Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
The physiological role of topoisomerase III is unclear for any organism. We show here that the removal of topoisomerase III in temperature sensitive topoisomerase IV mutants in Escherichia coli results in inviability at the permissive temperature. The removal of topoisomerase III has no effect on the accumulation of catenated intermediates of DNA replication, even when topoisomerase IV activity is removed. Either recQ or recA null mutations, but not helD null or lexA3, partially rescued the synthetic lethality of the double topoisomerase III/IV mutant, indicating a role for topoisomerase III in recombination. We find a bias against deleting the gene encoding topoisomerase III in ruvC53 or ΔruvABC backgrounds compared with the isogenic wild-type strains. The topoisomerase III RuvC double mutants that can be constructed are five- to 10-fold more sensitive to UV irradiation and mitomycin C treatment and are twofold less efficient in transduction efficiency than ruvC53 mutants. The overexpression of ruvABC allows the construction of the topoisomerase III/IV double mutant. These data are consistent with a role for topoisomerase III in disentangling recombination intermediates as an alternative to RuvABC to maintain the stability of the genome.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Topoisomerases are ubiquitous enzymes that are required for nearly all aspects of nucleic acid metabolism (Champoux, 2001; Wang, 2002; Corbett and Berger, 2004). Becoming covalently bound to the DNA, topoisomerases make transient breaks in DNA, pass DNA strands through the breaks, and then reseal the DNA. Topoisomerases allow cells to modulate three topological forms of DNA: linking number (Lk) (the number of times the two strands of DNA are intertwined), catenanes (intermolecular linking of multiple DNA molecules), and knots (intramolecular tangling of one DNA molecule). These enzymes are divided into two groups: those that alter Lk in steps of one (type-1) and those that alter Lk in steps of two (type-2).
Topoisomerase III is well conserved across evolution, yet the in vivo function of topoisomerase III remains elusive (reviewed in Wang, 2002). Biochemical studies have shown that E. coli topoisomerase III relaxes negative supercoils and unties knots that are single-stranded or that contain single-stranded regions (DiGate and Marians, 1988; Du et al., 1995). The enzyme is most efficient at decatenating DNA substrates containing a nick or a gap (DiGate and Marians, 1988). Topoisomerase III is also capable of unlinking what has been termed ‘precatenanes’ (Ullsperger et al., 1995) that may occur during DNA replication (Hiasa and Marians, 1994; Peter et al., 1998; Nurse et al., 2003). It has been reported that purified E. coli topoisomerase III and RecQ helicase together can catenate intact double-stranded supercoiled DNA (Harmon et al., 1999). This was the first evidence that a type-1 topoisomerase is capable of performing a type-2 topoisomerase function with covalently closed double-stranded DNA. In addition, a genetic study has suggested that topoisomerase III and topoisomerase I activity might be required for chromosomal segregation following homologous recombination in E. coli (Zhu et al., 2001). It has also been reported that the overexpression of topB can rescue cells lacking topA by disentangling R-loops (Broccoli et al., 2000).
Additional data imply that topoisomerase III plays an important cellular role. E. coli cells lacking topoisomerase III display an increase in recombination between regions of short homology and an increase in frameshift mutations (Schofield et al., 1992; Uematsu et al., 1997). The consequences of the loss of topoisomerase III activity have been studied more extensively in eukaryotic cells. Mice lacking topoisomerase IIIα die during the early stages of embryogenesis and are reabsorbed (Li and Wang, 1998). Removal of the β isoform of topoisomerase III is better tolerated. Top3β–/– mice survive to term, but have reduced lifespans and reduced fertility, compared with their wild-type littermates (Kwan and Wang, 2001). Further examination of cells from these mice revealed aneuploidy (Kwan et al., 2003). Saccharomyces cerevisiae and Schizosaccharomyces pombe topoisomerase III null mutants sporulate at a lower efficiency than wild-type strains (Goodwin et al., 1999). This defect in meiosis can be suppressed by preventing genetic recombination (Gangloff et al., 1999). S. cerevisiae cells lacking topoisomerase III display an increase in deletions of short repetitive sequences, an increase in recombination at regions of short homologies, and also exhibit a significant increase in homologous recombination-dependent rearrangements of rDNA arrays (Gangloff et al., 1994).
The data presented here support a model in which RecA and RecQ create Holliday junction intermediates that topoisomerase III resolves in an alternate pathway to the RuvABC pathway. This topoisomerase-mediated resolution would always result in non-crossover recombination thus preventing genomic rearrangements associated with crossover recombination.
Effect of topoisomerase III on DNA supercoiling, decatenating and unknotting in vivo
One possible cellular role of topoisomerase III is that it unlinks catenated intermediates of DNA replication (Hiasa and Marians, 1994; Harmon et al., 1999; Wu et al., 1999; Harmon et al., 2003; Nurse et al., 2003). We tested this directly by measuring the level of catenanes in a plasmid substrate, pJB3.5d (Bliska and Cozzarelli, 1987), in the absence of topoisomerase III. If topoisomerase III unlinks catenanes, then replication intermediates should accumulate in the cell. In a strain lacking the gene encoding topoisomerase III (topB), there were no catenanes (data not shown). At the same time, catenanes accumulated in the positive control, an isogenic topoisomerase IV temperature sensitive mutant at the non-permissive temperature. Steady-state plasmid DNA knotting levels were identical in the presence or absence of topoisomerase III (data not shown).
We reasoned that the strong decatenating activity of topoisomerase IV could mask a small, but potentially important, role of topoisomerase III. Therefore, to measure catenane accumulation when both topoisomerase III and topoisomerase IV activities were removed, we transduced the ΔtopB allele into a strain that contains a norfloxacin-resistant allele of gyrase, gyrAL83. The production of catenanes requires DNA replication and, thus, a functional gyrase (Adams et al., 1992). Normally, norfloxacin targets both gyrase and topoisomerase IV in E. coli, but topoisomerase IV is specifically inhibited and catenanes accumulate when the drug is added to the gyrAL83-containing strain (Khodursky et al., 1995). Non-synchronous plasmid replication takes place at a steady state level. It takes mere seconds to complete one round of plasmid replication; at any given time, ∼10% of the plasmids are undergoing DNA replication (Adams et al., 1992). In addition, the inhibition of topoisomerase IV with norfloxacin is not absolute. The enzyme retains activity at a rate ∼1/100 the normal level (Zechiedrich and Cozzarelli, 1995; Zechiedrich et al., 1997). Consequently, DNA replication intermediates represent a snapshot of the steady-state replication process. This snapshot shows that the accumulation of catenated plasmid DNA was the same in both the gyrAL83 (Fig. 1, lanes 2–6) and ΔtopB gyrAL83 (Fig. 1, lanes 8–12) cells. Therefore, topoisomerase III does not appear normally to assist topoisomerase IV in unlinking catenated intermediates of plasmid DNA replication.
