P2X7R/NLRP3 signaling pathway-mediated pyroptosis and neuroinflammation contributed to cognitive impairment in a mouse model of migraine

Seven to eight weeks old male C57BL/6 mice weighing 22-28 g were used in the study. All mice were purchased from the Charles River Laboratories. Mice were housed under specific pathogen-free (SPF) conditions in the Animal Experiment Center of Renmin Hospital of Wuhan University, where food and water were provided ad libitum with temperature 22 ± 2℃, humidity 60 ± 5% and 12-h light/dark cycle. All efforts were taken to minimize animal suffering and reduce the number of animals used. Experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Renmin Hospital of Wuhan University, with the IACUC issue No. WDRM animal (welfare) 20,201,101. The experiments were reported in accordance with the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines. In experiment 1, the mice were randomly assigned to each group: (1) sham, and (2) IS. In experiment 2, the mice were randomly assigned to the following groups: (1) sham-vehicle (VEH, 0.9% saline), (2) sham-BBG (a P2X7R antagonist), (3) IS-VEH, and (4) IS-BBG. Detailed grouping is shown in Fig. 1b. In all experiments, investigators were blind to animal grouping. The mice were acclimated for one week prior to experiment.

Mice were anesthetized initially by 3% isoflurane via anesthetic-specific vaporizers; once the paw pinch reflex of mice disappeared, the concentration of isoflurane was lowered to 1.5% throughout the surgery. The scalp was incised longitudinally and the connective tissues were removed to expose the bregma. A small skull hole (diameter 1 mm, 1 mm after and 1 mm lateral to the bregma) was drilled penetrating through the skull with the dura intact using a skull drill (Reward, 87,001). A plastic cylindrical container (diameter 5 mm, height 2 mm) without a bottom lid was used as a guide cannula. The top of the container has two holes (diameter 1 mm) to facilitate the insertion of the microinjector. The guide cannula was implanted over the previously drilled skull hole and sealed into place with glue. The skin incision was then closed with surgical suture. After surgery, topical erythromycin was administered to each mouse to prevent infection. Mice were returned to their home cage and allowed to recover for 24 h. The detailed procedures have been reported previously [35].

The mice were euthanized by inhaling a lethal dose of carbon dioxide (CO₂) and their brains were harvested. Their cerebral cortex and hippocampus were immediately separated on ice and were frozen in dry ice for immunoblot (n = 6 in each experimental group) and enzyme-linked immunosorbent assay (ELISA) (n = 4 in each experimental group). Another group of animals were sacrificed for histological analysis (n = 6 in each experimental group). Mice were deeply anaesthetized and transcardial perfusion with 40 mL of 0.9% chilled saline then followed by 40 mL of 4% chilled paraformaldehyde. After perfusion, the brains were harvested and fixed overnight in vials containing 4% paraformaldehyde solution.

Brain tissues were homogenized in RIPA buffer (G2002, Servicebio) containing protease inhibitor cocktail (G2006, Servicebio) and phenylmethylsulfonyl fluoride (PMSF, G2008, Servicebio). Protein samples were separated using 8–12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), and then transferred onto polyvinylidene-difluoride (PVDF) membrane to probe for proteins. After being blocked with 5% skimmed milk powder, the PVDF membranes were incubated with the following primary antibodies: P2X7 (1:500, APR-004, Alomone labs), NLRP3 (1:1000, AG-20B-0014, Adipogen), total caspase-1 and cleaved caspase-1 (1:1000, AG-20B-0042, Adipogen), total caspase-3 and cleaved caspase-3 (1:1000, 19,677–1-AP, Proteintech), full length GSDMD (1:1000, ab219800, abcam), N-terminal of GSDMD (1:1000, ab209845, abcam) and β-actin (1:5000, 20,536–1-AP, Proteintech) overnight at 4 °C with agitation. After primary antibody incubation, membranes were washed with 1xTBST (3 times for 10 min each) before incubating with horseradish peroxidase (HRP)-conjugated secondary antibodies (Goat Anti Rabbit, 1:5000, SA00001-2, Proteintech; Goat Anti Mouse, SA00001-1, 1:5000, Proteintech) for 1 h at room temperature with agitation. Then membranes were washed with 1xTBST (3 times for 10 min each). The substrate for HRP, enhanced chemiluminescence (BL523B, Biosharp) was applied before the membranes were visualized using chemiluminescence imaging system (BioRad Laboratories). The chemiluminescence intensity of protein bands was measured using Image J software (Version 1.46; National Institute of Health, Bethesda, MD, USA), and the relative expression of proteins was normalized to the corresponding β-actin.

