INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Escherichia coli contains at least two major systems for transporting inorganic phosphate (P_i). The low-affinity inorganic phosphate transporter (Pit) is dependent on the proton motive force for energy and is constitutively expressed (30, 31, 49). When P_i is plentiful, this is the major uptake system for phosphate, with a reported apparent K_m (K_m^app) of 25 µM (30) to 38 µM (50) in whole cells and 11.9 µM in membrane vesicles (43). If the external P_i concentration is below the millimolar range, the high-affinity phosphate-specific transport (Pst) system is induced. This has a K_m^app of around 0.2 µM (30, 50). The Pst system is a complex of four proteins, including a periplasmic binding protein, which is energized by ATP and belongs to the ABC transporter family (7, 15, 47). The pst operon contains five genes under pho regulon control (1, 40, 41), which induces a range of genes when the phosphate supply is limited. Both the Pit and Pst systems are highly specific for P_i (30). Another two transporters accept P_i as a low-affinity analogue for either glycerol-3-phosphate (glpT) (18) or glucose-6-phosphate (uhpT) (29, 53), but in the absence of Pit and Pst activity, these latter two systems cannot support cell growth when supplied with P_i (38).

Divalent cations, such as Mg²⁺ or Ca²⁺, were shown elsewhere to be essential for Pit activity (32), and experiments by van Veen et al. (43, 44) indicate that Pit forms a soluble neutral metal phosphate (MeHPO₄) complex which is symported with a proton. This is supported by the recent identification of a pitA mutant that accumulates reduced amounts of zinc(II), conferring resistance to toxic external concentrations of zinc (4). Efflux and homologous exchange of metal phosphate can occur under particular conditions, but there is no mixed exchange of metal phosphate for P_i, glycerol-3-phosphate, or glucose-6-phosphate (43). Interestingly, Beard et al. (4) suggest that PitA may also play a role in Zn²⁺ efflux when the ion reaches highly toxic external concentrations.

The Pit transport system was first reported by Willsky et al. (49) when mutations in the Pst system of several E. coli K-12 strains revealed a second P_i transporter. When a pst mutation was introduced into strain K-10, there was no measurable P_i transport, indicating that Pit is nonfunctional. This lesion was isolated as an -glycerol-3-phosphate (G3P) auxotroph and mapped at min 78.51 on the E. coli genome (37, 38) and is called pitA (Swiss Protein P37308). Since these studies were done, the sequencing of the E. coli genome has revealed the presence of another putative phosphate transporter gene, designated pitB, at min 67.44 (5) (Swiss Protein P43676). This gene has 75% sequence identity to pitA. If pitB is also a functional phosphate transporter, it may have contributed to the kinetic values and substrate specificities determined in earlier studies. These studies used strains in which the Pst system was repressed and/or mutated (30, 44, 50). Alternatively, the pitB gene may not encode a P_i transporter, as cysP is a sulfate permease from Bacillus subtilis which has some sequence identity with the Pit family of transporters but not with sulfate transporters (24).

Elvin et al. (10) located and cloned the pitA gene, overexpressed it, and partially purified the protein. This paper describes the cloning of pitB and the demonstration that pitB is a P_i transporter which appears to be repressed or inactivated by the pho regulon. In addition, the E. coli K-10 pitA mutation is identified, and the kinetic parameters of the PitA and PitB proteins are investigated.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. Strains and plasmids are listed in Table 1.

Cosmid cloning and sequencing of pitB. Chromosomal DNA was prepared from a recA derivative of K-12 strain AN2537 (pstC345) (39) and partially digested with HindIII to generate fragments of approximately 20 kb. These fragments were ligated into the cosmid cloning vector pHC79 and packaged into  heads with extracts prepared from the lysogen-induced strains BHB2690 (prehead donor) and BHB2688 (packaging protein donor) (33). Packaged cosmids were adsorbed to AN3020 (pitA pstC345) (47), a strain which shows no growth on minimal medium supplemented with 500 µM inorganic phosphate (P_i medium) (10). The mixture was spread onto Luria-Bertani plates containing 100 µg of ampicillin (AMP)/ml and 1 mM G3P. Either G3P or glucose-6-phosphate may be used as the phosphate source for pit pst strains, which cannot utilize P_i for growth (38). AMP-resistant colonies were replica plated onto P_i medium, and colonies were screened for growth in the absence of G3P, which indicated the presence of a functional P_i transport gene. Cosmid DNA was prepared from one of these isolates. Restriction analysis identified an approximately 3-kb ClaI/BamHI fragment that encoded a protein which complemented the pit mutation in AN3020, enabling growth of this strain on P_i medium. This DNA fragment was subcloned into the M13 vector mp18 and sequenced. The sequence is identical to the subsequently published pitB sequence (5). The pitA gene from plasmid pCE27 (10) was also sequenced and was found to be identical to the published sequence (37) (Swiss Protein P37308).

