The UL5 and UL52 subunits of the herpes simplex virus type 1 helicase-primase subcomplex exhibit a complex interdependence for DNA binding.

Herpes simplex virus type 1 encodes a heterotrimeric helicase-primase complex composed of the products of the UL5, UL52, and UL8 genes. The UL5 protein contains seven motifs found in all members of helicase Superfamily 1 (SF1), and the UL52 protein contains several conserved motifs found in primases; however, the contributions of each subunit to the biochemical activities of the subcomplex are not clear. In this work, the DNA binding properties of wild type and mutant subcomplexes were examined using single-stranded, duplex, and forked substrates. A gel mobility shift assay indicated that the UL5-UL52 subcomplex binds more efficiently to the forked substrate than to either single strand or duplex DNA. Although nucleotides are not absolutely required for DNA binding, ADP stimulated the binding of UL5-UL52 to single strand DNA whereas ATP, ADP, and adenosine 5'-O-(thiotriphosphate) stimulated the binding to a forked substrate. We have previously shown that both subunits contact single-stranded DNA in a photocross-linking assay (Biswas, N., and Weller, S. K. (1999) J. Biol. Chem. 274, 8068-8076). In this study, photocross-linking assays with forked substrates indicate that the UL5 and UL52 subunits contact the forked substrates at different positions, UL52 at the single-stranded DNA tail and UL5 near the junction between single-stranded and double-stranded DNA. Neither subunit was able to cross-link a forked substrate when 5-iododeoxyuridine was located within the duplex portion. Photocross-linking experiments with subcomplexes containing mutant versions of UL5 and wild type UL52 indicated that the integrity of the ATP binding region is important for DNA binding of both subunits. These results support our previous proposal that UL5 and UL52 exhibit a complex interdependence for DNA binding (Biswas, N., and Weller, S. K. (1999) J. Biol. Chem. 274, 8068-8076) and indicate that the UL52 subunit may play a more active role in helicase activity than had previously been thought.

DNA helicases catalyze the transient unwinding of dsDNA 1 to form ssDNA using the energy of NTP hydrolysis. Helicases are essential in many biological processes including replication, recombination, transcription, and DNA repair and have been isolated from prokaryotes, eukaryotes, and viruses. The helicase-primase complex of herpes simplex virus type 1 (HSV-1) is a heterotrimeric complex composed of the products of the UL5, UL52, and UL8 genes (1). All three genes are essential for viral DNA replication (2)(3)(4)(5)(6)(7). The UL5-UL52-UL8 complex possesses primase, ssDNA-dependent NTPase, and 5Ј to 3Ј DNA helicase activities (1, 8 -11). The HSV-1 helicase-primase complex can be isolated from insect cells that have been simultaneously infected with recombinant baculoviruses that express each of the three subunits (9). A subassembly consisting of the UL5 and UL52 gene products also exhibits all the enzymatic activities of the holoenzyme in vitro (12). The UL5 protein contains seven conserved motifs found in all members of Superfamily 1 (SF1) helicase proteins (13). The UL52 protein contains several conserved motifs found in other primases (14,15). Neither UL5 nor UL52 appears to possess any enzymatic activities when expressed alone (9,12). The UL8 gene product does not exhibit any enzymatic activities (10,12) but can stimulate both the helicase and primase activities of the helicase-primase complex (16 -19). Furthermore, UL8 may facilitate the entry of the heterotrimer into the nucleus of infected cells (20,21).
Although the molecular details of the mechanism of DNA unwinding is unknown for any helicase, it is likely that the unwinding reaction requires the coupling of several events such as ATP binding, ATP hydrolysis, single strand and double strand DNA binding, and translocation along the DNA. Many helicases function as multimers such as dimers (e.g. Escherichia coli Rep (22,23)) or hexamers (e.g. helicases of T4 and T7 bacteriophages (24,25) and SV40 large T antigen (26,27)). Although it has been suggested that oligomeric structures provide multiple DNA binding sites, which are required for helicase action (28), it appears that at least two helicases, E. coli DNA helicase II and Bacillus stearothermophilus PcrA, are active as monomers (29,30). Three models to explain the mechanism of helicase activity have been proposed. The inchworm model posits that conformational changes caused by binding and hydrolysis of ATP cause a helicase monomer to "inch" along the DNA (30,31). Monomeric helicases would presumably contain at least two nonidentical DNA binding sites on each monomer. The rolling model, which is based on the dimeric Rep protein, posits that a helicase must act as (at least) a dimer and that each subunit of the dimer can bind to either ssDNA or duplex DNA (23). According to this model, a helicase rolls along the DNA with alternating subunits binding first to ds then to ssDNA. A third model proposed for the hexameric helicases posits that the core of the hexameric unit provides a channel through which a single strand of DNA can be threaded (32)(33)(34). The protein would move along one strand with alternating subunits responsible for ATP hydrolysis. To distinguish between these models and to understand the mechanism of helicase action, it will be necessary to obtain more detailed information about how helicases contact DNA. Two members of SF1 helicases, Rep and PcrA, have been crystallized in the presence of DNA (30,35). The crystal structure of the E. coli Rep helicase bound to ssDNA, and ADP revealed putative contact residues for ssDNA on the protein (35); however, many of these assignments have not been confirmed by genetic analysis.
