Insights into migraine mechanisms and Ca_V2.1 calcium channel function from mouse models of familial hemiplegic migraine

Abstract

Migraine is a very common disabling brain disorder with unclear pathogenesis. A subtype of migraine with aura (familial hemiplegic migraine type 1: FHM1) is caused by mutations in Ca_V2.1 (P/Q-type) Ca²⁺ channels. This review describes the functional consequences of FHM1 mutations in knockin mouse models carrying the mild R192Q or severe S218L mutations in the orthologous gene. The FHM1 knockin mice show allele dosage-dependent gain-of-function of neuronal P/Q-type Ca²⁺ current, reflecting activation of mutant channels at lower voltages, and allele dosage- and sex-dependent facilitation of induction and propagation of cortical spreading depression (CSD), the phenomenon that underlies migraine aura. Gain-of-function of neuronal Ca²⁺ current, facilitation of CSD and post-CSD motor deficits were larger in S218L than R192Q knockin mice, in correlation with the more severe human S218L phenotype. Enhanced cortical excitatory neurotransmission, due to increased action potential-evoked Ca²⁺ influx and increased probability of glutamate release at pyramidal cell synapses, but unaltered inhibitory neurotransmission at fast-spiking interneuron synapses, were demonstrated in R192Q knockin mice. Evidence for a causative link between enhanced glutamate release and CSD facilitation was obtained. The data from FHM1 mice strengthen the view of CSD as a key player in the pathogenesis of migraine, give insight into CSD mechanisms and point to episodic disruption of excitation–inhibition balance and neuronal hyperactivity as the basis for vulnerability to CSD ignition in migraine.

Daniela Pietrobon graduated in Chemistry at the University of Padova in Italy. After postgraduate work on energy transduction in mitochondria in the same University, she became CNR researcher and spent 2 years as visiting scientist at the Weizmann Institute of Science in Israel (working with R. Caplan on kinetic modelling of mitochondrial proton pumps) and 3 years at Harvard Medical School in Boston (working with P. Hess on the biophysics of voltage-gated Ca²⁺ channels). Back in Italy, she became Associate and then Full Professor of Physiology at the University of Padova, and focused her research on neuronal voltage-gated Ca²⁺ channels and Ca_V2.1 channelopathies, in particular familial hemiplegic migraine.

Introduction

Migraine is a common disabling brain disorder that is characterized by recurrent attacks of unilateral headaches, often accompanied by nausea, phonophobia and photophobia; the headache may be preceded by transient neurological symptoms that are most frequently visual but may involve other senses (migraine with aura). The remarkably common occurrence of migraine (cumulative lifetime incidence of 43% in women and 18% in men in a recent population study; Stewart et al. 2008) suggests that it may involve relatively minor perturbations of normal brain function and may, therefore, have much to teach us about the basic physiology of the nervous system (Pietrobon & Striessnig, 2003; Charles, 2009).

The initiation of migraine pain requires the activation of meningeal nociceptive trigeminal sensory afferents; activation of meningeal nociceptors leads to release of vasoactive neuropeptides at their peripheral nerve terminals and to activation of second-order neurons in the trigeminal nucleus caudalis, followed by activation of brain structures involved in the processing and perception of pain. The maintenance of the severe prolonged pain of migraine headache involves sensitization of meningeal nociceptors and self-sustained sensitization of central neurons of the trigeminovascular system, whose incompletely understood mechanisms may include alterations of descending endogenous pain modulatory pathways. It is now generally accepted that the primary cause of migraine lies in the brain, but the nature and mechanisms of the primary brain dysfunction that leads to activation of the meningeal trigeminal nociceptors remain incompletely understood and controversial (Pietrobon & Striessnig, 2003; Charles, 2009; Goadsby et al. 2009; Levy et al. 2009). Recent findings point to cortical spreading depression (CSD) as a key player in the pathogenesis of migraine. CSD can be induced in animals by focal stimulation of the cerebral cortex and consists of a slowly propagating wave of strong neuronal and glial depolarization, whose initiation and propagation mechanisms remain unclear (Leao, 1944; Somjen, 2001; Charles & Brennan, 2009). Several lines of evidence, in particular neuroimaging findings, indicate that migraine aura is due to CSD-like events in cerebral sensory cortices (Bowyer et al. 2001; Hadjikhani et al. 2001; Ayata, 2009). Animal studies support the idea that CSD may also initiate the headache mechanisms (Bolay et al. 2002; Ayata et al. 2006; Ayata, 2009), but the connection between CSD and headache in patients (particularly those with migraine without aura) remains an open question. The mechanisms that make the brain of migraineurs susceptible to episodic ‘spontaneous’ CSDs in response to specific triggers are unknown. Migraineurs are hypersensitive to any kind of sensory overload and there is strong evidence for altered cortical excitability and abnormal processing of sensory information in their brain in the period between migraine attacks (Welch, 2005; Aurora & Wilkinson, 2007; Coppola et al. 2007). The mechanisms underlying the interictal abnormalities in cortical activity are controversial and their relationship to susceptibility and/or occurrence of CSD is unclear.

