The mechanism by which the membrane synthetic machinery might be co-organized with the cell-division architecture during the bacterial cell cycle remains to be investigated. We characterized a key enzyme of phospholipid and fatty acid synthesis in Bacillus subtilis, the acyl–acyl carrier protein phosphate acyltransferase (PlsX), and identified it as a component of the cell-division machinery. Comprehensive interaction analysis revealed that PlsX interacts with FtsA, the FtsZ-anchoring protein. PlsX mainly localized at the potential division site independent of FtsA and FtsZ and then colocalized with FtsA. By multidirectional approaches, we revealed that the Z-ring stabilizes the association of PlsX at the septum and pole. The localization of PlsX is also affected by the progression of DNA replication. PlsX is needed for cell division and its inactivation leads to aberrant Z-ring formation. We propose that PlsX localization is prior to Z-ring formation in the hierarchy of septum formation events and that PlsX is important for co-ordinating membrane synthesis with cell division in order to properly complete septum formation.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Two key cyclic events must be accomplished during the cell cycle of rod-shaped bacteria: cell elongation and cell division. During cell elongation, the cylindrical cell envelope is synthesized simultaneously with DNA replication and segregation prior to cell division. Elongation and division of bacterial cells must be tightly co-ordinated with cell envelope synthesis in order to maintain structural integrity and prevent lysis. However, the detailed mechanisms by which these growth processes are spatiotemporally co-ordinated remains incompletely understood.
Cell division is achieved by a macromolecular machine called the divisome. Divisome assembly is initiated by polymerization of the tubulin homologue FtsZ into the protofilament, which assembles to make a architecture called the Z-ring that lies close to the cytoplasmic membrane at the preseptal site (Errington et al., 2003). Polymerization of FtsZ into the proper ring structure and association of the ring with the inner face of the cytoplasmic membrane requires the aid of the FtsZ-binding protein FtsA (Jensen et al., 2005) in Bacillus subtilis. FtsA is dispensable, but ftsA mutants are highly filamentous and the frequency of functional Z-ring formation is severely reduced (Jensen et al., 2005). It is generally believed that FtsA localization to the division site occurs immediately after or concomitantly with Z-ring formation and that FtsA is responsible for recruiting the later-assembling membrane-bound division proteins such as DivIB, which mediates peptidoglycan synthesis at the septum (Errington et al., 2003; Daniel et al., 2006). In Escherichia coli and B. subtilis, lateral positioning of the Z-ring is mainly determined by two negative regulatory systems: the nucleoid occlusion and Min-systems (Bramkamp and van Baarle, 2009). Nucleoid occlusion counteracts assembly of Z-rings in the close vicinity of chromosomes and is mediated by the DNA-binding protein Noc in B. subtilis. Min-system proteins include MinC, MinD, MinJ and DivIVA, which localize at the cell pole to prevent inappropriate polar division in B. subtilis (Bramkamp and van Baarle, 2009).
The fluid mosaic model of membrane structure has been revised in recent years. The biological membrane is not a homogenous structure, but consists of domains enriched with specific lipids and proteins that are often referred to as lipid rafts (Edidin, 2003). Convincing evidence has indicated the presence of lipid domains in the microbial cytoplasmic membrane (Matsumoto et al., 2006). Specific probes revealed that the phospholipid cardiolipin (CL) and phosphatidylethanolamine (PE) locate preferentially at the cell poles and division sites (mid-cell regions) in B. subtilis owing to their chemical characteristics (Kawai et al., 2004; Nishibori et al., 2005). Moreover, all phospholipid synthases in the form of green fluorescent protein (GFP)-fused proteins, including enzymes for the major polar lipids PE and CL, locate at B. subtilis division sites and cell poles (Nishibori et al., 2005). They suggested that phospholipids were produced mostly in the septal membranes and were kept from diffusing out to lateral ones. The invagination of preseptal site creates a saddle structure and, as the radius of the FtsZ ring decreases, the requirements for phospholipids may change. Cells therefore might be needed to ensure the supply of appropriate lipids at the division site (Norris et al., 2002). The septally localized phospholipid synthases could meet this need by serving lipids with the appropriate nature at proper times and locations during the division process. If this hypothesis is correct, it raises the question how localization is achieved concurrent with the assembly of cell division proteins in order to synthesize the phospholipid membrane in concert with synthesis of the cell wall at the leading edge of the invaginating cell envelope.
In this report, we show evidence to support the idea of organization among proteins involved in lipid synthesis and cell division. Yeast two-hybrid (Y2H) matrix analysis revealed that PlsX, an essential acyl–acyl carrier protein (ACP):phosphate acyltransferase which catalyses the production of the acyl phosphate (acyl-PO4) required as an acyl donor for the first step of phospholipid synthesis (Lu et al., 2006; Paoletti et al., 2007), interacts with FtsA. PlsX localizes at the potential division site independent of FtsA and FtsZ. We also observed that PlsX localization at the potential division site depends on DNA replication initiation. Thus, PlsX localization is co-ordinated with both cell division and DNA replication. The plsX mutant showed aberrant septum formation co-occurring with delocalization of FtsA and FtsZ. These results suggest that PlsX functions as a member of the cell-division machinery and mediates the co-ordination of lipid synthesis and cell wall synthesis.
