Familial hemiplegic migraine mutations increase Ca²⁺ influx through single human Ca_V2.1 channels and decrease maximal Ca_V2.1 current density in neurons

Abstract

Insights into the pathogenesis of migraine with aura may be gained from a study of human Ca_V2.1 channels containing mutations linked to familial hemiplegic migraine (FHM). Here, we extend the previous single-channel analysis to human Ca_V2.1 channels containing mutation V1457L. This mutation increased the channel open probability by shifting its activation to more negative voltages and reduced both the unitary conductance and the density of functional channels in the membrane. To investigate the possibility of changes in Ca_V2.1 function common to all FHM mutations, we calculated the product of single-channel current and open probability as a measure of Ca²⁺ influx through single Ca_V2.1 channels. All five FHM mutants analyzed showed a single-channel Ca²⁺ influx larger than wild type in a broad voltage range around the threshold of activation. We also expressed the FHM mutants in cerebellar granule cells from Ca_V2.1α₁^−/− mice rather than HEK293 cells. The FHM mutations invariably led to a decrease of the maximal Ca_V2.1 current density in neurons. Current densities were similar to wild type at lower voltages because of the negatively shifted activation of FHM mutants. Our data show that mutational changes of functional channel densities can be different in different cell types, and they uncover two functional effects common to all FHM mutations analyzed: increase of single-channel Ca²⁺ influx and decrease of maximal Ca_V2.1 current density in neurons. We discuss the relevance of these findings for the pathogenesis of migraine with aura.

Migraine is a debilitating illness, characterized by attacks of severe, unilateral headache, which afflicts 10–15% of the population. Attacks can be preceded by an “aura,” most frequently consisting of scintillations that slowly drift across the visual field. Functional brain imaging and magnetoencephalography in patients (1, 2) have provided evidence that the visual aura is the result of cortical spreading depression (CSD), a wave of neuronal depolarization that spreads slowly across the cerebral cortex (3, 4). Evidence from both animal models and humans indicates that the headache is a consequence of activation of trigeminovascular afferents innervating the meninges, causing neurogenic inflammation and activation of trigeminal nucleus caudalis and brainstem nuclei involved in the perception of pain (5, 6). Recently, Bolay et al. (7) have shown that CSD in the rat cortex activates trigeminovascular afferents and evokes a series of alterations in the meninges and brainstem consistent with the development of headache. These data, together with earlier work, point to CSD as the critical event in the pathogenesis of migraine with aura, but it remains unknown what would make the brain of patients more susceptible to CSD.

Insights may be gained from the study of familial hemiplegic migraine (FHM), a dominantly inherited subtype of migraine with aura associated with ictal hemiparesis. In half of the families tested, FHM is caused by missense mutations in CACNA1A (8), the gene encoding the pore-forming α₁-subunit of Ca_V2.1 (voltage-gated P/Q-type Ca²⁺) channels (9–12). Ca_V2.1 channels are located in presynaptic terminals throughout the brain (13) and play a prominent role in controlling neurotransmitter release at most synapses (14). Their localization in somatodendritic membranes points to additional postsynaptic roles (15–17). Studies of mutant Ca_V2.1 channels in heterologous expression systems revealed that FHM mutations alter both the single-channel biophysical properties and the density of functional channels in the membrane (18–20). Alterations in single-channel function included an almost general increase in open probability, a reduction in conductance (by some FHM mutations), and variable effects on channel inactivation. The density of functional channels in the membrane was reduced by most mutations but was increased by one mutation. Overall, the changes induced by the FHM mutations led to the prediction of a decreased or increased whole-cell Ca²⁺ influx, depending on the mutation (18).

To investigate the possibility of changes in Ca_V2.1 function common to all FHM mutations, we calculated the product of single-channel current and open probability as a measure of the Ca²⁺ influx through single human Ca_V2.1 channels and extended the previous single-channel analysis to mutation V1457L (20). We also expressed FHM mutants in Ca_V2.1-deficient neurons rather than HEK293 cells. We uncovered two common changes in Ca_V2.1 function caused by FHM mutations: an increased single-channel Ca²⁺ influx and a decreased maximal Ca_V2.1 current density in neurons, and discuss their relevance for the pathogenesis of migraine with aura.

Methods

Cell Culture and Transfection.

