Adrenomedullin 2 Improves Early Obesity-Induced Adipose Insulin Resistance by Inhibiting the Class II MHC in Adipocytes

Obesity, as a result of the expansion of adipose tissues, disturbs the insulin sensitivity of insulin target organs and induces the onset of type 2 diabetes (1). There is a strong interaction between the chronic innate immune response and insulin resistance (2). Previous reports showed that the expression of the proinflammatory cytokines TNFα, mainly derived from macrophages, was upregulated in obese mice and exerted an important role in adipose and hepatic insulin resistance (3–5). However, to date the role of the adaptive immune response in adipose inflammation is largely unknown.

It has been reported that a short-term high-fat diet (HFD) elicited substantial numbers of CD4⁺ T cells in adipose tissues. Importantly, the increased percentages of CD4⁺ T cells in adipose tissues occurred prior to macrophages, which suggests that the adaptive immune response may play an important role in early adipose inflammation and insulin resistance (6). The resident naïve CD4⁺ T cells require antigens presented in an MHC class II (MHCII)-dependent manner by antigen-presenting cells (APCs) as the first signaling. CD28 and CD40L on the membrane of T cells interact with the costimulatory molecules CD80/86 and CD40 on APCs, forming the secondary signaling. The naïve T cells secret IL2, which acts in an autocrine manner to activate T cells, resulting in T-cell activation and proliferation (7). The activation of adipose resident T cells triggers the infiltration of more T cells as well as other immune cells, which initiate adipose inflammation (8).

Adrenomedullin 2 (ADM2)/intermedin is a widely expressed bioactive peptide belonging to the calcitonin gene-related peptide (CGRP)/calcitonin family (9,10). Calcitonin receptor–like receptor (CRLR) is an ADM2 receptor, and receptor activity–modifying proteins (RAMPs) are the coreceptors of CRLR (9). The interactions of CRLR with different RAMPs (RAMP1, -2, and -3) form different receptor complexes (11). ADM2 has been shown to exert protective effects on cardiac ischemia/reperfusion injury, vascular calcification, and immunoglobulin A nephropathy by inhibiting inflammation, oxidative stress, and endoplasmic reticulum stress (12–14). We and other groups reported that ADM2 treatment reduced atherosclerosis by inhibiting foam cell formation and improving dyslipidemia (15–17). Recently, our group discovered that ADM2 had a protective effect on HFD-induced established obesity by increasing thermogenesis in adipocytes (18). However, whether ADM2 has a protective effect on early insulin resistance is still unknown. In this study, we demonstrate that adipose-specific overexpression of ADM2 substantially improves early obesity-induced adipose inflammation and insulin resistance in mice. Mechanistic studies revealed that ADM2 inhibits adipose proinflammatory T-cell activation mainly through decreasing MHCII-mediated antigen presentation in adipocytes. Furthermore, ADM2 treatment upregulated B lymphocyte–induced maturation protein 1 (Blimp1) and inhibited the class II transactivator (CIITA)-MHCII axis in a CRLR/RAMP1-cAMP–dependent manner.

Human ADM2_1–53, ADM2_17–47, CGRP_8–37, ADM_22–52, and the anti-ADM2 antibody were purchased from Phoenix Pharmaceuticals (Belmont, CA). The anti-Blimp1 antibody was purchased from Cell Signaling Technology (Boston, MA). The anti-MHCII (M5/114.15.2) and anti-MHCII (OX-6) antibodies were purchased from BD Biosciences (San Jose, CA). Human γ-interferon (IFNγ) was purchased from R&D Systems (Minneapolis, MN). All other chemicals and drugs were purchased from Sigma-Aldrich (St. Louis, MO).

The study was approved by the ethics committee of Beijing Chao-Yang Hospital and complied with the principles outlined in the Declaration of Helsinki. All subjects gave written informed consent prior to participation. The blood samples from all subjects were placed in tubes containing EDTA and aprotinin (500 kIU/mL) and centrifuged immediately. The plasma was stored at −80°C.

