Cloning of RPB5-binding protein by far-Western blotting.
To identify cDNA encoding proteins that bind to RPB5, we applied far-Western blotting, a cloning method to detect protein-protein interaction in situ (13
). A HepG2 cDNA library in a λ phage expression vector was screened with a labeled recombinant human RPB5 probe. After the third screening, seven positive clones were selected from a total of 107
plaques and subjected to sequence analysis of the insert termini. Of the five clones two groups had identical sequences in the inserts (Fig. A). The remaining two clones with short stretches of sequence in frame have not been subjected to further analysis.
FIG. 1 DNA and amino acid sequence of RMP. (A) Schematic representation of five positive RMP cDNA clones screened by far-Western blotting. Open boxes indicate the predicted open reading frames. The RPB5-binding region was included in all the inserts of the positive (more ...)
All inserts of the five positive clones were subcloned and completely sequenced. Two different fragments covered 1.2 kb of cDNA encoding 306 continuous amino acids in frame with the LacZ gene. 5′ and 3′ rapid amplification of cDNA ends with HepG2 mRNA was applied to clone the full-length cDNA (see Materials and Methods), and a putative full-length clone of 2,088 bp was obtained. The cDNA harbors an open reading frame of 1.5 kb, beginning at a consensus sequence for initiation at nucleotide 468 and ending at nucleotide 1992. A nucleotide homology search of the GenBank and EMBL databases revealed that several human and rodent sequences of cDNA from different tissues matched different parts of the cloned cDNA with high scores. The 508-aa polypeptide RMP is a novel protein which has no homology to known sequences in protein databases. There is an Asp-rich region, including 13 contiguous Asp residues, in the central part of RMP which shows a high degree of homology to several unrelated proteins with different functions. RMP has no known motif except for two putative nuclear localization signals (NLS) from aa 98 to 111 and from aa 339 to 344 (Fig. B).
Expression of RMP mRNA and protein in mammalian cells.
The expression of RMP mRNA was examined in RNA samples from various sources by Northern blot analysis (Fig. A). RMP mRNA (approximately 2.4 kb in length) was detected at different levels in all human tissues examined. An additional weak band of 1.6 kb was detected in some tissues, such as testis (Fig. , lane 4). Low levels of RMP expression were observed in peripheral tissue, leukocytes, and lung. Under high-stringency conditions, RMP mRNA was also detected in RNA samples of mouse and rat tissues (data not shown). These results suggest that the RMP gene is conserved among mammals and ubiquitously expressed in various tissues.
FIG. 2 RMP is ubiquitously expressed in mammalian cells. (A) Poly(A) RNA fractions of various human tissues were subjected to Northern blotting with RMP (upper panels) and human β-actin cDNA probes (lower panels) as described in Materials and Methods. (more ...)
The presence of RMP was examined in different cell lines by immunoblotting with polyclonal antibody raised against bacterially expressed recombinant His-tagged RMP. The anti-RMP antibody recognized a 69-kDa protein in human cell lines, such as HepG2 and HeLa, and also in the monkey cell line COS1. To confirm the specificity of the endogenous RMP detected with anti-RMP antibody, these cells were transiently transfected with FLAG-tagged RMP. A protein migrating slightly slower than the endogenous RMP was detected by anti-RMP antibody in the FLAG-RMP-transfected cell lysates but not in cells transfected with the vector alone (Fig. B). This protein was also detected with anti-FLAG M2 antibody (data not shown). The apparent molecular sizes of the endogenous and FLAG-tagged RMPs were much larger than the calculated molecular mass (57 kDa). The migration behavior of RMP detected by SDS-PAGE was due not to modification of the protein but probably to the high content of charged amino acid residues, since similar discrepancies were observed with bacterial recombinant RMPs.
RMP specifically interacts with RPB5 in vitro.
Since the positive clones of RMP cDNA were selected from a λgt11 library with a GST-RPB5 probe, it remained possible that the LacZ-fused RMP portion but not the RMP moiety was recognized by RPB5. To exclude this possibility, GST-RMP was used as a probe to examine the interaction between RPB5 and RMP by far-Western blot analysis under the same binding conditions as those used to isolate the RMP clones (Fig. ). Comparable amounts of purified bacterial recombinant proteins were fractionated by SDS-PAGE (Fig. A) and subjected to far-Western blot analysis (Fig. B). Only GST-RPB5 (Fig. B, lane 2) bound to the RMP probe. The other GST-fused proteins, such as CTD and TBP, showed no binding to RMP. Interestingly, RMP did not bind HBx (Fig. , lanes 1). These results indicated the specificity of the binding of RMP and RPB5 in vitro.
