Characterization of the Chromosome Dimer Resolution Site in Caulobacter crescentus

Bacteria utilize site-specific recombination for a variety of purposes, including the control of gene expression, acquisition of genetic elements, and the resolution of dimeric chromosomes. Failure to resolve dimeric chromosomes can lead to cell division defects in a percentage of the cell population. The work presented here shows the existence of a chromosomal resolution system in C. crescentus. Defects in this resolution system result in the formation of chains of cells. Further understanding of how these cells remain linked together will help in the understanding of how chromosome segregation and cell division are linked in C. crescentus.

ABSTRACT

Chromosome dimers occur in bacterial cells as a result of the recombinational repair of DNA. In most bacteria, chromosome dimers are resolved by XerCD site-specific recombination at the dif (deletion-induced filamentation) site located in the terminus region of the chromosome. Caulobacter crescentus, a Gram-negative oligotrophic bacterium, also possesses Xer recombinases, called CcXerC and CcXerD, which have been shown to interact with the Escherichia coli dif site in vitro. Previous studies on Caulobacter have suggested the presence of a dif site (referred to in this paper as dif1_CC), but no in vitro data have shown any association with this site and the CcXer proteins. Using recursive hidden Markov modeling, another group has proposed a second dif site, which we call dif2_CC, which shows more similarity to the dif consensus sequence. Here, by using a combination of in vitro experiments, we compare the affinities and the cleavage abilities of CcXerCD recombinases for both dif sites. Our results show that dif2_CC displays a higher affinity for CcXerC and CcXerD and is bound cooperatively by these proteins, which is not the case for the original dif1_CC site. Furthermore, dif2_CC nicked substrates are more efficiently cleaved by CcXerCD, and deletion of the site results in about 5 to 10% of cells showing an altered cellular morphology.

IMPORTANCE Bacteria utilize site-specific recombination for a variety of purposes, including the control of gene expression, acquisition of genetic elements, and the resolution of dimeric chromosomes. Failure to resolve dimeric chromosomes can lead to cell division defects in a percentage of the cell population. The work presented here shows the existence of a chromosomal resolution system in C. crescentus. Defects in this resolution system result in the formation of chains of cells. Further understanding of how these cells remain linked together will help in the understanding of how chromosome segregation and cell division are linked in C. crescentus.

INTRODUCTION

Most bacteria possess a single circular chromosome which is replicated bidirectionally during the cell cycle. Chromosome replication and segregation occur simultaneously in bacteria (1), and segregation is completed after the end of replication and before the closure of the septum (2, 3). The circularity of the bacterial chromosome and the high number of homologous recombination events that occur during replication, mainly due to broken or stalled replication forks (4), result in the creation of chromosome dimers in a fraction of the cell population. In Escherichia coli, chromosome dimer formation occurs in 15% of the cell population during replication, which needs to be resolved before the closure of division septum (5, 6). In most of the well-studied bacterial species, chromosome dimers are resolved by a highly conserved site-specific recombination system that employs two tyrosine recombinases, XerC and XerD, and a recombination site located in the terminus region of the bacterial chromosome called dif, for deletion-induced filamentation (7). This two-recombinase site-specific recombination system is not universal, as some bacterial species have been shown to use a single tyrosine recombinase to resolve chromosome dimers (8, 9).

In E. coli, chromosome dimer resolution occurs in a synaptic nucleoprotein complex containing four tyrosine recombinase monomers bound to two directly repeated dif sites. Each dif site contains two binding sites, separated by a 6-bp spacer, for the two tyrosine recombinase monomers, XerC and XerD. Resolution of chromosome dimers is catalyzed by the alternating activation of XerC and XerD monomers, which results in the formation, isomerization, and resolution of a Holliday junction between two synapsed dif sites (10–12).

The temporal and spatial organization of the XerCD/dif site-specific recombination system, along with its activation, is regulated by a cell division transmembrane protein, FtsK. After the initiation of septum formation and during cell division, the FtsK N-terminal domain, FtsK_N, anchors to the septal inner membrane, while its C-terminal domain, FtsK_C, translocates the chromosome using KOPS (FtsK-orienting polar sequences). As a result, FtsK_C drives the newly replicated sister chromatids into the new daughter cells, and in the case of dimeric chromosomes, it brings together both dif sites into the midcell, where it activates XerD constituent monomers of the XerCD/dif synaptic complex to begin the chromosome dimer resolution reaction (11–13). Further studies on XerCD/dif site-specific recombination systems have revealed that the bacterial endogenous Xer machinery not only is needed for the intact vertical transfer of genetic heritage but also is involved in horizontal genetic exchange (14, 15).

In the well-studied cell cycle model bacterium Caulobacter crescentus, chromosome dimer resolution appears to be similar to that of E. coli; the Xer recombinases, here referred to as CcXerC and CcXerD, display 54.8 and 60.5% similarity, respectively, to their E. coli homologues. Maltose binding protein (MBP) fusions to both CcXerC and CcXerD were shown to bind cooperatively to the E. coli dif site in vitro (16).

