Overexpression, purification, sequence analysis, and characterization of the T4 bacteriophage dda DNA helicase.

The bacteriophage T4 dda protein is a 5'-3' DNA helicase that stimulates DNA replication and recombination reactions in vitro and seems to play a role in the initiation of T4 DNA replication in vivo. Oligonucleotide probes based on NH2-terminal amino acid sequence were used to precisely map the location of the dda gene on the T4 chromosome. Using polymerase chain reaction techniques, the dda gene was then cloned into an expression vector, and the overproduced protein was purified in two chromatography steps. Both the genomic and cloned dda genes were sequenced and found to be identical, encoding a protein of 439 amino acids. The dda protein contains amino acid sequences resembling those of other known helicases, and is most homologous to the Escherichia coli recD protein. Protein affinity chromatography was used to show a direct interaction between the dda protein and the T4 uvsX protein (a rec A-type DNA recombinase).

Helicases have been isolated from a wide variety of eukaryotic and prokaryotic cells (reviewed by Matson and Kaiser-Rogers, 1990). Exact physiological roles have yet to be determined for many of these enzymes. However, they vary in a broad range of biochemical properties, including substrate unwound (RNA or DNA helices), direction of strand movement (5' to 3' or 3' to 5'), and nucleotide cofactor hydrolyzed, reflecting their variety of functions inside the cell.
Bacteriophage T4 has been found to encode all of its own replication proteins. These include two distinct DNA helicases, the gene 41 protein and the dda protein, which appear to be important in both DNA replication and recombination reactions (Krell et al., 1979;Alberts et al., 1980;Jongeneel et al., 1984b;Kodadek and Alberts, 1987). The gene 41 protein, which is essential for T4 DNA replication, is a highly processive DNA helicase that moves along a single-stranded DNA template in the 5' to 3' direction (Le. along the lagging strand of a replication fork) (Liu and Alberts, 1981a;Venkatesan et al., 1982). The gene 61 protein (the DNA primase that makes the RNA primers for Okazaki fragment synthesis) and the gene 41 protein interact to form the T4 primosome (Liu and * This work was supported by National Institutes of Health Grant GM24020. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. To whom correspondence should be addressed. Alberts, , 1981bNossal, 1980). The dda protein was originally isolated as a DNA-dependent ATPase by Ebisuzaki and co-workers (Debreceni et al., 1970;Behme and Ebisuzaki, 1975). Mutant dda-phage show a substantial delay in DNA synthesis, but because near normal amounts of DNA are eventually produced phage burst size is reduced only slightly (Little, 1973).' No UV sensitivity or defects in recombination have been detected in dda mutant infections (Behme and Ebisuzaki, 1975).

to the GenBankTM/EMBL Data Bank with accession number(s) The nucleotide sequence($ reported in thispaper has
The dda protein and the gene 41 protein share some properties at a biochemical level. Both DNA helicases run in the 5' to 3' direction along single-stranded DNA. In the absence of the gene 41 protein, the dda protein stimulates the rate of DNA strand-displacement DNA synthesis at an in vitro replication fork (Jongeneel et al., 1984b). Since no increase in this rate is observed when the ddaprotein is added to reactions that have been stimulated by the gene 41 protein, the two DNA helicases do not appear to act synergistically at the fork (Jongeneel et al., 1984b).
The dda protein differs from the gene 41 protein in acting distributively (continuously dissociating and reassociating with the DNA molecule being unwound) rather than processively (Jongeneel et al., 1984a). In addition, the dda helicase does not form a primosome with the 61 protein. Unlike the 41 protein, the dda protein binds tightly to the T4 gene 32 protein (helix-destabilizing or single-stranded DNA-binding protein), and it is retained when T4 infected cell lysates are passed over a uvsX protein agarose column (uvsX is a recA analog with a central role in T4 genetic recombination) (Jongeneel et al., 1984a;Formosa and Alberts, 1984). A role in recombination is further suggested by dda protein's 4-fold acceleration of the rate of uvsX protein-catalyzed DNA branch migration in in vitro reactions (Kodadek and Alberts, 1987).
In general, one suspects that the two T4 DNA helicases can partially substitute for each other for some of the helicase functions inside the T4 bacteriophage-infected cell. Evidence for this assertion comes from studies on the T4 gene 59 protein. A T4 gene 59 amber mutant alone on a nonsuppressing strain shows normal DNA synthesis early in infection, followed by DNA synthesis arrest at late times of infection (Cunningham and Berger, 1977). However, if the phage is also dda-, almost no DNA is made.' Thus, without the 59 gene product, the dda protein is essential for any phage growth. Recent biochemical characterization of the gene 59 protein has shown that it loads the T4 gene 41 helicase onto singlestranded DNA.3 Thus, the combined biochemical and genetic data suggest that the dda protein plays an important role in DNA metabolism in uiuo but that its function can be partly P. Gauss, personal communication. P. Gauss, unpublished observation. J. Barry, personal communication. replaced with that of the gene 59-41 protein complex.
