Regulation of human B19 parvovirus promoter expression by hGABP (E4TF1) transcription factor.

The genetic expression of human B19 parvovirus is only dependent on one promoter in vivo and in vitro. This is the P6 promoter, which is located on the left side of the genome and is a single-stranded DNA molecule. This led us to investigate the regulation of the P6 promoter and the possible resulting variability of the nucleotide sequence. After analysis of the promoter region of 17 B19 strains, only 1.5% variability was found. More exciting was the finding of mutations that were clustered around the TATA box and defined a highly conserved region (nucleotides 113-210) in the proximal part of the P6 promoter. HeLa and UT7/Epo cell extracts were found to protect this region, which contained a core motif for Ets family proteins, with YY1 and Sp1 binding sites on either side. Gel mobility shift assays performed with nuclear proteins from HeLa and UT7/Epo cells identified DNA-binding proteins specific for these sites. By supershift analysis, we demonstrated the binding of the hGABP (also named E4TF1) protein to the Ets binding site and the fixation of Sp1 and YY1 proteins on their respective motifs. In Drosophila SL2 cells, hGABPalpha and -beta stimulated P6 promoter activity, and hGABPalpha/hGABPbeta and Sp1 exerted synergistic stimulation of this activity, an effect diminished by YY1.

B19 parvovirus is the only member of the Parvoviridae family that is pathogenic for humans (1). It has been associated with a wide range of clinical symptoms and is responsible for erythema infectiosum in children and arthropathy in adults. B19 infections can be particularly severe, leading to hydrops fetalis during pregnancy, transient aplastic crisis in patients with underlying hemolytic diseases, or chronic bone marrow infection in immunocompromised patients (2). In vivo and in vitro, the infection of human bone marrow cells leads to the depletion of the immature erythroid progenitor cells, i.e. burstforming unit erythroid and cluster-forming unit erythroid (3,4). In the latter cells, replication occurs and results in cell cytotoxicity (5). However, despite such remarkable erythroid tropism, which is still unexplained, B19 infection can also impair megakaryocytopoiesis (6,7). Whereas virus replication is responsible for the disruption of erythropoiesis, only viral transcription occurs in megakaryocytes. In these cells, the accumulation of the nonstructural protein NS1 seems to be responsible for cell lysis (8).
B19 virus, like other parvoviruses, is a nonenveloped icosahedral virus with a single-stranded DNA linear genome composed of 5596 nucleotides that encode one nonstructural protein (NS1), two structural proteins (VP1 and VP2), and several small polypeptides of unknown function (9 -11). Both ends of the genome are composed of identical inverted repeat sequences of 383 nucleotides (12). The distal 365 nucleotides are imperfect palindromes that can form a hairpin structure. The transcription map of the B19 parvovirus has been determined in infected human bone marrow cells (9,10). Its only known promoter, named P6 and located in the 5Ј-terminal region, directs the synthesis of up to nine viral transcripts (13,14). Although the mRNAs encoding for the capsid proteins and the small polypeptides are spliced, the NS1 mRNA is not (10).
The regulation of the P6 promoter by viral or cellular proteins has not been extensively studied. In erythroid-permissive cells, this regulation might be preponderant. Thus, a recombinant adeno-associated virus, a defective parvovirus in which the P5 promoter has been substituted for the B19 P6 promoter, is able to replicate specifically and autonomously in erythroid cells (15). However, isolated in front of a reporter gene, the P6 promoter exhibits strong activity in many cell lines, as demonstrated after transfection (14,16,17). Like other parvoviruses, the nonstructural protein NS1 can up-regulate the P6 promoter (14, 18 -20), but the exact mechanism of this up-regulation is not yet clear. The result of a recent study argues in favor of an indirect effect involving Sp1 and cAMP-response element binding proteins, as already demonstrated for other parvoviruses. 1 The Sp1 transcription factor has been implicated in the regulation of the P6 promoter (22). Indeed, two GC box motifs located upstream of the TATA box have been implicated in the in vitro up-regulation of promoter transcription. The YY1 transcription factor also binds the P6 promoter to three different motifs (23), which results in a positive P6 promoter regulation.
