EGR1-associated inflammatory and neurovascular signatures suggest a potential link between migraine and ischemic stroke

Abstract

Background

Migraine is associated with an increased risk of ischemic stroke, but the molecular mechanisms linking these two disorders remain unclear.

Methods

We performed an integrated analysis of bulk RNA-seq data from ischemic stroke and single-cell RNA-seq data from a mouse migraine model. Differential expression, Gene Ontology enrichment, cell-cell communication, protein-protein interaction, disease association, and drug-gene interaction analyses were conducted to identify shared molecular signatures and pathways. A nitroglycerin-induced migraine mouse model was further used to validate neurovascular alterations in vivo.

Results

Integrated transcriptomic analysis identified shared upregulated genes between migraine and ischemic stroke, with IL1B and EGR1 emerging as key candidates. In ischemic stroke, enriched pathways were mainly related to immune and inflammatory responses, particularly immune response-regulating cell surface receptor signaling and interleukin-1-mediated signaling, with IL1B and EGR1 emerging as prominent candidates in the enriched network context. Single-cell analysis showed that EGR1 was the only significantly shared upregulated gene in migraine, with elevated expression in PEP neurons, NF neurons, vascular cells, and fibroblasts, while the interleukin-1 production pathway was activated in most of these cell types. Cell-cell communication analysis revealed enhanced interactions among neuronal, vascular, and fibroblast populations, especially through ANGPTL signaling. Network analysis highlighted EGR1, IL1B, TLR4, and ANGPTL2 as candidate hub-associated molecules. In vivo, the migraine model showed increased neuronal activation, persistent mechanical hypersensitivity, and reduced ZO-1 expression in the trigeminocervical complex, indicating vascular tight junction impairment.

Conclusion

These findings identify shared inflammatory and neurovascular signatures across migraine-related and ischemic stroke-related datasets, with EGR1 emerging as a candidate molecule associated with these convergent changes. Our results support a hypothesis-generating model in which inflammatory signaling and altered neurovascular communication may represent potential links between migraine and stroke-related vascular vulnerability. Further functional studies are required to determine whether EGR1 or related pathways play a causal role.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s10194-026-02447-3.

Introduction

Migraine is a complex brain disorder characterized by recurrent headaches accompanied by various neurological and vascular symptoms. Patients typically experience photophobia, phonophobia, nausea, vomiting, as well as cognitive and emotional disturbances. Although traditionally regarded as a “benign” functional disorder, epidemiological evidence indicates that migraine is associated with a modest but reproducible increase in the risk of cerebrovascular and cardiovascular events, particularly ischemic stroke and, in some studies, myocardial infarction. This effect is generally moderate at the population level and appears to be more pronounced in patients with migraine with aura [1–3]. This association cannot be attributed to a single risk factor. Instead, it reflects a multi-level imbalance within the neurovascular unit [4]. Abnormal activation of the trigeminovascular system [5], impaired endothelial function, enhanced platelet reactivity, low but persistent immune activation, and accumulated oxidative stress are all considered to be the causes of this phenomenon. Understanding how these molecular and cellular dysfunctions interact with each other to link migraine to vascular fragility has become a key issue in the interdisciplinary field of neurology and vascular biology.

Chronic inflammation and endothelial dysfunction have been identified as the key mechanism link between migraine and increased risk of cerebrovascular events [6, 7]. Under normal circumstances, endothelial cells maintain the homeostasis of cerebral microcirculation by regulating the bioavailability of nitric oxide, vascular tension, the expression of adhesion molecules, and the balance between coagulation and fibrinolysis [8]. When inflammatory mediators and reactive oxygen species persistently increase chronically, endothelial cells will switch to an activated state, characterized by increased permeability, enhanced expression of adhesion molecules, and a pro-thrombotic tendency [7, 9]. Immune cells play a dual role in this process. Cytokines released by monocytes/macrophages, T lymphocytes, and natural killer cells can sensitize trigeminal nerve endings and promote vasogenic pain [10]. At the same time, immune-endothelial interactions mediated through ligand-receptor signaling pathways can alter the behavior of endothelial cells, promote microthrombosis, and promote uneven perfusion [11]. Previous transcriptomic analyses have reported that inflammatory and coagulation-related genes in the peripheral blood of migraine patients have been upregulated; however, these findings mostly come from overall-level data and lack cellular-level analysis or multi-layer validation across different independent datasets.

