Solute Carrier 11 Cation Symport Requires Distinct Residues in Transmembrane Helices 1 and 6*

Ubiquitous solute carriers 11 (SLC11) contribute to metal-ion homeostasis by importing Me2+ and H+ into the cytoplasm. To identify residues mediating cation symport, Escherichia coli proton-dependent manganese transporter (MntH) was mutated at five SLC11-specific transmembrane (TM) sites; each mutant activity was compared with wild-type MntH, and the biochemical results were tested by homology threading. Cd2+ and H+ uptake kinetics were analyzed in whole cells as a function of pH and temperature, and right-side out membrane vesicles were used to detail energy requirements and to probe site accessibility by Cys replacement and thiol modification. This approach revealed that TM segment 1 (TMS1) residue Asp34 couples H+ and Me2+ symport and contributes to MntH forward transport electrogenicity, whereas the TMS6 His211 residue mediates pH-dependent Me2+ uptake; MntH Asn37, Asn250, and Asn401 in TMS1, TMS7, and TMS11 participate in transporter cycling and/or helix packing interactions. These biochemical results fit the LeuT/SLC6 structural fold, which suggests that conserved peptide motifs Asp34-Pro-Gly (TMS1) and Met-Pro-His211 (TMS6) form antiparallel “TM helix/extended peptide” boundaries, lining a “pore” cavity and enabling H+-dependent Me2+ import.

Members of the natural resistance-associated macrophage protein (Nramp/SLC11) family (1,2) are structurally conserved transporters catalyzing cellular uptake of redox metals such as Fe 2ϩ and Mn 2ϩ . Mutations of NRAMP1 orthologs in terrestrial vertebrates were linked to host phagocyte innate response to infections and immune diseases (3,4). Genetic defects in NRAMP2 (aka DMT1) affect iron homeostasis (intestinal absorption, erythropoiesis, and tissue distribution (2,5,6). Despite their medical importance, insight into the structure and function of SLC11 transporters is largely missing. Elucida-tion of the molecular mechanism underlying electrogenic H ϩ and Me 2ϩ symport represents a milestone in the pursuit of future therapeutic approaches to treat Me 2ϩ homeostasis disorders, including brain diseases (7).
Prokaryotic orthologs of essential eukaryotic membrane transport functions represent attractive models to advance understanding of the mechanism of transport (8,9). Studies from different groups showed that Escherichia coli proton-dependent manganese transporter (MntH) 2 is a valuable system for structure/function studies of H ϩ and Me 2ϩ symport (10 -13). MntH TM topology was established and selected mutations resulted in similar phenotypes in MntH and Nramp2 variants (6,14,15). In addition, E. coli MntH wild-type (WT) lacks cysteine residues, and site-directed introduction of Cys moieties combined to thiol modifications allows for detailed structural and functional mapping (16). SLC11-dependent transport is typified by broad selectivity (e.g. Mn 2ϩ , Cd 2ϩ , Fe 2ϩ , Co 2ϩ , and Ni 2ϩ ) and Me 2ϩ -specific interactions, including a range of Me 2ϩ /H ϩ stoichiometries depending on external conditions (10,(17)(18)(19). The H ϩdependent Me 2ϩ transport mechanism of eukaryotic Nramp homologs was deduced from studies of Me 2ϩ uptake and Me 2ϩ -evoked currents, the external pH altering the transporter affinity for Me 2ϩ and the stoichiometry of H ϩ and Me 2ϩ fluxes (2,5). Kinetic models for H ϩ -coupled transport usually imply that H ϩ binding or uptake depends at least partly on the membrane potential (⌬, and at high proton concentration (pH 5.5), large Me 2ϩ -induced currents are mainly due to H ϩ charge transfer across the membrane (17,19).
To identify features defining the mechanism of transport of the SLC11 family, sites representing evolutionary type II rate shifts were targeted because they represent a radical shift in amino acid properties, which can contribute to functional divergence among homologous proteins including membrane proteins (20 -22). Phylogenetic analyses distinguished the SLC11 family from a group of distantly related sequences (out-* This work was supported in part by Canadian Institutes of Health Research group Ͻ30% amino acid identity). Four type II rate shifts were identified at TM sites displaying polar or charged SLC11-specific amino acids matched by distinct outgroup-specific residues (10). Reciprocal residue exchange and additional mutations at each of these sites in TMS1, TMS6, and TMS11 were characterized, showing individual roles of each site in the Me 2ϩ and/or H ϩ symport (10); among them residue Asp 34 appeared crucial for transport activity but its functional role remains undefined (10,11).
