Abstract
Mutations in the ac gene of bacteriophage T4 confer resistance to acridine-inhibition of phage development. Previous studies had localized the ac gene region; we show that inactivation of T4 Open Reading Frame 52.2 confers the Acr phenotype. Thus, 52.2 is ac. The resistance mechanism is unknown. The ac gene provides a convenient forward mutagenesis assay. Its compact size (156 bp) simplifies mutant sequencing and diverse mutant types are found: base substitutions leading to missense or nonsense codons, inframe deletions or duplications within the coding sequence, deletion or duplication frameshifts, insertions, complex mutations, and large deletions extending into neighboring sequences. Comparisons of spontaneous mutagenesis between phages bearing the wild-type or tsL141 alleles of DNA polymerase demonstrate that the impact of the mutant polymerase is cryptic when total spontaneous mutant frequencies are compared, but the DNA sequences of the ac mutants reveal a substantial alteration of fidelity by the mutant polymerase. The patterns of base substitution mutagenesis suggest that some site-specific mutation rate effects may reflect hotspots for mutagenesis arising by different mechanisms. A new class of spontaneous duplication mutations, having sequences inconsistent with misaligned pairing models, but consistent with nick-processing errors, has been identified at a hotspot in ac.
ACRIDINES inhibit bacteriophage development at concentrations below those that inhibit the bacterial host (reviewed by Adams 1959). The inhibition in vivo is largely manifest at steps other than phage adsorption to the host or macromolecular synthesis. The number of independent targets for in vivo inhibition is unknown. However, based on differences between acridine derivatives and the impact of specific phage mutations influencing acridine resistance or sensitivity, the intracellular targets are likely to be multiple.
The mutagenic effects of acridines stimulated investigations of the interaction of these drugs with DNA and the enzymes that act on it. The mutagenic effects in T4 are mediated by a phage-encoded type II topoisomerase which is stimulated by acridine to cleave the DNA (Ripleyet al. 1988). The nicked DNA intermediate produced in these reactions is converted to frameshift mutations by a nick processing mechanism. At the 3′ end of the nick, bases are added or deleted by the T4-encoded DNA polymerase (Kaiser and Ripley 1995) and this processed end is religated to the unmodified 5′ end. Mutations in the topoisomerase enzyme can confer resistance (Huffet al. 1990) or hypersensitivity (Hessleret al. 1967) to acridines. Acridine hypersensitivity has also been found in T4 phage with mutations in a variety of genes, many of which are DNA synthesis and repair genes, implying that some inhibitory effects of acridines act at this level (Hessleret al. 1967; Woodworth and Kreuzer 1996). Recombinational repair of DNA damage (Altman and Lerman 1970) caused by the acridine-topoisomerase combination is a likely target (Neeceet al. 1996).
Some mutations in gene 17, called q, confer quinacrine and acridine resistance (see Piechowski and Susman 1967, and references therein). The products of gene 17 are required for efficient head-full packaging of T4 DNA (reviewed in Black 1994; Franklin and Mosig 1996). Acridine treatments have been shown by electron microscopy to prevent DNA packaging, consistent with acridine-induced inhibition of this step of phage development (Kellenberger and Sechaud 1957). Mutations at a second locus, called ac, confer acridine resistance by an independent mechanism proposed to be due to decreased permeability of the phage-infected cell to acridines (Silver 1967).
Despite the undefined mechanism responsible for the Acr phenotype, acriflavin selection has been successfully used to monitor T4 mutagenesis (e.g., Koch and Drake 1973; Reha-Krantz and Lambert 1985; Hobbs and Nossal 1996). Although the rII genes of bacteriophage T4 offer remarkable advantages for measuring mutagenesis by reversion, their use for measuring forward mutagenesis (gene inactivation) is significantly arduous and rarely done (Drake and Ripley 1983), except in modified assays using only sequences from the N-terminus of the rIIB gene (Singer 1984).