Previous DNA binding studies revealed that the UL5-UL52 subcomplex binds to ssDNA more effectively than to dsDNA and that the minimum length of ssDNA that can bind and stimulate its ATPase activity is about 12 nucleotides (36). Herein we show that the UL5-UL52 subcomplex binds much more efficiently to a forked substrate than to either ss or dsDNA. The fact that the HSV-1 helicase is part of a multiprotein complex complicates the analysis of the DNA binding sites of the individual subunits. We have previously shown that both subunits can contact single-stranded DNA in a photocrosslinking assay (37). Moreover, we have shown a complex interdependence on both subunits for DNA binding, in that a mutation in the putative Zinc binding domain of the UL52 subunit has drastic effects on the ability of UL5 to cross-link singlestranded DNA. In this paper we have taken two approaches to study the interaction of the UL5-UL52 subcomplex with DNA. Cross-linking studies using forked substrates with substitutions of deoxyuridine (dIU) in three different positions indicate that the UL5 and UL52 subunits contact the forked substrates at different positions; UL5 appears to contact DNA near the fork, whereas UL52 appears to contact the ss tail of the forked substrate. Neither subunit appears to directly contact dsDNA. In a second approach, we performed DNA binding and crosslinking assays on a series of UL5 mutants whose mutations lie in conserved helicase motifs shared by other SF1 members (38,39). The results confirm a complex interdependence between the two subunits and indicate that the UL52 subunit may play a more active role in helicase activity than had previously been thought. Furthermore, these studies suggest that the HSV-1 helicase-primase may act as a monomer (one heterotrimer per replication fork) and favor the inchworm model for the mechanism for helicase activity.

EXPERIMENTAL PROCEDURES
Reagents-Supplemented Graces's medium and 10% Pluronic ® F were purchased from Life Technologies, Inc. Fetal calf serum was obtained from Atlanta Biologicals. Penicillin-streptomycin solution, ampicillin, phenylmethylsulfonyl fluoride, leupeptin, and pepstatin were purchased from Sigma. The 20-ml HiLoad 16/10 SP-Sepharose fast flow column was from Amersham Pharmacia Biotech. Inc. The 12-ml Uno Q (Q-12) column was from Bio-Rad. The 25-ml Superose 12 HR column was from Bio-Rad. Radiolabeled nucleotides were purchased from Amersham Pharmacia Biotech. Substituted oligonucleotides were synthesized from Cruachem. A polyclonal antibody (1248) directed against the C-terminal 10 amino acids of UL52 was a kind gift from Dr. Mark Challberg (National Institutes of Health, Bethesda, MD).
Protein Expression and Purification-2 liters of Sf9 cells were grown in suspension at 27°C in Graces's insect medium as described previously (39). The wild type and variant UL5-UL52 subcomplexes were purified essentially as described earlier with an additional gel filtration step. Cells were dounced using 15 strokes of a tight fitting pestle in Buffer A, and the cytosolic extracts were clarified by centrifugation at 35,000 ϫ g for 30 min. UL5-UL52 subcomplexes were precipitated from the cytosolic extract by the addition of an equal volume of Buffer B containing 0.2 M NaCl and 2 M ammonium sulfate and incubation on ice for 4 h. The resultant protein pellets were resuspended in Buffer B containing 0.1 M NaCl and dialyzed against the same buffer. The dialyzed sample was loaded onto an SP-Sepharose column equilibrated with Buffer B containing 0.1 M NaCl, and the column was washed with 5 column volumes of the equilibration buffer. Fractions containing the UL5-UL52 subcomplex were identified by SDS polyacrylamide gel electrophoresis. The UL5-UL52 subcomplex elutes from the column in the void volume. Pooled fractions from the SP-Sepharose column were loaded onto a 12-ml Uno Q column equilibrated with Buffer B containing 0.1 M NaCl. The column was washed with five column volumes of Buffer B containing 0.1 M NaCl, and the protein was eluted using a 185-ml linear gradient of Buffer B containing 0.1-1 M NaCl. Pooled fractions from the Uno Q column were concentrated using a centrifugal concentrator with a 10-kDa cut-off (Microsep TM , Pall Filtron) and loaded onto a 25-ml Superose 12 HR column equilibrated with Buffer B containing 0.1 M NaCl. The fractions containing the peak activities were pooled, concentrated, and frozen at Ϫ70°C.
DNA Substrates-An 18-mer of oligo(dT), PCdT18(5), with a dIU substitution at the 5th T from the 5Ј end was synthesized by Cruachem and end-labeled with [␥-32 P]ATP. Forked DNA substrate A was constructed by heat denaturing and annealing 80 pmol of the helicase 48C/FS oligonucleotide (5Ј CGAAAGTACGTTATTGCGACTGGCCGT-CGCTCTACAACGTCGTGACTG 3Ј) radiolabeled at its 5Ј end with [␥-32 P]ATP and 80 pmol of unlabeled 48FS oligonucleotide (5Ј CAGTC-ACGACGTTGTAGAGCGACGGCCAGTCGGTTATTGCATGAAAGC 3Ј). The underlined residues are complementary and create the duplex region of the molecule. After annealing, the products were subjected to electrophoresis on an 8% nondenaturing polyacrylamide gel, and the forked substrate was purified by electroelution and ethanol precipitation. Forked substrates (FS B, FS C, and FS D) were prepared by annealing 80 pmol of each of the end-labeled 48C/FSM oligonucleotide (5Ј CGAAAGdIUACGTTATTGCGACTGGCCGTCGCTCTACAACGTCGT-GACTG 3Ј), 48C/FSM27 oligonucleotide (5Ј CGAAAGTACGTTATTGC-GACTGGCCGdIUCGCTCTACAACGTCGTGACTG 3Ј), or 48C/FSM15 oligonucleotide (5Ј CGAAAGTACGTTATdIUGCGACTGGCCGTCGCT-CTACAACGTCGTGACTG 3Ј), respectively, to 80 pmol of the unlabeled 48FS oligonucleotide. In FS B, the substitution is in the 7th position from the 5Ј end of the lower (labeled) strand (see Fig. 2). In FS C, the dIU substitution is within the duplex region (see Fig. 2), and in FS D, the substitution is within the ss region of the lower (labeled) strand very near the ss/dsDNA junction. The duplex DNA substrate was prepared in a similar manner; 80 pmol of 32 S oligonucleotide (5ЈCAGTCACGA-CGTTGTAGAGCGACGGCCAGTCG3Ј) was annealed to the complementary 32 CS oligonucleotide (5ЈCGACTGGCCGTCGCTC-TACAACGTCGTGACTG3Ј).
