Standard housing temperature but not β3-adrenoreceptor agonism drives weight-gain and adipose tissue deposition in rats following diet induced obesity at thermoneutrality

Background and aim Rodents are commonly housed below thermoneutrality and this exposure to ‘cold’ (i.e. 20°C) activates thermogenic brown (BAT) and beiging of white adipose tissue. Here, we examined whether a standard housing temperature (i.e. 20°C, a reduction in temperature of ∼8°C) or YM-178, a highly-selective β3-adrenoreceptor agonist, in obese animals raised at thermoneutrality, would impact differently on classical BAT or subcutaneous inguinal (IWAT) beige depots. Methods Eighteen weanling Sprague-Dawley rats were housed at thermoneutrality (28°C) and fed a high-fat diet. At 12 weeks, 6 animals were randomised to either standard housing temperature (20°C, n=6) or to β3-A R agonist administration (28°C+ β3, 0.75mg/kg/d, n=6) for 4 weeks. Metabolic assessment was undertaken during the final 48h, followed by interscapular, perivascular BAT and IWAT sampling for the analysis of thermogenic genes and the proteome. Results Exposure to 20°C increased weight gain, BAT and IWAT mass. Proteomic analysis of BAT revealed novel pathways associated with cold-induced weight gain (i.e. histone deacetylation, glycosaminoglycan degradation and glycosphingolipid biosynthesis) whilst β3- adrenoreceptor agonism impacted on proteins involved in skeletal muscle contraction and cell differentiation. IWAT of cold-exposed animals exhibited an enrichment of proteins involved NAD+ binding, plus retinol and tyrosine metabolic pathways whilst β3-AR agonism downregulated ribosomal and upregulated acute phase response proteins. Conclusion Following diet-induced obesity at thermoneutrality, exposure to 20°C promotes subcutaneous fat deposition in order to reduce heat loss and defend body temperature. In contrast, chronic administration of β3-AR agonist has minimal metabolic-related effects on adipose tissue.


Introduction
Therapeutic activation of thermogenic brown adipose tissue (BAT) may be feasible to prevent, or treat, cardiometabolic disease [1]. In rodent models of obesity, the activation of BAT and UCP1-positive beige adipocytes in white adipose tissue (WAT) by cold exposure and sympathomimetics (i.e. β 3-agonists) can attenuate or reverse obesity, diabetes and atherosclerosis, thus eliciting beneficial effects on metabolic health [1]. One factor which could influence these outcomes is that animals are typically housed at temperatures well below their thermoneutral zone (which for a rodent is c. 28°C) [2]. Under these conditions, not only is BAT active, but UCP1+ beige adipocytes are readily seen in the inguinal WAT (IWAT) depot which is considered 'beige'.
Housing animals at thermoneutrality modifies the thermogenic and metabolic response. For example, fasting glucose, food intake and metabolic rate are reduced at a higher ambient temperature [3]. Furthermore, markers of 'browning' and BAT activation are reduced at thermoneutrality, whilst the primary organ responsible for glucose uptake becomes skeletal muscle and not BAT [3]. Long term housing at thermoneutrality reduces sympathetic drive in BAT, as demonstrated by reduced tyrosine hydroxylase and norepinephrine turnover [4].
This results in increased adiposity at thermoneutrality, despite a lower energy intake compared to animals housed at 20°C. Thermoneutral housing also attenuates the response to common thermogenic stimuli. For instance, the thermogenic effects of thyroxine become independent of UCP1 due to the onset of pyrexia [5]. Similarly, the induction of 'browning' seen with exercise training at standard housing is attenuated in animals housed at thermoneutrality [6].
The 'cold-stressed' animal has been widely studied but much less is known about the underlying adaptations in adipose tissue in animals maintained at thermoneutrality. Usually, temperatures as low as 4°C, which represents an 'extreme cold', are used to activate BAT, with the induction of UCP1 and subsequent thermogenic response primarily seen in subcutaneous IWAT and other 'beige' depots [7]. Conversely, when animals which have been housed at thermoneutrality and are then exposed to 20°C, the induction of UCP1 is primarily seen in BAT. These differences highlight that there are two key steps to the 'browning' process and emphasise the need to study metabolism under more translationally relevant conditions [7]. Importantly, it was recently demonstrated that BAT from thermoneutrally housed, obese animals closer resembles human BAT and, that this model of 'humanised BAT' represents the best choice for modelling BAT physiology [8]. Here, we aimed to determine the metabolic response to modest cold exposure (i.e. standard housing temperatures (20°C)), which constitutes a drop of c. 8°C and chronic β 3-AR agonism in animals raised from weaning at thermoneutrality. Using this model, in which animals are not 'cold-stressed' and BAT is chronically inactive, we have demonstrated UCP1 mRNA is absent in subcutaneous IWAT (the classical 'beige' depot) and not induced with exercise training, a common response at standard housing temperatures [9]. We hypothesised that exposure to a mild cold stimulus in animals raised at thermoneutrality would have modest effects on systemic metabolism but not weight per se. Instead, we demonstrate rapid weight gain in response to an 8°C reduction in ambient temperature and deposition of substantial amounts (125%) of subcutaneous adipose tissue through an, as yet, unexplained mechanism. These effects were not seen in animals treated with the highly-selective β 3adrenoreceptor agonist, YM-178, suggesting a direct effect of ambient temperature which is uncoupled from sympathetic activation. Using adipose tissue proteomics, we then uncovered novel pathways associated with this cold-induced weight gain illustrating a potentially novel mechanism through which animals defend body temperature in the absence of a thermogenic response.

