The UPR (unfolded protein response), a cellular defence mechanism against misfolded protein accumulation in the ER (endoplasmic reticulum), is associated with many human diseases such as aging, cancer and diabetes. XBP1 (X-box-binding protein 1), a key transcription factor of the UPR, is critical in maintaining ER homoeostasis. Nevertheless, the mechanism by which XBP1 transcriptional activity is regulated remains unexplored. In the present study we show that XBP1s, the active spliced form of XBP1 protein, is SUMOylated, mainly by PIAS2 [protein inhibitor of activated STAT (signal transducer and activator of transcription) 2] at two lysine residues located in the C-terminal transactivation domain. Ablation of these SUMOylation events significantly enhances the transcriptional activity of XBP1s towards UPR target genes. Thus our results reveal an unexpected role for SUMO (small ubiquitin-related modifier) in the regulation of UPR activation and ER homoeostasis.
The IRE1α (inositol-requiring enzyme 1α)/XBP1 (X-box-binding protein 1) pathway, the most highly conserved signalling branch of the UPR (unfolded protein response) [1–3], is important for ER (endoplasmic reticulum) biogenesis and the secretory capacity of cells. Once activated, mammalian IRE1α splices 26 nucleotides from the Xbp1u (unspliced XBP1) mRNA, leading to a frameshift and the generation of an active bZIP (basic leucine zipper) transcription factor called XBP1s (spliced XBP1) that contains a C-terminal transactivation domain absent from XBP1u [4–9]. The XBP1s protein subsequently translocates to the nucleus and activates the transcription of genes involved in lipid biosynthesis, protein folding, degradation and trafficking [6,7,10,11].
Studies in animal models have revealed that the IRE1α/XBP1 pathway is essential for the development and function of the liver and heart, as well as ‘professional’ secretory cells, such as pancreatic and plasma cells [12–16]. XBP1s has also been shown to regulate hepatic lipogenesis in an UPR-independent manner . We showed recently that the IRE1α/XBP1 pathway is indispensable for adipocyte differentiation via XBP1s-mediated regulation of a key adipogenic factor C/EBPα (CCAAT/enhancer-binding protein α) . Despite its physiological significance, the mechanism by which the transcriptional activity of XBP1s protein is regulated remained largely unknown.
The post-translational modification of proteins by SUMO (small ubiquitin-like modifier) is an important transient regulatory mechanism in many cellular processes, most notably transcriptional regulation, DNA damage and signal transduction [19,20]. SUMO proteins have three isoforms (SUMO1–3), with 50% identity between SUMO1 and SUMO2/3. Similar to the process of ubiquitination, SUMO is first activated by the E1 enzyme Sae1 (SUMO1-activating enzyme subunit 1)–Uba2 (ubiquitin-like modifier activating enzyme 2), then is transferred on to the E2 enzyme Ubc9 and subsequently conjugated to target proteins via the activity of an E3 ligase, such as PIAS [protein inhibitor of activated STAT (signal transducer and activator of transcription)], RanBP2 (Ran-binding protein 2) or the polycomb group protein Pc2 [19,20]. In the present study we show that the transcriptional activity of XBP1s is negatively modulated by SUMOylation. Our results reveal an unexpected role for SUMO in the regulation of UPR activation.
Full-length expression constructs for XBP1s, XBP1u, ATF6 (activating transcription factor 6) and the ATF6 active form ATF6N (nuclear ATF6) were cloned from a liver cDNA library and subcloned into either pcDNA3/FLAG or pcDNA3/HA or pcDNA3/Myc vectors to tag the proteins. The truncated XBP1s encodes amino acids 1–260 of full-length XBP1s. XBP1s–GFP (green fluorescent protein) fusion constructs were made by subcloning the XBP1s PCR fragment into the pEGFP-N1 vector (Clontech). To construct the plasmid encoding the luciferase gene driven by four XBP1-binding sites, a synthetic oligonucleotide containing four XBP1-binding sites, 5′-aagctagccgcgTGGATGACGTGTACATGGATGACGTGTACATGGATGACGTGTACATGGATGACGTGTACAaagctttt-3′ (italicized sequences represent the XBP1-binding sites; lowercase sequences were added to provide restriction enzyme sites for NheI and HindIII, which are underlined), was digested with NheI/Hind III and then ligated into the NheI/HindIII site of the pGL3 vector (Promega). mPIAS1/2 (mouse PIAS1/2) was cloned from a mouse liver cDNA library template and subcloned into the pcDNA3/FLAG vector using the NotI and XbaI sites. Plasmids encoding hPIAS2a (human PIAS2a) and hPIAS2b were gifts from Jorma Palvimo (University of Kuopio, Kuopio, Finland). Plasmids encoding FLAG-tagged HDAC (histone deacetylase) 1, 2, 4, 5 and 6 in the pBJ5 vector, FLAG-tagged HDAC3 in the pcDNA3 vector and the β-galactosidase expression plasmid were gifts from Marc Montminy (Salk Institute, La Jolla, CA, U.S.A.). The FLAG–hPIAS4-, Myc–Ubc9- and HA (haemagglutinin)–SUMO1-expressing plasmids were gifts from Ling Cai (University of California San Diego, La Jolla, CA, U.S.A.). HA–SUMO2- and HA–SUMO3-expressing plasmids were gifts from Katherine Kieckhafer (Department of Molecular Biology and Genetics, Cornell University, New York, U.S.A.). The same amount of plasmids was used in all transfection experiments.
