The Role of the Zinc Motif in Sequence Recognition by DNA Primases

The DNA primase of bacteriophage T7 has a zinc-bind- ing motif that is essential for the recognition of the sequence 3 (cid:42) -CTG-5 (cid:42) . The T7 primase also catalyzes helicase activity, a reaction coupled to nucleotide hydroly- sis. We have replaced the zinc motif of the T7 primase with those found in the gene 61 primase of phage T4 and the DnaG primase of Escherichia coli . The T4 and E. coli primases recognize the sequences 3 (cid:42) -T(C/T)G-5 (cid:42) and 3 (cid:42) -GTC-5 (cid:42) , respectively. Both chimeric proteins can partially replace T7 primase in vivo . The two chimeric pri- mases catalyze the synthesis of oligoribonucleotides albeit at a reduced rate and DNA dependent dTTPase activity is reduced by 3–10-fold. Both chimeric proteins recognize 3 (cid:42) -(A/G)CG-5 (cid:42) sites on single-stranded DNA, sites that differ from those recognized by the T7, T4, or E. coli primases, indicating that the zinc motif is only one determinant in site-specific recognition. DNA primases catalyze the template-directed de novo synthesis of oligoribonucleotides on single-stranded DNA (ssDNA) 1 for use as primers for DNA polymerases to initiate DNA synthesis at origins of replication and on the lagging strand side of the replication fork (1). In addition to their interaction with DNA polymerases, DNA primases are also physically and functionally associated with DNA helicases, in part, to make use of the unidirectional translocation activity of the helicase to access recognition sites

The DNA primase of bacteriophage T7 has a zinc-binding motif that is essential for the recognition of the sequence 3-CTG-5. The T7 primase also catalyzes helicase activity, a reaction coupled to nucleotide hydrolysis. We have replaced the zinc motif of the T7 primase with those found in the gene 61 primase of phage T4 and the DnaG primase of Escherichia coli. The T4 and E. coli primases recognize the sequences 3-T(C/T)G-5 and 3-GTC-5, respectively. Both chimeric proteins can partially replace T7 primase in vivo. The two chimeric primases catalyze the synthesis of oligoribonucleotides albeit at a reduced rate and DNA dependent dTTPase activity is reduced by 3-10-fold. Both chimeric proteins recognize 3-(A/G)CG-5 sites on single-stranded DNA, sites that differ from those recognized by the T7, T4, or E. coli primases, indicating that the zinc motif is only one determinant in site-specific recognition.
DNA primases catalyze the template-directed de novo synthesis of oligoribonucleotides on single-stranded DNA (ssDNA) 1 for use as primers for DNA polymerases to initiate DNA synthesis at origins of replication and on the lagging strand side of the replication fork (1). In addition to their interaction with DNA polymerases, DNA primases are also physically and functionally associated with DNA helicases, in part, to make use of the unidirectional translocation activity of the helicase to access recognition sites for primer synthesis.
All DNA primases, whether from bacteriophage, viral, prokaryotic, or eukaryotic sources, have a potential metal binding site (20,21). In the bacteriophage and prokaryotic primases, the potential metal binding site is located in the N terminus of the polypeptide. Two DNA primases, the phage T7 gene 4 primase (21) and the E. coli DnaG primase (22) have been shown to be zinc metalloproteins. It is reasonable to postulate that the zinc motif of primases is important in the recognition of primase sites since the recognition of specific sequences in DNA is known to occur in several other biological processes. For example, the Cys 2 His 2 zinc fingers of the Zif268 (23) and the human oncogene product GLI (24) make contacts with three bases. An alternative metal-binding motif, the Cys 4 zinc motif, is found in many proteins such as the human elongation factor TFIIS and interacts with both DNA and RNA (25,26). In one instance, gene 4 proteins with single amino acid substitutions of Ser at each of the four conserved Cys residues cannot support phage DNA replication and growth, and the altered proteins cannot catalyze template-directed synthesis of oligoribonucleotides (21).
