HCV core sensitizes ATRA-induced cell death. To investigate the effect of the core on sensitivity to apoptosis induced by a variety of stimuli, we established MCF-7 cells constitutively producing the core. Examination of the response of these cells to a variety of apoptotic stimuli determined that ATRA-induced cell death was enhanced in core-producing cells (Fig. ). Similar results were obtained for MCF-7 cells transiently producing the core protein (Fig. , bars 1 to 6). This enhancement of apoptotic induction was accompanied by an enhancement of the expression of tTGase (Fig. , lanes 4 and 5), which is downstream of ATRA and involved in ATRA-mediated apoptosis (
19,
35,
36). It was confirmed that suppression of tTGase function by a tTGase inhibitor, monodansylcadaverine (
36), reduced ATRA-induced cell death in core-producing MCF-7 cells (Fig. , bars 3, 4, 11, and 12). Thus, the core appears to sensitize cells to ATRA-mediated cell death, accompanied by an enhancement of proapoptotic tTGase gene induction.
HCV core protein activates RAR-mediated transcription. ATRA-induced apoptosis and tTGase expression are reportedly mediated by the transcriptional activity of RAR, the nuclear receptor for ATRA (
11,
19,
36). Therefore, we next examined the effect of core expression on the activity of RAR-mediated transcription by a reporter assay. We used the RAR-responsive reporter plasmid pRARE-Luc, which includes a basal promoter with three copies of RARE upstream of the luciferase gene, to assess the RARα-mediated transcriptional activity of the cells producing the core either transiently (Fig. ) or constitutively (Fig. ). Luciferase activity increased in a manner dependent on core concentration. Similar results were obtained with additional cell lines, including HeLa, Huh-7, and MCF-7 cells, producing the core transiently (data not shown). Both the whole HCV polyprotein and the core alone activated luciferase activity (data not shown). In Huh-7 cells carrying an HCV full-genome replicon, established in our laboratory (unpublished data), RARα activity increased approximately threefold over the levels observed in parental Huh-7 cells (data not shown). Since the luciferase activity driven by a reporter plasmid lacking RARE was not affected by core expression (data not shown), these results suggested that the core protein activates RARα-mediated transcription independently of cell type.
To identify the region of the core protein responsible for RARα activation, we performed a detailed deletion analysis as described previously (
54). We prepared carboxy-terminal deletion constructs fused with HCV p7 containing a signal sequence to allow localization to the ER. We confirmed that these fused proteins reside on ER membranes, as seen with the wild-type core (
27,
54) (data not shown). Cells producing the core constructs lacking the amino-terminal 100 aa [coreΔ(101-191)] or aa 21 to 80 [coreΔ(1-20 + 81-191)] demonstrated reporter activities similar to those of negative-control cells transfected with the empty vector (Fig. ). Production of the region from aa 21 to aa 80 [coreΔ(21-80)-p7] upregulated reporter activity significantly from the levels observed in negative-control cells and cells producing p7 alone. These data indicated that the region from aa 21 to aa 80 of the core [coreΔ(21-80)] is responsible for the activation of RARα-mediated transcription.
Identification of cellular factors interacting with HCV core by yeast two-hybrid screening. To clarify the molecular mechanisms governing transcriptional activation of RARα by the core protein, we tried to identify cellular factors interacting with coreΔ(21-80) by yeast two-hybrid screening. Using the amino-terminal 80 aa of the core (core80) fused to the LexA DNA binding region as bait, we screened a human prostate cDNA library for interacting proteins. Six independent candidate clones for binding with the bait were selected from 5 × 106 transformants. Sequencing analysis revealed that one of the candidate clones encoded Sp110b. A splicing variant of this protein, Sp110, reportedly activates RARα-mediated transcription. A schematic diagram of the molecular structures of Sp110 and Sp110b is shown in Fig. (left).
Sp110b interacts with HCV core through its central region. To confirm an interaction between the core and Sp110b, we performed a GST pulldown assay in vitro. Recombinant Sp110 and Sp110b fused to GST (GST-Sp110 and GST-Sp110b) were incubated with in vitro-translated wild-type 35S-labeled core. The 35S-labeled core was copurified with Sp110b but not with Sp110 (Fig. , top panel). In vitro-translated Sp110b was efficiently pulled down with GST-core80, but only a small amount of Sp110 could be observed in the pulled-down fraction (Fig. ). These results suggested that the core interacts more efficiently with Sp110b than with Sp110. The efficient interaction of the core with Sp110b appeared to be specific, as no association was observed with Sp100, which exhibits significant homology to Sp110 and Sp110b (Fig. ).