In this investigation, we first studied the genetic diversity of the B19 P6 promoter. A highly conserved region was characterized after sequencing 17 B19 strains. Within this region, a large sequence protected by erythroid or nonerythroid nuclear proteins was observed using in vitro footprinting analysis. For the first time, as far as we know, we demonstrated the presence of an Ets binding site (EBS) 2 in the conserved protected region using electrophoretic mobility shift assays (EMSA). By supershift analysis, we characterized the binding of hGABP proteins, an Ets-related transcription factor so far not found to be in-volved in regulating a parvoviral promoter. In addition to the YY1 transcription factor described above, we demonstrated the fixation of the Sp1 factor to a GC box placed just downstream of the EBS. We then defined a 3-fold sequence composed of the YY1, Ets, and Sp1 binding sites. By transfection analysis of a Drosophila cell line, we studied the effect of the B19 P6 promoter regulation by YY1, hGABP, and Sp1 factors. We showed that Sp1 and hGABP activated transcription synergistically throughout this 3-fold sequence. This synergy was abolished by YY1. Of greater interest was the fact that we observed the same results with the P6 native promoter.
Oligonucleotides-Oligonucleotides were synthesized and purified by Genset (France). Complementary strands were phosphorylated with T4 polynucleotide kinase (New England Biolabs), denatured at 88°C for 2 min, and annealed at room temperature. A HindIII restriction site was introduced at the end of each oligonucleotide to facilitate radioactive labeling. The NFB probe sequence was 5Ј-ACGTTACAAGGGACTT-TCCGCTA-3Ј (27).
Plasmid Construction-The pTK plasmid containing the Ϫ50 to ϩ55 region of the herpes simplex virus thymidine kinase (TK) promoter linked to the luciferase gene was constructed as follows. The XbaI/ HincII fragment corresponding to the TK promoter was obtained from pTK-50 (a gift from Dr. F. Thierry, Institut Pasteur, Paris, France) (28). The pTK-50 plasmid was digested with HincII and filled in with the Klenow fragment of DNA polymerase I (New England Biolabs). After digestion with the HindIII enzyme, the 105-bp TK fragment was inserted into the pGL2 basic vector (Promega) at the NheI/HindIII (blunt made) sites to obtain the pTK-LUC plasmid. Oligonucleotides D to G (see Fig. 2) were filled in with the Klenow fragment of DNA polymerase I. They were then introduced into the pTK-LUC plasmid at the SmaI site to obtain the pX-TK-LUC series of plasmids. The orientation and sequence of the recombinant constructs were verified by DNA sequence analysis (T7 sequencing kit, Amersham Pharmacia Biotech).
The pPacNdeI plasmid containing the D. melanogaster actin 5C promoter was a gift from Dr J. Ghysdael (Institut Curie, Orsay, France) (29). YY1 cDNA was obtained by the digestion of the pGEX-YY1 plasmid by EcoRI/BamHI (a gift from Dr. T. Shenk, Princetown, NJ). After filling in with the Klenow fragment of polymerase I, the 1274-bp fragment was inserted at the blunt end-made NdeI site of pPacNdeI to form the pPac-YY1 plasmid. The pPac-ETS1 and pPac-Sp1 plasmids were a gift from Dr. J. Ghysdael (29).
The reporter plasmid, pP6-LUC, carried the P6 promoter and its upstream region followed by firefly luciferase gene. It was constructed as follows. The P6 sequence was extract from the BP06 plasmid (a gift from Dr. P. Beard, ISREC, Lausanne, Switzerland) (14) with the Hin-dIII enzyme and subcloned at the HindIII site of the pGL2 basic vector (Promega). Next, the first 102 bp of the P6 promoter were removed using the BglII and AatII enzymes. After Klenow treatment, the construction was self-ligated to give the pP6-LUC plasmid. The orientation and sequence of the recombinant construct were verified by DNA sequence analysis (T7 sequencing kit, Amersham).
