A mutation in the promoter region of zipA, a component of the divisome, suppresses the shape defect of RodZ-deficient cells

RodZ is important for maintaining the rod shape of Escherichia coli. Loss of RodZ causes conversion of the rod shape to a round shape and a growth rate slower than that of wild-type cells. Suppressor mutations that simultaneously restore both the growth rates and the rod shape were isolated. Most of the suppressor mutations are found in mreB, mrdA, or mrdB. One of the mutations was in the promoter region of zipA, which encodes a crucial component of the cell division machinery. In this study, we investigated the mechanism of the suppression by this mutation. ZipA was slightly but significantly increased in the suppressor cells and led to a delay in cell division. While round-shaped mreB and mrdA mutants lose cell bipolarity, we found that round-shaped rodZ mutants retained cell bipolarity. Therefore, we concluded that a delay in the completion of septation provides extra time to elongate the cell laterally so that the zipA suppressor mutant is able to recover its ovoid or rod shape. The suppression by zipA demonstrates that the regulation of timing of septation potentially contributes to the conversion of morphology in bacterial cells.


Introduction
Bacterial cells with a rod shape are composed of two geometrical parts: a central cylinder and hemispherical caps at both ends of the cylinder. Formation of the rod shape is controlled by two distinct modes of synthesis of peptidoglycan: elongation and septation. For elongation, peptidoglycan, which is newly synthesized by more than 10 enzymes (den Blaauwen et al. 2008;Sauvage et al. 2008), is inserted homogeneously into the central cylinder, while peptidoglycan of the cell poles is inert (de Pedro et al. 1997). In Escherichia coli, penicillin-binding protein (PBP) 2, encoded by mrdA, is a specific enzyme required for the synthesis of the cylindrical wall (Spratt 1975;Spratt and Pardee 1975). It has been proposed that the bacterial actin MreB along with MreC and MreD form a complex with PBP2 and its functional partner RodA, and that the complex plays a role in the synthesis of the cylindrical wall van den Ent et al. 2006). In contrast, septation relies on the bacterial tubulin FtsZ and associated proteins consisting of cell division machinery called the divisome. One of them is PBP3, encoded by ftsI, a specific enzyme required for the synthesis of septal peptidoglycan (Spratt 1975;Spratt and Pardee 1975;Begg et al. 1986). It has been thought that these two modes of synthesis of the cell wall are independent of each other. However, it has been shown recently that FtsZ contributes to cell elongation at the midcell and sidewall near the cell poles (Varma et al. 2007;Varma and Young 2009), and that ZipA, a component of the divisome, is required for cell elongation at the midcell which has been called PBP3-independent peptidoglycan synthesis (PIPS) (Nanninga 1991;Potluri et al. 2012). Furthermore, it has been shown that MreB, MreC, MreD, PBP2, and RodA localize to the midcell during the cell division cycle in E. coli (Vats and Rothfield 2007;Vats et al. 2009;van der Ploeg et al. 2013), suggesting that these components of cell elongation can play some role at the division site. Thus, the separation of the synthesis of peptidoglycan for elongation and division is not as clear as previously thought. In any case, as in eukaryotes, cytoskeletal proteins in prokaryotes are involved in both elongation and septation, and hence the determination of morphology is possibly through the regulation of localization and the activity of enzymes such as PBP2 and PBP3 that synthesize peptidoglycan (Dye et al. 2005;van den Ent et al. 2006;Vats et al. 2009).
RodZ is also an important determinant for cell shape (Shiomi et al. 2008;Alyahya et al. 2009;Bendezu et al. 2009). Escherichia coli cells lacking either functional MreB or RodZ become round, although the mreB mutant cell is larger than the rodZ mutant (Shiomi et al. 2008). The bacterial actin MreB is involved in the determination of cell bipolarity as well as in cell elongation (Nilsen et al. 2005;Shih et al. 2005;Pradel et al. 2007). Defects of mreB in E. coli cause the mutant cells to become round without "poles" instead of rod shaped with bipoles (Shih et al. 2005). Although RodZ colocalizes with MreB in vivo (Shiomi et al. 2008;Alyahya et al. 2009;Bendezu et al. 2009), and they interact with each other in vitro (van den Ent et al. 2010), some of their functions are different (Shiomi et al. 2008).
To investigate the function of RodZ, suppressors were isolated that grew faster and were found to also restore the rod shape (Shiomi et al. 2013). Whole genomic sequencing of these mutants identified suppressor mutations. Twenty-six of the 29 suppressor mutations were in mreB, mrdA, or mrdB, whose gene products are involved in peptidoglycan synthesis for cell elongation (Shiomi et al. 2013). These mutations change the properties of the MreB protein so that the suppressors can synthesize the cylindrical peptidoglycan without RodZ. Unexpectedly, another mutation was found in the upstream region of the zipA gene, encoding an essential membrane protein involved in cell division to stabilize the assembled Z ring (Hale and de Boer 1997;Pichoff and Lutkenhaus 2002). Here, we address the mechanism of suppression by zipA to restore the cell shape of the rodZ deletion mutant.

