Sp1 Family Proteins Recognize the U5 Repressive Element of the Long Terminal Repeat of Human T Cell Leukemia Virus Type I through Binding to the CACCC Core Motif*

We have identified several nuclear proteins binding to the U5 repressive element (U5RE) at the U5 region of the human T cell leukemia virus type I (HTLV-I) long termi- nal repeat (LTR). In gel mobility shift assays with the U5RE DNA probe, Jurkat T cell nuclear proteins gener- ated five different complexes, named U5RE binding protein complexes (U5RP)-A1, -A2, -A3, -B, and -C. Only U5RP-C was affected by pretreatment with an excess of poly(dI-dC) and was immunodepressed by anti-Ku/p80 antibodies, suggesting that U5RP-C is a nonspecific complex involving Ku antigen. UV cross-linking showed at least six nuclear proteins involved in the other complexes, including U5RP-A1, -A2, -A3, and -B. The sequence of the binding core element of these specific complexes, determined by competition assays and gel mobility shift assays using a series of the U5RE mutants, is CACCC which is identical to that for the Sp1 transcription fac- tor. LTR with a mutant U5RE, which has no ability to bind with the nuclear proteins, showed stronger pro- moter activity than LTR with the wild U5RE, suggesting that the specific interaction of these U5RE-binding pro- teins might result in the U5-mediated repression. transfection Luciferase activity measured Ger-many) PicaGene kit Anti-Ku/p80 Monoclonal Antibody Preparation— Anti-Ku/p80 monoclonal antibodies were generated by injecting mice with bacterially expressed Ku/p80 proteins. Briefly, the cDNA fragment comprising the coding region for the Ku/p80 protein (26) was ligated with a pGEX-2T expression vector (Pharmacia) to obtain the plasmid capable of express-ing a Ku/p80 protein as a fusion protein with glutathione S -transferase (GST-Ku/p80). The recombinant GST-Ku/p80 protein expressed in E. coli was purified using a prepacked glutathione-Sepharose 4B column (Pharmacia) and loaded onto a preparative polyacrylamide gel (Bio-Rad model 491). Fractions containing GST-Ku/p80 were collected and used as immunogens in mice. We screened a hybrid-myeloma cell clone from which the culture supernatant specifically reacts with both the recom- binant Ku/p80 protein and the natural Ku/p80 protein in HeLa cells (ATCC no. CCL2) by Western blotting. The supernatant from this clone was used as anti-Ku/p80 monoclonal antibodies.

Human T cell leukemia virus type I (HTLV-I) 1 is an etiological agent of adult T cell leukemia and HTLV-I-associated myelopathy or tropical spastic paraparesis (1)(2)(3)(4)(5)(6). After infection into humans, however, the virus has a long latent period to induce such diseases. The mechanism of the viral latency has not yet been uncovered, although there are many studies on the regulation mechanisms of HTLV-I gene expression (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19). The expression of the viral genes, integrated into host chromosomal DNA, is regulated by various viral and host nuclear factors through the viral 5Ј-long terminal repeat (LTR). Specific inter-action of host cell transcription factors with the U3 region of the LTR is crucial in gene regulation. The 5Ј-U3 region contains a transcriptional enhancer composed of the 21-bp repeat and other transcription factor-binding motifs. The 21-bp repeat elements are required for the transactivation of the viral regulatory protein, p40tax, which is reported to bind indirectly to the enhancer elements through host cell nuclear factors such as ATF and NF B proteins (8,14,18). The p40tax recently has been shown to increase the in vitro DNA binding activity of multiple ATF proteins and many other bZIP proteins such as AP-1 and CREB (20). The cellular nuclear factors such as SP1, TIF-1, Ets, and Myb interact with the LTR at a region located between two proximal 21-bp repeats (7,11,13). The R region of the LTR also contains the enhancer activity, and YB-1, a cellular binding factor of this enhancer region, was recently characterized (9,19). An element at the boundary of the R-U5 region was proposed to control virus basal gene expression (10), although this region represses HTLV-I gene transcription in presence of the human cytomegalovirus IE2 protein (21). The U5 region of the 5Ј-LTR was shown to contain a repressive element for the viral gene expression (16). Seiki et al. (17) have shown that the region exerts its repressive effects at the posttranscriptional level. We recently reported that the U5-mediated repression also occurred at a transcriptional level and that the U5RE binding protein involved at least the autoantigen Ku protein complex p70/80 and an unknown protein p110 (15).
