Translesion DNA polymerases are required for spontaneous deletion formation in Salmonella typhimurium Sanna Koskiniemi and Dan I. Andersson1 Department of Medical Biochemistry and Microbiology, Uppsala University, S-751 23 Uppsala, Sweden Communicated by John R. Roth, University of California, Davis, CA, April 22, 2009 (received for review April 19, 2008)
How spontaneous deletions form in bacteria is still a partly unresolved problem. Here, we show that deletion formation in Salmonella typhimurium requires the presence of functional translesion polymerases. First, in wild-type bacteria, removal of the known translesion DNA polymerases, PolII (polB), PolIV (dinB), PolV (umuDC), and SamAB (samAB), resulted in a 10-fold decrease in the deletion rate, indicating that 90% of all spontaneous deletions require these polymerases for their formation. Second, overexpression of these polymerases by derepression of the DNA damage-inducible LexA regulon caused a 25-fold increase in deletion rate that depended on the presence of functional translesion polymerases. Third, overexpression of the polymerases PolII and PolIV from a plasmid increased the deletion rate 12- to 30-fold, respectively. Last, in a recBCⴚ mutant where dsDNA ends are stabilized due to the lack of the end-processing nuclease RecBC, the deletion rate was increased 20-fold. This increase depended on the translesion polymerases. In lexA(def) mutant cells with constitutive SOS expression, a 10-fold increase in DNA breaks was observed. Inactivation of all 4 translesion polymerases in the lexA(def) mutant reduced the deletion rate 250-fold without any concomitant reduction in the amount of DNA breaks. Mutational inactivation of 3 endonucleases under LexA control reduced the number of DNA breaks to the wild-type level in a lexA(def) mutant with a concomitant 50-fold reduction in deletion rate. These findings suggest that the translesion polymerases are not involved in forming the DNA breaks, but that they require them to stimulate deletion formation. bacteria 兩 DNA homology 兩 gene loss 兩 RecA protein
S
pontaneous deletions are formed at relatively high frequencies in growing bacteria, but they rarely become fixed in the population because of their normally deleterious nature. The different mechanisms that are involved in deletion formation are well described with regard to deletions that form between long sequence homologies, whereas the mechanism for deletion formation involving little or no sequence homology is still quite unclear. Deletion formation can be divided into 3 classes depending on the mechanism of formation. First, deletions can be formed by RecA-dependent homologous recombination, which requires at least 20–40 bp of homology to be effective (1). However, several studies suggest that many spontaneous deletions are formed via RecA-independent pathways (2–4) where short (⬍25 bp) or no repeats are found at the deletion endpoints. Such recombination is often referred to as illegitimate recombination (5–8), and it can further be divided into 2 subgroups: short homology-dependent illegitimate recombination (SHDIR) and short homology-independent illegitimate recombination (SHIIR), where the former involves some short homology at the endpoint and the latter does not (9). SHDIR has been proposed to occur by 2 mechanisms, single strand annealing and slipped strand mispairing, explaining formation of small deletions in a replication fork, but failing to explain how large spontaneous deletions could be formed between distant sites with no or little homology (5, 10). Last, SHIIR, or true illegitimate recombina10248 –10253 兩 PNAS 兩 June 23, 2009 兩 vol. 106 兩 no. 25
tion, can occur between distant sites and requires no homology at all. One mechanism for SHIIR is nonhomologous end-joining (NHEJ), a widely used mechanism in the repair of broken chromosomes in eukaryotes, where 2 DNA ends are directly ligated together by a special DNA ligase (Ku ligase). Recently, this type of recombination has been found in certain bacteria harboring these special DNA ligases (7). In bacteria that do not posses the Ku ligase (e.g., Salmonella typhimurium and Escherichia coli), SHIIR has been suggested to occur via DNA gyrase-mediated illegitimate recombination (11, 12), where 2 separate DNA gyrase molecules covalently bound to DNA are involved in a reciprocal gyrase subunit exchange, resulting in DNA loss (12). From other studies, genes involved in the repair of damaged DNA and in replication have been shown to influence deletion formation (10, 13–17), and DNA breaks have been associated with increased deletion formation in several systems (18, 19). In a bacterial cell, DNA repair is a constantly ongoing process due to the generation of various DNA damages. DNA damages will induce a repair system in the cell, the SOS response (20), that is activated on RecA filamentation on ssDNA, and where filamented RecA mediates proteolytic cleavage of the LexA repressor protein (21). Induction of the LexA regulon by DNA damage changes expression of ⬇40 genes, most of which are up-regulated (20, 22). Among these genes, several endonucleases and the Y family of DNA polymerases can be found. Polymerases in the Y family are responsible for bypassing several different types of DNA lesions in a mutagenic translesion process. In S. typhimurium, 4 translesion DNA polymerases are known: PolII (polB), PolIV (dinB), PolV (umuDC), and the PolV homologue SamAB (samAB). Under noninduced conditions, these polymerases are present in low numbers, whereas during the SOS response, a 10- to 20-fold increase in levels is observed depending on the particular polymerase (23). All 4 polymerases have a lower processivity (typically adding on 300–400 nt before dissociating) (24) than the normal replicative DNA polymerase PolIII (dnaE) (23). With regard to the SOS response and deletion formation, previous results suggest that induction of the SOS response does not increase deletion formation between tandem repeats (13, 14). However, constitutive expression of the SOS regulon has been observed in several mutants that show increased deletion formation (13–16), possibly implying an involvement of the SOS system in deletion formation. Several approaches have been used to study deletion formation, most of them involving systems where artificial homology and/or deletion targets are present on a plasmid or on the Author contributions: S.K. and D.I.A. designed research; S.K. performed research; S.K. and D.I.A. analyzed data; and S.K. and D.I.A. wrote the paper. The authors declare no conflict of interest. 1To
whom correspondence should be addressed. E-mail:
[email protected].
