An essential, multicomponent signal transduction pathway required for cell cycle regulation in Caulobacter

Abstract

Cell differentiation and division in Caulobacter crescentus are regulated by a signal transduction pathway mediated by the histidine kinase DivJ and the essential response regulator DivK. Here we report genetic and biochemical evidence that the DivJ and DivK proteins function to control the activity of CtrA, a response regulator required for multiple cell cycle events, including flagellum biosynthesis, DNA replication, and cell division. Temperature-sensitive sokA (suppressor of divK) alleles were isolated as extragenic suppressors of a cold-sensitive divK mutation and mapped to the C terminus of the CtrA protein. The sokA alleles also suppress the lethal phenotype of a divK gene disruption and the cold-sensitive cell division phenotype of divJ mutants. The relationship between these signal transduction components and their target was further defined by demonstrating that the purified DivJ kinase phosphorylates CtrA, as well as DivK. Our studies also showed that phospho-CtrA activates transcription in vitro from the class II flagellar genes and that their promoters are recognized by the principal C. crescentus sigma factor σ⁷³. We propose that an essential signal transduction pathway mediated by DivJ, DivK, and CtrA coordinates cell cycle and developmental events in C. crescentus by regulating the level of CtrA phosphorylation and transcription from σ⁷³-dependent class II gene promoters. Our results suggest that an unidentified phosphotransfer protein or kinase (X) is responsible for phosphoryl group transfer to CtrA in the proposed DivJ ⇒ DivK ⇒ X ⇒ CtrA phosphorelay pathway.

Bacterial two-component signal transduction systems control a wide array of physiological processes in response to a variety of environmental conditions. These systems typically contain a sensor kinase, which is autophosphorylated on a conserved histidine residue, and a cognate response regulator containing a conserved aspartate residue to which the phosphoryl group is transferred (1, 2). This same protein family functions in multistep phosphorelay pathways (3). Recent results have provided evidence that sensor kinases and response regulators also play essential roles in the coordination of cell cycle and developmental events in the aquatic bacterium Caulobacter crescentus (4). The histidine kinase DivJ (5) and response regulator DivK (6) have been implicated in a signal transduction pathway required for cell division initiation. Another response regulator, CtrA, has been identified as a transcription factor responsible for the expression of multiple cell cycle-regulated genes (7). Our results now indicate that DivJ and DivK proteins function as part of a multicomponent signal transduction pathway to control the transcriptional activity of CtrA during the cell cycle.

Members of two-component signal transduction pathways regulating cell cycle events in C. crescentus were originally identified in a pseudoreversion analysis of pleC, a pleiotropic developmental gene required for motility and polar morphogenesis (8). Several of the pleC suppressors displayed conditional cell division defects (9) and were mapped to divJ (5), which, like pleC (10), encodes a histidine protein kinase, and to divK (6), which encodes the first essential response regulator identified in bacteria. DivK is a 125-residue polypeptide that belongs to the subfamily of single-domain response regulators that includes the chemotactic protein CheY of Salmonella typhimurium (11) and the sporulation protein SpoOF of Bacillus subtilis (12). Although DivJ and DivK appear to play central roles in regulating initiation of cell division in C. crescentus (6), the PleC kinase may be more directly involved in regulating motility, chemotaxis, and stalk formation (13). Isolation of pleC suppressors mapping to divJ and divK is consistent with a tight interconnection of cell cycle and developmental regulation in C. crescentus, as indicated originally by an analysis of developmental defects in conditional cell division cycle mutants (14, 15).

CtrA is also an essential response regulator required for cell cycle regulation in C. crescentus (7). It functions in vivo as a global regulator controlling transcription of class II flagellar genes, as well as the DNA methylase gene ccrM (16) and hemE, whose expression is associated with cell type-specific initiation of chromosome replication in the stalked cell (17). Promoters of these cell cycle-regulated genes contain a conserved CtrA-binding site (7), or “CtrA-Box”, and an unusual consensus sequence originally thought to be recognized by a novel sigma factor (18, 19).

