Functional exofacially tagged N-type calcium channels elucidate the interaction with auxiliary α₂δ-1 subunits

Significance

The auxiliary α₂δ-1 subunits of voltage-gated calcium (Ca_V) channels are important therapeutic targets, representing the receptor for gabapentinoid drugs in neuropathic pain therapy. It is therefore important to understand their function. Because α₂δ subunits augment calcium currents, it is believed that they increase cell-surface expression of these channels. Here, using exofacially tagged Ca_V2.2 constructs, we now show this to be the case. However, recent proteomic analysis found that α₂δ subunits are associated only loosely and nonquantitatively with Ca_V2 channels, challenging their role as calcium channel subunits. In contrast, we find that Ca_V2.2 and α₂δ-1 are intimately and completely associated at the plasma membrane and that this is not disrupted by the α₂δ-1 ligand gabapentin, which reduces cell-surface expression of both Ca_V2.2 and α₂δ-1.

Abstract

Ca_V1 and Ca_V2 voltage-gated calcium channels are associated with β and α₂δ accessory subunits. However, examination of cell surface-associated Ca_V2 channels has been hampered by the lack of antibodies to cell surface-accessible epitopes and of functional exofacially tagged Ca_V2 channels. Here we report the development of fully functional Ca_V2.2 constructs containing inserted surface-accessible exofacial tags, which allow visualization of only those channels at the plasma membrane, in both a neuronal cell line and neurons. We first examined the effect of the auxiliary subunits. Although α₂δ subunits copurify with Ca_V2 channels, it has recently been suggested that this interaction is easily disrupted and nonquantitative. We have now tested whether α₂δ subunits are associated with these channels at the cell surface. We found that, whereas α₂δ-1 is readily observed at the plasma membrane when expressed alone, it appears absent when coexpressed with Ca_V2.2/β1b, despite our finding that α₂δ-1 increases plasma-membrane Ca_V2.2 expression. However, this was due to occlusion of the antigenic epitope by association with Ca_V2.2, as revealed by antigen retrieval; thus, our data provide evidence for a tight interaction between α₂δ-1 and the α₁ subunit at the plasma membrane. We further show that, although Ca_V2.2 cell-surface expression is reduced by gabapentin in the presence of wild-type α₂δ-1 (but not a gabapentin-insensitive α₂δ-1 mutant), the interaction between Ca_V2.2 and α₂δ-1 is not disrupted by gabapentin. Altogether, these results demonstrate that Ca_V2.2 and α₂δ-1 are intimately associated at the plasma membrane and allow us to infer a region of interaction.

Purification of L-type voltage-gated calcium (Ca_V) channels from skeletal muscle shows that they consist of a pore-forming α₁ subunit, Ca_V1.1, associated with three accessory subunits, β1, α₂δ-1, and γ₁ (1, 2). Cardiac L-type channels have a similar subunit composition, although the α₁ subunit is α1C and the γ subunit is not present (3). However, the study of cell surface-associated N-type (Ca_V2.2) and P/Q-type (Ca_V2.1) calcium channels has been hampered by the lack both of antibodies to cell-surface epitopes and of functional exofacially tagged Ca_V2 channels. Here we report the development of fully functional Ca_V2.2 constructs containing inserted surface-accessible exofacial tags, which allow visualization of only those channels at the cell surface, in both cell lines and neurons. Using this methodological advance, we can now examine directly the effect of the auxiliary subunits on cell-surface expression of Ca_V2 channels.

Although α₂δ subunits have been shown to be associated with Ca_V2.1 and Ca_V2.2 following purification (4, 5), it has recently been suggested that the α₂δ subunits are associated only very loosely and nonquantitatively with Ca_V2 channels (6), calling into question their role as calcium channel subunits. This study found that the α₂δ proteins α₂δ-1, α₂δ-2, and α₂δ-3 could only be copurified using digitonin for tissue solubilization, and not with other detergents; even with digitonin, the study found α₂δ present at less than 10% of the molar ratio of Cav2 α₁ and β subunits (6). One feature of such proteomic techniques is that they capture calcium channel complexes at all stages of maturation as they are trafficked and degraded, as well as the mature proteins, in this case the channels at the plasma membrane. Furthermore, glycosyl phosphatidylinositol anchoring of α₂δ subunits (7) may result in their partial separation from Ca_V2.2 during purification procedures (6).

