Migraine Mutations Increase Stroke Vulnerability by Facilitating Ischemic Depolarizations

Abstract

Background

Migraine is an independent risk factor for stroke. Mechanisms underlying this association are unclear. Familial hemiplegic migraine (FHM), a migraine subtype that also carries an increased stroke risk, is a useful model for common migraine phenotypes because of shared aura and headache features, trigger factors, and underlying glutamatergic mechanisms.

Methods and Results

Here, we show that FHM type 1 (FHM1) mutations in Ca_V2.1 voltage-gated Ca²⁺ channels render the brain more vulnerable to ischemic stroke. Compared to wild-type, two FHM1 mutant mouse strains developed earlier onset of anoxic depolarization and more frequent peri-infarct depolarizations, associated with rapid expansion of infarct core on diffusion-weighted MRI and larger perfusion deficits on laser speckle flowmetry. Cerebral blood flow required for tissue survival was higher in the mutants, leading to infarction with milder ischemia. As a result, mutants developed larger infarcts and worse neurological outcomes after stroke, which were selectively attenuated by a glutamate receptor antagonist.

Conclusions

We propose that enhanced susceptibility to ischemic depolarizations akin to spreading depression predisposes migraineurs to infarction during mild ischemic events, thereby increasing the stroke risk.

Migraine is the most common neurological condition affecting young to middle-age adults. Up to a third of migraineurs experience transient neurological symptoms called aura. Migraine, particularly with aura, is associated with increased stroke risk both during and between attacks, especially in women^1–4. The biological basis for this association is unknown. Stroke risk is also increased in familial hemiplegic migraine (FHM), a monogenic migraine subtype with hemiplegic auras in addition to the common aura forms⁵. FHM is a useful model for common migraine with aura because of shared clinical features and trigger factors, and because two-thirds of FHM patients and their first degree relatives also have attacks of common migraine with or without aura^{6, 7}. Neuronal network hyperexcitability and enhanced glutamate release have been implicated in both FHM and common forms of migraine⁸.

FHM type 1 (FHM1) is caused by mutations in the CACNA1A gene, which encodes the pore-forming α_1A subunit of neuronal Ca_V2.1 voltage-gated Ca²⁺ channels⁹. Presynaptic Ca_V2.1 channels are major regulators of excitatory neurotransmitter release. FHM1 mutant channels open with smaller depolarizations and stay open longer¹⁰, which augments presynaptic Ca²⁺ entry and glutamate release thereby enhancing brain excitability¹¹. Transgenic mice expressing the human R192Q or S218L FHM1 mutation show an increased susceptibility to spreading depression, the electrophysiological substrate for migraine, and display characteristic clinical features of FHM such as transient hemiplegia^12–14. Glutamatergic mechanisms and hyperexcitability have been implicated in the pathogenesis of common forms of migraine as well^{15, 16}. Genetic support for this link was recently obtained in a genome-wide association study identifying the astrocyte elevated gene 1 (AEG1), encoding a regulator of glial glutamate transporter EAAT2, as the first migraine gene¹⁶.

Glutamate excitotoxicity also plays a pivotal role in the pathogenesis of stroke. We, therefore, hypothesized that genetic mutations conferring cerebral hyperexcitability and migraine susceptibility increase the vulnerability to ischemic stroke as one mechanism to explain the migraine-stroke association, and tested this using Ca_V2.1 S218L and R192Q transgenic mouse models of FHM1. The results reveal electrophysiological and hemodynamic mechanisms that accelerate hyperacute stroke evolution and worsen ischemic outcome in FHM1 mutants, and suggest a pivotal role for enhanced glutamatergic transmission in increasing the vulnerability to ischemic stroke in susceptible migraineurs.

Methods

Experimental animals

Experimental procedures were approved by the institutional review board. A total of 267 male and female mice were used. Transgenic knockin Cacna1a migraine mouse models homozygous (HOM) or heterozygous (HET) for R192Q or S218L FHM1 mutations were generated by a gene targeting approach¹⁴. The R192Q mutant strain was compared to C57BL6/J, backcrossed for 10 generations. The S218L mutants were compared to their wild type (WT) littermates. Because stroke risk is highest in young adult migraineurs, mice were studied between 2–6 months of age. All experiments were carried out with the investigators blinded, and confirmatory genotyping was done.

