A tenth atp gene and the conserved atpI gene of a Bacillus atp operon have a role in Mg²⁺ uptake

Abstract

The atp operon of alkaliphilic Bacillus pseudofirmus OF4, as in most prokaryotes, contains the eight structural genes for the F-ATPase (ATP synthase), which are preceded by an atpI gene that encodes a membrane protein of unknown function. A tenth gene, atpZ, has been found in this operon, which is upstream of and overlapping with atpI. Most Bacillus species, and some other bacteria, possess atpZ homologues. AtpZ is predicted to be a membrane protein with a hairpin topology, and was detected by Western analyses. Deletion of atpZ, atpI, or atpZI from B. pseudofirmus OF4 led to a requirement for a greatly increased concentration of Mg²⁺ for growth at pH 7.5. Either atpZ, atpI, or atpZI complemented the similar phenotype of a triple mutant of Salmonella typhimurium (MM281), which is deficient in Mg²⁺ uptake. atpZ and atpI, separately and together, increased the Mg²⁺-sensitive ⁴⁵Ca²⁺ uptake by vesicles of an Escherichia coli mutant that is defective in Ca²⁺ and Na⁺ efflux. We hypothesize that AtpZ and AtpI, as homooligomers, and perhaps as heterooligomers, are Mg²⁺ transporter, Ca²⁺ transporter, or channel proteins. Such proteins could provide Mg²⁺, which is required by ATP synthase, and also support charge compensation, when the enzyme is functioning in the hydrolytic direction; e.g., during cytoplasmic pH regulation.

Prokaryotic atp operons encode the cell membrane F-type ATPase (ATP synthase) that couples the energy of an electrochemical H⁺ gradient (or sometimes Na⁺), to the synthesis of ATP, from ADP and P_i. In the reverse reaction, the ATPase hydrolyzes ATP concomitant with H⁺ (or Na⁺) efflux, thereby contributing to cytoplasmic pH regulation and/or generation of a transmembrane electrochemical gradient under fermentative conditions (1-4). Most atp operons, like that of Escherichia coli, contain the eight structural genes for the ATPase, atpBEFHAGDC, which are preceded by atpI (5). The Escherichia coli atpI is expressed, and its product associates with the membrane, as predicted from its deduced sequence (6-9). Whereas there is no demonstrated effect of AtpI on expression or assembly of the ATPase, an atpI deletion strain of E. coli has been reported to have a reduced growth yield (7). There is no function established for this “mysterious ninth gene” (10) that accounts for such an effect. We report here the finding of another gene, encoding a membrane protein, that is upstream of the atpI gene, and within the atp operon of alkaliphilic Bacillus pseudofirmus OF4. This gene, designated atpZ, was discovered during attempts to introduce site-directed changes in alkaliphile-specific motifs of the membrane-embedded F-ATPase subunits of B. pseudofirmus OF4 (11). A cassette introduced just upstream of the putative atp promoter abolished atp expression. This finding led us to reexamine the location of the atp operon promoter, to the inclusion of atpZ in the extended operon, and then to an exploration of the effects of deleting atpI as well as atpZ. The results suggest a cation translocation function for AtpI and AtpZ, which could be widely relevant to the diverse prokaryotes that possess homologues of one or both of these proteins.

Materials and Methods

Bacterial Strains, Plasmids, and Growth Conditions. Alkaliphilic Bacillus pseudofirmus OF4 strain 811M, a methionine auxotroph of B. pseudofirmus OF4 (12), was the parent strain for deletion mutants in atpZ, atpI, and atpZI. Routine cloning was performed in Escherichia coli DH5α, except for pG+host4 plasmids (Appligene, Pleasanton, CA), which were cloned in E. coli XL-1 Blue (Promega). The phenotypic effects of atpZ, atpI, and atpZI were examined in a Na⁺/H⁺ and Ca²⁺/H⁺ antiporter mutant of E. coli, KNabc (ΔnhaAΔnhaBΔchaA) (13), and in the Mg²⁺ uptake mutant of Salmonella typhimurium MM281 (corA45::MudJ mgtA21::MudJ mgtB10::MudJ) (14). Inducible gene expression was carried out in E. coli C43(DE3) (15). All plasmid inserts and PCR products were completely sequenced at the Utah State Biotechnology Center (Logan, UT), or at the DNA Core at the Mount Sinai School of Medicine. Ampicillin (Ap) was used at 100 μg·ml^-1 in E. coli and S. typhimurium, and erythromycin (Em) selection was conducted at 250 μg·ml^-1 in E. coli. For B. pseudofirmus OF4811M in complex or defined media, Em was used at 0.6 μg·ml^-1, and in protoplast regeneration DM3 plates (16), 0.1-0.3 μg·ml^-1 Em was used.

