Structural Determinants of the High Affinity Extracellular Zinc Binding Site on Cav3.2 T-type Calcium Channels*

Cav3.2 T-type channels contain a high affinity metal binding site for trace metals such as copper and zinc. This site is occupied at physiologically relevant concentrations of these metals, leading to decreased channel activity and pain transmission. A histidine at position 191 was recently identified as a critical determinant for both trace metal block of Cav3.2 and modulation by redox agents. His191 is found on the extracellular face of the Cav3.2 channel on the IS3-S4 linker and is not conserved in other Cav3 channels. Mutation of the corresponding residue in Cav3.1 to histidine, Gln172, significantly enhances trace metal inhibition, but not to the level observed in wild-type Cav3.2, implying that other residues also contribute to the metal binding site. The goal of the present study is to identify these other residues using a series of chimeric channels. The key findings of the study are that the metal binding site is composed of a Asp-Gly-His motif in IS3–S4 and a second aspartate residue in IS2. These results suggest that metal binding stabilizes the closed conformation of the voltage-sensor paddle in repeat I, and thereby inhibits channel opening. These studies provide insight into the structure of T-type channels, and identify an extracellular motif that could be targeted for drug development.

Block of voltage-dependent Ca 2ϩ channels by metal ions has been extensively studied as it provides insight into structural changes in channel conformations during gating, and serves as a pharmacological tool to distinguish the various channel types. Among trace metals, Cd 2ϩ has been shown to selectively block high voltage-activated (HVA) 2 Ca 2ϩ channels. Combined experiments using molecular biology and electrophysiology revealed that the Cd 2ϩ binding site in HVA Ca 2ϩ channels is composed of a Glu-Glu-Glu-Glu (EEEE) motif in the pore loops, thereby providing insights into Ca 2ϩ channel permeation and selectivity (5,6). Pharmacological studies have shown that LVA T-type currents are inhibited by much higher concentrations of Cd 2ϩ than HVA channel currents (7,8). Sensitivity to nickel inhibition of T-type currents varies greatly between cell types, implying the existence of multiple types of T-type channels. Molecular cloning of three T-type channel ␣1 subunits (Ca v 3.1, Ca v 3.2, and Ca v 3.3) allowed a comparison of nickel sensitivities of the three T-type channel isoforms, revealing that only Ca v 3.2 was sensitively inhibited by low micromolar concentrations of nickel (9). All three of the T-type channel isoforms share the Glu-Glu-Asp-Asp (EEDD) motifs in their pore regions corresponding to those of HVA Ca 2ϩ channels. Replacement of the third and fourth aspartate with glutamate residues enhances cadmium block of Ca v 3.1 T-type channels, suggesting that LVA and HVA channels share a similar pore structure (10). Detailed biophysical studies of metal block have also found evidence that LVA and K ϩ channels share an internal activation gate (11), which in the case of Zn 2ϩ binding to Ca v 3.3 (12), can lead to a "foot-in-the-door" block reminiscent of Rb ϩ block of K ϩ channels (13).
Pharmacological studies of cloned T-type channels reconstituted in expression systems have shown that Ca v 3.2 is more sensitively inhibited by not only nickel, but also zinc and copper, relative to either Ca v 3.1 or Ca v 3.3 (9,12,14). Notably, Ca v 3.2 channels are among the most sensitive targets of ion channel targets to zinc block (12). Our recent experiments using chimeric channels between Ca v 3.1 and Ca v 3.2 channels identified His 191 in the extracellular loop connecting S3 and S4 of domain I as a major structural determinant for inhibition of the Ca v 3.2 by nickel, zinc, copper, and redox agents (15)(16)(17).
Metal binding sites in the voltage-sensor paddle have also been reported in Ca v 2.3, Na v 1.2, K v 2.1, and H v 1 channels (18 -23). Two histidine residues in the IS3-IS4 loop of the Ca v 2.3 channel were identified to be critical for the nickel-sensitive inhibition (18). The interaction sites of ␣-scorpion and sea anemone toxins, which cause slowing of fast inactivation, were localized to the S3-S4 loop of domain IV of the Na v 1.2 channel (19,20). The ability of hanatoxin to reduce channel activity and shift channel gating was also localized to the S3-S4 loop of K v 2.1 channels (21). Hanatoxin interaction with the S3-S4 loop of the K v 2.1 channel was thought to stabilize a closed state of the channel thereby reducing channel opening (21,22). We previously proposed that the mechanism of nickel inhibition of Ca v 3.2 channel activity may be similar to the mechanism of hanatoxin inhibition (15). These findings establish that not only pore regions including S5 and S6, but also voltage-sensor regions including S1-S4 and their connecting loops could be potential binding sites for inhibitors.
