The Herpes Simplex Virus Type 1 Origin-binding Protein

Origin-dependent replication of the herpes simplex virus type 1 genome requires the virally encoded origin-binding protein, UL9. UL9 binds specifically to the herpes simplex virus type 1 replication origin at two high affinity binding sites on the DNA, Boxes I and II. UL9 also has ATP-dependent DNA helicase and DNA-stimulated ATPase activities that are used to unwind the origin DNA. Origin-specific binding is mediated by the C-terminal domain (C-domain) of the enzyme. ATPase and helicase activities are mediated by the N-terminal domain (N-domain). Previous studies have shown that single-stranded DNA is a good coeffector for ATPase activity. We have analyzed several DNAs for their ability to stimulate the ATPase activity of UL9 and of a truncated UL9 protein (UL9/N) consisting only of the N-domain. We report here that duplex Box I DNA specifically and potently stimulates the ATPase activity of UL9 but not of UL9/N. We also find that removal of the C-domain significantly increases the ATPase activity of UL9. We have incorporated these results into a model for initiation in which the C-domain of UL9 serves to regulate the enzymatic activity of the N-domain.

A recurrent theme in the initiation of DNA replication is the assembly of a highly organized nucleoprotein complex at the replication origin, initiated by an origin-specific DNA-binding protein (OBP) 1 (reviewed in Refs. 1 and 2). The OBPs found in certain eukaryotic DNA viruses (simian virus 40 (SV40), bovine and papilloma viruses, and herpes simplex viruses) also actively unwind the origin DNA to nucleate assembly of a complete replication complex at the origin (3)(4)(5)(6). We have been interested in the mechanism by which the OBP of herpes simplex virus type 1 (HSV-1), UL9, coordinates its DNA binding and DNA unwinding activities to initiate replication of the HSV-1 genome at the replication origin.
The HSV-1 genome contains three highly homologous origins of replication, OriL and two copies of OriS (reviewed in Refs. 6 and 7). The minimal sequence necessary for origin-specific replication, OriS, contains three homologous inverted repeat sequences known as Boxes I, II, and III, each containing a partially overlapping pentanucleotide dyad (8 -11) (Fig. 1). Boxes I and II flank an 18-bp A ϩ T-rich region (ATR). UL9 binds cooperatively and with high affinity to Boxes I and II (8,12) and binds with very low affinity to Box III (9,11,13,14). High affinity binding by UL9 is critically dependent upon residues within the pentanucleotide dyad (8,9,(15)(16)(17). The ATR undergoes structural changes as a result of UL9 binding and is required for normal replication (15,18,19). In addition to UL9, six virally encoded proteins are required for origin-specific replication of the HSV-1 genome in vivo (6,20,21): a singlestranded DNA (ssDNA) binding protein (ICP8), a processive heterodimeric DNA polymerase, and a heterotrimeric primosome with 5Ј 3 3Ј DNA helicase activity. A generally accepted model for initiation of HSV-1 replication proposes that protein-protein interactions between Box Iand Box II-bound dimers of UL9 cause formation of a nucleoprotein core complex in which the ATR is distorted into a partially single-stranded conformation (13,15,18,(22)(23)(24)(25)(26). ICP8 binds to the distorted ATR, stabilizing the singlestranded conformation, and also interacts with UL9. This UL9-ICP8-ssDNA initiation complex opens the origin DNA, rendering it accessible to other replication proteins. The ensemble of replication proteins that subsequently assembles at the origin then carries out the subsequent unwinding, priming, and elongation events of replication. In this model, formation of a ssDNA segment at the ATR is essential for UL9 activity contributing to origin unwinding.
UL9 has ATPase and ATP-dependent 3Ј 3 5Ј helicase activities that are specifically stimulated by ICP8 but not by heterologous ssDNA-binding proteins (23,(27)(28)(29)(30). The ATPase activity has previously been reported to be greatly stimulated by the presence of ssDNA but not by the presence of duplex DNA (28,29). Helicase activity has also been found to be dependent on ssDNA; UL9 unwinds partially single-stranded DNA substrates in the presence of ATP but has not been observed to unwind fully duplex DNAs or plasmids, even if they contain origin DNA or if ICP8 is present (23,26,30). These observations further support a model in which distortion of the ATR is a prerequisite for initiation of replication.
