Familial Hemiplegic Migraine Mutations Change α1ACa2+ Channel Kinetics*

Missense mutations in the pore-forming human α1A subunit of neuronal P/Q-type Ca2+channels are associated with familial hemiplegic migraine (FHM). The pathophysiological consequences of these mutations are unknown. We have introduced the four single mutations reported for the human α1A subunit into the conserved rabbit α1A(R192Q, T666M, V714A, and I1819L) and investigated possible changes in channel function after functional expression of mutant subunits inXenopus laevis oocytes. Changes in channel gating were observed for mutants T666M, V714A, and I1819L but not for R192Q. Ba2+ current (I Ba) inactivation was slightly faster in mutants T666M and V714A than in wild type. The time course of recovery from channel inactivation was slower than in wild type in T666M and accelerated in V714A and I1819L. As a consequence, accumulation of channel inactivation during a train of 1-Hz pulses was more pronounced for mutant T666M and less pronounced for V714A and I1819A. Our data demonstrate that three of the four FHM mutations, located at the putative channel pore, alter inactivation gating and provide a pathophysiological basis for the postulated neuronal instability in patients with FHM.

Missense mutations in the pore-forming human ␣ 1A subunit of neuronal P/Q-type Ca 2؉ channels are associated with familial hemiplegic migraine (FHM). The pathophysiological consequences of these mutations are unknown. We have introduced the four single mutations reported for the human ␣ 1A subunit into the conserved rabbit ␣ 1A (R192Q, T666M, V714A, and I1819L) and investigated possible changes in channel function after functional expression of mutant subunits in Xenopus laevis oocytes.
Changes in channel gating were observed for mutants T666M, V714A, and I1819L but not for R192Q. Ba 2؉ current (I Ba ) inactivation was slightly faster in mutants T666M and V714A than in wild type. The time course of recovery from channel inactivation was slower than in wild type in T666M and accelerated in V714A and I1819L. As a consequence, accumulation of channel inactivation during a train of 1-Hz pulses was more pronounced for mutant T666M and less pronounced for V714A and I1819A. Our data demonstrate that three of the four FHM mutations, located at the putative channel pore, alter inactivation gating and provide a pathophysiological basis for the postulated neuronal instability in patients with FHM.
␣ 1A subunits, in a complex with a ␤ and ␣ 2 ␦ subunit (1, 2), constitute the pore-forming subunit of neuronal voltage-gated P/Q-type Ca 2ϩ channels. This channel type is not only located on nerve cell bodies and dendrites but is also present in presynaptic terminals (3) where it controls depolarization-induced Ca 2ϩ influx tightly coupled to neurotransmitter release (4). Its gating properties are modulated by neurotransmitters (5,6) and affected by ␤ subunits in an isoform-specific manner (7,8). This suggests that a tight control of P/Q-type Ca 2ϩ channel activity is a prerequisite to fine tune its physiological function.
Missense mutations in the gene encoding human ␣ 1A (CACNL1A4) have recently been found to segregate with patients suffering from familial hemiplegic migraine (FHM) 1 (9), an autosomal dominant disorder. Although FHM represents a rare form of migraine, a detailed analysis of the functional consequences of this channelopathy may provide insight into the pathophysiology of migraine. Mutations in the CACNL1A4 gene could also underly more common forms of migraine with and without aura (10).
The pathophysiology of migraine remains to be fully understood and the mechanisms triggering an attack are unknown. Recent advances in brain imaging techniques (positron emission tomography and magnetic resonance spectroscopy) support a "primary neuronal theory" where attacks originate on the basis of a neuronal hyperexcitability of unknown origin (11)(12)(13)(14)(15). This may be the underlying cause of cortical spreading depression and hypoperfusion, phenomena associated with migraine attacks (12,13). Neuronal instability within central pain-modulating serotoninergic systems could not only serve as a "brainstem generator" of attacks but also initiate the headache and the events of neurogenic inflammation in the trigeminovascular system (12). It is therefore attractive to speculate that the four single ␣ 1A mutations found in FHM patients lead to such a neuronal instability by changes in P/Q-type Ca 2ϩ channel function.
As the direct analysis of changes in ␣ 1A Ca 2ϩ channel gating in human tissue samples is not feasible, we introduced the corresponding mutations into rabbit ␣ 1 subunits, which shares 94% sequence identity with the human ␣ 1A , and analyzed the biophysical properties of the mutant channels after heterologous expression in Xenopus laevis oocytes.

