Potential Role of Regulatory T Cells in Reversing Obesity-Linked Insulin Resistance and Diabetic Nephropathy

The study included 79 obese patients undergoing weight-reducing surgery and 14 lean patients who underwent elective cholecystectomy. Lean patients with elevated C-reactive protein were excluded from the study. Participants were included if they had fasting blood glucose levels <7.0 mmol/L, no history of diabetes or use of glucose- or lipid-lowering drugs, and no weight changes >3% during the previous 2 months. All study participants provided informed consent, and study protocols were approved by the local ethics committee. BMI, fasting blood glucose, and insulin were determined as described. Adipose tissue biopsies were collected in RNAlater (Ambion, Austin, TX) and stored at −80°C until RNA isolation (22).

Five-week-old male db/db leptin receptor–deficient mice (BKS.Cg-Dock7^m +/+ Lepr ^db/J) were obtained from Charles River (Sulzfeld, Germany) and maintained in a virus/antibody-free central animal facility of the Innsbruck Medical University. At 6 weeks of age, the animals underwent surgical uninephrectomy. Body weight, fasting glucose levels, and urinary albumin excretion were measured regularly once a week, and 24-h urine samples were collected in metabolic cages. Mice were treated with either 200 μg of an anti-CD25 monoclonal antibody (mAb; clone PC61; BD Biosciences, San Diego, CA) or an isotype control antibody administered intraperitoneally the day after uninephrectomy and 28 days thereafter. In a second set of experiments, db/db mice were treated intravenously with either 0.5–2 × 10⁶ CD4⁺FoxP3⁺GFP⁺ or CD4⁺GFP⁻ cells on days 4, 18, 32, and 46 after uninephrectomy. Adoptively transferred cells were isolated from pooled spleen and lymph node suspensions (minced and passed through a 70-μm mesh) of B6.Cg-Foxp3^tm2Tch/J reporter mice (Jackson Laboratories, Bar Harbor, ME). After staining for CD4 (BD Biosciences), CD4⁺FoxP3⁺GFP⁺ Tregs were purified by fluorescence-activated cell sorter (FACS) using a BD FACSAria cell sorter (BD Biosciences). A purity of 99.14 ± 0.43% was achieved.

All animal experiments were approved by Austrian veterinary authorities and corresponded to the European Commission Directive 86/609/EEC.

Single cell suspensions from liver, spleen, mVAT, and murine subcutaneous adipose tissue (mSAT) were costained for surface CD4, CD8, and CD69 (all BD Biosciences) and intracellular FoxP3⁺ (eBiosciences, San Diego, CA) strictly adhering to the manufacturer’s instructions. Data collection and analysis were done on a FACSCalibur (BD Biosciences).

Blood glucose levels were measured using OneTouch Ultra2 (LifeScan, Vienna, Austria), urine creatinine was quantified spectrophotometrically using a Picric acid–based method (Sigma, St. Louis, MO), and urinary albumin was determined by a double-sandwich ELISA (Abcam, Cambridge, MA) as reported previously (23). The insulin tolerance test was performed on day 56 after uninephrectomy. In brief, we injected 0.75 IU/kg body wt (in the Treg-depletion experiments) and 1.5 IU/kg body wt (in the Treg-treatment experiments) of human insulin aspartat (Novo Nordisk, Sorgenfri, Denmark) and measured the blood glucose level prior to and 15, 30, 45, 60, 75, and 90 min after insulin application. On the same day, we also measured fasting insulin levels (Crystal Chem, Downers Grove, IL) by ELISA. The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated by using the glucose-insulin product divided by the constant 22.5. Furthermore, we measured the fat cell diameter in mSAT and mVAT (epididymal) and the glomerular diameter at a magnification of 100 using a ProgRes C12 camera and the ProgRes Capture Pro 2.5 software with the straight line tool (Jenoptik GmbH, Jena, Germany). At least 200 adipocytes from both mSAT and mVAT or 50 glomeruli were evaluated for each mouse in a blinded fashion.