Sequence analysis and mutagenesis of the 851-amino acid UL9 polypeptide have led to the assignment of functional domains (Fig. 2). The domain comprised of the N-terminal twothirds of the polypeptide chain (the N-domain) contains conserved ATP binding and helicase motifs that are critical for origin-specific replication (31)(32)(33). The N-domain also contains the regions critical for dimerization (12,13), cooperative binding to OriS (13,34), and interaction with the UL8 subunit of the HSV-1 primosome (35). The N-domain has been expressed independently and comprises a functional helicase (36). On the other hand, the domain comprised of the C-terminal one-third of the protein (the C-domain) is responsible for OriS-specific binding. This domain, when expressed independently, is an OriS-specific DNA-binding protein (37,38) that is monomeric both in solution and when bound to OriS (12,13,15,39). The C-domain also contains the region of interaction with ICP8 (40). Thus, the critical DNA binding and helicase functions of UL9 are present on two separable but linked domains. This structural arrangement may provide for regulation of UL9 activity, as discussed later in this report.
Although the various functions of UL9 as an OBP, an ATPase, and a helicase have been identified and characterized, the exact mechanism by which UL9 functions to unwind DNA at the origin is not well understood. The enzymatic functions of UL9 have been consistently observed to require a singlestranded or partially single-stranded DNA substrate, implying that distortion of the origin to form a ssDNA region is a prerequisite for UL9 enzymatic activity in vivo. In this study, we have more rigorously examined the stimulation of UL9 ATPase activity by DNA coeffectors. We present evidence that duplex DNA containing Box I can act to specifically and dramatically stimulate UL9 ATPase activity. We also show that this DNA is a poor coeffector for the ATPase activity of the isolated Ndomain and that the isolated N-domain is a more active enzyme than full-length UL9. The implications of these findings for the role and regulation of UL9 in unwinding origin DNA are discussed.

EXPERIMENTAL PROCEDURES
Materials-Restriction and DNA-modifying enzymes were from New England Biolabs and Life Technologies, Inc. Spodoptera frugiperda Sf9 cells were from Invitrogen. Reagents for insect cell culture were from Life Technologies, Inc. The baculovirus transfer vector (pBacPAK9), viral DNA (BacPAK6), and transfection reagent (Bacfectin) were from CLONTECH. Heparin (Sigma) was coupled to Sepharose CL-4B (Amersham Pharmacia Biotech) as described (41). DEAE-Sepharose and ATP were from Amersham Pharmacia Biotech. HEPES buffer, pepstatin A, and leupeptin were from U. S. Biochemical Corp. HEPPS buffer was from Calbiochem. Malachite green base and ammonium molybdate (VI) tetrahydrate (used in the colorimetric ATPase assay) were from Sigma, as was pyrogallol red (used in sample preparation for SDS-PAGE). Benzamidine HCl monohydrate was from ICN Biomedicals Inc. Protein A-horseradish peroxidase conjugate and 4-chloro-1-naphthol were from Bio-Rad. Plasmid pGAD-UL9t and the recombinant Autographa californica nuclear polyhedrosis virus (baculovirus) AcNPV/ UL8/H expressing HSV-1 UL8 protein were generous gifts from N. Constantin (University of Arizona). Polyclonal antiserum to UL9 was a generous gift from Dr. I. R. Lehman (Stanford University).
