A Destructive Interaction Mechanism Accounts for Dominant-Negative Effects of Misfolded Mutants of Voltage-Gated Calcium Channels

Abstract

Channelopathies are often linked to defective protein folding and trafficking. Among them, the calcium channelopathy episodic ataxia type-2 (EA2) is an autosomal dominant disorder related to mutations in the pore-forming Ca_v2.1 subunit of P/Q-type calcium channels. Although EA2 is linked to loss of Ca_v2.1 channel activity, the molecular mechanism underlying dominant inheritance remains unclear. Here, we show that EA2 mutants as well as a truncated form (D_I-II) of the Ca_v3.2 subunit of T-type calcium channel are misfolded, retained in the endoplasmic reticulum, and subject to proteasomal degradation. Pulse-chase experiments revealed that misfolded mutants bind to nascent wild-type Ca_v subunits and induce their subsequent degradation, thereby abolishing channel activity. We conclude that this destructive interaction mechanism promoted by Ca_v mutants is likely to occur in EA2 and in other inherited dominant channelopathies.

Introduction

Episodic ataxia type 2 (EA2) [on-line mendelian inheritance in man (OMIM) #108500] is a prototypical episodic neurological syndrome linked to mutations in the pore-forming Ca_v2.1 (α_1A) subunit of P/Q-type Ca²⁺ channels (Ophoff et al., 1996; Pietrobon, 2002). These channels play a major role in neurotransmitter release (Urbano et al., 2003) throughout the nervous system, particularly in the cerebellar Purkinje cells. It is believed that P/Q-type mutations associated with EA2 reduce the precision of intrinsic Purkinje cell pace making, which likely accounts for the spontaneous attacks of ataxia in EA2 patients (Donato et al., 2006; Walter et al., 2006). The affected gene, CACNA1A, is localized on human chromosome 19p13 (Diriong et al., 1995). CACNA1A is also linked to familial hemiplegic migraine type 1 (FHM1) (OMIM #141500) and spinocerebellar ataxia type 6 (OMIM #183086) (Ophoff et al., 1996; Pietrobon, 2002). These allelic diseases are autosomal dominant disorders, yet the molecular mechanisms involved are unknown. Regarding EA2, >20 CACNA1A nonsense or missense mutations have been identified, and the general consensus is that these mutations, in contrast to FHM1, result in loss of channel activity (Ophoff et al., 1996; Jen, 2000; Guida et al., 2001; Wappl et al., 2002; Pietrobon, 2005). Haploinsufficiency has been suggested as a mechanism for EA2 (Guida et al., 2001; Wappl et al., 2002). This should cause a ∼50% reduction of Ca_v2.1 activity in patients with completely inactive EA2 mutations. However, analysis of Ca_v2.1^(+/−) heterozygous mice, which exhibit a ∼50% inhibition of the P/Q current, revealed no apparent neurological abnormalities (Jun et al., 1999; Fletcher et al., 2001). Additionally, it has also been shown that some EA2 mutants exert a dominant-negative effect (Jouvenceau et al., 2001; Page et al., 2004; Jeng et al., 2006). Several mechanisms have been proposed for this effect, including competition of mutant channels for auxiliary subunits and/or the cell trafficking machinery (Arikkath et al., 2002; Cao et al., 2004; Wan et al., 2005; Raike et al., 2007; Jeng et al., 2008) and translational arrest (Page et al., 2004).

In this study, we explore the folding, trafficking, and stability of Ca_v channels in the presence of mutants. Pulse-chase analysis reveals that wild-type Ca_v2.1 channels are quickly degraded in the presence of EA2 mutants and similarly for Ca_v3.2 channels in the presence of truncated mutant (D_I-II). Altogether, endoplasmic reticulum (ER) retention and proteasomal degradation of Ca_v mutants appear to be the basis of the dominant-negative activity. From this study, a model emerges where misfolded mutants interact with wild-type channels and promote their premature proteasomal degradation. This destructive interaction mechanism (DIM) provides a likely framework to decipher the principle of dominance in EA2 as well as other calcium channelopathies that involve misfolded mutants.

Materials and Methods

Molecular biology.

Green fluorescent protein (GFP) N-terminally fused Ca_v2.1 (GFP-Ca_v2.1) was constructed by cloning human Ca_v2.1 cDNA in pbA-eGFP expression vector. Hemagglutinin-tagged Ca_v2.1 (Ca_v2.1-HA) was constructed by replacing amino acids 1726–1733 in the domain IV pore loop of Ca_v2.1 in pbA-eGFP expression vectors (AF004883) (Mullner et al., 2004) by the HA epitope tag sequence YPYDVPDYA using standard splice overlap extension PCR. The construct was confirmed by sequencing. GFP-Ca_v2.1 and Ca_v2.1-HA cDNA were transferred in pCIneo expression vector (Promega, Madison, WI). R1279X, G293R, and AY1593/94D Ca_v2.1 mutant constructs have been described previously (Wappl et al., 2002). CD4 constructs have also been described previously (Zerangue et al., 2001). The generation of the HA-tagged Ca_v3.2 subunit has been described previously (Dubel et al., 2004; Vitko et al., 2007). Truncated isoforms of the Ca_v3.2 subunit were created using PCR techniques, automated DNA sequencing, and cloning into the pEGFP-C1 expression vector (BD Biosciences, Franklin Lakes, NJ).

Antibodies and reagents.

The 12CA5 (mouse) and 3F10 (rat) monoclonal antibodies were used to detect HA-tagged proteins (Roche Applied Science, Indianapolis, IN). The murine anti-CD4 antibody was obtained from Beckman Coulter (Fullerton, CA). Anti-GFP monoclonal and polyclonal antibodies were obtained from Clontech (Mountain View, CA). Anti-Ca_v2.1 polyclonal antibodies were purchased from Alomone Labs (Jerusalem, Israel). Secondary HRP-conjugated antibodies were obtained from The Jackson Laboratory (Bar Harbor, ME). Alexa-Fluor-conjugated antibodies and cholera toxin (CT) were obtained from (Invitrogen, Carlsbad, CA). MG-132 (carbobenzoxy-l-leucyl-l-leucyl-l-leucinal), leupeptin, and NH₄Cl were obtained from Calbiochem (San Diego, CA).

