Inhibition of islet immunoreactivity by adiponectin is attenuated in human type 1 diabetes.

CONTEXT
Adiponectin is an adipocyte-derived cytokine with insulin-sensitizing and antiinflammatory properties. These dual actions have not previously been examined in the context of human disease.


OBJECTIVES
Our objective was to examine the adiponectin axis in type 1 diabetes (T1D). T1D is an autoimmune inflammatory disease resulting from pancreatic β-cell destruction, in which insulin resistance associates with progression to disease.


DESIGN, PATIENTS, AND INTERVENTIONS
We measured circulating adiponectin and adiponectin receptor expression on blood-immune cells from 108 matched healthy, T1D, and type 2 diabetic subjects. We tested adiponectin effect on T cell proliferation to islet antigens and antigen-presenting function of monocyte-derived dendritic cells (mDCs). Lastly, we assessed the effect of a 3-week lifestyle intervention program on immune cell adiponectin receptor expression in 18 healthy subjects.


RESULTS
Circulating concentrations of adiponectin were not affected by T1D. However, expression of adiponectin receptors on blood monocytes was markedly reduced and inversely associated with insulin resistance. Reduced adiponectin receptor expression resulted in increased T cell proliferation to islet-antigen presented by autologous mDCs. We demonstrated a critical role for adiponectin in down-regulating the costimulatory molecule CD86 on mDCs, and this function was impaired in T1D. We proceeded to show that lifestyle intervention increased adiponectin receptor but reduced CD86 expression on monocytes.


CONCLUSIONS
These data indicate that T cells are released from the antiinflammatory effects of adiponectin in T1D and suggest a mechanism linking insulin resistance and islet immunity. Furthermore, we suggest that interventions that reduce insulin resistance could modulate the inflammatory process in T1D.

A diponectin is the most abundant serum adipokine (1,2). Adiponectin binds to at least 2 receptors, Adi-poR1 and AdipoR2, which signal via ceramide and AMP kinase (3,4). These receptors have been described on a variety of human tissues, including peripheral blood mononuclear cells (PBMCs) (5).
In contrast to most adipokines, adiponectin is distinguished by its insulin-sensitizing functions. Obesity and insulin resistance-related diseases such as type 2 diabetes (T2D) (6) and coronary artery disease (7) are associated with low levels of total adiponectin, and overexpression of adiponectin in animal models of these diseases confers protection (8,9).
Adiponectin is also distinguished by its antiinflammatory functions. Many of these functions have been demonstrated on cells of monocyte lineage. Chief among these are a change in the balance of proto-antiinflammatory cytokine production (10,11) and a reduction in maturation, migration, and phagocytosis (12,13). Recently adiponectin has also been demonstrated to affect the phenotype and function of murine monocyte-derived dendritic cells (DCs) (14). Here, adiponectin-treated DCs expressed lower levels of costimulatory molecules and showed less allostimulatory capacity when cocultured with T cells.
Although informative, these studies do not illustrate how the adiponectin pathway is modulated in human immunity, or how its antiinflammatory and unique insulinsensitizing functions could interact in the context of human disease. We chose to examine these features of adiponectin in the context of type 1 diabetes (T1D). T1D is characterized by T cell-mediated destruction of insulinproducing pancreatic ␤-cells. It is a disease that requires professional antigen presentation by DCs, cells of myeloid lineage that are recognized to be adiponectin responsive (15,16). It is also a disease in which insulin resistance associates with a more rapid onset of disease in genetically at-risk subjects (17)(18)(19).
We show that although circulating levels of adiponectin were unchanged in T1D, there was a striking reduction in adiponectin receptor expression by peripheral blood monocytes. We used serum-free conditions to demonstrate that reduced receptor expression resulted in protection from the antiinflammatory effects of adiponectin and a net proinflammatory effect on islet immunity. To dissect the mechanism of adiponectin effect, we examined the influence of adiponectin on antigen-presenting cell (APC) function and in particular on DCs. We demonstrated that adiponectin down-regulated expression of the costimulatory molecule CD86 by DCs, and this down-regulation was associated with suppression of antigen-specific T cell proliferation. We went on to show that an increase in adiponectin receptor expression by lifestyle intervention modulated CD86 expression on monocytes in vivo, thus illustrating translational application for our findings.

