INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Since bacteria are too small to sense a gradient along their length, they employ a system of temporal sensing. This allows them to detect concentration changes over distances equivalent to several body lengths. Many bacterial species possess a chemosensory system that allows the comparison of the current concentration of chemoeffectors in the environment to that encountered a few seconds previously. Bacterial chemotaxis has been studied most extensively in Escherichia coli and Salmonella enterica serovar Typhimurium (for reviews, see references 2, 8, and 46). In a homogeneous environment E. coli shows a random swimming pattern of "runs" (periods of smooth swimming) and "tumbles" (periods of reorientation caused by flagellum bundle disruption). When the concentration of an attractant is increased, the cells exhibit a greatly reduced frequency of tumbling, increasing the chance that they will swim toward the source of attractant. After a few seconds the cells adapt, reverting to their prestimulus tumbling bias, and are capable of responding to a subsequent change in chemoeffector concentration. Sensory adaptation is essential for temporal sensing and for the accumulation of bacteria in environments that are optimal for growth.

Methyl-accepting chemotaxis proteins (MCPs) are both the site of initial signal transduction and of adaptation in the chemosensory pathway (for reviews, see references 10 and 28). These proteins function as homodimers. They have two transmembrane helices flanking a periplasmic ligand-binding domain and a cytoplasmic signaling domain. The latter domain is highly conserved among the MCPs of eubacterial and archaeal species (27). It contains specific glutamate residues in two regions, K1 and R1, which are methylated during adaptation. Methyl groups are transferred to the specific glutamate residues of the MCPs from S-adenosylmethionine by a constitutively active methyltransferase, CheR (44). These groups are removed by the methylesterase, CheB (54), and released as methanol (18). CheB activity is increased on its phosphorylation by CheA-P (15). CheB has a regulatory N-terminal domain and a C-terminal catalytic, amidase-esterase domain (9). Inhibition of the catalytic domain is reduced upon phosphorylation of the regulatory domain by CheA-P (23). The ligand occupancy of the periplasmic domain of the MCP reflects the current chemoeffector concentration, whereas the methylation state of the cytoplasmic domain is a "record" of the concentration a few seconds previously. Methylated MCPs are more proficient at CheA activation than unmethylated MCPs (5). When a repellent binds or an attractant leaves the ligand-binding domain of the MCP, a signal is transmitted through a cytoplasmic linker protein CheW to CheA (6, 7, 29). CheA autophosphorylates and the phosphoryl group is transferred either to CheY or, at a slower rate, to CheB. CheY-P binds to FliM, effecting a switch in the direction of flagellar motor rotation and hence a tumble. CheB-P demethylates the MCPs, reducing the level of CheA activation by the MCPs. Conversely, when the attractant concentration increases, CheY-P and CheB-P levels decrease. In response to a persistent stimulus, the cells achieve a steady state of MCP methylation. This restores the prestimulus pattern of runs and tumbles, allowing a response to subsequent changes in concentration of chemoeffector.

By modulating the methylation state of the MCPs, CheR and CheB-P enable adaptation of the chemosensory system to a background level of chemoeffector. CheR and CheB have been extensively studied in E. coli (for recent reviews, see references 9 and 17). CheR is predominantly found associated with the extreme C-terminal pentapeptide, NWETF, of the high-abundance receptor proteins Tsr and Tar in a 1:1 molar ratio (53). The Trp at the second position and the Phe at the last position are critical (38). The NWETF motif is not present in the "low-abundance" receptors Tap and Trg. Cells expressing only low-abundance receptors exhibit poor swarming, impaired MCP methylation, and compromised adaptation (11, 51). Fusion of the NWETF motif to the C terminus of Trg greatly enhances methylation and chemotaxis in a strain expressing Trg as the sole receptor (12). This suggests that the binding of CheR to this motif is critical for methylation and thus adaptation.