In order to further characterize the dda helicase and its interaction with other proteins, we precisely mapped its 10cation within the T4 chromosome, sequenced and cloned the dda gene, and overexpressed and purified the dda protein. In addition, the purified dda protein was chromatographed on a uvsx protein affinity column to test for a direct uvsX-dda protein interaction.

MATERIALS AND METHODS
Reagents and Enzymes-All restriction and DNA modifying enzymes (including the Taq DNA polymerase) were purchased from New England Biolabs unless otherwise noted. Polynucleotide kinase and dideoxyribonucleoside triphosphates were obtained from Pharmacia LKB Biotechnology Inc., avian myeloblastosis virus reverse transcriptase from Life Sciences (St. Petersburg, FL), Sequenase was from United States Biochemical Corp., ampicillin from Roerig/Pfizer, lysozyme from Worthington, formamide from Fluka, and agarose from FMC Bioproducts. Dimethyl sulfate, piperidine, and hydrazine were from Aldrich, formic acid from Fisher Scientific, and [ Y -~~P ] A T P from Amersham Corp. The sequence of the 25 NH2terminal amino acids of the dda protein that we purified from T4infected cells by published procedures (Jongeneel et al., 1984a) was determined by Ken Williams (Yale University). Oligonucleotide primers were synthesized by the Biomolecular Resource Center at the University of California, San Francisco. The uvsX protein affinity column was prepared by Scott Morrical in this laboratory.
Plasmids-The plasmid vector pTL18xwd was obtained from Dr. T.-C. Lin (Yale University). This vector contains the large EcoR1-BamHI fragment from the pBR322 derivative, pUC19, ligated to the EcoRI-BamHI fragment from a pGW7 derivative that contains the X late promoter control region carrying the gene encoding the repressor cIa5' (Lin et al., 1987). The X-DNA has the rexA and rexB genes (map position 37,000-36,110) deleted, since they inhibit T4 infection when contained on a multicopy plasmid (Shinedling et al., 1987). The plasmids Bluescript M13+ and M13-, which contain the M13 bacteriophage origin, were purchased from Stratagene.
Isolation of DNA and RNA-TI cytosine-containing DNA (T4cDNA)' was obtained from laboratory stocks prepared according the Pribnow method (Pribnow et al., 1981). RNA from T4 phageinfected E. coli BE cells was prepared as described by McPheeters et al. (1986), with the modifications of Selick et a1.' Chemical Sequencing of DNA-To obtain the T4 ClaI-Hind111 DNA fragment (map position 10.608-10.295) for sequencing, T4cDNA was digested with EcoRV. The DNA fragments were separated by electrophoresis on a 0.8% agarose, 10 mM Tris-borate, pH 7.4, 1 mM Na3EDTA (TBE) gel. The 3.6-kb DNA fragment (map position 7.3-10.78) was cut out of the gel, and the DNA was recovered by electroelution onto a dialysis membrane followed by phenol extraction and ethanol precipitation. To purify the DNA from contaminants that inhibit the calf intestinal phosphatase, the DNA was purified on a NAC column (Bethesda Research Laboratories), following the manufacturer's instructions. The EcoRV fragment was then digested with HindIII and the 5' phosphates removed with calf intestinal phosphatase. The phosphatase was inactivated by SDS and phenol treatments, and the 5' ends of the DNA fragment were then labeled with 3zP by standard methods (Maniatis et al., 1982). The labeled fragments were digested with ClaI and separated by electrophoresis on a 1.7% low temperature gelling agarose, 10 mM Trisacetate, pH 7.4, 1 mM Na3EDTA (TAE) gel. The DNA was recovered by melting the agarose at 65 "C, followed by phenol extraction and ' The abbreviations used are: T4cDNA, T4 cytosine-containing DNA; cv, column volume; PCR, polymerase chain reaction; SDS, sodium dodecyl sulfate; kb, kilobase(s).
' H. E. Selick, G. D. Stormo, R. L. Dyson, and B. M. Alberts, manuscript submitted ethanol precipitation. The 300-base pair ClaI-HinIII fragment, labeled at its HindIII end with 32P, was then sequenced according to the Maxam-Gilbert method (Maxam and Gilbert, 1980). Enzymatic Sequencing of T4 RNA and DNA-Sequencing from RNA templates by primer extension with avian myeloblastosis virus reverse transcriptase and termination by dideoxyribonucleoside triphosphates was performed by the method of Inoue and Cech (1985) as modified by McPheeters et al. (1986) and by Selick et aL5 Sequencing from DNA templates was carried out as described previously for RNA templates: with the following changes: 1) 56 fmol of T4cDNA (as molecules) was used per 12 pl of reaction; 2) the molar ratio of labeled primer to T4cDNA was 30:l; 3) T4cDNA and primer were heated for 3 min at 90 "C and quickly chilled in a dryice/ethanol bath just prior to the enzymatic reaction.