Although His 191 was identified to be critical for rendering the nickel or zinc inhibition sensitivity to Ca v 3.2 T-type channels (15), reverse introduction of a histidine residue into the corresponding locus (Gln 172 ) of Ca v 3.1 channels only slightly increased trace metal block, suggesting that other residue(s) are involved in metal block of Ca v 3.2 channels in addition to His 191 . Therefore, we investigated additional residue(s) involved in zinc block of Ca v 3.2, focusing on zinc inhibition rather than copper to avoid complications of redox reactions (16). We found that the residues that precede His 191 , Asp 189 , and Gly 190 , were also critical residues for determining the high zinc sensitivity of Ca v 3.2. Additionally, we found an important role of negatively charged residues at the outer portion of the IS2 segment in zinc block of Ca v 3.2. These findings provide the structural basis of the high affinity extracellular metal binding site on Ca v 3.2, providing a novel therapeutic target for the treatment of neuropathic pain (24).

EXPERIMENTAL PROCEDURES
Chemicals-Chemicals were purchased from either Sigma or Amresco (Solon, OH). A zinc-chloride stock solution (100 mM; Sigma) was made in deionized water, and then stored at room temperature. A series of zinc solutions (in M: 0.3, 1, 3, 10, 30, 100, 300, 1000, and 3000) were prepared by diluting the zinc stock solution with 10 mM Ba 2ϩ solution just before experiments and their pH were adjusted to 7.6 if necessary.
Construction of Chimeras between Ca v 3.1 and Ca v 3.2-The serial chimeric channels constructed between the rat Ca v 3.1 (␣ 1G ; GenBank accession number AF027984) and human Ca v 3.2 (␣ 1H ; GenBank accession number AF051946) channels were previously reported (15). Additional chimeric and point mutant channels were made by the same PCR method described previously (15). All PCRs were performed using Pfu DNA polymerase (Vivagen, Seoul, Korea) and amplified fragments were verified by sequencing analysis. The restriction sites were marked by numbers in parentheses by indicating the 5Ј-terminal nucleotide generated by cleavage. Silent and nonsilent mutations for restriction sites used for construction of chimeric channels are indicated by asterisks and crosses, respectively.
Ca Ca v 3.2/ D189A and Ca v 3.2/ D189E -The forward primer to amplify the upper cassettes for D189A and D189E was TAAT-ACGACTCACTATAGGG (T7 promoter) and the reverse primers were TGTGTCCAGCCAACGAGTACTCCATCAT-GCCC and TGTGTCCCTCCAACGAGTACTCCATCATG-CCC, respectively. The forward primers to amplify the lower cassettes were CTCGTTGGCTGGACACAACGTGAGCCTC and CTCGTTGGAGGGACACAACGTGAGCCTC, respectively, and the reverse primer was CAGGATCCGCATGCT-AGG. The upper and lower cassettes were purified using the PCR purification kit and then combined by second-step PCR. Each point mutant channel was constructed by ligating the NotI-and BamHI-digested PCR fragments and BamHI (730, Ca v 3.2)-SalI (4634, Ca v 3.2) fragment into the NotI-(342, Ca v 3.2) and SalI-digested (4635, Ca v 3.2) plasmid Ca v 3.2 pGEM-HEA.
Ca v 3.1/ L171GϩQ172H and Ca v 3.1/ Q172HϩF176L -Plasmid Ca v 3.1/ Q172H was used as a PCR template, and the forward primer used to amplify the upper cassettes for L171GϩQ172H and Q172HϩF176L was TAATACGACTCACTATAGGG (T7 promoter). The reverse primers were GTTGTGTCCGTCCA-GCGAATACTCCAG and TGCGGAGAGGCTGACGTTGT-GCAGGTC, respectively. The forward primers to amplify the lower cassettes were CTGGACGGACACAACGTCAGCTT-CTCC and AACGTCAGCCTCTCCGCAGTCAGGGTC, respectively, and the reverse primer was GAGAAGCTTGCC-AGGGTGCTAGC. The upper and lower cassettes were purified using the PCR purification kit and then combined by second-step PCR. Each point mutant channel was constructed by ligating the ClaI-and HindIII-digested PCR fragments into ClaI-(5Ј-polylinker) and HindIII-digested (1755, Ca v 3.1) plasmid Ca v 3.1 pGEM-HEA.
