Activation of the SK potassium channel-calmodulin complex by nanomolar concentrations of terbium

Abstract

Small conductance Ca²⁺-activated K⁺ (SK) channels sense intracellular Ca²⁺ concentrations via the associated Ca²⁺-binding protein calmodulin. Structural and functional studies have revealed essential properties of the interaction between calmodulin and SK channels. However, it is not fully understood how the binding of Ca²⁺ to calmodulin leads to channel opening. Drawing on previous biochemical studies of free calmodulin using lanthanide ions as Ca²⁺ substitutes, we have used the lanthanide ion, Tb³⁺, as an alternative ligand to study the activation properties of SK channels. We found that SK channels can be fully activated by nanomolar concentrations of Tb³⁺, indicating an apparent affinity >100-fold higher than Ca²⁺. Competition experiments show that Tb³⁺ binds to the same sites as Ca²⁺ to activate the channels. Additionally, SK channels activated by Tb³⁺ demonstrate a remarkably slow deactivation process. Comparison of our results with previous biochemical studies suggests that in the intact SK channel complex, the N-lobe of calmodulin provides ligand-binding sites for channel gating, and that its ligand-binding properties are comparable to those of the N-lobe in isolated calmodulin.

Small conductance Ca²⁺-activated K⁺ (SK, K_Ca2) channels sense Ca²⁺ concentrations via the associated Ca²⁺-binding protein calmodulin (CaM) (1). CaM consists of two globular lobes connected by a flexible linker, referred to as the N-lobe and the C-lobe. Each of the two lobes contains two EF hand motifs that bind Ca²⁺. Ca²⁺ binding to the EF hands of CaM constitutively associated with channel subunits leads to rapid opening of the SK channel pore (2). Structural and functional studies have provided important insights into the molecular mechanism for the coupling between CaM and SK channels (2). The crystal structure of the intracellular CaM-binding domain (CaMBD) (96 aa) of the SK channel was solved in association with Ca²⁺-loaded CaM. This structure suggests that the C-lobe of CaM mediates the constitutive interaction with the CaMBD of SK channels. As a result of this interaction, the C-lobe loses its ability to bind Ca²⁺. The N-lobe sites are loaded with Ca²⁺ in the structure, suggesting that Ca²⁺ activation of SK channels is the result of Ca²⁺ binding to the two EF hands in the N-lobe of CaM (3). Functional studies with CaM mutants are consistent with this idea: Mutations at the N-lobe of CaM that abolish Ca²⁺ binding dramatically affect the Ca²⁺ gating of SK channels, whereas the equivalent mutations at the C-lobe have no effect (4). However, the molecular mechanisms for the CaM-SK coupling emerging from these studies are less than conclusive because of some experimental limitations. The structure only includes a relatively small segment of the SK channel (3); therefore, it is necessary to verify that it correctly depicts the interaction between CaM and full-length SK channels. Also, the interpretation of CaM mutational studies was complicated by the presence of endogenous wild-type CaM in the expression system.

Understanding of the gating mechanisms for ligand-gated ion channels has been facilitated by the use of a collection of chemically related ligands that interact differently with the channels. Such an approach has not been useful in the study of SK channels, however. In contrast to other channel types gated by small organic molecules for which many ligands are often available, SK channels are activated by an ion. Furthermore, SK channels have high selectivity for Ca²⁺ over other divalent ions, making these less useful as alternative ligands (5). However, trivalent lanthanide ions such as terbium (Tb³⁺) and europium (Eu³⁺) ions have been frequently used to substitute for Ca²⁺ in biochemical studies of Ca²⁺-binding proteins such as CaM. Lanthanide ions can bind to the Ca²⁺-binding sites because of their similar ionic size and coordinating properties as Ca²⁺. The unique spectroscopic properties of some lanthanide ions allow for optical measurement of ligand binding, which has helped elucidate the affinity and order of Ca²⁺ binding to the four Ca²⁺-binding sites of CaM (6–9). Given that CaM serves as the Ca²⁺ sensor for SK channels, we have investigated the activation of SK channels by lanthanide ions (mainly Tb³⁺), with the expectation that they may serve as alternative ligands to Ca²⁺, and that functional differences between lanthanide and Ca²⁺ ions can enhance the understanding of the gating mechanisms of SK channels. By directly applying Tb³⁺ to heterologously expressed SK channels, we found that like Ca²⁺, Tb³⁺ can fully activate SK channels. Comparing the functional effects of Ca²⁺ and Tb³⁺ on the activation and deactivation of SK channels provides insights into the coupling mechanism between CaM and SK channels.