PlsX associates with components of the cell division machinery
To explore possible interactions between cell division proteins and phospholipid synthases, we performed Y2H matrix analysis of the related proteins. We selected 12 enzymes involved in phospholipid synthesis (GpsA, PlsX, PlsY, PlsC, DgkA, CdsA, PssA, Psd, UgtP, PgsA, YfiX and YwnE) and 27 proteins related to cell division (FtsZ, FtsA, DivIB, FtsL, DivIC, FtsW, PbpB, DacA, DivIVA, EzrA, FtsE, FtsX, MinC, MinD, SecA, Soj, SpoIIE, SpsK, PbpD, PbpF, PbpX, PonA, YfhF, YfhK, YjoB, YppD and ZapA), constructed 53 DNA fragments that correspond to full-length and/or truncated forms of these proteins (without the trans-membrane region), and performed one-to-one analysis (53 × 53 = 2809 crosses). We observed several direct interactions between the two groups (Fig. S1). Our analysis revealed that PlsX, an acyl-ACP:phosphate acyltransferase that is essential for cell growth (Paoletti et al., 2007), interacted with the pivotal cell division proteins, FtsA, DivIB, DivIC and FtsL (Fig. S1). In B. subtilis, PlsX is required for phospholipid and fatty acid synthesis, suggesting PlsX is the key enzyme to co-ordinate membrane phospholipids synthesis (Paoletti et al., 2007; Yoshimura et al., 2007; Hara et al., 2008). We also detected the following known interactions among proteins related to cell division (indicated by white squares in Fig. S1): self-interaction of FtsA (Feucht et al., 2001), MinD-MinC and self -interaction of MinC but not MinD (Bramkamp et al., 2008) and FtsL-DivIC (Daniel et al., 2006), assuring that our Y2H analysis was effective.
Next, to confirm the interaction between PlsX and cell division proteins, we used in vivo cross-linking, followed by affinity co-purification of oligohistidine-tagged PlsX and liquid chromatography-tandem mass spectrometry analysis (LC-MS/MS) to identify proteins associated with PlsX (Fig. S2). We detected FtsA in PlsX-including complex, but not DivIB, DivIC and FtsL. Interactions confirmed by Y2H analysis are not always detected by pull-down assays, due to the different sensitivities of the analytical methods and the strength of the reaction. In addition, several other proteins, such as cell division proteins (EzrA, DivIVA), metabolic enzymes (PgcA, PyrG, GlmS, GlpD, PckA, GuaA, GuaB, MetK, MurAA, MurD, MetA GapA, SucC, CitZ, GapB, Ndh, YumC, MurG), and cytoskeletal proteins (MreB, Mbl, RodZ), were identified as candidates associating with PlsX. In contrast, Tkt, TufA, RpsB and Spo0M, which have been detected as non-specific backgrounds in FtsA and DnaC complexes purified by similar methods (Ishikawa et al., 2006), were also detected. Thus, we focused on the interaction between PlsX and FtsA to investigate how the membrane synthetic machinery is organized with the divisome.
PlsX localizes at the potential division site
The interaction between PlsX and FtsA raised the possibility that PlsX exhibits a localization pattern similar to that of cell division proteins. In order to investigate the localization of PlsX, we fused gfp to the 5'-terminus of plsX under the control of a xylose-inducible promoter (Pxyl) and integrated this construct into the amyE locus of the plsX+B. subtilis chromosome (NBS800). Expression levels of the GFP–PlsX were compared on Western blots probed with either antibodies to PlsX or GFP. While GFP–PlsX accumulates to a level 40% lower than endogenous PlsX level in the presence of xylose, endogenous PlsX level was not affected (Fig. 1A). Thus, the cellular function of PlsX supposed not be impaired. To test the function of GFP–PlsX, a plsX deletion was generated by replacing the entire plsX gene with a spectinomycin (spc)-resistance gene cassette introduced into a recipient strain containing an ectopic copy of Pxyl–gfp–plsX. Transformants (NBS1362) were readily obtained in xylose-containing media and retained the inducible plsX constructs at the amyE locus. With the Pxyl–gfp–plsX construct, the transformants grew normally with 0.5% xylose supplementation, but were capable of forming only small colonies in the absence of xylose (Fig. 1B). In liquid medium containing 0.5% xylose, growth rate of NBS1362 and NBS800 was equivalent (Fig. 1C). Therefore, Pxyl–gfp–plsX fully complemented the plsX disruption phenotype. Fluorescence microscopy experiments revealed majorly three distinct pattern of GFP–PlsX localization in populations of cells: (i) septum (indicated by arrows 2 and 4), (ii) cell pole (indicated by arrows 1, 3 and 6) that were already detectable with the FM4-64 membrane stain, and (iii) potential division site (indicated by arrow 5) which generally corresponds to the inter-nucleoid spaces detected by the DAPI nucleoid stain (Fig. 1D). We have noticed in some of our pictures that the GFP–PlsX exhibited some peripheral membrane staining (indicated by asterisk in Fig. 1D). As is clear in the line-scan of the cell perimeter (Fig. 1D bottom), the signal intensity of GFP–PlsX rose at the septum (the regions that there were dips in the signal intensity of DAPI, blue line, and rises in the signal intensity of FM4-64, red line) and the potential division site (the region that there was a dip in the signal intensity of DAPI but not