Animals, cold exposure and YM-178 treatment
All studies were approved by the University of Nottingham Animal Welfare and Ethical Review Board, and were carried out in accordance with the UK Animals (Scientific Procedures) Act of 1986. Eighteen male Sprague-Dawley rats aged 3 weeks were obtained from Charles River (Kent, UK) and housed immediately at thermoneutrality (c.28°C), on a high-fat diet (HFD; 45%, 824018 SDS, Kent, UK), under a 12:12-hour reverse light-dark cycle (lights off at 08:00, on at 20:00). These conditions were chosen so as to closer mimic human physiology [7], minimise animal stress and maximise data quality and translatability [10]. At 12 weeks of age, all animals were randomised to 4 weeks of standard housing temperature (20°C, n=6), YM-178 (28°C+β3, 0.75mg/kg/day, n=6) administration or HFD controls (28°C, n=6) from 12 weeks of age. In adherence to the NC3Rs, this experiment was ran alongside our work looking at the effect of exercise training on 'browning' and utilised the same cohort of control HFD animals.

Metabolic cages
All animals were placed in an open-circuit calorimeter (CLAMS: Columbus Instruments, Linton Instrumentation, UK) for 48h towards the end of the study. Assessment of whole body metabolism was carried out as previously described [11], after which all animals were weighed and fasted overnight prior to euthanasia by rising CO 2 gradient. BAT, perivascular BAT (PVAT) and IWAT were then rapidly dissected, weighed, snap-frozen in liquid nitrogen and stored at -80°C for subsequent analysis.

Gene expression analysis
Total RNA was extracted from each fat depot using the RNeasy Plus Micro extraction kit (Qiagen, West Sussex, UK) using an adapted version of the single step acidified phenolchloroform method. RT-qPCR was carried out as previously described [11] using rat-specific oligonucleotide primers (Sigma) or FAM-MGB Taqman probes [11]. Gene expression was determined using the GeNorm algorithm against two selected reference genes; RPL19:RPL13a in BAT and IWAT (stability value M = 0.163 in BAT and 0.383 in IWAT) and RPL19:HPRT1 in PVAT (stability value M = 0.285).