In this report, we further analyzed the binding complex with the U5RE in detail, determined the sequence of the binding core motif, and identified some of the specific binding proteins to U5RE, such as Sp1 and Sp3. Finally, we propose that the specific interaction of these binding proteins to U5RE might result in the U5-mediated repression.

EXPERIMENTAL PROCEDURES
Cell Lines-The human T cell line Jurkat (15) was cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (R10F medium).
Crude Nuclear Extracts-Nuclear extracts were prepared from Jurkat cells according to the method of Dignam et al. (22). Briefly, cells were washed with phosphate-buffered saline, suspended in 5 packed volumes of ice-cold lysis buffer A (10 mM HEPES, pH 8.0; 10 mM KCl; 1.5 mM MgCl 2 ; 0.5 mM dithiothreitol; 0.5 mM phenylmethylsulfonyl fluoride), kept on ice for 10 min, and centrifuged at 1,000 ϫ g for 10 min. The cell pellet was resuspended in 2 volumes of ice-cold buffer A, homogenized 10 -20 strokes with a Dounce homogenizer, and centrifuged at 1,000 ϫ g for 10 min at 4°C. After discarding the supernatant, the precipitate was recentrifuged at 25,000 ϫ g for 20 min at 4°C. The nuclear pellet was resuspended in ice-cold extraction buffer 1 (20 mM HEPES, pH 8.0; 0.5 M NaCl; 20% glycerol; 1.5 mM MgCl 2 ; 0.2 mM EDTA; 0.5 mM dithiothreitol; 0.5 mM phenylmethylsulfonyl fluoride) at a ratio of 2.5 ml/10 9 cells, homogenized 10 -20 strokes with a Dounce homogenizer, and kept on ice for 30 min. After centrifugation at 25,000 ϫ g for 30 min at 4°C, the supernatant was dialyzed against dialysis buffer 2 (20 mM Tris-HCl, pH 8.0, 20% glycerol, 0.1 M KCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) for more than 8 h at 4°C. The dialyzed extract was cleared by centrifugation at 8,000 ϫ g for 15 min, and the supernatant was used directly for further analysis or kept in aliquots at Ϫ80°C until use.
Sucrose Density Gradient Sedimentation-One ml of the Jurkat nuclear extract was applied on 4 ml of a 5-30% sucrose gradient bed and centrifuged at 100,000 ϫ g for 18 h at 4°C using SW55i rotor (Beckman; L8 -60 M). Each fraction from the bottom puncture of the centrifuge tube was recovered step wise in 500-l amounts and analyzed.
Oligonucleotides for Probes and Competitors-The oligonucleotides were synthesized using a DNA synthesizer (Cyclone Plus DNA Synthesizer, model 391 PCR-MATE DNA synthesizer, Applied Biosystems, Foster City, CA). The sequence of the DNA is indicated in the Figs. 5B and 6C. The sequence of a nonspecific DNA competitor is 5Ј-AGCT-TCAGGTAGACTGCTTCGATCACTAGAGA-3Ј as described previously (15). The synthesized DNAs were purified by electrophoresis on a 20% polyacrylamide, 7 M urea gel and DE52 ion exchange columns, precipitated in cold ethanol, and suspended in TE solution (10 mM Tris-HCl, pH 8.0, 1 mM EDTA).
Gel Mobility Shift Assay-The crude nuclear extracts or partially purified fractions were incubated with 32 P-end-labeled DNA probes in binding buffer 1 (20 mM Tris-HCl, pH 7.4, 50 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 5% glycerol) after preincubation with or without poly(dI-dC)⅐poly(dI-dC) (designated poly(dI-dC) DNA; Pharmacia, Uppsala, Sweden) for 30 min at 25°C. Aliquots of the reaction mixtures were loaded onto a 5% polyacrylamide gel, followed by electrophoresis for 90 min at 150 V in the electrophoresis buffer (TAE; 40 mM Tris acetate buffer, 1 mM EDTA) as described elsewhere (15). The mobilityretarded DNA bands were visualized by autoradiography with x-ray film (X-Omat AR; Eastman Kodak Co.) or by using a BAS 2000 BioImage analyzer (Fuji Film; Tokyo, Japan).