an apparent rise in the signal intensity of FM4-64). The result of the same experiment using the plsX− strain NBS1362 showed localizations of GFP–PlsX similar to those in the plsX+ strain NBS800 (Fig. 1D). Recently, weak dimerization behaviour of GFP was shown to stimulate clustering of GFP-fusion proteins, leading to artifact localization (Landgraf et al., 2012). To check this possibility, we introduced an A206K mutation which diminishes dimerization of GFP (NBS1876). In this strain, PlsX localization was similar to those of NBS800, suggesting that the localizations of GFP–PlsX are not an artifact of GFP dimerization (Fig. S3). However, PlsX localization was different from that observed in an immunofluorescence microscopy (IFM) experiment, showing that PlsX exhibited a punctate pattern and peripheral localization (Paoletti et al., 2007). The reason for this difference is not clear; however, one possibility is that the fixation and permeabilization procedures required for IFM could damage the structure of the cell, as observed in the case of MinD (Marston et al., 1998). We observed that the cellular amount of PlsX was relatively high at the exponential phase (Fig. S4A and B), during which we also observed a punctate pattern of staining and peripheral localization of PlsX in an IFM study of haemagglutinin (HA)-tag-fused PlsX (Fig. S4C). In contrast, after both our and the previously reported (Paoletti et al., 2007) fixation procedures, GFP–PlsX appeared to be distributed throughout the cell (Fig. S4D). Thus, we concluded that the localization patterns of GFP–PlsX reflected the more native localization patterns of PlsX.
PlsX colocalizes with Z-ring
The data in Fig. 1D indicates that PlsX associates with the divisome. To confirm this and to more precisely assess the timing of PlsX association with the divisome, we examined the colocalization of PlsX and FtsA using the strain expressing GFP–PlsX and cyan fluorescent protein (CFP)–FtsA fusion proteins (NBS1779). CFP was fused to ftsA at its endogenous locus, so that the fusion protein was expressed from the endogenous ftsA promoter as the sole form of FtsA in the cell. The constructs appeared to be functional as judged by normal septum formation on LB medium at 37°C (as indicated by arrows 2 and 4 in Fig. 1E). Analysis of this strain showed that PlsX and FtsA colocalized at the septum and cell pole (indicated by arrows 1–4 in Fig. 1E). However, some PlsX localized at the mid-cell without FtsA (indicated by asterisks in Fig. 1E). Thus, we postulated that PlsX localized at the potential division site prior to Z-ring formation and remained at the septum during constriction. Moreover, PlsX still associated with the cell pole, whereas Z-ring had already disassembled (indicated by arrow 5 in Fig. 1E). Although we could hardly distinguish the relocalized GFP–PlsX from the remaining GFP–PlsX at the cell pole, this observation indicated the regulation of PlsX localization is different from that of FtsA localization.
Preseptal localization of PlsX is independent of FtsA and FtsZ
We next asked if GFP–PlsX localization at the potential division site depends on FtsA. We calculated the ratio of the potential division site colocalized with GFP–PlsX to total potential division sites in the wild-type (NBS800) and the ftsA mutant (strain NBS1371) and found that GFP–PlsX mostly localized between nucleoids at the potential division site (wild-type: 84.7%; the number of sites counted are 188/222, ftsA: 94.0%; 332/353, indicated by arrows 1–4 in Fig. 2A). We also examined the localization of GFP–PlsX in FtsZ-depleted cells (strain NBS1341) to clarify whether its localization was dependent on FtsZ. 1.5 hours after depletion of FtsZ, Quantitative Western analyses revealed the greatly diminished cellular concentration of FtsZ (Fig. 2C). At this time point, GFP–PlsX fluorescence was mainly observed at the potential division site (91.7%; 288/314, indicated by arrows 1–5 in Fig. 2D). To confirm its independency on Z-ring, we observed the PlsX localization after the induction of MciZ, which is an inhibitor of FtsZ polymerization (Handler et al., 2008). In both before and after the induction of MciZ, GFP–PlsX localized at the potential division site (89.2%; 312/350, indicated by arrows 1–5 in Fig. S5A). Thus, we suggested PlsX localizes at the potential division site independently of FtsA and FtsZ. But bright transverse bands of GFP–PlsX (indicated by arrow 4 in the picture of NBS800 in Fig. 1D) were rarely formed in the elongating filaments; instead, the fusion protein mainly showed dot-like localization at potential division sites. Three-dimensional reconstruction images confirmed that some bands correspond to rings in the wild-type strain (Movie S1). On the other hand, in the cells with defective Z-ring formation (ftsA mutant, ftsZ-depleted and mciZ-overexpressed cells), the foci were discrete structures, located close to the cell periphery and probably not a part of the higher-order (e.g. ring) structure (Movie S2-4). Transverse bands of GFP–PlsX were mostly observed at the septum. Thus, Z-ring formation or septum formation might affect the transition of PlsX from dot-like localization to band-shaped localization.