Oligonucleotides-Synthetic DNAs were obtained from Genosys Biotechnologies, Inc. The sequences of the oligonucleotides used as coeffectors in ATPase assays are shown in Fig. 3 Construction of Recombinant Baculovirus Expressing UL9/N-Plasmid pGAD-UL9t contains the UL9 gene derived from pVL941/UL9 (28) in pGAD-424 (CLONTECH), flanked by EcoRI and XbaI sites. A partially self-complementary DNA (5Ј-GATCCCTAACCGGTTAGG-3Ј) containing a stop codon (TAA) and BamHI ends was self-annealed and then inserted into the distal BamHI site in the UL9 gene on pGAD-UL9t to truncate the gene at amino acid 536. The 3651-bp EcoRI-XbaI DNA fragment containing the truncated UL9 gene (UL9/N) was then inserted into the baculovirus transfer vector pBacPAK9. The sequence of the entire UL9/N gene in the resultant recombinant plasmid pBP-UL9/N was confirmed by the sequencing laboratory at the University of Arizona. pBP-UL9/N was then used to generate the recombinant baculovirus AcNPV/UL9/N as described by the CLONTECH manual. Stocks of recombinant baculovirus were propagated in Sf9 cells grown in TMNFH containing 10% fetal calf serum, 50 g/ml gentamycin sulfate, and 2 g/ml fungizone (TMNFH medium) (42).
Infection of Sf9 Cells and Preparation of Infected Cell Cytosolic Extracts-A 400-ml suspension culture of Sf9 cells was grown to a cell density of 2.4 ϫ 10 6 cells/ml in TMNFH medium containing 0.1% Pluronic F-68 (TMNFH/P medium). The cells were pelleted by centrifugation at 100 ϫ g for 15 min and resuspended in fresh TMNFH/P medium. Fifty ml of AcNPV/UL9/N viral stock (at approximately 2 ϫ 10 8 plaque-forming units/ml) were added to the cells for a final concentration of 9.4 ϫ 10 6 cells/ml. After 1 h at 24°C, the cells were divided into two 2-liter roller bottles, and fresh TMNFH/P medium was added to dilute the cells to 9.4 ϫ 10 5 cells/ml. The roller bottles were placed upright in an orbital shaker at 27°C and shaken at 140 rpm with loosened caps. Seventy hours later, the cells were pelleted by centrifugation at 2000 ϫ g for 10 min at 4°C. The cells were gently resuspended in 20 mM HEPES-NaOH, pH 7.6, containing 1.0 mM DTT, 150 mM NaCl, and 5% (w/v) glycerol. Centrifugation as above yielded 2.8 g wet cells that were frozen in liquid nitrogen and stored at Ϫ80°C. The cells were subsequently thawed at 37°C, immediately resuspended in 30 ml of cold Buffer A, transferred to a chilled Dounce homogenizer, and lysed using 10 strokes with a tight pestle. The suspension was centrifuged at 300 ϫ g at 4°C for 15 min. The cytosolic extract was frozen in liquid nitrogen and stored at Ϫ80°C.
Purification of Recombinant UL9/N Protein-All purification steps were performed at 4°C. After thawing on ice, the cytosolic extract was brought to 100 mM NaCl (as determined by conductivity measurements) by dropwise addition of 4 M NaCl while stirring. The extract was cleared by centrifugation at 24,000 rpm in a Beckman Ti50 rotor for 30 min. The outlet of a 12-ml DEAE-Sepharose column was connected directly to a 2-ml heparin-Sepharose CL-4B column, and both columns were equilibrated with Buffer B. The cleared extract was loaded onto the columns at 10 -20 ml/h. The columns were washed extensively with Buffer B (about 68 ml) until A 280 returned to base line, at which point the DEAE-Sepharose column was removed. The heparin-Sepharose column was eluted with a 20-ml gradient from 100 to 800 mM NaCl in Buffer A at 10 ml/h, and 700-l fractions were collected. After removal of analytical samples from selected fractions, all fractions were frozen in liquid nitrogen and stored at Ϫ80°C. Conductivity measurements were recorded for the analytical samples, which were then prepared for SDS-PAGE using a pyrogallol red-molybdate precipitation reagent as described (43). UL9/N protein eluted at 250 -650 mM NaCl, as determined by conductivity measurements. Densitometry of peak fraction 18 (described below) showed that the full-length UL9/N protein and its degradation products comprised 26 and 23% of the total protein, respectively.
Mock Purification of UL9/N Protein-As a negative control, the above procedures for expression and purification of UL9/N protein were performed using a recombinant baculovirus strain expressing UL8 protein (AcNPV/UL8/H) in place of AcNPV/UL9/N.
