Duplex signaling by CaM and Stac3 enhances Ca_V1.1 function and provides insights into congenital myopathy

Central to excitation–contraction (EC) coupling in skeletal muscle, Ca_V1.1 is an L-type voltage-gated calcium (Ca²⁺) channel that senses transmembrane depolarization to initiate Ca²⁺ release from the SR RYR1 (Schneider and Chandler, 1973; Bannister and Beam, 2013). Although its cardiac counterpart Ca_V1.2 communicates with RYR2 via freely diffusing Ca²⁺ ions, Ca_V1.1 is conformationally coupled to RYR1, obviating the intermediary second messenger (Armstrong et al., 1972; Tanabe et al., 1990a; Ríos et al., 1992). This intimate physical linkage warrants a precise geometric arrangement of the two partners in the skeletal myotube: four Ca_V1.1s, termed tetrads, are disposed in ordered arrays that parallel RYR1 arrays at the surface–membrane/SR (peripheral-couplings) or tubular-membrane/SR (triad) interfaces (Franzini-Armstrong and Jorgensen, 1994; Lamb, 2000).

Fitting with this physiology, a cohort of auxiliary subunits such as β_1A (Schredelseker et al., 2009), α₂δ (Obermair et al., 2005), γ₁ (Freise et al., 2000), and various SR proteins including RYR1 (Nakai et al., 1996; Avila and Dirksen, 2000; Bannister et al., 2016), JP45 (Anderson et al., 2006), and junctophilin (Golini et al., 2011) tune Ca_V1.1 function. To identify essential signaling partners, a top-down approach using primary cultures of skeletal myotubes obtained from gene knockout (KO) models (Obermair et al., 2008) and cell lines derived from dysgenic and normal myotubes have been insightful (Powell et al., 1996). However, such analyses have often revealed overlapping functions whereby loss of a single protein dramatically alters Ca_V1.1 localization and/or gating to ultimately disrupt EC coupling. These effects may be either direct or indirect depending on other proteins present in the complex. Thus, quantifying the role of a given modulator on Ca_V1.1 and the underlying regulatory mechanism is challenging. Intriguingly, recent studies have revealed that both calmodulin (CaM; Ohrtman et al., 2008; Stroffekova, 2008) and stac3 regulate Ca_V1.1, although underlying mechanisms remain to be fully elucidated (Horstick et al., 2013; Polster et al., 2015; Linsley et al., 2017a).

The Ca²⁺-binding protein CaM has emerged as a dynamic regulator of neuronal and cardiac Ca²⁺ channels (Ca_V1.2/3/4 and Ca_V2.1/2/3; Halling et al., 2006; Minor and Findeisen, 2010; Ben-Johny et al., 2015). The binding of Ca²⁺-free CaM (apoCaM) up-regulates the baseline open probability (P_O), whereas Ca²⁺–CaM interaction relieves this initial enhancement manifesting as Ca²⁺-dependent inactivation (CDI; Adams et al., 2014). For Ca_V1.1, however, CaM regulation has evaded consensus. Exogenously expressed CaM localizes to the skeletal muscle triad (Rodney and Schneider, 2003). CaM interaction with Ca_V1.1 has been controversial in biochemical studies, however, with some reporting weak to no binding (Ohrtman et al., 2008), whereas in vitro surface plasmon resonance measurements and crystallographic analysis suggest a high-affinity interaction with the channel carboxy tail (CT) in the presence of Ca²⁺ (Sencer et al., 2001; Black et al., 2005; Halling et al., 2009). Similarly, functional analysis of Ca_V1.1 in skeletal myotubes has revealed the ultra-slow and variable extent of CDI, casting doubt as to whether CaM is relevant for Ca_V1.1 function (Tanabe et al., 1990b; Ohrtman et al., 2008; Stroffekova, 2008). Interestingly, mutations of the CaM binding interface in the Ca_V1.1 CT strongly reduce EC coupling (Stroffekova, 2011).

Likewise, stac3 was recently identified as a component of the EC coupling machinery in association with debilitating congenital human myopathies (Stamm et al., 2008; Horstick et al., 2013; Nelson et al., 2013; Grzybowski et al., 2017). This autosomal-recessive disease was identified in a culturally isolated population of Native Americans (Stamm et al., 2008) but has since been observed in Middle Eastern, African, and South American individuals (Grzybowski et al., 2017; Telegrafi et al., 2017). Patients present with symptoms of muscle weakness, including short stature, kyphoscoliosis, talipes deformities, and drooping facial features, and increased susceptibility to malignant hyperthermia (Stamm et al., 2008; Grzybowski et al., 2017; Telegrafi et al., 2017). Functionally, homozygous KO of stac3 in mouse and zebrafish models led to markedly diminished Ca_V1.1 surface–membrane trafficking, reduced tetrad formation, loss of retrograde signaling, and a near-complete loss of EC coupling (Horstick et al., 2013; Nelson et al., 2013; Polster et al., 2015, 2016; Linsley et al., 2017a,b). However, overexpression of a myopathy-associated mutant stac3 partially rescued channel trafficking, although EC coupling remained reduced (Polster et al., 2016; Linsley et al., 2017a). Moreover, the structural determinants of Ca_V1.1 that mediate stac binding also remain unknown (Campiglio and Flucher, 2017). Thus, stac3 may elicit multiple regulatory functions of Ca_V1.1 through direct interactions with the channel or mediated by other triadic proteins (Polster et al., 2016; Linsley et al., 2017a).

To resolve these complex channel-regulatory mechanisms, a bottom-up approach whereby the effects of individual signaling molecules on Ca_V1.1 gating and trafficking are probed in a simplified system without an elaborate SR or t-tubules would be greatly beneficial (Dascal et al., 1992; Polster et al., 2015; Perni et al., 2017). However, functional analysis of Ca_V1.1 and its modulation by various signaling molecules in nonmuscle cell systems remains challenging (Perez-Reyes et al., 1989; Dascal et al., 1992; Johnson et al., 1997; Polster et al., 2015). Although homologous Ca_V1.2, Ca_V1.3, and Ca_V1.4 all exhibit reliable surface–membrane trafficking in heterologous systems in the presence of α₂δ and β auxiliary subunits (Mikami et al., 1989; Catterall, 2000; Xu and Lipscombe, 2001; McRory et al., 2004), Ca_V1.1 is thought to be retained in internal organelles (Polster et al., 2015; Linsley et al., 2017b). Countering this classical purview, however, a recent functional study demonstrated that the cytosolic adapter protein stac3 with the α₂δ/β subunits enabled Ca_V1.1 expression in human-derived tsA201 cells (Polster et al., 2015). Further analysis suggested that additional factors including the transmembrane γ₁ subunit may also permit Ca_V1.1 expression in tsA201 cells (Polster et al., 2016). The contrasting molecular requirements that permit Ca_V1.1 expression in heterologous systems obfuscate general principles that underlie channel trafficking and preclude systematic analysis of channel gating.

In this study, using a combination of whole-cell electrophysiology, FRET two-hybrid binding assay, and external-epitope labeling with flow cytometry (Yang et al., 2010), we demonstrate that Ca_V1.1, in fact, traffics to the plasma membrane of recombinant cell systems in the presence of auxiliary α₂δ and β subunits alone. However, this baseline expression is lower than that for homologous L-type channels. Moreover, electrophysiological analysis reveals tiny ionic currents, suggesting that Ca_V1.1 has a low baseline P_O. Both CaM and stac3 enhance both surface–membrane trafficking and baseline P_O of Ca_V1.1. Moreover, we demonstrate that stac3 binds to the CT of Ca_V1.1, and stac3 mutations associated with congenital myopathy weaken this interaction, resulting in reduced channel surface–membrane trafficking. Delivery of CaM to the channel complex can partially reverse this trafficking defect. Interestingly, long-term application of small-molecule Ca_V modulators diltiazem and verapamil yields a partial rescue of channel trafficking. These results highlight the utility of reconstituted Ca_V1.1 in HEK293 cells as a simplified platform to distinguish regulatory effects of individual triadic signaling molecules. Additionally, the flow cytometric analysis of plasmalemmal expression may be an attractive venue for high-throughput screens of small molecules that modulate Ca_V trafficking. In all, our findings illustrate parallel signaling mechanisms that tune Ca_V1.1 trafficking and gating and shed light on pathophysiological mechanisms for stac3-associated congenital myopathies.

Ca_V1.3_S was unmodified from previously published rat Ca_V1.3_Δ (GenBank Accession No. AF370009.1; Liu et al., 2010). GFP-Ca_V1.1 was a gift from Kurt Beam (University of Colorado at Denver, Denver, CO). Stac3 human isoform 2 was purchased from Origene. RYR1 P2 domain was synthesized by Genscript (sequence in Table S1). Ca_V1.1 CT chimeras were generated by first using PCR amplification with primers P01 and P02 (primers listed in Table S1) and restriction enzyme cutting sites XhoI and KpnI to generate a silent mutation to create a unique XbaI site ∼1–2 aa upstream of the EF hand and add an MssI restriction enzyme cutting site. Ca_V1.3 CI region was added to this construct by PCR amplification (P03 and P04) and inserted via XbaI and MssI restriction enzyme cutting sites. The Ca_V1.1 variant was generated by PCR amplification (P05 and P06) and cutting sites BglII–KpnI to insert an XbaI cutting site in place of the stop codon. Then, glycine-(12)-CaM_WT was PCR amplified (P07 and P08) and inserted into stopless Ca_V1.1 with XbaI and KpnI. Ca_V1.1_ΔCT was ordered from Genscript with the CT truncated after residue 1,397 (i.e., SILGPH*) and inserted with XhoI and KpnI. Ca_V1.1 (BBS) was generated by overlap PCR (P09–P12) and restriction enzyme sites SalI–XhoI to insert BBS. β_2A-glycine-(8)-CaM_WT was unchanged from previously published rat β_2A modifications (Yang et al., 2014). Using PCR amplification, we cloned CaM₁₂₃₄ (P08 and P13) into NotI–BsrGI to generate β_2A-glycine-(8)-CaM₁₂₃₄. β_2A-glycine-(12)-RYR1 P2 was generated from PCR amplification (P14 and P15) and inserted into β_2A-glycine-(32)-CaM_WT from a previously published construct (Sang et al., 2016) with BsrGI and compatible ends NheI–XbaI. C1 of stac3 was PCR amplified (P16 and P17) and cloned into pcDNA3 with NheI and BsrGI. Native American myopathy mutation was generated by QuikChange mutagenesis (P18 and P19). Venus- and Cerulean-tagged constructs were generated by PCR amplification (P20–P23) and inserted via NotI and XbaI restriction enzyme cutting sites into previously published constructs (Sang et al., 2016). All constructs were verified with DNA sequencing.

For whole-cell electrophysiology, HEK293 cells were cultured on glass coverslips in 10-cm dishes and transfected using a calcium phosphate method (Peterson et al., 1999) with the following DNA combinations: 8 µg α₁ subunit of Ca²⁺ channel, 8 µg rat β_2A (GenBank Accession No. M80545; Perez-Reyes et al., 1992) or β_1A from mouse (NP112450.1), and 8 µg rat α₂δ (NM012919.2; Tomlinson et al., 1993). 3 µg SV40 T antigen was also cotransfected to enhance expression, and 8 µg CaM variants, stac3 variants, and RYR1 P2 variants were transfected for overexpression of trafficking agent. Similarly, for bungarotoxin labeling, HEK293 cells were cultured in 60-mm dishes and transfected by calcium phosphate precipitation. DNA concentration used was half that for electrophysiology conditions.

For FRET two-hybrid experiments, HEK293 cells were cultured on glass-bottom dishes and transfected using a standard polyethylenimine protocol (Lambert et al., 1996). Epifluorescence was collected 1–2 d after transfection.

For the drug study, drugs were purchased from Sigma-Aldrich. Nifedipine and diltiazem were diluted to 1 mM in DMSO, and verapamil, ranolazine, and mexiletine were diluted to 10 mM in DMSO before being added to cell culture media. Cells were incubated in the respective concentration of drugs for 24 h before bungarotoxin labeling.

Whole-cell electrophysiology was performed at room temperature 1–4 d after transfection with Axopatch 200A (Axon Instruments). Glass pipettes were made from borosilicate glass (BF150-86-10; Sutter Instrument) at 1–3 MΩ resistance with a horizontal puller (P-97; Sutter Instrument) and fire polisher (microforge; Narishige). We low-pass filtered recordings at 2 kHz, sampled at 10 kHz, and used P/8 leak subtraction with 70% series resistance and capacitance compensation. Internal solution contained (in mM): CsMeSO₃ 114, CsCl₂ 5, MgCl₂ 1, MgATP 4, HEPES 10, and 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid 10, adjusted to 295 mOsm with CsMeSO3 and pH 7.4 with CsOH. External solution contained (in mM): TEA-MeSO3 140, HEPES 10, and CaCl2 40, adjusted to 300 mOsm with TEA-MeSO₃ and pH 7.4 with TEA-OH. For measuring charge movements, we added 0.2 mM LaCl₃ and 1.0 mM CdCl₂ to the external solution. We used a holding potential of −80 mV, family of test pulses from −30 mV to +80 mV in 10-mV increments, and repetition interval of 20 s for all whole-cell recordings. Custom MATLAB (MathWorks) software was used to determine peak current, and mean peak current densities are plotted with SEM.

Peak current density–voltage curves were fitted with the following equation:

where J_peak is the peak current density at test potential V, G_max is maximal channel conductance, V_rev is the reversal potential, V_1/2 is the half-activation voltage, and k_G is the slope factor (Table S2).

Normalized gating charge–voltage curves were fitted with the following equation:

where Q_norm is the gating charge movement at voltage V normalized to value at +80 mV. Gating charge movement is composed of a double Boltzmann relation, with Q_max as saturating normalized gating charge; V_1/2.a and V_1/2.b are half-activating potentials for the two components; and SF_a and SF_b are slope factors for the two components.

Three-cube FRET fluorescence of transfected HEK293 cells was measured on an inverted fluorescence microscope in 2 mM Ca²⁺ Tyrode’s under resting Ca²⁺ intracellular concentrations and 10 mM Ca²⁺ Tyrode’s incubated with 4 µM ionomycin (Sigma-Aldrich) under Ca²⁺/CaM conditions. Different concentrations and ratios of DNA were transfected to achieve a range of donor molecule (D_free) concentrations. FRET efficiency (E_A) for each individual cell was calculated (Erickson et al., 2001), and effective dissociation constants (K_d,EFF) were computed by fitting the binding curve E_A = [D_free]/(K_d,EFF + [D_free]) · E_A,max iteratively. For stac3 Native American myopathy constructs where plateaus of FRET binding curves were not clearly defined by data, we assumed that stac3 adopts the same conformation and possesses the same E_A,max (Ben Johny et al., 2013).

First, we washed transfected cells twice with DPBS (with Mg²⁺ and Ca²⁺; MediaTech). Then, we blocked nonspecific binding sites with 3% BSA/DMEM for 30 min at room temperature. We incubated cells with 1 µM α-bungarotoxin-biotin (Invitrogen) in 3% BSA/DMEM for 1 h at room temperature in the dark. On ice and in the dark, cells were washed twice with DPBS, incubated three times for 5 min with DPBS, and incubated for 1 h with 10 nM Qdot655 for flow cytometry or Qdot605 for confocal imaging (Invitrogen) in 3% BSA/DMEM. Finally, cells were washed with DPBS and imaged on the confocal microscope in 2 mM Ca²⁺ Tyrode’s or harvested with trypsin, washed with PBS (without Mg²⁺ and Ca²⁺), and resuspended for flow cytometry.

The total GFP fluorescence is proportional to the number of channels in a cell,

where α_G corresponds with the brightness of single GFP given the imaging setup, and I₀ is the intensity of the excitation lamp. Similarly, the number of channels at the plasma membrane is given by

where α_R corresponds with the brightness of a single quantum dot (QD) molecule when assessed through our imaging setup, I₀ is the intensity of the excitation lamp, and 4 corresponds with the stoichiometry for biotin–streptavidin interaction. The factor ε is the efficiency of QD labeling. The ratio of the two equations yields Eq. 5 and is proportional to the fraction of surface–membrane channels.

We captured exemplar images of bungarotoxin-labeled HEK cells with the Olympus FluoView FV300 confocal laser scanning microscope and FluoView software (Olympus). Using the Olympus Plan Apochromat 403 or 603 oil objective (NA 1.40, PLAPO60XO3; Olympus), GFP was excited with an argon laser (488 nm), and Qdot-605 streptavidin conjugate (Invitrogen) was excited with a helium neon (HeNe) green laser. Olympus optical filters used include 442/515-nm excitation splitter (FV-FCV), 570-nm emission splitter (FV-570CH), BA510 immunofluorescence and BA530RIF for GFP emission channel, and 605BP filter for Qdot channel. Images were converted and merged in ImageJ (National Institutes of Health).

Fluorescence of harvested cells was measured with an Attune acoustic focusing flow cytometer (Life Technologies) in high-sensitivity mode with a flow rate of 100 µl/min. We used the blue (488 nm) laser to excite GFP and Qdot to collect green and red fluorescence, respectively. Green fluorescence was measured through the 574/26 optical filter. Likewise, red fluorescence was measured through the 640LP optical filter. Flow cytometer was calibrated and maintained as previously published (Lee et al., 2016). Control experiments included untransfected cells, GFP-only cells, cells transfected with Ca_V1.1 and stac3 as a negative control, and cells with CaV1.1_BBS and stac3 as a positive control. Data were exported as FCS files and analyzed with custom MATLAB software.

Raw data were gated by forward- and side-scatter signals to filter for single and healthy cells, and green signals >1.5 × 10⁵ units were excluded because of nonlinearities in flow cytometer measurements (Lee et al., 2016). Red signals >2 × 10⁵ units were excluded because of PMT saturation and accounted for <1% of total collected points. To correct for the true green (S_G) and red (S_R) signals, we averaged red signal (Ŝ_R,blank) and green signal (Ŝ_G,blank) of blank cells. We also calculated the slope for GFP bleed-through into the red channel (f_RED,GFP) to be ∼2.65% because of the broadness of the GFP emission spectrum, yielding two equations: S_G = Ŝ_G − Ŝ_G,blank and Ŝ_R = S_R − Ŝ_R,blank − f_RED,GFP · S_G, where Ŝ_G is the raw green signal and S_RED is the red signal. Welch’s t test was used to statistically compare two trafficking conditions, and P values report the probability for the null hypothesis that the respective ϕ_max for conditions compared are equal. To ensure robustness, we also used a rank-sum test. In all cases, the P value for rank-sum test was similar to that with Welch’s t test. The number of independent trials and total number of cells analyzed are listed in Table S3.

The supplemental text explicitly derives the Langmuir relationship between plasmalemmal trafficking and affinity of binding for stac3 pertaining to Fig. 7 l. Fig. S1 shows individual peak current densities from individual cells pertaining to Fig. 1. Fig. S2 shows additional trafficking data and electrophysiological parameters for Ca_V1.1 channels in the presence of β_2A subunit alone Fig. 6 a. Fig. S3 shows additional justification for distinct interfaces for CaM versus stac3 within Ca_V1.1. Fig. S4 shows extended data with the extent of enhancement in trafficking of Ca_V1.1 in the presence of pharmacological chaperones. Table S1 lists primers used. Table S2 summarizes electrophysiological parameters pertaining to Fig. 1. Table S3 provides supplemental information for trafficking experiments including sample size.

In comparison with other L-type Ca channels, Ca_V1.1 expresses poorly in heterologous cell systems (Perez-Reyes et al., 1989; Polster et al., 2015). Fig. 1 a shows an exemplar inward Ca²⁺ current elicited in response to a voltage-step depolarization from a HEK293 cell transiently expressing Ca_V1.3 pore-forming α₁ subunit with auxiliary β_2A and α₂δ subunits. Population data of mean peak current densities elicited in response to a family of step depolarizations further illustrate robust expression of Ca_V1.3 in HEK293 cells (Fig. 1 a). In contrast, when Ca_V1.1 α₁ subunit is coexpressed with both β_2A and α₂δ auxiliary subunits, we observe minimal ionic currents (Figs. 1 b and S1). Given the functional difference between Ca_V1.1 and Ca_V1.3 despite their overall structural similarity, we sought to identify requirements for functional expression of Ca_V1.1 in heterologous systems.

First, we reasoned that the Ca_V channel cytoplasmic domains may contain critical motifs that differentially enhance channel function (Fang and Colecraft, 2011). In this regard, for related Na_V1.9 sodium channels that also fail to express in nonexcitable cells, a chimeric approach that replaced the CT of Na_V1.9 with that from Na_V1.4 yielded robust currents (Goral et al., 2015). Paralleling this approach, we exchanged the CT of Ca_V1.1 α₁ subunit with that of Ca_V1.3. Electrophysiological analysis revealed robust currents for the chimeric channels (Fig. 1 c), suggesting that the CT is a key determinant for functional expression.

Second, key Ca_V1.1-interacting proteins may modulate channel function by either serving as chaperones to promote plasmalemmal trafficking or enhancing channel activity. Indeed, recent research shows that Ca_V1.1 currents can be reestablished in HEK293 cell systems by coexpression of stac3, an adapter protein essential for skeletal muscle function (Polster et al., 2015). Exemplar current trace and population data of Ca_V1.1 after coexpression of stac3 in HEK293 cells further confirm these findings (Fig. 1 d). As CaM is a canonical interacting partner for the CT of various Ca_Vs, we reasoned that CaM might also permit functional expression of Ca_V1.1. Indeed, overexpression of CaM alone (Fig. 1 e) or localization of CaM to the Ca_V1.1 complex via fusion to β_2A subunit (Fig. 1 f) reveals a marked enhancement in Ca²⁺ currents. Thus, multiple seemingly disparate manipulations permit Ca_V1.1 expression in nonexcitable cells.

Thus informed, we sought to dissect molecular mechanisms that enable Ca_V1.1 function in HEK293 cells. The functional expression of ion channels may be enhanced from changes in three vital parameters: (1) the number of channels at the surface membrane dictated by protein trafficking, (2) ion permeation, and (3) channel gating.

To quantify the relative fraction of channels at the cell surface membrane, we used a dual-labeling approach (Yang et al., 2010) whereby the α₁ subunit is tagged with both a GFP on the cytoplasmic amino terminus and an external epitope composed of a 13-aa α-bungarotoxin–binding site (BBS) inserted into the extracellular loop between transmembrane segments 5 and 6 (S5 and S6) of domain II (Fig. 2 a, Ca_V1.1_BBS). To label surface membrane channels, we incubated cells with cell-impermeable biotin-conjugated α-bungarotoxin and visualized using streptavidin covalently attached to a red QD; the total expression of Ca_V1.1 in a cell is determined by monitoring the GFP fluorescence (Sekine-Aizawa and Huganir, 2004). The high affinity and specificity of bungarotoxin for the BBS site facilitates reliable detection of surface–membrane Ca_V1.1 with minimal background fluorescence (Sekine-Aizawa and Huganir, 2004). We first verified the functionality of Ca_V1.1_BBS by cotransfecting β_2a-CaM_WT into HEK cells. The resultant Ca²⁺ currents exhibited properties comparable with those of unmodified Ca_V1.1 (Fig. 2 a). We probed baseline plasmalemmal expression for Ca_V1.1_BBS in the presence of β_2A and α₂δ subunits using confocal imaging (Fig. 2 b). The left subpanel shows the transmitted light image of an exemplar cell, and the middle subpanels show green (S_G) and red (S_R) fluorescence images indicating GFP from total channels and QD emissions from extracellular channels, respectively. The far-right merged image showcases the difference in intracellular and extracellular labeling of Ca_V1.1_BBS. Although strong GFP fluorescence is evident, external QD labeling is sparse, indicating poor surface–membrane expression of Ca_V1.1 (Fig. 2 b). That said, we did observe some surface–membrane labeling in a few cells, suggesting that Ca_V1.1 with just α₂δ and β subunits might be sufficient for surface–membrane trafficking, albeit with a lower efficacy. Analysis of external epitope labeling from a multitude of individual cells would help resolve such ambiguities.

Accordingly, we used flow cytometric analysis to quantify surface–membrane trafficking at the population level by determining the total GFP (S_G) and QD (S_R) fluorescence from individual cells. For a given cell, the ratio of red to green fluorescence (ϕ = S_R/S_G) is proportional to the fraction of surface–membrane channels (f_mem) and serves as a quantifiable metric for trafficking efficacy:

The factors α_R and α_G are brightness of single QD and GFP fluorophores, respectively, and ε is the efficiency of labeling. Given this framework, we plotted S_R versus S_G obtained from individual cells expressing Ca_V1.1_BBS with β_2A and α₂δ subunits after 1 d of transfection (Fig. 2 c). Consistent with confocal imaging data, flow-cytometric analysis showed a mixed population of cells: one fraction exhibited minimal surface–membrane labeling (S_R = 0), and the other demonstrated reliable QD labeling (S_R > 0). Binned data reveal a saturating relationship for S_R as S_G increases, with a maximal value of ∼1,923 ± 51 fluorescence units. We estimated the saturating surface–membrane trafficking limit (ϕ_max) as the mean ratio ϕ for individual cells exhibiting high GFP fluorescence (i.e., 5.4 × 10⁴ ≤ S_G ≤ 1.4 × 10⁵) to be 0.0225 ± 0.0006 (Fig. 2 f). We excluded values >1.4 × 10⁵ GFP fluorescence units because of previously identified nonlinearities in fluorescence measurements from our flow cytometer above this value (Lee et al., 2016). In comparison, ϕ_max for Ca_V1.2_BBS is ∼0.0276 ± 0.0004 (Fig. 2 f). Similarly, coexpression of Ca_V1.1_BBS with β_1A subunit that is endogenous to skeletal myotubes also resulted in weak but detectible QD labeling based on confocal imaging (Fig. 2 d) and population analysis (Fig. 2, e and f). In contrast, Ca_V1.1_BBS exhibited minimal plasmalemmal expression in the absence of β subunits (Fig. 2 f). These findings further demonstrate that β and α₂δ subunits are sufficient for plasmalemmal trafficking of Ca_V1.1 in nonexcitable cells; however, this baseline trafficking efficacy is diminished in comparison with related Ca_V channels.

With baseline plasmalemmal expression levels established, we probed the effect of CaM and stac3 on overall Ca_V1.1 plasmalemmal trafficking. When CaM is delivered locally to Ca_V1.1_BBS via β_2A-CaM, QD labeling is markedly enhanced, suggesting improved plasmalemmal localization (Fig. 3 a). Flow cytometric analysis of Ca_V1.1 coexpressed with β_2A-CaM revealed an overall enhancement in the QD labeling (Fig. 3, b and e) in comparison with levels with the β_2A subunit alone (P < 10⁻⁵; Figs. 2 c and 3 e) or with β_2A fused to a sham payload, the P2 domain of RYR1 (P < 10⁻⁵; Fig. 3 e). Likewise, the surface membrane expression of Ca_V1.1 bound to the skeletal muscle β_1A subunit was also enhanced significantly upon coexpression of CaM as a separate molecule (P < 10⁻⁵; Fig. 3 f).

We next explored whether stac3, like CaM, enhances plasmalemmal trafficking of Ca_V1.1. Confocal imaging (Fig. 3 c) and flow cytometric analysis (Fig. 3 d) revealed substantial enhancement in QD labeling for Ca_V1.1_BBS with stac3 in the presence of β_2A (P < 10⁻⁵), consistent with improved channel trafficking. Similarly, stac3 enhanced Ca_V1.1 trafficking in the presence of β_1A subunit (P < 10⁻⁵; Fig. 3 f). In contrast, coexpression of mutant stac3 containing only the C1 domain only partially enhanced surface–membrane trafficking (P < 10⁻⁵; Fig. 3 f). Together, these results demonstrate that both CaM and stac3 enhance plasmalemmal trafficking of Ca_V1.1.

We next sought to identify key channel elements that mediate CaM- and stac3-dependent enhancement in Ca_V1.1 trafficking. As the carboxy terminus is critical for Ca_V1.1 functional expression, we tested the binding of CaM and stac3 to this channel domain (Fig. 4 a). Indeed, for nearly all Ca_V1/2 channels, CaM is a well-established partner for the CT known to modulate channel function (Peterson et al., 1999; Qin et al., 1999; Zühlke et al., 1999; Lee et al., 2000; Pitt et al., 2001; Liang et al., 2003; Singh et al., 2006; Yang et al., 2006). Consequently, we used a FRET two-hybrid binding assay (Erickson et al., 2001) in live cells to quantify CaM binding. We coexpressed cerulean-tagged CaM (Cer-CaM_WT) with Venus-tagged Ca_V1.3 CT, including the dual vestigial EF hands and the pre-IQ and IQ domains (Ven-Ca_V1.3 Ca²⁺ inactivation [CI]), and measured FRET efficiency (E_A) between the donor–acceptor pairs (Fig. 4 b). Strong binding of Cer-CaM_WT to Ven-Ca_V1.3 CI was observed under both basal and elevated Ca²⁺ conditions (Fig. 4 b), consistent with prior research (Ben Johny et al., 2013). In contrast, FRET two-hybrid analysis of Venus-tagged Ca_V1.1 CI (Ven-Ca_V1.1 CI) and Cer-CaM_WT showed weak binding under both basal and elevated Ca²⁺ conditions (Fig. 4 c). This weak affinity is consistent with a significant fraction of Ca_V1.1 lacking prebound CaM in endogenous conditions in HEK293 cells. In like manner, FRET two-hybrid analysis of Cer-tagged stac3 with Ven-Ca_V1.1 CI also revealed strong binding (Fig. 4 d), with K_d,EFF ∼12,000 D_free units ∼400 nM. Fig. 4 e compares relative binding affinities for both CaM and stac3 with Ca_V CT obtained from FRET two-hybrid experiments. Altogether, these findings raise the possibility that the binding of CaM or stac3 to the Ca_V1.1 CT may be critical for its function.

Accordingly, we reasoned that the deletion of the CT would abrogate stac3- and CaM-mediated enhancement in Ca_V1.1 trafficking, a prediction that could be assessed readily using the flow cytometric assay. With β_1A and α₂δ coexpressed, Ca_V1.1(ΔCT)_BBS with a truncated CT showed significant enhancement in trafficking in comparison with the WT channels (P < 10⁻⁵; Fig. 4 f). However, coexpression of either β_2A-CaM (P = 0.4) or stac3 (P = 0.11) did not further enhance surface–membrane trafficking of Ca_V1.1 (Fig. 4 f). These results suggest that CaM and stac3 binding to the CT is functionally critical to enhance plasmalemmal trafficking. Moreover, FRET two-hybrid experiments in Fig. 4 c revealed that CaM binding affinity to the Ca_V1.1 CT was substantially enhanced in the presence of Ca²⁺. Functionally, this difference in affinity would suggest that abrogating Ca²⁺-binding to CaM would diminish CaM-dependent enhancement in Ca_V1.1 plasmalemmal trafficking. Indeed, coexpression of Ca_V1.1_BBS with β_2A fused to a mutant CaM lacking Ca²⁺ binding (β_2A-CaM₁₂₃₄) resulted in minimal enhancement in QD labeling (Fig. 4 g; P < 10⁻⁵ for Ca_V1.1 with β_2A-CaM₁₂₃₄ compared with β_2A-CaM_WT). In all, these results suggest that the occupancy of Ca_V1.1 CT is closely linked to proper channel function, and an emerging repertoire of CT-binding proteins may modify Ca_V function via parallel mechanisms (Park et al., 2010; Wang et al., 2010; Marshall et al., 2011; Flynn and Altier, 2013; Hall et al., 2013).

Given that both CaM and stac3 bind to the channel CT to enhance surface–membrane expression, we examined whether these agents act independently or through a shared endpoint. Consequently, to further delineate the CT binding interface for stac3 and CaM, we parsed the CT into three distinct segments: dual vestigial EF hands and pre-IQ and IQ domains. Using FRET two-hybrid assay, we probed binding between Venus-tagged channel segments and cerulean-tagged CaM or stac3. Ca²⁺/CaM exhibits a markedly higher affinity to the pre-IQ and IQ domains in comparison with the dual vestigial EF hand segments (Fig. 5, a and b). In contrast, stac3 preferentially binds to the dual vestigial EF hand segments in comparison with the pre-IQ and IQ domains (Fig. 5, c and d). Importantly, these findings are in contrast with a recent study that suggested direct IQ binding based on reduced colocalization of Ca_V1.1 and stac3 after mutations in the IQ domain (Campiglio et al., 2018). In light of our present findings, it is possible that mutations in the IQ may indirectly alter stac3 interaction with upstream elements. In all, these findings demonstrate that stac3 and CaM prefer distinct CT interfaces.

Thus, we probed surface–membrane labeling of Ca_V1.1_BBS in the presence of both β_2A-CaM and stac3. If the two agents act through a shared endpoint, then their combination will not further increase trafficking. However, flow cytometric analysis revealed that the two agents combinatorially enhance the trafficking of Ca_V1.1_BBS nearly sixfold, suggesting that they act independently through distinct sites on the CT (P < 10⁻⁵; Fig. 5, e–g). In contrast, coexpression of freely diffusing CaM with Ca_V1.1_BBS and β_2A-CaM did not further enhance trafficking (P = 0.13; Fig. 5 g), suggesting that the additive effect here did not result from incomplete saturation of channel CT by CaM. Together, these findings suggest that Ca_V1.1 plasmalemmal trafficking is enriched by a duplex signaling mechanism.

With the role of CaM and stac3 on Ca_V1.1 trafficking established, we probed their effects on channel gating. However, as the activation of Ca_V1.1 is right-shifted to near its reversal potential (Table S2), detecting single-channel openings reliably in an on-cell configuration is challenging as the unitary currents at these voltages are small. Thus, to estimate changes in the maximal open probability, we analyzed macroscopic I_tail and overall gating charge movement. More specifically, the peak I_tail is linearly proportional to both the steady-state P_O of the channel at the activating prepulse potential and the number of surface–membrane channels. However, the total gating charge moved at the reversal potential (q_rev) is proportional to the number of surface–membrane channels. Gating charges can be isolated by blocking ion currents with heavy metals Cd²⁺/La³⁺. Thus, the ratio I_tail/q_rev is linearly proportional to P_O and serves as a convenient proxy to estimate changes in P_O,max under various conditions.

Although our initial functional research failed to detect appreciable Ca_V1.1 currents with auxiliary β_2A and α₂δ subunits coexpressed (Fig. 1 b), these experiments were conducted 1 d after transient transfection. Our trafficking research instead showed that Ca_V1.1 surface–membrane expression with the same subunits is substantially enhanced (P < 10⁻⁵) several days after transient transfection (Fig. S2 a). As such, we conducted whole-cell patch-clamp experiments of Ca_V1.1 with auxiliary β_2A and α₂δ subunits 2–4 d after transfection (Fig. S2 b). Scrutiny of current recordings revealed substantial gating currents in response to a 100-ms activating pulse to +80 mV, indicating the presence of surface membrane channels (Fig. 6 a, labeled Q). The duration of the activating pulse was chosen to accommodate the ultra-slow activation of Ca_V1.1, but the tail currents (I_tail) elicited at 0 mV after this activation pulse were comparatively small. Moreover, blockade of ionic currents revealed both ON gating current, in response to a depolarizing pulse, and OFF gating current during repolarization (Fig. 6 b). Computing I_tail/q_rev demonstrated low saturating values consistent with a diminished baseline P_O,max of Ca_V1.1 channels (Fig. 6 c). Moreover, normalized ON and OFF gating charges plotted as a function of voltage overlays on each other demonstrated that Q_ON and Q_OFF were similar in magnitude and voltage dependence (Figs. 6 d and S2 c). In contrast, with CaM or β_2A-CaM coexpressed, Ca_V1.1 produce markedly enhanced I_tail (Fig. 6, e and i) with similar gating currents (Fig. 6, f and j). Further analysis shows that the saturating value of I_tail/q_rev is approximately fivefold enhanced in the presence of CaM (Fig. 6 g, P = 0.006) or β_2A-CaM (Fig. 6 k, P = 0.004), suggesting that CaM up-regulates P_O,max. Reassuringly, normalized Q_ON and Q_OFF were similar in magnitude in the presence of CaM and β_2A-CaM (Fig. 6, h and l). In like manner, overexpression of stac3 also resulted in enhanced I_tail/q_rev (P = 0.006) for Ca_V1.1 (Fig. 6, m–p). These results indicate that both CaM and stac up-regulate the maximal P_O of Ca_V1.1 (Fig. 6 q). Reassuringly, further quantification of gating charge density at +80 mV (Q_density[+80]) showed a significant increase for β_2A-CaM (P = 0.039) and stac3 (P = 0.045), confirming that modulatory agents also enhance trafficking of Ca_V1.1 to the plasma membrane (Fig. 6 r). Together, these data suggest that both modulators not only boost surface–membrane expression but also up-regulate the activity of Ca_V1.1. The CaM-dependent change in maximal P_O is reminiscent of findings with related Ca_V1.3 channels (Adams et al., 2014).

Recent genetic screens have identified multiple mutations within stac3 that are associated with severe congenital myopathies as illustrated in Fig. 7 a. The first autosomal recessive mutation observed in patients of the Lumbee Native American tribe were homozygous autosomal recessive (W[284]S) in the first SH3 domain of stac3 (Stamm et al., 2008). Subsequently, compound heterozygous variants (K[288]* and L[111]Δ) were identified in a patient of Turkish heritage (Grzybowski et al., 2017). Given that stac3 binds to the Ca_V1.1 CT, we considered whether myopathy-associated mutants may disrupt this interaction and diminish surface–membrane trafficking.

Using a FRET two-hybrid assay, we assessed the binding of Ven-tagged Ca_V1.1 CI and Cer-tagged stac3 variants (Fig. 7 b). In comparison with WT, all three disease-associated stac3 variants exhibited a spectrum of weakened binding affinities (Fig. 7 c; black, WT; red, mutant). Stac3 variants L[111]Δ and W[284]S showed a nearly 10-fold weakened affinity, whereas the mutation K[288]* resulted in a twofold reduced binding of Ca_V1.1 carboxy terminus (Fig. 7 j). To discern analogous functional changes, we compared the surface–membrane trafficking of Ca_V1.1_BBS with WT or mutant stac3 in the presence of both β_1A and α₂δ subunits. Upon coexpression of WT stac3, Ca_V1.1_BBS showed strong QD labeling, confirmed by confocal imaging (Fig. 7 d) and flow cytometric analysis (Fig. 7, e and k), suggesting robust surface–membrane expression. In contrast, coexpression of stac3 variant W[284]S with Ca_V1.1_BBS resulted in sharply diminished QD labeling visualized via confocal imaging (Fig. 7 f). Population analysis using flow cytometric analysis further confirmed this result (P < 10⁻⁵; Fig. 7, g and k). Likewise, analysis of two additional disease-associated stac3 variants, L[111]Δ (P < 10⁻⁵) and K[288]* (P < 10⁻⁵), revealed variably diminished channel surface–membrane trafficking as evident from reduced ϕ_max (Fig. 7 k). Quantitatively, if the binding of stac3 to Ca_V1.1_BBS genuinely underlies the enhancement in channel surface–membrane trafficking, then this functional increase will follow a Langmuir function with the binding affinity of the stac–channel interaction as follows:

where ϕ_stac3 and Λ are constants and $ϕ max stac3$ and $ϕ max baseline$ represent the saturating surface–membrane trafficking limit in the presence and absence of stac3, respectively (see supplemental text). For stac3 variants, we assume that their relative binding affinity for the Ca_V1.1 CI deduced from FRET two-hybrid binding assays (K_a,EFF) is proportional to that for the holochannel interface (K_a). This theoretical framework for channel trafficking mirrors individually transformed Langmuir analysis previously developed to deduce binding interfaces critical for channel gating (Ben Johny et al., 2013). Plotting the experimentally determined saturating surface–membrane trafficking ratio ϕ_max versus the relative Ca_V1.1 CI binding affinities (K_d,EFF) for the stac3 variants reveals the predicted Langmuir relationship (Fig. 7 l). These results demonstrate that stac3 binding to Ca_V1.1 promotes plasmalemmal trafficking and that myopathy-associated stac variants exhibit weakened trafficking resulting from disrupted binding to the Ca_V1.1 CT.

Given that both CaM and stac3 independently enhance surface–membrane trafficking of Ca_V1.1, we next investigated whether CaM might rescue the defects in trafficking associated with myopathy-associated stac3. Consequently, we assessed surface–membrane trafficking of Ca_V1.1_BBS in the presence of both stac3 W[284]S and CaM_WT. Confocal imaging showed an increase in QD labeling (Fig. 7 h), and flow cytometry confirmed a modest rescue at the population level (P < 10⁻⁵; Fig. 7 i). Similar analysis with other myopathy-associated stac3 variants (L[111]Δ and K[288]*) further confirmed the partial rescue of Ca_V1.1_BBS trafficking when CaM_WT is coexpressed (P < 10⁻⁵ for both variants; Fig. 7 k). Intriguingly, the net magnitude of CaM-dependent enhancement in Ca_V1.1 surface–membrane trafficking is similar in the presence of all stac3 variants irrespective of their binding affinities (Fig. S3). These results suggest that the CaM effect on channel trafficking is independent of stac, consistent with the two regulatory proteins using distinct binding interfaces (Fig. S3), and raise the possibility that CaM delivery to Ca_V1.1 furnishes an orthogonal strategy for partially reversing functional defects resulting from myopathy-associated mutations in stac3. Moreover, Ca_V1.1 CT represents a prime interface for screening small molecules that promote Ca_V1.1 trafficking and function.

Recently, pharmacological chaperones have emerged as a promising strategy to rescue surface–membrane trafficking deficits observed in a variety of genetic disorders involving both G protein–coupled receptors (Beerepoot et al., 2017) and ion channels such as cystic fibrosis transmembrane conductance regulator associated with cystic fibrosis (Hanrahan et al., 2013), K_ATP channels associated with congenital hyperinsulinism of infancy (Martin et al., 2013), and Na_V1.5 channels associated with Brugada syndrome (Valdivia et al., 2004; Moreau et al., 2012). In many of these cases, small-molecule modulators that alter channel gating may offer a dual purpose as chaperones by stabilizing key channel conformations. Moreover, as Ca²⁺ influx through Ca_V1.1 channels is not necessary to trigger muscle contraction (Armstrong et al., 1972; Dayal et al., 2017), we reasoned that clinically relevant small-molecule Ca_V1 antagonists that traditionally block Ca²⁺ influx may be repurposed to reverse trafficking defects of Ca_V1.1 observed in the presence of myopathy-associated mutant stac3 (Fig. 8 a). To evaluate this possibility, bungarotoxin-labeling assays and flow-cytometric analysis were used to quantify drug-induced changes in Ca_V1.1 trafficking coexpressed with mutant stac3 W[284]S, the most prevalent myopathy-associated stac variant, and α₂δ and β_1A auxiliary subunits. We tested three L-type Ca²⁺-channel modulators, nifedipine, diltiazem, and verapamil (Fig. 8 b, cyan bars), as well as two Na channel modulators, mexiletine and ranolazine (Fig. 8 b, blue bars), clinically approved for various cardiovascular conditions, at two concentrations reflecting typical low and high therapeutic plasma concentrations. Remarkably, among Ca_V channel modulators, incubation with verapamil resulted in ∼40% recovery of Ca_V1.1 trafficking (Fig. 8 b) at low (>60%) drug concentration and ∼67% recovery at high (>80%) drug concentration (P < 10⁻⁵). Diltiazem also increased channel trafficking by ∼34% at high (>80%) drug concentration (P < 10⁻⁵). In contrast, incubation with nifedipine, mexiletine, and ranolazine resulted in minimal change (<10%) in the saturating fraction of surface–membrane channels (ϕ_max; Fig. 8 b). Of note, in all five conditions, the total GFP fluorescence remained the same, suggesting that the increase in the fraction of surface–membrane channels (ϕ_max) observed in the presence of verapamil and diltiazem reflects genuine potentiation of channel plasmalemmal trafficking. Verapamil application increased Ca_V1.1 trafficking in the absence of stac3 by approximately twofold (Fig. 8 b) but increased channel trafficking in the presence of CaM and stac3 by only 24% and 33%, respectively (Fig. S4 a). Importantly, as stac3 is also thought to directly mediate EC coupling, the partial rescue of trafficking observed in this study may not suffice to rescue deficits in muscle contraction. Nonetheless, these results highlight the utility of the bungarotoxin-labeling assay for small-molecule screens of pharmacological chaperones.

Ca_V1.1 has often appeared atypical among L-type channels with seemingly poor conservation of regulatory mechanisms and idiosyncratic requirements for membrane expression manifesting as a loss of function in heterologous systems. Our results indicate that reduced function stems from two deficits. First, quantitative flow-cytometric analyses of surface–membrane expression show that Ca_V1.1 with β and α₂δ subunits traffics to the plasma membrane, albeit at reduced levels in comparison with related L-type channels. Second, electrophysiological analyses reveal that Ca_V1.1 exhibits a low P_O. Both deficits in function depend on the channel CT harboring distinct binding interfaces for CaM and stac3, and coexpression of these proteins markedly enhances channel function. In addition, multiple myopathy-associated mutations weaken stac3 binding to Ca_V1.1 CT and fail to promote channel trafficking. Further analysis of trafficking demonstrated that clinically used Ca_V1 antagonists verapamil and diltiazem reverse trafficking defects of Ca_V1.1. In all, these findings highlight parallel mechanisms that buttress Ca_V1.1 function in heterologous expression systems, lend insight into pathophysiological deficits of Ca_V1.1 associated with congenital myopathy, and posit pharmacological strategies for rescue of channel function.

The requirements for Ca_V1.1 surface–membrane trafficking in heterologous systems have long evaded consensus. Although related Ca_V1/2 channels exhibit robust plasmalemmal trafficking with β and α₂δ subunits, additional components such as cytosolic stac3 and the transmembrane γ₁ subunit are thought to be obligatory for Ca_V1.1 currents in heterologous systems (Tuluc et al., 2009; Bannister and Beam, 2013; Polster et al., 2015, 2016). How do these modifications at disparate channel interfaces influence trafficking? Our results point to a unified trafficking scheme (Fig. 9), with the requirements for Ca_V1.1 trafficking paralleling those for related Ca_V channels (Fang and Colecraft, 2011). Specifically, the β subunit is a dominant effector necessary for Ca_V1.1 plasmalemmal trafficking (Fig. 9 a). This requirement of β subunits for Ca_V1.1 trafficking fits well with the reduced channel expression and diminished tetrad formation observed in β_1A-KO mice (Schredelseker et al., 2005). Upon binding the β subunit, however, Ca_V1.1 exhibits only low baseline trafficking (Fig. 9 b). The binding of either CaM or stac3 alone leads to only a partial enhancement in membrane trafficking (Fig. 9, c and d). Finally, the binding of both CaM and stac3 to the Ca_V1.1 CT yields a supralinear increase in membrane trafficking (Fig. 9 e). Interestingly, complete removal of the CT results in a basal increase in channel trafficking, suggesting that there may be retention motifs encoded within the CT that are masked upon the interaction of either stac3 or CaM (Fig. 4 f). This simplified scheme captures the experimentally observed effects of stac3 and CaM on Ca_V1.1 and provides a platform for other indirect mechanisms to be assessed.

Mechanistically, the Ca_V1.1 CT is a critical determinant for surface–membrane trafficking by harboring both CaM and stac3, a finding that resonates with early research that identified a vital role for this domain in triad localization (Flucher et al., 2000). As CaM is enriched in the triad via transient association with cytoplasmic loops of RYR1 (Mochca et al., 2001; Sencer et al., 2001; Xiong et al., 2002), its weak binding to Ca_V1.1 may promote colocalization of the channels at the tubular or surface membranes (Rodney and Schneider, 2003). Recurrent large-amplitude Ca²⁺ transients in the triadic space may further reinforce this localization. Indeed, the role of Ca²⁺/CaM in mediating activity-dependent trafficking has emerged as a pervasive theme in Ca_V channel physiology, yet the precise motifs that orchestrate this phenomenon are yet to be elucidated (Wang et al., 2007; Hall et al., 2013; Tseng et al., 2017). Similarly, our results indicate that stac3 potentiates Ca_V1.1 trafficking also via interaction with the CT. Even so, CaM and stac3 likely act through distinct sites as their combination supraadditively enhanced channel trafficking. In this regard, recent studies have shown that multiple channel segments including the II–III loop could bind stac3 (Wong King Yuen et al., 2017; Polster et al., 2018), although with weak affinity. It is possible that stac3 interaction with multiple Ca_V1.1 segments may concurrently enhance its affinity. Analysis of stac3^−/− zebrafish and mouse skeletal myotubes revealed a partial reduction of Ca_V1.1 at the triad, leading to incomplete tetrads and a loss of EC coupling (Polster et al., 2015; Linsley et al., 2017a,b). The magnitude of reduction varied between the two models, suggesting that other regulators such as CaM may play a role in channel trafficking in the muscle. As various stac isoforms promote trafficking of Ca_V1.2 and Ca_V3 (Rzhepetskyy et al., 2016), stac may be a shared modulator across the Ca_V family (Weiss and Zamponi, 2017). Our quantitative framework and flow-cytometric analysis of external-epitope labeling may delineate vital signals for membrane trafficking of Ca_V channels in skeletal muscle and other native cell types.

The ability to resolve Ca_V1.1 currents in heterologous systems in the presence of β and α₂δ subunits alone enables systematic analysis of channel gating modulation by regulatory partners. For nearly all Ca_V1/2 channels, CaM confers a potent feedback mechanism (Halling et al., 2006; Minor and Findeisen, 2010; Ben-Johny et al., 2015). Our analysis shows that local enrichment of CaM to Ca_V1.1 results in a fivefold increase in maximal P_O (Fig. 6 k). As potentiation in gating occurs at high voltages where channels convey minimal Ca²⁺ influx, this effect likely depends on apoCaM interaction. Excitingly, these results are evocative of recent findings that apoCaM binding augments the baseline P_O of Ca_V1.3 variants (Adams et al., 2014), hinting at a conserved mechanism across the Ca_V superfamily (Ben-Johny et al., 2015). Of note, effects on channel gating and trafficking were both elicited by CaM fused to β subunit. As β subunits have a 1:1 stoichiometry with α subunits (Dalton et al., 2005; Wu et al., 2016), a single CaM mediates both functional effects. Thus, CaM signaling may be bifurcated, whereby the apo form enhances channel gating and the Ca²⁺-bound form enriches channels at the plasma membrane.

Stac3 coexpression up-regulates the baseline P_O of Ca_V1.1 to the same extent as CaM depending on the CI module (Fig. 6 o). Interestingly, in skeletal myotubes, homozygous stac3 KO (Polster et al., 2015; Linsley et al., 2017a) and Ca_V1.1 mutants with weakened CaM binding (Stroffekova, 2011) lead to a dramatic loss of EC coupling despite the presence of gating charge movements. Thus, robust EC coupling may require a permissive Ca_V1.1 CT conformation along with that for the II–III loop (Tanabe et al., 1990a). Synthesizing a general framework of Ca_V modulation by CaM and stac is an exciting frontier, and the ability to express Ca_V1.1 in heterologous systems under a wide range of conditions facilitates this pursuit.

Stac3 has been identified as a vital genetic locus for debilitating congenital myopathy that encompasses an expanding list of mutations. Patients exhibit a plethora of myopathy-associated symptoms including facial weakness with ptosis, hypotonia, small stature, scoliosis, cleft palate, and susceptibility to malignant hyperthermia (Stamm et al., 2008; Zaharieva et al., 2014; Grzybowski et al., 2017; Telegrafi et al., 2017). Current treatment strategies focus on early diagnosis and symptom management, particularly anticipatory management of malignant hyperthermia, and novel small-molecule agents that reverse pathogenesis are highly desired.

Our analysis reveals that disease-associated stac3 variants weaken binding to the CT, resulting in variably diminished Ca_V1.1 surface–membrane trafficking, highlighting potential pathogenic mechanisms. Indeed, reconstitution of myopathy-associated mutant stac3 (W[284]S) in stac3^−/− KO zebrafish and mouse models led to diminished trafficking, triadic organization, and activity of Ca_V1.1, resulting in marked loss of EC coupling (Polster et al., 2016; Linsley et al., 2017a,b). As patients are either homozygous or compound heterozygous for stac mutations, it is likely that the weakened affinity of stac3 for Ca_V1.1 CT results in incomplete saturation of Ca_V1.1 by this regulatory protein. Our findings point to three distinct avenues for developing effective pharmacological strategies. First, given that CaM can both partially rescue reduced Ca_V1.1 trafficking and enhance Ca_V1.1 activation gating, local enrichment of CaM may be an effective strategy for reversing the pathophysiology of stac3-associated myopathies. In this regard, a CRISPR-interference approach was recently developed to selectively manipulate CaM expression for a subset of cardiac arrhythmogenic long-QT syndrome (Limpitikul et al., 2017). Second, as we identify Ca_V1.1 CT as the primary effector interface for stac3, FRET two-hybrid binding assay may be repurposed to devise high-throughput screens for small-molecule modulators that enhance this interaction and tune skeletal muscle function (Janzen, 2014). Third, certain Ca_V channel antagonists such as diltiazem and verapamil at low therapeutic plasma concentrations partially rescue these trafficking defects, depending on continual exposure to the drug. Structurally, phenylalkylamines such as verapamil bind pore-lining residues of the domain III–IV S6 segments adjacent to the beginning of the carboxy terminus (Striessnig et al., 1990; Tang et al., 2016). Moreover, mutations within the CaM-binding IQ domain in the Ca_V1 carboxy terminus have been shown to allosterically modify binding of phenylalkylamines and other Ca channel antagonists (Dilmac et al., 2004; Huang et al., 2013). Thus, it is possible that the binding of verapamil may either directly stabilize the CT or allosterically switch its conformation to ultimately promote plasmalemmal trafficking. As EC coupling in skeletal muscle does not depend on freely diffusing Ca²⁺ ions, blockade of Ca²⁺ influx resulting from Ca_V antagonists may not significantly alter the strength of EC coupling (Dayal et al., 2017). Paradoxically, recent research has shown that verapamil can potentiate contractions in mouse skeletal muscle (Dayal et al., 2017). Nonetheless, as stac3 is thought to be directly involved in EC coupling, a simple rescue of plasmalemmal channels may be insufficient to reverse pathogenesis in the case of congenital myopathies. Further functional analysis of EC coupling in stac3 mutant animal models after long-term application of verapamil is necessary to assess therapeutic potential. Nonetheless, the quantitative flow-cytometric assay promises to facilitate discovery of small-molecule trafficking modulators. Indeed, similar pharmacological chaperones have emerged as a potential therapeutic avenue for rescue of trafficking deficits associated with cystic fibrosis (Hanrahan et al., 2013), congenital hyperinsulinism of infancy (Martin et al., 2013), and Brugada syndrome (Valdivia et al., 2004; Moreau et al., 2012).

In all, our results hint at a conserved mechanism by which multiple signaling molecules tune Ca_V1.1 gating and localization, inform on mechanisms of disease pathogenesis for congenital myopathy, and suggest potential avenues for development of therapeutic strategies.

We thank Dr. Gordon Tomaselli and Dr. Ivy E. Dick for helpful comments on experimental design and the manuscript. We are also grateful for insightful discussions from the Calcium Signals Laboratory. Finally, we could not have completed this work without the inspirational example of our late mentor David T. Yue, who taught us to pursue science with a passion for the truth.

This work was supported by grants from the National Institute of Neurological Disorders and Stroke (grant NS085074 to D.T. Yue and T. Inoue), the National Institute of Mental Health (grant MH065531 to D.T. Yue and M. Ben-Johny), and the National Science Foundation (J. Niu).

The authors declare no competing financial interests.

Author contributions: J. Niu, M. Ben-Johny, D.T. Yue, and T. Inoue conceived and designed the study. J. Niu made the constructs and collected electrophysiological and FRET data with assistance from M. Ben-Johny and W. Yang. J. Niu and W. Yang performed bungarotoxin labeling and flow cytometry experiments. J. Niu and M. Ben-Johny analyzed all data. J. Niu and M. Ben-Johny wrote the manuscript with input from T. Inoue.

Richard W. Aldrich served as editor.