Coordination of Zn2+ (and Cd2+) by Prokaryotic Metallothionein

In mammalian metallothionein Zn2+ is exclusively coordinated to Cys-thiolate to form clusters in which the metal is thermodynamically stable but also kinetically labile. By contrast, little is known about coordination to prokaryotic metallothionein, SmtA. 3 nmol of Zn2+nmol−1 SmtA were displaced by 8 nmol ofp-(hydroxymercuri)phenylsulfonate implicating eight of the nine Cys in the coordination of three metal ions. None of the Zn2+ associated with SmtA was accessible to 4-(2-pyridylazo)resorcinol prior to the addition ofp-(hydroxymercuri)phenylsulfonate. An unusual feature of SmtA is the presence of three His residues, and we have investigated whether these contribute to metal coordination. Less Zn2+was associated with purified SmtA(H40R/H49R/H55R), in which all three His residues were substituted with Arg, and approximately one equivalent of Zn2+ was immediately accessible to 4-(2-pyridylazo)resorcinol. Following incubation of SmtA with111Cd, three 111Cd resonances were detected, two in a range expected for CdS4 and the third indicative of either CdNS3 or CdN2S2coordination. Two-dimensional TOCSY 1H NMR and111Cd-edited 1H NMR showed two His residues bound to 111Cd, confirming CdN2S2coordination. The pH of half-dissociation of Zn2+ increased from 4.05 for SmtA to 5.37 for SmtA(H40R/H49R/H55R). Equivalent values for single His mutants SmtA(H40R), SmtA(H49R), and SmtA(H55R) were 4.62, 4.48, and 3.81, respectively, revealing that conversion of His40 or His49 to Arg impairs Zn2+binding at the CdN2S2 and CdS4sites. Only approximately two equivalents of Zn2+ were associated with purified SmtA(H49R). The appearance of a fourth111Cd resonance at lower pH suggests that an alternative CdN2S2 site also exists.

MTs 1 bind metals of the copper and zinc triads, typically in metal thiolate clusters (reviewed in Ref. 1). Indeed, the presence of such clusters is often cited as a defining characteristic of MTs. In mammalian MTI and MTII, seven Zn 2ϩ (or Cd 2ϩ ) ions are tetrahedrally coordinated to thiolate ligands in two distinct Zn 3 S 9 and Zn 4 S 11 clusters. In comparison, little is known about metal coordination by bacterial MTs. Electronic absorption spectra suggested the presence of (some) metal thiolate coordination in MT purified from Synechococcus sp. (2), whereas the 113 Cd NMR spectrum of a low M r MT-like protein from Cd 2ϩ -resistant Pseudomonas putida was suggestive of Cys-thiolate and His-imidazole coordination (3). The smtA gene from the cyanobacterium Synechococcus PCC 7942 encodes an MT that contains three His residues (4), a feature uncommon among eukaryotic MTs. The possibility that these His residues may be involved in metal coordination has never been investigated. An atypical (for an MT) coordination chemistry could be significant in view of the unusually high affinity of SmtA for Zn 2ϩ (5) and the exclusive role of SmtA in the sequestration/ metabolism of Zn 2ϩ (and detoxification of Cd 2ϩ ) but not copper ions (6).
A comparison of the pH at which 50% of metal ions dissociate is a criterion that has been used to distinguish MTs from other metal binding proteins. The pH of half-dissociation of Zn 2ϩ from equine renal MT has been estimated to be 4.50 (7,8), whereas values of 4.10 and 4.50 for a GST-SmtA fusion protein and equine renal MT, respectively, were obtained in a comparative study (5). This implies that SmtA has a higher affinity than other MTs for Zn 2ϩ , whereas metal displacement curves for Cd 2ϩ and copper ions have indicated that SmtA has a lower affinity than equine renal MT for these ions (5). Consistent with these observations, expression from the smtA operatorpromoter is maximally induced by Zn 2ϩ , in comparison with other metals at maximum permissive concentrations (4). Furthermore, mutants deficient in SmtA are hypersensitive to Zn 2ϩ (and to some extent Cd 2ϩ ) but have normal tolerance to copper ions (6). In contrast, yeast mutants deficient in the MT gene CUP1 are hypersensitive to copper ions but not Zn 2ϩ (9), whereas transgenic animals in which MTI and MTII genes are disrupted are hypersensitive to hepatic poisoning by Cd 2ϩ (10).