Ca v 3.1/ D122AϩL171GϩQ172H -Plasmid Ca v 3.1/ L171GϩQ172H was used as a PCR template and the forward and reverse primers to amplify the upper cassettes for D122A ϩ L171G ϩ Q172H were TAATACGACTCACTATAGGG (T7 promoter) and AAAGATGAAGGCATCGAAGGCCTGCAGGAT, respectively. The forward and reverse primers to amplify the lower cassettes were GCCTTCGATGCCTTCATCTTTGCC-TTCTTT and GAGAAGCTTGCCAGGGTGCTAGC, respectively. The upper and lower cassettes were purified using the PCR purification kit and then combined by second-step PCR. Each point mutant channel was constructed by ligating the ClaI-and HindIII-digested PCR fragments into the ClaI-(5Јpolylinker) and HindIII-digested (1755, Ca v 3.1) plasmid Ca v 3.1 pGEM-HEA.
To synthesize capped cRNAs, all cDNAs encoding Ca v 3.1, Ca v 3.2, and chimeric channels were linearized by AflII and in vitro transcribed using T7 RNA polymerase (Ambion, Austin, TX) in accordance with the manufacturer's instructions. The cRNAs were injected into oocytes at concentrations of 10 -50 ng/50 nl using a Drummond Nanoject pipette injector (Parkway, PA) attached to a Narishige micromanipulator (Tokyo, Japan). The SOS solution was changed daily.
Electrophysiology and Data Analysis-Ba 2ϩ currents were measured using a two-electrode voltage clamp amplifier (OC-725C, Warner Instruments, Hamden, CT) between the 3rd and 5th days after cRNA injection. Microelectrodes were pulled using a pipette puller and filled with 3 M KCl. The electrode resistance was 0.5-1.0 megohm. The bath solution contained (in mM) 10 Ba(OH) 2 , 90 NaOH, 1 KOH, and 5 HEPES (pH 7.4 with methanesulfonic acid). The currents were sampled at 5 kHz and low pass-filtered at 1 kHz using the pClamp system (Digidata 1322A and pClamp 8; Molecular Devices, Palo Alto, CA). Data analysis and graphs were obtained with Clampfit software and Prism software (GraphPad, San Diego, CA), respectively. Dose-response curves were fitted using the Hill equation in Prism: B ϭ (1 ϩ IC 50 /(Zn 2ϩ ) n ) Ϫ1 , where B is the normalized block, IC 50 is the concentration of Zn 2ϩ giving halfmaximal inhibition, and n is the Hill coefficient. Data are presented as mean Ϯ S.E. Statistical significance was measured using Student's unpaired t test.
Whole cell patch clamp recordings were obtained at room temperature using an Axopatch 200A amplifier equipped with a CV201A headstage. The amplifier was connected to a computer through a Digidata 1200 A/D converter, and controlled using pCLAMP 9.2 software. Whole cell currents were recorded using the following external solution (in mM): 15 CaCl 2 , 155 tetraethylammonium chloride (TEA-Cl), and 10 HEPES, pH adjusted to 7.4 with TEA-OH. The internal pipette solution contained the following (in mM): 125 CsCl, 10 EGTA, 2 CaCl 2 , 1 MgCl 2 , 4 Mg-ATP, 0.3 Na 3 GTP, and 10 HEPES, pH adjusted to 7.2 with CsOH. Pipettes were made from TW-150-3 Currents were elicited by a test potential to Ϫ20 mV from a holding potential of Ϫ90 mV every 15 s. A-D, representative current traces of Ca v 3.1, Ca v 3.2, Ca v 3.1/ Q172H , and Ca v 3.2/ H191Q before and after cumulative application of concentrations of zinc were superimposed. Scale bars on the x and y axes represent 20 ms and 1 A, respectively. E, dose-response curves of zinc inhibition on Ca v 3.1 (Ⅺ) and Ca v 3.1/ Q172H (f). Currents were normalized to the peak current measured before application of zinc solutions, and the normalized percent inhibition was plotted against zinc concentrations. The smooth curves were obtained from fitting the average data with the Hill equation. Dose-response curves of zinc inhibition on Ca v 3.2 (E) and Ca v 3.2/ H191Q (F) were obtained from analyzing data in a similar way and shown in F. All data are presented as mean Ϯ S.E. (n ϭ 4 -42). The voltage dependence of activation and steady-state inactivation of wild-type and mutant channels are summarized in Table 1. capillary tubing (World Precision Instruments, Inc., Sarasota, FL). Under these solution conditions the pipette resistance was ϳ2.4 megaohms. Access resistance and cell capacitance were calculated using on-line exponential fits to the capacitance transient induced by a 20-mV depolarization (Membrane Test, pCLAMP software). Cell capacitance averaged 10 picofarads. Access resistance averaged 4 megaohms, and was compensated 70% using the series resistance prediction and compensation circuit. Data were filtered at 2 kHz and digitized at 5 kHz for ionic currents or filtered at 10 kHz and digitized at 20 kHz for gating currents. The voltage protocol to measure gating currents included P/Ϫ8 subtraction to remove residual capacitance. Gating currents were integrated (pA ϫ ms), and compared using a paired Student's t test. Due to variability between cells, the data were normalized to the control value.

