Selective Metal Binding to a Membrane-embedded Aspartate in the Escherichia coli Metal Transporter YiiP (FieF)*

The cation diffusion facilitators (CDF) are a ubiquitous family of metal transporters that play important roles in homeostasis of a wide range of divalent metal cations. Molecular identities of substrate-binding sites and their metal selectivity in the CDF family are thus far unknown. By using isothermal titration calorimetry and stopped-flow spectrofluorometry, we directly examined metal binding to a highly conserved aspartate in the Escherichia coli CDF transporter YiiP (FieF). A D157A mutation abolished a Cd2+-binding site and impaired the corresponding Cd2+ transport. In contrast, substitution of Asp-157 with a cysteinyl coordination residue resulted in intact Cd2+ binding as well as full transport activity. A similar correlation was found for Zn2+ binding and transport, suggesting that Asp-157 is a metal coordination residue required for binding and transport of Cd2+ and Zn2+. The location of Asp-157 was mapped topologically to the hydrophobic core of transmembrane segment 5 (TM-5) where D157C was found partially accessible to thiol-specific labeling of maleimide polyethylene-oxide biotin. Binding of Zn2+ and Cd2+, but not Fe2+, Hg2+, Co2+, Ni2+, Mn2+, Ca2+, and Mg2+, protected D157C from maleimide polyethylene-oxide biotin labeling in a concentration-dependent manner. Furthermore, isothermal titration calorimetry analysis of YiiPD157A showed no detectable change in Fe2+ and Hg2+ calorimetric titrations, indicating that Asp-157 is not a coordination residue for Fe2+ and Hg2+ binding. Our results provided direct evidence for selective binding of Zn2+ and Cd2+ for to the highly conserved Asp-157 and defined its functional role in metal transport.

Membrane transporters in the cation diffusion facilitator family are found both in eukaryotes and prokaryotes (1). This protein family of more than 400 genetically related members is characterized by a homologous hydrophobic N-terminal domain followed by a hydrophilic C-terminal domain that is variable both in sequence and length (2). Despite sequence variability, all CDF 3 family members exclusively transport zinc and other divalent metal ions across the cytoplasm or organelle membranes, thus playing a variety of important roles in cellular zinc homeostasis controls (3)(4)(5)(6)(7). Phenotype analyses of gene deletion and complementation have demonstrated that bacterial CDF transport-ers are involved in cellular resistance to a broad spectrum of divalent metal cations, including Zn 2ϩ , Cd 2ϩ , Co 2ϩ , Mn 2ϩ , Ni 2ϩ , and Fe 2ϩ (8 -13). To date, molecular identities of substrate-binding sites in CDF transporters have yet to be identified, and even less is known about metal binding selectivity at a molecular level.
The Escherichia coli metal transporter YiiP (FieF) and its homolog ZitB are the two CDF proteins that were shown to function as an obligatory Zn 2ϩ /H ϩ antiporter (12,14). This proton-linked antiport mechanism permits the free energy derived from the downhill H ϩ influx to be coupled to the uphill pumping of Zn 2ϩ out of the cells. The kinetics of Zn 2ϩ /Cd 2ϩ transport was described by a two-step process involving an initial binding step followed by a conformational transition that moves the metal ion across the membrane (14). The binding of metal ions to YiiP exhibited distinctive heat reactions in response to calorimetric titrations of the Zn 2ϩ 3 Cd 2ϩ 3 Hg 2ϩ series, leading to thermodynamic categorization of at least one mutually competitive binding site common to Zn 2ϩ , Cd 2ϩ , and Hg 2ϩ , and a set of noncompetitive binding sites (15). Other than binding to group 12 metal ions, YiiP was implicated to play a role in bacterial iron detoxification, suggesting that Fe 2ϩ may be an additional substrate (12). Both Zn 2ϩ and Fe 2ϩ are borderline soft metal ions with a similar donor preference for a combination of three protein ligand groups as follows: the cysteine thiolate, histidine imidazole, and aspartate/glutamate carboxylate (16). Phenotype analyses of two homologous CDF transporters, CzcD from Ralstonia metallidurans and ZitB from E. coli, demonstrated that mutating a highly conserved Asp residue in the putative TM-5 rendered host cells hypersensitive to zinc, probably because of a loss of zinc efflux pumping activities (17). The Asp appeared to be essential because expression of CzcD D158E , CzcD D158A , ZitB D163E , and ZitB D163A mutant proteins conferred no zinc resistance. It is not clear, however, whether this conserved CDF aspartate is directly involved in binding and transport of Zn 2ϩ and/or Fe 2ϩ .
