Present addresses: Rachel J. Jackson, School of Health and Related Research, The University of Sheffield, Regent Court, Sheffield S1 4DA, UK. Lucy J. Lee, School of Medicine and Biomedical Sciences, The University of Sheffield, Beech Hill Road, Sheffield S10 2RX, UK. Renli Ma, School of Applied Sciences, Northumbria University, A311 Ellison Building, Newcastle upon Tyne, NE1 8ST, UK.
Editor: Klaus Hantke
Correspondence: Robert K. Poole, Department of Molecular Biology and Biotechnology, The University of Sheffield, Western Bank, Sheffield S10 2TN, UK. Tel.: +44 114 222 4447; fax: +44 114 272 8697; e-mail: firstname.lastname@example.org
Escherichia coli possesses two major systems for inorganic phosphate (Pi) uptake. The Pst system (pstSCAB) is inducible by low phosphate concentrations whereas the low-affinity transporter (pitA) has been described as constitutively expressed. PitA catalyses transport of metal [Mg(II), Ca(II)]–phosphate complexes, and mutations in pitA confer Zn(II) resistance. Here we report that pitA transcription is not constitutive; activity of a single-copy pitA–lacZ transcriptional fusion (monolysogen) was maximal at high extracellular Zn(II) (150 μM), in the absence of added Pi, and in a well-defined pitA mutant strain. Intracellular zinc levels were unaffected by adding Zn(II) to the medium for both the wild-type and mutant strains. However, in the wild-type strain, Mg levels (per gram of dry biomass) fell by eightfold in cells grown with added Zn(II) and by 20-fold when Zn(II) and Pi were added to cultures. Mutation of pitA reduced the effects of external Zn(II) and phosphate levels on Mg pools, consistent with competition or inhibition by Zn(II) of PitA. The mechanism of pitA regulation by extracellular Zn(II) and Pi is unknown but appears not to involve Fur or other well-characterized regulators.
The biological functions of inorganic phosphate (Pi) and zinc are well defined: both are essential elements for all life forms and their transport mechanisms may be linked. Zinc has distinctive structural and thermodynamic properties: it has no redox functions, but is a good Lewis acid and fulfils a number of control, regulatory and structural roles (Berg & Shi, 1996; Frausto da Silva & Williams, 2001). Although normally classed as a trace element, zinc accumulates in Escherichia coli cells to levels comparable to those of calcium and iron (Outten & O'Halloran, 2001) and zinc-binding proteins account for 5–6% of the total proteome (Andreini et al., 2006). The major inducible high-affinity Zn(II) uptake system is the ABC transporter, ZnuABC (Patzer & Hantke, 1998), but a further broad substrate, low affinity, constitutive zinc uptake system exists, ZupT (Grass et al., 2005), and Zn(II) also enters via the Pit system (Beard et al., 2000). Because Zn(II) is an essential element, its sequestration may be an antimicrobial mechanism: the principal E. coli-killing protein of human skin, psoriasin, shows diminished antimicrobial activity when the protein is Zn(II)-saturated (Glaser et al., 2005).
Phosphate occurs always in oxidation state V as phosphate, free or combined, and plays a decisive role in numerous metabolic processes (Frausto da Silva & Williams, 2001). Free phosphate (HPO4−), typically at millimolar concentrations within cells, must be imported from outside to sustain these levels for functioning of acid–base reactions. Escherichia coli possesses two major systems for Pi uptake. The pst operon is part of the pho regulon and is inducible by below millimolar concentrations of phosphate. The Pst system is a complex of four proteins energized by ATP and is a member of the ABC superfamily (Webb et al., 1992; Chan & Torriani, 1996) with a high affinity for Pi (apparent Km around 0.2 μM) (Rosenberg et al., 1977; Willsky & Malamy, 1980). In contrast, the low-affinity Pi transporter (Pit) is dependent on the proton motive force for energization and has been described as constitutively expressed (Willsky et al., 1973; Rosenberg et al., 1977, 1979). When Pi is plentiful, this is the major uptake pathway with an apparent Km of 25–38 μM, when measured in intact cells (Rosenberg et al., 1977; Willsky & Malamy, 1980). A third Pi transport gene, pitB, encodes a functional Pi transporter that may be repressed at low Pi levels by the pho regulon (Harris et al., 2001). Two further transporters (encoded by glpT and uhpT) accept Pi with low affinity (Hayashi et al., 1964; Pogell et al., 1966; Winkler, 1966) but, in the absence of Pst and Pit, these systems cannot support growth when phosphate is provided as Pi (Sprague et al., 1975).
