Molecular determinants of Ca²⁺/calmodulin-dependent regulation of Ca_v2.1 channels

Abstract

Ca²⁺-dependent facilitation and inactivation (CDF and CDI) of Ca_v2.1 channels modulate presynaptic P/Q-type Ca²⁺ currents and contribute to activity-dependent synaptic plasticity. This dual feedback regulation by Ca²⁺ involves calmodulin (CaM) binding to the α₁ subunit (α₁2.1). The molecular determinants for Ca²⁺-dependent modulation of Ca_v2.1 channels reside in CaM and in two CaM-binding sites in the C-terminal domain of α₁2.1, the CaM-binding domain (CBD) and the IQ-like domain. In transfected tsA-201 cells, CDF and CDI were both reduced by deletion of CBD. In contrast, alanine substitution of the first two residues of the IQ-like domain (IM-AA) completely prevented CDF but had little effect on CDI, and glutamate substitutions (IM-EE) greatly accelerated voltage-dependent inactivation but did not prevent CDI. Mutational analyses of the Ca²⁺ binding sites of CaM showed that both the N- and C-terminal lobes of CaM were required for full development of facilitation, but only the N-terminal lobe was essential for CDI. In biochemical assays, CaM₁₂ and CaM₃₄ were unable to bind CBD, whereas CaM₃₄ but not CaM₁₂ retained Ca²⁺-dependent binding to the IQ-like domain. These findings support a model in which Ca²⁺ binding to the C-terminal EF-hands of preassociated CaM initiates CDF via interaction with the IQ-like domain. Further Ca²⁺ binding to the N-terminal EF-hands promotes secondary CaM interactions with CBD, which enhance facilitation and cause a conformational change that initiates CDI. This multifaceted mechanism allows positive regulation of Ca_v2.1 in response to local Ca²⁺ increases (CDF) and negative regulation during more global Ca²⁺ increases (CDI).

P/Q-type Ca²⁺ currents conducted by presynaptic Ca_v2.1 channels initiate neurotransmitter release at many central synapses. In nerve terminals, activity-dependent increases in intracellular Ca²⁺ ions cause an initial facilitation followed by a progressive inactivation of P/Q-type Ca²⁺ currents, which can alter synaptic efficacy (1-3). In transfected cells, Ca_v2.1 channels undergo a similar dual feedback regulation by Ca²⁺ that is mediated by calmodulin (CaM) (4, 5). CaM binds to the poreforming α₁ subunit of Ca_v2.1 channels (α₁2.1), causing an initial Ca²⁺-dependent facilitation (CDF) and, on a longer time scale, Ca²⁺-dependent inactivation (CDI) of these channels during repetitive stimuli (4, 5).

How CaM binding to α₁2.1 results in two opposing forms of channel regulation is not yet clear. A CaM-binding domain (CBD) in the cytoplasmic C-terminal tail of α₁2.1 binds CaM in a Ca²⁺-dependent manner, and deletion of this site diminishes but does not completely abolish CDF and CDI (4, 5). In addition, a second site on the N-terminal side of the CBD also participates in CaM regulation of Ca_v2.1 (6). This IQ-like domain is similar to the one that mediates CDI of L-type Ca²⁺ currents through Ca_v1.2 channels (7-9) but lacks the key Q residue at position 2 and hydrophobic residue at position 8 that characterize other IQ domains (10). Peptides corresponding to the IQ-like domain of α₁2.1 bind CaM in vitro, and mutations in the IQ-like domain impair CDF of Ca_v2.1 channels (6, 7, 11). Moreover, fluorescence resonance energy transfer experiments show that CaM mutants incapable of binding Ca²⁺ interact with Ca²⁺ channels, suggesting that apoCaM is constitutively associated with Ca_v2.1 (12). In contrast, biochemical experiments that require stable association of CaM show strictly Ca²⁺-dependent binding of CaM to the CBD and the IQ-like domain (4, 7).