Denaturing Polyacrylamide Gel Electrophoresis and Immunodetection-The protein samples were subjected to SDS-PAGE on identical 10% polyacrylamide gels that were stained with Coomassie Brilliant Blue or transferred to nitrocellulose for immunoblotting. The nitrocellulose was probed with UL9 antiserum followed by protein A-horseradish peroxidase conjugate; 4-chloro-1-naphthol was used as the substrate for color development. The Coomassie-stained gels were scanned with a Amersham Pharmacia Biotech UltraScan XL densitometer. UL9/N and its degradation products comprised approximately 49% of the total protein.
Expression and Purification of Recombinant UL9 Protein-UL9 protein was expressed in Sf9 cells and purified from nuclear extracts as described (28). The purity was 95% as determined by densitometry of a Coomassie-stained SDS-polyacrylamide gel loaded with approximately 6 g of protein.
ATPase Assays-Unless otherwise noted, the reaction mixtures contained 40 mM HEPPS buffer, pH 8.3, 2 mM ATP, 1 mM DTT, 5 mM MgCl 2 , 10% (w/v) glycerol, 100 g/ml bovine serum albumin, and varying amounts of purified UL9 or partially purified UL9/N. The DNA-dependent ATPase reactions were initiated by the addition of oligonucleotides to a final concentration of 0 -400 nM (for duplex DNAs) or 0 -800 nM (for ssDNAs) and concomitant transfer of the reaction tubes from ice to a temperature-controlled water bath. After 1 h at 37°C (UL9 reaction mixtures) or 15 min at 30°C (UL9/N reaction mixtures), the reactions were stopped by the addition of 0.75 ml of acidic molybdenum/malachite green solution (44). (Because a significant loss of UL9/N activity was observed upon extended incubation at 37°C, both the time and temperature of the assay reaction were reduced compared with the UL9 reaction. At 30°C approximately 80% of UL9/N ATPase activity was retained over 15 min. UL9/N was found to be significantly more active at 37 than 30°C, however.) Color was allowed to develop for 5 or 6 min at room temperature; color development was stopped by the addition of 100 l of 34% (w/v) sodium citrate, and the formation of inorganic phosphate was determined by measurement of A 650 . The blank contained all reaction components except DNA. To compensate for variations in enzyme concentration from run to run, an internal standard reaction was included in each run, and all measured activities were normalized relative to this standard. For UL9 assays, the standard reaction contained 572 nM BoxI/DS. For UL9/N assays, the standard reaction contained 800 nM (dT) 22 . For each DNA substrate, the normalized data from 3 to 6 replicate series were combined and then fit to the Michaelis-Menten equation using the curve fit feature of Kaleidograph 3.08 (Synergy Software). The standard errors for the curve fit parameters and the averages and standard deviations for the data were also calculated by Kaleidograph. Assays of UL9/N or UL8 column fractions were as described above for UL9/N except that the reaction mixtures were made up with or without DNA, and reactions were initiated by the addition of 0.5 l of a column fraction. Reactions were performed in duplicate in both the presence and absence of 800 nM (dT) 22 .

Duplex Box I DNA Is a Potent Activator of UL9 ATPase
Activity-We examined the stimulation of UL9 ATPase activity by various duplex and single-stranded 22-mer DNAs (Fig. 3) derived from the Box I sequence of OriS (Fig. 1). These results are summarized in Fig. 4 and Table I. Of the duplex DNAs examined, only BoxI/DS greatly stimulated UL9 ATPase activity, displaying both a high relative V max of 1.18 (corresponding to 2.8 ϫ 10 3 nmol of ATP hydrolyzed/nmol enzyme⅐h) and the highest apparent affinity (K d(app) ϭ 29 nM) ( Fig. 4a and Table  I). In contrast, duplex DNAs in which 4 bases were mutated to eliminate specific binding by UL9 (Mut/DS) or in which the base sequence was scrambled (Ran/DS) were relatively poor activators, exhibiting lower activities (V max ϭ 0.76 and 0.49, respectively) and significantly lower apparent affinities (K d(app) ϭ 690 and 240 nM, respectively). The Box I, random, and mutant ssDNAs exhibited poor to moderate stimulatory capacity (Fig. 4b). Although the upper strands exhibited approximately the same relative V max as BoxI/DS (ranging from 1.09 to 1.31), their apparent affinities were significantly reduced (by 25-51-fold). The lower strands (BoxI/L, Mut/L, and Ran/L) displayed significantly reduced relative activities (V max ϭ 0.27, 0.61, and 0.34, respectively) and intermediate affinities (reduced by 9 -21-fold). The specific and dramatic effect of duplex Box I on UL9 ATPase activity is most clearly illustrated by a comparison of V max /K d(app) for the various DNAs (Fig. 4c). This measure of activation efficiency, which is analogous to k cat /K m as a measure of catalytic efficiency, is 20 -56-fold higher for duplex Box I than for any of the other DNAs examined.
