Clamp Loader Processing Is Important during DNA Replication Stress

ABSTRACT

The DNA clamp loader is critical to the processivity of the DNA polymerase and coordinating synthesis on the leading and lagging strands. In bacteria, the major subunit of the clamp loader, DnaX, has two forms: the essential full-length τ form and shorter γ form. These are conserved across bacterial species, and three distinct mechanisms have been found to create them: ribosomal frameshift, transcriptional slippage, and, in Caulobacter crescentus, partial proteolysis. This conservation suggests that DnaX processing is evolutionarily important, but its role remains unknown. Here we find a bias against switching from expression of a wild-type dnaX to a nonprocessable τ-only allele in Caulobacter. Despite this bias, cells are able to adapt to the τ-only allele with little effect on growth or morphology and only minor defects during DNA damage. Motivated by transposon sequencing, we find that loss of the gene sidA in the τ-only strain slows growth and increases filamentation. Even in the absence of exogenous DNA damage treatment, the ΔsidA τ-only double mutant shows induction of and dependence on recA, likely due to a defect in resolution of DNA damage or replication fork stalling. We find that some of the phenotypes of the ΔsidA τ-only mutant can be complemented by expression of γ but that an overabundance of τ-only dnaX is also detrimental. The data presented here suggest that DnaX processing is important during resolution of DNA damage events during DNA replication stress. Although the presence of DnaX τ and γ forms is conserved across bacteria, different species have developed different mechanisms to make these forms. This conservation and independent evolution of mechanisms suggest that having two forms of DnaX is important. Despite having been discovered more than 30 years ago, the purpose of expressing both τ and γ is still unclear. Here, we present evidence that expressing two forms of DnaX and controlling the abundance and/or ratio of the forms are important during the resolution of DNA replication stress.

IMPORTANCE Though the presence of DnaX τ and γ forms is conserved across bacteria, different species have developed different mechanisms to make these forms. This conservation and independent evolution of mechanisms suggest that having two forms of DnaX is important. Despite having been discovered more than 30 years ago, the purpose of expressing both τ and γ is still unclear. Here, we present evidence that expressing two forms of DnaX and controlling the abundance and/or ratio of the forms is important during the resolution of DNA replication stress.

INTRODUCTION

The clamp loader is an essential component of the DNA replication machinery. It performs two critical functions: coordinating DNA synthesis on the leading and lagging strands (1) and loading the DNA clamp (2, 3), which increases the processivity of the DNA polymerase. There is also evidence that the clamp loader modulates the access to the replication fork of factors important for DNA damage bypass (4), DNA repair (5–7), DNA replication termination (8, 9), and replication restart (10) (Fig. 1).

FIG 1.

The bacterial clamp loader aids in resolution of replication fork stalling by allowing restart factors to access the replication fork. The DNA replication fork can encounter several types of replication blocks that cause replication fork stalling (e.g., DNA damage, collisions with transcription machinery, or physical tension caused by unwinding DNA). DnaX γ plays a role in allowing replication restart factors to access the replication fork (e.g., factors for DNA damage bypass [4], DNA repair [5–7], and DNA replication termination [8, 9]). This is critical for DNA replication restart and therefore cell survival. The clamp loader is shown here with three DnaX subunits (τ₂γ₁ in light gray), as well as the δ and δ′ subunits (darker grays).

The bacterial clamp loader is composed of five major subunits, three of which are DnaX (2, 11, 12). The single gene dnaX encodes two different forms of this subunit: the full-length τ form and truncated γ form. These forms were first discovered in Escherichia coli, where the γ form is created upon a ribosomal frameshift, which results in premature translation termination (13–15). While the two DnaX forms are conserved across many bacterial species, different bacteria create the γ form in different ways. For example, Thermus thermophilus creates a γ form upon RNA polymerase slippage, which results in premature transcriptional termination (16). Here, we study dnaX in the alphaproteobacterium Caulobacter crescentus, where a full-length τ is translated and partially proteolyzed by the AAA+ protease ClpXP to create two γ forms that lack the C terminus (17).

The conservation of the γ form across bacteria suggests that is important to cell survival; however, very little is known about its role. The DNA clamp loader complex binds the alpha subunits (DnaE) (18, 19) and helicase (DnaB) (20) of the DNA polymerase III holoenzyme to coordinate replication on the leading and lagging strands (1). Since the C-terminal domain of DnaX is responsible for binding the DnaE (18), only the full-length τ can perform this job. Thus, out of the three DnaX subunits in the clamp loader complex, at least two of these must be τ to maintain two polymerases for lagging and leading strand synthesis. There is some debate in the field about whether this replicative clamp loader contains τ₃ or τ₂γ (4, 21–23).

