Purification and Characterization of OF-1, a Host Factor Implicated in Herpes Simplex Replication*

A human cellular factor (OF-1) has been previously implicated in replication of herpes simplex virus, type 1. This protein binds to three conserved regions (Boxes I, II, and III) in the viral replication origin and appears to be required for viral DNA synthesis (Dabrowski, C. C., Carmillo, P. J., and Schaffer, P. A. (1994) Mol. Cell. Biol. 14, 2545–2555). In the present study, we have partially purified and characterized OF-1 from human cells. This protein appears to consist of a tetramer composed of two heterodimers with subunits of 73 and 90 kDa. The smaller subunit contains the DNA binding activity. We have investigated the binding specificity of OF-1 using a mobility shift analysis. These studies reveal that binding is specific for both duplex and single-stranded Box I sequences and that the strongest preference is for the bottom strand of Box I. We present evidence suggesting that the binding of OF-1 to Box I DNA is enhanced in the presence of the herpes simplex-encoded UL9 protein, which also binds to Box I in oriS and is required for viral replication. Implications of these findings for the initiation step in viral replication are discussed.

Initiation of DNA replication in eukaryotic cells occurs within localized chromosomal regions and involves binding by a series of origin-binding proteins. Although initiation occurs at specific DNA sequences in yeast (Saccharomyces cerevisiae) (1), the corresponding process in higher eukaryotes is less well understood (2), and only recently has evidence for the existence of specific origin sites in mammalian cells been presented (3). Hence, studies of human DNA viruses, such as herpes simplex virus, type 1 (HSV-1), 1 have been useful as models for understanding the role of origin recognition sequences in the initiation of DNA synthesis in human cells.
HSV-1 replication (reviewed in Refs. 4 and 5) initiates at one of three viral replication origins, called oriL and oriS (present in two copies). This process requires the viral initiator protein, UL9, which binds the origins in a sequence-specific manner. The minimal sequence necessary for origin function contains three homologous sequences, designated Boxes I, II, and III (see Fig. 1). Dimers of UL9 bind specifically to Boxes I and II and are thought to form a nucleoprotein complex by proteinprotein interactions between the dimers bound at both sites. This structure presumably promotes origin unwinding utilizing the ATP-dependent DNA helicase activity associated with UL9 (6,7). The unwound origin then facilitates the recruitment of additional replication proteins to initiate replication.
Viral replication also appears to require host factors, since attempts to establish replication of oriS-containing plasmids in vitro with purified viral replication proteins have been unsuccessful (discussed in Ref. 4). A recently identified human host protein, OF-1 (8), appears to be one of these required factors. This possibility is based on observations that OF-1 specifically binds the viral replication origins at the Box I, II, and III sites and that plasmids carrying oriS mutations that inhibit OF-1 binding, but not UL9 binding, fail to replicate efficiently in virally infected cells. Hence, OF-1 likely plays a role in viral replication in vivo.
The binding properties of OF-1 implicate this factor during the initiation step of replication. OF-1 specifically binds within oriS (8) to sequences that overlap the sites bound by UL9. The finding that OF-1 interacts with multiple origin sites suggests that OF-1, like UL9, may also participate in a nucleoprotein complex at the origin. However, although both OF-1 and UL9 bind with the highest affinity to the Box I site, their relative affinity for the other boxes differs significantly (4,8). Binding by OF-1 shows decreasing relative affinities of Box I Ͼ Box III Ͼ Box II, whereas binding by UL9 shows affinities of Box I Ͼ Box II Ͼ Ͼ Box III. Thus, the complexes formed by UL9 and OF-1 are likely to differ and may play distinct roles during viral replication.
We describe the first partial purification and physical characterization of OF-1 from human cells. We have characterized the binding specificity of this protein, and we find that it binds to the Box I site in oriS in a manner that involves specific interactions with one of the two DNA strands. This binding appears to be enhanced in the presence of UL9. Our results support the possibility that OF-1 participates in the initiation step of viral replication.
