reactor R1 ( 1 mm, PFA, 14 mL: 37 minutes residence time). A
100 psi back pressure regulator was employed to regulate the pressure of the system. After
exiting the coil the flow stream passed through a FlowIR spectrometer in order to detect
the presence of the Grignard product 51. The detector was linked to valve V4 (Uniqsis
FlowSyn integrated loop-switching valve) by software control (see Section 4.3.5, page 90)
and the output stream was directed to the next stage only when this product was detected.
Pump P5 (Knauer WellChrom K120) delivered Solvent Stream A at 0.5 mL/min through
valve V4 when the reactor output was being discarded.
Experimental 195
Quenching Stream C A saturated solution of NH4Cl combined in a ratio of 3 : 1 (v/v)
with distilled water (250 mL).
Stage 2 The reaction solution was combined with Quenching Stream C, delivered at
a flow rate of 0.5 mL/min, (P3, Knauer Smartline S100) using a tube-in-tube mixer (M2,
see Section 4.3.4); the output was directed into reactor R2 ( 2.4 mm, 600 mm length,
ETFE). A micro vibration motor was attached to R2, reducing the occurrence of blockages
at M2. The biphasic plug flow system formed was delivered to a glass column ( 6.6 mm,
100 mm length, glass) packed with charcoal and glass beads in order to remove magnesium
aggregates by filtration (see Section 4.3.3, page 85). The resulting clear biphasic system
was then delivered to liquid/liquid separating column S1 ( 10 mm, 100 mm length, glass).
The aqueous phase was withdrawn continuously by pump P4 (Knauer Smartline S100) at a
flow rate based on the position of the interface (see Section 4.3.4, page 88). The organic
layer was forced out of the top of the column into the next stage. A pressure differential
created by the two back-pressure regulators prevented undesired siphoning of liquid through
pump P4.
Stage 3 The organic output from S1 was combined at a tee mixer M3 ( 0.5 mm, PEEK)
with a stream of MeCN delivered at 1.2 mL/min by pump P6 (Knauer WellChrom K120).
The mixture was introduced into a nebulising evaporation device at 20 ◦C (see Section 4.4,
page 92) along with a stream of N2 gas (1.4 bar). The remaining solution was withdrawn
from the bottom of the evaporation chamber using a peristaltic pump (P7) at a flow rate of
1 mL/min. The solution was collected in reservoir S2. The volume of material within S2
was measured by a digital camera (see Section 4.5.1, page 94) and this data was used to
control the flow rate of pump P8 (Vapourtec R2+) in order to match the input and output
flow rates.
Stage 4 This solution was pumped at a flow rate of 0.1 to 1.4 mL/min by pump P8 and
combined at mixer M4 ( 0.5 mm, ETFE) with a stream of concentrated sulfuric acid that
was delivered at 0.05 mL/min by pump P9 (Vapourtec R2+ with acid-resistant fittings).
The mixture was passed through reactor R3 ( 1 mm, PFA, 10 mL: minimum 7 minutes
residence time).
Quenching Stream D A 4 M solution of KOH in EtOH / H2O (4 : 1 v/v).
196 Experimental
Stage 5 The output of reactor R3 was combined with Quenching Stream D in a second
tube-in-tube mixer (M5). Quenching Stream D was delivered at 2 mL/min by pump P10
(Vapourtec R2+); the output was directed into reactor R4 ( 2.4 mm, 300 mm length,
ETFE). Another vibrating device was attached to R4, reducing the occurrence of blockages
at M5. The stream exiting this reactor was directed onto the continuous filtration device
(see Section 4.5.3, page 99) that removed the salts formed by the quenching of the acid to
leave a basic ethanolic solution of compound 52 which was collected in reservoir S3.
Stage 6 The solution collected in reservoir S3 was pumped (P13, Vapourtec R2+) to
reactor R5 ( 2.5 mm, stainless steel, 50 mL: minimum 25 minutes residence time) heated
to 120 ◦C. The reactor output was collected and the solvent removed. The residue was
redissolved in ethyl acetate and washed with water, dried (MgSO4), and then the solvent
removed.
Stage 7 A solution of 2-methyl-4-(adamantane-2′-spiro)-5-methylidene oxazoline (53)
in acetone / H2O (5 : 1 v/v, 0.25 M) was delivered at 2 mL/min by pump P14 (Knauer
WellChrom K120) to meet a stream of ozone (0.5 bar, flow rate 500 mL/min) and these
were combined in a Y-mixer ( 0.5 mm, PEEK) and directed into a 1 mL tube reactor
R6 ( 2.5 mm, 180 mm length, PFA, 20 s residence time). The solution was directed to
column R7 ( 10 mm, 100 mm length, glass) from which the gases can escape, containing
amorphous MnO2 (2 g) through which the liquids percolate and the solution was collected.
This solution was pumped at 0.25 mL/min P15 (Vapourtec R2) and combined at a tee
mixer ( 0.5 mm, ETFE) with a stream of HCl / AcOH / H2O (1 : 5 : 8 v/v/v) delivered
at 0.25 mL/min by pump P16 (Vapourtec R2) and directed into reactor R8 ( 1 mm,
10 mL, PFA) heated to 150 ◦C. The output of the reactor was finally directed into a solvent
evaporator device. The solid recovered from the evaporator was washed with hot acetonitrile
(2×50 mL per gram of product) and diethyl ether (2×50 mL per gram of product), before
being recrystallized from MeOH (0 ◦C) to give 2-aminoadamantane-2-carboxylic acid
(135).
Experimental 197
2-Ethynyladamantan-2-ol (51)298,299
HO
Collected as an off-white solid, by concentration of the solution in acetonitrile m.p. 102 ◦C
(acetonitrile) [lit: 100 – 102 ◦C (ether)300].
An analytical sample was recrystallised from ethanol / water.
1H NMR (400 MHz, CDCl3): δ /ppm = 2.56 (s, 1 H), 2.21 – 2.16 (m, 4 H), 1.96 – 1.96 (m,
3 H), 1.82 – 1.77 (m, 4 H), 1.70 (br. s, 2 H), 1.55 – 1.58 (m, 3 H). 13C NMR (100 MHz,
CDCl3): δ /ppm = 88.51, 72.70 (CH), 72.50, 39.31 (CH), 37.60 (CH2), 35.37 (CH2), 31.54
(CH2), 26.40 (CH), 26.23 (CH). IR (neat): ν /cm−1 = 3477 (w), 3363 (w), 3312 (m), 3179
(m), 2928 (m), 2898 (s), 2858 (m), 2091 (w), 1466 (w), 1449 (w), 1407 (w), 1349 (w),
1326 (w), 1309 (w), 1284 (w), 1244 (w), 1194 (w), 1118 (w), 1101 (m), 1061 (s), 1042
(w), 1023 (s), 998 (s), 943 (w), 925 (m), 879 (w), 835 (w), 802 (w), 755 (m), 693 (w).
LC-MS: tR = 4.63 min, m/z 159.00 ([M – H2O]
+). HRMS: m/z calc. for [C12H17O]
+
([M + H]+) 177.2550, found 177.2558, ∆ = 4.5 ppm. Microanalysis: calc. (found) for
C12H16O C 81.77% (81.76%), H 9.15% (9.10%). The structure was confirmed by X-ray
crystallographic analysis, and deposited at Cambridge University with the unique reference
SL1406.
198 Experimental
2-Ethynyl-2-acetamidoadamantane (52)299
NHO
Collected by concentration of the reaction solution, partitioning between ethyl acetate
and water, then drying and concentration of the organic phase. The title compound was
collected as a white solid, m.p. 147 ◦C (ethyl acetate) [lit: 138 – 139 ◦C (ethanol)299].
1H NMR (400 MHz, CDCl3): δ /ppm = 5.41 (br. s, 1 H), 2.46 (s, 1 H), 2.40 (br. s, 2 H),
2.35 (br. s, 1 H), 2.31 (br. s, 1 H), 2.01 (s, 3 H), 1.82 – 1.77 (m, 4 H), 1.70 (br. s, 2 H),
1.55 – 1.58 (m, 3 H). 13C NMR (100 MHz, CDCl3): δ /ppm = 169.12, 86.18, 71.73 (CH),
56.47, 37.64 (CH2), 35.06 (CH), 34.37 (CH2), 32.18 (CH2), 27.24 (CH), 26.76 (CH), 24.23
(CH3). IR (neat): ν /cm−1 = 3287 (m), 2903 (m), 2166 (w), 1645 (s), 1537 (s), 670 (m);
LC-MS: tR = 4.39 min, m/z 218.14 ([M + H]+). HRMS: m/z calc. for [C14H20NO]
+ ([M +
H]+) 218.1545, found 218.1548, ∆ = 1.4 ppm. Microanalysis: calc. (found) for C14H19NO
C 77.78% (77.75%), H 8.81% (8.75%), N 6.45% (6.40%).
2-Methyl-4-(adamantane-2′-spiro)-5-methylidene oxazoline (53)301
N O
Collected as a yellow solid after Stage 6, m.p. 62 ◦C (ethyl acetate).
1H NMR (400 MHz, CDCl3): δ /ppm = 4.79 (d, 1 H, J = 2.4 Hz), 4.52 (d, 1 H, J = 2.4 Hz),
2.44 (d, 2 H, J = 12.4 Hz), 2.21 (d, 2 H, J = 12.4 Hz), 2.03 (s, 3 H), 1.86 (br. s, 3 H), 1.74 (s,
2 H), 1.69 – 1.55 (m, 5 H). 13C NMR (100 MHz, CDCl3): δ /ppm = 166.00, 158.67, 87.92
(CH2), 76.03, 38.70 (CH2), 37.08 (CH), 34.06 (CH2), 31.40 (CH2), 27.67 (CH), 26.41
(CH), 14.11 (CH3). IR (neat): ν /cm−1 = 3342 (w), 3031 (w), 2938 (w), 2801 (w), 1643
(w), 1495 (w), 1453 (w), 1363 (w), 1337 (w), 1282 (w), 1239 (w), 1212 (w), 1167 (w),
1139 (w), 1062 (s), 1042 (m), 1017 (m), 979 (w), 953 (w), 911 (w), 849 (w), 789 (m), 737
Experimental 199
(s), 698 (s), 662 (m). LC-MS: tR = 4.49 min, m/z 218.07 ([M + H]+). HRMS: m/z calc. for
[C14H20NO]
+ ([M + H]+) 218.1545, found 218.1541, ∆ = −1.8 ppm. Microanalysis: calc.
(found) for C14H19NO C 77.78% (77.80%), H 8.81% (8.77%), N 6.45% (6.44%).
2-Aminoadamantane-2-carboxylic acid hydrochloride (44)194
H2N OH
O
HCl
Collected as an off-white solid after recrystallisation from methanol, m.p. >258 ◦C (de-
comp.) [lit: 312 – 314 ◦C (decomp.)192].
1H NMR (400 MHz, CDCl3): δ /ppm = 8.70 (br. s, 3 H), 2.23 (s, 2 H), 2.16 (d, 2 H, J =
13.2 Hz), 1.93 (d, 2 H, J = 13.2 Hz), 1.73 – 1.90 (m, 4 H), 1.62 – 1.66 (m, 2 H). 13C NMR
(100 MHz, CDCl3): δ /ppm = 171.59, 63.92, 37.54 (CH2), 34.41 (CH), 32.25 (CH2), 31.22
(CH), 26.27 (CH). IR (neat): ν /cm−1 = 3223 (w), 2930 (m), 2874 (m), 2775 (m), 1729 (s),
1589 (m), 1507 (s), 1476 (m), 1373 (m), 1282 (w), 1239 (m), 1221 (m), 1146 (s), 1119 (w),
1100 (s), 1087 (m), 1066 (w), 1006 (w), 966 (w), 931 (w), 904 (w), 857 (m), 837 (m), 752
(w), 717 (s), 676 (w). LC-MS: tR = 0.31 min, m/z 196.10 ([M + H]+). HRMS: m/z calc. for
[C11H18NO2]
+ ([M – Cl]+) 196.1338, found 196.1336, ∆ = −1.0 ppm. Microanalysis: calc.
(found) for C11H18NO2Cl: C 57.02% (56.90%), H 7.83% (7.97%), N 6.04% (6.06%).
200 Experimental
Chemical Probes for BRD9
tert-Butyl (3,6-dichloropyridazin-4-yl)carbamate (87)302
N
N
Cl
H
Nt-BuO
O
Cl
For reactor scheme, see Figure 5.16, page 120.
Reagent Stream A: 3,6-Dichloropyridazine-4-carboxylic acid (8.0 g, 40 mmol), tri-
ethylamine (11.6 mL, 80 mmol, distilled over CaCl2 before use) and tert-butanol (18.5 g,
250 mmol) in toluene / acetonitrile (7 : 3 v/v, 125 mL).
Reagent Stream B: Diphenyl phosphoryl azide (13.5 mL, 75 mmol) in toluene / ace-
tonitrile (7 : 3 v/v, 125 mL).
Method: The reagents were stored in sealed flasks, pressurised to 1 bar under a nitrogen
atmosphere, and each stream was drawn through a pump at 0.179 mL/min. The streams
are combined ( 1 mm PFA tubing) in a tee mixer ( 0.5 mm, ETFE) and then heated at
120 ◦C in a continuous flow coil ( 2.5 mm, 2×25 mL volume, SS) before passing through
a 100 psi back-pressure regulator.
The output was collected and the solvent removed. The residue was extracted into toluene
(2×250 mL) from 0.1 M aqueous citric acid (50 mL). The combined organic extracts
were dried (MgSO4) and concentrated. The resultant mixture was purified by column
chromatography (50 g KP-NH silica), using a solvent system of ethyl acetate / hexane (0 : 1
to 2 : 8 over 20 CV, then 2 : 8 for 5 CV) to afford the title compound (8.5 g, 32 mmol, 39 %)
as a white solid, m.p. 110 – 111 ◦C (ethyl acetate) [lit: 81 – 83 ◦C302].
1H NMR (400 MHz, CDCl3): δ /ppm = 8.36 (s, 1 H), 7.20 (br. s, 1 H, NH), 1.56 (s, 9 H).
13C NMR (100 MHz, CDCl3): δ /ppm = 156.51, 150.88, 144.86, 137.28, 114.58 (CH),
84.11, 28.00 (CH3). IR (neat): ν /cm−1 = 3286 (w), 3186 (w), 3146 (w), 3086 (w), 2978
(w), 2933 (w), 2216 (w), 2168 (w), 2030 (w), 1994 (w), 1979 (w), 1760 (w), 1732 (s), 1693
Experimental 201
(w), 1638 (w), 1590 (w), 1547 (s), 1489 (s), 1457 (m), 1411 (w), 1393 (w), 1367 (m), 1341
(m), 1277 (m), 1265 (w), 1237 (s), 1188 (m), 1151 (s), 1131 (s), 1118 (s), 1056 (m), 1026
(m), 1010 (m), 959 (s), 882 (m), 860 (s), 788 (m), 753 (m), 740 (m), 688 (m). HRMS: m/z
calc. for C9H12O2N3[
35Cl]2 ([M + H]
+) 264.0301, found 264.0292, ∆ = −3.3 ppm.
3,6-Dichloropyridazin-4-amine (88)302
N
N
Cl
H2N
Cl
tert-Butyl (3,6-dichloropyridazin-4-yl)carbamate (87) (10 g, 37.8 mmol) is dissolved in
dichloromethane (40 mL). 4 M HCl in dioxane (125 mL) was added, with rapid formation
of a precipitate. The reaction mixture was allowed to stir at 25 ◦C for 1 day, and then the
solid was collected by filtration and resuspended in DCM (100 mL). 2 M NH3 in methanol
(50 mL) was added and stirred at 25 ◦C for 30 minutes. The solid product was collected by
filtration, washing with DCM and then dried in vacuo. The title compound (6.2 g, quant.)
was collected as a white solid, m.p. 204.6 – 204.8 ◦C (dichloromethane) [lit: 194 – 196 ◦C
(ethyl acetate)302].
1H NMR (400 MHz, DMSO-d6): δ /ppm = 7.1 (br. s, 2 H, NH2), 6.8 (s, 1 H).
13C
NMR (100 MHz, DMSO-d6): δ /ppm = 154.53, 146.26, 143.72, 108.49 (CH). IR (neat):
ν /cm−1 = 3459 (w), 3290 (w), 3209 (w), 3094 (w), 3049 (m), 2795 (w), 2694 (w), 1635
(s), 1563 (s), 1517 (w), 1439 (w), 1383 (m), 1293 (m), 1125 (s), 1072 (m), 1053 (s), 962
(s), 882 (s), 727 (s), 684 (m). HRMS: m/z calc. for C4H4N3[
35Cl]2 ([M + H]
+) 163.9782,
found 163.9783, ∆ = 0.6 ppm. Microanalysis: calc. (found) for C4H4N3Cl2 C 29.30%
(29.41%), H 1.84% (1.82%), N 25.62% (24.61%), Cl 43.23 (43.44%).
202 Experimental
3,6-Dichloro-3-hydrazinylpyridazin-4-amine (91)302
N
N
Cl
H2N
HN
NH2
A mixture of 3,6-dichloropyridazin-4-amine (88) (3.0 g, 18.2 mmol), hydrazine hydrate
(27 mL) and water (8 mL) was heated in two portions in sealed vials under microwave
irradiation at 100 ◦C for 3 h. The reaction mixture was allowed to cool to room temperature,
and the resulting slurry collected by filtration, washing on the filter with water (10 mL).
The title compound (2.2 g, 13.6 mmol, 75 %) was collected as a white solid.