Here we present genetic and biochemical evidence that the DivJ and DivK proteins function to control CtrA activity in the transcription of cell cycle-regulated genes. Moreover, our in vitro transcriptional analysis of class II flagellar genes indicates that CtrA-regulated promoters are recognized by the principal C. crescentus sigma factor, σ⁷³. These results strongly support a model in which DivJ, DivK, and CtrA, probably in conjunction with an unidentified transphosphorylase or kinase, function in an essential signal transduction pathway to regulate the activity of σ⁷³ in the transcription of cell cycle and developmental genes.

MATERIALS AND METHODS

Bacterial Strains, Plasmids, and Media.

C. crescentus strains were derived from wild-type strain CB15 (ATCC19089) and grown in either PYE medium (20) or minimal M2-glucose medium (20) supplemented with antibiotics as indicated. The cold-sensitive (cs) divK341, divJ331, and divJ332 alleles have been described (9). Escherichia coli strain DH5α was used for propagating plasmids and cultured at 37°C in ML medium (6) supplemented with ampicillin (50 μg/ml), tetracycline (15 μg/ml), or kanamycin (50 μg/ml) as required.

Isolation of Suppressors of divK Mutations and a Linked Tn5 Insertion.

Colonies of cs divK mutants grown on PYE plates at the permissive temperature (37°C) were stabbed in 0.35% agar swarm plates (20), and revertants were isolated from flares formed after incubation for 3–4 days at the nonpermissive temperature (24°C). Mot⁺ cells were purified and examined at 24°C and 37°C on swarm plates to identify temperature-sensitive (ts) revertants. Two revertants containing the ts sokA301 (PC3241) and sokA302 (PC3242) mutations, which were isolated from 300 revertants of cs strain PC4743 (divK356), displayed a filamentous phenotype at 37°C. None of the 1,000 revertants of cs strains PC4744 (divK357) and PC4746 (divK358) displayed a ts phenotype.

Tn5 insertion zzz471∷Tn5, which is ca. 15% linked to the sokA301 mutation, was isolated as described previously (21) and displayed the same genetic linkage to sokA302 by transduction. This Tn5 insertion was used to construct the isogenic sokA strains PC3247 (sokA301) and PC3248 (sokA302) in wild-type strain CB15.

Complementation and Cloning of sokA301 and sokA302.

Cosmids from a pLAFR1–7 library containing random inserts of C. crescentus DNA (5) were isolated that complemented the ts sokA301 mutation of strain PC3247 at 37°C. DNA fragments from the cosmid were subcloned in plasmid pRK2L1 (22) or pRK2L10, a derivative of pRK290 (23) that contains the polylinker sites from lambda phage tg131 (Amersham). DNA fragments containing the wild-type ctrA gene (7) and the sokA301 and sokA302 alleles of ctrA were cloned by PCR amplification of DNA from strain CB15 and the two mutant strains, respectively. DNA-sequencing reactions were carried out on double-stranded DNA cloned in pBluescript IIKS+ by using the Sequenase 2.0 sequencing kit with 7-deaza-dGTP (United States Biochemical).

Protein Purification.

The C-terminal fragments of histidine kinases DivJ (DivJ′) and PleC (PleC′) and the full-length response regulator DivK were purified by using published procedures (6). Purification of the principal C. crescentus sigma factor, σ⁷³, and the RNA polymerase (RNAP) core has also been described (24). Native CtrA protein was overproduced and purified from isopropyl β-d-thiogalactopyranoside (IPTG)-induced E. coli strain BL21 (DE3) carrying the ctrA gene fused to the initiation codon of the T7 gene 10 in the T7–7 vector (25). Inclusion bodies formed by the CtrA protein were solubilized with 6 M guanidine-HCl. The protein was greater than 95% pure, as judged by Coomassie blue staining of SDS/PAGE gels after renaturation and purification by chromatography on DEAE-cellulose and heparin-agarose columns.