Purification of OF-1-Throughout the purification, OF-1 activity was monitored by a mobility shift assay showing specific binding to a duplex Box I oligomer (see "Mobility Shift Assays in Non-denaturing Gels"), and protein concentrations were determined by the method of Bradford (11). Where indicated, the species of protein present in a given fraction were visualized by electrophoresis on 6% polyacrylamide gels containing SDS, followed by silver staining (9). Gels were scanned by a Hewlett-Packard optical scanner to create a digitized image. Where required, the fraction of protein present in OF-1-specific bands was determined relative to the sum of the intensities of all bands within the lane, using the NIH Image program. Ammonium Sulfate Precipitation-Suspension-grown HeLa S3 cells (National Cell Culture Center) were obtained as cell pellets that had been washed twice in 2.8 mM NaH 2 PO 4 , 13.6 mM Na 2 HPO 4 , 145 mM NaCl. Pellets totaling 2.3 ϫ 10 9 cells were resuspended in 10 ml of Buffer A, containing 50 mM Tris-HCl (pH 7.8), 100 mM NaCl, 1 mM EDTA, 0.5 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 2 g/ml pepstatin A, 2 g/ml leupeptin, and 2 g/ml antipain. Cell suspensions were frozen (Ϫ80°C) and thawed 3 times. Cell lysates were clarified by centrifugation at 10,000 ϫ g in a Sorvall SS-34 rotor for 10 min at 4°C. The supernatant was collected and the debris pellet extracted 3 times by resuspension in 2 ml of Buffer A followed by centrifugation, as above. The four supernatant fractions were then pooled, mixed with (NH 4 ) 2 SO 4 to a final concentration of 20% (w/v), held on ice for 15 min, and centrifuged as above. The supernatant was collected. To recover additional OF-1, the pellet was re-extracted three times with 1 ml of Buffer B, containing 20 mM HEPES (pH 7.6), 0.5 mM DTT, 50 mM NaCl, 1 mM EDTA, 20% (v/v) glycerol, and supplemented with 20% (w/v) (NH 4 ) 2 SO 4 . Each extraction was followed by centrifugation. The four supernatants were then combined and diluted with an equal volume of Buffer B to bring the final concentration of (NH 4 ) 2 SO 4 to 10%.
Chromatography-The protein suspension was then loaded onto a 150-ml phenyl-Sepharose (Amersham Pharmacia Biotech) column equilibrated in Buffer B containing 10% (NH 4 ) 2 SO 4 . The column was washed with 600 ml of equilibration buffer and eluted with a decreasing linear gradient of (NH 4 ) 2 SO 4 (10 to 0%) in Buffer B. OF-1 eluted between 7 and 3% (NH 4 ) 2 SO 4 . Fractions containing peak OF-1 activity were pooled and dialyzed against Buffer B. The dialyzed protein was loaded onto a 50-ml phosphocellulose column that had been equilibrated with Buffer B. The column was washed with 150 ml of Buffer B and eluted with a linear gradient from 50 to 700 mM NaCl in Buffer B. OF-1 eluted between 150 and 500 mM NaCl. Fractions containing peak OF-1 activity were pooled and stored at Ϫ80°C. Pooled fractions were thawed, diluted in Buffer B to bring the salt concentration to 300 mM NaCl, and separated into two equal portions. Each portion was loaded onto a 3-ml heparin-Sepharose (Amersham Pharmacia Biotech) column equilibrated in Buffer B containing 325 mM NaCl. The columns were washed with 15 ml of the equilibration buffer and eluted with a linear gradient from 325 to 1000 mM NaCl in Buffer B. OF-1 eluted between 375 and 600 mM NaCl. Fractions containing peak OF-1 activity were pooled, concentrated, de-salted in Centricon-100 spin cartridges, and stored in Buffer B at Ϫ80°C.
Mobility Shift Assays in Non-denaturing Gels-Reaction mixtures (10 l) contained heparin-Sepharose-purified OF-1 protein (0.175 g) and 32 P-labeled oligomers at the indicated concentrations in DNAbinding buffer (50 mM HEPES (pH 7.5), 0.1 mM EDTA, 0.5 mM DTT, 100 mM NaCl, 10% (v/v) glycerol). Reactions were incubated on ice for 30 min. When included, competitor DNA was added after the first 20 min, and incubation was then continued for the remaining 10 min. Bromphenol blue gel loading buffer type I (9) was added, and samples were subjected to non-denaturing electrophoresis at 4°C on 6% polyacrylamide gels (Hoeffer/ Amersham Pharmacia Biotech) in 50 mM Tris (pH 8.3), 45 mM boric acid, and 50 mM EDTA. Gels were scanned on a Molecular Dynamics PhosphorImager and the bands quantified using the program ImageQuant (Molecular Dynamics).
Two-dimensional Electrophoresis-A mobility shift assay was performed as described above using heparin-Sepharose-purified OF-1 (3 g) and a 32 P-labeled duplex Box I oligomer. The radioactive band containing OF-1-DNA complexes was visualized by scanning on the PhosphorImager. A gel slice containing the OF-1 complex was then excised from the gel and soaked briefly in SDS-gel running buffer (9). The gel slice was then placed into the well of a denaturing 6% polyacrylamide gel containing SDS and subjected to electrophoresis. Molecular weight markers were run in the same gel, and protein bands were visualized by staining with silver salts (9).
