Introduction

Adipose tissue (AT) distribution is a major determinant of mortality and morbidity in obesity [1]. Specifically, accumulation of intra-abdominal or visceral AT (VAT) is associated with type 2 diabetes, dyslipidaemia and hypertension [2]. This relationship was originally attributed to increased NEFA flux from VAT into the portal circulation [3]. However, VAT makes only a minor contribution to portal NEFA concentrations [4], suggesting that other factors, such as adipokines [5], are also involved.

Independently of VAT, greater amounts of subcutaneous AT (SAT), particularly in the lower body, may protect against glucose intolerance, insulin resistance (IR), dyslipidaemia and atherosclerosis [6]. Gluteal–femoral SAT, for example, prevents lipotoxicity in non-adipose tissues by sequestering meal-derived NEFA [7, 8], and resists deleterious gene expression changes in obesity [9]. Understanding how different AT depots influence the risk of glucose intolerance, IR and dyslipidaemia may yield new therapeutic targets for metabolic disease [10].

Obesity and IR are characterised by chronic inflammation [11], with elevated proinflammatory cytokine production in AT [12]. Adipocyte hypertrophy in AT is accompanied by macrophage infiltration [13] and polarisation towards a classically activated proinflammatory (M1) type rather than ‘alternatively’ activated, anti-inflammatory macrophages (M2) [14]. Proinflammatory cytokines impair insulin action both locally and in other tissues, such as liver [15]. AT from obese, insulin-resistant humans also has a higher proinflammatory T cell content [16]. Anti-inflammatory treatment improves glycaemic control in humans with type 2 diabetes [17].

We and others have investigated the relationship between regional adiposity and metabolism by performing syngeneic AT transplantation in mice [1822]. Intra-abdominal transplantation of epididymal AT was initially reported to improve glucose tolerance and reduce insulinaemia within 4–8 weeks [18]. In mice fed a high-fat diet (HFD) for 13 weeks, however, we found that intra-abdominal transplantation of inguinal AT, but not epididymal AT, improved glucose tolerance and reduced adiposity [19]. This did not occur when inguinal AT was transplanted subcutaneously. Our findings were confirmed by Tran et al [20], who also reported differences in AT glucose uptake between mice receiving intra-abdominal AT transplants and sham-operated mice, under euglycaemic–hyperinsulinaemic clamp conditions. More recently, intra-abdominal transplantation of either inguinal or epididymal AT has been reported to reverse HFD-induced glucose intolerance and liver triacylglycerol accumulation within 4 weeks [21].

We aimed to address these discrepancies and identify the underlying mechanisms. While the beneficial effects of inguinal SAT transplantation on glucose tolerance are highly reproducible [1922], epididymal AT transplantation is linked with both beneficial [18, 21] and negligible [19, 20] effects on metabolism, relative to sham-operated mice. Therefore, we performed time course studies to investigate the short- and long-term effects of inguinal subcutaneous → visceral (SubQ → Vis) and epididymal visceral → visceral (Vis → Vis) AT transplantation in HFD-fed mice.

Methods

Animals and surgery

Six-week-old male C57BL6/J mice were obtained from Australian Bioresources (Moss Vale, NSW, Australia) and housed as described previously [23]. Studies were approved by the Garvan Institute/St Vincent’s Hospital Animal Experimentation Ethics Committee, and were performed according to guidelines set by the National Health and Medical Research Council of Australia.

Inguinal and epididymal AT depots were dissected out from donors and transplanted into anaesthetised recipient littermates via a midline incision. AT was sutured onto the peritoneal surface of the anterior abdominal wall using Vicryl 4-0 sutures (Ethicon, Somerville, NJ, USA). The incision was closed with wound clips. Sham-operated mice received identical surgical treatment without transplant. After regaining their pre-surgical weight, mice were provided ad libitum with either chow (8% of energy from fat, Gordon’s Specialty Stock Feeds, Yanderra, NSW, Australia) or HFD (45% of energy from fat, based on D12451, Research Diets, New Brunswick, NJ, USA). At the end of the experiment, mice were killed by cervical dislocation.