Results

Although lanthanide ions including Tb³⁺ have been shown to bind to the Ca²⁺-binding sites and activate purified CaM protein (8, 10–13), their functional effects on CaM-dependent ion channels have not been studied. We directly applied Tb³⁺ ions to excised patches from Xenopus oocytes heterologously expressing SK channels to test their effects on activation.

In previous studies on Ca²⁺-activated channels including SK channels, Ca²⁺ chelators have been used in internal solutions to control free Ca²⁺ concentrations. However, these chelators have extremely high affinity for Tb³⁺ and other lanthanide ions (stability constants >10¹⁵ M⁻¹) (14). It is not feasible to achieve the desired Tb³⁺ concentrations in the presence of any of these Ca²⁺ chelators. Additionally, appropriate selective chelators for lanthanide ions at nanomolar to micromolar concentration are not available. We decided to use chelator-free solution (CFS) to test the effect of Tb³⁺. However, without Ca²⁺ chelators we need to reduce the contaminating Ca²⁺ in our solution (usually a few μM, sufficient to saturate the activation of SK channels). By using columns made of Chelex 100 resin (see Materials and Methods), we were able to reproducibly reduce the Ca²⁺ contamination to approximately 300 nM level, as measured with Fura-2. Tb³⁺ or Ca²⁺ was then added to the decalcified CFS to achieve the desired concentrations.

To verify our approach, we first measured the activation of SK channels by Ca²⁺ using CFS and added Ca²⁺. Contaminating Ca²⁺ (≈300 nM) alone in CFS did not significantly activate SK channels (Fig. 1A, trace 1), but larger SK currents emerged when additional Ca²⁺ was added (Fig. 1A traces 2–7 and B). Current levels at −80 mV are normalized and plotted as a function of total Ca²⁺ concentration (contaminating + added), which is fitted with the Hill equation I/I_max = 1/(1+(EC₅₀/[Ca²⁺])^h) (Fig. 1C). Average results from 11 patches show that Ca²⁺ activates SK channels with a half activation concentration (EC₅₀) of approximately 1.2 μM, and Hill coefficient (h) of approximately 3.7 (Table 1). To compare these results with previous measurements in the presence of Ca²⁺ chelators, in eight of the 11 experiments, we also measured Ca²⁺ dose-response relationships by using chelator-containing Ca²⁺ solutions on the same patches (see Materials and Methods). Results from one such experiment are shown in Fig. 1D. The dose-response relationships obtained using the two methods are very similar (Table 1), suggesting that the use of CFS provides reliable measurements of the dose-response relationship for SK channel activation, and that the results using CFS are comparable to the measurements using Ca²⁺ chelators. Our measured EC₅₀ values for Ca²⁺ activation of SK channels are somewhat different from previous studies (≈0.3–0.5 μM) (1, 4, 15), likely a result of different ways to determine free Ca²⁺ concentrations in the presence of Ca²⁺ chelators. Alternatively, this discrepancy could have resulted from differences in the phosphorylation status of CaM (16).

Fig. 1.

Activation of SK channels by Ca²⁺. (A) Representative SK current traces elicited by a voltage ramp from −80 to 100 mV at different total Ca²⁺ concentrations ([Ca²⁺]). Currents were recorded from an inside-out patch pulled from an oocyte expressing SK channels. To achieve desired [Ca²⁺], calculated amounts of Ca²⁺ stock solution were added to the bath (CFS with 350 nM contaminating Ca²⁺) in a Petri dish before thorough mixing by pipeting. The [Ca²⁺] (350 nM + added amount) for individual traces are: 1, 0.35 μM; 2, 0.55 μM; 3, 0.75 μM; 4, 0.95 μM; 5, 1.15 μM; 6, 1.35 μM; and 7, 6.15 μM. (B) SK current level at −80 mV measured every three seconds while Ca²⁺ was added to the bath. Data points during mixing were noisy and removed from the plot. Numbers correspond to the traces in A. (C) Mean current level at −80 mV at each [Ca²⁺] normalized to the maximal value at 6.15 μM and plotted as a function of [Ca²⁺]. Solid line represents fit of the data with the Hill equation (EC₅₀ = 1.01 μM, Hill coefficient h = 3.66). (D) Ca²⁺ dose-response relationship was measured with both CFS (solid circles) and chelator-containing Ca²⁺ solutions (open circles) in a same patch (see Materials and Methods). Fits with Hill equation yield: CFS, EC₅₀ = 1.35 μM, h = 2.87 (solid line); with chelators, EC₅₀ = 1.29 μM, h = 3.21 (dashed line).