Subjects
Male Caucasian subjects with T1D and T2D were recruited from diabetic outpatient clinics at the University Hospitals Birmingham National Health Service Foundation Trust (United Kingdom) after informed consent. Matched healthy volunteers, with confirmed normal glucose tolerance and on no regular medications, were recruited on the same day. Full ethical approval was obtained for this study. Patients with T1D had been diagnosed according to guidelines outlined by the American Diabetes Association. These included presentation with osmotic symptoms, weight loss, ketosis, and requirement for insulin within 3 months of diagnosis or presence of glutamic acid decarboxylase autoantibodies. Patients were presumed to have T2D if they did not have the above features, together with the absence of a family history of maturity-onset diabetes of the young, or a diagnosis of diabetes before the age of 40 years. They did not have associated endocrine autoimmune conditions and were all insulin treated. Specifically, insulin treatment was commenced more than 6 years after initial diagnosis, making the diagnosis of latent autoimmune diabetes of adulthood unlikely in our T2D cohort (20). Exclusion criteria include abnormal liver enzymes, microalbuminuria, poorly controlled diabetes [glycated hemoglobin (HbA1C) Ͼ 10%, 86 mmol/mol], acute illness and infection, heart failure, and use of thiazolidinediones [which independently increase AdipoR expression on monocytes (21)].

Measurement of adiponectin receptor expression of PBMCs by flow cytometry and quantitative PCR
The measurement of AdipoR1 and AdipoR2 has been described previously (5). Briefly, PBMCs were isolated from freshly drawn peripheral blood by density gradient centrifugation and preincubated with FcR blocker prior to staining with antibody. AdipoR1/2 antibodies (Phoenix Pharmaceuticals, Karlsruhe, Germany) were polyclonal, and their staining detected with a secondary goat antirabbit antibody (with preaddition of goat serum to avoid nonspecific binding). Specificity of AdipoR1/2 staining was confirmed by preincubating the antibodies with blocking peptides (Adipo R1/2; Phoenix Pharmaceuticals) before PBMC staining. Dead cells were excluded with propidium iodide staining. For each sample, at least 20 000 live cells were captured on flow cytometry (FACSCalibur; BD Biosciences, Abingdon, United Kingdom). The net mean fluorescence intensity (MFI) for AdipoR staining is the geometric mean fluorescence of the gated population, subtracted from fluorescence of the cells from the same subject stained with the secondary antibody only.
Quantitative RT-PCR was used to measure AdipoR1 and AdipoR2 gene expressions. Total RNA was extracted and reverse transcribed before analysis using previously published sequences of primers and probes for AdipoR1 and AdipoR2 (Eurogenetec, Southampton, United Kingdom), standardized against 18S RNA endogenous control.
Specimens from all subjects were processed on the same day as phlebotomy.

Serum adiponectin by ELISA
Serum was stored at Ϫ80°C until analysis. Serum adiponectin levels were measured using a quantikine human total adiponectin immunoassay (R&D Systems, Abingdon, United Kingdom).
(Hamberg, Germany). Dr Hilary Murray (The Islet Research Laboratory, Worcestershire Acute Hospitals National Health Service Trust, Worcestershire, United Kingdom) prepared human islets from cadaveric specimens. Four preparations, each containing at least 20 000 islet equivalents IEQ, were sonicated and mixed. The OKT3 antibody was a kind gift from Dr Dave Sansom (University of Birmingham, Birmingham, United Kingdom).

