Fat-to-blood recirculation of partially dysfunctional PD-1+CD4 Tconv cells is associated with dysglycemia in human obesity

Summary Obesity is characterized by the accumulation of T cells in insulin-sensitive tissues, including the visceral adipose tissue (VAT), that can interfere with the insulin signaling pathway eventually leading to insulin resistance (IR) and type 2 diabetes. Here, we found that PD-1+CD4 conventional T (Tconv) cells, endowed with a transcriptomic and functional profile of partially dysfunctional cells, are diminished in VAT of obese patients with dysglycemia (OB-Dys), without a concomitant increase in apoptosis. These cells showed enhanced capacity to recirculate into the bloodstream and had a non-restricted TCRβ repertoire divergent from that of normoglycemic obese and lean individuals. PD-1+CD4 Tconv were reduced in the circulation of OB-Dys, exhibited an altered migration potential, and were detected in the liver of patients with non-alcoholic steatohepatitis. The findings suggest a potential role for partially dysfunctional PD-1+CD4 Tconv cells as inter-organ mediators of IR in obese patients with dysglycemic.


INTRODUCTION
Insulin resistance (IR) is a pathological condition associated with obesity characterized by a reduced responsiveness of insulin-sensitive tissues -such as visceral adipose tissue (VAT), liver, and skeletal muscles-to the effects of insulin. 1,2The excessive accumulation of lipids in adipocytes can trigger the tissue recruitment of immune cells that differentiate into proinflammatory cytokine-producing cells and disrupt the insulin signaling pathway. 3The inability of pancreatic b-cells to compensate for the loss of insulin sensitivity results in dysglycemia, which progresses to the development of type 2 diabetes (T2D). 4n preclinical models of obesity, the dynamics of immune cells in VAT were studied.CD8 T cells have been described to precede the recruitment of macrophages into the VAT and contribute to IR by recognizing stress markers on the surface of adipocytes. 5,68][9] In addition, resident CD4 T cells shift from a T helper 2 (Th2) to a T helper 1 (Th1) phenotype and contribute to IR in both animal models of obesity 10,11 and humans. 124][15][16] Notably, CD4 T cells are resistant to cytokine-mediated suppression 13 and have a restricted T cell receptor (TCR) repertoire. 10,17Collectively, these findings suggest that chronic stimulation by unknown cognate antigens may be responsible for the phenotypic and functional specialization of CD4 T cells in obese VAT, which eventually leads to the development of dysglycemia.
Currently, there is limited understanding of tissue-specific characteristics of VAT-derived CD4 T cells and their role in causing dysglycemia in human obesity.In this study, we have identified a subset of PD-1-expressing CD4 conventional T cells (CD4 + CD25 À FoxP3 -; Tconv) which are decreased in the VAT of obese patients with concomitant dysglycemia (OB-Dys) compared to those without dysglycemia (OB-ND) and display a transcriptomic and functional profile compatible with partially dysfunctional cells.Our findings indicate that PD-1 + CD4 Tconv cells from OB-Dys are non-clonally expanding cells that circulate from VAT to peripheral blood and are enriched in the liver of patients with dysglycemic with non-alcoholic steatohepatitis (NASH).

