Cloning and sequencing.

The Tn5-derived kanamycin resistance marker and IS50L along with adjacent R. meliloti DNA was cloned into plasmid pNEB193 (New England Biolabs, Beverly, Mass.) from SalI-digested genomic DNA of the exoS96::Tn5 mutant and sequenced. The exoS gene carrying the 5.3-kb BglII fragment of pM13 (10) was cloned into pMCS5 (17) and named pHC50. The 5.3-kb region was then subcloned into the same vector in three fragments, and nested deletions from opposite directions were made from the subclones. Both strands of DNA of the exoS gene and part of the downstream region have been sequenced from these nested deletion constructs manually with a sequencing kit from U.S. Biochemicals (Arlington Heights, Ill.). Parts of the region were also sequenced from pHC50 with an automated sequencer from Perkin-Elmer ABI (Branchburg, N.J.).

Protein purifications and analyses.

All the proteins that were used in this study were cloned into the pET16b vector from Novagen (Madison, Wis.) so that they were expressed with an N-terminal 10-histidine tag attached. The cloned genes were sequenced for confirmation. The overexpressed protein was purified over a nickel column with the HisBind buffer kit from Novagen in accordance with the protocol for cell lysis and subsequent chromatography provided by the manufacturer. Protein products were resolved in polyacrylamide gel with the Hoffer minigel system (San Francisco, Calif.). The concentrations of polyacrylamide gel are indicated in the figure legends.

In the case of the incomplete and complete predicted cytoplasmic domains of ExoS, ExoS_cyto67, and ExoS_cyto72, respectively, both proteins had to be solubilized from inclusion bodies in 6 M urea and purified on a nickel column in 6 M urea. Both purified proteins were renatured as described by Jin et al. (20) by being dialyzed, without stirring for 18 h and then with stirring for 18 h, against the renaturing buffer (20 mM HEPES [pH 7.4], 50 mM KCl, 5 mM dithiothreitol, 50% glycerol).

A membrane protein fraction was prepared in accordance with the protocol described by Chervitz et al. (6) with some modification. Cells from 400 ml of saturated culture were collected and resuspended in 5 ml of lysis buffer. Cells were lysed with a microtip powered by a Sonicator from Heat Systems (Farmingdale N.Y.) at power level 3 with 2.5 s on and 2.5 s off. The total on time was 7 min for E. coli cells and 10 min for R. meliloti cells. The suspension was then subjected to centrifugation at 7,600 × g for 20 min to remove intact cells; this was followed by centrifugation at 27,000 × g for 20 min to collect membrane protein embedded in the cytoplasmic membrane.

Antibody generation and immunoblot analysis.

To obtain polyclonal antibodies against ExoS, the purified cytoplasmic domain of ExoS was resolved on a polyacrylamide gel. The gel was stained in 0.05% Coomassie brilliant blue R-250 prepared in water for 10 min at room temperature with shaking. The lightly stained protein band was then cut out. The gel slices containing protein were used by HRP Inc. (Denver, Pa.) to raise antibodies against ExoS in rabbits. Western blot analyses with ExoS-specific antibodies were conducted by use of the Western Light chemiluminescent detection system from Tropix, Inc. (Bedford, Mass.) according to the manufacturer’s instructions.

Autophosphorylation and kinase assay of ExoS_cyto72.

The purified ExoS_cyto72 protein was mixed with [γ-³²P]ATP in the reaction buffer (30 mM HEPES [pH 7.4], 10 mM MgCl₂, 5 mM CaCl₂, 50 mM KCl) along with enough unlabeled ATP to give a final concentration of 0.4 μM. The reaction mixture was incubated at room temperature for 10 min. The volumes of the reaction mixtures were reduced under centrifugation in Microcon-10 units (Amicon, Beverly, Mass.). The concentrated, labeled protein was mixed with or without the purified ChvI protein for an additional 5 min before being mixed with an equal volume of 2× loading buffer to stop the reactions. The samples containing ExoS_cyto72 were boiled before being resolved on gels. The samples containing labeled ChvI protein were not boiled because boiling causes dephosphorylation of phosphorylated ChvI (data not shown). To test the pH stability of phosphorylated ExoS_cyto72, the labeled protein was further incubated in 0.1 N HCl or 0.1 N NaOH for an additional 5 min before the addition of 2× loading buffer.

