Endoplasmic Reticulum Stress Regulates Adipocyte Resistin Expression

Six-week-old wild-type male C57Bl/6 mice were placed on high-fat diet (HFD) from Research Diets (45 kcal% fat) or normal chow (NC) (6 kcal% fat) for 30 weeks. At the end of the study, blood samples for serum resistin measurement were collected. Epidydimal white adipose tissue was isolated and processed for RNA or protein. All animal experiments were performed at the University of Pennsylvania according to protocols approved by the Institutional Animal Care and Use Committee.

3T3-L1 preadipocytes (American Type Culture Collection) were cultured in growth medium (high-glucose Dulbecco's modified Eagle's medium; Invitrogen) supplemented with 10% FBS (U.S. Biotechnologies) and 100 units/ml penicillin and 100 μg/ml streptomycin (Invitrogen). Two days postconfluence, the cells were induced to differentiate with standard cocktail consisting of growth medium with 1 μmol/l dexamethasone, 10 μg/ml bovine insulin, and 0.5 mmol/l isobutyl-1-methylxanthine (Sigma). After 3 days in differentiation medium, the cells were treated with growth medium with 10 μg/ml bovine insulin for 2 days and then maintained in growth medium alone. Cells were considered mature adipocytes 8 days postinduction of differentiation, when knockdown experiments or treatment with tunicamycin, thapsigargin, or actinomycin D (all from Sigma) were performed.

Short interfering RNA (siRNA) oligos targeting C/EBPα, PPARγ, CHOP10, and resistin (l-051631-00), as well as a nontarget negative control, were obtained from Dharmacon. Transfection was performed by electroporation with Nucleofector II and Cell Line Nucleofector Kit V (AMAXA). Electroporated cells were reseeded in growth medium and harvested or used for treatment at 24 h or 48 h post-transfection. Sense sequences for the specific target oligos are as follows:

• C/EBPα_1 GAGCCGAGAUAAAGCCAAAUU

• C/EBPα_2 CCUGAGAGCUCCUUGGUCAUU

• C/EBPα_3 GGAGUUGACCAGUGACAAUUU

• C/EBPα_4 CUAUAGACAUCAGCGCCUAUU

• CHOP10_1 CAACAGAGGUCACACGCACUU

• CHOP10_2 GCACCAAGCAUGAACAGUGUU

• CHOP10_3 GAGCAAGGAAGAACUAGGAUU

• CHOP10_4 GAAACAGAGUGGUCAGUGCUU

• PPARγ CAACAGGCCUCAUGAAGAAUU

Resistin-luc was generated by inserting −13580/+243 bp fragment of the mouse resistin promoter/enhancer into pGL3-basic vector (Promega) as described previously (31). 3T3-L1 adipocytes were transfected by electroporation with Nucleofector II (AMAXA). Briefly, mature adipocytes (day 10 after differentiation) were detached from culture dishes with 0.25% trypsin, washed twice with 1× PBS, and resuspended in electroporation buffer (solution V, AMAXA). Approximately 1 × 10⁶ were electroporated (electroporation program T-020) with 2 μg of pGL3-basic or resistin-luc and 0.3 μg of β-galactosidase expression vector and seeded into 3 wells of a 24-well plate. After 16 h, the medium was replaced with fresh culture medium with vehicle or 5 μg/ml tunicamycin. The cells were incubated for 24 additional hours, and luciferase activity was measured by a luciferase assay kit (Promega). Light units were normalized to β-galactosidase activity. Fold activations relative to the pGL3-basic and vehicle were calculated, and the results of triplicate samples were plotted.

Resistin levels in tissue culture media, whole cell lysates, and EWAT lysates (homogenized in PBS) were measured using a mouse resistin ELISA (Millipore) following the manufacturer's instructions. Samples were diluted appropriately before loading. Serum resistin levels in the mouse diet-induced obesity study were measured using a mouse Adipokine LINCOplex Assay (Millipore) according to the manufacturer's instructions. Total cell protein was measured using diluted samples on a NanoDrop Spectrophotometer (Thermo Scientific).

