Dysregulation of macrophage PEPD in obesity determines adipose tissue bro-inammation and insulin resistance

Fibrosis is a hallmark of adipose tissue (AT) dysfunction and obesity-associated insulin resistance that results from an impaired collagen turnover. Peptidase D (PEPD) plays a vital role in collagen turnover by degrading proline-containing dipeptides. Nevertheless, its specic function and importance in AT is unknown. GWAS identied the rs731839 variant in the locus near PEPD that uncouples obesity from insulin resistance and dyslipidaemia, thus indicating that defective PEPD might impair AT remodelling and exacerbate metabolic complications. Here we show that in human and murine obesity, PEPD expression and activity decrease in AT, coupled to the release of PEPD systemically. Both events, in turn, are associated with the accumulation of brosis in AT and insulin resistance. Using pharmacologic and genetic animal models of PEPD down-regulation, we show that whereas dysfunctional PEPD activity provokes AT brosis, it is the PEPD secreted by AT the main contributor to inammation, insulin resistance and metabolic dysfunction. Also, PEPD originated in inammatory macrophages (M(cid:0)), plays an essential role promoting bro-inammatory responses via activation of EGFR in M(cid:0) and preadipocytes. Using genetic ablation of pepd in M(cid:0) that prevents obesity-induced PEPD release, also averts AT bro-inammation and obesity-associated metabolic dysfunctions. Taking advantage of factor analysis, we have identied the coupling of prolidase decreased activity and increased systemic levels of PEPD as the essential pathogenic triggers of AT brosis and insulin resistance. Thus, PEPD produced by M(cid:0) qualies as a biomarker of AT bro-inammation and a therapeutic target for AT brosis and obesity-associated insulin resistance and type 2 diabetes.


