|Home | About | Journals | Submit | Contact Us | Français|
Bacterial cell morphogenesis requires coordination among multiple cellular systems, including the bacterial cytoskeleton and the cell wall. In the vibrioid bacterium Caulobacter crescentus, the intermediate filament-like protein crescentin forms a cell envelope-associated cytoskeletal structure that controls cell wall growth to generate cell curvature. We undertook a genetic screen to find other cellular components important for cell curvature. Here we report that deletion of a gene (wbqL) involved in the lipopolysaccharide (LPS) biosynthesis pathway abolishes cell curvature. Loss of WbqL function leads to the accumulation of an aberrant O-polysaccharide species and to the release of the S layer in the culture medium. Epistasis and microscopy experiments show that neither S-layer nor O-polysaccharide production is required for curved cell morphology per se but that production of the altered O-polysaccharide species abolishes cell curvature by apparently interfering with the ability of the crescentin structure to associate with the cell envelope. Our data suggest that perturbations in a cellular pathway that is itself fully dispensable for cell curvature can cause a disruption of cell morphogenesis, highlighting the delicate harmony among unrelated cellular systems. Using the wbqL mutant, we also show that the normal assembly and growth properties of the crescentin structure are independent of its association with the cell envelope. However, this envelope association is important for facilitating the local disruption of the stable crescentin structure at the division site during cytokinesis.
Most bacterial species display a particular cellular morphology that is generally preserved across generations. The production and maintenance of shape require coordination among multiple cellular systems positioned at different places within the cell. The peptidoglycan cell wall, located external to the cytoplasmic membrane, is an important structural element that is required for shape maintenance. The processes governing the localization and timing of cell wall growth and turnover are likewise critical (8, 9, 23). The bacterial cytoskeleton is thought to play a central role in cell morphogenesis by exerting spatiotemporal control over peptidoglycan growth (8, 9, 23). In order for it to do so, there are numerous proteins that are required to connect cytoskeletal control mechanisms to the periplasmic enzymes that directly synthesize and modify the peptidoglycan cell wall. These proteins, such as MreC, MreD, RodA, and RodZ, are essential for maintenance of cell shape and are positioned in the cytoplasmic membrane to presumably link cytoskeletal elements in the cytoplasm to the activities of peptidoglycan-modifying enzymes in the periplasm (2, 5, 9, 23, 29). Some bacterial species also contain additional components that make important contributions to cell shape, such as cell wall teichoic acids in Gram-positive bacteria (9) and periplasmic flagella in spirochetes (41).
In the vibrioid bacterium Caulobacter crescentus, an intermediate filament-like protein, crescentin, is required for cell curvature (3). Crescentin forms an intracellular filamentous structure that is associated with the cell wall and is thought to mechanically govern cell wall growth to produce cell curvature (7). The crescentin structure is localized along the inner curvature of the cell under the cytoplasmic membrane (3, 7) and is highly stable, with no detectable subunit exchange (10). The association between the crescentin structure and the cell envelope appears essential for its function, since an attachment-defective crescentin mutant is unable to support cell curvature (7). The function of the actin-like protein MreB is also critical for the envelope association of the crescentin structure (10), and MreB may provide one part of the connection between the crescentin structure and the peptidoglycan cell wall.
Since bacterial morphogenesis requires multiple cellular components and systems, we used a genetic screen to find other factors important for cell curvature in C. crescentus. Surprisingly, we found that an alteration in the lipopolysaccharide (LPS) biosynthesis pathway can have a catastrophic effect on the ability of the crescentin structure to associate with the cell envelope and govern cell curvature.
