The guanylate kinase domain of the β-subunit of voltage-gated calcium channels suffices to modulate gating

Abstract

Inactivation of voltage-gated calcium channels is crucial for the spatiotemporal coordination of calcium signals and prevention of toxic calcium buildup. Only one member of the highly conserved family of calcium channel β-subunits—Ca_Vβ—inhibits inactivation. This unique property has been attributed to short variable regions of the protein; however, here we report that this inhibition actually is conferred by a conserved guanylate kinase (GK) domain and, moreover, that this domain alone recapitulates Ca_Vβ-mediated modulation of channel activation. We expressed and refolded the GK domain of Ca_Vβ_2a, the unique variant that inhibits inactivation, and of Ca_Vβ_1b, an isoform that facilitates it. The refolded domains of both Ca_Vβ variants were found to inhibit inactivation of Ca_V2.3 channels expressed in Xenopus laevis oocytes. These findings suggest that the GK domain endows calcium channels with a brake restraining voltage-dependent inactivation, and thus facilitation of inactivation by full-length Ca_Vβ requires additional structural determinants to antagonize the GK effect. We found that Ca_Vβ can switch the inactivation phenotype conferred to Ca_V2.3 from slow to fast after posttranslational modifications during channel biogenesis. Our findings provide a framework within which to understand the modulation of inactivation and a new functional map of Ca_Vβ in which the GK domain regulates channel gating and the other conserved domain (Src homology 3) may couple calcium channels to other signaling pathways.

Calcium signals mediate various cellular processes, including neurotransmission, excitation-contraction coupling, hormone secretion, and gene expression (1). Voltage-gated calcium channels (VGCCs) are activated and inactivated on membrane depolarization, allowing transient increases in cytosolic Ca²⁺ concentration. Voltage-dependent activation and inactivation of VGCCs depend strongly on the association of the ancillary β-subunit (Ca_Vβ) to a highly conserved sequence within the intracellular loop joining the first and second repeats (loop I-II) of the pore-forming subunit (Ca_Vα₁), known as the α-interaction domain (AID) (2). Ca_Vβ is encoded by four nonallelic genes (β_1–4), each with multiple splice variants. Except for a few short splicing forms (3), all of these genes share a common structural arrangement, consisting of two highly conserved regions separated and flanked by shorter variable sequences (Fig. 1A). Crystallographic studies have revealed that whereas the first region encompasses a Src homology 3 (SH3) domain, the second region encompasses a guanylate kinase (GK) domain. The AID sequence forms an α-helix that fits into a hydrophobic cleft of the GK module that lies on the opposite side of the SH3 domain (Fig. 1B) (4–6). This suggests that the isolated GK module may preserve at least some of the modulatory capabilities of the full Ca_Vβ.

Fig. 1.

Domain structure, purification, and binding assay of Ca_Vβ constructs. (A) Schematic representation of Ca_Vβ_2a, Ca_Vβ_1b, and derived protein constructs used in this study. Ca_Vβ consists of two highly conserved regions, D1 and D2 (boxes), which are connected and flanked by variable regions (continuous lines). The gray box is the SH3 module; the black box is the GK module. The two cysteine residues at the N terminus of Ca_Vβ_2a that undergo palmitoylation are indicated by arrows. (B) A ribbon diagram of the crystal structure of Ca_Vβ in complex with AID (PDB accession code 1T3L). (C) Size-exclusion chromatography elution profile on the Superdex 200 10/30 column (GE Healthcare) of refolded Ca_Vβ_2a-GK and Ca_Vβ_1b-GK. Here 1 indicates the void volume, 2 indicates the elution volume of albumin (67 kDa), and 3 indicates the elution volume of ovalbumin (43 kDa). The inset shows Coomassie-stained sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS/PAGE) gels of the indicated proteins. Numbers indicate the molecular mass of standards, in kDa. (D) SDS/PAGE gel of the binding reaction with the indicated proteins. Yellow fluorescence protein (YFP)-Ca_Vβ_2a-GK is seen to bind specifically to GST–loop I-II (last lane). The binding assay was repeated three times.