Sensory adaptation in nonenteric bacteria shows some variations on the E. coli paradigm. For example, in the gram-positive bacterium Bacillus subtilis, CheY-P causes periods of smooth-swimming as opposed to tumbling. In addition, this bacterium shows CheB-dependent methanol release both upon addition and removal of attractant (19, 47). Remethylation of the MCPs appears to be dependent on phosphorylated CheY (20). B. subtilis has two further proteins, CheC and CheD, which are required for normal methylation (33, 34). CheC inhibits CheR activity and hence reduces MCP methylation. CheD is thought to facilitate CheA activation and is required for CheR activity (33). The archaeon Halobacterium salinarum also shows increased turnover of methyl groups in response to both positive and negative chemical and light stimuli (1, 43). Together, these differences suggest that chemotaxis and adaptation are more complex in many other species than they are in enteric bacteria.

The purple, nonsulfur, -subgroup bacterium Rhodobacter sphaeroides provides an alternative model system for the study of chemotaxis. It can grow aerobically, photoheterotrophically, and anaerobically in the dark by using a variety of alternative electron acceptors. It exhibits taxis toward various chemoeffectors, light, and oxygen (4). Transport and partial metabolism are required for some chemosensory responses (16), and the extent of the responses can depend on the growth conditions. Unlike E. coli and B. subtilis, R. sphaeroides changes swimming direction by interrupting flagellar rotation (2, 3). Upon an increase in attractant concentration the frequency of stops decreases (32). R. sphaeroides possesses multiple homologues of the E. coli chemotaxis genes arranged in three operons and at other unlinked loci. che Op₁ contains cheD, cheY₁, cheA₁, cheW₁, cheR₁, and cheY₂, and che Op₂ contains cheY₃, cheA₂, cheW₂, cheW₃, cheR₂, cheB₁, and tlpC (13, 37, 49). Thirteen chemoreceptors, including both membrane-spanning and cytoplasmic or transducer-like proteins (Tlps), have been identified to date. These are differentially expressed according to the environmental condition (14). It is not known whether the products of the che operons operate through independent, linear pathways or whether there is significant cross talk between the components of these operons. It is interesting that, while most of the receptors identified in R. sphaeroides have conserved glutamate residues which could be involved in adaptation, only two (McpA and TlpT) contain a motif with limited homology to the CheR-binding site (50; www.jgi.doe.gov/JGI_microbial/html/rhodobacter/rhodob_homepage.html). R. sphaeroides also possesses a homologue of the B. subtilis adaptation protein, CheD, but no CheC encoding gene has been identified.

The discovery of CheR and CheB homologues in R. sphaeroides and the identification of MCPs with conserved glutamyl residues suggest a role for methylation-dependent chemotaxis. In this study we set out to examine the function of CheR₁, CheR₂, and CheB₁ in R. sphaeroides chemotaxis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. Bacterial strains and plasmids are shown in Tables 1 and 2. R. sphaeroides strains were grown in succinate medium (39) at 30°C, either aerobically or photoheterotrophically as described previously (26). E. coli strains were grown at 37°C in Luria-Bertani (LB) medium. The antibiotics kanamycin, nalidixic acid, and streptomycin were used at 25 µg/ml, and ampicillin was used at 100 µg/ml.

Molecular genetic techniques All cloning steps were carried out by standard methods (35). Sequencing-quality plasmid DNA was extracted by using the Plasmid Midi-Kit (Qiagen) or the WizardPlus kit (Promega), sequenced by the University of Oxford Biochemistry sequencing service, and analyzed with the GCG software package (University of Wisconsin). PCRs were carried out by using Pfu DNA polymerase (Stratagene and Promega) and purified as described previously (26). All primers were synthesized by Genosys Biotechnologies, Inc.