Proinflammatory cytokines (IL-1β and IL-18) were quantified using ELISA. Brain tissues were homogenized on ice in phosphate buffer saline (PBS, pH 7.4) containing protease inhibitors. Tissue homogenates were centrifuged at 12,000 rpm for 15 min at 4℃. The supernatant was then centrifuged at 12,000 rpm for another 15 min at 4℃. The protein concentration of the supernatant of each sample was measured using bicinchonic acid (BCA) assay. ELISA kit for measuring IL-1β (catalog number: MLB00C; R&D Systems) and IL-18 (catalog number: ML002294, Mlbio) were used to quantify its content in the samples according to the manufacturer’s instructions. The quantity of IL-1β and IL-18 in each brain sample were standardized to the protein concentration.

After post-fixed in 4% paraformaldehyde solution overnight at 4℃, mouse brain tissues were processed into paraffin wax blocks and then cut into coronal Sects. (5 µm-thick). Paraffinized brain sections were dewaxed, hydrated and heat-induced antigen retrieval. Then all brain sections were permeabilized with 0.3% Triton X-100 solution for 10 min and blocked with 5% bovine serum albumin (BSA) solution for 1 h. The sections were immunostained with P2X7 (1:50, APR-004, Alomone labs), caspase-1 (1:200, AG-20B-0042, Adipogen), caspase-3 (1:200, 19,677–1-AP, Proteintech), GSDMD (1:500, ab219800, abcam), microtubule-associated protein 2 (MAP2, 1:200, GB11128-2, Servicebio), myelin basic protein (MBP, 1:200, GB12226, Servicebio), neuronal nuclei (NeuN, 1:200, ab104224, abcam), glial fibrillary acidic protein (GFAP, 1:200, GTX40988, GeneTex), ionized calcium-binding adaptor molecule-1 (Iba-1, 1:500, 019–19,741, Wako) overnight at 4 °C. Subsequently, they were incubated with FITC- (Goat Anti Rabbit, 1:200, GB22303, Servicebio; Goat Anti Mouse, 1:200, GB22301, Servicebio) or CY3- (Goat Anti Rabbit, 1:200, GB21303, Servicebio; Goat Anti Mouse, 1:200, GB21301, Servicebio) conjugated secondary antibodies for 1 h at room temperature in complete darkness. Furthermore, the nuclei were stained using DAPI (G1012, Servicebio). Images were captured under × 20, × 40 and × 100 magnification using an Olympus upright Fluorescence Microscope BX53. Morphological and quantitative analysis were performed in a blinded manner using Image J software (Version 1.46; National Institute of Health, Bethesda, MD, USA). For quantification of positively staining, three randomly microscopic fields were acquired in each section and three adjacent sections were examined from each mouse. The average percentage of the positively stained area was calculated to reflect the degree of positive immunoreactivity.

