The short-chain fatty acid receptor GPR43 is transcriptionally regulated by XBP1 in human monocytes

G-protein coupled receptor 43 (GPR43) recognizes short chain fatty acids and is implicated in obesity, colitis, asthma and arthritis. Here, we present the first full characterization of the GPR43 promoter and 5′-UTR. 5′-RACE of the GPR43 transcript identified the transcription start site (TSS) and a 124 bp 5′-UTR followed by a 1335 bp intron upstream of the ATG start codon. The sequence spanning -4560 to +68 bp relative to the GPR43 TSS was found to contain strong promoter activity, increasing luciferase reporter expression by >100-fold in U937 monocytes. Stepwise deletions further narrowed the putative GPR43 promoter (−451 to +68). Site-directed mutagenesis identified XBP1 as a core cis element, the mutation of which abrogated transcriptional activity. Mutations of predicted CREB, CHOP, NFAT and STAT5 binding sites, partially reduced promoter activity. ChIP assays confirmed the binding of XBP1 to the endogenous GPR43 promoter. Consistently, GPR43 expression is reduced in monocytes upon siRNA-knockdown of XBP1, while A549 cells overexpressing XBP1 displayed elevated GPR43 levels. Based on its ability to activate XBP1, we predicted and confirmed that TNFα induces GPR43 expression in human monocytes. Altogether, our findings form the basis for strategic modulation of GPR43 expression, with a view to regulate GPR43-associated diseases.

S tudies on knockout mice have identified Free Fatty Acid Receptor 2 (FFAR2 or GPR43) as a critical gene in the prevention of obesity, colitis, asthma and arthritis [1][2][3][4][5][6][7][8] . GPR43 is a G protein coupled-receptor that is activated by mid-micromolar concentrations of short-chain fatty acids (SCFAs) -namely acetate, propionate and butyrate. While the liver metabolism of ethanol can generate micromolar concentrations of acetate in the blood, by far the most abundant source of SCFAs in the human body is the colonic lumen, where hundreds of millimolars are continuously being produced during the anaerobic fermentation of dietary fibre by saccarolytic gut bacteria 9,10 . These gut SCFAs have been found to beneficially modulate blood glucose and lipid levels, the colonic environment, and immune functions [11][12][13] . As an SCFA receptor, GPR43 has already been shown to mediate some of these beneficial effects, with knockout mice studies confirming a role in obesity and inflammation (Table 1).
Consistent with its role as an SCFA receptor, GPR43 expression is found in cells that are exposed to the highest concentrations of SCFAs. These include the cells of the distal ileum, colon and adipose tissue, with the highest expression found in immune cells such as monocytes and neutrophils [14][15][16] . In addition, GPR43 expression appears to be modulated during inflammation since immune challenge by lipopolysaccharide (LPS) or treatment with granulocyte-macrophage colony stimulating factor (GM-CSF) raises GPR43 transcript levels in human monocytes 17 . This tissue-specificity suggests that GPR43 expression is tightly regulated and may be important for its function. Indeed, the compelling outcomes of Gpr43 knockout (Table 1) imply that proper regulation of GPR43 expression is pertinent to the normal functioning of a range of physiological processes, and consequently, targeting of GPR43 expression in diseases would provide new therapeutic potential. Thus, details on the factors involved in cell-type specific expression of the GPR43 gene under pathophysiological conditions remain an unexplored and intriguing area of investigation. Here, we characterized the human GPR43 gene, identifying the promoter and enhancer sequence elements, as well as the critical transcription factors and signalling pathways that regulate GPR43 expression.

Results
PMA-differentiated U937 monocytes are a suitable model for GPR43 expression. To understand the mechanisms underlying the specific expression of GPR43, we first confirmed the cell type with the highest level of GPR43 mRNA. Consistent with previous studies 14- 16 , we found human peripheral blood monocytes and neutrophils to express the highest level of GPR43 mRNA (Fig. 1). PMA-mediated monocytic differentiation of the human promonocytic cell line, U937 18,19 , led to a 100-fold increase in the transcription of GPR43, yielding mRNA levels that was comparable to the peripheral blood monocytes and neutrophils. Hence, we reasoned that the U937 cell line would be a suitable model to study GPR43 transcriptional regulation in leukocytes.
