Three New Familial Hemiplegic Migraine Mutants Affect P/Q-type Ca 2 1 Channel Kinetics*

Missense mutations in the pore-forming human a 1A subunit of neuronal P/Q-type Ca 2 1 channels are associated with familial hemiplegic migraine. We studied the functional consequences on P/Q-type Ca 2 1 channel func-tion of three recently identified mutations, R583Q, D715E, and V1457L after introduction into rabbit a 1A and expression in Xenopus laevis oocytes. The potential for half-maximal channel activation of Ba 2 1 inward cur rents was shifted by > 9 mV to more negative potentials in all three mutants. The potential for half-maximal channel inactivation was shifted by > 7 mV in the same direction in R583Q and D715E. Biexponential current inactivation during 3-s test pulses was significantly faster in D715E and slower in V1457L than in wild type. Mutations R583Q and V1457L delayed the time course of recovery from channel inactivation. The decrease of peak current through R583Q (30.2%) and D715E (30.1%) but not V1457L (18.7%) was more pronounced during 1-Hz trains of 15 100-ms pulses than in wild type (18.2%). Our data demonstrate that the mutations R583Q, D715E, and V1457L, like the previously reported mutations T666M, V714A, and I1819L, affect P/Q-type Ca 2 1 channel Electrophysiological Ba (I

ataxia type 6, and familial hemiplegic migraine (FHM) with and without cerebellar ataxia. These mutations may provide important insight into how altered Ca 2ϩ signaling and neuronal excitability can lead to neurodegeneration and episodic neurological diseases such as migraine.
Four nonsense mutations (6 -8), three splice site mutations, and four deletions in CACNA1A (5,7) have been found to segregate in patients with EA-2. Small CAG expansions were observed in a large series of patients with spinocerebellar ataxia type 6 (9), and a further CACNA1A missense mutation was identified in a patient with severe progressive ataxia (10). At least seven missense mutations have been identified in families with FHM (5,(11)(12)(13). Defects in the ␣ 1A gene are also responsible for the phenotypes (absence epilepsy and ataxia) of tottering (tg) and leaner (tg la ) mutant mice (14) and may also occur in more common forms of migraine (15).
The mechanisms by which these mutations cause these abnormal phenotypes is unclear. Mutations causing EA-2, an autosomal dominant disease, are predicted to give rise to truncated, presumably nonfunctional ␣ 1A proteins. 2 This must result in a partial loss of P/Q-type Ca 2ϩ channel function.
In contrast, we (16) and others (17) have shown recently that ␣ 1A missense mutations causing FHM do not prevent channel activity. FHM is a rare autosomal dominant form of migraine with aura, associated with ictal hemiparesis and, in some families, with cerebellar ataxia and atrophy (18). Functional expression of rabbit ␣ 1A subunits containing the FHM mutations T666M, V714A, and I1811L revealed mutation-induced changes in gating kinetics altering the extent to which P/Qtype channels accumulate in inactivation during trains of depolarizing pulses. We therefore proposed that this could alter Ca 2ϩ influx and signaling during episodes of high neuronal activity. This in turn might result in a long term activation of neurons within the proposed "migraine generator" in the brainstem discovered by brain imaging in migraine patients (19).
Essentially the same changes in gating kinetics were reported by Hans et al. (17) after introduction of the same FHM mutations in human ␣ 1A followed by heterologous expression in human embryonic kidney 293 cells and patch clamp analysis. In addition, they reported mutation-induced changes in single channel kinetics and expression density.
In the current study we examined the functional effects of three recently published FHM mutations, R583Q, D715E, and V1457L (11)(12)(13) to address further the important questions of whether all FHM mutations yield functional Ca 2ϩ channels and if altered channel gating represents a key pathophysiological principle in FHM as proposed from initial studies.
Single mutants R583Q and D715E were constructed according to the previously described procedure for generation of single mutants T666M and V714A (16).
Mutation V1457L (corresponding to V1465L in rabbit ␣ 1A , see legend to Fig. 1) as well as a silent mutation (codon for rabbit Asp-1545 GAC to GAT) were introduced simultaneously into rabbit class A cDNA by polymerase chain reaction to yield a ClaI restriction sequence. The mutated polymerase chain reaction fragment was cut SfiI (4290)-ClaI* (4925) and coligated with a NheI (3543)-SfiI (4290) fragment of BI-II into AL20 NheI (3543)-ClaI (homologous position to ClaI*) (22) to yield complete rabbit ␣ 1A cDNA sequence.
All polymerase chain reaction-generated fragments were sequenced completely to confirm sequence integrity.
