The CorA Mg2+ Transport Protein of Salmonella typhimurium

Salmonella typhimurium CorA is the archetypal member of the largest family of Mg2+transporters of the Bacteria and Archaea. It contains three transmembrane segments. There are no conserved charged residues within these segments indicating electrostatic interactions are not used in Mg2+ transport through CorA. Previous mutagenesis studies of CorA revealed a single face of the third transmembrane segment that is important for Mg2+ transport. In this study, we mutated hydroxyl-bearing and other conserved residues in the second transmembrane segment to identify residues involved in transport. Residues Ser260, Thr270, and Ser274appear to be important for transport and are oriented such that they would also line a face of an α-helix. In addition, the sequence276YGMNF280, found in virtually all CorA homologues, is critical for CorA function because even conservative mutations are not tolerated at these residues. Finally, mutations of residues in the second transmembrane segment, unlike those in the third transmembrane segment, revealed cooperative behavior for the influx of Mg2+. We conclude that the second transmembrane segment forms a major part of the Mg2+ pore with the third transmembrane segment of CorA.

Mg 2ϩ is the most abundant divalent cation within prokaryotes and eukaryotes, and accordingly it has many functions in the cell. It contributes to membrane stability and is a cofactor with molecules such as ATP (1,2), tRNA (3) and many other enzymes such as ribonuclease H (4,5). It is also a signaling molecule (6) for the PhoP/PhoQ sensor/signaling system required in Salmonella pathogenesis (7)(8)(9). In eukaryotic systems, Mg 2ϩ flux has been shown to be hormonally responsive (10 -14). This suggests Mg 2ϩ may also be a signaling molecule in eukaryotes (15,16). Despite these many functions, little is known about the regulation of intracellular Mg 2ϩ and the mechanism of Mg 2ϩ transport.
The CorA class of Mg 2ϩ transporters is by far the most abundant class of prokaryotic Mg 2ϩ transporter (17)(18)(19)(20). CorA homologues are found in almost all Gram-positive and Gramnegative Bacteria and in Archaea such as Archeoglobus fulgidis and Methanococcus jannaschii. There are also distantly related proteins in yeast and possibly other eukaryotes (21)(22)(23). It is likely that most of the members of this family are true homologues of the Salmonella typhimurium CorA and transport Mg 2ϩ . This is evidenced by the CorA homologue from the Archeon M. jannaschii. When expressed in S. typhimurium, this homologue appears to transport Mg 2ϩ with essentially identical properties as the S. typhimurium CorA, despite only 22% sequence identity and a considerably different native membrane environment (24).
CorA is apparently constitutively expressed and transports Mg 2ϩ with high capacity. CorA is functionally distinct from the MgtA/MgtB and MgtE classes of Mg 2ϩ transporters because it also transports Co 2ϩ and Ni 2ϩ and has been shown to efflux Mg 2ϩ at very high external Mg 2ϩ concentrations (25)(26)(27). Mg 2ϩ influx through CorA should not require energy other than the membrane potential; however, the transport mechanism is unknown. The topology of CorA is quite unusual among transporters in that it contains a very large periplasmic domain at its amino terminus and a small membrane domain at the carboxyl terminus (28). The membrane domain consists of only three TMs. 1 TM2 and TM3 are highly conserved, and only one of over forty homologues contains even a single charged residue in these two segments. TM1 contains a variable number of charged residues, none of which are conserved (21). The S. typhimurium CorA contains a single negatively charged residue in TM1 that is not required for transport (29). This lack of charge within the membrane is highly unusual for a cation transporter as most, if not all, contain charged residues in one or more membrane helices to counterbalance the charge of the ion being transported. Therefore, CorA-mediated transport of Mg 2ϩ appears quite different from known mechanisms of selective ion transport.
Previous mutagenesis studies targeting TM3 indicated that a single face of the presumed ␣-helix contained three residues important for transport and therefore appeared to form part of a "Mg 2ϩ pore". This report describes the mutagenesis of TM2. In contrast to studies on TM3, several residues within TM2 are important for transport, and their kinetic properties have led to insights of the transport process.

MATERIALS AND METHODS
All media were obtained from Difco. All other reagents were from Sigma unless otherwise specified. Oligonucleotides were purchased from Genosys (The Woodlands, TX). Supplemented N-minimal medium (N-Min) (30) contains 0.1% casamino acids and 0.4% (w/v) glucose. MgSO 4 was used when medium was supplemented with Mg 2ϩ . The concentrations of antibiotics used were 50 g/ml ampicillin, 20 g/ml chloramphenicol, and 50 g/ml kanamycin.
