Biochemical analysis of mutant T7 primase/helicase proteins defective in DNA binding, nucleotide hydrolysis, and the coupling of hydrolysis with DNA unwinding.

We characterized nine helicase-deficient mutants of bacteriophage T7 helicase-primase protein (4A') prepared by random mutagenesis as reported in the accompanying paper (Rosenberg, A. H., Griffin, K., Washington, M. T., Patel, S. S., and Studier, F. W. (1996) J. Biol. Chem. 271, 26819-26824). Mutants were selected from each of the helicase-conserved motifs for detailed analysis to understand better their function. In agreement with the in vivo results, the mutants were defective in helicase activity but were active in primase function. dTTP hydrolysis, DNA binding, and hexamer formation were examined. Three classes of defective mutants were observed. Group A mutants (E348K, D424N, and S496F), defective in dTTP hydrolysis, lie in motifs 1a, 2, and 4 and are possibly involved in NTP binding/hydrolysis. Group B mutants (R487C and G488D), defective in DNA binding, lie in motif 4 and are responsible directly or indirectly for DNA binding. Group C mutants (G116D, A257T, S345F, and G451E) were not defective in any of the activities except the helicase function. These mutants, scattered throughout the protein, appear defective in coupling dTTPase activity to helicase function. Secondary structural predictions of 4A' and DnaB helicases resemble the known structures of RecA and F1-ATPase enzymes. Alignment shows a striking correlation in the positions of the amino acids that interact with NTP and DNA.

DNA helicases catalyze unwinding of duplex DNA to singlestranded DNA, a process energetically coupled to NTP hydrolysis. Helicases are an important class of proteins required in almost all the processes of DNA and RNA metabolism. Recently, a large number of putative helicases have been identified mainly from amino acid sequence homologies. Known helicases have homologous amino acid sequences, confined to small regions in the protein, that are used as signature motifs for identifying helicases (1,2). Since a high resolution structure of a helicase is not known at the present time, the roles of these conserved motifs remain largely unclear.
We are studying the mechanism of bacteriophage T7 DNA helicase, which is involved in DNA replication. Bacteriophage T7 is a model system used to study the detailed mechanisms of DNA replication because of its simplicity. A minimum of two proteins, T7 DNA polymerase and T7 DNA primase/helicase, have been shown to reconstitute duplex DNA replication in vitro. T7 gene 4 encodes the two primase/helicase proteins, 4A and 4B (3). The full-length 63-kDa 4A protein has both helicase and primase activities, whereas the shorter 56-kDa 4B protein that begins at a second initiation codon has only helicase activity (4,5). The helicase activity unwinds double-stranded DNA during leading strand DNA replication, and the primase catalyzes synthesis of tetraribonucleotides that serve as primers for lagging strand DNA replication (4).
T7 DNA helicase belongs to the general class of hexameric helicases. Its low resolution structure, studied in detail using electron microscopy and image averaging, shows that both 4AЈ and 4B proteins form ring-shaped hexamers, and the ssDNA 1 binds through the central hole of the ring (6). This mode of DNA binding results in protection of about 25 bases of ssDNA from nuclease digestion (7) and likely confers high processivity to DNA unwinding. The helicase forms hexamers only in the presence of nucleotide ligands such as dTDP, dTTP, ATP, and dTMP-PCP (8,9). DNA binds tightly only to the hexameric species and requires the presence of dTTP or dTMP-PCP (7). The various activities of the helicase protein such as NTP binding/hydrolysis, protein oligomerization, and DNA binding are linked (9). Therefore, it is likely that amino acids responsible for these activities also may be close in space or perhaps lie in the same motif.
Regions of T7 gene 4 protein show sequence homology to several bacterial and bacteriophage primase/helicase and primase-related helicases that belong to the DnaB family of helicases. Comparison of amino acid sequences in this family of helicases has led to the identification of five conserved motifs denoted 1A, 1a, 2B, 3, and 4 (2). Conserved motif 1A is the well known GXXGXGKT/S sequence found in numerous nucleotidebinding proteins and shown in many ATPases to be involved in binding the diphosphate or the triphosphate moiety of nucleotides (10). Motif 2B is most likely the conserved motif B sequence involved in binding nucleotide via Mg 2ϩ confirmed from this study. The remaining three motifs, 1a, 3, 4 have unknown functions.
Site-directed mutagenesis has been used in the past to probe the function of several of the conserved motifs. Mutations have been made in the 1A motif, GXGKS sequence. Replacement of lysine 318 in this motif with an alanine (11) and replacement of glycine 317 with valine and lysine 318 with methionine (12) result in reduction of the dTTP hydrolysis activity and elimination of the helicase activity. Subsequent mutagenesis efforts focused on motif 4. H475A and D485G mutant proteins display reduced oligomerization ability particularly in the absence of nucleotide (13). These mutants also have reduced dTTPase (lower k cat /K m ), ssDNA binding, and duplex DNA unwinding activities; thus, the precise role of this motif was not clear.
In the accompanying paper (14), random mutagenesis and genetic selection were used to obtain lethal mutants in the cloned T7 primase/helicase gene (gene 4AЈ). The 76 mutants selected were distributed throughout much of the protein, and most of them provided sufficient primase function for T7 growth but were defective in helicase function. Many of the mutants in the C-terminal half of the protein lie in or close to conserved helicase motifs, and the mutants should be useful for understanding the function of these motifs. In this paper, we present the biochemical characterization of nine of the mutant proteins.
Proteins and Enzymes-4AЈ and the mutant 4AЈ proteins were purified as described below. T4 polynucleotide kinase (10 units/l) used to 5Ј-radiolabel oligodeoxynucleotides was purchased from Boehringer Mannheim. T7 DNA polymerase (exo Ϫ and exo ϩ ) and Escherichia coli thioredoxin proteins were purified as described (19).