Despite several attempts to use different Ca_Vβ constructs and experimental approaches (7–11), the functional competence of isolated GK and its ability to mimic Ca_Vβ function remain incompletely understood. A contributing factor may be the reduced stability of some GK-containing constructs (11). To overcome this difficulty, we expressed and refolded the GK domain of two Ca_Vβ isoforms, Ca_Vβ_1b and Ca_Vβ_2a. These isoforms share modulatory effects on voltage-dependent activation and exhibit opposite actions on voltage-dependent inactivation. We studied the effect of the refolded GK modules on Xenopus oocytes expressing two types of α₁ pore-forming subunits (Ca_V1.2 and Ca_V2.3) and compared it with the action of the recombinant full length and also the core of the Ca_Vβ protein containing both SH3 and GK domains. Whereas Ca_Vβ_2a is unique in its ability to inhibit voltage-dependent inactivation, the other Ca_Vβ isoforms facilitate this inactivation (12–17). Ca_Vβ_2a decelerates inactivation, increases the fraction of noninactivating current, and shifts the steady-state inactivation curve toward more positive potentials. These distinguishing modulatory properties of Ca_Vβ_2a have been broadly attributed to palmitoylation of the two contiguous cysteine residues at positions 3 and 4 in the N terminus region (15, 18–20). Here, however, we report that instead, the GK modules derived from both Ca_Vβ_2a and Ca_Vβ_1b inhibit inactivation of Ca_V2.3 channels. This finding indicates that the structural determinants of inhibition of inactivation by Ca_Vβ_2a are encoded not in variable regions but rather within the GK domain. GK appears to endow calcium channels with a brake to impair voltage-dependent inactivation; therefore, masking the inhibitory effect of GK facilitates inactivation. We show that Ca_Vβ acquires this capability when co-expressed with Ca_Vα₁ but not when added later during channel biogenesis. Moreover, Ca_Vβ_2a-GK increases peak currents and shifts the activation curve toward more negative potentials of Ca_V1.2 channels. Thus, GK emerges as a functional unit that recapitulates the hallmarks of Ca_Vβ modulation.

Results

Refolding and Binding Assay of Ca_vβ-GK Domain.

The GK domain derived from Ca_Vβ_1b (Ca_Vβ_1b-GK) and Ca_Vβ_2a (Ca_Vβ_2a-GK) (Fig. 1A) were expressed in bacteria, where they accumulated in inclusion bodies and were refolded by batch dilution. The purified GK domains were concentrated up to 0.1 mg/ml, because further concentration resulted in progressive protein aggregation, as demonstrated by high–molecular mass peaks appearing in the void volume of the size-exclusion chromato graphy column. At 0.1 mg/ml, Ca_Vβ_2a-GK eluted from the size-exclusion chromatography in a predominant peak, whereas Ca_Vβ_1b-GK exhibited a shoulder at this position and a main peak eluting near the albumin peak (Fig. 1C). Both GK constructs migrated with an apparent molecular mass larger than that predicted by their amino acid sequences. This may reflect either an aberrant migration or formation of higher oligomeric states. To determine whether the refolded GK domains form multimers, we conducted a sucrose gradient analysis [supporting information (SI) Figs. S1–S3]. The Ca_Vβ_2a-GK distribution overlapped with a protein standard of 66 kDa, whereas Ca_Vβ_1b-GK appeared over a wider range, reaching a second standard of 29 kDa. Thus, our data are consistent with Ca_Vβ_2a-GK being essentially a dimer and Ca_Vβ_1b-GK being a mixture of dimers and monomers.

To assess the binding ability of Ca_Vβ_2a-GK to loop I-II, we fused the former to YFP (YFP-Ca_Vβ_2a-GK) and the latter to GST (GST–loop I-II). Ca_Vβ_2a-GK was seen to bind specifically to loop I-II; no association between GST and YFP-Ca_Vβ_2a-GK or between YFP and GST–loop I-II was detected (Fig. 1D).

Ca_vβ_2a-GK Increases Peak Current Amplitude and Shifts the Current–Voltage Relationship of Ca_v1.2 Channels.

