Bex, the Bacillus subtilis Homolog of the Essential Escherichia coli GTPase Era, Is Required for Normal Cell Division and Spore Formation

Abstract

The Bacillus subtilis bex gene complemented the defect in an Escherichia coli era mutant. The Bex protein showed 39% identity and 67% similarity to the E. coli Era GTPase. In contrast to era, bex was not essential in all strains. bex mutant cells were elongated and filled with diffuse nucleoid material. They grew slowly and exhibited severely impaired spore formation.

Proteins of the GTPase superfamily function as molecular switches for diverse cellular processes. The Era protein of Escherichia coli is a member of the superfamily (1, 18). Although Era was originally named for its similarity to Ras, it is no more closely related to Ras than to any other small G protein (6). Purified E. coli Era protein binds GTP and GDP and has GTPase activity in vitro (18, 36). The exact pathways in which Era takes part have not been established. However, phenotypic studies of era mutants have suggested that Era is involved in several physiological processes. It is thought that Era might act as a cell cycle regulator, functioning after segregation of the nucleoids and before or during cell division (4). Era contains a KH-like RNA binding motif (7, 12). Era proteins of E. coli and Streptococcus pneumoniae are associated with 16S RNA in vivo, and RNA binding increases their GTPase activity in vitro, suggesting a link to protein synthesis (12, 19, 32). It has been hypothesized that the RNA-bound form is in an active state and signals cell division, possibly coordinating cell division with protein synthesis (19). Era has also been associated with energy metabolism (24, 28). We report here that Bex is the functional Bacillus subtilis homolog of Era. We find that bex has a central role in cell propagation, but it is not essential in all strains.

Isolation of the bex gene.

Using a pBR322-based library (kindly supplied by D. Henner) constructed with total DNA from B. subtilis strain W168, we selected clones that complemented loss of the essential era gene from E. coli cells, essentially as described previously (27). Strain BSP123, which is potentially resistant to streptomycin and has had the native era gene deleted, contains a λcI857 era⁺ rpsL⁺ prophage (λBP4) that can be cured by increasing the temperature; the presence of rpsL⁺ in the prophage renders the strain streptomycin sensitive. Library transformants of BSP123 that survived curing and subsequent growth at 42°C were selected on Luria-Bertani (LB) agar containing streptomycin (50 μg/ml). One transformant contained a plasmid (pBP17; also known as pWR1-1) bearing a 2.1-kb insert (comprised of two EcoRI segments) inserted at the EcoRI site of pBR322. DNA sequence analysis (by routine methods; GenBank accession number U18532) revealed two open reading frames (ORFs) oriented in the same direction. The first ORF was cdd, previously mapped to 225° on the B. subtilis chromosome and shown to encode cytidine deaminase (35). The second ORF overlapped the 3′-terminal 20 nucleotides of cdd and encodes a putative 301-amino-acid protein that was 39% identical (67% similar) to E. coli Era. The 5′ portion of this ORF was reported previously, but the gene product was not predicted because of a frameshift error (35). Deletion analysis of pBP17 (not shown) established that this Era-like gene was necessary and sufficient to complement an E. coli strain (BJ24 [12]) containing the well-characterized rnc-40::mini-Tn10 mutation of E. coli, in which era expression is dependent on induction by tetracycline (37, 38). On the basis of these results, the gene was termed bex (a homonym of becS [for bacillus era complementing segment]). pBP17 did not complement mutations in either the rnc or recO gene of E. coli, which flank the era gene in that organism.

In its N-terminal region, Bex contains the conserved G1, G3, G4, and G5 sequences typical of bacterial GTPases (17) and the G2 sequence (QTTR) characteristic of the Era GTPase subgroup. Bex also exhibits extensive homology to Era in its C-terminal region, including the KH-like domain (12, 41). No other B. subtilis protein (including the small, essential GTPases Obg [39], YsxC [29], and YloQ [3]) shows comparable similarity to Era. Proteins homologous to Era are found encoded in nearly all bacterial genomes that have been completely sequenced and have also been found in several eukaryotic genomes; they have a highly conserved structure. They also appear to have a highly conserved function, as Era depletion in temperature-sensitive mutants of E. coli can be complemented by the homologous genes from a number of species (20, 25, 40-42). Era is essential for E. coli (9, 15), and its homologs in other species have been reported to be essential (2, 12, 31).

Disruption of the bex gene.

