The CorA Mg²⁺ Transporter Does Not Transport Fe²⁺

Bacterial strains, plasmids, and growth conditions.Bacterial strains and plasmids used or constructed in this study are listed in Table 1. Unless indicated otherwise, all bacterial strains are derived from S. enterica serovar Typhimurium SL1344. The molecular biology methods used were those of Sambrook et al. (33). The high-frequency generalized transducing bacteriophage P22 mutant HT 105/1, int-201 was used for all transductional crosses, and phage-free, phage-sensitive transductants were purified as described previously (27). The corA mutant was constructed by P22 transduction of a corA::Tn10 insertion cassette from MM385 into MM2089 as a recipient, generating MM2242. The insertional mutation of corA in the SL1344 background was further confirmed by a Co²⁺ resistance assay (17), Western blot analysis, and PCR. To create a corA feoB strain, the feoB gene of MM2089 was deleted using the technique of Datsenko and Wanner (7) to give MM2985. The deletion was confirmed by PCR. The corA::Tn10 insertion allele was subsequently introduced by P22 transduction, giving MM2986. Bacteria were routinely grown in Luria-Bertani (LB) broth at 37°C with shaking and with antibiotic supplementation when required as described previously (16, 46).

Iron toxicity assays.The liquid iron killing assays were adapted from the methods of Chamnongpol and Groisman (4) and Hantke (15), and two different protocols were used that were based on the adaptation. For protocol 1, strains were grown overnight at 37°C with shaking in 2 ml of N-minimal medium [13.14 g of Tris HCl, 0.196 g of Tris Base, 0.136 g of KH₂PO₄, 0.372 g of KCl, 1.0 g of (NH₄)₂SO₄, 1.0 liters of double-distilled water] supplemented with 100 mM MgCl₂-0.1% Casamino Acids-0.4% glucose (pH 7.0). Overnight cultures were washed twice in 1 ml of N-minimal medium containing 10 μM MgCl₂ (pH 7.0), resuspended in 200 μl of the same medium, and used to inoculate 10 ml of N-minimal medium with 10 μM MgCl₂-0.1% Casamino Acids-0.4% glucose (pH 5.8 or 7.0) in 125-ml flasks. Cells were incubated at 37°C with shaking for 6 to 7 h. For toxicity assays, a 0.1 optical density (OD) unit of cells was removed and washed three times in 1 ml of N-minimal medium containing 10 μM MgCl₂ (pH 5.8 or pH 7.0). Cells were resuspended in 1 ml of N-minimal medium with 10 μM MgCl₂ (pH 5.8 or 7.0). The OD at 600 nm (OD₆₀₀) was read, and cells were adjusted to the same OD and diluted, initially at 1:100 and then twice at 1:1,000. A total of 100 μl of each diluted cell sample was mixed with 100 μl of 0, 10, 100, 250, or 500 μM FeCl₂ (freshly dissolved in N-minimal medium containing 10 μM MgCl₂ [pH 5.8] preincubated at 37°C) and incubated at 37°C with shaking for 15 min. Cells were immediately plated on N-minimal plates containing histidine (strain SL1344 is a histidine auxotroph) and 0.4% glucose or on LB and incubated overnight at 37°C, and colonies were counted. All experiments were performed in duplicate. In addition to the experiments described, other experiments were conducted in medium lacking carbon and nitrogen sources and in liquid medium containing 1 mM sodium ascorbate (data not shown). In all cases, although Fe²⁺ clearly killed cells, no differences were seen between wild-type and corA cells (data not shown).