Functionally, the main effect of α₂δ subunits on both Ca_V1 and Ca_V2 channels is to increase macroscopic calcium currents, which is likely to involve an effect on trafficking of the channels to the plasma membrane or on endocytosis, because no effect has been observed on single-channel conductance and little effect on open probability (8–10). However, there are effects of α₂δ subunits on voltage dependence and kinetics of inactivation (7, 11, 12), indicating that the channel complex is likely to remain intact at the plasma membrane, although these effects might also relate to altered maturation of the channel induced by transient interaction with α₂δ subunits. Thus, evidence that α₂δ remains tightly associated with the channels at the plasma membrane is sparse, and it is possible that the primary function of α₂δ subunits may be as trafficking proteins for calcium channels.

Here we have directly tested whether α₂δ subunits represent trafficking chaperone proteins for Ca_V2 channels, rather than remaining associated with the channels at the cell surface. Our evidence indicates conclusively that Ca_V2.2 and α₂δ-1 are intimately associated at the plasma membrane, and that this interaction is not disrupted by the α₂δ-1 ligand gabapentin.

Results

A tandem HA tag (Ca_V2.2-HA) or a tandem bungarotoxin-binding site (BBS) tag (Ca_V2.2-BBS) was inserted in an extracellular loop in domain II of Ca_V2.2 (Fig. 1A, Inset). These constructs exhibited Ba²⁺ current density–voltage (IV) relationships with no significant differences compared with wild-type (WT) Ca_V2.2 when expressed with α₂δ-1 and β1b in tsA-201 cells (Fig. 1 A and B and Table S1), and also showed identical steady-state inactivation parameters (Fig. 1C). Furthermore, Ca_V2.2-BBS was also inhibited by ω-conotoxin GVIA to the same extent as WT Ca_V2.2 (Fig. 1D). Moreover, the tags of both constructs were available on the cell surface (Fig. 1E and Fig. S1). Single tags in the same position were not cell surface-accessible, although the channels were otherwise functional, in terms of generating Ba²⁺ currents.

Fig. 1.

Properties of Ca_V2.2-HA and Ca_V2.2-BBS constructs. (A) Examples of I_Ba currents (the voltage protocol is shown at the top). Data shown are steps between −20 and +50 mV in 10-mV steps, for tsA-201 cells expressing Ca_V2.2 (Left; black traces), Ca_V2.2-BBS (Center; red traces), or Ca_V2.2-HA (Right; blue traces), all with α₂δ-1/β1b. (Scale bars refer to all panels.) (Inset) Schematic diagram of Ca_V2.2 with the location of the tag site (HA and BBS) identified. (B) Mean IV plots for Ca_V2.2/α₂δ-1/β1b (black squares; n = 30), Ca_V2.2-BBS/α₂δ-1/β1b (red circles; n = 17), and Ca_V2.2-HA/α₂δ-1/β1b (blue triangles; n = 13). Individual IV relationships were fit by a modified Boltzmann function. Mean G_max, V_{50, act}, V_rev, and k values showed no significant differences (Table S1). (C) Mean steady-state inactivation data for Ca_V2.2/α₂δ-1/β1b (black squares; n = 9), Ca_V2.2-BBS/α₂δ-1/β1b (red circles; n = 6), and Ca_V2.2-HA/α₂δ-1/β1b (blue triangles; n = 5). Mean data were fit by Boltzmann functions, with V_{50, inact} values of −59.2, −61.6, and −60.4 mV, respectively. (D) Application of ω-conotoxin GVIA (1 µM for 2 min) produced a complete block of both WT Ca_V2.2 (Upper; black traces) and Ca_V2.2-BBS (Lower; red traces) I_Ba (both representative of n = 5). Currents were elicited by a 50-ms test pulse to +10 mV from −80 mV holding potential. (Scale bars, 200 pA and 10 ms.) Tail current transients have been curtailed for clarity. (E) Representative images showing cell-surface expression of Ca_V2.2-BBS in Neuro2A cells visualized with α-BTX-AF 488. Ca_V2.2 +α₂δ-1-HA/β1b, +β1b, +α₂δ-1-HA, and +α₂δ-1-MIDAS-HA/β1b. (Scale bars, 10 μm.) (F) Bar chart of mean (±SEM) cell-surface Ca_V2.2-BBS density for Ca_V2.2-BBS/α₂δ-1-HA/β1b (black bar; n = 612), Ca_V2.2-BBS/β1b (red bar; n = 493), Ca_V2.2-BBS/α₂δ-1-HA (green bar; n = 238), α₂δ-1-HA/β1b (blue bar; n = 45), and α₂δ-1-HA alone (rightmost bar; n = 265). Data are from five separate transfections. Statistical differences were determined by one-way ANOVA and Bonferroni post hoc tests. ***P < 0.001 for Ca_V2.2-BBS/α₂δ-1-HA/β1b compared with all other conditions. Cells were selected that were positive for internal Ca_V2.2. (G) Bar chart of mean (±SEM) cell-surface Ca_V2.2-BBS density for Ca_V2.2-BBS/α₂δ-1-HA/β1b (black bar; n = 133), Ca_V2.2-BBS/β1b (red bar; n = 111), and Ca_V2.2-BBS/α₂δ-1-MIDAS-HA (gray bar; n = 107). Data were obtained from three separate transfections. Statistical differences were determined by one-way ANOVA and Bonferroni post hoc tests. ***P < 0.001 for Ca_V2.2-BBS/α₂δ-1-HA/β1b compared with the other two conditions.

