|Home | About | Journals | Submit | Contact Us | Français|
In the model organism Escherichia coli, the coupling protein CheW, which bridges the chemoreceptors and histidine kinase CheA, is essential for chemotaxis. Unlike the situation in E. coli, Borrelia burgdorferi, the causative agent of Lyme disease, has three cheW homologues (cheW1, cheW2, and cheW3). Here, a comprehensive approach is utilized to investigate the roles of the three cheWs in chemotaxis of B. burgdorferi. First, genetic studies indicated that both the cheW1 and cheW3 genes are essential for chemotaxis, as the mutants had altered swimming behaviors and were non-chemotactic. Second, immunofluorescence and cryo-electron tomography studies suggested that both CheW1 and CheW3 are involved in the assembly of chemoreceptor arrays at the cell poles. In contrast to cheW1 and cheW3, cheW2 is dispensable for chemotaxis and assembly of the chemoreceptor arrays. Finally, immunoprecipitation studies demonstrated that the three CheWs interact with different CheAs: CheW1 and CheW3 interact with CheA2 whereas CheW2 binds to CheA1. Collectively, our results indicate that CheW1 and CheW3 are incorporated into one chemosensory pathway that is essential for B. burgdorferi chemotaxis. Although many bacteria have more than one homologue of CheW, to our knowledge, this report provides the first experimental evidence that two CheW proteins co-exist in one chemosensory pathway and that both are essential for chemotaxis.
Chemotaxis allows motile bacteria to swim towards a favorable environment or away from one that is toxic. The signaling transduction system controlling bacterial chemotaxis has been extensively studied in two model organisms, Escherichia coli and Salmonella enterica [for recent reviews, see (Wadhams and Armitage, 2004;Sourjik and Armitage, 2010;Hazelbauer et al., 2008)]. The core structural unit in the chemotaxis signaling pathway consists of a ternary complex of chemoreceptors (often referred to as methyl-accepting chemotaxis proteins, MCPs), a histidine autokinase CheA, and a coupling protein CheW (Gegner et al., 1992;Liu and Parkinson, 1989). CheW is a single-domain cytoplasmic protein (Griswold and Dahlquist, 2002).
MCPs sense various environmental signals, which control the activity of CheA. Activated CheA (CheAP) transfers its phosphoryl group to CheY, a response regulator that controls the rotational direction of flagellar motors. The phosphorylated CheY (CheY-P) diffuses from the core complex to the flagellar motors, where it binds motor-switch complex proteins to promote a switch in the rotational direction from counterclockwise (CCW) to clockwise (CW). CCW rotation results in smooth swimming (also referred to as run), and CW rotation leads to tumbling. Cells responding to a positive response (binding of an attractant to MCPs) lengthen the intervals between tumbling events and hence have longer runs that allow the bacteria to swim preferentially toward higher concentrations of attractants (Sourjik and Armitage, 2010;Porter et al., 2011). In the enteric bacteria, there are single homologues of cheA, cheW and cheY, and null mutations in any of these genes cause cells to run constantly and to become deficient in chemotaxis (Parkinson, 1977;Parkinson and Houts, 1982).
Borrelia burgdorferi, the causative agent of Lyme disease (Burgdorfer et al., 1982), is highly motile and shows chemotactic responses to several attractants produced by the hosts (Charon and Goldstein, 2002; Bakker et al., 2007;Shih et al., 2002). Our recent study shows that chemotaxis is involved in the pathogenicity of B. burgdorferi (Sze et al., 2012). Chemotaxis in B. burgdorferi differs from that of E. coli and S. enterica in several important respects [for recent reviews, see (Charon and Goldstein, 2002; Charon et al., 2012)]. B. burgdorferi cells are relatively long (10 to 20 μm in length) and thin (0.3 μm in diameter), and two flat ribbons of periplasmic flagella (PFs) arise in the subpolar region at each cell end (Charon et al., 2009;Liu et al., 2009). Motility is powered by the coordinated rotation of the PFs. This architecture requires that the swimming behavior of spirochetes is very different from that of the peritrichously flagellated enteric bacteria (Dombrowski et al., 2009;Yang et al., 2011;Harman et al., 2012;Goldstein et al., 1994;Li et al., 2002;Motaleb et al., 2011b;Motaleb et al., 2005). B. burgdorferi has three swimming modes: run, flex, and reversal. A run occurs when the bundle of PFs at the anterior end rotates CCW and that at the posterior end rotates CW. A reversal happens when both bundles change their rotational direction nearly simultaneously. A flex represents a non-translational mode when the two bundles of PFs rotate in the same direction (both CCW or both CW).
