|Home | About | Journals | Submit | Contact Us | Français|
The rap gene of bacteriophage λ was placed in the chromosome of an Escherichia coli K-12 strain in which the recBCD gene cluster had previously been replaced by the λ red genes and in which the recG gene had been deleted. Recombination between linear double-stranded DNA molecules and the chromosome was tested in variants of the recGΔ red+ rap+ strain bearing mutations in genes known to affect recombination in other cellular pathways. The linear DNA was a 4-kb fragment containing the cat gene, with flanking lac sequences, released from an infecting phage chromosome by restriction enzyme cleavage in the cell. Replacement of wild-type lacZ with lacZ::cat was monitored by measuring the production of Lac-deficient chloramphenicol-resistant bacterial progeny. The results of these experiments indicated that the λ rap gene could functionally substitute for the E. coli ruvC gene in Red-mediated recombination.
The rap gene of phage λ has a number of properties suggesting that it functions in homologous recombination. It is required for efficient formation or stability of RecBCD-generated cointegrates between λ and plasmids bearing homologous sequences (3). This observation is responsible for its name: recombination adept with plasmids. The rap gene also influences both the nature and clustering of recombination events in λ crosses (17). On the basis of the rap gene's map location and apparent involvement in recombination, it was proposed that rap is a functional analog (though it is not a homolog) of the rusA genes of phage 82 and a cryptic Escherichia coli prophage (6). The RusA protein is a Holliday junction resolvase, which can substitute for the RuvC protein of E. coli in promoting recombination (7, 15). Consistent with this proposal, it was found that the rap gene encodes a nuclease which cleaves at the branch points of Holliday junctions; in addition, it cleaves three-stranded junctions (D-loops) (16).
An E. coli strain in which the RecBCD recombinase is replaced with the Red system of phage λ is hyper-rec. In such a strain recombination involving short stretches of sequence identity (30 to 1,000 bp) occurs at an elevated level, relative to wild-type E. coli, and is strongly dependent upon both Red function and double-strand breaks (2, 9, 11, 18, 19). This hyper-rec state of a bacterium expressing some of the phage recombination genes (gam, bet, and exo) partially models the conditions which prevail in a phage-infected cell. It has been found that expression in a red+ cell of genes from the nin region of the λ chromosome, including rap, does not further stimulate recombination but makes recombination partially independent of recF, recO, recR, and ruvC (12). In this study we show that the rap gene accounts for some of this activity, complementing the recombination defect of a ruvC mutant.
The rap gene was PCR amplified from λ cI857 S7 DNA with primers having the sequences 5′-CACGAGGAAGCATATGATGGCTA-3′ and 5′-GTTTCAACGAGCTCTTATGGAATGGTT-3′. Following digestion with NdeI and SacI restriction endonucleases (sites are underlined in the primer sequences), it was ligated between the NdeI and SacI sites of an expression vector bearing a promoter/ribosome binding site cassette with the sequence GGGCCCGCACCCCAGGCTTTACATTGTGAGCGGATAACAATATAATGAAGCTTAATAAGGAGGAAAAACATATG. The promoter (−35 and −10 hexamers are in bold type) is designated Pmac. Its design is based on the studies of Lanzer and Bujard (4). It promotes a moderate level of transcription and is controllable by the lac repressor (unpublished observations). The Pmac-rap assembly was sequenced and found to contain wild-type rap (data not shown). It was cloned between the ApaI and SacI sites of a derivative of pTP838 (10) in which the NdeI site had been removed by digestion with NdeI, filling in with E. coli DNA polymerase I large fragment, and religation. The resulting plasmid, pTP914, contains the following elements in order: an AatII site; galK amino terminal-encoding sequences; Pmac-rap; a kanamycin resistance determinant derived from Tn903; galK carboxy terminal-encoding sequences; a BamHI site; and the pBR322 replication origin and beta lactamase gene. pTP914 was digested with BamHI and was partially digested with AatII (the rap gene contains an AatII site). The digested DNA was introduced into E. coli strain TP656 via electroporation. Kanamycin-resistant transformants were screened for ampicillin sensitivity and were tested by colony PCR for the presence of an insertion of the expected size in galK. An otherwise identical strain, bearing the gfp (green fluorescent protein) gene of pGreenTIR (8) in place of rap, was constructed as a rap mutant control.
Starting with an E. coli strain in which the recBCD gene cluster was replaced with Ptac-gam-bet-exo pae cI, variants lacking Red functions were constructed by transformation with linear plasmids. One plasmid, pTP963, was constructed by replacing the Bgl2 site-bounded Ptac-gam-bet-exo module of plasmid pTP822 (13) with a segment of DNA containing the tetracycline resistance determinant of Tn10, previously cloned in plasmid pTP802 (10). The other plasmid, pTP967, was constructed in two steps. First, DNA between the HpaI sites in bet and exo was replaced by a linker for the restriction enzyme NotI. Then a segment of DNA containing the tetracycline resistance determinant of Tn10, previously cloned in plasmid pTP857 (10), was inserted into the NotI site.