Construction of deletion strains Regions upstream and downstream from the gene of interest were cloned together into pK18mobsacB, as detailed below, and sequenced to ensure that they were in frame and contained no PCR misincorporation errors. These constructs were introduced into the chromosome of R. sphaeroides by allelic exchange as described previously (13, 36).

cheR1 A 0.42-kb region immediately upstream of cheR₁ was cut from pCHE3.1 with EcoRI and PstI and cloned into pUC19 to generate plasmid pACM35. A 0.4-kb fragment immediately downstream of cheR₁ was cut from pCHE3.1 with BglII and PstI, pACM35 was linearized with PstI, the single-stranded ends of both fragments were filled in with T4 DNA polymerase, and the two were ligated together. The resultant plasmid with the upstream and downstream regions in the correct orientation was designated pACM36. The entire insert from this pUC19 derivative was excised with EcoRI and HindIII and ligated into similarly cut pK18mobsacB to generate the final construct pAG1.

cheR2 A 0.54-kb region immediately upstream of cheR₂ was amplified by PCR by using primers that encompassed the start codon and that included 5' EcoRI and 3' BamHI sites. A 0.54-kb region that contained the downstream flanking sequence of cheR₂, including the ribosome-binding site of cheB₁, was amplified by PCR with primers that included 5' BamHI and 3' HindIII sites. The first PCR product was cloned into appropriately cut pK18mobsacB to produce plasmid pHV3. The second PCR product was ligated into pHV3 cut with BamHI and HindIII to generate pHV4.

cheB1 A 1.4-kb region immediately upstream of cheB₁ was cut from pJPA112 with PstI and PvuII. A 1.4-kb region immediately downstream of cheB₁ was cut from pJPA113 with EcoRI, the single-stranded ends filled in with T4 DNA polymerase and then cut again with BamHI. These were cloned together into pUC19 cut with PstI and BamHI to generate pTJC10. The entire insert from this plasmid was excised with PstI and BamHI and ligated into similarly cut pK18mobsacB to generate pTJC11.

R. sphaeroides behavioral assays Swarm plates containing 0.25% agar (BiTek; Difco) and 100 µM attractant were inoculated and incubated as described elsewhere (26). Each experiment was performed in triplicate and repeated three times to generate nine data sets.

Construction of expression plasmids. The coding sequences (excluding the ATG start codon) of cheR₁, cheR₂, and cheB₁ were amplified by PCR with primers incorporating exogenous 5' and 3' restriction sites facilitating in-frame cloning into pQE30 (Qiagen). The fragments were cloned such that expression was under the control of the IPTG (isopropyl--D-thiogalactopyranoside)-inducible tac promoter and that the protein products would have an N-terminal tag of six-histidine residues to facilitate purification. The cheR₁, cheR₂, and cheB₁ fragments had 5' BamHI and 3'HindIII sites and, once cloned, generated expression plasmids pQE30R1, pQE30R2, and pQE30B1, respectively. The cloning of cheA₁ and cheA₂ in pQE30 has been described previously (37). Sequence analysis of the clones revealed no PCR misincorporation errors.

E. coli behavioral assay. E. coli swarming assays were carried out by a modification of the method of Wolfe et al.(52). RP437, RP1254, and RP4972 were each transformed with pREP4 prior to transformation with the cheR₁, cheR₂, or cheB₁ expression plasmids. pREP4, which expresses the LacI^q repressor protein, minimizes "leaky" expression from P_tac. The strains carrying the appropriate expression plasmids were grown overnight in LB broth with ampicillin and kanamycin at 30°C. LB swarm plates containing 0.25% agar, antibiotics as appropriate, and IPTG at 0, 1, 10, 100, or 1,000 µM were inoculated in triplicate with 5 µl of the stationary-phase culture and incubated at 30°C. RP437 swarm plates were photographed after 8 h, and swarm plates of the mutant strains were photographed after 15 h of incubation.