While the γ subunit is not essential in E. coli (24), strains lacking γ are more sensitive to DNA damage and show a lower rate of DNA damage-induced mutagenesis (4). From this previous work, it was suggested that E. coli γ plays a role in the DNA damage response, possibly by allowing the recruitment of error-prone DNA polymerases (4) (Fig. 1). It has also been hypothesized that a γ-only clamp loader might be responsible for unloading clamps (25), which would be also useful in resolving stalled replication forks by removing the DnaE-bound DNA clamp (Fig. 1). The elimination of γ from E. coli does not result in any observable phenotype in the absence of DNA damage (24). In contrast, our original characterization of Caulobacter crescentus suggested that γ was important even during normal growth (17).

Here, we investigate the consequences of changes in DnaX forms in Caulobacter crescentus. We demonstrate that switching from a processible form of DnaX to a nonprocessible variant (τ-only DnaX) causes a fitness defect. The strain grows normally once made but is less tolerant of replication stress in the presence of DNA-damaging agents. To explore this phenomenon further, we performed transposon sequencing to identify synthetic interactions with the τ-only allele, one of which was interruption of the damage-inducible cell division inhibitor sidA. Subsequent experiments show that the ΔsidA τ-only strain has a fitness defect compared to the wild-type strain or single mutants under normal growth conditions and intolerance to DNA-damaging agents. Consistent with this, we find that the ΔsidA τ-only strain has an elevated level of RecA protein and is dependent on recA for survival. This strain also has excess replication clamp foci during normal growth, suggesting a role for DnaX in preventing or resolving stalled replication forks, even in the absence of exogenous DNA damage treatment.

These data indicate that there is bias to maintain DnaX processing even in the absence of exogenous DNA damage. While Caulobacter can adapt to and tolerate the absence of DnaX processing, these cells are far less robust during DNA replication stress, during DNA damage treatment, or when DNA replication-cell division coordination is altered by deletion of sidA.

RESULTS

A bias against the τ-only allele revealed during strain construction.

Our lab has previously reported on a form of DnaX that is not proteolyzed by ClpXP (DnaX_np) (8). This DnaX variant contains two frameshift mutations that alters the amino acid sequence of a large portion of DnaX adjacent to the predicted ClpXP recognition site (17). This variant is recognized by ClpXP but fails to generate partial proteolysis fragments. To eliminate ClpXP recognition, we substituted two adjacent amino acids (AA544 and 545 to DD) in the predicted ClpXP recognition site, with the rest of the sequence remaining intact. This variant is unable to be processed at all by ClpXP in vitro (see Fig. S1 in the supplemental material).

A two-step double-recombination method was used to replace the Caulobacter dnaX gene with the τ-only allele. We first engineered point mutations in the center of a 2,000-bp region of homology cloned into the pNPTS138 suicide vector, which cannot propagate in Caulobacter and contains a counterselectable sacB marker along with a kanamycin resistance cassette (26) (Fig. 2A, top). The homology region starts at nucleotide 630 in the dnaX coding sequence (1,000 bp upstream of the mutations) and ends 822 nucleotides after the end of the dnaX reading frame (1,000 bp downstream of the mutations). Based on sequence alone, approximately half of primary integrations were expected to occur within the homology region before the mutations, leading to expression of the τ-only allele from the chromosomal promoter (Fig. 2A, middle). The other half were expected to occur in the homology region after the mutations, resulting in expression of the wild-type dnaX allele from the chromosomal promoter (Fig. 2A, middle). Western blot analysis was used to determine the percentage of isolates expressing wild-type dnaX and the percentage expressing the τ-only variant. The wild-type dnaX allele expresses three DnaX forms (one τ form and two γ forms), while the τ-only allele expresses only the largest DnaX form (Fig. 3A). Surprisingly, about 90% of primary integration isolates (18/20) expressed the wild-type dnaX allele from the chromosomal promoter (Fig. 2A, middle). Since there is no reason to believe that a bias exists in the locations of recombination events, this suggests a strong bias exists against eliminating DnaX processing.

FIG 2.

