Pancreatic fat cells of humans with type 2 diabetes display reduced adipogenic and lipolytic activity.

Obesity, especially visceral fat accumulation, increases the risk of type 2 diabetes (T2D). The purpose of this study was to investigate the impact of T2D on the pancreatic fat depot. Pancreatic fat pads from 17 partial pancreatectomized patients (PPP) were collected, pancreatic preadipocytes isolated and in vitro differentiated. Patients were grouped using HbA1c into normal glucose tolerant (NGT), prediabetic (PD) and T2D. Transcriptome profiles of preadipocytes and adipocytes were assessed by RNAseq. Insulin sensitivity was estimated by quantifying AKT phosphorylation on western blots. Lipogenic capacity was assessed with oil red O staining, lipolytic activity via fatty acid release. Secreted factors were measured using ELISA. Comparative transcriptome analysis of preadipocytes and adipocytes indicates defective upregulation of genes governing adipogenesis (NR1H3), lipogenesis (FASN, SCD, ELOVL6, FADS1) and lipolysis (LIPE) during differentiation of cells from T2D-PPP. In addition, the ratio of leptin/adiponectin mRNA was higher in T2D than in NGT-PPP. Preadipocytes and adipocytes of NGT-PPP were more insulin sensitive than T2D-PPP cells in regard to AKT phosphorylation. Triglyceride accumulation was similar in NGT and T2D adipocytes. Despite a high expression of the receptors NPR1 and NPR2 in NGT and T2D adipocytes, lipolysis was stimulated by ANP 1.74-fold in NGT cells only. This stimulation was further increased by the PDE5 inhibitor dipyridamole (3.09-fold). Dipyridamole and forskolin increased lipolysis receptor-independently 1.88-fold and 1.48-fold, respectively, solely in NGT cells. In conclusion, the metabolic status persistently affects differentiation and lipolysis of pancreatic adipocytes. These alterations could aggravate the development of T2D.


INTRODUCTION
Obesity is a major risk factor for the development of type 2 diabetes (T2D) (1). While the role of liver steatosis and nonalcoholic fatty liver disease (NAFLD) in the development of insulin resistance, metabolic syndrome, and T2D is well accepted, the role of excessive lipid accumulation in the pancreas is controversially discussed (2,3). Using noninvasive imaging methods for the quantification of tissue fat content, clinical association studies do not give a uniform picture (2,(4)(5)(6)(7)(8). A previous clinical study of our group suggests that increased pancreatic fat is associated with impaired insulin secretion in patients with prediabetes (PD, comprising "impaired glucose tolerant and/or impaired fasting blood glucose") but not in humans with normal glucose tolerance (NGT) (4). In a following study, a negative association of pancreatic fat content with insulin secretion was also found in nondiabetic humans with a high genetic risk score for insulin resistance and diabetes (9). These observations suggest that pancreatic fat may take part in the progressive deterioration of islet function in a subtype of humans only.
A more detailed, histological analysis of the human pancreatic tissue revealed a substantial infiltration of adipocytes which contributes to the higher percentage of pancreatic fat content (7). Adipocytes store fuel in a large intracellular lipid droplet, but they are particularly secretory cells, which release cytokines, chemokines, and adipokines in addition to metabolites, e.g., fatty acids (10)(11)(12). Consequently, secreted factors of infiltrating adipocytes may exert paracrine effects on neighboring cells. Previously, we described that pancreatic preadipocytes and adipocytes produce and secrete chemokines and cytokines when challenged with palmitate and fetuin-A, a hepatokine increasingly secreted by fatty liver (7). These results could at least partially explain the divergent outcome of human association studies and suggest that stimulation of local adipocytes by factors released from fatty liver alter the adipocyte secretome and consequently the paracrine impact of local fat cells on islet function.
To unravel the specific pancreatic fat cell functions, we characterized preadipocytes and in vitro differentiated adipocytes obtained from resections of partial pancreatectomized patients (PPP). This in vitro approach by using standardized protocols enables a comparative analysis of purified cell populations and their differentiation capacity. The highly variable genetic and environmental background among the patients, together with the variable anatomical location and the heterogeneous composition of the resected tissue, including fibrosis and inflammation, disables a comparative transcriptome analysis. Therefore, preadipocyte populations were prepared from fat pads using differential centrifugation. The absence of immune cells (CD68 positive) and endothelial cells (CD31 positive) was confirmed by fluorescence-activated cell sorting. The preadipocyte cell population retains proliferative capacity, which permits the expansion of the cells (13). This expansion increases the material of a defined cell population obtained from small surgical leftovers. The assessment of the metabolic status of the tissue donors allows a correlation of preadipocyte and adipocyte characteristics with metabolic/diabetic traits.
Here, we present the first analysis of human pancreatic fat cells which suggests that a diabetic environment persistently alters the functional characteristics of adipocytes.