In the present study, we sought to establish the functional role of the equivalent Asp residue in YiiP (Asp-157) by examining the effects of D157A and D157C mutations on metal binding and transport by using direct biophysical measurements. Furthermore, Asp-157 was localized to the hydrophobic core of TM-5 to establish a structural connection with the substrate translocation pathway where metal ions are selected and transported across the membrane. Metal binding to Asp-157 was found to be highly specific, with a strict selectivity for Zn 2ϩ and Cd 2ϩ over Hg 2ϩ , Fe 2ϩ , and other divalent metal ions that are thought to be the frequent CDF substrates. Because Asp-157 is one of the most conserved residues in the CDF family, selective metal binding to Asp-157 has broad structural and mechanistic implications.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis-Cloning and construction of the expression plasmid pYiiP-His were described previously (15). Site-directed mutagenesis was performed using the QuickChange site-directed mutagenesis kit (Stratagene). A mutant C287S was first prepared using * This work was supported by the Laboratory Directed Research and Development Program of the Brookhaven National Laboratory, Project 05-064 (to Y. W. and D. F.), and by a National Institutes of Health Grant RO1 GM65137 (to D. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Goldhabor postdoctoral fellow. 2  the pYiiP-His plasmid DNA as template and an anti-parallel pair of primers . The resultant C287S plasmid DNA served as the parent for an  additional 13 mutants, each containing a single cysteine substitution  mutation at positions 10, 36, 70, 107, 144, 150, 153, 155, 157, 171, 172, 174, or 177. Sequences of these mutants were verified by DNA sequencing of both strands.
Overexpression and Purification-YiiP and mutants were overexpressed with a C-terminal extension containing a thrombin cleavage site followed by six tandem histidine residues to facilitate protein purification. The expression host cells, BL21 (DE3) pLyS, were cultured in an auto-inducing medium for unattended protein overexpression (18). Cells from overnight cultures were harvested, and membrane proteins were extracted using 7% n-dodecyl-␤-D-maltopyranoside (DDM) as described previously (15). The detergent-solubilized proteins were absorbed by three passages through a Ni 2ϩ -nitrilotriacetic acid superflow column (Qiagen), which was washed free of contaminants and eluted with an elevated imidazole concentration at 500 mM. The column eluate was immediately applied to a PG-10 gel filtration column (Amersham Biosciences), yielding a desalted sample that was subsequently subjected to overnight thrombin digestion (Novagen) at a ratio of 0.5 units of thrombin per mg of protein. The completeness of thrombin digestion was confirmed by matrix-assisted laser desorption ionization time-of-flight mass spectrometric analysis, showing a complete conversion of the His tagged to a tag-free mass species. The resultant tag-free protein was incubated with 10 mM EDTA for 30 min and then applied to an TSK 3000SW XL size-exclusion HPLC column (TosoHaas), preequilibrated with a degassed HPLC buffer (20 mM HEPES, pH 7.0, 100 mM NaCl, 12.5% glycerol, 0.05% DDM, 0.2 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP)). The purified protein was collected as a discrete chromatographic fraction using a Beckman SC100 fraction collector.
Isothermal Titration Calorimetry-Protein aggregates and trace amounts of metal contaminants were removed by size exclusion HPLC prior to calorimetric titrations as described previously (15). Protein concentrations were determined by BCA protein assay (Pierce). Calorimetric titrations were carried out on a Microcal MCS titration calorimeter (Microcal) at 25°C. Metal titrants (chloride salts), typically in the concentration range of 0.25-0.5 mM, were dissolved in the same HPLC mobile phase used for protein purification. 1 mM ascorbate was added when FeCl 2 was titrated. Titrants and protein samples were thoroughly degassed, and then 30 -50 injections of an indicated titrant were made successively into a protein sample in 5-l increments at 210 -360-s intervals. Heats of titrant dilutions were measured by making identical injections into the HPLC buffer, subtracted from the corresponding total heats of reaction to yield net reaction heats. The titration data were deconvoluted based on a binding model containing either one or two sets of noninteracting binding sites by a nonlinear least squares algorithm using the Microcal Origin software. The binding enthalpy change ⌬H, association constant K a and the binding stoichiometry n were permitted to float during the least squares minimization process and were taken as the best fit values.
Reconstitution and Stopped-flow Transport Assay-HPLC-purified YiiP, YiiP D157C , or YiiP D157A was reconstituted into liposomes made of E. coli polar lipids (Avanti Polar Lipids) as described previously (14). Control liposomes were prepared following exactly the same procedure without adding protein. 200 M fluorescence indicator fluozin-1 (Molecular Probes) was encapsulated by two freeze-thaw cycles, followed by gel filtration to remove the untrapped dye. Transport experiments were performed at 8°C on a stopped-flow apparatus (KinTek Corp.). Proteoliposomes and a transport assay buffer (20 mM HEPES, 50 mM K 2 SO 4 , and either ZnSO 4 or CdSO 4 at an indicated concentration ranging from 0 to 4 mM, pH 7.3) were loaded into two separate syringes of equal volume, and transport reactions were initiated by pushing 60-l fresh reactants at a 1:1 ratio through the 12-l mixing cell at a flow rate of 20 ml/s. Zn 2ϩ or Cd 2ϩ concentrations in reaction mixtures were half of the concentrations in the initial transport assay buffers. Stopped-flow traces were the cumulative average of five successive recordings at 525 nm (excited at 490 nm). Liposome traces were collected as base lines and subtracted from proteoliposome traces to yield net fluorescence changes ⌬F. ⌬F/⌬F max was obtained by normalizing ⌬F to the maximum proteoliposome response elicited by a transport assay buffer containing 4 mM ZnSO 4 or CdSO 4 plus 2% n-octyl-␤-D-glucoside used to solubilized proteoliposomes.