For protocol 2, strains were grown overnight at 37°C with shaking in tryptone yeast extract (TY) medium (8.0 g of tryptone, 5.0 g of yeast extract, and 5.0 g of NaCl per liter). Overnight cultures were centrifuged, and the cell pellets were resuspended in a HEPES-salts buffer (30.0 g of HEPES, 4.65 g of NaCl, 1.5 g of KCl, 1.0 g of NH₄Cl, and 0.425 g of Na₂SO4 per liter at pH 7.2) supplemented with 0.2% glucose and 10 μM MgSO₄. Then ∼10⁹ cells were incubated with 30 μM FeSO₄ for 0, 5, 20, 40, and 60 min. At each time point, cells were diluted and plated on TY plates and incubated overnight at 37°C, and colonies were counted. All assays were conducted in the presence of 1 mM sodium ascorbate and in duplicate. For plate iron killing assays, a lawn of cells was plated on LB plates. A 6-mm-diameter filter paper disk in the center of the dish received 15 μl of freshly made 1 mM FeCl₂ with 1 mM sodium ascorbate. After overnight incubation, the diameter of the ring around the disk devoid of cells was evaluated (data not shown).

Transport assays.The uptake of ⁶³Ni²⁺ (NEN, Boston, Mass.) was assayed instead of Mg²⁺ uptake, as ²⁸Mg²⁺ is prohibitively expensive and not readily available. Methods for transport have been described in detail previously (13, 40, 42). Briefly, cells were grown aerobically overnight in LB with appropriate antibiotics, washed twice in cold cell wash buffer (N-minimal medium with 0.4% glucose and 0.2% Casamino Acids at either pH 5.8 or 7.0), resuspended to an OD₆₀₀ between 1.0 and 2.0, and placed on ice. Cells were added to tubes containing various concentrations of inhibitor cation plus 200 μM NiCl₂ and 0.3 to 1 μCi ⁶³Ni²⁺ in a final volume of 1 ml. The reaction mixtures were incubated for 15 min at 37°C, stopped by the addition of 5 ml of ice-cold N-minimal medium containing 10 mM MgSO₄ and 0.5 mM EDTA, filtered immediately on nitrocellulose filters (Schleicher & Schuell, Keene, N.H.) and washed once with 5 ml of the same solution. The filters were placed in 3 ml of Biosafe II scintillation cocktail (Research Products International Corp, Mount Prospect, Ill.), and radioactivity was measured by scintillation counting.

All solutions containing iron were made fresh immediately before use. Fe²⁺ solutions always contained 1 mM ascorbate unless noted, and ⁵⁵Fe²⁺ transport assays were always conducted in solutions containing 1 mM freshly made sodium ascorbate. The final Fe²⁺ concentration was 1 μM when ⁵⁵Fe²⁺ transport was measured, but the transport assay was otherwise similar to that for ⁶³Ni²⁺ transport assays. When Fe³⁺ was tested, ascorbate was not used to make stock solutions and no ascorbate was added to the incubation or wash buffers. In all cases, the concentration of sodium ascorbate used was chosen based on methods previously described; most likely some sodium ascorbate was consumed during the experiment. However, throughout the duration of all experiments up to 12 h postexperiment, oxidation of ferrous iron or precipitation of ferric iron in stock solutions was not observed.

While our transport assays were very similar to those used by Hantke (15), there were some differences. Hantke analyzed S. enterica serovar Typhimurium LT2, whereas we studied S. enterica serovar Typhimurium SL1344 and 14028s. However, CorA-mediated transport was initially characterized by this laboratory using S. enterica serovar Typhimurium LT2 as a strain background (13, 16, 17). corA expression and CorA-mediated transport results appear to be identical for the LT2, 14028s and SL1344 strain backgrounds. corA is a single gene locus, and both the promoter and coding sequences are identical for the different S. enterica serovar Typhimurium strains. In addition, Hantke grew cells overnight in TY medium and resuspended in a medium buffered with HEPES and containing NaCl, KCl, NH₄Cl, and Na₂SO₄ and no added magnesium. In our hands and with our methods, cells can be grown overnight in LB or N-minimal medium with or without various monovalent cations and resuspended in several buffers containing any monovalent cation without altering transport properties (13). Thus, these minor differences should not alter interpretation of the transport data.