In an attempt to identify the dif site in C. crescentus CB15N, Jensen defined a putative chromosome dimer resolution site, which we call dif1_CC, located in the terminus region by analyzing highly skewed polar sequences (17). dif1_CC is located in an intergenic region upstream of CCNA-01840, which encodes an Xre transcriptional regulator (Fig. 1A). The deletion of dif1_CC results in the filamentation of 2 to 4% of the cell population. Compared to the XerCD/dif deficiency in E. coli, which results in the death and the filamentation of 15% of the cells (18), and the low degree of sequence similarity to the dif consensus sequence (Fig. 1B), we hypothesized that the lower filamentation frequency in C. crescenuts dif1_CC deletion is due to the presence of a resolution site elsewhere in the terminus region.

FIG 1.

(A) Schematic of the C. crescentus genomic region which contains hypothesized dif1_CC and dif2_CC sites. Red boxes are dif1_CC and dif2_CC, upstream and downstream, respectively, of a coding sequence that belongs to the Xre family transcriptional regulator. The sequences of both strands of the dif sites are indicated and are enclosed in red to indicate their 5′ to 3′ orientation. (B) A comparison between dif2_CC-dif1_CC and the bacterial dif consensus (20). Stars show identical nucleotides between dif2_CC-dif1_CC and the bacterial dif consensus. The XerC and XerD binding sites are indicated above the sequences, along with the central spacer region.

Using recursive hidden Markov modeling, another potential chromosome dimer resolution site, in this paper called dif2_CC, was found to be 2.1 kb away from dif1_CC in the terminus region of the C. crescentus CB15N chromosome (19). dif2_CC is located downstream of CCNA-01840 and upstream of CCNA-01837, which encodes an IS30 family transposase and exhibits more similarity to the bacterial dif consensus sequence than dif1_CC (Fig. 1) (20).

Here, we used a combination of in vitro approaches, including electrophoretic mobility shift assays (EMSA) and nicked suicide substrate cleavage assays, to analyze the affinity of CcXerCD recombinases for dif1_CC and dif2_CC and their ability to catalyze the first step in site-specific recombination at these two putative dif sites. Furthermore, to address the in vivo role of dif2_CC in chromosome dimer resolution, we created a C. crescentus strain with a deletion in dif2_CC, and the mutant phenotype was analyzed by phase-contrast and fluorescence microscopy.

RESULTS

CcXerD shows stronger affinity for dif2_CC than dif1_CC.

To investigate the functional significance of dif1_CC and dif2_CC in chromosome dimer resolution, the affinity of CcXerD and CcXerC for dif1_CC and dif2_CC was tested using electrophoretic mobility shift assays (EMSAs). Two fragments of 310 bp and 270 bp, containing dif1_CC and dif2_CC, were amplified and cloned into pUC19 to create pAF205 and pAF204, respectively. These fragments were amplified and labeled from pAF204 and pAF205 by performing PCR using the M13F and the 5′-6-HEX-labeled M13R universal primers. The CcXerD coding sequence was cloned in the expression plasmid pET28a, and His-tagged CcXerD protein was expressed and purified under native conditions as described in Materials and Methods. The CcXerC coding sequence was cloned in pMALc2 and purified as an MBP fusion protein under native conditions as described in Materials and Methods.

The affinity of CcXerD and MBP-CcXerC for 6-HEX-labeled dif1_CC and dif2_CC fragments was tested by EMSAs. At most of the CcXerD concentrations used, specific retardation of dif2_CC was observed, which was not the case for dif1_CC unless the protein was at its highest concentration (Fig. 2). Similar results were seen when an MBP fusion to CcXerD was used instead of the histidine-tagged version (see Fig. S1 in the supplemental material). Interestingly, unlike CcXerD, MBP-CcXerC could bind to both dif1_CC and dif2_CC, albeit with a slightly stronger affinity for dif2_CC (Fig. 2, lanes 8 to 10 versus 22 to 24). It is interesting that at high MBP-CcXerC concentrations (Fig. 2, lanes 8 and 22) multiple protein-DNA complexes were observed, which suggests that this recombinase is able to bind to both halves of dif1_CC or dif2_CC, leading to the formation of these additional complexes. Similar complexes were not observed with excessive amounts of CcXerD (Fig. 2, lanes 1 and 14).

FIG 2.

CcXerD and CcXerC binding to dif1_CC and dif2_CC. The concentrations of CcXerD for each lane are the following: lanes 1 and 14, 5,450 nM; lanes 2 and 15, 1,816 nM; lanes 3 and 16, 605 nM; lanes 4 and 17, 201 nM; lanes 5 and 18, 67 nM; lanes 6 and 19, 22 nM; and lanes 20, 0 nM. The concentrations of MBP-CcXerC for each lane are the following: lanes 7 and 21, 1,650 nM; lanes 8 and 22, 825 nM; lanes 9 and 23, 412 nM; lanes 10 and 24, 206 nM; lanes 11 and 25, 103 nM; lanes 12 and 26, 51 nM; lanes 13 and 27, 0 nM. All reaction mixtures contained 33 ng of HEX-labeled dif DNA and 500 ng of salmon sperm DNA.

CcXerC and CcXerD can cooperatively bind to dif2_CC but not dif1_CC.

