INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

While complex multicellular behavior in bacteria is obvious in a relatively small number of species, e.g., fruiting body formation in myxobacteria and sporulation in streptomycetes (21, 23), the ability to form organized colonial communities is common (10, 32). Swarming, the coordinated migration of multicellular colonies, is best known in Proteus mirabilis (2, 20, 30, 32, 35) but is also evident to various degrees in other motile flagellated species (2, 12, 20, 29). In Proteus, it is characterized by the differentiation of short motile vegetative cells at the edge of a growing colony into extremely elongated hyperflagellated swarm cells. These align closely along their long axes, forming two-dimensional rafts that migrate by coordinated flagellar action within a film of hydrated polysaccharide secreted by the bacteria, causing rapid extension of the colony boundary (2, 35). In Proteus, migration ceases periodically, and continued growth is accompanied by increased septation and decreased flagellar density (consolidation). Regular cycles of migration and consolidation generate a characteristic pattern of concentric terraces (30, 35).

Flagellar gene expression is strongly upregulated and septation is repressed during differentiation into swarm cells (2, 4, 16). These conditions result primarily from increased transcription of the flagellar master operon flhDC, which is the principal regulator of flagellar assembly and which also modulates cell division, integrating a variety of signals (11-13). Characterization of Proteus transposon mutants and multicopy suppressors of these mutants has identified many genes involved in swarming, and these have been shown or are presumed to influence differentiation and flagellar gene expression (3, 6, 7, 11, 16, 18, 19). In contrast, only one swarming-defective transposon mutant has been shown to be unimpaired in differentiation and cell motility. This mutant, FC18, mutated in the putative sugar transferase gene cmfA, was unable to make an extracellular polysaccharide required for mass cell movement and so was impaired in the mechanics of translocation (17). Here we describe a second swarming-defective transposon mutant of this class, MNS185, which is similarly unimpaired in differentiation and cell motility. We show that the disrupted gene, so far unique to Proteus, influences the shape of swarm cells and so enhances the multicellular alignment essential for population migration.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Mutagenesis and characterization of the mutant. Wild-type P. mirabilis U6450 was mutagenized with mini-Tn5Cm (19). Mutants were selected on chloramphenicol (80 µg ml¹). Swarming was assessed on 1.5% Luria-Bertani (LB) agar plates, while vegetative cell motility was assessed on 0.3% LB agar. The swarming inhibitor glycerol (0.5%) or -p-nitrophenyl-glycerol (100 µg ml¹; Sigma) was added to isolate single colonies. Populations of synchronously differentiating cells were obtained by seeding 200 µl of a stationary-phase LB broth culture (ca 5 × 10⁸ cells ml¹) onto 8-cm LB agar plates and incubating the plates at 37°C (16). Cell length and shape were assessed by phase-contrast microscopy of at least 100 cells. The quantity of surface flagellin was determined by vortexing washed cells, separating supernatant proteins by sodium dodecyl sulfate (SDS)-12% polyacrylamide gel electrophoresis (PAGE), and Coomassie brilliant blue staining. Cell-associated hemolysin activity, normalized to total cell mass (A₆₀₀), was quantitated as described previously (4).

DNA manipulation and sequence analysis. Plasmid DNA was manipulated by standard techniques and maintained in Escherichia coli XL1-Blue (recA1 [F lacI^qZM15 Tn10]). Specific DNA sequences were detected with digoxigenin-labelled probes (Boehringer). Exonuclease III digestion was done with an Erase-a-Base kit (Promega). DNA was sequenced with a T7 kit (Pharmacia), and DNA and protein sequences were analyzed as described previously (11).

Construction of a ccm chromosomal null mutant. Plasmid pBluescript SK (Stratagene) carrying the ccm locus on a 4.9-kbp ClaI fragment was digested with HindIII and SmaI, blunt ended, and religated, removing EcoRV and PstI sites in the polylinker. The 570-bp EcoRV-PstI fragment internal to ccmA was replaced with the SmaI  interposon of pHP45 as previously described (11). The deleted locus was subcloned as a 6-kbp SacI-KpnI fragment into suicide plasmid pGP704 (26), which was then transferred into P. mirabilis U6450. After selection for the interposon Spc^r, ccmA:: integrants were confirmed by Southern hybridization.