Formalin-fixed renal tissue was embedded in paraffin, cut in 4-μm sections, and stained with hematoxylin-eosin and Periodic acid-Schiff using standard staining techniques. The three-layer immunoperoxidase technique was used for the detection of macrophages and T cells in frozen kidney sections. Macrophages were stained using rat anti-mouse F4/80 and CD68 antibodies (Serotec, Oxford, U.K.). A semiquantitative scoring system was performed as follows: 0, 0 to 4 cells; 1+, 5 to 10 cells; 2+, 10 to 50 cells; 3+, 50 to 200 cells; and 4+, >200 cells stained positive per low-power field. For the detection of CD4⁺ and CD8⁺ T cells, rat anti-mouse CD4 and CD8-α mAb (Serotec) were used. Biotin-conjugated goat anti-rat IgG antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) was used as a secondary antibody, followed by incubation with a streptavidin-biotin complex and subsequent development with 0.4% 3-amino-9-ethylcarbazole for 6 min and counterstaining with Gill's Hematoxylin No. 3 (Polysciences, Warrington, PA). Quantification of T cells was done by counting the number of cells in six adjacent high-power fields of renal cortex and medulla. Scoring of cells was performed by a blinded observer.

Human and murine tissue was stored at −80°C in RNAlater. Total RNA was isolated from the respective tissue using Trizol (Sigma) according to a standard protocol. The integrity of RNA was detected by RNA-gel electrophoresis in stratified random RNA samples. Thereafter, 2 μg of total RNA was reverse transcribed using Superscript III Transcription Kit (Nitrogen, Carlsbad, CA) and random primers (Roche, Basel, Switzerland). Real-time PCR was performed on a StepOnePlus real-time instrument (Applied Biosystems, Foster City, CA) or a CFX96 Real-Time System (Biorad, Hercules, CA). For amplification of murine FoxP3 and TNF-α, SYBR Green Master Mix (Nitrogen) and the following primers were used: Foxp3 forward 5′-TCT TGC CAA GCT GGA AGA CT-3′, reverse 5′-AGC TGA TGC ATG AAG TGT GG-3′; and TNF-α forward 5′-GAA CTG GCA GAA GAG GCA CT-3′, reverse 5′-AGG GTC TGG GCC ATA GAA CT-3′. The chosen primer pairs span the exon-intron border. The Foxp3 primer pair detects all murine Foxp3 splicing variants. For quantification of human FOXP3, Helios and TNF-α Taqman Master mix (Applied Biosystems) and the gene expression assays Hs01085834_m1, Hs00212361_m1, and Hs00174128_m1 (Applied Biosystems) were used. The human FOXP3 gene expression assay detects all human FOXP3 splicing variants. For quantification of murine IL-6, interferon-γ (IFN-γ), Gata-3, IL-10, the gene expression assays Mm00446190_m1, Mm00901778_m1, Mm00484683_m1, and Mm00439616_m1 (Applied Biosystems) were used. In murine samples, 18S mRNA served as reference. For the human studies, we evaluated the mean expression of three different housekeeping genes, namely, 18S, β-actin, and HPRT-1, which was then set as reference. We tested for intra- and interassay variance in our real-time PCR experiments. The coefficient of variation for the intra-assay variance was <0.0005 and for the interassay variance, <0.001.

Enhanced green fluorescent protein–positive cells were detected using the following primers: forward 5′-AAG TTC ATC TGC ACC ACC G-3′, reverse 5′-TCC TTG AAG AAG ATG GTG CG-3′. Nested PCR was performed with the following primer pair: 5′-GTG TCC GGC GAG GGC GAG G-3′, reverse 5′-CGT GCT GCT TCA TGT GGT CG-3′. cDNA isolated from spleens of B6.Cg-Foxp3^tm2Tch/J served as positive control. RT-PCR was carried out using HotStarTaq Polymerase (Qiagen, Hilden, Germany).

