The cytoplasmic kinase domain of PhoR is sufficient for the low phosphate-inducible expression of Pho regulon genes in Bacillus subtilis


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


PhoP–PhoR, one of three two-component systems known to be required to regulate the pho regulon in Bacillus subtilis, directly regulates the alkaline phosphatase genes that are used as pho reporters. Biochemical studies showed that B. subtilis PhoR, purified from Escherichia coli, was autophosphorylated in vitro in the presence of ATP. Phosphorylated PhoR showed stability under basic conditions but not acidic conditions, indicating that the phosphorylation probably occurs on a conserved histidine residue. Phospho–PhoR phosphorylated its cognate response regulator, PhoP in vitro. B. subtilis phoR was placed in the Bacillus chromosome under the control of the Pspac promoter, which is IPTG inducible. The wild-type phoR, under either native promoter or Pspac promoter with IPTG induction, resulted in a similar level of alkaline phosphatase production. Under high phosphate conditions, strains containing wild-type phoR, or phoR mutant gene products that lacked either the periplasmic domain, or both N-terminal transmembrane PhoR sequences or various extended N-terminal sequences, showed no significant APase production. Under phosphate starvation conditions, in the presence of IPTG, all strains containing mutated phoR genes showed alkaline phosphatase induction patterns similar to that of the wild-type strain, although the fully induced level was lower in the mutants. The decrease in total alkaline phosphatase production in these mutant strains can be compensated completely or partially by increasing the copy number of the mutant phoR gene. These in vivo results suggest that the C-terminal kinase domain of PhoR is sufficient for the induction of alkaline phosphatase expression under phosphate-limited conditions, and that the regulation for repression of APase under phosphate-replete conditions remains intact.


The phoP–phoR operon of Bacillus subtilis encodes one of the three two-component signal transduction systems responsible for phosphate-dependent control of a number of genes involved in the phosphate deficiency response (Hulett, 1996; Sun et al., 1996). The operon was shown to be transcribed at a low level under high phosphate conditions from a promoter proximal to phoP gene, but transcription increased three- to fourfold when the phosphate concentration in the culture medium decreased to 0.08–0.1 mM (Hulett, 1996; Sun et al., 1996; Qi et al., 1997). The induction of phoPR transcription observed during growth in low phosphate medium is dependent on PhoP and PhoR (Hulett et al., 1994a; Qi et al., 1997; Sun et al., 1996). The increased level of phoPR transcription results in about a 500-fold increase in total alkaline phosphatase (APase) specific activity, and over 5000-fold increase in pst operon transcription (Hulett et al., 1994b; Qi et al., 1997). Genes that are directly regulated by PhoP/R are called Pho regulon genes. These genes include two structural genes, phoA and phoB, encoding the two major APases that account for 98% of the total APase activity induced under low-phosphate conditions, a third APase structure gene, phoD, encoding an enzyme with both alkaline phosphodiesterase and APase activity (Eder et al., 1996), the tuaABCDEFGH operon responsible for synthesis of an anionic cell wall polymer, teichuronic acid, that replaces teichoic acid in the cell walls of phosphate-starved cells (Liu et al., 1998a), the pstSACB1B2 operon encoding the high-affinity phosphate transport system (Qi et al., 1997), the divergently transcribed operons tagAB and tagDEF encoding products involved in teichoic acid synthesis, whose transcription was repressed by PhoP and PhoR (Liu et al., 1998a), and the phoPR operon, encoding PhoP and PhoR (Seki et al., 1987; Seki et al., 1988). The activation/repression of the Pho regulon gene transcription requires PhoP-phosphate, which binds to the promoter region of the phoA, phoB, pstS, tuaA and tagA/D genes (Liu and Hulett, 1997; Liu et al., 1998a,b; Qi and Hulett, 1998a,b). Alkaline phosphatases have been used as reporters in the study of Pho regulon genes in B. subtilis.

PhoP and PhoR show significant sequence similarity with PhoB and PhoR of Escherichia coli (Seki et al., 1987; 1988), both of which belong to the superfamily of conserved histidine kinase/response regulator signal transduction proteins originally identified in many diverse bacteria (Parkinson and Kofoid, 1992), as well as in several eukaryotic cells (Chang and Meyerowitz, 1995). A null mutation in the gene encoding the response regulator protein PhoP abolishes the low phosphate-inducible expression of APase genes and other Pho regulon genes.

The B. subtilis PhoR protein has an N-terminal membrane associated domain, which contains two hydrophobic transmembrane sequences flanking a large ‘periplasmic’ domain, an N-terminal structure similar to that of a number of sensor kinases (for a review, see Parkinson and Kofoid, 1992; Stock et al., 1989). The ‘periplasmic’ domain of sensor kinases has been implicated in proteins such as NarX, NarQ and VirA as the region receiving environmental signals or interacting with other signal-binding proteins to elicit an adaptive response (Cavicchioli et al., 1990; Chang and Winans, 1992).