The objective of this study was to elucidate the individual roles of the TM residues identified as evolutionary type II rate shift sites by analyzing the impact of the SLC11/outgroup reciprocal mutations on Me 2ϩ and H ϩ uptake kinetics and thermodynamics in vivo and in vitro, by probing the in situ accessibility of targeted sites, and by selecting a tridimensional structural fold obtained by threading that fits our experimental results.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis of E. coli MntH-Site-directed mutagenesis was carried out as described previously (10). The oligonucleotide primers used to generate novel mutants are listed in supplemental Table 1. Mutant clones were selected by profiling in vivo Me 2ϩ sensitivity conferred upon expression in a metallo-dependent E. coli background (12). Full-length DNA sequence was determined for one clone of several that were functionally tested (CEQ 2000XL, Beckman Coulter, Mississauga, ON). The 12TMS-His 10 construct was generated by adding the 12th TMS of the Lactobacillus casei MntH C␤1 homolog (23) and a His 10 tag. A StuI restriction site was introduced prior to the mntH stop codon to enable swapping of a StuI-XbaI flanked fragment of mntH C␤1 (residues 469 -530, followed by a poly-His coding sequence).
Fluorescent Measurements of H ϩ Transport in Vivo-Metalinduced intracellular acidification was measured using a pHdependent ratiometric green fluorescent protein, pHluorin, and cells resuspended at an A 600 of 0.2 in 50 mM citrate-phosphate buffer (pH 4.7-5.7) (14). Fluorescence emission (wavelength 520 nm) was measured on a Fluoromax-2 spectrophotometer (Jobin-Yvon, SPEX) after dual excitation at 410 and 470 nm. MntH-dependent intracellular acidification was deduced from [H ϩ ] changes post-metal addition and monitored for 1000 s. Energy of activation (E a ) values were deduced from [H ϩ ] changes 250 s after addition of 10 M Cd 2ϩ at temperatures varying from 15 to 37°C.
Preparation of Right-side Out and Inside Out Membrane Vesicles-E. coli strain G536 (24) lacking several Fe 2ϩ and Mn 2ϩ transport systems (W3110 ⌬fecABCDE::kan ⌬zupT::cat ⌬mntH ⌬entC ⌬feoABC) was transformed with derivatives of pBAD24 expressing native or mutant MntH. Individual clones were cultured in LB medium containing ampicillin (100 g/ml) at 37°C and 250 rpm until an A 600 of 0.6 was reached. MntH expression was induced for 1 h using 0.1% arabinose. Right-side out vesicles (RSOV) were prepared by osmotic lysis (25), except that spheroplasts were lysed in 0.1 M potassium P i , pH 7.5, washed, and resuspended in 0.1 M Pipes-Mes, pH 7.5. Inside out vesicles (ISOV) were prepared by one passage of the cells (2.5 mg wet weight/ml) through an Aminco French Press at 16,000 p.s.i. in 0.1 M potassium P i , pH 7.5, buffer (26). RSOV and ISOV were resuspended in 0.1 M Pipes-Mes, pH 7.5, at a concentration of 10 -15 mg of protein/ml. The polarity of these preparations was tested by fluorescence spectrophotometry (supplemental Fig. S3) using indicators of (i) changes in ⌬ or ⌬pH resulting from respiration (negative/alkaline or positive/acidic inside, respectively, for RSOV or ISOV), or (ii) topological accessibility of single Cys residues introduced in MntH extramembranous loops formerly assigned to either side of the membrane (14). These Cys mutants (MntH 12TMS-His 10 ) catalyzed Cd 2ϩ -induced intracellular H ϩ uptake (supplemental Fig. 3, E and F). 109 Cd 2ϩ Transport Assays-We used Cd 2ϩ as substrate for MntH because it triggers larger intracellular acidification compared with Mn 2ϩ , Fe 2ϩ , and Co 2ϩ (10). 109 Cd (specific activity 60 mCi/mol; GE Healthcare, Baie d'Urfé, Quebec, Canada) uptake was performed by a quick filtration assay in 0.1 (vesicles preparations) or 0.05 M Pipes-Mes (intact cells), pH 6.5 or 7.5, at 24°C unless otherwise specified. Samples (0.1 ml) were removed at the indicated times and immediately filtered through 0.45-m Metricel GN-6 filters (PALL, East Hills, NY; whole cells), 0.75-m borosilicate GF75 microfiber filters (Advantec MFS Inc., Dublin, CA; RSOV), and 0.22-m nitrocellulose filters (Whatman, Florham Park, NJ; ISOV). Filters were washed with 5 ml of ice-cold uptake buffer containing 1 mM CdCl 2 . All experiments were performed at least in triplicate, using protein concentrations of 50 (RSOV), 200 (ISOV), and ϳ700 g/ml (whole cells, corresponding to an OD 420 of 10) as determined by protein assay using a modified Lowry procedure (27). Vesicles were energized by adding 2 mM L-ascorbate and 0.02 mM N-methylphenazonium methyl sulfate (PMS/ Asc), 10 mM D-lactate, or 5 mM NADH, for 3.5 min prior to the addition of Cd 2ϩ . Ionophores (1 M valinomycin, 0.1 M nigericin, and 10 M carbonyl cyanide m-chlorophenylhydrazone, CCCP) were added 1 min prior to PMS/Asc addition. E a (28) was measured using whole cells at pH 6.5 during the initial, linear phase of Cd 2ϩ uptake (after 10 s or 1 min) at temperatures varying from 24 to 37°C. MntH specificity for Mn 2ϩ was verified using varying concentrations of Mn 2ϩ to compete Cd 2ϩ uptake. By fitting data to the Cheng-Prusoff equation, an apparent inhibitory constant, K Mn i ϭ 1.2 Ϯ 0.2 M at pH 6.5, and [S] ϳ K Cd 0.5 ϭ 2.6 Ϯ 0.5 M Cd 2ϩ , was obtained (29) (supplemental Fig. 1A), consistent with previous studies using 54 Mn 2ϩ (13).