Autophosphorylation assays of ChvI.

The autophosphorylation of ChvI protein was assayed as previously described (35), with some modifications. [³²P]acetylphosphate was synthesized with 0.3 U of E. coli acetate kinase (Sigma) in a buffer containing 25 mM Tris-HCl (pH 7.4), 60 mM KOAc, 10 mM MgCl₂, and 10 μCi of [γ-³²P]ATP. The reaction was conducted at room temperature for 20 min; acetylphosphate was separated from the enzyme with a Microcon-10 microconcentrator (10,000-molecular-weight cut-off; Amicon), 3 μg of purified His-ChvI protein was added, and the sample was further incubated for 10 min. The phosphorylated ChvI was further incubated with 0.1 N HCl or 0.1 N NaOH for 5 min and neutralized by the addition of 0.1 N NaOH or 0.1 HCl, respectively. All reactions were stopped by the addition of equal volumes of 2× sodium dodecyl sulfate loading buffer.

R. meliloti exoS96 mutation is suppressed by the A. tumefaciens chvG gene.

In order to determine whether R. meliloti ExoS and A. tumefaciens ChvG are functionally related, as suggested by the high level of homology between the two, plasmids carrying either the A. tumefaciens chvG⁺ gene or the chvG gene interrupted by a transposon were introduced into the R. meliloti exoS96 mutant. The levels of succinoglycan production by the resulting strains were then examined on plates. The presence of the A. tumefaciens chvG⁺ gene caused the reduction of succinoglycan production by the R. meliloti exoS96 mutant, while the transposon-interrupted chvG gene did not cause a reduction. This indicates that the A. tumefaciens chvG gene can suppress the elevated, constitutive level of succinoglycan synthesis observed in an exoS96 mutant, suggesting that the R. meliloti ExoS and A. tumefaciens ChvG proteins are functionally interchangeable in R. meliloti cells.

Additional evidence that these two proteins are functionally interchangeable came from the work of Østerås et al. They interrupted the cloned R. meliloti chvG gene, which we have now shown is actually the exoS gene, and introduced it into A. tumefaciens chvG mutants. They found that only the uninterrupted R. meliloti exoS gene can suppress the mutant phenotype of the A. tumefaciens chvG mutant, which is sensitive to complex growth media (31). They have also shown that the R. meliloti chvI gene can suppress the A. tumefaciens chvI mutant. These results suggest that the heterologous pairs of sensors and response regulators from the two organisms can interact effectively.

ExoS cofractionates with the membrane.

In the hope of obtaining sufficient ExoS protein for raising antibodies and for biochemical characterization, we expressed N-terminal His-tagged ExoS in E. coli under the control of the T7 promoter. However, after induction no extra protein band corresponding to the predicted molecular weight of ExoS was detected when all the proteins from the induced cells were resolved on a polyacrylamide gel. For increased sensitivity, we separated the proteins from the induced cells into cytoplasmic and membrane fractions. One small extra band having the predicted molecular weight for ExoS could be seen when membrane proteins were separated on a gel. The very small amount of ExoS protein present in the induced cells would have made it difficult to purify enough protein for raising antibodies. Therefore, we instead constructed a plasmid that expressed an N-terminal His-tagged fragment of ExoS, ExoS_cyto67, consisting of amino acids 301 to 547, the region constituting most of the predicted histidine kinase domain. This ExoS_cyto67 protein was produced at high levels but was insoluble. We thus solubilized inclusion bodies in urea, purified the ExoS_cyto67 protein over a nickel column, resolved the purified protein on a polyacrylamide gel (Fig. ​(Fig.3),3), and raised antibodies against ExoS with the gel-purified protein. The removal of urea from the purified protein resulted in the precipitation of the protein.

Open in a separate window

FIG. 3

Comassie blue-stained sodium dodecyl sulfate-polyacrylamide gel of protein samples from the purification of the cytoplasmic domain of ExoS, ExoS_cyto67. Lane 1, protein size standards; lanes 2 and 3, total cellular proteins from cells carrying cloning vectors only that were grown under uninduced and induced conditions, respectively; lanes 4 and 5, total cellular proteins from cells carrying the plasmid encoding ExoS_cyto67 that were grown under uninduced and induced conditions, respectively; lane 6, ExoS_cyto67 purified over a nickel column.

