Mapping the interaction surface between CaVβ and actin and its role in calcium channel clearance

Abstract

Defective ion channel turnover and clearance of damaged proteins are associated with aging and neurodegeneration. The L-type Ca_V1.2 voltage-gated calcium channel mediates depolarization-induced calcium signals in heart and brain. Here, we determined the interaction surface between actin and two calcium channel subunits, Ca_Vβ₂ and Ca_Vβ₄, using cross-linking mass spectrometry and protein-protein docking, and uncovered a role in replenishing conduction-defective Ca_V1.2 channels. Computational and in vitro mutagenesis identified hotspots in Ca_Vβ that decreased the affinity for actin but not for Ca_V1.2. When coexpressed with Ca_V1.2, none of the tested actin-association-deficient Ca_Vβ mutants altered the single-channel properties or the total number of channels at the cell surface. However, coexpression with the Ca_Vβ₂ hotspot mutant downregulated current amplitudes, and with a concomitant reduction in the number of functionally available channels, indicating that current inhibition resulted from a build-up of conduction silent channels. Our findings established Ca_Vβ₂-actin interaction as a key player for clearing the plasma membrane of corrupted Ca_V1.2 proteins to ensure the maintenance of a functional pool of channels and proper calcium signal transduction. The Ca_Vβ-actin molecular model introduces a potentially druggable protein-protein interface to intervene Ca_V-mediated signaling processes.

Fig. 1. Workflow diagram illustrating the strategy used to identify the interaction surface between Ca_Vβ and actin by combining cross-linking mass spectrometry (XL-MS) with protein-protein docking and experimental validation.

The overall strategy can be divided in four phases; a pre-phase, named protein preparation, aimed to generate the proteins under study, and XL-MS, modeling and experimental and functional validation phases. The optimization of the XL-reaction conditions was done by testing different amounts of the XL (DSBU and DSSO) at different reaction times, as shown in Supplementary Fig. 1, and visualized by SDS-PAGE (1). Replicas of the optimal XL-reaction conditions were carried out for further analysis (2). The band containing the specific cross-linked product between Ca_Vβ and actin was excised and in gel trypsinized (3). The mix of peptides were subjected to C18 stage tip cleaning (4) and LC-MS/MS (5). XL-MS data analysis was performed using three different search engines (6). Unique intermolecular cross-links that were identified by at least two search engines (7) were considered for filtering using DisVis (8) to obtain a set of self-consistent XL-MS derived distance restraints (9). XL-MS restraints together with bioinformatics-predicted restraints from NACCESS and CPORT programs, considered as unambiguous and ambiguous restraints, respectively (10), served as input files for the HADDOCK hybrid docking protocol (11) to generate Ca_Vβ–actin models (12). The top four HADDOCK models of the selected docking cluster were used to predict hotspots via Computational Alanine Scanning (CAS) with six predictors (Anchor, BeAtMuSiC, Bude Alanine Scan, Mutabind2, Robetta and SAAMBE-3D) (13). Hotspot mutations selected based on the average ΔΔG over the six predictors were introduced by site-directed mutagenesis into Ca_Vβ (14). The impact of the Ca_Vβ hotspot mutants on its binding affinity for actin were assessed by in vitro assays (15), and the mutants with decreased affinity were tested for its effect on calcium channel function in living cells by electrophysiology (16).

Fig. 2. Chemical XL-MS and computational analysis define a set of distance constraints for modeling the Ca_Vβ–actin complex.