Uptake of Pi via the Pit system is reliant on cotransport with divalent metal cations such as Mg(II) or Ca(II) through the formation of a soluble, neutral, metal–phosphate complex, which is the transported species (van Veen et al., 1994). There is evidence that the PitA transporter is capable of transporting alternative metals, such as Zn(II), because a pitA mutant accumulates reduced amounts of the metal and confers resistance to external zinc (Beard et al., 2000). Because Pit activity contributes to both phosphate and metal uptake, and because mutations of pitA confer resistance to zinc, we investigated whether the pitA gene was under transcriptional control. Here we report for the first time that pitA transcription is not constitutive, as previously supposed, but regulated in response to levels in the medium of zinc and phosphate and also a functional pitA gene. We also report major changes in cellular metal content, particularly Mg(II), under these conditions, which suggest that Zn(II) competes with, or inhibits, Mg(II) uptake.
Materials and methods
Bacterial strains and construction of a pitA operon fusion
Escherichia coli strain RKP2914 is a monolysogen harbouring the Φ(pitA–lacZ) transcriptional fusion. The operon fusion was constructed on a plasmid and then transferred to λ phage by recombination in vivo (Simons et al., 1987). The promoter was amplified by PCR using Accuzyme and oligonucleotide primers RP105 (5′-TTGGAATTCGGGATTCTGGCTCAGATAAGCGCCTG) and RP106 (5′-TGCGGATCCGTTGAATACCGCCGCCATAACCACGGCG) (EcoRI and BamHI sites underlined) from chromosomal DNA extracted from E. coli strain MG1655. The resulting purified fragment was ligated into the site created by digestion of pRS528 with the same enzymes and the required recombinant plasmid isolated by transformation of strain RK4353 (Δlac) (Poole et al., 1996). The fusion was combined onto λRS45 (Simons et al., 1987) and single-copy fusions to the chromosome of strain VJS676 (Δlac) were isolated and verified using β-galactosidase assays and Ter tests as before (Poole et al., 1996). Strain RKP2935, in the same background [Φ(pitA–lacZ)], carries additionally a pitA mutation, constructed by P1 transduction from the pitA::Tn10dCam mutant, obtained after random transposon mutagenesis and selection for resistance to zinc (Beard et al., 2000). The precise location of Tn10dCam has been determined by sequencing and is consistent with Hfr mapping. The downstream ORF (yhiO, now uspB) is transcribed towards pitA and therefore most unlikely to be subject to polar effects (Beard et al., 2000).
Media and culture conditions
Cells were grown in sidearm flasks at 37 °C, in a shaker at 150 r.p.m., in glycerol–glycerophosphate medium (GGM) in which all sources of Pi were eliminated (Beard et al., 1997). Buffering was achieved by MES, a zwitterionic buffer with low metal-chelating constants (Good et al., 1966), and phosphate was provided by β-glycerophosphate. The medium contained (all final concentrations): 40 mM glycerol, 40 mM 2-(N-morpholino)ethanesulfonic acid (MES), 18.7 mM NH4Cl, 13.4 mM KCl, 4.99 mM K2SO4, 68 μM CaCl2, and trace elements [134 μM EDTA disodium salt, 18.5 μM FeCl3·6H2O, 6.1 μM ZnO, 0.59 μM CuCl2·2H2O, 0.34 μM Co(NO3)2·6H2O, 1.6 μM H3BO3] in distilled MilliQ water, essentially as described before (Poole et al., 1979). After adjusting the pH to 7.4 and autoclaving, 1.0 mM MgCl2 and 7.64 mM β-glycerophosphate (final concentrations) were added aseptically. Chloramphenicol (30 μg mL−1) was added to plates for selection of pitA::Tn10dCam. For β-galactosidase assays, cultures were harvested by centrifugation at 4400 g for 10 min when they had reached exponential phase (50 Klett units; red filter, Manostat Corporation). Spectinomycin was added (final concentration 300 μg mL−1) 10 min before harvest to inhibit protein synthesis, because the pitA mutant is chloramphenicol-resistant.