This article contains supporting information online at www.pnas.org/cgi/content/full/ 0904389106/DCSupplemental.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0904389106
chromosome (8, 10, 25). Here, we studied spontaneous deletion formation in a specific region of the S. typhimurium chromosome. (Fig. 1; Materials and Methods) (4). In this system, deletion formation is easily detected and the deletable region lacking essential genes extends at least 200 kbp, making this deletable region a useful system to study spontaneous deletion formation on the chromosome without any severe restrictions on deletion size or the presence of certain recombination homologies (4). Also, because no direct repeats ⬎25 bp in length are present in this region, RecA-dependent homologous recombination is likely to be a minor contributor to deletion formation. Mutations in the normal replicative polymerase (PolIII) have been shown to increase deletion formation (13, 14). However, to our knowledge, the role of other polymerases in deletion formation has not been studied to any greater extent. One study has investigated the role of the translesion DNA polymerases Pol II, PolIV, and PolV on RecA-dependent deletion formation between long (787 bp) tandem repeats (26), but no effect on deletion formation was seen in wild-type background for any of the polymerases, even though removal of PolIV did reduce deletion formation moderately in a dnaBts mutant background. Here, we specifically determined how the translesion DNA polymerases PolII, PolIV, PolV, and SamAB influenced deletion formation when no artificial long DNA repeats are present. Results show that spontaneous deletion formation requires the presence of these DNA polymerases, and that their overproduction strongly stimulates deletion formation. Results Spontaneous Deletion Formation Requires Functional Translesion DNA Polymerases. We examined how the translesion polymerases
influenced deletion formation by mutationally inactivating the 3
Derepression of Translesion Polymerases Increases Deletion Formation. Because inactivation of the translesion polymerases de-
creased deletion formation in the wild type, we determined whether increased expression would stimulate deletion formation. These polymerases are under control of the LexA repressor, and a lexA(def) mutant with constitutive expression of the translesion polymerases was used to confer increased expression. As shown in Fig. 3, a lexA(def) mutation increased the deletion rate 25-fold. To determine whether the translesion polymerases were causing this induction, the polymerases PolII, PolIV, PolV, and SamAB were inactivated separately and combined. Individual inactivation of each one of the 4 polymerases reduced the relative increase in deletion rate from 25-fold to 5- to 10-fold, suggesting that all 4 polymerases contribute approximately equally to the increased deletion rate seen in the lexA(def) background (Fig. 3). Inactivation of combinations of 2 polymerases did not decrease deletion formation further, whereas combinations where any 3 polymerases were inactivated decreased the deletion rates to wild-type level (Fig. 3). Last, inactivation of all 4 polymerases in the lexA(def) mutant reduced the relative deletion rate to 0.1, approximately a 250-fold reduction as compared with the lexA(def) mutant (Fig. 3). Overexpression of Certain Individual Translesion Polymerases Causes an Increase in Deletion Formation. Because an increased deletion
rate was seen in a lexA(def) mutant expressing all of the translesion polymerases at increased levels (Fig. 3), we determined whether overexpression of each separate polymerase could stimulate deletion formation. Overproduction of PolII and PolIV from an arabinose-inducible plasmid (pBAD30) (27) increased the deletion rate 12- and 30-fold, respectively (Fig. 4). In contrast, overexpression of PolV and SamAB caused no increase in deletion rate (Table S1). Both UmuDC and SamAB contain the RecA cleavage site between Cys-24 and Gly-25 at the N terminus of UmuD and SamA, respectively, and cleavage and removal of the 24 N-terminal amino acids are necessary to
Fig. 2. Effect of translesion DNA polymerases on spontaneous deletion formation. Rates are given relative to the mean deletion rate of the wild type (9 ⫻ 10⫺10 ⫾ 0.1 ⫻ 10⫺10), calculated from 21 independent experiments. Error bars are SEM.
Koskiniemi and Andersson
PNAS 兩 June 23, 2009 兩 vol. 106 兩 no. 25 兩 10249
GENETICS
Fig. 1. Selection for spontaneous deletions. A loss-of-function mutation in the mod operon makes cells chlorate resistant under anaerobic growth conditions. A loss-of-function mutation in the gal operon makes colonies white on MacConkey galactose agar plates. The mod and gal operons are present on the S. typhimurium chromosome at the location indicated. Loss of the kanamycinresistance cassette inserted between the mod and gal operons was used to verify chlorate-resistant white colonies as deletions.
chromosomally encoded polymerases (PolII, PolIV, and PolV) and the PolV homologue SamAB, a fourth translesion polymerase that is encoded by the plasmid-borne (pSLT) samAB genes in S. typhimurium. Separately, or in combinations of 2, these mutations caused no significant decrease in deletion rate, whereas simultaneous inactivation of any 3 polymerases resulted in a 2- to 5-fold decrease in deletion rate (Fig. 2). Also, simultaneous inactivation of all 4 translesion polymerases resulted in a 10-fold decrease in deletion formation as compared with the wild type (Fig. 2), suggesting that 90% of the spontaneous deletions seen in wild-type cells require functional translesion DNA polymerases for their formation.