To clarify the role of the CBD and IQ-like domains in CDF and CDI, we analyzed the contributions of the CBD, the IQ-like domain, and the N- and C-terminal lobes of CaM in Ca²⁺-dependent regulation of Ca_v2.1. Our results show that CDF and CDI result from a series of complex interactions between Ca²⁺ binding to specific lobes of CaM and distinct molecular contacts with the IQ-like domain and CBD.

Experimental Procedures

Molecular Biology. α₁2.1 constructs containing alterations of the CBD or IQ-like domain were generated by PCR from cDNA encoding rat α₁2.1 (rbA) (13). α₁2.1 lacking amino acids 1969-2000 (α₁2.1_ΔCBD) has been described (5). α₁2.1_IM-AA, α₁2.1_IM-EE, and α₁2.1_ΔCBD/IM-AA were constructed by subcloning EcoRV/PmlI fragments incorporating mutations at positions 1913 and 1914 into the corresponding sites of α₁2.1 or α₁2.1_ΔCBD in a pBluescript SK+ shuttle vector. From this construct, a SgrAI/MluI fragment containing the mutation(s) was subcloned into rba/pMT2XS. α₁2.1_1965t was constructed by amplifying the EcoRV/PmlI fragment with primers incorporating a stop codon after position 1965 and subcloning the truncated fragment into rba/pMT2XS. cDNAs encoding CaM mutants (14), provided by John Adelman (Vollum Institute, Oregon Health and Science University, Portland, OR), were subcloned into BamHI sites of pcDNA3.1+ (Invitrogen) for mammalian cell expression or pGEX4T1 for GST-fusion protein expression. His-tagged fusion constructs were generated by subcloning α₁2.1 C-terminal fragments into BamHI sites of pTrcHisA (Invitrogen).

Electrophysiological Recording and Data Analysis. tsA-201 cells were grown to ≈70% confluence and transfected by the calcium phosphate method with an equimolar ratio of cDNAs encoding WT or mutant α₁2.1, β_2a, and α₂δ (15). For electrophysiological experiments, cells plated on 35-mm dishes were transfected with a total of 5 μg of DNA including 0.3 μg of a CD8 expression plasmid for detection of transfected cells. Twenty-four hours later, cells were plated at low density for electrophysiological recording.

At least 48 h after transfection, cells were incubated with CD8-antibody-coated microspheres (Dynal, Oslo) to identify transfectants and washed in extracellular solution before recording. Whole-cell Ca²⁺ or Ba²⁺ currents were recorded with a List EPC-7 patch clamp amplifier driven by PULSE software (HEKA Electronics, Lambrecht/Pfalz, Germany). Leak and capacitive transients were subtracted by using a P/-4 protocol. Extracellular recording solutions contained 150 mM Tris, 1 mM MgCl₂ and 10 mM CaCl₂ or BaCl₂. Intracellular solutions consisted of 120 mM N-methyl-d-glucamine, 60 mM Hepes, 1 mM MgCl₂, 2 mM Mg-ATP, and 0.5 mM EGTA. The pH of all solutions was adjusted to 7.3 with methanesulfonic acid. Normalized tail current-voltage curves were fit with a single Boltzmann function: A/{1 + exp[(V - V_1/2)/k] + b}, where V is test pulse voltage, V_1/2 is the midpoint of the activation curve, k is a slope factor, A is the amplitude, and b is the baseline. Data analysis was done with IGOR PRO (WaveMetrics, Lake Oswego, OR).

Binding Assays. GST-CaM and His-α₁2.1 fusion proteins were expressed in BL21 Escherichia coli and purified according to standard protocols. GST-CaM immobilized on glutathione agarose beads was incubated with 5 μg of purified His-α₁2.1 fusion protein for 3 h at 4°C. After extensive washing, bound proteins were eluted and subjected to SDS/PAGE and immunoblotting with monoclonal anti-His or -GST antibodies.