RNA was isolated from cells with the RNeasy Mini Kit or from adipose tissue with the RNeasy Lipid kit (both from Qiagen). Reverse transcription of ∼0.5 μg of RNA was performed with the RT FOR PCR ADVANTAGE KIT (Clontech) following the manufacturer's instructions. Quantitative PCR (QPCR) was performed using TaqMan Polymerase Universal Master Mix or Power SYBR Green PCR Mastermix (Applied Biosystems) and the PRISM 7500 instrument (Applied Biosystems). Data were analyzed using the standard curve method and normalization of all genes of interest to the house-keeping control gene Arbp/36b4. Analysis of activating transcription factor 3 (ATF3), PPARγ, mouse, and human C/EBPα expression was conducted using TaqMan Gene Expression Assays (Applied Biosystems). Primer sequences used for QPCR analysis of CHOP10 and BiP mRNA were obtained from Rutkowski et al. (25). The remaining primer sequences are as follows:

• F-resistin: TCATTTCCCCTCCTTTTCCTTT

• R-resistin: TGGGACACAGTGGCATGCT

• F-Arbp/36b4: CAACCCAGCTCTGGAGAAAC

• R-Arbp/36b4: CCAACAGCATATCCCGAATC

Cell protein extracts were generated in cold whole-cell extract buffer (0.15 M NaCl, 0.05 M Tris, pH 7.4, 0.005 M EDTA, 0.5% NP-40) with Complete protease inhibitors (Roche). Immunoblotting following SDS-PAGE was performed using the following antibodies: anti-C/EBPα (sc-61, Santa Cruz), anti-PPARγ (sc-7273, Santa Cruz), anti-HDAC2 (sc-7899, Santa Cruz), and anti-Ran (610341, BD Biosciences), anti-phospho-eIF2α (Ser51) (119A11, Cell Signaling), total eIF2α (sc-11386, Santa Cruz).

Cross-linking of adipocyte proteins and chromatin was performed in 1% Formaldehyde (Sigma), followed by quenching in 125 mmol/l glycine and two washes with 1× PBS (Invitrogen). Nuclear pellets were obtained following dounce homogenization in nuclear lysis buffer (20 mmol/l HEPES, 0.25 M sucrose, 3 mmol/l MgCl₂, 0.2% NP-40, 3 mmol/l β- mercaptoethanol, 0.4 mmol/l phenylmethylsulfonyl fluoride, complete protease inhibitor tablets). Sonication was carried out with Bioruptor (Diagenode) in chromatin immunoprecipitation (ChIP) SDS lysis buffer (50 mmol/l HEPES, 1% SDS, 10 mmol/l EDTA). Immunoprecipitations were performed using ∼100 μg/ml chromatin and 10 μg of antibody: anti-C/EBPα (sc-61, Santa Cruz), anti-PPARγ (sc-7196, Santa Cruz), or nonspecific rabbit IgG control (Sigma). After cross-link reversal, DNA was isolated with phenol/chloroform/isoamyl alcohol, and enrichment was measured by QPCR, using Power SYBR Green PCR Mastermix (Applied Biosystems) and the PRISM 7500 instrument (Applied Biosystems). Analysis was performed by the standard curve method and normalization to a nontarget control region of the Arbp/36b4 gene. Primers used for QPCR analysis are as follows:

• F-dnstr-resistin-50bp TCCCTCCTCTGGGACCTCTA

• R-dnstr-resistin-50bp CCCATCCTGCCTTGGATAAT

• F-upstr-resistin-9kb GTAAGGGGGTGGCCTGATAG

• R-upstr-resistin-9kb ATTCCCTTCTCCCACCAAGT

• F-Arbp/36b4 CTGGGACGATGAATGAGGAT

• R-Arbp/36b4 AGCAGCTGGCACCTAAACAG

• F-ins1 CTTCAGCCCAGTTGACCAAT

• R-ins1 AGGGAGGAGGAAAGCAGAAC

• F-albumin CTTCAGCCCAGTTGACCAAT

• R-albumin AGGGAGGAGGAAAGCAGAAC

Student's t test was used to determine P values.