Visceral adipose tissue-derived PD-1+CD4 Tconv cells from obese patients with dysglycemia display transcriptomic and functional characteristics of partially dysfunctional cells
To comprehensively characterize PD-1 + CD4 Tconv cells and discern pathways potentially implicated in VAT-specific glucose dysmetabolism, we determined the transcriptomic profile of VAT-derived PD-1 + CD4 Tconv cells from OB-ND and OB-Dys.We focused our analysis on the PD-1 + compartment guided by our transcriptomic findings that revealed a diminished relevance of the PD-1 -cell subset in the context of dysglycemia.This was substantiated by the relatively low number of differentially expressed genes (DEGs) in OB-Dys compared to OB-ND (n = 33) and the absence of relevant functional enrichments, as observed through both EnrichR and GSEA platforms (data not shown).PD-1 + CD4 Tconv from patients with OB-Dys showed 58 genes upregulated and 19 downregulated compared to OB-ND (Figures 2A and 2B).The complete list of DEGs is provided in Table S1.Relevant DEGs include molecules involved in immune responses, such as IRAK2, TNFRSF4, IL1RN, and PTGDR (Figure 2A).PD-1 + CD4 Tconv cells from OB-Dys upregulated genes related to cytokine responses (Figure 2C), while downregulating pathways associated with glucocorticoid signaling and with the inhibition of fat differentiation (Figure 2D).Furthermore, PD-1 + CD4 Tconv cells from OB-Dys displayed a proinflammatory phenotype as indicated by the enrichment of gene signatures of ''inflammatory response'' (Figure 2E) and ''TNFA signaling via NFKB'' (Figure 2F).Consistent with this observation, the signature of effector cells was found enriched in PD-1 + CD4 Tconv cells from OB-Dys (Figure 2G) as well as an increased frequency of activated PD-1 + HLA-DR + CD4 Tconv cells was observed in OB-Dys VAT (Figure 2H). 19otably, the signature of tumor-infiltrating dysfunctional T cells was also enriched in VAT-derived PD-1 + CD4 Tconv cells from OB-Dys (Figure 2I).PD-1+CD4 Tconv cells from OB-Dys exhibited the expression of T cell exhaustion markers, including CTLA-4 and LAG-3, which were absent in patients with OB-ND (Figure 2J).Nonetheless, other exhaustion-associated inhibitory receptors and transcription factors such as TCF7, BATF, and TOX displayed comparable expression levels in both patient groups (Figure 2J).Additionally, an elevated frequency of PD-1 + TIGIT + and PD-1 + LAG-3 + co-expressing CD4 Tconv cells was observed in OB-Dys (Figure 2K, left and middle panels), while the frequency of CD4 Tconv cells expressing multiple inhibitory receptors (i.e., PD-1+TIGIT+LAG-3+, Figure 2K, right panel) remained similar between the two groups.Markers of senescence remained unaltered in the presence of dysglycemia, as demonstrated by comparable expression levels of genes associated with senescence (Figure S4A) 21 as well as an equivalent frequency of CD57+PD-1+CD4 Tconv cells between OB-Dys and OB-ND (Figure S4B) Notably, no difference in the frequency of IFN-ɣ, TNF-a, and IL-17-producing PD-1 + CD4 Tconv cells between VATs could be detected following PMA/Ionomycin stimulation (Figure 2L), indicating that the engagement of the TNFA/effector signaling pathway by VAT-derived PD-1 + cells does not result in increased release of proinflammatory soluble molecules.Two-hundred twenty-seven genes were found to be positively correlated with HOMA-IR levels in PD-1 + CD4 Tconv cells.These genes were enriched in pathways associated with inflammatory responses, including TNF-ɑ and Interferon-ɑ responses (Figure S5A; Table S2).Among those genes, TNFRSF4 and HMOX1, which are DEGs between PD-1 + CD4 Tconv cells of OB-ND and OB-Dys, have been previously described to be mechanistically linked to IR 22,23 (Figure S5B).However, protein content of TNFRSF4 and HMOX1 was similar between OB-ND and OB-Dys in PD-1 + CD4 Tconv cells (Figure S5C).
These data indicate that VAT-derived PD-1 + CD4 Tconv cells from OB-Dys are heterogeneous and display a transcriptional and functional profile compatible with partially dysfunctional cells.