Main
We measured PEPD mRNA expression in adipose tissue (AT) from seven human cohorts spanning a broad spectrum of BMIs, ages and degrees of insulin resistance (Extended data Figure 1a). PEPD expression was lower in visceral white AT (VsW) of obese individuals than in lean ones and also lower in obese type 2 diabetics than in obese normoglycemic patients (Figure 1a, b). PEPD was lower in VsW depot than in subcutaneous white AT (ScW) in a paired sample cohort study ( Figure 1c) and lower PEPD expression in VsW was associated with higher BMI and collagen content in AT from obese individuals (measured as hydroxyproline levels, HPro) ( Figure 1d). Moreover, low PEPD expression in VsW was also associated with lower expression of genes associated with AT insulin sensitivity such as PLIN1, PPARG and GLUT4 (Extended data Table 1). These results suggested that PEPD may regulate AT collagen remodelling and metabolism in AT. The reduction of VsW PEPD expression in obese individuals was tightly coupled with higher PEPD serum levels than lean individuals and associated with higher total cholesterol and aspartate transaminase (ASAT) levels (Cohort 2e, Figure 1e; Extended data Figure 1b). Among obese subjects, those with the higher amount of PEPD released from VsW were more proned to insulin resistance and type 2 diabetes (Figure 1f; Extended data Figure 1c). Also, the ROC curve analysis highlighted the PEPD released from VsW (but not ScW) as a predictor to type 2 diabetes with a cut off for PEPD levels of 1.33µM (cohort 2e,f; Figure 1g). Re-analysis of serum proteomic data from insulin resistant and sensitive obese subjects 6 further con rmed the predictive value of PEPD for insulin resistance and that serum PEPD levels were positively correlated with lipid metabolism (e.g. APOA proteins), in ammation and ECM remodelling proteins (e.g. DPP4, TNC, COL4A2) (Figure 1h, i; Extended data Figure 1d). Altogether, these data from human cohorts provided strong evidence for potential relevance of PEPD as a marker of bro-in ammation and insulin resistance in obese individuals. Similar pattern of low AT PEPD activity coupled with high circulating levels of PEPD was recapitulated in diet-induced obese mice with AT brosis and metabolic disturbances (e.g. high glycaemia, liver steatosis) (Extended data Figure 1e-i). Accordingly, serum levels of PEPD predicted the degree of brosis (peri-Ad collagen) and in ammation (tnfα expression) in GnW, with cut-off values for PEPD of 1.32 µM (Extended data Figure 1j).
To uncouple the effects of PEPD from obesity, we treated chow-fed lean mice with CBZ-Proline (CBZ-Pro), a selective pharmacological inhibitor of PEPD activity 7 . CBZ-Pro-treated mice showed higher dipeptides Gly-Pro serum levels con rming inhibition of PEPD activity and also exhibited higher PEPD serum levels than control mice (Figure 1j Figure 2g). Altogether, these results show that both, downregulation of the intracellular PEPD activity in AT and concomitant increase in the serum levels of PEPD, were necessary and su cient to induce AT broin ammation and insulin resistance. Furthermore, in line with the human GWAS data, dysregulation of PEPD uncoupled bro-in ammation and metabolic dysfunctions from obesity [3][4][5] .
Given this dual role of PEPD, it remained unclear which event -i.e. decreased PEPD activity or increased PEPD released from the cells-was primarily responsible for AT brosis and/or metabolic alterations. We initially performed an Exploratory Factor Analysis (EFA). This is an unbiased statistical method, which establishes the cluster of biological variables with high loadings (correlation equivalent) on each speci c factor in a reduced number of underlying variables ("factors") considered as "superfamilies of variables" 8 .
Using EFA, we investigated how serum PEPD level/PEPD activity co-varied with the metabolic parameters measured in the CBZ-Pro-treated mice and controls (Figure 1p). EFA showed that factor 1, representing a cluster of AT brosis-related variables, co-varied with PEPD serum levels and fasting glucose levels.
Moreover, factor 2, the cluster of "obesity"-related variables (BW, steatosis) also co-varied with PEPD serum levels, whereas factor 3 showed that the activity of GnW PEPD also co-varied with fasting glucose levels. Therefore, EFA indicated that the serum level of PEPD might explain the " brotic" pro le and glycaemic status observed in CBZ-Pro mice.
The link between serum PEPD, brosis and glycaemia was further strengthened when we included the CBZ-Pro mice fed HFD 58% data in the EFA (Factor 1, Extended data Figure 2h). However, the co-variation between PEPD serum level and "obesity"-related variables was not sustained (Extended data Figure 2h).
We rationalised that since HFD per se, dysregulates PEPD activity/secretion, it was likely that HFD-fed control mice might have developed as strong bro-in ammation as in CBZ-Pro-treated mice, masking the effects of this PEPD pharmacological inhibitor. Con rming this interpretation, untreated control and CBZ-Pro-treated mice fed HFD 58% exhibited similar severe AT collagen deposition, fasting insulin, FFA blood levels, glucose and insulin intolerance (Extended data Figure 2i-k).
Thus, to dissect the role of inhibition of PEPD activity from the extracellular action of the released PEPD, we next phenotyped the global pepd heterozygote (HET, +/-) and knock-out (KO, -/-) mice against their wild type (WT, +/+) littermates. PEPD activity and serum levels were not detected in pepd KO. Of interest, Pepd HET mice showed 50% decreased PEPD activity but no difference in PEPD serum levels compared to WTs (Figure 2a, b; Extended data Figure 3a, b) and were morphologically similar to WT. In contrast, pepd KO exhibited a runty phenotype characterised by short length, low body weight and decreased fat mass (Extended data Figure 3c-e) which complicated subsequent phenotyping. Thus, we preferentially used the pepd HET for metabolic analysis. We observed higher collagen accumulation in the ScW (HET/KO) and GnW (KO) compared to WT littermates in chow diet, supporting the role of reduced PEPD activity in promoting brosis ( Figure 2c). This association was supported by correlation matrix analysis (Extended data Figure 3f). Moreover, EFA analysis of pepd KO mice fed chow diet showed that Factor 1, clustering GnW or ScW brosis-related variables, had strong loading of glucose blood levels and negative loading of PEPD activity in AT depots (Figure 2d).
