Two-component signal transduction systems (TCS) are an important mechanism by which bacteria sense and respond to their environment. Although each two-component system appears to detect and respond to a specific signal(s), it is now evident that they do not always act independently of each other. In this paper we present data indicating regulatory links between the PhoPR two-component system that participates in the cellular response to phosphate limitation, and the essential YycFG two-component system in Bacillus subtilis. We show that the PhoR sensor kinase can activate the YycF response regulator during a phosphate limitation-induced stationary phase, and that this reaction occurs in the presence of the cognate YycG sensor kinase. Phosphorylation of YycF by PhoR also occurs in vitro, albeit at a reduced level. However, the reciprocal cross-phosphorylation does not occur. A second level of interaction between PhoPR and YycFG is indicated by the fact that cells depleted for YycFG have a severely deficient PhoPR-dependent phosphate limitation response and that YycF can bind directly to the promoter of the phoPR operon. YycFG-depleted cells neither activate expression of phoA and phoPR nor repress expression of the essential tagAB and tagDEF operons upon phosphate limitation. This effect is specific to the PhoPR-dependent phosphate limitation response because PhoPR-independent phosphate limitation responses can be initiated in YycFG-depleted cells.
Bacteria use multiple two-component signal transduction systems (TCS) to sense the prevailing physical, chemical and nutritional conditions and transduce this information into the cell so that an appropriate response can be effected. The prototypical TCS consists of two proteins, a sensor kinase and a response regulator (Hoch and Silhavy, 1995). Sensor kinases are composed of a variable sensing module, often with multiple transmembrane domains, and a cytoplasmically located transmitter module that has protein kinase activity. The response regulator is composed of a conserved receiver module that interacts with the protein kinase and an output module that is usually a DNA binding domain. Perception of a specific signal induces autophosphorylation of the protein kinase at a conserved histidine residue. Phosphorylation of the response regulator via phosphotransfer from the protein kinase to a conserved aspartate residue within the receiver module of the response regulator leads to its activation by one of the variety of mechanisms reviewed in Stock et al. (2000). Sensor kinases often have phosphoprotein phosphatase activity that allows the cellular response to be turned off. An essential feature of TCS is that specific histidine kinases and response regulators function as cognate pairs. A cognate pair will usually fulfil the following criteria: (i) the genes encoding the histidine kinase and response regulator are part of the same operon; (ii) null mutants of the histidine kinase or response regulator display similar phenotype, although the phenotype of the former mutant is often less extreme than the latter; and (iii) phosphotransfer between a sensor kinase and its cognate response regulator usually occurs efficiently in vitro. By functioning as a cognate pair, it is ensured that the elicited cellular response is appropriate to the perceived signal.
Most bacteria have multiple TCS encoded in the genome (e.g. Bacillus subtilis has 34 such systems). TCS can be used in combination to form regulatory networks, thereby increasing the complexity and subtlety of cellular responses to the prevailing conditions. Several different levels of interaction between TCS have been established that can be exemplified by work from B. subtilis (Bijlsma and Groisman, 2003). A biological process can be regulated by more than one TCS: competence development in B. subtilis requires the participation of ComPA, DegSU and the hybrid Kin/Spo0A TCS (Grossman, 1995; Hamoen et al., 2003). Expression of one TCS can be regulated by a second TCS: under phosphate-limiting conditions, expression of the resDE and phoPR TCS requires both ResDE and PhoPR in B. subtilis (Birkey et al., 1998). Other instances of such interactions were reported in the transcriptomic study of Kobayashi et al. (2001). A TCS can have multiple histidine protein kinases and response regulator phosphatases that permit integration of a wider diversity of signals: five kinases and six phosphatases are reported to participate in the hybrid Spo0F-Spo0B-Spo0A TCS (Hoch, 2000; Jiang et al., 2000a,b; Perego, 2001). The most contentious and least understood form of interaction between TCS is cross-talk, i.e. phosphorylation of a response regulator by a non-cognate histidine kinase. As cells usually have many TCS with conserved transmitter and receiver domains, the specificity of interaction between cognate/non-cognate pairs has prompted much investigation. Several studies have established that such cross-phosphorylation can occur in vitro. A recent study reports 21 instances of in vitro cross-talk from 692 non-cognate pairings of 27 (truncated) histidine kinases and 34 response regulators from Escherichia coli (Yamamoto et al., 2005). However, the occurrence of cross-talk in vivo and its physiological relevance is more difficult to assess. Several instances of in vivo cross-talk have been reported but usually only under non-physiological conditions where critical parameters of the histidine kinase–response regulator interaction were altered.
Recent work from this laboratory suggested that there may be a functional relationship between YycFG and the PhoPR TCS that regulates a cellular response to phosphate limitation (Howell et al., 2003). In this paper we extend these observations, showing that the PhoR kinase can phosphorylate the non-cognate YycF response regulator even in the presence of the cognate YycG kinase. However, the YycG histidine kinase does not appear to be able to phosphorylate the PhoP response regulator. In addition, we show that cells depleted for YycFG are unable to mount a normal PhoPR-dependent phosphate limitation response, even though two separate PhoPR-independent phosphate limitation responses are induced under these conditions.