Visceral adipose tissue-derived CD4 Tconv cells are a heterogeneous population, with enrichment in virus-specific cells within the public clonotypes
As PD-1 can mark antigen-specific T cells [24][25][26] and clonality of VAT-derived T cells has been described in preclinical models of obesity, 17,27 we wondered whether PD-1 + CD4 Tconv cells undergo clonal expansion toward specific antigens in the VAT of OB-Dys.To ascertain the extent of clonality under physiological conditions, for these experiments we included a cohort of n = 4 non-obese non-diabetic donors undergoing living kidney donation named lean controls (LC).Sequencing of the T cell receptor b-chain complementarity determining region 3 (TCRb CDR3) of PD-1 + and PD-1 -CD4 Tconv cells from VAT of OB-ND, OB-Dys, and LC revealed a similar distribution of the TCR-V b gene families suggesting a superimposable composition of TCRb repertoires among the 3 groups (Figure S6).This was confirmed by the comparable richness of the TCRb repertoire among the three groups as indicated by the Shannon's diversity index (Figure S7).These data are further supported by the similar frequency of CD137 + PD-1 + CD4 Tconv, which identify recently antigen-activated T cells, 28,29 in patients with OB-ND and OB-Dys (Figure S8).
In the attempt to infer the antigen specificity of clonotypes identified in VAT-derived PD-1 + CD4 Tconv from OB-Dys, we investigated whether public clones contributed to their TCRb repertoire.At first, we searched in our dataset for the presence of TCRs with known antigen specificity reported in the VDJdb database.Results showed that 0.4% amino acid sequences of PD-1 + CD4 Tconv cells from 4/7 OB-Dys, 3/6 OB-ND, and 2/4 LC matched with public clonotypes responding to viral, pathogen, and wheat antigens annotated in VDJdb (Figure 3A), with specificities toward Influenza A and T. Aestivum.To increase the chance to find public clonotypes in VDJdb, we exploited the GLIPH2 algorithm that clusters together TCR sequences with similar motifs (called patterns) which are predicted to bind the same antigens [32].Amino acid sequences that did not meet GLIPH2 criteria (see STAR methods section) were excluded from the analysis.Twenty-six percent of PD-1 + CD4 Tconv patterns identified in the VAT were public (Figure 3B), with specificities toward Influenza A, T. aestivum and CMV.A total of 10.81% of patterns showed specificities for multiple antigens, including Influenza A, CMV, HIV-1, and the tumor antigen NY-ESO1.The distribution of public clonotypes (Figure 3C) and public pattern-associated sequences (Figure 3D) among the three groups is shown.Similar matches of TCRb clones could be observed in VATderived PD-1 -cells (Figure S9A), with 24.4% of PD-1 -TCRb patterns being public and specific for Influenza A, T. aestivum, CMV, RSV and HIV-1 (Figure S9B).Notably, the VAT of obese patients showed a trend of enrichment in CD4 Tconv public clonotypes compared to LC, regardless of PD-1 expression or diabetes status (Figures 3C, 3D, and S9C).
These data indicate that VAT-derived CD4 Tconv cells exhibit a predominant virus-specificity among public clonotypes, irrespective of PD-1 expression or diabetes status.

The T cell receptor b-chain repertoire of visceral adipose tissue-derived CD4 Tconv cells is selectively altered in obese patients with dysglycemia
Although VAT-derived PD-1 + CD4 Tconv cells demonstrate a polyclonal constitution in the context of obesity, we sought to elucidate whether the presence of dysglycemia engenders a distinctive profile in the TCRb repertoire.We specifically focused on clones that exhibit presence in at least two individuals, effectively excluding private clonotypes from the analysis.Notably, discernible patterns emerge as TCRb clones identified within the OB-Dys subgroup exhibit a proclivity to be intra-patient common (Fisher's exact test, OB-Dys vs. LC and OB-Dys vs. OB-ND, p value <0.0001), while their convergence with the lean control (LC) (OR = 0.141) and OB-ND (OR = 0.0007) cohorts is infrequent (Figures 3E and S10).This observation underscores the marked divergence in the TCRb repertoire of VAT-derived PD-1 + CD4 Tconv cells as dysglycemia becomes a feature.In contrast, TCRb clones tend to be casually distributed between the LC and OB-ND groups (Fisher's exact test, p value 0.4).TCRb patterns identified by the GLIPH2 algorithm differed significantly between OB-ND and OB-Dys, while no significant difference was observed in LC vs. OB-ND and LC vs. OB-Dys (Figure 3F).Similar trends were also observed in the PD-1 -compartment (Figure S10).