Mass spectrometry
Protein extraction, clean up and trypsinisation was carried out as previously described [11].
Briefly, 50-100 mg of frozen tissue was homogenised in 500 μL CellLytic MT cell lysis buffer (Sigma, C3228) prior to removal of lipid and other contaminants using the ReadyPrep 2D cleanup Kit (Biorad, 1632130). Samples were then subjected to reduction, alkylation and overnight trypsinisation following which they were dried down at 60°C for 4 h and stored at 80°C before resuspension in LCMS grade 5% acetonitrile in 0.1% formic acid for subsequent analysis. Analysis by mass spectrometry was carried out as previously described on a SCIEX TripleTOF 6600 instrument [12]. Briefly, samples were analysed in both SWATH (Data Independent Acquisition) and IDA (Information Dependent Acquisition) modes for quantitation and spectral library generation respectively. IDA data was searched together using ProteinPilot 5.0.2 to generate a spectral library and SWATH data was analysed using Sciex OneOmics software [13] extracted against the locally generated library as described previously [11].

Statistical analysis
Statistical analysis was performed in GraphPad Prism version 8.0 (GraphPad Software, San Diego, CA). Data are expressed as Mean±SEM with details of specific statistical tests in figure legends. Functional analysis of the proteome was performed using the Advaita Bioinformatic iPathwayGuide software (www.advaitabio.com/ipathwayguide.html) (fold change ± 0.5 and confidence score cut-off of 0.75). Significantly impacted biological processes, molecular interactions and pathways were analysed in the context of pathways obtained from the Kyoto Encyclopaedia of Genes and Genomes (KEGG) database (Release 84.0+/10-26, Oct 17) [14] and the Gene Ontology Consortium database (2017-Nov) [15].
The Elim pruning method, which removes genes mapped to a significant GO term from more general (higher level) GO terms, was used to overcome the limitation of errors introduced by considering genes multiple times [16].

Exposure to standard housing temperature drives weight gain and adiposity
Four weeks of exposure to 20°C increased weight gain and body weight with greater IWAT and BAT mass, an effect not seen with YM-178 ( Fig. 1A-E). When body mass was incorporated into an ANCOVA, there was a significant relationship between body mass and energy expenditure (r 2 0.926, p=0.038, Fig.1F) and body mass and energy intake (r 2 =0.852, p=0.025, Fig. 1H) in animals exposed to standard housing temperatures, whilst there was a change in the direction of the relationship between body mass and energy intake in animals treated with YM-178 (Fig. 1H). Cold-exposed animals exhibited an increase in light-phase, and a reduction in dark-phase ambulatory activity which was also seen in animals treated with YM-178 (Fig. 1G). There was no change in serum metabolites (i.e. glucose, triglycerides and NEFA) or hormones (i.e. insulin and leptin) ( Fig. 1I-N).
Next, we analysed a panel of primary thermogenic genes in BAT and PVAT as we have previously shown these two anatomically and developmentally distinct BAT depots exhibit a divergent response to both a brief period of nutrient excess and exercise training [9,11].
Despite an increase in BAT mass with cold, there was no reduction in thermogenic genes (i.e. UCP1) in this depot or in PVAT with either cold or YM-178 ( Fig. 2A-B). Expression of the BAT marker CITED1 and beige marker TMEM26 were reduced in BAT of both cold exposed and YM-178 treated animals whilst the beige marker P2RX5 was upregulated in PVAT of both groups. Using targeted arrays to screen for key genes involved in adipose tissue metabolism we saw an increase in the expression of FASN mRNA in both BAT and PVAT but not IWAT following cold exposure ( Fig. 2A-C). There was also an increase of genes involved in glycolysis (i.e. HK2 and PDK), fatty acid oxidation (i.e. ACACA and ACACB) and insulin resistance (i.e. AdipoR1 and MAPK9) in PVAT only following cold exposure ( Fig. 2D -K). This suggests that the metabolic response to cold in PVAT is uncoupled from BAT activation. We then analysed the same panel of thermogenic genes in IWAT to assess the 'browning' capacity of each intervention (Fig. 2C). UCP1 mRNA was absent in IWAT of all animals and, despite a c.125% increase in IWAT mass of cold exposed animals, there was no change in the expression of other key genes associated with thermogenesis (i.e. ADRβ3), beige adipocytes (i.e. TMEM26) or lipogenesis (i.e. FASN).