Min system indirectly act on PlsX localization
We wanted to understand the physiological reason for interaction between PlsX and FtsA; therefore, we investigated the effectors of PlsX localization, focussing on MinCD, which prevents polar Z-ring assembly. Although most of the fusion protein localized at potential division sites (Fig. 3A), GFP–PlsX was missing at many potential division sites in the minCD mutant (indicated by arrows 1, 3 and 6 in Fig. 3A). The ratio was significantly lower in the minCD mutant than in the wild-type strain (minCD: 60.4%; 119/197, wild-type: 84.7%; 188/222). In the absence of the Min system, the Z-ring components, FtsA, FtsL and PBP2B, do not disassemble and remains in close proximity to recently used division sites (van Baarle and Bramkamp, 2010). Thus, we hypothesized that PlsX also remained at the septum and cell pole together with FtsA in the minCD mutant. To test this idea, we investigated simultaneous localization of GFP–PlsX and CFP–FtsA in the minCD mutant (Fig. 3C). Consistent with this idea, GFP–PlsX colocalized with CFP–FtsA at the septum and pole (indicated by arrows 1 and 3 in Fig. 3C). It was reported that the majority of the minCD mutant cells formed only single FtsZ rings and the cell division defect was suppressed in minimal medium (Levin et al., 1998). To compare the localization pattern of GFP–PlsX in both the rich (LB) medium and the minimal (S7 minimal salt) medium, we fused gfp to the 52- terminus of plsX under the control of an IPTG-inducible promoter (Phy-spank) and integrated this construct into the thrC locus (NBS1877). In the wild-type strain, GFP–PlsX was mostly localized at the potential division site in both mediums (Fig. S6A and B, LB: 86.3%; 158/183, S7: 93.3%; 56/60). The cell division defect in the minCD mutant was mostly suppressed in the minimal medium (Fig. S6C). As with Z-ring formation, in the minCD mutant, the majority of GFP–PlsX was localized at the potential division site in S7 minimal salt medium (Fig. S6A and B, LB: 63.7%; 114/179, S7: 88.7%; 55/62). Since MinC and MinD did not interact with PlsX in our Y2H analysis, these proteins might indirectly act on PlsX localization (Fig. S1).
FtsA stabilizes the association of PlsX at the septum
In the minCD mutant, the ratio was recovered by the induction of MciZ, the inhibitor of FtsZ (T0: 69.8%; 282/404, T1.5: 89.2%; 330/370). GFP–PlsX was localized at the preseptal site (indicated by arrows 1–6 in Fig. S7A). Thus, the abnormal pattern of PlsX localization in the minCD mutant was caused by the stabilization of Z-ring.
To confirm the relationship between PlsX and FtsA, we investigated the effect of ftsA overexpression on PlsX localization. We placed ftsA under the control of an IPTG-inducible promoter (Phy-spank) at the thrC locus (NBS1569), and examined the cells pre- and post-induction. In E. coli, it was reported that overexpression of ftsA results in cell division defects (Wang and Gayda, 1990). We observed that overexpression of ftsA led to an abnormal septum (irregular, twisted septum), minicell formation (Fig. S8B) and a drastic decrease of cell viability (Fig. S8A, a plating efficiency of 3.24 ± 1.47 × 10−3 % was observed on IPTG-containing plates). Induction of FtsA was confirmed by Western blotting (Fig. S8C). Importantly, similar division septa have been reported in previous studies, and are thought to form through the constriction of helical or non-ring FtsZ assemblies, which direct invagination of the cell envelope in an abnormal pattern (Feucht and Errington, 2005). In the presence of IPTG, GFP–FtsA excessively remained at cell poles and septum and exhibited an arc–like pattern of localization (indicated by arrows 1–6 in Fig. 4A). Therefore, the induction of FtsA, the membrane anchor for FtsZ, led to excess-stabilization of Z-ring at septum. Aberrant septum formation might be due to a failure to disassemble the Z-ring that results in another round of cytokinesis close to the original cell division site. Under the same condition, the localization of GFP–PlsX resembled that of GFP–FtsA (indicated by arrows 1–6 in Fig. 4C). We observed the colocalization of GFP–PlsX and CFP–FtsA at the septum when ftsA was overexpressed (Fig. 4E). We again checked the relationship between PlsX and FtsA by another approach. 3-Methoxybenzamide (3-MBA) and its derivatives were previously shown to enhance both the assembly and stability of FtsZ polymers, causing filamentation with punctate colocalized foci of cell division proteins in B. subtilis (Ohashi et al., 1999; Adams et al., 2011). Following growth with 10 mM 3-MBA, GFP–PlsX also exhibited an irregular spotty pattern and colocalized with CFP–FtsA (Fig. S9). From these results, we concluded that FtsA may not be required to recruit PlsX, but stabilizes the association of PlsX with the Z-ring or septum.