The wild-type (WT) mice, aP2-driven adipose-specific human ADM2 transgenic (aADM2-tg) mice (a mouse line constructed by the authors), systemic MHCII knockout (MKO) mice (MARC, Nanjing, China), and OTII mice (a kind gift from Yu Zhang, Peking University) were on a pure C57BL/6J background and housed as previously described (18). The animals were littermates, mixed housed in a cage, and fed a normal chow diet (NCD) (20% calories from fat) or HFD (60% calories from fat; Research Diets, New Brunswick, NJ) for 4 weeks. All the animal protocols were approved by the Animal Care and Use Committee of Peking University.

Mature adipocytes and adipose precursor cells were isolated as previously described (18).

The inflammatory cytokine levels in plasma were investigated using a cytometric bead array inflammation kit (BD Biosciences). For flow cytometry, the mature adipocytes or stromal vascular fractions (SVFs) were isolated as described in cell culture. Cells were filtered and eliminated red blood cells. Fixation and permeabilization were needed when intracellular protein was stained. Then cells were stained with different antibodies and analyzed by flow cytometer. When adipocytes were analyzed, the cells should be mixed constantly to avoid floating of adipocytes. The stained cells were analyzed using a FACSCaliber (BD Biosciences) and Beckman Gallios (Beckman Coulter, Brea, CA) with FlowJo software.

For in vitro antigen presentation assay, the differentiated 3T3-L1 adipocytes were seeded in 12-well plates and handled as previously described (19).

The murine total bone marrow hematopoietic progenitor donor cells were harvested and were transplanted via tail vein injection into the lethally irradiated WT and MKO mice. The transplanted mice were maintained for 6 weeks and treated with vehicle or ADM2 subcutaneously in saline through an Alzet Mini-osmotic Pump (DURECT, Cupertino, CA) at a rate of 300 ng/kg/h.

The extraction of total RNA, reverse transcription, and real-time PCR were undertaken as previously described (18).

The Western blot analyses were undertaken as previously described (18).

The data were expressed as means ± SEM and analyzed using GraphPad Prism software as previously described (18). P < 0.05 was considered significant.

For investigation of the role of ADM2 during the pathogenesis of insulin resistance, the plasma level of ADM2 in human was firstly determined. The human plasma ADM2 level was negatively correlated with HOMA of insulin resistance (HOMA-IR) in human (Fig. 1A). Western blot and quantitative PCR (qPCR) analysis further showed that ADM2 was highly expressed in the epididymal white adipose tissues (eWATs) and was substantially downregulated in the mice fed an HFD for 8 weeks, whereas the expression of ADM2 was much lower and remained unchanged in the liver and skeletal muscle (Supplementary Fig. 1A and B). The plasma level of ADM2 was also decreased after an HFD (Supplementary Fig. 1C).

For determination of whether ADM2 exerts a role in eWAT during insulin resistance, aADM2-tg mice were developed (18). Western blot and qPCR analysis showed that the expression of ADM2 was robustly induced in the eWAT of the aADM2-tg mice, while there were no differences in the macrophages, liver, and skeletal muscle (Supplementary Fig. 1D and E). Then, the eWATs from the WT and aADM2-tg mice were separated into mature adipocytes and SVFs, and the expression of ADM2 was examined. The ADM2 level was enhanced in the mature adipocytes of aADM2-tg mice but not in the SVFs (Supplementary Fig. 1F).