Complementation of the Mg²⁺-uptake defect of S. typhimurium MM281 was determined by passing 100-μl portions of overnight cultures (grown in LB/Ap medium plus 100 mM MgSO₄) that were pelleted and resuspended in LB medium (with no MgSO₄ or Ap) into 2 ml of test medium (LB/Ap with different concentrations of MgSO₄). The tubes were shaken at 37°C for 6.5 h, and the absorbance at 600 nm was determined. Alkaliphile strains were grown in semidefined (16) or defined media at 30°C, with glucose or malate as the carbon source. The media were buffered with 0.1 M Mops-NaOH at pH 7.5, and 0.1 M Na₂CO₃/NaHCO₃ at pH 10.5. The [Na⁺] of the pH 7.5 buffer was raised to 0.2 M Na⁺ by the addition of NaCl. In the defined media, QA, the yeast extract was omitted, and 0.1% glutamine and 0.1% alanine, 1 μg/ml thiamine and biotin, and 10 μg/ml methionine were added. For growth experiments with transformed alkaliphile strains, overnight cultures grown in a rich medium at pH 7.5 (16), containing 50 μg/ml kanamycin (Km), were diluted 200-fold into defined medium to start the experiment. The richer medium overcame deleterious effects of the plasmids on growth. Carryover from this medium resulted in the faster growth of transformed, but not of untransformed strains, in complementation experiments (Fig. 4 B vs. A).

Fig. 4.

Complementation of the Mg²⁺ phenotype of S. typhimurium MM281 by atpZ, atpI, and atpZI. S. typhimurium MM281 transformants were grown for 6.5 h in LB/Ap medium, supplemented with the indicated concentrations of MgSO₄ as described in Materials and Methods. The averages and SD of five independent growth experiments, each done in duplicate, are shown.

Transcriptional Start Site Mapping. RNA was isolated as described (17) from late-logarithmic cultures of B. pseudofirmus OF4 grown in semidefined media at pH 10.5. Reverse transcriptase reactions were carried out by using the SUPERSCRIPT preamplification system (GIBCO/BRL, Life Technologies, Carlsbad, CA), according to the manufacturer's instructions. RNA (50 μg) and 50 pmol of a primer that had been end-labeled with [³²P]ATP by using T4 polynucleotide kinase were used in each reaction. Two primers were used: 2700R in atpZ, and 3024R in atpI (see Fig. 1B). The latter primer corresponds to that used in the earlier study (17) that mapped the start site to base 2925 (all numbering is based on the deposited sequence in the GenBank database, accession number AF330160). The reverse transcriptase reactions were resolved on an 8 M urea/6% polyacrylamide denaturing gel (12 × 16 cm), along with ³²P-labeled 100-bp markers. They were further analyzed on DNA sequencing gels, by using a Bio-Rad apparatus, along with a sequencing ladder generated by using primer 2700R or primer 3024R, and a template consisting of the region from 2389 to 3108, which was obtained by PCR, and cloned into pGEM7Zf(+). RT-PCRs were carried out by annealing primer 3378R, corresponding to the 5′ end of atpB, to 1 μg of RNA. The reverse transcription (RT) reaction (0.5-2 μl of 20 μl) was then used as the template for PCR, by using the HotStarTaq master mix kit (Qiagen, Valencia, CA).

Fig. 1.

atp operon and upstream region with proposed promoter elements of B. pseudofirmus OF4. (A) atp operon structure showing an upstream gene, vpr (an incomplete ORF), and the atpBEFHAGDC structural genes, as well as the Z and I genes that are proposed to be part of the atp transcriptional unit. The solid arrow indicates the proposed promoter from this article, the dotted arrow is the previously described promoter (17), and the upward arrowhead indicates the site of insertion of a spectinomycin-resistance cassette in a preliminary experiment that led to a malate-negative phenotype, and hence to this article. (Upper) The bar is marked in kb. Below the operon is an expansion of the region from ≈2 to 5 kb that shows the regions deleted in the mutants constructed for this article. (B) The sequence from bases 2541-3052. The primers used in the transcriptional start mapping (3024R and 2700R) are shown. The start site from the earlier study is shown with a downward dotted arrow (base 2925), and the new start site is indicated with a downward solid arrow (base 2613). The proposed -35 promoter and -10 promoter elements are bold. The primers used in conjunction with primer 3378R in the RT-PCR of Fig. 2B are also shown (2568F and 2619F). The one-letter amino acid designations are shown above (atpZ) or below (atpI) their codons.

Construction of Mutants. Mutant constructs were made in pG⁺host4 carrying a temperature-sensitive replicon, and recombinant plasmids were introduced into the alkaliphile by protoplast transformation. The general strategy for making the mutant construction was the one described by Biswas et al. (18), as applied to alkaliphiles (16, 19, 20).

For construction of in-frame ΔatpZ, ΔatpI, and ΔatpZI mutations, two sets of PCRs for each mutant were performed on wild-type chromosomal DNA to amplify and introduce appropriate restriction sites into segments upstream and downstream of the region to be deleted: nucleotides 2740-2889 for ΔatpZ, nucleotides 2924-3292 for ΔatpI, and nucleotides 2740-3292 for ΔatpZI. Digests of the upstream and downstream pair of PCR products were prepared, and then were ligated to the appropriately digested low-copy Gram-negative vector pMW118 (Nippon Gene, Toyama, Japan). EcoRI digests of each of these plasmids were ligated with EcoRI-digested pG⁺host4 to produce a chimeric plasmid that was used to create each of the deletions in the alkaliphile. A growth temperature of 40°C was used during manipulations in E. coli, so that maintenance of the plasmid depended on the low-copy replicon of pMW118, thus minimizing potential toxicity of the hydrophobic gene products.

Cloning of atpZ, atpI, and atpZI. A PCR product generated from B. pseudofirmus OF4 chromosomal DNA with designed BamHI and EcoRI sites was cloned into pMW118. The product contained atpZ and 1.3 kb of upstream region. atpZI was cloned in a similar way. For the cloning of atpI, the same strategy was followed, except that the template for the PCR was chromosomal DNA from the ΔatpZ mutant. Expression of atpI in this construct would be under the control of the native promoter of the operon. For overexpression of atpZ, the gene was cloned into the vector pET-3a (Novagen), by using designed NdeI and XhoI sites at the 5′ and 3′ end of atpZ, respectively. Everted membrane vesicles were isolated from IPTG-induced cultures of E. coli C43(DE3) harboring the empty plasmid or pET-3a-atpZ. For the expression of atpZ in B. pseudofirmus OF4, a PCR product (using Vent DNA polymerase, New England Biolabs) that contained atpZ and 1.3 kb of upstream sequence was blunt-end-cloned into the shuttle vector pYH56 (16).

ATPase and Protein Assays. Everted membrane vesicles were prepared from strains grown to mid-logarithmic stage in defined media at pH 7.5 and 10.5 with malate as the carbon source, and were assayed for octyl glucoside-stimulated ATPase activity (21). Protein was measured by the Lowry method (22).

Assays of ⁴⁵Ca²⁺ Accumulation and Uptake by E. coli Transformants. E. coli KNabc transformed with pMW118, or the recombinant plasmid with atpZ, atpI, or atpZI were grown overnight in LBK (LB medium containing KCl instead of NaCl) (23) and Ap. The cells were washed once and were resuspended in 50 mM 1,3-bis[tris(hydroxymethyl)methylamino]propane buffer, pH 7.5. Three milliliters of cell suspension (≈1 mg of protein per ml) were shaken in 15-ml conical tubes at 37°C with 100 μM ⁴⁵CaCl₂ (0.1 μCi/mmol; 1 Ci = 37 GBq). At the end of a 1-h incubation, with 1 mM glucose where indicated, 1 ml was removed and vacuum filtered by using 25-mm GF/F glass microfiber filters (Whatman). The air-dried filters were counted by liquid scintillation counting. Experimental values were corrected for binding. Binding was assessed in the presence of 5% butanol, which released internalized solute. For ⁴⁵Ca²⁺ uptake assays, right-side-out membrane vesicles were prepared from the same transformants of E. coli KNabc by the method of Kaback (24) in 10 mM Bis-Tris propane, pH 7.5. ⁴⁵Ca²⁺ uptake was measured by adding 100 μM ⁴⁵CaCl₂ to a 1-ml suspension containing 0.25 mg vesicle protein. At 5-sec intervals during incubation at 37°C, the vesicle suspension was rapidly filtered through 0.45-μm HAWP filters (Millipore) and washed with 2.5 ml of the Bis-Tris propane buffer. Radioactivity was measured by scintillation counting.

Western Analysis. Everted membrane vesicles were prepared from E. coli transformants grown on LB/Ap medium, and the indicated alkaliphile strains grown on pH 7.5 QA media, with malate and 50 μg/ml kanamycin. The vesicles, denatured in SDS sample buffer, were resolved on mini-SDS/12% polyacrylamide gels (25). After transfer to nitrocellulose membranes, Western blots were developed by a chemiluminescence protocol (Amersham Pharmacia Biosciences, Piscataway, NJ), according to manufacturer's recommendations. A synthetic peptide was prepared that corresponded to the last 15 amino acids of AtpZ, with an additional cysteine at the N terminus. The peptide, conjugated to keyhole limpet hemocyanin, was injected in rabbits to raise polyclonal antisera (Covance Research Products, Denver, PA), from which the IgG fraction was purified (26), and used for these analyses.

Results

Reexamination of the Transcriptional Start and Probable Promoter of the atp Operon. Introduction of a cassette in the chromosomal region just upstream of atpI, in the position within atpZ indicated by the arrowhead in Fig. 1 A, resulted in a malate-minus growth phenotype, and loss of ATPase expression in B. pseudofirmus OF4, even though the promoter had earlier been predicted to be downstream of the cassette-insertion site, just upstream of atpI (17). To reexamine that promoter identification, primer extension was carried out with the same primer, 3024R, that was used in the earlier work (17). This primer sequence is in atpI, as shown in Fig. 1B. The product of the RT reaction was electrophoresed on a polyacrylamide/urea gel that enabled approximate sizing of the product from ≈50 to 500-600 nucleotides, a broader range than would have been detected in the earlier experiments (17). Two products of the reaction were visualized: one that corresponded approximately to the size originally determined, which was 85 nucleotides, and a second product of ≈400 nucleotides. To precisely size the larger product by primer extension, it was desirable to use a primer upstream of 3024R, i.e., 2700R. The resulting smaller reverse transcriptase product was determined to be 67 or 68 nucleotides on a 6% polyacrylamide/urea sequencing gel (data not shown). This result indicated that the transcriptional start site is base 2613 or 2614 of the deposited sequence, which is 293 (292) nucleotides upstream of the predicted start codon of atpI. The smaller product observed with the 3024R primer presumably resulted from either premature termination of the RT reaction, because of RNA secondary structure, or from a processed atp transcript fragment.

Additional support for an atp transcriptional start near base 2613 was obtained by RT-PCR. The reverse transcriptase reaction was carried out by using a primer corresponding to a sequence from the 5′ end of atpB (3378R), which was the first ATPase structural gene in the operon. The RT product was then subjected to PCR, by using two sets of primers. For the first set, 3378R was used with 2619F, whose 5′ end is six nucleotides downstream of the proposed start site. In the second set, 3378R was used with 2568F, whose 3′ end is 25 nucleotides upstream of the start site (Fig. 1B). As shown in Fig. 2A, a product of the right size, 760 base pairs, was observed for the wild-type with the set 1 primers, whereas no product was generated from primer set 2. The set 1 product was absent when the RT enzyme was omitted from the RT reaction, indicating that the observed product was not due to DNA contamination of the RNA preparation. RT-PCR was also carried out on RNA from the ΔatpI mutant, and similar results were obtained, except that the size of the set 1 product was approximately the predicted size of 392 base pairs, taking into account the deleted region.

Fig. 2.

Demonstration of a promoter upstream of atpZ and of an AtpZ protein. (A) RT-PCR was carried out by using primer 3378R for the RT reaction. The reaction was subjected to PCR by using 3378R, with either primer 2568F (set 1) or 2619F (set 2; see Fig. 1B for precise locations of these primers). The reverse transcriptase enzyme was left out (-) or included (+) in the RT reaction as indicated. The control lanes show the product generated from PCR with the appropriate primer pairs, with chromosomal DNA as the template. The standards are labeled in base pairs. Below the gel is a diagram showing the location of the primers, with the reverse primer in atpB, and the forward primers either just after the proposed start site (set 1) or before the start site (set 2), as indicated by the numbered arrows. (B) Western blot of atpZ expression in E. coli and B. pseudofirmus OF4. In E. coli, everted membrane vesicles were isolated from induced cultures of strains with the empty vector, or with the cloned atpZ. In B. pseudofirmus OF4, everted membrane vesicles were prepared from the wild-type carrying pYH56 (empty vector), or ΔatpZ carrying the empty vector, or with cloned atpZ. Lane 1, E. coli with empty vector (0.25 μg); lane 2, E. coli with atpZ (0.25 μg); lane 3, B. pseudofirmus OF4 wild-type with empty vector (100 μg); lane 4, ΔatpZ with empty vector (100 μg); lane 5, ΔatpZ with atpZ (100 μg). The AtpZ band is marked on the left, and the other bands represent nonspecific reactions. Molecular weight markers are indicated on the left. (C) Alignment of AtpZ homologues. Boxed residues are charged amino acids shared by three of four of the candidate proteins. Other residues shared by three or four proteins are shaded. The two predicted transmembrane helices (TMS) are overlined. Bps, B. pseudofirmus OF4; Bh, Bacillus halodurans C-125; Bm, Bacillus megaterium; Ba, Bacillus anthracis. According to the National Center for Biotechnology Information ORF finder, the B. anthracis candidate has an additional 10 residues at the N terminus for which no strong ribosome binding site (RBS) can be identified; there is a good RBS for the leucine codon that lines up with the first amino acid of B. pseudofirmus OF4 AtpZ. The B. megaterium candidate shown is an incomplete ORF with at least 14 more amino acids at the N terminus. In this alignment, these extended segments have not been included.

The suggested promoter sequence shown in Fig. 1B has the consensus sequence for the Bacillus subtilis -35 element, TTGACA, and a -10 element, TACGAT, which is an infrequently used element in B. subtilis (27). The transcriptional start site is four nucleotides upstream of the predicted start codon of atpZ (National Center for Biotechnology Information ORF finder), a putative ORF upstream of atpI that had not earlier been considered part of the operon. A more likely start for AtpZ is that shown in Fig. 1B, which is based on the presence of a potential ribosome-binding site, and the size of the product detected in Western blots by antibodies raised against a synthetic peptide corresponding to the C-terminal sequence of AtpZ (see Fig. 2B). An alignment of several possible homologues of AtpZ from Bacillus species is shown in Fig. 2C, and additional homologues are found in the database entries for genomes of Clostridium difficile and Desulfitobacterium hafniense.

Detection of AtpZ. Initial assessments of whether atpZ actually produces a protein product were conducted by Western blot analysis of membranes from wild-type B. pseudofirmus OF4 grown in semidefined malate medium at pH 7.5 and 10.5, but no AtpZ was detected. Apparently the antibody could not detect the low levels of AtpZ found in such cells. Therefore, atpZ was cloned into the plasmid pET-3a behind an IPTG-inducible promoter, by using the start codon deduced from this article. As shown in Fig. 2B, AtpZ was expressed at high levels in E. coli C43(DE3) (lane 2); the negative control lacked a reactive product of the same size. atpZ was also expressed in the alkaliphile ΔatpZ mutant transformed with recombinant pYH56 containing the atpZ coding sequence, and 1 kb of upstream sequence. A band was observed that was the same size as that seen in E. coli (lane 5), supporting the conclusion that the actual start site corresponds to that shown in Fig. 1B. No AtpZ signal was detected from either the ΔatpZ, or from the wild-type strains harboring the empty plasmid (lanes 3 and 4).

Phenotype of Deletion Mutants. The ΔatpZ and ΔatpI mutant strains of B. pseudofirmus OF4 grew like the wild-type strain on semidefined media, with either malate or glucose as the carbon source. The specific ATPase activities in membrane vesicles from ΔatpZ and ΔatpI were similar to the wild-type (data not shown), which was consistent with the nonfermentative growth on malate. This result confirmed the nonpolar nature of the deletions. Little growth deficit was observed on defined medium at pH 10.5, ruling out a specific role of atpZ or atpI in alkaliphily. However, the mutant strains showed a pronounced growth deficit when grown in a defined medium at pH 7.5. Recent studies (28-31) suggest a role in Mg²⁺ transport for several putative 2 TMS proteins. This finding led us to consider the possibility that one or both of the first atp operon genes was involved in Mg²⁺ acquisition. Because of the poor solubility of Mg²⁺ in high-pH settings, the alkaliphile is likely to require special Mg²⁺ acquisition mechanisms for growth at pH 10.5, but might have a lower overall capacity for this important function at pH 7.5. As shown in Fig. 3A, all three deletion strains exhibited a greatly increased requirement for Mg²⁺ in malate-containing QA-Mops media at pH 7.5. Although not shown, a growth deficit was also observed when glucose was the primary carbon source instead of malate. As shown in Fig. 3B, transformation of the ΔatpZ strain with a plasmid-bearing atpZ resulted in nearly wild-type levels of growth over a broad range of MgSO₄ concentrations. There were differences in pregrowth conditions and Mg²⁺ carryover (see Materials and Methods), which were necessitated by effects of plasmids on growth. This finding led to the apparent ability of the wild-type transformant and complemented mutant to grow at lower Mg²⁺ concentrations than the untransformed wild-type shown in Fig. 3A.

Fig. 3.

Effect of [MgSO₄] on the growth of deletion strains of B. pseudofirmus OF4. (A) The MgSO₄ concentration was varied for overnight growth in the pH 7.5 QA-Mops-defined media described in Materials and Methods. (B) Growth of the indicated transformants of wild-type B. pseudofirmus OF4 and the atpZ mutant was monitored as in A. The enhanced response of the wild-type with the control vector to low [MgSO₄] was due to a carryover of Mg²⁺ from the richer medium used for the pregrowth of transformed alkaliphile strains, as described in Materials and Methods. The results are from two independent experiments, each carried out in duplicate.

Effects of atpZ, atpI, or atpZI Expression on the Phenotypes of Heterologous Transport Mutants. S. typhimurium MM281 is deficient in Mg²⁺ uptake by virtue of disruptions in the genes for three different transport systems, CorA, MgtA, and MgtB (14); complementation of this strain has proved useful in identifying and developing information about new, heterologous Mg²⁺ transporters (32, 33). As shown in Fig. 4, low-copy plasmids expressing atpZ, atpI, or atpZI complemented the growth phenotype of S. typhimurium MM281. The AtpZI combination supported significantly greater growth of the mutant on contaminating and low added concentrations of Mg²⁺ than did either AtpI or AtpZ alone.

Because of the lack of availability of radioactive isotopes of Mg²⁺, assays of Mg²⁺ transport are usually performed as the Mg²⁺-dependent inhibition of transport of other divalent cations. This method was used to characterize the MgtE Mg²⁺ transporter of B. pseudofirmus OF4 (32), which exhibited Mg²⁺ -inhibitable uptake of ⁵⁷Co²⁺ in S. typhimurium MM281 transformants. However, no evidence for AtpZ-, AtpI-, or AtpZI-dependent Co²⁺ uptake was obtained, although such activity was conferred by the positive control plasmid in which the alkaliphile mgtE gene was cloned (data not shown). Because crossinhibition occurs between Mg²⁺ and Ca²⁺ with some transporters (34), the possibility that AtpZ and AtpI might confer Mg²⁺-inhibitable ⁴⁵Ca²⁺ accumulation was assessed in E. coli KNabc, which is deficient in Ca²⁺ efflux (13). As shown in Table 1 for a 1-h accumulation experiment, the control transformant (cells harboring the empty vector) exhibited little Mg²⁺-inhibitable ⁴⁵Ca²⁺ accumulation. The transformant expressing atpI exhibited a significant increase in activity only when glucose was also added, but not as much as the transformants expressing atpZ and atpZI, both of which exhibited enhanced accumulation even in the absence of glucose. In followup experiments, right-side-out membrane vesicles from the same transformants were assayed for ⁴⁵Ca²⁺ uptake in the absence of any intravesicular solute. Initial experiments showed that uptake was extremely rapid, such that the highest uptake was observed at the shortest time points that could be taken (5 sec). Subsequent time points indicated a loss of ⁴⁵Ca²⁺ from the vesicles, as would be expected under unenergized conditions. Addition of an electron donor, D-lactate, did not cause more sustained uptake, but instead inhibited uptake overall. This result was presumed to reflect energization of the remaining Ca²⁺ efflux pathway(s), which were expected to be present in the chaA deletion background of E. coli KNabc (35). Subsequent vesicle experiments were conducted in the absence of added electron donors at 5-sec points. As shown in Table 2, AtpZ-, AtpI- and AtpZI-dependent ⁴⁵Ca²⁺ uptake was observed. This uptake, after subtracting the background binding and uptake in the control transformant, was completely inhibited by addition of either 1 mM MgCl₂ or 200 μM of the general cation channel inhibitor GdCl₃ (36) to the extravesicular space, just before the addition of ⁴⁵Ca²⁺ to start the reaction.

Table 1. AtpZ-, AtpI-, and AtpZI-enhanced ⁴⁵Ca²⁺ accumulation by transformants of Escherichia coli KNabc.

	45Ca2+ accumulation, nmol/mg of cell protein*
	−Glucose		+Glucose
Transformed with plasmid	No add	+MgCl2	No add	+MgCl2
pMW118	2.1 ± 0.3	2.4 ± 0.4	2.8 ± 0.1	1.9 ± 0.2
pMW-atpZ	4.5 ± 0.8	2.3 ± 0.4	11.8 ± 0.1	1.9 ± 0.1
pMW-atpl	2.1 ± 0.2	2.2 ± 0.2	7.0 ± 0.1	2.7 ± 0.2
pMW-atpZI	11.0 ± 0.1	2.2 ± 0.2	16.3 ± 0.2	2.2 ± 0.1

^*Transformant cells were incubated with 100 μM ⁴⁵Ca²⁺ for 1 h, with or without 1 mM MgCl₂, after which the accumulation of ⁴⁵Ca²⁺ was assayed as described in Materials and Methods. All values are corrected for binding controls, which were carried out by butanol treatment. The results are the averages of six independent experiments conducted in duplicate and are shown ± SD. No add, no MgCl₂ added.

Table 2. AtpZ-, AtpI-, and AtpZI-dependent ⁴⁵Ca²⁺ uptake in right-side-out membrane vesicles of Escherichia coli KNabc.

Vesicles from transformants with plasmid	45Ca2+ uptake, nmol/5 sec per mg of vesicle protein*
	No add	+1 mM MgCl2	+200 μM GdCl3
pMW-atpZ	8.5 ± 1.3	0.5 ± 0.1	0.8 ± 0.2
pMW-atpI	9.5 ± 1.5	0 ± 0.2	0.8 ± 0.2
pMW-atpZI	6.6 ± 1.1	0.1 ± 0.2	0.6 ± 0.2

^*The values shown were corrected for the ⁴⁵Ca²⁺ bound and taken up by the vesicles from the control vector (pMW118) transformant. The values subtracted were 18.8 ± 1.6, No add; 12.8 ± 1.2, +1 mM MgCl₂; and 6.0 ± 0.6, +GdCl₃. The results, shown with SD, are the averages from six independent vesicle experiments assayed in duplicate. No add, no MgCl₂ added.

Discussion

The results indicate that a tenth gene, designated atpZ, is part of the B. pseudofirmus OF4 atp operon. Because atpZ homologues are found in a comparable position in atp operons of several nonalkaliphilic Bacillus species, as well as C. difficile and D. hafniense, the indications that AtpZ and AtpI have a role in divalent cation translocation may have broad applicability. In addition to the phenotypes of the atpZ, atpI, and atpZI mutants of the alkaliphile, the enhancement of Mg²⁺ acquisition in the Mg²⁺ uptake-deficient mutant strain, S. typhimurium MM281, and of Mg²⁺-inhibitable ⁴⁵Ca²⁺ uptake in E. coli KNabc, are further evidence that AtpZ and AtpI support inward Mg²⁺ transport. In the whole-cell experiments in the heterologous systems, the relative effectiveness of the alkaliphile genes was atpZI > atpZ > atpI in enhancing both growth of the Mg²⁺ uptake mutant of S. typhimurium and the Mg²⁺-sensitive accumulation of Ca²⁺ in E. coli. Only in the vesicle uptake experiments did atpI seem to confer higher activity than the other genes, with atpZI now conferring the lowest. The likely explanation for this deviation from the whole-cell experiments is that the measurements in vesicles were too slow to capture the peak points of the fastest transporters. The faster the initial transport, the further the 5-sec assay point was down the slope of the subsequent loss of internalized ⁴⁵Ca²⁺. Such a loss would reflect an expected reequilibration of the Ca²⁺ in the absence of an imposed potential.

Different properties conferred by AtpZ, AtpI, and AtpZI suggest that these molecules constitute the translocation pathway themselves, rather than acting through an indirect mechanism, by affecting some independent transporter in the cells. AtpZ and AtpI would be likely to function as homooligomers and, perhaps, AtpZ and AtpI can also form heterooligomers. Homooligomeric assemblies of CorA (37), and the 2 TMS ALR1 and Mrs2 transporters of yeast (30, 38), have been reported, and may function as channels. Although no conclusion can yet be drawn about the transport mechanism of AtpZ and AtpI, the properties observed thus far raise the possibility of a cation channel.

The unusual and diverse nature of the proteins to which a Mg²⁺ transport function has been ascribed is notable (39). The current findings add to that diversity. Neither AtpZ nor AtpI has the conserved motif of the CorA family of Mg²⁺ transporters (39) and the 2-TMS Mrs2-type transporters (28-31), nor do they show sequence similarity to other putative Mg²⁺ transporters such as MgtE, which was first isolated from an alkaliphilic B. pseudofirmus DNA library (32). The CorA family possesses a putative magnesium transporter signature, Y/FGMNF, which is critical for activity (30, 31, 37). There is no evident counterpart in AtpZ or AtpI. We note, however, that the AtpZ proteins of two alkaliphilic Bacillus species possess three contiguous acidic residues at their C terminus. A discontinuous DDE motif has been shown to function in divalent metal binding (40), and contiguous acidic residues have been associated with Ca²⁺ binding and translocating proteins (34, 41).

Coexpression of a cation translocation pathway with the F-ATPase from a single operon is a functionally satisfying combination. The nucleotide substrates for the enzyme are usually complexed with Mg²⁺, and the cation has been suggested to play a role in establishing the asymmetry of the catalytic sites (42). In addition, when the ATPase is functioning hydrolytically to lower the cytoplasmic pH (43), a capacity of AtpZI, AtpZ, or AtpI to support cation flux could allow charge-compensating, inward movement of cations that would maximize the proton efflux. AtpI gene products from different organisms do not exhibit strong sequence conservation, and are less conserved than are deduced AtpZ sequences. Perhaps inactivation, loss, or modifications in AtpZ and AtpI function will ultimately be found to relate to different physiological needs of diverse organisms.

It is expected that extremely alkaliphilic bacteria require well adapted mechanisms for acquiring needed concentrations of ions such as Fe²⁺ and Mg²⁺, when growing at pH values as high as 10.5. The transporter complement encoded in the genome of alkaliphilic Bacillus halodurans C-125 is predicted to include four distinct MgtE-like proteins (www.membranetransport.org). The finding that the phenotype of the atpZI deletion is more pronounced at near-neutral pH than at pH 10.5, suggests that the major Mg²⁺ uptake systems of B. pseudofirmus OF4 are better adapted for function at pH 10.5 than at pH 7.5. Although a B. pseudofirmus OF4 MgtE conferred significant Mg²⁺ transport capacity at pH 7.0 in a heterologous system (32), pH optima of MgtE proteins may be higher in the native alkaliphile membrane. Several examples of such hard-wired adaptations to extreme alkaliphily have been found to disadvantage B. pseudofirmus OF4 at near-neutral pH (20).