Southern Hybridization-The 0.7-pg samples of T4cDNA were digested individually with 64 units of either ClaI, EcoRV, EcoRI, HindIII, NdeI, or XbaI. After electrophoresis on a 0.75% agarose/ TBE gel, Southern transfer and hybridization were performed as described in Maniatis et al. (1982).
Sequencing of Plasmid DNA-Plasmid pKHdda was digested with BamHI and SalI. The fragments were separated by electrophoresis on a 1% low temperature gelling agarose/TAE gel and the 1.3-kb BamHI-Sal1 fragment containing the dda gene was recovered by melting the agarose at 65 "C followed by phenol extraction. The DNA was ethanol precipitated, resuspended in digest buffer, and cut with HindIII. The 165-base pair BamHI-Hind111 and 1.1-kb HindIII-Sal1 fragments were ligated into separate Bluescript M13 origin-containing plasmids. After the constructs were transformed into E. coli DG98 cells, ssDNA was isolated and sequenced using the Sequenase protocol based on the Sanger method (Sanger et al., 1977).
Amplification and Modification of the dda Gene by Exploiting the Polymerase Chain Reaction (PCR)-The PCR amplification reaction was based on the protocol of Kogan et al. (1987). Reactions were performed in 1.5-ml screw-capped Sarstedt tubes. The reaction mixture had a final volume of 100 pl and contained 250 pmol (as molecules) of each primer, 1.0 X pmol of T4cDNA molecules, 1.5 mM each of dATP, dCTP, dGTP, TTP, 10% dimethyl sulfoxide, and Taq polymerase buffer (16.6 mM ammonium sulfate, 67 mM Tris-HC1, pH 8.8, at 25 "C, 6.7 mM MgClz, 6.7 pM Na3EDTA, 170 pg/ml bovine serum albumin, and 10 mM P-mercaptoethanol). The DNA was denatured by incubation at 95 "C for 5 min, and the tube was spun briefly. Two units of Taq DNA polymerase were added, and the mixture was layered with mineral oil. The primers were annealed at 42 "C for 24 s, DNA synthesis was allowed at 65 "C for 4 min, and the DNA was denatured at 95 "C for 1 min. This cycle was repeated 30 times and stopped by placing the tubes in an ice-cold water bath. The amplified DNA was electrophoresed on a 1% low temperature gelling/TAE gel. The predominant 1.3-kb band was cut from the gel, and the DNA was recovered from the gel piece by melting the agarose at 65 "C followed by extraction with phenol. The DNA was ethanol precipitated, resuspended in digest buffer, and cut with BamHI and SalI. After deproteinization, the 1.3-kb fragment with BamHI and SalI ends was ligated into the pTLl8xwd plasmid following standard methods (Maniatis et al., 1982).
Plating of Phage-E. coli Tab32-4 cells containing either plasmid pKHdda or pTL19xwd in LB media were incubated at 30 "C. For complementation at 35 "C, samples were switched to a water bath at 35 "C 15 min prior to infection with a T4ts75(gene 32)sudl double mutant. Cells were infected, plated in T4 soft agar, and incubated at the appropriate temperature (30 or 35 "C).
Overproduction and Purification of the Cloned dda Gene Product-An E. coli SG934 culture containing plasmid pKHdda was grown to 6 X 10' cells/ml in LB media containing 50 pg/ml ampicillin. The temperature was then quickly switched to 38 "C, and the incubation was continued for 3 h. After harvesting, the cells were stored at -20 "C.
Cells were lysed using a procedure (Alberts and Frey, 1970) modified by J. Barry in this laboratory? Cells (28 g) were thawed and resuspended in 136 ml of buffer containing 20 mM Tris-HC1, pH 8.1, 1 mM Na3EDTA, 1 mM P-mercaptoethanol, 10 mM benzamidine-HCl, 1 mM phenylmethylsulfonyl fluoride. T4 lysozyme was added to a final concentration of 200 +g/ml and the suspension was incubated until viscous. The DNA was degraded by adding DNase I to 10 pg/ ml and adjusting the extract to 10 mM MgC12, 1 mM CaCIZ. The suspension was incubated in ice water, gently mixed until the viscosity decreased, and then sonicated with repeated 1-min blasts from the 0.5-inch horn of a Branson sonifer (40% duty) until the OD,, had dropped to 15% of its original value. Another 10 pg/ml of DNase I was added, and the extract was incubated at 15 "C for 20 min. The extract was centrifuged at 20,000 rpm (48,200 X g ) in a Sorvall 34 rotor for 20 min to remove cell debris and then further clarified by centrifugation at 35,000 rpm (111,000 X g ) in a Beckman vTi 50.2 rotor for 3 h. The supernatant was dialyzed against 4 two-liter changes of buffer A (20 mM Tris-HC1, pH 8.1, 5 mM Na3EDTA, 1 mM Pmercaptoethanol, 2 mM benzamidine-HCl, and 10% (v/v) glycerol) to remove the divalent cations necessary for DNase I activity.