For immunodepression assays, 1 l of the appropriate antibody was simultaneously added into the binding reaction mixtures. After incubation for 15 min at 25°C, the reaction mixtures were further incubated with protein G and A agarose (Oncogene Science, Manhasset, NY) for 30 min; thereafter, the supernatants of these reaction mixtures were processed as described above. For supershift assays, 1 l of the appropriate antibody was added to the binding reaction mixture 20 min prior to the loading of the gel.
For UV cross-linking, the binding reaction mixtures were irradiated at 254 nm for 30 min at 25°C at a distance of 3 cm and then loaded onto a 5% polyacrylamide gel. From the gel slice including the target band, proteins were eluted in SDS-PAGE loading buffer, heated at 90°C for 5 min, and loaded onto an SDS-10% polyacrylamide gel. After electrophoresis, the gel was dried on Whatman 3MM paper and exposed on the x-ray film.
Transfection and Chloramphenicol Acetyltransferase (CAT) and Luciferase Assays-Three luciferase (luc) expression plasmids derived from the HTLV-I LTR promoter with the U5RE (wild type), or the U5RE M13 or M17 (mutant type; see Fig. 6C) were constructed as described previously (15); pBLTR-Wt-luc, pBLTR-M13-luc, or pBLTR-M17-luc, in which the luc gene was under the control of the U3-R-U5 (Ϫ321 to ϩ316) region of the wild type LTR or the mutant type LTR which is introduced as a single point mutation (C to G or A) at nucleotide ϩ275 or ϩ279 of the U5RE region as shown in Fig. 7A, respectively.
The mixture of 1 g of pBLTR-Wt-luc, pBLTR-M13-luc, or pBLTR-M17-luc expression plasmids and 2 g of the internal control CAT expression plasmid pRSV-CAT (15) was transfected into Jurkat cells using a Lipofectin KS reagent (Life Technologies, Inc.) in serum-free culture medium, followed by culturing for 16 h, and was further cultured for 48 h in R10F medium. Thereafter, the cells were harvested, and whole cell lysates in 100 l of reporter lysis buffer (Promega, Madison, WI) were prepared. The lysates were cleared by centrifugation at 15,000 ϫ g for 10 min. The supernatants were recovered and assayed for CAT and luciferase activities. CAT activity was measured as described previously (25) using L-threo-[dichloracetyl-1-14 C]chloramphenicol (Amersham Int., plc., Buckinghamshire, UK), by which transfection efficiency was normalized. Luciferase activity was measured using a luminometer (Lumat model LB9501; Berthold, Wildbad, Ger-many) with a PicaGene kit (TOYO INKI, Tokyo, Japan).
Anti-Ku/p80 Monoclonal Antibody Preparation-Anti-Ku/p80 monoclonal antibodies were generated by injecting mice with bacterially expressed Ku/p80 proteins. Briefly, the cDNA fragment comprising the coding region for the Ku/p80 protein (26) was ligated with a pGEX-2T expression vector (Pharmacia) to obtain the plasmid capable of expressing a Ku/p80 protein as a fusion protein with glutathione S-transferase (GST-Ku/p80). The recombinant GST-Ku/p80 protein expressed in E. coli was purified using a prepacked glutathione-Sepharose 4B column (Pharmacia) and loaded onto a preparative polyacrylamide gel (Bio-Rad model 491). Fractions containing GST-Ku/p80 were collected and used as immunogens in mice. We screened a hybrid-myeloma cell clone from which the culture supernatant specifically reacts with both the recombinant Ku/p80 protein and the natural Ku/p80 protein in HeLa cells (ATCC no. CCL2) by Western blotting. The supernatant from this clone was used as anti-Ku/p80 monoclonal antibodies.