Overall, these data suggest that an altered composition of the TCRb repertoire of VAT-derived CD4 Tconv cells can be observed in OB-Dys regardless of PD-1 expression.

PD-1 + CD4 Tconv cells from obese patients with dysglycemia show increased recirculation of clones with a non-restricted T cell receptor b-chain repertoire from visceral adipose tissue to PB
To explain why a shortage of partially exhausted PD-1 + CD4 Tconv was observed in the VAT of OB-Dys, we tested the hypothesis that this cell subset develops the ability to recirculate from VAT to PB when dysglycemia occurs.A clonal tracking analysis was performed to determine the frequency of clones shared between PB and VAT within the same individuals referred to as ''recirculating clones.''The median percentage of recirculating clones was higher in the VAT of OB-Dys compared to OB-ND and LC in PD-1 + , but not PD-1 -, CD4 Tconv (Figure 4A).This observation was corroborated by two key lines of evidence: firstly, the TCRb patterns of PD-1 + CD4 Tconv cells from patients with OB-Dys exhibited the most pronounced rate of recirculation between PB and VAT, an observation not mirrored by their PD-1 -counterparts (Figure 4B); secondly, it was evident that recirculating PD-1 + CD4 Tconv clones from OB-Dys were expanded in the PB (Figure 4C).Notably, recirculating clones from both the PD-1 + and PD-1 -compartments were found to be less expanded in obese VATs compared to LC (Figure 4D), suggestive of a non-clonal TCR repertoire of recirculating clones in obesity regardless of diabetes status.To discern the trajectory (VAT to PB or PB to VAT) of clone recirculation, we ascertained whether the clones were preferentially expanded within the VAT or PB compartments.The results revealed a distinctive pattern: while 97% and 77% of recirculating PD-1 + CD4 Tconv cells from LC and OB-ND, respectively, were expanded in the VAT, the majority (58%) of recirculating clones from OB-Dys exhibited expansion within the PB compartment (Figures 4E and 4F).This underscores a shift in the directionality of recirculating clones within the context of OB-Dys, deviating from the conventional PB to VAT trajectory observed in physiological/normoglycemic conditions.A similar inversion of clones' expansion could be observed in the PD-1 -counterpart (Figure S11A), with 96% and 72% clones expanded in VAT of LC and OB-ND, respectively, and 53% expanded in PB of OB-Dys (Figure S11B).Further evidence of increased PD-1 + CD4 Tconv cell recirculation in OB-Dys was provided by the enrichment of the ''leukocyte migration'' signature in OB-Dys compared to OB-ND (Figure 4G).Of note, the expression of S1PR4, which regulates T cell migration, 30 could be found among upregulated DEGs in PD-1 + cells from OB-Dys compared to OB-ND (Figures 2A and 2B) and positively correlated with HOMA-IR levels (Table S2).