Body composition, respirometry and food intake

Body composition was measured by dual-energy x-ray absorptiometry (DEXA; Lunar PIXImus, GE Medical Systems, New York, NY, USA) under isoflurane anaesthesia. Twelve weeks after transplantation, food intake, energy expenditure and RQ were measured as described previously [23].

Glucose and insulin tolerance tests

Intraperitoneal glucose tolerance tests (GTTs; 2 g/kg) were performed in conscious 6-h-fasted mice. For insulin tolerance tests (ITTs), mice were fasted for 4 h before injection of 0.75 U/kg body weight i.p. (Actrapid, Novo Nordisk, Baulkham Hills, NSW, Australia). Blood glucose concentrations were measured using an Accu-Chek meter (Roche Diagnostics, Castle Hill, NSW, Australia).

For measurement of tissue-specific glucose uptake, 6-h-fasted mice were injected with 1.5 g/kg glucose containing tracer amounts of [U-14C]glucose (370,000 Bq/mouse) and [3H-2]deoxyglucose (37,000 Bq/mouse; Perkin-Elmer, Glen Waverley, VIC, Australia). Tail blood was collected into heparinised tubes (Microvette CB300LH, Sarstedt, Mawson Lakes, SA, Australia), and plasma was separated by centrifuging at 2,000×g for 5 min. Glucose uptake in skeletal muscle and individual AT depots was normalised for the specific activity of glucose in plasma as described previously [24].

AT adipocyte sizing and immunohistochemistry

Tissues were fixed in 3.7% (vol./vol.) formaldehyde, paraffin-embedded, cut into 4 μm sections and stained with haematoxylin and eosin. Adipocyte size was determined in three representative fields from each section (100-150 adipocytes/field) using ImageJ software (National Institutes of Health, USA). Adipocyte size distribution was determined for each depot by assigning adipocytes to 500 μm2 bins. Differences between groups were determined by χ 2 test. Graft apoptosis was assessed by cleaved caspase-3 immunohistochemistry [25].

Biochemical methods

Tissue triacylglycerol and glycogen content, and plasma insulin and NEFA concentrations were measured as described [23]. Plasma leptin and adiponectin concentrations were measured by ELISA (Crystal Chem, Downers Grove, IL, USA and R&D Systems, Minneapolis, MN, USA, respectively). Plasma lactate was measured using a YSI 2300 Analyzer (YSI Life Sciences, Yellow Springs, OH, USA).

Microarrays

Tissues were obtained in our previous 13 week study [19], which included mice receiving subcutaneous inguinal AT in the dorsal subcutaneous space (SubQ → SubQ). RNA was extracted from the following depots (n = 5 of each): SubQ → Vis and SubQ → SubQ grafts, endogenous inguinal SAT from SubQ → Vis mice (SubQ → Vis ING), endogenous epididymal AT from SubQ → Vis mice (SubQ → Vis EPI) and inguinal SAT from a sham-operated mouse (Sham ING).

cDNA preparation and array hybridisation were performed according to the manufacturer’s instructions (Affymetrix Mouse Genome 430 2.0, Santa Clara, CA, USA). Gene expression differences >twofold were analysed using Genepattern software [26]. Gene Set Enrichment Analysis (GSEA) was performed using the c3_all and c5 collections in the Molecular Signatures Database (MSigDB) [27, 28]. Results were deposited in a public functional genomics data repository (Gene Expression Omnibus, www.ncbi.nlm.nih.gov/geo).

