Independence and interdependence of Dif and Frz chemosensory pathways in Myxococcus xanthus chemotaxis

Summary

Dif and Frz, two Myxococcus xanthus chemosensory pathways, are required in phosphatidylethanolamine (PE) chemotaxis for excitation and adaptation, respectively. DifA and FrzCD, the homologs of methyl-accepting chemoreceptors in the two pathways, were examined for methylation in the context of chemotaxis and inter-pathway interactions. Evidence indicates that DifA may not undergo methylation but signals transmitting through DifA do modulate FrzCD methylation. Results also revealed that M. xanthus possesses Dif-dependent and Dif-independent PE sensing mechanisms. Previous studies showed that FrzCD methylation is decreased by negative chemostimuli but increased by attractants such as PE. Results here demonstrate that the Dif-dependent sensory mechanism suppresses the increase in FrzCD methylation in attractant response and elevates FrzCD methylation upon negative stimulation. In other words, FrzCD methylation is governed by opposing forces from Dif-dependent and Dif-independent sensing mechanisms. We propose that the Dif-independent but Frz-dependent PE sensing leads to increases in FrzCD methylation and subsequent adaptation, while the Dif-dependent PE signaling suppresses or diminishes the increase in FrzCD methylation to decelerate or delay adaptation. We contend that these antagonistic interactions are crucial for effective chemotaxis in this gliding bacterium to ensure that adaptation does not occur too quickly relative to the slow speed of M. xanthus movement.

Introduction

The proteobacterium Myxococcus xanthus utilizes its surface motility in both the vegetative and the developmental stages of its life cycle (Zusman et al., 2007). During vegetative growth, this Gram-negative bacterium moves on surfaces to seek nutrients in the form of organic matter or other bacteria as prey (Reichenbach, 1999, Berleman et al., 2006). Under nutrient limitation, M. xanthus initiates a developmental process in which up to 10⁵ cells employ their motility to aggregate and form a multicellular structure called a fruiting body. Approximately 20% of the cells eventually differentiate into myxospores within fruiting bodies (Wireman & Dworkin, 1975, Berleman & Kirby, 2007). The myxospores, which are more resistant than vegetative cells to environmental stress, can germinate and reenter the vegetative cell cycle when conditions become favorable for growth.

The surface motility of M. xanthus, which involves no flagella, is known as gliding (McBride, 2001, Spormann, 1999). It entails two distinct components: the adventurous (A) and the social (S) motility systems. The A-motility engine powers the movement of isolated cells and the S-motility engine functions only when cells are in close proximity or in cell groups (Hodgkin & Kaiser, 1979b, Hodgkin & Kaiser, 1977, Hodgkin & Kaiser, 1979a). The two motility systems, which appear to function synergistically to enable surface translocation of M. xanthus (Kaiser & Crosby, 1983, Spormann & Kaiser, 1999), are essential for both vegetative swarming and developmental aggregation. It has been proposed that a response to self-generated chemoattractants is at the core of developmental aggregation (Kearns et al., 2001, Shi et al., 1993).

Frz and Dif systems, both indispensable for developmental aggregation as well as vegetative swarming, are the most extensively studied among the eight chemosensory pathways in M. xanthus (Zusman et al., 2007). The Frz pathway controls the reversal frequency of moving M. xanthus cells (Blackhart & Zusman, 1985, McBride et al., 1992) and mediates tactic or behavioral responses to various chemostimuli including certain phosphatidylethanolamine (PE) species (Shi et al., 1993, Kearns & Shimkets, 1998, Xu et al., 2007). Cells initially decrease their reversal frequency upon PE exposure but eventually adapt and return to the prestimulus level of reversal. The initial suppression of cell reversal by the PE attractants in M. xanthus has been referred to as the excitation response (Bonner & Shimkets, 2006, Kearns & Shimkets, 2001), and this convention will be followed here as well. Although the Dif pathway plays a central role in regulating the production of exopolysaccharide (EPS) (Yang et al., 1998b, Black & Yang, 2004, Bellenger et al., 2002), it is also involved in the regulation of cell reversal in response to PE. Previous results suggested that the Dif system is essential for excitation while the Frz system is essential for adaptation in the response to PE (Kearns & Shimkets, 1998, Bonner et al., 2005). Therefore, the functions of the Dif and the Frz chemosensory pathways must converge at the regulation of cell reversal behavior (Kearns & Shimkets, 2001).