Cell cultures and transfections.

Human embryonic kidney 293 (HEK293) and NG108-15 cells were cultured as described previously (Chemin et al., 2002; Vitko et al., 2007). For optimal transfection, cells were plated at 50–70% confluence. Cell lines were transfected using the JetPEI transfection reagent (QBiogene, Irvine, CA) according to the manufacturer protocol. The Ca_v2.1 constructs were cotransfected with α₂/δ₁ and β_1b subunits (1:2:2 molar ratio), unless otherwise indicated.

Surface expression measurement.

Luminometry-ELISA analysis was performed as described previously (Vitko et al., 2007). Briefly, cells were transfected in 24-well plates. Forty hours after transfection, the cells were fixed for 5 min in 4% paraformaldehyde followed by two washes in PBS. Four wells (10⁵ cells per well) for each condition were incubated for 30 min in blocking solution (PBS plus 1% fetal bovine serum). The expression of HA-tagged channels was measured using a rat anti-HA (3F10, 1/1000) and secondary goat anti-rat coupled to horseradish peroxidase (1/5000). After extensive washes in PBS, SuperSignal substrate (femto; Pierce Chemical, Rockford, IL) was added, and luminescence was measured (Victor 2 luminometer). The surface expression was measured in nonpermeabilized cells, and total expression was measured after Triton X-100 (0.1%) permeabilization. The percentage of surface expression corresponds to the ratio of surface/total relative light unit values.

Immunocytochemistry.

NG108-15 cells were seeded on poly-ornithine-coated coverslips the day before transfection. Two days after transfection, cells were washed in PBS, fixed for 5 min in 4% paraformaldehyde at room temperature, and permeabilized with 0.5 mg/ml saponin in 10 mm glycine PBS. Cells were then incubated with blocking solution [5% fetal calf serum (FCS)] for 30 min. Incubation of the primary antibody was performed at room temperature for 1 h. After washes in PBS, incubation of the secondary antibody was performed at room temperature for 30 min. Coverslips were next mounted onto a microscope slide and were observed on Leica (Wetzlar, Germany) SP2 confocal microscope using a 63× oil immersion objective.

Western blotting.

Forty-eight hours after transfection, cells cultured in 35 mm dishes were lysed on ice for 20 min with NP-40 buffer containing 10 mm Tris-HCl, pH 7.4, 120 mm NaCl, 1% NP-40, and mixture inhibitors (Roche Applied Science). Cell lysates were spun at 10,000 × g for 30 min at 4°C. Protein content of the supernatant was determined using the BCA Protein Assay kit (Pierce Chemical). Sixty micrograms of proteins were mixed with a 4× loading buffer and then loaded on SDS-PAGE. Proteins were then transferred onto nitrocellulose membranes and blocked with 5% powdered nonfat milk. Primary antibody was incubated for 1 h at room temperature in PBS-T (Tween 0.05%). After two washes in PBS-T, secondary HRP-coupled antibodies were incubated for 1 h in PBS-T. The signal was detected using the Super Signal West Pico Chemiluminescent system (Pierce Chemical).

Pulse-chase and immunoprecipitation.

Pulse-chase analyses were performed as described previously (Mezghrani et al., 2001). Briefly, cells were labeled with 250 μCi/ml promix (GE Healthcare Bio-Sciences, Little Chalfont, Buckinghamshire, UK) for 20 min in a free methionine/cysteine medium. Next, cells were washed two times with complete medium and chased in the presence of a fivefold excess of methionine/cysteine for the indicated time. At the end of the chase, cells were lysed in NP-40 buffer (1% NP-40, 150 mm NaCl, 10 mm Tris-HCl, pH 7,6, and mixture inhibitors; Roche Applied Science) supplemented with 10 mm N-ethyl maleimide for 30 min on ice. Lysates were centrifuged at 10,000 × g for 20 min, precleared for 1 h with FCS-Sepharose beads, and incubated overnight with specific antibodies and protein-A Sepharose beads (GE Healthcare Bio-Sciences). Beads were washed three times with lysis buffer, and bead-bound proteins were resolved by standard SDS-PAGE and quantified by autoradiography using automated densitometric scanning (GE Healthcare). Individual bands were quantified by the ImageQuant software.

Electrophysiology.

Whole-cell Ca²⁺ currents were recorded 2–4 d after transfection. Ca_v2.1 and Ca_v3.2 currents were elicited by test pulses (TPs) to 0 and −30 mV, respectively. The current density was calculated according to the capacitance of the cell and expressed in pA/pF. Extracellular solution contained the following (in mm): 2 CaCl₂, 160 tetraethylammonium (TEA)Cl, and 10 HEPES (pH to 7.4 with TEAOH). Pipettes (2–3 MÙ) were filled with a solution containing the following (in mm): 110 CsCl, 10 EGTA, 10 HEPES, 3 Mg-ATP, and 0.6 GTP (pH to 7.2 with CsOH). Detailed acquisition and analysis procedures can be found in a previous study (Chemin et al., 2002).

