Ni²⁺ Block of Ca_V3.1 (α1G) T-type Calcium Channels

Many ions block channels by binding within the pore, preventing flux of permeant ions. In particular, several di- and trivalent cations bind with high affinity to the ion selectivity filter of voltage-dependent cation channels. Indeed, that is how calcium itself blocks permeation of monovalent cations, introducing calcium selectivity into a channel that has a high intrinsic conductance for monovalent cations (Almers et al., 1984; Hess and Tsien, 1984). High affinity blockers, which can block not only flux of monovalent cations but also current carried by calcium ions, have been especially useful as probes for the functional architecture of the calcium channel pore (Yang et al., 1993; Obejero-Paz et al., 2004).

Although the fundamental mechanism of ion selectivity is conserved among the Ca_V family of voltage-dependent calcium channels, it has long been recognized that there are significant quantitative differences in selectivity for permeation and block. Ni²⁺ has received particular attention. Originally thought to be a selective blocker of T-type calcium channels (Narahashi et al., 1987; Hagiwara et al., 1988), it is now clear that only one of the three cloned mammalian T-channels, Ca_V3.2, is strongly inhibited by Ni²⁺ (Lee et al., 1999), while one high voltage-activated channel, Ca_V2.3, is quite sensitive to Ni²⁺ (Soong et al., 1993). However, it has been suggested that Ni²⁺ acts in part by binding to a site on the extracellular side of the channel, affecting current through the channel allosterically, not only by binding within the permeation pathway itself (Zamponi et al., 1996). Lacinova et al. (2000) reported that Ni²⁺ accelerated deactivation of Ca_V3.1, interpreted as a modification of activation gating, but Diaz et al. (2005) concluded instead that this effect resulted from time-dependent block by Ni²⁺.

Recently, mutational analysis has found that high affinity inhibition by Ni²⁺ of Ca_V3.2 requires a histidine near the extracellular side of the S4 transmembrane segment of domain I (Kang et al., 2005). Replacement of histidine with glutamine (the corresponding residue in Ca_V3.1) increased the IC₅₀ from 5 to 300 μM in Ca_V3.2. The reverse mutation (Q172H in Ca_V3.1) had a partial effect, changing the IC₅₀ from 300 to 60 μM. This suggests that high affinity inhibition by Ni²⁺ results from binding to an extracellular site (Kang et al., 2005). That result also reopens the question of the mechanism of the relatively low affinity inhibition of Ca_V3.1 by Ni²⁺. One clue is that the effect of Ni²⁺ on Ca_V3.2 is essentially voltage independent (Diaz et al., 2005; Kang et al., 2005), while pore block by di- and trivalent cations is typically voltage dependent for Ca_V channels, including Ca_V3.1 (Lee et al., 1999; Serrano et al., 2000; Obejero-Paz et al., 2004).

We examine here not only the voltage dependence of Ni²⁺ inhibition of Ca_V3.1, but also the dependence on the chemical nature and the concentration of the permeant ion. The potency of pore block is strongly influenced by competitive interactions among ions in the pore; for example, block is much stronger when monovalent ions are the charge carrier (Yang et al., 1993). We have also found that block by Mg²⁺ is considerably stronger in 2 mM Ba²⁺ than in 2 mM Ca²⁺ for Ca_V3.1, even though currents carried by Ba²⁺ and Ca²⁺ are quite comparable (Serrano et al., 2000). The nature of the ionic effects may vary among blocking ions, however, as the rate of block by Y³⁺ is sensitive to Ca²⁺ vs. Ba²⁺ for closed but not open channels (Obejero-Paz et al., 2004).

We report here that Ni²⁺ has two inhibitory effects on Ca_V3.1 channels, a rapid inhibition that is ion dependent but weakly voltage dependent, and a slower block (reflected in tail current acceleration) that is both ion and voltage dependent. Both effects are most consistent with interactions of Ni²⁺ with the ion conduction pathway, as opposed to an extracellular allosteric site (see Fig. S1, available). Ni²⁺ blocks Ca_V3.1 by mechanisms fundamentally different from the voltage-independent high-affinity block of Ca_V3.2.

General methods for whole-cell patch clamp recording from HEK 293 cells stably transfected with Ca_V3.1 (α1G) were specified by Khan et al. (2008) and are only summarized here.