Serum-free PBMC proliferative assay
The 2 ϫ 10 5 carboxyfluorescein-succinimidyl-ester (CFSE)labeled (1.25 M) PBMCs in 100 L of serum-free Cellgro SCGM media (Cellgenix, Freiberg, Germany), supplemented with penicillin 100 U/mL Ϫ1 , streptomycin 100 U/mL Ϫ1 , and glutamine 2 mM, were cultured in round bottomed 96-well plates. Incubation conditions were controlled at 37°C and 5% humidified O 2 . Each condition was tested in triplicate. The antigens and recombinant adiponectin were added at the beginning of the culture. IL-2 at 10 Ϫ4 g/mL Ϫ1 was added to cultures stimulated by islet lysates. The cells were fed complete media on day 4 and read on day 8. Stimulation index (SI) was calculated as the ratio of percentage of live CFSE low cells in a stimulated condition over that of the unstimulated control (22).

Serum-free DC generation and culture
CD14 ϩ monocytes were positively selected from whole PB-MCs by magnetic sorting (Miltenyi, Church Lane Bisley, United Kingdom). The 5 ϫ 10 5 cells were cultured in 500 L of serumfree media Cellgro DC (Cellgenix), supplemented with 1 g of granulocyte macrophage colony-stimulating factor and IL-4 and refed on day 2. Adiponectin at 10 g/mL Ϫ1 was added at the beginning of culture. The DCs were harvested on day 6 and phenotyped with anti-CD80, CD86, human leukocyte antigen (HLA)-DR, dendritic cell-specific intercellular adhesion mole-cule-3-grabbing non-integrin, CD1a, CD11c, and CD14 antibodies.
In experiments assessing the role of AMP kinase and p38 MAPK, compound C (Sigma, Dorset, United Kingdom), SB 203580 (Merck, Nottingham, United Kingdom), or dimethylsulfoxide was added to CD14ϩ cells at 0.01 nM and then washed after an hour-long incubation.
To test the T cell stimulatory capacity, DCs were added at a ratio of 1:10 to 2 ϫ 10 5 CFSE-labeled CD4 ϩ CD25 Ϫ cells in a final volume of 200 L Cellgro SCGM complete media (Cellgenix). In experiments in which OKT3 antibodies were used (0.5 g mL Ϫ1 ), the effectors were obtained by sorting allogeneic whole PBMCs, first by CD4ϩ negative selection and then CD25ϩ depletion. With tetanus toxoid, fresh allogeneic PBMCs were used for sorting. Proliferation was assessed on days 4 -6 by flow cytometry.

Exercise study in healthy individuals
Study participants were 45-64 years old, were nonsmokers, had a body mass index (BMI) of 25-35 kg m Ϫ2 , and were not on medication. Female subjects were all premenopausal. Prior to the study, they performed no regular structured exercise (defined as 30 minutes of moderate intensity activity for more than 5 days a week). All subjects followed a hypocaloric diet (5000 kcal/week) and had 5 supervised treadmill sessions per week for 3 weeks with 600 kcal of energy expended per session. Subjects were asked to perform at moderate intensity [50% maximal oxygen consumption (VO 2 max), estimated based on target heart rate] for approximately 75 minutes per session. Venous peripheral blood samples were collected with full informed consent, in the fasting state, on the morning of day 1 (start) and day 22 (completion). Abbreviations: ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BP, blood pressure; WHR, waist to hip ratio. All subjects were Caucasian males. Mean Ϯ SD values are represented for continuous variables. By t test, there were no significant differences in age, BMI, and WHR between T1D and HC groups. HbA1C, duration of diabetes, lipids, and rates of vascular complications were similar between T1D and T2D groups. The T2D subjects were older and had greater BMI and WHR compared with both HC and T1D subjects. By 2 test, ACEI/ARB and statin use was significantly higher (P Ͻ .05) in the T2D group, compared with the T1D subjects (P Ͻ .0001 and P Ͻ .01, respectively). a P Ͻ.0001.