Recovery of cell shape by a suppressor mutation of zipA
We refer to the suppressor mutation in the upstream region of zipA as zipA p56 hereafter. After whole-genome sequencing of the suppressor, we confirmed the mutation by resequencing of the chromosomal region. Furthermore, we transferred the zipA p56 mutation to the wild-type and to the rodZ deletion mutant using transduction of P1 vir phage. The suppressor strain DS631 (ΔrodZ zipA p56 ) was isolated as one of the spontaneous mutants that grew faster than the rodZ deletion mutant in rich medium (Table  S1). This suppressor mutation simultaneously suppressed the defect in the cell shape as seen with other suppressor mutations ( Fig. 1A) (Shiomi et al. 2013). We analyzed the cell size of DS679 (ΔrodZ) and DS631 (ΔrodZ, zipA p56 ) (Fig. 1B). The cell width did not significantly change between DS679 and DS631: 1.83 AE 0.45 and 1.70 AE 0.32 lm, respectively, which was wider than that of the wild type (1.24 AE 0.09 lm). However, the cell length of DS631 (4.47 AE 1.70 lm) was increased compared with DS679 (3.17 AE 0.98 lm), and was similar in length to the wild type (4.45 AE 0.95 lm). The suppressor mutation zipA p56 restored not only cell growth but also the cell shape of the rodZ mutant.
Previously, we showed that most of the mreB, mrdA, and mrdB mutants isolated as suppressors of rodZ in rich medium could not recover the cell growth and shape of the rodZ mutant in minimal medium (Shiomi et al. 2013). Likewise, the suppressor strain carrying zipA p56 was not rod shaped but round in minimal medium, while the wild-type strain remained rod shaped in the same medium (Fig. S1), indicating that the effect of the suppressor mutations is dependent on the medium. We also have shown that suppressor strains carrying a suppressor mutation in mreB, mrdA, or mrdB grow at low temperatures and show swarming ability while the rodZ mutant doesn't (Shiomi et al. 2013). The zipA p56 mutation restored not only growth and cell shape at low temperatures but also the swarming ability of the rodZ mutant (Figs. S2A, S2B, and S3).