There is a bias against eliminating DnaX processing, but rare isolates make strain construction possible. (A) We used a two-step recombination strategy to construct the τ-only strain. The suicide vector was designed to produce a 1:1 ratio of wild-type dnaX to τ-only allele expression at the primary integration and secondary recombination steps; however, both steps favored maintaining the wild-type dnaX over recombining to switch to expression of the τ-only allele. (B) While the wild-type dnaX allele rescues growth of the dnaX_Ts strain at the nonpermissive temperature, the τ-only allele was not able to fully rescue. The experiment was performed in triplicate; representative images are shown. (C) Overnight growth was back-diluted to an OD₆₀₀ of 0.3 and grown at the nonpermissive temperature for 1.5 h. The OD₆₀₀ was measured, and cells were back-diluted to an OD₆₀₀ of 0.3. Measurement and back-dilution were repeated every 1.5 h for 9 h. A cumulative change in OD₆₀₀ was calculated at each time point. Results of a single experiment are shown. As in the plating assay, the wild-type dnaX allele rescued the growth of the dnaX_Ts strain at the nonpermissive temperature, but the τ-only allele did not.

FIG 3.

Absence of DnaX processing does not affect cell growth or morphology. (A) The final τ-only strain was analyzed by Western blotting using the Caulobacter anti-DnaX antibody. Instead of the three DnaX forms observed in the wild-type strain, only the longest DnaX form (τ) is observed. The growth (B) and morphology (C) of this strain are comparable to those of the wild-type strain. Experiments in panel B were performed in triplicate; the mean and standard deviation are shown. Where error bars are smaller than the width of the marker, error bars were omitted. The experiment in panel C was performed in triplicate; representative images are shown. Scale bars represents 5 μm. (D and E) Exponential-phase cells were treated with various amounts of UV light (D) or various concentrations of mitomycin C (MMC) for 1 h (E) The τ-only strain demonstrated a consistent but mild defect in mitomycin C survival but not in survival to UV light. Experiments were performed in triplicate; the mean and individual data are shown. Two-way ANOVA was performed to identify the effect of the strain on survival with mitomycin C (P = 0.078).

Secondary recombination events that had eliminated the plasmid from the chromosome were then identified by growing cells in nonselective medium and plating on sucrose-containing medium, which selects for loss of sacB function, and then verifying kanamycin sensitivity, which indicated loss of the nptI gene function. Because dnaX is essential, these isolates must have kept one of the two homologous regions surrounding the dnaX mutation site but lost the plasmid sequence with the second homologous sequence. Based on a model where the secondary recombination occurs randomly, the secondary recombinants were expected to have a roughly equal probability of maintaining either the dnaX or τ-only allele, despite which allele was expressed in the primary integration (Fig. 2A, bottom). When a dnaX-expressing primary integrant was used, almost 90% of isolates expressed wild-type dnaX (48/57) (Fig. 2A, bottom left). One of the τ-only-expressing isolates from this process is used in the remainder of this article and is called the τ-only strain.

When a τ-only-expressing primary integration was used for secondary recombination, there was a strong bias for maintaining the τ-only allele (83/103) (Fig. 2A, bottom right). Five of the final τ-only strains were analyzed using short-read whole-genome sequencing, which did not reveal any suppressor mutations (see Tables S3 to S7). These data suggest that eliminating DnaX processing by converting from a wild-type dnaX to the τ-only allele causes a fitness defect, but that Caulobacter can adapt to expression of the τ-only allele without the presence of a suppressor mutation.

To explore this phenomenon further, we introduced plasmid-borne copies of these alleles into a temperature-sensitive dnaX (dnaX_Ts) strain. We had previously reported that the dnaX_np allele was unable to rescue growth of a temperature-sensitive allele of dnaX (17). It was unclear whether the lack of processing itself or the large change in amino acid sequence of the DnaX_np was the cause of this phenotype. Unlike the allelic replacement result, we find that the τ-only allele is insufficient to completely rescue growth of the dnaX_Ts strain at nonpermissive temperature (Fig. 2B and C). These data support our observation above that the replacement of wild-type dnaX with the τ-only allele is detrimental, a consequence accentuated during acute loss of DnaX activity in a temperature-sensitive background. These data also align with what had previously been observed by our lab for the dnaX_np allele (17).

The τ-only strain is mildly sensitive to DNA damage by mitomycin C, but not UV light.

