Saturation of DNA Mismatch Repair and Error Catastrophe by a Base Analogue in Escherichia coli

Corresponding author: Roel M. Schaaper, E3-01, National Institute of Environmental Health Sciences, P.O. Box 12233, 111 TW Alexander Dr., Research Triangle Park, NC 27709., schaaper{at}niehs.nih.gov (E-mail)

Communicating editor: P. L. FOSTER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Deoxyribosyl-dihydropyrimido[4,5-c][1,2]oxazin-7-one (dP) is a potent mutagenic deoxycytidine-derived base analogue capable of pairing with both A and G, thereby causing G · C A · T and A · T G · C transition mutations. We have found that the Escherichia coli DNA mismatch-repair system can protect cells against this mutagenic action. At a low dose, dP is much more mutagenic in mismatch-repair-defective mutH, mutL, and mutS strains than in a wild-type strain. At higher doses, the difference between the wild-type and the mutator strains becomes small, indicative of saturation of mismatch repair. Introduction of a plasmid containing the E. coli mutL⁺ gene significantly reduces dP-induced mutagenesis. Together, the results indicate that the mismatch-repair system can remove dP-induced replication errors, but that its capacity to remove dP-containing mismatches can readily be saturated. When cells are cultured at high dP concentration, mutant frequencies reach exceptionally high levels and viable cell counts are reduced. The observations are consistent with a hypothesis in which dP-induced cell killing and growth impairment result from excess mutations (error catastrophe), as previously observed spontaneously in proofreading-deficient mutD (dnaQ) strains.

NUCLEOSIDE analogue mutagens are a unique class of mutagens that are metabolized like normal nucleosides or bases but disturb accurate transmission of genetic information because of their increased mispairing potential. Deoxyribosyl-dihydropyrimido[4,5-c][1,2]oxazin-7-one (dP; Fig 1) is a potent mutagenic deoxycytidine analogue capable of pairing with both A and G (LIN and BROWN 1989 ; NEGISHI et al. 1997 ; HILL et al. 1998 ). This deoxycytidine analogue had been synthesized as a model compound for the anti conformer of N⁴-methoxydeoxycytidine, a major product of the reaction of methoxyamine with DNA, whose ambiguous base-pairing properties were reported previously (STONE et al. 1991 and references therein; NEDDERMAN et al. 1993 ). dP can be incorporated into Escherichia coli DNA, thereby causing replicational errors resulting in G · C A · T and A · T G · C transition mutations (NEGISHI et al. 1997 ). In this study, we have analyzed some of the E. coli repair activities that might control or interfere with the mutagenic action of the analogue. No significant effect was observed of the activities associated with the recA, umuDC, or uvrA genes. However, the mutH, -L, -S DNA mismatch-repair system was found to strongly protect cells against dP mutagenesis, particularly at low dP dose levels. At high doses, dP causes growth delay and cell killing due to an excess of mutations and saturation of mismatch repair, similar to the "error catastrophe" brought about by increased spontaneous replication errors in proofreading-deficient mutD5 cells (SCHAAPER and RADMAN 1989 ; FIJALKOWSKA and SCHAAPER 1996 ).

View larger version (16K): [in this window] [in a new window] [Download PPT slide]   Figure 1. The structure of dP (A) and proposed (mis)pairing schemes (B) consistent with the ability of dP to induce transition base-pair errors.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains and media:
The strains used in this study are listed in Table 1. All strains are derivatives of KA796 (ara, thi, prolac; SCHAAPER et al. 1985 ). F'CC102 and F'CC106 are the F'prolac episome from strains CC102 or CC106 (CUPPLES and MILLER 1989 ), respectively. These F' contain a defined lacZ missense mutation that can revert to lac⁺ by a single base substitution only. The lac (E461G) allele of F'CC102 reverts only by G · C A · T transition and that of F'CC106 (E461K) only by A · T G · C transition (CUPPLES and MILLER 1989 ). The mutH, mutL, and mutS deficiencies were introduced by P1 transduction using strains GM6470 (mutH471::Tn5; obtained from M. Marinus, University of Massachusetts), ES1293 (mutL211::Tn5; SIEGEL et al. 1982 ), and ES1301 (mutS201:: Tn5; SIEGEL et al. 1982 ) as donors, respectively. Selection was for kanamycin resistance followed by testing for mutator phenotype. The recA56 allele was introduced by P1 transduction using linkage with srl360::Tn10. The (umuDC595::cat) marker was transferred by P1 transduction from strain RW82 (WOODGATE 1992 ) selecting for chloramphenicol resistance. The uvrA277::Tn10 allele was transferred from strain N3055 (E. coli Genetic Stock Center, Yale University) selecting for tetracycline resistance and testing for UV sensitivity. F'CC102 and F'CC106 were transferred by conjugation from CC102 and CC106 (CUPPLES and MILLER 1989 ). Plasmids pMQ350 (mutL⁺), pMQ315 (mutS⁺), and pRH71-17 (mutH⁺; all from M. Marinus, University of Massachusetts) are pBR322 derivatives carrying, respectively, the E. coli mutL, mutS, and mutH wild-type genes. They were introduced into NR10832 and NR10836 by calcium cell transformation and selection for ampicillin resistance.