Table 1.

Different functional effects of Ca²⁺ and Tb³⁺ on SK channel gating

	Apparent affinity for activation			Deactivation kinetics
	EC50	Hill coefficient, h	No. of trials	Time constant, τ	# of trials
Ca2+ + CFS	1.17 ± 0.24 μM	3.68 ± 0.82	n = 11	51.6 ± 10.6 ms	n = 6
Ca2+ + Chelator	1.27 ± 0.24 μM	3.45 ± 0.73	n = 8	78.4 ± 18.5 ms	n = 7
Tb3+	5.17 ± 2.86 nM	1.74 ± 0.47	n = 37	8.28 ± 3.45 s	n = 10

Values are reported as mean ± SD

Activation of SK channels by Tb³⁺ was similarly measured by adding Tb³⁺ to the bath solution. Fig. 2 A and B shows that addition of 2 nM Tb³⁺ was sufficient to activate significant amount of SK channels, although with a slower time course compared with μM range Ca²⁺ (compare Figs. 2B and 1B). Greater SK currents were activated with increasing Tb³⁺ concentration, until saturation at 40–80 nM total Tb³⁺ (Fig. 2 A and B). The SK currents activated by Tb³⁺ have similar current-voltage relationship to those activated by either Ca²⁺ in CFS (Fig. 1A), or Ca²⁺ solutions with chelators (data not shown), and to SK currents in previous studies (for example, 1, 4). As shown by trace 6 (gray) in Fig. 2A and the time course in Fig. 2B (compare 6 and 5), after saturation of activation with 80 nM total Tb³⁺, addition of 10 μM Ca²⁺ in the presence of Tb³⁺ did not elicit further SK current, although it slightly reduced the current at positive potentials, presumably because of Ca²⁺ block (5). This result suggests that like Ca²⁺, Tb³⁺ is a full agonist for SK channels.

Fig. 2.

Activation of SK channels by Tb³⁺. (A) Representative SK current traces elicited by a voltage ramp from −80 to 100 mV at different total Tb³⁺ concentration ([Tb³⁺]). The recording conditions were as in Fig. 1A. The bath solution (CFS) contained 300 nM contaminating Ca²⁺. To achieve desired [Tb³⁺], appropriate amounts of 1 μM Tb³⁺ stock solution were added to the bath before thorough mixing by pipeting. The [Tb³⁺] for individual traces are: 1, 0 nM; 2, 2 nM; 3, 5 nM; 4, 10 nM; 5, 80 nM; and 6, 80 nM +10 μM Ca²⁺ (gray trace). (B) SK current level at −80 mV measured every three seconds while Tb³⁺ was added to the bath. Data points during mixing were removed from the plot. Numbers correspond to the traces in A. (C) Mean current levels at −80 mV normalized to the maximal current and plotted as a function of [Tb³⁺]. Solid line represents fit with the Hill equation (EC₅₀ = 3.47 nM, h = 1.93). (D) Tb³⁺ (solid squares) and Ca²⁺ (solid circles) dose-response relationships measured in a same patch (see Materials and Methods). Solid lines are fits with the Hill equation (Tb³⁺: EC₅₀ = 3.60 nM, h = 2.53; Ca²⁺: EC₅₀ = 0.81 μM, h = 4.06).

The dose-response relationship for Tb³⁺ activation of SK channels shown in Fig. 2C is fitted with the Hill equation (EC₅₀ = 3.47 nM and h = 1.93). In this fit, the contribution by the contaminating Ca²⁺ (≈300 nM) is ignored. Exactly how the contaminating Ca²⁺ affects the measured Tb³⁺ activation of SK channels depends on the unknown interaction between Ca²⁺ and Tb³⁺ when they both bind to the same channels. However, it is likely that the effect of contaminating Ca²⁺ on the values of EC₅₀ and h for Tb³⁺ is small, because the concentration of contaminating Ca²⁺ in CFS is only approximately 1/4 of the measured EC₅₀ for Ca²⁺ activation, and it alone usually activates <1% of the SK channels. Regardless of the exact mechanism for the Ca²⁺-Tb³⁺ interaction, it is evident that approximately 5 nM Tb³⁺ can substitute for approximately 900 nM Ca²⁺ to half-maximally activate the SK channels in the presence of approximately 300 nM contaminating Ca²⁺. Therefore, Tb³⁺ activates SK channels with >100-fold higher apparent affinity than Ca²⁺. Based on individual fits of data from 37 patches (neglecting the contribution by contaminating Ca²⁺), the average results for Tb³⁺ activation of SK channels yielded EC₅₀ = 5.17 ± 2.86 nM, h = 1.74 ± 0.47. The h values for Tb³⁺ activation are significantly lower compared with values for Ca²⁺ activation of SK channels (Student's t test, P < 0.01) (Table 1), suggesting less cooperativity in activation by Tb³⁺ than by Ca²⁺. However, it should be kept in mind that Hill equation is only a simple empirical description of the dose-response relationship. For an SK channel complex with multiple CaMs, each containing multiple binding sites, we cannot make further conclusions about cooperativity without a more thorough understanding of the activation mechanism.