EXPERIMENTAL PROCEDURES
Mutant ␣ 1A cDNAs-Nucleotide numbering of restriction sites is given in parentheses. Mutants were constructed by applying the "gene-SOEing" technique (16) as described previously (17).
Single mutants T666M and V714A were constructed by using a XhoI (1689)-HindIII (2503) cassette in the subclone after elimination of the 5Ј-polylinker HindIII restriction site. The single mutations were subsequently introduced into pSPCBI-2 by exchanging a XhoI (1689)-NheI (3543) fragment in pSPCBI-2 for the respective mutant sequence. Mutation I1819L (corresponding to the human FHM mutation I1811L) was constructed by exchanging the KpnI-BglII cassette of construct AL22 (19) for the respective mutant BI-II sequence. All polymerase chain reaction-generated fragments were sequenced completely to confirm sequence integrity.
Expression of ␣ 1A Mutants in X. laevis Oocytes-Preparation of stage V-VI oocytes from X. laevis and injection of cRNA are described in detail elsewhere (17). Capped run-off poly(A ϩ ) cRNA transcripts from XbaIlinearized cDNA templates were synthesized according to the procedures of Krieg and Melton (20). ␣ 1 cRNAs were coinjected with ␤ 1a (21) and ␣ 2 ␦ (22) subunit cRNAs. To exclude effects of endogenous Ca 2ϩ -activated Cl Ϫ * This work was supported by Fonds zur Förderung der Wissenschaftlichen Forschung Grants P12641-MED (to J. S.), P12689 (to H. G.), and P12649-MED (to S. H.), the Ö sterreichische Nationalbank (to J. S.), and the Austrian Academy of Sciences (to R. L. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
currents on current kinetics experiments were also 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 (17). Similar current amplitudes were obtained with mutant and wild type ␣ 1A subunits. Oocytes expressing peak I Ba smaller than 400 nA or larger than 1.6 A were excluded from analysis. Data analysis and acquisition was 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 , 40 mM N-methyl-D-glucamine, 10 mM HEPES, 10 mM glucose, 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 during 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 steady state inactivation (k inact ), and the slope factor of the curve (V 0.5,inact ) were obtained by fitting the data to the Boltzmann equation.
Data Analysis-Nonlinear least square fitting and statistical calculations were performed using Origin R (Microcal). Data are given as means Ϯ S.E. for the indicated number of experiments.

RESULTS
To study the functional consequences of single amino acid mutations associated with human FHM we introduced the corresponding mutations into the rabbit ␣ 1A subunit (BI-II, Ref. 18). Their positions are illustrated in Fig. 1A. Human mutation I1811L corresponds to I1819L in rabbit ␣ 1A . Wild type and mutant ␣ 1A subunits were functionally expressed in X. laevis oocytes (together with accessory ␤ 1a and ␣ 2 ␦ subunits) and macroscopic channel properties measured using the two microelectrode voltage-clamp technique.
The half-maximal voltage for activation (V 0.5act ) was slightly, but significantly, shifted toward more negative potentials for mutants T666M, V714A, and I1819L (Table I). The midpoint voltage for steady-state inactivation was not significantly affected (Table I). The effects of mutations on I Ba decay during a 3-s pulse applied from a holding potential of Ϫ80 mV to ϩ10 mV is illustrated in Fig. 1, B and C. For wild type and mutant channels, current decay could be well described by a double exponential time course. The fast component of current decay was significantly (p Ͻ 0.01) faster for mutants T666M and V714A but not for I1819L and R192Q (Fig. 1C). No significant changes were found for the slow component (see legend to Fig. 1C).
The mutational effects on current inactivation could affect the accumulation of channels in inactivation during frequent depolarizations at high firing rates in neurons. To test this possibility we applied trains of pulses (1 Hz) from a holding potential of Ϫ60 mV to a test potential of ϩ10 mV. As illustrated in Fig. 2A, I Ba decreased by 18 Ϯ 1% (n ϭ 14) during a train of 15 pulses in wild type channels. In mutants T666M, V714A, I1819L, but not in R192Q, the amount of accumulation in an inactivated state was significantly (p Ͻ 0.01) different from wild type. Peak I Ba decrease during the pulse train was about 2-fold larger in T666M (34 Ϯ 2%, n ϭ 15) and about 2-fold smaller in I1819L (7.2 Ϯ 0.8%, n ϭ 6) than in wild type. Less accumulation in inactivation was also found for mutant V714A (13.2 Ϯ 0.5%, n ϭ 14). At the more negative holding potential of Ϫ80 mV I Ba current decay of mutants T666M (12.7 Ϯ 1%, n ϭ 15) and I1819L (4.5 Ϯ 0.6%, n ϭ 6) was also significantly different from wild type (7.5 Ϯ 0.4%, n ϭ 14).