Denaturing Gel Electrophoresis of Cross-linked Protein-DNA Complexes-Reaction mixtures (10 l) contained 32 P-labeled oligomer and OF-1 and/or UL9 in DNA-binding buffer. In these reactions, the total volume of UL9 plus UL9 storage buffer remained constant. Reactions were incubated 30 min on ice and then cross-linked by exposure to ultraviolet light (254 nm) for 15 min at an intensity of 48 milliwatts/ cm 2 . SDS-loading buffer (9) was added; the samples were boiled for 3 min, and the reactions were subjected to electrophoresis on 6% polyacrylamide gels containing SDS (9). Radioactive bands were visualized by scanning on a PhosphorImager. Pre-stained protein molecular weight markers were run on each gel, and their positions were used to determine the molecular weights of protein-DNA complexes. Where indicated, the 75-kDa OF-1-associated band was quantitated using the program ImageQuant, after adjusting the contrast to allow resolution of individual bands.
Gel Filtration Chromatography-Protein molecular weight markers (Sigma) of sizes 650 (thyroglobulin), 443 (apoferritin), 200 (␤-amylase), 150 (alcohol dehydrogenase), and 66 kDa (bovine serum albumin) were applied to an 18-ml Sepharose CL-4B (Amersham Pharmacia Biotech) column (height 22 cm, diameter 1 cm) equilibrated in Buffer B. Proteins were detected during elution by absorbance at 280 nm using a UVvisible detector (ISCO), and a standard plot of log molecular weight versus fraction number was generated from the elution pattern. OF-1 (20 g) was cross-linked for 30 min to 625 pmol (3.1 M) of 32 P-labeled duplex Box I DNA in Buffer B, as described above. The cross-linked material was then applied to the same column, and the elution positions of OF-1-DNA complexes were detected by scintillation counting of the resulting fractions.
ATPase Assays-The DNA-dependent formation of inorganic phosphate was determined as described (12). Briefly, 50-l reactions containing 40 mM HEPES buffer (pH 8.3), 4 mM ATP, 1 mM DTT, 5 mM MgCl 2 , 10% (v/v) glycerol, 100 g/ml bovine serum albumin, 6 nM UL9, varying amounts of Box I DNA, and OF-1 or OF-1 buffer were incubated for 1 h at 37°C, treated with acidic molybdenum/malachite green solution and sodium citrate to develop color, and subjected to measurement of absorbance at 650 nm. A reaction without DNA was used as a blank.
Calculation of Binding and Kinetic Parameters-Amounts of bound and unbound oligomer from electrophoretic analyses on non-denaturing (mobility shift assays) and denaturing (cross-linking assays) gels were determined using ImageQuant. Where more than one bound complex was observed (e.g. in cross-linking analyses), the intensities of the bound complexes were combined. As appropriate, lines of best fit were applied to the data points, using the curve-fit function of Kaleidograph (Synergy Software). Where indicated, Scatchard plots were produced, and K d or K d(app) values were calculated from the negative reciprocals of the slopes of the lines. For analysis of the kinetic data from the ATPase assay, the data were fit to a sigmoidal curve derived from a rearrangement of the Hill equation (13). The equation used was v ϭ (V max [S] n )/

FIG. 1. Sequences of the HSV-1 replication origin (oriS) and of Box I and Box III oligonucleotides used in mobility shift and cross-linking assays.
Upper panel shows the oriS sequence and depicts the conserved Boxes I, II, and III (underlined) with their relative orientations (indicated by arrows). Lower panel shows top and bottom oligomer strands (designated with respect to their relative orientation to oriS). The Random Box I oligomers have the same base composition as the Alternate Box I oligomers but are of randomized sequence. Alternate Box I contains one more base and is shifted to the left relative to Box I.
where v ϭ reaction velocity, V max ϭ maximum reaction velocity, K 0.5 ϭ substrate concentration at one-half V max , [S] ϭ substrate concentration, and n ϭ the Hill coefficient, a measure of the cooperativity of the system.

RESULTS
Purification of OF-1-To gain insight into the role of host factor OF-1 in HSV-1 replication, we have purified this factor from human cells and characterized it with respect to polypeptide composition, DNA binding specificity, and its potential interaction with the HSV-1 origin-binding protein, UL9.
OF-1 activity was isolated from HeLa cell extracts as shown in Table I and was purified 3400-fold. It was detected by its ability to produce a mobility shift during electrophoresis of a duplex Box I oligomer. This activity remained in the supernatant after ammonium sulfate (20% w/v) precipitation of crude extracts. Further purification was achieved by sequential chromatography on phenyl-Sepharose, phosphocellulose, and heparin-Sepharose, as shown in Fig. 2, panel A. We noticed that early steps in the purification resulted in an increase in total activity (see Table I), most likely due to removal of contaminating nucleases. Heparin-Sepharose chromatography resulted in substantial loss of activity but removed several major contaminating polypeptides. Since only a single peak of activity was recovered from this purification step (see Fig. 2, panel A), and since we observed only a single band in our mobility shift assay throughout the purification (not shown), we believe that OF-1 represents a single DNA binding activity.
