A CaV2.1 N-terminal fragment relieves the dominant-negative inhibition by an Episodic ataxia 2 mutant

Episodic ataxia 2 (EA2) is an autosomal dominant disorder caused by mutations in the gene CACNA1A that encodes the pore-forming CaV2.1 calcium channel subunit. The majority of EA2 mutations reported so far are nonsense or deletion/insertion mutations predicted to form truncated proteins. Heterologous expression of wild-type CaV2.1, together with truncated constructs that mimic EA2 mutants, significantly suppressed wild-type calcium channel function, indicating that the truncated protein produces a dominant-negative effect (Jouvenceau et al., 2001; Page et al., 2004). A similar finding has been shown for CaV2.2 (Raghib et al., 2001). We show here that a highly conserved sequence in the cytoplasmic N-terminus is involved in this process, for both CaV2.1 and CaV2.2 channels. Additionally, we were able to interfere with the suppressive effect of an EA2 construct by mutating key N-terminal residues within it. We postulate that the N-terminus of the truncated channel plays an essential part in its interaction with the full-length CaV2.1, which prevents the correct folding of the wild-type channel. In agreement with this, we were able to disrupt the interaction between EA2 and the full length channel by co-expressing a free N-terminal peptide.


Introduction
Ca V 2.1 (P/Q-type) and Ca V 2.2 (N-type) channels are voltage-gated calcium channels that are expressed in both the central and peripheral nervous system, where they are preferentially localized in presynaptic terminals and play a central role in neurotransmitter release (Turner et al., 1993). Mutations in the CACNA1A gene that encodes the poreforming Ca V 2.1 α1 subunit cause three neurological disorders: familial hemiplegic migraine type 1 (FHM1), spinocerebellar ataxia type 6 (SCA6) and Episodic ataxia 2 (EA2) (Pietrobon, 2010). EA2 is a rare autosomal dominant disorder characterized by prolonged episodes of ataxia, which are commonly triggered by emotional and physical stress (Jen, 2008). Interestingly, while all FHM1 mutations reported so far are missense mutations, localized to important functional regions of Ca V 2.1 such as the pore and the voltage sensors, EA2 is frequently associated with nonsense, deletion or insertion mutations (Jeng et al., 2008;Mantuano et al., 2010). Indeed the majority of EA2 mutations described to date are predicted to form truncated proteins resulting from a premature stop codon (Pietrobon, 2010).
It has been found that the functional expression of the full-length Ca V 2.1 channel is substantially suppressed when it is co-expressed with truncated constructs mimicking EA2 mutations (Jouvenceau et al., 2001;Page et al., 2004), indicating that EA2 may not be simply a result of haploinsufficiency. Furthermore, heterologous expression of the wild-type Ca V 2.2 channel together with corresponding truncated constructs similarly suppressed wild-type channel function (Raghib et al., 2001). Our evidence suggests that the truncated proteins are recognized as misfolded proteins, and retained in the endoplasmic reticulum where they trigger endoplasmic reticulum stress (Page et al., 2004), and are also targeted for proteasomal degradation (Mezghrani et al., 2008). Furthermore, the suppression effect requires interaction between the fulllength and the mutant protein, to induce both synthesis arrest and channel degradation, thereby reducing functional expression of the full-length channel (Mezghrani et al., 2008;Page et al., 2004).
Strikingly, the suppressive effect mediated by the truncated channel proteins has also been described for other calcium channels and may play a physiological role in regulating current density. Indeed, two-domain truncated forms of Ca V 1.2 channel have been identified. These splice variants are predominantly expressed in fetal and neonatal rat heart (Wielowieyski et al., 2001). Furthermore a truncated two domain form of Ca V 2.1 has been identified to occur in brain (Arikkath et al., 2002). Moreover, a truncated Ca V 1.3 splice variant, Cav1.3 33 L, consisting of Dom I, II, III and a portion of domain IV, affects the function of the full-length channel (Liao et al., 2015). Thus, the suppressive effect of the truncated protein appears to play a physiological role in regulating Ca V 1.3 function during cardiac development (Liao et al., 2015). Recently, Ca V 1.2 was also shown to undergo proteolytic cleavage resulting in two complementary fragments. This mid-channel proteolysis is described as an activity-dependent feedback inhibition of voltagedependent calcium channels (Michailidis et al., 2014).