Most previous studies relied on a single data layer, usually based on data from a peripheral blood microarray cohort, which made it difficult to rule out the influence of cohort-specific factors and platform-related variability [12]. Secondly, evidence at the cellular resolution level is relatively limited [13], which leaves the question of which immune cell populations drive these interrelated pathways unanswered. Thirdly, endothelial dysfunction are rarely studied jointly within the same analytical framework, and there is a lack of systematic integration of population-level transcriptomics, single-cell analysis, pathway and cell communication analysis, as well as cross-dataset validation research [14]. Constructing a stepwise evidence chain from overall differential expression, cell type localization, cell-to-cell communication, and cross-dataset pathway convergence will help strengthen mechanism inference and better meet the clinical need for actionable and cell-specific therapeutic targets.

In the present study, we integrated bulk RNA-seq data from ischemic stroke with single-cell RNA-seq data from a mouse migraine model to identify shared genes, pathways, and cellular interaction networks. We further combined bioinformatic analyses with in vivo validation in a nitroglycerin-induced migraine model to test whether migraine is associated with vascular barrier disruption. Using this strategy, we sought to identify shared inflammatory and neurovascular signatures across migraine-related and ischemic stroke-related datasets and to generate testable hypotheses regarding potential molecular links between these conditions.

Methods

Animals

Male C57BL/6J mice, aged 10 weeks, were acquired from the Experimental Animal Center of the First Affiliated Hospital of Zhengzhou University. The animals were housed in specific pathogen-free (SPF) conditions at a controlled temperature of 22 ± 2 °C, with a 12-hour light/dark cycle and ad libitum access to standard chow and water. After a one-week acclimatization period, 12 mice were randomly assigned to experimental and control groups based on a sequence generated using R (version 4.0.2). The study protocol was reviewed and approved by the Animal Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval number ZZU-LAC2025042202), and all experimental procedures followed the ethical guidelines set forth by the International Association for the Study of Pain [15].

NTG-induced model preparation

A chronic migraine model induced by NTG was established using slightly modified protocols from previously validated methods [16, 17]. The NTG stock solution (5 mg/mL) was prepared in a vehicle composed of 30% ethanol, 30% propylene glycol, and 40% distilled water. Before administration, the stock solution was diluted with sterile 0.9% saline to achieve a final concentration of 1 mg/mL. Mice in the model group were administered intraperitoneal injections of NTG at a dose of 10 mg/kg, while control animals received an equivalent volume of saline. The injections were given every two days for a total of five sessions over a 9-day period to establish a stable migraine-like hypersensitivity phenotype.

Behavioral assessment

Mechanical nociceptive thresholds were assessed using von Frey filaments with calibrated bending forces ranging from 0.008 to 2 g, following the up-down method described previously [18]. Two areas were tested: the periorbital region and the hind paw. Before behavioral testing, mice were habituated to the behavioral testing room for 30 min per day for 3 consecutive days. On each testing day, mice were allowed to acclimate to the testing environment for an additional 30 min before assessment. During testing, the filament was applied perpendicularly to the skin surface for approximately 3 s, starting with a 0.16 g filament. A positive response was defined as an immediate head withdrawal in the periorbital region or a brisk paw flick/withdrawal in the hind paw. If no response occurred, the next stronger filament was applied, whereas a positive response led to application of the next weaker filament, according to the staircase logic of the up–down method.

To minimize sensitization, a 3-minute interval was maintained between consecutive stimulations. The 50% mechanical withdrawal threshold (g value) was determined using an online analysis tool (https://bioapps.shinyapps.io/von_frey_app/), as described by Christensen et al. [19]. This algorithm is based on Dixon’s non-parametric optimization method and reduces the influence of individual response variability on threshold estimation. Behavioral measurements were obtained 2 h before and 2 h after each injection. All assessments were performed under double-blind conditions, with both experimenters and data analysts unaware of group allocation.