PlsX localization is co-ordinated with DNA replication
To elucidate the mechanism by which PlsX localized at the potential division site independently of FtsZ and FtsA, we investigated the relationship between PlsX localization and DNA replication. Determination of the position of cell division is tightly linked to DNA replication status. Inhibiting replication initiation leads to the formation of elongated cells with acentric Z-rings adjacent to centrally positioned nucleoids, suggesting the key role of early stages of replication in promoting Z-ring assembly precisely at the mid-cell (Harry et al., 1999). We observed PlsX localization during temporal inhibition of replication initiation by using the sirA-inducible strain (NBS1585). SirA binds to DnaA and inhibits the binding of DnaA to the origin of replication (Wagner et al., 2009; Rahn-Lee et al., 2011). To exclude cells with on-going replication elongation, cells were cultured for 1.5 h after induction of SirA and GFP–PlsX localization was observed. Cells exhibited filamentous morphology and dramatically changed their DNA distribution. Nucleoids were separated by large spaces. Surprisingly, GFP–PlsX mainly localized at the lateral membrane and cell pole, but rarely on the top of the nucleoid (indicated by arrows 1–4 in Fig. 5A). We also observed GFP–FtsA localization when sirA was overexpressed (Fig. S10A). As shown in the line-scan of the cell perimeter, while GFP–FtsA tightly localized near the edges of the DNA masses (Fig. S10A), GFP–PlsX distributed farther from the nucleoid (Fig. 5A). These results indicate that one of the systems that senses replication/chromatin status and distribution is also the key in promoting PlsX localization precisely at the mid-cell.
plsX mutants exhibited cell division defect
To investigate the physiological relevance of the localization of GFP–PlsX, we attempted to screen a plsX temperature-sensitive mutant by random PCR mutagenesis. From more than 10 000 transformants we isolated a single mutant plsX103, NBS1010, which showed a severe growth defect on plates at 39°C (Fig. 6A). In liquid medium, strain NBS1010 grew normally at the permissive temperature (30°C), but shortly after the shift to the restrictive temperature (45°C), it showed a severe growth defect (Fig. 6B). Sequence analysis revealed 2 nucleotide substitutions converting Asp59 to Gly and Leu104 to Ser in the N-terminal region of PlsX, the site of enzymatic activity. To determine whether one or both of the substitutions contributes to the temperature sensitivity of NBS1010, we constructed single mutation strains NBS1328 (plsXR59G) and NBS1329 (plsXL104S). Both strains grew normally at 39°C but showed a growth defect on plates at temperatures greater than 39°C (Fig. S11). This result indicates that the growth defect of strain plsX103 is caused by the synergistic effect of both mutations.
To examine the effect of plsX mutation, we observed cell morphology over time by fluorescence microscopy. Just before the shift to 45°C (T0), cell length (as the distance between adjacent septa) of the plsX103 strain appeared slightly longer than of the wild-type strain (Fig. 6C). FM4-64-stained preparations showed that this elongation was mainly due to the disturbed cell division with long aseptate regions, which occurred irregularly. To quantify these phenotypes, we measured the length of more than 150 individual cells. As shown in Fig. 6D, plsX103 cells were longer than the wild-type cells with a broad length distribution, and a significant population of cells (29%) formed filaments with lengths of 7 μm or longer. Next, to determine whether this phenotype was due to defective nucleoid segregation, we measured the number of nucleoids per cell; the number was higher in plsX103 than in the wild-type (Fig. 6E). Thus, nucleoid segregation proceeded normally but septum formation was moderately defective in plsX103 cells at T0.
One hour after the shift to 45°C (T1), we observed a more defective phenotype in plsX103 cells. Cell length measurements indicated that plsX103 cells were longer than the wild-type cells (Fig. 6D). We also observed an increase in nucleoid bisection in plsX103 cells at the non-permissive temperature (indicated by the arrowhead in Fig. 6C). Bisection of unsegregated chromosomes by the division septum is usually an extremely rare event; it occurred at a frequency < 0.42% (0/233) in our strain background, consistent with previous reports (Ireton et al., 1994; Lee and Grossman, 2006). In contrast, chromosome bisection occurred > 17.4-fold more frequently at the non-permissive temperature (7.32%: 18/246). Our result is indicative of the defective coupling of cell division to chromosome segregation in plsX103 cells at the non-permissive temperature. Intracellular PlsX decreased after the shift to 45°C in the plsX103 mutant (Fig. 6F), but some PlsX remained, suggesting these phenotypes are caused by qualitative and quantitative changes of PlsX.