We also measured DNA supercoil relaxation following gyrase inhibition with norfloxacin (Zechiedrich et al., 2000) in the presence or absence of topoisomerase III (data not shown). The rate or extent of DNA supercoil relaxation by topoisomerases I and IV was identical with and without topoisomerase III. The rate or extent of DNA supercoil relaxation by topoisomerase I plus or minus topoisomerase III was also identical following inhibition of gyrase and topoisomerase IV activity with norfloxacin in the gyrA+parC+ strain. The removal of topoisomerase IV activity alone with norfloxacin in the gyrAL83 parC+ strain causes a shift to more negatively supercoiled plasmid DNA (Zechiedrich et al., 2000). We found that the shift in the presence or absence of topoisomerase III was identical. Thus, topoisomerase III seems to play no role in regulating DNA supercoiling in E. coli.
Synthetic lethality of a topoisomerase III null allele with topoisomerase IV temperature-sensitive alleles
We attempted to transduce the topB deletion (ΔtopB::kan) into a topoisomerase IV temperature-sensitive mutant, parEts, at the permissive temperature, 30°C (Kato et al., 1990). We obtained very few, slowly growing colonies that were not viable at 37°C (data not shown). This result indicates that the combination of the two mutant alleles was problematic to the cell. To test quantitatively whether the topB deletion could coexist with the topoisomerase IV mutation, we carried out genetic linkage experiments using bacteriophage P1-mediated transduction as outlined in Fig. 2A. In these experiments, an antibiotic resistance-encoding gene (Tn10tet) was transduced near ΔtopB::kan to make strain CRL6. We used CRL6 as the P1 donor to transduce Tn10tetΔtopB::kan into cells containing either the parE ts allele or its isogenic wild type (parE+), selected for tetracycline resistance, and screened for kanamycin resistance. If the topB deletion can be tolerated by the parE ts strain, then the percentage of tetracycline-resistant colonies that are also kanamycin resistant will be equal in both the parE ts and parE+ cells. Based upon their physical distance on the E. coli chromosome, these two markers should be transduced together ∼50% of the time.
When selecting for tetracycline resistance, we found that the parE+ recipients also acquired the kanamycin resistance gene with a frequency of about 50% (Fig. 2B, columns i, iii, v and vii) in four different strain backgrounds (W3110, C600, KD100 and LE234). In contrast, the parEts cells that were resistant to tetracycline were rarely (5/201) also resistant to kanamycin (Fig. 2B, column ii). Chi-square statistical analysis showed a significant difference (P < 0.0001) between ΔtopB transduction into the parE ts strain compared with the parE+ strain at 30°C (Fig. 2C). The resultant 5/201 ΔtopB parE ts cells were dead at 37°C, whereas ΔtopB and parE ts cells were fully viable at 37°C as measured by plating efficiency assays (data not shown). Therefore, ΔtopB and parE ts are synthetically lethal. This was surprising because ΔtopB alone is viable and heretofore the robust parE ts mutant strain (Kato et al., 1990) had appeared to retain wild-type topoisomerase IV activity at 30°C and 37°C (Adams et al., 1992; Zechiedrich and Cozzarelli, 1995; Zechiedrich et al., 2000; Deibler et al., 2001). The synthetic lethal phenotype of ΔtopB is not restricted to the mutant parE subunit of topoisomerase IV, but is also seen with the parC ts allele (Fig. 2B, column iv). We were unable to combine ΔtopB with parC ts (P < 0.0001) (Fig. 2C). Thus, the removal of topB in either topoisomerase IV temperature-sensitive mutant background is lethal.
Although we had not previously observed any defect in transduction in the parE ts mutant (Zechiedrich and Cozzarelli, 1995; and see below), it was possible that this particular set of double antibiotic resistance markers was affected by the mutant allele of parE. Therefore, we tested the cotransduction efficiency of a pair of transposon markers in approximately the same chromosomal position as topB by transducing Tn10kan Tn10tet from strain CRL47 into parE ts and its isogenic wild-type strain. The transduction and cotransduction frequencies were equal (data not shown).
It was shown previously that the overproduction of topoisomerase III rescues the temperature-sensitivity of topoisomerase IV mutant strains (Nurse et al., 2003). We utilized this finding to explore potential genetic interactions with topoisomerase III. We used IPTG to express topB from plasmid pPH1243 (Broccoli et al., 2000), which rescued both parC ts and parE ts cells at their respective non-permissive temperatures, 37°C and 42°C, and did not affect the viability of the parC+(C600), parE+ (W3110) or the topB null strain (in W3110). topB overexpression from pPH1243 allowed construction of the ΔtopB parE ts strain, which was now viable at 37°C (used below) and even 42°C (data not shown) in the presence of IPTG. In the absence of the inducer, the double topoisomerase mutant cells were not viable at either temperature (Fig. 5). That the ΔtopB parE ts strain was viable at 30°C in the absence of IPTG likely results from the well known leaky expression from the trc promoter. A lower copy number plasmid, pPH1095, containing the topB gene under its natural promoter (Broccoli et al., 2000), rescued 93 ± 8.5% of parE ts colonies at 42°C (data not shown).
Gyrase and topoisomerase IV are the type-2 topoisomerases in E. coli. These enzymes can perform many of the same reactions in a test tube (Steck and Drlica, 1984; Marians, 1987; Ullsperger and Cozzarelli, 1996), yet they perform distinct roles in the cell (Deibler et al., 2001). We tested whether temperature-sensitive gyrase mutations could coexist with ΔtopB using the same genetic linkage experiment as above. We found no difference (P = 0.4539, Fig. 2C) in the frequency of transducing ΔtopB into gyrB+ (KD100) or gyrB225ts (GP100) (Steck and Drlica, 1984) at the permissive temperature, 30°C (Fig. 2B, compare vi with v). There was also no difference in the viability or colony size of the ΔtopB gyrB225ts colonies compared with gyrB225ts colonies at 30°C, 37°C or 42°C (data not shown).
Recognizing that this particular gyrase mutant may not have impaired function at 30°C, we repeated the experiment with the gyrB134ts allele in strain LE316 (Orr et al., 1979), which exhibits a supercoiling deficiency at 30°C (data not shown). There was a significant decrease (P = 0.0053) in the percentage of gyrBts cells that also contained the topB null allele relative to gyrBts (Fig. 2B, compare column viii with vii). However, unlike what was seen in topoisomerase IV mutants, the overproduction of topoisomerase III, with IPTG concentrations ranging from 0.5 to 5 mM, did not rescue the gyrB134ts strain. In addition, there was no difference in the viability of the ΔtopB gyrB134ts double mutant strain at 30°C or 37°C (data not shown). Using Tn10tet linked to Tn10kan from strain CRL47 as described above, and we found a similar twofold decrease in the likelihood that gyrB134ts cells contained both markers compared with gyrB+ cells. We conclude that the gyrB134ts allele decreases general cotransduction and is not synthetically lethal with ΔtopB.