Partial Purification of Recombinant UL9/N-A recombinant protein consisting of the N-domain of UL9 (UL9/N) (Fig. 2) was partially purified from Sf9 cytosolic extracts by DEAE-Sepharose and heparin-Sepharose chromatography. A major band running at the expected molecular weight of UL9/N (60,000) was apparent on an SDS-polyacrylamide gel of the column fractions, peaking in fractions 14 -18 (Fig. 5a). This band also reacted with anti-UL9 antiserum (Fig. 5b). In addition, three moderate intensity bands of lower apparent molecular weight were visible by Coomassie staining and immunodetection with anti-UL9 antiserum, indicating that they were products of UL9/N degradation. Indeed, significant degradation of UL9/N was observed throughout the purification procedure (data not shown), despite the presence of multiple protease inhibitors. The full-length UL9/N protein and its degradation products comprised approximately 49% of the total protein. As a control, an identical purification procedure was carried out on cytosolic extracts from cells infected with a baculovirus expressing a different HSV-1 protein (UL8). Column fractions from this procedure yielded no major 60-kDa band visible by Coomassie staining and no bands visible by immunodetection with anti-UL9 antiserum (data not shown).
Adenosine Triphosphatase Activity of UL9/N-The UL9/N column fractions were assayed for ATPase activity in the presence and absence of 800 nM (dT) 22 . Both activities were found to copurify with the immunoreactive 60-kDa polypeptide, peaking in fractions 14 -18 (Fig. 6a). The addition of (dT) 22 was found to stimulate the ATPase activity of the peak fractions by 10 -16fold. Quantitation of the amount of full-length UL9/N in fraction 18 by densitometry allowed the DNA-dependent ATPase activity of UL9/N to be estimated at 3.6 Ϯ 1.5 ϫ 10 4 nmol of ATP hydrolyzed per nmol of UL9/N polypeptide⅐h; the specific activity of UL9/N is thus 15-fold higher on a molar basis than that of UL9. UL9/N appeared to be extremely labile; in addition to the degradation described above, a significant loss of ATPase activity was observed upon incubation of the column fractions at 30 or 37°C (data not shown). The control UL8 column fractions yielded no detectable ATPase activity in the presence or absence of DNA (Fig. 6b).
Duplex Box I DNA Is a Poor Activator of UL9/N ATPase Activity-The results of UL9/N ATPase activity assays carried out in the presence of various DNAs are summarized in Fig. 7 and Table II. The single-stranded and duplex mutant DNAs yielded the highest relative activities overall, with V max ranging from 0.97 to 1.10 (Fig. 7, a and b). In contrast, the singlestranded and duplex Box I DNAs yielded the lowest activities overall, with V max ranging from 0.33 to 0.47. The singlestranded and duplex random DNAs were intermediate in activity but displayed the highest apparent affinities (with K d(app) ranging from 36 to 49 nM), yielding the highest activation efficiencies overall (Fig. 7c). The apparent affinities of the mutant DNAs were lowest, with K d(app) ranging from 110 nM (for Mut/DS) to 270 nM (for Mut/U and Mut/L), whereas the apparent affinities of the Box I DNAs were intermediate, with K d(app) ranging from 73 (for BoxI/DS) to 123 nM (for BoxI/U). Single-stranded and duplex random DNAs exhibited very little difference in their activation efficiencies. In contrast, the higher affinity observed for the duplex Box I and mutant DNAs  Table I). These parameters were derived from fits of the data to the Michaelis-Menten equation by the curve-fit feature of Kaleidograph, as described under "Experimental Procedures."

