Predictability of mutant sequences. Relationships between mutational mechanisms and mutant specificity.

Spontaneous mutations are rare and are produced by multiple biochemical mechanisms. Nonetheless, studies of these mechanisms have revealed striking examples in which mutational specificity can be regularly related to a characteristic of the surrounding DNA sequence and/or the enzymes participating in mutagenesis. Thus, to an increasing degree the DNA sequences of mutants are "predictable." This report considers some examples of predictable sequence changes, evidence for their contribution to mutagenesis in populations, and how the predictability of mutant sequences may be useful to improve our interpretation of the molecular course of evolution from DNA sequence comparisons.

Primer-template misalignments during leading strand DNA synthesis account for the most frequent spontaneous mutations in a quasipalindromic region in Escherichia coli.

Spontaneous mutant sequences which differ from the starting DNA sequence by the specific correction of quasipalindromic to perfect palindromic sequence are hallmarks of mutagenesis mediated by misalignments directed by palindromic complementarity. The mutant sequences are specifically predicted by templated, but ectopic, DNA polymerization on a misaligned DNA substrate. In a previous study, we characterized a spontaneous frameshift hotspot near a 17 bp quasipalindromic DNA sequence within the mutant chloramphenicol acetyl transferase (CAT) gene of plasmid pJT7. A one base-pair insertion hotspot, ectopically templated by misalignment mediated by palindromic complementarity, was shown to occur more frequently during synthesis of the leading than the lagging DNA strand. Here we analyze the misalignment mechanisms that can account for the DNA sequences of 123 additional spontaneous frameshift mutations (22 distinct genotypes) occurring in the same quasipalindromic DNA region in plasmids pJT7 and p7TJ (a pJT7 derivative with the CAT gene in the inverse orientation). Approximately 80% of the small frameshift mutants in each plasmid are predicted by palindromic misalignments of the leading strand. Smaller numbers of mutations are consistent with other DNA misalignments, including those predicted by simple slippage of the nascent DNA strand on its template. The results show that remarkable changes in the mutation spectra of a reporter gene may not be revealed by measurements of mutation frequency.

The spectrum of acridine resistant mutants of bacteriophage T4 reveals cryptic effects of the tsL141 DNA polymerase allele on spontaneous mutagenesis.

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, in-frame 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.

DNA nick processing by exonuclease and polymerase activities of bacteriophage T4 DNA polymerase accounts for acridine-induced mutation specificities in T4.

Acridine-induced frameshift mutagenesis in bacteriophage T4 has been shown to be dependent on T4 topoisomerase. In the absence of a functional T4 topoisomerase, in vivo acridine-induced mutagenesis is reduced to background levels. Further, the in vivo sites of acridine-induced deletions and duplications correlate precisely with in vitro sites of acridine-induced T4 topoisomerase cleavage. These correlations suggest that acridine-induced discontinuities introduced by topoisomerase could be processed into frameshift mutations. The induced mutations at these sites have a specific arrangement about the cleavage site. Deletions occur adjacent to the 3' end and duplications occur adjacent to the 5' end of the cleaved bond. It was proposed that at the nick, deletions could be produced by the 3'-->5' removal of bases by DNA polymerase-associated exonuclease and duplications could be produced by the 5'-->3' templated addition of bases. We have tested in vivo for T4 DNA polymerase involvement in nick processing, using T4 phage having DNA polymerases with altered ratios of exonuclease to polymerase activities. We predicted that the ratios of the deletion to duplication mutations induced by acridines in these polymerase mutant strains would reflect the altered exonuclease/polymerase ratios of the mutant T4 DNA polymerases. The results support this prediction, confirming that the two activities of the T4 DNA polymerase contribute to mutagenesis. The experiments show that the influence of T4 DNA polymerase in acridine-induced mutation specificities is due to its processing of acridine-induced 3'-hydroxyl ends to generate deletions and duplications by a mechanism that does not involve DNA slippage.

Deletion and duplication sequences induced in CHO cells by teniposide (VM-26), a topoisomerase II targeting drug, can be explained by the processing of DNA nicks produced by the drug-topoisomerase interaction.

