Mechanism of auxiliary β-subunit-mediated membrane targeting of L-type (Ca_V1.2) channels

Non-technical summary

Voltage-dependent L-type calcium (Ca_V1.2) channels are critical gateways for Ca²⁺ entry into excitable cells such as heart myocytes and neurons. This Ca²⁺ signal controls many essential physiological responses including triggering the heartbeat and regulating gene expression in nerve cells. Ca_V1.2 channels are multi-subunit proteins, comprising α_1C, β, and α₂δ subunits, and must target to the cell surface to function. Association of a pore-forming α_1C and cytosolic β is necessary for targeting Ca_V1.2 channels to the cell surface through poorly understood mechanisms. Here, using a chimeric channel strategy, we provide data that suggest β binding to the α_1C intracellular I–II loop causes a global rearrangement of intracellular domains, shifting a balance of power between export signals on the I–II loop and retention signals elsewhere on the channel. The results provide novel insights into the mechanism of a protein–protein interaction that is vital for forming functional Ca_V1.2 channels.

Abstract

Abstract

Ca²⁺ influx via Ca_V1/Ca_V2 channels drives processes ranging from neurotransmission to muscle contraction. Association of a pore-forming α₁ and cytosolic β is necessary for trafficking Ca_V1/Ca_V2 channels to the cell surface through poorly understood mechanisms. A prevalent idea suggests β binds the α₁ intracellular I–II loop, masking an endoplasmic reticulum (ER) retention signal as the dominant mechanism for Ca_V1/Ca_V2 channel membrane trafficking. There are hints that other α₁ subunit cytoplasmic domains may play a significant role, but the nature of their potential contribution is unclear. We assessed the roles of all intracellular domains of Ca_V1.2-α_1C by generating chimeras featuring substitutions of all possible permutations of intracellular loops/termini of α_1C into the β-independent Ca_V3.1-α_1G channel. Surprisingly, functional analyses demonstrated α_1C I–II loop strongly increases channel surface density while other cytoplasmic domains had a competing opposing effect. Alanine-scanning mutagenesis identified an acidic-residue putative ER export motif responsible for the I–II loop-mediated increase in channel surface density. β-dependent increase in current arose as an emergent property requiring four α_1C intracellular domains, with the I–II loop and C-terminus being essential. The results suggest β binding to the α_1C I–II loop causes a C-terminus-dependent rearrangement of intracellular domains, shifting a balance of power between export signals on the I–II loop and retention signals elsewhere.

Introduction

The entry of calcium ions into excitable cells through voltage-dependent Ca_V1 (Ca_V1.1 − 1.4) and –2 (Ca_V2.1 − 2.3) channels constitutes a prevalent and versatile signal transduction paradigm in biology. This basic mechanism is used to evoke neurotransmitter release that underlies synaptic transmission (Catterall & Few, 2008), control neuronal excitability by coupling to Ca²⁺-sensitive K⁺ channels (Fakler & Adelman, 2008), trigger excitation–contraction coupling in heart muscle (Bers, 2002), and regulate gene expression (Deisseroth et al. 2003). Hence, myriad biological processes critically depend on the proper cell surface targeting and function of Ca_V1/Ca_V2 channels. Functional Ca_V1/Ca_V2 channels are multi-subunit protein complexes containing a membrane-spanning α₁ subunit assembled with auxiliary proteins that include β (β₁−β₄) and α₂δ (α₂δ1−3) subunits, and calmodulin (Catterall & Few, 2008). While the α₁ subunit contains the voltage sensor and channel pore, its subcellular localization and biophysical properties are profoundly influenced by the accessory proteins. In particular, β subunits are essential determinants of channel behaviour, affecting both trafficking and gating (Dolphin, 2003; Buraei & Yang, 2010). Auxiliary βs induce a hyperpolarizing shift in the voltage dependence of channel activation, increase channel open probability (P_o), and determine channel inactivation properties (Perez-Reyes et al. 1992; Neely et al. 1993; De Waard & Campbell, 1995; Colecraft et al. 2002; Dolphin, 2003; Takahashi et al. 2004; Buraei & Yang, 2010). In addition to the biophysical modifications, a cardinal feature of Ca_V1/Ca_V2 channels is their reliance on association with a β for effectual targeting to the cell surface (Gao et al. 1999; Dolphin, 2003; Kanevsky & Dascal, 2006; Buraei & Yang, 2010; Obermair et al. 2010; Yang et al. 2010). Because this trafficking step is fundamental to the formation of functional Ca_V1/Ca_V2 channels, much effort has focused on elucidating how β subunits promote membrane-targeting of Ca_V channels.

Ca_Vβs contain a src homology 3 (SH3)/guanylate kinase (GK) structural module (Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004), identifying them as members of the membrane-associated guanylate kinase (MAGUK) family of scaffold proteins (Funke et al. 2005). The β GK domain binds with high affinity to a conserved 18-residue region (termed the α₁ interaction domain, or AID) in the α₁ subunit intracellular I–II loop (Pragnell et al. 1994; Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004). Mutations that selectively disrupt the β-AID interaction prevent β-induced channel targeting to the membrane, suggesting a dominant role for this association in regulating Ca_V channel trafficking (Van Petegem et al. 2008; Bourdin et al. 2010; Buraei & Yang, 2010; Obermair et al. 2010). One prevalent idea, based on experiments carried out on Ca_V2.1 channels, is that Ca_V1/Ca_V2 α₁ subunits possess an ER retention signal on the I–II loop that is masked upon β-binding, thus allowing forward trafficking of the channel to proceed (Bichet et al. 2000). However, several observations challenge the sufficiency and generality of this model to account for β-induced membrane targeting of Ca_V1/Ca_V2 channels (Buraei & Yang, 2010). First, the putative ER retention sequence on the α₁ I–II loop has not been identified in any Ca_V1/Ca_V2 channel (Bichet et al. 2000). Second, in contrast to Ca_V2.1, neither the Ca_V1.2 nor Ca_V2.2 α₁ subunit I–II loop displayed ER retention properties when fused to CD4 (Altier et al. 2011). Third, deletions in the Ca_V1.2 α_1C C-terminus that do not impact β binding to the channel can, nevertheless, severely diminish membrane trafficking (Gao et al. 2000). Fourth, point mutations in the α_1C C-terminus that disrupt apo-calmodulin, but not β subunit binding, also significantly impair channel trafficking (Wang et al. 2007; Bourdin et al. 2010). These discrepancies underscore a clear need for a unifying model that accounts for both the essential role of β binding to the I–II loop and the apparent importance of the C-terminus for Ca_V1.2 channel membrane trafficking. Furthermore, the potential contribution of other α₁ subunit intracellular domains (N-terminus, II–III loop, and III–IV loop) to β-induced trafficking has not been rigorously explored in any Ca_V1/Ca_V2 channel. This omission critically compromises the necessary dataset required to formulate a more complete model of β-induced trafficking of Ca_V1/Ca_V2 channels.

Identifying whether and how the various intracellular domains of α₁ contribute to β-dependent channel trafficking is a daunting task given their multiplicity and likely complex three-dimensional spatial arrangement. One typical approach to the problem has been to mutate or delete portions of α₁ intracellular domains and assess whether they compromise channel trafficking (Bichet et al. 2000; Wang et al. 2007). While such loss-of-function approaches provide important information about channel regions that may be involved in trafficking, the dataset produced typically yield only modest insights into the mechanistic basis for β-dependent regulation of Ca_V channel trafficking. For example, observations that deletions in the Ca_V1.2 α_1C C-terminus may impair channel trafficking simply informs that this intracellular domain is important for the process, but not how it is involved. An alternative reductionist approach involves splicing individual α₁ subunit intracellular domains to the C-terminus of the trans-membrane protein CD8 (or CD4) and observing how they impact targeting of this protein to the cell surface (Bichet et al. 2000; Cornet et al. 2002; Wang et al. 2007; Altier et al. 2011). This is a useful method that, nevertheless, suffers from two main disadvantages. First, in the context of the channel, three of the α₁ intracellular loops (the I–II, II–III, and III–IV loops) are geometrically constrained by having transmembrane regions bracketing either side. This configuration is lost when the loops are fused to CD8, and the resulting change in conformation could impact their functional properties. Second, this approach neglects possible interactions among the α₁ intracellular domains that could give rise to new emergent properties.