We noticed a rather large variation in EC₅₀ and h values for Tb³⁺ activation of SK channels among different patches. The low concentrations of Tb³⁺ used and the absence of chelators in our solutions make our measurements more sensitive to even trace amounts of contamination and inaccuracies in determination of concentration. Other possible reasons for this variation include variable local Tb³⁺ concentrations because of different levels of endogenous chelating molecules and surface charges across patches, and variability in the properties of the SK channel complex related to the phosphorylation status of CaM (16). Occasionally (<10%), we came across EC₅₀s for Tb³⁺ activation that were >3-fold higher than the average value. Those experiments were considered as outliers and excluded from further analysis (Chauvenet's criterion).

To compare Ca²⁺ and Tb³⁺ activation of SK channels more directly, we measured both Ca²⁺ and Tb³⁺ dose-response relationships from the same patches (see Materials and Methods). Fig. 2D shows the Ca²⁺ and Tb³⁺ dose-response relationships from one such experiment. Average results from five such patches show that Ca²⁺ activates SK channels with EC₅₀ = 0.91 ± 0.21 μM, h = 4.30 ± 1.51; and for Tb³⁺, EC₅₀ = 3.47 ± 1.05 nM, h = 2.16 ± 0.84, confirming that within the same patches Tb³⁺ activates SK channels with >100-fold higher apparent affinity than Ca²⁺.

Previous biochemical studies indicated that the dissociation rates for Tb³⁺ from CaM are much slower than Ca²⁺ (17, 18), as would be expected by the difference in affinities (19). To test whether this difference in dissociation is reflected in the kinetics of SK channels, we compared the deactivation of SK channels when Ca²⁺ or Tb³⁺ is removed from CaM. To minimize rebinding of ligand during deactivation, we quickly moved the patch into Ca²⁺-free solution (5 mM EGTA). As shown in Fig. 3A, when SK channels are activated by Ca²⁺ alone, the deactivation process demonstrates a fast decay of current, which can be fitted well by using a single exponential time course with time constants of approximately 50 ms (Table 1). This time constant is comparable with the measured values using an automated fast solution changing system (20). When SK channels are activated by a saturating concentration of Tb³⁺, the majority of the current decay during deactivation can also be fitted with a single exponential time course (Fig. 3B), but with a time constant of approximately 10 s (Table 1). The deactivation process for Tb³⁺-activated SK channels is therefore >100-fold slower than that for SK channels activated by Ca²⁺. This difference would be expected from the >100-fold higher apparent affinity under the conditions where the deactivation rate is limited by ligand dissociation, and binding rates for Tb³⁺ and Ca²⁺ are similar. In some experiments, including the one shown in Fig. 3B, a small fraction of the current at the beginning of the deactivation follows a faster time course, which is likely because of the presence of the contaminating approximately 300 nM Ca²⁺ in the CFS (see below).

Fig. 3.

Deactivation kinetics of SK channels activated by Ca²⁺ or Tb³⁺. (A) After the activation of SK channels by CFS +3 μM Ca²⁺ stabilized, current (gray) at −80 mV was recorded at 100-μs intervals while the recording pipette was quickly moved into a laminar flow of the Ca²⁺-free solution (5 mM EGTA, see Materials and Methods). Time 0 is arbitrarily chosen for illustration purpose in this and the following figures. Dark solid line is a fit with single exponential time course (τ = 58 ms). (B) In a different patch, after SK channel activation stabilized in CFS + 80 nM Tb³⁺, deactivation was measured as in A. Dark solid line is a fit with single exponential time course (τ = 10.5 s).