Gel Mobility Shift Assay-Gel mobility shift assays were essentially performed as described previously (39). The reaction mixture (25 l) contained 20 mM Na ϩ HEPES (pH 7.6), 1 mM DTT, 0.1 mg/ml bovine serum albumin, 10% glycerol, 5 mM MgCl 2 , 1.2 pmol (molecules) of the DNA substrates labeled with [␥-P 32 ]ATP and 4 pmol of the UL5-UL52 subcomplex with or without UL8 protein (12 pmol), ATP (5 mM), ADP (5 mM), or ATP␥S (5 mM). The reaction was allowed to proceed for 10 min on ice and terminated by the addition of one-tenth of a volume of a loading solution (80% glycerol, 0.1% bromphenol blue). Reaction products were analyzed on a 4% nondenaturing acrylamide, 0.11% bisacrylamide gel at 150 V at 4°C. The gels were dried and exposed to film at Ϫ70°C.
Photocross-linking-Photocross-linking experiments were performed essentially as described previously with 1.2 pmol of the indicated DNA substrate molecules and 4 pmol of UL5-UL52 subcomplex in 20 mM Na ϩ HEPES (pH 7.6), 1 mM DTT, 0.1 mg/ml bovine serum albumin, 10% glycerol, and 5 mM MgCl 2 (37). The samples were incubated on ice for 10 min before irradiation. An IK series He-Cd laser (IK 3302R-E, KIMMON; Kimmon Electric Co., Ltd.) was used to achieve monochromatic 325-nm light. The laser beam output was 34 milliwatts measured with a power meter, Mentor MA10, Scientech® (Scientech, Inc., Boulder, CO). Samples were irradiated in a methacrylate cuvette (catalog number 14-385-938; Fisherbrand) at room temperature. At different time points aliquots were withdrawn, boiled for 5 min in SDS-PAGE loading buffer, and subjected to SDS-PAGE on an 8% gel. The gels were dried and exposed to film at Ϫ70°C.

UL5-UL52 Subcomplex Binds to a Forked Substrate More
Efficiently than to either Single-stranded or Double-stranded DNA-Previous studies with the UL5-UL52 subcomplex indicated a preference for ss versus dsDNA; in a filter-binding experiment the subcomplex bound ssDNA about 5-fold more effectively than it did dsDNA (36). We previously showed that the UL5-UL52 subcomplex could bind a forked substrate generated by the annealing of partially complementary oligonucleotides (39). Here we compare the binding efficiencies of UL5-UL52 to the forked substrate and to single-and doublestranded DNA using a mobility shift experiment. Fig. 1 shows that the UL5-UL52 subcomplex can bind forked, ss, and dsDNA ( Fig. 1, lanes b, i, and p, respectively). The gel shift data indicate that the binding of the UL5-UL52 subcomplex to a forked substrate is at least 8-fold higher than to ssDNA and is at least 35-fold higher than to dsDNA. Addition of UL8 to the binding reaction resulted in a supershift to a slower migrating species (Fig. 1, lanes c, j, and q); in the case of the forked substrate (Fig. 1, lane c), the supershifted band is somewhat smeared, perhaps reflecting the complex interactions exhibited by UL5-UL52 with forked DNA (see below). Quantification of the gel shift data indicates that the UL8 stimulates the binding of UL5-UL52 to both forked (1.8-fold) and ss (2.2-fold) DNA substrates (Table I). To determine whether nucleotide di-and triphosphates play a role in the DNA binding properties of UL5-UL52, the binding of the subcomplex with ss and forked DNA substrates was tested in the presence of ATP, ADP, and ATP␥S ( Fig. 1). Single strand DNA binding was stimulated 1.6-fold in the presence of ADP, and the binding of UL5-UL52 to forked substrate was stimulated in the presence of ATP (1.4-fold), ADP (1.3-fold), and ATP␥S (1.5-fold) ( Table I). In summary, it appears that the UL5-UL52 subcomplex binds much more efficiently to the forked substrate than to either single-stranded or double-stranded DNA; the addition of UL8 or nucleotide cofactors exhibited modest but reproducible stimulatory effects on DNA binding of the subcomplex. The similar levels of stimulation observed with ADP, ATP, and ATP␥S suggest that the binding of ATP but not its hydrolysis is important for optimal binding of UL5-UL52 to the forked substrate.