Effect of topB overexpression on the cell morphology of topoisomerase IV temperature-sensitive mutants
Originally, the parC ts and parE ts mutant strains were identified by a defect in DNA nucleoid partitioning and inviability at high temperatures (Hirota et al., 1968; Kato et al., 1988; Kato et al., 1990). These mutants also exhibit a septation defect, resulting in cells that are longer than normal. To try to understand how topoisomerase III rescues the conditional lethality of the topoisomerase IV mutants, we analysed the effect of topB overexpression on cell length and DNA partitioning in the parC ts and parEts compared with isogenic wild-type strains. Shown in Fig. 3 are light (left) and fluorescence (right) micrographs of DAPI stained cells. IPTG-induced overexpression of topB did not affect the morphology of the wild-type strains (Fig. 3A a–d and data not shown), as expected from previous experiments with arabinose-induced topB overexpression (Zhu et al., 2001).
The parC ts mutant exhibited significant DNA segregation and cell septation defects, even at the permissive temperature (30°C) (Fig. 3A e and f), with ∼50% of the population abnormal (Fig. 3B, compare panel c with panel a). At 37°C, the non-permissive temperature, most cells were filamented and contained non-partitioned DNA (Fig. 3A g, h and Fig. 3B d). The overexpression of topB had a dramatic effect on the parC ts mutant at 30°C, returning its morphology nearly to that of wild type (Fig. 3A i, k and Fig. 3B, compare panel e with a) and rescuing nucleoid partitioning in ∼75% of the cells. At 37°C (Fig. 3A l, m and 3B f), the overexpression of topB increased the amount of partitioned DNA from almost none (Fig. 3A e–h and Fig. 3B d) to ∼45% (Fig. 3A i–m and Fig. 3B f); whereas, ∼90% of the wild-type cells were partitioned at 37°C. However, topB overexpression did not rescue the filamentation defect (Fig. 3B compare d with f). The parE ts cells were longer than their isogenic wild type at 30°C and contained what appeared to be discrete, partitioned nucleoids (Fig. 3A n, o). Again, the overexpression of topB did not rescue the filamentation phenotype of the parE ts cells (Fig. 3A, compare panel n with r). At 42°C, the parE ts cells exhibited a severe DNA segregation defect, similar to the parC ts cells, which was partially rescued by the overexpression of topB (Fig. 3A t, u). Thus, over-production of topoisomerase III partially compensates for the DNA partitioning defect, but not for the filamentation defect, observed in the topoisomerase IV temperature-sensitive mutants. This is in agreement with the results reported by Nurse et al. (2003) for the parE ts strain using an arabinose-inducible overexpression system.
RecQ is required for the synthetic lethality of the ΔtopB parEts double mutant
In S. cerevisiae and S. pombe, deletion of the genes that encode the respective RecQ homologues rescued the topoisomerase III null phenotypes (Gangloff et al., 1994; Goodwin et al., 1999; Maftahi et al., 1999). Based upon these results and the finding that purified RecQ and topoisomerase III can catenate double-stranded DNA (Harmon et al., 1999), it was hypothesized that RecQ helicases create a DNA substrate that topoisomerase III resolves (Gangloff et al., 1994; Wu et al., 1999). Therefore, we examined the effect of RecQ on the synthetic lethal phenotype of the ΔtopB parE ts double mutant.
We first constructed the ΔrecQ::cat parE ts and ΔrecQ::cat parE+ strains, for which the transduction efficiency and growth were identical (data not shown). Then, using the linkage efficiency assay depicted in Fig. 2A, we asked whether the absence of RecQ would allow the topB deletion to be combined with the parE ts allele at 30°C. Removal of the recQ gene increased the cotransduction efficiency of ΔtopB into the parE ts strain ∼7.5 times (compare Fig. 2B, column ii with Fig. 4A, column ii) (P ≤ 0.0001) (Fig. 4B). Thus, the removal of recQ partially rescues the synthetic lethal phenotype of the ΔtopB parE ts mutant cells.
Removing another 3′→5′ helicase, helicase IV (Mendonca et al., 1995), did not rescue the synthetic lethality of the ΔtopB parE ts strain (compare Fig. 4A iv with Fig. 2B ii) because the combination of the two topoisomerase mutations was still lethal. There was no detectable difference in transduction efficiency or strain growth between parE ts and ΔhelD parE ts strains (data not shown).
Effect of recA on ΔtopB, parEts and ΔtopB parEts viability
The filamentation and chromosome segregation defects of the ΔtopAΔtopB double mutant were suppressed by deleting the gene encoding the strand-invasion protein, RecA (Zhu et al., 2001). Furthermore, the prevention of meiotic recombination in S. cerevisiae was able to rescue the sporulation defect seen in Δtop3 cells (Gangloff et al., 1999). It has been suggested that topoisomerase III, in conjunction with Sgs1, suppresses the production of crossover recombinants during double strand break repair in S. cerevisiae (Ira et al., 2003). Thus, it is possible that topoisomerase III could play a role in RecA-mediated recombination (Wang et al., 1990; Bailis et al., 1992; Wu et al., 1999; Zhu et al., 2001; Wang, 2002).
To determine whether RecA plays a role in the synthetic lethality of the ΔtopB parE ts, we measured the viability of ΔtopB parE ts in the absence recA. Because plasmid pPH1243 allowed the construction of the ΔtopB parE ts double mutant strain, we transduced the ΔrecA::Tn10 allele into ΔtopB parE ts, ΔtopB, parE ts and their isogenic wild-type strain in the presence of pPH1243 at 30°C. There was no difference in transduction frequency among these strains. The number of colonies formed on Luria–Bertani (LB) agar plates containing ampicillin with (right) or without (left) 0.5 mM IPTG was divided by the number of cells spread as determined by cell counts using a Petroff-Hauser counter. The viability of the various otherwise isogenic strains is shown (Fig. 5). All strains lacking the recA gene were less viable compared with recA+ strains (Fig. 5). Despite this overall reduction in viability, the ΔtopBΔrecA parE ts triple mutant was viable at 37°C in the absence of IPTG-induced topB expression, whereas the ΔtopB parE ts strain did not grow under these conditions (P = 0.0095). Thus the deletion of recA is able to partially rescue the ΔtopB parE ts double mutant at 37°C.
ΔrecA cells are deficient in both the SOS response and in homologous recombination. Thus, we utilized a lexA mutant allele, lexA3, that produces a non-cleavable LexA protein to distinguish which RecA function was involved. Cells that contain lexA3 are unable to derepress the LexA/SOS regulon in response to DNA damage (Little et al., 1980). Employing P1 phage, we moved lexA3 into the ΔtopB parE ts strain (containing pPH1243) in the presence of 0.5 mM IPTG and measured the number of colonies/viable cells. Unlike the ΔrecAΔtopB parEts strain, the lexA3ΔtopB parE ts strain was not viable at 37°C without IPTG (data not shown). Therefore, lexA3 does not rescue the synthetic lethality of the ΔtopB parE ts mutant. We conclude that the synthetic lethality of the ΔtopB parE ts strain does not involve LexA regulon induction. These results support the idea that the recombination function of RecA is required. Based upon these data, it appears likely that preventing the formation of recombination intermediates allows the survival of topoisomerase III/IV double mutant cells.