Determination of Transport Kinetic Parameters-The rate of the fluorescence rise k obs was determined by least squares fit of the kinetic trace to a single exponential function using the data analysis software SigmaPlot 4.0 (SPSS Inc., Chicago, IL). The following two-step kinetic scheme (Scheme 1) was used to describe the YiiP transport process (14), where T 1 and T 2 are different conformational states of YiiP, and M is the metal ion substrate. k 1 , k 2 , and k 3 are the rate constants. The dissociation constants of metal binding K d ϭ k 2 /k 1 . The relationship among k 1 , k 2 , and k 3 is defined as K m ϭ (k 2 ϩ k 3 )/k 1 . Application of the steady-state condition to the species MT1 gives Equation 1 as described previously (19). MPB Labeling-MPB labeling was carried out with cells that expressed YiiP or a YiiP variant as indicated. Cells (1 ml) from overnight cultures in the auto-inducing medium were pelleted by centrifugation and resuspended in a reaction buffer (0.5 ml) containing 20 mM HEPES, 100 mM NaCl, 2 mM MgCl 2 , 10% sucrose, 0.25 mM TCEP, pH 7.5. A freshly prepared MPB stock solution (50 mM) was added to a final concentration of 2 mM, or an equal volume of double deionized water (20 l) instead of MPB was added to a control sample as indicated. Cells were incubated with MPB at room temperature for 30 min with or without sonication (30 s), and then ␤-ME (20 mM) was added to quench the unreacted MPB. The resulting cells were pelleted again and washed, and membrane proteins were extracted using a solubilization buffer (20 mM HEPES, 100 mM NaCl, 1% DDM, 20% glycerol, 0.25 mM TCEP, pH 7.5) with brief sonication (1 min), followed by a 30-min incubation at 10°C to achieve complete detergent solubilization. Cellular debris was removed by centrifugation (10,000 ϫ g for 30 min), and supernatants were collected and incubated with 20 l of Ni 2ϩ -nitrilotriacetic acid superflow resin for 30 min. The resin was washed free of contaminants with 2 ml of wash buffer (20 mM HEPES, 300 mM NaCl, 40 mM imidazole, 12.5% glycerol, 0.05% DDM, and 0.25 mM TCEP, pH 7.0), and then eluted with 50 l of elution buffer (20 mM HEPES, 100 mM NaCl, 500 mM imidazole, 12.5% glycerol, 0.05% DDM and 0.25 mM TCEP, pH 7.0). The purified proteins obtained were immediately subjected to fluorescence labeling.
Fluorescence Labeling and Western Blot-The MPB-treated and purified proteins (ϳ0.5 mg/ml) were incubated with 0.1 mM fluorescein 5-maleimide (FM, Molecular Probes) in the presence of 10% SDS at room temperature for 20 min, and then ␤-ME was added to 10 mM to terminate the reaction. The resulting FM-treated proteins were subjected to SDS-PAGE on an 8 -16% Tris-HCl precast polyacrylamide gel (Bio-Rad). The gel was visualized on a UV transilluminator and documented using a BioDoc-It System (Ultraviolet Products). After fluorescence detection, proteins were transferred to nitrocellulose using a Trans-blot semi-dry transfer cell (Bio-Rad) and were exposed to a peroxidase-conjugated monoclonal anti-biotin antibody (Sigma) for detection of MPB labeling by a SuperSignal West Pico chemiluminescent substrate (Pierce). For FM labeling to membrane-bound YiiP variants, cells hosting the overexpression of a YiiP variant were harvested and resuspended in the reaction buffer, and then 1 mM FM was added with a brief sonication. After incubation for 30 min at room temperature, 10 mM ␤-ME was added to quench unreacted FM. The resulting cells were pelleted and washed, and the FM-treated proteins were purified as described above. Aliquots of purified proteins were either directly subjected to SDS-PAGE or subjected to another exposure of 0.1 mM fresh FM in the presence of 10% SDS, followed by SDS-PAGE. Proteins in the gels were visualized under UV light for fluorescence detection before being stained with Coomassie Blue for estimation of the total amount of proteins.