To examine the effect of PlsX dysfunction on septum formation by another approach, we observed cell morphology upon PlsX depletion at 37°C. The IPTG-inducible plsX strain showed the immediate arrest of growth in the absence of IPTG (Fig. S12A). Quantitative Western-blot analysis revealed that the population of PlsX reduced in the absence of IPTG (Fig. S12B). The PlsX-depleted cells also exhibited the aberrant septum formation that was not apparent in the presence of IPTG (Fig. S12C). Cell length was slightly longer and the number of nucleoids per cell was greater in the absence of IPTG (Fig. S12D and E). Some cells (14.4%) formed filaments with lengths of 8 μm or longer. Therefore, the aberrant septum formation in plsX103 cells was not an indirect effect of heat stress. Nucleoid bisection was not observed in the PlsX-depleted cells (Fig. S12C). We do not understand why no nucleoid bisection was observed under PlsX depletion. One possibility is that there are qualitative and quantitative differences in the protein level of PlsX. Some amount of PlsX was detected in the plsX103 mutant (Fig. 6F), but not in the PlsX-depleted cells (Fig. S12B).
PlsX affects Z-ring formation
To study the relationship between cell division and PlsX, we constructed merodiploid strains in the wild-type or plsX103 backgrounds harbouring intact ftsAZ and the ftsA–gfp fusion. Plasmid pFTSA8G, encoding the ftsA ORF with the promoter region of ftsA, was integrated at the chromosomal ftsA locus by single cross-over, so that the FtsA–GFP and FtsA were expressed from the endogenous ftsA promoter. In the wild-type cells, bands of FtsA–GFP corresponding to Z-rings formed at regular positions between segregating nucleoids before and after the shift to 45°C (Fig. 7A). In contrast, its localization was disrupted in plsX103 cells after the shift to 45°C, but seemed normal before the shift (Fig. 7A). In plsX103 cells after the temperature shift, a few clear bands of FtsA–GFP correctly positioned between well-separated nucleoids were observed (indicated by arrowheads in Fig. 7A). However, most bands were missing or only faint, and there appeared to be more dispersed fluorescence than in the normal state. To gain more insight, we focused on the localization of GFP–FtsZ under these conditions. We constructed wild-type and plsX103 strains carrying gfp–ftsZ under control of the natural ftsAZ promoter at the amyE locus. The bands of GFP–FtsZ in plsX103 cells disappeared from the potential division site in the same way as FtsA–GFP (Fig. 7B). Similar results were obtained in the PlsX-depleted strain (Fig. S13A and B). We postulated that collapse of anchoring of FtsZ to the membrane by FtsA caused cell division defects in plsX103 and the PlsX-depleted strains. Immunoblot analysis revealed that GFP–FtsZ and FtsA–GFP remained mostly intact in the presence of PlsX, while their amounts decreased, particularly in the plsX103 strain at the non-permissive temperature (Fig. S14A and B).
For several decades, it remained unclear how phospholipid synthases act spatiotemporally during the cell cycle. Our study provides insights into their involvement in the cell division control.
First, we identified PlsX, acyl-ACP:phosphate acyltransferase, as a component of the cell division machinery in B. subtilis (Figs S1 and S2). PlsX interacts with FtsA, the FtsZ-anchoring protein, and colocalizes with FtsA at the site of cell division. This observation is consistent with the previous report that phospholipids are synthesized mainly in the septal membrane (Nishibori et al., 2005). The formation of acyl-PO4 by PlsX is a critical step in the initiation of phospholipid biosynthesis, coupling fatty acid and phospholipid synthesis (Paoletti et al., 2007; Yoshimura et al., 2007; Hara et al., 2008). Thus, we concluded that PlsX is the key example to investigate the relationship between cell division proteins and phospholipid synthases.
Second, we postulated that PlsX is cytologically upstream of FtsA and FtsZ. PlsX is localized at the potential division site independently of FtsA and FtsZ (Fig. 2 and Fig. S5). This is a surprising observation, as it was previously thought that self-assembly of the Z-ring determines the location of the division site and serves as a framework for assembly of the division apparatus. Some proteins including phospholipid synthases are localized at the potential division site independently of FtsZ in E. coli, indicating the existence of another framework for assembly of phospholipid synthases (Janakiraman et al., 2009). On the other hand, in the minCD mutant PlsX is missing at the potential division site (Fig. 3A), but instead PlsX and FtsA co-associate with the septum and pole (Fig. 3C). The same phenomenon was observed under the condition that Z-ring is stabilized by ftsA overexpression or addition of 3-MBA (Fig. 4E and Fig. S9). At first, this seems to contradict its independence from FtsZ. However, phospholipid synthases should remain at division sites to synthesize membrane phospholipid until septum formation is complete. Thus, these results indicate the existence of co-ordination between membrane synthesis and cell division machinery. The current data do not lead specifically to any particular model at the molecular level. To clarify the timing of PlsX localization against Z-ring formation, a time-lapse experiment will be effective. Thus, the timing of PlsX localization at the potential division site is not conclusive while its localization is indeed independent of FtsA and FtsZ. Nevertheless, we postulate that FtsA interacts with PlsX after Z-ring formation and prevents PlsX from moving to the next division site until the completion of septum formation in order to couple septum formation (Z-ring formation) with PlsX localization.
