A New System for Studying Molecular Mechanisms of Mutation by Carcinogens*

A new system for studying the molecular mecha- nisms of mutation by carcinogens is described. The system involves (a) site-specific modification of the essential gene G in +X174 replicative form DNA by a combination of chemical and enzymatic steps; (b) pro- duction of mutant virus carrying a change at a single preselected site by transfection of spheroplasts with the site modified +X174 DNA; (c) detection and propa- gation of mutants using a host carrying the plasmid, p+XG, that rescues all type of gene G mutants by com- plementation; (d) identification of the mutation in the progeny virus by isolating and sequencing mutant +X174 DNA in the region that carried the parental, site- specific change. To demonstrate that this system operational, +X174 G + at 2401 the (plus) This pre-planned, (amber) octadeoxynucleotide, strand elongating this enzymatically with Escherichia

A new system for studying the molecular mechanisms of mutation by carcinogens is described. and nonpermissive (without p+XG) host cells. About 1% of the progeny virus were mutants. Out of 15 isolates, 11 were suppressible by an amber Sul+ (serine) or an ochre Su8' (glutamine) suppressor strain, but not by an amber Su3 ' (tyrosine) suppressor. The other 4 isolates were not suppressed at all. Replicative form DNA produced from one of the suppressible mutants was shown (by sequencing) to contain the expected C + T change at the preselected site in the viral strand. Replicative form DNA from one of the nonsuppressible mutants was partially sequenced. No change was found at or around position 2401. The nature of the mutation(s) in these isolates is still unknown. The occurrence of mutations outside the preselected sites represent a potential problem for our projected studies, but additional data is required before the problem can be fully evaluated.
In spite of this, it should be possible to study, in uiuo, the biological effects of any site-specific modification (including covalent modifications by carcinogens) that can be introduced into gene G of +X174 DNA via a synthetic, oligonucleotide primer.
Most carcinogens are mutagens (2). They are also electrophilic reagents that react with DNA to form a variety of stable, covalent adducts (3). It is generally believed that these covalent adducts cause mutations, but the details of this process are poorly understood at the molecular level. With this fundamental problem in mind, we have developed a system to explore the following questions: (a) Which of the different carcinogen-induced covalent modifications of DNA produce mutations?
(b) What kind of mutation(s) does each different kind of premutational lesion produce? (c) What role do the various DNA repair systems play in this mutation process? Briefly, the system consists of four parts: (a) introduction of the site-specific, covalent modification to be studied at a preselected site in gene G of bacteriophage +X174 replicative form DNA by a combination of chemical and enzymatic steps; (b) expression of the modification in viuo by transfection of spheroplasts carrying different DNA repair backgrounds; (c) identification of any mutants that are produced; (d) characterization of the mutations by isolating mutant DNA and sequencing it in the region that carried the site-specific modification in the parental DNA.
While this approach is general, in principle, some important technical requirements must be satisfied in order to have a practical system. For our initial studies, we selected the virulent bacteriophage, +X174, because of its relative simplicity (4,5). We have focused on an essential gene (gene G) in order to maximize the biological effect(s) of mutations produced by site-specific modifications.
Gene G codes for a viral spike protein (6). This protein is required for assembly of an infectious virus particle (5). In addition, gene G is necessary for production of a singlestranded viral DNA from RF' DNA (5, 7). Thus, we can anticipate that many gene G mutations will be lethal. Since we will never know ahead of time what kind of mutation a given site-specific modification will produce, we need a system that is permissive for all kinds of mutation at the preselected site in this gene. Previously, we have described a plasmid system that is permissive for gene G mutations (8,9). The plasmid, p+XG, carries a functional copy of gene G that is expressed in cells carrying the plasmid, even without virus infection. This provides a source of normal gene G product that can rescue gene G mutants by complementation.
We have demonstrated that the system is permissive for temperature-sensitive mutants at the nonpermissive temperature, for lethal missense mutants, for amber nonsense mutants, and for a lethal, site-specific deletion/fiameshift mutant constructed in z&-o. These experiments validate the biological part of the system. This paper deals with the biochemical part of the system. Our objective is to show that a single, predetermined change can be introduced into infectious 4X RF DNA via a small, synthetic, oligonucleotide primer, and that the biological expression of this change can be identified by isolating and sequencing the progeny DNA.