Expression of ␣ 1A Mutants in Xenopus laevis Oocytes-Preparation of stage V-VI oocytes from X. laevis and injection of cRNA are described in detail elsewhere (21). Capped run-off poly(A) ϩ cRNA transcripts from XbaI-linearized cDNA templates were synthesized according to the procedures of Krieg and Melton (23). ␣ 1 cRNAs were coinjected with ␤ 1a (24) and ␣ 2 -␦ (25) subunit cRNAs. To exclude effects of endogenous Ca 2ϩ -activated Cl Ϫ currents on current kinetics, experiments were carried out in oocytes previously injected with 50 -100 nl of a 0.1 M BAPTA solution.
Electrophysiological Recordings-Inward Ba 2ϩ currents (I Ba ) through expressed channel complexes were measured using the twomicroelectrode voltage clamp technique as described previously (21). Similar current amplitudes were obtained with mutant and wild-type ␣ 1A subunits. Oocytes expressing peak I Ba smaller than 400 nA or larger than 1.8 A were excluded from analysis. Data analysis and acquisition were performed by using the pClamp software package (version 6.0, Axon Instruments).
Recordings were carried out at room temperature in a bath solution containing 40 mM Ba(OH) 2 , 50 mM NaOH, 2 mM CsOH, 5 mM HEPES, adjusted to a pH of 7.4 with methanesulfonic acid. Voltage recording and current injecting microelectrodes were filled with 2.8 M CsCl, 0.2 M CsOH, 10 mM EGTA, 10 mM HEPES (adjusted to pH 7.4 with HCl) and had resistances of 0.3-2 megohm.
Recovery of I Ba from inactivation was studied using a double-pulse protocol. After a 3-s depolarizing prepulse to ϩ10 mV (holding potential Ϫ80 mV) the time course of I Ba recovery was determined at Ϫ60 mV by applying 300-ms test pulses to ϩ10 mV at various time intervals after the prepulse. Peak I Ba was normalized to the peak current amplitude measured during the prepulse. I Ba was then allowed to recover for 1 min at Ϫ100 mV. This double-pulse protocol was repeated individually for each recovery time interval in the same oocyte.
The voltage dependence of inactivation (steady-state inactivation) was determined from normalized inward currents elicited during steps to ϩ10 mV after 10-s steps to various holding potentials. The voltage dependence of activation was determined from I-V curves obtained by step depolarizations from a holding potential of Ϫ80 mV to various test potentials. The half-maximal voltage for activation (V 0.5,act ), the slope factor of the curve at V 0.5,act (k act ,), the half-maximal voltage for steadystate inactivation (V 0.5 , inact ), and the slope factor of the curve (k inact ) were obtained by fitting the data to the Boltzmann equation. Apparent reversal potentials were calculated by extrapolation from I-V relationships.
Data Analysis-Nonlinear least square fitting and statistical calculations were performed using Origin R 5.0 (Microcal). Data are given as means Ϯ S.E. for the indicated number of experiments.

RESULTS
Mutations R583Q, D715E, and V1457L, illustrated in Fig.  1A, are located in highly conserved and functionally important regions of the human ␣ 1A subunit of neuronal P/Q-type Ca 2ϩ channels. Mutation R583Q neutralizes a positive charge in transmembrane segment S4 of the channel in domain II (IIS4). S4 segments form part of the voltage sensor of voltage-gated Ca 2ϩ channels (26). D715E is located in IIS6 adjacent to mutation V714A analyzed in our previous study and V1457L in the S5-S6 linker of domain III. Segments S5 and S6 and their connecting linkers are assumed to form the pore of the channel (26). We introduced the single mutations into the corresponding positions of the highly homologous rabbit ␣ 1A subunit (wild type, Ref. 1) and analyzed mutant channels for changes in their biophysical properties after functional expression in X. laevis oocytes (together with accessory ␤ 1 and ␣ 2 -␦ subunits) using the two-microelectrode voltage clamp technique.
The potential for half-maximal activation (V 0.5,act ) was significantly (p Ͻ 0.01) shifted to hyperpolarized potentials for all three mutants without changing the steepness of the steadystate activation curve (Table I). This effect was most pronounced in D715E. The midpoint voltage for steady-state inactivation was not altered in mutant V1457L, but a significant shift to more negative potentials occurred in R583Q and D715E (Table I). Apparent reversal potentials were similar for all constructs (53-59 mV) ruling out major changes in Ba 2ϩ permeability.