To investigate the modulation of voltage-dependent activation by Ca_Vβ_2a-GK, we used the Ca_V1.2α₁ subunit, because it exhibits little inactivation, and the coupling between voltage sensor and channel opening is extremely sensitive to the presence of Ca_Vβ. This Ca_Vα₁ isoform also is more suitable for monitoring channel expression through gating current measurements (21). Injection of recombinant full-length Ca_Vβ_2a into oocytes expressing Ca_V1.2 channels results in an increase in the ionic current to charge movement ratio (I/Q) and a leftward shift in the current–voltage relationship (22). Ca_Vβ_2a-SH3-GK proved to be equally robust in modulating the activation of Ca_V1.2 channels (Fig. 2A, Fig. S4, and Table S1). Here, as in a previous report (22), we measured charge movement by integrating outward transient currents at the onset of depolarizing pulses to the ionic current reversal potential. Injection of Ca_Vβ_2a-SH3-GK into Ca_V1.2-expressing oocytes increased the I/Q from 0.3 ± 0.06 nA/pC (n = 15) to 4.6 ± 0.7 nA/pC (n = 17), comparable to the values obtained after injection of Ca_Vβ_2a (4.5 ± 0.9 nA/pC; n = 18). The effect of Ca_Vβ_2a-GK was seen more clearly at reduced channel expression levels, where gating currents were barely visible and thus I/Q plots were not readily measurable. Nevertheless, the normalized current–voltage relationship was shifted to more-negative potentials and to similar extents with Ca_Vβ_2a (as protein or cRNA), Ca_Vβ_2a-SH3-GK, and Ca_Vβ_2a-GK (Fig. 2B).

Fig. 2.

Refolded Ca_Vβ_2a-GK and Ca_Vβ_2a-SH3-GK shift the current–voltage relationship of Ca_V1.2-mediated currents. (A) Representative gating and ionic currents traces from oocytes expressing Ca_V1.2 cRNA alone and after injection of either Ca_Vβ_2a-SH3-GK or Ca_Vβ_2a-GK. Currents were evoked by 50-ms pulses to −30, 0, and +30 mV from a holding potential of −80 mV. (B) Normalized current–voltage plot from oocytes expressing the subunit combinations Ca_V1.2 cRNA (n = 11), Ca_V1.2 + Ca_Vβ_2a-SH3-GK (n = 14), and Ca_V1.2 + Ca_Vβ_2a-GK (n = 16). For comparison, normalized current–voltage curves for Ca_V1.2 + Ca_Vβ_2a, either injected as a protein (dashed line) or co-injected as cRNA (continuous line) are shown (see Table S1 for details).

We attribute the limited activity of refolded Ca_Vβ_2a-GK to the low concentration of this protein, calculated as 0.3 μM for a 525-nl oocyte. In addition, at high expression levels, a large amount of Ca_V1.2 subunit remained in the cytoplasm and may have acted as a sink. Another contributing factor could be the protein's decreased stability (see below). To overcome this potential problem, and inspired by a previous experiment demonstrating that Ca_Vβ_2a recapitulates channel modulation when attached to the C-termini of Ca_Vα₁ (23), we covalently linked Ca_Vβ-GK to Ca_V1.2 (Ca_V1.2-Ca_Vβ_2a-GK) and compared this with Ca_V1.2 linked to Ca_Vβ_2a (Ca_V1.2-Ca_Vβ_2a). Fig. 3 shows gating and ionic currents from oocytes expressing Ca_V1.2-Ca_Vβ_2a or Ca_V1.2-Ca_Vβ_2a-GK. When covalently linked to Ca_V1.2, Ca_Vβ_2a-GK appears to be as efficient as full-length Ca_Vβ_2a in increasing I/Q and shifting the voltage dependence of activation. This effect was abolished by a mutation in the conserved tryptophan within the AID (W470S) shown to prevent binding to Ca_Vβ (22), indicating that the changes in gating properties occurred through specific association of the fused GK moiety to the AID site and not from changes in channel activity generated by the GK linkage.

Fig. 3.