Experiments with Ca_V3.2 used transiently transfected HEK 293 cells. Cells were transfected at 80% confluence in 35-mm dishes using FuGENE6 (Roche) with 1.5 μg plasmid DNA encoding rat Ca_V3.2 and 0.5 μg CD8 pcDNA (gifts from E. Perez-Reyes [University of Virginia, Charlottesville, VA] and E. Ficker [Case Western Reserve University], respectively). One day after transfection, cells were dispersed with trypsin and transferred to coverslips. On the day of the experiment, usually 2 d after plating, transfected cells were identified using 1 μl Dynabeads CD8 (Dynal Biotech).

Pipet resistances were 1.8–2.3 MΩ. Access resistances in the whole-cell configuration were 5 ± 1 MΩ and were compensated by ∼90% (“correction”) and ∼50% (“prediction”). Ni²⁺ was added in the concentrations indicated (as NiCl₂) without other change in the extracellular solution.

Effects of Ni²⁺ were examined by two basic protocols: the I-V protocol, where currents were recorded directly upon depolarization from a holding potential of −100 mV to the voltages indicated, and the IIV protocol, where the voltage was varied following a 2-ms step to +200 mV to maximally activate channels with minimal inactivation. These two protocols, taken together, help separate effects of voltage and ions on gating vs. permeation. Both protocols were delivered sequentially before, during, and after recovery from Ni²⁺ exposure, to assess current rundown. Two versions of the I-V protocol were run, brief (5 ms) voltage steps from −90 to +200 mV, and longer (40 ms) steps from −90 to 0 mV (+30 mV in 110 mM Ca²⁺_o or Ba²⁺_o), since channel activation is not complete in 5 ms at some negative voltages. I-V relations shown in the figures are the peak current (from either the 5 or 40 ms protocol), except in nominally Ca²⁺_o -free conditions (Fig. 9), where the current at the end of 5-ms steps was used.

Ni²⁺_o block was described by a simplified Woodhull (1973) model (Khan et al., 2008), with parameters estimated by minimizing the sum of squared errors using the Solver function in Microsoft Excel. Data are shown as mean ± SEM.

Fig. S1 (available) illustrates our proposal for the mechanism of Ni²⁺ block. Fig. S2 compares one- vs. two-exponential fits to tail currents in Ni²⁺, and supplemental Materials and methods describes how rate constants for Ni²⁺ block were calculated from tail currents. Figs. S3–S7 show additional sample current records in control vs. Ni²⁺. Figs. S8 and S9 show additional current records and I-V relations for effects of Ni²⁺ on Ca_V3.2.

Fig. 1 shows effects of 1 mM Ni²⁺ on Ca_V3.1 using the IIV protocol, where channels are first activated by strong depolarization (2 ms at +200 mV) and then repolarized to a wide range of potentials (+100 to −150 mV). Several effects are apparent. First, currents at +200 mV are slower to activate, and are slightly smaller after 2 ms. Second, the “instantaneous” current measured following repolarization is reduced, only slightly at +100 mV, but more substantially at −150 mV. Third, tail currents are faster in Ni²⁺, especially at the more negative voltages. Fourth, the effect of Ni²⁺ appears to be greater with Ba²⁺ than with Ca²⁺, especially on the time course of the inward tail currents.

Fig. 2 illustrates effects of Ni²⁺ on the IIV relationships for several concentrations of Ni²⁺, in 2 mM Ca²⁺ (Fig. 2, A–C) or 2 mM Ba²⁺ (Fig. 2, D–F). Outward currents are reduced only slightly, while inward currents are more sensitive to Ni²⁺, in a concentration-dependent manner (Fig. 2, A and C). The effects are shown more clearly when currents are converted to chord conductances (Fig. 2, B and E). The voltage dependence of the inhibition is unusual (Fig. 2, C and F), as inhibition seems to depend primarily on the direction of current (inward vs. outward). For example, there is little systematic variation in block between −50 and −150 mV, or from +50 to +100 mV. Furthermore, inhibition is not substantially different between 2 mM Ca²⁺ and Ba²⁺ (compare red and blue symbols, 0.3 mM and 1 mM Ni²⁺, respectively, in Fig. 2, C and F).