During a chemotaxis response, the spirochetes must coordinate the rotation of the motors at the two ends of cells (i.e., repressing the time spent in flexing and reversing, and increasing the time spent in running). A long-standing question about the spirochete chemotaxis is how the cells achieve this coordination (Li et al., 2002;Charon and Goldstein, 2002;Charon et al., 2012). In the spirochetes, the motors at the two ends of the cells are located at a considerable distance from one another (at least 10 μm), and the MCPs form clusters that are in close proximity to the motors (Xu et al., 2011;Briegel et al., 2009;Charon et al., 2009;Liu et al., 2009). It would seem too slow to transmit signals from one end of the cell to the other simply by diffusion of CheY-P (Motaleb et al., 2011b;Sarkar et al., 2010;Porter et al., 2011).
Unlike E. coli and S. enterica, B. burgdorferi contains more than one homologue of cheA, cheW, and cheY: two cheAs (cheA1 and cheA2), three cheWs (cheW1, cheW2 and cheW3), and three cheYs (cheY1, cheY2 and cheY3) (Fraser et al., 1997;Charon and Goldstein, 2002). Many of these genes reside within two gene clusters: the flaA operon (flaA-cheA2-cheW3-cheX-cheY3) and the cheW2 operon (cheW2-bb0566-cheA1-cheB2-bb0569-cheY2) (Ge and Charon, 1997;Li et al., 2002). We have recently identified several genes that are essential for the chemotaxis of B. burgdorferi, including cheA2, cheY3, and cheX (an analogue of cheZ from E. coli). The cheA2 and cheY3 mutants fail to reverse and constantly run, whereas the cheX mutant constantly flexes. None of these mutants is able to carry out chemotaxis (Motaleb et al., 2011b;Motaleb et al., 2005;Li et al., 2002;Bakker et al., 2007;Sze et al., 2012).
In contrast to the flaA operon, the genes studied to date in the cheW2 operon are not required for the chemotaxis of B. burgdorferi, e.g., the cheA1 and cheY2 mutants have a chemotaxis phenotype that is similar to wild type (Li et al., 2002;Motaleb et al., 2011b). It has been speculated that B. burgdorferi may possess two chemotaxis pathways that function in different hosts during the infection cycle (Li et al., 2002;Charon and Goldstein, 2002;Sze et al., 2012). For example, the chemotaxis genes (cheA2-cheW3-cheX-cheY3) in the flaA operon may form a pathway that executes chemotaxis in mammalian hosts, whereas the genes in the cheW2 operon (cheW2-cheA1-cheY2) may constitute a pathway that controls chemotaxis in the tick vector. In E. coli, CheW interacts with both MCPs and CheA and plays a pivotal role in chemotaxis and formation of the MCP-CheW-CheA ternary complexes (Gegner et al., 1992;Liu and Parkinson, 1989;Vu et al., 2012;Boukhvalova et al., 2002b). Thus, elucidating the roles of the three CheWs of B. burgdorferi in chemotaxis will help us determine whether this organism has two different chemotaxis pathways.
In this report, the three cheW genes of B. burgdorferi were separately inactivated by allelic exchange mutagenesis, and their roles in chemotaxis and chemoreceptor assembly were investigated by an approach consisting of computer-based bacterial tracking analysis, swim plate and capillary assays, immunofluorescence assay (IFA), and cryo-electron tomography (cryo-ET). Furthermore, the interactions between the two CheAs and three CheWs were studied by co-immunoprecipitation (co-IP). The results support the idea that B. burgdorferi has two different chemosensory pathways: CheW1/CheW3-CheA2-CheY3, which form a pathway that is essential for chemotaxis under the tested in vitro conditions; CheW2-CheA1-CheY2 and/or CheY1, whichform another pathway that is either required for chemotaxis under other conditions or is involved in a different signaling pathway.
Among the three cheW genes, cheW2 (bb0565) is the first gene in the cheW2 operon, cheW3 (bb0670) is the third gene in the flaA operon, and cheW1 (bb0312) is located in a gene cluster where no other putative chemotaxis or motility genes are evident (Fraser et al., 1997;Charon and Goldstein, 2002;Li et al., 2002). CheW1 consists of 176 amino acids (aa) with a predicated molecular weight (MW) of 20 kDa. CheW2 is 180 aa in length with a predicted MW of 21 kDa. CheW3 contains 466 aa, and its predicted MW is 53 kDa. A Blast search showed that the N-terminus of CheW3 is a conserved CheW domain (aa 26 to 165) and that its C-terminus (aa 196 to 466) contains a CheR-like domain (Figure S1) (Djordjevic and Stock, 1997;Djordjevic and Stock, 1998;Shiomi et al., 2002). The E. coli CheA contains a CheW-like domain, P5, which mediates the interaction between CheA and CheW (Bilwes et al., 1999;Park et al., 2006). Sequence alignment showed that the three CheW proteins also share certain similarity to the P5 domains from the CheA proteins of E. coli and B. burgdorferi (Figure S2).