Despite the apparent bias observed in its construction, the τ-only strain shows no defects in growth or morphology (Fig. 3B and C and Fig. S2A). E. coli τ-only strains are sensitive to ultraviolet (UV) radiation (4), but the τ-only strain of Caulobacter does not show the same sensitivity (Fig. 3D). This difference might be explained by the fact that over Caulobacter’s evolution, it has naturally been exposed to high doses of UV light in its native freshwater environment. The τ-only strain does exhibit a mild sensitivity to mitomycin C, with ~65 to 75% of the survival rate of mitomycin C treatment of the wild-type cells (two-way analysis of variance [ANOVA]; P = 0.078) (Fig. 3E). This evidence suggests that, while the cells can adapt to the absence of DnaX processing, this change leaves them susceptible to certain forms of DNA replication stress, such as by treatment with mitomycin C.

Deletion of cell division inhibitor sidA is detrimental in the absence of DnaX processing.

To gain insight into the role of DnaX processing, we performed transposon sequencing to identify synthetic interactions that would help us determine how cells are adapting to the τ-only allele. We created and sequenced EZ-Tn5 transposon libraries in the wild-type and τ-only backgrounds and then compared the distribution of insertions. Each library contained ~200,000 unique insertion sites (27) (see Materials and Methods).

We found significantly smaller number of insertions in the sidA gene for the τ-only background compared to the wild-type background (log₂ fold difference = −4.8). We found that the bulk of the sidA insertions in the wild type library were located at a single site (Fig. 4A), which was not surprising due to the small size of the sidA coding sequence (123 bp). There were no insertions at this site in the τ-only library. The sidA gene encodes a LexA/RecA-controlled protein that is thought to inhibit cell division during the DNA damage response (28). SidA is one of two known DNA damage-induced Caulobacter cell division inhibitors, which are thought to stall cell division and allow cells to repair damaged DNA before it is passed on to a daughter cell (28, 29).

FIG 4.

Cell division inhibitor sidA is important in the absence of DnaX processing. (A) A transposon-sequencing experiment showed fewer insertions in the gene sidA in a τ-only background compared to in a wild-type background. The y axis of each plot measures the number of transposon insertion on a linear scale from 0 to 200. Each x axis shows genome location of nucleotides 2152211 to 2152696 in Caulobacter strain NA1000 (40). Deletion of sidA in a τ-only strain results in a growth defect (B and C) and filamentation (D). All experiments were performed in triplicate. In panel B, the mean and standard deviation are shown. Where error bars are smaller than the width of the marker, error bars were omitted. Growth rate quantification is shown in Fig. S2A. The ΔsidA τ-only strain has a significantly lower slope than each of the other strains (P < 0.05). (C) When each test strain is competed against a wild type Venus-expressing fluorescent reporter in coculture, only the ΔsidA τ-only mutant shows a competitive disadvantage. The mean and results of a two-tailed t test are shown. This is consistent with the growth defect observed in panel B. In panel D, representative images are shown. Quantifications of cell length are shown in Fig. S2B. Scale bars represent 5 μm.

We constructed a ΔsidA τ-only strain and found that this strain grows more slowly than either parent or the wild-type strain (Fig. 4B and Fig. S2A). To explore this growth defect further, we used a coculture competition experiment. The ΔsidA τ-only strain has a significantly lower ability to compete with a fluorescently labeled wild-type reporter strain than the wild-type strain or single mutant strains (Fig. 4C). To confirm that this is not an artifact of the wild-type fluorescent reporter strain, we also tested the single mutants against a ΔsidA τ-only fluorescent reporter and vice versa (Fig. S2C and D). In both cases, the individual mutant strains have a higher ability to compete in coculture than the ΔsidA τ-only strain (Fig. S2C and D).

τ-only ΔsidA cells show increased RecA dependence.

The ΔsidA τ-only cells are also longer (Fig. 4D and Fig. S2B) than cells of the single mutants and wild-type strain. As cell filamentation can be a sign of cellular distress such as that seen during DNA damage (28), we hypothesized that deletion of the sidA cell division inhibitor affects coordination between DNA replication and cell division, leading to increased DNA replication stress in the absence of DnaX processing. Consistent with this interpretation, RecA levels in the double ΔsidA τ-only mutant were elevated (Fig. 5A; quantification in Fig. S3A). Furthermore, deletion of recA in this strain severely inhibited growth (Fig. 5B; Fig. S3B), indicating that a functional DNA damage response is required for normal growth of the double ΔsidA τ-only mutant. These data suggest that the cell division inhibitor gene sidA is necessary in the absence of DnaX processing, likely due to DNA replication stress. SidA and its function as a cell division inhibitor are likely one mechanism by which the cell adapts to the τ-only allele during this stressful time.

FIG 5.