Genetic Analysis-Seventeen B19-PCR positive sera were collected between 1972 and 1995 in our laboratory. They were obtained from 4 blood donors and 12 patients with erythema, arthralgia, and acute or chronic anemia. One patient was asymptomatic. To amplify a 247-bp fragment corresponding to nucleotides 113-360 of the sequence published by Shade et al. (31), we used the primers 5Ј-AAATGACGTAAT-TGTCCGCCATCT-3Ј (nt 113-136) and 5Ј-AGCCCAGAAAGAAA-GAGC-3Ј (nt 360 -343). PCR was run for 30 cycles, each cycle consisting of 30 s at 94°C, 30 s at 52°C, and 30 s at 72°C using Taq polymerase (Boehringer) in a buffer containing 10 mM Tris-HCl, pH 8.3, 50 mM KCl, and 1.5 mM MgCl 2 in a 9600 Perkin-Elmer thermal cycler. For each serum sample, two PCR products were obtained by two independent amplifications. After purification of the 247-bp DNA fragment (Wizard DNA clean-up system; Promega), nucleotide sequences were directly determined by the enzymatic method of Sanger with Taq polymerase (fmol DNA sequencing system; Promega) using [ 35 S]dATP. The primers used for the sequence reaction were the same as for the PCR step, thus allowing both the sense and nonsense DNA strands to be read. Each sequence was determined with two distinct PCR products.
In Vitro DNase I Footprinting Analysis-Nuclear extracts of HeLa and UT7/Epo cells were used for this analysis. The P6 probe was prepared by PCR with a 3Ј 32 P-labeled primer. A 226-bp fragment corresponding to nucleotides 78 -304 of the P6 promoter sequence (31) was amplified from the BP06 plasmid using the primers 5Ј-ATTTCCT-GTGACGTCATTTCCTG-3Ј (nt 78 -100) and 5Ј-ACGCTCCGCCCA-TTTT AACCG-3Ј (nt 304 -283). PCR was run for 35 cycles at 94°for 30 s, 56°C for 30 s, and 68°C for 2 min/cycle with Klen Taq polymerase, according to the manufacturer's instructions (Advantage-GC cDNA PCR kit; CLONTECH). The labeled fragment was purified by native agarose gel electrophoresis. About 0.5 ng of probe was incubated for 15 min at room temperature with 10 g of nuclear extract in 25 l of binding buffer (8 mM HEPES pH 7.9, 8% glycerol, 40 mM KCl, 0.08 mM EDTA, 0.08 mM phenylmethylsulfonyl fluoride, and 0.2 mM dithiothreitol). The probe was then digested for 2 min on ice with various concentrations of DNase I (Sigma) ranging from 20 to 160 g/ml. The reaction was stopped by adding 250 l of 0.05% SDS, 2.5 M EDTA, and 300 M NaCl. After phenol/chloroform extraction and ethanol precipitation, samples were loaded on an 8 M urea, 6% polyacrylamide sequencing gel. Autoradiography was performed for 2-8 days.