We used deletion analysis to determine the region of Sp110b interacting with the core. The in vitro-synthesized fragment of Sp110b from aa 277 to 453, but not fragments of Sp110 from aa 1 to 276 or aa 454 to 689, was copurified with GST-core80 (Fig. ). Sp110b lacking the region from aa 277 to 453 could not be copurified with GST-core80 (Fig. ), suggesting that the region of Sp110b from aa 277 to 453 is both necessary and sufficient for the interaction with core80. Dissection of this region determined that the region of Sp110b from aa 389 to 453, but not that from aa 277 to 388, was essential for this interaction (Fig. ). Therefore, we have named this region the core-binding region (CBR). Although the CBR is shared by Sp110b and Sp110, the affinity of Sp110 to the core was low in vitro, as described above. Similar results were obtained when the C-terminal fragments of Sp110b and Sp110, which include the CBR, were examined in this assay system (Fig. ). From these results, it appears to be possible that the C-terminal region of Sp110, containing the plant homeobox domain (PHD) and the bromodomain, has an inhibitory effect on core association. Analysis of additional deletion mutants showed that the entire C-terminal domain harboring both the PHD and the bromodomain, rather than either domain alone, was likely to be responsible for blocking the interaction with the core (data not shown).
To further confirm the interaction of Sp110b with the core, we performed a coimmunoprecipitation assay. The core was detected in immune complexes pulled down with an anti-FLAG antibody, but not with normal mouse IgG, from the lysates of COS-7 cells producing both FLAG-tagged Sp110b and the core simultaneously (Fig. ). Cumulatively, these results indicate that Sp110b interacts specifically with the core.
Sp110b is expressed ubiquitously in human tissues and more abundantly than Sp110. Sp110b expression in human tissues was examined by Northern blot analysis using the 32P-labeled 5′-terminal cDNA region of Sp110b as a probe. A 1.9-kb band was detected in all the human tissues investigated (Fig. ). From the information that the Sp110 and Sp110b mRNAs are approximately 2.3 and 1.9 kb, respectively, this band seemed to correspond to Sp110b mRNA. The 2.3-kb band of Sp110 mRNA was not observed in this experiment. To compare the expression levels of Sp110 and Sp110b mRNAs in these specimens, a semiquantitative analysis was conducted by RT-PCR. The band for Sp110 mRNA was detected in total RNAs from human spleens and Jurkat cells, a human T-lymphoma cell line (Fig. , lanes 11 and 14). Little or no signal was observed in other tissues and cell lines (Fig. , lanes 10 to 15). We observed, however, that Sp110b mRNA was more abundant than Sp110 mRNA in all the tissues and cell lines investigated. In vitro-synthesized Sp110 and Sp110b mRNAs were mixed at variable ratios as templates in this system (Fig. , lanes 1 to 9); the expression of Sp110b mRNA was estimated to be about 10- and 3- to 5-fold higher than that of Sp110 mRNA in nonleukocytes and leukocytes, respectively. These results were consistent with data in which Sp110b protein but not Sp110 protein was detected in lysates of HeLa (Fig. ) and Huh-7 (data not shown) cells by using an antibody recognizing both Sp110 and Sp110b protein. Thus, Sp110b mRNA is expressed ubiquitously in all the human tissues examined, more abundantly than Sp110.
Sp110b is a potent transcriptional corepressor of RAR. Although Sp110 is reported to activate RARα-mediated transcription (
6), the molecular function of Sp110b remains unknown. To characterize the function of Sp110b, we examined the effect of this protein on RARα-mediated transcription by a reporter assay. As reported previously (
6), ectopic expression of Sp110 augmented reporter activity in a dose-dependent manner (Fig. ). In contrast, luciferase activity was drastically reduced in cells producing Sp110b in a dose-dependent manner, irrespective of ectopic RARα production (Fig. ). Reporter activity from the reporter plasmid lacking RARE was not affected by ectopically produced Sp110 or Sp110b (data not shown). Similar results were obtained in all the cell lines examined, including Huh-7, MCF-7, and HeLa cells (data not shown). Moreover, we observed the downregulation of RAR-responsive genes, including tTGase and transforming growth factor β2 (TGF-β2), in cells producing Sp110b (data not shown). These data suggest that Sp110b suppresses RARα function. To investigate the function of endogenous Sp110b, we introduced an siRNA, Sp110(b)-siRNA, to specifically knock down the expression of Sp110 and Sp110b in HeLa cells in which Sp110b was expressed but Sp110 protein was not detected (Fig. ). Endogenous Sp110b levels were dramatically reduced, while α-tubulin levels remained the same (Fig. ). Under these conditions, RARα-mediated transcription was activated two- to threefold over that in cells transfected with a randomized siRNA (Fig. , bars 1 and 3). These findings indicated that Sp110b suppresses the transcriptional activity of RARα without altering RARα production levels (Fig. , lower panel).