Here we describe analyses of metal coordination by recombinant SmtA and mutants thereof to determine the number of metal ions bound and the number of Cys residues involved. Most importantly, the results of experiments that implicate His-imidazole groups in metal binding by SmtA are reported.

EXPERIMENTAL PROCEDURES
Production and Purification of Recombinant Polypeptides-DNA restriction and modification enzymes were supplied by New England Biolabs Inc., Taq DNA polymerase was supplied by Life Technologies, Inc., and other reagents were purchased from Sigma Chemical Co. All generated plasmid constructs were checked by sequence analysis as described previously (4), and reaction products analyzed using an Applied Biosystems 370A DNA sequence analyzer.
A mutant of smtA was generated in which all three His codons were converted to Arg codons. A three-stage PCR reaction was performed (to minimize mispriming) using plasmid pJHNR49 as template DNA with primers P1 and P3 (5Ј-CCGGAATTCTGATTAGCCGCGGCAGTTACA-GCCGGTGCGGCCGCAGCCTTTGCTACCACCGGTGCGGCCATCGG-C-3Ј, designed to convert codons 40, 49, and 55 from CAC to CGC): Stage 1, three cycles of denaturation (95°C, 1 min) and annealing (50°C, 1 min) were performed with a 50-l reaction containing pJHNR49 (100 ng) and P3 (2 mM) in PCR buffer; stage 2, the reaction volume was increased to 90 l with the addition of Taq DNA polymerase (5 units) and dNTPs (0.22 mM) in PCR buffer, and three cycles of denaturation (94°C, 1.5 min), annealing (45°C, 1.5 min), and extension (73°C, 2 min) were performed; stage 3, the reaction volume was increased to 100 l with the addition of P1 (1 mM) in PCR buffer and 12 further cycles of PCR performed (as for stage 2). The PCR amplification product containing mutated smtA was digested with BamHI and EcoRI and cloned into the BamHI/EcoRI site of pGEX-3X to create pMDNR1.2.
Recombinant fusion proteins were expressed in Escherichia coli (JM101) grown in the presence of 0.5 mM Zn 2ϩ and purified as described previously (11). Recombinant proteins and three residues of GST (Gly, Ile, and Leu), were released from glutathione-Sepharose-bound GST by incubation with factor Xa (Amersham Pharmacia Biotech). Proteins were resolved on 15% SDS-polyacrylamide gels and visualized with Coomassie Brilliant Blue. An aliquot of SmtA was hydrolyzed and analyzed for amino acid composition (Alta Bioscience, University of Birmingham) to allow calibration of colorimetric estimations of SmtA. Purified SmtA was concentrated (to 0.9 mM) for NMR spectroscopy using Centriprep-3 concentrators (Amicon) according to the manufacturer's protocols.
Metal Binding Studies-Purified recombinant proteins (in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM CaCl 2 ) were incubated at room temperature for 1 h with 353.52 kBq 65 Zn before fractionation on Sephadex G-25 equilibrated to pH 7.0 (in 0.05 M KH 2 PO 4 , 0.029 M NaOH). Fractions (0.5 ml) were collected, and aliquots (0.1 ml) were analyzed for protein content and radioactivity. Protein-containing fractions were pooled, and aliquots (625 l) were incubated at room temperature for 1 h with 1.875 ml of 0.05 M KH phthalate/HCl (pH 2-5) or 0.05 M KH 2 PO 4 /NaOH (pH 5-7), followed by fractionation on Sephadex G-25 equilibrated with the same buffers. Fractions (0.5 ml) were collected, and aliquots (0.1 ml) were again analyzed for protein content and radioactivity.
The amount of Zn 2ϩ bound to 1 nmol of recombinant protein (in 0.1 M sodium phosphate buffer, pH 7.0) was determined by addition of 0.67 mM PAR and incubation for 5 min with increasing amounts (1 nmol increments) of PMPS. The metallochromic indicator PAR is known to generate colored chelate compounds with Zn 2ϩ (12). Metal ion release was monitored by the increase in absorbance at 492 nm, and the values were calibrated by reacting known amounts of Zn 2ϩ with PAR.
