Involvement of recF, recO, and recR Genes in UV-Radiation Mutagenesis of Escherichia coli

Abstract

The recF, recO, and recR genes were originally identified as those affecting the RecF pathway of recombination in Escherichia coli cells. Several lines of evidence suggest that the recF, recO, and recR genes function at the same step of recombination and postreplication repair. In this work, we report that null mutations in recF, recO, or recR greatly reduce UV-radiation mutagenesis (UVM) in an assay for reversion from a Trp⁻ (trpE65) to a Trp⁺ phenotypes. Introduction of the defective lexA51 mutation [lexA51(Def)] and/or UmuD′ into recF, recO, and recR mutants failed to restore normal UVM in the mutants. On the other hand, the presence of recA2020, a suppressor mutation for recF, recO, and recR mutations, restored normal UVM in recF, recO, and recR mutants. These results indicate an involvement of the recF, recO, and recR genes and their products in UVM, possibly by affecting the third role of RecA in UVM.

Mutagenesis by UV irradiation is not a passive process but rather requires the active participation of cellular proteins other than those in the replication complex. In Escherichia coli, mutagenesis by UV irradiation and many carcinogens requires the induction of SOS regulons (18, 43). This category of mutagenesis has been frequently referred to as SOS mutagenesis.

SOS mutagenesis in E. coli has been shown to be dependent on four genes, lexA, umuD, umuC, and recA (reviewed in reference 10). The LexA protein, encoded by lexA, is the repressor for SOS regulons (17, 18). It inhibits transcription by binding to operator sequences, called SOS boxes, located upstream of SOS genes (18). The UmuD and UmuC proteins are encoded by the umuD and umuC genes, which are organized in an operon under the regulation of lexA (35). In wild-type cells, UmuDC proteins are expressed at a low basal concentration. Among the many SOS-regulated genes, only the umuDC operon must be induced for SOS mutagenesis (36). While induction of the umuDC operon is an essential process in SOS mutagenesis, SOS mutagenesis also requires that UmuD be cleaved to the active UmuD′ form (1, 29, 34). Two UmuD′ molecules combine with one UmuC molecule to form a complex named UmuD′C, which is required for SOS mutagenesis (3, 50). RecA appears to play at least two roles in SOS mutagenesis; one is its regulatory role. In response to an SOS-inducing treatment, such as UV irradiation, RecA becomes activated and mediates or facilitates the proteolytic cleavage of LexA at an Ala-Gly bond, thus inducing the expression of umuDC and other SOS-regulated genes (17). A second role of activated RecA in SOS mutagenesis is promoting the proteolytic cleavage of UmuD at an Ala-Gly bond, producing two protein fragments of which the larger, COOH-terminal fragment (UmuD′) is required for SOS mutagenesis (29, 34). RecA has a third essential role in SOS mutagenesis (1, 7, 37): allowing translesion DNA replication, possibly by complexing with UmuD′C (9). The mechanism by which RecA performs its third function in SOS mutagenesis is presently unknown.

The recF (14), recO (16), and recR (27, 28) genes were originally identified as those affecting the RecF pathway of recombination in E. coli. Mutations in recF, recO, or recR conferred recombination-deficient and extremely UV-sensitive phenotypes in both recB recC sbcA and recB recC sbcB sbcC genetic backgrounds (14, 16, 27, 28). In the recBC⁺ sbcBC⁺ genetic background, mutations in recF, recO, or recR produced deficiencies in plasmid recombination and in the repair of DNA daughter-strand gaps and increased the cells’ sensitivity to UV irradiation, but the mutations did not appear to reduce conjugational or transductional recombination (6, 16, 19, 20, 24, 39). The deficiency caused by recF, recO, and recR mutations can be partially suppressed by a common suppressor mutation for recF, recA(Srf) (4, 42, 46). Accumulating genetic evidence suggests that the recF, recO, and recR gene products function at the same step of recombination and postreplication repair (4, 19, 33, 39, 46). The biochemical properties of purified RecF, RecO, and RecR have been studied. RecF binds to single-stranded DNA (ssDNA) and, in the presence of ATP or ATP-γS, double-stranded DNA (dsDNA) (11, 25, 26) but does not appear to have any positive effect on RecA-catalyzed reactions in vitro (25, 41). RecR has not been shown to exhibit any biochemical activity by itself. An interaction of RecR with RecO in the absence of any DNA or nucleotide cofactor (40) and an ATP- and dsDNA-dependent interaction of RecF and RecR (47) have been reported. RecO binds to both ssDNA and dsDNA and can promote renaturation of complementary ssDNA (23) and homologous pairing of ssDNA and superhelical dsDNA (22). RecO interacts with RecR and Ssb (40). This interaction is thought to help RecA adhere to Ssb-coated ssDNA, overcoming the Ssb inhibition of joint molecule formation in vitro (40, 41). More recently, RecF has been shown to interact with RecO, and the existence of a RecF-RecO-RecR complex in vitro has been demonstrated (13). The biochemical roles of the RecF-RecO-RecR complex in DNA repair and recombination, however, remain an open question.

