Cloning and Expression of Gene 4 of Bacteriophage T7 and Creation and Analysis of T7 Mutants Lacking the 4A Primase/Helicase or the 4B Helicase”

T7 gene 4, which is required for DNA replication, specifies two proteins whose coding sequences overlap in the same reading frame: the 4A protein, a 566-amino acid primase/helicase, and the 4B protein, a 503-amino acid helicase whose initiation codon is the 64th codon of the 4A protein. To study better the individual functions of these two overlapping proteins, we made clones that express both 4A and 4B proteins, only 4B protein, or only what we refer to as the 4A‘ protein, in which methionine 64 is replaced by leucine, thereby elimi- nating the 4B initiation codon. These clones provide considerably more gene 4 protein for biochemical analysis than do infected cells. They can also be used to isolate and propagate T7 gene 4 deletion mutants, and we have made T7 mutants which lack all gene 4 coding sequences, or which express 4A’ protein but no 4B protein, or 4B protein but no 4A protein. Analysis of these phage mutants shows that 4A‘ protein without any 4B protein can support essentially normal replication and growth, whereas 4B protein without any 4A protein supports little replication or growth. Apparently, the primase activity of the 4A protein is essential for replication, but the 4B protein is dispen-sable, presumably because the 4A protein also supplies helicase activity. The mutation at amino acid 64 of 4A’ appears to have little effect on 4A function. The rate of replication during normal T7 infection appears to be limited by

T7 gene 4 , which is required for DNA replication, specifies two proteins whose coding sequences overlap in the same reading frame: the 4A protein, a 566-amino acid primase/helicase, and the 4B protein, a 503-amino acid helicase whose initiation codon is the 64th codon of the 4A protein. To study better the individual functions of these two overlapping proteins, we made clones that express both 4A and 4B proteins, only 4B protein, or only what we refer to as the 4A' protein, in which methionine 64 is replaced by leucine, thereby eliminating the 4B initiation codon. These clones provide considerably more gene 4 protein for biochemical analysis than do infected cells. They can also be used to isolate and propagate T7 gene 4 deletion mutants, and we have made T7 mutants which lack all gene 4 coding sequences, or which express 4A' protein but no 4B protein, or 4B protein but no 4A protein. Analysis of these phage mutants shows that 4A' protein without any 4B protein can support essentially normal replication and growth, whereas 4B protein without any 4A protein supports little replication or growth. Apparently, the primase activity of the 4A protein is essential for replication, but the 4B protein is dispensable, presumably because the 4A protein also supplies helicase activity. The mutation at amino acid 64 of 4A' appears to have little effect on 4A function. The rate of replication during normal T7 infection appears to be limited by the amount of gene 4 protein, but too high a level of either 4A or 4B protein is inhibitory to growth. T7 gene 4 is required for replication (Studier, 1969), and it specifies two proteins that are made in approximately equal amounts (Studier, 1972;Dunn and Studier, 1981). The nucleotide sequence indicates that the two proteins are translated in the same reading frame from separate initiation codons and a common termination codon (Dunn and Studier, *Work performed at Brookhaven National Laboratory was supported by the Office of Health and Environmental Research of the United States Department of Energy and by Public Health Service grant GM21872. Work performed at The Pennsylvania State University was supported by National Institutes of Health Grant GM44613 (to K. A. J.) and National Institutes of Health Postdoctoral Fellowship GM13135 (to s. S. PJ. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  1983). The two gene 4 proteins co-purify from infected cells and have primase and DNA helicase activities that are needed for replication (reviewed in Richardson et al., 1987). The 503amino acid 4B protein, expressed and purified from the cloned coding sequence, has helicase but not primase activity (Bernstein and . The 566-amino acid 4A protein was not obtained free of 4B, but the behavior of a mixture enriched for 4A suggests that the 4A protein itself may have both primase and helicase activity (Bernstein and Richardson, 1989).
To understand better the role of the 4A and 4B proteins during T7 infection and their possible biochemical functions and interactions, it is useful to be able to express the two proteins individually or together from the cloned coding sequences. Not only should clones provide a better source of protein than infected cells, but active protein from a clone should allow T7 mutants deficient in 4A or 4B to be propagated, thereby allowing a genetic dissection of the role of the two proteins in T7 infection.
A difficulty in this approach is the toxicity of gene 4 clones to Escherichia coli: clones that can express 4B are quite toxic (Studier, 1991), and our initial attempts to obtain clones that express both 4A and 4B were unsuccessful. Clones have been used by others to produce the 4B and mixed 4AB proteins (Bernstein and Richardson, 1988;Nakai and Richardson, 1988) but have not been described. We also wanted to express 4A alone, which requires some means of preventing the normally associated synthesis of 4B. In this paper we describe solutions to these problems, and we analyze the effect of loss of either 4A or 4B protein on T7 replication and growth. Purification and biochemical analysis of the proteins expressed from clones are described in the accompanying paper (Pate1 et al., 1992).