Here, we took a gain-of-function chimeric channel approach seeking to reconstitute β-dependent trafficking in a normally β-independent channel. While the chimeric channel strategy has its own inherent limitations, the approach offered the potentially unique advantage that Ca_V channel α₁ subunit intracellular domains could be used in a manner that preserved their conformations and spatial inter-relationships. This was achieved by systematically generating a series of 31 separate chimeric channels featuring a swap of all possible permutations of the intracellular domains of the β-dependent Ca_V1.2 α_1C into the analogous positions in the β-independent Ca_V3.1 channel α_1G subunit. Functional analyses of these chimeras yielded results that provided a different perspective of the determinants underlying β-mediated membrane trafficking of Ca_V1.2 channels, and suggest a new mechanistic model for this physiologically crucial phenomenon.

Methods

Generation of plasmid constructs

All the experiments comply with the policies and regulations of The Journal of Physiology given by Drummond (2009). Generation of plasmids encoding α_1C[BBS]–yellow fluorescent protein (YFP) and β_2a–cyan fluorescent protein (CFP) have been previously described (Yang et al. 2010). Chimeric channels featuring a systematic swap of intracellular domains of rabbit Ca_V1.2 α_1C (accession no.: X15539) into the rat Ca_V3.1 α_1G (accession no.: AF027984) backbone were generated using the in-fusion cloning technique (Clontech, Mountain View, CA, USA) according to the manufacturer's instructions. To generate chimeras in which only one intracellular domain of α_1G-YFP was replaced by the corresponding intracellular domain from α_1C, four specially designed primers were used – two primers for PCR amplification of the vector containing the α_1G segment, and two primers for PCR amplification of the relevant α_1C segment as the insert (online Supplemental Material, Table S1). The in-fusion reaction was carried out with purified vector and insert in a 1:2 molar ratio, and the resulting chimeric constructs subsequently cloned. Replacement of multiple intracellular domains was achieved by combining the in-fusion technique with traditional restriction enzyme digestion and ligation methods. The precise boundaries of the N- and C-termini, and the intracellular loops of Ca_V1.2 and Ca_V3.1 used to generate the chimeras are shown in the online Supplemental Material, Fig. S1.

To generate C-terminus-truncated α_1C[BBS]-YFP constructs, we used PCR to introduce a XhoI site immediately downstream of the sequence for the last desired residue (G1540, D1632, L1732, or K1906) in the context of α_1C[BBS]. YFP was then PCR amplified and cloned in-frame and downstream of truncated α_1C using XhoI and XbaI sites. Point mutations changing acidic residues (D or E) to alanines in the α_1C I–II loop were accomplished using the QuikChange Lightning Site-Directed Mutagenesis Kit (Stratagene) according to manufacturer's instructions. All constructs were verified by sequencing.

Cell culture and transfection

Low-passage-number human embryonic kidney (HEK) 293 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 100 μg ml⁻¹ penicillin–streptomycin at 37°C. HEK 293 cells were transiently transfected with appropriate constructs – 6 μg wild-type or modified α_1C or α_1G, 6 μg β_2a (as needed), and 1 μg T antigen using calcium phosphate precipitation – and cultured in supplemented DMEM at 37°C for 48–72 h before use.

Electrophysiology

Whole-cell recordings were conducted 48–72 h after transfection using an EPC-10 patch clamp amplifier (HEKA Electronics, Lambrecht, Germany) controlled by PULSE software (HEKA). Micropipettes were prepared from 1.5 mm thin-walled glass (World Precision Instruments, Sarasota, FL, USA) with a filament micropipette puller (P-97, Sutter Instrument Co., Novato, CA, USA). The internal solution contained (in mm): 135 caesium methanesulphonate (MeSO₃), 5 CsCl, 5 EGTA, 1 MgCl₂, 4 MgATP (added fresh) and 10 Hepes (pH 7.3). External solution contained (in mm): 140 tetraethylammonium-MeSO₃, 5 BaCl₂ and 10 Hepes (pH 7.4). When filled with internal solution, the resistance of the pipette was typically 1.5–2 MΩ. Whole-cell I–V curves were generated from a family of step depolarizations (−50 to +50 mV from a holding potential of −90 mV for α_1C constructs, or −100 to +40 mV from a holding potential of −100 mV for wild-type α_1G and chimeric constructs). Currents were sampled at 25 kHz and filtered at 5 or 10 kHz. Traces were acquired at a repetition interval of 6 s. Leak and capacitive currents were subtracted using a P/8 protocol. I−V curves from individual cells were fit using a least-squares method to the following modified Boltzmann equation:

where, G is the specific conductance, V_rev is the reversal potential, V_0.5 is the potential of half-maximal activation, and k is a slope factor.

Detection and quantification of cell surface Cav1.2 channels with quantum dots

Relative surface expression of epitope-tagged α_1C subunits was quantitatively determined using quantum dot labelling and flow cytometry as previously described (Yang et al. 2010). Briefly, HEK 293 cells transfected with BBS-tagged α_1C-YFP constructs in six-well tissue culture dishes were washed twice with phosphate-buffered saline (PBS) containing Ca²⁺ and Mg²⁺, and sequentially incubated with 1 μm biotinylated α-bungarotoxin (BTX) in DMEM–3% BSA at room temperature for 1 h and 5 nm streptavidin-conjugated quantum dot (QD₆₅₅, Invitrogen) at 4°C for 1 h in the dark. Surface labelled HEK 293 cells were harvested with trypsin, washed with PBS and assayed by flow cytometry using a BD LSRII Cell Analyzer (BD Biosciences, San Jose, CA, USA). CFP- and YFP-tagged proteins were excited at 407 and 488 nm, respectively, and red quantum dot signal was excited at 633 nm. For each group of experiments, we used isochronal untransfected and single colour controls to manually set the appropriate gain settings for each fluorophore to ensure signals remained in the linear range, and to set threshold values. The same gain settings were then used for assaying all isochronal transfection samples.

Flow cytometry data were analysed using FlowJo software. To normalize for protein expression, analysis of QD fluorescence was conducted over a window selected such that the mean YFP fluorescence intensity registered 1000 arbitrary units across all groups. In the case of α_1CΔ1632-YFP and α_1CΔ1540-YFP, there was an apparent decrease in protein expression. Therefore, for these constructs the comparisons to control were done over an analysis window with a mean YFP fluorescence intensity of 400 arbitrary units (Supplemental Material, Fig. S6). For each experimental condition, the ratio of the mean QD fluorescence intensity to the mean YFP fluorescence intensity (R_QD/YFP) was calculated and normalized to the R_QD/YFP obtained for isochronal control cells expressing α_1C+β_2a channels.

Fluorescence imaging

Fluorescence images of labelled (YFP and quantum dot) HEK 293 cell suspensions were obtained using an inverted Nikon Eclipse Ti microscope equipped for epifluorescence. Fluorophores were excited with the appropriate wavelength using a DeltaRAM Random Access Monochromator (Photon Technology International, Birmingham, NJ, USA) and the emitted light imaged with a QuantEM CCD camera (Roper Scientific, Trenton, NJ, USA).