Detection of Five Distinct Binding
Complexes to the U5RE-We previously described that one major shift band was detectable in gel mobility shift assays with the 32 P-labeled U5RE DNA probe. In this study, however, at least five bands were distinguishable when assayed under the conditions with modifications as described under ''Experimental Procedures.'' The slowest three mobility complexes appear to be very closely retarded bands, which addressed U5RP-A1, -A2, and -A3. Here, we name Group A inclusive of these three bands. The other two separated bands are designated as U5RP-B and -C (Fig. 1). The complex formation of these bands appeared to be very fragile, because the intensity of these bands decreased when the nuclear extract proteins were used after freezing and thawing (data not shown). Thus, the Jurkat nuclear extracts without treatment by freezing and thawing were used in this study. As shown in Fig. 1, lanes 1 and 6, the band intensity of U5RP-C was shown to decrease after preincubation of the nuclear extracts with the poly(dI-dC) DNA for 10 min on ice; however, the intensity of the others, U5RP-A1, -A2, -A3, and -B, was not affected. Competition gel shift assays revealed that the bands, Group A and U5RP-B, were clearly competed with the U5RE DNA but not with the nonspecific DNA ( Fig. 1). In addition, these complexes were competed with similar efficiencies by the U5RE competitor. These results suggested the presence of specific binding proteins in U5RP-A1, -A2, -A3, and -B, but not in U5RP-C. To characterize these complexes further, Jurkat nuclear extracts were subjected to sucrose gradient sedimentation, and each fraction was analyzed by gel mobility shift assays (Fig. 2). The highest peak of the complexes of Group A and U5RE-B was detected in the same fraction (fraction 3; Fig. 2, lane 4); whereas that of the other U5RP-C was in another fraction (fraction 4; Fig. 2, lane 5), suggesting that DNA-protein complexes contained in Group A and U5RP-B are distinct from those in U5RP-C. Our previous studies have shown that Ku or Ku-related proteins are involved in the U5RE-binding protein complex (15). Thus, to clarify which complex involves Ku, we performed gel mobility immunodepression assays using monoclonal antibody raised against Ku/p80 antigen as described under ''Experimental Procedures.'' The anti-p80/Ku antibody appeared to depress DNA binding only in U5RP-C; whereas neither the anti-p40tax monoclonal antibody nor a negative control culture supernatant appeared to affect binding in these five complexes (Fig. 3). To investigate the kind of proteins that are included in these specific binding complexes, we performed a UV cross-linking assay. After PAGE of the irradiated binding reaction mixtures, the two gel slices including Group A and U5RE-B, respectively, were prepared and analyzed. Group A appeared to contain at least four proteins of about 64, 70, 76, and 110 kDa (indicated with a large arrowhead in Fig. 4, lanes 1 and 3). U5RP-B contains at least two proteins of about 52 and 95 kDa (indicated with a small arrowhead in Fig. 4, lanes 2 and 4), and these two are also observed in Group A (indicated with a small arrowhead in Fig. 4, lanes  1 and 3). These results indicated that each of these complexes of Group A and U5RP-B contains mostly different proteins.
Identification of the Binding Core Motif(s) of U5RP-A1, -A2, -A3, and -B-To identify the binding core motif of Group A and U5RP-B, competition analysis was performed in gel mobility shift assays using a series of mutant U5RE competitor DNAs, M1-M5 (Fig. 5B). Group A and U5RP-B complexes binding to the wild U5RE probe were competed with M1, M3, and M5 as well as with the wild U5RE but not with M2 nor M4 (Fig. 5A). From these results summarized in Fig. 5B, we suspected that a binding core motif of Group A and U5RP-B was involved in the TTCCACCC sequence. To further analyze the Group A and U5RP-B binding core motif in detail, we performed a competition study with another series of mutant U5RE competitor DNAs, M11-M21 (Fig. 6C) in gel mobility shift assays. M13, M15, M16, and M17 had no effect on the U5RE binding ( Fig.  6A; lanes 6, 8, 9, and 10, respectively). M19 and M20 (lanes 12 and 13, respectively) were less competitive than was wild U5RE. Furthermore, gel mobility shift assays were performed using the 32 P-labeled DNAs from M11 to M21 as a probe (Fig.  6B). Compared with the wild U5RE probe, neither M16 nor M17 was detectable the Group A and U5RP-B complexes. The complexes were detectable with the other mutants, and the intensity of these bands with M11, M12, M14, M18, M19, and M20 were relatively comparable to that with the wild U5RE.   1 and 3) and U5RP-B (lanes 2 and 4) bands, respectively, were analyzed on 10% SDS-PAGE. After electrophoresis, the gel was dried on Whatman paper and exposed on the x-ray film for short (lanes 3 and 4) or long periods (lanes 1 and 2). The molecular masses (kDa) of the markers ( 14 C-labeled methylated protein mixture, high molecular mass range; Amersham Corp.) are indicated to the right.