The involvement of the recF, recO, and recR genes in SOS induction and in UV-radiation mutagenesis (UVM) has been investigated. Mutations in the recF, recO, and recR genes delay the induction of several SOS-regulated genes (12, 38, 48). Mutations in recF decrease UVM of ssDNA phages but have no effect on reversion of chromosomal hisG4 mutations (5, 15). The recR mutants are normal with regard to UV-induced reversion of hisG4 and argE3 mutations (27). To the best of our knowledge, there has been no report on the involvement of the recO gene in UVM. We initiated a study to investigate the effect of a recO mutation on UV-induced reversion of trpE65, and much to our surprise, we observed that recO mutants are grossly deficient in UVM. In view of the fact that recF, recO, and recR mutants have common phenotypes and that the recF, recO, and recR gene products function at the same step of recombination and postreplication repair, this observation raises the question of whether recF and recR mutants are deficient in the UV-induced reversion of trpE65. In this work, the possible involvement of recF, recO, recR, and other RecF pathway recombination genes in UVM was investigated.

MATERIALS AND METHODS

Bacterial strains and media.

The bacterial strains used are listed in Table 1. The construction of these strains by transduction has been described (39). All of the strains used in this study are isogenic; i.e., they differ only in the genes being investigated. The plasmid pGW2122, which produces truncated UmuD′ (29), was kindly provided by G. Walker. The supplemented minimal medium (SMM) and DTM buffer have been described (45). The complex media used were Luria broth (1% tryptone, 0.5% yeast extract, and 1% NaCl) and YENB (0.75% yeast extract and 0.8% nutrient broth). Reversion of from the Trp⁻ phenotype to the Trp⁺ phenotype was assayed in SMM containing 200 μg of nutrient broth per ml. Media were solidified by adding Difco Bacto Agar at 1.5%. Phosphate buffer contained Na₂HPO₄ at 5.83 g/liter and KH₂PO₄ at 3.53 g/liter and had a pH of 7.0.

TABLE 1.

E. coli strains used^a

Parent strain (relevant genotype)b	Strain with additional mutationc
	recF332	recO1504	recR252	recG258	recJ284	recQ63
EWRP-1A (recA+)	VW48	VW40	VW32	VW101	VW75	VW150
EWRP-1 (recA+lexA51)	VW16	VW174	VW175
VW56 (recA2020)	VW57	VW54	VW55

^aAll strains are also uvrA155 trpE65 lon-11 sulA1. 

^bParent strains, except VW56, were obtained from E. Witkin (36). Strain VW56 was constructed by P1 transduction of recA2020 srlA300::Tn10 (44) into strain EWRP-1A. 

^cStrains were constructed by P1 transduction as previously described (38). 

UV irradiation.

The source (254 nm) and measurement of the fluence rate of UV irradiation have been described (44). Survival curves were determined and assays for mutagenesis were carried out as previously described (31).

Quantitation of mutagenesis.