Correlation between Cd 2ϩ
Binding and Transport-The transport of metal ions is a sequential process of equilibrium binding and energized movement of metal ions along one or more binding sites in a translocation pathway across the membrane (20). To establish the functional role of Asp-157 in metal binding and transport, we examined the effects of mutating Asp-157 to a cysteine or alanine residue, corresponding to a metal coordination or noncoordination residue. His-tagged YiiP and mutants were overexpressed and purified to homogeneity by nickel affinity chromatography, followed by thrombin cleavage of the affinity tag, EDTA chelation of bound metal ions, and size exclusion HPLC purification to yield protein samples suitable for metal calorimetric titrations (15). Cd 2ϩ binding to YiiP, YiiP D157A , and YiiP D157C was examined directly by ITC at 25°C, pH 7.0, as described under "Experimental Procedures." Examples of heat changes resulting from binding of incremental additions of Cd 2ϩ and plots of the integrated heat per mol of Cd 2ϩ as a function of the Cd 2ϩ /protein molar ratio are displayed in Fig.  1A. The heat effects generated by Cd 2ϩ binding to YiiP and YiiP D157C dropped sharply near a stoichiometric equivalence point of 2.5, whereas the midpoint of binding heat changes for YiiP D157A occurred at 1.5, indicating a loss of 1 eq Cd 2ϩ -binding site by the D157A mutation. Accordingly, binding isotherms were fitted with a two-site model for YiiP and YiiP D157C and a one-site model for YiiP D157A . As shown in TABLE ONE, binding affinities, stoichiometries, and ⌬H changes of site 1 and site 2 for YiiP and YiiP D157C are nearly identical and within experimental errors. Fit of YiiP D157A binding isotherm to a one-site model resulted in K a ϭ 7.3 Ϯ 1.9 M Ϫ1 , n ϭ 1.5 Ϯ 0.1, and ⌬H ϭ Ϫ5.7 Ϯ 0.2 kcal/mol, in excellent agreement with the binding parameters of site 1 of both YiiP and YiiP D157C . Thus, a D157C mutation caused no change to both sites 1 and 2, whereas a D157A mutation completely abolished Cd 2ϩ binding to site 2 but had no effect on site 1. Furthermore, the binding stoichiometries for site 2 were close to unity, suggesting that a D157A mutation disrupted one Cd 2ϩ -binding site per YiiP subunit.
The ITC data indicated the presence of at least two Cd 2ϩ -binding sites, each of which could play a structural, functional, or regulatory role. To examine a possible correlation between Cd 2ϩ binding to Asp-157 and Cd 2ϩ transport, HPLC-purified YiiP, YiiP D157A , and YiiP D157C were reconstituted into proteoliposomes, and the kinetics of Cd 2ϩ transport was analyzed by stopped-flow measurements of fluorescence changes of an encapsulated Zn 2ϩ /Cd 2ϩ -sensitive indicator, fluozin-1, in response to rapid mixtures of proteoliposomes with Cd 2ϩ exterior to vesicles. SDS-PAGE analysis of proteoliposome samples confirmed that approximately the same amount of YiiP, YiiP D157A , and YiiP D157C was reconstituted into vesicles. Mixing proteoliposomes with external Cd 2ϩ evoked rapid and progressive fluorescence increases that were dependent of Cd 2ϩ concentrations ranging from 0 to 2 mM in the reaction mixture as described under "Experimental Procedures." Liposomes prepared in parallel to proteoliposomes only yielded negligible background fluorescence responses. A linear correlation between the initial rates of fluorescence responses and the molar ratios of YiiP/lipid was observed, indicating a linear relationship between the initial rate and the transport activity (data not shown). As shown in Fig. 1B, the fluorescence responses of YiiP and YiiP D157C were comparable in amplitudes within the Cd 2ϩ concentration range, in contrast to greatly diminished   Fig. 2A displayed a mixed heat reaction that began with exothermic and followed by late endothermic heat changes. In addition to the main exothermic-to-endothermic transition, another enthalpic transition was evident at the beginning of Zn 2ϩ titrations where exothermic heat effects increased progressively. The overall profile of the YiiP D157C binding isotherm was comparable with that of YiiP, but the binding isotherm of YiiP D157A showed a leftward shift of the exothermic-to-endothermic transition by about 1 stoichiometry unit, qualitatively corresponding to the loss of one exothermic binding site. To a first approximation, Zn 2ϩ binding isotherms were fitted with two sets of independent binding sites, accounting for the endothermic and exothermic heat reactions with the respective binding stoichiometries of 0.6 Ϯ 2.4 and 2.0 Ϯ 0.1 for YiiP, 0.6 Ϯ 1.8 and 1.9 Ϯ 0.1 for YiiP D157C , and 0.7 Ϯ 3.1 and 1.1 Ϯ 0.1 for YiiP D157A (TABLE ONE). Compared with YiiP, YiiP D157C appeared to retain all Zn 2ϩ -binding sites, whereas YiiP D157A was short of one exothermic site. The presence of multiple heat transitions precluded fitting of Zn 2ϩ binding isotherms with certainty, as indicated by significant fitting errors. Thus a quantitative comparison of Zn 2ϩ binding parameters was not amenable. Nevertheless, it was evident that a Zn 2ϩ exothermic binding site was disrupted in YiiP D157A , whereas the same site remained intact in YiiP D157C. Furthermore, multiple transitions observed in the YiiP D157A binding isotherm indicated the presence of at least two additional Zn 2ϩ -binding sites after disruption of the Asp-157 site. In corroboration of the effects of Asp-157 mutations on Zn 2ϩ binding, the rate of Zn 2ϩ transport was unchanged with k 3 values of 34 Ϯ 5 and 33 Ϯ 3 s Ϫ1 for YiiP and YiiP D157C , respectively, making a sharp contrast to YiiP D157A responses that were reduced to the background level (Fig. 2B). The K m values estimated from three experiments were 310 Ϯ 32 and 358 Ϯ 36 M for YiiP and YiiP D157C , respectively, indicating no significant change in Zn 2ϩ binding.