C. crescentus was grown at 30°C in PYE (peptone-yeast extract) or M2G supplemented with 1% PYE (M2G+) (16). Vanillic acid (0.5 mM) or xylose (0.3%) was added to the medium to induce the Pvan or Pxyl promoter, respectively, when appropriate. Plasmids were mobilized from Escherichia coli strain S17-1 into C. crescentus by conjugation (16) or electroporation. Plasmids and strains are listed in Table Table1,1, and their construction and primers are described in the text of the supplemental material. Exponentially growing cultures were used for all experiments unless otherwise noted. For transposon mutagenesis, a plasmid bearing the mini-Tn5 Km2 (LS1680) was mated into CJW940 or a plasmid bearing the Himar1 element (pSC189) (11) was electroporated into CJW1034. UV mutagenesis was performed on strain CJW1126. Two 100-μl drops of an overnight culture (in stationary phase) were placed on a petri dish, set atop a Spectroline TVC-312R UV transilluminator, and irradiated for 40 s at the maximum intensity setting.
Cells were imaged at room temperature (~22°C) or at 31°C using an objective heater on either a Nikon E1000 microscope fitted with 100× differential interference contrast (DIC) and phase-contrast objectives and a Hamamatsu Orca-ER LCD camera or a Nikon E80i microscope with similar objectives, an Andor iXon+ camera, and a Hamamatsu Orca-IIER LCD camera. The latter setup and the Andor camera were used for photobleaching with a Photonic Instruments Micropoint laser system with a 481-nm dye cell. Photobleaching was performed with 10 pulses with power 2 at the attenuator plate and 18% transmission at the internal attenuator, using an ND8 filter. Cells were immobilized on 1% agarose-PYE or M2G+ agarose-padded slides containing vanillic acid or xylose inducers when appropriate. Immunofluorescence microscopy was performed as previously described (3) using anticrescentin antibodies at 1:800. Images were taken and analyzed with the Metamorph software program (MDS Analytical Technologies). For live-cell immunofluorescence, cells were incubated with 1:200 anti-“smooth” LPS (S-LPS) in PYE medium for 15 min at 30°C, washed twice with PYE, incubated with 1:200 fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit immunoglobulin in PYE for 15 min, washed twice with PYE, resuspended in M2G+, and mounted on M2G+-1% agarose pads for microscopy.
Automated cell curvature analysis was performed using Matlab as described previously (7). Briefly, cell outlines were obtained from phase-contrast images, and the center line of each cell was detected. In order to find the curvature of the cells, multiple points located equidistantly along the center line were fitted to an arc of a circle, minimizing the sum of the squared distances to the centerline points from their projections to the arc. The reciprocal of the radius of the best-fit arc is the curvature value. The few cells where the centerline shape was not representative of cell curvature (e.g., dividing cells that bend at the constriction site) were removed prior to statistical analysis by applying a cutoff to the curvature distribution. Statistical calculations were done using Matlab.
Affinity-purified anticrescentin antibodies were used for immunoblotting at a dilution of 1:15,000. The load volume was normalized to optical density at 600 nm (OD660) values. Polysaccharide isolation for LPS immunoblot analysis was performed as previously described (38). Specifically, cells were washed with 10 mM HEPES, pH 7.2, resuspended in 10 mM Tris-1 mM EDTA, and frozen in a volume of 250 μl. Samples were thawed, combined with 1 μl DNase (0.5 mg/ml), 20 μl lysozyme (10 mg/ml), and 3 μl 1 M MgCl2, and incubated at room temperature for 15 min. The above mixture (38.75 μl) was then combined with 10 μl of 5× SDS dye, incubated at 100°C for 10 min, and cooled to room temperature. Proteinase K (1.25 μl of a 20-mg/ml stock solution) was added, and the sample was incubated at 60°C for 1 h. Samples were analyzed on 10% or 12% acrylamide gels by SDS-PAGE. Gels were silver stained according to a previously described method (36) with minor modifications (38). Blots were generally probed with anti-S-LPS antibodies (37) at a dilution of 1:100,000 for 15 to 30 min. For anti-RsaA (S-layer) blotting, total cell extracts (normalized by OD660) or isolated shed material was used. Shed material was collected on a nylon mesh, washed extensively with 1 M Tris (pH 7.5), and then dissolved in 0.1 M Tris (pH 7.5)-8 M urea. Anti-RsaA was used at 1:10,000.