Mutant Construction and Expression-Mutations were created using one of two similar vectors. Mutations of corA in plasmid pRS170 (29) were made using the Altered Sites ® II kit (Promega, Madison, WI). The QuikChange TM Mutagenesis System (Stratagene, La Jolla, CA) was used to create mutations of corA in plasmid pMAS29, which is pRS170 with the ampicillin gene repaired. Mutations were verified by sequencing the 3Ј-portion of the gene, which included the entire membrane domain. The plasmids were propagated as described (29) and growth by CorA mutants were analyzed in MM281, a Mg 2ϩ transport-deficient strain (18), with Western blot analysis performed as described previously (29). Briefly, overnight cultures of the mutants were lysed by French Press, and the membranes were pelleted by ultracentrifugation at 100,000 ϫ g for 1 h. Protein quantitation was done using the Pierce BCA Assay. Membrane samples were analyzed by loading 10 g of protein on 10% SDS-polyacrylamide gel electrophoresis gels and transferring to nitrocellulose. CorA expression was visualized using an antibody directed to the 16 residues at the amino terminus (29). For those mutants exhibiting altered migration on Western blots, the entire coding region of the gene was sequenced to verify the absence of secondary mutations.
Growth in Minimal Medium-Complementation of the Mg 2ϩ transport-deficient strain MM281 was determined by streaking a single colony from an LB plate onto an N-Min plate containing 0.25% (w/v) glucose and 0.5 mM leucine and incubating at 37°C for 48 h. The growth assay can detect mutants able to take up sufficient Mg 2ϩ to grow but whose transport capacity is too low to be measured by the Ni 2ϩ transport assay. The growth assay also can detect mutations that hinder cell growth yet retain measurable amounts of cation uptake.
Growth in supplemented N-Min medium with varying concentrations of Mg 2ϩ was determined by adding 5 ϫ 10 3 cells/well in a 96-well microtiter plate, assuming 2 ϫ 10 8 cells ϭ 1 A 600 nm . The total volume in each well was 100 l. The inside plate cover was coated with Never Fog ® (North American, Atlanta, GA) to prevent condensation during the incubation. MgSO 4 concentrations used were: 0, 0.01, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 50, and 100 mM. The OD 650 nm was measured every 30 min for 16 h using a Molecular Devices THERMOmax TM plate reader held at 37°C. The Mg 2ϩ transport-deficient strain, MM281, required at least 2.5 mM Mg 2ϩ to grow. MM281 with wild type CorA on plasmid pMAS29 required 10 M Mg 2ϩ to grow in this assay.
Transport-The uptake of 63 Ni 2ϩ (NEN Life Science Products) was assayed instead of Mg 2ϩ uptake, as 28 Mg 2ϩ is prohibitively expensive and not readily available (19). Methods for transport were as described previously (29,31,32). Briefly, cells were grown overnight in LB with 100 mM MgSO 4 and appropriate antibiotics. The MgSO 4 concentration was 100 mM because the background strain MM281 requires high concentrations of Mg 2ϩ for growth. Cells were washed twice in supplemented N-Min (without Mg 2ϩ ) and resuspended in the same medium to 1-2 A 600 nm . Cells were added to tubes containing varying concentrations of inhibitor cation plus 200 M NiCl 2 and 1-1.2 Ci of 63 Ni 2ϩ in a final volume of 1 ml. The reactions were incubated for 5 min at 37°C, stopped by the addition of 5 ml of ice-cold N-Min containing 10 mM MgSO 4 and 0.5 mM EDTA, filtered immediately on nitrocellulose filters (Schleicher & Schuell), and washed once with 5 ml of the same solution. The filters were placed in a 3-ml Biosafe II scintillation mixture (Research Products International Corp., Mount Prospect, IL) and counted using a Beckman LS 6500 scintillation counter with 80% efficiency.
Transport was assayed at 200 M NiCl 2 , the approximate K m for wild type CorA (19). If the affinity of a mutant for Ni 2ϩ has decreased, the substrate concentration is no longer near the K m , and therefore the velocity attained will be a smaller fraction of the V max of the mutant. Unfortunately, the uptake values could not be corrected for affinity changes. As the Ni 2ϩ dose-response curves were irregular for many mutants as described below, the Ni 2ϩ affinity could not be determined accurately. Therefore, reported maximal uptake of each mutant (Table  I) as a percentage of wild type CorA represents a minimum value.