GPR43 transcription start site is located at 1459 bp upstream of the ATG start codon. To define the GPR43 putative promoter site, we first performed a 59-RACE of the GPR43 transcript, generating an approximately 450 bp product which was sequenced to reveal a 124 bp 59-UTR upstream of the GPR43 ATG start codon (Fig. 2a). Mapping of this sequence with the human genome database (NCBI Ref Seq: NC_000019.10) revealed the presence of a 1335 bp intron flanked by the GPR43 59-UTR and the start codon (Fig. 2b). Thus, the transcription start site (TSS) is located 1459 bp upstream of the ATG start codon. While it was difficult to judge the presence of other 59-RACE products (Fig. 2a, Lane 1), we note that a previously reported northern blot analysis (Senga et al., 2003 17 , Fig. 2h), detected two GPR43 transcripts, with the shorter (by about a few hundred bases) product being many times more abundant than the longer transcript.
It is possible that the two transcript sizes described by Senga et al., 2003 17 are due to different 59-UTR lengths, which can possibly result from alternate splicing 20 or promoters 21 . The transcript detected by the 59 RACE here is likely the shorter, more abundant, transcript. Our attempts to detect the presence of other possible 59-UTR species via RACE were thus far unsuccessful (data not shown), presumably due to the low abundance. Overall, these results characterize the 59 region of the human GPR43 gene and allow the putative core promoter to be identified for subsequent analysis of transcriptional elements.
The core and proximal promoter of GPR43 are located within 2451 to 233 from the TSS. Upon identification of the TSS, we cloned a putative promoter region (spanning 24560 to 168 bp relative to the TSS) from primary human monocyte genomic DNA and analyzed its promoter activity using dual luciferase reporter system. This putative promoter contained aligned sequences with greater than 70% sequence identity to the corresponding mouse sequence (Fig. 3a), suggesting conservation with involvement in transcriptional regulation 22 . Indeed, compared to the vector alone, the promoter insert raised the luciferase reporter activity by .100fold (Fig. 3b). Deletion from positions 24560 to 2451 revealed no significant change (p-values . 0.05) in luciferase activity, implying that this region lacks regulatory elements, which is consistent with the general lack of sequence conservation observed (Fig. 3a). Although two short sequence alignments do appear in this region, we note that the corresponding mouse sequences are much further away from the mouse Gpr43 gene (0.6 Mbp and 1.7 Mbp, respectively), suggesting a lack of involvement in mouse Gpr43 transcriptional control. Deletion from positions 2451 to 2182 resulted in a marked drop (p-value 5 0.07) in activity while further deletions downstream of position 2182 resulted in significant reductions in activity (p values , 0.005), suggesting that the region spanning 2451 to 258 might contain crucial activator elements. This is consistent with the observed interspecies sequence conservation (Fig. 3a). Strikingly, the near abrogation of promoter activity upon deletion of 258 to 233 suggests that the core promoter element(s) lies within this 25 bp region.
XBP1 is part of the core promoter while CREB, CHOP, NFAT and STAT5 act as enhancers. By in silico predictions using the Mat-Inspector program 23 , we identified 8 transcription factor (TF) recognition sites with high matrix similarity of $90% within the 519 bp putative promoter (Fig. 4a). The null mutations of Tax/CREB, CHOP, NFAT and STAT5/5B resulted in significant losses in activity compared to the wild type (WT) promoter (Fig. 4b), indicating that these TFs may be important enhancers for the control of GPR43 expression. Conversely, no significant loss in activity was found for mutations in the IRF4, PU.1 and c-Rel recognition sites. Mutation of the XBP1 recognition site resulted in near abrogation of promoter activity (Fig. 4b), hence identifying it as a crucial core cis element. Notably,  53 Increased lipolysis and plasma free fatty acids Bjursell et al. 2011 6 Improved glucose control and reduced body fat mass on a high fat diet Tolhurst et al. 2012 7 Impaired glucagon-like peptide-1 secretion and glucose tolerance Kimura et al. 2013 8 Increased fat accumulation and obesity on a normal diet  this XBP1 recognition site resides within the 25 bp region that was deleted from the 126 bp promoter (Fig. 3b). XBP1 binding also appears to be required for the low level GPR43 expression observed in A549 adenocarcinomic human alveolar basal epithelial cells ( Supplementary Fig. S1) since the XBP1 null mutation also led to near abrogation in promoter activity in the A549 cells.