All together these data indicate that dysglycemia in human obesity is associated with increased recirculation of PD-1 + CD4 Tconv cells with unrestricted TCRb repertoire from VAT to PB.The reduced frequency of PD-1 + CD4 Tconv cells in VAT of OB-Dys was expected to be associated with their accumulation in PB.However, the frequency of PD-1 + CD4 Tconv cells was found to decrease in PB of OB-Dys compared to OB-ND (Figure 5A) and not imputable to apoptosis (Figure S12).The transcriptomic analysis of PB-derived PD-1 + CD4 Tconv cells showed differential expression of 40 genes (18 upregulated and e 22 downregulated) between OB-Dys and OB-ND (Figure 5B), with the deregulation of the ''integrin family cell surface interaction'' pathway (NCI Nature 2016).An altered integrin expression profile was evident in PD-1 + CD4 Tconv cells from OB-Dys, which included higher expression levels of ITGAX -enriched in T cells with high migratory potential 31 -and ITGAV -associated with cancer progression 32 -(Figure 5C), and lower expression of ITGA1, tissue-resident memory T cell marker 33 (data not shown).Furthermore, a trend of increased protein expression of the chemokine receptor CCR7, regulator of T cell trafficking and recruitment into inflamed tissues 34,35 -was observed in OB-Dys compared to OB-ND (p = 0.08) (Figure 5D), while the proportion of CCR7+ cells, i.e., naive and central memory PD-1+ CD4 Tconv cells, remain unchanged (Figure S13).Notably, cells co-expressing PD-1 and CD4 were found to be enriched in the liver of patients with NASH and concomitant dysglycemia compared to patients with normoglycemic NASH (Figure 5E).Collectively, these findings underscore the linkage between dysglycemia and a diminished prevalence of PB-derived PD-1 + CD4 Tconv cells characterized by augmented migratory potential.Furthermore, these results concomitantly highlight the accumulation of PD-1 + CD4 + cell populations within the hepatic milieu of patients with dysglycemic NASH.

DISCUSSION
14,16 This study, by evaluating CD4 Tconv phenotype, RNA transcriptome, and receptor clonality in the VAT and PB of patients with and without dysglycemia, attempts to elucidate the link between adaptive immunity and T2D.
While Th1 cells for a long time have been imputed as drivers of tissue inflammation in obesity, more recent studies have found that CD4 T cells from obese VAT are functionally exhausted, with PD-1 + cells endowed with senescent characteristics. 14,16We hereby demonstrate that PD-1 + CD4 Tconv cells are reduced in the VAT of obese patients with dysglycemia and recirculate into the bloodstream.We found that PD-1-expressing CD4 Tconv cells in the VAT of OB-Dys are a heterogeneous population with transcriptomic and functional characteristics compatible with partially dysfunctional cells.T cell exhaustion is a differentiation state occurring in the presence of persistent chronic TCR stimulation with dysfunctional cells expressing arrays of inhibitory molecules, distinctive patterns of transcription factors, and impaired effector molecules.Recent studies reported a gradient of exhaustion, 36 with dysfunctional T cells being a heterogeneous population that is part of a wide differentiation spectrum, spanning from transitional, through early dysfunctional, toward highly dysfunctional T cells. 20,37,38Dysfunctional cells exhibit a spectrum that encompasses the retention of certain effector functionalities, 39,40 manifestation as early-stage or partially exhausted states, 20 which concurrently manifests an intermediate level of the TOX transcription factor. 41Within the VAT microenvironment of OB-Dys, PD-1 + CD4 Tconv cells: (i) manifest a transcriptomic profile that bridges effector and dysfunctional phenotypes, (ii) following stimulation, do not show increased proinflammatory cytokine synthesis, (iii) demonstrate co-expression of TIGIT or LAG-3, but not multiple inhibitory receptors, and (iv) exhibit an intermediate expression level of TOX.Collectively, this evidence implies the likelihood of these cells being in the earlier stages of partial dysfunction.Such a profile could correspond to cells that are less sensitive to modulation or immunosuppression as previously shown by our group on bulk CD4 Tconv cells from OB-Dys. 13Single cell resolution of the PD-1 + CD4 Tconv cell transcriptome will help elucidate the heterogeneity of this cell subset.However, one could hypothesize that localized soluble factors and cell-to-cell interactions ultimately intercede with proinflammatory/effector signaling pathways, thereby fostering the development of a subset of PD-1 + CD4 Tconv cells with a partially exhausted phenotype.Also, it cannot be excluded that the PD-1 signaling is not fully functional in this context, thus resulting in the freezing of PD-1 + CD4 Tconv cells in a transitional state of differentiation.The pivotal question revolves around whether this partial impairment of PD-1 + CD4 Tconv cells in OB-Dys is beneficial or impedes the resolution of chronic low-grade inflammation in adipose tissue.Our hypothesis posits that PD-1 + CD4 Tconv cells may undergo partial exhaustion in the context of dysglycemia, acting as a physiological feedback mechanism to attenuate metabolic inflammation sustained by macrophages.However, it is plausible that unresolved T cell stimulation through the TCR could result in a lack of inhibitory signals to macrophages, potentially perpetuating a sustained inflammatory response.