Human data were analyzed for normal distribution using Kolmogorov-Smirnov test with Lilliefors correction. Since the data were not normally distributed, the bivariate associations between Helios/FOXP3 mRNA levels, BMI, fasting glucose levels, and HOMA-IR were calculated using the Spearman correlation coefficient. In a case of comparing three groups, the significance level was lowered to P < 0.025.

Results from animal experiments are presented as mean ± SEM. Normal distribution of data were assessed by the Kolmogorov-Smirnov test with Lilliefors correction. Both groups were compared by either nonparametric Mann-Whitney U test or unpaired t test as appropriate, depending on the distribution of the tested variable. A two-tailed P < 0.05 was considered statistically significant. For the insulin-sensitivity assay in Treg-depleted mice versus controls, the Fisher exact test was performed. All statistical analyses were done with SPSS 15.0 for Windows (SPSS, Chicago, IL).

To test whether Tregs accumulate in visceral obesity, we analyzed hVAT from 14 lean control subjects, 39 obese patients without insulin resistance (HOMA-IR <2.5), and 39 obese patients with insulin resistance (HOMA-IR >5). We evaluated the mRNA expression of the thymic Treg marker Helios as well as of the master regulator of Tregs FOXP3 in hVAT. We found the mRNA expression of Helios to be downregulated in both obese patient groups, whereas FOXP3 was significantly downregulated only in obese patients without insulin resistance. In obese patients with insulin resistance, mRNA expression of FOXP3 was similar to lean control subjects (Fig. 1A). The Helios-to-FOXP3 ratio correlated inversely with BMI (Fig. 1B), HOMA-IR (Fig. 1C), and fasting glucose levels (Fig. 1D). Significance was reached in the correlation with fasting blood glucose level and HOMA-IR.

Since Treg lineage plasticity is influenced by insulin resistance, we next studied their pathogenic role in a murine model of type 2 diabetes. We therefore used a well-established model of murine type 2 diabetes (i.e., the leptin-deficient db/db mice). Tregs were eliminated in these mice using the Treg-depleting anti-CD25 antibody, which was injected intraperitoneally on days 1 and 28 after uninephrectomy. We chose this application schedule since we could not detect any CD4⁺CD25⁺ T cells in the inguinal lymph nodes 3 weeks after application of a single dose of the antibody (data not shown). Treg-depleted mice and the respective controls had a comparable body weight throughout the observation period of 56 days (Fig. 2A); however, fasting blood glucose levels were significantly increased as early as 21 days after uninephrectomy in Treg-depleted db/db mice when compared with vehicle-treated db/db mice (Fig. 2B). On day 56, insulin-sensitivity testing revealed that insulin lowered blood glucose levels more efficiently in vehicle-treated db/db mice compared with Treg-depleted db/db mice (Table 1). In addition, Treg-depleted db/db mice displayed a significantly increased HOMA-IR on day 56 as compared with control animals (Fig. 2C). To analyze whether these metabolic differences were also reflected by changes in the inflammatory profile within the mVAT and mSAT compartment, real-time PCR for mRNA expression of IFN-γ, IL-6, TNF-α, Gata-3, and IL-10 was performed. We could detect a prominent increase of the Th1 cytokines IFN-γ, IL-6, and TNF-α in mVAT of Treg-depleted db/db mice as compared with control animals (Fig. 3A). No difference in the mRNA expression of Gata-3 and IL-10 was detected in mVAT (Fig. 3A). Furthermore, we did not detect any difference in the expression of the respective transcripts in mSAT (Fig. 3B). In parallel, flow cytometry revealed increased numbers of activated CD4⁺CD69⁺ T cells in the mVAT of Treg-depleted mice (Fig. 3C and D), while no difference in the abundance of this cell population was found in mSAT (Fig. 3C). No difference in the percentage of CD8⁺ T cells and CD11c⁺ dendritic cells in mVAT was detected (data not shown).