For further validation of the role of adipose ADM2 in insulin resistance, WT and aADM2-tg mice were fed an NCD or an HFD for 4 weeks (4w-HFD). The aADM2-tg mice fed an NCD or a 4w-HFD displayed a body weight gain similar to that of the WT mice fed an NCD or a 4w-HFD (Supplementary Figs. 1G and 2A). However, the glucose tolerance test (GTT) and insulin tolerance test (ITT) revealed that adipose-specific ADM2 overexpression noticeably improved the HFD-induced systemic insulin resistance (Fig. 1B and C) but not in the aADM2-tg mice fed an NCD (Supplementary Fig. 2B and C). Although the fasting plasma level of glucose was not changed in the WT or aADM2-tg mice fed an NCD or a 4w-HFD (Fig. 1D and Supplementary Fig. 2D), the 4w-HFD–induced increased fasting plasma insulin level and HOMA-IR were significantly reversed in the aADM2-tg mice fed a 4w-HFD (Fig. 1E and F) but not in the aADM2-tg mice fed an NCD (Supplementary Fig. 2E and F). Insulin signaling was further investigated in the eWAT. The 4w-HFD–induced insulin signaling impairment in the eWAT of the WT mice was substantially improved, as revealed by an increased level of AKT phosphorylation (Fig. 1G), which was not observed in the aADM2-tg mice fed an NCD (Supplementary Fig. 2G). The plasma levels of inflammatory cytokines were also measured to determine the extent of the systemic inflammation. The 4w-HFD–induced elevation of plasma proinflammatory cytokines levels, including those of IL2, IL17a, IL6, IL12p70, TNFα, and MCP1, in the WT mice was markedly restored in the aADM2-tg mice (Fig. 1H). These results suggest that adipose ADM2 protects against early obesity-induced insulin resistance and inflammation in eWAT of mice.

The increased percentage of immune cells in white adipose tissue (WAT) is one of the important causes of adipose inflammation (20). Thus, the percentages of immune cells in eWAT were determined by flow cytometry. The increased proportion of CD45⁺CD3⁺ T cells in the eWAT of the WT mice induced by a 4w-HFD was substantially reversed in the aADM2-tg mice (Fig. 2A). However, the proportions of other immune cells, such as CD45⁺F4/80⁺ macrophages, CD45⁺CD19⁺ B cells, and CD45⁺CD11c⁺ dendritic cells, were not reduced (Fig. 2A). Furthermore, the proportion of CD3⁺CD4⁺ T cells was increased by a 4w-HFD treatment and was noticeably blunted in the aADM2-tg mice compared with that of the WT mice (Fig. 2B), whereas there was no significant alteration in the proportion of CD3⁺CD8⁺ T cells (Fig. 2B). The percentages of different subsets of CD4⁺ T cells in eWAT were further assessed. The 4w-HFD–induced increased percentages of proinflammatory CD4⁺IFNγ⁺ T helper (Th)1 and CD4⁺IL17a⁺ Th17 cells, as well as the decreased percentages of anti-inflammatory CD4⁺IL4⁺ Th2 and CD4⁺Foxp3⁺ regulatory T cells in the eWAT, were dramatically reversed in the aADM2-tg mice (Fig. 2C).

The activation and proliferation of CD4⁺ Th cells depend on the MHCII-mediated antigen presentation function in APCs (21). Thus, the expression of MHCII in eWAT was determined. The mRNA levels of the related MHCII family genes H2-Eb1 and Ciita as well as levels of the MHCII (H2-A/E) protein were substantially induced in the eWAT of the WT mice treated with a 4w-HFD (Fig. 2D and E) but were downregulated in the eWAT of the aADM2-tg mice fed an NCD or a 4w-HFD (Fig. 2F and G and Supplementary Fig. 3A and B). Flow cytometry analysis of SVFs and adipocytes in eWAT was performed to further investigate the type of cells responsible for the decreased MHCII expression in aADM2-tg mice. The 4w-HFD–induced increased proportion of MHCII⁺ adipocytes was substantially blunted in the aADM2-tg mice but not the MHCII⁺ SVFs (Fig. 2H). Immunofluorescence staining of eWAT further confirmed that the MHCII protein level was downregulated in the aADM2-tg mice compared with those of the WT mice on a 4w-HFD (Fig. 2I). Taken together, these results suggest that adipose-specific ADM2 overexpression noticeably reduces adipocyte-derived MHCII expression and T-cell activation in eWAT.

