Pho signal transduction network reveals direct transcriptional regulation of one two-component system by another two-component regulator: Bacillus subtilis PhoP directly regulates production of ResD

Authors

  • Stephanie M. Birkey,

    1. Laboratory for Molecular Biology, Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607USA.
    Search for more papers by this author
    • †Metabolism Branch/NCI, NIH, 10 Center Drive, RM 6B-05, Bethesda, MD 20892, USA.

    • §These authors contributed equally to this work.

  • Wei Liu,

    1. Laboratory for Molecular Biology, Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607USA.
    Search for more papers by this author
    • ‡Department of Molecular Pharmacology, Stanford University School of Medicine, Stanford, CA 94305, USA.

    • §These authors contributed equally to this work.

  • Xiaohui Zhang,

    1. Laboratory for Molecular Biology, Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607USA.
    Search for more papers by this author
  • Mary Fran Duggan,

    1. Laboratory for Molecular Biology, Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607USA.
    Search for more papers by this author
  • F. Marion Hulett

    1. Laboratory for Molecular Biology, Department of Biological Sciences, University of Illinois at Chicago, Chicago, IL 60607USA.
    Search for more papers by this author

F. Marion Hulett. E-mail Hulett@uic.edu; Tel. (312) 996 5460; Fax (312) 413 2691.

Abstract

The Bacillus subtilis ResD–ResE two-component system is responsible for the regulation of a number of genes involved in cytochrome c biogenesis and haem A biosynthesis, and it is required for anaerobic respiration in this organism. We reported previously that the operon encoding these regulatory proteins, the resABCDE operon, is induced under several conditions, one of which is phosphate starvation. We report here that this transcription requires the PhoP–PhoR two-component system, whereas other induction conditions do not. The PhoP∼P response regulator directly binds to and is essential for transcriptional activation of the resABCDE operon as well as being involved in repression of the internal resDE promoter during phosphate-limited growth. The concentration of ResD in various phoP mutant strains corroborates the role of PhoP in the production of ResD. These interactions result in a regulatory network that ties together the cellular functions of respiration/energy production and phosphate starvation. Significantly, this represents the first evidence for direct involvement of one two-component system in transcription of a second two-component system.

Introduction

Sensing environmental conditions and responding to them with a timely and appropriate response is vital to the survival of any organism, including bacteria. Towards that end, bacteria often use two-component signal transduction systems that typically consist of a histidine protein kinase (HK), which ‘senses’ the environmental signal and transmits that signal to the second component, the response regulator (RR), through a phosphorylation cascade (Nixon et al., 1986; Ronson et al., 1987; Kofoid and Parkinson, 1988; Msadek et al., 1993; Alex and Simon, 1994; Swanson et al., 1994; Parkinson, 1995). The RR is often responsible for activating or repressing transcription of the genes involved in responding to the individual environmental signal in order to facilitate survival of the cell.

Within its natural environment, the soil, Bacillus subtilis must constantly monitor changing conditions and integrate multiple environmental signals in order to elicit the most appropriate response. We have reported that multiple two-component systems form a signal transduction network to regulate the phosphate starvation (Pho) response in B. subtilis (Hulett, 1996; Sun et al., 1996a). The interconnected pathways involve the PhoP–PhoR system, whose primary role is the phosphate deficiency response, the Spo0A phosphorelay required for the initiation of sporulation, and a recently discovered signal transduction system, ResD–ResE, which also has an essential role in aerobic and anaerobic respiration (Nakano and Hulett, 1997). The PhoP–PhoR and ResD–ResE systems positively regulate the Pho response, whereas Spo0A∼P represses both activator signal transduction systems, thereby repressing the Pho response (Sun et al., 1996a). PhoP directly binds to Pho regulon promoters, where it functions to activate or repress Pho promoter expression (Liu, 1997; Liu et al., 1998a, b; Liu and Hulett, 1997; 1998; Qi and Hulett, 1998a, b) that is essential for the Pho response. The ResD–ResE two-component system activates the phoPR operon by an unknown mechanism (Hulett, 1996; Sun et al., 1996a).

The resABCDE operon encodes gene products, including the ResD–ResE two-component system, that are involved in aerobic and anaerobic respiration. Genes requiring ResD–ResE during aerobic growth include resABC (Sun et al., 1996b), whose products show similarity to cytochrome c biogenesis proteins, ctaA, which is essential for haem A biosynthesis (Svensson and Hederstedt, 1994), and petCBD, which encodes subunits of the cytochrome bf complex (Sun et al., 1996b). ResD–ResE is also required for anaerobic induction of fnr transcription, which regulates the narK and narG operons involved in nitrate respiration (Nakano et al., 1996; Nakano and Hulett, 1997; Nakano and Zuber, 1998).