Systemic physiological monitoring

Arterial pH, pO₂, pCO₂, and blood pressure were measured via a femoral artery catheter under isoflurane anesthesia (2.5% induction, 1.5% maintenance, in 70% N₂O and 30% O₂; Supplemental Table 1). Rectal temperature was controlled at 37°C during ischemia, and intermittent monitoring continued for 6 hours in a subset of mice.

Transient filament occlusion of the middle cerebral artery (fMCAO)

A nylon monofilament was inserted into internal via the external carotid artery followed by reperfusion after 30 or 60 minutes, under isoflurane anesthesia (2.5% induction, 1.5% maintenance, in 70% N₂O and 30% O₂) and laser Doppler monitoring. Mice were placed in a temperature-controlled incubator with easy access to food and water. Neurological outcomes were scored 24 hours after reperfusion, using a five-point scale: 0, normal; 1, forepaw monoparesis; 2, circling to left; 3, falling to left; 4, no spontaneous walking and depressed consciousness; 5, death. Infarct volume was calculated by integrating the infarct area in ten 1 mm-thick 2,3,5-triphenyltetrazolium chloride (TTC)-stained coronal sections. Infarct volumes were calculated by subtracting the volume of ipsilateral non-infarcted tissue from contralateral hemisphere. Ischemic swelling volumes were calculated by subtracting the volume of contralateral hemisphere from the volume of ipsilateral hemisphere. Glutamatergic mechanisms were tested by administering MK-801 (1mg/kg, intraperitoneal; Sigma, St Louis, MO) 15 minutes before fMCAO.

MRI

Apparent diffusion coefficient (ADC) maps were acquired under isoflurane anesthesia with a 9.4-Tesla MRI scanner (Bruker Biospin, Inc., Billerica, MA) 30 and 60 minutes after fMCAO (TR/TE=3000/27 ms, b=154 & 1294 s/mm², 180•180 μm² in-plan resolution, 1 mm slice thickness, NA=8). Mean and standard deviation of ADC in cortex, striatum, hippocampus and thalamus were extracted from the normal hemisphere, and thresholds defined as mean minus 2 standard deviations to calculate lesion volumes. Normal systemic physiological parameters were confirmed under simulated MRI conditions in a separate group of mice (not shown).

Electrophysiological recordings

After fMCAO, isoflurane-anesthetized mice were intubated, ventilated, and femoral artery catheterized for blood pressure and blood gas monitoring. Two intracortical glass micropipettes were placed and extracellular recordings (depth 250 μm) started within 15 minutes after the onset of ischemia and continued for approximately 2 hours.

Receptor autoradiography

The density and distribution of glutamate and GABA_A receptors and glutamate re-uptake sites were assessed on 10 μm frozen sections using tritium-sensitive storage phosphor screens (GE Healthcare) as described¹⁷.

Laser speckle-flowmetry of transient distal middle cerebral artery occlusion (dMCAO)

Spontaneously breathing mice (S218L HOM) were anesthetized with isoflurane as above, femoral artery catheterized for blood pressure and gas measurements. Mice were placed in a stereotaxic frame, a temporal burr hole (2 mm diameter) was drilled above the zygomatic arch, and distal middle cerebral artery was occluded using a microvascular clip for 60 minutes. Cortical perfusion was imaged during dMCAO using laser speckle flowmetry through intact skull¹⁸. Cerebral blood flow (CBF) changes were calculated for each pixel relative to pre-ischemic baseline, and the area of cortex with residual CBF ≤30% was determined by thresholding. Neurological outcomes and infarcts were assessed 48 hours later as described above. In addition, CBF threshold for tissue viability was estimated by superimposing the images of CBF and infarct. In the R192Q HOM strain, mice were intubated and ventilated to ensure normal arterial blood gas values, precluding survival for neurological and infarct assessment in this strain.

Anatomical analysis of the circle of Willis and pial collaterals

Mice were transcardially perfused with carbon black. Diameter of cerebral arteries, patency of the posterior communicating artery, and the number of pial arterial anastomoses between the anterior, posterior and middle cerebral arteries and their distance from midline were determined.