The τ-only allele and sidA deletion results in a synthetic phenotype of a deficiency in DNA damage resolution. (A) Western blot analysis shows an increased level of RecA protein in the ΔsidA τ-only strain compared to the wild-type or single mutant strains. Variability of relative RecA concentration is ~10%. Full quantification of RecA levels shown in Fig. S3A. A comparison of growth curves of these parental (recA⁺) strains (Fig. 4B) and recA knockouts in these strains (B) indicates that recA is critical for growth of the ΔsidA τ-only strain compared to the wild-type or single mutant strains. The mean and standard deviation are shown. Where error bars are smaller than the width of the marker, error bars were omitted. Quantification of these growth rates shown in Fig. S3B. (C) When DnaN-YFP foci are observed in the parental (recA⁺) strains, the ΔsidA τ-only strain has an increased proportion of cells that contain >2 DnaN-YFP foci. This indicates a greater than normal number of replisomes. The extra foci are likely due to replication-dependent DNA repair. All experiments were performed in triplicate. The mean and standard deviation shown. (D) We find that the ΔsidA τ-only mutant is more sensitive to mitomycin C than the wild-type strain or single mutants. Cells were spotted on 0 or 1 μg/mL mitomycin C. Spots in a row represent 2-fold serial dilutions. Representative images of triplicate experiments shown.

To directly obtain insight into replisome dynamics in the ΔsidA τ-only strain, we monitored fluorescently labeled DNA clamp (DnaN-yellow fluorescent protein [YFP]) foci as a proxy for replisomes. In asynchronous wild-type and single mutant strains, most cells had 1 to 2 replication foci, likely corresponding to active replication forks and/or a small number of DNA damage-related foci (e.g., stalled replication forks or lesion bypass or lesion repair sites). A small population of cells contain more than two DnaN foci (<10%) (Fig. 5C; examples of wild-type cells with 1, 2, or >2 foci are depicted in Fig. S3C). These extra foci are likely to be related to DNA damage or replication fork stalling, since only a single DNA replication initiation event occurs in one Caulobacter cell cycle. Importantly, we find that the ΔsidA τ-only strain has twice as many cells with >2 foci as the other strains (~20%) (Fig. 5C). Given the RecA induction and importance of recA to the growth of this strain, these extra foci are likely related to DNA damage and/or replication fork stalling. Consistent with this interpretation, the strain is more sensitive to mitomycin C than the wild-type strain or single mutants (Fig. 5D).

These data can be explained by two possibilities: either the ΔsidA τ-only strain has a larger amount of damaged DNA, even in the absence of an exogenous DNA-damaging treatment, or DNA damage-related foci are more persistent in this strain. Either possibility explains the filamentation, RecA induction, and dependence on recA that we observed above. While Caulobacter can tolerate either the absence of DnaX processing or the absence of the cell division inhibitor sidA, both mutations lead to an increase in DNA damage and/or replication fork stalling events. This likely contributes to the growth and fitness defects that we observe in this strain as well as its sensitivity to the DNA-damaging agent mitomycin C.

Phenotypic defects arise from both lack of γ and surplus of τ.

Since the τ-only strain has a larger total amount of DnaX in the cell (Fig. S4), it is unclear whether bias against the τ-only allele and the phenotypes observed in the τ-only and ΔsidA τ-only strains are due to the absence of the γ forms or due to the increased abundance of DnaX or τ in the strain. To address this question, we attempted to complement the ΔsidA τ-only strain’s mitomycin C sensitivity by expressing a truncated DnaX form that mimicked the estimated size of the longer Caulobacter γ form. The mitomycin C sensitivity of the ΔsidA τ-only strain is partially rescued by expression of γ [DnaX(1–492)] at a second locus (Fig. 6A). While we estimate that the ΔsidA τ-only strain has ~16- to 32-fold greater sensitivity than the ΔsidA strain, expression of γ [DnaX(1–492)] results in only a 4- to 8-fold greater sensitivity than the ΔsidA strain.

FIG 6.

Mitomycin C sensitivity in the ΔsidA τ-only strain is partially complemented by expression of γ, but poor competition is not. (A) The mitomycin C (MMC) sensitivity in the ΔsidA τ-only strain is partially complemented by expressing of γ (amino acids 1 to 492) at an alternate locus (the xylose locus). Cells were spotted on 0 or 1 μg/mL mitomycin C. Spots in a row represent 2-fold serial dilutions. Representative images of triplicate experiments shown. Other replicates are shown in Fig. S5. (B) The competitive disadvantage of the ΔsidA τ-only strain is not complemented by expression of γ (amino acids 1 to 492) at an alternate locus (the xylose locus).

