Structure guided prediction of Pyrazinamide resistance mutations in pncA

Change log
Karmakar, Malancha 
Rodrigues, Carlos HM 
Horan, Kristy 
Denholm, Justin T 

Pyrazinamide plays an important role in tuberculosis treatment; however, its use is complicated by side-effects and challenges with reliable drug susceptibility testing. Resistance to pyrazinamide is largely driven by mutations in pyrazinamidase (pncA), responsible for drug activation, but genetic heterogeneity has hindered development of a molecular diagnostic test. We proposed to use information on how variants were likely to affect the 3D structure of pncA to identify variants likely to lead to pyrazinamide resistance. We curated 610 pncA mutations with high confidence experimental and clinical information on pyrazinamide susceptibility. The molecular consequences of each mutation on protein stability, conformation, and interactions were computationally assessed using our comprehensive suite of graph-based signature methods, mCSM. The molecular consequences of the variants were used to train a classifier with an accuracy of 80%. Our model was tested against internationally curated clinical datasets, achieving up to 85% accuracy. Screening of 600 Victorian clinical isolates identified a set of previously unreported variants, which our model had a 71% agreement with drug susceptibility testing. Conclusion: Here, we have shown the 3D structure of pncA can be used to accurately identify pyrazinamide resistance mutations. SUSPECT-PZA is freely available at:

Amidohydrolases, Amino Acid Substitution, Antitubercular Agents, DNA Mutational Analysis, DNA, Bacterial, Datasets as Topic, Drug Resistance, Bacterial, Humans, Machine Learning, Microbial Sensitivity Tests, Models, Genetic, Models, Molecular, Mutation, Mycobacterium tuberculosis, Protein Structure, Tertiary, Pyrazinamide, Structure-Activity Relationship, Tuberculosis
Journal Title
Scientific Reports
Conference Name
Journal ISSN
Volume Title
Nature Publishing Group
Medical Research Council (MR/M026302/1)
M.K. and C.M.H.R. were funded by the Melbourne Research Scholarship. Funding for genomic sequencing was provided by the Department of Health and Human Services, Victoria. D.B.A. was funded by a Newton Fund RCUK-CONFAP Grant awarded by The Medical Research Council (MRC) and Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) (MR/M026302/1), the Jack Brockhoff Foundation (JBF 4186, 2016), and a C. J. Martin Research Fellowship from the National Health and Medical Research Council (NHMRC) of Australia (APP1072476). This work was supported in part by the Victorian Government's OIS Program.