Mutations that inactivate the ac gene can be directly selected, and, as we show in this report, are in a 156-bp Open Reading Frame (ORF). In contrast, mutations that inactivate the T4 rII genes must be identified using a two-step phenotypic screen. Moreover, the isolated rII mutants are distributed in >3,100 bp of DNA. Thus, the ac assay offers a powerful tool for the analysis of mutagenesis in T4. The number of base pairs that were sequenced to describe the ~250 mutants described in this report would have identified only ~15 rII mutants, if the entire rII genetic region was sequenced for each mutation. The small size of the gene not only decreases the sequencing effort, but also increases the density of mutations in the gene after equal numbers of sequences have been determined. Because of this density and the absence of extraordinary spontaneous hotspots, the sequences of a modest number of mutants often yield multiple examples of the same mutation at multiple sites in the ac gene. Thus, the ac spectrum gives a quantitative view of site-specific variations in mutation rate, which cannot be achieved in larger genes, even with an extensive sequencing effort, since at most sites only a single example of a mutant is found.
We chose to compare spontaneous mutations in ac induced in the presence of wild type DNA polymerase and in the presence of tsL141 DNA polymerase. Previous studies of the tsL141 allele using reversion assays have shown that it increases or decreases spontaneous mutagenesis at specific sites in the rII gene in patterns that are, in part, mutation pathway specific (Drakeet al. 1969; Ripley 1975; Ripley and Shoemaker 1983; Ripleyet al. 1983; Reha-Krantz 1995). It is the ability of this and a few other mutantalleles of the T4DNA polymerase to strongly reduce spontaneous A·T-site transitions (and base analogue-induced transitions of all types) that led to tsL141 being called an “antimutator” allele (Drakeet al. 1969; Drake and Greening 1970). Altogether, the data suggest (Ripley 1981) that spontaneous A·T→G·C base pair substitutions (A·T-site transitions) are strikingly reduced, but that A·T→C·G or T·A base pair substitutions (A·T-site transversions) can be increased. All possible quantitative effects have been seen for frameshifts at different sites, i.e., increased, decreased or unchanged mutation frequencies, in patterns that clearly support the view that quantitative effects reflect mechanism-specific effects as well as site-specific effects (Ripley 1990). Spontaneous G·C-site transition mutation rates have not regularly appeared to be affected, and G·C-site transversions have not been well tested. Only one site (rII amP7) with a particularly low spontaneous rate is definitively identified as reverting only by this pathway (Ripley 1975). However, these mutagenesis pathway interpretations are indirect and not based on mutant sequencing.
The only regular and large influence on spontaneous mutagenesis produced by tsL141 is the reduction of A·T-site transitions. The absence of a detectable generalized antimutator effect, measured by forward mutation (gene inactivation) assays in the r genes of T4, is suspected to arise in large part from the contradictory effects of the tsL141 allele on different mutation pathways (Drake 1993). However, this reasonable supposition has not been directly demonstrated by an examination of mutant sequences. Thus, we have used the Acr phenotype to collect mutants whose sequences will allow us to simultaneously examine the sequence-specific details of the influence of the tsL141 allele on spontaneous mutagenesis.
MATERIALS AND METHODS
Bacteriophage T4 strains and Escherichia coli hosts: T4B is the wild-type strain used in this study. The tsL141 DNA polymerase allele was originally isolated in T4D but was transferred to T4B by extensive backcrossing (Ripley and Shoemaker 1983). This polymerase allele has been used extensively in fidelity studies, and its properties are due to an alanine to valine amino acid substitution at position 737 in the T4 DNA polymerase (Reha-Krantz 1994). Phage were grown and assayed using E. coli BB cells.
Enzymes and chemicals: Taq DNA polymerase (Boehringer Mannheim, Indianapolis) was used in sequencing reactions. Sequencing primers were labeled by T4 polynucleotide kinase (Stratagene, La Jolla, CA) using γ32P-ATP (Amersham, Arlington Heights, IL). Stock solutions of acriflavin neutral (Aldrich Chemical, Milwaukee) at a concentration of 20 mg/ml were prepared in sterile water and stored at 4° in a glass tube protected from light with aluminum foil. Acriflavin is a mixture of 3,6-diamino-10-methylacridinium chloride and 3,6-diaminoacridine.
Media: One liter of M9 broth contains a base solution of 3 g KH2PO4, 6 g Na2HPO4, 1 g NH4Cl, 3.5 g NaCl, and 1 ml of a stock solution containing 0.16 mg FeCl3/ml stored frozen. At the time of use, 2 ml of a separately autoclaved solution containing 0.665 g of MgSO4 and 20 g of dextrose/100 ml, and 4 ml of a separately autoclaved solution containing 20% Bacto Casamino acids were added to each 100 ml of M9 buffer. Plating was carried out on plates containing 30 ml of Drake bottom agar, made with 10 g Bacto tryptone, 1 g Bacto yeast extract, 5 g NaCl, 0.2 g dextrose and 10 g Bacto agar/liter. Top agar contained the same ingredients as the bottom agar in the same proportions at a concentration of dry ingredients of 16 g/liter. Top agar was predispensed in aliquots of 3.0 ml per plate.