Azevedo et al. (1993 ) reported detection of an mRNA transcript including resABCDE and one including only resDE in RNA from cultures grown under phosphate-replete conditions. Sun et al. (1996b ) showed that resABCDE was induced under several conditions, which included post-exponential induction in Schaeffer's sporulation medium (SSM with glucose; SSG), and during P i limited growth in low phosphate defined medium (LPDM), and that transcription in either case was initiated at the same site. These authors also determined the transcription initiation site for the resDE internal promoter under phosphate-replete conditions but could detect no transcript in RNA from phosphate-starved cells.

Studies reported here were initiated to determine whether PhoP/PhoR was involved in the Pi-limited resABCDE induction. The data reveal that both transcripts encoding the ResD–ResE two-component system, from the resABCDE operon promoter and the internal resDE operon promoter (Sun et al., 1996b), are directly regulated by the PhoP–PhoR two-component system under phosphate starvation conditions, and that PhoP transcriptional regulation influences ResD concentrations in the cell. The resABCDE operon requires PhoP–PhoR for induction under phosphate starvation conditions, whereas the internal resDE promoter is repressed by PhoP∼P under these same conditions. Furthermore, PhoP binds to both promoters. These results indicate that one two-component system, PhoP–PhoR, directly affects the transcription of another two-component system, ResD–ResE, under phosphate starvation conditions.

Results

Both ResD and PhoP are essential for resABCDE operon induction in response to phosphate starvation

The ResD–ResE two-component system is encoded by the final two genes of a five-gene operon (Sun et al., 1996b). This operon, resABCDE, has been shown to be induced in response to several different growth conditions, including anaerobic growth (Nakano and Hulett, 1997; Nakano and Zuber, 1998), nitrogen starvation and phosphate starvation (Sun et al., 1996b). To investigate further the phosphate starvation induction of the resABCDE operon, we grew strain MH5229 (resA–lacZ, Table 1) in LPDM with an initial phosphate concentration (0.4 mM) that becomes limiting during culture growth (Hulett et al., 1990; Qi et al., 1997). After initial Pho induction of the resA–lacZ fusion (at a time similar to that following hour 10 in Fig. 1A), the culture was separated into two flasks. Increasing the Pi concentration to 10 mM (HPDM) in one flask stopped the low phosphate induction of the resA–lacZ fusion, resulting in decreased β-gal specific activity as growth continued (data not shown), indicating that the induction of the resA–lacZ fusion is due to phosphate starvation. Only low levels of resA–lacZ expression (< 2% of phosphate starvation levels) were seen during growth of this same strain in DM with excess phosphate (10 mM; HPDM), and that was during early exponential growth. Taken as a whole, these data suggest that the resABCDE operon is expressed in response to phosphate deprivation and pose a question concerning the role of PhoP and PhoR in this induction.

Table 1. . Bacterial strains and plasmids. a. All strains used in this study are isogeneic and are derived from JH642. b. All resA–lacZ and resD–lacZ fusions are integrated in single copy at the amyE locus using the pDH32 plasmid, Sun et al. (1996b). c. resD ΩpRC1211 denotes a Campbell integration (insertion duplication) of the plasmid at the internal resD promoter which blocks transcription upstream of this internal promoter (// resD  ).Thumbnail image of
Figure 1.

. Phosphate starvation induction of transcription of the resABCDE operon requires both PhoP and ResD. Strains were grown in LPDM as previously described ( Hulett et al., 1990 ; Qi et al., 1997 ). Growth (•) and β-gal (○) ( Ferrari et al., 1988) from (A), MH5229 ( resA–lacZ  ); (B), MH5202 ( resD resA–lacZ  ); (C), MH5417 ( phoP resA–lacZ  ).

All known Pho regulon genes are induced or repressed under the specific growth conditions described above and are known to be regulated by the PhoP–PhoR two-component system (Hulett et al., 1990; Eder et al., 1996; Hulett, 1996; Sun et al., 1996b; Qi et al., 1997). As it was reported that the resABCDE operon is subject to autoregulation by ResD–ResE during stationary growth in a complex sporulation medium (Sun et al., 1996b), it was necessary to ask whether autoregulation also occurred during phosphate starvation. To test these possibilities, we introduced the resA–lacZ fusion into strains containing mutations within the response regulators of these two regulatory systems; a non-polar insertion mutation within the resD gene (strain MH5202) or an in-frame deletion within the phoP gene (MH5417). These strains were grown, along with the parental strain (MH5229), under phosphate starvation conditions. Figure 1 shows that the parental strain, MH5229 (Fig. 1A) was induced in response to phosphate depletion, as previously noted, but neither the resD mutant strain, MH5202 (Fig. 1B), nor the phoP mutant strain, MH5417, (Fig. 1C) induced the resA–lacZ fusion. Similarly, expression of the resA–lacZ fusion in either a resE or phoR mutant strain was reduced, although to a lesser extent (resE < 16% and phoR < 10% of the parental strain). When the phoP mutant strain (MH5417) was transformed back to phoP + (MH5524), the phosphate starvation-induced resA–lacZ expression returned to levels observed in the parental strain, MH5229(data not shown). These data suggest that under phosphate starvation induction conditions the resABCDE operon is regulated by the PhoP–PhoR two-component system as well as being autoregulated by the ResD–ResE two-component system. Although each of these two-component systems is essential, neither is sufficient for induction during phosphate starvation.