Electrophoretic Mobility Shift Assay-The EMSA was performed with nuclear extract or in vitro translated protein and the D probe corresponding to the region spanning nucleotides 126 -158 of the P6 promoter sequence (see Fig. 2). This oligonucleotide probe was labeled by filling in with the Klenow fragment of DNA polymerase I in the presence of [␣-32 P]dCTP. First, 2.5 g of cell extract or 1 l of in vitro translated protein were incubated for 10 min at room temperature in the binding buffer (4% Ficoll, 20 mM HEPES, pH 7.5, 70 mM NaCl, 2 mM dithiothreitol, 100 g/ml bovine serum albumin, and 0.01% Nonidet P-40) with 1 g of poly(dI-dC)⅐poly(dI-dC) (Amersham) and 0.5 g of salmon sperm DNA. Next, either competitor or antibodies were added when indicated and then 1 l of the 32 P-labeled probe (about 20,000 cpm). The preparation was left to stand for 25 min at room temperature and then underwent electrophoresis on 5% polyacrylamide gel in 0.5 ϫ Tris borate/EDTA buffer (45 mM Tris borate, 1 mM EDTA). Lastly, the gel was dried and autoradiographed.
Transfection and Luciferase Assay-One day before transfection, SL2 cells were seeded at 2 ϫ 10 6 cells/35-mm well. They were co-transfected with the indicated amount of expression plasmid and 1 g of reporter plasmid using the calcium phosphate coprecipitation method (33). The total amount of DNA was kept constant at 10 g by adding nonrecombinant expression plasmid (control plasmid). After the addition of DNA, the plates were left undisturbed until the time of harvest 48 h later. Cells were then washed once with phosphate-buffered saline, and luciferase activity was measured in a luminometer (Lumat LB 9501, Berthold), as described previously (20). To normalize the luciferase assay, the total protein concentration was evaluated for each cell lysate (protein assay reagent, Bio-Rad).

RESULTS
Genetic Diversity Analysis-As stated above, the 17 B19 PCR-positive sera collected between 1972 and 1995 in our laboratory were obtained from 4 blood donors and 12 patients with erythema, arthralgia, and acute or chronic anemia. One patient was asymptomatic. To analyze the nucleotide composition of the P6 promoter, we amplified a 247-bp fragment including the TATA box. This fragment corresponded to nucleotides 113-360 of the sequence published by Shade et al. (31). For each serum, two PCR products were sequenced independently. Twenty nucleotide modifications (18 substitutions, 1 deletion, and 1 insertion) were observed in comparison with the reference sequence (31) given an average of 1.5% of variation (data not shown). However, these mutations were not equally distributed on the P6 promoter and allowed two regions to be distinguished. All the mutations were situated between nucleotides 210 and 340, and a highly conserved region comprising nucleotides 113-210 was observed on the 5Ј side of the amplified promoter. Even though this conserved region corresponded to the palindromic sequence of the hairpin terminus indispensable for parvovirus replication, we could not rule out the possibility that this region has an important role in regulating transcription, as recently suggested. 1 Hence, we examined the cellular factors interacting with the conserved sequence comprising nucleotides 113-210 of the B19 P6 sequence.
Protection Analysis of the Highly Conserved P6 Region-The ability of cellular proteins to bind the highly conserved P6 region was explored using in vitro DNase I footprinting assays. The double-stranded P6 probe covered nucleotides 78 -304 of the sequence previously described (31). Only one extremity of the double strand probe was labeled alternatively and incubated with nuclear extracts from epithelial HeLa and erythroid UT7/Epo cells. A large protected region was observed within the conserved region with the nuclear extract from UT7/Epo cells (Fig. 1). By comparison with a DNA ladder, the protected region covered the nucleotides 129 -146 (Ϫ220/Ϫ203) on the sense strand and nucleotides 129 -150 (Ϫ220/Ϫ199) on the nonsense one (Fig. 1). The same footprint was observed with the extract from HeLa cells (data not shown). In the protected region, a YY1 binding site was described previously between nucleotides Ϫ220 and Ϫ212 (23) and an Sp1 binding site between nucleotides Ϫ200 and Ϫ195 (16,22). Between these two protected regions, we noticed for the first time as far as we know, an Ets family consensus binding site CCGGAAGT located between nucleotides Ϫ208 and Ϫ201 (Fig. 2). These results suggested that nuclear proteins from HeLa or UT7/Epo cells were able to protect the nucleotide sequence spanning nucleotides 129 -150 (Ϫ220/Ϫ199) in the highly conserved re-gion of the P6 promoter.