Growth profile of strain AH9912 and accumulation of YycG in strains 168 and AH057 in low phosphate defined medium
Our previous study indicated a possible interaction between the PhoPR TCS that controls one of the cellular responses to phosphate limitation and the essential YycFG TCS in B. subtilis (Howell et al., 2003). To further examine whether YycFG plays a role in the PhoPR-dependent phosphate limitation response, we sought to establish the conditions under which cells could be YycFG-depleted and phosphate-limited simultaneously. Therefore, we examined growth of strain AH9912 (PspacyycFGHIJyyxA) when expressing different levels of yycFGHIJyyxA in low phosphate defined medium (LPDM). Separate starter cultures of strain AH9912 were grown in high phosphate defined medium (HPDM) containing the following IPTG (isopropyl-β-D-thiogalactopyranoside) concentrations: 30 µM, 40 µM, 50 µM, 75 µM, 100 µM and 1 mM. Each starter culture was used to inoculate LPDM without IPTG to an OD600 of 0.02. An LPDM culture containing 1 mM IPTG was similarly inoculated with strain AH9912 as a control. The growth profiles are shown in Fig. 1. When grown in LPDM the culture of strain AH9912 grown with 1 mM IPTG reaches a phosphate limitation-induced stationary phase at OD600 = 1.0, identical to the profile observed for the wild-type strain 168. Similar growth profiles are observed for cultures grown without IPTG that were inoculated from starter cultures containing 75 µM, 100 µM and 1 mM IPTG – these profiles are indistinguishable from that observed for wild-type cells. The phosphate limitation medium and growth conditions used in this study are very similar to those used by Pragai et al. (2004), who demonstrated experimentally that the concentration of phosphate in the medium had decreased to a low level when growth became attenuated at OD600 = 1.0. The similarity in growth kinetics, and in the OD600 at which growth of the culture was attenuated between this study and that of Pragai et al. (2004) allows us to be confident that the cultures grown without IPTG but inoculated from starter cultures containing 75–100 µM IPTG are indeed phosphate-limited. In contrast, the growth of those cultures that were inoculated from starter cultures containing lower amounts of IPTG was attenuated before phosphate limitation was reached, i.e. the transition phase was reached at OD600 values lower than 1 (Fig. 1). Interestingly, the time of growth attenuation correlated with the IPTG level of the starter culture: the 30 µM IPTG culture stopped growing first, followed by the 40 µM and then 50 µM cultures, suggesting that growth limitation is caused by YycFG depletion in these cultures. Therefore when a culture, grown in HPDM containing IPTG levels between 75 and 100 µM IPTG, is used to inoculate LPDM medium without IPTG, both the time and OD600 at which the culture reaches phosphate limitation is indistinguishable from that observed in wild-type cultures. It is also evident that neither the YycFG-limited cultures (i.e. those grown from starter cultures containing 30–75 µM IPTG) nor the cultures that are phosphate-limited and YycFG-depleted (i.e. those grown from starter cultures containing > 75 µM IPTG) lyse during the extended stationary phase. This contrasts with cultures grown in Luria–Bertani (LB) where significant lysis occurs when cells become YycFG-limited (Fabret and Hoch, 1998; Howell et al., 2003).
To examine YycG accumulation during growth in LPDM and to ensure that cultures grown under this regimen were also depleted for YycFG, we established the cellular YycG level in strains 168 and AH057 (ΔphoR) during a growth cycle in LPDM and compared it with YycG accumulation in cultures grown in LPDM without IPTG (but inoculated from started cultures containing 80–100 µM IPTG). The results of this Western analysis are shown in Fig. 2. Confirmation that the antiserum is specific for YycG is shown in Fig. 2A: (i) it specifically reacts with purified truncated ′YycG protein (data not shown); (ii) it reacts with a protein of the predicted Mr of YycG in extracts of strain AH9912 cells (lanes 2–4); and (iii) the level of that band (which co-migrates with the band seen in wild-type cells) increases with increasing IPTG concentration as expected during culture of strain AH9912 (PspacyycFGHIJyyxA) (lanes 1–4). The cellular YycG levels in strains 168 and AH057 (ΔphoR) during a growth cycle in LPDM are shown in Fig. 2B. It is evident that YycG is present during both the exponential growth and phosphate limitation-induced stationary phases of the growth cycle. With the exception of time points T−3 and T1 where the level appears to be somewhat lower, the level of YycG is fairly constant during the remaining periods of the growth cycle. Importantly for this study, deletion of phoR has little detectable effect on the cellular YycG level (compare the WT lane with the Δ lane at each period of the growth cycle in Fig. 2B). To verify that cells become YycFG-depleted during growth in the double-depletion regimen previously established, a culture of strain AH9912 (PspacyycFGHIJyyxA) was grown overnight in HPDM containing 80 µM IPTG and diluted into LPDM without IPTG. This culture grew exponentially until a phosphate limitation-induced stationary phase was reached at OD600 = 1.0. The cellular YycG level in this culture was determined by Western analysis and is shown in Fig. 2C. YycG is detectable only during the early growth phases at times T−3 and T−2, and at a level significantly lower than in wild-type cells during exponential growth (see Fig. 2B). However, YycG could not be detected at any subsequent period of the growth cycle in LPDM, even with increased exposure periods. These data confirm (i) that expression of the yycFGHIJyyxA operon in strain AH9912 can be controlled by IPTG addition; (ii) that YycG is present during both the exponential and phosphate limitation-induced stationary phases of growth in LPDM at similar levels and that deletion of phoR does not alter the YycG level detectably during these growth conditions; and (iii) that under the growth regimen established for simultaneous YycFG depletion and phosphate limitation, the level of YycG is undetectable at T−1, showing that the cells are indeed YycFG-depleted upon entry into the phosphate limitation-induced stationary phase of the growth cycle.
The expression of yocH following phosphate limitation is dependent on both YycF and PhoR
To determine the role of the YycF response regulator during growth in a limiting phosphate defined medium, we examined expression of yocH in strain AH9912 where expression of the yycFGHIJyyxA operon is under the control of an IPTG-inducible Pspac promoter. As yocH expression is positively regulated by the phosphorylated form of YycF, the level of yocH transcript reflects the level of YycF activation. We grew strain AH9912 in LPDM (inoculated from a HPDM starter culture containing 100 µM IPTG) containing no IPTG, 100 µM IPTG and 1 mM IPTG monitoring growth turbidimetrically (Fig. 3A) and yocH expression by Northern analysis (Fig. 3B and C). All three cultures grow at very similar rates until T−60: at T−30 growth has slowed for all three cultures while at T0 all cultures have reached a plateau. The plateau OD600 of the culture grown without IPTG was slightly lower than those of the cultures containing 100 µM and 1 mM which are superimposable (Fig. 3A). Because neither the 100 µM nor the 1 mM IPTG cultures are YycFG-limited, retardation of cell growth at T−30 must represent detection of phosphate limitation by all three cultures. Expression of yocH in the 100 µM and 1 mM IPTG-containing cultures is shown in Fig. 3B, while Fig. 3C shows expression in 0 µM and 1 mM IPTG-containing cultures from a separate experiment. During growth in 1 mM IPTG, expression of yocH is high during exponential growth (T−90) and decreases as the culture approaches the phosphate limitation-induced stationary phase. However, it is evident from the cultures grown in 100 µM and 1 mM IPTG that yocH expression increases after phosphate limitation and remains high for up to 240 min. The level of yocH expression in cultures grown without IPTG differs substantially: a low expression level is observed during exponential growth with only a small increase at the point of phosphate starvation and the yocH transcript stays at this level for the remainder of the experiment. These data suggest that the phosphate limitation-induced expression of yocH is dependent on one or more of the genes of the yycFGHIJyyxA operon.