Migraine has a strong genetic component, but causative genes have not been identified, except for familial hemiplegic migraine (FHM), a rare monogenic subtype of migraine with aura of childhood onset. Apart from the motor weakness or hemiparesis during aura, typical FHM attacks resemble migraine with aura attacks and both types of attacks may alternate in patients and co-occur within families, suggesting that FHM and migraine with aura may be part of the same spectrum and may share some pathogenetic mechanisms (Thomsen et al. 2002, 2003). FHM is genetically heterogeneous: missense mutations in CACNA1A and SCN1A, the genes encoding the pore-forming subunits of the neuronal voltage-gated Ca²⁺ channel Ca_V2.1 (or P/Q-type) and Na⁺ channel Na_V1.1, cause FHM type 1 (FHM1; Ophoff et al. 1996) and type 3 (FHM3, Dichgans et al. 2005), respectively; mutations in ATP1A2, the gene encoding the Na⁺–K⁺-ATPase α₂ subunit, cause FHM type 2 (FHM2; De Fusco et al. 2003). The different FHM mutations and their functional consequences on recombinant human Ca_V2.1 channels in heterologous expression systems (and, for some FHM1 mutations, in transfected neurons) have been recently reviewed (Pietrobon, 2007) and will not be discussed in detail here.

Two different FHM1 knockin (KI) mouse models have been generated by introducing the human R192Q and S218L mutations into the orthologous cacna1a Ca_V2.1 channel gene (van den Maagdenberg et al. 2004, 2010). While mutation R192Q in humans causes pure FHM, mutation S218L causes a particularly dramatic clinical syndrome, that may consist of, in addition to attacks of hemiplegic migraine, slowly progressive cerebellar ataxia and atrophy, epileptic seizures, coma or profound stupor and severe, sometimes fatal, cerebral oedema which can be triggered by only a trivial head trauma (Kors et al. 2001; van den Maagdenberg et al. 2010). While homozygous R192Q (RQ/RQ) and heterozygous S218L (SL/WT) KI mice did not exhibit an overt phenotype, homozygous S218L (SL/SL) mice exhibited mild permanent cerebellar ataxia, spontaneous attacks of hemiparesis and/or (sometimes fatal) generalized seizures, and brain oedema after only a mild head impact, thus modelling the main features of the severe S218L clinical syndrome (van den Maagdenberg et al. 2004, 2010).

Ca_V2.1 channels

Ca_V2.1 channels are located in presynaptic terminals and somatodendritic membranes throughout the mammalian brain (Westenbroek et al. 1995) and play a prominent role in initiating action potential-evoked neurotransmitter release at central nervous system synapses (Pietrobon, 2005a). At many central synapses P/Q-, N- and R-type Ca²⁺ channels cooperate in controlling neurotransmitter release, but P/Q-type channels have a dominant role, partly because of a more efficient coupling to the exocytotic machinery (Mintz et al. 1995; Wu et al. 1999; Qian & Noebels, 2001; Li et al. 2007); the relative contribution of P/Q-type channels increases with postnatal age at many synapses (Iwasaki et al. 2000). The somatodendritic localization of Ca_V2.1 channels points to additional postsynaptic roles, e.g. in neural excitability (Pineda et al. 1998; Womack et al. 2004).

Ca_V2.1 channels are expressed in all brain structures that have been implicated in the pathogenesis of migraine and/or migraine pain, including the cerebral cortex, the trigeminal ganglia, and brainstem nuclei involved in the central control of nociception; their expression is particularly high in the cerebellum, which explains the cerebellar symptoms produced by some FHM1 mutations (Pietrobon & Striessnig, 2003, for review and references).