The function of CheW has been extensively studied in E. coli, and the key residues involved in the CheW/MCP and CheW/CheA interactions have been identified (Boukhvalova et al., 2002a; Boukhvalova et al., 2002b;Liu and Parkinson, 1989;Liu and Parkinson, 1991;Cardozo et al., 2010;Vu et al., 2012). B. burgdorferi CheWs share 28% (CheW1), 28% (CheW2), and 30% (the CheW domain in CheW3) sequence identity with E. coli CheW (CheWEc). Sequence alignment disclosed that the majority of the residues essential for the function of CheWEc are conserved among the three CheWs (Figure 1), including I33, E38, G57, R62, G63, G99, V108, and G133. A few residue variations were also observed (e.g., V36/I in CheW1, V88/M and V105/I in CheW3; see Figure 1). These similarities suggest that all three CheWs may function like CheWEc.
The structure of T. maritima CheW (designated as CheWTm) has been determined by nuclear magnetic resonance (NMR) (Griswold and Dahlquist, 2002;Park et al., 2006), and CheWEc and CheWTm appear to share a very similar 3D structure (Li et al., 2007). To reveal the structural features of CheW1, CheW2, and CheW3, homology modeling analysis was conducted using CheWTm as a structure template. Like CheWTm, all three CheW proteins are predicted to contain two β-sheet domains (domain 1 and domain 2), and each domain consists of a five-stranded β-barrel (Figure 2). In addition, five highly variable regions (HVR) were identified (Figure 2). Structural alignment revealed that the root-mean-square deviations (RMSD) of backbone atoms between CheWTm (blue) and CheW1 (yellow), CheW2 (orange), or the N-terminal CheW domain of CheW3 (red) were 0.566 Å, 1.617 Å, and 0.347 Å, respectively. In contrast to CheW1 and CheW3, CheW2 had a long loop inserted near the N-terminus of β strand 6 in domain 2 (Figure 2B), within the binding interface predicted for CheA (Griswold and Dahlquist, 2002;Park et al., 2006). These structural features suggest that CheW1 and CheW3 are more structurally similar to CheWEc and CheWTm than is CheW2.
As a coupling protein, CheW interacts with both MCPs and CheA. In E. coli, CheW plays a critical role in chemotaxis; a cheW null mutant constantly runs and is deficient in chemotaxis (Parkinson, 1977;Liu and Parkinson, 1989;Liu and Parkinson, 1991). To investigate the roles of CheW1, CheW2, and CheW3 in chemotaxis, the genes encoding these three proteins were inactivated by allelic exchange mutagenesis (described in Materials and Methods). A PCR analysis showed that the individual cheW genes were targeted by the antibiotic resistant makers as expected (Figure S3).
A single clone representing each mutation (ΔW1, ΔW2, and ΔW3, which represent the cheW1, cheW2, and cheW3 mutants, respectively) was selected for further characterizations. Immunoblot analyses using anti-CheW antisera (designated asαCheW1, αCheW2, and αCheW3) showed that CheW1, CheW2, and CheW3 were all detected in the wild-type strain B31A but not in the corresponding mutant clones (Figure 3). Among these three mutants, as ΔW1 and ΔW3 had altered chemosensory behaviors, these two mutants were complemented using the vectors CheW1/pBSV2G and CheW3/pBSV2G, which were constructed as described in the Materials and Methods. Immunoblot analyses showed that the complementation of cheW1 (ΔW1+) and cheW3 (ΔW3+) by the corresponding wild-type genes restored the synthesis of CheW1 (Figure 3A) and CheW3 (Figure 3C).
Chemotaxis in the ΔW1, ΔW2, and ΔW3 mutants was characterized using swim plate and capillary assays. In the swim plate assay, the ΔW2 mutant formed similar-sized colonies as the B31A strain (Figure 4B). However, the ΔW1 and ΔW3 mutants formed considerably smaller rings that were similar to that of a ΔflaB strain (Figure 4A & C), a previously documented non-motile mutant (Motaleb et al., 2000). Thus, cheW1 and cheW3, but not cheW2, are critical for chemotaxis under the tested conditions. Consistent with the results of swim plate assay, the capillary assay demonstrated that ΔW1 and ΔW3 do not respond to GlcNAc as an attractant (Figure 4D & F), whereas the ΔW2 mutantshowed the same response to GlcNAc as the wild-type strain (Figure 4E). The cognate complemented strains, ΔW1+ and ΔW3+, exhibited spreading on the swim plates and chemotactic responses to GlcNAc at wild-type levels (Figure 4A, C, D, and F). Collectively, these results indicate that cheW1 and cheW3 are required for B. burgdorferi chemotaxis, whereas cheW2 is dispensable for chemotaxis.