Selection for acriflavin resistance: Each Acr mutant was isolated from an independently grown T4 stock. Each stock was initiated by inoculating 5 ml of M9 broth containing log phase E. coli BB cells at a concentration of 7 × 107/ml with phage from a single plaque at 37° for 6 hours. Lysis was completed by the addition of a few drops of chloroform. Selection for Acr was carried out by streaking samples from each T4 stock on acriflavin-containing plates seeded with log phase E. coli BB cells. Cells and acriflavin were delivered to the plates in the top agar layer: 0.3 ml of cells at 2 × 108/ml were premixed with 15 μl of acriflavine at a concentration of 1 mg/ml and added to a tube containing 3 ml of melted top agar (46°). The top agar was allowed to solidify, and the phage were streaked on the surface with sterile paper strips. After overnight incubation at 37°, one resistant mutant was picked from each stock. A 5-ml stock of each Acr mutant was grown in E. coli BB cells in M9 broth. The phage were concentrated by centrifugation and DNA was prepared for sequencing.
Quantitative measurements of Acr frequencies were determined by plating the phage on Drake plates in the presence or absence of acridine, using the same acridine and cell concentrations described for selection; the ratio of T4 plaques is reported as the Acr frequency.
Sequencing of 52.2 gene: PCR-amplified sequencing was performed. The DNA primer used for sequencing ac mutants is complementary to sequence 165,291–312 in genomic T4 DNA (Kutteret al. 1994). It lies in the adjacent 52.1 ORF and has the sequence 5′-CTACCAATAAAGCAGCAAGGGC-3′. The PCR cycling protocol used was 1 min at 94°, 2 min at 54° and 1 min at 72°.
RESULTS
Identification of the bacteriophage T4 52.2 gene as ac: Studies of the stp gene of T4, which maps near to the ac gene, led to the cloning of the region, and the tentative localization of the ac gene to the region based on the ac phenotype of a large deletion that includes stp and neighboring ORFs 52.2 and 52.1 (Chapmanet al. 1988). Based on this information, we sequenced ORF 52.1 and 52.2 in a few spontaneous and 4-(9-acridinylamino)-methane-sulfone-m-anisidide (m-AMSA)-induced Acr mutants (sequences not shown). The Acr mutant phenotype uniformly correlated to mutations predicted to inactivate ORF 52.2. Immediately beyond ORF52.1, is gene 52, which is a subunit of the type II topoisomerase of T4. Mutations in gene 52 have been shown to confer resistance to inhibition of plaque formation by high levels of the acridine, m-AMSA (Huffet al. 1990). Indeed, our initial motivation for identifying the ac gene location was to be assured that physiological phenotypes associated with acridine-resistant topoisomerase alleles were not due in part to ac gene mutations (and they were not). We subsequently found that the ac gene provides a powerful forward assay for mutations in bacteriophage T4. The small size of the ac gene sequence, the ease of mutant selection and the diverse mutants included in our initial sample suggested that a modest effort could broadly sample the molecular characteristics of spontaneous mutations in ac. We present here a comparison of spontaneous Acr mutations arising in wild-type phage T4 and in phage having the mutant tsL141 allele of DNA polymerase.
The frequencies of Acr mutations in wild-type and tsL141 DNA polymerase mutant backgrounds: The frequencies of Acr mutants in independently grown stocks of T4 wild type and in T4 DNA polymerase mutant tsL141 were measured and found to be indistinguishable (Table 1). Thus, the tsL141 allele does not strongly influence the total frequency of mutants in ac. This result is consistent with the influence of tsL141 on r plaque frequencies (Drakeet al. 1969; Drake 1993). However, several mutant polymerase alleles, including tsL141, can sharply reduce spontaneous mutation rates at particular DNA sites. Because of their improved fidelity for some spontaneous and some induced mutations, they were called antimutator alleles (Drakeet al. 1969).