To investigate whether the FHM mutations affect the time course of channel inactivation we analyzed the current decay during 3-s test pulses elicited from a holding potential of Ϫ80 mV to ϩ10 mV (Fig. 1B). Next we tested whether the mutations also change the extent of peak I Ba decrease during pulse trains which reflects accumulation of channels in inactivation. Application of 15 100-ms pulses from a holding potential of Ϫ60 mV to a test potential of ϩ10 mV at a frequency of 1 Hz caused a significant (p Ͻ 0.01) increase of accumulation in inactivation for mutants R583Q and D715E but not V1457L. Current decay after 15 pulses was 1.6-fold larger in R583Q (30.2 Ϯ 0.8%; n ϭ 35) and D715E (30.1 Ϯ 1.5%; n ϭ 18) than in wild type (19.1 Ϯ 1%; n ϭ 19) (Fig.  2, A and B).
The fraction of channels inactivating during frequent depolarizations not only depends on the inactivation rate during the pulses but also on the rate of recovery from inactivation between pulses. Therefore recovery from inactivation was measured employing a double-pulse protocol (Fig. 3A). Channels were inactivated by a 3-s conditioning prepulse from Ϫ80 to ϩ10 mV. The time course of recovery was then determined at Ϫ60 mV by applying 300-ms test pulses to ϩ10 mV after various time intervals after the prepulse (Fig. 3A). Between single double-pulse experiments the oocytes were held at Ϫ100 mV for 60 s to allow full recovery of I Ba . Recovery was determined at Ϫ60 mV to maximize the difference between wild-type and mutant channels. In wild-type and mutant channels about 90% of I Ba recovered after 20 s. In all constructs recovery of I Ba followed a biexponential time course (Fig. 3B). In both R583Q and V1457L the fraction of recovered current at all time intervals measured was significantly smaller (p Ͻ 0.01) than in wild type. No change was observed for D715E (Fig. 3B).
In summary, our experiments convincingly show that all newly discovered ␣ 1A mutations in patients with FHM cause abnormal gating behavior of P/Q-type Ca 2ϩ channels. Gating changes therefore seem to represent an elementary mechanism underlying P/Q-type Ca 2ϩ channel dysfunction in FHM. DISCUSSION We have studied the functional consequences of three recently identified FHM missense mutations, R583Q, D715E, and V1457L within the ␣ 1A subunit of neuronal P/Q-type Ca 2ϩ channels. None of the mutations resulted in a nonfunctional channel as proposed for EA-2 mutations in the ␣ 1A subunit gene. EA-2 mutations (5, 7) are believed to be incompatible with the expression of a functional protein. In the presence of an unaffected gene, it is therefore likely that the observed

TABLE I Effects of mutations on activation and inactivation properties
The half-maximal voltage for activation (V 0.5,act ), the steepness of the curve at V 0.5,act (k act ), the half-maximal voltage for steady-state inactivation (V 0.5,inact ), and the steepness of the curve at V 0.5,inact (k inact ) were obtained by fitting the data to the Boltzmann equation. Data are means Ϯ S.E. for n ϭ 9 -26. Asterisks indicate a statistically significant (p Ͻ 0.01) difference from wild type. z g , apparent gating charge obtained by dividing 25.26 (ϭRT/F at 20°C)/k act . WT, wild type.
Ϫ5.5 Ϯ 0.8* Ϫ3.9 Ϯ 0.2 6.5 Ϫ27.6 Ϯ 0.7* 7.6 Ϯ 0.3 D715E Ϫ12 Ϫ8.5 Ϯ 1.1* Ϫ3.8 Ϯ 0.2 6.6 Ϫ17.1 Ϯ 0.6 6.0 Ϯ 0.2* neurological phenotype in EA-2 results from a reduced activity of P/Q-type Ca 2ϩ channels in the central nervous system. Instead, two independent mechanisms can affect P/Q-type currents in FHM patients: altered expression density and changes in channel gating. Hans et al. (17) have recently found that FHM mutations decrease or increase the density of functional P/Q-type currents after heterologous expression in Xenopus oocytes or mammalian cells. It is difficult to predict if these changes of expression density also occur in vivo where, in addition to the accessory ␣ 2 -␦ and ␤ subunits, ␣ 1A interacts with a number of other modulatory proteins such as G-proteins (27), calmodulin (28), and synaptic vesicle proteins (29). Clearly, this important question can only be addressed in animal models containing the respective mutations. A second mechanism by which FHM mutations can affect P/Q-type Ca 2ϩ currents is by changing channel gating. Our electrophysiological analysis provides convincing evidence that such changes also occur in three recently identified FHM mutations. Together with our previous results (16) this allows us to conclude that, irrespective of changes in expression density, this represents an elementary functional alteration underlying P/Q-type Ca 2ϩ channel dysfunction in FHM.