Tissue perfusion and cryosection preparation

Following deep anesthesia, mice underwent transcardial perfusion with pre-chilled 0.9% saline, followed by 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS, pH 7.4). The brains were carefully extracted and post-fixed in 4% PFA at 4 °C for 12–16 h. The segment of the medulla containing the trigeminal nucleus caudalis (TNC) was isolated and cryoprotected sequentially in 20% and 30% sucrose solutions in PBS until the tissue sank. Once dehydrated, the samples were embedded and frozen for sectioning. Coronal cryosections (10 μm thick) were made using a cryostat microtome (Leica, Japan) at low temperature, and the sections were mounted onto adhesive microscope slides for subsequent immunofluorescence analysis.

Immunofluorescence staining and imaging

Cryosections were allowed to equilibrate to room temperature and were washed gently with phosphate-buffered saline (PBS). To minimize nonspecific antibody binding, the sections were preincubated with 5% bovine serum albumin (BSA) for 1 h at room temperature. Following this, primary antibodies diluted 1:500 in PBS containing 1% BSA were applied and incubated overnight at 4 °C. After thorough washing with PBS, fluorophore-conjugated secondary antibodies were added and incubated for 2 h at room temperature in the dark. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) for 10 min.

Once staining was complete, the slides were mounted with an antifade reagent and examined using a confocal laser scanning microscope (Zeiss 980, Germany). Image acquisition and quantitative analysis of fluorescence intensity and positive signal area were performed using ImageJ software (version 2.3.0, NIH, USA), ensuring that the exposure parameters were consistent across all samples for reliable comparison.

Data collection

Publicly available peripheral blood bulk and single-cell RNA sequencing datasets were utilized for this analysis. Batch bulk data related to cerebrovascular diseases were obtained from the GEO database (GSE22255 and GSE58294), both of which include clinical cases as well as matched healthy controls from peripheral blood samples. The differentially expressed genes in the peripheral blood RNA-seq of migraine patients were sourced from the study by Timea Aczél et al. [20]. For the mouse single-cell data, it was obtained from the research by Lite Yang et al., available in the NCBI GEO database under accession number GSE197289 [21]. All datasets were analyzed in compliance with public data access and privacy regulations.

Data preprocessing and differential expression analysis

Raw microarray probe signals were mapped to standard human gene symbols using the platform annotation file. For genes with multiple probes, the median probe value was taken to obtain a unique expression value. Each dataset underwent independent background correction, normalization, and quality control, with outlier samples removed prior to integration.

To remove technical batch effects, we applied an empirical Bayes-based batch correction method, ensuring expression distributions and clustering patterns were primarily driven by biological rather than technical variation. Differential expression analysis was conducted using the limma package’s empirical Bayes moderated linear model, followed by multiple testing correction. Significant genes were defined as those with |Log2 fold change| > Log2(1.5) and adjusted P-value < 0.05.

Single-cell data analysis

The single-cell data were processed using Seurat, including quality control, normalization, dimension reduction and clustering [22]. Low-quality cells were removed, and the dataset was integrated through anchor-based alignment to correct batch differences. Cell maps were constructed through principal component analysis and nonlinear embedding. Cell subpopulations were annotated based on standard immune cell marker genes, and the annotations were confirmed and refined with the help of SingleR and a reference peripheral blood mononuclear cell dataset. Then, the expression patterns of pathways and key genes related to migraine identified from the batch analysis were examined in immune subpopulations to determine the cellular sources and activation characteristics of peripheral immune signals in migraine.

Functional enrichment

Functional enrichment analysis of differentially expressed genes (DEGs) was performed using the clusterProfiler package (version 4.16.0) in R [23]. Gene Ontology (GO) enrichment analysis was conducted to categorize genes into biological process, molecular function, and cellular component terms (http://www.geneontology.org/) [24]. GO enrichment analysis was conducted to classify genes into biological process, molecular function, and cellular component categories. Adjusted p values were calculated using the Benjamini-Hochberg false discovery rate (BH-FDR) correction, and adjusted p < 0.05 was considered statistically significant. The minimum gene set size was set to 10 (minGSSize = 10). Unless otherwise specified, standard analysis settings were used.

To further evaluate the disease, trait, and drug-target relevance of the candidate gene set, enrichment analyses were performed using the Enrichr web platform (https://maayanlab.cloud/Enrichr/) [25]. The DisGeNET, GWAS Catalog, and DGIdb libraries were used for disease, trait, and drug-target enrichment analyses, respectively. Statistical significance for these analyses was calculated using Fisher’s exact test as implemented by Enrichr, and the returned enrichment results were used for ranking and visualization.