Site-directed Thiol Fluorescence Labeling-Each MntH 12TMS-His 10 Cys derivative conferred in vivo Me 2ϩ sensitivity (supplemental Fig. 4B), consistent with respective data from prior mutagenesis (supplemental Fig. 2) (10), implying that Cys accessibility will reflect the functional structure of native MntH. RSOV (0.5 mg of protein) were labeled by incubation with 0.1 mM fluorescein 5-maleimide (F5M, Vector Laboratories, Burlingame, CA) at 23°C for 10 min, prior to or after solubilization in detergent. When indicated RSOV were pretreated for 10 min with 0.5 mM 4-acetomido-4Ј-maleimidylstilbene 2,2Јdisulfonic acid, disodium salt (Molecular Probes, Eugene, OR), or 0.5 mM N-ethylmaleimide (NEM) and then washed and resuspended or solubilized before labeling with F5M. ISOV (0.5 mg of proteins) were incubated for 5 min in the presence of 2.5 M F5M; labeling was stopped by a 15-min incubation in 20 mM NEM. Membranes were solubilized, and the constructs purified as described under supplemental methods before SDS-PAGE (31) and Coomassie or silver nitrate staining (32). Prior to gel staining, F5M fluorescence was measured under a UV lamp equipped with a green filter using an Alpha Imager 3400 (Innotech, San Leandro, CA).
Western Blot Analyses-C-terminal tagged (-c-Myc or -12TMS-His 10 ) MntH variants were immunodetected in membrane preparations or in purified form as described (10) using anti-c-Myc (PerkinElmer Life Sciences) or anti-His (GE Healthcare) monoclonal antibodies.
Homology Modeling-Several approaches that use an alignment to map residues in a query sequence to sites in candidate template crystals structures were tested using the parameters specified by the developers, including Modeler (33), mGenThreader (34), and several others that are proposed on the meta-server LOMETS, including I-TASSER (9,33,35). Four sequences representative of the SLC11 family (two prokaryotic MntH and two eukaryotic Nramp) were tested and the results compiled to produce a consensus prediction. The PDB coordinates calculated were used visualize three-dimensional models using the freeware viewer PyMol (60). The root mean square deviation and Z-score of the LeuT models were verified using Combinatorial Extension (36), DaliLite (37), and MArkovian TRAnsition of Structure evolution (38).

Targeting Five Type II Evolutionary Rate Sites to Study the
Mechanism of Me 2ϩ and H ϩ Symport-The functional impact of mutating four SLC11-specific TM sites (MntH Asp 34 , Asn 37 , His 211 , and Asn 401 ; Ref. 10) suggested that further kinetic and thermodynamic analyses of mutants at these sites would inform understanding of structure-function relationships. We thus studied mutants in which the SLC11 residues were exchanged for the matching outgroup moiety. The presence of two Asn residues at selected sites suggested also include MntH TMS7 Asn 250 , which is invariant in the SLC11 family and substituted for Thr or Ser in the outgroup. Because replacement of MntH Asn 401 with Thr produces a phenotype similar to the exchange for the matching outgroup moiety (Gly) (10), we analyzed MntH mutants N37T, N250T, and N401T. The TM location of the targeted sites is schematized (Fig. 1A) based on previous predictions and experimental determinations (11,14,15,39). The mutant proteins displayed membrane expression levels similar to MntH-WT, indicating that the substitutions were structurally well tolerated (Fig. 1B).