Our approach, summarized in Fig. 1, is patterned after the classical synthesis of infectious +X RF DNA (11,12). At the time we began this work, only a short segment of the $X sequence was known (13). This began 25 residues before gene G and extended 27 residues into the translated portion of the gene (13,14). Since it is much easier to make plus (viral) strands for template than minus strands, we chose to use a minus strand primer. Inspection of the plus strand sequence ( Fig. 1) showed that a C -+ T transition in the 7th residue of the translated portion of gene G changes a glutamine codon to nonsense. Since we could be virtually certain that this change would be conditionally lethal, it seemed a good choice for working out the biochemistry and testing the entire system.
To produce, ultimately, a site-specific C + T change in the viral DNA, we introduced a G + A change into the 4th residue of a minus strand, octadeoxynucleotide primer ( Fig.  1). This corresponds to a change at position 2401 in the current 4X sequence (15) and creates a mismatch between the primer and the template at this position. In order to achieve specific priming at this preselected region of the genome, the octanucleotide primer was elongated enzymatically to a 17mer. The 17-mer was isolated, sequenced, and used to reprime the synthesis of infectious, site-altered $X RF DNA. After transfection of spheroplasts, mutant phage were isolated using a screen that employed cells carrying a functional copy of 4X gene G on a plasmid (mXG) as the permissive host (Fig. 1). The expected nonsense mutant was isolated and used to produce +X RF DNA. This DNA was sequenced and shown to contain a C + T transition at position 2401, exactly as planned.
While these experiments do not deal with covalent modifications by carcinogens per se, they demonstrate that all the methodology that is necessary for these projected studies is operational.

The Priming Reaction
Choice of Primer-For the reasons described above, we selected the minus strand sequence, d-(pT-C-T-A-A-A-A-C), as the initial primer. This octadeoxynucleotide, containing a site-specific change (G + A) at the 4th nucleotide, was synthesized by a modification of the diester route. The purity of the product was established by chromatography on DEAEcellulose in the presence of 7 M urea, nucleoside composition analysis, and sequencing ( 10 occurring at the lower temperature. A computer search of the +X sequence indicated that although there are no sites on the template where this octanucleotide could prime with perfect base pairing, there are 7 sites where it could primer with a single base mismatch. This is illustrated in Fig. 2. In addition to the desired site in gene G, there is another possible priming site in gene G and 5 possible sites in other genes. In order to see which of these sites are actually primed in the enzymatic synthesis, we elongated the 5'-32P-labeled octanucleotide primer with DNA polymerase I (large fragment) (16) in the presence of dATP, dGTP, and dTTP. The expected products are shown in Fig. 2. Note that priming at the preselected site in gene G gives a 17-mer. Priming at other sites gives shorter products that are easily distinguished from the desired product by gel electrophoresis.
One of the possible priming sites cannot be measured in this particular experiment since it does not elongate in the absence of dCTP. Two products from different sites are the same length (lo-mer). Fig. 3 shows the products formed by the elongation reaction in the absence of dCTP. Lane 1 shows the purity of the 8-mer primer by itself. Lane 2 shows the degradation of the primer (in the absence of template) by the enzyme preparation we used. Lane 3 shows the elongation synthesis. Four bands running slower than the primer are clearly visible on this radioautograph.
The slowest (top) band corresponds to the expected 17-mer. The next two bands correspond to a 14-mer and a 13-mer, respectively. Another band, corresponding to a lo-mer is just ahead of the labeled primer (8-mer). Densitom- The sequences in +X174 DNA complementary to the synthetic primer with no more than one base mismatch were located with a computer. The gene involved and the position of the mismatch is shown at the left. The synthetic primer is underlined and its polarity is indicated by the arrowhead. The mismatched bases are boxed. The reading frame is indicated by the bracket over the template strand. The nature of the mutation expected from the mismatch is shown on the right. Priming was measured by enzymatic elongation of the "'P-labeled primer with E. coli DNA polymerase I (large fragment) (16) in the absence of dCTP. The sequence of each elongated product is shown; the length of the product is indicated at the right.
eter tracings gave a ratio of 1:5:8 for the 17-mer, 14-mer, and 13.mer, respectively; the lo-mer could not be measured accurately because of interference from the excess primer (Bmer) .