RESULTS
Inward Ba 2ϩ currents were not detected from control oocytes injected with either H 2 O or 0.1 M KCl, whereas robust inward transient currents were recorded from oocytes injected with cRNA encoding Ca v 3.1, Ca v 3.2, or chimeric channels. We first confirmed the Xenopus oocyte expression system could be used to measure the higher potency of zinc to inhibit Ca v 3.2 over Ca v 3.1 channels, which was originally observed in mammalian cells (14,17). Application of serial zinc solutions demonstrated that Ca v 3.2 currents were sensitively inhibited by low micromolar concentrations of zinc, whereas Ca v 3.1 currents required much higher concentrations (Fig. 1, A and B). The average IC 50 values for the Ca v 3.1 and Ca v 3.2 currents were 82.2 Ϯ 6.9 and 3.0 Ϯ 0.2 M, respectively (n ϭ 13, 42; open symbols in Fig. 1, E and F), being consistent with the previously reported findings performed in HEK-293 cells (14,17). The half-potentials for their activation and steady-state inactivation were not significantly changed by zinc (Table 1). We also repeated our recent finding that the presence of His 191 in the IS3-IS4 loop is critical for determining the zinc-and nickelsensitive inhibition of the Ca v 3.2 channel (15,17). Application of serial zinc solutions inhibited Ca v 3.2/ H191Q currents in a concentration-dependent manner (Fig. 1F, F), and the IC 50 value was 103.8 Ϯ 14.3 M (n ϭ 7). This result indicated that the H191Q mutation greatly diminished the zinc sensitivity of Ca v 3.2, lowering it to the levels observed with Ca v 3.1.
We next tested how reverse mutation of the corresponding glutamine residue to histidine (Q172H) in Ca v 3.1 altered zinc sensitivity of Ca v 3.1. The zinc inhibition profile of the Ca v 3.1/ Q172H mutant revealed an IC 50 value of 42.5 Ϯ 7.6 M (n ϭ 4), indicating that the mutation increased sensitivity by about 2-fold relative to wild-type Ca v 3.1, but still 15-fold less sensitive than Ca v 3.2 (Fig. 1E, f). These results suggested that adoption of other residue(s) into Ca v 3.1/ Q172H were required to gain the zinc-sensitive inhibition to the levels observed with Ca v 3.2 besides His 191 .