The activities of p38 and PLC are required for GPR43 transcription.
A number of the identified TFs are known to be regulated by MAPK and PLC/PKC pathways [24][25][26] . To determine if these pathways are involved in the modulation of GPR43 expression, we blocked the putative pathways with small molecule inhibitors (Fig. 4c, 4d). The relative expression of housekeeping genes was not markedly affected (Supp. Fig. 2). Both SB203580 and U73122, inhibitors of p38 and PLC respectively, attenuated GPR43 transcription in LPS-challenged monocytes while only SB203580 attenuated GPR43 transcription in unchallenged monocytes (Fig. 4c, 4d). This suggests that the p38 pathway is required for basal transcription of GPR43 which is active with or without LPS-challenge while the PLC-PKC pathway is important only during LPS-mediated up-regulation of GPR43. This may be due to p38 pathway activation of XBP1 26 , which is part of the core promoter regulating basal transcription (Fig. 4b); while the PLC-PKC pathway activates NFAT 25 , which we find to act as an enhancer (Fig. 4b). The p38 pathway also activates CHOP 24 . Separately, the proteasome inhibitor, MG132, reduced GPR43 mRNA levels with or without LPS treatment while Wortmannin, an AKT inhibitor, increased GPR43 transcript levels. Overall, our findings indicate that activation of the GPR43 expression is mediated by the p38 and PLC/ PKC pathways, and transcriptional regulation may possibly also involve the proteasome degradation and PI3/Akt pathways.
ChIP analysis confirmed XBP1 binding to GPR43 promoter in vivo. Since the XBP1 binding sites were found to be necessary for promoter activity, we investigated XBP1 binding to the endogenous GPR43 promoter. Chromatin immunoprecipitation (ChIP) was performed with antibodies against XBP1. The GPR43 promoter XBP1 binding site was significantly enriched in the resulting immunoprecipitate, consistent with strong association of XBP1 to that site (Fig. 5).
The pull down appears specific as no significant enrichment was detected for the negative control genomic regions and no GPR43 promoter enrichment was observed when the isotype control antibody was used.

XBP1-siRNA knockdown or overexpression alters GPR43 expression.
Next, we sought to confirm that XBP1 activity is required for the endogenous transcription of GPR43. Stable knockdown of XBP1 in U937 cells led to a reduction in GPR43 expression levels relative to the control siRNA ( Fig. 6a, b). Conversely, in the A549 adenocarcinomic human alveolar basal epithelial cell line, which expresses low levels of XBP1-dependent GPR43 mRNA ( Fig. 1 and Supplementary  Fig. S1), the overexpression of the active spliced form of XBP1 (XBP1s) raised GPR43 expression levels ( Fig. 6c, d). Thus, we have confirmed that XBP1 modulates endogenous GPR43 expression.
GPR43 expression is up-regulated by XBP1 activators. The XBP1mediated transactivation of GPR43 prompted us to examine whether GPR43 expression is increased by known activators of XBP1. LPS is a known activator of XBP1 27 while GM-CSF is a known positive regulator of dendritic cell development 28 , a process which requires XBP1 29 . We found that LPS treatment increased the GPR43 transcription by at least 7-fold while GM-CSF raised transcription by 5fold ( Fig. 7a). Both stimuli were previously reported to induce GPR43 expression via an unknown mechanism 17 . However, our results now revealed this to be attributable to XBP1 activation. Besides LPS, treatment with TNFa, another known activator of XBP1 30 , also up-regulated GPR43 transcription by 5-fold. Thus, we have demonstrated that pathways leading to the activation of XBP1 consistently up-regulate GPR43 transcription.