Frameshift mutations induced by acridines in bacteriophage T4 have been shown to be due to the ability of these mutagens to cause DNA cleavage by the type II topoisomerase of T4 and the subsequent processing of the 3' ends at DNA nicks by DNA polymerase or its associated 3' exonuclease followed by ligation of the processed end to the original 5' end. An analysis of the ability of nick-processing models is presented here to test the ability of nick processing to account for the DNA sequences of duplications and deletions induced in the aprt gene of CHO cells by teniposide (VM-26) [Han et al. (1993) J. Mol. Biol., 229, 52]. Although teniposide is not an acridine, it induces topoisomerase II-mediated DNA cutting in aprt sequences in vitro and mutagenesis in vivo. Although the previous study noted a correlation between mutation sites and nearby DNA discontinuities induced by the enzyme in vitro, neither the nick-processing model responsible for T4 mutations, nor double-strand break models alone were able to account for most of the mutant sequences. Thus, no single model explained the correlation between teniposide-induced DNA cleavage and mutagenic specificity. This report describes an expanded analysis of the ways that nick-processing models might be related to mutagenesis and demonstrates that a modified nick-processing model provides a biochemical rationale for the mutant specificities. The successful nick-processing model proposes that either 3' ends at nicks are elongated by DNA polymerase and/or that 5' ends of nicks are subject to nuclease activity; 3'-nuclease activity is not implicated. The mutagenesis model for nick-processing of teniposide-induced nicks in CHO cells when compared to the mechanism of nick-processing in bacteriophage T4 at acridine-induced nicks provides a framework for considering whether the differences may be due to cell-specific modes of DNA processing and/or due to the precise characteristics of topoisomerase-DNA intermediates created by teniposide or acridine that lead to mutagenesis.

DNA sequence effects on single base deletions arising during DNA polymerization in vitro by Escherichia coli Klenow fragment polymerase.

Most single base deletions detected after DNA polymerization in vitro directed by either Escherichia coli DNA polymerase I or its Klenow fragment are opposite Pu in the template. The most frequent mutations were previously found to be associated with the consensus template context 5'-PyTPu-3'. In this study, the predictive power of the consensus sequence on single base deletion frequencies was directly tested by parallel comparison of mutations arising in four related DNAs differing by a single base. G, a deletion hotspot within the template context 5'-TTGA-3', was substituted by each of the 3 other bases. Previous studies had shown that deletions opposite the G were frequent but that deletions opposite its neighboring A were never detected. Based on the predictions of the consensus, the substitution of T for G should produce frequent deletions opposite the neighboring A due to its new 5'-TTTA-3' template context. This prediction was fulfilled; no deletions of this A were detected in the other templates. The consensus further predicted that deletions opposite template C would be lower than those opposite either A or G at the same site and this prediction was also fulfilled. The C substitution also produced a new hotspot for 1 bp deletions 14 bp away. The new hotspot depends on quasi-palindromic misalignment of the newly synthesized DNA strand during polymerization; accurate, but ectopically templated synthesis is responsible for this mutagenesis. Mutations templated by quasi-palindromic misalignments have previously been recognized when they produced complex sequence changes; here we show that this mechanism can produce frequent single base deletions. The unique stimulation of misalignment mutagenesis by the C substitution in the template is consistent with the singular ability of C at that site to contribute to extended complementary pairing during the DNA misalignment that precedes mutagenesis.

A proflavin-induced frameshift hotspot in the thymidylate synthase gene of bacteriophage T4.

Twenty-one independent thymidylate synthase deficient (td) mutants were isolated after proflavin mutagenesis of T4D0 phage. A strikingly high proportion of these mutations (17 of 21; 80%) mapped in a small 122 nucleotide (nt) region which spans the 5' splice site of this intron-containing gene. This region comprises only 14% of the total td exon sequence. RNA sequence analysis of these mutants identified a series of frameshift insertion/deletion mutations and indicated a hotspot for proflavin-induced mutations in the 3' end of exon I of the td gene. The mutant sequences at the hotspot site fully support a previously proposed mutagenic mechanism for proflavin-induced mutations in which frameshifts are produced as a consequence of exonuclease or DNA polymerase activity at the 3' ends of nicks in the DNA produced by perturbation of the T4-encoded type II topoisomerase activity by the acridine. Sixteen of the seventeen DNA mutations in the hotspot region can be explained by the model as a consequence of enzymatic processing of nicks at two phosphodiester bonds staggered by 4 base pairs (bp) and located on opposite strands of the DNA. Thus, these mutants exhibit precisely the symmetry expected of topoisomerase-mediated mutagenesis. The DNA sequences of the td hotspot mutants, when considered with the sequences of proflavin-induced mutants in the T4 rIIB and lysozyme genes, confirm the view that proflavin-induced mutations in diverse bacteriophage T4 DNA sequences are all produced by the topoisomerase-dependent mechanisms and do not support the view that classical misalignments in DNA repeats are hotspots for proflavin-induced mutagenesis in T4.