In contrast to the results during phosphate starvation, PhoP is not required for resA induction during anaerobic growth or post-exponential growth in SSG. 2Figure 2A shows that resA is induced post-exponentially in the parent strain, MH5229 and in the isogeneic phoP mutant strain, MH5417, but not in the resD mutant strain, MH5229. Similarly, during anaerobic induction of resA, ResD is required (Nakano and Zuber, 1998), but not PhoP (Fig. 2b). These data indicate that the PhoP requirement for resA induction is restricted to the condition of phosphate starvation.

Figure 2.

. PhoP is not required for post-exponential induction of resABCDE during aerobic or anaerobic growth in SSG. Strains were grown in SSG and monitored for β-gal ( Ferrari et al., 1988). β-Gal from strain MH5229 ( resA–lacZ  ) (▪); from strain MH5417 ( phoP resA–lacZ  ) (□). A. Aerobic growth conditions. B. Anaerobic growth conditions.

The resABCDE transcript is the major transcript for ResD–ResE during phosphate starvation

The ResD–ResE two-component system is known to be encoded on two different transcripts; a major resABCDE operon transcript and a minor resDE internal transcript (Azevedo et al., 1993; Sun et al., 1996b). The same resA transcription initiation site was identified using RNA isolated from cells during phosphate starvation as during post-exponential induction of resA in cells grown in SSG (Sun et al., 1996b), a medium in which cells exit exponential growth because of nitrogen limitation (Wray et al., 1998). The level of message in RNA from phosphate-starved cells was ≈threefold higher than in SSG grown cells. Although no resD transcript could be identified in RNA from the phosphate-starved cells, similar low levels of transcript initiating at identical start sites could be detected in RNA from cells growing exponentially in SSG or LB medium, a stage of growth when resD–lacZ promoter fusions indicated low levels of promoter activity in those media (Sun et al., 1996b). Northern Blot analysis (Azevedo et al., 1993) detected low resDE transcript levels in RNA from exponential and post-exponential cells grown in Spizzen's minimal medium (SMM), a defined medium similar to LPDM but buffered with phosphate such that SMM cultures never become phosphate limiting. To assess the relative promoter strength of resA and resD before and during phosphate starvation conditions in LPDM, we compared the expression of the lacZ promoter fusion in strain MH5229 (resA–lacZ, Table 1) with MH5004 (resD–lacZ, Table 1) as shown in Fig. 3. The resD promoter fusion was expressed only at low levels and only during exponential growth. The β-gal from the resA fusion transcript is much more abundant than that from the resD promoter fusion under phosphate starvation conditions. These data corroborate the primer extension data of Sun et al. (1996b), which detected no transcript from resD during phosphate starvation but an abundant resA message, suggesting the major transcript encoding resDE under phosphate starvation conditions is from the resA promoter.

Figure 3.

. resA and resD promoter fusion data show that the resA promoter is a stronger promoter under phosphate starvation conditions than the resD promoter. Strains were grown in LPDM ( Hulett et al., 1990 ; Qi et al., 1997 ) and monitored for β-gal ( Ferrari et al., 1988 ). Growth (filled symbols) and β-gal (open symbols) from strains (•, ○), MH5229 ( resA–lacZ  ); or (▪, □) MH5004 ( resD–lacZ  ).