Increased expression of ZipA by the zipA p56 mutation
The mutation zipA p56 , which is a base substitution from T (thymine) to A (adenine), is located 56 bases upstream of the initiation codon of zipA. The position is not in the putative À10 and À35 promoter regions, which are located at 185 and 211 bases upstream of the zipA gene, respectively (Hale and de Boer 1997). Nevertheless, we thought that the mutation would be in a regulatory region and affect the amount of cellular ZipA. Indeed, immunoblotting analysis revealed that the amount of ZipA was slightly but statistically increased in both DS554 (zipA p56 ) and DS631 mutants (ΔrodZ zipA p56 ) compared with DS645 (wild-type) and DS679 (ΔrodZ): a 1.26-fold and 1.64-fold increase, respectively ( Fig. 1C and D), while the amount of FtsZ was comparable in these strains ( Fig. 1C and D). Measurements of transcribed mRNA by real-time PCR revealed that the amount of mRNA of zipA was statistically higher in DS554 (zipA p56 ) and DS631 (ΔrodZ zipA p56 ) cells than in DS645 (wild-type) or DS679 (ΔrodZ) (Fig. 1E and Fig. S4). In addition, rodZ mutant cells harboring a multicopy plasmid encoding zipA expressed from arabinose promoter P BAD (pDS1019) grew faster than mutant cells carrying an empty vector (Fig. S5A), and were restored the cell shape (Fig. S5B  The amounts of ZipA in WT or in rodZ mutant cells carrying the zipAp56 mutation was significantly higher than those in WT or in the rodZ mutant. The asterisk shows P value < 0.05 as determined by a T-test. (E) Relative quantity of mRNA of zipA in DS645 (WT), DS554 (zipA p56 ), DS679 (ΔrodZ), and DS631 (ΔrodZ, zipA p56 ) as calculated by real-time PCR shown in (Fig. S4). The quantity was normalized by the amount of mRNA of zipA in WT. (F) Phase contrast images of DS631 (ΔrodZ, zipA p56 ) carrying pWM2784 (vector) or pWM2787 (ftsA*). Scale bar represents 2 lm. (G) Suppression of thermosensitivity of ftsZ84 by introduction of zipA p56 . Five microliter of overnight culture of DS708 (WT), DS709 (zipA p56 ), DS710 (ftsZ84), and DS711 (fts84, zipA p56 ) were spotted onto L plates and incubated at 30 or 42°C. The amount of zipA may be increased by the gene dosage effect. These results indicate that the suppressor mutation was within the regulatory region of the zipA gene and caused increased expression of zipA.
Overproduction of ZipA, even a twofold, is toxic to cells, resulting in filaments in wild-type cells (Fig. S5B) (Geissler et al. 2003). A mutant FtsA (FtsA-R286W), known as FtsA*, abolishes the negative effect of excess ZipA by an unknown mechanism, and long-term lethal filamentation is inhibited (Geissler et al. 2003). By analogy, FtsA* should abolish the effect of zipA p56 on the suppression of the rodZ mutant. A vector plasmid or a plasmid encoding ftsA* was introduced into the suppressor strain DS631 (ΔrodZ, zipA p56 ). Cells carrying a plasmid encoding ftsA* became oval or round (Fig. 1F), which was very similar in shape to the rodZ cells, suggesting that the "negative" effect of increased ZipA was canceled by FtsA*. In addition, the growth rate of DS631 carrying ftsA* was slower than that of DS631 carrying an empty vector (44 vs. 36 min). Thus, upregulation of zipA was sufficient to suppress the deficiency in cell shape maintenance.
The zipA p56 mutation is also located in a putative regulatory region of the cysZ gene, whose function is unknown. So far, there is no evidence of relationship between CysZ and FtsA. Growth of rodZ cells expressing extra cysZ from the arabinose promoter P BAD was similar to that of the rodZ cells harboring pBAD24 (data not shown). In addition, as we showed above, the gene dosage effect of ZipA caused suppression of the rodZ cells and production of FtsA* in the DS631 antagonizes the effect of the suppression. Thus, we concluded that the suppression by the zipA p56 mutation occurs through the regulation of expression of zipA, and not cysZ.
zipA p56 can suppress temperature-sensitive growth of cells carrying the ftsZ84 mutation It has been reported that higher expression of zipA can suppress temperature sensitivity of the ftsZ84 mutation by stabilizing the labile FtsZ84 ring structure (Ray-Chaudhuri 1999). To confirm whether a slight upregulation of zipA has the same effect, the zipA p56 mutation was introduced into wild-type (WM1074) or ftsZ84 mutant (WM1125) cells, yielding DS709 (zipA p56 ) and DS711 (ftsZ84, zipA p56 ). As seen in Fig. 1(G), DS711 cells (ftsZ84, zipA p56 ) grew on L plates at the restrictive temperature of 42°C, while WM1125 cells (ftsZ84) failed to form colonies. These results indicate that the increased amount of ZipA expressed in strains carrying the zipA p56 mutation is sufficient to stabilize the assembly of FtsZ84.
The zipA p56 mutation could not restore viability to the mreB or mrdA deletion mutants It has been shown that overproduction of FtsZ can restore viability to round-shaped mreB or mrdA mutants (Bendezu and de Boer 2008). Thus, we examined whether the zipA p56 mutation rescued the lethality and defect in the cell shape of the mreB or mrdA mutants. The zipA p56 mutation did not suppress the lethality of either of the strains in L broth (Fig. S6A). We could not conclude whether the zipA p56 mutation restored their shape in L broth. The mreB or mrdA mutants are viable in minimal synthetic media. In M9 minimal medium, the zipA p56 mutation did not suppress the cell shape defect of mreB or mrdA mutants or that of the rodZ mutant (Fig. S6B). Thus, differences in the causes of conversion from rod cells to round cells between the mreB or mrdA mutants and the rodZ mutant were evident.

