The Periplasmic Loop Provides Stability to the Open State of the CorA Magnesium Channel*

Background: Members of the CorA/Mrs2 family mediate Mg2+ uptake in prokaryotic cells and mitochondria. Results: Mutations in the periplasmic loop disrupt the integrity of the CorA pentamer in low [Mg2+] but not high [Mg2+]. Conclusion: The CorA pentamer is stabilized differently in the presence and absence of Mg2+. Significance: The periplasmic loop stabilizes the open state of the channel. Crystal structures of the CorA Mg2+ channel have suggested that metal binding in the cytoplasmic domain stabilizes the pentamer in a closed conformation. The open “metal free” state of the channel is, however, still structurally uncharacterized. Here, we have attempted to map conformational states of CorA from Thermotoga maritima by determining which residues support the pentameric structure in the presence or absence of Mg2+. We find that when Mg2+ is present, the pentamer is stabilized by the putative gating sites (M1/M2) in the cytoplasmic domain. Strikingly however, we find that the conserved and functionally important periplasmic loop is vital for the integrity of the pentamer when Mg2+ is absent from the M1/M2 sites. Thus, although the periplasmic loops were largely disordered in the x-ray structures of the closed channel, our data suggests a prominent role for the loops in stabilizing the open conformation of the CorA channels.

Ion transport in a cell is mediated by channels and transporters that are embedded within membranes. These proteins alternate between different conformations in response to changes in substrate concentration or other regulators in the intra-/extracellular milieu (1,2). Unfortunately, it has proven difficult to capture and study different conformational states of these proteins, particularly by x-ray crystallography. An illustrative example is the CorA/Mrs2 family of cation channels, which are responsible for magnesium (Mg 2ϩ ) uptake in prokaryotic cells and mitochondria (3)(4)(5). Three separate x-ray structures of CorA from Thermotoga maritima (TmCorA) 2 have been solved (6 -8), but they are all in a single closed conformation, and many questions about the uptake mechanism remain unanswered.
The structures of TmCorA revealed that it is a homopentamer with a centrally located ion conduction pathway and a large soluble domain in the cytoplasm (Fig. 1). The soluble domain contains two Mg 2ϩ binding sites at the interface between each protomer (termed M1 and M2), which appear to stabilize the pentamer in a closed conformation by negating static repulsion ( Fig. 1) (6 -8). Subsequent biochemical characterization indicated that these sites act as "homeostasis sensors" that gate the opening and closing of the ion-conducting pore in response to changes in intracellular Mg 2ϩ (9). An additional Mg 2ϩ binding site was identified at the cytoplasmic pore exit and was termed the aspartate ring (Fig. 1). Unfortunately, no x-ray structure of a metal-free protein has been captured. Thus, our understanding of the role of the cytoplasmic metal-binding sites in mediating conformational transitions is limited.
Another issue that cannot be answered by the x-ray structures is the role of the periplasmic loop. This loop contains a conserved motif (YGMNFXXMPEL) that is required for Mg 2ϩ uptake (10,11). As the five loops in the pentamer constitute the periplasmic entrance to the pore, it was presumed that they form the ion selectivity filter. However, in the x-ray structures of the closed channel, the loops were largely disordered and, thus, poorly resolved. A subsequent study employing electron paramagnetic resonance spectroscopy and molecular dynamics simulations have helped generate a model for the conserved periplasmic loop in the closed conformation (12), but the structure of the loop in the open state is still unclear. Curiously, two conserved residues in the loop are important for the pentameric structure (9). This latter observation caught our interest, as it cannot be easily explained by the x-ray structures of the closed channel, where considerable structural stability appears to be mediated by metal coordination to the M1/M2 sites (6 -9).
In this study, we have investigated the role of the periplasmic loop in stabilizing the TmCorA pentamer. Our data reveal that when Mg 2ϩ is absent from the M1/M2 sites, conserved residues in the periplasmic loop are essential for maintaining a pentameric state. This suggests that there is a structurally uncharacterized conformation of the TmCorA pentamer in which packing interactions in the periplasmic loops are critical for structural stability. Revealingly, we found that when Mg 2ϩ is added, the pentamer is reassembled via M1/M2-mediated interactions in the cytoplasmic domain. Together, these data indicate that the pentameric state can be maintained either through interactions mediated by the periplasmic loops or the cytoplasmic M1/M2sites. We propose that these two alternative stabilities reflect the molecular mechanism by which the channel stabilizes the open and closed state.

