RESEARCH ARTICLE Role of Adipose Tissue Nutrient/Vitamin Metabolism in Physiological and Altered Metabolic Settings Glucose-dependent insulinotropic polypeptide promotes lipid deposition in subcutaneous adipocytes in obese type 2 diabetes patients: a maladaptive response

Glucose-dependent insulinotropic polypeptide promotes lipid deposition in subcutaneous adipocytes in obese type 2 diabetes patients: a maladaptive response. Physiol doi:10.1152/ajpendo.00347.2016.—Glu-cose-dependent insulinotropic polypeptide (GIP) beyond its insulinotropic effects may regulate postprandial lipid metabolism. Whereas the insulinotropic action of GIP is known to be impaired in type 2 diabetes mellitus (T2DM), its adipogenic effect is unknown. We hypothesized that GIP is anabolic in human subcutaneous adipose tissue (SAT) promoting triacylglycerol (TAG) deposition through reesteriﬁcation of nonesteriﬁed fatty acids (NEFA), and this effect may differ according to obesity status or glucose tolerance. Twenty-three subjects categorized into

GIP has other important extrapancreatic metabolic functions, with receptors expressed in such tissues as bone, brain, stomach, and adipose tissue, where it may modulate postprandial lipid metabolism (7). In animal models of obesity-induced insulin resistance, genetic and chemical disruption of GIP signaling protects against the deleterious effects of high-fat feeding by preventing lipid deposition, adipocyte hypertrophy, and expansion of adipose tissue mass and reducing triglyceride deposition in liver and skeletal muscle, maintaining insulin sensitivity (25,31). Thus, if GIP has a potential proadipogenic effect, selective GIP antagonists may be beneficial in treating obesity and type 2 diabetes mellitus (T2DM) (17).
There is evidence that plasma GIP concentrations are increased in obesity. Given that dietary fat consumption chronically stimulates the production and secretion of GIP, inducing K cell hyperplasia (8,36), higher GIP concentrations may reflect consumption of an energy dense, high-fat diet. Early rodent studies demonstrated that a GIP infusion during an intraduodenal lipid infusion decreased plasma triglyceride levels (14), whereas GIP has been shown to enhance insulininduced fatty acid incorporation in rat adipose tissue (9). Thus GIP mediated through the adipocyte GIP receptor is anabolic in adipose tissue promoting fat deposition.
It is important to distinguish between direct effects of GIP on fatty acid metabolism and indirect effects based on its insulinotropic action. Acute GIP infusion in lean healthy males (with hyperinsulinemia and hyperglycemia) increases adipose tissue blood flow, triacylglycerol (TAG) hydrolysis, and nonesterified fatty acid (NEFA) reesterification, thus promoting triglyceride deposition (5,6). In healthy obese men, acute GIP infusion reduced expression and activity of 11␤-hydroxysteroid dehydrogenase type 1, a fat-specific glucocorticoid me-tabolism enzyme that may enhance lipolysis in subcutaneous adipose tissue (SAT) (20). In addition, it has been suggested that GIP contributes to the induction of adipocyte and SAT inflammation (and thus insulin resistance), increasing production of proinflammatory adipokines such as monocyte chemoattractant protein-1 (MCP-1) (21), IL-6, IL-1␤, and osteopontin (1,37). Thus, from the available animal model and human data, GIP appears to have a key regulatory role in lipid metabolism and adipose tissue.
To date, very few studies have investigated the effects of GIP on human adipose tissue, and none have involved subjects with T2DM, although the reported presence of functional GIP receptors on adipocytes strongly suggests that GIP modulates human adipose tissue metabolism (41). GIP has also been proposed to modulate other adipose tissue depots, and excessive GIP secretion may underlie excessive visceral and liver fat deposition (33,34). In support of this, results from a crosssectional study of Danish men demonstrated an association between higher levels of GIP (during a glucose tolerance test) and a metabolically unfavorable phenotype (higher visceral: subcutaneous fat and a higher waist/hip ratio) (32).
We hypothesized that GIP would have an anabolic action in SAT promoting NEFA reesterification, which we speculated may be mediated either by enhancing lipoprotein lipase (LPL) expression/activity (a lipogenic enzyme) (15,26) or by reducing adipose tissue triglyceride lipase (ATGL) and hormonesensitive lipase (HSL) expression/activity, two key lipolytic enzymes. We postulated that this effect may be different according to obesity status or glucose tolerance. Thus, we set out to determine the acute in vivo effects of intravenous GIP on 1) plasma/serum insulin and nonesterified fatty acid (NEFA) concentrations, and 2) TAG content and gene expression of the key lipid-regulating genes LPL, ATGL, and HSL in SAT in obese individuals with different categories of glucose regulation [normoglycemic, impaired glucose regulation (IGR), and type 2 diabetes mellitus (T2DM)] vs. lean, normoglycemic controls.
Lean and obese were defined according to a BMI of Յ25 and Ն30 kg/m 2 , respectively. Allocation to glucose regulation categories was based on recent medical records combined with a fasting plasma glucose concentration. Obese subjects were allocated to the obese IGR group if they had one or more of the following: fasting hyperglycemia, impaired glucose tolerance on a 75-g oral glucose tolerance test (OGTT), or Hb A 1C in the prediabetes range (6 -6.5% or 42-47 mmol/mol). Obese subjects with T2DM (according to World Health Organization diagnostic criteria) (40) and not on pharmacological treatment for diabetes were allocated to the obese T2DM group. Homeostatic model assessment (HOMA)-2 was used to estimate whole body insulin resistance (23); adipose tissue insulin resistance (Adipo-IR) was calculated from fasting NEFA (mmol/l) and insulin (pmol/l) concentrations (19). Baseline demographic, anthropometric, and biochemical parameters of all participants are shown in Table 1.

Ethical Approval
Ethical approval for this project was obtained from the Northwest Research Ethics Committee (UK; REC ref. no. 08/H1001/20). All subjects were studied after informed and written consent was obtained.

Study Protocol
Each subject was studied on two separate occasions 1-3 wk apart. After overnight fasting, subjects were infused with either GIP (2 pmol·kg Ϫ1 ·min Ϫ1 in 0.9% saline) or placebo (0.9% saline alone). GIP was dosed based on the rate infused in previous studies (16,35,38). Subjects were randomly assigned to either GIP or Values are means Ϯ SD. T2DM, type 2 diabetes mellitus; BMI, body mass index; NEFAs, nonesterified fatty acids; HOMA-IR, homeostasis model assessment-insulin resistance; Adipo-IR, adipose tissue insulin resistance. a P Ͻ 0.05, b P Ͻ 0.01, c P Ͻ 0.001, and d P Ͻ 0.0001, statistically significant difference vs. lean group; e P Ͻ 0.05, significant difference vs. obese group. placebo infusion on their initial visit and received the alternate infusion subsequently. Anthropometric assessments were recorded during each visit. Percent body fat estimation was determined by whole body bioelectrical impedance analysis (Tanita, Tokyo, Japan).
GIP infusions, hyperglycemic clamp, and blood sampling. Intravenous cannulae were inserted into both antecubital fossae for blood sampling and infusions (GIP or placebo). GIP (Polypeptide Laboratories, Strasbourg, France) was sterile-filtered and dispensed by Stockport Pharmaceuticals (Stepping Hill Hospital, Stockport, UK). A blood glucose concentration of~8.0 mmol/l was maintained during a hyperglycemic clamp using a priming dose of 20% glucose bolus (based on weight and fasting glucose) given in the first 5 min, followed by a variable rate infusion of 20% glucose adjusted according to whole blood glucose levels measured every 5 min on a YSI blood glucose analyzer (YSI UK). Intravenous infusion of GIP/ placebo was continued from 30 min after initiation of hyperglycemic clamp until 240 min. Ten-milliliter blood samples were taken at baseline (before hyperglycemic clamp) and at 15, 30, 60, 120, 180, and 240 min following the initiation of GIP/placebo infusion. To minimize protein degradation, aprotinin was added to the tubes before sample collection. Samples were centrifuged immediately, and serum was stored at Ϫ80°C until further analysis.
SAT biopsies. Subcutaneous adipose tissue (SAT) biopsies were obtained at baseline and after 240 min of the GIP/placebo infusion on the contralateral site. Under local anesthesia (1% lidocaine, adrenaline 1:200,000), a small incision was made through the skin and fascia 10 cm lateral to the umbilicus. Adipose tissue samples (50 -150 mg wet wt) were collected and snap-frozen in liquid nitrogen and stored at Ϫ80°C until further analysis.

Laboratory Analysis: Biochemical Analysis
Plasma glucose concentration, lipid profile, liver function parameters, and Hb A 1c were measured using a Cobas 8000 modular analyzer (Roche Diagnostics). Blood glucose concentrations during hyperglycemic clamp were measured using YSI 2300 STAT glucose analyzer (YSI UK; and Fleet, Hampshire, UK). Serum insulin was measured by ELISA method (Invitrogen, Fisher Scientific, Loughborough, UK). Nonesterified fatty acids (NEFAs) were measured from plasma by Randox kit on a Biostat BSD 570 analyzer (Randox Laboratories, London, UK). Intact GIP was measured at the University of Copenhagen (Copenhagen, Denmark); the assay is specific for the intact NH 2 terminus of GIP (biologically active peptide) (13).

SAT Analysis
SAT lipid content. Lysates were prepared by homogenization of fat biopsies in a buffer containing 50 mM Tris·HCl, pH ϭ 7.5, 150 mM NaCl, 1% Triton X-100, and standard protease inhibitor cocktail (Complete Mini protease inhibitor cocktail; Roche Diagnostics). Triacylglycerol (TAG) was quantified by measuring free glycerol output following overnight lipase treatment at 37°C (Sigma). The values were normalized according to protein content.
SAT gene expression. Gene expression of LPL, ATGL, and HSL was quantified through RNA extraction and real-time quantitative PCR. Total RNA was isolated using RNeasy Lipid Tissue Mini Kit (Qiagen). Real-time quantitative PCR was conducted in triplicate using a Bio-Rad CFX-connect real-time PCR instrument (Bio-Rad Laboratories) using prevalidated TaqMan probes (Life Technologies) as follows: endogenous control ␤-actin (Hs99999903_m1) and target genes lipoprotein lipase (lpl; Hs00173425_m1), ATGL (pnpla2; Hs00386101_m1), and hormone-sensitive lipase (lipe; Hs00193510_m1). Relative quantification was carried out using the ⌬⌬CT method with ␤-actin gene expression as an internal control.

Statistical Analysis
Participant demographics, baseline biochemical parameters, and blood glucose concentrations during hyperglycemic clamp are expressed as means Ϯ SD; all other results are expressed as means Ϯ SE. One-way analysis of variance (ANOVA) and Tukey's t-tests were performed to compare participant demographics and baseline biochemical parameters among the four groups in this study. Area under the curve for insulin and NEFA concentrations over a 4-h period of infusion (AUC 0 -4 h) were calculated by trapezoidal rule using GraphPad Prism software. Paired t-tests were performed on changes in gene expression and lipid content (SAT-TAG) parameters to explore whether the change over the two time points differed between GIP and placebo. P Ͻ 0.05 (2-tailed) was considered to be significant. A Pearson product-moment correlation coefficient was computed to assess the relationship between degree of NEFA reduction and other variables [fasting plasma glucose and adipose tissue insulin resistance (Adipo-IR)].
A linear mixed-effects model was also used to model insulin secretion and NEFA concentrations using three time points (baseline, 120 min, and 240 min). Main effects for the four different groups are included along with a two-way interaction between treatment and group. This allows that the overall effect of GIP infusion in comparison with the placebo infusion can be assessed individually for different groups. Results are expressed in estimated average unit changes in insulin and NEFAs during GIP vs. placebo infusion.

Baseline Characteristics: Patient Demographics
Twenty-three individuals completed the study protocol in four subgroups: lean (n ϭ 6), obese (n ϭ 6), obese IGR (n ϭ 6), and obese T2DM (n ϭ 5). Waist circumference and percent body fat mass were significantly higher in obese, obese IGR, and obese T2DM compared with the lean group. The duration of diabetes in obese T2DM group was 7 Ϯ 5.5 mo (means Ϯ SD), with a mean Hb A 1c of 54 Ϯ 8.5 mmol/mol (7.1 Ϯ 0.8%), and all participants were naive to oral or injectable diabetes medications.

Baseline Biochemistry: Plasma Glucose and Insulin Concentrations
As expected, mean fasting glucose was higher in obese IGR and obese T2DM groups compared with the two other groups. Fasting insulin and HOMA-IR were significantly higher in obese, obese IGR, and obese T2DM groups vs. the lean group. Adipo-IR was significantly higher in obese T2DM group vs. lean and obese groups but not vs. obese IGR group (Table 1).

Metabolic Parameters
All subjects in obese IGR and obese T2DM groups had metabolic syndrome based on International Diabetes Federation 2006 criteria (2) with most consequently treated for hypertension and dyslipidemia: ACE inhibitors or angiotensin receptor blockers (3 subjects in the obese IGR group, 5 subjects in the obese T2DM group), ␤-blockers (2 obese IGR, 2 obese T2DM), and calcium channel blocker (1 obese T2DM). Three subjects in each of the above two groups were on statins. Two subjects in the obese group had metabolic syndrome (one on ACE inhibitors and one a fibrate).

Biochemistry Changes During Infusions
Blood glucose. The blood glucose concentrations were maintained at~8.0 mmol/l during the hyperglycemic clamp with both GIP and placebo infusions in all four groups (Fig.  1, A-D) Fig. 2E). The change in insulin concentration over 240 min, compared with baseline values, differed by 63, 70, and 121 IU/ml with GIP infusion vs. placebo in the lean, obese, and obese IGR groups, respectively. In the obese T2DM group, there was only a 9 IU/ml increase in insulin concentration with GIP vs. placebo infusion (Fig. 2F).

Serum triacylglycerol concentration.
There were no significant alterations in serum triacylglycerol (TAG) concentrations with either GIP or placebo in any of the four groups (data not shown).

SAT Changes
SAT-TAG content. The changes in lipid content after 240 min of GIP vs. placebo infusion relative to respective baselines on each visit are shown in Fig. 5. In the obese T2DM group, the SAT-TAG content increased 1.78 Ϯ 0.4-fold (means Ϯ SE) from baseline with GIP infusion compared with 0.86 Ϯ 0.1-fold with placebo (95% CI: 0.1, 1.8, P ϭ 0.043). The changes in TAG content in the other three groups were not statistically significant (data shown in Fig. 5).
Gene expression of enzymes involved in lipid metabolism. The changes in mRNA expression (LPL, ATGL, and HSL) in SAT after 240 min of GIP vs. placebo infusion relative to respective baselines on each visit are shown in Fig. 6.
LPL. The LPL mRNA expression in the T2DM group was 1.25-fold higher from baseline with GIP infusion compared with the 0.94-fold change with placebo, but this was not statistically significant (P ϭ 0.27). In the other three groups, the changes in LPL mRNA expression with GIP and placebo were comparable (Fig. 6A).
ATGL. In the T2DM group, ATGL mRNA expression was higher with GIP infusion compared with placebo (1.5-vs. 1.1-fold, P ϭ 0.12), but this was not statistically significant. In the other three groups, the changes in ATGL gene expression with GIP vs. placebo were comparable (Fig. 6B). HSL. The changes in HSL gene expression with GIP did not differ significantly compared with placebo in all four groups (Fig.  6C). Fold change data for the three enzymes in all four groups is shown in Fig. 6D.

DISCUSSION
We demonstrate that acute GIP infusion during fasting under hyperglycemic conditions reduced plasma NEFAs, concomitantly increasing SAT triacylglycerol (TAG) content in obese patients with T2DM. This anabolic effect was not observed in the lean patients, obese patients, or obese patients with IGR. In contrast, whereas GIP was able to stimulate insulin secretion in the lean patients, obese patients, or obese patients with IGR, its insulinotropic action was not observed in obese patients with T2DM. Thus, in obese patients with T2DM, there is a dissociation of the effects on GIP on ␤-cells and adipocytes, with blunted insulinotropic but preserved lipogenic actions, respectively.
Expression of the GIP receptor (GIPR) is somehow glucose dependent and downregulated in response to hyperglycemia (24). In patients with T2DM, the blunted incretin effect (involving both incretin hormones glucagon-like peptide-1 and GIP) may be due in part to reduced islet cell expression of GIP receptors (GIPR) secondary to chronic hyperglycemia (16,29,35,39). The physiological role of GIP in adipose tissue in T2DM remains unclear, although adipose GIPR expression may be similarly downregulated in insulin-resistant human subjects and may represent a compensatory mechanism to reduce fat storage in insulin resistance, considering the interference of NEFAs on insulin signal transduction (10,22). However, energy-dense, high-fat diets in obese individuals with T2DM could result in exaggerated fat storage (through exaggerated GIP release), even in the absence of adequate insulin secretion (Fig. 7). Although we did not measure GIPR, the lipogenic action of GIP at the adipocyte appears to be more pronounced in T2DM (Fig. 5). Studies in patients with nonalcoholic fatty liver disease suggest that elevated GIP secretion is also associated with intrahepatocellular lipid deposition (33).
Several factors may explain the differential ability of GIP to increase NEFA reesterification in SAT in obese T2DM subjects vs. other groups. In lean individuals, obese individuals, and obese individuals with IGR, where insulin secretion is potently stimulated and adipose tissue insulin sensitivity is preserved (lower Adipo-IR), insulin independently suppressed lipolysis, lowering NEFAs and perhaps leaving GIP's effects trivial. However, in T2DM, when insulin secretion is impaired and adipose tissue is insulin resistant (high Adipo-IR), the effect of GIP assumes greater importance, promoting lipid accumulation in adipocytes. This is consistent with animal data. GIP does not promote fat accumulation in adipocytes with normal insulin sensitivity, with GIPR Ϫ/Ϫ mice showing similar adiposity to wild-type mice on control diet (31). However, under conditions of diminished insulin action using insulin receptor substrate-1 (IRS-1)-deficient mice, when the effects of GIP were examined (by disrupting GIP signaling, GIP Ϫ/Ϫ vs. GIPR ϩ/ϩ ), GIP was shown to promote SAT and VAT expansion and decrease fat oxidation, with greater SAT and VAT mass and lower fat oxidation in IRS-1 Ϫ/Ϫ /GIPR Ϫ/Ϫ vs. IRS-1 Ϫ/Ϫ /GIPR ϩ/ϩ mice (42). A few human studies examined the metabolic effect of an acute GIP infusion in lean and obese individuals but none reported in people with T2DM. In studies to date, the effects of GIP have been examined under experimental conditions different from those here, for example, during concomitant intralipid infusion and/or with hyperinsulinemic hyperglycemic clamp conditions and measuring arteriovenous concentrations of metabolites. These data demonstrated that in lean people, GIP in combination with hyperinsulinemia and hyperglycemia increased adipose tissue blood flow, glucose uptake, and NEFA reesterification, thus resulting in increased abdominal SAT-TAG deposition (4 -6). The same group showed that, in obese and IGR subjects, GIP infusion did not have the same effect on adipose tissue blood flow or TAG deposition in adipose tissue (3). However, the independent contributions of insulin vs. GIP to these metabolic effects are difficult to dissect, although GIP per se appeared to have little effect on human subcutaneous adipose tissue in  lean insulin-sensitive subjects, with an effect apparent only when GIP was coadministered with insulin during hyperglycemia. Thus it would appear that there are direct and indirect effects of GIP. During nutrient excess, lipogenesis is