Ethical Approval
The study protocols were approved by the Ethics Commission of the Medical Faculty and the University Hospital of the University of T€ ubingen in accordance with the Declaration of Helsinki for pancreatic (Nos. 697/2011BO1 and 563/2019BO2) and subcutaneous (No. 446/2016BO2) tissue sampling. Written informed consent was given from all patients.

Measurement of Insulin Secretion and Sensitivity
Fasting blood samples were collected from patients before surgery. Insulin secretion and sensitivity were calculated using the HOMA2 method (14). Insulin secretion (HOMA2% B) was computed from fasting glucose and C-peptide levels. Insulin sensitivity (HOMA2%S) was assessed using fasting glucose and specific insulin levels. HOMA2 disposition index, a measure of b-cell function adjusted for insulin sensitivity, was calculated using the following formula: HOMA2%B Â HOMA2%S/10,000. No correlation was found between the parameters of insulin secretion and sex, age, BMI, and visceral adipose tissue (VAT, data not shown).

Abdominal Fat Quantification Using CT
Patients (n = 12) underwent an abdominal computed tomography (CT) examination before surgery. Quantification of total, visceral, and subcutaneous fat area was performed from a single slice at the umbilical level, which has been shown to be a reliable representative slice for total abdominal fat volumes (15). The fat areas were manually segmented by measuring pixels with densities between À30 and À190 Hounsfield units to delineate total, visceral, and subcutaneous compartments and to compute the cross-sectional area of each in cm 2 .

Preadipocyte Isolation and In Vitro Differentiation
Pancreatic and subcutaneous preadipocytes were isolated from fat biopsies as previously described (16,17).

Transcriptome Analysis (RNAseq)
Aliquots of preadipocytes and adipocytes were used for RNA extraction (NucleoSpin RNA, Macherey-Nagel, D€ uren, Germany). RNA integrity (RIN ! 9) was measured with Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). Sequencing was performed as described previously at the Center for Molecular and Cellular Bioengineering (CMCB, Technical University Dresden, Germany). In short, mRNA was isolated from 1 mg RNA by poly-dT enrichment using the NEBNext Poly(A) mRNA Magnetic Isolation Module. After fragmentation, the samples were subjected to the workflow for strand-specific RNAseq library preparation (Ultra Directional RNA Library Prep II, NEB) and 75-bp single-read sequencing was performed on Illumina NextSeq500. After sequencing, FastQC was used to perform quality control. Differential expression between preadipocytes and adipocytes was tested with the R package DESeq2 v2.7.R (18). Selected genes have been published previously (19). The complete data set is available under GEO accession GSE169514.
Gene ontology (GO) enrichment analysis was performed with the Metascape database (http://metascape.org) using a P < 0.01 as a cutoff criterion (20). Fold-change of !2 and À2 and Benjamini-Hochberg adjusted P values of 0.05 were used as criteria to select differentially expressed genes (DEG).

Oil Red O Staining
Mature adipocytes were fixed with 4% paraformaldehyde for 30 min at RT and subsequently incubated with oil red O for further 30 min. Excess of oil red O was washed with PBS and lipid droplets were examined using a light microscope (EVOS M5000, Thermo Fischer Scientific Invitrogen, Karlsruhe, Germany). Oil red O-positive area was analyzed using ImageJ (NIH, Bethesda, MD).