If Tb³⁺ and Ca²⁺ bind to the same sites in CaM associated with SK channels, their binding should be competitive. The large difference in deactivation kinetics allowed us to test directly for competition. For this purpose, we added 10 μM Ca²⁺ to the bath solution in the presence of 80 nM Tb³⁺ after the SK channels were fully activated with Tb³⁺. When equilibrium was reached after thorough mixing and >3-min incubation, no further SK current was activated by this extra Ca²⁺ (data not shown). However, the deactivation kinetics were very different from when the channels were activated by 80 nM Tb³⁺ alone (compare Figs. 4A and 3B). Fitting the current decay requires at least two major exponential components, one with a time constant of approximately 1 s, the other approximately 10 s. Average results from five similar experiments indicate that approximately 70% of the SK channels deactivate with the slow time course, whereas approximately 30% deactivate with the fast time course. The significant presence of this fast component in deactivation suggests that Ca²⁺ can indeed compete with Tb³⁺ for the same ion-binding sites in CaM. The slower time course is comparable with the deactivation of SK channels activated by Tb³⁺ alone, whereas the fast time course is still much slower than expected for channels activated by Ca²⁺ alone (≈50 ms). To measure the deactivation of SK channels activated by Ca²⁺ alone in the same patches, we reactivated SK channels by using chelator-containing solutions with saturating free Ca²⁺ concentration. Presumably all Tb³⁺ ions in the solution were chelated by Ca²⁺ chelators because of their extremely high affinity, and the channels should be activated only by Ca²⁺. As expected, deactivation after this treatment shows a fast, single-exponential decay, with time constants close to those for SK channels activated by Ca²⁺ alone in CFS (Fig. 4B and Table 1).

Fig. 4.

Effects of the competition between Ca²⁺ and Tb³⁺ on the deactivation kinetics of SK channels. (A) Experiment was done as in Fig. 3B except that 10 μM Ca²⁺ was added to the bath after 80 nM Tb³⁺. Bath was mixed thoroughly and incubated for >3 min before the pipette was moved into Ca²⁺-free solution. Current decay (gray) is fitted with two exponential components (dark solid line) with τ₁ = 0.93 s (49%), τ₂ = 9.91 s (51%). (B) With the same patch in A, deactivation was measured again after reactivating the channels with chelator-containing solution (15.4 μM free Ca²⁺). Current trace (Left, gray) is plotted with the trace and fit from A (Right, gray trace and dark solid line) on an expanded time scale after normalized to the current level before deactivation. The new deactivation current trace is fitted with a single exponential time course (τ = 65 ms, dark solid line). Brief downward spikes in this and other traces likely resulted from the activation of stretch-activated channels. They do not affect the exponential fits. (C) Deactivation measured after stabilized activation of SK channels by 80 nM Tb³⁺ +50 μM Ca²⁺. Current decay (gray) is fitted with two exponential time course (τ₁ = 0.19 s, 64%; τ₂ = 1.06 s, 36%; dark solid line).

From Fig. 4B, it is clear that the faster component (τ = ≈1 s) of the deactivation process for SK channels activated by Tb³⁺ + Ca²⁺ is still slower than the deactivation of SK channels activated by Ca²⁺ alone (τ = ≈50 ms). The channel deactivation cannot be described by the combination of two components with the measured fast time constant for Ca²⁺ alone and the slow time constant for Tb³⁺ alone, suggesting that Ca²⁺ and Tb³⁺ can collectively activate a single channel and thus producing an intermediate time course in deactivation. Each SK channel has 4 associated CaMs, which together have at least eight functional ion-binding sites for activation (3). SK channels could have many combinations of different numbers of bound Ca²⁺ and Tb³⁺, potentially resulting in different deactivation kinetics. Fitting the deactivation with exponential time courses does not allow us to distinguish more than three components. The intermediate time constants from the fit of the deactivation likely reflect the combined result of multiple species with different numbers of bound Tb³⁺ and Ca²⁺. Consistent with the competition between Ca²⁺ and Tb³⁺, when 50 μM Ca²⁺ was added after 80 nM Tb³⁺, the slow component was virtually absent from the deactivation process. As shown in Fig. 4C, the time course of deactivation can be still fitted with two exponential components, with the longer time constant at approximately 1 s, and the shorter one at approximately 200 ms, which approaches the fast deactivation process with Ca²⁺ alone. We found that the deactivation of SK channels activated by nonsaturating Tb³⁺ concentrations (e.g., 2 nM and 5 nM) demonstrates both a slow (τ = ≈10 s) and a significant fast (τ = ≈1 s) component (data not shown), likely reflecting competition between Tb³⁺ and the contaminating Ca²⁺ (≈300 nM) in our solution. We also measured deactivation of SK channels over a range of different concentrations of competing Ca²⁺ (1–20 μM) in the presence of 80 nM Tb³⁺. Although the deactivation can always be fitted with a fast (τ = ≈1 s) and a slow (τ = ≈10 s) component, the relationship between Ca²⁺ concentration and the fraction of each component in the deactivation was too variable across different patches to allow detailed quantitative comparison. It is likely that this variability is partly a result of variable Tb³⁺ apparent affinity among different patches, and partly reflects the inadequacy of using exponential fits with only a few components to characterize a complicated deactivation process. Nevertheless, the trend from these results indicates clearly that increased Ca²⁺ concentration leads to a decrease in the amplitude of the slow component in the deactivation process, as expected for competition between fast unbinding Ca²⁺ and slow unbinding Tb³⁺.