Of note, Treg depletion by anti-CD25 antibody did not compromise endogenous insulin production since mice that received the CD25-depleting antibody had even slightly higher insulin levels as compared with control mice (control: 3.63 ± 0.62 ng/mL; Treg depleted: 3.93 ± 0.67 ng/mL).

The urinary albumin-to-creatinine ratio as a marker of renal impairment was increased in Treg-depleted versus nondepleted db/db mice starting on day 28 and reached statistical significance on day 42 after uninephrectomy (Fig. 4A). Accordingly, the glomerular diameter was found to be significantly increased in Treg-depleted mice on day 56 after uninephrectomy (Fig. 4B). Semiquantitative assessment of renal inflammatory cell infiltration revealed a profound increase in CD4⁺ and CD8⁺ T cells in the kidneys of Treg-depleted db/db mice (Fig. 4C), whereas F4/80⁺ cells and CD68-expressing macrophages were not different (Fig. 4D). Real-time analysis revealed a significantly increased mRNA expression of the Th1 cytokine TNF-α in the kidneys of Treg-depleted db/db mice (Fig. 4E). IFN-γ and IL-10 tended to be increased in Treg-depleted mice, whereas IL-6 and Gata-3 were not differentially regulated (Fig. 4E).

Db/db mice underwent uninephrectomy at the age of 6 weeks. On days 4, 18, 32, and 46 after uninephrectomy, CD4⁺FoxP3⁺GFP⁺ (Tregs) or CD4⁺FoxP3⁻GFP⁻ (control) T cells isolated from healthy donor animals were adoptively transferred into db/db mice. No difference in the body weight was observed between the two groups (Fig. 5A). The blood glucose levels were significantly decreased only in the Treg-treated mice on day 28, whereas on the other evaluated time points, no difference between the groups was detectable (Fig. 5B). Moreover, significantly improved insulin sensitivity was observed in db/db mice treated with Tregs compared with db/db mice injected with control T cells (Fig. 5C). A trend toward an improved HOMA-IR in Treg-treated obese animals was detected, but no statistical significance was reached (Fig. 5D). In addition, Treg-treated db/db mice displayed a significantly decreased mVAT cell diameter as compared with controls (Fig. 6A), whereas we could not detect any difference in the mSAT cell diameters (data not shown). Treg transfer increased the expression of the Treg marker FoxP3 and the Th1 cytokine IFN-γ in mVAT, whereas IL-6, TNF-α, Gata-3, and IL-10 mRNA were not differentially regulated (Fig. 6B). It is interesting that the percentage of proinflammatory mVAT-infiltrating CD8⁺CD69⁺ effector T cells was significantly decreased in Treg-treated db/db mice (Fig. 6C and D), whereas no difference of these cells could be detected in mSAT (Fig. 6C). Of note, no difference in the mVAT and mSAT infiltration of CD4⁺CD69⁺ T cells was observed (Fig. 6C).

Adoptive transfer of Treg significantly decreased urinary albumin-to-creatinine ratio in db/db mice when compared with respective controls on days 42 and 56 after uninephrectomy (Fig. 7A). In addition, the glomerular diameter, which reflects hyperfiltration, was also significantly decreased in kidneys of Treg-treated mice (Fig. 7B). In line with data from mVAT T-cell infiltration, CD8⁺ T cells were significantly less abundant in the kidneys of Treg-injected db/db mice (Fig. 7C), whereas the infiltration of CD4⁺ T cells and macrophages was not affected (data not shown). Despite a trend toward higher expression of the proinflammatory Th1 markers IFN-γ, IL-6, and TNF-α in kidneys of Treg-treated db/db mice, there was a predominant and statistically significant effect on the expression of the anti-inflammatory markers Gata-3, IL-10, and FoxP3 (Fig. 7D). Of note, we were not able to detect adoptively transferred GFP⁺ Tregs in harvested spleens, livers, kidneys, mSAT, and mVAT of uninephrectomized db/db mice on day 56 (i.e., 14 days after the final Treg-transfer) using either flow cytometry or a highly sensitive PCR for GFP (data not shown).