Determination of Protein Concentration-Protein concentrations were determined by spectroscopic measurements at 280 nM in TE buffer, 8 M urea using the extinction coefficient of 4AЈ equal to 76,100 M Ϫ1 cm Ϫ1 (20). Protein concentrations were also determined by the Bio-Rad Protein Assay using bovine serum albumin as a standard. Both methods gave consistent values of protein concentrations.
Measurement of Primase Activity-Primase activity was measured in reaction buffer containing 50 mM Tris acetate, pH 7.5, 50 mM sodium acetate, 10 mM magnesium acetate, 1 mM DTT, and 0.1 mg/ml bovine serum albumin. Wild-type 4AЈ or the mutant protein (1 M) was incubated at 22°C for 2 min in reaction buffer with 500 M ATP, 200 M CTP, [␣-32 P]CTP (2 Ci), and 6 mM dTTP. Primer synthesis was initiated by addition of 2 M 60-mer DNA, which contains the primase recognition sequence 3Ј-CTGGT. Reactions were quenched with 1 N HCl and chloroform at various time intervals, from 0 to 60 min. After neutralization to pH 7 with 1 M Tris base, 1 M NaOH, the samples were mixed with an equal volume of sequencing gel loading buffer (95% formamide and 0.05% bromphenol blue) and applied to a 25% polyacrylamide, 3% bisacrylamide, 3 M urea sequencing gel (Bio-Rad apparatus; 35 ϫ 43 cm ϫ 0.2 mm thickness). The gel was run at 100 watts until the tracking dye was halfway through the gel. The unreacted CTP and the RNA products were quantitated with the PhosphorImager instrument (Molecular Dynamics).
Measurement of the Helicase Activity Using Fork Unwinding Assay-The 5Ј-tail (5 M) was radiolabeled with 32 P using [␥-32 P]ATP (50 Ci) and T4 polynucleotide kinase. Excess [␥-32 P]ATP was removed by Bio-Gel P-30 spin gel filtration (Bio-Rad). The 5Ј-tail (1 M) was incubated with 3Ј-tail (1 M) in annealing buffer (50 mM Tris acetate, 50 mM sodium acetate, and 100 mM NaCl). The solution was rapidly heated to 95°C for 5 min and slowly cooled to room temperature over several hours.
The ssM13 DNA/60-mer fork DNA was prepared by a similar procedure. The 60-mer was radiolabeled and annealed to ssM13 DNA to form 30 bp of duplex (M13 positions 6202-6235) and 30-nucleotide nonhomologous 3Ј-tail (20). The M13 DNA (50 nM) and the 60-mer (10 nM) annealing mix were heated to 95°C for 15 min and slowly cooled to 37°C. It was incubated at 37°C for 24 h to complete the annealing.
Measurement of Helicase Activity-The helicase activity was measured at 22°C in a buffer consisting of 50 mM Tris-Cl, pH 7.5, 50 mM NaCl, 10 mM MgCl 2 , and 10% glycerol. 4AЈ or a mutant 4AЈ (2.5 M) was incubated for 2 min with 5 mM dTTP. The reaction was initiated by adding 5Ј-32 P-labeled fork (2 nM). Low DNA concentration was used to reduce the rate of reannealing of unwound DNAs. At time points from 0 to 60 min, reaction aliquots were mixed with SDS Quench dye (3% SDS, 100 mM EDTA, 40% glycerol, and 0.1% bromphenol blue), and the samples were immediately loaded on a 7% native polyacrylamide gel prepared in TBE buffer and run at 8°C at 40 V. The remaining fork DNA and unwound ssDNA were quantitated on the PhosphorImager instrument (Molecular Dynamics). The quantitated DNAs at various time points were normalized to the heat-denatured sample, and the zero point (no protein) was subtracted from each.
The same procedure was used for the M13/60-mer fork DNA substrate. This reaction was initiated by the addition of the M13/60-mer fork DNA to a final concentration of 5 nM M13 DNA (1 nM of M13 with the radiolabeled 60-mer annealed).
Measurement of Primase and Helicase Activity During DNA Synthesis-The primase and helicase activities were measured at 22°C using the RNA-primed DNA synthesis assay described by Nakai and Richardson (21) with the following modifications. The reactions contained 500 M ATP, CTP, dATP, dGTP, and dCTP (2 Ci of [␣-32 P]dCTP), and 6 mM dTTP, 1 M T7 DNA polymerase exo Ϫ complexed with 5 M thioredoxin, and 1 M of 4AЈ or a 4AЈ mutant, and 20 nM single-stranded M13 DNA. Reactions were quenched with equal volume of 0.5 M EDTA. The radiolabeled DNA products were resolved on a 0.7% alkaline agarose gel (400 mA, 10 h) as described (22). After fixing in 7% trichloroacetic acid and drying, the DNA products were analyzed using the PhosphorImager instrument.
Measurement of dTTP Hydrolysis-The dTTPase activity of the wildtype 4AЈ (1 M) and the mutant proteins (1 M) was measured at 22°C. Proteins were preincubated with 100 M dTTP for 10 min with or without M13 DNA (50 nM). The reactions were initiated by adding a mixture of 5 mM dTTP and [␣-32 P]dTTP (2 Ci). At various time intervals, aliquots were quenched with an equal volume of 0.5 M EDTA. The unreacted dTTP and the hydrolysis product, dTDP, were separated on polyethyleneimine-cellulose thin layer chromatography plates (polyethyleneimine-cellulose TLC plates, Whatman) using 0.3 M potassium phosphate, pH 3.4, as the running buffer. The separated dTTP and dTDP were quantitated using the Molecular Dynamics Phosphor-Imager instrument.