Cross-linking of UL5-UL52 to a Forked Substrate-We have previously used a photocross-linking assay to show that both UL5 and UL52 subunits of the wild type UL5-UL52 subcomplex can contact a short ss oligomer (37). During replication, the helicase-primase would presumably contact a replication fork consisting of double-and single-stranded DNA; we therefore initiated cross-linking studies using a series of forked substrates shown in Fig. 2. A He-Cd light source that emits at 325 nm was used to photocross-link the UL5-UL52 subcomplex to a 32 P end-labeled forked substrate in which dIU was substituted for one of the thymidine residues. In FS B, dIU was placed in the ss portion of the substrate at a position 7 nucleotides from the 5Ј end of the lower strand (Fig. 2). The UL5-UL52 subcomplex was cross-linked to a 5Ј 32 P-labeled single strand oligonucleotide (Fig. 3A, lane a) or to a 5Ј 32 P-labeled forked substrate (FS B) (Fig. 3A, lane b). As previously reported (37), when the subcomplex was cross-linked to the single strand oligonucleotide, two labeled bands were observed by SDS-PAGE, one migrating at ϳ100 kDa corresponding to UL5 and a slower band migrating at 120 kDa corresponding to UL52 (Fig.  3A, lane a). When the UL5-UL52 subcomplex was cross-linked to FS B, however, slower migrating bands were observed (Fig.  3A, lane b); the uppermost band migrates at a position corresponding to ϳ220 kDa and a lower set of smeared bands, which may contain two or more species migrating at a position corresponding to 170 -195 kDa. A time course of binding in which the UL5-UL52 subcomplex was incubated with FS B irradiated for varying lengths of times was performed to determine whether the pattern of cross-linked bands changes with time. Fig. 3B shows that the time of irradiation correlates with the amount of cross-linked material and that the pattern of bands, a 220-kDa band and two or more bands migrating between 170 a DNA binding was measured with a phosphorimager system. The relative DNA binding efficiencies were calculated with respect to the total number of counts; 100% is defined as the net number of counts in the aliquot lacking any enzyme. The values represent the average of three independent experiments. Mean deviations are shown. and 195 kDa, remains constant throughout the experiment.
To further characterize the cross-linking to a forked substrate, substrates were generated that contain dIU at various positions (Fig. 2). FS A does not contain any substitutions, FS B was described above and contains a substitution entirely within the single-stranded portion of the substrate, FS C contains a dIU substitution within the duplex portion of the forked substrate at the 27th position from the 5Ј end of the lower labeled strand (11 base pairs from the ss/ds junction), and FS D contains dIU substitution at the 15th position from the 5Ј end of the lower strand very close to the ss/ds junction (Fig. 2). In a gel shift assay, it is apparent that the UL5-UL52 subcomplex can bind both FS B and FS C with equal efficiencies (Fig. 4A,  lanes b and d). However, the cross-linking experiment shown in Fig. 4B demonstrates that UL5-UL52 can be cross-linked to FS B much more efficiently than it can be cross-linked to FS C (Fig. 4B, compare lanes a and b to lanes c and d). Quantification of the cross-linked bands indicates that cross-linking was 4.4fold more efficient to FS B than FS C at the 15-min time point and 6.2-fold more efficient at the 30-min time point. This result suggests that although FS C can be bound to the subcomplex as assessed by the gel mobility assay, neither UL5 nor UL52 are located in close proximity to the duplex portion of the substrate. In the experiment shown in Fig. 5, forked substrates B and D were compared. By gel mobility assay, both substrates were bound with equal efficiency (Fig. 5A, lanes b and d). In this case the UL5-UL52 subcomplex could be cross-linked to FS B and FS D with more or less equal efficiency (Fig. 5B); however, the mobility of the bands cross-linked to FS D was very different from those cross-linked to FS B. In this case two or more bands that migrate in the range of 115-180 kDa were observed (see below). The observation that the subcomplex cross-links to both FS B and FS D but is unable to cross-link to FS C suggests that the subcomplex is bound to the single-stranded rather than the duplex region of the forked substrate.
DNase1/S1 Nuclease Digestion of Cross-linked Species-SDS-PAGE analysis of the FS B and FS D cross-linked species revealed the presence of multiple radioactive bands that migrate more slowly than UL5 and UL52 (Figs. 4 and 5). The forked substrate itself has a molecular mass of 30 kDa, and the slow mobilities of the cross-linked species may be a result of binding multiple substrate molecules to either UL5 or UL52. To characterize the composition of the slower migrating crosslinked species, they were treated with DNase1 and S1 nuclease for increasing amounts of time. Treatment with nucleases is expected to degrade the DNA substrates allowing identification of the proteins present in the high molecular weight complexes. Fig. 6A demonstrates that after treatment of the FS B crosslinked material with nucleases, two radiolabeled bands appear that migrate at the same position as UL52 and UL5. At 5, 10, and 20 min of digestion, the strongest band corresponds to UL52. By 2 h, the signals corresponding to both UL5 and UL52 disappeared almost entirely (Fig. 6A, lane g). In Fig. 6B, the FS D cross-linked material was treated with nucleases for various periods of time. At the 20-and 40-min time points, a signal corresponding to UL5 was predominant (Fig. 6B, lanes d and e). This experiment suggests that the high molecular weight crosslinked bands represent bound forms of UL5 and UL52 to a forked substrate. Furthermore, the signals obtained after partial nuclease digestion suggest that UL5 preferentially crosslinks to FS D, which contains the dIU substitution at a position very close to the ss/ds junction, whereas UL52 preferentially cross-links to FS B, which contains the dIU residues close to the 5Ј end of the ssDNA tail.