The UV radiation-induced mutant frequency was calculated per average mutant selection plate according to a rearranged version of the formula of Bridges (2): MF = (M_x × 10⁸)/(S_c × volume plated). M_x is defined as M_t − M_po + M₀ (1-SF), where M_t is the number of mutant colonies arising from irradiated cells on mutant selection plates, M_po is the number of mutant colonies arising from nonirradiated cells on mutant-selection plates, M₀ is the number of mutant colonies arising from nonirradiated cells on plates lacking the growth-limiting nutrient, SF is the surviving fraction of irradiated cells, M_x is the number of radiation-induced mutants per mutant selection plate, and S_c is the number of CFU per milliliter in the cell suspension. Data were generally compiled from three experiments per UV radiation fluence, with four mutant selection plates and three viability plates per fluence.

RESULTS

Effects of recF, recG, recJ, recO, recQ, and recR mutations on UV-induced reversion from the Trp⁻ phenotype to the Trp⁺ phenotype.

The involvement of the recF, recO, and recR genes in UVM was examined by an assay for reversion from the Trp⁻ phenotype (trpE65) to the Trp⁺ phenotype. For comparison, we included three other RecF pathway genes, recG, recJ, and recQ, in the study. Similar levels of UV induction of Trp⁺ reversion were detected in recJ, recG, and recQ mutants and the rec⁺ control, indicating that mutations in recJ, recG, or recQ have no effect on UVM (Fig. 1 and data not shown). On the other hand, the recF, recO, and recR mutants exhibited altered levels of UVM. As shown in Fig. 1, the levels of UV-induced Trp⁺ reversion in these three mutants are comparable to those of rec⁺ organisms at low fluences of UV radiation, e.g., 0.4 J/m² or lower. At higher fluences of UV radiation, however, the number of UV-induced revertants is greatly reduced.

FIG. 1.

UVM of E. coli uvrA155 strains to a Trp⁺ phenotype. The reversion from a Trp⁻ phenotype to a Trp⁺ phenotype was assayed as described in Materials and Methods. At 1.2 J/m², UVM was not detected for strain VW32, VW40, or VW48. A value of 50 Trp⁺ mutants per 10⁸ survivors is indicated with an arrow. Symbols: ▿, EWRP-1A (rec⁺ uvrA155); □, VW32 (uvrA155 recR252); ○, VW40 (uvrA155 recO1504); ▵, VW48 (uvrA155 recF332); ×, VW101 (uvrA155 recG258). The data for strains VW75 (uvrA155 recJ284) and VW150 (uvrA155 recQ63) (not shown) are essentially congruent with those for strains EWRP-1A and VW101.

Effects of the lexA51(Def) mutation and UmuD′ on UV mutagenesis of recF, recO, and recR mutants.

From what is known about SOS mutagenesis, a deficiency in UVM may be caused by a deficiency in the induction of SOS regulons, in the cleavage of UmuD to active UmuD′, or in RecA’s third function in UVM. To determine if the observed deficiency in UVM of recF, recO, and recR mutants may be caused by a deficiency in the induction of SOS regulons, we examined the effect of lexA51(Def) (the lexA51 mutation which produces defective LexA and allows cells to express SOS regulons constitutively) on UVM of recF, recO, and recR mutants. The presence of the lexA51 mutation has no effect on UVM of recA⁺ cells (compare rec⁺ in Fig. 1 with lexA51 in Fig. 2) and fails to restore normal UVM in recF, recO, and recR mutants (Fig. 2), indicating that the deficiency in UVM of recF, recO, and recR mutants is not due to a failure to express SOS regulons. To examine if the observed deficiency in UVM of recF, recO, and recR mutants may be caused by a deficiency in the proteolytic cleavage of UmuD, we transformed a recombinant plasmid, pGW2122, into recF, recO, recR, lexA51 recF, lexA51 recO, and lexA51 recR strains. The presence of pGW2122 failed to restore normal UVM in recF, recO, and recR strains (data not shown) and in lexA51 recF, lexA51 recO, and lexA51 recR strains (Fig. 3). Therefore, deregulation of SOS regulons and supply of UmuD′ are insufficient to restore normal UVM in recF, recO, and recR mutants.

FIG. 2.