Results

Roles of the CBD and the IQ-Like Domain in CDI. We constructed α₁2.1 subunits with alterations in the IQ-like domain and the CBD (Fig. 1a) and analyzed CDI of Ca_v2.1 channels transfected in tsA-201 cells by using Ca²⁺ and Ba²⁺ as charge carriers and a relatively low concentration of EGTA (0.5 mM) in the recording pipette to minimize Ca²⁺ buffering (4, 5). Inactivation of WT and mutant channels was compared quantitatively as the ratio of the residual current amplitude at 800 ms to the peak current amplitude (I_res/I_pk; Fig. 1b), and CDI was expressed as the difference between the average I_res/I_pk for Ca²⁺ and Ba²⁺. For α₁2.1, CDI is manifested as a faster decay of the Ca²⁺ current (I_Ca), and therefore, a significant decrease in I_res/I_pk for I_Ca compared to I_Ba (0.40 ± 0.03, n = 9 for I_Ca vs. 0.67 ± 0.04, n = 7 for I_Ba; CDI = 0.27; P < 0.01). α₁2.1 subunits lacking the CBD (α₁2.1_ΔCBD) gave rise to channels with significantly reduced CDI (I_res/I_pk = 0.55 ± 0.02, n = 7 for I_Ca vs. 0.65 ± 0.04, n = 9 for I_Ba; CDI = 0.10, P < 0.01). We obtained similar results for α₁2.1 subunits truncated immediately before the CBD (α₁2.1_1965t; I_res/I_pk = 0.56 ± 0.02, n = 10 for I_Ca vs. 0.65 ± 0.05, n = 9 for I_Ba; CDI = 0.09, P < 0.01). Despite the significant reduction in CDI for both α₁2.1_ΔCBD and α₁2.1_1965t, I_Ca still inactivated faster than I_Ba, suggesting that a second site such as the IQ-like domain may also be involved in CDI.

Fig. 1.

Contributions of the CBD and IQ-like domain to CDI. (a) Schematic showing constructs with alterations affecting the IQ-like domain (IM) and CBD in the C-terminal domain of α₁2.1; α₁2.1_IM-AA containing AA in place of ¹⁹¹³I¹⁹¹⁴M; α₁2.1_1965t truncated after amino acid 1965; α₁2.1_ΔCBD lacking amino acids 1969-2000; α₁2.1_ΔCBDIM-AA with both mutations; and α₁2.1_IM-EE with EE in place of¹⁹¹³I¹⁹¹⁴M. (b) CDI of Ca_v2.1 channels containing α₁ subunits with altered C termini. Normalized current traces are shown for currents evoked by 1-s test pulses from a holding voltage of -80 mV to +10 or 0 mV with 10 mM extracellular Ca²⁺ (black trace) or Ba²⁺ (gray trace), respectively. Intracellular solutions contained 0.5 mM EGTA. The current amplitude was measured at 800 ms (I_res) and normalized to the peak current amplitude (I_pk). Resulting I_res/I_pk values were averaged (n = 5-10) and plotted (±SEM). Asterisks indicate significant differences between CDI [(I_res/I_pk for I_Ca) - (I_res/I_pk for I_Ba)] for mutant compared to WT channels (P < 0.05). (c) Enhanced inactivation of α₁2.1_IM-EE. Normalized currents evoked by 0.5-s test pulses from -80 to +10 (I_Ca, black trace) or 0 mV (I_Ba, gray trace). I_res was measured at 50 ms, and I_res/I_pk was plotted as in b.

To assess the role of the IQ-like domain, alanine substitutions were made for the first two residues (IM) in this region (α₁2.1_IM-AA). These residues are critical for CaM regulation in conventional IQ domains in many other target proteins, including Ca_v1.2 channels (16, 17), and the IM-AA mutation in α₁2.1 lowers the affinity for binding CaM by 30-fold in vitro (6). However, CDI was not significantly reduced in channels with α₁2.1_IM-AA (I_res/I_pk = 0.45 ± 0.03, n = 10 for I_Ca vs. 0.66 ± 0.04 for I_Ba, n = 6; CDI = 0.21, P = 0.11). Double mutant channels lacking the CBD and containing the IM-AA substitution (α₁2.1_ΔCBDIM-AA) exhibited significantly slower inactivation for I_Ca and I_Ba compared to WT (CDI = 0.12, P < 0.01; Fig. 1b), but the reduction in CDI was not greater than that caused by deletion of the CBD alone.