α₂δ-1 Increases Cell-Surface Expression of Ca_V2.2.

We used the neuronal cell line Neuro2A for subsequent imaging experiments to provide a relevant environment for the neuronal Ca_V2.2 channels. In initial studies using Ca_V2.2-HA, we found that, as expected (13), β subunits are essential for expression of Ca_V2.2 at the plasma membrane, and that almost no surface staining was obtained with Ca_V2.2-HA alone, indicating there are no endogenous β subunits in these cells (Fig. S1 A and B).

In subsequent experiments in Neuro2A cells, we used the equivalent Ca_V2.2-BBS, which was detected at the cell surface with α-bungarotoxin (BTX)-AF 488. When Ca_V2.2-BBS was expressed only with β1b, the amount of Ca_V2.2-BBS at the cell surface was 54% of that in the presence of all subunits (Fig. 1 E and F). Surprisingly, when Ca_V2.2-BBS was coexpressed with α₂δ-1 alone, it showed some surface expression, reaching 14% of the control plus β and α₂δ-1 (Fig. 1 E and F). These results demonstrate clearly that α₂δ-1 does increase the amount of Ca_V2.2 protein at the plasma membrane, and although its effect is particularly exerted after the β subunit interacts with the channel, surprisingly α₂δ-1 does have some effect alone.

We have previously shown that an intact metal ion-dependent adhesion site (MIDAS) in the Von Willebrand factor A (VWA) domain of α₂δ-1 and α₂δ-2 is essential for the enhancement of Ca_V1 and Ca_V2 calcium currents (12, 14). We found that an α₂δ-1 construct with three MIDAS residues mutated to alanine (α₂δ-1-MIDAS^AAA) did not increase the amount of Ca_V2.2-BBS at the cell surface (Fig. 1 E and G), which parallels its inability to enhance Ca_V2.2 calcium currents (14). We also found that in this neuronal cell line, α₂δ-1-MIDAS^AAA itself exhibits reduced cell-surface density compared with WT α₂δ-1 when expressed alone (Fig. S2), indicating that this site may be involved in the interaction with a protein that is important for trafficking α₂δ-1.

Paradoxical Loss of α₂δ-1 Labeling at the Cell Surface When Coexpressed with Ca_V2.2 and β.

Our studies have shown previously that α₂δ subunits can reach the cell surface when expressed alone (12). We therefore examined the cell-surface expression of α₂δ-1-HA, and the effect of Ca_V2.2 coexpression on this, to gain insight into the interaction site of the auxiliary subunits, with the aim of determining whether they remained associated as judged by colocalization at the cell surface. We used α₂δ-1-HA for these studies, as we have shown previously that it supports calcium channel currents with identical properties to WT α₂δ-1 (15).

Surprisingly, we found that the amount of α₂δ-1-HA detected on the cell surface was strongly reduced by coexpression of Ca_V2.2-BBS, both in the presence and absence of the β subunit (Fig. 2 A and B, conditions 1 and 3 compared with 5). In contrast, α₂δ-1-HA cell-surface expression was not reduced by coexpression with only β1b, with which it does not interact directly (Fig. 2 A and B, condition 4 compared with 5). The low detection of α₂δ-1-HA on the cell surface, when coexpressed with Ca_V2.2-BBS, is surprising, because it is effective to increase the amount of Ca_V2.2 at the cell surface (Fig. 1 E–G). A possible contributing factor is that there is likely to be intracellular interaction between α₂δ-1 and Ca_V2.2, particularly in the absence of β, which results in complex formation and intracellular retention (Fig. 2 A and B, condition 3). However, this would not be the explanation in condition 1, when all three subunits are transfected. The same result, loss of cell-surface staining for α₂δ-1 in cells transfected with Ca_V2.2-BBS/α₂δ-1/β1b, was also obtained when using α₂δ-1 without an HA epitope tag (Fig. 2C) and when using WT Ca_V2.2 without an epitope tag (Fig. S3).