Secretome Analysis
Growth factors, cytokines, and adipokines secreted into the medium were quantified using the Bio-Plex Pro Human Cytokine, Chemokine, and Growth Factor Assay (BioRad Laboratories). Secreted adiponectin was measured using the Bio-Plex Pro Human Diabetes Assay (BioRad Laboratories).

Statistics
Statistical analysis was performed using GraphPad Prism (GraphPad Software Inc, La Jolla, CA). Differences between two groups were assessed by Student's t test. ANOVA with Tukey posttesting was used when more than two groups were compared. In addition, categorical variables were analyzed using JMP (JMP software, SAS Institute Inc., Cary, NC). Person chi-square test was used to show the relationship between two categorical variables. Differences were considered statistically significant at P 0.05.

Diabetogenic Environment Impairs Pancreatic Adipocyte Differentiation
To understand whether chronic hyperglycemia impacts on function and differentiation of pancreatic fat cells, preadipocytes were isolated from fat pads collected from PPP and differentiated in vitro into adipocytes. The patients were stratified into three groups, NGT, PD, and T2D according to HbA1c values (Table 1). HOMA2%B and HOMA2 disposition index were lower in T2D compared with NGT, indicative for impaired insulin secretion in T2D. Comparative transcriptome analysis was performed with preadipocytes and in vitro differentiated adipocytes (Fig. 1). Principal component analysis (PCA) displayed two distinct clusters (PC1) separating the preadipocyte and adipocyte populations. This indicates efficient in vitro differentiation (Fig. 1A). The variation observed among the preadipocytes of different donors (PC2) was less pronounced than the variation among the respective adipocytes. The differentiation evoked an upregulation of 398 and a downregulation of 625 genes in the transcriptome of NGT-preadipocytes (Fig. 1B). Particularly, genes typical for differentiated fat cells were upregulated, such as transcription factors PPARG, CEBPA, CEBPB, and NR1H3, ensuing upregulation of adiponectin mRNA levels (ADIPOQ; Fig. 1, C-E). Unexpectedly, leptin mRNA levels (LEP) were lower after differentiation (Fig. 1D). Furthermore, enzymes involved in triglyceride storage (e.g., FASN, ELOVL6, SCD, PLIN1, and PLIN4), as well as lipolytic enzymes and proteins enabling fatty acid secretion, were upregulated upon differentiation (e.g., PNPLA2, LIPE, and FABP4; Fig. 1, F-I).
In an attempt to detect differences between adipocytes from PPP with NGT, PD, and T2D, the transcriptomes of these three groups were compared (Fig. 2). Despite a substantial overlap of 255 upregulated and 397 downregulated genes (Fig. 2, A and B), several upregulated genes were lower expressed in adipocytes of T2D-than of NGT-PPP among them the induced genes GPD1, PLIN1, and CIDEC and a large variety of genes involved in lipogenesis, lipid storage, lipid uptake, and lipolysis (Fig. 2, C-N). These results indicate T2D-dependent impaired lipid metabolism and reduced differentiation capacity of human pancreatic adipocytes. The mRNA levels of SLC2A4 (GLUT4), an important differentiation marker of adipocytes was low but increased in cells of NGT-PPP upon differentiation (Fig. 2O). In contrast, SLC2A1 (GLUT1) and SREBF1 mRNA levels were high in all cells (Fig.  2, P and Q).
The successful in vitro differentiation of preadipocytes into adipocytes was further supported by the GO enrichment analysis (Fig. 2R). Most of the significantly enriched GO terms are attributed to fatty acid and lipid metabolism ( Fig.  2R; Supplemental Table S3 and Supplemental Fig. S1). The comparison between NGT, PD and T2D samples supports the assumption that T2D negatively interferes with adipocyte differentiation. Thus, GO terms enriched to a lesser extent or not enriched in T2D compared with NGT were linked to triglyceride (TG) synthesis, lipolysis, and adipokine secretion, for example, PPAR signaling pathway, fat cell differentiation, fatty acid metabolism, protein kinase B signaling, and glycerol-3-phosphate metabolic process (Supplemental Table S3). Furthermore, the downregulated DEG were enriched in GO terms related to angiogenesis, extracellular matrix remodeling, cytoskeleton organization, and cell motility [(Supplemental Fig. S2 and Ref. (19)]. In addition, evaluation of transcription factor interaction with upregulated genes inferred by TRRUST (transcriptional regulatory relationships unraveled by sentence-based text-mining) shows that adipocyte-specific factors (CEBPB, PPARG, and SREBF1) are more active in NGT than in T2D (Fig. 2S, marked in red).
In conclusion, in vitro differentiation of preadipocytes into adipocytes is impaired in T2D-PPP compared with NGT-PPP.