Ca_Vβ_2a-GK covalently linked to Ca_V1.2 WT, but not to Ca_V1.2 W470S, increases peak current amplitudes and shifts the current–voltage relationship. (A) Representative gating and ionic current traces from oocytes expressing Ca_V1.2 WT covalently linked to either Ca_Vβ_2a (Ca_V1.2-Ca_Vβ_2a) or Ca_Vβ_2a-GK (Ca_V1.2-Ca_Vβ_2a-GK), and Ca_V1.2 W470S covalently linked to Ca_Vβ_2a-GK (Ca_V1.2 W470S-Ca_Vβ_2a-GK). Currents were evoked by 50-ms pulses to −30, 0, and +30 mV from a holding potential of −80 mV. (B) Ionic current from oocytes expressing the different constructs were normalized by charge movement (I/Q) and plotted versus voltage. For Ca_V1.2-Ca_Vβ_2a, the peak I/Q was 2.44 ± 0.44 nA/pC (n = 17); for Ca_V1.2-Ca_Vβ_2a-GK, it was 2.71 ± 0.52 nA/pC (n = 17); and for Ca_V1.2 W470S-Ca_Vβ_2a-GK, it was 0.35 ± 0.03 nA/pC (n = 17). For comparison, the average I/Q from 15 oocytes expressing Ca_V1.2 alone are shown as dashed lines (0.30 ± 0.06 nA/pC). (C) Normalized tail currents from oocytes expressing the different constructs. The continuous lines correspond to the fit of the sum of two Boltzmann distributions, and the dashed line corresponds to the fit obtained from Ca_V1.2-expressing oocytes (see Fig. S4 and Table S1 for details). The fit to Ca_V1.2 W470S-Ca_Vβ_2a-GK was excluded from the plot for clarity.

Ca_vβ_2a-GK Inhibits Inactivation of Ca_v2.3 WT Channels.

Next, we investigated the ability of Ca_Vβ_2a-GK to regulate inactivation kinetics, using the fast-inactivating Ca_V2.3 channels (Fig. 4). Injection of refolded Ca_Vβ_2a-GK into Ca_V2.3-expressing oocytes resulted in a sixfold increase in the decay time to half-peak current amplitude (t½; Fig. 4 A and B); however, Ca_Vβ_2a-GK co-injected as cRNA failed to modulate the Ca_V2.3-mediated currents. A greater increase in t½ was observed after injection of full-length Ca_Vβ_2a, either as protein or co-injected as cRNA, and Ca_Vβ_2a-SH3-GK. Consistent with our earlier findings, Ca_Vβ_2a-SH3 decreased ionic currents without changing the time course (24). The effect of refolded Ca_Vβ_2a-GK vanished 5 h after injection, indicating some degree of protein instability (Fig. 4C); this finding also may explain the lack of effect of Ca_Vβ_2a-GK cRNA.

Fig. 4.

Ca_Vβ-GK slows inactivation of Ca_V2.3-mediated currents. (A) Representative current traces from oocytes expressing Ca_V2.3 cRNA alone or after injection of the specified protein during a 10-s pulse to 0 mV from a holding potential of −90 mV. (B) Average decay times to half-peak current amplitude (t½) for the different subunit combinations: Ca_V2.3 cRNA, t½ = 0.33 ± 0.03 s (n = 26); Ca_V2.3 + Ca_Vβ_2a-GK cRNA, t½ = 0.41 ± 0.05 s (n = 13); Ca_V2.3 + Ca_Vβ_2a-GK, t½ = 1.85 ± 0.44 s (n = 13); Ca_V2.3 + Ca_Vβ_2a-SH3-GK, t½ = 3.57 ± 0.42 s (n = 21); Ca_V2.3 + Ca_Vβ_2a cRNA, t½ = 4.11 ± 0.65 s (n = 12); Ca_V2.3 + Ca_Vβ_2a, t½ = 4.76 ± 0.70 s (n = 16); Ca_V2.3 + Ca_Vβ_2a-SH3, t½ = 0.35 ± 0.04 s (n = 13). The t½ values for Ca_V2.3 + Ca_Vβ_2a-GK, Ca_V2.3 + Ca_Vβ_2a-SH3-GK, and Ca_V2.3 + Ca_Vβ_2a were significantly different from those measured in oocytes expressing Ca_V2.3 alone (t test; P < .01). (C) Time course of inhibition of inactivation by Ca_Vβ_2a-GK. Each bar corresponds to the average t½ measured at different time intervals after protein injection. The first bar includes recordings from 12–50 min (n = 4), and the second bar includes recordings from 51–100 min (n = 6) and every 100 min thereafter (n = 7, 2, and 4, respectively). The dashed line corresponds to t½ for Ca_V2.3 alone.