Identification of differentially regulated proteins in BAT and WAT associated with cold-induced increases in fat mass
Given the increase in BAT and IWAT mass seen in response to standard housing temperature, we carried out proteomic analysis to elucidate potential contributors to this physiological response identifying 175 differentially regulated proteins in BAT of coldexposed animals (Table 1 for top 20 proteins, Supp. Data for full list). This method, which quantifies the 30-40% most abundant proteins across samples, did not detect UCP1 in BAT.
There was, however, an increase in numerous other metabolic proteins in BAT of cold exposed animals including a pronounced increase in the mitochondrial citrate transporter proteins (SLC25a1), together with glucose-6-phosphate dehydrogenase (G6PD), phosphogluconate dehydrogenase (PGD), succinate dehydrogenase complex flavoprotein subunit a (SDHA), fatty acid binding protein 5 (FABP5) and the muscle isoforms of phosphoglycerate mutase (PGAM2) and creatine kinase (CKm). The increase in these metabolic proteins occurred alongside a downregulation of proteins involved in the import of membrane (TIMM9) and cytoplasmic (TIMM13) proteins to the inner mitochondrial membrane, mitochondrial amino acid production (BCAT2) and redox state (PRDX4, GLRX3 and SOD2).
Conversely, 137 proteins were differentially regulated in BAT of YM-178 treated animals (Table 1 for  In IWAT, we detected 116 proteins differentially regulated in cold-exposed animals (Table 2 for top 20 significant proteins, Supp. Data for full list) with the upregulation of numerous lipid metabolic proteins including fatty acid binding proteins 1 (FABP1) and 3 (FABP3), fatty acid synthase (FASN) and ATP citrate lyase (ACLY) as well as those involved in beta oxidation including hydroxynacyl-CoA dehydrogenase (HADH), hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit beta (HADHB) and acyl-CoA dehydrogenase long chain (ACADL). In addition, there was an upregulation of glycerol kinase (GK), pyruvate carboxylase (PC) and monocarboxylic acid transporter 1 (MCT1) and these alterations in IWAT metabolism were accompanied by a downregulation of multiple proteins involved in mRNA processing and splicing (i.e. RNA binding motif protein 8B, RBM8B; dexd-Box helicase 39B, DDX39B and poly(u) binding splicing factor 60, PUF60).

Functional analysis of the BAT and WAT proteome
We then carried out functional analysis of the BAT and IWAT proteome in order to elucidate potential biological processes, functions and pathways in adipose tissue that may be contributing to the significant increase in BAT and IWAT mass seen following cold-exposure.
The differentially regulated proteins in BAT of cold exposed animals enriched GO terms involved 'glucose import', 'ATP-dependent helicase activity' and 'regulation of protein phosphorylation' whilst there was also an enrichment of nuclear related GO terms including 'histone deacetylation', 'nucleosomal DNA binding', 'nuclear chromatin' and the 'nucleosome' (Table 3 Impact analysis, meanwhile, demonstrated that cold-exposure was associated with an impact on both retinol and tyrosine metabolic pathways (Table 6