A 55-ml single-stranded DNA-cellulose column was constructed (Alberts and Herrick, 1971), containing 1.5 mg of DNA/packed ml. The dialyzed extract (fraction I, Table 11) was pumped at 1 column volume (cv)/h through the column, washed with 1 cv of buffer A containing 0.10 M NaCI, followed by 1.5 cv of buffer A containing 0.25 M NaC1, and then eluted with a 2-cv linear gradient of 0.24-2 M NaCl in buffer A. Fractions of 4 ml were collected. ATPase activity was determined using a charcoal adsorption assay, as described previously (Liu and Alberts, 1981).
The fractions containing the dda protein were pooled and dialyzed against buffer B (20 mM Tris-HCI, pH 8.1, 20 mM NaC1, 1 mM NaaEDTA, 2 mM P-mercaptoethanol, and 10% (v/v) glycerol). A 5ml DEAE-cellulose column was constructed and equilibrated with buffer B. The pooled, dialyzed DNA cellulose fractions (fraction 11, see Table 11) were loaded at 1 cv/h onto the column, and then eluted with 2 cv of buffer B containing 60 mM NaCI. Fractions of 1.2 ml were collected. The fractions containing dda protein were pooled and concentrated in a Centricon 30 ultrafiltration device (Amicon Corp., Danvers, MA), and adjusted to storage buffer (20 mM Tris-HC1, pH 7.7, 50 mM NaCl, 1 mM p-mercaptoethanol, 0.2 mM Na3EDTA, and 45% (v/v) glycerol. The dda preparation was free of detectable nuclease activities as evaluated by sensitive DNA gel electrophoresis assays (Morris et al., 1979).
UUSX Protein Affinity Chromatography-Highly purified T4 uvsX protein was covalently linked to Affi-Gel-10 (Bio-Rad) to produce a matrix containing 1.7 mg uvsX protein/packed ml (Formosa et al., 1983). A 0.5-ml uvsX-Affi-Gel column was prewashed with 2 cv of column buffer (20 mM Tris-Cl, pH 8.1, 1 mM Na,EDTA, 1 mM 6mercaptoethanol, and 10% (v/v) glycerol) containing 2 M NaCl and then equilibrated with column buffer containing 25 mM NaCl at 4 "C. A mixture of 25 pg of purified dda protein and 37 pg of albumin in 0.2 ml of column buffer containing 25 mM NaCl was applied to the column at 1 cv/h. The column was washed with 2 cv of column buffer containing 25 mM NaCl and eluted successively with 50 mM, 100 mM, and 2 M NaCl in column buffer (1 cv each). Fractions of 100 p1 were collected, subjected to electrophoresis on a 13.5% SDS-polyacrylamide gel, and visualized by Coomassie Blue staining.

RESULTS
The dda Gene Is Located a t the 10.5-9.0-kb Position on the T4 Chromosome- Behme and Ebisuzaki (1975) mapped the dda gene by showing that its DNA-dependent ATPase activity was present in cells infected with the T4 del(39-56)l mutant (deletion map position 6.15-10.34 kb), but was missing in cells infected with the T4 del(39-56)lO mutant (deletion map position 6.3-10.7 kb).
We used two 20 nucleotide-long probes derived from the NHz-terminal amino acid sequence that we determined for the purified dda protein to further define the location of the dda gene in the T4 chromosome. Oligonucleotide 1 was derived from the first 6 amino acids of dda and was 128-fold degenerate. Oligonucleotide 2 was derived from amino acids 8-13 and was 192-fold degenerate.
Blot hybridization to restriction nuclease-digested T4cDNA was performed separately with oligonucleotides 1 and 2 according to standard methods (Maniatis et al., 1982). Each probe hybridized to multiple restriction fragments, but only restriction fragments that hybridized with both probes were judged to contain the dda gene. Both oligonucleotides hybridized to a 2.0-kb Hind111 fragment at map position 10.3-12.35 on the T4 linear chromosome and with a 1.7-kb EcoRI fragment at map position 8.94-10.59; thus the 5' end of the dda gene must lie between map position 10.3-10.6. Assuming that dda is transcribed in the counter-clockwise direction on the T 4 circular map like other early and middle T4 genes (Kutter and Ruger, 1983), the dda gene should lie somewhere between map positions 10.6 and 8.8.