Non-chemotactic mutants often show altered swimming behaviors, e.g., the cheA2 and cheY3 mutants of B. burgdorferi fail to reverse and constantly run (Motaleb et al., 2011b;Li et al., 2002). The tracking analysis using a computer-assisted cell tracker coupled with video microscopy disclosed that the ΔW2 mutant had swimming behavior indistinguishable from (Video 2, Table 1) the wild type (Video 1), whereas the ΔW1 and ΔW3 mutants had altered swimming behaviors. The ΔW3 mutant failed to reverse and constantly ran in one direction (Video 3, Table 1), like the cheA2 and cheY3 mutants of B. burgdorferi. The behavior of the ΔW1 mutant is mixed (Video 4): approximately half of the cells (21 out of 50) failed to reverse and swam exclusively in one direction. The remainder of the cells (29 out of 50) reversed, but at a lower reversal frequency (9 reversals/min) compared to the wild type (23 reversals/min). A similar pattern was observed in a reconstructed ΔW1 mutant, suggesting that the observed mixed phenotype is stochastic and not caused by genetic heterogeneity. The complemented mutants (ΔW3+and ΔW1+) had a similar swimming behavior as the wild type (Video 3A, Video 4A, and Table 1). All three cheW mutants had similar swimming velocities as the wild type (Table 1), ranging from 9 to12 μm/sec. Thus, none of the cheW mutations causes a decrease in the propulsive force generated by the flagella.
CheW3 possesses a CheR-like domain at its C-terminus (Figure S1). In E. coli, CheR functions as a methyltransferase that is involved in chemoreceptor adaptation (Djordjevic and Stock, 1997;Djordjevic and Stock, 1998;Porter et al., 2011). Searching large sets of CheW homologues from microbial genome databases revealed that only CheWs from some spirochete species have a similar domain composition as CheW3, including CheW1 (TP_0364) of Treponema pallidum and CheW1 (TDE_1492)of Treponema denticola (Fraser et al., 1998;Seshadri et al., 2004). To determine whether the CheR-like domain is required for normal chemotaxis, the ΔW3 mutantwas complemented with a plasmid producing only the N-terminal CheW domain of CheW3 (aa 1–210). Immunoblotting using αCheW3 showed that the expression of the N-terminal CheW domain was restored in the complemented clone (ΔW3N+) (Figure 3C). The swim plate (Figure 4C), capillary (Figure 4F), and tracking (Table 1) assays demonstrated that chemotaxis in the ΔW3N+ strain was indistinguishable from that of the wild-type and ΔW3+ strains, indicating that deletion of the CheR-like domain does not affect the chemotactic function of CheW3 under the conditions tested.
In E. coli, CheW is essential for the assembly of chemoreceptor arrays at the cell poles (Studdert and Parkinson, 2005;Maddock and Shapiro, 1993;Sourjik and Berg, 2000). Our previous studies showed that B. burgdorferi MCPs also form arrays at the cell poles (Xu et al., 2011). To determine whether the B. burgdorferi cheW mutants are defective in chemoreceptor assembly, the cellular location of the MCPs in the three mutants was determined by IFA using an antibody targeted specifically against B. burgdorferi MCP3 (Xu et al., 2011). As expected, bright fluorescent loci were observed at both cell poles in wild-type cells (Figure 5A). A similar pattern was observed in ΔW2 cells (Figure 5C), but not in ΔW3 cells, in which the fluorescence was diffused (Figure 5D). Although fluorescent loci were still evident at the poles of ΔW1 mutant cells, the fluorescence signals were considerably reduced, and even absent in many cells (Figure 5B). The IFA results suggest that CheW1 and CheW3, but not CheW2, are involved in the assembly and localization of the chemoreceptor arrays.
Cryo-ET was conducted to determine the cellular locations and ultrastructures of the chemoreceptor arrays in the three cheW mutants more precisely. Chemoreceptor arrays could be readily recognized as prominent ‘basal plate’-like structures (Zhang et al., 2004;Briegel et al., 2009;Briegel et al., 2012;Xu et al., 2011;Liu et al., 2012) at the poles of wild-type (Figure 6A) and ΔW2 cells (Figure 6B). The arrays had an average length of 159 ± 86 nm (n=19 cells, Table 2). No chemoreceptor arrays were observed in any of the ΔW3 cells examined (0 out of 25 cells, Figure 6C). However, the arrays could be readily detected in its complemented strain ΔW3+ (12 out of 30 cells, Figure 6D). With the ΔW1 mutant, the arrays were still evident in a small portion of the cells (4 out 31 cells, Figure 6E), but their sizes were substantially reduced (average length of 75 ± 7 nm, n=4 cells) compared to those of the wild type or the complemented ΔW1+strain (Figure 6F, 152 ± 58 nm, n = 10 cells). The cryo-ET results are consistent with the IFA data and thus further confirm that both CheW1 and CheW3 are involvedin assembly of the chemoreceptor arrays, whereas CheW2 is not.