To narrow down other regions contributing to the zinc sensitivity of Ca v 3.2, we systematically constructed serial chimeric channels by adopting the adjacent regions(s) of His 191 into either Ca v 3.1 or Ca v 3.1/ Q172H . We first constructed Ca v 3.1/ 3.2:N-IS4 by introducing the region from the amino terminus to the IS4 segment of Ca v 3.2 into the Ca v 3.1. The IC 50 value of Ca v 3.1/ 3.2:N-IS4 was 2.6 Ϯ 0.3 M (n ϭ 6; Fig. 2), which was similar to that of Ca v 3.2. Consistent with our previous report (15), this result strongly suggested that residue(s) determining the zinc sensitivity of Ca v 3.2 were present in the voltage-sensor paddle region (IS1-IS4) of domain I of Ca v 3.2 rather than the pore region. Next, we constructed chimeric channels with smaller substitutions, Ca v 3.1/ 3.2:N-IS12L and Ca v 3.1/ 3.2:IS34L , by transferring the NH 2 terminus to the connecting loop between IS1 and IS2 of Ca v 3.2 and the connecting linker between IS3 and IS4 of Ca v 3.2 into the Ca v 3.1, respectively (Fig. 2). Zinc inhibition profiles of the chimeric channels showed that the IC 50 value of Ca v 3.1/ 3.2:N-IS12L was  Fig. 2). The enhanced sensitivity suggests that residues in IS2 and IS3 that precede the IS3-IS4 loop may participate in zinc inhibition of Ca v 3.2 currents. The reverse single mutation of Q172H increased the zinc sensitivity of Ca v 3.1 by about 2-fold. In contrast, adoption of the IS3-IS4 loop of Ca v 3.2 into the corresponding loop of Ca v 3.1 increased zinc sensitivity 15-fold, suggesting that not only His 191 , but also other residues in the IS3-IS4 loop, participate in the zinc sensitivity of Ca v 3.2. Comparing the amino acid sequences of the IS3-IS4 loops of Ca v 3.1 with those of Ca v 3.2 revealed that Leu 171 , Gln 172 , and Phe 176 in Ca v 3.1 were different from the corresponding residues, Gly 190 , His 191 , and Leu 195 , in the Ca v 3.2 channel (Fig. 3A). To examine whether these residues influence zinc sensitivity, we additionally introduced individual mutations of L171G and F176L into Ca v 3.1/ Q172H . Analysis of their zinc inhibition profiles showed that zinc sensitivity of Ca v 3.1/ Q172HϩF176L was not significantly different from that of Ca v 3.1/ Q172H (Fig. 3, B and C). In contrast, the zinc sensitivity of Ca v 3.1/ L171GϩQ172H was significantly augmented (IC 50 ϭ 6.4 Ϯ 0.6 M, n ϭ 4; p Ͻ 0.001; Fig. 3, B and C), suggesting that the double mutation of L171G as well as Q172H into Ca v 3.1, was critical to the increase in zinc sensitivity of wild-type Ca v 3.1. Consistently, reciprocal mutation of Gly 190 of Ca v 3.2 into leucine strongly reduced the potency of zinc (Ca v 3.2/ G190L ; IC 50 ϭ 47.8 Ϯ 6.5 M, n ϭ 12; Fig.  3, B and C), supporting our hypothesis that glycine as well as histidine residues in the IS3-IS4 loop are crucial elements affecting zinc sensitivity of Ca v 3.2.
Sequence comparison of the IS3-IS4 loops of Ca v 3 isoforms identified an aspartate residue that was commonly present at the preceding position to the identified two critical residues (Gly 190 and His 191 ). Because the carboxyl acid side chains of aspartate could be another potential ligand involved in zinc binding, we tested possible participation of Asp 189 in the zinc sensitivity of Ca v 3.2. Point mutation of D189A into Ca v 3.2 decreased its zinc sensitivity by about 10-fold (IC 50 ϭ 28.2 Ϯ 8.8 M, n ϭ 13; Fig. 3, B and C), whereas point mutation of D189E enhanced its zinc sensitivity by about 3-fold (IC 50 ϭ 1.0 Ϯ 0.1 M, n ϭ 6; p Ͻ 0.001; Fig. 3, B and C). These results indicated that an acidic residue just before the critical Gly 190 and His 191 residues is also required to endow the Ca v 3.2 with high-affinity zinc inhibition. Taken together, these findings indicate that Asp 189 , Gly 190 , and His 191 compose a major part of the zincbinding motif.
Crystallographic studies of the voltage-sensor paddle of voltage-gated K ϩ channels show that the S1-S2 loop and the S3 and S4 loops are in close proximity (25). Therefore, we hypothesized that the Asp-Gly-His motif in the IS3-IS4 loop is likely to be structurally close to the IS1-IS2 loop of Ca v 3.2 (Fig. 4A). To test the possibility that the region from the amino-terminal end to the IS1-IS2 loop influences zinc sensitivity, we constructed Ca v 3.1/ 3.2:N-IS12LϩL171GϩQ172H . The zinc sensitivity of this mutant was similar to that of Ca v 3.1/ L171GϩQ172H , displaying an IC 50 of 5.5 Ϯ 0.5 M (n ϭ 10) (Fig. 4, B and C). This finding suggests that non-conserved residues in the NH 2 terminus to IS1-IS2 loop do not contribute to the zinc metal binding site.