Elongation of rodZ mutant cells during the cell division cycle
It seemed that suppression of the rodZ mutant by overproduction of ZipA was related to cell division. We thus examined the manner of division of the rodZ mutant. Cells were cultured in M9 synthetic medium, because cells in this medium were more round than those in rich medium, although oval-shaped cells were observed in both synthetic and rich media ( Fig. 2A). We carried out timelapse observations of rodZ mutant cells grown in synthetic medium every 2 min (Fig. 2B). The cells did not swell, but they elongated along the long axis of the cell and divided at the midcell (Fig. 2B). Daughter cells were almost completely round and then elongated gradually. As a result, round daughter cells became oval. To confirm the lateral elongation of cells, we measured the length and width of dividing cells. As seen in Fig. 2(C), the width was almost constant (mean: 1.7 lm), while the length continuously elongated and reached twofold of the width (maximum length: 3.4 lm) (Fig. 2C). These results indicate that cells elongated unidirectionally while the cell width was constant. In addition, this result suggests that, even though the rodZ mutant cells are round, the cells retain cell bipolarity; that is, a determinant of the direction of elongation.

Directional oscillation of Min protein in round cells
To confirm that rodZ mutant cells retain cell bipolarity, we observed the oscillation of the MinD protein. Min proteins, which consist of MinC, MinD, and MinE, oscillate between the end poles to control assembly of FtsZ at the midcell, and then Z ring formation is inhibited at the cell poles (Raskin and de Boer 1999;Hale et al. 2001). Indeed, GFP-MinD rapidly oscillated from one pole to the other in a rod-shaped wild-type E. coli cell (Fig. 3A). It has been previously shown that oscillations of MinD are often random in spherical mrdB or mreB mutant cells (Begg and Donachie 1998;Corbin et al. 2002;Shih et al. 2005). Similar results were obtained in spherical wild-type cells after the addition of A22 that inhibits binding of ATP to MreB (Fig. 3B). Random oscillation patterns suggest a loss of polarity in mrdB, mreB round cells, or spherical wild-type cells by A22. However, oscillation of MinD was not random in spherical rodZ mutant cells in which a polar cap of GFP-MinD was observed (Fig. 3C). The visible polar cap rapidly moved to the opposite side, and the direction of oscillation of GFP-MinD was constant in the rodZ cells. This movement of GFP-MinD supports the hypothesis that the rodZ cells retain a fixed polarity and can divide symmetrically at the midcell.

Cell elongation by arrest of cell division
If the direction of cell elongation is transmitted to daughter cells, division planes that are perpendicular to the axis should be parallel to each other in daughter cells, and then descendants would form chains. In order to observe cells, we used a gelatin-mounting method (Mason and Powelson 1956;Yamaichi and Niki 2004) to avoid completely separating cells from each other after cell division.
In fact, newly divided cells were arranged linearly when rodZ mutant and wild-type cells were mounted in medium containing gelatin (Fig. 4A). In contrast, a cluster was formed by round cells of the mreB mutant, which lose polarity after cell division (Shih et al. 2005) (Fig. 4A). Furthermore, the effect of inhibition of cell division on cell morphology was examined to confirm the fixation of the direction of cell elongation. Cephalexin, which inactivates PBP3 that synthesizes peptidoglycan at the septum during cell division, inhibits cytokinesis of bacteria and results in elongation of the cells Weiss et al. 1997) (Fig. 4B). On the other hand, when the division of the rodZ mutant cells was inhibited by cephalexin, the mutant cells became warped filaments with swollen balloons, extending along the long axis (Fig. 4B). It is currently unclear what constitutes these balloons. In addition, cells treated with cephalexin were wider than untreated cells. Interestingly, round-shaped mreB mutant cells became larger spheres after the addition of cephalexin (Fig. 4B). It seems likely that the mreB mutant cells swelled as a whole because they had lost polarity for the direction of cell elongation and expanded in all directions. Likewise, cells with both rodZ and mreB deleted formed clusters after cell division and became larger spheres upon inhibition of cell division ( Fig. 4A and B). FtsZ localizes to the division site and forms a ring-like structure called the Z ring (Bi and Lutkenhaus 1991;Ma et al. 1996). GFP-tagged FtsZ (FtsZ-GFP) formed a Z ring at the division plane located at the midcell of short cells or at the quarter sites of elongated cells in the wild-type (Fig. 4C). FtsZ-GFP was still localized at the midcell of the rodZ mutant cells in the absence and presence of cephalexin, and additionally, FtsZ rings were perpendicular to the long axis as in the wild type (Fig. 4C). Thus, directional elongation by inhibition of cell division indicates that constant directionality of the long axis clearly exists in the rodZ mutant cells. These results also indicate that inhibition of cell division of the rodZ mutant results in recovery of the cell shape to a rod.