PhoR of B. subtilis also has a large cytoplasmic domain, connecting the N-terminal domain and the C-terminal highly conserved kinase domain, which has been referred to as C2 (Scholton and Tommassen, 1993). This domain, absent from many sensor kinases, shares homology with the same domain found in E. coli PhoR. Unlike B. subtilis PhoR, E. coli PhoR does not have a large periplasmic domain. Several E. coli phoR mutants have been used to explore its sensing function. One mutant E. coli phoR gene encoded a PhoR(PhoR1084) protein that lacked the amino-terminal hydrophobic region of the intact protein. This protein was autophosphorylated in the presence of ATP, and the phosphate group on the protein was rapidly transferred to PhoB (Makino et al., 1989). The soluble protein, PhoR1084, retained only the positive function in vivo because the strain producing this protein continued to produce APase under both low- and high-phosphate-culturing conditions. In fact, the APase activity under excess Pi conditions was higher than that of the wild-type strain (Yamada et al., 1990). In another study (Scholton and Tommassen, 1993), deletions of the sequence within the large cytoplasmic domain (C2) between the membrane region and the kinase domain caused constitutive expression of APase in a phoR phoM mutant background, indicating that the C2 domain might serve as a signal-sensing domain, and the mutant PhoR proteins might be locked in the kinase function of the protein (Scholton and Tommassen, 1993).

The presence of the large periplasmic domain and the extended cytoplasmic domain in B. subtilis PhoR provides two potential signal-sensing domains leading to the Pho response in B. subtilis. It is not known whether either of the two domains might be responsible for signal sensing, nor is it known whether other factors, if any, interact with the PhoR protein. The unique structural features of B. subtilis PhoR make it an interesting subject to study the diversity in the mechanism for phosphate regulation in Gram-positive and Gram-negative bacteria.

In this study, we provide evidence that PhoR acts as a histidine kinase in vitro. We show that neither deletions of the periplasmic domain nor the complete membrane-associated domain changed the inducibility of the Pho response. Strains containing the mutant proteins retained their dependency on the level of inorganic phosphate in the culture medium to initiate the Pho response, suggesting that there are more complex mechanisms, in addition to autophosphorylation and phosphotransfer initiated by signal sensing of the PhoR protein, involved in phosphate regulation in B. subtilis.


Overproduction and characterization of PhoR

To test the kinase activity of PhoR suggested by the sequence similarity to other kinases, we PCR-amplified the complete B. subtilis phoR gene and cloned it into plasmid pBluescript SK, under the control of a T7 promoter. This plasmid, pLS6, was transformed into the E. coli expression strain BL21(DE3). Cells harbouring pLS6 were harvested after induction by IPTG for 2 h. The E. coli-produced PhoR protein remained in the insoluble fraction, and in the presence of [γ-32P]-ATP showed autophosphorylation (Fig. 1, Lane 1), whereas the insoluble fraction of non-induced E. coli cells did not (Fig. 1, Lane 3). When 1 μg of PhoP protein was included in the reaction mixture containing the membrane fraction of PhoR and [γ-32P]-ATP, PhoP was phosphorylated (Fig. 1, Lane 4), whereas PhoP alone could not be phosphorylated by [γ-32P]-ATP (Fig. 1, lane 2). This result showed that the B. subtilis PhoR protein produced in E. coli functions as an autokinase and a kinase of PhoP in vitro.

Figure 1.

. Autophosphorylation of PhoR and phosphorylation of PhoP by PhoR. Insoluble protein (0.3 μg) prepared from an IPTG induced culture of Escherichia coli strain containing IPTG-inducible phoR genes (Lane 1 and 4) or from a non-induced culture of the same E. coli strain (2 and 3) was incubated with (lanes 2 and 4) or without (lane 1 and3) 1 μg of purified PhoP in the presence of 5 μCi [γ-32P]-ATP (0.05 nM) as described in Experimental procedures. The results were analysed by SDS–PAGE followed by autoradiography. Lane 5, protein size standard. The numbers on the right show the molecular mass markers in kDa.

Overexpression and purification of GST–*PhoR

As the insoluble nature of the complete PhoR protein complicated in vitro studies, we constructed an N-terminal-truncated PhoR, *PhoR, in a glutathione-S-transferase fusion vector. This construct provides a simplified purification of the soluble form of the protein. The *PhoR shares homology with a similarly truncated form of the E. coli PhoR protein, PhoR1084 (Seki et al., 1988; Makino et al., 1989; Yamada et al., 1990). Having its N-terminal 230 amino acids deleted, *PhoR retains the majority of this extended cytoplasmic domain, and the conserved kinase domain. The GST–*PhoR protein was successfully overexpressed in E. coli, and it accounted for ≈15% of the total cellular protein (Fig. 2, lane 2). Approximately 60% of the GST–*PhoR protein expressed at 30°C was in the soluble fraction after centrifugation of 100 000 × g for 1 h. The soluble fusion protein was bound to a glutathione–agarose column, eluted by 10 mM glutathione and shown to be a 67 kDa protein by SDS–PAGE (data not shown). Alternatively, glutathione–agarose beads with a bound fusion protein were mixed with 10 U of thrombin. After thrombin cleavage, ≈95% of the truncated *PhoR (41 kDa) protein was collected in the flowthrough fraction from the mixture (Fig. 2, lane 5). The bound GST protein (27 kDa) was eluted by 10 mM glutathione (Fig. 2, lane 4). The thrombin-cleaved *PhoR, based on sequence prediction, has the deleted N-terminal 225 aa replaced with 8 aa (Gly-Ser-Pro-Gly-Ile-Gln-Ser-Leu) followed by Thr-226 of the wild-type PhoR. PhoR173, purified to homogeneity using the same procedure, has the N-terminal membrane-bound domain (172 aa) deleted, but has the extended cytoplasmic domain and the conserved kinase domain intact, and PhoR350, which has the N-terminal 349 aa deleted, leaving only the kinase domain with the conserved histidine residue. The predicted N-terminal aa sequence of PhoR173 is Gly-Ser-Pro-Gly-Lys-Glu-Ala-Tyr-Val-Met-173 and of PhoR350 is Gly-Ser-Pro-Gly-Lys-Glu-Ala-Tyr-Gln-Met-350 (Fig. 3).