To investigate the potential link between ER stress and resistin downregulation, mouse 3T3-L1 adipocytes were incubated with thapsigargin, which causes ER stress by inhibiting the sarco/ER Ca²⁺ pump, and led to dramatically reduced resistin mRNA levels (Fig. 1,A). To confirm ER stress as the mechanism underlying resistin downregulation, adipocytes were treated with tunicamycin, which induces ER stress by inhibiting N-linked glycosylation of newly synthesized proteins. Tunicamycin treatment markedly downregulated resistin mRNA levels in a time-dependent fashion (Fig. 1,B). By contrast, the same treatment induced expression of ATF3, which is activated by ER stress (26). The decrease of resistin mRNA by tunicamycin was also dose-dependent (Fig. 1,C). To examine the effects of tunicamycin on resistin protein levels, cells were treated for 24 h to lower resistin mRNA as in Fig. 1,C, then treated with fresh tunicamycin- or vehicle-containing media, after which the accumulation of secreted resistin as well as the intracellular resistin levels were measured by ELISA. Resistin protein was substantially decreased in the media and cell lysate of tunicamycin-treated cells (Fig. 1,D), similar to that observed with resistin knockdown, which reduced the mRNA to similar extent but did not induce upregulation of BiP mRNA that would have signified ER stress (Fig. 1 E). Finally, the effects of tunicamycin on resistin secretion also appeared to be dose dependent (data not shown). Collectively, these results indicate that ER stress is a potent regulator of resistin mRNA and protein levels in 3T3-L1 adipocytes.

It has been reported previously that in obese mice, resistin mRNA levels are decreased while protein levels in the circulation are increased compared with lean mice, raising the possibility that during obesity there is dissociation between resistin mRNA and protein levels. A reasonable prediction, then, would be that protein levels in the fat pad may also be higher in obese versus lean mice, similar to what is seen in the circulation. To address this question, C57Bl/6 mice were fed HFD or NC for 30 weeks, at which point resistin mRNA levels were decreased in epididymal white adipose tissue (EWAT) of the HFD mice (Fig. 2,A), despite increased serum resistin levels (Fig. 2,B). Surprisingly, EWAT resistin protein levels were not increased, but rather tended to be lower in the HFD mice (Fig. 2,C). This indicates that locally in the fat pad, resistin protein levels correspond to the mRNA changes, similar to what was observed in the tunicamycin-treated cells. Furthermore, consistent with previous reports of ER stress in obesity (21,27), markers of ER stress such as phospho-eIF2α (Fig. 2,D) and BiP mRNA levels (Fig. 2 E) were elevated in the adipose tissue from the HFD-fed obese mice. Taken together, these data suggest that ER stress may be a relevant mechanism in the downregulation of resistin in vivo in the setting of mouse obesity.

An initial step in dissecting the mechanism by which ER stress regulates resistin expression was to determine whether the downregulation of resistin by tunicamycin is transcriptional. For this purpose, 3T3-L1 adipocytes were treated with 5 μg/ml tunicamycin in the presence of 5 μg/ml of the transcriptional inhibitor Actinomycin D. Tunicamycin treatment did not reduce the half-life of resistin mRNA as would have been expected if ER stress reduced resistin mRNA by a post-transcriptional mechanism (Fig. 3,A). The effect of ER stress was further explored using a luciferase reporter vector (resistin-luc) driven by a large fragment of the resistin gene including the promoter and transcriptional start site (−13,580 bp to +243 bp). The resistin-luc reporter was active in mature adipocytes, but most of this activity was lost when the cells were treated with 5 μg/ml tunicamycin for 24 h (Fig. 3 B), further demonstrating that ER stress causes reduced resistin gene transcription.

One candidate transcription factor that could explain the effects of ER stress on resistin expression is C/EBPα. C/EBPα is critical for adipocyte differentiation (28) and contributes to activation of adipocyte-specific genes such as adiponectin (29), PPARγ (30), and resistin (31). A binding site for C/EBPα on the resistin promoter has been characterized, and it has been shown that ectopic C/EBPα expression in nonadipogenic cells could drive luciferase expression from the resistin promoter (31). Furthermore, C/EBPα was recently reported to be downregulated on the mRNA level by the inducers of ER stress tunicamycin and thapsigargin (32), and indeed we confirmed that C/EBPα mRNA and protein are reduced in tunicamycin-treated adipocytes (Fig. 4,A and B). This treatment also abolished C/EBPα occupancy at the resistin gene as measured by ChIP both at the previously characterized C/EBPα binding site (31) as well as an additional upstream site identified in a recent genome-wide ChIP-chip study (33) (Fig. 4 C). Similar decreases in occupancy were noted on the known C/EBPα binding sites on the promoters of PPARγ and adiponectin (data not shown).