Fig. 2.

Cell-surface localization of α₂δ-1: effect of Ca_V2.2 and β1b. (A) Representative images showing cell-surface expression of Ca_V2.2-BBS (row 1, green), internal Ca_V2.2 following permeabilization (row 2, red), and cell-surface α₂δ-1-HA (row 3, white) in Neuro2A cells. The transfected subunits in conditions 1–5 correspond to those in B. (Scale bars, 10 μm.) Cells were selected that were positive for internal Ca_V2.2. (B) Bar chart of mean (±SEM) cell-surface α₂δ-1-HA density for Ca_V2.2-BBS/α₂δ-1-HA/β1b (1, black bar; n = 612), Ca_V2.2-BBS/β1b (2, red bar; n = 493), Ca_V2.2-BBS/α₂δ-1-HA (3, green bar; n = 238), α₂δ-1-HA/β1b (4, blue bar; n = 64), and α₂δ-1-HA alone (5, orange bar; n = 265). Data are from five separate transfections. Statistical differences were determined by one-way ANOVA and Bonferroni post hoc tests for the conditions shown ±Ca_V2.2-BBS. ***P < 0.001. (C) Representative images showing cell-surface expression of α₂δ-1 (red, using α₂δ-1 mAb; Left) and Ca_V2.2-BBS (green; Center) and nuclear staining with DAPI (Right) in Neuro2A cells transfected with α₂δ-1 (without an HA tag) either together with Ca_V2.2-BBS and β1b (Upper) or alone (Lower). (Scale bars, 10 μm.) N/D, not determined.

We therefore examined the relationship between Ca_V2.2-BBS expression and α₂δ-1-HA expression in 522 individual cells (Fig. 3 A and B). Cell-surface Ca_V2.2-BBS and α₂δ-1-HA staining were negatively correlated (Fig. 3B), with cells exhibiting the highest surface Ca_V2.2 staining showing very low surface α₂δ-1 staining, and vice versa. It is possible that many of the cells with elevated staining for α₂δ-1 [>0.6 normalized arbitrary units (a.u.)] were transfected with only low levels or no Ca_V2.2, as this cDNA is the largest and most difficult to cotransfect. However, for the cells exhibiting strong cell-surface staining for Ca_V2.2 (>0.7 normalized a.u.), only 8/522 (1.53%) showed staining for α₂δ-1 of >0.6 normalized a.u., and in these cells there was little colocalization with Ca_V2.2 (Fig. 3A). It is unlikely that this population of cells did not become transfected with α₂δ-1, because we have shown that α₂δ-1 promotes the cell-surface expression of Ca_V2.2 (Fig. 1 E–G).

Fig. 3.

α₂δ-1 epitope occlusion by Ca_V2.2 and lack of effect on endocytosis. (A) Representative images following transfection of Ca_V2.2-BBS/α₂δ-1-HA/β1b showing cell-surface expression of Ca_V2.2-BBS (i, green) or cell-surface α₂δ-1-HA (ii, red) in Neuro2A cells. A few cells showed staining for both subunits, either colocalized (yellow staining) or in separate domains (red and green) (iii and iv). (Scale bars, 10 μm.) (B) Scatter plot of cell-surface staining for both α₂δ-1-HA and Ca_V2.2-BBS in individual cells (n = 522 from four experiments). Cells are categorized as described in SI Materials and Methods. The dashed lines represent the criteria used for the symbols. All cells were included that exhibited surface staining above the background in cells not transfected with the relevant subunit. (C) Representative images showing cell-surface expression of α₂δ-1-HA following antigen retrieval as described in SI Materials and Methods (red; Upper) and nuclear staining with DAPI (Lower) in Neuro2A cells transfected with Ca_V2.2-BBS/α₂δ-1-HA/β1b (Left), Ca_V2.2-BBS/β1b (Center), or α₂δ-1-HA alone (Right). (Scale bars, 10 μm.) (D) Bar chart of mean (±SEM) cell-surface α₂δ-1-HA density following antigen retrieval, for cells such as those shown in C, for Ca_V2.2-BBS/α₂δ-1-HA/β1b (black bar; n = 82), Ca_V2.2-BBS/β1b (red bar; n = 6), and α₂δ-1-HA alone (orange bar; n = 60). Data were obtained from two separate transfections. Statistical differences were determined by one-way ANOVA and Bonferroni post hoc tests for the conditions shown. ***P < 0.001. (E) Bar chart of mean normalized (±SEM) internal Ca_V2.2 (identified by II-III loop Ab; solid bars) and α₂δ-1-HA (open bars) density measured in the same cells, for the subunit combinations Ca_V2.2-BBS/α₂δ-1-HA/β1b (1, black bars; n = 169), Ca_V2.2-BBS/β1b (2, red bars; n = 205), Ca_V2.2-BBS/α₂δ-1-HA (3, green bars; n = 156), and α₂δ-1-HA alone (4, orange bars; n = 213). Data were obtained from three separate transfections. Statistical differences were determined by one-way ANOVA and Bonferroni post hoc tests for the conditions shown ±Ca_V2.2-BBS or ±β1b. *P < 0.05, ***P < 0.001. (F) Ca_V2.2-BBS was measured on the cell surface after 0–30 min incubation at 37 °C for cells expressing Ca_V2.2-BBS/β1b/α₂δ-1-HA (black squares; n = 3 experiments) or Ca_V2.2-BBS/β1b (red circles; n = 3 experiments). In each experiment, 50–90 cells were analyzed per time point. The decay time constant, τ, for the plotted fits was 14.1 min for Ca_V2.2-BBS/β1b/α₂δ-1-HA (black line) and 12.8 min for Ca_V2.2-BBS/β1b (red line). As a control, cells were incubated at 17 °C for 30 min (open triangle; n = 139).