Phosphorylation Assays.

In vitro phosphorylation assays have been described (6). Histidine kinases (1 μM) were incubated at 25°C for 15 min in phosphorylation buffer containing 5 μCi [γ-³²P]ATP followed by the addition of the response regulator(s) (10 μM). The phosphotransfer reaction was carried out at 25°C for 20 min. ³²P-labeled proteins were assayed by electrophoresis on 12.5% SDS/PAGE gels and visualized by autoradiography.

In Vitro Transcription Assays.

The formation of open complexes at promoters was measured in run-off transcription assays (26) by using as DNA templates either the 290-bp BamHI-HindIII fragment from plasmid pJW001 containing the fliF promoter (27) or a 315-bp BamHI-HindIII PCR fragment from plasmid pJW002 containing the fliL promoter (28). RNAP holoenzyme was reconstituted by addition of purified σ⁷³ protein to purified C. crescentus RNAP core and incubation at 4°C for 10 min (24). CtrA (10 μM), FlbD (10 μM), and DivJ (1 μM) proteins were added as indicated. RNA products were analyzed by 7 M urea/PAGE with end-labeled Sau3A fragments of pUC18 as size markers.

The accession number AF021339 has been assigned for the partial ctrA sequence described here.

RESULTS

Isolation of Suppressors of divK Cell Division Mutations.

To identify downstream components in the signal transduction pathway mediated by the response regulator DivK we carried out a pseudoreversion analysis of conditional divK strains PC4743, PC4744, and PC4746 (Materials and Methods). Because these cell division mutants have a severe cs phenotype and exhibit extensive filamentation at the nonpermissive temperature (9), they form tight swarms on 0.35% agar plates at 24°C, as shown for the divK356 mutant (Fig. 1A). We isolated a total of 1,300 independent revertants from flares on swarm plates at 24°C that were Mot⁺, Div⁺. The revertants were then screened for Mot⁻, Div⁻ defects at 37°C. Two such ts mutants, sokA301 strain PC3241 (Fig. 1B) and sokA302 strain PC3242 (data not shown), were isolated.

Figure 1.

Phenotype of sokA suppressor strains. (A–E) Single colonies were stabbed in swarm plates and incubated at 24°C (Left) or 37°C (Right) for ca. 40 hr. (A) PC4743 (divK356); (B) PC3241 (divK356, sokA301); (C) CB15 (wild type); (D) PC3247 (sokA301, zzz-471∷Tn5); (E) PC3248 (sokA302, zzz-471∷Tn5). (F) Electron micrographs of sokA301 mutant cells grown in PYE medium at 24°C and shifted to the nonpermissive temperature of 37°C for 12 hr. Cells were stained with 0.5% uranylacetate and were examined by transmission electron microscopy (JEOL JEM-100C). Bar in E represents 1 μ.

To map the sokA301 mutation, we isolated the linked Tn5 insertion zzz471∷Tn5 and showed that it is also linked by transduction to the sokA302 mutation. The Tn5 insertion is unlinked to the divK locus, however, indicating that the sokA alleles are outside suppressors. We confirmed this conclusion by showing that the sokA301 and sokA302 mutations conferred a ts swarm defect (Fig. 1 D and E) in the wild-type, strain CB15 background and that they also suppressed the cs cell division phenotype of the well characterized divK341 allele (ref. 6; data not shown).

Electron micrographs of the sokA301, divK⁺ strain PC3247 showed that it forms long, curled filaments ca. 12 hr after shifting from 24°C to 37°C (Fig. 1F) and arrests growth upon longer incubation at 37°C. Cultures of the sokA301 mutant grown at the permissive temperatures of 24°C, by contrast, contained mostly single cells and short filaments of two to three cell lengths. The cells were flagellated, highly motile, and, like supermotile pleD mutants (29), failed to form stalks at 24°C (Fig. 1F). The isogenic sokA302 strain PC3248 displayed similar phenotypes at 24°C and 37°C (data not shown), but it reverted frequently at 37°C.

sokA Mutations Map to ctrA.