HEK293 cells were grown and cotransfected with human Ca_V2.1α₁ (α_1A-2), β_2e, and α_2bδ cDNAs, as described (18). The constructs encoding mutants T666M, V714A, I1815L, and R192Q were those described in (18). Mutation V1457L was introduced into the human Ca_V2.1α₁ cDNA described in ref. 21 and differs from α_1A-2 by one splice deletion (V⁷²⁶EA). To construct the mutant clone V1457L, fusion PCR was applied to generate an EcoRI–KpnI (nucleotides 4319–4450) fragment carrying a G-to-C (nucleotide 4366) substitution. This PCR fragment was added to an AatII–EcoRI (nucleotides 2875–4319) subclone of Ca_V2.1α₁ inserted in pSport-1 (Invitrogen). Finally, the SplI–KpnI (nucleotides 3039–4450) fragment of this new subclone was ligated into the corresponding sites of Ca_V2.1α₁ in pGFP⁻ (21). Cells transfected with α_1A-2 were incubated at 28°C for 14–20 h before recording, because this procedure increased expression (18). Incubation at 28°C was not necessary for cells transfected with α_1A-2(-VEA), given the higher expression level. Identical normalized whole-cell current-voltage (I-V) relationships, and also identical single-channel current, i, and open probability, po, as a function of voltage, were measured in cells expressing the α_1A-2 and α_1A-2(-VEA) subunits. The data were then pooled together.

Cerebellar granule cells were grown in primary culture from 6-day-old Ca_V2.1α₁^−/− mice as described (12). Experiments were performed on cells grown from 5 to 7 days in vitro. Cells were transfected with wild-type (wt) or mutant human α_1A-2 cDNA at day 2 by using the modified calcium phosphate procedure described in ref. 22, except for a higher concentration of DNA per plate (for each 35-mm diameter plate, 4 and 8 μg for the reporter GFP and the α₁-subunit, respectively), and for the absence of ionotropic glutamate receptor inhibitors in the transfection medium (Eagle's minimal essential medium with Hanks' salts/25 mM Hepes, pH 7.85). Typically, the incubation with the DNA/calcium phosphate precipitate was stopped 15 min after precipitate was formed.

Patch-Clamp Recordings and Data Analysis.

Whole-cell patch-clamp recordings were performed as in ref. 12 and single-channel recordings as in ref. 18. Currents were sampled at 5 kHz and low-pass filtered at 1 kHz.

External solution for whole-cell recordings was 5 mM BaCl₂ (or CaCl₂), 148 mM tetraethylammonium (TEA)-Cl, 10 mM Hepes (pH 7.4 with TEA-OH) (and 5 μM nimodipine and 0.1 mg/ml cytochrome C, when recording from neurons). Internal solution was 100 mM Cs-methanesulfonate/5 mM MgCl₂/30 mM Hepes/10 mM EGTA/4 mM ATP/0.5 mM GTP/1 mM c-AMP (pH 7.4 with CsOH). Compensation (typically 70–80%) for series resistance was used, and I-V curves were obtained only from cells with a voltage error <5 mV. The average normalized I-V curves were multiplied by the average maximal current density obtained from all cells. I-V curves in HEK cells were obtained with voltage ramps (0.85 mV/msec) from a holding potential of −80 mV. I-V ramps for leak subtraction were obtained after blockade of Ca²⁺ channels with 5–10 mM Ni²⁺. I-V curves were fitted with Eq. 1:

1

Single-channel recordings were obtained in cell-attached configuration. Pipette solution was 90 mM BaCl₂/10 mM TEA-Cl/15 mM CsCl/10 mM Hepes or 10 mM BaCl₂ (or CaCl₂)/130 mM TEA-Cl/15 mM CsCl/10 mM Hepes (pH 7.4 with TEA-OH). Bath solution was 140 mM K-gluconate/5 mM EGTA/35 mM l-glucose/10 mM Hepes (pH 7.4 with KOH). Quartz pipettes were used with 10 mM Ca²⁺ as charge carriers. Single-channel currents and open probabilities were obtained as in ref. 18.

The density of functional channels in the membrane for mutant V1457L relative to that of wt channels was obtained from Eq. 2:

2

with the current densities, I_mt and I_wt, and the products of single-channel current and open probability, i_mtpo_mt and i_wtpo_wt, measured at the peak of the I-V and ipo-V curves obtained for mutant and wt channels, respectively.