There was no significant difference in viability between any of the ΔrecA-containing strains at 30°C. At this temperature, the addition of IPTG to the medium increased the viability of the ΔrecA, ΔtopBΔrecA and ΔtopBΔrecA parE ts strains. In general, at all temperatures tested, the overexpression of topoisomerase III had a positive influence on cell viability in cells lacking recA, indicating that the increase in topoisomerase III levels was somehow advantageous.
At 37°C, in the absence of IPTG, wild-type, ΔtopB, parEts and ΔrecA strains were as viable as at 30°C. The ΔtopB parE ts double mutant was not viable, and the ΔtopBΔrecA strain was more viable compared with 30°C. This strain is also more viable than the ΔrecA strain (P = 0.041). The viability of the ΔrecA parE ts and ΔtopBΔrecA parE ts strains was decreased; however, these strains were viable in the absence of IPTG induced topB expression. This suggests that the deletion of recA, although somewhat deleterious to the parE ts strain (as suggested by the decreased viability of the ΔrecA parE ts strain compared with the ΔrecA strain), obviates the need for topB.
At 42°C, in the absence of IPTG, none of the strains containing the parE ts allele grew. The rest of the strains showed no change in viability compared with 37°C. In addition, topB overexpression rescued the temperature-sensitivity of all the strains with the parE ts allele. These strains were just as viable as they were at 37°C + IPTG. These results were unexpected in the ΔtopBΔrecA parE ts and ΔrecA parE ts strains. Because deleting recA allowed ΔtopB parE ts cells to live at 37°C, it appeared that topB was acting down-stream of recA in the parE ts mutants. Based upon this, we expected that the overexpression of topB should have no effect on cells lacking recA.
Transduction of ΔtopB into ruvC53 or ΔruvABC
Human topoisomerase IIIα has been shown to cleave a synthetic double Holliday junction with the help of BLM helicase (Wu and Hickson, 2003). If topoisomerase III unlinks such recombination intermediates in vivo, then the cell should not tolerate the removal of both the major Holliday junction resolving enzyme, RuvC (Dunderdale et al., 1991; Iwasaki et al., 1991; reviewed in West, 1997), and topoisomerase III. ruvC+ and ruvC53 (a loss of function mutant, Sharples and Lloyd, 1993) strains (in the 594 genetic background) were transformed with pPH1243. ΔtopB was transduced into these cells in either the absence or presence of IPTG. ΔtopB::kan could be transduced into ruvC+, but not ruvC53 cells, in the absence of IPTG at 30°C (Fig. 6A). Next, we employed the linkage assay schematized in Fig. 2A. ΔtopB::kan Tn10tet was transduced into the ruvC53 mutant strain and its isogenic (SMR4562) wild-type strain, selecting for tetracycline resistance. The percentage of ruvC+ cells that contained both kanamycin and tetracycline resistance markers was 50% (Fig. 6B). This percentage was reduced by approximately threefold in the strain containing the ruvC53 allele and approximately twofold in the ΔruvABC mutant (Fig. 6B; P = 0.0001 and P = 0.0295 respectively). These data suggest that the removal of both topoisomerase III and RuvC is detrimental to cells.
Throughout our experiments, we found a reduced number of transductants in cells containing the ruvC53 allele compared with ruvC+ cells, which agrees with previous results (Lloyd, 1991). It was possible that the role of RuvC in resolving the recombination intermediates that arise during transduction, or the chromosomal location of ruvC (41.9′) could account for the reduced efficiency of ΔtopB::kan cotransduction with Tn10tet. To test both of these possibilities, we performed linkage efficiency assays using three different pairs of linked transposons (Table 1). We tested the cotransduction frequencies of transposon pairs from strains CRL45 (with the transposons located at 53.7′ and 54.1′), CRL46 (86.4′ and 86.8′) and CRL47 (39.5′ and 40.3′) (topB is located at 39.7′) and found no difference between ruvC+ and ruvC53 containing strains (Table 1). Because the probability of transducing both markers was the same, neither the chromosomal position of ruvC nor its role in transduction was responsible for the bias against ΔtopB::kan cotransduction into ruvC53 mutant strains.
Table 1. Cotransduction efficiencya in ruvC+ and ruvC53 strains.
ruvC53 mutants are characterized by an increased sensitivity to DNA damaging agents, as well as a decrease in transduction efficiency (Lloyd, 1991). If topoisomerase III acts in a pathway that provides an alternative route of Holliday junction disentangling to the RuvABC pathway, then the ΔtopB ruvC53 double mutant strain should be even more sensitive to DNA damage and further reduced in transduction efficiency than the ruvC53 mutant.
We measured the UV sensitivity of two different ΔtopB ruvC53 mutants from the cotransduction assay shown in Fig. 6B compared with an isogenic set of strains in an FC40 genetic background. Both double mutants were up to ∼20 times more sensitive to UV irradiation than the cells containing the ruvC53 allele alone (Fig. 7A). ΔtopB cells displayed the same sensitivity to UV light as wild-type cells (Fig. 7A). We repeated these experiments in an MG1655 background and found identical results (data not shown).
As shown previously, a plasmid encoding the ruvABC genes is able to complement a ruvABC null strain (Seigneur et al., 1998; Flores et al., 2001; Grompone et al., 2002). This plasmid, pGB ruvABC, also complemented the UV sensitivity of the ruvC53 strain (Fig. 7B). The overexpression of ruvABC rescued the UV sensitive phenotype of ΔtopB ruvC53 cells to the wild-type level. The vector, pGB2, had no effect.
We measured the effect of overexpressing topB on the UV hypersensitive phenotype of the ΔtopB ruvC53 mutants. Plasmid pTBE302 contains topB under the control of an arabinose-inducible promoter (Zhu et al., 2001). topB overexpression had no effect on the UV sensitivity of the ruvC53 mutant in either the presence or absence of 0.2% arabinose (Fig. 7C). However, when topB was induced in the ΔtopB ruvC53 mutant, the tolerance to UV irradiation was increased to that of the ruvC53 mutant. This was not seen in the absence of arabinose, nor with vector (pBAD30) alone (Fig. 7C, first three panels).
In addition, the ΔtopB ruvC53 mutant was approximately five times more sensitive to mitomycin C (0.5 µg ml−1) than the ruvC53 mutant (Fig. 8A). The ΔtopB strain was the same as wild type. The transduction efficiency of the ΔtopB ruvC53 (CRL76) was also approximately twofold decreased relative to the ruvC53 (CRL75) strain (Fig. 8B); a Poisson statistical analysis showed these values to be distinct with a confidence limit of 95%.