3][44][45][46] This evidence was confirmed by our results showing that public clonotypes, predominantly specific for viral antigens, were observed in obese VATs regardless of diabetes status.Notably, the VAT-derived TCRb repertoire of CD4 Tconv was divergent in OB-Dys compared to the other groups.This evidence is supported by the outcomes of hierarchical clustering analysis arising from unsupervised flow cytometry data, which delineates the preferential clustering of obese VATs encompassing prediabetic and type 2 diabetic subgroups over their OB-ND counterparts.The confluence of these observations collectively suggests that dysglycemia elicits substantive modifications within the CD4 Tconv cell compartment.This study is not designed to elucidate if changes in CD4 Tconv cells determine dysglycemia or rather dysglycemia induces CD4 Tconv cell skewing; however, considering that the natural history of T2D consists in the progression from a normoglycemic state to a transient phase of prediabetes preceding the development of full-blown disease, 47 the homogeneity of CD4 Tconv cells infiltrating the VAT in patients with dysglycemia suggests that changes of CD4 Tconv cells may occur long before the development of T2D.Furthermore, it is unlikely that the mild hyperglycemia occurring in prediabetes could explain the phenotypic and transcriptomic changes, as well as alterations of the TCR repertoire, observed in VAT-derived CD4 Tconv cells.One could speculate that gradual changes of PD-1 + CD4 Tconv cells may occur at the target site of inflammation of obesity in response to metabolic and environmental stressors which induce a phenotypic shift toward an early dysfunctional state resulting in the engagement of a transcriptional programming of cell migration.
The phenotypic profile of PD-1 + CD4 Tconv cells in the VAT of OB-Dys was expected to be driven by self-antigens and associated with clonal expansion, as previously suggested in different settings. 17,20,27While TCR clonality in PD-1 + CD4 Tconv cells within the VAT was not observed, the presence of dysglycemia yielded distinctive outcomes: (i) an exclusive TCR repertoire configuration, (ii) increased adipose-to-circulation recirculation, and (iii) a diminished frequency yet heightened migratory potential within the PB.Furthermore, recirculating clones of PD-1 + CD4 Tconv cells showed lower expansion in obese VAT regardless of diabetes status, suggesting that they likely undergo bystander cytokine-mediated activation.While it was not expected that the recirculation of 2% of cells from VAT to PB would lead to a clear-cut increase in PD-1 + CD4 Tconv cells in the PB of OB-Dys, the observed decrease was equally unexpected.This diminished frequency of PD-1 + CD4 Tconv cells in the PB of OB-Dys could not be explained by increased apoptosis.However, the accumulation of this cell subset in patients with NASH with dysglycemia supports the hypothesis that the bloodstream might not inherently constitute the ultimate reservoir for this cell subset and that they may further recirculate in the liver.However, additional analyses are required to substantiate this claim.Our working hypothesis underscores the conceivable role of PD-1 + CD4 Tconv cells in effecting an inter-organ propagation of a proinflammatory milieu, thereby contributing to insulin resistance across various insulin-sensitive tissues (Figure 5F).