In E. coli, XerC and XerD cooperatively bind to the dif site (21). To test the cooperativity of CcXerC and CcXerD in binding to dif2_CC and dif1_CC, we performed EMSAs using a constant concentration of CcXerD and increasing concentrations of MBP-CcXerC. The results showed a significant cooperativity between MBP-CcXerC and CcXerD in binding to dif2_CC (Fig. 3, lanes 11 to 16). The MBP-CcXerC concentrations we used in our EMSAs ranged from 25 to 825 nM. At low concentrations, i.e., 51 and 103 nM, MBP-CcXerC was not able to bind to dif2_CC when it was the only protein in the reaction mixture. However, in the presence of CcXerD, MBP-CcXerC could bind to dif2_CC even at the lowest concentration, 25 nM, in this experiment. This strongly suggests that having CcXerD bound to dif2_CC can stimulate the interaction between dif2_CC and MBP-CcXerC. The same experiment was performed to test CcXerC and CcXerD cooperativity in binding to dif1_CC, and no cooperativity was observed (Fig. 3, lanes 17 to 24). We also tested the cooperativity between CcXerC and CcXerD by adding a constant amount of MBP-CcXerC to increasing amounts of CcXerD. Again, a significant degree of cooperativity was seen between MBP-CcXerC and CcXerD in binding to dif2_CC but not dif1_CC (Fig. 3, lanes 1 to 8 and 25 to 32).

FIG 3.

CcXerD and MBP-CcXerC cooperative binding to dif2_CC and dif1_CC. The concentrations of MBP-CcXerC used in the samples are the following: lanes 2 to 8 and 26 and 32, 687 nM; lanes 11 and 19, 25 nM; lanes 12 and 20, 51 nM; lanes 13 and 21, 103 nM; lanes 14 and 22, 206 nM; lanes 15 and 23, 412 nM; and lanes 16 and 24, 825 nM. The concentrations of CcXerD used in the samples are the following: lanes 10 to 16 and 18 to 24, 605 nM; lanes 27 and 3, 22 nM; lanes 28 and 4, 67 nM; lanes 29 and 5, 201 nM; lanes 30 and 6, 605 nM; lanes 31 and 7, 1,816 nM; lanes 32 and 8, 5,450 nM. All reaction mixtures contained 33 ng of HEX-labeled dif DNA and 500 ng of salmon sperm DNA.

CcXerCD can efficiently cleave dif2_CC but not dif1_CC.

The resolution of chromosome dimers starts with the Xer-mediated cleavage of a single strand of each synaptic dif site within a recombinase-DNA nucleoprotein complex. However, after the first recombinase cleavage and prior to the subsequent strand exchanges, the reaction can be blocked by using a suicide substrate which contains a nick in the central region two or three nucleotides away from the first cleavage site (Fig. 4E). After the first recombinase cleavage of the dif site a small fragment diffuses away, blocking the subsequent strand exchanges. The stable recombinase-DNA covalent complex then can be isolated by electrophoresis on a denaturing polyacrylamide gel. To test the ability of CcXerCD to cleave the hypothetical dif_CC sites, we constructed two suicide substrates for each of the dif1_CC and dif2_CC sites that contained a nick either at the top or the bottom strand. The left-half site of the top nicked substrate and right-half site of the bottom nicked substrate were 5′-6-HEX labeled. dif suicide substrates were incubated alone and also in the presence of both recombinases. Both recombinases were used together at concentrations at which more than 90% of dif2_CC substrates were bound based on our EMSAs.

FIG 4.

CcXerCD activity on dif1_CC and dif2_CC suicide substrates. The top of the figure indicates the sequences of the dif substrates, with the arrow indicating the position of the nick in either the top (TN) or bottom strand (BN). (A) dif1_CC bottom-nicked substrate; (B) dif1cc top-nicked substrate; (C) dif2cc bottom-nicked substrate; (D) dif2_CC top-nicked substrate. For panels A to D, lane 1 contains MBP-CcXerC, lane 2 contains CcXerD, lane 3 contains CcXerD and MBP-CcXerC, and lane 4 contains no protein. CcXerD and MBP-CcXerC were used at concentrations of 605 and 687 nM, respectively. All samples contained 100 pmol of labeled dif suicide substrate and 500 ng of salmon sperm DNA. The cartoons show MBPXerC-bound and free dif suicide substrate. (E) Diagram of the Xer recombinase reaction on suicide substrates. The reactive hydroxyl group of the tyrosine residue of XerC (red) is indicated as the protruding chemical group. For brevity, only the top-strand reaction is indicated. The presence of XerD (green) is required for the XerC cleavage reaction, but it is not catalytically active. XerD can also perform the cleavage reaction on the bottom strand in the presence of XerC (not shown). The 5′ ends of the DNA are indicated as balls, and the stars indicate the position of the fluorescent label on the 5′ end of the DNA. After cleavage, XerC becomes chemically linked to dif DNA via a 3′-phosphotyrosyl linkage, and a small DNA fragment is liberated after cleavage (shown at a 45° angle).