Both Ca_Vβ_2a-GK and Ca_Vβ_2a-SH3-GK shifted the steady-state inactivation curve of Ca_V2.3-mediated currents to more positive potentials. A residual current at the end of the pulse also emerged, but it was only a fraction of that seen with full-length Ca_Vβ_2a (Fig. 5). Half-inactivation voltages (V½), derived from the fit to a Boltzmann distribution plus the residual component, were similar with Ca_Vβ_2a-GK, Ca_Vβ_2a-SH3-GK, and full-length Ca_Vβ_2a but were significantly more positive than those with Ca_V2.3 alone (Fig. 5B and Table S2). Overall, Ca_Vβ_2a derivatives appear to be less effective than full-length Ca_Vβ_2a in inhibiting inactivation and may reflect a contribution of unoccupied Ca_V2.3 subunits. Nevertheless, inactivation of Ca_V2.3 channels bearing the W386S mutation that disrupt binding to Ca_Vβ (25) were not modulated by Ca_Vβ_2a or Ca_Vβ_2a-GK (Fig. S5), indicating that the action of Ca_Vβ_2a-GK is AID-dependent and does not involve nonspecific binding. As in previous work (25), we also found that this substitution results in Ca_V2.3 channels that inactivate more slowly than wild-type channels. Taken together, these findings indicate that the N terminus of Ca_Vβ_2a is not mandatory for inhibiting inactivation and predict (owing to the highly conserved nature of the GK domain) that inhibition of inactivation is a property shared by all GKs in the various Ca_Vβ isoforms.

Fig. 5.

Ca_Vβ-GK shifts midpoint voltage for the steady-state inactivation of Ca_V2.3-mediated currents. (A) Representative traces of Ca_V2.3-mediated currents in the presence of the specified protein during a steady-state inactivation pulse protocol. This consisted of a 10-s conditioning period to voltages of increasing amplitude, from −120 mV to +30 mV in 15-mV increments, followed by a 0.4-s test pulse to 0 mV. Pulses were delivered once every 50 s from a holding potential of −90 mV. (B) Average steady-state inactivation curves from oocytes expressing Ca_V2.3 alone (n = 22) or after injection of full-length Ca_Vβ_2a (n = 13), Ca_Vβ_2a-SH3-GK (n = 23), or Ca_Vβ_2a-GK (n = 15). The continuous lines correspond to Boltzmann distributions plus a noninactivating current component that best described each set of data. For comparison, the Boltzmann distributions that best described the Ca_V2.3 + Ca_Vβ_2a cRNA data (dashed line) also are shown. The V½ values for Ca_V2.3 + Ca_Vβ_2a-GK and Ca_V2.3 + Ca_Vβ_2a-SH3-GK are significantly different than the V½ measured in oocytes expressing Ca_V2.3 alone (t test; P < .01) (see Table S2 for details).

Ca_vβ_1b-GK Resembles Ca_vβ_2a in Inhibiting Inactivation of Ca_v2.3 WT Channels.

We studied the effect of a GK module derived from Ca_Vβ_1b, an isoform that accelerates inactivation and shifts the steady-state inactivation curve to more negative potentials when co-expressed as cRNA (12). Injection of refolded Ca_Vβ_1b-GK into oocytes expressing Ca_V2.3 channels yielded currents that inactivated slowly (t½ = 4.2 ± 0.9 s, n = 16; Fig. 6A). Ca_Vβ_1b-GK also shifted the steady-state inactivation curves toward more depolarizing potentials and, like Ca_Vβ_2a-GK, induced a residual current (Fig. 6 B and C). Based on these findings, we conclude that the GK module encompasses the minimal structural requirements to inhibit voltage-dependent inactivation. A question that naturally arises is what determines the facilitation of inactivation by full-length Ca_Vβ proteins.