Zinc inhibition profiles of Ca v 3.1/ 3.2:IS34L and Ca v 3.1/ 3.2:IS2-IS4 showed that, despite both mutants containing the Asp-Gly-His motif, the former was still about 2-fold less sensitive to zinc than Ca v 3.2, whereas the latter was about 2-fold more sensitive (Fig. 2). Compared with Ca v 3.1/ 3.2:IS34L , Ca v 3.1/ 3.2:IS2-IS4 additionally contained the IS2-IS3 and IS4 regions from Ca v 3.2, implying that these regions are responsible for the difference in zinc sensitivity. Sequence comparison from IS2 to IS4 revealed that "Phe-Asp-Ala 141 " containing one negatively charged residue was present at the outer portion of IS2 of Ca v 3.2, whereas "Phe-Asp-Asp 122 " containing two negatively charged residues was at the corresponding portion in Ca v 3.1 (Fig. 4A). We hypothesized that the position and number of aspartate residues at this position might alter zinc sensitivity. To test this hypothesis, we first replaced Asp 122 in Ca v 3.1/ L171GϩQ172H with a non-polar alanine residue, converting "Phe-Asp-Asp" into "Phe-Asp-Ala." Ca v 3.1/ D122AϩL171GϩQ172H was about 2and 4-fold more sensitive to zinc than wild-type Ca v 3.2 and Ca v 3.1/ L171GϩQ172H , respectively (IC 50 ϭ 1.4 Ϯ 0.2 M, n ϭ 14; p Ͻ 0.001; Fig. 4, B and C). In contrast, Ca v 3.2/ A141D was 4-fold less sensitive to zinc than wild-type Ca v 3.2 (Ca v 3.2/ A141D , IC 50 ϭ 12.5 Ϯ 2.1 M, n ϭ 8; Fig. 4, B and C). In addition, mutation of D140A (Ca v 3.2/ D140A ) reduced zinc sensitivity of Ca v 3.2 by 2.6-fold (IC 50 ϭ 7.9 Ϯ 1.3 M, n ϭ 8; Fig. 4, B and C), whereas mutation of D140E (Ca v 3.2/ D140E ) did not significantly alter zinc sensitivity (IC 50 ϭ 3.4 Ϯ 0.7 M, n ϭ 10; p ϭ 0.4601; Fig. 4B). These consistent changes in zinc sensitivity by diverse mutations of aspartate residues in this region support the hypothesis that the acidic residue(s) in the outer portion of IS2 may also play a role in zinc binding. Another implication is that the presence of only one negatively charged residue in this region is more likely to form a structural conformation favorable to zinc block.
Based on the experimental results and potassium channel models (25,26), we developed a model to illustrate how zinc coordinates to the Asp-Gly-His motif and the acidic residue at the outer portion of IS2 (Fig. 5). We chose a model of the closed channel (26) rather than the crystal structure of the open channel (25), because our previous studies demonstrated that nickel had a lower affinity for the open state (9). The model suggests zinc might inhibit channel opening by disrupting the function of the voltage-sensor paddle in repeat I, rather than a direct action on Ca 2ϩ ion permeation as observed with Cd 2ϩ block of HVA channels. We next sought biophysical data to support this hypothesis, reasoning that Zn 2ϩ might alter charge movement that precedes channel opening. Due to limitations in the two microelectrode voltage clamp of oocytes, we used patch clamp recording of HEK-293 cells in the whole cell mode. On-gating currents can be measured at the reversal potential after series resistance compensation and proper cancellation of residual capacitance charge, and quantitated by integrating the area of the outward gating current (27). Expression of recombinant Ca v 3.2 in HEK-293 cells generated ϳ4000 pA of inward current during step depolarizations to Ϫ20 mV and ϳ750 pA outward current during step depolarizations to ϩ55 mV (Fig. 6). Addition of 10 M zinc inhibited the inward current by 91% (Ϯ1%, n ϭ 6), and inhibited the outward gating current 23% (Ϯ1%, n ϭ 6, p ϭ 0.08, paired t test; Fig. 6, C and D). Time matched controls (Control 2) showed no change in gating current, ruling out effects due to rundown. These results indicate that zinc is capa- ble of inhibiting gating charge movement, and is consistent with a model whereby zinc binding stabilizes the closed conformation of the voltage-sensor paddle.