Cell communication analysis

Cell-cell communication analysis was performed using CellChat (version 1.6.1) to infer ligand-receptor interaction networks from single-cell transcriptomic data [26]. Based on the transcriptional expression levels of ligands and receptors, this model infers the signal strength and pathway activity between cell subpopulations, and determines the main signaling axes as well as the changes in the sending and receiving capabilities of specific cell populations. Significantly increased communication signals in the migraine group were defined using a significance threshold of p < 0.05. The analyses focused on communication changes among neuronal, vascular, immune, and fibroblast-related cell populations.

Protein-protein interaction analysis

Protein-protein interaction (PPI) analysis was performed using the STRING database. A Mus musculus interaction network centered on EGR1 was constructed using a minimum required interaction score of 0.4. The final network contained 32 proteins and 105 interactions. Candidate core molecules were interpreted according to network connectivity together with their recurrence across the integrative transcriptomic analyses.

Statistical analysis

All analyses were conducted in the R environment. The key packages used include limma (version 3.64.3), Seurat (version 5.4.0), SingleR (version 2.12.0), CellChat (version 2.1.2), and clusterProfiler (version 4.16.0), which were employed for data processing, differential expression analysis, functional enrichment, and communication modeling. The statistical tests were two-sided tests, and the P-values were corrected for multiple comparisons. Data visualization was carried out using basic plotting functions and ggplot2 (version 4.0.1). The analysis process underwent multiple independent reviews to ensure the reproducibility and stability of the results.

Results

Integrated transcriptomic analysis identifies shared upregulated genes between migraine and ischemic stroke

To investigate the molecular association between migraine and ischemic stroke, we performed an integrated analysis of bulk RNA-seq and single-cell RNA-seq datasets, followed by experimental validation in a migraine mouse model (Fig. 1A). Differential expression analysis of the ischemic stroke RNA-seq dataset revealed a distinct transcriptional profile, as illustrated by the volcano plot (Fig. 1B, Supplemental Table S2). Cross-condition comparison further identified a subset of commonly upregulated genes, including IL1B and EGR1, as shown in the Venn diagram (Fig. 1C, Supplemental Table S2). These findings indicate that migraine-related and ischemic stroke-related datasets share overlapping molecular signatures. These convergent signals suggest potential common inflammatory features, but they should be interpreted as associative observations rather than evidence of a defined pathogenic mechanism.

Fig. 1.

Integrative workflow for identifying shared molecular features between migraine and ischemic stroke. (A) Overview of the experimental workflow. Bulk RNA-seq and single-cell RNA-seq datasets were analyzed to explore the molecular association between ischemic stroke and migraine. Differentially expressed genes (DEGs) were identified and subjected to network analysis, highlighting key pathways such as EGR1 regulation, interleukin-1 production, and smooth muscle cell (SMC) proliferation. A migraine mouse model was utilized for experimental validation, where drug administration (NTG) was performed across multiple time points (1, 3, 5, 7, and 9 days) following habituation, with immunofluorescence (IF) imaging for validation of endothelial cell (EC) damage. (B) Volcano plot of the ischemic stroke RNA-seq dataset. The plot illustrates the log2 fold changes (x-axis) and -log10 adjusted P-values (y-axis) for differentially expressed genes. Upregulated genes are highlighted in red, downregulated genes in blue, and non-significant genes in gray. (C) Venn diagram showing the overlap of genes upregulated in both stroke and migraine conditions

Immune-inflammatory functional modules are prominently enriched in ischemic stroke

Next, we performed Gene Ontology enrichment analysis to characterize the biological functions of the differentially expressed genes identified in ischemic stroke. The most significantly enriched biological process terms were predominantly related to immune and inflammatory responses, including the immune response-regulating cell surface receptor signaling pathway, interleukin-1-mediated signaling pathway, and cellular response to mechanical stimulus (Fig. 2A). We further visualized the relationships between these pathways and representative genes, including CD177 and IL1B, in a circular plot (Fig. 2B). Network analysis further showed that the enriched functional network was primarily organized around immune- and inflammation-related biological processes, including immune response-regulating cell surface receptor signaling and interleukin-1-mediated signaling (Fig. 2C). Genes such as IL1B and EGR1 were linked to these pathways, but this panel was intended to emphasize the inflammatory nature of the enriched network rather than to define gene centrality based on visual node size alone. Together, these results support a central role for immune-inflammatory dysregulation and cellular stress responses in ischemic stroke.