Absolute resting CBF

Mice were anesthetized with α-chloralose (50 mg/kg) and ventilated. Femoral artery and external jugular vein were cannulated. Arterial blood was withdrawn continuously (0.3 mL/min). One μCi of N-isopropyl-[methyl-1,3-¹⁴C]-p-iodoamphetamine was injected in 0.1 mL saline over 10 seconds. Twenty seconds after injection, animal was decapitated and the blood withdrawal terminated simultaneously. The brain was removed, frozen and dissected. CBF was calculated from the radioactivity in tissue and blood measured by liquid scintillation spectrometry.

Statistical Analysis

Data were analyzed using SPSS (v11.0), and presented as mean ± standard error, or median [interquartile range]. We used chi-square test to compare proportions, ANOVA to compare mean values of continuous measures according to genotypes, and general linear models for repeated measures to compare mean values over time according to genotypes. Comparisons of disability scores were performed by using non-parametric rank tests, cumulative peri-infarct depolarization (PID) incidence by log rank test, and PID frequency vs. the area of CBF deficit by Pearson correlation test. P values are two-tailed, and P<0.05 was considered statistically significant.

Results

Enhanced susceptibility to anoxic and peri-infarct depolarizations

Anoxic depolarization is characterized by a sudden loss of membrane ionic gradients, uncontrolled glutamate release and cell swelling, triggered by the failure of Na⁺/K⁺ ATPase under ischemic conditions. We found significantly earlier onset of anoxic depolarization in FHM1 mutants after fMCAO, by monitoring its vasoconstrictive effect on cerebral vasculature as previously described (Fig. 1a, b)^{18, 19}. Importantly, the magnitude of CBF reduction in the ischemic core did not differ among groups in this fMCAO model, eliminating the possibility that faster anoxic depolarization rates were due to more severe ischemia (residual CBF 10–17% of baseline in both S218L and R192Q; P=0.634 and 0.599, respectively, data not shown).

Figure 1.

Faster anoxic depolarization rates and more frequent peri-infarct depolarizations in FHM1 mutant mice. a) Representative laser Doppler tracings show CBF reduction upon common carotid artery occlusion (CCAO) followed by fMCAO. A further decline in CBF marks the onset of anoxic depolarization (AD), and is due to the vasoconstrictive effect of tissue depolarization on ischemic microvasculature¹⁸. Scale bars: vertical 10%, horizontal 1 min. b) The latency to AD was shorter in S218L and R192Q mutants (p<0.001 and p=0.035, respectively), with a trend for allele-dosage effect. *p<0.05 vs. WT. c) Representative electrophysiological tracings show more frequent PIDs in S218L HOM compared to WT. Scale bars: vertical 20 mV, horizontal 4 min. d) PIDs (round symbols shown as a function of time) occurred in S218L HOM more frequently and sometimes in clusters (rectangular boxes). Horizontal lines indicate the time of onset and end of electrophysiological recordings in each mouse (n=5 each). When calculating the average PID frequency, these minor differences in recording duration were taken into account. e) Pooled cumulative PID numbers as a function of time after fMCAO was more than doubled in S218L HOM (*p<0.001; n=5 each). f) Experimental setup showing two intracortical glass micropipettes (E1, E2) placed outside the ischemic territory to detect PIDs after fMCAO. Shaded area indicates typical distribution of CBF deficit after fMCAO.

PIDs are recurrent propagating depolarization waves akin to spreading depression that exacerbate the metabolic mismatch in penumbra and promote infarct growth during hyperacute stroke^18–21. We reasoned that FHM1 mutations that enhance spreading depression susceptibility¹³ might also facilitate the occurrence of PIDs. Using intracortical microelectrode recordings during fMCAO, we indeed found a two-fold increase in the frequency of PIDs in mutants over WT (5.3±1.3 vs. 2.6±0.3 PIDs/h, P=0.028; Fig. 1c–e). In the mutants, PIDs sometimes occurred in clusters, possibly reflecting circling around the ischemic core (see below)²², but otherwise did not differ from WT in terms of their durations and amplitudes (not shown). Altogether, these data suggest that genetically enhanced susceptibility to spreading depression^11–14 facilitates the occurrence of anoxic depolarization and PIDs during acute stroke as a novel mechanism to explain increased stroke vulnerability in migraineurs.