Fig. 3 examines the effects of Ni²⁺ on the kinetics of tail currents, approximated by single exponential fits. In the absence of Ni²⁺, the time constant reflects a combination of channel deactivation (closing) and inactivation. Voltage-independent inactivation dominates above −50 mV, and voltage-dependent deactivation dominates at more negative voltages (Serrano et al., 1999). Ni²⁺ accelerates tail currents at negative voltages, especially in Ba²⁺ (compare red and blue symbols, 0.3 and 1 mM Ni²⁺, respectively, in Fig. 3, A and B).

One mechanism by which Ni²⁺ might affect tail kinetics is by voltage- and time-dependent block (Obejero-Paz et al., 2004). We estimated the pseudo first-order rate constant for block by Ni²⁺ (k_B) from either one- or two-exponential fits, as described in supplemental Materials and methods. The calculated rates of Ni²⁺ block are described reasonably well by bimolecular kinetics (straight lines in Fig. 3, C and D), especially with 2 mM Ba²⁺ as the charge carrier, over the range of voltage and [Ni²⁺]_o where block can be accurately measured.

The calculated rates of Ni²⁺ block exhibit several interesting features. Block was rather strongly voltage dependent, with rates varying e-fold for 32 mV, corresponding to an apparent electrical distance δ = 0.4 for the transition state for Ni²⁺ entry into the pore. The bimolecular rate constants were several orders of magnitude lower than expected for diffusion-limited entry of Ni²⁺ into the pore, at −80 mV (1.0 ± 0.1) × 10⁶ M⁻¹s⁻¹ in Ca²⁺ (n = 12, 0.3–3 mM Ni²⁺) and (3.1 ± 0.3) × 10⁶ M⁻¹s⁻¹ at −80 mV in Ba²⁺ (n = 11, 0.1–1 mM Ni²⁺). Note that block was approximately threefold faster in Ba²⁺ than in Ca²⁺.

Does the faster entry of Ni²⁺ into the pore in Ba²⁺ produce a more potent block? This was examined using I-V relationships, measured as the effect on peak inward or outward current at each voltage (Fig. 4). With this protocol, inward currents were very strongly inhibited in both Ca²⁺ and Ba²⁺, while outward currents were much less affected (see Fig. S3 for sample current records). Inhibition by Ni²⁺ could be roughly approximated by a Woodhull (1973) model with K_d = 0.2 mM for Ba²⁺ and 0.3 mM for Ca²⁺ (at 0 mV), with weak voltage dependence (e-fold for 64 mV, corresponding to δ = 0.19 for the site of Ni²⁺ block) (Fig. 4, C and F). The Woodhull analysis assumes that the voltage dependence of inhibition by Ni²⁺ results from voltage-dependent pore block, vs. (e.g.) a shift in channel gating to more positive voltages. This issue (and interpretation of δ values) will be considered further in the Discussion.

Estimates of the K_d for block from the IIV protocol (data of Figs. 1 and 2) and the I-V protocol (Fig. 4) are summarized in Fig. 5. With the I-V protocol, Ni²⁺ blocks more strongly in Ba²⁺ than with Ca²⁺ (Fig. 5, open symbols), but the effect is less dramatic than observed for the rate of block by Ni²⁺ (Fig. 3, C vs. D).

Ni²⁺ appears to slow activation of outward currents (Fig. 1 and Fig. S3). One explanation is voltage-dependent relief of resting block by Ni²⁺, as observed for Y³⁺ (Obejero-Paz et al., 2004). That was examined using current ratios (Ni²⁺/control) (Fig. 6). The strongest depolarizations produced nearly complete relief of block by 1 mM Ni²⁺ (Fig. 6, A and B), but relief was incomplete even at +200 mV in 3 mM Ni²⁺ (Fig. 6 C).

Fig. 7 shows the analysis of current ratios, measured as in Fig. 6. The extent of Ni²⁺ block remaining after 5 ms of depolarization is shown in Fig. 7 A (in Ca²⁺) and Fig. 7 D (for Ba²⁺). The time constant for the current ratios accelerated with depolarization (Fig. 7, B and E). That time constant (τ_Ni) should depend on both the on and off rates (Ni²⁺ entry and exit):

where k_B is the pseudo first-order rate constant for block by Ni²⁺, and k_-B is the rate constant for exit of Ni²⁺ from the pore. Assuming that the Ni²⁺/control ratio at 5 ms (r₅) (Fig. 7, A and D) reflects steady-state occupancy of Ni²⁺ at each voltage,

