Escherichia coliNucleoside Diphosphate Kinase Mutants Depend on Translesion DNA Synthesis To Prevent Mutagenesis

Abstract

Escherichia colinucleoside diphosphate (NDP) kinase mutants have an increased frequency of spontaneous mutation, possibly due to uracil misincorporation into DNA. Here we show that NDP kinase mutants are dependent on translesion DNA synthesis, often a mutagenic form of DNA synthesis, to prevent mutagenesis.

TEXT

The process of DNA replication requires the coordinated efforts of multiple proteins in order to maximize the accuracy of genome duplication. In Escherichia coli, replication forks are loaded at a defined origin of replication and travel bidirectionally around the circular chromosome and resolve at a defined terminus (14). However, DNA lesions that often occur throughout the chromosome can block fork progression. Many such lesions give rise to regions of single-stranded DNA (ssDNA) (2). The RecA protein, which catalyzes homologous recombination, binds to regions of ssDNA, forming a nucleoprotein filament (RecA-ssDNA) that facilitates the autoproteolytic cleavage of the LexA protein, a direct repressor of numerous genes involved in the repair/tolerance of DNA damage (SOS response). Two sets of genes that are part of the SOS regulon, umuDCand dinB, encode Y family DNA polymerases that are capable of bypassing multiple DNA lesions that would normally halt progression of polymerase III (Pol III), the replicative polymerase, a process known as translesion synthesis (TLS) (3, 10, 12, 21, 26). Normally, there are ∼17, 180, and 250 molecules per cell of UmuC, UmuD, and DinB, respectively. All of these concentrations increase ∼10- to 12-fold in SOS-induced cells (13, 28). UmuD forms a dimer, UmuD₂, that also undergoes RecA-ssDNA-dependent autoproteolytic cleavage, converting it into UmuD′₂, a form that is active for UmuC-dependent TLS (11, 23, 25).

By-products and/or intermediates generated during the synthesis of deoxynucleoside triphosphates (dNTPs) can lead to mutagenic DNA lesions. For example, the de novosynthesis of thymidylate requires the formation of dUTP (19). Incorporation of deoxyuridine (dU) in place of deoxyribosylthymine (dT) can be mutagenic, since dU can base pair with deoxyguanosine (dG). Furthermore, excision of uracil by the DNA glycosylase Ung results in the formation of an abasic site, which itself can be mutagenic (6). Loss of nucleoside diphosphate (NDP) kinase activity in E. coliresults in a modest imbalance in dNTP pool levels and an increase in mutation frequency (16, 22). NDP kinase, the product of the ndkgene, is capable of generating all nucleoside triphosphates from their corresponding nucleoside diphosphates with little or no substrate specificity (15). While it is unclear why loss of NDP kinase activity is mutagenic, it has recently been shown that loss of dUTPase activity in an ndkmutant background results in a synergistic increase in the frequency of mutagenesis (20). NDP kinase has also been shown to interact with Ung, thereby lowering the K_mof Ung for uracil-containing DNA (9). Taken together, these results suggest that mutagenesis associated with loss of NDP kinase function occurs through a defect in uracil metabolism. Here we demonstrate that SOS induction of Y family polymerases is necessary to prevent mutagenesis in an ndkmutant background.