Purification of CcmA for antibody production. Primers (see Fig. 4, panel ii) were used to amplify ccmA and the 5' 35 bp, introducing an upstream EcoRI site and downstream BglII and HindIII sites. The EcoRI-HindIII fragment was inserted into pBAD18 (15) to form pBAD-ccm, in which ccmA was controlled by the arabinose-inducible ara promoter and the CcmA C terminus was extended by an Arg-Ser dipeptide. The BglII-XbaI fragment carrying the histidine tag was excised from pQE16 (Qiagen) and inserted into pBAD-ccm, generating pCcmHis, in which the CcmA C terminus was extended by Arg-Ser followed by six His residues. E. coli MC1061 (pCcmHis) was grown in LB broth to an A₆₀₀ of 0.5, arabinose was added to 0.2%, and the induced culture was grown for 3 h. Cells were pelleted, resuspended in 50 mM sodium phosphate (pH 7.8)-300 mM NaCl, and lysed in a French pressure cell at 800 lb/in², and unlysed cells and debris were removed. The supernatant was centrifuged for 1 h at 180,000 × g, and an equal volume of 16 M urea in 50 mM sodium phosphate (pH 7.8)-300 mM NaCl was added before incubation with Ni-nitrilotriacetic acid resin (Qiagen) at 4°C overnight. The resin was washed with 8 M urea-0.1 M sodium phosphate-0.01 M Tris (pH 8), with 0.1 M sodium phosphate-0.01 M Tris (pH 6), and finally with this buffer containing 0.025 M imidazole. CcmA-His was eluted with 0.125 M imidazole in the same buffer, dialyzed against 8 M urea, and purified by isoelectrofocusing in the presence of 8 M urea and 2% ampholytes of pH 3 to 10. Fractions in the pH range of 7.5 to 8.5 were pooled and dialyzed against urea at 2, 0.8, 0.2, and 0.1 M and then against water. Protein was concentrated with a Centricon 3-kDa filter to 0.5 µg ml¹ and used for rabbit immunization. CcmA-reactive antibody was enriched by affinity chromatography and used for immunoblotting at a dilution of 1:500 relative to serum.

Immunoblotting of CcmA. Cells from seeded LB agar plates or LB broth were harvested into ice-cold 0.9% saline-1 mM EDTA, pelleted, and resuspended in the same solution to an A₆₀₀ of 10. An equal volume of loading buffer (8 M urea, 50 mM Tris [pH 6.8], 2% SDS, a trace of bromophenol blue) was added, and samples were fractionated by SDS-12.5% PAGE prior to immunoblotting with the anti-CcmA serum and a SuperSignal system (Pierce).

Isolation of membrane and soluble cellular fractions. Cells were resuspended in 10 mM sodium phosphate buffer [pH 7.2]-1 mM EDTA-1 mM Pefablock (Boehringer) and lysed in a French pressure cell at 800 lb/in², and debris was removed. The supernatant was centrifuged at 180,000 × g for 1 h. The pellet was suspended in either 10 mM sodium phosphate buffer (pH 7.2) or 0.1 M Na₂CO₃ (pH 11.5) and incubated for 15 min on ice. Supernatant (soluble) and pellet (membrane) fractions were collected after 1 h of centrifugation at 180,000 × g.

mRNA hybridization. As previously described (11), RNA (10 µg) isolated with hot phenol was fractionated by agarose gel electrophoresis and transferred to nitrocellulose filters (Hybond C; Amersham). An EcoRV-PstI fragment spanning most of ccmA was used as a probe.