A hallmark of bacterial chemotaxis regulation is the methylation of chemoreceptors (Falke et al., 1997, Szurmant & Ordal, 2004). These chemoreceptors, commonly known as methyl-accepting chemotaxis proteins or MCPs, share high similarities in their methylation and signaling domains at the C-termini. The methylation of chemoreceptors is crucial for chemotactic adaptation which enables bacteria to respond to concentration differences instead of absolute concentrations of chemoeffectors. The two MCP homologs in the Frz and the Dif systems are FrzCD and DifA, respectively (McBride et al., 1989, Yang et al., 1998b). The cytoplasmic FrzCD is known to be methylated (McCleary et al., 1990, McBride et al., 1992). FrzCD methylation is likely critical in chemotactic adaptation because it is modulated by chemicals that affect the motility behavior of M. xanthus (McBride et al., 1992, Kearns & Shimkets, 1998). DifA, likely a transmembrane protein (Yang et al., 1998b), has not been examined for methylation. We are interested in understanding how the Dif and the Frz pathways interact by studies of DifA and FrzCD methylation. Here we present evidence that DifA is unlikely to be methylated. More importantly, there are apparently two mechanisms for sensing PE in M. xanthus: one is Dif-dependent and the other Dif-independent. The two sensing mechanisms interact antagonistically in the regulation of FrzCD methylation. We suggest that the dual regulation of FrzCD methylation is to correlate chemotactic adaptation with the slow speed of M. xanthus movement.

Results

DifA may not undergo methylation

First, the methylation of DifA, the MCP homolog in the Dif pathway, was examined for its involvement in EPS production. It has been shown that type IV pili (TFP) function upstream of the Dif pathway in EPS regulation. Mutations in pilA, which encodes the pilin subunit of the pilus, lead to a TFP^- and EPS^- phenotype (Black et al., 2006). DifA methylation in a ΔpilA mutant (YZ690) was analyzed in comparison with the wild type (DK1622) by immunoblotting (see Experimental procedures). It is known that MCP methylation, which occurs on glutamates, is base labile (Van Der Werf & Koshland, 1977). As shown in Figure 1A, DifA appeared as a single band in both the ΔpilA mutant and the wild type regardless of NaOH treatment. This initial experiment was performed in a nutrient medium (CYE) in which pilA deletion eliminated EPS production (Black et al., 2006). Nutrient deprivation is also known to upregulate EPS production (Shimkets, 1986, Behmlander & Dworkin, 1991, Bonner & Shimkets, 2006). DifA methylation in the wild type was additionally examined under different nutrient conditions (see Experimental procedures). Again, the banding pattern of DifA looked the same in MOPS buffer (starvation condition) as in CYE (rich nutrient medium) with no signs of methylation (Fig. 1A). As controls for the methylation assays, FrzCD in the same samples as those in Figure 1A displayed multiple bands as clear indications of methylation (Fig. 1B). In addition, NaOH treatment abolished methylated forms of FrzCD (faster migrating bands) as expected (McBride et al., 1992, McCleary et al., 1990). These results (Fig. 1A and 1B) suggested that DifA may not undergo methylation during the regulation of EPS production.

Figure 1.

Analysis of receptor methylation in EPS regulation. Lysates were prepared from WT and ΔpilA cells incubated in rich medium (CYE) or starvation buffer (MOPS). Cell lysates were treated with (+) or without (-) 0.2 M NaOH (OH^-) before analysis by immunoblotting using αDifA (A) or αFrzCD (B) antibodies. Refer to Experimental procedures for details. Strains: WT, DK1622; ΔpilA, YZ690.

Next, the involvement of DifA methylation during chemotaxis was investigated. Methylation of MCPs is usually associated with adaptation in chemotaxis (Falke et al., 1997, Szurmant & Ordal, 2004). Since EPS production is not chemotaxis, lack of DifA methylation in EPS regulation may not be too surprising. On the contrary, DifA and the Dif pathway are involved in the taxis of M. xanthus towards certain PE molecules and adaptation is part of the response (Kearns et al., 2000, Bonner et al., 2005). The possible methylation of DifA in response to 12:0 PE, one of the known attractants for M. xanthus, was examined using similar assays as in Figure 1A. Again, DifA appeared as a single band, regardless of the presence of PE, alkaline treatment or nutrient conditions (Fig. 2A). These results indicated that DifA may not be methylated even during the regulation of chemotaxis in M. xanthus.

Figure 2.

Analysis of DifA methylation in chemotactic responses. Lysates were prepared from cells incubated in rich medium (CTT) or starvation buffer (TPM) with (+) or without (-) 12:0 PE (A) or nitrate (B). Cell lysates were treated with (+) or without (-) 0.2 M NaOH (OH^-) before analysis by immunoblotting using αDifA antibodies. Strains: WT, DK1622; ΔdifA/nafA, YZ724. See Fig. S2 for a more complete version of Fig. 2B.