Figure 2.

. Purification of overexpressed GST–*PhoR from Escherichia coli. Fractions from each purification step were analysed by SDS–PAGE and the gel was stained with Coomassie blue. The amount of protein loaded in each lane is give below in parentheses. Lanes: 1, soluble fraction of sonication lysate (35 μg) of induced cells harbouring vector pGEX-2T only; 2, soluble fraction of sonication lysate (35 μg) of induced cells harbouring plasmid pLS21 (GST–*PhoR); 3, flowthrough fraction from the mixture of glutathione–agarose beads and samples in lane 2 (30 μg); 4, eluate of GST by 10 mM glutathione (5 μg); 5, flowthrough collected after thrombin cleavage on glutathione agarose beads (3 μg).

Figure 3.

. Deletion mutants of PhoR proteins. A. The predicted topology of PhoR and PhoR proteins with deletions. The predicted topology of the wild-type PhoR protein is shown at the top. Numbers indicate the approximate boundaries of each domain. Amino acid numbers in parentheses indicates the number of extra amino acids predicted in proteins produced in E. coli but absent in the B. subtilis expression system. Solid boxes represents the transmembrane sequence. Hatched boxes represents the conserved kinase domain. The bold face indicates the extra amino acid (Ile) in the periplasmic domain-deleted PhoR 37-137 protein in vivo. TM1, the first transmembrane sequence; P, periplasmic domain; TM2, the second transmembrane sequence; C2, the cytoplasmic linking domain; HK, the conserved histidine kinase domain; ▾, the conserved histidine residue; O, outside of membrane; I, inside of membrane. B. The chromosomal structure of the strains containing inducible phoR genes. The solid boxes indicate chromosomal position of the phoP/R operon. The open boxes indicate the position of antibiotic-resistant genes. The hatched box indicates the position of mutant phoR genes. Arrows indicate the direction of transcription.

Autophosphorylation of PhoR proteins

To determine whether the truncated form of PhoR proteins function as autokinases, the purified truncated PhoR proteins, PhoR173, *PhoR and PhoR350 were also mixed with [γ-32P]-ATP. All purified proteins showed autophosphorylation (data not shown). *PhoR was used for further in vitro characterization.

Phosphotransfer from PhoR∼P to PhoP

When PhoP, the cognate response regulator for PhoR, was mixed with *PhoR∼P, PhoP was phosphorylated (Fig. 4A). The rate of this reaction was rapid as the level of PhoP∼P reached equilibrium within 10 s (Fig. 4A).

Figure 4.

. Phosphorylation of PhoP of B. subtilis by truncated PhoR proteins in the presence of ATP. GST–*PhoR bound to glutathione–agarose was phosphorylated by incubating with [γ-32P]-ATP. After excess [γ-32P]-ATP was removed, 5 U of thrombin was added to the beads and was mixed at room temperature for 10 min The thrombin-liberated PhoR∼phosphate was separated from the beads. PhoR-phosphate was mixed with an equal molar ratio of PhoP or PhoB at t = 0. Samples were removed at intervals of time, and the reaction was stopped with SDS loading buffer. Samples were then run on SDS–PAGE. The gels were dried and the radioactivity was quantified by a PhosphorImager. Phosphorylated protein was measured in arbitrary units. A. *PhoR∼P incubated with PhoP. B. PhoR173∼P incubated with PhoP. C. Pho350∼P incubated with PhoP. ▪, PhoR-phosphate; ♦, PhoP-phosphate.

When PhoP was added to autophosphorylated PhoR173 (Fig. 4B) or the PhoR350 protein (Fig. 4C), PhoP could be phosphorylated as well (Fig. 4B and C respectively). However, in the experiment with PhoR173 or PhoR350, the phosphotransfer reaction between phospho-PhoR and PhoP was much slower, with the maximal level of PhoP-phosphate obtained in about 45 s (PhoR173) (Fig. 4B) and 2 min (PhoR350) (Fig. 4C) instead of 10 s. Thus, the efficiency of phosphotransfer from either PhoR173-phosphate or PhoR350-phosphate to PhoP in vitro was also much lower than that for *PhoR (Fig. 4B and C). The results also showed a loss of total radioactive label on PhoP∼P and PhoR∼P over time. This is probably the result of a PhoR phosphatase activity for PhoP∼P protein, because all PhoR∼P and PhoP∼P proteins are quite stable, with a half-life longer than 2 h (data not shown).