In both whole-cell (12) and single-channel recordings (90 mM Ba²⁺), the liquid junction potentials were such that a value of 12 mV should be subtracted from all voltages to obtain the correct values of membrane potential. Averages are given ± SEM.

Results

Human α_1A-2-subunits containing the five different mutations linked to FHM shown in Fig. 1 were coexpressed with human α_2bδ- and β_2e-subunits in HEK293 cells. Fig. 1A shows the relative whole-cell Ba²⁺ current densities measured as a function of voltage in cells expressing wt channels (wt) or Ca_V2.1 mutants containing mutations T666M (TM) and V1457L (VL), located close to glutamate residues forming the high-affinity binding site for Ca²⁺ ions in the pore. Near the threshold of activation, the current density of the VL mutant was similar to wt, whereas it was much smaller at higher voltages. Single-channel recordings revealed that a decreased single-channel current, i, and conductance (from 19.6 ± 0.2 pS, n = 19 to 12.6 ± 0.2 pS, n = 11) contribute to the smaller current density; on the other hand, a shift to more negative voltages of the voltage-dependence of the channel open probability, po, accounts for the current density similar to wt at low voltages (Fig. 1A). Given the similar maximal po of wt and VL channels and the less than 50% decreased unitary conductance of VL, a reduction of the density of functional channels in the membrane must contribute to the almost 70% decrease of the maximal whole-cell current density of the mutant. The estimated density of functional mutant channels with respect to wt channels was 0.45 ± 0.10 (from Eq. 2 in Methods). As reported for human Ca_V2.1 channels containing the β_3a-subunit (18), the whole-cell current density of the TM mutant was smaller than wt at all voltages, as a consequence of an even smaller unitary current and conductance and lower density of functional channels in the membrane than VL (18). For both pore mutants the reversal potential, E_rev, of the Ba²⁺ current was 15 mV smaller than for wt, indicating that both mutations decrease the selectivity of the channel for Ca²⁺ with respect to monovalent ions (compare similar effect on E_rev with Ca²⁺ as charge carrier; Fig. 2).

Figure 1.

FHM mutations increase Ba²⁺ influx through single human Ca_V2.1 channels. (A Top Left) Whole-cell recordings with 5 mM Ba²⁺ as charge carrier. Whole-cell current densities during voltage ramps were measured in HEK293 cells expressing human wt and mutant V1457L (VL) and T666M (TM) Ca_V2.1 channels. Maximal current densities in cells transfected with α_1A-2 and α_1A-2 (-VEA) cDNAs were 70.6 ± 12.0 pA/pF, n = 52, and 71.6 ± 11.3 pA/pF, n = 31, respectively. Identical normalized current density-voltage (I-V) curves were measured for the two wt isoforms (n = 9 and n = 12, respectively). Maximal current densities of the α_1A-2TM and α_1A-2 (-VEA) VL mutants were 10.7 ± 0.9 pA/pF (n = 53) and 22.7 ± 3.1 pA/pF (n = 32), respectively. The I-V curves of the mutants were divided by the maximal current density of the corresponding wt channels. (A Bottom Left) Single-channel cell-attached recordings on HEK cells expressing wt and mutant VL channels; 90 mM Ba²⁺ as charge carrier. Holding potential = −80 mV. Representative current traces from single-channel patches containing a wt Ca_V2.1 channel (cell X24A) or a VL mutant (cell A07B). (Bars = 0.5 pA, 40 ms.) (A Right) Unitary current, i, and open probability, po, as a function of voltage, for wt and VL mutant channels. (Top) Average I-V curves from 19 (wt) and 11 (VL) patches. (Inset) Unitary activity at 20 mV on an expanded time scale. (Bars = 0.3 pA, 10 ms.) (Bottom) Average po-V curves from 12 (wt) and 10 (VL) patches containing a single channel. The data points were best fitted by Boltzmann distributions with V_1/2 = 34.5 mV (k = 6.2) for wt and V_1/2 = 27.4 mV (k = 7.2) for VL. (B) Single-channel cell-attached recordings (90 mM Ba²⁺) on HEK cells expressing wt and mutant TM, VL, R192Q (RQ), V714A (VA), and I1815L (IL) Ca_V2.1 channels. (Top) Product of i and po, as a function of voltage, for the wt channel and the five FHM mutants. All ipo values were divided by the maximal ipo value of the wt channel (−0.21 ± 0.01 pA). Values of i and po for RQ, VA, and IL mutants, and i for TM, were taken from ref. 18; po values for TM were obtained as described in the text. For the wt channel, the normalized ipo values were well fitted by the normalized whole-cell I-V curve, shifted by 26 mV toward more positive voltages (thick line). This shift accounts for the difference in surface potential in the single-channel and whole-cell recording solutions (90 vs. 5 mM Ba²⁺). (Bottom) Comparison of ipo values for the VL mutant, calculated by using either the po values measured in cell-attached recordings (○) or the po values derived from the wt po-V curve in Fig. 1A, shifted in the hyperpolarizing direction of 7.1 mV, corresponding to the difference in V_1/2 activation estimated from fitting the whole-cell current densities of wt and VL channels with Eq. 1 (▴).