Substitution of negatively charged glutamate residues for IM in the IQ-like domain had substantial effects on Ca_v2.1 function independent of Ca²⁺ (α₁2.1_IM-EE; Fig. 1c). First, the IM-EE mutation caused a nearly 5-fold reduction in current, which necessitated the use of higher extracellular Ca²⁺ and Ba²⁺ concentrations to resolve effects on CDI. Second, α₁2.1_IM-EE inactivated rapidly, with I_Ca and I_Ba decaying completely by 200 ms, a time point at which neither I_Ca nor I_Ba through WT is strongly inactivated. This marked enhancement of inactivation of α₁2.1_IM-EE would partially occlude CDI because it exceeds the rate at which Ca²⁺/CaM can induce CDI. However, comparison of the residual current amplitude at 50 ms revealed faster inactivation of I_Ca than I_Ba (Fig. 1c; I_res/I_pk = 0.26 ± 0.04, n = 5 for I_Ca vs. 0.41 ± 0.05, n = 5 for I_Ba; P = 0.06), suggesting that CDI remains at least partially intact after the IM-EE mutation. Therefore, our results support a prominent role for the CBD, but not the IQ-like domain, in CDI of Ca_v2.1.

Roles of the CBD and the IQ-Like Domain in CDF. To analyze CDF of Ca_v2.1, we used two protocols similar to those used to assess facilitation of synaptic transmission (Fig. 2). During trains of repetitive stimuli, I_Ca increases to a new steady level, whereas I_Ba remains relatively constant (Fig. 2a). During paired-pulse experiments, I_Ca evoked after a conditioning prepulse (P2) is greater than I_Ca elicited before the prepulse (P1) (Fig. 2f). These two measures of CDF provide complementary information: the paired-pulse protocol shows CDF evoked by a single episode of sustained Ca²⁺ influx, and the cumulative increase of I_Ca during trains of stimuli also reflects the stability of CDF in the interval between pulses. Our previous results show that the stability of CDF is strongly enhanced by Ca²⁺ and CaM (5).

Fig. 2.

Contributions of the IQ-like domain and CBD to CDF. (a-d) Facilitation of I_Ca during 100-Hz trains of pulses from -80 to +10 or 0 mV, as indicated in the voltage protocol shown above the panels. Test current amplitudes (± SEM, n > 4) were normalized to the first in the train, and every third point is plotted against time. (e) Averaged normalized current amplitudes between 0.5 and 1 s, (I_n/I₁). Asterisks indicate significant difference between mutant and WT channels (P < 0.05). (f-i) CDF in paired-pulse protocols, as indicated above the panels, with extracellular 10 mM Ca²⁺ and intracellular 0.5 mM EGTA. Tail current amplitudes were measured on repolarization to -40 mV, normalized to the mean tail current after P1 pulses between +50 to +70 mV, and plotted against test pulse voltage. Insets show representative P1 and P2 currents. Plotted points are mean ± SEM for P1 (□) and P2 (▪) for α₁2.1 (n = 5), α₁2.1_IM-AA (n = 5), α₁2.1_ΔCBD (n = 4), and α₁2.1_ΔCBDIM-AA (n = 7). (j) Means for I_Ca or I_Ba of P2_max/P1_max, where PX_max is the mean value of tails after pulses between +50 and +70 mV.

Compared to WT Ca_v2.1 (Fig. 2f), deletion of the CBD reduced the paired-pulse facilitation of I_Ca (Fig. 2 h and j). Similarly, deletion of the CBD caused a partial loss of facilitation measured in the repetitive pulse protocol (Fig. 2 c and e). By contrast, there was no significant facilitation of α₁2.1_IM-AA or α₁2.1_ΔCBDIM-AA in paired-pulse experiments (Fig. 2 g, i, and j) and no difference between facilitation of I_Ca and I_Ba in repetitive pulse experiments (Fig. 2 b, d, and e). These results demonstrate a primary requirement for the IQ-like domain in CDF and a modulatory role for the CBD in this process.