Characterization of E. coli MntH Cd 2ϩ and H ϩ Transport in Vivo-To study the effects of the external pH on the kinetics of MntH, we measured 109 Cd 2ϩ uptake and Cd 2ϩ -dependent intracellular acidification in intact E. coli cells. Cd 2ϩ uptake by MntH-WT (Fig. 1C) revealed K Cd 0.5 values at 37°C of 2.6 Ϯ 0.5 M at pH 6.5 and 9.1 Ϯ 2.9 M at pH 7.5 (data fitted to the Michaelis-Menten equation, Table 1). The maximal velocity (V max ) of Cd 2ϩ transport was lower at pH 6.5 compared with pH 7.5. MntH-catalyzed Cd 2ϩ -induced H ϩ uptake was activated by increasing amounts of Cd 2ϩ , with K 0.5 values at 23°C ranging from 0.78 to 1.4 M Cd 2ϩ (pH 4.7-5.7, respectively, Table 2 and supplemental Fig. 1, B and C). In both systems, lowering the external pH increased MntH affinity for Cd 2ϩ .
To discriminate carrier-type from channel-mediated transport the Arrhenius E a values were measured (28, 40) for Cd 2ϩ FIGURE 1. Thermal activation of E. coli MntH mutants activity. A, low resolution E. coli MntH topological model. Arrows indicate the predicted location of MntH residues corresponding to SLC11-specific sites. B, Western blot of c-Myc-tagged mutants (80 g of membrane proteins/lane). C, Cd 2ϩ uptake by E. coli GR536 cells expressing MntH-WT and -mutants at SLC11 type II evolutionary sites. Uptake of 10 M 109 Cd 2ϩ (60 mCi/mol) was measured for 10 s at 37°C, pH 6.5. D and E, Arrhenius E a plots of MntH activity (15-37°C) deduced using the equation k ϭ A ϫ exp (ϪEa/RT) , where k is the rate coefficient, A is a constant, R the universal gas constant and T, temperature in K, for Cd 2ϩ uptake quantified by radio-filter assay (D), and Cd 2ϩ -induced H ϩ uptake measured by ratiometric fluorescence analysis of intracellular pH (E).  (16). MntH Cd 2ϩ uptake activity was drastically reduced for four of the mutants studied (Fig. 1C). Further analyses of MntH Cd 2ϩ /H ϩ symport kinetics and thermodynamics revealed various types of effects (Tables 1 and 2), which can be summarized as: (i) low affinity Me 2ϩ uptake and little H ϩ transport (Asp 34 and Asn 37 ), (ii) reduced Me 2ϩ uptake but significant residual H ϩ uptake (Asn 250 , His 211 ), or (iii) slight opposite variations in Cd 2ϩ and H ϩ uptake (Asn 401 ), and (iv) impaired transporter cycling (Asn 37 , Asn 250 , and His 211 ). Individual phenotypes are detailed below.
MntH TMS1 Asp 34 Is Essential for Coupling H ϩ and Cd 2ϩ Uptake-Kinetic analyses of the MntH-D34G mutant revealed dose-dependent Cd 2ϩ uptake, and K Cd 0.5 values that were increased about 2-and 10-fold compared with MntH-WT at pH 7.5 and 6.5, respectively, but Cd 2ϩ -induced H ϩ uptake was not detected even when using 1 mM Cd 2ϩ (supplemental Fig.  2D). The unique properties of the MntH-D34G mutant, such as K Cd 0.5 values little affected by the external pH and lack of H ϩ transport, suggested that exchange of Asp 34 uncoupled Cd 2ϩ uptake from H ϩ influx.
To determine whether a conformational rearrangement was involved we tested the E a of Cd 2ϩ uptake by MntH-D34G, which was similar to MntH-WT (Fig. 1D). The E a of H ϩ uptake could not be measured because no significant activity was detected for any Asp 34 variant (supplemental Fig. 2D). This indicated that mutation of Asp 34 could directly affect H ϩ binding versus transporter cycling, which was confirmed in vitro by lack of a Cd 2ϩ -induced MntH-dependent variation of ⌬ in right-side out vesicles (see below "Electrogenicity of MntH-dependent Forward Cd 2ϩ Transport " and Fig. 3E).
MntH TMS6 His 211 Role in Cd 2ϩ Transport is pH-dependent-The exchange of TMS6 His 211 for Tyr had pH-dependent effects on MntH Cd 2ϩ uptake. Compared with MntH-WT, the K Cd 0.5 value of MntH-H211Y doubled at pH 6.5, but at pH 7.5, Cd 2ϩ uptake levels were indistinguishable from cells lacking a functional MntH (Table 1). At pH 6.5, the MntH-H211Y V max value was reduced (Table 1), and the E a value possibly affected (Fig. 1D). Similar strongly pH-dependent Me 2ϩ uptake activity has been reported for Nramp2 mutation at the homologous site, H267A, which was compensated for by lowering the external pH (41).