Kinetics of Pho regulon induction are preserved in strains carrying phoR truncation mutations

To study the domain function of PhoR protein in vivo, an inducible PhoR production system was constructed. The various 5′-deleted phoR genes (Fig. 3) were constructed and cloned into plasmid pDH88, which contains an IPTG-inducible promoter, Pspac, and an artificial ribosome binding site in front of the ATG initiation codon. The encoded N-terminal truncated/deleted proteins are shown in 3Fig. 3A. The plasmids containing the mutated phoR genes were then transformed into B. subtilis strain MH5124 (Hulett et al., 1994b) and were integrated in single copy in the chromosome via a Campbell-like insertion duplication. The parent strain, MH5124, has a deletion-insertion (tet ) at the phoR locus and does not induce the Pho response, including APase under phosphate starvation conditions. The chromosomal structure in these strains is illustrated in 3Fig. 3B. APase was used as a reporter of Pho induction in the parent strain (JH642), the phoR deletion strain (MH5124), and the five strains containing phoR or the phoR genes containing deletion mutations under the IPTG-inducible Pspac promoter integrated at the phoR locus. Transformants showing IPTG-dependent APase production under phosphate-limiting conditions were chosen, and 12 h growth curves were performed with or without IPTG at 1 mM. Under low phosphate conditions without IPTG, neither the parent phoR mutant strain (MH5124) nor any of the strains containing Pspac-controlled phoR genes produced APases (data not shown). When IPTG was added to the culture at time 0, the strain MH5701 containing Pspac-controlled wild-type phoR produced APase at post-exponential stage when phosphate became limited between hours 5 and 6 (Fig. 5B) just as the JH642 (wild-type) strains did, and the total induced APase activity reached a similar level (Fig. 5A and B). In strain MH5702 (PhoRΔ37–137), which produces the PhoRΔ37–137 protein without the periplasmic domain, the induction pattern was not altered and the specific APase activity reached about 50% compared with the APase activity in wild type (Fig. 5C). For strain MH5703 (PhoR173), the level of induced APase activity in this strain in the presence of IPTG was decreased to ≈40% of that of the wild type (Fig. 5D). Strain MH5704 (*PhoR) in low phosphate defined medium (LPDM) with IPTG produced APase post-exponentially, with its level reaching ≈45% of the wild type (Fig. 5E). The strain MH5710 (PhoR350), in the presence of IPTG, produced APase at 10–15% of the wild-type level (Fig. 5F). It should be noted that when any of these strains were grown in LPDM containing 0.4 mM phosphate (Fig. 5A–F), the cells did not induce APase synthesis in logarithmic growth, when phosphate was still sufficient to repress the Pho response in the wild-type strain (hour 0 to hour 5) (Fig. 5A–F), even though the PhoR protein was constitutively produced by IPTG in the culture medium (Fig. 6). Further, none of these strains induced APase during 12 h of growth in Pi-replete culture medium that contained 5 mM phosphate (HPDM) when IPTG was present (data not shown).

Figure 5.

. Growth and APase production of B. subtilis strains cultured 12 h in LPDM with or without IPTG. +×, Cell growth of MH5124 (phoR parent); ○, APase specific activity of MH5124; *, cell growth of JH642 (wild-type control); •, APase specific activity of JH642. B–F. Strains containing phoR mutants genes in single or multiple copy. □, Cell growth of strains containing phoR gene in single copy; ▪, cell growth of strains containing phoR gene in multicopy; ◊, APase specific activity of strains containing phoR gene in single copy; ♦, APase specific activity of strains containing phoR gene in multicopy. Arrow indicates the time when IPTG was added. PhoR allele in strain is given at the top of each graph. The predicted protein topology is shown on the upper left corner of the graph for each pair of experimental strains.

Figure 6.

. Immunoblotting of B. subtilis cells producing truncated PhoR proteins. All samples of B. subtilis cells producing truncated PhoR proteins were taken during logarithmic growth (hour 0), post-exponential growth (hour 7) or stationary growth (hour 11) and lysed by incubating with lysozyme. Total cellular protein (150 μg) of each sample was separated by SDS–PAGE and transferred to Immobilon-P (Millipore); the membrane was incubated by polyclonal anti-*PhoR antibody.

Inducible expression of phoR genes from a multicopy plasmid in B. subtilis increases the magnitude of the Pho response

The decreased level of total induced APase activity in the phoR truncation strains could possibly be due to a reduced level of active PhoR protein in the cell. To determine whether the in vivo PhoR protein concentration contributes to the reduced APase levels, multicopy plasmids containing Pspac-controlled phoR mutants were transformed into phoR strain MH5124. Each transformed cell should carry 5–10 copies of the plasmid (Haima et al., 1990). The strains containing the phoR genes (wild-type or mutated) on a multicopy plasmid were cultured for 12 h in LPDM in the presence of IPTG. The results of these growth curves and APase induction patterns showed that (Fig. 5B–F, ♦) strains harbouring multiple copies of phoR mutant genes in the cell have increased APase production by ≈50% to ≈150% compared with strains containing the same phoR mutation in single copy. In the case of phoR173, the level of APase was restored to wild-type level (Fig. 5D). The pattern of APase induction (Pho induction) was independent of the copy number of the phoR allele. When wild-type phoR was constructed on a multicopy plasmid, in contrast to the increase of APase levels in mutant phoR strains, the level of total induced APase decreased by about 30%. This is probably the result of the overexpression of wild-type phoR gene, which would shift the ratio of PhoP to PhoR protein in vivo, a phenomenon that has been observed during overexpression of phoP in B. subtilis (Liu, 1997).