Fig. 6.

Ca_Vβ_1b-GK slows inactivation of Ca_V2.3-mediated currents and shifts the steady-state inactivation toward more depolarized potentials. (A) Representative current traces from a Ca_V2.3-expressing oocytes injected with Ca_Vβ_1b-GK during a 10-s pulse to 0 mV from a holding potential of −90 mV. (B) Current traces evoked with the steady-state inactivation pulse protocol from the same oocytes shown in (A). (C) Average steady-state inactivation curve from oocytes expressing Ca_V2.3 and injected with Ca_Vβ_1b-GK protein (n = 14). For comparison, the Boltzmann distributions that best describe Ca_V2.3 and Ca_V2.3 + Ca_Vβ_2a data from Fig. 5 also are shown (dashed lines). The V½ value for Ca_V2.3 + Ca_Vβ_1b-GK was significantly different from that measured in oocytes expressing Ca_V2.3 alone (t test; P < .01) (see Table S2 for details).

Full-Length Ca_Vβ Proteins Switch Ca_V2.3-Inactivation Phenotype Depending on the Time of Injection.

Our finding that the GK module inhibits inactivation implies that full-length proteins that facilitate inactivation must be able to counteract the GK effect. This property has been documented for Ca_V2.3 channels co-expressed with cRNA encoding full-length Ca_Vβ_1b or a palmitoylation-deficient mutant of Ca_Vβ_2a bearing two cysteine-to-serine substitutions at positions 3 and 4 (Ca_Vβ_2a C3,4S) (12, 18). Here we report that injection of these Ca_Vβ constructs as proteins into oocytes already expressing Ca_V2.3 channels (late injection) slowed the time to inactivation compared with Ca_V2.3 alone (Fig. 7 A and B). Furthermore, Ca_Vβ C3,4S shifted the steady-state inactivation curve toward more positive potentials (Fig. 7 C and D and Table S2). Assuming that these full-length proteins were folded correctly, we conjectured that posttranslational modifications requiring longer times than allowed by the experiment were needed to confer their native phenotype. Consequently, we co-injected these proteins together with the cRNA-encoding Ca_V2.3 subunit (co-injection) and indeed found that inactivation was accelerated (Fig. 7 A and B) and the steady-state inactivation curve shifted toward more negative potentials (Fig. 7 C and D and Table S2). These findings indicate that the recombinant proteins are properly folded, and that after posttranslational modifications during biogenesis of the channel complex, Ca_Vβ can switch the phenotype conferred to Ca_V2.3 from slow-inactivating to fast-inactivating.

Fig. 7.

The Ca_V2.3-inactivation phenotype induced by full-length Ca_Vβ_1b and Ca_Vβ_2aC3,4S depends on the time of injection. (A) Representative current traces from oocytes expressing Ca_V2.3 channels with Ca_Vβ_1b or Ca_Vβ_2aC3,4S injected either 2–7 h before recording (late injection) or co-injected with Ca_V2.3-encoding cRNA (co-injection). Currents were evoked by a 10-s pulse to 0 mV from a holding potential of −90 mV. (B) Average t½ for both of the subunit combinations shown in (A). Using Ca_Vβ_1b, the t½ values for co-injection (0.29 ± 0.02 s; n = 11) and late injection (2.19 ± 0.25 s; n = 14) differed significantly. This was true for experiments with Ca_Vβ_2aC3,4S as well (co-injection: 0.82 ± 0.08 s, n = 13; late injection: 2.86 ± 0.48 s, n = 16; t test; P < .01). (C) Steady-state inactivation curves from oocytes either co-injected or late-injected with Ca_Vβ_1b. The continuous lines correspond to the Boltzmann distributions that best describe each set of data. For comparison, the Boltzmann distributions that best describe the Ca_V2.3 data from Fig. 5 are shown (dashed lines). (D) As in C, but for Ca_Vβ_2aC3,4S. With both proteins, the V½ values from the co-injection experiments were significantly more negative than those from the late-injection experiments (t test; P < .01). For Ca_Vβ_2a C3,4S, the V½ values from both the late-injection and co-injection experiments were significantly different from those values for Ca_V2.3 alone (t test; P < .01). The V½ value for Ca_Vβ_1b differed from that of Ca_V2.3 alone only in the co-injection experiments. With both proteins, t½ values in late-injection and co-injection experiments differed significantly from each other and from those values for Ca_V2.3 alone (t test; P < .01) (see Table S2 for details).