TABLE I Kinetic parameters for UL9 ATPase activity in the presence of various duplex and single-stranded oligodeoxyribonucleotides
The formation of inorganic phosphate upon incubation of purified recombinant UL9 in the presence of DNA and 2 mM ATP was determined as described under "Experimental Procedures." The resulting data, which are summarized in Fig. 4, a and b,

DISCUSSION
As the initiator of replication at the HSV-1 replication origin, UL9 must accomplish two distinct tasks. First, it must accurately locate the replication origin, and second, it must initiate replication at this site. Clearly, UL9 can locate the replication origin by virtue of its high affinities for Boxes I and Box II.
There is much evidence to indicate that cooperative binding of UL9 at Boxes I and II results in assembly of an OriS-UL9 nucleoprotein core complex in which the ATR structure is distorted (13,15,18,(23)(24)(25)(26). The role of UL9 in the subsequent events in the initiation process is not well understood, however. Are additional molecules of UL9 recruited to the distorted ATR to act as helicases, as suggested by the finding that UL9 acts as a stoichiometric helicase (45)? If additional UL9 molecules are not required, what is the signal for the origin-bound dimers to switch from a binding mode to an unwinding mode? Formation of the complex might serve as the signal for the bound UL9 molecules to become active helicases or, alternatively, might serve as the signal for additional UL9 molecules to begin unwinding the DNA at the origin. In this report we have demonstrated that UL9 ATPase activity is specifically activated by duplex Box I DNA. This specific activation of ATPase activity by Box I, which is not a substrate for helicase activity (26), indicates that UL9 dimers bound to the origin may be enzymatically functional as ATPases but not as helicases. We speculate that this phenomenon may be part of a regulatory mechanism for controlling UL9 unwinding activity at the origin, as described below.
We examined several duplex and single-stranded 22-mer DNAs for their ability to act as coeffectors of the DNA-stimulated ATPase activity of UL9. We examined upper and lower strands of Box I as well as the Box I duplex. We also examined the component single strands and duplex forms of two variants of Box I as follows: (a) a mutant form, in which 4 base changes were made at the sites that are known to be critical for high affinity binding of UL9 to Box I (15,17), and (b) a random form, in which the base composition remained the same but the sequence was randomized. Of all the DNAs examined, BoxI/DS was by far the most effective at stimulating ATPase activity of UL9, displaying a relative activation efficiency at least 20-fold greater than any of the other duplex or ssDNAs (Fig. 4c). One salient feature of this potent stimulatory activity is that the apparent affinity of UL9 for BoxI/DS is much higher (by 9 -51fold) than for the other DNAs examined. Double-stranded Box I acts therefore as both the preferred ligand for UL9, as shown previously by binding studies, and as its preferred coeffector for ATPase activity.
We also examined the above DNAs for their ability to stimulate the ATPase activity of the isolated N-domain of UL9 (UL9/N). This protein, which consists of amino acids 1-536 of UL9 and contains all of the helicase and ATP-binding site motifs, was found to be extremely labile when partially purified from recombinant baculovirus-infected cells. The extreme lability of UL9/N prohibited its further purification but was not unexpected, as removal of a large domain of a globular protein is likely to expose a previously solvent-protected hydrophobic interface. We found that UL9/N is a weak DNA-independent ATPase and that its ATPase activity is increased 10 -16-fold upon addition of 800 nM (dT) 22 . This result is consistent with a previous report that UL9/N is a DNA-stimulated ATPase and ATP-dependent helicase, with an approximately 5-10-fold increase in ATPase activity in the presence of 2 g/ml activated calf thymus DNA (36).