Fig. 2.

GO enrichment and network context of immune-inflammatory pathways in ischemic stroke. (A) Gene Ontology (GO) enrichment analysis of differentially expressed genes in ischemic stroke. The most significantly enriched biological processes are shown, primarily associated with immune and inflammatory responses. Terms such as immune response-regulating cell surface receptor signaling pathway, interleukin-1-mediated signaling pathway, and cellular response to mechanical stimulus are highlighted. Red indicates upregulated genes, while blue indicates downregulated genes. (B) Circular plot visualizing the relationships between enriched biological processes and representative genes in ischemic stroke. (C) Network analysis of immune-inflammatory pathways in ischemic stroke. Gene size corresponds to the degree of connectivity, with larger nodes representing more central genes in the network

Single-cell transcriptomics identifies EGR1 as a shared upregulated gene in migraine-relevant cell populations

Single-cell RNA-seq data from a mouse migraine model were analyzed to define the cellular context of shared molecular signals. UMAP visualization delineated the overall cellular composition and revealed disease-associated shifts across cell populations (Fig. 3A and B), while the dot plot identified characteristic marker genes for each cell type (Fig. 3C). Grouped differential expression analysis showed that EGR1 was the only shared gene significantly upregulated in the migraine condition relative to naive controls, with increased expression observed in PEP neurons, NF neurons, vascular cells, and fibroblasts (Fig. 3D). In parallel, pathway enrichment analysis demonstrated activation of the interleukin-1 production pathway in the majority of these cell populations. These observations show that EGR1 is broadly upregulated across several migraine-relevant cell populations and may represent a shared candidate marker associated with inflammatory activation across datasets. However, the present data do not establish whether EGR1 functions as a causal driver of this process.

Fig. 3.

Single-cell transcriptomic profiling reveals EGR1 as a key upregulated gene in migraine-related cell populations. (A) UMAP plot of single-cell RNA-seq data from a mouse migraine model. Each cell is colored according to its identified cell type, including cLTMR, Fibroblast, Immune, NF, PEP, Satglia, Schwann, SST, TRPM8, and Vascular. This plot highlights the overall cellular composition and shifts in cell populations associated with the migraine condition. (B) UMAP visualization showing the comparison between migraine and naive control conditions. Cells are colored by group (migraine vs. naive), revealing disease-associated shifts across cell populations. (C) Dot plot of marker genes for each cell type in the migraine model. The plot shows the percent of cells expressing each gene (dot size) and the average expression level (color intensity) across cell types, with distinct markers highlighted for each population. (D) Differential expression analysis comparing the migraine condition to naive controls. The plot shows log2 fold change (x-axis) for each gene across different cell types (y-axis). Pathway enrichment analysis reveals activation of the interleukin-1 production pathway in these cell populations, suggesting a broad inflammatory activation linked to EGR1

Enhanced neurovascular and fibroblast communication characterizes the migraine-associated inflammatory microenvironment

To further explore potential intercellular communication changes in the migraine model, we first performed CellChat analysis to compare ligand-receptor-based communication patterns between the naive and migraine groups. This global analysis showed significantly increased predicted communication signals among NF neurons, PEP neurons, vascular cells, immune cells, and fibroblasts in the migraine group (Fig. 4A). Within this communication network, ANGPTL signaling was notably enhanced, particularly in interactions between NF neurons and immune fibroblasts as well as vascular cells (Fig. 4B). To place these communication-related findings into a broader molecular context, we further constructed a STRING-based protein-protein interaction network of candidate molecules associated with the inflammatory-neurovascular analysis framework. This complementary network analysis highlighted EGR1, IL1B, TLR4, and ANGPTL2 as candidate hub-associated molecules (Fig. 4C). Together, these panels provide a stepwise view from global inferred communication changes, to pathway-level refinement, to candidate molecular interaction context.

Fig. 4.

