- Split View
-
Views
-
Cite
Cite
Jeffrey H Miller, Pauline Funchain, Wendy Clendenin, Tiffany Huang, Anh Nguyen, Erika Wolff, Annie Yeung, Ju-Huei Chiang, Lilit Garibyan, Malgorzata M Slupska, Hanjing Yang, Escherichia coli Strains (ndk) Lacking Nucleoside Diphosphate Kinase Are Powerful Mutators for Base Substitutions and Frameshifts in Mismatch-Repair-Deficient Strains, Genetics, Volume 162, Issue 1, 1 September 2002, Pages 5–13, https://doi.org/10.1093/genetics/162.1.5
- Share Icon Share
Abstract
Nucleoside diphosphate (NDP) kinase is one of the enzymes that maintains triphosphate pools. Escherichia coli strains (ndk) lacking this enzyme have been shown to be modest base substitution mutators, and two members of the human family of NDP kinases act as tumor suppressors. We show here that in E. coli strains lacking NDP kinase high levels of mispairs are generated, but most of these are corrected by the mismatch-repair system. Double mutants that are ndk mutS, lacking both the NDP kinase and mismatch repair, have levels of base substitutions 15-fold higher and levels of certain frameshifts up to 10-fold higher than those of the respective mutations in mutS strains that are NDP kinase proficient. A sequence analysis of the specificity of base substitution mutations generated in ndk and ndk mutS backgrounds as well as other experiments suggests that NDP kinase deficiency stimulates polymerase errors that lead to A:T → G:C transitions and that the editing capacity of cells may be affected, leading to additional uncorrected mispairs and to A:T → T:A transversions.
NUCLEOSIDE diphosphate (NDP) kinase, a ubiquitous enzyme that maintains deoxynucleotide triphosphate pools by converting deoxynucleotide diphosphates to triphosphates, is highly conserved between bacteria, Drosophila, and humans (see reviews by Lacombe et al. 2000 and Lombardi et al. 2000). In Escherichia coli, the ndk gene encodes NDP kinase. A number of studies have shown that deoxynucleotide triphosphate levels are important for genetic stability (see review by Kunz et al. 1994). E. coli cells lacking NDP kinase are weak mutators, with moderate increases in mutations, as measured by the frequency of rifampicin-resistant (Rifr) and nalidic acid-resistant (Nalr) mutants in cell populations (Lu et al. 1995). Intracellular pools of dCTP are elevated 20-fold, and dGTP pools are elevated 7-fold (Lu et al. 1995). Defects in the Drosophila gene altered wing direction (awd) encoding NDP kinase lead to developmental abnormalities (Biggs et al. 1988, 1990). In humans, the nm23 family of genes (nm23-H1 to nm23-H8) encodes NDP kinases that are expressed in a tissue-specific manner and appear to play a role in normal development and differentiation (Lacombe et al. 2000; Lombardi et al. 2000 and references therein). The nm23-H1- and nm23-H2-encoded NDP kinases have been identified as tumor suppressors involved in tumor metastasis and pathogenesis (Leone et al. 1991, 1993a). The NM23-H2/NDP kinase appears to be a transcription factor for the c-MYC oncogene, among others (Postel et al. 1993, 2000). Identification of a cleavage activity at a sequence in the c-MYC promoter and partial sequence homologies with certain DNA glycosylases has led to the speculation that the NM23-H2/NDP kinase B might also play a role in DNA repair (Postel et al. 2000).
We show here that in NDP kinase (NDK)-deficient strains of E. coli very high levels of mispairs are generated, but most of these are corrected by the mismatch-repair system and to some degree by the editing function (ε) of DNA polymerase. These two systems normally safeguard the cell against polymerase errors (Scheuermann and Echols 1984; Modrich and Lahue 1996). However, we find that double mutants lacking both mismatch repair and NDP kinase have extraordinarily high levels of base substitutions and also of frameshifts at repeat-tract sequences that are 15- and 10-fold higher than those of the respective mutations in cells defective for mismatch repair but with normal NDK. We have analyzed the nature of these mutations with genetic reversion systems (Cupples and Miller 1989; Cupples et al. 1990) and by DNA sequencing of mutations in rpoB that lead to resistance to rifampicin (Rifr). In doing so we have extended the development of the rpoB/Rifr system (Jin and Gross 1988; Severinov et al. 1993) as a useful tool for analyzing chromosomal mutations. These and other experiments suggest that NDK deficiency provokes polymerase errors that lead to A:T → G:C transitions. The altered triphosphate pools may hinder the editing capacity of cells, leading to additional uncorrected mispairs. In the absence of mismatch repair (MMR-), the additional mispairs result in extremely high levels of mutations that are more than an order of magnitude greater than those seen in NDK-proficient MMR- strains. We consider these results with respect to the tumor suppressor activity of the human NDK protein and to the interaction of different alleles in generating cancer susceptibilities.
MATERIALS AND METHODS
Bacterial strains: CC101-108, CC110, and CC111 have been described previously (Cupples and Miller 1989; Cupples et al. 1990). Each of these strains is ara Δ(gpt-lac)5 thi/F′128 lacIZ proA+B+. The nature of the lacZ mutation differs in each case. In CC101-106 the lacZ mutation results from a base substitution at coding position 464 in lacZ. Reversion to Lac+ occurs by a different base substitution in each case (see Table 2). Strains CC107, -108, -110, and -111 carry a frameshift mutation in lacZ. Each reverts only via a specific frameshift (see Table 4). ndk, mutS, and ndk mutS derivatives of each of the CC101-111 strains were prepared by transduction of each strain with P1vir lysates (Miller 1992) grown on strains carrying a mini-Tn10 in ndk or a miniTn10cam in mutS (J. H. Miller, P. Funchain and A. Yeung, unpublished results) and selecting for either Tetr or Camr cells. Double mutants ndk::minTn10 mutS::minTn10cam were prepared in two successive transduction steps. All genetic procedures were carried out as described in Miller (1992). Derivatives of CC105 and CC107, carrying either pBR329 or pBR329dnaQ66 (Slupska et al. 1998), were prepared by transformation of electrocompetent cells. dnaQ66 is a weak mutator allele constructed by site-directed mutagenesis and results in a glycine-for-histidine substitution at position 66 in the ε subunit of DNA polymerase III (Slupska et al. 1998). Strain J93 (Funchain et al. 2000) is ara Δ(lac)RV thi.
Detection of insertions in ndk: We employed mini-Tn10 derivatives to generate insertions in chromosomal genes. We looked for weak frameshift mutators using the same procedures that are described in detail in Slupska et al. (2000) for the detection of rearrangement mutators, except that strain CC107 (see above) was used. The exact position of each insertion was determined by the method of Higashitani et al. (1994), as described in detail in Slupska et al. (2000). The junction point of the mini-Tn10tet insertion in ndk used here was at 229 bp from the start of the coding sequence.
Determination of mutation rates: Rifr and Nalr mutants and Lac+ revertants were determined as described previously (Cupples and Miller 1989; Miller 1992). Briefly, overnight cultures were seeded with ≤100 cells and grown at 37°. Dilutions were plated on selective media and on Luria broth (LB; for determining the viable cell titer). For determining Lac+ revertants in mutS derivatives of CC107 and CC108, cultures were seeded with 10 cells or less. For determinations in ndk mutS derivatives of CC107, cultures used for plating were seeded with 1 or 2 cells by diluting a starting culture to the point where a drop delivered to a set of cultures resulted in one-third of the cultures receiving no cells. Lac+ frequencies were determined with a cell density of 109 cells per plate to prevent additional growth of cells leading to additional revertants. When <109 cells from the actual culture were plated, 109 cells of a scavenger (J93; Δlac, see above) were added. Under these conditions 10-fold dilutions of the culture being tested yielded 10-fold fewer revertants. Mutant frequencies were determined, and the median frequency (f) from a set of cultures (the number of cultures varied from 11 to >50) was used to calculate the mutation rate (μ) per replication by the method of Drake (1991), using the formula: μ= f/ln Nμ, where N is the number of cells in the culture. Ninety-five percent confidence limits were determined according to Dixon and Massey (1969).