CB15N cells (800 ml) at an OD660 of 0.3 were pelleted (12,000 × g) and resuspended in 28 ml of 10 mM Tris (pH 7.5)-0.75 M sucrose at 4°C, at which temperature all the following steps were performed. Lysozyme was added to 50 μg/ml, an EDTA-free complete protease inhibitor cocktail tablet (Roche) was added, and then 56 ml of cold 1.5 mM EDTA (pH 7.5) was added dropwise with constant swirling. The cells were incubated on ice for 10 min before being lysed in a French press set to ~750 lb/in2 (a very low pressure that results in only a fraction of the cells being lysed; higher pressures cause the formation of predominately mixed inner/outer membrane vesicles that cannot be separated by density). The lysate was cleared by centrifugation (16,000 × g), and the cleared lysate was ultracentrifuged at 160,000 × g for 1.5 h. The membrane pellet was resuspended by Dounce homogenization in 3.3 mM Tris (pH 7.5)-1 mM EDTA-0.25 M sucrose, and 500 μl was carefully laid over a 35 to 60% sucrose (in 5 mM EDTA [pH 7.5]) step gradient in Beckman-Coulter SW41 tubes. The tubes were centrifuged for 40 h at 140,000 × g at 4°C. Fractions (500 μl) were collected by hand, and 4 volumes of each fraction were mixed with 1 volume of 5× SDS loading buffer. The samples were separated by 10% PAGE, transferred to polyvinylidene difluoride (PVDF) membranes, and immunoblotted.
Sacculi were isolated from 1-liter PYE cultures grown without antibiotics at 30°C and harvested in mid-log phase (OD660 between 0.3 and 0.4). Under these conditions, CB15N doubles in 90 min, and CJW1090 and CJW1249 double in 95 min. Isolation, identification, and high-performance liquid chromatography (HPLC) analysis of muropeptides were performed as described previously (19) with previously published modifications (34).
To find genes involved in the generation or maintenance of normal cell morphology, we made libraries of mutant C. crescentus cells by UV or transposon mutagenesis. To increase the randomness of transpositions, we independently used either the Tn5-derived mini-Tn5 Km2 or the mariner-based Himar1. We then visually screened these mutant libraries for cell shape defects by DIC microscopy. UV-generated mutations of interest were mapped by complementation using a cosmid library in a 96-well plate format, while transposon insertions were identified by cloning the transposon and surrounding sequence into a vector and by sequencing from the transposon sequence.
From more than 5,000 transposon mutants screened, 8 had a cell curvature defect. This defect was quantified by computerized image analysis (7) to obtain numerical values for cell curvature (see Materials and Methods). For comparison, wild-type cells have a median cell curvature of 0.39 μm−1 (Fig. (Fig.11 A, panel ii), whereas crescentin-null mutant cells, which have virtually no detectable curvature, have a median cell curvature of 0.06 μm−1 (Fig. (Fig.1A,1A, panel i). Four of the 8 mutants with curvature defects had a Tn5 insertion in the crescentin-encoding gene creS. Transposon insertions in rcdA, encoding a CtrA degradation regulator, the gene cc_0884, encoding a predicted transcriptional regulator, and the cell division gene ftsH also produced cells with reduced curvature, but these mutants were elongated compared to the wild type (data not shown) and therefore were not analyzed further. However, a Himar1 insertion in the gene cc_0632 of hypothetical function produced cells with normal lengths but with a considerable reduction of cell curvature (0.2 μm−1; Fig. Fig.1A,1A, panel iii).
