Subcutaneous Adipose Tissue Cell-derived Cytokines Are Early Markers of Impaired Glucose Tolerance in Severe Obesity

92 severely obese (SO) clinical Following an Oral Glucose Tolerance Test, participants were included in three groups, based on the presence of normal glucose tolerance (NGT), impaired glucose tolerance (IGT) or Type 2 Diabetes (T2D). Serum and subcutaneous AT (SAT) biopsies were obtained and Mesenchymal Stem Cells (MSCs) were isolated, characterized and differentiated in adipocytes in vitro. TNFA and PPARG mRNA levels were determined by qRT-PCR. Circulating, adipocyte- and MSC-released cytokines, chemokines and growth factors were assessed by multiplex ELISA.


Introduction
Type 2 diabetes (T2D) is a large health and socio-economic burden worldwide. In 2019, about 463 million people were estimated to be living with diabetes and, approximately 90% of those were affected by T2D. This number is expected to increase to 578 million in 2030 and 700 million in 2045 1 . The global epidemics of T2D is largely driven by the dramatic increase of obesity and severe obesity observed during the past 35 years 2 . In particular, the prevalence of obesity with Body Mass Index (BMI) ≥ 40 kg/m 2 has quadrupled, and that of obesity with BMI ≥ 50 kg/m 2 increased vefold 3 . Recently, it has been shown a 40% prevalence of T2D in severely obese (SO) population 4 .
Adipose tissue (AT) has a central role in the connection between obesity and T2D. For a long time, AT has been envisioned as a site of energy storage. To date, AT is considered an active endocrine organ, capable of regulating systemic energy and metabolic homeostasis through a complex network of endocrine, paracrine, and autocrine signals. Within AT, mesenchymal stem cells (MSCs) are located in perivascular niches (stromal vascular fraction -SVF) and participate to cell turnover and to the vascular network for AT tropism 5,6 .
Subcutaneous AT (SAT) is by far the largest adipose depot and very relevant for the quantitative contribution to lipid storage/release and endocrine function 7 . Obesity drives the pathological expansion of AT. Unhealthy AT harbours enlarged hypertrophic adipocytes and displays a state of low-grade chronic in ammation. The formation of new vasculature to support AT growth is also impaired, and enhanced brosis and hypoxia are evident. Such events contribute to the development of insulin resistance 8 .
Dysfunctional AT secretes pro-in ammatory cytokines that antagonize insulin action, promoting local and systemic in ammation 7,8 . For instance, tumor necrosis factor-α (TNF-α) plays a well-known role in the connection between obesity, in ammation and insulin resistance 9 .
Thus, although excess weight is an established risk factor for T2D, yet a large part of obese individuals do not develop T2D and the precise mechanisms linking the two conditions remain unclear 10,11 .
Moreover, there are still few studies focused on the progression toward T2D in SO individuals 4 .
To clarify whether AT-related low-grade in ammation may contribute to metabolic derangement in severe obesity, we enrolled a cohort of SO individuals undergoing bariatric surgery, with or without impairment of glucose tolerance. We found that SAT adipocytes and their precursor MSCs derived from SO individuals with abnormal glucose tolerance, displayed increased production of in ammatory molecules compared to those with normal glucose tolerance (NGT). Deranged production of cytokines by MSCs was at least in part reverted by the avonoid quercetin.

Material And Methods
Patient enrollment and tissue collection.
Abdominal human SAT biopsies and serum samples were obtained from N = 92 subjects undergoing laparoscopic gastric banding from October 2013 to April 2015. Exclusion criteria were: metabolic and endocrine diseases, gastrointestinal in ammatory diseases, previous or current malignancies. All patients enrolled underwent detailed clinical phenotyping, including measurement of height, weight, waist, blood pressure, complete clinical chemistry, in ammatory and hormone determination (thyroid, adrenal and sexual). Moreover, patients were subjected to an Oral Glucose Tolerance Test (OGTT) with the administration of 75 g of glucose and blood drawn at 0, 30, 60, 90 and 120 min for plasma glucose and insulin determination. IGT and T2D were diagnosed based on American Diabetes Association criteria 12 . Informed consent was obtained from every subject before the surgical procedure. Protocols were approved by the ethical committee of the University of Naples (prot. n. 225_2013).
Cell isolation and differentiation.
Mesenchymal stem cells (MSCs) were isolated from SAT biopsies as previously described 13 . Brie y, SAT specimens were chopped into small pieces and were enzymatically digested by collagenase solution RNA isolation and Quantitative RT-PCR.