Chromosomal DNA isolation and sequencing: Chromosomal DNA was isolated using DNAzol reagent (GIBCO BRL, Rockville, MD) following the protocols supplied by the manufacturer. Portions of the rpoB gene were PCR amplified for cloning. The sequence of primers for amplifying the rpoB region encoding the main group of mutations (cluster II) was as follows: 5′ GATATGATCAACGCCAAGCCGATTT 3′ and 5′ TTACACCGGAGTCAACGGCAACAGC 3′. The sequence of primers for amplifying the earlier region (cluster 1) was 5′ AATGTCAAATCCGTGGCGT 3′ and 5′ CCAACCGCAGACA AGTCATA 3′. For some of the mutations in cluster II, the amplified fragments were cloned into the pCRII-TOPO vector using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA) and sequenced using vector primers. DNA sequencing was carried out by using [α-32P]dATP and a SequiTherm cycle sequencing kit (Epicentre Technologies, Madison, WI) with reagents and protocols supplied by the manufacturer. About one out of every five sequences yielded a secondary mutation, presumably introduced during the cloning step. In these cases the rpoB gene was cloned again from the same Rifr colony and resequenced. In each case one of the two mutations disappeared. We then confirmed many of the sequences by sequencing a PCR-amplified sample in the absence of cloning. Here, the rpoB gene was PCR amplified from genomic DNA using a different set of primers, which allowed us to sequence directly from the PCR product. The following primers were used for PCR: 5′ CGTCGTATCCGTTCCGTTGG 3′ and 5′ TTCACCC GGATACATCTCGTC 3′. The PCR product was purified using the MinElute PCR purification kit (QIAGEN, Valencia, CA). The purified PCR products were then sequenced using the SequiTherm EXCEL II DNA sequencing kit (Epicentre Technologies). The following primers were used for sequencing straight from the PCR product: 5′ CGTGTAGAGCGTGCGGT GAAA 3′ for cluster II and 5′ GAAGGCACCGTAAAAGACAT 3′ for cluster I.
Purification of NDK and NM23-H2: The E. coli ndk gene was amplified by PCR from E. coli genomic DNA and cloned into the NcoI and HindIII sites of the bacterial expression vector pQE60 (QIAGEN). The construct pQE60ndk allows the expression of a recombinant E. coli NDK protein that is the same as the native E. coli NDK protein. The open reading frame for the human NM23-H2 gene was amplified by PCR from the Hela cell line S3 cDNA library (Stratagene, La Jolla, CA). Standard site-directed mutagenesis (Sambrook et al. 1989) was used to eliminate a NcoI restriction site in the NM23-H2 coding region, while maintaining the same protein sequence. The NM23-H2 gene was then subcloned between the NcoI and HindIII sites of pQE60. The construct pQE60NM23-H2 allows the expression of a recombinant human NM23-H2 protein that is the same as the native human NM23-H2 protein. The ndk derivative of E. coli strain CC107 (see above), CC107ndk, was first transformed with the plasmid pREP4 (QIAGEN), which constitutively expresses the Lac repressor protein encoded by the lacI gene, to reduce the basal level of expression from pQE60 plasmids. As a second step, the CC107ndk strain was transformed with either pQE60ndk or pQE60NM23-H2. These new constructs were grown at 37° in LB medium (Miller 1992) with 100 μg/ml ampicillin and 25 μg/ml kanamycin.
. | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|
Locus . | Wild type . | ndk . | mutS . | ndk mutS . |
rpoB (Rifr) | 1.5 | 27 | 120 | 1900 |
(1.1-1.7)a | (19-44) | (110-140) | (1500-2400) | |
gyrB (Nalr) | 0.25 | 4.8 | 35 | 310 |
(0.01-0.28) | (2.6-6.4) | (27-40) | (250-340) |
. | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|
Locus . | Wild type . | ndk . | mutS . | ndk mutS . |
rpoB (Rifr) | 1.5 | 27 | 120 | 1900 |
(1.1-1.7)a | (19-44) | (110-140) | (1500-2400) | |
gyrB (Nalr) | 0.25 | 4.8 | 35 | 310 |
(0.01-0.28) | (2.6-6.4) | (27-40) | (250-340) |
The mutation rate (μ) was determined in strain CC107 by the method of Drake (1991; see materials and methods).
Values in parentheses are 95% confidence limits.
. | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|
Locus . | Wild type . | ndk . | mutS . | ndk mutS . |
rpoB (Rifr) | 1.5 | 27 | 120 | 1900 |
(1.1-1.7)a | (19-44) | (110-140) | (1500-2400) | |
gyrB (Nalr) | 0.25 | 4.8 | 35 | 310 |
(0.01-0.28) | (2.6-6.4) | (27-40) | (250-340) |
. | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|
Locus . | Wild type . | ndk . | mutS . | ndk mutS . |
rpoB (Rifr) | 1.5 | 27 | 120 | 1900 |
(1.1-1.7)a | (19-44) | (110-140) | (1500-2400) | |
gyrB (Nalr) | 0.25 | 4.8 | 35 | 310 |
(0.01-0.28) | (2.6-6.4) | (27-40) | (250-340) |
The mutation rate (μ) was determined in strain CC107 by the method of Drake (1991; see materials and methods).
Values in parentheses are 95% confidence limits.
RESULTS
Mutation rates in strains lacking NDK and MMR: Base substitutions: We detected inserts of miniTn10 in the ndk gene after a random search for weak frameshift mutators and verified their structure by DNA sequencing (see materials and methods). We prepared double mutants, ndk mutS, by P1 transduction and assayed them initially for increases in Rifr and Nalr mutant frequencies. Table 1 presents these data in terms of mutation rate per replication, using the method of Drake (1991; see materials and methods). Strains that are ndk have rates of rpoB (Rifr) and gyrB (Nalr) mutations modestly higher than those of wild type, as reported by Inouye and co-workers (Lu et al. 1995). Strains that are mutS and thus lack mismatch repair have greater increases in mutation rates using these assays. However, what is so striking is the increase in ndk mutS double mutants. The mutation rates in the double mutant are 16-fold higher than those found in a mutS strain, when the Rifr assay is used, and 9-fold higher when the Nalr assay is used.
Base substitutions in lacZ: To analyze further the specificity of base substitutions generated in ndk and ndk mutS strains, we first looked at different derivatives of the series of strains CC101-CC106, which measure each of the specific base substitutions at one site in the lacZ gene (Cupples and Miller 1989). Table 2 shows these results. In ndk strains, A:T → T:A transverisons are the most prominent, although→marginal effects are seen with other base substitutions. A:T → G:C transitions are elevated 10-fold at the one site examined → (CC106), but the absolute rate of these mutations is still very low compared with other rates (see below). However, in ndk mutS double mutants A:T → G:C transitions are clearly elevated over the levels found in mutS strains alone. The level of A:T → G:C and A:T → TA changes in ndk mutS double mutants appears too low to account for the mutations in rpoB that generate Rifr mutants (Table 1), even when one takes into account the fact that a number of sites in rpoB can be mutated to yield the Rifr phenotype (Jin and Gross 1988). Therefore, we analyzed the sequence changes that result in the Rifr phenotype, as described below.
Base substitutions in rpoB: Sequence studies from a number of laboratories have analyzed mutations in rpoB leading to Rifr. Jin and Gross (1988) and subsequently Severinov et al. (1993) have compiled the sequence changes that they and others detected so far. Virtually all of these fall into two nearby subregions of the >4-kb rpoB gene, permitting sequencing by a single pair of primers. These previous studies identified 34 single-base substitutions at 24 sites (specific nucleotide pairs) distributed among 18 coding positions. Our work, described below, has detected an additional 13 single-base substitutions, 5 new sites, and 3 new coding positions. With these additions, the rpoB/Rifr system now allows the monitoring of 47 base substitutions at 29 sites distributed among 21 coding positions. Table 3 lists the base changes arranged according to the category of substitution. This system tests for A:T → G:C changes at 8 sites, G:C → A:T changes at 12 sites, A:T → TA changes at 7 sites, A:T → C:G changes at 7 sites, G:C → T:A changes at 10 sites, and G:C → C:G changes at 3 sites. (A more complete description of the extended rpoB/Rifr system will be published elsewhere.)