Figure 2.

The T666M mutation increases more the single-channel Ca²⁺ than the Ba²⁺ influx. (A Left) Whole-cell recordings with 5 mM Ba²⁺ or 5 mM Ca²⁺ as charge carriers, on HEK cells expressing wt Ca_V2.1 channels. Voltage ramps were applied after perfusion of each cell with Ca²⁺ and then again Ba²⁺ solution. The Ca²⁺ and Ba²⁺ current curves were normalized to the maximal value of the Ba²⁺ current in each cell (n = 9). (Inset) Representative traces of Ca²⁺ and Ba²⁺ current recorded at 20 mV (Cell S93C). (Bars = 100 pA, 20 ms.) (A Right and traces) Single-channel cell-attached recordings with 10 mM Ba²⁺ or 10 mM Ca²⁺ as charge carrier, from HEK cells expressing wt channels. Holding potential = −80 mV. Average I-V relationships (n = 8 with Ba²⁺ and n = 4 with Ca²⁺; slope conductances: g = 14.7 and 7.2 pS) and representative current traces at 10 mV of a wt Ca_V2.1 channel with Ba²⁺ (Left, cell R43C) and Ca²⁺ (Right, cell R37A) as charge carrier. (Bars = 0.5 pA, 40 ms.) (B) Whole-cell recordings as in A but on cells expressing the TM mutant (n = 11). (Inset) Representative traces recorded at 10 mV (cell S85F). (Bars = 50 pA, 20 ms.) (C) Single-channel Ca²⁺ influx of the TM mutant relative to that of the wt channel, as a function of voltage, was estimated as follows. The whole-cell Ca²⁺ (5 mM) current density of the TM mutant relative to that of wt was divided by the density of functional TM mutants in the membrane relative to that of wt, N_mt/N_wt = 0.31 ± 0.03 (obtained in two independent ways, see text). The Ca²⁺ current densities as a function of voltage were obtained by multiplying the average normalized Ca²⁺ currents recorded during voltage ramps (as in A) by the average peak Ca²⁺ current density of 38.8 ± 6.4 pA/pF for the wt (55 ± 2% of the Ba²⁺ current density in 52 cells) and 8.0 ± 0.7 pA/pF for TM (75 ± 1% of the Ba²⁺ current density in 53 cells).

If one considers the macroscopic current densities in Fig. 1 and those obtained by Hans et al. (18) for mutations V714A (VA), I1815L (IL), and R192Q (RQ), one would not be able to predict a common functional effect of FHM mutations on Ca²⁺ influx. In fact, the Ba²⁺ current density may be larger or smaller than wt at any voltage, depending on the mutation. Nor can one find a common denominator among the different mutations if one considers their effect on inactivation properties (18–20). On the other hand, because a common property of FHM mutants seems to be an increased po over a broad voltage range (Fig. 1A and ref. 18), we wondered whether, despite the decreased unitary conductance of VL (Fig. 1A) and of both TM and VA mutants (18), a common functional effect of the FHM mutations was an increased single-channel Ca²⁺ influx, as given by the product of single-channel current and open probability, ipo.

Fig. 1B shows the ipo values calculated as a function of voltage for the wt and the five mutant channels. Indeed, at low voltages, around the threshold of activation, the single-channel Ba²⁺ influx, as measured by the product ipo, was larger than wt for all FHM mutants. Thus, even for the two pore mutants (TM and VL) with considerably reduced single-channel conductance, the shift of channel activation to more negative voltages produces an increase in po that is more than sufficient to compensate for the reduction in unitary current over a broad voltage range. Moreover, all FHM mutants exhibit a significant single-channel Ba²⁺ influx at voltages where the open probability of wt channels is insignificant. For VA mutants, similar ipo values were obtained for the prevalent mutants with reduced conductance and for the minority of mutants with wt conductance, given the lower po of the latter (ref. 18, and data not shown).