ResD protein levels during phosphate starvation are regulated by PhoP

The reduction in transcription from the resA promoter, the major transcript encoding ResD, in the phoP  strain (Fig. 1) may result in reduction in ResD protein levels in these strains. Using antibodies raised against ResD, we performed Western immunoblotting on samples taken hourly after Pho induction from strains MH5229 (parent), MH5417 (phoP ), MH5561 (resDΩpRC1211) and MH5594 (phoP resDΩpRC1211) (Table 1). The results, as illustrated in Fig. 4 from samples taken during maximal resA–lacZ induction, support the resA–lacZ data and show that the ResD protein levels were greatly reduced (≈20% of the parent) in the phoP mutant background (MH5417). A strain in which the transcription from resA is blocked, such that the resDE operon is transcribed only from the internal resD promoter, MH5561 (resDΩpRC1211), showed ResD at levels that were significantly lower than either the parental strain (MH5229) (≤ 3%) or the phoP mutant strain (MH5417). This result suggested that the internal resD promoter is responsible for only minute levels of ResD–ResE production in the wild-type background during Pi-limited post-exponential growth, and corroborates the data in Fig. 2, which show that the resA promoter is the major promoter for production of the ResD–ResE two-component system under phosphate starvation conditions. These data raise the question of why the ResD concentration is not equal in the phoP mutant strain (MH5417) and the strain producing ResD only from the transcript from the internal resD promoter (MH5561). Surprisingly, when strain MH5561 (resDΩpRC1211), which expresses resDE only from the internal promoter, carries a second mutation in phoP (MH5594 phoP resDΩpRC1211), significantly higher levels of the ResD protein accumulated than in the phoP + parental strain (MH5561), suggesting that PhoP represses expression from this internal promoter. This observation explains the origin of the ResD (20% of parental MH5229) found in the phoP mutant strain (MH5417), whereas the transcriptional induction from the resA promoter decreased at least 600-fold. In the same phoP mutant strain (MH5417), expression from the resD internal promoter is relieved of the repression by PhoP and thus can express ResD at this level equal to ≈20% that of its parental strain, MH5229. Cumulatively, these data support the hypothesis that the resA promoter, and thus production of ResD and ResE, is positively regulated by PhoP under phosphate starvation conditions, and also suggest that the internal resDE promoter is repressed by PhoP under these same conditions.

Figure 4.

. Quantfication of ResD protein levels by Western Immunoblot analysis demonstrates that the resDE promoter contributes only a small fraction to total ResD levels. Strains were grown in LPDM ( Hulett et al., 1990 ; Qi et al., 1997 ) and samples (200 μg total protein) were removed hourly. The immunoblot was performed as described in Experimental procedures . The top panels depict the actual immunoblot using samples taken 11 h after Pho induction, whereas the lower panel represents quantification of these blots by a Bio-Rad Imaging Densitometer700 using the MOLECULAR IMAGER software. The amount of ResD protein is expressed as a percentage of the parental strain, which is considered to express 100% ResD. Strains are (parent), MH5229; ( phoP  ), MH5417; (// resD  ), MH5561; (// resD phoP  ), MH5594.// resD denotes that transcripts originating upstream of resD (from resA ) are blocked ( see Table 1 ).

PhoP binds directly to both the resA promoter and the internal resD promoter

The above data indicate that both the resA promoter and the internal resD promoter are subject to regulation by PhoP, the response regulator of the PhoP–PhoR two-component system, under phosphate starvation conditions. This could be the result of direct regulation by PhoP binding to either, or both of the promoters, or indirect regulation through an intermediate protein(s). To test whether PhoP binds to the resA promoter as it binds to other Pho-regulon promoters, gel shift assays were conducted by incubating the labelled resA promoter fragment from pES16 with various concentrations of PhoP and 1 μM *PhoR (Liu and Hulett, 1997), in the presence or absence of 4 mM ATP (Fig. 5). Both PhoP∼P (in the presence of *PhoR and ATP) and unphosporylated PhoP (in the presence of *PhoR, but absence of ATP) caused a mobility shift of the resA promoter fragment. However, PhoP∼P binds to the resA promoter a lower concentration than unphosphorylated PhoP (Fig. 5).

Figure 5.

. Gel shift assays of the resA promoter bound by PhoP and Pho∼P. The 208 bp resA promoter fragment was incubated with PhoP and *PhoR in the presence or absence of ATP. The concentration of PhoP used in each case is indicated above the respective lane. Each reaction contained 1 μM *PhoR. If PhoP∼P was needed, a final concentration of 4 mM ATP was added. The gel shift assay was performed as described previously ( Liu and Hulett, 1997 ).

To investigate this binding by PhoP and identify its binding site(s), we performed DNase I footprinting analysis on the resA promoter region from pES16 with the PhoP protein.