The mechanism by which the prokaryotic DNA primases recognize a trinucleotide sequence on ssDNA is at present unknown as is the precise role of the zinc motif in this process. One approach to defining the role of the zinc motif in sequence recognition is to construct chimeric primases in which the zinc motif of one is substituted for another. The gene 4 primase of phage T7 has utility in this approach in that it is also a helicase, thus circumventing the dependence of the primase reaction on the presence of a separate specific helicase, as is the case for the phage T4 and E. coli priming systems (12,(27)(28)(29)(30). Gene 4 actually encodes two colinear proteins, a 56-kDa protein and a 63-kDa protein, the former protein arising as the result of an internal translation initiation sequence in the gene 4 transcript (31). The 56-kDa gene 4 protein is a helicase that translocates 5Ј to 3Ј on ssDNA, a reaction coupled to the hydrolysis of nucleotides, and catalyzes the unwinding of duplex DNA that it encounters (32). The 63-kDa gene 4 protein contains an additional 63 amino acids at its N terminus, and it is within this sequence that the zinc motif is located. Consequently, the 63-kDa gene 4 protein has primase activity in addition to the helicase activity found in both molecular mass species of the protein (3,32). Since the 63-kDa protein can provide both primase and helicase activities, it is both necessary and sufficient for productive infection by T7 phage (33,34). However, gene 4 protein functions as a hexamer (35)(36)(37), and most likely hexamers found in wild-type phage infected cells contain both molecular forms of gene 4 protein (38).
We recently constructed a T3/T7 chimeric primase in which the zinc motif of the phage T3 primase, a primase whose recognition site is not known, was substituted for the zinc motif of the T7 primase (39). The recognition sites used by the chimeric primase were identical to those used by the T7 primase, and the chimeric primase was functional in vivo, not a surprising result since T3 and T7 are closely related and their gene 4 proteins highly conserved (40,41). The ability of the T3/T7 chimeric primase to function both in vivo and in vitro provided the incentive for us to examine chimeric proteins in which the zinc motif of the primase of either phage T4 or E. coli is substituted for that of the phage T7 primase. Since the trinucleotide recognition sequences for these three primases differ from one another, the identity of any functioning chimeric protein will provide definitive information on the role of the zinc motif in sequence recognition and will perhaps yield insight into the importance of specific recognition sequences in a particular replication system. In this report we show that the zinc motifs of both the T4 and E. coli primases can replace the zinc motif of the T7 gene 4 protein to yield a functional primase but that the zinc motif alone does not dictate the sequence specificity.

EXPERIMENTAL PROCEDURES
Bacterial Strain of Bacteriophage-E. coli DH5␣ (Life Technologies, Inc.) was used for cloning of DNA, and E. coli C600 was used for complementation analysis. E. coli HMS 174 and bacteriophage CE6 were used for expression of recombinant genes. Wild-type bacteriophage T7 and T7 ⌬4-1 containing a deletion of the entire gene 4 were kindly provided by S. Notarnicola (Harvard Medical School). Bacteriophage T4 was obtained from American Type Culture Collection.
Enzymes and Biochemicals-Wild-type T7 56-kDa gene 4 protein was provided by B. Beauchamp (Harvard Medical School) and wild-type T7 DNA polymerase (T7 gene 5 protein and E. coli thioredoxin complex) was provided by S. Tabor (Harvard Medical School). Wild-type T7 63-kDa gene 4 protein was purified as described previously (37). T4 DNA polymerase, restriction enzymes, and other enzymes were purchased from Amersham Corp. and New England Biolabs.
Protein Purification-Purification of the chimeric primases was per-formed as described previously by Notarnicola et al. (37). Three liters of E. coli cells carrying the appropriate expression plasmids were cultured in 2X YT medium (16 g/liter Bacto-Tryptone, 10 g/liter yeast extract, and 5 g/liter NaCl) at 37°C until A 600 ϭ 1.0 and infected with bacteriophage CE6 which encode RNA polymerase of bacteriophage T7 at a multiplicity of infection of 10. The cells were cultured for an additional 3 h and harvested. The cells were suspended in buffer L (50 mM Tris-Cl (pH 7.4), 5 mM EDTA, 100 mM NaCl) containing 50 mg/ml lysozyme and 400 M phenylmethanesulfonyl fluoride and incubated on ice for 1 h. The suspension was frozen, thawed twice, and centrifuged at 30,000 ϫ g for 1 h at 4°C, and the supernatant was collected. The precipitate was reextracted twice with buffer L-II (50 mM Tris-Cl (pH 7.4), 5 mM EDTA, 500 mM NaCl), and the supernatant obtained after centrifugation (15,000 ϫ g for 15 min at 4°C) was combined with the first supernatant. The combined lysate (fraction I) was adjusted to a NaCl concentration of 0.5 M, and the protein was precipitated by the addition of polyethylene glycol 4000 to a final concentration of 10%. The precipitate was dissolved in buffer P (20 mM KPO 4 (pH 6. Nucleotide Hydrolysis Assay-The assay for measuring nucleotide hydrolysis by the primases was performed essentially as described previously (37). The reactions (20 l) contained 40 mM Tris-Cl (pH7.5), 10 mM MgCl2, 10 mM DTT, 50 g/ml bovine serum albumin, 50 mM NaCl, 5 mM [␣-32 P]dTTP, and varying concentrations of primase. After incubation at 30°C for 10 min in the presence or absence of 50 M M13 ssDNA (nucleotide phosphorous), the reaction was stopped by the addition of 5 l of 150 mM EDTA and 10 mM dTDP. The products of the reactions were separated by CEL 300 polyethyleneimine/UV thin layer chromatography (Brinkmann Instrument Inc.), and the amount of radioactivity in the dTDP spot was determined by scintillation counting.
