Single-Channel Monitoring of Reversible L-Type Ca²⁺ Channel Ca_Vα₁-Ca_Vβ Subunit Interaction

Abstract

Voltage-dependent Ca²⁺ channels are heteromultimers of Ca_Vα₁ (pore), Ca_Vβ- and Ca_Vα₂δ-subunits. The stoichiometry of this complex, and whether it is dynamically regulated in intact cells, remains controversial. Fortunately, Ca_Vβ-isoforms affect gating differentially, and we chose two extremes (Ca_Vβ_1a and Ca_Vβ_2b) regarding single-channel open probability to address this question. HEK293α_1C cells expressing the Ca_V1.2 subunit were transiently transfected with Ca_Vα₂δ1 alone or with Ca_Vβ_1a, Ca_Vβ_2b, or (2:1 or 1:1 plasmid ratio) combinations. Both Ca_Vβ-subunits increased whole-cell current and shifted the voltage dependence of activation and inactivation to hyperpolarization. Time-dependent inactivation was accelerated by Ca_Vβ_1a-subunits but not by Ca_Vβ_2b-subunits. Mixtures induced intermediate phenotypes. Single channels sometimes switched between periods of low and high open probability. To validate such slow gating behavior, data were segmented in clusters of statistically similar open probability. With Ca_Vβ_1a-subunits alone, channels mostly stayed in clusters (or regimes of alike clusters) of low open probability. Increasing Ca_Vβ_2b-subunits (co-)expressed (1:2, 1:1 ratio or alone) progressively enhanced the frequency and total duration of high open probability clusters and regimes. Our analysis was validated by the inactivation behavior of segmented ensemble averages. Hence, a phenotype consistent with mutually exclusive and dynamically competing binding of different Ca_Vβ-subunits is demonstrated in intact cells.

Introduction

Voltage-dependent Ca²⁺ (Ca_V) channels are found in all types of excitable cells and play a unique role in the transduction of electrical signaling into various physiological Ca²⁺-dependent processes including muscle contractions, synaptic plasticity, and gene expression (1). Ten distinct pore-forming (Ca_Vα₁) subunits identified by molecular cloning lead to functional diversity through differences in their biophysical properties (1). L-type Ca²⁺ channels are heteromultimeric complexes composed of a pore-forming Ca_Vα₁ subunit (from the Ca_V1 family) and auxiliary subunits Ca_Vβ and Ca_Vα₂δ (1). Ca_Vβ subunits are ∼500-amino-acid cytoplasmic proteins encoded by four genes (Ca_Vβ_1–4) with multiple alternatively spliced variants (2). Ca_Vβ subunits bind to a conserved 18-residue sequence in the Ca_Vα₁ I-II intracellular loop called the α-interaction domain (AID) (3,4). Among the auxiliary subunits, Ca_Vβ subunits play a central role in functional aspects of high-voltage-activated (HVA) Ca²⁺ channels including Ca_V1 and Ca_V2 channels: e.g., in surface expression, activation, inactivation, and regulation by other proteins; and the effects of Ca_Vβ-subunits are isoform-specific (2). Therefore, the association of Ca_Vα₁ and Ca_Vβ subunits is crucial for proper channel biophysical phenotype and cell surface expression of HVA Ca²⁺ channels.

Sequence homology analysis, confirmed by high-resolution x-ray crystallography, revealed that Ca_Vβ subunits consist of five domains arranged as V1-C1-V2-C2-V3, where V1, V2, and V3 are variable N-terminus, HOOK, and C-terminus domains, respectively, and C1 and C2 form conserved Src homology 3 and guanylate kinase (GK) domains, respectively (reviewed in Buraei and Yang (2)). The high-affinity Ca_Vα₁-Ca_Vβ interaction takes place between the AID α-helix and the α-binding pocket (ABP) in the Ca_Vβ GK domain (5–7).

Despite the high (2–54 nM, reviewed in Buraei and Yang (2)) affinity of the ABP-AID interaction found in vitro, whole-cell electrophysiology experiments provide indirect evidence that the interaction between Ca_Vα₁ and Ca_Vβ can be quite dynamic in intact cells. For instance, modulation of gating by adding (or removing) a (different) Ca_Vβ can be observed within <1 h (8), within some 10 min (9), or even within a few seconds if the binding affinity is reduced by mutagenesis (10). In addition, the concentration dependence of the interaction between different Ca_Vβ isoforms argues in favor of a competitive replacement at a unique high-affinity binding site (11). Altogether, the interaction between Ca_Vα₁ and Ca_Vβ can be assumed to be rather rapidly reversible in a physiological context (reviewed by Buraei and Yang (2)).

However, to date, a direct demonstration and mechanistic explanation for reversible binding, and hence the possibility of competition between Ca_Vβ subunits in modulating HVA Ca²⁺ channel gating, has not been accomplished in intact mammalian cells. In this work, we studied competition of Ca_Vβ subunits for recombinant Ca_V1.2 L-type Ca²⁺ channels in heterologous expression system of intact HEK293 cells. We employed the fact that different Ca_Vβ subunits and their splice variants lead to distinct single-channel gating properties of recombinant Ca_V1.2 channels (12–14). For example, we have previously identified that a prevalent Ca_Vβ subunit isoform in human heart, Ca_Vβ_2b (12), is capable to inhibit inactivation of Ca_V1.2 channels (13,14). In contrast, Ca_Vβ_1a promotes rapid inactivation of Ca_V1.2 channels (15) and confers a much lower open probability (14).