We sequenced the rpoB mutations responsible for the Rifr phenotype in a total of >200 independent mutants from the following four backgrounds: wild type, ndk, mutS, and mutS ndk. The results are shown in Table 3. In the data set from the wild-type strain, all types of mutations are detected, although the A:T → G:C change at prominent. one particular site (base 1547) is the most In the other data sets, this is the predominant mutation detected. The percentage of mutations at this hotspot (A:T → G:C, 1547) varies from 15% of the spontaneous base substitutions detected to 77% in the ndk background. Although G:C → A:T changes are seen in the spontaneous and mutS backgrounds, none of these appear in the ndk or the mutS ndk backgrounds. Table 4 shows the surrounding sequence for each of the sites at which A:T → G:C substitutions are measured. Both the hotspot site→and the second most represented A:T → G:C site have a 5′-G as nearest neighbor to the T that changes to a C. This 5′-G preference has been seen before for T → C transitions in a MMR- stain. (Schaaper and Dunn 1987; see discussion).
. | . | Rate (μ) per replication ×109 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC101 | AT → CG | 0.9 | 1.5 | 3.5 | 0.9 |
(0.4-1.6) | (0.9-1.8) | (1.1-8.5) | (0.3-2.3) | ||
CC102 | GC → AT | 1.8 | 2.5 | 150 | 160 |
(0.7-2.3) | (1.5-3.4) | (110-210) | (130-200) | ||
CC103 | GC → CG | <0.1 | <0.1 | <0.1 | <0.1 |
CC104 | GC → TA | 3.2 | 4.3 | 8.9 | 4.0 |
(2.6-6.6) | (3.1-5.4) | (6.4-13) | (2.8-5.2) | ||
CC105 | AT → TA | 1.1 | 59 | 2.9 | 55 |
(1.5-1.7) | (36-96) | (2.0-4.5) | (44-80) | ||
CC106 | AT → GC | 0.13 | 1.4 | 35 | 130 |
(0.11-0.17) | (0.6-3.6) | (27-44) | (110-180) |
. | . | Rate (μ) per replication ×109 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC101 | AT → CG | 0.9 | 1.5 | 3.5 | 0.9 |
(0.4-1.6) | (0.9-1.8) | (1.1-8.5) | (0.3-2.3) | ||
CC102 | GC → AT | 1.8 | 2.5 | 150 | 160 |
(0.7-2.3) | (1.5-3.4) | (110-210) | (130-200) | ||
CC103 | GC → CG | <0.1 | <0.1 | <0.1 | <0.1 |
CC104 | GC → TA | 3.2 | 4.3 | 8.9 | 4.0 |
(2.6-6.6) | (3.1-5.4) | (6.4-13) | (2.8-5.2) | ||
CC105 | AT → TA | 1.1 | 59 | 2.9 | 55 |
(1.5-1.7) | (36-96) | (2.0-4.5) | (44-80) | ||
CC106 | AT → GC | 0.13 | 1.4 | 35 | 130 |
(0.11-0.17) | (0.6-3.6) | (27-44) | (110-180) |
Values in parentheses are 95% confidence limits.
. | . | Rate (μ) per replication ×109 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC101 | AT → CG | 0.9 | 1.5 | 3.5 | 0.9 |
(0.4-1.6) | (0.9-1.8) | (1.1-8.5) | (0.3-2.3) | ||
CC102 | GC → AT | 1.8 | 2.5 | 150 | 160 |
(0.7-2.3) | (1.5-3.4) | (110-210) | (130-200) | ||
CC103 | GC → CG | <0.1 | <0.1 | <0.1 | <0.1 |
CC104 | GC → TA | 3.2 | 4.3 | 8.9 | 4.0 |
(2.6-6.6) | (3.1-5.4) | (6.4-13) | (2.8-5.2) | ||
CC105 | AT → TA | 1.1 | 59 | 2.9 | 55 |
(1.5-1.7) | (36-96) | (2.0-4.5) | (44-80) | ||
CC106 | AT → GC | 0.13 | 1.4 | 35 | 130 |
(0.11-0.17) | (0.6-3.6) | (27-44) | (110-180) |
. | . | Rate (μ) per replication ×109 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC101 | AT → CG | 0.9 | 1.5 | 3.5 | 0.9 |
(0.4-1.6) | (0.9-1.8) | (1.1-8.5) | (0.3-2.3) | ||
CC102 | GC → AT | 1.8 | 2.5 | 150 | 160 |
(0.7-2.3) | (1.5-3.4) | (110-210) | (130-200) | ||
CC103 | GC → CG | <0.1 | <0.1 | <0.1 | <0.1 |
CC104 | GC → TA | 3.2 | 4.3 | 8.9 | 4.0 |
(2.6-6.6) | (3.1-5.4) | (6.4-13) | (2.8-5.2) | ||
CC105 | AT → TA | 1.1 | 59 | 2.9 | 55 |
(1.5-1.7) | (36-96) | (2.0-4.5) | (44-80) | ||
CC106 | AT → GC | 0.13 | 1.4 | 35 | 130 |
(0.11-0.17) | (0.6-3.6) | (27-44) | (110-180) |
Values in parentheses are 95% confidence limits.
Frameshifts: We originally detected the miniTn10 insertion in ndk during a search for frameshift mutators. Table 5 provides mutation rates for frameshifts using strains CC107, CC108, CC110, and CC111. These strains revert from the addition or loss of a single base pair from a run of six G’s or six A’s (see Table 7 and Cupples et al. 1990). The biggest effect is seen in strain CC107, which reverts from lacZ to wild type by the addition of a G to a run of six -G’s. In MMR--deficient strains (mutS) frameshifts at repeat-tract sequences such as that found in lac in CC107 occur at very high levels as reflected in Table 7. However, even these high levels are elevated an additional 8- to 10-fold in ndk mutS double mutants. The levels reported in Table 5 (see also materials and methods) are so high that they can be verified by direct visualization of Lac+ colonies on indicator plates without the use of any selection (data not shown). Strain CC108, which reverts to Lac+ by the loss of a G from a run of six G’s, shows smaller increases in a ndk and a mutS ndk strain. One of two frameshifts at runs of A’s is also affected. Thus, CC110 shows an increase, but the absolute rate of frameshifts is low compared to that seen in CC107 and CC108. CC111 does not show a significant increase in the mutS ndk strain vs. the mutS background.
Effect of mutD alleles: We tested the effect of a partially impaired polymerase editing function, ε, on an NDP kinase-deficient strain. We previously constructed a set of single-base changes in a dnaQ insert on the plasmid pBR329 that result in varying levels of mutator activity (mutD; Slupska et al. 1998). These effects are seen when the plasmid with the altered dnaQ gene is carried by a strain with a normal chromosomal copy of dnaQ. Those that generate weak mutators have reduced activity of the editing function. Table 6 shows the effects of one of these alleles, mutD66, which results in a glycine-for-histidine substitution at position 66 in the ε subunit. In otherwise wild-type strains, pBRmutD66 results in a moderate increase in mutations as measured by the Rifr assay. However, a double mutant that is ndk/pBRmutD66 has significantly increased mutation rates for rpoB (Rifr). We also tested whether the increase in A:T → T:A transversions in a ndk strain (Table 2) might be the result of a secondary effect of saturating the mutD-encoded certain conditions (ε) editing function, since A:T → T:A transversions are prominent in a mutD strain under (Schaaper 1988). Therefore, we measured Lac+ revertants in strain CC105, which reverts only by an A:T → T:A tranversion, in ndk, ndk/pBR, and WT/pBRmutD66 strains. As can be seen in Table 6, the ndk/pBRmutD66 strain displays a significant increase in reversion of the lacZ mutation in CC105 over that seen in ndk or ndk/pBRmutD66 strains alone. Although it is possible that the NDP kinase deficiency generates more polymerase errors that lead directly to A:A (or T:T) mispairs, the data in Table 6 are consistent with the increase in A:T → T:A transversions occurring indirectly as a result of inactivating the editing function (see discussion).