Effect of mutation of corA on survival after exposure to iron.During encounter with host cells, S. enterica serovar Typhimurium encounters significant stress due to exposure to increased reactive oxygen and nitrogen species (9, 18). In addition, stress due to iron limitation is well known to affect virulence. A reported phenotype of a corA strain is its lack of sensitivity to Fe²⁺ toxicity (15). Therefore, the survival rates of wild-type (strains SL1344, LT2, and 14028s) and corA and phoP strains were determined by challenge with different concentrations of Fe²⁺ in liquid medium and protocols adapted both from Hantke and from Chamnongpol and Groisman (Fig. 1 and Table 2). A marked increase in Fe²⁺ sensitivity in three different phoP strains, as had been reported by Chamnongpol and Groisman (4), was observed; however, altered sensitivity to iron as a function of corA allele status, as reported by Hantke (15), could not be demonstrated (Fig. 1 and Table 2). Only a slight effect of a 20 to 30% difference in survival could be observed for wild-type compared to corA cells. A similar lack of effect of a corA mutation on iron sensitivity was also noted by Chamnongpol and Groisman (4). Iron toxicity was independent of S. enterica serovar Typhimurium strain background, whether LT2, SL1344, or 14028s (Table 2). Similar results were also obtained from plate diffusion assays used to detect iron toxicity (data not shown). Reported changes in Fe²⁺ sensitivity were attributed to an ability of the CorA transporter to mediate influx of Fe²⁺. However, these interpretations were inferred from indirect assays rather than from direct demonstration of Fe²⁺ interaction with CorA. We therefore examined the effect of expression of CorA on Fe²⁺ accumulation.

• Open in new tab
• Download powerpoint

FIG. 1.

Fe²⁺ toxicity in S. enterica serovar Typhimurium. Iron toxicity was measured in wild-type SL1344, LT2, and SL1344 corA strains grown overnight in TY medium and exposed to 30 μM Fe²⁺ in 1 mM ascorbate as described in Materials and Methods. A 20 to 30% decrease in survival occurred for both wild-type strains, whereas the corA strain showed no decreased survival. Data represent the averages of the results of two independent experiments performed in duplicate.

CorA and iron accumulation.The only characterized primary Fe²⁺ uptake system in S. enterica serovar Typhimurium, E. coli, and Helicobacter pylori is FeoB (14, 19, 49). Although other transport systems such as MntH and SitABCD transport Fe²⁺ when it is present at very high concentrations, Fe²⁺ is not their primary substrate cation. We therefore measured the ability of wild-type, corA, feoB, and corA feoB strains to accumulate iron.

If CorA provided any significant amount of Fe²⁺ uptake, its mutation should have diminished total Fe²⁺ accumulation after growth on typical laboratory medium. While iron uptake results differed between experiments, the average amount (n = 11) of ⁵⁵Fe²⁺ accumulation did not differ significantly between a wild-type versus a corA strain or between a wild-type and an feoB strain (Fig. 2). While Hantke observed no difference in Fe²⁺ uptake between wild-type and feoB strains, he did report a difference between wild-type and corA strains. We suggest that this difference in results can be attributed to the very high variability in iron accumulation from experiment to experiment, as indicated by the error bars in Fig. 2. We examined several parameters to reduce variability in an attempt to determine whether there was a real difference in iron accumulation, including type of buffer, presence or absence of Mg²⁺, iron concentration, use of EDTA in wash buffers, and time of incubation, but could never detect consistent differences or markedly reduce variability, most likely because iron binds tightly to many biological molecules and cannot be readily removed without more severe (and likely deleterious) means. Similar issues arise in measurement of ⁴⁵Ca²⁺ uptake, as we have previously noted (13). Regardless, under the growth and assay conditions used, the presence or absence of CorA had no effect on Fe²⁺ accumulation, an indication that CorA cannot mediate the uptake of iron. Interestingly, by an unknown mechanism, the corA feoB strain accumulated more Fe²⁺ than the wild type or the single mutants (Fig. 2).