Deletion of PBP2 in the rodZ mutant
It has been thought that lateral peptidoglycan synthesis requires PBP2 (Spratt 1975;Spratt and Pardee 1975), and that PBP2 is usually indispensable for cell growth in E. coli. However, PBP2 should be dispensable in the rodZ mutant because, presumably, lateral elongation of the cell wall is unnecessary if the rodZ mutant consists of only polar caps. A deletion mutant for the mrdA gene encoding PBP2 can grow on synthetic medium, but not on nutritionally rich medium (Bendezu and de Boer 2008) (Fig. S6A). Cells deleted for both rodZ and mrdA, as well as cells deleted for only mrdA, failed to grow on nutritionally rich medium, suggesting that the rodZ mutation could not suppress the growth deficiency caused by the mrdA mutation (Fig. S6A). The inability to delete mrdA from the rodZ mutant may be because: (1) the rodZ mutant retains a very small portion of cylindrical peptidoglycan that is synthesized by PBP2, and/or (2) PBP2 plays a role in other essential functions such as the establishment of cell polarity. As seen in Figure 4(A and B), cells deleted for mrdA or both mrdA and rodZ formed aggregates after division and became larger spherical cells, suggesting that cells with deleted mrdA lost cell polarity.

Active peptidoglycan synthesis only at the midcell of the rodZ mutant
The elongation of a rod cell is dependent on the synthesis of lateral peptidoglycan at the cylindrical part.
We predicted that the active site of peptidoglycan synthesis for cell growth should be changed in the round cells of the rodZ mutant. Fluorescent vancomycin was used to detect the active sites of peptidoglycan synthesis (Varma et al. 2007;Varma and Young 2009). Vancomycin specifically binds to terminal D-alanyl-D-alanine moieties of peptides that are to be closely linked to sugar chains in alternating N-acetylglucosamine and N-acetylmuramic acid. Even though D-Ala-D-Ala are only present in E. coli cells as 1% or less of total muropeptides (Glauner et al. 1988), Young and colleagues applied a vancomycin-labeling method of isolated sacculi to visualize the sites of peptidoglycan insertion (Varma et al. 2007;Varma and Young 2009). Isolated single giant molecules of the peptidoglycan layer, which maintains the rod shape of living cells, were labeled by fluorescent vancomycin. Fluorescent spots were distributed as spirals along the lateral side of rod-shaped sacculi (Fig. 4D). However, fluorescence in the rodZ mutant was localized only at the midcell plane and septum in dividing cells (Fig. 4E), indicating that peptidoglycan synthesis actively occurs at the midcell of the rodZ mutant. This result is consistent with the assumption that the lateral part of the rod cell is remarkably shortened or lost in the rodZ mutant.

Delay of cytokinesis by increased ZipA
So far, we found that inhibition of cell division of the rodZ mutant resulted in its apparent rod shape (Fig. 4C). It has been previously shown that overproduction of ZipA results in elongation of a cell, possibly because division is inhibited (Geissler et al. 2003). These results allow us to hypothesize that suppression of the rodZ mutant by zipA p56 is due to inhibition or delay of cell division of the mutant. As shown in Figure 2(B), the rodZ cells elongate laterally during the division process. Thus, we examined whether cell division of rodZ + cells is inhibited by zipA p56 . Cell growth was measured by the following two independent methods: measurement of the turbidity of the culture medium that reflects an increase in cell mass, and cell counting using a flow cytometer. The profiles of turbidity were similar between wild-type and zipA p56 (Fig. 5A), and the mass doubling times were the same (Table S1). The similar increase in cell mass between wild-type and zipA p56 indicates that both strains elongated at almost the same rate. On the other hand, differences were observed in the increase in cell number between wild type and zipA p56 (Fig. 5B). The number of wild-type cells increased more rapidly than that of the zipA p56 cells. This inconsistency between cell mass and cell number can be explained by the zipA p56 cells taking a longer time to divide into daughter cells compared to the wild-type, while their cell volume increased at a similar rate. Thus, these results indicate that the zipA p56 mutation leads to a delay of cell division. Consequently, we propose that the zipAp56 mutation causes a delay of cell division in the rodZ mutant, and hence the mutant becomes apparently rod shaped.