Because the very small current (i = 0.3 pA at 30 mV) and fast gating of the TM mutant precluded a reliable measurement of its po as a function of voltage, the po values for TM were obtained in an indirect manner by combining single-channel and whole-cell data. The current density curves in Fig. 1A revealed that the TM mutant activates at more negative voltages than the wt channel, and a 6.5-mV difference in V_1/2 activation was estimated from fitting the curves with Eq. 1 in Methods. A po-V curve for the TM mutant was obtained by shifting 6.5 mV in the hyperpolarizing direction the po-V curve of the wt channel; the po values thus obtained were multiplied by the measured unitary current of the TM mutant at each voltage to calculate the ipo values shown in Fig. 1B. The validity of this method was tested by applying it to the VL mutant: Fig. 1B Bottom shows a very good agreement between the measured and calculated ipo values. The good agreement also implies that the effect of the mutation on channel activation is independent of permeant ion concentration. The same is true for its effect on unitary conductance, because a similar decrease was measured with 90 and 10 mM Ba²⁺ (not shown). To infer the voltage-dependence of the single-channel Ba²⁺ influx with 5 mM Ba²⁺ as charge carrier, one can then simply shift the curves in Fig. 1B of 26 mV in the hyperpolarizing direction to account for the larger surface potential (see Fig. 1 legend).

The data in Fig. 2 suggest that the TM mutation increases more the single-channel Ca²⁺ than Ba²⁺ influx. With Ca²⁺ (5 mM), the maximal whole-cell current of wt channels was 45 ± 2% (n = 9) smaller than with Ba²⁺ (Fig. 2A), mainly as a consequence of a smaller single-channel conductance. Indeed, the single-channel conductance of Ca_V2.1 channels was 14.7 ± 0.2 pS (n = 8) with 10 mM Ba²⁺ and 7.3 ± 0.2 pS (n = 4) with 10 mM Ca²⁺ (Fig. 2A). A similar decrease in conductance can be inferred from the ascending parts of the whole-cell I-V curves (fitting with Eq. 1, gives G_Ca/G_Ba = 0.51). For the TM mutant, the maximal Ca²⁺ current was only 25 ± 1% (n = 11) smaller than the Ba²⁺ current, and the I-V fits gave G_Ca/G_Ba = 0.76 (Fig. 2B). This finding is consistent with the conclusion that, with respect to the wt channel, the mutant has a larger unitary conductance for Ca²⁺ relative to Ba²⁺, and, therefore, that mutation TM decreases less the single-channel conductance for Ca²⁺ than for Ba²⁺ ions. Single-channel recordings, recently performed with 90 mM Ca²⁺ as charge carrier, confirmed this conclusion (not shown).

The single-channel Ca²⁺ influx of the TM mutant relative to that of the wt channel, with nearly physiological concentrations of Ca²⁺ ions, was estimated by combining whole-cell and single-channel recordings, as follows. The whole-cell Ca²⁺ current density of the TM mutant relative to that of the wt channel was divided by the density of functional channels in the membrane of the mutant relative to that of wt, N_mt/N_wt (Fig. 2C). Hans et al. (18) obtained N_mt/N_wt = 0.37 ± 0.02 by counting the number of channels per patch in cell-attached recordings. A lower value (N_mt/N_wt = 0.25 ± 0.06) was obtained by using Eq. 2 and the data for TM in Fig. 1. The two values were averaged to obtain Fig. 2C. With either value, the estimated single-channel Ca²⁺ influx of the TM mutant was much larger than wt over a broad voltage range. Thus, our data support the conclusion that the FHM mutations increase Ca²⁺ influx through single human Ca_V2.1 channels.