The hypersensitivity to DNA damaging agents, as well as the decreased transduction efficiency of the ΔtopB ruvC53 double mutant strain compared with the ruvC53 strain suggest that topoisomerase III is involved in processing recombination intermediates in an alternate pathway to RuvC.
Effect of ruvABC overexpression on topoisomerase double mutants
RuvABC might be expected to prevent the synthetic lethality of the topoisomerase III topoisomerase IV double mutant, but this rescue would require increased cellular levels of ruvABC. We tested whether an increase in RuvABC levels would allow the construction of the ΔtopB parE ts or ΔtopB parC ts mutants. The cotransduction efficiency assay described in Fig. 2A was carried out in parE+, parE ts, parC+ and parC ts containing strains that had been transformed with either pGBruvABC or pGB2 (Fig. 9). In general, the frequency of ΔtopB::kan cotransduction into wild-type strains was reduced in the presence of pGBruvABC, compared with pGB2 alone or no plasmid (Fig. 2B). Perhaps ruvABC overexpression increases RuvC-mediated DNA cleavage to yield smaller chromosomal replacements by P1 transduction.
Despite the resulting decrease in the transduction of both markers, the overexpression of ruvABC from pGBruvABC allowed the cotransduction of ΔtopB into the parE ts strain at the same efficiency as the parE + strain (Fig. 9A). ruvABC overexpression also allowed the construction of the ΔtopB parC ts strain (Fig. 9A), and the frequency of ΔtopB cotransduction was similar to the parC+ strain. The presence of pGB2 (in the presence or absence of spectinomycin) itself partially relieved the synthetic lethality of ΔtopB parE ts; however, there was still a statistically significant difference in cotransduction efficiency between parE ts and parE+ cells. pGB2 had no effect on the cotransduction of ΔtopB into the parC ts strain. No other plasmid tested (pBR322 or pACYC184) produced such an effect. The number of colony forming units in the ΔtopB parE ts strain harbouring pGBruvABC was twofold greater than in the ΔtopB parE ts pGB2 strain at both 30°C and 37°C (data not shown).
Additional results are difficult to resolve with a model for topoisomerase III in decatenation. First, if at normal cellular levels, topoisomerase III unlinks precatenanes, it does so poorly because it leaves 90% of replicated plasmid DNA as closed circular catenane links that only topoisomerase IV can resolve (Zechiedrich and Cozzarelli, 1995). Indeed, the overproduction of topoisomerase III at any level cannot rescue fully a parE null mutant strain (Nurse et al., 2003). On the cellular level, the overproduction of topoisomerase III rescued the partition defect in only ∼45% of the parEts and parCts cells and did not rescue the filamentation phenotype at the non-permissive temperature (Fig. 3B and Nurse et al., 2003). These results suggest that topoisomerase III cannot efficiently perform the essential function of topoisomerase IV, which is either a subtle, but important, role in DNA supercoiling maintenance (Zechiedrich et al., 2000), the unlinking of daughter chromosomes at the end of replication (Adams et al., 1992; Zechiedrich and Cozzarelli, 1995) or unknotting DNA (Deibler et al., 2001). Second, we find that the removal of topoisomerase III does not cause catenated plasmid replication intermediates to accumulate (data not shown). Even when the activity of the main cellular decatenating enzyme, topoisomerase IV, is absent, the removal of topoisomerase III activity does not cause an additional accumulation of catenanes when compared with the removal of only topoisomerase IV (Fig. 1). Third, topoisomerase III does not have the ability to preferential unlink rather than link DNA that the type-2 topoisomerases have (Rybenkov et al., 1997). Therefore, topoisomerase III is as likely to link as to unlink DNA. Fourth, unlinking of replicated chromosomal DNA must occur in all living cells, yet topoisomerase III in a parEts background, is dispensable at the permissive temperature when recA or recQ is eliminated (Figs 4 and 5). Finally, it is difficult to see how a role in decatenation can result in the phenotypes of increased mutation and genome rearrangement observed for topoisomerase III mutant cells.
In general, topoisomerases are in high abundance. In E. coli, topoisomerase I, gyrase and topoisomerase IV are thought to range from 500 to 10 000 mol cell−1. Topoisomerase III is much less abundant (1–10 mol cell−1; DiGate and Marians, 1989). The topB gene encodes rare codons; this might explain the low protein abundance because the charged tRNAs for these codons are rate limiting (DiGate and Marians, 1989). The codon usage of the other topoisomerases is skewed to the abundant tRNAs (Table 2). The mean codon bias index (CBI) (Bennetzen and Hall, 1982) for the five known topoisomerase genes was 0.403 ± 0.094 (a value of zero would indicate random codon choice; a value of one would indicate preferred codon choice). Proteins involved in DNA repair are often in low abundance (discussed in Friedberg et al., 1995). Excluding recA (0.558), the mean CBI for E. coli DNA repair and recombination genes (see complete list in Experimental procedures) was 0.175 ± 0.079. The values for topoisomerases and repair genes were distinct (P < 0.001) by the student's t-test. Topoisomerase III, with a CBI = 0.192, is as likely to encode rare codons as preferred codons, and more closely resembles the DNA repair and recombination genes (Table 2).
We hypothesize that topoisomerase III acts in a separate pathway involved in processing recombination intermediates (Fig. 10A). We cannot envision a role for topoisomerase III in resolving single Holliday junctions. Indeed, synthetic single Holliday junctions are not cleaved by human topoisomerase IIIα (Wu and Hickson, 2003). Instead, we show a model in which RecQ helicase migrates two Holliday junctions toward each other, resulting in a build up of positive DNA supercoils (Fig. 10B ii). Based upon previous biochemical studies, topoisomerase IV could remove the positive supercoils (Crisona et al., 2000) (or DNA gyrase could introduce negative DNA supercoils to prevent the formation) to allow additional branch migration of the Holliday junctions towards each other until they meet (Fig. 10B ii–iii). The Holliday junctions would most likely converge in a parallel conformation (Schwacha and Kleckner, 1995). This would result in the formation of a structure that has three single-stranded links, indicated by asterisks, two initial (Fig. 10B iii) and one that would remain when the first two were unlinked (Fig. 10B iv). Topoisomerase III might unlink this substrate, which is similar to its preferred in vitro substrate (DiGate and Marians, 1988). In support of this, human topoisomerase IIIα was shown to cleave a synthetic double Holliday junction in the presence of BLM helicase (Wu and Hickson, 2003). Although not perhaps its normal reaction, when overexpressed (Wallis et al., 1989) or when topoisomerase III is removed (Broccoli et al., 2000; Zhu et al., 2001), topoisomerase I may also unlink the merged Holliday junctions through a separate pathway. Topoisomerase-mediated resolution of Holliday junctions would always result in non-crossover products (Fig. 10B v). A reasonable prediction of this model is that the removal of topB from any organism would result in an increase in chromosomal rearrangements. Over generations, this could lead to the shuffling of the genome.