The protein species present after each purification step were analyzed by denaturing gel electrophoresis, as shown in Fig. 2, panel B. Four major polypeptide species predominated after heparin-Sepharose purification (Fig. 2, panel B, lane 5). As discussed below, OF-1 consists of two of these polypeptides (73 and 90 kDa), designated by arrows in Fig. 2, panel B. Based on densitometry scans of these protein bands, we estimate that the OF-1 polypeptides comprised 28% of the total protein in the heparin-purified sample.
OF-1 Contains Two Subunits and May Exist as a Tetramer-To determine which of the peptides in our OF-1 preparation are involved in binding to HSV-1 origin (Box I) sequences, we employed a modified two-dimensional electrophoresis protocol. OF-1 protein was first run on a non-denaturing gel in the presence of labeled duplex Box I oligomer under mobility shift conditions. Polypeptides associated with the shifted band were then separated by denaturing gel electrophoresis. As shown in Fig. 3, this analysis resulted in recovery of two polypeptides of 73 and 90 kDa. These peptides correspond to two of the four species seen in the purified OF-1 preparation (arrows in Fig. 2, panel B). Densitometry of the OF-1-associated bands showed that the two polypeptides are present in approximately equimolar amounts, suggesting that OF-1 exists as a heterodimer.
Further insight into the subunit structure of OF-1 was obtained using gel filtration chromatography. A 32 P-labeled duplex Box I DNA oligomer was chromatographed on Sepharose CL-4B either alone or after being cross-linked to OF-1. As shown in Fig. 4, a comparison of these two chromatographs revealed a large peak from both columns representing the free DNA and two additional smaller peaks from the column loaded with the cross-linked material. These smaller peaks presumably represent OF-1-DNA complexes and exhibit apparent mo-TABLE I Purification of OF-1 OF-1 was purified by ammonium sulfate precipitation and chromatography on the indicated matrices, as described under "Experimental Procedures." After each step, peak OF-1 fractions were pooled and assayed for protein concentration and specific activity. To measure specific activity (nmol of DNA bound/mg of protein), protein was diluted and assayed for a mobility shift using a 32 P-labeled duplex Box I oligomer (12.5 nM), and the fraction of shifted DNA was determined from scans of the gels. For each measurement, several protein dilutions were used, and specific activity was determined with those dilutions that showed a linear increase in binding proportional to concentration. Since nucleases may have been present during purification and since the DNA concentrations used did not exceed the subsequently determined K d value for binding, the specific activity determined at each step is a minimal estimate of activity. The relative intensities of gel mobility-shifted products from assays performed on column fractions are shown. The gel shift autoradiograms for the peak fractions are indicated above the corresponding plots. Panel B, aliquots (3 g) from purification steps shown in Table I were run on denaturing polyacrylamide gels and silver-stained, as described under "Experimental Procedures." Lane 1, crude cell extract; lane 2, pooled supernatant fraction from ammonium sulfate precipitation; lane 3, pooled peak fractions from phenyl-Sepharose chromatography; lane 4, pooled peak fractions from phosphocellulose chromatography; lane 5, pooled peak fractions from heparin-Sepharose chromatography. Molecular masses in kDa from standards are shown at the left. Arrows at the right indicate OF-1 subunits, determined as described in the text. lecular masses of 165 (fraction 30) and 395 (fraction 27) kDa. The 165-kDa peak is consistent with a complex containing a protein heterodimer with both the 73-and 90-kDa subunits plus DNA (13 kDa). The 395-kDa peak may represent a tetramer consisting of two heterodimers plus one or more copies of Box I DNA. Hence, we conclude that OF-1 most likely exists as a tetramer containing two heterodimers.
OF-1 Binds Specifically to Both Duplex and Single-stranded Box I Sequences-Dabrowski et al. (8) previously showed that OF-1 from crude extracts binds specifically to the Box I region of oriS. Consequently, we wished to know whether our partially purified preparation of OF-1 retained this binding specificity and to determine the substrate requirements for binding. In particular, since replication origins must unwind to initiate replication, we considered the possibility that OF-1 might bind to single-stranded DNA sequences within oriS.