Next, we tried to complement the ΔsidA τ-only strain's inability to compete in coculture by expressing γ [DnaX(1–492)], but γ expression did not complement this phenotype (Fig. 6B). These data indicate that while the absence of γ is likely involved in the phenotypes of the ΔsidA τ-only double mutant, it is not the full explanation for these phenotypes.

Our results so far suggest that a surplus of the τ form of DnaX may be toxic to cells lacking sidA. To explicitly test this, we introduced a second copy of dnaX on a low-copy plasmid into the wild-type, τ-only, and ΔsidA strains. In all parental strains, addition of an exogenous τ-only allele resulted in a significant increase in cell length (Fig. 7A and Fig. S6A) and decrease in growth rate (Fig. 7B, Fig. S6B, and Table S8), while an exogenous copy of the wild-type dnaX gene had little to no effect. Additionally, we find that the effects of an exogenous τ-only allele on cell length and growth rate are much larger in the ΔsidA strain than in the wild-type and τ-only strains (Fig. 7A and B, Fig. S6A and B, and Table S8), despite total DnaX levels being similar in the two parental strains for each plasmid (Table S5C).

FIG 7.

Excess nonprocessible DnaX is toxic in the absence of sidA. Introduction of excess τ-only results in (A) elongation and (B) slower growth of ΔsidA cells, while introduction of excess wild-type dnaX has no significant effect. All experiments were performed in triplicate. Representative images are shown in panel A; quantification of cell length is shown in Fig. S6A. In panel B, the mean and standard deviation are shown. Where error bars are smaller than the width of the marker, error bars were omitted. Growth rate quantification is shown in Fig. S6B.

These data suggest that phenotypes associated with the τ-only allele are not simply due to the absence of γ. They also suggest that the bias shifting to the τ-only allele in strain construction was not due to a simple elevation in total DnaX levels. It appears that an excess of a nonprocessible form of DnaX is detrimental to the cell, particularly in the absence of sidA. We propose that DnaX processing is important to preventing or resolving instances of DNA damage and/or replication fork stalling and that sidA can compensate for the loss of DnaX processing, likely by adjusting cell division. An excess of τ-only DnaX interferes with this role and results in a sensitivity to DNA replication stress.

DISCUSSION

Here, we find that there is selection to maintain DnaX processing in Caulobacter. Although Caulobacter cells can survive without this processing, they are less able to survive stress related to DNA replication. These findings are in line with previous work in E. coli, where ribosomal frameshifting during DnaX translation aids survival of exogenous sources of DNA damage (4). This need for γ is explained by the higher preference of E. coli τ-only clamp loader for DnaE, the replicative polymerase, which restricts access for alternative DNA polymerase loading during replication fork stalling (4). Our results here showing increased DnaN-YFP foci, dependence on recA, and increased mitomycin C sensitivity in the ΔsidA τ-only strain suggest that the clamp loader may play a similar role in resolving replication fork stalling events in Caulobacter as it does in E. coli (Fig. 8).

FIG 8.

Clamp loader composition is important for resolution of DNA damage. Our data support the model that the τ-only clamp loader is less effective at allowing alternative translesion DNA polymerases (TLS Pols) and possibly other replication restart factors (e.g., DNA repair factors and DNA replication termination factors) to access the fork than the wild-type τ₂γ₁ clamp loader. This deficient access of replication restart factors causes a buildup of stalled replication forks, which triggers a sidA-dependent mechanism to compensate for this, likely involving inhibition of cell division. The expression of γ in the τ-only strain improves its DNA damage tolerance, possibly due to a γ-only clamp unloader assisting in removal of stalled DnaE-associated DNA clamps, which allows some restart machinery to better access the clamp.

We find that expression of γ can partially rescue the mitomycin C sensitivity in the ΔsidA τ-only but cannot suppress this strain’s reduced fitness under unstressed conditions. It has been suggested that γ could primarily participate in clamp unloading (25), and eukaryotes encode the clamp unloader Elg1/ATAD5, which removes PCNA clamp at stalled replication sites during DNA damage (reviewed in reference 30). Based on this, we speculate that a γ-only clamp unloader is providing some relief to the mitomycin C-treated ΔsidA τ-only strain by helping to unload clamp and DnaE at stalled replication forks. This would allow replication restart machinery to access and repair or bypass the DNA damage.