Western-blotting assay for in vivo level of PhoR proteins shows increased PhoR from multicopy vector

To determine whether the increased APase production in strains carrying truncated phoR genes is indeed correlated with the level of IPTG-induced PhoR protein in vivo, Western-blotting assays were performed. Cells from the initial exponential phase (hour 0), the post-exponential phase (hour 7) and stationary phase (hour 11) of the growth in LPDM were collected. Strains MH5703 (phoR173), MH5704 (*phoR) and MH5710 (phoR350), which contain only one copy of the respective mutated phoR gene, contained very low levels of PhoR protein, which in the case of MH5710 was almost non-detectable. Each strain carrying the mutated phoR gene on a multicopy plasmid, MH5713 (phoR173), MH5714 (*phoR) and MH5720 (phoR350), maintained a higher PhoR protein concentration than the PhoR concentration with a single copy. Thus, PhoR protein concentration is correlated with the level of APase production in these strains (Fig. 6).


Little is known about whether and how the PhoR protein of B. subtilis senses the phosphate starvation signals in order to regulate alkaline phosphatase, cell wall anionic polymer biosynthesis and proteins involved in phosphate transport. A comparison of the domain structure of B. subtilis with other sensor kinases provides two domains previously implicated in sensing, the periplasmic domain and the extended cytoplasmic domain called C2. It has been proposed that the histidine sensor kinase proteins can receive the environmental signal through their periplasmic domain, triggering a conformational change at the kinase domain, causing the switch of phosphorylation state at the conserved histidine residue (Hess et al., 1988; Kofoid and Parkinson, 1988; Stock et al., 1989). Point mutations in the periplasmic domain of NarX and NarQ, which locked the kinase in an on or off state (Cavicchioli et al., 1990), support such a model. N-terminal deletion in E. coli phoR leading to constitutive expression has also been interpreted to support this model (Makino et al., 1989; Yamada et al., 1990).

The linker region (C2) between the N-terminal and the C-terminal kinase domains of B. subtilis is also of interest, because studies of E. coli phoR by Scholton and Tommassen (1993) suggested a model and presented data consistent with the idea that the C2 domain may serve a sensing function. This model was attractive because it had been suggested that the periplasmic region of E. coli PhoR protein may be too small to serve as a signal-sensing area. The existence in B. subtilis PhoR of a large C2 region with sequence similarity to E. coli PhoR suggested that C2 could also be a potential domain for sensing an intracellular signal.

This study was designed to assess the role of the two N-terminal domains, ‘periplasmic’ or C2, in sensing the low phosphate signal in B. subtilis PhoR by using deletion/truncation mutagenesis. In vitro data presented here showed that using signal transduction components of the B. subtilis Pho regulators produced in E. coli, PhoR showed autophosphorylation and phosphorylation of PhoP. The N-terminal truncated *PhoR-phosphate phosphorylate PhoP rapidly, whereas the other mutated proteins, PhoR173 and PhoR350, phosphorylate PhoP protein more slowly and less efficiently. The phosphorylated *PhoR protein, which shows a half-life of longer than 2 h (data not shown), displayed base stability and acid lability (data not shown), suggesting that the PhoR protein is phosphorylated at a histidine residue. Consistent with the histidine phosphorylation, a H360Q mutant protein, *PhoRH360Q, has no autokinase activity (data not shown).

We used in vivo expression of mutated phoR genes to study the possible sensing function of the two N-terminal domains. A sensor region in PhoR is involved in regulation, and its removal would lead to either defective or constitutive expression of Pho response. However, deleting 102 of the 132 aa of the periplasmic domain, or the entire N-terminal sequence up to 10 aa upstream of the conserved histidine, did not alter the APase induction pattern. Although each of the mutant phoR genes encoding the various mutated proteins was constitutively expressed from the Pspac promoter from time 0, the Pho reporters, APases, were not produced in any of these strains until approximately hour 6, an induction time identical to that of a strain containing the wild-type phoR gene controlled by its own promoter or the Pspac promoter. All of these strains, containing wild-type phoR or mutant phoR, failed to initiate the Pho response until the Pi concentration became limited. These data show that the time of PhoR transcription/translation does not dictate the time of Pho induction, and that all strains, independent of the phoR allele, retain Pi-dependent Pho regulatory control.