A structural model for sequence-specific proflavin-DNA interactions during in vitro frameshift mutagenesis.

Molecular models describing intermediates that may lead to proflavin-induced 1 bp deletions during in vitro polymerization by E. coli DNA polymerase I Klenow fragment are proposed. The models provide structural explanations for the fact that the induced frameshifts always occur opposite template bases that are adjacent to 5' pyrimidines and are based on the underlying hypothesis that the deletions arise because the polymerase passes by a template base without copying it. Because the most frequent mutations are opposite Pu in the template sequence 5' Py Pu 3', a single-strand loop-out model was constructed for this sequence and proflavin was added, using structures found in crystalline oligonucleotides and their complexes with proflavin. The model seeks to rationalize the roles of the 5' pyrimidine and proflavin in facilitating the bypass. Four potential roles for proflavin in mutagenesis are described: 1) stacking on the looped-out base; 2) stacking on the base pair immediately preceding the site of mutation; 3) hydrogen bonding with the 5' pyrimidine; 4) hydrogen bonding with the phosphate backbone. These models point to the possibility that a number of proflavin-DNA interactions may be involved. In contrast, modeling does not suggest a role for classically intercalated proflavin in frameshift mutagenesis arising during in vitro DNA polymerization.

An in vitro approach to identifying specificity determinants of mutagenesis mediated by DNA misalignments.

In vitro, misalignments of the newly synthesized (primer) strand during DNA polymerization lead to deletion and/or complex frameshift mutations. In vivo, similar misalignments of repeated and quasipalindromic DNA sequences are predicted to be intermediates of mutagenesis. The mutagenic misalignments are mediated by complementary pairing between the sequence at the 3'-OH end of the newly synthesized DNA strand and sequences in the template or in the newly synthesized DNA. Mutant sequences are produced when the misaligned primers act as substrates for DNA polymerization. The misalignments responsible for detected mutant sequences were compared to similar misalignments that were not implicated in mutagenesis, and all misalignment possibilities were compared to the position of pausing during polymerization by Escherichia coli polymerase I or its Klenow fragment. These comparisons revealed three characteristics of in vitro misalignment specificity. First, the termini produced by pausing are likely to be precursors to mutagenic misalignments. Second, the absence of some potential misalignments from the detected spectrum is explained well by the predicted undetectability of the mutant sequences they produce. Third, factors distinct from pausing and mutant detectability are responsible for differences in the specificity of misalignment mutagenesis mediated by E. coli DNA polymerase I and Klenow polymerase during in vitro synthesis.

The specificity of topoisomerase-mediated DNA cleavage defines acridine-induced frameshift specificity within a hotspot in bacteriophage T4.

Acridine-induced frameshift mutations in bacteriophage T4 occur at the precise location in the DNA at which acridines stimulate DNA cleavage by the T4-encoded type II topoisomerase in vitro. The mutations are duplications or deletions that begin precisely at the broken phosphodiester bond. In vivo, acridine-induced frameshift mutagenesis is reduced nearly to background levels when the topoisomerase is genetically inactivated. These observations are consistent with a model in which cleaved DNA, induced by the topoisomerase and acridine, serves as the substrate for the production of frameshift mutations at the same site. Our model predicts that the specificity and frequency of cleavage direct the specificity and frequency of mutagenesis. This prediction was tested by examining the influence of DNA sequence changes on topoisomerase-mediated cleavage and on mutagenesis in the T4 rIIB gene. The model successfully predicted the results. When DNA sequence changes altered the position of acridine-induced, topoisomerase-mediated DNA cleavage in vitro, frameshift mutations were found at the new positions. DNA sequence changes that strongly decreased in vitro cleavage also reduced mutagenesis at that site. These results demonstrate that acridine-induced frameshift mutation specificity is directed by the characteristics of the acridine-topoisomerase reaction and do not suggest that slipped pairing in repeated sequences plays a major role in acridine-induced frameshifts in bacteriophage T4.

Effects of proflavin and photoactivated proflavin on the template function of single-stranded DNA.