Discussion
In this study, we identified and characterized the GPR43 promoter, revealing that XBP1 acts as a core cis promoter element while CREB, CHOP, NFAT and STAT5 act as enhancers (Fig. 2, 4a, 4b). Based on its known ability to activate XBP1, we accurately predicted that TNFa up-regulates GPR43 expression (Fig. 7b). This suggests that other regulators of XBP1 activity may likewise be involved in the regulation of GPR43 expression. We also showed that GPR43 expression was altered by inhibition of pathways involved in the activation of CHOP and NFAT (Fig. 4c, 4d). Our identification of the pathways known to activate these GPR43 promoter elements would facilitate the application of stimuli and conditions to alter GPR43 expression and its subsequent functions, with a view to therapeutic developments. The indispensability of XBP1 for GPR43 promoter activity has important implications since XBP1 is involved in a number of physiological processes and diseases which include B cell development and the unfolded protein response (UPR) 31 . It now appears likely that these physiological processes may engage the expression and function of GPR43. Notably, mutations in the XBP1 gene locus of human patients has been associated with increased risk of inflammatory bowel disease (IBD), Crohn's disease and ulcerative colitis, while Xbp1 2/2 mice have increased susceptibility to colitis 32 . Xbp1 2/2 mice also display impaired glucose and insulin tolerance upon high fat diet-induced obesity 33 . The association of XBP1 with gut inflammation and obesity may be partially attributable to the disrupted expression of GPR43, since Gpr43 2/2 mice are similarly more susceptible to colitis 1-4 as well as exhibiting impaired glucose tolerance and increased obesity 7,8 . Considering the implications of XBP1 and GPR43 in inflammation and obesity, it might be pertinent to further investigate this hypothetical link.
Besides XBP1, the GPR43 promoter was also found to be up-regulated by Tax/CREB, CHOP, NFAT and STAT5 (Fig. 4). Tax/CREB is a heterodimer consisting of viral oncoprotein, Tax, and the host derived CREB 34 . NFAT and STAT5 have been extensively implicated in immunity and cancer, through regulating cell development, growth and cytokine production [35][36][37][38] . CHOP plays a role in ER stress-induced apoptosis and cytokine production 39 . However, in comparison with XBP1, mutations of the promoter binding sites for the aforementioned four TFs only resulted in relatively modest decrease of ,30-40% in the promoter activity (Fig. 4b). It is hence likely that these TFs are involved in fine-tuning the expression levels of GPR43 in monocytes, or, they may be more active in upregulating the gene under the appropriate external stimuli (e.g. cytokine stimulation, ER stress or viral infection). Following the same line of argument, we cannot rule out the potential involvement of other predicted transcription factors, e.g. IRF4, c-Rel (component of NF-kB), and Pu.1 (Fig. 4b), in the expression of GPR43 under the appropriate immune contexts. This will be a subject of future investigation. The ability of inflammatory stimuli such as LPS, TNFa and GM-CSF (Fig. 7a) to up-regulate GPR43 expression infers a number of interesting possibilities. Our findings imply that the positive correlation between GPR43 expression and inflammation at the fetal cell membranes 40 or TNFa levels in adipocytes 41 , is likely due to the upregulation of GPR43 by the inflammatory stimuli present. Notably, GM-CSF has been found to be beneficial for the treatment of IBD 42,43 . Since GPR43 is also implicated in the etiology of IBD 1-4 , it would be  interesting to investigate whether GM-CSF exerts any beneficial therapeutic effects through modulating GPR43 expression. In view of the low plasma levels of SCFA (,200 mM) 9,10 and low potencies of the ligands in activating GPR43 (generally in the mid micromolar range) [14][15][16] , the higher levels of expression of GPR43 in the peripheral blood monocytes might thus enhance the potency of SCFAs on the GPR43-mediated inflammation.
The increase in GPR43 expression observed in phosphoinositide-3 kinase (PI3K) inhibition under basal or immune-challenged conditions (Fig. 4c, 4d) suggests that there was a relief of suppression that is mediated by the PI3K/Akt pathway. However, the current information is insufficient to deduce whether this is a direct or indirect effect from the PI3K/Akt signalling. Nevertheless, it should be noted that activation of the PI3K/Akt pathway may lead to the inhibition of p38 44 . Hence, the increase may be due to the recovery of p38, causing an indirect activation of GPR43 expression.