Immunoblotting

Tissues were homogenised in RIPA buffer (65 mmol/l Tris · HCl, 150 mmol/l NaCl, 5 mmol/l EDTA, 1% NP-40 [vol./vol.], 0.5% sodium deoxycholate [wt/vol.], 0.1% SDS [wt/vol.] and 10% glycerol [vol./vol.], pH 7.4) containing 1 mg/l aprotinin, 1 mg/l leupeptin, 10 mmol/l NaF, 1 mmol/l Na3VO4 and 1 mmol/l PMSF. Cleared lysates (20 μg protein) were electrophoresed in 10% polyacrylamide gels and transferred to PVDF membranes (GE Healthcare, Buckinghamshire, UK). Antibodies for stearoyl-CoA desaturase-1 (SCD1; 2438), fatty acid synthase (FAS; 3180), acetyl-CoA carboxylase (ACC; 36625) and acyl-CoA synthetase long-chain family member 1 (ACSL1; 4047) were obtained from Cell Signaling Technologies (Danvers, MA, USA). The sterol regulatory element-binding protein (SREBP1) antibody (MS1207P1ABX) was obtained from Thermo Scientific (Scoresby, VIC, Australia). Glucose 6-phosphatase (G6Pase) and 14-3-3 (sc-25840 and sc-629, respectively) antibodies were supplied by Santa Cruz Biotechnology (Santa Cruz, CA, USA). Appropriate HRP-conjugated secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA, USA). Densitometry was performed using ImageJ software.

Cytokine profiling

Plasma was collected at 4 and 10 weeks post-transplantation. Freshly thawed plasma was filtered using sterile Ultrafree-MC GV centrifuge filters (Millipore, Kilsyth, VIC, Australia) and cytokine concentrations were determined using a Bio-Plex Pro Mouse Cytokine Assay (M60-009RDPD) and a Bio-Plex 200 instrument (Bio-Rad, Hercules, CA, USA). The intra-assay CV for all cytokines was >5%.

Statistical analysis

All data are presented as mean ± SEM. Glucose excursions during the GTT were expressed as incremental AUC (iAUC), calculated using Microsoft Excel. Differences between groups were analysed by Student’s t test, one-way or two-way ANOVA (or their non-parametric equivalents) using GraphPad Prism (La Jolla, CA, USA). Multiple comparisons were corrected using the Holm–Sidak (parametric) or Dunn’s (non-parametric) tests. Relationships between continuous variables were assessed using Pearson’s or Spearman correlations, as appropriate. A two-sided p < 0.05 was considered significant.

Results

Glucose tolerance in SubQ → Vis mice was improved from 6 weeks post-transplantation

We previously reported beneficial effects of SubQ → Vis transplantation at 12–13 weeks [19]. Here, we studied HFD-fed Sham, SubQ → Vis and Vis → Vis mice for 3, 6 and 17 weeks (Fig. 1a). AT transplantation did not significantly affect body weight (Fig. 1b), fat mass (Fig. 1c) or lean mass (Fig. 1d). Masses of AT depots at 3, 6, and 17 weeks after transplantation are shown in Table 1. No significant differences in energy expenditure, RQ or food intake were observed (Electronic supplementary material [ESM] Table 1).

Fig. 1
figure 1

Whole-body effects of intra-abdominal AT transplantation. (a) Cohorts: 22 mice (n = 6–8/group) were studied for 3 weeks, 27 mice (n = 7–10/group) for 6 weeks and 24 mice (n = 7–9/group) for 17 weeks. (b) Body weight: Sham, SubQ → Vis and Vis → Vis mice are indicated by black circles, white circles and black squares, respectively. (c) Fat mass; (d) lean mass; (c, d) Sham, SubQ → Vis and Vis → Vis mice are indicated by black, white and striped columns, respectively (n = 14–19/group at 6 weeks and n = 7–9/group at 13 weeks). (e–h) Glucose tolerance at 3, 6, 10 and 14 weeks, respectively; symbols are as for Fig. 1b. (i) iAUC at each time point. *p < 0.05 vs Sham, p < 0.05 and †† p < 0.01 vs Vis → Vis (n = 6–8, n = 14–19, n = 7–9 and n = 7–9/group at 3, 6, 10 and 14 weeks, respectively); bar colour is as for Fig. 1c, d. (j) ITT at 7 weeks; symbols are as for Fig. 1b; p < 0.05 for SubQ → Vis vs Vis → Vis and **p < 0.01 for SubQ → Vis vs Sham mice. (k) ITT expressed relative to baseline; *p < 0.05 for Sham vs Vis → Vis mice. EE, energy expenditure; IPGTT, intraperitoneal GTT; T, transplantation