In contrast, expression of γ does not rescue the ΔsidA τ-only mutant’s failure to compete with wild type in the absence of DNA damage. Our interpretation that overabundance of τ-only DnaX or an imbalance of τ/γ ratio may play a role in this phenotype. We note that excessive τ-only results in slower growth even in wild-type cells (Fig. 7B), suggesting that an excess of the DnaX C-terminal domain may have negative effects even in wild-type backgrounds, such as titration of the DnaE away from the replication fork.

The mitomycin C sensitivity of the τ-only strain is amplified when sidA is deleted, with defects in growth and dependence on recA that are not seen in the parent strains. While sidA is highly upregulated during DNA damage (28), cells lacking sidA show no deficiency in DNA damage tolerance (Fig. 5D), presumably due to redundant pathways, such as didA (29). The only published effect for sidA alone is that overexpression is sufficient to halt cell division, even in the absence of DNA damage (28). We hypothesize that sidA inhibition of cell division is required in the τ-only strain to compensate for deficiencies in clamp loader activities, such as alterative polymerase switching, needed for repair under DNA-damaging conditions (Fig. 3E, 6, and 8). Similarly, elevation of τ levels in a ΔsidA mutant results in a severe growth defect and filamentation (Fig. 7), consistent with a need for sidA to adjust cell division in response to DNA replication needs.

Our overall model is that the τ-only clamp loader is not as effective in its role in replication restart. This may be because the τ-only clamp loader is less effective in allowing restart factors, such as alternative DNA polymerases, to access the fork (Fig. 8), as has been found for the τ-only clamp loader and alternative DNA polymerases in E. coli (4). The switch of wild-type clamp loaders to τ-only clamp loaders is initially very detrimental, as we see in our strain construction (Fig. 2A) and temperature sensitivity complementation (Fig. 2B and C). Although there is a bias against switching to the τ-only allele, cells can adapt and grow normally (Fig. 3B and C). One such adaptation seems to rely on sidA to adjust the coordination of DNA replication and cell division (Fig. 7).

While our results indicate that the capacity to process DnaX is important, particularly during DNA replication stress, they also show that DnaX processing is not essential for cell survival and that cells can eventually adapt to a τ-only allele. This may explain why the τ and γ forms of DnaX are generally conserved in bacteria, but the pathways to the shorter γ form vary. A possible scenario is that early bacteria used a clamp loader with identical DnaX subunits, which would have more closely resembled the T4 phage clamp loader (reviewed in reference 31). Over time, perhaps coinciding with the increased risk to DNA damage that came with an oxygen-rich atmosphere, bacteria may have independently evolved mechanisms to create γ. Eukaryotic clamp loaders took a different route to evolve a similar system, where the multiple DnaX-like subunits of the clamp loader are encoded by unique genes. This might also explain why there are bacterial species that do not appear to produce a γ form, such as Streptococcus pyogenes, Aquifex aeolicus, and Bacillus subtilis (32–34).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. All bacterial strains in this study are Caulobacter crescentus NA1000 derivatives and were grown in PYE medium (2 g/L peptone, 1 g/L yeast extract, 1 mM MgSO₄, and 0.5 mM CaCl₂) at 30°C. Solid medium was made with 1.5% agar. The following antibiotics were used as indicated below: kanamycin (liquid culture, 10 μg/mL; solid medium, 25 μg/mL), spectinomycin (liquid culture, 25 μg/mL; solid medium, 200 μg/mL), oxytetracycline (liquid culture, 1 μg/mL; solid medium, 2 μg/mL). Antibiotics were used when growing all strains containing self-replicating plasmids or single-recombination plasmid integrations.

Constructing the τ-only strain using two-step recombination.

The HinDIII-EcoRI fragment of pNPTS138 was replaced by a 2,000-bp homologous region of the Caulobacter chromosome surrounding codons 544 and 545 of the dnaX coding sequence. These two codons were mutated by site-directed mutagenesis (codon 544, GCG to GAC; codon 545, GCC to GAC) to replace DnaX alanine residues 544 and 545 with aspartic acid. This suicide vector was introduced into Caulobacter NA1000 by electroporation, and primary integrations were selected for by plating on kanamycin. Single colonies were used to inoculate liquid cultures, and lysates were analyzed by Western blotting to determine which form of DnaX was expressed. These liquid cultures were also plated on 3% sucrose to select for cells that had undergone secondary recombination to remove the plasmid sequence from the chromosome. Single isolates were then patched on antibiotic-free plates and on kanamycin to confirm the loss of the plasmid sequence. Kanamycin-sensitive isolates were grown in liquid culture, and lysates were analyzed by Western blotting to determine which DnaX form was expressed.

Western blot.