We have demonstrated that expression of a human NDP kinase homologue, Nm23-H2, suppresses the mutagenic phenotype of an E. colindkmutant without correcting the dNTP pool imbalance, indicating that the dNTP pool imbalance is not the cause of increased mutation frequency (20). Additionally, we have shown that mutagenesis in an E. colindkmutant is likely due to uracil misincorporation into DNA, combined with a potential defect in the repair of uracil-containing DNA. We hypothesized that abasic sites, generated as an intermediate of the uracil excision process, could be the underlying cause of increased mutagenesis in an ndkmutant. Based on this hypothesis, we wanted to determine if recombinational repair and/or SOS induction has any effect on the mutagenesis associated with loss of ndkfunction. To this end, we created a recA ndkdouble mutant and compared its frequency of mutagenesis to that in wild-type and single mutant cells by assaying the frequency of spontaneous rifampin-resistant mutants. We found that loss of RecA activity in an ndkmutant background resulted in a dramatic increase in the already elevated frequency of spontaneous mutagenesis in an ndknull mutant, while loss of RecA activity had no effect on the mutation frequency in wild-type cells (Table 1). To begin to address which of the activities of RecA are necessary to prevent excessive mutagenesis in an ndkmutant background, we combined an ndkmutation with a lexA3mutation, which makes LexA refractory to RecA-mediated cleavage. Thus, in lexA3mutant cells, DNA damage does not induce the SOS response (18). lexA3 ndkdouble mutant cells displayed the same high level of spontaneous mutagenesis observed in the ndk recAdouble mutant, while a lexA3single mutation had no effect on the frequency of mutagenesis in ndk⁺cells (Table 1). There are multiple possibilities to explain the dependence of NDP kinase mutants on RecA and LexA activities to prevent mutagenesis. NDP kinase mutants could be dependent on SOS induction to prevent mutagenesis or, alternatively, loss of RecA activity or the ability of LexA to autoproteolyze could decrease the basal levels of SOS-induced genes. To distinguish between these two possibilities, we measured the level of SOS induction in an ndkmutant to determine if it was induced for the SOS response.

Table 1.

Frequencies of spontaneous mutation

ED8566 genotype	No. of Rifrcolonies/108CFUa	95% CIb	Fold increasec
Wild type	1.89	0.4–3	1
recA::cat	2.41	1.3–15	1.2
lexA3	1.32	0.6–3.5	0.7
umuDC595::cat	2.08	1.3–4.1	1.1
umuC773::kan	3.11	0.5–9.7	1.7
umuC122::Tn5	5.0	1.9–180	2.7
dinB::cat	0.47	0.3–2.3	0.3
Δndk	27.2	11.3–53.6	14.4
Δndk recA::cat	1,525	467–1,940	809
Δndk lexA3	3,118	952–3,482	1,653
Δndk umuDC595::cat	2,268	1,621–2,388	1,202
Δndk umuC773::kan	205	28.6–553	109
Δndk umuC122::Tn5	404	99–824	214
Δndk dinB::cat	227	78–274	120
Δndk umuC::Tn5/pNDK	0.87	0.4–1.1	1
Δndk umuC::Tn5/pBAD33	272	100–1,142	312
Δndk umuC::Tn5/pNm23-H2	4.7	3.6–28	5.4

^aValues represent medians of results from 8 independent cultures, each grown overnight, serially diluted, and plated onto LB-rifampin (Rif) plates for determination of total numbers of CFU. Mutation frequencies were determined by dividing the number of Rif^rcolonies by the total number of CFU.

^bCI, confidence interval.

^cFold increase relative to the level in the wild type.

To this end, we measured β-galactosidase activity in cells harboring plasmid pSE200, which contains a gene fusion between the SOS-inducible mucBgene and the lacZgene (4). Our results indicate that loss of NDP kinase function does not result in a significant increase in the SOS response, whereas control cells treated with a DNA-damaging agent were significantly induced (Fig. 1). These results suggest that basal-level expression of SOS-inducible genes is required to prevent excessive mutagenesis in an NDP kinase mutant.

Fig. 1.

ndkmutants are not induced for the SOS response. β-Galactosidase activities of wild-type (SQ15) and ndk(JN319) mutant cells harboring plasmid pSE200 were measured from 4 independent cultures grown to mid-exponential phase (optical density at 600 nm [OD₆₀₀], 0.2 to 0.3) in L broth at 37°C (4). For UV-treated control cells, 2 independent cultures of each strain were UV radiated with 50 J/m²UV in sterile buffer using a Stratagene UV Stratalinker 2400, diluted 1:100 in L broth, and grown for 3 h before being harvested (OD, ∼0.3). The graph represents the means, and error bars represent the standard errors of the means.