Inhibition of 63 Ni 2ϩ transport was determined for Ni 2ϩ , Mg 2ϩ , and Co 2ϩ . Inhibition curves are plotted as a percent of the maximal transport of the mutant, that is, the amount of 63 Ni 2ϩ uptake at 200 M NiCl 2 with no competing cation present. These plots allow a visual determination of affinity changes, assuming a single cation binding site, and are independent of changes in uptake capacity. A change in affinity for Ni 2ϩ will also alter inhibition curves for Mg 2ϩ and Co 2ϩ because they are in turn inhibiting 63 Ni 2ϩ transport. Because a mutation could affect the affinity of each cation differently, a brief delineation of possible changes is relevant. When Ni 2ϩ is used to inhibit 63 Ni 2ϩ uptake, a shift of the dose-response curve to the right would indicate a decrease in affinity for Ni 2ϩ , whereas a shift to the left would indicate an increase in affinity. For Co 2ϩ or Mg 2ϩ inhibition of 63 Ni 2ϩ uptake, the shift in the dose-response curve is dependent upon the affinities of both Ni 2ϩ and the inhibitory cation. We have never observed an increase in the affinity for Ni 2ϩ with any mutant, so only three general patterns of affinity changes are likely for the cations assayed. First, if there is no change in affinity for either Ni 2ϩ or the competing cation, there will be no shift in the dose-response curves. Second, if the affinity for Ni 2ϩ does not change and that for the competing cation decreases, the Ni 2ϩ inhibition curve remains the same and the competing inhibition curve of the cation shifts to the right. If the affinity for Ni 2ϩ remains the same and that for the competing cation increases, there will be a shift to the left in the dose-response curve of that cation. Third, if the affinity for Ni 2ϩ is decreased, there would be a shift to the right in the Ni 2ϩ doseresponse curve. If the affinity for Ni 2ϩ is decreased and the affinity for the competing cation remains the same as the wild type, the dose response curve of the competing cation will exhibit a shift to the left. However, if the affinity for the competing cation also decreases, the dose response curve of the competing cation may demonstrate a shift to the right, a slight shift to the left, or no apparent shift depending on the relative extent of the affinity change for the two cations. If the affinity for CorA of the competing cation increases, it will exhibit a significant left shift. These interpretations assume that there is a single binding site. If there is more than one cation binding site, the kinetics of transport become more complex than the interpretations described above, unless mutations affect each site similarly.
Transport K m and V max Studies-The assay used for K m and V max studies was similar to the dose-response curve transport protocol above. The uptake of Ni 2ϩ was assayed over a concentration range of 8 M to 1 mM NiCl 2 . The final concentrations of 63 Ni 2ϩ were 8 -88 Ci/mol NiCl 2 . Each Ni 2ϩ concentration was assayed in triplicate, and blanks with 10 mM MgSO 4 were subtracted for each individual Ni 2ϩ concentration.

Residue Selection and Mutant Expression-
The residues selected for mutagenesis studies were based on conserved and hydroxyl-bearing residues from the alignment shown in Fig. 1 and from additional alignments using other CorA homologues (21).
The expression of the CorA mutants was determined by Western blot analysis of isolated membranes. All mutants were expressed to a similar extent as wild type CorA (Fig. 2). The S260A, S260V, S263A, and F266A mutants exhibited slightly altered migration. These mutants were therefore completely sequenced, and none contained secondary mutations. The S260V mutant in particular migrated at a significantly lower apparent molecular mass. This suggests that the S260V mutant may be less stable than wild type CorA, possibly being proteolytically cleaved at the carboxyl terminus during membrane isolation.
Nickel Inhibition Studies: Evidence for Cooperative Kinetics-With many of the TM2 mutants assayed, Ni 2ϩ inhibition of 63 Ni 2ϩ transport curves demonstrated an increase in uptake with increasing Ni 2ϩ concentration even though the 63 Ni 2ϩ specific activity was simultaneously decreasing (Fig. 3). This phenomenon will be referred to as an "induction" of transport. This induction was also seen for some Co 2ϩ inhibition curves but to a smaller extent. Very few Mg 2ϩ inhibition curves displayed induction, and this was also to a much smaller extent, such as that shown for the F266Y mutant (Fig. 4). This difference between the three cations is likely because of their relative affinities. Ni 2ϩ has the lowest affinity at approximately 200 FIG. 1. Conserved residues in the second transmembrane segment of CorA. The sequences shown above represent a sample of CorA-like proteins based upon genomic sequence data (21). The numbering is based upon CorA from S. typhimurium and may not be correct for the other sequences shown. Conserved residues are marked in bold type.
M, whereas Co 2ϩ and Mg 2ϩ have affinities of 30 and 15 M, respectively. Therefore the affinities of Co 2ϩ and Mg 2ϩ would have to be drastically altered to see induction with the addition of micromolar quantities of Co 2ϩ or Mg 2ϩ . Conversely, the affinity of Ni 2ϩ is sufficiently low to allow changes in affinity to demonstrate the induction at micromolar levels of added Ni 2ϩ .