For both Ca V 2.1 and Ca V 2.2 it has been shown that a motif in the Nterminus plays an important role in channel function and modulation by second messengers (Page et al., 1998). Initially, it was established that the substitution of just two arginine residues in this motif completely abolished G-protein modulation (Canti et al., 1999). Later, this motif was also found to be essential for the process underlying dominant-negative suppression of Ca V 2.1 and Ca V 2.2 currents (Page et al., 2010).
In this study we wished to explore whether overexpressing these key N-terminal residues as a separate peptide would impede the dominant-negative effect of the truncated EA2 protein, and thus restore the function of the wild-type Ca V 2.1 channels. If so, this would provide a potential route towards therapeutic intervention.

Immunocytochemistry
48 h after transfection, the Neuro-2a cells were washed twice with PBS and fixed with 4% paraformaldehyde (PFA) for 10 min. The cells were then incubated with rat anti-HA antibody (Roche) at a dilution of 1:500 in PBS with 1% Bovine Serum Albumin (BSA) for 1 h at room temperature. The cells were washed and incubated with a secondary anti-rat antibody conjugated to fluorescein isothiocyanate. Detection of the intracellular channel pool following surface labelling was performed after cell permeabilization with 0.2% Triton X-100 for 10 min. The coverslips were then washed with PBS and incubated with rabbit anti-Ca v 2.2 antibody (polyclonal, affinity-purified antibody directed against the intracellular II-III linker) at a dilution of 1:500 for 1 h in PBS with 1% BSA. The cells were then incubated with secondary antibody, anti-Rabbit Alexa 594 (Abcam) at a dilution of 1:500.
Four days after transfection, the DRG neurons were washed and fixed with 4% PFA for 10 min. The neurons were then permeabilized with 0.2% Triton X-100 for 10 min. The cells were incubated in PBS blocking solution containing 10% goat serum (GS) and 5% BSA, for 30 min. The primary anti-Ca V 2.1 antibody (Alomone Labs) in PBS, containing 5% GS and 2.5% BSA, was applied at 4°C overnight. The cells were washed and the secondary antibody anti-rabbit Alexa 594 (Abcam) was applied in PBS containing 5% GS and 2.5% BSA, for 1 h at room temperature. For all cells, the coverslips were washed and mounted on slides using Vectashield (Vector Laboratories) to reduce photobleaching. Imaging was performed using a confocal laser-scanning microscope (Zeiss).
Cell surface expression was measured using ImageJ, by manually tracing the surface of the cells represented by the HA signal. The background signal was measured in an area of the image lacking cells and subtracted from the measurements.

Immunoprecipitation and western-blotting
Cells were lysed using 1% Igepal in PBS in the presence of 25 mM Nethylmaleimide (NEM) and protease inhibitors for 30 min on ice, and whole cell lysates (WCL) were collected after centrifugation (14,000 ×g, for 30 min at 4°C). The total protein concentration was determined (Bradford assay, Bio-Rad) for each sample. For co-immunoprecipitation, WCL (containing 1 mg protein) was pre-cleared on 30 μL protein A/G Plus Agarose (Santa Cruz) for 2 h at 4°C. Agarose beads were discarded and supernatants were incubated with rat anti-HA (Roche) at a dilution of 1:100 overnight at 4°C. Immunoprecipitated proteins were captured on 50 μL of protein A/G beads for 2 h at 4°C. The beads were washed three times with PBS containing 0.1% Igepal. The beads, as well as the WCL samples, were incubated for 15 min at 55°C with 100 mM DTT, 25 mM NEM and 2× Laemmli sample buffer. Eluted co-immunoprecipitated proteins and WCL samples were resolved by SDS-PAGE 3-8% Tris-Acetate gels and then transferred to polyvinylidene fluoride membranes. Membranes were incubated in blocking buffer (10 mM Tris pH 7.4, 500 mM NaCl, 0.5% Igepal and 3% BSA) for 1 h, followed by incubation with rabbit anti-Ca V 2.2 (against the II-III loop) overnight at a dilution of 1:1000 at 4°C. The secondary antibody, goat anti-rabbit coupled to horseradish peroxidise, was incubated at a dilution of 1:2000 for 1 h at room temperature. Proteins were detected using the enhanced ECL Plus reagent (GE Healthcare) on a Typhoon 9410 scanner (GE Healthcare).