Consistent with our EMSA results, suicide substrate assays on nicked dif substrates showed that the Caulobacter XerCD recombination system can efficiently cleave the top strand of dif2_CC when both MBP-CcXerC and CcXerD are present in the reaction mixture, generating a recombinase-dif covalent complex (Fig. 4D, lane 3). A small amount of covalent complex was also observed when only MBP-CcXerC was added to either dif1_CC top-nicked (TN) or bottom-nicked (BN) substrate (Fig. 4A and B, lane 1) or to dif2_CC TN or BN substrate (Fig. 4C and D, lane 1). Similar phenomena have been described previously with MBP-XerC from E. coli (22) and MBP-CodV from Bacillus subtilis (23). The faint, faster-migrating complex seen in Fig. 4C, lane 3, is the CcXerD-dif2_CC complex, which requires the presence of both MBP-CcXerC and CcXerD, similar to what has been reported for both E. coli and B. subtilis, which both display a faint XerD-dif covalent complex on bottom nicked suicide substrates. The addition of CcXerD to the dif1_CC reactions does not increase the amount of complex observed with MBP-CcXerC alone (Fig. 4A and B, lane 3). Trace amounts of a faster-migrating complex can be seen in Fig. 4A, lane 2. This appears to be an artifact with the dif1_CC substrate, as it can also be seen in the control lane without protein (Fig. 4A, lane 4). Cleavage assays using an MBP-CcXerC^Y-F mutant where a phenylalanine replaces the active tyrosine residue did not produce any covalent complexes for the dif2_CC top-nicked suicide substrate, confirming that the strong complex formed in Fig. 4D, lane 3, is comprised of MBP-CcXerC covalently linked to dif2_CC (Fig. S2).

Deletion of dif2_CC induces abnormal cell division in a subpopulation of mutant cells.

Our results show that CcXerC and CcXerD can cooperatively bind to dif2_CC and efficiently cleave the dif2_CC top strand to start the Xer recombination reaction. This strongly suggests that dif2_CC is the main site where CcXerC and CcXerD resolve chromosome dimers by site-specific recombination. It has been shown that the deletion of dif in E. coli and B. subtilis induces chromosome partitioning failure in 15% and 25% of the cell population, respectively (23, 24). In order to address the impact of dif2_CC deletion in C. crescentus cells, we constructed a mutant lacking a 45-bp chromosomal segment containing dif2_CC in the intergenic region between CCNA1840 (xre) and CCNA1837 (IS30-like tnpase). The dif2_CC deletion mutant was constructed by double crossover using a pNPTS138-derived plasmid, pAF202, as described in Materials and Methods.

C. crescentus dif2_CC-deleted cells were grown in peptone-yeast extract (PYE) to mid-log phase, stained with 4′,6-diamidino-2-phenylindole (DAPI), and examined by phase-contrast and fluorescence microscopy. About 5% to 10% of cells in the dif2_CC-deleted mutant cultures showed cell chains containing 3 to 10 cells (Fig. 5b and c), some of which were connected with threadlike appendices (Fig. 5c). Fluorescence microscopy revealed the presence DAPI-stained DNA in the threadlike appendices, suggestive of aberrant DNA segregation (Fig. 5f). However, we could not confirm the existence of chromosomal DNA in all examined threadlike structures by DAPI staining, probably due to the limitation of DAPI staining in revealing the presence of trivial amounts of DNA trapped in these thin structures. To test whether the cell chain phenotype observed in the dif2_CC mutant was a result of the inability of cells to resolve chromosome dimers, we disrupted the C. crescentus xerD gene (CcxerD) by plasmid integration using pAF208, and the same cell chain phenotype was observed in the CcxerD background (Fig. 5d), which confirms dif2_CC as the chromosome dimer resolution site in C. crescentus. This was further confirmed in recA mutant cells lacking dif2_CC, where the chain phenotype was almost eliminated and suppressed (Fig. S3) due to the elimination of RecA-dependent homologous recombination in the double mutant recA dif2_CC cells. It should be noted that even though the cell chain phenotype was drastically reduced in the recA dif2_CC double mutant, about 0.5% of cells still displayed a mild cell chain phenotype containing 3 to 5 cells held in a chain. Crossover events through RecA-independent recombination mechanisms have been reported previously in E. coli (25, 26). To the best of our knowledge, in this regard, no report has been documented for Caulobacter so far. Nevertheless, taking into consideration that Caulobacter possesses a RecA-independent checkpoint mechanism for delaying cell division in the case of DNA damage (27), we cannot rule out the possibility that, like E. coli, similar RecA-independent recombination mechanisms exist in Caulobacter, explaining the possible reason behind the remaining mild chain phenotype observed in 0.5% of cells in the recA dif2_CC double mutant.

FIG 5.