We isolated cosmid clones that complement the ts cell division defect of the sokA301 allele at 37°C (Materials and Methods). Fragments of the 20-kb DNA insert from cosmid pNOR506 were subcloned to generate plasmid pNOR549 containing a 9-kb SstIa-SstIb fragment and plasmid pNOR546 containing an overlapping 8.5-kb EcoRI-SstIb fragment (Fig. 2). Because only plasmid pNOR549 rescued the sokA301 mutation, we surmised that the 482-bp SstIa-EcoRI fragment was responsible for the observed marker rescue in the rec⁺ background (Fig. 2). The 5′ portion of this DNA fragment was found to encode a partial ORF of 73 residues, which is 100% identical to the corresponding C-terminal amino acid sequence of CtrA (7). We obtained direct evidence that the suppressor mutation maps to ctrA by complementing the ts growth and division phenotypes of the sokA301 allele in a rec⁻ strain with a 1,154-bp HindIII-NcoI fragment cloned in plasmid pNOR552, which contains only the ctrA coding and 5′ regulatory sequences (Fig. 2).

Figure 2.

Cloning and mapping of the sokA301 allele to ctrA. pNOR546 and pNOR549 contain subcloned fragments from complementing cosmid clone pNOR506. Plasmid pNOR552 contains the 450-bp 5′ regulatory sequence and the coding sequence of ctrA (7). Complementation (Compl.) was assayed at 37°C in the rec⁻ strain by the formation of uniform large swarms in soft agar (+) and a Div⁺ phenotype by light microscopy. Recombination (Recomb.) was determined by the appearance of flares around the tight swarm at 37°C in the rec⁺ strain (+), but not in the rec⁻ strain (−).

To locate the suppressor mutation, a 700-bp fragment containing the entire ctrA coding sequence was cloned from suppressor strains PC3247 and PC3248 by PCR amplification. DNA sequence analysis of these clones demonstrated that both sokA mutant genes contain the identical A-to-G change at nucleotide 637 of the ctrA coding sequence, which corresponds to a T213A change in the CtrA protein (7). The ctrA sequence from the sokA302 mutant strain also contains a G-to-T change at nucleotide 107. This second mutation may contribute to the unstable phenotype of the allele at 37°C, but the sokA302 mutation has not been examined further.

sokA301 Suppresses the Lethality of a divK Disruption and Cell Division Defects of divJ Mutations.

The divK gene is essential and can be disrupted only in cells that carry two copies of the divK⁺ allele (6). If the essential function of DivK results solely from its regulation of CtrA, it should be possible to disrupt divK in strains containing a sokA mutation. We confirmed this prediction by transducing the sokA301 strain PC3247 with a phage φCr30 lysate prepared on the divK∷aacC1/divK⁺ merodiploid strain PC4781 and recovering drug-resistant recombinants carrying the Gm^R marker of the divK∷aacC1 disruption. Gm^R recombinants were also recovered in crosses with the divK⁺/divK⁺ strain PC1123, which carries a second copy of divK on a plasmid. No drug-resistant recombinants with the divK∷aacC1 disruption were recovered, however, in wild-type strain CB15 that carried only the vector plasmid pRK2L1 (data not shown).

If the sensor kinase DivJ is involved in the regulation of DivK, as indicated by previous genetic and biochemical studies (5, 6), we expected that divK suppressors mapping to ctrA should also suppress the cell division defects of divJ mutations. The sokA301 allele was transduced into divJ332 strain PC4212 (Fig. 3 E and F) and, as a positive control, into the divK341 strain PC4160 (Fig. 3 A and B). The sokA301 mutation suppressed the cell division defect of the divJ strain (Fig. 3G) as well as the divK strain at 24°C (Fig. 3C). The recombinant strains displayed extensive filamentation at 37°C, however, indicating the presence of the sokA301 allele (Fig. 3 D and H). In similar experiments, the sokA301 allele also suppressed the cell division defect of the divJ331 allele (data not shown). These results indicate that CtrA activity is regulated by both DivJ and DivK.