1H NMR (400 MHz, DMSO-d6): δ /ppm = 7.27 (br. s, 1 H), 6.36 (s, 1 H), 6.24 (br. s,
2 H), 4.23 (br. s, 2 H). 13C NMR (100 MHz, DMSO-d6): δ /ppm = 143.93, 137.58, 103.59,
91.83 (CH). IR (neat): ν /cm−1 = 3317 (w), 3285 (w), 3101 (m), 3045 (w), 2984 (m), 2931
(w), 2763 (w), 2272 (w), 2254 (w), 2216 (w), 2199 (w), 2168 (w), 2138 (w), 2027 (w),
1988 (w), 1980 (w), 1972 (w), 1958 (w), 1670 (w), 1577 (s), 1558 (m), 1456 (m), 1417
(w), 1394 (w), 1344 (w), 1308 (w), 1276 (m), 1177 (w), 1145 (w), 1100 (m), 991 (m), 960
(m), 879 (m), 824 (m), 761 (w), 695 (m). HRMS: m/z calc. for C4H7N5[
35Cl] ([M + H]+)
160.0390, found 160.0396, ∆ = 3.7 ppm.
6-Chloro-3-methyl-[1,2,4]triazolo[4,3-b]pyridazin-8-amine (76)235
N
N
N N
Cl
H2N
A mixture of 3,6-dichloro-3-hydrazinylpyridazin-4-amine (91) (2.4 g, 15 mmol) and acetic
acid (30 mL) was heated in two portions in sealed vials under microwave irradiation at
120 ◦C for 2 h. The solvent was removed and the combined residue recrystallized from
10 % water in ethanol. The title compound (1.35 g, 7 mmol, 49 %) was collected as a pale
yellow crystalline solid, m.p. 233.5 – 234.0 ◦C (decomp.).
1H NMR (400 MHz, DMSO-d6): δ /ppm = 7.85 (br. s, 2 H), 6.09 (s, 1 H), 2.48 (s, 3 H).
13C NMR (100 MHz, DMSO-d6): δ /ppm = 149.77, 147.16, 144.31, 139.79, 94.01 (CH),
Experimental 203
9.90 (CH3). IR (neat): ν /cm−1 = 3451 (m), 3299 (w), 3214 (w), 3103 (m), 3073 (m), 1651
(s), 1596 (w), 1563 (s), 1529 (m), 1466 (s), 1421 (s), 1411 (s), 1379 (m), 1339 (m), 1243
(m), 1188 (w), 1129 (s), 1051 (m), 994 (w), 929 (s), 810 (s), 794 (m), 745 (m), 718 (s),
669 (m). HRMS: m/z calc. for C6H7N5[
35Cl] ([M + H]+) 184.0390, found 184.0392, ∆ =
1.1 ppm.
2-Methyl-5-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)aniline (100)303
B
OO
NH2
(3-Amino-4-methylphenyl)boronic acid (0.50 g, 3.3 mmol) and pinacol (0.43 g, 3.6 mmol)
in acetonitrile (5 mL) was stirred at 25 ◦C for 18 h. The solvent was removed and the
resulting solid triturated with CH2Cl2 to afford the title compound (0.5 g, 65 %) as a pale
orange crystalline solid.
An analytical sample was recrystallised from ethanol, m.p. 115.2 – 115.4 ◦C (ethanol) [lit:
107 – 108 ◦C303].
1H NMR (400 MHz, CDCl3): δ /ppm = 7.16 (d, J = 7.2 Hz, 1 H, H-4), 7.12 (s, 1 H, H-8),
7.07 (d, J = 7.2 Hz, 1 H, H-5), 3.57 (br. s, 2 H, NH2), 2.18 (s, 3 H, H-6
′), 1.33 (s, 12 H,
H-1). 13C NMR (100 MHz, CDCl3): δ /ppm = 144.18, 130.12 (CH), 125.99, 125.35 (CH),
121.17 (CH), 83.70, 24.98 (CH3), 17.72 (CH3), (One carbon resonance was not visible). IR
(neat): ν /cm−1 = 3421 (w), 3352 (w), 3003 (w), 2979 (w), 2931 (w), 1632 (w), 1607 (w),
1564 (w), 1516 (w), 1454 (w), 1419 (m), 1378 (m), 1366 (m), 1349 (s), 1320 (m), 1297 (m),
1285 (m), 1260 (s), 1214 (w), 1203 (m), 1169 (m), 1137 (s), 1104 (m), 1064 (w), 1035 (w),
997 (m), 967 (m), 922 (w), 892 (w), 855 (s), 832 (w), 812 (s), 787 (w), 753 (w), 727 (m),
701 (m), 685 (s), 672 (m). HRMS: m/z calc. for C13H21[
11B]NO2 ([M + H]
+) 234.1665,
found 234.1662, ∆ = −1.3 ppm.
204 Experimental
N-(2-methyl-5-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)phenyl)
methanesulfonamide (102e)
4
5
8
B
OO
1
N
H
S
9
OO
6'
Compound 100 (1.5 g, 6.6 mmol), mesyl chloride (1.0 mL, 13 mmol, 2 equiv.), and pyridine
(1.0 mL, 13 mmol) in acetonitrile (4 mL) were stirred at 25 ◦C for 1 h. The solvent was
removed and the solid product partitioned between ethyl acetate (10 mL) and aqueous
HCl (0.2 M, 10 mL). The aqueous phase was extracted with ethyl acetate (10 mL), and the
combined organic phases dried (MgSO4), concentrated and recrystallized from ethanol
to afford the title compound (637 mg, 50 %) as an off-white solid, m.p. 146.5 – 146.8 ◦C
(ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.77 (s, 1 H, H-8), 7.60 (d, J = 7.6 Hz, 1 H, H-
4), 7.26 (d, J = 7.6 Hz, 1 H, H-5), 6.10 (br. s, 1 H, NH), 3.06 (s, 3 H, H-9), 2.38 (s,
3 H, H-6′), 1.33 (s, 12 H, H-1). 13C NMR (100 MHz, CDCl3): δ /ppm = 135.76, 133.30,
130.99, 130.25, 115.64, 84.11 (CH), 40.37 (CH3), 25.02 (CH3), 18.60 (CH3), (One carbon
resonance was not visible). IR (neat): ν /cm−1 = 3277 (w), 2980 (w), 1611 (w), 1512
(w), 1474 (w), 1449 (w), 1405 (m), 1370 (m), 1352 (m), 1317 (s), 1283 (w), 1271 (w),
1215 (w), 1191 (w), 1161 (m), 1149 (s), 1131 (m), 1089 (m), 1038 (w), 1000 (w), 985
(m), 963 (m), 931 (w), 919 (w), 853 (m), 838 (w), 828 (w), 797 (w), 776 (m), 722 (w),
697 (w), 683 (w), 671 (m). HRMS: m/z calc. for C14H23[
11B]NO4S ([M + H]
+) 312.1441,
found 312.1435, ∆ = −1.9 ppm. Microanalysis: calc. (found) for C13H20BNO2 C 54.03%
(53.83%), H 7.13% (7.07%), N 4.50% (4.51%).
Experimental 205
General Procedure A for sulfonamide formation
Compound 100, the aryl sulfonyl chloride (1.1 equiv.), and pyridine (0.5 mL, 2 equiv.)
in ethanol (4 mL) were stirred at 25 ◦C for 18 h. The solvent was removed and the solid
product partitioned between ethyl acetate (5 mL) and water (5 mL). The organic phase was
dried (MgSO4), concentrated and recrystallized from ethanol.
2-Chloro-N-(2-methyl-5-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)
phenyl)benzenesulfonamide (102a)
4
5
8
B
OO
1
N
H
S
13
12
11
10
Cl
OO
6'
Prepared according to General Procedure A on 3.3 mmol scale. Recrystallisation from
ethanol (two rounds) afforded the title compound (914 mg, 68 %) as a colourless crystalline
solid, m.p. 177.6 – 177.8 ◦C (ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.99 (dd, J = 8.0 Hz and 1.6 Hz, 1 H, H-10), 7.54
(dd, J = 8.0 Hz and 1.3 Hz, 1 H, H-13), 7.51 – 7.41 (m, 3 H, H-4, H-8 and H-12), 7.32 (td,
J = 8.0 Hz and 1.3 Hz, 1 H, H-11), 7.13 (d, J = 7.4 Hz, 1 H, H-5), 6.72 (br. s, 1 H, NH),
2.29 (s, 3 H, H-6′), 1.27 (s, 12 H, H-1), (One carbon resonance was not visible). 13C NMR
(100 MHz, CDCl3): δ /ppm = 137.35, 136.01, 133.98 (CH), 133.68, 132.92 (CH), 132.07
(CH), 131.75 (CH), 131.67, 130.69 (CH), 130.34 (CH), 127.19 (CH), 83.91, 24.97 (CH3),
18.27 (CH3). IR (neat): ν /cm−1 = 3371 (w), 2980 (w), 1615 (w), 1453 (w), 1436 (w),
1380 (s), 1372 (s), 1354 (s), 1331 (s), 1317 (s), 1284 (w), 1251 (w), 1235 (w), 1217 (w),
1160 (s), 1144 (s), 1127 (s), 1111 (s), 1087 (s), 1042 (s), 996 (w), 967 (s), 947 (w), 916
(w), 880 (w), 853 (s), 828 (w), 771 (s), 751 (s), 719 (s), 710 (w), 696 (w), 683 (w), 665
(s). HRMS: m/z calc. for C19H23[
11B][35Cl]NO4S ([M + H]
+) 408.1208, found 408.1199,
∆ = −2.2 ppm. Microanalysis: calc. (found) for C19H22BClNO4S C 55.97% (55.92%), H
5.69% (5.61%), N 3.44% (3.49%), Cl 8.69% (8.89%).
206 Experimental
3-Chloro-N-(2-methyl-5-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-
yl)phenyl)benzenesulfonamide (102b)
4
5
8
B
OO
1
N
H
S
9
10
11
13OO
6'
Cl
Prepared according to General Procedure A on 3.3 mmol scale. Recrystallisation from
ethanol (two rounds) afforded the title compound (741 mg, 55 %) as a colourless crystalline
solid, m.p. 138.4 – 138.6 ◦C (ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.73 (t, J = 1.9 Hz, 1 H, H-14), 7.64 – 7.47 (m,
4 H, H-4, H-8, H-10 and H-12), 7.37 (t, J = 7.9 Hz, 1 H, H-11), 7.15 (d, J = 7.5 Hz, 1 H,
H-5), 6.22 (br. s, 1 H, NH), 2.10 (s, 3 H, H-6′), 1.32 (s, 12 H, H-1). 13C NMR (100 MHz,
CDCl3): δ /ppm = 141.36, 136.70, 135.29, 133.59 (CH), 133.46, 133.11 (CH), 132.19
(CH), 130.74 (CH), 130.35 (CH), 127.64 (CH), 125.58 (CH), 84.05, 25.00 (CH3), 18.14
(CH3), (One carbon resonance was not visible). IR (neat): ν /cm−1 = 3371 (w), 2980 (w),
1615 (w), 1578 (w), 1560 (w), 1510 (w), 1453 (w), 1436 (w), 1380 (s), 1372 (s), 1354
(s), 1331 (s), 1317 (s), 1284 (m), 1251 (m), 1235 (w), 1217 (w), 1160 (s), 1144 (s), 1127
(s), 1111 (m), 1087 (m), 1042 (m), 996 (m), 967 (m), 947 (m), 916 (w), 880 (w), 853
(s), 828 (m), 771 (s), 751 (s), 719 (m), 710 (m), 696 (w), 683 (m), 665 (s). HRMS: m/z
calc. for [C19H23[
11B][35Cl]NO4S]
+ ([M + H]+) 408.1208, found 408.1202, ∆ = −1.5 ppm.
Microanalysis: calc. (found) for C19H22BClNO4S C 55.97% (55.93%), H 5.69% (5.64%),
N 3.44% (3.54%), Cl 8.69% (8.71%).
Experimental 207
4-Chloro-N-(2-methyl-5-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)
phenyl)benzenesulfonamide (102c)
4
5
8
B
OO
1
N
H
S 11
10OO
6' Cl
Prepared according General Procedure A on 2.7 mmol scale. Recrystallisation from ethanol
(two rounds) afforded the title compound (852 mg, 77 %) as a colourless crystalline solid,
m.p. 199.1 – 199.3 ◦C (ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.62 (d, J = 8.6 Hz, 2 H, H-10), 7.55 (d, J =
7.5 Hz, 1 H, H-4), 7.47 (s, 1 H, H-8), 7.40 (d, J = 8.6 Hz, 2 H, H-11), 7.15 (d, J = 7.5 Hz,
1 H), 6.14 (br. s, 1 H, NH), 2.09 (s, 3 H, H-6′), 1.32 (s, 12 H, H-1). 13C NMR (100 MHz,
CDCl3): δ /ppm = 139.54 (C-12), 138.10 (C-9), 136.94 (CAr), 133.58 (CH, CAr), 133.52
(CAr), 132.47 (CH, CAr), 130.73 (CH, CAr), 129.33 (CH, C-11), 128.96 (CH, C-10), 84.05
(C-2), 24.99 (CH3, C-6
′), 18.16 (CH3, C-1), (One carbon resonance was not visible). IR
(neat): ν /cm−1 = 3219 (w), 2977 (w), 2162 (w), 2033 (w), 2010 (w), 2000 (w), 1988
(w), 1976 (w), 1614 (w), 1586 (w), 1575 (w), 1515 (w), 1477 (w), 1446 (w), 1409 (m),
1383 (m), 1372 (m), 1350 (s), 1333 (s), 1278 (m), 1218 (w), 1188 (m), 1164 (s), 1145
208 Experimental
(s), 1129 (s), 1111 (m), 1089 (s), 1037 (w), 1013 (m), 997 (w), 965 (m), 936 (m), 920
(m), 884 (w), 854 (m), 836 (m), 822 (m), 798 (m), 756 (s), 723 (m), 696 (s), 681 (s).
HRMS: m/z calc. for C19H23[
11B][35Cl]NO4S ([M + H]
+) 408.1208, found 408.1198, ∆ =
−2.5 ppm. Microanalysis: calc. (found) for C19H22BClNO4S C 55.97% (55.94%), H 5.69%
(5.64%), N 3.44% (3.41%), Cl 8.69% (8.56%). The structure was confirmed by X-ray
crystallographic analysis, and deposited at Cambridge University with the unique reference
SL1242.
3,4-Dichloro-N-(2-methyl-5-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-
2-yl)phenyl)benzenesulfonamide (102d)
4
5
8
B
OO
1
N
H
S
10
11
14OO
6' Cl
Cl
Prepared according General Procedure A on 3.3 mmol scale. Recrystallisation from ethanol
(two rounds) afforded the title compound (922 mg, 63 %) as a colourless crystalline solid,
m.p. 176.2 – 176.4 ◦C (ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.81 (s, 1 H, H-14), 7.57 (d, J = 7.5 Hz, 1 H, H-4),
7.51 (m, 2 H, H-10 and H-11), 7.40 (s, 1 H, H-8), 7.18 (d, J = 7.5 Hz, 1 H, H-5), 6.29 (br. s,
1 H, NH), 2.17 (s, 3 H, H-6′), 1.31 (s, 12 H, H-1). 13C NMR (100 MHz, CDCl3): δ /ppm =
139.28, 137.86, 137.23, 133.83 (CH), 133.69, 133.18, 132.35 (CH), 131.11 (CH), 130.91
(CH), 129.56 (CH), 126.57 (CH), 126.36, 84.09, 24.99 (CH3), 18.29 (CH3). IR (neat):
ν /cm−1 = 3202 (w), 2975 (w), 1617 (w), 1520 (w), 1452 (w), 1424 (m), 1394 (w), 1386
(w), 1356 (s), 1339 (s), 1310 (m), 1272 (m), 1217 (w), 1192 (m), 1167 (s), 1141 (s), 1129
(s), 1090 (s), 1033 (m), 998 (w), 963 (m), 942 (m), 914 (w), 889 (w), 855 (m), 822 (s),
813 (s), 798 (m), 767 (w), 700 (s), 677 (s). HRMS: m/z calc. for C19H24[
11B][35Cl]2NO4S
([M + H]+) 442.0818, found 442.0806, ∆ = −2.7 ppm. Microanalysis: calc. (found) for
C19H23BCl2NO4S C 51.61% (51.46%), H 5.02% (4.92%), N 3.17% (3.14%), Cl 16.03%
(15.76%).
Experimental 209
General Procedure B for Aryl Coupling
Triazolopyridazine 76 (0.1 mmol), the boronic ester (1.1 equiv.), potassium phosphate (4
equiv.) and [1,1′-bis(di-tert-butylphosphino)ferrocene]dichloropalladium(II) (10 mol%)
were combined with n-butanol / water (7 : 3 v/v, 2 mL) in a sealed vial. The solvent was
degassed under vacuum by three freeze-thaw cycles, and the vial was then backfilled with
N2. The reaction mixture was heated to 100
◦C for 4 h under microwave irradiation. The
solvent was removed, and the resulting mixture redissolved in methanol / water (1 : 1 v/v)
and purified by preparative HPLC using a methanol / water gradient.
N-(5-(8-Amino-3-methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)-2-meth-
ylphenyl)-2-chlorobenzenesulfonamide (103b)
5 N
N
N N
1
12
9
8
10'
N
H
S 17
16
15
14
OO
H2N
Cl
Prepared according to General Procedure B above. Purification by HPLC afforded the title
compound (15 mg, 35 %) as a white solid.