For this analysis, we conducted a series of binding experiments with 32 P-labeled single-stranded or duplex Box I oligomers. To demonstrate specificity of binding, excess unlabeled competing DNA was added, as indicated. This competitor DNA consisted of single-stranded or duplex sequences corresponding to either Box I, a randomized sequence with an overall base composition similar to Box I (Random Box I), or poly(dC)-oli-go(dG). Reaction products were then analyzed by non-denaturing polyacrylamide gel electrophoresis and autoradiography.
Consistent with the previous results of Dabrowski et al. (8), we find that OF-1 binds specifically to duplex Box I DNA. As shown in Fig. 5, OF-1 bound to the labeled Box I probe yielding a single-shifted band, in the absence of competing DNA (lane 1). Specificity of binding was revealed in competition experiments that showed no shift when the competing DNA was the same duplex Box I sequence (lane 2) but a complete shift when the randomized duplex oligo (lane 3) or poly(dC)-oligo(dG) homopolymer (lane 4) was used.
Our results also indicate that OF-1 binds specifically to single-stranded Box I sequences. For this analysis, we conducted similar mobility shift experiments to those described above, except that the labeled probe consisted of the bottom strand of Box I, and the competing DNA was also single-stranded. As shown in Fig. 5, in the absence of competition, OF-1 bound to the single-stranded probe producing a mobility shift (lane 5). This binding is specific, since it was eliminated by an excess of unlabeled Box I bottom strand (lane 6) but not by an excess of the randomized duplex oligomer which had been denatured prior to addition to the reaction (lane 7).
A similar mobility shift analysis with a related Box I sequence revealed that residues at the 3Ј end of Box I are critical for binding. This related sequence (called Alternate Box I in Fig.  1) is shifted to the left relative to the original Box I sequence and, hence, lacks three nucleotides from the right end (AAT).  3. Composition of OF-1. Heparin-purified OF-1 protein (3 g) was subjected to a modified two-dimensional gel electrophoresis protocol, as described under "Experimental Procedures." Arrows indicate the positions of the two major peptides recovered and correspond to molecular masses of 73 (right) and 90 (left) kDa, based on standard markers (kDa) that were run on the same gel. Stained material (to right of arrows) is an artifact that appeared across the gel, even in the absence of protein, and may be due to formation of an SDS-complex (60 -70 kDa) described previously (30). The inset below the gel depicts an intensity scan (using NIH Image to create a graphical plot) of the region of the stained gel containing the OF-1-associated peptides.

FIG. 4. Gel filtration chromatography of OF-1.
Heparin-purified OF-1 protein that had been cross-linked to radioactive duplex Box I DNA (छ) or unbound DNA (q) was applied to a Sepharose CL-4B column, and the radioactivity of the column fractions was measured, as described under "Experimental Procedures." Arrows indicate molecular masses (in kDa) of protein standards run on the same column. This result is significant since Dabrowski et al. (8) showed that OF-1 from crude extracts fails to bind an oligomer carrying deletions or mutations in the residues CAAT at the 3Ј end of the Box I sequence. They further showed that base substitutions at the 3Ј CA residues eliminated replication of oriScontaining plasmids transfected into HSV-1-infected cells. Our finding indicates that our purified OF-1 retains a binding requirement for the 3Ј AAT residues in this sequence and further suggests that these residues may be critical for OF-1 function during viral replication.

OF-1 Binds Box I Sequences Preferentially Over Box III Sequences and Shows the Highest Overall Affinity for the Box I
Bottom Strand-To define and quantitate the binding preferences of OF-1, we conducted the experiments shown in Fig. 6. Mobility shift assays similar to those described above were performed, and dissociation constants (K d values) were calculated from Scatchard plots of the data. The plots obtained were biphasic (except for panel F), suggesting that OF-1 contains two sites with different affinities for DNA. Since the K d value for the higher affinity site in each case is substantially lower than that for the second site, we have reported only the higher affinity values.
First, we examined the preference of OF-1 for Box I and III duplex sequences, since Dabrowski et al. (8) had previously shown that OF-1 from crude extracts binds preferentially to Box I duplexes. Our results confirm this finding using partially purified OF-1 and indicate that OF-1 has a 3.2-fold higher affinity for duplex Box I (panel A) than for duplex Box III (panel D) (K d ϭ 24 nM and 78 nM, respectively).
Next, we compared the binding affinities of OF-1 for duplex and single-stranded Box I DNAs. This comparison reveals that OF-1 binds with the strongest affinity to the bottom strand of Box I. The K d value for binding to the Box I bottom strand was 10 nM (panel C) and was 2.4-fold lower than that seen for duplex Box I (K d ϭ 24 nM, panel A) and 6.2-fold lower than seen for the top strand of Box I (K d ϭ 62 nM, panel B). Hence, we suspect that OF-1 may prefer a single-stranded Box I site for binding in vivo.