The accumulation of channels in inactivation during a pulse train depends on how fast inactivation is removed between pulses. We therefore investigated the effects of the mutations on the time course of recovery by employing a double pulse protocol (Fig. 3). Wild type and mutant channels were inacti- vated by a 3-s conditioning prepulse from Ϫ80 mV to ϩ10 mV (Fig. 1B). The time course of recovery from inactivation at Ϫ60 mV was determined by subsequent test pulses applied at various times after the prepulse as described under "Experimental Procedures." In wild type and mutant channels about 90% of I Ba recovered within 20 s. The time course followed a biexponential function suggesting that recovery occurs from more than one inactivated state. In wild type the fast component ( fast ϭ 0.432 Ϯ 0.033 s, n ϭ 7) accounted for 68 Ϯ 2% of recovered I Ba and was about 8-fold faster than the slow component ( slow ϭ 3.1 Ϯ 0.34 s). Mutations T666M, V714A, and I1819L profoundly altered the time course of channel recovery from inactivation (Fig. 3, A and B). Mutant T666M slowed I Ba recovery mainly by significantly increasing fast and, to a smaller extent, also slow (Fig. 3B). In contrast to T666M, mu-tants V714A and I1819L recovered much faster than the wild type channel. These mutations significantly increased the contribution of the fast recovering component, decreased fast 3-4fold (Fig. 3B) and slightly reduced slow . The slower recovery from inactivation of T666M explains the enhanced accumulation of these channels in inactivation during pulse trains, whereas accelerated recovery of I1819L decreases such accumulation (Fig. 2). Mutation V714A also accelerated recovery from inactivation but yet exhibited a larger I Ba decrease during the train than I1819L (Fig. 2B). This can be explained by the finding that mutation V714A, but not I1819L, accelerated inactivation during a single test pulse (Fig. 1B), which can also favor accumulation in inactivation during pulse trains.
Taken together our data clearly demonstrate that FHM mutations T666M, V714A, and I1819L affect ␣ 1A Ca 2ϩ channel inactivation kinetics. This can cause significant changes in channel availability at higher depolarization frequencies. DISCUSSION Our data provide convincing evidence that three of the four mutations reported in FHM patients (9) affect the kinetic properties and the voltage dependence of ␣ 1A Ca 2ϩ channel activation. These mutations especially changed channel recovery from inactivation and thereby altered the extent to which mutant channels accumulate in an inactivated state during rapid depolarizations. Our experiments show that these mutations yield at least two functional phenotypes, leading to either an increase or a decrease in Ca 2ϩ channel availability. Mutants V714A and I1819L are located at almost identical positions at the intracellular end of the homologous helices S6 in repeats II (V714) and IV (I1819). Both accelerated recovery from inactivation and slightly shifted the activation curve toward more negative potentials. Therefore these mutations should increase Ca 2ϩ channel availability and promote voltage-dependent Ca 2ϩ influx into neurons. Mutation T666M is located in the linker IIS5-S6. It also slightly shifted the voltage dependence of activation toward more negative potentials but simultaneously slowed channel recovery from inactivation. The latter effect can decrease channel availability at high stimulation frequency (Fig. 2). Our data are in accordance with previous findings describing changes in the inactivation properties by site-directed mutations in the S6 segments of voltage-gated Ca 2ϩ (19,(23)(24)(25), potassium (26,27), and sodium channels (28,29). According to present folding models of voltage-gated cation channels (30), the S5-S6 linkers (containing mutation T666M) and S6 segments (containing V714A and I1819L) participate in the formation of the ion pore. Our data obtained with the FHM mutations therefore further support the hypothesis (24) that pore-forming residues play an important role for Ca 2ϩ channel inactivation.
Mutation R192Q eliminates a conserved positive charge within the amphipathic helix IS4, one of the putative voltage sensors (31). The finding that its mutation does not cause detectable functional changes under our experimental conditions is surprising. A recent analysis of charge neutralizing S4 mutations in a L-type Ca 2ϩ channel ␣ 1 subunit (32) has demonstrated that a large number but not all conservative positive charges in S4 contribute to channel gating. Charge neutralizations with no effect on V 0.5 were mainly seen in IIS4 and IVS4 but also included a residue in IS4. Our data also do not rule out possible effects of R192Q on the time course of current activation, which was not analyzed in our study.