Immunoblot Analysis of Cross-linked Complexes-Immunoblotting was used to confirm the identity of the slower migrating species observed when the UL5-UL52 subcomplexes are cross-linked to forked substrates B and D (Figs. 4B and 5B, described above). UL5-UL52 subcomplexes cross-linked either to a single strand substrate or forked substrate or not crosslinked were subjected to SDS-PAGE in duplicate; one-half of the gel was processed for autoradiography, and the other half was subjected to immunoblot analysis with antisera raised against either UL5 or UL52. Fig. 7A shows the autoradiogram of the cross-linked samples; as in Fig. 3, slower migrating bands of ϳ170 -195 and 220 kDa are seen when the UL5-UL52 subcomplex is cross-linked to FS B (Fig. 7A, lane 2). In these cross-linking experiments, only a small proportion of the UL5 and UL52 proteins are actually cross-linked; therefore, it is expected that immunoblotting will detect uncross-linked UL5 and UL52. As predicted, in the experiments shown in Fig. 7B, antisera against UL5 (lanes 4, 5, and 6) reacts primarily with a band corresponding to uncross-linked UL5, although a weak band corresponding to UL52 is also observed, presumably because of cross-reactivity of the antisera with UL52. Antisera against UL52 (Fig. 7C, lanes 9, 10, and 11) primarily reacts with a band corresponding to UL52 and a weak band corresponding to UL5, again probably because of cross-reactivity. Interestingly, in the material cross-linked to the forked substrate, three slower migrating bands corresponding to 170 -195 kDa (marked with an *) were detected with the UL52 antibody (Fig. 7C, lane 10). No slower migrating bands were detected in the ss cross-linked sample or in the uncross-linked protein sample (Fig. 7C, lanes 9 and 11, respectively). This result indicates that UL52 is present in the complexes cross-linked to FS B. We cannot rule out that some UL5 is also present; however, the predominant signal appears to be UL52. In the experiment shown in Fig. 8, the UL5-UL52 subcomplex was cross-linked to FS D. Fig. 8A shows the autoradiogram of the subcomplex cross-linked to ssDNA or FS D or uncross-linked (Fig. 8A, lanes 3, 2, and 1, respectively). With the UL5 antibody, a strong UL5 uncross-linked band was seen in all three lanes (Fig. 8B, lanes 5-7). A faint band migrating at UL52 was seen in the sample cross-linked to ssDNA; this is probably because of cross-reactivity of UL52 with the UL5 antibody (Fig.  8B, lane 7). Several more intense slower migrating bands could be detected in the material cross-linked to FS D, indicating that UL5 is present in the cross-linked material (Fig. 8B, lane 6). With the UL52 antibody, only the uncross-linked UL52 was detected (Fig. 8C, lanes 9 -11). Thus it appears that, as described above, UL52 is cross-linked preferentially to FS B, and UL5 is cross-linked preferentially to FS D; however, we cannot rule out that these complexes also contain small amounts of the other subunit.
To confirm that the high molecular weight cross-linked species represent complexes containing UL5 and UL52, as the complexes with single strand DNA clearly do, competition experiments with single-stranded 48-mer DNA oligonucleotide (either unlabeled or labeled) were also performed. This experiment indicates that the slower migrating bands of the complex with FS B disappear in the presence of unlabeled ssDNA (data not shown). If labeled ssDNA is used as the competitor, the high molecular weight bands are decreased in intensity, and bands corresponding to UL5 and UL52 bound to ssDNA are observed, although the UL52 band is stronger (data not shown). These results confirm that the high molecular weight species seen with the FS B substrate contained both the UL5 and the UL52 subunits, although UL52 was predominant as demonstrated above.
Cross-linking of Mutant UL5-Wild Type UL52 Subcomplexes  FSB (A, lane b) in a methacrylate cuvette (light path, 10 mm) at room temperature with an He-Cd laser emitting 34 milliwatts at 325 nm. Samples were removed after 30 min of irradiation. Samples were boiled for 5 min in 1ϫ SDS-PAGE loading buffer and subjected to electrophoresis on a 8% SDS-PAGE gel, which was then dried and exposed to film overnight at Ϫ70°C. Arrows in A indicate UL5 and UL52 cross-linked to ssDNA. The approximate molecular mass of each radioactive band was calculated from the standard graph of the 10-kDa protein ladder. In to ss and Forked Substrates-The DNA binding sites within the helicase-primase complex have not been mapped. We previously reported the isolation and characterization of UL5 mutants bearing mutations in the conserved motifs shared among SF1 members (39). In that study, we found that mutant subcomplexes were able to bind forked substrates using a gel shift assay as well or better than wild type subcomplexes; however, this assay reflects the DNA binding of the whole complex, and it was not possible to determine the individual contributions of either subunit. For instance, if one of the UL5 mutant proteins was defective for DNA binding, it may not have been apparent using this assay, because a defect may have been masked by the binding of the UL52 subunit. To characterize UL5 binding more directly and map regions of UL5 responsible for contacting DNA, we tested subcomplexes containing mutant UL5 proteins and wild type UL52 for their ability to cross-link various substrates. In the experiment shown in Fig. 9, wild type and motif mutant subcomplexes were irradiated for 10 and 30 min by a He-Cd laser in the presence of the labeled single-stranded 18-mer oligo(dT) substrate, which contains one dIU substitution. SDS-PAGE and subsequent autoradiography revealed that the UL5 subunit with a mutation in motif I (Fig. 9A, lanes c and d) or motif III (Fig. 9A, lanes g and h) is somewhat defective in cross-linking to ssDNA compared with wild type (Fig. 9A, lanes a and b). Fig. 9B is a Coomassie-stained gel showing that approximately the same amount of protein was loaded into each cross-linking reaction. Quantification demonstrates that the DNA binding ability of the motif I mutant UL5 subunit was 1.7-fold less than wild type, and the motif III mutant was 1.6-fold less than wild type (Fig. 9C). The other UL5 mutant proteins cross-link to ssDNA with approximately wild type efficiency. Fig. 9A also shows that in some of the mutant subcomplexes, the UL52 subunit fails to bind ssDNA effectively. For instance, in the subcomplexes containing the motif I mutation (Fig. 9A, lanes c and d), the motif II mutation (Fig. 9A, lanes e and f), or the motif III mutation (Fig. 9A, lanes  g and h), cross-linking to ssDNA of the UL52 subunit was significantly lower than wild type. Quantification indicates that UL52 subunit binding is decreased 6.5-fold in motif I, 3.5-fold in motif II, and 2.7-fold in motif III (Fig. 9C). Motif IV, motif V, and motif VI exhibited wild type levels of cross-linking of both the UL5 and UL52 subunits to ssDNA substrate (Fig. 9,  A and C).