Stepwise analysis of altered communication and related molecular interaction patterns in migraine. (A) Dot plot of intercellular communication signaling pathways in the migraine and naive groups. The plot shows increased interaction strength in the migraine group compared to the naive group, with significant alterations observed in communication among NF neurons, PEP neurons, vascular cells, and fibroblasts. The size of the dots represents the p-value, with darker shades indicating stronger significance. (B) ANGPTL signaling pathway network in both naive and migraine groups. The network diagram illustrates enhanced ANGPTL signaling in the migraine group, particularly between NF neurons and immune fibroblasts, as well as vascular cells. The lines indicate the communication strength between cell types, with thicker lines representing stronger interactions. (C) Protein-protein interaction network constructed from the STRING database. Nodes represent individual proteins; node color encodes the STRING combined interaction score between each protein and Egr1, displayed as a continuous blue gradient (white: score = 0, no direct interaction; dark blue: score ≥ 0.65), as indicated by the color scale bar on the right. Egr1 is highlighted in red. Edges represent predicted functional associations; red edges connect Egr1 directly to its six first-order interactors (Il1b, Sdc4, Tlr4, Vegfa, Pdgfa, and Insr), while gray edges represent interactions among the remaining proteins. Edge width is proportional to the STRING combined score. Proteins with no direct interaction with Egr1 are shown in light gray

Complementary disease, trait, and drug-target annotation analyses of the candidate gene set

To further contextualize the candidate gene set identified from the integrative analyses, we performed three complementary database-based enrichment analyses using the same input gene list. DisGeNET enrichment analysis was used to examine disease-category associations and showed enrichment for inflammation-, arteriosclerosis-, and atherosclerosis-related entries (Fig. 5A, Supplemental Table S3). GWAS Catalog enrichment analysis was then used to assess trait-level associations and linked the candidate gene set to cytokine network levels, dysmenorrheic pain severity, and cardiovascular risk-related traits (Fig. 5B, Supplemental Table S4). Finally, drug-target enrichment analysis using the DGIdb library identified compounds associated with this gene set, among which SUNITINIB ranked among the top enriched agents (Fig. 5C).

Fig. 5.

Complementary disease, trait, and drug-target annotation analyses of the candidate gene set. (A) DisGeNET disease enrichment analysis of the identified hub genes. The plot shows the association of the hub gene set with various disease categories. Significant enrichments are observed in inflammation, arteriosclerosis, and atherosclerosis, with the most prominent disease categories indicated. The -log(p value) is plotted on the y-axis. (B) GWAS Catalog enrichment analysis of the hub gene set. The plot displays associations with various traits, including cytokine network levels, dysmenorrheic pain severity, and cardiovascular risk factors. The size of the circles corresponds to the odds ratio, and the color intensity represents the significance level (-log(p value)). (C) Drug-gene interaction analysis using the DGIdb database. The bar plot shows the -log(p value) for candidate compounds targeting the hub gene network. SUNITINIB, ranked among the top predicted agents, is identified as a potential therapeutic candidate for targeting the gene network associated with inflammation and vascular pathology

These three panels should be interpreted as complementary annotation layers that provide exploratory disease, phenotype, and drug-association context for the same candidate gene set. They do not constitute a stepwise demonstration of therapeutic relevance, and the drug-associated results should not be interpreted as evidence of efficacy or clinical applicability.

Migraine induces neuronal activation and vascular tight junction disruption in vivo

To validate the vascular effects of migraine in vivo, we established a nitroglycerin-induced mouse model. Immunofluorescence analysis of the trigeminocervical complex (TNC) demonstrated a significant increase in both the number of cFos-positive activated neurons and the proportion of cFos⁺/NeuN⁺ neurons in NTG-treated mice compared with vehicle controls (Fig. 6A and B). Behavioral assessment further confirmed successful model induction, as NTG-treated mice exhibited a sustained reduction in periorbital mechanical threshold after treatment (Fig. 6C and D). Importantly, expression of the vascular tight junction marker ZO-1 was markedly reduced in the TNC of the migraine model group (Fig. 6E and F). These results support the presence of migraine-associated neuronal activation and reduced ZO-1 expression in the trigeminocervical complex. While these in vivo observations are consistent with neurovascular alteration, they do not by themselves establish a causal link to ischemic stroke risk.

Fig. 6.