Sp1, YY1, and GAPB Bind to the Conserved Part of the P6 Region-To confirm the binding of eukaryotic proteins to nucleotides 129 -150 (Ϫ220/Ϫ199) of the conserved region, we performed EMSA with the radiolabeled D probe (Fig. 2) corresponding to the binding sites of YY1 (Ϫ220/Ϫ212), EBS (Ϫ208/ Ϫ201), and Sp1 (Ϫ200/Ϫ195). When the D probe was incubated with nuclear extracts from HeLa and UT7/Epo cells, it generated seven retarded complexes (Fig. 3A). The migration of six of these complexes was the same in each cell type, whereas the migration of the complex V was more delayed with HeLa proteins than with UT7/Epo proteins (compare lanes 2 and 8). All these seven complexes proved to be specific, since competition with a 30-fold excess of an unrelated sequence (the NFB region of the HIV-1 long terminal repeat) did not affect their formation (Fig. 3A, lanes 4 and 10). This was confirmed by a competition assay with the cold D probe, which inhibited nucleoprotein complex formation (lanes 3 and 9). Experiments involving competition between the D probe and the cold A, B, C probes were used to link binding sites of the YY1, EBS, or Sp1  1-4). The probe (nt 78 -304) was only labeled on the upper or sense strand (lanes 1 and 2) or only on the lower or nonsense strand (lanes 3 and 4). The protected DNA sequences are shown as solid bars. transcription factors with the cell protein complexes. The formation of complexes I and II was specifically inhibited by the presence of the cold C probe corresponding to the Sp1 binding site (lanes 7 and 13). The intensity of complex I was not the same with the two cell protein extracts, as the HeLa extract produced a stronger signal than the UT7/Epo, suggesting that different concentrations of the proteins were involved in this complex. Complexes III, IV, and V disappeared when the B probe corresponding to the EBS was used (lanes 6 and 12). Several complexes are usually observed for the EBS since different Ets proteins recognize the same DNA sequence (34). Nevertheless, complex V did not have the same mobility pattern with the two nuclear extracts, suggesting the binding of two different members of the Ets family. Finally, the A probe corresponding to the YY1 binding site inhibited the formation of complexes I, VI, and VII (lanes 5 and 11). Complex I formation was also inhibited by competition with the C probe ( lanes  7 and 13), suggesting that its formation involved the binding of the two proteins YY1 and Sp1 to the D probe. The two other complexes, VI and VII, probably corresponded, respectively, to the binding of the complete and the truncated forms of the YY1 protein, as previously suggested (35,36).
To identify the components of the seven complexes, we performed a gel shift assay with the D probe in the presence of specific antisera. Complex I was supershifted by the two polyclonal antisera raised against Sp1 and YY1 (Fig. 3, B and C,  lanes 3 and 4). Complexes VI and VII, ascribed to the YY1 binding site, were shifted by the polyclonal antibody that recognized the YY1 factor (lane 3). Complex II was supershifted by the polyclonal antibody raised against Sp1 (lane 4). Among the eight antisera raised against the proteins of the Ets family, only anti-GABP␣ and GABP␤ inhibited the formation of complexes III and IV (lanes 7 and 8). None of these antisera modified the migration of either of the two V complex. All these results were obtained with nuclear extracts from either nonerythroid HeLa cells (Fig. 3B) or erythroid UT7/Epo cells (Fig.  3C). These results provided evidence that Sp1, YY1, and GABP␣ and -␤ transcription factors all bind the D probe cor-  c, a, and b, respectively. responding to the 3-fold sequence YY1-EBS-Sp1 of the B19 P6 promoter.