Native PAGE to Assay for DNA Binding and Hexamer Formation-Hexamer formation and DNA binding activities were assayed by native PAGE (6%) containing MgdTMP-PCP as described (7). The protein samples (10 l) contained 12 M 4AЈ or mutant protein, 2 M 5Јradiolabeled 30-mer DNA, 1 mM dTMP-PCP in binding buffer (50 mM Tris acetate, pH 7.5, 10 mM magnesium acetate, 5 mM sodium acetate, and 1 mM DTT). Native gel loading buffer (2 l of 50% glycerol, 0.25% bromphenol blue) was added to each sample, and the gel was run at constant current (18 mA) for 2 h. The protein species were detected by Coomassie Blue staining, and the DNA-bound species were observed by using the PhosphorImager instrument. Hexamer formation in the absence of DNA was also assayed by the same native gel experiment. The running conditions and sample composition were identical except that 30-mer DNA was omitted.
Nitrocellulose-DEAE DNA Binding Assay-DNA binding by nitrocellulose/DEAE filter binding assay was performed as described (7). Radiolabeled 30-mer (1 M) was mixed with increasing 4AЈ or the mutant protein (1-20 M). The buffer contained 50 mM Tris acetate, pH 7.5, 10 mM magnesium acetate, 5 mM sodium acetate, 1 mM DTT, and 1 mM TMP-PCP. After 20 min incubation, samples were filtered through the membrane assembly and washed with 100 l of buffer. The proteinbound DNA on the nitrocellulose membrane and the free DNA on the DEAE membrane were quantitated using the PhosphorImager instrument.

Expression and Purification of the 4AЈ Mutant Proteins-
Representative mutants were chosen from the N terminus half of 4AЈ (the presumed primase domain), the linking region between the primase and helicase domains, and the C terminus helicase domain for purification and biochemical analysis. Mutants from the C-terminal helicase domain were chosen from each of the conserved helicase motifs, identified from amino acid sequence homology between the DnaB family of helicases ( Fig. 1), with the hope that the biochemical activities of these mutants will lead to a better understanding of the role of these motifs. The selected 4AЈ mutants were overexpressed under the T7lac promoter in BL21 (DE3) cells using the T7 expression system (18). The level of overexpression of all the mutants was approximately two to five times greater than that of wild-type 4AЈ protein. The proteins were purified to greater than 95% purity as judged by SDS-PAGE (Fig. 2) using the modified 4AЈ purification scheme. The 4AЈ mutant proteins behaved similarly to the wild-type protein during the entire purification process.
Primase Activity of the 4AЈ Mutant Proteins-Since all the mutants complemented T7⌬4A phage (14), it suggested that the mutants either have sufficient primase activity to support phage growth or are capable of restoring the primase activity to 4B protein. The primase activity of the mutant proteins was assayed by measuring the steady-state kinetics of RNA primer synthesis. A 60-mer ssDNA containing the primase recognition sequence 3Ј-CTGGG/T was used as the template. The primase synthesizes 2-to 4-mer RNA products complementary to sequence 3Ј-TGGT and requires the presence of either dTTP or dTMP-PCP, which promote stable hexamer formation and DNA binding. At 22°C in the presence of dTTP, 4AЈ synthesizes 2-to 4-mer RNAs with a steady-state rate constant of 3.3 ϫ 10 Ϫ3 s Ϫ1 , and 26% of the RNA products synthesized by 4AЈ were full-length 4-mer products. All the mutant proteins were competent in their primase function (Table I). The rates ranged from close to wild-type (3.5 ϫ 10 Ϫ3 s Ϫ1 ) for D424N mutant to about 10-fold lower (3.9 ϫ 10 Ϫ4 s Ϫ1 ) for G488D mutant protein. Similarly, the processivities ranged from close to wild-type (92% of the wild-type) for S345F mutant to 2-fold lower for D424N (46% of wild-type). The RNA-primed DNA synthesis assay also showed that the mutant proteins are capable of synthesizing RNA primers that support DNA synthesis by T7 DNA polymerase (Fig. 4).
The mutants with low primase activity, R487C, G488D, and S496F (16, 12, and 26% of 4AЈ, respectively), all lie in the conserved helicase motif 4. Interestingly, this low primase activity is still enough to complement the T7⌬4A phage in vivo (14). The lower primase activity of these mutants we believe is due to their defect in DNA binding as shown below. Consistent with this, when the primase activity was remeasured in the presence of a higher concentration of the template DNA, the activity of R487C and S496F increased to 40 and 50% of 4AЈ. The activity of G488D, however, remained low at 10%. Overall, there does not appear to be a severe defect in the primase function in any of the mutant proteins, consistent with the ability of the mutant proteins to provide primase activity in vivo (14).
Helicase Activity of the 4AЈ Mutant Proteins-We have used three different DNA substrates to assay for the helicase activity of the mutant proteins. Two assays measured the intrinsic helicase activity of the protein, and one assay measured the helicase activity in the presence of DNA polymerase. T7 DNA helicase requires two noncomplementary ssDNA tails at one end of a duplex DNA (fork DNA) to initiate DNA strand sepa- Nine mutants selected for biochemical characterization were purified ("Experimental Procedures") and analyzed by 8% SDS-PAGE. Lanes 1 and 10 contain the high range molecular mass standards (Life Technologies, Inc.). Lanes 2-9 contain 2.5 g each of 4AЈ, G116D, A257T, S345F, G451E, R487C, G488D, and S496F proteins, respectively. The 4AЈ mutants are greater than 95% pure and have the same mobility as the wild-type 4AЈ protein.
ration. In one assay, we used a synthetic fork DNA prepared by annealing two 60-mer ssDNAs that form 33 bp of duplex region and 27-nucleotide 5Ј and 3Ј nonhomologous DNA tails at one end. This is the most direct method for measuring the intrinsic helicase activity. In the second assay, the DNA substrate was prepared by annealing a 60-mer to ssM13 DNA, which resulted in 30-bp duplex region. The ssM13 DNA provided the 5Ј ssDNA tail, and the 30 nucleotides of the 60-mer provides the noncomplementary 3Ј-tail. The M13/60-mer fork DNA substrate was used to test if some of the mutant proteins needed a "running start" or needed to translocate on ssDNA prior to DNA unwinding. The third assay consisted of measuring strand displacement coupled to DNA synthesis by T7 DNA polymerase.