Western blotting

Transiently transfected HEK 293 cells were harvested with trypsin, washed with PBS and solubilized in lysis buffer (20 mm Tris base, 1 mm EDTA, 150 mm NaCl, 1% SDS, 0.1% Triton X-100, pH 7.5), supplemented with Complete Mini protease inhibitor cocktail (Roche), by brief sonication. Protein concentrations of the cell lysates were determined with Pierce BCA Protein Assay Kit (Thermo Scientific). Following addition of sample buffer (50 mm Tris base, 10% glycerol, 2% sodium dodecyl sulfate (SDS), 50 mm dithiothreitol (DTT), 0.2 mg ml⁻¹ bromphenol blue, pH 7.5), cell lysates were resolved by SDS–PAGE (NuPage 10% gel, Invitrogen) at a constant voltage of 200 V for 1 h. Protein bands were then electrotransferred to a 0.45 μm nitrocellulose membrane for 2.5 h at 4°C at a constant voltage of 30 V in transfer buffer (96 mm glycine, 12 mm Tris base, 0.01% SDS, 20% methanol, pH 8.3). The membranes were blocked for 1 h at room temperature with 5% milk in TBS-T buffer (20 mm Tris base, 140 mm NaCl, 0.1% Tween-20, pH 7.6) and incubated overnight at 4°C with rabbit anti-GFP antibody (1:10,000) in TBS-T. The blots were washed with TBS-T and incubated with goat anti-rabbit antibody (1:10,000) for 1 h at room temperature. After further washing, the immune complexes were visualized using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific).

Data and statistical analyses

Electrophysiological data were analysed off-line using built in functions in PulseFit (HEKA), Microsoft Excel and Origin software. Statistically significant differences between means (P < 0.05) were determined using one-way ANOVA followed by pairwise comparisons using the Bonferroni test for multiple group comparisons, and either Student's unpaired t test or Wilcoxon's rank sum test for comparisons between two groups. Data are presented as means ± SEM.

Results

Differential impact of Ca_Vβ on functional expression of Ca_V1.2 (α_1C) and Ca_V3.1 (α_1G) channels

We examined the impact of auxiliary β subunits on functional expression of Ca_V1.2 and Ca_V3.1 channels reconstituted in HEK 293 cells by transient transfection. Cells transfected with the Ca_V1.2 α_1C-YFP subunit alone displayed barely detectable currents across the relevant range of test pulse potentials (Fig. 1A and B). Co-expression of α_1C-YFP with β_2a-CFP resulted in a dramatic 15-fold increase in whole-cell current amplitude, accompanied by a 10 mV hyperpolarizing shift in the voltage dependence of channel activation (Fig. 1A and B; Table 1). Ca_V1.2 channels activated at a threshold voltage of −30 mV and peaked at either 0 mV (with β_2a) or +10 mV (no β) (Fig. 1A and B). In sharp contrast, cells transfected with Ca_V3.1 α_1G-YFP subunit alone displayed large whole-cell currents that activated at a threshold potential of −60 mV and peaked at −30 mV (Fig. 1C and D), consistent with its classification as a low voltage-activated calcium channel (Perez-Reyes et al. 1998). Co-expression of α_1G-YFP with β_2a-CFP had no impact on either the whole-cell current amplitude or the voltage dependence of channel gating compared to cells expressing α_1G alone (Fig. 1C and D; Table 1). These results recapitulate the well-known impact of β subunits on the functional expression of high-voltage-activated Ca_V1 and Ca_V2 channels, and the relative β independence of low voltage-activated Ca_V3 channels.

Figure 1. Divergent regulation of Ca_V1.2 and Ca_V3.1 channels by auxiliary β subunits.

A, top, schematic diagram illustrating topology of Ca_V1.2 α_1C subunit. Four homologous transmembrane domains (boxes labelled I–IV) are connected and flanked by five intracellular modules (red lines). Bottom, exemplar whole-cell currents from Ca_V1.2 channels reconstituted without (left) or with (right) auxiliary β subunits. Displayed currents were elicited by voltage steps to −30, −10, +10 and +30 mV. B, population peak current density versus voltage (I−V) relationship for Ca_V1.2 channels reconstituted with either α_1C alone (▵, n = 6 for each point) or α_1C+β_2a (▴, n = 5). C, top, topological illustration of Ca_V3.1 α_1G subunit. Bottom, exemplar currents from Ca_V3.1 channels reconstituted without (left) or with (right) auxiliary β subunits. D, population I−V relationship for Ca_V3.1 channels reconstituted with either α_1G alone (□, n = 12) or α_1G+β_2a (▪, n = 12). E, fluorescence images of HEK 293 cells expressing either α_1C[BBS]-YFP alone (top) or α_1C[BBS]-YFP +β_2a (bottom). Left, YFP fluorescence shows total α_1C expressed in the cells. Right, quantum dot (QD₆₅₅) fluorescence specifically labels α_1C channels at the cell surface. F, left, representative flow cytometry results from live HEK 293 cells transiently transfected with either α_1C[BBS]-YFP alone (top) or α_1C[BBS]-YFP +β_2a (bottom), and with surface channels labelled with QD₆₅₅. Right, relative surface density of Ca_V1.2 channels as reported by normalized QD₆₅₅ fluorescence intensity.

Table 1.

Gating parameters for reconstituted channels

Construct	Gmax (pS)	V0.5 (mV)	k	n
α1C	0.19 ± 0.12	13.91 ± 11.36	13.59 ± 3.78	6
α1C+β2a	1.98 ± 0.70†	−7.81 ± 2.95	6.19 ± 0.83	5
α1G	0.96 ± 0.15	−38.86 ± 0.90	5.10 ± 0.49	12
α1G+β2a	1.04 ± 0.09	−43.45 ± 1.59	4.56 ± 0.23	12
α1G[cgggg]	0.23 ± 0.10*	−35.09 ± 4.47	13.57 ± 7.20	6
α1G[cgggg]+β2a	0.19 ± 0.07	−41.92 ± 3.30	6.68 ± 1.37	4
α1G[gcggg]	4.08 ± 0.70*	−68.61 ± 1.75*	1.25 ± 0.35*	8
α1G[gcggg]+β2a	1.35 ± 0.23†	−69.76 ± 2.37	1.24 ± 0.53	5
α1G[ggcgg]	0.55 ± 0.12	−40.39 ± 0.52	5.59 ± 0.20	10
α1G[ggcgg]+β2a	0.35 ± 0.09	−39.07 ± 1.16	5.67 ± 0.24	10
α1G[gggcg]	0.17 ± 0.06*	−33.26 ± 1.55*	7.71 ± 0.94*	7
α1G[gggcg]+β2a	0.23 ± 0.11	−33.82 ± 1.55	6.57 ± 1.05	6
α1G[ggggc]	0.30 ± 0.12*	−38.65 ± 2.08	7.89 ± 1.33*	8
α1G[ggggc]+β2a	0.21 ± 0.11	−44.43 ± 2.64	7.40 ± 0.33	4

^†P < 0.05 compared to the corresponding –β data, Wilcoxon's rank sum test.

^*P < 0.05 compared to α_1G data using Wilcoxon's rank sum test.