Our data obtained with FHM mutations T666M, V714A, and I1819L agree with the hypothesis (9, 12) that mutations in the ␣ 1A subunit underly the neuronal instability which renders patients susceptible to migraine attacks that can be triggered by neural stimuli, such as stress or sensory afferentiation (12, TABLE I Effects of mutations on activation and inactivation properties V 0.5, act , half-maximal voltage for activation; k act , slope factor of the curve at V 0.5, act ; V 0.5, inact , half-maximal voltage for steady state inactivation; k inact , slope factor of the curve at V 0.5, inact .  (15), only a fraction of the channels should be affected by the mutations. With the exception of cerebellar degeneration, FHM (and other migraine) patients show no other major disturbances in neurological function, suggesting that the functional consequences of the mutations only become physiologically relevant under certain conditions. This assumption agrees with our observation that the functional consequences of altered inactivation properties become especially obvious only at higher stimulation frequency. Presynaptic V714A and I1819L channels can be considered "gain-of-function" mutants under these conditions. Their relative contribution to Ca 2ϩ entry would gradually increase with firing rate because they are expected to accumulate to a smaller extent in inactivation than wild type ␣ 1A . This could result in higher than normal Ca 2ϩ entry with even more pronounced effects on neurotransmitter release, which can rise with the fourth power of intracellular Ca 2ϩ concentrations. Such an enhanced Ca 2ϩ entry could eventually also lead to episodes of neuronal Ca 2ϩ overloading and explain the cerebellar neurodegeneration observed in some FHM patients, including patients with the I1811L (rabbit I1819L) mutation (9). The situation is even more complex because mutant P/Q-type Ca 2ϩ channels must also be localized on cell bodies and dendrites (3). There alterations in Ca 2ϩ entry at high firing rates could affect neuronal Ca 2ϩ -dependent processes, such as Ca 2ϩ -dependent phosphorylation/dephosphorylation and gene transcription. Decreased Ca 2ϩ entry through Ca 2ϩ channels (such as expected for mutant T666M) into cell bodies can also increase the firing rate of a neuron. Inhibition of Ca 2ϩ current can diminish neuronal spike after hyperpolarizations (e.g. by decreased activation of Ca 2ϩ -activated K ϩ -channels, 33) as has recently been shown, e.g. for P/Q-type channels in caudal raphe neurons (34). Among others, these neurons seem to play a crucial role in the pathophysiology of migraine (12).
Our data prompt further experiments to assess the patho-physiological consequences of the FHM mutations under conditions that more closely resemble neuronal activity in vivo. It will be especially important to study the effects of the mutations at 37°C. At this temperature faster channel kinetics would allow higher stimulation rates than in our experiments in X. laevis oocytes. Higher stimulation frequency may lead to considerable accumulation in inactivation even during trains of much shorter pulses than used in our study. Although our data clearly demonstrate that three of the four FHM mutations lead to significant alterations in ␣ 1A subunit function, several questions remain to be answered. The rabbit and human ␣ 1A share high sequence identity (93.5%) with sequence heterology limited to the C-terminal tail and the long cytoplasmic loops (9,18). Only a total of 4 amino acid differences exists in the putative pore forming region (consisting of S5, S5-S6 linkers, S6 of the four repeats). Despite this high conservation quantitative or qualitative differences of mutational effects on rabbit and human ␣ 1A cannot be excluded.
The three mutations affecting channel gating are located in conserved regions participating in the formation of the channel pore. They are not located within other known functional domains of the channel, such as ␣ 1 subunit interaction domains for accessory subunits (35) or G-proteins (36,37). This suggests that their effects are not indirectly caused by interfering with subunit or G-protein interactions. However, we cannot rule out the possibility that the mutational effects are affected by other factors such as the ␤ subunit isoform (␤ 1 -␤ 4 ) associated with the mutant ␣ 1A (8,38) or the level of G-protein activation. The biophysical characteristics of the mutants may also be affected by the permeating ion.
Our data demonstrate that residues in putative pore-forming regions of Ca 2ϩ channel ␣ 1A subunits determine inactivation properties. Further experiments are required to prove that FHM mutations alter Ca 2ϩ entry and neurotransmitter release preferentially at high firing rates in intact neurons.