Next, we investigated whether knockdown of C/EBPα could recapitulate the effects seen by tunicamycin treatment. For this purpose, mature adipocytes were electroporated with siRNA oligonucleotides against C/EBPα or a nontarget control. This strategy reduced C/EBPα protein to levels similar to those seen in control cells treated with 5 μg/ml tunicamycin (Fig. 4,D). Resistin mRNA levels, measured 48 h after electroporation, appeared greatly reduced when C/EBPα had been knocked down (Fig. 4,E). Interestingly, measurement of C/EBPα mRNA (Fig. 5,A) and protein (Fig. 5 B) in WAT of obese mice revealed that C/EBPα levels were also decreased relative to WAT of lean mice. This finding suggests that in vivo as well as in vitro, ER stress activation is associated with decreased C/EBPα levels, which may account for the effects seen on resistin expression. Of note, C/EBPβ levels were not significantly changed in adipocytes by tunicamycin treatment or by HFD feeding in WAT (data not shown).

Our recent genome-wide study of C/EBP and PPARγ binding in adipocytes (33) demonstrated that PPARγ binds near C/EBPα at the upstream resistin enhancer. PPARγ, which is crucial for adipogenesis (34,35), also induces resistin expression during adipocyte differentiation (36). Therefore, we examined whether changes in PPARγ levels may also mediate the effects of ER stress on resistin expression. Indeed, tunicamycin treatment of adipocytes substantially reduced PPARγ gene expression (Fig. 6,A) as has been reported previously (24). Moreover, tunicamycin treatment nearly abolished recruitment of PPARγ to the resistin gene (Fig. 6,B). Furthermore, knockdown of PPARγ in mature adipocytes reduced resistin mRNA levels by ∼90% (Fig. 6,C). Efficiency of PPARγ knockdown was assessed by measuring PPARγ mRNA levels by QPCR (Fig. 6 D). Thus, PPARγ downregulation contributes to the effects of ER stress on resistin gene expression.

Another transcription factor that is active in adipocytes under conditions of stress and was recently shown to regulate expression of the adipokine adiponectin is CHOP10 (24). This protein bears significant homology to other C/EBPs and is induced by various stressors, including hypoxia and the unfolded protein response (37). Although it does not bind DNA alone, it can have dominant negative interactions with other C/EBPs leading to decreased transcription of their targets (37,38). As expected from other cell types, CHOP10 gene expression was elevated in the setting of ER stress in 3T3-L1 adipocytes (Fig. 7,A). Knockdown of CHOP10 (Fig. 7,B) partially prevented the tunicamycin-induced decrease in adipocyte resistin mRNA (Fig. 7,C). CHOP10 depletion also increased C/EBPα expression in tunicamycin-treated adipocytes (Fig. 7 D), suggesting that CHOP10 inhibits resistin expression in part by downregulating C/EBPα.

The present study demonstrates for the first time that ER stress, which exists in obese adipose tissue, can downregulate resistin in vitro in mouse adipocytes. The effect of ER stress induction appeared to be primarily transcriptional, and three transcription factors were implicated in mediating the effects of ER stress on resistin levels. C/EBPα and PPARγ, which are known activators of resistin, are downregulated by treatment with tunicamycin and have diminished binding at the resistin gene. Knockdown of these factors mimicked the effects of ER stress induction. On the other hand, CHOP10 is a transcriptional repressor that is activated by ER stress in adipocytes and can interact in a dominant negative fashion with C/EBPα (37,38). Knockdown of this protein was able to partially rescue the effects of tunicamycin on resistin expression, suggesting that CHOP10 may also be responsible for the changes in resistin mRNA. Thus, the effects of ER stress on resistin levels in mouse adipocytes appear to be a combination of decreased expression of activating transcription factors, including C/EBPα and PPARγ, and increased expression of repressors such as CHOP10 and possibly others.

A remaining question is how activation of the unfolded protein response leads to downregulation of C/EBPα and PPARγ. One possibility is that repressors of gene transcription such as CHOP10 and ATF3 that are activated as part of the ER stress response may be directly involved. For example, it is known that both C/EBPα and PPARγ genes are activated by C/EBPα binding to their promoters (30,39), and therefore CHOP10 may decrease their expression via its dominant negative interactions with C/EBPα. Notably, however, CHOP10 knockdown was not able to fully rescue the effects of tunicamycin treatment on C/EBPα, indicating that other factors must be involved. Knockdown of ATF3, which is another transcriptional repressor (40) that is activated by ER stress (26), did not affect the ability of tunicamycin to reduce resistin levels (data not shown), suggesting that it does not play a role in this process.

