Asymmetric contribution of a selectivity filter gate in triggering inactivation of Ca_V1.3 channels

The high voltage–activated Ca²⁺ channels (Ca_V1/2) serve as vital conduits for Ca²⁺ entry responsible for orchestrating diverse physiological processes, ranging from excitation–contraction coupling in muscle (Bers, 2002) to vesicle secretion and gene transcription in neurons (Berridge et al., 2000; Dolmetsch, 2003). Molecularly, the eukaryotic Ca_V1/2 channels possess a conspicuous pseudotetrameric architecture that is thought to permit asymmetric conformational changes to fine-tune channel activity (Catterall et al., 2017). For related voltage-gated Na⁺ (Na_V) channels, this asymmetric architecture allows distinct voltage-sensing domains (VSDs) to differentially impact activation versus inactivation and pore conformation (Goldschen-Ohm et al., 2013). By comparison, although Ca_V channels have VSDs with distinct kinetics, the importance of the overall pseudotetrameric architecture and potential asymmetric conformational changes are not fully established, though likely important (Tuluc et al., 2016; Fernández-Quintero et al., 2021; Savalli et al., 2021). Like Na_V, Ca_V channels are subject to exquisite feedback regulatory mechanisms, including both voltage-dependent inactivation (VDI) and Ca²⁺-dependent inactivation (CDI) that limits excessive Ca²⁺ entry into the cell (Minor and Findeisen, 2010; Ben-Johny and Yue, 2014). Both processes are biologically consequential as Ca²⁺ channelopathies involve deficits in inactivation that disrupt electrical signaling and Ca²⁺ homeostasis, thereby contributing to neurological, neurodevelopmental, and cardiac arrhythmogenic disorders such as Timothy syndrome (TS; Barrett and Tsien, 2008; Dick et al., 2012; Limpitikul et al., 2014; Pinggera et al., 2015). Both inhibitory processes have disparate molecular origins. VDI is initiated by voltage-dependent conformational changes triggered by the movement of the four homologous transmembrane voltage-sensing domains (Stotz and Zamponi, 2001; Stotz et al., 2004; Tadross et al., 2010). By contrast, CDI is largely independent of voltage (Tadross and Yue, 2010) and, instead, ensues from Ca²⁺ binding to the two lobes of calmodulin (CaM) tethered to the Ca_V C-terminus (Peterson et al., 1999; Qin et al., 1999; Zühlke et al., 1999). These processes converge on distinct “gates” in the pseudosymmetric pore domain to obstruct ion influx (Tadross et al., 2010; Abderemane-Ali et al., 2019). Even so, the exact nature of conformational changes that underlie channel feedback inhibition and the contribution of the pseudosymmetric channel architecture are not fully understood.

Three end-stage mechanisms of inactivation have been proposed, paralleling findings with related ion channels (Fig. 1 A). These include (1) a hinged-lid mechanism that involves a cytosolic inactivation particle that occludes the ion conduction pathway (Stotz et al., 2000) similar to N-type inactivation of Shaker K⁺ channels (Hoshi et al., 1990) or fast inactivation of Na_V channels (West et al., 1992; Yan et al., 2017), (2) an allosteric closure or destabilization of the S6 activation gate (Imredy and Yue, 1994; Tadross et al., 2010) similar to allosteric regulation of K_V7 channels (Zaydman and Cui, 2014), and (3) a conformational rearrangement of the selectivity filter (SF; Abderemane-Ali et al., 2019), akin to P-type (or C-type) inactivation in Shaker K⁺ channels (Hoshi et al., 1991) or slow inactivation of Na_V channels (Balser et al., 1996). For VDI, a hinged-lid mechanism has been proposed (scheme 1), with the N-terminus of the intracellular domain I–II (DI–DII) linker serving as an inactivation particle that binds to a transmembrane receptor site formed by the S6 gate (Stotz et al., 2000; Dafi et al., 2004; Tadross et al., 2010). For CDI, initial studies indicated an allosteric mechanism, where channels switch to a distinct gating pattern with sparse openings (scheme 2; Imredy and Yue, 1994; Tadross et al., 2010). More recently, structure-guided mutagenesis experiments point to the involvement of an alternate SF gate for both VDI and CDI (scheme 3; Abderemane-Ali et al., 2019). Even so, the extent of SF conformational changes in channel feedback inhibition and key factors that govern this process remains to be fully understood.

Here, we sought to delineate the functional importance of conformational changes in the SF in mediating both VDI and CDI. We focused on Ca_V1.3 channels for two reasons. First, these channels exhibit characteristically slow VDI kinetics owing to the presence of a “shield” that counteracts a hinged-lid inactivation mechanism (Tadross et al., 2010). This permits a bottom-up approach, whereby VDI may be engineered, thereby allowing us to probe the functional role of the SF gate. Second, these channels also demonstrate robust CDI that depends on distinct conformational changes involving both the N- and C-lobes of CaM. We considered an ultrahighly conserved Trp residue (W[+2]) within the SF found at the interface between the pore helices from different domains (Payandeh and Minor, 2015; Fig. 1, B and C). Systematic mutagenesis of the corresponding residue in each domain revealed an asymmetric switch in the pore domain that upregulates VDI. Furthermore, in depth analysis of CDI showed that mutating the W[+2] residue in distinct domains asymmetrically diminished CDI mediated by distinct CaM lobes. Of pathophysiological relevance, we further show that destabilizing the SF gate could reverse deficits in inactivation linked to TS. In all, these findings highlight the fundamental importance of the asymmetric arrangement of the pseudotetrameric Ca_V1.3 pore domain for channel inactivation.

Mutagenesis Ca_V1.3 was performed using the truncated variant of rat Ca_V1.3 (AF370009.1), Ca_V1.3_Δ1626, as previous published (Tadross et al., 2010; Ben-Johny et al., 2013). Briefly, we first PCR-amplified subsegments of wild-type Ca_V1.3 containing SF domains of DI and DII (R1, flanked by restriction enzyme sites BsiWI and Eco47III), DIII (R2, flanked by restriction enzyme sites Eco47III and BglII), and DIV (R3, flanked by restriction enzyme sites BglII and XbaI), respectively. These subsegments were inserted into the pCRBlunt II-TOPO vector using the Zero Blunt TOPO PCR cloning kit (Invitrogen). Point mutations in respective Ca_V1.3 subsegments were then generated using Quikchange II kit (Agilent). We generated W357A and W718A and used R1 plasmid, while R2 plasmid was used for W1114A and R3 plasmid for W1404A, W1404F, W1404V, and W1404T mutations. Subsequently, the mutated segments were ligated into the wild-type Ca_V1.3 following restriction digest with the specific enzymes for a given subsegment. Ligates were then transformed into either XL10-Gold Ultracompetent Cells (Agilent) or DH5α Competent Cells (Thermo Fisher Scientific), plated, and cultured in selective LB broth. DNA was extracted and purified from cultures using either QIAprep Spin Miniprep Kit (Qiagen) or GeneJET PCR Purification Kit (Thermo Fisher Scientific). Sanger sequencing was used to verify each mutant. For T1107Q/W1404A double mutant, we first generated the T1107Q mutation using Quikchange II (Agilent) in the R2 subsegment. Subsequently, this region was ligated into the W1404A mutant channel following restriction digest using Eco47III and BglII. Human Ca_V1.2 and the G406R mutant were generously gifted by Dr. Ivy Dick as were previously published (Dick et al., 2016). DIV W[+2]A mutation corresponding to residue 1,466 was generated on the G406R mutant background using Quickchange II (Agilent).

For whole-cell electrophysiology, HEK293 cells were cultured on glass coverslips in 60-mm dishes and transfected using a calcium phosphate method. We applied 2–8 μg of cDNA encoding the desired channel α₁ subunit (WT or engineered variant), along with 4 μg of rat β subunit (β_1b, β_2a, β₃, or β₄) and 4 μg of rat brain α₂δ₁ (NCBI accession no. NM012919.2). In experiments that required CaM expression, we transfected 2–4 μg of cDNA encoding CaM WT or mutant variants (CaM₁₂, CaM₃₄, or CaM₁₂₃₄). To enhance the expression, cDNA for simian virus 40 T antigen (0.5 μg) was cotransfected. Electrophysiology recordings were done at room temperature 1–3 d after transfection.

Whole-cell voltage-clamp recordings for HEK293 were collected at room temperature using an Axopatch 200B amplifier (Axon Instruments). Glass pipettes (MTW 150-F4, World Precision Instruments) were pulled with a horizontal puller (P97; Sutter Instruments Co.) and fire-polished (Microforge; Narishige), resulting in 1–3-MΩ resistances, before series resistance compensation of 70%. Internal solutions contained 135 mM CsMeSO₃, 5 mM CsCl₂, 1 mM MgCl₂, 4 mM MgATP, 10 mM HEPES, and 10 mM BAPTA, adjusted to 290–295 mOsm with CsMeSO₃ and pH 7.4 with CsOH. The external solutions contained 140 mM TEA-MeSO₃, 10 mM HEPES (pH 7.4), and either 40 mM CaCl₂ or BaCl₂. This external solution composition was chosen based on previous studies to ensure that local Ca²⁺ signals are saturated to drive maximal local CDI (Tadross et al., 2008). For experiments using Na⁺ as a charge carrier, we used an external solution containing 140 mM NaCl, 10 mM HEPES (pH 7.4), 10 mM TEA-MeSO₃, and 0.5 mM EGTA. Solutions were adjusted to 300 mosM with TEA-MeSO₃ and pH 7.4 with TEA-OH. For CDI and VDI measurements, we used a family of test pulses from −50 to +50 mV with repetition intervals of 20–120 s from a holding potential of −80 mV. Custom MATLAB (Mathworks) software was used to determine peak current and fraction of peak current remaining after either 300 ms (r₃₀₀) of depolarization or 800 ms (r₈₀₀) of depolarization. Residual currents after depolarization were measured at +10 mV after an initial pulse of +10 mV for 15 ms followed by a family of test pulses from −60 to +50 mV for 800 ms. Custom MATLAB (Mathworks) software was used to determine peak current and I₂/I₁ ratio. As W[1404]A mutation itself enhanced VDI, quantifying CDI (i.e., the CaM-driven component of inactivation) by comparing inactivation with Ba²⁺ versus Ca²⁺ as the charge carrier is imprecise. Accordingly, we estimated baseline VDI for W[1404]A mutation in the presence of Ca²⁺ by coexpressing CaM₁₂₃₄ (⁠$r300,CaM1234=I300ms/Ipeak$ and $r800,CaM1234=I800ms/Ipeak$⁠), where I_peak is the peak current, and I₃₀₀ and I₈₀₀ are current levels following 300 ms or 800 ms of depolarization, respectively. We then measured total inactivation (CDI and VDI) with Ca²⁺ as the charge carrier with endogenous CaM, CaM₁₂, or CaM₃₄ to isolate total, C- and N-lobe components of CDI, respectively (⁠$r300,Ca=I300ms/Ipeak$ and $r800,Ca=I800ms/Ipeak$⁠, with Ca²⁺ as the charge carrier). Subsequently, CDI was estimated as (⁠$CDI300=1−r300,Ca/r300,CaM1234$⁠) or (⁠$CDI800=1−r800,Ca/r800,CaM1234$⁠). For all other mutations, we observed minimal VDI. As such CDI was measured as $CDI300=1−r300,Ca/r300,Ba$ and $CDI800=1−r800,Ca/r800,Ba$⁠. For these experiments, the experimenter was not blinded to the specific experimental condition as transfected cells included a fluorescent reporter. Data from different conditions were acquired in a random order. For each condition, we obtained data from at least two independent transfections.

As starting structures for our simulations, we used the recently published cryo-EM structure of the Ca_V1.3 Ca²⁺ channel complex (PDB accession code: 7UHG; Yao et al., 2022). The structures of the W1404A and the W718A mutants were derived from the wild-type model by replacing the mutated residue followed by a local energy minimization using MOE (Molecular Operating Environment, version 2020.09; Molecular Computing Group, Inc.). The C- and N-terminal parts of each domain were capped with acetylamide (ACE) and N-methylamide to avoid perturbations by free-charged functional groups. The structure model was embedded in a plasma membrane consisting of POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) and cholesterol in a 3:1 ratio using the CHARMM-GUI Membrane Builder (Jo et al., 2008, 2009). Water molecules and 0.15 M CaCl₂ were included in the simulation box. Energy minimizations of wild-type and mutant structures in the membrane environment were performed. The topology was generated with the LEaP tool of the AmberTools21 using force fields for proteins and lipids, ff14SBonlysc and Lipid14, respectively (Dickson et al., 2014). The wild-type and mutant structures were gradually heated from 0 to 300 K in two steps, keeping the lipids fixed, and then equilibrated over 1 ns. Then molecular dynamics simulations were performed for 3 × 300 ns, with time steps of 2 fs at 300 K and in anisotropic pressure scaling conditions. Van der Waals and short-range electrostatic interactions were cut off at 10 Å, whereas long-range electrostatics were calculated by the Particle Mesh Ewald (PME) method. PyMOL was used to visualize the key interactions and point out differences in the wild-type and mutant structures (The PyMOL Molecular Graphics System; Version 2.0 Schrödinger, LLC).

Fig. S1 shows that mutations in Ca_V1.3 DI–DIV SF mutations minimally alter channel activation. Fig. S2 shows an extended biophysical characterization of Ca_V1.3 DIV W[+2]A mutant, including an in-depth assessment of inactivation using Ca²⁺ and Na⁺ as charge carriers.

Previous studies have identified the presence of a highly conserved Trp (W[+2]) residue, two residues downstream of the selectivity filter of Ca_V, Na_V, and NALCN channels (Payandeh and Minor, 2015). Fig. 1 B shows sequence alignment confirming the preservation of this residue within all four domains of Ca_V channels, while Fig. 1 C shows its structural location within the interdomain interfaces. We individually substituted W[+2] residues in DI–DIV of Ca_V1.3 with alanine, yielding W[357]A, W[718]A, W[1114]A, and W[1404]A mutant channels, respectively. For convenience, we refer to these mutations by the nomenclature: Dx W[y]A where x denotes the particular domain and y refers to the position relative to the central Glu that forms the SF. To probe changes in VDI, we transfected either wild-type or mutant channels in HEK293 cells and undertook whole-cell current recordings utilizing Ba²⁺ as the charge carrier. Here, we cotransfected both α₂δ₁ and β_2A subunits as they are essential for trafficking and proper function of the Ca_V1.3 channel. Importantly, the β subunit is well known to tune inactivation properties. As such for these experiments, we chose β_2A, a palmitoylated variant that strongly reduces inactivation compared with other β subunit isoforms (Olcese et al., 1994; Chien et al., 1996; Colecraft et al., 2002; Van Petegem et al., 2004; He et al., 2007). As Ba²⁺ binds poorly to calmodulin (Chao et al., 1984), the use of this charge carrier permits quantification of VDI independent of CDI. For the wild-type Ca_V1.3 short variant, an 800-ms voltage step to 0 mV revealed rapid activation and minimal inactivation, consistent with previous studies (Fig. 1 D, gray trace). Similarly, analysis of DI W[+2]A, DII W[+2]A, and DIII W[+2]A mutant channels also revealed minimal VDI (Fig. 1, D–F, black traces). By contrast, exemplar current recordings with the DIV W[+2]A mutant showed a rapid decay of Ba²⁺ currents following initial activation, demonstrating increased inactivation (Fig. 1 G, black trace). To quantify the steady-state extent of inactivation, we computed the ratio of the current remaining following 800 ms of depolarization with that of the peak value (r₈₀₀). Population data confirmed a significant increase in inactivation with the DIV W[+2]A mutation, manifesting as a reduced r₈₀₀ value, in comparison with wild-type and other mutant channels (Fig. 1 H). For all mutants, we observed minimal changes in the normalized current–voltage relationship suggesting a minimal effect of the SF mutations on channel activation (Fig. S1). These findings suggest that the pore domain constitutes a key structural determinant for VDI in Ca_V1.3. Furthermore, the selective effect of DIV W[+2]A in enhancing VDI highlights the functional asymmetry of pore domain conformational changes.

Canonical VDI of Ca_V1/2 channels follow a Boltzmann relationship with voltage, whereby currents evoked at more depolarizing potentials exhibit stronger inactivation (Imredy and Yue, 1994). We, therefore, sought to determine whether enhanced Ca_V1.3 VDI resulting from the pore-domain mutation also follows a similar trend. As such, the inactivation of Ca_V1.3 DIV W[+2]A mutant channels was determined by measuring fractional decay of whole-cell Ba²⁺ currents evoked in response to a family of step-depolarizations (Fig. 2, A and B). Normalizing each current trace to its peak value showed that the extent of inactivation was reduced for DIV W[+2]A mutant at higher potentials but not wild-type channels (Fig. 2, A and B; and Fig. S2 A). To further quantify the voltage dependence of inactivation, we utilized a two-pulse protocol (Fig. 2, C–F; Patil et al., 1998). A brief 15-ms prepulse to 10 mV is used to probe the available current prior to VDI (I_pre; Fig. 2 C and Fig. S2 B). A family of 800-ms step depolarizations to various test-pulse potentials (V_test) is then used to elicit steady-state levels of inactivation, while a 15-ms postpulse to 10 mV measures the residual current, reflecting ion influx through non-inactivated channels. The strength of VDI is quantified as the ratio of peak current during the postpulse (I_post) to that during the prepulse (I_pre; Fig. S2 B). For wild-type Ca_V1.3, we observed a modest increase in peak currents during the postpulse as compared with the prepulse at more depolarized test-potential, indicating the presence of weak voltage-dependent facilitation (Fig. 2, C and E; and Fig. S2 B). By comparison, for the DIV W[+2]A mutant, we observed a U-shaped voltage-dependence of VDI with the largest reduction in peak current observed with modest depolarizations (Fig. 2, D and F; and Fig. S2 C). We further assessed voltage dependence of inactivation with either Ca²⁺ or Na⁺ as the charge carrier. In both cases, we observed a U-shaped voltage dependence (Ca²⁺, Fig. S2, D–F; Na⁺, Fig. S2, G–I). Of note, for these experiments, we coexpressed Ca²⁺-insensitive mutant CaM (CaM₁₂₃₄) whose Ca²⁺ binding sites in its EF hands were disabled (Xia et al., 1998; Peterson et al., 1999). As this maneuver disables CDI, it allows us to dissect the CaM-independent effects of Ca²⁺ on channel gating.

Conventional VDI of Ca_V1/2 channels is highly tuned by the auxiliary β-subunit associated with the channel. More specifically, membrane-localized β-subunits (e.g., palmitoylated β_2A) reduce VDI, while other isoforms yield strong VDI. For Ca_V1.3, we previously identified mutations in the distal S6 gate that enhanced VDI (Tadross et al., 2010). However, the magnitude of VDI for these mutations varied depending on the specific β-subunit coexpressed. Here, to determine whether VDI conferred by SF mutation is also tuned by the auxiliary β-subunit, we coexpressed Ca_V1.3 DIV W[+2]A mutant with non-membrane localized β₁, β₃, or β₄ subunits. In all cases, we found that the extent of VDI was not further accentuated compared to that with the β_2A subunit (Fig. 2, G and H). Taken together, these findings suggest that VDI conferred by SF mutation is a biophysically distinct process compared with conventional VDI of Ca_V1 and Ca_V2 channels.

Given the asymmetric effect of DIV W[+2] residue and its location at the interface of two domains, we considered whether this mutation destabilizes the conformation of the SF. To probe this possibility, we performed molecular dynamics (MD) simulations (3 × 300 ns) based on the recently published cryo-EM structure of the Ca_V1.3 channel (Yao et al., 2022). We considered both WT and the mutant channels embedded in a lipid bilayer. This cryo-EM structure has a global resolution of ∼3 Å and the local quality of the electron density maps suggests that this region is well resolved. To quantify potential changes in flexibility upon substitution of tryptophan residues with alanine, we used the obtained simulations to calculate residue-wise B-factors. B-factors are an alignment-dependent measure to identify areas with higher fluctuations or flexibility. In wild-type channels, the two helices of the pore loop are all very rigid (low B-factors; Fig. 3, A, C, E, and G, green), except in DIV, where the P1 helix displays elevated flexibility (higher B-factor; Fig. 3 A, red). In comparison with wild-type channels, the introduction of the DIV W[+2]A mutation increased the local conformational flexibility of this region evident as increased B-factor mapped on the structural model, particularly in the P2 helix (Fig. 3, A and B). This suggests that this mutation destabilizes the SF conformation to potentially initiate inactivation. To evaluate this possibility further, we considered whether similar mutations in the other domains also yielded increased conformational flexibility. In contrast to the DIV W[+2]A, we found that W[+2]A mutations in all other repeats did not appreciably increase the structural variability of the SF and only minimally altered the B-factors (Fig. 3, C–H), suggesting that the conformational stability of the SF is largely preserved by these mutations. Apparently, the increased conformational flexibility of the SF of DIV is not the direct consequence of the amino acid substitution in the P2 helix, rather it depends on the specific environment of the SF in this domain. Interestingly, further scrutiny of the DIV W[+2]A mutant revealed key alterations in the conformation of the Ca²⁺ coordinating E[0] residues that line the SF (Fig. 3 I). Specifically, the DIV E[0] residue shifts away from the center of the ion permeation pathway by ∼4 Å, forming a contact with DI W[+2] residue (W357). This change may disrupt the high-affinity coordination of a Ca²⁺ ion in the SF, ultimately impacting ion influx through the channel. Taken together, these results highlight the asymmetric effect of SF mutations on conformational stability and the unique role of DIV W[+2] mutation for VDI of Ca_V1.3.

Having inferred that the DIV W[+2]A mutant undergoes a distinct form of inactivation, we sought to dissect its underlying mechanisms. Structurally, the W[1404] residue is wedged at the interface between selectivity filter helices from DIV and DIII (Fig. 1 C). The alanine mutation would remove this wedge and potentially promote conformational changes in the selectivity filter that manifests as channel inactivation. As such, we reasoned that the bulkiness of the W[1404] side chain may be essential to stabilize the selectivity filter and tune inactivation kinetics. We thus replaced the tryptophan residue with a threonine (DIV W[+2]T) and undertook whole-cell recordings. We found that DIV W[+2]T also increased VDI; however, the net magnitude of increase was blunted in comparison to the DIV W[+2]A (Fig. 4 A), consistent with the intermediate size of the threonine side-chain. In like manner, substitution of the DIV W[+2] residue with a valine (DIV W[+2]V) also yielded an intermediate increase in VDI (Fig. 4 B). By contrast, replacing the DIV W[+2] residue with a bulky phenylalanine residue (DIV W[+2]F) yielded no appreciable increase in VDI as compared with wild-type channels (Fig. 4 C). These findings qualitatively confirm the trend that increased bulkiness of the DIV W[+2] residue reduces the magnitude of VDI. To quantify this trend, we plotted the r₈₀₀ values for various DIV W[+2] substitutions as a function of the side-chain accessible surface area (ASA). We found that the r₈₀₀ values increased with increasing ASA and with the substitution of bulkier residues resulting in diminished VDI (Fig. 4 G). This relationship was empirically fit using a Boltzmann function. These findings are consistent with the DIV W[+2] residue serving to stabilize the DIII–DIV SF interface.

To further test this possibility, we hypothesized that introducing bulky side chains to residues in the DIII interface could compensate for the decreased stability resulting from DIV W[+2]A mutation. Accordingly, we identified candidate residues in DIII selectivity filter helices that are within 5 Å of the DIV W[+2] residue. These included residues F[1106] (DIII F[−6]), T[1107] (DIII T[−5]), T[1110] (DIII T[−2]), and E[1112] (DIII E[0]; Fig. 4 E). Examination of corresponding residues in domain I, II, and IV showed that F[−6], T[−2], E[0] are conserved. By contrast, the analogous position of DIII T[−5] residue in domains I, II, and IV is occupied by either a glutamine or an arginine (Fig. 4 E). As such, we replaced T[1107] with a glutamine (T[1107]Q) in DIV W[+2]A mutant channel and probed changes in VDI. Indeed, we found that the DIV W[+2]A/DIII T[−5]Q double mutant exhibited substantially diminished inactivation as compared to DIV W[+2]A mutant channels (Fig. 4, F and G). Taken together, these results suggest that the symmetric W[+2] residues determine the stability of the selectivity filter in concert with the asymmetric interdomain interface residues. The two together are critical to prevent the collapse of the ion conduction pathway, and thus confer a unique role for DIV in suppressing VDI in Ca_V1.3 channels.

Distinct from VDI, Ca_V1.3 channels undergo rapid CDI mediated by CaM. Recent studies suggest that the end-stage mechanism of CDI relies on conformational changes in the selectivity filter (Abderemane-Ali et al., 2019). Given the proximity of the tryptophan residues in the structure of the selectivity filter to the Ca²⁺ coordinating residues, we considered whether these mutations also alter CDI. At baseline, Ca_V1.3 channels exhibit strong CDI (Fig. 5 A, rose curve) quantified as the fraction of excess inactivation with Ca²⁺ versus Ba²⁺ as charge carrier measured following 300 ms of depolarization (⁠$CDI=1−r300,Ca/r300,Ba$⁠). Here, CDI measurements were performed at ambient CaM levels. Analysis of DI W[+2]A, DII W[+2]A, and DIII W[+2]A mutants revealed minimal changes in CDI (Fig. 5, A–C). As DIV W[+2]A mutation enhances VDI, comparison of inactivation kinetics with Ca²⁺ versus Ba²⁺ as permeant ion is insufficient to quantify CDI. To obviate this limitation, we estimated CDI as excess inactivation of Ca²⁺ currents with endogenous CaM compared to with Ca²⁺-binding deficient CaM₁₂₃₄ (see Materials and methods). The latter maneuver disables CDI and provides an estimate of average VDI with Ca²⁺ as permeant ion (Fig. 5 D, blue). Thus measured, the DIV W[+2]A mutant showed a modest but statistically significant reduction in CDI (P < 0.001, Fig. 5 E).

As Ca_V1.3 CDI is mediated by both N- and C-lobes of CaM (Yang et al., 2006), we next considered whether pore domain mutations may differentially impact regulation by the distinct CaM lobes. To evaluate this possibility, we measured C-lobe CDI by co-expressing mutant CaM₁₂ with Ca²⁺ binding to its N-lobe disabled, or N-lobe CDI in the presence of CaM₃₄ whose Ca²⁺ binding to the C-lobe is disabled. Indeed, we found that DI W[+2]A, DII W[+2]A, and DIII W[+2]A mutants had minimal or no effect on C-lobe CDI (Fig. 5, F–H and J). However the DIV W[+2]A mutation strongly inhibited C-lobe CDI (Fig. 5, I and J). By contrast, we observed that N-lobe CDI was sharply diminished for DI W[+2]A, DII W[+2]A, and DIII W[+2]A mutants (Fig. 5, K–M and O), while the DIV W[+2]A mutant exhibited an increase in CDI (Fig. 5, N and O). These findings highlight the asymmetric contribution of pore domain residues in differentially coupling CDI triggered by distinct CaM lobes.

Human mutations in Ca_V1.2 cause a severe multisystem disorder known as TS, marked by an increased likelihood of cardiac arrhythmia and neurological and neurodevelopmental deficits (Splawski et al., 2004). Mechanistically, these channelopathic variants are well established to exhibit reduced inactivation (both VDI and CDI) resulting in prolonged Ca²⁺ entry that delays action potential repolarization and disrupts Ca²⁺ homeostasis (Barrett and Tsien, 2008; Dick et al., 2016; Calorio et al., 2019). As such, identifying mechanisms that reverse this change would be beneficial from the perspective of devising new therapeutics. Since DIV W[+2]A mutation in Ca_V1.3 enhanced VDI, we probed whether the analogous mutation in a TS Ca_V1.2 variant could reverse deficits in VDI observed for mutant channels. We considered the G406R variant, known to exhibit a strong functional phenotype. Consistent with previous studies (Barrett and Tsien, 2008; Dick et al., 2016), we found that this variant exhibited reduced both VDI and CDI compared with wild-type Ca_V1.2 channels (Fig. 6, A and B). Introduction of the DIV W[+2]A mutation to the G406 variant resulted in a marked enhancement in VDI (Fig. 6 C), restoring overall inactivation to near wild-type levels. This finding suggests that enhancing the conformational flexibility of the SF may be a potential approach to reverse the pathophysiological reduction in inactivation linked to TS.

Ca_V channel inactivation is a physiologically consequential and mechanistically rich ion channel feedback regulation that serves to limit excess Ca²⁺ entry into neurons and cardiomyocytes (Minor and Findeisen, 2010; Ben-Johny and Yue, 2014). Here, we probed the functional importance of the SF in tuning both VDI and CDI. Mutagenesis of an ultrahighly conserved W[+2] residue (Payandeh and Minor, 2015) in DI–DIV revealed an unexpected asymmetric role of the pore domain in tuning channel inactivation. First, we found that mutating the W[+2] residue in DIV but not DI–DIII enhanced VDI. MD simulations showed that DIV W[+2]A mutation increased the local conformational flexibility of the pore domain, hinting at a distinct role for the SF gate in tuning VDI. Although the W[+2] residue is symmetric in all four domains, the asymmetric effect of mutating DIV W[+2] residue on VDI stems from distinct differences in neighboring residues that stabilize the interdomain interfaces. Second, an in-depth analysis of CDI revealed a structural bifurcation of CaM signaling, whereby the DIV mutation selectively diminished CDI mediated by the C-lobe of CaM, while mutations in DI–DIII preferentially reduced N-lobe CDI. Taken together, these findings demonstrate that the pseudotetrameric arrangement of the eukaryotic Ca_V channel pore enables asymmetric conformational changes that are fundamentally important for its distinct inactivation mechanisms.

For VDI, a hinged-lid mechanism has been proposed based on multiple lines of evidence (Stotz et al., 2000; Dafi et al., 2004; Tadross et al., 2010): (1) chimeric analyses of Ca_V1/2 channels pointed to the involvement of both the DI S6 segment and the intracellular DI–DII linker in tuning inactivation kinetics (Stotz and Zamponi, 2001; Stotz et al., 2004). (2) Kinetics of VDI is dependent on the specific auxiliary Ca_Vβ subunit associated with the channel DI–DII linker, with membrane-tethered Ca_Vβ subunits slowing VDI (Olcese et al., 1994; Chien et al., 1996). (3) Mutagenesis of the presumed S6 transmembrane receptor site also alters VDI (Stotz and Zamponi, 2001; Dafi et al., 2004; Tadross et al., 2010). (4) Exogenous expression of DI–DII linker as a peptide accelerates VDI (Cens et al., 1999). Structurally, the DI–DII linker forms a continuous helix with the S6 gate and disruption of the helix by a polyglycine linker promotes VDI, findings that are inconsistent with a hinged-lid model (Findeisen and Minor, 2009; Wu et al., 2016). Recent studies instead point to the involvement of the SF for VDI of Ca_V channels, with select mutations in the SF diminishing VDI (Abderemane-Ali et al., 2019). In this study, the slow baseline VDI of Ca_V1.3 enabled us to discern the potential role of the SF using a gain-of-function approach by introducing mutations to upregulate VDI. Our results suggest that the SF interface between DIII and DIV may be a previously unrecognized locus for VDI. Although the W[+2] residue is symmetrically present in all four domains, only the DIV W[+2]A mutation increased local conformational flexibility and enhanced VDI. Further mutagenesis of this residue showed that the kinetics of VDI varied depending on the bulkiness of the residue at the DIV [+2] position, with bulkier residues slowing inactivation kinetics. This asymmetric effect of the DIV mutation reflects distinct differences in residues at the DIII–DIV interdomain interface (Fig. 7 A). Specifically, the W[+2] residue interacts with a T[−2] residue that is present in all four domains. Beyond this, the DI–DII, DII–DII, and DIV–DI interdomain interfaces are also stabilized by additional interactions between either a Q[−5] or R[−5] and a Y[+7] residue from the neighboring domain. However, this additional stabilizing interaction is absent in the DIII–DIV interface, with a T[−5] in the DIII and an L[+7] in the DIV locations. Consequently, the DIV W[+2]A mutation destabilizes the SF conformation. Consistent with this possibility, the DIV W[+2]A/DIII T[−5]Q double mutation partially reversed the increase in VDI. Taken together, the interdomain interaction of DIV W[+2] with neighboring residues may be envisioned as a brake that stabilizes the SF conformation allowing sustained ion influx during prolonged depolarization.

For Ca_V1.3, mutating residues in the distal S6 domain below the bundle crossing have also been shown to enhance VDI (Tadross et al., 2010). However, this form of VDI exhibited a conventional Boltzmann-shaped voltage dependence and its magnitude was susceptible to further modulation by Ca_Vβ subunits. By contrast, VDI uncovered by SF mutations exhibited an atypical U-shaped voltage dependence and was insensitive to modulation by the Ca_Vβ subunits. Mechanistically, this feature of SF-dependent inactivation may reflect its dependence on pore occupancy by the permeant ion, which may be reduced at higher voltages owing to a reduced driving force. Nonetheless, the disparate functional signatures suggest that the two forms of VDI are distinct. Interestingly, Ca_V2.2 channels have been shown to also undergo preferential intermediate closed-state inactivation with a similar U-shaped voltage dependence (Patil et al., 1998). Recent cryo-EM studies identified a cytosolic helical segment known as W-helix unique to Ca_V2 DII–DIII linker that occludes the ion conduction pathway (Dong et al., 2021; Gao et al., 2021). Either deletion or disruption of this domain by alternative splicing was shown to diminish preferential intermediate closed-state inactivation (Dong et al., 2021). It is possible that altered conformational changes in the SF of Ca_V2.2 may also contribute to this process.

CDI is initiated by CaM-dependent conformational changes in the cytosolic domain of Ca_V channels (Budde et al., 2002; Halling et al., 2005; Minor and Findeisen, 2010; Ben Johny and Yue, 2015). Ca²⁺ binding to the individual N- and C-lobes of CaM can trigger distinct forms of channel regulation (Lee et al., 2000; DeMaria et al., 2001; Liang et al., 2003), a phenomenon known as “functional bipartition” of CaM (Kink et al., 1990; Saimi and Kung, 1994). For Ca_V1.3, the N- and C-lobe of CaM evoke kinetically distinct forms of inactivation (Yang et al., 2006; Dick et al., 2008). Molecularly, these effects are initiated by a Ca²⁺-dependent switching of CaM interaction with distinct channel interfaces–the Ca²⁺-bound N-lobe engages the NSCaTE domain on the channel amino-terminus (Dick et al., 2008; Simms et al., 2013), while the C-lobe associates with the C-terminus and evokes a conformational change (Kim et al., 2004; Black et al., 2005; Van Petegem et al., 2005; Asmara et al., 2010; Ben-Johny et al., 2013; Banerjee et al., 2018). How these disparate changes in the cytosolic domains are coupled to the pore domain is not fully understood. Two end-stage mechanisms of CDI have been proposed: (1) an allosteric change in the activation gate corresponding to a change in modal gating (Tadross et al., 2010) or (2) an altered selectivity filter conformation (Abderemane-Ali et al., 2019). First, single-channel recordings show that Ca²⁺/CaM regulation switches channels from a high-activity gating with “flickery” openings to a low-activity gating mode with “sparse” openings (Imredy and Yue, 1994). Furthermore, both engineered and human disease-linked mutations in the S6 gate that enhance channel activation proportionately weaken CDI, all consistent with an allosteric mechanism (Tadross et al., 2010). Second, Ca_V1.2 inactivation is sensitive to extracellular Ca²⁺ and CDI alters the propensity for pore blockade by Gd³⁺, suggesting an involvement of the SF (Babich et al., 2005, 2007). More concretely, mutations in the SF alter CDI. Specifically, SF DII D[+1]A diminishes CDI of various Ca_V1/2 channels without any apparent effect on VDI (Abderemane-Ali et al., 2019). Our present findings further corroborate the essential role of the SF in mediating CDI and uncover an unexpected asymmetry in this process. The DIV W[+2]A mutation that destabilizes the SF, evokes contrasting effects on CDI mediated by distinct CaM lobes (Fig. 7 B). While this mutation enhances N-lobe CDI, it also suppresses C-lobe CDI. By contrast, mutating corresponding W[+2] residues in DI–DIII revealed a selective effect on N-lobe CDI but not C-lobe CDI. This divergent effect suggests that the specific interdomain interactions of the W[+2] residue in the SF may be involved in the transduction of cytosolic conformational changes to the end-stage mechanism. Of note, MD simulations did not show a marked increase in conformational flexibility with DI–DIII W[+2]A mutations. However, these simulations did not include the channel C- and N-terminal domains, which are essential for CaM regulation. It is possible that CaM rearrangement in the cytosolic domains may evoke additional changes in the SF domain. These findings also suggest that there may be distinct SF conformational changes associated with N- versus C-lobe CDI. How does one reconcile an allosteric mechanism evident from single-channel studies with a change in SF conformation? One possibility emerges from studies of Kcsa channels where altered SF conformation has been shown to result in shifts in modal gating (Chakrapani et al., 2011). The activation gate of these channels is also coupled to the SF gate (Cuello et al., 2010; Heer et al., 2017), suggesting that the two mechanisms may overlap. Similar findings have also been reported with other ion channels (Boiteux et al., 2014; Kopec et al., 2019; Coonen et al., 2020).

From a pathophysiological perspective, channelopathic mutations in both Ca_V1.2 and Ca_V1.3 are known to reduce both VDI and CDI (Dick et al., 2016; Ortner et al., 2020; Bamgboye et al., 2022a). This deficit in inactivation is thought to result in sustained Ca²⁺ influx that is proarrhythmic. From a pharmacological perspective, traditional small molecule blockers of Ca_V1 channels have been largely inadequate as a therapy. First, mutant channels often exhibit a reduced affinity for many small molecule blockers (Bamgboye et al., 2022b). Second, small molecule Ca_V1 inhibitors impact the peak Ca²⁺ current, which in turn inhibits the overall Ca²⁺ influx causing adverse negative inotropic effects (Angelini et al., 2021). Instead, Ca_V1 gating modifiers that enhance inactivation have been proposed as a new and potentially more effective class of antiarrhythmic drugs (Angelini et al., 2021). In this study, we found that the introduction of the DIV W[+2]A mutation can in fact enhance the VDI of TS-linked mutant channels and restore overall inactivation to near wild-type levels. This increase in VDI points to the possibility that there may be a coupling between S6 gate and the SF region. Of note, we did not observe a significant reduction in overall CDI here with the SF mutation, as these experiments were conducted with endogenous CaM with both lobes intact. This is consistent with results in Ca_V1.3 where N-lobe CDI is enhanced even as C-lobe CDI is reduced resulting in only a modest reduction in CDI in the presence of endogenous CaM. Altogether, this finding suggests that either destabilizing or increasing the conformational flexibility of the DIV SF domain may suffice to upregulate Ca_V1 inactivation and potentially reverse some of the pathophysiology linked to TS.

Of broader relevance, the W[+2] residue is highly conserved across eukaryotic Ca_V, Na_V, and NALCN leak channels, as well as prokaryotic BacNa_V, suggesting that asymmetric SF conformational changes may constitute an ancient and general mechanism for channel inactivation (Payandeh and Minor, 2015). For Na_V1.4, the DI W[+2]C mutation was shown to inhibit slow inactivation while mutations of analogous residues in DII–DIV had minimal effect (Balser et al., 1996). In like manner, mutating the DIV A[0] residue that forms part of the SF of Na_V1.4 introduces an ultraslow component of inactivation with a U-shaped voltage-dependence, much like the Ca_V1.3 DIV W[+2]A mutation (Todt et al., 1999). For the homotetrameric bacterial Na_V channels, asymmetric pore collapse is thought to be the underlying structural mechanism for inactivation (Pavlov et al., 2005; Payandeh et al., 2012). The W[+2] residue serves as an important anchor for the SF that allows hydrogen bonding interactions with T[−2] residues of neighboring subunits (Payandeh and Minor, 2015). This interaction has been proposed to be important to allow SF conformational changes to be influenced by neighboring subunits (Payandeh and Minor, 2015). Structurally, the slow inactivation of BacNa_V channels increases the conformational flexibility of the selectivity filter, resulting in a reorientation of the pore helices that change the intersubunit cavity volumes and accessibility of lipids (Chatterjee et al., 2018).

Overall, this study furnishes new insights into the role of the SF in VDI and CDI of Ca_V1.3 and it uncovers how asymmetric changes in the SF allow structural bifurcation of CaM signaling, a scheme that may be broadly relevant (Saimi and Kung, 2002; Liang et al., 2003; Ben-Johny et al., 2014).

All data are available in the article itself or in the supplementary materials.

Crina M. Nimigean served as editor.

We thank Dr. Ryan Mahling, Dr. Nourdine Chakouri, Dr. Ivy Dick, Dr. Po Wei Kang, and Dr. Filip van Petegem for their helpful feedback.

This study is supported by funding from the National Institute of Neurological Disorders and Stroke (R01 NS110672) to M. Ben-Johny, American Heart Association Predoctoral Fellowship (award ID 835091) to P.J. del Rivero Morfin, a grant from the Austrian Science Fund (FWF) P35618 to B.E. Flucher, and Austrian Academy of sciences APART-MINT postdoctoral fellowship to M.L.I. Fernández-Quintero and the Austrian Science fund (P34514). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the American Heart Association. The computational results presented here have been achieved in part using the Vienna Scientific Cluster (VSC). We thank PRACE for awarding us access to Piz Daint at CSCS, Switzerland.

Author contributions: P.J. del Rivero Morfin, M.L.I. Fernández-Quintero, K.R. Liedl, B.E. Flucher, and M. Ben-Johny designed research, P.J. del Rivero Morfin, M.L.I. Fernández-Quintero, A.L. Kochiss, and M. Ben-Johny performed research, acquired and analyzed data, P.J. del Rivero Morfin, M.L.I. Fernández-Quintero, K.R. Liedl, B.E. Flucher, and M. Ben-Johny contributed new reagents/analytic tools; M. Ben-Johny, M.L.I. Fernández-Quintero, and B.E. Flucher funding acquisition; P.J. del Rivero Morfin, M.L.I. Fernández-Quintero, B.E. Flucher, and M. Ben-Johny made figures and wrote the original draft; and all authors revised the manuscript.

All data are available in the article itself or in the supplementary materials.