Genetic functions promoting homologous recombination in Escherichia coli: a study of inversions in phage lambda.

We have studied homologous recombination in a derivative of phage lambda containing two 1.4-kb repeats in inverted orientation. Inversion of the intervening 2.5-kb segment occurred efficiently by the Escherichia coli RecBC pathway but markedly less efficiently by the lambda Red pathway or the E. coli RecE or RecF pathways. Inversion by the RecBCD pathway was stimulated by Chi sites located to the right of the invertible segment; this stimulation decreased exponentially by a factor of about 2 for each 2.2 kb between the invertible segment and the Chi site. In addition to RecA protein and RecBCD enzyme, inversion by the RecBC pathway required single-stranded DNA binding protein, DNA gyrase, DNA polymerase I and DNA ligase. Inversion appeared to occur either intra- or intermolecularly. These results are discussed in the framework of a current molecular model for the RecBC pathway of homologous recombination.

Analysis of mutagenesis induced by expression of retroviral reverse transcriptase in Escherichia coli.

We showed that expression of reverse transcriptase from HIV or MuLV resulted in exceptionally high levels of mutagenesis in E coli. We observed high rates of mutagenesis in plasmid genes when reverse transcriptase was expressed off that plasmid. Although very high rates were observed in cis, our experiments could not detect mutagenic events in markers in other replicons (ie in trans). These results suggest mutagenic events occur preferentially on the same replicon in which the reverse transcriptase is encoded.

Selective inhibition of RecA functions by the Hc1 nucleoid condensation protein from Chlamydia trachomatis.

During the normal biphasic life cycle of Chlamydia trachomatis, the histone-like protein Hc1 promotes the condensation of nucleoids in elementary bodies, it may also displace nucleoproteins, including repair functions from chromatin. Hc1 was found to effectively inhibit the recombination and repair of the weak binding RecA430 mutant protein from Escherichia coli, but had minimal effects on the parental RecA(+) protein. Expression of Hc1 was also found to inhibit the repair activities of the C. trachomatis RecA protein but not recombination. These results suggest that chlamydial RecA may have evolved mechanisms to minimize Hc1 competition for recombinational activities.

The recA gene of Chlamydia trachomatis: cloning, sequence, and characterization in Escherichia coli.

The recA gene of Chlamydia trachomatis was isolated by complementation of an Escherichia coli recA mutant. The cloned gene restored resistance to methyl methanesulfonate in E. coli recA mutants. The DNA sequence of the chlamydial gene was determined and the deduced protein sequence compared with other RecA proteins. In E. coli recA deletion mutants, the cloned gene conferred moderate recombinational activity as assayed by Hfr matings. The chlamydial recA gene was efficient in repairing alkylated DNA but less so in repairing of UV damage when compared with the E. coli homologue. As detected by an SOS gene fusion, a small but measurable amount of LexA co-cleavage was indicated.

Analysis of recA mutants with altered SOS functions.

The Escherichia coli RecA protein has at least three roles in SOS mutagenesis: (1) derepression of the SOS regulon by mediating LexA cleavage; (2) activation of the UmuD mutagenesis protein by mediating its cleavage; and (3) targeting the Umu-like mutagenesis proteins to DNA. Using a combined approach of molecular and physiological assays, it is now possible to determine which of the three defined steps has been altered in any recA mutant. In this study, we have focused on the ability of six particular recA mutants (recA85, recA430, recA432, recA433, recA435 and recA730) to perform these functions. Phenotypically, recA85 and recA730 were similar in that in lexA+ and lexA(Def) backgrounds, they exhibited constitutive coprotease activity towards the UmuD mutagenesis protein. Somewhat surprisingly, in a lexA(Ind-) background, UmuD cleavage was damage inducible, suggesting that the repressed level of the RecA* protein cannot spontaneously achieve a fully activated state. Although isolated in separate laboratories, the nucleotide sequence of the recA85 and recA730 mutants revealed that they were identical, with both alleles possessing a Glu38-->Lys change in the mutant protein. The recA430, recA433 and recA435 mutants were found to be defective for both lambda mutagenesis and UmuD cleavage. lambda mutagenesis was fully restored, however, to the recA433 and recA435 strains by a low copy plasmid expressing the mutagenically active UmuD' protein. In contrast, lambda mutagenesis was only partially restored to a recA430 strain by a high copy UmuD' plasmid, suggesting that RecA430 may also be additionally defective in targeting the Umu proteins to DNA. Sequence analysis of the recA433 and recA435 alleles revealed identical substitutions resulting in Arg243-->His. The recA432 mutation had a complex phenotype in that its coprotease activity towards UmuD depended upon the lexA background: inducible in lexA+ strains, inefficient in lexA(Ind-) cells and constitutive in a lexA(Def) background. The recA432 mutant was found to carry a Pro119-->Ser substitution, a residue believed to be at the RecA subunit interface; thus this complex phenotype may result from alterations in the assembly of RecA multimers.