Figure 3.

Suppression of cs divK and divJ alleles by sokA301. Overnight cultures were grown in PYE medium, diluted 25-fold in duplicate, and incubated for ca. 12 hr at either 24°C (Left) or 37°C (Right). Culture samples were applied to a slide coated with 0.1% polylysine and examined by phase-contrast microscope at 630-fold magnification. (A and B) PC4160 (divK341); (C and D) PC3251 (divK341, sokA301, zzz-471∷Tn5); (E and F) PC4212 (divJ332, aux∷Tn5-132); and (G and H) (divJ332, sokA301, aux∷Tn5-132, zzz-471∷Tn5).

The DivJ Kinase Phosphorylates CtrA in Vitro.

We examined the ability of the C-terminal kinase domains, DivJ′ and PleC′, to phosphorylate the CtrA protein in vitro by using purified components. DivJ′ and PleC′ were autophosphorylated in the presence of [γ-³²P] ATP (Fig. 4, lanes 1 and 5), and the phosphate was efficiently transferred to the response regulator DivK (Fig. 4, lanes 2, 3, 6, and 7), as shown previously (6). More importantly for the present studies, DivJ′, but not PleC′, phosphorylated the purified CtrA protein under the same assay conditions (cf. Fig. 4, lanes 4 and 8). These results are consistent with the genetic results described in the previous section and support a model in which DivJ, DivK, and CtrA mediate a common signal transduction pathway, as considered below (see Fig. 7).

Figure 4.

Phosphorylation of response regulators DivK and CtrA by histidine kinases DivJ and PleC. All experiments were carried out with C-terminal fragments DivJ′ and PleC′ and the full-length DivK and CtrA proteins as described in Materials and Methods. The presence of each protein, including DivJ′ (lanes 1–4) and PleC′ (lanes 5–8), is indicated by +. Results of autophosphorylation and phosphotransfer were visualized by autoradiography.

Figure 7.

Model for an essential, multicomponent signal transduction pathway in the regulation of cell division and development in C. crescentus. The C. crescentus cell cycle is diagrammed at the top of the figure. As drawn, the signal transduction pathway is initiated by unidentified cues provided during cell cycle progression to regulate the activity of DivJ and possibly other kinases. Cell cycle cues may also modulate the pathway by controlling the activity of phosphoprotein phosphatase. DivJ and DivK are visualized as mediating phosphotransfer to CtrA via an unidentified component (X). Phospho-CtrA regulates transcription from class II flagellar gene promoters and presumably from other cell cycle-regulated genes with a similar σ⁷³ promoter architecture, including those involved in DNA replication (see hemE and ccrM in Fig. 5) and possibly cell division.

We also noted that DivJ′ was more efficient in the phosphorylation of DivK (Fig. 4, lane 2) than CtrA (Fig. 4, lane 4). Consistent with this result is the observation that in a reaction containing DivK and CtrA, DivJ′ selectively phosphorylated DivK (Fig. 4, lane 3). The failure to detect phospho-CtrA in this last reaction strongly suggests that DivK cannot phosphorylate CtrA directly (see Discussion).

Phospho-CtrA Regulates in Vitro Transcription from Class II Flagellar Gene Promoters by Eσ⁷³.