Methanol release experiments A modified version of the methanol release assay developed for R. sphaeroides was used (21). All experiments were carried out under photoheterotrophic conditions. Cells were grown to an optical density at 660 nm (OD₆₆₀) of between 0.8 and 1.2, and a volume adjusted for cell number was harvested by centrifugation. The pellet was washed once in 1 ml of HEPES-Cm buffer (10 mM Na-HEPES [pH 7.2]-50 µg of chloramphenicol ml¹, sparged with nitrogen), resuspended in 1 ml of HEPES-Cm, and incubated for 40 min with illumination at 50 µM m² s¹ with 40 µl of L-[methyl-³H]methionine (0.06 µmol ml¹; specific activity, 185 mCi mmol¹; Amersham). The cells were harvested, washed in HEPES-Cm buffer, and loaded onto a sterile 0.45-µm (pore size) syringe filter (Nalgene). HEPES-Cm buffer was passed through the filter for 30 min to remove the excess radiolabel. The flow rate was adjusted to 1 ml min¹. Each experiment included monitoring of (i) prestimulus behavior (HEPES-Cm buffer for 10 min), (ii) the response to 1 mM sodium propionate in HEPES-Cm (10 min), and (iii) the response to its replacement by HEPES-Cm without attractant (10 min). Next, 0.5-ml samples were collected and subjected to vapor-phase transfer into 7 ml of Optiphase Hi-Safe scintillation fluid (Fisher) for 16 h, and the radioactivity was measured in a Beckman 5000TD scintillation counter. At least three data sets were collected for each experiment.

Purification of CheA₁, CheA₂, and CheB₁ Cells containing the appropriate expression plasmids were grown in 2YT medium with antibiotics as appropriate to an OD₆₀₀ of 0.8 at 37°C. Expression was induced by the addition of 0.1 mM IPTG for 20 h at 18°C. After induction, cells were harvested by centrifugation (6,000 × g, 15 min) and resuspended in lysis buffer (10% [vol/vol] glycerol; 50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 10 mM imidazole; 1 mM dithiothreitol [DTT]) in a volume of 60 ml per liter of the original culture. Cells were lysed by sonication on ice for six 20-s intervals (Vibracell; Sonics and Materials, Inc.). Lysates were cleared by centrifugation (35,000 × g, 15 min) and filtration of the supernatant through a 0.45-µm (pore-size) syringe filter. The filtered supernatant was applied to a Ni-nitrilotriacetic acid agarose column (Qiagen) equilibrated with lysis buffer. The column was washed with 60 column volumes of lysis buffer prior to elution of the protein in lysis buffer supplemented with 250 mM imidazole.

Phosphotransfer assays. All reactions were performed in TGMNKD buffer (50 mM Tris-HCl, pH 8.0; 10% [vol/vol] glycerol; 5 mM MgCl₂; 150 mM NaCl; 50 mM KCl; 1 mM DTT) at 20°C. Then, a 5 µM concentration of either CheA₁ or CheA₂ was preincubated with 0.5 mM [-³²P]ATP (14 GBq mmol¹; Amersham) for 15 min before the addition of CheB₁ in a final reaction volume of 100 µl. A 10-µl sample was taken prior to addition of CheB₁ (T = 0). After the addition of CheB₁, 10-µl samples were taken at the intervals shown and quenched in 5 µl of 3× sodium dodecyl sulfate (SDS)-EDTA loading dye (7.5% [wt/vol] SDS; 90 mM EDTA; 37.5 mM Tris-HCl, pH 6.8; 37.5% [vol/vol] glycerol; 3% [vol/vol] -mercaptoethanol). Samples were heated to 65°C for 30 s prior to SDS-polyacrylamide gel electrophoresis (PAGE) on 15% gels according to the method of Laemmli (22). Gels were dried and exposed to phosphor screens (Kodak), and the radioactive bands were detected by using an SF-PhosphorImager with ImageQuant version 5.0 software (Molecular Dynamics).

Construction of egfp fusions and fluorescence microscopy. A pK18mobsacB derivative, pGW41, that contained mcpG fused in-frame to egfp from pEGFP-N1 (Clontech), together with downstream flanking sequences, was used to generate strain JPA500 in a recent study (48). This strain contained mcpG-egfp in place of the wild-type gene in the chromosome. The pK18mobsacB derivatives for the deletion of cheR₁, cheR₂, and cheB₁ (pAG1, pHV4, and pTJC11, respectively) were introduced into JPA500 to generate deletion mutants in the mcpG-egfp fusion background (JPA569, JPA564, and JPA516, respectively).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Deletion analysis of cheR₁, cheR₂, and cheB₁. The predicted amino acid sequences of CheR₁ (encoded in che Op₁, accession no. X80205) and of CheR₂ and CheB₁ (encoded in che Op₂, accession no. AJ000977) were compared with the equivalent proteins from E. coli and B. subtilis. The CheR₁, CheR₂, and CheB₁ of R. sphaeroides have 43, 37, and 47% identity, respectively, with the corresponding E. coli proteins and were therefore designated homologues of these adaptation enzymes.