• Open in new tab
• Download powerpoint

FIG. 2.

Fe²⁺ uptake in S. enterica serovar Typhimurium. Iron accumulation was measured in wild-type, corA, feoB, and corA feoB strains after 20 min of incubation using 1 μM ⁵⁵Fe²⁺ with no added Mg²⁺ as described in Materials and Methods. Ascorbate was present at 1 mM throughout. Preliminary experiments had shown that maximal intracellular ⁵⁵Fe²⁺ was reached after 15 to 20 min of incubation. The data represent the averages of the results of 11 experiments. There was no significant difference in uptake seen with any strain comparisons except for the corA feoB strain, which exhibited about a 50% increase in total accumulation compared to the other three strains (paired t test; P < 0.01).

Effect of iron on CorA-mediated transport.We next determined the ability of Fe²⁺ and Fe³⁺ to interact with CorA. If CorA is capable of transporting iron, then iron must inhibit the uptake of other CorA substrates. ⁶³Ni²⁺ has been used as a surrogate due to the high cost and poor availability of ²⁸Mg²⁺ (13, 17, 41). The apparent K_0.5 of Mg²⁺ for CorA is 10 to 20 μM, while that of Ni²⁺ is 200 μM. If the total [⁶³Ni²⁺] in the assay solution is equal to the K_0.5, then the concentration of a cation required to inhibit 50% of control ⁶³Ni²⁺ uptake (IC₅₀) is exactly double the true K_i for that cation by simple Mass Action Law consideration. The ability of Mg²⁺ with and without 1 mM ascorbate, Fe²⁺ with 1 mM ascorbate, and Fe³⁺ to inhibit ⁶³Ni²⁺ uptake was measured for MM2089, a wild-type SL1344 S. enterica serovar Typhimurium strain. Although this strain carries functional mgtA and mgtCB alleles in addition to corA, corA is constitutively expressed, and cells for assay were grown overnight in LB, which contains sufficient Mg²⁺ to completely repress transcription of mgtA and mgtCB (37, 47).

Neither Fe²⁺ nor Fe³⁺ could inhibit ⁶³Ni²⁺ uptake via CorA at pH 7.0. At concentrations up to 20 μM Fe³⁺ (data not shown) or 100 μM Fe²⁺ (Fig. 3), neither iron species had any effect on ⁶³Ni²⁺ uptake. At the assay pH of 7.0, higher concentrations of Fe²⁺ and Fe³⁺ precipitated during filtering; they did not precipitate in stock solutions (Fig. 3). Thus, it could not be determined whether very high concentrations of iron could inhibit CorA at pH 7.0. Nonetheless, biological systems are highly unlikely to have free iron concentrations of this magnitude; therefore, for practical purposes, iron is not an effective inhibitor of the CorA Mg²⁺ transporter. We also performed the same experiment at pH 5.8, since S. enterica serovar Typhimurium can encounter acid conditions in some growth situations. Acid pH increases the apparent V_max of CorA about twofold (40, 46) and slightly decreases the K_0.5 for Ni²⁺ to about 100 μM. Again, neither Fe²⁺ nor Fe³⁺ had any effect on uptake via CorA at concentrations of up to 300 μM Fe²⁺ (Fig. 3) or Fe³⁺ (data not shown). Since we have previously shown that S. enterica serovar Typhimurium LT2, 14028s, and SL1344 strains have identical CorA transport properties and since coding and promoter sequences are identical for all three strains, these results are strain independent.

• Open in new tab
• Download powerpoint

FIG. 3.