Preparation of cells for uptake. Cells were grown and prepared as described in the work of Rosenberg et al. (30), with the addition of 5% Luria-Bertani medium to the overnight growing medium. AMP (50 µg/ml) was added where appropriate. All cells were washed three times, and suspensions were stored at 4°C for less than 30 min.

Measurement of phosphate uptake. Measurement of ³³P_i (NEN) uptake by cells was carried out under conditions in which P_i uptake was linear over time. These conditions were 1 to 15 µM P_i (PitA) or 10 to 100 µM P_i (PitB) over a 25-s period as previously described (30), except that the phosphate-free buffered "uptake" medium was pH 6.6 or 7.0 and the magnesium concentration was 1.8 or 10 mM. Data from kinetic experiments were analyzed by nonlinear regression using the Michaelis-Menten equation and the Graphpad Prism program.

Genetic techniques. Plasmid DNA was prepared using the Magic Minipreps DNA purification system (Promega). Site-directed mutagenesis was carried out using the Amersham Sculptor in vitro mutagenesis system and PitA- or PitB-specific oligonucleotide primers. The nucleotide sequence at each mutation was checked by DNA sequencing. Transductions using phage PI_kc were performed as previously described (28).

Inactivation of genomic pitB. The chloramphenicol resistance gene from pBR328 was excised with the restriction endonucleases AatII and SauI, 5' overhangs were filled in, and 3' overhangs were digested by T4 DNA polymerase 1; then the gene was ligated into the StuI site in pitB in pAN656 to create pAN1193. Genomic replacement of wild-type pitB was carried out by transforming pAN1193 into the recB21 recC22 sbcB15 sbcC201 mutant strain JC7623 as previously described (26). The insertion of pitB::Cat was checked by PCR. P1 transduction was then used to transfer pitB::Cat into pstC (AN2537) and pitA pstC (AN3020) strains. Recipients were selected by resistance to 50 µg of chloramphenicol/ml. Successful removal of wild-type pitB was confirmed by PCR. The rapid spray alkaline phosphatase assay (6) showed that alkaline phosphatase was derepressed under high-phosphate conditions (Luria-Bertani medium plus 1 mM G3P), indicating the presence of pstC345.

Inactivation of the pho regulon. The phoB-phoR operon deletion from ANCH1 (54) was transduced into AN3020 and AN3902 and selected for by resistance to kanamycin. Inactivation of the pho regulon was confirmed by a negative result with the rapid spray alkaline phosphatase assay from cells grown on Luria-Bertani medium.

Isolation of membrane fraction. Two-liter cultures of E. coli strains were grown to stationary phase in Luria-Bertani medium plus 34 mM glucose, 100 µg of AMP/ml, and 1 mM G3P, as required. The membrane fraction was prepared by passing the cells through a Sorvall Ribi cell fractionator, followed by centrifugation and ammonium sulfate precipitation, as previously described (9). The protein concentration of each sample was determined.

Polyclonal antipeptide antibody. The PitA peptide ARIHLTPAEREKKDC (from A188 to D201) and the equivalent sequence in PitB, DRIHRIPEDRKKKKC (from D188 to K201), are from an extramembranous loop in the putative folded structure. Rabbits were immunized with the PitA peptide coupled to a lysine core matrix via a C-terminal cysteine (22). The multiple antigen peptide system conjugate was dissolved in phosphate-buffered saline (PBS) and emulsified with an equal volume of Freund's adjuvant (12), and 200 µg was injected subcutaneously. Standard sampling and injecting protocols were followed (16). Sera were isolated by centrifugation, and initial positive responses determined by enzyme-linked immunosorbent assay (11) were screened by Western blotting of the sera against the membrane fractions of AN3903 (pitA pitB), AN3904 (pitB), and AN3905 (pitA). This showed that there was no cross-reactivity between PitA antibody and the PitB protein. The PitB peptide was attached to Imject maleimide-activated keyhole limpet hemocyanin (Pierce) according to the manufacturer's instructions. The conjugate was partially dissolved in dimethyl sulfoxide with sonication and then diluted with 1 volume of PBS. Injections and screenings were carried out as previously described for the PitA peptide.