In addition, we took advantage of a NarX-DifA (NafA) chimera (Xu et al., 2005) to examine the presence of DifA methylation. The primary reason for performing experiments with NafA is because the exact function of DifA in either EPS regulation or PE taxis is yet to be elucidated. It may be argued that the inability to detect DifA methylation in the experiments in Figures 1A and 2A was because DifA is not a direct signal transducer under those experimental conditions. The NafA chimera contains the sensory module of NarX (the nitrate sensor kinase from E. coli) and the signaling module of DifA; it was shown to regulate EPS production and modulate cellular reversal frequency in M. xanthus in a nitrate-dependent manner (Xu et al., 2005, Bonner et al., 2005). This NafA chimera, which directly signals through the Dif pathway (Xu et al., 2005), contains the predicted methylation domains of DifA (Yang et al., 1998b, Bonner et al., 2005) and is recognized by anti-DifA polyclonal antibodies (Xu et al., 2005). As shown in Figure 2B (also see Fig. S2), NafA showed no mobility change indicative of methylation in response to nitrate and alkaline treatment. These results indicated that DifA is not modified by methylation even when it is directly involved in signal perception and transmission.

Finally, we examined the effects of mutations of potential methylation sites on the electrophoretic properties of DifA. All five glutamate (E) residues identified as possible methylation sites (Bonner et al., 2005) were mutated to aspartate (D). In addition, glutamine 346 (Q346), which could be deamidated to become a glutamate, was changed to either an aspartate or alanine (A). As show in Fig. 3, DifA with these amino acid substitutions showed the same electrophoretic mobility as the wild-type protein. If any of these six residues were deamidated and/or methylated as in classical MCPs, a change in mobility would have been observed. The results in Figures 1, 2 and 3 collectively led to the conclusion that DifA is unlikely a methylated protein despite its homology to bacterial MCPs.

Figure 3.

Effect of amino acid substitutions on electrophoretic mobility of DifA. Lysates from cells incubated either in the nutrient-rich medium CYE (upper panel) or the MOPS starvation buffer (lower panel) were analyzed by immunoblotting using αDifA antibodies. The strains all have the original difA gene deleted at the dif locus. They either express the wild-type difA (WT) or difA with indicated mutations at the MX8 attachment site. Strains: ΔdifA/difA (WT), YZ619; ΔdifA/difA(E110D), YZ712; ΔdifA/difA(E135D), YZ707; ΔdifA/difA(Q346D), YZ713; ΔdifA/difA(Q346A), YZ746; ΔdifA/difA(E352D), YZ714; ΔdifA/difA(E359D), YZ715; ΔdifA/difA(E380D), YZ708.

Signaling through the Dif pathway modulates FrzCD methylation

The NarX-DifA (NafA) chimera and nitrate (Xu et al., 2005, Bonner et al., 2005) were utilized to examine the modulation of FrzCD methylation by signaling through the Dif pathway. Previous studies indicated that the tactic response of M. xanthus to PE involves both the Frz and the Dif chemosensory pathways: the Dif pathway is crucial for the perception of or excitation to PE and the Frz pathway is essential for temporal adaptation (Kearns & Shimkets, 1998). It is plausible that signal input through the Dif pathway may modulate the methylation of FrzCD to bring about adaptation. Nitrate had no effect on the electropheretic mobility of FrzCD in the wild-type strain as expected (Fig. 4); on the other hand, nitrate clearly increased the methylation of FrzCD in YZ724, a difA deletion mutant containing the NafA chimera (ΔdifA/nafA). This indicated that there is crosstalk from the Dif pathway to the Frz pathway detectable at the level of FrzCD methylation.

Figure 4.

Influence of signaling through DifA on FrzCD methylation. Cells with or without the NarX-DifA (NafA) chimera were incubated in CTT liquid medium for 3 hours with (+) or without (-) 700 μM of nitrate. Whole cell lysates were analyzed by immunoblotting using αFrzCD antibodies. Strains: WT, DK1622; ΔdifA Δpgi, YZ1010; ΔdifA Δpgi/nafA, YZ1012.

It is known that nitrate also results in EPS production in YZ724 (ΔdifA/nafA) (Xu et al., 2005). This above crosstalk from NafA to FrzCD could be the result of direct signaling through DifA or through EPS production. pgi, a gene essential for EPS biosynthesis in M. xanthus (Cadieux and Yang, unpublished data), was deleted from the ΔdifA/nafA mutant. The resulting ΔdifA Δpgi/nafA mutant (YZ1012), which produces no EPS with or without nitrate (Cadieux and Yang, unpublished data), displayed similar changes in FrzCD methylation in response to nitrate as its parental strain YZ724 (ΔdifA/nafA) (Fig. 4). These results demonstrated that it is the signaling through NafA and the Dif pathway, not EPS production per se, that leads to increased FrzCD methylation.

