A Molecular Determinant of Nickel Inhibition in Cav3.2 T-type Calcium Channels*

Molecular cloning studies have revealed that heterogeneity of T-type Ca2+ currents in native tissues arises from the three isoforms of Cav3 channels: Cav3.1, Cav3.2, and Cav3.3. From pharmacological analysis of the recombinant T-type channels, low concentrations (<50 μm) of nickel were found to selectively block the Cav3.2 over the other isoforms. To date, however, the structural element(s) responsible for the nickel block on the Cav3.2 T-type Ca2+ channel remain unknown. Thus, we constructed chimeric channels between the nickel-sensitive Cav3.2 and the nickel-insensitive Cav3.1 to localize the region interacting with nickel. Systematic assaying of serial chimeras suggests that the region preceding domain I S4 of Cav3.2 contributes to nickel block. Point mutations of potential nickel-interacting sites revealed that H191Q in the S3–S4 loop of domain I significantly attenuated the nickel block of Cav3.2, mimicking the nickel-insensitive blocking potency of Cav3.1. These findings indicate that His-191 in the S3–S4 loop is a critical residue conferring nickel block to Cav3.2 and reveal a novel role for the S3–S4 loop to control ion permeation through T-type Ca2+ channels.


MATERIALS AND METHODS
Chemicals-Nickel (II) chloride hexahydrate (NiCl 2 ⅐6H 2 O) was obtained from Sigma (St. Louis). Most of the other chemicals were purchased from Sigma and USB (Cleveland, OH). A nickel stock solution (100 mM) was made in deionized water and stored at room temperature. A series of nickel solutions (in M: 1, 3, 10, 30, 100, 300, 1000, 3000) were prepared by diluting the nickel stock solution with 10 mM Ba 2ϩ solution just before experiments, and their pH values were adjusted to 7.6.

Construction of Chimeras between Ca v 3.1 and Ca v 3.2-
The chimeric channels were constructed by modification of the cDNAs encoding the rat Ca v 3.1 (␣ 1G , GenBank TM accession number AF027984) and human Ca v 3.2 (␣ 1H , GenBank TM accession number AF051946) channels, for which PCR was used to add silent or non-silent restriction enzyme sites. The chimeric channels were subcloned into the pGEM-HEA vector containing 5Ј and 3Ј untranslated regions of a Xenopus ␤-globin gene for better expression in oocytes (20). The methods for construction of the chimeric channels have been reported previously (21). Details of their construction are given below, where the restriction endonuclease sites and the nucleotide positions in the respective channel cDNAs are designated in parentheses. In some cases a restriction site was introduced by PCR, and these are indicated by asterisks for silent mutations and crosses for mutations that change the protein sequence.
Site-directed Mutagenesis-Point mutations were generated using two-step PCR methods (22).
HGGG/E137Q-The forward and reverse primers to amplify the upper fragments covering from the 5Ј-polylinker to 499 (nucleotide number of Ca v 3.2) were 5Ј-TAATACGACTCACTATAGGG-3Ј (T7 promoter sequence) and 5Ј-TCAAAGGCCTGCAGGATGTTGCAG-CGCTC-3Ј, respectively. The forward and reverse primers to amplify the lower fragments covering from 479 to 1228 (nucleotide number of Ca v 3.2) were 5Ј-AACATCCTGCAGGCCTTTGACGCCTTCATT-3Ј and 5Ј-GATGTCGACCCAGCCTTCCAG-3Ј, respectively. The amplified DNA fragments were purified and then combined by additional PCR. The plasmid HGGG/E137Q was constructed by inserting the PCR-generated DNA cassette digested with ClaI and SalI into the plasmid Ca v 3.1 pGEM-HEA, which was opened by ClaI (5Ј-polylinker) and SalI (1552, Ca v 3.1).
HGGG/H191Q-The forward and reverse primers to amplify the upper fragments covering from the 5Ј-polylinker to 682 (nucleotide number of Ca v 3.2) were T7 promoter sequence and 5Ј-GGCTCACGT-TCTGTCCGTCCAACGAGTACTC-3Ј, respectively. The forward and reverse primers to amplify the lower fragments covering from 641 to 1228 (nucleotide number of Cav3.2) were 5Ј-TTGGACGGACAGAA-CGTGAGCCTCTCGGCTAT-3Ј and 5Ј-GATGTCGACCCAGCCT-TCCAG-3Ј, respectively. The upper and lower DNA fragments were purified and then combined by additional PCR. The plasmid HGGG/ H191Q was constructed by inserting the PCR-generated DNA cassette digested with ClaI and SalI into the plasmid Ca v 3.1 pGEM-HEA, which was opened by ClaI (5Ј-polylinker) and SalI (1552, Ca v 3.1).