In strain AH9912 the entire yycFGHIJyyxA operon is under the control of the Pspac promoter, and potentially any gene could be responsible for the phosphate limitation-induced expression of yocH (see Fig. 3B and C). To refine this analysis, we focused our attention on the involvement of the YycFG and PhoPR TCS in controlling phosphate limitation-induced expression of yocH, specifically concentrating on the YycF response regulator and the PhoR kinase. In strain AH047 (PxylyycF, ΔphoP), the yycFGHIJyyxA operon is intact and expressed normally, but there is an additional copy of the yycF gene under the control of a xylose-inducible promoter positioned at the threonine locus on the chromosome. Importantly this strain also has an in frame deletion in the phoP gene. Therefore, we can establish the contribution of YycF alone to the yocH expression profile during growth in LPDM and determine whether the PhoP response regulator contributes to the phosphate limitation-induced expression. The results (Fig. 4A) show that in the absence of xylose inducer, expression of yocH is high between T−180 minutes and T30 minutes post phosphate limitation. The level of yocH transcript then decreases to a low level at T60 and T90, which is followed by an increased transcript level at T120 and T180 minutes post phosphate limitation. The level of yocH expression is higher in this strain than that observed in AH9912 (see Fig. 3B and C) especially during exponential growth and at the point of phosphate limitation. This is probably due to the fact that in strain AH047, YycF levels reflect expression both from its natural promoter and from the Pxyl-inducible promoter that we know from previous experiments to be leaky (Jester et al., 2003). Upon addition of 1% xylose to increase the cellular level of YycF only, there is a very considerable increase in the level of yocH transcript. This is especially evident between T0 and T60, the period of phosphate limitation. For example, the reduced level of transcript normally observed in the absence of xylose at T60 and T90 post phosphate limitation is increased substantially. This result shows that increasing the cellular level of YycF alone results in an increased the level of yocH transcript upon phosphate limitation. As a control to ensure (i) that equal amounts of RNA were loaded in both the –xylose and +xylose cultures; (ii) that the RNA was not degraded; and (iii) that both cultures were phosphate-limited at the same time and to the same extent, we stripped the yocH probe from the filter of the Northern shown in the top panel of Fig. 4A and reprobed the membrane with a yxiE probe. Expression of yxiE is induced by phosphate limitation in a PhoPR-independent manner with a characteristic induction profile (Antelmann et al., 2000). The profile of yxiE induction for the –xylose and +xylose cultures are shown beneath the corresponding yocH profiles. It is clear that both cultures were phosphate-limited at similar times, and that phosphate limitation resulted in comparable levels of yxiE transcript in the –xylose and +xylose cultures. This result shows that the decreased yocH transcript level at T60 and T90 in the –xylose culture and T90 in the +xylose cultures are not due to total RNA degradation. These results are especially important, because they demonstrate that expression of yocH during the phosphate limitation-induced stationary phase is YycF-dependent and does not depend on the PhoP response regulator (AH047 has a deletion in the phoP gene).
We then sought to determine which sensor kinase was responsible for activation of the YycF response regulator. As we previously showed using the YycF′–′PhoP hybrid response regulator that the YycG histidine kinase does not detect phosphate limitation (Howell et al., 2003), we examined the involvement of PhoR in the phosphate limitation-induced and YycF-dependent expression of yocH. Strain AH048 contains the additional copy of yycF under xylose-inducible control at the threonine locus (the same as in strain AH047), but both the phoP and phoR genes are deleted. Cultures of strain AH048 were grown in the presence and absence of 1% xylose in LPDM and expression of yocH was determined by Northern analysis, as already described for strain AH047. Results are shown in Fig. 4B. It is evident that expression of yocH during exponential growth is approximately equal in both cultures (i.e. with and without xylose inducer) and that the levels are similar to those found for AH047 (see Fig. 4A). However, the elevated level of yocH expression that was observed for up to 3 h after the point of phosphate limitation in strain AH047 is almost completely abolished in strain AH048. There is a very low level of yocH expression post phosphate limitation in strain AH048, with little detectable difference in the cultures grown without and with the xylose inducer. To establish (i) that these cultures of strain AH048 were phosphate-limited; (ii) that limitation occurred at the same times in each culture; and (iii) that equal amounts of RNA were indeed loaded onto each lane, the membrane shown in the top panel was stripped of the yocH probe and reprobed with the yxiE probe as already described. The similar expression profiles in the –xylose and +xylose cultures confirm that the cells were phosphate starved at approximately the same times, that equivalent amounts of RNA were loaded in each lane and most importantly that the reduced level of yocH transcript observed post phosphate limitation is not due to RNA degradation. In summary, these results indicate that YycF is activated during a phosphate limitation-induced stationary phase and that this activation of YycF is mediated by the PhoR kinase.
YycF can be phosphorylated by PhoR but YycG cannot detectably phosphorylate PhoP in vitro
The genetic evidence already presented suggests that YycF can be phosphorylated by the non-cognate PhoR, although the reciprocal phosphorylation of PhoP by YycG seems not to occur (this study and Howell et al., 2003). As truncated forms of the YycG and PhoR kinases (i.e. ′YycG and PhoR231) are active, capable of autophosphorylation and phosphorylation of their purified cognate regulators, YycF and PhoP respectively (Howell et al., 2003; Pragai et al., 2004), we tested the capacity of the PhoR231 kinase to phosphorylate the YycF regulator in vitro. The results are shown in Fig. 5. It is clear that the ′YycG kinase is active, capable of autophosphorylation and of phosphorylating the YycF response regulator (Fig. 5A, lanes 1–4). Similar results are shown for PhoR231 and PhoP (Fig. 5B, lanes 5–8). In a reaction containing the ′YycG kinase and the PhoP response regulator (Fig. 5B, lanes 1–4) the ′YycG autophosphorylates, but there is no detectable phosphorylation of PhoP, a result obtained in multiple experiments. However, in a reaction containing the PhoR kinase and YycF (Fig. 5A, lanes 5–8), it is evident that in addition to a band of autophosphorylated PhoR23 a second band corresponding to phosphorylated YycF is seen. Although the level of YycF phosphorylation by PhoR231 is low, it was observed repeatedly. The results of these experiments show that PhoR can phosphorylate YycF albeit at a low level in vitro while YycG cannot phosphorylate PhoP under the same conditions, observations that are in concordance with the genetic evidence.