Given these results, we wanted to determine if a specific SOS-inducible gene was necessary for the increased mutagenesis associated with the lexA3 ndkdouble mutant. Based on the hypothesis that AP sites contribute to mutagenesis in an ndkmutant, we asked if UmuD′₂C, the SOS-inducible translesion polymerase that is known to bypass abasic sites, is involved in the increased frequency of mutation (26). First, we constructed a umuD umuC ndktriple mutant and measured its mutation frequency relative to that of control cells. To our surprise, we found that loss of UmuDC activity in an ndkmutant background resulted in the same high level of spontaneous mutation seen in either recA, ndk, or lexA3 ndkdouble mutant cells (Table 1). This finding is counterintuitive, given the fact that UmuD′₂C-dependent TLS is thought to be error prone since SOS-induced mutagenesis is dependent on UmuD′₂C activity (5). To further explore this finding, we measured the level of spontaneous mutation in umuC ndkdouble mutant cells, which still express UmuD. Although there was a significant level of mutagenesis seen in this double mutant, it was 10-fold reduced compared to that in the umuD umuC ndktriple mutant (Table 1). We also examined the effect of a mutated umuCallele, umuC122, resulting in a truncated carboxy terminus, which is defective for UV-induced mutagenesis but has been shown to promote cell survival upon dNTP depletion (3, 8). umuC122 ndkdouble mutant cells were as susceptible to mutagenesis as umuC ndkdouble mutant cells (Table 1). Furthermore, loss of DinB activity, another Y family DNA polymerase, in an ndkmutant background resulted in the same frequency of mutation seen in umuC ndkor umuC122 ndkdouble mutant cells (Table 1). All together, these results are consistent with the notion that ndkmutants depend on Y family DNA polymerases to keep the level of mutagenesis within certain limits. These results also suggest that UmuD has an antimutagenic effect in the umuC ndk, umuC122 ndk, and dinB ndkmutant backgrounds (Table 1). To our knowledge, this is the first time that UmuD′₂C-dependent TLS has been demonstrated to be antimutagenic in vivo, although DinB has been shown to be capable of accurately bypassing a N²-furfuryl-deoxyguanosine lesion (10). This is not entirely without precedent in eukaryotic cells. Patients with the disease xeroderma pigmentosum (XP) develop skin tumors in response to sunlight, due to a defect in nucleotide excision repair. A variant group of XP cells, XP-V, is proficient for excision repair but defective for TLS (1). XP-V cells are known to have an increased frequency of mutation in response to UV radiation and harbor mutations in the Y family polymerase (Pol η) (1, 17).

We have assumed that the increased levels of mutagenesis described in this work are due to specific DNA lesions, possibly abasic sites, generated in the absence of NDP kinase function. However, we have not ruled out the possibility that the altered dNTP pool levels associated with loss of NDP kinase activity are responsible for this phenotype. To address this issue, we transformed the ndkmutant strain with a plasmid expressing human Nm23-H2, which has been shown to complement the mutagenic phenotype of an ndkmutant without affecting the dNTP pool levels (20). We then introduced a umuCmutation into this strain by P1 transduction and compared its mutation frequency with that of isogenic strains expressing NDP kinase or containing an empty vector control. Expression of Nm23-H2 suppressed the increased frequency of mutation seen in the umuC ndkdouble mutant, indicating that dNTP pool imbalances are not responsible for this phenotype (Table 1).

How could the activities of UmuC and DinB be antimutagenic in an ndkmutant? In vitro, both UmuD′₂C and DinB have been shown to bypass specific DNA lesions (10, 26). UmuD′₂C has been shown to bypass abasic lesions, preferentially incorporating an A residue opposite the abasic site, a process referred to as the “A rule” (24, 27). If UmuD′₂C bypasses abasic lesions derived from uracil incorporation in place of thymidine, the outcome would be nonmutagenic. Abasic sites are known to block Pol III progression in vitro, but there may be a UmuD′₂C-independent bypass mechanism in vivothat does not preferentially incorporate A residues opposite abasic sites. The outcome of such a bypass mechanism would be mutagenic with respect to abasic sites derived from the excision of uracil and would explain the phenotypes seen here. DinB has not been demonstrated to bypass abasic sites in vitro(26). However, in light of numerous interactions that regulate the activity of DinB in vivo, it seems possible that DinB function could be modulated to tolerate numerous DNA lesions in vivo(7).