Assays were performed at room temperature as described previously (Poole et al., 1996). Activities are expressed in terms of the OD600 nm of the cell suspensions used for permeabilization using the formula of Miller (1972). Each culture was assayed in triplicate and using a range of sample volumes to ensure linearity of the assay.
Trace element analysis using emission spectroscopy
Aliquots (80 mL) of GGM, supplemented where specified with up to 150 μM Zn(II) and 1 mM Pi, were inoculated with starter cultures (1.5% by volume) of strains RKP2914 and RKP2935 (also grown in GGM but without supplements). Cells were collected by centrifugation and resuspended in 1 mL MilliQ water, then transferred to Eppendorf tubes. Samples were vortexed for 30 s before centrifugation using a Sigma microcentrifuge for 2 min at 15 000 g. The wash with water was repeated, and then the pellet was washed twice with 1 mL 0.5% HNO3 (Aristar nitric acid, 69% v/v) to preferentially remove metal bound to the cell surface as established before (Beard et al., 1997). Supernatants collected from the washes were retained for analysis. Pellets were digested by resuspension in 750 μL conc HNO3 before transfer to nitric acid-washed test tubes previously sterilized by heating to 200 °C. Samples were heated for 45 min at 60 to 90 °C, taking care to prevent excessive bubbling to minimize expulsion of phosphorus and sulphur. Samples were analysed by inductively coupled plasma emission spectroscopy (ICP-ES) using a Spectro CIROSCCD (Spectroanalytical UK Ltd) with background correction. Calibration curves were established for each test element using standard solutions of 0.1, 0.2, 1, 5 and 10 mg L−1. Wavelengths tested for each element were as follows: Mn (293.930 nm), Fe (259.940 nm), Cu (224.700 nm), Zn (213.856 nm), Mg (285.213 nm), Ca (183.856 nm), Co (228.615 nm), P (178.287 nm) and S (182.034 nm), using MilliQ water as a blank solution and to dilute cell digests before ICP analysis.
To allow calculation of element concentrations on a dry mass basis, the weights of cells were determined by filtering 10–30 mL culture, grown to OD600 nm 0.55–0.65 (Jenway 6100 spectrophotometer, 1 cm path length) in GGM, through preweighed cellulose nitrate filters (47 mm diameter; pore size, 0.45 μm) that had been dried at 105 °C for 25 h to constant weight. Filters were then reweighed to calculate dry cell mass per millilitre of culture.
Φ(pitA–lacZ) activities in the presence of zinc and phosphate
The widely accepted view of regulation of phosphate transport in E. coli is that Pit is constitutively expressed (Rosenberg et al., 1977). Nevertheless, the finding that divalent cations, particularly Mg(II) and Ca(II) (van Veen et al., 1994), are essential for Pit activity and that Zn(II) ions can be transported via this system (Beard et al., 2000), prompted an investigation of transcriptional regulation of pitA in response to Zn(II) and phosphate provision. Previous studies with the zntA promoter have emphasized the importance of employing single-copy fusions rather than plasmid-based constructs. Thus, use of a Φ(zntA–lacZ) monolysogen revealed marked regulation of zntA [encoding the Zn(II) exporter] by Zn(II), Cd(II) and Pb(II), consistent with the ATP-driven transport of these ions (Beard et al., 1997; Rensing et al., 1997, 1998). In contrast, a pUC-based fusion did not respond even to 0.5 mM Cd(II) and only slightly to 0.5 mM Pb(II) (Brocklehurst et al., 1999). The discrepancies in these data sets may also include a contribution from the complex Luria broth medium used (Brocklehurst et al., 1999), the constituents of which can dramatically influence metal bioavailability (Hughes & Poole, 1991).
For the present work, we therefore constructed a Φ(pitA–lacZ) monolysogen using well-established methods, and grew this strain in a chemically-defined medium (GGM) to explore the effects of extracellular Zn(II) and Pi on pitA expression. This medium (Beard et al., 1997; Lee et al., 2005) not only allows phosphate concentrations to be experimentally manipulated, but also minimizes formation of metal phosphates and consequent unintentional loss of metal bioavailability (Hughes & Poole, 1991). β-Galactosidase activities of strain RKP2914 [Φ(pitA–lacZ), pitA+] are shown in Fig. 1a. Fusion activity (c. 250 Miller units) was unaffected by decreasing phosphate concentration (from 1 to 0 mM added phosphate) both when the medium was not supplemented with Zn(II) [nominally 6.1 μM Zn(II)] and with 50 μM additional zinc. However, when the Zn(II) was increased to 150 μM, particularly in the absence of added Pi, there was a significant increase (typically 1.9-fold over the basal level, P<0.001) in Φ(pitA–lacZ) activity.