Phosphorylation of YycF by PhoR in vivo during growth in LPDM
The accumulated evidence thus far shows that PhoR can phosphorylate YycF under three non-physiological conditions: (i) PhoR can phosphorylate the YycF′–′PhoP hybrid response regulator (Howell et al., 2003); (ii) PhoR can phosphorylate YycF when its cellular levels are elevated due to induced expression of an additional copy of the yycF gene (in strain AH047 thr::PxylyycF); and (iii) purified (truncated) PhoR231 kinase can phosphorylate YycF in vitro. To assess the physiological relevance of this putative cross-phosphorylation between the non-cognate PhoR-YycF histidine kinase and response regulator pair, we compared the yocH expression profile of wild-type strain 168 with the profile of strain AH057 (ΔphoR) both grown in LPDM. It is important to emphasize that the yycFGHIJyyxA operon and its expression have not been modified in any way in either strain. In addition it has already been established that YycG is present throughout the growth cycle in LPDM and that deletion of phoR does not detectably alter the YycG level at any stage (see Fig. 2B). The profile of yocH expression in wild-type cells during a growth cycle in LPDM is shown in Fig. 6 (top panel): there is a low level of expression during exponential growth, which increases dramatically at the onset of phosphate limitation (T0). The level then decreases somewhat in the samples harvested between T30 and T90, followed by increased expression at T150 and T210. The expression of yocH in strain AH057 (ΔphoR) differs significantly from the wild-type profile in two important respects: (i) the level of yocH transcript is substantially increased during exponential growth relative to the levels seen in wild-type cells during the same growth period; and (ii) there is only a very low level of yocH expression upon phosphate limitation. Transcript is present at an almost undetectable level at T30 and T60 with a small increase in transcript level between T90 and T210. This expression profile was observed in RNA preparations from three separate growth experiments. These results indicate that the PhoR kinase has a negative effect on yocH expression during exponential growth in LPDM and a positive effect on its expression in response to phosphate limitation.
Repression of the tagAB/tagDEF divergon and activation of phoA expression in response to phosphate limitation are both YycFG-dependent
The YycF protein binds to the promoter region of the tagAB/tagDEF divergon (Howell et al., 2003). Intriguingly, the tagAB/tagDEF genes, encoding the enzymes for teichoic acid biosynthesis, are essential for viability and their expression is turned off by the PhoPR two-component system under phosphate-limiting conditions (Karamata et al., 1987; Mauel et al., 1989; Liu et al., 1998; Bhavsar et al., 2001). In view of the relationship between the PhoR kinase and YycF, and the essential nature of both the YycFG two-component system and the teichoic acid biosynthetic genes, we decided to investigate whether YycFG participates in regulating expression of the tagAB/tagDEF genes during exponential growth or during phosphate limitation. Using conditions already established to obtain cultures depleted for YycFG and phosphate-limited simultaneously, two HPDM starter cultures of strain AH9912 were grown in HPDM containing 100 µM and 1 mM IPTG. The HPDM culture containing 100 µM IPTG was used to inoculate an LPDM culture without IPTG. As a control, the HPDM culture containing 1 mM IPTG was used to inoculate a LPDM culture containing 1 mM IPTG. The growth profiles of both cultures are almost superimposable with phosphate limitation occurring at approximately OD600 = 1.0 (Fig. 7A). Total RNA, made from cells harvested at the indicated time points, was used as template for primer extension reactions to monitor expression of the tagAB and tagDEF operons. Results are shown in Fig. 7B. In the culture grown in 1 mM IPTG, the levels of transcript are high during exponential growth and persist for up to 30 min post phosphate limitation, at which point they decrease to an undetectable (for tagAB, Fig. 7B upper panel) or very low (for tagDEF, Fig. 7B, lower panel) level. In cultures grown without IPTG, both the tagAB and tagDEF transcript levels are observed during exponential growth but persist during the phosphate limitation-induced stationary phase. The tagAB transcript was readily observed at all time points up to T270 in the cultures grown without IPTG. This result was observed in three separate experiments. These data show that the PhoPR-dependent repression of the tagAB and tagDEF divergon normally observed upon phosphate limitation requires the YycFG TCS. As the tagAB/DEF transcripts persist upon YycFG depletion, it is unlikely that the essential nature of the YycFG TCS can be explained by its control of this divergon.
While expression of the tagAB/DEF genes are negatively regulated by PhoPR in response to phosphate limitation, expression of the phoPR operon (encoding the PhoPR TCS) and the phoA gene (encoding alkaline phosphatase) are both activated under these conditions. Therefore we sought to establish whether the activation of phoPR and phoA in response to phosphate limitation might also require YycFG. Cultures were grown in LPDM under conditions that resulted in simultaneous depletion of YycFG and phosphate (–IPTG) and depletion for phosphate only (+IPTG) as already established. The growth profiles of strain AH9912 under these two conditions are virtually superimposable with the culture grown without IPTG entering the transition phase at OD600 = 0.562 and the culture grown with IPTG entering the transition phase at OD600 = 0.522 (Fig. 8A). During the stationary phase the culture grown without IPTG reached a steady state OD600 = 1.1 while the culture grown with IPTG reached a steady state OD600 = 1.01. The levels of both phoPR and phoA RNA transcripts in cells from the transition state (T0) up to T180 were determined by Northern analysis and are shown (Fig. 8B). In the culture grown with IPTG, phoA transcript first appears at T80 and rises to a maximum at T180. However, in the culture grown without IPTG, phoA transcript is only observed at T120 and T140 but at levels very significantly lower those in the plus IPTG culture. The profile of phoPR transcript accumulation is similar to that of phoA. In the culture containing IPTG, a low level of phoPR transcript was observed between T0 and T60 with a significant increase between T80 and T140. In the culture grown without IPTG, phoPR transcripts are barely detectable between T0 and T80; this increases at T100 and T120 before decreasing at subsequent time points. These data show that when YycFG is present, expression of both phoPR and phoA (both positively regulated by PhoPR) rises to significant levels within 80 min of phosphate depletion: in contrast, depletion of YycFG under the same conditions results in a delay of the response and a very significant decrease in the levels of transcripts. A similar response was observed in five separate experiments.