DISCUSSION
We recently identified His 191 in the IS3-IS4 extracellular linker as a critical determinant of nickel, copper, zinc, and redox sensitivity of Ca v 3.2 (15)(16)(17). To localize other partners of His 191 required for interacting with high affinity zinc, we systematically transferred various portions in domain I of Ca v 3.2 into Ca v 3.1/ Q172H , which was only slightly more sensitive to zinc than wild-type Ca v 3.1. Analysis of their zinc doseresponse relationships revealed that introduction of the region(s) only including the IS3-IS4 loop of Ca v 3.2 into Ca v 3.1/ Q172H (or Ca v 3.1) dramatically increased its zinc sensitivity, indicating that the IS3-IS4 loop contains additional residues involved in metal binding. Consequent point mutations uncovered the role of an "Asp-Gly-His" motif involved in zinc inhibition of Ca v 3.2. Chimeras and point mutations also revealed an important role of aspartate residues in IS2 in zinc block, thereby providing the basis for a structural model where trace metals bind and block movements of the repeat I voltage-sensor paddle. This hypothesis was supported by measuring the ability of zinc to reduce on-gating charge movements.
The critical role of Gly in the middle of the Asp-Gly-His motif was shown by two experimental results: 1) single mutation of Q172H into Ca v 3.1 only partially increased zinc sensitivity, but double mutations of L171G and Q172H into Ca v 3.1 had a larger effect, bringing zinc potency close to that of wildtype Ca v 3.2, and 2) single mutation of G190L decreased the zinc sensitivity of Ca v 3.2. Glycine residues are known for providing flexibility for polypeptide(s) because glycine has hydrogen as its side chain, rather than a carbon, as is the case in all other amino acids (28). Based on its structural property, we simply interpret that the Gly in the Asp-Gly-His motif provides not only a deprotonated amide nitrogen atom, but also flexibility for its neighboring Asp 189 and His 191 residues to interact with zinc more efficiently. This may explain why introduction of histidine into the corresponding position of Ca v 3.1 was not sufficient to convert zinc block to the potency observed in Ca v 3.2. We hypothesize the bulky leucine residue in Ca v 3.1 interferes with zinc coordination by Asp 170 and His 172 .
The amino-terminal Cu(II)-and Ni(II)-binding (ATCUN) motif of serum albumin is composed of an "Asp/Glu-Ala-His" (29). Based on the similarity of the Ca v 3.2 Asp-Gly-His motif to the ATCUN motif, we predicted that the acidic residue in the motif would also be critical for the micromolar zinc sensitivity of Ca v 3.2. This prediction was tested by mutating Asp 189 in Ca v 3.2 into either a neutral residue or another acidic residue, followed by examining their zinc sensitivities. Point mutation   (26). For clarity, only the voltagesensor paddle is shown (left panel). The zinc interacting residues are shown by individual amino acids, of which oxygen, carbon, and nitrogen atoms are marked red, white, and blue, respectively. A possible location of the zinc ion is displayed with a blue sphere. Because the amino acid sequence of the IS3-IS4 linker of Ca v 3.2 is profoundly different from the Shaker potassium channel, we modeled the Asp-Gly-His motif to that predicted for the albumin ATCUN motif (29).  (41). Currents were measured using 10 mM Ca 2ϩ in the external solution and 155 TEA-CI in the internal solution as charge carriers. A, ionic currents recorded during a step pulse to Ϫ20 mV from a holding potential of Ϫ100 mV. Currents were measured before (control) and after application of 10 M ZnCl 2 . B, average results of the peak currents were measured in control and zinc. C, gating currents were measured from the same cell using a step depolarization to ϩ55 mV. Each trace represents the average of 20 consecutive sweeps. As in A, data recorded in the presence of zinc is represented by a thick line. D, average gating currents were normalized to control. Control 2 is a time-matched control where cells were continuously perfused with the 10 mM external solution.
of D189A decreased the zinc sensitivity by about 10-fold. Notably, point mutation of D189E enhanced the zinc sensitivity by about 3-fold. These results support that, in addition to Gly 190 and His 191 , the preceding Asp 189 is another requirement for the high zinc sensitivity to Ca v 3.2. The structure of the ATCUN motif (Asp/Glu-Ala-His) was proposed to be a pentacoordinated structure formed by the carboxyl group of the acidic residue and four nitrogen ligands (one from the amino terminus, two from the peptide backbone, and one from the imidazole nitrogen) (29). In addition to an ATCUN motif, our data support the hypothesis that Ca v 3.2 includes an additional ligand, coming from the carboxyl group of the aspartate residue in IS2, thereby providing a model of the zinc binding site in Ca v 3.2.