RNA-primed DNA Synthesis-Analysis of primase stimulation of DNA synthesis on ssDNA by T7 DNA polymerase was performed using a modification of the method described previously (43). The reactions (20 l) contained 40 mM Tris-Cl (pH 7.5), 10 mM MgCl 2 , 10 mM DTT, 50 mM potassium glutamate, 0.3 mM rNTP, 0.3 mM dTTP, 0.3 mM [␣-32 P]dTTP, 5 nM T7 polymerase, 10 nM M13 ssDNA, and varying concentrations of primases for 20 min at 37°C. After incubation the reactions were stopped by the addition of 5 l of 0.2 mM EDTA (pH 8.0) and spotted onto DE81 filters. The filters were washed with 0.3 M ammonium formate (pH 8.0) four times and then washed with 95% ethanol once. After drying, the radioactivity retained on the filter was determined by scintillation counting.

Generation of Chimeric Primases-
The DNA primases of phage T7, phage T4, and E. coli all contain a Cys 4 -type zinc motif, and each primase recognizes a different trinucleotide sequence (see Introduction). We have replaced the zinc motif of the T7 63-kDa gene 4 primase with those found in the T4 gene 61 primase and the E. coli DnaG primase in order to examine the properties of the resulting chimeric proteins in vivo and in vitro. The expression plasmids used to produce the chimeric primases were constructed by replacing the DNA sequence encoding a 15-amino acid sequence of the T7 primase with the corresponding regions derived from T4 gene 61 or E. coli DnaG (Fig. 1). The N-terminal fragments derived from T4 and E. coli encode 24 or 17 amino acids, respectively, and constitute the zinc motifs of these primases. The T4 and E. coli DNA fragments were obtained by PCR amplification of the appropriate sequences from T4 and E. coli DNA. The resulting DNA fragments were then inserted into the T7 gene 4 expression vector pGP4 -6G64 replacing a segment of T7 gene 4 encoding a 15 amino acid residue found in the loop region of the zinc motif. The sequences were confirmed by DNA sequence analysis. The N-terminal amino acid sequences, derived from the DNA sequence, of the chimeric primases are shown in Fig. 1.
Complementation Analysis of Chimeric Primases-In order to determine if the chimeric primases function in vivo, plating efficiencies of T7 phage lacking gene 4 (T7 ⌬4-1) were compared on E. coli cells containing the expression vectors for T7, T4/T7, or E. coli/T7 chimeric proteins (Table I). T7 ⌬4-1 cannot grow on E. coli in the absence of exogenous gene 4. In agreement with earlier studies (33, 34) E. coli cells containing plasmids encoding both the 56-and 63-kDa gene 4 protein or the 63-kDa gene 4 protein alone support the growth of T7 ⌬4-1. However, a plasmid encoding the 56-kDa helicase does so less efficiently (0.8 ϫ 10 3 p.f.u./ml). A mechanism allowing T7 ⌬4-1 to grow at all in cells expressing only the 56-kDa helicase is presently unknown. The essential function of the zinc motif is illustrated by the reduction of growth of T7 ⌬4-1 in cells harboring a plasmid that encodes a 63-kDa gene 4 protein in which two of the four cysteines of the zinc motif have been replaced with serine, a finding in agreement with earlier studies (21).
As shown in Table I a plasmid expressing the chimeric protein in which the zinc motif of the T4 primase is substituted for that of the gene 4 primase can support the growth of T7 ⌬4-1 phage, albeit at a lower efficiency (2.4 ϫ 10 5 p.f.u./ml). On the other hand, the E. coli/T7 chimeric primase appears to support the growth of T7 ⌬4-1 at an efficiency similar to that observed with the plasmid expressing only the 56-kDa helicase (1.8 ϫ 10 3 p.f.u./ml).