The pandemic of type 2 diabetes is one of the major health burdens in the industrial world today, and diabetic nephropathy due to type 2 diabetes has become the leading cause of end stage renal disease. Thus, new therapeutic options are eagerly awaited. Insulin resistance and type 2 diabetes are driven by chronic inflammation (24) where not only macrophages but also T cells are critically involved (24). Here, we present evidence that Tregs are key regulatory cells in the pathogenesis of insulin resistance and that they have the capacity to improve insulin resistance and diabetic nephropathy when transferred in vivo.

It is long known that macrophages accumulate in the mVAT of obese mice (25), but recently Winer et al. (15) found that T cells drive macrophage infiltration. By blocking T cells, they detected significantly improved insulin resistance in their murine type 2 diabetes model. In a similar approach, Ilan et al. (16) showed that the ratio between TH1 effector cells and Tregs changed toward Tregs as a result of oral administration of anti-CD3 and β-glucosylceramide. The importance of T cells and Tregs in insulin resistance was suggested by observational human studies. Nevertheless, published human data on Tregs in hVAT have been, thus far, conflicting (17–19). While Zeyda et al. (19) found T cells together with Tregs to accumulate in hVAT of obese patients, others detected decreased infiltrating Treg numbers (17,18). Our patient population consisted of lean, obese insulin-sensitive, and obese insulin-resistant patients. We detected decreased FOXP3 mRNA signals only in obese insulin-sensitive patients as compared with lean control subjects. Obese patients with insulin resistance had similar FOXP3 mRNA expression as lean control subjects. Most interesting, Helios—a new marker for thymic-derived natural Tregs (13)—differed significantly from our FOXP3 data, since Helios expression was significantly lowered in hVAT of all obese patients irrespective of their insulin resistance. Our data might thus explain the differences observed in the previous human studies. Obese patients seem to have decreased thymic-derived Tregs in their hVAT. Moreover, peripherally induced adaptive FoxP3⁺ Tregs seem to accumulate in the inflammatory microenvironment of hVAT in obese patients with overt insulin resistance. Limitations for these suggestions come from the technique used throughout our studies since we had no access to fresh hVAT to perform FoxP3/Helios double staining. Since FoxP3⁺ cell populations have also been shown to produce IL-17 or IFN-γ (26,27), we cannot exclude that our FOXP3 mRNA expression reflects other CD4⁺ T cells infiltrating hVAT of obese patients with insulin resistance. To gain further insights in the functional role of Treg in insulin resistance, we used db/db mice, which lack signaling capacity of the leptin receptor and therefore develop hyperphagia, diet-induced obesity, and type 2 diabetes (20). After treatment with a CD25-depletion antibody, which has been used for Treg depletion in various disease models (28–30), db/db mice displayed augmented insulin resistance and a more severe diabetic nephropathy. In line with recent reports (15,17), we detected significantly increased TH1 cytokines and activated T cells in the mVAT of Treg-depleted db/db mice, whereas no changes were detected in mSAT. These data strongly support the idea that VAT but not SAT is crucial for impaired insulin sensitivity. Independent studies in lean mice demonstrate that Treg depletion induced by a genetic approach aggravated insulin resistance (17). Crossing adipose mouse strains with FoxP3-DTR mice is a difficult issue because of infertility of homozygous db/db mice. Therefore, we chose to transfer sorted FoxP3⁺ Tregs into db/db mice.