Oligo-fork Unwinding Assay-Helicase activity was measured using radiolabeled fork DNA at 22°C in the presence of 5 mM dTTP. Unwound ssDNA strands were resolved from the fork DNA using native PAGE, as shown in Fig. 3. The kinetic assay showed that 4AЈ protein completely unwound the fork DNA in 2 min, the shortest time point used in the assay. However, none of the mutant proteins showed any DNA unwinding activity, even after 1 h of reaction, indicating that the mutants are severely defective in helicase function. Since reannealing of the fork at 1 nM DNA concentration occurs with a rate of 2 ϫ 10 Ϫ4 s Ϫ1 or with a half-time of 70 min, unwinding occurring with a rate comparable with reannealing would be barely detectable. Therefore, relative to the unwinding rate of 4AЈ equal to 3.2 bp/s at 18°C, 2 we estimate that the helicase activity of the mutants is at least 600-fold lower than 4AЈ.
RNA-primed DNA Synthesis and Strand Separation Assay for Helicase Activity-RNA-primed DNA synthesis assay was used to investigate the ability of the mutant proteins to provide helicase activity to support DNA synthesis by T7 DNA polymerase. In this assay, ssM13 was used as the template. Since all the mutant proteins had primase activity, DNA products up to 7 kb in length can be formed by elongation of RNA primers by T7 DNA polymerase (Fig. 4). The polymerase does not perform strand displacement DNA synthesis (see control reaction in Fig. 4), and it needs the helicase activity to synthesize DNA products longer than 7 kb. These longer DNA products (Ͼ40 kb) are formed in reactions containing 4AЈ but not present in reactions containing some of the mutant proteins. A few mutant proteins, however, did show helicase activity in this assay. These mutant proteins included the ones in the N-terminal domain, the linker region, and some in the C-terminal domain, G116D (N-terminal), A257T (linker region), S345F (motif 1a), and G451E (motif 3). The mutant proteins that showed no helicase activity are all in the C-terminal domain, D424N (motif 2B), R487C, G488D, and S496F (all in motif 4). Note that the latter three are also the ones that have reduced primase activity.
By quantitating the amount of the Ͼ7-kb products, we can determine the relative helicase activity of the mutants. Note that this is a very rough estimate of the defect in helicase activity, as this helicase assay is indirect and not quantitative. The mutant proteins that showed helicase activities, G116D, A257T, S345F, and G451E, have 45, 46, 25, and 16% activity, respectively, relative to wild-type 4AЈ. The above results suggest that T7 DNA polymerase is able to partially restore the helicase activity of at least some mutant proteins. Since these 2 P. Ahnert and S. S. Patel, manuscript in preparation.  mutants are scattered throughout the primary sequence of the protein, it is difficult to postulate the nature of this restoration in helicase activity. More detailed biochemical analysis of the mutant proteins will be needed to fully understand the result.
M13/Oligodeoxynucleotide Helicase Assay-It is clear that mutants, D424N (motif 2B), R487C, G488D, and S496F (all in motif 4) had no helicase activity by both assays described above. The mutants G116D (N terminus), A257T (linker region), S345F (motif 1a), G451E (motif 3) show no activity with the small fork DNA unwinding substrate but showed measurable helicase activity in the presence of the polymerase protein.
To understand the different results, we measured the helicase activity of the four mutants using a third assay. One potential problem with using the small fork DNA as the substrate can be the short ssDNA tail region available to initiate DNA unwinding. The other problem might be the reannealing of the unwound DNA which would preclude detection of low levels of helicase activity. The ssM13/60-mer fork DNA substrate should circumvent both of these problems. This assay is, however, not quantitative and does not provide the intrinsic rate of DNA unwinding, as unwinding is dependent on a number of steps prior to strand separation. Fig. 5 shows the helicase activity of 4AЈ, G116D, A257T, S345F, and G451E using the ssM13/60-mer DNA substrate. 4AЈ unwound the 60-mer primer with a rate constant of 0.06 bp/s Ϫ1 . Note that the measured unwinding rate of wild-type 4AЈ using this assay was about 50-fold lower than with the small fork DNA unwinding assay. Mutant proteins, G116D (N-terminal) and G451E (motif 3), showed no detectable helicase activity even in this assay. But the mutant proteins A257T (linker region) and S345F (motif 1a) did show detectable unwinding with a rate 10-fold slower (0.006 bp/s Ϫ1 ) than 4AЈ. Detection of helicase activity by this assay as compared with the small fork assay suggests that these mutant proteins may need a longer ssDNA 5Ј-tail to initiate unwinding.