Sp1 and Sp3 Binding to HTLV-I U5 Repressive Element
The intensity of the bands with M13 and M15 became weaker. These results are summarized in Fig. 6C, indicating that the CACCC sequence is the core binding motif of both of the Group A and U5RP-B. A single point mutation (A to T at the 276 nucleotide of the U5RE region) maintained the binding activity with the U5RE binding proteins (Fig. 6), suggesting that the C(A/T)CCC sequence is the consensus core binding motif.
U5RE-binding Proteins Involve in the Repression of the LTRdirected Expression-As described above, the mutant M17 had no binding ability with the U5RE-binding proteins, but the mutant M13 had weak binding ability. To test whether the U5 region with the M17 or M13 mutation exerts its repressive effect on the LTR-directed expression, three luciferase (luc) expression plasmids derived from the HTLV-I LTR promoter with the wild type, the M13, or M17 mutant type within the U5RE region were constructed as described under ''Experimental Procedures'' and designated pBLTR-Wt-luc, pBLTR-M13luc, or pBLTR-M17-luc, respectively (Fig. 7A). The luciferase activities of these three reporter genes were measured and compared in at least four independent experiments. Fig. 7B shows that the activities of pBLTR-M17-luc appeared to be approximately twice those of the wild pBLTR-Wt-luc or of pBLTR-M13-luc. Therefore, a single point mutation (C to A at the 279 nucleotide of the U5RE region) diminishes not only the binding activity with the U5RE binding proteins but also the repressive effect on the LTR-directed expression. On the other hand, another single point mutation (C to G at the 275 nt) decreases the binding activity but still retain the repressive effect. Thus, we argued that the protein components in U5RP-A1, -A2, -A3, and -B play an important role in the repression of HTLV-I gene expression.
Involvement of Sp1 Family Proteins in U5RP-A1, A2, -A3,  2-12, respectively). U5RP-A1, -A2, -A3, -B, and -C are indicated by A1, A2, A3, B, and C, respectively, at the left. C, the nucleotide sequence of the competitors and probes, and results from the competition and the gel shift assays. The sequence differences between the wild and the mutant competitors and probes are indicated by underlining. The strong or weak positive, or negative reaction is shown by ϩϩ or ϩ, or Ϫ, respectively. and -B-The CCACCC sequence in U5RE is known to be the binding motif of the transcription factors, the Sp1 family (27,28). To test whether Sp1 or Sp3 are included in these complexes, we performed supershift assay with anti-Sp1 or -Sp3 antibody.
Supershift assays with anti-Sp3 antibodies revealed that both U5RP-A2 and -B were not only supershifted but also completely disappeared (Fig. 8, lane 4). This result was further confirmed by the results from supershift assays with a mixture of anti-Sp1 and -Sp3 antibodies (Fig. 8, lane 5). These data showed that both U5RP-B and -A2 include Sp3 or Sp3-related protein and further suggested that U5RP-A3 contains neither Sp1 nor Sp3 proteins.