Cresyl violet staining was performed to detect the neuronal loss in the cerebral cortex and hippocampus. In brief, de-waxed rehydrated sections were immersed in Cresyl Violet solution (G1036, Servicebio) for 5 min and washed in ultrapure water. Then the sections were dehydrated in a graded series of ethanol (70–100%) and finally cleared in xylene. Luxol-fast-blue (LFB) staining was performed in order to evaluate the white matter integrity. Sections were immersed in the LFB solution (G1030, Servicebio) at 37℃ overnight. The excess staining was first removed with 95% ethanol and then washed by ultrapure water. Finally, to differentiate the white matter from the gray matter, the sections were immersed in 0.05% aqueous lithium carbonate for 20 s followed by 70% ethanol until the nuclei are decolorized. The bright field images were captured using an Olympus upright Fluorescence Microscope BX53 with a 4x, 20 × or 40 × objective lens. The number of Nissl-positive cells in cerebral cortex and hippocampal CA1, CA2, CA3, dentate gyrus (DG) regions was used to represent the severity of neuronal loss in the corresponding brain regions. The number of Nissl-positive cells was counted by three examiners who were blinded to experimental conditions, as previously described [27]. The white matter of five brain regions was evaluated: the corpus callosum (Medial), corpus callosum (Paramedian), caudoputamen, internal capsule and optic tract. The severity of the white matter lesions was classified into three levels: Grade 0-normal, Grade 1- disarrangement of the nerve fibers, Grade 2—the formation of visible vacuoles, Grade 3—the disappearance of myelinated fibers.

We detected the expression levels of the key proteins of P2X7R-NLRP3 signaling pathway in the cerebral cortex and hippocampus following repeated dural IS stimulation (Figs. 2 and 3). Expression of P2X7R (p < 0.01), NLRP3 (inflammasome receptor, p < 0.001), total caspase-1 (inflammasome priming, p < 0.01), cleaved caspase-1 (CC1, marker of inflammasome activation, p < 0.001) were all increased in the cerebral cortex, while only expression of P2X7R (p = 0.033) was increased in the hippocampus after dural IS stimulation (Figs. 2a-b and 3a-b). Proinflammatory cytokines IL-1β (p < 0.001) and IL-18 (p < 0.001), direct downstream products of inflammasome activation, were also higher in the cerebral cortex in IS mice than sham controls (Fig. 5e-f). These results indicated the activation of P2X7R-NLRP3 signaling pathway following repeated dural IS stimulation.

We further examined the cellular specificity of P2X7R and NLRP3 inflammasome activation using double immunofluorescence labeling. The P2X7R was mainly expressed in the membrane of NeuN-positive neurons in the cerebral cortex and hippocampus (Fig. 2c-e), while CC1 was mainly expressed in NeuN-positive neurons and Iba-1 positive microglia of cortex and hippocampus (Fig. 3f-g, Supplementary Figs. 1a, d and 2). These results indicated that the dural IS stimulation activated P2X7R on neurons, as well as inflammasome on microglia and neurons.

The active CC1 can promote cell death by activating apoptosis and pyroptosis pathways [24, 25]. We next detected protein expression levels of cleaved caspase-3 (CC3, apoptotic marker) and N-terminal GSDMD (GSDMD-NT, pyroptotic marker) and their precursor proteins (Fig. 3c-e). GSDMD-NT expression was increased in the cerebral cortex after repeated dural IS stimulation (p < 0.001, Fig. 3c, e), but no significant changes in CC3 expression (Fig. 3c-d), suggesting that dural IS stimulation induced pyroptosis-predominant cell death in the cerebral cortex. Consistent with the cellular localization of CC1, GSDMD was also localized in NeuN-positive neurons and Iba-1-positive microglia of cortex and hippocampus (Fig. 3f-g, Supplementary Figs. 1c, f and 4), supporting the occurrence of inflammasome-mediating pyroptosis in cortical and hippocampal neurons and microglia of the IS-induced migraine mouse model.