The ExoS-specific antibodies were used to determine the cellular location of the ExoS protein. The total cellular protein of the wild-type strain Rm1021 cells was fractionated into cytoplasmic and membrane proteins after sonication and centrifugation. Equivalent amounts of protein from both fractions were resolved on a polyacrylamide gel, blotted, and probed with ExoS-specific antibodies. As shown in Fig. ​Fig.4A,4A, one strong band could be seen in the lane containing membrane proteins while one band of much lower intensity migrated at the same position in the lane containing cytoplasmic proteins. This lower-intensity band disappeared when the cytoplasmic protein fraction was subjected to further centrifugation at higher speed, which suggests that the band represents a small amount of ExoS that was associated with the smaller vesicle of the cytoplasmic membrane. These results indicate that ExoS is a membrane protein. This conclusion is consistent with the computer analyses of its putative peptide sequence and with the fact that ExoS cofractionated with membrane proteins in E. coli. Thus, it appears that ExoS is a sensor protein located in the cytoplasmic membrane.

Open in a separate window

FIG. 4

Cellular distribution of the ExoS protein examined in an immunoblot by using ExoS-specific antibodies. (A) Total cellular proteins from wild-type strain Rm1021 were separated into membrane and cytoplasmic fractions and examined in an immunoblot by using ExoS-specific antibodies. The ExoS protein was found to cofractionate predominantly in the membrane protein fraction. The cross-reacting material in the cytoplasmic protein fraction can be removed by additional high-speed centrifugation, suggesting that it represents ExoS protein which associated with smaller vesicles of cytoplasmic membrane. (B) Membrane proteins from Rm1021 and the exoS96 mutant were examined in an immunoblot by using ExoS-specific antibodies. Lane 1, wild-type strain proteins; lane 2, exoS96 mutant proteins. The ExoS band is missing from the lane containing ExoS96 membrane proteins. A faint new band, which is about 1/20 of the intensity of that of ExoS is seen; the band represents a protein that is 5 kDa smaller than ExoS.

exoS96 encodes an N-terminal truncated version of ExoS that cofractionates with the membrane.

We were interested in the properties of the exoS96::Tn5 allele for several reasons. We had initially assumed that it was a null allele because it was caused by a Tn5 insertion and because the exoS96::Tn5 allele was recessive to exoS⁺ carried on pM13 (10), an IncP-based cosmid vector. However, we were unsuccessful in homogenotizing a construct of the exoS gene in which part of the open reading frame had been replaced by a drug marker, despite several attempts employing minimal medium and the procedures that we have used routinely to generate hundreds of disruptions in R. meliloti. We were also unable to interupt the chvI gene by using the same approach. Furthermore, Østerås et al. (31) had reported that they were unsuccessful in homogenotizing chvG::Tn5 mutations into the genome. These observations raised the possibility that the R. meliloti chvI and exoS genes could not be deleted in a wild-type R. meliloti background and that there was something special about the exoS96 allele.

In the exoS96::Tn5 mutant, the site of insertion is within the exoS open reading frame, 57 bp downstream from the putative start site of the open reading frame. There is an in-frame ATG start codon 135 bp downstream from the beginning of the open reading frame. Since there is an E. coli outward-facing promoter within 69 bp of the end of IS50R (7), it seemed possible that the rest of the exoS open reading frame could still be transcribed in R. meliloti at a low level and then translated. This would result in the synthesis of a truncated ExoS missing the first 45 amino acids at the amino terminus, which would make the ExoS96 protein 5 kDa smaller than the ExoS protein. To examine this possibility, equal amounts of membrane proteins from the exoS96 mutant were resolved side by side with those from the wild-type strain and were probed with anti-ExoS antibodies (Fig. ​(Fig.4B).4B). For the exoS96 mutant, the wild-type ExoS band was missing. However there was a band which was about 5 kDa smaller. Its intensity was only 1/20 that of the ExoS band. These results suggest that in the exoS96 mutant there is a truncated ExoS being made at approximately 1/20 the level of ExoS in the wild-type strain. The size reduction detected in the experiment was consistent with our prediction, which suggests that the ATG located 136 bp inside the exoS open reading frame is being used to start the translation. The lower level of the ExoS96 protein compared to the ExoS protein could be due to a low level of transcription, a lack of a proper ribosomal binding site in front of the ATG start codon, or both.