To examine whether this SLC11 invariant His residue is crucial for pH-dependent Me 2ϩ binding we measured Cd 2ϩ -induced variations in intracellular pH ( Table 2, supplemental Fig.  1, B and C). The apparent K 0.5 of MntH H211Y was several 100-fold increased at pH 5.3 compared with WT. The V max value for H ϩ accumulation showed that this mutant still catalyzed H ϩ uptake, although the E a value of H ϩ transport was elevated (Fig. 1E). The results thus indicated that replacement of MntH TMS6 His 211 alters H ϩ binding and the catalytic cycle of Cd 2ϩ binding and transport.
MntH Asn 37 , Asn 250 , and Asn 401 Contribute Indirectly to Cotransport Activity-The MntH Asn 250 mutant also showed unique effects compared with MntH-WT: up to an 100-fold increase in the apparent K 0.5 value for Cd 2ϩ -induced H ϩ uptake at pH 4.7 and reduced V max values ( Table 2 and supplemental Fig. 1, B and C); strong reduction in the V max values for Cd 2ϩ uptake, especially at pH 7.5 and a lesser increase of K Cd 0.5 values, mainly at pH 6.5 (Table 1); and elevated E a values of both Cd 2ϩ and H ϩ transport (Fig. 1, D and E). Because N250Q exchange was not conservative (compared with N401Q, supplemental Fig. 2) these results indicated a role of Asn 250 in transporter cycling.
Although replacement of Asn 37 affected Cd 2ϩ and H ϩ transport kinetics similarly to Asp 34 mutation, maybe due to proximity of these sites, important differences were noted: the MntH-N37T mutation did not abrogate Cd 2ϩ -induced H ϩ uptake (supplemental Fig. 2D), whereas the E a value of Cd 2ϩ transport was elevated (Fig. 1D), and K Cd 0.5 values remained pHdependent ( Table 1). The MntH Asn 37 residue may, similarly to Asn 250 , take part in conformational changes during the transport cycle.
MntH mutation Asn 401 to Thr had less impact on transport kinetics than N37T and N250T exchanges (Fig. 1), consistent with Asn to Gln replacement that was conservative only at site 401 (supplemental Figs. 1 and 2). Main effects were on MntH-N401T V max , which was reduced for Cd 2ϩ uptake especially at pH 6.5 (Table 1), but increased ϳ2-fold for Cd 2ϩ -induced H ϩ uptake, independent of the external pH (Table 2, supplemental Fig. 1, B and C). pH-dependent variations of K Cd 0.5 were not correlated with Cd 2ϩ transport V max values. Such effects supported previous observations linking nonconservative mutations of Asn 401 to intracellular acidification in the absence of added Cd 2ϩ (10), due to apparently opposite effects on the V max values of Me 2ϩ and H ϩ uptake.
Characterization of MntH-catalyzed Forward Metal Uptake in Vitro-To confirm the uncoupling effect of the MntH D34G mutation observed in vivo we studied Cd 2ϩ transport in vitro using RSOV. ISOV were used for comparison. MntH-dependent 109 Cd 2ϩ uptake in 0 trans conditions showed different requirements for RSOV or ISOV preparations (Fig. 2, A and B). Uptake into RSOV was stimulated about 10-fold of the background levels by addition of millimolar amounts of Ca 2ϩ and the respiratory substrate PMS/Asc (42), whereas 10-fold more metal was necessary to obtain about 3-fold stimulation of back- ground uptake in ISOV (Fig. 2, A and B). However, both Cd 2ϩ uptake activities were saturable, temperature-dependent (data not shown), and required an active transporter: mutants at sites thought to interact directly with Cd 2ϩ and H ϩ showed little uptake in conditions otherwise favorable for MntH-WT (Fig. 2, C and D). MntH uptake in RSOV was stimulated by a potent electron donor (PMS/Asc versus lactate or NADH, Fig. 3A). In the presence of 1 unit of ⌬pH (10-fold variation in [H ϩ ]), Cd 2ϩ uptake was increased by adding the H ϩ /K ϩ ionophore nigericin, which converts ⌬pH into increased ⌬ (43), and it was abrogated using the K ϩ ionophore valinomycin, which dissipates ⌬ (Fig.  3C). Without external K ϩ , Cd 2ϩ transport still required PMS/ Asc and Ca 2ϩ but was not affected by nigericin, slightly stimulated with valinomycin, and the protonophore CCCP abrogated it (data not shown). Thus, MntH forward Me 2ϩ uptake activity is proton-dependent.