MM (minimal medium) and LB (Luria-Bertani) media were standard recipes (MILLER 1972 ). MM contained 1x VB salts (VOGEL and BONNER 1956 ) supplemented with 0.4% glucose or lactose as carbon source, 5 µg of thiamine/ml, and 50 µg of amino acids/ml, when required. Antibiotics were added as follows: rifampicin at 100 µg/ml, tetracycline at 15 µg/ml, ampicillin at 100 µg/ml, kanamycin at 50 µg/ml, and chloramphenicol at 20 µg/ml. dP was prepared as described and used in liquid media as described (NEGISHI et al. 1997 ). N⁴-Aminocytidine was prepared and used as described (NEGISHI et al. 1987, NEGISHI et al. 1988 ).

Mutant frequency determination:
Overnight cultures were diluted 10⁵-fold with fresh LB broth to give suspensions of 10⁴ cells/ml. From there, 0.1-ml aliquots were taken and added to 1 ml of LB broth containing various concentrations of dP. The cultures were grown overnight at 37° on a rotator wheel. Ampicillin (100 µg/ml) was included for strains harboring the plasmids indicated in Table 1. The saturated cultures (0.1 ml) were spread on LB Rif plates to determine the number of rifampicin-resistant colonies and on MM lactose plates to determine the number of lac revertants. A 10^-6 dilution (0.1 ml) was spread on MM plates to determine the total cell count. Five to 12 independent tubes were used for each dose level. Mutant frequencies were calculated by dividing the median number of mutants by the average number of total cells. Ninety-five percent confidence limits for the mutant frequencies were calculated on the basis of the binomial distribution as described by ZAR 1984 .

Base-analogue-induced growth inhibition:
An overnight culture of bacteria (0.1 ml) was mixed with molten 0.7% agar containing 0.9% NaCl and poured onto a minimal glucose plate. After solidification, a paper disc soaked with an aqueous solution of one of the mutagens was placed onto the top agar, and the plate was then incubated overnight at 37°. The next day, the plates were inspected for a zone of growth inhibition around the disc.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Effects of mismatch-repair deficiency on dP mutagenesis:
The deoxycytidine analogue dP (see Fig 1) is a recently synthesized (LIN and BROWN 1989 ) mutagenic nucleoside that specifically induces transition mutations (NEGISHI et al. 1997 ). Among the six strains developed by CUPPLES and MILLER 1989 to detect each of the six base change mutations, only strains CC102 (reverting to lac⁺ by G · C A · T transition) and CC106 (reverting by A · T G · C transition) showed a mutagenic response to dP (NEGISHI et al. 1997 ), consistent with the proposed ambiguous base-pairing properties of the modified base (see Fig 1). Here, we have carefully analyzed the dose dependence for dP mutagenesis in strains carrying these two lac alleles. The results are shown in Table 2 and Fig 2. For both alleles we found the dose response to be nonlinear, with relatively little mutagenesis occurring at the lower doses but with disproportionately increased mutagenesis occurring at the higher levels. A shape like this could reflect the presence of a system that prevents dP-induced errors at low dose and becomes inhibited or saturated at higher dP concentrations. Alternatively, it could reflect the induction by dP of an error-prone system, such as the SOS system, which would disproportionately increase mutagenesis at higher dose levels.

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   Figure 2. dP-induced mutant frequencies in wild-type and mismatch-repair-defective mutL strains and in wild-type strains carrying the MutL-overproducing plasmid pMQ350. (A) F'CC102 () and F'CC102 mutL (•); (B) F'CC106 () and F'CC106 mutL (); (C) F'CC102 (), F'CC102 (pBR322) (), and F'CC102 (pMQ350) (); (D) F'CC106 (), F'CC106 (pBR322) (), and F'CC106 (pMQ350) (). The insets show the mutant frequency response at low dP concentrations. Strains used were NR10832 (F'CC102), NR10836 (F'CC106), NR11102 (F'CC102 mutL), and NR11106 (F'CC106 mutL) as described in Table 1. The curves were calculated by power regression fitting. Error bars showing 95% confidence limits were calculated on the basis of a binomial distribution of the data as described by ZAR 1984 .