Eqs. 1 and 2 give two equations for two unknowns (k_B and k_-B). Calculated k_-B values are shown in Fig. 7 C (in Ca²⁺) and Fig. 7 F (in Ba²⁺). Ni²⁺ exit was faster with depolarization, as expected for relief of pore block. The other noteworthy effect is that Ni²⁺ exit was slower in 3 mM Ni²⁺ than in lower concentrations of Ni²⁺ (Fig. 7 C). The final level of inhibition at +50 to +100 mV (Fig. 7, A and D) agrees reasonably well with that measured with the IIV protocol (Fig. 2 C). It is especially noteworthy that strong depolarization did not fully reverse inhibition by 3 mM Ni²⁺ (Fig. 6 C and Fig. 7 A), even though the kinetics were fast (τ < 1 ms, Fig. 7 B). Taken together, these results demonstrate that inhibition of Ca_V3.1 by Ni²⁺ cannot be explained by simple voltage-dependent pore block alone. The kinetics of block (from tail currents at negative voltages) and relief of block (from activation at positive voltages) are consistent with a contribution of slow pore block. Slow block can also explain the observation that block is much stronger with the I-V protocol than with the IIV. However, the instantaneous block observed with the IIV and the incomplete relief of block by depolarization to +200 mV in 3 mM Ni²⁺ suggest a faster, lower-affinity process with different ion and voltage dependence (Fig. S1 A). Is the low-affinity site in the pore, or an allosteric site, perhaps related to the site that binds Ni²⁺ with high affinity in Ca_V3.2?

We next examined effects of Ni²⁺ at higher concentrations of Ca²⁺ and Ba²⁺ (110 mM). Pore block should be reduced (and slowed) in those conditions, while an allosteric Ni²⁺ site might not be sensitive to Ca²⁺ or Ba²⁺.

Averaged IIV relationships are shown in Fig. 8 (A and B) and conductance ratios (Ni²⁺/control) in Fig. 8 C. Chord conductances were reduced by ∼15% in Ca²⁺ at all voltages, although measurement was less accurate near the reversal potential. Effects of Ni²⁺ were indistinguishable in Ca²⁺ vs. Ba²⁺ above 0 mV, but peak tail current amplitudes were reduced more strongly in Ba²⁺ at negative voltages. One crucial result is that “instantaneous” block by 3 mM Ni²⁺ is substantially weaker in 110 mM Ca²⁺ (Fig. 8 C) than in 2 mM Ca²⁺ (Fig. 2 C). Sample records are shown in Fig. S4.

Tail currents were accelerated by 3 mM Ni²⁺, and the effect was clearly greater in 110 mM Ba²⁺ vs. 110 mM Ca²⁺ (Fig. 8, D vs. E). Analyzed on the assumption that the faster tail currents reflect time-dependent block, the rate of block by Ni²⁺ was voltage dependent, 2–3× faster in Ba²⁺ than in Ca²⁺ (Fig. 8 F). From −80 to −10 mV, the rate of block increased e-fold for 35 mV (δ = 0.36). Assuming bimolecular kinetics, the rate of block at −80 mV in 3 mM Ni²⁺ was (1.1 ± 0.1) × 10⁵ M⁻¹s⁻¹ in Ca²⁺ (n = 5) and (2.7 ± 0.1) × 10⁵ M⁻¹s⁻¹ in Ba²⁺ (n = 4). Raising either Ca²⁺ or Ba²⁺ from 2 mM to 110 mM slowed Ni²⁺ entry by ∼10-fold.

Effects of 3 mM Ni²⁺ on currents carried by 110 mM Ca²⁺ or Ba²⁺ with the I-V protocol are shown in Fig. 8, G and H (and sample current records, Fig. S5). As with 2 mM Ca²⁺ or Ba²⁺, Ni²⁺ inhibited inward currents more strongly, and inhibition was stronger in Ba²⁺ than in Ca²⁺. However, block was less voltage dependent with 110 mM Ca²⁺ or Ba²⁺ (e-fold for 86 mV, δ = 0.15; K_d at 0 mV = 3.7 mM for Ca²⁺ and 2.2 mM for Ba²⁺), noting that the data deviate in detail from a Woodhull model. Inhibition of inward currents was stronger with the I-V protocol (Fig. 8 I) than with the IIV protocol (Fig. 8 C), but again the difference was less dramatic than with 2 mM Ca²⁺ or Ba²⁺.