In order to determine unequivocally if the zinc motifs of the T4 and E. coli primases of the chimeric primases can support T7 growth, we replaced Cys-36 and Cys-39 in each primases with serine, an alternation that inactivates T7 primase ( Table  I). As shown in Table II the mutation in the plasmids encoding each of the chimeric proteins reduced greatly the ability of the chimeric protein to support the growth of T7 ⌬4-1. These results show that the chimeric primases can function in vivo, although at greatly reduced efficiency. The apparent inability of the E. coli/T7 primase to support T7 ⌬4-1 growth (Table I) is most likely due to our finding that the chimeric protein has significantly lower dTTPase activity for translocation (see below). Hence, a direct comparison between T7 ⌬4-1 growth on cells expressing the T7 helicase alone and the chimeric primase can not be made.
Purification of Chimeric Primases-The reduced ability of the T4/T7 and E. coli/T7 chimeric primase to support T7 growth could arise from several defects in the proteins. One simple interpretation is that the conformation of the zinc motif in the protein is altered in such a way that its interaction with the primase recognition sites is weakened. More interesting would be a requirement for recognition of the T7 primase recognition sequence 3Ј-CTG-5Ј for efficient DNA replication, a sequence different from that recognized by the T4 or E. coli primases. Alternately, the defect could reside in the helicase portion of the protein, limiting translocation to primase sites and perhaps even reducing helicase activity. Finally the chimeric protein could be defective in its interaction with other replication proteins such as the T7 DNA polymerase or gene 2.5 protein. In order to explore these possibilities we have purified the two chimeric proteins and characterized them biochemically.
b The mutant T7 bacteriophage which lack the both of gene 4 proteins.
c Plaque-forming units on E. coli C600 cells with plasmids encoding the indicated gene 4 proteins. (Fig. 2). The purification procedure was essentially that previously described for wild-type gene 4 protein with the exception that the cell pellet was extracted with high salt since the overexpressed chimeric proteins were less soluble than the wild-type gene 4 proteins.
Oligoribonucleotide Synthesis-The ability of the two chimeric primases to function even to a limited extent in vivo suggests that they catalyze the synthesis of oligoribonucleotides that can be used as primers by T7 DNA polymerase. The oligoribonucleotide synthesis assay provides a direct analysis of RNA primer synthesis (3,10). As shown in Fig. 3, the wildtype gene 4 primase catalyzed the synthesis of di-, tri-, and tetraribonucleotides on M13 DNA. In this assay, triphosphate was removed from 5Ј-termini by phosphatase, and the resulting products were separated by denaturing PAGE. Oligoribonucleotide products were visualized by [␣-32 P]NTP incorporation and autoradiography.
Both of the chimeric primases catalyzed the synthesis of oligoribonucleotides, but the amount of protein required to detect oligoribonucleotides was considerably more than that required for wild-type gene 4 protein (Fig. 3). Under the condition of this assay the T4/T7 chimeric primase was 2.5-fold more active than the E. coli/T7 primase. However, the amounts of products with 320 nM T4/T7 primase and E. coli/T7 primase were only 0.78 and 0.29%, respectively, of that obtained with wild-type gene 4 protein, as measured by a Bio-Imaging analyzer. In this assay trimers and tetramers were not detected, but as will be shown below, these products are synthesized under appropriate conditions. dTTP Hydrolysis-The 63-kDa gene 4 primase, like the 56-kDa gene 4 protein, catalyzes the hydrolysis of nucleotides, a reaction stimulated greatly by the presence of ssDNA (44). The energy of nucleotide hydrolysis drives the unidirectional translocation of the protein on ssDNA. As shown in Fig. 4A the two chimeric proteins catalyzed the hydrolysis of dTTP, the nucleotide preferred by T7 gene 4 protein, in the absence of ssDNA at rates similar to those catalyzed by the wild-type T7 primase. However, the hydrolysis of dTTP catalyzed by the two chimeric proteins was not stimulated by ssDNA to the same extent as that observed with the wild-type T7 primase (Fig. 4B). Whereas the rate of hydrolysis catalyzed by the wild-type T7 primase was stimulated approximately 38-fold by ssDNA, the T4/T7 and E. coli/T7 chimeric primases were stimulated only 16-and 5-fold, respectively.
Stimulation of Oligoribonucleotide Synthesis by 56-kDa Gene 4 Helicase-The reduced dTTPase activity observed with the chimeric proteins in the presence of ssDNA indicates either a defect in the binding of the proteins to ssDNA, a reaction that require the presence of a NTP but not its hydrolysis (45), or a diminished ability to translocate 5Ј to 3Ј on ssDNA. In either case, the decreased primase activity on M13 DNA observed with the chimeric proteins could result from an inability of the proteins to translocate to primase recognition sites. For example, mutations in the dTTP binding site of the 63-kDa primase eliminate dTTP hydrolysis, helicase activity, and primer syn-thesis on ssDNA (46). The latter activity can be recovered by wild-type 56-kDa helicase.