View larger version (34K): [in this window] [in a new window]   FIG. 1.   Comparison of the swarm diameters of strains JPA568 (cheR1), JPA565 (cheR2), and JPA517 (cheB1) with WS8N (wild type) under aerobic (a) and photoheterotrophic (b) conditions to 100 µM concentrations of attractants (control is no added attractant). The swarm diameters of strains WS8N (wild type), JPA568 (cheR1), JPA565 (cheR2), and JPA517 (cheB1) are shown from left to right for each attractant. The error bars represent 1 standard error of the mean (SEM) from nine experiments. Strains JPA565 and JPA517 have significantly reduced swarm diameters to most attractants compared to strains WS8N and JPA568 under both growth conditions.

View larger version (28K): [in this window] [in a new window]   FIG. 2.   Behavior of wild-type (26) and mutant strains in the tethered cell assay. The strains and the environmental conditions are indicated above each graph in panels a to h. Addition and removal of 1 mM propionate is shown by arrows. The wild-type (WS8N) and cheR1 (JPA568) strains showed normal responses, whereas the cheR2 (JPA565) and cheB1 (JPA517) strains failed to respond. Each strain was assayed at least three times. Graphs show the average response of a field of view that included at least three cells. Individual cells showed significant variability in their unstimulated rotation rates. However, this result was not strain dependent.

Expression of CheR₁, CheR₂, and CheB₁ of R. sphaeroides in wild-type and mutant strains of E. coli CheR₁, CheR₂, and CheB₁ of R. sphaeroides were expressed as N-terminal His₆ fusions from plasmid pQE30 in E. coli strains RP437 (wild type), RP1254 (cheR), and RP4972 (cheB). In each case, upon induction with IPTG, bands of the correct size were detected by SDS-PAGE (data not shown). The effects of this expression on swarming behavior are shown in Fig. 3. CheR₁ did not restore swarming to RP1254 at any level of induction (data not shown). CheR₂ complemented RP1254 when induced at low levels (up to 10 µM IPTG [Fig. 3a]), restoring the ability of RP1254 to form chemotaxis rings in swarm plates. CheR₂ inhibited swarming of RP437 at IPTG concentrations of 100 µM, but there was a severe inhibition of growth at these levels of induction (data not shown). At lower levels of induction there was no effect on swarming (Fig. 3b). CheB₁ complemented RP4972 only in the absence of IPTG (Fig. 3c). IPTG concentrations of 1, 10, and 100 µM abolished complementation by CheB₁ but did not affect growth rate (data not shown). CheB₁ inhibited swarming of RP437 in the absence of IPTG without affecting the growth rate (Fig. 3b). Thus, there is some expression from the noninduced P_tac promoter even in the presence of pREP4. Complementation of both the E. coli mutant strains required an incubation period of 15 h to achieve swarm sizes comparable to those obtained with the wild-type strain, RP437, after 8 h. However, swarm rings were observed in all cases where partial complementation was observed, indicating restoration of some chemotaxis. Microscopic examination revealed that the presence of R. sphaeroides CheB₁ reduced the high tumbling bias of RP4972 and CheR₂ reduced the smooth bias of RP1254 (data not shown). Together, these data suggest that CheR₂ and CheB₁ from R. sphaeroides function as adaptation enzymes and are able, at least in part, to complement the equivalent E. coli mutations.

View larger version (50K): [in this window] [in a new window]   FIG. 3.   Effect of expression of pQE30, pQE30R2, and pQE30B1 in E. coli strains RP1254 (cheR) (a), RP437 (wild type) (b), and RP4972 (cheB) (c). The plasmids in RP1254 were induced with 10 µM IPTG but were not induced in RP437 and RP4972. The swarm plates were incubated for 15 h at 30°C.