Complementation of E. coli ndk with human nm23-H2: We expressed native NDK protein from both E. coli genomic and human cDNA (see materials and methods), using the inducible bacterial expression vector pQE60 (QIAGEN) and the plasmid pREP4 (QIAGEN) that expresses the Lac repressor protein. Here, expression from pQE60 is increased in the presence of isopropyl thiogalactoside (IPTG). Table 7 shows the mutation rate for rpoB mutations leading to Rifr in strains lacking a functional chromosomal ndk gene. With the pQE60 lacking an insert, the mutation rate is high in the presence and absence of IPTG, since the NDK deficiency leads to a modest mutator phenotype (see Table 1). With the pQE60 carrying the E. coli ndk gene, the mutator effect is eliminated. The residual expression of the ndk gene on pQE60 in the absence of IPTG is still enough to provide sufficient NDK function. With the pQE60 carrying the human nm23-H2, the mutator effect is only partially overcome in the absence of IPTG, but completely nullified in the presence of IPTG, since the increased expression provides enough NDK function. This experiment shows that the human enzyme can provide NDK function to E. coli.
. | . | . | Occurrence of mutations . | |||
---|---|---|---|---|---|---|
Site (bp) . | AA change . | bp change . | Wild type . | ndk . | mutS . | ndk mutS . |
443a | Q148R | AT → GC | 3 | 1 | 0 | 0 |
1532 | L511P | AT → GC | 0 | 0 | 4 | 3 |
1534 | S512P | AT → GC | 6 | 2 | 8 | 7 |
1538 | Q513R | AT → GC | 0 | 0 | 3 | 0 |
1547 | D516G | AT → GC | 9 | 46 | 29 | 22 |
1552a | N518D | AT → GC | 0 | 0 | 2 | 0 |
1577 | H526R | AT → GC | 0 | 0 | 0 | 0 |
1598 | L533P | AT → GC | 0 | 0 | 0 | 0 |
1535 | S512F | GC → AT | 2 | 0 | 0 | 0 |
1546 | D516N | GC → AT | 2 | 0 | 6 | 0 |
1565 | S522F | GC → AT | 3 | 0 | 0 | 0 |
1576 | H526Y | GC → AT | 3 | 0 | 0 | 0 |
1585b | R529C | GC → AT | 0 | 0 | 0 | 0 |
1586 | R529H | GC → AT | 2 | 0 | 1 | 0 |
1592 | S531F | GC → AT | 1 | 0 | 1 | 0 |
1595 | A532V | GC → AT | 0 | 0 | 0 | 0 |
1600a | G534S | GC → AT | 0 | 0 | 3 | 0 |
1691 | P564L | GC → AT | 4 | 0 | 0 | 0 |
1721 | S574F | GC → AT | 0 | 0 | 0 | 0 |
2060b | R687H | GC → AT | 0 | 0 | 0 | 0 |
443a | Q148L | AT → TA | 6 | 4 | 0 | 0 |
1538 | Q513L | AT → TA | 1 | 2 | 0 | 0 |
1547 | D516V | AT → TA | 0 | 0 | 0 | 1 |
1568 | E523V | AT → TA | 0 | 0 | 0 | 0 |
1577a | H526L | AT → TA | 0 | 0 | 0 | 0 |
1598a | L533H | AT → TA | 1 | 0 | 0 | 0 |
1714 | I572F | AT → TA | 3 | 0 | 0 | 0 |
1525a | S509R | AT → CG | 1 | 0 | 0 | 0 |
1532a | L511Q | AT → CG | 0 | 1 | 0 | 0 |
1538 | Q513P | AT → CG | 0 | 0 | 0 | 0 |
1577 | H526P | AT → CG | 1 | 0 | 0 | 0 |
1687b | T563P | AT → CG | 0 | 0 | 0 | 0 |
1714a | I572L | AT → CG | 7 | 1 | 0 | 0 |
1715a | I572S | AT → CG | 0 | 2 | 0 | 0 |
1537 | Q513K | GC → TA | 0 | 0 | 0 | 0 |
1546 | D516Y | GC → TA | 1 | 0 | 0 | 0 |
1576a | H526N | GC → TA | 1 | 0 | 0 | 0 |
1578 | H526Q | GC → TA | 0 | 0 | 0 | 0 |
1585b | R529S | GC → TA | 0 | 0 | 0 | 0 |
1586 | R529L | GC → TA | 0 | 0 | 0 | 0 |
1592 | S531Y | GC → TA | 0 | 0 | 0 | 0 |
1595 | A532E | GC → TA | 0 | 0 | 0 | 0 |
1600a | G534C | GC → TA | 2 | 0 | 0 | 0 |
1601 | G534D | GC → TA | 0 | 0 | 0 | 0 |
1574 | T525R | GC → CG | 0 | 1 | 0 | 0 |
1576a | H526D | GC → CG | 3 | 0 | 0 | 0 |
1578 | H526Q | GC → CG | 0 | 0 | 0 | 0 |
Total | 62 | 60 | 57 | 33 |
. | . | . | Occurrence of mutations . | |||
---|---|---|---|---|---|---|
Site (bp) . | AA change . | bp change . | Wild type . | ndk . | mutS . | ndk mutS . |
443a | Q148R | AT → GC | 3 | 1 | 0 | 0 |
1532 | L511P | AT → GC | 0 | 0 | 4 | 3 |
1534 | S512P | AT → GC | 6 | 2 | 8 | 7 |
1538 | Q513R | AT → GC | 0 | 0 | 3 | 0 |
1547 | D516G | AT → GC | 9 | 46 | 29 | 22 |
1552a | N518D | AT → GC | 0 | 0 | 2 | 0 |
1577 | H526R | AT → GC | 0 | 0 | 0 | 0 |
1598 | L533P | AT → GC | 0 | 0 | 0 | 0 |
1535 | S512F | GC → AT | 2 | 0 | 0 | 0 |
1546 | D516N | GC → AT | 2 | 0 | 6 | 0 |
1565 | S522F | GC → AT | 3 | 0 | 0 | 0 |
1576 | H526Y | GC → AT | 3 | 0 | 0 | 0 |
1585b | R529C | GC → AT | 0 | 0 | 0 | 0 |
1586 | R529H | GC → AT | 2 | 0 | 1 | 0 |
1592 | S531F | GC → AT | 1 | 0 | 1 | 0 |
1595 | A532V | GC → AT | 0 | 0 | 0 | 0 |
1600a | G534S | GC → AT | 0 | 0 | 3 | 0 |
1691 | P564L | GC → AT | 4 | 0 | 0 | 0 |
1721 | S574F | GC → AT | 0 | 0 | 0 | 0 |
2060b | R687H | GC → AT | 0 | 0 | 0 | 0 |
443a | Q148L | AT → TA | 6 | 4 | 0 | 0 |
1538 | Q513L | AT → TA | 1 | 2 | 0 | 0 |
1547 | D516V | AT → TA | 0 | 0 | 0 | 1 |
1568 | E523V | AT → TA | 0 | 0 | 0 | 0 |
1577a | H526L | AT → TA | 0 | 0 | 0 | 0 |
1598a | L533H | AT → TA | 1 | 0 | 0 | 0 |
1714 | I572F | AT → TA | 3 | 0 | 0 | 0 |
1525a | S509R | AT → CG | 1 | 0 | 0 | 0 |
1532a | L511Q | AT → CG | 0 | 1 | 0 | 0 |
1538 | Q513P | AT → CG | 0 | 0 | 0 | 0 |
1577 | H526P | AT → CG | 1 | 0 | 0 | 0 |
1687b | T563P | AT → CG | 0 | 0 | 0 | 0 |
1714a | I572L | AT → CG | 7 | 1 | 0 | 0 |
1715a | I572S | AT → CG | 0 | 2 | 0 | 0 |
1537 | Q513K | GC → TA | 0 | 0 | 0 | 0 |
1546 | D516Y | GC → TA | 1 | 0 | 0 | 0 |
1576a | H526N | GC → TA | 1 | 0 | 0 | 0 |
1578 | H526Q | GC → TA | 0 | 0 | 0 | 0 |
1585b | R529S | GC → TA | 0 | 0 | 0 | 0 |
1586 | R529L | GC → TA | 0 | 0 | 0 | 0 |
1592 | S531Y | GC → TA | 0 | 0 | 0 | 0 |
1595 | A532E | GC → TA | 0 | 0 | 0 | 0 |
1600a | G534C | GC → TA | 2 | 0 | 0 | 0 |
1601 | G534D | GC → TA | 0 | 0 | 0 | 0 |
1574 | T525R | GC → CG | 0 | 1 | 0 | 0 |
1576a | H526D | GC → CG | 3 | 0 | 0 | 0 |
1578 | H526Q | GC → CG | 0 | 0 | 0 | 0 |
Total | 62 | 60 | 57 | 33 |
The DNA sequence change in rpoB was determined in each case.