In this work, Ca_Vβ_1a or Ca_Vβ_2b subunits or their 1:1 and 2:1 (plasmid ratio) mixtures were overexpressed with Ca_V1.2 channels in HEK293 cells. We used similar transfection protocols to record both whole-cell and single-channel current. At whole-cell level, Ca_Vβ subunits increased current density, displaced the peak of voltage current relationship to hyperpolarized potentials, and modulated steady-state inactivation. As predicted, the inactivation rate was gauged by varying the proportion of the (co-)transfected Ca_Vβ_1a subunit. Single-channel recording and elaborate statistical data analysis provided more direct evidence of dynamic replacement of Ca_Vβ subunits, which spontaneously took place at single-channel level on the timescale of minutes.

Materials and Methods

Expression vectors for L-type Ca²⁺ channel auxiliary subunits

See the Supporting Material.

Cell culture and transfection

For patch-clamp experiments, HEK293α_1C cells stably expressing human cardiac Ca_V1.2 subunits were transiently transfected with pIRESpuro3-α₂δ1 and either pIRES2-DsRed2-β_1a, pIRES2-EGFP-β_2b or the mixtures of pIRES2-DsRed2-β_1a and pIRES2-EGFP-β_2b (2:1 or 1:1 by DNA mass). See the Supporting Material for details.

Patch-clamp measurements and analysis

All recording were made using an Axopatch 1D amplifier (Axon Instruments, Foster City, CA) with pClamp 5.5 data acquisition software (Axon Instruments). Recordings were performed 48–72 h after transfection at room temperature (19–23°C) using whole-cell or cell-attached voltage-clamp of successfully transfected cells, which were identified on the basis of either EGFP, DsRed2, or both EGFP and DsRed2 fluorescence. Whole-cell and single-channel Ba²⁺ currents were performed as reported in Herzig et al. (13) and Jangsangthong et al. (15).

Whole-cell external recording solution contained 10 mM BaCl₂, 1 mM MgCl₂, 65 mM CsCl, 65 mM TEA-Cl, 10 mM dextrose, and 10 mM HEPES (pH 7.3 with TEA-OH). Whole-cell internal solution contained 140 mM CsCl, 1 mM MgCl₂, 4 mM Mg-ATP, 10 mM EGTA, and 9 mM HEPES (pH 7.3 with CsOH). Holding potential in whole-cell experiments was −100 mV. To obtain current-voltage (I/V) relationships, cells were depolarized for 160 ms from the holding potential to voltages ranging from −40 to +60 mV, applied every 5 s. See the Supporting Material for analysis.

Single-channel depolarizing bath solution consisted of 120 mM K glutamate, 25 mM KCl, 2 mM MgCl₂, 1 mM CaCl₂, 2 mM EGTA, 1 mM Na₂-ATP (for sake of comparison with previous studies), 10 mM dextrose, and 10 mM HEPES (pH 7.3 with KOH). Single-channel pipette solution contained 110 mM BaCl₂ and 10 mM HEPES (pH 7.3 with TEA-OH). Single-channel Ba²⁺ currents were evoked in cell-attached configuration by rectangular voltage pulses to the test potential 10 mV from the holding potential –100 mV. The pulse duration and repetition rate were 150 ms and 1.67 Hz (100 sweeps/min), respectively. Single-channel data were processed as described previously (13,15). Experiments were analyzed only when the channel activity persisted for at least 108 s (≥180 sweeps) and no stacked openings were observed. Routinely, at least 9 out of 10 experiments had to be discarded because of the stacked openings.

In the remaining datasets, we calculated the probability p_N=1 that the absence of stacked openings firmly indicates the absence of multiple functional channels (not shown). Although this value was low on the case of Ca_Vβ_1a, robust estimates were obtained in the groups with higher overall single channel open-probability (2:1 mixture of Ca_Vβ_b1a and Ca_Vβ_2b: p_{N = 1} = 0.62 + −0.15, n = 5; 1:1 mixture of Ca_Vβ_1a and Ca_Vβ_2b: p_N=1 = 0.76 + −0.08, n = 4; Ca_Vβ_2b alone: p_N=1: 0.90 + −0.06, n = 8).

Statistics

Unless otherwise stated, data were reported as mean ± SE. Statistical comparison of several groups to a control group was performed by one-way ANOVA followed by Dunnett's post-test. Statistical comparison of several groups to each other was performed by the Holm-Bonferroni method. The significance level was p < 0.05. The term “n” refers to the number of experiments.