To elucidate the mechanisms by which the P2X7R-NLRP3 signaling pathway mediates IS-induced cognitive impairment, we examined cognition-related pathological changes including gliosis (microglia and astrocyte) and neuronal loss in the cerebral cortex and hippocampus, and white matter damage. Mice receiving dural IS stimulation exhibited microgliosis and astrogliosis in the cerebral cortex and hippocampus, indicated by increased immunoreactivity of Iba-1 (marker of microglia) and GFAP (marker of astrocyte) (Fig. 4a-d). The cerebral cortex and hippocampus of IS mice also underwent neuronal loss, evidenced by reduced immunoreactivity of MAP2 and decreased Nissl-positive cells in cresyl violet staining (Fig. 4e-f, g-h). The white matter damage was assessed by MBP and LFB staining. No obvious white matter lesions were observed in the cerebral cortex and hippocampus measured by MBP staining and five white matter regions (including medial corpus callosum, paramedian corpus callosum, caudoputamen, internal capsule and optic tract) measured by LFB staining after repeated dural IS stimulation (Fig. 4e-f, i-j). These findings indicated that repeated dural IS stimulation induces microgliosis, astrogliosis and neuronal loss in the cerebral cortex and hippocampus, but no significant white matter damage.

Given the upregulation of P2X7R and activation of its downstream NLRP3 inflammasome signaling pathway in the IS-induced migraine mice model, inhibition of P2X7R with BBG (a specific P2X7R antagonist) is a potential effective therapeutic approach. The IS-induced elevation of P2X7R expression was conspicuously restrained with the administration of BBG (Fig. 5a-b), suggesting that the P2X7R was effectively inhibited. The IS-induced NLRP3 inflammasome activation was attenuated after inhibition of P2X7R, evidenced by a decrease in protein expression levels of the components of NLRP3 inflammasome (NLRP3, total caspase-1) and downstream products of its activation (CC1, IL-1β and IL-18) after application of BBG (Fig. 5a-b, e–f). These findings supported the hypothesis that the IS-induced NLRP3 inflammasome activation was mediated by P2X7R. Furthermore, the expression of pro-pyroptotic protein (GSDMD-NT) was lower of IS-BBG mice than IS-VEH mice (Fig. 5c-d), indicating the IS-induced pyroptosis was attenuated after inhibition of P2X7R. The decreased number of CC1-positive and GSDMD-positive neurons and microglia in the IS-BBG mice compared to IS-VEH mice provided evidence for the role of P2X7R in the inflammasome activation and pyroptosis in neurons and microglia (Fig. 6a, c, d, f, Supplementary Figs. 5a, c, d, f, 6 and 8).

We further explored the role of P2X7R in IS-induced cognition-related pathological changes (i.e., microgliosis, astrogliosis and neuronal loss). We found that inhibition of P2X7R was resistant to IS-induced gliosis, indicated by the lower immunoreactivity of Iba-1 (marker of microglia) and GFAP (marker of astrocyte) in the cortex and hippocampus of the IS-BBG mice than IS-VEH mice (Fig. 7a-b, e–f). Inhibition of P2X7R was also resistant to IS-induced by neuronal loss, evidenced by the IS-induced loss of MAP2- and Nissl-positive cells was partially reversed after administration of BBG (Figs. 7c, g and 8a, d). The role of P2X7R in white matter integrity was assessed by MBP and LFB staining. Immunoreactivity of MBP and the white matter lesions index measured by LFB staining displayed no difference among the four groups, including sham-VEH, sham-BBG, IS-VEH and IS-BBG group (Figs. 7d, h and 8f-j). These results indicated that the IS-induced gliosis and neuronal loss were associated with the IS-induced upregulation of P2X7R.

Finally, we examined the role of P2X7R in IS-induced nociceptive behavior and cognitive deficits. Nociceptive behavior was assessed by periorbital withdrawal thresholds and the number of head-scratching. There was no significant difference in baseline periorbital withdrawal thresholds measured before the first drug administration among four groups (Fig. 9a). The IS-VEH mice showed a significant decrease in post-treatment periorbital withdrawal thresholds measured 1-h after the last drug administration (p < 0.001, Fig. 9b) and an increase in head-scratching within 1-h measured immediately after the last drug administration (p < 0.001, Fig. 9c) compared to sham-VEH mice, indicating successful modeling of the IS-induced migraine mouse model. These IS-induced nociceptive behaviors were partially prevented by pretreatment with BBG, indicated by the higher post-treatment periorbital withdrawal thresholds (p = 0.04, Fig. 9b) and less head-scratching (p < 0.001, Fig. 9c) of IS-BBG mice than IS-VEH mice.