Bacterial strains were grown overnight with aeration in Luria-Bertani (LB; 1% tryptone, 0.5% yeast extract, 0.5% NaCl, 1 mM NaOH) medium supplemented with 25 μg of kanamycin/ml and, in the case of tetracycline-resistant strains, 25 μg of tetracycline/ml. Overnight cultures were diluted 100-fold into LB supplemented with 10 mM MgSO4 plus 0.2% maltose, grown with aeration at 37°C, chilled on ice, and adjusted, if necessary, to a density of approximately 2 × 108/ml. Phage was added at a multiplicity of 10, along with isopropyl-β-d-thiogalactopyranoside (IPTG) to a concentration of 1 mM. Tubes containing mixtures of phage and bacteria were incubated on ice for 10 min, shifted to a 37°C water bath for 5 min, and then aerated by rolling at 37°C for 55 min. Cultures were plated on LB agar for determination of viable titer and were plated on LB agar supplemented with 10 μg of chloramphenicol/ml plus 1 mM IPTG plus 80 μg of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal)/ml for determination of lac::cat819 recombinants.
The recombination event monitored in these crosses involves replacement of a small segment of the lacZ gene with the chloramphenicol resistance-conferring cat gene. The cat gene is brought into the cell by infection with λ lac::cat819 nin5. The phage injects its chromosome, which circularizes but is not transcribed or replicated due to the presence of cI repressor in the cell. The PaeR7 restriction endonuclease, also present in the cell, cuts the phage chromosome in two places, releasing a double-stranded DNA fragment consisting of the cat gene flanked on either side by 1.5-kb segments of lac sequence.
To verify that recombination between the cell chromosome and λ lac::cat819 nin5 is dependent upon Red function in the Red-for-RecBCD-substituted E. coli strain TP656, variants lacking Red functions were constructed (Table (Table1).1). In one strain the entire Ptac-gam-bet-exo module was eliminated; in another, the last 64 codons of bet and the first 46 codons of exo were eliminated. Formation of lac::cat819 recombinants was reduced approximately 700- and 400-fold by these two modifications, respectively (averages of two measurements; data not shown).
Starting with an E. coli strain bearing a (recC-ptr-recB-recD)Δ:: Ptac-gam-bet-exo-pae-cI substitution as well as deletions of recG and sulA, two isogenic sets of rap+ and mutant rap strains with mutations in various other recombination genes were constructed. The reason for employing a background lacking recG is that deletion of recG elevates the frequency of Red-mediated recombination; deletion of sulA was found to enhance the viability of recBCDΔ::red strains lacking recF, recO, or recR function (12). The results of crosses in these strains are summarized in Table Table2.2. As seen previously (12), recombination in the non-rap-containing background depends on recA, recF, recO, recQ, recR, ruvAB, and ruvC. The data in Table Table22 indicate that mutation of recJ in the red+ recGΔ background had little effect on recombination; in contrast, it significantly reduced recombination in the previous study. We attribute the difference in outcomes to the absence of sulA in the strains constructed for this study, which improves the viability of the recG recJ double mutant. In addition, the data in Table Table22 show that deletion of recN has little effect on Red-mediated recombination.
Expression of rap alters the genetic dependency of Red-mediated recombination, as shown in Table Table2.2. Recombination in the rap+ cell is much less dependent upon ruvC, as predicted. In addition, recombination is somewhat less dependent upon recF. Slight differences are also seen in recA, recQ, recR, and ruvAB mutants, but these are of doubtful significance, representing small changes in low recombination frequencies.
The restoration of recombination proficiency to a recGΔ ruvC mutant by rap suggested that rap might restore some degree of resistance to DNA-damaging agents as well, but this appears not to be the case. Testing UV sensitivity, as described previously (12), we found that there was no reliably measurable difference between rap+ and mutant rap versions of the recGΔ ruvC strain (data not shown).
Introduction of the ruvC53 allele into E. coli (recC-ptr-recB-recD)Δ::Ptac-gam-bet-exo-pae-cI recGΔ sulAΔ produces a cell which grows quite slowly relative to its ruvC+ parent. When rap is expressed in the cell, the growth defect is less pronounced but is still detectable. However, cultures of both rap+ and non-rap-containing variants are frequently taken over by fast-growing, recombination-proficient, and relatively UV-resistant mutants. We assume these mutants are pseudorevertants, with mutations activating the rusA gene as described by Mandal et al. (7), but we have not tested them further. To ensure that the results shown in Table Table22 were not influenced by the presence of such revertants, ruvC mutant strains were reconstructed by P1 transduction for each experiment. Only crosses which showed no evidence of fast-growing revertants after plating were used in determining recombination frequencies.
None of the phenotypes of the rap+ strains is affected by the presence or absence of the lac inducer IPTG, although the control gfp+ strain fluoresces more brightly when grown in the presence of IPTG (data not shown). We infer that the basal level of expression from Pmac is sufficient to suppress the recombination and cell growth phenotypes of recBCDΔ:: red-pae-cI recGΔ sulAΔ ruvC53, but neither the basal nor induced level suppresses the UV sensitivity phenotype. The inability of the wild-type, chromosomal lacI gene to exert tight negative control over rap expression may be due to the presence in the cell of four different promoters capable of binding (and thus partially titrating) the lac repressor: wild-type Plac, Ptac-gam-bet-exo, PlacUV5-cI, and Pmac-rap. However, this hypothesis has not been tested.
The properties of rap+ cells suggest that Rap can substitute for RuvC in some cellular pathways but not in others. On the other hand, the Rap protein's partial complementation of a recF mutant (Table (Table2)2) and its broader range of nucleolytic activities (16) suggests that, conversely, Rap may carry out an additional function(s), most likely in phage replication, repair, or recombination, that RuvC does not.
We thank Kenan Murphy for helpful discussions.
This study was supported in part by grant GM51609 from the National Institutes of Health.