Statistical analysis
Statistical analysis between groups was performed using one-way ANOVA with Bonferroni post-hoc test, using Graphpad Prism software. Statistical analysis between two groups ( Fig. 4) was carried out using Student's t test.

Lack of cell surface expression of truncated channels
The Ca V 2.1 and Ca V 2.2 N-termini show a high degree of homology (Fig. 1A). Here we assessed the role of the two conserved arginine residues (indicated by the red rectangles in Fig. 1A) in the dominant-negative suppression of Ca V 2 channels by truncated two domain constructs. We initially monitored the surface expression of the full-length and truncated Ca V 2.2 harbouring an exofacial HA tag (Cassidy et al., 2014), co-expressed with the auxiliary subunits α 2 δ-1 and β1b. Ca V 2.2 was used to monitor cell surface expression, since in our hands Ca V 2.1 with an exofacial tag  was not well exposed on the cell surface. While the full-length Ca V 2.2 channel was found to be expressed on the cell surface of non-permeabilized Neuro2A cells, as shown by the HA signal (Fig. 1B), a truncated Ca V 2.2 construct, consisting of Dom I-II and the II-III loop was retained intracellularly, and no surface HA signal was detected (Fig. 1B). Ca V 2.2 Dom I-II R52A, R54A , in which the key arginine residues identified in Fig. 1A were substituted by alanine, was also not trafficked to the surface ( Fig 1B). The fact that the truncated two domain Ca V 2.2 constructs are not expressed at the cell surface is not surprising as it has already been shown that the truncated mutant Ca V 2.1 channels alone do not form functional channels (Jouvenceau et al., 2001). However, for Ca V 2.2 and Ca V 1.2, co-expression of complementary truncated domain pairs was found to give rise to currents, although their amplitude was smaller than the full-length channel current (Raghib et al., 2001;Michailidis et al., 2014), indicating that channel fragments can interact and fold appropriately, when all the domains are present.
3.2. Role of the N-terminal arginine motif in the suppressive effect of truncated channels We next examined the suppressive effect of the truncated constructs on Ca V 2.1 and Ca V 2.2 channel function and trafficking. We first investigated the effect of the substitution of R57A, R59A in an EA2 mutant (Ca V 2.1-P1217fs), containing only the first two domains and the intracellular II-III loop; this mutant will be referred as EA2 throughout the paper. In all our studies, the equivalent first two domains of Ca V 3.1 were used as a negative control as this was shown previously not to cause dominant-negative inhibition of Ca V 2.2 channels (Page et al., 2010). This was also found in the present study for Ca V 2.1 as shown by the unchanged current densities at +5 mV in the presence of either Ca V 3.1 Dom I-II (−66.6 ± 8.0 pA/pF) or GFP-CAAX (−75.3 ± 13.0 pA/ pF) (data from Fig. 2C and Fig. 5C).
As shown in Fig. 2A-C, while the EA2 mutant exerted a strong dominant-negative effect, significantly reducing Ca V 2.1 current density at + 5 mV by 63% (control: − 66.6 ± 8.0 pA/pF; EA2: − 27.4 ± 4.8 pA/ pF), the EA2 R57A, R59A construct (−47.2 ± 9.6 pA/pF) produced a nonsignificant dominant-negative inhibition (Fig. 2C). There was no difference in the voltage for 50% activation (V 50 ) (data not shown). This differential effect was confirmed for cell surface expression of full-length Ca V 2.2, which was significantly reduced by co-expression with Ca V 2.2 Dom I-II but not with Ca V 2.2 Dom I-II R52A, R54A (Fig. 2D, E). This confirms that the N-terminal RAR motif in Ca V 2.1 and Ca V 2.2 is involved in the process underlying the dominant-negative suppression by the truncated channels.