Interaction with Ca_V2.2 Occludes the Detection of the α₂δ-1 Antigenic Epitope at the Cell Surface.

Two possibilities might therefore explain this unexpected finding. First, α₂δ-1 might deliver Ca_V2.2 to the plasma membrane but not remain closely associated with it, being rapidly endocytosed and recycled separately. It has been described previously that α₂δ subunits are only loosely associated with Ca_V2 channels during purification (6, 16), and that α₂δ subunits partition into lipid raft domains (7, 17), and also undergo endocytosis (18). A second possibility is that the epitopes for both the internal HA tag in α₂δ-1 and the α₂δ-1 monoclonal antibody used in these experiments are occluded by a tight association between α₂δ-1 and Ca_V2.2.

With the aim of examining whether the antigenic epitope for α₂δ-1-HA was hidden by association with the Ca_V2.2 α₁ subunit on the cell surface, we used an antigen retrieval method (7, 19). We found that the level of α₂δ-1-HA epitope detected on the cell surface is markedly increased from 15% of the level seen for α₂δ-1 alone, without antigen retrieval (Fig. 2B), to 65% with antigen retrieval (Fig. 3 C and D). This result clearly shows that when α₂δ-1 and Ca_V2.2 are coexpressed together, they must be closely and almost completely associated at the cell surface, sufficient to occlude both the HA antibody from binding to its epitope tag and the binding site of the monoclonal antibody on α₂δ-1. We then mapped the epitope on α₂δ-1 recognized by the monoclonal antibody used in this study. We found that its recognition site requires amino acids 751–755 in the α₂ moiety of α₂δ-1 (Fig. S4). This epitope is downstream of the VWA domain, and is within the bacterial chemosensory-like domains (20), as is the HA epitope in α₂δ-1-HA (which is inserted between amino acids 549 and 550). Therefore, this region may be involved in interaction with Ca_V2.2. In agreement with this, when α₂δ-1-MIDAS^AAA was coexpressed with Ca_V2.2 and β1b, the level of α₂δ-1-MIDAS^AAA on the cell surface was significantly lower than when it was expressed alone (Fig. S2), which indicates that Ca_V2.2 is still able to interact with α₂δ-1-MIDAS^AAA intracellularly. Therefore, the MIDAS motif on the VWA domain of α₂δ-1 is unlikely to be key to the interaction between these two proteins.

Effect of Ca_V2.2 on Intracellular Detection of α₂δ-1.

To determine whether an intracellular interaction of α₂δ-1 with Ca_V2.2 also occluded the detection of intracellular α₂δ-1, we permeabilized cells and quantified internal Ca_V2.2 and α₂δ-1 in parallel experiments to those in which staining for cell-surface Ca_V2.2-BBS was determined. We found that detection of intracellular α₂δ-1 was not reduced in the Ca_V2.2/β1b/α₂δ-1 condition, compared with α₂δ-1 alone (Fig. 3E and Fig. S5, condition 1 compared with 4). Thus, the interaction between Ca_V2.2 and α₂δ-1 in intracellular trafficking compartments is either sufficiently flexible that the α₂δ-1-HA epitope is not occluded or the interaction is loosened by the permeabilization procedure.