Statistical analysis
Data are expressed as mean Ϯ SD. The Mann-Whitney U test was used for nonparametric data and the 2 test for categorical data. For comparisons between more than 2 groups, ANOVA and Bonferroni adjustment for multiple testing was used. We used the paired t test for mean SI, of the triplicate from the same subject, between serum-free and adiponectin supplementation conditions. Two-tailed Pearson correlation analysis was used to analyze possible relationship between 2 set variables. Statistical analyses were performed on SPSS version 15.0 (SPSS Inc, Chicago, Illinois). Graphical plots were done using GraphPad Prism version 5.03 (San Diego, California) and Flowjo version 8.87 (Ashland, Oregon). Contour plots were done at the 5% probability level with the outliers shown. Dot plots depict a minimum of 8000 gated events.

Total serum adiponectin concentrations are not affected by T1D
To examine the adiponectin axis in T1D, we first measured circulating adiponectin in subjects with this condition. Subjects with T1D were compared with age and healthy BMI-matched male Caucasian subjects (Table 1). We recruited a second comparator group of male Cauca-sian subjects with insulin-treated T2D, matched for duration of diabetes, HbA1C, and exogenous insulin dose. Total serum adiponectin concentrations were not significantly different between T1D subjects and matched healthy controls (HCs) (T1D: 9.6 Ϯ 6.3 g/mL Ϫ1 ; HC: 7.6 Ϯ 2.3 g/mL Ϫ1 , P ϭ .17) ( Figure 1A). In contrast, circulating adiponectin levels in T2D were significantly reduced (T2D: 4.2 Ϯ 2.8 g/mL Ϫ1 , P ϭ .0003 vs HC). Within patients with T1D, there was a significant inverse correlation between serum adiponectin and BMI (r ϭ Ϫ0.397, P ϭ .033). Other than having BMI in the lowest quintile, the 3 T1D subjects with the highest serum adiponectin levels had no other distinguishing clinical characteristics.

Expression of AdipoR1 and AdipoR2 by peripheral blood monocytes is reduced in T1D
To further explore the effect of T1D on the adiponectin axis, we measured the expression of adiponectin receptors by PBMCs in subjects with this condition. We have previously described the presence of AdipoR1 and AdipoR2 on PBMCs, in particular on monocytes (5). We reproduced these findings (Supplemental Figure 1, published on The Endocrine Soci- ety's Journals Online web site at http://jcem.endojournals.org), including demonstrating a close correlation between expression at the level of protein and RNA (Supplemental Figure 2, A and C-E). T-cadherin, although reported to act as an adiponectin receptor on endothelial cells (23), was not detected on PBMCs (Supplemental Figure 2B). Because monocytes expressed the highest levels of adiponectin receptors, we compared receptor expression on these cells between the T1D, healthy, and T2D subjects ( Table 1). Expression of both AdipoR1 and AdipoR2 was decreased significantly in T1D, by approximately 40% when compared with matched healthy controls (Figure 1, B-E) (MFI AdipoR1, HC ϭ 34.58 Ϯ 18.47, T1D ϭ 19.46 Ϯ 11.16, P Ͻ .01; AdipoR2, HC ϭ 23.14 Ϯ 10.63, T1D ϭ 11.12 Ϯ 9.36 P Ͻ .01). The reduction in adiponectin receptor expression was also seen in all PBMCs (MFI AdipoR1, HC ϭ 4.0 Ϯ 1.6, T1D ϭ 2.3 Ϯ 1.8, P Ͻ .01; MFI AdipoR2, HC ϭ 3.5 Ϯ 1.7, T1D ϭ 1.3 Ϯ 1.1 P Ͻ .01, Supplemental Figure 2F), confirmed at the RNA level (T1D AdipoR1 relative gene expression to HC ϭ 0.27, P Ͻ .01; AdipoR2 ϭ 0.55, P Ͻ .05) (Supplemental Figure 2G).
Within patients with T1D, monocytic expression of AdipoR correlated with estimated glucose disposition ratio, a clinical surrogate of insulin sensitivity in T1D (24) (Figure 1, F and G, AdipoR1, r ϭ 0.4180 P ϭ .0240; AdipoR2, r ϭ 0.4949, P ϭ .0063), and with BMI in patients with T2D (Supplemental Figure 2H). There was no significant correlation between AdipoR expression with HbA1C, insulin dose per kilogram, duration of disease, presence of vascular complications, or medication use in subjects with T1D.