In mammalian cells, β subunits markedly enhance the membrane trafficking of Ca_V1/Ca_V2 channels to the plasma membrane, and this represents a major mechanistic contributor to the β-induced increase in whole-cell current amplitude. This effect of β on Ca_V1.2 channels can be readily demonstrated in our reconstituted system in two independent ways. First, the maximal gating current recorded in cells expressing α_1C either in the absence or the presence of β provides a measure of the number of channels with moveable voltage sensors in the membrane (Neely et al. 1993; Takahashi et al. 2004). Hence, based on the assumption that β subunits do not alter the unitary gating charge required to open the channel, the time integral of the maximum gating charge (Q_max) provides an index of the number of channels in the membrane. Using this metric, cells expressing α_1C-YFP +β_2a-CFP display a significantly larger Q_max, and, therefore, more surface channels than cells transfected with α_1C-YFP alone (Fig. 1E). By contrast, β_2a-CFP had no effect on Q_max obtained from cells expressing α_1G subunits (Fig. 1F), fitting with the lack of impact of the auxiliary subunit on whole-cell current amplitude in Ca_V3.1 channels. A second, more direct way of visualizing Ca_V1.2 channels at the membrane involves labelling cell surface channels with quantum dots followed by quantification of signals using flow cytometry (Yang et al. 2010). Here, a 13-residue bungarotoxin-binding site (BBS) (Sekine-Aizawa & Huganir, 2004) is introduced into the extracellular domain II S5–S6 loop of α_1C-YFP, and surface channels are selectively labelled in non-permeabilized cells by sequential exposure to biotinylated bungarotoxin and streptavidin-conjugated quantum dot (Fig. 1G). The relative cell surface density of Ca_V1.2 channels is then quantified in a high throughput manner by flow cytometry. Using this method, the presence of β_2a-CFP resulted in a 12-fold increase in QD₆₅₅ fluorescence intensity compared to cells expressing BBS epitope-tagged α_1C-YFP alone (Fig. 1H).

The stark contrast between Ca_V1.2 α_1C and Ca_V3.1 α_1G subunits could potentially be exploited to identify the important determinants and mechanisms underlying β-dependent trafficking. Given that the β subunit is entirely intracellular, we hypothesized that cytoplasmic domains of the α_1C subunit play a prominent role in Ca_V1.2 channel ER retention and β-dependent ER export. Accordingly, we systematically generated 31 separate chimeric channels featuring all possible permutations of the five α_1C intracellular domains (N-terminus, I–II loop, II–III loop, III–IV loop and C-terminus) swapped into the analogous regions of α_1G and assessed the resulting functional outcomes.

Functional outcomes of single intracellular domain-substituted chimeras

We first investigated whether a single α_1C subunit intracellular domain could potentially confer ER retention and β-dependent ER export when transplanted into α_1G. We generated five individual chimeras (termed α_1G[cgggg], α_1G[gcggg], α_1G[ggcgg], α_1G[gggcg] and α_1G[ggggc]) in which just one intracellular region of α_1C was swapped into the α_1G backbone (Fig. 2). We adopted a naming convention in which the backbone channel subunit is followed by a square bracket containing letters that denote the configuration of the intracellular domains. Hence, for example, α_1G[cgggg] refers to the chimeric channel in which the N-terminus of α_1G is replaced with the analogous segment from α_1C. To assess the functional impact of the single-domain substitutions, we transiently expressed the chimeric channels in HEK 293 cells in either the absence or the presence of β_2a-CFP and recorded whole-cell currents (Fig. 2). All chimeras, here and throughout, were tagged with YFP enabling direct visual confirmation of expression to be used as a criterion for cell selection. All the chimeras expressed full-length channels as assessed by Western blot using an anti-GFP antibody (Fig. 2K, inset). Cells expressing α_1G[cgggg] alone displayed whole-cell currents that were significantly smaller at all test voltages compared to α_1G alone channels (Fig. 2A and B, open symbols; Table 1). When peak amplitudes are compared at −30 mV to ensure a similar driving force, α_1G[cgggg] channels exhibited a substantive fourfold decrease in current amplitude (Fig. 2K). Co-expressing β_2a had no impact on either current amplitude or voltage dependence in this chimeric channel (Fig. 2A, B and K).

Figure 2. Functional outcomes of chimeras featuring substitutions of individual intracellular domains from Ca_V1.2 α_1C into Ca_V3.1 α_1G subunit.

A, top, topological illustration of the chimeric channel α_1G[cgggg], which features the sole substitution of the α_1C N-terminus into α_1G. Bottom, exemplar whole-cell currents from cells expressing either α_1G[cgggg] alone (left) or α_1G[cgggg]+β_2a (right). Displayed currents were elicited by voltage steps to −50, −30, −10 and +10 mV. B, I−V curves for channels reconstituted with α_1G[cgggg] alone (□) or α_1G[cgggg]+β_2a (▪). Data for wild-type α_1G (cyan trace) have been reproduced to facilitate direct visual comparison. C−J, topological illustrations, exemplar currents and population I−V relationships for the remaining single intracellular domain substituted chimeras α_1G[gcggg], α_1G[ggcgg], α_1G[gggcg] and α_1G[ggggc], respectively. Same format as A and B. K, comparison of peak current densities acquired at −30 mV for the various chimeras in the absence and presence of β_2a. The average current density at −30 mV for wild-type α_1G alone channels is represented (cyan dashed line) to facilitate visual comparison. **P < 0.005 for grouped data (±β_2a) compared to α_1G (±β_2a) using one-way ANOVA followed by means comparisons using the Bonferroni test. *P < 0.001 compared to α_1G (±β_2a), one-way ANOVA followed by Bonferroni test. #P < 0.05, Wilcoxon's rank sum test. Inset, Western blot: 1-α_1G; 2-α_1G[cgggg]; 3-α_1G[gcggg]; 4-α_1G[ggcgg]; 5-α_1G[gggcg]; 6-α_1G[ggggc].

In striking contrast to the results obtained with α_1G[cgggg], cells expressing α_1G[gcggg], a chimera featuring a swap of the entire α_1C I–II loop into α_1G, demonstrated a dramatic increase in current amplitude and a 30 mV leftward shift in the I−V relationship (Fig. 2C and D, open symbols; Table 1). When compared at a −30 mV test pulse, α_1G[gcggg] channels exhibited an over threefold increase in current density compared to wild-type α_1G-alone channels (Fig. 2F; I_peak = 62.6 ± 12.1 pA pF⁻¹, n = 12, for α_1G, and I_peak = 197.4 ± 23.0 pA pF⁻¹, n = 8 for α_1G[gcggg] channels, P < 0.05, one-way ANOVA). The surprising result with the α_1G[gcggg] chimera confirms recent findings (Arias et al. 2005; Fan et al. 2010), and are seemingly at odds with the notion that the I–II loop of Ca_V1 and Ca_V2 channels harbours a dominant ER retention signal. Co-expressing β_2a unexpectedly diminished, rather than increased α_1G[gcggg] currents, without affecting the voltage dependence of channel gating (Fig. 2C, D and K; Table 1).

Similar to the results obtained with α_1G[cgggg], cells transfected with α_1G[ggcgg], α_1G[gggcg], or α_1G[ggggc] displayed significantly smaller currents compared to wild-type α_1G, and were β independent (Fig. 2E–K; Table 1). Additionally, these other chimeras had only modest, if any, effects on the voltage dependence of channel activation (Table 1). These functional results were not correlated with differences in protein expression (Fig. 2K, inset). Overall, these results suggested that no single α_1C intracellular domain could confer β-dependent plasma membrane targeting to the α_1G backbone. Moreover, the data also raised the radical possibility that the α_1C I–II loop could actually facilitate ER export rather than retention in the context of Ca_V channel α₁ subunits. We next sought to test this possibility.

Identification of a putative ER export signal in the α_1C I–II loop

It was possible that the variance in current density between wild-type α_1G and the single-domain-swapped chimeras was due to changes in channel gating (such as distinctions in channel open probability) rather than differences in channel trafficking. As such, it was important to determine whether the different chimeras displayed divergent propensities to target to the cell surface. Unfortunately, the quantum dot labelling method for surface channels was unsuccessful in the context of Ca_V3.1, probably due to geometric constraints that limited accessibility to the surface epitope tag. Fortunately, however, the α_1G subunit exhibits robust gating currents, thus rendering it possible to measure Q_max as an index of channel trafficking to the membrane. Compared to wild-type α_1G, α_1G[gcggg] displayed a significantly larger gating current and Q_max, while all the other single-domain-swap chimeras (except α_1G[ggcgg]) exhibited a diminished Q_max (Fig. 3A). This result indicates that α_1G[gcggg] shows markedly improved membrane targeting, whereas the other chimeras show a relative retention compared to wild-type α_1G.