That a purified topoisomerase can cleave Holliday junctions in a test tube has been shown (Sekiguchi et al., 1996; Palaniyar et al., 1999; Wu and Hickson, 2003). Topoisomerase-mediated cleavage in all of these studies was measured by incubating a synthetic Holliday junction with topoisomerase and a divalent cation. Cleavage is trapped by adding SDS to denature the topoisomerase. Proteinase K is added to digest the covalently bound topoisomerase. It is not clear in any of these cases whether a topoisomerase would actually finish the resolution reaction: break a DNA strand, pass a DNA strand through the break, and religate the break. This is required for Holliday junction resolution in vivo. Merged Holliday junctions form a hemi-catenane. Hemi-catenanes have been shown previously to be excellent substrates for topoisomerase III-mediated DNA strand passage.
Branch migration of Holliday junctions by helicases has been seen in T4 and T7 bacteriophages (Kong et al., 1997). In addition to RuvAB, the E. coli helicases DnaB, RecG and RecQ have also shown branch migration activity in vitro (Harmon and Kowalczykowski, 1998; Bolt and Lloyd, 2002; Kaplan and O’Donnell, 2002; McGlynn and Lloyd, 2002). The actions of a helicase could disrupt D-loops to prevent the formation of Holliday junctions; however, it is unclear how branch migration by a helicase could process an existing Holliday junction to unlink the DNA strands. Alternately, RecQ might unwind various DNA structures (reviewed in Bennett and Keck, 2004) into a substrate that topoisomerase III recognizes and can disentangle, or RecQ-mediated branch migration of two Holliday junctions can lead to the formation of a DNA structure similar to a hemi-catenane that topoisomerase III unlinks as shown in Fig. 10B.
Perhaps, in the absence of RecQ, the topoisomerase-mediated disentangling mode of Holliday junction resolution is either not used or is inefficient. This could explain why the deletion of the recQ gene can relieve partly the synthetic lethality of the ΔtopB parE ts double mutant (Fig. 4) and may explain why the deletion of the RecQ homologues, rqh1 or sgs1, in yeast relieves the defects of topoisomerase III mutations (Gangloff et al., 1994; Goodwin et al., 1999; Maftahi et al., 1999; Oh et al., 2002). The BLM helicase has been shown to stimulate the activity of human topoisomerase IIIα (Wu and Hickson, 2002).
Our data show that RuvABC can act on recombination intermediates normally metabolized by the topoisomerase pathway, as evidenced by the rescue of ΔtopB ruvC53 to wild-type levels by ruvABC overexpression (Fig. 7B). Whether RuvAB can act after RecQ, as denoted as an arrow in Fig. 10A or competes more readily with RecQ for a recombination intermediate is unknown. In contrast, topoisomerase III cannot act upon the DNA substrate once processed by RuvAB for RuvC because topoisomerase III overproduction did not have any effect on ruvC53 mutant cells (Fig. 7C). What decides the fate of the Holliday junction is unknown; relative binding affinities, enzyme abundance, and preferred sites of enzyme activity may all come into play. It is also possible that normally the topoisomerase pathway might be used exclusively for resolving double Holliday junctions whereas RuvC might be preferred for the resolution of single Holliday junctions.
Because of RuvC, topoisomerase I, RecG, and sometimes RusA, the absence of topoisomerase III is normally tolerated by E. coli. However, when topoisomerase IV is mutated and RecA and RecQ are present, the removal of topoisomerase III may lead to cell death because the chromosomes become intertwined and trapped in the recombination intermediate structures. Failure to unlink the DNA would prevent chromosomal segregation, thus impeding cell division and resulting in a partition defective phenotype. A similar partition defective phenotype was seen in S. pombe and S cerevisiae cells that lack topoisomerase III (Gangloff et al., 1999; Maftahi et al., 1999). Interestingly, similar cell morphology also resulted from the overproduction of the RecQ homologue from S. pombe, Rqh1 (Oh et al., 2002); more Rqh1 might lead to more intermediates funnelled into the topoisomerase pathway than those enzymes can process.
Our data suggest that the parEts and parC ts alleles cause an increase in DNA lesions that must be processed by topoisomerase III or RuvC. The topoisomerase IV mutants at their non-permissive temperatures result in: (i) increased negative supercoiling of the DNA (Zechiedrich et al., 2000), which might explain; (ii) increase in cell growth and DNA replication rates; (iii) increased catenation (Adams et al., 1992; Zechiedrich and Cozzarelli, 1995; Zechiedrich et al., 1997) and knotting (Deibler et al., 2001). Any of these could lead to an increase in DNA damage, which in turn might overwhelm topoisomerase III and RuvC, both of which, as discussed above, are not abundant in the cell. Indeed, even at 30°C, the SOS response is activated in parE ts strains (data not shown), indicating the presence of DNA damage. In addition, preliminary results show that ruvC53 parE10 cells are more sensitive to UV light than ruvC53 single mutants and that parE10 cells are also more sensitive to UV irradiation than wild-type cells at higher (but still permissive) temperatures (C. R. Lopez, J. Galloway and E. L. Zechiedrich, unpubl. results). We show here that recA parE10 cells are less viable than recA cells (Fig. 5B). This has been reported previously (Grompone et al., 2004). Grompone et al. also found that a priA parE10 double mutant displays a synthetic lethal phenotype. This suggests that the parE10 allele causes a hindrance to DNA replication. These data all point to parE10 causing DNA damage.
There also appears to be a recA independent repair pathway that is involved in dealing with the problems caused by the parE10 allele. Although the recA parE10 double mutant has reduced viability, it is still viable. If the tolerance of the damage caused by the parE10 allele involved only recA, then the recA parE10 would not be viable. The stochastic nature as to which repair pathway is chosen (McCool et al., 2004) would explain why the deletion of recA rescues some of the topB parE10 mutants, but not all of them. We think that it is also possible that topA mutants could lead to DNA damage that requires recA, topB and possibly ruvC for repair.
The functional complexity of the roles of proteins involved in DNA segregation, DNA repair, DNA recombination, and the modulation of DNA topology exemplifies the adaptive nature of bacteria. Rather than these enzymes having redundant functions, each has a specific role. Multiple enzymes functioning in multiple alternative pathways probably allow cells to adapt to a variety of conditions.
Chemicals and reagents
All antibiotics, DNAseI, RNAseA and DAPI were from Sigma. Supercoiled DNA and 1KB plus DNA ladders were from Invitrogen. Proteinase K was from Boeringer Mannheim. Nylon Zeta-Probe Blotting Membranes were from Bio-Rad. IPTG and formaldehyde were from Fisher Scientific. [α-32P] dCTP and Megaprime DNA Labelling System were purchased from Amersham Pharmacia.