Induction of Ni 2ϩ uptake was displayed for all TM2 mutants except S263A and S275A. The F266Y, P269A, S274A, and S274T mutants demonstrated the largest induction, each reaching approximately 140% of maximal uptake defined at 200 M Ni 2ϩ . The peak of induction for most mutants was at approximately 300 M added NiCl 2 , corresponding to a total Ni 2ϩ concentration of 500 M. Exceptions were the S274A and S274T mutants, which had a peak of induction with 1 mM added Ni 2ϩ , corresponding to a total of 1.2 mM Ni 2ϩ . Such induction may be indicative of cooperativity as has been shown with a Na ϩ , K ϩ -ATPase mutant (33). To investigate this possibility, 63 Ni 2ϩ uptake at varying concentrations of NiCl 2 was determined for wild type CorA and one of the mutants. This experiment is different from the Ni 2ϩ inhibition curves because there was a greater range of Ni 2ϩ concentration, and the specific activity of the isotope was maintained. The P268A mutant was chosen for this assay because it maintained a significant amount of uptake, had a normal growth phenotype, yet demonstrated the Ni 2ϩ induction. The P268A mutant had a 5-fold shift in the Ni 2ϩ and a 4-fold shift for Co 2ϩ dose-response curves with no shift in the Mg 2ϩ dose-response curve. The velocity versus substrate concentration curve was sigmoidal for both wild type CorA and the P268A mutant (Fig. 5), consistent with positive cooperativity. The V max and Hill coefficient were estimated using a least square fit. The estimate of V max is 1800 pmol/A 600 nm /min for wild type and 935 pmol/A 600 nm /min for the P268A mutant. The n H values are 2.5 for the control and 2.0 for the mutant. Therefore, CorA may contain two or more binding sites for Mg 2ϩ , and the mutants alter the binding properties of these sites. When Ni 2ϩ inhibition for several alanine mutants was assayed with 500 M NiCl 2 , rather than 200 M NiCl 2 , the induction was either diminished or eliminated altogether (data not shown). Consequently, the induction effect is revealed by the decreased affinity of the mutated CorAs for Ni 2ϩ . Although the kinetics of transport are more complex than simple inhibition, the dose-response curves are nonetheless appropriate for determining relative affinity changes for the mutants, because all other data follow the wild type simple inhibition curve because the affinities of the different binding sites appear to be altered to a similar extent.
Residue Ser 260 : Mutations Affect Cation Selectivity-The S260A and S260T mutants displayed an unusual affinity profile for the cations tested, suggesting that Ser 260 plays a role in cation selectivity. Both mutants maintained significant amounts of 63 Ni 2ϩ uptake and displayed normal growth. However, both mutants also displayed a shift to the left of 4 -5-fold in the Mg 2ϩ dose-response curve as shown in Fig. 6. No other CorA mutants to date have demonstrated such a left shift. Co 2ϩ inhibition was similar to wild type, and Ni 2ϩ inhibition displayed 10-and 3-fold shifts to the right for the S260A and S260T mutants, respectively (data not shown). This is in contrast to the behavior seen with other CorA mutants, where the dose-response curves for all three cations shifted in the same direction.
The S260V mutant had no measurable transport. However, it was functional because it required only 50 M Mg 2ϩ to grow in supplemented N-Min medium, as shown in Table I. The increased Mg 2ϩ requirement could be related to the decreased stability of the S260V mutant as noted above. Alternatively, this may be because of a slight decrease in affinity for Mg 2ϩ with a severely decreased affinity for Ni 2ϩ , causing no measurable 63 Ni 2ϩ transport. The latter seems more likely because it correlates with the properties of other Ser 260 mutants. Thus, Ser 260 is likely involved in cation selectivity, but the hydroxyl moiety is not essential.