The slow rates of Ni²⁺ entry could reflect competitive interactions with permeant divalent cations, but it is also possible that dehydration of Ni²⁺ could be rate limiting for entry into the pore (Winegar et al., 1991). To test this, we recorded currents in the nominal absence of Ca²⁺_o (and with no extracellular EGTA, which binds Ni²⁺ with high affinity). Contaminating Ca²⁺_o was estimated to be <3 μM; Khan et al., 2008), and symmetrical 145 mM Na⁺ was the charge carrier. With the IIV protocol (Fig. S6 A), inward tail currents were accelerated, and reduced slightly in amplitude, by 30 μM Ni²⁺. That same concentration of Ni²⁺ produced strong inhibition of inward, but not outward, currents in the I-V protocol (Fig. 9, F–I; Fig. S6 B).

Averaged IIV relationships are shown in Fig. 9 A, for 10, 30, and 100 μM Ni²⁺. Detailed interpretation is difficult, since even in the control condition there was partial block by Ca²⁺_o (maximal near −30 mV) and by Mg²⁺_i (maximal at the most positive voltages), visible in the control chord conductances (Fig. 9 B) (see Khan et al., 2008). Qualitatively, Ni²⁺ block was weak at all voltages at 10 μM, but reached ∼50% inhibition below −100 mV for 100 μM. Block was much stronger with the I-V protocol (Fig. 9, F–I), and could be well described by a Woodhull (1973) model with K_d = 13 μM at 0 mV, e-fold for 28 mV (δ = 0.45) (Fig. 9 I).

With Na⁺ as charge carrier, tail currents were markedly accelerated by micromolar concentrations of Ni²⁺ (Fig. 9 D and Fig. S5). Ni²⁺ block was voltage and concentration dependent (Fig. 9 E). At −80 mV, the bimolecular block rate was (3.0 ± 0.3) × 10⁷ M⁻¹s⁻¹ (n = 12, 10–100 μM Ni²⁺). The rate of block, fitted from −50 to −100 mV, varied e-fold for 78 mV (δ = 0.16).

Fig. 10 summarizes the bimolecular rate constants for Ni²⁺ block in the ionic conditions examined in this study. The rate of Ni²⁺ entry varied by >1,000-fold, from (4.8 ± 0.3) × 10⁷ M⁻¹s⁻¹ (nominally Ca²⁺ free, −120 mV) to (3.1 ± 0.2) × 10⁴ M⁻¹s⁻¹ (110 mM Ca²⁺, −20 mV).

Fig. 11 demonstrates that high-affinity inhibition of Ca_V3.2 by 20 μM Ni²⁺ does not exhibit the key kinetic signatures of pore block. Inhibition of Ca_V3.2 was nearly independent of voltage over a 350-mV range, and was comparable with the I-V and IIV protocols (Fig. 11, D and E; Fig. S8). Inhibition was not any stronger in 2 mM Ba²⁺ than in Ca²⁺ (Fig. 11, B and C; Fig. S9). There was little effect on the kinetics of tail currents (Fig. 11, F and G); tails were ∼20% faster in Ni²⁺ with Ba²⁺ (P = 0.02–0.04 from −80 to −120 mV), possibly indicating a small amount of slow pore block. These results suggest that the kinetic and molecular mechanisms of Ni²⁺ action are fundamentally different for high-affinity inhibition of Ca_V3.2, vs. pore block of Ca_V3.1.

We propose that Ni²⁺ blocks the Ca_V3.1 pore at two sites, a relatively superficial low affinity site, and a site deeper within the pore where Ni²⁺ entry is strongly affected by permeant ions (Fig. S1 B). We will first describe how this hypothesis can explain the major features of our results, as well as a substantial number of subtle features. We next consider alternative interpretations, and relate our findings to previous views of Ni²⁺ action on calcium channels.