To check the first possibility, we investigated the dP mutability of mismatch-repair-defective mutH, -L, and -S derivatives. We found that in the mismatch-repair-defective strains, the mutant frequency for both lac alleles was significantly elevated above that of the repair-proficient controls (Fig 2A and Fig B). In addition, in the repair-deficient strains the mutant frequency increased as a linear function of the dP concentration. These observations are consistent with a hypothesis that mismatch repair protects cells from the mutagenic action of dP, especially at the lower dP concentrations. At the higher doses, the mismatch-repair system may become saturated, leading to a similar mutability of the mutL and mutL⁺ strains (see below). When rifampicin resistance was used to score mutagenesis, dP was also significantly more mutagenic for the mutL strain at the lower dose levels, but less so at the higher dose level (Table 2, experiment 1).

We also noted that in the mismatch-repair-deficient background the mutability of the CC102 and CC106 lacZ alleles was similar, whereas in the wild-type background the CC102 allele is much less mutable than the CC106 allele (Table 2, experiment 1; Fig 2A and Fig B, see insets for low doses). This suggests that mismatch repair may be more effective in preventing the dPinduced G · C A · T transition of F'CC102 than the A · T G · C transition of F'CC106. Overall, the results show that mismatch repair can effectively prevent mutagenesis induced by dP and that it may be more effective in doing so for G · C A · T transitions than for A · T G · C transitions.

Saturation of mismatch repair:
At higher dP concentrations, the differences between the mismatch-repair-proficient and -deficient strains become smaller and the slopes of the curves become similar. This suggests that the mismatch-repair system is capable of preventing dP-induced mutations effectively at the lower dose, but that its overall capacity may be limited and become overwhelmed by larger amounts of dP incorporation. For example, using the data of Table 1, experiment 1, the efficiency of mismatch repair for correcting the G · C A · T transitions of the CC102 allele can be calculated to be 99.3, 98.8, 94.3, and 63.0% at 0, 1, 3, and 10 µg/ml dP, respectively, indicating that, at these increasing doses, 0.7, 1.2, 6.7, and 37% of all mismatches escape repair. For the CC106 allele, the respective efficiencies are (>97%), 75, 61, and 9%, indicating that (<3%), 25, 39, and 91% escape repair, respectively.

To further corroborate the limited capacity of mismatch repair in dP mutagenesis, we transformed the mismatch-repair-proficient strains with pMQ350, a multicopy plasmid carrying the E. coli mutL gene (ARONSHTAM and MARINUS 1996 ). Previous studies on mismatch-repair saturation have shown the MutL protein to be one of the rate-limiting factors (SCHAAPER and RADMAN 1989 ; FIJALKOWSKA and SCHAAPER 1996 ). As shown in Fig 2C and Fig D, the dP-induced mutant frequencies were strongly suppressed by this plasmid, supporting the idea that mismatch repair is being saturated. As in previous studies (SCHAAPER and RADMAN 1989 ), overproduction of MutS protein did not affect the frequencies, while MutH overproduction had an intermediate effect (data not shown). Consistent with the greater susceptibility to mismatch correction of the dP-induced G · C A · T transitions of the E461G allele of F'CC102, MutL overproduction was very effective in reducing the mutant frequency for this allele, making the strain virtually immutable at low dP concentrations (Fig 2C).

The lacZ allele of strain CC107 (CUPPLES et al. 1990 ) detects +1 frameshift mutations (G₆ G₇). This reversion is induced efficiently by the base analogue 2-aminopurine, and this effect was postulated to occur indirectly through saturation of the mismatch-repair system (CUPPLES et al. 1990 ). We found that dP is also highly mutagenic for this lac allele. The presence of 5 or 10 µg/ml dP in the growth medium increased the reversion frequency from 0.6 x 10^-6 to 24 x 10^-6 and 74 x 10^-6, respectively (data not shown), consistent with the proposed ability of dP to saturate mismatch repair. Complete loss of mismatch repair (in a mutH strain) has been reported to yield a mutant frequency for this lacZ allele of 120 x 10^-6 (CUPPLES et al. 1990 ).

2-Aminopurine is known to be highly toxic to dam mutants (GLICKMAN et al. 1978 ; GLICKMAN and RADMAN 1980 ), and this hypersensitivity is taken to indicate that 2-aminopurine-containing base pairs are a target for mismatch repair. Since dam strains lack the 6-methyladenine strand-discrimination signal, mismatch-repairmediated incisions in both strands are thought to create lethal double-strand DNA breaks in this strain. Consistent with this hypothesis, dam mutL or dam mutS double mutant strains are resistant to 2-aminopurine (GLICKMAN and RADMAN 1980 ). We used this criterion of dam strains hypersensitivity to corroborate the susceptibility of incorporated dP to mismatch repair. A filter disc containing 50 µg of dP placed in the middle of a plate seeded with dam strain NR3742 (Table 1) formed a zone of inhibition in the bacterial lawn of 24 mm in diameter, while no inhibition zone was observed for a wild-type strain (Fig 3A). In contrast, 50 µg of N⁴-aminocytidine, another potent nucleoside analogue (NEGISHI et al. 1988 ), whose mutagenicity is little affected by mismatch repair (NEGISHI et al. 1988 ; SCHAAPER and DUNN 1998 ), formed a 20-mm inhibitory zone in the dam mutant and an 18-mm zone in the wild-type strain (Fig 3). Furthermore, the dP sensitivity of the dam strain is greatly reduced in the corresponding dam mutL strain (Fig 3B), as expected.