Residue Thr 270 : Size of the Residue Affects Transport-For the Thr 270 mutants constructed, as the size of the side chain increased, both transport capacity and affinity for cation also increased. The T270A mutant had no measurable transport activity. The T270C mutant had 4% of wild type uptake, a greater than 5-fold shift in the Mg 2ϩ dose-response curve (Fig.  7) and a 10-fold shift to the right in the Ni 2ϩ inhibition curve (data not shown), indicating significant decreases in affinity for both Mg 2ϩ and Ni 2ϩ . The T270S mutant, which maintains the hydroxyl moiety, had 17% of wild type uptake and demonstrated a smaller shift to the right in the Mg 2ϩ dose-response curve of about 3-4-fold (Fig. 7). The shift for Ni 2ϩ inhibition was also moderate at 3-fold (data not shown). The T270V mutant, which maintains the relative size of the residue while eliminating the hydroxyl group, maintained 35% of wild type ion uptake and had a minimal shift to the right in both Mg 2ϩ and Ni 2ϩ dose-response curves. All three mutants appeared to have the same 5-fold shift to the right for Co 2ϩ (data not shown). The growth data indicated that all Thr 270 mutants, including T270A, were functional because they all complemented MM281. The T270A mutant required 50 M Mg 2ϩ to grow, whereas the T270C, T270S, and T270V mutants required only 10 M Mg 2ϩ , the same as wild type. Thus, the T270V mutant maintained the greatest amount of CorA function of all the mutants made at this residue. Valine best approximates the size of threonine but is hydrophobic. This suggests that the size of the residue at Thr 270 is more important for transport than the hydroxyl group.
Residue Ser 274 : The Hydroxyl Moiety Is Important-All mutations at Ser 274 displayed considerably decreased cation affinity and decreased transport capacity. S274A and S274T mu- Growth on plates was the ability to grow on N-Min plates supplemented with leucine and glucose. Mutants that grew as well as wild type are designated by "ϩϩ" and mutants with slow growth are designated by "ϩ". Mutants that did not grow on plates are indicated by "-".
b The minimum concentration of Mg 2ϩ required for growth was tested in N-Min medium supplemented with casamino acids and glucose. tants maintained measurable cation transport, whereas the S274C and S274V mutants had no measurable ion uptake. The S274A mutant had only 5% of wild type transport and demonstrated an asymmetric shift to the right for the Mg 2ϩ doseresponse curve that reached about 10-fold at high Mg 2ϩ concentrations (Fig. 8). This was the only mutant that exhibited this behavior. Both Ni 2ϩ and Co 2ϩ gave a greater than 10-fold shift to the right for the S274A mutants (data not shown). The Ni 2ϩ and Co 2ϩ inhibition curves of the S274A mutant displayed an even shift throughout the curve and did not reflect the asymmetric Mg 2ϩ inhibition curve. The S274T mutant, a functionally conservative mutation, had a symmetric Mg 2ϩ shift to the right that was approximately 10-fold and maintained 17% wild type transport, the greatest amount of transport for any of the Ser 274 mutants. The Ni 2ϩ and Co 2ϩ patterns of cation inhibition were similar to the Mg 2ϩ inhibition curve for the S274T mutant, 5-fold for Ni 2ϩ and 10-fold for Co 2ϩ (data not shown). Thus all of the mutations at Ser 274 affected the transport properties of CorA, with the S274T mutant retaining the greatest amount of uptake.
Growth on minimal medium reflected the ion uptake experiment for the S274V mutant as it did not complement MM281 and required 2.5 mM Mg 2ϩ to grow. The S274C mutant allowed slow growth on minimal plates and grew as well as wild type in the growth curve assay, although uptake was undetectable by the transport assay. The S274A mutant and the S274T mutant grew as well as wild type. Therefore, all mutations at Ser 274 greatly affected transport properties. The transport data reflected an order of Ser Ͼ Thr Ͼ Ala Ͼ Ͼ Cys ϭ Val in terms of maintaining 63 Ni 2ϩ transport, whereas the order for the ability to grow in minimal medium was Ser ϭ Thr ϭ Ala Ͼ Cys Ͼ Ͼ Val. These data suggest that a hydroxyl moiety provides optimal transport and that size is not the dominant factor.
Residues 276 YGMNF 280 : Critical Structural Motif for Mg 2ϩ Transport-This five amino acid sequence has the highest conservation among the CorA homologues, effectively acting as a CorA signature sequence (21). In keeping with this high degree of conservation, this motif is critical for CorA function as even the most conservative mutations at these residues abolished or markedly diminished transport.
A Y276A CorA mutant had no measurable 63 Ni 2ϩ uptake, nor did the conservative Y276F mutant. This would imply input from the hydroxyl group of the tyrosine; however, the Y276W mutant maintained 50% ion transport capacity. The Y276W mutant also had a small shift to the right in the Mg 2ϩ dose-response curve, a 2-fold shift for Ni 2ϩ , and a 5-fold shift for Co 2ϩ (data not shown). The amount of Mg 2ϩ required for growth also varied with mutations at Y276. The Y276A mutant required 1 mM, the Y276F mutant required 250 M, and the Y276W mutant required only the wild type level of 10 M Mg 2ϩ to grow. This roughly correlates with transport, as the Y276W mutant, the only mutant with measurable transport, required the least Mg 2ϩ for growth.