It was reported that IFNγ induced MHCII expression in adipocytes (19,22). Similarly, IFNγ treatment dramatically increased the mRNA and protein level of MHCII in cultured adipocytes in a dose-dependent manner (Supplementary Fig. 4A–C). Flow cytometry analysis confirmed that the proportion of membrane MHCII⁺ adipocytes was substantially expanded after IFNγ stimulation (Supplementary Fig. 4D). Immunofluorescence staining also demonstrated an increase of MHCII expression in the adipocytes treated with IFNγ (Supplementary Fig. 4E).

Primary adipocytes treated with ADM2 were investigated by RNA-Seq to explore the mechanism by which adipose-specific ADM2 overexpression markedly downregulated MHCII expression. The RNA-Seq analysis showed that ADM2 stimulation reduced the mRNA levels of related MHCII genes, Rt1-B, Rt1-D, Rt1-DM, Rt1-DO, Cd74, and Ciita, as well as the costimulatory molecules, Cd80 and Icam1 (Fig. 3A). qPCR analysis confirmed that ADM2 substantially reversed the IFNγ-induced expression of the MHCII family genes, including H2-Aa, H2-Ab1, H2-Eb1, H2-Eb2, H2-Ma, H2-Mb, H2-Oa, Cd74, and Ciita, but not the costimulatory factors and adherent molecule, including Cd40, Cd80, Cd86, and Icam1, in 3T3-L1 adipocytes (Fig. 3B). Furthermore, the ADM2 treatment inhibited the expression of MHCII mRNA and protein levels in the primary adipocytes in a dose-dependent manner under both basal and IFNγ-induced conditions (Fig. 3C–F). Immunofluorescence staining further validated that the IFNγ-induced upregulation of MHCII in adipocytes was abolished after ADM2 treatment (Fig. 3G). The image-capture flow cytometry analysis also showed that the ADM2 treatment markedly reversed the IFNγ-induced upregulation of MHCII protein level at the adipocyte membrane (Fig. 3H).

For further identification of the role of ADM2 in adipocyte-mediated antigen presentation function, an antigen presentation assay was performed in vitro. 3T3-L1 adipocytes were pretreated with IFNγ and ADM2 and were subsequently cocultured with naïve T cells isolated from OTII mice with ovalbumin (OVA). The ability of adipocytes to activate naïve T cells was assessed by the levels of IL2 and IFNγ secreted into the culture media by the T cells. The IFNγ-induced secretion of IL2 and IFNγ was partially reversed by ADM2 treatment (Fig. 3I and J). Overall, these results indicate that ADM2 prevents the activation of T cells in WAT primarily by downregulating MHCII expression in adipocytes.

In atypical APCs, IFNγ is the only well-characterized inducer of MHCII expression (23). For exploration of the mechanism by which ADM2 reduced the expression of MHCII in adipocytes, whether the ADM2-inhibited expression of MHCII depended on IFNγ pretreatment was examined. ADM2 was observed to inhibit the basal expression of MHCII in primary adipocytes without IFNγ treatment (Fig. 3C–E). Moreover, the effect of ADM2 on the IFNγ signaling pathway was also detected. The IFNγ-induced phosphorylation of signal transducer and activator of transcription 1 (STAT1) remained similar after ADM2 treatment (Supplementary Fig. 5A), suggesting that the ADM2-mediated inhibition of MHCII was independent of the IFNγ signaling pathway. CYT387, an IFNγ receptor adaptor protein Jak1/2 inhibitor, was also used to block the IFNγ signal pathway (Supplementary Fig. 5B). CYT387 blocked both the basal and IFNγ-induced MHCII expression in adipocytes (Supplementary Fig. 5C–E). However, ADM2 had a further inhibitory effect on MHCII expression after pretreatment of CYT387 (Supplementary Fig. 5C–E), which suggesting that the inhibition of MHCII expression by ADM2 is independent of IFNγ.