The above result demonstrates that removal of the C-terminal domain of UL9 does not eliminate the ability of the enzyme to act as a DNA-stimulated ATPase. It does, however, dramatically alter the specificity characteristics of this stimulation. Whereas BoxI/DS is a specific and potent coeffector for UL9 ATPase activity, it is a very poor coeffector for UL9/N ATPase activity. In fact, the duplex and single-stranded Box I and mutant sequences are all poor coeffectors in terms of activation efficiencies, with the single-stranded sequences being approximately half as effective as the duplex DNAs. Interestingly, the most potent activators of UL9/N ATPase activity, in terms of both apparent affinity and activation efficiency, are the singlestranded and duplex random DNAs. These sequences exhibit apparent affinities for UL9/N (K d(app) ϭ 36 -49 nM) of the same order of magnitude as that of BoxI/DS for UL9 (K d(app) ϭ 29 nM). The overall pattern of activation is quite different from that of UL9; UL9/N is activated most effectively by those DNAs (Ran/DS, Ran/U, and Ran/L) that least resemble Box I in sequence. UL9/N also exhibits a much higher DNA-dependent Error bars represent the propagated standard error for V max /K d(app) , which was derived from the standard errors for V max and K d(app) (see Table II). These parameters were derived from fits of the data to the Michaelis-Menten equation by the curve-fit feature of Kaleidograph, as described under "Experimental Procedures."

TABLE II
Kinetic parameters for UL9/N ATPase activity in the presence of various duplex and single-stranded oligodeoxyribonucleotides The formation of inorganic phosphate upon incubation of partially purified recombinant UL9/N in the presence of DNA and 2 mM ATP was determined as described under "Experimental Procedures." The resulting data, which are summarized in Fig. 7, a and  ATPase activity than UL9, even at a reduced assay temperature. By using our most conservative estimates, we find that UL9/N has a molar specific activity (obtained at 30°C with (dT) 22 ) of 2.5-15-fold greater than that of UL9 (obtained at 37°C with BoxI/DS). These results demonstrate that UL9 has two discrete DNAbinding sites with different binding specificities. One site, located on the C-domain, specifically binds Box I (and Box II) DNA, as shown previously, and is required for the Box I-specific stimulation of the ATPase activity of full-length UL9. The second site, located on the N-domain, mediates the activation of the truncated enzyme, UL9/N, upon binding of duplex or ssD-NAs that do not resemble Box I. The different DNA binding characteristics of the two domains may serve complementary purposes in origin-specific replication. First, appreciable enzymatic activity of the full-length enzyme can begin only upon binding to Box I (or possibly to Box II), which ensures that replication initiates at the correct site on the genome. Second, the preference of the N-domain for non-Box I ligands may assist the UL9 molecule in binding to the origin in the correct orientation (i.e. with Box I bound to the C-domain) rather than in the opposite, nonfunctional orientation (i.e. with Box I bound to the N-domain). The increased activity of UL9/N relative to UL9 suggests that the presence of the C-domain inhibits the enzymatic activity of the N-domain. This proposition is consistent with that of Simonsson et al. (16) who have previously suggested that the C-domain actively controls helicase activity and is not simply a passive anchor to DNA.
Integrating previous models (16,18,25,26) with our results, we propose the tentative model shown in Fig. 8. In this model, the C-domain of UL9 acts to regulate the potential ATPase activity of the N-domain to ensure accurate and efficient replication. In this scheme, UL9 is a poor ATPase in the absence of DNA because the N-domain is trapped in an enzymatically incompetent conformation (Fig. 8a). When the UL9 C-domain binds to Box I, the conformation of the enzyme changes to relieve the inhibition of the enzymatic activity of the N-domain (Fig. 8b). At this point ATP hydrolysis can occur, but the enzyme cannot translocate along the DNA because it is locked into place on the DNA by the tight association between the C-domain and the DNA. Since the enzyme cannot translocate, it cannot use the energy derived from ATP hydrolysis to unwind the DNA. At some point during subsequent assembly of the complete initiation complex, the enzyme undergoes a second conformational change (Fig. 8c). This second conformational change disrupts the tight UL9-Box I interaction, thus eliminating the block to translocation, and allows unwinding to begin. The most likely candidate for the trigger of this second conformational change is the association of ICP8 with the Cdomain of UL9, promoted by the availability of a ssDNA region that also interacts with ICP8. Other possible events triggering this second