The data in Fig. 4, show that a region of the resA promoter from −6 to −72 on the coding strand, relative to the transcription start site, and from −3 to −88 on the non-coding strand was protected by PhoP. There was one hypersensitive site on each strand upon PhoP∼P binding. Again, both phosphorylated and unphosphorylated PhoP bound to this promoter, but PhoP∼P showed better protection, especially on the non-coding strand. In addition, PhoP∼P expanded the binding region further upstream and downstream than the unphosphorylated PhoP, albeit the highest concentration (6.7 μM) of unphosphorylated PhoP showed some protection in these extended regions. These data are consistent with the pattern of PhoP binding to most of the Pho promoters (Liu, 1997; Liu and Hulett, 1997; 1998; Liu et al., 1998a, b) and support the results of the mobility shift assays that suggested that PhoP binds directly to the resA promoter. Coupled with the lacZ fusion data, these results indicate that PhoP binds directly to the resA promoter and is essential for induction of the major operon encoding the ResD–ResE two-component system, through regulation of the resABCDE operon under phosphate starvation conditions.

Investigation of possible interactions between PhoP and the internal resD promoter, which were suggested by the Western immunoblot data, was performed by DNase I footprinting analysis of the resD promoter region from pWL42 with PhoP protein. The data, shown in Fig. 7, indicate that PhoP∼P does bind to the internal resD promoter, protecting a region from −45 to +132 on the coding strand, and from −50 to +125 on the non-coding strands, relative to the transcription start site. Multiple hypersensitive sites were created after PhoP∼P bound to the promoter. Footprinting analysis of a resD promoter clone that contained DNA 5′ of that in pWL42, pRC1211, showed no additional PhoP binding (data not shown). Interestingly, only phosphorylated PhoP bound to this promoter at all concentrations tested. The PhoP binding data, coupled with the data obtained from the Western immunoblots, suggest that PhoP∼P binds directly to the internal resD promoter and represses expression from this internal promoter under phosphate starvation conditions.

Figure 7.

. DNase I footprinting analysis of the PhoP or PhoP∼P protein on the internal resD promoter. Top, varying amounts of PhoP incubated with *PhoR (1.4 μg) in the presence or absence of 4 mM ATP were mixed with the 222 bp resD promoter probe and DNase I footprinting experiments were performed on both end-labelled coding and non-coding strands. The concentration of PhoP used in each reaction is, from left to right, 0 nM, 55 nM, 275 nM, 1.38 μM and 6.7 μM. The lanes in which PhoP∼P was used are labelled as ‘+ATP’ across the top, F, PhoP-free lanes; G, ‘G’ sequencing reactions. The dashed lines represent sites bound by PhoP∼P. The hypersensitive sites are marked with a dark arrowhead. Bottom, the sequence of the resD promoter region, which is marked according to the top panel. The transcription start site for resD is underlined and marked with a ‘+1’. Likewise, the −10 region is underlined and marked with a ‘−10’. Consensus 6 bp sequences for PhoP binding are indicated by bold underline. The translation start site is labelled with an arrow.

Discussion

Many B. subtilis Pi starvation-regulated promoters, in fact all that have been studied in detail, require PhoP∼P for transcription induction or repression both in vivo and in vitro (Chesnut et al., 1991; Jensen et al., 1993; Hulett et al., 1994; Eder et al., 1996; Hulett, 1996; Sun et al., 1996a, b; Liu and Hulett, 1997; 1998; Qi et al., 1997; Liu et al., 1998a, b; Qi and Hulett, 1998a, b). However, recent results from analyses of proteins induced under Pi starvation were interpreted as showing that not all Pi starvation-induced proteins were dependent on the Pho regulon signal transduction regulators, PhoP and PhoR (Eymann et al., 1996). These observations, coupled with the knowledge that ResD is required for resA transcription under other inducing conditions known (Nakano and Zuber, 1998; Sun et al., 1996b), made it important to ask what, if any, was the role of PhoP/PhoR in transcription of resA during Pi starvation induction.

The results presented here show an essential role for PhoP–PhoR in the production of ResD during phosphate starvation. First, the lacZ reporter studies showed that the resABCDE transcript was the major transcript encoding ResD–ResE during phosphate starvation (Fig. 3), corroborating the primer extension data of Sun et al. (1996b), which found an abundant transcript initiated at resA, but none initiated from the internal resD promoter in RNA from phosphate-starved cells. Further, both PhoP and ResD were essential, but neither was sufficient for this resA transcription (Fig. 1), although PhoP has no role in resA transcription under other resA induction conditions (Fig. 2). Cellular concentration of ResD protein in various mutant strains confirmed that the resABCDE transcript was responsible for ResD production in phosphate-starved cells, but that a PhoP mutant strain produced low levels of ResD dependent on the internal resDE transcript (Fig. 4). The latter observation suggests PhoP represses the internal resDE promoter. Finally, the role of PhoP in regulation of the resA and resD promoters appears to be direct (Figs 5–7) as evidenced from DNA footprinting analysis.

Figure 6.