Purification of antipeptide antibody. Sera containing PitA antipeptide antibody were not purified. The PitB equivalent was isolated from sera by immunoaffinity purification (SulfoLink kit; Pierce) following a two-stage ammonium sulfate precipitation (16) and dialysis against PBS. The synthetic peptide column was prepared according to the manufacturer's instructions, and the dialyzed PitB antipeptide antibody solution was passed through the column three times.

Western blots. The membrane fraction of various E. coli strains was solubilized at 150 µg (PitA Western blots) or 1 mg (PitB Western blots) of protein per ml. Electrophoresis was performed according to the method of Laemmli (20) using a sodium dodecyl sulfate-polyacrylamide gel. The proteins were then transferred onto a polyvinylidene difluoride membrane by electroblotting. After blocking in 10% milk powder-10 mM Tris-HCl (pH 7.5)-0.9% NaCl (blocking buffer), the polyvinylidene difluoride membrane was incubated with either polyclonal antipeptide PitA antibody diluted 1/500 in blocking buffer or polyclonal PitB antibody diluted 1/100. Alkaline phosphatase conjugated with goat anti-rabbit immunoglobulin (DAKO) was applied at a 1/1,500 (PitA) or a 1/1,000 (PitB) dilution in blocking buffer. The blot was immunostained with Western Blue stabilized alkaline phosphatase substrate (Promega).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of pitA and pitB. The pitA and pitB genes from E. coli strain K-12 were isolated and sequenced in this laboratory as described in Materials and Methods, and the ORFs and putative promoter regions were identical to the published sequences of pitA at nucleotides 3635272 to 3636771 (37) and pitB at nucleotides 3132887 to 3134386 (5) on the E. coli K-12 genome. The ORFs of pitA and pitB contain 1,497 nucleotides each, with 75% identity in the nucleotide sequences and 81% identity in the deduced amino acid sequences. Topological models suggest 10 putative transmembrane helices, with most sequence variation occurring in the putative hydrophilic loop regions.

Confirmation of pitA and pitB translation start codons. To identify the start codons, attempts were made to overexpress pitA or pitB and purify the protein products. However these attempts were hindered by the toxicity of the overexpressed genes (unpublished data). Instead, the translation starts were identified by functional analysis of alleles carrying mutated start codons (34). Both pitA and pitB have an ATG codon at the beginning of a 1,497-nucleotide ORF, and the nucleotide sequence similarity between the two DNA fragments exists only over this ORF region, suggesting that these ATG codons may initiate translation of the Pit genes. These putative ATG start codons were changed to GTG and CTG, causing a progressive drop in the P_i uptake activity of these mutants (Table 3). Changing an ATG start codon to GTG or CTG would be expected to progressively reduce expression of these proteins. In addition, two termination codons inserted in frame only 13 nucleotides after the putative ATG start of pitB completely abolished all P_i uptake activity. pitA has a second in-frame ATG codon 232 nucleotides after the putative ATG start codon, but termination codons placed before (at 118 nucleotides) or after this methionine also abolished all uptake activity. The combination of the above results indicates that translation of the pit genes commences at the beginning of the identified ORFs.

Kinetic parameters of PitA and PitB. To characterize the kinetic properties of PitA and PitB with respect to P_i uptake, AN3066 cells expressing either protein from the plasmid vector pBR322 were grown, deprived of phosphate, and assayed as described in Materials and Methods. The background strain AN3066 (pitA1 pstC345) does not grow in minimal medium supplemented with 500 µM P_i (P_i medium) but will grow with the addition of G3P. Both pitA and pitB can support cell growth when expressed on plasmids in AN3066. The PitA protein has a K_m^app about 10-fold lower than that of PitB on pAN656 in whole cells (Table 4). While the V_max^app for PitA was relatively stable, PitB V_max^app values were variable, as indicated by the high error values. Magnesium concentration and pH were altered to allow comparison with kinetic parameters measured by other researchers. While the K_m^app values remained similar under these conditions, the V_max^app was lower for both PitA and PitB when the magnesium concentration was increased from 1.8 to 10 mM and the pH was increased from 6.6 to 7.0 (Table 4).

View larger version (46K): [in this window] [in a new window]   FIG. 1.   Expression of PitB protein in different strains. Shown is a Western blot of purified polyclonal antipeptide PitB antibody against the membrane fraction of the indicated pitA1 pstC345 strains (pitB genotype of the chromosome and plasmid). Lane 1, AN3903 (pitB, vector pBR322); lane 2, AN3514 (pitB+, vector pBR322); lane 3, AN3135 (pitB+, plasmid pitB+ with 1,403 upstream nucleotides); lane 4, AN3905 (pitB, plasmid pitB+ with 206 upstream nucleotides); lane 5, AN3904 (pitB, plasmid pitA+). Strains in lanes 2 and 3 have wild-type pitB on the genome, which produces no Pi uptake activity.