Competition gel shift assays were performed to further confirm the presence of Sp1 and Sp3 in Group A and U5RP-B. Radiolabeled U5RE probe binding to protein in the nuclear extract was competed with a consensus Sp1 sequence, the wild U5RE and mutant M17 DNAs as a competitor. The Sp1 consensus sequence (5Ј-ATTCGATCGGGGCGGGGCGAGC-3Ј, purchased from Promega) competed more effectively than the wild U5RE for proteins involved in complexes U5RP-A1, -A2 and -A3 as well as complex U5RP-B (data not shown), whereas the M17 sequence had no effect on either complex formation as described above. This result suggested that the binding affinity of U5RP-A1, -A2, -A3, and -B for the Sp1 consensus sequence is much higher than that for U5RE. Therefore, U5RE seems to be distinguishable from the Sp1 consensus element. DISCUSSION Here, we described five different binding complexes to U5RE, namely U5RP-A1, -A2, -A3, -B, and -C. At first, we named Group A inclusive of U5RP-A1, -A2, and -A3 because they were hardly distinguishable without supershift assays using anti-Sp1 and -Sp3 antibodies. These complexes, except for U5RP-C, were found to specifically bind to U5RE. The immunodepression assays revealed that only U5RP-C reacted to the anti-p80/Ku antibody, suggesting that the Ku antigen is involved in U5RP-C. In contrast to this finding, we previously reported that Ku antigen is involved in U5RPs and its binding to U5RE is specific (15). This discrepancy may be explained by the difference in these two experimental conditions; previously, we simultaneously analyzed for binding specificity of the other specific complexes to U5RE involving the nonspecific complexes. Here, we conclude that nonspecific binding of Ku antigen to U5RE occurs. The nonspecific binding affinity of U5RP-C to U5RE might be due to the possibility that Ku antigen also binds to the termini of DNA fragments. Some groups have shown that Ku antigen has binding activity at the double-stranded DNA end (29 -32).
The binding core element of U5RP-A1, -A2, -A3, and -B was determined by competition assays and gel shift mobility assays using a series of mutant U5RE competitor DNAs (Figs. 5 and 6). The results showed that U5RP-A1, -A2, -A3, and -B recognize the CACCC sequence as a core motif with identical affinities. Thereafter, a single point mutation (C to A at the 279 position of the U5RE region) diminished not only the binding activity with the U5RE binding proteins but also the repressive effect on the LTR-directed expression. In addition to the previous report, in which we have observed a 2-5-fold increase in basal promoter activity when the U5RE domain was deleted (15), this evidence strongly suggested that the protein components in U5RP-A1, -A2, -A3, and -B play an important role in the repression of HTLV-I gene expression.
The CACCC sequence is well known as the motif bound to some transcription factors such as Sp1 and Kruppel type zincfinger proteins (28,33). Group A and U5RP-B were then subjected to supershift assays with anti-Sp1 and anti-Sp3 antibod- FIG. 7. Involvement of U5RE-binding proteins in the repression on the LTR-directed expression. A, schematic diagram of three luciferase (luc) expression plasmids, pBLTR-Wt-luc, pBLTR-M13-luc, and pBLTR-M17-luc, in which the luc gene was under the control of the U3-R-U5 (Ϫ325 to ϩ316) region of the wild type LTR and the mutant type LTRs which are introduced as a single point mutation (C to G or A) at nucleotide ϩ275 or ϩ279 of the U5RE region, respectively. B, significant differences in the promoter activity of pBLTR-M13-luc or pBLTR-M17-luc, compared with that of pBLTR-Wt-luc. The mixture of 1 g of each expression plasmids and 2 g of the internal control CAT expression plasmid pRSV-CAT was transfected into Jurkat cells, followed by culturing for 16 h, and was further cultured for 48 h in medium containing 10% fetal bovine serum. Thereafter the supernatants of the cell lysates were recovered and assayed for CAT and luciferase activities. Transfection efficiency was normalized by the CAT activity. Results are the means Ϯ S.E. of the mean for four or six independent experiments. ies. The results indicated that U5RP-A1, corresponding to the slowest band, was found to include Sp1 or Sp1-related protein.