The spectrum of spontaneous Acr mutations in wild type bacteriophage T4: A spectrum of Acr mutations was selected as described in materials and methods. Importantly, each mutant was selected from an independently grown stock of T4, thus assuring the independence of each mutation. Among 77 independent spontaneous Acr mutants selected from wild-type phage, 59 were small mutations with sequence changes lying wholly within the coding region of ac (Figure 1A). These mutations include base substitution mutations predicted to lead to amino acid substitutions in the small, 51 amino acid polypeptide predicted by the ac gene sequence. The remaining ac mutations are duplication or deletion frameshifts and base substitution mutations leading to nonsense codons; each predicts a truncated and/or altered ac gene product. A few larger deletions and duplications, sometimes extending into sequences flanking the ac gene, were also found and can be seen along with those produced by tsL141 in Figures 2 and 3. The calculated frequencies of mutants separated by category are presented in Table 2. The specificity of all the mutations will be described and compared to those arising spontaneously in phage bearing the DNA polymerase mutation tsL141 in the discussion.
Comparison of spontaneous Acr frequencies in tsL141 and wild-type T4 phages
The spontaneous spectrum of Acr mutants in the presence of the tsL141 DNA polymerase allele reveals major effects on mutagenesis: Figure 1B shows the spectrum of small spontaneous mutations occurring wholly within the ac gene in tsL141 bearing phages. Larger mutations are compared to those produced by the wild-type phage in Figures 2 and 3. Mutant frequencies in tsL141 are quantitatively different from those produced by the wild-type polymerase for some mutant categories (Figure 1, Table 2). Thus, despite the absence of a large impact on total Acr mutant frequencies, the tsL141 polymerase allele does quantitatively influence spontaneous mutagenesis. It is the specificity of ac mutations that reveals these otherwise cryptic mutagenesis effects. Indeed, it appears that compensating effects of mutation reduction at some sites and increases at others largely account for the failure to see a change in the total frequency of Acr mutations.
The distribution of the spontaneous Acr mutations lying wholly within the ac gene sequence. (A) Spontaneous ac mutant sequences isolated from the wild type DNA polymerase background; (B) Sequences from the tsL141 DNA polymerase background. The coding sequence of the ac gene begins at position 1 and the non-transcribed strand is illustrated. The T4 genomic coordinates are indicated at the beginning and near the end of the gene (Kutteret al. 1994; http://www.evergreen.edu/user/T4/home.html). Base substitutions are indicated by showing the substituted base immediately below the wild type sequence. Each entry is an independently isolated and sequenced mutation with one exception. An exceptional warm spot for multiple deletions and duplications was found with identical sequences in the vicinity of bp 70 in thetsL141 spectrum, and the number of isolates are given in parentheses, adjacent to these mutants. Single base deletions are indicated by open boxes; larger deletions by lines labeled with Δ and the number of base pairs deleted. Duplications are indicated by filled boxes under the duplicated bases. In some cases deletion or duplication end points are ambiguous because of repeated sequences and can be drawn differently. Three mutants were found which are insertions of a base which is not identical to either of its neighbors and thus not a duplication. These insertions are shown above the sequence, indicated by an I before the nucleotide inserted and the position of insertion is indicated by ▾.
Large deletion or duplication mutations lying wholly within the ac gene sequence implicate DNA misalignments between ac sequences. (A) Deletion or duplication mutations explained by misalignments between the underlined sequences in regions 1, 2 and 3 in the ac gene. A break in the illustrated sequence is indicated by //. The numbers of mutants in the wild-type and tsL141 spectra, respectively, are indicated in parentheses before each sequence. Deletions are indicated by Δ; duplications by +. The sequence in region 1 differs from that in region 2 by a single G which is shown with a double underline. This G provides a potential template for the insertion of a G (iG) in region 1. This insertion is also shown in Figure 1B. Misalignments between regions 1 and 2 produce mutations of either 31 or 32 bp, depending on whether the G in region 2 is retained in deletions or copied in duplications. A misalignment between regions 1 and 3 appears to account for the 80-bp duplication. (B and C) Schematic illustration of misalignments predicted to produce deletions of 31 bp and 32 bp, respectively. See text.