Ni²⁺ inhibited currents measured “instantaneously” following strong depolarization. This suggests a site accessible to Ni²⁺ at near the diffusion-limited rate. However, this component of Ni²⁺ block was low affinity, with K_d = 1–3 mM (in 2 mM Ca²⁺ or Ba²⁺). Fast block had a weak voltage dependence, suggestive of a primary dependence on the direction of current flow (Fig. 2, C and F). Fast block was similar with Ca²⁺ or Ba²⁺ as the charge carrier, but was reduced in 110 mM Ca²⁺ or Ba²⁺ (Fig. 8 C). The effect of high Ca²⁺ or Ba²⁺, and the voltage or current dependence, suggests that this site is in the ion permeation pathway (as opposed to a completely extracellular site that affects ion flow allosterically). This is supported by the observation that relief of slow block is inhibited by Ni²⁺ at the high concentrations that produce significant fast block (Fig. 7 C), suggesting a weak “lock-in” effect. That effect should be weak, since the “steady-state” outward current is inhibited by <50% at >+150 mV in 3 mM Ni²⁺ (Fig. 6 C), implying <50% occupancy of the fast block site.

The most plausible location for the fast block site is therefore near the outer end of the permeation pathway, perhaps in an outer vestibule, which Ni²⁺ might be able to access readily (perhaps even without dehydrating). If that site is outside the electrical field of the membrane, Ni²⁺ occupancy would not be directly voltage-dependent, but it seems plausible that an outward current could relieve Ni²⁺ block at this superficial site.

Slow block by Ni²⁺ shares several properties of block by other di- and trivalent cations, but there are interesting differences as well. Block is relieved by strong depolarization, as observed for Ca²⁺ (Lux et al., 1990), Mg²⁺ (Serrano et al., 2000), and Y³⁺ (Obejero-Paz et al., 2004). The rate of Ni²⁺ entry is voltage dependent, and is much slower with Ca²⁺ than with Ba²⁺, as observed for block of closed channels by Y³⁺. However, the apparent electrical distance (δ) is 0.4, vs. 0.1 for Y³⁺ (Obejero-Paz et al., 2004), in 2 mM Ca²⁺. Note that this δ value, estimated from the entry rate, is the apparent electrical distance to the energy barrier that limits Ni²⁺ entry, not the distance to the actual binding site. Curiously, δ varied with ionic conditions (see below). The bimolecular rates for Ni²⁺ entry are much lower than for Ca²⁺ (3 × 10⁸ M⁻¹s⁻¹ near −60 mV; Khan et al., 2008) or Y³⁺ (10⁹ M⁻¹s⁻¹ at −80 mV; Obejero-Paz et al., 2004).

Taken together, the most plausible site for slow block by Ni²⁺ is the selectivity filter for Ca²⁺. Ni²⁺ appears to have considerable difficulty in entering this site, even though it is thought to be physically near the extracellular side of the pore. This may involve both slow dehydration and competitive ion–ion interactions. These two factors may be related, since compared with many other di- and trivalent cations, Ni²⁺ entry might require more complete replacement of waters of hydration with the side chains of the glutamate and aspartate residues of a Ca_V3 selectivity filter, which could increase the ability of ions within the pore to prevent Ni²⁺ entry. Perhaps some ions (e.g., Y³⁺) can aggressively push any competing Ca²⁺ or Ba²⁺ ions through the pore, while Ni²⁺ must wait patiently in the low-affinity fast-block site until Ca²⁺ or Ba²⁺ ions permeate. In that view, most ions pass quickly by the site where Ni²⁺ occupancy produces a fast block. That could explain the lack of Ca²⁺–Ba²⁺ selectivity for effects on fast block by Ni²⁺.

Why is the δ for Ni²⁺ entry so large, and so ion dependent? Those two features together suggest that the large δ is a consequence of strong interactions with other ions within the pore, rather than an accurate measure of an electrical location within the pore. The ion dependence predicts that the δ for Ni²⁺ entry should approach the “true” value in the absence of competition, which would explain the observation that δ is lower in Na⁺ (0.16, vs. ∼0.4 in 2–110 mM Ca²⁺ or Ba²⁺).

It is noteworthy that the extent of block, measured from I-V relations, was less ion dependent than the rate of block. Furthermore, the apparent voltage dependence of block depended in the opposite direction on ion concentration, with δ = 0.45 in Na⁺, δ = 0.19 in 2 mM Ca²⁺ or Ba²⁺, and δ = 0.15 in 110 mM Ca²⁺ or Ba²⁺. This can be explained in part by the low-affinity Ni²⁺-binding site, which is relatively insensitive to ionic conditions and to voltage. That site would make the largest contribution to the net block by Ni²⁺ when “high-affinity” slow block is the weakest (in high Ca²⁺). In addition, the measured δ could be in error if Ni²⁺ affects gating, or if slow Ni²⁺ block has not reached steady state at the time of peak current.