Rapid growth of hyperacute ischemic core on MRI

To assess whether enhanced susceptibility to anoxic and peri-infarct depolarizations accelerate the hyperacute stroke evolution in FHM1 mutants, we performed serial diffusion-weighted MRI during fMCAO. Reduced ADC values on diffusion-weighted MRI reflect anoxic depolarization, loss of transmembrane ionic gradients, and cell swelling (i.e., ischemic core). We found that the ADC lesion volumes expanded more rapidly in FHM1 mutant strains compared to WT controls (Fig. 2). Although larger ADC lesion volumes were primarily due to more severe cortical involvement, the hyperacute lesion also encompassed hippocampus and thalamus in S218L mutants.

Figure 2.

FHM1 mutant mice show accelerated lesion growth on MRI during hyperacute stroke. a) ADC lesion (i.e., ischemic core with restricted water diffusion, purple) was larger on diffusion-weighted MR images in S218L (left panel) and R192Q (right panel) mutants compared to WT during the hyperacute phase after fMCAO. b) ADC lesion volumes shown as a function of time after fMCAO were 40–50% larger in S218L and R192Q HOM compared to WT as early as 30 minutes after fMCAO suggesting faster growth of ischemic core (p=0.019 and p=0.018, respectively). The difference remained significant at 60 minutes. Vertical and horizontal error bars reflect the standard errors for total ADC lesion volume and timing of MRI scans, respectively. c) Thirty minutes after stroke onset, enlarged ADC lesion volumes in S218L HOM and R192Q HOM were primarily due to more severe cortical involvement (p=0.015 and p=0.023, respectively), although S218L HOM mutants showed hyperacute ADC changes in the hippocampus and thalamus as well (P=0.018 and P=0.117, respectively). Of note, the average regional ADC values in the center of ischemic core did not significantly differ between FHM1 mutants and their WT controls, suggesting that cytotoxic cell swelling is complete in ischemic core in all groups (60±7% vs. 56±6% in R192Q WT and HOM, and 57±5% vs. 57±6% of contralateral hemisphere in S218L WT and HOM, respectively, 60 minutes after stroke onset). *p<0.05 vs. WT.

Larger perfusion deficit during hyperacute stroke

Ischemic depolarizations compromise residual CBF within the territory supplied by the occluded artery via vasoconstrictive (i.e., inverse) neurovascular coupling^{18, 19}, as a major determinant of outcome in cerebral ischemia. Using laser speckle flowmetry, we found larger cortical perfusion deficits after dMCAO in FHM1 mutants (Fig. 3a, b; only S218L shown), associated with an increased frequency of PIDs (0.9±0.3 vs. 4.6±1.2 PID/h in WT and S218L HOM, respectively; Fig. 3c) which circled around the hypoperfused core in 38% of S218L mutants, but not in the WT (Movies 1 and 2)²². In fact, higher PID frequencies were associated with larger cortical CBF deficits (Fig. 3d), as well as bigger infarcts (Fig. 3e, f) and worse neurological outcomes (deficit score 1 [1–1] vs. 0 [0–0.25] in S218L HOM and WT mice respectively; P=0.019). These data suggest that ischemic depolarizations adversely influence the perfusion deficits and exacerbate the metabolic and O₂ supply-demand mismatch in FHM1 mutants, in part via vasoconstrictive (i.e., inverse) neurovascular coupling as an additional hemodynamic mechanism for infarct growth^{18, 19, 23}.

Figure 3.

FHM1 mutant mice develop larger areas of CBF deficit during dMCAO because of increased susceptibility to ischemic depolarizations. a) Representative laser speckle contrast images show the area of cortex with ≤30% residual CBF compared to pre-ischemic baseline (blue pixels), 60 min after dMCAO in WT and S218L HOM. Similar data were obtained using the R192Q strain (n=6 mutant and 6 WT; data not shown). Imaging was performed over the right hemisphere (light gray shaded rectangle in the inset), through intact skull. Arrowheads indicate clip occlusion. b) The area of CBF deficit expanded rapidly in S218L HOM throughout the 60 minute dMCAO (*p=0.004). c) Representative tracings show cortical blood flow reductions in penumbra (measured within the gray squares shown in ‘a’) after dMCAO. Anoxic depolarization triggers the first peri-infarct depolarization (blue arrowheads) marking a second abrupt reduction in perfusion. Each subsequent peri-infarct depolarization (red arrowheads) causes a characteristic blood flow transient. d) The frequency of PIDs was higher in S218L HOM, and correlated with the area of hypoperfused cortex 60 minutes after dMCAO (^#p<0.001). Each symbol represents the PID frequency in individual mice. e) Representative TTC-stained whole brains show enlarged infarcts in S218L HOM 48 hours after 60 min dMCAO. f) The area of infarcts in 1 mm-thick coronal slices (0=anterior, 9=posterior) were larger in S218L HOM compared to WT (*p=0.016 S218L HOM vs. WT for infarct areas). Integrated total infarct volumes were also larger in the mutants (19±3 vs. 11±2 mm³, respectively; p=0.008).