Topological Mapping of Asp-157-A mechanistic understanding of the functional role of Asp-157 depends heavily on the ability of mapping it to a reliable topology model. The membrane topology of any CDF transporter has not yet been determined experimentally, but six segments of hydrophobic residues are suggested by YiiP hydropathy analysis. To determine whether each of these hydrophobic segments actually traversed the membrane and, if so, to determine their membrane spanning polarity, we introduced a series of cysteine substitution mutations, each located at the beginning or end of a hydrophobic segment. When intact cells were exposed to an impermeant thiol-specific probe, the probe accessibility to a reporter cysteine residue could indicate its extracellular or intracellular location. Two maleimide derivatives, FM and MPB, were used in this study. The maleimide group in FM is directly attached to a bulky and charged fluorescein moiety, and the maleimide in MPB is tethered to its biotin moiety through a linear polyethylene oxide chain (Fig. 3A). The bulkiness and the hydrophilic nature of both maleimide derivatives make them impermeant to the cytoplasmic membrane (21). YiiP contains two native cysteine residues at positions 127 and 287. FM or MPB labeling to Cys-127 was not detected when YiiP C287S was exposed to 1 mM FM or MPB, even in the presence of 10% SDS (Fig. 3B). Thus a panel of single cysteine substitution mutants was constructed on the YiiP C287S background. Only one reactive cysteine residue is present in each of the following YiiP variants at a position as numbered: S10C/C287S, S36C/C287S, N70C/C287S, S107C/C287S, S144C/C287S, R177C/C287S, Cys-287 (equivalent to wild type YiiP). All these mutants were found to be fully functional by stopped-flow analysis, thereby ascertaining the structural and functional relevance of the topological mapping.
MPB labeling was carried out with intact cells, from which MPBtreated YiiP variants were purified, and then exposed to FM in 10% SDS. The resultant MPB-FM doubly treated YiiP variants were separated on a SDS-polyacrylamide gel for transillumination of FM labeling and then Western blot detection of MPB labeling. As shown in Fig. 3C, Western blots revealed a strong protein band in lane 1 for S36C, S107C, and R177C (top panel) but no detectable signal for S10C, N70C, S144C, and  Fig. 3C, lane 1, for S36C, S107C, and R177C, as opposed to an intense fluorescence signal in lane 1 for S10C, N70C, S144C, and Cys-287. Because MPB labeling was directed to cysteine residues on the extracellular surface, whereas FM labeling was directed to any cysteine residue that survived the first run of MPB labeling, the location of an MPB-or FM-reactive residue was interpreted to be extracellular or intracellular, respectively. Correlating MPB and FM labeling with the positions of seven cysteine residues in a sequence from the N to C terminus revealed an alternating pattern of MPB reactivities and a reversed alternating pattern of FM reactivities, both of which were consistent with a topology model depicted in Fig. 3C (middle panel). In a control experiment shown in Fig. 3C, lane 2, of each gel, MPB labeling with sonicated cells yielded a strong and uniform MPB chemiluminescence signal regardless of the cysteine position. The subsequent FM labeling showed a background level of fluorescence signal, indicating the completeness of MPB labeling to all seven cysteine residues when the membrane barrier was disrupted by sonication. Likewise, FM labeling of the purified YiiP variants without MPB pretreatment yielded a strong and uniform fluorescence signal at all positions (Fig. 3C, lane 3), indicating that the seven cysteine residues were fully reactive with FM when the plasma membrane was disrupted by detergent solubilization. Taken together, these control experiments ensured that the alternating pattern observed with intact cells truly reflected the transmembrane location of the cysteine residues. The alternating accessibility of positions Ser-144 and Arg-177 established a transmembrane spanning domain (TM-5) between these two positions with an N 3 C polarity from the cytoplasm to periplasm. Therefore, Asp-157 was localized to TM-5.
Localization of Asp-157 to the Hydrophobic Core of TM5-We further determined whether Asp-157 is located in the membrane-embedded region of TM-5 by fine topological mapping. The boundary of the hydrophobic core of TM-5 was probed by examining FM reactivities of a set of YiiP variants, each containing a single reactive cysteine residue at one of the following positions: 150, 153, 155, 157, 171, 172, and 174. All these mutants are active Zn 2ϩ /Cd 2ϩ transporters (data not shown). FM  ). Thus, the discontinuous pattern of FM labeling in the membrane was due to neither a lack of FM reactivity for the three central positions nor different levels of protein expression but to a change of local environments surrounding the cysteine residues as TM-5 traversed the plasma membrane through a membrane-embedded region. Accordingly, the region from position 155 to 171 was localized to the hydrophobic core of TM-5, within which is located Asp-157 close to the cytoplasmic boundary (Fig. 4B).