The existence of a liganded nuclear receptor binding motif (LXXLL) (
17) in Sp110 and Sp110b (aa 525 to 529; LGELL) raised the possibility that Sp110 and Sp110b may interact with the nuclear receptor RARα in the presence of RARα ligands. We investigated this possibility using a GST pulldown assay. RARα was copurified with both GST-Sp110 and GST-Sp110b in the presence, but not in the absence, of ATRA (Fig. , top and center panels), indicating that RARα interacts with Sp110 and Sp110b in vitro in a ligand-dependent manner. The interaction between Sp110b and RARα in the presence of ATRA was confirmed by a coimmunoprecipitation assay with 293T cells (Fig. ). Further analysis on whether Sp110b, together with RARα, associates with the target promoter was performed by a DNA-protein complex immunoprecipitation assay. A RARE-containing promoter was immunoprecipitated with either Sp110b or RARα in the presence of ATRA (Fig. ). These data suggest that, in conjunction with RARα, Sp110b associates with the target promoter containing RARE in the presence of ATRA. In addition, Sp110 and Sp110b were observed in the nuclei of cells in a dense granular staining pattern (Fig. , panels 2, 9, and 30), as previously reported for Sp110 (
6). RARα was also located primarily in the nucleus (Fig. , panels 3 to 5 and 10 to 12), as reported previously (
7).
These results suggest that Sp110b is a transcriptional corepressor downregulating RARα activity in the nucleus.
The ability of HCV core to activate RARα was reduced by RNAi elimination of endogenous Sp110b. The accumulated evidence suggests that Sp110b, a potent transcriptional corepressor of RARα, may play a role in the activation of RARα-mediated transcription by the core. We therefore examined the capacity of the core to activate RARα upon elimination of endogenous Sp110b protein in HeLa cells by the RNAi technique. When endogenous Sp110b protein was knocked down by siRNA treatment (Fig. ), the effect of core expression on RARα-mediated transcription (about 1.30-fold activation) was significantly reduced from levels observed in cells treated with a control siRNA (about 2.69-fold activation) (Fig. ). This result suggests that endogenous Sp110b plays an important role in the activation of RARα-mediated transcription by the core protein.
HCV core changes the subcellular localization of Sp110b from the nucleus to the cytoplasmic surface of the ER. Our finding that Sp110b interacts with the core and plays an important role in RARα activation by the core presented a paradoxical problem in that Sp110b was likely to be a nuclear factor but the core was located mainly around the perinuclear region in the cytoplasm (Fig. , panel 1). To investigate this question, we examined the subcellular localization of each protein by indirect immunofluorescence analysis. Upon coproduction of the core protein, the localization of Sp110b changed dramatically from the nucleus to the perinuclear region of the cytoplasm, where the core was located (Fig. , panels 9 and 13 to 15). In contrast, Sp110 was observed in the nucleus, irrespective of core production (Fig. , panels 2 and 6 to 8). Similar results were obtained in 293T cells (data not shown). The subcellular localization of other well-known nuclear proteins—p53 (
12), poly(ADP-ribose) polymerase (
45), and RARα (
7)—was not influenced by core expression, in a manner similar to that observed for Sp110 (data not shown). These results indicate that the core specifically alters the subcellular localization of Sp110b. We further confirmed this phenomenon in HeLa cells by observation of the alteration of the location of endogenous Sp110b by the core. In the absence of core expression, a nuclear staining pattern was seen (Fig. , panel 30) as described above. However, perinuclear localization was observed in certain cells producing the core, similar to the pattern seen for exogenous Sp110b (Fig. , panels 31 to 33). Since Sp110b but not Sp110 was detected in HeLa cells by using this antibody (as shown in Fig. ), the endogenous protein detected around the perinuclear region in the presence of the core appears to be primarily Sp110b. Thus, core expression altered the subcellular localization of Sp110b.