The implication of these findings is that ER stress activation may provide an explanation for the decrease in resistin mRNA in adipose tissue of obese mice in the setting of elevated serum resistin levels. Indeed, this study demonstrates that decreased resistin expression co-exists with markers of ER stress activation such as increased BiP mRNA and phospho-eIF2α in adipose tissue from obese mice fed an HFD. Furthermore, C/EBPα mRNA and protein were decreased under these conditions, consistent with the role of this transcription factor as an important regulator of resistin (31). In addition, the study demonstrates that in EWAT the levels of resistin protein reflect the downregulation in the mRNA, similar to what is observed for 3T3-L1 adipocytes in vitro. This raises the possibility that the discrepancy between adipose tissue and plasma resistin levels may not be occurring at the level of individual adipocytes but rather results from various global defects characteristic of obesity and insulin resistance. For example, it has been shown that the development of obesity is associated with an increase in fat cell number (41,42), and therefore the net effect in obesity may be elevated resistin release into the circulation even if resistin secretion is decreased on a per-cell basis. In addition, a number of recent studies have demonstrated a negative correlation between renal function and resistin levels (43,44), suggesting that resistin may be cleared through the kidney. Thus, in the setting of diabetic nephropathy, resistin clearance may be impaired leading to accumulation of the protein in the circulation. Another possibility is that resistin half-life in obesity may be increased because of oligomerization. A number of studies have shown that both mouse and human resistin can form oligomers, which can be detected in the circulation (45) and have different biological actions compared with the monomer form (45,46); and the propensity to oligomerize is concentration dependent (47).

It has been previously hypothesized that the discrepancy between resistin mRNA and circulating protein levels may be because of the hyperinsulinemia associated with obesity and insulin resistance (15). In vitro experiments have demonstrated that insulin treatment of mature adipocytes downregulates resistin expression (15,48,49) although the mechanism has not been elucidated. However, there is evidence that insulin can potently decrease C/EBPα expression in differentiated 3T3-L1 cells leading to decreased C/EBPα binding at target DNA sequences (50). This suggests that both ER stress and hyperinsulinemia may contribute to the decreased resistin mRNA levels in obesity by converging on C/EBPα. Moreover, insulin treatment of 3T3-L1 adipocytes did not lead to a discrepancy between resistin mRNA and protein secretion (data not shown), suggesting that at least in this model system both insulin and ER stress downregulated the protein levels along with the mRNA of resistin. It should be noted that the effects of insulin and ER stress in vivo may be different from those observed in vitro in cultured 3T3-L1 cells. However, 3T3-L1 cells, which are derived from immortalized mouse embryonic fibroblasts and can be differentiated into adipocytes with a hormonal cocktail, are generally considered a valid model for adipocyte function. Moreover, transplantation of 3T3-L1 cells into nude mice has been shown to result in formation of adipose tissue that is essentially identical to normal fat (51), suggesting that this cell line is fully capable of reproducing the in vivo adipocyte phenotype.

The overall significance of the findings presented here is that ER stress, which develops in obese adipose tissue, can affect adipocyte function on many levels including dysregulation of adipokine production. It was previously shown that ER stress can impair insulin signaling both in vitro in adipocytes and in vivo in obese XBP^+/− mice, which are unable to respond properly to ER stress, and develop dramatically worse adipose tissue insulin resistance compared with wild-type controls (21). Importantly, treatment of obese mice with chemical chaperones that alleviate ER stress improves signaling through the insulin receptor (52), indicating that the effects of ER stress on adipose tissue insulin resistance may be reversible. Thus, targeting ER stress may constitute a feasible strategy for treating obesity and insulin resistance, although it will be critical to understand the various mechanisms by which ER stress affects adipose tissue, including its effects on adipokine expression and function.

This work was supported by National Institutes of Health Grant P01 DK-49210 (to M.A.L.).

M.A.L. has consultative or advisory arrangements for which monetary compensation is received with Wyeth, Eli Lilly, Dr. Reddy's Laboratories, and Rigei Pharmaceuticals as well as grant support with Glaxo-Smith Kline. No other potential conflicts of interest relevant to this article were reported.

We thank Neal G. Copeland for kindly providing materials for recombineering for the resistin-luc construct.