Discussion
Mechanism of suppression of the rodZ mutant by zipA p56 Morphological mutations that convert rod-shaped E. coli to round shaped have been found in some genes, including mreB, mrdA, and rodZ. The former two mutations result in relatively larger round cells with defects in the maintenance of cell polarity (Begg and Donachie 1998;Corbin et al. 2002;Shih et al. 2005) (Fig. 4A and B). On the other hand, the rodZ round cell may still have a defined single cell polarity and elongates laterally, but probably with a shortened central cylinder of a normal rod cell or squat rod (Shiomi et al. 2009) (Fig. 6A).
An excess amount of ZipA perturbs the physiological function of the Z ring in the divisome and hence inhibits or delays the closure of the divisome for cytokinesis (RayChaudhuri 1999;Geissler et al. 2003). Inhibition of (1) Rod-shaped cells of Escherichia coli consist of polar caps and a central cylinder, retain cell polarity, and elongate laterally. Peptidoglycan is synthesized at the entire central cylinder so that cells elongate laterally until the cell length is twice as long as a newborn cell. Polar caps are derived from septa, and the peptidoglycan is synthesized during cell division. After that, polar caps inert with respect to synthesis of the peptidoglycan layer (de Pedro et al. 1997).
(2) Newly synthesized peptidoglycan is inserted at the midcell in ΔrodZ cells so that a round cell gradually elongates along a single axis. Synthesis of the peptidoglycan layer at septum is efficient in forming ovoid-shaped cells. cell division by the zipA p56 mutation could give enough time for "lateral" peptidoglycan synthesis at the midcell to elongate the cell, resulting in recovery of a rod shape (Fig. 6A). Thus, this "negative" effect of ZipA on the Z ring results in elongation of the rodZ mutant to recover the rod shape. How can overproduced ZipA delay cell division? Because ZipA stabilizes the Z ring possibly through direct interaction or indirect effects through ClpXP (RayChaudhuri 1999;Pazos et al. 2013), and thus overproduced ZipA would overstabilize the Z ring, which can cause a delay in cell division.
Synthesis of the lateral cell wall to maintain the rod shape is executed by the elongasome, which is a protein complex that includes PBP2 and MreB (den Blaauwen et al. 2008). In the rodZ mutant, the activity of the elongasome may be reduced. In the rodZ, zipA p56 mutant, presumably elongasome activity is prolonged at the Z ring during the duration of cell division, and thus cells look rod shaped. This effect of the elongasome can be expected only in the rodZ mutant, but not in the mreB, mrdA, or presumably the mrdB mutant. Thus, RodZ is required for the elongasome to work away from the septum but is not required for it to work at the septum.
Moreover, it has recently been reported that FtsZ and ZipA are required for preseptal peptidoglycan synthesis at the septum, known as PIPS in E. coli as well as in Caulobacter crescentus (Aaron et al. 2007;Varma and Young 2009;Potluri et al. 2012). Peptidoglycan in the rodZ mutant is mainly synthesized at the septum (Figs. 4E and 6A). Thus, overproduction of ZipA might activate peptidoglycan synthesis at the septum via PIPS. It has been suggested that some of the components of the elongasome are involved in PIPS (Potluri et al. 2012;van der Ploeg et al. 2013). It seems likely that in the rodZ, zipA p56 mutant, both activities of the elongasome and PIPS cooperatively help round cells to recover the rod shape.
Our results suggest an indirect relationship between RodZ and FtsZ. Bendezu et al. (2009) have shown a genetic interaction; that is, overproduced FtsZ improves the growth of the rodZ mutant, and rodZ cells overproducing FtsZ are not round. Although the mechanisms behind these phenomena are unclear, the results suggest relationships between RodZ and FtsZ-induced cell division.

Analogy between round-shaped E. coli and other round-shaped bacteria
The manner of cell growth of the rodZ mutant is analogous to that of the ovococcus bacteria, Streptococcus pneumoniae, and Lactococcus lactis, as follows: (1) round or ovoid cells, (2) retainment of polarity, and (3) active peptidoglycan synthesis only at the midcell but not within caps (Fig. 6B). Interestingly, inhibition of cell division of ovoid-shaped Lactococcus lactis results in rod-shaped cells (Perez-Nunez et al. 2011). This transition from ovoid cells to rod is similar to the mechanism of recovery of the cell shape in the rodZ mutant by the zipA p56 mutation. Thus, the rodZ mutant can be referred as a "pseudoovoid" shaped mutant.
In contrast, the manner of cell growth of the mreB or mrdA mutants is analogous to staphylococcus-type bacteria. Mutant cells increased as a whole (Figs. 4B and 6B), and larger round cells were divided into smaller daughter cells as seen in Staphylococcus aureus (Tzagoloff and Novick 1977). In addition, mutants of mreB and mrdA formed in grape-like clumps (Fig. 4A). The formation of the grape-like clump after cell septation indicates that the planes of cytokinesis are not constant in each cell. MinD moves often randomly in mreB mutants (Begg and Donachie 1998;Corbin et al. 2002;Shih et al. 2005). Thus, the mreB and mrdA mutants do not retain cell bipolarity correctly. Although both RodZ and MreB are colocalized to maintain the rod shape, each plays a different role in maintaining cell shape.