Impaired Insulin Signaling in Pancreatic Adipocytes of Human with T2D
Since transcriptome analysis suggested impaired adipogenesis in adipocytes derived from PPP with T2D and insulin is the most prominent factor promoting adipocyte differentiation, we hypothesize that preadipocytes of PPP with T2D are insulin resistant and retained this phenotype in cell culture, following expansion and in vitro differentiation.
To examine insulin sensitivity, phosphorylation of AKT was stimulated with increasing concentrations of insulin (1, 10, and 100 nmol/L; Fig. 3). Stimulation of AKT phosphorylation by the highest supramaximal concentration of insulin (100 nmol/L) was used as reference and set to 100% for the respective cell preparations. Subcutaneous adipocytes of insulin-sensitive humans were used as control. Insulin, at 10 nmol/L, elicited a comparable AKT phosphorylation in pancreatic preadipocytes and adipocytes of NGT-PPP [37 ± 9.3% and 38 ± 8.2% (n = 5), respectively] as well as in subcutaneous preadipocytes and adipocytes [45 ± 12.5% and 54 ± 18.4% (n = 5-6), respectively]. However, AKT phosphorylation induced by 10 nmol/L insulin was significantly lower in preadipocytes and adipocytes of T2D-PPP [19 ± 3.5% and 20 ± 6% (n = 5), respectively; P < 0.01 vs. SC; Fig. 3, A-F]. These data support the results of the expression profiles and show that, also functionally, preadipocytes from T2D-PPP retained the insulin-resistant phenotype even after in vitro expansion and differentiation.