1H NMR (400 MHz, CDCl3): δ /ppm = 8.07 (dd, J = 7.9 Hz and 1.5 Hz, 1 H, H-19), 7.79
(d, J = 1.8 Hz, 1 H, H-12), 7.64 – 7.44 (m, 3 H, H-8, H-15 and H-9 or H-22), 7.36 (t,
J = 7.4 Hz, 1.5 Hz, 1 H, H-21), 7.25 (d, J = 7.4 Hz, 1 H, H-9 or H-22), 6.98 (br. s, 1 H,
NH), 6.41 (s, 1 H, H-5), 5.57 (br. s, 2 H, NH2), 2.78 (s, 3 H, H-1), 2.34 (s, 3 H, H-10
′).
13C NMR (126 MHz, CDCl3): δ /ppm = 154.16, 148.03, 141.15, 140.02, 136.97, 134.84,
134.70, 134.24 (CH), 132.92, 131.82 (CH), 131.82 (CH), 131.54, 131.49 (CH), 127.29
(CH), 124.45 (CH), 121.18 (CH), 93.69 (CH), 17.69 (CH3), 10.00 (CH3). HRMS: m/z calc.
for C19H18N6O2S[
35Cl] ([M + H]+) 429.0900, found 429.0901, ∆ = 0.2 ppm.
210 Experimental
N-(5-(8-Amino-3-methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)-2-meth-
ylphenyl)-3-chlorobenzenesulfonamide (103c)
5 N
N
N N
1
12
9
8
10'
N
H
S
18
16
15
14
OO
H2N
Cl
Prepared according to General Procedure B above. Purification by HPLC afforded the title
compound (1 mg, 4 %) as a white solid.
1H NMR (400 MHz, CDCl3): δ /ppm = 7.85 (d, J = 1.9 Hz, 1 H, H-12), 7.81 (t, J = 1.9 Hz,
1 H, H-18), 7.73 (dd, J = 8.0 Hz and 1.9 Hz, 1 H, H-14), 7.64 (d, J = 8.0 Hz, 1 H, H-16),
7.55 (d, J = 8.2 Hz, 1 H, H-8), 7.40 (t, J = 8.0 Hz, 1 H, H-15), 7.26 (d, J = 8.2 Hz, 1 H, H-9),
6.48 (s, 1 H, H-5), 6.41 (br. s, 1 H, NH), 5.43 (br. s, 2 H, NH2), 2.81 (s, 3 H, H-1), 2.10
(s, 3 H, H-10′). 13C NMR (126 MHz, CDCl3): δ /ppm = 153.99, 148.13, 141.23, 141.16,
140.02, 135.33, 135.07, 134.42, 133.62, 133.28 (CH), 131.47 (CH), 130.44 (CH), 127.25
(CH), 125.29 (CH), 125.22 (CH), 123.27 (CH), 93.72 (CH), 17.50 (CH3), 10.00 (CH3).
HRMS: m/z calc. for C19H18N6O2S[
35Cl] ([M + H]+) 429.0900, found 429.0909, ∆ =
2.1 ppm.
N-(5-(8-Amino-3-methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)-2-meth-
ylphenyl)-4-chlorobenzenesulfonamide (103d)
5 N
N
N N
1
12
9
8
10'
N
H
S
14
15
OO
H2N
Cl
Prepared according to the General Procedure B above. Purification by HPLC afforded the
title compound (12 mg, 32 %) as a white solid.
Experimental 211
1H NMR (500 MHz, DMSO-d6): δ /ppm = 9.81 (br. s, 1 H, NH), 7.73 – 7.52 (m, 7 H),
7.46 (br. s, 2 H, NH2), 7.27 (d, J = 8.0 Hz, 1 H, H-9), 6.45 (s, 1 H, H-5), 2.61 (s, 3 H, H-1),
2.06 (s, 3 H, H-10′). 13C NMR (126 MHz, DMSO-d6): δ /ppm = 153.35 (C-6), 146.70
(C-2), 142.85 (C-3), 140.12 (C-16), 139.86 (C-4), 137.24 (C-13), 135.62 (C-10), 134.08
(C-7), 131.10 (CH, C-9), 129.29 (CH, C-15), 128.50 (CH, C-14), 123.69 (CH, C-12), 91.45
(CH, C-5), 17.66 (CH3, C-10
′), 9.53 (CH3, C-1), (C-8 and C-11 were not visible). HRMS:
m/z calc. for C19H18[
35Cl]N6O2S ([M + H]
+) 429.0900, found 429.0894, ∆ = −1.4 ppm.
N-(5-(8-Amino-3-methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)-2-meth-
ylphenyl)-3,4-dichlorobenzenesulfonamide (103e)
5 N
N
N N
1
12
9
8
10'
N
H
S
18
15
14
OO
H2N
Cl
Cl
Prepared according to the General Procedure B above. Purification by HPLC afforded the
title compound (9 mg, 13 %) as a white solid.
1H NMR (400 MHz, DMSO-d6): δ /ppm = 10.00 (br. s, 1 H, NH), 7.84 (d, J = 2.1 Hz,
1 H, H-18), 7.75 (d, J = 8.3 Hz, 1 H, H-15), 7.64 (dd, J = 8.4 Hz and 2.1 Hz, 1 H, H-14),
7.54 (d, J = 1.9 Hz, 1 H, H-12), 7.41 (br. s, 2 H, NH2), 7.46 – 7.33 (m, 1 H, H-8), 7.19
(d, J = 7.8 Hz, 1 H, H-9), 6.44 (s, 1 H, H-5), 2.60 (s, 3 H, H-1), 2.10 (s, 3 H, H-10′). 13C
NMR (126 MHz, DMSO-d6): δ /ppm = 154.13 (C-6), 146.67 (C-2), 142.74 (C-3), 139.97
(C-4), 134.49 (C-11), 133.97 (C-13), 133.82 (C-7), 131.47 (C-16), 131.19 (CH, C-15),
130.58 (CH, C-9), 128.19 (CH, C-18), 126.67 (CH, C-14), 91.73 (CH, C-5), 18.23 (CH3,
C-10′), 9.58 (CH3, C-1), (C-8, C-10, C-12 and C-17 were not visible). HRMS: m/z calc.
for C19H17[
35Cl]2N6O2S ([M + H]
+) 463.0511, found 463.0505, ∆ = −1.3 ppm.
212 Experimental
Ethyl (6-(3-((2-chlorophenyl)sulfonamido)-4-methylphenyl)-3-meth-
yl-[1,2,4]triazolo[4,3-b]pyridazin-8-yl)carbamate (108b)
5 N
N
N N
1
12
9
8
10'
N
S 17
16
15
OO
H2N
O O
Cl
20
21
To a solution of compound 103b (1 mg), N,N-dimethylaminopyridine (3 mg) and triethy-
lamine (8 µL) in tetrahydrofuran (5 mL) under an N2 atmosphere was added a solution
of ethyl chloroformate in tetrahydrofuran (0.1 M, 0.4 mL) in two portions. The mixture
was stirred at room temperature for 20 minutes. Aqueous citric acid (0.1 M, 5 mL) was
added, and the mixture concentrated under reduced pressure. The residue was dissolved in
methanol / water (1 : 1 v/v) and purified by HPLC to afford the title compound (<1 mg).
1H NMR (400 MHz, CDCl3): δ /ppm = 8.43 (d, J = 8.1 Hz and 1.9 Hz, 1 H), 8.10 (d, J =
1.9 Hz, 1 H), 7.88 (dd, J = 8.0 Hz and 1.9 Hz, 1 H), 7.66 – 7.56 (m, 2 H), 7.56 – 7.49 (m,
1 H), 7.45 (d, J = 8.0 Hz, 1 H), 6.52 (s, 1 H), 5.45 (s, 2 H), 4.25 – 4.04 (m, 2 H), 2.83 (s,
3 H), 2.49 (s, 3 H), 1.13 (t, J = 7.1 Hz, 3 H).
Experimental 213
Ethyl (6-(3-((3-chlorophenyl)sulfonamido)-4-methylphenyl)-3-meth-
yl-[1,2,4]triazolo[4,3-b]pyridazin-8-yl)carbamate (108c)
5 N
N
N N
1
12
9
8
10'
N
S
18
16
15
OO
H2N
O O
Cl
20
21
To a solution of compound 103c (1 mg), N,N-dimethylaminopyridine (5.6 mg) and triethy-
lamine (8 µL) in tetrahydrofuran (5 mL) under an N2 atmosphere was added a solution of
ethyl chloroformate in tetrahydrofuran (0.1 M, 0.2 mL). The mixture was stirred at room
temperature for 10 minutes. Aqueous citric acid (0.1 M, 5 mL) was added, and the mixture
concentrated under reduced pressure. The residue was dissolved in methanol / water (1 : 1
v/v) and purified by HPLC to afford the title compound (<1 mg).
1H NMR (400 MHz, CDCl3): δ /ppm = 8.10 (t, J = 1.9 Hz, 1 H, H-18), 8.03 (d, J = 7.8,
1 H, H-14), 7.91 (dd, J = 8.0 Hz and 1.9 Hz, 1 H, H-8), 7.73 – 7.64 (m, 2 H, H-12 and H-16),
7.55 (t, J = 8.0 Hz, 1 H, H-15), 7.46 (d, J = 8.0 Hz, 1 H, H-9), 6.50 (s, 1 H, H-5), 5.50 (s,
2 H, NH2), 4.31 – 4.07 (m, NH2, H-20), 2.82 (s, 3 H, H-1), 2.37 (s, 3 H, H-10
′), 1.19 (t,
J = 7.1 Hz, 3 H, H-21). 13C NMR (126 MHz, CDCl3): δ /ppm = 153.73, 151.65, 148.15,
141.26, 140.79, 140.55, 140.00, 135.58, 134.99, 134.92, 134.12 (CH), 131.77 (CH), 130.14
(CH), 129.16 (CH), 128.12 (CH), 127.79 (CH), 127.46 (CH), 93.66 (CH), 64.00 (CH2),
18.10 (CH3), 14.08 (CH3), 10.03 (CH3).
214 Experimental
Ethyl (6-(3-((4-chlorophenyl)sulfonamido)-4-methylphenyl)-3-meth-
yl-[1,2,4]triazolo[4,3-b]pyridazin-8-yl)carbamate (108d)
5 N
N
N N
1
12
9
8
10'
N
S
14
15
OO
H2N
ClO O
18
19
To a solution of compound 103d (10 mg) and triethylamine (10 µL) in tetrahydrofuran
(10 mL) was added a solution of ethyl chloroformate in tetrahydrofuran (0.1 M, 0.5 mL).
The mixture was stirred at room temperature for 5 minutes. Aqueous citric acid (0.1 M,
5 mL) was added, and the mixture concentrated under reduced pressure. The residue
was extracted into ethyl acetate (2×10 mL), and the combined organic phases were dried
(MgSO4), filtered, and concentrated under reduced pressure to afford the title compound
(11 mg, 95 %) as a white solid.
1H NMR (400 MHz, CDCl3): δ /ppm = 8.06 (d, J = 8.7 Hz, 2 H, H-14), 7.89 (dd, J =
8.0 Hz and 1.9 Hz, 1 H), 7.65 (d, J = 1.9 Hz, 1 H), 7.57 (d, J = 8.7 Hz, 2 H, H-15), 7.45
(d, J = 8.0 Hz, 1 H, H-9), 6.49 (s, 1 H, H-5), 5.50 (br. s, 2 H, NH2), 4.27 – 4.06 (m, 2 H,
H-18), 2.82 (s, 3 H, H-1), 2.34 (s, 3 H, H-10′), 1.18 (t, J = 7.1 Hz, 3 H, H-19). 13C NMR
(126 MHz, CDCl3): δ /ppm = 153.78, 151.77, 148.10, 141.24, 140.81, 140.42, 139.99,
137.51, 135.61, 134.88, 131.71 (CH), 130.72 (CH), 129.19 (CH), 128.06 (CH), 127.88
(CH), 93.66 (CH), 63.87 (CH2), 18.11 (CH3), 14.12 (CH3), 10.01 (CH3). HRMS: m/z calc.
for C22H22[
35Cl]N6O4S ([M + H]
+) 501.1112, found 501.1133, ∆ = 4.2 ppm.
Experimental 215
Ethyl ((3,4-dichlorophenyl)sulfonyl)(5-(8-((ethoxycarbonyl)amino)-3-
methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)-2-methylphenyl)carbamate
(109e)
5 N
N
N N
1
12
9
8
10'
N
S
18
15
OO
H
N
Cl
Cl
O
OO
21
20
O24
23
To a solution of 103e (4 mg) and triethylamine (10 µL) in dichloromethane (5 mL) was
added ethyl chloroformate (0.1 mL). The mixture was stirred at room temperature for
5 minutes. Water (2 mL) was added, and the solvent removed under reduced pressure.
Purification by HPLC afforded the title compound (2 mg, 38 %).
1H NMR (500 MHz, DMSO-d6): δ /ppm = 8.20 (d, J = 2.2 Hz, 1 H, H-18), 8.09 (s, 1 H,
H-5), 8.05 (d, J = 8.5 Hz, 1 H, H-15), 8.01 (dd, J = 8.0 Hz and 1.9 Hz, 1 H, H-8), 7.94 (dd,
J = 8.5 Hz and 2.2 Hz, 1 H, H-14), 7.76 (d, J = 1.9 Hz, 1 H, H-12), 7.63 (dd, J = 8.0 Hz,
1 H, H-9), 4.27 (q, J = 7.1 Hz, 2 H, H-23), 4.23 – 4.02 (m, 2 H, H-20), 2.73 (s, 3 H, H-1),
2.32 (s, 3 H, H-10′), 1.30 (t, J = 7.1 Hz, 3 H, H-24), 1.09 (t, J = 7.1 Hz, 3 H, H-21). 13C
NMR (126 MHz, DMSO-d6): δ /ppm = 171.59, 156.02, 152.86, 148.75, 147.50, 140.57,
139.09, 138.28, 137.98, 135.36, 134.32, 132.26, 132.08 (CH), 131.95 (CH), 130.62 (CH),
128.58 (CH), 128.21 (CH), 127.58 (CH), 100.75 (CH), 64.56 (CH2), 64.09 (CH2), 17.57
(CH3), 14.39 (CH3), 13.85 (CH3), 9.56 (CH3), (One carbon resonance was not visible).
HRMS: m/z calc. for C25H25N6O6S[
35Cl]2 ([M + H]
+) 607.0933, found 607.0912, ∆ =
−3.5 ppm.
216 Experimental
Chemical Probes for CECR2
6-Chloro-3-methyl-[1,2,4]triazolo[4,3-b]pyridazine (129)233
4
5 N
N
NN
1
Cl
To 3-chloro-6-hydrazino pyridazine (250 mg, 1.75 mmol, Alfa Aesar) was added acetic
acid (2.5 mL) and the mixture heated at 120 ◦C for 1 hour in a sealed vial under microwave
irradiation. The solvent was removed, and the remaining acetic acid removed by azeotropic
drying with toluene. The title compound was obtained (270 mg, 1.60 mmol, 92 %) as a
white solid, m.p. 105 – 107 ◦C (ethyl acetate / hexane) [lit: 106 – 107 ◦C233].
1H NMR (400 MHz, CDCl3): δ /ppm = 8.02 (d, J = 9.6 Hz, 2 H, H-5), 7.07 (d, J = 9.6 Hz,
2 H, H-4), 2.77 (s, 3 H, H-1). 13C NMR (100 MHz, CDCl3): δ /ppm = 149.19 (C-6), 147.36
(C-2), 142.74 (C-3), 126.39 (CH, C-5), 121.76 (CH, C-4), 9.78 (CH3, C-1). IR (neat):
ν /cm−1 = 3029 (w), 1813 (w), 1605 (w), 1538 (w), 1513 (m), 1464 (m), 1441 (w), 1406
(w), 1383 (m), 1364 (m), 1330 (m), 1172 (w), 1152 (m), 1139 (w), 1081 (m), 1030 (m),
996 (m), 930 (m), 813 (s), 784 (m), 755 (s), 669 (w), 658 (m). LC-MS: tR = 1.04 min, m/z
169.15 ([M + H]+). HRMS: m/z calc. for [C6H6N4
35Cl]+ ([M + H]+) 169.0276, found
169.0270, ∆ = −3.1 ppm.
General Procedure A for the synthesis of sulfonamides
The sulfonyl chloride (256 mg, 1 mmol) was dissolved in n-butanol, CPME or DME (8 mL).
QuadraPure-dimethylamine (250 mg, 0.75 equiv.) was added and the mixture stirred at
25 ◦C for 15 minutes. The amine (1 mmol) was added, with rapid formation of a crystalline
precipitate. The mixture was stirred at 25 ◦C for 1 hour, and then water (1 mL) was added.
The amine resin was removed by filtration, and then the solvents were removed to afford
the product as a colourless crystalline solid.
Experimental 217
4-((4-Bromophenyl)sulfonyl)morpholine (125a)
3
2
Br
S N
O
6
5
O O
Prepared according to General Procedure A above, and collected as the hydrochloride salt
(357 mg, quant.). Free base m.p. 150.2 – 150.5 ◦C (dimethoxyethane / water).
1H NMR (400 MHz, CDCl3): δ /ppm = 9.97 (br. s, NH+), 7.69 (d, J = 8.7, 2 H, H-2 or
H-3), 7.62 (d, J = 8.7, 2 H, H-2 or H-3), 3.78 – 3.70 (m, 4 H, H-6), 3.04 – 2.96 (m, 4 H,
H-5). 13C NMR (100 MHz, CDCl3): δ /ppm = 134.40 (C-1 or C-4), 132.61 (CH, C-2 or
C-3), 129.43 (CH, C-3 or C-3), 128.36 (C-1 or C-4), 66.18 (CH2, C-6), 46.07 (CH2, C-5).