Bacterial allele construction and plasmids
ΔrecQ::cat contains the chloramphenicol acetyltransferase (cat) gene flanked by FRT sites replacing codon 32–567 of the recQ gene. This allele was constructed by linear replacement (Datsenko and Wanner, 2000). The primers used to generate the linear fragment were recQ1: 5′-TAAA CAGGTTTTACAAGAAACCTTTGGCTACCAACAGTTTCGC CCCGGCCGTGTAGGCTGGAGCTGCTTC and recQ2: 5′-CATGCGCACGAATCAGCGCCATAAACGGTTTGCCAAAGCGTTCCAGCTTGCGTATGAATATCCTCCTTAGT where the underlined sequence corresponds to recQ coding sequence and the bold sequence corresponds to plasmid pKD3 (Datsenko and Wanner, 2000). Plasmid pPH1243 (Broccoli et al., 2000) contains the topB gene with 200 bp of 5′ upstream sequence under the control of the trc promoter and is derived from the pTRC99a vector (provided by Marc Drölet, University of Montreal, Canada). pTBE302 (Zhu et al., 2001) contains the topB gene under the control of an arabinose-inducible promoter. It is based on the pBAD30 vector. pGBruvABC (Flores et al., 2001) contains the ruvAB and ruvC genes cloned into the low copy number vector pGB2 (generously provided by Benedicte Michel, Genetique Microbienne, Institute National de la Recherche Agronomique, Jouy en Josas, France). Plasmid pJB3.5d has been described previously (Bliska and Cozzarelli, 1987).
All bacterial strains used are listed in Table 3. P1 transductions were performed with P1virA according to Miller (Miller, 1972). To make strain CRL6, first the Tn10tet allele from strain CAG18465 was transduced into strain QZ103 (ΔtopB:: kan). The resulting ΔtopB::kan Tn10tet alleles from this strain were transduced (selecting for tetracycline resistance) into the following strains: W3110, C600, ParE10, ParC1215, LE234, LE316, CRL31, CRL32, SAR3, SAR4, SMR6047 and SMR4562. These transductants were then screened for kanamycin resistance at 30°C to determine whether the ΔtopB allele was cotransduced (all subsequent experiments with these strains were performed at 30°C), and we verified the presence of the original mutation in these strains by measuring temperature sensitivity at 42°C (ParE10, ParC1215 and LE316) or UV sensitivity (SMR6047).
To construct CRL22 and CRL36, the ΔtopB::kan Tn10tet alleles from CRL6 and the ΔtopB::kan allele from QZ103 were transduced into ParE10 [pPH1243]parE10 +[pPH1243] in the presence of 0.5 mM IPTG and 100 µg ml−1 of ampicillin in addition to the selection agent (tetracycline or kanamycin respectively). This allowed the ΔtopB allele to coexist with the parE10 allele at 30°C. Tetracycline and kanamycin resistance, and temperature sensitivity were verified.
Strain LZ2369 was made by transducing the ΔtopB::kan allele from QZ103 into a strain LZ5 (gyrAL83). To construct strains CRL38 and CRL40, the ΔrecA and lexA3 alleles were transduced into strains LZ2549 (W3110 + pPH1243) and CRL36 (ΔtopB:: kan parE10 + pPH1243). CRL45, CRL46 and CRL47 were constructed by transducing Tn10 from strains CAG18468, CAG18491 and CAG 18465 into strains CAG18522, CAG18557 and CAG18518. Kanamycin and tetracycline resistance were verified. CRL2 [P1(QZ103) X ParE10] and CRL35 [P1(CRL6) X ParE10] were constructed as above. Throughout the text we refer to these strains by their relevant phenotype.
Linkage efficiency assay
Tn10 was linked to ΔtopB::kan, as described above. ΔtopB::kan Tn10 from donor strain CRL6 was then transduced into an isogenic set of mutants, selecting for tetracycline resistance. Transductants were given up to 160 h to appear at 30°C. The subsequent transductants were then screened for kanamycin resistance. A contingency chi-square statistical analysis was performed to determine whether the difference in linkage values was significant. P-values are given. A-value ≤ 0.05 was considered significant, with a 95% confidence level. Transduction frequency was identical in parE+ and parEts strains when selecting for tetracycline resistance (CRL6) or chloramphenicol resistance (SMR6201 and SWM1001).
Tn10 and Tn10kan were transduced into ruvC + (SMR4562) and ruvC53 (SMR6047) strains. The cotransduction frequency was ∼32% from CRL45, 67% from CRL46 and 17% from CRL47.
More ΔtopB parEts colonies do arise if given more recovery time to appear after transduction; however, greater incubation time (∼150 h) does not result in a greater proportion of ΔtopB parE ts colonies (data not shown). Even given more time after the transduction to grow, none of the late arising tetracycline-resistant colonies were also kanamycin resistant in the parC ts background.
ParE10 and ParC1215 strains (and their isogenic wild-type strains) were transformed with pGBruvABC or empty vector (pGB2). ΔtopB::kan Tn10 from CRL6 was then transduced into these strains as above.
Plating efficiency assays
Cultures were grown to mid-logarithmic phase (A600 = 0.4) in LB medium containing 100 µg ml−1 ampicillin and 0.5 mM IPTG. A 50 µl aliquot of a 10−4 dilution of the culture was spread onto two sets of LB agar plates containing 100 µg ml−1 ampicillin with and without 0.5 mM IPTG. The plates were incubated up to 72 h at 30°C, 37°C or 42°C. The number of colonies on each plate was counted after each overnight incubation.
WT (C600 and W3110), parCts and parEts mutant strains containing plasmid pPH1243 were grown in LB medium with 100 µg ml−1 ampicillin and either 0.5 mM or no IPTG at 30°C. Mid-logarithmic phase cells (A600 = 0.4) were diluted to A600 = 0.01 and shifted to 37°C (parCts and C600) or 43°C (parEts and W3110) and allowed to grow for two additional hours. Cells were prepared for microscopy essentially as described in (Pringle et al., 1991). Briefly, cells were fixed by the addition of formaldehyde (37% w/v stock) to 5% w/v and incubated for at least 1 h at 25°C. Cells were then washed with 0.1 M KH2PO4/K2HPO4 (pH 7.0), placed on poly-l-lysine treated glass slides and overlaid with mounting medium that contained 225 ng ml−1 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI). A Zeiss Axioplan fluorescence microscope was used. Images were captured using a Photometrics CoolSnap HQ camera with MetaVue v.6r6 software.
Analysis of catenane accumulation by high-resolution gel electrophoresis
Strains LZ5 and LZ2369, containing plasmid pJB3.5d, were grown at 37°C in LB medium containing 100 µg ml−1 ampicillin. At A600 = 0.4, a 5.7 ml aliquot was taken (0 min). Norfloxacin was added to a final concentration of 60 µM to inhibit topoisomerase IV activity (Khodursky et al., 1995). Additional 5.7 ml aliquots were taken 20, 30, 40 and 60 min after addition of drug. Plasmid DNA was isolated by alkaline lysis immediately after taking the aliquots. The DNA was nicked with DNAseI, subjected to high-resolution gel electrophoresis (Sundin and Varshavsky, 1981), transferred to nylon membranes, and probed with [α-32P] dCTP labelled linearized pJB3.5d DNA.