Behavioral analysis indicated that nitrate functions as a repellent that signals through NafA and the Dif pathway (Bonner et al., 2005). One complicating factor in the previous behavior assay (Bonner et al., 2005) was that nitrate also stimulates the ΔdifA/nafA strain to produce EPS (Xu et al., 2005). The lack of EPS is known to decrease the reversal frequency of M. xanthus cells (Kearns et al., 2000). To verify that nitrate is a bona fide repellent for NafA-expressing cells, the ΔdifA Δpgi/nafA (YZ1012) strain, which produces no EPS with or without nitrate (Cadieux and Yang, unpublished data), was examined for its behavioral response to nitrate. Nitrate at 100, 350 and 1000 μM clearly reduced the colony expansion of YZ1012 but had no effect on that of the ΔdifA Δpgi and the wild-type strains (Fig. S1). This is consistent with nitrate being a repellent mediated by NafA, as repellents are known to reduce the colony expansion of M. xanthus due to its lack of adaptation to negative stimuli (Xu et al., 2007). Isolated YZ1012 cells also showed increased cell reversal in the presence of nitrate (data not shown), supporting the notion that nitrate functions as a repellent for this strain. These observations (Fig. 4 and S1, Bonner et al., 2005) indicated that there is a Dif-directed modulation of FrzCD methylation and that a repellent that signals through Dif leads to increased FrzCD methylation.

Attractant PE can be sensed independently of Dif-mediated excitation pathway

PE attractants are known to be sensed and transduced through Dif for the elicitation of an excitation response and Frz is proposed to be responsible for adaptation. The demonstration of Dif-directed FrzCD methylation (Fig. 4) suggested that perception of PE by the Dif pathway may first lead to Dif-dependent changes in cell reversal followed by adaptation due to increased FrzCD methylation. As expected, FrzCD methylation increased in response to PE in the wild type (Fig. 5A). However, increased FrzCD methylation was also readily detected in a difA mutant upon exposure to 12:0 PE (Fig. 5A). This suggests that there is a Dif-independent sensing mechanism for PE in M. xanthus for the modulation of FrzCD methylation.

Figure 5.

FrzCD methylation upon treatment with 12:0 PE. Cells were incubated in CTT liquid medium with (+) or without (-)12:0 PE. Whole cell lysates were probed with αFrzCD antibodies in immunoblotting. (A) Strains: WT, DK1622; ΔdifA, YZ601; ΔpilA, YZ690; ΔsglK, YZ811; fibA, LS2200. (B) Strains: WT, DK1622; ΔdifA, YZ601; ΔdifC, SW403; ΔdifE, YZ603; ΔdifB, YZ602; ΔdifD, YZ613; ΔdifG, YZ604.

The EPS matrix and its associated protein FibA are also known to be part of the PE excitation pathway upstream of the Dif proteins (Kearns et al., 2002, Kearns et al., 2000). The above observation with the difA mutant (Fig. 5A) would predict that other EPS^- and fibA mutants would behave similarly as the difA mutants in response to PE at the level of FrzCD methylation. ΔpilA (YZ690) and ΔsglK (YZ811) mutants, both EPS^- (Weimer et al., 1998, Yang et al., 1998a, Black et al., 2006), as well as a fibA mutant (LS2200) were examined for FrzCD methylation upon PE treatment (Fig. 5A). FrzCD in these three mutants was also readily methylated after PE treatment. These results demonstrate that there is a PE sensing mechanism that is independent of the Dif-mediated excitation pathway.

PE signaling through Dif suppresses the increase in FrzCD methylation

The results in Figure 5A additionally showed that mutants without the Dif-dependent excitation pathway increased FrzCD methylation even further than the wild type upon PE exposure. The demethylated form of FrzCD in the wild type (Fig. 5A), represented by the slowest migrating species in SDS-PAGE, is not detected in ΔdifA, ΔpilA, ΔsglK and fibA mutants. Similarly, PE treatment resulted in more FrzCD methylation in ΔdifC and ΔdifE mutants than in the wild type (Fig. 5B). Because DifA, DifC and DifE are all essential Dif components for 12:0 PE signaling, and because ΔpilA, ΔsglK and fibA mutants lack components upstream of Dif in the PE excitation pathway, these results (Fig. 5) demonstrated that mutants defective in Dif-dependent PE sensing increases their FrzCD methylation even further than the wild type after PE treatment. It can be concluded therefore that in attractant responses such as PE taxis (Fig. 5), Dif-dependent signaling suppresses the increases in FrzCD methylation whereas a repellent signaling through Dif, as is the case with nitrate and NafA, elevates or increases FrzCD methylation (Fig. 4 and 5).

Dif-dependent modulation of FrzCD requires DifA, DifC and DifE, but not DifB, DifD or DifG

Mutations in all dif genes (difB, difC, difD, difE and difG) were examined for the requirement of Dif-directed FrzCD modulation (Fig. 5B). In the presence of PE, FrzCD is similarly methylated in ΔdifB, ΔdifD and ΔdifG mutants as in the wild type. On the other hand, in ΔdifA, ΔdifC and ΔdifE mutants, FrzCD is consistently more methylated than in the wild type upon exposure to PE. This indicates that in PE taxis, difA, difC and difE, but not difB, difD or difG, are required for the excitation- or Dif-directed suppression of FrzCD methylation. Similar experiments were performed to examine the requirement for NafA-directed modulation of FrzCD methylation. As shown in Fig. 6, the deletion of difB, difD and difG had no effect on NafA-directed FrzCD methylation upon nitrate treatment. In contrast, such crosstalk from NafA to FrzCD was no longer detectable when difC or difE was deleted. The results in Figures 5 and 6 showed that Dif-directed modulation of FrzCD methylation requires DifA, DifC and DifE, but not DifB, DifD or DifG. It is known that DifA, DifC and DifE, which are homologous to MCP, CheW and CheA, likely form a ternary signaling complex in M. xanthus (Yang & Li, 2005). The results here suggest that it is the DifACE signaling complex that is essential for the apparent crosstalk from Dif to FrzCD.