DNA context-specific effects of the association of proflavin, single-stranded DNA and DNA polymerase on DNA polymerization reactions were examined. Frameshift mutations induced by the presence of proflavin during in vitro DNA replication of a single-stranded DNA template by the Klenow fragment of Escherichia coli DNA polymerase I were sequenced. More than 80% of the frameshifts were one base-pair deletions opposite purine bases that were immediately 3' to pyrimidines. Purines (Pu) that were not adjacent to pyrimidines (Py) were not deletion sites. The remaining deletions were opposite template pyrimidines that were also immediately 3' to another pyrimidine. All pyrimidine site deletions occurred in the context 5' PyPyPu 3'. In additional experiments, the site-specific inhibition of processive DNA polymerization by proflavin was examined. A novel inhibition of polymerization was found opposite all pyrimidines in the template when proflavin-template complexes were exposed to ten seconds of white light. This inhibition of polymerization is reversible. Longer photoactivation led to an altered pattern of DNA sequence-specific inhibition that was not reversible. The role of DNA sequence-specific interactions of proflavin with DNA in proflavin mutagenesis is discussed.

Polymerase-specific differences in the DNA intermediates of frameshift mutagenesis. In vitro synthesis errors of Escherichia coli DNA polymerase I and its large fragment derivative.

The sequences of more than 600 frameshift mutations produced as a consequence of in vitro DNA replication on an oligonucleotide-primed, single-stranded DNA template by the Escherichia coli polymerase I enzyme (PolI) or its large fragment derivative (PolLF) were compared. Four categories of mutants were found: (1) single-base deletions, (2) base substitutions, (3) multiple-base deletions and (4) complex frameshift mutations that change both the base sequence and the number of bases in a concerted mutational process. The template sequence 5'-Py-T-G-3', previously identified as a PolLF hotspot for single-base deletions opposite G, is also a hotspot for PolI. A PolI-specific warm spot for single-base deletions was identified. Among base substitutions, transitions were more frequent than transversions. Transversions were mediated by (template)G.G, (template)G.A, and (template)C.T mispairs. Multiple-base deletions were found only after PolI replication. Although each of these deletions can be explained by a misalignment mediated by directly repeated DNA sequences, deletion frequencies were often different for repeats of the same length. Both PolI and PolLF produced many complex frameshift mutants. The new sequences at the mutant sites are exactly complementary to nearby DNA sequences in the newly synthesized DNA strand. In each case, palindromic complementarity could mediate the misalignment needed to initiate the mutational process. The misaligned DNA synthesis accounts for the nucleotide changes at the mutant site and for homology that could direct realignment of the DNA onto the template. Most of the complex mutant sequences could be initiated by either intramolecular misalignments involving fold-back structures in newly synthesized DNA or by strand-switching during strand-displacement synthesis. The striking differences between the specificities of complex frameshift mutations and multiple-base deletions by PolI and PolLF identify the existence of polymerase-specific determinants that influence the frequency and specificity of misalignment-mediated frameshifts and deletions.

Estimation of in-vivo miscoding rates.

The replication of premutagenic DNA lesions generates mutant progeny in patterns that distinguish lesions that rarely produce a mutation per DNA replication from those that frequently do so. The quantitative aspects of this distinction were tested in studies of heat-mutagenized bacteriophage T4. Previous T4 studies had demonstrated that transition mutations produced at G.C base-pairs depended upon heat-induced DNA lesions distinct from those responsible for transversions at G.C pairs. In this study the transversion mutations are shown to arise in patterns predicted for mutations produced from lesions that miscode rarely (fewer than 10% per replication). In contrast, the transition mutations arise in patterns predicted for mutations produced from lesions that miscode at about 20 to 60% per replication. The fact that the two classes of DNA lesions are distinguishable as predicted by the quantitative model suggests that such studies may in general be useful in quantifying the behavior of mutation-generating DNA lesions. The method employed also estimates the frequency of premutagenic lesions in DNA.

Hotspot sites for acridine-induced frameshift mutations in bacteriophage T4 correspond to sites of action of the T4 type II topoisomerase.

The type II topoisomerase of bacteriophage T4 is a central determinant of the frequency and specificity of acridine-induced frameshift mutations. Acridine-induced frameshift mutagenesis is specifically reduced in a mutant defective in topoisomerase activity. The ability of an acridine to promote topoisomerase-dependent cleavage at specific DNA sites in vitro is correlated to its ability to produce frameshift mutations at those sites in vivo. The specific phosphodiester bonds cleaved in vitro are precisely those at which frameshifts are most strongly promoted by acridines in vivo. The cospecificity of in vitro cleavage and in vivo mutation implicate acridine-induced, topoisomerase-mediated DNA cleavages as intermediates of acridine-induced mutagenesis in T4.