Segmentation of the open-probability diary

Single-channel open-probability diary was constructed by calculating open probability for every sweep in a single-channel experiment. Sweep open probability was defined as the total open time in a given sweep divided by the pulse duration (150 ms). Open-probability diary was segmented into clusters with statistically different single-channel activity using a bottom-up merging strategy (similar to Kuzmenkina et al. (16)). The following algorithm was employed by our homemade software. To begin, the open-probability diary was split into groups of five consecutive sweeps. Then, the statistical distance, d_{k, k+1}, between neighboring groups, k and k+1, was calculated as

$$d_{k,k+1}=p_{diffk,k+1}\times f,$$ (1)

where the value p_{diff k, k+1} was the probability that the average open probabilities of two compared groups were different, which was calculated by either exact or Monte Carlo permutation tests. Monte Carlo permutation test with 20,000 trials was applied whenever the number of possible permutations was >7000. The second factor in Eq. 1, f, was an empirical correction factor, which forced the program to merge short groups first and was negligible for large groups. The correction factor, of the following form, was tried and was not optimized further:

$$f=1-\left(1-p_{diff}k,k+1\right)\left(\frac{2}{N_k}+\frac{2}{N_{k+1}}\right).$$ (2)

Here, N_k and N_k+1 were the number of sweeps in the compared groups.

The groups with the smallest statistical distance between them were fused. Then new distances were calculated and statistically closest groups were merged until all d_{k, k+1} were larger than 0.999. In some cases, d_k, k+1 > 0.9995 or 0.9999 were used. Such high values were required to correct for multiple testing during the segmentation procedure.

To finally prove that identified temporal clustering of the open-probability was not an artifact due to multiple comparisons, we generated 200 randomized open probability diaries for every experiment and applied our segmentation procedure to them. Randomized open-probability diaries were obtained from the original open-probability diaries by the bootstrapping (17). In every experiment, the fraction of such randomized open-probability diaries, which were split into two or more clusters, was <5%.

Open-probability clusters were classified into low- and high-open-probability regimes using a threshold of the mean open probability within a cluster at 0.0037. Importantly, there were no secular changes in open-probability over time in our experiments (i.e., “run down”, not shown).

Calculation of lifetimes

Average lifetimes of open-probability clusters were calculated as the total observation time divided by the number of the transition between clusters. Average lifetimes of low- or high-open-probability regimes were calculated as the total observation time of the respective type of the channel activity divided by the number of transitions from the given type of the channel activity to a different channel activity (from low to high, or from high to low). Standard errors were estimated by the jackknife method.

Quantitative real-time polymerase chain reaction

See the Supporting Material.

Results

Whole-cell currents revealed concurrent heterologous expression of Ca_Vβ_1a and Ca_Vβ_2b subunits

To study how different Ca_Vβ subunits modulate gating of Ca_V1.2 channels, either Ca_Vβ_1a, Ca_Vβ_2b or mixtures of Ca_Vβ_1a and Ca_Vβ_2b subunits were heterologously coexpressed with Ca_Vα₂δ1 subunit in HEK293α_1C cells that stably expressed a Ca_v1.2 subunit cloned from human heart. Fluorescent imaging revealed red fluorescence of DsRed2 colocalization with green fluorescence of EGFP that reported expression of Ca_Vβ_1a and Ca_Vβ_2b subunits, respectively, indicating the presence of both Ca_Vβ_1a and Ca_Vβ_2b subunits in those cells (Fig. 1 A). Whole-cell Ba²⁺ current density of Ca_V1.2 channel complexes containing Ca_Vβ subunits were 3.5−4-fold higher compared to the channels from mock-transfected cells (Fig. 1, B and C, and see Table S2 in the Supporting Material). Furthermore, expression of Ca_Vβ subunits shifted the voltage of half-maximal activation to hyperpolarizing potentials, with a significantly stronger shift induced by Ca_Vβ_1a subunits in comparison to Ca_Vβ_2b subunits (Fig. 1 D and see Table S2). Steady-state inactivation (induced by a 5-s conditioning pulse) was also shifted to more negative voltages by Ca_Vβ subunits, with the strongest shift induced by Ca_Vβ_1a subunits (Fig. 1 E and see Table S2). In Ca_Vβ_1a and Ca_Vβ_2b competition experiments, inactivation curves lay between Ca_Vβ_1a and Ca_Vβ_2b curves in agreement with the relative amount of Ca_Vβ_1a subunits being transfected (Fig. 1 E and see Table S2).

Figure 1.

Whole-cell current properties of Ca_V1.2 channels modulated by Ca_Vβ_1a and Ca_Vβ_2b subunits and their mixtures. (A) Micrographs showing HEK293α_1C cells after transfection with vectors encoding Ca_Vβ and Ca_Vα₂δ1 subunits. (Red and green fluorescences) DsRed2 and EGF proteins reporting expression of Ca_Vβ_1a and Ca_Vβ_2b subunits, respectively, in the cells. (B) Representative whole-cell Ba²⁺ current traces. Ca_Vβ subunits used for transfections are indicated above the traces. (C–E) Properties of Ca_V1.2 channels in β⁻ cells (open circles, n = 5), and in cells cotransfected with Ca_Vβ_1a (up-pointing triangles, n = 9), Ca_Vβ_1aCa_Vβ_2b mixture (2:1) (solid circles, n = 8), Ca_Vβ_1aCa_Vβ_2b mixture (1:1) (solid diamonds, n = 6), and Ca_Vβ_2b (down-pointing triangles, n = 6). Symbols with error bars represent mean ± SE. Curves show fits of the average data. (C) I/V curves were obtained by depolarizing from –100 mV holding potential to test voltages between –40 and +60 mV. The data were fitted with a Boltzmann-Ohm function. (D) Activation curves were obtained from fitting of normalized tail currents with a Boltzmann function. (E) To study the steady-state inactivation, peak current amplitudes at +10 mV were measured immediately after a 5-s conditioning potential. The data were fitted using the Boltzmann equation.