The cc_0632 gene is the penultimate gene of a predicted operon encompassing cc_0635 to cc_0631 (Fig. (Fig.1B),1B), suggesting a potential polar effect on the downstream cc_0631 gene. Consistent with this notion, a UV-generated mutant with a straight cell shape (0.11 μm−1) was found to have a point mutation in cc_0631 (Fig. (Fig.1A,1A, panel v). A cosmid (cos1-10-A) containing cc_0631 and flanking regions complemented the UV mutant (0.39 μm−1; Fig. Fig.1A,1A, panel vi), whose point mutation in cc_0631 should have no effect on the expression of cc_0632. Additionally, the cc_0632::Himar1 cell shape defect could be rescued with a plasmid bearing the cc_0631 sequence driven by Plac (0.32 μm−1; Fig. Fig.1A,1A, panel iv), suggesting that a polar effect on the expression of cc_0631 was responsible for the moderate curvature defect of cc_0632::Himar1 cells. To confirm this, we specifically interrupted the cc_0631 gene by plasmid insertion (generating strain CJW1090) (Fig. (Fig.1A,1A, panel vii). This gene disruption resulted in the same loss of curvature as observed for the UV mutant (0.11 μm−1) and could be complemented by plasmid-borne cc_0631 (0.38 μm−1) (Fig. (Fig.1A,1A, panel viii). Collectively, these results indicate that the proper functioning of the cc_0631 gene product is required for cell curvature.
We noticed that stationary-phase liquid cultures of cc_0631 mutant cells contained debris that was visible on the sides of culture tubes (Fig. (Fig.1C).1C). In C. crescentus, the surface of the cell is covered by an S layer, a hexagonal protein lattice (32). Since there are some C. crescentus mutants that are known to shed S-layer proteins (37), we tested to see whether this debris contained S layer, which is made of the RsaA protein (15, 31). Immunoblots of whole-cell extracts of wild-type CB15N cells displayed a clear band of RsaA at ~130 kDa (15) (Fig. (Fig.1D,1D, lane 1), while cells with an antibiotic resistance cassette in the rsaA gene (JS1003) lacked detectable protein (Fig. (Fig.1D,1D, lane 2). Along with these controls, we analyzed the debris shed by the cc_0631 disruption mutant and the cc_0631 UV mutant. The debris was isolated from stationary-phase cultures on nylon mesh, washed, and solubilized in urea. The lanes containing this debris exhibited the characteristic RsaA band (Fig. (Fig.1D,1D, lanes 3 and 4), showing that the shed material contained S layer. Some S-layer material remained associated with the cell (Fig. (Fig.1D,1D, lane 5).
Since the paracrystalline array of S-layer protein is attached to the cell surface through the O-polysaccharide portion of lipopolysaccharide (LPS) in the outer membrane and since mutants in LPS synthesis shed S layer (37), we tested whether cc_0631 mutants produced altered LPS. To do so, we used an antibody that is specific for “smooth” LPS (S-LPS) (37). S-LPS is composed of lipid A, to which core sugar and O-polysaccharide are linked. Since this antibody does not recognize “rough” LPS containing only lipid A and core sugar (26), we assume that this antibody is specific for the O-polysaccharide portion of LPS. In Escherichia coli, the lipid-bound core and O-antigen oligosaccharides are synthesized separately at the inner face of the cytoplasmic membrane before being flipped onto the periplasmic face (40). There, multiple O-antigen oligosaccharides are ligated to the lipid-bound core, which is then shuttled to the outer membrane (24). Anti-O-polysaccharide immunoblots of C. crescentus polysaccharide extracts typically display a single band, corresponding to a homogeneous O-polysaccharide length (37). We confirmed this length homogeneity in wild-type cells (Fig. (Fig.22 A, lane 1). As a negative control, we used a transposon mutant of wbqP (4) that does not produce a detectable O-polysaccharide band (lane 2, Fig. Fig.2A).2A). Consistent with an absence of O-polysaccharide, WbqP has been proposed, based on sequence homology, to initiate the formation of O-polysaccharide in the cytoplasm by conjugating the first sugar to the undecaprenol lipid carrier (4). Both cc_0631 mutants (UV-generated and gene disruption; lanes 3 and 4 in Fig. Fig.2A,2A, respectively) showed heterogeneity in O-polysaccharide length, with multiple bands of shorter length than that for the wild type, with the shortest band being the most abundant. Addition of cc_0631 on a low-copy-number plasmid in the cc_0631 disruption mutant (strain CJW1117) restored both wild-type cell curvature (Fig. (Fig.1A,1A, panel viii) and wild-type O-polysaccharide length uniformity (Fig. (Fig.2A,2A, lane 5). These data indicate that the CC0631 protein plays a role in LPS biosynthesis in C. crescentus and that defects in LPS production are responsible for the S-layer shedding phenotype of cc_0631 mutants.