Ca v 3.1/Q172H (GGGG/Q172H)-The forward and reverse primers to amplify the upper fragments covering from the 5Ј-polylinker to 1008 (nucleotide number of Ca v 3.1) were T7 promoter sequence and 5Ј-AG-CTGACGTTGTGCAGGTCCAGCGAATACTCC-3Ј, respectively. The forward and reverse primers to amplify the lower fragments covering from 988 to 1762 (nucleotide number of Ca v 3.1) were 5Ј-TG-GACCTGCACAACGTCAGCTTCTCCGCA-3Ј and 5Ј-GAGAAGC-TTGCCAGGGTGCTAGC-3Ј, respectively. The amplified upper and lower DNA fragments were purified and then combined by additional PCR. The plasmid Ca v 3.1/Q172H was constructed by inserting the PCR-generated DNA cassette digested with ClaI and HindIII into the plasmid Ca v 3.1 pGEM-HEA, which was opened by ClaI (5Ј-polylinker) and HindIII (1755, Ca v 3.1).
All PCRs were performed using Pfu Ultra DNA polymerase (Stratagene), and the entire region derived from PCR products was sequenced to verify correct introduction of point mutated site(s) and that there were no inadvertent mutations.
All cDNAs encoding Ca v 3.1, Ca v 3.2, and chimeric channels were linearized by AflII and used as templates. Capped cRNAs were synthesized in vitro using T7 RNA polymerase provided in the mMessage mMachine transcription kit in according to the manufacturer's instruction (Ambion, Austin, TX). 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). SOS solution was changed daily.
Electrophysiology and Data Analysis-Barium currents were measured using a two-microelectrode voltage clamp amplifier (OC-725C, Warner Instruments, Hamden, CT) between the third and eighth day after cRNA injection. Microelectrodes (Warner Instruments) were broken to decrease the electrode resistance to 0.2-1.0 megohms and filled with 3 M KCl. The bath solution contained 10 mM Ba(OH) 2 , 90 mM NaOH, 1 mM KOH, and 5 mM HEPES (pH 7.4 with methanesulfonic acid). The currents were acquired at 5 kHz and low pass-filtered at 1 kHz using the pClamp system (Digidata 1320A and pClamp 8, Axon instruments, Foster City, CA). Data were analyzed using the Clampfit software (Axon instruments) and presented graphically using the Prism software (GraphPad, San Diego, CA). Dose-response curves were fitted using the Hill equation: B ϭ (1 ϩ IC 50 /(Ni 2ϩ ) n ) Ϫ1 , where B is the normalized block, IC 50 is the concentration of Ni 2ϩ giving half-maximal inhibition, and n is the Hill coefficient. Data are presented as means Ϯ S.E. and tested for significance using Student's unpaired t test.

RESULTS
Prior to testing the chimeric channels, we first confirmed the effects of nickel on wild-type Ca v 3.1 and Ca v 3.2 channels. Peak Ba 2ϩ currents were elicited by test pulses to Ϫ20 mV from a holding potential of Ϫ90 mV every 15 s. Expression of the T-type Ca 2ϩ channels were detected as robust inward currents from the third day after cRNA injection. Application of serial nickel solutions inhibited Ba 2ϩ currents through the Ca v 3.1 or Ca v 3.2 channels in a dose-dependent manner, and the inhibited currents could be reversed by washing (Fig 1, A and B). The Ca v 3.1 currents required high concentrations of nickel to be blocked. In contrast, the Ca v 3.2 currents were highly sensitive to nickel block. On aver- age, the IC 50 values for inhibiting the Ca v 3.1 and Ca v 3.2 channels were 304.8 Ϯ 6.2 and 4.9 Ϯ 2.0 M, respectively ( Fig. 1, C and D), being consistent with previous studies (19). Comparison of current-voltage (I-V) relationships of Ca v 3.1 or Ca v 3.2 currents before and after nickel treatment showed that nickel inhibition positively shifted the I-V relationships ( Fig. 1, E and F). Consistently, the half-activation potentials were found to be positively shifted by nickel (Table 1).