Scanning and transmission electron microscopy. Cells were resuspended in 0.9% NaCl containing 2% glutaraldehyde fixative, rinsed in piperazine-N,N'-bis(2-ethanesulfonic acid) (PIPES) buffer (pH 7.4)-2 mM CaCl₂, fixed in 1% osmium tetroxide, and stained with 2% uranyl acetate. They were dehydrated in ethanol, dried in a Polaron E5000, sputter coated with 2-nm gold particles, and viewed in a Philips XL30 FEG-SEM microscope. For transmission electron microscopy, fixed cells were embedded in Araldite (Taab Laboratories), sectioned at 50 nm, stained twice with uranyl acetate, and viewed in a Philips CM100 microscope. For immunogold labelling, a 0.2-ml aliquot of concentrated cells was placed on a 100-mesh, Formvar-coated gold grid, quench frozen, transferred to 0.1% uranyl acetate in pure methanol under liquid nitrogen, placed in the cold chamber of a Leica AFS freeze substitution unit, and warmed to 90°C. After 24 h, the grids were warmed to 70°C for 24 h and finally to 50°C and then were infiltrated with Lowicryl HM20 that had been UV polymerized. Cells were sectioned at 50 nm in phosphate-buffered saline containing 10% fetal calf serum. Bound anti-CcmA antibody was visualized with a goat anti-rabbit immunoglobulin-10-nm colloidal gold particle conjugate (British Biocell) diluted 1:100 in phosphate-buffered saline-fetal calf serum. Sections were stained with uranyl acetate-lead citrate.

Nucleotide sequence accession number. The sequence for the ccm locus has been deposited in the EMBL database under accession no. AJ000084.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

A motile, differentiating, nonswarming mutant that has curved cells. Following mutagenesis of P. mirabilis U6450 with mini-Tn5Cm, mutant MNS185 was identified as nonswarming among the Cm^r colonies on 1.5% LB agar (Fig. 1, panel i). This mutant nevertheless retained virtually wild-type individual cell motility (Fig. 1, panel ii). We have previously shown that cell populations can be induced to undergo synchronous differentiation in the absence of migration throughout a normal 4- to 5-h differentiation cycle by seeding short, vegetative cells at high cell density onto the entire surface of multiple 1.5% LB agar plates (16). With this assay, it was shown that levels of cell surface flagellin (FliC) and HpmA hemolysin, which is coinduced with flagellin during swarm cell differentiation (4), were comparable in wild-type and MNS185 cells (Fig. 1, panels iii and iv). Consistently more background cell proteins were evident, however, in the supernatants of MNS185 cells (Fig. 1, panel iii), suggesting increased cell fragility. Differentiated MNS185 cells were still elongated, to at least half the length of the parent cells (Fig. 2, panels i and ii). Nonswarming MNS185 was therefore not substantially impaired in its capacity to differentiate into swarm cells.

View larger version (34K): [in this window] [in a new window]   FIG. 1.   Phenotypic comparison of the P. mirabilis wild type (wt) and mutant MNS185. (i) Swarming following central inoculation of 1.5% LB agar plates with stationary-phase broth cultures and 10 h of incubation. (ii) Motility after 8 h on 0.3% LB agar. (iii and iv) Surface flagellin and hemolysin activities (mean ± standard error [10%]), respectively, of differentiating cells following seeding and growth on 1.5% LB agar. Size markers were ovalbumin (50 kDa) and carbonic anhydrase (34 kDa).

View larger version (70K): [in this window] [in a new window]   FIG. 2.   Differentiated cells harvested from a seeded LB agar plate after 4 h of incubation. (i and ii) Cells of the wild type and MNS185, respectively, viewed by phase-contrast light microscopy. Magnification, ×1,000. (iii) Separate MNS185 cells viewed by scanning electron microscopy.

View larger version (155K): [in this window] [in a new window]   FIG. 3.   Transmission electron micrograph of sections of wild-type (wt) or MNS185 cells harvested from a seeded LB agar plate after 4 h.