The ClaI-Hind111 DNA fragment, encompassing map positions 10.7-10.4, was isolated from a digest of T4 cytosinecontaining DNA (T4cDNA). This fragment was sequenced using the Maxam-Gilbert technique to determine the precise location of the dda gene's 5' end and confirm the expected orientation of the gene on the T4 chromosome. The results showed that the 5' end of the ddu gene lies at the 10.5 position on the chromosome with the predicted orientation; therefore, the 3' end must lie at position 9.0 (assuming that the gene encodes a 50-kDa protein and does not contain an intron).
Sequence Analysis of the dda Gene-Repeated attempts to clone restriction fragments encompassing the entire dda gene by conventional methods failed, suggesting that the dda gene product is deleterious to E. coli when under the control of its own promoter. Fortunately, both T4cDNA and T4 mRNA are easily obtained from T4-infected cells, and by employing these nucleic acids as single-stranded templates we were able to use the dideoxy method to determine the sequence of the dda gene. The strategy used for sequencing is presented in Fig. 1. The dda gene sequence is shown in Fig. 2.
Sequence analysis predicts a dda protein sequence of 439 amino acids. The calculated mass of 49,947 daltons agrees with the estimated mass from SDS-polyacrylamide gel electrophoresis of 56,000 (Krell et al., 1979), 50,000 (Purkey andEbisuzaki, 1977), or 48,000 daltons (Jongeneel et al., 1984a). The dda sequence contains a consensus nucleoside triphosphate-binding site and six regions of homology with other helicases (see "Discussion").
Amplification, Modification, and Cloning of the dda Gene for Overproduction of the Gene Product-We used the PCR technique to amplify the dda gene while simultaneously changing its Shine-Dalgarno sequence in preparation for cloning in a tightly regulated expression vector. The sequences of the two primers used for PCR are shown in Fig. 2. T4cDNA was used as the template, and the PCR reaction was carried out as described by Kogan et al. (1987), with the following modifications: the time and temperature used for polymerase incubation was increased to 4 min at 65 "C, primers were allowed to anneal at 42 "C for 24 s, and no additional Taq DNA polymerase was added during the cycles. On a 1% agarose gel, more than 90% of the DNA product migrated at 1.3 kb, the length of the dda gene; 30 rounds of synthesis yielded a 13,000fold amplication (1.3 pg of DNA). The 1.3-kb DNA fragment containing the dda gene was placed in an expression vector, pTLlSxwd, downstream of the h late promoter. This promoter is controlled by the c1857 repressor produced by the vector. The plasmid contains the 39,178-34,500 region of bacteriophage X with the r e d and B genes deleted; the rex gene products inhibit T4 growth when contained on a plasmid (Shinedling et al., 19871, which would complicate genetic studies. DNA strands from the cloned dda gene, complementary to those sequenced from T 4 genomic nucleic acids (see Fig. lA), were sequenced to confirm the sequence obtained from the T4 genomic nucleic acids and to determine if any mutations were introduced during the cloning of the dda gene. The approach used to sequence the cloned dda gene, pKHdda, is shown in Fig. 1B. The sequence of both the genomic and the cloned dda gene were found to be identical, and the sequence of the first 25 amino acids encoded by the gene is the same as that determined for the dda protein isolated from phage T4 infected cells.
The production of functional dda protein was tested genetically by assaying the ability of the cloned dda gene to com- Each arrow represents the sequence information obtained from an individual oligonucleotide primer with the arrow pointing in the direction of sequencing from the primer. The sequence information required to prepare the most 5' primer was determined by Maxam-Gilbert sequencing of the T4 ClaI-Hind111 DNA fragment, map position 10.608-10.295 (see "Materials and Methods"). The information required for the primer used at the 3' end of the gene was obtained by dideoxy sequencing, starting from the published EcoRI-EcoRI fragment sequence at map position 7.6-8.942 (Gauss et al., 1987). Denatured double-stranded T4 genomic DNA was the template for dideoxy sequencing at the start of the dda gene (arrows pointing to the right), whereas both this T 4 DNA and total RNA from T4-infected E. coli cells were used as templates for dideoxy sequencing for arrows pointing to the left. B, the plasmid containing the ddu gene, pKHdda, was cut with BamHI, HindIII, and SdI. The BamHI-Hind111 and HindIII-Sal1 fragments were cloned separately into M13 origin-containing Bluescript plasmids a s shown by the separate open boxes. Single-stranded DNA was isolated from M13-infected cells containing these plasmids, and the sequence of all DNA strands complementary to those sequenced from the T4 genome-derived nucleic acids shown in A were determined by the dideoxy sequencing method (thus, Q~~O U S in B point opposite to arrows in A for each part of the gene). plement a bacteriophage T4 gene 32 temperature-sensitive, sud deletion double mutant (T4ts75sudl). The sud and dda genes are believed to be the same (Jongeneel et al., 1984;Doherty et al., 1982). Whereas either the T4 ts75 (gene 32) or sudl mutant alone will grow at both 30 and 35 "C, the T4ts75(gene 32)sudl double mutant phage will not grow at either temperature on a Tab32-4 E. coli strain (Doherty et al., 1982); this strain restricts the growth of many gene 32 temperature-sensitive mutants at normally permissive temperatures, without affecting wild-type T4 phage (Nelson and . As shown in Table I, the T4ts75(gene 32)sudl mutant phage grows on Tab32-4 containing pKHdda but not on Tab32-4 containing only the vector (pTLlSxwd), at both 30 and 35 "C. This result shows that the dda gene product encoded by the plasmid is active in uiuo, and it further supports the previous evidence that sud and dda are the same gene.