Finally, we observed that OF-1 also binds to single-stranded Box III sequences but greatly prefers the top strand. In this case, we obtained K d values of 61 nM for the Box III top strand (panel E) and 780 nM for the bottom strand (panel F). Since Box I and Box III are inverted with respect to each other in oriS (see Fig. 1), OF-1 shows a preference for the equivalent strand in each case. The affinity of OF-1 for the preferred Box III strand is higher than that seen for the Box III duplex DNA, although the preference in this case is not as striking as that seen with Box I DNA.
The 73-kDa Subunit of OF-1 Is the DNA Binding Subunit-To identify the polypeptide(s) in OF-1 that directly contact DNA in OF-1-DNA complexes, we cross-linked OF-1 to 32 P-labeled duplex Box I or Box III oligomers, boiled the complexes in SDS to denature the protein and DNA, and analyzed the products by SDS-polyacrylamide gel electrophoresis and autoradiography, as shown in Fig. 7, panel A. OF-1 formed three major bands following cross-linking to either Box I or Box III DNA (lanes 2 and 7, respectively). The most intense band (lowest arrow) had an apparent molecular mass of 75 kDa, suggesting that it represents a complex between the smaller OF-1 subunit (73 kDa) and a single DNA strand (6.6 kDa). Most likely, this DNA strand is the bottom strand in the case of Box I and the top strand in the case of Box III, since OF-1 preferentially interacts with these specific strands. A second band migrated with an apparent molecular mass of 82 kDa, consistent with a complex between the smaller OF-1 subunit and two DNA strands. The third band was broad with a maximum molecular mass of 168 kDa. We suggest that this band consists of a complex between DNA (1 or 2 strands) and both the smaller (73 kDa) and larger (90 kDa) OF-1 subunits. Taken together, these results suggest that the smaller subunit binds DNA, whereas the larger subunit makes only incidental contact with the DNA through interaction with this smaller subunit.
Effects of UL9 Protein on DNA Binding by OF-1-Since the OF-1-binding site in Box I overlaps that of the viral UL9 protein (8,14), we suspected that these proteins might interact at this site. In contrast, since UL9 binds poorly to Box III DNA (14), we did not predict an interaction at the Box III site. To test these possibilities, we measured the binding of OF-1 to Box I and Box III oligomers in the presence of UL9, using the crosslinking procedure described above.
As shown in Fig. 7, panel A, the OF-1 preparation forms cross-linked complexes with Box I in both the absence and presence of UL9 (compare lane 2 with lanes 3-5). The mobilities of these bands are the same in the absence and presence of UL9, suggesting that the same complexes are being formed in both cases. However, the intensities of the bands increase as the concentration of UL9 increases. We have quantitated these increases (as shown in Fig. 7, panel B) by determining the relative intensities of the major (75 kDa) OF-1-associated band  11-167 nM). Oligomer sequences and strand designations are shown in Fig. 1. The reaction mixtures were run on native polyacrylamide gels and analyzed as described under "Experimental Procedures." The concentration of bound DNA for each reaction was determined, and the K d values were obtained from Scatchard plots of the data.
in the absence and presence of UL9. This comparison reveals increases of up to 12.5-fold at the highest UL9 concentration used. In addition, we have determined apparent K d values for the formation of these complexes, as shown in Fig. 8. When the concentration of OF-1 was held constant and the Box I concentration was varied, K d(app) values of 42 and 7 nM were obtained in the absence and presence of UL9, respectively. Hence, the relative affinity of the OF-1-associated proteins for Box I DNA is increased 6-fold in the presence of UL9.
We further observed that UL9 has no significant effect on the formation of cross-linked complexes between the OF-1 prepa-ration and Box III DNA (Fig. 7, panel A, lanes 8 -10, and Fig. 7,  panel B). Since UL9 binds poorly to Box III, its ability to facilitate binding of OF-1 to a particular DNA sequence appears to correlate with its affinity for that DNA.
Effect of OF-1 on UL9 Binding to Box I DNA-As shown in Fig. 7, panel A, our cross-linking experiments are consistent with previous findings that UL9 binds strongly to Box I and poorly to Box III sequences (14). With Box I DNA (lane 1), we detected two UL9-DNA complexes (95 and 208 kDa), consistent with complexes containing a UL9 monomer (94 kDa) and dimer, respectively. We also observed that the vast majority of the Box I DNA was apparently trapped in higher order UL9 complexes and did not enter the gel, as previously reported (15). In contrast, few or no UL9-cross-linked products were observed when the Box III probe was used (lane 6). In the presence of our OF-1 preparation, we observed a large reduction in UL9-DNA complexes (especially those Box I complexes trapped in the wells), and we failed to observe any supershifted OF-1-associated bands indicative of a UL9-OF-1-DNA complex. These findings suggest that OF-1 may compete with UL9 for Box I binding.