Mutations not described previously in Jin and Gross (1989), Severinov et al. (1993), or Rangarajan et al. (1997).
Mutations that lead to a temperature-sensitive phenotype between 30° and 42° and may not have been detected at 37°.
. | . | . | Occurrence of mutations . | |||
---|---|---|---|---|---|---|
Site (bp) . | AA change . | bp change . | Wild type . | ndk . | mutS . | ndk mutS . |
443a | Q148R | AT → GC | 3 | 1 | 0 | 0 |
1532 | L511P | AT → GC | 0 | 0 | 4 | 3 |
1534 | S512P | AT → GC | 6 | 2 | 8 | 7 |
1538 | Q513R | AT → GC | 0 | 0 | 3 | 0 |
1547 | D516G | AT → GC | 9 | 46 | 29 | 22 |
1552a | N518D | AT → GC | 0 | 0 | 2 | 0 |
1577 | H526R | AT → GC | 0 | 0 | 0 | 0 |
1598 | L533P | AT → GC | 0 | 0 | 0 | 0 |
1535 | S512F | GC → AT | 2 | 0 | 0 | 0 |
1546 | D516N | GC → AT | 2 | 0 | 6 | 0 |
1565 | S522F | GC → AT | 3 | 0 | 0 | 0 |
1576 | H526Y | GC → AT | 3 | 0 | 0 | 0 |
1585b | R529C | GC → AT | 0 | 0 | 0 | 0 |
1586 | R529H | GC → AT | 2 | 0 | 1 | 0 |
1592 | S531F | GC → AT | 1 | 0 | 1 | 0 |
1595 | A532V | GC → AT | 0 | 0 | 0 | 0 |
1600a | G534S | GC → AT | 0 | 0 | 3 | 0 |
1691 | P564L | GC → AT | 4 | 0 | 0 | 0 |
1721 | S574F | GC → AT | 0 | 0 | 0 | 0 |
2060b | R687H | GC → AT | 0 | 0 | 0 | 0 |
443a | Q148L | AT → TA | 6 | 4 | 0 | 0 |
1538 | Q513L | AT → TA | 1 | 2 | 0 | 0 |
1547 | D516V | AT → TA | 0 | 0 | 0 | 1 |
1568 | E523V | AT → TA | 0 | 0 | 0 | 0 |
1577a | H526L | AT → TA | 0 | 0 | 0 | 0 |
1598a | L533H | AT → TA | 1 | 0 | 0 | 0 |
1714 | I572F | AT → TA | 3 | 0 | 0 | 0 |
1525a | S509R | AT → CG | 1 | 0 | 0 | 0 |
1532a | L511Q | AT → CG | 0 | 1 | 0 | 0 |
1538 | Q513P | AT → CG | 0 | 0 | 0 | 0 |
1577 | H526P | AT → CG | 1 | 0 | 0 | 0 |
1687b | T563P | AT → CG | 0 | 0 | 0 | 0 |
1714a | I572L | AT → CG | 7 | 1 | 0 | 0 |
1715a | I572S | AT → CG | 0 | 2 | 0 | 0 |
1537 | Q513K | GC → TA | 0 | 0 | 0 | 0 |
1546 | D516Y | GC → TA | 1 | 0 | 0 | 0 |
1576a | H526N | GC → TA | 1 | 0 | 0 | 0 |
1578 | H526Q | GC → TA | 0 | 0 | 0 | 0 |
1585b | R529S | GC → TA | 0 | 0 | 0 | 0 |
1586 | R529L | GC → TA | 0 | 0 | 0 | 0 |
1592 | S531Y | GC → TA | 0 | 0 | 0 | 0 |
1595 | A532E | GC → TA | 0 | 0 | 0 | 0 |
1600a | G534C | GC → TA | 2 | 0 | 0 | 0 |
1601 | G534D | GC → TA | 0 | 0 | 0 | 0 |
1574 | T525R | GC → CG | 0 | 1 | 0 | 0 |
1576a | H526D | GC → CG | 3 | 0 | 0 | 0 |
1578 | H526Q | GC → CG | 0 | 0 | 0 | 0 |
Total | 62 | 60 | 57 | 33 |
. | . | . | Occurrence of mutations . | |||
---|---|---|---|---|---|---|
Site (bp) . | AA change . | bp change . | Wild type . | ndk . | mutS . | ndk mutS . |
443a | Q148R | AT → GC | 3 | 1 | 0 | 0 |
1532 | L511P | AT → GC | 0 | 0 | 4 | 3 |
1534 | S512P | AT → GC | 6 | 2 | 8 | 7 |
1538 | Q513R | AT → GC | 0 | 0 | 3 | 0 |
1547 | D516G | AT → GC | 9 | 46 | 29 | 22 |
1552a | N518D | AT → GC | 0 | 0 | 2 | 0 |
1577 | H526R | AT → GC | 0 | 0 | 0 | 0 |
1598 | L533P | AT → GC | 0 | 0 | 0 | 0 |
1535 | S512F | GC → AT | 2 | 0 | 0 | 0 |
1546 | D516N | GC → AT | 2 | 0 | 6 | 0 |
1565 | S522F | GC → AT | 3 | 0 | 0 | 0 |
1576 | H526Y | GC → AT | 3 | 0 | 0 | 0 |
1585b | R529C | GC → AT | 0 | 0 | 0 | 0 |
1586 | R529H | GC → AT | 2 | 0 | 1 | 0 |
1592 | S531F | GC → AT | 1 | 0 | 1 | 0 |
1595 | A532V | GC → AT | 0 | 0 | 0 | 0 |
1600a | G534S | GC → AT | 0 | 0 | 3 | 0 |
1691 | P564L | GC → AT | 4 | 0 | 0 | 0 |
1721 | S574F | GC → AT | 0 | 0 | 0 | 0 |
2060b | R687H | GC → AT | 0 | 0 | 0 | 0 |
443a | Q148L | AT → TA | 6 | 4 | 0 | 0 |
1538 | Q513L | AT → TA | 1 | 2 | 0 | 0 |
1547 | D516V | AT → TA | 0 | 0 | 0 | 1 |
1568 | E523V | AT → TA | 0 | 0 | 0 | 0 |
1577a | H526L | AT → TA | 0 | 0 | 0 | 0 |
1598a | L533H | AT → TA | 1 | 0 | 0 | 0 |
1714 | I572F | AT → TA | 3 | 0 | 0 | 0 |
1525a | S509R | AT → CG | 1 | 0 | 0 | 0 |
1532a | L511Q | AT → CG | 0 | 1 | 0 | 0 |
1538 | Q513P | AT → CG | 0 | 0 | 0 | 0 |
1577 | H526P | AT → CG | 1 | 0 | 0 | 0 |
1687b | T563P | AT → CG | 0 | 0 | 0 | 0 |
1714a | I572L | AT → CG | 7 | 1 | 0 | 0 |
1715a | I572S | AT → CG | 0 | 2 | 0 | 0 |
1537 | Q513K | GC → TA | 0 | 0 | 0 | 0 |
1546 | D516Y | GC → TA | 1 | 0 | 0 | 0 |
1576a | H526N | GC → TA | 1 | 0 | 0 | 0 |
1578 | H526Q | GC → TA | 0 | 0 | 0 | 0 |
1585b | R529S | GC → TA | 0 | 0 | 0 | 0 |
1586 | R529L | GC → TA | 0 | 0 | 0 | 0 |
1592 | S531Y | GC → TA | 0 | 0 | 0 | 0 |
1595 | A532E | GC → TA | 0 | 0 | 0 | 0 |
1600a | G534C | GC → TA | 2 | 0 | 0 | 0 |
1601 | G534D | GC → TA | 0 | 0 | 0 | 0 |
1574 | T525R | GC → CG | 0 | 1 | 0 | 0 |
1576a | H526D | GC → CG | 3 | 0 | 0 | 0 |
1578 | H526Q | GC → CG | 0 | 0 | 0 | 0 |
Total | 62 | 60 | 57 | 33 |
The DNA sequence change in rpoB was determined in each case.