Third, we found that PlsX localization at the next division sites depends on DNA replication. Although PlsX can be localized near the edges of the DNA mass in the absence of DNA replication initiation, the precise localization of PlsX at the next division sites is dependent on progress through the early stages of DNA replication, including completion of replication initiation. Independence of PlsX localization at the potential division site from Z-ring formation suggests that one or more unidentified proteins upstream of FtsZ in the assembly hierarchy of the divisome might act on the coupling of PlsX localization and DNA replication status. Recent studies suggest several mechanisms in addition to the Noc-dependent cell division inhibition, and the Min system regulates Z-ring localization in rod-shaped bacteria (Bernard et al., 2010; Moriya et al., 2010). This may be why many bacteria do not have Min or Noc, and other mechanism(s) must exist in these organisms to determine the division site (Margolin, 2001). Therefore, it is possible that this hypothetical mechanism also regulates PlsX localization. Moreover, like cell division proteins (Strahl and Hamoen, 2010), the localization of GFP–PlsX was disrupted by the dissolution of the membrane potential (Fig. S15). Although the mechanistic details remain unclear, Z-ring formation and DNA replication play an important role in guiding PlsX to the potential division site and allowing it to act there.
Fourth, PlsX might take part in the precise Z-ring formation. Cells defective in phospholipid or fatty acid synthesis exhibit a small cell phenotype (Hunt et al., 2006). We again confirmed this phenotype using the plsY-inducible strain and the plsC-temperature sensitive strain (Figs S16 and S17). At the non-permissive temperature, however, the plsX103 mutant and PlsX-depleted cells exhibit the aberrant septum formation co-occurring with the defective Z-ring formation. Under these conditions, both FtsA and FtsZ are delocalized. FtsA can tether the Z-ring to the membrane, and indeed, FtsA has a membrane-targeting amphipathic helix that is essential for its function (Pichoff and Lutkenhaus, 2005). Thus, the division defect of the plsX mutant could be due to the delocalization of FtsA, although it is unclear whether the effect of PlsX on FtsA is direct or indirect. In support of this idea, we also observed an increase in nucleoid bisection in plsX103 cells at the non-permissive temperature, indicating a defect in the proper Z-ring formation (Fig. 6C).
Our findings might explain only a part of the physiological roles of phospholipid synthases in the bacterial cell cycle. The synthesis of cell wall is orchestrated by penicillin-binding proteins (PBPs). Rod-shaped bacteria have two distinct modes of cell wall synthesis, involved in cell elongation and cell division, which are believed to employ different sets of PBPs. These are positioned by 2 cytoskeletal systems that involve homologues of actin (MreB) and tubulin (FtsZ) (Carballido-Lopez and Formstone, 2007). On the other hand, most steps in the membrane metabolism consist of one set of enzymes. Therefore, regulating the localization of these enzymes might differ from the regulation of peptidoglycan synthase. Further research is necessary to resolve these questions.
Bacterial strains, plasmids, primers and genetic techniques
The bacterial strains and plasmids used in this study were listed in Table S1 and primers were summarized in Table S2. Drug concentrations for selection were: erythromycin, 1 μg ml−1; kanamycin, 50 μg ml−1; chloramphenicol, 5 μg ml−1; tetracycline, 15 μg ml−1; spectinomycin, 100 μg ml−1. B. subtilis or E. coli was grown in Luria–Bertani (LB) broth or S7 minimal salts medium supplemented with glucose (1%) and glutamate (0.1%) as described previously (Vasantha and Freese, 1980), except that MOPS buffer was used at 50 mM rather than at 100 mM. Required amino acids (tryptophan and threonine) were added at 50 μg ml−1.
Yeast two-hybrid analysis was carried out as described elsewhere (Fukushima et al., 2006). PCR-amplified fragments from B. subtilis genomic DNA were cloned into pGBTK or pGADT7. Oligonucleotide primers used in this specificity test were listed in Table S2. The pGBTK derivatives were transformed into yeast PJ69-4Aa (MATa, trp1-901, leu2-3, 112ura3-52, his-200, gal4Δ, gal80Δ, LYS2::GAL1-HIS3, GAL2-ADE2, met2::GAL7-lacZ), and the pGADT7 derivatives, into PJ69-4Aα (MATα, trp1-901, leu2-3, 112ura3-52, his-200, gal4Δ, gal80Δ, LYS2::GAL1-HIS3, GAL2-ADE2, met2::GAL7-lacZ), with a haploid strain, using TRP1 and LEU2 as selective markers. Transformants were mated in the appropriate liquid medium using flat-bottomed, 96 well plates. After mating, the cultures were collected, washed with sterilized water, and spotted onto synthetic complete (SC) agar plates that lacked leucine and tryptophan (SC-LW) for selection of LEU2 and TRP1 diploid cells. Selected cells were cultured in liquid SC-LW for 1 day and were then replica plated on the selection agar medium, SC-LWHA plates containing 1 mM 3-aminotriazol.