Migraine induces neuronal activation and vascular tight junction impairment in vivo (A) Immunofluorescence images of the trigeminocervical complex (TNC) showing cFos (red) and NeuN (green) staining in vehicle and NTG-treated mice. The DAPI stain (blue) is used to visualize cell nuclei. The merge panel shows the colocalization of cFos and NeuN in activated neurons. Scale bar = 50 μm. (B) Quantification of the proportion of cFos+/NeuN+ double-positive neurons in the TNC. NTG-treated mice show a significant increase in the number of activated neurons compared to vehicle controls (**p < 0.01). (C) Basal mechanical threshold response in the periorbital test. NTG-treated mice exhibit a significant reduction in the mechanical threshold over the course of the study compared to vehicle-treated mice (****p < 0.001). (D) Post-treatment response in the periorbital test. NTG-treated mice show a sustained reduction in mechanical threshold, confirming the development of hypersensitivity after treatment (****p < 0.001). (E) Immunofluorescence images showing the expression of ZO-1 (red), a vascular tight junction marker, in the TNC of vehicle and NTG-treated mice. DAPI (blue) highlights cell nuclei. Scale bar = 50 μm. (F) Quantification of the mean fluorescence intensity of ZO-1 expression. NTG-treated mice display significantly reduced ZO-1 expression compared to vehicle controls (**p < 0.01)

Discussion

Our study integrates bulk transcriptomic data, single-cell transcriptomic data, and in vivo observations to identify convergent inflammatory and neurovascular features potentially shared between migraine and ischemic stroke. Rather than establishing a definitive mechanism, these findings provide a hypothesis-generating framework for understanding how migraine-related inflammatory and vascular alterations may overlap with stroke-related pathways.

Among the shared signals, EGR1 emerged as a recurrent candidate molecule across analyses. However, given that EGR1 is a well-known immediate early response gene and that the present study did not include perturbation-based validation, its role should be interpreted cautiously as a candidate marker or candidate regulatory node rather than a confirmed mechanistic driver.

The present results provide a framework for exploring how migraine-related molecular alterations may overlap with ischemic stroke-related signatures by identifying shared transcriptional features and cell-type-associated changes. First, our integrated transcriptomic analysis revealed that IL1B and EGR1 are commonly upregulated in both ischemic stroke and migraine, indicating a conserved inflammatory axis (Fig. 1B and C). Previous studies have established IL1B as a key mediator of neuroinflammation and ischemic injury, promoting leukocyte recruitment, blood–brain barrier disruption, and neuronal damage [27, 28]. Similarly, EGR1 has been implicated as an immediate early response gene that regulates stress-induced transcriptional programs in vascular and neuronal cells [29]. Our findings extend these observations by showing that these molecules are upregulated in both conditions, supporting the presence of shared inflammatory signatures rather than a confirmed shared pathogenic pathway.

Second, our Gene Ontology enrichment analysis highlighted immune response-regulating receptor signaling and interleukin-1-mediated signaling as central pathways in ischemic stroke, with IL1B and EGR1 acting as hub genes (Fig. 2). These results are consistent with prior reports emphasizing the critical role of innate immune activation in stroke progression [30–32]. However, our study provides new insight by linking these pathways to migraine-related molecular changes, suggesting that repeated inflammatory activation during migraine attacks may be associated with neurovascular changes potentially relevant to ischemic vulnerability. This aligns with emerging theories proposing that chronic neuroinflammation contributes to vascular dysfunction and stroke risk [33–35].

Third, single-cell RNA sequencing revealed that EGR1 is the only significantly shared upregulated gene across migraine-relevant cell populations, including neurons, vascular cells, and fibroblasts. Importantly, the interleukin-1 production pathway was activated in most of these cell types, indicating a multicellular inflammatory response (Fig. 3D). Previous single-cell studies have shown cell-type-specific inflammatory signatures in neurological disorders [36], whereas our findings suggest a coordinated inflammatory signature associated with EGR1 across neuronal and vascular compartments in the migraine model. Therefore, in the context of the present study, EGR1 is more appropriately interpreted as a candidate marker or candidate regulatory node associated with convergent inflammatory activation, rather than a validated mechanistic driver.

In addition, our cell–cell communication analysis identified enhanced interactions among neurons, vascular cells, and fibroblasts, particularly through ANGPTL signaling (Fig. 4). ANGPTL family proteins have been implicated in angiogenesis, vascular permeability, and inflammation [37–39]. The observed increase in ANGPTL-mediated signaling is consistent with a pro-inflammatory and pro-angiogenic communication state in the migraine model that may be relevant to altered vascular integrity. This is further supported by our protein–protein interaction network, which identified EGR1, IL1B, TLR4, and ANGPTL2 as core molecules. TLR4 signaling has been widely associated with neuroinflammation and stroke severity [40–43], reinforcing the relevance of our identified network.