Table 1 Fat depot weightsa in Sham, SubQ→Vis and Vis→Vis mice at 3, 6 and 17 weeks after transplantation

Glucose tolerance was sustainably improved in SubQ → Vis mice from 6 weeks onwards (Fig. 1e–i). At 7 weeks after transplantation, SubQ → Vis mice had the lowest basal and 15 min blood glucose concentrations during an ITT (Fig. 1j, k).

Grafted subcutaneous fat regresses, while grafted visceral fat expands intra-abdominally

Comparable amounts of AT were initially transplanted into SubQ → Vis and Vis → Vis mice, but over the 17 weeks, grafted inguinal SAT gradually reduced in mass, while grafted epididymal VAT expanded (Table 1 and Fig. 2a). Apoptosis was generally absent (ESM Fig. 1). SubQ → Vis grafts had smaller adipocytes than Vis → Vis grafts, particularly at 17 weeks (Fig. 2b–j); and Vis → Vis grafts expressed more leptin (data not shown). Plasma concentrations of leptin (Fig. 2k) and adiponectin (Fig. 2l) did not differ among groups.

Fig. 2
figure 2

Effects of transplantation on grafted AT. (a) Relative changes in graft mass. SubQ → Vis mice (n = 6–7/cohort) are indicated by white circles and Vis → Vis mice (n = 8–10/cohort) by black squares. (b) Adipocyte area in grafts at 3 weeks; p < 0.0001 for difference, χ 2 test. (c) Adipocyte area in grafts at 6 weeks; p = 0.044, χ 2 test. (d) Adipocyte area in grafts at 17 weeks; p < 0.0001, χ 2 test. (e–g) SubQ → Vis grafts at 3, 6 and 17 weeks, respectively. Scale bars, 100 μm. (h–j) Vis → Vis grafts at 3, 6 and 17 weeks, respectively. Scale bars, 100 μm. (k) Plasma leptin concentrations. Sham, SubQ → Vis and Vis → Vis mice are indicated by black, white and striped columns, respectively (n = 8, n = 5–6 and n = 8–9, respectively, for Sham, SubQ → Vis and Vis → Vis mice at each time point). Effect of time: p < 0.0001; effect of group: p = 0.17; two-way ANOVA. (l) Plasma adiponectin concentrations; bars are as for Fig. 2k; effect of time: p = 0.055; effect of group: p = 0.63

Improved glucose tolerance in SubQ → Vis mice does not involve altered skeletal muscle or endogenous AT glucose uptake

Next, we investigated tissue-specific glucose uptake in chow-fed mice and HFD-fed Sham, SubQ → Vis and Vis → Vis mice at 13 weeks after transplantation. HFD feeding increased adiposity (Fig. 3a), worsened glucose tolerance (Fig. 3b, c) and increased fasting insulinaemia (Fig. 3d).