Purification of protein complexes
An overnight liquid culture of B. subtilis cells expressing histidine (His)-tagged PlsX on the LB medium plate containing 100 μg ml−1 spectinomycin at r.t was inoculated into 400 ml of the same medium at the initial OD600 of 0.01, and cultured at 37°C. When the cells reached an OD600 of 0.4, the culture was treated with formaldehyde (1% final concentration) for 30 min. Purification of the protein complex with the His-tagged protein was performed as described (Ishikawa et al., 2006). Separated proteins were visualized by colloidal Coomassie staining. Polyacrylamide gels were divided into slices, and the presence of proteins was confirmed by peptide-mass fingerprinting (Kasahara et al., 2012). Several proteins (e.g. TufA and ribosomal proteins) have already been identified as representative of a non-specific backgrounds (Ishikawa et al., 2006), and we also detected these proteins in the PlsX-complex (Fig. S2).
ftsZ depletion studies
Strain NBS1341 (trpC2 amyE::(Pxyl–gfp–plsX spc) Pspac–ftsZ erm; Pxyl and Pspac were xylose- and IPTG-inducible promoters respectively) was inoculated in LB medium supplemented with 0.5% (w/v) xylose and 1 mM IPTG at OD600 = 0.05 and grown at 37°C to OD600 = 1.0. The culture was diluted to 1:20 in LB medium and divided into 2 portions. Half of the culture was supplemented with 0.5% (w/v) xylose to induce gfp–plsX and 1 mM IPTG to induce ftsZ, whereas the other half was supplemented only with 0.5% (w/v) xylose.
For fluorescence microscopy, samples of living cells were examined on glass slides pre-treated with polylysine or on a thin film of 1% agarose. Membranes were stained with FM4-64 (Molecular Probes, Eugene, Oregon, USA). Nucleoids were stained with DAPI (Sigma, St Louis, Missouri, USA). Images were acquired on an Olympus BX-71 system with the DP71 digital camera and DP Controller software (Olympus, Tokyo, Japan). Confocal images were obtained using Zeiss LSM 710 and LSM780 laser scanning microscope systems (Carl Zeiss MicroImaging GmbH, Jena, Germany). For 3D reconstruction, 14–16 images (spacing between 0.214 and 0.188 μm) were taken through the focal plane. Fluorescence signals were quantified using MetaMorph v. 7.7 software (Universal Imaging Corporation, West Chester, Pennsylvania, USA).
Measurements of cell length and nucleoid number per cell
Cell length as the distance between adjacent septa was measured using MetaMorph v. 7.7 software. Nucleoid number was counted as the number of DAPI-stained nucleoid foci between adjacent septa.
Samples were prepared as described (Weart et al., 2007). Briefly, harvested cells were resuspended in 50 mM Tris pH 8.0, 1 mM EDTA, 2 mg ml−1 lysozyme, and 1 tablet 10 ml−1 of complete Mini EDTA-free (Protease Inhibitor Cocktail Tablets, Roche). Cells were incubated at 37°C for 20 min and lysed with sodium dodecyl sulphate (SDS). Cell lysates were normalized by OD600 at cell harvest and subjected to SDS-polyacrylamide gel electrophoresis. PlsX-HA was detected using the monoclonal anti-HA antibody (COVANCE, Princeton, New Jersey, USA). GFP was detected using the polyclonal rabbit anti-GFP antibody (MBL, Woburn, Nagoya, Japan). SigA was detected using the polyclonal anti-SigA antibody (gifted from F. Kawamura). FtsA and FtsZ were detected using polyclonal anti-FtsA and anti-FtsZ antibodies (gifted from S. Ishikawa). WalR was detected using the polyclonal anti-WalR antibody (gifted from R. Utsumi). Anti-PlsX antibodies were prepared by immunizing rabbits with the synthetic oligopeptides [C + VIEPTDEPVRAVRR; the first cysteine residue was added synthetically in order to conjugate the oligopeptide to Keyhole Limpet Hemocyanin (KLH) protein] corresponding to the amino acid positions 61–74 of PlsX.
This work was supported by Grant-in-Aid for Scientific Research on Priority Area (B) from the Ministry of Education, Culture, Sports, Science and Technology of Japan to H.Y., and Research Fund for the Advancement of the Graduate School, Tokyo University of Agriculture to H.T. We thank T. Araya-Kojima for critical reading of the manuscript, S. Ishikawa for his kind gift of the unpublished strain SD100 and anti-FtsA and anti-FtsZ antibodies, R. Utsumi for his kind gift of anti-WalR, F. Kawamura for his kind gift of anti-SigA antibody, and M. Fujita for his kind gift of plasmid pDR200.
Conflict of interest
The authors declare that they have no conflict of interest.