Effect of iron on CorA-mediated transport. Using a final Ni²⁺ concentration of 200 μM, ⁶³Ni²⁺ uptake was measured in wild-type cells (MM2089) as previously described (13, 41). The effect of Fe²⁺ with 1 mM ascorbate at pH 7 (▪) and Fe²⁺ at pH 5.8 (□) is shown. Inhibition by Mg²⁺ with (•) and without (○) 1 mM ascorbate is shown for comparison. Iron solutions were made fresh immediately before each experiment. The experiment shown is representative of three additional experiments. For each of the iron dose response curves, iron precipitated at concentrations higher than level represented by the last point shown. Experiments using Fe³⁺ also showed no inhibition (data not shown).

Effect of Mg²⁺, Co(III)hexaammine, and CorA on ⁵⁵Fe²⁺ uptake.Mg²⁺ inhibited uptake of ⁵⁵Fe²⁺ (1 μM) in wild-type cells with a K_0.5 of 50 to 100 μM, a result severalfold higher than the K_0.5 of Mg²⁺ for CorA. However, Mg²⁺ inhibition of ⁵⁵Fe²⁺ uptake was identical in a corA strain (Fig. 4), an indication that the ability of Mg²⁺ to inhibit ⁵⁵Fe²⁺ uptake was not due to interaction with the CorA transporter. The ability of Mg²⁺ to inhibit ⁵⁵Fe²⁺ uptake presumably reflects a low-affinity interaction of Mg²⁺ with one or more iron transporters. Finally, Co(III)hexaammine is a selective inhibitor of CorA, with a K_i of 1 μM. However, its half-maximal inhibitory concentration for ⁵⁵Fe²⁺ uptake was around 1 mM, about 3 logs higher than its K_i for CorA (Fig. 4). These data clearly indicate that binding of these compounds to CorA is not involved in any effect they may have on Fe²⁺ toxicity.

• Open in new tab
• Download powerpoint

FIG. 4.

Effect of Mg²⁺ on ⁵⁵Fe²⁺ uptake. The ability of Mg²⁺ (•) and Co(III)hexaammine (▪) in a wild-type strain (MM2089) and of Mg²⁺ (○) in a corA strain (MM2242) to inhibit ⁵⁵Fe²⁺ uptake was measured in the same manner as for ⁶³Ni²⁺ uptake (Materials and Methods). The final ⁵⁵Fe²⁺ concentration was 1 μM. Ascorbate was present at 1 mM. The Mg²⁺ data shown represent the averages of the results of at least four independent experiments. The average initial uptake with no Mg²⁺ present was 1.03 nmol of ⁵⁵Fe²⁺/OD₆₀₀. For clarity, standard error bars are not shown but were generally slightly larger than the size of the symbols. For Co(III)hexaammine, the data are from a single experiment representative of one additional experiment.

Chemistry of Fe²⁺ and Mg²⁺.A cation that is a substrate for a transport system must in turn inhibit the transport of any other substrate cation. However, from the viewpoint of the respective chemical properties of Fe²⁺ and Mg²⁺, it is highly unlikely that Fe²⁺ would be transported by a system capable of selectively interacting with and transporting Mg²⁺. Mg²⁺ has a far larger hydration shell than does Fe²⁺. Mg²⁺ is a very hard Lewis acid, whereas Fe²⁺ is an intermediate-to-soft Lewis acid. While Mg²⁺ readily interacts with carboxylate and phosphate anions in solution or on proteins (1, 26), its transport by CorA (40, 46) or by MgtB (D.G. Kehres and M. E. Maguire, unpublished observations) does not require electrostatic interactions, in apparent contrast to transport of iron (3, 24, 34, 52). Mg²⁺ is invariably hexacoordinate, with resulting Mg²⁺-ligand bond angles of 90 ± 5° in the octahedron (6). In contrast, Fe²⁺ can be tetra- or hexacoordinate, with far more flexibility in bond angles (6). Iron prefers in many cases to bind to sulfur-containing ligands, whereas Mg²⁺ binds almost exclusively to oxygen-containing ligands (6, 26). Finally, although Mg²⁺ does not possess d orbital electrons, the other two known substrates of CorA, Co²⁺ and Ni²⁺, are high-spin d⁷ and d⁸ coordinate ligands whereas Fe²⁺ is usually a low spin d⁶ ligand in biological complexes (6, 8, 26). Overall, this combination of properties makes Fe²⁺ a very unlikely substrate for CorA. We note, however, that this does not preclude simple inhibition of transport by a nonsubstrate cation, since an inhibitory cation need only interact with a part of the transport protein.