Furthermore, in vivo validation using a nitroglycerin-induced migraine model demonstrated increased neuronal activation and reduced expression of the tight junction protein ZO-1 in the trigeminocervical complex (Fig. 6E and F). Disruption of tight junctions is a hallmark of blood–brain barrier dysfunction and has been strongly linked to ischemic injury [44–46]. Our findings provide in vivo support for vascular barrier-related alterations in the NTG model, which may reflect neurovascular changes potentially relevant to stroke vulnerability; however, whether these alterations increase stroke susceptibility remains to be determined.

The nitroglycerin-induced migraine model used in this study is a well-established and reproducible preclinical model of migraine-like hypersensitivity. However, current concepts of migraine pathophysiology increasingly emphasize cortical and trigeminovascular activation mediated by neuropeptides such as CGRP and PACAP. Therefore, although the NTG model provides useful biological context for the present analyses, future studies should examine whether the inflammatory and neurovascular signatures identified here can be reproduced in peptide-based migraine models with greater pathophysiological specificity.

Several limitations should be acknowledged. First, the integration of human peripheral blood bulk data and mouse migraine single-cell data enables identification of convergent signals across systems, but it also introduces biological heterogeneity and precludes direct mechanistic inference. Second, the computational approaches used here, including enrichment analysis, CellChat, and protein-interaction network analysis, are inherently associative and do not establish causality. Third, although the in vivo model supports migraine-associated neurovascular changes, we did not directly test ischemic outcomes or functionally manipulate EGR1. Therefore, future studies using genetic or pharmacologic perturbation will be necessary to determine whether EGR1 has a causal role in migraine-stroke pathobiology. Finally, only male mice were used in the present in vivo experiments. Given the marked female predominance of migraine, this limits the translational relevance of the model, and future studies should include female animals to determine whether the observed molecular and neurovascular alterations are influenced by sex.

Future studies should aim to validate these findings in human clinical samples and longitudinal cohorts to determine whether EGR1 and IL1B can serve as predictive biomarkers for stroke risk in migraine patients. In addition, targeting the identified inflammatory and neurovascular pathways may offer new therapeutic strategies. For example, pharmacological modulation of interleukin-1 signaling or ANGPTL-mediated communication could potentially reduce vascular vulnerability. Ultimately, a deeper understanding of the interplay between neuronal activity, inflammation, and vascular function will be essential for developing effective interventions to mitigate stroke risk in individuals with migraine.

In conclusion, our integrative transcriptomic and single-cell analyses, together with in vivo observations, identify shared inflammatory and neurovascular features across migraine-related and ischemic stroke-related datasets. EGR1 emerged as a recurrent candidate marker associated with these convergent changes, but its specific mechanistic role remains to be determined. Collectively, these findings provide a hypothesis-generating framework and nominate EGR1, IL1B, TLR4, and ANGPTL2 as candidate molecules for future functional validation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

WZR and YMX designed the study and established the animal models. They also performed the behavioral experiments and confirmed the hypersensitivity phenotypes. JX and BL were responsible for tissue dissociation, library construction, and sequencing quality assessment. XFZ and YPL conducted the bioinformatic analyses, processed the multi-omics data, and interpreted the biological significance of the results. WZR and BS prepared the initial draft of the manuscript and contributed substantial mechanistic interpretation. All authors participated in revising the manuscript and approved the final version for publication.

Funding

The National Natural Science Foundation of China [Grant U1904207], National Key R&D Program of China [Grant 2017YFA0105003], Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences [Grant 2020-PT310-01], and Innovative and Scientific and Technological Talents Training Project of Henan Province [Grant YXKC2021062] all provided funding for this work.

Data availability

The data generated and analyzed in this study are available as Supplemental Materials. The code used for the analyses is available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All experimental procedures involving animals were approved by the Animal Ethics Committee of the First Affiliated Hospital of Zhengzhou University. Animal suffering was minimized as far as possible by using minimally invasive behavioral testing, optimized pain management strategies, and humane euthanasia in compliance with institutional ethical guidelines.

Informed consent

Not applicable.

Competing interests

The authors declare no competing interests.