In E. coli, the ternary complex of MCP-CheW-CheA is the core structural unit in the signaling pathway of chemotaxis (Wadhams and Armitage, 2004;Hazelbauer et al., 2008). B. burgdorferi has two CheA homologues, CheA1 and CheA2. Identifying the interactions between the two CheAs and the three CheWs will help us understand the complexity of chemotaxis signaling pathways in B. burgdorferi. Co-IP experiments were carried out to reveal the interactions between the two CheAs and the three CheWs. For the co-IP assays, either CheA1 antibody (αCheA1) or CheA2 antibody (αCheA2) was first co-incubated with whole cell lysates of the B31A wild type and a previously constructed double cheA1cheA2 mutant (designated as ΔA1A2 and used as a negative control) (Li et al., 2002). The co-precipitated products were then probed with αCheW1, αCheW2, or αCheW3, respectively. As shown in Figure 7, CheW1 (Figure 7A) and CheW3 (Figure 7C) were detected in the samples precipitated by αCheA2 (left panel, Figure 7) but not by αCheA1 (right panel, Figure 7), whereas CheW2 was detected in the samples precipitated by αCheA1 (right panel, Figure 7B) but not by αCheA2 (left panel, Figure 7B), suggesting that both CheW1 and CheW3 interact with CheA2, whereas CheW2 binds CheA1. To confirm that CheW1 and CheW3 interact with CheA2, αCheW1 and αCheW3 were used in the co-IP assays, and the co-IP samples were probed with αCheA2. As expected, CheA2 was detected in the co-precipitated products from the wild type but not from the ΔW1 and ΔW3 mutants (Figure 7D). Collectively, the results of the co-IP assays show that CheW1 and CheW3 interact with CheA2 but not with CheA1, whereas CheW2 interacts with CheA1 but not with CheA2.
As a coupling protein, CheWEc has four known activities: binding to CheA, binding to MCPs, promoting formation of MCP-CheW-CheA ternary complexes and chemoreceptor arrays, and enabling MCPs to modulate CheA autokinase activity (Gegner et al., 1992;Cardozo et al., 2010;Liu and Parkinson, 1989). In this report, a comprehensive approach has been applied to investigate the roles of the products of the three cheW genes in B. burgdorferi. The results indicate that CheW1 and CheW3 play a similar role as the CheW of E. coli, because the ΔW1 and ΔW3 mutants showed an altered swimming behavior (Table 1 and Videos 3 & 4) and failed to respond to attractant stimuli (Figure 4D & F). Also, the IFA and cryo-ET studies showed that these two mutants are unable to assemble intact chemoreceptor arrays at the cell poles of B. burgdorferi (Figures 5 & 6). In contrast to ΔW1 and ΔW3, the ΔW2 mutant behaved like the wild type with respect to chemotactic response to attractants (Figure 4E), swimming behavior (Table 1), and chemoreceptor assembly (Figure 5C & Figure 6B). Collectively, these results indicate that CheW1 and CheW3 are essential for the chemotaxis of B. burgdorferi, whereas CheW2 is dispensable for the chemotaxis under the tested in vitro conditions. Consistent with this proposition, the homology modeling analysis predicts that CheW2 shares the least structural similarity to CheWEc and CheWTm (Figure 2). It is noteworthy to point out that CheW2 has a long loop insertion near the binding interface of CheW and CheA (Figure 2B). This insertion may disrupt the local environment of the CheA binding surface and consequently prevent CheW2 from interacting effectively with CheA2, a histidine kinase that is essential for chemotaxis of B. burgdorferi (Li et al., 2002;Sze et al., 2012).
IFA and cryo-ET assays demonstrate that CheW3 plays a more important role than CheW1 in the assembly of chemoreceptor arrays at the cell poles of B. burgdorferi. The IFA results showed that the polar-localized chemoreceptor arrays were completely disrupted in ΔW3 cells(Figure 5D & Figure S5), nor did cryo-ET analyses find any array-like structures in the mutant (Figure 6C, Table 2). The observed phenotype of the ΔW3 mutant is very similar to that of an E. coli cheW mutant (Zhang et al., 2004; Sourjik and Berg, 2000;Maddock and Shapiro, 1993). Unlike the situation in ΔW3, IFA still detected weak polar localized signals in ΔW1 cells (Figure 5B), and arrays could still be observed by cryo-ET in a small portion of the ΔW1 cells (Figure 6E). The average length of the chemoreceptor arrays observed in the ΔW1 cells was approximately two fold less than those in the wild type and its complemented strain, ΔW1+ (Table 2). Recent cryo-ET studies of E. coli MCPs show that the basal plates of the arrays consist primarily of CheA and CheW (Briegel et al., 2009;Briegel et al., 2012;Liu et al., 2012). Thus, it is conceivable that CheW1 contributes to the stability of the basal plates but is not essential for their formation. Approximately one half of the ΔW1 cells swim smoothly, and the other half still reverse but with a lower frequency than wild type (Video 4 and Table 1). The observed heterogeneous phenotype of the ΔW1 mutant is not due to genetic heterogeneity because when the mutation was recloned, the same mixed phenotype of the original ΔW1 mutant was observed. Moreover, genetic complementation totally restored the wild-type phenotype (Video 4A and Table 1).