Fig. 3. Effect of LexA and the translesion DNA polymerases on spontaneous deletion formation. Rates are given relative to the mean deletion rate of the wild type (9 ⫻ 10⫺10 ⫾ 0.1 ⫻ 10⫺10). Error bars are SEM. The lexA(def) mutant (DA10598, lexA(def), sulA46::spc, cured of fels-2, gifsy-1, and gifsy-2), and the wild type (DA10212) are not isogenic, but deletion rates for the lexA(def) mutant are given relative to the wild type. As a control, deletion rates of an isogenic lexA⫹ strain (DA14668) sulA46:.spc, cured of fels-2, gifsy-1, and gifsy-2 were measured and found to be similar to the wild type.
convert the proteins to their active form (28). However, overproduction of the truncated polymerase subunits, UmuD⬘C and SamA⬘B, from the same pBAD30-based plasmid, resulted in no increase in deletion rate (Table S1). Similarly, overproduction of the truncated polymerase subunits, UmuD⬘C and SamA⬘B in mutant strains that had the chromosomal wild-type copies of the umuDC and samAB genes inactivated caused no increase in deletion formation (Table S1). One possible explanation for the lack of enhancement of mutagenesis is that expression from the plasmid pBAD30 is not high enough to allow deletion formation mediated by PolV and SamAB, whereas expression is sufficient for PolII and PolIV to increase the deletion rates. Control experiments demonstrated that arabinose induction of the pBAD30 plasmid with a nonfunctional PolIV, carrying an inactivating point mutation in the active site [dinB(D108G)], had no effect on deletion rates, demonstrating that neither the plasmid itself, nor overexpression of protein, nor the arabinose caused the increased deletion rates.
examined its role in formation of polymerase-induced deletions. First, when introduced into a wild-type genetic background a recA⫺ mutation caused a 5-fold decrease in deletion formation (Fig. 2). Second, when a recA null mutation was introduced into a lexA(def) mutant, derepression of the LexA regulon caused only a 6-fold increase in deletion rate as compared with the 25-fold increase seen in the recA⫹ lexA(def) single-mutant strain, indicating a partial requirement for RecA (Fig. 3). Third, when separately overproducing PolII and PolIV from the pBAD plasmid, a recA⫺ mutation had no effect on deletion formation. In the recA mutant, a 20- and 30-fold induction in deletion formation, respectively, could be observed, and this increase was similar to that seen in the wild type (12- and 30-fold increases, respectively) (Fig. 4). These results show that deletion formation caused by the translesion polymerases in the wild-type and lexA(def) mutant partly depended on RecA, whereas the deletions induced by plasmid-overproduction of PolII and PolIV were RecA-independent (see Discussion).
Role of RecA in Formation of Polymerase-Dependent Deletions. Be-
Stabilized dsDNA Ends Increase Deletion Formation, and This Increase Depends on the Presence of Functional Translesion Polymerases.
cause RecA is a key protein in homologous recombination, we
Because an increased deletion formation has been associated with mutants that accumulate DNA breaks (18, 19), we determined whether such an increase could be observed in our assay system. RecBC is the major ds-exonuclease in S. typhimurium and a key component of recombinational repair. This enzyme possesses a high affinity for dsDNA ends, and it rapidly degrades broken DNA. A recBC⫺ mutant is defective in processing dsDNA ends, which results in an increased number of DNA ends in the cells due to stabilization of the generated dsDNA breaks (29). In a recBC⫺ mutant, the deletion rate was increased 20-fold as compared with the wild type (Fig. 5A), and removal of all 4 translesion polymerases in the recBC⫺ background decreased the relative deletion rate to ⬇0.1, an ⬇200-fold reduction (Fig. 5B). Thus, the increased deletion rate seen in the recBC⫺ mutant completely depended on the translesion polymerases. Increased Levels of DNA Breaks Due to Derepression of the LexA Regulon Do Not Require Functional Translesion DNA Polymerases. The
Fig. 4. Effect of overexpression of PolII (Left) and PolIV (Right), from an arabinose-inducible plasmid on deletion formation in different genetic backgrounds. ⫹, Presence and induction of araBAD promoter with 0.01% arabinose; ⫺, control cells without plasmid, grown in the same medium and the same arabinose concentration. Rates are given relative to the mean deletion rate of the wild type (9 ⫻ 10⫺10 ⫾ 0.1 ⫻ 10⫺10). Mean deletion rate of recA⫺ and lexAIND are 2 ⫻ 10⫺10 ⫾ 0.01 ⫻ 10⫺10 and 6 ⫻ 10⫺10 ⫾ 0.15 ⫻ 10⫺10, respectively. Relative difference between arabinose and no arabinose for each background is indicated in the figure as fold change. Error bars are SEM. 10250 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0904389106
deletion-inducing effect of the translesion DNA polymerases could be explained in at least 2 ways: either they increase the amount of recombination substrate (i.e., DNA ends) or they are involved in later processing of the DNA ends. With regard to the first possibility, because of the low processivity of these polymerases, their overproduction could induce replication pausing and, as a result, DNA breaks that stimulate recombination (18, 25, 30–32). To examine this idea, we determined whether overproduction of the translesion polymerases caused an inKoskiniemi and Andersson