To confirm these results, we tested in vitro translated hGABP␣, hGABP␤1, and hGABP␤2 proteins for binding to the EBS motif of the 3-fold sequence (Fig. 3D). hGABP␣ produced a complex c (lane 3), whereas neither hGABP␤1 nor hGABP␤2 alone bound to the D probe (lanes 4 and 5). In contrast, when a mixture of hGABP␣ and hGABP␤1 or of hGABP␣ and hGABP␤2 was tested, an intense band a or b of lower mobility was produced along with the abolition of complex c (lanes 6 and  7). The mobility of complexes a and b was very similar to that of bands III and IV produced by HeLa or UT7/Epo nuclear proteins (lanes 12 and 13). Furthermore, these complexes produced by hGABP proteins were specific, because competition with a 30-fold excess of probe A, corresponding to the EBS, inhibited their formation (lanes 9 -11).
These data clearly demonstrate that bands III and IV formed by the nuclear extracts corresponded to the heterodimers of hGABP␣/hGABP␤1 and hGABP␣/hGABP␤2, respectively. In addition, the YY1 protein formed complexes VI and VII, and the Sp1 protein formed complex II. A band comprising Sp1 and YY1 was also detected (I), but no complex was composed of three proteins. Lastly, we were unable to identify the Ets factor involved in complex V.
Effect of Overexpression of YY1, hGABP, and Sp1 on Transcription-The transcriptional effect of the YY1, hGABP, and Sp1 factors on the 3-fold sequence was evaluated by transfection. We isolated the wild-type 3-fold sequence (nt Ϫ223/Ϫ192, corresponding to oligonucleotide D in Fig. 2) or the sequences mutated on the three different sites (oligonucleotides E, F, G in Fig. 2) upstream of the minimal promoter of the herpes simplex virus TK and the gene encoding the firefly luciferase. The resulting constructs, pD-TK-LUC, pE-TK-LUC, pF-TK-LUC, and pG-TK-LUC were co-transfected with expression vectors for Sp1, YY1, hGABP␣, and -␤ (pPac-Sp1, pPac-YY1, pPac-GABP␣, and pPac-GABP␤, respectively) into D. melanogaster SL2 cells, which are devoid of endogenous Sp1 and YY1 (29). After 48 h of incubation, luciferase activity was then estimated in cell extracts. Luciferase activity levels were calculated in relation to the activity in cells into which only the reporter plasmid was transfected. The results illustrated in Fig. 4A show that Sp1 protein activated LUC expression in a dose-dependent manner from the pD-TK-LUC plasmid harboring the wild-type 3-fold sequence, and that this activation was impaired by a mutation that suppressed the binding of Sp1 (Fig.  4A, compare lanes 1-3 with lane 4). On the other hand, YY1 protein had no significant effect on pD-TK-LUC transcription (Fig. 4A, lanes 5-7). When, the effect of hGABP proteins on the activity of the 3-fold sequence was similarly investigated, no activation was observed when each subunit was present alone (Fig. 4B, lanes 1-3). In addition, transcription was activated when hGABP␣ was coexpressed with hGABP␤1 or ␤2 (lanes  4 -9). Nevertheless, the formation of an increased amount of ␤ subunits reduced transcription activation. Such an effect was not observed with another viral promoter, the adenovirus E4 promoter (30). All these results indicate that Sp1 and hGABP complexes are able to transactivate the 3-fold sequence in SL2 cells. The YY1 transcription factor had no effect on transcription whatever the concentration of expression plasmid used.