Results

The EA2 mutant (R1279X) exerts a specific dominant-negative effect on the Ca_v2.1 protein

Several EA2 mutations result in truncated forms of the Ca_v2.1 subunit of P/Q-type channels, such as the R1279X mutant (Fig. 1A) described previously (Wappl et al., 2002). To investigate the properties of R1279X in coexpression experiments with the wild-type Ca_v2.1 subunit, various epitope-tagged constructs were engineered, especially an R1279X mutant N-terminally fused to GFP (R1279X) (Fig. 1A). Functional expression of Ca_v2.1-HA was obtained both in HEK293 cells (Fig. 1B) and in undifferentiated NG108-15 cells (Fig. 1C) that do not express high-voltage-activated (HVA) calcium currents (Chemin et al., 2002). Typical HVA P/Q-type currents (Fig. 1, insets) were recorded in the presence of 2 mm extracellular Ca²⁺. In these two heterologous expression systems, the coexpression of R1279X with Ca_v2.1-HA (1:1) resulted in a strong reduction of the Ca²⁺ current density (>80% in HEK293 cells and ∼70% in NG108-15 cells) (Fig. 1B,C, black bars). The remaining current (Fig. 1B,C, insets) displayed no significant electrophysiological difference in terms of steady state and kinetic properties (data not shown). Similar results were obtained using untagged Ca_v2.1 and the R1279X mutant, excluding a role of the different tags used in the described effects (supplemental Fig. 1A,B, available at www.jneurosci.org as supplemental material). Whether this R1279X mutant exhibits inhibitory activity onto other types of Ca²⁺ channels was evaluated in differentiated NG108-15 cells, which display T-type and HVA currents, essentially L- and N-types (Lukyanetz, 1998; Chemin et al., 2002). Figure 1D shows that transfection of R1279X induces no change in the T-type current density both in proliferative and in differentiated NG108-15 cells (white bars), as well as no significant change in native HVA current density (black bars). In addition, coexpression in undifferentiated NG108-15 cells of R1279X with human Ca_v2.2 that codes for N-type channels produced no significant change in N-type current (10.3 ± 4.7 pA/pF, n = 11 for Ca_v2.2 alone; 11.7 ± 2.5 pA/pF, n = 10 for Ca_v2.2 in the presence of R1279X).

Figure 1.

Inhibition of Ca_v2.1 Ca²⁺ currents by the R1279X EA2-truncated form of Ca_v2.1 protein. A, Schematic cartoon representing the main Ca_v2.1 constructs used in this study. B, The Ca²⁺ current density in HEK293 cells expressing the Ca_v2.1 channel protein. The cells were cotransfected with plasmids encoding the human Ca_v2.1 subunit, the β_1b subunit, and the α₂/δ₁ subunit. Mean ± SEM values of the current density for cells expressing the Ca_v2.1 subunit alone (white bar; n = 18) or in the presence of the R1279X mutant (+R1279X) (black bar; n = 22) are presented (***p < 0.001; Student's t test). The inset shows representative current traces at TP −40 and 0 mV [holding potential (HP), −80 mV]. C, Same experiments as in B performed in the neuroblastoma cell line NG108-15. The HP was −50 mV (***p < 0.001; Student's t test). Ca_v2.1 subunit alone (white bar; n = 15) or in the presence of the R1279X mutant (+R1279X) (black bar; n = 22). D, Measurements of native T-type and HVA Ca²⁺ current densities in R1279X-transfected NG108-15 cells. The T-current density (left) was measured in proliferative NG108-15 cells (Prolif.; 2–3 d after transfection; n = 9) and in differentiated NG108-15 cells (Diff.; 6 d after transfection; 3–4 d after switching to differentiation medium; n = 11) using HP −100 mV and TP −30 mV. The HVA current density, mainly corresponding to L- and N-type channel activities, was measured using HP −50 mV and TP 0 mV (right).

The R1279X mutant impairs trafficking of the wild-type Ca_v2.1 protein

This dominant-negative effect could be caused by either a functional inhibition of wild-type channels at the plasma membrane or by an alteration of Ca_v2.1 trafficking. To visualize the pool of wild-type Ca_v2.1 at the plasma membrane in the absence or presence of R1279X (+GFP and +R1279X, respectively), we performed immunofluorescence studies in HEK293 cells coexpressing wild-type Ca_v2.1 (Ca_v2.1-HA) and the auxiliary β_1b and α₂/δ₁ subunits (Fig. 2A). With coexpressed GFP (control), wild-type Ca_v2.1 can be detected at the surface of the cells (Fig. 2A, top). However, after coexpression of R1279X, no HA staining is detected in nonpermeabilized cells (Fig. 2A, middle). The absence of Ca_v2.1 at the cell surface suggests that R1279X alters trafficking of the channel complex. Immunofluorescence in permeabilized cells revealed a preferential distribution of wild-type Ca_v2.1 in intracellular compartments (Fig. 2A, bottom). Interestingly, despite the low expression of the wild-type Ca_v2.1 protein, it appeared to strongly colocalize with the R1279X mutant (Fig. 2A). A similar pattern of expression was observed in NG108-15 cells (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). A more prominent punctuate labeling can bee seen in NG108-15 cells (supplemental Fig. 2, top, available at www.jneurosci.org as supplemental material), that likely reflects specific membrane domains and/or channel clustering.

Figure 2.

Effect of R1279X mutant on surface expression of Ca_v2.1 channel. A, HEK293 cells were cotransfected with the HA-tagged wild-type Ca_v2.1 subunit alone (with free GFP) or together with the R1279X mutant (fused to GFP). Left column, GFP fluorescence. Middle column, Staining of cells with monoclonal rat anti-HA antibody (primary antibody) and Alexa594 (secondary antibody). NP, Nonpermeabilized; P, permeabilized. Images were acquired using a Leica SP2 confocal microscope, with a 63× oil immersion objective. B, Luminometric ELISA assays to quantify surface expression of the HA-tagged Ca_v2.1 in the presence of the R1279X mutant and auxiliary subunits (**p < 0.01, ***p < 0.001; Student's t test). C, Western blots performed on HEK293 cells expressing Ca_v2.1-HA alone (−) or in the presence of R1279X (+) without (left) or with (right) auxiliary subunits. β_1b-HA was used in these experiments (right).