. DNase I footprinting analysis of the PhoP or PhoP∼P protein on the resA promoter. Top, varying amounts of PhoP incubated with *PhoR (1.4 μg) in the presence or absence of 4 mM ATP were mixed with the 201 bp resA promoter probe and DNase I footprinting experiments were performed on both end-labelled coding and non-coding strands. The concentration of PhoP used in each reaction is, from left to right, 0 nM, 55 nM, 275 nM, 1.38 μM and 6.7 μM. The lanes in which PhoP∼P was used are labelled as ‘+ATP’ across the top. F, PhoP-free lanes; G, ‘G’ sequencing reactions. The solid lines represent the PhoP and PhoP∼P binding regions, whereas the dashed lines represent sites bound only by PhoP∼P. The hypersensitive site is marked with a dark arrowhead. Bottom, the sequence of the resA promoter region and is marked according to the top panel. The transcription start site for resA is underlined and marked with a ‘+1’. Likewise, the −10 region is underlined and marked with a ‘−10’ Consensus 6 bp sequences for PhoP binding are indicated by bold underline. The translation start site is labelled with an arrow.

The PhoP binding pattern within the resA promoter is similar to other PhoP-regulated promoters(Liu, 1997; Liu and Hulett, 1997; 1998; Liu et al., 1998a, b), in that (i) both PhoP and PhoP∼P bind to these promoters; (ii) PhoP∼P extends the binding site further upstream and/or downstream, and has a higher binding affinity; and (iii) the core PhoP-binding region (region bound by PhoP and PhoP∼P) is located from about −22 to −60 in all these promoters(Liu and Hulett, 1998), including resA. A feature unique to the resA promoter among PhoP-regulated promoters is that it contains only two of the consensus PhoP binding repeats in the core binding region (Fig. 6), instead of at least four as seen in other Pho promoters studied. Ongoing expression and footprinting analyses of a site-specific mutagenized PhoP regulated promoter (S. Eder and F. M. Hulett, unpublished) indicate that PhoP binds to 6 bp repeated TT(A/T/C)ACA sequences separated by 4–5 bp, and that PhoP binding to one set of consensus repeats requires higher concentrations of PhoP or PhoP∼P (> 200 nM) than to promoter regions containing four or more repeats (55 nM) (Liu and Hulett, 1998). These observations are in good agreement with PhoP binding reported here for the resA promoter, which contains only two consensus repeats. In addition, it must be noted that PhoP alone is essential but insufficient for expression of resA during Pi starvation, and that the role of ResD, direct or indirect, remains an interesting question.

Analysis of the internal resDE operon promoter revealed a second role, that of a repressor, for PhoP∼P in the control of resD expression. PhoP has previously been reported to repress expression of operons involved in teichoic acid synthesis, tagAB and tagDEF (Liu et al., 1998a; Qi and Hulett, 1998b). The binding patterns between these operons are similar in that PhoP∼P binds to regions that overlap the transcription initiation sites as well as far into the coding regions of these genes (up to +125 for resD, +115 for tagA and +168 for tagD ) (Liu et al., 1998a). This binding pattern suggests that PhoP∼P may repress expression of these genes by blocking RNA polymerase from binding to the promoter (a competition mechanism demonstrated for tagD repression by PhoP∼P (Qi and Hulett, 1998a), or possibly by stopping the progression of the transcription apparatus. It is interesting to note that in two promoters repressed by PhoP, tagD and resD, each promoter contains only one set of PhoP binding consensus sequences located in the promoter region (Fig. 7), and that only PhoP∼P shows DNA binding. This cannot, however, be viewed as a phenomenon specific for repression by PhoP∼P because both PhoP and PhoP∼P bind to the tagA promoter region, a promoter that contains four repeats of the consensus sequence, albeit on the non-coding strand.

The studies reported here show that PhoP∼P has a dual role in controlling the expression of resD and the resulting concentration of ResD within the cell during phosphate starvation. Without repression by PhoP∼P, either in a PhoP mutant strain or during growth with adequate phosphate, the resDE internal promoter can supply low levels of ResD. When PhoP is phosphorylated, it functions, along with ResD, to dramatically increase the expression of the resABCDE operon such that transcription from the internal resDE promoter is not needed, or is perhaps even detrimental, and PhoP∼P represses that transcription.