A few Acr mutations are not in ac: A total of eleven Acr mutants, eight in the wild-type background and three in tsL141, had no mutation in either the ac gene or in gene 51.1. Preliminary mapping studies suggest that these Acr mutations are genetically linked to ac but that they can be readily separated from ac by recombination. The non-ac gene mutants recombine with each other to produce wild type recombinants at frequencies of ≤1% (probably equivalent to ~100–200 bp or less, unless the mutant allele influences recombination frequencies) and thus probably lie in a single gene. We do not know whether the apparently lower frequency of these mutations in tsL141 reflects mutation-specificity effects or some other aspect of Acr selection.
Our standard plating protocol for Acr mutant selection does not distinguish ac from the non-ac mutants either phenotypically (plaque size) or numerically (plating efficiency) when the phage are titred on plates with or without acriflavin. We have recently defined conditions for distinguishing the level of Acr conferred by Acr alleles of ac from the level conferred by the anonymous locus (or loci) by burst size when phage are grown for a single cycle in broth. The resistance to acriflavin conferred during growth in broth by mutations at the anonymous locus is significantly less than that conferred by ac mutations (data not shown). The lower level of resistance conferred by the anonymous mutations may account for the fact that plate selection tests were unsuccessful in identifying double-mutant recombinants from crosses between the two classes of Acr loci. We do not know whether the Acr loci are epistatic. The anonymous Acr mutations compared to ac mutations do not currently interfere with the ac mutation assay, but might be lowered further by modifying the Acr selection protocol.
Large Acr deletion and complex mutations with only one end point in the ac gene sequence. The sizes of the deletions and their coordinates in the T4 genome are indicated; positions in ac are also given to facilitate identifying the portion of the ac gene which is deleted. Mutant sequences A–F implicate mutagenesis mediated by misalignment of imperfect direct repeats. The misalignments leave one copy of the repeat behind and delete the other; the bases known to be deleted are in fact the imperfect portions of these repeats; some repeated bases are ambiguous. The implicated repeats; are shown in boxes. The deleted repeat is indicated by a box in the deleted sequence; the repeat that remains is indicated by a box located in either the 5′ or 3′ flanking sequence. Sequence G is a deletion with repeats at both the ends that remain and thus is not explained by classical misalignment; see text. Mutant sequence H is complex. As illustrated, there is a deletion of 53 bp and an insertion of 6 bp. The sequence could alternatively be described as two noncontiguous mutations in which GG is inserted and the underlined CGTC sequence is retained, making the sequence a Δ 49 bp, +2 bp.
DISCUSSION
A major challenge to an improved understanding of spontaneous mutagenesis is the separate identification of the diverse contributing mutagenesis mechanisms. Clues to these contributing mechanisms lie in the mutant sequences. DNA sequence specificity of mutations may reflect various links to the underlying mechanism. Important general possibilities include sequence-specific or sequence-preferred patterns of DNA damage or repair, as well as sequence-specific or sequence-preferred activities of the enzymes that create or repair the mutations.
Additional clues to mutagenesis mechanisms may be detected by examining the influence of a mutant enzyme. Perturbation of mutagenesis by the mutant enzyme may reflect a direct alteration of the activity or specificity of that enzyme. However, perturbation may alternatively or additionally reflect indirect effects, such as changes in DNA metabolism that compensate for an enzymatic deficiency of a mutant. The multiplicity of possibilities suggests that a single mutant enzyme could be expected to influence fidelity in alternative mechanism-specific (or site-specific) ways.
The tsL141 allele has been studied from many perspectives, both in vitro and in vivo. The details of these characterizations, especially if they can be related to DNA sequence-specific behavior of the enzyme, offer the exciting opportunity to explain site-specific and/or mutagenesis mechanism-specific fidelity effects of this mutant allele at the level of its biochemical properties and the interaction of polymerase with the other proteins that participate in phage DNA replication. A successful test of this possibility showed that the hyperactive 3′→5′ exonuclease of the tsL141 enzyme, previously postulated to primarily influence fidelity through its improved proofreading of misincorporated nucleotides (Lo and Bessman 1976), enhances deletion mutagenesis in vivo at the expense of duplication mutagenesis in a reaction that does not involve proofreading (Kaiser and Ripley 1995).