The cognitive functions of mice were assessed by novel object recognition assay (non-spatial recognition memory) and Morris water maze test (spatial learning and memory). In the novel object recognition assay, all mice showed similar explorations of two identical objects in the training stage (Fig. 9e). In the test stage, IS-VEH mice spent less time sniffing the novel object than sham-VEH mice (p < 0.001, Fig. 9f), suggesting a poor non-spatial recognition memory of IS-VEH mice. The impaired recognition memory of IS-VEH mice was improved after BBG application (IS-VEH vs IS-BBG: p < 0.001, Fig. 9f). As for the Morris Water Maze test, there was no difference in the average latency to find a visible platform on day 1 among the four groups of mice (Fig. 9g), suggesting the similar vision and motor ability of all mice. From day 2 to day 5 (learning period), the escape latency gradually decreased in all mice and the slope of the decrease was not significantly different among the four groups of mice (Fig. 9g), suggesting the similar spatial learning ability of all mice. On day 6 (probe trial), IS-VEH mice displayed a shorter time spent in target quadrant (p = 0.024, Fig. 9h) and less platform crossings (p = 0.038, Fig. 9i) than sham-VEH mice, suggesting the impaired spatial memory retention in IS-VEH mice. Although IS-BBG mice displayed a trend with a longer time spent in target quadrant (Fig. 9h) and more platform crossings (Fig. 9i) than IS-VEH mice, the difference did not reach statistical levels. These data showed that IS-induced cognitive impairment was characterized by impaired non-spatial recognition memory and spatial memory retention. Moreover, the impaired recognition memory can be improved by inhibition of P2X7R.

P2X7R, a purinergic receptor family member, has been shown to be upregulated in TNC (a central area related to migraine pain) in a mouse model of NTG-induced migraine [12, 13]. Inhibition of P2X7R can alleviate NTG-induced mechanical and thermal hyperalgesia by negatively modulating the autophagic pathway and reducing proinflammatory cytokine IL-1β release [12, 13]. However, the role of P2X7R in migraine-related cognitive impairment has not been disclosed. Migraineurs frequently exhibited structural and functional abnormalities in the cerebral cortex and hippocampus, which is one of the possible mechanisms accounting for cognitive decline in migraineurs [5, 31]. In the present study, we observed the elevated expression level of P2X7R in cerebral cortex and hippocampus, brain regions related to cognition, following repeated dural IS stimulation (Fig. 2), suggesting a potential role of P2X7R in migraine-related cognitive deficit.