Importantly, we found no difference in absolute resting CBF values between WT and R192Q HOM mice in cortex, striatum and cerebellum using [¹⁴C]iodoamphetamine method (Supplemental Table 2), indicating that differences in pre-ischemic resting CBF did not influence our measurements. This was also confirmed in WT and S218L HOM mice under isoflurane anesthesia using the correlation time values obtained by laser speckle imaging, which allow direct comparison of resting CBF among groups of mice (data not shown).^{24, 25} Moreover, the incidence of incomplete circle of Willis, diameter of its major branches, and the number and location of pial arterial anastomoses did not differ between WT and S218L HOM mice, suggesting that developmental differences in cerebrovascular anatomy did not contribute to worse perfusion deficits in the mutant (Supplemental Fig. 1; Supplemental Table 3).

Higher CBF threshold for tissue survival

In order to determine the critical tissue perfusion level below which infarction ensued (i.e., viability threshold), we calculated the regional CBF at the infarct margin by spatially co-registering the laser speckle perfusion map during dMCAO with the infarct that developed 48 hours later (Fig. 4). We found that cortical tissue in S218L HOM mutants required a higher CBF level for survival compared to WT mice (42±3% vs. 35±2% of baseline CBF, respectively; P=0.048). These data underscore the importance of parenchymal mechanisms, such as neuronal hyperexcitability and ischemic depolarizations, as the main cause for increased vulnerability to ischemic stroke in FHM1 mutants, independent of the severity of CBF deficit.

Figure 4.

Elevated blood flow threshold for tissue survival in FHM1 mutant mice. a) Representative laser speckle contrast images (LSCI) during dMCAO (left) and TTC-stained brain showing the infarct 48 hours after 60 min dMCAO (right) are shown for WT and S218L HOM mice. Imaging field was positioned as shown in Figure 3a. Images were spatially co-registered using surface landmarks. Line profiles (blue and green oblique lines, labeled in mm) were drawn between lambda and the clip occluding the middle cerebral artery branch (yellow arrowheads). b) For each animal, cortical blood flow (CBF) was plotted along these line profiles as a function of distance from lambda using laser speckle images, and the blood flow level corresponding to the infarct edge was determined (red dotted lines). This value represented the CBF threshold for viability, below which the tissue infarcted in each mouse. c) The average viability threshold was significantly higher in S218L HOM mutants compared to WT controls (p=0.048) indicating that FHM1 mutant brains are more vulnerable to ischemia and require higher blood flow to survive. Numbers of mice are shown on each bar. *p<0.05 vs. WT.

Worse stroke outcomes

Enhanced susceptibility to anoxic and peri-infarct depolarizations and accelerated hyperacute infarct growth with more severe CBF deficits translated into worse stroke outcomes in FHM1 mutants. Transient fMCAO for 1 hour produced larger infarcts in both S218L and R192Q mutant mice compared to their WT controls (Fig. 5a). Larger infarcts predominantly reflected more severe cortical involvement in both mutants (more than 70% of total infarct volume); however, the incidence of hippocampal or thalamic infarction also tended to be higher in the S218L mutant (present in 33% of S218L HET mice compared to 13% of WT; P=0.1; data not shown), consistent with higher incidence of subcortical infarction observed on MRI in this strain (see above). Functional outcomes, assessed using a combined death and neurological disability score as a clinically relevant endpoint, were worse in mutants compared to WT (Table 1). Indeed, mortality rate was significantly higher in the S218L mutants, reaching 100% in the HOM within 24 hours after stroke onset (Supplemental Fig. 2a). The timing of death after stroke was variable (12±3 hours after stroke onset), and not associated with overt seizure activity. Immediate postmortem examination revealed two-fold larger infarcts in S218L HOM when compared to WT mice sacrificed at the same time point of death of each mutant after 60 min fMCAO (95±15 vs. 44±7 mm³, respectively, n=5 and 4; p=0.031), suggesting that selection bias due to high mortality in the mutants diminished the strain differences in outcome. These data were excluded from the overall comparisons among genotypes (Fig. 5) because of variable time of death. Ischemic brain swelling tended to be more severe in the mutants in proportion to the actual infarct volume, and might have contributed to the high mortality in the S218L mutants.