As seen in Fig. 1 B, expression of Ca_Vβ_1a subunits accelerated inactivation of Ca_V1.2 channels. This inactivation was not purely monoexponential on the second timescale (data not shown). To address the early component of the inactivation, we determined the percentage of the whole-cell Ba²⁺ current that had inactivated after 150 ms of depolarization (Fig. 2 A). For test potentials –20 mV and above, Ca_Vβ_1a subunits induced faster inactivation than Ca_Vβ_2b subunits (note that steady-state inactivation with Ca_Vβ_2b at –20 mV was 80%, Fig. 1 E). The mixtures of Ca_Vβ_1a and Ca_Vβ_2b subunits showed an intermediate extent of the fast inactivation. From 0 mV and above, the extent of the fast inactivation in Ca_Vβ_1a Ca_Vβ_2b mixture experiments was larger when more plasmid DNA encoding Ca_Vβ_1a subunits was transfected. To exemplify these findings, Fig. 2 B shows normalized representative whole-cell Ba²⁺ currents evoked by 160-ms depolarization to +10 mV. Fig. 2 C presents the average time course of inactivation at +10 mV over 5 s of depolarization. According to our previous observations, the inactivation was promoted by Ca_Vβ_1a; accordingly, the lower the ratio of Ca_Vβ_1a subunits to Ca_Vβ_2b subunits, the slower the inactivation.

Figure 2.

Inactivation kinetics of Ca_V1.2 channels modulated by Ca_Vβ subunits and their mixtures. (A) Voltage dependence of the inactivation rate (determined as the fraction of peak current inactivated after 150 ms of depolarization) of Ca_V1.2/α₂δ1 coexpressed with Ca_Vβ_1a subunit (up-pointing triangles, n = 6), Ca_Vβ_1aCa_Vβ_2b combination (2:1) (solid circles, n = 5), Ca_Vβ_1aCa_Vβ_2b combination (1:1) (solid diamonds, n = 5), and Ca_Vβ_2b subunit (down-pointing triangles, n = 6), respectively. (B) Normalized representative Ba²⁺ currents through Ca_V1.2 channels traces evoked by 160 ms depolarization to +10 mV. (C) Percent of the current inactivation at 0.1-, 0.5-, 1-, and 5 s during 5-s test pulses to +10 mV. (Open, light-shaded, medium-shaded, and solid bars) Ca_Vβ_1a subunit (n = 5), Ca_Vβ_1aCa_Vβ_2b combination (2:1) (n = 4), Ca_Vβ_1aCa_Vβ_2b combination (1:1) (n = 4), and Ca_Vβ_2b subunit (n = 6) experiments, respectively. ^∗p < 0.05; †p < 0.01, one-way ANOVA followed by Dunnett's post-test compared with Ca_V1.2 channels containing Ca_Vβ_2b subunits. The apparent differences between the two mixtures of Ca_Vβ_1aCa_Vβ_2b are not significant.

Taken together, the gating properties of whole-cell currents in Ca_Vβ_1aCa_Vβ_2b mixtures were intermediate between the properties of Ca_Vβ_1a- and Ca_Vβ_2b-modulated channels, verifying concurrent heterologous expression of Ca_Vβ_1a and Ca_Vβ_2b subunits.

Coexpression of Ca_Vβ_1a and Ca_Vβ_2b subunits gradually modulated average single-channel characteristics

Effects of Ca_Vβ subunits on unitary currents were studied in cell-attached patches, with 110 mM Ba²⁺ in the pipette solution (holding potential –100 mV, test potential +10 mV). Fig. 3 A shows representative current traces and ensemble average currents of single Ca_V1.2 channels with the two Ca_Vβ subunits and their combinations. Transfection of Ca_Vβ subunits modified gating of single channels as compared to the mock-transfection: Ca_Vβ_1a and Ca_Vβ_1aCa_Vβ_2b (2:1) transfections resulted in higher time-dependent inactivation, whereas Ca_Vβ_1aCa_Vβ_2b (1:1) and Ca_Vβ_2b led to higher channel availability and open probability (see Table S3). Within the series of different Ca_Vβ experiments, all average single-channel characteristics progressively changed according to the relative amount of transfected Ca_Vβ_2b. The most pronounced effects were for channel availability, open probability within active sweeps, and peak current (from low values for Ca_Vβ_1a to high values for Ca_Vβ_2b) and for inactivation (from high inactivation for Ca_Vβ_1a to low inactivation for Ca_Vβ_2b).

Figure 3.