The addition of wild-type 56-kDa gene 4, a protein that has helicase activity but no primase activity, to reactions containing either T4/T7 or E. coli/T7 chimeric primase stimulated the synthesis of oligoribonucleotides 2.4-or 3.1-fold, respectively (Fig. 5), whereas the wild-type T7 primase was stimulated by only 1.3-fold. Not only were the amounts of dimer increased but the presence of tri-and tetranucleotide products were now observed. The latter are the only species that can serve as primers for T7 DNA polymerase either in vivo or in vitro (11,47). These results show that the chimeric proteins can still interact physically with the 56-kDa helicase, presumably to form hexamers and to provide translocation activity for the chimeric primase in the complex. The addition of T7 DNA polymerase to the reaction mixtures had no effect on oligonucleotide synthesis (data not shown).

FIG. 3. Oligoribonucleotide synthesis by T7 and chimeric primases.
Oligoribonucleotide synthesis reactions were carried out as described under "Experimental Procedures" using [␣-32 P] rCTP and the indicated concentration of primase. The radioactively labeled products of oligoribonucleotide synthesis assays were treated with alkaline phosphatase, separated by denaturing 15% PAGE, and visualized by autoradiography. The identities of the oligoribonucleotide species are indicated. mases catalyze the synthesis of oligoribonucleotides on M13 DNA. However, to our surprise neither chimeric primase catalyzed the synthesis of oligoribonucleotides on synthetic templates containing the recognition sequences for the T7, T4, or E. coli primases (data not shown). Apparently the chimeric proteins recognize entirely different sequences on ssDNA.
In order to identify the sites at which the chimeric primases initiate the synthesis of oligoribonucleotides we have determined the sequence of the oligoribonucleotides synthesized on M13 DNA. We first identified the 3Ј-nucleotide of the dimers (Fig. 6A) synthesized by the primases by measuring the incorporation of radioactivity from each of the four [␣-32 P]rNTPs in the presence of the other three unlabeled rNTPs into dimers using the gel assay procedure described above. Only [ 32 P]phos-phate incorporated into a phosphodiester bridge is expected to appear in the dimer, since the 5Ј-triphosphate was removed by treatment with phosphatase prior to gel analysis. As expected from the known sequence of dimers, pppApC, synthesized by the T7 gene 4 primase, rCMP was the predominant nucleotide in the second position of the dimer. As shown in Fig. 6A, rCMP was also the predominant nucleotide found in this position of the dimers synthesized by the two chimeric primases.
The knowledge that the sequence of the chimeric dimers was pppNpC permitted the determination of the first nucleotide by oligoribonucleotide synthesis in the presence of each of the four ribonucleoside 5Ј-triphosphate in the presence of [␣-32 P]rCTP (Fig. 6B). Again, as expected from the known sequence of dimers synthesized by T7 primase, synthesis occurred predominantly in the presence of rATP and rCTP. Results with the chimeric primases were equally unambiguous, synthesis of dimers by both the T4/T7 and E. coli/T7 primases occurring in the presence of rGTP and rCTP. The presence of rGTP as the first nucleotide of the dimers synthesized by the chimeric primases was confirmed by carrying out synthesis in the presence of [␣-32 P]rCTP and two of the remaining three rNTPs (Fig. 6C). In the case of the T7 primase, omission of rATP eliminated oligoribonucleotide synthesis and the omission of rGTP eliminated synthesis by the two chimeric proteins. We conclude that both chimeric primases synthesize the dimers pppGpC.
Template Recognition Site-As discussed in the Introduction, T7, T4, and E. coli primases each catalyze the synthesis of oligoribonucleotides at trinucleotide sequences unique for each primase. The 3Ј-nucleotide of the recognition sequence is cryptic in that it is not copied into the primer whereas the other two nucleotides are copied in a template-mediated reaction. It seemed likely therefore that oligoribonucleotide synthesis catalyzed by the chimeric primases was occurring at a trinucleotide sequence, 3Ј-NCG-5Ј, in which N is the cryptic nucleotide. In order to confirm this hypothesis and identify the cryptic nucleotide, we synthesized four 33-mer oligoribonucleotides, each containing the trinucleotide sequence 3Ј-NCG-5Ј; each oligonucleotide contained one of the four deoxynucleotides in the N position. The length of the flanking sequences exceeded those required by the T7 primase (3). As shown in Fig. 6D, both chimeric primases preferred the purines A or G in the 3Јposition but the stringency was not as great as that found for the wild-type T7 primase. For example, quantitation of the activity on each templates (see the legend to Fig. 6) showed that primer synthesis was only reduced by half even T was present at the 3Ј-position.