Mutations not described previously in Jin and Gross (1989), Severinov et al. (1993), or Rangarajan et al. (1997).
Mutations that lead to a temperature-sensitive phenotype between 30° and 42° and may not have been detected at 37°.
Site (bp) . | 5′ sequence 3′ . |
---|---|
1532 | 5′ CCAGCTGTCTC 3′ |
3′ GGTCGACAGAG 5′ | |
1534 | 5′ AGCTGTCTCAG 3′ |
3′ TCGACAGAGTC 5′ | |
1538 | 5′ TAAACTGAGAC 3′ |
3′ ATTTGACTCTG 5′ | |
1547 | 5′ TCTGGTCCATA 3′ |
3′ AGACCAGGTAT 5′ | |
1552 | 5′ GTTGTTCTGGT 3′ |
3′ CAACAAGACCA 5′ | |
1577 | 5′ GTTTGTGCGTA 3′ |
3′ CAAACACGCAT 5′ | |
1598 | 5′ CGCACTCGGCC 3′ |
3′ GCGTGAGCCGG 5′ |
Site (bp) . | 5′ sequence 3′ . |
---|---|
1532 | 5′ CCAGCTGTCTC 3′ |
3′ GGTCGACAGAG 5′ | |
1534 | 5′ AGCTGTCTCAG 3′ |
3′ TCGACAGAGTC 5′ | |
1538 | 5′ TAAACTGAGAC 3′ |
3′ ATTTGACTCTG 5′ | |
1547 | 5′ TCTGGTCCATA 3′ |
3′ AGACCAGGTAT 5′ | |
1552 | 5′ GTTGTTCTGGT 3′ |
3′ CAACAAGACCA 5′ | |
1577 | 5′ GTTTGTGCGTA 3′ |
3′ CAAACACGCAT 5′ | |
1598 | 5′ CGCACTCGGCC 3′ |
3′ GCGTGAGCCGG 5′ |
The sequences in rpoB have been oriented to show the base 5′ to the T undergoing the T → C change.
Site (bp) . | 5′ sequence 3′ . |
---|---|
1532 | 5′ CCAGCTGTCTC 3′ |
3′ GGTCGACAGAG 5′ | |
1534 | 5′ AGCTGTCTCAG 3′ |
3′ TCGACAGAGTC 5′ | |
1538 | 5′ TAAACTGAGAC 3′ |
3′ ATTTGACTCTG 5′ | |
1547 | 5′ TCTGGTCCATA 3′ |
3′ AGACCAGGTAT 5′ | |
1552 | 5′ GTTGTTCTGGT 3′ |
3′ CAACAAGACCA 5′ | |
1577 | 5′ GTTTGTGCGTA 3′ |
3′ CAAACACGCAT 5′ | |
1598 | 5′ CGCACTCGGCC 3′ |
3′ GCGTGAGCCGG 5′ |
Site (bp) . | 5′ sequence 3′ . |
---|---|
1532 | 5′ CCAGCTGTCTC 3′ |
3′ GGTCGACAGAG 5′ | |
1534 | 5′ AGCTGTCTCAG 3′ |
3′ TCGACAGAGTC 5′ | |
1538 | 5′ TAAACTGAGAC 3′ |
3′ ATTTGACTCTG 5′ | |
1547 | 5′ TCTGGTCCATA 3′ |
3′ AGACCAGGTAT 5′ | |
1552 | 5′ GTTGTTCTGGT 3′ |
3′ CAACAAGACCA 5′ | |
1577 | 5′ GTTTGTGCGTA 3′ |
3′ CAAACACGCAT 5′ | |
1598 | 5′ CGCACTCGGCC 3′ |
3′ GCGTGAGCCGG 5′ |
The sequences in rpoB have been oriented to show the base 5′ to the T undergoing the T → C change.
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC107 | 6 G’s → 7 G’s | 8.9 | 74 | 3,600 | 28,000 |
(+1) | (7.4-12) | (59-120) | (3,000-3,900) | (24,000-42,000) | |
CC108 | 6 G’s → 5 G’s | 7.3 | 23 | 1,900 | 4,500 |
(-1) | (6.0-9.4) | (13-29) | (1,800-2,100) | (3,300-5,000) | |
CC110 | 6 A’s → 7 A’s | 0.5 | 0.5 | 38 | 160 |
(+1) | (0.2-0.8) | (0.4-1.0) | (34-115) | (110-230) | |
CC111 | 6 A’s → 5 A’s | 2.8 | 7.9 | 320 | 400 |
(-1) | (2.3-4.2) | (5.9-9.5) | (270-370) | (220-470) |
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC107 | 6 G’s → 7 G’s | 8.9 | 74 | 3,600 | 28,000 |
(+1) | (7.4-12) | (59-120) | (3,000-3,900) | (24,000-42,000) | |
CC108 | 6 G’s → 5 G’s | 7.3 | 23 | 1,900 | 4,500 |
(-1) | (6.0-9.4) | (13-29) | (1,800-2,100) | (3,300-5,000) | |
CC110 | 6 A’s → 7 A’s | 0.5 | 0.5 | 38 | 160 |
(+1) | (0.2-0.8) | (0.4-1.0) | (34-115) | (110-230) | |
CC111 | 6 A’s → 5 A’s | 2.8 | 7.9 | 320 | 400 |
(-1) | (2.3-4.2) | (5.9-9.5) | (270-370) | (220-470) |
Base additions (+1) or deletions (-1) that lead to the Lac+ phenotype are scored. Values in parentheses indicate 95% confidence limits.