conformational change could include the N-domainmediated association of Box I-and Box II-bound UL9 dimers and/or binding of an as yet unidentified cellular factor to UL9. a, in the absence of DNA, UL9 exists in an enzymatically incompetent conformation. b, the binding of duplex Box I DNA to UL9 causes a conformational change in the enzyme that enables it to hydrolyze ATP efficiently. The tight binding of the C-domains to Box I prevents translocation and thus no unwinding occurs despite the hydrolysis of ATP. The conformational change required for ATPase activity is not effectively promoted by non-Box I DNA. c, subsequent events in the assembly of the complete initiation complex, such as binding of ICP8 to the C-domains, cause a second conformational change in UL9 that disrupts the binding of the C-domains to Box I. With the block to translocation removed, unwinding occurs. Helicase activity is presumed to involve binding of ssDNA to the DNA-binding site on the N-domain. In this model, high affinity binding of UL9 to the origin is inhibitory to the unwinding activity of UL9, and this inhibition can be relieved by the subsequent association of another replication factor (such as ICP8) with the origin-bound UL9. These predictions are supported by several results of Lee and Lehman (26,46). First, they have observed that UL9 can specifically unwind a partial duplex Box I substrate only in the presence of ICP8 and that mutation of the DNA substrate to prevent high affinity binding of UL9 results in significant helicase activity in the absence of ICP8 (26). We surmise that in the case of the Box I substrate, the binding of ICP8 to UL9 triggers the conformational change in UL9 which disrupts its interaction with the DNA, thus allowing unwinding to occur. We surmise that in the case of the mutated substrate, the above-mentioned ICP8-mediated conformational change is not required for helicase activity because the enzyme is not locked into place on the DNA substrate. These interpretations are further supported by the recent observations of Lee and Lehman (46) that ICP8 weakens the interaction of UL9 with the partial duplex Box I substrate. Furthermore, they report that incubation of UL9 with an antibody that prevents association with ICP8 inhibits unwinding of the Box I substrate in the presence of ICP8, and that preincubation of UL9 with Box I before addition of the antibody increases the inhibitory effect (46). This result suggests that binding of UL9 to Box I causes a significant conformational change in the protein, as we have suggested here.
The enzymatic characteristics of UL9/N can be explained in terms of our model (Fig. 9). Like full-length UL9, UL9/N is a poor ATPase in the absence of bound DNA (Fig. 9a). Binding of non-Box I DNA is proposed to convert UL9/N to a more active conformation (Fig. 9b), just as binding of Box I DNA is proposed to convert UL9 to a more active conformation (Fig. 8b). The increased ATPase activity of UL9/N relative to the full-length enzyme may be due to greater accessibility of ligand-binding sites in the absence of the C-domain. Alternatively, UL9/N may assume an optimally active conformational state that is inaccessible to the full-length enzyme.
In this work we have shown that the ATPase activity of UL9 is specifically and dramatically stimulated by duplex DNA containing Box I and that the C-domain is responsible for this stimulation. We have integrated this result into a model for initiation that proposes that (a) the differential preferences of the C-and N-domains for Box I DNA aid in orienting the UL9 molecule at the origin, and (b) both the correct binding of UL9 to the origin and the association of an additional replication factor with UL9 are necessary to switch the enzyme to a conformation that can unwind the origin DNA. These stepwise changes in conformation could serve to coordinate the enzymatic activities of UL9 with the various associative events and structural changes involving UL9 during initiation. The possible advantage of ATP hydrolysis in the absence of unwinding activity is not clear; perhaps a stable initiation complex can only form if the UL9 molecules are in an enzymatically competent conformation. We are currently conducting experiments to examine further the requirements for enzyme activation, including the effects of Boxes II and III, and to identify additional factors and/or associative events required for helicase activity of Box I-bound UL9.
The specific activation of ATPase activity by binding of origin DNA has not thus far been observed for the OBPs of other eukaryotic viruses such as SV40 and bovine papilloma virus. In light of the possibility that the specific activation of UL9 by Box I may be an example of a general regulatory mechanism, it would be interesting to determine the effect of origin sequence binding on the enzymatic activities of the OBPs in these other viral systems.