Interactions with T7 DNA Polymerase-Specific interactions between the T7 56-and 63-kDa gene 4 proteins and T7 DNA polymerase are necessary for efficient and coordinated DNA synthesis at the T7 replication fork (48,49), and a physical interaction between the two proteins has been demonstrated (50). In addition to catalyzing the synthesis of oligonucleotides, the gene 4 protein stabilizes them on ssDNA (50) and facilitates their use as primers by T7 DNA polymerase (10,32). In order to determine if the oligoribonucleotides synthesized by the chimeric DNA primases can be stabilized and used as primers, we have used an assay that couples oligoribonucleotide synthesis to DNA synthesis catalyzed by T7 DNA polymerase. This assay measures both the ability of the proteins to synthesize oligoribonucleotides and to provide functional primers for the DNA polymerase. With wild-type T7 DNA primase, maximal stimulation of DNA synthesis occurred at approximately 60 nM protein (Fig. 7). Neither of the two chimeric proteins stimulated nearly the same level of DNA synthesis; the T4/T7 protein was 0.71% as efficient as the wild-type T7 primase and the E. coli/T7 only 0.39%. Under the conditions of this assay, the concentration of primases, T7 DNA polymerase, and M13 ssDNA were 60 nM, 5 nM, and 10 nM, respectively. Therefore the molar ratio of primase (hexamer), DNA polymerase (monomer), and M13 ssDNA is 2:1:2. We do not know the basis of the biphasic curve obtained in Fig. 7. Such a curve is often seen with wild-type enzyme albeit at lower concentrations. The reaction is difficult to characterize kinetically since it involves at least three activities, helicase, primase, and DNA polymerase. Although the efficiency of priming of DNA synthesis was considerably lower than that observed with wild-type T7 primase, our results demonstrate that both of the chimeric primases can provide functional primers for T7 DNA polymerase. DISCUSSION All DNA primases must have a catalytic site at which phosphodiester bridges are formed between nucleotides. All primases identified to date contain a zinc motif. In the case of bacteriophage T7 the two molecular mass forms of gene 4 protein dramatically illustrate this point in that the catalytic site is located within the 56-kDa gene 4 protein. The 56-kDa helicase cannot synthesize functional tetranucleotide primers, but it does catalyze the template-independent synthesis of random dinucleotides. By contrast, the colinear 63-kDa gene 4 protein that has a zinc motif at its N terminus catalyzes a template directed synthesis of oligoribonucleotides at specific primase recognition sites. As is the case for the T4 and E. coli primases, the recognition site is a trinucleotide sequence in which the 3Ј-nucleotide is cryptic, being required for recognition but not for incorporation into the primer.
What is the precise role of the zinc motif in sequence recognition and primer synthesis? The inability of the 56-kDa gene 4 protein to carry out template-directed synthesis of oligoribonucleotides, even at nonspecific sites, indicates that one role of the zinc motif is to mediate the interaction of the catalytic domain with the template. Such a role is also supported by the finding that substitution of a serine residue for any one of the four cysteine residues in the T7 primase abolishes template-FIG. 6. Identification of the dinucleotides synthesized by the chimeric primases. A, In order to identify the 3Ј-terminal nucleotide of the dimers synthesized by each primase oligonucleotide synthesis assays were carried out in reaction mixtures containing M13 ssDNA and the indicated [␣-32 P]rNTP alone with the remaining three unlabeled rNTPs. After incubation with either T7, T4/T7, or E. coli/T7 primase the products were analyzed as described under "Experimental Procedures" and an autoradiogram is shown above. B, in order to identify the 5Ј-terminal nucleotide of the dimers synthesized by each primase oligonucleotide synthesis assays were carried out in presence of [␣-32 P]rCTP and the indicated unlabeled rNTPs. The products of the reaction were analyzed as above. C, the identify of the 5Ј-terminal nucleotide was confirmed by carrying out oligonucleotide synthesis assays in presence of [␣-32 P]rCTP and two of remaining three rNTPs. The rNTP omitted from the reaction mixture is indicated. The products of the reaction were analyzed as above. D, Identification of 3Ј-cryptic nucleotides of the template sequence recognized by the chimeric primases. The oligonucleotide synthesis reactions contained one of four synthetic 33-mer oligodeoxynucleotides containing the trinucleotide sequence 3Ј-NCG-5Ј where N represents one of the four nucleotides. The sequences of each synthetic oligonucleotide is given under "Experimental Procedures." The identity of the N nucleotide is indicated above each lane. The reaction mixtures contained [␣-32 P]rCTP and the other three unlabeled rNTPs. The products of the reaction were analyzed as above. In addition, the gel shown in D was analyzed using BAS100 Fuji Bio-Imaging analyzer and the amount of radioactivity in the dimers synthesized by the chimeric primases were: T4/T7 GCG (100%), CCG (32%), ACG (102%), TCG (45%). E. coli/T7 GCG (100%), CCG (27%), ACG (89%), TCG (41%). In all experiments the products synthesized by wild-type T7 primase were diluted 5-fold prior to gel analysis. The identity of the oligonucleotide products is shown. FIG. 7. RNA-primed DNA synthesis catalyzed by T7 primase and chimeric primases. The ability of T7, T4/T7, and E. coli/T7 primases to mediate RNA-primed DNA synthesis catalyzed by T7 DNA polymerase was measured as described under "Experimental Procedures" using the indicated amounts of each protein.