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC107 | 6 G’s → 7 G’s | 8.9 | 74 | 3,600 | 28,000 |
(+1) | (7.4-12) | (59-120) | (3,000-3,900) | (24,000-42,000) | |
CC108 | 6 G’s → 5 G’s | 7.3 | 23 | 1,900 | 4,500 |
(-1) | (6.0-9.4) | (13-29) | (1,800-2,100) | (3,300-5,000) | |
CC110 | 6 A’s → 7 A’s | 0.5 | 0.5 | 38 | 160 |
(+1) | (0.2-0.8) | (0.4-1.0) | (34-115) | (110-230) | |
CC111 | 6 A’s → 5 A’s | 2.8 | 7.9 | 320 | 400 |
(-1) | (2.3-4.2) | (5.9-9.5) | (270-370) | (220-470) |
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Lac- strain . | Base change . | Wild type . | ndk . | mutS . | ndk mutS . |
CC107 | 6 G’s → 7 G’s | 8.9 | 74 | 3,600 | 28,000 |
(+1) | (7.4-12) | (59-120) | (3,000-3,900) | (24,000-42,000) | |
CC108 | 6 G’s → 5 G’s | 7.3 | 23 | 1,900 | 4,500 |
(-1) | (6.0-9.4) | (13-29) | (1,800-2,100) | (3,300-5,000) | |
CC110 | 6 A’s → 7 A’s | 0.5 | 0.5 | 38 | 160 |
(+1) | (0.2-0.8) | (0.4-1.0) | (34-115) | (110-230) | |
CC111 | 6 A’s → 5 A’s | 2.8 | 7.9 | 320 | 400 |
(-1) | (2.3-4.2) | (5.9-9.5) | (270-370) | (220-470) |
Base additions (+1) or deletions (-1) that lead to the Lac+ phenotype are scored. Values in parentheses indicate 95% confidence limits.
DISCUSSION
Understanding pathways of mutagenesis is relevant to the study of repair systems and to certain human diseases. As a tool, mutators, strains with a higher than normal mutation rate, have played an important role in the elucidation of DNA repair systems and the characterization of new pathways of mutagenesis, both in bacteria and in higher cells (see review by Miller 1998). For instance, the definition of mutH, mutL, mutS, and mutU (now termed uvrD) has been instrumental in understanding the mismatch-repair system in E. coli (Radman and Wagner 1986; Modrich and Lahue 1996; Modrich 1997). Subsequent studies identified the human counterpart to this system and showed its involvement in the inherited form of colon and ovarian cancer (Fishel et al. 1993; Leach et al. 1993; see also Kolodner 1996; Modrich 1997). The importance of this area of research is underscored by the fact that some tumor lines are mutators (Aaltonen et al. 1993; Leach et al., 1993; Peltomäki et al. 1993; Rampino et al. 1997), and several authors have postulated that an early step in the origin of a cancer cell is the creation of a mutator phenotype (Nowell 1976; Loeb 1991), since this would accelerate the accumulation of the five to seven mutations needed to result in a complete tumor cell. Thus, up to 30% of colon cancer lines are mutators lacking the mismatch-repair system. Detecting new pathways of mutagenesis or mutagenesis enhancement can thus open up new avenues of research into the causes of cancer.
It is well established that perturbing the dNTP pool balance can result in increased mutation and is relevant to human disease (see review by Kunz et al. 1994). In E. coli, the ndk gene encodes NDP kinase that plays a key role in maintaining the triphosphate pool (see Lu et al. 1995). This enzyme is highly conserved from bacteria to Drosophila to humans and appears to be involved in a number of important cellular processes. For instance, in Drosophila, defects at the NDP kinase-encoding awd locus result in abnormal development (Biggs et al. 1988, 1990). In humans, a family of NDP kinase-encoding genes (nm23-H1 to nm23-H8) is expressed in a tissue-specific manner and is also postulated to play a role in normal development and in some cases a direct involvement in the proliferation of tumors (Lacombe et al. 2000; Lombardi et al. 2000). In vivo experiments have shown that nm23-H1 suppresses metastasis of human breast carcinoma cells (Leone et al. 1993a), and point mutations have been found in nm23-H1 and nm23-H2 in several tumor cell lines (Leone et al. 1993b; Chang et al. 1994; Hamby et al. 1995). The nm23-H2-encoded NDP kinase is also a regulator of at least the c-MYC oncogene (Postel et al. 1993, 2000) and makes a sequence-specific cleavage at the c-MYC promoter sequence (Postel et al. 2000). On the basis of sequence homologies with certain glycosylases, it has been postulated that the NM23-H2/NDP kinase may function as a repair protein in humans (Postel et al. 2000). The NM23-H1/NDP kinase has been shown to regulate Rac1 GTPase by interaction with Tiam1 (Otsuki et al. 2001).
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Locus . | Wild type . | ndk . | ndk/pBR . | WT/pBRmutD66 . | Ndk/pBRmutD66 . |
rpoB (Rifr)a | 1.5 | 27 | 40 | 32 | 470 |
(1.1-1.7) | (19-44) | (17-58) | (26-49) | (400-760) | |
lacZ (Lac+)b | 0.11 | 5.9 | 2.1 | 6.2 | 140 |
AT → TA | (0.05-0.17) | (3.6-9.6) | (1.3-3.3) | (5.3-9.6) | (100-190) |
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Locus . | Wild type . | ndk . | ndk/pBR . | WT/pBRmutD66 . | Ndk/pBRmutD66 . |
rpoB (Rifr)a | 1.5 | 27 | 40 | 32 | 470 |
(1.1-1.7) | (19-44) | (17-58) | (26-49) | (400-760) | |
lacZ (Lac+)b | 0.11 | 5.9 | 2.1 | 6.2 | 140 |
AT → TA | (0.05-0.17) | (3.6-9.6) | (1.3-3.3) | (5.3-9.6) | (100-190) |
Wild-type and ndk strains are shown. In some cases these strains carry either pBR329 or pBR329 with a mutated dnaQ gene (pBR329mutD66). Values in parentheses are 95% confidence limits.
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Locus . | Wild type . | ndk . | ndk/pBR . | WT/pBRmutD66 . | Ndk/pBRmutD66 . |
rpoB (Rifr)a | 1.5 | 27 | 40 | 32 | 470 |
(1.1-1.7) | (19-44) | (17-58) | (26-49) | (400-760) | |
lacZ (Lac+)b | 0.11 | 5.9 | 2.1 | 6.2 | 140 |
AT → TA | (0.05-0.17) | (3.6-9.6) | (1.3-3.3) | (5.3-9.6) | (100-190) |
. | . | Rate (μ) per replication ×108 . | |||
---|---|---|---|---|---|
Locus . | Wild type . | ndk . | ndk/pBR . | WT/pBRmutD66 . | Ndk/pBRmutD66 . |
rpoB (Rifr)a | 1.5 | 27 | 40 | 32 | 470 |
(1.1-1.7) | (19-44) | (17-58) | (26-49) | (400-760) | |
lacZ (Lac+)b | 0.11 | 5.9 | 2.1 | 6.2 | 140 |
AT → TA | (0.05-0.17) | (3.6-9.6) | (1.3-3.3) | (5.3-9.6) | (100-190) |
Wild-type and ndk strains are shown. In some cases these strains carry either pBR329 or pBR329 with a mutated dnaQ gene (pBR329mutD66). Values in parentheses are 95% confidence limits.
We have examined the involvement of the E. coli ndk gene in mutagenesis. Earlier studies showed a moderate mutator activity for base substitutions of NDP kinase-deficient strains in E. coli (Lu et al. 1995). We detected minitransposon inserts in the E. coli ndk gene as part of a search for frameshift mutators (see materials and methods) and found that in the absence of mismatch repair NDP kinase-deficient strains (mutS ndk) generate extremely high levels of both base substitutions and frameshifts that reflect mutation rates 16- and 8-fold higher, respectively, than those seen in MMR-deficient (mutS) strains alone and rates 75-fold higher than those seen in MMR-proficient ndk strains (see Tables 1 and 5). What can account for this very high level of mutator activity? Is this due simply to the effects of altered nucleotide triphosphate pool sizes on polymerase errors or to the lack of a repair activity provided by the NDP kinase? We tested purified native E. coli NDK protein, as well as purified native human NM23-H2 protein for glycosylase activity, using a complete collection of single-base mispairs, as well as several frameshift mispairs, and failed to detect any glycosylase activity on these mispairs (H.-J. Yang and J. H. Miller, unpublished results). On the other hand, several lines of evidence point to the combined effects of increased mispairs caused by the change in nucleotide triphosphate pools and the subsequent partial saturation of the dnaQ/mutD-encoded editing subunit (ε) of DNA polymerase.