In the experiment shown in Fig. 10, we asked whether the ability of motif mutant subcomplexes to cross-link forked DNA was different from their ability to cross-link the ssDNA substrate described above. Six different UL5 helicase motif mutants (motifs I, II, III, IV, V, and VI) were compared with wild type for their ability to cross-link to forked substrate B (Fig. 10,  A and B). The overall cross-linking efficiency as measured by adding the intensities of the two radiolabeled bands was slightly (1.3-fold) lower in the motif I mutant subcomplex compared with wild type (Fig. 10A, compare lanes b and d). The motif V and motif VI mutant subcomplexes showed slightly higher efficiencies than wild type (Fig. 10A, lanes j and l,  respectively), whereas motif III and motif IV mutant proteins showed 2.4-to 2.8-fold higher cross-linking efficiencies than wild type (Fig. 10, lanes f and h, respectively). The motif II mutant binds with approximately wild type efficiency to the forked substrate (Fig. 10B, compare lanes c and d to a and b). Thus the ability of the mutant subcomplexes to cross-link to a single-stranded DNA substrate differs considerably from their ability to bind to forked substrates; even subcomplexes with defects in ssDNA binding appear to be stabilized on forked substrates. DISCUSSION In this paper we have studied DNA binding of the HSV-1 helicase-primase by analyzing the substrate preferences of the  Fig. 4 using forked substrate B (lanes a and  b) and forked substrate D (lanes c and d). Lanes a and c (A) represent the reaction in the absence of any protein. In cross-linking reaction samples were taken out at the 15- min (B, lanes a and c) , lanes b-g, respectively). In B, aliquots were taken at 5-, 10-, 20-, and 40-min intervals (lanes b-e, respectively). Samples were boiled for 5 min in 1ϫ SDS-PAGE loading buffer and subjected to electrophoresis on a 8% SDS-PAGE gel, which was then dried and exposed overnight at Ϫ70°C to film. UL5-UL52 cross-linked with ssDNA were shown in lane h and i (A) and in lane f (B).
helicase-primase complex and by determining the binding properties of subcomplexes containing various mutant forms of the UL5 subunit. Several observations were made: 1) UL5-UL52 binds preferentially to a forked substrate over ss or dsDNA substrate in a mobility shift assay. 2) Although nucleotides are not absolutely required for DNA binding, ADP stimulates the binding of UL5-UL52 to ssDNA, whereas ATP, ADP, and ATP␥S stimulates the binding to a forked substrate. 3) The UL5-UL52 subcomplex can be cross-linked to a forked substrate, and the composition of the resulting cross-linked species varies depending on the position of the dIU substitution. When the substitution is within the ss region of the substrate, UL52 is preferentially cross-linked; however, when the substitution is near to the ss/ds junction, UL5 appears to be preferentially cross-linked. 4) UL5 proteins bearing mutations in the conserved helicase motifs varied in the ability of subcomplexes containing mutant UL5 and wild type UL52 to bind ss or forked substrates. These results suggest that the ATP binding region is important for DNA binding of both subunits and confirm our previous finding of a complex interdependence between UL5 and UL52 subunits for their DNA binding properties (discussed below).
Substrate Preferences and Effects of Nucleotides on DNA Binding-Previous reports indicated that HSV-1 UL5-UL52 can bind ssDNA 5-fold more effectively than ds plasmid DNA by filter binding assay (36). To our knowledge, this paper presents the first comparison between forked, ss, and dsDNA substrates for the HSV helicase-primase. The observation that DNA binding to the forked substrate is much better than to the single strand substrate can be explained in at least three nonmutually exclusive ways: 1) It is possible that the helicaseprimase needs to bind first to single-stranded DNA to be able to recognize dsDNA. In other words, the helicase needs to be loaded onto dsDNA. Thus, the previously observed low affinity for dsDNA may reflect the fact that the enzyme can only bind dsDNA after it has contacted ssDNA. 2) Each subunit (UL5 and UL52) appears to have the capacity to contact DNA individually as determined by cross-linking studies to single strand substrates (37); however, binding to the forked substrate may reflect cooperativity between binding site(s) on each of the subunits resulting in more stable binding to the forked substrate.
3) The structure of the forked substrate itself may act to promote binding of the subcomplex. For instance, the presence of a joint region between the ds and ss regions of the substrate may provide a binding surface that greatly stabilizes the binding of one or both subunits.