Levels of chromosomally encoded Umu proteins and requirements for in vivo UmuD cleavage.

Most of the inducible mutagenesis observed in Escherichia coli after treatment with many DNA damaging agents is dependent upon the products of the umuD,C operon. RecA-mediated proteolytic processing of UmuD yields a carboxyl-terminal fragment (UmuD') that is active for mutagenesis. Processing of UmuD is therefore a critical step in the fixation of mutations. In this paper we have analyzed the requirements for UmuD processing in vivo. Standard immuno-detection assays, coupled with a sensitive chemiluminescence detection assay, have been utilized to probe levels of chromosomally encoded Umu proteins from whole-cell E. coli extracts. We found that the derepression of additional SOS gene products, other than RecA, was not required for UmuD processing. Moreover, efficient cleavage of UmuD was observed only in the presence of elevated levels of activated RecA, suggesting that efficient processing would occur only under conditions of severe DNA damage. Detection of chromosomally encoded Umu proteins has allowed us, for the first time, to measure directly the cellular steady-state levels of these proteins under various SOS inducing conditions. UmuD was present at approximately 180 copies per uninduced cell and was measured at approximately 2400 copies per cell in strains that lacked a functional repressor. Induced levels of UmuC were approximately 12-fold lower than UmuD with approximately 200 molecules per cell. These levels of cellular UmuC protein suggest that it functions through specific protein-DNA or protein-protein interactions, possibly as a lesion recognition protein or by interacting with DNA polymerase III.

Novel mechanism for UV sensitivity and apparent UV nonmutability of recA432 mutants: persistent LexA cleavage following SOS induction.

The recA432 mutant allele was isolated (T. Kato and Y. Shinoura, Mol. Gen. Genet. 156:121-131, 1977) by virtue of its defect in cellular mutagenesis (Mut-) and its hypersensitivity to damage by UV irradiation (UVs), which were phenotypes expected for a recA mutant. However, we found that in a different genetic background (lexA51 sulA211 uvrB+), recA432 mutants expressed certain mutant phenotypes but not the Mut- and UVs phenotypes (D.G. Ennis, N. Ossanna, and D.W. Mount, J. Bacteriol. 171:2533-2541, 1989). We present several lines of evidence that these differences resulted from the sulA genotype of the cell and that the apparent UVs and Mut- phenotypes of the sulA+ derivatives resulted from lethal filamentation of induced cells because of persistent derepression of sulA. First, transduction of sulA(Def) mutations into the recA432 strains restored cellular mutagenesis and resistance to UV. Second, recA432 sulA+ strains underwent filamentous death following SOS-inducing treatments. Third, cleavage of LexA repressor in a recA432 strain continued at a rapid rate long after UV induction, at a time when cleavage of the repressor in the recA+ parental strain had substantially declined. Fourth, we confirmed that a single mutation (recA432) conferring both the UVs and Mut- phenotypes mapped to the recA gene. These findings indicate that the RecA432 mutant protein is defective in making the transition back to the deactivated state following SOS induction; thus, the SOS-induced state of recA432 mutants is prolonged and can account for an excess of SulA protein, leading to filamentation. These results are discussed in the context of molecular models for RecA activation for LexA and UmuD cleavage and their roles in the control of mutagenesis and cell division in the SOS response.

Increased expression of the Escherichia coli umuDC operon restores SOS mutagenesis in lexA41 cells.

The lexA41 allele of Escherichia coli encodes a semidefective mutant repressor that is also resistant to RecA facilitated cleavage. Cells harboring the lexA41 allele were found previously to repress only a subset of operons in the SOS regulon. lexA41 cells cannot promote SOS mutagenesis, presumably because one or more operons required for mutagenesis are repressed by this mutant repressor. Using the lac regulatory system to increase the expression of the umuDC operon, we were able to restore mutagenesis in the lexA41 mutant. We conclude that the products of the umuDC operon appear to be uniquely limiting in this mutant.

Genetic separation of Escherichia coli recA functions for SOS mutagenesis and repressor cleavage.