Based on the different potencies of nickel block between the Ca v 3.1 (GGGG) and Ca v 3.2 (HHHH), we investigated what structural portion(s) endowed Ca v 3.2 with nickel sensitivity. In this regard, sensitivities of nickel block were examined for a series of chimeric channels (Fig.  2). Of the two half-half chimeras, the GGHH currents were blocked by relatively high concentrations of nickel. On average, the IC 50 value of GGHH was 307.3 Ϯ 8.1 M (n ϭ 6), similar to that for Ca v 3.1 (GGGG). On the contrary, the HHGG currents were sensitively blocked by low concentrations of nickel. On average, the IC 50 value was 7.3 Ϯ 2.2 M (n ϭ 5), similar to that for the Ca v 3.2. These findings suggested that the structural element(s) contributing to high nickel sensitivity were located in the first half (domain I and II) but not on the second half (domain III and IV) of the Ca v 3.2.
Our next step was to transfer a single domain of Ca v 3.2 into Ca v 3.1. The HGGG currents were found to be blocked by low concentrations to nickel. In contrast, the GHGG currents required much higher concentrations of nickel to be blocked. The IC 50 values for blocking HGGG and GHGG were 4.7 Ϯ 1.8 M (n ϭ 5) and 291.1 Ϯ 5.2 M (n ϭ 6), respectively. Taken together, the potency of nickel block for the HGGG was very close to that of the wild-type Ca v 3.2, indicating that the domain I of the Ca v 3.2 contains the essential structural element(s) determining the nickel-sensitive block.
The identified domain I of the Ca v 3.2 was further dissected to narrow down the exact region(s) contributing to the nickel block. We initially hypothesized that the pore loop and S6 of the domain I, known to be essential for ion permeation and selectivity, are involved in the nickel block. However, it is unlikely that these structural portions contribute to the high nickel sensitivity, because the Ca v 3.1 and Ca v 3.2 contain identical amino acid sequences in these regions. Our next hypothesis was that the extracellular loop connecting the S5 and the pore is involved in the nickel block, because the extracellular loop sequences are quite different between the two T-type channels. However, the extracellular loop mutant channel, HGGG/G IS5-Ipore (where the IS5-pore loop of HGGG was replaced with the corresponding one of the Ca v 3.1) was still sensitive to nickel (IC 50 ϭ 4.7 Ϯ 1.9 M, n ϭ 5). These results restricted the nickel interacting site(s) within the remaining region from the amino terminus to S4 (IS4) of domain I of the Ca v 3.2. To examine the relevance of this region, GGGG/H N-IS4 was constructed. As expected, the chimeric channel currents were blocked by low concentrations of nickel (IC 50 ϭ 3.9 Ϯ 2.1 M, n ϭ 9), for the GGGG/H N-IS4 , which was slightly lower than that for the wild-type Ca v 3.2. These findings indicate that essential structural determinant(s) for the nickel-sensitive block reside between the amino terminus and IS4.
Next, we postulated that nickel may interact with regions preceding to IS4, such as with residue(s) in the extracellular loops between IS1 and IS2, and/or IS3 and IS4. To identify putative nickel-interacting residue(s), we aligned the amino acid sequences in the regions prior to IS4 of the three T-type channel isoforms (Fig. 3A). Ni 2ϩ can interact with histidine (H) and cysteine (C) residues and the acidic amino acids, aspartic acid (D) and glutamic acid (E) (23)(24)(25)(26). Involvement of Glu-127 and Glu-131 in the IS1-IS2 loop of the Ca v 3.2 seems unlikely because aspartate (D), a negatively charged amino acid similar to glutamate (E), is found in the corresponding position of the nickel-insensitive Ca v 3.3   FEBRUARY 24, 2006 • VOLUME 281 • NUMBER 8 , similar to that of HGGG, suggesting that Glu-137 in the IS1-IS2 loop is not a crucial residue determining nickel block (Fig. 3, B and D). In contrast, HGGG/H191Q currents required much higher concentrations of nickel to be blocked, showing an IC 50 of 312.5 Ϯ 4.2 M (n ϭ 10). These findings show that the single point mutation of H191Q induced a 25-fold change in nickel sensitivity and suggest that His-191 accounts for the high nickel sensitivity observed with the HGGG chimera. We next sought to confirm the critical role of His-191 by mutating this residue in wild-type T-type channels. The application of nickel solutions dose-dependently inhibited Ca v 3.2/H191Q (HHHH/H191Q), and the inhibited currents could be reversed by washing (Fig 4A). On average, the IC 50 for the nickel block was 306.6 Ϯ 7.1 M (n ϭ 6), indicating that the H191Q mutation greatly reduced the nickel sensitivity of the channel. Another point mutation of H191A at the same location of the Ca v 3.2 also reduced the nickel sensitivity to a similar level (IC 50 ϭ 285.9 Ϯ 3.1 M, n ϭ 7, Fig. 4B). I-V relationships of Ca v 3.2/ H191Q currents in the absence and presence of 300 M nickel showed that nickel shifted the I-V relationship to more depolarized potentials (Fig. 4C, Table 1). These results support our findings in the chimeric channels and show that His-191 in the IS3-IS4 loop confers the high nickel sensitivity to the Ca v 3.2.