dTTP Hydrolysis Activity of the 4AЈ Mutant Proteins-Nucleotide triphosphate binding and hydrolysis are necessary for the helicase activity. Since all 4AЈ mutants had significant defects in their helicase function, we have investigated the dTTPase activity of all the mutant proteins. (T7 DNA helicase prefers dTTP as the nucleotide substrate for its helicase activity (23).) The dTTPase activity of the mutant proteins was measured both in the absence and in the presence of ssM13 DNA. It is necessary to measure the DNA-independent dTTPase activity as this provides information about the mutant's intrinsic ability to bind and hydrolyze dTTP. The dTTPase activity in the presence of ssM13 DNA is a measure of the DNA-stimulated activity of the mutant protein. To estimate the k cat value, all the assays have been carried out at very high dTTP concentrations (5 mM). The 4AЈ dTTPase K m in the absence of DNA was 5 M and in the presence of DNA about 100 M at 22°C (data not shown). Table II lists the dTTPase activities of the mutant proteins. To our surprise, several mutant proteins including G116D (Nterminal), A257T (linker region), S345F (motif 1a), and G451E (motif 3) hydrolyzed dTTP in the absence of DNA with rate constants about 10-fold higher than 4AЈ. The reason for the increased intrinsic rate of dTTPase activity is unclear but suggestive of an uncoupling between NTPase and DNA unwinding activities as the mutant proteins are defective in helicase function. One mutant, R487C (motif 4), also had a higher intrinsic dTTPase activity, but the neighboring mutant, G488D, had the same intrinsic dTTPase activity as the wild type. The only mutants that had extreme defects of their dTTPase activity are the E348K (motif 1a), D424N (motif 2B), and S496F (motif 4). These mutants are most likely involved in direct interactions with the dTTP nucleotide.
The DNA-dependent dTTPase activity of the mutants was assayed in the presence of ssM13 DNA. The wild-type 4AЈ protein showed about a 100-fold stimulation of dTTPase activity in the presence of ssM13 DNA (Table II). Mutant proteins, G116D (N terminus), A257T (linker region), S345F (motif 1a), and G451E (motif 3), which had elevated intrinsic dTTPase activity all showed DNA-stimulated dTTPase activity. All hydrolyzed dTTP with slightly lower (2-5-fold) DNA-stimulated dTTPase rates than 4AЈ. The reduction in DNA-stimulated dTTPase activity of the above mutants was relatively slight compared with the large decrease in the helicase activity of these mutant proteins. The inability of these four mutants to unwind duplex DNA, therefore, cannot be due to their inability to hydrolyze the dTTP substrate.
All the mutants in motif 4, R487C, G488D, and S496F, showed no DNA-stimulated dTTPase activity. Interestingly, the dTTPase activities of R487C and G488D mutants were actually lower in the presence of DNA. Evidently DNA somehow inhibits their dTTPase activity. A similar observation was made with the site-directed mutant R487A (13). These two mutant proteins are the ones that had lower affinity for ssDNA (see Fig. 6 and Fig. 7), and the failure of M13 ssDNA to stimulate dTTP hydrolysis activity correlates with their DNA binding defect. Finally, E348K (motif 1a), S496F (motif 4), and D424N (motif 2B) mutants that were defective in intrinsic dTTPase activity were also defective in dTTPase activity in the presence of ssDNA consistent with their role in interactions with dTTP.  Protein Oligomerization and DNA Binding-Both helicase and primase activities require hexamer formation and DNA binding. The 4AЈ mutants were assayed for these functions by native PAGE in the presence of dTMP-PCP, since the nonhydrolyzable dTTP analogue promotes stable protein oligomerization and DNA binding by 4AЈ (7). All the mutants assembled into oligomers ranging from dimers to hexamers and beyond similar to 4AЈ (Fig. 6a). These results indicate that none of the mutant proteins are defective in oligomerization. Interestingly, one mutant, R487C, appeared to form more stable hexamers relative to 4AЈ.
Previous studies have shown that ssDNA binding and hexamer formation are linked processes, that is hexamer formation is necessary for DNA binding (7). In addition, DNA binds to 4AЈ on a native gel only in the presence of dTMP-PCP, most likely because dTMP-PCP does not get hydrolyzed and thus promotes stable hexamer formation. In addition, hexamers are stabilized on the native gel in the presence of ssDNA. To assay for DNA binding, the native PAGE experiment was carried out in the presence of radiolabeled 30-mer DNA and MgdTMP-PCP. The protein species were detected by Coomassie Blue staining, and DNA binding was detected by PhosphorImaging. Fig. 6b shows the Coomassie-stained gel showing oligomerization of the mutant proteins in the presence of DNA. All the mutant proteins oligomerized in the presence of dTMPPCP and DNA. In the presence of DNA, the mutant proteins G116D (N terminus), A257T (linker region), and S345F (motif 1a) formed more stable hexamers and fewer lower order oligomers similar to 4AЈ. However, oligomerization of the mutant proteins, E348K (motif 1a), G451E (motif 3), D424N (motif 2B), R487C, G488D, and S496F (all motif 4), was unaffected by the presence of DNA. The smaller, middle domain is the N terminus primase domain, and the smallest, bottom domain is the zinc-binding domain, which is only present on the 4A subunit, needed for primase template recognition. The 60-mer ssDNA is shown binding through the hexamer hole, and the directionality is indicated. Possible interactions with the DNA-binding site, located inside the hole, explain the requirement of a tail flanking the 3Ј side of the primase template sequence. The 60-mer DNA image was constructed using the HyperChem program; the 4B hexamer image was provided by Dr. Egelman (University of Minnesota) and derived from electron microscopic analysis, and the 63-amino acid zinc-binding domain was drawn to scale and placed in its most likely position.