directed oligoribonucleotide synthesis but not random dinucleotide synthesis (21). An earlier approach to dissecting the multiple roles of the zinc motif involved the substitution of the zinc motif of the primase of the closely related phage T3 for that of the T7 primase (39). The chimeric primase functioned relatively normally in vivo and recognized the same trinucleotide sequence in vitro as did the wild-type T7 primase. This result was not surprising, since the zinc motifs of the two primases are 50% homologous.
In the present study we examined the properties of two T7 chimeric primases whose zinc motifs were derived from two primases, phage T4 and E. coli, that do not share a large degree of homology with the T7 primase and that are known to interact with template sequences that differ from those recognized by the T7 primase. Despite the difference in the sequences recognized by the three primases, they share certain molecular structures and biochemical properties that make a plausible case for the creation of functional chimeric primases. For example, the three primases contain six conserved regions (boxes 1 through 6) (20) and they all function as a complex with their cognate DNA helicase. More important, the phage T4 gene 61 and the E. coli DnaG primases have a C-X 2 -C-X 2 3-C-X 2 -C and a C-X 2 -H-X 17 -C-X 2 -C, respectively, motifs that are strikingly similar to that of the T7 63-kDa gene 4 primase, C-X 2 -C-X 15 -C-X 2 -C ( Fig. 1) (31,51,52). However, as described in the Introduction, while the T7 primase recognizes the trinucleotide sequence 3Ј-CTG-5Ј, the T4 and E. coli primases recognize the sequences 3Ј-T(C/T)G-5Ј and 3Ј-GTC-5Ј, respectively. The similarity in size and shape of the zinc motifs of these three primases suggests that a substitution of only the zinc loop region of the T4 and E. coli primases for that of T7 primase would not drastically alter the conformation of the T7 enzyme.
The effect of changes in the zinc motif on the function of gene 4 protein can be monitored by the ability of the chimeric proteins to catalyze the hydrolysis of dTTP, an activity that resides in the helicase portion of the gene 4 protein (32,38,53). In the absence of DNA both chimeric primases hydrolyze dTTP at approximately equal rates, suggesting that no major alterations in the active site have occurred. In the presence of ssDNA the hydrolysis of dTTP by wild-type T7 primase is stimulated approximately 40-fold, indicative of its ability to translocate on ssDNA. The hydrolysis of dTTP catalyzed by the chimeric primases is also stimulated by ssDNA albeit to a lesser extent. This result confirms earlier studies suggesting that changes in the zinc motif of the T7 primase can affect helicase activity. For example, substitution of a serine for one of the cysteine residues in the zinc motif not only eliminates primase activity but also reduces helicase activity by 10-fold (21). These results, taken together, suggest that the zinc motif maintains intimate contacts with the helicase domain during translocation on ssDNA.
Although both chimeric primases are drastically impaired in their ability to catalyze template-dependent synthesis of oligoribonucleotides, they do function sufficiently to partially support T7 DNA replication and growth. The addition of wild-type helicase to the chimeric primases significantly enhances oligonucleotide synthesis in vitro, presumably by providing for translocation on ssDNA so that the primase gains access to primase recognition sites. Precedence for this rescuing of primase activity comes from studies with T7 primase defective in translocation due to a mutation in its nucleotide binding site (46). Wild-type 56 kDa gene 4 protein can form heterohexamers with 63 kDa protein (38) and thus provide translocation for the primase. However, coexpression of 56 kDa helicase with the chimeric primases did not increase its ability to provide for T7 growth in vivo (data not shown).