Methanol release during the chemotactic response. To investigate the methylation of R. sphaeroides MCPs, the pattern of methanol release from the wild-type strain, WS8N, was assessed. WS8N cells showed a release of methanol upon the addition of 1 mM propionate but not upon its removal (Fig. 4a). This is different from the pattern of methanol release seen both in E. coli and in B. subtilis. To determine the roles of CheR₁, CheR₂, and CheB₁ in this methanol production, the three deletion strains JPA568, JPA565, and JPA517 were investigated. In cells deleted for cheR_1, methanol was detected upon both the addition and the removal of 1 mM propionate (Fig. 4b). Interestingly, a mutant lacking all of che Op₁ (JPA117) showed a wild-type response to the addition of propionate with methanol release upon addition but not upon its removal (Fig. 4c). At least three data sets were collected for each strain. Thus, the methanol released upon the removal of propionate in the cheR₁ strain must depend on one or more components encoded by che Op₁ other than cheR₁.

View larger version (22K): [in this window] [in a new window]   FIG. 4.   Methanol release from wild-type and mutant strains upon the addition and removal of 1 mM propionate (shown by arrows). (a) WS8N (wild-type) showed release upon the addition of attractant. (b) cheR1 (JPA568) showed release upon attractant removal as well as upon addition. (c) che Op1 showed a wild-type methanol release profile. (d and e) Methanol release was not detected in cheR2 (JPA565) (d) or in cheB1 (JPA517) (e).

Phosphorylation of CheB₁ by CheA₁ and CheA₂. The transfer of phosphoryl groups from purified CheA₁ and CheA₂ to purified CheB₁ in the presence of ATP was measured in vitro. CheA₁ and CheA₂ both autophosphorylated when preincubated with [-³²P]ATP. CheB₁ was added to each of the phosphorylated CheA preparations, and the transfer of label was monitored (Fig. 5). CheA₂-³²P was able to transfer phosphoryl groups to CheB₁, but there was no detectable phosphotransfer from CheA₁-³²P. Control reactions showed that there was no change in the level of label on the CheAs when CheB₁ was not added. Furthermore, CheB₁ was not phosphorylated in reaction mixtures containing labeled ATP but lacking either of the CheAs (data not shown).

View larger version (73K): [in this window] [in a new window]   FIG. 5.   CheA2-dependent phosphorylation of CheB1 in the presence of [-32P]ATP. Phosphorimages after separation of reaction products by SDS-15% PAGE are shown for CheA1 and CheB1 (a) and for CheA2 and CheB1 (b). Both of the CheA preparations contain phosphorylatable proteolytic fragments of CheA, which are similar in size to CheB1 (visible in the T = 0 samples). Control experiments in which CheB1 was omitted showed that these bands did not change in intensity throughout the experiment (data not shown).

Effect of deletions on MCP localization. In E. coli MCPs are localized to the poles of the cell, and this localization is dependent on CheW and CheA but not on CheR and CheB (24, 25, 40). A chemoreceptor of R. sphaeroides, McpG, also forms clusters at a single pole of the cell. Correct localization of this MCP is dependent upon components of che Op₂ (26, 48). We investigated the effect of CheR₁, CheR₂, and CheB₁ on the localization of McpG in R. sphaeroides cells by generating strains containing the mcpG-egfp fusion in mutants deleted of cheR₁ (JPA569), cheR₂ (JPA564), and cheB₁ (JPA516). The pattern of fluorescence of the McpG-GFP fusion strain in a WS8N background (JPA500) was compared to that shown by the fusion in the deletion backgrounds (Table 3). The localization of the fusion protein was scored as "one pole" (as seen in 93.7% of cells with a wild-type background [JPA500])or "aberrant." Aberrant localization included cells showing similar levels of fluorescence at both poles and fluorescence not associated with the cell pole. Wild-type cells that did not contain the mcpG-egfp fusion did not fluoresce. The McpG-GFP fusion localized to one pole in 95.4% of the cheR₁ cells, 94.9% of the cheR₂ cells, and 97.0% of the cheB₁ cells (Table 3). A similar localization profile was observed with photoheterotrophically grown cells (data not shown). Therefore, localization of McpG does not depend upon CheR₁, CheR₂, or CheB₁. This indicates that the adaptation proteins of R. sphaeroides are not essential for normal McpG clustering at a pole.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Several observations have led to the suggestion that MCPs and methylation-dependent adaptation may have a role in R. sphaeroides chemotaxis. In this study we examined the R. sphaeroides adaptation enzyme homologues CheR₁, CheR₂, and CheB₁ by genetic and biochemical approaches.

