Leptomeningeal collaterals constitute an immune-responsive vascular niche in ischemic stroke

Abstract

Background:

Leptomeningeal collaterals form a critical vascular network that enlarges to support retrograde reperfusion after ischemic stroke, yet the cellular mechanisms governing their structural plasticity remain poorly defined. Here we identify an immune-responsive vascular niche and uncover a central role for bone marrow–derived monocytes in regulating collateral remodeling.

Methods:

GFP⁺ bone marrow chimeric mice and inducible monocyte-specific Ccr2-CreERT2/EphA4^f/f and Ccr2-CreERT2/EphA4^f/f/Tie2^f/f mice underwent permanent middle cerebral artery occlusion (pMCAO) to assess monocyte recruitment, collateral remodeling, cerebral blood flow, infarct volume, and functional recovery. Mechanistic studies evaluated EphA4/Tie2-PI3Kα signaling in macrophages, while serum soluble Tie2 (sTie2) levels and immune transcriptomic profiles were analyzed in patients with acute large-vessel occlusion and correlated with angiographic collateral grade.

Results:

We observed rapid recruitment of EphA4-expressing monocytes to pial collateral vessels following permanent middle cerebral artery occlusion (pMCAO). EphA4 knockout (KO) chimeric mice show a marked increase in monocyte recruitment, enhanced collateral diameters, improved cerebral blood flow, reduced infarct volume, and accelerated motor recovery. EphA4-null macrophages exhibited elevated Tie2, pAkt, and PI3Kα, a phenotype reversed by PI3Kα inhibition or soluble (s)Tie2. We further show enhanced pMCAO-induced collateral enlargement, neuroprotection, and monocyte recruitment using Ccr2-Cre^ERT2/EphA4^f/f mice, which is attenuated in Ccr2-Cre^ERT2/EphA4^f/f/Tie2^f/f double knockout mice. Notably, we find that serum sTie2 levels are elevated in human patients with large-vessel occlusion (LVO), and these levels correlate with improved digital subtraction angiography (DSA) collateral scoring and key immune-specific bulk transcriptomic changes.

Conclusions:

Together, these findings establish immune cell–intrinsic EphA4/Tie2 signaling as a key regulator of leptomeningeal collateral enlargement and reveal a therapeutic axis for augmenting perfusion after stroke.

Brief summary:

Immune cells shape collateral remodeling after stroke. Monocyte EphA4 restrains Tie2 signaling, limiting collateral enlargement. Human serum Angpt/Tie2 levels support translational relevance for collateral biomarkers.

Graphical Abstract

Ischemic stroke induces rapid recruitment of bone marrow–derived monocytes/ macrophages to leptomeningeal collateral vessels. Monocyte-intrinsic EphA4 limits arteriogenic remodeling by suppressing Tie2–PI3K–Akt signaling, promoting inflammation, collateral dysfunction, and cerebral blood flow loss. Genetic deletion of EphA4 enhances monocyte recruitment, activates Tie2–PI3K–Akt signaling, drives collateral enlargement, restores perfusion, and improves neurological outcome.

Background

The extent of tissue injury after ischemic stroke is shaped not only by the site of arterial occlusion but by the capacity of leptomeningeal collaterals to sustain retrograde reperfusion ¹. These pre-existing arteriolar connections between distal branches of the middle cerebral artery (MCA), anterior cerebral artery (ACA), and posterior cerebral artery (PCA) form a critical hemodynamic buffer, yet the mechanisms that govern their structural plasticity after stroke remain incompletely understood ^2,3. Although peripheral monocytes are established drivers of arteriogenesis in other vascular beds, where they coordinate endothelial activation, matrix remodeling, and vessel enlargement ^4–8, it is unclear whether immune cells directly regulate pial collateral remodeling after cerebral ischemia remains unclear. This question is particularly relevant given the rapid recruitment of monocytes into ischemic tissue and their established roles in inflammation, angiogenesis, and tissue repair.

Eph receptors are a family of receptor tyrosine kinases involved in vascular remodeling and inflammatory signaling ^9,10. Among these, ephrin receptor A4 (EphA4) has emerged as a negative regulator of adaptive vascular responses and macrophage-mediated inflammation after stroke ^11,12. EphA4 signaling exacerbates neuronal injury, disrupts blood–brain barrier integrity, and restrains post-stroke arteriogenesis ^13–16. However, the role of immune cell–intrinsic EphA4 signaling in regulating leptomeningeal collateral remodeling remains unknown. Defining these mechanisms is important because monocyte-derived macrophages can exert both detrimental inflammatory and reparative pro-angiogenic effects after ischemic injury ^17–22.

The Tie2–phosphatidylinositol 3-kinase (PI3K)/Akt signaling axis is a central regulator of vascular stability and endothelial remodeling ¹⁶, and macrophage reparative functions ^23,24. Angiopoietin-1 (Angpt-1) promotes Tie2-dependent PI3K/Akt signaling associated with vascular stabilization, whereas Angiopoietin-2 (Angpt-2) destabilizes the vascular niche and modulates immune–vascular interactions ^25–28. In ischemic stroke, Tie2-expressing monocytes (TEMs), a pro-angiogenic monocyte subset, are rapidly mobilized and associated with improved vascular remodeling and functional recovery ^29–32. Elevated Tie2 expression on circulating CD14+ monocytes correlates with favorable clinical outcomes, highlighting the translational relevance of this pathway ^32,33. Despite these findings, the upstream mechanisms regulating Tie2 signaling in monocytes during stroke remain poorly defined, including whether EphA4 constrains TEM-mediated reparative responses and collateral vessel expansion.

Here, we define the role of the immune cell–intrinsic EphA4 in regulating leptomeningeal collateral remodeling after ischemic stroke. Using GFP+ bone marrow chimeric mice and Ccr2-CreERT2–driven single- and double-knockout models, we demonstrate monocyte-specific EphA4/Tie2 regulation of immune cell recruitment and collateral expansion, supported by mechanistic studies in bone marrow-derived macrophages. We further evaluate whole-blood transcriptomics and Angpt/Tie2 protein levels in human stroke serum in relation to DSA collateral scores to establish their translational relevance as biomarkers of collateral status and large vessel occlusion stroke.

Data availability.

Data is available upon reasonable request to the corresponding author.

Methods

Animals.

All mice were bred and group-housed at 3–5 mice/cage in an AAALAC-accredited, specific pathogen-free facility under a controlled 12-h light-dark cycle, with a standard irradiated chow diet, and water was provided ad libitum. Donor mice, EphA4^+/+/ROSA^GFP, and EphA4^f/f/ROSA^GFP were maintained on the CD1 strain (Charles River Laboratories) and backcrossed for at least 10 generations. These lines were crossbred to generate experimental mice (via ROSA26-loxP-STOP-loxP-GFP insertion). Adult mice were used at 8–10 weeks. For Ccr2-driven monocyte-specific EphA4 and Tie2 genetic inducible (i) deletion experiments, Ccr2-Cre^ERT2gfp (iWT), Ccr2-Cre^ERT2gfp/EphA4^f/f (iEphA4 KO), and Ccr2-Cre^ERT2gfp/EphA4^f/f/Tie2^f/f (idKO) on CD1 background. Tamoxifen was administered by intraperitoneal injection at 50 mg/kg body weight once daily for 5 consecutive days, beginning at 8 weeks of age. Two weeks after the last tamoxifen injection, genotyping PCR was performed as previously described ^34,35.