View larger version (126K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Sensitivity of E. coli dam to dP. Experiment A demonstrates that the dam strain is dP hypersensitive compared to the wild-type strain, consistent with the postulated susceptibility of dP-containing intermediates to mismatch repair. N4-aminocytosine, to which dam and wild-type strains appear equally sensitive, is shown as an example of a base analogue that is not significantly subject to mismatch repair. Experiment B shows that the dP sensitivity of dam is due to the action of mismatch repair, because the corresponding dam mutL strain is largely resistant to dP. Strains used were NR3742 (dam) and NR13393 (dam mutL; Table 1).

Effect of repair deficiencies other than DNA mismatch repair:
To check if dP mutagenesis might involve participation of the inducible SOS response, we studied dP mutagenesis in recA and umuDC strains. As shown in Table 2 (experiments 4 and 5), the mutational responses in recA or umuDC strains were, likewise, not linear with dP concentration, and dP remained highly mutagenic at high concentration (10 µg/ml) in these strains. Therefore, SOS induction is not essential for dP mutagenesis and not responsible for the nonlinearity of the response in the wild-type strains. Next, we investigated whether dP can be removed from DNA by uvrABC nucleotide excision repair by comparing our strains to their uvrA counterparts. As shown in Table 2 (experiments 6 and 7), the dP-induced mutant frequencies in the uvrA strains were not greater than those in the wild-type strains and their dose responses were again not linear. Thus, nucleotide excision repair does not counteract dP-induced mutagenesis and the nonlinear dose response is not related to the excision repair of the mutagenic nucleotides.

Toxicity of dP:
When cells were cultured at high concentrations of dP (10–20 µg/ml), final cell titers were reduced significantly (about an order of magnitude, see Fig 4), and mutant frequencies reached extremely high values, approaching or exceeding 10^-4 (Table 2). Such high frequencies are comparable to those observed spontaneously in the proofreading-deficient mutD5 strain, which is the strongest known spontaneous mutator (SCHAAPER 1988 ; SCHAAPER and RADMAN 1989 ). Mutation rates in mutD5 and related strains are excessive because of (i) lack of polymerase proofreading and (ii) saturation of mismatch repair caused by the increased level of incorporation errors (SCHAAPER and RADMAN 1989 ). MutD5 strains also show reduced final cell counts, which have been ascribed to the accumulation of lethal mutations (error catastrophe; FIJALKOWSKA and SCHAAPER 1996 ). Therefore, we suggest that cells exposed to high concentrations of dP, likewise, die or become nonculturable because of the accumulation of deleterious mutations in essential genes.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Increase in cell numbers in cultures of wild-type (F'CC106) strains without () or with () dP and F'CC106mutS strains without () or with (•) dP. The inset shows the increase in optical density of the cultures during the later part of the growth curve. Strains used were NR10836 (F'CC106) and NR12898 (F'CC106mutS) as described in Table 1.

The growth of our strains in the presence of dP was studied in more detail as shown in Fig 4. Cells reached stationary phase after 8 hr, at which time the number of colony-forming units in the dP-exposed cultures was one order of magnitude lower than those in the controls. On the other hand, there was relatively little difference between dP-treated and control cultures when cell density was monitored by absorbance at 560 nm. This discrepancy suggests that in dP-treated cultures a majority of the final cells might be nonviable. Alternatively, dP-exposed cells might form long filaments. Microscopic inspection showed that dP-exposed cells were generally somewhat elongated, but on average by only about twofold (not shown). We conclude that a large fraction (>80%) of the dP-exposed cells may not be able to form colonies. Loss of colony-forming ability was more severe in the mismatch-repair-defective mutS strain than in the wild type (Fig 4), suggesting that loss of mismatch repair contributes to the viability problem. These results corroborate the protective effect of mismatch repair toward dP toxicity and that dP toxicity results from the accumulation of lethal mutations.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The interest in DNA mismatch repair in both prokaryotes and eukaryotes has expanded in recent years after discovery of a link between human cancer and mismatch repair. Mismatch repair has now been established to promote genetic stability in a broad sense, by functioning not only in repairing replication errors but also by preventing recombination between nonidentical sequences and by interacting with damaged DNA (for a review, see HARFE and JINKS-ROBERTSON 2000 ). Most studies on the interaction of mismatch repair with damaged DNA have focused on the apoptotic response that can be induced in mammalian cells by binding of mismatch-repair proteins to damaged DNA. One part of this response may be the "futile cycling" mediated by mismatch repair through repetitive excision of and resynthesis across an alkylated or otherwise damaged template base. This mismatch-repair-mediated response to DNA damage may find a useful analogue in the hypersensitivity of E. coli dam bacteria, which are actively killed by exposure to certain DNA-damaging agents, including alkylating agents (JONES and WAGNER 1981 ; KARRAN and MARINUS 1982 ). This hypersensitivity is readily suppressed by a DNA mismatch-repair deficiency, analogous to the alkylation tolerance (lack of apoptosis) of mismatch-repair-deficient tumor cells.