DNA Binding by Nitrocellulose-DEAE Assay-The native gel provides a rapid means to assay for both protein oligomerization and DNA binding. However, since this assay only detects stable or very tight DNA binding, it is not a sensitive or a quantitative assay for measuring DNA binding. We have used the nitrocellulose-DEAE binding assay, which can provide quantitative information about DNA binding, to further investigate the DNA binding properties of mutant proteins that do not bind DNA on the native gel. The DNA binding assay was carried by titrating a constant amount of radiolabeled 30-mer DNA with increasing protein. DNA binding was quantitated from the protein-DNA complex bound on nitrocellulose membrane and free DNA on the DEAE membrane. The results of these experiments with 4AЈ and some of the mutant proteins are shown in Fig. 7. All the mutant proteins in motif 4, R487C, G488D, and S496F showed weak or no DNA binding, whereas other mutant proteins that did not bind DNA on the native gel, E348K (motif 1a), D424N (motif 2B), or G451E (bound DNA weakly), showed stoichiometric DNA binding that was comparable with 4AЈ. DISCUSSION To unwind duplex DNA, a helicase requires a number of subactivities, including NTP binding, NTP hydrolysis, oligomerization, DNA binding, and coordination among these various subactivities. A defect in any of these functions can lead to a defect in the unwinding activity. We have used random mutagenesis and genetic selection to obtain helicase-defective mutants of T7 primase/helicase with the goal of identifying amino acids or regions of the protein participating in these various activities. A detailed biochemical analysis of these mutants has allowed us to understand the basis for the helicase defect at the level of these various subactivities.
Random mutagenesis created a number of single amino acid mutations distributed throughout the protein that produced defects in the helicase function required in vivo (14). All of the mutants provided sufficient primase function, except those that produced shortened proteins. The biochemical properties of the nine mutant proteins analyzed in this paper are consistent with their in vivo properties; all were able to synthesize RNA primers but had severe defect in helicase activity.
Relationship between Primase and Helicase Activities-In bacteriophage T7, both the helicase and primase activities can be provided by the same protein (4A), and from amino acid sequence homology to primases and helicases, it appears that the N-terminal half of the protein contains all the primase motifs and the C-terminal domain the helicase motifs. The two activities are coupled to a certain extent because amino acid changes in the N-terminal domain affected the helicase activity and vice versa. The mutants in the C-terminal domain, R487C, G488D, and S496F, that had lower affinity for ssDNA showed a measurable reduction in primase activity. This suggests that motif 4, likely involved in DNA binding necessary for the helicase function, may also be important for efficient RNA primer synthesis. The zinc-binding motif (N-terminal 63 amino acids) that recognizes the primase site may not be the only DNA interacting site necessary for primase activity. Previous work (6) has shown that ssDNA binds through the central hole of the hexamer as shown in Fig. 8. The 3Ј-end of the DNA binds on the side of the hexamer containing the larger domain, which we have identified using limited proteolysis as the C-terminal domain or the helicase domain. 3 It has been noted that a 5-base tail on the 3Ј side of the primase recognition site is not sufficient for primer synthesis, and a 10-base tail on the 3Ј side of the primase recognition site is required (24). We therefore propose that the 10-base tail on the 3Ј side is bound by the DNA-binding site inside the hole of the hexamer when the zinc-binding domain binds to the primase recognition sequence (Fig. 8). This mode of DNA binding explains the lower primase activity of the mutants, since R487C, G488D, and S496F have impaired DNA binding ability. This may also partly explain the activation of DnaG primase by the DnaB helicase in E. coli and the activation of gp61 primase by the gp41 helicase in bacteriophage T4.
Functional Classification of the Mutants-We have chosen nine representative mutants from both the N terminus and from the various conserved motifs of the C terminus domain for biochemical studies. Based on their defects with respect to the various subactivities, the nine 4AЈ mutants have been classified into three groups (see Table III). A previously character- Here is tabulated their DNA-induced hexamer formation activity as measured by nondenaturing PAGE. ϩϩ indicates that the hexamer is present to the same extent as the wild type on the native gel in the presence of 30-mer ssDNA. ϩ indicates that the hexamer is present, but it is not as predominant as the wild type in the presence of 30-mer ssDNA.
b ϩ indicates tight, stoichiometric binding of ssDNA to protein in the NC/DEAE membrane binding assay. Weak indicates binding weaker than stoichiometric binding. Ϫ indicates no detectable binding by the membrane binding assay.
ized mutant, K318A, is also included for discussion. Group A mutants are defective in intrinsic dTTPase activity. These mutants are confined to the C-terminal domain and lie in the various conserved helicase motifs. Group B mutants are defective in DNA binding but competent in intrinsic dTTPase activity. These mutants are also confined to the C-terminal domain. Group C mutants are defective in coupling dTTPase and helicase activity. These mutants have normal subactivities, and the mutants are not confined to the C-terminal domain but are found throughout the protein. We have not identified any mutants defective in oligomerization. Perhaps protein-protein interactions that stabilize the hexamer occur over a large surface area; hence, multiple mutations would be required to affect oligomerization to a significant level.
Nucleotide Triphosphate Binding-The group A mutants defective in dTTP binding/hydrolysis are found in four different helicase motifs. Mutants K318A, E348K, D424N, and S496F lie within motifs 1A, 1a, 2B, and 4, respectively. Motifs 1A and 2B are the well known Walker A and B sequences that have been shown in a number of proteins to be involved in NTP binding. Amino acids in region 1A form a P-loop that interacts with the ␤-␥-phosphate of NTP and Mg 2ϩ in proteins such as Ef-Tu, F 1 -ATPase, and RecA (25)(26)(27). Asp-424 in motif 2B is found to interact with the NTP via the Mg 2ϩ that binds to the ␤-␥phosphate of NTP (26,27). In addition to these regions, our mutagenesis results show that part of motif 1a and motif 4 also participates in dTTP binding/hydrolysis, as mutants E348K in motif 1a and S496F in motif 4 are severely defective in their intrinsic dTTPase activity.