Discussion

We have demonstrated that the GK module of the VGCC β subunit recapitulates key modulatory properties of the full-length protein, and thus Ca_Vβ-GK emerges as a functional competent domain. This implies that Ca_Vα₁–GK interaction suffices for channel gating regulation. Moreover, we have shown that regulation by GK is abolished by a mutation of the AID sequence, consistent with the idea that the AID–GK binding surface is critical for channel modulation (26). Although our size-exclusion chromatography and sucrose gradient analysis revealed that purified GK domains formed dimers in solution, the finding that Ca_Vβ_2a-GK covalently linked to a Ca_V1.2 pore-forming subunit fully recapitulated the activation properties of the channel (Fig. 3) proves that the functional unit is a single GK molecule. Our experiments also demonstrate that Ca_Vβ-GK can sustain the inhibition of voltage-dependent inactivation of Ca_V2.3 channels. The prevailing view is that this phenotype is conferred uniquely by Ca_Vβ_2a, because it is anchored to the membrane and constrains the movement of the inactivating particle encoded by loop I-II (15, 27). Our findings show that instead, inhibition of inactivation is not a unique feature of Ca_Vβ_2a, but rather is an inherent property of the GK module that appears to act as a brake to impair voltage-dependent inactivation. A corollary of this conclusion is that facilitation of inactivation by full-length Ca_Vβ requires additional structural determinants to antagonize GK's brake-like effect. Full-length proteins, Ca_Vβ_1b and Ca_Vβ_2a C3,4S, were able to counteract the action of GK only when they were co-injected with the Ca_V2.3-encoding cRNA, as if these determinants were acquired within a restricted time window during channel biogenesis.

We envision that in cRNA or protein co-injection experiments, formation of the Ca_Vα₁-Ca_Vβ complex in early compartments, such as the endoplasmic reticulum, allows the necessary chemical or structural modifications to counteract GK's brake-like effect. Within this framework, palmitoylation may sequester Ca_Vβ_2a to other membranous compartments early in the course of biogenesis, hindering the formation of Ca_Vα₁-Ca_Vβ complexes in the compartment, which is permissible for these structural modifications. Alternatively, Ca_Vβ protein may gradually switch to a fast-inactivation–conferring phenotype over a period of several days independent of its location or association state. In any case, we conclude that fast inactivation relies on further posttranslational modifications of Ca_Vβ, although the precise molecular mechanism remains to be identified.

In clear contrast to the protein injection experiments, we observed no modulation when Ca_Vβ-SH3-GK or Ca_Vβ-GK were co-injected as cRNA. Most likely, different parts of Ca_Vβ are important to either efficient translation or stability of the protein in the oocyte cytoplasm. Indeed, modulation of inactivation by Ca_Vβ_2a-GK, injected as protein, is rather transient compared with that of the full-length protein, indicating a reduced lifetime. This may explain why previous attempts by either co-expressing or co-injecting the protein more than 24 h before the recordings yielded seemingly contradictory results (7–11). Although the time course of Ca_Vβ_2a-GK effect on Ca_V1.2 activation could not be determined, linking GK to Ca_V1.2 proved to be an effective strategy for stabilizing this module and increasing its potency. So far, we have been unable to express a concatamer of Ca_V2.3 with either the full-length Ca_Vβ or the GK domain.

As new protein partners are discovered, the functional role of Ca_Vβ is expanding rapidly (28, 29). We recently found that the SH3 module of Ca_Vβ binds to the endocytotic protein dynamin (24), and now we report that the GK module regulates calcium channel function. Together, these findings introduce a new perspective on Ca_Vβ. Calcium entry through VGCCs on membrane depolarization ensures a transient change in intracellular calcium concentration that regulates diverse cellular functions. Integration of these different cellular processes must be tightly coordinated in living cells, and the domain architecture of Ca_Vβ, with its two functionally independent modules, appears particularly well suited for orchestrating calcium signaling. We suggest that whereas GK regulates calcium entrance, the SH3 domain links channel activity to other cellular processes by binding to additional proteins.