Spontaneous Acr mutagenesis by different mutation pathways in phage T4 with wild type or L141 DNA polymerase
Advances in the biochemical and structural description of the T4 DNA polymerase and its replication complex (e.g., Spacciapoli and Nossal 1994a,b; Reha-Krantz 1995) now permit us not only to see some of the reasons for the increased 3′→5′ exonuclease activity of the tsL141 polymerase, but also to consider the possibility that some fidelity effects may arise from specific protein-protein interactions in the replication fork. Meanwhile, advances in understanding the in vivo properties of DNA replication (Mosiget al. 1995) provide a vastly improved context within which to consider polymerization-mediated errors in contrast to DNA replication perturbations. Thus, we separately consider the influence of the tsL141 allele on each mutant sequence category to facilitate a consideration of the alternative fidelity explanations, and to point out the diversity of mechanisms likely to contribute to the mutation rate at the ac locus.
Basepair substitution mutations: Table 3 shows the catalogue of ac sites at which an amino acid codon can be changed to nonsense by a base substitution. All substitutions except A·T→G·C transition mutations, which are inherently undetectable in nonsense mutation assays, can produce nonsense codons in ac. Nonsense mutations were found at eight ac sites, although substitutions at 29 sites could have produced them. The substitution pattern (Table 3) shows that this is due to a much lower rate of spontaneous mutation at A·T sites than at G·C sites in T4. This is not merely due to the absence of A·T-site transitions; transversions at G·C sites account for half of the G·C-site mutations. In contrast, only one A·T-site transversion was found, despite the opportunity to detect these transversions at 20 sites.
In contrast to nonsense mutations, the Acr phenotype of missense mutations cannot be predicted a priori. However, the low rate of transversions at A·T sites leading to nonsense mutations predicts that transversions at A·T sites leading to missense mutations will also be rare. Indeed, none were found in our sample (Table 4).
Although A·T-site transitions could not produce nonsense mutations, missense mutations along this pathway are produced by the wild-type DNA polymerase. The A·T-site transition frequency is sharply reduced (Table 2) consistent with an antimutator effect of the tsL141 polymerase on this pathway, as seen in reversion studies. The small number of sites at which A·T-site transitions were detected was influenced by the low frequency of these mutants in the tsL141 background, and further exacerbated by the fact that we sampled fewer mutants in the wild-type polymerase background. A larger sample of spontaneous ac mutant sequences from the wild-type polymerase background would be expected to improve the quantitative interpretation of the extent of the reduction of mutagenesis along this pathway by tsL141 and would furthermore be expected to reveal additional examples of missense mutation sites detectable in the Acr assay.
Sites in ac at which single base pair substitutions create nonsense codons
Spontaneous missense mutations in ac
The patterns of substitutions at G·C-sites (Tables 2, 3 and 4) show that transitions and G·C→T·A transversions are detected at similar frquencies in ac, while G·C→C·G transversions are somewhat less frequent. The relative patterns for G·C-site transversions are similar in the lacI gene of wild-type E. coli (http://eden.ceh.uvic.ca/bigblue/bacteria.htm) and in the ac gene of T4. However, in E. coli the G·C-site transitions outnumber either transversion by three- to sixfold, while in T4 transitions they occur at similar frequencies to transversions at the G·C-sites. It will be interesting to learn whether such differences will prove to be generally organism specific, or whether gene-specific characteristics strongly influence apparent patterns. In neither lacI nor ac do these ratios appear to be due to a single “kind” of hotspot.
The largest changes in G·C-site base substitution frequencies are found at particular sites in a particular polymerase background. A notable example of site-specific, polymerase-specific differences can be seen at neighboring G·C base pair positions 112 and 113 in the ac gene (Figure 1 and Table 4). G·C→C·G transversions at 112 arise at a frequency of 11 × 10−7 in the wild type background, but were undetected at this site in tsL141 (i.e., < 1 × 10−7). In contrast, G·C→A·T transitions at 113 but were found were undetected in the wild type, in tsL141 at a frequency of 6 × 10−7. Although most G·C→T·A transversions were distributed similarly in the two polymerase backgrounds, they were uniquely frequent in the tsL141 spectrum at bp 122.
It is likely that base pair substitutions arising at different sites are sometimes mediated by different mechanisms. In vivo site-specific effects in ac offer a powerful opportunity to identify DNA sites at which the rate of spontaneous base substitutions depends primarily on different mechanisms, and ultimately to dissect spontaneous base substitution rates into their mechanistic components.