Cells and reagents
XBP1−/− MEFs (mouse embryonic fibroblasts; a gift from Laurie Glimcher, Harvard Medical School, Boston, MA, U.S.A.), macrophage RAW264.7 and HEK-293T [HEK (human embryonic kidney)-293 cells expressing the large T-antigen of SV40 (simian virus 40)] cells were maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) FBS (fetal bovine serum; Hyclone). The ER stress inducer Tg (thapsigargin; EMD Calbiochem) was dissolved in DMSO and used at 300 nM.
Site-directed mutagenesis to generate XBP1s mutants was performed using pcDNA3/FLAG–XBP1s as the template. The PCR included the mutagenic primers, as well as the Pfu polymerase (Stratagene) and the program consisted of 13 cycles of 95 °C for 30 s, 55 °C for 30 s and 72 °C for 5 min. After DpnI digestion for 2–3 h, the PCR product was transformed into XL-Blue competent cells by incubating plasmids or PCR products with the cells for 30 min on ice, followed by a heat-shock at 42 °C for 30 s. The correct introduction of the mutants was confirmed by DNA sequencing. All primers used in mutagenesis are listed in Supplementary Table S1 (at http://www.BiochemJ.org/bj/429/bj4290095add.htm).
HEK-293T cells were placed on to sterile cover slips coated with poly-L-lysine (Sigma–Aldrich). Cells were transfected with plasmids encoding the wild-type or mutant XBP1s–GFP fusion proteins. Cover slips were removed from dishes and mounted on to slides with Vectashield® and DAPI (4′,6-diamidino-2-phenylindole). The images were taken on a Leica DM5000 B microscope using a 63× oil objective.
HEK-293T cells were plated on to 10-cm-diameter dishes the day before transfection. Transfection was performed by preparing PEI (polyethylenimiene) (1 mg/ml in 0.9% NaCl) with plasmids at a 4:1 (μl/μg) ratio and adding to the cell culture. Transfected cells were scrapped off the plates with cold lysis buffer [50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1% (v/v) Triton X-100, 1 mM EDTA, 0.5 mM DTT (dithiothreitol) and protease inhibitor cocktail], containing 10 mM N-ethylmaleimide (Sigma–Aldrich) when detecting SUMOylation, and sonicated with a Branson digital sonifier 25 for 10 s. Lysate were incubated on ice for 15 min and centrifuged at 16000 g for 15 min. Anti-FLAG–agarose beads (Sigma–Aldrich) were added to the supernatants and rocked gently at 4 °C overnight. Following a wash with lysis buffer and two washes with wash buffer [20 mM Tris/HCl, pH7.5, 137 mM NaCl, 1% (v/v) Triton X-100, 2 mM EDTA, 10% (v/v) glycerol and 0.5 mM DTT], the beads were boiled and subjected to centrifugation (300 g for 1 min). Supernatants were analysed by SDS/PAGE (8% gels) and Western blotting.
Western blotting was performed using 15–30 μg of total cell lysates as described in . The antibodies were anti-FLAG antibody and anti-HA–HRP (horseradish peroxidase) antibody (both Sigma–Aldrich), rabbit anti-XBP1 antibody (Santa Cruz Biotechnology) and goat anti-(rabbit IgG)–HRP (BioRad).