A useful distinction may be made between DNA damage that is present in the template strand, as introduced by a DNA-damaging agent, and damage resulting from incorporation of a damaged or modified base, such as a base analogue, in the nascent strand. In the case of DNA template damage, mismatch repair can have an antimutagenic effect by preferentially removing incorrect insertions opposite the lesion, as demonstrated, for example, for UV mutagenesis in E. coli (LIU et al. 2000 ). In general, this effect may be limited by the inability of mismatch repair to remove the offending lesion from the template strand. On the other hand, mismatch repair has the potential to play an even more significant antimutagenic (or anticarcinogenic) role by directly removing altered bases incorporated in the newly synthesized strand. Interest in modified components of the deoxynucleoside triphosphate pool has increased with the discovery of powerfully mutagenic components, such as 8-oxo-dGTP (MAKI and SEKIGUCHI 1992 ), 2-hydroxy-dATP (KAMIYA and KASAI 2000 ), and 5-hydroxy-dCTP (FEIG et al. 1994 ), which can arise readily in the dNTP pool by oxidative stress. Base analogues such as dP, but also 2-aminopurine, 5-bromouracil, N⁴-hydroxy-cytosine, N⁴-aminocytosine, N⁶-hydroxylaminopurine and others (for a review, see KHROMOV-BORISOV 1997 ) may provide useful model systems for the interaction of the mismatch repair system with such mutagenic components.

One previous case where mismatch repair was shown capable of preventing mutagenesis by a base analogue is 5-bromouracil, for which mutL and mutS strains were shown to be hypermutable (RYDBERG 1977 , RYDBERG 1978 ). Frameshift mutagenesis by reversibly binding intercalators, such as 9-aminoacridine (SKOPEK and HUTCHINSON 1984 ) or oxazolopyridocarbazoles (RENE et al. 1988 ) was also shown to be under control of the mutHLS system. In each of these cases, a curved dose response was observed for the wild-type strain but a straight response for the mismatch-repair-deficient strains, arguing for the saturability of the system's capability to handle mismatches. (Note that in the case of the reversible intercalators, the target for mismatch repair may be a frameshift intermediate where the intercalating molecule that caused it is no longer present.) The current case of dP confirms and extends these earlier observations. In particular, we demonstrate that (i) overproduction of mismatch-repair proteins, notably the limiting component MutL, is capable of counteracting the saturability and that (ii) saturation of mismatch repair by a highly mutagenic base analogue can readily reach the level of error catastrophe, where cells can divide but the daughter cells may no longer be viable due to excessive mutation levels.

Our experiments show that dP was capable of inducing the G · C A · T and A · T G · C errors at roughly similar efficiency, as measured in the absence of mismatch repair. In the presence of mismatch repair both are significantly reduced. However, the G · C A · T transitions appear significantly more susceptible to correction than the A · T G · C. It remains to be demonstrated whether this differential effect of mismatch repair will be generally applicable beyond the two lacZ alleles investigated here. Despite this caveat, the scheme in Fig 5 can be used productively to show how (i) P incorporation can readily yield both transitions and (ii) mismatch repair can differentially affect the two pathways. The G · C A · T pathway proceeds through two steps involving G · P and P · A base pairs in this order (here, and below, we state the template base first), whereas the A · T G · C pathway involves A · P and P · G base-pair intermediates. Correction of the first pair in each of the pathways (G · P for G · C A · T and A · P for A · T G · C) would have the largest effect, as it would result in removal of the offending base. Thus, more efficient correction of the G · P pair than of the A · P pair could account for the preferential prevention of G · C A · T transitions. Indeed, bandshift assay experiments using purified MutS protein and oligonucleotides containing the various matches and mismatches (D. MAEHARA, L. WORTH, D. LOAKES, T. NEGISHI, R. SCHAAPER and K. NEGISHI, unpublished data) show strongest binding to the G · P mismatch, at a level comparable to MutS binding to the generally well-corrected G · T mismatch. This preferential binding of MutS to G · P compared to A · P is consistent with the observed preferred correction of G · C A · T substitutions. The (G · P) mismatch would also be formed in the A · T G · C pathway, but only as P · G in the second step of the A · T G · C pathway and mismatch repair would not be able to remove the modified base.