All group A mutants bind DNA, except S496F (motif 4), which showed no DNA binding either in the presence of dTTP or dTMPPCP. The mutant S496F is different from the other mutants in that it shows a defect both in dTTPase and DNA binding. This is interesting because other mutants in motif 4 are also defective in DNA binding. Thus the amino acids in motif 4 may play a role both in DNA binding and dTTP binding/hydrolysis. DNA Binding-The group B mutants show a defect in DNA binding. Thus far, we have identified only two group B mutants, and both lie in motif 4. Mutants R487C and G488D were competent in binding and hydrolyzing dTTP, forming hexamers (as assayed by native gel), but these mutants bound DNA weakly. Results from three different assays are consistent with a DNA binding defect in these mutants. First, these mutants did not bind DNA by native gel assay. Second, weaker DNA binding was observed by a nitrocellulose-DEAE DNA binding assay. Third, although these mutants have intrinsic dTTPase activity, they did not show DNA-stimulated dTTPase activity. Since the DNA binding regions of helicases are not known, we cannot rule out the possibility that these amino acids in motif 4 may be indirectly involved in interactions with the DNA. In indirect interactions, these amino acids would play a role in the conformational changes necessary for DNA binding that ultimately lead to translocation and strand separation.
We have shown that in T7 helicase, DNA binding is modulated by dTTP binding and hydrolysis (7). 4AЈ protein binds DNA with a high affinity in the presence of dTMPPCP, the analogue of dTTP. But the affinity of 4AЈ for the DNA is much lower in the presence of dTDP, the product of dTTP hydrolysis. Since the state of the nucleotide bound, triphosphate or diphosphate, affects the affinity for DNA, there is probably a switch region that allows communication between the nucleotidebinding site and the DNA-binding site, similar to the switches observed in GTPases such as EF-Tu (25). We postulate that motif 4 involved in both dTTP binding/hydrolysis and DNA binding may be part of that switch region.
A previous report hypothesized that motif 4 is important for hexamer formation based on characterization of three sitedirected mutants, H475A, D485G, and R487A (13). It is known that nucleotide binding, hexamer formation, and DNA binding are all linked processes. Therefore, the idea that this domain is involved in hexamer formation may be an oversimplification of the function of the residues in this motif. In addition, the role of this motif in hexamer formation is not consistent with the characterization of random mutants reported in this paper. We have shown that the R487C and G488D mutants were not defective in oligomerization. In fact R487C formed hexamers in the absence of DNA that are more stable than the wild-type 4AЈ. Note that the site-directed mutant R487A was reported also to have no defect in oligomerization but clearly a defect in DNA binding. It was suggested that these defects were due to improper orientation of the subunit within the hexamer (13). However, since this is the most severely affected mutant of the three site-directed ones, we prefer the explanation that the Arg-487 and Gly-488 residues play a role in DNA binding. The effect of mutating residues in this region on hexamer formation may be an indirect one because of the coupling between DNA binding and hexamer formation.
Energy Transduction-Group C mutants did not show a severe defect in any of the subactivities. These mutants hydrolyzed dTTP and bound DNA and formed stable hexamers. However, none had the ability to unwind the small fork DNA. We have classified these mutants as having a defect in coupling the subactivities to helicase function. It is not uncommon in helicases that point mutations result in a defect in coupling nucleotide hydrolysis and DNA unwinding. Mutations affecting coupling have been reported recently for E. coli transcription termination factor rho (28) and RNA helicase eukaryotic initiation factor-4A (29). In eukaryotic initiation factor-4A, the mutants defective in coupling ATPase to helicase activity are found among the A motif, B motif, and a C-terminal region (region III) conserved among DEAD box RNA helicases.
The group C mutants are either intrinsically defective in DNA unwinding or have a defect in coupling the energy of nucleotide hydrolysis to DNA unwinding. Because of their large number and their distribution throughout the protein in both domains, it is more likely that these mutants are defective in energy transduction brought about by a defect in conformational changes of the protein. Uncoupling of activities can occur through a variety of mechanisms. The dTTPase activity may be uncoupled from translocation activity necessary for DNA unwinding. In such a case, the proteins will hydrolyze dTTP but fail to move along the DNA, since hydrolysis fails to cause the necessary conformational changes. Alternatively, these mutants may be severely defective in processive translocation. This would lead to frequent dissociation of the protein which would impair both translocation and DNA unwinding. The two types of defects described above may be caused by improper coordination of dTTP binding and/or DNA binding within the same subunit of the hexamer, or improper coordination among the various subunits. Interestingly, all group C mutants have a characteristic ϳ10-fold increase in the intrinsic dTTPase activity relative to wild-type 4AЈ. The increased dTTPase activity is either the consequence of or the cause for the uncoupling of activities. More detailed experiments will be required to determine the exact reason for the uncoupling and the increased dTTPase activity of this group of mutants.
One peculiar property of the group C mutants is that they possessed helicase activity in the presence of the DNA polymerase, although somewhat reduced compared with the wildtype. This was surprising as these mutants were completely defective in the fork unwinding assay. Since T7 DNA polymer-ase can partially restore the helicase activity of these mutants suggests that the polymerase can somehow overcome the defect resulting from the mutations. Since T7 primase/helicase protein forms a specific complex with T7 DNA polymerase (30,31), it must be the formation of this complex that allows restoration of some activity of the group C mutants. The mechanism by which DNA polymerase partially restores the helicase activity of these mutants is unknown. Restoration of the helicase activity of group C mutants by the polymerase is consistent with our proposal that these mutants may be defective in energy transduction, because it is easier to imagine the DNA polymerase partially restoring the ability of the helicase mutant that is defective in either translocation or processivity rather than the activity of the mutant defective in intrinsic DNA unwinding activity.