We further identified downstream signaling pathways following P2X7R activation in the IS-induced migraine model. P2X7R regulates multifaceted cellular signaling pathways including cytokine and chemokine secretion, NLRP3 inflammasome activation, cell death and autophagy [42]. Among them, the P2X7R-NLRP3 signaling pathway mediating inflammatory response and cell death has been shown to be involved in cognitive impairment in multiple neurological diseases such as AD, VaD and diabetes [28‐30]. P2X7R activation can initiate the assembly of the NLRP3 inflammasome complex [16, 17], an intracellular multimeric protein complex consisting of a stimulus-detecting sensor NLRP3, the adaptor protein ASC and effector protein total caspase-1 [43]. Stimulation of NLRP3 activates its effector (total caspase-1) into biologically active CC1 [18‐20]. The active CC1 facilitates inflammatory response by releasing proinflammatory cytokines IL-1β and IL-18 [21]. The active CC1 may also promote cell death by activating apoptosis and pyroptosis pathways [24, 25]. CC1 induces pyroptosis and apoptosis by cleavage of total caspase-3 and full length GSDMD into CC3 and GSDMD-NT respectively [24, 25]. GSDMD-NT can cause pore formation in the plasma membranes, resulting in leakage of proinflammatory cytokines and cellular content [44, 45]. The P2X7R-NLRP3 signaling pathway was depicted in detail in Fig. 10. It has been shown that NLRP3 inflammasome was activated in the TNC (a central region related to migraine pain) and apoptotic marker (CC3) was increased in the trigeminal ganglia (a peripheral region related to migraine pain) in an NTG-induced migraine mouse model, suggesting that NLRP3 inflammasome activation and apoptosis may be at least partially responsible for migraine pain sensitization (including peripheral and central sensitization) [33, 46]. Our present data indicated that repeated dural IS stimulation led to increased expression of NLRP3 inflammasome components (NLRP3 and total caspase-1), marker of inflammasome activation (CC1), downstream proinflammatory cytokines (IL-1β and IL-18) and pro-pyroptotic protein (GSDMD-NT) in the cerebral cortex (Fig. 3). The pro-apoptotic protein CC3 did not change significantly after dural IS stimulation. The IS-induced NLRP3 inflammasome activation, proinflammatory cytokines release and pyroptosis were attenuated by BBG, a specific P2X7R antagonist (Fig. 5). These findings supported our hypothesis that P2X7R-NLRP3 signaling pathway was activated in the cognition-related brain regions (i.e., cortex and hippocampus) following dural IS stimulation. The present study was the first to demonstrate that NLRP3 inflammasome-mediating pyroptosis occurs in cerebral cortex and hippocampus of a mouse model of IS-induced migraine. This study also provides a possibility that the different ways to induce migraine attacks may activate different cell death pathways in different brain regions and mediate different clinical manifestations of migraine.

To explore the mechanisms by which the P2X7R-NLRP3 signaling pathway mediates IS-induced cognitive impairment, we examined cognition-related pathological changes including gliosis (microgliosis and astrogliosis) and neuronal loss in the cerebral cortex and hippocampus, and white matter damage. It has been shown that inflammasome activation can mediate cognitive decline by inducing gliosis and neuronal loss in the cerebral cortex and hippocampus, and white matter damage in a mouse model of VaD [27]. Our previous study showed that microgliosis and astrogliosis occurring in the medullary dorsal horn after repeated dural IS stimulation were associated with nociceptive behavior in an animal model of IS-induced migraine.[34]. Here, we found that microgliosis, astrogliosis and neuronal loss were present in the cerebral cortex and hippocampus following dural IS stimulation (Fig. 4), which may account for the structural and functional alterations in the cerebral cortex and hippocampus of migraine patients previously observed in clinical neuroimaging studies [5, 31]. These pathological changes were attenuated by pretreatment with BBG (Figs. 7 and 8), suggesting a role for P2X7R in IS-induced gliosis and neuronal loss.

No obvious white matter lesions were observed in multiple white matter regions following repeated dural IS stimulation (Fig. 4), which was in contrast to four studies using diffusion-weighted magnetic resonance imaging (MRI) [47, 48] and magnetization transfer imaging (MTI) [49, 50], but in line with two MTI studies [32, 51]. A possible explanation might be that migraine patients with white matter lesions were those who had more severe migraine attacks and longer disease duration. Moreover, the MBP and LFB staining used in the present study can only detect visible white matter lesions. Focal invisible microstructural white matter changes might occur in migraine patients preceding visible focal white matter lesions [49, 50]. Thus, our results could not rule out the potential involvement of white matter damage in migraine-related cognitive impairment. Further studies are needed to evaluate the white matter lesions in a more severe migraine model with more sensitive indicators.