The development of metabolic complications in the pepd HET mice (i.e. insulin resistance, increased FFA levels and liver steatosis) required challenging with HFD. Again, when on HFD no differences in BW and collagen deposition were observed among the genotypes (extended data Figure 3g-j and 4a). Feeding the pepd HET mice with HFD 45%, resulted in higher fed glucose levels and insulin intolerance compared to WT mice, despite maintaining similar fasted glucose, insulin and FFA levels (Extended data Figure 4b-d).
When challenged with HFD 58% 9 promoting greater bro-in ammation in GnW than HFD 45% (Extended data Figure 4e, f), pepd HET mice were more insulin resistant and exhibited a trend to be more glucose intolerant despite still showing signi cantly lower fasting glucose than WT mice (Figure 2e-g). On HFD 58%, pepd HET mice showed signi cantly higher FFA levels compared to WT mice, had AT insulin resistance according to the AT-insulin resistance index (AT-IR index) 10 (Figure 2g, Extended data Figure  4g), and exhibited more severe liver steatosis than WT (Extended data Figure 4h). Collectively, these results indicate that reduced PEPD activity in pepd HET mice was enough to develop AT brosis, in the absence of obesity, and that upon challenge with HFD -which downregulates PEPD activity, increases its release and promotes brosis -further exacerbated metabolic complications in HET mice. Moreover, EFA revealed that in pepd HET mice fed chow+HFD (45% and 58%), their insulin resistance and liver steatosis could be determined by BW and GnW brosis, since these variables clustered and co-varied together (Factor 2, Extended data Figure 5a, b).
To unmask the molecular mechanisms underlying the brogenic and metabolic pathogenic effectors driven by the downregulation of pepd, we performed RNA-Sequencing of the GnW of pepd HET and KO mice fed chow and HFD 45 %. We selected the milder nutritional challenge to prevent the confounding bro-in ammatory effect of 58% HFD by down-regulating pepd in WT mice (Extended data Figure 5c). Analysis of DEGs and the top pathways differentially regulated in GnW pepd HET and KO mice fed 45 %HFD vs. WT revealed that both pepd HET and KO mice had higher expression of genes involved in the regulation of actin cytoskeleton and cell cycle, immune system, in ammatory-related pathways, and ECM/ECM organisation-related proteins (Figure 2h, Extended data Figure 5d; Supplementary Information, Supplementary Table S2-S5). Also, pepd HET mice fed HFD 45% showed lower expression of genes involved in metabolic pathways including fatty acids, leptin and insulin signalling. The pepd KO mice also showed higher expression of genes in the cluster of pro-diabetes-related pathway. Validation of these data using additional in vitro and ex-vivo experiments con rmed that defects in PEPD activity impaired adipogenesis from GnW progenitors, lipolysis in mature adipocytes and leptin secretion from GnW tissue explants (Extended data Figure 5e Table S6-S8). In response to HFD 45% and compared with chow diet, GnW of the pepd HET mice exhibited higher expression of pro-brotic, ECM components, ECM remodelling genes, and in ammatory and immune cells genes, whereas it had lower expression of AT metabolism-related genes. Transcriptomic analyses con rmed the exacerbated bro-in ammation and AT dysfunction in both the KO and HET mice when challenged with HFD 45%. This bro-in ammatory response to HFD was more exacerbated in HET mice than in WT, indicating that pepd HET mice were prone to bro-in ammation in response to HFD. These results also indicated that M and immune cell-derived factors might contribute actively to the metabolic phenotypes associated with the genetic ablation of pepd, pointing to a pathogenic link between PEPD dysregulation and immunity regulation. Cellular fractionation of GnW from lean mice revealed that M (CD11b+ fraction) reported the highest levels of pepd expression compared to that of mature adipocytes (AD), other immune (CD45+), endothelial cells (CD31+) or stroma-vascular fractions (SVF) (Figure 3a). In agreement with the results from total AT, the expression of pepd was lower in AT M from 16 weeks-old ob/ob mice and also from mice fed 20 weeks with HFD 45% compared with their controls (genetically lean and chow fed mice, respectively) (Extended data Figure 6a, b). Moreover, PEPD activity increased during M differentiation using a model of bone-marrow-derived M (BMDMs) (Extended data Figure 6c).
These ndings were validated in humans by showing PEPD enrichment in the immune cell fraction (CD45 + ) of human AT (Extended data Figure 6d). Also, proteomic analysis of human induced pluripotent stem cell-derived M (iPSDM), con rmed the relatively high abundance of PEPD in M compared to undifferentiated iPSCs (Extended data Figure 6e). Given that obesity is associated with an imbalance between classically activated M (considered as in ammatory) and alternatively activated M (considered as non-in ammatory), we assessed the modulation of pepd expression by speci c M polarising agents 11 . In agreement with the reduction of pepd expression in the in amed AT of obese mice, pepd expression was also lower in pro-in ammatory M(LPS) cells, whereas it was higher in M(GC), and not modulated in M(IL-4) treated cells compared to unstimulated M (Extended data Figure 6f). Additional supportive evidence of human transferability was provided by RNA-Seq comparing undifferentiated human induced pluripotent stem cells (iPSCs) and differentiated 12 and analysis of the effect of LPS on human macrophage-derived monocytes (MDM) and iPSDM. In agreement with murine models, stem cell transcriptomic analysis con rmed that i) PEPD expression increases during human M differentiation; ii) iPSDM and MDM display similar PEPD expression levels, and that iii) PEPD expression decreases in response to LPS treatment in both iPSDM and MDM(Extended data Figure 6g).