The 17-mer band was eluted from the gel and sequenced by the . The results are shown in Fig. 4. The first 3 nucleotides (T-C-T) at the 5'-end of the elongated primer do not show up on the radioautograph.
The cleavage reactions are not as specific as is usually found with this method, perhaps because we were working with a singlestranded oligonucleotide rather than a duplex. Nevertheless, the sequence can be read without difficulty. The only possible ambiguities are at residues 9 and 10 (dashed lines in Fig. 4). Comparison of the intensities of bands 7, 9, and 11 indicates 9 corresponds to A, not G. Band 10 corresponds to the 10th residue of the elongated 8-mer. This cannot be C since there was no dCTP in the reaction mixture; it must be T. The 17mer sequence is TCTAAAACATGATTAAA.
Comparison of this sequence with the possible elongated products shown in Fig. 2 indicates that the slowest band in Fig. 3 is the desired 17-mer.
The sequences of the other primed products determined by the Maxam-Gilbert procedure were ambiguous. However, these data (not shown) combined with the other available information were sufficient to correlate the bands marked in Fig. 3 with the elongated products shown in Fig. 2.
We have been able to increase both the efficiency and the specificity of the priming with ribosomal protein Sl. The rationale for this experiment is as follows. The preselected priming site in gene G is part of a strong, ribosomal binding site (13). From inspection of sequence data, it has been suggested that this region of +X DNA has a secondary structure shown in Fig. 5 (13). It seemed likely that this structure might be responsible for the low yield of 17-mer in the elongation reaction. Since ribosomal protein Sl is known to destabilize part of the ordered structure in +X174 DNA (19, 20), we hoped that this protein would bind asymmetrically to this ribosome binding site and destabilize the ordered structure so that priming and elongation could take place more efficiently. Part of the sequence (boned region in Fig. 5) is homologous to a known bacteriophage Q/I RNA ribosomal binding site (21). A quantitative study of priming in the presence of ribosomal protein Sl is shown in Fig. 6. The amount of 17-mer increases to a maximum as the concentration of ribosomal protein Sl is increased. Priming at other sites decreases. At high levels of ribosomal protein Sl, all priming is inhibited.
These results indicate that ribosomal protein Sl binds asymmetrically to the priming region at the beginning of +X gene G, perhaps to the sequence shown by the boxed area in Fig. 5. This binding apparently destabilizes the ordered structure of the template molecule in such a manner that the 8mer can bind and elongate leftward (counterclockwise) giving the 17-mer. Binding of ribosomal protein Sl at other priming sites is apparently nonspecific and inhibitory.
At high concentrations of ribosomal protein Sl, nonspecific binding occurs at the 17-mer priming site as well, and elongation of the 8-mer is inhibited at all priming sites. This effect seems to be specific for ribosomal protein Sl since bacteriophage Tq gene 32 protein, a known DNA helix destabilizing protein (22) as well as E. coli DNA helix destabilizing protein (22), did not stimulate formation of the 17-mer with E. coli DNA polymerase I (large fragment). The results (not shown) indicate that these proteins inhibited priming at all sites.

Synthesis of the Site-specific Gene G Mutant
Enzymatic Synthesis of Infectious +X174 RF DNA Containing a G + A Change at Position 2401 of the Minus Strand-In order to avoid the possible complication of multiple priming, we utilized the 17-mer described above to reprime the synthesis of RF DNA on a wild type, plus strand template in the presence of E. coli DNA polymerase I (large fragment), Tq DNA ligase, ATP, and all 4 dNTP's. After an appropriate incubation, the reaction mixture was treated with single strand-specific nuclease Sl to destroy template DNA that remained and the incomplete duplex molecules (23). This mixture was used directly to transfect E. coli C600 Su2+ spheroplasts.
The Sl nuclease step proved to be important since our initial attempts to find the mutant against a very high wild type background failed.  clease step gives an effective enrichment of only 50. Even this proved important in detecting the mutant. Detection of @IX Gene G Mutants-Phage produced by spheroplasts were screened in two different ways using our pqbXG105 system (8). One method employed a double layer technique utilizing a soft agar layer of cells carrying the plasmid, p+XG, (HF4740 recA/p+XGlOFi) to which an aliquot of phage had been added poured over a solidified layer of the same strain without p+XG (HF4740 recA). The results were never unequivocal, but any plaque with even a slightly turbid appearance was picked. Out of 855 plaques examined, 155 were suspected of being mutant. Of these, 12 were confiied by replating on E. coli C and HF4740 recA/p+XGlOB.