We next performed luminometric assays in HEK293 cells to quantify Ca_v2.1 surface expression in the presence/absence of the R1279X mutant and the presence/absence of auxiliary subunits (Fig. 2B,C). Approximately 12% of total Ca_v2.1-HA was present at the cell surface when expressed alone, whereas in the presence of the auxiliary subunits, Ca_v2.1 surface expression reached ∼40% (Fig. 2B). R1279X significantly interfered with surface expression under both conditions (Fig. 2B). The remaining Ca_v2.1 surface expression in the presence of R1279X (∼5–7%) was not further modulated by the presence of auxiliary subunits.

Total expression was also strongly reduced (∼50%) in the presence of the R1279X mutant, again regardless of the presence of auxiliary subunits (data not shown). The ability of R1279X to downregulate Ca_v2.1 expression was verified in Western blot experiments with Ca_v2.1-HA (Fig. 2C). Again, we observed a strong inhibition of Ca_v2.1 expression, regardless of the presence of auxiliary subunits. In these experiments, we used an HA-tagged version of the β_1b subunit that allowed us to visualize both Ca_v2.1 and β_1b subunits. The β_1b subunit expression was unchanged in the presence of R1279X, indicating that the observed downregulation of the Ca_v2.1 subunit is not attributable to a decrease in β subunit expression (Fig. 2C) and that the dominant-negative effect of R1279X acts specifically on Ca_v2.1. We also verified by Western blot that the untagged R1279X mutant downregulated Ca_v2.1 expression (supplemental Fig. 3A,B, available at www.jneurosci.org as supplemental material). To further address that this effect was not caused by the saturation of the transcription/translation machinery, we coexpressed the Ca_v2.1 subunit with the β2-adrenergic receptor and a CD4 protein with an ER retention site (KKXX) (supplemental Fig. 3A,B, available at www.jneurosci.org as supplemental material). No detectable downregulation with unrelated membrane proteins was observed supporting the specificity of this dominant-negative effect. In addition, we report that R1279X is able to slightly downregulate (∼20%) Ca_v2.2 channel expression (supplemental Fig. 3C,D, available at www.jneurosci.org as supplemental material). In agreement with previous reports (Page et al., 2004), we found that coexpression of R1279X with the Ca_v2.2 subunit in HEK293 cells resulted in a small but not significant reduction of N-type current (16.2 ± 6.1, n = 10 and 12.8 ± 5.0, n = 12 for Ca_v2.2 alone and Ca_v2.2 plus R1279X, respectively). Together, these results reveal that R1279X induces a preferential downregulation of Ca_v2.1 channel.

The presence of R1279X mutant induces Ca_v2.1 protein instability

The lower protein amount could result from a decrease in protein synthesis and/or from an increase in protein degradation/instability. We therefore designed pulse-chase experiments to study Ca_v2.1 channel stability in the presence of R1279X (Fig. 3A,B). These experiments were performed using in vivo labeling of cells with a [³⁵S]methionine-cysteine mixture (see Materials and Methods), followed by wash out for an indicated time (1–4 h) and subsequent immunoprecipitation. In the absence of R1279X, the Ca_v2.1 protein decreases slightly (in the range of 25–35%) during 4 h chase (Fig. 3A, left, C, open squares). In contrast, in the presence of R1279X, the Ca_v2.1 signal decreased rapidly and significantly during the chase (down to 10% after 4 h) (Fig. 3A, right, C, filled squares). Notably, a 140 kDa protein coimmunoprecipitated with wild-type Ca_v2.1 (Fig. 3A, right, arrow). This protein, which was only present after R1279X coexpression, most likely corresponds to the R1279X construct. Similar experiments were performed in the presence of auxiliary subunits (Fig. 3B,C, circles). In the absence of R1279X, a ∼60 kDa band was observed, which most likely corresponds to the β_1b subunit, because it is known to coimmunoprecipitate with Ca_v2.1 (Fig. 3B, arrow), and it is absent in transfections without auxiliary subunits (Fig. 3A). In the presence of R1279X, Ca_v2.1 was rapidly degraded (Fig. 3B, right, C), regardless of the presence of auxiliary subunits. Quantification revealed a large and marked change in channel stability as evident from the reduction of the half-life of the Ca_v2.1 subunit from >4 h to 1 h (Fig. 3C). This destabilization cannot be rescued by coexpression of the auxiliary subunits (Fig. 3C). An alternative possibility is that R1279X expression could affect cellular homeostasis, leading to a nonspecific dominant effect. To test this hypothesis, we analyzed the effect of R1279X expression on CD4 receptor stability by pulse chase (Fig. 3D). No significant change of CD4 stability was observed after 4 h of chase (Fig. 3D, quantifications), suggesting that Ca_v2.1 protein stability is selectively affected.

Figure 3.

Channel misfolding and instability induced by the R1279X mutant. A, B, Pulse-chase experiments were performed on HEK293 cells 48 h after transfection (see Materials and Methods). Cells were labeled with ³⁵S-methionine-cysteine for 20 min. The chase was performed for 0, 1, 2, or 4 h as indicated. After lysis, the Ca_v2.1 subunit was immunoprecipitated with an anti-HA antibody. The arrow indicates the band corresponding to the EA2 mutant that coimmunoprecipitates with the wild-type Ca_v2.1 subunit. In B (left), the β_1b is detected (arrow). C, Quantification of three independent experiments using GE Healthcare software. D, Pulse-chase analyses performed on HEK293 cells transfected with CD4-AAXX alone or in the presence of R1279X. Quantification was performed as described above. E, Immunoprecipitations performed on cells coexpressing various subunit arrangements (as indicated) in the absence or presence of an untagged R1279X mutant (right) using standard SDS-PAGE (6%). Western blots (WBs) were performed with the indicated antibodies. F, Immunoprecipitations were performed between β_1b-HA and GFP-R1279X as described in E. MW, Molecular weight.