NMR Spectroscopy-Samples were prepared by adding 4 mol eq of 111 Cd (as a 50 mM solution, prepared by dissolving 111 CdO in a minimum amount of HCl and diluted with D 2 O) dropwise to 0.9 mM SmtA (in 50 mM Tris-HCl, pH 8.4) and equilibrated for Ͼ12 h. Excess 111 Cd was removed by ultracentrifugation (Amicon) and washed four times with 50 mM Tris-d 11 (pH 8.4) for ϳ2 h. 111 Cd and 1 H{ 111 Cd} NMR spectra were obtained at 298 K on a Bruker DMX 500 MHz NMR spectrometer operating at 106.04 and 500.13 MHz for 111 Cd and 1 H, respectively. Typically parameters for 111 Cd were: spectral width, 240 ppm, centered at 620 ppm (relative to Cd(ClO 4 ) 2 ), with 16, 384 data points; pulse width, 8 s (60°pulse); pulse delay, 0.6 s. Gated CPD 1 H decoupling (during acquisition time) was used to avoid negative NOEs. The total acquisition time for each spectrum was 6 -8 h. An exponential function (equivalent to a line broadening of ϳ10 -20 Hz) was used prior to Fourier transformation. Shifts are referenced to external 0.1 M Cd(ClO 4 ) 2 (0 ppm). 111 Cd-edited 1 H NMR spectra (4,096 data points) were obtained by using the first increment of a two-dimensional [ 1 H, 111 Cd] heteronuclear multiple quantum coherence experiment (13), and the coherence transfer was selected by pulsed field gradients (14). Data were acquired with different mixing times (1/2J[ 111 Cd, 1 H]) because of markedly different vicinal coupling constants ( 3 J[ 111 Cd, 1 H]) (15). 111 Cd GARP decoupling (centered at 620 ppm) (16) was carried out during the acquisition time to remove 111 Cd coupling. The 1 H spectral width was 8 ppm with the H 2 O resonance in the center of the spectrum. Unshifted Gaussian functions were used for processing. Two-dimensional TOCSY spectra were acquired using a spin-lock time (MLEV-17) of 60 ms with 2,048 data points in the second frequency domain and 256 increments in the first frequency domain. Data were zero-filled to 2,048 ϫ 1,024 data points. Typically 32 transients were acquired for each increment.

Production and Purification of Recombinant Polypeptides-
Extracts from E. coli cells containing plasmids pMDNR1.1 and pMDNR1.2 were fractionated on glutathione-Sepharose 4B, and proteins of ϳ35 kDa, corresponding to the predicted size of GST-SmtA, were detected in fractions eluted with buffer containing 5 mM glutathione (data not shown). It was noted that GST-SmtA(H40R/H49R/H55R) migrated slightly faster than GST-SmtA, which does not correlate with a difference in molecular mass. Following overnight incubation of recombinant proteins immobilized on glutathione-Sepharose 4B with factor Xa, smaller proteins of ϳ6.5 kDa (corresponding to the predicted size of SmtA) were released in factor Xa cleavage buffer (devoid of glutathione).
The pH Stability of Zn 2ϩ Binding to SmtA Is Reduced in a His Mutant-In vitro incubation of either SmtA or SmtA(H40R/H49R/H55R) for 1 h with 65 Zn gave rise to radioactivity and protein that co-chromatographed on Sephadex G-25 (Fig. 1, A and B). At pH 4.28 65 Zn remained associated with SmtA but dissociated from SmtA(H40R/H49R/H55R) (Fig.  1, C and D). Fig. 1E shows the proportion of 65 Zn associated with either protein as a function of pH. The mean estimated pH values (and standard deviations) of half-dissociation of Zn 2ϩ from SmtA and SmtA(H40R/H49R/H55R) from three independent experiments were 4.05 Ϯ 0.09 and 5.37 Ϯ 0.04, respectively. The former value is in good agreement with that reported previously (5), whereas the latter implies that substitution of His residues with Arg reduces the affinity for Zn 2ϩ .