Luciferase reporter assay
HEK-293T cells and XBP1−/− MEF cells were transfected on to 24-well for 24 h via PEI and Lipofectamine™ 2000 respectively. Cells were lysed in 100 μl of extraction buffer [Gly-Gly buffer (Fisher Scientific), 1% (v/v) Triton X-100 and 1 mM DTT]. Lysates (50 μl) were transferred on to a 96-well plate containing 50 μl of assay mix (Gly-Gly buffer, 16 mM K2HPO4, 2 mM DTT and 2.5 mM ATP). Luciferin mix (50 μl of a 0.4 mM solution; Xenogen) was added to each well through an injector and readings were collected using the Synergy 2 plate reader (BioTek). Luciferase activity was normalized to activity from a co-transfected CMV (cytomegalovirus)-promoter-driven β-galactosidase expression plasmid.
RNA extraction and qRT-PCR (quantitative real-time PCR)
Total RNA was extracted using TRIzol® according to the supplier's protocol (Molecular Research Center) and reverse transcribed using the Superscript III kit (Invitrogen). cDNAs were analysed using the SYBR Green PCR system on the Roche LightCycler 480 machine. All qRT-PCR results were normalized to the level of the ribosomal l32 gene. Primer sequences are listed in Supplementary Table S1.
Results are expressed as means±S.E.M. Comparisons between groups were made by unpaired two-tailed Student's t tests. P<0.05 was considered as statistically significant. All experiments were repeated at least twice and representative results are shown.
XBP1s of the UPR is SUMOylated
To identify potential UPR transcription factors that could be SUMOylated, we transfected HEK-293T cells with vectors encoding XBP1s, ATF6 and ATF6N, plus the E2 enzyme Ubc9 and SUMO1. Among the three factors, XBP1s protein was the only one which underwent SUMOylation (Figure 1A). To confirm further the specificity of SUMOylation in the UPR, we found that XBP1s, but not XBP1u, was SUMOylated by all three isoforms of SUMO, i.e. SUMO1 and SUMO2/3 (Figure 1B). Indeed, in a recent genome-wide analysis of SUMO2 modifications during the heat-shock response, XBP1s was identified as a protein that was SUMOylated by SUMO2 . In line with previous reports [22–24], SUMOylation increased the molecular mass of XBP1s protein by approx. 20 kDa. XBP1s SUMOylation was not cell-type specific as it was also detected in MEFs and COS7 cells (results not shown). Taken together, our results show that the XBP1s protein of the UPR is SUMOylated.
XBP1s is SUMOylated
XBP1s is SUMOylated at lysine residues in the C-terminal transactivation domain
SUMOylation often occurs at the lysine residues in the motif ΨKXE/D (where Ψ is a hydrophobic residue, X is any residue and E/D is aspartate/glutamate) [25,26]. Two highly conserved SUMOylation motifs were identified at residues Lys276 and Lys297 within the transactivation domain of the XBP1s protein, a region that is absent from the XBP1u protein (Figure 2A). We performed site-directed mutagenesis to target these residues and found that introducing a K276R mutation in XBP1s abolished one SUMOylation event (Figure 2B). Interestingly, the K297R mutation completely abolished SUMOylation on XBP1s, suggesting that SUMOylation at Lys297 is required for SUMOylation at Lys276. To exclude the possibility of non-specific modification events, we determined that the K298R mutation, i.e. the X in the ΨKXE motif, had no effect on the SUMOylation of the XBP1s protein (Figure 2B). Thus our results show that the XBP1s protein is SUMOylated at two lysine residues located in the C-terminal transactivation domain.
XBP1s is SUMOylated at Lys276 and Lys297 in the C-terminal transactivation domain
XBP1s protein is SUMOylated endogenously
To determine whether the endogenous SUMOylation machinery could mediate SUMOylation of XBP1s, cells were transfected with various XBP1s plasmids in the absence of exogenous Ubc9 and SUMO1. Indeed, both wild-type and K298R XBP1s were SUMOylated; the K276R mutant was modified to a lesser extent and no SUMOylation occurred in the K297R mutant or K276R/K297R double-mutant (Figure 3A). The differences between the XBP1 and SUMO blots were highly reproducible and specific; therefore we speculated that this was probably because the anti-SUMO antibody preferentially recognizes the bi-SUMOylated protein (Figure 3A). Furthermore, SUMOylated endogenous XBP1s protein could be detected following immunoprecipitation of XBP1 in macrophage cells treated with the ER stress inducer Tg (Figure 3B). Nonetheless, ER stress itself did not increase the SUMOylation of XBP1s as Tg treatment failed to alter the amount of SUMO–XBP1s in HEK-293T cells transfected with XBP1s (Figure 3C). Thus our results suggest that the XBP1s protein may be constitutively SUMOylated during ER stress.