Single-channel recordings of Ca_V1.2 channels modulated by Ca_Vβ_1a and Ca_Vβ_2b subunits and their mixtures. (A) Exemplary single-channel Ba²⁺ current traces of Ca_V1.2 channels after transient coexpression of Ca_Vβ_1a and Ca_Vβ_2b subunit as indicated. (Top row) Pulse protocol (150-ms pulses applied every 600 ms). (Middle) Ten representative consecutive traces are shown for each Ca_Vβ subunit combination. (Bottom rows) Respective single-channel average currents (from at least 180 traces recorded per experiment). (B) Single-channel open probability fluctuates on a minute timescale. Diaries of single-channel open probability (lower panels of each graphs with noisy lines) were automatically split into temporal clusters with statistically significantly different mean open probability values (upper panels of each graph).

Open probability of single Ca_V1.2 channels fluctuated on the minute timescale

By visual inspection of the open-probability diaries of single Ca_V1.2 channels, we had the impression that the channel open probability fluctuated on the minute timescale (Fig. 3 B). To statistically prove and analyze in detail changes in the channel activity over time, we segmented open-probability diaries into temporal clusters demarcated by statistically significant differences of open probability. As negative controls, we constructed 200 randomized open-probability diaries for each experiment and applied the segmentation procedure to them. In all experiments, the fraction of negative controls where clusters were found was <5%.

With our segmentation procedure, we found fluctuations of the open probability only in half of the experiments with Ca_Vβ_1a, in the majority of the experiments with Ca_Vβ_2b, and in all experiments where Ca_Vβ_1a and Ca_Vβ_2b were coexpressed (see Fig. S1 in the Supporting Material). Moreover, modulation of the open probability in the case of Ca_Vβ_1a alone was much smaller than in the presence of Ca_Vβ_2b subunits (Fig. 3 B).

Average lifetimes of the resolved clusters were similar (Ca_Vβ_1a: 3.9 ± 1.7 min, n = 8 experiments; Ca_Vβ_1aCa_Vβ_2b (2:1): 2.1 ± 0.4 min, n = 5; Ca_Vβ_1aCa_Vβ_2b (1:1): 3.3 ± 1.6 min, n = 4; and Ca_Vβ_2b: 2.3 ± 0.4 min, n = 8). However, we will demonstrate below that, for more accurate analysis, clusters should be grouped according to the level of channel activity.

The following analysis was performed using open-probability clusters from our segmentation procedure. The term “cluster open probability” refers to the mean open probability of a given cluster.

Low- and high-activity regimes of the channel gating

Interestingly, we found that distribution of the cluster-open probabilities from Ca_Vβ_1a experiments could be described as a single Gaussian (Fig. 4). By contrast, distribution of the cluster-open probabilities from Ca_Vβ_2b experiments were spread over two decades and could not be fitted either by one (Fig. 4) or by two (data not shown) Gaussians. This indicates that Ca_Vβ_2b subunits induced multimodal gating behavior and agrees with the observation that periods of high channel activity in Ca_Vβ_2b experiments alternates with periods of low activity (Fig. 3 B).

Figure 4.

Multiregime behavior in the presence of Ca_Vβ_2b subunits. Symbols show distribution of open probabilities and durations of resolved clusters (left Y axis). (Solid lines) Cumulative distribution of cluster open probabilities (right Y axis). (Dashed lines) Cumulative normal distributions with means and variances obtained from the statistics of cluster's open-probability values, weighted by the durations of the clusters. In the case of Ca_Vβ_1a, a good agreement between the experimental and the theoretical normal open-probability distributions is seen. By contrast, in the case of Ca_Vβ_2b, the predicted normal distribution is much narrower than the observed open-probability distribution, indicating that channels switch between high- and low-open-probability regimes. The threshold to separate high-open-probability (Ca_Vβ_2b-like) from low-open-probability (Ca_Vβ_1a-like) regimes was set at open probability = 0.0037 (vertical shaded lines), which is equal to the mean + SD of the open-probability distribution from Ca_Vβ_1a experiments.

Note that, for further analysis, we prefer to use the word “regime” instead of “mode”. This is because the term “mode” is habitually used to describe features of the gating behavior that become evident on a single-sweep basis, rather than the average activity during a sequence of sweeps (open-probability clusters). To separate a high-open-probability behavior typical for Ca_Vβ_2b experiments from a low-open-probability regime found in Ca_Vβ_1a experiments, we used a threshold at the cluster-open probability = 0.0037. The value of the threshold was set to the mean + SD of the cluster-open-probability distributions from Ca_Vβ_1a experiments (weighted by cluster length; and see Fig. 4).

Single Ca_V1.2 channels in Ca_Vβ_2b and Ca_Vβ_1aCa_Vβ_2b mixture experiments robustly exhibited gating in the high-open-probability regime (Fig. 5 A). Whereas single Ca_V1.2 channels in Ca_Vβ_2b experiments stayed predominantly in the high-open-probability regime, the fraction of time spent in the high-open-probability regime decreased with more of Ca_Vβ_1a being cotransfected (Fig. 5 B). Consistently, in the case of Ca_Vβ_1a transfections, channels were almost exclusively in the low-open-probability regime. Altogether, these results suggest that the low- and high-activity regimes reflect association of Ca_Vβ_1a and Ca_Vβ_2b subunits with the channel complex, respectively.