The magnitude of the Pho response assessed by APase specific activity was less in single-copy phoR mutant strains than in the strain containing the wild-type phoR gene. At least three factors can contribute to the reduced Pho response. The first is the efficiency of PhoR autophosphorylation, the second is the efficiency of phosphotransfer from PhoR to PhoP, and the third is the turnover rate of PhoR protein affecting PhoR cellular concentrations. The in vivo result is presumably dependent on the combination of at least these three factors. Therefore, a comparison of any one factor, such as autophosphorylation, or phosphotransfer in vitro among these PhoR proteins, may not reflect their comparative catalytic ability in vivo. However, increasing the cellular concentration of any one of these proteins in vivo may help us to understand the correlation between active PhoR concentration in vivo and the level of Pho response. Our results showed that (in Fig. 5) multicopy expression strain MH5713 (phoR173) dramatically improved the total APase production to wild-type level; multicopy expression strains MH5712 (phoRΔ37–137 ), MH5714 (*phoR) and MH5720 (PhoR350) also improved their total APase production by 50–100%. Consistently, the multicopy expression strains produced more PhoR proteins as predicted than the single-copy strains, among which the PhoR173 and PhoR350 proteins were almost non-detectable (Fig. 6). Although PhoR173 and PhoR350 proteins appeared to be at similar concentration during multicopy expression, recovery of Pho expression was markedly different, albeit their in vitro activities appeared quite similar (Fig. 4). This may, at least in part, be due to the condition of purified protein from E. coli used in in vitro experiments, because PhoR350 was purified as a soluble protein and PhoR173 was largely insoluble. The contribution of the C2 region in PhoR173 to the amplitude of the Pho response may suggest it is required for the full Pho expression by sensing a cytosolic signal, or that it is necessary for the maintenance of the conformation of the autophosphorylation site, which is only 10 aa from the N-terminal of PhoR350. What is clear from these data is that increasing the cellular concentration of the PhoR protein through multicopy expression increases the magnitude of the Pho response in the mutant strains, but does not influence the timing of initiation of induction.

The mechanism for the repression of pho regulon gene expression remains unclear. Although the best candidate for the sensor region in this case may be the C2 domain, we cannot rule out, based on our results, the possibility that the real sensor for the low phosphate signal triggering Pho response resides in the PhoR kinase domain, thus it is not deleted in any of the PhoR mutants. This is possible because the low phosphate signal could be cytosolic phosphate or a small molecule derived from phosphate that could directly interact with the sensor region within in the kinase domain of PhoR to modulate either its autokinase or autophosphatase activity without the intervention of any additional regulator. However, we consider this unlikely as the kinase domain is highly conserved among histidine kinases. The C-terminal of the sensor kinase, without the N-terminal domain, is often considered to be constitutively active as a kinase (Ninfa et al., 1988; Makino et al., 1989; Yamada et al., 1990; Magasanik, 1996). Assuming the sensor region on PhoR has been removed in the strain producing PhoR350, yet the high phosphate repression of Pho response is unaffected, one might propose that an additional regulator(s) is involved. This regulator may affect the autophosphorylation of PhoR and/or phosphotransfer between PhoR and PhoP in a number of ways. Such regulator(s) might affect the stability of the phosphorylation state of PhoP or PhoR, as is expected for a phosphatase. Precedence for phosphatases of the phosphorylated response regulator is well established in B. subtilis (Ohlsen et al., 1994; Perego et al., 1994; 1996; Perego and Hoch, 1996). Spo0E is a phosphatase specific to Spo0A, whereas RapA and RapB are two aspartate phosphatases specific to Spo0F, two key proteins leading to the initiation of sporulation. An E. coli protein, SixA, exhibiting phosphohistidine phosphatase activity towards the HPt domain of the ArcB sensor was identified (Ogino et al., 1998), and it has a negative regulatory function on the EnvZ-independent pathway for the osmolarity response in certain mutant backgrounds (Ogino et al., 1998). Alternatively, other unknown proteins may regulate the autophosphorylation of PhoR or may interact with either PhoP or PhoR to affect the phosphotransfer. In the case of phosphotransfer, such protein may bind to PhoR to block or promote the phosphotransfer to PhoP, or bind to PhoP to block or promote phosphorylation by PhoR. The PII protein in the nitrogen regulation system of E. coli acts to stimulate an phosphatase activity of NRII, the histidine kinase (Jiang et al., 1997) in vivo. Indeed, the purified *PhoR protein showed low levels of phosphatase activity in vitro (L. Shi and F. M. Hulett, unpublished), but an additional factor could stimulate a higher phosphatase activity or reverse the course of phosphotransfer back to PhoR and ATP. In the case of autophosphorylation, an inhibitor of the autokinase would prevent the kinase from being phosphorylated. Such an antikinase factor has recently been found by Wang et al. (1997). KipI prevents the autophosphorylation of KinA sensor kinase required for sporulation initiation in B. subtilis, but has no effects on the phosphotranferase reaction of KinA to Spo0F.

Based on the results shown in this study, an additional level(s) of regulation, mediating the cellular concentration of PhoP∼phosphate, is likely to be involved in Pho regulon activation. The function of the large N-terminal membrane domain of PhoR is unclear. Our data do not exclude the possibility that it may be involved in sensing a signal or in interacting with a sensing protein.

