Potentiation of ERα transcriptional activity by XBP-1S and XBP-1U
To investigate the effects of the XBP-1 proteins on ERα transcriptional activity, human breast cancer MDA-MB-435 cells, which lack the ERα, were co-transfected with the estrogen response element-containing reporter ERE-LUC, ERα and increasing amounts of XBP-1S or XBP-1U. As shown in Figure A, in the absence and presence of E2, 0.5 µg of XBP-1S enhanced the transcriptional activity of ERα 32- and 15-fold, respectively, whereas 0.5 µg of XBP-1U only enhanced ERα transcriptional activity 2.1- and 2.2-fold, respectively. This might be explained by the structural differences between the C-terminal transactivation domains of XBP-1S and XBP-1U. Both XBP-1S and XBP-1U increased ERα transcriptional activation in a dose-dependent manner (Fig. A). It is important to note that the magnitude of the ligand-independent activation of the ERα by XBP-1S (0.5 µg) was 5-fold higher than that observed with E2 (10 nM), and that XBP-1S and estrogen were synergistic in stimulating estrogen-regulated transcription. The activity of the ERE-LUC reporter activated by XBP-1S and XBP-1U decreased dramatically in the absence of exogeneous ERα (Fig. B), indicating that ERα itself was required for the maximal effects of XBP-1S and XBP-1U on ERE-LUC reporter transcription.
Figure 1 XBP-1S and XBP-1U enhance ERα-mediated transactivation in MDA-MB-435 cells. (A) Effects of XBP-1S and XBP-1U on ERα-mediated transactivation. Cells were co-transfected with 0.2 µg of ERE-LUC, 50 ng of the expression plasmid for (more ...)
To examine the effects of antiestrogens on ERα transactivation by XBP-1S and XBP-1U, MDA-MB-435 cells were co-transfected with the ERE-LUC reporter, ERα and XBP-1S or XBP-1U, and subsequently treated with antiestrogens, 4-hydroxytamoxifen (4-OHT) and ICI 182,780 (Fig. C). Both ICI 182,780 and 4-OHT completely blocked the effects of XBP-1U on ERα transcriptional activity in the presence or absence of E2, whereas both ICI 182,780 and, to a lesser extent, 4-OHT reduced but did not abolish the ability of XBP-1S to transactivate ERα.
To determine whether the observed effects of XBP-1S and XBP-1U on ERα transactivation were specific to MDA-MB-435 cells, human breast cancer cell line MCF-7 and human embryonic kidney cell line 293T were used in co-transfection experiments. Although ERα transactivation by XBP-1S in MCF-7 and 293T cells was of a lower magnitude than in MDA-MB-435 cells, similar results were observed (Table ). The transcriptional activity of ERα co-activated by XBP-1S was always greater than that co-activated by XBP-1U. Thus, XBP-1S and XBP-1U can act as positive regulators of ERα-dependent transcriptional activation in a variety of mammalian cell lines. To verify that this E2-independent enhanced transcriptional activity was not a result of increased ERα protein production, we examined protein expression in whole-cell extracts using western blotting analysis. Figure shows that ERα levels were not increased by XBP-1S or XBP-1U expression. In addition, the greater effects of XBP-1S on ERα transcriptional activation were also not attributable to its high expression level. Conversely, the expression level of XBP-1S was lower than that of XBP-1U (Fig. ). Together, our results suggest that XBP-1S and XBP-1U could function as a co-regulator to enhance ERα transcriptional activity in a ligand-independent manner.
XBP-1S and XBP-1U enhance ERα-mediated transactivation
Figure 2 Western blotting showing the ERα, XBP-1S and XBP-1U protein levels in 293T cells. Cells were transfected as in Table . Whole-cell extracts were prepared, and equivalent amounts of each extract were probed with anti-ERα (more ...)
To determine the specificity of XBP-1S and XBP-1U in ERα-mediated transactivation, we tested the effects of XBP-1S and XBP-1U on AR-mediated transactivation using the ARE-containing reporter construct PSA-LUC. As expected, transcription of the reporter was induced by an androgen, R1881, in MDA-MB-435, MCF-7 and 293T cells (Table ). XBP-1S and XBP-1U failed, however, to enhance AR-mediated transcription in both the presence and absence of R1881. Conversely, in the presence and absence of R1881, XBP-1S decreased AR transcriptional activity ~2- and 4-fold, respectively, in MCF-7 and 293T cells (Table ). This result suggests that XBP-1S and XBP-1U may not be general co-activators for steroid receptors.
XBP-1S and XBP-1U do not enhance AR-mediated transactivation
Both the N- and C-terminal domains of ERα contribute to ERα transactivation by XBP-1S and XBP-1U
To determine which domain of ERα is involved in the co-activation of ERα by XBP-1S and XBP-1U, ERα constructs containing N-terminal AF-1 and DNA-binding domain (DBD) (ABC domain) or C-terminal AF-2 and DBD (CDEF domain) were co-expressed with XBP-1S or XBP-1U in MDA-MB-435 cells. XBP-1S and, to a lesser extent, XBP-1U enhanced both the constitutive transactivation activity of AF-1 and ligand-dependent transcriptional activity of AF-2 (Fig. ). However, co-expression of XBP-1S or XBP-1U with full-length ERα resulted in a more efficient enhancement of the reporter transcriptional activity (2- to 5-fold), compared with ABC and CDEF domains expressed separately (Fig. ). Thus, both N- and C-terminal domains of ERα contribute to ERα transcriptional activity by XBP-1S and XBP-1U.