An in vitro assay for frameshift mutations: hotspots for deletions of 1 bp by Klenow-fragment polymerase share a consensus DNA sequence.

The fidelity of in vitro DNA synthesis catalyzed by the large fragment of DNA polymerase I was examined. The templates, specifically designed to detect shifts to the +1 or to the -1 reading frame, are composites of M13mp8 and bacteriophage T4 rIIB DNA and were designed to assist in the identification of the types of frameshifts that are the specific consequence of DNA polymerization errors. In vitro polymerization by the Klenow fragment produced only deletions, rather than the mixture of duplications and deletions characteristic of in vivo frameshifts. The most frequent frameshifts were deletions of 1 bp opposite a template purine base. Hotspots for these deletions occurred when the template purine immediately preceded the template sequence TT. The highest mutation frequencies were seen when the TTPu consensus sequence was adjacent to G:C rich sequences in the 3' direction. The nature of the consensus sequence itself distinguishes this 1-bp deletion mechanism from those operating in DNA repeats and attributed to the misalignment of DNA primers during synthesis. Deletions that were larger than 1 or 2 bp isolated after in vitro replication were consistent with the misalignment of the primer. Deletions of 2 bp and complex frameshifts (the replacement of AA by C) were also found. Mechanisms that may account for these mutations are discussed.

Spectrum of spontaneous frameshift mutations. Sequences of bacteriophage T4 rII gene frameshifts.

The DNA sequences of 185 independent spontaneous frameshift mutations in the rIIB gene of bacteriophage T4 are described. Approximately half of the frameshifts, including those at hot spot sites, are fully consistent with classical proposals that frameshift mutations are produced by a mechanism involving the misaligned pairing of repeated DNA sequences. However, the remaining frameshifts are inconsistent with this model. Correlations between the positions of two base-pair frameshifts and the bases of DNA hairpins suggest that local DNA topology might influence frameshift mutation. Warm spots for larger deletions share the property of having endpoints adjacent to DNA sequences whose complementarity to sequences a few base-pairs away suggest that non-classical DNA misalignments may participate in deletion mutation. A model for duplication mutation as a consequence of strand displacement synthesis is discussed. In all, 15 frameshifts were complex combinations of frameshifts and base substitutions. Three of these were identical, and have extended homology to a sequence 256 base-pairs away that is likely to participate in the mutational event; the remainder are unique combinations of frameshifts and transversions. The frequency and diversity of complex mutants suggest a challenge to the assumption that the molecular evolution of DNA must depend primarily upon the accumulation of single nucleotide changes.

Frameshift mutations produced by proflavin in bacteriophage T4: specificity within a hotspot.

Frameshift mutations were induced by proflavin in the rIIB gene of bacteriophage T4. rIIB DNA from each of 48 independent frameshifts was inserted into M13mp8 and sequenced. Two-thirds of the frameshifts (33/48) lie contiguous to one another in 10 base pairs of the rIIB sequence. This hotspot differs markedly from previously characterized mutagen-induced frameshift hotspots. Distinctive features of the hotspot include the absence of locally repetitive sequences, particularly G X C runs, and the fact that many different sequence changes are induced within the hotspot sequence at appreciable frequencies. Among the 33 mutants at the hotspot, 8 distinguishable DNA sequence changes were seen. All of the mutations were deletions of a single base or duplications of one or more bases. Duplications were more frequent than deletions. The patterns of the base sequence changes suggest that two specific phosphodiester bonds within the hotspot sequence are sites at which proflavin-induced mutation is initiated.

Bacteriophage T4 particles are refractory to bisulfite mutagenesis.

Bisulfite-induced deamination of cytosine produces uracil, a thymine analog reported to be mutagenic both in vitro and in vivo. Although deamination of cytosine in DNA should produce G:C----A:T transitions, treating bacteriophage T4 particles with 0.9 M bisulfite at pH 5 at 37 degrees C produced no more mutations than did the equivalent buffer without bisulfite. Lack of bisulfite mutagenicity is fully consistent with the reported resistance of 5-substituted cytosines to bisulfite-induced deamination, since T4 DNA contains glucosylated 5-hydroxymethylcytosine. However, bisulfite also failed to induce mutations in T4 particles whose DNA contained unmodified cytosine. The lack of mutagenesis persisted in E. coli hosts deficient in uracil glycosylase, an enzyme expected to participate in the repair of the putative bisulfite-generated uracil. Cytosine in T4 DNA may be largely protected from bisulfite attack within phage particles.