Evidence is presented that recA functions which promote the SOS functions of mutagenesis, LexA protein proteolysis, and lambda cI repressor proteolysis are each genetically separable from the others. This separation was observed in recombination-proficient recA mutants and rec+ (F' recA56) heterodiploids. recA430, recA433, and recA435 mutants and recA+ (F' recA56) heterodiploids were inducible for only one or two of the three functions and defective for mutagenesis. recA80 and recA432 mutants were constitutively activated for two of the three functions in that these mutants did not have to be induced to express the functions. We propose that binding of RecA protein to damaged DNA and subsequent interaction with small inducer molecules gives rise to conformational changes in RecA protein. These changes promote surface-surface interactions with other target proteins, such as cI and LexA proteins. By this model, the recA mutants are likely to have incorrect amino acids substituted as sites in the RecA protein structure which affect surface regions required for protein-protein interactions. The constitutively activated mutants could likewise insert altered amino acids at sites in RecA which are involved in the activation of RecA protein by binding small molecules or polynucleotides which metabolically regulate RecA protein.

In vivo stability of the Umu mutagenesis proteins: a major role for RecA.

The Escherichia coli Umu proteins play critical roles in damage-inducible SOS mutagenesis. To avoid any gratuitous mutagenesis, the activity of the Umu proteins is normally kept to a minimum by tight transcriptional and posttranslational regulation. We have, however, previously observed that compared with an isogenic recA+ strain, the steady-state levels of the Umu proteins are elevated in a recA730 background (R. Woodgate and D. G. Ennis, Mol. Gen. Genet. 229:10-16, 1991). We have investigated this phenomenon further and find that another coprotease-constitutive (recA*) mutant, a recA432 strain, exhibits a similar phenotype. Analysis revealed that the increased steady-state levels of the Umu proteins in the recA* strains do indeed reflect an in vivo stabilization of the proteins. We have investigated the basis for the phenomenon and find that the mutant RecA* protein stabilizes the Umu proteins by not only converting the labile UmuD protein to the much more stable (and mutagenically active) UmuD' protein but by directly stabilizing UmuD' itself. In contrast, UmuC does not appear to be directly stabilized by RecA* but is instead dramatically stabilized in the presence of UmuD'. On the basis of these observations, we suggest that formation of a UmuD'C-RecA*-DNA quaternary complex protects the UmuD'C proteins from proteolytic degradation and as a consequence helps to promote the switch from error-free to error-prone mechanisms of DNA repair.

Regulation of SOS mutagenesis by proteolysis.

DNA damage-inducible mutagenesis in Escherichia coli is largely dependent upon the activity of the UmuD (UmuD') and UmuC proteins. The intracellular level of these proteins is tightly regulated at both the transcriptional and the posttranslational levels. Such regulation presumably allows cells to deal with DNA damage via error-free repair pathways before being committed to error-prone pathways. We have recently discovered that as part of this elaborate regulation, both the UmuD and the UmuC proteins are rapidly degraded in vivo. We report here that the enzyme responsible for their degradation is the ATP-dependent serine protease, Lon. In contrast, UmuD' (the posttranslational product and mutagenically active form of UmuD) is degraded at a much reduced rate by Lon, but is instead rapidly degraded by another ATP-dependent protease, ClpXP. Interestingly, UmuD' is rapidly degraded by ClpXP only when it is in a heterodimeric complex with UmuD. Formation of UmuD/UmuD' heterodimers in preference to UmuD' homodimers therefore targets UmuD' protein for proteolysis. Such a mechanism allows cells to reduce the intracellular levels of the mutagenically active Umu proteins and thereby return to a resting state once error-prone DNA repair has occurred. The apparent half-life of the heterodimeric UmuD/D' complex is greatly increased in the clpX::Kan and clpP::Kan strains and these strains are correspondingly rendered virtually UV non-mutable. We believe that these phenotypes are consistent with the suggestion that while the UmuD/D' heterodimer is mutagenically inactive, it still retains the ability to interact with UmuC, and thereby precludes the formation of the mutagenically active UmuD'2C complex.

Escherichia coli umuDC mutants: DNA sequence alterations and UmuD cleavage.

The products of the chromosomally encoded umuDC genes are directly required for mutagenesis in Escherichia coli. Strains with either umuD or umuC mutations are rendered phenotypically non-mutable. To ascertain the molecular basis of this non-mutability, we determined the DNA sequence alterations of seven chromosomal umuDC mutants. Six mutants (umuD1, umuD44, umuD77, umuC36, umuC25, and umuC104) were found to be single base-pair substitutions that resulted in missense mutations. The Tn5 transposon insertion mutation (umuC122) resulted in a missense mutation followed immediately by a termination codon, producing a truncated UmuC protein lacking 102 carboxyl-terminal amino acids. All of the mutations were found to reside in regions of the UmuD and UmuC proteins that share high homology with analogous proteins. Chemiluminescent immunoassays revealed that the umuD1, umuD44, and umuD77 mutations all resulted in a non-cleavable UmuD protein. Because UmuD cleavage is a prerequisite for mutagenesis, the lack of UmuD processing appears to be the molecular basis for the non-mutable phenotype in these strains. These studies re-emphasize the critical nature of the RecA-mediated cleavage of UmuD for inducible mutagenesis and provide insights into the functional domains of the UmuC protein.