MATERIALS AND METHODS
Bacteria, Phage Strains, Plasmids, and Expression System-E. coli strains BL21, HMS174, C1757, 011' and R594 have been described (Studier, 1975;Studier and Moffatt, 1986), as have wild-type and amber mutants of bacteriophage T7 and methods for working with them (Studier, 1969). The entire nucleotide sequence of T7 DNA and locations of its genetic elements are given by Dunn and Studier (1983).
The PET vectors are derivatives of pBR322 and contain a promoter that directs transcription by T7 RNA polymerase counterclockwise across the BamHI cloning site (Rosenberg et al., 1987;Dubendorff and Studier, 1991) (refer to Fig. 1 for orientations of various elements in the plasmids): PET-1 contains the strong $10 promoter; PET-10 contains the T7lac promoter, which is tightly controlled by lac repressor supplied by the lacl gene in the vector. Genes under control of a T7 promoter, as in a PET vector, can be transcribed by the T7 RNA polymerase made during T7 infection, thereby complementing defects in the phage (Campbell et al., 1978; 1. Plasmids pAR3729 (4A,B), pAR5018 (4A'), and pAR3708 (4B). Positions and direction of transcription are shown for the A1 promoter for E. coli RNA polymerase (pAR3729), the T7hc promoter (pAR5018), the T7 610 promoter (pAR3708), and the T 7 44c and 64.3 promoters. The plasmids derive ultimately from pBR322. Predicted plasmid sizes: pAR3729,6,569 bp; pAR5018,7,566 bp; pAR3708, 6,252 bp. Homologous plasmids with changes in promoter composition are listed in Table 11.

1981). They can also
be induced by IPTG' in hosts such as BL21(DE3), HMS174(DE3), or C1757(DE3), which contain a chromosomal copy of the gene for T7 RNA polymerase under control of the lacUV5 promoter (Studier and Moffatt, 1986;Studier et al., 1990). The anti-sense cloning vector pBR322(Al), isolation number pAR1960, carries a 286-bp fragment of T7 DNA (bp 258-543) containing the strong A1 promoter for E. coli RNA polymerase, which directs transcription clockwise across the BamHI site (further details available from A. H. R.).
Gene 4 Clones-The gene 4 clones used in this work were constructed by standard techniques (Studier and Rosenberg, 1981;Maniatis et al., 1982). All of the clones and vectors are based ultimately on pBR322, and genes were typically cloned in the counterclockwise orientation in the BamHI site by means of BamHI linkers. Cleavage at the unique PstI site in the bla gene, and partial BamHI cleavage was a convenient way to obtain fragments from clones and vectors that could be joined to create homologous plasmids with changes in promoter composition (see Table 11). Enzymes used in modifying DNA were obtained from New England Biolabs, Boehringer Mannheim, or Bethesda Research Laboratories. Oligonucleotides were synthesized by a Systec 1450 or AB1 380A synthesizer. Nucleotide sequencing was done chemically (Maxam and Gilbert, 1979) or enzymatically (Sequenase, U. S. Biochemical Corp.).
A clone of a T7 DNA fragment carrying the complete wild-type coding sequence of genes 4A and 4B was obtained in the anti-sense cloning vector pBR322(A1) but not in pBR322 itself. This clone, pAR3729 (Fig. l ) , carries a PuuII-PflMI fragment (bp 11,483-13,384 of T7 DNA) that begins within the gene 3.8 coding sequence 82 bp ahead of the 4A initiation codon and ends within the gene 4.3 coding sequence 121 bp past the shared 4AB termination codon (Fig. 2). This fragment also contains the complete coding sequences of the putative gene 4.1 and 4.2 proteins, whose coding sequences overlap the gene 4 coding sequences in a different reading frame (Fig. 2). The 610 promoter was inserted ahead of the 4AB coding sequence by fusing the appropriate PstI-BamHI fragments from PET-1 and pAR3729 to create pAR3730.
Codon 64 of the 4A coding sequence in pAR3730 was changed from ATG (methionine) to TTA (leucine), thereby eliminating the initiation codon for the 4B coding sequence. At the same time, a silent change was introduced into codon 63 (proline, CCA to CCG) to create a new HpaI site (GTTAAC). The total change was from CCA ATG ACT to CCG TTA ACT. This set of changes in 3 base pairs is referred to as the gene 4A-ML64 mutation, and the resultingprotein is referred to as the 4A' protein.  region of T7 DNA (Dunn and Studier, 1983) are shown at the top of the figure. Precise locations of the cloned DNAs, the phage deletions, and the 4A-amS4 and 4A' mutations ( X ) are given under "Materials and Methods." pAR3730 and replacing it with a synthetic double-stranded DNA having the above changes. Replacement involved partial digestion with Bsu36I (two sites in pAR3730, both in gene 4) and complete digestion with HpaI (unique site). Mutant plasmids were detected by colony hybridization, using one of the mutant synthetic DNA strands that had been 5' end labeled with 32P, and identification was confirmed by cutting with HpaI. Determination of the nucleotide sequence of the complete 4A' coding sequence of the mutant plasmid selected for further work (pAR5000) verified that only the above changes had been introduced. The A1 anti-sense promoter was removed by joining appropriate PstI-BamHI fragments from pAR5000 and pBR322 to create pAR5001. The 410 or T7Zac promoter was inserted ahead of the 4A' coding sequence by fusing the appropriate Pstl-BamHI fragments from PET-1 or PET-10 and pAR5001 to create pAR5007 and pAR5018 (see Table I1 and Fig. 1).