Adiponectin inhibits antigen-driven CD4؉ T cell proliferation, and this effect is reduced in T1D
We wanted to determine the functional significance of a reduced AdipoR expression in T1D. To do so, we first developed a standardized serum (adiponectin)-free assay of CD4ϩ T cell proliferation in which different PBMCs  could be cultured. Division of CD4ϩ T cells within whole PBMC cultures was tracked by CFSE dilution and concomitant up-regulation of IL-2 receptor (CD25) expression. In healthy subjects (n ϭ 8), we demonstrated a consistent reduction in tetanus toxoid (TT)-stimulated CD4ϩ T cell proliferation when adiponectin was supplemented at a physiological concentration of 10 g/mL Ϫ1 (Figure 2, A and B, TT: serum free SI ϭ 7.6 Ϯ 5.1, adiponectin supplementation SI ϭ 3.9 Ϯ 2.6, P ϭ .0085). This inhibition was also observed in response to human islet lysate (Figure 3, A and B) (serum free SI ϭ 3.0 Ϯ 0.6, adiponectin supplementation SI ϭ 1.8 Ϯ 0.3, P ϭ .0004).
To determine the functional significance of reduced adiponectin receptor expression in T1D, we compared the effect of adiponectin on TT and islet antigen stimulated T cell proliferation in these patients. An adiponectin concentration of 10 g mL Ϫ1 was used to model the similar serum levels seen in vivo in health and T1D. Adiponectin inhibited T cell proliferative responses to TT and islet lysate in T1D, but this inhibitory effect was significantly lower than in HC (Figures 2 and 3, 3% reduction in SI after adiponectin treatment: TT, HC ϭ 48.5 Ϯ 7.4%, T1D ϭ 22.5 Ϯ 2.9%, P ϭ .0124; islet lysate, HC ϭ 38.7 Ϯ 2.9%, T1D ϭ 27.0 Ϯ 5.6%, P ϭ .0430). Similar distributions of CD14ϩ monocytes were observed in both groups, again matched in age, sex, ethnicity, and BMI.

Adiponectin-treated mDCs have reduced CD86 expression and CD4؉ T cell stimulatory capacity
Resting T cells do not express adiponectin receptors (5). To explore the mechanism of adiponectin effect on T cell proliferation, we therefore examined the effect of adiponectin on antigen presenting function. We chose to study DCs because they are professional APCs critical to the development of T1D. Also, they can be generated from blood monocytes that express high levels of adiponectin receptors.
Monocyte-derived DCs (mDCs) were expanded ex vivo in serum-free conditions and on day 6 had the phenotype of immature DC generated in 10% fetal calf serum, the standard approach to generating mDCs (Figure 4, A, C, and E; and Supplemental Figure 3). They were CD1a high , CD14 low , and expressed HLA class II and the costimulatory molecules CD80 and CD86. The addition of physiological concentrations of adiponectin at the beginning of culture significantly reduced the expression of CD86 but not CD80 or HLA-DR (Figure 4, A-F).
The inhibition of CD86 was adiponectin dose dependent (Supplemental Figure 4, C and D) but not observed if adiponectin was added after the first 18 hours of the 6-day culture (Supplemental Figure 4B). Adiponectin is an established activator of AMP kinase and p38 MAPK in immune cells (10,25). Inhibition of AMP kinase, but not p38 MAPK, ameliorated adiponectin effect on CD86 expression by CD1aϩmDC ( Supplemental Figure 4, E and F).
CD86 provides a costimulatory signal for T cell activation. Consistent with reduced CD86 expression, adiponectin-treated mDCs had reduced CD4ϩT cell stimulatory capacity (Figure 4, G-I). The proliferation of CFSEϩCD4ϩCD25Ϫ effector cells was significantly re- Figure 3. Adiponectin suppresses islet antigen-stimulated CD4ϩ T cell proliferation in healthy individuals more than in people with T1D. Enumeration of proliferating CD4ϩ T cells in serum-free, whole-PBMC cultures stimulated by human islet lysates (at 5 g/ml Ϫ1 ) and IL-2 (10 Ϫ4 g/ml Ϫ1 ) from a healthy subject (A) and (B) after adiponectin supplementation. Islet lysate stimulation in serum-free culture (C) and with adiponectin supplementation in a T1D subject (D). E, SI of CD4ϩ T cells in serum-free culture and with adiponectin supplementation plotted as a joint pair. * P Ͻ .05, *** P Ͻ .001, paired t test.