As mutational changes of functional channel densities in HEK cells may not reflect changes occurring in neurons, we expressed wt and mutant human Ca_V2.1α₁ (α_1A-2) subunits in cerebellar granule cells in primary culture from Ca_V2.1α₁^−/− mice. We have previously shown that these neurons completely lack P/Q Ca²⁺ channels and up-regulate both N- and L-type but not R-type channels, as compared with control (12). The Ca_V2.1 current density in transfected neurons was obtained from the amount of whole-cell current inhibited by ω-CTx-MVIIC applied after the specific blocker of N-type channels ω-CgTx-GVIA, as in ref. 12. In neurons expressing the wt human α_1A-2-subunit, a maximal Ca_V2.1 current density of 24.6 ± 2.4 pA/pF (n = 32) was measured at −10 mV (Fig. 3), which is a voltage more than 20 mV more negative than that of the maximal current in HEK cells cotransfected with the same α_1A-2-subunit and human α_2bδ- and β-subunits, and which is similar to that of the maximal endogenous P/Q current in Ca_V2.1^+/+ neurons (ref. 12 and Fig. 4, which is published as supporting information on the PNAS web site, www.pnas.org). In neurons expressing mutant human α_1A-2-subunits, the Ca_V2.1 current density at −10 mV was invariably smaller (Fig. 3A). Whereas the reduction of maximal current density in neurons was similar to that measured in HEK cells for the TM, VA, and IL mutants, the reduced maximal current density of the RQ mutant is in striking contrast with the 93% larger density in HEK cells (18).

Figure 3.

Functional effects of FHM mutations in Ca_V2.1α₁^−/− neurons expressing human Ca_V2.1α₁-subunits. Whole-cell recordings (5 mM Ba²⁺) on cerebellar granule cells of Ca_V2.1α₁^−/− mouse transfected with wt or mutant human Ca_V2.1α₁ (α_1A-2) subunits. (A Left) Plot of peak current vs. time for a representative experiment on a neuron transfected with wt α_1A-2-subunits (cell U280B). Depolarizations at −10 mV were delivered every 10 s from −80 mV. ω-CgTx-GVIA (1 μM) and ω-CTx-MVIIC (3 μM) were sequentially applied in the continuous presence of nimodipine (5 μM). (Inset) Representative traces at increasing voltages (from −50 to −10 mV) taken at times a and b. Ca_V2.1 currents were obtained as the difference between traces at times a and b. (Bars = 20 ms, 50 pA.) (A Right) Average density of Ca_V2.1 current at −10 mV in neurons transfected with the wt human α_1A-2-subunit (n = 32) or the RQ (n = 22), TM (n = 19), VA (n = 14), and IL (n = 12) mutants, from 7, 5, 3, 2, and 4 neuronal cultures, respectively. (Inset) Corresponding pooled Ca_V2.1 current traces (n = 10 for wt, n = 10 RQ, n = 8 TM, n = 6 VA, n = 6 IL). (Bars = 20 ms, 5 pA/pF.) (B) Average normalized Ca_V2.1 current (Left) and average Ca_V2.1 current density (Right) as a function of voltage in neurons expressing wt α_1A-2 (n = 8), RQ (n = 10), VA (n = 5), or IL (n = 5) mutant subunits. For comparison, the normalized I-V curve measured in HEK cells transfected with the same human α_1A-subunit (and human α_2bδ- and β_2e-subunits) is shown as a dotted line. (C) Single-channel Ba²⁺ influx of the RQ, VA, and IL mutants relative to that of the wt channel, (ipo)_mt/(ipo)_wt, as a function of voltage, obtained from the current densities in B, as described in the text.

In neurons expressing mutant human α_1A-2-subunits, the voltage dependence of the Ca_V2.1 current was invariably shifted toward hyperpolarized voltages with respect to that in neurons expressing the wt subunit (Fig. 3B). As a consequence, the current densities of the mutants were similar to wt at low voltages. Fitting of the normalized I-V curves gave V_1/2 values of −17.8 ± 1.0 (n = 8), −21.9 ± 1.6 (n = 9), −29.3 ± 0.8 (n = 5) and −31.3 ± 1.9 (n = 5) mV for wt, RQ, VA, and IL, respectively. Although the very small P/Q current density of the TM mutant precluded a reliable measurement of a complete I-V curve and an estimation of V_1/2, its maximal current density was measured at 5 mV more negative voltage than wt (not shown), suggesting activation at lower voltages also for this mutant. The negatively shifted activation of the mutants expressed in granule cells implies that, also in neurons, a common functional effect of FHM mutations is to increase the open probability of Ca_V2.1 channels in a broad voltage range. An increased Ca²⁺ influx through single mutant channels in neurons can be inferred from the values of (ipo)_mt/(ipo)_wt larger than 1, as in Fig. 3C. These values were calculated at each voltage from the ratio of the current densities of mutant and wt channels, I_mt/I_wt, divided by the ratio of the density of channels in the membrane, N_mt/N_wt (given that I = Nipo). If the mutations do not affect the maximal open probability, po_mx, N_mt/N_wt can be obtained from I_mt/I_wt measured at voltages where po = po_mx for mutants with the same unitary conductance as wt and from the same ratio divided by i_mt/i_wt for mutants with reduced conductance. Considering I_mt/I_wt at 0 mV (where po ≅ po_mx; see Fig. 5, which is published as supporting information on the PNAS web site), N_mt/N_wt values of 0.17 ± 0.03, 0.48 ± 0.08, and 0.27 ± 0.03 were obtained for IL, RQ, and VA, respectively. These values are upper limits, and the values of (ipo)_mt/(ipo)_wt in Fig. 3C are lower limits for the RQ and VA mutants if, as found in HEK cells, their po_mx is larger than wt, and a fraction of VA mutants has wt conductance (18). The validity of this method for deriving N_mt/N_wt and ipo_mt/ipo_wt is supported by the good agreement between the values obtained in HEK cells from whole-cell current densities and from single-channel recordings (see Fig. 5B).