Identification of the E. coli K-10 Pit mutation in phosphate transport. While E. coli K-10 has been characterized as deficient in Pit activity (49), the mutation has not been defined. Copies of pitA and pitB nucleotide sequences from the K-10 strain AN3066 were made by PCR. DNA sequencing revealed that the pitB ORF and the 319 upstream nucleotides are wild type. The pitA sequence contains a point mutation of G to A at nucleotide 658 in the ORF of the published sequence, which creates an amino acid change of G220D in the coding sequence (pitA1). To check that this single base change could account for the loss of Pit function in K-10, site-directed mutagenesis was used to recreate this mutation in a K-12 pitA⁺ gene. Both sequences were cloned into pBR322 and then transformed into AN3066, which is deficient in P_i transport (pitA pstC). Neither the mutant genomic pitA from AN3066 nor the site-directed pitA(G220D) allowed growth on P_i medium. This indicates that the single point mutation found in the AN3066 pitA is sufficient to create a nonfunctional Pit gene. Western blotting on the membrane fraction of whole cells, using polyclonal antipeptide PitA antibody, showed the level of expression of PitA protein inserted in the cell membrane (Fig. 2). While cells containing a wild-type pitA plasmid showed a strong band representing PitA protein (lane 1), cells containing the AN3066 genomic pitA (lane 4) or the site-directed pitA(G220D) (lane 5) had low levels of protein similar to the pitA mutant background strain (lane 3). Therefore, the G220D mutation seems to affect insertion of the protein into the membrane.

View larger version (59K): [in this window] [in a new window]   FIG. 2.   PitA protein expression from the AN3066 pitA1 mutation. Shown is a Western blot of polyclonal antipeptide PitA antibody against the membrane fraction of the indicated strains. All plasmids are in the AN3066 (pitA1) background unless otherwise stated. Lane 1, pAN920 (pitA+); lane 2, AN248 (genomic pitA+); lane 3, pBR322 (vector control); lane 4, pAN1243 (pitA1 PCR from AN3066); lane 5, pAN1244 (pitA G220D).

Functional characterization of PitA and PitB. Plasmids containing either the pitA or pitB gene conferred P_i uptake activity on strain AN3066, which contains pitA1 pstC345 and a wild-type pitB gene sequence. These assays show that a functional P_i transporter can be produced from the pitB gene when it is located on a plasmid but not when it is present only on the AN3066 genome. However, P_i uptake activity from the AN3171 strain, previously attributed to pitA, could in fact result from an interaction between the plasmid's PitA protein and the PitB protein expressed from the genome. To explore this further, a pitA⁺ pitB pstC mutant was created by inserting a chloramphenicol resistance gene into the pitB coding region of a pstC strain. This strain (AN3926) grew on P_i medium, indicating that the PitA protein can transport P_i in the absence of PitB. Wild-type pitA or pitB plasmids could also restore growth on P_i medium when transformed into a pitA pitB pstC triple mutant strain (AN3902). Hence, both PitA and PitB can transport P_i independently of each other.

Effect of the pho regulon on pitB activity. The discovery that E. coli K-10, which lacks Pit activity (49), contains pitB which is wild type for the coding region and at least 319 upstream nucleotides poses an interesting problem. pitB was cloned due to the ability of the 3-kb fragment to complement for P_i uptake when in vector pHC79 (Materials and Methods), and P_i uptake assays confirmed that a functional P_i transporter was encoded on this fragment. Therefore, genomic pitB may be under regulation that is disrupted when the gene is placed on a plasmid.