Overproduction and Purification of the Cloned dda Gene Product-The dda protein was overproduced from plasmid pKHdda in E. coli SG934 cells, which contain a mutation in the htpR gene. The htpR gene is essential for the transcription of heat shock genes, and proteases normally induced upon heat shock are not expressed (for review, see Neidhardt et al., 1984). The expression of the dda protein was induced at 38 "C for 3 h. After induction, the dda protein represented 1% of the total soluble protein, which is 10-fold more than obtained from T4-infected cells. Further overexpression of the dda protein could be obtained at higher temperatures, but this resulted in the formation of insoluble dda protein aggregates and a reduced final yield.
Purification of the Overexpressed dda Protein-The overexpression of the dda protein allowed us to simplify our previous procedure for dda purification from T4-infected cells, which involved five columns and resulted in a 5% yield of pure dda protein (Jongeneel et al., 1984a). From our induced cells, the dda protein can be purified free of nucleases after only two columns with a yield of 60%, as detailed in Table 11.
In the new procedure, the crude lysate is passed through a single-stranded DNA-cellulose column, from which the dda protein is eluted with a steep NaCl gradient. The fractions that contain the highly purified dda protein are pooled and chromatographed over a DEAE-cellulose column under conditions in which the dda protein flows through. The results of an SDS-polyacrylamide gel analysis at each stage of the purification are shown in Fig. 3.
The Ouerexpressed dda Protein Has DNA Helicase Actiuity-An assay was carried out to determine whether the overexpressed dda protein purified from T4-uninfected cells has DNA helicase activity, since it is conceivable that the dda protein requires post-translational modification or some other component picked up during T4 infection to become an active helicase.
To assay the dda protein for helicase activity, we constructed a DNA substrate that contains a fully complementary 5' end-labeled 393-nucleotide DNA fragment annealed to single-stranded genomic M13 DNA. The unwinding of this substrate by a helicase changes the mobility of the labeled DNA fragment on a non-denaturing agarose gel and is readily detected by autoradiography (Jongeneel et al., 1984a).
The unwinding of the DNA substrate by the overexpressed  (Fig. 1B) and wild-type (Fig. 1A) dda gene were identical. This result was unexpected since the clone was obtained using DNA amplified by the PCR technique with the Taq polymerase which (under slightly different reaction conditions) has a reported error rate of 0.25% (Saiki et al., 1988). At this error rate, we would have expected to find three to four nucleotide changes in the clone compared to the wild-type sequence. purified dda protein is presented in Fig. 4. The dda protein does not require post-translation modification by phage T4 proteins for helicase activity. The percentage of DNA unwound is a nonlinear function of dda protein concentration, as also observed for the dda protein purified from phage T4infected cells (Jongeneel et al., 1984a). The percentage of the DNA unwound greatly increases when the dda concentration is raised from 4 to 8 pg/ml, which increases the ratio of dda protein molecules to DNA nucleotides from 1:2 to 1:l. Three dda protein molecules/nucleotide are required for the unwinding of all of the DNA molecules. One dda protein molecule from phage T4-infected cells/three DNA nucleotides was needed to unwind 84% of the somewhat different DNA molecules used in our earlier study (Jongeneel et al., 1984a).

Ala Gly Thr Gly Lys Thr Thr Leu Thr Lys Phe Ile Ile Glu Ala Leu Ile Ser Thr
The dda Gene Product Binds Directly to the T4 UVSX Protein-Overexpression of the dda gene product made it possible to obtain large enough quantities of the protein to produce a dda protein affinity column. Affinity chromatography with other T4 proteins involved in DNA metabolism attached to an agarose matrix showed a tight interaction of the dda protein with the gene 32 protein (Formosa et al., 1983). To extend this analysis, the dda protein was covalently coupled to an agarose matrix, Affi-Gel 10, as described by Formosa et al. (1983). Although a column containing 2 mg of dda protein/ packed ml was prepared, we did not detect any binding of the purified gene 32 protein to this column (data not shown). This result suggests that the dda protein is inactivated during its attachment to the column. Interestingly, of the eight other T 4 proteins that have previously been attached to this agarose matrix, the only one that was similarly inactivated was the other DNA helicase, the T4 gene 41 protein?