We next used full-length channel constructs carrying the arginineto-alanine substitutions (Ca V 2.1 R57A, R59A and Ca V 2.2 R52A, R54A ), and assessed the dominant-negative effect of the truncated constructs. As shown in Fig. 3A-C, co-expression of the EA2 mutant significantly reduced Ca V 2.1 R57A, R59A current by 60% (control: − 93.5 ± 15.0 pA/pF; EA2: −37.5 ± 6.3 pA/pF), indicating that the full-length channel still interacts with the EA2 mutant despite the fact that its own arginine motif was disrupted. However there was still a dominant-negative effect when Ca V 2.1 R57A, R59A was co-expressed with EA2 R57A, R59A (− 61.1 ± 6.4 pA/pF), albeit reduced compared to the EA2 condition (Fig. 3B). It is of interest that Ca V 2.1 R57A, R59A exhibited larger currents than wildtype Ca V 2.1 (Fig. 3C, compared to Fig. 2C), suggesting that there may be a tonic inhibitory effect that is relieved by mutating the N-terminus.
Similar findings were obtained for cell surface expression of Ca V 2.2. The cell surface expression of full-length Ca V 2.2 R52A, R54A was significantly reduced by Ca V 2.2 Dom I-II, and also by Ca V 2.2 Dom I-II R52A, R54A (Fig. 3D, E).
Thus the RAR motif in the N-terminus of the truncated two domain constructs, but not the full-length channels, is directly involved in the suppressive effect of the truncated constructs. Because this inhibition is mediated by the interaction between the full-length and truncated constructs, we hypothesised that the N-terminus of the truncated construct is essential for the truncated channels to interact with the full-length channels. We therefore tested whether the coexpression of the key N-terminal peptide would prevent this deleterious interaction.

Co-expression of an N-terminus construct disrupted the interaction of the full-length Ca V 2.2 with the truncated domains
In this study we used an N-terminal construct with a CAAX myristoylation motif attached to its C-terminus to allow its association with intracellular lipid bilayers (Page et al., 2010). An N-terminally truncated N-terminus construct still containing the RAR motif was used, as it has been shown that the full-length N-terminus itself significantly reduced calcium currents (Page et al., 2010). In Ca V 2.2-(43-95)-CAAX, the first 42 residues containing the glycine-rich region were removed (Fig. 1A).
We performed co-immunoprecipitation experiments to determine if Ca V 2.2-(43-95)-CAAX interferes with the association of full-length Ca V 2.2 HA with Ca V 2.2 Dom I-II. WCL were immunoprecipitated with HA antibody and then probed with a Ca V 2.2 antibody that targets the II-III loop, present in both the full-length and the truncated channel (Fig. 4A). As shown in Fig. 4, full-length Ca V 2.2 HA was able to co-immunoprecipitate Ca V 2.2 Dom I-II (Fig. 4A, band indicated in lane 2). Ca V 2.2-(43-95)-CAAX (lane 4), but not GFP-CAAX used as a control (lane 3), interfered with the interaction between full-length Ca V 2.2 HA and Ca V 2.2 Dom I-II, as indicated by the absence of the band corresponding to Ca V 2.2 Dom I-II in lane 4 (Fig. 4A). Three independent experiments were performed and the bands corresponding to Ca V 2.2 Dom I-II were quantified and normalised to full-length Ca V 2.2 expression for each condition, in both co-immunoprecipitation and WCL western-blots (Fig. 4A, B). The mean data (Fig. 4C) show the normalised values with respect to the Ca V 2.2 HA + Ca V 2.2 Dom I-II condition (lane 2 in Fig. 4A and B). The results show that Dom I-II co-immunoprecipitation with Ca V 2.2 was   significantly decreased compared to the WCL in the presence of Ca V 2.2-(43-95)-CAAX, indicating that the N-terminus disrupted the association of full-length Ca V 2.2 and Ca V 2.2 Dom I-II. In addition, the presence of the N-terminus construct did not affect the total amount of Ca V 2.2 full-length channel, indicated by the unchanged amount of full-length channel band in lane 3 and 4 (Fig. 4B).  We next examined whether uncoupling the deleterious interaction and thus releasing the full-length channel from its association with the truncated protein by expressing the key N-terminus residues can restore the channel's function and trafficking.

Co-expression of the free N-terminus did not alter the function of Ca V 2.1
First, we evaluated whether there was any direct effect of the N-terminal constructs on channel function. When co-transfected with the full-length channel, the truncated Ca V 2.1 N-terminus Ca V 2.1-(46-100)-CAAX corresponding to Ca V 2.2-(43-95)-CAAX did not affect Ca V 2.1 function, as shown by the unchanged current densities compared to control cells co-transfected with the control protein GFP-CAAX (control: − 75.3 ± 13.0 pA/pF; Ca V 2.1-(46-100)-CAAX: −69.0 ± 6.1 pA/pF; Fig. 5A-C). This indicates that the N-terminal fragment, lacking the initial residues 1-45, does not alone interact with the full-length channel sufficiently to cause inhibition.