According to ARRIVE and ARRIVE 2.0 guidelines ^36,37, (Table S1. Major Resource Table) all animals were assigned anonymized codes at the time of surgery, and investigators conducting surgeries, behavioral testing, imaging, and data analysis were blinded to group allocation. Sample sizes were determined based on prior studies from our laboratory and published literature demonstrating sufficient sensitivity to detect biologically meaningful differences in infarct volume, vascular remodeling, and behavioral outcomes. No formal a priori power calculation was performed. Male and female CD1 mice were initially included to assess potential sex differences in pial collateral enlargement following permanent MCA occlusion (pMCAO) (Supplemental Figure 1). No significant sex differences were detected in these preliminary analyses (n=8/sex). In accordance with NIH guidance permitting the use of a single sex when scientifically justified, subsequent experiments were conducted using male mice to reduce animal use and control for variability. All procedures complied with institutional IACUC protocols and followed national regulations for the ethical care and use of laboratory animals.

A total of 479 mice were enrolled across all study protocols. Of these, 17 mice died or were euthanized for humane reasons before reaching their planned experimental endpoints, yielding a mortality rate of 3.55%. The remaining 462 mice successfully completed their respective protocols; all were included in the final analyses.

Inclusion and Exclusion Criteria. Animals were included if they successfully completed the surgical procedure and survived to the designated experimental endpoint. Animals were excluded a priori in cases of surgical mortality, technical complications during surgery, or inability to complete outcome assessments. Animals were randomly assigned to experimental groups prior to surgery and treatment allocation.

Surgical Procedures.

Ischemic stroke was induced in adult mice using pMCAO as previously described ³⁸. Physiological Monitoring: Animals were monitored throughout surgical procedures to maintain anesthesia, body temperature, and general physiological status. Core body temperature was maintained at 37 ± 0.5°C using a homeothermic blanket system (Harvard Apparatus). Respiration rate, body weight, and postoperative recovery were monitored throughout the study. No significant differences in physiological parameters were observed between experimental groups. Bone marrow chimeric mice were generated as previously described ^39,40. Cerebral Blood Flow Measurement was assessed using a laser speckle contrast imaging system (RFLSI III Laser Speckle Imaging System, RWD Life Science, Dover, DE, USA) as previously described ⁴¹.

Vessel Painting, Collateral and Infarct Quantification.

Vessel painting was performed as previously described ^42,43. Nikon C2 confocal microscope (Tokyo, Japan) images of the brains were imported into Fiji-ImageJ (NIH) for quantification of the number and diameter of collaterals, as previously described ³⁸. The number of GFP+, Iba-1+, Ly6G+, and PCNA+ cells was quantified from maximum z-projected images ³⁵. Infarct volume was determined utilizing the Cavelieri Estimator feature of the non-biased StereoInvestigator software (MicroBrightField, Williston, VT, USA), following our established protocols ⁴⁴.

Immunostaining of Cortical Whole Mounts and Western Blots.

Cortical whole mounts were prepared following our established protocols ³⁸. Briefly, whole mounts underwent blocking in 2% fish gelatin (Sigma Aldrich, St. Louis, MO) with 0.2% Triton-X, then incubated with primary rabbit anti-EphA4 (Proteintech 21875–1-AP), abbit anti-PCNA antibody (1:300; Cell Signaling, Danvers, MA, USA; Cat# 13110), rabbit anti-Iba1 (Wako), and rat anti-Ly6G (Biolegend) at 1:200 in the blocking solution. Cell lysates were harvested in RIPA buffer (ThermoFisher) and centrifuged at 14,000 x g for 20min at 4°C as previously described ^45,46. Blot subjected to rabbit anti-Tie2 (R&D Systems), rabbit anti-AKT, rabbit anti-phospho-AKT, and mouse anti-β-actin (Cell Signaling), then incubated with HRP-conjugated secondary antibodies (Jackson ImmunoResearch). Protein bands were visualized using chemiluminescent detection (Thermo Scientific Pierce) on the Bio-Rad ChemiDoc system and quantified using Fiji-ImageJ (NIH).

Flow Cytometry Analysis.

Whole blood was collected via cardiac puncture, subjected to ACK lysing buffer (ThermoFisher) then 10⁶ cells/100 μl in 1x PBS (Fisher Scientific; Waltham, MA) were stained with Zombie aqua (BioLegend, San Diego, CA) followed by incubation with 2% Fc blocker (1% FBS/2mM EDTA in PBS), then incubated with appropriate conjugated or unconjugated primary monoclonal antibodies (BioLegend, PE-CD11b; BV421-Ly6C; BV650-CCR2; PE/Cy7-Ly6G; APC-CD45; PerCP/Cy5.5-CX3CR1; PerCP/Cy7-CD49b; PE-CD3; EphA4 (Proteintech, Rosemont, IL)). Control groups were incubated with isotype antibodies, followed by fluorochrome-conjugated secondary antibodies as necessary. Analysis was performed using a BD FACSAria^™ II Flow Cytometer, with each measurement containing 2 × 10⁵ cells. Data analysis was conducted using FlowJo v10 (BD, Ashland, Oregon, USA).

Bone marrow-derived macrophage culture.

BM was obtained from the femurs of 8- to 10-week-old mice and cultured at a concentration of 1 × 10^6 cells/ml in complete DMEM medium supplemented with fetal bovine serum (10%), L-glutamine (2 mM), penicillin/streptomycin (1%), and M-CSF (10 ng/ml) as previously described ³⁴. On day 5, BMDMs were either supplemented with fresh DMEM media or pre-treated with 50 nM of the PI3K-alpha inhibitor Alpelisib (BYL-719, Adooq Bioscience), or 20 μg/ml of recombinant mouse sTie2-Fc or Fc-control (Sino Biological, Wayne, PA, USA) for 4 hours. Cells were washed with cold sterile PBS prior to RNA isolation and subsequent analyses.

Patient Cohort and Sample Collection.

We conducted a prospective pilot study at Carilion Roanoke Memorial Hospital (Carilion Clinic), Roanoke, VA, USA, enrolling 15 patients with acute ischemic stroke due to large-vessel occlusion who underwent mechanical thrombectomy between November 2021 and July 2023. The study was approved by the Carilion Clinic Institutional Review Board (IRB #19–386), and written informed consent was obtained from legally authorized representatives. Patients were aged 29–90 years (median 65 years). Inclusion criteria included confirmed ischemic stroke by computed tomography or magnetic resonance imaging, age ≥18 years, and symptom onset within 24 hours of admission. Patients were stratified by DSA collateral scores (1, 2, or 3) according to the American Society of Interventional and Therapeutic Neuroradiology/Society of Interventional Radiology criteria. Exclusion criteria comprised hemorrhagic stroke, active infection, malignancy, or immunosuppressive therapy. Whole blood was collected via venipuncture immediately before mechanical thrombectomy (MT-pre-intervention), immediately after thrombectomy (post-intervention), and 24 hours post-procedure during hospitalization. Samples were drawn into serum separator tubes, allowed to clot at room temperature for 30 minutes, and centrifuged at 1,500g for 10 minutes; serum was aliquoted and stored at −80°C until analysis. For RNA-based analyses, samples were collected into PAXgene Blood RNA Tubes (PreAnalytiX, BD) for RNA stabilization and stored at −80°C until extraction. Age-, sex-, and race-matched control serum samples were obtained from healthy volunteers without vascular disease (Discovery Life Sciences, Huntsville, AL). Demographic data (age, sex, race) and stroke severity, assessed by the National Institutes of Health Stroke Scale at admission, were recorded.