In the presence of R1279X, the loss of β subunit interaction in pulse-chase experiments (Fig. 3B, right) suggests that its binding to Ca_v2.1 protein is disabled by R1279X. To further support this observation, we next performed coimmunoprecipitation assays using cells coexpressing β_1b-HA and GFP-Ca_v2.1 (Fig. 3E). As already observed in pulse-chase experiments, β_1b-HA coimmunoprecipitated with GFP-Ca_v2.1 (Fig. 3E, lane 3), but no GFP-Ca_v2.1 coimmunoprecipitated with β_1b-HA (Fig. 3E, lane 4). As predicted from the pulse-chase experiments, no β_1b-HA coimmunoprecipitated with GFP-Ca_v2.1 in the presence of R1279X (Fig. 3E, lane 6), providing further support that R1279X expression prevents Ca_v2.1-β_1b interaction. Next, we tested for a possible interaction between R1279X and the β_1b subunit (Fig. 3F). No detectable interaction between β_1b-HA and R1279X was detected by coimmunoprecipitation (Fig. 3F, lanes 2, 3). Collectively, these data demonstrate that R1279X destabilizes the wild-type Ca_v2.1 channel and interferes with its association with the β_1b subunit. R1279X coimmunoprecipitates with wild-type Ca_v2.1, suggesting that the destabilization process requires a direct interaction between the R1279X and wild-type Ca_v2.1 proteins.

Dominant-negative effect with a similar truncated form of Ca_v3.2

To investigate whether this mechanism also applies to other Ca_v channels, notably T-type Ca²⁺ channels that do not depend critically on β and α₂/δ subunits for correct functionality (Perez-Reyes, 2003), we constructed two truncated forms, the two first domains (termed D_I-II) and the two last (termed D_III-IV) (Fig. 4A) of the Ca_v3.2 channel. Immunofluorescence experiments were performed in NG108-15 cells using an HA-tagged Ca_v3.2 construct (Ca_v3.2-HA) described previously (Dubel et al., 2004). Similar to that observed with R1279X, D_I-II suppressed surface expression of wild-type Ca_v3.2 (Fig. 4B, middle). In contrast, D_III-IV did not alter Ca_v3.2 surface expression (Fig. 4B, bottom). Furthermore, wild-type Ca_v3.2 and D_III-IV colocalized at the plasma membrane.

Figure 4.

Opposite effects of truncated Ca_v3.2 forms on recombinant and endogenous Ca_v3.2 channels. A, Cartoon representing the main Ca_v3.2 constructs used in this study. B, Immunofluorescence was performed on nonpermeabilized NG108-15 cells expressing Ca_v3.2-HA in the presence of truncated forms (D_I-II or D_III-IV) with a monoclonal rat anti-HA (3F10) antibody and an Alexa-594 secondary antibody. Images were acquired using a Leica SP2 confocal microscope with a 63× oil immersion objective. C, Measurements of the T-type Ca²⁺ current density (I_Ca,T) in NG108-15 cells (white bars) transfected with D_I-II (n = 19; black bar) or D_III-IV (n = 13; gray bar) Ca_v3.2 constructs. Note that expression of D_I-II, but not D_III-IV, significantly downregulated the endogenous T-type current density. D, Effect on the T-current density measured in HEK293 cells of the coexpression of D_I-II (n = 15; black bar) or D_III-IV (n = 7; gray bar) with the wild-type Ca_v3.2 protein. Note that coexpression of the D_I-II construct with the wild-type Ca_v3.2 protein completely downregulated the T-type current density (see the corresponding insets). E, F, Luminometry assay was performed to quantify surface (E) and total expression (F) of the HA-tagged Ca_v3.2 in the presence of D_I-II and D_III-IV mutants. The data correspond to the average of three independent experiments (**p < 0.01, ***p < 0.001; Student's t test).

Proliferative NG108-15 cells endogenously express T-type channels related to the Ca_v3.2 subunit (Chemin et al., 2002). We performed patch-clamp recordings of the native T-type current (I_Ca,T) in NG108-15 cells after transfection of either D_I-II or D_III-IV constructs (Fig. 4C). These experiments reveal that overexpression of D_I-II, but not of D_III-IV, significantly affects the density of I_Ca,T in NG108-15 cells. In HEK293 cells, coexpression of D_I-II with wild-type Ca_v3.2 caused abolishment of I_Ca,T (Fig. 4D, black bar). No change in the T-current density was observed after cotransfection of D_III-IV with wild-type Ca_v3.2 (Fig. 4D, gray bar). In agreement with the electrophysiological data, luminometry analysis confirmed the inhibition of Ca_v3.2 surface expression (Fig. 4E) and revealed a strong inhibition of total expression (Fig. 4F) in the presence of D_I-II. Note that coexpression of the wild-type Ca_v3.2 subunit with D_III-IV actually results in a slight increase of Ca_v3.2 surface expression (Fig. 4E). This is in contrast with the dominant activity that has been observed for the D_III-IV form of Cav2.2 (Raghib et al., 2001).

The D_I-II truncated form induces Ca_v3.2 channel instability

We next performed pulse-chase analysis to study whether the loss of Ca_v3.2 channel expression in the presence of D_I-II relates to protein instability. When expressed alone, Ca_v3.2-HA remained stable (Fig. 5A, left). In accordance with the observations with R1279X, D_I-II drastically destabilized wild-type Ca_v3.2 with only 10% remaining after 4 h (Fig. 5A, middle, B). A band corresponding to D_I-II (120 kDa) coimmunoprecipitated with Ca_v3.2-HA and disappeared during the chase. Anti-GFP immunoprecipitation from the same fractions identified this band as D_I-II (Fig. 5A, right, arrowhead).

Figure 5.

Misfolding induction of D_I-II truncated form by direct interaction with the Ca_v3.2 channel. A, Pulse-chase experiments were performed on HEK293 cells transfected with Ca_v3.2 alone (left) and in the presence of D_I-II (middle and right) for the indicated time periods. After lysis, Ca_v3.2-HA subunits were immunoprecipitated with anti-HA antibody. The arrow (right) indicates the D_I-II-truncated form that coimmunoprecipitates with the wild-type channel. B, Quantification of three independent experiments as shown in A was performed using GE Healthcare software. C, Coimmunoprecipitations were performed on cells expressing the Ca_v3.2-HA subunit in the presence of the two D_I-II and D_III-IV truncated forms. Immunoprecipitated samples (IP) were resolved on 6% SDS-PAGE, and Western blots (WB) were performed with indicated antibodies on the same membranes. MW, Molecular weight.

Coimmunoprecipitation experiments were then designed to identify whether the two truncated forms of Ca_v3.2 interact with the wild-type channel (Fig. 5C). D_I-II could be coimmunoprecipitated with Ca_v3.2 and conversely Ca_v3.2 coimmunoprecipitated with D_I-II (Fig. 5C, lanes 3 and 4, respectively). D_III-IV did not coimmunoprecipitate with Ca_v3.2 (Fig. 5C, compare lanes 9 and 10). These data extend our analysis of the dominant-negative effects of R1279X by showing that a truncated Ca_v3.2 protein suppresses functional expression of its wild-type channel counterpart. Notably, the dominant-negative effect appeared to be specific for constructs comprising domains I and II, because D_III-IV did not affect channel stability.

EA2 missense mutants exert a dominant-negative effect through induction of protein misfolding and instability

Missense mutations also exist in EA2 and progressive ataxia (Pietrobon, 2005). We therefore analyzed the effects of two of these mutants, G293R and AY1593/94D (Yue et al., 1997; Wappl et al., 2002). When expressed in NG108-15 cells, only the untagged and tagged G293R mutants produced a detectable current (in 5 of 33 cells) (Fig. 6A) (supplemental Fig. 1C, available at www.jneurosci.org as supplemental material), in agreement with data reported previously (Wappl et al., 2002). The coexpression of these missense mutants with Ca_v2.1-HA (1:1) resulted in a marked loss of the channel activity (Fig. 6A), whereas T-type channel activity was not affected (Fig. 6B).

Figure 6.

EA2 missense mutants act in a dominant-negative manner by misfolding and instability induction. A, Histograms of the mean Ca²⁺ current density (± SEM) obtained in a representative batch of NG108-15 cells expressing wild-type subunit alone (Ca_v2.1-HA, n = 18); the EA2 mutants alone (Ca_v2.1-HA-G293R, n = 33, 5 with detectable current; Ca_v2.1-HA-AY1593/94D, n = 6); and the wild-type and EA2 mutants together (Ca_v2.1-HA+Ca_v2.1-HA-G293R, n = 23, 3 with detectable current; Ca_v2.1-HA+Ca_v2.1-HA-AY1593/94D, n = 10). These experiments were conducted in the presence of the auxiliary β_1b and α₂/δ₁ subunits. B, Normalized native T-type Ca²⁺ current densities (mean ± SEM) in the NG108-15 cells analyzed in A. The differences are not statistically significant between the various conditions tested (Student's t test). C, Western blots with the indicated antibodies (WB) were performed on HEK293 cells coexpressing Ca_v2.1-HA alone or in the presence of G293R and AY1593/94D mutants together with auxiliary subunits. The bottom panel shows the Western blot from cells expressing missense mutants alone. D, Pulse-chase analyses were performed on HEK293 cells transfected with wild-type Ca_v2.1 alone or in the presence of G293R. Quantification of three independents experiment was performed as described above. MW, Molecular weight.

The ability of missense EA2 mutants to downregulate Ca_v2.1 protein expression was investigated by Western blot (Fig. 6C). We observed a strong inhibition of the Ca_v2.1 expression, independent of the presence of auxiliary subunits (Fig. 6C, top). Mutant expression was verified by Western blot with anti-Ca_v2.1 polyclonal antibodies (Fig. 6C, bottom). Next, we performed a pulse-chase analysis to study whether the loss of Ca_v2.1 expression in the presence of missense channels was also related to the induction of protein instability (Fig. 6D). As obtained above with truncated mutants, G293R drastically destabilized wild-type Ca_v2.1, reducing its half-life to <1 h (Fig. 6D). Altogether, these data indicate that the G293R mutant acts in a dominant-negative manner, similar to the truncated Ca_v mutants.

Dominant activity implies ER retention and proteasomal degradation

The destabilization mechanism depicted above is likely to depend on the intrinsic instability of the Ca_v mutants and their ER localization (Gelman and Kopito, 2003). Therefore, we analyzed the subcellular localization of the EA2 mutants and the truncated Ca_v3.2 constructs in NG108-15 by confocal microscopy (Fig. 7A). R1279X, G293R, and D_I-II were not detected at the cell surface, because no colocalization with CT (a plasma membrane marker) was observed (Fig. 7A, magnification). The finding of a trafficking defect of the G293R mutant is in good agreement with that observed previously (Wan et al., 2005) and was further validated by luminometry experiments (supplemental Fig. 1D, available at www.jneurosci.org as supplemental material).

Figure 7.

Endoplasmic reticulum retention and proteasomal degradation of the dominant-negative Ca_v mutants. A, Confocal images of nonpermeabilized NG108-15 cells expressing EA2 mutants and truncated Ca_v3.2 subunits. Alexa 594-coupled CT was used as plasma membrane marker (0.5 μg/ml). B, Confocal images of immunofluorescence staining performed on permeabilized NG108-15 cells expressing indicated EA2 mutants and truncated Ca_v3.2. Polyclonal anti-protein disulfide isomerase (Assay Designs, Ann Arbor, MI) was used as ER marker. C, Pulse-chase experiments were performed as in Figure 3 on cells transfected with EA2 mutants (R1279X and G293R) and Ca_v3.2 truncated forms. Chase was done for the indicated time (hours) with or without MG-132 proteasome inhibitor (50 μm). After lysis, the truncated Ca_v channels were immunoprecipitated with anti-GFP antibody. D, Representative pulse chase performed on HEK293 cells transfected with Ca_v2.1 and the R1279X mutant and the corresponding quantification (n = 3). During the chase, cells were treated with MG-132 (50 μm), leupeptin (20 μm), and NH4Cl (10 mm).