The N- and R-type current components were similar in neurons from Ca_V2.1α₁^−/− mice expressing wt (N-type: 5.4 ± 0.5 pA/pF; R-type: 16.1 ± 1.1 pA/pF, n = 32) or mutant human α_1A-2-subunits (N-type: 6.0 ± 0.7, 7.4 ± 0.9, 6.5 ± 0.9, 5.1 ± 0.4; R-type: 16.5 ± 1.4, 14.8 ± 1.6, 17.5 ± 1.6, 16.0 ± 1.2 pA/pF, with RQ: n = 22, TM: n = 19, VA: n = 14, and IL: n = 12, respectively), and both components were not significantly different from those measured in the same neurons from Ca_V2.1^+/+ mice (6.1 ± 0.4 and 18.2 ± 1.1 pA/pF, n = 23).

Discussion

The findings reported in this article support the following conclusions.

1.

FHM mutations increase the Ca²⁺ influx through single human Ca_V2.1 channels. A common functional effect of FHM mutations is to shift the activation curve of Ca_V2.1 channels to hyperpolarized voltages and thus increase their open probability. This result has two important consequences: (i) an increased single-channel Ca²⁺ influx in a large voltage range, despite the reduction in unitary current produced by some mutations; (ii) Ca²⁺ influx through mutant channels in response to small depolarizations insufficient to open wt channels. FHM mutants expressed in neurons from Ca_V2.1α₁^−/− mice or HEK cells showed similar alterations in single-channel function.

2.

Another common functional effect of FHM mutations is to decrease the density of functional Ca_V2.1 channels and the maximal Ca_V2.1 current density in cerebellar granule cells from Ca_V2.1α₁^−/− mice expressing human Ca_V2.1α₁-subunits. Current densities were similar to wt close to the threshold of activation as a consequence of activation at more negative voltages of the mutant channels.

3.

The changes in the density of functional channels in the membrane caused by the FHM mutations can be different (even opposite, in the case of RQ) depending on whether the mutant is expressed in a neuron or in an HEK cell. Based on this observation, it cannot be ruled out that changes might be different even among different types of neurons.

The importance of the neuronal environment for studying Ca_V2.1 channels is stressed by the striking (>20 mV) difference in the voltage range of activation of human Ca_V2.1 channels expressed in neurons or HEK cells. It is unlikely that different auxiliary subunits (mouse vs. human) are responsible for the shift, because only minor changes in channel activation have been measured with different auxiliary subunits in heterologous expression systems (23, 24). Clearly though, some interacting protein and/or biochemical process, present in neurons but not in HEK cells, strongly affects the activation of Ca_V2.1 channels.

Our data point to two apparently contradictory functional effects common to all FHM mutations analyzed: an increased Ca²⁺ influx through single Ca_V2.1 channels, with relatively small depolarizations (V < −10 mV in neurons) but a decreased neuronal Ca_V2.1 current density with relatively strong depolarizations (V > −20 mV). How can these alterations of Ca_V2.1 channel function explain the pathogenesis of FHM and, in particular, lead to its typical episodic neurological symptoms, including aura and headache?