Figure 5.

Properties of low- and high-open-probability regimes. (A and B) Experiments with Ca_Vβ_2b subunits (alone or in mixture experiments) robustly showed high-open-probability clusters (A). Moreover, the fraction of time spent in the high-open-probability clusters increased with more of Ca_Vβ_2b being cotransfected (B). Open circles show values from each experiment. (Diamonds with error bars) Logarithmic (A) or arithmetic (B) group means ± SE. All group means were compared using the Holm-Bonferroni method. ^∗p < 0.05, †p < 0.01, and ‡p < 0.001 compared with Ca_Vβ_1a experiments; §p < 0.001 compared with Ca_Vβ_2b experiments. (C and D) High-open-probability clusters were characterized by a lower inactivation of the average current. (C) Exemplary average currents of the low-open-probability (left) and high-open-probability (right) regimes from the same single-channel experiments. For better visibility, a moving average of 15 points is shown. Graphs are scaled to the same peak current level (the lowest value of moving average of 100 points). (Scale bars) Vertical, 2fA; and horizontal, 20 ms. (D) Inactivation of average currents after 150 ms during low (half-down diamonds) and high (half-up diamonds) open-probability clusters (mean ± SE). For each experiment, all clusters with either high- or low-open-probability were pooled together. Data were used for the analysis only if the pooled duration of the respective clusters was longer than 100 episodes. The value n is the number of experiments with sufficient statistics. All group means were compared using Holm-Bonferroni method. ^∗p < 0.05 and †p < 0.01 compared with high-open-probability clusters of Ca_Vβ_2b experiments. (E and F) Lifetimes of low- and high-open-probability regimes. Low- or high-activity regimes were defined as consecutive clusters of the same type of the channel open probability (low or high, respectively). Mean lifetimes and errors were determined as described in Materials and Methods. Errors for the lifetime of the low-open-probability regime in Ca_Vβ_1a experiments could not be estimated because transitions from the low- to high-activity regime occurred only in one experiment.

Low- and high-open-probability regimes of the channel activity revealed different inactivation kinetics

To substantiate our analysis by a second and rather different biophysical parameter, we studied time-dependent inactivation during low-and high-open-probability regimes. To reduce data noise, for each experiment, we pooled all low- or high-open-probability clusters together to construct respective low- or high-open-probability subset ensemble average single-channel currents. In half of Ca_Vβ_2b and Ca_Vβ_1aCa_Vβ_2b mixture experiments, which showed both low and high channel activity, the extent of the inactivation of the average single-channel currents was higher for the low-open-probability regime (Fig. 5 C). In the rest of the experiments, the inactivation properties were not distinguishable within statistical noise. Accordingly, the mean extent of the inactivation was higher in the low- than in the high-open-probability regimes (Fig. 5 D). Subset ensembles—as far as they could be analyzed—mirrored inactivation properties of Ca_Vβ_1a and Ca_Vβ_2b subunits, respectively (Fig. 5 D and see Table S3).

Apparent lifetimes of the low- and high-open-probability regimes

Next, we determined average lifetimes of low- (Ca_Vβ_1a-like) and high- (Ca_Vβ_2b-like) open-probability regimes to estimate the exchange rate of Ca_Vβ-subunits (Fig. 5, E and F). Additionally, Fig. S2 shows lifetimes of resolved clusters within low- or high-channel-activity regimes. Clusters were defined by our segmentation procedure as a sequence of sweeps with similar channel activity. By contrast, for the lifetime analysis, low- and high-open-probability regimes were defined as consecutive clusters of the same type of the channel activity (low or high). Both ways of the analysis produced similar results. From our data, lifetime of the high-open-probability regime from Ca_Vβ_1aCa_Vβ_2b (2:1) experiments can serve to estimate the lifetime of Ca_Vβ_2b subunits in the channel complex. The increase of the apparent lifetimes of the high-open-probability regime at higher relative Ca_Vβ_2b amounts (Ca_Vβ_1aCa_Vβ_2b (1:1) and Ca_Vβ_2b experiments) may have resulted from the several Ca_Vβ_2b subunits consecutively binding to the channel complex. The higher the ratio of the transfected Ca_Vβ_1a/Ca_Vβ_2b subunit, the more we observed an increase of the apparent lifetime of the low-open-probability regime.

Endogenous expression of Ca_V subunits in HEK293 cells

Endogenous Ca_Vβ₃ subunits were found in HEK tsA-201 cells (18). The presence of endogenous Ca_Vβ subunits can explain why, when only Ca_Vβ_2b subunits were transfected, Ca_V1.2 channels spent a small fraction of time in the low-activity regime (Fig. 5 B). Indeed, using a quantitative real-time polymerase chain reaction, we found mRNA for endogenous Ca_Vβ₁, Ca_Vβ₂, Ca_Vβ₃, and Ca_Vβ₄ subunits as well as for the pore-forming Ca_Vα_1C subunits in HEK293 and HEK293α_1C cells (see Fig. S3, A and B). In HEK293α_1C cells, mRNA level for the pore-forming subunits was elevated by a factor of 46 (see Fig. S3 C).