IR (neat): ν /cm−1 = 2971 (w), 2920 (w), 2904 (w), 2867 (w), 2761 (w), 2720 (w), 2161
(w), 1574 (w), 1466 (w), 1453 (m), 1431 (w), 1390 (w), 1347 (s), 1332 (m), 1299 (w), 1276
(w), 1261 (m), 1228 (w), 1216 (w), 1188 (w), 1179 (w), 1161 (s), 1123 (m), 1108 (s), 1093
(s), 1069 (s), 1049 (w), 1021 (w), 1009 (m), 970 (w), 942 (s), 925 (m), 911 (w), 872 (w),
849 (m), 826 (s), 753 (s), 715 (m). LC-MS: tR = 4.52 min, m/z 307.94 ([M + H]+). HRMS:
m/z calc. for [C10H13NO3S
79Br]+ ([M + H]+) 305.9800, found 305.9801, ∆ = 0.3 ppm.
4-((3-Bromophenyl)sulfonyl)morpholine (122a)
4
5
13
S N
O
8
7
Br
O O
Prepared according to General Procedure A above, and collected as the hydrochloride salt
(338 mg, quant.). Free base m.p. 151.3 – 151.5 ◦C (CPME).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.89 (t, J = 1.5 Hz, 1 H, H-1), 7.75 (dt, J = 7.9 Hz
and 1.5 Hz, 1 H, H-3), 7.68 (dt, J = 7.9 Hz and 1.5 Hz, 1 H, H-5), 7.43 (t, J = 7.9 Hz, 1 H,
H-4), 3.78 – 3.71 (m, 4 H, H-8), 3.06 – 2.98 (m, 4 H, H-7). 13C NMR (100 MHz, CDCl3):
δ /ppm = 137.18 (C-2 or C-6), 136.10 (CH, C-3), 130.64 (CH, C-1 or C-4), 130.56 (CH,
C-1 or C-4), 126.30 (CH, C-5), 123.27 (C-2 or C-6), 66.02 (CH2, C-8), 45.95 (CH2, C-7).
2868 (w), 2860 (w), 2161 (w), 1979 (w), 1574 (w), 1467 (w), 1453 (m), 1390 (w), 1363
218 Experimental
(w), 1347 (s), 1332 (m), 1299 (w), 1276(w), 1261 (m), 1217 (w), 1188 (w), 1178 (w), 1161
(s), 1123 (w), 1108 (s), 1093 (s), 1069 (s), 1021 (w), 1009 (m), 970 (w), 942 (s), 925 (m),
849 (m), 826 (s), 753 (s), 716 (m), 708 (m). LC-MS: tR = 4.48 min, m/z 307.88 ([M +
H]+). HRMS: m/z calc. for [C10H13NO3S
79Br]+ ([M + H]+) 305.9800, found 305.9814,
∆ = 4.6 ppm.
2-((4-Bromophenyl)sulfonyl)-1,2,3,4-tetrahydroisoquinoline (125b)
3
2
Br
S N
5
6
13 11
10
9
8
O O
Prepared according to General Procedure A above, and collected as the hydrochloride salt
after recrystallisation from ethanol (176 mg, 0.453 mmol, 45 %). Free base, m.p. 153.2 –
154.4 ◦C (ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.70 (d, J = 8.7 Hz, 1 H, H-2 or H-3), 7.66 (d, J =
8.7 Hz, 1 H, H-2 or H-3), 7.20 – 6.99 (m, 4 H, H-8 to H-11), 4.28 (s, 1 H, H-13), 3.39 (t, J =
6.0 Hz, 1 H, H-5), 2.93 (t, J = 6.0 Hz, 1 H, H-6). 13C NMR (100 MHz, CDCl3): δ /ppm =
135.82 (C-4), 133.04 (C-7), 132.53 (CH, C-2), 131.43 (C-12), 129.26 (CH, C-3), 128.99
(CH, C-8), 128.03 (C-1), 127.03 (CH, C-10), 126.60 (CH, C-11), 126.45 (CH, C-9), 47.60
(CH2, C-13), 43.83 (CH2, C-5), 28.89 (CH2, C-6). IR (neat): ν /cm−1 = 2160 (w), 1576
(w), 1498 (w), 1467 (w), 1428 (w), 1393 (w), 1356 (w), 1334 (m), 1272 (w), 1244 (w), 1162
(m), 1134 (w), 1121 (w), 1090 (m), 1068 (m), 1023 (m), 1009 (m), 954 (m), 921 (m), 838
(w), 822 (m), 771 (m), 743 (s), 706 (m). HRMS: m/z calc. for [C15H14O2N
79Br23Na32S]+
([M + Na]+) 373.9821, found 373.9803, ∆ = −4.8 ppm.
4-((4-(4,4,5,5-Tetramethyl-1,3,2-dioxaborolan-2-yl)phenyl)sulfonyl)
morpholine (126a)
3
2
B
S N
5
6
O
O
O
2'
O O
Experimental 219
4-Bromobenzenesulfonyl chloride (256 mg, 1 mmol) was dissolved in n-butanol (4 mL) and
heated to 50 ◦C. Morpholine (0.2 mL, 2 mmol) was added, and the mixture allowed to cool
to room temperature. A precipitate formed upon cooling.
Bis(pinacolato)diboron (279 mg, 1.1 mmol, 1.1 equiv.), KOAc (294 mg, 3 mmol, 3 equiv.),
Pd2(dba)3 (21 mg, 20 µmol, 0.02 equiv.) and X-Phos (19 mg, 40 µmol, 0.04 equiv.) were
added. The resulting mixture was degassed by sparging with N2 (1 bar) for 15 minutes,
and then heated at 150 ◦C for 1 hour in a sealed vial under microwave irradiation, and then
allowed to cool to room temperature.
The solvent was removed and the solid product recrystallised from ethanol to afford the title
compound (43 mg, 0.12 mmol, 12 %) as a colourless crystalline solid, m.p. 170.1 – 170.4 ◦C
(ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.97 (d, J = 7.87 Hz, 2 H, H-2), 7.73 (d, J =
7.87 Hz, 2 H, H-3), 3.76 – 3.69 (m, 4 H, H-6), 3.03 – 2.95 (m, 4 H, H-5), 1.36 (s, 12 H,
H-2′). 13C NMR (100 MHz, CDCl3): δ /ppm = 137.41 (C-4), 135.46 (CH, C-2), 126.91
(CH, C-3), 84.63 (C-2′), 66.22 (CH2, C-6), 46.12 (CH2, C-5), 25.02 (CH3, C-2′), (C-1
was not observed). IR (neat): ν /cm−1 = 2161 (w), 1449 (w), 1392 (m), 1352 (s), 1295
(w), 1260 (m), 1212 (w), 1168 (s), 1138 (m), 1115 (s), 1101 (m), 1079 (s), 1018 (m), 964
(w), 940 (s), 852 (m), 822 (w), 745 (s), 732 (s), 703 (m), 670 (m). LC-MS: tR = 4.89 min,
m/z 354.00 ([M + H]+). HRMS: m/z calc. for C16H25
11BNO5
32S]+ ([M + H]+) 354.1547,
found 354.1546, ∆ = −0.3 ppm. Microanalysis: calc. (found) for C16H24BNO5S C 54.54%
(54.39%), H 6.85% (6.81%), N 3.97% (4.11%).
2-((4-(4,4,5,5-Tetramethyl-1,3,2-dioxaborolan-2-yl)phenyl)sulfonyl)-
1,2,3,4-tetrahydroisoquinoline (126b)
3
2
S N
5
6
13 11
10
9
8
BO
O
2'
O O
Compound 125c (105 mg, 0.3 mmol) was combined with bis(pinacolato)diboron (114 mg,
0.45 mmol, 1.5 equiv.), KOAc (88 mg, 0.9 mmol, 3 equiv.), Pd2(dba)3 (6 mg, 6 µmol, 0.02
equiv.) and X-Phos (6 mg, 12 µmol, 0.04 equiv.). CPME (12 mL) was added and the
220 Experimental
resulting mixture degassed by sparging with N2 (1 bar) for 15 minutes, and then heated at
110 ◦C for 1 hour in a sealed vial under microwave irradiation. The title compound was
obtained as a colourless crystalline solid (98 mg, 244 mmol, 49 %) after recrystallisation
from hot ethanol, m.p. 168.5 – 169.7 ◦C (ethanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 7.95 (d, J = 8.0 Hz, 2 H, H-2 or H-3), 7.82 (d,
J = 8.0 Hz, 2 H, H-2 or H-3), 7.19 – 6.97 (m, 4 H, H-8 to H-11), 4.25 (s, 2 H, H-13),
3.36 (t, J = 5.9 Hz, 2 H, H-5), 2.93 (q, J = 5.9 Hz, 2 H, H-6), 1.35 (s, 12 H, H-2′). 13C
NMR (100 MHz, CDCl3): δ /ppm = 138.66 (C-4), 135.46 (CH, C-2), 133.17 (C-7), 131.67
(C-12), 128.95 (CH, C-8), 126.89 (CH, C-10), 0 126.77 (CH, C-3), 126.51 (CH, C-11),
126.47 (CH, C-9), 84.59 (C-1′), 47.63 (CH2, C-13), 43.87 (CH2, C-5), 29.03 (CH2, C-6),
25.02 (CH3, C-2
′), (C-1 was not observed). IR (neat): ν /cm−1 = 1391 (w), 1353 (s), 1335
(s), 1270 (w), 1165 (s), 1144 (m), 1121 (w), 1096 (m), 1077 (m), 1069 (m), 1024 (w),
1017 (w), 953 (m), 922 (m), 856 (w), 839 (w), 823 (w), 814 (w), 766 (m), 733 (s), 714
(w), 671 (w), 656 (s). LC-MS: tR = 5.42 min, m/z 400.01 ([M + H]+). HRMS: m/z calc.
for [C21H26O4N
11B23Na32S]+ ([M + Na]+) 422.1568, found 422.1555, ∆ = −2.97 ppm.
Microanalysis: calc. (found) for C21H25O4NBS C 63.17% (63.24%), H 6.56% (6.38%), N
3.51% (3.80%).
General Procedure B for one-pot synthesis of target molecules
The sulfonyl chloride (256 mg, 1 mmol) was dissolved in n-butanol (8 mL). QuadraPure-
dimethylamine (250 mg, 0.75 equiv.) was added and the mixture stirred at 25 ◦C for 15
minutes. The amine (1 mmol) was added, with rapid formation of a crystalline precipitate.
The mixture was stirred at 25 ◦C for 1 hour, and then water (1 mL) was added. The amine
resin was removed by filtration, and then the solvents were removed to afford a solid
product.
The solid (generally approx. 0.75 mmol) was combined with bis(pinacolato)diboron (286 mg,
1.13 mmol, 1.5 equiv.), KOAc (221 mg, 2.25 mmol, 3 equiv.), Pd2(dba)3 (16 mg, 15 µmol,
0.02 equiv.) and X-Phos (14 mg, 30 µmol, 0.04 equiv.). CPME (12 mL) was added and the
resulting mixture degassed by sparging with N2 (1 bar) for 15 minutes, and then heated at
110 ◦C for 1 hour in a sealed vial under microwave irradiation, and then allowed to cool to
room temperature.
The solvents were removed under reduced pressure and then the reaction mixture was redis-
Experimental 221
solved in n-butanol / water (9 : 1 v/v, 10 mL). Triazolopyridazine 129 (76 mg, 0.46 mmol,
0.9 equiv.) and K3PO4 (386 mg, 1.8 mmol, 3.6 equiv.) were added and the resulting mixture
degassed by sparging with N2 (1 bar) for 15 minutes, and then heated at 100
◦C for 2 hours
in a sealed vial under microwave irradiation.
The reaction mixture was allowed to cool to room temperature, concentrated, loaded onto
KP-NH silica, and then purified by column chromatography using KP-NH silica, using a
gradient of ethyl acetate / methanol (9 : 1 v/v) and n-hexane: 0 : 1 for 4 CV, then 0 : 1 to 2 : 3
over 40 CV.
4-((4-(3-Methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)phenyl)sulfonyl)
morpholine (128a)
8
9
5
4
N N
N
N
1
S N
O
12
11
O O
Obtained using General Procedure B above. Purification by column chromatography
afforded the title compound (7 mg, 0.019 mmol, 2 % over 3 steps) as a white solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 8.22 (d, J = 9.7 Hz, 1 H, H-4), 8.18 (d, J = 8.3 Hz,
2 H, H-8), 7.94 (d, J = 8.3 Hz, 2 H, H-9), 7.56 (d, J = 9.7 Hz, 1 H, H-5), 3.80 – 3.74 (m,
4 H, H-12), 3.12 – 3.03 (m, 4 H, H-11), 2.90 (s, 3 H, H-1). 13C NMR (126 MHz, CDCl3):
δ /ppm = 151.84 (C-6), 147.84 (C-2), 143.34 (C-3), 138.98 (C-10), 137.55 (C-7), 128.77
(CH, C-8), 128.17 (CH, C-9), 125.82 (CH, C-4), 118.41 (CH, C-5), 66.23 (CH2, C-12),
46.12 (CH2, C-11), 10.01 (CH3, C-1). IR (neat): ν /cm−1 = 2860 (w), 2160 (w), 2024 (w),
1551 (w), 1524 (w), 1454 (w), 1392 (w), 1346 (m), 1329 (m), 1309 (m), 1260 (m), 1166
(s), 1128 (m), 1113 (s), 1071 (m), 1042 (w), 1014 (w), 990 (w), 946 (s), 848 (m), 827 (m),
763 (s), 738 (s), 710 (m). LC-MS: tR = 4.18 min, m/z 360.02 ([M + H]+). HRMS: m/z calc.
for [C16H18N5O3
32S]+ ([M + H]+) 360.1130, found 360.1135, ∆ = 1.4 ppm.
222 Experimental
2-((4-(3-Methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)phenyl)sulfonyl)-
1,2,3,4-tetrahydroisoquinoline (128b)
8
9
5
4
N
N
N
N
1
S N
11
12
19 17
16
15
14
O O
Obtained using General Procedure B above. Purification by column chromatography and
subsequent recrystallisation from chloroform / ethanol afforded the title compound (7 mg,
0.017 mmol, 2 % over 3 steps) as a white solid, m.p. 232.3 – 233.4 ◦C (decomp.).
1H NMR (400 MHz, CDCl3): δ /ppm = 8.20 (d, J = 9.7 Hz, 1 H, H-4), 8.15 (d, J = 8.3 Hz,
2 H, H-8), 8.02 (d, J = 8.3 Hz, 2 H, H-9), 7.55 (d, J = 9.7 Hz, 1 H, H-5), 7.21 – 7.02 (m, 4 H,
H-14 to H-17), 4.35 (s, 2 H, H-19), 3.46 (t, J = 5.9 Hz, 2 H, H-11), 2.95 (t, J = 5.9 Hz, 2 H,
H-12), 2.89 (s, 3 H, H-1). 13C NMR (100 MHz, CDCl3): δ /ppm = 151.91 (C-6), 147.81
(C-2), 139.05 (C-10), 138.68 (C-7), 133.05 (C-13), 131.39 (C-18), 129.03 (CH, C-14),
128.56 (CH, C-8), 128.11 (CH, C-9), 127.09 (CH, C-16), 126.65 (CH, C-17), 126.44 (CH,
C-15), 125.73 (CH, C-4), 118.45 (CH, C-5), 115.63 (C-3), 47.62 (CH2, C-19), 43.89 (CH2,
C-11), 28.95 (CH2, C-12), 10.00 (CH3, C-1). IR (neat): ν /cm−1 = 2160 (w), 2035 (w),
1968 (w), 1524 (w), 1388 (w), 1339 (m), 1308 (w), 1274 (w), 1164 (m), 1131 (w), 1117
(w), 1106 (w), 1092 (w), 1073 (w), 1048 (w), 1025 (w), 998 (w), 956 (w), 919 (m), 857
(w), 841 (w), 820 (m), 769 (m), 746 (s), 724 (w), 707 (w), 672 (w). LC-MS: tR = 4.69 min,
m/z 405.98 ([M + H]+). HRMS: m/z calc. for [C21H20N5O2
32S]+ ([M + H]+) 406.1338,
found 406.1349, ∆ = 2.7 ppm.
Experimental 223
6-(4-((2-Ethylpiperidin-1-yl)sulfonyl)phenyl)-3-methyl-[1,2,4]triazolo
[4,3-b]pyridazine (128c)
8
9
5
4
N N
N
N
1
S N
14
13
12
11
17
O O
Obtained using General Procedure B above. Purification by column chromatography
afforded the title compound (16 mg, 0.042 mmol, 4 % over 3 steps) as a white solid, m.p.