The mutants were then tested on a variety of suppressor strains. The growth pattern showed that two classes of mutants were present. They were designated OBI and OBII. The properties of these mutants on various host strains are summarized in Table II of Mutation for the site at the start of gene G that binds ribosomes (13). The synthetic minus strand primer is aligned with its complementary sequence.
The boxed area is homologous to part of the strong Q/3 ribosome binding site Sl (21) shown in c. b, opening the hairpin loop allowing the primer to bind more stably to the template.
The direction and maximum extent of elongation in the absence of dCTP is shown by the curved arrow.
should be noted that Sul+ suppressors should insert serine instead of the wild type glutamine. The results indicate that this substitution is permissible at 32"C, but not at 38°C. Insertion of tyrosine for glutamine (Su3+) apparently inactivates the gene G protein. We did not have a +X-sensitive Su2+ (glutamine) strain (the "correct" suppressor) available. However, OBI mutants grow on the ochre suppressor, WWU Su2+. This is expected since this ochre suppressor suppresses both amber and ochre mutations and is thought to insert glutamine, the wild type amino acid (30).
Thus, the spheroplast lysate contains an amber mutant with the expected biological properties (OBI type) as well as a second, unexpected mutant (OBII type) that seems to contain a lethal mutation in gene G since it grows only by complementation in p+XG-bearing strains. The second screening method uses a gridding procedure that detects not only gene G mutants, but also differentiates types OBI and OBII in the same assay. Out of 500 plaques examined, 3 were found to be mutants.3 Two of these 3 mutants (OBI type) grew on appropriate nonsense suppressor hosts as well as on p+XG-bearing hosts; the other (OBII type) grew only on the plasmid-bearing host. Out of 1355 plaques examined by these two assays, 15 confirmed mutants have been isolated. Of these, 11 are OBI type, 4 are OBII type. These data are summarized in Table I  (Experiment  7). The Nature of the Mutation in OBI-All 11 isolates designated as OBI type mutants give exactly the same suppression pattern (see Table II). We conclude from this and from the method used for their construction that they all contain the same mutation. One of these was investigated further. Mutant virus was purified from single plaques grown on the permissive strain HF4738 Su,,, recA. Stocks containing lop6 wild type ' Some may have been missed because of high wild type content in the initial plaques obtained on HF4738 Su' recA/p+XG105. We have reason to believe that the particular culture used in this experiment may have reverted-to recA+. We have described the problem of wild type contamination with certain polar mutants grown in recA+ p+XG105-bearing strains (8).
+X174 were grown in liquid culture and used to infect either E. coli HF4738 or C in the presence of chloramphenicol to produce RF DNA. This DNA was sequenced by the Maxam-Gilbert procedure as described under "Experimental Procedures." The results are shown in Fig. 7a. Residue 2401 is clearly a T rather than the wild type C. The remaining sequence is wild type. Thus, the minus strand A at the position 2401, introduced into heteroduplex DNA via a short, synthetic primer of known sequence, gives rise in viuo to a T in the viral DNA, as expected, producing a nonsense mutation at the preselected site in gene G. This experiment demonstrates that all phases   (2) 3 (1) a Experiments 1 to 4 were not carried through any of the steps used in the final enzymatic synthesis. Experiments 5 to 7 were carried through the entire synthesis protocol. * A 17-mer (see Fig. 1  The Nature of the Mutation in OBII-The sequence of OBII type mutant DNA in the 2401 region was determined in the same manner as OBI. The results are shown in Fig. 7b. The sequence is wild type throughout the region complementary to the 17-mer used as a primer. The mutation must be in gene G because OBII grows only on p+XG-carrying strains. The position and the nature of the mutation are unknown. The DNA's from the other 3 OBII isolates have not yet been sequenced.