Single base frameshift mutations: No influence of the DNA polymerase on this pathway of mutagenesis is revealed when the frameshifts were divided into deletion, duplication or base pair categories (Table 2). Groupings based on suspected mutagenesis mechanisms may be more helpful. Mutations arising by misaligned pairing in monotonic runs are expected to contribute to this mutant class (Streisinger and Owen 1985). The longest monotonic runs in ac are 4 bp long. Previous studies in rII revealed that frameshifts in monotonic runs were more frequent in repeats of 3 and 4 bp than in repeats of 2 bp or non-repeated single bases (Ripleyet al. 1986). Comparison of the tsL141 spectrum in ac with the wild-type spectrum in a frameshift-specific assay in rIIB, suggests that tsL141 produces an excess of non-repeat 1-bp frameshifts. (There are not yet enough frameshifts in the ac spectrum to carry out this analysis for the wild-type polymerase). This result is consistent with two properties already observed for spontaneous frameshifts induced by tsL141: modestly decreased mutations in runs of five consecutive A·T-bp sites, although sequences of the revertants were not determined (Ripley and Shoemaker 1983) and novel hotspots (Ripleyet al. 1983) now known to include Δ1-bp frameshifts that are not in repeats (L. S. Ripley and J. G. de Boer, unpublished results).
The tsL141 spectrum contains three base additions that insert a base different from either nearest neighbor. Insertions are unusual in T4 (Ripleyet al. 1986), but can be consistent with misaligned synthesis mechanisms, involving imperfect direct repeats or palindromes which serve as templates for the insertion (Ripley 1990). A templated explanation, involving a nearby imperfect direct repeat, accounts for the G insertion between bases 54 and 55 in ac (Figures 1B and 2A). However, we find no attractive template for the insertions of A between bases 108 and 109 in the T4 genomic sequence (http://www.evergreen.edu/user/T4/home.html). Nonetheless, a templated explanation is suggested by the occurrence of two examples of such a mutation in a modest-sized spectrum. A polymorphism in the genome of our parent strain has not been ruled out as a template.
Additions or deletions larger than 1 bp: The tsL141 allele increases the frequency of base additions or duplications of modest size within the ac gene (Table 2) by about twofold. At the level of resolution offered by our current sequencing effort, the increases appear to be broad rather than to be focused at one or two specific hotspots.
The warm spot for multibase duplications in the region around bp 70 (Figure 1) is of particular note because of their specificity. These duplications are unlikely to be due to misaligned pairing. Instead the arrangement of the mutant sequences suggests that most of the mutations share one end point. Indeed, adjacent 2-bp deletions may also be part of this warm spot. The phenomenon of clustered frameshift end points is observed at DNA nick sites induced by topoisomerases, not only in T4 but also in mammalian cells (Ripley and Clark 1986; Ripley 1990; Ripley 1994). This ac duplication site appears to be an excellent candidate for a spontaneous frameshift hotspot initiated by a specific DNA nick. The involvement of topoisomerase or other candidate endonucleases is under investigation. Because mutants at this site are also found in the wild-type spectrum (Figure 1A), it is unlikely that nicking is limited to phage bearing the mutant polymerase.
Figure 2A shows a group of mutations created by alternative misalignments between sets of repeated sequences within the ac gene. An influence of tsL141 on deletion specificity is suggested by the mutant distributions. Although the total frequencies of the 31-bp and 32-bp deletions or duplications are not influenced by the polymerase mutant, the larger fraction of 32-bp deletions in the tsL141 spectrum suggests that fidelity is nonetheless altered.
One possibility is that the mutant polymerase may convert misaligned DNA intermediates that would have produced 31-bp deletions into those that produce 32-bp deletions by virtue of its stronger 3′→5′ exonuclease activity. The 31-bp and 32-bp mutations differ by whether the G in the midst of the repeat (81–94) in region 2 of Figure 2 is included or not in the deletion or duplication.
Making the assumption that the entire repeat is initially included in most misalignments, Figure 2B shows the misalignment leading to a 31-bp deletion, while Figure 2C shows the misalignment leading to a 32-bp deletion. However, the misalignment in Figure 2B would produce a 32-bp deletion if the exonuclease removed the bases from positions 81–86 (6 bp). The misalignment still involves 8 bp. In contrast, the misalignment in Figure 2C would require the removal of bases 55–62 (8 bp), and the misaligning repeat retains only 5 bp.