Statistical Analysis.
Statistical analyses were performed with R statistical platform and with GraphPad Prism 6.0 software (GraphPad Software Inc., La Jolla, CA). Results are presented as mean ± standard deviation (SD) for continuous variables if normally distributed and as median and range if not. For comparisons between 2 groups, a two-tailed t-test for independent samples (for normally distributed data) or a Mann-Whitney U test (for not normally distributed data) was used. Multiple comparisons among the three groups were made using the ANOVA test with Tukey correction (for normally distributed data) or the Kruskal-Wallis test followed by the Dunn test (for not normally distributed data). p value of < 0.05 were considered statistically signi cant.

Results
Anthropometric and clinical characteristics of the study population Severely obese (SO) individuals enrolled in the study (N=92) were subjected to OGTT and grouped as NGT (N= 44), IGT (N= 25) and T2D (N= 23). NGT, IGT and T2D groups did not display signi cant differences for mean age, BMI, abdominal circumference, arm circumference, systolic and diastolic blood pressure (Table 1). HOMA-IR was signi cantly higher in T2D patients compared to IGT and NGT (Table 1).
No differences in serum levels of the gastro-intestinal hormones Ghrelin, GIP, GLP-1 and Glucagon were found among the 3 groups (Suppl. Table 1). Finally, NGT, IGT and T2D groups had similar concentrations of pituitary, thyroid and sexual hormones (data not shown).

Systemic in ammation in NGT, IGT and T2D subjects
Serum samples of the 3 groups of patients were screened for the concentration of C-reactive Protein (CRP), Fibrinogen and Ferritin, as classical markers of systemic in ammation and no signi cant differences were identi ed among the 3 groups of SO subjects (Suppl. Table 2). However, the screening of a panel of cytokines, chemokines and growth factors revealed that T2D patients displayed an increase of IL-9, IL-13 and MIP-1β compared to IGT (1.2, 1.4, 1.5-fold, respectively) and of IL-13 compared to NGT (1.5-fold). Thus, higher levels of speci c cytokines characterize the serum of T2D SO subjects. Moreover, no signi cant difference was found between IGT and NGT groups, beside a slight reduction of IL-9 levels in SO individuals with IGT (Table 2).
Adipose tissue in ammation in NGT, IGT and T2D SO subjects In order to assess whether subcutaneous adipose tissue (SAT) may contribute to the worsening of glucose tolerance in SO individuals, SAT biopsies from a subset of patients were collected and analyzed for quanti cation of speci c mRNAs. As shown in g. 1A, no differences in mRNA levels of the canonical isoforms of adipogenic master gene PPARG were found. However, SAT from IGT and T2D patients displayed a 3-fold increased mRNA levels of the in ammatory marker TNF-α, compared with NGT ( g. 1B).
Next, SAT-derived mesenchymal stem cells (MSCs) were isolated and characterized, as described in Materials and Methods and in Supplementary gure 1. MSCs from the 3 groups displayed the same ability to differentiate in adipocytes (Suppl. g. 2). However, adipocytes obtained from IGT SO patients released higher levels of TNF-α (pval= 0.01), IL-1β (pval=0.04) and IP-10 (pval=0.08) compared to NGT SO patients ( g. 2). TNF-α, IL-1β and IP-10 secretion was signi cantly higher also in adipocytes from T2D compared with those of NGT subjects ( g. 3). No detectable differences between T2D and IGT were identi ed ( g. 2).
Next, in order to assess whether the in ammatory phenotype of adipocytes was already displayed by their mesenchymal precursor cells, TNF-α levels were evaluated in MSCs obtained from the 3 groups. Interestingly, signi cantly increased TNF-α mRNA and protein levels were observed in MSCs from IGT, compared to NGT patients, with no signi cant differences between IGT and T2D ( g. 3). Thus, SAT precursor MSCs from SO subjects with abnormal glucose tolerance are primed to develop an in ammatory phenotype.