FIG. 3 RMP specifically binds RPB5 in vitro. Partially purified bacterial recombinant proteins were fractionated by SDS–12.5% PAGE and stained with Coomassie brilliant blue (A) or transferred onto nitrocellulose membranes for far-Western blot (more ...) The RPB5-binding region resides in the region shared by the positive clones.
To further confirm the specific binding of RPB5, mapping of the RPB5-binding site in RMP was carried out with a GST-RPB5 probe (Fig. B). The GST-fused full-length and truncated RMPs were bacterially expressed and purified (Fig. A and C). The RPB5 probe bound strongly to the full-length RMP (Fig. A and B, lanes 1) as well as RMP-D1, -D2, -D6, -14D2, and -D15 (lanes 8, 5, 7, 4, and 14, respectively) and weakly to RMP-D13, -D14, and -D16 (lanes 10, 16, and 15, respectively) but not to the other truncated proteins or GST (lanes 2, 3, 6, and 11). Since RMP-D1 strongly bound RPB5, the RPB5-binding site was within the region from aa 137 to 231, the sequence of which was included in the two different inserts of the positive λgt11 phage clones (Fig. A). Further examination of the region around RMP-D1 with constructs RMP-D13 (aa 88 to 150), RMP-D14 (aa 176 to 231), RMP-D15 (aa 150 to 231), and RMP-D16 (aa 151 to 198) showed that only RMP-D15 had strong GST-RPB5-binding ability. RMP-D13, -D14, and -D16 had much weaker RPB5-binding ability than RMP-D15. Internal-deletion mutants of RMP, RMP-Id150 and RMP-Id175, did not exhibit RPB5 binding (Fig. A and B, lanes 12 and 13). These results indicated that the region covered by RMP-D16 (aa 151 to 198) is essential for RPB5 binding and that the region from aa 198 to 231 may have an accessory role in this binding (Fig. B, lane 14).
FIG. 4 Delineation of the RPB5-binding region of RMP. (A) Coomassie brilliant blue staining of the fractionated full-length and truncated RMPs in GST-fused forms. (B) GST-RMP and truncated mutant proteins were subjected to far-Western blot analysis with a GST-RPB5 (more ...)
The mapping results by far-Western blot analysis were further confirmed by GST resin pull-down assay (Fig. D). Bacterially expressed FLAG-RMP (Fig. D, lane 3) and FLAG-RMP-14D2 (aa 1 to 231 [lane 9]) harboring the RPB5-binding region were efficiently recovered by the immobilized GST-RPB5 resin, whereas neither a C-terminal part, RMP-D11 (aa 232 to 508 [lane 12]), nor RMP-Id150 (lane 6) (lacking aa 151 to 231) was recovered. The negative-control GST bound none of the RMPs. From the consistent results obtained by the two different methods in vitro, we concluded that RMP specifically interacts with RPB5 through its central region.
The RMP-binding region covers the TFIIB- and HBx-binding sites in RPB5.
We used a similar approach to delineate the RMP-binding region of RPB5 by far-Western blot analysis (Fig. ). The GST-RMP probe strongly bound to full-length RPB5 and RPB5-d5 (aa 1 to 160) but showed no or only weak binding to all other truncated forms of RPB5 (Fig. B, lanes 3 to 9). The region of RPB5 necessary for RMP binding was confirmed by GST resin pull-down assay (Fig. D). In this experiment, bacterially expressed and purified FLAG-RMP was incubated with immobilized GST fusion proteins harboring different portions of RPB5, and the recovered FLAG-tagged RMP was detected with an anti-FLAG antibody. Consistent with the results of far-Western blotting, RPB5-d5 as well as RPB5, but not the other truncated forms, associated with FLAG-RMP. These results suggest that RMP binding requires a region of more than two-thirds of RPB5, which completely covers both the TFIIB- and HBx-binding sites (aa 1 to 46 and 47 to 120, respectively) (13
FIG. 5 Mapping of the RMP-binding region of RPB5. (A) Coomassie brilliant blue staining of the applied GST-RPB5 and its mutant proteins. (B) GST-fused RPB5 and its truncated mutants were subjected to far-Western blot analysis with the GST-RMP probe. (C) Schematic (more ...) RMP binds RPB5 in vivo.