Alternatively, the mutant polymerase could preferentially promote formation of intermediates of the type in Figure 2C, while suppressing the formation of intermediates of the type in Figure 2B. General paradigms for this scenario are available from in vitro experiments showing that sites of frequent polymerase pausing correlate to sites at which mutagenic misalignments occur more frequently (Papanicolaou and Ripley 1991), or from in vivo experiments showing that mutagenic misalignments often occur more frequently on the lagging strand during DNA synthesis (Trinh and Sinden 1991) and that many perturbations, including mutant polymerases of T4, influence the sites at which replication is initiated (Mosiget al. 1995).
The largest deletions with the Acr phenotype have one end point outside of ac. The sequenced examples are not notably changed in frequency or specificity (Table 2, Figure 3) by the tsL141 polymerase. Most deletion endpoints can be explained by models that invoke misaligned pairing between small direct repeats with interruptions, just as seen in studies of E. coli lacI-lacZ gene fusions (Albertiniet al. 1982). Sequence B, a 1,121-bp deletion, is precisely predicted by the indicated repeat; its 1-bp shorter cousin, sequence C, is presumably due to either a second mutation, or an imprecise matching at the ends during the production of the deletion from the same misalignment (Figure 3). This single example does not permit us to draw any conclusion about the probability of imprecision of deletion end points relative to repeats.
Sequence G, a deletion of 731 bp, is not explained by any classical description of slipped pairing, although it does juxtapose the two TAGT sequences shown on the 5′ and 3′ flanking sides of the deletion in Figure 3. The endpoints of this deletion are very close to those that account for sequence E, a 742-bp deletion. The position of the left-hand repeat element implicated by sequence E is underlined in the deleted sequence of the 731-bp deletion (Figure 3). Perhaps imperfect processing of DNA ends during deletion formation produced this mutant sequence.
Sequence H, a complex mutation (a 53-bp deletion and a 6-bp insertion), is not explained by misalignment. There is no single template source for the 6-bp insertion in the T4 genome. The sequence could be alternatively described as a GG insertion and a 49-bp deletion (see Figure 3).
However, the alternative description does not produce a better explanation for either the identity of the insertion or the position of the deletion.
Conclusions:We identified the specific bacteriophage T4 sequence which corresponds to the ac gene. We demonstrate that spontaneous Acr mutant sequences include a balanced mix of mutations of all mutation categories in ac. Because the ac spectrum is not dominated by a single hotspot, a sequenced mutant collection of ~200 can provide an excellent overview of mutational specificity.
We identified a warm spot for spontaneous duplications in ac that has the specificity expected for mutations arising as a result of nick-processing errors, independent duplications which have one terminus in common (the nick site) and extend for different distances. Although a nick-processing mechanism has been well-documented for mutagen-induced frameshifts (Ripley 1990), the mechanism has not yet been shown to contribute to spontaneous mutagenesis. This spontaneous warm spot site in ac represents an excellent candidate for demonstrating that nick-processing errors contribute to frequent spontaneous frameshift mutagenesis and may provide an explanation for some of the substantial fraction of spontaneous duplications whose sequences are inconsistent with misalignment mechanisms.
The ac gene spectrum shows that spontaneous transversion mutations at A·T sites occur at least 20-fold less frequently than do spontaneous transversions at G·C sites in both DNA polymerase backgrounds. The genomic impact, including the ac gene, is further augmented because there are approximately twice as many A·T base pairs in T4 DNA as there are G·C base pairs. The comparison of ac mutant sequences arising in wild-type polymerase and mutant tsL141 polymerase backgrounds confirms the ability of the mutant polymerase to sharply reduce spontaneous A·T-site transition mutations. However, the influence of this mutant polymerase on the other base pair substitution pathways is different. Quantitatively the average frequency for all the mutations sharing a substitution pathway is affected only slightly. However, substantial site-specific effects are seen, suggesting that either these mutant polymerase-mediated effects are very DNA-sequence-context dependent, or that the mechanisms producing substitutions at different sites differ. The site-specific patterns of these changes point to the level of detail at which both substitution mechanisms and DNA polymerase characteristics will need to be known, before the extent of direct perturbation of polymerization fidelity by polymerase mutants can be accurately assessed and distinguished from indirect effects due to altered DNA metabolism in genetically perturbed backgrounds.
Acknowledgments
This work was supported by grant MCB-9305474 from the National Science Foundation and by grant GM-53105 from the National Institutes of Health. Preliminary experiments were supported by grant CN-50 from the American Cancer Society.
- Copyright © 1998 by the Genetics Society of America