Effect of OF-1 on the ATPase Activity of UL9 -Our results in Fig. 7 suggested that OF-1 might compete with UL9 for binding to Box I DNA. Hence, to test the possibility that OF-1 might also interfere with UL9 enzymatic activities that require DNA binding, we assayed UL9 ATPase activity in the presence and absence of OF-1. This activity is specifically and potently stimulated by a DNA duplex containing Box I (12). As shown in Fig.  9, the presence of a 5-fold molar excess of OF-1 substantially reduced UL9 ATPase activity at low DNA concentrations, resulting in a 3.9-fold increase in the K 0.5 over the value in the absence of OF-1. In contrast, relative V max values in the presence and absence of OF-1 were similar (0.81 and 0.88, respectively). Hence, our OF-1 preparation appears to affect the affinity of UL9 for Box I DNA without altering the catalytic activity of UL9.  5 and 10), based on our determinations of protein purity in each preparation and the assumption that OF-1 exists as a heterodimer. Reactions were cross-linked and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography, as described under "Experimental Procedures." Since the specific radioactivities of the Box I and Box III DNAs differed and since the Box I and Box III experiments were conducted independently, the intensities of the Box I-associated bands cannot be compared directly with the Box III-associated bands. Molecular weights of each band were determined using standard protein markers run on the same gel (not shown).

DISCUSSION
This study examines the properties of OF-1, a human protein that has been implicated in the initiation of replication by HSV-1. Our results support a role for OF-1 in viral replication and suggest that OF-1 might participate in a nucleoprotein complex at the viral origin.
DNA Binding Specificity of OF-1 and Formation of Origin Complexes- Dabrowski et al. (8) previously showed that human cell extracts contain a factor, OF-1, that binds to Box I, II, and III regions within oriS and shows the strongest preference for Box I. We have partially purified this factor and have characterized its properties. Our results indicate that OF-1 binds specifically to duplex Box I, thus confirming previous studies using crude extracts (8). We also show that OF-1 exhibits a 3-fold lower affinity for Box III than for Box I.
A novel finding from our studies is that OF-1 binds specifically to single-stranded Box I DNA and shows a high affinity for a Box III single strand. Binding is significantly stronger for one of the two strands at each site (the bottom strand of Box I and the top strand of Box III). Since these strands contain functionally equivalent sequences, due to inversion of the Box I and III sequences within oriS, the binding to OF-1 in both cases may involve similar protein-DNA interactions. Furthermore, since OF-1 strongly prefers the single-stranded form of Box I over the duplex form of Box I, it seems likely that some of the OF-1 interactions with oriS in vivo involve single-stranded DNA. Although several eukaryotic proteins are known that bind single-stranded DNA at specific origin sequences (16 -20), the target sequences of these proteins show no apparent correlation with the OF-1-binding sites.
The sequence specificity of OF-1 binding was demonstrated by binding studies with various single-stranded and duplex competitor DNAs. These experiments also rule out the possibility that OF-1 binds nonspecifically to generic DNA structures such as duplex DNA ends or single strands. In addition, our binding affinity assays show a significant decrease in the affinity of OF-1 for Box III relative to Box I, demonstrating that binding affinity is sequence-dependent. It is possible, however, that the observed specificity is partially due to a preference of OF-1 for a secondary structure stabilized by the sequence of Box I.
The Quaternary Structure of OF-1 and Identification of the DNA Binding Subunit-OF-1 appears to be a tetramer containing two identical heterodimers. Evidence for a heterodimeric structure is provided by a two-dimensional electrophoresis procedure that identified two OF-1-associated polypeptides (73 and 90 kDa) present in approximately equal molar amounts. Additional evidence for a heterodimeric structure is provided by determination of the sizes of cross-linked OF-1-DNA complexes by denaturing gel electrophoresis. The heterodimers appear to dimerize to form a tetrameric structure, as shown by the profile of cross-linked complexes observed during non-denaturing gel filtration chromatography. The smaller OF-1 subunit (73 kDa) is probably responsible for the DNA binding activity of this protein. This conclusion is based on our observation that the sizes of cross-linked products are consistent with complexes between DNA and either the heterodimer or the smaller subunit but never between DNA and the larger subunit.
The fact that OF-1 binds to multiple sites within the HSV-1 origin suggests that OF-1 may form a large nucleoprotein complex involving interactions between proteins bound at multiple sites. This possibility is further supported by our finding that OF-1 may exist in a tetrameric form. Initiation complexes involving protein-protein interactions at multiple DNA sites have been observed in other replication systems. For example, initiation in Escherichia coli and bacteriophage requires binding of initiator proteins (dnaA and O, respectively) at multiple origin sites (22,23). Electron microscopy studies of the E. coli and systems reveal that the initiator proteins bound at different sites interact by protein-protein contacts to form a larger structure (22,23). Similar complexes have been proposed for the HSV-1 initiator protein, UL9, that binds as a dimer to Boxes I and II origin sequences (24) and may form a larger complex via tetrameric interactions (reviewed in Ref. 4).