Fig. 3
figure 3

Radiolabelled GTTs. (a) Endogenous and grafted AT masses. Results from chow (vertical striped bars, n = 10), HFD Sham (black bars, n = 12), HFD SubQ → Vis (white bars, n = 8) and Vis → Vis (diagonally striped bars, n = 10) mice are shown. **p < 0.01 and ***p < 0.001 vs Sham, p < 0.05 vs Vis → Vis mice. (b) Plasma glucose concentrations. Results from chow (white squares, n = 10), HFD Sham (black circles, n = 12), HFD SubQ → Vis (white circles, n = 8) and HFD Vis → Vis (black squares, n = 10) mice are shown. (c) Plasma glucose iAUC; bars are as for Fig. 3a; *p < 0.05 vs Sham; p < 0.05 vs Vis → Vis. (d) Plasma insulin concentrations; symbols are as for Fig. 3b; *p < 0.05, ANOVA. (e) Plasma insulin AUC; bars are as for Fig. 3a; overall p = 0.092, Kruskal–Wallis test. (f) Plasma NEFA concentrations; symbols are as for Fig. 3b; effect of time: p < 0.0025; effect of group: p = 0.35, two-way repeated measures ANOVA. (g) Immunoblot for hepatic gluconeogenic enzymes PEPCK and G6Pase. Pooled samples were used (n = 8–12/group). PEPCK and G6Pase expression was normalised to 14-3-3 (loading control). (h) Skeletal muscle and AT glucose uptake; bars are as for Fig. 3a; *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001 vs Sham; †† p < 0.01 and ††† p < 0.001 for graft vs endogenous AT. DPM, disintegrations per minute; IPGTT, intraperitoneal GTT

SubQ → Vis mice were completely protected against HFD-induced glucose intolerance (Fig. 3b, c); and transplantation of either type of AT significantly reduced fasting insulinaemia (Fig. 3d, e). Suppression of NEFA concentrations was used as a surrogate measure of AT insulin sensitivity: fasting and 120 min NEFA concentrations tended to be higher in HFD-fed Sham mice compared with the other groups (Fig. 3f). SubQ → Vis transplantation also reversed HFD-induced expression of the hepatic gluconeogenic enzymes phosphoenolpyruvate carboxykinase (PEPCK) and G6Pase (Fig. 3g).

Glucose uptake into quadriceps muscle and endogenous AT (epididymal, inguinal, retroperitoneal, intrascapular brown) was blunted by HFD feeding but did not differ among HFD-fed Sham, SubQ → Vis and Vis → Vis mice (Fig. 3h). Grafted SAT and VAT displayed similar glucose uptake; however, when compared with their endogenous depots, glucose uptake was increased by 2.8-fold and 2.0-fold in SubQ → Vis and Vis → Vis grafts, respectively. Chronically increasing AT glucose uptake increases lactate release in the fed state and improves insulin sensitivity [29]: SubQ → Vis mice tended to have the highest postprandial plasma lactate concentrations at 17 weeks (ESM Fig. 2). Tracer incorporation into glycogen and triacylglycerol is shown in ESM Fig. 3.

Grafted and endogenous subcutaneous depots of SubQ → Vis mice display an induction of skeletal muscle-specific genes

To better understand the metabolic effects of SubQ → Vis transplantation, we compared gene expression in SAT implanted either intra-abdominally (SubQ → Vis) or subcutaneously (SubQ → SubQ) [19] (Fig. 4a). Twenty-eight genes displayed a >fivefold increase in expression in SubQ → Vis compared with SubQ → SubQ grafts (Fig. 4b); many were cytoskeletal components of skeletal muscle (Fig. 4c). GSEA identified a common transcription factor regulating 12 of these genes: myocyte enhancer factor 2A (MEF2A). MEF2 transcription factors have established roles in myogenesis, muscle metabolism, nervous system development/repair and more recently, immune response [30].

Fig. 4
figure 4

Gene expression analysis of grafted and endogenous AT. (a) Endogenous and transplanted AT depots studied. (b). Skeletal muscle-specific genes, including MEF2A targets, were induced in both the grafted and endogenous inguinal AT of SubQ → Vis mice. Genes displaying a >fivefold increase in expression (relative to SubQ → SubQ grafts) are listed. Genes increased in both SubQ → Vis grafts and the endogenous inguinal SAT of SubQ → Vis mice (SubQ → Vis ING) are highlighted in green, including apredicted targets of MEF2A. (c) Top ten sets of genes over-represented in SubQ → Vis grafts vs SubQ → SubQ grafts. (d) Top ten sets of genes over-represented in endogenous inguinal AT of SubQ → Vis vs Sham mice. (e) MEF2 subtype expression in transplanted and endogenous AT. MEF2C and MEF2D were selectively expressed in the endogenous inguinal depot of SubQ → Vis mice (q < 0.005 vs SubQ → Vis EPI or Sham ING). (f) Adipocyte size in the endogenous inguinal AT of Sham, SubQ → Vis and Vis → Vis mice. Sham, SubQ → Vis and Vis → Vis mice are indicated by black circles, white circles and black squares, respectively (n = 7–9 per group), p < 0.0001, χ 2 test