The MNS185 mutant locus and gene products. The mutant locus was mapped to a unique chromosomal fragment by Southern hybridization with a transposon restriction fragment as a probe. It was cloned into pBluescript SK as a BglII chromosomal fragment by use of Tn5-encoded chloramphenicol resistance for selection. A fragment of this cloned region was then used to probe a  Dash II phage library of partial Sau3A fragments of the wild-type P. mirabilis chromosome (16). Several phage hybridized and were used to assemble a restriction map of the locus (Fig. 4, panel i), which was subcloned as a 4.9-kb ClaI fragment into pBluescript SK. A total of 1.722 kb was sequenced in both directions (Fig. 4, panel ii). The transposon was inserted 32 codons from the 3' end of a gene that we called ccmA (curved cell morphology), immediately creating a stop codon. The first three ATG codons at which CcmA could initiate are at nucleotides 185, 254, and 359 (Fig. 4, panel ii). None is immediately preceded by a consensus Shine-Dalgarno or promoter sequence. Translation from the first ATG would produce a basic protein of 228 amino acids (25 kDa). The N terminus of this putative protein has no apparent secretion signal or signal peptidase site but is predominantly hydrophobic, with two potential transmembrane segments at residues 10 to 27 and 35 to 52 (Fig. 4, panel ii). The hypothetical Helicobacter pylori protein Hp1542 has 22% identity over CcmA residues 73 to 210, but no other significant similarities were found in any published sequence, including that of the entire E. coli genome. An open reading frame (ORF) that we called pat (putative acetyltransferase), 69 bp downstream of ccmA, encodes a predicted 185-amino-acid protein that is related to acetyltransferases (22). Further downstream, reading in the direction opposite that of pat, is the potential C terminus of a protein with 72% identity to E. coli YgbA, a protein of unknown function (31). No ORFs were found in the extensive AT-rich region upstream of ccmA. An ORF 1.1 kb upstream and reading in the orientation opposite that of ccmA (Fig. 4, panel i) showed identity to the Yersinia pestis virF transcription activator (8).

View larger version (48K): [in this window] [in a new window]   FIG. 4.   (i) ccm locus. ccmA, curved cell morphology; pat, putative acetyltransferase; ygbA, homolog of E. coli ygbA; virF, homolog of Y. pestis virF; Tn, site of transposon insertion; C, ClaI; H, HindIII; EV, EcoRV; S, SacII; N, NsiI; X, XbaI; Sp, SpeI; P, PstI; B, BglII. The box indicates the region of sequence shown in panel ii. (ii) DNA and protein sequences for the ccm locus. Codons are numbered from the first ccmA Met residue, where CcmA1 presumably initiates. The first three Met residues are indicated in bold. Met 59 is the probable N terminus of CcmA2. Tn, transposon insertion. Primers used to amplify ccmA and incorporate EcoRI, BglII, and HindIII sites are indicated by broken arrows. (iii) Hydrophobicity plot (33), showing positions at which a shorter CcmA initiates translation (M59) and Tn196 is inserted. (iv) Hybridization of a ccmA probe to total RNA from wild-type P. mirabilis cells harvested 4 h after seeding. RNA markers are indicated in kilobases.

Two membrane-associated forms of CcmA in wild-type P. mirabilis. CcmA is predicted (33) to be an integral inner membrane protein with a 9-residue cytoplasmic N terminus, with a 176-residue cytoplasmic C terminus, and with only residues 28 to 34 exposed to the periplasm (Fig. 5). C-terminally His-tagged CcmA was expressed in E. coli, and the purified protein was used to raise polyclonal rabbit antisera. N-terminal sequence analysis of the His-tagged product showed that it initiated at the third methionine, Met 59 (Fig. 4, panel ii), generating a predicted protein of 170 amino acids and lacking both of the putative transmembrane domains. Gel mobility corresponded to the predicted size of 19 kDa (Fig. 6, panel i, first lane). Immunoblotting of total proteins from differentiated wild-type Proteus cells with this antiserum revealed two CcmA proteins (Fig. 6, panel i), of 25 kDa (CcmA1) and 19 kDa (CcmA2). CcmA1 has the size predicted for the full-length translation product, while CcmA2, the most abundant product, presumably initiates at Met 59. Two proteins corresponding in size to the predicted C-terminally truncated forms of CcmA1 and CcmA2 were detected at wild-type levels in mutant MNS185 (Fig. 6, panel i).

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Predicted topology for CcmA1, the full-length translation product of ccmA, indicating residues marking the termini of the putative transmembrane segments, Met 59 (the start of CcmA2), and the transposon insertion (triangle).

View larger version (30K): [in this window] [in a new window]   FIG. 6.   (i) ccmA gene products detected by Western blotting following SDS-12% PAGE of total cell proteins from the P. mirabilis wild type (wt) and MNS185 grown for 4 h on seeded plates. CcmA-His is the purified His-tagged product isolated from E. coli. MNS185 and MNS185 l represent shorter and longer exposures of products from MNS185. (ii) Localization of CcmA proteins. The P. mirabilis wild type was harvested from a seeded plate after 4 h, and cells were fractionated as indicated by the flow diagram (see Materials and Methods). c, total cell protein; s, soluble fraction; m, membrane or insoluble fraction; bs and bm, supernatant and pellet, respectively, following washing of the membrane fraction in sodium phosphate buffer (pH 7.2); ws and wp, supernatant and pellet, respectively, following washing of the membrane fraction in 0.1 M Na2CO3 (pH 11.5). Protein markers were soybean trypsin inhibitor (28 kDa) and lysozyme (19 kDa).