We then validated at the cellular level that BMDMs treated with CBZ-Pro or LPS, both known to decrease PEPD activity (Extended data Figure 6h), resulted in higher levels of PEPD released to the medium compared to non-activated or alternatively activated Mφ (Figure 3b; Extended data Figure 6i). These results strengthen the functional coupling between the lower enzymatic activity of PEPD in AT and higher release of PEPD to the extracellular compartment.
Based on these previous results, we investigated whether the released PEPD was a determinant of M polarisation. Treatment of M with puri ed PEPD protein-induced phosphorylation of NF-κB (ser536) and expression of in ammatory markers, such as cox2 and il-1β, while non-in ammatory markers mgl1 and mrc2 were downregulated ( Figure 3c; Extended data Figure 7a-d). Phospho-kinase proteome analysis using a pro ler array on PEPD-treated BMDMs con rmed that PEPD phosphorylated EGFR (Extended data Figure 7e) 13 . Pre-treatment of M with Erlotinib, an EGFR-speci c tyrosine kinase inhibitor, attenuated PEPD-induced NFkB phosphorylation, as well as cox2 and il1β expression (Figure 3d; Extended data Figure 7f, g). We rationalised that besides its effects on M , secreted PEPD could also signal to other adipose tissue cells. We characterised the effect of extracellular PEPD on differentiated and non-differentiated pre-adipocytes, as critical pro-brotic cellular effectors in the AT 14,15 . Supporting a direct PEPD pro-in ammatory role, puri ed PEPD increased il-6 expression in differentiated preadipocytes, promoted the production of collagen I and prevented lipid accumulation in pre-adipocytes in part through a mechanism involving EGFR signalling ( To dissect the pivotal role of PEPD secreted from M driving bro-in ammation and metabolic disturbances, we performed a bone marrow transplant (BMT) from pepd KO into WT recipient mice to ablate pepd exclusively in hematopoietic cells (HCs) (Figure 3f; Extended data Figure 8a). BMT KO mice showed 85% reduced prolidase activity in peritoneal M and 50% reduction of prolidase activity in the whole AT (Figure 3g, h). Reduction in PEPD activity by genetic manipulation was not associated with increased PEPD levels in serum, opposite to what we observed in CBZ-Pro-treated mice or dietary models of obesity (Extended data Figure 8b).
As previously observed in models with decreased PEPD activity, the BMT KO model also had increased collagen accumulation in AT compared to BMT WT ( Figure 3i). However, BMT KO mouse maintained carbohydrate metabolic homeostasis in contrast to the chow-fed CBZ-Pro-treated mice, in which the decreased PEPD activity was inversely associated to its secretion (Extended data Figure 8c-h). Furthermore, when fed 58% HFD, the BMT KO mice exhibited more signi cant brosis than BMT WTs but were resistant to obesity and associated bro-in ammation and metabolic disturbances (i.e. glucose and insulin tolerance, and liver steatosis) (Extended data Figure 8e-l). Thus, the BMT KO model exhibited increased AT brosis uncoupled from insulin resistance and liver steatosis. These results showed that brosis and metabolic complications could be uncoupled and pointed to the relevance of the extracellular PEPD produced by M as a critical trigger of bro-in ammation and metabolic complications. Moreover, EFA among BMT mice fed chow, further con rmed the negative association between AT brosis and PEPD activity (Factor 1, Figure 3m). Also, factor 3 showed that AT insulin resistance-related parameters (i.e. AT-IR and FFA levels) co-varied with PEPD serum levels ( Figure 3m). Thus, these results supported that extracellular PEPD was the main trigger for AT related metabolic dysfunctions.
Finally, we performed a global integrative EFA including the three PEPD in vivo experimental models (pharmacological, genetic and macrophage-speci c genetic ablation) to dissect the main pathogenic factors linking speci c PEPD dysregulation and metabolic and bro-in ammatory differences. EFA, including chow-fed and chow+HFD fed animals, showed that Factor 1, clustering AT brosis-related variables had negative loading of AT PEPD activity and body weight (Figure 4a These ndings demonstrate for the rst time that dysregulation of PEPD in obesity determines AT broin ammation and exacerbates obesity-associated metabolic comorbidities through two independent but coupled mechanisms: i) PEPD enzymatic activity, promoting the last step in the degradation of collagen, is inhibited in M from obese AT leading to brosis and metabolic disturbances, i.e. liver steatosis and insulin resistance; and ii) secreted PEPD from M , acting as a non-canonical EGFR ligand promoting AT insulin resistance and bro-in ammation in an autocrine and paracrine manner.
This study emphasises the pathogenic importance of defects in collagen degradation in obesityassociated complications. It provides a strong rationale for the measurement of serum PEPD in obese individuals to identify and stratify those at a higher metabolic risk -by recognising their susceptibility to adipose tissue brosis and in ammation. Taking advantage of murine in vivo models modelling dysregulated PEPD activity and secretion combined with EFA integrative analysis, our results strengthen the conclusion that dysregulation of PEPD elicits a dual role mediating AT brosis and metabolic risk through complementary mechanisms that open new avenues for both diagnostic and therapeutic approaches aiming at uncoupling obesity from its associated metabolic complications.