Cell shape and cell growth
The surface-to-volume ratio is one of the important parameters that characterizes prokaryotes. Because of their smaller size than eukaryotic cells, bacteria have a large surface-to-volume ratio. This guarantees that nutrients and metabolites can be easily transported between all parts of the cell. When the cell volume is equal between a round cell and a rod cell, the surface-to-volume ratio is naturally larger in the rod cell. This suggests that a rod shape should be advantageous to cell growth for relatively larger bacteria. As shown in a previous study (Shiomi et al. 2013) and in this study, recovery of the defect in growth is always accompanied by conversion of a round cell to a rod cell. Furthermore, whatever the suppressors of the rodZ mutation are, recovery of the rod shape restores the cold-sensitive growth and swarming inability of the rodZ mutant.
Indeed, the cell volume of streptococci and staphylococci is smaller than that of E. coli. The conversion of a rod cell to a round cell in E. coli means a reduction in the surfaceto-volume ratio. Thus, only one mutation of the rodZ gene causes a metabolic revolution in the ability to utilize some carbon sources and other metabolites (Ito et al. 2005) and subsequently results in growth deficiency. Mutating only one base pair, such as in the promoter of the zipA gene (zipA p56 ) and mreB gene, would be a relatively simple way to recover the rod shape. Such simple methods might be utilized to convert cell shapes from round to rod.

Isolation of suppressors and genomic sequencing
Details were provided previously (Shiomi et al. 2013). Independent single colonies of DS6 (ΔrodZ) were grown in L broth containing kanamycin at 37°C. Cells were transferred every day to fresh L broth containing kanamycin and kept at 37°C. Five to 7 days after splitting cells, cells were plated on L plates containing kanamycin and incubated at 37°C. Bigger colonies were isolated as suppressors of the slow growth phenotype of ΔrodZ. Chromosomal DNAs of wild-type and suppressor strains were extracted and sequenced by the Solexa system.

Microscopy and measurements of the length and width
Cells were grown at 37°C in L broth or M9 medium containing 0.25% glucose overnight. The culture was diluted in L broth or M9 glucose and exponentially grown at 37°C. Cells were observed under an epifluorescence microscope (Zeiss, AXIO, Oberkochen, Germany). To take fluorescent images, delta vision microscopic system (Applied Precision, Issaquah, WA) was applied. The lengths of the long and short axes were measured as previously reported (Shiomi et al. 2008). Briefly, digital images were processed by the software Metamorph and changed into binary images to detect cell outlines. Each cell in the binary images was automatically measured for the length of the longest line (the long axis of the cell) and maximum breadth perpendicular to the longest line (the short axis of the cell).

Isolation of saculli and vancomycin labeling
Cells were grown in M9 medium containing 0.25% glucose to log phase at 37°C and harvested by centrifugation for 10 min at room temperature. The pellet was resuspended in 3 mL of M9 medium containing 0.25% glucose and added dropwise to a tube containing 6 mL of 6% SDS. The samples were boiled for 3 h and incubated overnight at RT with gentle stirring. The next morning, the tubes were boiled for 1 h and sacculi were sedimented by ultracentrifugation at 173,970 g for 15 min at 30°C. The pellets were resuspended in 2.5 mL of 4% SDS in a closed tube and incubated in boiling water for 2 h. Sacculi were sedimented as above and washed with 1% SDS. Sacculi were resuspended with PBS containing 300 lg/mL a-chymotrypsin and incubated overnight at 37°C. Then, the next morning, the sacculi were pelleted as above, and resuspended in 1% SDS, incubated in boiling water for 2 h, and pelleted. The final pellet was resuspended in 100 lL H 2 O and stored at 4°C. The isolated sacculi were incubated with Van-FL (Molecular Probe, Carlsbad, CA) at room temperature for 1 h in the dark. The samples were washed with H 2 O and resuspended in H 2 O.