179.4 – 179.8 ◦C (ethyl acetate / hexane / methanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 8.19 (d, J = 9.7 Hz, 1 H, H-4), 8.13 (d, J = 8.5 Hz,
2 H, H-8), 8.02 (d, J = 8.5 Hz, 2 H, H-9), 7.57 (d, J = 9.7 Hz, 1 H, H-5), 4.00 (q, J = 6.8 Hz,
1 H, H-15), 3.83 (dd, J = 14.0 Hz and 4.6 Hz, 1 H, H-11), 3.04 (td, J = 14.0 Hz and 2.7 Hz,
1 H, H-11), 2.91 (s, 3 H, H-1), 1.73 – 1.47 (m, 7 H, H-12, H-13, H-14 and H-16), 1.38 – 1.13
(m, 1 H, H-12), 0.90 (t, J = 7.5 Hz, 3 H, H-17). 13C NMR (100 MHz, CDCl3): δ /ppm =
151.95 (C-6), 147.76 (C-2), 144.39 (C-3), 143.37 (C-10), 137.84 (C-7), 127.97 (CH, C-8 or
C-9), 127.81 (CH, C-8 or C-9), 125.57 (CH, C-4), 118.47 (CH, C-5), 54.98 (CH, C-15),
40.93 (CH2, C-11), 27.31 (CH2, C-14), 24.76 (CH2, C-12, C-13 or C-16), 22.65 (CH2,
C-12, C-13 or C-16), 18.47 (CH2, C-12, C-13 or C-16), 11.15 (CH3, C-17), 10.00 (CH3,
C-1). IR (neat): ν /cm−1 = 3123 (w), 3068 (w), 2941 (w), 2871 (w), 2527 (w), 2418 (w),
2179 (w), 2164 (w), 2144 (w), 2033 (w), 2019 (w), 1962 (w), 1609 (w), 1552 (w), 1527
(w), 1493 (w), 1468 (w), 1397 (m), 1374 (m), 1360 (w), 1330 (s), 1311 (m), 1291 (m),
1270 (w), 1218 (w), 1194 (w), 1162 (s), 1152 (s), 1092 (m), 1071 (w), 1058 (m), 1038 (m),
1012 (m), 993 (m), 950 (s), 917 (m), 904 (m), 865 (m), 851 (w), 837 (m), 818 (s), 775
(m), 760 (m), 734 (s), 686 (m). LC-MS: tR = 4.69 min, m/z 386.02 ([M + H]+). HRMS:
m/z calc. for [C19H24N5O2
32S]+ ([M + H]+) 386.1651, found 386.1649, ∆ = −0.5 ppm.
Microanalysis: calc. (found) for C19H23N5O2S C 59.20% (59.16%), H 6.01% (5.88%), N
18.17% (17.75%).
224 Experimental
6-(4-((4-Ethylpiperazin-1-yl)sulfonyl)phenyl)-3-methyl-[1,2,4]triazolo
[4,3-b]pyridazine (128d)
8
9
5
4
N
N
N
N
1
S N
11
12
N 13
14
O O
Obtained using General Procedure B above. Purification by column chromatography and
subsequent recrystallisation from ethanol afforded the title compound (11 mg, 0.028 mmol,
3 % over 3 steps) as a white solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 8.21 (d, J = 9.7 Hz, 1 H, H-4), 8.14 (d, J = 8.3 Hz,
2 H, H-8), 7.93 (d, J = 8.3 Hz, 2 H, H-9), 7.54 (d, J = 9.7 Hz, 1 H, H-5), 3.17 – 3.03 (m, 4 H,
H-11), 2.90 (s, 3 H, H-1), 2.59 – 2.50 (m, 4 H, H-12), 2.41 (q, J = 7.2 Hz, 2 H, H-13), 1.03
(t, J = 7.2 Hz, 3 H, H-14). 13C NMR (126 MHz, CDCl3): δ /ppm = 151.93 (C-6), 147.84
(C-2), 143.35 (C-3), 138.78 (C-10), 137.57 (C-7), 128.79 (CH, C-8 or C-9), 128.04 (CH,
C-8 or C-9), 125.78 (CH, C-4), 118.44 (CH, C-5), 52.03 (CH2, C-13), 51.91 (CH2, C-12),
46.25 (CH2, C-11), 12.06 (CH3, C-14), 10.01 (CH3, C-1). IR (neat): ν /cm−1 = 1547 (w),
1521 (w), 1454 (w), 1388 (w), 1330 (m), 1307 (m), 1262 (w), 1168 (s), 1152 (s), 1101
(m), 1065 (w), 951 (s), 926 (m), 820 (m), 761 (m), 740 (s). LC-MS: tR = 3.52 min, m/z
387.03 ([M + H]+). HRMS: m/z calc. for [C18H23N6O2
32S]+ ([M + H]+) 387.1603, found
387.1594, ∆ = −2.3 ppm.
Experimental 225
6-(3-((2-Ethylpiperidin-1-yl)sulfonyl)phenyl)-3-methyl-[1,2,4]triazolo
[4,3-b]pyridazine (132a)
8
9
10
12
5
4 N
N
NN
1
S N 16
15
14
19
O O
Obtained using General Procedure B above. Purification by column chromatography using
KP-NH silica afforded the title compound (43 mg, 0.122 mmol, 12 % over 3 steps) as a
white solid, m.p. 160.9 – 161.3 ◦C (ethyl acetate / hexane / methanol).
1H NMR (400 MHz, CDCl3): δ /ppm = 8.44 (s, 1 H, H-12), 8.29 – 8.10 (m, 2 H, H-4 and
H-8), 8.00 (d, J = 7.7 Hz, 1 H, H-10), 7.68 (t, J = 7.7 Hz, 1 H, H-9), 7.57 (d, J = 9.7 Hz, 1 H,
H-5), 3.99 (q, J = 7.1 Hz, 1 H, H-17), 3.79 (dd, J = 14.0 Hz and 4.2 Hz, 1 H, H-13), 3.03
(td, J = 14.0 Hz and 2.5 Hz, 1 H, H-13), 2.87 (s, 3 H, H-1), 1.78 – 1.14 (m, 8 H, H-14, H-15,
H-16 and H-18), 0.88 (t, J = 7.4 Hz, 3 H, H-19). 13C NMR (100 MHz, CDCl3): δ /ppm =
223.24 (C-6), 152.04 (C-2), 143.52 (C-3), 135.59 (C-11), 130.68 (CH, CAr), 130.11 (CH,
CAr), 129.03 (CH, CAr), 125.71 (CH, CAr), 125.62 (CH, C-4), 118.43 (CH, C-5), 115.59
(C-7), 54.95 (CH, C-17), 40.89 (CH2, C-13), 27.33 (CH2, C-16), 24.73 (CH2, C-14, C-15
or C-18), 22.63 (CH2, C-14, C-15 or C-18), 18.46 (CH2, C-14, C-15 or C-18), 11.15 (CH3,
C-19), 9.97 (CH3, C-1). IR (neat): ν /cm−1 = 2933 (w), 2166 (w), 1980 (w), 1546 (w),
1522 (w), 1472 (w), 1389 (m), 1332 (s), 1315 (s), 1195 (w), 1154 (s), 1104 (m), 1036
(m), 994 (m), 947 (s), 913 (m), 838 (m), 804 (s), 761 (m), 737 (s), 689 (s). LC-MS: tR =
4.68 min, m/z 386.02 ([M + H]+). HRMS: m/z calc. for [C19H24N5O2
32S]+ ([M + H]+)
386.1651, found 386.1641, ∆ = −2.6 ppm. Microanalysis: calc. (found) for C19H23N5O2S
C 59.20% (59.39%), H 6.01% (5.97%), N 18.17% (18.01%).
226 Experimental
2-((3-(3-Methyl-[1,2,4]triazolo[4,3-b]pyridazin-6-yl)phenyl)sulfonyl)-
1,2,3,4-tetrahydroisoquinoline (132b)
9
10
128
5
4 N
N
NN
1
S N
21
14 16
17
18
19O O
Obtained using General Procedure B above. Purification by column chromatography and
subsequent recrystallisation from ethyl acetate / hexane afforded the title compound (15 mg,
0.037 mmol, 4 % over 3 steps) as a white solid, m.p. 222.9 – 224.5 ◦C (ethyl acetate /
hexane).
1H NMR (500 MHz, CDCl3): δ /ppm = 8.40 (dd, J = 1.9 Hz and 1.6 Hz, 1 H, H-12), 8.22
(ddd, J = 7.8, 1.9 Hz and 1.1 Hz, 1 H, H-8), 8.19 (d, J = 9.7 Hz, 1 H, H-4), 8.01 (ddd, J =
7.8, 1.6 Hz and 1.1 Hz, 1 H, H-10), 7.73 (t, J = 7.8 Hz, 1 H, H-9), 7.51 (d, J = 9.7 Hz, 1 H,
H-5), 7.16 – 7.12 (m, 2 H, HAr), 7.10 – 7.01 (m, 2 H, HAr), 4.38 (s, 2 H, H-21), 3.49 (t, J =
5.9 Hz, 2 H, H-13), 2.93 (t, J = 6.0 Hz, 2 H, H-14), 2.89 (s, 3 H, H-1). 13C NMR (126 MHz,
CDCl3): δ /ppm = 151.76 (C-6), 147.65 (C-2), 143.22 (C-3), 138.42 (C-7 or C-11), 135.71
(C-7 or C-11), 132.90 (C-15), 131.31 (CH, CAr), 131.24 (C-20), 130.13 (CH, CAr), 129.52
(CH, CAr), 128.91 (CH, C-14), 126.98 (CH, C-18), 126.51 (CH, C-19), 126.28 (CH, C-17),
126.20 (CH, CAr), 125.60 (CH, C-4), 118.27 (CH, C-5), 47.51 (CH2, C-21), 43.70 (CH2,
C-13), 28.61 (CH2, C-14), 9.88 (CH3, C-1). IR (neat): ν /cm−1 = 3062 (w), 2161 (w),
1553 (w), 1524 (w), 1465 (w), 1397 (w), 1373 (w), 1356 (m), 1335 (s), 1271 (w), 1164 (s),
1122 (m), 1091 (m), 1073 (m), 1020 (m), 995 (w), 957 (m), 926 (m), 835 (m), 796 (m),
759 (s), 742 (s), 723 (s), 686 (s), 675 (m), 658 (m). LC-MS: tR = 4.66 min, m/z 406.00 ([M
+ H]+). HRMS: m/z calc. for [C21H20N5O2
32S]+ ([M + H]+) 406.1338, found 406.1339,
∆ = 0.2 ppm.
Experimental 227
3-Methyl-6-(3-(piperidin-1-ylsulfonyl)phenyl)-[1,2,4]triazolo[4,3-b]
pyridazine (132c)
4
5 N
N
NN
1
8
9
10
12
S N
O O
15
14
Obtained using General Procedure B above. Purification by column chromatography
afforded the title compound (3.5 mg, 0.01 mmol, 1 % over 3 steps) as a white solid.
1H NMR (500 MHz, CDCl3): δ /ppm = 8.34 (dd, J = 1.8 Hz and 0.6 Hz, 1 H, H-12), 8.24
(ddd, J = 7.8 Hz, 1.8 Hz and 1.1 Hz, 1 H, H-4), 8.20 (d, J = 9.7 Hz, 1 H, H8 or H-10), 7.93
(ddd, J = 7.8 Hz, 1.8 Hz and 1.1 Hz, 1 H, H-5), 7.74 (td, J = 7.8 Hz and 0.6 Hz, 1 H, H-9),
7.57 (d, J = 9.7 Hz, 1 H, H8 or H-10), 3.13 – 3.03 (m, 4 H, H-13), 2.90 (s, 3 H, H-1),
1.73 – 1.63 (m, 4 H, H-14), 1.52 – 1.38 (m, 2 H, H-15). 13C NMR (126 MHz, CDCl3):
δ /ppm = 151.88 (C-6), 147.64 (C-2), 143.23 (C-3), 138.04 (C-7 or C-11), 135.57 (C-7 or
C-11), 131.07 (CH, CAr), 130.00 (CH, CAr), 129.59 (CH, CAr), 126.23 (CH, CAr), 125.60
(CH, C-4), 118.29 (CH, C-5), 46.96 (CH2, C-13), 25.15 (CH2, C-14), 23.45 (CH2, C-15),
9.88 (CH3, C-1). LC-MS: tR = 4.49 min, m/z 358.04 ([M + H]
+). HRMS: m/z calc. for
[C17H20N5O2
32S]+ ([M + H]+) 358.1338, found 358.1352, ∆ = −3.9 ppm.
228 Experimental
Frontal Affinity Chromatography
Pierce Streptavidin Ultralink resin (P/N PL-53113) was obtained from Fisher Scientific.
Biotinylated BRD9 and CECR2 were kindly provided by the Structural Genomics Con-
sortium, University of Oxford. Upchurch 1 mm×20 mm 15 µL microbore HPLC guard
column (P/N C-128) was obtained from Kinesis Ltd.
Preparation of Immobilised Enzymes
The proteins were supplied in individual vials each containing 10 µL of 0.5 mg/mL protein
suspension in aqueous buffer. The contents of 16 vials were combined, and an additional
16×10 µL was used to rinse the vials. A suspension of streptavidin-functionalised acry-
lamide resin (60 µL) was added, and the mixture shaken at 4 ◦C for 1 hour. The mixture
was then loaded into a 15 µL column using a syringe, with occasional packing using the
HPLC pump.
FAC-UV / FAC-MS experiments
FAC assays were run on a modified Agilent 1100 HPLC system with a 100 µL injection
syringe, a 600 µL storage loop and 100 mM ammonium acetate (when the MS was in use),
or 100 mM phosphate-buffered saline (PBS) solution (for UV-only experiments) was used
as the eluent. Two wavelengths were monitored simultaneously: 220 nm (DMSO) and
254 nm (analyte) using a UV Diode Array Detector.
The output from the HPLC was combined with a stream of 0.1 % formic acid in methanol
(200 µL/min) and directed into an Expression CMS mass spectrometer (Advion). The CMS
was used in positive ESI mode, with selective ion monitoring (SIM, span 1 AMU, dwell
time 500 ms).
Stock solutions of compounds were prepared at a concentration of 50 mM in DMSO and
stored at −20 ◦C. The HPLC system was set to inject 600 µL of 4 µM analyte at 30 µL/min
for 120 minutes.
References and Notes
1. Szostak, J. W. Chem. Rev. 1997, 97, 347–348, doi: 10.1021/cr9700080.
2. Nefzi, A.; Ostresh, J. M.; Houghten, R. A. Chem. Rev. 1997, 97, 449–472, doi:
10.1021/cr960010b.
3. Newman, D. J.; Cragg, G. M. J. Nat. Prod. 2007, 70, 461–77, doi: 10 . 1021 /
np068054v.
4. Feher, M.; Schmidt, J. J. Chem. Inf. Comput. Sci. 2003, 43, 218–227, doi: 10.1021/
ci0200467.
5. Spring, D. R. Org. Biomol. Chem 2003, 1, 3867–3870, doi: 10.1039/b310752n.
6. Schmink, J. R.; Bellomo, A.; Berritt, S. Aldrichim. Acta 2013, 46, 71–80.
7. Accelerating chemical process development at GSK. Radleys Press Release Feb.
2005, http://www.radleys.co.uk/pages/press/accelerating_chemical.html.
8. Sanderson, K. Bloggers put chemical reactions through the replication mill. Nature
News Jan. 2013, http://www.nature.com/news/bloggers-put-chemical-reactions-
through-the-replication-mill-1.12262.
9. Bosch, C. Nobel Lecture: The development of the chemical high pressure method
during the establishment of the new ammonia industry. 1932.
10. Phillips, P. Manufacture of sulphuric acid. British Patent, 6096, 1831.
11. Stankiewicz, A. I.; Moulijn, J. A. Process intensification: transforming chemical
engineering. Chem. Eng. Progress 2000, 96, 22–34.
12. Ramshaw, C. The Incentive for Process Intensification In 1st International Confer-
ence on Process Intensification for the Chemical Industry, BHR Group: London,
UK, 1995.
13. Nagaki, A.; Imai, K.; Kim, H.; Yoshida, J.-i. RSC Adv. 2011, 1, 758–760, doi:
10.1039/c1ra00377a.
14. Bogdan, A. R.; Poe, S. L.; Kubis, D. C.; Broadwater, S. J.; McQuade, D. T. Angew.
Chem. Int. Ed. 2009, 48, 8547–8550, doi: 10.1002/anie.200903055.
15. Ley, S. V.; Ladlow, M.; Vickerstaffe, E., The Use of Polymer-Assisted Solution-Phase
Synthesis and Automation for the High-Throughput Preparation of Biologically
Active Compounds. In Exploiting Chemical Diversity for Drug Discovery, Bartlett, P.
229
230 References and Notes
A., Entzeroth, M., Eds.; The Royal Society of Chemistry: 2006; Chapter 1, pp 3–32,
doi: 10.1039/9781847552556-00001.
16. Brocklehurst, C. E.; Lehmann, H.; La Vecchia, L. Org. Process Res. Dev. 2011, 15,
1447–1453, doi: 10.1021/op200055r.
17. Audiger, L.; Watts, K.; Elmore, S. C.; Robinson, R. I.; Wirth, T. ChemSusChem
2012, 5, 257–260, doi: 10.1002/cssc.201100372.
18. Brodmann, T.; Koos, P.; Metzger, A.; Knochel, P.; Ley, S. V. Org. Process Res. Dev.
2012, 16, 1102–1113, doi: 10.1021/op200275d.
19. Baumann, M.; Baxendale, I. R.; Ley, S. V.; Nikbin, N.; Smith, C. D.; Tierney, J. P.
Org. Biomol. Chem. 2008, 6, 1577–1586, doi: 10.1039/b801631n.
20. Künzle, N.; Solèr, J.-W.; Baiker, A. Catal. Today 2003, 79U˝-80, 503–509, doi:
10.1016/S0920-5861(03)00075-0.
21. Johnson, M. D.; May, S. A.; Calvin, J. R.; Remacle, J.; Stout, J. R.; Diseroad, W. D.;
Zaborenko, N.; Haeberle, B. D.; Sun, W.-M.; Miller, M. T.; Brennan, J. Org. Process
Res. Dev. 2012, 16, 1017–1038, doi: 10.1021/op200362h.
22. Elvira, K. S.; Casadevall i Solvas, X.; Wootton, R. C. R.; DeMello, A. J. Nature
Chem. 2013, 5, 905–915, doi: 10.1038/nchem.1753.
23. Wegner, J.; Ceylan, S.; Kirschning, A. Adv. Synth. Catal. 2012, 354, 17–57, doi:
10.1002/adsc.201100584.
24. Baumann, M.; Baxendale, I. R.; Ley, S. V. Mol. Divers. 2011, 15, 613–630, doi:
10.1007/s11030-010-9282-1.
25. Ley, S. V.; Baxendale, I. R.; Bream, R. N.; Jackson, P. S.; Leach, A. G.; Longbottom,
D. A.; Nesi, M.; Scott, J. S.; Storer, R. I.; Taylor, S. J. J. Chem. Soc. Perkin Trans. 1
2000, 3815–4195, doi: 10.1039/b006588i.
26. Wasik, B. In the Programmable World, All Our Objects Will Act as One. Wired USA
May 2013, http://www.wired.com/2013/05/internet-of-things-2/.
27. Coughlan, S. Curriculum changes ‘to catch up with world’s best’. BBC News July
2013, http://www.bbc.co.uk/news/education-23222068.
28. Raspberry Pi Foundation. Website: http://www.raspberrypi.org/.
29. Ingham, R. J.; Riva, E.; Nikbin, N.; Baxendale, I. R.; Ley, S. V. Org. Lett. 2012, 14,
3920–3923, doi: 10.1021/ol301673q.
30. Saczewski, J.; Paluchowska, A.; Klenc, J.; Raux, E.; Barnes, S.; Sullivan, S.; Duszyn-
ska, B.; Bojarski, A. J.; Strekowski, L. J. Heterocycl. Chem. 2009, 46, 1259–1265,
doi: 10.1002/jhet.236.
31. El-Deeb, I. M.; Lee, S. H. Bioorg. Med. Chem. 2010, 18, 3860–3874, doi: 10.1016/j.
bmc.2010.04.037.
32. Stanetty, P.; Hattinger, G.; Schnürch, M.; Mihovilovic, M. D. J. Org. Chem. 2005,
70, 5215–5220, doi: 10.1021/jo0505223.
References and Notes 231
33. Zimmermann, J. Pyrimidine derivatives and processes for the preparation thereof.
US Patent, 5521184, 1996.
34. Agarwal, A.; Srivastava, K.; Puri, S. K.; Chauhan, P. M. S. Bioorg. Med. Chem.
2005, 13, 4645–4650, doi: 10.1016/j.bmc.2005.04.061.
35. Wang, X.; Chakrabarti, P. P.; Ognyanov, V. I.; Pettus, L. H.; Tamir, R.; Tan, H.;
Tang, P.; Treanor, J. J. S.; Gavva, N. R.; Norman, M. H. Bioorg. Med. Chem. Lett.
2007, 17, 6539–6545, doi: 10.1016/j.bmcl.2007.09.080.
36. Hopkin, M. D.; Baxendale, I. R.; Ley, S. V. Org. Biomol. Chem. 2013, 11, 1822–
1839, doi: 10.1039/c2ob27002a.
37. Almario Garcia, A.; Ando, R.; Aritomo, K.; Frost, J. R.; Li, A. T.; Shoda, A.;
Uehara, F.; Watanabe, K. Preparation of 2-(arylalkylamino)pyrimidones and 2-
(heteroarylalkylamino)pyrimidones for preventive and/or therapeutic treatment of a
neurodegenerative disease caused by abnormal activity of GSK3β . World Patent,
2001070727, 2001.
38. Masquelin, T.; Sprenger, D.; Baer, R.; Gerber, F.; Mercadal, Y. Helv. Chim. Acta
1998, 81, 646–660, doi: 10.1002/hlca.19980810315.
39. Gayo, L. M.; Suto, M. J. Tetrahedron Lett. 1997, 38, 211–214, doi: 10.1016/S0040-
4039(96)02256-3.
40. Baekeland, L. H. J. Ind. Eng. Chem. 1909, 1, 149–161, doi: 10.1021/ie50003a004.
41. Adams, B. A.; Holmes, E. L. Improvements in and relating to the removal of
components or constituents from liquids by adsorption or absorption. British Patent,
450308, 1936.
42. Adams, B. A.; Holmes, E. L. Improvements in and relating to the treatment of liquids
and gases by absorption and in particular to treatment of liquids for the removal or
exchange of anions. British Patent, 450309, 1936.
43. Herr, D. S. Ind. Eng. Chem. 1945, 37, 631–634, doi: 10.1021/ie50427a011.
44. Merrifield, R. B. J. Am. Chem. Soc. 1963, 85, 2149–2154, doi: 10.1021/ja00897a025.
45. See references within Akelah, A.; Sherrington, D. C. Chem. Rev. 1981, 81, 557–587,
doi: 10.1021/cr00046a003.
46. Baxendale, I. R.; Ley, S. V. Bioorg. Med. Chem. Lett. 2000, 10, 1983–1986, doi:
10/d7jxzq.
47. Ley, S. V.; Massi, A. J. Comb. Chem. 2000, 2, 104–107, doi: 10.1021/cc9900697.
48. Habermann, J; Ley, S. V.; Scott, J. S. J. Chem. Soc. Perkin Trans. 1 1999, 1253–1255,
doi: 10.1039/a901802f.
49. Hinzen, B.; Ley, S. V. J. Chem. Soc. Perkin Trans. 1 1997, 1907–1908, doi: 10.1039/
A703461J.
50. Vickerstaffe, E.; Warrington, B.; Ladlow, M.; Ley, S. J. Comb. Chem. 2004, 6, 332–
339, doi: 10.1021/cc049977g.
232 References and Notes
51. Baxendale, I. R.; Ley, S. V.; Mansfield, A. C.; Smith, C. D. Angew. Chem. Int. Ed.
2009, 48, 4017–4021, doi: 10.1002/anie.200900970.
52. Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem. Int. Ed. 2001, 40, 2004–
2021, doi: 10/br9nj7.
53. Sherrington, D. J. Assoc. Lab. Autom. 2000, 5, 75–78, doi: 10 . 1016 / S1535 -
5535(04)00066-8.
54. Svec, F.; Fréchet, J. M. J. Ind. Eng. Chem. Res. 1999, 38, 34–48, doi: 10.1021/
ie970598s.
55. Svec, F.; Fréchet, J. M. J. Anal. Chem. 1992, 64, 820–822, doi: 10.1021/ac00031a022.
56. Petro, M.; Svec, F.; Fréchet, J. M. J. Biotechnol. Bioeng. 1996, 49, 355–363, doi:
10.1002/(SICI)1097-0290(19960220)49:4<355::AID-BIT1>3.0.CO;2-O.
57. Altava, B.; Burguete, M. I.; Fraile, J. M.; García, J. I.; Luis, S. V.; Mayoral, J. A.;
Vicent, M. J. Angew. Chem. Int. Ed. 2000, 39, 1503–1506, doi: 10/bwj6f5.
58. Krause, J. O.; Lubbad, S. H.; Nuyken, O.; Buchmeiser, M. R. Macromol. Rapid
Commun. 2003, 24, 875–878, doi: 10.1002/marc.200300024.
59. Solodenko, W.; Wen, H.; Leue, S.; Stuhlmann, F.; Sourkouni-Argirusi, G.; Jas, G.;
Schönfeld, H.; Kunz, U.; Kirschning, A. Eur. J. Org. Chem. 2004, 2004, 3601–3610,
doi: 10.1002/ejoc.200400194.
60. Burguete, M. I.; Cornejo, A.; García-Verdugo, E.; Gil, M. J.; Luis, S. V.; Mayoral,
J. A.; Martínez-Merino, V.; Sokolova, M. J. Org. Chem. 2007, 72, 4344–4350, doi:
10.1021/jo070119r.
61. Baumann, M.; Baxendale, I. R.; Ley, S. V.; Nikbin, N.; Smith, C. D. Org. Biomol.
Chem. 2008, 6, 1587–1593, doi: 10.1039/b801634h.
62. Roper, K. A. Synthesis of Heterocyclic Molecules using Flow and Microwave
Technologies. Master’s Thesis, University of Cambridge, 2008.
63. Peters, E. C.; Petro, M.; Svec, F.; Fréchet, J. M. J. Anal. Chem. 1997, 69, 3646–3649,
doi: 10.1021/ac970377w.
64. Sherrington, D. C., Preparation, Functionalisation and Characteristics of Polymer
Supports. In Polymer Supported Reactions in Organic Synthesis, Hodge, P., Sher-
rington, D. C., Eds., 1st ed.; Wiley: Chichester, 1980; Chapter 1, pp 1–82.
65. Olah, G. A.; Burrichter, A.; Rasul, G.; Prakash, G. K. S. J. Am. Chem. Soc. 1997,
119, 12929–12933, doi: 10.1021/ja972553o.
66. Hennel, J. W.; Klinowski, J., Magic-Angle Spinning: a Historical Perspective. In
New Techniques in Solid-State NMR, Klinowski, J., Ed.; Topics in Current Chemistry,
Vol. 246; Springer Berlin Heidelberg: 2005, pp 1–14, doi: 10.1007/b98646, http:
//dx.doi.org/10.1007/b98646.
67. Carberry, J. J.; Bretton, R. H. Axial dispersion of mass in flow through fixed beds.
AIChE J. 1958, 4, 367–375, doi: 10.1002/aic.690040327.
68. Liebeskind, L. S.; Srogl, J. Org. Lett. 2002, 4, 979–981, doi: 10.1021/ol0200091.
References and Notes 233
69. Lengar, A.; Kappe, C. O. Org. Lett. 2004, 6, 771–774, doi: 10.1021/ol036496h.
70. Dunn, P. J. Chem. Soc. Rev. 2012, 41, 1452–1461, doi: 10.1039/c1cs15041c.
71. O’Boyle, N. Python - the scripting language of chemistry., 2008, http://baoilleach.
blogspot.co.uk/2008/03/python-scripting-language-of-chemistry.html.
72. Personal communication from Vapourtec Ltd.
73. Zimmermann, J. Pyrimidin derivatives and process for their preparation. EU Patent,
0564409, 1993.
74. Zimmermann, J.; Sutter, B.; Bürger, H. M. Crystal modification of a N-phenyl-2-
pyrimidineamine derivative, processes for its manufacture and its use. US Patent,
6894051, 2005.
75. Buchdunger, E.; Capdeville, R.; Demetri, G. D.; Dimitrijevic, S.; Druker, B. J.;
Fletcher, J. A.; Heinrich, M. C.; Joensuu, H.; Silberman, S. L.; Tuveson, D. Treatment
of gastrointestinal stromal tumors. US Patent, 6958335, 2005.
76. Szczepek, W.; Luniewski, W.; Kaczmarek, L.; Samson-Lazinska, D.; Szelejew-
ski, W.; Skarzynski, M. A process for preparation of imatinib base. World Patent,
2006071130, 2006.
77. Loiseleur, O.; Kaufmann, D.; Abel, S.; Buerger, H. M.; Meisenbach, M.; Schmitz, B.;
Sedelmeier, G. N-phenyl-2-pyrimidine-amine derivatives. World Patent, 2003066613,
2003.
78. Kankan, R. N.; Rao, D. R. A process of preparing imatinib. World Patent, 2004074502,
2004.
79. Kompella, A.; Bhujanga Rao, A. K. S.; Venkaiah Chowdary, N.; Srinivas, R. Process
for the preparation of the anti-cancer drug imatinib and its analogues. World Patent,
2004108699, 2004.
80. Szakács, Z.; Béni, S.; Varga, Z.; Orfi, L.; Kéri, G.; Noszál, B. J. Med. Chem. 2005,
48, 249–255, doi: 10.1021/jm049546c.
81. Kompella, A. K.; Rachakonda, S.; Gampa, V. G. K.; Bhujanga Rao, A. K. S.;
Nannapaneni, V. C. Processes for the preparation of imatinib base and intermediates
thereof. World Patent, 2013035102, 2013.
82. Ma Wang, Z.; Zunyuan, W.; Zhen, M.; Zhengrong, S. Process for preparing N-
phenyl-2-pyrimidyl amine derivative. Chinese Patent, 1900073, 2007.
83. Xing, L.; Xungui, H.; Wang, Y.; Bekhazi, M.; Krivonos, S.; Danon, E. Imatinib
production process. World Patent, 2008135980, 2008.
84. Macdonald, P.; Rossetto, P. Process for the preparation of imatinib. World Patent,
2008051597, 2008.
85. Anli, H.; Xing, L.; Zelikovitch, L.; Kaspi, J. Process for preparing Imatinib. US
Patent, 7507821, 2009.
86. Maiti, D.; Fors, B. P.; Henderson, J. L.; Nakamura, Y.; Buchwald, S. L. Chem. Sci.
2011, 2, 57–68, doi: 10.1039/c0sc00330a.
234 References and Notes
87. Liu, Y.-F.; Wang, C.-L.; Bai, Y.-J.; Han, N.; Jiao, J.-P.; Qi, X.-L. Org. Process Res.
Dev. 2008, 12, 490–495, doi: 10.1021/op700270n.
88. Leonetti, F.; Capaldi, C.; Carotti, A. Tetrahedron Lett. 2007, 48, 3455–3458, doi:
10.1016/j.tetlet.2007.03.033.
89. Lange, H.; Carter, C. F.; Hopkin, M. D.; Burke, A.; Goode, J. G.; Baxendale, I. R.;
Ley, S. V. Chem. Sci. 2011, 2, 765–769, doi: 10.1039/c0sc00603c.
90. Sahoo, H. R.; Kralj, J. G.; Jensen, K. F. Angew. Chem. Int. Ed. 2007, 46, 5704–5708,
doi: 10.1002/anie.200701434.
91. O’Brien, A. G.; Horváth, Z.; Lévesque, F.; Lee, J. W.; Seidel-Morgenstern, A.;
Seeberger, P. H. Angew. Chem. Int. Ed. 2012, 51, 7028–7030, doi: 10.1002/anie.
201202795.
92. O’Brien, M.; Koos, P.; Browne, D. L.; Ley, S. V. Org. Biomol. Chem. 2012, 10,
7031–7036, doi: 10.1039/c2ob25912e.
93. Rubacha, M.; Rattan, A. K.; Hosselet, S. C. J. Lab. Autom. 2011, 16, 90–98, doi:
10.1016/j.jala.2009.01.002.
94. IDBS E-Workbook Suite, IDBS Business Solutions Ltd. Website: http://www.idbs.
com/en/platform-products/e-workbook/.
95. King, A. Waving goodbye to the paper lab book. Chemistry World Nov. 2011, 46–49.
96. King, A. Out on a LIM. Chemistry World Feb. 2014, http : / / www . rsc . org /
chemistryworld/2014/02/lims-lab-information-management.
97. Lab Information Management Systems (LIMS), Thermo Fisher Scientific Ltd. Web-
site: http://www.thermoscientific.com/en/products/lab-information-management-
systems-lims.html.
98. Delaney, N. F.; Rojas Echenique, J. I.; Marx, C. J. J. Lab. Autom. 2013, 18, 171–177,
doi: 10.1177/2211068212460237.
99. Peak Analysis and Automation Ltd. (PAA). Website: http://www.paa-automation.
com/.
100. PerkinElmer Integration Solutions. Website: http://www.perkinelmer.com/Catalog/
Category/ID/OnPoint%20Integration%20Solutions.
101. infoteam Software AG. Website: http://infoteam.de/.
102. Standards in Laboratory Automation (SiLA). Website: http://www.sila-standard.
org/.
103. Bär, H.; Hochstrasser, R.; Papenfub, B. J. Lab. Autom. 2012, 17, 86–95, doi: 10.
1177/2211068211424550.
104. McMullen, J. P.; Stone, M. T.; Buchwald, S. L.; Jensen, K. F. Angew. Chem. Int. Ed.
2010, 49, 7076–7080, doi: 10.1002/anie.201002590.
105. McMullen, J. P.; Jensen, K. F. Org. Process Res. Dev. 2010, 14, 1169–1176, doi:
10.1021/op100123e.
References and Notes 235
106. National Instruments LabVIEW. Website: http://www.ni.com/labview/ni-labview/.
107. Uniqsis Ltd. Website: http://www.uniqsis.com/.
108. Syrris Ltd. Website: http://www.syrris.com/.
109. OpenAPC (Open Advanced Process Control). Website: http://www.openapc.com/.
110. MathWorks MATLAB. Website: http://www.mathworks.co.uk/products/matlab/.
111. Parrott, A. J.; Bourne, R. A.; Akien, G. R.; Irvine, D. J.; Poliakoff, M. Angew. Chem.
Int. Ed. 2011, 50, 3788–3792, doi: 10.1002/anie.201100412.
112. Smith, C. J.; Nikbin, N.; Ley, S. V.; Lange, H.; Baxendale, I. R. Org. Biomol. Chem.
2011, 9, 1938–1947, doi: 10.1039/c0ob00815j.
113. Moore, J. S.; Jensen, K. F. Org. Process Res. Dev. 2012, 16, 1409–1415, doi: 10.
1021/op300099x.
114. Rueping, M.; Bootwicha, T.; Sugiono, E. Beilstein J. Org. Chem. 2012, 8, 300–307,
doi: 10.3762/bjoc.8.32.
115. Python Programming Language. Website: http://www.python.org/.
116. Wilson, J. A. J. Benefits Of Open Source Code., 2013, http://oss-watch.ac.uk/
resources/whoneedssource.
117. SciPy: open-source software for mathematics, science and engineering. Website:
http://www.scipy.org.
118. Experimental Flux reactor developed by Syrris Ltd.
119. Hook, B. D.; Dohle, W.; Hirst, P. R.; Pickworth, M.; Berry, M. B.; Booker-Milburn,
K. I. J. Org. Chem. 2005, 70, 7558–7564, doi: 10.1021/jo050705p.
120. Vapourtec UV-150 Photochemical Reactor. http://www.vapourtec.co.uk/products/
photochemicalreactor.
121. For example, the FlowSYNTH reactor from Milestone Inc. Website: http://www.
milestonesci.com/.
122. User Manual for VICI Microelectric Actuator. http://www.vici.com/support/tn/
tn415.pdf.
123. A wide range of remote desktop tools are available, including: VNC (such as Re-
alVNC, http://www.realvnc.com/), NoMachine NX (http://www.nomachine.com/),
Microsoft Remote Desktop (Included with Microsoft Windows) and TeamViewer
(http://www.teamviewer.com/). These products can allow portable devices to be
used for the remote operation of existing control software packages.
124. Arduino Uno Image by SparkFun Electronics. Reproduced under a Creative Com-
mons license from: http://en.wikipedia.org/wiki/File:Arduino_Uno_-_R3.jpg.
125. Microsoft .NET Gadgeteer and button Module image by Multichill. Reproduced
under a Creative Commons license from: http : / / en . wikipedia . org / wiki / File :
SytechMainboardAndGHIButton.png.
236 References and Notes
126. Raspberry Pi image by Jwrodgers, adapted under the Creative Commons Attribution
- Share Alike License, from: http://en.wikipedia.org/wiki/File:RaspberryPi.jpg.
127. Intel Galileo image by Regi51. Reproduced under a Creative Commons license from:
http://en.wikipedia.org/wiki/File:Embedded_World_2014_Intel_Galileo_01.jpg.
128. Arduino Uno Image by SparkFun Electronics. Reproduced under a Creative Com-
mons license from: http://en.wikipedia.org/wiki/File:Arduino_Uno_-_R3.jpg.
129. Microsoft .NET Gadgeteer and button Module image by Multichill. Reproduced
under a Creative Commons license from: http : / / en . wikipedia . org / wiki / File :
SytechMainboardAndGHIButton.png.
130. Raspberry Pi image by Jwrodgers, adapted under the Creative Commons Attribution
- Share Alike License, from: http://en.wikipedia.org/wiki/File:RaspberryPi.jpg.
131. Intel Galileo image by Regi51. Reproduced under a Creative Commons license from:
http://en.wikipedia.org/wiki/File:Embedded_World_2014_Intel_Galileo_01.jpg.
132. Arduino prototyping platform. Website: http://www.arduino.cc/.
133. Microsoft Research .NET Gadgeteer prototyping tool. Website: http://research.
microsoft.com/en-us/projects/gadgeteer/.
134. Intel Galileo Development Board. Website: http://www.intel.com/content/www/us/
en/intelligent-systems/galileo/galileo-overview.html.
135. Intel Edison Development Board. Website: http://www.intel.co.uk/content/www/uk/
en/do-it-yourself/edison.html.
136. Phidgets Single Board Computer (SBC), P/N 1073_0. Available from http://www.
phidgets.com/.
137. Octopus, version 0.1. Ingham, R. 2014. http://www.github.com/richardingham/
octopus/.
138. Ousterhout, J. Why Threads are a Bad Idea (for most purposes) In USENIX Winter
Technical Conference, Jan. 1996, http://www.cs.utah.edu/~regehr/research/ouster.
pdf.
139. Erb, B. Concurrent Programming for Scalable Web Architectures. Diploma Thesis,
Institute of Distributed Systems, Ulm University, Apr. 2012.
140. Dahl, R. node.js In JSConf, 2009, http://s3.amazonaws.com/four. livejournal/
20091117/jsconf.pdf.
141. Twisted event-driven networking engine. Website: http://www.twistedmatrix.com/.
142. Open Babel: The Open Source Chemistry Toolbox. Website: http://www.openbabel.
org/.
143. Python Data Analysis Library. Website: http://pandas.pydata.org/.
144. SimpleCV: Computer Vision platform using Python. Website: http://www.simplecv.
org/.
References and Notes 237
145. Syme, D.; Petricek, T.; Lomov, D. The F# Asynchronous Programming Model In
Principles and Applications of Declarative Languages, ACM SIGPLAN: 2011,
http://research.microsoft.com/apps/pubs/default.aspx?id=147194.
146. NumPy (Numerical Python). Website: http://www.numpy.org/.
147. Åström, K. J.; Murray, R. M., PID Control. In Feedback Systems: An Introduction
for Scientists and Engineers; Princeton University Press: Princeton, N.J., 2012;
Chapter 10, pp 293–314.
148. Ingham, R. J.; Battilocchio, C.; Hawkins, J. M.; Ley, S. V. Beilstein J. Org. Chem.
2014, 10, 641–652, doi: 10.3762/bjoc.10.56.
149. Carey, J. S.; Laffan, D.; Thomson, C.; Williams, M. T. Org. Biomol. Chem. 2006, 4,
2337–2347, doi: 10.1039/b602413k.
150. Dugger, R. W.; Ragan, J. A.; Ripin, D. H. B. Org. Process Res. Dev. 2005, 9, 253–
258, doi: 10.1021/op050021j.
151. Constable, D. J. C.; Dunn, P. J.; Hayler, J. D.; Humphrey, G. R.; Leazer, Jr., J. L.;
Linderman, R. J.; Lorenz, K.; Manley, J.; Pearlman, B. a.; Wells, A.; Zaks, A.;
Zhang, T. Y. Green Chem. 2007, 9, 411–420, doi: 10.1039/b703488c.
152. García-Álvarez, R.; Crochet, P.; Cadierno, V. Green Chem. 2013, 15, 46–66, doi:
10.1039/c2gc36534k.
153. Saaby, S.; Knudsen, K. R.; Ladlow, M.; Ley, S. V. Chem. Commun. 2005, 2909–
2911, doi: 10.1039/b504854k.
154. Battilocchio, C.; Hawkins, J. M.; Ley, S. V. Org. Lett. 2014, 16, 1060–1063, doi:
10.1021/ol403591c.
155. Battilocchio, C.; Hawkins, J. M.; Ley, S. V. Org. Lett. 2013, 15, 2278–2281, doi:
10.1021/ol400856g.
156. Chorghade, R.; Battilocchio, C.; Hawkins, J. M.; Ley, S. V. Org. Lett. 2013, 15,
5698–5701, doi: 10.1021/ol4027107.
157. Eichhorn, E.; Roduit, J.-P.; Shaw, N.; Heinzmann, K.; Kiener, A. Tetrahedron:
Asymm. 1997, 8, 2533–2536, doi: 10/cp2p2d.
158. Piperazine-2-carboxamide (39) listed by Fluorochem (P/N 068934) at £233/g on
14th Apr 2013.
159. U.S. Food and Drug Administration (FDA), PAT Guidance; tech. rep.; 2004.
160. Fabry, D. C.; Sugiono, E.; Rueping, M. Isr. J. Chem. 2014, 54, 341–350, doi: 10.
1002/ijch.201300080.
161. Routh, M. W.; Swartz, P. A.; Denton, M. B. Anal. Chem. 1977, 49, 1422–1428, doi:
10.1021/ac50017a033.
162. Morgan, E.; Burton, K. W.; Nickless, G. Chemometr. Intel. Lab. 1990, 8, 97–107,
doi: 10.1016/0169-7439(90)80127-R.
238 References and Notes
163. Schrijver, A., Theory of Linear and Integer Programming; Wiley-Blackwell: Chich-
ester, UK, 1998.
164. Cyr, P.; Deng, S. T.; Hawkins, J. M.; Price, K. E. Org. Lett. 2013, 15, 4342–4345,
doi: 10.1021/ol4018134.
165. Hogan, P. J.; Hopes, P. A.; Moss, W. O.; Robinson, G. E.; Patel, I. Org. Process Res.
Dev. 2002, 6, 225–229, doi: 10.1021/op0101052.
166. Jamieson, C.; Congreve, M. S.; Emiabata-Smith, D. F.; Ley, S. V. Synlett 2000, 2000,
1603–1607, doi: 10.1055/s-2000-7917.
167. Ley, S. V.; Ingham, R. J.; O’Brien, M.; Browne, D. L. Beilstein J. Org. Chem. 2013,
9, 1051–1072, doi: 10.3762/bjoc.9.118.
168. Ingham, R. J. figshare 2013, doi: 10.6084/m9.figshare.881895.
169. Vaxelaire, C.; Winter, P.; Christmann, M. Angew. Chem. Int. Ed. 2011, 50, 3605–
3607, doi: 10.1002/anie.201100059.
170. Gothard, C. M.; Soh, S.; Gothard, N.; Kowalczyk, B.; Wei, Y.; Baytekin, B.; Grzy-
bowski, B. Angew. Chem. Int. Ed. 2012, 51, 7922–7927, doi: 10 . 1002 / anie .
201202155.
171. Baxendale, I.; Schou, S.; Sedelmeier, J.; Ley, S. Chem. Eur. J. 2010, 16, 89–94, doi:
10.1002/chem.200902906.
172. Storer, R. I.; Takemoto, T.; Jackson, P. S.; Brown, D. S.; Baxendale, I. R.; Ley, S. V.
Chem. Eur. J. 2004, 10, 2529–2547, doi: 10.1002/chem.200305669.
173. Müller, E.; Berger, R.; Blass, E.; Sluyts, D.; Pfennig, A., Liquid–Liquid Extraction.
In Ullmann’s Encyclopedia of Industrial Chemistry, 7th ed.; Wiley-VCH Verlag
GmbH & Co. KGaA: 2011, doi: 10.1002/14356007.b03_06.pub2.
174. Ripperger, S.; Gösele, W.; Alt, C.; Loewe, T., Filtration, 2. Equipment. In Ullmann’s
Encyclopedia of Industrial Chemistry, 7th ed.; Wiley-VCH Verlag GmbH & Co.
KGaA: 2011, doi: 10.1002/14356007.n11_n01.pub2.
175. Stichlmair, J., Distillation, 2. Equipment. In Ullmann’s Encyclopedia of Industrial
Chemistry, 7th ed.; Wiley-VCH Verlag GmbH & Co. KGaA: 2011, doi: 10.1002/
14356007.o08_o01.
176. Billet, R., Evaporation. In Ullmann’s Encyclopedia of Industrial Chemistry, 7th ed.;
Wiley-VCH Verlag GmbH & Co. KGaA: 2011, doi: 10.1002/14356007.b03_03.
177. Hartman, R. L. Org. Process Res. Dev. 2012, 16, 870–887, doi: 10.1021/op200348t.
178. Koos, P.; Browne, D. L.; Ley, S. V. Green Process. Synth. 2012, 1, 11–18, doi:
10.1515/greenps-2011-0501.
179. Browne, D. L.; Deadman, B. J.; Ashe, R.; Baxendale, I. R.; Ley, S. V. Org. Process
Res. Dev. 2011, 15, 693–697, doi: 10.1021/op2000223.
180. Adamo, A.; Heider, P. L.; Weeranoppanant, N.; Jensen, K. F. Ind. Eng. Chem. Res.
2013, 52, 10802–10808, doi: 10.1021/ie401180t.
References and Notes 239
181. Syrris Asia FLLEX Module. Website: http : / / syrris . com/flow- products /asia -
modules/asia-fllex.
182. Biotage universal phase separator. Website: http:/ /www.biotage.com/product-
page/phase-separators.
183. Kashid, M. N.; Harshe, Y. M.; Agar, D. W. Ind. Eng. Chem. Res. 2007, 46, 8420–
8430, doi: 10.1021/ie070077x.
184. Aota, A.; Nonaka, M.; Hibara, A.; Kitamori, T. Angew. Chem. Int. Ed. 2007, 46,
878–880, doi: 10.1002/anie.200600122.
185. Deadman, B. J.; Battilocchio, C.; Sliwinski, E.; Ley, S. V. Green Chem. 2013, 15,
2050–2055, doi: 10.1039/c3gc40967h.
186. Borukhova, S.; Noël, T.; Metten, B.; de Vos, E.; Hessel, V. ChemSusChem 2013, 6,
2220–2225, doi: 10.1002/cssc.201300684.
187. Snead, D. R.; Jamison, T. F. Chem. Sci. 2013, 4, 2822–2827, doi: 10 . 1039 /
c3sc50859e.
188. Battilocchio, C.; Baxendale, I. R.; Biava, M.; Kitching, M. O.; Ley, S. V. Org.
Process Res. Dev. 2012, 16, 798–810, doi: 10.1021/op300084z.
189. Labbé-Jullié, C; Deschaintres, S; Gully, D; Le Fur, G; Kitabgi, P J. Pharmacol. Exp.
Ther. 1994, 271, 267–276.
190. Battilocchio, C.; Deadman, B. J.; Nikbin, N.; Kitching, M. O.; Baxendale, I. R.;
Ley, S. V. Chem. Eur. J. 2013, 19, 7917–7930, doi: 10.1002/chem.201300696.
191. Baxendale, I. R.; Cheung, S; Kitching, M. O.; Ley, S. V.; Shearman, J. W. Bioorg.
Med. Chem. 2013, 21, 4378–4387, doi: 10.1016/j.bmc.2013.04.075.
192. Nagasawa, H. T.; Elberling, J. A.; Shirota, F. N. J. Med. Chem. 1973, 16, 823–826,
doi: 10.1021/jm00265a016.
193. Pascal, R.; Taillades, J.; Commeyras, A. Tetrahedron 1980, 36, 2999–3008, doi:
10.1016/0040-4020(80)88025-2.
194. Paventi, M.; Chubb, F. L.; Edward, J. T. Can. J. Chem. 1987, 65, 2114–2117, doi:
10.1139/v87-351.
195. Ritter, J. J.; Minieri, P. P. J. Am. Chem. Soc. 1948, 70, 4045–4048, doi: 10.1021/
ja01192a022.
196. Pantano, M.; Formaggio, F.; Crisma, M.; Bonora, G. M.; Mammi, S.; Peggion, E.;
Toniolo, C.; Boesten, W. H. J.; Broxterman, Q. B. Macromolecules 1993, 26, 1980–
1984, doi: 10.1021/ma00060a028.
197. Dummann, G.; Quittmann, U.; Gröschel, L.; Agar, D. W.; Wörz, O.; Morgenschweis,
K. Catalysis Today 2003, 79-80, 433–439, doi: 10/frgnqr.
198. Calcinated Kieselguhr obtained from Sigma-Aldrich Co.
199. Activated charcoal powder obtained from Breckland Scientific Supplies Ltd.
240 References and Notes
200. Hu, D. X.; O’Brien, M.; Ley, S. V. Org. Lett. 2012, 14, 4246–4249, doi: 10.1021/
ol301930h.
201. Prat, D.; Pardigon, O.; Flemming, H.-W.; Letestu, S.; Ducandas, V.; Isnard, P.;
Guntrum, E.; Senac, T.; Ruisseau, S.; Cruciani, P.; Hosek, P. Org. Process Res. Dev.
2013, 17, 1517–1525, doi: 10.1021/op4002565.
202. Control script for the 3-step 2-aminoadamantane-2-carboxylic acid synthesis plat-
form: https://gist.github.com/richardingham/4ac997b5d720a4f7384f.
203. DIN 53170. Solvents for paints and varnishes – Determination of the evaporation
rate. Deutsches Institut für Normung., http://www.din.de/.
204. Schmidt, E. Y.; Vasil’tsov, A. M.; Mikhaleva, A. I.; Zaitsev, A. B.; Afonin, A. V.;
Toryashinova, D.-S. D.; Klyba, L. V.; Arndt, J.-D.; Henkelmann, J. ARKIVOC 2003,
35–44.
205. Mascia, S.; Heider, P. L.; Zhang, H.; Lakerveld, R.; Benyahia, B.; Barton, P. I.;
Braatz, R. D.; Cooney, C. L.; Evans, J. M. B.; Jamison, T. F.; Jensen, K. F.; Myerson,
A. S.; Trout, B. L. Angew. Chem. Int. Ed. 2013, 52, 12359–12363, doi: 10.1002/
anie.201305429.
206. Deadman, B. J. New Tools for Flow Chemistry and the Machine Assisted Synthesis
of Pharmaceuticals. PhD Thesis, University of Cambridge, 2013.
207. Schiaffo, C. E.; Dussault, P. H. J. Org. Chem. 2008, 73, 4688–4690, doi: 10.1021/
jo800323x.
208. Sorge, A. R.; Turco, M.; Pilone, G.; Bagnasco, G. J. Propul. Power 2004, 20, 1069–
1075, doi: 10.2514/1.2490.
209. Djaballah, H. Chemical space, high throughput screening and the world of block-
buster drugs. Drug Discov. World 2013, 14, 73–78.
210. Rees, D. C.; Congreve, M.; Murray, C. W.; Carr, R. Nat. Rev. Drug Disc. 2004, 3,
660–672, doi: 10.1038/nrd1467.
211. Zartler, D. E.; Swain, D. C.; Pearce, S. Fragment library design: the evolution of
fragment-based lead discovery. Drug Discov. World 2012, 13, 19–24.
212. Jhoti, H.; Williams, G.; Rees, D. C.; Murray, C. W. Nat. Rev. Drug Disc. 2013, 12,
644–645, doi: 10.1038/nrd3926-c1.