Discussion

Activation of SK Channels by Nanomolar Tb³⁺.

In the present study, we established conditions where activation of SK channels by lanthanide ions can be directly measured with patch-clamp recording. Our results demonstrate that Tb³⁺ activates SK channels at nanomolar concentrations. We have also tested europium (Eu³⁺) and lanthanum (La³⁺) ions, and obtained results qualitatively similar to Tb³⁺ in terms of apparent affinity and kinetics (data not shown). Lanthanide ions such as Tb³⁺ and Eu³⁺ were shown with biochemical assays to bind the Ca²⁺-binding sites in purified CaM protein with higher affinity than Ca²⁺ (9, 19). The higher affinity was generally attributed to the fact that lanthanide ions are similar in size to Ca²⁺ but carry an extra charge, enhancing the association with the negatively charged EF hands. In particular, lanthanide ions were found to bind very tightly to the N-lobe of CaM (8, 21). With the help of the high quantum yield of the Eu³⁺ ion in D₂O, the binding affinity of Eu³⁺ at the N-lobe was directly measured by using 50 nM purified CaM, yielding K_d of approximately 6–12 nM (19). In our study, we used the activation of SK channels as a reporter to show that, in a functionally complete SK channel complex CaM is indeed effectively activated by a few nanomolar lanthanide ions. Our experiments did not directly measure lanthanide binding to CaM, and the opening of SK channels is unlikely to be linearly correlated to the number of bound ligands. However, regardless of the exact number of lanthanide ions required for channel opening, our measurement is in excellent agreement with the earlier estimates of the high-affinity binding of lanthanide ions to CaM (19).

In contrast to the strong biochemical interest in using lanthanide ions as a tool to study Ca²⁺ binding, it has been rarely tested whether binding of lanthanide ions in CaM is functionally equivalent to the binding of Ca²⁺ in a complete biological context. Our data demonstrate that lanthanide ions can activate SK channels through binding to the associated CaM. Competition between Ca²⁺ and Tb³⁺, as demonstrated in the measurements of channel deactivation (Fig. 4), established that lanthanide ions bind to the same sites as Ca²⁺ to activate SK channels. Both Tb³⁺ and Ca²⁺ can fully activate SK channels, and the appearance of SK currents activated by Tb³⁺ is similar to those activated by Ca²⁺ (Figs. 1 and 2). Additionally, our results indicate that Tb³⁺ and Ca²⁺ can collaboratively activate a single channel (Fig. 4B). All of the above evidence collectively suggests that lanthanide ions induce similar conformational changes in CaM as does Ca²⁺, leading to the opening of the SK channel pore. This hypothesis is consistent with previous NMR studies on EF-hand proteins suggesting that substitution of lanthanide ions for Ca²⁺ does not significantly alter the conformation of the proteins (reviewed in ref. 22).

Ligand Binding at the N-lobe of CaM Gates SK Channels.

Interestingly, previous biochemical studies showed that lanthanide ions bind to the two lobes of free CaM with very different affinities. Whereas lanthanide ions bind to the N-lobe of CaM with nanomolar affinity, they bind to the C-lobe of CaM with only slightly higher affinity than Ca²⁺, with K_d in the μM range (19, 21). In contrast, Ca²⁺ binds to the C-lobe of CaM with slightly higher affinity than to the N-lobe, but all four binding sites in both C- and N-lobe of CaM have K_d for Ca²⁺ in the μM range (23, 24). In light of this distinction, the high apparent affinity for the activation of SK channel by lanthanide ions has important implications on the gating mechanism. The nanomolar apparent affinity measured in this study suggests that lanthanide ions specifically bind to the N-lobe of CaM to activate SK channels, because at nanomolar concentration range, binding at the C-lobe does not occur.