DISCUSSION
The purpose of this work was to establish the feasibility of using an essential gene of phage +X174 for studying the biological effects of site-specific, covalent modification of DNA by carcinogens. By producing a previously unknown, nonsense mutant (+X Gam2401 OBl ), carrying a single base change at a preselected position (2401) in gene G of $X174, we have shown that the +XG system ( Fig. 1) is technically operational. The use of a short, chemically synthesized primer for the enzymatic synthesis of infectious I$X RF DNA carrying a single base mismatch at position 2401 suggests that a wide variety of interesting covalent modifications can be introduced by this route. Our earlier work has demonstrated that host cells carrying the plasmid, p+XG105, are permissive for all types of mutation (missense, nonsense, frameshift, and large deletion) in gene G (89). Thus, it should be possible to isolate any mutant phage derived from a site-specific, covalent, gene G modification of infectious RF DNA providing the mutation produces a recognizable biological effect. The successful isolation of c$X Gam2401 OBl using a host carrying p+XG105 and the identification of the mutation by sequencing indicates that the +XG system is ready for our projected studies on the molecular mechanisms of mutation by carcinogens.
Our results also have more general implications for the technology of site-specific mutagenesis that has been developed during the past 2 years. The particular synthesis described here utilized a chemically synthesized primer. Although not novel, this is important for our work because of the flexibility the chemical approach provides. Our results show that a very short primer (8mer) can be used for the enzymatic synthesis of site-modified, infectious, RF DNA even when the initial priming reaction is nonspecific and inefficient. We have shown that the AT-rich primer, d-(pT-C-T-A-A-A-A-C), anneals with the viral strand template at several places, including the preselected site, with a single base mismatch (Fig. 2). Specificity for the preselected site was achieved by elongating the 8mer using E. coli DNA polymerase I (large fragment) (lacking 5' + 3' exonuclease activity) in the presence of dATP, dGTP, and dTTP. The elongated product (a 17-mer) was isolated and used to reprime the synthesis of RF DNA. This relay approach should be fairly general. Not only does it provide a means to achieve specific priming with a short obligomer, but it may provide a convenient way of cleaning up chemically synthesized primers. This is important because mistakes in the primer can lead to mutations that have nothing to do with the modification under investigation.
Our priming results can be compared with those reported recently by Hutchison et al. (23) since our synthesis of +X Gam2401 OBl was very similar to their site-specific synthesis of the well known +X Earn3 mutant. These workers examined priming by a 7-mer and a 12-mer carrying the Earn3 nonsense mutation. The desired priming was achieved with the 12-mer on either a heterologous (wild type) or homologous (am3) template.
The am3:7-mer, d-(pG-T-A-T-C-C-T), did not prime at the am3 site on the heterologous (wild type) template. The authors attribute this to the T,G mismatch between the 3' terminus of the primer and the template at the am3 site. The am3:7-mer did prime on the homologous template at a site in gene H. The authors attribute this to the perfect base pairing between the primer and the template that occurs fortuitously at the gene H site. Our data (Fig. 3) show that the &mer, d-(pT-C-T-A-A-A-A-C), primes at several sites, each having 7 base pairs and a single mismatch (Fig. 2). It may be significant that we found no priming in gene C (Fig.  2) where there is a purine-purine mismatch between the primer and the template. Priming in gene F (lo-mer, Fig. 2) would also have a purine-purine mismatch. The data for this site are ambiguous. Thus, it appears that short primers with a purine-pyrimidine or a pyrimidine-pyrimidine mismatch will work providing the mismatch is not too close to the 3'-end where it can be edited by the 3' -+ 5' exonuclease activity of the polymerase (32).
The poor efficiency of priming by d-(pT-C-T-A-A-A-A-C) at a preselected site in gene G cannot be attributed solely to the primer sequence. The results with the helix destabilizing protein, ribosomal protein Sl (Figs. 5 and 6), show that ordered structure of the template in this region of +X DNA plays an important role in inhibiting the priming reaction. Even in the presence of this helix destabilizing protein, the isolated yield of the desired 17-mer was only about 5% of the template input. This probably does reflect the length and the composition of the primer. Thus, it appears that both ordered structure of the template DNA and the primer sequence influence the priming reaction.
The poor yields, both in the elongation reaction and the RF synthesis from the 17-mer, caused considerable difficulty and made it impossible for us to characterize the RF DNA product biochemically.