Impaired Lipolysis of Differentiated Pancreatic Adipocytes with T2D
Next, the adipocyte secretome was examined, since putative paracrine effects of adipocytes are exerted via locally secreted factors. Fatty acids, which are components of the adipocyte secretome, are important cofactors of glucoseinduced insulin secretion ensuring proper stimulation of secretion (7). Sympathetic nerve activation and glucagon, especially during fasting and exercise, stimulate fatty acid secretion from adipocytes. Therefore, we examined the lipolytic activity of differentiated pancreatic adipocytes in response to isoproterenol, a specific b-adrenergic receptors agonist, and glucagon. TG accumulation was visible in pancreatic adipocytes after differentiation (Fig. 4A). Since lipid accumulation of pancreatic adipocytes was much lower than that of subcutaneous adipocytes, pancreatic adipocytes were cultured in MUFA-enriched medium which increased cellular lipid accumulation to levels detected in subcutaneous adipocytes (Fig. 4, A and B). No differences in lipid accumulation between pancreatic adipocytes from PPP with NGT and T2D were observed.
Basal lipolysis of pancreatic adipocytes was similar among the NGT and T2D groups (80.14 ± 17.78 nmol/L Â mg protein -1 Â h À1 and 80.70 ± 51.82 nmol/L Â mg protein -1 Â h À1 (n = 5), respectively; n.s.; t test). Lipolysis was stimulated neither by isoproterenol nor by glucagon (Fig. 4C). In adipocytes from PPP with NGT only forskolin, 5 mmol/L, which activates adenylyl cyclase by bypassing receptors, increased lipolysis 1.48-fold (Fig. 4C). Addition of the phosphodiesterase inhibitor IBMX had no further stimulatory effect. The lipolytic rate of pancreatic adipocytes was not improved when cells were cultured in the presence of MUFAs (Supplemental Fig. S3). In adipocytes of PPP with T2D, none of the stimuli elicited a lipolytic response (Fig. 4C and Supplemental Fig. S3). In contrast, fatty acid release of subcutaneous adipocytes was stimulated 12.1-fold by forskolin [basal lipolysis 39.73 ± 23.29 nmol/L Â mg protein À1 Â h À1 (n = 5)], significantly augmented by isoproterenol (2.7-fold) and inhibited by insulin (Fig. 4D). In pancreatic fat cells, mRNA levels of the glucagon receptor (GCGR) and b-adrenergic receptors (ADRB1-3) were low in comparison to the high amount of insulin receptor (INSR) mRNA (Fig. 4, E and F). This low expression of the respective receptor could explain the absence of detectable effects of glucagon and isoproterenol on lipolysis.
In contrast to the adrenoceptors, pancreatic adipocytes contained high mRNA levels of the natriuretic peptide receptors 1 and 2 (NPR1 and NPR2; Fig. 4G). Furthermore, expression of PDE5A, a selective cGMP-hydrolyzing molecule, was much higher than PDE3B, which preferentially hydrolyzes cAMP (Fig. 4H). Receptor and phosphodiesterase expression profiles could suggest that the cGMP pathway is more important in eliciting lipolysis than the cAMP pathway. Indeed, exposure of NGT adipocytes to ANP stimulated lipolysis 1.74-fold, inhibition of PDE5 by dipyridamole 1.88-fold, and both together increased lipolysis 3.09-fold (Fig. 4I). Insulin, an activator of PDE3B, did not counteract the stimulatory effect of ANP on lipolysis. In T2D-PPP adipocytes, lipolysis was not stimulated by ANP. These results suggest that in isolated in vitro differentiated human pancreatic adipocytes lipolysis is preferentially stimulated by ANP and persistently impaired in T2D.

Secretome of Pancreatic Preadipocytes and Adipocytes
Additional components of the adipocyte secretome, which may interfere with insulin secretion, were evaluated via the transcriptome and via measurement of the respective product released into the supernatant. The mRNA levels have been presented previously (19). Of note, differentiation was accompanied by a 13-fold induction of ADIPOQ mRNA levels [ Fig. 1D; Ref. (19)]. In contrast, LEP mRNA levels were already high in preadipocytes and declined during differentiation [ Fig. 1D; Ref. (19)]. The ratio of leptin/adiponectin mRNA correlated positively with HbA1c (Fig. 5A). In agreement with ADIPOQ mRNA levels, secreted adiponectin was detected in the supernatant of adipocytes only (Fig. 5B). Leptin secretion was detected in the supernatants of preadipocytes of PD and T2D and of adipocytes of NGT, PD, and T2D (Fig. 5C). T2D fat cells secreted more leptin than NGT cells. Growth factors (VEGF and HGF) were secreted from preadipocytes and adipocytes, whereas chemokines (IL-6 and MCP-1) were mainly produced by preadipocytes (Fig. 5, D-G). In comparison, the secretome of subcutaneous fat cells changed in a similar direction during differentiation as the NGT-PPP secretome: leptin and adiponectin secretion increased, whereas VEGF and IL-6 secretion decreased (Fig.  5, B-G). These results suggest that T2D pancreatic fat cells (preadipocytes and adipocytes) secrete more leptin than NGT, whereas the secretion of other factors including adiponectin, growth factors, and chemokines/cytokines were not different.
In conclusion, diabetes, that is, chronic hyperglycemic episodes reflected by increased HbA1c, is associated with changes of the transcriptome and function of pancreatic fat cells, which may elicit an altered fat cell-islet cross talk and further accelerate disease development.