Anti-FtsZ antiserum production
To raise antiserum against FtsZ, a 6x His (His 6 ) tag was added to the N-terminus of FtsZ. KRX carrying pDS996 (a plasmid encoding his 6 -ftsZ) was grown in L broth supplemented with kanamycin and 0.25% glucose overnight. Cells were diluted 100 times into fresh L broth supplemented with kanamycin and incubated for 3 h at 37°C. Then cells were transferred to 25°C and incubated for 30 min. 0.1% rhamnose was added to the culture and the cells were incubated at 25°C overnight. Cells were harvested by centrifugation and resuspended in Buffer A (50 mmol/L NaPO 4 [pH 7], 300 mmol/L NaCl, 8 mol/L urea) supplemented with one tablet of EDTA-free protease inhibitor cocktail (Roche, Penzberg, Upper Bavaria, Germany). The mixture was incubated for 1 h at 37°C, and centrifuged at 13,000 g for 20 min at 4°C. The supernatant containing His 6 -FtsZ was mixed with Talon metal affinity resin (Clontech, Mountain View, CA) equilibrated by Buffer A. The mixture was incubated at 4°C overnight. The mixture was centrifuged and the supernatant removed. The resin was resuspended in Buffer A and centrifuged. This step was repeated three times. His 6 -FtsZ was eluted by Buffer A containing 150 mmol/L imidazole. The protein was used to immunize rabbits, and anti-FtsZ antiserum was produced by Biogate (Gifu, Japan).

Immunoblotting
Immunoblotting was performed with a blot using an antibody against ZipA (Geissler et al. 2003), which was provided by Dr. Margolin, or anti-FtsZ. Three clones for each strain were independently grown to mid log phase (OD 600 = 0.6). The amounts of samples loaded onto electrophoresis gels were equally adjusted, monitoring by Coomassie Brilliant Blue (CBB) staining.

Measurements of cell growth and cell number
Cells were grown at 37°C in L broth overnight. The culture was diluted 1:100 in fresh L broth and grown at 37°C. The Klett value was measured every 15 min. Culture samples (1 mL) were harvested at the indicated time points and stored on ice to measure cell number. Stored cultures (300 lL) were precipitated, washed with 500 lL of TE (pH 8), and resuspended with 90 lL of TE (pH 8). Next, 900 lL of 77% ethanol was added to the cells, followed by incubation at 4°C for 30 min. Following the incubation, 200 lL of 0.1 mol/L phosphate buffer (pH 9) was added to 20 lL of the samples. The samples were washed twice with 200 lL of 0.1 mol/L phosphate buffer (pH 9). The pellets were resuspended in 100 lL of 0.1 mol/L phosphate buffer (pH 9) and 100 lL of 3 lg/mL of FITC was added. The samples were incubated at 4°C overnight, then washed twice with 20 mmol/L PBS (pH 9), and resuspended in 1 mL of 20 mmol/L PBS. The number of cells in 250 lL of sample was measured by a cytometer (Cell Lab Quanta; BECKMAN, Indianapolis, IN).

Real-time PCR
Total RNA was purified using an RNeasy Mini kit (QIAGEN, Venlo, Netherlands) following the supplied protocol. Then the RNA was reverse transcribed to obtain cDNA using a PrimeScript RT reagent kit (TAKARA, Shiga, Japan). Real-time PCR was performed and analyzed using SYBR Premix Ex Taq and a Thermal Cycler Dice Real Time System (TAKARA). mRNA was purified from three independent clones for each strain.
for sequencing and construction of plasmids and strains. This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan to H. N., and a Grant-in-Aid for Young Scientists (Start-up and B), Grant-in-Aid for Scientific Researcch on Innovative Areas, and an NIG Postdoctoral Fellowship to D.S. Some strains were supplied by the National Bioresource Project, Japan: E. coli (NIG, Mishima).

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. Morphology of the mutants grown in minimal medium. Figure S2. Suppression of the cold-sensitive growth of the rodZ mutant by the zipA p56 mutation. Figure S3. Suppression of the swarming defect of the rodZ mutant by the zipA p56 mutation. Figure S4. Quantitative real-time PCR analysis of cells carrying WT or the mutant promoter of the zipA gene. Figure S5. Growth and cell morphology of BW25113 (WT) or DS290 (ΔrodZ) cells carrying pBAD24 (vector; blue) or pDS1019 (pBAD24-zipA; red). Figure S6. Viability of the mreB or mrdA mutants carrying the zipA p56 mutation. Table S1. Mass doubling time (min) of strains carrying suppressor mutations grown in L broth at 37°C. Table S2. Strains and plasmids used in this study.