Remarkably, expression of many of these genes was also induced in the endogenous inguinal AT of SubQ → Vis mice, relative to Sham-operated mice (Fig. 4b). GSEA of endogenous inguinal AT revealed enrichment for genes related to RNA processing and immune responses (Fig. 4d). Mef2c and Mef2d were selectively induced in endogenous inguinal AT of SubQ → Vis mice (Fig. 4e); and SubQ → Vis mice had smaller adipocytes in this depot (Fig. 4f).

SubQ → Vis transplantation protects against HFD-induced hepatic triacylglycerol accumulation

AT transplantation has been previously reported to reduce HFD-induced hepatic triacylglycerol accumulation [21, 22]. At 17 weeks, hepatic triacylglycerol content was reduced in SubQ → Vis mice, relative to Vis → Vis mice, but no differences were observed at 6 weeks, when glucose tolerance was first improved (Fig. 5a). Hepatic glycogen content was unaffected (Fig. 5b).

Fig. 5
figure 5

SubQ → Vis transplantation protects mice against HFD-induced hepatic triacylglycerol accumulation. (a, b) Hepatic triacylglycerol (a) and glycogen (b) content at 3, 6 and 17 weeks after transplantation; p < 0.05 vs Vis → Vis. (c, d) Immunoblots for hepatic SREBP1 (c) and hepatic SCD1 (d) levels at 17 weeks after transplantation. (e, f) Relative levels of SREBP1 (e) and SCD1 (f); p = 0.12 (e) and p = 0.24 (f), ANOVA. (g) Immunoblot for hepatic FAS, ACC and ACSL1 levels at 17 weeks after transplantation. (h) Relative FAS, ACC and ACSL1 levels; p = 0.33, 0.32 and 0.70 for differences in FAS, ACC and ACSL1 content, ANOVA. (i, j) Quadriceps muscle triacylglycerol (i) and glycogen (j) content at 3, 6 and 17 weeks after transplantation; p < 0.05 vs Vis → Vis. In all cases, Sham (n = 8–10/group), SubQ → Vis (n = 6–8/group) and Vis → Vis (n = 8–10/group) mice are indicated by black, white and striped columns, respectively. AU, arbitrary units

Next, we measured the levels of the insulin-regulated lipogenic factor SREBP1 and its target genes (Fig. 5c–h). SREBP1 content and that of its target SCD1 tended to be highest in Vis → Vis mice (Fig. 5e, f), but the differences were not statistically significant. Levels of other SREBP targets, FAS, ACC and ACSL1, were not altered (Fig. 5g, h). Quadriceps muscle triacylglycerol content was significantly elevated in SubQ → Vis mice at 6 weeks after transplantation, but not at 17 weeks (Fig. 5i). Transplantation did not affect muscle glycogen content (Fig. 5j).

We also investigated gene expression using PCR arrays (ESM Fig. 4). Stat3 mRNA was elevated in livers from SubQ → Vis mice compared with Vis → Vis mice (p < 0.05), and tended to be higher than in livers from Sham mice (p = 0.056). Relative to Sham mice, Socs3, Adipor2, Lpl and Foxa2 mRNA were all significantly downregulated in livers from Vis → Vis mice. In human hepatocarcinoma cells, ADIPOR2 expression may be negatively regulated by SREBP1 [31].