SmtA Binds 3 mol of Zn 2ϩ mol Ϫ1 via Eight Cys Residues-Upon Cys modification of 1 nmol of SmtA in the presence of PAR, an increase in A 492 (⌬A 492 ) was observed up to the addition of 8 nmol of PMPS ( Fig. 2A). In each of three experiments using independent preparations of recombinant SmtA, no further increase in A 492 was detected upon addition of more than 8 nmol of PMPS ( Fig. 2A). This implies that only eight of the nine Cys residues are involved in Zn 2ϩ coordination and that the remaining Cys residue is the least accessible to PMPS. Use of calibration curves for Zn 2ϩ reacting with PAR gave estimates of 3.06 Ϯ 0.37 nmol of Zn 2ϩ released nmol Ϫ1 of SmtA following the addition of saturating amounts of PMPS. This implies that each molecule of SmtA coordinates three Zn 2ϩ ions.
One mol of Zn 2ϩ mol Ϫ1 Protein Is Lost, and a Second Is Accessible to PAR, in a Triple His Mutant of SmtA-When SmtA(H40R/H49R/H55R) was incubated with PAR, an immediate color change (prior to the addition of PMPS) was observed. Thus, unlike Zn 2ϩ associated with SmtA, a proportion of the Zn 2ϩ associated with SmtA(H40R/H49R/H55R) is immediately accessible to PAR with no requirement for Cys modification. It is known that PAR, a metal chelator, can remove Zn 2ϩ from some proteins (17). The amount of Zn 2ϩ removed from the mutant protein by PAR was estimated to be 1.22 Ϯ 0.21 nmol Zn 2ϩ nmol Ϫ1 protein, a value close to one (Fig. 2B). These data suggest that the coordination of (at least) one Zn 2ϩ ion is sufficiently weakened by the substitution of all three His residues by Arg such that PAR can now directly compete for this metal ion. The amount of metal subsequently displaced by PMPS was variable, possibly reflecting some variable loss of metal during purification, but the total stoichiometry is less than three and does not significantly deviate from 2 nmol Zn 2ϩ nmol Ϫ1 protein.
His 40 and His 49 but Not His 55 Mutants Have a Reduced Apparent Affinity for Zn 2ϩ -The pH lability (Fig. 1) and accessibility to PAR (Fig. 2) of Zn 2ϩ associated with SmtA(H40R/ H49R/H55R) implicates His residues in metal coordination. Mutants were therefore generated in which each His residue was individually substituted by Arg to define which of the three residues contribute(s) toward metal binding. Following overnight incubation with factor Xa, proteins of ϳ6.5 kDa (corresponding to the predicted size of SmtA) were released from glutathione-Sepharose 4B-immobilized extracts from E. coli containing plasmid pMDNR1.3, pMDNR1.4, or pMDNR1.5. Purified SmtA(H40R), SmtA(H49R), or SmtA(H55R) was incubated with 65 Zn, fractionated on Sephadex G-25, and then incubated and re-chromatographed in buffers of differing pH. The amount of metal displaced by PMPS did not significantly deviate from 3 nmol of Zn 2ϩ nmol Ϫ1 protein for either SmtA(H40R) or SmtA(H55R), whereas the stoichiometry was close to 2 nmol of Zn 2ϩ nmol Ϫ1 SmtA(H49R) (Fig. 2B). An increase in A 492 was observed with the addition of up to 6 nmol of PMPS nmol Ϫ1 of SmtA(H49R). In each of three experiments using independent preparations of recombinant SmtA(H49R), no further increase in A 492 was detected upon the addition of more than 6 nmol of PMPS. This implies that only six Cys residues coordinate Zn 2ϩ in SmtA(H49R) and that the other three Cys are less accessible to PMPS. 111 Cd and 1 H{ 111 Cd} NMR Spectra Show Cys-thiolate and His-imidazole Metal Coordination-The pH stability and accessibility to PAR of Zn 2ϩ associated with His mutants of SmtA suggest that His-imidazole is involved in metal binding in addition to Cys-thiolate. To investigate the types of residues (especially