Distinct Functions of the N- and C-Terminal EF-Hands of CaM in CDI. We have shown previously that CDI but not CDF is blocked by 10 mM EGTA, a high-affinity but slow Ca²⁺ chelator, whereas both processes are blocked by 10 mM BAPTA, a rapid chelator (5). The distinct Ca²⁺ sensitivities for the two processes indicate that CDF is triggered by local Ca²⁺, which is intercepted rapidly by BAPTA, whereas CDI is induced by global Ca²⁺ elevations, which are removed more slowly by EGTA. The molecular basis for this difference between CDF and CDI could reside in the N- and C-terminal lobes of CaM, which bind Ca²⁺ ions with different affinities (18, 19). To test this, we used CaM constructs containing alanine substitutions for critical aspartate residues that impair Ca²⁺ coordination in the paired EF-hands of the N-terminal lobe (CaM₁₂), the C-terminal lobe (CaM₃₄), or both lobes of CaM (CaM₁₂₃₄) (Fig. 3) (14). Although coexpression of Ca_v2.1 with CaM or CaM₃₄ did not influence CDI during 1-s depolarizing pulses, CaM₁₂ virtually eliminated the difference in inactivation between I_Ca and I_Ba (Fig. 3). As expected, there was also substantially reduced CDI when CaM₁₂₃₄ was expressed, although its effect was less complete than CaM₁₂ (Fig. 3). These results point to a specific requirement for the N-terminal EF-hands of CaM for CDI.

Fig. 3.

Requirement for N- and C-terminal EF-hands of CaM for CDI. (Upper) Schematic of CaM. (Lower) CDI for Ca_v2.1 channels cotransfected with CaM mutants. Normalized current traces are shown for 1-s test pulses from a holding voltage of -80 mV to +10 or 0 mV for I_Ca (black traces) or I_Ba (gray traces) with 0.5 mM EGTA intracellular. Mean I_res/I_pk values (±SEM; n = 5-12) for I_Ca or I_Ba as in Fig. 1b for the indicated CaM constructs. Asterisks indicate significant differences from WT CaM (P < 0.01).

Dual Requirement for the N- and C-Terminal Lobes of CaM in CDF. In a previous study (6), CaM₁₂ supported CDF in paired-pulse experiments, whereas CaM₃₄ did not, leading to the proposal that CDF was mediated primarily by the C-terminal EF-hands of CaM. We further investigated the function of the N- and C-terminal lobes of CaM by using the paired-pulse and repetitive-stimulation protocols (Fig. 4). As in the previous work, we found nearly complete loss of CDF in the paired-pulse protocol (Fig. 4 h and j) and in the repetitive-stimulation protocol (Fig. 4 c and e) with the CaM₃₄ mutant. However, we also found a significant impact of the CaM₁₂ mutant on facilitation in both protocols. In the paired-pulse protocol, facilitation of I_Ca was substantially reduced by CaM₁₂, but facilitation of I_Ba was reduced to a similar extent (Fig. 4 g and j; see also ref. 6). In the repetitive pulse protocol, there was nearly complete loss of facilitation (Fig. 4 b and e), and differences between I_Ca and I_Ba with CaM₁₂ were not significant (1.04 ± 0.02, n = 13 vs. 0.98 ± 0.04, n = 13; P = 0.22). CaM₁₂₃₄ also caused nearly compete loss of CDF in both protocols (Fig. 4 d, e, i, and j). Together, these results confirm an absolute requirement for Ca²⁺ binding to the C-terminal lobe of CaM for CDF and reveal a significant role for the N-terminal lobe as well.

Fig. 4.