This two-site model for pore block by Ni²⁺ provides an attractive qualitative explanation for the behavior of Ni²⁺ block under a variety of ionic conditions, and a wide range of voltages. However, we are well aware that this will need to be confirmed by molecularly based modeling of ion permeation in calcium channels, at a sufficient level of detail to account for the complex organometallic chemistry of Ni²⁺ ions.

The Ni²⁺ entry rate was sensitive to the identity of the permeant divalent cation (Ca²⁺ vs. Ba²⁺), even at 110 mM concentrations where flux through the pore is saturated (Khan et al., 2008). It is often assumed that saturation of flux implies saturation of binding site(s) within the pore, but that is not necessarily true.

As for any pharmacological agent, there are several possible mechanisms by which an ion such as Ni²⁺ can affect the behavior of a voltage-dependent ion channel, including pore block, shifts in gating by screening or binding to surface charge, and allosteric modulation by binding to extra-pore sites.

There is good reason to expect that Ni²⁺ will shift channel activation to more positive potentials, at least in part by a surface charge mechanism. Effects of three other divalent cations (Ca²⁺, Ba²⁺, and Mg²⁺) on gating were well described by screening a surface charge of 1 e⁻ per 98 Ų, which predicts (e.g.) that addition of 1 mM divalent cation to 2 mM Ca²⁺ or Ba²⁺ would shift channel activation by ∼3 mV. It is possible that Ni²⁺ affects surface charge more strongly than Ca²⁺, Ba²⁺, and Mg²⁺; for example, Fe²⁺ and Mn²⁺ shift gating as expected for both screening and low-affinity binding to surface charge (Lopin, K.V., I.P. Gray, C.A. Obejero-Paz, and S.W. Jones. 2007. Biophysical Society meeting. Abstract 2867; Jamieson, Q., C.A. Obejero-Paz, and S.W. Jones. 2008. Biophysical Society meeting. Abstract 3129).

The effect of Ni²⁺ on tail kinetics is in the direction expected for a surface charge effect: that is, faster tail currents (measured at a particular voltage) could in principle result from a positive shift along the voltage axis. However, where the effect of Ni²⁺ on tail current time constants was the strongest (2 mM Ba²⁺, Fig. 3 B; nominally Ca²⁺ free, Fig. 9 D), the effect was not a pure parallel shift along the voltage axis. Furthermore, the effect of Ni²⁺ to speed tail currents more profoundly in Ba²⁺ vs. Ca²⁺ would not be expected from a surface charge mechanism. Finally, this interpretation would imply that it is merely coincidental that the effect on tail kinetics is well approximated by bimolecular kinetics with a voltage-dependent rate for Ni²⁺ block (Fig. 3, C and D, Fig. 8 F, and Fig. 9 E). Thus, we conclude that Ni²⁺ speeds tail currents primarily by voltage- and time-dependent channel block, not by effects on channel gating.

Several figures in this paper present the fractional inhibition by Ni²⁺ of the peak current measured from I-V relationships (Fig. 4, C and F, Fig. 8 I, and Fig. 9 I). Inhibition can result from voltage-dependent block by Ni²⁺ within the ion permeation pathway, but a positive shift in channel activation would also lead to a voltage-dependent reduction in current amplitudes. This complicates interpretation of the Woodhull (1973) fits to the data from the I-V protocol. Specifically, if Ni²⁺ shifts activation to more positive voltages (even by a few mV), inhibition at more negative voltages would reflect a combination of pore block and a reduction in the probability that a channel is open. It is important to recognize that a positive shift in the I-V curve (or derived measures such as the G-V) could result either from a true shift in channel activation, or from voltage-dependent block. For a fast blocker (e.g., Mg²⁺) pore block can be measured as a reduction in the IIV, and shifts in gating from the I-V/IIV ratio (P_O,r), but separation of the two effects is difficult for a slow blocker such as Ni²⁺. If block is very slow, the block observed at the time of peak current with the I-V protocol may not fully reflect steady-state block, as noted previously for other channel blockers (Thévenod and Jones, 1992; Block et al., 1998; Babich et al., 2007).

In principle, surface charges could also affect ion permeation, if the charges are close enough to the permeation pathway to affect the local concentration of permeating and blocking cations. However, this effect appears to be minimal for voltage-dependent calcium channels (Kuo and Hess, 1992; Block et al., 1998).