Endogenous SUMOylation of XBP1s
SUMO E3 ligase PIAS2 interacts with and mediates SUMOylation of XBP1s
To identify the E3 ligase responsible for SUMOylation of XBP1s, we transfected HEK-293T cells with XBP1s and various E3 ligase PIAS constructs (PIAS1, 2a/b, 3 and 4). Strikingly, XBP1s protein interacted specifically with PIAS2 proteins, either mPIAS2 or its human homologue hPIAS2b. Interactions with other PIAS proteins, including PIAS1, 3, and 4 were either much weaker or not detected (Figure 4A and results not shown). Similar observations were observed for XBP1u (Figure 4B), suggesting that the N-terminal region, containing the DNA-binding domain shared by both XBP1u and XBP1s (Figure 2A), may mediate the interaction between XBP1s and PIAS2. The interaction of XBP1s or XBP1u protein with hPIAS2a was much weaker compared with that for mPIAS2 or hPIAS2b (Figures 4A and 4B). It is worth pointing out that the same amount of plasmids for XBP1s was used in all experiments and the difference in the level of XBP1 protein (Figures 4A and 4B) was probably due to the stabilization of XBP1s proteins by the interacting proteins.
SUMO E3 ligase PIAS2 interacts with and increases SUMOylation of XBP1s
As splicing variants of PIAS2a (PIASxα) and PIAS2b (PIASxβ) differ from each other at the C-terminal serine/threonine (S/T)-rich domain (Figure 4C), our results suggest that this domain of the PIAS protein may partially mediate the interaction with XBP1s. Furthermore, and supporting its role in SUMOylation of XBP1s protein, overexpression of hPIAS2b increased SUMOylation of wild-type but not K276R/K297R XBP1s proteins (Figure 4D). Thus our results suggest that mPIAS2 or hPIAS2b interacts with and mediates SUMOylation of XBP1s.
SUMOylation does not alter the nuclear localization of XBP1s
SUMOylation of transcription factors often leads to alterations in the intracellular localization of target proteins . To address this, we expressed wild-type or mutant XBP1s–GFP fusion proteins in HEK-293T cells. As with wild-type XBP1s proteins, SUMOylation mutant proteins predominantly localized to the nucleus and did not exhibit altered intracellular distribution (Figure 5A). Thus our results suggest that SUMOylation does not affect the nuclear localization of XBP1s protein.
SUMOylation of XBP1s does not affect its nuclear localization nor its interaction with HDAC5
SUMOylation does not affect the interaction between XBP1s and HDAC5
SUMOylation of transcription factors may also alter their interaction with transcriptional co-repressors such as HDACs [27–29]. Indeed, out of all the HDACs (HDAC1–6), XBP1s strongly interacted with HDAC5 (Figure 5B). This interaction was not affected by mutations at the SUMOylation sites (Figure 5C). Thus our results show that XBP1s interacts with HDAC5 and this interaction is not affected by SUMOylation of XBP1s.
SUMOylation negatively regulates transcriptional activity of XBP1s
We next determined the physiological significance of SUMOylation on the transcriptional activity of XBP1s. First, we transfected wild-type or mutant XBP1s into XBP1−/− MEFs together with a luciferase reporter driven by a synthetic promoter containing four-tandem XBP1-binding sites (5′-TGACGT-3′) . Wild-type XBP1s increased luciferase activity over 10-fold (Figure 6A). In line with the requirement of the C-terminal domain of XBP1s for transcriptional activation, loss of the C-terminal 100 amino acids of XBP1s (i.e. a truncated form of XBP1s protein) reduced the activity by 50% (Figure 6A). Mutations at either Lys276 or Lys297 increased XBP1s transcriptional activity by 50 and 100% relative to wild-type XBP1s respectively, whereas mutations at both residues further increased its transcriptional activity (Figure 6A). In contrast, mutation of Lys298 had no effect. Similar observations were obtained in HEK-293T cells (Figure 6A). It remains unclear why K276R/K297R XBP1s has higher transcriptional activity compared with K297R XBP1s as both mutants abolish SUMOylation of XBP1s (Figure 2B). We speculate that the Lys276 residue in the K297R mutant may still be SUMOylated at a low efficiency beyond the detection limit or be modified via a SUMO-independent mechanism, which may negatively regulates its transcriptional activity.
SUMOylation represses transcriptional activity of XBP1s
To exclude further the possibility of SUMO-independent effects at the two lysine residues (Lys276 and Lys297), we mutated the glutamate residue in the SUMO motifs (ΨKXE) and replaced it with an alanine or glutamine residue, i.e. E278A/Q and E299A/Q. These mutants exhibited similar transcriptional activity as the lysine mutant located in the same motif (Figure 6B), pointing to a key role of both SUMO motifs in regulating XBP1s activity. Thus these results demonstrate unequivocally that SUMOylation negatively regulates transcriptional activity of XBP1s.