The cloned 4B coding sequence contains bp 11,575-13,384 from T7 DNA, beginning 10 bp downstream of the 4A initiation codon and ending at the same downstream PflMI site as the other gene 4 clones. Thc fragment was made by limited Bal-31 digestion of a BglII-BglI fragment from T7 DNA (bp 11,515-13,523) followed by cleavage with PflMI and attachment of BamHI linkers. Cloning this fragment in the BamHI site of pBR322 created pAR3701, in which the nucleotide sequence across the vector/T7 DNA junction at the Bal-31 end was determined. The 610 or T7lac promoter was inserted ahead of the 4 8 coding sequence by fusing the appropriate PstI-BarnHI fragments from PET-1 or PET-10 and pAR3701 to create pAR3708 and pAR5019 (see Table I1 and Fig. 1). 7'7 Gene 4 Mutants-Mutations that altered one or both of the gene 4A or 4B coding sequences were constructed in plasmids and then introduced into T7 phage by homologous genetic recombination. In some cases, enrichment for the desired recombinant was accomplished by selecting for a nearby recombination between a mutant phage and wild-type T7 DNA in the plasmid. Restriction analysis of the phage DNA was used where feasible for identifying the recombinant. The correctness of each mutation was confirmed by determining the nucleotide sequence of the mutant phage DNA in the region of the mutation. Pertinent information about the construction and growth of the gene 4 phage mutants is given in Table I. An amber mutation was placed at the fourth codon of the 4A coding sequence by replacing serine codon TCG with TAG to create the 4A-amS4 mutation. To do this, two polymerase chain reaction products were obtained from T7 DNA by use of Taq DNA polymerase (Cetus) and oligonucleotide primers that introduced the amber mutation (C to A at bp 11,575), an NcoI site CCATGG (G to C mutation at bp 11,563, 2 bp ahead of the 4A initiation codon), and BamHI sites at the ends. The two polymerase chain reaction products, containing bp 11,229-11,568 and 11,563-11,972 of T7 DNA, were joined through their NcoI sites and cloned in the BamHI site of pBR322. The mutations were introduced into T7 by recombination with wild type, and 2 out of 650 progeny phage carried the amber mutation, as determined by spot test on BL21 and C1757 (supD). The 4A-ML64 mutation was introduced into T7 by recombination between the gene 4 amber mutant 4-245 and plasmid pAR5001. About 5% of the progeny phage lost the amber mutation, and HpaI restriction analysis indicated that three out of six of these recombinants acquired the 4A-ML64 mutation.
T7 DNA fragments used for creating deletion mutants were cloned by means of BamHI linkers into the BamHI site of pBR322, all in the silent orientation, (Studier and Rosenberg, 1981). Deletions were constructed in a plasmid by joining cloned T7 DNA fragments from two parental plasmids, using PstI and a second restriction enzyme to generate the appropriate fragments for joining. Cloned T7 fragments were joined through their BarnHI linkers (A3.8-4.5) or at restriction sites within the fragments (AseI and BglII for A3.8 AseI and Bsu36I for A3.8-4.5; AseI and HpaI for A3.8-4B). Incorporation of the junction into an infecting phage DNA by homologous genetic recombination with the DNA flanking the junction creates the deletion mutant. The lengths of deletion ranged from 353 to 2,327 bp (Table   I), and the lengths of flanking homologies ranged from 500 to 634 bp to the left of the junction and from 234 to 1,780 bp to the right. Lysates were sometimes enriched for the deletion phage by diluting 100-fold in tryptone broth containing 10 mM EDTA and heating at 55 "C for 15 minutes (Studier, 1973a). Lethal deletion mutants require a complementing clone for propagation and can therefore be readily identified by spot tests using complementing and noncomplementing hosts.
We knew from previous work' that gene 3.8 is not essential for growth on laboratory hosts and that a gene 3.8 mutant would grow well on BL21. The lysozyme amber mutant 3.5-lys13a (Silberstein et al., 1975) was the parental phage for A3.8, and 6 out of 60 heattreated phage lost the lysozyme mutation and acquired the A3.8 deletion, as determined by ability to grow on BL21 and by restriction analysis of the phage DNAs.
Defective gene 4 deletion mutants require complementing plasmids for growth, but plasmids that supply active gene 4 proteins can be toxic to both E. coli and T7, causing cell death or inhibiting T7 growth to various degrees dependent on both the plasmid and the host. Both HMS174 and C1757 hosts tolerate gene 4 plasmids well enough to propagate gene 4 deletion mutants; C1757 strains produce higher titer stocks. Plasmid pAR5018, which supplies 4A' protein, is perhaps the most convenient complementing plasmid for growing gene 4 deletion mutants but should not be used to propagate A3.8-4A, which picks up the 4A' mutation by recombination; pAR3729 is used to propagate A3.8-4A. All of the deletions extend farther into gene 3.8 than the cloned fragment in the complementing plasmids, so that homologous recombination between complementing plasmid and the deletion phage should not be able to generate wild-type phage. The parental phage for A3.8-4A and A3.8-4B was wild-type T7, and after heat enrichment 13 out of 270 and 81 out of 270 plaques, respectively, were deletion mutants by spot test. The parental phage for A3.8-4.5 was the gene 5 amber mutant 5-28. About 25% of the progeny phage lost the amber mutation, and 14 out of 90 recombinants acquired the deletion, as determined by spot test.

RESULTS
Cloning the Natural 4AB Coding Sequence-We readily obtained a clone of t h e 4B coding sequence in the BamHI site of pBR322, but several attempts to obtain a clone of the coding sequence for both the 4A and 4B proteins were unsuccessful, whether directly or by reconstituting the complete gene from separately cloned partial gene fragments. Because * Plaque size was either large, small, or tiny.
we had been successful in cloning almost all T7 genes in this site, some of which are very toxic to the cell3 (Studier and Rosenberg, 1981;Studier, 19911, we suspected that something peculiar to the 4A-4B sequence was preventing the cloning. Possible expression of the putative gene 4.1 protein, whose coding sequence lies within the 4A coding sequence in a different reading frame, or the putative gene 4.2 protein, whose coding sequence overlaps the 3' end of the 4AB coding sequence in a different reading frame (Fig. 2), was unlikely to be the problem because fragments containing these coding sequences could be cloned individually.
Based on our previous experience, the most likely impediment to cloning would be the presence of a promoter for E. coli RNA polymerase in the 4A coding sequence ahead of the 4B initiation site. Constitutive expression of 4B protein, known to be toxic, would prevent the clone from being established. Such a promoter would have to be near the beginning of t h e 4A coding sequence, since the 4B clone we obtained begins only 10 bp downstream of the 4A initiation codon.
The consensus promoter sequence for E. coli RNA polymerase has a -35 sequence of TTGACa and a -10 sequence of TAtaaT, with approximately 16-18 base pairs between the two sequences (Hawley and McClure, 1983 and references therein). Examination of the nucleotide sequence near the 4A start reveals a reasonable match to this consensus sequence. A potential -35 sequence TGGACA begins with the second base pair of the ATG initiation codon for 4A, and a potential -10 sequence TAGTGT is separated from it by 16 base pairs. If this potential promoter functions in the plasmid, it should produce translatable 4B mRNA from a 4AB clone but not from the 4B clone (because the putative -35 sequence is missing from the 4B clone).
To overcome the apparent problem of an internal promoter, we tried cloning the 4AB coding sequence with a strong opposing anti-sense promoter downstream of the gene, an approach others have found successful in reducing expression of cloned genes (reviewed by Green et al., 1986). As an antisense promoter we used the A1 promoter from T7 DNA, one of the strongest promoters known for E. coli RNA polymerase (Deuschle et al., 1986b). This cloning was successful, producing plasmid pAR3729 (Table I1 a n d Fig. 1). As discussed in more detail in a later section, the ability of this clone to complement a T 7 m u t a n t that lacks gene 4 demonstrates that the cloned gene is active. Attempts to remove the A1 promoter A. H. Rosenberg and F. W. Studier, unpublished results. from this clone and leave an active 4AB coding sequence were unsuccessful.
Cloning the 4A' Coding Sequence-To make a clone that produces the 4A but not the 4B protein, initiation of 4B protein synthesis must be prevented. The 4B initiation codon is located an appropriate distance from a good ribosomebinding sequence in the sequence CAGGAGGTAAACCA-ATG. Changing the ATG initiation codon would presumably prevent initiation of 4B protein synthesis but would necessarily change amino acid 64 of the 4A protein, possibly altering its activity. The ribosome-binding sequence could be altered without changing the amino acid sequence of the 4A protein, but such alterations seemed less likely to prevent synthesis of 4B protein completely.
We elected to change the 4B initiation codon from ATG to TTA, which changes amino acid 64 of the 4A protein from methionine to leucine. Leucine was chosen as the replacement amino acid because it is the most likely amino acid to replace methionine in equivalent positions in related proteins, according to the Mutation Data Matrix of Dayhoff et al., 1978. The replacement, referred to as ML64, was made in the cloned 4AB fragment from pAR3730. Once the 4B initiation codon was removed, the anti-sense promoter could also be removed. We refer to this mutant 4A protein as the 4A' protein. Results in succeeding sections, and the activities of the purified protein (Patel et al., 1992), indicate that the 4A' protein probably has little if any deficiency relative to wild-type 4A protein.
Expression of Gene 4 Proteins from Clones-Clones capable of expressing an active gene 4 protein, whether 4A' or 4B alone or the combination of 4A and 4B, seem to be toxic to the E. coli hosts we have tried, presumably because of basal gene 4 expression. Even when a gene 4 clone can be established in a particular host, maintenance of the plasmid can be a problem. Some combinations of clone and host seem to have an appreciable fraction of dead cells in saturated cultures (Studier, 1991). Of the hosts we have worked with, HMS174 and C1757 seem somewhat less sensitive than BL21 and 011' to gene 4 toxicity.
The gene 4 clones listed in Table I1 all have two promoters that can be used by T7 RNA polymerase, the 44c promoter within the gene 4 coding sequence and the 44.3 promoter following the coding sequence (Fig. 2). Transcription from these promoters should proceed around the plasmid to produce gene 4 mRNA (Fig. l), so the cloned genes should be transcribed by the T7 RNA polymerase produced during T7 infection (Studier and Rosenberg, 1981;McAllister et aL, 1981). These promoters can also be used to direct expression of the cloned genes in host cells that provide T7 RNA polymerase (Studier et al., 1990). To try to increase expression levels, we also placed the cloned gene 4 fragments immediately downstream of the strong $10 promoter of PET-1 or the regulated T7lac promoter of PET-10 (Table I1 and Fig. 1).
Attempts to produce gene 4 proteins for purification and biochemical characterization have used DE3 lysogens, which provide IPTG-inducible T7 RNA polymerase (Studier and Moffatt, 1986). HMS174(DE3) and C1757(DE3) were used because they are less sensitive to the toxic effects of basal gene 4 expression than BL21(DE3). The compatible plasmid pLysS was used as a source of T7 lysozyme to reduce basal gene 4 expression by T7 RNA polymerase in uninduced cells (Studier, 1991). In general, pLysS was helpful for stabilizing plasmids having gene 4 under control of the strong 410 promoter but was not needed when gene 4 was under control of the regulated T7hc promoter (Dubendorff and Studier, 1991).
Examples of expression of the 4A' and 4B proteins are given in Fig. 3. As expected, no 4B protein was detectable from the 4A' clone.
We have yet to detect any 4B protein from any 4A' construction, whether in a clone or in T7 itself, and we expect that none is produced. The most sensitive tests we have made (Fig. 3 B ) would have detected at least 1 molecule of 4B protein in 80 molecules of 4A' protein.
The 4A' protein is synthesized at a somewhat higher rate than the 4B protein, and about 10 times as much accumulates after induction (Fig. 3). However, neither the 4A' nor 4B protein is synthesized well compared with many other target proteins that are made efficiently in the T7 expression system, and neither continues to be synthesized at its maximal rate beyond an hour or so. This relatively poor expression is presumably caused by the toxic effects of the accumulating gene 4 protein. Consistent with this interpretation, we have observed that inactive partial or mutant gene 4 proteins can accumulate to considerably higher levels. Nevertheless, expression from the cloned genes produces 40-400 times more gene 4 protein than does T7 infection (compare Kolodner et al., 1978;Fischer and Hinkle, 1980;and Patel et al., 1992).  (1990). Samples (left to right) in each set were prepared by labeling 50 pl of culture for 2 min at 37 "C with 1 pCi of [35S]methionine (425 Ci/ mmol), just before induction, and 30 min, 2 h, and 3 h after induction. 25 pl of 3 X concentrated SDS sample buffer was added directly to the 50 pl of labeled culture, samples were heated at 100 "C for 2 min, and 25 pl of the sample was loaded on a 10-20% SDS-polyacrylamide gradient gel with a 5% stack, as described (Studier, 1973b). Proteins were visualized by autoradiography. Positions are shown for the 4A' and 4B proteins and for the p-lactamase proteins. B, accumulation of 4A' and 4B proteins. HMS174(DE3) cells carrying 4A'-pAR5018 or 4B-pAR5019 were grown shaking a t 37 'C in M9TB medium + ampicillin (20 pg/ml) to A-= 0.8 and induced by adding IPTG to 0.4 mM. Samples were collected 2 h after induction, centrifuged, suspended in 0.10 volume of SDS sample buffer, and heated at 100 "C for 2 min. Extracts were run on an 8-25% SDS-polyacrylamide gradient gel (Pharmacia PhastGel), and the gel was blotted to a nitrocellulose filter (0.45 p~, Bio-Rad) by capillary action. Western analysis was carried out using a 1:lOOO dilution of anti-4A' antibody induced in rabbits. Goat anti-rabbit IgG conjugated to alkaline phosphatase was used as the second antibody, and bands were detected by an alkaline phosphatase colorimetric assay (Bio-Rad). Samples are (left to right) 3-fold serial dilutions of extracts of 4A' and 4B cells followed by a sample of purified 4A' protein.

T7 Mutants Defective in Expression
gation of the phage requires functional gene 4. All of the amber mutants we tested affected both the 4A and 4B proteins, and we did not try to determine whether any temperature-sensitive mutants might be defective in 4A but not 4B.
Tests with gene 4 amber mutants indicated that the 4AB clone pAR3729 can support the growth of gene 4-defective mutants, but experiments with amber mutants are complicated by recombination between the phage mutant and the cloned gene, which produces a significant fraction of wildtype phage. However, this complementing plasmid could be used to isolate and propagate a set of deletion mutants of T7 that entirely or partially eliminate the gene 4 coding sequences ( Table I and Fig. 2). Each deletion was made so that homologous genetic recombination between the mutant phage and the complementing clone would be unable to restore an active gene 4 to the mutant. In practice, this means that the deletion mutant must lack homology with at least one end of the complementing clone. Such mutants can be propagated on the complementing host without accumulating a significant fraction of wild-type (or pseudo wild-type) recombinants and can therefore be studied under nonpermissive conditions without any background from wild-type phages or low level suppression of amber mutations.
Construction of the appropriate deletion mutants was facilitated because the genes flanking gene 4 , namely genes 3.8, 4.3, and 4.5, are not essential for T7 growth' (Studier, 1981). The deletion mutants are named by the extent of the coding sequences they affect, namely A3.8, A3.8-4A, A3.8-4B, and A3.8-4.5. The A3.8 deletion retains 50 bp ahead of the 4A intiation codon, and this mutant would be expected to make both 4A and 4B proteins. The A3.8-4A deletion retains 16 bp ahead of the 4B initiation codon, and this mutant would be expected to make 4B but not 4A protein. The A3.8-4B deletion extends 18 codons into the 4B coding sequence, and this mutant should be unable to make either 4A or 4B protein. The A3.8-4.5 deletion completely removes all gene 4 coding sequences.
The A3.8-4.5 mutant is completely unable to form plaques on normal laboratory hosts but has normal specific infectivity when a host carries pAR3729 (Table 11). Presumably, the cloned gene 4 is expressed by T7 RNA polymerase during infection, beginning at the 44c and 44.3 promoters and proceeding around the plasmid (Fig. 1). Transcription from the anti-sense promoter, which is by E. coli RNA polymerase, should have shut off by the time gene 4 proteins would normally be made during infection (Dunn and Studier, 1983). On the other hand, too much gene 4 protein from a clone can be inhibitory to T7 growth, as indicated by the decreased plaque size and/or plating efficiency of wild-type T7 when the clone has a strong 410 promoter in front of the coding sequences for 4AB, 4A', or 4B (Table 11).
The uninduced T7lac promoter, which is blocked by lac repressor (Dubendorff and Studier, 1991), seems to relieve a slight reduction of plating efficiency caused by the cloned 4A' or 4B coding sequences themselves (compare pAR5018 with pAR5001 and pAR5019 with pAR3701 in Table 11). This slight reduction in plating efficiency is not accompanied by a decrease in plaque size and seems likely to be a result of an increase in the fraction of dead cells in the culture because of a higher basal level of gene 4 protein. Lac repressor bound at the T7lac promoter would presumably block a transcribing E. coli RNA polymerase (Nakamura and Inouye, 1982;Deuschle et al., 1986a), thereby reducing basal gene 4 expression. However, bound lac repressor would have little effect on a transcribing T7 RNA polymerase that started at the 44c and 44.3 promoters (Giordano et al., 1989), and therefore the uninduced T7lac promoter should have little effect on levels of gene 4 protein produced from the clone during infection.
Clones that supply appropriate amounts of 4A' protein support apparently normal growth of A3.8-4.5, but equivalent clones that supply 4B protein support only tiny plaques at low efficiency, a severe growth deficiency (Table 11). Clearly, the 4A' protein is functional and can support the growth of T7 in the absence of 4B protein, but 4B protein without 4A protein is insufficient.
Since 4A' protein from a clone can support T7 growth, we expected that a mutant T7 phage that has the 4A-ML64 mutation, and which therefore should make the 4A' protein but no 4B protein, would be able to grow on normal laboratory hosts. We transferred this mutation from the clone into T7 itself by recombination with a gene 4 amber mutant. About half of the progeny that lost the amber mutation gained the ML64 mutation, and the mutant and wild-type recombinant plaques were indistinguishable. We refer to the mutant as T7-4A', or simply 4A'.
When plated on the complementing host C1757/4A'-pAR5018, purified 4A' mutant phage and each of the deletion mutants had the same specific plating efficiency as wild-type T7. Relative plating efficiencies of these mutant phages on c1757 itself are given in Table 111. A3.8-4.5, which lacks both the 4A and 4B proteins, cannot make plaques at all. It has no homology with the complementing clone upon which it was propagated, and any recombinant phages that might have picked up an active gene 4 by illegitimate recombination constituted less than lo-' of the population. Similarly, A3.8-4B also cannot produce either the 4A or 4B protein and also does not form plaques. However, this mutant has considerable homology with its complementing clone on one end (Fig. 2), so that active gene 4 can be acquired by one homologous and one illegitimate crossover with the complementing plasmid. Such recombinant phages were observed at a frequency of about The A3.8 and 4A' mutant phages give normal plaques on C1757, and the A3.8-4A mutant gave only tiny plaques at low efficiency. Thus, the results with phage mutants (Table 111) and with complementation by gene 4 clones (Table 11) agree that phage infection is completely defective when neither 4A nor 4B protein is present, is very defective when 4B but not 4A is present, and is essentially normal when 4A' but not 4B is present.
These results also predict that an amber mutation in the 4A coding sequence upstream of the 4B start would make 4B but not 4A protein in a nonpermissive host and would be unable to grow. We constructed an amber mutation at codon 4 of the 4A coding sequence in a plasmid by changing the TCG serine codon to TAG and inserted it into the phage by recombination. In contrast to A3.8-4A, the 4A amber mutant formed plaques on nonsuppressing hosts at normal efficiency,

T7 Primase/Helicase Clones and Mutants
with a plaque size that depended on the host: a host such as R594, which has very low amber suppression, gave tiny plaques whereas less stringent hosts such as BL21 or HMS174 gave small to medium plaques. Typically, most of the phage in a plaque remained amber and were not overgrown with wild-type phage. The most likely interpretation is that very low levels of 4A protein are sufficient to allow T7 growth when normal amounts of 4B protein are made.
Expression of Gene 4 Proteins during T7 Infection-Patterns of protein synthesis during infection of BL21 by wildtype T7 and the 4A', A3.8-4A) and A3.8-4B mutants are shown in Fig. 4. In a wild-type infection, 4A and 4B proteins are made in small amounts about 8-12 min after infection, the same time as other replication proteins such as T7 DNA polymerase (gene 5 ) and the single-stranded DNA-binding protein (gene 2.5). Patterns of the gene 4 mutants are similar to wild-type except for the gene 4 region of the gel (we are not able to identify the putative 3.8 or 4.1 proteins): the 4A' mutant makes approximately normal amounts of a 4A protein but no apparent 4B protein; A3.8-4A makes larger than normal amounts of 4B protein but no apparent 4A protein; and A3.8-4B makes no apparent 4A or 4B protein. These are the proteins expected from the genotypes of the infecting phages (Table 111). An unexpected feature of the protein patterns is the larger than normal production of 4B protein during infection by A3.8-4A. Such an increase was not observed in the 4A amber mutant (not shown). The increased rate of 4B synthesis in the deletion mutant seems likely to be the result of translational coupling to the gene 3.5 coding sequence in this mutant: the A3.8-4A deletion places the gene 3.5 termination codon 17 base pairs ahead of the 4B initiation codon, just upstream of the 4B ribosome-binding sequence.
To study the effects of 4A' or 4B protein supplied from a clone during T7 infection, C1757 rather than BL21 was used as a host because it is less sensitive to the toxic effects of the gene 4 proteins. The time course of T7 infection in C1757 is delayed a few min relative to that in BL21, as judged by patterns of protein synthesis, rates of replication, and time of lysis (which is not only slightly delayed but also less abrupt). However, the overall progression of T7 infection is similar in the two hosts. Patterns of protein synthesis during infection of (21757 itself or carrying 4A'-pAR5018 or 4B-pAR5019, where the gene 4 coding sequences are preceded by the T7hc promoter, are shown in Fig. 5. Infection was by wild-type T7 or by the 4A' mutant. The timing and amounts of 4A' and 4B protein produced from these plasmids seem rather similar to the timing and amounts of 4A and 4B protein produced in a normal infection. In the patterns shown in Fig. 5, labeling of the 4A band increased above normal when either phage infected a cell carrying the 4A' clone; the 4B band increased when wild-type infected the 4B clone; and a 4B band appeared when the 4A' mutant infected cells carrying the 4B clone. Measurements of protein synthesis during infection, combined with plating efficiencies of defective mutants (Tables I1  and 111)) suggest that the 4A'-pAR5018 and 4B-pAR5019 clones supply gene 4 proteins during infection at times and in amounts appropriate for growth of phage that lack such proteins and that the increase supplied to a phage which itself makes the protein does not generate a level that becomes inhibitory to growth.
Effects of 4A and 4B Deficiency on Replication"T7 replication is severely deficient when both the 4A and 4B proteins are lacking (Studier, 1969). We tested replication during infection by deletion and point mutants that lack either the 4A or 4B protein.