Materials and Methods

Construction of cDNA and Protein Expression.

cDNA encoding the GK domain (residues 201–422), the SH3-GK core (residues 24–422) of rat β_2a (Swiss-Prot Q8VGC3–2), and the GK domain (residues 209–413) of the rat β_1b (Swiss-Prot P54283) were subcloned by polymerase chain reaction methods into a pRSET vector (Invitrogen). The predicted molecular masses of the Ca_Vβ_1b-GK, Ca_Vβ_2a-GK, and Ca_Vβ_2a-SH3-GK constructs, including the N-terminal His Tag, a transcript-stabilizing sequence, and the enterokinase cleavage recognition site, were 26.9 kDa, 28.6 kDa, and 48.2 kDa, respectively. Ca_Vβ_2a-SH3 was prepared as described previously (24). Full-length Ca_Vβ_2a, the mutant bearing two substitutions at positions 3 and 4 (Ca_Vβ_2a C3,4S) and Ca_Vβ_2a-SH3-GK also were prepared as described previously (22). Ca_Vβ_1b, Ca_Vβ_1b-GK, and Ca_Vβ_2a-GK were expressed in bacteria and recovered from inclusion bodies as reported previously (30). The GK domains were refolded by batch dilution (11-fold dilution) in refolding buffer (400 mM l-arginine, 2 mM NaEDTA, 0.5 mM glutathione oxidized, and 100 mM Tris base [pH 7.0]) and subsequently purified by size-exclusion chromatography onto a Superdex S-200 column (GE Healthcare) preequilibrated with nondenaturing buffer (20 mM Tris, 300 mM NaCl, and 1 mM EDTA [pH 8.0]). Proteins were concentrated up to 0.1 mg/ml by ultrafiltration (Amicon Ultra-4 10 kDa MWCO), rapidly frozen, and stored at −80°C until use. The identity of the purified proteins was confirmed by mass spectrometry analysis performed in the mass spectrometry laboratory of the Department of Pharmacology and Toxicology, Medizinische Hochschule Hannover. The protein was digested by trypsin, and the peptides were analyzed with an Ultraflex MALDI-TOF/TOF mass spectrometer (Bruker Daltonics).

The loop I-II of Ca_V1.2 (Swiss-Prot P15381) fused to GST (GST–loop I-II) was prepared as described previously (30). YFP was fused to Ca_Vβ_2a-GK at its N terminus (YFP-Ca_Vβ_2a-GK) and subcloned into the pcDNA 3.1 vector. The cDNA encoding Ca_V1.2 was fused to the GK domain at the carboxyl-terminal end (Ca_V1.2- Ca_Vβ_2a-GK) by overlapping extension polymerase chain reaction, which incorporated the sequence “MGRDLYDDDDKD” at residue 2164 of Ca_V1.2. All constructs were verified by DNA sequencing.

Binding Assay.

First, tsA201 cells were transfected with a YFP-Ca_Vβ_2a-GK or YFP-alone encoding vector and lysed after 24–36 h. Precleared cell extracts were incubated for 1 h with glutathione beads coupled to either GST–loop I-II or GST alone. The beads were pelleted and washed, and bound proteins were eluted with SDS/PAGE loading buffer. Proteins were then resolved on SDS/PAGE and visualized by fluorescence scanning (Typhoon imager; GE Healthcare).

Oocyte Injection and Electrophysiologic Recordings.

cRNA was synthesized and Xenopus laevis oocytes were prepared, injected, and maintained as reported previously (30). The Ca_V2.3 encoding cDNA was sequenced; compared with the Swiss-Prot entry Q15878, the following changes were noted: I649M, W837L, P838A, and insertion of a glycine residue at position 839. The Ca_V1.2-subunit used in this study bears 60 aa deletion at the amino terminal (31). Electrophysiologic recordings were performed with the cut-open oocyte technique 4–6 days after cRNA injection and 1–7 hours after protein injection, as described previously (22). For details see SI Materials and Methods.