Synergistic Activation by hGABP and Sp1-To observe the possible functional interplay between the Sp1, YY1, and hGABP factors, we performed co-transfection experiments in SL2 cells using the pD-TK-LUC plasmid, the reporter plasmid construction pP6-LUC, which included the entire promoter region (nt 102-480) and the expression plasmids for the factors. When the reporter plasmid pD-TK-LUC was used, as depicted in Fig. 5A, the concomitant production of Sp1 and hGABP␣ ϩ ␤1 resulted in reporter gene levels that were seven to ten times higher than those achieved by each factor alone (compare lanes 1, 3, and 5). This effect was not observed when Sp1 and YY1 or YY1 and hGABP were co-expressed (Fig. 5A, lanes 4 and 6). Nevertheless, when the plasmid expressing YY1 was added to the Sp1 and hGABP plasmids, the synergistic effect of Sp1 and hGABP was 2 times inhibited (78-fold activation, lane 5 versus 32-fold activation, lane 7). Similar results were obtained when using a plasmid expressing hGABP␤2 for co-transfection (data not shown). When we transfected the pP6-LUC reporter plasmid, we observed the same results as with pD-TK-LUC construct (Fig. 5B). Thus, we found a synergistic activation of the P6 promoter by Sp1 and hGABP␣ ϩ ␤1 (compare lanes 1 and 3 with lane 5 in Fig. 5B). We also observed that this activation was inhibited by the YY1 transcription factor (compare lanes 5 and 7 in Fig. 5B). All these results indicate that Sp1 and the dimer formed by hGABP␣ and hGABP␤ cooperate in the activation of the 3-fold sequence formed in SL2 cells by the binding sites Sp1, EBS, and YY1. Furthermore, this cooperation seemed to be partially inhibited by the YY1 protein. Nevertheless, these effects were weaker when the 3-fold sequence was situated in the promoter context (compare Fig. 5, A and B), possibly due to the involvement of other factors that might regulate other regions of the P6 promoter. The fact that the synergistic effect between Sp1 and hGABP was only partially affected by the mutation of one of their binding sites (data not shown) could suggest a mechanism involving protein-protein interaction. Nevertheless, this hypothesis is actually under investigation in our laboratory.

DISCUSSION
Contrary to other parvoviruses whose genetic expression is controlled by two functional promoters located on the left and the middle of the genome, only one promoter has been described for the human parvovirus B19. We therefore attempted to analyze its variability by PCR amplification and sequencing of the promoter region of 17 B19 virus strains. Our data show an average variation of 1.5%, in agreement with previous results. 1 The mutations were clustered around the TATA box, and a highly conserved region was observed in the area proximal of the P6 promoter. This region was located between nucleotides 113 and 210. When it was deleted, there was a 90% loss in transcriptional activity. 1 Therefore we decided to examine the cellular factors that interact with this region. By DNA footprinting assays on both strands, a part of the conserved region was seen to be protected by nuclear extracts from the nonerythroid HeLa cell line or the erythroid UT7/Epo cell line.
Previous footprinting experiments showed that this region interacts with the proteins of HeLa and CEM cells (16,22). Here, we noticed that an 8-nucleotide EBS site CCGGAAGT (Ϫ208/ Ϫ201) is located between the binding sites YY1 (Ϫ220/Ϫ212) and Sp1 (Ϫ200/Ϫ195). An EBS was described earlier in the P4 promoter of the minute virus of mice and was also located immediately upstream from the GC box that binds the Sp1 transcription factor (37). Using EMSA with synthetic oligonucleotidic probes and competition assays with the corresponding probes, we confirmed the binding of YY1 previously observed by Momoeda et al. (23). Binding was also detected at this site for Sp1, contrarily to the findings of Liu et al. (16). The results were similar whether the extract used was from HeLa or UT7/Epo cells. Supershift analysis allowed us to establish that the Ets motif at nucleotides Ϫ208/Ϫ201 in the conserved B19 promoter region is recognized by hGAPB␣, a ubiquitously expressed Ets protein. In addition, antibodies confirmed that HeLa and UT7/Epo binding complexes contain proteins