Previous studies have established that negatively charged residues in transmembrane segments S1, S2, and S3 also act as parts of voltage sensing machinery by electrostatically interacting with positively charged residues in the S4 segment of voltage-gated ion channels (30). The electrostatic interaction between charged residues helps fold the channel in a proper conformation that efficiently targets the channels to the plasma membrane (31,32). Papazian and colleagues (33,34) also found that divalent ions such as nickel and magnesium can bind to the extracellular ion binding pocket formed by the negatively charged residues at S2 and S3 in ether-à-go-go K ϩ channels, decelerating activation kinetics and/or inhibition of currents. Based on these previous findings, we tested whether the negatively charged residue(s) at the outer portion of IS2 were involved in zinc sensitivity. Sequence comparison showed that Ca v 3.1 has Phe-Asp-Asp, whereas Ca v 3.2 has Phe-Asp-Ala at this position. Ca v 3.1/ D122AϩL171GϩQ172H , constructed by replacing Phe-Asp-Asp of Ca v 3.1/ L171GϩQ172H with Phe-Asp-Ala, was 4.6-fold more sensitive to zinc than Ca v 3.1/ L171GϩQ172H . Ca v 3.2/ A141D constructed by mutating Phe-Asp-Ala into Phe-Asp-Asp became 4.2-fold less sensitive to zinc than wild-type Ca v 3.2. These data imply that the presence of one acidic residue in the middle of this triplet renders higher zinc sensitivity than that of two acidic residues. This implication is further supported by the findings that the presence or absence of a negatively charged residue in the middle of the triplet significantly affected zinc inhibition sensitivity (Ca v 3.2/ D140E versus Ca v 3.2/ D140A , IC 50 ϭ 3.4 Ϯ 0.7 versus 7.9 Ϯ 1.3 M). These results suggest that Phe-Asp-Ala at the outer portion of IS2 is more favorable than the other sequence combinations in improving zinc inhibition sensitivity, together with the major contribution of the Asp-Gly-His motif.
All of the mutant channels where the negatively charged residue at the outer portion of IS2 was neutralized showed a common positive shift of their activation and inactivation curves compared with those of wild-type Ca v 3 (Table 1). For example, activation and inactivation curves of Ca v 3.2/ D140A were positively shifted by ϳ10 mV, and its functional expression in Xenopus oocytes was dramatically decreased as well (data not shown). The voltage dependence of activation and inactivation of all the other mutants were similar to their respective wildtype channels, ruling out any artifactual shift in sensitivity due to less activation under the voltage protocols used. By analogy to findings with K ϩ channels, the decrease in expression may suggest that Asp 140 is electrostatically interacting with posi-tively charged residues in IS4, influencing structural conformation of the channel and trafficking of the channels to the plasma membrane (30,35).
We identified two structural elements critical for rendering the high zinc sensitivity to Ca v 3.2: the Asp-Gly-His motif in the IS3-IS4 loop and an Asp residue in IS2. Based on the chimeric approach used in this study, we cannot rule out possible involvement of conserved residues in other parts of the channel, such as S4 voltage-sensor paddle regions, as identified in Cu 2ϩ block of BK channels (36). In addition, this study does not localize the lower affinity zinc binding sites involved in block of Ca v 3.1. Although the underlying mechanism of how the negatively charged residue in IS2 contributes to zinc sensitivity remains to be further investigated, a simple interpretation is that the acidic residue acts as another member for zinc coordination with the Asp-Gly-His motif, based on the findings that its mutation into a neutral residue decreases zinc sensitivity. The ability of zinc to decrease gating charge 25% is consistent with immobilization of one of the four voltage-sensor paddles. It is interesting to note that repeat I plays a dominant role in the opening of both HVA and LVA channels (37,38). These studies also provide evidence for considerable structural similarity between voltage-gated K ϩ and Ca 2ϩ channels, and combined with the established role of Ca v 3.2 in pain and epilepsy, provide a structural model for the development of novel therapeutics (17,24,39,40).