His) involved in binding, 111 Cd was used to displace Zn 2ϩ from the protein, and both 111 Cd (Fig. 4) and 111 Cd-edited 1 H (Fig. 5) NMR were used to probe the binding sites. Fig. 4 shows the 1 H{ 111 Cd } NMR spectrum of a solution of SmtA (0.9 mM in 50 mM Tris-d 11 , 90% v/v H 2 O) with resonances in the range expected for 111 Cd bound, in full or in part, by thiolate ligands (18,19). Three major resonances were observed at pH 8.4, two with chemical shifts of 654 and 661 ppm and a third displaced at 572 ppm. At pH 7.6 a fourth resonance appeared at 567 ppm. No 111 Cd resonances were observed from 0 to 500 ppm. Fig. 5 shows 1 H{ 111 Cd} NMR spectra recorded with different mixing times (0.5/ 3 J[ 111 Cd, 1 H]). At very short mixing times, several resonances were observed around 2.7-3.0 ppm. This region is typical of the ␤ proton resonances of Cys residues. Four apparent singlets at 6.65, 7.31, 7.78, and 7.98 ppm appeared in the aromatic region of the 1 H{ 111 Cd} one-dimensional heteronuclear multiple quantum coherence NMR spectrum. These peaks decreased in intensity with decreasing mixing times. A two-dimensional TOCSY 1 H NMR spectrum (with 111 Cd decoupling; mixing time, 65 ms) showed cross-peaks between the peaks at 6.65 and 7.78 ppm, 7.31 and 7.98 ppm, and 7.28 and 8.17 ppm (Fig. 5A), assignable to His C␦H/C⑀H connectivities. It is evident from Fig. 5B that only the two His residues giving rise to the former two sets of cross-peaks and not the latter are coordinated to Cd 2ϩ . The 1 H NMR spectra of Cd 2ϩ -SmtA and Zn 2ϩ -SmtA (data not shown) show very similar well dispersed resonances in the aromatic and NH region (6 to 11 ppm) and low frequency-shifted methyl resonances in the aliphatic region (ϳ0.5 ppm), features that are typical of folded proteins. This suggests that Cd 2ϩ -SmtA and Zn 2ϩ -SmtA proteins have similar structures. DISCUSSION Our results show that SmtA coordinates to three Zn 2ϩ ions via eight Cys residues. Analyses of metal binding to mutants of SmtA in which His residues were substituted by Arg, suggest that at least two His residues (specifically His 40 and His 49 ) are also involved in the coordination of metal ions. 111 Cd and 111 Cd-edited 1 H NMR spectra confirmed that SmtA has at least three distinct metal binding sites. Two metal sites contain exclusively Cys-thiolate ligands, whereas the third contains both Cys-thiolate and His-imidazole ligands. This novel (for an MT) coordination chemistry is of significance in view of (i) the unusually high apparent affinity of SmtA for Zn 2ϩ and (ii) observations that SmtA is exquisitely adapted to roles in the intracellular handling of Zn 2ϩ rather than other metal ions.  111 Cd NMR has been used previously to probe metal coordination by eukaryotic MTs (19). The 111 Cd (or 113 Cd) NMR shifts of the CdS 4 clusters of eukaryotic MTs usually lie within the range of 600 -700 ppm (18 -20), whereas the shifts of isolated CdS 4 centers are usually at the high frequency end of this range (680 -750 ppm) (21,22). The 111 Cd NMR spectrum of 111 Cd-SmtA contains two high frequency peaks at 654 and 661 ppm, which are both in the chemical shift region normally associated with the CdS 4 clusters of eukaryotic MT. These two 111 Cd NMR peaks for Cd 2ϩ -SmtA are relatively sharp. Unlike the spectra of eukaryotic MT, there is no evidence for Cd 2ϩ -Cd 2ϩ coupling (ϳ29 -48 Hz for MT) (19) that could suggest that clusters are not present in Cd 2ϩ -SmtA. Alternatively the couplings may be small or the bridging ligands may be involved in fluxional processes, for example involving a dynamic equilibrium between structures (ii) and (iii) in Fig. 6A.