(a) Phase-contrast microscopy of wild-type C. crescentus during log phase at 400× magnification. (b) Phase-contrast microscopy of C. crescentus dif2_CC-deleted mutant during log phase at 400× magnification. Arrows show the subpopulation of the cells demonstrating the cell chain phenotype. (c) Phase-contrast microscopy of C. crescentus dif2_CC-deleted mutant during log phase at 1,000× magnification. The white arrow shows a threadlike appendix which links two chained cells. (d) Phase-contrast microscopy of C. crescentus CcxerD mutant during log phase at 1,000× magnification. (e) Phase-contrast and fluorescence microscopy of DAPI-stained C. crescentus wild type during log phase at 1,000× magnification. (f) Phase-contrast and fluorescence microscopy of DAPI-stained C. crescentus dif2_CC-deleted mutant during log phase at 1,000× magnification. Arrows show a delicate threadlike appendix that contains chromosomal DNA and links two chained cells.

DISCUSSION

Replication of circular chromosomes in bacteria generates chromosome dimers in a subpopulation of cells due to an odd number of crossover events that occur during DNA damage repair by homologous recombination. A previous study, based on searching for similarity with other bacterial dif sites, had suggested that a chromosomal site, dif1_CC, is involved in chromosome dimer resolution in C. crescentus.

Here, our findings reveal that it is unlikely that chromosome dimers are resolved at dif1_CC, since our EMSA results showed that even at high protein concentrations, dif1_CC is not bound by CcXerD. Also, our in vitro cleavage assays demonstrated that the CcXerCD recombination system could not cleave dif1_CC. On the other hand, our data confirm that the other Caulobacter dif candidate, dif2_CC, is the correct recombination site for chromosome dimer resolution in Caulobacter. We observed a high level of cooperativity between CcXerC and CcXerD in binding to dif2_CC. Furthermore, our in vitro cleavage assays showed that the Caulobacter Xer recombination system was much more effective in cleaving the top strand of dif2_CC when both recombinases were present in the reaction mixture. This is in line with the results that have been reported for dif cleavage in both E. coli and B. subtilis, in which the largest amount of recombinase-DNA complex was generated when both XerC and XerD were added to the reaction (22, 23). These authors also observed small but detectable XerD-specific cleavage of dif bottom strand, which we were also able to detect (Fig. 4C, lane 3).

It should be noted that even though our results strongly suggest that dif1_CC is not implicated in chromosome dimer resolution, the binding affinity of MBP-CcXerC for dif1_CC is not negligible (Fig. 2). The existence of two dif sites in the bacterial chromosome is not unique to C. crescentus. For instance, the Vibrio cholerae larger chromosome, ChrI, contains two overlapping dif sites, which serve as substrates for two different XerC/D-mediated processes: one dif site for the integration of a toxin-encoding bacteriophage, CTXФ, and the other for chromosome dimer resolution (14). Recent studies have shed light on the integration mechanisms of IMEXs (integrative mobile elements exploiting Xer), which hijack the host XerCD recombination system for their integration into the host chromosome (28). We have not found any evidence of recombination between the dif_CC sites by either standard or nested PCR experiments (data not shown). Furthermore, the product of an open reading frame found between the dif sites displays similarity to the Xre family of transcriptional regulators and does not show any similarity to recombinases or transposases, which leads us to believe that the region between the dif_CC sites is not an IMEX element. In addition, the dif_CC sites are in an inverse orientation (Fig. 1A), making the excision/reintegration of this region unlikely.

If dif2_CC is the main chromosome dimer resolution site, then the deletion of dif2_CC would result in chromosome partitioning failures in a subpopulation of cells that generate chromosome dimers. To test this, we constructed a markerless deletion in which an intergenic region of 45 bp including dif2_CC is deleted. As expected, about 5% to 10% of cells showed abnormal DNA distributions, suggestive of altered chromosome segregation and failures in the completion of cell division (Fig. 5). The dif2_CC deletion caused a noticeable cell chain phenotype where some cells in the population, with uneven distribution of DNA, are connected to each other by delicate and threadlike appendices (Fig. 5f). Interestingly, this phenotype is similar in morphology to the phenotype that has been reported for Caulobacter when the cells are depleted of the C terminus of FtsK (12). In E. coli, among other things, FtsK_C is implicated in chromosome dimer resolution by directly activating XerD strand exchange within the synaptic complex (11). It is likely that at least one possible source of the phenotype observed in Caulobacter FtsK_C-deficient cells is the inability of cells to activate CcXerD strand exchange within the synaptic complex.

The dif2_CC phenotype was different from what has been reported for the E. coli dif-deleted mutant in which twin filaments in the latter are produced as a result of the guillotining of chromosome dimers during cell division (29, 30). Apparently, E. coli does not have a system to detect division site blockage by unresolved chromosome dimers; therefore, when the latter becomes trapped at the division plane during cell division, it is guillotined and sheared by the constricting septum. Thus, the newborn daughter cells inherit damaged chromosomes which, in turn, will induce the SOS response, leading to the filamentation of the two newborn daughter cells, called twin filaments. In E. coli filamentation is performed by the action of SulA protein, an SOS-induced cell division inhibitor which directly inhibits the polymerization of the FtsZ ring (31). Although sulA is restricted to gammaproteobacteria, other bacterial species have evolved different checkpoint mechanisms for delaying cell division in the case of DNA damage.