EXPERIMENTAL PROCEDURES
Mutant Construction-The corA gene from T. maritima was cloned with an N-terminal 6ϫ His affinity tag and a tobacco etch virus (TEV) protease cleavage site (MHHHHHHENLY-FQGM) into pGFPi (a modified version of the pGFPe expression vector (13,14) where the gfp gene has been removed). The coding region was therefore under the control of the T7 promoter. This construct is referred to as wt. All mutants were obtained by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit or PfuUltra II Fusion HS DNA polymerase (Stratagene, Sweden) using wt as a template. The Salmonella typhimurium MM281 strain used for the in vivo complementation assay does not have a chromosomal T7 RNA polymerase gene. For that reason, wt and alanine mutants were recloned (excluding the N-terminal 6ϫ His affinity tag and TEV protease cleavage site) into the pBAD/HisA vector (Invitrogen) downstream of an arabinose-inducible promoter. Primers were obtained from MWG (Germany). Cloning was performed in MACH1 cells. Mutations were confirmed by sequencing (MWG).
Protein Purification-Plasmids were transformed into the E. coli strain BL21(DE3) and grown at 37°C in Luria Bertani broth (LB). Protein expression was induced for 3 h by the addition of isopropyl 1-thio-␤-D-galactopyranoside (0.4 mM) when A 600 0.7 was reached. Cells were harvested, washed with 1ϫ PBS (pH 7.4) and resuspended in lysis buffer (1ϫ PBS (pH 7.4) with EDTA-free complete protease inhibitor mixture (Roche) and deoxyribonuclease I from bovine pancreas (Sigma)). All subsequent steps were carried out at 4°C. Cells were broken using an Emulsiflex C3 (Avestin, Canada). Unbroken cells were removed by centrifugation at 9300 ϫ g for 20 min, and membranes were isolated by centrifugation at 210 000 ϫ g for 2 h. Membranes were washed in 1ϫ PBS (pH 7.4). Membrane proteins were solubilized with 1% n-dodecyl-␤-D-maltopyranoside (DDM) (Affymetrix) in 7 ml of solubilization buffer (20 mM Tris-HCl (pH 8), 0.3 M NaCl, 20 mM imidazole) for 1 h. Unsolubilized material was removed by centrifugation at 142000 ϫ g for 30 min. The supernatant containing the solubilized proteins was applied to 2 ml of nickel-nitrilotriacetic acid-agarose (Invitrogen) and incubated for 30 min to allow for His tag binding to the resin. The column was washed with wash buffer (20 mM Tris-HCl (pH 8), 0.3 M NaCl, 40 mM imidazole, 0.03% (w/v) DDM) before elution with 3 ml of elution buffer (20 mM Tris-HCl (pH 8), 0.1 M NaCl, 300 mM imidazole, 0.03% DDM). The elution buffer was changed to gel filtration buffer (20 mM Tris-HCl (pH8), 0.3 M NaCl, 0.03% (w/v) DDM) using prepacked PD-10 columns (GE Healthcare). Amicon Ultra Centrifugal Filters 30K (Millipore) was used to concentrate the proteins to Ϸ5 mg/ml. The purity and integrity of the concentrated proteins was verified by denaturing PAGE (NuPAGE 4 -12% Bis-Tris gels (Invitrogen)). The protein was stored at 4°C for immediate use or snap-frozen in liquid nitrogen and stored at Ϫ80°C.
BN-PAGE-The oligomeric state of the purified proteins was verified by (non-denaturating) BN-PAGE. 5 l of purified protein was diluted with 80 l of ACA 750 buffer (750 mM n-aminocaproic acid, 50 mM Bis-Tris, 0.5 mM Na 2 EDTA (pH 7)) and 15 l of G250 solution (5% (w/v) Coomassie G250 in ACA750 buffer) and loaded onto gels with dimensions of 14 cm ϫ 20 cm ϫ 1.5 mm. High molecular mass markers were obtained from Amersham Biosciences (Sweden). Buffers and gel compositions used were prepared as described previously (15). The experiment was repeated at least twice for every construct.
Size Exclusion Chromatography-Purified proteins were subjected to analytical size exclusion chromatography (aSEC) on a Superdex TM 200 5/150 GL column (GE Healthcare) in the presence and absence of Mg 2ϩ . Protein concentration was adjusted to 0.5 mg/ml in gel filtration buffer (20 mM Tris-HCl (pH8), 0.3 M NaCl, 0.03% (w/v) DDM Ϯ 100 mM MgCl 2 ) and incubated for a minimum of 1 h at 4°C prior to analyses. To estimate the fraction of pentamer and lower oligomers, QtiPlot version 0.9.7. was used to fit two Gaussians to the elution profile. To allow for a qualitative comparison between elution profiles Ϯ Mg 2ϩ , each absorption curve was divided by its peak value.