that are immunologically related to both hGABP␣ and hGABP␤. The complexes produced by the hGABP proteins were specific, as evaluated by probe competition. hGABP (also named E4TF1) is indeed composed of three distinct polypeptides: hGABP␣ (60 kDa), hGABP␤1 (53 kDa), and hGABP␤2 (47 kDa), all of which are required for high affinity DNA binding (␣ subunit) and transcriptional activation (␣ and ␤ subunits) (38,39). hGABP binds to a purine-rich cis-regulatory element required for the VP16-mediated activation of herpes simplex virus immediate early gene and regulates adenovirus E4 gene transcription (39,40). We therefore investigated the possible involvement of hGABP in the regulation of the B19 promoter. hGABP was shown to activate a 3-fold sequence comprising YY1-GABP-Sp1 binding sites with the TK minimal promoter in Drosophila SL2 cells; this activation was also found with the P6 promoter. It is noteworthy that our GABP binding site was immediately adjacent to the SP1 site. Ets-related transcription factors such as hGABP are often found in large complexes with other transcription factors (41)(42)(43)(44)(45)(46)(47). For example, Ets-1 and Sp1 interact to activate synergistically the human T-cell lymphotrophic virus long terminal repeat (29). In addition, Sp1 activity is known to be modulated by factors that recognize the DNA elements flanking or overlapping a GC box (48,49). In the present work, Sp1 transactivated the 3-fold sequence and the P6 promoter and displayed synergistic activation with hGABP. Similarly, by co-transfection experiments using also Drosophila SL2 cells, the P4 promoter of minute virus of mice was found to be transactivated synergistically by Ets-1, the prototype member of the Ets family of transcription factors, and the Sp1 factor that binds to a GC box flanking the EBS motif (37). In our study, the mutations of the GABP and Sp1 sites suggest that the combined synergistic effect of the corresponding transcription factors seems to incriminate DNA binding but also protein interactions. Whatever the precise mechanism under investigation, this cooperation was partially inhibited by YY1 protein. In adeno-associated virus, YY1 was found to act as a repressor of transcription from the adeno-associated virus P5 promoter, which is relieved by EIA proteins (21). For B19 parvovirus, the positive effect of YY1 on transcription was described by Momoeda et al. in HeLa cells but was very weak, i.e. 1.3-1.9-fold above basal transcription (23). We did not find that YY1 had any effect on the 3-fold sequence or the P6 promoter in SL2 cells. This difference may be due to the type of cells transfected.
The present study demonstrated, for the first time as far as we know, that the specific DNA-binding proteins for the CCG-GAAGT motif of the human B19 parvovirus promoter is very likely to be hGABP, as indicated by the following results. (i) The DNA protein complex detected in the gel shift assay was abolished by the competition assay, (ii) antibodies against GABP␣ and -␤ subunits supershifted this complex, and (iii) the combination of in vitro translated hGABP␣ and -␤ proteins produced a complex with essentially the same mobility as that FIG. 5. Synergistic activation by hGABP and Sp1. A, 1 g of pD-TK-LUC was co-transfected in SL2 cells with 0.1 g of pPac-Sp1, 0.5 g of pPac-YY1, and 1 g of both pPac-GABP␣ and pPac-GABP␤1 as indicated under the graph. The levels of luciferase activity are expressed relative to the activity of cells in which only pD-TK-LUC was transfected. B, SL2 cells were co-transfected with 1 g of pP6-LUC plasmid and 0.5 g of pPac-Sp1, 0.5 g of pPac-YY1, and 1 g of both pPac-GABP␣ and pPac-GABP␤1 plasmids, as indicated under the graphic. The levels of luciferase activity are expressed relative to the luciferase activity of cells in wich pP6-LUC was transfected alone. Data are means Ϯ S.E. of three separate experiments. produced by the HeLa or UT7/Epo cell extract. Lastly, our results clearly demonstrated that in nonerythroid cells, hGABP proteins, ubiquitously expressed Ets protein, stimulate the expression of the human B19 parvovirus promoter. The precise mechanism of the synergy exerted by hGABP and Sp1, which is diminished by YY1, is currently under investigation.