Our study clearly shows that the kinase domain is sufficient for the low phosphate-inducible expression of Pho regulon genes. The repression of APase production in high phosphate remains intact even in the absence of the N-terminal 349 amino acids containing the membrane domain and the extended C2 domain. Activation of PhoP/PhoR leading to Pho induction probably involves additional factors that would prevent the Pho regulon genes from being turned on/off before it is absolutely necessary as phosphate is an essential nutrient for cell growth and physiological functions. Future studies are focused on identifying a possible regulator(s), positive or negative, for PhoP activated by PhoR and determining the possible physiological function of the N-terminal membrane domain and the extended C2 domain.

Experimental procedures

Strains and construction of plasmids expressing phoR genes

E. coli DH5α was used as the host for plasmid construction (see Table 1). E. coli BL21(DE3) (Novagen) served as the host for overexpressing the PhoR proteins. A series of derivatives of the pDH88 or pHB201 vector were constructed to facilitate expression of various deletion or truncation mutants of phoR gene in B. subtilis. The limits of the coding region of each mutant, as defined by the sequences of the oligonucleotides used in PCR amplification, were chosen on the basis of computer alignments of the PhoR sequence with those of other sensor kinases of the two-component system family by using the program package PROSIS (Hitachi). The oligonucleotides were purchased from Operon Technologies. To construct pLS6, pES5 (Jensen et al., 1993) was digested with SacI and BamHI enzymes and the 1.8 kb fragment containing the complete phoR gene was recovered and ligated to pBluescript SK digested with the same pair of restriction enzymes.

Table 1. . Bacterial strains and plasmids (Hulett et al., 1994b; Jensen et al., 1993; Henner, 1990). a. Amp, ampicillin; Erm, erythromycin; Cm, chlromphenicol; Δ, deletion.Thumbnail image of

To construct pLS16-Tm1, primers FMH127 (TTAAGCTT58GGTATAAACTGGAGGAAGC) containing a HindIII site, FMH212 (205CCTCTGGGA196TCCTTTGATC187) (in which the bases in bold italic generate the restriction site mentioned below) containing a BamHI site were used to PCR out the phoR sequence encoding the first transmembrane sequence and clones into pCRII vector. The number in the sequence refers to the number in the operon sequence of GenBank database accession number M23549. The HindIII/BamHI fragment was cloned first into pBluescript SK digested with the same enzymes to generate pLS18.

To construct pLS17-Tm2, primers FMH213 (482CAGGATATGGGATC496CTTTCCGCC), containing a mutagenized BamHI site, and FMH214 (GAAGATCT1881CATCCTACCAGCAT), containing a BglII site, were used to PCR out the phoR sequence encoding the second transmembrane sequence and the cytoplasmic domain and cloned into pCRII vector. Clones capable of releasing a 1.3 kb BamHI fragment were selected.

To construct pLS19, the 1.3 kb fragment was released by digesting pLS17 DNA with BamHI and ligated to pLS18 linearized by BamHI. The clones capable of releasing a 1.5 kb HindIII/BglII fragment were selected. The 1.5 kb HindIII/BglII fragment was subsequently cloned into pDH88 (Henner, 1990) digested by HindIII and BglII to yield pLS20.

To construct pLS21, primers FMH105 (CAAAGCTT759GACGCGAACGCAGG) and FMH104 (GGAGATCT1881CATCCTACCAGCATAGG) were used to PCR out the part of the phoR gene from pES5. The PCR product was treated with the Klenow fragment of E. coli polymerase I, and ligated to pGEX-2T (Pharmacia Biotech) digested with EcoRI and treated with Klenow. The correctness of inserting *phoR gene was verified by sequencing.

To construct pLS73 and pLS77, pLS71 was first constructed by using primers FMH241A (CCCGGGCAAGGAGGCCTA494TGTGATGCTGTCC) and FMH104 to PCR out the part of phoR gene from pES5. The PCR product was digested with SmaI and BglII and ligated to pGEX-2T digested with the same enzymes. pLS73 and pLS77 were obtained by using primers FMH243 (AGGCC595TCCAGCATGACCTCA) for pLS73, or primers FMH247 (AGGCCTAT1129CAGATGAGGAAG) for pLS77, and FMH104 to PCR out the part of phoR gene from pES5. The PCR product was cloned in pCRII and the StuI–BglII fragment containing part of phoR gene was ligated to the 4.9 kb vector fragment of pLS71 digested by the same enzymes.

To construct pLS23-2, the BamHI–BglII fragment of pES5 (Hulett et al., 1994b) was ligated to pDH88 linearized with BglII. The mutant phoR gene was amplified from JH642 chromosomal DNA by PCR using primers at the 3′ end of the gene. To obtain pLS59; pLS23-2 was digested with SmaI and BglII. The 1.9 kb fragment containing the complete phoR gene was treated with Klenow and ligated to pLS68 linearized by SmaI.

To construct pLS68, pDH88 was digested with BamHI and PvuI, and the 2.3 kb fragment containing lacI and Pspac was ligated to pHB201 digested with the same enzymes to generate the 8.9 kb pLS68.