Strains W3110, ParE10 and CRL3, containing pBR322, were grown at 30°C in LB medium containing ampicillin (100 µg ml−1). Upon reaching A600 = 0.4, a 5.7 ml aliquot was taken (0 min) and the cultures were shifted to 42°C. Aliquots of 5.7 ml were then taken at 15 and 30′ after this temperature shift. Plasmid DNA was isolated by alkaline lysis immediately after taking the aliquots. The DNA was nicked with DNAseI, subjected to high-resolution gel electrophoresis (Sundin and Varshavsky, 1981), transferred to nylon membranes, and probed with [α-32P] dCTP-labelled linearized pBR322 DNA.
Analysis of plasmid supercoiling levels
Strains LZ5, LZ6, LZ2369 and LZ2371, containing plasmid pJB3.5d, were grown at 37°C as described above and previously (Zechiedrich et al., 2000). At A600 = 0.4, a 5.7 ml aliquot was taken (0 min). Norfloxacin was added to a final concentration of 60 µM to inhibit topoisomerase IV activity (Khodursky et al., 1995). Additional 5.7 ml aliquots were taken 20, 30, 40 and 60 min after addition of drug. Plasmid DNA was isolated by alkaline lysis immediately after taking the aliquots. The DNA was subjected to gel electrophoresis on a 1.2% agarose gel containing either 2 µg ml−1 (for relaxed DNA) or 4 µg ml−1 (for supercoiled DNA) chloroquine (Zechiedrich et al., 2000). The DNA was then transferred to nylon membranes, and probed with [α-32P] dCTP labelled linearized pJB3.5d DNA.
UV sensitivity assay
Cells were grown to A600 = 0.4 in the LB broth at 30°C. After normalizing the amount of cells to A600 = 0.4, serial dilutions were performed. Aliquots (50 µl) of 10−4 dilutions were spread onto LB plates (unless otherwise noted). For cells containing the ruvC53 allele, a 50 µl aliquot of a 10−3 dilution was also plated. Cells were allowed to grow on the plates for 30 min at 24°C. The plates were then exposed to the indicated dosage of UV light using a CL-1000 ultraviolet crosslinker and incubated overnight at 30°C. Colonies were allowed to grow for 72 h at 30°C, followed by another 72 h at 24°C. Colonies were counted after each night of incubation. Each strain was tested a minimum of three times.
Mitomycin C sensitivity
Cells were grown to mid-logarithmic phase (A600 = 0.4). After normalization to 0.4, the cultures were serially diluted. Fifty microlitres of both a 10−3 and 10−4 dilution (for strains with the ruvC53 allele) or 50 µl of the 10−4 dilution were spread onto LB plates containing 0.5 µg ml−1 of mitomycin C, as well as an LB plate without drug. The plates were incubated for up to 72 h at 30°C.
A P1 lysate of CAG18468 was titered, and an MOI of 0.1 was used to infect the strains tested. There was no significant difference in cell density at A600 = 0.4 for WT, ruvC53, ΔtopB and ΔtopB ruvC53 strains. After normalizing 1 ml of culture to 0.4, the cells were pelleted and resuspended in 0.5 ml of 0.03 M MgSO4 0.015 M CaCl2. In addition to cells incubated with phage, a tube of each strain was also incubated without phage. After 20 min at 30°C, the cells were pelleted and resuspended in 150 µl of LB + sodium citrate. The cells were then allowed to recover for 45 min at 30°C. All 150 µl were spread onto LB-Tet plates. In addition to this, 25 µl of non-transduced cells were also spread onto a plain LB plate. After plating, cells were incubated overnight at 30°C. After recording the number of transductants, the plates were incubated an additional 72 h at the same temperature. Each strain was tested in duplicate. There appeared to be no significant difference in viability among the strains tested. The number of possible transductants = number of colonies present without selection agent × dilution used (10−5). Efficiency = number of transductants/number of possible transductants.
The viability of the indicated strains was measured by diluting an overnight culture in fresh LB broth and growing to A600 = 0.4. After normalization to 0.4, the culture was serially diluted in 1% NaCl. Fifty microlitres of a 10−3 (for recA strains) or 10−4 dilution were spread onto plates with or without 0.5 mM IPTG and incubated for up to 72 h at 30°C, 37°C or 42°C. The number of colonies was counted after each overnight incubation period. After plating the diluted cells, the 5 µl of the normalized culture were placed onto a Petroff-Hauser slide to determine the number of cells ml−1. Viability = colonies/cells plated.
Codon bias index calculation
Codon bias index was calculated according to Bennetzen and Hall (1982). The sequence of each gene in question was retrieved from the NCBI website and inserted into DNA Strider (version 1.01). DNA Strider then determined the frequency of each codon. From this information, we calculated: (i) number of preferred codons used (as defined by Bennetzen and Hall), (ii) number of preferred codons by random chance (sum of the total number of codons encoding each particular amino acid divided by the number of possible codons for each amino acid. For example, if there were 20 phenylalanines encoded, then 20 ÷ 2 (because there are two codons that encode phenylalanine) and so forth for each amino acid and (iii) the total number of codons (not counting Met, Asp or Trp).
CBI = (number of preferred codons − number of preferred codons by random chance) ÷ (total number of codons − number of codons encoding Met, Asp and Trp) −number of preferred codons by random chance).
In addition to the genes shown in Table 2, CBI was calculated for mutH (0.181), mutL (0.194), mutS (0.254), phrB (0.066), recB (0.210), recC (0.211), recD (0.112), recF (0.026), recJ (0.204), recO (0.081), and recR (0.224).
We dedicate this work to the memory of Dr Gisela Mosig. We thank her for critical reading of the manuscript and many stimulating discussions. We thank M. Drölet and B. Michel for plasmids, and C. A. Gross and S. W. Matson for strains. We are grateful to R. A. Hull and J. S. Butel for helpful advice, and A. Bacolla for comments on the manuscript. This work was supported by National Science Foundation Grant MCB-0090880, National Institutes of Health Grant R01-AI054830, a New Investigator Award in the Toxicological Sciences from the Burroughs Wellcome Fund, and a Curtis Hankamer Research Award to E. L. Z., a National Institutes of Health Grant R01-CA85777 to S. M. R, and a National Institutes of Health Grant R01-GM064022 to P. J. H. C. R. L and J. M. P. were supported by predoctoral fellowships from the Department of Defense Breast Cancer Research Program (DAMD17-02-1-0287) and (DAMD17-02-1-0289) respectively. S. Y. was supported by a National Research Service Award (CA09197) from the National Cancer Institute. R. W. D. was supported by a predoctoral fellowship from the Program in Mathematics and Molecular Biology at Florida State University with funding from the Burroughs Wellcome Fund. C. R. L., R. W. D., J. M. P and S. A. R. were supported, previously, by NIH T32 G08231.