Plasmid in CC107 ndk . | IPTG . | Mutation rate (μ) ×108 (95% confidence limits) . |
---|---|---|
pREP4/pQE60 | - | 47 (35-56) |
pREP4/pQE60 | + | 30 (24-75) |
pREP4/pQE60 ndk | - | 0 (0-5.6) |
pREP4/pQE60 ndk | + | 0 (0-5.9) |
pREP4/pQE60 nm23-H2 | - | 10 (6.7-19) |
pREP4/pQE60 nm23-H2 | + | 0 (0-2.3) |
Plasmid in CC107 ndk . | IPTG . | Mutation rate (μ) ×108 (95% confidence limits) . |
---|---|---|
pREP4/pQE60 | - | 47 (35-56) |
pREP4/pQE60 | + | 30 (24-75) |
pREP4/pQE60 ndk | - | 0 (0-5.6) |
pREP4/pQE60 ndk | + | 0 (0-5.9) |
pREP4/pQE60 nm23-H2 | - | 10 (6.7-19) |
pREP4/pQE60 nm23-H2 | + | 0 (0-2.3) |
Plasmid in CC107 ndk . | IPTG . | Mutation rate (μ) ×108 (95% confidence limits) . |
---|---|---|
pREP4/pQE60 | - | 47 (35-56) |
pREP4/pQE60 | + | 30 (24-75) |
pREP4/pQE60 ndk | - | 0 (0-5.6) |
pREP4/pQE60 ndk | + | 0 (0-5.9) |
pREP4/pQE60 nm23-H2 | - | 10 (6.7-19) |
pREP4/pQE60 nm23-H2 | + | 0 (0-2.3) |
Plasmid in CC107 ndk . | IPTG . | Mutation rate (μ) ×108 (95% confidence limits) . |
---|---|---|
pREP4/pQE60 | - | 47 (35-56) |
pREP4/pQE60 | + | 30 (24-75) |
pREP4/pQE60 ndk | - | 0 (0-5.6) |
pREP4/pQE60 ndk | + | 0 (0-5.9) |
pREP4/pQE60 nm23-H2 | - | 10 (6.7-19) |
pREP4/pQE60 nm23-H2 | + | 0 (0-2.3) |
Using a reversion system that detects specific base substitutions or frameshifts in the lacZ gene (Cupples and Miller 1989; Cupples et al. 1990), we found that in otherwise wild-type backgrounds, only the A:T → T:A and the A:T → G:C changes are significantly increased in ndk strains (see Table 2). In a MMR- (mutS) background, ndk strains display a much greater increase of A:T → G:C transitions. No such effect of ndk was found for G:C → A:T transitions in a mutS background. However, the lacZ system utilizes only one site for each of the six base substitutions. Because of the possibility of surrounding sequence effects on mutation rate, we decided to investigate a wider spectrum of base substitution mutations by sequencing a portion of the rpoB gene in Rifr mutants. This rpoB/Rifr system has been developed and improved by several groups (Jin and Gross 1988; Severinov et al. 1993 and references therein) and has been extended by the work reported here and elsewhere (Garibyan et al. 2002). This is a system that now offers a number of interesting advantages for mutational analysis. First, the mutations responsible for the Rifr phenotype are virtually all within a small segment of the rpoB gene that can be amplified and sequenced with two primer pairs. Second, the locus is on the chromosome, eliminating any variables in using mutational sites on an extrachromosomal element such as an F′ plasmid. Third, only mutants in the culture at the time of plating are scored, since Rifr colonies are not generated by events taking place after plating. The current extended system measures all six possible base substitutions distributed among 47 single-base substitution events at 29 sites (base pairs) with different sequence contexts (see Tables 3 and 4). The region of the rpoB-encoded β-subunit of RNA polymerase that contains the binding site for rifampicin is very highly conserved among several microorganisms, including pathogens such as certain mycobacteria (see review by Musser 1995). The Rifr mutations in M. tuberculosis and M. smegmatis are at sites equivalent to their counterparts in E. coli (Musser 1995; Karunakaran and Davies 2000).
The mutations detected by DNA sequencing of Rifr mutants (Table 3) show that whereas spontaneous mutations are distributed among many sites, 49 of 60 (82%) of the mutations seen in an ndk strain and 32/33 (97%) of the mutations in an ndk mutS strain are A:T G:C transitions. All of the mutations seen in the sequenced sample from the mutS strain are transitions, of which 46/57 (81%) are A:T → G:C, and 11/57 (19%) are G:C → A:T). Although G:C → A:T are found in the mutS spectrum, they are transitions absent from the ndk and the mutS ndk spectrum. This propensity for increased A:T → G:C transitions in ndk and mutS ndk strains is consistent with increased pool sizes of dCTP and dGTP found in ndk strains (Lu et al. 1995). In vitro studies have shown that when the concentrations of dGTP and dCTP were increased by 20-fold, the rate of stable misincorporation of dCMP and dGMP opposite adenine and thymine nucleotides increased (Roberts and Kunkel 1988). One particular hotspot site in rpoB (Table 3) represents 51% of all the mutations in the mutS, 77% of those in the ndk, and 67% of those in the mutS ndk spectrum. This and the second most prominent site in these spectra, also A:T → G:C, have a 5′ G as the nearest neighbor to the T that is changed to a C (Table 4). This same preference for a 5′ G nearest neighbor has been reported for lacI mutations in a mutS strain (Schaaper and Dunn 1987) and also for lacI mutations in mutD5 grown under conditions where the MMR system is saturated (Schaaper 1988). This latter result indicates that the preference for the 5′ G emanates from the polymerase errors themselves rather than being imprinted by the editing function or MMR.
The A:T → T:A substitutions seen in ndk strains (see Tables 2 and 3) would be difficult to explain by the 20-fold increase in dCTP pools or the 7-fold increase in dGTP pools reported by Inoue and co-workers for ndk strains (Lu et al. 1995). On the other hand, the A:T → T:A transversion is one of the signatures of a mutD strain, being the most prominent base change in the spectrum of lacI mutations in a mutD strain under certain conditions (Schaaper 1988). The experiment depicted in Table 6 shows that a mutD allele (mutD66) causing a partial mutator phenotype (Slupska et al. 1998) results in a mutator effect much greater in strains that are ndk than that seen in either ndk or mutD66 strains alone. This suggests that the defect in ndk strains may hinder proofreading. It is possible that an increase in the pool size of the next correct nucleotide might increase the rate of polymerization and thus inhibit proofreading (Mendelman et al. 1990; Roberts et al. 1991).
The finding that strains lacking NDP kinase are hypermutable in mismatch-repair backgrounds provides an example of synergistic effects of different alleles, in this case ndk and mutS. Given the involvement of both NDP kinase deficiency (see Lacombe et al. 2000 and Lombardi et al. 2000) and mismatch-repair deficiency (Fishel et al. 1993; Leach et al. 1993) in human cancers, we can ask the question of whether alleles giving weaker effects by themselves might still produce strong effects in the presence of certain other alleles or conditions. Thus, certain alleles of mutS might result in high mutation rates only in the presence of altered pool sizes caused by growth conditions or by genetic inactivation of NDP kinase. The challenge is to be able to detect mutations that result in cancer susceptibilities only in certain genetic backgrounds or only under certain conditions. One example is the finding that a single-nucleotide polymorphism in the RAD51 gene modifies cancer risk in BRCA2 carriers (Levy-Lahad et al. 2001). An example of how environmental challenges can unmask susceptibilities by hidden alleles is provided by the variation in benzo[a]pyrene diol epoxide susceptibilities in controls and lung carcinoma patients (Wu et al. 1998).
Acknowledgement
We thank Tomas Lindahl and Edith Postel for helpful discussions. J.H.M. was supported by a grant from the National Institutes of Health (ES-0110875).
Footnotes
Communicating editor: H. Ochman
LITERATURE CITED