At least two possible hypotheses for the interaction of OF-1 with the full-length origin in vivo are suggested by the DNA binding preferences of OF-1. First, since OF-1 prefers Box I single-stranded DNA, it is possible that OF-1 binds to singlestranded DNA resulting from unwinding of the origin by UL9. The UL9 helicase activity is known to unwind a partial duplex DNA containing Box I in the presence of ATP and ICP8 (7) and to form extruded single-stranded loops of DNA upon binding to oriS in the presence of ATP (27). In the second hypothesis, OF-1 may bind to an alternative conformation of the origin, known as oriS*. This conformation, recently proposed by Aslani et al. (28), consists of an extruded stem-loop structure formed by the top strand of oriS. In this structure, which is stabilized by UL9 binding (28), the top strand of the Box I motif is paired with the top strand of Box II. An analogous stem-loop structure is presumably formed by the bottom strand of oriS. This analogous structure would contain the Box I bottom strand, which is the preferred strand for OF-1 binding, and, thus, could serve as an optimal binding site for OF-1.
Possible Functional Interactions between OF-1, UL9, and oriS-When OF-1 was cross-linked to Box I DNA in the presence of UL9, we obtained substantial increases in the yield of cross-linked products compared to similar reactions without UL9. Hence, UL9 appears to enhance DNA binding by proteins in the OF-1 preparation. We believe that the increased products consist of OF-1-DNA complexes, although we cannot rule out the possibility that UL9 promotes binding of other proteins in our OF-1 preparation. However, arguing against this possibility is our finding that the increased products have the same mobilities as OF-1-DNA complexes formed without UL9. FIG. 9. The effect of OF-1 on the Box I-stimulated ATPase activity of UL9. ATPase activity of UL9 was measured as described under "Experimental Procedures" in the presence (⅜) and absence (OE) of an apparent 5:1 molar ratio of phosphocellulose-purified OF-1 over UL9. The molar ratio is based on our determination of protein purity of each preparation and the assumption that OF-1 exists as a heterodimer. Reactions contained UL9 (6 nM), DNA (0 -80 nM), and OF-1 (200 ng) or an equivalent volume of OF-1 buffer. K 0.5 values from these data are shown on the graph. For the reaction with UL9 alone, K 0.5 ϭ 3.3 Ϯ 0.2 nM, V max ϭ 0.88 Ϯ 0.02, and the correlation coefficient for the curve fit ϭ 0.9845. For the reaction with UL9 and OF-1, K 0.5 ϭ 9.3 Ϯ 0.6 nM, V max ϭ 0.81 Ϯ 0.03, and the correlation coefficient ϭ 0.9850.
We suggest two possible mechanisms that might explain the putative functional interactions between OF-1 and UL9. On the one hand, UL9 may physically interact with OF-1 to increase its affinity for DNA. Alternatively, UL9 may alter the conformation of the DNA to favor OF-1 binding. Consistent with this latter possibility, UL9 causes an increase in crosslinking of our OF-1 preparation to Box I but not to Box III. Since UL9 binds strongly to Box I but poorly to Box III sequences (14,25,26), the enhancement effect appears to correlate with UL9 binding to the DNA.
Our experiments further suggest that UL9 is displaced from Box I DNA in the presence of our OF-1 preparation. Evidence for displacement is provided by our observations that the formation of UL9-DNA complexes is reduced in the presence of OF-1 and that the Box I-stimulated UL9 ATPase activity is suppressed by OF-1. However, we presently cannot rule out the possibility that proteins other than OF-1 in our preparation are responsible for the apparent displacement. Competition between OF-1 and UL9 for binding to oriS sequences could regulate viral replication by affecting formation of a nucleoprotein initiation complex at the viral origin.
Summary-We have partially purified and characterized OF-1, a human protein that binds specifically to the HSV-1 replication origins (8). We find that OF-1 binds specifically to both duplex and single-stranded sequences from oriS with the strongest preference for the lower strand of the Box I region. Based on its interaction with multiple origin sites, we speculate that OF-1 participates in a large origin complex. We present evidence that UL9, the viral initiation protein, facilitates binding of OF-1 to Box I DNA. This result suggests that UL9 either interacts with OF-1 directly or alters the DNA structure to favor OF-1 binding. The OF-1-origin complex may be involved in initiation of viral replication.