Structural and functional studies on SK channels have suggested that the N-lobe of associated CaM is solely responsible for ligand binding (3, 4). However, in light of some experimental limitations in these studies (see Introduction), it is important to verify this proposed molecular mechanism in a more intact and complete system. In our study, SK channels are functionally associated with endogenous wild-type CaM. Although our results do not directly indicate whether the C-lobe of CaM associated with SK channels can still bind ligands, it is clear that ligand binding at the N-lobe of CaM is sufficient to fully open the channels.

Our kinetic measurements of SK channel deactivation also indicate that ligand binding at the N-lobe of CaM is responsible for channel gating. Previous biochemical experiments with purified CaM and its tryptic fragments containing individual lobes showed that Tb³⁺ dissociates from the N-lobe of CaM with time constants of a few seconds. This rate is significantly slower than the dissociation rate of Tb³⁺ from the C-lobe of CaM (τ ≈0.1–1 s), and much slower than the dissociation rate of Ca²⁺ from either lobe of CaM (τ ≈4–40 ms) (17, 18). Our data show that when SK channels are activated purely by saturating concentration of Tb³⁺, the deactivation process has a time constant of approximately 10 s. The closing of SK channels during deactivation does not provide a direct measurement of the dissociation rate of Tb³⁺. However, by comparison with the fast deactivation in the case of Ca²⁺, it is evident that dissociation of Tb³⁺ from CaM associated with SK channels has time constants on the order of a few seconds. The time course of the slow deactivation is in excellent agreement with the biochemically measured dissociation rate of Tb³⁺ from the N-lobe of CaM, but slower than would be expected for the dissociation rate from the C-lobe (18).

The above discussion was based on the assumption that lanthanide-binding properties of the two lobes of CaM (if they can still bind ligand) are not dramatically altered by the interaction with SK channels. Although cases exist where ligand binding of CaM is affected by interaction with its effector proteins (24, 25), it is unlikely that the C-lobe of CaM will be modified by the interaction with SK channel such that it mimics both the affinity and kinetic properties of the N-lobe for lanthanide binding. In light of the previous structural and mutational studies on SK channels (3, 4), and the biochemical findings that individual N and C-lobes of CaM in tryptic fragments maintain their affinity and kinetics for ligand binding as in intact protein (17, 18, 23), it is reasonable to conclude that the activation of SK channels by lanthanide ions is the result of binding to the N-lobe of CaM, which preserves its lanthanide binding properties when CaM is associated with SK channels.

Lanthanide ions have been instrumental in understanding the mechanisms of Ca²⁺-binding proteins including CaM. Our finding that lanthanide ions can fully activate SK channels by binding to the Ca²⁺-binding sites in CaM promises new approaches to study the gating mechanisms of SK channels. Related ligands with distinctive functional effects have become powerful tools in elucidating the gating mechanisms of ligand-gated ion channels such as glutamate receptor channels (reviewed in ref. 26). Lanthanide ions may prove to be the equivalent tools for the study of SK channels. Furthermore, the unique spectroscopic features of lanthanide ions in combination with electrophysiological recordings will likely provide very useful approaches to directly correlate ligand binding to channel opening in the functionally complete SK channel complex.

Materials and Methods

Channel Expression.

All experiments were performed on the rSK2 clone of the SK channel kindly provided by the Adelman lab (Vollum Institute, Oregon Health and Science University). The clone was introduced into the pOX vector, from which cRNA was transcribed in vivo. Approximately 10–30 ng of rSK2 cRNA was injected into Xenopus laevis oocytes 2–6 days before recording.

Electrophysiology.

All recordings were performed in the inside-out patch clamp configuration as described in ref. 27. Data were sampled at 20-μs intervals after low-pass filtering at 5 KHz using the built-in filter of the amplifier. For long recordings data were sampled at 100-μs intervals. No correction was made for the small (<5 mV) voltage errors because of junction potential and series resistance. All experiments were performed at 22 °C. Data analysis, curve fitting and plotting were performed with IGOR Pro software (WaveMetrics). Average results are presented as mean ± SD.

Solutions.

The extracellular (pipette) solution contained (in mM): 140 KMeSO₃, 5 Hepes, 2 KCl, and 2 MgCl₂, pH 7.20. The base internal (bath) solution (BIS) contained: 136 KMeSO₃, 5 Hepes, and 6 KCl, pH 7.20. For solutions containing Ca²⁺ chelators, 5 mM chelator (HEDTA for [Ca²⁺] of 0.77 μM and above, EGTA for 0.55 μM and below) was added to the BIS. To achieve the desired free [Ca²⁺], CaCl₂ was added based on calculations using the WEBMAXC program (Stanford University). The actual free [Ca²⁺] in the final solutions was determined by measurements with a Ca²⁺-sensitive electrode (Orion Research, Inc.). Solution with 5 mM EGTA and no added Ca²⁺ was considered Ca²⁺ free.