Structural Determinant of Ca v 3.2 for Nickel-sensitive Block
Finally, we examined whether Ca v 3.1 could be transformed into a nickel-sensitive channel by simply switching the corresponding glutamine (Q) of the Ca v 3.1 to histidine (H). Accordingly, Ca v 3.1/Q172H (GGGG/Q172H) was constructed, and its nickel sensitivity was assayed. Ca v 3.1/Q172H currents were inhibited by nickel solutions in a dose-dependent manner, and the inhibited currents were rapidly recovered by washing (Fig. 4D). Consistent with our hypothesis, the nickel blocking sensitivity of the GGGG/Q172H was increased 5-fold (IC 50 ϭ 61.3 Ϯ 3.7 M, n ϭ 9), although it was not as sensitive as that of the Ca v 3.2 (Fig. 4,   D and E). These results clearly show that His-191 is a key residue in the nickel binding pocket (Fig. 5). Alternatively, mutation of His-191 might have led to a rearrangement of the channel that disrupted the nickel binding pocket. Data arguing against this latter hypothesis are that the biophysical (Table 1) and pharmacological properties (mibefradil doseresponse studies, results not shown) were not altered by H191Q mutation.
Voltage-gated ion channels contain many conserved amino acids and are likely to be similar in structure-function. Therefore we have modeled repeat I of Ca v 3.2 using current models for the Shaker K ϩ channel (27). Voltage-dependent gating is thought to begin with outward movement of S4 segments, which in turn leads to opening of the channel walls formed by S6 segments. In these models the S3-S4 linker is in close proximity to the extracellular face of S4, S5, and S6 segments; therefore it is likely that there is a nickel binding pocket on the extracellular surface of Ca v 3.2 channels. If so, then nickel should be able to bind to closed channels in the rested state and block their transition to open states. To test this prediction we exposed oocytes expressing Ca v 3.2 channels to nickel at a holding potential of Ϫ100 mV, then tested for channel availability (Fig. 6). Nickel evoked the same degree of block in the absence and presence of depolarizing test pulses, indicating that it could bind to closed channels.

DISCUSSION
Voltage-gated calcium channels are highly selective for Ca 2ϩ ions because they bind Ca 2ϩ in the pore, thereby preventing permeation by monovalent cations. This binding site also binds other divalent cations such as Cd 2ϩ with higher affinity, leading to the block of Ca 2ϩ permeation. This binding site was localized to the pore loops by site-directed mutagenesis (10) and in HVA channels is formed by glutamates (E) in each of the four repeats (EEEE locus). The mutation of these residues in Ca v 1.2 channels significantly decreased the affinity for cadmium binding (28). In low voltage-activated channels two of these glutamates were replaced by aspartates (D), creating an EEDD locus. Replacement of these aspartates in Ca v 3.1 channels with glutamates confers a cadmiumsensitive block to Ca v 3.1, which requires much higher concentrations of cadmium to be blocked than HVA calcium channels (29). Our previous studies indicated that nickel blocks Ca v 3 channels in part by binding within the permeation path (19), presumably because of binding at the  (Durell et al.,27) is shown. In these models the S3-S4 linker comes in close proximity to the extracellular surface of segments S5 and S6 and the linker joining these segments to the pore (P). A Ni 2ϩ binding pocket could be formed in this region, and its binding could disrupt movements of these segments that occur during channel gating. In addition to His-191, the Ni 2ϩ binding pocket might be composed of other histidine or cysteine residues in the S5-P and P-S6 loops. Also shown is the P loop from repeat 3, and the glutatmate (E) and aspartate (D) residues that form the EEDD locus in the permeation pathway. The S5-P loop contains 103 amino acids and has been truncated. EEDD locus. Our present finding that nickel binds to the S3-S4 loop was unexpected and reveals a second site and mode of action. We propose that binding to the S3-S4 linker prevents movements associated with channel gating effectively stabilizing channels in closed states.