Figure 5.

FHM1 mutant mice develop larger infarcts after experimental stroke selectively attenuated by NMDA receptor antagonist MK-801. Left panel shows representative infarcts (unstained white tissue) 24 hours after fMCAO. Right panel shows infarct and ischemic swelling volumes (grey and white bars, respectively). a) After 60 minute fMCAO, infarcts were larger in both R192Q and S218L mutants compared to WT (p=0.007 and p=0.019, respectively). One of 9 WT, 8 of 23 S218L HET, and all 8 S218L HOM (†) mice died within 24 hours (Table 1), and were excluded from infarct volume analysis. Because genetic backgrounds and infarct volumes significantly differed between WT controls of the two mutant strains, we did not directly compare S218L and R192Q mutants in this study. MK-801 (1 mg/kg, intraperitoneally 15 minutes before fMCAO) significantly reduced infarct volume in S218L HET mutants (P<0.001 vs. untreated S218L HET shown in ‘a’) but not in the WT (P=0.061 vs. untreated WT shown in ‘a’). Therefore, MK-801 was more efficacious in FHM1 mutants (P=0.026 for infarct reduction by MK-801 between WT and FHM1 mutants). As a result, after MK-801, infarct and swelling volumes were comparable between WT and S218L HET mice (P=0.367), as were neurological outcomes (Table 1). b) Thirty minute fMCAO also resulted in larger infarcts in male and female S218L mutants compared to WT (p=0.028). Despite shorter ischemia, 3 of 4 S218L HOM mutants died within 24 hours and were again excluded from infarct volume analysis; the data from the only surviving HOM mutant are shown. There was no mortality in WT and HET groups after 30 min fMCAO. Numbers of mice are shown on each bar. Standard error bars and p values refer to total volume (i.e., infarct plus swelling). *p<0.05 vs. WT, #p<0.05 vs. untreated S218L HET after 60 min fMCAO.

Table 1.

Death and neurological disability after transient fMCAO in S218L mutant mice.

Experiment	60 min fMCAO*			60 min fMCAO + MK-801†#		30 min fMCAO‡
Gender, Genotype	Male WT	Male HET	Male HOM	Male WT	Male HET	Male WT	Male HET	Male HOM	Female WT	Female HET
N	9	23	8	9	10	9	5	4	6	8
Mortality	11%	35%	100%	0%	10%	0%	0%	75%	0%	0%
Functional outcome score	2 [2–3]	3 [2–5]	5 [5–5]	2 [1–2]	2 [1–2]	1 [1–2]	2 [1–2]	5 [4–5]	1 [0.3–1]	2 [1–2]

Functional outcome scores are shown as median [interquartile range]. 0 = best, 5 = worst, see Methods for the scoring system.

^*p<0.001 and p=0.008 for the effect of genotype on mortality and functional outcome score, respectively.

^†p=0.353 and p=0.757 for the effect of genotype on mortality and functional outcome score, respectively.

^#p=0.303 and p=0.315 for the effect of MK-801 on mortality and functional outcome score, respectively, in WT, and p=0.022 and p=0.009 for the effect of MK-801 on mortality and functional outcome score, respectively, in the S218L HET, compared to the untreated group.

^‡p<0.001 and p=0.086 for the effect of genotype on mortality and functional outcome score in males, respectively, and p=0.02 for the effect of genotype on functional outcome score in females, after 30 min fMCAO. In the R192Q mutant strain, the mortality rate was 17%, 0% and 14% in WT, HET and HOM, respectively; neurological disability was not studied in this mutant strain.