View larger version (80K): [in this window] [in a new window]   FIG. 7.   Transmission electron microscopy of P. mirabilis wild-type cells harvested from seeded plates at 4 h. Cells were sectioned and labelled with anti-CcmA antibody-immunogold. Arrowheads indicate gold particles.

CcmA is preferentially expressed in swarm cells. To compare cells before and during differentiation, wild-type vegetative P. mirabilis cells were inoculated at equivalent cell densities (about 10⁷ cells/ml) either into liquid medium or onto solid 1.5% agar medium. Growth kinetics were very similar (Fig. 8, panel i), and total cell protein profiles on Coomassie brilliant blue-stained SDS-polyacrylamide gels were similar (data not shown). However, cells harvested from solid medium at 3 to 4 h, the time of maximal differentiation, were elongated (Fig. 8, panel ii) and produced approximately 20-fold more CcmA (Fig. 8, panel ii). This increase is comparable to the increase seen for flagellin FliC expression (Fig. 8, panel ii), which is diagnostic of swarming differentiation. A coupling of CcmA expression to swarm cell differentiation was confirmed by comparison of wild-type cells with cells of a class II flhA flagellar assembly mutant (16), which differentiate very poorly due to negative feedback upon expression of the flhDC master operon (13). Cells of the flhA mutant from 3-h seeded plates (Fig. 8, panel ii) had barely detectable CcmA, a level comparable to that in the 3-h broth cultures (Fig. 8, panel ii).

View larger version (32K): [in this window] [in a new window]   FIG. 8.   (i) Growth of wild-type cultures in liquid LB broth or on seeded LB agar plates. (ii) CcmA and FliC expression and cell elongation for liquid- and agar-grown cultures. Equal weights of cells, equivalent to 0.1 ml of culture at an A600 of 0.05, were analyzed at each point. (Upper panels) Western blots of CcmA. The panel on the right shows the P. mirabilis flhA mutant isolated after 3 h on seeded plates. Protein markers were soybean trypsin inhibitor (28 kDa) and lysozyme (19 kDa). (Middle panels) Western blots of flagellin (FliC). The protein markers were ovalbumin (50 kDa) and carbonic anhydrase (34 kDa). (Lower panels) Cells (magnification, ~×195) grown for 4 h in liquid or on solid medium.

A ccmA null mutant is only mildly misshapen and is able to swarm. Because C-terminally truncated CcmA proteins were expressed at normal induced levels in differentiating cells of transposon mutant MNS185, it seemed possible that the mutant phenotype was caused by the truncated proteins, rather than by the straightforward loss of the wild-type proteins. To test this possibility, a ccm null mutant was constructed by replacing with an interposon the EcoRV-PstI fragment that comprises 80% of ccmA (Fig. 4, panel i). Unlike mutant MNS185, this ccmA null mutant retained the ability to swarm, albeit less vigorously than the wild type. Migration initiated about 30 min later than in wild-type cells, and while the wild-type 4-h swarm cycle was maintained, migration was slower, resulting in narrower swarm zones and colonies with only half the diameter seen for wild-type colonies. Flagellin expression was indistinguishable from that of the wild type in differentiating cells in the seeded plate assay (data not shown), and cell length appeared at most only marginally reduced (Fig. 9). About 70% of elongated ccmA null mutant cells were nevertheless modestly but distinctly bent, a defect considerably less severe than that seen for mutant MNS185. Only about 5% of wild-type differentiated cells appeared bent.