Figure 3. Identification of an acidic sequence that functions as an ER export signal in α_1C I–II loop.

A, top, exemplar gating currents for wild-type α_1G and single domain-substituted chimeric channels. Bottom, comparison of maximal gating charge (Q_max) for the different channel types. *P < 0.05 compared to α_1G. B, sequence of the α_1C I–II loop showing candidate acidic-residue ER export signals that were mutated to alanines in the context of the α_1G[gcggg] chimera to generate five distinct mutants, AXA1−AXA5. C, Western blot: 1, AXA1; 2, AXA2; 3, AXA3; 4, AXA4; 5, AXA5. D, population I−V curves for chimeric mutant channels AXA1 (○, n = 9), AXA2 (▵, n = 6), AXA4 (▿, n = 6) and AXA5 (⋄, n = 6). Data for wild-type α_1G (cyan trace) and α_1G[gcggg] (red trace) are reproduced to facilitate visual comparison. E, I−V curve for AXA3 mutant chimeric channels (□, n = 8). F, comparison of Q_max for the different AXA mutant chimeric channels. Mean data for wild-type α_1G (dashed cyan line) and α_1G[gcggg] (dashed red line) are represented. *P < 0.05 compared to α_1G[gcggg], Wilcoxon's rank sum test.

We hypothesized that the α_1C I–II loop contains an ER export signal that could potentially explain the enhanced membrane targeting of α_1G[gcggg]. Di-acidic motifs (DXD, DXE, EXD and EXE, where X is any amino acid) have been found to act as ER export signals in many membrane proteins (Ma et al. 2001; Ma et al. 2002; Mikosch & Homann, 2009). We searched for possible acidic residue ER export signals by scanning the α_1C I–II loop sequence (Fig. 3B; Supplemental Material, Fig. S2). We identified five candidate regions with acidic residue motifs and generated five distinct mutants (termed AXA1–AXA5) in which acidic amino acids (D or E) in the α_1C I–II loop were changed to alanine, all within the context of the α_1G[gcggg] chimera (Fig. 3B). All five chimeras expressed full-length channels (Fig. 3C). Whole-cell currents were recorded from HEK 293 cells transfected with each mutant in the absence of β, and I−V curves were generated (Fig. 3D). The mutant chimeras AXA1, AXA2, AXA4 and AXA5 displayed current densities that were only modestly reduced compared to α_1G[gcggg] (Fig. 3D), indicating that the acidic residues present in the associated regions do not account for the ER export capabilities of the α_1C I–II loop. By contrast, the mutant chimera AXA3 exhibited a current density that was essentially identical to wild-type α_1G (Fig. 3E), indicating that the nine acidic residues mutated in this cluster accounted wholly for the increased current density observed with the α_1G[gcggg] chimera. Significantly, AXA3 still displayed a V_1/2 that was substantially left-shifted compared to wild-type α_1G (Fig. 3E). Therefore, the increase in current density and leftward-shift in activation are separable functions conferred by the α_1C I–II loop in the context of α_1G[gcggg]. Gating charge analyses (Fig. 3F) demonstrated that AXA3 displayed a Q_max that was significantly smaller than obtained with α_1G[gcggg] (Q_max = 6.01 ± 1.16 fC pF⁻¹, n = 8 for AXA3, and Q_max = 14.48 ± 2.83 fC pF⁻¹, n = 7, for α_1G[gcggg]-alone channels, P < 0.05), and essentially identical to wild-type α_1G (Q_max = 6.95 ± 1.22 fC pF⁻¹, n = 7). The remaining mutated chimeras (AXA1, AXA2, AXA4 and AXA5), by contrast, exhibited Q_max values that were not significantly different from α_1G[gcggg] (Fig. 3F). Taken together, the results suggest that the acidic residue cluster mutated in AXA3 is a putative ER export region (PEER; Supplemental Material, Fig. S2) in the α_1C I–II loop, and that the other α_1C intracellular domains may either confer ER retention propensities or increase the rate of channel removal from the cell surface.

Functional interactions among putative α_1C ER export and retention modules probed with double and triple domain-substituted chimeras

How do the putative ER retention and export signals functionally interact with each other, and what is the minimum combination of domains required to confer β dependency to channel regulation? To address these questions we generated chimeric channels featuring multiple intracellular domains of α_1C swapped into α_1G, starting with the 10 possible chimeras containing double-domain substitutions (Fig. 4). For these double-domain chimeras, there were two types of questions of primary interest. First, how did the putative ER retention characteristics of the other intracellular domains functionally interact with the export capabilities of the I–II loop? Second, were the putative ER retention properties of individual intracellular domains synergistic when present on the same molecule? Examination of the electrophysiological properties of the four chimeras in which α_1C I–II loop was paired with each of the other intracellular domains suggested varying strengths among the distinct ER retention regions (Fig. 4). Cells expressing α_1G[ccggg] alone displayed an intermediate current density that lay between wild-type α_1G and α_1G[gcggg] (Fig. 4A and E), suggesting α_1C N-terminus moderately opposed the ability of the α_1C I–II loop to increase channel surface density. Cells expressing α_1G[gccgg] alone displayed large currents that were indistinguishable from α_1G[gcggg] channels (Fig. 4B and E), indicating that the α_1C II–III loop is unable by itself to diminish the trafficking function of α_1C I–II loop. At the other extreme, α_1G[gcgcg] and α_1G[gcggc] channels exhibited relatively small currents similar in amplitude to wild-type α_1G channels (Fig. 4C–E). Hence, α_1C III–IV loop and C-terminus completely neutralized the α_1C I–II loop enhanced trafficking effect. All other chimeras that contained two α_1C intracellular domains, exclusive of the I–II loop, exhibited exceptionally small currents (Fig. 4E; Supplemental Material, Fig. S3). None of the double domain-substituted chimeras showed a β-dependent increase in current density –α_1G[gccgg] channels displayed a β-dependent decrease in current density (Fig. 4B and E) similar to that observed with α_1G[gcggg]. These results were not due to differences in protein expression among the distinct chimeric channels (Fig. 4E, inset). Overall, these results indicated that the putative ER retention signals on the α_1C intracellular loops acted synergistically, and could counteract the propensity of α_1C I–II loop to increase channel surface density, but with differing efficacies. Moreover, no two α_1C intracellular domains were sufficient to reconstitute β-dependent increased channel trafficking.

Figure 4. Functional outcomes of chimeras featuring two intracellular domains from α_1C substituted into α_1G.

A−D, topological illustrations and population I−V curves (±β) for those double intracellular domain swapped chimeras that include the α_1C I–II loop. Data for wild-type α_1G (cyan trace) and α_1G[gcggg] (red trace) are reproduced for comparison. E, peak current densities acquired at −30 mV for all double intracellular domain swapped chimeras compared to wild-type α_1G (dashed cyan line). **P < 0.005 for grouped data (±β_2a) compared to α_1G (±β_2a) using one-way ANOVA followed by means comparisons using the Bonferroni test. *P < 0.001 compared to α_1G (±β_2a), one-way ANOVA followed by the Bonferroni test. Inset, Western blot: 1, α_1G[ccggg]; 2, α_1G[cgcgg]; 3, α_1G[cggcg]; 4, α_1G[cgggc]; 5, α_1G[gccgg]; 6, α_1G[gcgcg]; 7, α_1G[gcggc]; 8, α_1G[ggccg]; 9, α_1G[ggcgc]; 10, α_1G[gggcc].