To clarify whether inflammasome activation and cell death are cell-specific, we examined the cellular localization of markers of inflammasome activation (CC1), apoptosis (CC3) and pyroptosis (GSDMD). The results indicated that inflammasome CC1 and pro-pyroptotic GSDMD were mainly expressed in neurons and microglia, whereas pro-apoptotic CC3 was mainly expressed in neurons (Fig. 3). The similar cellular localization of CC1 and GSDMD provides evidence for the occurrence of inflammasome-mediating pyroptosis in cortical and hippocampal neurons and microglia in an IS-induced migraine mouse model. The pro-apoptotic CC3 was mainly expressed in cortical and hippocampal neurons, but its expression level was not affected by dural IS stimulation. A possible explanation for this is that the apoptosis in neurons may be induced by cannula implantation surgery, independent of the dural IS stimulation. Despite increased apoptosis was reported in the trigeminal ganglion in an NTG-induced migraine mouse model [46], there is no evidence of apoptosis in central nervous system of migraine patients or animal models of migraine, which is consistent with the present study. No colocalization of CC1, CC3 and GSDMD was observed in astrocytes, indicating inflammasome activation and cell death did not occur in astrocytes, which is in contradiction with astrogliosis following dural IS stimulation. The precise molecular mechanisms for this remain unclear. It is possible that the astrogliosis might be a compensatory mechanism secondary to IS-induced brain injury, and might play a role in neuronal repairment. The number of CC1-positive and GSDMD-positive neurons and microglia decreased after application of BBG (Fig. 6), indicating that the inflammasome activation and pyroptosis in neurons and microglia were mediated by P2X7R.

We further evaluated the effects of P2X7R on nociceptive and cognitive behaviors in the IS-induced migraine model. The role of P2X7R for migraine-related pain symptoms has been elucidated in an NTG-induced migraine mouse model, which reported that inhibition of P2X7R can alleviate NTG-induced mechanical and thermal hyperalgesia [12, 13]. Our results indicated that the reduced periorbital mechanical pain threshold and increased spontaneous pain behavior of the IS mice can be alleviated by inhibition of P2X7R (Fig. 9), supporting the previous findings [52, 53]. The results of cognitive behavior tests showed that the IS mice exhibited a poor recognition memory in novel object recognition test, and impaired spatial memory retention in probe trial of Morris water maze, but no obvious impairment of spatial learning ability in the learning period of Morris water maze (Fig. 9). Inhibition of P2X7R can improve the IS-induced selective cognitive impairment, indicating that P2X7R is a key contributor to migraine-related cognitive impairment. The selective cognitive impairment may be due to the more severe IS-induced brain injury in the cerebral cortex than in the hippocampus. It has been shown that the spatial learning ability in the Morris water maze critically depends on hippocampal function [54], while the recognition memory in novel object recognition test mainly relies on limbic-cortical function [55]. A recent clinical study reported that the impaired visuospatial processing of migraine patients was associated with abnormal recruitment of cortical areas including the anterior insula, medial frontal and orbitofrontal cortex [5], demonstrating a key role of cortical areas in migraine-related cognitive deficits. However, these cognitive behavioral results were in contrast with a previous study conducted on a genetic mouse model of familial hemiplegic migraine type 1 (FHM-1) [56]. This study reported that FHM-1 gain-of-function mutations selectively impaired the spatial memory in Morris water maze by enhancing hippocampal long-term potentiation, but had no effect on non-spatial recognition memory in novel object test [56]. Different animal models of migraine used in the two studies might be a major cause for the contradictory results. Several clinical neuropsychological researches reported that cognitive symptoms in migraine patients are mainly presented as visuospatial abilities, processing speed, attention and executive functions [3‐6]. Moreover, neuroimaging studies have revealed that the structural and functional alterations in the cerebral cortex and hippocampus were involved in cognitive impairment in migraine patients [5, 31]. Since any animal models of migraine cannot fully mimic the complex pathogenesis and clinical manifestation of migraine, clinical studies are needed to explore the correspondence between the affected cognitive domains and specific brain regions in migraine patients.