The strand displacement ability of the group C mutants in the presence of the polymerase is not entirely consistent with the in vivo complementation results presented in the companion paper (14). The mutants are likely complexed with the T7 polymerase in vivo, but they still fail to provide enough helicase activity to complement the T7 helicase/primase deletion phage. There could be a difference in the in vitro and the in vivo assay conditions, especially in 4AЈ concentration which was 1 M in the in vitro assay. It is likely that under in vivo conditions, T7 DNA polymerase is unable to sufficiently restore the defect of these mutants to support T7 growth, or the mutants may be unable to function in other processes during phage growth, for example initiation of bidirectional DNA replication where helicase activity has been shown to be important in vitro (32).
Possible Structure of the NTP-binding Site-No high resolution structure of a helicase is known at the present time. An attempt has been made here to predict the location of the above mutants with respect to the NTP-binding site by secondary structure predictions of 4AЈ and alignment with the known structure of RecA (33) and F 1 -ATPase (26). Interestingly, F 1 -ATPase shows a great deal of sequence homology to E. coli rho protein which is a hexameric protein that has RNA/DNA un-winding activity (34). Both F 1 -ATPase and RecA proteins are structurally related to the 4AЈ protein. The F 1 -ATPase is a ring-shaped hexamer with six nucleotide-binding sites, and it interacts with the ␥ subunit of the ATPase within the central cavity of the ring, in an analogous manner to the mode of ssDNA binding to 4AЈ. The RecA protein forms helical filaments around ssDNA which binds in the interior cavity of the protein helix. In addition, the loops of RecA protein that interact with the DNA correspond to the loops of the F 1 -ATPase that contact the ␥ subunit (35,36).
The secondary structures of 4AЈ and the DnaB protein were derived using SOPMA (Self-Optimised Prediction Method from Alignments) (37,38). A consensus secondary structure was derived from five methods (38 -42). Despite little sequence homology between the F 1 -ATPase and the RecA protein, both structures are very similar in their secondary structures and tertiary folds (26,33). The secondary structural prediction of 4AЈ also corresponded well to that of F 1 -ATPase and RecA, particularly around the Walker A and Walker B nucleotide binding motifs.
Only the C-terminal half of 4AЈ, which contains the NTP and proposed DNA binding motifs, has been aligned with the known secondary structure of the F 1 -ATPase. Residues around Walker A motif (motif 1A) and Walker B motif (motif 2B) have been used as guides for alignment. The resulting alignment is shown in Fig. 9. The motif 1A begins with a sheet spanning residues 306 -311 and is followed by the P-loop containing the GXGKS sequence. Conserved motif 1a follows as a sheet spanning residues 337-342 and the following loop. The secondary structural prediction of 4AЈ for these residues is a helix, but for the very similar DnaB protein, this region is a sheet that corresponds with the F 1 -ATPase structure. The region between motif 1a and motif 2B in F 1 -ATPase and 4AЈ or DnaB shows little similarity; thus, the structural comparison was resumed at conserved motif 2 (residues 420 to 424) that forms a loop following a ␤-sheet. Conserved motif 3 is a helix followed by a loop and a short sheet. The F 1 -ATPase structure and the pre- FIG. 9. Possible structure of 4A helicase domain. The predicted secondary structure of 4AЈ is aligned with the known structure of the F 1 -ATPase as described in the text. The numbers indicate the beginning and ending residues of each secondary structural element. Boxes represent ␣-helices, and arrows represent ␤-sheets. The helicase-conserved motifs are indicated above the 4AЈ sequence. Above the 4AЈ structure, black arrows indicate the positions of group A mutants K318A, E348K, D424N, and S496F. White arrows indicate the position of group B mutants R487C and G488D. Gray arrows indicate the position of the group C mutants S345F and G451E. Above the F 1 -ATPase structure, downward pointing arrows indicate the position of amino acid residues that interact with the nucleotide bound primarily on the ␤ subunit. Upward pointing arrows indicate the position of amino acid residues that interact with the nucleotide bound primarily on the adjacent ␣ subunit. The location of Walker A and B nucleotide binding motifs and the R-loop are indicated above the structure. dicted structure of 4AЈ correspond well also from residue 481 to 504 that constitutes motif 4. The coil of residues 486 -490 corresponds to loop L2 in the RecA structure and the R-loop in the F 1 -ATPase structure. This predicted structure is entirely consistent with the biochemical properties of the mutants in these conserved motifs.
The four group A mutants K318A, E348K, D424N, and S496F (defective in nucleotide hydrolysis) are in regions that correspond to residues of the F 1 -ATPase protein that directly contact the nucleotide. There are five regions in F 1 -ATPase that contact ATP, the Walker A motif, residues forming a loop and a helix analogous to conserved motif 1a, the Walker B motif, the region analogous to motif 4 in helicase (which in F 1 -ATPase contacts nucleotide bound to the adjacent subunit in the hexameric ring), and a few residues at the very C terminus that contact the nucleotide base. Lys-318 is in Walker A motif; the Asp-424 is in the Walker B motif; Glu-348 is in motif 1a; and the Ser-496 is in the helix in motif 4. Analogous to F 1 -ATPase, it is also possible that 4AЈ may bind NTP at the interface. This would be consistent with the fact that nucleotide binding is linked to hexamer formation. If such is the case, some of the above residues, especially the ones in motif 4 in 4AЈ, may also be involved in interactions with the nucleotide bound to the adjacent subunit. Examination of the above structural prediction makes assignment of motif 4 as part of the DNAbinding site and the effector switch region between the nucleotide-binding site and the DNA-binding site even more likely. The corresponding loop where these two mutations occur is the R-loop in the F 1 -ATPase, which contacts the ␥ subunit in the central cavity of the ring-shaped hexamer, and the L2 loop in the RecA protein that interacts with the DNA.