These observations were consolidated and extended by considering the functional outcomes of chimeras in which three α_1C intracellular domains were swapped into α_1G (Fig. 5). All the triple-substituted chimeras which included the α_1C I–II loop exhibited currents with amplitudes that were either on par with or less than wild-type α_1G (Fig. 5). When the α_1C I–II loop was absent, the triple mutant chimeras did not register discernible currents (Fig. 5G; Supplemental Material, Fig. S4), consistent with the idea that the discrete putative ER retention signals acted synergistically to retain channels inside the cell. All the triple-substituted chimeras expressed full-length proteins (Fig. 5G, inset), and the functional results were not correlated with differences in protein expression. Finally, none of the triple-substituted chimeras displayed β-dependent increase in current density, further emphasizing the surprisingly complex nature of this phenomenon.

Figure 5. Functional outcomes of chimeras featuring three intracellular domains from α_1C substituted into α_1G.

A−F, topological illustrations and population I−V curves (±β) for those triple intracellular domain-swapped chimeras that include the α_1C I–II loop. Data for wild-type α_1G (cyan trace) and α_1G[gcggg] (red trace) are reproduced for comparison. G, peak current densities acquired at −30 mV for all triple intracellular domain-swapped chimeras compared to wild-type α_1G (dashed cyan line) and α_1G[gcggg] channels (dashed red line). **P < 0.005 for grouped data (±β_2a) compared to α_1G (±β_2a) using one-way ANOVA followed by means comparisons using the Bonferroni test. Inset, Western blot: 1, α_1G[cccgg]; 2, α_1G[ccgcg]; 3, α_1G[ccggc]; 4, α_1G[cgccg]; 5, α_1G[cgcgc]; 6, α_1G[cggcc]; 7, α_1G[gcccg]; 8, α_1G[gccgc]; 9, α_1G[gcgcc]; 10, α_1G[ggccc].

Chimeras displaying successful reconstitution of β-dependent channel regulation

Arguably, the most profound insights into the mechanistic bases of β-dependent channel regulation may be provided by successfully reconstituting this phenomenon in a normally β-independent channel background. As such, it was instructive that three distinct chimeras (α_1G[cccgc], α_1G[ccgcc] and α_1G[gcccc]) characterized by four intracellular domains of α_1C substituted into the α_1G background recapitulated a β-dependent increase in current density that is normally only seen in Ca_V1/Ca_V2 channels (Fig. 6). Two other quadruple domain-substituted chimeras, α_1G[cgccc] and α_1G[ccccg], displayed no current in either the absence or presence of β (Fig. 6A, D and F). The fact that α_1G[cgccc] produced no currents was not surprising based on the emerging concept that this chimera essentially contains four α_1C ER retention modules while lacking the robust ER export services of the α_1C I–II loop. Less predictable was the absence of current observed with the α_1G[ccccg] chimera. Together with results from other chimeras, this finding suggests that the α_1C C-terminus is absolutely essential for channel trafficking to the membrane and β-dependent channel regulation, but only when it is assembled with at least three other α_1C intracellular domains on the same channel molecule. A chimera in which all five intracellular domains of α_1C were placed into α_1G also displayed currents which trended higher in the presence of β, although this effect did not reach statistical significance (Fig. 6F and G). All the chimeric channels expressed well (Fig. 6G, inset), ruling out differences in protein expression as a potential trivial explanation for the results.

Figure 6. Reconstitution of β regulation of current density in specific chimeric channels containing four α_1C intracellular domains swapped into α_1G.

A−F, topological illustrations and population I−V curves (±β) for quadruple and quintuple intracellular domain-swapped chimeras. Data for wild-type α_1G (cyan trace) and α_1G[gcggg] (red trace) are reproduced for comparison. *P < 0.05 compared to corresponding −β data, Wilcoxon's rank sum test. G, peak current densities acquired at −30 mV for quadruple and quintuple intracellular domain-swapped chimeras compared to wild-type α_1G (dashed cyan line) channels. Inset, Western blot: 1, α_1G[ccccg]; 2, α_1G[cccgc]; 3, α_1G[ccgcc]; 4, α_1G[cgccc]; 5, α_1G[ccccc].

Overall, these results demonstrate that a minimum of four α_1C intracellular domains that must include the I–II loop and C-terminus are necessary and sufficient to reconstitute β-dependent regulation of current density in α_1G/α_1C chimeric channels.

Trafficking role of the I–II loop and C-terminus in the α_1C subunit context

Two remaining questions were: (1) does the newly identified PEER in the α_1C I–II loop regulate channel surface density in the context of wild-type Ca_V1.2 α_1C subunit? and (2) how and why is the C-terminus important for Ca_V1.2 channel targeting? To address the hypothesized ER export function of the α_1C I–II loop in the native channel, we introduced the distinct AXA1–AXA5 mutations (Fig. 3B) in the context of α_1C[BBS]-YFP (Fig. 7A), and monitored channel targeting to the cell surface using the quantum dot labelling method (Fig. 7B). Compared to wild-type α_1C[BBS]-YFP +β_2a, the α_1C[AXA3] mutant (+β_2a) exhibited a 60% decrease in QD₆₅₅ fluorescence intensity (Fig. 7C), supporting the idea that the PEER is important for channel trafficking within the context of the native Ca_V1.2 channel. The fact that β-dependent channel trafficking was not completely abolished in the AXA3 mutant channel may indicate that other ER export signals reside elsewhere on the α_1C subunit. Channels bearing mutations of other acidic residues in the I–II loop (AXA2, AXA4 and AXA5) also displayed moderate decreases in trafficking (Fig. 7C). However, combining mutations (AXA345) did not decrease the trafficking beyond what was observed with AXA3 alone (data not shown).

Figure 7. Role of the I–II loop and C-terminus in β-dependent trafficking of Ca_V1.2 α_1C subunit.

A, topological illustration of α_1C with AXA1–AXA5 mutations in the I–II loop. B, exemplar flow cytometry results examining surface expression of α_1C[AXA1] (top) and α_1C[AXA3] (bottom) ±β_2a using the quantum dot labelling method. C, relative surface expression of (normalized QD₆₅₅ fluorescence intensities) for the distinct α_1C AXA mutants. *P < 0.05 compared to α_1C[BBS]-YFP +β_2a. #P < 0.05 compared to the corresponding –β_2a data. D, left, topological illustration of serial C-terminus truncation mutants of α_1C showing the relative position of CaM-binding pre-IQ and IQ sites. Right, impact of C-terminus truncations on relative surface expression of α_1C subunits ±β_2a. E and F, impact of combining C-terminus truncations and AXA1–AXA5 mutations on relative surface expression of α_1C subunits, n = 3−4. Dashed lines represent means of data for α_1CΔ1906 +β_2a and α_1CΔ1732 +β_2a, respectively, and the corresponding 95% confidence intervals (shaded regions). *P < 0.05 compared to α_1CΔ1906 +β_2a and α_1CΔ1732 +β_2a, respectively. #P < 0.05 compared to the corresponding –β_2a data.