Reduced adiponectin receptor expression on T1D monocytes is associated with a proinflammatory response
We then asked whether the reduction in AdipoR expression in T1D would affect mDC phenotype and function. We recruited a new cohort of T1D and matched HC subjects (n ϭ 30) and again confirmed reduction in monocyte expression of AdipoR1 and AdipoR2 (data not shown). We demonstrated that adiponectin did not modulate CD86 expression as much in T1D as it did in HC (ratio CD86 MFI adiponectin-treated DC: serum free con-trol DC, HC ϭ 0.45 Ϯ 0.13, T1D ϭ 0.59 Ϯ 0.17, P ϭ .0207), whereas expression of CD80 and HLA-DR remained unchanged (Figure 5, A-G). The yield of CD1a mDC was not altered by diabetes or adiponectin supplementation in media.
The T cell stimulatory capacity of mDCs from adiponectin-supplemented conditions was compared in subjects with and without T1D (n ϭ 17). Consistent with the effect of adiponectin on costimulatory molecule expression in T1D mDCs, there was reduced inhibition of T cell proliferation with adiponectin-treated mDC from T1D compared with healthy control subjects ( Figure 5H, median reduction HC ϭ 23.6%, T1D ϭ 6.1%, P ϭ .0079).

Up-regulation of adiponectin receptors associates with down-regulation of CD86 on PBMCs
To explore the translational application for our findings, we examined whether a clinical intervention that increased AdipoR expression by immune cells had the potential to modulate their proinflammatory phenotype. In humans, an aerobic exercise programs that lead to weight loss or increase in VO 2 max, up-regulate adiponectin receptor expression on skeletal muscle (26,27). We tested the association between adiponectin signaling and CD86 expression on PBMCs observed in vitro through an exercise trial with healthy nondiabetic subjects. We compared the effect in responders who had achieved an increase in VO 2 max greater than 10% or weight loss greater than 5% with those whose did not (Supplemental Table 1). At baseline, clinical characteristics between the 2 groups were similar. After intervention, responders achieved increases in AdipoR gene expression on PBMCs (95% confidence interval fold change, AdipoR1 1.13-2.39; AdipoR2 0.94 -2.91) ( Figure 6A). These changes were accompanied by significant reduction in waist circumference and fasting insulin, in addition to the changes in VO 2 max (ϩ3.0 Ϯ 0.83 ml/kg Ϫ1 ⅐min Ϫ1 ) and weight (Ϫ2.9 Ϯ 0.6 kg). The increase in AdipoR was seen on monocytes at a protein level ( Figure 6B). The increased AdipoR expression was associated with a decrease in CD86 expression on whole PBMCs (fold change 0.47-0.90) ( Figure 6A). Serum adiponectin levels were not significantly altered at the end of the study in both groups.