To obtain pLS58, pLS26 was digested with HindIII and BglII to release a 1.1 kb fragment containing *phoR with RBS in front. The fragment was then treated with Klenow and ligated to pLS68 linearized by SmaI.

pLS53 was obtained by ligating the SmaI–BglII fragment containing phoR173 and pDH88 digested with the same enzymes. To construct pLS63, pLS53 was digested with PvuI and BamHI; the 3.7 kb fragment containing Pspac, phoR173, and LacI was ligated with pLS68 digested with the same enzymes.

pLS57 was obtained by ligating the SmaI–BglII fragment containing phoR350 and pDH88 with the same enzymes. To construct pLS67, pLS57 was digested with PvuI and BamHI; the 3.1 kb fragment containing Pspac, phoR350 and lacI was ligated to pLS68 digested with the same enzymes.

Overexpression and purification of PhoR

Escherichia coli BL21(DE3)pLS21 was incubated overnight at 37°C in LB medium containing penicillin (150 μg ml−1) and was then inoculated into 2 l of M9 medium at a ratio of 1:40. The cells were grown at 30°C until the optical density at 600 nm (OD600) of the culture reached 0.6. Then 1 mM IPTG (isopropylthio-β-D-galactoside) was added to the culture, and growth was continued for another 3 h. The cells were harvested by centrifugation at 4°C and washed with P buffer (50 mM HEPES, 50 mM KCl, 5 mM MgCl2, pH 8.0). The cell pellet was stored at −70°C.

The pelleted cells were resuspended in P buffer and disrupted by sonication. The lysis of cells was checked using a microscope. After centrifugation at 33 000 × g in a Beckman 50Ti rotor, the insoluble fraction was resuspended in P buffer. A 0.3 μg sample of total insoluble protein was used in each reaction in the presence of 5 μCi of [γ-32P]-ATP (0.05 nM).

The thawed cells were suspended on ice in 40 ml of P buffer containing 1 mM phenylmethylsulphonyl fluoride (PMSF) and were immediately subjected to sonication. After a 5 min sonication on ice at an output of 100 W, the cell lysate was centrifuged at 100 000 × g for 1 h at 4°C. The supernatant was applied to a 2 ml glutathione–agarose (Sigma) affinity column equilibrated with P buffer. The column was washed with P buffer until the OD280 of the eluate was less than 0.03; the bound GST–*PhoR protein was then eluted with P buffer containing 10 mM glutathione and 20% (w/v) glycerine. To obtaining *PhoR without glutathione-S-transferase (GST), the protein-bound beads were added with P buffer containing 20% (w/v) glycerine and 20 U of thrombin (Sigma) and gently shaken at 4°C overnight; the flowthrough was collected by centrifugation. The protein was aliquoted and stored at −70°C.

Phosphorylation and transphosphorylation assays

Phosphorylation assays were carried out at room temperature in the presence of 5 μCi of [γ-32P]-ATP (specific activity 6000 Ci mmol−1; Amersham Biotech) in P buffer. The reactions were initiated by the addition of [γ-32P]-ATP to 0.05 nM in a 20 μl reaction mixture), terminated by the addition of an equal volume of 6× sodium dodecyl sulphate (SDS) sample buffer, and subjected to SDS–PAGE) on 10% or 12.5% polyacrylamide gels (Laemmli, 1970). The radioactivities of proteins resolved on gels were determined qualitatively by autoradiography of dried gels with X-Omat AR (Kodak). A PhosphorImager (Molecular Dynamics) was used for quantitative analyses.

Stability of *PhoR∼P

Purified GST–*PhoR was used to bind glutathione–agarose, and 5 μCi [γ-32P]-ATP for 10 min, and 5 U of thrombin was then added to the reaction mixture and stood at room temperature for 10 min; the flowthrough containing *PhoR∼P was collected. Aliquots were mixed with 0.5 M NaOH, 0.5 HCl or P buffer for 20 min at room temperature. The reaction was stopped by adding 6× SDS sample buffer. After electrophoresis, the gel was dried and was exposed to a PhosphorImager screen for quantitative analysis.

Detection of protein concentration

Protein concentration was detected by the Bradford method (Bradford, 1976) using the Bio-Rad protein assay kit as instructed by the manufacturer.

Growth conditions and enzyme assays

APase activity was measured in the cells that had been grown in the low phosphate defined medium described previously (Hulett et al., 1990) with the following modification: for strains containing Pspac promoter-controlled construct, IPTG was added at 1 mM final concentration throughout the growth. To determine when APase induction occurred, hourly determination of culture density and APase activity was made on cells grown under culture conditions described previously (Chesnut et al., 1991).

Immunoblotting assays

B. subtilis cells growing at exponential phase (hour 0), post-exponential phase (hour 7) and stationary phase (hour 11) were collected by centrifugation. The cells were resuspended in water containing 1 mg ml−1 lysozyme and stood at room temperature for 10 min. Total cellular protein content was calculated using the formula: total protein = OD540* 83 μg per ml culture. After the samples were denatured by adding SDS sample buffer and boiling, 150 μg of total cellular protein was separated by SDS–PAGE. Polyclonial antibody against *PhoR was used as the primary antibody, and the induced PhoR protein was visualized by the second APase-conjugated goat anti-rabbit antibody (Bio-Rad).

Antibody production

The purified *PhoR was injected into rabbits to generate PhoR antibody. The antibody was made by Cocalico Biological, (Reamstown, PA, USA).


We thank Dr W. Liu for purified PhoP protein in this study. This work was supported by a grant (GM33471) from the National Institutes of Health to F.M.H.