Discussion

Switching between low and high channel activity likely reflects an exchange of Ca_Vβ subunits

In this work, we showed that simultaneous heterologous overexpression of different Ca_Vβ subunits (Ca_Vβ_1a and Ca_Vβ_2b) produced calcium channels with gating phenotypes intermediate to the gating phenotypes of calcium channels containing only one type of transfected Ca_Vβ subunit. These intermediate average gating characteristics resulted from the switching between Ca_Vβ_1a- and Ca_Vβ_2b-like behaviors of individual channels on the minute timescale.

We provided statistical evidence of the switching by the data segmentation analysis. Our segmentation analysis allowed us to investigate long-time correlation in the channel activity. This is in contrast to the run analysis, which focuses on the short-time correlations. Because definition of the gating modes is associated with the categorization of individual sweeps eventually followed by the analysis of runs or correlations between consecutive sweeps (19–24), we consistently avoided the term “mode” in our article. Our low- and high-activity “regimes” can include several modes with a particular equilibrium between them reflected by the mean cluster open-probability.

Interestingly, we found that, when Ca_Vβ_2b as an only Ca_Vβ subunit was used for the transfection, channels still spent a minor fraction of time in the low-activity regime. To compare, Luvisetto et al. (23) observed that Ca_V2.1 channels coexpressed with one type of Ca_Vβ subunits in HEK293 cells switched between different types of the channel activity also on the minute timescale (no detailed analysis on the switching kinetics was performed). The authors favored the hypothesis that the switching reflected a slow equilibrium between modes (specific for Ca_V2.1) rather than dynamic exchange of heterologously expressed Ca_Vβ subunits with endogenous ones. However, the latter possibility cannot be excluded as the expression of endogenous Ca_Vβ subunits in HEK cells was reported previously (18) and confirmed in this work. Of note, mode switching as induced by covalent modification through cAMP-dependent phosphorylation typically occurs on a faster timescale of a few seconds (20,25).

Importantly, we found that coexpression of Ca_Vβ_1a and Ca_Vβ_2b subunits in 2:1 by the DNA mass ratio largely shortened the high-open-probability regimes indicating that, at this condition, the duration of the high-activity regime was limited by the lifetime of a Ca_Vβ_2b subunit in the channel complex.

Thus, our study suggests dynamic exchange of Ca_Vβ subunits as an additional mechanism in regulation of Ca_V channels in vivo.

Affinity and reversibility of the Ca_Vβ and Ca_Vα₁ binding

The main anchoring site for Ca_Vβ is AID in the intracellular I-II loop of Ca_Vα₁ subunits. The dissociation constant (K_d) of Ca_Vβ subunits with AID peptide or full-length I-II linker is 2–54 nM (2). Affinity of Ca_Vβ to nascent Ca_Vα₁ subunits (judged by the current increase after injection of Ca_Vβ subunits in Xenopus oocytes) appears to be the same (26,27). However, the affinity to Ca_Vα₁ incorporated in the plasma membrane (addressed by voltage shifts of the activation and inactivation) remains less certain: Geib et al. (26) measured K_d values of 3.5 to tens of nM, depending on isoform, whereas other groups reported values of 120–350 nM (11,24,27).

Multiple works suggest that the Ca_Vα₁-Ca_Vβ interaction is reversible in intact cells. For example, endogenous Ca_Vβ subunits aid trafficking of Ca_Vα₁ to the plasma membrane (28) but allow rapid gating modulation by exogenously delivered Ca_Vβ subunits (8,9). Gating properties of Ca_Vα₁ in oocytes in the presence of endogenous Ca_Vβ_3XO are different from the gating properties when Ca_Vβ_3XO is additionally exogenously overexpressed (22). From this, it was proposed that endogenous Ca_Vβ subunits dissociated from the Ca_Vα₁ channels after chaperoning them to the plasma membrane. Likewise, an injection of Ca_Vβ_2a subunits into oocytes with Ca_V2.3/Ca_Vβ_1b channels converted the fast channel inactivation induced by Ca_Vβ_1b subunits to the slow inactivation, indicating that Ca_Vβ_2a subunits took over the channels (11).

Competition conditions may accelerate full dissociation of a Ca_Vβ-Ca_Vα₁ complex

Initial experiments showed that the formation and dissociation of Ca_Vβ-AID complexes required hours and more than tens of hours, respectively (4). However, surface plasmon resonance studies revealed that the association of Ca_Vβ with AID or the I-II linker occurred with the rate of 2–6 × 10⁵ M⁻¹ s⁻¹ (26,27). For our estimate of the dissociation time of ∼30 s, this gives numerically an affinity constant of ∼100 nM. In the surface plasmon resonance experiments, the dissociation rates were on the minute timescale. It was pointed out that the dissociation rates from such experiments may be underestimated due to the rebinding of the ligand (27). Indeed, competition conditions can accelerate dissociation of Ca_Vβ-AID/Ca_Vα₁ complexes. For example, a half-dissociation time of Ca_Vβ₃ bound to the I-II linker in the presence of AID-derived peptides was only 28 s (29). Similarly, although modulation of the current through Ca_Vα₁ subunits by Ca_Vβ subunits was stable in excised inside-out patches for up to 30 min (10,30), the presence of the competitive AID peptide quickly removed this modulation (30). It was proposed that multiple Ca_Vα₁-Ca_Vβ interaction sites led to formation of a ternary complex of Ca_Vα₁, Ca_Vβ and the AID peptide with lower stability (30). It was further suggested that a similar step-by-step binding mechanism can lead to exchange of Ca_Vβ subunits in Ca_V channels in vivo.