Next, we determined whether the interaction of RMP and RPB5 occurs in vivo by immunoprecipitation and Western blot analysis. COS1 cells were transiently cotransfected with mammalian expression vectors of RPB5 and FLAG-tagged RMP, and the cell lysates were precipitated with anti-FLAG M2 resin (Fig. A). The nontagged RPB5 protein was efficiently recovered in the immunoprecipitates with FLAG-RMP and the FLAG-RMP-14D2 mutant (Fig. A, lanes 1 and 3) but was not coprecipitated at all with FLAG-RMP-Id150 or RMP-D11 lacking the RPB5-binding region (lanes 2 and 4). Conversely, when the lysate of the COS1 cells cotransfected with FLAG-tagged RPB5 and RMP was precipitated with anti-FLAG M2 resin, FLAG-RPB5 also coimmunoprecipitated RMP (Fig. B, lane 1). RMP was barely recovered in the precipitates with anti-FLAG M2 resin in lysates of nontransfected cells (Fig. B, lane 2). These results clearly indicate that RMP interacts with RPB5 in mammalian cells and that the RPB5-binding region of RMP delineated in vivo is consistent with the mapping results obtained in vitro (Fig. ).
FIG. 6 RMP binds RPB5 in vivo. (A) COS1 cells were cotransfected with mammalian expression plasmids of RPB5 and FLAG-RMP (F-RMP), FLAG-RMP-Id150 (F-Id150), FLAG-RMP-14D2 (F-14D2), or FLAG-RMP-D11 (F-D11). Total cell lysates were immunoprecipitated with anti-FLAG (more ...)
The interaction of RMP and RPB5 described above was detected in the lysates of cells overexpressing both proteins, but it may not reflect the interaction of endogenous RMP and RPB5. Therefore, we examined whether endogenous RMP interacts with RPB5 (Fig. ). At first, relative amounts of endogenous RMP and RPB5 in HepG2 cells were determined immunologically (Fig. A). As determined by comparison of the band intensities of the various amounts of the purified recombinant RMP and RPB5, the amount of endogenous RMP in 60 μg of the total proteins is close to 2.5 ng of the recombinant RMP (Fig. A, lanes 3 and 4), and the amount of endogenous RPB5 in 20 μg of the total protein is close to 5 ng of the recombinant RPB5 (lanes 6 and 9). Therefore, the relative molecular ratio of endogenous RMP to RPB5 is approximately 1:10, indicating that RMP is substoichiometric to RPB5 or to RNA polymerases. Next, the possible interaction of endogenous RMP and RPB5 in HepG2 cells was examined by coimmunoprecipitation with anti-RMP antibody. RPB5 was weakly detected in the anti-RMP immunoprecipitates, and two other polymerase II subunits, RPB1 and RPB6, were also faintly detected in the immunoprecipitates, whereas all of the examined subunits were not recovered in the control immunoprecipitates (Fig. B, lanes 3 and 4). This result suggests that RMP, at least in part, is complexed with RNA polymerase II. To confirm this notion, coimmunoprecipitation with anti-RPB6 antibody was used to detect the RMP-polymerase II interaction, since the anti-RPB5 antibody we raised was not able to coimmunoprecipitate RMP and RNA polymerase subunits. Anti-RPB6 coimmunoprecipitated RMP and the other RNA polymerase subunits (Fig. B, lane 2). This result indicates that endogenous RMP is complexed with RNA polymerase II through RPB5 in HepG2 cells. The RPB5-binding region of RMP mapped in vitro is necessary for the association to RNA polymerase II, since anti-RPB6 antibody coimmunoprecipitated RMP but not RMP-Id150 in the overexpressed cell lysates (data not shown).
FIG. 7 Endogenous RMP associates with RNA polymerase II complex. (A) Western blot analysis of molar ratio of endogenous RMP to RPB5 in HepG2 cells. Proteins were loaded in lanes as follows: lanes 4 to 6, 60, 40, and 20 μg, respectively, of HepG2 total (more ...) RMP counteracts transactivation by HBx.