Signal transduction pathways in B. subtilis cannot be viewed as linear regulatory pathways, but rather as signal transduction networks that gather diverse input from environmental and intracellular signals and process that information to determine the appropriate response based on the signals received in a given situation, thereby establishing the hierarchy of the environmental signal responses. Complex ‘two-component’ systems have been reported in B. subtilis, most notably the phosphorelay leading to the phosphorylation of Spo0A (Burbulys et al., 1991). A variety of signals with either positive or negative effects on the concentration of cellular levels of Spo0A∼P are accommodated by multiple kinases, regulators of kinases, aspartate phosphatases and regulators of aspartate phosphatases adjusting the flow of phosphate through proteins in the his-asp-his-asp phosphorelay (Perego et al., 1994; 1996; Perego and Hoch, 1996a, b; Wang et al., 1997). Other two-component systems such as ComA–ComP exert their influence on Spo0A∼P levels indirectly via transcriptional regulation of intermediate proteins such as RapA, one of the aspartate phosphatases. In contrast, data presented here implicate one two-component system as essential and directly involved in transcriptional regulation of another two-component system under a specific induction condition.

The regulatory coupling of the PhoP–PhoR and ResD–ResE signal transduction systems that functions during phosphate-depleted growth is illustrated (Fig. 8) in a working model consistent with our current cumulative data. Expression of the Pho regulon genes, including the phoPR operon, is dependent on PhoP–PhoR. Low constitutive expression of phoPR, but no induction upon Pi limitation, is observed in either a phoP or phoR mutant (Hulett et al., 1994), which indicates that PhoR is required to respond to a signal to activate the Pho deficiency response via PhoP∼P. ResD–ResE are required for full induction of the Pho response; a non-polar mutation in resD results in reduced induction of Pho regulon genes to levels ≈20% that of the wild-type induction. Thus, autoinduction of the phoPR operon is insufficient to initiate a full Pho response, a response which also requires ResD. (The role of ResD in full induction of the Pho response is undetermined. It may directly affect phoPR or all Pho promoters, or indirectly regulate unknown genes whose products affect the activation state of PhoR or PhoP or affect the transfer of phosphate between them.) A co-dependent induction of phoPR and resDE in a positive feed back loop was revealed when PhoP was shown essential for resABCDE induction initiated in response to phosphate limitation. The role of PhoP is judged essential but insufficient because ResD, made available from low level expression of the internal resDE promoter, is also essential for resABCDE induction. Increased expression of resABCDE results in increased levels of ResD protein, promoting a full induction of the Pho response that provides a high-affinity Pi transport system (pstSCAB1B2) (Qi et al., 1997), degradative enzymes (phoA, phoB, phoD ) to harvest phosphate from teichoic acids (Hulett et al., 1990; Eder et al., 1996; Hulett, 1996), and induction of a non-Pi-containing replacement cell wall anionic polymer (teichuronic acid, tuaA) while it represses synthesis of teichoic acids (tagD ) (Ellwood and Tempest, 1969; Liu and Hulett, 1998; Liu et al., 1998a; Qi and Hulett, 1998b) and the internal resDE promoter. During phosphate starvation, the resABCDE operon provides components essential for electron transport, such as haem A biosynthesis and cytochrome c biogenesis (Azevedo et al., 1993; Sun et al., 1996b), needed for assimilation of the transported Pi into ATP. When faced with ever decreasing concentrations of Pi, the failing Pho system is repressed by Spo0A∼P (Jensen et al., 1993), which inhibits transcription of the phoPR operon via AbrB (whose participation is essential for the 20% phoPR induction remaining in the resD mutant strain) as well as through ResD–ResE (Hulett, 1996; Sun et al., 1996a) and positions the cell to initiate sporulation. Further studies are required (i) to understand the mechanism(s) of action of ResD, in the shared responsibility with PhoP, for induction of resABCDE and phoPR ; (ii) to determine what signal ResE responds to during phosphate limited growth; and (iii) to determine the mechanism of Spo0A repression of resABCDE transcription.

Figure 8.

. Model depicting the signal transduction network leading to Pho induction in B. subtilis , a phosphate deficiency response. Proteins are indicated by ovals, whereas genes/operons are symbolized by rectangles. Solid lines indicate direct interaction has been demonstrated. Dashed lines are used for interactions that could either be direct or indirect. Positive regulation is labelled with ↑ and +, whereas repression is labelled with ⊥ and −. Two-component system members are labelled as either the histidine kinase (HK) or the response regulator (RR) for each case. The interactions described in this paper are noted by a thick black line.