The cytoplasmic domain of the ExoS protein can autophosphorylate at a histidine residue in vitro in the presence of ATP.

The high level of homology of ExoS to the histidine kinase domains of E. coli PhoR and EnvZ strongly suggests that ExoS is a histidine kinase which would autophosphorylate at the conserved histidine residue in the presence of ATP (Fig. ​(Fig.1).1). To confirm biochemically that ExoS has histidine kinase activities, plasmid pHC72 was constructed to express the entire cytoplasmic domain of ExoS, ExoS_cyto72, with an N-terminal histidine tag (Fig. ​(Fig.2).2). The ExoS_cyto72 protein consists of amino acids 281 to 578 from ExoS and forms inclusion bodies when it is overexpressed. It was purified under denaturing conditions and renatured as described in Materials and Methods. When the purified and renatured ExoS_cyto72 protein was incubated with [γ-³²P]ATP and then resolved on a polyacrylamide gel, a band appeared on the autoradiograph at the position corresponding to that of ExoS_cyto72 (Fig. ​(Fig.5).5). This indicated that the cytoplasmic domain of ExoS can autophosphorylate in the presence of [γ-³²P]ATP, a property which is a characteristic of all tested sensor proteins of two-component regulatory systems.

Open in a separate window

FIG. 5

Autoradiograph demonstrating autophosphorylation of the ExoS and ChvI proteins. (A) ExoS_cyto72 protein autophosphorylates in the presence of [γ-³²P]ATP. The phosphorylated ExoS_cyto72 loses its phosphate at acidic pH (0.1 N HCl), but not at neutral (buffer; pH 7.4), and alkaline (0.1 N NaOH) pHs. (B) ChvI protein autophosphorylates in the presence of [³²P]acetylphosphate. The phosphorylated ChvI protein loses its phosphate at alkaline pH (0.1 NaOH), but not at neutral (pH 7.4) and acidic (0.1 N HCl) pHs.

To test whether a histidine residue is the site of phosphorylation, we examined the stability of phosphorylated ExoS_cyto72 protein at different pHs (28). Histidyl-phosphate is stable at neutral and alkaline pH but is not stable at acidic pH. The phosphorylated ExoS_cyto72 protein was further incubated for 5 min at neutral (no additional buffer), acidic (0.1 N hydrochloric acid), or alkaline (0.1 N sodium hydroxide) pH (Fig. ​(Fig.5A).5A). The ExoS_cyto72 only lost the phosphate label after the acid treatment, an observation consistent with a histidine residue being the site of phosphorylation.

The ChvI protein can autophosphorylate at an aspartate residue in vitro in the presence of acetylphosphate.

On the basis of its sequence homology, and because of the genetic analysis of its A. tumefaciens homolog, the R. meliloti ChvI protein has been said to function as a response regulator in a two-component regulatory system (31). Because many response regulators have been shown to autophosphorylate in the presence of acetylphosphate (26, 45), we decided to test whether the R. meliloti ChvI protein shared this distinctive property.

We therefore cloned the chvI open reading frame and expressed a derivative carrying an N-terminal oligohistidine tag in E. coli. The resulting soluble protein was purified through a nickel column. The purified ChvI protein was then incubated with ³²P-acetylphosphate. As shown in Fig. ​Fig.5B,5B, the ChvI protein became radiolabeled when [³²P]acetylphosphate was present, suggesting that ChvI can autophosphorylate by using acetylphosphate as the phosphate donor. The phosphorylated ChvI is stable under acidic but not alkaline conditions; this is consistent with the phosphorylation occurring at an acidic residue (35), most likely Asp-52 by analogy to other response regulators. The ChvI protein was not phosphorylated when it was incubated with [γ-³²P]ATP (Fig. ​(Fig.6).6). All of these results are consistent with ChvI being a response regulator of a two-component regulatory system.

Open in a separate window

FIG. 6

Autoradiograph showing that ExoS_cyto72 phosphorylates ChvI protein in the presence of [γ-³²P]ATP. The ChvI protein became phosphorylated only when ExoS_cyto72 was present as well.

We thank John A. Leigh, Stephen C. Winans, Turlough M. Finan, and Tevor C. Charles for strains, unpublished results, and helpful suggestions. We thank members of this laboratory for many helpful suggestions and critical discussions.

This work is supported by Public Health Service grant GM31030 to G.C.W. from the National Institutes of Health.