creased number of DNA breaks. Using the TUNEL assay, we measured breaks present in the chromosome of the different mutant strains. Because double-strand breaks (DSB) are rapidly degraded by the RecBC exonuclease in wild-type cells (25), this assay should mainly detect single-strand gaps, and to allow measurements of DSB, assays were also performed in the recBC⫺ mutant, where both types of breaks are detected. However, it is possible that dsDNA breaks are also detectable in the recBC⫹ background because of the better affinity of terminal transferases for dsDNA ends. For all of the TUNEL assays, wild-type cells treated with mitomycin C (MMC), a drug that creates both ss- and dsDNA breaks, was used as a positive control. With MMC, a 20-fold increase in break formation could be measured, representing the positive control for the increase in number of DNA breaks discernable with this method. For the constitutively induced lexA(def) mutant, the number of ssDNA gaps was increased 10-fold (Fig. 6), a finding compatible with the idea that the translesion polymerases stimulate deletion formation by inducing DNA breaks. However, if overproduction of the translesion polymerases caused the DNA breaks and the associated increase in deletion rates, one would expect that inactivation of the translesion polymerases in the lexA(def) background should also, concomitantly with the observed reduction in the deletion rate (Fig. 3), reduce the number of DNA breaks. No such reduction in the number of breaks was observed (Fig. 6), even though the deletion rate was reduced below wild-type levels (Fig. 3), implying that the polymerases do not stimulate deletion formation by inducing DNA breaks. Similarly, for the recBC⫺ mutant, dsDNA breaks increased 20-fold and inactivation of all 4 polymerases in the recBC⫺ mutant reduced the deletion rate ⬇200-fold, without any associated decrease in DNA breaks. These results imply, first, that neither ss nor ds breaks require the translesion polymerases for their formation (even though both types of breaks might act as a template for the polymerases during deletion formation), and second, that for DNA breaks to stimulate deletion formation, functional translesion polymerases are required. Finally, which functions in the LexA regulon might, when overproduced, cause the DNA breaks? One possibility is that the extensive ss breaks observed in the lexA(def) mutant were formed by endonucleases under LexA control (e.g., RuvC, Cho, MutY, Nfo, UvrABC, and Tag, or genes with unknown function; recN and dinI). This idea is compatible with the observation that certain endonucleases when present in high numbers may attack undamaged DNA (33), and as a consequence cause DNA breaks. Koskiniemi and Andersson
Fig. 6. DNA gaps/breaks in wild type and various mutants measured by using the TUNEL assay. The wild-type strain DA10212 treated with MMC was used as positive control. The data are presented as relative number of gaps/breaks as compared with the wild type (set to 1). Error bars are SEM.
Mutational inactivation of the above genes, alone or in combinations, revealed that inactivation of uvrB, uvrC, and cho in combination reduced the number of DNA breaks seen in a lexA(def) mutant to the wild-type level, with uvrB and uvrC conferring the largest effect (Fig. S1). The decrease in DNA breaks seen in the triple mutant was associated with a concomitant 50-fold reduction in deletion formation (Fig. 3). Thus, in the lexA(def), uvrB⫺, uvrC⫺, cho⫺ mutant, the translesion DNA polymerases are still expressed at a high level, but they cannot stimulate deletion formation when DNA break formation is prevented by inactivation of these endonucleases. Discussion The presented results show that 90% of the deletions seen in wild-type cells require functional translesion DNA polymerases to form (Fig. 2). Conversely, increased expression, via SOS induction or artificial overproduction of these polymerases, increased the rate of deletion formation ⬇10- to 30-fold (Figs. 3 and 4). Here, we provide direct evidence showing that deletion formation depends on the translesion polymerases. Previous reports suggest an involvement of the normal replicative PolIII in deletion formation between short and long tandem repeats (34, 35), and in illegitimate recombination (10). Also, an important role for the translesion polymerases, in particular PolIV, has been shown for frameshift mutations (36), and in deletion formation between long tandem repeats, but the mechanism of frameshift formation and RecA-dependent homologous recombination is different from that observed for large deletions with no or little homology at deletion endpoints (1, 37). In view of previous data, involvement of the LexA regulon and the translesion polymerases in deletion formation is unexpected, because results from specific model systems containing recombining tandem repeats indicate that induction of the SOS response itself does not increase deletion formation (13, 16, 38). One interpretation of these conflicting results is that deletions formed between tandem repeats are not mediated via the same mechanism as spontaneous deletions in a normal bacterial chromosome. As shown here, deletion rates in wild-type cells as well as in a lexA(def) mutant (constitutive SOS induction) and recBC⫺ mutant (increased dsDNA breaks) were reduced 10-, 250-, and 200-fold, respectively, when all 4 translesion polymerases were inactivated. However, the polymerases do not contribute to deletion formation in a simple additive manner, and all 4 polymerases need to be inactivated to achieve a substantial reduction in deletion rate (Figs. 2, 3, and 5 A and B). One PNAS 兩 June 23, 2009 兩 vol. 106 兩 no. 25 兩 10251
GENETICS
Fig. 5. Effect of RecBC (A) and the translesion polymerases on deletion formation (B). Rates are given relative to the mean deletion rate of the wild type (0.9 ⫻ 10⫺10 ⫾ 0.1 ⫻ 10⫺10). Error bars are SEM.
