Polar Localization of CheA₂ in Rhodobacter sphaeroides Requires Specific Che Homologs

Abstract

Rhodobacter sphaeroides is a motile bacterium that has multiple chemotaxis genes organized predominantly in three major operons (cheOp₁, cheOp₂, and cheOp₃). The chemoreceptor proteins are clustered at two distinct locations, the cell poles and in one or more cytoplasmic clusters. One intriguing possibility is that the physically distinct chemoreceptor clusters are each composed of a defined subset of specific chemotaxis proteins, including the chemoreceptors themselves plus specific CheW and CheA proteins. Here we report the subcellular localization of one such protein, CheA₂, under aerobic and photoheterotrophic growth conditions. CheA₂ is predominantly clustered and localized at the cell poles under both growth conditions. Furthermore, its localization is dependent upon one or more genes in cheOp₂ but not those of cheOp₁ or cheOp₃. In E. coli, the polar localization of CheA depends upon CheW. The R. sphaeroides cheOp₂ contains two cheW genes. Interestingly, CheW₂ is required under both aerobic and photoheterotrophic conditions, whereas CheW₃ is not required under aerobic conditions but appears to play a modest role under photoheterotrophic conditions. This suggests that R. sphaeroides contains at least two distinct chemotaxis complexes, possibly composed of proteins dedicated for each subcellular location. Furthermore, the composition of these spatially distinct complexes may change under different growth conditions.

It has been well documented that the localization of some bacterial proteins to specific regions in the cell is essential for their correct functioning. Examples of these include proteins involved in Caulobacter crescentus development, Bacillus subtilis sporulation, and Escherichia coli cell division (for a review, see reference 20). The chemotaxis pathway, which allows bacteria to move in a favorable direction, also has components that are specifically localized within the cell (14). In enteric bacteria, chemotaxis is mediated by a classical two-component signal transduction pathway (for reviews, see references 2, 4, and 26). The protein kinase CheA is phosphorylated on a conserved histidine residue due to a change in the signaling state of a trans-membrane chemoreceptor (trans-MCP). The phosphoryl group is transferred to the response regulator CheY. CheY-P binds to the flagellar switch protein FliM, causing the direction of flagellar rotation to change from counterclockwise to clockwise and ultimately resulting in a change of swimming direction. Adaptation to stable chemoeffector concentrations is accomplished by modification of the chemoreceptors using two enzymes working antagonistically. The constitutively active methyltransferase CheR adds methyl groups to specific glutamate residues of the trans-MCPs, thereby increasing the activity of CheA (1). CheB, when phosphorylated by CheA, removes these methyl groups, thus decreasing CheA activity (12).

In all bacteria and archaea examined thus far, the chemoreceptor complexes are clustered (6, 9, 13, 14). Although the function of clustering is currently unknown, one possibility is that clustering may allow cooperative interactions between receptors, facilitating signal generation, signal amplification and/or adaptation (3, 5, 10, 11, 21). In E. coli, the kinase CheA and the scaffolding protein CheW are also specifically localized to the cell poles (14), where they may form higher order signaling arrays with the chemoreceptors (3, 10, 11, 21).

Many motile bacteria sense and respond to environmental changes by employing variations of the E. coli paradigm. The α-subgroup bacterium Rhodobacter sphaeroides is a metabolically diverse species that has multiple homologs of the E. coli signaling proteins encoded in three operons and at other unlinked loci (www.jgi.doe.gov/JGI_microbial/html/rhodobacter). cheOp₁ contains cheY₁, cheA₁, cheW₁, cheR₁, and cheY₂. cheOp₂ contains cheY₃, cheA₂, cheW₂, cheW₃, cheR₂, cheB₁, and tlpC. cheOp₃ contains cheA₄, cheR₃, cheB₂, cheW₄, slp, tlpT, cheY₆, and cheA₃. In total, there are four CheAs, four CheWs, six CheYs, three CheRs, and two CheBs. In addition, there are a CheBRA fusion protein (encoded at a separate locus) and 13 chemoreceptors. Nine chemoreceptors are membrane-spanning and four are cytoplasmic, known as transducer-like proteins (Tlps). Immunoelectron microscopy using an antibody against the highly conserved domain of trans-MCPs showed that receptor proteins are clustered at both the cell poles and in the cytoplasm (9). Specific green fluorescent protein (GFP) fusions showed that the trans-membrane receptor, McpG, was located at the poles (24) and the putative cytoplasmic receptor, TlpC, formed discrete foci within the cytoplasm of the cell (25). Defining the distribution of these Che proteins in the cell and determining their interplay is critical for truly understanding chemotaxis in R. sphaeroides.

CheA₂ (encoded in cheOp₂) is essential for aerotaxis, phototaxis, and chemotaxis to all compounds tested and for the localization of McpG to the cell pole (15). Deletion of CheA₂ results in some, but not total, delocalization of TlpC (25). In contrast, deletion of cheA₁ has only minor effects on chemosensing and is not required for either the localization of McpG (15) or TlpC (25). In this study we examined the subcellular localization of CheA₂ in R. sphaeroides and systematically investigated the requirement for other signaling proteins in that localization.

MATERIALS AND METHODS

Strains and growth conditions.

Strains of R. sphaeroides (Table 1) were grown in succinate medium (22) containing nalidixic acid (25 μg/ml) at 30°C either aerobically with shaking in the dark or anaerobically with illumination at 50 μmol m⁻² s⁻¹.

TABLE 1.

Strains used in this study

Strain	Characteristic	Source or reference
WS8	Wild-type R. sphaeroides	Gift from W. Sistrom
WS8N	Spontaneous nalidixic acid-resistant mutant of WS8	23
JPA117	ΔcheOp1 derivative of WS8N	7
JPA1301	ΔcheOp3 derivative of WS8N	17
JPA1340	ΔcheBRA derivative of WS8N	Gift from S. L. Porter
JPA1349	Δ cheBRA cheY7 derivative of WS8N	Gift from S. L. Porter
JPA211	ΔcheA2 derivative of WS8N	7
JPA470	ΔtlpC derivative of WS8N	25
JPA514	ΔcheW2 derivative of WS8N	15
JPA517	ΔcheB1 derivative of WS8N	16
JPA527	ΔcheW3 derivative of WS8N	15
JPA531	WS8N containing an Ω cartidge interrupting transcription and translation of mcpG	24
JPA565	ΔcheR2 derivative of WS8N	16

Antibody production.

Purified His-tagged CheA₂ was made as described previously (17). Rabbit polyclonal antibodies to His-tagged CheA₂ (Eurogentec) detect both His-tagged CheA₂ protein and a protein from R. sphaeroides WS8N extracts of the expected molecular mass (69.4 kDa). This immunoreactive protein was absent from the ΔcheA₂ control strain (JPA211). The antiserum was immunodepleted before use with acetone powders prepared from JPA211 by standard methods (8).

Electron microscopy.

Motile cultures of R. sphaeroides were fixed, embedded in LR-White resin, sectioned and placed on nickel grids as described previously (9). Immunoelectron microscopy was performed using a 1:500 dilution of primary antibody and a 1:30 dilution of secondary antibody (12-nm-diameter colloidal gold particles conjugated to goat antibody to rabbit immunoglobulin G; Jackson Immunoresearch) as described previously (9).

The intracellular positions of all gold particles in longitudinal sections of predivisional cells were recorded. Gold particles within 20 nm of the membrane were scored as being membrane associated. These were further subdivided into those along the lateral membrane (lateral) and those associated with the polar membrane (polar). We also tracked the colocalization (clustering) of gold particles. For this study, a cluster was defined as three or more gold particles each located no more than 20 nm from its neighbor, together with any outlying particles that were no more than 40 nm from the core cluster. Statistical analysis was performed using the χ² test.

Immunoblotting.

Because the packing of proteins can influence the number of gold particles, using immunoelectron microscopy, CheA₂ levels were monitored by immunoblotting. Motile cells (1 ml, optical density at 700 nm = 0.6) were harvested and resuspended in 100 μl of sample buffer (0.05 M Tris-Cl [pH 6.8], 10% glycerol, 1% sodium dodecyl sulfate, 0.05 M dithiothreitol, 0.01% bromophenol blue), and 10 μl was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (12% polyacrylamide) and electroblotted by standard methods (18). The membranes were blocked in 5% dried milk, incubated for 1 h in preabsorbed sera diluted 1/2,000 in phosphate-buffered saline (PBS) containing 1% dried milk, and washed extensively with PBS. The membrane was then blocked in PBS containing 0.2% Tween 20, incubated for 1 h with a 1/1,000 dilution of anti-rabbit horseradish peroxidase conjugate (Dako), and washed, and bands were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech).

To determine the copy number of CheA₂ in the cell, quantitative immunoblots of dilution series of WS8N extracts (prepared as above) and CheA₂ protein were performed. The mean of results frp, three independent experiments was taken.

RESULTS

Localization of CheA₂.

To determine the localization of CheA₂ in aerobically grown cells, immunoelectron microscopy experiments were performed using CheA₂ antibody on ultrathin sections of wild-type cells. Of the gold particles, 78% were associated with the membrane, and 86% of these were polar (67% of the total particles) (Fig. 1; Table 2). Analysis of the polar signal revealed that typically, more gold particles were seen at one pole than at the other (71% of cells had at least two more particles at one pole than at the other), perhaps reflecting a bias for the localization of CheA₂ to the older pole. Polar CheA₂ was moderately clustered (52%). There were on average 8.6 gold particles per cell section, with very little noise evident (0.4 particles per cell in the ΔcheA₂ control strain, JPA211). Therefore, the antibody does not cross-react with the other CheA species. Some CheA₂ is found in the cytoplasm (1.9 cytoplasmic particles per cell compared to 0.2 in JPA211), but no specific positioning of the gold particles was observed and they were not clustered. Thus, the majority of CheA₂ is localized to the poles of the R. sphaeroides cell.

FIG. 1.

CheA₂ localizes predominantly to the cell poles in R. sphaeroides cells. Ultrathin sections of wild-type cells grown aerobically (A) and photoheterotrophically (B) were incubated with an antibody to CheA₂ and detected with an anti-rabbit colloidal gold conjugate. Few gold particles are present in JPA211, a strain lacking cheA₂ (C). The micrographs show the location of CheA₂ as detected by immunoelectron microscopy experiments using an antibody to CheA₂.

TABLE 2.

Spatial distribution of CheA₂ in aerobically grown wild-type strains and deletion mutants^a

Strain	No. polar	Polar in clusters	No. lateral	No. cytoplasmic	Cytoplasmic in clusters	No. of particles/section	No. of polar particles/section	No. of cytoplasmic particles/section
WS8N	930	482	146	304	9	8.6	5.8	1.9
ΔcheA2	33	10	14	33	0	0.4	0.2	0.2
ΔcheR2	838	364	100	186	3	7.0	5.2	1.2
ΔcheB1	817	321	90	171	6	6.7	5.1	1.1
ΔtlpC	781	320	89	199	0	6.7	4.9	1.2
ΔcheW2	175	9	222	743	0	7.0	1.0	4.6
ΔcheW3	651	259	171	258	12	6.8	4.1	1.6

^aFor each strain 160 longitudinal sections were scored.

Dependence of CheA₂ localization on components of cheOp₁ and cheOp₃, on CheBRA, and on McpG under aerobic conditions.

To determine whether Che proteins encoded by cheOp₁, cheOp₃, or cheBRA were required for the polar localization of CheA₂, immunoelectron microscopy was performed on sections of strains with deletions of cheOp₁ (JPA117), cheOp₃ (JPA1301) and cheBRA (JPA1340). Since both CheA₂ and CheW₂ are required for the localization of McpG under aerobic conditions (15), we also examined the requirement for McpG in CheA₂ localization in a strain in which mcpG had been insertionally inactivated by an Ω cartridge (JPA531) (24). The data obtained from all these strains were very similar to those obtained from wild-type cells (data not shown). Therefore, none of the components of cheOp₁ or cheOp₃ (which include genes encoding CheA₁, CheA₃, CheA₄, CheW₁, and CheW₄), cheBRA, or mcpG are required for CheA₂ localization.

Dependence of CheA₂ localization on components of cheOp₂ under aerobic conditions.

Proteins encoded within the same operon as CheA₂ were examined, individually, for any role in CheA₂ localization. The pattern of CheA₂ localization in strains from which cheW₃ (JPA527), cheR₂ (JPA565), cheB₁ (JPA517), and tlpC (JPA470) were deleted was not significantly different from that of the wild-type strain (P > 0.05) (Table 2). In the absence of CheW₂ (JPA514), however, the number of polar gold particles decreased dramatically (1.0 and 5.8 polar membrane particles in cells lacking cheW₂ and in the wild type, respectively) (Table 2; Fig. 2). Only 15% of the signal was polar (compared to 67% for the wild type) and very few polar particles were in clusters (5%). There was a concomitant 2.4-fold increase in the number of cytoplasmic particles and an apparent 1.5-fold increase in the number of lateral membrane particles. Despite there being an increase in the number of cytoplasmic and lateral particles, they were not clustered. These data show that CheW₂ is required for normal CheA₂ polar localization and clustering.

FIG. 2.

The localization of CheA₂ in aerobically grown cells is significantly affected by a deletion of cheW₂ (A), only slightly affected by the deletion of cheW₃ under aerobic conditions (B), and moderately affected by the deletion of cheW₃ under photoheterotrophic conditions (C). The micrographs show the location of CheA₂ as detected by immunoelectron microscopy experiments using an antibody to CheA₂.

Dependence of CheA₂ localization under photoheterotrophic conditions.

We have previously shown that CheW₃ has a more pronounced role in chemotaxis and in McpG localization under photoheterotrophic conditions than under aerobic conditions (15). Therefore, we addressed whether CheW₃ had a role in the localization of CheA₂ in cells grown photoheterotrophically.

The pattern of CheA₂ localization in photoheterotrophically grown wild-type cells was very similar (P > 0.05) to that observed in aerobically grown cells (Fig. 1). The amount of signal, however, was approximately twofold lower (3.9 and 8.6 spots per cell section under photoheterotrophic and aerobic conditions, respectively). When grown photoheterotrophically, 75% of the gold particles were associated with the membrane and 88% of these were polar (Table 3). Clustering of the polar particles was lower than in aerobically grown cells (27 and 52%, respectively) but may be a consequence of the lower signal observed (2.6 and 5.8 polar membrane particles under photoheterotrophic and aerobic conditions, respectively).

TABLE 3.

Spatial distribution of CheA₂ in photoheterorophically grown wild-type strains and deletion mutants^a

Strain	No. polar	Polar in clusters	No. lateral	No. cytoplasmic	Cytoplasmic in clusters	No. of particles/section	No. of polar particles/section	No. of cytoplasmic particles/section
WS8N	415	113	59	154	3	3.9	2.6	1.0
ΔcheA2	105	0	14	50	0	0.5	0.1	0.3
ΔcheW2	103	0	123	444	23	4.2	0.6	2.8
ΔcheW3	368	188	104	188	16	4.1	2.3	1.2

^aFor each strain 160 longitudinal sections were scored.

In the absence of CheW₂, under photoheterotrophic conditions, the percentage of particles that were polar decreased from 66 to 15% (P < 0.001) (Table 3), consistent with a critical requirement for CheW₂ under these conditions. The 4.3-fold decrease in the average number of polar particles was accompanied by a 2.8-fold increase in the cytoplasmic signal and a 2.1-fold increase in the lateral membrane signal. Interestingly, in the absence of CheW₃, the percentage of polar particles decreased to 56%, a much smaller but nevertheless statistically significant reduction (P < 0.05) (Table 3). In addition, a concomitant increase in the number of cytoplasmic and lateral particles was observed in cells lacking CheW₃ (Fig. 2). These data demonstrate that CheW₂ and, to a lesser extent, CheW₃ are required for the normal localization of CheA₂ under photoheterotrophic conditions.

Expression of CheA₂ is not affected by the deletion of other che genes.

Quantitative immunoblots of CheA₂ in the appropriate deletion strains were performed to ensure that any observed differences in the pattern of CheA₂ localization was not due to changes in CheA₂ expression levels. In aerobic cells the level of CheA₂ remained constant in all the strains used in this study, with the exception of JPA211, the negative control (Fig. 3). Under photoheterotrophic conditions, the total level of CheA₂ was reduced, but at a consistent level, in all strains except the negative control strain. Therefore, the CheA₂ localization pattern in the deletion strains was not influenced by CheA₂ levels under either environmental condition.

FIG. 3.

Expression of CheA₂ is not affected by the deletion of other Che proteins. Shown is a quantitative immunoblot of CheA₂ in the wild type (lane 1) and in all the appropriate deletion strains under aerobic conditions: ΔtlpC (lane 3), ΔcheB₁ (lane 4), ΔcheR₂ (lane 5), ΔcheW₃ (lane 6), ΔcheW₂ (lane 7), ΔcheBRA (lane 8), ΔcheOp₃ (lane 9), and ΔcheOp₁ (lane 10). There is no band in the ΔcheA₂ negative control (lane 2).

Quantitative immunoblots using known amounts of purified CheA₂ protein showed that the number of copies of CheA₂ in R. sphaeroides was approximately 7,000 and 1,000 per cell under aerobic and photoheterotrophic conditions, respectively. This is consistent with the lower signal observed in photoheterotrophic cells by immunoelectron microscopy.

DISCUSSION

CheA₂ is the most highly expressed CheA in R. sphaeroides (19) and is essential for all measured responses under both aerobic and photoheterotrophic conditions (15). Here we show that CheA₂ is clustered at the cell pole. It is now well established that the chemoreceptors form clusters or signaling complexes at the cell poles of a variety of bacterial species (6, 14). In R. sphaeroides, however, receptors are clustered at the cell poles and in the cytoplasm (9). We have previously shown using GFP fusions that McpG predominately localizes to the poles (24) whereas TlpC localizes to cytoplasmic foci. It seems probable that CheA₂ forms part of a polar signaling complex that includes McpG.

CheA₂ was only moderately clustered at the cell pole (52% of polar membrane particles are clustered). One intriguing possibility is that CheA₁, CheA₃, and/or CheA₄ is in the same signaling array and, therefore, reduces the density of CheA₂ within this array. It remains to be determined, however, if these other CheA homologs localize to the cell poles or to the cytoplasmic cluster. Interestingly, the pattern of localization of CheA₂ was the same in cells grown aerobically and photoheterotrophically even though the copy number of CheA₂ protein was sevenfold higher in aerobically grown cells, suggesting no significant redistribution of CheA₂ under these different environmental conditions.

Deletion of cheW₂ resulted in a major reduction in the polarity and clustering of CheA₂ in both aerobic and photoheterotrophic cells. In the absence of CheW₂, CheA₂ was predominantly in the cytoplasm although some CheA₂ was also associated with the lateral membrane (Fig. 2). These data are consistent with those from E. coli, where CheW (the only CheW in E. coli) was found to be required for the polarity of both the chemoreceptors and CheA (14). Deletion of cheW₃ resulted in a modest reduction in CheA₂ polar localization only in photoheterotrophically grown cells, suggesting that CheW₂ is absolutely essential for CheA₂ localization while CheW₃ may only be required for optimal CheA₂ localization in photoheterotrophic cells. These results are consistent with previously published data that described the localization of McpG (15). There is evidence that the chemoreceptors of R. sphaeroides are differentially expressed according to the environmental condition (9); therefore, the receptors that are more highly expressed under photoheterotrophic conditions may require both CheW₂ and CheW₃ for optimal packing in the array.

Deletion of mcpG does not affect CheA₂ localization. The absence of a single, albeit highly expressed chemoreceptor, is insufficient to disorder the polar CheA₂ cluster. In E. coli, CheA and CheW polar clustering is disrupted in the absence of all of the chemoreceptors (14); however, expression of one chemoreceptor is sufficient to sequester both CheA and CheW to the cell poles (13, 14). Thus, CheA is targeted to the pole provided that at least one chemoreceptor is present (14). Given that R. sphaeroides has eight additional predicted membrane receptors, the localization of CheA₂ in the absence of McpG is not surprising.

In R. sphaeroides, there appear to be two discrete regions in the cell that are absolutely required for chemotactic response: the cell poles and the cytoplasmic foci. Whether there is communication between the two loci to produce a balanced response at the single flagellar motor is the subject of ongoing experimentation. In the absence of CheA₂ there was a small, yet consistent, delocalization of TlpC-GFP fluorescence. It seems likely that deletion of cheA₂ affects the stoichiometry of other CheA, CheW, and receptor proteins, resulting in a modest TlpC delocalization as CheA₂ was not found associated with the cytoplasmic clusters and therefore probably does not have a major role in signaling from the cytoplasmic receptors.

These data are consistent with a recently proposed model (15) which suggests that there are polar signaling arrays containing up to nine chemoreceptor homologs, CheA₂, CheW₂, and, under photoheterotrophic conditions when different receptors may be expressed, CheW₃. Similarly, the cytoplasmic clusters may be composed of up to four soluble receptors, including TlpC (25) and other dedicated CheA and CheW proteins. Interestingly, it is also likely that these spatially distinct complexes also share one or more components and may be more dynamic, as evidenced by the dual role that CheW₃ plays in signaling from both polar and cytoplasmic clusters (15, 25).

We have shown that genes in both cheOp₂ and cheOp₃ are essential for chemotaxis and that proteins encoded by the two operons are found in two locations, the cell poles and a cytoplasmic cluster. Disruption of either the polar or cytoplasmic cluster appears to result in the loss of taxis (15). The data presented here show that CheA₂, itself essential for all tactic responses (15), is found only in the polar clusters and not in the cytoplasmic cluster. This suggests that essential chemosensory proteins may be differentially targeted to different cellular locations.