View larger version (11K): [in this window] [in a new window] [Download PPT slide]   Figure 5. A model for dP-induced G · C A · T and A · T G · C mutations. Shown are the GGG to GAG conversion of the E461G lacZ allele on F'CC102 by G · C A · T transition and the AAG to GAG conversion of the E461K lacZ allele on F'CC106 by A · T G · C transition.

An interesting question is to what extent different base analogues are subject to interaction with (or correction by) the mismatch repair system. Such interaction can be observed experimentally, for example, by a diminishment of mutation rates by mismatch repair or by hypertoxicity of the compound to a dam strain. On the basis of such criteria base pairs containing dP (this work), 2-aminopurine (GLICKMAN and RADMAN 1980 ; CUPPLES et al. 1990 ), 5-bromouracil (RYDBERG 1977 , RYDBERG 1978 ), and N⁴-hydroxycytosine (SLEDZIEWSKA-GOJSKA and JANION 1982 ) are recognized to varying extents, whereas N⁴-aminocytosine (NEGISHI et al. 1988 ; SCHAAPER and DUNN 1998 ) and 8-oxoguanine (SCHAAPER et al. 1989 ) are poorly recognized or not recognized at all. The structural basis for such differential discrimination remains to be determined. Studies of these compounds may also aid in understanding how the mismatch-repair proteins are capable of discriminating between the various correct and incorrect pairs under conditions of normal DNA replication, a subject of current interest (JUNOP et al. 2001 ).

As far as we know, our study is the first to reveal the inactivation or killing of E. coli by a chemical through error catastrophe. Under conditions of error catastrophe (i) sufficient mutations are induced by the base analogue to saturate mismatch repair and (ii) the level of mutations induced under these conditions is high enough to prevent further propagation of the cells. This phenomenon had previously been demonstrated for spontaneously occurring mutations in proofreading- defective strains (SCHAAPER 1988 ; SCHAAPER and RADMAN 1989 ; FIJALKOWSKA and SCHAAPER 1996 ). An example of inactivation of yeast cells by the base analogue 6-hydroxyaminopurine has also been described (PAVLOV et al. 1988 ). As base analogues can be toxic to cells by mechanisms other than lethal mutation induction, this type of inactivation can occur only in the case of analogues that do not produce significant alternative toxicity effects. Analogues like this would be essentially indistinguishable from normal bases, except for their propensity to frequently mispair. The term "ideal mutagens" has been used to describe them (KHROMOVBORISOV 1997 ) and dP may be a useful example of this class.

Recently, it was demonstrated that base analogues could be used to generate a sufficient level of lethal mutations to inactivate HIV virus (LOEB et al. 1999 ). No saturation of mismatch repair has yet been demonstrated in mammalian cells (HARFE and JINKS-ROBERTSON 2000 ), but such mechanisms may well exist. The apparently limited capacity of E. coli mismatch repair is a somewhat puzzling phenomenon that may have physiological significance. In our experiments, saturation of mismatch repair can already be seen at low levels of dP when mutation rates are only moderately elevated (Fig 2). Under those conditions, a simple overproduction of MutL protein can reduce mutation to essentially background levels. MutL overproduction has also been shown to reduce the level of spontaneous mutations under conditions of adaptive mutagenesis (HARRIS et al. 1997 ; FOSTER 1999 ). Thus, limitations in mismatch-repair capacity can affect mutation rates of cells under normal or near-normal conditions.

	ACKNOWLEDGMENTS

We thank Youra Pavlov and Polina Shcherbakova of the National Institute of Environmental Health Sciences for helpful comments on the manuscript for this article.

Manuscript received June 1, 2001; Accepted for publication March 15, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ARONSHTAM, A. and M. G. MARINUS, 1996  Dominant negative mutator mutations in the mutL gene of Escherichia coli.. Nucleic Acids Res. 24:2498-2504.[Abstract/Free Full Text]

CUPPLES, C. G. and J. H. MILLER, 1989  A set of lacZ mutations in Escherichia coli that allow rapid detection of each of the six base substitutions. Proc. Natl. Acad. Sci. USA 86:5345-5349.[Abstract/Free Full Text]

CUPPLES, C. G., M. CABRERA, C. CRUZ, and J. H. MILLER, 1990  A set of lacZ mutations in Escherichia coli that allow rapid detection of specific frameshift mutations. Genetics 125:275-280.[Abstract]

FEIG, D. I., L. C. SOWERS, and L. A. LOEB, 1994  Reverse chemical mutagenesis: identification of the mutagenic lesions resulting from reactive oxygen species-mediated damage to DNA. Proc. Natl. Acad. Sci. USA 91:6609-6613.[Abstract/Free Full Text]

FIJALKOWSKA, I. J. and R. M. SCHAAPER, 1996  Mutants in the Exo I motif of Escherichia coli dnaQ: defective proofreading and inviability due to error catastrophe. Proc. Natl. Acad. Sci. USA 93:2856-2861.[Abstract/Free Full Text]