UVM of E. coli uvrA155 lexA51 strains to a Trp⁺ phenotype. Experimental details were as described in Materials and Methods. UVM was not detected for strain VW174 or VW175 at 1.2 J/m². A value of 50 Trp⁺ mutants per 10⁸ survivors is indicated with an arrow. Symbols: ▿, EWRP-1 (uvrA155 lexA51); □, VW174 (uvrA155 lexA51 recR252); ○, VW175 (uvrA155 lexA51 recO1504); ▵, VW16 (uvrA155 lexA51 recF332).

FIG. 3.

Effect of UmuD′ on UVM of E. coli uvrA155 lexA51 strains to a Trp⁺ phenotype. The plasmid pGW2122 was transformed into strains EWRP-1, VW16, VW174, and VW175, and the reversion from a Trp⁻ phenotype to a Trp⁺ phenotype was assayed as described in Materials and Methods. UVM was not detected for strain VW16/pGW2122, VW174/pGW2122, or VW175/pGW2122 at 1.2 J/m². A value of 50 Trp⁺ mutants per 10⁸ survivors is indicated with an arrow. Symbols: ▿, EWRP-1/pGW2122; □, VW174/pGW2122; ○, VW175/pGW2122; ▵, VW16/pGW2122.

Effects of recA(Srf) mutations on UVM of recF, recO, and recR mutants.

In addition to participating in the regulation of lexA regulons and in the cleavage of UmuD, RecA has a third essential role in SOS mutagenesis (1, 7, 37). It is possible that the deficiency in UVM observed in recF, recO, and recR mutants may be related to the third essential role of RecA in SOS mutagenesis. The postreplication repair deficiency in recF, recO, and recR mutants has been previously shown to be partially suppressed by recA(Srf) mutations, such as recA2020 (39, 45). To determine if the recA(Srf) mutations may suppress the deficiency of recF, recO, and recR mutants in UVM, we examined the effect of the recA2020 mutation on UVM of these mutants. Interestingly, we observed that the presence of the recA2020 mutation restored normal UVM in these mutants (Fig. 4).

FIG. 4.

UVM of E. coli uvrA155 recA2020 strains to a Trp⁺ phenotype. Experimental details were as described in Materials and Methods. Symbols: ▿, VW56 (uvrA155 recA2020); □, VW55 (uvrA155 recA2020 recR252); ○, VW54 (uvrA155 recA2020 recO1504); ▵, VW57 (uvrA155 recA2020 recF332).

DISCUSSION

In this work, we show that among the several RecF pathway genes investigated in this study, only mutations in the recF, recO, or recR gene produced a deficiency in UV-induced reversion of tryE65. The deficiency in UVM caused by recF, recO, and recR mutations appears to occur at higher fluences of UV radiation (Fig. 1). Derepression of the SOS regulon by lexA51(Def) and supply of UmuD′ failed to restore normal UVM in recF, recO, and recR mutants (Fig. 2 and 3). Interestingly, the presence of the recA2020 mutation restored normal UVM in recF, recO, and recR mutants (Fig. 4). These results suggest that the recF, recO, and recR genes do not act by affecting RecA’s functions in the induction of SOS regulons or in the cleavage of UmuD but may affect the third role of RecA in UVM. Several speculations have been made on the activities of RecA involved in its third role in UVM. These activities include inhibition of the 3′-to-5′ proofreading exonuclease of DNA polymerase III (21), binding to small, single-stranded regions of DNA (37), and directing the UmuD′ and UmuC proteins to the site of the lesion in DNA (1, 37, 49). Recent studies on the biochemical functions of the RecF, RecO, and RecR proteins suggest that RecF-RecO-RecR may help RecA displace Ssb at a DNA daughter-strand gap, which was produced after replication of a UV-damaged DNA template and, presumably, a DNA lesion for postreplication repair and SOS repair (13, 40). The multiprotein RecA-RecF-RecO-RecR-Ssb bound at the gapped DNA (13, 40) may form a presynaptic filament which initiates recombination repair. Alternatively, the multiproteins bound at the gapped DNA may help RecA complex with UmuD′C and allow translesion DNA synthesis to take place. According to this postulate, the role of RecF-RecO-RecR in UVM may be (i) to chaperone RecA to gapped DNA bound by Ssb so that RecA can perform its third function in UVM or (ii) to moderate the interaction of RecA with UmuD′C. Recent findings of a major role of RecA in stabilizing Umu proteins in vivo (8) suggest the alternative possibility that RecF-RecO-RecR may affect RecA’s ability to stabilize Umu proteins.