All experiments with Tb³⁺ and some with Ca²⁺ were conducted using chelator-free solution (CFS). Great care was taken in the preparation of this solution to reduce contamination by Ca²⁺ or Ca²⁺ chelators. All surfaces in contact with CFS were washed thoroughly with purified water from a Nanopure water system (Barnstead International). To remove contaminating Ca²⁺, BIS was passed through columns made of Chelex 100 resin (Bio-Rad). The columns were prepared and used according to the manufacturer's instructions. Before use the column (≈15 g of resin) was first washed with 500 ml of water, then with BIS until the pH of the effluent stabilized at 7.2. Solution was passed through a column at least 10 times, then passed through a second column repeatedly until the contaminating Ca²⁺ level reached a minimum (see below).

The contaminating Ca²⁺ concentration in the final solution was measured with a Ca²⁺-sensitive dye, Fura-2 (Invitrogen), using a spectrophotometer (Beckman). The absorption spectra of Fura-2 in the Ca²⁺-free form (CFS + 2 μM Fura-2 + 1 mM EGTA) and in the Ca²⁺-loaded form (CFS + 2 μM Fura-2 + 200 μM CaCl₂) were used to linearly fit the absorption spectrum of Fura-2 in CFS to estimate contaminating Ca²⁺ levels. With enough passes through the columns, we were able to routinely reduce the contaminating Ca²⁺ levels to approximately 300 nM, whereas more passes did not lead to further reduction. Different batches of CFS used in this study had contaminating Ca²⁺ levels within ± 100 nM of 300 nM. CFS was filtered with 0.02-μm syringe filters (Anotop 25, Whatman International Ltd.) immediately before use to remove possible free resin particles. Open probability of SK channels in CFS was usually so low that only single-channel openings were visible. In some experiments the baseline current levels in CFS were considered as leak for correction. In other cases where they were available, current levels in Ca²⁺-free solution (5 mM EGTA) were used for leak correction. Leak corrections (usually 5–20 pA at −80 mV) had little effect on the fits of dose-response relationships.

Recording Conditions.

For most experiments, desired amounts of Tb³⁺ (TbCl₃) or Ca²⁺ (CaCl₂) were added directly to CFS by using stock solutions. CaCl₂ stock (0.1 mM and 1 mM) in CFS was prepared from the 0.1 M calcium standard (Orion). 10 mM TbCl₃ was prepared in CFS at pH 2 with added HMeSO₃, from which 1 μM TbCl₃ stock in CFS was prepared by serial dilutions immediately before use. Because of the low final concentrations used, adding TbCl₃ did not result in changes of pH in the bath solution. Tb³⁺ was found to be difficult to wash off from all surfaces in contact. Because of this limitation, all Tb³⁺ dose-response relationships were only measured with increasing concentrations. To avoid contamination between experiments, in many cases a freshly washed disposable Petri dish was used as the recording chamber for each experiment. In other cases where dose-response relationships were measured under two different conditions, a reusable perfusion chamber was used. In these experiments, CFS with increasing amounts of added Ca²⁺ or Tb³⁺ were first applied to the patch, followed by chelator-containing solutions with different free Ca²⁺ concentrations. Solution of >10 times the volume of the recording chamber was washed through to achieve a complete solution change. This chamber was extensively washed after each experiment to remove residual Tb³⁺ or Ca²⁺ chelator. Additionally, oocytes were washed three times with purified water before being transferred to the recording chamber to avoid carrying over Ca²⁺ or chelators.

In experiments to measure the deactivation of SK channels, a syringe connected to a quartz sewer pipe (100 μm in diameter) was used to deliver Ca²⁺-free solution (5 mM EGTA). After the SK current stabilized, the recording pipette was manually moved into a laminar flow of Ca²⁺-free solution from the sewer pipe during continuous recording of current. By using an open electrode, we estimated that it routinely requires only a few ms to move the electrode across the solution interface, judged by the current changes because of junction potential difference. Complete solution change around an inside-out patch takes longer time (a few tens of ms, as judged by changes in SK current when moving a patch into solutions with different K⁺ concentrations). However, only a small fraction of 5 mM EGTA on the inside of the patch is necessary to effectively prevent ligand rebinding during deactivation. This should require much less time to achieve than a complete solution change. It is likely that our measured deactivation time constants are not significantly affected by the solution change, although a slight overestimate is expected in the fast time constants for the deactivation of SK channels activated by Ca²⁺.