Prior to this work, a screen had been performed to identify mutants that were defective in attachment of the S layer (4). Through this screen, a wbqL transposon mutant was identified (4). At the time, the mutant was characterized, but the sequence or reference number of the wbqL gene was not made publicly available until after we had determined, by comparing the gene length and those of neighboring genes, that their mutation was in cc_0631, which will here be called wbqL. Awram and Smit (4) showed that a wbqL::Tn5 mutant sheds S layer and produces altered LPS but made no mention of a defect in cell shape. We confirmed that their wbqL::Tn5 mutant had the same curvature defect that we observed with our wbqL mutants (data not shown).
BLAST alignment of the predicted WbqL protein sequence revealed that it matches the COG4421 superfamily of bacterial capsular polysaccharide biosynthesis proteins and shows substantial homology to a number of these proteins. The UV point mutation in wbqL that abolishes cell curvature (Fig. (Fig.1A,1A, panel v) causes a tryptophan-to-arginine substitution at position 138 (W138R). Tryptophan 138 is located in a region of the sequence displaying high levels of conservation among different polysaccharide biosynthesis proteins (data not shown), in accord with this substitution affecting protein function. Consistent with an earlier alignment analysis suggesting that WbqL is a glycosyltransferase (4), HHpred structural prediction analysis (33) matched WbqL to general glycosyltransferase structure as well as to E. coli WaaF, which is ADP-heptose-LPS heptosyltransferase II. WaaF is responsible for ligating the second heptose unit to the core sugar of E. coli LPS (25). The composition of the core sugar of C. crescentus LPS is slightly different from E. coli (27), and its assembly pathway has not yet been determined. Because genes putatively involved in C. crescentus LPS biosynthesis have been identified solely based on abnormal or absent LPS in mutants and by sequence comparison to known LPS biosynthesis proteins (4), it is difficult to assign WbqL a specific function in LPS biosynthesis. In any case, it is clear that without WbqL, the cells accumulate an aberrant O-polysaccharide species (4).
Since the altered LPS species and associated S-layer shedding in wbqL mutant strains were correlated with a straight cell shape, we next examined whether cell curvature required the production of either S layer or O-polysaccharide. To test the dependence of curvature on S layer, we used an rsaA mutant strain (JS1003). These cells, which produce no detectable S layer (Fig. (Fig.1D,1D, lane 2), had a curvature value of 0.32 μm−1 (Fig. (Fig.2B),2B), which is close to the wild-type value (0.39 μm−1). Therefore, cell curvature does not require S-layer production. Similarly, synthesis of O-polysaccharide is dispensable for cell curvature, since wbqP::Tn5 cells (CJW1249) displayed near-normal curvature (0.35 μm−1) (Fig. (Fig.2B2B).
Even though wbqL mutants shed the S layer and have altered O-polysaccharide, disruption of S layer or O-polysaccharide synthesis does not result in a defect in cell curvature. It was not clear whether the alteration in O-polysaccharide was directly related to the cell curvature defect. It was possible that WbqL might have a second role, unrelated to O-polysaccharide synthesis, which might result in the straight-rod shape in the wbqL mutant. Because wbqP mutant cells do not produce any detectable O-polysaccharide, wbqP would be predicted to be upstream of wbqL in the pathway of O-polysaccharide biosynthesis. We therefore created a wbqP wbqL double mutant (CJW1908), which would produce no O-polysaccharide (and therefore no altered O-polysaccharide either), and examined the shape of the cells. If a wbqP mutation in the wbqL null background could restore wild-type curved cell morphology, this would indicate that the altered O-polysaccharide species likely causes the straight-rod shape. If a wbqP mutation in the wbqL null background could not rescue wild-type curved cell morphology, this would support a model in which WbqL has dual roles, and the abrogation of WbqL's second role would cause the straight-rod shape independently from its effect on LPS.