Alternatively, low-affinity interaction sites can transiently trap a Ca_Vβ subunit released from AID, thus allowing its effective rebinding. Ca_Vβ subunits from cytoplasm can interfere with this mechanism by competing for the binding to the lower-interaction sites or to the freed AID. Low-affinity interaction sites for Ca_Vβ subunits were found in the C- and N-terminal tails of Ca_Vα₁ subunit, and in the I-II linker outside of AID (28,31–34) Moreover, Ca_Vα₁ and Ca_Vβ subunits may interact nonspecifically (for example, a current through Ca_Vα₁ was upregulated by a random 43-amino-acid peptide (5)). This work demonstrated that, indeed, under conditions facilitating competition, Ca_Vβ subunits can displace each other on a physiologically relevant short timescale of minutes. However, the exact mechanism and, in particular, a role of secondary binding sites, therein require further investigations.

Oligomerization of Ca_Vβ subunits

So far, we considered the widely accepted 1:1 stoichiometry of Ca_Vα₁-Ca_Vβ complex (2), thus assuming that Ca_Vβ subunits are monomeric proteins. However, it was recently demonstrated that Ca_Vβ subunits can form functional homo- and heterooligomers (35,36), and our data do not firmly rule out this possibility. Lao et al. (35) found that the current-voltage relationship and inactivation kinetics of Ca_V1.2 channels with mixture of Ca_Vβ_2d and Ca_Vβ₃ subunits could not be explained by a mixture of Ca_V1.2 channels with either Ca_Vβ_2d or Ca_Vβ₃ subunits. By contrast, in our work, all whole-cell and average single-channel characteristics for mixtures of Ca_Vβ_1a and Ca_Vβ_2b subunits could be interpreted in terms of a mixture of populations of Ca_V channels with either Ca_Vβ subunits. However, very wide distribution of single-channel cluster open probability in the presence of Ca_Vβ_2b subunits may be explained by Ca_Vβ_2b subunit oligomers with different modulation of the channel activity. An existence of Ca_Vβ subunit oligomers may lead to more comprehensive mechanisms for Ca_Vβ subunit displacement from Ca_V channels but it does not change the conclusions of our work.

Limitations and perspectives

Dynamic regulation of Ca_V channels by Ca_Vβ subunits allows modification of the channel gating. Furthermore, Ca_V channels can be regulated by other proteins via their interaction with Ca_Vβ subunits, e.g., by G-protein βγ subunits, Ras-related RGK GTPases, many types of kinases, and bestrophin-1 (2). Moreover, by preventing ubiquitination or interacting with dynamin, Ca_Vβ subunits control degradation and internalization of Ca_V channels (36–38). Akt-dependent phosphorylation of Ca_Vβ₂ subunits prevents proteolysis of Ca_V channels, importantly, the phosphorylation site being conserved only in Ca_Vβ₂ isoforms (39,40). The number of proteins regulating Ca_V channels will continue to grow: comprehensive proteomic studies identified 207 proteins that interact with and form a nanoenvironment of Ca_V2-Ca_Vβ channel complexes (41). Results of our studies implicate that the subunit composition of Ca_V channels can be changed on a physiological timescale, allowing rapid adjustment to particular cellular requirements.

In future experiments on the competition between Ca_Vβ subunits, it would be interesting to modify the strength of Ca_Vα₁-Ca_Vβ interaction, e.g., by site-directed mutagenesis of ABP (42). However, if this will result in an even faster exchange of Ca_Vβ subunits, this might not be resolved statistically by the segmentation of the open-probability diaries. The statistical resolution could perhaps be improved by choosing Ca_Vβ subunits with more distinct gating properties. Nevertheless, here, we preferred to use Ca_Vβ_2b instead of Ca_Vβ_2a subunits, which also strongly enhances open probability and strongly inhibits voltage-dependent inactivation (13,14,43), because palmitoylation of Ca_Vβ_2a leads to its tethering to the cell membrane and, as a result, Ca_Vβ_2a subunits can interact with Ca_Vα₁ bypassing binding to AID (18). Alternatively, one could think to increase open probability applying channel agonists, e.g., (S)-BayK 8644. On the other hand, this can reduce the isoform difference in gating modulation by Ca_Vβ subunits (13). Another source for methodological improvement would be complete elimination of endogenous subunits, either by choosing a different cell line, or RNA silencing strategies. Furthermore, modifications of the pulse protocol maybe used to address whether the switching kinetics are inherently voltage-dependent.

We frankly admit that the complex natural proteomic environment of Ca_V channels (41) will not be fully reconstituted within any heterologous (over-)expression system. Yet, we believe that our experimental approach, followed by careful tuning of the analytical strategy to monitor single-channel behavior in real-time, will pave the way to address the subunit interaction dynamics in intact native cells.