Co-expression of the N-terminal construct partially restored the function of Ca V 2.1
We next assessed whether the expression of the Ca V 2.1 N-terminal construct, Ca V 2.1-(46-100)-CAAX, might restore Ca V 2.1 function by interfering with the dominant-negative effect of EA2.
3.6. Co-expression of the N-terminus construct restored the expression of Ca V 2.2 at the cell surface We next wanted to examine whether the suppression of Ca V 2.2 trafficking by Ca V 2.2 Dom I-II is also disrupted by co-expression of the equivalent Ca V 2.2 N-terminal peptide. In parallel with the result found for Ca V 2.1 currents, we found firstly that Ca V 2.2-(43-95)-CAAX, unlike Ca V 2.2 Dom I-II, did not alter the cell-surface expression of Ca V 2.2 when they were co-expressed, as shown by the unchanged mean intensity of the HA signal at the cell surface (Fig. 6A, B). Furthermore, co-expression of Ca V 2.2-(43-95)-CAAX restored Ca V 2.2 cell surface expression by decreasing the deleterious effect of Ca V 2.2 Dom I-II (Fig.  6C, D). These results nicely recapitulate the Ca V 2.1 results shown in Fig. 5. 3.7. Co-expression of the N-terminal construct did not restore the function of Ca V 2.1 R57A, R59A Since we found in this study that the EA2 mutant still exerted a dominant-negative effect on Ca V 2.1 R57A, R59A (Fig. 3A-C), we next examined whether it was possible to perturb this suppressive effect by co-expressing Ca V 2.1-(46-100)-CAAX. We first showed that this construct did not have a direct effect on Ca V 2.1 R57A, R59A currents (control: 114.0 ± 28.0 pA/pF; Ca V 2.1-(46-100)-CAAX: 110.0 ± 20.0 pA/pF; Fig. 7A-C). Furthermore, the dominant-negative effect of EA2 on Ca V 2.1 R57A, R59A currents was not significantly reversed by Ca V 2.1-(46-100)-CAAX (control: 109.9 ± 19.2 pA/pF; EA2 + GFP-CAAX: − 36.8 ± 6.0; EA2 + Ca V 2.1-(46-100)-CAAX: − 65.0 ± 8.8 pA/pF; Fig.7D-F). This likely indicates that the suppressive effect of EA2 on Ca V 2.1 R57A, R59A is more robust, and less easily disrupted.
3.8. Co-expression of Ca V 2.1 N-terminus restored P/Q current suppressed by EA2 in DRG neurons We then examined whether Ca V 2.1-(46-100)-CAAX could restore the endogenous P/Q-type current in DRG neurons co-transfected with the EA2 mutant. Firstly, DRG neurons were transfected with YFP as a transfection marker and empty vector, EA2 or EA2 plus Ca V 2.1-(46-100)-CAAX, and stained with a Ca V 2.1 antibody that targets the II-III loop present in EA2, to demonstrate its expression (Fig. 8A). The native Ca V 2.1 could not be detected, probably because of its low level of expression. We performed experiments 4 days after transfection in the presence of 1 μM nifedipine and 1 μM ω-conotoxin GVIA to block Land N-type channels, respectively, in order to isolate native P/Q-type calcium currents. The electrophysiological data showed that the expression of EA2 mutant induced, as expected, a reduction of the native P/Qtype current greater than 50% (control: − 66.0 ± 10.9 pA/pF; EA2: −27.6 ± 4.3 pA/pF; Fig. 8B-D). Importantly, this reduction was almost completely prevented when Ca V 2.1-(46-100)-CAAX was co-expressed (− 65.0 ± 8.8 pA/pF; Fig. 8B-D). This finding further reinforces the view that key N-terminal residues interfere with the dominant-negative effect of EA2.