SubQ → Vis transplantation reduced circulating markers of inflammation

Plasma cytokine profiling was also performed at 4 and 10 weeks after transplantation (Fig. 6 & ESM Tables 2 and 3). SubQ → Vis transplantation significantly attenuated HFD-induced increases in plasma concentrations of TNF-α, IL-6 and IL-17 (Fig. 6a–c). Plasma IL-6 concentrations were uniquely elevated in Vis → Vis mice at 10 weeks after transplantation (Fig. 6b). HFD feeding is associated with increased plasma concentrations of IL-12p70 and the chemokines monocyte chemotactic protein-1/chemokine (C-C) motif) ligand 5 (MCP-1/CCL5) and macrophage inflammatory protein-1 beta/chemokine (C-C) ligand 4 (MIP-1β/CCL4) [3234]; all were significantly reduced in SubQ → Vis mice relative to Sham mice (Fig. 6d–f). Concentrations of the Th1 chemokine RANTES/CCL5 (regulated on activation, normal T cell expressed and secreted/chemokine (C-C) ligand 5) were suppressed in SubQ → Vis and Vis → Vis mice compared with Sham mice (Fig. 6g).

Fig. 6
figure 6

(al) Plasma cytokine concentrations at 4 and 10 weeks post-transplantation. Cytokines displaying significant differences among groups at either 4 or 10 weeks are shown. Each graph shows the mean ± SEM plasma cytokine concentration in chow (vertical striped bars), HFD-fed Sham (black bars), HFD-fed SubQ → Vis (white bars) and HFD-fed Vis → Vis (diagonally striped bars) mice. The remainder are shown in ESM Table 3. *p < 0.05 vs Sham; p < 0.05 vs Vis → Vis

Th2 cells promote anti-inflammatory (M2) macrophage polarisation via the production of IL-4, IL-10 or IL-13 [35]. Plasma IL-10 and IL-4 concentrations were lower in SubQ → Vis mice than in Sham mice at 10 weeks (Fig. 6h, i). Transplantation of both types of AT suppressed plasma concentrations of IL-13, and those of the Th1 cytokines IL-5 and IL-2 (Fig. 6j–l).

Next, we examined relationships between the above cytokines and either the response to a GTT (iAUC GTT) or fasting insulinaemia at 13 weeks in chow or HFD-fed Sham, SubQ → Vis and Vis → Vis mice (n = 40). All cytokines in Fig. 6, except IL-6, predicted fasting insulinaemia (Fig. 7a); however, only TNF-α and IL-17 concentrations predicted both iAUC GTT and fasting insulinaemia (Fig. 7a–c). None of the cytokines measured at 4 weeks significantly predicted subsequent changes in glucose tolerance (data not shown).

Fig. 7
figure 7

Relationships between plasma cytokine concentrations, glucose tolerance and insulinaemia. (a) Correlations between individual cytokines at 10 weeks post-transplantation and glucose tolerance or fasting insulinaemia at 13 weeks; N = 40 mice (10 chow, 12 HFD Sham, 8 HFD SubQ → Vis and 10 HFD Vis → Vis). (b) Correlation between plasma TNF-α concentrations and iAUC GTT. Chow (white squares, n = 10), HFD Sham (black circles, n = 12), HFD SubQ → Vis (white circles, n = 8) and HFD Vis → Vis (black squares, n = 10) mice are shown; Pearson r = 0.34, p = 0.033. (c) Correlation between plasma IL-17 concentrations and iAUC GTT; symbols are as for Fig. 7b; Spearman r = 0.34, p = 0.032. (d) Tnfa expression in grafted and endogenous AT by microarray. Normalised gene expression is provided in arbitrary units. (ej) Expression of Il6 (e); Il17 (f); Il12a (g); Ccl2 (h); Ccl4 (i); and Ccl5 (j); **p < 0.01 for comparison

Finally, to determine whether changes in AT gene expression influenced systemic cytokine concentrations, we examined the mRNA expression of Tnfa, Il6, Il17, Il12a, Ccl2, Ccl4 and Ccl5 in endogenous and grafted AT from our microarray data. Differences in plasma cytokine concentrations were not reflected in AT gene expression (Fig. 7d–j), and staining of SubQ → Vis and Vis → Vis grafts for F4/80, a marker of AT macrophages [36], did not reveal marked differences in macrophage content (ESM Fig. 5).