Assuming the data in Table I are representative and that the OBI mutants are all the same, we calculate that the yield of site altered RF DNA (before Sl nuclease treatment) was 0.03% based on template input. Goulian et al. reported a yield of about 30% in a similar synthesis using a boiled extract of E. coli as a source of primers (12). Razin et al. report a yield of 0.1% using a synthetic 17-mer primer derived from a +X gene E (lysis gene) sequence (33). We suspect our poor yield reflects not only inefficiency of priming with a 17-mer in this particular region of the $X genome, but also the difficulty in copying through ordered structure in the template to complete the circle. In spite of this very low yield, we were able to isolate the desired mutant virus without difficulty because of the enrichment provided by the Sl nuclease step, the biological amplification provided by the spheroplasts, and the simplicity of the p+XG screen for finding gene G mutants.
Once this is done, the mutant DNA can be obtained in large amounts from the phage.
The most important feature of the $XG system for sitespecific mutagenesis is its generality. This is true not only for the biochemical part of the system, but also for the biological part. For example, the transfection experiments were carried out with spheroplasts derived from the E. coli K12 derivative, C600 Su2'. This contains an amber suppressor that can read the site-specific amber mutation in the third codon of gene G and insert the correct amino acid, glutamine. However, this suppressor is probably unnecessary when a heteroduplex DNA containing a wild type template strand and a mutant complementary strand is used, as was the case in our experiments. Wild type RF DNA produced from the template strand should produce wild type gene G product so complementation should be set up automatically.
Thus, infectious virus should be produced regardless of the nature of the modification in the RF DNA.
Rescue of mutant phage from spheroplast lysates with host cells carrying the plasmid p+XG also seems to be general. The results reported here reinforce our earlier data (8,9). They also demonstrate that one needs no prior knowledge concerning the nature of the gene G mutation in order to isolate it with p@XG-bearing hosts. For example, we had originally intended to use the suppressor approach to isolate the sitespecific, amber mutant reported here since the appropriate +X-sensitive, Su2+ host had been described (E. coli WWU Su2+) (31). However, this particular mutant was not available. We did have +X-sensitive Sulf (amber -+ Ser) and Su3+ (amber --\ Try) suppressor strains on hand, but we had no way of knowing ahead of time whether the missense substitutions that should result with these suppressors would give a functional gene G product. Therefore, we used a p$XGbearing host to detect and isolate the gene G mutants from the spheroplast lysate. Once the nonsense mutant +X

Gum2401
OBl was isolated, we were able to show that it was suppressed by Sul+, but not by Su3+. It could easily have turned out that neither of these suppressors worked, and without p+XG we would not have been able to isolate the desired mutant even though it was present in the spheroplast lysate.
These results demonstrate another important use of sitespecific nonsense mutations.
The finding that 4X Gum2401 OBl is suppressed by Sul', but not Su3' hosts shows that the third amino acid (Gln) in the +X174 gene G spike protein can be replaced by serine, but not by tyrosine. Recent experiments (not reported here) suggest that the serine replacement produces a temperature-sensitive viral spike protein. Another feature of the p+XG-bearing system is its sensitivity and the ease with which it can be used as a mutant screen. The isolation of OBII type mutants illustrates this. Four isolates out of the 1355 plaques examined grow only on p+XGbearing hosts. One of the OBII mutants has been partially sequenced. It does not contain a mutation at the preselected site (position 2401). Since the nature and the origin of these mutants are unknown, it is possible that all four isolates are different mutants. Therefore, mutant frequency is between 7 x 10m4 and 2 x lo-" yet there was no difficulty in detecting these mutants.
Assuming all OBI type mutants are the same, their apparent frequency compared to wild type phage was about 1% of the 1355 plaques examined. This is much less than the 15% reported by Hutchison et al. (23) for the +X Earn3 mutant synthesized from a 12-mer primer, but about the same as reported by Razin et al. (33) for the Earn3 revertant synthesized from a 17-mer primer by a procedure similar to that described by Hutchison et al. The data in Table I suggest that increased digestion with Sl nuclease would have raised the efficiency since we only reduced the template DNA by a factor of 320; Hutchison et al. reduced it by a factor of 1000. None of the data from the three site-specific syntheses reported so far (23, 33, this paper) should be regarded as true efficiencies of mutant production from site-modified +X RF DNA because in no case has the DNA actually used for transfection been isolated and characterized biochemically.