A Domain architecture (top) and crystal structure (bottom) of rat Ca_Vβ₂ (UniProtKB Q8VGC3-2; PDB 5V2P). Src homology 3 (SH3) and guanylate kinase (GK) domains forming the protein core are labeled. Numbers represent the first and last residues. B Size exclusion chromatography profile of the Ca_Vβ₂-core; inset, purified protein resolved by SDS-PAGE. Numbers denote the size of molecular weight (MW) standards. The experiments were repeated at least 3 times. C Representative SDS-PAGE images of optimal cross-linking reactions for the Ca_Vβ₂-core and actin using DSSO, DSBU, and control reaction with only vehicle, no cross-linker (Non-XL) from 5, 4 and 3 independent replicas, respectively. DSSO: lane 1, MW standards; lane 2, Ca_Vβ₂-core–actin cross-linking; lane 3, actin alone incubated with DSSO; and lane 4, Ca_Vβ₂-core alone incubated with DSSO. DSBU: lane 1, Ca_Vβ₂-core–actin cross-linking; lane 2, actin alone incubated with DSBU; and lane 3, Ca_Vβ₂-core alone incubated with DSBU. MW standards lane was omitted and the full image is shown in Supplementary Fig. 1. The red arrow and asterisk denote the band that was excised from the gel for tryptic digestion and LC-MS/MS analysis. Non-XL control: lane 1, MW standards; lane 2, Ca_Vβ₂-core and actin together incubated with DMSO; lane 3, actin alone incubated with DMSO; and lane 4, Ca_Vβ₂-core alone incubated with DMSO. D Linear representation depicting the location and residue number of unique intermolecular cross-links between Ca_Vβ₂-core and actin (UniProtKB P68135, PDB 5OOE). E Three-dimensional ribbon representation of two adjacent actin monomers in one protofilament, named n and n + 2, shown in yellow and pink, respectively. Actin amino acids participating in the unique intermolecular cross-links (shown as spheres, with those in monomer n + 2 labeled with a prime (ʹ) symbol) form a contiguous patch across two adjacent actin subunits (black contour). F Structural mapping of inter-cross-links (red lines) that satisfied data consistency upon DisVis filtering for one Ca_Vβ₂ molecule and two adjacent actin monomers. Actin is shown in space-filling and ribbon mode, but for clarity Ca_Vβ₂ is displayed in ribbon mode only.

Fig. 3. Model of Ca_Vβ₂ bound to actin shows an extended contact surface and electrostatic complementarity.

A Space-filling and ribbon diagram showing two views related by a 90° rotation of the best-scoring docking model of Ca_Vβ₂ complexed with actin. One molecule of Ca_Vβ₂ (blue) interacts with two adjacent monomers in one actin protofilament, named n (yellow) and n + 2 (pink). Actin subunits are labeled based on their position along the filament from the pointed end (−) to the barbed end (+). For clarity, Ca_Vβ₂ is shown in ribbon mode only in the right panel. The four actin subdomains, SD1–SD4, are labeled in the n monomer. Subdomains in monomer n + 2 are indicated with a prime (ʹ); SD2ʹ and SD3ʹ were omitted. B Surface models showing the contact frequency of Ca_Vβ₂ and actin residues. C Same as in B but showing the electrostatic potential map of both proteins. D SDS-PAGE analysis of a representative F-actin cosedimentation assay using the indicated NaCl concentrations. Briefly, the Ca_Vβ₂-core was incubated with phalloidin-stabilized actin filaments, the mixture was centrifuged, and the supernatant (S) and pellet (P) fractions separated. The pellet was resuspended in the same volume of SDS-loading buffer as the volume of supernatant, and the supernatant and pellet fractions were resolved by denaturing SDS-PAGE. Control assays included the Ca_Vβ₂-core only (lower panel). For clarity, irrelevant lanes were cropped from the source images (as indicated by the white space); full gel images are shown in Supplementary Fig. 8. E Fraction of Ca_Vβ₂-core bound to actin (calculated by densitometry) against NaCl concentration. Box plot shows the mean (dashed line), median (continuous line), interquartile range (25th–75th percentiles, box edges) and whiskers (1.5× interquartile range). Each dot represents an independent experiment; p-values, two-tailed t-test. Each experiment was repeated three times. Also see Supplementary Fig. 8.

Fig. 4. Alanine substitution of predicted PPI hotspots decreases the in vitro affinity of Ca_Vβ₂ for actin but not for the AID peptide.

A Structure of Ca_Vβ₂ (blue ribbon), indicating the eight predicted hotspot residues (red spheres) mutated to alanine in the Ca_Vβ₂-core. Two actin subunits interacting with Ca_Vβ₂ are shown in space-filling mode, oriented from the barbed end (n + 2, pink) to the pointed end (n, yellow). B Linear diagram of the Ca_Vβ₂-core domain, indicating the positions of hotspot mutations and size exclusion chromatography profiles of the hotspot mutants, Ca_Vβ₂-core 6Ala (orange) and Ca_Vβ₂-core 8Ala (green). The arrow denotes the elution volume for Ca_Vβ₂-core WT. C SDS-PAGE of representative F-actin cosedimentation assays for the Ca_Vβ₂-core WT (lanes 2–5), Ca_Vβ₂-core 6Ala (lanes 6–9), and Ca_Vβ₂-core 8Ala (lanes 10–13). MW standard (lane 1). Ca_Vβ₂-core-derived proteins were centrifuged either alone (control) or after incubation with polymerized actin (actin). S, supernatant; P, pellet. Asterisks denote bands of Ca_Vβ₂-core-derived proteins that pelleted with actin. Each experiment was repeated three times; full images are shown in Supplementary Fig. 11A. D Fraction of the indicated Ca_Vβ₂-core proteins bound to actin. Box plot shows the mean (dashed line), median (continuous line), interquartile range (25th–75th percentiles, box edges) and whiskers (1.5× interquartile range). Each dot represents an independent experiment; p-values, unpaired two-tailed t-test. E Ca_Vβ₂ complexed with the AID peptide (PDB 5V2P). Ca_Vβ₂ is depicted as a blue ribbon with hotspots for actin binding as red van der Waals spheres and the critical residues for Ca_Vβ association (L352, I343, and M245)^, in purple. The AID is shown as a pink ribbon. F SDS-PAGE of a representative pull down assay using as bait GST, either alone or fused to AID (GST-AID) and as prey Ca_Vβ₂-core WT (WT), Ca_Vβ₂-core 6Ala (6Ala), or Ca_Vβ₂-core 8Ala (8Ala). MW standards (lane 1), Ca_Vβ₂-core proteins used as input (lanes 2–4), elution fractions from the GST (control, lanes 5, 7, and 9) and GST-AID pull down assays (lanes 6, 8, and 10). Each experiment was repeated three times; full images are shown in Supplementary Fig. 11C.

Fig. 5. Ca_Vβ₄ R482X hotspot mutations blunt association with actin.

A Superimposition of the actin-bound structures of Ca_Vβ₂ (cyan ribbon) and Ca_Vβ₄ (purple ribbon). Each Ca_Vβ structure represents the average of the four best-scoring models of the selected docking cluster. For clarity, two actin subunits used as a reference for the alignment are shown in space-filling mode (yellow and pink). B Linear diagram and comparison of the in vitro actin-binding affinities by F-actin cosedimentation assay for Ca_vβ₄-core (lanes 2–5) and Ca_vβ₄ R482X (lanes 6–9). Numbers denote the positions of predicted hotspots substituted by alanine in the Ca_Vβ₄ R482X background according to human Ca_Vβ₄ (UniProtKB O00305-1). S and P, supernatant and pellet fractions, respectively. Single and double black arrows indicate the amount of corresponding protein that cosedimented with actin. The assays were repeated at least 3 times (Supplementary Fig. 15A). C Size exclusion chromatography profiles for Ca_Vβ₄ R482X (carrying no Ala substitutions) (black), Ca_Vβ₄ R482X 6Ala (orange), and Ca_Vβ₄ R482X 8Ala (green). D SDS-PAGE of F-actin cosedimentation assays for Ca_Vβ₄ R482X (lanes 2–5), Ca_Vβ₄ R482X 6Ala (lanes 6–9), and Ca_Vβ₄ R482X 8Ala (lanes 10–13). S, supernatant; P, pellet. Asterisks indicate protein bands that pelleted with actin. Each experiment was repeated three times; full images are shown in Supplementary Fig. 15C. E Fraction of the indicated Ca_Vβ₄ R482X proteins bound to actin. In box plot: mean (dashed line), median (continuous line), interquartile range (25th–75th percentiles, box edges), and whiskers (1.5× interquartile range). Each dot represents an independent experiment; p-values, unpaired two-tailed t-test. F SDS-PAGE of a representative pull-down assay using as bait GST, either alone or fused to AID (GST-AID), and as prey Ca_Vβ₄ R482X (–), Ca_Vβ₄ R482X 6Ala (6Ala), or Ca_Vβ₄ R482X 8Ala (8Ala). Molecular weight standard (lane 1), Ca_Vβ₄ R482X-derived proteins used as input (lanes 2–4), elution fractions from the pull down assay using either GST (lanes 5, 7, and 9) as control or GST-AID (lanes 6, 8, and 10) and the indicated proteins. Each experiment was repeated three times; full images are shown in Supplementary Fig. 15D.

Fig. 6. An actin-association-deficient Ca_Vβ₂ mutant decreases Ca_V1.2-mediated ionic current but not gating currents.

A Schematic of the membrane-associated Ca_Vβ₂ fused to mRFP (Ca_Vβ₂-mRFP) and laser scanning confocal images of HEK293 cells expressing either the wild-type full-length Ca_Vβ₂ (WT) or the eight hotspot mutant (8Ala). Fluorescence images for the indicated Ca_Vβ₂ (a;d, magenta) and the plasma membrane marker CellMask^TM (b;e, green); overlapping pixels appear as white (c;f). Scale bar, 15 µm (for all images). B Colocalization analysis between the indicated Ca_Vβ₂ protein and CellMask^TM. Box plot shows the mean (dashed line), median (continuous line), interquartile range (25th–75th percentiles, box edges) and whiskers (1.5 × interquartilee range). Each dot represents the Mander’s overlap coefficient (MOC) for a field of view containing between 150 and 200 cells; p-value, two-tailed t-test. C Western blot (WB) of crude lysates from cells expressing either Ca_Vβ₂ WT or Ca_Vβ₂ 8Ala. Lysates from three separate experiments were probed with anti-Ca_Vβ₂ and anti-GAPDH antibodies. D Representative ionic current traces and plots of ionic current versus voltage (I/V) and fraction of activated channels versus voltage obtained from HEK293 cells cotransfected with Ca_V1.2 and either Ca_Vβ₂ WT or Ca_Vβ₂ 8Ala. Only the traces induced by −40, +15, and +40 mV pulses are shown. Currents were elicited by voltages of −50 to +50 mV in 5 mV increments from a holding potential of −90 mV; Data are presented as mean ± SEM; number of recorded cells = 32, 29 and 5 for Ca_V1.2/Ca_Vβ₂ WT, Ca_V1.2/Ca_Vβ₂ 8Ala, and Ca_V1.2 alone, respectively. E Representative gating currents obtained from cells in panel (D) and total charge movement (Qon) calculated from the integral of the On gating current (shaded area) during the voltage step to the reversal potential for the carrier ion, Ba²⁺. In box plot: mean (dashed line), median (continuous line), interquartile range (25th–75th percentiles, box edges) and whiskers (1.5 × interquartile range). Each dot represents an individual recorded cell; p-values, unpaired two-tailed t-test. Mean values ± S.E.M: 363 ± 33 fC (n = 32) and 304 ± 28 fC (n = 29) for Ca_V1.2/Ca_Vβ₂ WT and Ca_V1.2/Ca_Vβ₂ 8Ala, respectively.

Fig. 7. The Ca_Vβ₂ actin association-deficient mutant reduces the number of functional Ca_V1.2 channels and does not alter the channel conductance or maximal open probability.

A Representative ionic current traces, with red lines delineating the time window used for noise analysis and plotting the relationship between variance and mean macroscopic current $\left(\frac{{\sigma }^2}{⟨I⟩\left(Vm-Vrev\right)}\right)$ versus $\left(\frac{⟨I⟩}{Vm-Vrev}\right)$ obtained at different voltages for two representative cells expressing Ca_V1.2 complexed with either Ca_Vβ₂ WT or Ca_Vβ₂ 8Ala. The number of functional channels n and the unitary conductance γ were calculated from the slope and y-intercept of the linear fit, respectively (Eq. 6). B Box plots for n, γ, and Po_max. The box edges represent interquartile ranges (25th–75th percentiles), the continuous and dashed lines the median and mean values and the ranges of the whiskers denote 1.5× the interquartile range. Each data point represents an individual recorded cell (number of cells = 18 and 17 for Ca_V1.2/Ca_Vβ₂ WT and Ca_V1.2/Ca_Vβ₂ 8Ala, respectively). P-values were determined by unpaired two-tailed t-test. The smaller unitary conductance value for Ca_V1.2 reported here is explained by the absence of Ca_V agonists in our whole-cell current recordings. C Bootstrap sample distributions for n, γ, and Po_max (bootstrap sample N = 500,000). The mean values for n, γ, and Po_max obtained from the corresponding bootstrap distributions are; n, 12233 ± 1219 for Ca_V1.2/Ca_Vβ₂ WT and 6580 ± 899 for Ca_V1.2/Ca_Vβ₂ 8Ala; γ, 3.9 ± 0.1 pS for Ca_V1.2/Ca_Vβ₂ WT and 4.0 ± 0_.1 pS for Ca_V1.2/Ca_Vβ₂ 8Ala; and Po_max, 0.59 ± 0.02 for Ca_V1.2/Ca_Vβ₂ WT and 0.60 ± 0.02 for Ca_V1.2/Ca_Vβ₂ 8Ala; p-values were determined by bootstrap t-test.

Fig. 8. Ca_Vβ₄ R482X hotspots mutant keeps Ca_V1.2-mediated ionic current intact.

A Representative laser scanning confocal images of HEK293 cells expressing either Ca_Vβ₄ R482X (a;c) or Ca_Vβ₄ R482X 8Ala (d;f) fused to mCherry. Plasma membrane was stained with CellMask^TM (b;e, green). Merged images (c;f). Scale bar, 15 µm (for all images). B Western blot of three separate crude lysates from cells expressing either Ca_Vβ₄ R482X (--) or the 8Ala mutant (8Ala) probed with anti-Ca_Vβ₄ and anti-GAPDH antibodies. Uncropped images are shown in Supplementary Fig. 18A. C Representative ionic current traces (induced by −35, +20, and +45 mV pulses) from cells co-expressing Ca_V1.2 with either Ca_Vβ₄ R482X or Ca_Vβ₄ R482X 8Ala. D Plots of ionic current and fraction of activated channels versus voltage, and total charge movement (Qon) from the indicated channel subunit combinations. Ionic currents were elicited by voltages of −50 to +50 mV in 5 mV increments from a holding potential of −90 mV; data are presented as mean ± SEM. In the box plot, the edges represent interquartile ranges (25th–75th percentiles), the continuous and dashed lines, the median and mean values and whiskers denote 1.5× the interquartile range. Each data point represents an individual recorded cell. Qon mean values ± SEM: 234 ± 24 fC and 259 ± 37 fC (number of cells = 24 and 20) for Ca_V1.2/Ca_Vβ₄ R482X and Ca_V1.2/Ca_Vβ₄ R482X 8Ala, respectively. E Box plots for n, γ, and Po_max for the indicated channel subunit combinations with interquartile ranges (25th–75th percentiles), median (continuous line), mean (dashed lines) and whiskers (1.5× the interquartile range). Each individual data point represents an individual recorded cell (number of cells = 13 and 15 for Ca_V1.2/Ca_Vβ₄ R482X and Ca_V1.2/Ca_Vβ₄ R482X 8Ala, respectively); p-values, unpaired two-tailed t-test. F Bootstrap distributions from the data shown in panel (E). Bootstrap sample N = 500,000. The mean values for n, γ, and Po_max are: n, 2901 ± 509; γ, 3.4 ± 0.1 pS and Po_max, 0.61 ± 0.02 for Ca_V1.2/Ca_Vβ₄ R482X, and n, 3425 ± 437; 3.8 ± 0.3 pS and 0.63 ± 0.03 for Ca_V1.2/Ca_Vβ₄ R482X 8Ala; p-values, bootstrap t-test.

Fig. 9. Model for the clearance of corrupted Ca_V1.2 channels from the plasma membrane.

Conformationally impaired Ca_V1.2 proteins at the plasma membrane can be generated by diverse cellular stimuli, which change the channel environment leading to local unfolding, or by exocytic insertion of misfolded channels that evade cytosolic quality control (biosynthetic misfolding). Ca_Vβ–actin interaction is mandatory for the endocytic removal of corrupted Ca_V1.2 channels, which prevents their build-up and maintains stable Ca_V1.2 currents, but not for the insertion of functionally folded channels. PM, plasma membrane.