In contrast, and contrary to the expected inhibition by ⌬ (positive inside) of a proton-motive force-dependent mechanism of Cd 2ϩ uptake, MntH-dependent activity in ISOV persisted in the presence of 10 mM D-lactate or 5 mM NADH (Fig. 3B) and was indifferent to nigericin, valinomycin, or both as well as up to 100 M CCCP (Fig. 3D). Also, compared with RSOV, PMS/Asc and Ca 2ϩ induced little stimulation of Cd 2ϩ uptake in ISOV, which was nullified by CCCP (consistent with ϳ90% homogeneous ISOV preparations, supplemental Fig. 3D and data not shown). These data distinguished MntH "reverse " transport from proton-dependent forward transport activity.

Electrogenicity of MntH-dependent Forward Cd 2ϩ Transport-
Observation of Me 2ϩ uptake driven by ⌬ only (negative inside) into RSOV preparations (Fig. 3C) was consistent with in vivo data (supplemental Fig. 4A) using an E. coli ⌬unc strain, unable to convert ATP and the proton-motive force, in which MntH Mn 2ϩ uptake at pH 7.4 (⌬pH ϳ 0) was independent of ATP and abrogated by micromolar amounts of CCCP. These data suggested that MntH forward transport activity, symporting Me 2ϩ and H ϩ , is electrogenic, i.e. it produces charge imbalance across the membrane.
Consequently, MntH activity should contribute directly to modulate ⌬; using conditions where RSOV are energized with ⌬ only (negative inside), MntH-dependent symport of Me 2ϩ and H ϩ should in turn reduce ⌬. Such MntH-dependent charge movement across the membrane was quantified by following Cd 2ϩ -induced depolarization of ⌬. Because addition of 100 M Cd 2ϩ to RSOV harboring MntH-D34G yielded low level Cd 2ϩ uptake at pH 6.5 (data not shown), this concentration was used to follow the Cd 2ϩ -induced variation in ⌬ (less negative inside). Similar depolarization was observed with the MntH-WT and -N401T mutants (Fig. 3E), reflecting electrogenic, non-compensated charge movement across the vesicle membrane. Deficient electrogenicity of MntH-D34G supported the loss of the Cd 2ϩ -induced H ϩ symport in this mutant.  MntH Asp 34 and His 211 Are Accessible to Solvent and Predicted to Line a Permeation Pathway-Direct involvement of Asp 34 in coupled Cd 2ϩ and H ϩ uptakes suggested that this residue could interact with either cation through a water-filled "pore, " and should thus be solvent-accessible. To address this possibility, single Cys mutants were produced using a MntH construct that had been modified to add a 12th TMS and a C-terminal cytoplasmic His tag that allowed affinity purification. The MntH 12TMS-His 10 construct exhibited transport activity similar to MntH-WT (supplemental Figs. 3F and 4B). RSOV harboring each of the MntH single Cys mutants at the five Nramp/SLC11-specific sites (34, 37, 211, 250, and 401) were probed in situ with sulfhydryl reagents, either bulky and polar, F5M, or small and permeable, NEM.
Strikingly, only two sites were reachable in situ: Cys 34 was freely accessible to F5M (Fig. 4A) and Cys 211 reacted solely with NEM (Fig. 4, A and B). These results seemed consistent with the Asp 34 and His 211 respective predicted locations in TMS1 and TMS6 and their distinct direct roles in Me 2ϩ /H ϩ symport. Also, TMS1 Cys 37 and TMS7 Cys 250 became labeled with F5M after solubilizing MntH (Fig. 4, A and B), supporting possible roles in the transporter motion. TMS11 Cys 401 was marginally accessible (Fig. 4, A and B) implying that it participates to tight inter-helical contacts; exchanging Asn 401 for a smaller side chain increased cation uptake, whereas a similar sized moiety limited it (e.g. supplemental Fig. 4B and 2, A and C, respectively, Cys and Gln; prior data using Gly and Thr, see Ref. 10). The results of site-directed labeling thus converged with functional data to demonstrate that SLC11-specific residues contribute critically to MntH transport by either binding substrates directly (Asp 34 , His 211 ) or contributing to inter-helices contacts (Asn 37 , Asn 250 , and Asn 401 ).