Poor yields in the primed, enzymatic synthesis made it impossible for us to do this. By gel electrophoresis, Razin et al. demonstrated the formation of RF DNA in their enzymatic synthesis, but they used a crude reaction mixture for transfection (33). So did Hutchison et al. (23).
After the classical enzymatic synthesis of $X174 Earn3 RF DNA in 1967 (11) and the demonstration that the DNA was infectious (la), Goulian and Kornberg pointed out, "The implication for studies of mutagenesis becomes clear at once, in as much as a variety of base analogs or ribonucleotides can now be incorporated into an infectious molecule" (11). Before meaningful studies of this kind could be attempted, however, three technical problems had to be solved: (1) development of techniques for obtaining sequence data from interesting regions of biologically active DNA; (2) development of methodology for introducing any desired site-specific modification at a preselected site in a suitable DNA; (3) development of methods for detecting and propagating any kind of mutation at the preselected site.
Through the efforts of several laboratories, all the necessary technology is now available and a new range of well controlled experiments bearing on important questions in molecular biology is possible. How covalent modifications of DNA by carcinogens produce mutations is one such question. Before we can use the +XG system to its fullest potential some additional calibration studies are necessary since the biological properties of some missense mutant that may be produced are unknown at present. Even without this, it appears that considerable progress can be made with the information on hand. A complete description of our strategy is beyond the scope of this paper, but the site-specific amber mutant described here plays a key role because it is unique. For example, the viral DNA from this mutant can be used as a template for the synthesis of RF DNA carrying a site-specific modification with a carcinogen in the third codon of the gene G minus strand. Any mutant produced in uiuo from this RF DNA, regardless of the nature of the mutation, can be distinguished from amber mutant virus produced from the template strand. The new mutants can be grouped further as wild type and pseudo-wild type, temperature-sensitive, ochre, and lethal. Out of nine possible transitions and transversions that can arise from site-specific modifications of the bases in this amber codon, the biological properties of five are already known (wild type, ochre, Ser = temperature-sensitive, Tyr = lethal). We expect frameshifts, large deletions, and rearrangements in this essential gene to be lethal. Therefore, only the properties of missense mutants carrying Trp, Leu, Glu, and Lys, instead of wild type Gln, are uncertain. These missense mutants can be constructed using the general approach described in this paper, and their biological properties can be determined. Even without this information it should be possible to identify biologically two of the three possible transitions arising from modifications of the three bases in this codon since they lead to wild type, ochre, and Trp codons, respectively.
Similarly, three of the six possible transversions can be detected biologically since we know that Ser missense is temperature-sensitive and Tyr missense is lethal.
The lethal mutation group is expected to be the most complex. We already know it will contain two transversions leading to Tyr. We expect frameshifts, large deletions, and rearrangements to be lethal as well. At present these different mutations will have to be identified by sequencing. This presents a problem only if the mutation is a minor event compared to other lethal mutations.
Our approach is not limited to this amber codon, of course. As new mutants are isolated, either by synthesis of site-altered RF DNA or as a result of a site-specific mutation from a modified residue, the mutant DNA becomes easily available for use as a template. The only limitation that is clear at present is that virus from the template strand should be easily distinguished from the mutants one hopes to find. This is why nonsense codons in the template are useful.
From our results, as well as theoretical considerations, we are convinced that the questions posed at the beginning of this paper can be approached in a direct manner using the +XG system described here.

Synthesis of d-(""PT.C-T-A-A-A-A-C-A-T-G-A-T-T-A-A-A)
The chemical synthesis of the octanucleotide primer, had traveled by about half a distance on the gel. The product bands were visualized by autoradiography. For sequencing, the required bands (0.1 x 1 x 0.15 cm) were cut, crushed with the help of a glass rod, and extracted with 2 ml of 0.5 M NaCl, 10 mM Tris-HCl, pH 8.0, 10 mM EDTA, and 1 ml of distilled phenol (saturated with 5 x TE buffer) at 37°C overnight with shaking. The aqueous layer was extracted three times with ether and concentrated to 0.5 ml. The solution was extensively dialyzed against 1 mM Tris-HCl, pH 7.5. The elongated (single-stranded) product was subjected to four different chemical treatments that promote random cleavage at specific base, e.g. A + G, G, C, C + T as described by Maxam and Gilbert (18). The partial chemical degradation products were fractionated on a 25% polyacrylamide 7 M urea gel as described.