CIITA was shown to be the core transcription factor of MHCII with three types of transcripts, Ciita-PI, Ciita-PIII, and Ciita-PIV, expressing in different types of cells (23). Therefore, the expression of different Ciita transcripts was examined. The expression levels of Ciita-PI and Ciita-PIV in adipocytes were restrained by ADM2 under the basal and IFNγ-treated conditions (Fig. 4A). These results indicate that ADM2 inhibits MHCII expression via direct downregulation of Ciita-PI and Ciita-PIV transcription.

The precise molecular mechanism by which ADM2 downregulated CIITA expression in adipocytes was then determined. It is reported that Blimp1 is a powerful transcription factor that negatively regulates MHCII expression and downregulates all three Ciita transcripts (24–26). Our results showed that ADM2 markedly increased the expression of Blimp1 mRNA and protein in adipocytes in a time-dependent manner (Fig. 4B and C). Consistently, the aADM2-tg mice displayed a striking increase of Blimp1 mRNA and protein levels in the eWAT compared with those of the WT mice on an NCD or a 4w-HFD (Fig. 4D and E and Supplementary Fig. 6A and B). For determination of whether Blimp1 is involved in ADM2 inhibition of CIITA expression, Blimp1 expression was knocked down in adipocytes with a specific small interfering (si)RNA. The knockdown efficiency of BLIMP1 expression in adipocytes was >50% at the protein level (Fig. 4F). The ADM2-mediated inhibition of the IFNγ-induced MHCII expression was totally abolished after the Blimp1 siRNA treatment (Fig. 4G). Collectively, these results suggest that the downregulation of the CIITA transcription via the increase of Blimp1 expression mediates the ADM2-mediated inhibition of MHCII expression in adipocytes.

For validation that the CRLR/RAMPs that are responsible for the ADM2-mediated inhibition of MHCII expression, three different peptide antagonists—ADM2_17–47 for the inhibition of CRLR/RAMP1, -2, and 3; CGRP_8–37 for the inhibition of CRLR/RAMP1; and ADM_22–52 for the inhibition of CRLR/RAMP2 and -3—were used. Pretreatment with ADM2_17–47 and CGRP_8–37 eliminated the ADM2-mediated inhibition of the IFNγ-induced MHCII expression as well as increased Blimp1 expression in adipocytes but not ADM_22–52 (Fig. 5A–D). These results imply that CRLR/RAMP1 might mainly mediate the effects of ADM2 on the Blimp1-CIITA-MHCII axis. It is reported that the activation of CRLR/RAMP1 triggers various signal transduction pathways (27). In the current study, two signaling pathways were firstly determined with the following inhibitors: LY294002 to inhibit Akt and compound C to inhibit AMPK. The ADM2-mediated downregulation of MHCII and upregulation of Blimp1 were not inhibited by LY294002 or compound C (Fig. 5F–I). CRLR/RAMP1, as a G-protein–coupled receptor complex, increased the cellular cAMP level (Fig. 5E). Blocking the cAMP pathway by Rp-cAMPS eliminated the ADM2-mediated inhibition of the MHCII expression and Blimp1 upregulation (Fig. 5F–I). These results reveal that ADM2 inhibits the Blimp1-CIITA-MHCII axis in a CRLR/RAMP1-cAMP–dependent manner.

For further identification of the role of adipocyte MHCII in the ADM2-induced early adipose insulin resistance improvements, MKO mice were used. The levels of the MHCII mRNA and protein were almost undetectable in the macrophages and eWAT of the MKO mice (Supplementary Fig. 7A and B). However, adipose macrophages, B cells, and dendritic cells as APCs derived from bone marrow infiltrate into adipose tissue under physiological and pathological conditions (28–30). The antigen-presenting function of these cells in the eWAT should be considered. Therefore, the hematopoietic cells from the WT mice were transplanted into the WT and MKO mice to generate the chimeric WT and MKO mice (WT-WT and WT-MKO) (Supplementary Fig. 7C). PCR analysis confirmed the highly chimeric tissue in WT-MKO mice (Supplementary Fig. 7D).