In the cerebral cortex, excitatory synaptic transmission at pyramidal cell synapses in different cortical areas depends predominantly on P/Q-type Ca²⁺ channels (Iwasaki et al. 2000; Koester & Sakmann, 2000; Rozov et al. 2001; Ali & Nelson, 2006; Zaitsev et al. 2007; Tottene et al. 2009) with a notable exception at synapses between layer 5 pyramidal cells and burst-firing bipolar interneurons of motor cortex (Ali & Nelson, 2006). Relatively little is known about the Ca²⁺ channels initiating inhibitory neurotransmission in the cortex, since the Ca²⁺ channel pharmacology has been investigated only at the synapses between fast-spiking (FS) interneurons and pyramidal cells; neurotransmission was found to be exclusively dependent on P/Q-type channels in many cortical areas (Zaitsev et al. 2007; Kruglikov & Rudy, 2008; Tottene et al. 2009), but again with the exception of layer 5 of the motor cortex, where it was exclusively dependent on N-type channels (Ali & Nelson, 2006). P/Q-type channels also contribute to the regulation of the intrinsic firing of layer 2/3 cortical pyramidal neurons, via activation of different Ca²⁺-dependent K⁺ channels (Pineda et al. 1998), and are essential in thalamocortical neurons for the generation of high-frequency subthreshold gamma-band oscillations, that are characteristic of the aroused attentive state and considered a functional prerequisite to cognitive states (Llinás et al. 2007).

In the trigeminovascular system, P/Q channels are involved in the control of release of vasoactive neuropeptides from perivascular terminals of meningeal nociceptors (Hong et al. 1999; Asakura et al. 2000; Akerman et al. 2003), and of glutamate release from trigeminal ganglion neurons in culture (Xiao et al. 2008). P/Q channels are also involved in controlling tonic inhibition of trigeminal nucleus caudalis neurons with input from the dura (Ebersberger et al. 2004), and in descending inhibitory and facilitatory pathways that regulate trigeminal and spinal pain transmission (Knight et al. 2002; Urban et al. 2005). A possible important role of P/Q channels in central sensitization of the trigeminovascular system is suggested by pharmacological evidence in the spinal cord (Vanegas & Schaible, 2000), and by the finding of reduced responses to inflammatory and neuropathic pain in Ca_V2.1^−/+ mice, carrying a genetic ablation of the Ca_V2.1α₁ subunit that reduces to half the Ca_V2.1 channels (Luvisetto et al. 2006).

Knockin mouse models of familial hemiplegic migraine

The FHM1 KI mouse models allowed the first analysis of the functional consequences of FHM1 mutations on Ca_V2.1 channels and synaptic transmission in neurons expressing the channels at the endogenous physiological level (van den Maagdenberg et al. 2004, 2010; Tottene et al. 2009). In cerebellar granule cells and cortical pyramidal cells of R192Q and S218L KI mice the P/Q-type Ca²⁺ current density was larger than in wild-type (WT) neurons in a wide range of relatively mild depolarizations, reflecting activation of mutant Ca_V2.1 channels at more negative voltages than the corresponding WT channels; P/Q current densities were similar in KI and WT neurons at higher voltages (that elicit maximal Ca_V2.1 channel open probability), indicating similar densities of functional Ca_V2.1 channels (van den Maagdenberg et al. 2004, 2010; Tottene et al. 2009). Thus, in two different FHM1 mouse models and two different types of neurons the functional consequences of the FHM1 mutations on native neuronal mouse Ca_V2.1 channels were quite similar to those on single recombinant human Ca_V2.1 channels (Tottene et al. 2002, 2005). In fact, the analysis of the single-channel properties of human Ca_V2.1 channels carrying eight different FHM1 mutations revealed a consistent increase in channel open probability and single-channel Ca²⁺ influx (as measured by the product of single-channel current and open probability) in a wide range of depolarizations, mainly due to a shift to lower voltages of channel activation (Tottene et al. 2002, 2005; Catterall et al. 2008; author's unpublished observations). In agreement with the lower threshold of activation of human S218L Ca_V2.1 channels compared to that of human R192Q Ca_V2.1 channels (Tottene et al. 2005), the gain-of-function of the P/Q current at low voltages was larger in SL/SL than in RQ/RQ KI mice (van den Maagdenberg et al. 2004, 2010). The shift to lower voltages of Ca_V2.1 channel activation and the gain-of-function of the neuronal Ca_V2.1 current were about twice as large in homozygous compared to heterozygous KI mice, revealing an allele–dosage effect consistent with dominance of the mutation in FHM1 patients (van den Maagdenberg et al. 2010). The densities of the other Ca²⁺ current components (L, N and R) were similar in KI and WT neurons, indicating the absence of compensatory changes in the other Ca²⁺ channel types (van den Maagdenberg et al. 2004, 2010; Tottene et al. 2009).