The wbqP wbqL double mutant strain exhibited curvature (0.37 μm−1) virtually indistinguishable from that of the wild type (Fig. (Fig.2B)2B) and did not produce any detectable O-polysaccharide (Fig. (Fig.2C,2C, lane 5). Supplying wbqP on a low-copy-number plasmid in this strain background (CJW1917) allowed the production of the aberrant O-polysaccharide species (Fig. (Fig.2C,2C, lane 6) and led to cell curvature loss (0.11 μm−1; Fig. Fig.2B)2B) identical to that of a wbqL mutant alone (Fig. (Fig.1A,1A, panel vii). This result suggested that it is the generation of an altered O-polysaccharide species, rather than the loss of normal O-polysaccharide, that causes the curvature defect. This strongly argues against a secondary function of WbqL that specifically relates to cell curvature, instead suggesting that a loss of cell curvature in a wbqL mutant is specifically due to the production of an altered O-polysaccharide species.
Since crescentin plays an essential role in cell curvature, we asked whether crescentin production, structure, or function was disrupted in wbqL mutants. Immunoblot analysis showed that crescentin was present in wbqL mutant extracts, with no visible degradation or size shift (Fig. (Fig.33 A), indicating that the cell curvature defect was not caused by the absence or degradation of crescentin. Immunofluorescence microscopy revealed that while crescentin formed filamentous structures in wbqL mutant cells, these structures were S shaped and seemingly in the cytoplasm (Fig. (Fig.3B)3B) instead of being linear and membrane juxtaposed along one side of the cells as seen in wild-type cells (3). This mislocalization was also visible in live cells (CJW2876) when crescentin was coproduced with monomeric green fluorescent protein-tagged crescentin (crescentin mGFP, which labels the endogenous crescentin structure ) (Fig. (Fig.3C).3C). This localization pattern was reminiscent of envelope-dissociated crescentin structures generated by drug treatment or crescentin truncation (7, 10). We reasoned that such dissociated structures should display increased mobility within the cell, as previously observed (7, 10). Time-lapse imaging of mGFP-labeled crescentin structures revealed visible mobility within these cells (Fig. (Fig.3D).3D). Thus, the loss of curvature in wbqL mutant cells is caused by a defect in the association of the crescentin structure with the cell envelope.
Collectively, our data suggest that an altered O-polysaccharide species in a wbqL mutant interferes with the association of the crescentin structure and the cell envelope. How might this interference occur? Since very little is known about LPS biogenesis in C. crescentus, we can only speculate. O-polysaccharide can still be exported in wbqL mutant cells, as evidenced by the presence of anti-O-polysaccharide-accessible epitopes on the surfaces of intact cells (see Fig. S1 in the supplemental material). However, it is unclear whether these epitopes represent the aberrant O-polysaccharide species or the small amount of normal-length O-polysaccharide produced in the mutant (Fig. (Fig.2C).2C). As previously observed (4), silver staining for LPS revealed a faster-migrating band in the wbqL mutant that stained with intensity similar to that of the wild type (see Fig. S2 in the supplemental material). Since undecaprenyl-linked O-antigen precursors in E. coli have been reported to stain less efficiently than lipid A-linked LPS with silver (35), this raises the possibility that the aberrant band in the wbqL mutant represents a mature (lipid A-linked) LPS species that has been exported to the cell surface. Without more information, it is difficult to envision how an exported (i.e., noncytoplasmic) aberrant LPS product could interfere with the cytoplasmic crescentin structure.