In Caulobacter, cell division becomes inhibited when the DNA is damaged through two redundant mechanisms: SOS induced and SOS independent (27, 32). The SOS-induced checkpoint involves the RecA induction of a small cell division inhibitor protein, SidA, which interacts with the cell division protein FtsW and thereby blocks the divisome activity and constriction. However, the SOS-independent mechanism includes another small protein, DidA, which is induced in a RecA-independent manner and blocks the constriction of the divisome apparatus by making a direct interaction with another late recruited divisome component, FtsN. In the case of DNA damage in Caulobacter, the inhibition of cell division by either mechanism results in a typical filamentation phenotype which is definitely distinguishable from the Dif phenotype seen in dif2_CC deletion, ruling out the possibility of chromosome dimer guillotining by septum constriction in dif2_CC cells.

Our microscopic analysis of the dif2_CC mutant showed that about 5% to 10% of cells bore failures in cytokinesis. This Dif phenotype observed in Caulobacter is milder than what has been reported for E. coli and B. subtilis dif-deleted mutants in which 15% and 25% of cells, respectively, bore evidence of partitioning failures. One of the reasons that may explain this dissimilarity is that in Caulobacter, unlike E. coli and B. subtilis, chromosome replication initiates once per cell cycle, ruling out the opportunity of performing multifork replication and rear-end collisions, which leads to a lower number of homologous recombination events during the cell cycle (33, 34).

Curiously, a very similar phenotype has been reported for the wbqL mutant when crescentin is exogenously overproduced. Crescentin is an intracellular microfilamentous structure which is located along the inner membrane and plays a major role in the upkeep of the crescent shape in C. crescentus (35). In wild-type cells, crescentin is depolymerized at the division site during cell division and, thus, the septum closure can be fulfilled. It has been shown that WbqL, a protein involved in lipopolysaccharide metabolism, is necessary for crescentin association with the inner membrane as well as its degradation at the division site during septum constriction. Thus, in the wbqL mutant the overproduced crescentin filaments do not degrade at the division site during cell division; as a result, the cells fail to accomplish the last stages of cell division and remain in chains linked by threadlike appendices which contain crescentin filaments encased by the cell envelope (36). This phenotype is morphologically similar to what we have reported for dif2_CC deletion, except that in the wbqL mutant septum closure is blocked by the presence of nondepolymerized crescentin, but in the dif2cc cells the division plane is impassable for the constricting septum because of unresolved chromosomal DNA trapped at the division site. This suggests that the presence of unresolved chromosomes or nondepolymerized crescentin at the cell septum leads to cell separation failures in Caulobacter, resulting in the formation of threadlike appendices which hold the cells in chains but do not block cell cycle progression.

Recent studies have uncovered the role of a divisome protein, FtsE, in the late stages of cytokinesis in C. crescentus (37). FtsE is involved in the organization of FtsZ at the midcell and simultaneously in inner membrane fusion by activating the remodeling of peptidoglycan synthesis at the division site in the final stages of cell division. Cells lacking FtsE generate chained cells interconnected by thin structures morphologically comparable to threadlike appendices observed in our dif2_CC-deleted mutant.

Several questions regarding the Caulobacter Dif phenotype remain to be answered in the future, including how Caulobacter dif-deleted mutants can detect the DNA that lingers between two incompletely separated chromosomes at the cell division plane during cell division and avoid guillotining and also how the development of threadlike structures is regulated in the Caulobacter dif mutant.

MATERIALS AND METHODS

Bacterial strains, plasmids, and primers.

All C. crescentus strains used in this study were grown at 29°C in peptone-yeast extract (PYE). Medium was supplemented when required with 3% sucrose and 5 μg/ml kanamycin, except for solid medium, where 25 μg/ml of kanamycin was used. All E. coli strains were grown at 37°C in LB and supplemented with 50 μg/ml of kanamycin and 100 μg/ml of ampicillin when applicable. All bacterial strains and plasmids used in this study are listed in Table 1, and all primer sequences used in this study are listed in Table 2. Genomic DNA was extracted using a Presto Mini gDNA bacterium kit (Geneaid) by following the manufacturer’s instructions.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Feature(s)	Reference or source
Plasmids
pUC19	Cloning vector	NEB
pET28a	N terminus His tag fusion expression plasmid	Novagen
pMAL-c2	Maltose-binding protein fusion expression plasmid	NEB
pMEG375	DNA exchange vector, sacB, Cmr	41
pNPTS138	DNA exchange vector, sacB, Kanr	M. R. K. Alley, unpublished data
pLJ1	CcxerD in pMAL-c2	16
pLJ2	CcxerC in pMALc2	16
pAF201	dif2CC flanking regions in pMEG375	This study
pAF202	pNPTS138-derived plasmid for deletion of dif2cc	This study
pAF204	dif2CC fragment in pUC19	This study
pAF205	dif1CC fragment in pUC19	This study
pAF206	CcxerD in pET28a	This study
pAF207	pNPTS138-derived plasmid for disruption of recA	This study
pAF208	pNPTS138-derived plasmid for disruption of CcxerD	This study
Strains
DH5α	Cloning strain	NEB
T7 express	E. coli protein overexpression strain	NEB
NA1000	C. crescentus synchronizable strain	42
CAF01	NA1000 Δdif2CC	This study
CAF02	NA1000 CcxerD::pAF208	This study
CAF03	NA1000 recA::pAF207	This study

TABLE 2.