To ascertain whether these YycFG-depleted/phosphate-limited cultures were non-viable, incapable of sensing a phosphate limitation response or indeed whether the RNA in such cells was degraded, we stripped the membranes and separately reprobed to detect yxiE and yknZ (SigB-dependent phosphate limitation induction, Antelmann et al., 2000; Pragai and Harwood, 2002) transcripts. The results are shown in Fig. 8B. A similar profile of yxiE induction was observed in both cultures with comparable levels of transcript at each stage of the growth cycle. The yknZ transcript was first observed at T40 in both cultures and increased to high levels during the remainder of the experiment, although there were differences in transcript level at some time points during the experiment. However, it is clear that both yxiE and yknZ are induced as expected upon phosphate limitation in a manner independent of the YycFG status of the cell. In contrast, expression of genes both positively (phoPR and phoA) and negatively (tagAB tagDEF) regulated by PhoPR upon phosphate limitation is affected by the YycFG status of the cell. These data confirm that cells depleted for YycFG and phosphate-limited simultaneously are capable of sensing and responding to phosphate limitation by the SigB pathway and by the unknown pathway controlling induction of yxiE but are incapable of mounting a normal PhoPR-dependent phosphate starvation response. Cumulatively, these results show that to execute a PhoPR-dependent phosphate limitation response, the cell requires expression of the yycFGHIJyyxA operon encoding the YycFG two-component system.
YycF binds to the phoPR promoter region
The finding that expression of the yycFGHIJyyxA operon, encoding the YycFG TCS, is required for a normal PhoPR-dependent phosphate limitation response prompted us to examine the possibility that YycF may bind directly to the phoPR promoter. The binding of purified PhoP and YycF to the phoPR promoter was assessed using gel shift assays as described in Experimental procedures and the results for YycF are shown in Fig. 9. A small retardation of the phoPR promoter fragment is observed at 13 µM YycF while an almost complete shift of the probe is observed at 19.5 µM YycF. Addition of a 25-fold excess of non-biotinylated phoPR promoter fragment results in an almost complete inhibition of promoter retardation (lanes 6–10). To establish the concentration at which PhoP binds to the phoPR promoter under these same conditions, purified PhoP was substituted for YycF in the above experiment. Retardation of the phoPR promoter fragment was first observed at a PhoP concentration of 8.5 µM with complete retardation occurring at 17 µM, levels comparable with those observed for YycF (data not shown). As a negative control, the htrA promoter whose expression is regulated by the CssRS TCS was tested under the above conditions and no retardation of the DNA fragment was observed even at the highest concentration of YycF protein used. These data show that YycF can bind directly to the phoPR promoter.
In this paper we address the relationship between the YycFG and PhoPR two-component systems and the possibility that they may interact to regulate the cellular response to phosphate limitation in B. subtilis. We show that the YycFG and PhoPR TCS interact at two levels to control gene expression both during exponential growth and during a phosphate starvation-induced stationary phase. The first level of interaction is demonstrated by the ability of PhoR to phosphorylate and possibly to dephosphorylate the YycF response regulator. The supporting evidence for this interaction is threefold: (i) overexpression of either the entire yycFGHIJyyxA operon or of the yycF gene alone results in increased expression of the yocH gene during a phosphate limitation-induced stationary phase and this expression is abolished in strains where the phoR gene is deleted; (ii) deletion of the phoR gene in an otherwise wild-type background abolishes phosphate limitation-induced expression of yocH and also results in increased expression of yocH during exponential growth; and (iii) PhoR231, an amino-terminally truncated version of PhoR, can phosphorylate the YycF response regulator in vitro. Together these genetic and biochemical evidence provide strong evidence that PhoR can phosphorylate YycF when activated by phosphate limitation. It also suggests that PhoR can dephosphorylate YycF during exponential growth in LPDM (although this remains to be verified by additional data), indicating a role for PhoR in regulating gene expression during this phase of the growth cycle. As cross-phosphorylation of YycF by PhoR occurs in wild-type cells during phosphate limitation where (i) the expression of YycFG and PhoPR has not been modified in any way; and (ii) the cognate YycG kinase is present throughout the period of this cross-phosphorylation (Fig. 2B), we conclude that the non-cognate interaction between them is physiologically relevant under these conditions. The phosphorylation of YycF by PhoR does not appear to be just a consequence of the close phylogenetic relationship between PhoPR and YycFG because the reciprocal phosphorylation i.e. phosphorylation of PhoP by YycG does not detectably occur. This is demonstrated in vitro in this paper (Fig. 5) and is also supported by the fact that Pho regulon genes are not expressed during exponential growth in LB where we know YycG to be activated. While instances of cross-talk between non-cognate sensor kinases and response regulators have been reported, most were observed under unusual conditions: for example where the cognate kinase is deleted or where the normal stoichiometry of a cognate pair is altered by over- or under-expression of constituent proteins or domains (Bijlsma and Groisman, 2003). In their review evaluating the evidence for, and physiological relevance of, this form of interaction, Bijlsma and Groisman (2003) conclude that the cumulative data ‘suggest(s) that cross-talk is rare, if it occurs at all in a way that is significant biologically’. The interaction between PhoR and YycF reported here suggests that perhaps instances of physiologically relevant cross-talk will be observed but only between TCS that are very closely related phylogenetically (as PhoPR and YycFG are) and under very particular physiological conditions.