To investigate the colocalization of Sp110b with the core, we fractionated cell extracts into nuclear, microsomal-membrane, and cytosolic fractions for detection of Sp110b (Fig. ). Fraction identities were confirmed by detection of SC-35 (
49), the core (
27,
54), and α-tubulin for the nuclear, microsomal-membrane, and cytosolic fractions, respectively (Fig. , lower panels). Following coproduction of the core, Sp110b shifted into the microsomal-membrane fraction (Fig. , top panel). Core expression thus altered the subcellular localization of Sp110b from the nucleus to the cytoplasmic surface of the ER, where the core was originally located.
To investigate the molecular mechanisms governing the change in Sp110b subcellular localization by the core, we utilized a core point mutant, core(6162M). This protein, containing substitutions of arginines at positions 61 and 62 for glycine and leucine, respectively, has little affinity for Sp110b (Fig. , center panel). Production of the core(6162M) mutant had no effect on the subcellular localization of Sp110b (Fig. , panels 16 to 19). In addition, Sp110b(1-276 + 454-539), which lacks the CBR and does not associate with the core in vitro (Fig. ), remained in the nucleus irrespective of core production (Fig. , panels 20 to 23). We also performed an interaction-competition analysis using a peptide fragment containing the CBR of Sp110b (myc-CBR fragment), a fragment capable of colocalizing with the core in the cytoplasm (Fig. , panels 24 to 26). Excess myc-CBR fragment inhibited the colocalization of Sp110b with the core (Fig. , panels 27 to 29), suggesting that molecular interaction of Sp110b with the core is essential for the alteration of Sp110b subcellular localization by core production.
Transcriptional activation of RARα by HCV core was exerted via sequestration of the transcriptional corepressor Sp110b away from the nucleus. The above results raised the possibility that Sp110b, a potent nuclear transcriptional corepressor of RARα, was functionally modulated by the core through sequestration from the nucleus. To investigate this possibility, we examined by a reporter assay whether the core modulates the suppressive effect of Sp110b on RARα activity (Fig. ). Sp110b-induced transcriptional suppression was overcome by production of the core in a dose-dependent manner (Fig. ). As the expression levels of both ectopically produced Sp110b and RARα were not affected by core production (data not shown), the core likely inactivates the function of Sp110b suppressing RARα-mediated transcription.
To determine if the sequestration of Sp110b from the nucleus is essential for the activation of RARα-mediated transcription by the core, we performed an interaction-competition analysis. The CBR fragment, which itself did not affect the transcriptional activity of RARα in the absence of the core (Fig. , bars 7 and 8), served as a competitor for Sp110b sequestration by the core (Fig. , panels 24 to 29). Coproduction of this CBR fragment with the core reversed the core-induced transcriptional activation in a dose-dependent manner (Fig. , bars 3, 4, and 9 to 14). Moreover, we observed that the core mutant core(6162M), which could not alter the localization of Sp110b, also lacked the ability to augment transcriptional activity (Fig. , bars 5 and 6). From the above results, we concluded that the activation of RARα-mediated transcription by the core results from the sequestration from the nucleus and subsequent inactivation of the transcriptional corepressor Sp110b.
Sequestration of Sp110b from the nucleus by HCV core caused sensitization to ATRA-mediated cell death. We analyzed the relationship of Sp110b sequestration by the core to sensitization to ATRA-induced cell death and enhancement of tTGase gene induction (Fig. ) by an interaction-competition analysis using the CBR fragment. tTGase mRNA levels, increased by core expression in the presence of ATRA, decreased following coproduction of the CBR fragment (Fig. , upper panel, lanes 4 to 6). When the core and the CBR fragment were simultaneously produced in transiently transfected cells, the observed levels of ATRA-induced cell death were similar to those found in cells with no ectopic protein production (Fig. , bars 7 and 8). This result suggested that the CBR fragment reverses the promotion of ATRA-induced cell death mediated by the core. Moreover, the core(6162M) mutant did not enhance cell death induced by ATRA (Fig. , bars 9 and 10).
The data cumulatively suggest that the interaction of HCV core with Sp110b and the subsequent sequestration of Sp110b from the nucleus play a significant role in the sensitization to ATRA-mediated cell death induced by HCV core protein.
This article has been 