MPB Labeling to D157C-We also evaluated the accessibility of six TM-5 positions to MPB labeling. As shown in Fig. 4C   D157C as shown in Fig. 4A, suggesting that the maleimide group in a more flexible MPB might reach a buried D157C that was not accessible to the maleimide group directly attached to a rigid and charged fluorescein moiety of FM (Fig. 3A). The observed partial MPB labeling to D157C appeared to be position-specific because MPB labeling to two neighboring cysteine residues at positions 155 and 171 was not detectable.
Selective Metal Binding to D157C-Because YiiP D157C exhibited intact Zn 2ϩ /Cd 2ϩ binding and full Zn 2ϩ /Cd 2ϩ transport activity, the selectivity and binding affinity of metal binding to Cys-157 may reflect metal binding to the native Asp-157. Metal binding to Cys-157 was indicated by metal protection of Cys-157 from MPB-labeling, because metal binding to a membrane-embedded Cys-157 is expected to incur steric hindrance to MPB labeling. YiiP D157C/C287S in the membrane was exposed to MPB labeling with a brief sonication in the presence of one of the following divalent cations: Zn 2ϩ , Cd 2ϩ , Hg 2ϩ , Mg 2ϩ , Ca 2ϩ , Mn 2ϩ , Fe 2ϩ , Ni 2ϩ and Co 2ϩ , each at a concentration of 0.2 mM. The MPBtreated YiiP D157C/C287S was then purified and subjected to Western blot analysis. Among the nine divalent metals tested, only Zn 2ϩ and Cd 2ϩ appeared to abolish the corresponding Western blot signals (Fig. 5A,  lower panel), whereas Coomassie Blue staining of a duplicate gel indicated that approximately an equal amount of protein was loaded to each lane (Fig. 5A, upper panel). Additional experiments were carried out to examine the concentration dependence of Zn 2ϩ or Cd 2ϩ protection. MPB labeling was carried out with sonicated cells in the presence of Zn 2ϩ or Cd 2ϩ with a concentration ranging from 0 to 128 M as indicated. The intensity of MPB labeling began to decrease at 16 M for Zn 2ϩ (Fig. 5B) and 32 M for Cd 2ϩ (Fig. 5C), and then rapidly approached a background level at 64 M, indicating that the equilibrium binding of both metal ions could effectively protect the irreversible MPB labeling at 32-64 M, at which Zn 2ϩ /Cd 2ϩ binding to Cys-157 must have approached saturation. Thus, the binding affinity of Zn 2ϩ /Cd 2ϩ to Cys-157 might be in a micromolar range.
Hg 2ϩ and Fe 2ϩ Binding to YiiP-It was unexpected that Hg 2ϩ and Fe 2ϩ did not protect MPB labeling to Cys-157, because the cysteine thiolate is a strong coordination group for both metal ions. To directly evaluate Hg 2ϩ or Fe 2ϩ binding to Asp-157, calorimetric titrations of YiiP and YiiP D157A were carried out with FeCl 2 (Fig. 6A) or HgCl 2 (Fig.  6B). The resultant Fe 2ϩ and Hg 2ϩ binding isotherms were fitted with a two-site binding model. Between YiiP and YiiP D157A , binding stoichiometries, affinities, and binding enthalpy changes were found nearly identical (TABLE TWO). Thus, a D157A mutation altered neither Fe 2ϩ nor Hg 2ϩ binding to Asp-157, in contrast to the loss of one Zn 2ϩ -or Cd 2ϩ -binding site caused by the same mutation. This result suggests that Asp-157 is a highly specific coordination residue for Zn 2ϩ and Cd 2ϩ but not for Hg 2ϩ and Fe 2ϩ .