Discussion

Transplantation of inguinal, but not epididymal, AT uniquely and reproducibly protects mice against HFD-induced glucose intolerance and may reduce adiposity [19, 20]. Here, SubQ → Vis transplantation improved glucose tolerance at 6 weeks, and was not explained by differences in adiposity, leptin, adiponectin or tissue-specific glucose uptake. Instead, SubQ → Vis transplantation reduced hepatic triacylglycerol accumulation and normalised gluconeogenic enzyme expression. Reduced hepatic triacylglycerol content in SubQ → Vis mice was not present at 6 weeks, however, suggesting that this occurs subsequent to improved glucose tolerance. SubQ → Vis transplantation also suppressed HFD-induced systemic inflammation: plasma concentrations of TNF-α, IL-17, IL-12p70, MCP-1/CCL2 and MIP-1β/CCL4 were all significantly reduced in SubQ → Vis mice, relative to Sham mice. Plasma IL-17 and MIP-1β concentrations were reduced in SubQ → Vis mice as early as 4 weeks post-transplantation, possibly preceding improvements in glucose tolerance.

This study is the first to show anti-inflammatory effects of SubQ → Vis transplantation, although previous studies have suggested that Vis → Vis transplantation has proinflammatory effects [20, 3739]. Immune cell composition differs between AT depots and is highly influenced by obesity [40]. In C57BL6/J mice, epididymal fat is skewed towards innate immunity, while the inguinal depot contains higher proportions of adaptive immune cells (CD4+ and CD8+ T cells and B cells) [40]. Obesity is characterised by an increase in proinflammatory Th1 cells, particularly in VAT [35]. SubQ → Vis transplantation likely prevents HFD-induced glucose intolerance and IR by altering the balance of these cell types specifically in the intra-abdominal compartment, as SubQ → SubQ transplantation does not affect glucose tolerance [19, 20]. While further studies will be required to fully understand how this procedure affects immune cell populations, it will be important to consider migration of cells from donor AT to other tissues, including liver and spleen, following transplantation [41]. We were unable to relate differences in plasma cytokine concentrations to mRNA expression in grafted or endogenous AT (Fig. 7).

The early differences in IL-17 concentrations, and their relationship with glucose intolerance and insulinaemia, support a role for T helper 17 (Th17) cells in the pathogenesis of HFD-induced IR. Th17 populations expand in obesity and are a major source of proinflammatory IL-17 [42]. Th17 expansion depends on IL-6 and leptin in the AT microenvironment [43]. Genetic deletion [44] or antibody-mediated neutralisation [45] of IL-17 prevents IR, glucose intolerance and liver injury in HFD-fed mice. Human studies also implicate IL-17 in the pathogenesis of IR [46, 47]. IL-17 may link AT inflammation with hepatic IR, as pre-treating human hepatocytes with IL-17 inhibited insulin-stimulated Akt phosphorylation and suppression of gluconeogenesis [16].

Beneficial effects of Vis → Vis transplantation (with caval drainage, not portal) on glucose tolerance have been reported at 4–8 weeks post-transplantation [18, 21, 39]. In the short term (≤6 weeks), this additional VAT may reduce endogenous adiposity (Table 1), and protect against lipotoxicity (Fig. 5); but these effects are not sustained. Relative to Sham mice, Vis → Vis transplantation reduces HFD-induced hyperinsulinaemia, elevates plasma IL-6 concentrations, but does not prevent glucose intolerance or hepatic triacylglycerol accumulation.

Subcutaneous AT transplantation reproducibly prevents HFD-induced glucose intolerance in mice. While the mechanisms are incompletely understood, we have identified sustained anti-inflammatory effects of this procedure that predicted metabolic improvements. Future studies will be required to identify the cell type(s) responsible, and potential sites of intervention for the treatment and prevention of diet-induced glucose intolerance and IR.