One possible explanation of the low-affinity inhibition by Ni²⁺ of the IIV relationship is that Ca_V3.1 retains a low-affinity extracellular site for Ni²⁺ block, homologous to the high-affinity site in Ca_V3.2. Perhaps Ni²⁺ traps a voltage sensor in the “down” position, analogous to voltage sensor toxins (Li-Smerin and Swartz, 1998; McDonough, 2007). That could slow activation, and/or keep some channels from opening at all. We cannot fully exclude this, but that mechanism would not easily explain the reduced fast block in 110 mM Ca²⁺ or Ba²⁺. Furthermore, concentrations of Ni²⁺ that occupy that low-affinity site “lock in” Ni²⁺ block at the deep pore site, suggesting a more direct interaction (which we propose to be at the extracellular side of the permeation pathway). Note also that high-affinity inhibition of Ca_V3.2 is not reversed even by depolarization to +200 mV (Fig. 11), consistent with simple voltage-independent inhibition (Diaz et al., 2005; Kang et al., 2005).

This study illustrates how the basic I-V and IIV protocols, combined with an analysis of ion dependence, can go a long way toward determining the mechanisms of action of an ion such as Ni²⁺ on a calcium channel. A discrepancy between the extent of inhibition measured from the I-V and IIV protocols can be explained either by time-dependent block, or by modified channel activation. Acceleration of tail currents may result either from time-dependent open-channel block or from a shift in activation, but differential effects on tail kinetics in Ca²⁺ vs. Ba²⁺ strongly suggest a pore block mechanism. That is especially true for Ca_V3.1, where changes in Ca²⁺ and Ba²⁺ concentration shift the voltage dependence of gating in the same manner, implying similar “surface charge” effects of Ca²⁺ and Ba²⁺ on channel activation. Analysis of tail current acceleration in terms of pore block can test that hypothesis by comparison to expectations for a bimolecular reaction, with voltage-dependent rates. Conversely, analysis based solely on effects on I-V relationships may not be sufficient for mechanistic analysis. For example, Lacinova et al. (2000) concluded that Ni²⁺ modified activation kinetics for Ca_V3.1, by slowing channel deactivation (more strongly in 20 mM Ba²⁺ vs. Ca²⁺). Their experimental results are consistent with ours, but we conclude instead that Ni²⁺ accelerates tail currents primarily by time- and voltage-dependent block, vs. a true modulation of the kinetics of channel closing.

We propose that Ni²⁺ block of Ca_V3.1 involves a low affinity site with rapid kinetics, and a high affinity site within the “deep” pore. This agrees quite well with single-channel data on Ni²⁺ block of L-type calcium channels (Winegar et al., 1991), where Ni²⁺ not only reduced the apparent single channel conductance, but also produced discrete blocking events. The rate for discrete block of L channels by Ni²⁺ was also quite slow, 6 × 10⁴ M⁻¹s⁻¹ at 0 mV in 110 mM Ba²⁺ (Winegar et al., 1991), compared with 3 × 10⁴ M⁻¹s⁻¹ for Ca_V3.1 (from Fig. 8 F). In addition, the rates of block and relief of block both increased with hyperpolarization (Winegar et al., 1991), consistent with our observation that the rate of block by Ni²⁺ was more voltage dependent than the steady-state block (which depends on the off/on rate ratio). We do not have direct measurements of the exit rate, but note that an exit rate that increases with hyperpolarization suggests that Ni²⁺ can exit the pore to the intracellular site (i.e., Ni²⁺ is a weakly permeant ion).

Many previous studies, on T channels (Lee et al., 1999; Kang et al., 2005) and other calcium channels (Zamponi et al., 1996; McFarlane and Gilly, 1998; Magistretti et al., 2001), concluded that Ni²⁺ modified channel gating, most likely by an extra-pore site. Our results do not provide compelling evidence for such an effect for Ca_V3.1. This is consistent with the interpretation that the high-affinity site near the extracellular side of S4 in Ca_V3.2 is essentially absent in Ca_V3.1 (Kang et al., 2005).

We thank Dr. Ed Perez-Reyes for the gift of cells stably transfected with Ca_V3.1, and for a plasmid encoding Ca_V3.2.

This work was supported by National Institutes of Health grant NS24771 to S.W. Jones.

Olaf S. Andersen served as editor.