potential explanation for this observation is provided by the fact that these polymerases compete hierarchically (PolIII ⬎ PolI ⬎ PolII ⬎ PolIV ⫽ PolV) for binding to the B-clamp of the replication fork (39). Thus, if one of the polymerases early in the hierarchical chain is inactivated by mutation, another polymerase downstream in the chain might have better access to the replication fork; therefore, the overall effect on the deletion rate might be relatively small until all 4 polymerases are inactivated. Previous studies have shown that deletion rates only weakly depend on RecA, implying that spontaneous deletions often form via RecA-independent homologous recombination (2–4). In our experimental system, RecA influenced deletion formation, but the extent of the effect depended strongly on the genetic background. In an otherwise wild-type background, a recA⫺ mutation caused a 5-fold decrease in deletion formation (Fig. 2). Because RecA is required for SOS induction, and it has been shown that for logarithmically growing wild-type cells a fraction of the cells induce the LexA regulon to some degree (40), this decrease in deletion rate could be explained by an inability of the recA mutant to induce the SOS response. This hypothesis is also supported by the fact that a lexAIND mutant, incapable of inducing the LexA regulon, showed an ⬇2-fold decrease in deletion rate (Fig. 2), indicating that at least half of the deletions formed in the wild-type background during normal growth depend on spurious SOS induction (40). In a lexA(def) genetic background, absence of RecA caused a 4-fold reduction in deletion formation (Fig. 3). In the lexA(def) background, the SOS regulon is constitutively expressed, and should, therefore, be unaffected by the recA⫺ mutation. However, because RecA mediates proteolytic cleavage of the UmuD protein of PolV, resulting in activation of the UmuD2C complex (41), the lack of RecA could reduce the deletion rate by preventing the activation of PolV, which, as shown by our data, contributes to deletion formation (Fig. 3). Last, when overexpressing PolII and PolIV, absence of RecA had no effect on the deletion rate. Overall these data suggest that RecA is not necessarily required for translesion polymerase mediated deletion formation, but that it can have an indirect effect on the deletion rate due to its involvement in proteolytic activation of the UmuD2C complex and cleavage of the LexA protein to induce the SOS regulon. By which mechanism could the translesion polymerases promote deletion formation? One potential explanation is that, because of the low processivity of these polymerases, overproduction of the polymerases induces replication pausing (32), and as a result DNA breaks and an associated increase in recombination (25, 31). However, this explanation is less likely, because inactivation of all 4 translesion polymerases in a lexA(def) genetic background reduced deletion formation below wild-type rates without being associated with any measurable reduction in DNA break formation. However, it is formally possible that the translesion polymerases could generate a rare specific type of break that promotes deletion formation and that cannot be distinguished in the bulk measurement of breaks assessed by the TUNEL assay. Alternatively, based on recent work indicating a role of PolIV in DSB repair (R) (26, 42), one could hypothesize that deletions result from misaligned break repair complexes. However, because RecA is required for break repair (29), one would expect a dependence of RecA on deletion formation, which is not observed when PolII and PolIV are overexpressed. However, increased amounts of breaks seem to be associated with increased deletion rates, as shown by previous studies (18, 31), and our demonstration that in a lexA(def) mutant overexpressing the translesion polymerases, removal of 3 LexA-controlled endonucleases (uvrB, uvrC, and cho) causes both a reduction in deletion formation (Fig. 3) and DNA breaks (Fig. S1). Also, because our experiments show an important role for the translesion polymerases in deletion formation (Fig. 3), together, these 10252 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0904389106
results suggest that both DNA breaks and translesion DNA polymerases are required for increased deletion formation. Therefore, we hypothesize that the polymerases act on DNA ends generated by breaks, and allow replication from an invading misaligned ssDNA end in a RecA-independent manner. This model has some precedence, because structural data suggest that the translesion polymerases can replicate from misaligned primers (43). By such a mechanism, deletion formation between distant chromosomal sites with little or no homology could be promoted, explaining the poor homology at the endpoints of many spontaneously formed deletions (4). Materials and Methods Bacterial Strains, Genetic Methods, and Growth Conditions. The bacterial strains used in this study were derived from Salmonella enterica var typhimurium