The α_1C C-terminus is a hub of protein–protein interactions that modulate the Ca_V1.2 channel (Supplemental Material, Fig. S5). In particular, the proximal C-termini of Ca_V1 and 2 channels contain pre-IQ and IQ regions (Fig. 7D; Supplemental Material, Fig. S5) which serve as binding sites for apo-calmodulin (CaM) and Ca²⁺–CaM (Romanin et al. 2000; Pitt et al. 2001; Erickson et al. 2003; Kim et al. 2004; Van Petegem et al. 2005), and are critical for feedback regulation of Ca_V channels (inactivation or facilitation) by Ca²⁺ ions (Lee et al. 1999; Peterson et al. 1999; Zuhlke et al. 1999). Point mutations in the pre-IQ region that ablate apo-CaM binding to the C-terminus also markedly depress β-dependent targeting of the Ca_V1.2 channel to the cell surface (Wang et al. 2007; Bourdin et al. 2010). We sought to gain a deepened appreciation of the role of the α_1C C-terminus in Ca_V1.2 channel targeting by determining whether the important determinants of β-dependent trafficking were pervasively distributed throughout the C-terminus, or were localized to the pre-IQ/IQ region. Accordingly, we tested the impact of a series of C-terminus truncations on Ca_V1.2 channel trafficking to the plasma membrane. Serial truncations at the α_1C C-terminus led to graded decreases in Ca_V1.2 channel targeting to the cell surface – channels truncated at residues 1906 and 1732 displayed a 60% and 80% reduction in surface density, respectively (Fig. 7D). Longer truncations at residues 1632 and 1540, respectively, virtually eliminated β-dependent membrane targeting (Supplemental Material, Fig. S6). Importantly, QD₆₅₅ fluorescence was normalized for YFP expression under all conditions, explicitly ruling out the potential trivial explanation that differences in trafficking propensity among the distinct α_1C species could be due to variations in protein expression. Both α_1CΔ1906 and α_1CΔ1732 functionally retain Ca²⁺-dependent inactivation (CDI) gating (Erickson et al. 2001; Kim et al. 2004), as might be expected given that the pre-IQ and IQ regions remain intact (Supplemental Material, Fig. S5). The results indicate that critical determinants of β-dependent trafficking of Ca_V1.2 channels are widely distributed throughout the C-terminus, rather than being locally limited to the proximal CaM-binding preIQ/IQ region.

Finally, we examined the impact of combining the α_1C I–II loop AXA1–AXA5 mutations and the C-terminus truncations on Ca_V1.2 channel trafficking to the cell surface (Fig. 7E and F). In both α_1CΔ1906 and α_1CΔ1732, only AXA3 caused a further significant decrease in channel surface density, while the other AXA mutations either caused no change, or even slightly enhanced channel trafficking (Fig. 7E and F). These results offer further evidence that the PEER in the α_1C I–II loop is an important determinant of Ca_V1.2 channel surface density.

Discussion

The fundamental importance of auxiliary β subunits to the functional evolution of Ca_V1/Ca_V2 channels is underscored by the severe phenotypes of β-null mice: β₁– lethal at birth due to asphyxiation (Gregg et al. 1996); β₂– embryonic lethal due to a compromised heartbeat (Ball et al. 2002); and β₄– development of a lethargic epileptic phenotype (Burgess et al. 1997). A prominent function of β subunits is to promote the cell surface trafficking of Ca_V1/Ca_V2 channels. In this work we have re-examined the important molecular determinants and mechanism underlying β-subunit-mediated plasma membrane targeting of Ca_V channels. There are three major new findings: (1) the Ca_V1.2 α_1C subunit I–II loop is a putative ER export rather than retention module; (2) the α₁ C-terminus plays a dual role in channel trafficking; and (3) β-dependent increase in channel surface density is an emergent property that requires multiple α_1C intracellular domains inclusive of the I–II loop and C-terminus. We discuss these aspects of the work in the context of previous results.

Putative ER export versus retention role of the α_1C I–II loop in Ca_V channel trafficking

Auxiliary β subunits bind with nanomolar affinity to the AID in the I–II loop of Ca_V1/Ca_V2 α₁ subunits (Pragnell et al. 1994; Canti et al. 2001; Opatowsky et al. 2003; Butcher et al. 2006; Van Petegem et al. 2008). High-resolution crystal structures show that β subunits interact with the AID using an α₁-binding pocket formed from non-contiguous residues localized in the GK domain (Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004). Point mutations within the AID that disrupt the interaction with β subunits prevent β-induced trafficking of α₁ subunits to the plasma membrane in mammalian cells (Bourdin et al. 2010; Obermair et al. 2010) and Xenopus oocytes (Van Petegem et al. 2008). How does β binding to the I–II loop promote trafficking of α₁ subunits to the cell surface? The prevailing notion, based on experiments carried out in Ca_V2.1 channels, is that the α₁ subunit I–II loop contains an ER retention signal that becomes masked once β binds, thereby permitting forward trafficking (Bichet et al. 2000). The evidence presented in this work suggests that in Ca_V1.2 channels the I–II loop does not act as an ER retention module in the context of the Ca_V channel α₁ subunit. Rather, the opposite situation was discovered – i.e. within the context of the channel, the α_1C I–II loop may have a net ER export function. A cluster of acidic residues situated just downstream of the AID fully accounted for the capability of the I–II loop to enhance chimeric channel surface density, and reprised a similar function within the context of the native Ca_V1.2 channel. It is important to note here that the loss of Ca_V channel trafficking resulting from mutations in the PEER differ fundamentally from previously identified mutations in the I–II loop that also prevent β-induced channel targeting to the cell surface. The previous loss-of-function mutations are within the AID, and prevent channel trafficking by abolishing the α₁−β interaction. By contrast, the AXA3 mutations prevent channel trafficking in a β-independent manner, as uniquely demonstrated in the α_1G[gcggg] chimeric channel context. Di-acidic ER export motifs have been demonstrated in membrane proteins from plant and animal cells (Ma et al. 2001, 2002; Mikosch & Homann, 2009). The basis of their ER export function is believed to be their recognition by components of the coat protein II (COP II) complexes that mediate anterograde ER-to-Golgi vesicular transport (Mikosch & Homann, 2009). We hypothesize that the acidic sequence identified in the α₁ subunit I–II loop may act in such a fashion. This hypothesis will need to be tested in future experiments.

The Ca_V2.1 α₁ subunit I–II loop was proposed as an ER retention module based on its ability to retard membrane targeting of either Shaker K⁺ channels or CD8, when fused to their intracellular C-termini (Bichet et al. 2000). However, in similar experiments, neither the Ca_V1.2 nor Ca_V2.2 α₁ subunit I–II loop displayed ER retention properties, suggesting there may be a genuine difference among distinct Ca_V1/Ca_V2 channel isoforms (Altier et al. 2011). Given that the experiments described in this work have focused exclusively on trafficking determinants in Ca_V1.2 channels, it is worth considering whether the I–II loop may function similarly in other Ca_V1/Ca_V2 channel types. In this regard, it is reassuring that the I–II loops of both Ca_V2.2 (α_1B) and Ca_V2.1 (α_1A) result in a substantial increase in current density when transplanted into to α_1G (Arias et al. 2005; Fan et al. 2010). Moreover, similar to α_1C, the α_1A and α_1B I–II loops contain an acidic residue rich region immediately downstream of the AID (Supplemental Material, Fig. S2). Based on these similarities we speculate that the I–II loop may serve to enhance channel surface density similarly in all Ca_V1/Ca_V2 channels. This prediction will need to be verified experimentally in future experiments.