We tried several ways to disrupt the bex gene. Our studies with B. subtilis strain IS75 (his leu met; obtained from J. Hoch) and its derivatives suggested that bex was essential. First, we tried to transform the strain with chloramphenicol-resistant (Cm^r) plasmids that were unable to replicate in B. subtilis and contained different portions of the bex gene. Plasmids that bore either an intact bex gene or only its distal portion easily transformed strain IS75 so that it exhibited stable Cm^r, whereas a plasmid containing only an internal portion of the bex gene yielded no transformants (data not shown). Second, we tried to disrupt bex in IS75 derivatives by double crossovers using a linearized plasmid, pRB3248 (containing a cat gene flanked by portions of the bex gene), as the donor. (Details of strain and plasmid constructions are available on request.) The linearized plasmid efficiently transformed a merodiploid derivative of strain IS75, RS8740, which had a second copy of bex inserted at amyE together with the spc marker, and conferred Cm^r on the strain (Table 1). In contrast, a haploid derivative, RS8739, with only spc inserted at amyE, did not yield viable transformants with linearized pRB3248. Interestingly, transformants of RS8739 appeared as small colonies after 5 to 7 days of incubation. Microscopic analysis (not shown) revealed that these transformants formed very long filaments. However, colony development ceased at this point, and these transformants could not be propagated on fresh medium. We were not successful in determining the genetic structure of these transformants by PCR, suggesting that their DNA was degraded.

TABLE 1.

Stable transformants obtained with haploid and merodiploid strains

Transforming plasmida (relevant genotypeb)	No. of stable transformants (per μg of DNA) obtainedc
	RS8739 (bex+)	RS8740 (bex+/bex+)	BR151 (bex+)
pRS3248 (bex′-cat-′bex)	0d	10,000	15,900e
pPP537 (bex′-neo-′bex)	ND	ND	6,400e
pVK215 (ppsD′-neo-′ppsD)	ND	ND	8,400
pVK2 (amyE′-neo-′amyE)	ND	ND	20,300

^aPlasmids replicated in E. coli but not in B. subtilis. They were linearized before transformation to favor integration by double crossover. pRS3248 was constructed in two steps. First, the 1.5-kb BamHI fragment bearing the cat gene (encoding Cm^r) from pMI1101 (obtained from S. Roels) was inserted into the unique BclI site in the bex gene on pBP17 to give pRS3247. Next, the 1.8-kb EcoRI-SalI fragment bearing the neo gene (encoding Neo^r/Km^r) from pDG364-Kan (also from S. Roels) was rendered flush ended and inserted into the EcoRV site of pRS3247 to give pRS3248. To construct pPP537, a neo cassette (11) was first inserted into the MscI site of pPP525 to give pPP532. A 450-bp region of the cdd gene, located 5′ to bex, was amplified by PCR and ligated to SnaBI, ClaI-digested pPP532 to yield pPP537. In pPP537, the 5′ 68% of bex was replaced by the neo cassette. pVK2 and pVK215 are integrative plasmids that were used to test for homologous recombination at amyE and ppsD, respectively.

^bcat, chloramphenicol resistance; neo, kanamycin/neomycin resistance; ′, truncation or disruption.

^cSee text for strain descriptions. Plasmid DNA was prepared by routine methods from E. coli, quantified by absorbance, and used to transform the indicated strains as described previously (8). Selection was made at 30°C on LB medium containing neomycin (12 μg/ml) or chloramphenicol (5 μg/ml). ND, not determined.

^dSmall colonies appeared after 5 to 7 days but could not be subcultured (see text).

^eTransformants appeared after several days, of which ∼70% could be subcultured.

To explore the function of bex, we placed it under the control of the tightly regulated, xylose-inducible P_xyl promoter (14) in strain BR151 (trpC2 lys-3 metB10). The coding region of bex with its ribosome-binding site but without its promoter was amplified by PCR using Pfu polymerase with DNA from B. subtilis strain BR151 and inserted into the EcoRV site of pBluescript KS (Stratagene) to give pPP525. A HincII-SmaI fragment from pPP525 was ligated into the filled-in BamHI site of pX (14). The resulting plasmid, pPP528, was used to transform B. subtilis BR151, yielding a strain (SL7745) with bex under the control of the P_xyl promoter inserted by double crossover at the amyE locus. The bex gene in SL7745 was then inactivated at its natural locus with a neo cassette (11), which replaced the 5′ 68% of the bex gene (to the MscI site). The resulting strain was named SL7976, and the bex null mutation was confirmed by PCR. Strain SL7976 grew normally on LB (30) and SSA (33) solid media in the presence of xylose (2% wt/vol), while in the absence of the inducer it formed slow-growing, small colonies of irregular shape. The doubling time of SL7976 in liquid MSSM (modified Schaeffer's sporulation medium) (23) with xylose was 45 min; without xylose, it was about 120 min (Fig. 1c). Without xylose, SL7967 formed long filaments (not shown). The growth rates of BR151 and the BR151 derivative SL7745, containing two intact copies of the bex gene, one of which was under P_xyl control, were not affected by xylose (Fig. 1a and b).