Discussion
We demonstrate that adiponectin down-regulates APC expression of CD86, and this associates with a reduction in antigen-specific T cell stimulation. Furthermore, we show that the reduced expression of adiponectin receptors in T1D is associated with release from inhibition by adiponectin, thus promoting an inflammatory T cell response to islet antigen. Lastly, we confirmed, in vivo, an increase adiponectin receptor expression on PBMCs through lifestyle interventions is coupled with a concomitant decrease in CD86.
We recognize mechanisms other than CD86 on myeloid APC may also play a role in the dampening of T cell antigen reactivity. For example, adiponectin can induce monocytes to secrete antiinflammatory cytokines IL-10 and IL-1RA (10) and also suppress the chemokine CCL2 (28). However, we have focused on CD86 because of its importance in autoimmune diabetes. CD86 expression on APCs has a critical role in priming of diabetogenic CD4ϩ T cells, and its knockout in the NOD mouse model of T1D prevents disease (29). Recent clinical trial data also confirm that blocking T cell costimulation preserves ␤-cell function in human T1D (30). Our results therefore demonstrate the potential for adiponectin signaling to modulate islet autoimmunity.
Our data are in agreement with recently published work in a murine system in which adiponectin was reported to alter the phenotype of bone marrow-derived DCs (14). A global reduction in costimulatory molecule expression was reported, in contrast to our findings of a decrease confined to CD86. We suggest that the potential for adiponectin effect in culture serum (in comparison with our serum free system) may explain the differences in results. Adiponectin is known to bind avidly to endotoxin (31), which if present at low levels can be tolerogenic and could confound our in vitro findings (32). However, we believe this to be not the case. The phenotypic change on APCs after endotoxin priming would involve other surface molecules including major histocompatibility complex class II expression (32), which was unchanged with recombinant human adiponectin. In addition, endotoxin contamination in our adiponectin preparation was undetectable by the LAL test.
In contrast to other reports (33,34), we find that the serum total adiponectin levels is not affected by T1D. However, our T1D subjects were carefully matched for age, sex, ethnicity, and BMI as well as renal function and lipid profile. These factors have been reported to affect adiponectin levels in T1D (33, 34) but have not been ac- Figure 6. Exercise up-regulates adiponectin receptors and decreases CD86 expression in healthy nondiabetic subjects. A, Fold change in gene expression on day 21 (after the intervention) relative to baseline (day 0). Responders are defined by a change in VO 2 max greater than 10% or weight loss greater than 5%, at the end of the study. ** P Ͻ .01; * P Ͻ .05 t test. B, Change in MFI of AdipoR1 staining on CD14ϩ monocytes. * P Ͻ .05 paired t test.
counted for in other comparisons between T1D and health populations. In agreement with current knowledge, however, we found an inverse correlation of serum adiponectin with BMI within patients with T1D (35).
The mechanism leading to reduced adiponectin receptor expression by peripheral blood myeloid cells in T1D is unclear. From twins studies, heritability for AdipoR1 and AdipoR2 gene expression is estimated at 39%-78% (36). However, genes coding for AdipoR1/AdipoR2 are not located within regions associated with T1D risk (http:// t1dbase.org). In our hands and in contrast to animal systems (37,38), insulin, glucose, and adiponectin did not affect AdipoR expression in vitro (data not shown). Although adiponectin receptor expression is down-regulated in transgenic mice in which adiponectin is overexpressed (38), we do not believe this mechanism explains our findings because circulating adiponectin was not increased in our subjects, nor did adiponectin correlate inversely with receptor expression in our diabetic subjects (data not shown). Although we show that insulin resistance associates with low adiponectin receptor expression within patients with T1D, insulin resistance alone is unlikely to explain this disease specific decrease because receptor expression is higher in T2D (and these patients are likely to be more insulin resistant). It is likely that other mechanisms contribute, and we speculate that the broad activation of innate immunity associated with T1D (39) underlies the reduced expression described in PBMCs. As an example, inflammatory cytokines inhibit expression of AdipoR1 and AdipoR2 through the endoplasmic reticular stress-inducible transcriptional factor ATF3 (40).
As with other studies involving human subjects, a limitation of our study is the correlative nature of the presented data. We cannot therefore state that the reduced adiponectin receptor expression leads directly to islet autoimmunity in T1D. Our findings do, however, describe a mechanism whereby an imbalance in adipokines such as adiponectin could contribute to the association between insulin resistance, inflammation, and the development of T1D in those at high genetic risk. Thus, we see confirmation of the mediatory role of AdipoR on APC costimulation and islet immunity in animal models to be an important focus for future studies. Also, our finding that adiponectin receptors are up-regulated by exercise suggests a physiological approach to modulating islet immunity in T1D, and this we believe requires further clinical investigation.