stimulated via lipoprotein lipase (LPL), hydrolyzing circulating lipoprotein-de-rived triglycerides and promoting NEFA esterification into TAG and storage within lipid droplets of adipose tissue. During periods of fasting, mobilization of NEFAs from fat depots relies on the activity of key hydrolases, including hormone-sensitive lipase (HSL) and adipose triglyceride lipase (ATGL). In SAT, insulin stimulates NEFA esterification by    7. In healthy people, GIP acts on its receptors on ␤-cells and adipocytes to promote insulin secretion (insulinotropic action) and lipid deposition (adipogenic action) (left). In obesity, with consumption of an energy-dense, higher-fat diet, there is enhanced insulin secretion (which may help overcome peripheral insulin resistance) and increased lipid deposition (which will further enhance fat storage) (middle). In T2DM, the effects of GIP on ␤-cell are impaired with reduced insulin secretion; the effects on the adipocyte seem to be preserved, further promoting lipid deposition (right). enhancing lipoprotein lipase (LPL),and inhibits lipolytic process (18). The majority of the animal studies have shown that GIP potentiates the role of insulin in regulation of LPL and NEFA incorporation into adipose tissue (9,15,27,31). GIP enhances LPL gene expression in cultured subcutaneous human adipocytes through pathways involving protein kinase B and AMP-activated protein kinase (26,28). Trying to determine the molecular mechanism by which SAT-TAG content changed, we measured SAT mRNA expression of LPL, ATGL, and HSL; surprisingly, we observed no significant changes in expression to account for altered serum NEFAs or SAT-TAG content. This may represent a time course phenomenon (changes in gene expression with GIP in human adipose tissue may occur over a longer interval). This speculation is consistent with the slow temporal onset of the molecular responses in adipose tissue in animal studies. GIP infusion may affect enzyme activity rather than gene expression and therefore results may differ if activity/phosphorylation was measured.
To better appreciate the physiological effects of GIP administration on human SAT, stable isotope studies to determine dynamic changes in fat metabolism with serial tissue biopsies are required. All studies were performed under hyperglycemic clamp conditions to achieve comparable hyperglycemia and to mimic postprandial increases in GIP and insulin. The peak GIP concentrations achieved in our study during GIP infusions were comparable with levels achieved elsewhere (3). We believe the changes in NEFAs and SAT lipid content in our obese T2DM are more likely due to the effect of GIP, particularly in the absence of excess insulin secretion. Reductions in NEFA correlated positively with fasting glucose and Adipo-IR in all the subjects across the four groups, suggesting that the effects of GIP are more pronounced in hyperglycemic and insulin-resistant states. We recognize that higher ⌬NEFA would be expected in subjects with higher fasting NEFA levels, however correlation with Adipo-IR was seen only with GIP but not with placebo infusion (Fig. 4).
Studying four distinct groups (with differing BMI and glucose tolerance) facilitates evaluation of the differential effects of GIP in insulin-sensitive and -resistant individuals. However, we acknowledge limitations, including small group sizes and the degree of obesity; there was limited pilot data in humans before the initiation of this study, and subsequently published human studies on GIP infusion had small number of subjects (3)(4)(5). Findings from our study may differ in less severely obese individuals. Lean subjects were younger compared with others and may have increased insulinotropic activity to GIP (30), but there was no significant difference in insulin AUC between the groups, except in obese T2DM. Unrecognized interactions between antihypertensive or lipid-modifying medication and the effects of GIP cannot be excluded.
In conclusion, we demonstrate that in obese patients with T2DM, acute GIP infusion in a fasting state during hyperglycemia lowers serum NEFA and increases the SAT lipid content despite reduced insulinotropic activity. In lean, obese, and obese with IGR, despite the intact insulinotropic response to GIP, no lipogenic effect was observed. This anabolic effect of GIP further exacerbates obesity and insulin resistance.