FOSTER, P. L., 1999  Are adaptive mutations due to a decline in mismatch repair? The evidence is lacking. Mutat. Res. 436:179-184.[Medline]

GLICKMAN, B. W., 1979  Spontaneous mutagenesis in Escherichia coli strains lacking 6-methyladenine residues in their DNA: an altered mutational spectrum in dam^- mutants. Mutat. Res. 61:153-162.[Medline]

GLICKMAN, B., P. VAN DEN ELSEN, and M. RADMAN, 1978  Induced mutagenesis in dam^- mutants of Escherichia coli: a role for 6-methyladenine residues in mutation avoidance. Mol. Gen. Genet. 163:307-312.[Medline]

GLICKMAN, B. W. and M. RADMAN, 1980  Escherichia coli mutator mutants deficient in methylation-instructed DNA mismatch correction. Proc. Natl. Acad. Sci. USA 77:1063-1067.[Abstract/Free Full Text]

GRAFSTROM, R. H. and R. H. HOESS, 1987  Nucleotide sequence of the Escherichia coli mutH gene. Nucleic Acids Res. 15:3073-3084.[Abstract/Free Full Text]

HARFE, B. D. and S. JINKS-ROBERTSON, 2000  DNA mismatch repair and genetic instability. Annu. Rev. Genet. 34:359-399.[Medline]

HARRIS, R. S., G. FENG, K. J. ROSS, R. SIDHU, and C. THULIN et al., 1997  Mismatch repair protein MutL becomes limiting during stationary-phase mutation. Genes Dev. 11:2426-2437.[Abstract/Free Full Text]

HILL, F., D. LOAKES, and D. M. BROWN, 1998  Polymerase recognition of synthetic oligodeoxyribonucleotides incorporating degenerate pyrimidine and purine bases. Proc. Natl. Acad. Sci. USA 95:4258-4263.[Abstract/Free Full Text]

JONES, M. and R. WAGNER, 1981  N-methyl-N'-nitro-N-nitrosoguanidine sensitivity of E. coli mutants deficient in DNA methylation and mismatch repair. Mol. Gen. Genet. 184:562-563.[Medline]

JUNOP, M. S., G. OBMOLOVA, K. RAUSCH, P. HSIEH, and W. YANG, 2001  Composite active site of an ABC ATPase: MutS uses ATP to verify mismatch recognition and authorize DNA repair. Mol. Cell 7:1-12.[Medline]

KAMIYA, H. and H. KASAI, 2000  2-hydroxy-dATP is incorporated opposite G by Escherichia coli DNA polymerase III resulting in high mutagenicity. Nucleic Acids Res. 28:1640-1646.[Abstract/Free Full Text]

KARRAN, P. and M. G. MARINUS, 1982  Mismatch correction at O⁶-methylguanine residues in E. coli DNA. Nature 296:868-869.[Medline]

KHROMOV-BORISOV, N. N., 1997  Naming the mutagenic nucleic acid base analogs: the Galatea syndrome. Mutat. Res. 379:95-103.[Medline]

LIN, P. K. T. and D. M. BROWN, 1989  Synthesis and duplex stability of oligonucleotides containing cytosine-thymine analogues. Nucleic Acids Res. 17:10373-10383.[Abstract/Free Full Text]

LIU, H., S. R. HEWITT, and J. B. HAYS, 2000  Antagonism of ultraviolet-light mutagenesis by the methyl-directed mismatch-repair system of Escherichia coli.. Genetics 154:503-512.[Abstract/Free Full Text]

LOEB, L. A., J. M. ESSIGMANN, F. KAZAZI, J. ZHANG, and K. D. ROSE et al., 1999  Lethal mutagenesis of HIV with mutagenic nucleoside analogs. Proc. Natl. Acad. Sci. USA 96:1492-1497.[Abstract/Free Full Text]

MAKI, H. and M. SEKIGUCHI, 1992  MutT protein specifically hydrolyses a potent mutagenic substrate for DNA synthesis. Nature 355:273-275.[Medline]

MILLER, J. H., 1972 Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

NEDDERMAN, A. N. R., M. J. STONE, D. H. WILLIAMS, P. K. T. LIN, and D. M. BROWN, 1993  Molecular basis for methoxyamine-initiated mutagenesis: ¹H nuclear magnetic resonance studies of oligonucleotide duplex containing base-modified cytosine residues. J. Mol. Biol. 230:1068-1076.[Medline]

NEGHISHI, K., M. KAWAKAMI, K. KAYASUGA, J. ODO, and H. HAYATSU, 1987  Improved synthesis of N⁴-aminocytidine. Chem. Pharm. Bull. 35:3884-3887.