View larger version (72K): [in this window] [in a new window]   FIG. 9.   P. mirabilis wild-type (wt) and ccmA null mutant (ccm) cells harvested from seeded plates after 4 h of incubation. Magnification, ×320.

ccmA overexpression causes ellipsoidal morphology in Proteus and E. coli. Wild-type P. mirabilis expressing multicopy ccmA had abnormal morphology, particularly in cells isolated from the center of a colony that had swarmed for several hours (Fig. 10, panel i). This level of expression had little effect on E. coli morphology (data not shown). To further assess the effects of ccmA overexpression, pBAD-ccm was constructed; in this construct, ccmA is under the control of the arabinose promoter, which is repressed by glucose and positively regulated by arabinose in both E. coli and Proteus (19). E. coli MC1061 carrying this plasmid was examined at different levels of induction in broth cultures. Normal rod-shaped morphology was maintained in 0.2% glucose (Fig. 10, panel ii), but low-level induction by 0.002% arabinose resulted in cells with a curved morphology reminiscent of that of the Proteus MNS185 transposon mutant (Fig. 10, panel ii). At higher levels of induction, cells became larger and predominantly ellipsoidal or spherical (Fig. 10, panel ii). This pattern was seen 1 h after induction and became more pronounced as induction continued. Enlarged, rounded cells were also evident in E. coli cultures overexpressing CcmA2 from pRVCCM, and comparable effects were observed upon arabinose-induced overexpression of full-length ccmA in P. mirabilis (data not shown).

View larger version (70K): [in this window] [in a new window]   FIG. 10.   (i) Cells of wild-type (wt) P. mirabilis carrying the high-copy-number plasmid vector pBluescript SK (pBSK) or a derivative carrying ccmA (pBSK-ccm). Cells were collected from the center of swarming colonies 6 h after inoculation onto 1.5% LB agar plates, fixed in 2% formaldehyde-0.9% NaCl, and viewed by phase-contrast light microscopy. Magnification, ×1,000. (ii) Effect of ccmA expression on the morphology of E. coli MC1061 carrying plasmid pBAD-ccm. After growth overnight in LB broth supplemented with 0.2% glucose, the culture was diluted 1:100 into fresh LB broth and grown to an A600 of 0.2. The culture was then subdivided, glucose or arabinose was added, and the cultures were grown for a further 5 h. glu, 0.2% glucose; +ara, 0.002% arabinose; ++ara, 0.2% arabinose. Cells were fixed in 2% formaldehyde and viewed at a magnification of ×320.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Proteus swarms on solid growth media by regular cycles of mass migration following differentiation into highly elongated, linear, hyperflagellated swarm cells. Flagellar gene transcription is upregulated ca. 50-fold and septation is almost completely suppressed in these differentiated cells. Swarming requires the formation of two-dimensional multicellular rafts in which swarm cells are closely juxtaposed and flagellar function is tightly coordinated. Intimate cell-cell contact in rafts may be essential for the coordination of flagellar function and intercellular communication (32), and single cells detached from swarm rafts do not migrate. This aspect of swarming is reminiscent of the close alignment of migrating cells needed for multicellular gliding motility by the fruiting bacterium Myxococcus xanthus (21).

The majority of swarming-defective transposon mutants that have been characterized, both motile and nonmotile, are compromised in the regulation of the flagellar hierarchy and septation. In contrast, cells of mutant MNS185, described here, although incapable of migration, retained normal motility and differentiation, including the induction of flagellin as well as the coregulated hemolysin to normal levels and with normal kinetics, and nearly normal suppression of septation. These features are reminiscent of those of mutant FC18, the only other transposon mutant reported to be specifically impaired in the mechanics of swarming. FC18 displays normal cell morphology and differentiation and forms rafts but migrates slowly due to the loss of an extracellular polysaccharide component of the gel proposed to act as an essential lubricant for the migrating rafts (17). In contrast, although the short vegetative cells of mutant MNS185 appeared morphologically normal, most of the elongated cells were visibly misshapen, usually curved or irregularly twisted, and failed to form rafts. We suggest that the failure to form migration rafts and thus to swarm is a consequence of the morphological defect preventing alignment.