Enzyme-linked immunosorbent assays (ELISA) of human serum.

Serum concentrations of Angpt-1, Angpt-2, and sTie2 were quantified using commercially available ELISA kits (EK0519, EK1296, EK0938, Boster Bio, Pleasanton, CA) according to the manufacturer’s instructions. All samples were assayed in duplicate, with absorbance measured at 450 nm on a microplate reader (BioTek). Standard curves were generated using four-parameter logistic regression, and concentrations were interpolated in ng/mL.

RAS Pulldown Activation Assay and qPCR.

The concentration of BMDM protein lysate was quantified using the BCA Protein Assay Kit (Thermo Fisher Scientific), and 500 μg of protein was used for the RAS-binding domain (RBD) assay following the manufacturer’s protocol for the Ras Pull-down Activation Assay Biochem Kit (BK008, Cytoskeleton, Inc.). Briefly, 30 μL of RAF-RBD beads and protein were incubated, then centrifuged and washed with 500 μL of wash buffer, followed by centrifugation; the pellet was then resuspended in 20 μL of 2× Laemmli sample buffer and boiled at 95°C for 5 minutes. Samples were run on a Western blot as described previously⁴⁶. Controls included GDP-incubated samples (negative control), GTPγS (positive control), and His-tagged RAS protein (control protein), with total protein input used as the loading control. RNA Isolation and Quantitative Real-time PCR. Total RNA was extracted from cultured BMDMs according to the manufacturer’s instructions using TRIzol reagent (Life Technologies), and qPCR was performed as previously reported.

Behavioral Testing.

As performed previously ^47,48. Gross motor function was assessed using the rotarod (Economex; Columbus Instruments), with mice pre-trained for 4 days before surgery and tested at baseline and days 3, 7, and 14 post-pMCAO. Neurological deficits were evaluated using the modified neurological severity score (mNSS; 0–14 scale). Sensorimotor function was assessed using the adhesive removal test, with asymmetry scores calculated from contralateral and ipsilateral forepaw removal times. Fine motor coordination was evaluated using the pole test by measuring orientation and descent times. Mice were habituated/trained prior to testing, and all assessments were performed at baseline and days 3, 7, and 14 post-injury.

Library Preparation and Bulk RNA Sequencing.

Ribosomal RNA-depleted libraries were prepared from 1 μg total RNA and sequenced (2 × 150 bp) on an Illumina NovaSeq platform (MedGenome, Foster City, CA, USA). Reads were quality-checked, aligned to the human reference genome (GRCh38) using STAR, and quantified at the gene level. Differential expression analysis was performed using DESeq2, with significance defined as |log₂ fold change| ≥ 0.6 and adjusted p ≤ 0.05. Gene Ontology enrichment analysis was conducted using clusterProfiler, and enriched pathways were visualized using dot plots and chord diagrams. Bulk RNA-seq deconvolution was performed using CIBERSORTx with the LM22 leukocyte signature matrix and recommended parameters ^49,50 and inferred cell-type-specific expression profiles were compared between DSA1 and DSA3 groups. Significant genes were visualized using custom Python-based analyses.

Statistical Analysis.

Data were analyzed using GraphPad Prism 10, and flow cytometry data were processed in FlowJo 10.8.1 using identical blinded gating strategies with fluorescence-minus-one controls. Sample size was determined by power analysis for two-way ANOVA (power = 0.83, α = 0.05). Comparisons between two groups were performed using unpaired two-tailed Student’s t-tests, while multiple groups were analyzed by one- or two-way ANOVA with appropriate post hoc tests. Data are presented as mean ± SEM with individual biological replicates shown. Statistical significance was defined as P<0.05.

Availability of data.

All datasets supporting the conclusions and generated or analyzed during this study are included in this published article and its supplementary information files. Raw data supporting the findings of this study are available from the corresponding author upon reasonable request.

Study approval.

All procedures adhered to the NIH standards outlined in the guideline for Care and Use of Laboratory Animals and were approved by Virginia Tech (IACUC: 24–032; 24–033). Human pilot studies were approved by IRB 19–386.

Results

Stroke induces upregulation of EphA4 in circulating and on collateral-recruited immune cells.

Peripheral immune cell recruitment and activation are key mediators of the acute inflammatory response after ischemic injury ⁵¹. We analyzed whole-blood immune populations by flow cytometry to assess EphA4 expression and acute immune alterations following stroke. Myeloid leukocytes (CD45⁺/CD11b⁺), agranulocytes (CD45⁺/CD11b⁺/Ly6G⁻), granulocytes (CD45⁺/CD11b⁺/Ly6G⁺), and Ly6C⁺ monocyte subsets were quantified in naïve, sham, and pMCAO-injured CD1 mice (Fig. 1A–1B). At 24 hours post-pMCAO, both classical Ly6Chigh and intermediate Ly6C+/CCR2+/CX3CR1+ monocytes were significantly reduced, consistent with rapid monocyte mobilization after stroke (Fig. 1B).

Figure 1. EphA4 expression and immune cell dynamics in whole blood following ischemic stroke.

(A) Flow cytometry gating strategy for identifying single, live, CD45+/CD11b+ cells, and subset into Ly6G⁻/Ly6C^high and Ly6G⁻/Ly6C^low monocytes, and Ly6G⁺ neutrophils expressing EphA4 in whole blood from sham- or pMCAO-injured WT mice (n = 5–14 mice/group). (B) Percentage (%) of agranulocytes (CD45⁺/CD11b⁺/Ly6G⁻/Ly6C^bigh and CD45⁺/CD11b⁺/Ly6G⁻/Ly6C^low), granulocytes (CD45⁺/CD11b⁺/Ly6G⁺), T cells (CD45⁺/CD3⁺/CD49b⁻), and NK cells (CD45⁺/CD3⁻/CD49b⁺) at 24hrs post-pMCAO (n = 5–14 mice/group). (C) Representative confocal images of cortical whole-mounts immunostained for EphA4 in +WT^BMCs at 24hrs post-pMCAO. (D) Quantification of EphA4-positive, GFP+ immune cells within the leptomeningeal collateral niche at 24 h post-pMCAO using IHC. Total numbers are shown as associated with the ipsilateral versus contralateral pial collaterals of +WT^BMCs mice (n = 5 mice/group). (E) Flow cytometry percentage of EphA4⁺ cells within granulocytes (CD45⁺/CD11b⁺/Ly6G⁺) and agranulocytes (CD45⁺/CD11b⁺/Ly6G⁻/Ly6C, Cx3cr1 and Ccr2-expressing), as well as T cell (CD45⁺/CD11b⁺/CD3⁺/CD49^+/−) populations in whole blood from sham vs. pMCAO-injured WT mice (n = 5–14 mice/group). Data were analyzed by two-way ANOVA followed by Tukey’s post-hoc. *P<0.05, **P<0.01, ***P<0.001, ****P<0.00001.