Interference could be indirect, and we considered an effect on the peptidoglycan, since chemical perturbations of peptidoglycan synthesis can cause a similar phenotype of crescentin structure detachment (7, 10). However, the HPLC profiles of peptidoglycan digests (muropeptides) of wild-type, wbqL mutant, and wbqP mutant cells were very similar (Fig. (Fig.44 A), showing identical peptidoglycan cross-linking and only a minor increase in glycan chain length in the wbqL mutant relative to the wild type (Fig. (Fig.4B;4B; see also Tables S1 and S2 in the supplemental material). Since this small increase was also present in a wbqP mutant, which produces no LPS and has normal crescentin structure localization and function, these data strongly argue against a perturbation in peptidoglycan structure or synthesis interfering with crescentin structure localization in the wbqL mutant. Still, it remains possible that some aberrant LPS intermediates might be inefficiently exported across the cytoplasmic membrane, causing them to accumulate on the inner face of the cytoplasmic membrane, where they might interfere with the crescentin structure. Direct evidence for such a disruptive interaction is difficult to obtain, since membrane fractionation methods do not differentiate between membrane leaflets, and overlap of inner and outer membrane peaks is exacerbated in wbqL mutants (data not shown), complicating interpretation. However, in membrane fractionation of wild-type cells, a faster-migrating O-polysaccharide band detectable by anti-S-LPS is enriched in cytoplasmic-membrane fractions (see Fig. S3 in the supplemental material). This result raises the possibility that the faster-migrating band may represent an intermediate that normally faces the cytoplasm. If such intermediates accumulated in wbqL mutant cells, they might conceivably occlude or disrupt interactions between the crescentin structure and cytoplasmic membrane-associated factors that keep the crescentin structure localized at the cell envelope.
While the precise mechanism disrupting the crescentin structure localization in wbqL mutants remains speculative, we clearly show that a mutation altering LPS biosynthesis affects crescentin-mediated cell curvature. The intersection of these two usually unrelated cellular processes highlights the importance of physiological balance in cells. Perturbation of one pathway can upset this balance, possibly leading to production of an aberrant product and inducing a novel, disruptive interaction with another cellular component. Disruptive interactions of this kind have also been reported for other systems. In Staphylococcus aureus, disruptions in the teichoic acid biosynthesis pathway that are usually lethal are bypassed if the first enzyme in the pathway is inactivated (13). In this case, aberrant buildup of toxic teichoic intermediates is probably at play, possibly analogous to the aberrant O-polysaccharide species seen in a wbqL mutant. Similarly, an O-polysaccharide transporter knockout in E. coli inhibits cell growth and division but not when a gene upstream in the synthesis pathway is inactivated (12). The potential for creating harmful interactions may be an evolutionary constraining force, favoring changes that refine the harmony between different cellular systems as an organism adapts to its environment.
The dissociation of crescentin structures in a wbqL disruption background provides a useful tool for probing the configuration and behavior of dissociated crescentin structures in the absence of crescentin mutations or drug treatments. Consistent with earlier reports (7, 10), we observed primarily left-handed (91%; n = 204) helical structures by both immunofluorescence and mGFP tagging in elongated cells made by depleting the cell division protein FtsZ (Fig. (Fig.55 A to C). Addition of new crescentin subunits along the entire length of the crescentin structure was independent of envelope association (data not shown). Similarly, biphasic growth, where longitudinal extension is replaced by primarily lateral thickening once crescentin structures reach the cell poles (10), was normal and thus independent of membrane association in the wbqL mutant (data not shown).
During normal cell division in C. crescentus, the crescentin structure is cleaved or locally depolymerized at the division site so that each daughter cell inherits part of the crescentin structure. This process works well even when crescentin (tagged with a tetracysteine motif [TC]) is strongly overproduced from a medium-copy-number plasmid, pJS14 (Fig. (Fig.66 A). We therefore tested whether the cell division-dependent disruption of the crescentin structure was related to envelope association. We similarly produced crescentin-TC from pJS14 in a wbqL disruption background (CJW3295). In these wbqL mutant cells, the overproduced crescentin-TC formed a brightly fluorescent, nearly straight filamentous structure within cells (Fig. (Fig.6B).6B). The straight rather than helical configuration of the crescentin-TC structure likely results from crescentin overproduction, since dissociated crescentin structures in FtsZ-depleted cells (CJW3332) also tend to lose their helicity during the course of overproduction (Fig. (Fig.5D).5D). This loss of crescentin structure helicity may be due to its becoming stiffer as a result of continued lateral addition of crescentin subunits.