A similar hypothesis was proposed by Zamponi et al. (11) for nickel effects on HVA channels, where nickel was found to shift activation gating at lower concentrations than it blocked permeation. Notably the HVA channel that showed the greatest shift in the voltage dependence of gating was Ca v 2.3. Comparison of the amino acid sequences of all 10 Ca 2ϩ channel ␣1 subunits shows that only Ca v 3.2 and Ca v 2.3 contain a histidine in this region of the IS3-IS4 loop (His-179, GenBank TM accession number Q15878). Therefore, it is tempting to speculate that this residue might explain why Ca v 2.3 is more sensitive to the gating effects of nickel than other HVA channels.
Previous studies have shown that ␣-scorpion toxins or sea anemone toxins significantly slow fast inactivation of voltage-gated sodium channels and slightly inhibit their current amplitude (30,31). The main binding site of toxins was identified to be in the S3-S4 loop of domain IV of sodium channels, indicating that the structural determinant for prolonging the fast inactivation and inhibiting the activity of sodium channels is analogous to that for that nickel block of the Ca v 3.2. The structural and functional analogy between the two channels suggests that nickel might slow inactivation kinetics of Ca v 3.2. Analysis of current kinetics before and after nickel block shows that this was indeed the case for wild-type channels (19), and in nickel-sensitive chimeras (Table 1), but was lost in the H191Q mutant of Ca v 3.2.
The H191Q or H191A mutation in Ca v 3.2 lowered its nickel sensitivity to that observed for Ca v 3.1, and conversely, the Q172H mutation in Ca v 3.1 increased its nickel sensitivity 5-fold (IC 50 ϭ 61.3 Ϯ 3.7 M) approaching, but not quite matching, the sensitivity observed for Ca v 3.2. These results indicate that His-191 is a critical residue in nickel binding and that most of the other residues that form the nickel binding pocket are conserved. Because of their conservation these residues cannot be identified using the chimeric approach. The finding that Ca v 3.1 channels with the Q172H mutation are not as sensitive as Ca v 3.2 could be because either 1) the S3-S4 linker adopts a different conformation in Cav3.1 channels, thereby positioning the histidine in a slightly different orientation, or 2) that there are other residues involved in binding nickel in Ca v 3.2 that are not conserved in Ca v 3.1. A similar case has been reported for the nickel-induced augmentation of cyclic nucleotidegated channels, where mutation of the nickel-sensitive retinal cyclic nucleotide-gated channel at His-420 to the corresponding amino acid found in nickel-insensitive olfactory isoforms (Q) almost completely abolished the effect of nickel (32). Conversely, the Q to H mutation in olfactory channels conferred some nickel sensitivity, but it was still less sensitive than the rod isoforms. These results were interpreted as showing that His-420 might require other residues for coordinated interaction with Ni 2ϩ , because Ni 2ϩ has 4 -6 ligands and interacts weakly (K d ϭ 1 mM) with a single imidazole of histidine (33).
The time course of nickel inhibition of both wild-type and mutant channels did not show any concentration dependence (Figs. 1 and 4). Therefore, we were unable to calculate the apparent affinity constant K D from the K on and K off rates. Apparently the on rate of block is faster than our perfusion speed. A second limitation of the present study is the relatively long time it takes to clamp the entire oocyte membrane, thereby precluding detailed studies of activation kinetics. Future analysis of the activation and blocking kinetics would be better studied using patch clamp electrophysiology of mammalian cells with a faster perfusion system.
In summary, we have identified the histidine at position 191 as a key molecular determinant contributing to the high affinity block of Ca v 3.2 channels by nickel. This residue was localized using a series of chimeras between channels that show high affinity block (Ca v 3.2) and low affinity block (Ca v 3.1). Interestingly His-191 resides in the short (9-amino-acid long) loop that connects IS3 to the IS4 voltage sensor, rather than residing in the pore loops as might be expected from work on cadmium binding sites. Based on observations that nickel appears to block Ca v channels at two sites (11,19), we propose that nickel binding to His-191 blocks the gating of channels to the open state by interrupting the coupling between S4 and the pore.