If CSD is the critical event in the pathogenesis of migraine with aura, as indicated by the recent findings of Bolay et al. (7) and earlier work, then the alterations in Ca_V2.1 channel function induced by FHM mutations should favor the spontaneous occurrence of CSD in the brain of patients. Although the mechanisms for initiation and propagation of CSD are not completely understood, it is clear that an increased neuronal excitability precedes the regenerative depolarization typical of CSD (4). The importance of P/Q type channels in the initiation and spread of CSD has been revealed by Ayata et al. (25), who found a striking elevation of the threshold for initiating CSD and a marked reduction in glutamate levels released by K⁺-induced depolarization in the neocortex of mouse mutants with spontaneous Ca_V2.1α₁ mutations (tg and tg^la). Tg^la mutants also showed a slower velocity and frequent failure of propagation of CSD and a much larger reduction of excitatory with respect to inhibitory neurotransmitter release. It has been shown that tg and tg^la mutations lead to a decreased P/Q current density in both native cerebellar neurons and in heterologous expression systems, and that the tg^la mutation shifts the activation curve of Ca_V2.1 channels to depolarized voltages and reduces their single-channel open probability (26–28). These data support the conclusion that a reduced Ca²⁺ entry through Ca_V2.1 channels reduces neuronal cortical network excitability and makes the cortex more resistant to CSD. It seems unlikely, then, that the functional defect relevant to explain enhanced CSD susceptibility in FHM patients is the reduced neuronal Ca_V2.1 current density of FHM mutants. Instead, the functional defect relevant for FHM pathogenesis (or at least its aura phase) could be the increased Ca²⁺ influx through single Ca_V2.1 channels that clearly distinguishes the FHM from the tg^la mutations. As argued below, this increase could lead to increased local Ca²⁺ influx at presynaptic active zones, despite a global reduction of Ca²⁺ entry in the cell.

Neurotransmitter release at each release site is controlled by a cluster of only a few Ca²⁺ channels located sufficiently close to the Ca²⁺ sensor to contribute to the local Ca²⁺ increase that triggers release in response to single-action potentials (29). Ca_V2.1 channels are generally more effective than other Ca²⁺ channel types in triggering release (30–32) as a consequence of their preferential localization at the release sites (31). Binding of Ca_V2.1 channels to syntaxin and synaptosomal-associated protein of 25 kDa contribute to this preferential localization (31, 33). It seems unlikely, then, that the number of Ca_V2.1 channels at the release sites would be decreased by FHM mutations in a similar manner as the density of channels in the soma. If a similar number of Ca_V2.1 channels are located at the release sites in wt and mutant synapses, the hyperpolarized activation and increased ipo of mutant channels may well lead to an increased action potential-evoked Ca²⁺ influx at the active zones (34) and to a large increase in neurotransmitter release (given the supralinear dependence of release on Ca²⁺ influx and the nonsaturation of the Ca²⁺ sensor during an action potential; ref. 35). If this is the case, and if, as suggested by the data on tg and tg^la mice (25, 36), the control of release by Ca_V2.1 channels is larger in excitatory than inhibitory synapses, then the FHM mutations are expected to increase neuronal cortical network excitability and make the cortex more susceptible to CSD.

There is some clinical evidence that the migraine aura and the headache are not necessarily sequential, and that the aura may not be the trigger for the pain (37). Functional imaging of patients has provided evidence for a role of the brainstem, particularly the periaqueductal gray (PAG) region, in attacks of migraine without aura (38, 39). Inhibition of P/Q-type channels in PAG has been shown to facilitate trigeminal nociception, thus pointing out a role of these channels in the descending pain-inhibitory system that regulates the perception of pain (40). It is not clear whether the localization of the relevant P/Q channels is presynaptic or postsynaptic, and the mechanism by which their inhibition leads to facilitation (or rather disinhibition) of the second order trigeminal neurons remains unknown. If FHM aura and headache are not sequential but parallel processes, one might even consider the hypothesis that the two apparently contradictory functional effects of FHM mutations may actually underlie the two parallel processes: a decreased Ca_V2.1 current density in the PAG may contribute to the headache pathogenesis, whereas an increased Ca²⁺ entry through single Ca_V2.1 channels in synaptic terminals in the cortex may underlie the aura.

FHM knock-in mice will allow the testing of these different hypotheses.

Abbreviations

CSD

cortical spreading depression

FHM

familial hemiplegic migraine

I-V

current–voltage relationship

wt

wild type

TEA

tetraethylammonium