Requirement for N- and C-terminal EF-hands for CDF. (a-d) CDF during 100-Hz trains of depolarizations as in Fig. 2 a-d [mean ± SEM for CaM (n = 11), CaM₁₂ (n = 13), CaM₃₄ (n = 9)], and CaM₁₂₃₄ (n = 7). (e) Averaged normalized current amplitudes between 0.5 and 1 s as in Fig. 2e. (f-i) CDF during paired-pulse protocols measured as in Fig. 2 e-i [mean ± SEM for CaM (n = 10), CaM₁₂ (n = 6), CaM₃₄ (n = 7), and CaM₁₂₃₄ (n = 5)]. (j) Mean values for P2_max/P1_max for I_Ca or I_Ba determined as in Fig. 2j. Asterisks indicate significant differences from WT (P < 0.05).

Differential Binding of the IQ-Like Domain and CBD to CaM. To determine whether differences in CaM-binding to the CBD and IQ-like domain contribute to their different functional effects, we measured the direct interaction of the CBD and IQ-like domains with CaM, CaM₁₂, and CaM₃₄ in a binding assay that detects stable, high-affinity CaM-binding interactions. Using GST-tagged CaMs immobilized on glutathione agarose beads, we observed Ca²⁺-dependent binding of His-tagged CBD and IQ-like domain peptides, as reported (4, 7, 11) (Fig. 5 a and b). Neither CaM₁₂ nor CaM₃₄ associated with the CBD, indicating that functional EF-hands in both the N- and C-terminal lobes are required for stable Ca²⁺-dependent CaM interaction with this site (Fig. 5c). By contrast, CaM binding to the IQ-like domain was prevented by inactivation of the N- but not C-terminal EF-hands (Fig. 5c), consistent with results of yeast-two-hybrid analyses (20). These results eliminate the possibility that blockade of CDF by CaM₃₄ (Fig. 4) was caused by its inability to interact with the IQ-like domain. Evidently, Ca²⁺ binding to the N-terminal lobe is sufficient for stable CaM binding to the IQ-like domain but not for induction of CDF.

Fig. 5.

Binding of CaM to the IQ-like domain and the CBD. (a) His-tagged C-terminal fragments including the IQ-like domain (IM, amino acids 1848 - 1964), CBD (1959-2035), or an upstream sequence (1764-1848) used as a negative control. (b)Ca²⁺-dependent binding of His-tagged C-terminal fragments (His) to GST-tagged CaM (input, GST) immobilized on glutathione agarose in the presence of 2 μM Ca²⁺ or 10 mM EGTA was detected by immunoblot with anti-His antibodies. (c) Differential binding of CaM₁₂ and CaM₃₄ to the IQ-like domain and CBD. GST-tagged CaM₁₂ or CaM₃₄ was used to pull-down His-tagged α₁2.1 C-terminal fragments in 2 μMCa²⁺ as in b.(d) Model for CDI and CDF based on sequential interactions of the IQ-like domain and CBD with CaM.

Discussion

Specific Roles of the CBD and IQ-Like Domain in Ca²⁺-Dependent Regulation of Ca_v2.1. Our previous results showed that Ca_v2.1 channels undergo prominent CDF and CDI and identified the CBD as a CaM-binding motif that contributes to these forms of regulation (4, 5). The experiments presented here also demonstrate a significant loss of CDI in Ca_v2.1 channels lacking the CBD (Fig. 1) and provide the first evidence for an important role of the CBD in CDF measured in paired-pulse and repetitive stimulation paradigms (Fig. 2). Deletion of the CBD selectively in α₁2.1_ΔCBD or by truncation of the C-terminal domain in α₁2.1_1965t caused a similar reduction in CDF and CDI. Evidently, the CBD has functionally significant interactions with CaM in both CDF and CDI, but these are not the exclusive interactions that mediate these regulatory processes because measurable CDF and CDI remain in CBD-deletion mutants.

Our results also extend the work of DeMaria and colleagues, who discovered a major role for the IQ-like domain in CDF of Ca_v2.1 (6). Like these authors, we find that mutation of the first two amino acid residues in the IQ-like domain to alanine (IM-AA) completely prevents CDF (Fig. 2 b and g), indicating that the IQ-like domain is necessary for CDF of Ca_v2.1 channels. In contrast to these results with CDF, we did not resolve a significant effect of the IM-AA mutation on CDI, in agreement with previous work (6). Because the IM-AA mutation causes a 30-fold reduction in affinity for CaM (6), these results suggest that CDF requires high-affinity binding of CaM that is prevented by the IM-AA mutation. In contrast, CDI either does not require the IQ-like domain or requires only low-affinity interactions that are retained in the IM-AA mutant.