Overproduction of crescentin in wbqL mutant cells also caused a striking cell chaining phenotype, with many cells in the population appearing to be linked together by a single crescentin structure (Fig. (Fig.6B).6B). Additionally, thin cellular extensions, visible by phase-contrast microscopy, appeared to link some chained cells together (Fig. (Fig.6B,6B, arrowheads). These extensions often contained a crescentin-TC structure (Fig. (Fig.6B,6B, inset), suggesting that the presence of a dissociated, thicker crescentin structure caused a block in cell division. However, it was also possible that this inability of cells to divide efficiently in the presence of a thick crescentin structure was specific to the inactivation of wbqL. To rule this out, we used an N-terminally truncated mutant of crescentin, crescentinΔN27, which is able to form a filamentous structure but is dissociated from the cell envelope (7). We overexpressed creS(ΔN27)-tc using the Pxyl promoter on the low-copy-number plasmid pMR20 in a crescentin-null background (CJW1537), so that it was the only source of crescentin in the cell. Upon induction of crescentinΔN27-TC production, a strong cell chaining phenotype developed. Just as in the wbqL mutant background, crescentinΔN27-TC filamentous structures were straight (consistent with overproduction) and spanned the length of multiple chained cells (Fig. (Fig.6C).6C). The cell chains displayed clear constrictions that contained crescentinΔN27-TC structures (Fig. (Fig.6C,6C, arrows), giving the impression that the cells had attempted to divide but could not separate the crescentin structure.
Since crescentin has not yet been conclusively identified in electron cryotomography (ECT) experiments (6), we reasoned that the thinness of these crescentinΔN27-containing constrictions, combined with protein overproduction, might allow good visualization of the crescentinΔN27 structure. Indeed, ECT revealed thick and straight filament bundles within the cytoplasm of cells that were sometimes bent at stalled division sites (Fig. (Fig.6D).6D). In the image shown, the division site appears to have constricted until it reached the approximate diameter of the crescentin bundle but no further. Fully developed flagellar motors (FM, Fig. Fig.6D)6D) and chemoreceptor arrays (CR, Fig. Fig.6D)6D) are present at each side of the constriction, as might be expected if daughter cells failed to separate but then advanced through their respective cell cycles. Collectively, these data suggest that the thicker crescentin structures formed by crescentin overproduction interfere with the completion of cytokinesis when they are dissociated from the cell envelope.
How might envelope association assist in successful division of the crescentin structure during cell division? One possibility is that a protein or factor that promotes degradation or disassembly of crescentin filaments at the division site is membrane associated and therefore cannot act efficiently on the crescentin structure when this structure is dissociated from the membrane, as in a wbqL mutant. Interestingly, E. coli displays a cell chaining defect when crescentin is exogenously expressed even though the crescentin structure is envelope associated (7). It is therefore possible that the evolutionarily distant E. coli lacks an important factor for splitting the crescentin structure during division. C. crescentus also exhibits a considerable period of midcell elongation preceding division that is not shared by E. coli (1). It is tempting to hypothesize that midcell elongation in C. crescentus would tend to stretch the envelope-associated crescentin structure, making its disruption at the division site more favorable.
We thank J. Smit for gifts of strains, anti-O-polysaccharide, and anti-RsaA antibodies, O. Sliusarenko and T. Emonet for assistance with cell curvature analysis, H. Lam for plasmids, J. Wertz for technical help, and the members of the Jacobs-Wagner laboratory for critical reading of the manuscript.
This work was supported by the National Science Foundation GRFP (to M.T.C.), the Mustard Seed Foundation (to M.T.C.), a Howard Hughes Medical Institute predoctoral fellowship (to M.A.M.), NIH grants AI067548 (to G.J.J.) and GM076698 (to C.J.-W.), gifts to Caltech from the Gordon and Betty Moore Foundation and Agouron Institute, the Pew Scholars Program in the Biological Sciences, sponsored by the Pew Charitable Trust (to C.J.-W.), and the HEALTH-F3-2009-223431 DIVINOCELL collaborative project grant from the Commission of the European Communities (to W.V.). C.J.-W. and G.J.J. are Howard Hughes Medical Institute investigators.
Published ahead of print on 30 April 2010.
§Supplemental material for this article may be found at http://jb.asm.org/.