Our results differ from those of Hantke (15), who reported that mutation of corA made E. coli and S. enterica serovar Typhimurium relatively resistant to Fe²⁺-mediated killing. Neither we (this report) nor Chamnongpol and Groisman (4) observed any difference in Fe²⁺ toxicity between a wild-type and a corA strain of S. enterica serovar Typhimurium. We have no explanation for this discrepancy. We (this report) and previously Chamnongpol and Groisman (4) also observed a difference in iron toxicity between phoP and wild-type strains. (Those authors also reported that introduction of a corA mutation into a phoP strain restored wild-type Fe²⁺ resistance, a phenotype we did not investigate.) However, an interpretation from such experiments that corA cells were transporting iron because they were resistant to iron toxicity is premature without direct transport assays. Furthermore, our data do not show that a corA strain accumulates less iron than a wild-type strain, again differing from Hantke's results. We attribute this latter difference to the highly variable nature of iron transport (see error bars in Fig. 2). During preliminary experiments, transport assay conditions were adjusted considerably in an effort to minimize such variability with ⁵⁵Fe²⁺, with incomplete success. Subsequently, of 11 independent experiments in which we evaluated ⁵⁵Fe²⁺ uptake, a corA strain accumulated slightly more iron than wild-type in 5 experiments, slightly less iron in 4 experiments, and an approximately equivalent amount in 2 experiments. Thus, it would be possible to observe a decrease (or an increase) in iron uptake in a corA strain if only a few experiments were performed. These results are in contrast to those comparing wild-type or corA strains with a corA feoB strain, for which 10 or 11 experiments showed a marked increase in iron uptake in the corA feoB strain compared to the results seen with the others.

Regardless of these accumulation experiments, the K_0.5 of Mg²⁺ inhibition of ⁵⁵Fe²⁺ uptake is severalfold greater than would be expected if Mg²⁺ were interacting with CorA, the dose response curve for Mg²⁺ inhibition of ⁵⁵Fe²⁺ uptake is independent of the presence of the CorA transporter, and the ability of the CorA-selective inhibitor Co(III)hexaammine to inhibit ⁵⁵Fe²⁺ uptake does not correlate with its K_i for inhibition of CorA (Fig. 4). Thus, even with the experimental limitations noted, mutation of corA has no significant effect on ⁵⁵Fe²⁺ uptake. These conclusions are emphasized by the demonstration, by direct transport assay, that Fe²⁺ does not affect CorA-mediated transport of a known substrate, ⁶³Ni²⁺ (Fig. 3). One substrate of a transporter must inhibit flux of another substrate. Fe²⁺ does not inhibit CorA-mediated transport of ⁶³Ni²⁺ even at high nonphysiological iron concentrations (Fig. 3). Thus, we conclude that CorA cannot transport Fe²⁺.

Despite the reasonable presumptions that CorA might be considered a “housekeeping” system and, further, that its mutation leads to no significant Mg²⁺-dependent growth phenotype, loss of corA gives a surprising array of phenotypes. These include altered transcription of PhoPQ-regulated genes, altered expression of genes encoded by Salmonella pathogenicity island I, increased sensitivity to heat shock and peroxide, and diminished virulence (J. Lin, D. G. Kehres, and M. E. Maguire, unpublished data). Thus, while it is possible that loss of corA may be a cause of one or more phenotypes related to iron (4, 15), the association is indirect and does not arise from an ability of CorA to transport iron.