Flow cytometry revealed increased EphA4 expression on agranulocyte and granulocyte populations in sham and pMCAO-injured mice compared with naïve controls, including classical, non-classical, and intermediate monocyte subsets (Fig. 1E). EphA4 was also expressed on circulating NK cells (21.8%) and T cells (8.18%), although these populations were unchanged at 24 hours post-pMCAO (Fig. 1E). Consistent with these findings, immunohistochemical analysis of vessel-painted cortical whole mounts from GFP⁺ bone marrow chimeric mice demonstrated robust recruitment of GFP⁺/EphA4⁺ immune cells to the ipsilateral leptomeningeal collateral niche at 24 hours post-pMCAO, with EphA4 also localized along collateral vessels (Fig. 1C–1D). Quantification confirmed significant enrichment of GFP⁺ and EphA4⁺ immune cells within the ipsilateral collateral niche compared with the contralateral hemisphere (Fig. 1D). Together, these findings demonstrate that EphA4-expressing peripheral immune cells are preferentially recruited to the pial collateral niche following ischemic stroke.

Immune cell-specific EphA4 deletion augments leptomeningeal collateral enlargement following ischemia

Given the enrichment of EphA4-expressing immune cells within the leptomeningeal collateral niche after pMCAO, we next investigated whether EphA4–intrinsic signaling regulates collateral vessel remodeling following ischemic injury. First, sex differences in collateral remodeling were assessed after ischemic injury; male and female mice underwent pMCAO, and collateral vessel diameters were quantified at 24 hours post-injury. No significant sex differences were observed in collateral number or diameter across MCA-ACA, MCA-PCA, or ACA-PCA territories in either hemisphere (Supplementary Fig. 1A–1E), indicating that collateral enlargement is not sexually dimorphic in this model. Next, GFP BM chimeric mice in which EphA4 BMC deficiency (+KO^BMCs) was compared with wild type (+WT^BMCs) (Fig. 2Ai–iii), we evaluated collateral size using vessel-painted at 24hrs post-pMCAO (Fig. 2B). Under naïve conditions, loss of EphA4 in bone marrow–derived immune cells did not alter collateral vessel size or number across MCA–ACA, MCA–PCA, or ACA–PCA territories (Supplementary Fig. 2A–2E). Following pMCAO, both +WTBMCs and +KOBMCs mice exhibited ipsilateral collateral enlargement relative to the contralateral hemisphere (Fig. 2Ci–vi). However, immune cell-specific EphA4 deletion significantly increased MCA–ACA collateral diameter compared with +WTBMCs controls (37.11 ± 0.61 μm vs. 32.49 ± 0.56 μm; Fig. 2D), whereas MCA–PCA collateral diameters were unchanged between groups (Fig. 2E). Collateral size distribution analysis further demonstrated a greater proportion of enlarged ipsilateral vessels in +KOBMCs mice, including increased numbers of vessels >31 μm and >40 μm in diameter (Fig. 2F). Together, these findings identify peripheral immune cell EphA4 as a negative regulator of pial collateral enlargement following ischemic stroke.

Figure 2. EphA4 deletion in adoptively transferred bone marrow cells promotes collateral vessel enlargement 24hrs after pMCAO.

(A i) Schematic representation of experimental timeline. (A ii) GFP BM chimeric +WT^BMCs and +KO^BMCs mice were generated by irradiation and reconstitution with either wild-type or EphA4 knockout BMCs. (A iii) qPCR confirmation of EphA4 deletion in whole blood cells (WBCs) and BMCs from +KO^BMCs mice relative to +WT^BMCs controls (n = 3 mice/group). (B) Representative tiled 4X confocal images of vessel-painted +WT^BMCs brains 24hrs post-pMCAO. (C i-vi) Representative tiled 4X confocal images of the MCA-ACA pial collateral niche, and GFP+ peripheral immune cell infiltration in +WT^BMCs and (C v) +KO^BMCs mice (C vi). Scale bars,1 mm (C i-iv) and 500 μm (C v and C vi). (D, E) Quantification of pial collateral diameter in MCA–ACA and MCA–PCA anastomoses post-pMCAO. +KO^BMCs mice exhibit significantly larger ipsilateral MCA–ACA collaterals than +WT^BMCs controls. (F) Analysis of MCA-ACA collateral vessel size distribution (n=11–14 mice/group). *Denotes collateral vessels. Contra, contralateral; Ipsi, ipsilateral. Data in (D, E) were analyzed by two-way ANOVA with Tukey’s post hoc test; data in (F) by two-way ANOVA with Bonferroni’s post hoc test. *P<0.05, ****P<0.0001 compared to corresponding +WT^BMCs mice. Scale bars = 1mm.

Immune cell–specific EphA4 deletion enhances neuroprotection and functional recovery after ischemic stroke

Pial collateral vessels undergo outward growth and remodeling, promoting restoration of CBF and functional recovery after ischemic stroke ⁵². To determine whether immune cell-specific EphA4 deletion improves post-stroke perfusion, CBF was measured by laser speckle contrast imaging before and at 10 minutes, 1 day, and 3 days after pMCAO (Fig. 3Ai). Although no differences were observed between +WTBMCs and +KOBMCs mice at 10 minutes post-pMCAO, +KOBMCs mice exhibited significantly greater recovery of ipsilateral CBF at 1 and 3 days post-injury (Fig. 3Aii). Enhanced perfusion in +KOBMCs mice was associated with reduced infarct volume at 24 hours post-pMCAO compared with +WTBMCs controls (17.44 ± 1.16 mm3 vs. 24.03 ± 1.62 mm3; Fig. 3B i–ii). Sham-operated +WTBMCs and +KOBMCs mice did not differ across behavioral assessments, indicating no baseline effect of bone marrow genotype on neurological function (Fig. 3C–3D). Following pMCAO, however, +KOBMCs mice demonstrated significantly improved motor coordination, neurological recovery, sensorimotor integration, and motor planning across rotarod, mNSS, adhesive removal, and pole tests compared with +WTBMCs mice (Fig. 3C–3D). Together, these findings demonstrate that immune cell-derived EphA4 exacerbates ischemic injury by limiting post-stroke perfusion and recovery, whereas EphA4 deletion promotes collateral-associated CBF restoration, neuroprotection, and sustained functional improvement after stroke.

Figure 3. Enhanced cerebral perfusion, behavioral outcomes, and neuroprotection in +KO^BMCs mice following pMCAO.