In E. coli, CheW tethers CheA to the MCPs and affects the activity of CheA, which in turn controls the level of CheY-P. The inactivation of cheW completely blocks production of CheY-P. Thus, the flagellar motors are locked in CCW rotation, and a cheW mutant constantly runs (Gegner et al., 1992;Wadhams and Armitage, 2004;Hazelbauer et al., 2008). Our previous studies show that, of the two CheAs and three CheYs of B. burgdorferi, only CheA2 and CheY3 are involved in chemotaxis (Li et al., 2002; Motaleb et al., 2011b). The cheA2 and cheY3 mutants are smooth swimming and non-chemotactic, suggesting that CheY3-P directly controls the rotation of flagellar motors. Because the chemoreceptor arrays in the ΔW1 cells are only partially disrupted (Figure 5B and Figure 6E), it is possible that CheW1 plays an auxiliary role in coupling CheA2 to the MCPs.The decrease in CheY3-P associated with the reduced coupling of CheA2 results in a baseline concentration that spans the threshold required to elicit reversals.
The results presented here raise the possibility that B. burgdorferi may have two different chemosensory pathways (Li et al., 2002;Motaleb et al., 2011b;Charon and Goldstein, 2002). Among the multiple homologues of cheA, cheW, and cheY, only cheA2, cheY3, cheW1, and cheW3 are essential for chemotaxis in vitro. Other than cheW1, all of the essential che genes are located within the flaA operon (cheA2, cheW3, cheX, and cheY3), whereas most of the che genes that are dispensable for chemotaxis reside within anoperon that contains cheA1, cheY2, and cheW2 (Charon and Goldstein, 2002;Ge and Charon, 1997;Li et al., 2002;Fraser et al., 1997). Co-IP assays demonstrated that CheW1 and CheW3 interact with CheA2, whereas CheW2 binds CheA1. Thus, we favor the idea that B. burgdorferi has two chemosensory pathways: CheW1/CheW3-CheA2-CheY3 form the pathway that is essential for chemotaxis under the conditions usually used in vitro, and CheW2-CheA1-CheY2 and/or CheY1 form another pathway that may be used only under other conditions that have yet to be duplicated in the laboratory.
Why might B. burgdorferi have two chemosensory pathways? In nature, B. burgdorferi is maintained via an enzootic cycle comprising both mammalian hosts and an Ixodes tick vector [for recent reviews, see (Radolf et al., 2012;Samuels, 2011;Rosa et al., 2005;Steere et al., 2004)]. The enzootic cycle begins with the feeding by an uninfected tick on an infected vertebrate. After the feeding, the spirochetes remain in the tick gut throughout the molting process. At the time that the infected tick takes a blood meal on a mammal, the spirochetes begin to multiply and migrate from the tick gut to the salivary glands, from which they are transmitted to a new host, thereby completing the enzootic cycle. To adapt to different hosts and complete its enzootic cycle, B. burgdorferi may need one chemosensory pathway, perhaps represented by CheW1/CheW3-CheA2-CheY3, for chemotaxis in mammalian hosts. The pathway involving CheW2-CheA1-CheY2 and/or CheY1 may be activated in the tick vector and/or during the transmission from tick to mammal. Our recent study of the role of cheA2 in the enzootic cycle of B. burgdorferi (Sze et al., 2012) is consistent with the proposal that the CheW1/CheW3-CheA2-CheY3 pathway is important in the mammalian host. Inactivation of cheA2 decreased the ability of B. burgdorferi to establish infection in mice, but not in ticks. The true function of the CheW2-CheA1-CheY2 and/or CheY1 pathway remains obscure. It could be involved in chemotaxis in the tick vector, perhaps in migration of the spirochetes to the salivary glands, or it may function in a signal transduction pathway that regulates gene expression of B. burgdorferi.
The incorporation of two coupling proteins (CheW1 and CheW3) into one chemosensory pathway is different from the situation in other bacteria that have more than one homologue of CheW, such as Vibrio cholera and Rhodobacter sphaeroides [for recent review, see (Porter et al., 2011;Butler and Camilli, 2005;Alexander et al., 2010;Rao et al., 2008)]. In these organisms, either only one CheW homologue functions as a key coupling protein essential for chemotaxis, or CheW homologues are functionally redundant. For instance, V. cholera has three CheW homologues, but only CheW-1 is required for chemotaxis (Butler et al., 2006). Among the four CheW homologues of R. sphaeroides, CheW2 is essential for chemotaxis and chemoreceptor clustering; and deletions of other three cheWs either have no impact on chemotaxis or only conditionally affect chemotactic responses and chemoreceptor localization (Martin et al., 2001;Hamblin et al., 1997a;Hamblin et al., 1997b). It is intriguing to think that the requirement for two CheW proteins in B. burgdorferi may have to do with the extra task of coordinating flagellar reversals at the two ends of an elongated cell body.
High-passage, avirulent Borrelia burgdorferi sensu stricto strain B31A (wild type) (Bono et al., 2000) and its derivative mutants were grown in BSK-II liquid medium or on semi-solid agar plates at 34°Cin a humidified incubator in the presence of 3.4% CO2, as previously documented (Li et al., 2002). The E. coli strains were grown in LB medium at 37°C with appropriate antibiotics.
The cheW1, cheW2 and cheW3 genes were inactivated by allelic exchange mutagenesis as illustrated in Figure S6. To construct the vector for inactivation of cheW1 (gene locus bb0312; gene length, 531 bp), a 120 bp HindIII fragment was deleted and replaced by a kanamycin-resistance cassette (aphI) (Elias et al., 2003). To construct the vector for inactivation of cheW2 (bb0565; gene length, 543 bp), the aphI cassette was directly inserted into an EcoRV restriction cut site within the gene. To construct the vector for inactivation of cheW3 (bb0670; gene length, 1,401 bp), the entire open reading frame (orf) was deleted and replaced with a promoterless streptomycin resistance marker (aadA1), as recently described (Frank et al., 2003;Motaleb et al., 2011a). The resultant constructs were designated as W1::aphI, W2::aphI, and W3::aadA1 (Figure S6), respectively. The PCR primers for constructing these vectors are listed in Table S1. To knock out the cheW genes, these vectors were first linearized and then separately transformed into B31A competent cells via electroporation as previously reported (Samuels, 1995). Transformants were selected on BSK-II agar plates containing 350 μg/ml kanamycin (for W1::aphI and W2::aphI) or 50 μg/ml streptomycin (for W3::aadA1).
To construct the vector for the complementation of the cheW3 mutant, the entire cheW3 gene and its native promoter (Pami) (Yang and Li, 2009) were first amplified by PCR with two pairs of primers (P17/P18 for Pami; P19/P20 for cheW3). The resultant PCR products were then fused together via PCR using primers P17/P20. The resultant PamicheW3 fragment was first cloned into the pGEM®-T Easy vector (Promega, Madison, WI) and then subcloned into pBSV2G, a shuttle vector of B. burgdorferi that contains a gentamicin-resistance cassette (aacC1) (Elias et al., 2003;Rosa et al., 2005). The final construct was named CheW3/pBSV2G (Figure S6). A similar strategy was used to construct a vector for complementation of the cheW1 mutant (CheW1/pBSV2G) and the vector for complementation of the cheW3 mutant (CheW3N+/pBSV2G) with the N-terminal domain of CheW3 (1–210 amino acids). The PCR primers for constructing the complementation vectors are listed in Table S1.
The entire orfs (without the translation initiation ATG/GTG codon) of cheW1, cheW2 and cheW3 were amplified by PCR (the primers are listed in Table S1). The obtained PCR products were first cloned into the pGEM®-T Easy vector (Promega), and then subcloned into the pQE30 expression vector (Qiagen, Valencia, CA), which encodes an amino-terminal histidine tag. The expression of these three genes was induced using 1 mM isopropyl-β-D-thiogalactoside (IPTG). The recombinant proteins were purified by a nickel agarose column and concentrated in 10 kDa molecular weight cut off Amicon Ultra centrifugal concentrators (Millipore, Billerica, MA). Rats (for rCheW1 and rCheW2) and rabbits (for rCheW3) were immunized with 1 to 5 mg of purified recombinant proteins during a one-month period using standard methods. The obtained polyclonal antisera were further purified using affinity chromatography with the AminoLink Plus Immobilization Kit (Thermo Scientific, Rockford, IL) and eluted as recommended by the manufacturer.
The swimming velocity of B. burgdorferi cells was measured using a computer-based motion tracking system. Swim plate assays were carried out using 0.35% agarose with BSK-IImedium diluted 1:10 with Dulbecco’s phosphate-buffered saline (DPBS, pH 7.5) without divalent cations, as previously documented (Motaleb et al., 2000;Li et al., 2002). The plates were incubated for 3–4 days at 34°C in the presence of 3.4% CO2. Diameters of the swim rings that appeared on the plates were measured and recorded in millimeters (mm). A previously constructed non-motile flaB− mutant (ΔflaB) (Motaleb et al., 2000) was used as a negative control to determine the initial inoculum size. Capillary assays were carried out as previously documented with minor modifications (Li et al., 2002;Bakker et al., 2007). Briefly, B. burgdorferi cells were grown to late-logarithmic-phase (~5–7 × 107 cells/ml) and harvested by low-speed centrifugations (1,800 × g). The harvested cells were then resuspended in the motility buffer (Bakker et al., 2007). Capillary tubes filled with either the attractant (0.1 M N-acetyl-glucosamine [GlcNAc] dissolved in the motility buffer) or only motility buffer (negative control) were sealed and inserted into microcentrifuge tubes containing 200 μl of resuspended cells (7 × 108 cells/ml). After 2 hrs incubation at 34°C in a humidified chamber, the solutions were expelled from the capillary tubes, and the spirochete cells were enumerated using Petroff-Hausser counting chambers under a dark-field microscope. A positive chemotactic response was defined as at least twice as many cells entering the attractant-filled tubes as the buffer-filled tubes. For the swim plate, motion tracking, and capillary assays, results are expressed as means ± standard errors of the means (SEM). The significance of the difference between different strains was evaluated with an unpaired Student t test (P value < 0.01).
Sodium-dodecyl-sulfate polyacrylamide-gel electrophoresis (SDS-PAGE) and immunoblotting using the enhanced chemiluminescent detectionsystem were carried out as described before (Li et al., 2010;Sze et al., 2011). B. burgdorferi cells were grown at 34°C and harvested at early stationary phase (approximately108 cells/ml). The whole cell lysates were prepared by washingcells once in PBS buffer (phosphate-buffered saline, pH 7.5) and then boiling for 5 min in Laemmlisample buffer. The same amount of cell lysates (~10–20 μg) were separated on SDS-PAGE gels and transferred to PVDF membrane (Bio-Rad Laboratories, Hercules, CA). The immunoblots were probed with specific antibodies against various proteins (CheA1, CheA2, CheW1, CheW2, and CheW3) and developed using horseradish peroxidase-coupled secondary antibody with an ECL luminol assay.
The co-IP assay was carried out as previously described (Motaleb et al., 2004). Briefly, 200 ml of the late-logarithmic-phase (~5–7 × 107 cells/ml) B. burgdorferi cultures were harvested by centrifugation and washed twice with PBS buffer containing 5 mM MgCl2. The resultant cell pellets were resuspended in TSEA buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.05% sodium azide, pH 7.5) containing Nonidet P-40 (1%, v/v) and phenylmethylsulfonyl fluoride (50 μg/ml) and then incubated at 37°Cfor 1 hr. After the incubation, the obtained samples were centrifuged (1,600 × g for 30 min, 25°C). The resultant cell pellets were resuspended in the PBS buffer and French pressed followed by centrifugation (15,000 × g for 30 min, 25°C). Approximately 200 μl of the obtained supernatants were incubated with 50 μl of the polyclonal anti-CheAs (αCheA1 and αCheA2) or anti-CheWs (αCheW1, αCheW2, and αCheW3) for 1 hr at 25°C in the presence of 1% bovine serum albumin (BSA). After the incubation, 50 μl of protein A (Calbiochem-Behring Corporation, La Jolla, CA) was added to each sample and further incubated at 25°C for 1 hr. The immunoprecipitates and controls were centrifuged at 1,600 × g at 25°C and washed three times with 1 ml of TSEA buffer containing 0.05% Tween-20. The final pellets were suspended in 100 μl of electrophoresis sample buffer, boiled for 5 min, and briefly centrifuged. For the immunoblots, 10 μl of the supernatants was applied to each lane of SDS-PAGE gels as described above.
IFA and cryo-ET assays were carried out to determine the cellular locations of MCPs in B31A and the three cheW mutants as previously described (Xu et al., 2011). For the IFA, αMCP3, a specific antibody against B. burgdorferi MCP3, was used. For the cryo-ET analysis, freshly prepared B. burgdorferi cultures were deposited onto a glow-discharged holey carbon EM grid, blotted, and rapidly frozen in liquid ethane. The frozen-hydrated specimens were imaged at −170°C using a Polara G2 electron microscope (FEI Company, Hillsboro, Oregon) equipped with a field emission gun and a 4K × 4K CCD camera (TVIPS; GMBH, Germany). The microscope was operated at 300 kV with a magnification of 31,000×. Low-dose single-axis tilt series were collected from each bacterium at −6 μm defocus with a cumulative dose of ~100 e−/Å2 distributed over 65 images with an angular increment of 2°, covering a range from −64° to +64°. The tilt series images were aligned and reconstructed using the IMOD software package (Kremer et al., 1996). In total, cryo tomograms of B31A (30 cells), a cheW1 mutant (31 cells) and its complemented strain (20 cells), a cheW2 mutant (30 cells), a cheW3 mutant (25 cells) and its complemented strain (30 cells) were reconstructed and visualized using IMOD (Kremer et al., 1996).
The NMR structure of T. maritima CheW (Protein Data Bank ID: 1K0S) (Griswold and Dahlquist, 2002) was selected as a template for the homology modeling analysis of CheW1, CheW2, and N-terminus CheW like domain of CheW3. Pairwise sequence alignment of CheW homologues was conducted using Clustal X. Automodel module in Modeller 9v7 (Sali and Blundell, 1993) was applied to obtain the final refined structures. All structures were analyzed and visualized in PyMol (The PyMol Molecular Graphic System, Version 184.108.40.206, Schrodinger, LLC). The qualities of the models were evaluated by PDBsum (Laskowski, 2001).
This research was supported by Public Health Service grants (AI073354 and AI078958) to C. Li, GM072004 to C. Wolgemuth; J. Liu was supported in part by grants AI087946 from the National Institute of Allergy and Infectious Diseases (NIAID) and AU-1714 from the Welch Foundation.