DISCUSSION
The experiments described here used calorimetric titrations and stopped-flow spectrofluorometry to examine directly the metal binding energetics and transport kinetics. By correlating these biophysical measurements with site-directed mutagenesis, we demonstrated a straightforward dependence of metal binding and transport on the residue type at position 157 of YiiP. Replacing an aspartate with a cysteine residue at position 157 resulted in intact Zn 2ϩ /Cd 2ϩ binding as well as full Zn 2ϩ /Cd 2ϩ transport activity, whereas an alanine substitution caused a complete disruption of Zn 2ϩ /Cd 2ϩ binding, accompanied by impeding Zn 2ϩ /Cd 2ϩ transport activity. This side chain dependence could be explained by coordination chemistry of Zn 2ϩ and Cd 2ϩ , both are isomorphous group 12 metal ions that can bind to a combination of carboxylate, thiolate, and imidazole groups to form polyhedral coordination complexes. Common to aspartate and cysteine residues is the ability of providing a Zn 2ϩ /Cd 2ϩ coordination ligand. ITC measurements of Zn 2ϩ /Cd 2ϩ binding to YiiP Asp-157 and YiiP D157C showed that substituting the Asp oxygen donor with a Cys sulfur donor caused only marginal changes to Zn 2ϩ and to a less degree to Cd 2ϩ binding (TABLE  ONE). It appears that Asp-157 and Cys-157 are largely interchangeable, likely due to the similar space filling volume of both residues. Further supporting evidence for a direct Zn 2ϩ /Cd 2ϩ -Asp-157 coordination comes from the observation that replacing Asp-157 with a noncoordination Ala residue completely disrupted a Cd 2ϩ /Zn 2ϩ site. Comparing Cd 2ϩ binding isotherms of YiiP Asp-157 , YiiP D157C , and YiiP D157A suggests that a D157A mutation disrupts site 2 of Cd 2ϩ binding while leaving site 1 unchanged. Thus, site 2 can be assigned to position 157 with an equivalent K d value of 0.91 M for an aspartate or cysteine residue. The Cd 2ϩ binding affinity obtained by direct ITC measurements agrees well with the micromolar range binding affinity estimated based on Cd 2ϩ protection against MPB labeling to YiiP D157C . A similar Zn 2ϩ binding affinity is suggested by the Zn 2ϩ protection analysis. It is noted that the affinity of Zn 2ϩ /Cd 2ϩ binding to Asp-157 is several orders of magnitude lower than Zn 2ϩ /Cd 2ϩ affinities found in many zinc metalloproteins (22). Zinc exchanges from those metalloproteins in general occur over a period of hours or days. Exchange rates of this order are not suitable for a more rapid on-off binding reaction required to keep the flow of a zinc ion when it is transported. Thus, a lower Zn 2ϩ /Cd 2ϩ affinity is anticipated to reconcile metal ion mobility in YiiP. In this regard, the Asp 3 Cys functional convertibility at position 157 may reflect a loose metal coordination interaction resulting from some side chain flexibility within the binding site. Likewise, the Asp-157-binding site is sufficiently flexible to accommodate various ionic radii from 0.74 Å for Zn 2ϩ to 0.97 Å for Cd 2ϩ . The correlation between Zn 2ϩ /Cd 2ϩ binding and transport distinguishes Asp-157 from a regular metal coordination residue. Besides supplying an oxygen donor for coordination binding, Asp-157 also acts as a pivotal structural component that couples metal binding to a transport process. Thus, the local chemical environment surrounding Asp-157 may contribute to binding-transport coupling. Asp-157 was localized to TM-5 by coarse topological mapping using a set of cysteine substitution mutations introduced such that each pair of cysteine residues was positioned in two solvent-accessible regions flanking each of the six putative TMs in YiiP. Thiol-specific labeling using impermeant probes MPB and FM revealed an alternating pattern of thiol reactivities when intact cells were probed, indicating that the polypeptide chain of YiiP traverses the membrane six times with both N and C termini in the cytoplasm. A fine topological mapping then focused on TM-5 by using cysteine-scanning mutagenesis to probe the positional dependence of cysteine thiol reactivities toward a water-soluble probe, FM, which reacts more readily with solvent-exposed cysteines than with those embedded in the membrane. A central region of low and two peripheral regions of higher FM reactivities were identified with two sharp boundaries occurring between residues 153-155 and 171-172. The lower FM reactivities could be attributed to inaccessibility (23) and/or elevated pK a of cysteine residues (24) in a hydrophobic environment. Thus Asp-157 was localized to a stretch of membrane-embedded residues. The mechanistic implications of a membrane-embedded Asp-157 are 2-fold. 1) The transport of metal ions is thought to be associated with a global protein conformational change originating from the metal-binding site. The hydrophobic surroundings of Asp-157 can greatly strengthen the Asp-157-Zn 2ϩ /Cd 2ϩ interaction and facilitate electrostatic propagation of a local conformational change through a hydrophobic medium. 2) Zn 2ϩ transport in YiiP is coupled to antiport of proton(s). The localization of Asp-157 to a hydrophobic environment raises the possibility that the pK a of this membrane-embedded aspartate may be sufficiently elevated to remain protonated in the absence of Zn 2ϩ /Cd 2ϩ binding. Upon Zn 2ϩ /Cd 2ϩ binding, Asp-157 may be deprotonated to trigger Zn 2ϩ /Cd 2ϩ transport with a separation of the released proton moving in an opposite direction. In agreement with this speculation, our recent ITC study indicated that binding one Cd 2ϩ to YiiP yielded 1.23 protons (15). Further experiments are underway to investigate the involvement of Asp-157 in binding-deprotonation coupling.