The 4A' mutant, which lacks the 4B protein, grows essen- tially normally, as indicated by plating behavior and burst size, but it appears to incorporate only perhaps 60% as much [3H]thymidine during infection as does wild type (Fig. 6). This mutant also has a slightly delayed lysis, another indicator of possible replication deficiency (Studier, 1969). The apparent rate of replication in BL21 has an early peak and a later peak (Fig. 6), in contrast to a single peak observed in C1757 (Fig. 7) or in B (Studier, 1969). Apparently, the appearance of a single or double peak depends on the host; the reason for the difference has not been determined, but one possibility might be a difference in kinetics of release of unlabeled nucleotides from host DNA. The A3.8-4A and 4A amber mutants, which make 4B but not 4A protein, have little detectable replication in BL21 (not shown). Thus, 4A protein is needed for replication, and 4B protein alone is not sufficient. The replication results are entirely consistent with the plating behavior of these mutants, if one assumes that the  Fig. 4. 100 p1 of culture was labeled for 2 min with 10 pCi of [3H]thymidine (84.2 Ci/mmol) before and at 2-min intervals after infection. 2 ml of cold 5% trichloroacetic acid was added to stop the labeling and precipitate the DNA. The precipitate was filtered through Whatman GF/C filters and washed several times with 5% trichloroacetic acid and then 95% ethanol. Filters were dried and counted in a liquid scintillati fluor (RPI corp.).  Fig. 6, except that cells were labeled for 2 min at 4-min intervals after infection. moderate decrease from wild-type replication levels shown by the 4A' mutant has little effect on burst size.
We wanted to see if supplying 4B protein to the 4A' mutant would restore wild-type levels of replication. We used the 4A-pAR5018 and 4B-pAR5019 plasmids in C1757 for these tests, complementing conditions that do not inhibit plating efficiency of wild-type T7 (Table 11). As in BL21, incorporation of [3H]thymidine during 4A' infection of (21757 is only about 60% that during wild-type infection (Fig. 7). Supplying either 4A' or 4B protein from the clone not only made [3H]thymidine incorporation by 4A' and wild-type T7 equivalent, but enhanced incorporation 2-3-fold over the normal wild-type level. Lysis was also accelerated, a further indication of the correlation between the efficiencies of replication and lysis.
A simple interpretation of these relative rates of replication would be that the level of gene 4 protein normally limits the rate of replication and that supplying additional gene 4 protein from a clone relieves this limitation. In this interpretation, the 4A' protein would have no intrinsic deficiency, and the reduced replication in the 4A' mutant would simply reflect a reduced total amount of gene 4 activity because no 4B protein is made.

DISCUSSION
Previous work (Bernstein and Richardson, 1989) and results presented in the accompanying paper (Patel et al., 1992) indicate that the gene 4A protein has both primase and helicase activities and that the 4B protein has helicase but not primase activity. Our results demonstrate that 4A protein is required for appreciable replication and growth of T7 and that 4B protein is not required. Thus, the primase activity of the 4A protein appears to be essential, but the helicase activity of the 4B protein is dispensable, presumably because it can be replaced by the helicase activity of 4A.
The rate of replication of wild-type T7 or the 4A' mutant is enhanced by the addition of either 4A' or 4B protein, suggesting that replication during a normal infection is limited by the amount of gene 4 protein. Addition of either 4A' protein or 4B protein would be expected to increase both primase and helicase activity, 4A' because it has both activities and 4B because it has helicase activity and can stimulate the primase activity of 4A (Bernstein and Richardson, 1989;Patel et al., 1992). However, only very small amounts of 4A protein are needed for T7 growth, as demonstrated by the ability of a 4A amber mutant (but not a 4A deletion mutant) to make plaques with normal efficiency on nonsuppressing hosts, in which normal amounts of 4B are made, but 4A protein is made only by basal suppression of the amber mutation. Therefore, primase activity seems unlikely to be limiting normal T7 replication and, by default, helicase activity seems likely to be limiting.
Both the 4A and 4B proteins are very toxic to E. coli cells, and too much of either protein is toxic to T7 growth as well. Since the 4B protein has no primase activity, helicase activity may be primarily responsible for the toxicity of both proteins. Apparently, the levels of the gene 4 proteins must be carefully regulated during T7 infection to have enough to support efficient replication but not so much as to prevent growth. The overlapping 4A and 4B proteins may have evolved to provide an appropriate ratio of helicase to primase activity during infection. If mutants of 4A' that retain primase but lack helicase activity could be isolated, they might permit independent variation of primase and helicase activities and thereby help to define the relative roles of the two activities in replication and toxicity.
The toxicity of the gene 4 proteins made cloning the natural gene difficult and also appears to limit the amount of protein that can be produced from gene 4 clones. However, sufficient protein can be produced from these clones for biochemical and structural analysis (Patel et al., 1992). Our solution to obtaining 4A protein without 4B was to change the methionine initiation codon of 4B to leucine. This 4A' mutant protein is produced without detectable amounts of 4B protein and appears to have normal 4A activity in the tests we have made. We chose what appeared to be a conservative replacement and do not know whether replacing the 4B initiation codon with other amino acids would generate other 4A proteins that have equivalent activities or whether only a limited set of replacements can be tolerated at this position. We have not tried to make the wild-type 4A protein without 4B protein by synonymous codon replacements in the ribosome-binding region upstream of the 4B initiation codon. The ability of appropriate 4A' clones to complement total gene 4 deficiency of T7 should make them useful in further genetic analysis of gene 4 functions.