We provide the first direct evidence that FoxP3 Tregs have the capacity to improve insulin resistance. Transfer of CD4⁺FoxP3⁺ cells induces Treg infiltration and/or FoxP3 upregulation in mVAT. The increased Treg activity inhibited CD8⁺ T-cell infiltration into mVAT. CD8⁺ T cells can be divided in different subpopulations. Naïve T cells provide the potential to respond to new pathogens, whereas central-memory and effector-memory T cells can mount rapid and efficient secondary proliferate and effector responses to recall antigens (31). In our experimental setup, CD8⁺ T cells infiltrating mVAT mainly expressed the T-cell activation marker CD69. Since T-cell activation markers such as CD69 and CD25 are expressed only on a small fraction of the two memory subsets (32), it can be proposed that Treg transfer decreased the mVAT infiltration of effector CD8⁺ T cells, which originated from naïve CD8⁺ T cells. This is further supported by the recent observation that Tregs dramatically decrease the expansion and differentiation of naïve CD8⁺ T cell precursors in an experimental murine model of repopulation of lymphopenic environment (33). CD8⁺ effector T cells are involved in the pathogenesis of obesity and infiltrate adipose tissue (34). Nishimura et al. (34) provided evidence that CD8⁺ T cells lead to the recruitment of macrophages, thereby mediating insulin resistance. In contrast, Winer et al. (15) found CD8⁺ T cells did not influence diet-induced obesity in lymphocyte-free Rag1-null mice. In our murine model, transfer of CD4⁺FoxP3⁺ Tregs not only improved insulin resistance but also resulted in less end organ damage, such as diabetic nephropathy. In the kidney, we observed a shift toward a TH2 response, reflected by increased mRNA expression of Gata-3 and IL-10, as well as significantly increased Treg numbers. Thus, IL-10 produced by Tregs might be of crucial importance for the Treg-mediated protective effect, but further studies using Tregs from IL-10 knockout mice are necessary. The role of IL-10 needs further evaluation since renal IL-10 mRNA expression also tended to be increased in kidneys of Treg-depleted mice. Again, we found significantly decreased numbers of CD8⁺ T cells to infiltrate the kidneys, whereas no change in the number of renal CD4⁺ T cells was noted. It is interesting that no significant difference in the macrophage infiltration of kidneys was detected. It has been proposed that CD8⁺ T cells improve insulin resistance by recruiting macrophages (34), which seems not to be the case in the pathogenesis of diabetic nephropathy. Further studies depleting CD8⁺ T cells in diabetic nephropathy are needed to shed more light on the role of CD8⁺ T cells in diabetic nephropathy. It might be further speculated that Tregs, independently from improving insulin resistance, improve diabetic nephropathy by exerting their anti-inflammatory capacities, but our study design does not allow discrimination between the two mechanisms.

By transferring FoxP3⁺-GFP⁺ Tregs, we transferred a selective fluorescent Treg population. Contrary to the murine model of nephrotoxic nephritis (2), we failed to track these fluorescent Tregs by using molecular genetic methods and flow cytometry. The absence of FoxP3⁺-GFP⁺ Tregs in mVAT, liver, kidney, and spleen on day 56 is probably due to the long interval of 14 days after the final transfer. It seems that at the tested time point, all transferred Tregs were already apoptotic and that they had previously, in their short lifetime, primed the immune system toward an anti-inflammatory milieu.

Taken together, we prove that transfer of CD4⁺FoxP3⁺ Tregs improves insulin resistance and diabetic nephropathy by tipping the balance toward anti-inflammation and inhibiting CD8⁺ effector cell infiltration in VAT and kidneys. Our results confirm the importance of Tregs in the pathogenesis of insulin resistance and might lead to new therapeutic concepts of type 2 diabetes.

W.P. was supported by the Wiener Wissenschafts-, Forschungs- und Technologiefonds (Project LS07-058). A.R.R. was supported by the Austrian Research Fund (P-21402).

No potential conflicts of interest relevant to this article were reported.

K.E. researched data and wrote the manuscript. A.K. researched data. A.M.W. and S.S. researched data and contributed to discussion. A.T. and U.S. researched data. D.W., W.P., and A.R.R. contributed to discussion and wrote the manuscript. P.E. researched data and wrote the manuscript.