To mimic transient ischemic attacks and circumvent the high mortality rate in S218L HOM mice, we subjected this mutant strain to 30 minutes of fMCAO. With shorter duration of ischemia, we did not detect overt infarcts in 27% of WT mice using TTC staining, whereas all S218L HET and HOM mutant mice developed conspicuous territorial infarcts (p=0.077). Selective ischemic changes in scattered neurons were nevertheless present upon histological examination of brains without an overt infarct (not shown). Infarct volumes were once again larger in the S218L mutants compared to WT (Fig. 5b). The volume of subcortical infarction, limited to the striatum in this shorter ischemia model, was also larger in the S218L HET compared to WT (17±4 and 5±1 mm³, respectively, in males, P=0.002; 16±3 and 4±2 mm³, respectively in females, p=0.013). Because classical, sporadic and FHM are more prevalent in women of reproductive age,^{5, 26–29} and because susceptibility to spreading depression is higher in female FHM1 mutant mice compared to males,^{13, 30} we also studied female mice and found an even more striking increase in infarct volumes in S218L HET compared to WT (Fig. 5b). Despite the shorter ischemia duration, mortality was still high in the S218L HOM mutants (75%), but all HET mutants survived for at least 24 hours and showed a trend for worse functional outcome compared to WT (P=0.086; Table 1). To assess long term tissue and neurological outcome (2 weeks) in FHM1 mutants we subjected female S218L HET and WT mice to 30 min fMCAO. However, we observed more than 60% mortality in the mutants predominantly between 24 and 96 hours, which precluded outcome comparisons between the mutant and WT strains at this late time point (Supplemental Fig. 2b). Altogether, these data indicate that mice expressing FHM1 mutations are particularly susceptible to infarction when challenged by cerebral arterial occlusion.

Enhanced neuroprotective efficacy of glutamate receptor antagonist MK-801

FHM1 mutations enhance glutamate release¹¹, and glutamate plays a pivotal role in spreading depression and ischemic depolarizations, as well as in excitotoxic cell death, mainly via N-methyl-D-aspartate (NMDA) subtype of receptors. Therefore, we tested whether enhanced glutamatergic activity in FHM1 mutants is responsible for their vulnerability to infarction. Pre-ischemic treatment with the NMDA receptor inhibitor MK-801 abolished the differences in stroke phenotype between genotypes. MK-801 reduced infarct volume by 45% in S218L HET compared to only 23% in WT (Fig. 5a), and functional outcome was improved only in S218L mutants (Table 1).

As an important control, we also examined the density of glutamate and GABA_A binding sites in the FHM1 mutant and WT mice using quantitative in vitro autoradiography, and did not find overt differences (Supplemental Fig. 3; Supplemental Table 4). These data are consistent with recent proteomics analysis of cortical synapses in this mutant³¹, and suggest that enhanced susceptibility to ischemic depolarizations in FHM1 mutants is unlikely to reflect changes in neurotransmitter receptors and reuptake mechanisms.

Discussion

Our data provide a novel mechanism to explain the higher incidence of ischemic stroke in migraineurs. Two genetic mouse models expressing FHM1 mutations were at risk to develop large infarcts and worse neurological outcomes after transient focal cerebral ischemia. Consistent with the higher stroke risk in women compared to men with migraine with aura, we found more striking increases in infarct volume in female mutants compared to males. Faster AD rates, more frequent PIDs and enhanced neuroprotective efficacy of NMDA-antagonist MK-801 in the mutants implicated glutamatergic neuronal hyperexcitability as one mechanism, and larger perfusion defects linked to ischemic depolarizations implicated vasoconstrictive neurovascular coupling as another. Therefore, neuronal and vascular mechanisms together render migraineurs more vulnerable to cerebral infarction upon ischemia.

The data have clinical implications. In susceptible migraineurs, increased sensitivity to ischemia may predispose to strokes during mild ischemic events, which remain clinically silent or only manifest as transient ischemic attacks in non-migraineurs. Moreover, elevated CBF threshold for viability³² as a sign of increased vulnerability to ischemia, may promote rapid infarct expansion into tissue with milder perfusion deficits, diminish salvageable tissue at risk (i.e., ischemic penumbra), and shorten the therapeutic window of acute stroke interventions in migraineurs. Lastly, higher mortality among the S218L mutants may have implications for malignant infarcts, large supratentorial strokes characterized by progressive loss of consciousness over 48 hours with up to 80% mortality if untreated³³. To date, there has been no reliable predictor for malignant infarction, and history of migraine with aura (or its genetic determinants) may be one such marker increasing the risk of stroke progression, and in case of large territorial infarcts, the risk of death, as has recently been reported for hemorrhagic stroke in migraineurs³⁴.