A second level at which the PhoPR and YycFG TCS interact is seen in the requirement for expression of the yycFGHIJyyxA operon encoding YycFG to initiate the PhoPR-dependent phosphate limitation response. Exposure of cells to phosphate limitation results in PhoPR-dependent repression of the teichoic acid (a phosphate-rich cell wall polymer) biosynthetic genes (tagAB tagDEF), activation of phoA (encoding an alkaline phosphatase) and autoactivation of the phoPR operon. Our results show that such regulatory events do not occur under phosphate-limiting conditions when cells are depleted for the products of the yycFGHIJyyxA operon. While our results, especially the observation that YycF binds to the phoPR promoter, are suggestive that it is the YycFG two-component system that provides the critical link to the PhoPR two-component system and induction of the PhoPR-mediated phosphate limitation response, we cannot rule out the involvement of (some) other members of the yycFGHIJyyxA operon. It is important to emphasize that these YycFG-depletion effects are specific to the PhoPR-mediated phosphate limitation response because genes induced by phosphate limitation in a PhoPR-independent manner (yxiE, yknZ, ysnF) are induced normally. Our finding that YycF binds directly to the phoPR promoter suggests a mechanism for the yycFGHIJyyxA expression requirement during induction of the PhoPR response. More importantly, it provides a rationale both for the observed cross-talk between PhoR and YycF and why the reciprocal cross-phosphorylation does not occur. In this model, YycF∼P and PhoP∼P are both required for initiation and/or full induction of the PhoPR mediated response. However, YycG seems to be activated only during exponential growth and to be incapable of detecting phosphate limitation (Howell et al., 2003). During phosphate limitation when cells are not dividing, the YycG-specific stimulus is not present (although YycG itself is present, Fig. 2B). Therefore, the cellular YycF∼P levels can only be maintained during phosphate limitation if YycF is phosphorylated by a non-cognate kinase. Our results suggest that PhoR, the kinase that is activated by phosphate limitation, fulfils this role.
The recognition specificity between proteins lies in their interacting surfaces. Considerable insights into the recognition of response regulators by kinases have been gained from the analysis of the Spo0F:Spo0B crystal structure (Zapf et al., 2000; Hoch and Varughese, 2001; Varughese, 2002). Response regulators are single domain proteins and the site of phosphorylation is located in a shallow pocket on the top of the β-sheet (Fig. 10). The phosphotransferase domain of histidine kinases are four-helix bundles formed through the dimerization of two helical hairpin structures from the two ‘promoters’ that make up a kinase dimer (Fig. 10). The active histidine residues are located on the four-helix bundle and consequently most of the significant interactions of the response regulator during phosphoryl transfer are with the four-helix bundle. In view of the unidirectional cross-talk we observe between the YycFG and PhoPR TCS, it is interesting to examine those amino acids of the YycG and PhoR histidine kinases that interact with their respective response regulators. In histidine kinases, the interacting amino acids are situated around the site of phosphorylation located on helix 1 (Mukhopadhyay and Varughese, 2005). The most relevant for comparison between YycG and PhoR are positions −4, −3, −1, +1, +3, +4, +7, +8, +10, +11 and +12 shown in Table 1 (numbered relative to the phosphorylated His residue). Figure 10 shows that these residues are suitably positioned to interact with the response regulators. In YycG and PhoR, the residues at positions −4, −3, −1, +1, +4 and +7 are identical. The amino acids at +3, +8, +10 and +12, although different, are similar in nature: (Arg-Thr-Arg-Tyr versus Lys-Ser-Lys-Phe). Position 10 is especially interesting because it is different for all members of this HK family, suggesting that the amino acid at this position has a role in determining specificity. Interestingly YycG and PhoR have similar residues at this position (an Arg and a Lys) perhaps paving the way for cross-talk. It is noteworthy that YycG and PhoR do not have similarity in the nature of the amino acid at this position with any of the other four kinases in the family. However, the question then is, why can YycG not talk with PhoP? Of the 11 pairs of amino acids examined, six are identical, four are similar and one is different. That very conspicuous difference is at position 11 where YycG has a Ser whereas PhoR has a Gly, suggesting that this position plays a role in YycG discriminating against PhoP. Gly is the smallest amino. Therefore the presence of Gly in PhoR creates an additional space on the surface of the four-helix bundle for the response regulator to approach. YycG has a Ser at this position, which complements YycF but most likely causes steric repulsion of PhoP. Additionally, the residues in YycG at positions +3, +8, +10 and +12 (Arg-Thr-Arg-Tyr) are larger than the corresponding residues in PhoR (Lys-Ser-Lys-Phe), again implying possible steric repulsion of PhoP by YycG. These predictions are being tested experimentally.
Table 1. A comparison of the sequences around the site of phosphorylation in the YycG familya.