The fact that the chimeric primases recognize a template sequence that differs from that recognized by the T7 primase establishes that the zinc motif of the gene 4 protein is required, at least in part, for the recognition of the 3Ј-CTG-5Ј sequence. The earlier finding that primase activity was abolished by substitution of serine for any of the four cysteines in the T7 zinc motif demonstrates the importance of this motif in primer synthesis (21). However, those results did not allow us to distinguish between a role of the zinc motif in mediating the binding of the protein to the template or in sequence recognition. Likewise, the finding that a T3/T7 chimeric primase recognized the same trinucleotide sequence as did the T7 primase left this question unresolved (39).
An important finding of the current study is that the T4/T7 and E. coli/T7 chimeric primases fail to use the recognition sequences of either the T7 primase (3-CTG-5Ј), the T4 (3Ј-T(C/ T)G-5Ј), or the E. coli (3Ј-GTC-5Ј) primase. Instead, they both recognize the trinucleotide sequence 3Ј-(A/G)CG-5Ј which differs from that recognized by T7, T4, or E. coli primase. For example, while all three of these wild-type primases have a high specificity for a single 3Ј-cryptic nucleotide in the template trinucleotide recognition site, the chimeric primases appear to have loosened this specificity in that they can recognize either a G or A residue at this position. The only reservation in interpreting these results is the fact that, by necessity, we have, in part, deduced the template recognition sequence from the identity of the oligoribonucleotides synthesized by the chimeric primases. Hence, the slight possibility remains that the chimeric primases can bind to different recognition sequences but can not catalyze the formation of phosphodiester bonds at these sites.
A major role of the zinc motifs found in proteins other than DNA primases, for instance transcription factors, is their stable binding to specific sequences in DNA. Unlike the DNA primases these factors bind, in turn, to other proteins that mediate the catalytic activity occurring at this site. In the case of DNA primases the same polypeptide that contains the zinc motif also contains the catalytic site for oligoribonucleotide synthesis. Our results suggest that the zinc motifs found in DNA primases function in two capacities. In one case the zinc motif serves to bind the protein stably to the DNA such that the catalytic domain is properly positioned for the template directed condensation of nucleotides. Second, the zinc motif plays a role in selecting the site of primer synthesis by selectively binding to primase recognition sites which are unique to each primase. In the case of the chimeric primases the exogenous zinc motif fulfills its first role but it fails to address its cognate recognition site. Consequently, the sequence of the oligoribonucleotides is dictated by other interactions between the protein and the template, perhaps involving the catalytic site. The fact that G is the 5Ј-residue of the trinucleotide sequence recognized by T7 primase and by both of the chimeric primases suggests that the catalytic site or at least some other region that was also not altered in the construction of the chimeric primases is involved in the recognition of this position. In fact, it has been shown recently that the nucleotide substrate for primer synthesis can be cross-linked to both a residue within the zinc motif of the E. coli DnaG primase and to residues in the vicinity of boxes 3 and 4 that most likely comprise the catalytic domain (54). The latter region (amino acids 205-268) is homologous to a region in the T7 primase comprising amino acids 105-160 (20). Furthermore, in the case of the DnaG primase it is known that protein-protein interaction can influence primer specificity. In one instance an interaction of the DnaG primase with the DnaB helicase loosens the recognition sequence of the primase from 3Ј-GTC-5Ј to the more general sequence 3Ј-PuPyPy-5Ј (7).
Interestingly, the transcription factors Krox-20, Sp-1, and Zif268 contain zinc fingers that recognize specific trinucleotide sequences (55,56). However, the zinc motifs of these proteins differ from those of the DNA primases described here in several respects: the zinc fingers consist of a Cys 2 His 2 motif, there are multiple zinc fingers that interact with a relatively long stretch of DNA, and recognition occurs on double-stranded DNA. The Cys 2 His 2 zinc motif contains both an ␣-helix and a ␤-sheet with the base contacts occurring through residues in the a-helix (23,24,55).
Structurally, the zinc motifs of the DNA primases more likely resemble the zinc ribbon sequences that contain threestranded antiparallel ␤-sheets around a single Cys 4 motif (25). Examples of proteins that belong to this zinc ribbon family are TFIIS, TFIIE, RNA polymerase II large subunit, DNA polymerase ␣-subunit, and several bacterial repair proteins. Many members of this protein family bind to ssDNA as well as to dsDNA but, unlike the DNA primases, without sequence specificity. Similarly, proteins such as the T4 gene 32 protein and the retroviral nucleocapsid protein that contain another type of zinc motif, bind to ssDNA but in a sequence-independent manner (57). The unique ability of the DNA primases to recognize specific sequences on ssDNA lends support to our hypothesis that the sequence recognition arises as a result of an interaction of the zinc motif with another domain of the primase.