Chemical Cross-linking-The oligomeric state of the purified proteins was verified by chemical cross-linking using disuccinimidyl suberate (DSS) (Thermo Scientific). Purified protein was adjusted to 0.2 mg/ml in cross-linking buffer (20 mM Na-phosphate buffer (pH 7.5), 0.3 M NaCl, 0.03% (w/v) DDM). The reaction was allowed to equilibrate for 30 min at room temperature. Freshly prepared DSS cross-linker was dissolved in 100% dimethyl sulfoxide to a concentration of 0.1 M and then added to the reactions to a final concentration of 4 mM. Samples were incubated at 37°C for 30 min, followed by quenching with 100 mM Tris-HCl (pH 8) for 5 min. Samples were mixed with 4ϫ sample buffer and then loaded onto denaturing gels (NuPAGE 4 -12% Bis-Tris gels (Invitrogen)). The experiment was repeated at least twice for every protein. The cross-linking experiment entailing Mg 2ϩ was carried out essentially the same way, except that increasing concentrations (0.25-9 mM) of MgCl 2 (titration) or 5 mM MgCl 2 (all loop mutants) was included in the cross-linking buffer. The fraction pentamer and lower oligomers were quantified from three independent repetitions using Bio-Rad Image Lab software (version 4.0.1.).
In Vivo Complementation Assay-pBADa constructs were first introduced into Escherichia strain JR501 for plasmid preparation. The Mg 2ϩ -import deficient S. typhimurium strain MM281 was transformed with JR501-derived pBADa plasmids and grown in LB medium supplemented with 100 mM MgCl 2, 100 g/ml carbencillin, 100 g/ml ampicillin, 50 g/ml kanamycin, 30 g/ml chloramphenicol in 24-well growth plates at 37°C with vigorous shaking (1200 rpm) for 16 h. The cultures were back-diluted 1:100 into 4 ml of fresh LB media with the same composition. When cells reached A 600 Ϸ0.8, they were induced with 0.02% arabinose for 1 h before harvest. To remove Mg 2ϩ , the pellet was washed with 1 ml of buffer A (20 mM Tris-HCl (pH 8), 0.3 M NaCl, 20 mM imidazole) and resuspended in 4 ml of the same buffer (1:1). To verify protein expression, 5 l of cell culture was used for Western blotting using rabbit anti-TmCorA antisera (Innovagen AB, Lund, Sweden). For functional analysis of the mutants, a dilution series with 1:2 increments and an end dilution of 1:64 was performed. From each dilution step, 4 l was spotted on a dried LB agar plate supplemented with the appropriate antibiotics and 0.02% arabinose but no Mg 2ϩ . As a control, the cultures were also spotted on a plate containing 100 mM MgCl 2 . Following overnight incubation at 37°C, the plates were photographed, and growth density was quantified as described previously (16) using ImageGauge V4.23. The growth density of each spot was plotted with respect to dilution factor. The line of best fit was calculated, and the growth of each mutant was quantified as the integral beneath the best fit line. Growth was normalized using the following equation: Normalized growth mutant ϭ (growth mutant Ϫ growth empty vector)/(growth WT Ϫ growth empty vector). At least three independent growth measurements were made for each construct.

The Periplasmic Loop Is Important for Pentamer Integrity in
Low Mg 2ϩ -It has been reported previously that two conserved amino acids in the periplasmic loop of TmCorA (Met-313 and Asn-314) are vital for the isolation of stable pentameric species (9). This observation suggested that the conserved periplasmic loop plays an important structural role. To investigate this issue further, we performed an alanine scan of the loop region (including the flanks of the transmembrane helices) and then purified the resultant 20 proteins (each carrying a single alanine substitution).
To determine whether the loop is important for pentamer integrity when the cytoplasmic metal binding sites are unoccupied, we analyzed the oligomeric state of the 20 mutant proteins in the absence of Mg 2ϩ . Two non-denaturing methods were employed: blue native (BN) PAGE ( Fig. 2A) and aSEC (B and C). To substantiate our observations we used a third approach that entailed adding DSS (a chemical crosslinker) to the purified protein and analyzing the oligomeric state by denaturing PAGE (Fig. 2D). This latter method is non-reversible, thus allowing us to fix protomer interactions in their closest conformation.