Western blotting was performed using either an affinity-purified rabbit anti-Caulobacter DnaX primary antibody (1:10,000 dilution for 1 h) or a commercially available rabbit anti-E. coli RecA primary antibody (1:10,000 dilution for 16 h) (Abcam, Cambridge, MA). Blots were visualized using the goat anti-rabbit Alexa Fluor Plus 800-conjugated secondary antibody (1:10,000 dilution for 1 h) (Licor, Lincoln, NE). For normalization purposes, blots were then probed with a rabbit anti-Caulobacter ClpP serum (1:10,000 dilution for 16 h) and visualized with the same Licor secondary antibody. Bands were quantified using ImageJ (35).

Bacterial growth curves.

Cells in the stationary phase (Fig. 3B and 7B) or exponential phase (Fig. 4B and 5B) were diluted to an optical density at 600 nm (OD₆₀₀) of 0.1 in a 96-well plate. Cells were grown with shaking at 30°C in a plate reader, and the OD₆₀₀ was measured every 20 min for 15 h. To find the growth rate, the slope of the most linear portion of each replicate was taken.

Cell morphology measurements and competition assays.

For cell morphology measurements, cells were back-diluted and grown for 10 to 15 generations to an OD₆₀₀ of about 0.5. Cells were imaged by phase-contrast microscopy under ×1,000 magnification. ImageJ (35) and MicrobeJ (36) were used to identify cells and quantify cell length.

Overnight cultures of each test strain and the Venus-expressing reporter strain were mixed 1:1 according to OD₆₀₀ and serially diluted 30,000-fold in a final volume of 10 mL PYE. Undiluted cells were imaged by phase microscopy and fluorescence microscopy using a green fluorescent protein (GFP)-selecting filter. ImageJ was used to manually quantify the number of fluorescent and nonfluorescent cells in the images for each sample. Diluted samples were grown to stationary phase and imaged again. The competitive index was calculated as the log₂ {[(ratio test/Venus)_END/(ratio test/Venus)_START]/[(ratio wild type/Venus)_END/(ratio wild type/Venus)_START]}.

DNA damage susceptibility testing.

For the acute stress testing (Fig. 3D and E), cells were grown to the exponential phase and treated with UV light using a Stratalinker UV cross-linker or with mitomycin C (Millipore Sigma, St. Louis, MO) for 1 h. Cells were serially diluted in PYE, plated on PYE agar, and incubated at 30°C for 48 h. CFU were counted, and the CFU count of the original sample was calculated as (no. of CFU on plate) × (dilution factor). The percentage of survival was calculated as CFU count for treated cells/CFU count for untreated cells.

For chronic stress (Fig. 5D and Fig. S5), stationary-phase cultures were normalized to an OD₆₀₀ value of 0.5, diluted 100-fold (indicated by the leftmost spot in a row), and then serially diluted by 2-fold for 7 dilutions, and plated on PYE agar and PYE agar plus mitomycin C. Plates were incubated for 48 h (untreated) or 60 h (treated) at 30°C and then imaged using a Syngene G:Box.

Temperature sensitivity complementation.

For the plate-based method, colonies from fresh plates were struck onto PYE agar and inoculated at the permissive temperature (30°C) or nonpermissive temperature (42°C).

For liquid-based method, overnight growth was diluted and grown to exponential phase at permissive temperature. At time zero, cells were normalized to an OD₆₀₀ of 0.3 and shifted to the nonpermissive temperature. Every 2 h, the OD₆₀₀ was measured, and cells were diluted back to OD₆₀₀ of 0.3 in prewarmed PYE. The total increase in mass was calculated as the cumulative increase in OD₆₀₀.

Transposon sequencing.

Transposon libraries were created using the wild-type and τ-only parental strains as previously described (27, 37). The wild-type and τ-only libraries outgrown and sequenced in triplicate. Data were analyzed using edgeR (38). Differential analyses of the counts in these libraries are available in Table S2. The gene sidA (CCNA_02004) showed a significantly lower number of insertions in the τ-only library compared to the wild-type library [log₂ (τ only/wild type) = ~4.8] (Fig. 4).

Quantifying DnaN-YFP foci.

We used the single integration plasmid pNABC198 (a generous gift from the Badrinarayanan lab) (39) to tag DnaN with a C-terminal YFP tag at the native locus in the wild-type, τ-only, ΔsidA, and ΔsidA τ-only parental strains. Cells were grown to the exponential phase and imaged by phase-contrast and fluorescence microscopy. YFP foci were quantified manually using ImageJ (35).