An accumulating body of direct and indirect evidence supports a connection between YycFG and PhoPR and their involvement in the cellular response to phosphate limitation in B. subtilis and other bacteria. (i) The YycFG and PhoPR TCS are very closely related phylogenetically with the YycG, PhoR and ResE histidine kinases forming a subdivision of Group IIIA according to the classification of Fabret et al. (1999); (ii) The receiver and output domains of YycF and PhoP can be swapped to give two functional hybrid response regulators, PhoP′–′YycF and YycF′–′PhoP (Howell et al., 2003); (iii) The DNA binding sequences for PhoP and YycF are the complement of each other: the PhoP binding site is [5′..TT(A/T/C)ACA N5 TT(A/T/C)ACA..3′] with the number of repeats and whether they are on the coding or non-coding strand determining the efficiency of binding and whether activation or repression of expression is effected respectively (Liu and Hulett, 1998; Eder et al., 1999). The YycF DNA binding sequence is [5′..TGT(A/T)A(A/T/C) N5 TGT(A/T)A(A/T/C)..3′] in both B. subtilis and S. aureus (Howell et al., 2003; Dubrac and Msadek, 2004); (iv) YycF has been shown to control expression of genes involved in phosphate metabolism. In addition to the results presented in this paper and in Howell et al. (2003) for B. subtilis, Ng et al. (2004) have shown that the phosphate ABC transporter PstS and the regulator PhoU are members of the YycFG regulon in S. pneumoniae. Also, Martin et al. (1999) have shown that a strain carrying the yycF1 temperature-sensitive allele was hypersensitive to the lincosamide antibiotics clindamycin hydrochloride and lincomycin hydrochloride, a condition that was also observed in S. aureus strains carrying conditional lethal alleles of genes involved in phospholipids biosynthesis (P.K. Martin, unpublished: see Martin et al., 1999). In view of this observation it is interesting that the lincosamide antibiotic sensitivity of the strain carrying the yycF1 allele could be suppressed by adding 100 µM Pi to the medium. Cumulatively these data suggest that the YycFG TCS participates in regulating the response to the phosphate status of the cell. Indeed phosphate homeostasis might be a unifying theme to explain the diverse phenotypes and gene expression profiles observed upon YycFG overexpression, depletion or mutation in several studies on different bacteria (Fabret and Hoch, 1998; Martin et al., 1999; 2002; Fukuchi et al., 2000; Echenique and Trombe, 2001; Howell et al., 2003; Ng et al., 2003; Dubrac and Msadek, 2004; Mohedano et al., 2005). Both cytoplasmic membranes (phospholipids) and cell walls (teichoic acids) contain high phosphate levels and perturbation of phosphate homeostasis might be expected to affect expression of genes involved in the biosynthesis and maintenance of the components of these cellular structures. Experiments are under way to ascertain whether the YycFG and PhoPR relationships seen in B. subtilis are more generally observed.
The work presented here, in conjunction with the studies of Hulett and colleagues (Sun et al., 1996a,b; Birkey et al., 1998), shows that the YycFG, PhoPR and ResDE two-component systems form a signalling network that senses and executes changes in gene expression in response to phosphate limitation. Previous reports showed (i) that the resABCDE operon (encoding the ResDE two-component system) is induced by phosphate limitation in a manner that is dependent on both ResDE and PhoPR; (ii) that induction of alkaline phosphatase by phosphate limitation requires ResDE and that this ResDE-mediated control is mediated through PhoPR (Sun et al., 1996a,b; Birkey et al., 1998). In this report we show that YycFG is also part of this network through cross-talk between PhoR and YycF, and through YycF binding to the phoPR promoter, perhaps together accounting for the YycFG-dependence in effecting the phoPR-dependent phosphate limitation response. Therefore, activation of the PhoPR-dependent response upon phosphate limitation requires three two-component systems, PhoPR, ResDE and YycFG.
Bacterial strains, plasmids and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 2. E. coli K12 strain TG-1 (Gibson, 1984) was used for routine cloning and was transformed using the calcium chloride method (Sambrook et al., 1989). B. subtilis was transformed by the method first presented by Anagnostopoulos and Spizizen (1961). B. subtilis 168 trpC2 and derivatives were grown in LB medium (Miller, 1972), HPDM or LPDM (Muller et al., 1997). IPTG was added to the final concentrations as specified in the text. Antibiotics were added at the following concentrations: ampicillin, 100 µg ml−1; chloramphenicol, 3 µg ml−1; erythromycin, 2 µg ml−1; neomycin, 7 µg ml−1.
pDG782 derivative for deletion of the phoR locus. Contains fragment predominantly upstream of phoR (nucleotides −407 to +56 of phoR ORF cloned into StuI–BglII) and fragment downstream of phoR (a 409 bp fragment beginning 14 bp downstream of phoR ORF cloned into PstI–BamHI) either side of a neomycin resistance cassette.
Vector for overexpressing His-tagged proteins using a T7 bacteriophage promoter (ApR)
735 bp PCR fragment containing the yycF gene cloned into pET21a
Strains AH9912 and AH024 were constructed as described previously (Howell et al., 2003). Strain AH043 was generated by transforming strain AH024 with pEr::Nm, a plasmid which permits the replacement of erythromycin antibiotic resistance with neomycin (or kanamycin) resistance by double cross-over (Steinmetz and Richter, 1994). The pEr::Nm plasmid contains an intact neor cassette flanked by the front and back regions of an ermr cassette. Transformants in which a double cross-over had taken place were confirmed by screening for neomycin resistance and erythromycin sensitivity. Strains AH047 and AH048 were generated by transforming MH5117 and AH043 respectively, with chromosomal DNA from strain AE04. Transformants were screened for chloramphenicol resistance and erythromycin sensitivity to verify the integration of the PxylA-yycF (Cmr) construct into the chromosome. Strain AH057 was constructed by transforming 168 with plasmid pAH027. This plasmid was constructed by cloning a fragment upstream of phoR (amplified with primers phoRKO-1 [5′-AAAAGGCCTGCCGTTCAGTCCAAGGGAAG-3′] and phoRKO-2 [5′-GAAGATCTCACCAGAATCATACAGACAAC-3′]) and a fragment downstream of phoR (amplified with primers phoRKO-3 [5′-AAAACTGCAGGGTTTAACAATATCT TCATGC-3′] and phoRKO-4 [5′-CGCGGATCCCGCAACC AGCATGTGCGTCGG-3′] upstream and downstream, respectively, of the neor cassette in pDG782 (Guerout-Fleury et al., 1995). Underlined primer sequence denotes restriction sites added for cloning purposes, and the sequences 5′ to these sites are sequence clamps to aid in digestion of the polymerase chain reaction (PCR)-generated product. The phoR deletion was confirmed by Southern blotting (Sambrook et al., 1989).
Standard procedures were used for DNA manipulation (Sambrook et al., 1989). Restriction enzymes were supplied by New England Biolabs (Beverly, Massachusetts) and T4 DNA ligase was obtained from Boehringer (Mannheim, Germany).
All HPDM and LPDM cultures were inoculated using starter cultures grown in HPDM. LB cultures were inoculated from LB starter cultures grown overnight with the exception of strain AH9912, where the starter cultures were grown for only 4 h before inoculation. The starter cultures of strain AH9912 contained 100 µM IPTG unless otherwise stated. Cultures were grown at 37°C with shaking at 220 rpm in an orbital shaker (New Brunswick, Edison, NJ). Total RNA was prepared from B. subtilis cells as previously described except that the cells were broken using a Fastprep shaker (Bio101) (Noone et al., 2000). Northern blotting and detection was carried out as previously described before (Howell et al., 2003). Primer extension analysis was performed using 25 µg of total RNA isolated from cells harvested at appropriate times (Noone et al., 2000). The RNA was annealed to radioactively end-labelled primer tagA PE-1 (5′-TGTGAATA GTCTCTGTTTGC-3′) or tagD PE-1 (5′-AAGTTCCATATG TGATAAC-3′).