From the data reported in Fig. 2, we conclude that the wildtype TmCorA and 12 of the alanine-substituted proteins were pentamers. Intriguingly however, eight of the loop mutants were predominantly detected as lower oligomers (Fig. 2, A-D, I310A, Y311A, M313A, N314A, F315A, M318A, L321A, and Y327A). This observation suggests that each of the alaninesubstituted proteins has lost important interprotomeric interactions that are normally mediated by the native side chain. Note that the oligomeric state of N314A was influenced by the method. In the BN-PAGE it was only detected as lower oligomers ( Fig. 2A). However, when analyzed by aSEC and DSS cross-linking, it was detected in two populations, a high molecular mass population and a low molecular mass population (Fig.  2, B and D, and Ref. 9).
From this analysis we conclude that eight residues in the periplasmic loop of TmCorA are important for the integrity of the pentamer when Mg 2ϩ is absent. Remarkably, seven of these residues are part of either the CorA-family signature motif (YGMNFXXMPEL) or another conserved loop motif (GYP) present in many prokaryotic family members (see underlined residues, 310 IYGMNFEYMPELRWKWGYPV 329 ) (10,11,17). This latter observation suggests that our findings are not just relevant for TmCorA but that the loop mediates pentamer packing interactions in all CorA family members. Such a loop scaffold, or nexus, could potentially be important for ion conduction in an open state of the CorA channels, perhaps by ensuring a relatively defined fold for Mg 2ϩ selectivity.
Interactions To test this hypothesis, we expressed the 20 loop mutants in the S. typhimurium strain MM281. This strain is devoid of all Mg 2ϩ uptake systems and can only grow in LB medium when complemented with a functional TmCorA variant or when supplemented with 100 mM Mg 2ϩ (Fig. 3A) (18). In agreement with our in vitro analysis, we found that seven of the mutants that were detected as lower oligomers and were part of the family signature motifs (Y311A, M313A, N314A, F315A, M318A, L321A, and Y327A) were either unable to complement MM281 or displayed a significantly reduced growth phenotype compared with wild-type TmCorA (Fig. 3, A and B). Thus, there is a clear correlation between pentamer integrity and channel function. Surprisingly, the growth of the cells expressing the I310A mutant exceeded that of the cells expressing the wild-type TmCorA, even though I310A was detected as a lower oligomer in vitro (Fig. 3, A and B). Further work is required to explain this observation. We note, however, that I310 is not a conserved residue in the CorA family (i.e. it is not part of the CorA loop motifs). The apparent requirement for the seven conserved residues to scaffold the TmCorA pentamer provides a plausible explanation to why the equivalent residues have been found important for channel function in other CorA family members (11). In agreement with Payandeh et al. (9), we found that the cells expressing the G312A mutant could not complement the MM281 strain. Thus, the G312A mutant is a non-functional pentamer. We also confirm that the two negatively charged loop-residues, Glu-316 and Glu-320, are not essential for channel function (9,19).
Taken together, the data show that the five periplasmic loops in the pentamer play at least two roles: they stabilize the pentamer in low Mg 2ϩ conditions (when the cytoplasmic Mg 2ϩ binding sites are presumably unoccupied), and they are important for Mg 2ϩ uptake.
Mg 2ϩ Induces a Structural Response-The x-ray structures of TmCorA and subsequent biochemical studies suggested that bound Mg 2ϩ in the cytoplasmic domains stabilize the pentamer (9). Therefore, we hypothesized that addition of Mg 2ϩ might restabilize the eight destabilized loop mutants (i.e. to a pentameric state). To explore this possibility, we added 100 mM Mg 2ϩ to the purified proteins (i.e. a similar concentration to that used in the crystallization experiments (6 -8)) and then reanalyzed the oligomeric state by aSEC. BN-PAGE could not be employed because Mg 2ϩ perturbed running of the gels. As expected, the addition of Mg 2ϩ to the wild-type TmCorA made no difference to the oligomeric state, which remained pentameric (Fig. 4A). Revealingly, the addition of Mg 2ϩ to the eight destabilized loop mutants caused them to transit from lower oligomers toward pentamers (Fig. 4, B and C). Predictably, we found that the Mg 2ϩ -dependent transition from lower oligomers to pentamer was reversible. When we washed the samples with buffer containing EDTA (or buffer that simply lacked Mg 2ϩ ) during aSEC analysis, the mutants were again detected as lower oligomeric species (Fig. 4B, inset).