Primers and oligonucleotides used in this study

Primer or oligonucleotide	Sequence
Primers
NSmaIDifIFW	NNNNNNCCCGGGGAGCAGGTTCAGATAGCTGGC
NSmaIDifIRE	NNNNNNCCCGGGCTGCGTCATGGTCGTTCGATC
NSmaIDifIIFW	NNNNNNCCCGGGGTTCCTGAAGCGTGATATGGTCGT
NSmaIDifIIRE	NNNNNNCCCGGGGTCCTCACGACAGAAACTTTG
NNdeIXerDfw	NNNNNNCATATGAGCCAGGGCGAGGCCTG
NNheIXerDre2	NNNNNNGCTAGCCTACCCCTTCTTGCGCCCCA
BamHI-RfwdifII	NNNNNNGGATCCGTGTTGCACATCACCTGTCACG
NXbaI-RredifII	NNNNNNTCTAGAGCGGTCGATCTCAATGACGA
NSmaI-LfwdifII	NNNNNNCCCGGGTGGCTTGCAGAAGTAGCTGTCA
NBamHI-LredifII	NNNNNNGGATCCTGGATCTGCAACGGCGGATT
6NHindIIIxerDintFW2	NNNNNNAAGCTTCTCGAAATGATGGCGGTCGA
6NBamHIxerDintRE2	NNNNNNGGATCCCTTCCAGCAAGTGCGTGGCG
6NBAmHIRecAintRE	NNNNNNGGATCCGTTGATGAAGATGACGATGGTGT
6NHindIIIRecAintFW	NNNNNNAAGCTTGCAAGGTCGAGATCGAGTCG
Oligonucleotides (used in suicide substrate assays)
ccdif2TSRL	CATAGTTCGTCTAAGATATATTATCAGAAGAATG
ccdif2BSR-5'HEX	CATTCTTCTGATAATAT
ccdif2BSL	ATCTTAGACGAACTATG
ccdif2BSRL	CATTCTTCTGATAATATATCTTAGACGAACTATG
ccdif2TSL-5'HEX	CATAGTTCGTCTAAGAT
ccdif2TSR	ATATTATCAGAAGAATG
ccDif1_5'HEX_TSL	TCAAAGATCGACTTTGT
ccdif1_TSR	AATTTATGTAAAGTTGT
ccDif1_TSL+R	TCAAAGATCGACTTTGTAATTTATGTAAAGTTGT
ccDif1_5'Hex_BSR	ACAACTTTACATAAATT
ccDif1_BSL	ACAAAGTCGATCTTTGA
ccDif1_BSL+R	ACAACTTTACATAAATTACAAAGTCGATCTTTGA

pLJ1 and pLJ2, expressing N-terminal MBP fusions to CcXerD and CcXerC, respectively, were constructed as described by Jouan and Szatmari (16). The sequence of dif1_CC was derived from the work of Jensen (17), and the dif2_CC sequence was obtained from the work of Kono et al. (19). To create pAF205, a 310-bp fragment of the NA1000 genome carrying dif1_CC was amplified using primer pair NSmaIDifIFW and NSmaIDifIRE. The PCR product then was inserted into SmaI-linearized pUC19. For constructing pAF204, a 270-bp fragment that carries dif2_CC was amplified from NA1000 genomic DNA by using the primer pair NSmaIDifIIFW and NSmaIDifIIRE. The PCR product then was inserted into SmaI-linearized pUC19. pAF206 was created by amplifying the CcxerD coding sequence from NA1000 genomic DNA using primer pair NNdeIXerDfw and NNheIXerDre2, and then the PCR product was inserted into NdeI/NheI-linearized pET28a. pAF202 was constructed by amplifying the dif2_CC flanking regions using BamHI-RfwdifII, NXbaI-RredifII, NSmaI-LfwdifII, and NBamHI-LredifII primers. dif2_CC flanking regions were cloned into pMEG375 and ligated with a BamHI site to create pAF201. A fragment containing the ligated flanking regions then was amplified from pAF201 using primer pair NSmaI-LfwdifII/NXbaI-RredifII, followed by insertion into EcoRV-linearized pNPTS138 to create pAF202. pAF207 was created by inserting a 521-bp internal fragment of Caulobacter recA to EcoRV-linearized pNPTS138. A recA internal fragment was amplified from NA1000 genomic DNA using 6NHindIIIRecAintFW and 6NBamHIRecAintRE primers. pAF208 was created by amplifying a 744-bp internal fragment of CcxerD from NA1000 genomic DNA by using 6NHindIIIxerDintFW2 and 6NBamHIxerDintRE2 primers, and then the amplified fragment was inserted into EcoRV-linearized pNPTS138. The plasmid pMal-CcXerCYF, containing a mutation which substitutes a phenylalanine for the active-site tyrosine, was constructed by site-directed mutagenesis in plasmid pLJ2 as described previously (38).

Protein expression and purification.