Deletion of the che Op₁ encoded cheR₁ does not affect the ability of R. sphaeroides to respond to the compounds tested either on swarm plates or in tethered cell assays under aerobic or photoheterotrophic conditions, a finding consistent with the report of Hamblin et al. (13). The inability of CheR₁ to complement a cheR mutant of E. coli indicates that CheR₁ is unable to substitute for the CheR of E. coli. This is of interest since CheR₁ shares greater sequence identity with the E. coli protein than does CheR₂ (43 and 37%, respectively). The methanol release data do, however, suggest that CheR₁ has a role, as yet unidentified, in R. sphaeroides chemotaxis.

The deletion of either cheR₂ or cheB_1, which both lie within che Op₂, resulted in a nonchemotactic phenotype under aerobic and photoheterotrophic conditions, both on swarm plates and in tethered cell assays. Both CheR₂ and CheB₁ can substitute for the corresponding homologues in E. coli. These findings suggest that CheR₂ is the principal methyltransferase under laboratory conditions and that CheB₁ functions as a methylesterase. The formation of swarm rings by the complemented strains demonstrates that CheR₂ and CheB₁ support chemotaxis in E. coli and are not simply restoring the tumble bias. CheR₂ and CheB₁, unlike CheR₁, must therefore allow adaptation of these E. coli MCPs. The CheR of E. coli binds to the conserved NWETF motif on the high-abundance receptors, allowing methylation of both high- and low-abundance receptors (12). This motif has not been identified on any R. sphaeroides chemoreceptors, including the highly expressed McpG, although McpA and TlpT have similar motifs (GWEDF and GFEDF, respectively). However, McpA is expressed at low levels under the conditions tested, and its deletion has no effects on chemosensory behavior (unpublished data), whereas transmembrane prediction analysis shows that TlpT is probably a cytoplasmic protein. The ability of CheR₂ to function in the E. coli chemosensory pathway is interesting, since it suggests that CheR₂ is either able to interact with this motif, even though its native binding motif may be different, or CheR₂ may interact with a different, conserved region of the MCPs.

In E. coli, null mutants of cheR and cheB cause smooth-swimming and tumbly phenotypes, respectively (42, 54). However, these strains retain the ability to respond (although not adapt) to changes in chemoattractant concentration (45, 54). Thus, while the strains appear nonchemotactic on swarm plates, changes in flagellar switching frequency are observed on the addition or the removal of attractant in tethered cell and free-swimming assays. In R. sphaeroides, deletion of either cheR₂ or cheB₁ results in a strain that is nonchemotactic on swarm plates. However, unlike the equivalent E. coli mutants, these strains are incapable of responding to the addition or the removal of attractant in the tethered cell assay. In addition, these strains retain a stopping frequency similar to that of unstimulated wild-type cells. Therefore, while R. sphaeroides contains homologues of the E. coli adaptation enzymes that will complement the equivalent E. coli mutations, the phenotypes observed on their deletion are significantly different. Further investigations are ongoing to determine whether these differences are the result of the multiple homologues of the adaptation enzymes present in R. sphaeroides or reflect a more substantial difference in the adaptation pathway between E. coli and R. sphaeroides.