Several genes regulated by CtrA in vivo, including the class II flagellar genes, contain promoters with a conserved direct repeat recognized by the CtrA protein (ref. 7; Fig. 5). The unusual −35, −10 architecture of class II promoters has suggested that they are recognized by a novel sigma factor (18, 19). To characterize these promoters we examined transcription from class II flagellar genes by using RNAP holoenzymes reconstituted from purified C. crescentus components (24). In these experiments, we assayed for run-off transcripts from DNA fragments containing either the fliL or fliF promoter (Materials and Methods). Unexpectedly, the fliL promoter was recognized by Eσ⁷³ RNAP, which contains the principal σ-factor, and produced a run-off transcript of 88 nt (Fig. 6A, lane 2), the size predicted from the in vivo transcription initiation start site (28). Holoenzymes containing purified σ³² (26) or σ⁵⁴ (30) did not recognize the fliL promoter, however (data not shown). Importantly, transcription from the fliL promoter was stimulated by the addition of CtrA in the presence of DivJ and ATP (Fig. 6A, lane 3), conditions shown above to generate phospho-CtrA (Fig. 4). Unphosphorylated CtrA, by contrast, inhibited transcription from the fliL promoter (Fig. 6A, lane 4). Control experiments demonstrated that DivJ in the absence of CtrA did not affect the rate of transcription (data not shown). These results indicate that the fliL promoter is recognized by the Eσ⁷³ holoenzyme and regulated in vitro by CtrA.

Figure 5.

Alignment of CtrA-Boxes and σ⁷³-dependent class II promoters. The −10, −35 regions, based on the σ⁷³-dependnet consensus (40), and the highly conserved sequences within the spacer region are boxed. Nucleotides that are identical to the consensus sequences are underlined. Transcription start sites are shown in boldface. The two direct repeats of the conserved CtrA binding sequence (7) lie within conserved −35 and spacer regions, respectively, and are indicated by arrows at the top of the figure. Promoter sequences shown correspond to fliF (27), fliL (42), fliQ (43), rpoN (30), hemE (44), ccrM (45), and divK (A. K. Benson and A.N., unpublished data). N, any nucleotides; S, G or C; W, A or T.

Figure 6.

Phospho-CtrA regulates transcription initiation from class II fliL and fliF flagellar promoters by RNAP containing σ⁷³. (A) In vitro run-off transcription assays from the fliL promoter (42) by using purified C. crescentus RNAP core and sigma factor σ⁷³ (24). Reaction contained components as indicated by + (present) or − (absent). Final concentrations of proteins were 10 μM for CtrA and 1 μM for DivJ′. (B) Transcription initiation from the wild-type fliF promoter and a mutant promoter (see Results) was determined by the core RNAP alone or the reconstituted Eσ⁷³. CtrA, FlbD, and DivJ were added to the reactions as indicated by +. The transcripts were run on 7 M urea/PAGE gels with the size markers (M) and visualized by autoradiography.

More conclusive results were obtained from a transcription analysis of the fliF promoter, whose activation is required for synthesis of early flagellar components and two regulatory proteins required for transcription of class III and IV flagellar genes (reviewed in ref. 31). The fliF promoter was not recognized either by the RNAP core alone (Fig. 6B, lane 1), the reconstituted Eσ⁷³ holoenzyme (Fig. 6B, lane 2), or Eσ⁷³ with purified CtrA (Fig. 6B, lane 3). The fliF template was recognized by Eσ⁷³ holoenzyme only under conditions permitting phosphorylation of the CtrA protein: it was transcribed in reaction mixtures containing DivJ and CtrA (Fig. 6B, lane 4), but not DivJ alone (data not shown). The 76-nt run-off transcript produced was exactly the size predicted from the previously determined in vivo transcription start site (27).