Gly 277 was only mutated to alanine, as this was the only conservative mutation possible; even so, the G277A mutant demonstrated no transport of 63 Ni 2ϩ and required 2.5 mM Mg 2ϩ for growth, the amount required for the Mg 2ϩ transport-deficient strain. The G277A CorA mutant is therefore a nonfunctional protein, suggesting that any side chain at this position likely sterically interferes with a critical structure in CorA.
M278A and the conservative M278C and M278I mutants were not functional. None had measurable ion uptake, and all required 2.5 mM Mg 2ϩ for growth. Similarly, the N279A, the sterically conservative N279L, and the functionally conservative N279Q mutations were all nonfunctional in either assay. Mutations at Phe 280 also lacked measurable transport. The F280W and F280Y mutants, both conservative mutations maintaining an aromatic ring, required 1.0 mM Mg 2ϩ for growth suggesting minimal functionality. The F280A and F280R mutants required 2.5 mM Mg 2ϩ and therefore appeared completely nonfunctional.
The results from mutations in the 276 YGMNF 280 sequence clearly show that these residues are essential for CorA function. It is striking that in practically all cases even the most conservative mutations could not be tolerated. Because this sequence would comprise more than a full turn of an ␣-helix, all of these residues could not come in direct contact with a cation as it traversed the pore. Therefore it is likely that these residues form a critical structural motif within CorA.
Residues Phe 266 and Pro 269 : Unusual Growth Phenotype-For the mutants discussed above, the Mg 2ϩ requirement for growth reflected the ability of the mutants to transport cation. However, the F266A and P269A mutants exhibited significant transport but had a decreased the ability to grow in minimal medium. The P269A mutant maintained 40% of wild type ion uptake with no shift in the Mg 2ϩ (Fig. 4) and Co 2ϩ doseresponse curves and only a small shift in the Ni 2ϩ dose-response curve (data not shown). However, growth in liquid medium required 0.25 mM Mg 2ϩ . The F266A mutant also required 0.25 mM Mg 2ϩ for growth. It had approximately 47% of wild type ion uptake with no shift in the Mg 2ϩ inhibition curve (Fig.  4) and 2-and 5-fold changes in the affinity for Co 2ϩ and Ni 2ϩ (data not shown). To further investigate the phenotype of mutations at Phe 266 , an F266Y mutant was made. This mutation resulted in approximately a 3-fold shift in the Mg 2ϩ doseresponse curve (Fig. 4) with a 10-fold shift for Co 2ϩ and a 5-fold shift for Ni 2ϩ inhibition curves (data not shown). It retained 25% of wild type cation uptake. The growth phenotype, however, was similar to wild type. Therefore, the aromatic group may be important at this position. The disparity in transport and the growth phenotype seen with the F266A and P269A mutants could be because of an alteration in efflux of Mg 2ϩ , although this cannot be tested without 28 Mg 2ϩ (see "Discussion").
Residues Ser 263 and Ser 275 : Hydroxyl-bearing Residues Not Required for Transport-Ser 263 and Ser 275 were mutated to alanine with few changes in transport properties. There were no shifts in the dose-response curves for all cations tested, unlike results seen at other hydroxyl-bearing residues in TM2. The S263A and S275A mutants retained 68 and 85% wild type transport, respectively. Accordingly, the S263A and S275A mutants both had growth phenotypes similar to wild type CorA. FIG. 8. Mg 2؉ inhibition of 63 Ni 2؉ uptake for the S274A and S274T mutants. Transport was performed as described under "Materials and Methods." Curves are normalized to the maximal uptake of each mutant. These data are the average of a minimum of three separate uptake experiments.

DISCUSSION
Our working hypothesis is that CorA functions as a homooligomer with residues along the ␣-helical TM2 and TM3 segments, including many of the hydroxyl-bearing residues, coordinating Mg 2ϩ as it passes through a pore or channel. Preliminary data indicate that CorA is a pentamer, 2 and we have previously shown that three residues along a single face of the presumed TM3 ␣-helix appear to be required for Mg 2ϩ transport. Herein, we continued the mutagenesis of the membrane domain targeting highly conserved and hydroxyl-bearing residues in TM2 basing our selection upon alignments of homologues such as those shown in Fig. 1. We first used alaninescanning mutagenesis and then, for those residues that showed a significant effect, changed the residues to more conserved amino acids in size or functional moiety. Western analysis revealed that all mutants were expressed to a similar extent as CorA except S260V, which may not have been stable during membrane isolation. Functionality was determined by 63 Ni 2ϩ uptake and its inhibition by Mg 2ϩ , Co 2ϩ , and Ni 2ϩ . Phenotype was assayed by complementation of the Mg 2ϩ transport-deficient strain MM281 and by growth assays. We have determined that in TM2, conserved residues 276 YGMNF 280 and the hydroxyl-bearing residues Ser 260 , Thr 270 , and Ser 274 are important for transport through the S. typhimurium CorA. These latter residues may line part of a pore of an oligomeric CorA in conjunction with TM3. Furthermore, several mutations in TM2 alter the affinity for Ni 2ϩ and allow an induction to be seen in the dose-response curves revealing cooperative behavior.