Modified experimental conditions allowed us to demonstrate for the first time methanol release upon challenge with chemoeffector. Methanol production was observed upon the addition of 1 mM propionate but not upon its removal. This is in contrast to the pattern of methanol release in E. coli, where release is associated with the increased methylesterase activity of CheB-P, resulting in reduced activation of CheA by the MCPs. In R. sphaeroides methanol release upon the addition of propionate was abolished in both a cheB₁ strain and in a cheR₂ strain. Taken together with the complementation data, this suggests that CheR₂ is the principal methyltransferase and that CheB₁ is the principal methylesterase.

Unexpectedly, methanol release was observed upon both the addition and the removal of propionate in the cheR₁ strain. However, despite this difference in the pattern of methanol release, the cheR₁ strain was capable of exhibiting normal chemotaxis both on swarm plates and in the tethered cell assay. This is a similar pattern to that observed in wild-type B. subtilis and H. salinarum cells (19, 30, 47). In B. subtilis the additional proteins CheC and CheD are required for chemoreceptor methylation. orf₁ in che Op₁ (49) has 29% similarity with cheD from B. subtilis. The methanol release profile in the che Op₁ strain was similar to that of wild-type cells. Therefore, the methanol produced upon the removal of propionate in the cheR₁ strain depends on components of che Op_1, possibly the cheD homologue. This would suggest that there is an interaction between the CheD and CheR₁ of R. sphaeroides. In B. subtilis, CheD has been shown to interact with the protein CheC (34), although the mechanism by which CheD regulates chemotaxis in B. subtilis remains unclear. However, despite the presence of CheD, R. sphaeroides has no apparent CheC homologue. Investigations are under way to determine whether any CheR₁-CheD interaction exists and also the role of these proteins in adaptation in R. sphaeroides. It is clear that both CheR₂ and CheB₁ share some functional similarity with the equivalent adaptation enzymes of E. coli, as demonstrated by the complementation and methanol release data. However, adaptation to chemical stimuli in R. sphaeroides is more complex than in E. coli and also shows some similarity with the system of B. subtilis.

CheB₁ has an important role in methylation-dependent chemotaxis. It is essential for chemotaxis in swarm plates and in tethered cell assays. CheB₁ is also required for a normal pattern of methanol production. Complementation data show that it shares functional similarity with its E. coli counterpart and hence may be phosphorylated by E. coli CheA. In vitro phosphotransfer assays demonstrated that in R. sphaeroides CheB₁ is phosphorylated by CheA₂ but not by CheA₁ (although both CheA₁ and CheA₂ can autophosphorylate in vitro). This finding is surprising since it suggests that any signal mediated by CheA₁ cannot be subject to feedback adaptation via CheB₁-P. The intriguing disparity between the ability of the CheAs to phosphorylate CheB₁ is not readily identifiable from a comparison of the sequences of the two histidine protein kinases.

In agreement with a previous study in which it was shown that components of che Op₁were not required for McpG-GFP clustering (48), CheR₁ is not required for correct McpG-GFP localization. Furthermore, neither of the che Op₂-encoded proteins, CheR₂ and CheB₁, was required for the aggregation of McpG in a chemoreceptor complex. Similarly, in E. coli CheR and CheB are not required for receptor clustering (24). The methylation state of the deletion mutants is likely to be altered with respect to the wild-type and therefore, as in E. coli, the methylation state of the chemoreceptor complex does not affect receptor localization. In this regard at least, the CheR₁, CheR₂, and CheB₁ proteins of R. sphaeroides have properties similar to those of their counterparts in E. coli.

We have investigated the roles of CheR₁, CheR₂, and CheB₁ from R. sphaeroides. It is clear that adaptation to chemoeffectors in this bacterium will not conform to the E. coli paradigm. The mechanism of adaptation may share some similarity with the gram-positive organism B. subtilis and the archeon H. salinarum, but there are also crucial differences between the species. The data presented here illustrate selectivity and discrimination within the histidine protein kinase and the response regulator components of the R. sphaeroides che pathways. The completion of the R. sphaeroides genome indicates additional CheR and CheB homologues, and investigations into the role of the multiple proteins in behavior and adaptation are under way to derive an alternative model by which bacteria may adapt to chemotactic stimuli.