NEGISHI, K., K. TAMANOI, M. ISHII, M. KAWAKAMI, and Y. YAMASHITA et al., 1988  Mutagenic nucleoside analog N⁴-aminocytidine: metabolism, incorporation into DNA, and mutagenesis in Escherichia coli.. J. Bacteriol. 170:5257-5262.[Abstract/Free Full Text]

NEGISHI, K., D. M. WILLIAMS, Y. INOUE, K. MORIYAMA, and D. M. BROWN et al., 1997  The mechanism of mutation induction by a hydrogen bond ambivalent, bicyclic N⁴-oxy-2'-deoxycytidine in Escherichia coli.. Nucleic Acids Res. 25:1548-1552.[Abstract/Free Full Text]

PAVLOV, Y. I., V. N. NOSKOV, Y. O. CHERNOFF, and D. A. GORDENIN, 1988  The study of LYS2 gene mutability in diploids of the yeast Saccharomyces cerevisiae. Sov. Genet. 24:1219-1225.

RENÉ, B., C. AUCLAIR, and C. PAOLETTI, 1988  Frameshift lesions induced by oxazolopyridocarbazoles are recognized by the mismatch repair system in Escherichia coli.. Mutat. Res. 193:269-273.[Medline]

RYDBERG, B., 1977  Bromouracil mutagenesis in Escherichia coli. Evidence for involvement of mismatch repair. Mol. Gen. Genet. 152:19-28.[Medline]

RYDBERG, B., 1978  Bromouracil mutagenesis and mismatch repair in mutator strains in Escherichia coli.. Mutat. Res. 52:11-24.[Medline]

SCHAAPER, R. M., 1988  Mechanisms of mutagenesis in the Escherichia coli mutator mutD5: role of DNA mismatch repair. Proc. Natl. Acad. Sci. USA 85:8126-8130.[Abstract/Free Full Text]

SCHAAPER, R. M. and R. L. DUNN, 1998  Effect of Escherichia coli dnaE antimutator mutants on mutagenesis by the base analog N⁴-aminocytidine. Mutat. Res. 402:23-28.[Medline]

SCHAAPER, R. M. and M. RADMAN, 1989  The extreme mutator effect of Escherichia coli mutD5 results from saturation of mismatch repair by excessive DNA replication errors. EMBO J. 8:3511-3516.[Medline]

SCHAAPER, R. M., B. N. DANFORTH, and B. W. GLICKMAN, 1985  Rapid repeated cloning of mutant lac repressor genes. Gene 39:181-189.[Medline]

SCHAAPER, R. M., B. I. BOND, and R. G. FOWLER, 1989  A·T C·G transversions and their prevention by the Escherichia coli mutT and mutHLS pathways. Mol. Gen. Genet. 219:256-262.[Medline]

SIEGEL, E. C., S. L. WAIN, S. F. MELTZER, M. L. BINION, and J. L. STEINBERG, 1982  Mutator mutations induced by the insertion of phage Mu and the transposable resistance elements Tn5 and Tn10.. Mutat. Res. 93:25-33.[Medline]

SKOPEK, T. R. and F. HUTCHINSON, 1984  Frameshift mutagenesis of lambda prophage by 9-aminoacridine, proflavin and ICR-191. Mol. Gen. Genet. 195:418-423.[Medline]

SLEDZIEWSKA-GOJSKA, E. and C. JANION, 1982  Effect of proofreading and dam-instructed mismatch repair system on N⁴-hydroxycytidine-induced mutagenesis. Mol. Gen. Genet. 186:411-418.[Medline]

STONE, M. J., A. N. R. NEDDERMAN, D. H. WILLIAMS, P. K. T. LIN, and D. M. BROWN, 1991  Molecular basis for methoxyamine-initiated mutagenesis: ¹H nuclear magnetic resonance studies of base-modified oligonucleotides. J. Mol. Biol. 222:711-723.[Medline]

VOGEL, H. J. and D. M. BONNER, 1956  Acetylornithinase of Escherichia coli: partial purification and some properties. J. Biol. Chem. 218:97-106.[Free Full Text]

WOODGATE, R., 1992  Construction of a umuDC operon substitution mutation in Escherichia coli.. Mutat. Res. 281:221-225.[Medline]

WU, T.-H. and M. G. MARINUS, 1994  Dominant-negative mutator mutations in the mutS gene of Escherichia coli.. J. Bacteriol. 176:5393-5400.[Abstract/Free Full Text]

ZAR, J. H., 1984 Biostatistical Analysis, Vol. 26, Ed. 2, Table B, pp. 592–601. Prentice-Hall, Englewood Cliffs, NJ.

This article has been cited by other articles:

				I. Matic, A. Babic, and M. Radman 2-Aminopurine Allows Interspecies Recombination by a Reversible Inactivation of the Escherichia coli Mismatch Repair System J. Bacteriol., February 15, 2003; 185(4): 1459 - 1461. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Negishi, K. Articles by Schaaper, R. M. Search for Related Content PubMed PubMed Citation Articles by Negishi, K. Articles by Schaaper, R. M.