Effect of Helix Destabilizing
Proteins on Elongation of d-(32pT-C-T-A-A-A-A-C)-Elongation of the labeled octanucleotide on a viral DNA template was investigated under the reaction conditions described above except that the final concentration of MgC12 was 5 mM. The protein (ribosomal protein Sl, Tq gene 32 protein, or E. coli helix destabilizing protein) and viral DNA were mixed in the presence of Tris-HCl, pH 7.4, and 2-mercaptoethanol. After 20 min at room temperature, primer was added and the contents transferred to a bath at 7-8°C. After 2 to 3 h, the remaining contents of the reaction mixture and polymerase were added and the incubation continued for 12 to 14 h. After separation of the products by gel electrophoresis as described above, the radioautographs were scanned in a Beckmann 35 spectrophotometer equipped with a gel scanning accessory. The peak areas were measured by planimetry.
gridding procedure that detected all gene G mutants and partially classified them in the same assay. A suitable dilution of the above lysate was made so as to give about 100 pfu/plate on a lawn of HF4738 Su' recAl/p+XG. Individual plaques were picked with a sterile toothpick and gridded sequentially on E. coli C, HF4738 Su+ recA1 and HF4738 Su' recAl/p+XG. Mutants were identified by comparing the plates. They were classified as OBI if they grew on plates 2 and 3 but not 1; as OBII if they grew only on plate 3. All mutants were confirmed by picking 10 plaques at random from plate 3 and retesting by gridding on fresh test plates. In each case, all 10 plaques grew on the permissive strain(s); none grew on the nonpermissive strain(s). Partial Sequencing of RF DNA from OBI and OBII Type Mutants-The sequence around position 2401 of the plus strand, where a G + A transition had been introduced in the minus strand primer, was determined by the method of Maxam and Gilbert (18). RF DNA was isolated from the mutant as described above (two l-liter cultures). It was cleaved with the restriction endonuclease, Hha I. The reaction mixture (500 ~1) contained 10 mM Tris-HCl, pH 8.0, 10 mM M&12, 10 mM 2-mercaptoethanol, 600 units of endonuclease Hha I, and 100 pmol of RF DNA. Incubation was at 37°C for 1 h. The reaction was terminated by two phenol extractions followed by two ether extractions. The fragments produced were fractionated on 5% polyacrylamide slab gel (0.3 X 20 x 40 cm) in 90 mM Tris-borate. PH 8.3. 2.5 mM EDTA at 400 V, for 12 h. The bands in the gel were stained with 0.05% methylene blue. The 614 bp fragment, H3 (36), was extracted by shaking with 3 ml of 0.5 M NaCl, 10 mM Tris-HCl, pH 8.0, 10 mM EDTA equilibrated with 1 ml of freshly distilled phenol. The fragment was precipitated from the aqueous layer with ethanol.
The 5-phosphoryl groups were removed with alkaline phosphatase and relabeled with radioactive phosphate. A portion of the restriction fragment (-5 pmol based on starting RF DNA) in 50 ~1 reaction mixture contained 10 mM Tris-HCl, pH 8.5; 10 mu MgC12, 10 mM 2-mercaptoethanol, 5 PM [y-'"P]ATP (specific activity 2700 Ci/mmol), and 5 units of polynucleotide kinase. After incubation for 40 min at 37°C 50 ~1 of H20 were added and the reaction terminated by two phenol extractions followed by two ether extractions. After the removal of residual ether, 10 ~1 of buffer containing 100 mM Tris-HCl, pH 7.5, 100 mrvr MgC12, and 100 mM 2-mercaptoethanol were added to it. An 80 base pair fragment with a single labeled 5'-end in the plus strand was generated by cleaving the doubly labeled H3 fragment with 10 units of restriction endonuclease AluI at 37°C for 40 min. This produces four fragments (25, 33, 80, 476 base pairs). Two labeled bands, 476 and 80 base pairs were obtained on a 15% polyacrylamide gel by electrophoresis at 400 V for 12 h. The desired 80 base pair product was isolated as described above. This fragment, containing the expected mutation site, was sequenced.