Oncoprotein HBXIP Modulates Abnormal Lipid Metabolism and Growth of Breast Cancer Cells by Activating the LXRs/SREBP-1c/FAS Signaling Cascade

Cancer cells are distinct from normal cells in their ability to sustain chronic proliferation (1). Self-sufficiency in growth signals, insensitivity to antigrowth signals, and abnormal metabolism are three traits of cancer cells and account for the constitutive and rapid proliferation of cancer cells (2). FAS acts as a key enzyme of fatty acid biosynthetic pathway that catalyzes de novo lipid synthesis for lipogenesis and is exquisitely sensitive to nutritional signals regulation in the liver (3). FAS expression is very low or undetectable in most normal adult tissues for majority of fatty acids are obtained from dietary sources (4). FAS as a metabolic oncogene is highly expressed in cancers and tumor-associated FAS prefers to confer growth and survival advantages rather than functions as an anabolic energy storage pathway (5). Specially, FAS provides sufficient lipids for membrane biogenesis and lipid-based posttranslational modification of proteins in highly proliferating tumor cells (4). However, almost all fatty acids in tumors derived from de novo synthesis and abnormally activated lipogenesis pathways are insensitive to nutritional signals, despite adequate nutritional supply (6).

FAS is primarily regulated by transcriptional regulator sterol regulatory element binding protein-1c (SREBP-1c; ref. 7). Liver X receptors (LXR) belong to a class of ligand-dependent nuclear receptors (NR), which induce genes controlling cholesterol homeostasis and lipogenesis (8). LXRs mainly regulate transcription of SREBP-1c through LXR response element (LXRE) in SREBP-1c promoter, leading to a subsequent transcription of lipogenic genes (9). LXRs with ligands recruit nuclear coactivators with NR motif to activate target genes transcription, and LXRs lacking ligands suppress the transcription by recruiting nuclear corepressors with corepressor/nuclear receptor (CoRNR) motif (10, 11). Thus, the activation of LXRs/SREBP-1c system is tightly responsive to nutritional regulation for maintaining FAS homeostasis in normal cells.

Hepatitis B X-interacting protein (HBXIP), as a conserved protein among mammalian species, was originally identified by its interaction with hepatitis B virus X protein, suppressing the replication of hepatitis B virus (12) and regulating the centrosome duplication in HeLa cells (13). It has been reported that HBXIP is required for mTORC1 activation by amino acid in cell growth (14). Importantly, our study has revealed that HBXIP requires the interaction with c-Fos and phosphorylation of both proteins for nuclear import (15) and acts as a coactivator for transcription factors including CREB, TFIID, Sp1, E2F1, and STAT4 in the development of breast cancer (15–18). Therefore, we supposed that HBXIP might play a crucial role in abnormal lipid metabolism of breast cancer.

In this study, we investigated the effect of HBXIP on abnormal lipid metabolism in breast cancer. Our data show that HBXIP is able to upregulate FAS in breast cancer cells. Interestingly, HBXIP containing CoRNR motif with special flanking sequence can coactivate LXRs independent of ligand. Thus, HBXIP contributes to the abnormal lipid metabolism through triggering LXRs/SREBP-1c/FAS signaling in breast cancer cells.

Thirty human breast cancer tissues were collected for qPCR analysis. Breast carcinoma tissue microarray includes 40 cases of normal breast tissues, 10 cases of hyperplasia lobular tissues, 10 cases of fibroadenoma tissues, 10 cases of atypical duct hyperplasia tissues, 40 cases of ductal carcinoma in situ (DCIS) tissues, and 236 cases of clinical invasive breast cancer tissues with information of patients' overall survival and progression-free survival (Supplementary Table S1). Another breast carcinoma tissue microarray contained 87 cases of clinical breast carcinoma tissues were used for FAS × HBXIP cross-tabulation. Multiple organ tissue microarray contained tumor tissues and corresponding normal tissues (Supplementary Table S4), such as cerebrum cancer and adjacent cerebrum tissues (n = 26/16), thyroid cancer and adjacent thyroid tissues (n = 27/16), lung cancer and adjacent lung tissues (n = 41/16), colon cancer and adjacent colon tissues (n = 27/14), prostate cancer and adjacent prostate tissues (n = 20/17). The tissue microarrays (no. AM08C19, no. AM08C20, no. AM08C21, and no. AM08C22) were purchased from Aomeibio Company. All samples were approved by Ethics Committee of Hospital providing tissues.

The cell lines such as MCF-7, SK-BR3, HBL-100, MCF-10A, 184A1, Hs478T, T47D, BT-549, Chang liver, HepG2, BGC-823, HCT116, and SW480 were purchased from ATCC and cultured according to the ATCC protocol. We finished the related experiments within 6 months after we purchased the cell lines. All cell lines were analyzed and authenticated by targeted genomic and RNA sequencing. The SK-BR3 cells (or MCF-7 cells) stably overexpressing control or HBXIP were established by transfecting with the pCMV-tag2B empty vector or the pCMV-HBXIP vector, respectively, and G418 selection. To knock down specific genes, 2′-O-methyl siRNAs targeting HBXIP, SREBP-1c, LXRα or LXRβ, FAS, and control 2′-O-methyl siRNA were transfected into the SK-BR3 cells stably overexpressing control or HBXIP, respectively. The 2′-O-methyl siRNAs targeting HBXIP 3′-UTR, LXRα, LXRβ, and control siRNAs were purchased from Riobio Company. The sequences of siRNA for RXRα, HBXIP, FAS, and SREBP-1c have been described previously (13, 19–21).

BALB/c athymic nude mice (15 g/mouse; Experiment Animal Center of Peking, China) were housed and treated according to guidelines of the NIH Guide for the Care and Use of Laboratory Animals. We conducted the animal transplantation according to the Declaration of Helsinki. MCF-7 cells were harvested and resuspended at 5 × 10⁷ per mL with sterile PBS. Groups of 4-week-old female BALB/c athymic nude mice (each group, n = 6) were subcutaneously injected at the shoulder with 0.2 mL of the cell suspensions. Tumor growth was measured after 3 days from injection and then every 3 days. Tumor volume (V) was monitored by measuring the length (L) and width (W) with calipers and calculated with the formula (L × W²) × 0.5. After 21 days, tumor-bearing mice and controls were sacrificed, and the tumors were excised and measured.

Data were analyzed by Student t test. Statistically significant P values were indicated as follows: *, P < 0.05; **, P < 0.01; ***, P < 0.001; and NS, not significant. Spearman (or Spearman rank) correlation coefficient and Pearson correlation coefficient were used for relation statistics. Statistical analysis of the tissue microarray was performed with the χ² test or Kruskal–Wallis test using the SPSS software program (SPSS).

The details of other materials and methods [constructions, immunohistochemistry, assay of FAS activity, total RNA isolation, reverse-transcription PCR (RT-PCR) and quantitative real-time PCR (qPCR), nucleocytoplasmic fractionation and Western blot analysis, luciferase reporter gene assay, electrophoretic mobility shift assays (EMSA), immunoprecipitation and GST pull-down, immunofluorescence staining and confocal microscopy, Oil Red O staining] are described in the Supplementary Information.

We first examined the expression of HBXIP in multiple cancers by immunohistochemical analysis using human tissue microarrays. Our data showed that the positive rate of HBXIP was 87.5% (329/376) in breast cancer tissues (atypical duct hyperplasia, DCIS, and invasive breast cancer), in which the strong positive rate was 54.0% (203/376), but only 9.4% (6/64) in normal breast tissues (Supplementary Fig. S1A; Supplementary Table S1). In addition, the positive rates of HBXIP were 77.5% (31/40) in cerebrum tumor, 100% (41/41) in thyroid cancer, 87.7% (50/57) in lung cancer, 92.0% (138/150) in colon cancer, 98.0% (197/201) in prostate cancer, and 95.0% (19/20) in testis cancer (Supplementary Fig. S1A; Supplementary Table S2). Moreover, higher expression of HBXIP was related to the higher histologic grade and larger tumor size in 236 breast cancer patients (Supplementary Table S3). Then, the analysis of 236 breast cancer patients for 80 months indicated that the high level of HBXIP was significantly correlated with the decreased overall survival and progression-free survival (Fig. 1A). Our data indicate that the HBXIP is an important marker of evaluating risk of mortality and recurrence of breast cancer patients.

Many studies have demonstrated that FAS is highly expressed in human cancers and associated with the survival of cancer cells. Therefore, we speculated that HBXIP might be associated with FAS in the development of cancer. The immunohistochemical analysis manifested that the positive rate of FAS was 94.6% (70/74) in HBXIP-positive tissues of breast cancer (Fig. 1B; Supplementary Table S4). In addition, the protein expression level and enzyme activity of FAS in cancer tissues and their adjacent normal tissues revealed that FAS was biologically active in cancer tissues (Supplementary Fig. S1B and S1C). Moreover, the positive rates of FAS were 93.1% (95/102) and 92.9% (184/198) in HBXIP-positive tissues of colon carcinoma and prostate cancer (Supplementary Fig. S1D; Supplementary Table S4). To further explore the relevance of HBXIP and FAS, we performed qPCR analysis in 30 clinical breast cancer tissues. The levels of HBXIP mRNA exhibited a significant correlation with those of FAS (Spearman correlation coefficient of 0.7630, P < 0.001; Fig. 1C). In addition, we confirmed the correlation in 8 breast cancer cell lines with a Spearman correlation coefficient of 0.9666 (P < 0.001; Fig. 1D). Therefore, the expression of HBXIP is positively correlated with that of FAS in breast cancer and negatively associated with survival of breast cancer patients.

Next, we assessed the effect of HBXIP on FAS in breast cancer MCF-7 and SK-BR3 cells. The overexpressed HBXIP could upregulate FAS in MCF-7 cells, whereas the depletion of HBXIP downregulated FAS in SK-BR3 cells in a dose-dependent manner at the levels of mRNA, protein, and promoter (Fig. 2A), suggesting that HBXIP was able to upregulate FAS in breast cancer cells. Meanwhile, the transfection or interference efficiency of HBXIP was validated by Western blot analysis (Supplementary Fig. S2A). In addition, silencing HBXIP could downregulate FAS expression in colon cancer HCT116 and SW480 cells (Supplementary Fig. S2B).

Previously, we reported that HBXIP acted as an oncogenic coactivator for transcription factors in breast cancer (16, 22–24). Accordingly, we performed chromatin immunoprecipitation (ChIP) assays to determine whether HBXIP was involved in the transcriptional regulation of FAS. However, HBXIP was unavailable to bind to the core region of the FAS promoter in SK-BR3 cells (Fig. 2B). It has been reported that the transcription factors SREBP-1a, SREBP-1c, and ChREBP are responsible for FAS transcription (25). Thus, we proposed that HBXIP might activate FAS through those transcription factors. As expected, ChIP assays showed that HBXIP could bind to the SREBP-1c promoter (Fig. 2B). Then, knockdown of SREBP-1c abolished HBXIP-enhanced FAS promoter activities (Fig. 2C), and overexpressed HBXIP could upregulate SREBP-1c at the mRNA level in SK-BR3 cells (Supplementary Fig. S2C). These results suggest that HBXIP is able to upregulate SREBP-1c in breast cancer cells. We further validated that HBXIP could upregulate SREBP-1c including the precursor form (125 kDa) and the active form (68 kDa) in MCF-7 cells (Fig. 2D), whereas silencing HBXIP led to decrease of SREBP-1c in SK-BR3 cells (Fig. 2E and Supplementary Fig. S2D). In addition, knockdown of HBXIP remarkably decreased SREBP-1c promoter activities in hepatoma HepG2 cells, gastric cancer BGC-823 cells, colon cancer HCT116 and SW480 cells (Supplementary Fig. S2E). Moreover, HBXIP could upregulate other SREBP-1c target genes, such as acetyl-CoA carboxylase (ACC) and stearoyl-CoA desaturase-1 (SCD1), at the levels of mRNA (Supplementary Fig. S2F), suggesting that HBXIP was able to upregulate other SREBP-1c target genes involved in de novo fatty acid synthesis. Collectively, we conclude that HBXIP activates FAS through upregulation of SREBP-1c in breast cancer.

It has been reported that HBXIP is also known as LAMTOR5, which is an activator of mTORC1 signaling to activate SREBP-1 (14, 26, 27). Therefore, we detected the effect of HBXIP on SREBP-1c at the mRNA level in the cells treated with rapamycin, which could inhibit the mTORC1 signaling. The result indicated that rapamycin could not inhibit the upregulation of SREBP-1c mediated by HBXIP (Supplementary Fig. S2G), suggesting that HBXIP activated SREBP-1c independent of mTOR signaling. To clarify the other mechanism by which HBXIP activated SREBP-1c promoter, we mapped the specific HBXIP binding sites in the upstream loci of SREBP-1c promoter using electrophoretic mobility shift assay (EMSA). Four approximately 190 bp DNA fragments (termed C1–C4) with approximately 20 bp overlaps, derived from the core region (−703 ∼ +1) of SREBP-1c promoter, were used as probes in EMSA. Our data showed that two shifted bands of C2 fragment were remarkably impaired when HBXIP antibody was added (Fig. 3A). Then, we divided the C2 fragment into seven DNA fragments with a length of approximately 33 bp. Among these seven short probes, only two DNA fragments E1 (−260 ∼ −222) and E2 (−335 ∼ −296) containing LXREs could interact with HBXIP (Supplementary Fig. S3A). But HBXIP failed to bind to the LXRE mutants (E1-mut3 and E2-mut1) by EMSA (Fig. 3B). Luciferase reporter gene assays demonstrated that HBXIP could activate the reporters of E1 and E2 (4 × E1 and 4 × E2), but not their mutants (4 × E1-mut3 and 4 × E2-mut1; Fig. 3C and D). However, EMSA exhibited that the purified HBXIP alone failed to bind to the E1 and E2 DNA fragments (Supplementary Fig. S3B). Sequential ChIP experiments and qPCR analysis demonstrated that HBXIP and LXRs simultaneously occupied the loci contained LXRE (Fig. 3E and Supplementary Fig. S3C), suggesting that HBXIP coactivated LXRs in activation of SREBP-1c promoter. Moreover, the GST-LXRs were able to bind to LXRE when T-0901317 (a synthetic ligand of LXRs) and/or His-HBXIP was added (Supplementary Fig. S3D), suggesting that HBXIP could bind to LXRE of SREBP-1c promoter with LXRs. Therefore, we conclude that HBXIP occupies LXRE in SREBP-1c promoter involving LXRs.

Next, we discerned whether HBXIP could interact with LXRs in LXRs activation. LXRs belong to a class of ligand-dependent NRs, which induce certain genes controlling cholesterol homeostasis and lipogenesis (8). Coimmunoprecipitation (co-IP) assays in vivo demonstrated that HBXIP interacted with either LXRα or LXRβ in SK-BR3 and MCF-7 cells (Fig. 4A). Moreover, GST pull-down assays in vitro revealed that both GST-LXRα and GST-LXRβ interacted with fusion-protein His-HBXIP (Fig. 4B). Confocal images further validated that HBXIP and LXRα/β colocalized in the nucleus with a significant overlap (Fig. 4C), suggesting that HBXIP was able to directly bind to LXRs in breast cancer cells. Interestingly, ChIP assays revealed that HBXIP failed to bind to the SREBP-1c promoter when LXRα and LXRβ were simultaneously knocked down by siRNA in SK-BR3 cells, and vice versa (Fig. 4D and E). Meanwhile, the HBXIP-enhanced activities of SREBP-1c promoter could be attenuated in the event (Fig. 4F). Moreover, overexpression of LXRα and LXRβ failed to enhance the activities of SREBP-1c promoter when HBXIP was knocked down in the cells (Fig. 4G). The transfection efficiency of LXRα/β (and/or interference efficiency of HBXIP) was validated by qPCR analysis (Supplementary Fig. S4A). Meanwhile, the above observation was confirmed at the levels of mRNA (Supplementary Fig. S4A). Moreover, we performed a Gal4 fusion protein approach to detect the effect of HBXIP on Gal4-LXR constructs in MCF-7 cells. The results revealed that HBXIP was able to activate Gal4-LXRs as a coactivator (Supplementary Fig. S4B), further supporting that HBXIP could directly coactivate LXRs in breast cancer cells. However, Western blot analysis showed that overexpression or knockdown of HBXIP failed to affect the expression of LXRs in breast cancer cells (Supplementary Fig. S4C), suggesting that HBXIP activated SRBPE-1c promoter through coactivating LXRs, rather than elevating LXRs expression in the cells. Taken together, HBXIP coactivates LXRs through directly interaction with LXRs to enhance the transcription activity of SREBP-1c promoter.

To identify the function domain of HBXIP in coactivation of LXRs, we analyzed the amino acid sequence of HBXIP by bioinformatics. The coding region of the human HBXIP gene is 522 bp (GenBank ID NM_006402.2) and encodes a protein of 173 amino acids, in which the homologies at the C-terminal 91 amino acids (83–173) containing a leucine zipper relative to human HBXIP sequence are 100% for mouse, 98.9% for rat, and 84% for Xenopus, respectively (Fig. 5A), suggesting that the fragment of 83–173 amino acids in HBXIP may have important functions. Interestingly, we observed that there was a CoRNR motif (L/VXXI/VI) IVGV in the C-terminal 91 amino acids of HBXIP. Generally, the corepressors of NR with CoRNR motifs (L/VXXI/VI) bind to unliganded NR and suppress the target gene transcriptions (28, 29). In addition, Hu X and Lazar MA have reported that sequences flanking the CoRNR box determine NR specificity, the designed mutants of flanking sequences of the CoRNR motif are able to convert the corepressors of NR into the coactivators (29). Accordingly, we propose that HBXIP may bind to LXRs via its CoRNR motif with flanking sequences for coactivation of LXRs. Thus, we divided HBXIP into four fragments and found that the fragment of HBXIP (aa 83–144) containing a CoRNR motif could increase the activities of SREBP-1c promoter (Fig. 5B). Moreover, we designed three mutants of amino acids 83–144 of HBXIP. The mutant “AVGAA” (termed mut-1) replaced the wild-type CoRNR motif “IVGVL”. The mutant of N-terminal (mut-2) substituted the polar amino acids for nonpolar amino acids. The mutant of C-terminal (mut-3) was a truncation with amino acids deleted in C-terminal of CoRNR. co-IP assays showed that only mut-1 failed to interact with LXRα in SK-BR3 cells (Fig. 5C), and all of the mutants failed to activate the SREBP-1c promoter (Fig. 5D), suggesting that the CoRNR motif with its flanking sequences of HBXIP was responsible for interacting with and coactivating LXRs in breast cancer cells.

The LXR ligands can induce a conformation change of LXRs to disassociate LXRs and corepressors (9, 30). In this study, our data showed that T-0901317 dose dependently impaired the interaction of HBXIP and LXRs in vivo and in vitro (Fig. 5E and F). Physiologically, LXRs acting as a ligand-dependent transcription factor activate a number of the promoters of lipid metabolism gene through binding to LXRE (31). Therefore, we further ascertained whether HBXIP could also affect other target genes of LXRs in a ligand-independent manner in breast cancer cells. We selected some lipid metabolism genes involving LXRs, such as solute carrier family 27 (SLC27A2), clusterin (CLU), phosphatidylinositol transfer protein beta (PITPNB), ATP-binding cassette transporter A1 (ABCA1), ATP-binding cassette transporter G1,5,8 (ABCG1,5,8), apolipoprotein E (APOE), and cholesteryl ester transfer protein (CETP) by bioinformatics (http://wwwmgs.bionet.nsc.ru/mgs/gnw/trrd/thesaurus/search_hidden.html) and literature (32). Interestingly, ChIP assays demonstrated that HBXIP could occupy the promoters of these eight genes with LXRE through LXRs, but not ABCA1 of cholesterol-efflux in MCF-7 cells (Supplementary Fig. S5A), due to the fact that ABCA1 gene functioned in cholesterol-efflux may be not conducive to lipogenesis (33). Moreover, we observed that HBXIP could upregulate the expression of six genes in above eight positive genes by qPCR analysis in SK-BR3 cells (Supplementary Fig. S5B), supporting that HBXIP coactivated LXRs in a ligand-independent manner in breast cancer cells. The mRNA levels of SLC27A2, CETP, and APOE were significantly higher in clinical breast cancer tissues relative to their corresponding adjacent tissues by qPCR analysis (P < 0.05, Supplementary Fig. S5C). Therefore, our results indicate that CoRNR motif with its flanking sequences of HBXIP is required for coactivation of LXRs in a ligand-independent manner.

Next, we evaluated the effect of HBXIP on lipid metabolism of breast cancer cells. Oil red O staining revealed that overexpression of HBXIP stimulated the lipid accumulation in SK-BR3 cells. Notably, SREBP-1c knockdown abolished the HBXIP-enhanced lipogenesis in the cells (Fig. 6A). Meanwhile, the expression levels of HBXIP, SREBP-1c, and FAS were examined in SK-BR3 cells (Fig. 6A). Interestingly, SREBP-1c knockdown decreased the HBXIP expression relative to the control (Fig. 6A), implicating that SREBP-1c might be involved in upregulating HBXIP expression. Bioinformatics analysis revealed that there were two sterol response elements (SRE) in HBXIP promoter (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3). Then, we found that SREBP-1c was able to bind to HBXIP promoter (−3400 to −1673) by ChIP assays (Fig. 6B). SREBP-1c knockdown could decrease the promoter activities of HBXIP and downregulate the expression of HBXIP (Fig. 6C). In addition, the overexpression of SREBP-1c could rescue the suppression of lipogenesis by si-HBXIP in MCF-7 cells (Supplementary Fig. S6A). These data suggest that HBXIP enhances lipid metabolism through a positive feedback of HBXIP/LXRs/SREBP-1c/HBXIP in breast cancer cells. In addition, we detected the effect of HBXIP on other lipid classes, such as arachidonic acid (AA), high-density cholesterol (HDL-C), phosphatidylcholines (PC), and triglycerides (TG), in MCF-7 cells (Supplementary Fig. S6B). The data further showed that HBXIP was able to result in lipid abnormalities in breast cancer cells. Next, we tested the effect of FAS on the HBXIP-promoted proliferation of breast cancer cells in vitro by 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT), and 5-ethynyl-2′- deoxyuridine (EdU) assays. The data showed that deletion of FAS could abolish HBXIP-enhanced proliferation (Supplementary Fig. S6C). Moreover, we investigated the effect of SREBP-1c on the HBXIP-promoted tumor growth in athymic mice (BALB/c) model. Our data displayed that silencing SREBP-1c by siRNAs markedly abolished the HBXIP-enhanced tumor growth. (Fig. 6D and E). The expression levels of HBXIP, SREBP-1c, and FAS were examined by Western blot analysis in the tumor tissues from mice (Fig. 6F). Moreover, mouse weight measuring ruled out the possibilities of the effect of tumor on host (Supplementary Fig. S6D). Collectively, we conclude that HBXIP enhances the growth of breast cancer cells through modulation of abnormal lipid metabolism in vitro and in vivo, in which a positive-feedback loop of HBXIP/LXRs/SREBP-1c/HBXIP benefits to the upregulation of FAS.

Abnormal metabolism of cancer cells is one of the hallmarks of malignancies. Cancer cells reprogram their metabolic pathways to adapt to their abnormal demands for survival (34, 35). The best-known metabolic phenotype is the “Warburg effect” that cancer cells rely on oxygen-independent glycolysis instead of oxidative phosphorylation to metabolize glucose and generate metabolic energy (36, 37). Recent studies are bringing the biosynthetic-associated pathways into focus on cancer. In this study, we are interested in the role of the oncoprotein HBXIP in lipid metabolism of breast cancer.

FAS, as a metabolic oncogene, is highly correlated with cancer pathogenesis. In the current study, the analysis of breast cancer tissues identified that the expression levels of HBXIP were positively related to those of FAS. Moreover, we found that HBXIP greatly upregulated FAS in breast cancer cells. Our recent work illustrated that the oncoprotein HBXIP as a transcription coactivator appeared to play a crucial role in promoting the development of breast cancer (16, 22–24). Accordingly, we hypothesized that the HBXIP might be involved in modulating FAS transcription. Actually, HBXIP upregulated FAS through activating transcription factor SREBP-1c in breast cancer cells. Previous work showed that the transcription of SREBP-1c required LXRs through LXRE in SREBP-1c promoter, leading to a subsequent transcription of lipogenic genes (9, 38). Our results showed that HBXIP was also able to occupy the LXREs in SREBP-1c promoter.

LXRα/β belongs to a class of ligand-dependent NRs, which induce certain genes controlling lipogenesis (8). In general, LXRs with ligands recruit nuclear coactivators with NR motif to activate transcription, and LXRs lacking ligands suppress transcription by recruiting nuclear corepressors with CoRNR motif (10, 11). SMRT (silencing mediator for retinoid and thyroid receptors) and N-CoR (NR corepressor) act as typical nuclear corepressors, all containing a CoRNR motif, which is required for the interaction of NRs. Our data indicated that HBXIP could directly bind to and coactivate LXRα/β and subsequently activated SRBPE-1c promoter. However, we identified that HBXIP contained a CoRNR motif (IVGVL) without NR motif, and the CoRNR motif was responsible for interacting with LXRs. Thus, we wondered why HBXIP could coactivate LXRs and whether this process needed LXR ligands or not. Previous research indicated that the CoRNR or NR motifs were responsible for binding to unliganded or liganded NRs, and its different flanking sequences determined NR activity. They found that the flanking sequences of CoRNR motif of corepressors were replaced by the sequences flanking the NR motif of coactivators, which could be converted from suppression to activation of NR (29, 39). Our results revealed that HBXIP could bind to unliganded LXRs in SREBP-1c promoter, in which the sequence flanking the CoRNR was involved in the activation of LXRs. Therefore, we first report that HBXIP containing CoRNR motif as a unique coactivator of NR triggers LXRs/SREBP-1c/FAS axis in a ligand-independent manner. RXRα is the dimerization partner of LXRα/β, involving the interaction and activation of SREBP-1c promoter (40). Interestingly, we further found that HBXIP could interact with RXRα in SK-BR3 cells, and RXRα siRNA could block HBXIP-enhanced SREBP-1c promoter activity (Supplementary Fig. S4D). It suggests that RXRα is one of the members of HBXIP and LXRs complex.

SREBP-1c is the major transcription factor that regulates the synthesis of fatty acids within the liver (41, 42). Here, our data revealed that FAS as a fatty acid synthase was markedly upregulated by SREBP1-c for fatty acid synthesis in breast cancer cells. Given that polyunsaturated fatty acids (PUFA) as nutritional ligands were involved in nutritional regulation of cellular fatty acid levels via competing with ligands to inhibit LXRs binding to SREBP-1c promoter in a negative-feedback manner in the liver (43, 44), we validated that SREBP-1c was obviously sensitive to arachidonic acid (AA, ω6 PUFAs) inhibition, leading to a negative feedback in normal Chang liver cells, but SK-BR3 cells lost this negative feedback (Supplementary Fig. S6E), suggesting that normal human liver cells mainly used typical coactivator-activated ligand-dependent LXR/SREBP-1c system for keeping the balance of lipid metabolism, but it is insensitive to ω6-PUFA inhibition in cancer cells. Interestingly, overexpressed HBXIP abolished this inhibitory effect of AA in Chang liver cells (Supplementary Fig. S6E), suggesting the ligand-dependent pathway occupied the main position because of the low-level HBXIP in normal liver cells, whereas there is still a physiologic role for HBXIP in regulating metabolism in normal cells. Furthermore, treatment with AA could not affect the interaction of HBXIP with LXRs in SK-BR3 cells (Supplementary Fig. S6F), suggesting that overexpressed HBXIP fully occupied LXR-independent ligands, AA lost its competition with ligands, resulting in HBXIP-triggered abnormal lipid metabolism escaped from the control of a negative feedback. Thus, the ligand-dependent pathway, which is regulated by the negative-feedback control, plays a more important role in normal cells, whereas cancer cells mainly rely on the oncoprotein HBXIP-driven LXR/SREBP-1c system.

Finally, we evaluated the effect of HBXIP on modulating abnormal lipid metabolism and proliferation of breast cancer cells in vitro and in vivo. Our experiments indicated that HBXIP-triggered activation of LXRs/SREBP-1c/FAS contributed to promotion of lipid metabolism and tumor growth of breast cancer cells. Interestingly, we found that SREBP-1c knockdown markedly decreased HBXIP expression. Moreover, SREBP-1c was shown to bind to and activate HBXIP promoter, suggesting that a positive-feedback loop of HBXIP/LXRs/SREBP-1c/HBXIP was formed in maintaining the activation of SREBP-1c to upregulate FAS in breast cancer. During the development of cancer, the normal level of nutrients is hard to sustain the fast growth of cancer cells (40, 45). FAS appears to be an obligatory acquisition, selecting a subgroup of cancer cells that are capable of growth and survival upon stresses such as hypoxia, low pH, and/or nutritional deprivation (46, 47). Our finding provides strong evidence that the oncoprotein HBXIP accelerates the growth of breast cancer cells through modulating abnormal lipid metabolism.

Taken together, we summarize a model that the oncoprotein HBXIP modulates abnormal lipid metabolism in breast cancer (Fig. 7). HBXIP acting as a coactivator triggers the LXRs/SREBP-1c/FAS signaling in a LXRs ligand-independent manner, in which CoRNR motif of HBXIP is required for LXR activation. A signaling of HBXIP/LXRs/SREBP-1c/HBXIP forms a positive-feedback loop, leading to the sustaining overexpression of FAS to enhance the growth of breast cancer.

No potential conflicts of interest were disclosed.

Conception and design: Y. Zhao, H. Li, X. Zhang, L. Ye

Development of methodology: Y. Zhao, H. Li, W. Zhang, X. Zhang, L. Ye

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Zhao, H. Li, Y. Zhang, Y. Li, Q. Liu, L. Qiu, F. Liu, X. Zhang, L. Ye

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zhao, H. Li, Y. Zhang, L. Li, R. Fang, W. Zhang, X. Zhang, L. Ye

Writing, review, and/or revision of the manuscript: Y. Zhao, H. Li, L. Li, R. Fang, W. Zhang, X. Zhang, L. Ye

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zhao, H. Li, R. Fang, W. Zhang, X. Zhang, L. Ye

Study supervision: X. Zhang, L. Ye

Other (conducted several experiments in this article): L. Li

X.D. Zhang has been awarded the National Natural Scientific Foundation of China grant (no. 81272218, 31470756). L.H. Ye has been awarded the grants from the National Basic Research Program of China (973 Program, no. 2015CB553905), the National Natural Scientific Foundation of China (no. 81372186), and Tianjin Natural Scientific Foundation (no. 14JCZDJC32800).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.