Human Studies
Age, BMI and glycaemic status of the different cohorts can be found in Figure S1a.

Cohort 1 and 3
In cohort 1, a group of 154 [84 visceral (VsW) and 70 subcutaneous (ScW) adipose tissues] from participants with a wide range of adiposity (BMI between 20 and 68 kg/m2), were analysed. In cohort 3, 46 VAT and 36 SAT samples from morbidly obese subjects (BMI > 35 kg/m2), were analysed. Altogether these subjects were recruited at the Endocrinology Service of the Hospital of Girona "Dr Josep Trueta". All subjects were of Caucasian origin and reported that their body weight had been stable for at least three months before the study. Subjects were studied in the post-absorptive state. They had no systemic disease other than obesity and all were free of any infections in the previous month before the study. Liver diseases (speci cally tumoral disease and HCV infection) and thyroid dysfunction were speci cally excluded by biochemical work-up. All subjects gave written informed consent, validated and approved by the ethical committee of the Hospital of Girona "Dr Josep Trueta", after the purpose of the study was explained to them. Samples included in this study were partially provided by the FATBANK platform promoted by the CIBEROBN and coordinated by the IDIBGI Biobank (Biobanc IDIBGI, B.0000872), integrated in the Spanish National Biobanks Network and they were processed following standard operating procedures with the appropriate approval of the Ethics, External Scienti c and FATBANK Internal Scienti c Committees. AT samples were obtained from SAT and VAT depots during elective surgical procedures (cholecystectomy, surgery of abdominal hernia and gastric by-pass surgery).
Samples of AT were immediately transported to the laboratory (5-10 min). The handling of tissue was carried out under strictly aseptic conditions. AT samples were washed in PBS, cut off with forceps and scalpel into small pieces (100 mg), and immediately ash-frozen in liquid nitrogen before stored at -80ºC. Serum glucose concentrations were measured in duplicate by the glucose oxidase method using a Beckman glucose analyser II (Beckman Instruments, Brea, CA, United States). Intra-assay and inter-assay coe cients of variation were less than 4% for all these tests. Total serum triglycerides were measured by an enzymatic, colorimetric method with glycerol phosphate oxidase and peroxidase (Cobas TRIGL) using a Roche Hitachi Cobas c 711 instrument. approximatively daily dose of 60mg/kg of CBZ-Proline, as previously used in other murine studies 7 , treatment lasted for 6 weeks, control mice were offered regular pellets. CBZ-Proline does not alter viability or promote toxic effects in mice 7 . In our experimental conditions, CBZ-Pro-treated mice did not show evidence of hepatotoxicity/liver damage (alanine (ALAT) or aspartate (ASAT) transaminase levels) (Extended data Figure 2a).