The WT-WT and WT-MKO mice were treated with vehicle or ADM2 subcutaneously via mini-pumps and concurrently fed a 4w-HFD. The ADM2 treatment substantially increased the plasma ADM2 level in the WT-WT and WT-MKO mice at the end of the fourth week (Supplementary Fig. 7E). And the ADM2 treatment had no effect on the body weight gain of the WT-WT and WT-MKO mice fed a 4w-HFD (Supplementary Fig. 7F). However, ADM2 noticeably ameliorated insulin resistance in the WT-WT mice but not in the WT-MKO mice (Fig. 6A and B). No significant difference in the fasting plasma glucose levels was observed in the WT-WT and WT-MKO mice after the ADM2 treatment (Fig. 6C). The fasting plasma insulin level and HOMA-IR were substantially reduced in the ADM2-treated WT-WT mice comparatively (Fig. 6D and E). However, the WT-MKO mice were relatively unresponsive to the metabolic benefits of the ADM2 treatment (Fig. 6D and E). ADM2 treatment markedly increased the basal level of AKT phosphorylation in the WT-WT mice. Although the AKT phosphorylation level was increased in WT-MKO mice compared with WT-WT mice, ADM2 treatment cannot further increase the phosphorylation level of AKT (S.Y.-Z., Y.L., H.Z., C.J., X.W., unpublished data). Taken together, these results indicate that the protective role of ADM2 in 4w-HFD–induced insulin resistance in eWAT depends primarily on the inhibition of MHCII in adipocytes.

In the current study, we found that ADM2 exerts metabolic benefits on early obesity-induced adipose inflammation and insulin resistance by inhibiting MHCII expression in adipocytes (Fig. 7). During the pathogenesis of obesity, adipose CD4⁺ proinflammatory Th1 cells were activated and secreted large amounts of IFNγ (6,19,31). IFNγ stimulation substantially enhanced the expression of MHCII in adipocytes. Subsequently, adipocyte MHCII-mediated antigen presentation activated more proinflammatory CD4⁺ Th cells. ADM2 was observed to upregulate Blimp1 expression by activating the CRLR/RAMP1-cAMP pathway. The Blimp1-CIITA-MHCII axis mediated ADM2-induced improvement in early obesity-induced adipose inflammation and insulin resistance. These findings demonstrate that ADM2 is a potential drug candidate for early obesity-induced inflammation and insulin resistance.

In obese mice, the adipose T cells exhibited antigen-specific expansion, and the Th1 proportion was strikingly increased (32). The MHCII, CD40L, and CD80/86 systemic knockout mice displayed a lower susceptibility to HFD-induced adipose inflammation, suggesting that APC-induced CD4⁺ T-cell activation is required for adipose inflammation (19,33,34). However, the exact type of APC that presents the antigen to the adipose T cells in early obesity remains largely unknown. Previous results show that adipocytes have antigen presentation function and might be the APCs in WAT (19,22). Consistently, we also found the expression of MHCII in adipocytes.

Preadipocytes share a lot of features with fibroblasts and macrophages (35,36). Both macrophages and fibroblasts have been shown to express MHCII after IFNγ treatment (23). Therefore, it is difficult to exclude whether the expression of MHCII induced by IFNγ in differentiated adipocytes is derived from undifferentiated preadipocytes. Here we isolated primary adipocytes from eWAT of rat and treated with IFNγ. IFNγ markedly upregulated MHCII expression. MHCII expression was induced in the eWAT of mice fed a 4w-HFD. However, other groups did not observe this phenotype in the mice fed an NCD (28,37). This might result from the fact that these mice were not fed with an HFD and the expression of MHCII was too low to be observed. In addition, WAT is not uniform and not all adipocytes express MHCII. The immunofluorescence staining might miss the MHCII expression in WAT.