Recalling the result that coexpression of Ca_V2.2-BBS with α₂δ-1-HA in the absence of β reduced cell-surface expression of α₂δ-1 (Fig. 2 A and B, condition 3), which indicates that there must be intracellular interaction between these two moieties leading to their intracellular retention, we found that in the same condition (Fig. 3E, condition 3: Ca_V2.2-BBS/α₂δ-1-HA) there was a significant reduction of intracellular α₂δ-1-HA compared with cells transfected with α₂δ-1-HA alone (condition 4) and also a significant reduction of Ca_V2.2-BBS compared with coexpression of all three subunits (Fig. 3E, condition 1: Ca_V2.2-BBS/β1b/α₂δ-1-HA). Thus, when Ca_V2.2 and α₂δ-1 are cotransfected, they may both be subjected to degradation in the absence of the β subunit (21, 22). Together, these results indicate that α₂δ-1 is likely to interact intracellularly with Ca_V2.2, even in the absence of β, but that trafficking out of the endoplasmic reticulum to the plasma membrane is promoted by the β subunit.

We then examined endocytosis of Ca_V2.2-BBS from the cell surface by labeling with α-BTX at 17 °C and then incubating the cells for 10–30 min at 37 °C. We found that the presence of α₂δ-1 had no significant effect on the rate of removal of Ca_V2.2-BBS from the cell surface at 37 °C over the time period measured, the decay time constant (τ) being 12.2 ± 1.8 min for Ca_V2.2/β1b and 15.2 ± 3.2 min for Ca_V2.2-BBS/β1b/α₂δ-1 (n = 3 experiments, P > 0.05; Fig. 3F). As a control, there was only 13% reduction of labeling when cells were incubated for 30 min at 17 °C, a temperature at which endocytosis does not occur (18), indicating that unbinding of α-BTX is negligible over this time course.

Gabapentin Reduces Cell-Surface Expression of Ca_V2.2 and α₂δ-1.

As further evidence that α₂δ-1 influences Ca_V2.2 trafficking and cell-surface expression, we investigated the effect of the α₂δ-1 ligand gabapentin. We found that incubation of Neuro2A cells with gabapentin (100 µM for 24 h) significantly reduced cell-surface expression of Ca_V2.2-BBS by 54% for the Ca_V2.2-BBS/β1b/α₂δ-1-HA combination (Fig. 4 A and B). This result is in agreement with our previous electrophysiological results for Ca_V2.2 channels (23). In contrast, when a mutant α₂δ-1 that does not bind gabapentin (α₂δ-1^R241A) (24) was used in place of WT α₂δ-1, gabapentin had no effect on cell-surface expression of Ca_V2.2-BBS (Fig. 4 A and B). Furthermore, cell-surface expression of Ca_V2.2-BBS in the presence of α₂δ-1^R241A was reduced compared with that in the presence of WT α₂δ-1 (Fig. 4B), in agreement with the reduced functionality of this construct to support Ca_V2.2 currents noted previously (25). In this experiment, cell-surface Ca_V2.2 was increased by α₂δ-1 alone, in the absence of β subunits, as also seen in Fig. 1F, to a level that was 27.0 ± 4.1% (n = 106) of the control plus both auxiliary subunits, and we found that this effect was completely prevented by gabapentin.

Fig. 4.

Effect of gabapentin on cell-surface expression of Ca_V2.2 and α₂δ-1 and α₂δ-1^R241A. (A) Cell-surface expression of Ca_V2.2-BBS in Neuro2A cells transfected with Ca_V2.2-BBS/α₂δ-1-HA (WT)/β1b (Left) or Ca_V2.2-BBS/α₂δ-1^R241A-HA/β1b (Right). (Upper) Control cells. (Lower) Cells incubated with gabapentin (100 µM). (Scale bars, 10 μm.) Cells positive for internal Ca_V2.2 were analyzed. (B) Bar chart of mean (±SEM) cell-surface Ca_V2.2-BBS density in the absence (solid bars) and presence (open bars) of 100 µM gabapentin, for Ca_V2.2-BBS/α₂δ-1-HA (WT)/β1b (black bars; n = 259, 306) and Ca_V2.2-BBS/α₂δ-1^R241A-HA/β1b (red bars; n = 136, 56). Data were obtained from two to four separate transfections. Statistical differences ±gabapentin were determined by Student t test. ***P < 0.001; not significant (ns), P > 0.05. (C) Representative images showing cell-surface expression of α₂δ-1-HA (Left) and α₂δ-1^R241A-HA (Right) in Neuro2A cells transfected with the α₂δ-1 subunit alone. (Upper) Control cells. (Lower) Cells incubated with gabapentin (100 µM). (Scale bars, 10 μm.) (D) Bar chart of mean (±SEM) cell-surface α₂δ-1-HA density in the absence (solid bars) and presence (open bars) of 100 µM gabapentin for the same experiments quantified in B, with the subunit combinations Ca_V2.2-BBS/α₂δ-1-HA (WT)/β1b (black bars; n = 259, 306), Ca_V2.2-BBS/α₂δ-1^R241A-HA/β1b (red bars; n = 136, 56), α₂δ-1-HA (WT) (green bars; n = 175, 201), and α₂δ-1^R241A-HA (blue bars; n = 147, 80). Statistical differences ±gabapentin were determined by Student t test. ***P < 0.001; ns, P > 0.05. Cells were selected that were positive for internal Ca_V2.2, or α₂δ-1 when Ca_V2.2 was not transfected.