Reported findings indicate that dysglycemia, characterized by elevated levels of IR in human obesity, is associated with the recirculation of partially dysfunctional polyclonal PD-1 + CD4 Tconv from VAT to PB.This study has uncovered several novel findings regarding adipose tissue biology in the context of obesity and T2D, providing the rationale for proposing T cell recirculation as the underlying mechanism of human IR.Signals released by metabolic tissues -such as peptide hormones, cytokines, exosomes, and lipids -are described to mediate inter-organ communication in the control of systemic metabolism. 48However, T cells have all the required features to be vehicles of metabolic inflammation as they can (i) be carriers of a proinflammatory state induced by a variety of triggers, (ii) be individually (chemo)attracted toward specific tissues, (iii) easily adapt to local environments.Here, for the first time, we describe the link between T cell recirculation and whole-body metabolism.Further investigation is needed to determine if recirculation is the mechanism underlying the development of dysglycemia or vice versa and whether altered glucose metabolism induces T cell recirculation.This newly acquired knowledge provides the rationale for further testing pharmacological compounds halting T cell recirculation, 49,50 in the prevention or treatment of T2D.

Limitations of the study
This study posits PD-1 + CD4 Tconv cells as potential mediators of insulin resistance within insulin-sensitive tissues, substantiated by evidence from a diverse range of patient cohorts.Limitations encompass the inability to chronicle events during the transition from euglycemia to dysglycemia and uncertainty regarding the direct causation of dysglycemia by these cells in type 2 diabetes.Furthermore, our observations indicate a reduction of PD-1 + CD4 Tconv cells in VAT of obese patients with dysglycemic and an augmentation in the liver of individuals with NASH and type 2 diabetes.Validation of these data necessitates single-cell tracking analyses in both human subjects and murine models to elucidate the causal association between PD-1 + CD4 Tconv cells and dysglycemia.
program under the Marie Sklodowska-Curie grant agreement n. 704779.Furthermore, we are grateful to Filippo Cortesi who introduced the team to unsupervised flow cytometry analysis technique. of acute or chronic infections, concomitant autoimmune or chronic inflammatory diseases, including malignancies, and severe cardiovascular disease.Obese patients were stratified in two groups based on their diabetes status, i.e., patients with normoglycemia (OB-ND) and patients with dysglycemia (OB-Dys), according to the American Diabetes Association (ADA) guidelines. 51OB-Dys group included both patients with type 2 diabetes (OB-D) and with prediabetes (OB-PreD).Patients with fasting glucose <100 mg/dL and/or hemoglobin A1c (HbA1c) < 5.7% were classified as OB-ND.Patients with fasting glucose R126 mg/dL and/or hemoglobin A1c (HbA1c) R6.5% or undergoing treatment with antidiabetic medications were classified as OB-D.Patients with fasting glucose R100 and %125 mg/dL or with HbA1c 5.7-6.4% were classified as OB-PreD.Diabetes status was assessed based on blood tests available in the last 4 months before surgery.Healthy lean controls (LC) were enrolled for some experiments.They had a BMI %25 kg/m 2 and were selected from the waiting list for living kidney donation at OSR. LC were free from diabetes, concomitant infections, malignancies, autoimmune and chronic inflammatory diseases.Clinical data, including BMI, fat mass, lipid profile, renal function, fasting glycemia, fasting insulin, HbA1c and blood tests were obtained from all patients prior to surgery.The Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) was calculated using the formula: fasting glucose (mg/dL)*fasting insulin (mU/L)/405, as illustrated in Matthews et al. 52 The figure legend provides information on the quantity of subjects associated with each experiment.Liver biopsies from n = 12 patients with histological evidence of NASH were studied by the OSR Pathology Unit, 7/12 individuals were dysglycemic, while 5/12 were normoglycemic.The Bedossa score was available for 10 out of 12 patients and was 4 in 6/12, 6 in 2/12 and 7 in 2/12.BMI, available for 5/12 subjects, was in the overweight range.The study was approved by the OSR Ethics Committee.All the study participants signed the informed consent (protocols DRI006-FAT001 and DRI-FAT003).