Mutation of the same two amino acid residues in the IQ-like domain to glutamate (IM-EE) completely blocks CaM binding to the IQ-like domain and dramatically accelerates voltage-dependent inactivation of I_Ba (Fig. 1c and ref. 6). This result establishes an important role for these amino acid residues in control of voltage-dependent inactivation in the absence of Ca²⁺. Acceleration of voltage-dependent inactivation to the extent observed for the IM-EE mutation would also cause the channel to inactivate before Ca²⁺/CaM could initiate CDI, regardless of the role of the IQ-like domain in this process. We have shown that similar rapid, voltage-dependent inactivation of Ca_v2.1 caused by coexpression of auxiliary β_1b rather than β_2a subunits largely occludes CDI, even though these β subunits have no direct role in CDI itself (5). Despite this potential difficulty, we were able to measure residual CDI of the IM-EE mutant at an early (50 ms) time point (Fig. 1c). These results and the lack of effect of the IM-AA mutation on CDI shown here and in previous work (6) indicate that the IQ-like domain does not have a prominent role in CDI of Ca_v2.1 channels. Although it is possible that the IQ-like domain does have a significant role in CDI, its impact is not revealed by the IM-EE or IM-AA mutations that have been analyzed to date.

Our results differ from those of DeMaria et al. (6), who found that deletions of the CBD did not impact either CDF or CDI. These different results may in part be due to methodological differences. Our CDF protocols measure an increase in peak I_Ca rather than an increased rate of activation of I_Ca as in previous work (6), which may require more stable binding of CaM through interaction with the CBD. Our CDI experiments show that deletion of the CBD significantly reduced but did not abolish CDI (Fig. 1 and ref. 4), indicating that other regions of the channel contribute to CaM-dependent conformational changes leading to CDI. The functional roles of the CBD in CDI and CDF that we have detected by using the rat brain α₁2.1 (rbA isoform) may be less apparent with the human α₁2.1 isoform used in the previous work (6). Although the C-terminal regions, including the IQ-like domain and the CBD, are quite similar in the human and rat sequences, functionally significant differences may exist within these conserved sequences or in other parts of the channel that could influence channel modulation by CaM (21, 22).

Distinct Roles of the N- and C-Terminal Lobes of CaM. CaM mutants can disrupt Ca²⁺-dependent modulation of Ca²⁺ channels by displacing endogenous CaM binding to the channel and preventing Ca²⁺-dependent conformational changes leading to CDF and CDI (6-8). However, interpretation of these results is complicated by the presence of endogenous CaM at sufficient concentration to give both CDF and CDI. In this situation, the lack of effect of a mutant CaM may reflect inability to bind Ca_v2.1, thereby allowing endogenous CaM to function normally. Alternatively, a CaM mutant may have no apparent effect because, although it does interact with Ca_v2.1, it causes CDF and CDI that are indistinguishable from endogenous CaM. Moreover, some CaM mutants may have effects that are independent of their inability to bind Ca²⁺.

Mutation of the N-terminal EF-hands of CaM completely blocks CDI (Fig. 3 and ref. 6). These results indicate that CaM₁₂ does interact with Ca_v2.1 and prevent the action of endogenous CaM but cannot cause CDI itself. Because we do not detect stable high-affinity binding of CaM₁₂ in our biochemical experiments (Fig. 5c), reversible, low-affinity interactions of overexpressed CaM₁₂ are apparently sufficient to disrupt CDI by endogenous CaM. By contrast, CaM₃₄ has no effect on CDI (Fig. 3 and ref. 6), but our binding studies show that it can bind to the IQ-like domain (Fig. 5). These results indicate that CaM₃₄ is fully active in CDI and therefore that Ca²⁺ binding to the N-terminal EF-hands of CaM is sufficient for CDI.