Figure 3 Both N- and C-terminal domains contribute to ERα transcriptional activity regulated by XBP-1S and XBP-1U. MDA-MB-435 cells were co-transfected with 50 ng of the expression vector for ERα, ERα ABC domain or ERα CDEF domain, (more ...)
Cooperative co-activation of ERα by XBP-1 and SRC-1
SRC-1 is a member of the p160 steroid receptor coactivator (SRC) family (33
). SRC-1 interacts with ERα and stimulates E2
-mediated gene transcription. SRC-1 enhances the interaction between the N-terminal AF-1-containing and C-terminal AF-2-containing regions of the ERα, allowing for full ERα activation. To determine whether XBP-1 plays a role in SRC-1-mediated co-activation of the ERα, MDA-MB-435 cells were co-transfected with the ERE-LUC reporter construct and expression vectors for XBP-1 and SRC-1 (Table ). As expected, in the absence of E2
, transfection of XBP-1S or XBP-1U alone co-activated ERα transcriptional activity, whereas transfection of SRC-1 did not have a significant effect. Co-transfection with XBP-1S or XBP-1U, plus SRC-1 expression vectors gave greater than additive effects of XBP-1S or XBP-1U and SRC-1 measured independently. In the presence of E2
, XBP-1S, XBP-1U and SRC-1 all enhanced ERα transcriptional activity to different levels. When SRC-1 was co-expressed with either XBP-1S or XBP-1U, however, the ERα transcriptional activity was cooperatively enhanced. Therefore, the synergistic enhancement of ERα activity by co-expression of SRC-1 and XBP-1S or XBP-1U is ligand independent.
Synergistic enhancement of ERα activity by SRC-1 and XBP-1
Interaction of XBP-1S and XBP-1U with ERα in vitro and in vivo
Our observation that XBP-1S and XBP-1U could function as a co-regulator to enhance ligand-independent ERα transactivation suggested that XBP-1S and XBP-1U might physically interact with ERα. To test this possibility, GST pull-down experiments were performed in which in vitro translated [35S]methionine-labeled XBP-1S or XBP-1U was incubated with full-length GST–ERα. The binding of XBP-1S and XBP-1U to GST–ERα, but not to GST, was observed in both the absence and presence of E2 (Fig. A). E2 did not increase the interaction of XBP-1S and XBP-1U with GST–ERα. Notably, the binding of XBP-1S to ERα was stronger than that of XBP-1U to ERα.
Figure 4 XBP-1S and XBP-1U bind to ERα in vitro and in vivo. (A) Interaction of XBP-1S and XBP-1U with ERα in vitro. GST pull-down assay was performed as described in Materials and Methods. Full-length GST–ERα fusion proteins, (more ...)
To determine whether XBP-1S and XBP-1U interact with ERα in vivo, 293T cells were transfected with ERα and FLAG-tagged XBP-1S or XBP-1U and cultured in both the absence (Fig. B) and presence of 10 nM E2 (data not shown). FLAG-XBP-1S or XBP-1U was immunoprecipitated from cell lysates by an anti-FLAG antibody and analyzed for ERα binding by western blotting analysis. The results showed that ERα could be co-immunoprecipitated in a ligand-independent manner in the presence, but not in the absence, of FLAG-XBP-1S or FLAG-XBP-1U (Fig. B, and data not shown). Consistent with the co-activation results, the binding of XBP-1S to ERα was stronger than that of XBP-1U to ERα. The in vivo interaction of XBP-1S and XBP-1U with ERα was unlikely to be mediated by nucleic acids, as it was not affected by the treatment with ethidium bromide that disrupts DNA–protein interaction (data not shown).
Mapping of the ERα and XBP-1 interaction domains
To determine which region of ERα binds to XBP-1S or XBP-1U, GST pull-down experiments were performed. As shown in Figure C, the GST–ERα(180–282) containing the DBD bound specifically to in vitro translated [35S]methionine-labeled XBP-1S or XBP-1U, but the GST-ERα(1–185) containing the AF-1 and the GST–ERα(282–595) containing the AF-2 did not. Consistent with the in vivo binding results, XBP-1S interacted with GST–ERα(180–282) in vitro more strongly than XBP-1U.
XBP-1S and XBP-1U are proteins of 261 and 376 amino acids, respectively, with an identical N-terminus (amino acids 1–164). To delineate the domains in the XBP-1S and XBP-1U that mediate the protein–protein interaction with ERα, a series of mutant GST–XBP-1 fusion proteins were used in GST pull-down experiments (Fig. D). Deletion of the XBP-1 N-terminal amino acids (1–82) reduced but did not abolish the ability of the XBP-1 proteins to bind to the ERα. Either region (amino acids 1–101, containing the bZIP domain, or either amino acids 148–376 or amino acids 148–261, containing the transactivation domain) of XBP-1S or XBP-1U was sufficient for ERα binding. However, the full-length GST–XBP-1 interacted with ERα more strongly than any GST–XBP-1 fragments. In addition, GST–XBP-1S associated with ERα more strongly than GST–XBP-1U.
The N-terminal portion of XBP-1 is required for ERα transactivation function
To test the possibility that the maximal interaction of XBP-1S and XBP-1U with ERα is required for the enhancement of ERα transcriptional activation, mutants of XBP-1S and XBP-1U (ΔXBP-1S and ΔXBP-1U) were made in which the N-terminal region from amino acids 1 to 82 was deleted. MDA-MB-435 cells were co-transfected with the ERE-LUC reporter, ERα and either FLAG-tagged XBP-1S, XBP-1U, ΔXBP-1S or ΔXBP-1U. As shown in Figure A, the mutations lacking some of the ERα-binding sites completely abolished the ERα transcriptional activation in a ligand-independent manner. This is not attributable to decreased expression of the ΔXBP-1S and the ΔXBP-1U deletion mutants. In contrast, ΔXBP-1S and ΔXBP-1U were expressed at higher levels than XBP-1S and XBP-1U (Fig. B). Taken together, these findings suggest that the XBP-1S and XBP-1U action of ERα by their maximal binding contributes to the transactivation function of ERα.
Figure 5 The deletion mutants of XBP-1S and XBP-1U abolished the ERα transactivation. (A) MDA-MB-435 cells were co-transfected with 0.2 µg of ERE-LUC, 50 ng of the expression plasmid for ERα and 0.5 µg of the expression vector (more ...)
Neither XBP-1S nor XBP-1U binds to ERE
The transcription factor XBP-1 was found to bind preferably to the cAMP responsive element (CRE)-like element GATGACGTG(T/G)nnn(A/T)T (34
). ERα binds to the ERE GGTCAnnnTGACC. Since both XBP-1 and ERα target sequences have the sequence TGAC, we tested whether XBP-1S and XBP-1U bind to the consensus ERE, using a gel shift assay. As expected, the 32
P-labeled ERE, but not mutant ERE (EREM), bound to in vitro
-translated ERα in the absence or presence of E2
(Fig. and data not shown). The binding was specifically inhibited by a 100-fold molar excess of a cold ERE oligonucleotide. Moreover, neither XBP-1S nor XBP-1U bound to the ERE.
Figure 6 Neither XBP-1S nor XBP-1U binds to ERE. Gel shift assay was performed as described in Materials and Methods. The 32P-labeled ERE probe was incubated with the in vitro-translated ERα, XBP-1S and XBP-1U proteins as indicated, in the presence of (more ...)
XBP-1S and XBP-1U mRNAs are expressed in breast cancer cell lines
Human XBP-1 (XBP-1U) was originally isolated as a transcription factor which binds to the X2 box present in the promoter region of human major histocompatibility complex (MHC) class II genes (14
). More recently, XBP-1U mRNA was found to be spliced in response to the endoplasmic reticulum stress, resulting in XBP-1S (18
). To determine whether XBP-1S mRNA is expressed in breast cancer cells, we prepared mRNA from five ER-positive (MCF10A, T47D, MCF7, ZR75-1 and BT474) and five ER-negative (MDA-MB-436, MDA-MB-435, MDA-MB-231, SKBR3 and MDA-MB-453) breast cancer cell lines and performed RT–PCR using primers specific for the unique XBP-1S region. As shown in Figure , in addition to XBP-1U mRNA, we could also detect XBP-1S mRNA in all of the breast cancer cell lines tested. The identity of XBP-1S detected was further confirmed by DNA sequencing. XBP-1S was also expressed in the immortalized normal breast epithelial cell line MCF-10A, although at a lower level. Since a limited number of breast cancer cell lines were examined, there is no significant correlation between XBP-1 and ERα expression. However, both XBP-1S and XBP-1U were widely expressed in breast cancer cells. Interestingly, XBP-1S was also expressed in mammary and ovary two-hybrid cDNA libraries, which were prepared from the corresponding tissues of individuals who died from trauma, indicating that XBP-1S is more widely expressed than previously thought. After searching expressed sequence tags (ESTs) for XBP-1S in GenBank (www.ncbi.nlm.nih.gov/blast
), we also found that human XBP-1S mRNA is expressed in breast adenocarcinoma, lung tumor, stomach, brain, skin and pancreas, and mouse XBP-1S is expressed in mammary tumor, lung tumor metastatic to mammary, and neural retina. Taken together, these data suggest that, like XBP-1U, XBP-1S may play a critical role under physiological and pathological conditions.
XBP-1 mRNA expression in breast cancer cell lines. RT–PCR from the selected cell lines and cDNA libraries was performed as described in Materials and Methods. β-Actin was used as an internal control.