In addition to characterizing the promoter, we provide the first description of the 59-UTR of the human GPR43 gene, identifying an intron between the 59-UTR and the start codon. Our findings present a new perspective on the human GPR43 gene, which was initially described as being intronless, downstream of the ATG start codon 45 . Notably, this gene organization is also seen in the bovine GPR43 gene, the only other GPR43 homolog which has its 59-UTR sequenced 46 . Such cross-species similarities in the gene organization and high sequence conservation of the entire core promoter across human and mouse (Fig. 2a) suggests that the same regulatory elements are conserved in the bovine and murine promoters. Indeed, bovine and murine tissue specific expression of GPR43 have been found to be highly similar to that of the human counterpart [14][15][16]46 . A recent study reported transcriptional activity from a luciferase assay of a 500 bp sequence immediately upstream of the human GPR43 start codon in mouse RAW264.7 macrophages 47 . Interestingly, these researchers predicted an NF-kB binding site within this 500-bp sequence, although the mutation was reported to cause only 10% reduction in the promoter activity. Perhaps, inclusion of the GPR43 TSS (located more than 1 kb upstream of the predicted NF-kB binding site as reported by our study) would offer a more complete analysis of the effect of this putative enhancer.
While the findings on the GPR43 promoter are mostly obtained from human monocytes in this study, it would be interesting to explore if the promoter elements identified, regulate GPR43 in other cell types. Our findings on A549 cells suggest that this may indeed be the case. In A549 cells, the 519 bp GPR43 promoter construct upregulates luciferase expression; an up-regulation that is abrogated by XBP1 null mutation (Supp. Fig. S1). These findings mirror those of U937 cells (Fig. 4b) although differences are also noted. Notably, while a 50% knockdown of XBP1 mRNA led to an 80% drop in expression of GPR43 in the U937 cell, a 100-fold increase in XBP1 mRNA only resulted in a 2-fold increase in A549 expression of GPR43. This discrepancy in effect may be due to a variety of factors. XBP1 activity in A549 may already be near saturation and hence further increases in XBP1 expression will not proportionally increase GPR43 expression. Another possibility is that XBP1 may require further activation, such as by the p38 pathway 26 , to become fully activated. Such pathways may be more active in monocytes. The GPR43 promoter in A549 cells may also be epigenetically silenced through DNA methylation or chromatin remodeling, such that further increases in XBP1 expression may not fully overcome this silencing. Despite these differences, our findings confirm that XBP1 can serve as a regulator involved in controlling GPR43 expression in other cell types. Notably, both GPR43 15,41,48 and XBP1 49 are found to be up-regulated in adipocytes and adipocyte-derived cell lines, suggesting that XBP1 may also be involved in regulating the expression of GPR43 in adipocytes, an interaction that may have implications in adipocyte function.
In conclusion, we present the first full characterization of the human GPR43 promoter, revealing that GPR43 expression is regulated by XBP1 as a core cis element while CREB, CHOP, NFAT and STAT5 act as enhancers. We show that by distinguishing pathways known to activate these GPR43 promoter elements, it is possible to predict stimuli and conditions that may alter GPR43 expression and function, thus providing novel drug targets for the treatment of GPR43-associated diseases such as obesity, colitis, asthma and arthritis.

Methods
Isolation of peripheral blood monocytes, cell culture and differentiation of U937 cells. Peripheral blood mononuclear cells were isolated from the buffy coat of healthy adult donors (National University of Singapore Blood Donation Centre). Ficoll-Paque PREMIUM (GE Healthcare) gradient centrifugation was performed according to the manufacturer's instructions. Briefly, the mononuclear cell layer was isolated and washed with PBS supplemented with 2% FBS (GE healthcare) and 1 mM EDTA, to remove platelets. Monocytes were subsequently purified by negative selection using the Human Monocyte Enrichment Kit (StemCell Technologies) according to the manufacturer's instructions.  Primary monocytes and U937 cell line were cultured in RPMI (Life Technologies) while A549 cells were cultured in DMEM (Life Technologies) supplemented with 10% FBS and 1% (v/v) penicillin and streptomycin (Life Technologies). The cells were grown at 37uC and 5% CO 2 . For differentiation of U937 into monocytes, 5 3 10 5 cells/ mL of pre-differentiated cells were induced with 30 ng/mL phorbol 12-myristate 13acetate (PMA) (Sigma-Aldrich) for 24 h, before changing to fresh media. The cells were cultured for another 48 h to allow for full differentiation and thereafter used for downstream assays.