DISCUSSION
The development of T2D is associated with systemic insulin resistance (22). Our data show that both pancreatic preadipocytes and in vitro differentiated adipocytes of T2D-PPP display insulin resistance. In accordance, insulin-dependent AKT phosphorylation was reduced in pancreatic fat cells from T2D-PPP compared with NGT-PPP (Fig. 3). Furthermore, during adipogenesis, target genes of the insulin-sensitive transcription factor SREBP1c, such as FASN (FAS), ELOVL6, SCD and FADS1, were not upregulated in cells of T2D-PPP, contrary to NGT-PPP (Fig. 2, K-N). The reduced insulin sensitivity is not a consequence of reduced expression of INSR, in agreement with a previous observation in insulin-resistant subcutaneous adipocytes [ Fig. 4E; Ref. and adipocytes (F). Results are expressed as means ± SE, n = 5-6. ÃÃÃ P < 0.001 versus respective 100 nmol/L insulin; † P < 0.05, † † P < 0.01, † † † P < 0.001 versus respective 10 nmol/L insulin; ‡ P < 0.05 versus respective 1 nmol/L insulin; § §P < 0.01 versus SC 10 nmol/L insulin; two-way ANOVA followed by Tukey posttesting. NGT, normal glucose tolerance (white columns and symbols); SC, subcutaneous adipocytes (light gray columns and symbols); and T2D, type 2 diabetes (black columns and symbols). (23)]. It is likely that the lower insulin sensitivity observed in T2D-PPP is related to alterations downstream of the insulin receptor (23)(24)(25).
This insulin resistance did not translate to a reduced capacity of triglyceride storage, indicating that insulin resistance might reduce the rate of lipogenesis rather than the storage capacity. Although SLC2A4 mRNA level increased in NGT-PPP upon differentiation, GLUT4 expression remained most likely too low for a sufficient activation of de novo lipogenesis in in vitro differentiated pancreatic adipocytes. This feature seems to represent an intrinsic defect of pancreatic preadipocytes as subcutaneous preadipocytes retain a much higher de novo lipogenic activity in vitro than the pancreatic preadipocytes. Supplementation of culture medium with MUFAs increased triglyceride storage in pancreatic adipocytes, but the formation of a central lipid droplet was still incomplete.
Despite insulin resistance, the basal lipolytic activity of adipocytes from T2D-PPP was similar to that of NGT-PPP and refractory to stimuli. This feature is distinct from the lipolytic behavior of visceral adipocytes, which increases their lipolysis along with insulin resistance (26). Fatty acid release was stimulated 1.5-fold by forskolin, exclusively in NGT-PPP, while isoproterenol and glucagon were ineffective. The direct stimulation of adenylyl cyclase by forskolin bypasses receptor activation which in pancreatic adipocytes is most likely nonfunctional given the very low expression of b-adrenergic and the absence of glucagon receptors [Fig. 4, E and F; Refs. (27)(28)(29)]. Since IBMX, a phosphodiesterase inhibitor, had no significant stimulatory effect on lipolysis, increased phosphodiesterase activity seems not to contribute to the low responsiveness of the cells. The reduced lipogenic capacity was also not the reason for the low lipolytic response. Thus, pancreatic adipocytes differentiated in MUFA-supplemented medium did not secrete more fatty acids in spite of increased lipid storage. In comparison to the subcutaneous adipocytes, where forskolinstimulated lipolysis 12-fold, the cAMP-dependent stimulation of lipolysis is less efficient in human pancreatic adipocytes (Fig. 4D). In agreement, a recent study in mice reported lower mRNA levels of lipogenic and lipolytic markers in peripancreatic adipose tissue in comparison to subcutaneous and other visceral fat depots (30).