LT2 (designated S. typhimurium throughout this article) and are listed in Table S2. All gene transfers were made by P22 transduction (44). Bacteria were grown in standard Luria–Bertani broth (LB) or M9 minimal medium supplemented with 0.2% glucose or glycerol (45). When grown overnight, bacteria were incubated at 37 °C, and liquid cultures were shaken at 200 rpm. Antibiotic concentrations used were as follows: ampicillin (Amp) 100 mg/L, kanamycin (Kan) 50 mg/L, chloramphenicol (Cam) 20 mg/L, and rifampicin (Rif) 100 mg/L in both agar plates and culture media. PCR, Linear Transformation, and Mutant Construction. See SI Materials and Methods. All plasmids were prepared by using the EZNA plasmid mini kit (Omega Bio-Tek) and transferred between strains by using transformation. Deletions of the polymerase genes in S. typhimurium were constructed by linear transformation as described previously in strain DA6196 carrying the red system (46). Linear DNA was produced by PCR using primers F: 5⬘tgtaggctggagctgcttc-3⬘ and R: 5⬘-catatgaatatcctcctta-3⬘ for Cam and Kan cassettes and F: 5⬘-caccaaacaccccccaaaacc-3⬘ and R: 5⬘-cacacaaccacaccacaccac-3⬘ for the Rif cassette, with 40 bp of homologous DNA to insertion site flanking at the 5⬘ ends (Tables S3 and S4). All primers were from MWG-Biotech. As templates, plasmids pKD3 (Cam) and pKD4 (Kan) and chromosomal DNA from strain DA10194 (Rif) were used to create the respective antibioticresistance marker. DNA was subjected to gel electrophoresis, and bands of the appropriate size were cut out and purified with GFX-PCR purification kit (Amersham). Resistance markers from pKD3 and pKD4 inserted by linear transformation included FRT-recombination sites present on the template plasmids. Resistance markers were removed from strains by using plasmid pCP20 carrying the FLP-recombinase under thermal induction control (46).
mod–gal Assay. In the S. typhimurium chromosome, the mod– gal region can be used to study spontaneous deletion formation. In this region, 2 operons, mod and gal, are separated by 1 kbp of DNA. Loss of function mutation in any of the mod genes renders cells chlorate resistant because of inactivation of the molybdate transport pathway (47). Any inactivating mutation in the gal operon will make the cells unable to grow on galactose, which can be seen as white appearance on MacConkey agar plates containing galactose. To ensure that chlorate-resistant white colonies harbor deletions, we included a kanamycin-resistance gene between the 2 operons, that in the case of a deletion should be lost and the cells will become kanamycin sensitive (Fig. 1). To determine the rate of deletion formation in the mod– gal region, bacteria were grown in 9 independent 1-mL overnight cultures inoculated with 106 cells, in M9 minimal medium supplemented with 0.2% glucose or glycerol. Strains with arabinose-induced plasmid expression were grown in M9-glycerol to avoid inhibition of the araBAD promoter by glucose (for these strains, deletion rates of wild-type cells grown in the same media with the same arabinose concentration were used to calculate relative rates). For all other strains, M9-glucose was used; 900 L from each overnight culture was plated on MacConkey agar plates (Difco) supplemented with 0.2% galactose and 0.2% sodium chlorate. Plates were incubated at 37 °C for 24 h anaerobically and then for 6 h aerobically to select for chlorate-resistant colonies that carried mod mutations. These colonies were then scored for white appearance (gal mutations), and white colonies that were kanamycin susceptible were confirmed as deletions (48). For each independent culture, the number of chlorate-resistant, white, kanamycin-susceptible colonies was divided by the total number of cells plated and deletion rates were calculated with either the median or P 0 method (49). For strains containing the pBAD::dinB, pBAD30::polB, pBAD30::umuDC, and pBAD30::samAB plasmids, M9-glycerol supplemented with 0.01% arabinose was used to overexpress the PolII, PolIV,
Koskiniemi and Andersson
TUNEL Assay. To measure the amount of breaks in living cells, the TUNEL assay was used (51). Cells were grown in 1 mL of LB overnight at 37 °C and diluted 1:20 into 10 mL of fresh LB. After 1.5 h of growth at 37 °C, arabinose (0.01%) or 2 g/mL MMC (Fluka) was added to the cultures and cells were grown for another 1.5 h. Nontreated cultures were grown for an additional 1.5 h; 30 L of the cells was centrifuged, resuspended, and fixed with 4% paraformaldehyde (Fluka) for 30 min; 0.1% Triton X-100 (Sigma) in 0.1 M sodium citrate (Merck) was used to permeabilize the cells for 2 min on ice. The TUNEL reagents were diluted 1:9 as described by the manufacturer (Roche), and enzyme, labeling agents, and cells were incubated in the TUNEL mix for 1 h at