We next determined the effect of SUMOylation on the expression of known XBP1s target genes. To this end, we transfected wild-type or mutant XBP1s into XBP1−/− MEFs and measured the expression of endogenous UPR genes. Mutations of Lys276 or Lys297, but not of Lys298, strongly induced the expression of several well-known XBP1 target genes , including Edem (ER degradation enhancer, mannosidase), Herp (homocysteine-induced endoplasmic reticulum protein) and P58ipk, by nearly 30–50%, but not the ATF6 target gene Grp78 (glucose-regulated protein 78) (Figure 6C). Taken together, our results show that SUMOylation of XBP1s down-regulates its transcriptional activity.
In the present study we show that a key transcription factor of the UPR signalling pathway is subjected to covalent post-translational modifications via SUMOylation. Our results showed that XBP1s protein, generated upon activation of IRE1α RNase activity during the ER stress response or UPR, is SUMOylated at two lysine residues in the C-terminal transactivation domain and, that these SUMOylation events have additive suppressive effects on XBP1s transcriptional activity. Thus the transcriptional activity of XBP1s protein is regulated not only by the IRE1α-mediated splicing event, but also by post-translational modifications, such as SUMOylation.
Our results adds XBP1s to a growing list of transcription factors whose activity is regulated by SUMOylation and suggests an unexpected role for SUMOylation in regulating ER homoeostasis and UPR activity. As SUMOylation is a highly transient post-translational modification event, it may serve as a mechanism to further fine-tune the transcriptional activity of XBP1s and, in turn, affect the outcome of the UPR. Nonetheless, similar to the scenario for most other transcription factors that are regulated by SUMOylation [19,20], the detailed molecular mechanism by which SUMOylation affects XBP1s transcriptional activity requires further investigation.
Previous studies have established the indispensible role of the IRE1α/XBP1 branch of the UPR in plasma cell and adipocyte differentiation [15,16,18,31]. During plasma cell differentiation, accumulation of XBP1s protein induces a series of transcriptional events leading to enlarged cell size, expanded ER membrane/volume and increased secretory capacity . On the other hand, during adipocyte differentiation, accumulation of XBP1s protein increases the expression of C/EBPα, a key adipogenic factor, which subsequently controls cellular adaptation to a mature adipocyte phenotype . Further studies are required to determine whether SUMOylation of XBP1s plays a role in cellular differentiation processes.
The UPR, a critical cellular defence mechanism against the accumulation of misfolded proteins and aggregates in the ER, has been implicated in many human diseases, collectively termed ‘conformational diseases’, including diabetes, cancer and neurodegeneration . When the UPR is overwhelmed or fails, it often leads to cellular dysfunction and cell death [11,33]. As our knowledge of the role of UPR and XBP1s in human diseases improves [34,35], the ability to modulate the transcriptional activity of XBP1s may be an important first-step towards improving therapeutical strategies targeting dysfunctional UPR.
activating transcription factor 6
CCAAT/enhancer-binding protein α
Dulbecco's modified Eagle's medium
green fluorescent protein
glucose-regulated protein 78
HEK (human embryonic kidney)-293 cells expressing the large T-antigen of SV40 (simian virus 40)
homocysteine-induced endoplasmic reticulum protein
inositol-requiring enzyme 1α
mouse embryonic fibroblast
protein inhibitor of activated STAT
quantitative real-time PCR
Ran-binding protein 2
signal transducer and activator of transcription
small ubiquitin-related modifier
unfolded protein response
X-box-binding protein 1
Hui Chen and Ling Qi conceived the project. Hui Chen performed the experiments and analysed the results. Ling Qi wrote the paper and with assistance from Hui Chen.
We thank Laurie Glimcher, Jorma Palvimo, Marc Montminy, Ling Cai and Katherine Kieckhafer for cell lines and plasmids; Yin He for critical reading of the manuscript; and other members of the Qi laboratory for insightful comments and suggestions.
This study was supported by the American Federation for Aging Research [grant number RAG08061] and the National Institute of Diabetes and Digestive and Kidney Disease [grant number R01DK082582]. L.Q. is the recipient of the Rosalinde and Arthur Foundation/American Federation for Aging Research New Investigator Award in Alzheimer's Diseases and the American Diabetes Association (ADA) Junior Faculty Award.