Role of the α₁ C-terminus in Ca_V channel trafficking

In addition to the I–II loop, there has been accruing evidence that the α_1C C-terminus plays a crucial role in Ca_V channel trafficking. Deletion of residues 1623−1666 in the α_1C C-terminus abolishes β-dependent trafficking of Ca_V1.2 channels to the surface membrane, despite an enduring interaction between the two channel subunits (Gao et al. 2000; Bourdin et al. 2010). The stretch of deleted amino acids encompasses a demonstrated Ca²⁺–CaM binding IQ domain (Erickson et al. 2003; Kim et al. 2004; Van Petegem et al. 2005), leading to suggestions that Ca²⁺–CaM binding to the C-terminus may be necessary for Ca_V1.2 channel trafficking to the membrane (Wang et al. 2007). However, this idea is challenged by the finding that β-dependent Ca_V1.2 channel membrane trafficking remained intact when the α_1C subunit featured a more restricted, but still complete, deletion of the IQ domain (residues 1643–1666) (Bourdin et al. 2010). On the other hand, point mutations in an upstream pre-IQ site that disrupt apo-CaM binding to the α_1C C-terminus also interfere with β-induced membrane targeting of Ca_V1.2 channels (Wang et al. 2007; Bourdin et al. 2010). Hence, apo-CaM, rather than Ca²⁺−CaM, binding to the α₁ C-terminus may be critical for Ca_V channel membrane trafficking. Irrespective of which CaM species is paramount, the question arises as to how CaM molecules can promote β-dependent forward trafficking of Ca_V1.2 channels. It has been previously suggested that the α_1C C-terminus may contain an ER retention signal that is masked upon CaM binding, thereby permitting channel movement to the plasma membrane (Wang et al. 2007). Further, experiments in which α_1C C-terminus fragments were fused to CD4 identified three separate areas in the proximal C-terminus overlapping the EF-hand/pre-IQ/IQ motifs (Supplemental Material, Fig. S5) that exhibited ER retention properties (Altier et al. 2011). Aspects of our results support the idea that the α_1C C-terminus can function as a net ER retention module. Specifically, when swapped for the analogous domain in α_1G, the α_1C C-terminus caused a decrease in the surface density of Ca_V3.1 channels. Moreover, in the doubly substituted chimera α_1G[gcggc], the α_1C C-terminus completely neutralized the strong forward trafficking propensity conferred by the α_1C I–II loop. Nevertheless, the data also indicate a more complex role of the C-terminus in Ca_V1.2 channel trafficking beyond a simple masking of an ER retention sequence by CaM. Truncating α_1C at residue 1732 in the C-terminus results in an 80% decrease in channel trafficking even though both Ca_Vβ and CaM still associate with the channel (Gao et al. 2000; Erickson et al. 2001). This result argues against the simple explanation that CaM merely masks a local ER retention signal, and that this in concert with β binding is sufficient to promote channel trafficking to the membrane. Rather, the data support a more pervasive distribution of determinants on the C-terminus that are important for Ca_V1.2 channel trafficking. It is noteworthy that although α_1CΔ1906 and α_1CΔ1732 target less well than wild-type α_1C to the plasma membrane, they typically give rise to larger whole-cell currents in functional assays (Wei et al. 1994; Hulme et al. 2006). This discrepancy is due to an auto-inhibitory role of the distal C-terminus on channel gating, such that deletion of this region results in a large increase in Ca_V1.2 channel P_o (Hulme et al. 2006).

Interestingly, the C-termini of the distinct Ca_V1 and 2 channels display little sequence homology beyond the residues corresponding to L1732 in Ca_V1.2 (Supplemental Material, Fig. S5). Hence, it is possible that the C-terminus may play divergent roles in β-dependent trafficking among the different channel types. This will be an important and interesting question to address in future studies.

β-dependent trafficking as an emergent property supported by multiple α₁ intracellular domains

A striking result was that the double chimera, α_1G[gcggc], which featured both the α_1C I–II loop and the C-terminus swapped into α_1G, failed to display β-dependent channel targeting to the cell surface. Instead, β-dependent regulation of current density required the presence of at least four intracellular domains of α_1C that included the I–II loop and C-terminus. Therefore, the α_1C I–II loop and C-terminus are necessary but not sufficient to reconstitute β-dependent channel trafficking to the membrane. A limitation of our study is that the magnitude of β-dependent up-regulation of current density reconstituted in the chimeric channels (∼three-fold increase) is less than the 10- to 15-fold increase typically observed with wild-type Ca_V1.2. A likely contributing factor to this discrepancy is that β subunits increase the single-channel P_o of Ca_V1.2 channels (ranging from 2- to 8-fold in different studies) in addition to promoting channel trafficking to the plasma membrane (Kanevsky & Dascal, 2006). The increase in P_o relies on a β-dependent formation of a rigid helix spanning the AID and domain IS6 (Vitko et al. 2008; Findeisen & Minor, 2009). Our results do not support a similar β-dependent increase in single channel P_o in the chimeric channels since we do not observe a β-dependent enhancement of current in the majority of chimeras that contain the α_1C I–II loop. Hence, the magnitude of β-dependent increase in current density may be quite comparable between the relevant chimeras and Ca_V1.2 if the enhanced single-channel P_o effect on the latter is taken into account.

Overall, our results suggest the following hypothesized model for β-dependent regulation of Ca_V channel trafficking (Fig. 8). The α_1C I–II loop contains a putative ER export signal while the other intracellular loops and termini have net ER retention characteristics. In the absence of β, the intracellular domains are configured such that the multiple ER retention signals are exposed and functionally dominant, leading to channels being retained in the ER. Upon β-binding to the α_1C I–II loop, we propose that a conformational rearrangement of the intracellular domains occurs that diminishes the strength of ER retention signals relative to I–II loop export signals, leading to channel transport to the cell surface (Fig. 8). It is interesting to contemplate the possible nature of the proposed β-induced conformational change in intracellular domains. Our results do not support a model where the β subunit itself physically masks spatially distinct ER retention signals on the channel. If this were the case then it would be expected that β subunits would enhance trafficking in double and triple chimeras that included the I–II loop. However, this was not observed. Rather, the data invite speculation that β-binding to the I–II loop initiates a concerted motion of the intracellular segments that is coordinated by the C-terminus. CaM may participate in this process by promoting a permissive ternary conformation of the C-terminus. This model intimates that the α_1C intracellular domains are not independent entities, but rather engage in intra-molecular interactions among themselves. In this regard, it is noteworthy that in Ca_V2.1 channels the I–II loop has been demonstrated to interact with the N- and C-termini, and the III–IV loop (Restituito et al. 2000; Geib et al. 2002). The exact details of the β-induced rearrangement of α_1C subunit intracellular segments that we propose will need to be explored in future experiments, including structural studies.

Figure 8. Conceptual model of mechanism underlying β-mediated trafficking of Ca_V1.2 channels.

In the absence of β, an ER export signal present on the α₁ subunit I–II loop is functionally overcome by discrete ER retention signals present in the other intracellular domains, leading to channels being retained in the ER. Upon β-binding to the α_1C I–II loop, a C-terminus-dependent conformational rearrangement of the intracellular domains occurs that diminishes the strength of ER retention signals relative to I–II loop export signals, leading to channel transport to the cell surface.

Recently, two groups independently demonstrated that Ca_Vβ binding to either Ca_V1.2 (Altier et al. 2011) or Ca_V2.2 (Waithe et al. 2011) protects the respective α₁ subunit from ubiquitination and subsequent proteasomal degradation. In one study, blocking the proteasomal degradation pathway with MG132 was sufficient to rescue surface expression of Ca_V1.2 channels even in the absence of β (Altier et al. 2011). By contrast, MG132 did not lead to increased surface expression of Ca_V2.2 channels in the absence of Ca_Vβ (Waithe et al. 2011). Possibly, the β-dependent rearrangement of α_1C subunit intracellular segments we envision is necessary to prevent channel ubiquitination and targeting to the proteasome. Further work is clearly needed to reconcile these new results and to develop a more detailed mechanistic model for β regulation of Ca_V1/Ca_V2 channel trafficking.

Glossary

Abbreviations

AID

α interaction domain

BBS

bungarotoxin binding site

Ca_V channel

voltage-dependent calcium channel

Ca_Vβ

voltage-dependent calcium channel auxiliary β subunit

CaM

calmodulin

CCD

charge-coupled device

CFP

cyan fluorescent protein

C-terminus

carboxyl terminus

ER

endoplasmic reticulum

GK

guanylate kinase

I_Ca

calcium channel current

PEER

putative endoplasmic reticulum export region

QD

quantum dot

Q_max

maximal gating charge

YFP

yellow fluorescent protein

Author contributions

K.F. designed and conducted experiments in all areas of the work, analysed results, interpreted data, and helped write the paper. H.M.C. conceived of and designed experiments, analysed results, interpreted data, and wrote the paper. Both authors approved the final version of the manuscript.

Supplementary material

Figure S1

Figure S2

Figure S3

Figure S4

Figure S5

Figure S6

Table S1

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors