Replication Protein A Is a Component of a Complex That Binds the Human Metallothionein IIA Gene Transcription Start Site*

Previous studies revealed that sequences surround- ing the initiation sites in many mammalian and viral gene promoters, called initiator (Inr) elements, may be essential for promoter strength and for determining the actual transcription start sites. DNA sequences in the vicinity of the human metallothionein IIA (hMTIIA) gene transcription start site share homology with some of the previously identified Inr elements. However, in the present study we have found by in vitro transcrip- tion assays that the hMTIIA promoter does not contain a typical Inr. Electrophoretic mobility shift assays identi- fied several DNA-protein complexes at the hMTIIA gene transcription start site. A partially purified protein frac- tion containing replication protein A (RPA) binds to the hMTIIA gene transcription start site and represses tran- scription from the hMTIIA promoter in vitro . In addition, overexpression of the human 70-kDa RPA-1 protein represses transcription of a reporter gene controlled by the hMTIIA promoter in vivo . These findings suggest that hMTIIA transcription initiation is controlled by a mechanism different from most mammalian and viral promoters and that the previously identified RPA may also be involved in transcription regulation. Analysis of a large number of mammalian gene promoters revealed the existence of a conserved 5 (cid:57) -TATA-3 (cid:57) sequence centered at around (cid:50) 30 (relative

Previous studies revealed that sequences surrounding the initiation sites in many mammalian and viral gene promoters, called initiator (Inr) elements, may be essential for promoter strength and for determining the actual transcription start sites. DNA sequences in the vicinity of the human metallothionein IIA (hMTIIA) gene transcription start site share homology with some of the previously identified Inr elements. However, in the present study we have found by in vitro transcription assays that the hMTIIA promoter does not contain a typical Inr. Electrophoretic mobility shift assays identified several DNA-protein complexes at the hMTIIA gene transcription start site. A partially purified protein fraction containing replication protein A (RPA) binds to the hMTIIA gene transcription start site and represses transcription from the hMTIIA promoter in vitro. In addition, overexpression of the human 70-kDa RPA-1 protein represses transcription of a reporter gene controlled by the hMTIIA promoter in vivo. These findings suggest that hMTIIA transcription initiation is controlled by a mechanism different from most mammalian and viral promoters and that the previously identified RPA may also be involved in transcription regulation.
Analysis of a large number of mammalian gene promoters revealed the existence of a conserved 5Ј-TATA-3Ј sequence centered at around Ϫ30 (relative to the start of initiation) and a conserved 5Ј-CA-3Ј sequence with A at ϩ1 (1). The TATA element binds TFIID and directs RNA polymerase and other required basal initiation factors to assemble a stable preinitiation complex. The importance of a TATA element was demonstrated in studies where deletion of this element from a promoter resulted in spurious initiation and a low level of transcription (reviewed in Refs. 2 and 3). However, not all eukaryotic genes contain a TATA consensus sequence upstream of their transcription initiation. Promoter studies of some of these TATA-less genes suggest that transcription initiation relies heavily on sequences surrounding the start site called initiator (Inr) 1 elements (4 -7). In one study, a 17-base pair element surrounding the transcription start site of the terminal deoxynucleotidyltransferase gene was sufficient for accurate basal transcription of this gene both in vitro and in vivo. In the presence of either a TATA box or the SV40 21-base pair repeats, a greatly increased level of transcription initiates specifically from this element (7).
Inr elements can be divided into many categories. Although they were first discovered in TATA-less promoters, many TATA-containing promoters also possess Inr elements (7)(8)(9)(10)(11)(12). Some Inrs are weakly bound by a general transcription factor such as TFII-I, RNA polymerase II, or TFIID, but they appear to lack recognition sites for specific DNA binding proteins (9,11,13), whereas others contain high affinity recognition sites for specific DNA-binding proteins (6, 12, 14 -16).
There are many definitions for an Inr in the current literature. The original definition for an Inr was a discrete promoter element that can act alone or in concert with either a TATA box or upstream element to direct specific transcription initiation (7). Later, the term Inr was restricted to the terminal deoxynucleotidyltransferase Inr and elements that have sequence homology to the terminal deoxynucleotidyltransferase Inr (10). At another extreme, the Inr has been used to describe any promoter element that determines the transcription start site (17). More recently, Inr elements have been defined as elements that can localize a transcription start site and mediate the action of at least some upstream activators in the absence of a TATA box (18). Whatever the definition, Inr elements are widespread in mammalian and viral promoters. In one study, O'Shea-Greenfield and Smale (10) inserted start site regions from eight different promoters downstream of the SV40 21-base pair repeats and found that all except the HIV-1 promoter exhibited Inr activity. Subsequently, it was shown that the HIV-1 core promoter lacks a simple Inr element but contains a bipartite activator at the transcription start site (19). The important conclusion is that start site region from many but not all mammalian and viral genes contain Inr activity. In fact, before the discovery of Inr elements, it was found that many DNA elements surrounding the start site of transcription can activate transcription (for examples see Refs. 20 -29). Given the apparent importance of Inr elements and the wide distribution of them in many mammalian and viral gene promoters, there is a definite need to understand the mechanism by which they function and how they may regulate transcription.
We have previously shown that the adeno-associated virus P5 promoter initiation region binds the transcription factor YY1 (12,30). We have also found that this initiation sequence shares homology with the terminal deoxynucleotidyltransferase Inr and is important for accurate basal transcription. Further, partially purified YY1 can restore basal level transcription from a P5 Inr in a HeLa extract depleted of YY1 or a Drosophila embryo extract devoid of YY1 activity, whereas a YY1-specific antibody can block the reactivation. This was one of the first studies to identify a cellular factor that can mediate transcription through an Inr element and place the P5 Inr in the distinct class of Inrs that bind a sequence-specific protein.
Interestingly, a computer alignment of eukaryotic promoter initiation sequences (at the ϩ1 nucleotide) revealed that the sequence immediately surrounding the human metallothionein IIA (hMTIIA) gene transcription start site exhibits homology to the P5 Inr (P5, 5Ј-GGTCTCCATTTTG-3Ј; hMTIIA, GCACTC-CACCACG). This hMTIIA transcription start site is identical to a sequence found in a fish Xiphophorus maculatus promoter (31) and the promoter of the human ␣-2 macroglobulin gene (32). It also shares over 80% identity with a sequence located in the human androgen receptor promoter (33). It has not yet been established whether sequences in these other promoters are part of a transcription start site or whether they play an essential role in transcription. Nevertheless, this high conservation of DNA sequences suggests that the hMTIIA transcription start site may be important for accurate regulation of transcription of the hMTIIA and other genes and thus may represent an individual recognition site for a component of the transcriptional machinery. In fact, it has been suggested that a specific protein may bind to this site in the hMTIIA promoter (34,35).
In addition to the fact that the hMTIIA transcription start site contain DNA sequences similar to other Inrs, we were interested in examining the hMTIIA gene transcription start site for other reasons. The hMTIIA gene is a good example of the complexity of protein-DNA interactions that can occur at a single promoter to influence the transcription of a gene. Although extensive analysis of the 5Ј-flanking regions of this promoter revealed a variety of trans-acting factors that bind to different upstream cis-acting sites (34 -38), the role of the initiation region in basal transcription regulation is not known. A detailed study of the importance of the hMTIIA initiation site alone and characterization of possible factors that may interact with this region has not yet been done.
In spite of the homology to known Inrs in other mammalian and viral gene transcription start sites, we demonstrate in these studies that the hMTIIA transcription start site lacks a functional Inr element. More interestingly, we have demonstrated that several proteins bind to this transcription start site and repress transcription. One of these proteins has been identified as the human replication protein A (RPA), originally found as a multisubunit protein being absolutely required for SV40 DNA replication (39 -41). This observation suggests that in addition to its role in DNA replication, genetic recombination (42,43), and DNA repair (44 -46), RPA may function as a DNA sequence-specific binding transcription factor.
Monoclonal Antibodies-Mouse monoclonal antibodies that recognize different RPA subunits have been described (56). Monoclonal antibody to transcription factor Sp1 was obtained from Santa Cruz Biotechnology. All antibodies were purified by protein A-Sepharose chromatography, concentrated with ammonium sulfate, and dialyzed with 50 mM Tris, pH 8, 150 mM KCl, and 0.02% NaN 3 .
In Vitro Transcription and Primer Extension-HeLa nuclear extracts were prepared by the method of Dignam et al. (57). Each reaction (in a 25-l total volume) contained 500 ng of DNA template, 10 mM MgCl 2 , 40 ng of poly(dG-dC), 2.4 mM HEPES, pH 7.9, 2.4% (v/v) glycerol, 12 mM KCl, 0.024 mM EDTA, 0.06 mM DTT, 0.5 mM ATP, 0.5 mM GTP, 0.5 mM CTP, 0.5 mM UTP, 1 mM creatine phosphate, and approximately 60 -80 g of HeLa cell nuclear extracts. Reactions were incubated for 1 h at 30°C and terminated by the addition of 225 l of stop buffer (1% SDS, 10 mM Tris-HCl, pH 7.4, 10 mM EDTA, pH 8.0, and 5 g/ml yeast tRNA). Following phenol-chloroform extraction and ethanol precipitation, template DNAs were degraded with 5 units of RNase-free DNase (Promega) for 1 h at 37°C. In some reactions, different amounts of an oligodeoxynucleotides were added to the mixture prior to the addition of ribonucleotides.
The RNA products were analyzed by primer extension (58). Briefly, after ethanol precipitation, RNA pellets were resuspended in 15 l of RNase-free buffer containing 150 mM KCl, 10 mM Tris-HCl, pH 8.3, 1 mM EDTA, and radiolabeled oligodeoxynucleotide primer. RNA and primers were hybridized by incubation for 90 min at 60°C and then allowed to cool slowly to room temperature. For extension reactions, 30 l of reaction mixture (30 mM Tris-HCl, pH 8.3, 15 mM MgCl 2 , 8.3 mM DTT, 0.225 mg/ml actinomycin D, 0.22 mM of dNTPs, and 0.66 units/l avian myeloblastosis virus reverse transcriptase) were added to each sample and incubated for 1 h at 42°C. 32 P-Labeled extension products were extracted with phenol-chloroform, precipitated with ethanol, resuspended in formamide loading dye, and separated on 8% sequencing gels. After electrophoresis, gels were dried and subjected to autoradiography.
Electrophoretic Mobility Shift Assays-Single-stranded oligodeoxnucleotides were labeled individually with [␥ 32 P]ATP and T4 polynucleotide kinase, heated together at 65°C, and allowed to anneal by slow cooling to room temperature. Each reaction contained 20 fmol of labeled DNA, 12 mM HEPES, pH 7.9, 10% glycerol, 5 mM MgCl 2 , 60 mM KCl, 1 mM DTT, 50 g/ml bovine serum albumin, 0.5 mM EDTA, 0.05% Nonidet P-40, 0.1 or 1 g of poly(dI-dC), and approximately 7 g of HeLa cell extract or approximately 0.4 g of purified protein in a 12-l total volume. Unlabeled specific and nonspecific competitors or antibodies were included in some reactions. Reactions were incubated for 10 min at room temperature, separated on 4% nondenaturing polyacrylamide gel (0.0225 M Tris borate, and 0.0005 M EDTA), dried, and subjected to autoradiography.
Purification of hMTIIAϩ1 Binding Proteins-A nuclear extract was prepared from 100 liters of HeLa cells by the method of Dignam et al. (57). This extract (185 ml, 2.3 g) was dialyzed against buffer A (20 mM HEPES, pH 7.9, 10% glycerol, 50 mM KCl, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5 mM DTT) and then applied to a 2.6 ϫ 40-cm P11 phosphocellulose column. After washing with buffer A, bound proteins were eluted with a 1.5-liter gradient from 0.05 to 1 M KCl in buffer A. Protein-DNA complexes were detected by EMSA described above. Proteins forming complexes I and III eluted at about 250 mM KCl. These fractions were pooled (100 ml, 190 mg), dialyzed against buffer B (50 mM Tris-HCl, pH 7.5, 10% glycerol, 50 mM KCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 0.5 mM DTT), and then applied to a 1.6 ϫ 20-cm DEAE-Sephacel column. Bound proteins were eluted with a 400-ml gradient from 0.05 to 1 M KCl in buffer B. Proteins forming complex III (34 ml, 21 mg) eluted at 250 mM KCl were partially resolved from those forming complex I. After dialysis against buffer B, the sample was applied to an FPLC HR5/5 Mono Q column, and bound proteins were eluted with a 20 ml gradient from 0.05 to 0.75 M KCl in buffer B. Proteins forming complex III (8 ml, 14 mg) eluted at 350 mM KCl were resolved from those forming complex I. This fraction was then separated by gel filtration through a 1.6 ϫ 60-cm Superdex 200 column (two identical fractions of 4-ml samples) that had been pre-equilibrated with buffer B containing 0.25 M KCl. Active fractions (26 ml, 2 mg) were dialyzed against Buffer C (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.5 mM DTT, and 1 M ammonium sulfate) and then applied to an FPLC HR 5/5 phenyl superose column. Bound proteins were eluted with a 20-ml linear gradient from 1 to 0 M ammonium sulfate in buffer C. Active fractions (10 ml, 200 g) that eluted at about 0.3 M ammonium sulfate were dialyzed against buffer B and then fractionated by FPLC Mono Q chromatography as described above.
Proteolytic Digestion, HPLC Separation, and Microsequencing of hMTIIAϩ1 Binding Proteins-The most highly purified fractions that formed complex III were pooled and concentrated by ultrafiltration using a Centricon-10 filter (Amicon) as described by the manufacturer. Polypeptides were separated by electrophoresis through a 10% denaturing polyacrylamide gel and then transferred to a polyvinylidene difluoride membrane. Polypeptides on the membranes were visualized by staining with 0.1% Ponceau S in 5% acetic acid. After destaining with 5% acetic acid, the appropriate strips of membrane were excised, washed with H 2 0, and stored at Ϫ20°C prior to in situ digestion with trypsin (59). The resulting peptide mixture was separated by narrow bore high performance liquid chromatography using a Vydac C18 2.1 ϫ 150-mm reverse phase column on a Hewlett-Packard 1090 HPLC/1040 diode array detector. Optimum fractions from each peptide chromatogram were chosen based on differential UV absorbance at 210, 277, and 292 nm, peak symmetry, and resolution. Peaks were further screened for length and homogeneity by matrix-assisted laser desorption timeof-flight mass spectrometry on a Finnigan Lasermat 2000. Selected fractions were submitted to automated Edman degradation on an Applied Biosystems 477A or Hewlett Packard G1005 protein sequencer. Details of strategies for the selection of peptide fractions and their microsequencing have been previously described (60).
Transfection, CAT, and ␤-Galactosidase Assays-CV1 cells were grown on 60-mm plates in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transfections were done with the calcium phosphate method (61). 48 h after transfection, cells were lysed by repeated freeze/thaw cycles, and extracts were assayed for CAT activity as described previously (52). ␤-Galactosidase assays were performed using the Galacto-Light kit (Tropix).

RESULTS
The hMTIIA Promoter Does Not Contain a Typical Functional Inr-Previously, using in vitro transcription assays, we and others have demonstrated that many mammalian gene transcription start sites possess Inr activities. Here, we tested the ability of the hMTIIA transcription start site (Ϫ7 to ϩ11 [hMTIIAϩ1]) to direct and activate transcription in the presence or the absence of either the adenovirus major late TATA box (Fig. 1, lanes 2 and 3) or multiple Sp1 binding sites (Fig. 1,  lane 4). Compared with the P5 Inr, the hMTIIA initiation site gave no detectable transcription activity with or without upstream activator binding sites. This is somewhat surprising, given the earlier findings (12,47) that a major late TATA box or multiple Sp1 sites alone can confer spurious and low level initiations (although it has been suggested that these low level initiations may be artificial). All transcription reactions were repeated at least three times with consistent results using different nuclear extract and template preparations. Therefore, at least by two definitions (7,18), unlike many mammalian and viral promoters, the hMTIIA transcription start site does not possess an Inr.
Two models could be invoked to explain our observation of the lack of an hMTIIA Inr activity. First, it is possible that proteins, either a universal Inr-binding protein or others that may augment the strength of an Inr, do not bind the hMTIIA start site. Alternatively, one can imagine that similar to the SV40 major late promoter transcription start site (62), the hMTIIA start site could bind a transcription repressor. Of course, these two models need not be mutually exclusive; it is also possible that the hMTIIA start site binds to a combination of positive and negative factors, and the net result is a lack of Inr activity. Any rigorous test of these models requires the identification of DNA sequence that conveys Inr activity and characterization of proteins that interact with the hMTIIA start site DNA.
In Vitro Transcription Activity from the hMTIIAϩ1 Sequence Can Be Restored by Two Base Changes or Addition of hMTIIAϩ1 DNA-Five plasmids were constructed to contain DNA sequences mutated to mimic the P5 Inr. These plasmids were derived from pMLTATA/hMTIIAϩ1 (Fig. 1, lane 3, and Fig. 2A, lane 8), which contains the adenovirus major late TATA box upstream of the hMTIIAϩ1. In vitro transcription assays were performed, and the results are displayed in Fig.  2A. We found that mutations of the hMTIIA start site at nucleotides ϩ2 and ϩ3 (from CC to TT) imparted Inr activity. This is consistent with previous findings that the sequence requirement for Inr activity resides primarily between Ϫ1 and ϩ3 with CATT being a much stronger Inr than CACC (18). If the lack of hMTIIA Inr activity is a result of a repressor acting on the hMTIIAϩ1, then one would predict that only hMTIIAϩ1mt2 would have lost its ability to bind this putative repressor. In contrast, if the lack of hMTIIA Inr activity is a result of the absence of an activator, then one would have to predict that hMTIIAϩ1mt2 and no other mutants gained the ability to bind this putative activator.
We reasoned that if the lack of an Inr activity from the hMTIIAϩ1 is due to the binding of a repressor, then the addition of DNA containing the hMTIIAϩ1 sequences into the in vitro transcription mixture should titrate out the repressor binding activity and restore transcription. In contrast, if the lack of an Inr activity from the hMTIIA start site is due to the lack of binding of an Inr activator, then the addition of DNA containing the hMTIIA start site sequences into the in vitro transcription reactions should have no effect on transcription activity. As shown in Fig. 2B, oligodeoxynucleotide containing hMTIIAϩ1 increased transcription initiation from a pMLTATA/hMTIIAϩ1 template (compare lane 1 with lanes  2-4). This activation of transcription is highly specific and requires specific hMTIIAϩ1 sequences, and as a mutated sequence (which presumably can no longer bind the putative repressor) cannot compete with the repression activity (Fig. 2B,  lane 5); sequences that cannot restore transcription activity ( Fig. 2A, lane 7, which presumably retains binding to the putative repressor) effectively competed with the repression activity (Fig. 2B, lane 6). Furthermore, excess hMTIIAϩ1 oligodeoxynucleotides have no effect on transcription of the P5 Inr (Fig. 2B, compare lanes 7 and 8). Taken together, this competition experiment suggests that the hMTIIAϩ1 sequence binds FIG. 1. In vitro transcription from synthetic minimal promoters. Plasmids were constructed to contain hMTIIAϩ1 sequences with or without upstream TATA box or Sp1 binding sites. In vitro transcription assays were performed with HeLa nuclear extracts, and RNA was analyzed by primer extension with a SP6 promoter primer (Promega). The expected 79-nucleotide extension product is indicated by an arrow. a transcriptional repressor. In addition, it confirmed that the lack of transcription activity from pMLTATA/hMTIIAϩ1 is not a result of suboptimal reaction conditions.
To distinguish whether an activator could bind the hMTIIAϩ1mt2 sequence, we performed additional in vitro transcription assays with the pMLTATA/hMTIIAϩ1mt2 template in the presence of hMTIIAϩ1mt2 or hMTIIAϩ1 oligodeoxynucleotide competitors. As shown in lanes 10 and 11 of Fig.  2B, a marked decrease in transcription was observed in the presence of excess hMTIIAϩ1mt2 but not hMTIIAϩ1 oligodeoxynucleotides. This observation, together with the findings that the hMTIIAϩ1 sequence might bind a transcriptional repressor, implies that the hMTIIA start site binds to a negative factor and mutation of hMTIIAϩ1 sequence into hMTIIAϩ1mt2 sequence resulted in a decrease of binding to a repressor and at the same time increased binding to an activator.
The hMTIIA Transcription Start Site Can Form Several Protein-DNA Complexes-To examine the cellular DNA-binding proteins that interact with the hMTIIAϩ1 sequence, a 22-base pair oligodeoxynucleotide containing this sequence was used to detect specific DNA-binding proteins by EMSA. As shown in Fig. 3, when analyzed with HeLa nuclear extract, five sequence-specific DNA-protein complexes were detected. The formations of these complexes were prevented by the addition of excess unlabeled hMTIIAϩ1 oligodeoxynucleotide (Fig. 3, lanes  3-5) but not by an unrelated oligodeoxynucleotide containing the binding site for the transcription factor AP1 (Fig. 3, lanes  9 -11). Interestingly, an oligodeoxynucleotide consisting of the P5 Inr sequence can also effectively inhibit formation of one of the complexes (complex III, Fig. 3, lanes 6 -8). This complex probably does not contain YY1 protein, because complexes generated with binding sites corresponding to the P5 Inr did not migrate identically, and antisera directed against YY1 did not affect this complex (data not shown).
Replication Protein A Binds the hMTIIAϩ1 DNA Sequence-To identify the protein(s) responsible for the formation of complex III, the DNA binding activity was purified from a HeLa nuclear extract as described under "Materials and Methods." After the final FPLC Mono Q column, fractions were assayed for DNA binding activity (Fig. 4A). The protein content of these fractions was also analyzed by Coomassie Blue staining after separation by denaturing gel electrophoresis (Fig. 4B). From these results it was apparent that the DNA binding activity had not been purified to homogeneity. However, two polypeptides with molecular masses of 70 and 32 kDa co-eluted with the DNA binding activity (Fig. 4). The peak fractions (16 -19) were pooled and concentrated prior to separation by denaturing gel electrophoresis. After the transfer of polypeptides to a polyvinylidene difluoride membrane, the portion of the membrane that contained the 70-kDa polypeptide was excised and subjected to digestion with trypsin. The resultant peptides were separated by reverse phase HPLC, and the amino acid sequences of two peptides were determined. A search of the data bases with the peptides, IGNPVPYNEGL-GQPQ and SGGVGGSNTNWK, revealed identical sequences within the 70-kDa subunit (RPA-1) of the trimeric human RPA (48).
To determine whether complex III indeed contains RPA, we performed supershift experiments. A binding reaction mixture containing 32 P-labeled hMTIIAϩ1 DNA and a HeLa nuclear extract was incubated with a monoclonal antibody to the 70-kDa RPA-1, a monoclonal antibody to the 32 kDa RPA subunit, EMSA was performed as described under "Materials and Methods." Probes used for the assay were 32 P-5Ј end-labeled double-stranded oligodeoxynucleotide 5Ј-CTGCACTCCACCACGCCTCCTC-3Ј and its complement. Molar excess of unlabeled competitor was added as indicated. Sequences for P5 Inr and AP1 competitor are given in Seto et al. (12). No protein was added in the reaction presented in lane 1. or a monoclonal antibody to an unrelated protein (transcription factor Sp1). As shown in Fig. 5A, complex III was removed (supershifted and masked by complex I) when reacted with an antibody to RPA (lanes 3 and 4) but not an antibody to Sp1 (lane 5). As a negative control, a protein-DNA complex formed by YY1 was not affected by the RPA-1 antibody (compare lane  6 with lanes 7 and 8). Furthermore, protein-DNA complex produced by the partially purified DNA binding activity from HeLa cells also can be supershifted by an RPA antibody ( lanes   11 and 12). Taken together, these data provide strong evidence that one of the hMTIIAϩ1 binding activities (complex III) contains RPA.
To further confirm that RPA binds the hMTIIAϩ1 sequence, we repeated the EMSA using a separate source of RPA proteins obtained from HeLa cells (kindly provided by Jerard Hurwitz, Sloan-Kettering Cancer Center). This protein was purified to at least 95% homogeneity judged by SDS/polyacrylamide gel electrophoresis and silver staining. As expected, purified RPA binds the hMTIIAϩ1 sequence (Fig. 5B, compare lanes 1 and  2); and based on competition assays, the binding is highly specific (lanes 3 and 4).
RPA Can Repress Transcription-To determine the effect of RPA binding to the hMTIIAϩ1 DNA, we performed in vitro transcription experiments with nuclear extracts derived from HeLa cells in the presence and the absence of partially purified hMTIIAϩ1 binding proteins (partially purified RPA). When 1.2 or 2.4 g of partially purified RPA was added to a reaction mixture containing the hMTIIA promoter with sequences from Ϫ286 to ϩ73, the level of correctly initiated transcripts decreased (maximal repression was approximately 20 fold: Fig.  6A, compare lane 3 with lanes 4 and 5).
To be certain that RPA was not a contaminate in the partially purified protein fractions and that the repression effect was not due to another protein, we cotransfected a plasmid expressing the 70-kDa RPA-1 and an hMTIIA promoter reporter plasmid into CV1 cells and assayed for CAT activities. As shown in Fig. 6B, cotransfection with 4Ј5CAT plus pCMV-RPA(ϩ) but not pCMV-RPA(Ϫ) caused a modest decrease in CAT expression. This low level of repression may be due to the fact that there could be other factors, such as other subunits of RPA or other hMTIIAϩ1 binding proteins, which are required in combination with the transfected RPA-1 in order to produce maximum repression. Also, the intracellular concentration of RPA-1 may be sufficiently high to negate the full repression effect of an additional source of protein supplied by transfection. Nevertheless, these data argue that RPA-1 binds the hMTIIAϩ1 sequence and represses transcription.
To be sure that the observed repression by RPA-1 is not a result of nonspecific inhibition from overexpression of the RPA-1 protein, we repeated the cotransfection experiments with three other unrelated promoters. As shown in Table I, overexpression of RPA-1 did not reduce activities from the Rous sarcoma virus, the simian virus 40, and the human immunodeficiency virus promoters. Taken together, our data suggest that overexpression of RPA-1 can repress transcription from the hMTIIA promoter and is not generally cytotoxic.
To address whether the observed repression by RPA-1 is due to its ability to bind near a promoter, we constructed RPA-1 chimeric proteins with an added DNA binding specificity. When Gal4-RPA1(ϩ) was cotransfected with the target plasmid pGal4TKCAT, a marked repression of CAT activity was observed (Fig. 6C, compare lanes 1 and 2). Neither a Gal4 DNA-binding domain alone or a nonfusion RPA-1 protein repressed transcription (compare lanes 1 and 3 and lanes 4 and  5). Also, repression was dependent on the presence of the Gal4-binding sites because CAT expression was not affected when TKCAT, lacking Gal4-binding sites, was used as a target (compare lanes 6 and 7).
Partially Purified RPA Fractions Bind Double-stranded but Not Single-stranded hMTIIAϩ1 DNA-Human RPA has high affinity for single-stranded DNA and low affinity for RNA and double-stranded DNA (39 -41, 63-65). When binding to singlestranded DNA, RPA interactions are partially sequence dependent with an RPA monomer occupying about 30 nucleotides. To address the possibility that RPA was a contaminant in FIG. 5. RPA binds the hMTIIA؉1 DNA sequence. A, recognition of one of the hMTIIAϩ1 binding protein complexes by two different anti-RPA monoclonal antibody. EMSA with HeLa nuclear extract (lanes 1-8) was performed as described in the legend to Fig. 3 with the addition of 1 g of the indicated antibodies (lanes 3-5, 7, and 8). Lanes 1 and 9 contained no protein extract. EMSA displayed in lanes 9 -12 was performed identically, with the exception that 0.4 g of the partially purified hMTIIAϩ1 binding protein (combination of fractions 16 -19 in Fig. 4) was used in place of the HeLa nuclear extract. B, EMSA with HeLa extract (lane 5) and with homogeneous RPA proteins (lanes [1][2][3][4] in the presence and the absence of specific (hMTIIAϩ1) or nonspecific (5Ј-GCAGGATGGAGAGGAGACGCATCACCTCCGCTGCTCG-CCG-3Ј and its complement) oligodeoxynucleotides. The probe used in this experiment is identical to the hMTIIAϩ1 probe in panel A. Lane 1 contained no protein extract. the partially purified DNA binding activities and the binding of RPA to the hMTIIAϩ1 oligodeoxynucleotides resulted from a small percentage of single-stranded DNA in the reaction, we performed EMSA with excess single-stranded or doublestranded hMTIIAϩ1 DNA. As shown in Fig. 7, neither the hMTIIAϩ1 (ϩ) or (Ϫ) strand DNA affected complex III (com-pare lane 1 with lanes 2-5); but hMTIIAϩ1 double-stranded DNA effectively competed for the protein binding (lanes 6 and 7). Because purified recombinant RPA binds single-stranded hMTIIAϩ1 DNA effectively (data not shown), this result suggests that complex III may contain a modified form of RPA that recognizes double-stranded hMTIIAϩ1 sequence with high specificity. Alternatively, other proteins within complex III may contribute to its preference for specific double-stranded DNA binding activity. DISCUSSION We originally set out to test the hypothesis that the hMTIIAϩ1 sequence functions as a typical transcription Inr and to identify cellular factors that activate transcription of the hMTIIA promoter by binding to the site of initiation. Instead, we found that the hMTIIAϩ1 sequence binds proteins that mediate transcriptional repression. Five specific protein-DNA  complexes are formed when HeLa nuclear extract is incubated with the hMTIIAϩ1 sequence. In this report, we describe the partial purification of the proteins responsible for one of these complexes, complex III. Polypeptides of 70 and 32 kDa coeluted with complex III formation. Amino acid sequencing of peptides from the 70-kDa protein identified this polypeptide as the 70-kDa subunit of replication protein A (41) (also known as RFA (39) or SSB (40,65)). RPA is a trimeric DNA-binding protein with molecular weights of 70, 32, and 13 kDa that is evolutionarily conserved from yeast to human (67,68). Based on the amino acid sequence identity and the molecular masses of the polypeptides that co-elute with complex III formation, it appears that at least two subunits of replication factor A are present in the complex that binds specifically to the hMTIIAϩ1 double-stranded DNA sequence. RPA was initially identified as an essential factor for the in vitro replication of plasmid molecules containing the SV40 origin (39 -41). Subsequently, it has been shown that RPA is a single-stranded DNA-binding protein, which interacts with DNA helicases such as SV40 T antigen and DNA polymerases (69). The observation that the Saccharomyces cerevisiae genes encoding the three RPA subunits are all essential is consistent with the predicted requirement for this complex in chromosomal replication (42,68). More recently, RPA has been shown to be required for the incision step of DNA nucleotide excision repair (46), to be involved in genetic recombination (42,43), and to bind to transcriptional activators such as Gal4, VP16, and p53 (70 -72).
Our finding that the multifunctional RPA also binds specifically to the transcription initiation site of the hMTIIA promoter and may be involved in transcriptional repression is both exciting and intriguing. While our work was in progress, Singh and Samson (73) reported that S. cerevisiae RPA binds specifically to two URS elements in the 3-methyladenine DNA glycosylase gene. Because similar sequences that function either as URS or UAS elements have been identified in the promoters of at least 11 genes that are involved in DNA metabolism, in particular DNA repair, it suggests that RPA may regulate a group of genes that are involved in the cellular response to DNA damage. Interestingly, expression of the human metallothionein gene is also induced following DNA damage or treatment with heavy metals, hormones, and other agents (74 -83). Thus, a similar regulatory pathway involving RPA may exist in mammalian cells.
How might RPA complex bind to the hMTIIAϩ1 sequence and repress transcription? First, just as the SV40 T antigen binding to the SV40 early promoter inhibits binding of transcription initiation complex to DNA (84 -86), RPA or proteins associated with RPA may bind to the hMTIIAϩ1 sequence and prevent the binding of an Inr-binding protein that activates transcription. Second, it is possible that RPA may block the activation of an Inr-binding protein. For example, there may be activators that bind the hMTIIAϩ1 sequence simultaneously with the RPA complex and are inactivated by formation of a heterodimer with one of the RPA subunits. Finally, RPA may be a genuine transcription repressor composed of repression domains. Currently, repressing domains that have been identified in other proteins are commonly rich in alanine, glutamine, and/or proline, but no clear amino acid sequence similarities exist between the different repressing domains (reviewed in Ref. [87][88][89]. A causal inspection of the RPA protein sequences has failed to reveal any obvious resemblance to well characterized transcriptional repression motifs or to repression domains present in other transcription factors.
In addition to the complex that contains RPA, we have identified four other specific complexes that bind the hMTIIAϩ1 sequence. Currently, we do not know whether these other proteins are activators or repressors or whether they are important for the regulation of hMTIIA transcription. Our results show that one of the purified complex represses transcription but does not exclude the possibility that other complexes may actually activate transcription. Preliminary in vitro transcription assays with oligodeoxynucleotide competitors indeed suggest the participation of an activator. Further protein purification studies as well as in vitro transcription experiments should resolve these issues.
As we mentioned before, most mammalian and viral promoter transcription start sites contain Inrs that activate transcription. Here, we show that the hMTIIAϩ1 sequence not only failed to activate transcription but actually is regulated by a cellular negative factor. This repression phenomenon at the transcription start site is not unprecedented; Wiley et al. (62) described an activity (IBPs) that binds to the transcriptional initiation site of the SV40 major late promoter and represses transcription. Whether this type of regulation is only limited to a small set of promoters or whether it is a more widespread phenomenon remains to be determined. FIG. 7. EMSA and competition experiment of binding to oligodeoxynucleotides containing hMTIIA؉1. 32 P-Labeled hMTIIAϩ1 double-stranded oligodeoxynucleotides were used as substrates for binding with HeLa nuclear extract. Single-stranded DNA competitor in lanes 2 and 3 is an oligodeoxynucleotide containing the (ϩ) strand DNA of hMTIIAϩ1. Single-stranded DNA competitor in lanes 4 and 5 is an oligodeoxynucleotide containing the (Ϫ) strand DNA of hMTIIAϩ1. Double-stranded DNA competitor in lanes 6 and 7 is an oligodeoxynucleotide containing both strands of DNA of hMTIIAϩ1. The numbers above the lanes indicate the molar excess of unlabeled DNAs.