The interaction of the dda protein with the T4 uvsX protein (a recA protein analogue) was previously suggested when the dda protein in an extract of T4-infected cells was retained on a uvsX protein-agarose column. Since the dda protein coeluted with the gene 32 protein from this column, the interaction of the dda protein with the uvsX protein could have been indirect (Formosa and Alberts, 1984). To determine if the uvsX and dda proteins bind directly to one another, we chromatographed a mixture of the pure dda protein with albumin on a uvsX protein affinity column. As shown in Fig.  5 , the albumin is not retained by the column, whereas the dda protein binds to the column and is eluted by 50 mM NaC1. The dda protein and the albumin behaved identically on an agarose control column, and the dda protein was not retained on an albumin-agarose control column (data not shown). These results demonstrate that there is a direct, albeit weak, interaction between the uvsX and dda proteins. Similar weak interactions have been observed between the protein subunits of the T4 DNA polymerase holoenzyme. On protein affinity columns both the interactions between T4 DNA polymerase accessory proteins (the gene 45 protein and the 44/62 protein complex) and the interaction of the T4 DNA polymerase with the gene 45 protein are disrupted by washing with 50 mM NaCl (Formosa and Alberts, 1984;Formosa, 1985).

DISCUSSION
Direct sequencing of nucleic acids (DNA and RNA) produced from bacteriophage T4-infected cells has allowed us to use the PCR technique to engineer an appropriate vector to produce the dda protein in E. coli. The dda gene product was thereby overexpressed to approximately 1% of the total cel-

TABLE I
The protein produced by the cloned dda gene i s biologically active Plating efficiencies of T4 ts75 (gene 32) sudl double mutant on E. coli strain Tab32-4, with and without the cloned dda gene. Plating efficiency is expressed as the number of plaques observed divided by the number of plaques produced on E. coli Tab32-4 containing the pKHdda plasmid at 35 "C. The plasmid used for the "vector only" control is identical to pKHdda, except that it lacks the dda gene insert. overexpression of the dda protein greatly simplified its purification, since its association with nucleases in the infected cell and with the tightly binding T4 gene 32 protein was avoided. Thus, the dda protein purified from overexpressing cells is more than 99% pure and free of nucleases after only two chromatography steps. The yield is greatly increased such that 7 mg of dda protein is obtained from 28 g of cells. The overexpressed dda protein has DNA helicase activity, and a direct interaction of the this protein with the T4 uvsX protein (a recA analogue essential for T4 genetic recombination) was detected on a uvsX protein-agarose column. Both the genomic and cloned dda genes were sequenced using the dideoxy method. The dda amino acid sequence shares six conserved motifs with a superfamily of ATPases identified by Gorbalenya et al. (1988) and independently by Hodgman (1988). The consensus sequences for the superfamily and the alignment of dda with the family of E. coli helicases within the six regions is shown in Fig. 6. Motif I contains the sequence common to many GTP-and ATP-binding domains, originally described by Walker et al. (1982). This motif forms a loop that binds the ATP phosphate (La Cour et al., 1985;Jurnack, 1985;Fry et al., 1986). Motif I1 most likely binds the ATP phosphate indirectly via a magnesium ion (Jurnack, 1985). Motif I11 is also conserved among many DNA and RNA polymerases (Hodgman, 1986). Motif VI is believed to be involved in DNA binding because of its occurrence in putative DNA-binding proteins (Hodgman, 1988). Neither the structure nor function of motifs IV and V is known.
Among this superfamily of more than 20 proteins, the dda protein has the greatest homology with the E. coli recD protein. The recD protein is the most diverged member of the E. coli helicase family, which is composed additionally of the recB, rep, and uvrD proteins (Hodgman, 1988). recD is a subunit of the recBCD (exonuclease V) complex which plays a central role in homologous genetic recombination (for review, see Telander-Muskavitch et al., 1981;Taylor, 1988). Exonuclease V moves along the DNA creating looped structures that are periodically cut by the enzyme (Taylor, 1988). The dda and recD protein share 38% amino acid identity within the six conserved sequence motifs and a Monte Carlo score of 4.5 when the amino acids NH2-terminal to the first motif and COOH-terminal to the last motif are deleted from the analysis. For comparison, recD shares a 26, 31.5, and 30% amino acid identity within the conserved motifs with recB, rep, and uvrD, respectively, and Monte Carlo scores throughout the region encompassing all of the conserved motifs of 6.8, 4.4, 2.9, respectively.