(A i) Representative laser speckle contrast images of CBF in +WT^BMCs and +KO^BMCs mice. (A ii) Quantification of ipsilateral CBF before (pre), immediately after (post), and 1–3 days following pMCAO (n = 6–11 mice/group). (B i) Representative serial Nissl-stained brain sections at 24hrs post-pMCAO from +WT^BMCs and +KO^BMCs mice. (B ii) Infarct volume quantification reveals significant neuroprotection in +KO^BMCs mice. n = 9–14 per group. (C i) Rotarod assessment of +WT^BMCs and +KO^BMCs mice. +KO^BMCs mice performed significantly better than +WT^BMCs mice at 3 days after pMCAO. (C ii, C iii) Functional recovery evaluated using the modified neurological severity score (mNSS) and adhesive removal test from baseline and 3 to 14 days post-pMCAO shows improved neurological outcomes in +KO^BMCs mice compared with +WT^BMCs mice (n = 5–14 mice/group). (D i, ii) Pole test performance, including (D i) time to turn completely and (D ii) total time to descend the pole, assessed at baseline and 3, 7, and 14 days post-pMCAO. +KO^BMCs mice showed significantly improved performance compared with +WT^BMCs mice (n = 5–14 mice/group). Data in panels A ii, C, and D were analyzed by two-way ANOVA with Sidak's. * = +WT^BMCs pMCAO vs +KO^BMCs pMCAO mice. # = +WT^BMCs sham vs +WT^BMCs pMCAO mice. $ = +KO^BMCs sham vs +KO^BMCs pMCAO mice. *p<0.05, ^{**,##,$ $}p<0.01, ^***,###p<0.001, ^{****,####,$ $ $ $}p<0.0001. White rectangles in (A) indicate standardized regions of interest (ROIs) used for CBF quantification.

Immune-specific EphA4 restricts monocyte/macrophage recruitment to the pial collateral niche after pMCAO.

Recruitment of bone marrow–derived immune cells to the pial collateral niche after ischemic stroke has not previously been examined. Using GFP+ bone marrow chimeric mice, we observed significant accumulation of GFP+ immune cells along ipsilateral MCA–ACA collateral vessels at 24 hours post-pMCAO, with +KOBMCs mice exhibiting markedly greater recruitment compared with +WTBMCs controls and contralateral hemispheres (Fig. 4A i–vi). Although total GFP+ cell number did not correlate with collateral diameter, immunolabeling for the monocyte/macrophage marker Iba1 revealed significantly increased GFP+/Iba1+ cell recruitment along ipsilateral collateral vessels in +KOBMCs mice (6.72 ± 0.47 vs. 4.67 ± 0.63 cells/100 μm in +WTBMCs; Fig. 4B–F). GFP+/Iba1+ cells represented the majority of recruited GFP+ cells in both groups, with greater enrichment in +KOBMCs mice (81.4% vs. 69.8%; Fig. 4D). Importantly, collateral diameter positively correlated with GFP+/Iba1+ cell number in +KOBMCs mice, but not +WTBMCs controls (Fig. 4Fii), suggesting that EphA4-deficient monocyte/macrophage recruitment contributes to enhanced collateral enlargement. In contrast, Ly6G+ neutrophils were only sparsely associated with collateral vessels despite their presence within the meninges after pMCAO (Supplementary Fig. 3A–3F), and neutrophil number did not correlate with collateral diameter. Together, these findings identify monocyte/macrophages, rather than neutrophils, as the predominant immune population associated with the pial collateral niche after ischemic stroke and suggest that EphA4 limits their recruitment during collateral remodeling.

Figure 4. EphA4 knockout in bone marrow chimeras promotes GFP⁺ immune cell recruitment and pial collateral remodeling at 24hrs post-pMCAO.

(A i-iv) Representative maximum-intensity Z-projection confocal images of GFP⁺ cells in the contralateral (A i, ii) and ipsilateral (A iii, iv) hemispheres of BM chimeric mice at 24 hours post-pMCAO. Yellow outlines highlight pial collateral vessels. (A v) Quantification reveals a significant increase in GFP+ cells along middle cerebral artery-anterior cerebral artery (MCA-ACA) collateral vessels in +KO^BMCs versus +WT^BMCs mice (n = 6–11 mice/group). (A vi) Pearson correlation analysis shows a positive relationship between collateral vessel diameter and GFP+ cell number in the collateral niche, with P and R² values indicated (each dot represents one mouse). (B, C, E) Confocal images of vessel-painted (VP) pial collaterals, GFP, and Iba1+ monocytes/macrophages (arrowheads) in ipsilateral hemispheres of +WT^BMCs (B) and +KO^BMCs (C) mice at 24 hours post-pMCAO. (F i) Quantified data confirm higher GFP+/Iba1+ cell counts on pial collateral walls in +KO^BMCs mice. (D) The percentage of GFP+/Iba1+ cells is 81.4% in +KO^BMCs and 69.8% in +WT^BMCs mice. (Fii) Correlation analysis demonstrates a significant association between collateral diameter and GFP+/Iba1+ cell number in +KO^BMCs mice (R² = 0.4325, P < 0.05; each dot represents one mouse). Scale bars, 100 μm. Data in (A v, F i) analyzed by two-way ANOVA with Šídák’s multiple comparisons test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 7–13 mice per group.

EphA4-deficient bone marrow chimeras exhibit increased cell proliferation within the pial collateral niche

Cell proliferation contributes to early arteriogenic remodeling after ischemia ⁵³. Consistent with this, PCNA immunolabeling of vessel-painted cortical whole mounts at 24 hours post-pMCAO revealed significantly increased numbers of PCNA+ cells within ipsilateral collateral vessels of +KOBMCs mice compared with +WTBMCs and contralateral vessels (Fig. 5A–5F). +KOBMCs mice also exhibited greater recruitment of GFP+/PCNA+ immune cells to collateral vessels (Fig. 5E), consistent with previous findings that immune cell proliferation is enriched around enlarging pial collaterals after stroke. ^16,38. Together, these findings identify immune cell-specific EphA4 as a negative regulator of proliferative remodeling within the pial collateral niche after pMCAO.

Figure 5. Enhanced cell proliferation in the MCA-ACA pial collateral niche in +KO^BMCs mice.

(A-D) Vessel-painted brains were immuno-labeled with anti-PCNA and imaged by high magnification confocal imaging. Maximum z-projection analysis of the entire z-stacked vessel was used to evaluate cell division for evidence of ipsilateral pial collateral remodeling at 24hrs post-pMCAO in +KO^BMCs mice. MCA-ACA collateral vessels from +KO^BMCs mice show an increased number of PCNA⁺/GFP+ cells (yellow arrow heads) associated with the collateral vessel wall and PCNA+ staining within the collateral wall itself (white arrows) compared with +WT^BMCs mice. (E, F) Quantified data showing the number of GFP⁺/PCNA⁺ cells (E) and PCNA⁺ collateral vessels (F) were both increased in +KO^BMCs mice at 1 day after pMCAO (n = 5 mice/group). Scale bars=50μm. Two-way ANOVA with Šídák's multiple comparisons test. *P<0.05, **P<0.01, ****P<0.0001.