Cooperativity-The Ni 2ϩ induction seen with several residues in TM2 is an unexpected finding of this study because no similar transport phenomenon was seen with any mutation in TM3 (29). Such substrate stimulation has been seen previously in a mutant of the Na ϩ , K ϩ -ATPase (33), with the conclusion that it is because of positive cooperativity. Evidence for positive cooperativity is apparent in the sigmoidal uptake curve (Fig. 5), and accordingly, the Hill coefficient is greater than two. This cooperativity may be because of either binding or transport (33). Cooperative binding refers to the ability of an initial bound substrate molecule to facilitate the binding of subsequent substrate molecules. Cooperative transport refers to an increased rate of transport when all binding sites are filled. Possibly, two or more binding sites in CorA act in a cooperative manner. An alternative interpretation is that CorA is a channel in which transport occurs in a cooperative manner, with two or more binding sites. It is very difficult with this type of assay to distinguish which mechanism is responsible for this effect in CorA.
Growth Phenotype-The growth phenotype generally reflected the ability to transport Ni 2ϩ . However, for certain mutants, cell growth did not correspond with transport. The S260V, T270A, S274C, Y276A, Y276F, F280W, and F280Y CorA mutants had no measurable transport but required less Mg 2ϩ to grow than MM281, the transport-deficient strain. These mutants apparently permit sufficient Mg 2ϩ entry to sustain cell growth but do not have sufficient transport to allow measurement under the conditions assayed, the mutations decreasing Ni 2ϩ uptake to a level below the sensitivity of the assay. The most likely explanation for this difference is a markedly decreased maximal transport capacity, with possible contribution from decreased affinity. Conversely, but less likely, this phenotype could be because of a highly increased affinity for Ni 2ϩ in which the cation binds too tightly for transport to occur.
The opposite growth phenotype occurred with the F266A, F266Y, P269A, and Y276W mutants. These mutants maintained at least 25% total uptake compared with wild type CorA and had minimal changes in affinity for Mg 2ϩ . Yet these mutants required more Mg 2ϩ for growth than other mutants with significantly less ion transport capacity. We interpret this phenotype as most likely because of inappropriate Mg 2ϩ efflux. CorA mediates efflux of Mg 2ϩ (25), but efflux is "gated" by Mg 2ϩ and does not occur with wild type CorA except at high extracellular Mg 2ϩ concentrations, greater than or equal to 1 mM (26). If a mutation altered this gating mechanism to constitutively activate efflux or simply allowed leakage of Mg 2ϩ , Mg 2ϩ efflux would occur at low extracellular concentrations. If this loss of Mg 2ϩ were of sufficient magnitude, growth could be compromised. However, the unavailable isotope 28 Mg 2ϩ is required for efflux studies, as CorA does not mediate Ni 2ϩ or Co 2ϩ efflux (26), possibly because of high affinity binding of these metals within the cell. Therefore this supposition cannot currently be tested directly.
Hydroxyl-bearing Residues-Many of the hydroxyl-bearing residues in TM2 are not essential for Mg 2ϩ transport. The S263A and S275A mutants had significant uptake capacity and no apparent change in cation affinity. In addition, whereas Thr 270 is important for transport, the size of the side chain rather than the hydroxyl group is required for CorA function. Tyr 276 also plays a like role in cation transport. The Y276F mutant was not functional, which might suggest that the hydroxyl moiety is important. However, the Y276W mutant maintained a significant amount of activity with only a slight shift in affinity, suggesting that a hydroxyl moiety is not critical. The requirement here may be sufficient delocalization of charge within the aromatic ring. We conclude that the hydroxyl moiety itself at Ser 263 , Ser 275 , Thr 270 , and Tyr 276 does not play a role in transport, even though the latter two residues have important roles in transport.
In contrast, the hydroxyl moieties of Ser 260 and Ser 274 appear to play significant roles in Mg 2ϩ transport. Alanine mutations at either position had drastic phenotypes. Mutations even to relatively conserved residues had major effects. Alanine and threonine mutations at Ser 260 caused the affinity for Co 2ϩ and Ni 2ϩ to decrease significantly compared with the Mg 2ϩ affinity, whereas the valine mutant did not exhibit transport. This difference in cation affinities suggests that Ser 260 plays a role in cation selectivity with the hydroxyl moiety conferring optimal activity. Although the alanine mutant did have significant uptake, as did the threonine mutant, the alanine might simply be too small to hinder transport. Conversely, the valine mutant, which is similar in size to threonine, did not retain ion uptake in the transport assay. The valine may be too large for a nonhydrophilic side chain at that position.
Ser 274 is similar to Ser 260 in that the alanine and threonine mutants maintained transport but the valine and cysteine mutants did not. The threonine mutant did not have such drastic effects as the alanine mutant but did have a decrease in affinity for Mg 2ϩ . Therefore the Ser 274 hydroxyl may be required for optimal transport in a manner similar to the Ser 260 hydroxyl moiety. The alanine mutant may be too small to eliminate transport, whereas the valine mutant is too large to allow transport to occur. In addition, the asymmetry of the shift in the S274A mutant Mg 2ϩ inhibition curve may indicate that CorA contains multiple cation binding sites, such as the Na ϩ , K ϩ -ATPase, SR Ca 2ϩ -ATPase (34), and the KcsA ion channel (35). The alanine mutant may have altered the Mg 2ϩ affinity of only one of the binding sites. The presence of multiple binding sites is compatible with the cooperative behavior derived from the kinetic analysis above and suggests cooperative binding may occur, although other mechanisms cannot be ruled out.
The CorA Signature Sequence-The 276 YGMNF 280 sequence is the most highly conserved sequence in all CorA homologues. As might be expected, all of the residues in this sequence are required for transport; even conservative mutations completely eliminated cation transport. The only conservative mutation of the 276 YGMNF 280 sequence that retained activity in the ion uptake assay was the Y276W mutant, although the even more conservative Y276F mutant was completely nonfunctional. Similarly, aromatic substitutions at Phe 280 are functional, but only minimally. Thus, at Tyr 276 and Phe 280 , a bulky aromatic group is apparently necessary. In the center of this motif, the G277A mutant and functionally conservative mutations at Met 278 and Asn 279 were all nonfunctional, emphasizing the importance of this short sequence.
What then is the role of this motif? If TM2 is entirely an ␣-helix within the membrane, the 276 YGMNF 280 sequence would form more than a complete turn around the helix. Thus all of these residues could not face the pore and physically interact with Mg 2ϩ . Perhaps this sequence positions TM2 and TM3 relative to each other because they both have residues required for transport. There is a relatively short span of 7 or 8 residues between the two helices, and the size of this span is maintained in the large majority of CorA homologues (21). This span could act with the 276 YGMNF 280 sequence to form a structure that positions the membrane segments. This would in turn imply that the entire short loop between TM2 and TM3 is important, and preliminary mutagenesis results 3 indicate that several residues of this loop are crucial to CorA function.
Taken together with the previous mutagenesis studies of TM3, residues in both TM2 and TM3 appear to interact with Mg 2ϩ , and therefore it would appear that both TM2 and TM3 segments line a Mg 2ϩ pore. It is interesting that residues from the two TM segments that have similar effects appear to align (Fig. 9). That is, if the helices are in approximate register, residues having similar effects on cation affinity or selectivity could be at the same level or depth within the membrane. For example, the two residues that affect cation selectivity, Ser 260 and Tyr 307 , could be near each other. Together they might act as a gate or filter for Mg 2ϩ . Likewise, Thr 270 and Met 299 , and Ser 274 and Tyr 292 also would align if Ser 260 and Tyr 307 were aligned. Each of these latter pairs of residues exhibited comparable changes in cation affinity and capacity when replaced by conservative residues.
It is curious that some effects seen in TM2 are not seen with mutations in TM3, such as the induction of Ni 2ϩ transport. This implies that TM2 has a markedly different role than TM3. The structure of both the homo-tetrameric KcsA and homopentameric MscL ion channels shows a single TM segment lying within the channel and a second distinct TM segment behind it at an angle (35,36). This arrangement exposes a limited number of residues of the second peripheral TM segment to the channel. Our current data could fit these ion channel models, because we have preliminary evidence that CorA is a pentamer. How the TM segments that actually line the pore cannot, of course, be determined without definitive structural data. Regardless of the actual membrane domain structure of CorA, the results of our mutagenesis of TM2 coupled with those for TM3 indicate that three residues along a single face of each presumed ␣-helix have primary effects on Mg 2ϩ influx. In addition, the data indicate that other residues within the membrane domain affect transport, presumably by altering positioning of the other six more prominent residues. Finally, the highly conserved 276 YGMNF 280 sequence probably provides a crucial structural role within CorA necessary for transport.