Body Composition
Fat and lean masses were calculated by time-domain nuclear magnetic resonance (TD-NMR) by using a minispec Live Mice Analyzer LF50 (Bruker).

Glucose and Insulin Tolerance Tests
For glucose tolerance test, mice were fasted for overnight with free access to drinking water. Glucose was administered intraperitoneally (2 g/kg), and blood glucose levels were monitored from the tip of the tail with a glucometer. For insulin tolerance tests, insulin was administered intraperitoneally (0.75mU/g), and blood glucose was measured at various times after injection.

Serum Biochemistry
Triglycerides were measured on the Dimension RXL analyzer (Siemens Healthcare). Free fatty acids were measured using the Roche Free Fatty Acid Kit (half-micro test) (kit code 11383175001). Insulin was measured using electrochemical luminescence immunoassay on the MesoScale Discovery immunoassay platform.

Explants for conditioned medium
Approximately 100 mg of freshly dissected GnW cut into ne pieces from 30-week-old mice in chow and HFD conditions were incubated for 6h hour at 37oC in 5% CO2 in DMEM with 5% heat inactivated FBS, 20 mM HEPES, 100 units/mL penicillin, 100 μg/mL streptomycin, and 20 mM L-glutamine) (1 mL media per 100 mg of tissue).

Bone marrow derived M preparation and treatments
Femur and tibia bones from 10-16 weeks-old C57BL6 mice or pepd WT, HET and KO mice were isolated and cleaned, and 10 mL of RPMI-1640 was ushed through each bone using a syringe. Total bonemarrow cells were passed into a 100 μm cell strainer and counted using Countess II automated cell counter (Thermo sher). Cells were spun (400g, 5 min.), resuspended in BMDM culture medium (RPMI1640 supplemented with 20% of L929-conditioned cell medium, 10% heat-inactivated foetal bovine serum (HI-FBS), and 1% penicillin and streptomycin). Total bone-marrow cells were seeded in 10 cm nonculture treated plates (Falcon) at a density of 5x10 6 cells per plate per 10 ml of Mφ differentiation medium and cultured for 7 days at 37 °C in 5% CO2. On day 5 of differentiation, medium was removed and replace with 10 ml of fresh BMDM culture medium. On day 7, BMDMs were detached using ice-cold PBS-EDTA 1mM, spun (400xg, 5 min.) and resuspended in fresh BMDM culture medium. Differentiated BMDMs were counted using Countess II automated cell counter and cell concentration adjusted to 5x105 cells/ml. Immediately after, cells were plated for experiments at the following densities: 100μl/well of 96-well plate, 500 μl/well of 24-well plate, 1 ml/well of 12-well plate, 2 ml/well of 6-well plate and 10 ml per 10 cm plate. Cells were incubated for 16-24 h after plating before conducting experiments.
Mφ purity of the culture was routinely tested by the expression of CD11b and F4/80 by ow cytometry.

93-97% of the cells express high levels of CD11b and F4/80 after 7 days of differentiation.
To make L929-conditioned medium, L929 cells (CCL-1, ATCC) were seeded in DMEM supplemented with 10% heat-inactivated FBS, 100 U/ml penicillin-streptomycin and 2 mM L-glutamine (Sigma) at a density of 250,000 cells per 50 ml of medium per T175 tissue culture ask. Medium was harvested after 1 week of culture, and then 50 mL of fresh DMEM supplemented with 10% heat-inactivated FBS, 100 U/ml penicillin-streptomycin and 2 mM L-glutamine was added onto cells and harvested 1 week later. Batches obtained after the rst and second weeks of culture were mixed at a 1:1 ratio, aliquoted and stored at -20°C .
Erlotinib (5µM) was added to the culture medium 2h before and during the treatment with PEPD. We validated PEPD speci c effect by measuring cox2 expression in BMDMs, reported as a PEPD target gene 13 . In addition, we discarded the potential cytotoxic effects of puri ed PEPD on BMDMs and its potential endotoxin contamination by boiling the puri ed protein and tested on BMDMs for cox2 induction (Extended data Figure 7a Cell Culture and adipocyte differentiation Primary adipocytes isolated from GnW of 10 weeks-old pepd WT,HET, KO and 3T3L1 Cells were differentiated into adipocytes (day 9) accordingly to the protocol described by Roberts et al. 20 .