1. Lovett ST, Hurley RL, Sutera VA, Jr, Aubuchon RH, Lebedeva MA (2002) Crossing over between regions of limited homology in Escherichia coli. RecA-dependent and RecAindependent pathways. Genetics 160:851– 859. 2. Jones IM, Primrose SB, Ehrlich SD (1982) Recombination between short direct repeats in a recA host. Mol Gen Genet 188:486 – 489. 3. Albertini AM, Hofer M, Calos MP, Miller JH (1982) On the formation of spontaneous deletions: The importance of short sequence homologies in the generation of large deletions. Cell 29:319 –328. 4. Nilsson AI, et al. (2005) Bacterial genome size reduction by experimental evolution. Proc Natl Acad Sci USA 102:12112–12116. 5. Feschenko VV, Lovett ST (1998) Slipped misalignment mechanisms of deletion formation: Analysis of deletion endpoints. J Mol Biol 276:559 –569. 6. Ikeda H, Moriya K, Matsumoto T (1981) In vitro study of illegitimate recombination: Involvement of DNA gyrase. Cold Spring Harbor Symp Quant Biol 45:399 – 408. 7. Wilkinson A, Day J, Bowater R (2001) Bacterial DNA ligases. Mol Microbiol 40:1241– 1248. 8. Albertini AM, Hofer M, Calos MP, Tlsty TD, Miller JH (1983) Analysis of spontaneous deletions and gene amplification in the lac region of Escherichia coli. Cold Spring Harbor Symp Quant Biol 47:841– 850. 9. Shimizu H, et al. (1997) Short-homology-independent illegitimate recombination in Escherichia coli: Distinct mechanism from short-homology-dependent illegitimate recombination. J Mol Biol 266:297–305. 10. Dutra BE, Sutera VA, Jr, Lovett ST (2007) RecA-independent recombination is efficient but limited by exonucleases. Proc Natl Acad Sci USA 104:216 –221. 11. Ikeda H, Kawasaki I, Gellert M (1984) Mechanism of illegitimate recombination: Common sites for recombination and cleavage mediated by E. coli DNA gyrase. Mol Gen Genet 196:546 –549. 12. Ikeda H, Shiraishi K, Ogata Y (2004) Illegitimate recombination mediated by doublestrand break and end-joining in Escherichia coli. Adv Biophys 38:3–20. 13. Saveson CJ, Lovett ST (1997) Enhanced deletion formation by aberrant DNA replication in Escherichia coli. Genetics 146:457– 470. 14. Bierne H, Seigneur M, Ehrlich SD, Michel B (1997) uvrD mutations enhance tandem repeat deletion in the Escherichia coli chromosome via SOS induction of the RecF recombination pathway. Mol Microbiol 26:557–567. 15. Bierne H, Vilette D, Ehrlich SD, Michel B (1997) Isolation of a dnaE mutation which enhances RecA-independent homologous recombination in the Escherichia coli chromosome. Mol Microbiol 24:1225–1234. 16. Bruand C, Bidnenko V, Ehrlich SD (2001) Replication mutations differentially enhance RecA-dependent and RecA-independent recombination between tandem repeats in Bacillus subtilis. Mol Microbiol 39:1248 –1258. 17. Lovett ST, Drapkin PT, Sutera VA, Jr, Gluckman-Peskind TJ (1993) A sister-strand exchange mechanism for recA-independent deletion of repeated DNA sequences in Escherichia coli. Genetics 135:631– 642. 18. Bierne H, Ehrlich SD, Michel B (1997) Deletions at stalled replication forks occur by two different pathways. EMBO J 16:3332–3340. 19. Yang Y, Masker W (1997) Double-strand breaks increase the incidence of genetic deletion associated with intermolecular recombination in bacteriophage T7. Mol Gen Genet 255:277–284. 20. Fernandez De Henestrosa AR, et al. (2000) Identification of additional genes belonging to the LexA regulon in Escherichia coli. Mol Microbiol 35:1560 –1572. 21. Moreau PL (1985) Role of Escherichia coli RecA protein in SOS induction and postreplication repair. Biochimie 67:353–356. 22. Quillardet P, Rouffaud MA, Bouige P (2003) DNA array analysis of gene expression in response to UV irradiation in Escherichia coli. Res Microbiol 154:559 –572. 23. Delmas S, Matic I (2006) Interplay between replication and recombination in Escherichia coli: Impact of the alternative DNA polymerases. Proc Natl Acad Sci USA 103:4564 – 4569. 24. Wagner J, Fujii S, Gruz P, Nohmi T, Fuchs RP (2000) The beta clamp targets DNA polymerase IV to DNA and strongly increases its processivity. EMBO Rep 1:484 – 488. 25. Michel B, Ehrlich SD, Uzest M (1997) DNA double-strand breaks caused by replication arrest. EMBO J 16:430 – 438. 26. Lovett ST (2006) Replication arrest-stimulated recombination: Dependence on the RecA paralog, RadA/Sms and translesion polymerase, DinB. DNA Repair 5:1421–1427.
Koskiniemi and Andersson
37 °C in the dark. Nonlabeled, red [propidium iodide (PI)], green (TUNEL), and red-green (PI ⫹ TUNEL) controls were used. For all controls, wild-type (DA10212) cells were stained with: PI (Fluka), TUNEL, or PI and TUNEL. MMCtreated cells were used as positive controls. PI was used at a concentration of 0.5 g/L in 1⫻ PBS. After the TUNEL reaction, cells were spun down and stained with PI for 10 min on ice in the dark (51). Samples were studied with FacsAria (BD). For every sample, at least 100,000 counts were measured, and grids were set to contain viable single cells, and for green counts no more than 1% of wild-type (DA10212) cells were allowed to pass as green. ACKNOWLEDGMENTS. We thank Diarmaid Hughes, Linus Sandegren, and Santanu Dasgupta for comments on the manuscript. This work was supported by grants from the Swedish Research Council and Uppsala University (D.I.A.).