FIG. 1.

Effect of inactivation of the bex locus on growth of B. subtilis. Bacteria were grown at 37°C in MSSM (open squares) or MSSM supplemented with 2% (wt/vol) xylose (filled squares). The growth of parent strain BR151 (bex⁺) (a), SL7745 (bex⁺ amyE::P_xyl-bex) (b), SL7976 (bex::neo amyE::P_xyl-bex) (c), and SL7968 (bex::neo) (d) was monitored by measuring the A₆₀₀.

bex is not essential in strain BR151.

The growth of haploid transformants of RS8739 by linearized pRS3248, albeit terminal, had suggested that bex might not be essential under all conditions. Indeed, bex was found not to be essential in strain BR151, as we were able to obtain viable derivatives in which the sole copy of bex was inactivated. Linearized plasmid pPP537 was used to transform BR151 so as to obtain, by double crossover, a bex null mutant (SL7968). The bex null mutation was confirmed by PCR. The bex null transformants grew slowly and formed small colonies of irregular shape that appeared only after several days at 37°C; they were obtained at frequencies comparable to those for transformants for crosses at other loci (Table 1). We were also able to obtain viable transformants of BR151 with BamHI-linearized pRS3248, one of which was designated SL10137. The colony and cell morphologies of the bex null mutants SL7968 and SL10137 were similar to those of SL7976 on solid medium lacking xylose and the terminal transformants of RS8739 by linearized pRS3248. Strains SL7968 and SL10137 retained the mutant phenotype on initial subculture. However, on repeated subculture, the aberrant properties of the mutant strains became less pronounced, suggesting the acquisition of one or more suppressor mutations. This behavior has not been studied further. To avoid suppressor mutations, the bex mutant SL7968 was reconstructed by transformation before each set of experiments reported below; the bex disruption was confirmed by PCR after each reconstruction. Strain SL7968 grew slowly in liquid medium (Fig. 1d). In such growth experiments, the viable count was about 1 to 10% that of the parental strain, BR151. The doubling time (about 150 min) was consistently slightly lower than that of SL7976 grown in the absence of xylose, possibly because of some slight expression of bex in the latter strain.

The nucleoid morphology of a bex null mutant was studied by fluorescence microscopy of exponential-phase cells (grown in MSSM) fixed with 0.37% formalin and stained with 4′,6′-diamidino-2-phenylindole (DAPI) (5). The bex null mutant SL7968, as well as the conditional mutant SL7976 grown without xylose (not shown), formed chains of elongated cells with diffuse nucleoids which occupied most of the cell (Fig. 2B). In contrast, the cells of the parent, BR151, were much shorter and contained condensed nucleoids (Fig. 2A). In strain SL7976, the morphological effects of bex inactivation were reversed by expression of bex, inserted at amyE, from the P_xyl promoter (not shown), confirming that the morphological defect was the result of bex inactivation, and not a polar effect on genes downstream of bex.

FIG. 2.

Micrographs of B. subtilis strain BR151 (bex⁺) (A) and its isogenic derivative SL7968 (bex::neo) (B), showing nucleoid morphology. Mid-exponential-phase bacteria were grown in MSSM, fixed, and stained with DAPI. Bacteria were visualized by a combination of fluorescence microscopy and phase-contrast microscopy (5). The scale bar in panel B also applies to panel A.

Nucleic acid staining with propidium iodide was used to enhance the visibility of cell divisions (10) in order to assess cell length. The extent of cell elongation in the mutant is readily apparent (Fig. 3). More than 90% of the cells of the parent strain were less than 6 μm in length, whereas 70% of the bex::neo mutant cells were greater than 6 μm in length, with some being over 20 μm. The long mutant cells generally did not lie in one plane, so their lengths may have been underestimated. Membrane staining with FM4-64 (26) and peptidoglycan staining with wheat germ agglutinin (22) confirmed that cells of the mutant strain were grossly elongated compared to those of the parent strain (not shown).

FIG. 3.

Frequency distribution of cell lengths of the parental strain BR151 (bex⁺) and Bex null mutant SL7968 (bex::neo). The strains were in the exponential phase and were grown in MSSM. Image analysis was performed with a Macintosh Power Mac G4 computer. Images were scanned using a Minolta QuickScan35 scanner and a QS35 utility provided by Minolta in Adobe Photoshop version 6.0. The images were analyzed using Scion Image 1.62c software. The lengths of 400 BR151 cells and 200 SL7968 cells were measured.