Three-dimensional mapping of functional sites is crucial to understand catalytic mechanisms. We determined whether known transporter structures could help model a spatial distribution for the SLC11-specific functional sites. Several template structures, e.g. LeuT, glycerol-3-phosphate transporter, EmrD, and chloride channel, were frequently ranked as significant hits with most programs and for each SLC11 sequence tested. However, only the LeuT/SLC6 family structure (44) fully supported our functional and topological data ( Fig. 4C and supplemental  Fig. 5). Additional predictions obtained using Modeler and either single sequences or groups of sequences representing SLC11 phylotypes as queries produced 16 models, which were all based on the LeuT template, and with scores ranging from 0.01 to 0.76. The similarity of each of these models with the LeuT structure was quantified using DaliLite (37) yielding values in the ranges 44 -60.3 (Z-score), 322-439 (aligned residues), 0.8 -1.7 (Å, root mean square deviation), and 10 -17 (% sequence identity), which indicated that SLC11 transporters may share the same general fold. The model presented ( Fig. 4C; supplemental Fig. 5) exhibits little difference with the LeuT/ SLC6 structure, with values for root mean square deviations and Z-score of, respectively, 1.07 Å and 7.5 (45) or 1.0 Å and 52.9 (37). This LeuT fold places both Asp 34 and His 211 in the inner core of the MntH molecule, as parts of extended peptides interrupting TMS1 and TMS6, and lining a water-filled pore. The three Asn 37 , Asn 250 , and Asn 401 appeared directed toward inter-helix contacts and the outer fence of the model architecture ( Fig. 4C and supplemental Fig. 5). Structural modeling of SLC11 evolutionary distinct sites thus corroborates their key functional roles.

DISCUSSION
The present study showed that MntH Asp 34 and His 211 , two SLC11 family-specific and ionizable residues located in TMS1 and TMS6 are necessary for binding and transport of H ϩ and Me 2ϩ . Whereas interdependent transport of Me 2ϩ and H ϩ by MntH is consistent with properties reported for eukaryotic Nramp homologs (17,46), we demonstrated here for the first time that exchange of MntH TMS1 Asp 34 uncoupled H ϩ and Cd 2ϩ transport and also reduced affinity for Cd 2ϩ , but preserved pH-independent Cd 2ϩ uptake and transporter cycling. Because this site was also freely accessible in situ, it is likely part of an H ϩ translocation pathway. Exchange of MntH TMS6 His 211 rendered affinity for Cd 2ϩ more dependent on the external pH but preserved Cd 2ϩ -induced H ϩ transport; Cys 211 was accessible in situ only to a small reagent, supporting a direct role of His 211 (de)protonation in Cd 2ϩ transporter cycling. The SLC11-specific TMS1 Asp and TMS6 His residues constitute thus functional determinants of Me 2ϩ and H ϩ symport.
The TMS1 Asp residue is part of a conserved DPGN motif that has been subjected to mutagenesis in studies using MntH or Nramp2 homologs, which showed loss of Me 2ϩ uptake caused by Gly exchange (11,47). The carboxyl end of Nramp2 TMS1 and adjacent extra loop were implicated in Me 2ϩ binding and coupling of Me 2ϩ uptake to the proton-motive force (47). MntH scanning mutagenesis revealed that only Pro 35 exchange did not abrogate Mn 2ϩ uptake, and MntH-P35G had a marginally affected K Mn 0.5 (11). Here, the role of the flanking SLC11 family-specific site Asn 37 was detailed. Exchange of Asn 37 seemed to perturb interactions among TMS rather than between MntH and cations, as demonstrated by inaccessibility to a small soluble compound in situ, impaired Cd 2ϩ and H ϩ transporter cycling but residual Cd 2ϩ -induced H ϩ transport, and pH dependence of Cd 2ϩ uptake. We deduce that the SLC11 invariant tripeptide (DPG) may bear analogy to a known functional signature for Me 2ϩ/ϩ transport, i.e. ((C/S/T)P(C/H)) that is conserved in the TMS6 of P 1B -type ATPases, which pump heavy metal cations using energy provided by ATP hydrolysis (48,49). The SLC11-invariant Asp and Gly residues in TMS1 could thus contribute to Me 2ϩ binding, the acidic moiety being key for H ϩ -coupled transport.
The His residue in TMS6 was shown previously to regulate Me 2ϩ uptake via Nramp2 and MntH (10,41). We further showed that low Me 2ϩ uptake persisted with MntH-H211C (supplemental Fig. 4B), similar to the Nramp2 matching mutant H267C (41), and that MntH-H211Y low affinity for Cd 2ϩ did not prevent H ϩ uptake. It seems thus unlikely that this His residue could either bind Me 2ϩ directly, or simply be part of a relay or channel enabling H ϩ movement across the membrane (50,51). Instead, (de)protonation could favor a conformation facilitating Me 2ϩ translocation and transporter cycling; TonB His 20 is an example of a TM His moiety required for transport cycling and the energy transduction event (52). The SLC11-specific TMS6 His residue is part of another conserved motif (MPH) (with exceptions in plasmodia, plants, and fungi homologs) not found in the SLC11 phylogenetic outgroup (V(P/G)Y) (10), suggesting by analogy with TMS1 that the TMS6 motif (MPH) may represent another signature for metal transport.