CDF transporters were initially identified as Zn 2ϩ /Cd 2ϩ /Co 2ϩ pumps (25,26), and they subsequently were shown to transport Cd 2ϩ , Mn 2ϩ , Ni 2ϩ , and Fe 2ϩ . Categorization of CDF proteins based on multiple sequence alignment suggested three distinct subgroups, each apparently associated with a broad range of substrate specificities but with different metal preferences (3). The E. coli CDF transporters ZitB and YiiP belong to group 2 and group 3, respectively. ZitB functions as a Zn 2ϩ /Cd 2ϩ efflux pump, whereas the substrate specificity of YiiP remains obscure. YiiP (FieF) was recently implicated in a role of iron efflux pump based on the observations that both zinc and iron induced yiiP transcription and that overexpression of YiiP led to a loss of cytosolic iron content in E. coli and an increase of iron tolerance in a ⌬fur strain. However, only accumulation of Zn 2ϩ but not Fe 2ϩ in everted membrane vesicles was established, although Fe 2ϩ transport was observed in reconstituted YiiP proteoliposomes (12). In the present study, both MPB protection analysis and direct ITC measurement indicated that Fe 2ϩ did not bind to Asp-157 that was localized to the active site of the Zn 2ϩ /Cd 2ϩ transporter. Nevertheless, Fe 2ϩ calorimetric titrations suggested the presence of two independent Fe 2ϩ -binding sites in YiiP, both with modest binding affinities (2.2 and 0.22 M). The functional connection of these putative Fe 2ϩ -binding sites to Fe 2ϩ transport has yet to be established.
Contrary to the general expectation that YiiP is a broad range metal ion transporter, metal binding to Asp-157 is highly specific. Calorimetric measurements of metal binding to YiiP Asp-157 and YiiP D157A showed that a D157A mutation disrupted Zn 2ϩ or Cd 2ϩ binding to Asp-157 but caused no detectable change to Fe 2ϩ or Hg 2ϩ binding. Fe 2ϩ and Hg 2ϩ represent neighboring metal ions in the same transition period and in the same element group, respectively. The selectivity of Asp-157 for Zn 2ϩ /Cd 2ϩ over Fe 2ϩ and Hg 2ϩ is consistent with the selectivity of Cys-157 for Zn 2ϩ /Cd 2ϩ over Fe 2ϩ and Hg 2ϩ , as well as Co 2ϩ , Ni 2ϩ , Mn 2ϩ , Ca 2ϩ , and Mg 2ϩ , as suggested by metal binding protection of Cys-157 from MPB labeling. The lack of Hg 2ϩ protection is noteworthy because the cysteine thiolate is a very strong Hg 2ϩ donor group. It is difficult to see how selection of Zn 2ϩ /Cd 2ϩ by Asp-157 and Cys-157 is so specifically based on chemical consideration for metal ions, because the electrostatic binding by Zn 2ϩ /Cd 2ϩ is a property shared with Mg 2ϩ and Ca 2ϩ , whereas the Lewis acid strength, and thus the donor group preference of Zn 2ϩ /Cd 2ϩ , is within the same range shared by Co 2ϩ , Ni 2ϩ , Fe 2ϩ , and Mn 2ϩ . The mechanism of metal selectivity may be better appreciated in terms of coordination chemistry in the chemical context of the immediate binding site neighborhood. To a first approximation, the hydrophobic surroundings of Asp-157 may entail steric and electrostatic hindrances as suggested by the observation that Cys-157 was only partially accessible to thiol-specific modification by a maleimide group tethered to a flexible and neutral MPB as opposed to a rigid and charged FM molecule.
Previous ITC analyses of metal binding to YiiP suggested the presence of at least one mutually competitive binding site common to Zn 2ϩ , Cd 2ϩ , and Hg 2ϩ and a set of noncompetitive binding sites (15). In the present study, we physically assigned Asp-157 to one of the noncompetitive binding sites, because metal binding to Asp-157 is specific to Zn 2ϩ and Cd 2ϩ but not to Hg 2ϩ . After removing the Asp-157 site by a D157A mutation, a reduced Cd 2ϩ binding isotherm could be fitted with one set of independent binding sites with a binding stoichiometry of 1.5, suggesting that one or two additional Cd 2ϩ -binding sites are present in YiiP D157A . In light of our recent finding that YiiP is a homodimer (27), the 1.5 stoichiometric equivalence of Cd 2ϩ binding may be interpreted as two independent Cd 2ϩ -binding sites, one is located in each subunit and the other at the dimeric subunit interface. The molecular identities of these additional Cd 2ϩ -binding sites and their roles in metal transport have yet to be determined. The stoichiometries of Zn 2ϩ and Cd 2ϩ binding to Asp-157 are both close to unity, corresponding to two independent Asp-157-binding sites per YiiP homodimer. Because Zn 2ϩ /Cd 2ϩ binding to Asp-157 is directly linked to Zn 2ϩ /Cd 2ϩ transport, transport of Zn 2ϩ /Cd 2ϩ in a YiiP dimer may occur in two independent translocation pathways, each located in the center of a YiiP monomer. Alternatively, two Cd 2ϩ -binding sites may be located in a common translocation pathway at the dimer interface, thus transport of two Cd 2ϩ ions may occur in a shared translocation pathway. Further structural analysis of YiiP is underway to test these hypotheses.