1

1

1

1

1

1

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have shown that E. coli contains two pit genes encoding proteins able to transport inorganic phosphate. pitA has previously been characterized, but we have shown that pitB also encodes a functional P_i transporter. Previously, Pit activity was reported to be constitutive (30, 49). While pitA appears constitutive, pitB may be regulated by the amount of available P_i. Our data suggest that pitB repression/inhibition is mediated through the pho regulon, since deletion of the phoB-phoR operon activated P_i uptake by pitB. The pho regulon is normally induced under conditions of P_i limitation, and P_i-dependent activation requires an interaction between the sensor protein PhoR and the Pst transport system. PhoR phosphorylates the transcriptional regulator PhoB, which then binds to a pho box consensus sequence within the promoter of genes in the pho regulon (46). However, the pho regulon of the pstC345 strain used in our experiments is active at all P_i concentrations (8). The pitB⁺ strain AN3066 (pitA1 pstC345) was unable to grow on P_i medium until the phoB-phoR operon was deleted. The rate of P_i uptake obtained from PitB expressed from the genome was higher than the control rates and was similar to the lower rates produced from plasmid pAN656. P_i uptake by pitB on plasmids pAN656 and pAN1116 may be caused by the increased copy number of pitB titrating out pho regulon inhibition. Thus, our experiments show that genomic pitB encodes a functional P_i transporter repressed/inactivated by the pho regulon, either directly through PhoB or via another pathway controlled by the pho regulon. The most likely explanation is that PhoB represses the pitB gene because this has already been observed for a pitB-like pit gene in Rhizobium meliloti using a pit::lacZ fusion. The R. meliloti pit gene was repressed under conditions of P_i limitation, but repression was relieved in a phoB mutant (2). As in E. coli, mutating the Pst transport system equivalent (phoCDET) caused derepression of the pho regulon, repressing pit at all P_i concentrations. Transforming a phoC R. meliloti strain with plasmid-borne pit increases cell growth on P_i medium to normal levels, suggesting that the repression can be overcome by supplying multiple copies of the gene. Only a two- to threefold increase in pit expression was needed to suppress the phoC phenotype (3).

While most studies on the pho regulon in E. coli have focused on the activation of genes, there is mounting evidence that the pho regulon may also repress some genes involved in phosphate assimilation. Two-dimensional protein experiments have shown that the pho regulon induces about 118 proteins and represses around 19 proteins (all with pIs of less than 7) under conditions of P_i limitation (42). More specifically, Willsky and Mallamy (51) have shown that two proteins which are repressed under P_i-limiting conditions are not repressed in phoB or phoR strains under the same conditions. Smith and Payne (36) propose that these are the periplasmic peptide binding proteins OppA and DppA.

Putative pho box sequences have been identified within the promoter regions of the R. meliloti orfA-pit operon and the E. coli opp and dpp operons (3, 36). However, while genes activated by the pho regulon have a pho box located 10 bases upstream from the 10 region (reference 27 and references therein), those operons which may be inhibited by the pho regulon have pho boxes either upstream of or overlapping the proposed 35 region. It has been suggested previously that this atypical positioning may reflect the negative regulation by PhoB (3, 36). While the pitB promoter region contains several putative pho box sequences in atypical positions, pitA also has similar sequences in the equivalent locations. Any potential pho box involved in negative regulation should be unique to pitB, as pitA is not repressed by the pho regulon. The binding sites for PhoB repression have not been characterized for any gene, so other DNA motifs may be involved. Thus, binding studies will be needed to locate any PhoB interactions with the pitB gene.

PitB's change in K_m^app indicates that a different mode of enzyme activity is also occurring in the protein expressed from the shortened plasmid pAN1116. An inhibitor which interacts with the PitB protein may be diluted out by the increased PitB expression, altering the K_m^app. Alternatively, higher concentrations of PitB may allow the protein to form a complex of monomers with a more effective mechanism. Although most secondary transporters are thought to function as monomers, this is not always the case. The sodium proton antiporter from E. coli has been crystallized as a dimer (48) and exists in the cytoplasmic membrane as a homo-oligomer (14). Glutamate transporters from the human brain have been shown previously to form dimers and trimers (17). There are now several examples of transporters that have dual affinity for their substrate and/or two mechanisms (13, 19, 21, 35, 55). Further investigation is needed to determine if this applies to PitB.

It is also possible that PitA undergoes similar changes in substrate affinity when protein expression is increased. Western blotting indicates that pitA expression is greatly elevated by placing it on a plasmid (Fig. 2), and our K_m^app of around 2 µM for pAN686 is significantly lower than the 11.9 to 38 µM range obtained by researchers using genomic Pit (30, 43, 50). However, the previously reported K_m^app values for Pit have been measured under conditions which make it impossible to attribute activity to pitA and/or pitB.

The K-10 pitA lesion was identified as a replacement of glycine 220 with aspartic acid, disrupting membrane insertion of the PitA protein. While it is not surprising that replacement of glycine, a small uncharged amino acid, with an aspartic acid could disrupt membrane insertion of the protein, it was unexpected that this relatively stable lesion was caused by a single point mutation. However, recreation of this mutation by site-directed mutagenesis produced identical behavior.

Our results indicate that PitA is likely to be active under a greater variety of conditions than is PitB. This situation of multiple P_i transporters with overlapping activities has been reported previously for Saccharomyces cerevisiae (25) and Neurospora crassa (45).