Furthermore, gabapentin reduced cell-surface staining of α₂δ-1-HA (WT) when it was expressed alone (by 44%; Fig. 4 C and D) but had no effect on the cell-surface expression of α₂δ-1^R241A-HA (Fig. 4 C and D). In addition, gabapentin did not counteract the occluded cell-surface detection of α₂δ-1-HA, or α₂δ-1^R241A-HA, when it was coexpressed with Ca_V2.2-BBS and β1b (Fig. 4D), indicating that it does not prevent the interaction between α₂δ-1 and Ca_V2.2 subunits on the cell surface. If this interaction were disrupted by gabapentin, then increased detection of α₂δ-1-HA on the cell surface might have been expected.

Expression of Ca_V2.2-HA in Neurons.

When Ca_V2.2-HA was expressed in cultured dorsal root ganglia (DRG) neurons, together with α₂δ-1 and β1b, it could be visualized on the plasma membrane of nonpermeabilized DRG neuron somata, and extended down the neurites (Fig. 5A). Furthermore, similar to our finding in Neuro2A cells, the epitope for α₂δ-1 was hidden in all transfected DRG neurons examined (Fig. 5B), unless they were subjected to antigen retrieval (Fig. 5C).

Fig. 5.

Cell-surface localization of Ca_V2.2 and α₂δ-1 in DRG neurons. Cell-surface expression of Ca_V2.2-HA (A) and α₂δ-1 (B and C) in nonpermeabilized DRG neurons transfected with Ca_V2.2-HA/α₂δ-1/β1b and VAMP-mCherry. Transfected cells were identified by VAMP-mCherry (red). (Lower) Merged images). (A) Ca_V2.2-HA immunostaining (green); 58/71 mCherry-positive DRG examined (81.7%) had surface HA signal in this condition. (B) α₂δ-1 immunostaining (green); 0/20 mCherry-positive DRG had surface α₂δ-1 signal. (C) α₂δ-1 immunostaining (green) after antigen retrieval; 52/62 mCherry-positive DRG (85.5%) had surface α₂δ-1 signal in this condition. (Scale bar, 20 μm.) Representative of two separate transfections.

Discussion

Development of an Exofacially Tagged Ca_V2.2.

To examine the factors affecting the plasma-membrane expression and trafficking of Ca_V2.2, the development of fully functional exofacially tagged Ca_V2.2 constructs was essential. In previous studies, tagged Ca_V2.2 constructs have been used that were not described as functional (26), and the uncertainty remains that partial or complete lack of function may either result in, or be the result of, altered channel trafficking. The functional exofacially tagged Ca_V2.2 constructs described here thus represent important tools for the examination of Ca_V2.2 distribution and trafficking and the effect of auxiliary subunits and other factors. Expression of Ca_V2.2-HA in DRG neurons also results in robust expression on the cell surface, unlike the finding for an HA-tagged Ca_V2.1 construct (27), providing evidence that these constructs represent important tools for studying Ca_V2.2 trafficking and localization in these neurons.

Mechanism of Action of α₂δ-1 to Increase Cell-Surface Expression of Ca_V2.2.

Although it is believed that the major mechanism whereby α₂δ subunits increase the functional expression of calcium channels is due to an increase of the amount of channel protein at the plasma membrane (12), definitive evidence that this is the case has been lacking, particularly for Ca_V2 channels, with measurements for L-type channels mainly relying on determination of gating charge (28, 29). However, the single-channel conductance and open probability of Ca_V2.2, which are two other mechanisms whereby macroscopic current could be increased without affecting the number of channels in the plasma membrane, are little affected by α₂δ subunits (8, 10). Nevertheless, there are minor effects of α₂δ subunits on kinetic and voltage-dependent properties of the currents to increase voltage-dependent inactivation and to hyperpolarize the voltage dependence of steady-state inactivation, which might be attributed, either to an effect of α₂δ proteins on calcium channel folding and maturation or to ongoing association of the channels with α₂δ subunits, to form functional channel complexes on the plasma membrane (7, 12, 30).

We now provide definitive evidence for the increase by α₂δ-1 of cell-surface expression of Ca_V2.2, and also demonstrate that Ca_V2.2 and α₂δ-1 are completely associated at the cell surface when they are coexpressed (together with a β subunit), which is sufficient to occlude the binding of both the HA antibody and the monoclonal antibody to α₂δ-1. We also show that in cultured DRG neurons, antigen retrieval is required to detect α₂δ-1 when it is overexpressed with Ca_V2.2 and β1b, indicating that the epitope is also hidden in these neurons, as is also true for endogenous α₂δ-1 (7, 19). Furthermore, we demonstrate that α₂δ-1 has no effect on endocytosis, and is therefore likely to increase forward trafficking of the channels.

Site of Interaction Between α₂δ-1 and Ca_V2.2.

The epitope for the α₂δ-1 antibody and the inserted HA tag are both within the region spanning the two chemosensory-like domains of α₂δ-1, downstream of the VWA domain (20). It is tempting to speculate that this region forms part of the interaction site with the α₁ subunit, which results in masking of these epitopes when the two subunits interact. In agreement with this, our evidence also indicates that although α₂δ-1-MIDAS^AAA does not increase Ca_V2.2 cell-surface density, there is still an association of this mutant with Ca_V2.2, sufficient to completely prevent α₂δ-1-MIDAS^AAA cell-surface expression. Thus, an intact VWA domain may not be required for interaction with Ca_V2.2, but is required for correct trafficking of both α₂δ-1 and its complex with the pore-forming subunit, possibly via interaction with a trafficking protein(s). This domain of α₂δ-1 has also previously been shown to interact with secreted extracellular matrix proteins of the thrombospondin family (31).

Our results indicate that Ca_V2.2 can interact intracellularly with α₂δ-1, possibly before its β subunit-mediated exit from the endoplasmic reticulum, because the cell-surface expression of α₂δ-1 (which alone can readily reach the cell surface) is markedly reduced by coexpression with Ca_V2.2 in the absence of β, indicating that Ca_V2.2 must be causing α₂δ-1 to be retained intracellularly. Surprisingly, we also find a small but significant effect of α₂δ-1 to increase the amount of Ca_V2.2 on the cell surface, even in the absence of β subunits. This is unlikely to be a result of the influence of endogenous β, because no Ca_V2.2 reaches the cell surface in the absence of both β and α₂δ-1. As expected, the presence of a β subunit alone increased Ca_V2.2 cell-surface expression markedly from a very low level; this is in agreement with indirect evidence for Ca_V1.2 channels from most (32, 33) but not all (34) other studies.

Mechanism of Action of Gabapentin on N-Type Calcium Channel Cell-Surface Expression.

Gabapentin reduced the cell-surface expression of both Ca_V2.2 and α₂δ-1 in all conditions in which α₂δ-1 was coexpressed. Furthermore, our finding that gabapentin does not increase the detection of cell surface-expressed α₂δ-1 when it is occluded by coexpression with Ca_V2.2/β1b indicates that the interaction between Ca_V2.2 and α₂δ-1 is not disrupted by gabapentin, and that this does not therefore form part of its mechanism of action. All of the effects of gabapentin are via binding to α₂δ-1, as evidenced by the lack of effect of gabapentin when the α₂δ-1^R241A subunit is used in place of WT α₂δ-1. This mutation abrogates the binding and function of gabapentinoids in experimental models of neuropathic pain and epilepsy (24, 25, 35).

In conclusion, this study has identified Ca_V2.2-HA and Ca_V2.2-BBS to be important tools for research into factors affecting N-type calcium channel trafficking (36). It has allowed us to show that α₂δ-1 increases the plasma-membrane expression of N-type channels and remains closely associated with these channels on the cell surface, with the interaction possibly involving the α₂δ-1 chemosensory-like domains. This study has also increased our understanding of the mechanism of action of the gabapentinoid drugs on N-type calcium channel trafficking.

Materials and Methods

Molecular biology, cell culture, immunocytochemistry, imaging, electrophysiology and immunoblotting methods are given in SI Materials and Methods. The primers used for molecular biology and the full tag sequences are given in Table S2. The details of analysis for electrophysiology and imaging are also given in SI Materials and Methods.