213. Desai, B.; Dixon, K.; Farrant, E.; Feng, Q.; Gibson, K. R.; van Hoorn, W. P.; Mills, J.;
Morgan, T.; Parry, D. M.; Ramjee, M. K.; Selway, C. N.; Tarver, G. J.; Whitlock, G.;
Wright, A. G. J. Med. Chem. 2013, 56, 3033–3047, doi: 10.1021/jm400099d.
214. Schneider, G.; Lee, M.-L.; Stahl, M.; Schneider, P. J. Comput. Aided Mol. Des. 2000,
14, 487–494, doi: 10.1023/A:1008184403558.
215. Dial-a-Molecule EPSRC Grand Challenge Network. Website: http://www.dial-a-
molecule.org/.
216. Schönherr, H.; Cernak, T. Angew. Chem. Int. Ed. 2013, 52, 12256–12267, doi:
10.1002/anie.201303207.
References and Notes 241
217. Guetzoyan, L.; Nikbin, N.; Baxendale, I. R.; Ley, S. V. Chem. Sci. 2013, 4, 764–769,
doi: 10.1039/c2sc21850j.
218. Filippakopoulos, P.; Knapp, S. FEBS Letters 2012, 586, 2692–2704, doi: 10.1016/j.
febslet.2012.04.045.
219. Bromosporine, a chemical probe developed by the SGC. Biological data and synthe-
sis scheme available at: http://www.thesgc.org/chemical-probes/bromosporine.
220. Kasai, K.-i.; Ishii, S.-i. J. Biochem. 1975, 77, 261–264.
221. Tateno, H.; Nakamura-Tsuruta, S.; Hirabayashi, J. Nat. Protoc. 2007, 2, 2529–2537,
doi: 10.1038/nprot.2007.357.
222. Chan, N. W. C.; Lewis, D. F.; Rosner, P. J.; Kelly, M. A.; Schriemer, D. C. Anal.
Biochem. 2003, 319, 1–12, doi: 10/cfzrjt.
223. Ng, W.; Dai, J.-R.; Slon-Usakiewicz, J. J.; Redden, P. R.; Pasternak, A.; Reid, N. J.
Biomol. Screen. 2007, 12, 167–174, doi: 10.1177/1087057106297567.
224. Ng, E. S. M.; Chan, N. W. C.; Lewis, D. F.; Hindsgaul, O.; Schriemer, D. C. Nat.
Protoc. 2007, 2, 1907–1917, doi: 10.1038/nprot.2007.262.
225. Calleri, E. et al. J. Med. Chem. 2010, 53, 3489–3501, doi: 10.1021/jm901691y.
226. Beckett, D.; Kovaleva, E.; Schatz, P. J. Protein Sci. 1999, 8, 921–929, doi: 10.1110/
ps.8.4.921.
227. Green, N. M. Methods Enzymol. 1990, 184, 51–67, doi: 10.1016/0076-6879(90)
84259-J.
228. Rossotti, H., The study of ionic equilibria: An introduction; Longman: London,
1978.
229. Lineweaver, H.; Burk, D. J. Am. Chem. Soc. 1934, 56, 658–666, doi: 10.1021/
ja01318a036.
230. Guetzoyan, L.; Ingham, R. J.; Nikbin, N.; Rossignol, J.; Wolling, M.; Baumert,
M.; Burgess-Brown, N. A.; Strain-Damerell, C. M.; Shrestha, L.; Brennan, P. E.;
Fedorov, O.; Knapp, S.; Ley, S. V. MedChemComm 2014, 5, 540–546, doi: 10.1039/
c4md00007b.
231. Kocevar, M.; Pozgan, F.; Polanc, S. Heterocycles 2001, 54, 1011–1019, doi: 10.
3987/COM-00-S(I)73.
232. Cox, J. M.; Edmondson, S. D.; Harper, B.; Weber, A. E. Preparation of aminopiperid-
ines as dipeptidyl peptidase IV inhibitors for the treatment or prevention of diabetes.
World Patent, 2006039325, 2006.
233. Sunder, S.; Peet, N. P. J. Heterocycl. Chem. 1980, 17, 1527–1529, doi: 10.1002/jhet.
5570170734.
234. Monge-Vega, A.; Aldana, I.; Ferández-Alvarez, E. J. Heterocycl. Chem. 1981, 18,
1533–1536, doi: 10.1002/jhet.5570180808.
242 References and Notes
235. Alazawe, S.; Elvidge, J. A. J. Chem. Soc. Perkin Trans. 1 1974, 696–698, doi:
10.1039/p19740000696.
236. Albrecht, B. K.; Harmange, J.-C.; Cote, A.; Taylor, A. M. Bromodomain inhibitors
and uses thereof. World Patent, 2012174487, 2012.
237. Cugnon de Sevricourt, M.; Robba, M. J. Heterocycl. Chem. 1978, 15, 977–979, doi:
10.1002/jhet.5570150614.
238. Zaher, A. F. A.; Khalil, O. M.; Refaat, H. M. Med. Chem. Res. 2011, 21, 3146–3153,
doi: 10.1007/s00044-011-9847-2.
239. Bourotte, M.; Delhomel, J.-F.; Dubernet, M.; Gouy, M.-H. Preparation of derivs.
of 6-substituted triazolopyridazines as Rev-Erb agonists for treating inflammatory,
circadian rhythm-related, or cardiometabolic disorders. World Patent, 2013045519,
2013.
240. Mizzoni, R. H.; Spoerri, P. E. J. Am. Chem. Soc. 1951, 73, 1873–1874, doi: 10.1021/
ja01148a537.
241. Li, A.-H.; Steinig, A. G.; Kleinberg, A.; Weng, Q.; Mulvihill, M. J.; Wang, J.; Chen,
X.; Wang, T.; Dong, H.; Jin, M. Furo- and Thieno [3,2-c] Pyridines. US Patent,
20090197864, 2009.
242. Calderwood, D. J.; Bonafoux, D. F.; Burchat, A.; Ding, P.; Frank, K. E.; Hoe-
mann, M. Z.; Mullen, K. D.; Davis, H. M. Novel triazolopyridazines. World Patent,
2009005675, 2009.
243. Bacque, E.; Nemecek, C.; Ugolini, A.; Wentzler, S. 6-(alkyl- or cycloalkyl-triazolo-
pyridazine-sulfanyl) benzothiazole derivatives: preparation, application as medica-
ments and use as met inhibitors. World Patent, 2011121223, 2011.
244. Dewdney, N. J.; Lou, Y.; Soth, M. Novel imidazo[1,2-a]pyridine and imidazo[1,2-
b]pyridazine derivatives. World Patent, 2009077334, 2009.
245. Lucas, M. C.; Goldstein, D. M.; Hermann, J. C.; Kuglstatter, A.; Liu, W.; Luk, K. C.;
Padilla, F.; Slade, M.; Villaseñor, A. G.; Wanner, J.; Xie, W.; Zhang, X.; Liao, C. J.
Med. Chem. 2012, 55, 10414–10423, doi: 10.1021/jm301367c.
246. Structural Genomics Consortium, Unpublished results.
247. Billingsley, K.; Buchwald, S. L. J. Am. Chem. Soc. 2007, 129, 3358–3366, doi:
10.1021/ja068577p.
248. Nikbin, N. Unpublished results.
249. Newman, M. S.; Gildenhorn, H. L. J. Am. Chem. Soc. 1948, 70, 317–319, doi:
10.1021/ja01181a097.
250. Shioiri, T.; Ninomiya, K.; Yamada, S. J. Am. Chem. Soc. 1972, 94, 6203–6205, doi:
10.1021/ja00772a052.
251. Spano, M. M.; Bickler, J. R. Heterocyclic Amines Purification on a New FLASH
Media - KP-NH. Biotage Application Note #40., 2003, http://www.biotage.com/
literature/download/1583_app_note_40.pdf.
References and Notes 243
252. Suginome, M.; Ohmura, T., Alkylboronic acid derivative cross-coupling reactions.
In Science of Synthesis: Cross Coupling and Heck-Type Reactions 1: C-C Cross
Coupling using Organometallic Partners, Molander, G. A., Ed.; Thieme: Stuttgart,
2013; Chapter 1.1.1.2, pp 147–201.
253. Angew. Chem. Int. Ed. 2010, 49, 8300, doi: 10.1002/anie.201006286.
254. Suzuki, A. Angew. Chem. Int. Ed. 2011, 50, 6722–6737, doi: 10 . 1002 / anie .
201101379.
255. Chemler, S. R.; Trauner, D.; Danishefsky, S. J. Angew. Chem. Int. Ed. 2001, 40,
4544–4568, doi: 10/dcfn97.
256. Guram, A. S.; Milne, J. E.; Tedrow, J. S.; Walker, S. D., Alkylboronic Acid Cross-
coupling Reactions. In Science of Synthesis: Cross Coupling and Heck-Type Reac-
tions 1: C-C Cross Coupling using Organometallic Partners, Molander, G. A., Ed.;
Thieme: Stuttgart, 2013; Chapter 1.1.1.1, pp 9–146.
257. Gillis, E. P.; Burke, M. D. J. Am. Chem. Soc. 2007, 129, 6716–6717, doi: 10.1021/
ja0716204.
258. Molander, G. A.; Canturk, B. Angew. Chem. Int. Ed. 2009, 48, 9240–9261, doi:
10.1002/anie.200904306.
259. Clayden, J.; Greeves, N.; Warren, S.; Wothers, P., Organometallic chemistry. In
Organic Chemistry; Oxford University Press: Oxford, 2001; Chapter 48, pp 1311–
1343.
260. Cepanec, I., The Suzuki-Miyaura Reaction. In Synthesis of Biaryls; Elsevier: 2004;
Chapter 5, pp 139–207, doi: 10.1016/B978-008044412-3/50006-3.
261. Billingsley, K. L.; Anderson, K. W.; Buchwald, S. L. Angew. Chem. Int. Ed. 2006,
45, 3484–3488, doi: 10.1002/anie.200600493.
262. Martin, R.; Buchwald, S. L. Acc. Chem. Res. 2008, 41, 1461–1473, doi: 10.1021/
ar800036s.
263. Strieter, E. R.; Buchwald, S. L. Angew. Chem. Int. Ed. 2006, 45, 925–938, doi:
10.1002/anie.200502927.
264. Miyaura, N. J. Organomet. Chem. 2002, 653, 54–57, doi: 10/ftq26v.
265. Wuts, P. G. M., Greene’s protective groups in organic synthesis, 4th ed.; Wiley-
Interscience: Hoboken, N.J., 2007.
266. Itoh, T.; Sato, K.; Mase, T. Adv. Synth. Catal. 2004, 346, 1859–1867, doi: 10.1002/
adsc.200404159.
267. Guetzoyan, L. Unpublished results.
268. Compounds 112 and 113 are described in Rossignol, J. Design and synthesis of a
series of bromodomain inhibitors; tech. rep.; University of Cambridge, 2012.
269. Malet-Sanz, L.; Madrzak, J.; Ley, S. V.; Baxendale, I. R. Org. Biomol. Chem. 2010,
8, 5324–5332, doi: 10.1039/c0ob00450b.
244 References and Notes
270. Ikemoto, N.; Liu, J.; Brands, K. M.; McNamara, J. M.; Reider, P. J. Tetrahedron
2003, 59, 1317–1325, doi: 10/fb8zp5.
271. Birum, G. H.; Jansen, R. F. Preparation of arylsulfonamides. US Patent, 4739116,
1988.
272. Reid, J. R.; Dufresne, R. F.; Chapman, J. J. Org. Synth. 1997, 74, 217–223, doi:
10.15227/orgsyn.074.0217.
273. Wolling, M. Unpublished results.
274. Ishiyama, T.; Murata, M.; Miyaura, N. J. Org. Chem. 1995, 60, 7508–7510, doi:
10.1021/jo00128a024.
275. Ishiyama, T.; Miyaura, N. J. Organomet. Chem. 2000, 611, 392–402, doi: 10.1016/
S0022-328X(00)00470-8.
276. CMPE product literature, Zeon Corporation. http://www.zeon.co.jp/business_e/
enterprise/spechemi/spechemi5-13.html.
277. Adamo, C.; Amatore, C.; Ciofini, I.; Jutand, A.; Lakmini, H. J. Am. Chem. Soc.
2006, 128, 6829–6836, doi: 10.1021/ja0569959.
278. Fürstner, A.; Seidel, G. Org. Lett. 2002, 4, 541–543, doi: 10.1021/ol0171463.
279. O’Brien, C. J.; Kantchev, E. A. B.; Valente, C.; Hadei, N.; Chass, G.; Lough,
A.; Hopkinson, A. C.; Organ, M. G. Chem. Eur. J. 2006, 12, 4743–4748, doi:
10.1002/chem.200600251.
280. Kantchev, E. A. B.; O’Brien, C. J.; Organ, M. G. Angew. Chem. Int. Ed. 2007, 46,
2768–2813, doi: 10.1002/anie.200601663.
281. Matos, K.; Burkhardt, E. R. Chim. Oggi 2005, 23, 12–17.
282. Ishiyama, T.; Itoh, Y.; Kitano, T.; Miyaura, N. Tetrahedron Lett. 1997, 38, 3447–
3450, doi: 10.1016/S0040-4039(97)00642-4.
283. Glide, version 5.8. Schrödinger LLC, New York, NY, 2012.
284. Huang, D.; Rossini, E.; Steiner, S.; Caflisch, A. ChemMedChem 2014, 9, 573–579,
doi: 10.1002/cmdc.201300156.
285. For example, the Accendo Conjure platform performs automated segmented-flow
experiments. Website: http://www.accendocorporation.com/.
286. For example, Chemspeed Technologies AG produce a variety of automated synthesis
platforms. Website: http://www.chemspeed.com/.
287. Braslau, R.; Rivera, F.; Tansakul, C. React. Funct. Polym. 2013, 73, 624–633, doi:
10.1016/j.reactfunctpolym.2013.01.009.
288. Lücking, U.; Cleve, A.; Haendler, B.; Faus, H.; Köhr, S.; Irlbacher, H. Sustituted
(heteroarylmethyl) thiohydantoins as anticancer drugs. World Patent, 2011029537,
2011.
289. Schmidt, H.-W.; Koitz, G.; Junek, H. J. Het. Chem. 1987, 24, 1305–1307, doi:
10.1002/jhet.5570240514.
References and Notes 245
290. Bonnet, M.; Flanagan, J. U.; Chan, D. A.; Lai, E. W.; Nguyen, P.; Giaccia, A. J.; Hay,
M. P. Bioorg. Med. Chem. 2011, 19, 3347–3356, doi: 10.1016/j.bmc.2011.04.042.
291. Kumar, D.; Kommi, D. N.; Chopra, P.; Ansari, M. I.; Chakraborti, A. K. Eur. J. Org.
Chem. 2012, 2012, 6407–6413, doi: 10.1002/ejoc.201200778.
292. Hernández, S.; Moreno, I.; SanMartin, R.; Gómez, G.; Herrero, M. T.; Domínguez,
E. J. Org. Chem. 2010, 75, 434–441, doi: 10.1021/jo902257j.
293. Kamal, A.; Dastagiri, D.; Ramaiah, M. J.; Reddy, J. S.; Bharathi, E. V.; Reddy, M. K.;
Sagar, M. V. P.; Reddy, T. L.; Pushpavalli, S. N. C. V. L.; Pal-Bhadra, M. Eur. J. Med.
Chem. 2011, 46, 5817–5824, doi: 10.1016/j.ejmech.2011.09.039.
294. Liu, Y.; Bai, Y.; Zhang, J.; Li, Y.; Jiao, J.; Qi, X. Eur. J. Org. Chem. 2007, 2007,
6084–6088, doi: 10.1002/ejoc.200700577.
295. Leproult, E.; Barluenga, S.; Moras, D.; Wurtz, J.-M.; Winssinger, N. J. Med. Chem.
2011, 54, 1347–1355, doi: 10.1021/jm101396q.
296. Kagarlitskii, A.; Krichevskii, L.; Suvorov, B. Pharm. Chem. J. 1993, 27, 219–220,
doi: 10.1007/BF00773004.
297. Bach, F. L.; Kushner, S.; Williams, J. H. J. Am. Chem. Soc. 1955, 77, 6049–6051,
doi: 10.1021/ja01627a075.
298. Sum, Y. N.; Yu, D.; Zhang, Y. Green Chem. 2013, 15, 2718–2721, doi: 10.1039/
c3gc41269e.
299. Koval’skaya, S. S.; Dikusar, E. A.; Kozlov, N. G.; Popova, L. A. Russ. J. Org. Chem.
2001, 37, 1225–1227, doi: 10.1023/A:1013102815024.
300. Eguchi, S.; Arasaki, M. J. Chem. Soc. Perkin Trans. 1 1988, 1047–1050, doi: 10.
1039/p19880001047.
301. Kuroda, Y.; Ueda, H.; Nozawa, H.; Ogoshi, H. Tetrahedron Lett. 1997, 38, 7901–
7904, doi: 10/dh27rp.
302. Liscio, P.; Carotti, A.; Asciutti, S.; Karlberg, T.; Bellocchi, D.; Llacuna, L.; Mac-
chiarulo, A.; Aaronson, S. A.; Schüler, H.; Pellicciari, R.; Camaioni, E. J. Med.
Chem. 2014, 57, 2807–2812, doi: 10.1021/jm401356t.
303. Qiu, D.; Wang, S.; Tang, S.; Meng, H.; Jin, L.; Mo, F.; Zhang, Y.; Wang, J. J. Org.
Chem. 2014, 79, 1979–1988, doi: 10.1021/jo402618r.