In contrast, D_III-IV colocalized with CT (Fig. 7A, left column). In permeabilized cells (Fig. 7B), staining of R1279X and G293R, as well as D_I-II staining, appeared predominantly in the perinuclear region (i.e., the ER). Figure 7B shows that both R1279X and G293R mutants fully colocalized with an ER resident protein, protein disulfide isomerase. The same pattern was obtained with D_I-II (data not shown). These data establish a correlation between the dominant-negative effect, exerted by R1279X, G293R, and D_I-II and intracellular retention.

Because most misfolded proteins retained in the ER are degraded by the proteasome, a mechanism called endoplasmic reticulum-associated degradation (ERAD) (for review, see Gelman and Kopito, 2003), we performed pulse-chase experiments to evaluate the proteasome-dependent stability of the EA2 mutants and truncated Ca_v3.2 constructs. Fig. 7C shows that R1279X, G293R, and D_I-II were degraded rapidly (for quantification, see supplemental Fig. 4A, available at www.jneurosci.org as supplemental material) and protected from degradation by the proteasome inhibitor MG-132 (50 μm) when added during the chase (Fig. 7C). Such stabilization was not obtained in the presence of lysosomal degradation inhibitors (data not shown). In contrast with EA2 mutants and D_I-II, D_III-IV remained stable during the chase (Fig. 7C) (supplemental Fig. 4A, available at www.jneurosci.org as supplemental material). These results show that the truncated forms with dominant-negative activity are recognized as misfolded proteins and become degraded by the proteasome.

Next, we investigated whether misfolded mutant forms induce the proteasomal destruction of the wild-type channels (Fig. 7D). Pulse-chase experiments on cells coexpressing wild-type channels (Ca_v2.1-HA and Ca_v3.2-HA) and truncated forms revealed that MG-132 treatment, but neither leupeptin nor NH₄Cl (lysosomal protease inhibitors), stabilizes wild-type channels after 4 h (Fig. 7D) (supplemental Fig. 4B, available at www.jneurosci.org as supplemental material). We observed a similar effect of the proteasome inhibitor on Ca_v2.1 wild-type channel with the G256R missense EA2 mutant (supplemental Fig. 4C, available at www.jneurosci.org as supplemental material). Altogether, these data indicate that misfolded forms of different Ca_v isoforms promote misfolding-induced ERAD of the wild-type Ca_v subunits and their degradation by the proteasome. This mechanism critically relies on the instability of the Ca_v mutants themselves, which pull their wild-type protein counterparts into degradation pathways through a direct interaction.

Discussion

We provide evidence that the dominant-negative effect induced by Ca_v mutants associated with EA2 and progressive ataxia relies on a DIM of these mutants with wild-type Ca_v2.1 culminating in the degradation of the latter. This dominant-negative behavior appears as a general destructive mechanism for voltage-gated Ca²⁺ channels, because a construct made of the first two domains of the Ca_v3.2 subunit acts similarly to promote Ca_v3.2 degradation. From these data, we propose a model in which the newly synthesized mutant channels physically interact with wild-type Ca_v subunits to promote their degradation. The hypothesis of an interaction between wild-type and mutant channels was proposed in previous studies (Page et al., 2004; Jeng et al., 2006). Here, we show that these incorrectly folded mutant channels are subject to ERAD. These misfolded mutants are able to drive wild-type Ca_v proteins toward degradation (i.e., DIM), which results in a powerful dominant-negative mechanism. DIM requires concomitant translation of the mutant and wild-type proteins, interaction of these two proteins, and induction of instability and proteasomal degradation. Because DIM critically involves the ER retention machinery, it is expected that the dominant activity of EA2 mutants may depend on the expression system/experimental temperature used. Indeed, studies in Xenopus oocytes, for which the ER retention machinery is leaky for misfolded mammalian proteins (Denning et al., 1992), have revealed no (Wappl et al., 2002) or modest (Jeng et al., 2006) dominant activity effect of EA2 mutants. Importantly, ER retention of misfolded proteins can be overcome by lowering culture temperature in mammalian cells (Denning et al., 1992; Morello et al., 2000), as recently reported for EA2 mutants (Jeng et al., 2008). Other data suggest that Ca_v2.1 subunit splice variants might be differentially affected by DIM (Jeng et al., 2006) and that auxiliary subunit expression and stoichiometry may modulate DIM (Arikkath et al., 2002; Raike et al., 2007). Altogether, our data reconcile the apparently controversial data on EA2 mutant dominant activity that appear to be principally attributable to variation in the ER retention and cellular degradative capacity.

Another striking finding of this study is that the presence of both misfolded and wild-type Ca_v in the ER enables their physical interaction. Pulse-chase and immunoprecipitation experiments validate this interaction (Figs. 3, 5). Our data suggest that the interaction between misfolded and wild-type Ca_v proteins depends primarily on determinants distributed along the domain I-II region of the Ca_v subunits, rather than the region from domain III to the C terminus, because D_III-IV does not immunoprecipitate with wild-type Ca_v3.2. Interestingly, it has been shown that D_I-II and D_III-IV hemichannels of the Ca_v2.2 are able to form functional channels, suggesting their ability to associate with one another (Raghib et al., 2001). Altogether, these data, which indicate that a Ca_v–Ca_v interaction may be possible because of intramolecular as well as intermolecular interactions, open new questions regarding assembly and stochiometry of Ca_v proteins in a Ca²⁺ channel complex.