Elucidating the 'interactome' in BAT and IWAT
We then sought to discover putative hub proteins and signalling pathways in these tissues, through characterisation of the 'interactome' through protein-protein interactions.
Differentially regulated proteins in BAT of cold-exposed animals were involved in 22 networks (i.e. the 'continent' and 21 'islands') with the 'continent' consisting of 293 interacting proteins. We identified multiple hub proteins in this network including the G proteins G protein subunit alpha I3 (GNAI3) and G protein subunit alpha Q (GNAQ), protein tyrosine phosphatase non-receptor type 11 (PTPN11), actin gamma 1 (ACTG1) and serine/threonine kinase 24 (STK24) (Fig. 7A). Further analysis demonstrated that these hub proteins interacted with numerous other G proteins (i.e. Adenosine A1 receptor, ADORA1; Free fatty acid receptor 3, FFAR3 and neuropeptide Y receptor 2R; NPY2R), in addition to proteins involved in growth signalling pathways (i.e. phosphoinositide-3-kinase regulatory subunit 2; PIK3R2) and adipogenesis (platelet derived growth factor receptor beta; PDGFRB) to enrich 48 biological processes (Supp. Data) including 'anatomical structure

Discussion
Housing temperature, especially cold exposure, impacts on metabolic homeostasis as illustrated by the effects it has on BAT and 'browning'. However, much less is known on the effects of 'browning' interventions in animals maintained at thermoneutrality [2,17]. Our major finding is that chronic exposure to a mild-cold stimulus (i.e. standard housing temperature) drives weight gain and the deposition of large quantities of subcutaneous adipose tissue in obese animals rather than the activation of BAT and subsequent weightloss as would be expected. This effect is not seen in animals treated with the highly selective β 3-agonist YM-178, suggesting it is independent of sympathetic activation and a direct effect of ambient temperature, pointing towards a novel mechanism whereby animals increase body weight and fat mass in the absence of prior cold-stress to potentially reduce heat loss and thus defend body temperature.
Whilst we do not detect any change in energy expenditure and energy intake in animals exposed to 20°C, it is important to note that metabolic cages are not truly representative of the entire study period and that, even with an adaptation period, it is an artificial environment where changes in energy intake are not always be detected [18]. Furthermore, at 600-700g these animals are likely to be near the size limit of the metabolic cages, such that home cage systems would be better-suited to collect physiological data in these large animals.
These issues could also explain the relationship between body weight and both energy expenditure and energy intake, which suggests a reduction in both as body mass increases though the opposite is true in animals treated with the β 3-agonist YM-178. Nevertheless, there is accumulating evidence that, in the absence of adaptive thermogenesis, other mechanisms compensate in order to maintain body temperature. For instance, in UCP1 k/o mice, a reduction in BAT thermogenesis leads to a compensatory increase in shivering thermogenesis in skeletal muscle [19]. Similarly, when shivering thermogenesis is impaired in sarcolipin k/o mice, there is a compensatory increase in BAT activity [19]. In obese animals lacking BAT, there seems to be an entirely different homeostatic response to coldstress where following BAT lipectomy, obese, cold-exposed rats gain weight and exhibit nearly a one fold increase in adipose tissue mass [20]. There is also evidence to suggest that intermittent cold-exposure (from 20°C to 4°C) over a period of days increases weight gain and adiposity and that this is associated with transient, intermittent increases in energy intake [21]. Although other mechanisms such as intestinal growth and increased fatty acid absorption [22] may also play a role the impact of BAT, lipectomy on the response to coldexposure mirrors our own findings and would suggest a compensatory mechanism for heat production in obesity when adaptive thermogenesis is insufficient. Importantly, an increase of c. 0.5g in BAT and of c. 10g in IWAT cannot account for the c. 41g higher weight gain seen in the cold-exposed animals. Given that liver mass was not higher in cold-exposed animals, and that the increase in IWAT was highest, we speculate that this weight is not due to increased organ mass but due an increase in mass of dermal adipose tissue, given the putative role of this depot in regulating insulation and this should be investigated in future studies. A useful comparison here could be the Siberian hamster where obesity, and weight loss, can be manipulated by photoperiod. In this model, reductions in fat mass initially occur independently of reductions in energy intake and subcutaneous AT mass is preserved, likely due to its insulative properties highlighting how, at least in the short term, changes in weight and fat mass can be uncoupled from energy balance [23].