Discussion
In this study we show for the first time that a key N-terminal peptide is able to interfere with the suppressive effect of the two domain truncated Ca V 2 constructs mimicking EA2 mutant channels, and hence partially restore the functional expression of P/Q-type channels. This may pave the way for development of future treatments to prevent or disrupt the deleterious effect of the truncated channels.
Mutations in the CACNA1A gene encoding the pore-forming subunit of the Ca V 2.1 channel cause several neurological disorders, in particular FMH1, SCA6 and EA2 (Pietrobon, 2010;Jen, 2008). EA2 is a unique channelopathy, as it is frequently associated with mutations leading to truncation of the protein. Although EA2 is episodic in nature, it may also be associated with progressive symptoms, and there are limited treatment options available (Jen, 2008). EA2 is a dominant disorder, and although haploinsufficiency was originally thought to be its cause, there is increasing evidence for a dominant-negative mechanism (Jouvenceau et al., 2001;Page et al., 2004;Cao et al., 2004;Mezghrani et al., 2008). Cacna1a knockout mice, and the naturally occurring cacna1a mutant mouse tottering, have been widely studied and show absence epilepsy, ataxia and paroxysmal dyskinesia, with progressive loss of cerebellar Purkinje cells (Fletcher et al., 1996;Fletcher et al., 2001;Shirley et al., 2008). However the heterozygous knockout and mutant mice show no evidence of neurodegeneration, and thus the null mutation is recessive (Fletcher et al., 2001), except in the case of the semi-dominant tottering 5J (Miki et al., 2008). Furthermore postnatal knockout of Ca V 2.1 in Purkinje cells produces similar effects to the full knockout, in terms of neurological deficit (Mark et al., 2011). It is also of interest that siRNA knock-down of P/Q-type channels in adult mouse cerebellum resulted in mild impairment that could be enhanced by activation of β-adrenoceptors, mimicking stress (Salvi et al., 2014).
Our study confirms that an arginine-alanine-arginine (RAR) motif in the N-terminus is involved in the suppressive effect of the truncated channels. Interestingly, we found that disrupting the arginine motif in the full-length Ca V 2.1 R57A, R59A and Ca V 2.2 R52A, R54A channels did not impede channel function or expression at the cell surface, respectively. It is also worth mentioning that Ca V 2.1 R57A, R59A evoked a larger current compared to wild-type Ca V 2.1. The level of expression of Ca V 2.1 R57A, R59A and Ca V 2.2 R52A, R54A was found unchanged in respect to the wild type Ca V 2.1 and Ca V 2.2 respectively indicating that these proteins are not more stable or more highly expressed (data not shown, three independent experiments). The larger Ca V 2.1 R57A, R59A current can be explained by the lack of tonic G-protein inhibition, as it has been shown that disrupting this motif completely abolished Ca V 2.2 G-protein modulation (Canti et al., 1999). The truncated channels mimicking EA2 suppressed Ca V 2.2 R52A, R54A and Ca V 2.1 R57A, R59A trafficking and currents, in a similar manner to their effect on the wild-type channels. This indicates that disrupting the RAR motif in the full-length channel does not impede the deleterious interaction; in contrast, disrupting the same RAR motif present in the N-terminus of the truncated channels reduced their ability to produce dominant-negative suppression for both Ca V 2.1 and Ca V 2.2.