Experimental procedures

Bacterial strains and plasmids

The bacterial strains and plasmids used in this study are listed in Table 1. Plasmids were constructed as follows. Fragments are numbered according to the sequence submitted to GenBank with accession number L09228: pES16 was constructed by amplifying the resA promoter region contained within a 201 bp fragment of the B. subtilis chromosome from strain JH642. Primers FMH164 (5′-TTGAATTCG-18101ATGCCAGAGAGTTACG18116-3′) and FMH165 (5′-TTGGATCCG- 18299GTCCGAATGAATAAACGC18283-3′) were used to amplify this region. The PCR fragment was then digested with BamHI and EcoRI and cloned into the BamHI and EcoRI sites of pJM103 to create pES16; pRC1211 was constructed by amplifying a 222 bp region of the internal resD promoter using B. subtilis strain JH642 DNA as template and primers FMH120 (5′-TTGAATTC-21545GCCGTAATCGGTTTTGC21561-3′) and FMH121 (5′-TTGGATCC-21756GGCGGCGAATTCTGGC21739-3′). The PCR fragment was digested with EcoRI and cloned into the EcoRI site of pJM103 to create pRC1211. pWL42, containing the internal resD promoter and 5′ coding region, was constructed by amplifying a 257 bp fragment, using JH642 DNA as template and primers FMH200 (5′-GTGAATTC-21626GAAAGCGACACAGTTCT21643-3′) and FMH201 (5′-GTTTCGAA21883GGCATCATCAGATCAAG21867-3′), which was cloned into pCR2.1 and sequenced.

General methods

Transformation of B. subtilis was performed by the two-step transformation method of Cutting and Vander Horn (1990). Transformants were selected for drug resistance on tryptose blood agar base (Difco) plates containing glucose (0.5%) and either chloramphenicol (5 μg ml−1), tetracycline (10 μg ml−1) or spectinomycin (100 μg ml−1).

Media, growth conditions and enzyme assays

Spizizen's mimial medium (SMM, a chemically defined medium buffered with phosphate) and Schaeffer's sporulation medium (SSM or DSM; SSG, with glucose 0.5%) were formulated according to Harwood and Cutting (1990). For phosphate starvation induction of the resA or resD promoter fusions, the cells were cultured at 37°C in low phosphate defined medium (LPDM) (Hulett et al., 1990; Qi et al., 1997) with aeration. This medium was supplemented with 10 mM K2HPO4 for growth under high phosphate conditions (HPDM). β-Galactosidase-specific activity was determined according to the method of Ferrari et al. (1988) and is expressed in units per mg of protein. The unit used is equivalent to 0.33 nmol of o-nitrophenol produced per minute.

Western immunoblot analysis:

Proteins (200 μg total protein for each) were separated on sodium dodecyl sulphate–12% polyacrylamide gel electrophoresis (SDS–PAGE) gels and then were transferred to Immobilon membranes (Millipore) by electrophoretic blotting using a Trans-Blot cell (Bio-Rad) at 400 mA for 1 h in CAPS buffer [3-(cyclohexylamino)-1-propanesulphonic acid], which consists of 10 mM CAPS and 10% methanol brought to pH 11 with NaOH. The immunostaining procedure was performed with the Immun-Star Chemiluminescent Protein Detection System (Bio-Rad). The primary antibody was B. subtilis anti-ResD (raised against purified ResD by CoCalico, Reamstown, PA, USA). The secondary antibody was APase-conjugated goat anti-rabbit immunoglobulin G (Bio-Rad). The data were quantified by a Bio-Rad Densitometer700 using the MOLECULAR ANALYST software (Bio-Rad).

Purification of PhoP and *PhoP proteins

PhoP and *PhoR (the cytoplasmic domain of PhoR) were purified as described in (Liu and Hulett, 1997).

DNase I footprinting

The resA promoter fragment from pES16 was digested with either BamHI, for the coding strand, or EcoRI for the non-coding strand and was end-labelled with Klenow fragment in the presence of [α-32P]-dATP. The insert was then released by digestion with either EcoRI (for the coding strand) or BamHI (for the non-coding strand). The resD internal promoter fragment from pWL42 was digested with either BamHI, for the coding strand, or XbaI, for the non-coding strand, and was end-labelled with Klenow fragment in the presence of [α-32P]-dATP. The insert was then released by digestion with either XbaI (for the coding strand) or BamHI (for the non-coding strand). Purification of the probes and the DNase I Footprinting experiments were performed according to Liu and Hulett (1997).

Footnotes

  1. †Metabolism Branch/NCI, NIH, 10 Center Drive, RM 6B-05, Bethesda, MD 20892, USA.

  2. ‡Department of Molecular Pharmacology, Stanford University School of Medicine, Stanford, CA 94305, USA.

  3. §These authors contributed equally to this work.

Acknowledgements

We wish to thank Elena Sharkova for construction of pES16 containing the resA promoter, Ruth Chesnut for constructing pRC1211 which contains the resD internal promoter and Lei Shi for assistance with computer graphics. This work was supported by Public Health Service Grant GM33471 from the National Institutes of Health.

Ancillary