Mechanisms of ischemic vulnerability in FHM1 mice

Biological mechanisms underlying the association between migraine and stroke are unknown, although in both diseases dynamic interactions among the constituents of the neurovascular unit are important to the pathophysiology. At the onset of cerebral ischemia the gradual failure of Na⁺, K⁺-ATPase causes a gradual rise in extracellular K⁺ and loss of neuronal membrane potential until a critical threshold for initiation of anoxic depolarization is reached^{35, 36}. FHM1 mutations shift the Ca_V2.1 channel opening voltage to more negative membrane potentials so that channels open with smaller depolarizations triggering glutamate release, which can explain faster anoxic depolarization onset in the mutants. Glutamate is also critical for PIDs, which are spreading depolarization waves triggered in ischemic penumbra. Therefore, enhanced release in ischemic penumbra can also explain higher PID frequencies in FHM1 mutants¹¹. Delayed Ca_V2.1 channel inactivation may further exacerbate the excitotoxicity by prolonging the Ca²⁺ influx and glutamate release during PIDs in penumbra. Moreover, PIDs exacerbate the metabolic mismatch in penumbra by stimulating O₂ and glucose consumption, and by worsening tissue perfusion via vasoconstrictive (i.e., inverse) neurovascular coupling^{18, 19, 23}, particularly when they occur with high frequency and in clusters as observed in FHM1 mutants^{19, 37–39}. With each PID, more of the penumbra is incorporated into the core, accounting for concentric infarct growth over time^{22, 40–42}. PIDs do occur frequently in human brain after ischemic or hemorrhagic stroke and head trauma, and appear to worsen patient outcomes similar to experimental stroke^{38, 43–45}. Hence, migraine with aura may be a risk factor for increased occurrence of PIDs and worse outcomes in human stroke, as recently suggested in subarachnoid hemorrhage⁴⁶.

Association between migraine and stroke

Our data in FHM1 mutant mice support shared genetic risk factors enhancing susceptibility to spreading depression as one mechanism to explain migraine-stroke association. Shared genetic factors enhancing susceptibility to migraine and stroke have been described, such as NOTCH3 mutations in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL)^47–50. Indeed, NOTCH3 mutations, despite being exclusively expressed in vascular smooth muscle cells, augment susceptibility to spreading depression⁵¹. Mutations associated with FHM2 and FHM3 are also predicted to enhance neuronal excitability and spreading depression susceptibility possibly via glutamatergic mechanisms⁵². Glutamatergic mechanisms and hyperexcitability are implicated in common forms of migraine as well, by recent studies linking the AEG1 encoding a regulator of glial glutamate transporter EAAT2, and the KCNK18 encoding the TRESK potassium channel, to migraine^{15, 16}. Vascular mechanisms (e.g. endothelial dysfunction) have also been implicated in increasing stroke risk in migraineurs^{53, 54}. Indeed, functional Ca_v2.1 channel expression has been reported in renovascular smooth muscle cells^{55, 56}, but whether FHM1 mutations alter cerebrovascular physiology is not known. Together with parenchymal mechanisms that enhance vulnerability to perfusion deficits, vascular mechanisms might further augment stroke risk in migraineurs. For example, highly focal and mild ischemic vascular events such as microembolism⁵⁷ may trigger spreading depression more readily in migraineurs highly susceptible to ischemic depolarizations, providing a possible explanation for the origin of a subset of migraine auras.

In summary, a monogenic determinant of migraine with aura increases stroke vulnerability via glutamatergic mechanisms that enhance susceptibility to ischemic depolarizations akin to spreading depression, and accelerate stroke evolution. Hence, our data put FHM1 mutations among the shared genetic determinants of migraine with aura and stroke. More work is needed to extrapolate these data to other monogenic syndromes, as well as to the more common and genetically more complex forms of migraine with aura, and determine whether targeting hyperexcitability and spreading depression, such as migraine prophylaxis⁵⁸, confers ischemic protection in susceptible mouse strains or in migraineurs.