Prolidase activity
Prolidase activity was determined optimizing Myara's spectrophotometric procedure which was modi ed from the Chinard technique 21,22

Histological Analysis
AT and liver samples were xed in 4% paraformaldehyde for 24h, embedded in para n, sectioned into 5 μm sections, and processed for Sirius ( brosis) or haematoxylin and eosin (H&E) (liver steatosis) staining. The slides were scanned (Microscopy Zeiss Axioscan Z1 Slidescanner) and processed for brosis (Sirius staining excluding vessels) and steatosis (Vacuole % area) quanti cation using HALO™ Image Analysis Software.

Hydroxyproline Assay
Hydroxyproline measurement was done using a hydroxyproline colorimetric assay (BioVision) as previously described 23 . Brie y, frozen fat is weighted and heated in 6 N HCl at 110°C overnight in sealed tubes, as 10 μL of HCl/mg of WAT. Ten microliters are evaporated before incubation with chloramine-T and p-di-methyl-amino-benzaldehyde (DMAB) at 60°C for 90 min. The absorbance was read at 560 nm and the concentration was determined using the standard curve created with hydroxyproline.

ELISA assays
Murine and human PEPD protein concentration were measured using respectively an ELISA kit for Mouse Xaa-Pro dipeptidase (PEPD) ELISA kit (CSB-EL017784MO, CUSABIO) and Human PEPD (Peptidase D) ELISA Kit (E-EL-H5575.96, Elabscience) in AT explant (from which debris was removed by centrifugation) and serum according to the manufacturer's instructions. A standard curve was prepared according to the manufacturer's instructions, and the value associated with an unconditioned media blank was subtracted from that of conditioned media.

RNA Extraction and Real-Time PCR
RNA from cells extracted using RNeasy Mini columns (Qiagen) according to the manufacturer's instructions. RNA was harvested from frozen tissue using RNA-STAT-60TM (AMS Bio), and puri ed by chloroform extraction and isopropanol precipitation. Reverse transcription was performed using Reverse Transcriptase System (Promega) according to manufacturer's instructions. Real-time PCR was carried out using TaqMan or Sybr Green reagents using an Abi 7900 real-time PCR machine using default thermal cycler conditions. Primer sequences are described in Table S9. Reactions were run in duplicate checked for reproducibility, and then averaged. A standard curve generated from a pool of all cDNA samples was used for quanti cation. The expression of genes of interest was normalized using the geometric average of four housekeeping genes (18s, 36b4, βactin, and B2m), and data is expressed as arbitrary units.
Regarding human samples (Cohorts 1, 2, 3), RNA puri cation, gene expression procedures and analyses were performed, as previously described 24,25 . Brie y, Total RNA was extracted and puri ed using RNeasy Lipid Tissue Mini kit and integrity was checked by Agilent Bioanalyzer. Total RNA was quanti ed by means of a spectrophotometer. The same amount of total RNA was reverse transcribed to cDNA from all samples using High Capacity cDNA Archive kit following manufacturers' instructions. Gene expression was assessed by real-time PCR using a LightCycler 480 real-time PCR system, using TaqMan and SYBRgreen technology suitable for relative genetic expression quanti cation. The commercially available and pre-validated TaqMan primer/probe sets used are described in Supplementary Information, supplementary Table 9 and 10).

RNA sequencing
Library preparation and sequencing Total RNA of GnW was extracted using the miRNeasy mini kit (Quiagen) according to manufacturer's instructions. Per experiment, 4-10 independent biological repeats were used.
Total RNA was quality checked (RIN >7) via the Agilent Bioanalyser 2100 system, using the Agilent RNA  Table 1) were mapped to the most recent ENSEMBLE version GRCm38.p5 of the mouse reference genome sequence (GRCm38.p5) using STAR v2.5.1b 26 including the annotations as hints for exon-intron borders. Reads were considered as mapped, if the similarity was at least 95% over at least 90% of the read length as previously described 27 .
FeatureCounts v1.5 28 was applied for the generation of count tables based on the mapping les.
Customized python scripts 27 were deployed for downstream processing including the normalization of the raw counts to the total number of assigned reads per gene (TPMs) and to the combined exon length (FPKMs), respectively.