Quercetin reduces MSC in ammation in SO subjects
Finally, we investigated whether the avonoid quercetin was able to counteract MSC in ammation in IGT and T2D SO subjects. Quercetin treatment did not impair MSC viability (data not shown). However, quercetin exposure reduced TNFA mRNA levels by about 3-fold in MSCs from IGT subjects ( g. 4A). In addition, it signi cantly reduced TNF-α protein secretion in MSCs both from IGT and from T2D subjects ( g. 4B). Interestingly, in presence of quercetin a signi cant decrease of the in ammatory cytokines IL-4, IL-6, IL-7, INF-γ and MCP-1 in IGT and of IL-1β, IL-4, IL-6, IL-7, INF-γ, MCP-1 and FGFb in T2D-MSCs was also observed ( g. 4B). Quercetin did not modify the release of IL2, IL-17, CCL5, MIP-1α, MIP-1β and IP-10 both in IGT and T2D derived MSCs (data not shown). Notably, quercetin treatment induced a signi cant increase of PDGF and G-CSF in MSCs of IGT and T2D SO subjects ( g. 4B). Thus, quercetin is able to modify in ammatory signals in SAT precursor cells.

Discussion
Obesity is a clear risk factor for metabolic, cardiovascular and oncological diseases 2, 5 . In particular, the prevalence of T2D increases with increasing BMI. Hence, SO subjects have a higher risk than moderately obese individuals 19 . However, there are 7-28.3% SO subjects, who do not develop T2D 20 . Most of the studies investigated the metabolic abnormalities associated with the more common moderate obesity, while much fewer data is available for morbid obese patients 4 .
Here, we have shown, for the rst time, that in SO subjects: i) circulating IL-9, IL-13 and MIP-1β are associated to T2D; ii) in the progression of glucose tolerance impairment, in ammation is detectable in SAT earlier than in serum; iii) SAT-derived MSCs from IGT are already educated to secrete more TNF-α; iiii) quercetin is able to control MSC in ammation.
Obesity and T2D are characterized by systemic low-grade in ammation with higher levels of both proin ammatory and anti-in ammatory cytokines 10,[21][22][23] . However, con icting data and discrepancies about the composition of the cytokinome exist. First, we have observed that SO subjects affected by T2D displayed higher levels of serum IL-9, IL-13 and MIP-1β, as compared to IGT.
IL-9 is a pleiotropic in ammatory cytokine also implicated in pancreatic β-cell destruction in T1D 24,25 . MIP-1β, also known as chemokine CC motif ligand 4 -CCL4, is a potent macrophage attractant, which acts through the binding to CCR5, a transmembrane receptor involved in infection, in ammation and cancer 26 . At variance, IL-13 is an anti-in ammatory cytokine that promotes polarization of macrophages toward a M2 phenotype: it contributes to epicardial adipose tissue remodeling and increases glucose uptake and metabolism in skeletal muscle 27,28 . Previous studies have detected increased levels of IL-9, IL-13 and MIP-1β in serum of obese individuals compared with lean controls [21][22][23] . Other studies in T2D have also reported an increase of MIP-1β, a reduction or no change of IL-13 and no difference of IL-9 compared to NGT subjects 21,23,26,27,29 . However, at our knowledge, no data are available for T2D and IGT morbid obese individuals. In this population, the increase of IL-9 and MIP-1β is consistent with their possible involvement in β-cell damage and macrophage recruitment, respectively, and contribute to the progression from IGT toward T2D. On the other side, IL-13 increase may be due to a compensatory mechanism. Indeed, all T2D individuals enrolled in the study were either newly diagnosed or with a short duration of disease, also for their relatively young age. Moreover, no signi cant differences were detected in the systemic levels of cytokines of SO individuals with IGT, as compared to those with NGT.
AT has a central role in the association between obesity and T2D and is now considered a prominent target for new pharmacological approaches, including adipokine-based therapeutics 8 . Excess glucose, lipids and other nutrients lead to an impaired AT with altered plasticity and imbalanced production of adipokines and pro-in ammatory factors, which sustain a local low-grade in ammation that is a crucial determinant for T2D and obesity complications 5,11,30 . We have provided evidence that SAT and SATderived isolated cells of SO IGT and T2D patients expressed signi cantly higher levels of TNF-α compared to NGT patients, with no detectable differences between T2D and IGT. TNF-α is a well-known in ammatory cytokine, with a pivotal