The first indication that the gain-of-function of Ca_V2.1 channels produced by FHM1 mutations could lead to enhanced evoked neurotransmitter release was obtained by investigating neurotransmission at the neuromuscular junction; in KI mice, evoked transmission was unaltered at physiological Ca²⁺ ion concentrations but increased at 0.2 mm Ca²⁺ (van den Maagdenberg et al. 2004, 2010; Kaja et al. 2005). Recently, cortical excitatory neurotransmission was investigated in neuronal microcultures and in brain slices from R192Q KI mice (Tottene et al. 2009). The results show increased synaptic strength at physiological Ca²⁺ due to enhanced action-potential-evoked Ca²⁺ influx through mutant synaptic P/Q Ca²⁺ channels and enhanced probability of glutamate release at cortical pyramidal cell synapses. Short-term synaptic depression during trains of action potentials was enhanced. Neither amplitude nor frequency of miniature excitatory postsynaptic current was altered, indicating the absence of homeostatic compensatory mechanisms at excitatory synapses onto pyramidal cells.

Paired recordings of layer 2/3 pyramidal cells and fast spiking (FS) inhibitory interneurons in acute cortical slices revealed that, in striking contrast with the enhanced glutamatergic transmission, the inhibitory GABAergic transmission at FS synapses was not altered in FHM1 mice, despite being initiated by P/Q Ca²⁺ channels (Tottene et al. 2009). Given the evidence that the magnitude (or even the presence) of the negative shift in activation of human Ca_V2.1 channels produced by FHM1 mutations may depend on the particular recombinant Ca_V2.1α₁ splice variant and/or Ca_V2.1β subunit (Mullner et al. 2004; Adams et al. 2009), a possible explanation for the unaltered inhibitory transmission may be the presence at FS interneuron synapses of a Ca_V2.1 isoform that is little affected by the mutation. Alternative explanations may be near saturation of the presynaptic Ca²⁺ sensor and/or a minor effect of FHM1 mutations on action potential-evoked Ca²⁺ influx due to the very short duration of the action potential of FS interneurons (Cauli et al. 2000).

The demonstration that FHM1 mutations may affect synaptic transmission and short-term synaptic plasticity differently at different cortical synapses suggests that the neuronal circuits that coordinate and dynamically adjust the balance between excitation and inhibition during cortical activity are very likely altered in FHM1 (Tottene et al. 2009). Functional alterations in these circuits might underlie the abnormal processing of sensory information of migraineurs in the interictal period.

The investigation of experimental CSD, elicited either by electrical stimulation of the cortex in vivo or high KCl in cortical slices, revealed a lower threshold for CSD induction and an increased velocity of CSD propagation in R192Q and S218L KI mice (van den Maagdenberg et al. 2004, 2010; Tottene et al. 2009). Moreover, a single CSD, elicited by brief epidural application of high KCl, produced more severe and prolonged motor deficits (including hemiplegia) in FHM1 KI mice, and, in contrast with WT mice, CSD readily propagated into the striatum (Eikermann-Haerter et al. 2009b). In agreement with the higher incidence of migraine in females, the velocity of propagation and the frequency of CSDs, elicited by continuous epidural high KCl application, were larger in females than in males of both mutant strains; the sex difference was abrogated by ovariectomy and enhanced by orchiectomy, suggesting that female and male gonadal hormones exert reciprocal effects on CSD susceptibility (Eikermann-Haerter et al. 2009a,b;).

In correlation with the more severe S218L clinical phenotype and the larger gain-of-function of the neuronal P/Q Ca²⁺ current, the facilitation of both induction and propagation of CSD were larger (van den Maagdenberg et al. 2010) and the post-CSD neurological motor deficits were more severe (Eikermann-Haerter et al. 2009b) in S218L than R192Q mice in vivo. Moreover, S218L KI mice showed a unique increased susceptibility to repetitive successive CSD events following a single CSD-inducing stimulus: multiple CSDs after a threshold stimulation were observed in more than 70% of SL/SL and 50% of SL/WT mice, but only in 10–20% of WT and RQ/RQ KI mice (van den Maagdenberg et al. 2010). Although the exact mechanism of this unique increased propensity to recurrent CSDs remains to be determined, the findings of van den Maagdenberg et al. (2010) suggest that the high sensitivity of the S218L brain to even mild stimuli (e.g. low-impact head trauma) may be explained, at least in part, by the unique combination of a particularly low CSD trigger threshold and a high propensity for multiple CSD events.