EphA4 suppresses PI3Kα (PIK3CA) downstream of Tie2 signaling in bone marrow–derived macrophages (BMDMs)

Angpt/Tie2 signaling has recently emerged as a key player in monocyte/macrophage phenotypic characteristics ^54–56 and regulates downstream signaling through the PI3K/Akt pathway ⁵⁷. To determine whether EphA4 modulates this pathway, WT and EphA4-KO bone marrow–derived macrophages (BMDMs) were treated with soluble Tie2-Fc (sTie2-Fc) or Fc-control. EphA4-KO BMDMs exhibited significantly increased expression of PI3K catalytic subunits PIK3CA and PIK3CB, as well as regulatory subunits PIK3R1 and PIK3R2, whereas PIK3CD and PIK3R3 were unchanged (Fig. 6A–6B). Notably, sTie2-Fc treatment attenuated PIK3CA expression in EphA4-KO BMDMs, suggesting that EphA4 suppresses Tie2-dependent PI3Kα signaling. Consistent with this, EphA4-KO BMDMs displayed increased Tie2 and phosphorylated Akt levels under vehicle-treated conditions, both of which were reversed by the PI3K p110α inhibitor BYL-719 (Fig. 6C–6D), indicating that p110α is a major catalytic isoform downstream of Tie2 in macrophages. Although basal Ras-GTP activity was unchanged between groups, PI3Kα inhibition increased Ras-GTP levels in EphA4-KO BMDMs relative to WT cells (Fig. 6E), suggesting compensatory Ras pathway activation in the absence of EphA4. Together, these findings demonstrate that EphA4 negatively regulates Tie2-mediated PI3Kα/Akt signaling in macrophages (Fig. 6F).

Fig. 6. Loss of EphA4 in BMDMs enhances Tie2-mediated PI3K-Akt signaling.

(A, B) Relative mRNA expression levels of PI3K catalytic subunits (PIK3CA, PIK3CB, PIK3CD) and regulatory subunits (PIK3R1, PIK3R2, PIK3R3) in WT and EPhA4-null BMDMs treated with 20 μg/ml soluble Tie2-Fc (sTie2-Fc) or Fc-control for 5hrs. EphA4 KO BMDMs show increased PIK3CA expression compared to WT BMDMs. Blockade of Tie2 signaling with sTie2-Fc reduced PIK3CA expression. PIK3CB expression is increased in KO BMDMs treated with Fc-control. PIK3R1 and R2 expression increased in KO BMDMs under Fc-control conditions. No changes were observed in PIK3CD or PIK3R3. (C) Western blot analysis of BMDMs treated with or without BYL-719 (PIK3CA inhibitor, 24hrs). (D) Densitometric analysis revealed a significant increase in Tie2 expression and the p-Akt/total Akt ratio in KO BMDMs compared to WT BMDMs. These increases were diminished following BYL-719 treatment. (E) Western blot showing no difference in Ras-GTP levels in WT and KO BMDMs under basal conditions but increased more in KO after 4hrs of treatment with BYL-719. (F) Proposed schematic pathway illustrating EphA4 deficiency potentiates Tie2 receptor-mediated PI3K-Akt signaling. Statistical analysis was performed using two-way ANOVA followed by Šídák's post-hoc (*P<0.05, **P<0.01, ***P<0.001).

Loss of Ccr2 monocyte-specific EphA4 augments GFP recruitment, neuroprotection, and collateral enlargement after pMCAO

To define monocyte-specific contributions of EphA4 and Tie2 signaling, we generated tamoxifen-inducible Ccr2-CreERT2gfp/EphA4f/f (iEphA4 KO), Ccr2-CreERT2gfp/EphA4f/f/Tie2f/f double knockout (idKO), and Ccr2-CreERT2gfp wild-type (iWT) mice (Fig. 7A). Efficient deletion of EphA4 and/or Tie2 in circulating monocytes was confirmed by qPCR (Supplementary Fig. 4A–4B). Following pMCAO, iEphA4 KO mice exhibited significantly increased ipsilateral collateral diameters across MCA–ACA territories compared with iWT mice, whereas this effect was attenuated in idKO mice (Fig. 7B, 7E ii–iii). No differences were observed in contralateral vessels. Consistent with enhanced collateral remodeling, infarct volume at 24 hours post-pMCAO was markedly reduced in iEphA4 KO mice relative to iWT controls, with partial loss of protection in idKO mice (Fig. 7C).

Fig. 7. Ccr2-specific loss of EphA4 enhances monocyte recruitment, collateral enlargement, and prevents cortical damage, which is attenuated in Tie2/EphA4 double knockout mice at 24hrs post-pMCAO.

(A) Schematic representation of experimental timeline. (B) Tiled 4x magnification confocal images of the vessel-painted brain showing collateral vessels (white asterisk) of iWT, iEphA4 KO, and idKO mice. (B iv) Quantification of MCA-ACA collateral diameter shows increased diameters in iEphA4 KO compared to iWT, which are attenuated in iEphA4/Tie2 double KO mice (n = 4–14 mice/group). (C i) Infarct volume (mm³) was reduced in iEphA4 KO vs. iWT and idKO (n = 5–8 mice/group). (C ii-iv) Representative Nissl-stained coronal sections showing infarct damage (doted line). (D) Higher 20x magnification images of ipsilateral collaterals (red) with recruited VP/Ccr2-GFP+ monocytes (green; Iba1+, grayscale insects) in iWT (D i), iEphA4 KO (D ii), and idKO (D iii). (E i) Ccr2-GFP+ cells per collateral were markedly increased in ipsilateral iEphA4 KO compared to idKO and iWT (n = 3–13 mice/group). Collateral diameters in ACA–PCA (E i) and MCA–PCA (E ii) collateral beds. Scale bars: 2 mm (B), 50 μm (D). Data: mean ± SEM; two-way ANOVA with Šídák’s test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

High-magnification GFP imaging further demonstrated robust recruitment of Ccr2+ monocytes to the pial collateral niche in iEphA4 KO mice, which was significantly reduced in idKO mice (Fig. 7D–E i). Together, these findings demonstrate that monocyte-specific EphA4 deletion enhances monocyte recruitment, collateral enlargement, and neuroprotection after ischemic stroke, in part through Tie2-dependent mechanisms.

Serum levels of Angpt-1, Angpt-2, and sTie2 are altered in acute ischemic Stroke and associated with collateral vessel status

Recent findings demonstrate an increase in Tie2-expressing monocyte/macrophages in the iliac crest BM of patients within 2 days following acute ischemic stroke (AIS) ³². Further, our prior pre-clinical study showed that the Tie2 agonist, Vasculotide, promotes collateral remodeling, CBF, and improves outcomes in a pMCAO model ¹⁶. To determine whether the Angpt/Tie2 signaling axis is similarly altered in human stroke, we performed an exploratory pilot study in patients with large vessel occlusion (LVO) undergoing mechanical thrombectomy. Serum Angpt-1, Angpt-2, and soluble Tie2 (sTie2) levels were measured by ELISA in 15 patients and sex-matched healthy controls (Supplementary Fig. 5) at pre-thrombectomy, post-thrombectomy, and 24 hours after stroke onset. Collateral status was graded by DSA using the ASITN/SIR scale (1 = poor; 3 = good collateral flow). Compared with controls, stroke patients exhibited significantly elevated serum sTie2 and Angpt-2 with reduced Angpt-1 levels at the pre-thrombectomy time point, indicating acute dysregulation of the Angpt/Tie2 axis during ischemic stroke (Fig. 8A–C i). Biomarker levels remained stable across the first 24 hours after stroke (Fig. 8A–C ii). Stratification by collateral grade revealed significantly higher sTie2 levels in patients with good collaterals (DSA 3) compared with poor collaterals (DSA 1), whereas Angpt-1 and Angpt-2 levels did not differ between groups (Fig. 8A–C iii). Representative DSA images demonstrated minimal leptomeningeal filling in DSA 1 patients and robust retrograde collateral perfusion in DSA 3 patients (Fig. 8D i–ii).