We postulate that the N-terminus is normally involved in intramolecular docking with another part of the same channel, to form correctly folded and functional channels, and the RAR motif is involved in this interaction (Fig. 9A). This intramolecular docking is also involved in Gprotein-mediated inhibition of the Ca V 2 channels (Page et al., 1998;Canti et al., 1999). We propose that in the truncated channels this intramolecular interaction cannot occur, because the channel structure is incomplete and misfolded (Fig. 9B), with the consequence that the exposed N-terminus of the truncated channel is able to participate in competition for the N-terminal docking site on the full-length channel, to form a deleterious intermolecular interaction (Fig. 9B, C). When the RAR motif is mutated in the N-terminus of the truncated channels, it therefore reduces their ability to interact with docking site on the full length channel (Fig. 9B, C).  We have also shown that overexpression of an N-terminal peptide containing the RAR motif disrupted the deleterious interaction and interfered with the ability of the truncated domains to suppress trafficking and functional expression of the full-length channels (as represented in Fig. 9D). Although this effect was only partial in overexpression systems, strikingly, the same N-terminal peptide completely abolished the Fig. 8. Rescue by the free N-terminus of the suppression by EA2 of endogenous P/Q type current in DRG neurons. (A) Confocal images of DRG neurons transfected with empty pcDNA3 vector and YFP (control, top panel), EA2, pcDNA3 empty vector and YFP (middle panel) or EA2, Ca V 2.1-(46-100)-CAAX and YFP (bottom panel). The neurons were permeabilized and stained with Ca V 2.1 antibody that targets the II-III loop. (B) Representative traces of endogenous P/Q current. The DRG neurons were co-transfected with empty pcDNA3 vector and YFP (black); EA2 and pcDNA3 empty vector and YFP (red) or EA2, Ca V 2.1-(46-100)-CAAX and YFP (blue). The P/Q-type current was isolated pharmacologically. Currents were evoked by 50 ms step depolarizations between −50 and +60 mV from a holding potential of −80 mV. The charge carrier was 5 mM Ba 2+ . (C) Current-voltage relationships of DRG neurons transfected with pcDNA3 empty vector and YFP (black squares, n = 13), EA2, pcDNA3 empty vector and YFP (red circles, n = 10) or EA2, Ca V 2.1-(46-100)-CAAX and YFP (blue circles, n = 15). (D) Current density at +10 mV ± SEM for pcDNA3 empty vector and YFP (black, n = 13), EA2, pcDNA3 empty vector and YFP (red, n = 10) or EA2, Ca V 2.1-(46-100)-CAAX and YFP (blue, n = 15). Statistical analysis: *p b 0.05, **p b 0.01, ns = non-significant difference. deleterious effect of the EA2 mutant on the endogenous P/Q-type current in DRG neurons. This discrepancy is probably due to the fact that the interaction occurs very early during protein synthesis in the endoplasmic reticulum (Page et al., 2004;Mezghrani et al., 2008). Thus the complex, comprising full-length and truncated channels, is formed cotranslationally in expression systems, possibly involving additional sites as well as the N-terminus, and is retained in the endoplasmic reticulum with limited access to competition for its binding site by the free N-terminal construct. In contrast in DRG neurons the free N-terminus is probably expressed in excess compared to the endogenous channels, and better able to compete with EA2 for the N-terminal docking site on the full length channel. When the N-terminal RAR motif in the fulllength channel is mutated to AAA, we propose that the intramolecular interaction of the N-terminus with its own docking site may be weakened (Fig 9A). This would agree with our finding that the free N-terminus was not able to disrupt the deleterious interaction between Ca V 2.1 R57A, R59A and the EA2 mutant, and may indicate that the interaction is more robust and thus more difficult to disrupt.
The intramolecular docking mechanism we propose for Ca V 2 is reminiscent of the role of the tetramerization domain (T1 domain) in the Ntermini of Kv1-Kv4 potassium channels, which has been shown to be involved with the folding and oligomerization of the channels (Brueggemann et al., 2013). Oligomerization requires a threshold level of folding of the N-terminus and it has been suggested that this coupling between folding and assembly of Kv channels may be a common trait (Robinson and Deutsch, 2005). The interactions between T1 domains of Kv1.3 occur very early during protein synthesis in the endoplasmic reticulum, while the nascent peptides of different subunits are still attached to ribosomes (Lu et al., 2001).
In summary we have shown that uncoupling the deleterious interaction and thus releasing the full-length channel from its association with the truncated EA2 channel by expressing the key N-terminus residues Fig. 9. Diagram of the possible mechanism for involvement of the N-terminus in dominant-negative suppression. (A) The four domains of Ca V 2 channels are shown in light blue, orange, green and red. The N-terminus is shown interacting with an intramolecular docking site on the left, whereas on the right interaction is reduced by mutation of the RAR motif (**). (B) The two domains of the truncated constructs are shown in dark blue and brown. On the right the truncated construct has a mutated RAR (**). (C) The N-terminus of the truncated construct can interact with the docking site on the full-length channel, initiating aggregation. This interaction is reduced by mutation of the RAR motif. (D) The free N-terminus interacts with the docking site on the full-length channel reducing the ability of the truncated channel to interact with this site. partially restored channel functional expression in overexpression systems. More importantly the endogenous P/Q current in native neurons was fully recovered by expressing the same N-terminal peptide.