Compatible with the high expression of NPR1 and NPR2, ANP stimulated fatty acid release of NGT pancreatic adipocytes. The expression of NPR1 and NPR2 in other human fat depots was first described in the late 1990s (31). Lipolysis was stimulated by natriuretic peptides in isolated adipocytes and human in vivo studies using ANP infusion (32,33). The effect of ANP is exerted through the activation of guanylate cyclase, an increase of intracellular cGMP, and subsequent activation of PKG (34,35). Thus, the effect of ANP is not counteracted by insulin, which interferes with the cAMP pathway (32,33,36). Insulin activates PDE3B, while cGMP is mainly hydrolyzed by PDE5 (37)(38)(39). Adipocytes normally have a higher expression of PDE3B than of PDE5, and inhibition of PDE5 did not increase ANP stimulation in subcutaneous and visceral human adipocytes (34,(40)(41)(42). In pancreatic adipocytes, we found more PDE5A than PDE3B mRNA levels (Fig. 4H). In addition, dipyridamole increased lipolysis in the presence and absence of ANP. However, T2D adipocytes did not respond to ANP and dipyridamole despite the high mRNA levels of NPR1, NPR2, and PDE5A (Fig. 4, G and H). These observations suggest a metabolic/humoral, that is, through cGMP, rather than a neuronal regulation, that is, through cAMP, of fatty acid release from isolated, in vitro differentiated pancreatic adipocytes.
insulin hypersecretion under hypoglycemic situations. However, whether, in humans, pancreatic adipocytes exert paracrine effects on insulin secretion via fatty acid release needs further experimental evidence. In this study, we report functional differences between adipocytes of NGTand T2D-PPP as lipolysis of T2D-PPP adipocytes was refractory to any stimuli.
Besides metabolites, adipocytes secrete also adipokines (19). Multiple observations suggest beneficial effect of adiponectin on b-cell function and survival and inhibitory effects of leptin on insulin secretion (44). Adiponectin receptors are expressed in human b-cells (45). Leptin acts as an anorexogenic hormone mainly on NPY/AGRP and POMC neurons in the hypothalamus (46). Whether leptin acts directly on insulin secretion, in view of a low abundance of leptin receptor mRNA in human b-cells is a yet unresolved question (45). Leptin, however, affects insulin receptor signaling, which aggravates insulin resistance (24). Here, we found an increasing mRNA leptin/adiponectin ratio positively associated with HbA1c. Leptin secretion was higher in preadipocytes and adipocytes of T2D-PPP than of NGT-PPP. Increases in leptin/adiponectin plasma levels are observed in patients with T2D and related to impaired insulin sensitivity (47)(48)(49). Of note, secreted adiponectin levels were more than 1,000 times higher than that of leptin, suggesting that a beneficial paracrine effect of adiponectin would be more likely than an inhibitory action of leptin on insulin secretion (50).
Additional factors secreted by preadipocytes and adipocytes are cytokines and chemokines of which IL-6 and MCP-1 production was higher in the undifferentiated fat cells. Moreover, IL-6 secretion in PD and T2D adipocytes was higher than in NGT adipocytes, which could be an additional factor contributing to impaired adipogenesis. Increased adipose IL-6 secretion is observed in obese patients with insulin resistance and T2D and correlates with decreased adipogenic capacity (51). Previously, we found that fetuin-A and palmitate specifically increased the production of IL-6 and MCP-1 through TLR4 in pancreatic fat cells, which might lead to increased islet macrophage infiltration (7). Macrophages, in turn, can further impair preadipocyte differentiation and exacerbate adipocyte IL-6 and MCP-1 secretion while decreasing adiponectin release (52). Thus, exposure to diabetogenic factors triggers a proinflammatory state, which can ultimately contribute to b-cell damage.
The observation that DEG downregulated during adipogenesis is enriched in GO terms specific for angiogenesis, extracellular matrix remodeling, cytoskeleton organization, and cell motility, and that T2D-PPP display reduced adipogenesis, suggest that preadipocytes play an important role in matrix remodeling including fibrosis in T2D-PPP (53,54). How local changes in matrix remodeling impact on b-cell function remains to be tested.
In conclusion, the metabolic status impacts the secretome of local pancreatic fat cells and this may influence islet function via paracrine actions.