27. Boudsocq F, Campbell M, Devoret R, Bailone A (1997) Quantitation of the inhibition of Hfr x F- recombination by the mutagenesis complex UmuD⬘C. J Mol Biol 270:201–211. 28. Nohmi T, Battista JR, Dodson LA, Walker GC (1988) RecA-mediated cleavage activates UmuD for mutagenesis: Mechanistic relationship between transcriptional derepression and posttranslational activation. Proc Natl Acad Sci USA 85:1816 –1820. 29. Kowalczykowski SC, Dixon DA, Eggleston AK, Lauder SD, Rehrauer WM (1994) Biochemistry of homologous recombination in Escherichia coli. Microbiol Rev 58:401– 465. 30. Kusano K, Asami Y, Fujita A, Tanokura M, Kobayashi I (2003) Type I restriction enzyme with RecA protein promotes illegitimate recombination. Plasmid 50:202–212. 31. Kong D, Masker W (1994) Deletion between direct repeats in T7 DNA stimulated by double-strand breaks. J Bacteriol 176:5904 –5911. 32. Lovett ST (2007) Polymerase switching in DNA replication. Mol Cell 27:523–526. 33. Branum ME, Reardon JT, Sancar A (2001) DNA repair excision nuclease attacks undamaged DNA. A potential source of spontaneous mutations. J Biol Chem 276:25421– 25426. 34. Trinh TQ, Sinden RR (1993) The influence of primary and secondary DNA structure in deletion and duplication between direct repeats in Escherichia coli. Genetics 134:409 – 422. 35. Lovett ST, Feschenko VV (1996) Stabilization of diverged tandem repeats by mismatch repair: Evidence for deletion formation via a misaligned replication intermediate. Proc Natl Acad Sci USA 93:7120 –7124. 36. Kim SR, et al. (1997) Multiple pathways for SOS-induced mutagenesis in Escherichia coli: An overexpression of dinB/dinP results in strongly enhancing mutagenesis in the absence of any exogenous treatment to damage DNA. Proc Natl Acad Sci USA 94:13792–13797. 37. Tippin B, Pham P, Goodman MF (2004) Error-prone replication for better or worse. Trends Microbiol 12:288 –295. 38. Shanado Y, Kato J, Ikeda H (1998) Escherichia coli HU protein suppresses DNA-gyrasemediated illegitimate recombination and SOS induction. Genes Cells 3:511–520. 39. Matsui K, Yamada M, Imai M, Yamamoto K, Nohmi T (2006) Specificity of replicative and SOS-inducible DNA polymerases in frameshift mutagenesis: Mutability of Salmonella typhimurium strains overexpressing SOS-inducible DNA polymerases to 30 chemical mutagens. DNA Repair 5:465– 478. 40. Pennington JM, Rosenberg SM (2007) Spontaneous DNA breakage in single living Escherichia coli cells. Nat Genet 39:797– 802. 41. Ferentz AE, Walker GC, Wagner G (2001) Converting a DNA damage checkpoint effector (UmuD2C) into a lesion bypass polymerase (UmuD⬘2C). EMBO J 20:4287– 4298. 42. Ponder RG, Fonville NC, Rosenberg SM (2005) A switch from high-fidelity to errorprone DNA double-strand break repair underlies stress-induced mutation. Mol Cell 19:791– 804. 43. Ling H, Boudsocq F, Woodgate R, Yang W (2004) Snapshots of replication through an abasic lesion; structural basis for base substitutions and frameshifts. Mol Cell 13:751– 762. 44. Maloy SR, Cronan JE, Jr, Freifelder D (1994) Microbial Genetics (Jones and Bartlett, Sudbury, MA), 2nd Ed. 45. Miller JH (1992) A Short Course in Bacterial Genetics (Cold Spring Harbor Lab Press, Plainview, NY). 46. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97:6640 – 6645. 47. Maupin-Furlow JA, et al. (1995) Genetic analysis of the modABCD (molybdate transport) operon of Escherichia coli. J Bacteriol 177:4851– 4856. 48. Lejeune P, Danchin A (1990) Mutations in the bglY gene increase the frequency of spontaneous deletions in Escherichia coli K-12. Proc Natl Acad Sci USA 87:360 –363. 49. Lea DE, Coulson CA (1949) The distribution of the numbers of mutants in bacterial populations. J Genet 49:264 –285. 50. Maisnier-Patin S, et al. (2005) Genomic buffering mitigates the effects of deleterious mutations in bacteria. Nat Genet 37:1376 –1379. 51. Rohwer F, Azam F (2000) Detection of DNA damage in prokaryotes by terminal deoxyribonucleotide transferase-mediated dUTP nick end labeling. Appl Environ Microbiol 66:1001–1006.
PNAS 兩 June 23, 2009 兩 vol. 106 兩 no. 25 兩 10253
GENETICS
PolV, and SamAB proteins, respectively (50). All measured deletion rates are listed in Table S1.