As for T666M, V714A, and I1811L, all three new mutations significantly shifted the voltage dependence of activation to more negative potentials. In the absence of changes in the slope of the activation curve this must result in a more negative threshold of Ca 2ϩ channel activation. This could lead to altered Ca 2ϩ signaling by increasing P/Q-type Ca 2ϩ channel activity at weak depolarizations. Two of the mutations also caused a more pronounced decrease of I Ba during pulse trains, reflecting altered accumulation of channels in inactivation. This can result from either increased inactivation during the pulse or delayed recovery from inactivation between pulses. Our experiments demonstrate that it is due to slower recovery from inactivation in R583Q and faster inactivation in D715E. In V1457L decrease of I Ba during the train was not different from wild type. This can be explained by the slower inactivation kinetics, which are counteracted by the slowed recovery from inactivation. Altered accumulation of channels in inactivation during rapid depolarizations could cause changes in Ca 2ϩ influx especially during high but not during low neuronal activity. This may underlie the episodic character of FHM with attacks triggered by sensory or emotional stimuli.
In addition to the potential insight into the pathophysiology of migraine, mutations R583Q and D715E also provide us with interesting molecular information about channel function. As in the previously analyzed mutant R192Q (16), R583Q eliminates a conserved positive charge at the extracellular side of transmembrane S4-helix, which forms part of the voltage sensor of the channel. The charge neutralization at position 583 in IIS4 (R583Q) shifted the voltage dependence of activation (and inactivation) to more negative potentials and slowed recovery from inactivation. These findings indicate that not only mutations in the putative pore region (T666M, V714A, I1811L) but also in the S4 segments can alter ␣ 1A recovery from inactivation. This illustrates that conformational changes of voltagesensing portions of ␣ 1A are involved in this process.
Mutation R583Q caused a negative shift of V 0.5,act without a change in the apparent gating charge, z g (Table I) By assuming a model in which the voltage sensors in all four repeats move independently it can be predicted that Arg-583 (in IIS4) forms part of a voltage sensor which moves over potentials close to those causing channel opening. This is in contrast to data reported earlier for an ␣ 1S (skeletal muscle)/ ␣ 1C (cardiac muscle) chimera where this was observed for sensors in repeats I and III but not in repeat II. Therefore this naturally occurring mutation clearly demonstrates that voltage sensor movements vary not only between different voltage-dependent cation channels (31) but even between different Ca 2ϩ channel ␣ 1 subunits.
Mutation D715E is located adjacent to mutation V714A. Together with I1811L in IVS6 these are believed to be located close to the cytoplasmic mouth of the pore. Unlike V714A and I1811L, D715E did not affect recovery from inactivation and prominently accelerated current inactivation upon depolarization. These data indicate that the cytoplasmic end of S6 helices comprise a functionally relevant region within ␣ 1A which tightly controls the channel's inactivation properties. Although our data do not allow us to propose a defined molecular mechanism for this process they clearly show that even minor structural changes such as the introduction of a single side chain methyl group in mutant D715E are sufficient to disturb this functional domain.
The present work clearly shows that mutations in the human CACNA1A gene alter the gating properties of neuronal P/Qtype Ca 2ϩ channels in all seven FHM mutants analyzed so far.

FIG. 3. Mutations affect I Ba recovery from inactivation. Panel
A, recovery from inactivation was measured using a double-pulse protocol. After a 3-s depolarizing prepulse to ϩ10 mV (holding potential Ϫ80 mV) the time course of I Ba was determined at Ϫ60 mV by applying 300-ms test pulses to ϩ10 mV at various time intervals after the prepulse. A representative trace of a recovery experiment for R583Q (cell R1205027) is illustrated. Panel B, fractional I Ba recovery from inactivation after various periods of time (0.05, 0.1, 0.2, 0.5, 1, 2, 3, 5, 10, 15, 20 s). Note that fractional recoveries of R583Q and V715L were statistically different from wild type (WT) (p Ͻ 0.01) at all time points tested. The inset shows fractional current recovery after selected time points (0.5, 3, and 15 s) for clarity. Statistical significance (p Ͻ 0.01) is indicated by asterisks. Means Ϯ S.E. are given for n ϭ 5-20.
This provides a rational basis for the generation of mutant mice containing selected mutations. Introduction of FHM mutations differing with respect to their biophysical properties should enable electrophysiological analysis of the consequences of altered channel gating for neuronal Ca 2ϩ signaling in FHM and more common forms of migraine (15).