Figure 6.

Requirement for NafA-directed modulation of FrzCD methylation. Experiment was performed as in Fig. 4. Strains: WT, DK1622; ΔdifA/nafA, YZ724; ΔdifA Δpgi/nafA, YZ1012; ΔdifA ΔdifC/nafA, YZ730; ΔdifA ΔdifE/nafA, YZ732; ΔdifA ΔdifB/nafA, YZ754; ΔdifA ΔdifD/nafA, YZ731; ΔdifA ΔdifG/nafA, YZ733.

Discussion

In this study, we examined the methylation of DifA and FrzCD, two MCP homologs from the M. xanthus Dif and Frz chemosensory pathways, respectively. No DifA methylation could be detected in various genetic backgrounds or under different environmental conditions (Figs 1-3, S2). In contrast, FrzCD methylation was readily detected as expected (Fig. 1B; McBride et al., 1992). One important discovery from this study is the crosstalk between Dif and Frz since signaling through the Dif pathway can modulate FrzCD methylation. Interestingly, 12:0 PE, an attractant known to require Dif to elicit behavioral responses, increased FrzCD methylation when the Dif pathway was genetically inactivated (Fig. 5B). This reveals the existence of Dif-dependent and Dif-independent PE sensing in M. xanthus. Results (Figs 4-6, S1) here also demonstrated an attractant reduces whereas a repellent elevates FrzCD methylation when signaling through the Dif pathway. This is in contrast to the overall effects of chemoeffectors on FrzCD methylation, as attractants increase and repellents decrease FrzCD methylation in general (McBride et al., 1992, Zusman et al., 2007). Last but not least, Dif-directed modulation of FrzCD methylation was found to require DifA, DifC and DifE, but not DifB, DifD or DifG.

FrzCD methylation in chemotactic adaptation and two PE sensing mechanisms in M. xanthus

Modulation of FrzCD methylation is proposed to regulate chemotactic adaptation in M. xanthus (Zusman et al., 2007). Isoamyl alcohol (IAA), dimethylsulphoxide (DMSO), and spent media, all potential repellents for M. xanthus, decrease FrzCD methylation (McBride et al., 1992). Both PE and nutrients, attractants for M. xanthus, increase FrzCD methylation (McBride et al., 1992). In PE taxis, the Frz proteins constitute a pathway essential for adaptation but not excitation (Kearns & Shimkets, 1998). The observation that PE led to increased FrzCD methylation in various strains (Fig. 5) is consistent with this model and supports the notion that increases in FrzCD methylation are directly involved in adaptation to PE as attractants.

Remarkably, dif, pil, sglK and fibA mutants, all deficient in sensing and transducing PE for the excitation response (Fig. 5; Kearns and Shimkets, 1998; Kearns et al., 2000), showed increased FrzCD methylation in the presence of PE (Fig. 5). In other bacteria, chemotactic adaptation is always the consequence of excitation and shows an absolute dependency on the sensing mechanism for excitation (Falke et al., 1997, Szurmant & Ordal, 2004). In M. xanthus, however, there are apparently two independent sensing mechanisms for PE, one for the excitation pathway involving Dif and the other for the Frz-mediated adaptation response (Fig. 7). The existence of Dif-dependent and Dif-independent sensory mechanisms for PE may attest to the importance of PE responses in the biology of M. xanthus. Shimkets, Kearns and colleagues (Kearns & Shimkets, 1998, Kearns et al., 2001, Kearns et al., 2000, Bonner et al., 2005) have proposed that PE taxis plays roles in the recognition of prey organisms as non-self during vegetative growth and in self-recognition during fruiting body development to facilitate aggregation.

Figure 7.