As a whole, the studies of experimental CSD in FHM1 KI mice strengthen the view of CSD as a key player in the pathogenesis of migraine. Moreover, together with the previous study of Ayata et al. (2000) that revealed an increased threshold for initiation and a decreased velocity of propagation of experimental CSD in the spontaneous mouse mutants leaner and tottering with loss-of-function mutations in cacna1a (Pietrobon, 2002), the studies of CSD in FHM1 KI mice support a key role of Ca_V2.1 channels in the initiation and propagation of CSD.

Tottene et al. (2009) provided direct evidence that the gain-of-function of glutamate release at synapses onto pyramidal cells may explain the facilitation of experimental CSD in FHM1 mutant mice. In fact, the facilitation of induction and propagation of experimental CSD observed in acute slices of somatosensory cortex of R192Q KI mice was completely eliminated (both CSD threshold and velocity became similar to those in WT slices) when glutamate release at pyramidal cell synapses was brought back to WT values using a subsaturating concentration of the P/Q channel blocker ω-agatoxin IVA.

This finding gives insight into the controversial mechanisms of CSD initiation and propagation. The initiation of the positive feedback cycle that ignites CSD and almost zeroes the neuronal membrane potential depends on the local increase of the extracellular concentration of K⁺ ions ([K⁺]_o) above a critical value and on the activation of a net inward current at the apical pyramidal cell dendrites (Somjen, 2001); the nature of the cationic channels mediating this inward current remains unclear and controversial, although there is strong pharmacological support for a key role of NMDA receptors (e.g. Marrannes et al. 1988; Footitt & Newberry, 1998); another controversial issue is the mechanism of CSD propagation (Somjen, 2001; Charles & Brennan, 2009). The findings of Tottene et al. (2009) support a model of CSD initiation in which activation of presynaptic P/Q-type Ca²⁺ channels with consequent release of glutamate from recurrent cortical pyramidal cell synapses and activation of NMDA receptors (and possibly postsynaptic P/Q-type Ca²⁺ channels) are key components of the positive feedback cycle that ignites CSD (Pietrobon, 2005b). Regarding CSD propagation, the findings are consistent with a model based on interstitial K⁺ diffusion initiating this positive feedback cycle in adjacent dendrites.

In migraineurs CSD is not induced by experimental depolarizing stimuli, but arises ‘spontaneously’ in response to specific triggers that somehow create in the cortex the conditions for initiation of the positive feedback cycle that overwhelms the regulatory mechanisms controlling cortical [K⁺]_o and ignites CSD. Insights into how this might occur have been provided by the interesting finding that, in contrast with the gain-of-function of excitatory neurotransmission, inhibitory neurotransmission at synapses between fast-spiking interneurons and pyramidal cells was not altered in FHM1 KI mice (Tottene et al. 2009). A plausible working hypothesis is that the differential effect of FHM1 mutations on excitatory and inhibitory neurotransmission may, in certain conditions (cf. migraine triggers), lead to disruption of the cortical excitation–inhibition balance due to excessive recurrent excitation, resulting in overexcitation and neuronal hyperactivity, that may increase [K⁺]_o above the critical value for CSD ignition (Fig. 1). Within this hypothesis, the findings of reduced G protein-mediated inhibition of recombinant Ca_V2.1 channels carrying FHM1 mutations (Melliti et al. 2003; Weiss et al. 2008; Serra et al. 2009), if confirmed for action potential-evoked Ca²⁺ influx through native presynaptic Ca_V2.1 channels at glutamatergic synapses, may have important implications for understanding the occurrence of episodic ‘spontaneous’ CSDs in migraine.

Figure 1.

Proposed pathophysiological mechanism of FHM1 and migraine: episodic disruptions of the cortical excitation–inhibition balance and hyperactivity of cortical circuits in response to specific migraine triggers are supposed to underlie vulnerability to CSD ignition. In FHM1 (and probably, FHM2, FHM3 and a fraction of migraine cases) excitation–inhibition inbalance results from enhanced glutamatergic neurotransmission and enhanced recurrent cortical excitation. In migraine, different mechanisms, that remain to be elucidated, may result in disruption of the excitation–inhibition balance in the cortex (and/or subcortical areas that have been hypothesized to activate the trigeminovascular system).

The analysis of the functional consequences of FHM1 mutations in KI mouse models supports the view of migraine as an episodic disorder of brain excitability, with episodic disruptions of the excitation–inhibition balance and hyperactivity of cortical circuits in response to specific migraine triggers as the basis for vulnerability to CSD ignition (Fig. 1). Given the remarkable clinical heterogeneity of migraine, it is possible that episodic disruptions of the excitation–inhibition balance in response to migraine triggers may occur independently or in parallel in multiple brain regions and converge in activating the trigeminovascular system (Charles, 2009; Levy et al. 2009).