Taken together, we conclude that mutations in the periplasmic loop do not affect pentamer stability when sufficient concentrations of Mg 2ϩ are present. In accordance with the observation that Mg 2ϩ stabilizes the closed channel confor-mation (9), we conclude that the addition of Mg 2ϩ drives the destabilized loop mutants to adopt a structural arrangement similar to that captured in the crystal structures (i.e. a closed conformation) where the cytoplasmic domains pack together and the structural contribution of the periplasmic loops is negligible.
The M1/M2 Sites Drive the Mg 2ϩ -dependent Pentamerization-How does Mg 2ϩ influence the stability of TmCorA? Is the pentamerization effect brought about by Mg 2ϩ -binding to the M1/M2 sites (i.e. the putative homeostasis sensors), the aspartate ring (at the pore exit), or perhaps another unidentified binding site? To address these questions, we picked one protein with a destabilizing alanine substitution in the periplasmic loop, Y311A, and then incorporated a second alanine substitution that should prevent Mg 2ϩ binding to either the aspartate ring (Y311A:D277A) or the M1/M2 sites (Y311A:D253A) (Fig. 5, A and B). The mutant proteins were purified and analyzed by aSEC in the presence 100 mM Mg 2ϩ or 2 mM EDTA, as described previously. Strikingly, we found that the Y311A:D253A protein could not transit from a lower oligomer to a pentamer in the presence of 100 mM Mg 2ϩ , whereas the Y311A:D277A variant could (Fig. 5C). The reaction was Mg 2ϩ -specific, as when EDTA was present, the fraction of pentameric species was low for all mutant proteins. We therefore conclude that association of Mg 2ϩ to the M1/M2 sites can drive pentamerization of the loop-destabilized Y311A mutant. This observation is in complete agreement with previous studies suggesting that interactions within the M1/M2 sites stabilize the TmCorA protein in a closed conformation (6,9). It thus appears that by disrupting the loop scaffold, we have isolated the M1/M2 homeostasis sensor/gating machinery.

M1/M2-mediated Pentamerization Is Induced at Physiologically Relevant Mg 2ϩ Levels-At what concentration is Mg 2ϩ
sensed by the putative M1/M2-gating machinery? To address this question, we analyzed the oligomeric state of the purified Y311A loop mutant (which has intact M1/M2 sites) in increasing concentrations of Mg 2ϩ . To prevent disassembly during analysis, the oligomeric state was verified by adding the crosslinker DSS prior to analysis by denaturating PAGE. The titration revealed a strong increase in the amount of pentamer between 0.25-5 mM Mg 2ϩ (Fig. 6, A and B). When the concentration of Mg 2ϩ exceeded 5 mM, the reaction appeared saturated (fraction of cross-linked pentamers around Ϸ0.4). As expected for wild-type TmCorA, the fraction of cross-linked pentamers remained relatively constant in this assay (Ϸ0.5), regardless of surrounding Mg 2ϩ concentration (Fig. 6B). 5 mM Mg 2ϩ was also effective in causing the other loop mutants to transit from lower oligomers to pentamers (Fig. 6C). From this analysis we conclude that the M1/M2-driven pentamerization response is induced at physiologically relevant concentration of Mg 2ϩ (0.25-5 mM).
Taken together, the data indicate that Mg 2ϩ sensing by the M1/M2 sites can be triggered at concentrations below 5 mM. However, a higher concentration of Mg 2ϩ is required to keep all M1/M2 sites saturated (i.e. constantly occupied) in solution ( Fig. 6 versus Fig. 4B).
When the Periplasmic Loop Is Intact, TmCorA Is Not Dependent on Mg 2ϩ Binding to the M1/M2 Sites to Remain Pentameric-Our data show that the periplasmic loops are critical for the integrity of the TmCorA pentamer when Mg 2ϩ levels are low, suggesting that the loop scaffold plays an important role in stabilizing the open conformation of CorA. In line with this reasoning, wild-type TmCorA should remain pentameric even when Mg 2ϩ is absent from the M1/M2 sites. To test this hypothesis, we disrupted the cytoplasmic M1/M2 sites in the wild-type TmCorA. The resulting protein, D253A, was purified, and its oligomeric state was analyzed by aSEC. As expected, the fraction of D253A pentamer was similar to wildtype TmCorA, regardless of whether 2 mM EDTA or 100 mM Mg 2ϩ was present (Fig. 5D). This experiment shows that when the periplasmic loop is intact, TmCorA is not dependent on Mg 2ϩ binding to the M1/M2 sites to remain pentameric. It also implies that the putative open loop-mediated conformation is stable in both high and low Mg 2ϩ .