The Monte Carlo scores give a statistical evaluation of the homology of the motifs and the regions between them, encompassing a larger part of the gene than just the conserved motifs. It is calculated by aligning the sequences using the SS2 algorithm to produce a similarity score, subtracting from the original similarity score the mean scores from compari- One unit of ATPase activity is defined as the amount of enzyme required to hydrolyze 1 pmole of ATP in 1 min at 37 "C. Not possible to determine due to high ATPase activity by proteins other than dda; ND, not determined. Calculated by determining the amount of dda protein in fractions I and I1 by SDS-polyacrylamide gel analysis. The helicase activity of the dda protein as a function of its concentration. A DNA helicase assay was carried out as described by Jongeneel et al. (1989a). Reactions contained dda protein as indicated, 65 ng/ml DNA substrate constructed by annealing the fully complementary 5' end-labeled 393 nucleotide fragment from pJMCllO to single-stranded M13 DNA, 1 mM ATP, 33 mM Tris acetate, pH 7.8,66 mM potassium acetate, 10 mM magnesium acetate, 1 mM dithiothreitol, and 100 pg/ml human serum albumin. The reactions were incubated at 30 "C for 3 min, stopped by addition of sodium dodecyl sulfate and NasEDTA to 1% and 10 mM, respectively, and electrophoresed on a 1.2% agarose/TBE gel. The gel was then dried and autoradiographed.
sons with the randomized sequence, and then dividing by the standard deviation. A Monte Carlo score of between 3 and 6 indicates possible homology, whereas scores greater than 6 indicate a probable homology (Barker and Dayhoff, 1972;Argos and Vingron, 1990).
The overall homology between the recD and dda proteins A mixture of 25 pg of purified dda protein and 37 pg of albumin in column buffer containing 25 mM NaCl was applied to a 0.5-ml uvsx protein-Affi-Gel column containing 0.85 mg of covalently bound UVSX protein. The matrix had been previously equilibrated with the same buffer. The column was washed with 1 ml of column buffer containing 25 mM NaCl and then eluted with column buffer containing 50 mM NaC1. Fractions of 100 pl were collected, and 20 p1 of each fraction found to contain protein by Bradford assay (Bradford, 1872) was loaded on to a 13.5% SDS-polyacrylamide gel, electrophoresed, and visualized by Coomassie Blue staining. Note that the fractions at the end of the 25 mM NaCl wash contained no protein and are not shown.
is not enough to suggest a strong structural or functional homology. It is nevertheless worth nothing that their genetic phenotypes display some similarities. The dda-mutant phage show a DNA delay phenotype, but they eventually attain a phage burst size that is close to normal (Little, 1973; P. Gauss, personal communication). Only in a T4 gene 59-background is the dda gene essential for DNA synthesis.' The complex of recBC, missing recD, lacks exonuclease V activity in vitro.
But recD mutant cells show a hyper-recombination phenotype and are viable (Chaudhury and Smith, 1984). Thus, like the dda gene, recD is an nonessential gene. However, in a recJ mutant (recF pathway gene) background, the recD gene is required for chromosome recombination and UV resistance; this suggests that the recBCD and recF pathways are somewhat redundant (Lovett et al., 1988). Similarly, the dda DNA helicase seems to be partially replaceable by the gene 41 DNA helicase, providing that an accessory protein for 41 protein function, the gene 59 protein is present.
What are the physiological roles of the dda protein? The DNA delay phenotype of dda mutant phage, which is extended to a severe block in early DNA synthesis when the 59 protein is absent,' suggests an important function for the dda protein in the initiation of T4 DNA replication. Determination of its exact role in initiation is likely to require the reconstitution of the initiation process in a purified in vitro system containing a T4 replication origin (Kreuzer and Alberts, 1985;Kreuzer et al., 1988;Menkens and Kreuzer, 1988). In addition, the direct interaction of the dda and uvsX proteins reported here suggests that the dda protein serves as a specific accessory factor in T4 uvsX protein-catalyzed genetic recombination. A role in DNA recombination is also suggested by the dda protein's 4-fold stimulation of uvsX-catalyzed DNA branch migration rates (Kodadek and Alberts, 1987). However, the observed binding of the dda protein to both 32 protein and uvsX protein could have the alternative function of promoting access of the dda protein to single-stranded DNA, which is believed to be completely covered by one or the other of these proteins in a T4-infected cell.
In summary, we have cloned and sequenced the dda gene and used the sequence to overproduce the dda protein. The overproduced dda protein has allowed us to demonstrate its direct interaction with the UVSX protein. This result supports a role of the dda protein in DNA recombination, but many questions remain. The availability of the clone and large FIG. 6. Alignment of the dda protein with motifs conserved among a family of E. coli helicases. These motifs were independently identified by Gorbalenya et al. (1988) and by Hodgm a n (1988). The dda protein was aligned using the SS2 algorithm of Altschul and Erickson (1986). Residues placed in bones are absolutely conserved among six protein families of putative ATPases (Gorbalenya et al., 1988;Hodgman, 1988). Small letters are used in the consensus sequence to denote the range of observed amino acids at those positions where four or less alternatives exist among all members of the families. T h e numbers indicate the length of the gap between conserved motifs.