The primary neonatal human fibroblasts (Life Technologies) were routinely grown in medium 106 (Life Technologies) before RNA extraction.
59-Rapid Amplification of cDNA Ends (59 RACE). Amplification of 59 cDNA ends of mature GPR43 was performed with the GeneRacer TM kit (Life Technologies, Cat. #L1-502-02) following manufacturer's instructions. Briefly, 5 mg of U937 total RNA was dephosphorylated with Calf Intestinal Phosphatase (CIP) and ligated with a sequence specified RNA oligonucleotide at 37uC for 1 h. Following ligation, the 59 ends of GPR43 mRNA were reverse transcribed and amplified using nested PCR primers (Supplementary Table S1). The PCR product was then analyzed on a 1.5% agarose gel. The sequence of the GPR43 59 end was confirmed using Big Dye Terminator cycle sequencing kit and ABI 3100 Genetic Analyser (Life Technologies).
Construction of GPR43 promoter and deletion mutants. To generate promoter deletion constructs, primary monocyte genomic DNA was extracted from the interphase and phenol layer using TRIzol reagent (Life Technologies), according to the manufacturer's instructions. To obtain 59 end promoter deletion products, primers (Supplementary Table S1) flanking the desired promoter regions were used for PCR amplification, using the purified genomic DNA as template. PCR was performed using iProof High Fidelity DNA Polymerase (Bio-rad) under the following parameters: initial denaturation at 98uC; 40 cycles of amplification, denaturation at 98uC for 10 s, annealing at Tm of primer pair 13uC for 30 s, elongation at 72uC for 2.5 min; followed by a final extension at 72uC.
The isolated promoter lengths were separately cloned into pGL4.20 vector (Promega) using standard molecular cloning techniques. The restriction enzymes used were XhoI and HindIII (Thermo Fisher Scientific). T4 DNA ligase was from Roche. For small-scale purification of plasmids, AxyPrep Plasmid Miniprep kit (Axygen Biosciences) was used. Large-scale purification of plasmids intended for transfection was carried out using PureLink HiPure Plasmid Filter Purification kit (Invitrogen). The full-length sequence of each promoter construct was confirmed by sequencing using Big Dye Terminator cycle sequencing kit and ABI Prism 3100 Genetic Analyzer (Life Technologies).
In silico predictions of transcription factor binding sites and site-directed mutagenesis. Transcription factor binding sites (TFBS) along the delineated promoter regions were identified using the MatInspector programme available online at http://www.genomatix.de/ 23 . Analysis of the sequences was performed using the MatInspector TFBS, weight matrix library version 9.0. Only predicted TFBS with a core matrix similarity of $0.95 and an overall matrix similarity of $0.90, were considered significant.
Promoters containing the specific mutated TFBS were synthesized by Integrated DNA technologies. The mutated promoters were cloned into pGL4.20 vector (Promega) and confirmed by sequencing as described above.
Transient transfection and luciferase reporter assay. The full-length promoter, deletion and mutant constructs were transiently transfected into differentiated U937 cells at 48 h after removal of PMA, with X-tremeGENE HP DNA Transfection Reagent (Roche) according to the manufacturer's guidelines. Promoter-pGL4.20 vector constructs and pRL-CMV control Renilla luciferase vector (Promega) were transfected in the molar ratio of 1051.
Renilla luciferase reporter activities were assessed using the Dual Luciferase Reporter Assay System (Promega) 22 h after transfection. Luminescence was detected using the Glomax 20/20 Luminometer (Promega), and the resulting measurements from the Firefly luciferase were normalized to the Renilla luciferase. Relative light units were calculated using readouts from the pGL4.20-Basic (promoter-less) vector or the wild type promoter as baseline for deletion and mutation constructs, respectively.