P/Q-type Ca²⁺ channels are heteromultimeric complexes comprising a pore-forming Ca_v2.1 subunit and auxiliary β and α₂/δ subunits. The β subunit binds to the I-II loop of the Ca_v2.1 subunit [α1 interacting domain (AID)] and promotes its sorting from the ER (Isom et al., 1994; Gurnett and Campbell, 1996; Bichet et al., 2000). Interestingly, R1279X is retained in the ER and rapidly degraded, even in the presence of the β subunit. Coexpression of the β subunit does not attenuate DIM, indicating that the interaction of wild-type Ca_v2.1 with β subunit is lost in the presence of R1279X. Importantly, immunoprecipitation experiments indicate that the β subunit does not bind to the EA2 mutant, suggesting a conformational defect of the AID site. Again, the lack of β subunit binding can be interpreted as misfolding of the EA2 mutant. This indicates that interaction with EA2 mutants strongly affects the folding of nascent wild-type Ca_v2.1 subunits and interferes with β subunit interaction. Our data are in accordance with those of Page et al. (2004), who reported that dominant-negative suppression of Ca_v2.2 does not involve sequestration of the β subunit and is not prevented by expression of an increased amount of β subunit (Page et al., 2004).

Coexpression of EA2 mutants with the wild-type Ca_v2.1 subunit induces ER aggregation and rapid Ca_v2.1 degradation, which suggests induction of protein misfolding. At this point, Ca_v misfolding should be considered as a disease-related phenomenon and, reciprocally, EA2 may be considered as a misfolding protein disorder, as are many other neurodegenerative diseases. Defective protein trafficking has emerged as an important mechanism in many inheritable diseases (Kim and Arvan, 1998), and this applies to various channelopathies, including cystic fibrosis (Gelman and Kopito, 2003) and long QT2 syndrome (Delisle et al., 2004), as well as calcium channelopathies (Wan et al., 2005; Jeng et al., 2008). Within the Ca_v2.1 subunit, the ER quality control machinery may recognize many different regions (i.e., transmembrane domains and loops). At first glance, because the I-II loop contains multiple ER retention motifs (Bichet et al., 2000), it is tempting to speculate that a misfolded I-II loop may be responsible for EA2 mutant retention.

Once identified as misfolded proteins, the cellular machinery can activate several mechanisms of protein degradation, such as ATP-dependent ubiquitination and proteasomal degradation, or autophagic destruction by lysosomes (Schwartz and Ciechanover, 1999; Kostova and Wolf, 2003). To the best of our knowledge, the degradation process of misfolded Ca_v subunits is unknown. In this study, we provide evidence that the proteasome is critically involved in the degradation of both mutant and wild-type Ca_v subunit proteins. Page et al. (2004) suggested that accumulation of wild-type and truncated Ca_v proteins triggers a translation inhibition mechanism that is part of the unfolded protein response (UPR). This UPR activation would rely on pancreatic ER-resident, RNA-dependent protein kinase (PERK), because the dominant-negative effect was reduced partially after PERK inhibition (Page et al., 2004). UPR is composed of at least the PERK, inositol-requiring enzyme 1 (IRE1), and activating transcription factor-6 (ATF6) pathways (Harding et al., 2002). These proteins act as sensors of binding immunoglobulin protein (BiP), which is the major chaperone for protein folding. When BiP availability drops in the ER, the following UPR pathways are activated: PERK inhibits protein synthesis (Harding et al., 1999), whereas IRE1 and ATF6 increase ER chaperone expression that, in turn, increases ER exit and ERAD (Shamu and Walter, 1996; Yoshida et al., 2001). Persistent activation of PERK and IRE1 are well known to induce apoptosis (Schroder and Kaufman, 2006). Although we did not evaluate the level of persistent PERK activation, pulse-chase experiments that provide measurements of de novo Ca_v protein synthesis reveal no drastic reduction of both Ca_v2.1 (Fig. 3, compare line 0 in A and B) and Ca_v3.2 (Fig. 5, compare lane 0 in A, left and middle) translation in the presence of EA2 mutants and D_I-II, respectively. Second, because PERK is an ER stress-sensing component and mediates broad translational suppression (Ron, 2002), one would expect low-voltage-activated and HVA Ca²⁺ currents to be significantly reduced in NG108-15 cells transfected with EA2 mutants, which we did not observe. Finally, R1279X does not affect expression of an unrelated membrane protein, CD4, and modestly the Ca_v2.2 channel. Similarly, Jouvenceau et al. (2001) reported no difference in potassium current density in cells cotransfected either with K_v1.1/Ca_v2.1 (wild-type) or K_v1.1/Ca_v2.1 (R1820X). Altogether, we propose that ERAD is a critical mechanism for dominant suppression of Ca²⁺ channel activity in EA2, which could be reinforced by UPR activation (Page et al., 2004).

This destructive interaction is likely to occur in EA2 and other channelopathies that are caused by improper folding of mutant proteins. For example, misassembly of mutant and wild-type channels resulting in targeted degradation was described for human ether-a-go-go-related gene (HERG) potassium channels in long QT2 syndrome (Kagan et al., 2000). HERG subunits assemble as tetramers in the ER to form functional potassium channels. Trafficking-defective mutants of HERG subunits have dominant-negative effects that suppress surface expression of wild-type subunits (Delisle et al., 2004). Another striking example is the human immunodeficiency virus 1 (HIV1) accessory protein Vpu, which shares a significant homology with the N-terminal region of TWIK-related acid-sensitive K⁺ channel 1 (TASK-1) channel and inhibits potassium current through a destructive interaction with the TASK-1 subunit (Hsu et al., 2004). These authors demonstrated nicely that bidirectional destruction occurred as a consequence of TASK-Vpu oligomerization: on one hand, HIV1 Vpu inhibits TASK current, although in contrast, TASK expression restrains the Vpu-mediated virus release (Hsu et al., 2004). The Ca_v channel pore subunit is a single entity of four transmembrane domains, and there is no biochemical evidence for oligomerization of mature Ca_v proteins. Precise analysis of Ca_v subunit folding in the ER by trapping conformational intermediates and eventual transient oligomerization would be necessary to further elucidate the mechanism of EA2 mutant misfolding induction. As a perspective for pharmacological therapy, the search for drugs able to alleviate this destructive interaction, such as small molecule chemical chaperones (Ulloa-Aguirre et al., 2004; Yam et al., 2005) may offer new therapeutic strategies for the treatment of EA2 and other channelopathies.