Working model for Dif and Frz pathways in PE taxis. Dif components (in gray) constitute the excitation pathway and Frz proteins (in white) form the adaptation pathway. Solid lines depict known biochemical interactions and/or direct contacts whereas dashed lines indicate indirect or inferred interactions; arrows indicate positive or stimulatory effects and bars represent negative or inhibitory effects in the respective pathways in PE response only. 12:0 PE (and possibly 16:1 PE) molecules are perceived independently by two MCP homologs, DifA and FrzCD, either directly or indirectly. The signal sensed by DifA is transduced through DifC, DifE and DifD to inhibit cellular reversal (excitation pathway). The PE signal sensed by FrzCD leads to increased FrzCD methylation by unknown mechanisms. Methylated FrzCD signals through FrzA/B, FrzE and FrzZ to increase cellular reversal (adaptation pathway). In addition, signals from the DifACE complex counteract the PE-induced increases in FrzCD methylation to possibly guarantee that adaptation does not occur too quickly for the slow moving M. xanthus cells. The homologies of Dif and Frz proteins to bacterial chemotaxis proteins are as follows: DifA and FrzCD to MCP; DifC, FrzA and FrzB to CheW; DifE to CheA; FrzE to CheA-CheY hybrid; DifD to CheY; FrzZ to CheY-CheY hybrid.

Roles of Dif-dependent regulation of FrzCD methylation

Despite the two independent sensing mechanisms for PE, the Dif and the Frz pathways do communicate with each other as signaling through the Dif pathway clearly influences FrzCD methylation. As shown in Figure 4, nitrate, a repellent-like signal mediated by the NarX-DifA chimera (Fig. S1; Bonner et al., 2005), increased FrzCD methylation. In contrast, the attractant 12:0 PE suppressed FrzCD methylation when the signal is relayed through DifA, DifC and DifE, i.e., FrzCD is less methylated in the wild type than in the ΔdifA, ΔdifC or ΔdifE mutants after PE treatment (Fig. 5B). Mutations in components upstream of Dif in PE taxis (Kearns et al., 2000) also showed similar effects as these Δdif mutations on FrzCD methylation (Fig. 5A). These results indicate interdependence between the Dif and the Frz pathways despite their ability to sense the PE signal independently. The signal only appears to be transmitted in the direction of Dif to Frz but not the reverse since there are no clear quantitative differences between the wild type and various frz mutants in EPS production which is regulated by the Dif pathway (data not shown). The signal originating from the Dif pathway must branch three ways after the DifACE signaling complex. One branch regulates EPS through unknown mechanisms (Black & Yang, 2004). The remaining two regulate motility and taxis (Fig. 7): one modulates the gliding motors through DifD (Bonner et al., 2005), and the other communicates with the Frz pathway by modulating FrzCD methylation.

What is the function of the DifACE- or the excitation-dependent modulation of FrzCD methylation? As indicated earlier, an attractant (12:0 PE) decreases and a repellent (nitrate and NafA) increases FrzCD methylation when they signal through the Dif pathway (Figs 4, 5 and 6). One possibility is that DifACE-dependent modulation of the Frz pathway is meant to augment the excitation response to the signal from the Dif pathway. It is well known that the Frz system plays a primary role in regulating cell motility in M. xanthus. Although the Dif signal transduction pathway is able to affect cell reversal independently of the Frz pathway (Kearns & Shimkets, 1998, Bonner et al., 2005), Dif is the primary system that regulates the production of EPS (Yang et al., 2000, Black & Yang, 2004, Bellenger et al., 2002). As such, the Dif pathway may require assistance from the Frz system to mount an effective excitation response for PE taxis. Indeed, a frzCD^c mutant, which is constitutively hyperreversal (Blackhart & Zusman, 1985), is unresponsive to PE (Kearns & Shimkets, 1998). This suggests that the signal from the Frz pathway can overwhelm those from Dif in the regulation of cell reversal. Interestingly, the magnitude of reversal suppression by PE is similar in the wild type as in a frzCD null mutant (Kearns & Shimkets, 1998). This indicates that the Dif system may effectively regulate the reversal behaviors of single cells independently of its effect on FrzCD methylation.

Alternatively, the DifACE-dependent modulation of FrzCD methylation could be to manipulate the adaptation response instead of augmenting the excitation response. Such manipulation could be crucial for proper chemotactic responses in M. xanthus with slow motility and an excitation-independent sensing mechanism for adaptation. Temporal sensing is at the heart of bacterial chemotaxis and the timing of adaptation is of ultimate importance to effective chemotaxis (Macnab & Koshland, 1972, Lovdok et al., 2007, Bray et al., 2007). The rate or speed of adaptation has to match the speed of bacterial movement. Too quick of an adaptation would nullify excitation since cells would adapt before mounting a sufficient response. Too slow of an adaptation would lead cells away from desirable territories or trap them in unfavorable environments, since the cells would either keep moving in one direction or incessantly change directions with little net movement. M. xanthus gliding is quite slow, about three orders of magnitude slower than swimming bacteria (Spormann, 1999, Zusman et al., 2007). Adapting too quickly relative to its slow speed would lead to ineffective taxis at best. Depending on the strength of stimulation, adaptation in flagellated bacteria takes from seconds to minutes (Brown & Berg, 1974, Macnab & Koshland, 1972, Goldman & Ordal, 1981). In comparison, it takes M. xanthus about one and half hours to adapt to PE or other attractants (Kearns and Shimkets, 1998; Xu et al., 2007). There is evidence that the timing of adaptation in M. xanthus is regulated in response to PE by both Frz and Dif proteins: it took a frzE mutant more and a difB mutant less time to adapt to PE in comparison with the wild type (Bonner et al., 2005, Kearns & Shimkets, 1998).