DISCUSSION
The x-ray structures of TmCorA were obtained in the presence of high concentrations of divalent ions, and they provided an important framework for understanding how Mg 2ϩ regulation may be accomplished (6 -8). Unfortunately, all structures were locked in a single closed conformation because of Mg 2ϩ binding in the cytoplasmic domain, leaving many fundamental aspects of Mg 2ϩ uptake unanswered. Notably, when cytoplas-  The oligomeric state of the wild-type TmCorA was unaffected by the addition of Mg 2ϩ (p ϭ 0.24), whereas the eight mutants shifted from lower oligomers to pentamers upon addition of Mg 2ϩ (p Յ 0.03).
closed channel achieved? We have probed these questions by performing biochemical analysis on the TmCorA pentamer in the absence as well as in the presence of Mg 2ϩ .
We first wanted to understand which interprotomeric interactions are important to stabilize the TmCorA pentamer when the cytoplasmic Mg 2ϩ binding sites are unoccupied (i.e. a putative open conformation). A curious discrepancy directed our attention toward the conserved and functionally important periplasmic loop. Although the x-ray structures revealed that the five periplasmic loops in the pentamer were largely disordered (suggesting only a minor structural contribution), Pai and co-workers (9) reported that isolation of stable pentameric species was prevented when two residues in the loop were substituted to alanine (M313A and N314A). To determine whether the loop had a structural role when cytoplasmic Mg 2ϩ levels are low (as opposed to the high Mg 2ϩ conditions employed in the crystallization trials), we performed a systematic alanine scan of the entire loop region. Analysis of the 20 purified mutant pro-teins (each carrying a single alanine point mutation in the loop) revealed eight positions to be important for the integrity of the pentamer when Mg 2ϩ was absent, as they were primarily detected as lower oligomers. Correspondingly, we found a strong correlation between channel function and pentamer integrity, as revealed by an in vivo complementation assay. Taken together, our in vitro and in vivo data strongly indicate that when Mg 2ϩ is absent from the cytoplasmic binding sites, there is a structurally uncharacterized conformation of TmCorA in which the periplasmic loops stabilize the pentamer.
Intriguingly, we noted that seven scaffolding positions belonged to conserved CorA family signature loop motifs (see underlined residues 310 XYGMNFXXMPELXXXXGYPX 329 ). Given the high evolutionary retention of these residues, it seems reasonable to assume that the structural role of the loop determined in this study is a general property in the entire CorA family. Although the structural details of the putative loop scaffold have not been resolved, we envision that its ultimate pur- We next wanted to determine how Mg 2ϩ affects TmCorA stability. Revealingly, by simply adding Mg 2ϩ , the loop-destabilized mutants could transit from lower oligomers to pentam-ers. Importantly, this shows that Mg 2ϩ modulates how the TmCorA pentamer is stabilized and that the periplasmic loop is no longer critical for pentamer stability when Mg 2ϩ is present.
Closer inspection indicated that the Mg 2ϩ -dependent pentamerization was concentration-dependent, reversible, and relied on interprotomeric contacts supplied by the M1/M2 sites in the cytoplasmic domains. In light of the x-ray structures and biochemical and biophysical analysis (6 -9, 12), our conclusion is that Mg 2ϩ stabilized the loop mutants by inducing the closed channel conformation. Thus, if interpreted in terms of M1/M2mediated gating and Mg 2ϩ homeostasis, our data suggest that low concentrations of Mg 2ϩ (Ͻ 5 mM) are sensed by TmCorA and can induce a transition between an open and closed conformation. This is in good agreement with other studies that have shown low concentrations of Mg 2ϩ to have a stabilizing effect on the TmCorA protein (6). On the other hand, we find that high concentrations (Ͼ 100 mM) of Mg 2ϩ are required to saturate the M1/M2 sites. This suggests that a high local concentration of Mg 2ϩ is required to stabilize the closed conformation of the channel, perhaps explaining why a high concentration of divalent cations was reported to be a prerequisite for crystal growth.
In short, our data suggest that intracellular levels of Mg 2ϩ determine whether the periplasmic loops or M1/M2 sites scaffold the pentamer and that this dictates if the ion conduction pore is opened or closed.