Another potentially important finding of this study is that the increase in weight gain and subcutaneous AT mass seen in cold-exposed animals was not associated with impaired metabolic parameters (i.e. fasting glucose and lipids) or any discernible adipose tissue dysfunction. Here, chronic exposure to a mild cold stimulus seemingly drives a phenotypically healthy expansion of subcutaneous AT, increasing it by 125% compared to controls. By using proteomics, we were able to elucidate processes in both BAT and IWAT associated with this expansion of AT mass. An increase in genes that modulate metabolism, including those involved in glycolysis, the TCA cycle and lipogenesis, suggests that there is an increase in metabolic flux of these depots which ultimately results in net lipogenesis. In BAT, this is associated with an enrichment of GO terms associated with DNA metabolism (i.e. histone deacetylation) with CTBP1, HDAC1 and SIRT2 all previously shown to be negative regulators of lipogenesis by either binding or deacetylating PPARY [24][25][26]. In IWAT, we show an enrichment of proteins involved in retinol metabolism, a well-known pathway which drives both adipogenesis and lipogenesis which is dependent on an intact NAD+ binding motif [27]. The enrichment of multiple proteins involved in NAD+ binding suggests this pathway is intact and upregulated and may contribute to this increase in IWAT mass.
We were also able to elucidate the impact of sympathetic activation in the absence of any change in housing temperature by administering YM-178. This highly selective β 3-agonist increases BAT activity, whole body EE and induces 'browning' in humans [28][29][30][31].
Administration of YM-178 to mice for 4-6 weeks at similar dosage to this study (i.e. 0.8mg/kg) reduces fat mass, increases UCP1 in IWAT and glucose uptake in BAT at standard housing temperature (SHT) whilst driving atherosclerosis, through BAT lipolysis in ApoE-/-and LdlR-/-mice [32]. Here, we show YM-178 treatment in obese animals raised at thermoneutraility has no impact on thermogenesis although it does modulate specific proteins involved in cell differentiation and skeletal muscle contraction. YM-178 also impacted on proteins involved in the acute phase response and inflammation in both BAT and IWAT with many increasing in the latter. Taken together, this suggests that YM-178, when given to obese rats, acts negatively on adipose tissue by driving inflammatory processes. Another potentially important finding is that chronic YM-178 induces the downregulation of ribosomal proteins. Adrenergic agonists have previously been shown to modify protein synthesis in skeletal muscle but the effect of YM-178 has yet to be examined with human studies focussing on overactive bladder and associated symptoms [33,34]. The physiological relevance of a reduction in ribosomal proteins, and perturbed proteins synthesis in IWAT with YM-178 treatment and, whether this translates to human patients is as yet unclear.
Elucidating protein-protein interactions (PPI) is an important step in understanding both communication between proteins following an intervention and the identification of novel signalling pathways. We have previously shown that ribosomal proteins are important hub proteins in PPI networks following exercise training and, in the present study, show that these also respond to . Furthermore, these interact with multiple other mitochondrial ribosomal proteins and proteins governing energy metabolism such as mTOR and leptin. The role, and physiological significance, of these adaptations in putative ribosomal regulated signalling pathways remain to be determined but may assist in our understanding of the sympathetic control of adipose tissue. In BAT, the identification of GNAI3 and GNAC as hub proteins is of particular interest given the role of GPCRs and the Gq signalling pathway, in particular, given its role in inhibiting BAT metabolism and 'browning' [35].

Conclusion
We propose that chronic suppression of BAT from weaning, through the use of thermoneutral housing, changes the physiological response to cold in the obese state.
Instead, these animals deposit adipose tissue and gain weight, an effect not seen with YM-178, suggesting a direct effect of temperature, whereby an insulative mechanism is recruited when a thermogenic response is absent.

Availability of data and material:
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.