Strikingly, modeling SLC11 homologs on the LeuT/SLC6 structure suggests an internal symmetry (including two domains made of helices 1-5 and 6 -10, similarly folded but in inverted orientation with respect to the membrane (53), and thus a possible origin for the motifs in TMS1 and TMS6, which appear central to the model transporter architecture. These two motifs in anti-parallel orientation ((DPG) and ("HPM")) form a pair of extended peptides interrupting TMS1 and TMS6. Similar pairs of discontinuous helices (␣-helix-extended peptide-␣-helix) constitute a salient feature in five cation transporter structures, and pairs of antiparallel discontinuous TM helices may be part of a fold shared by several cation transporter families (54).
Burying such (DPG) and (HPM) "extended peptides/polar helix termini" elements within the low-dielectric core of the membrane is not energetically favored, partly due to charges, including polar backbone groups not engaged in intra-chain hydrogen bonds. Such elements should be stabilized by interacting with adjacent residues, transported substrates (e.g. dehydrated cations), or by mediating alternate conformation rearrangements (44,54). Our data indicate that interactions with Me 2ϩ and/or H ϩ may stabilize SLC11 TMS1 Asp and TMS6 His. Mapping to the LeuT model are other sites of potential divergence between the SLC11 family and outgroup, suggested clustering around these central motifs, (DPG) and (HPM) (supplemental Fig. 5, C and D), and possible inter-helix contacts, which could be considered in future structure/function studies using the present approach. The targeted Asn residues in TMS1, TMS7, and TMS11 affected transport without interacting with cations or being accessible to NEM in situ (Asn 37 , Asn 250 , and Asn 401 ). According to the LeuT fold, these residues may be oriented for inter-helix contacts, possibly key for transport cycling, e.g. TMS7 Asn 250 and TMS1b Asn 37 , or velocity (Asn 401 ), because TM Asn residues can drive strongly interhelix associations (55,56) or mediate structural motion during ion permeation (57).
Analysis of E. coli MntH activity in vitro yielded surprising results concerning activation of MntH forward Cd 2ϩ transport by millimolar amounts of Ca 2ϩ , and reverse Cd 2ϩ transport that appeared passive. Requirement for millimolar Ca 2ϩ in addition to RSOV energization was previously reported to stimulate Mn 2ϩ uptake in E. coli RSOV (42). Because MntH is the sole manganese uptake system known in E. coli laboratory strains, our data may suggest a possible link between MntH transport and external Ca 2ϩ , which might be reminiscent of interactions between a DMT1 mutant and Ca 2ϩ (6). Strong asymmetry observed between MntH forward and reverse uptakes supports the proposition that MntH functions as a pump driven by ⌬, accumulating cytosolic Me 2ϩ against their concentration gradient. Kinetic studies of H ϩ -coupled transport systems have shown that ⌬ regulates H ϩ binding or uptake, and that H ϩ coupling increases a transporter affinity for its substrates or provides some thermodynamic force for the translocation step (2). Such a mechanism is widespread in microbes and was conserved in higher eukaryotic systems where an acid microclimate produces a huge proton gradient (e.g. the epithelial brush-border membrane of the proximal intestine, site of uptake by Nramp2 of non-heme dietary iron (58), and the phagosomal lumen in phagocytes that is depleted from Me 2ϩ by Nramp1 (59)). It is thus expected that future work on the E. coli MntH mechanism of the H ϩ -dependent transport will reveal additional structure-function relationships key to the Nramp/SLC11 family.
Evolutionary targeting of sites for biochemical analyses and three-dimensional validation of the results provides a framework to locate distinct TM residues key to the mechanism of the SLC11 membrane transport family. Low levels of sequence identity between SLC11 and SLC6 families (Ͻ20%) implies that the "LeuT-like" structural models obtained probably indicate a possibility that different families of cation transporters may share a general fold that could include inverted symmetry and discontinuous TM helices. SLC11 crystals will be required to ultimately validate any predicted model with resolution and accuracy. Nevertheless, the high resolution of the LeuT structure, which was corroborated by extensive biochemical and mutagenesis data, provided a high quality modeling template, and the suggested fit between the deduced architecture and SLC11 transporters topological and functional data imply that the current model holds significant potential for future studies of the SLC11 mechanism of Me 2ϩ /H ϩ symport and investigations of naturally occurring mutants in relation to diseases.