Protein phosphorylation assays
Proteins (2 µg of each) were incubated for 5 min at room temperature in 9 µl of phosphorylation buffer (100 mM Tris HCl pH 8, 200 mM KCl, 4 mM MgCl2, 0.5 mM DTT, 0.1 mM EDTA, 3.5% glycerol). Phosphorylation reactions were initiated by addition of 1 µl of a radiolabelled ATP mixture (2.5 µM ATP, 4 µCi [γ-32P]-ATP) and performed for 30 min at room temperature. Reactions were stopped by adding 2 µl of SDS loading buffer and analysed by SDS-PAGE on 12% acrylamide gel, followed by autoradiography.
Expression and purification of YycFhisx6
Overproduction and purification of truncated kinases (′YycG and PhoR231) and their cognate regulators (YycF and PhoP) were performed essentially as in previous studies (Howell et al., 2003; Pragai et al., 2004). Briefly, to produce YycF, plasmid pDN1100 was generated by directionally cloning the yycF gene encoded on a PCR fragment generated with primers YYCFpetF (5′..GGAATTCCATATGGATAAAAAGATCCT TGTAG..3′) and YYCFpetR (5′..CCGCTCGAGGTCCTGTT CTGGGTTTCTC..3′) into plasmid pET21a using NdeI and XhoI sites. The YycF protein produced has six His amino acids at the carboxy-terminus. The fidelity of the PCR amplification was verified by sequencing. Plasmid pDN1100 was transformed into E. coli strain BL21 DE3 and YycF expression was induced by addition of 1 mM IPTG during exponential growth. YycF was purified from the soluble fraction of the lysate using Ni-NTA agarose beads (Qiagen) according to the manufacturer's instructions in a batch process and YycFhisx6 was eluted using 300 mM imidazole. The eluate containing YycFhisx6 was dialysed against 20 mM Tris pH 8, 300 mM NaCl, 50% glycerol as a concentrating and storage buffer.
Generation of antiserum against YycG
A mixture of two peptides from yycG, YYCG1, CKFD SKDYQFNREWI (amino acids 439–452) and YYCG2, CLPY KEEQEDDWDEA (amino acids 598–611) were designed, synthesized and injected into New Zealand white rabbits by Sigma Genosys (Cambridge, UK). Antibodies were affinity purified from the sera using peptides coupled to CNBr activated Sepharose 4 FastFlow(Pharmacia) following the recommendations of the manufacturer. A total of 5 mg of each peptide were bound to a single batch of 0.6 g of swollen equilibrated CNBr activated Sepharose 4 FastFlow(Pharmacia) resin. Antibodies were eluted using 0.2 M glycine pH 2.2 into tubes containing 2 M Tris pH 8.8, desalted and buffer exchanged using MicroSep concentration columns (Pall) into 1× PBS containing 40% glycerol, 1 mg ml−1 BSA and 0.1% sodium azide, aliquoted and stored at −20°C.
Gel mobility shift DNA binding assays
The promoter region of the phoPR operon was amplified by PCR with biotinylated primers PHOPRfwd (5′..TCAT TGAACTTGAACTGACAG..3′) and PHOPRrev (5′..AAATTTT CTTGTTCATGCTGTG..3) (Sigma-Genosys) giving a 316 bp fragment which was gel purified. Binding reactions were performed on ice in 10 mM Tris pH 7.4, 50 mM NaCl, 10% glycerol, 5 mM EDTA, 20 mM DTT in a final volume of 20 µl containing 0.5 µg poly (dI-dC) for 30 min. Each reaction was then electrophoresed using 5% native polyacrylamide gels in TBE buffer at 4°C and bands detected after electrotransfer to BioDyne membrane (Pall) using the Phototope-Star detection kit (NEB). For competition reactions an excess of non-labelled DNA fragment was added to reactions prior to addition of labelled fragment.
Cells were harvested by centrifugation at 14 k at 4°C for 2 min and the cell pellets were resuspended in an appropriate volume of lysis solution (10 mM Tris pH 8, 0.5 mM EDTA, lysozyme at 100 µg ml−1, Dnase at 10 µg ml−1 and Calbiochem protease inhibitor coctail set III) and incubated at 37°C for 20 min. One volume of 2× SDS/PAGE loading buffer (without bromophenol blue) was added to the lysate and the contents vortexed thoroughly followed by addition of one-fortieth vol. B-mercaptoethanol and the mixture boiled for 10 min. Protein concentration was determined using a non-interfering protein assay kit (Calbiochem) using a standard curve generated with BSA. Samples were separated on 10% SDS/PAGE Tris-Glycine polyacrylamide gels and transferred to PVDF membrane (Roche) by electroblotting. Membranes were blocked with 2% skim milk in Tris buffered saline pH 7.5, incubated with 1:1000 dilution of primary anti-YycG Ab for 1.5 h, washed and then incubated with 1:40 000 dilution of goat anti-rabbit HRP Ab conjugate for 1 h. YycG protein–antibody complexes were detected with the SuperSignal West Dura substrate (Pierce, Perbio).
We are grateful to Tarek Msadek (Institut Pasteur), in whose group part of this work was carried out. This work was supported by an SFI Investigator Programme Grant (03/IN3/B409), by Health Research Board (Grant RP/101/2002) and by Enterprise Ireland (Grant SC/02/109) to Kevin Devine. Sarah Dubrac's work was supported by research funds from the European Commission (Grant LSHG-CT-2004-503468 BACELL Health), the Centre National de la Recherche Scientifique and the Institut Pasteur (Grand Programme Horizontal n°9). K. I. Varughese's work is supported by the Grant GM54246 from National Institutes of General Medical Sciences, NIH, USPHS. We thank Marion Hulett and Wolfgang Schumann for the gift of strains.