The Pst system is repressed at Pi concentrations >1 mM and so Pit is considered to be physiologically relevant at these higher Pi concentrations. We therefore tested the effect of a mutation in Pit on pitA gene expression at a range of Pi concentrations. In a pitA mutant (Fig. 1b), there was a marked increase in Φ(pitA–lacZ) activity under all conditions tested. In the absence of Zn(II) or at 50 μM Zn(II), irrespective of phosphate concentration, there was typically a twofold increase in Φ(pitA–lacZ) activity relative to basal levels (Fig. 1a). However, maximal Φ(pitA–lacZ) activity occurred at high added Zn(II) (150 μM) and increased further as the phosphate concentration was reduced; maximal Φ(pitA–lacZ) activity was observed without additional phosphate and was approximately fivefold higher than basal levels.
In GGM, all phosphorus detected by ICP-ES could be accounted for by the organic phosphate (data not shown); because of the low Pi concentration, the Pst transport system will be derepressed (Rosenberg et al., 1977) and, even in the pitA mutant, intracellular phosphate levels will be adequate. The data of Fig. 1 are therefore consistent with pitA expression being positively regulated by Zn(II), not by phosphate depletion. The explanation for the higher Φ(pitA–lacZ) activities in the pitA mutant (Fig. 1b) is unclear but might be linked to a reduced level of Pi in that strain; in the wild-type strain, Zn(II) is transported as a soluble, neutral metal phosphate complex, MeHPO4, where Me represents the divalent cation (van Veen et al., 1994). That maximal Φ(pitA–lacZ) activity was observed in the absence of added Pi, but in the presence of additional Zn(II), probably reflects maximal bioavailability of Zn(II) in the medium allowed by the lack of formation of metal phosphate complexes.
Trace element analyses of strains RKP2914 and RKP2935 in the presence of zinc and phosphate
To correlate pitA transcription with total cellular levels of zinc and phosphate, we analysed cell digests prepared after careful washing of cells to remove adventitious and loosely-bound metals and nutrients according to established procedures. We recognize that all elements will be determined in these analyses, irrespective of discrete subcellular pools or compartments, and that metal binding to cell walls is not quantified. Intracellular zinc levels were unaffected in the wild-type strain by Zn(II) added to the medium (Table 1). This is anticipated, given (1) the Zn(II) homoeostatic mechanisms employed by E. coli and (2) the presence of adequate zinc levels in Zn(II)-unsupplemented GGM (6.1 μM calculated; typically 5.2 μM on analysis, not shown). The most marked change in elemental composition elicited by added Zn(II) was in Mg: levels (per gram dry biomass) fell by eightfold in cells grown with added Zn(II) and by 20-fold when Zn(II) and Pi were added (Table 1). Because Mg(II) is cotransported with Pi by the PitA system, the data are consistent with competition or inhibition by Zn(II) of this route of metal uptake. Presumably, if Zn(II) is taken up in competition with Mg, export mechanisms such as ZntA maintain intracellular Zn(II) homoeostasis. Cellular phosphorus also fell on supplementation with Zn(II) and Pi; although the fold differences were not large, they were statistically significant.
Table 1. Elemental composition of cells harvested from GGM supplemented with Zn and Pi and the effects of a pitA mutation
* Values are means of two growths and, for each, five replicate assays of the cell digests. Superscripts indicate statistical significance using the Student's t-test: 1P<0.001 , 2P<0.01 , 3P<0.02 , 4P<0.05 . For rows 2 and 3, the values are compared with row 1 (wild-type strain, no supplements); for row 6, the values are compared with row 4 (pitA mutant, no supplements); for row 4, the values are compared with row 1.