Fig. 7 presents our current model on the regulation of PE taxis. Signaling through the Dif and the Frz pathways in PE taxis is independent as well as interdependent. DifA and FrzCD, the MCP homologs of the Dif and the Frz pathways, can sense 12:0 PE independently of each other. The Dif-dependent signaling mainly activates the excitation (Dif) pathway while the Dif-independent PE-sensing, by virtue of increased FrzCD methylation, leads to adaptation through the Frz pathway. On the other hand, the Frz-mediated adaptation is partially dependent on or modulated by the Dif pathway. Specifically, PE signaling through the DifACE complex decreases or delays FrzCD methylation possibly to ensure that adaptation does not occur too quickly. This is also somewhat reminiscent of the roles proposed for the phosphorylation and dephosphorylation of CheV (a CheW-CheY hybrid) in B. subtilis chemotaxis (Karatan et al., 2001). It was suggested that the slow phosphorylation of CheV and the increased stability of CheV-P is to allow an excitatory signal enough time for a significant period of response before adaptation. In other words, the DifACE-directed modulation of the Frz pathway may be to change the timing of adaptation in M. xanthus chemotaxis. It is noted that this model needs further experimental validation and it is unclear at the present how the Dif pathway regulates FrzCD methylation at the molecular level.

DifA, an unmethylated “MCP” signal transducer?

DifA does not appear to be methylated despite its homology to methyl-accepting chemoreceptors. This perhaps should not be too surprising since Dif is the only chemosensory pathway in M. xanthus that lacks CheB and CheR homologs, which regulate MCP methylation (Zusman et al., 2007). In contrast, difG, a gene at the dif locus, encodes the only M. xanthus homolog of CheC, a phosphatase of CheY-phosphate (Szurmant et al., 2004). Functionally, the lack of DifA methylation is perhaps pertinent to the role of this protein in the regulation of EPS production (Yang et al., 1998b, Yang et al., 2000). In bacterial chemotaxis, temporal adaptation and therefore modulation of receptor methylation is essential. As such, the output of a chemotaxis pathway is regulated, not by absolute concentration, but by the change in the concentration of chemoeffectors. The end result is that even in the presence of a high concentration of a chemoeffector, the activity of the CheA kinase is kept at a basal level as long as the concentration stays uniform spatially and temporally. A change in concentration will result in a change in CheA activity only temporarily. In contrast, in M. xanthus EPS regulation, a response to a signal of a given strength must be sustained instead of temporary. If EPS were to be regulated by a chemosensory pathway capable of temporal adaptation, a stimulatory signal must increase its strength constantly to maintain the same level of EPS production as the output. A chemotaxis-type adaptation would only result in a spurt of EPS production which would hardly be sufficient for the critical role of EPS in S motility and fruiting body development. It appears more plausible therefore that EPS production, once stimulated by a signal of a given strength, will increase steadily to reach a plateau corresponding to the signal strength. In this context, it makes more biological sense for DifA not to be methylated.

Experimental procedures

Strains, plasmids and growth conditions

The M. xanthus strains used in this study are listed in Table 1. M. xanthus was grown at 32°C using CYE (1% Casitone, 0.5% yeast extract, 0.1% MgSO₄·7H₂O, 10 mM MOPS, pH 7.6) (Campos & Zusman, 1975) or CTT (1% casitone, 8 mM MgSO₄, 10 mM Tris·HCl, 1 mM K₂HPO₄-KH₂PO₄, pH 7.6) (Kaiser, 1979) media. The E. coli strain XL1-Blue (Stratagene), used for routine cloning and plasmid construction, was grown at 37°C using Luria-Bertani (LB) media (Miller, 1972). Liquid cultures were grown on a rotary shaker at 300 rotations per min (rpm). Plates contained 1.5% agar unless noted otherwise. Whenever applicable, media were supplemented with kanamycin at 100 μg/ml.

Table 1.