We confirmed the specificity of fliF promoter recognition by using a mutant DNA template (pUC18SV4#4) with a deletion of the TA residues at −24, −25 of the fliF promoter. This mutation, which removes two residues within the 3′ TTTAC element of the fliF CtrA-Box (Fig. 5), is known to abolish transcription from fliF promoter in vivo (27). The mutant template was not recognized by the reconstituted Eσ⁷³ holoenzyme either in the absence (Fig. 6B, lane 5) or presence of phospho-CtrA (Fig. 6B, lane 6). We also showed that this promoter is repressed in vitro by the addition of the transcription regulator FlbD (compare lanes 7 and 8, Fig. 6), which binds to the ftr4 sequence within the fliF promoter and negatively autoregulates transcription in vivo (27). Similar results have been reported by using a partially purified RNAP holoenzyme preparation (33). These results demonstrate that the class II fliF promoter is recognized by σ⁷³ and that its activation is regulated in vitro by phosphorylated CtrA and FlbD. Consistent with a role of DivJ and DivK in the regulation of CtrA activity are the results of an in vivo analysis of fliF-lacZ and fliL-lacZ fusions in divJ331 and divK341 mutant strains. Expression of the two transcription fusions was reduced 3- to 4-fold after the cs strains had been shifted from 37°C to the nonpermissive temperature of 24°C (data not shown).

DISCUSSION

Analysis of conditional cell division cycle mutants originally provided evidence for the close coupling of developmental events to cell cycle progression in C. crescentus (14, 15). The work presented here defines an essential signal transduction pathway that can account for the coordinate regulation of cell cycle and developmental events. The sensor protein kinase DivJ (5) and the essential response regulator DivK (6) had been shown to regulate initiation of the cell division cycle. Our genetic and biochemical analyses now indicate that these proteins control the activity of the transcription regulator CtrA (7). We propose a model in which DivJ, DivK, and CtrA, probably in conjunction with an unidentified phosphotransfer protein or kinase, mediate an essential signal transduction pathway that regulates the level of CtrA phosphorylation during the cell cycle. We also provide in vitro evidence that phospho-CtrA controls transcription from σ⁷³ promoters of cell cycle-dependent class II flagellar genes.

Organization of DivJ, DivK, and CtrA in a Common Signal Transduction Pathway.

The two sokA alleles isolated as suppressors of the conditional cell division defect of a divK mutation map to the ctrA gene and contain a T213A change in the CtrA protein. The T residue is highly conserved in the OmpR subfamily of transcriptional regulators, of which CtrA is a member. This conserved T residue in OmpR maps on the three-dimensional structure of the cytoplasmic domain in the β strand (β6, ref. 34, or β5, ref. 35), which is near the α3 DNA recognition helix, and one mutation at this site has been shown to affect OmpR-DNA binding (36). We have not investigated the mechanism of SokA suppression, but the location of the sokA allele suggests that it could affect DNA–protein binding and possibly interaction of the mutant regulator with RNA polymerase. Independent of the mechanism, the isolation of these suppressors provides strong genetic evidence that CtrA functions downstream of DivK and that CtrA is regulated directly or indirectly by DivK. The ability of the sokA301 allele to suppress the lethal phenotype of the divK disruption mutation (see Results) also indicates that divK is essential because of its role as an upstream regulator of CtrA, which in turn controls expression of essential cell cycle genes like ccrM (16).

Results of genetic experiments also support the conclusion that DivK is a downstream target of DivJ. Overexpression of divK confers a severe filamentous phenotype in C. crescentus and causes cell lysis, but this effect can be reversed by increasing divJ kinase gene expression in the same cells (N.O. and A.N., unpublished results). These results are consistent with the phosphorylation of DivK by the DivJ kinase in vivo, as well as in vitro (Fig. 4).

The catalytic domain of PleC also acts as a DivK kinase in vitro, as shown in Fig. 4 and noted previously (6). Although both PleC and DivJ are thought to function through DivK in vivo (6), PleC appears to function primarily in the regulation of cell motility and to play no direct role in cell division. Thus, disruption mutations in pleC produce viable, nonmotile cells that divide normally under most growth conditions (13). How can one response regulator, like DivK, differentially mediate the DivJ and PleC activities? We speculate that these two kinases function at different times in the cell cycle, with DivJ acting early in the initiation of cell division (5) and PleC functioning late in the cell cycle to turn on flagellum rotation (6, 37).