Mutation of the C-terminal lobe of CaM completely prevents CDF in both paired-pulse and repetitive-stimulation protocols (Fig. 4), as in previous work using different stimulus paradigms (6). CaM₃₄ can bind to the IQ-like domain in the presence of Ca²⁺ (Fig. 5c), so the block of CDF by CaM₃₄ evidently results from its inability to bind Ca²⁺ to EF-hands 3 and 4 and induce conformational changes within the IQ-like domain that favor CDF. We also found that mutation of the N-terminal lobe of CaM significantly reduces facilitation measured in the paired-pulse protocol and nearly completely prevents facilitation during trains of stimuli (Fig. 4). In contrast to the loss of CDF with CaM₃₄, CaM₁₂ reduces facilitation in the presence of Ca²⁺ but also decreases the limited facilitation observed when Ba²⁺ is the charge carrier. Ba²⁺ may bind weakly to the N-terminal EF-hands of CaM and partially support facilitation. In fact, Ba²⁺-dependent inactivation of Ca_v1.2 has been observed and proposed to be due to Ba²⁺ binding to CaM (23, 24). In any case, our results indicate that Ca²⁺ binding to EF-hands 1 and 2 is required for maximal facilitation when measured as an increase in peak I_Ca in paired-pulse or repetitive-stimulation protocols, perhaps because Ca²⁺-binding to EF-hands 1 and 2 increases the overall affinity of CaM binding to the CBD.

A Revised Molecular Model for CDF and CDI of Ca_v2.1. Lobe-specific regulation is a general feature of functional interactions of CaM with many targets (14, 20, 25, 26). Although a number of ion channels are modulated by direct interactions with CaM (27), Ca_v2.1 channels are unusual in that CaM binding mediates two opposing forms of modulation that are kinetically and mechanistically distinct (5). Our results support and extend a model (6) in which CDF and CDI result from Ca²⁺ binding to specific lobes of CaM and sequential changes in the interaction of Ca²⁺/CaM with the IQ-like domain and CBD. Although our biochemical experiments do not detect stable CaM binding to the IQ-like domain or the CBD in the absence of Ca²⁺ (Fig. 5b), the millisecond kinetics of CDF imply that CaM must be reversibly bound at resting Ca²⁺ levels in the cell. In our revised model (Fig. 5d), voltage-gated Ca²⁺ influx through Ca_v2.1 would promote local Ca²⁺ binding to the C-terminal lobe of preassociated CaM, which initiates or strengthens interaction with the IQ-like domain. More global increase in Ca²⁺ leads to binding of Ca²⁺ ions to the N-terminal lobe of CaM and also induces a conformational change that allows binding of fully Ca²⁺-liganded CaM to the CBD. Binding of Ca²⁺/CaM to the CBD then enhances CDF and initiates CDI. According to this model, when preassociated CaM is replaced by CaM₃₄, Ca²⁺ influx initiates Ca²⁺ binding to the N-terminal lobe of CaM, which supports only CDI. When CaM is replaced by CaM₁₂, Ca²⁺ influx initiates Ca²⁺ binding only to the C-terminal lobe of CaM that permits partial induction of CDF.

Considerable progress has been made toward understanding where CaM interacts with Ca_v2.1 and other voltage-gated Ca²⁺ channels and how this interaction can modulate channel function (4-6, 11, 17, 28-30). However, it is becoming increasingly clear that Ca²⁺-dependent regulation of Ca_v2.1 is complex, involving Ca²⁺-dependent and -independent interactions of CaM with multiple sites on the α₁2.1 subunit. It is possible that CaM regulation of Ca_v2.1 involves dynamic interactions between the CBD, IQ-like domain, and potentially other sites, as has been proposed for the multiple CaM-binding sequences in the ryanodine receptor (31, 32). Moreover, emerging evidence shows that Ca²⁺ sensors in addition to CaM are likely to participate in Ca_v2.1 regulation in neurons (33, 34). How such interactions are ultimately transduced into physiological forms of channel modulation is an intriguing question to resolve in future structural and functional analyses.