Because the RMP-binding region of RPB5 (aa 1 to 160) covers the regions necessary for TFIIB binding (aa 1 to 46) and HBx binding (aa 47 to 120), RMP may interfere with the trimeric interaction of RPB5, TFIIB, and HBx. Overexpression of RMP, therefore, may inhibit HBx transactivation. To examine this possibility, HepG2 cells were transiently cotransfected with RMP and HBx expression plasmids together with pHECx2CAT, an HBx-responsive CAT reporter, under the control of the X-responsive element derived from the core sequence in HBV Enh1 (Fig. A). Cotransfection of RMP inhibited the transactivation by HBx (Fig. A, bars 12 to 14), although the inhibition was not complete in the presence of excess RMP. Expression levels of HBx immunologically detected by anti-HBx antibody were similar among cotransfected cell lysates in the absence or presence of RMP (data not shown). Coexpression of RMP-Id150 and RMP-Id175, which lack 57 and 82 aa in the RPB5-binding region, respectively, exhibited no inhibitory effect on the transactivation by HBx (Fig. A, bars 16 to 18, and data not shown), implying that RPB5 binding is essential for the inhibitory effect. Because HBx has been reported to transactivate a wide variety of cis
), we examined the possibility that RMP may inhibit transactivation of HBx through the other X-responsive cis
element. A similar inhibitory effect of RMP was observed when pNFκBx3CAT, harboring three tandem repeats of the NFκB-binding site, another X-responsive cis
element, was used as the reporter (Fig. B). This result suggests that RMP inhibits transactivation by HBx of different X-responsive cis
FIG. 8 RMP inhibits HBx transactivation. (A) Transfection of HepG2 cells and CAT assays were described in Materials and Methods. Various amounts of HBx and RMP plasmids were transfected together with the reporter plasmid, pHECx2CAT, in which the CAT gene is (more ...) HBx counteracts the inhibitory effect of RMP.
The inhibitory function of RMP on activated transcription implies that HBx may counteract the negative regulatory role of RMP, which may be involved in HBx transactivation. To address this possibility, we examined whether HBx can release the corepressor effect of RMP in the pHECx2CAT reporter system, in which the CAT gene is under the control of the dimeric sequence of the HBV Enh1 core. HBx counteracted the corepressor activity of RMP in this system in a dose-dependent manner (Fig. A, bars 19 to 22). The anticorepressor function of HBx resides in its transactivation domain, since HBx-5D1 bearing the transactivation domain similarly counteracted the corepressor function of RMP (data not shown). Overexpression of HBx-3D5 lacking the RPB5 binding site had no counteracting effect on RMP (Fig. A, bars 23 to 25).
RMP inhibits activated transcription in mammalian cells.
Since RMP has no HBx-binding ability, the negative effect of RMP on transcription may occur in the absence of HBx. Therefore, we examined the effects of RMP on activated transcription by Gal-VP16, a chimeric activator with the Gal4 DNA-binding domain fused to the VP16 activation domain. pGalCAT, a CAT reporter with the simian virus 40 promoter driven by five Gal4-binding sites, was used as the reporter (Fig. ). The transactivation of Gal-VP16 was inhibited twofold by the coexpression of RMP, indicating that RMP negatively affects a wide variety of activated transcriptions (Fig. , bars 9 to 11). No such effect was observed with pSG9RMP in which the RMP cDNA was inserted in the reverse orientation (data not shown). If RMP inhibits activated transcription through the RPB5 binding of RMP, deletion mutants lacking RPB5-binding abilities may have no such inhibitory effect. The internal-deletion mutant RMP-Id150, lacking the 82 amino acid residues in the RPB5-binding region, had no effect on pGalCAT in the absence of Gal-VP16 (Fig. , bars 12 to 15) but augmented CAT activity in the presence of Gal-VP16 (lanes 16 to 19). These results suggest that RMP has a corepressor activity and that the mutant RMP lacking the RPB5-binding ability acts positively, probably by perturbing the negative regulation in which endogenous RMP is involved.
FIG. 9 RMP inhibits activated transcription by Gal-VP16. HepG2 cells were cotransfected with pGalCAT and Gal-VP16 together with HBx or RMP constructs. The transfected plasmid DNA was 5 μg of pGalx5CAT together with the following: bars 1 to 4, 0, 5, 10, (more ...)