M. xanthus strains

Strains	Genotype or description	Source or reference
DK1622	Wild type	(Kaiser, 1979)
LS2200	fibA	(Kearns et al., 2002)
SW403	ΔdifC	(Bellenger et al., 2002)
YZ601	ΔdifA	(Xu et al., 2005)
YZ602	ΔdifB	(Black & Yang, 2004)
YZ603	ΔdifE	(Black & Yang, 2004)
YZ604	ΔdifG	(Black & Yang, 2004)
YZ613	ΔdifD	(Black & Yang, 2004)
YZ619	ΔdifA/difA	(Bonner et al., 2005)
YZ690	ΔpilA	This study
YZ707	ΔdifA/difA(E135D)	(Bonner et al., 2005)
YZ708	ΔdifA/difA(E380D)	(Bonner et al., 2005)
YZ712	ΔdifA/difA(E110D)	(Bonner et al., 2005)
YZ713	ΔdifA/difA(Q346D)	(Bonner et al., 2005)
YZ714	ΔdifA/difA(E352D)	(Bonner et al., 2005)
YZ715	ΔdifA/difA(E359D)	(Bonner et al., 2005)
YZ746	ΔdifA/difA(Q346A)	This study
YZ724	ΔdifA/nafA	(Xu et al., 2005)
YZ730	ΔdifA ΔdifC/nafA	(Xu et al., 2005)
YZ731	ΔdifA ΔdifD/nafA	(Xu et al., 2005)
YZ732	ΔdifA ΔdifE/nafA	(Xu et al., 2005)
YZ733	ΔdifA ΔdifG/nafA	(Xu et al., 2005)
YZ754	ΔdifA ΔdifB/nafA	This study
YZ811	ΔsglK	This study
YZ1010	ΔdifA Δpgi	This study
YZ1012	ΔdifA Δpgi/nafA	This study

Construction of plasmids and mutants

Mutants with in-frame deletions of pilA, sglK and pgi genes were constructed using a positive-negative kanamycin/galactose (KG) selection cassette (Ueki et al., 1996). DNA fragments with in-frame deletions were generated by a two-step overlap PCR procedure (Sambrook & Russell, 2001) and cloned into pBJ113 (Julien et al., 2000) to generate pWB505 (ΔpilA, deletion of codons 7-218), pLZ406 (ΔsglK, deletion of codons 7-605) and pCC7425 (Δpgi, deletion of codons 41-526). These deletion plasmid constructs were electroporated (Kashefi & Hartzell, 1995) into DK1622 (wild type) and selected on plates with kanamycin. Deletion mutants including YZ690 (ΔpilA), YZ811 (ΔsglK) and YZ1001 (Δpgi) were subsequently identified by their resistance to galactose and sensitivity to kanamycin and further confirmed by PCR. Following the same procedure, a ΔdifA Δpgi mutant (YZ1010) was constructed using pCC7425 and a ΔdifA mutant (YZ601). pXQ719, which harbors the nafA chimeric gene (Xu et al., 2005), was electroporated (Kashefi & Hartzell, 1995) into appropriate M. xanthus mutants to construct NafA-expressing strains YZ1012 (ΔdifA Δpgi/nafA) and YZ754 (ΔdifA ΔdifB/nafA). YZ746, a mutant with the Q346A substitution in DifA, was constructed using a similar procedure described for the Q346D mutant YZ713 (Bonner et al., 2005).

Examination of DifA and FrzCD methylation

Three slightly different procedures were used to prepare samples for immunoblotting using either αDifA or αFrzCD antibodies (Xu et al., 2005, McCleary et al., 1990). For the examination of DifA methylation in different genetic backgrounds and nutrient conditions, cells in CYE liquid at 2.0 ∼ 3.0 × 10⁸ cells/ml were harvested, washed and resuspended at 2.5 × 10⁸ cells/ml in CYE liquid or MOPS buffer (10 mM MOPS, 2 mM MgSO₄, pH 7.6); these suspensions were incubated at 32°C for 3 h. For the examination of the effect of nitrate, cells in CTT liquid at 2.0 ∼ 3.0 × 10⁸ cells/ml were harvested, washed and resuspended at 2.5 × 10⁸ cells/ml in CTT liquid or TPM buffer [10 mM Tris·HCl (pH 7.6), 8 mM MgSO₄ and 1 mM K₂HPO₄-KH₂PO₄ (pH 7.6)] containing 700 μM of nitrate (KNO₃); these suspensions were incubated at 32°C for 3 h. For the examination of the effect of 12:0 PE, a 100 × stock solution containing 1.25% (w/v) 12:0 PE in a solvent of equal volumes of chloroform and methanol was prepared; cells in CTT liquid at 1.0 ∼ 2.0 × 10⁸ cells/ml were harvested, washed and resuspended at 6.0 × 10⁸ cells/ml in CTT liquid or TPM buffer; these suspensions with 12:0 PE or the solvent alone were incubated at 32°C for 100 min.

Cells from the above treatments were harvested, washed and resuspended in lysis buffer (10 mM Tris·HCl, 5 mM EDTA, pH 8.0) and lyzed by sonication. Protein concentration of the cell lysate was determined by the Protein Assay kit from Bio-Rad (Hercules, CA) with bovine serum albumin as the standard. To identify base-labile modifications, an aliquot of the cell lysate was treated with 0.2 M NaOH for 30 min and followed by neutralization with 0.2 M HCl (McBride et al., 1992, McCleary et al., 1990). Samples containing 30 μg of total protein were loaded in each well and analyzed by SDS-PAGE and immunoblotting as described previously (McCleary et al., 1990) using polyclonal antibodies against DifA (Xu et al., 2005) or FrzCD (McCleary et al., 1990).