Models for the mechanism of helicase unwinding have been proposed that make predictions concerning the types of DNA contacts a helicase is expected to make with its substrate and its stoichiometry of binding. According to the rolling model, each helicase subunit must be able to bind ssDNA, as well as dsDNA, but not both at the same time (28). According to the inchworm model, a monomer may need to bind ss and dsDNA at the same time at least during a portion of the reaction cycle (30). In the hexameric helicases, the helicases are proposed to contact ssDNA primarily. Thus, the binding affinities and stoichiometry of binding have important implications for the mechanism of helicase action. The stoichiometry of binding of the UL5-UL8-UL52 complex at the replication fork is not known; however, it is possible that the UL5-UL8-UL52 complex exists either as a monomer (one heterotrimer) or as a dimer (two molecules of the heterotrimer). A dimer of trimers might be expected to function either as a dimer consistent with the rolling mechanism or as hexamer. Two lines of evidence appear to rule out the rolling mechanism for the HSV-1 helicaseprimase. First, the UL5-UL52 subcomplex does not bind efficiently to dsDNA as would be predicted by the rolling mechanism. Second, genetic analysis from our laboratory suggest that UL5 does not function as a dimer or higher order structure like a hexamer; mutants in motifs I and II that abolish helicase and ATPase activity do not exert a transdominant effect on wild type UL5 function. 2 This is in contrast to mutants in motifs I and II of UL9 that inactivate helicase activity and are strongly transdominant; for UL9, the potent transdominant activity appears to be due at least in part to the ability of UL9 to dimerize (40,41). Another reason to suspect that the HSVhelicase-primase is unlikely to function as a hexamer is that only superfamily 3 and family 4 helicases have been shown to function as hexameric helicases (reviewed in Refs. 42 and 43). In summary, the genetic and DNA binding evidence lead us to favor the inchworm mechanism for this helicase-primase complex. According to the inchworm model, each monomeric unit would possess at least two binding sites. This implies either that UL5 has two or more binding sites or that one of the other 2 R. Zhou and S. K. Weller, unpublished data. Cross-linking reactions were carried out in a methacrylate cuvette (light path, 10 mm) at room temperature with an He-Cd laser emitting 34 milliwatts at 325 nm. Samples were removed after 10 and 30 min of irradiation for wild type (WT) and UL5 motif mutants (MI-MVI). Samples were boiled for 5 min in 1ϫ SDS-PAGE loading buffer and subjected to electrophoresis on a 8% SDS-PAGE gel, which was then dried and exposed overnight at Ϫ70°C to film. Panel A represents the autoradiogram, panel B represents the Coomassie-stained gel, and panel C represents the quantification of cross-linking data. subunits, most likely UL52, contributes to helicase activity by providing a second DNA binding site. We favor the later possibility as it is consistent with existing data (discussed below).
In this study we also confirmed and extended previous reports on effects of nucleotides on the DNA binding activities of the UL5-UL52 subcomplex. Healy et al. (36) had previously reported that the ssDNA binding activity of UL5-UL52 could be stimulated 1.7-fold in presence of ADP. In this paper we confirm this result, and in addition we show that ATP, ADP, and ATP␥S can stimulate the binding of the protein to a forked substrate. Modulation of DNA binding affinity by nucleotide cofactors have been reported for several helicases (23,44,45). Kinetic studies with the Rep helicase suggests that ADP favors binding to ssDNA whereas a nonhydrolyzable ATP analog favors the simultaneous binding to both ss and duplex DNA by the Rep dimer (23). The behavior of HSV-1 helicase-primase appears to resemble the Rep helicase in this respect, because ADP can stimulate the binding to both forked substrates and ssDNA, but ATP and ATP␥S stimulate the binding to the forked substrate only (Table I). Thus, binding of ATP-Mg 2ϩ but not its hydrolysis may be important for optimal binding of helicase to the forked substrate. The binding of ATP may allosterically regulate the affinity of the UL5-UL52 protein for different types of DNA substrates. The binding of ATP and other nucleotides may enhance the formation of protein-DNA complexes or increase their stability once formed, or both.
UL5-UL52 Interactions with Forked Substrates-Photocrosslinking experiments were designed to elucidate the contributions made by the individual subunits to DNA binding and to determine which part of the ss and ds junction of a replication fork is occupied by each subunit of the UL5-UL52 complex. The slow mobility of subcomplexes cross-linked to either FS B or FS D (Figs. 4B and 5B) indicates that there are two to three substrate molecules bound to each enzyme complex. These data suggest that there may be more than one DNA binding site per subcomplex. A combination of experiments including DNase treatment, Western blot analysis, and competition experiments indicate that the higher molecular weight radiolabeled bands are indeed composed of the UL5 and UL52 subunits. Furthermore, these experiments suggest that UL5 preferentially cross-links at a position close to the ss/ds junction, whereas UL52 preferentially crosslinks within the ss region of the forked substrate.
Within the HSV-1 helicase-primase, UL5 has long been assumed to be the helicase and UL52 to be the primase; however, several lines of evidence suggest a complex interdependence on both subunits for the activities of the subcomplex. For instance, Barrera et al. (46) analyzed an intertypic helicase-primase complex consisting of a UL5 subunit from HSV-1 and a UL52 subunit from HSV-2. This subcomplex exhibited decreased helicase and primase activities and diminished neurovirulence, indicating that small structural changes in the UL5 subunit could also affect primase activity. Furthermore, we previously showed that a mutation in the putative Zn binding region of the UL52 subunit abolished not only primase activity but also ATPase and helicase activities (37). In addition, both UL5 and UL52 subunits within the mutant subcomplex were totally defective in cross-linking to ssDNA (37). To explain these and other observations discussed below, we propose that UL52 may play a more important role than previously recognized in the helicase activity of the subcomplex. It is possible that binding of UL52 to ssDNA may be necessary to load the UL5 subunit. Alternatively, UL52 may play an even more active role in the helicase mechanism by providing a second DNA binding site necessary during the unwinding reaction.