Figure 8. Serum levels of Angpt-1, Angpt-2, and Tie2 in patients following ischemic stroke.

(A i, B i, C i) Serum concentrations of Tie2, Angpt-2, and Angpt-1 in healthy controls versus pre-mechanical thrombectomy (pre-MT) large vessel occlusion (LVO) stroke patients. Group comparisons were performed using the Wilcoxon rank-sum test due to non-homogeneous variances. Tie2 levels were significantly elevated in stroke patients (W = 130, p = 0.0005), Angpt-2 levels were significantly increased (W = 213.5, p = 0.033), and Angpt-1 levels were significantly reduced (W = 474, p = 0.031). (A ii, B ii, C ii) Temporal dynamics of serum Tie2, Angpt-2, and Angpt-1 measured at pre-MT, immediately post-MT, and 24 hours post-MT. Changes over time were assessed using repeated-measures ANOVA. (A iii, B iii, C iii) Associations between serum Tie2, Angpt-2, and Angpt-1 levels and digital subtraction angiography (DSA) collateral scores (1–3). Mixed-effects modeling revealed that Tie2 concentrations were highest in patients with excellent collaterals (DSA score 3) (F(2,2) = 3.31, p = 0.047), whereas Angpt-1 and Angpt-2 levels did not significantly differ across collateral grades (F(2,2) = 0.83, p = 0.458). (D i,ii) Representative DSA images illustrating differences in collateral circulation. (D i) A DSA collateral score of 1 indicates poor collateral flow, characterized by minimal leptomeningeal vessel perfusion. (D ii) A DSA collateral score of 3 reflects robust collateral distribution with extensive leptomeningeal anastomoses. (F) Volcano plot of differentially expressed genes (DEGs) from whole-blood bulk RNA-seq across DSA collateral groups. Genes with log2 fold-change ≥ 0.6 and p ≤ 0.05 were considered significantly upregulated, whereas genes with log2 fold-change < −0.6 and p ≤ 0.05 were considered significantly downregulated. Top DEGs are labeled. (E) Functional enrichment analysis of biological processes in DSA collateral 1 versus DSA collateral 3 groups. Gene Ontology (GO) chord plot illustrates immune activation, leukocyte signaling, and T-cell receptor related processes, with corresponding DEGs mapped to functional categories. *p < 0.05, ***p < 0.001.

To identify transcriptional programs associated with collateral status, bulk RNA sequencing was performed on whole blood from DSA 1 and DSA 3 patients. Differential expression analysis identified 621 genes associated with collateral status, including PTGDS, RPH3A, and NRG1 (Fig. 8F). Gene Ontology analysis demonstrated enrichment of immune-related pathways, including T-cell receptor signaling, leukocyte activation, and immune response-regulating pathways (Fig. 8E). CIBERSORTx deconvolution further revealed a myeloid-skewed immune profile in DSA 1 patients characterized by increased neutrophils and reduced naïve T cells, whereas DSA 3 patients exhibited more balanced innate-adaptive immune signatures (Supplemental Fig. 6). Together, these findings demonstrate acute dysregulation of Angpt/Tie2 signaling and heightened systemic inflammatory activation in patients with poor collateral circulation, suggesting coordinated immune–vascular mechanisms contribute to collateral responsiveness after stroke.

Conclusions

Our study elucidates the pivotal role of EphA4 in regulating peripheral immune-mediated collateral enlargement following ischemic stroke. A relatively high baseline fraction of circulating immune cells expresses EphA4. We interpret this as consistent with homeostatic immune surveillance functions, wherein Eph/ephrin signaling regulates leukocyte adhesion, cytoskeletal dynamics, and motility during vascular patrolling and tissue trafficking. In this context, EphA4 may function as a regulatory brake that restrains excessive activation and limits inappropriate tissue infiltration under steady conditions. Following ischemic stroke, we observed elevated EphA4 expression on immune cells in both whole blood and pial collateral vessels. These changes correlated with recruitment of EphA4-expressing, GFP⁺ immune cells to pial collateral vessels after ischemic stroke. While previous studies have linked EphA4 to central nervous system injury and inflammatory gene expression ^58–60, we find that EphA4 deficiency increases monocyte infiltration and proliferation in the vessel wall and derepresses Tie2/PI3K-p110α/Akt signaling in BMDMs. Inducible Ccr2-specific EphA4 deletion recapitulated these effects and identified monocyte-intrinsic Tie2 signaling as a key mechanistic mediator. Together with clinical evidence of altered Angpt1, Angpt2, and soluble Tie2 levels associated with collateral status, these findings support the EphA4/Tie2 axis as a potential therapeutic target to enhance collateral remodeling after ischemic stroke. Although Tie2 signaling emerged as a central pathway in our analyses, additional inflammatory pathways, such as NF-κB signaling, could also be explored. The collateral enlargement observed here reflects acute arteriogenesis of pre-existing pial collateral arterioles; however, because parenchymal capillary branching and microvascular angiogenesis were not evaluated, additional contributions from microvascular remodeling to improved cerebral blood flow and functional recovery cannot be excluded.

EphA4 BM deficiency promoted collateral vessel enlargement, increased cerebral blood flow, and facilitated functional recovery post-stroke. Mechanistically, EphA4 deficiency increased recruitment of BMC-derived GFP+ cells to the pial collateral niche, underscoring EphA4's role in constraining immune dynamics. Additionally, we find that EphA4 mediates pro-inflammatory signaling and uniquely controls PI3K-Akt signaling in BMDMs. The PI3K pathway, consisting of regulatory subunits (p85α, p85β, p55γ) and catalytic subunits (p110α, p110β, p110δ), is activated by Ras and tyrosine kinases ⁶¹. The p110α catalytic subunit, widely expressed in monocytes/macrophages ^62–66, plays a critical role in Tie2-mediated signaling ^62,67. Deficiency in p110α results in diminished Tie2 expression and vascular defects resembling those in Tie2 knockout models, as well as impaired RAS-p110α-mediated acute responses to inflammation ⁶⁷. Tie2, which regulates hematopoietic stem cells, neutrophils, and a subset of Tie2-expressing macrophages, has been implicated in tumorigenesis and inflammation, underscoring its broad functional importance ⁶⁸. These findings identify hematopoietic EphA4 as a regulator of monocyte/macrophage Tie2–PI3K-p110α/Akt signaling that influences post-stroke inflammation and collateral remodeling.