Treatment of cells with inflammatory stimuli and pathway activation/inhibition.
For stimulation of cells, Escherichia coli 055:B5 LPS and sodium acetate were purchased from Sigma-Aldrich while GM-CSF and TNFa were from Life Technologies. Purified human monocytes were seeded at a density of 4 3 10 6 cells/mL in 24-well plates and cultured in the presence of either LPS (100 ng/mL), PMA (30 ng/mL), TNFa (10 ng/mL), GM-CSF (100 ng/mL) or sodium acetate (10 mM) for 3 h before RNA extraction with TRIzol reagent (Life Technologies).
To identify regulatory pathways involved in the basal regulation of gene expression, the same conditions above were applied except that replicate wells of cells treated with MG132 (10 mM) and Forskolin (20 mM) were lysed and analyzed for mRNA expression after 4 h without LPS treatment. For treatment with other inhibitors, monocytes were lysed for RNA extraction and analyzed after 12 h in culture.
Quantitative PCR (qPCR). Total RNA was isolated using the TRIzol reagent according to the manufacturer's instructions and the procedure was repeated to ensure thorough removal of genomic DNA. cDNA synthesis of the extracted RNA was performed using Superscript III First Strand Synthesis kit (Life Technologies). Quantitative PCR was performed using GoTaq qPCR Master Mix (Promega) and forward and reverse primers (Supplementary Table S1) with LightCycler 480 system (Roche). Spliced XBP1 was detected using the Taqman gene expression assay (Life Technologies, Hs03929085_g1) and Light Cycler Probes Master (Roche). To obtain relative target mRNA folds, cycle thresholds (Ct) were normalized to the Ct of Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or Ribosomal protein L27 (RPL27) reference genes and expressed as fold-change by the 2 2nnCt method 50 .
Western Blotting. Cell lysates were separated by SDS-PAGE and transferred onto PVDF membranes (Bio-rad). The blots were then probed with antibodies against XBP1 (Abcam, ab37152), or b-actin (Sigma, a2066) followed by the corresponding secondary antibodies according to respective manufacturer's instructions. Following incubation with WesternBright ECL chemiluminescent substrate (Advansta), chemiluminescent signals was detected with an ImageQuant TM LAS 4000 mini (GE Healthcare).
ChIP assay. ChIP assay was performed as described previously 51 . Differentiated U937 cells were cross-linked with 1% formaldehyde for 10 min at room temperature and then quenched with 0.2 M of glycine. Chromatin extracts from lysed cells were sonicated to an average DNA fragment length of 350 bp and immunoprecipitated with Santa Cruz Biotechnology antibodies anti-XBP1 (sc-7160) or IgG isotype control (sc-2027). Following de-crosslinking, quantitative PCR was performed with multiple ChIP primers (Supplementary Table S1) to quantitate the relative occupancy of the GPR43 promoter region. Fold enrichment was obtained by normalizing the amount of precipitated DNA to that of the input sample and then expressed relative to the GPR43 negative control coding region.
siRNA knockdown and XBP1s overexpression. For stable knockdown of XBP1, siRNA sequences Sense: 59-AAAGACAGCAAGTGGTAGATTTA-39; Anti-sense: 59-AAAATAAATCTACCACTTGCTGT-39 52 was cloned into pFIV-H1/U6 vector (System Biosciences), while a non-targeting sequence was cloned as the negative control vector. The resulting siRNA vector construct or pFIV-H1/U6-copGFP vector (System Biosystems), which contains CopepodGFP coding sequence in replacement of the puromycin selection marker, was co-transfected with lentiviral packaging vectors pFIV-34N and pVSV-G into HEK-293T cells with TurboFect transfection reagent (Thermo Scientific). Culture supernatant was collected after 48 h and used to resuspend U937 cells, with the addition of polybrene to a final concentration of 8 mg/ mL. The resulting suspensions were then centrifuged at 1000 g for 2 h at 37uC, after which the transduced cells were re-suspended in 2 mL of fresh complete RPMI media and seeded into 6 well plates. Successfully transduced cells were selected with 1 mg/ mL puromycin (Life technologies) in culture medium 48 h after transduction, for 1.5 weeks.