In the pitA mutant, zinc levels appeared to be elevated relative to the mutant but the differences were not significant (P>0.05). Previously (Beard et al., 2000), we showed that the pitA mutant accumulated higher levels of zinc when in Zn-supplemented media [>2.5 mM Zn(II)] but that work was conducted on cultures grown in Luria–Bertani, where medium constituents markedly reduce metal bioavailability (Hughes & Poole, 1991). Table 1 shows that the most striking change was again in Mg levels; in the mutant, the values (0.27 mg g−1 dry cell mass) were less than one half of the wild type (0.6 mg g−1 dry cell mass) and decreased further on addition of Zn(II) and Pi. Phosphorus levels in the pitA mutant were slightly reduced compared with the wild-type strain but restored on growth with Zn(II) and Pi.
Divalent cations, such as Zn(II), Mg(II) and Ca(II) are essential for Pi uptake via Pit (Russell & Rosenberg, 1980). These cations, together with Co(II) and Mn(II), form with PO43− a soluble, neutral metal phosphate complex, MeHPO4, where Me represents the divalent cation (van Veen et al., 1994). In Bacillus subtilis, uptake of these metal ions is stimulated by Pi and a pitA mutant exhibited reduced transport of Ca(II) and Co(II) (Kay & Ghei, 1981). Although transport of Zn(II) via Pit was not initially reported (van Veen et al., 1994), we showed that a Zn(II)-resistant mutant carried a single insertion in pitA, and the mutant accumulated less zinc when grown at elevated Zn(II) levels (Beard et al., 2000). These data are consistent with the involvement of PitA in Zn(II) uptake, probably via formation of a ZnHPO4 complex. Thus, Zn(II), Mg(II) and Ca(II) are potential cosubstrates (with Pi) for PitA. The major alternative route for Pi transport is the inducible Pst system (Rosenberg et al., 1977; Willsky & Malamy, 1980). In Salmonella, Mg(II) enters via CorA, MgtA or MgtB (Smith & Maguire, 1998). We propose that, in a Pit+ strain, the elevation of extracellular Zn(II) dramatically diminishes cellular Mg levels (Table 1) by competing with these cations and elicits a modest increase in pitA transcription. Increasing Zn(II) concentration to 150 μM is not sufficient for elevated pitA transcription and requires additionally reduced phosphate provision, perhaps to maximize Zn(II) bioavailability.
Under all conditions tested, a mutation in pitA significantly increased Φ(pitA–lacZ) activity. In this mutant, the major route for Pi uptake will presumably be via Pst; in addition, Pi may enter as an analogue for either glycerol-3-phosphate or glucose-6-phosphate Pi via glpT or uhpT, respectively (Hayashi et al., 1964; Pogell et al., 1966; Winkler, 1966). In the pitA mutant, intracellular levels of Mg were less than one half of wild-type levels and further depressed in the presence of 150 μM zinc. This suggests that Zn(II) also affects a Pit-independent divalent cation transport pathway. Further work is required to explore this effect.
The mechanism of regulation of pitA transcription is not understood. In the putative promoter (200 bp upstream of the pitA start codon), no DNA-binding sites for the following transcription factors are evident: MerR, CueR, ZntR (all MerR family), Zur, NikR, CusR, ZraR, CpxR and ModE. There is a partial, poor match to the Fur consensus centred at −33.5: GATAATGCGCCGCGTTCAT (underlining representing a match with the published site; Lavrrar & McIntosh, 2003). However, microarray ChIP-on-Chip (chromatin immunoprecipitation-on-chip) data, combined with Prodoric prediction (http://prodoric.tu-bs.de/) of Fur sites does not reveal any Fur site for pitA (S. Andrews, pers. commun.).
In conclusion, the data in this paper demonstrate for the first time that pitA expression in E. coli is not constitutive. Transcription is modulated in a chemically-defined medium by the availability of Pi and of Zn(II), and is upregulated in a pitA deletion mutant. The possibility of interactions between metal ions and added phosphate and the failure to correlate readily intracellular element levels with Φ(pitA–lacZ) activity makes an understanding of the mechanism difficult. Nevertheless, the evidence points to positive regulation of pitA by free, intracellular Zn(II), with phosphate starvation being an additional determining factor. The data also support a model in which Zn(II), as well as other divalent cations, is cotransported with phosphate by PitA.
We are grateful to the BBSRC for a Research Grant to R.K.P. and C.W.McL. We are most grateful to Dr Simon Andrews for communicating his unpublished data on Fur sites.