The enhanced mutagenic potential of the MucAB proteins correlates with the highly efficient processing of the MucA protein.

Inducible mutagenesis in Escherichia coli requires the direct action of the chromosomally encoded UmuDC proteins or functional homologs found on certain naturally occurring plasmids. Although structurally similar, the five umu-like operons that have been characterized at the molecular level vary in their ability to enhance cellular and phage mutagenesis; of these operons, the mucAB genes from the N-group plasmid pKM101 are the most efficient at promoting mutagenesis. During the mutagenic process, UmuD is posttranslationally processed to an active form, UmuD'. To explain the more potent mutagenic efficiency of mucAB compared with that of umuDC it has been suggested that unlike UmuD, intact MucA is functional for mutagenesis. To examine this possibility, we have overproduced and purified the MucA protein. Although functionally similar to UmuD, MucA was cleaved much more rapidly both in vitro and in vivo than UmuD. In vivo, restoration of mutagenesis functions to normally nonmutable recA430, recA433, recA435, or recA730 delta(umuDC)595::cat strains by either MucA+ or mutant MucA protein correlated with the appearance of the cleavage product, MucA'. These results suggest that most of the differences in mutagenic phenotype exhibited by MucAB and UmuDC correlate with the efficiency of posttranslational processing of MucA and UmuD rather than an inherent activity of the unprocessed proteins.

The bacteriophage P1 HumD protein is a functional homolog of the prokaryotic UmuD'-like proteins and facilitates SOS mutagenesis in Escherichia coli.

The Escherichia coli umuD and umuC genes comprise an operon and encode proteins that are involved in the mutagenic bypass of normally replication-inhibiting DNA lesions. UmuD is, however, unable to function in this process until it undergoes a RecA-mediated cleavage reaction to generate UmuD'. Many homologs of umuDC have now been identified. Most are located on bacterial chromosomes or on broad-host-range R plasmids. One such putative homolog, humD (homolog of umuD) is, however, found on the bacteriophage P1 genome. Interestingly, humD differs from other umuD homologs in that it encodes a protein similar in size to the posttranslationally generated UmuD' protein and not UmuD, nor is it in an operon with a cognate umuC partner. To determine if HumD is, in fact, a bona fide homolog of the prokaryotic UmuD'-like mutagenesis proteins, we have analyzed the ability of HumD to complement UmuD' functions in vivo as well as examined HumD's physical properties in vitro. When expressed from a high-copy-number plasmid, HumD restored cellular mutagenesis and increased UV survival to normally nonmutable recA430 lexA(Def) and UV-sensitive DeltaumuDC recA718 lexA(Def) strains, respectively. Complementing activity was reduced when HumD was expressed from a low-copy-number plasmid, but this observation is explained by immunoanalysis which indicates that HumD is normally poorly expressed in vivo. In vitro analysis revealed that like UmuD', HumD forms a stable dimer in solution and is able to interact with E. coli UmuC and RecA nucleoprotein filaments. We conclude, therefore, that bacteriophage P1 HumD is a functional homolog of the UmuD'-like proteins, and we speculate as to the reasons why P1 might require the activity of such a protein in vivo.

Dual role for Escherichia coli RecA protein in SOS mutagenesis.

Induction of the Escherichia coli SOS system increases the ability of the cells to perform DNA repair and mutagenesis. Previous work has shown that this increased mutagenesis is the result of derepression of specific genes through a complex regulatory mechanism controlled by LexA and RecA proteins. One role of RecA protein in this process is to facilitate proteolytic cleavage of LexA protein (the repressor) in response to an inducing signal that reversibly activates RecA protein to perform this function. We show that activated RecA protein plays a second role in SOS mutagenesis, as revealed by analyzing repair of UV-damaged phage lambda in host mutants with alterations in the SOS regulatory system. First, phage mutagenesis was not expressed constitutively in a mutant that is derepressed through lack of functional LexA protein; activated RecA protein was still required. Second, phage mutagenesis was constitutively expressed in the presence of recA mutations that alter RecA protein so that it is activated in normally growing cells. There was also RecA-dependent constitutive expression of SOS mutagenesis in host mutants that lack functional LexA protein and carry plasmids. We discuss several possible biochemical mechanisms for this second role of activated RecA protein in SOS mutagenesis.