Gene expression and pathway enrichment analyses
Raw counts were subjected to differential gene expression analysis via DESeq2 29 Table 9). Bound primary antibodies were detected using peroxidase-coupled secondary antibodies and enhanced chemiluminescence (Millipore). Blots were exposed digitally using the ChemiDoc MP System (Bio-Rad), and bands were quanti ed using Image J software. The expression of proteins was normalized to protein levels of a housekeeping protein (β-actin or Tubulin), and data is expressed as arbitrary units.
Array based detection of phosphorylated receptor tyrosine kinase The Proteome Pro ler Mouse Phospho-RTK Array Kit (R&D Systems, USA, Catalog Number: ARY014) was employed to screen for the level of phosphorylation of receptor tyrosine kinase (RTKs) in BMDMs in response to puri ed PEPD, according to the manufacturer's instructions. Brie y, the array membranes were incubated 1h with an array blocking buffer prior incubation over night at 4C on an orbital shaker with 1.5 ml of cellular extract. The membrane were then washed and incubated 30 min. with streptavidin-HRP solution. Membranes were exposed digitally using the ChemiDoc MP System (Bio-Rad), and spots were quanti ed using the Image J software. One condition corresponds to a pool of cellular extracts from 4 independent experiments. All the arrays were measured three times; each spots were normalized to the positives controls. The results are presented in a heat map and considered relevant when the fold variation was >1.3 or <0.6.

Cytotoxicity assays
To determine the cytotoxic effect of compounds on M , cells were seeded in Roswell Park Memorial Institute (RPMI) 1640 Medium without FBS supplemented at a density of 15,000 cells per well in wells of a 96 well plate. Cells were treated with the given compounds at the given concentrations for 24 hours, and cytotoxicity was measured using an LDH-Cytotoxicity Calorimetric Assay Kit (BioVision) according to the manufacturer's instructions.

Data Analysis
All data from experiments is summarised by its mean, with error bars showing standard error of the mean. The number of replicates is reported in the gure legends. When the results of a pairwise comparison is expressed as a fold-change it is declared in the gure legends to what value the data was normalised. Statistical analysis was performed using Prism7 and Prism8 (GraphPad). Comparisons between two groups were conducted using an unpaired t-test. Comparison between more than two groups were conducted using a one-Way ANOVA followed by appropriate post hoc multiple comparisons tests. Comparisons between more than two groups and factors were conducted using a two-way ANOVA followed by appropriate post hoc testing. Multiple comparisons were corrected and the resulting p-values were adjusted (q-value) relying on the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli. Pearson's coe cients were evaluated to estimate the extent of correlation between series of data. Data points were excluded ifrom the correlation analysis f they exhibited a value of more than two SDs from the mean. For all metabolic tests, the model animals were randomly assigned to the different experimental settings (GTT, ITT). Areas under the receiver operating characteristic (ROC) curves were determined for each variable to identify the predictors of AT bro-in ammation and insulin resistance/type 2 diabetes. ROC curves is a plot of sensitivity (true positive) versus 1-speci city (false positive) showing the ability of biomarker (PEPD level) to discriminate between true positives (e.g. insulin resistant) and true negatives (e.g. insulin sensitive). The best marker has an ROC curve shifted to the left with area under the curve close to unity 33 . To determine the optimal cut-off values for bro-in ammatory status or insulin resistance indices, the Youden index was calculated (sensitivity + speci city−1), and the values for the maximum of the Youden index was considered as the optimal cut-off points using the Webtool easyROC 34 . Statistical signi cance was set at *p <0.05, **p<0.01 and ***p<0.001.

Exploratory Factor Analysis
Exploratory factor analysis (EFA) was conducted to determine the possible latent structure of the variables (listed, e.g., in Figure 1.p -rows of the depicted matrix) measured in each animal models, i. We want also to acknowledge the FATBANK platform promoted by the CIBEROBN and the IDIBGI Biobank (Biobanc IDIBGI, B.0000872), integrated in the Spanish National Biobanks Network, for their collaboration and coordination.
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Please note that the views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.    High PEPD serum levels is associated with AT insulin resistance and drives the differences between the pharmacologic and genetic animal models of PEPD down-regulation. a, b Heat map representing the four factors extracted through EFA performed among mice fed chow. The columns report the factors loadings of the observed variables. Mice from the three animal models (i.e. CBZ-Pro, PEPD and BMT) were plotted according to factor 1 and 2 (b). c, d. Heat map representing the four factors extracted through EFA