We called the mutated gene ccmA, for curved cell morphology. The 1.1-kb AT-rich region upstream of ccmA lacks open reading frames or recognizable promoter sequences. The size of the single transcript suggests initiation 150 to 200 bp upstream of the first Met codon in ccmA. Two CcmA proteins were detected. The size of the larger, CcmA1 species (25 kDa) is consistent with initiation at this first Met codon to produce the full-length 228-residue product. The size of the smaller, more abundant CcmA2 species (19 kDa) is identical to that of the product seen following the expression of ccmA in E. coli, which results from initiation at residue 59, producing a 170-residue N-terminally truncated product. CcmA1 is an integral cytoplasmic membrane protein, consistent with the predicted presence of two transmembrane segments near the N terminus of the full-length protein. These are predicted to be oriented so that the N terminus and the large C-terminal domain of CcmA1 lie on the cytoplasmic face. CcmA2 initiates 7 residues downstream of the second hydrophobic segment, so it includes essentially all of this C-terminal domain. It behaves as a peripheral membrane protein.

The synthesis of both CcmA proteins is induced approximately 20-fold during swarm cell differentiation, with the same induction kinetics as flagellin and HmpA, although CcmA2 accumulates more rapidly than CcmA1. The MNS185 transposon insertion causes a C-terminal truncation of both CcmA proteins by 32 codons, and these truncated proteins accumulate to normal levels in differentiating MNS185 mutant cells. A ccmA null mutant grows normally, showing that the ccmA gene is not necessary for cell viability. Unexpectedly, the morphological and swarming defects of this null mutant were considerably less severe than those of mutant MNS185. The expression of the C-terminally truncated CcmA proteins, therefore, is more detrimental to morphogenesis than is the complete absence of the normal CcmA proteins.

CcmA appears to play a role in morphogenesis that is principally restricted to swarm cells, as indicated by the regulation of its expression, a function that is disrupted by expression of the truncated CcmA proteins. The normal CcmA proteins, while not essential for swarming, clearly enhance its efficiency. The modest bending seen in the null mutant suggests that they do this by maintaining linearity in highly elongated swarm cells. Overexpression of the wild-type proteins, by introducing multiple copies of the wild-type ccmA gene, only partially suppressed the aberrant MNS185 morphological and swarming phenotypes. This result might have been due to competition between the normal and the truncated proteins for the morphogenetic machinery determining cell shape. Since the overexpression of CcmA2 alone was sufficient for complementation, this species either can function with C-terminally truncated CcmA1 or is sufficient on its own. The ccmA gene may have evolved so that it lacks strong translation initiation signals, resulting in normal expression of both proteins. This unusual aspect suggests that both species may be required for normal function, perhaps, for example, in a polymer anchored to the membrane by the CcmA1 species.

The mechanism by which the CcmA membrane proteins influence morphogenesis is unknown. High-level expression of the wild-type ccmA gene in nonswarm cells caused abnormal morphology in both Proteus and E. coli, perhaps indicating the conservation of interacting cell components. Bacterial cell morphology is primarily determined by the pattern of peptidoglycan synthesis and remodeling during growth (27), and CcmA might have a role in organizing peptidoglycan assembly in elongated swarm cells. It has been suggested that swarm cells differ in envelope composition from vegetative cells (5), but our preliminary attempts at identifying changes in the mutant cell wall were not fruitful. The penicillin binding protein (PBP) PAGE profile of E. coli MC1061 overexpressing the ccmA gene was indistinguishable from that of the wild type (data not shown), and we saw no alteration in susceptibility to PBP 2-specific antibiotics, such as mecillinam, or to various detergents or osmotic shock in these cells. Neither was any change in autolysin activity apparent (7a).

Genes affecting bacterial cell shape are typically essential for viability (9), but bloated and twisted cells have been reported for mbl mutants of Bacillus subtilis. Mbl is 50% homologous to MreB, a regulator of cell shape and septation-specific PBP function (1, 34). Curved variants have been described following overexpression of C-terminally truncated FtsA in E. coli (14) and of FtsZ in Rhizobium (25). FtsA and FtsZ cooperate in guiding centripetal septal cell wall growth (24, 27, 28), and cells overexpressing truncated FtsA have aberrant cytoskeletal elements. Although the use of a labelled antibody showed that the truncated form of CcmA in mutant MNS185 is located at the cytoplasmic membrane, no such cytoskeletal elements were visualized. A morphogenetic function specific for highly elongated cells remains consistent with the substantial increase in CcmA1 and CcmA2 abundance seen for swarm cells and is consistent with the absence of homologous proteins in related gram-negative bacteria that do not readily swarm.