Our results indicating an involvement of the recF, recO, and recR genes in UVM are inconsistent with the previous observations that recF and recR mutants are normal in regard to UVM (15, 27). This discrepancy may be attributed to the differences in the reversion assays and/or the genetic backgrounds of parent strains used. The hisG4 and argE3 mutations in K-12 strain AB1157’s genetic background were employed in the previous reversion studies, while the trpE65 mutation in the EWRP-1A (a K-12 and B hybrid) genetic background was used in the present study. All of the hisG4, argE3, and trpE65 mutations are known to be ochre nonsense mutations (32), yet the precise molecular base change in each of these mutations is not known, to the best of our knowledge. We have transduced the trpE65 mutation into a uvrA recF143 strain of AB1157 and have observed that the recF143 mutation reduced UV-induced reversion of trpE65 (data not shown). This result suggests that it is the different mutations used in the reversion assay which account for the observed difference. However, systematic studies on the effect of recF, recO, and recR mutations on the UV-induced reversions of hisG4 in the EWRP-1A genetic background and trpE65 in the AB1157 genetic background are needed to provide a definite answer.

There are several questions that need to be addressed even when the above discrepancy is resolved. First, why do mutations in recF, recO, and recR produce a deficiency in UV-induced reversion of trpE65 but not in that of hisG4 or argE3? Second, why can recA2020, which only partially suppresses both the UV sensitivity and the postreplication repair deficiency of recF, recO, and recR mutants (39, 45), fully restore the normal UVM of recF, recO, and recR mutants (Fig. 4)? We offer the following speculations based on the known defects caused by recF, recO, and recR mutations. The fact that mutations in the recF, recO, and recR genes are known to produce a 50% deficiency in the repair of DNA daughter-strand gaps (30, 39, 44) suggests that half of the repair of DNA daughter-strand gaps does not require the participation of the recF, recO, and recR genes. Assuming that UV-induced reversion of a given mutation requires that errors be made during the processing of a specific DNA daughter-strand gap, it is possible that the processing of such a gap requires the participation of the recF, recO, and recR genes. A provocative speculation, for example, is that the recF, recO, and recR genes are involved in the processing of DNA daughter-strand gaps produced in leading-strand synthesis but not for those produced in lagging-strand synthesis. According to such a postulate, reversion of a mutation requiring errors to be made during the processing of leading-strand DNA daughter-strand gaps would show a dependency on the recF, recO, and recR genes. On the other hand, reversion of a mutation requiring errors to be made during the processing of lagging-strand DNA daughter-strand gaps would be independent of the recF, recO, and recR genes. This may explain why there are different requirements for the recF, recO, and recR genes in different reversion assays. To account for the differential effects of recA2020 on the suppression of deficiencies both in postreplication repair and in UVM in recF, recO, and recR mutants, we suggest that among the DNA daughter-strand gaps that require the participation of the recF, recO, and recR genes for repair, some are processed through an error-free process and some are processed through an error-prone process. Restoration by recA2020 of the portion of postreplication repair which is error prone can explain why a partial restoration of postreplication repair proficiency can lead to a full restoration of UVM in recF, recO, and recR mutants. According to this postulate, the portion of postreplication repair restored by recA2020 is error prone. Future experiments shall test the validity of this hypothesis.

In conclusion, we have presented evidence for an involvement of the recF, recO, and recR genes in UVM. Our results indicate that these genes do not act by affecting RecA’s functions in the induction of SOS regulons or in the cleavage of UmuD. We suggest that the recF, recO, and recR gene products act by affecting the third role of RecA in UVM.