Although DivJ also phosphorylates CtrA (Fig. 4), CtrA is unlikely to be the immediate target of DivJ in this pathway. In addition to the relatively inefficient phosphorylation of CtrA by this histidine kinase, DivK did not catalyze phosphotransfer to CtrA in vitro (Fig. 4). Given the H1 ⇒ D1 ⇒ H2 ⇒ D2 architecture of multicomponent signal transduction pathways (3), where H2 is typically the histidine residue of a phosphotransferase, phospho-DivK (D1) would not be expected to phosphorylate CtrA (D2) directly. Thus, it seems probable that an additional, unidentified component or components (X) is responsible for transferring phosphoryl groups to CtrA in a DivJ ⇒ DivK ⇒ X ⇒ CtrA phosphorelay pathway (Fig. 7).

This provisional working model is similar in outline to the phosphorelay (KinA, KinB ⇒ SpoOF ⇒ SpoOB ⇒ SpoOA) established originally by Hoch and coworkers for the regulation of sporulation in Bacillus subtilis (reviewed in ref. 38). As in this B. subtilis phosphorelay, multiple sensor kinases and phosphoprotein phosphatases (38) could also regulate the signal transduction pathway in C. crescentus. Of particular interest in this regard is divL, another gene identified in the pleC pseudoreversion analysis whose function is required for initiation of cell division (9). The translated divL product is now known to contain sequence motifs conserved in the histidine protein kinase superfamily. The DivL protein is also a candidate for the phosphotransfer protein X postulated in Fig. 7, although preliminary results indicate that its presumptive C-terminal catalytic domain does not catalyze phosphate group transfer from phospho-DivK to CtrA (J.W. and A.N., unpublished data). The precise role of DivL in signal transduction will be of great interest given the observation that the conditional cell division defect of divL mutations can be suppressed by the sokA301 allele (N.O. and A.N., unpublished data).

Regulation of σ⁷³-Dependent Class II Promoters by CtrA.

The in vitro transcription assays indicate that the class II promoters of flagellar genes fliF and fliL are recognized by the principal sigma factor σ⁷³ (39) and require phospho-CtrA for activation (Fig. 6). We have confirmed that the phosphorylated form of CtrA activates transcription from the fliF and fliL promoters in experiments by using phosphoramidate. This low-molecular-weight phosphoryl group donor (32) fully substituted for DivJ in activating CtrA in in vitro transcription assays (J. W. and A.N., unpublished results).

DNA sequence analysis suggests that other CtrA-dependent genes may also be transcribed from σ⁷³-dependent promoters. Alignment of these cell cycle-regulated promoters reveals some conservation of the σ⁷³ consensus sequence at −35, but less similarity at −10 (Fig. 5). In addition, the −10 and −35 spacing of the developmentally regulated promoters is 17 to 19 bp, compared with the 10- to 14-bp spacing of σ⁷³-dependent C. crescentus housekeeping promoters (40). Fig. 5 shows that the 5′ TTAAC of the conserved CtrA-Box (TTAAC N6 TTAAC) overlaps the putative −35 element of the class II promoters and that the 3′ TTAAC element lies within spacer region. The CtrA-Box is not found in the σ⁷³-dependent housekeeping promoters, however (40). This unusual class II promoter architecture, including the conserved −35 element and the novel −10, −35 spacing, may account for the regulation of these σ⁷³-dependent promoters by CtrA.

Conclusion.

Our results strongly support a model in which DivJ, DivK, CtrA, and almost certainly additional proteins, are members of an essential, multicomponent signal transduction pathway that integrates cell cycle cues and possibly environmental signals to coordinate cell cycle and developmental events in Caulobacter (Fig. 7). CtrA has recently been shown to undergo cell cycle-regulated phosphorylation (41), and it seems likely that the signal transduction pathway described here plays a central role in this regulation. Challenging questions for the future will be to identify other players in this pathway and the sensory inputs to which it responds during the course of the cell cycle.

ABBREVIATIONS

RNAP

RNA polymerase

E

core RNA polymerase

cs

cold sensitive

ts

temperature sensitive