Mutations in the Helicase Motifs of UL5-We previously reported the biochemical analysis of mutants in conserved res-idues in the motifs of UL5. We found that motif I is directly involved in ATP binding and/or hydrolysis and that motif II appears to be required for coupling of DNA binding to ATP hydrolysis. Residues in motifs III, IV, V, and VI are involved in the coupling of ATP hydrolysis and DNA binding to the process of DNA unwinding (39). The defects in ATPase activity in the UL5 mutants can be explained in light of the recently solved crystal structure of two other SF1 family helicases, Rep and PcrA, which exhibit a remarkable degree of similarity to each other (30,35,47). Both contain two recA-like domains arranged such that the conserved helicase motifs all lie along a cleft between them. This arrangement has led to the suggestion that helicase activity may be carried out through conformational changes within the molecule in response to ATP binding, ATP hydrolysis, and binding of DNA (30,35,47). The severe defects in ATPase activity exhibited by the UL5 mutations in motifs I and II are consistent with a role in ATP binding and hydroylsis. Furthermore, the lack of coupling between ATPase and helicase activities of mutations in motifs III, IV, V, and VI (39) can be explained by the position of these motifs along the cleft between the two recA like domains. Our results support the proposal that the conserved motifs play a role in mediating conformational changes within the molecule in response to DNA and nucleotide binding.
The crystal structure of Rep and PcrA in the presence of single-stranded DNA has also been reported (30,35). In both cases, the ssDNA was found to bind along the top of the recAlike domains, and residues from motifs Ia, III, and V were shown to contact ssDNA. To confirm the predictions made from the structural information about Rep and PcrA for UL5, the motif mutants described above were analyzed for their ability to bind various substrates. Cross-linking data with ss substrates indicated that subcomplexes containing motif I mutations are defective not only in UL5 but also UL52 binding. This result was somewhat surprising, because motif I is not physically located near the putative ssDNA binding cleft in the other SF1 family helicases. The binding defects of this mutant may be explained by the fact that ATP is an allosteric effector of the DNA binding activity of the enzyme. The structural integrity of the ATP binding domain of UL5 may be essential for DNA binding of the entire complex. Alternatively, the integrity of ATP binding domain could be required for the proper folding or stability of UL5. However, the fact that all mutant UL5 proteins retain the ability to interact with UL52 and UL8 suggest that the mutation in motif I does not dramatically alter the overall structure of the protein.
In subcomplexes containing the UL5 motif II mutant, UL5 was able to cross-link ssDNA and forked DNA with wild type efficiency; however, the UL52 subunit was defective in ssDNA cross-linking. This may also be because of the fact that ATP is an allosteric effector of the entire complex; perhaps the subcomplex is affected by the inability of UL5 to bind and/or hydrolyze ATP. Interestingly, despite the apparent defect in the ability of UL52 to bind ssDNA, the motif II UL5 mutant subcomplex exhibits wild type levels of primase activity but no helicase activity (39). Thus, the DNA binding activity of UL52 in mutant subcomplexes does not necessarily correlate with primase activity supporting the notion that at least some of the DNA binding ability of UL52 contributes to helicase activity, not primase.
The defects in the ability of motif III mutant UL5 proteins to bind ssDNA support the structural prediction that motif III interacts directly with ssDNA in two other SFI family helicases (30,35). This result is consistent with a previous report showing that a motif III mutant of DNA helicase II is unable to form a stable binary complex with either DNA or ATP (48). Despite the fact that motif III subcomplexes were defective for binding ssDNA, motif III and IV subcomplexes could cross-link better than wild type to the forked substrate. This result confirms our previous observation that these two mutants were able to bind forked substrates 5-to 6-fold better than wild type, respectively, in a gel shift assay (39). This result may also reflect the fact that binding of the subcomplex to the forked substrate can be stabilized even in the presence of mutant subunits with decreased affinity for ssDNA. In general, subcomplexes containing mutants defective in binding ssDNA (Fig. 9) were less defective in their ability to bind to forked substrates either by cross-linking (Fig.  10) or by gel shift assays (33), again supporting the notion that subcomplexes can be stabilized on the forked substrate.
Interestingly, Motif III and IV subcomplexes also exhibited dramatic increases, 36-and 9-fold, in primase activity (39). This rather drastic effect on primase activity may reflect a complex regulation of helicase and primase activities within the helicase-primase complex. A single heterotrimeric helicaseprimase complex bound to DNA would not be expected to carry out helicase and primase activities simultaneously; helicase is believed to move in the 5Ј to 3Ј direction along the lagging strand template, whereas primase activity necessarily occurs in the 3Ј to 5Ј direction along the template. It is possible that that there may be competition between the helicase site and the primase site for binding to DNA. Thus, mutation of the helicase motifs of the UL5 polypeptide may disrupt binding of the UL5 helicase subunit to DNA, increasing the likelihood of DNA binding at the primase active site. Alternatively, as suggested above, when helicase is active, UL52 may act as a second DNA binding subunit contributing to helicase activity, precluding it from binding to the primase recognition site. When helicase activity is abolished, as in a helicase-defective mutant, UL52 is free to bind primase recognition sites thus resulting in increased primase activity and ability to bind to forked substrates. In the assays used in this paper, a 5Ј to 3Ј helicase (like the HSV-1 helicase-primase) would be expected to contact the lower strand of the forked substrates shown in Fig. 2 during lagging strand synthesis, whereas the primase would be expected to contact the upper strand. Thus, the substrates were not optimized for looking at contacts between primase and ssDNA within the forked substrates. To test our models fully, however, it will be important to study forked substrates bearing a dIU substitution on the top strand, as well as substrates that have only a 5Ј or 3Ј tail instead of two ssDNA tails. The presence of the preferred primase recognition site on one or the other tail will also be tested.
In summary, in this paper we have taken two approaches to the study of interactions between the helicase-primase subcomplex with DNA. We have studied substrate preferences for the wild type version helicase-primase subcomplex and have analyzed binding properties of UL5 motif mutants. The DNA binding data and the behavior of mutant subcomplexes in crosslinking assays have lead to the suggestion that the UL52 subunit may play a more active role in helicase activity than had previously been thought. Our results are also consistent with the inchworm mechanism for helicase activity for the HSV-1 helicase-primase. Further experiments will be required to test these models.