The CcxerD coding sequence was cloned into pET28a, introduced into T7 Express cells, and purified as an N-terminal 6×His fusion protein. Five hundred milliliters of Terrific Broth medium supplemented with 50 μg/ml of kanamycin was inoculated with 5 ml of overnight starter culture and grown up to an optical density at 600 nm of 0.4. Isopropyl-β-d-thiogalactopyranoside (IPTG) then was added to a final concentration of 1 mM and expression continued for 3 h. Cells were harvested and resuspended in 20 ml of binding buffer containing 50 mM NaH₂PO₄, 300 mM NaCl, and 20 mM imidazole. The solution was subjected to two freeze-thaw cycles followed by 10 cycles of sonication (10 s) separated by 20-s intervals, and the lysate was centrifuged at 10,000 × g for 15 min at 4°C. The supernatant was applied to a nickel-nitrilotriacetic acid Superflow cartridge by following the manufacturer’s instructions (Qiagen). Purified CcXerD-6×His was eluted in elution buffer containing 300 mM imidazole. MBP-CcXerC and MBP-CcXerD were purified as described by Jouan and Szatmari (16). The MBP-CcXerC^Y-F protein was purified as described in reference 38.

EMSA.

Electrophoretic mobility shift assay (EMSA) was performed by using specific DNA fragments previously cloned in pUC19 that were labeled by performing PCR using M13F and 5′-6-HEX-labeled M13R primers. Thirty-three-nanogram samples of labeled DNA fragments were incubated at 29°C in 1× TENg-binding buffer (20 mM Tris-HCl [pH 7.5], 1 mM EDTA, 25 mM NaCl, and 5% glycerol), 500 ng of salmon sperm DNA, and various concentrations of CcXerC/CcXerD recombinases. After 30 min of incubation, reaction mixtures were loaded on a native 6% polyacrylamide gel that had been run for 1 h in 0.5× Tris-borate-EDTA at 100 V. DNA fragments were visualized using a Typhoon 9410 imager and analyzed by ImageQuant software (GE Healthcare).

In vitro cleavage assay.

The ability of CcXerC/CcXerD to cleave hypothetical dif sites was addressed by using 5′-6-HEX-labeled double-stranded dif suicide substrates that contain a nick at the spacer, either at the top or the bottom strand. The suicide substrates were constructed by annealing 3 oligonucleotides (Table 2), as described by Leroux et al. (9). One hundred-picomole samples of 5′-6-HEX-labeled dif suicide substrates were incubated at 29°C in 1× TDMNG buffer (50 mM Tris [pH 7.5], 25 mM NaCl, 7.5 mM MgCl₂, 5 mM dithiothreitol, and 25% glycerol), 500 ng of salmon sperm DNA, and different concentrations of CcXerC/CcXerD recombinases. After 2.5 h of incubation, the reactions were stopped by adding 5 μl of 2% SDS and 5 μl of 11 mM EDTA and incubated at 95°C for 10 min. The reaction mixtures then were subjected to electrophoresis on a denaturing 6% polyacrylamide gel containing 0.1% SDS at 100 V and were scanned by a Typhoon 9410 imager.

Construction of dif2_CC-deleted, CcxerD-disrupted, and dif2_CC recA double mutant strains CAF01, CAF02, and CAF03.

CAF01 was constructed from C. crescentus CB15N by replacing a 45-bp intergenic region, downstream of CCNA-01840 and upstream of CCNA-01837, containing dif2_CC, with a BamHI restriction site. Markerless deletion of dif2_CC was performed by double crossover homologous recombination as described by Ried and Collmer (39), using pAF202, a pNPTS138-derived plasmid which contains two chromosomal dif2_CC flanking regions ligated together with a BamHI restriction site.

CAF02 was constructed from C. crescentus CB15N by inactivating CcxerD (CCNA-03101) via plasmid integration as follows. An internal fragment of CcxerD was amplified and cloned into pNPTS138 to create pAF208 as described above. pAF208 was electroporated to wild-type CB15N cells and integrated into CcxerD by homologous recombination. The integration of pAF208 into CcxerD was confirmed by PCR. CAF03 was constructed from a dif2_CC background by inactivating recA (CCNA-01141) via plasmid integration as follows. An internal fragment of recA was amplified and cloned into pNPTS138 to create pAF207, as described above. pAF207 was electroporated to dif2_CC deletion cells and integrated into recA by homologous recombination.

Microscopy.

The Dif phenotype in the dif2_CC-deleted mutant was analyzed by light and fluorescence microscopy using Nikon Eclipse E600 and Nikon Eclipse Ti2 microscopes, respectively. For this purpose, cells were treated as described by Chen et al. (40). In brief, cells were fixed during log phase by direct addition of formaldehyde and glutaraldehyde to the cell culture to the final concentrations of 3.2% and 0.0002%, respectively. The fixed cells then were washed three times by PBS and allowed to adhere to a slide pretreated with a 0.1%, wt/vol, poly-l-lysine solution. Afterwards, cells were incubated with 0.3 μg/ml of DAPI in PBS for 5 min and the slide was washed two times with PBS before microscopic analysis. Cells were analyzed by 40× and 100× objectives. Images were acquired by Nikon NIS-Elements imaging software.