Although adipocytes might be the APC-presenting antigen in WAT, the exact antigen presented by adipocytes during obesity is still unknown. MHCII is believed to present exogenous antigens, but nearly 20–30% of antigens presented by APCs in a MHCII-dependent manner are self-antigens (38). Previous results showed that in atherosclerosis heat shock protein 60, oxidized LDL and β2-glycoprotein I act as self-antigens (39–41). During the onset of adipose expansion in obesity, oxidative stress and endoplasmic reticulum stress induce the generation of lipid-binding protein, oxidative modified protein, and misfolded protein, which might be the self-antigen to MHCII.

The inhibition of adipocyte MHCII antigen presentation led to the ADM2-mediated improvements in 4w-HFD–induced early adipose inflammation, as revealed by the attenuated benefits of ADM2 in the WT-MKO mice. The WT-MKO mice also displayed an improvement of insulin resistance compared with that in the WT-WT mice, which is similar to that of the WT-WT mice treated with ADM2 and cannot be further improved by ADM2. This observation is consistent with previous results and suggests that the protective effect of ADM2 is at least partially through the inhibition of MHCII expression in adipocytes (19). However, other evidence revealed that adipose macrophages also present antigen and play a key role in the late stage of obesity-induced adipose T-cell activation and insulin resistance (28,42). It should be noted that the role of macrophages in adipose antigen presentation cannot be excluded in obesity-induced adipose inflammation and insulin resistance. Therefore, the role of MHCII in different cells and under different conditions needs to be investigated further.

ADM2 is highly expressed in WAT (43). Our unpublished data showed that ADM2 was expressed in adipocytes. In addition, ADM2 is expressed in vascular tissue (44), which might be another source of ADM2 in WAT. ADM2 shares receptor complexes with CGRP and ADM. Here, ADM2 was demonstrated to inhibit MHCII expression in adipocytes primarily through CRLR/RAMP1, which is the classical pharmacological CGRP₁ receptor (45). In agreement with our present findings, some reports have shown that CGRP reduced MHCII expression in dendritic cells and hair follicle dermal papilla (46,47), although the exact mechanism remains unclear.

The related MHCII family genes are tightly regulated primarily at the transcription levels by CIITA. Three CIITA transcripts were expressed in different cell types. In addition to the constitutive Ciita-PI and Ciita-PIII expression by professional APCs, IFNγ was shown to increase Ciita-PIV expression (48). Ciita-PI and Ciita-PIV were expressed in adipocytes, suggesting that adipocytes may have some features of professional APCs. Previous studies have shown that IL4, IL10, and TGF-β inhibit MHCII expression, while the exact molecular mechanism has not been described (49,50). In this study, ADM2 downregulated MHCII expression as a result of increased Blimp1 expression through the CRLR/RAMP1-cAMP pathway, which provided the exact molecular mechanism of MHCII downregulation.

Funding. This work was supported by the National Natural Science Foundation of the P.R. of China (91439206 and 31230035 to X.W. and 81470554 to C.J.), the National Basic Research Program (973 Program) of the P. R. of China (2012CB518002 to M.-J.X.), the 111 Project of the Chinese Ministry of Education (B07001), and the Center for Molecular and Translational Medicine (BMU20140475 to G.L.).

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

Author Contributions. S.-Y.Z., Y.L., and T.W. designed and performed the experiments and analyzed data. H.Z. and S.G. contributed to the human sample collection and analysis. J.F. designed the immunological experiments. Y.W. and G.L. constructed the aADM2-tg mouse line. M.-J.X., X.W., and C.J. designed and supervised the research. S.-Y.Z., X.W., and C.J. wrote the manuscript. All authors approved the final manuscript. C.J. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.