A molecular mass of Sp1 is known to be 100 ϳ 110 kDa (24), which is consistent with that of the slowest band in the UV cross-linking assay as described above, and a 110-kDa protein was previously identified in the U5RE-binding complexes as described (15). Cloning and functional analysis of three other major proteins of about 64, 72, and 76 kDa in Group A will be required to establish their role. Moreover, U5RP-A2 and -B were retarded by gel shift assays using anti-Sp3 antibodies, indicating the possible involvement of Sp3 or Sp3-related protein in both of the U5RP-A2 and -B complexes. A recent report (24) described that the anti-Sp3 antiserum, which we also used in this study, specifically recognized 97-, 60-, and 58-kDa proteins in a nuclear extract from HeLa cells. The 95-kDa protein shown in our UV cross-linking assay seems to correspond to the largest 97-kDa protein in that immunoblot, suggesting that it is Sp3. The intensity of the band of the 52-kDa protein observed to be the strongest. In supershift assays, anti-Sp3 antibodies completely supershifted the U5RP-B complex band. These findings might suggest that this 52-kDa protein associates with Sp3 binding through the U5RE DNA. Further characterization of this 52-kDa protein remains to be determined. The 95-kDa protein was also contained in Group A, and anti-Sp3 antibodies supershifted the U5RP-A2 complex band, suggesting that Sp3 is in U5RP-A2.
As in our results, a similar pattern of DNA specificity was observed using both the GC box (GGGGCGGGC) and the GT motif (GGGTGTGGC) as a probe (24,34). We additionally found the third complex, namely U5RP-A3, specifically binding to the CACCC motif, but these antibodies did not affect the binding of U5RP-A3. Thus, it is remained to be determined what proteins were involved in U5RP-A3. We speculate the possible involvement of the other CACCC binding protein family, such as a Sp1 family, Sp2 or Sp4. It is unlikely, however, that Sp4 is involved, because Sp4 is known to be a brainspecific expression protein (27,28,33).
Our demonstration that Sp-1 family proteins are involved in U5RE-binding will aid in examining the mechanism of the LTR U5-mediated repression. Originally identified as a cellular transcription factor required for SV40 gene expression, Sp1 stimulates transcription by binding to GC-rich promoter elements embedded in a wide variety of cellular and viral promoters (35)(36)(37)(38)(39)(40)(41). The CACCC motif was found in the ␤-globin gene promoter region and was required for efficient and accurate ␤-globin gene expression (42), indicating that the CACCC motif binding proteins have enhancer function. On the contrary, the CACCC motif of U5RE exerts its repressive effect in the LTRmediated expression as shown in the luciferase assays using the mutated promoter. We correlate the reduction of its repressive effect with diminished levels of competition and binding activities in gel mobility shift assays using a series of the one-point mutants within the CACCC motif of U5RE. Thus, the function of U5RE might be as a repressor in vivo. It has been suggested that the same enhancer element, such as CRE and TRE, acts in both positive and negative regulation (43)(44)(45). In this manner, the CACCC motif within U5RE might function as a repressor. Because Sp3 is known to be an inhibitory member of the Sp1 family (24,28), Sp3 inhibition might involve competition with Sp1 for occupancy of the CACCC motif. The finding of a cellular protein, p74, that binds Sp1 in vivo and in vitro via the Sp1 trans-activation domain and exerts its negative regulation in the Sp1-mediated transcription was recently reported (46). In addition, a common sequence, CCACCC, termed the retinoblastoma control element motif, has been identified as being important for conferring retinoblastoma-mediated tran-scriptional repression (47). The Sp1 consensus binding sequence, CCGCCC, can confer equal responsiveness to RB. The retinoblastoma protein is directly or indirectly involved in Sp1 binding. These findings provide evidence for a functional link between retinoblastoma and Sp1. Based on these ideas, there is a possibility that association of Sp1 with some cofactors might result in the U5-mediated repression. Furthermore, an unidentified binding protein(s) to U5RE, which is involved in the U5RP-A3 complex, might be involved in the repression.
Taken together, we propose that the U5RE plays an important role in the down-regulation of HTLV-I gene expression. Moreover, in viral latency, some contribution that HTLV-I gene expression is down-regulated at the transcriptional levels by the U5RE-binding proteins, such as Sp1, Sp3, and others, might be suggested.