METHOD DETAILS Sample collection and processing
Visceral adipose tissue (VAT) was collected from OB-ND, OB-Dys and LC during surgery.In particular, omental fat was collected from obese patients and perirenal fat from LC. Blood was withdrawn before general anesthetic injection.VAT was transferred to the laboratory in Dulbecco's Modified Eagle Medium (DMEM) supplemented with GlutaMAX (ThermoFisher -Scientific), bovine serum albumin (BSA), penicillin/streptomycin (P/S), and HEPES, where it was processed within 90 min from the collection.VAT was finely minced and digested with collagenase IV (Sigma-Aldrich) resuspended in PBS with a final concentration of 2 mg/mL for 40 min at 37 .The sample was then washed with DMEM high glucose (10% FBS, 1% P/S, 1% Glutamine) and treated with Red Blood Lysis buffer (Biolegend) to obtain the stromal vascular fraction (SVF).After washing, the pellet of SVF was filtered and counted.Whole blood (WB) was collected simultaneously with VAT and used for flow cytometry analysis.Peripheral blood mononuclear cells (PBMCs) were obtained using Lympholyte Cell Separation density gradient centrifugation media (Cedarlane).After centrifugation, the PBMC ring was collected and washed two times with PBS.

Flow cytometry analysis
SVF (1310 6 cells) and WB were suspended in X-Vivo medium supplemented with human serum and P/S.Stainings were performed within 24 h from tissue collection as described in Cardellini et al. 13 The list of anti-human monoclonal antibodies, including information on the company catalog number, clone and concentration used for our assays is provided in Table S3.Cells were counted and washed with PBS and stained with Fixable Viability Stain 575V (BD) before antibody incubation.Annexin staining was performed according to manufacturer's instructions (BD) on samples marked with antibody and live dead staining.Apoptotic cells were identified as Annexin positive within the live-dead negative fraction.Data were acquired on a BD LSRFortessa or FACSymphony instruments (BD Biosciences).Fluorescence intensity was standardized using multiple peak Rainbow Calibration Particles (Spherotech).For unsupervised analysis, T cells were extracted, fcs.files were imported and logicle transformed in R environment (v.4.0.3).Then, CD4 T cells were isolated using OpenCyto package (version 2.2.0). 53All markers satisfying the rules for normalization have been subjected to the landmark alignment procedure, using the normalize() function in the flow-Stats R package (version 3.40.1). 54Unsupervised analysis was realized using the FlowSOM algorithm, included in the CyTOF/CATALYST pipeline (version 1.14.1). 55Uniform Manifold Approximation and Projection (UMAP) and visual representations were realized using the CyTOF/ CATALYST pipeline.For the supervised analysis of flow cytometry data, FlowJo V.10 was used.

RNA extraction
Total RNA was obtained from sorted cells using RNeasy Micro Kit (Qiagen), according to manufacturer procedures.RNA integrity and concentration was determined using the 2200 TapeStation instrument (Agilent Technologies).

Figure 3 .
Figure 3.The TCRb repertoire of PD-1 + CD4 Tconv cells in VAT is altered in OB-Dys and is characterized predominantly by virus-specific public clonotypes (A) Pie chart representing the frequency of clones with a match for antigens annotated in the VDJ database.(B) Pie chart representing the frequency of patterns -obtained using the GLIPH2 algorithm -with a match for antigens annotated in VDJdb.(C) Number of clones or (D) pattern-associated sequences with a unique match in VDJdb.In ''multiple antigen specificities" bars indicate the number of patternassociated sequences matching with multiple antigens.(E) Venn diagrams showing the number of clones and (F) patterns shared among LC, OB-ND and OB-Dys individuals in VAT-derived PD-1 + CD4 Tconv cells.Matches with public clones annotated in VDJdb are reported.Exact Fisher's test has been adopted to evaluate the association between shared and unshared clones and patterns for each comparison (LC vs. OB-ND, LC vs. OB-Dys, OB-ND vs. OB-Dys).LC (n = 4), OB-ND (n = 6) and OB-Dys (n = 7).