Interestingly, while neutrophils rapidly infiltrate the brain following brain ischemia, their recruitment to the collateral vessel wall in the meninges is minimal compared with that of GFP⁺/Iba1⁺ monocyte/macrophages. This supports studies indicating that neutrophils primarily exacerbate cerebral injury through mechanisms such as oxidative stress, blood-brain barrier disruption, and extracellular matrix degradation ^69–71. In contrast, monocytes engage in cytokine secretion, phagocytosis, and angiogenic processes ^72,73. Ly6C^hi monocytes, for instance, foster pro-resolving macrophage polarization, underscoring their protective role in acute ischemic stroke and potential as therapeutic targets ⁷⁴. Monocyte-derived cytokines and chemokines may also serve as biomarkers for predicting patient outcomes and collateral status, thereby informing stratified treatments ^75,76.

The recruitment of monocytes/macrophages stimulates endothelial cell proliferation within the vessel wall, exerting pro-angiogenic and lymphangiogenic effects ^77,78. This occurs through mechanisms mediated by key cytokines and signaling pathways, including vascular growth factors, tumor necrosis factor-alpha, interleukin-1 beta, and platelet-derived growth factor ⁷⁹. These factors promote endothelial cell activation, angiogenesis, and vessel enlargement by enhancing cell migration, proliferation, and survival ^80,81. In this study, immune-specific deletion of EphA4 enhances proliferation within the collateral niche, a hallmark of arteriogenesis ⁴⁷. This pro-arteriogenic effect is correlated with a shift from a pro-inflammatory to a pro-resolving phenotype in EphA4-null monocytes/macrophages and with elevated p-Akt levels. Supporting studies revealed that monocyte injection enhances collateral growth by promoting arteriogenesis, as demonstrated in GM-CSF-activated monocytes, which improve matrix remodeling in hindlimb or coronary niches ^7,82,83. Thus, targeting EphA4 could further optimize immune cell-mediated vascular remodeling across collateral niches.

Our clinical observations provide preliminary, hypothesis-generating evidence that acute alterations in the Angpt/Tie2 signaling axis are associated with leptomeningeal collateral status in patients with LVO stroke. Although exploratory due to the small and heterogeneous thrombectomy cohort, these findings demonstrate elevated circulating Angpt-2 and sTie2 with reduced Angpt-1, consistent with the vessel-destabilizing profile previously reported in experimental stroke models and lacunar stroke cohorts ^30,44,84,85. Notably, we provide the first evidence in patients undergoing mechanical thrombectomy that higher pre-procedural sTie2 levels are associated with improved DSA collateral grades. Because circulating Angpt/Tie2 factors likely originate from multiple vascular and immune compartments, these measurements should be interpreted as systemic biomarkers rather than as cell-specific readouts, and they do not establish causality. Nevertheless, higher circulating sTie2 levels in patients with good collaterals may reflect enhanced Tie2 signaling that restrains inflammatory activation and supports leptomeningeal collateralization, consistent with the efficacy of delayed Tie2-activating therapies (e.g., Vasculotide, VE-PTP inhibitors) in experimental stroke models ^16,86. Together with our transcriptomic analyses, these findings suggest that systemic vascular and immune signaling influence the spatially restricted responses required for collateral remodeling after stroke. Future studies using larger cohorts and cell-resolved profiling will be needed to define the mechanistic and cell-specific contributions of Angpt/Tie2 signaling to collateral remodeling and stroke outcome.

Whole-blood transcriptomic analysis stratified by DSA collateral grade identified a robust inflammatory signature in patients with poor (DSA 1) versus good (DSA 3) collaterals. Among 621 differentially expressed genes, pathways related to leukocyte activation, T-cell receptor signaling, and immune regulation were enriched in the poor-collateral group. Deconvolution and cell-type-specific analyses further revealed a myeloid-skewed immune landscape characterized by neutrophil expansion, reduced naïve T cells, and inflammatory monocyte/neutrophil activation programs, whereas good collateral status was associated with more balanced innate-adaptive immune signaling. Several top upregulated genes in the poor-collateral cohort are mediators of impaired arteriogenesis and vascular–immune dysfunction: Prostaglandin D2 synthase (PTGDS), the most strongly upregulated gene, produces PGD2 and 15d-PGJ2, which exert anti-angiogenic effects and suppress shear-stress responses critical for collateral enlargement ^87,88; Rabphilin-3A (RPH3A) facilitates rapid granule exocytosis in cytotoxic T cells and neutrophils, potentially amplifying perivascular delivery of perforin, granzymes, and reactive oxygen species ⁸⁹; Neuregulin-1 (NRG1) isoforms prevalent in acute stroke can drive monocyte/macrophage-mediated inflammation via ErbB4, exacerbating Angpt-2-induced pericyte loss and barrier leakage ⁹⁰. These findings support the emerging concept that leptomeningeal collaterals represent an immune-responsive vascular niche ^16,91. In patients with poor collateral status, altered Angpt/Tie2 signaling and inflammatory programs involving PTGDS, RPH3A, NRG1, and T-cell signaling pathways may promote a feed-forward inflammatory state that limits arteriogenic remodeling.

A primary limitation of this study is the use of static vessel-painted whole-mount preparations to assess collateral diameter, which captures only a single time point and cannot distinguish sustained arteriogenesis from transient vasodilation or ongoing structural remodeling. Although EphA4-deficient mice exhibited significant collateral enlargement at 24 hours after pMCAO, future studies using two-photon intravital microscopy will be needed to define the temporal dynamics of collateral perfusion, leukocyte–endothelial interactions, and long-term vessel maturation in vivo. In addition, EphA4-mediated neuroinflammatory responses within the meninges and parenchyma may act in concert to support recovery, while the upstream mechanisms regulating EphA4 expression prior to monocyte mobilization remain unclear and will require temporally resolved, compartment-specific analyses. Another limitation is the restricted temporal window of this study, as extending analyses to later stages after stroke will be important to determine whether early EphA4-dependent immune modulation promotes durable collateral remodeling, sustained functional recovery, and changes in monocyte/macrophage and perivascular macrophage populations.

In summary, our findings identify monocyte/macrophage EphA4 as an immunoregulatory brake on leptomeningeal collateral remodeling after ischemic stroke. By constraining Tie2–PI3K/Akt signaling and monocyte recruitment, EphA4 limits arteriogenesis and collateral perfusion. Consistent with this, altered Angpt/Tie2 signaling and immune activation in patients with poor collateral status support a translationally relevant immune–vascular axis and identify Angpt/Tie2 signaling as a potential biomarker and therapeutic target to enhance collateral flow.

Supplementary Material

Supplemental Materials:

Figures S1–S6

Table S1. Major Resource Table

ARRIVE checklist

Non-Standard Abbreviations and Acronyms:

WT

wildtype

KO

knockout

VP

vessel painting

pMCAO

permanent middle cerebral artery occlusion

MCA

middle cerebral artery

PCA

posterior cerebral artery

ACA

anterior cerebral artery

Angpt

angiopoietin

BM

bone marrow

BMCs

bone marrow cells

BMDMs

bone marrow-derived macrophages

i.p.

intraperitoneal

CBF

cerebral blood flow

LVO

large vessel occlusion

MDMs

monocyte-derived macrophages

DSA

digital subtraction angiography

PI3K

phosphatidylinositol 3-kinase