Our observations are consistent with the bex null mutation having a marked effect on nucleoid segregation and cell division. It remains unclear whether the effect of Bex is direct or indirect. The very long cells were generally filled with DAPI-staining material (Fig. 2B), suggesting that chromosome replication continued in the absence of cell division. The large DAPI-staining regions could well consist of several genomes. The inhibition of division in these cells could be a consequence of nucleoid occlusion (21, 34) by the large diffuse nucleoids. It is not clear how the bex mutant divides, and it may be that nucleoid occlusion can occasionally be overcome. In the bex mutant, we rarely saw cells with bilobed nucleoids or with the two nucleoids that are typical of predivisional cells of the parent strain. The bex mutant had a low plating efficiency. The colonies recovered by plating the cultures used in these experiments retained the mutant phenotype, indicating that the heterogeneity in morphology was not caused by suppressor mutations.

We do not yet understand why bex is apparently essential in the strains with the IS75 background but not in strain BR151. This difference could be due to differences in the strains themselves, subtle differences in the growth conditions used, or the presence of unknown suppressor mutations in some backgrounds. Nevertheless, bex is required for normal growth and cell division in both strains and is able to complement the cell division defects of an E. coli era mutant. These observations show clearly that the function of the bex/era family is highly conserved.

Impairment of Era function in E. coli affects cell morphology. However, the morphology we have described for the bex null mutant is different from that described for E. coli with Era defects (4, 12, 16). A cold-sensitive era mutant grown at the nonpermissive temperature resulted in elongated cells with up to four condensed, well-segregated nucleoids (16); a similar phenotype was observed upon depletion of Era (4, 12). Bex shows extensive sequence similarity to Era, including all the identified domains of Era. Further, bex complements the defect in an E. coli era mutant. It is possible that Bex and Era have overlapping but different roles, as suggested by the different morphologies associated with impairment of their function and by the finding that bex is not essential. However, it is also possible that Bex and Era have similar functions. Because era is essential for E. coli, analysis of it is less direct. The phenotype associated with Era results from its depletion or from impaired function, and not from total loss of the protein, as is the case with Bex. Consequently, a more-severe phenotype may be associated with the bex null mutant.

Expression of bex during growth and sporulation.

A bex-gusA transcriptional fusion was used to monitor transcription of bex. β-Glucuronidase was assayed as described previously (13). There was substantial expression throughout growth and about twofold induction at the end of exponential growth (Fig. 4). The postexponential induction was abolished by a spo0A mutation. The time of bex induction was slightly earlier than that of the Spo0A-controlled spoIIA-lacZ fusion (13, 23) assessed in the strain containing the bex-gusA fusion (Fig. 4). The Spo0A dependence of this postexponential induction suggests that Bex may have a particular role in spore formation. Sporulation was assessed 20 h after the end of exponential growth in MSSM by heat treatment for 20 min at 80°C. The viable count of the bex mutant SL7968 at this time was about 0.1% of that of the parent strain. Sporulation of the mutant was severely impaired, with a frequency of less than 10⁻⁷ that of the parent strain. The block in spore formation may indicate a specific role in spore formation or may be a nonspecific consequence of the severely impaired growth of the bex null mutant.

FIG. 4.

Expression of bex-gusA and spoIIA-lacZ during the transition from exponential growth to sporulation in MSSM. Time is relative to the end of exponential growth. Samples were assayed for β-galactosidase and β-glucuronidase activity by using o-nitrophenyl-β-d-galactopyranoside and p-nitrophenyl-β-d-glucuronide, respectively, as the substrate (13). Specific activity is expressed as nanomoles of substrate hydrolyzed per minute per milligram of bacteria (dry weight). bex-gusA expression (circles) and spoIIA-lacZ expression (squares) are shown. The fusions are present in a spo⁺ strain (filled symbols) or a spo0A mutant (open symbols).

Together, these results show that era or its bex homolog is an essential or nearly essential gene in the diverse species E. coli and B. subtilis and that the biological function of these two homologs is conserved and appears to involve key aspects of cell division and/or sporulation. The ability to isolate viable bex knockout mutations in B. subtilis strain BR151, even if it requires the presence of a so-far-uncharacterized suppressor, should enable experimental insight into the physiological role(s) of the widely conserved and unique Era/Bex GTPase. No comparable knockout mutations have been reported for E. coli or any other species.