The third peak in the 111 Cd NMR spectrum lies at 572 ppm, outside the normal range for CdS 4 centers. Shifts of 630 -660 ppm have been documented for CdNS 3 centers composed of Cys and a single His residue, and even lower shifts in the region of 400 ppm observed for centers that include two nitrogen atoms (19,23,24). The third peak for Cd 2ϩ -SmtA is therefore suggestive of either CdNS 3 or CdN 2 S 2 coordination. Two-dimensional TOCSY 1 H NMR and 111 Cd-edited 1 H NMR data reveal that two His residues are bound to 111 Cd, and therefore CdN 2 S 2 coordination is inferred for the third site. In the only previous report of 113 Cd NMR of a prokaryotic MT-like protein isolated from P. putida, Higham et al. (3) reported 113 Cd NMR shifts of 615, 604, 483, and 476 ppm. These represent two groups of peaks separated by ϳ100 ppm, somewhat analogous to the present work.
The total number of Cys (nine) and His (three) residues in SmtA is sufficient for tetrahedral coordination of Zn 2ϩ in three independent sites. However, the requirement for only eight equivalents of PMPS to displace all Zn 2ϩ (Fig. 2) indicates that one Cys was not involved in metal binding. Furthermore, both the pH stability of Zn 2ϩ -SmtA(H55R) (Fig. 3) and the 111 Cdedited 1 H NMR spectrum (Fig. 5) imply that His 55 does not stabilize metal binding. Thus, only configurations for metal coordination that involve eight Cys and two His residues are shown (Fig. 6A).
The reduced stoichiometry of SmtA(H49R) (Fig. 2B) suggests that the His coordinated ion is lost from this protein and hence that His 49 is obligatory for the CdN 2 S 2 site, whereas His 40 (or His 55 ) is optional. Only six equivalents of PMPS are required to displace the Zn 2ϩ , which remains associated with SmtA(H49R) supporting models (ii) and (iii) (Fig. 6A). Although only one metal ion is coordinated to His residues at pH 8.4 (Fig. 4), all metal ions are more readily displaced at low pH upon substitution of all three His with Arg (Fig. 1). Furthermore, the two Zn 2ϩ that remain associated with SmtA(H49R) are also more readily displaced at low pH (Fig. 3), showing interaction between binding at the CdN 2 S 2 and CdS 4 sites and/or showing that residual metal coordination is abnormal in the mutant proteins because of abnormal folding. NMR analyses of the mutant proteins would be required to resolve this.
Future investigations of metal coordination by SmtA should not only consider the basis of its unusual specificity in the detoxification of Zn 2ϩ but also any specificity in the release of Zn 2ϩ to apo-proteins. The nature of Zn 2ϩ coordination by mammalian MT not only creates high affinity binding sites but also ones that are highly labile allowing rapid metal exchange (1). There is evidence to support a role for mammalian MT in the donation of Zn 2ϩ to Zn 2ϩ -binding proteins (25)(26)(27)(28)(29)(30). It has been hypothesized that SmtA may donate Zn 2ϩ to DNA primase (DnaG), which in Synechococcus PCC 7942 contains two Zn 2ϩ binding sites and is encoded by a gene that is located next to the smtA gene (31). Similarities between metal binding domains provided a clue to the existence of interaction between a copper chaperone and its receptor (32). Alignments between SmtA and the metal binding domains of DnaG do show similarities in the spacing of some His and Cys residues (Fig. 6, B and C). Both alignments include His 55 . It is formally possible that His 55 and the ninth Cys contribute to transient metal coordination states that are favored during metal transfer.
The 111 Cd-SmtA NMR peak at 572 ppm (Fig. 4) is the most broad resonance, which is possibly a reflection of the kinetic lability of the CdN 2 S 2 site. The pH dependence of the 111 Cd NMR spectrum of Cd 2ϩ -SmtA was investigated only over a limited range in view of the potential instability of the protein and limited availability. However, at pH 7.6 an additional peak appeared at 567 ppm. Because no additional 111 Cd was added to the protein when the pH was lowered, this implies a redistribution of Cd 2ϩ between available sites. This is consistent with flexible coordination modes involving a variable complement of the nine available Cys and three His ligands. Selective protonation of certain residues may encourage the redistribution of metal ions to other centers. We are intrigued by the possibility that such a process could "drive" metal release in vivo.