role in the onset and progression of insulin resistance 9 . TNF-α levels are increased in AT and plasma of obese individuals, and its expression is positively related to BMI with no signi cant relationship between fat cell size and mRNA levels 9,[31][32][33] . In addition to TNF-α, we have found that adipocytes from IGT SO people released higher levels of IL-1β and IP-10, other key cytokines in obesity and in etiology of T2D 34 . TNFα and IL-1β work together to both initiate and propagate the in ammatory process, also inducing the expression of multiple cytokines, among which IP-10 35,36 . IP-10, also known as CXCL10, has been shown to play a deleterious role in obesity as potential inhibitor of AT angiogenesis 37 . Thus, our data suggest that TNF-α, IL-1β and IP-10 are not only adipocytokines associated to obesity, but also molecules with a crucial role in the early phases of impairment of glucose metabolism in SO subjects.
SAT from obese subjects contains a dysfunctional pool of MSCs with impaired adipogenesis 38,39 . We have previously shown that a truncated isoform of PPAR (PPARΔ5), which acts as a dominant-negative isoform by reducing PPARγ activity, positively correlates -whereas canonical PPARG does not -with BMI in overweight or obese and T2D patients, possibly contributing to impaired adipogenesis 40 . Furthermore, an increased ratio between PPARΔ5 and canonical isoforms has been recently reported in a in vitro generated model of hypertrophic-like adipocytes, also displaying increasing IL-6 secretion, whereas only a modest reduction of canonical PPARG expression was observed 41 . Here, we detected similar levels of canonical PPARG in SAT biopsies of SO individuals with different glucose tolerance conditions, as well as similar levels of lipid accumulation in differentiated adipocytes. Thus, most likely, the difference in cytokine release does not relate to a different adipocyte differentiation program.
Notably, we found that MSCs from IGT subjects are already primed to express and secrete more TNFα, compared to MSCs from NGT subjects. High glucose exposure of MSCs leads to impaired cell functions, upregulation of in ammatory genes (IL-1 β, CCL5, IL-8, MCP-1) and alteration in DNA methylation 38,42,43 . Epigenetic modi cations represent a common mechanism through which both genetic and environmental exposures impact on the susceptibility to T2D 44 . It has been shown that distinct methylations of the promoter regulate TNF-α expression 45,46 . It could be hypothesized that impaired glucose levels may imprint speci c epigenetic signature on MSCs, upregulating TNF-α production, which in turn contributes to impaired glucose tolerance. However, further studies are necessary to elucidate the molecular mechanisms underlining increased TNF-α levels in IGT-derived MSCs.
To date, a special attention for both prophylactic and therapeutic interventions in T2D is directed to diet nutrients, able to regulate the metabolism and in uence the consumer's health 47 . Quercetin is a nutraceutical compound with anti-in ammatory and antioxidant effects both in vivo and in vitro. Possible mechanisms of actions include inhibition of COX-2, NF-kB and MAPK pathways, post-translational modi cations and elimination of senescent cells [47][48][49] . Here, we have shown that quercetin ameliorates the local low-grade in ammation by reducing the expression of TNF-α and the release of a number of in ammatory cytokines in MSCs isolated from IGT and T2D morbid obese subjects. Moreover, it increases the release of G-CSF, a cytokine able to counteract the activation of major in ammatory cytokines, including TNF-α, whose anti-obesity effects have been shown in animal models with diabetes and obesity 50 .

Conclusions
Speci c patterns of circulating and AT-released cytokines may have a crucial role in the worsening of glucose tolerance allowing the identi cation of subjects with unhealthy obesity that could develop T2D. Of course, many other factors may contribute to the progression from normal glucose tolerance to impaired glucose tolerance in SO subjects. For instance, even though abdominal SAT contributes more than visceral AT (VAT) to the pro-in ammatory milieu associated with SO 51 , VAT-derived cytokines, whose release correlates with epigenetic modi cations in SO-VAT 44 , may also play a relevant role, and thus require further investigations. and compared between groups using the non parametric Kruskall Wallis test followed by Mann Whitney U test for pairwise comparisons.