Results

CXCL16 is upregulated at day 3 after MCAO, with Cxcl16 transcripts enriched across multiple brain cell populations

To determine whether CXCL16 may be involved in the pathological process of cerebral ischemia, we examined its expression at day 3 post-MCAO, a time point associated with substantial brain edema19,20. Both Western blotting (Fig. 1A-B) and RT-qPCR (Fig. 1C) revealed markedly elevated CXCL16 levels at this time point compared to sham controls. To further characterize its cellular distribution, we reanalyzed the publicly available GSE227651 single-cell RNA-sequencing dataset. CXCL16 transcripts were enriched primarily in microglia, border-associated macrophages, and meningeal fibroblast populations, suggesting that these populations may represent potential cellular sources of CXCL16 after ischemic injury (Fig. 1D–F). Analysis of receptor distribution showed that Cxcr6 transcripts were predominantly detected in T/NK lymphocytes, whereas no detectable Cxcr6 expression was observed in the annotated microglial population (Supplementary Fig. 2G).

Fig. 1: CXCL16 upregulation after cerebral ischemia and the cellular distribution of Cxcl16 transcripts. (A-B) Western blot analysis of CXCL16 protein expression at day 3 after MCAO in SHAM and MCAO mice (n = 5). (C) RT-qPCR analysis of CXCL16 mRNA levels (n = 6). (D) UMAP visualization of major cell populations identified in the GSE227651 single-cell RNA-sequencing dataset. (E) Feature plot showing the distribution of CXCL16 expression. (F) Dot plot showing CXCL16 expression across major cell types. **p < 0.01, ****p < 0.0001. Abbreviations: OPCs, oligodendrocyte precursor cells; VSMCs, vascular smooth muscle cells; T/NK, T and natural killer lymphocytes.

rCXCL16 modulates inflammation- and repair-associated gene expression in primary microglia under OGD/R conditions

Microglia are central regulators of neuroinflammation following cerebral ischemia and express the CXCL16 receptor CXCR68,21. Given that CXCL16 was elevated at day 3 post-MCAO, we investigated whether it modulates microglial functional states. Primary microglia were isolated and characterized by immunofluorescence staining for Iba-1 and CD206 (Supplementary Fig. 2 A). Double immunofluorescence staining further demonstrated CXCR6 immunoreactivity in Iba1-positive cells under basal conditions (Supplementary Fig. 2B).

The OGD/R model used for primary microglia was optimized as follows. Cell viability was assessed after five different durations of OGD followed by 24 h of reoxygenation. Four hours of OGD reduced microglial viability to approximately 50% of the normoxic control level and was therefore selected for subsequent experiments (Supplementary Fig. 2 C). Using this condition, we examined the endogenous expression of both components of the CXCL16–CXCR6 axis. Western blotting showed that CXCL16 and CXCR6 were detectable in primary microglia under normoxic conditions and that the expression of both proteins was significantly increased following OGD/R (Supplementary Fig. 2D–F). These findings indicate that cultured primary microglia express both CXCL16 and CXCR6 under the experimental conditions used and may represent a potential source of CXCL16 under OGD/R conditions.

To examine whether rCXCL16 alters microglial responses under OGD/R conditions, we assessed the expression of inflammation-associated markers, including CD16, CD32, and iNOS, together with repair-associated markers, including CD206, Arg-1, and IL-10. Treatment with 120 ng/mL rCXCL16 significantly reduced the expression of the inflammation-associated markers CD16, CD32, and iNOS following OGD/R (Fig. 2A-C). In parallel, rCXCL16 increased the expression of the repair-associated markers CD206 and Arg-1 (Fig. 2D and E), while IL-10 expression was increased at both 60 and 120 ng/mL (Fig. 2F). These findings demonstrate that rCXCL16 modulates selected inflammation- and repair-associated genes in primary microglia under OGD/R conditions (see Fig. 2).

Fig. 2: Effects of rCXCL16 on inflammation- and repair-associated gene expression in primary microglia under OGD/R conditions. RT-qPCR analysis of inflammation-associated markers, including CD16, CD32, and iNOS (A–C), and repair-associated markers, including CD206, Arg-1, and IL-10 (D–F), in primary microglia exposed to OGD/R and treated with the indicated concentrations of rCXCL16 (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001.

rCXCL16 partially reverses LPS-induced changes in inflammation- and repair-associated gene expression of primary microglia

We next examined whether the modulatory effects of rCXCL16 were also observed in response to a non-ischemic inflammatory stimulus. LPS stimulation significantly increased the expression of CD16, CD32, and iNOS and reduced the expression of CD206, Arg-1, and IL-10 compared with the control group (Fig. 3A–F). rCXCL16 treatment significantly attenuated these LPS-induced changes. Nevertheless, all six markers remained significantly different from control levels, indicating that rCXCL16 partially, but not completely, reversed the transcriptional response induced by LPS.

Fig. 3: Effects of rCXCL16 on inflammation- and repair-associated gene expression in LPS-stimulated primary microglia. RT-qPCR analysis of selected inflammation-associated markers, including CD16, CD32, and iNOS (A–C), and repair-associated markers, including CD206, Arg-1, and IL-10 (D–F), in Control, LPS, and LPS + rCXCL16 (120 ng/mL) groups (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

rCXCL16-pretreated primary microglia enhance HT-22 cell viability in a co-culture model under OGD/R conditions

Given the effects of rCXCL16 on inflammation- and repair-associated gene expression in microglia under both LPS and OGD/R conditions, we next asked whether these changes were associated with neuronal protection. A co-culture model of primary microglia and HT-22 neurons was therefore established to evaluate the protective potential of rCXCL16-pretreated microglia under OGD/R conditions (Fig. 4A). An MTT assay was first performed in HT-22 cells to determine the optimal duration of OGD, and we chose 4 h for further assays (Fig. 4B). The HT-22 cells were then co-cultured with microglia during the reoxygenation phase. Under OGD/R conditions, rCXCL16-pretreated microglia markedly enhanced neuronal viability relative to neuron-only cultures or co-cultures with non-treated microglia (Fig. 4C). Given that apoptosis is the predominant mode of cell death within the ischemic penumbra, we performed flow cytometry (FACS) to quantify neuronal apoptosis (Fig. 4D). Consistently, under OGD/R conditions, the apoptotic rate was markedly decreased in the presence of rCXCL16-treated microglia versus both the neuron-only group and the co-culture group with untreated microglia (Fig. 4E).

Because the non-contact co-culture findings suggested the involvement of microglia-derived soluble factors, we performed complementary experiments to determine whether the protective effect could be reproduced using microglial conditioned medium and whether rCXCL16 acted directly on HT-22 cells. Direct treatment with rCXCL16 at 60, 120, or 240 ng/mL did not significantly improve HT-22 cell viability following OGD/R (Supplementary Fig. 2H). In contrast, conditioned medium derived from rCXCL16-treated microglia significantly increased HT-22 cell viability compared with conditioned medium from vehicle-treated microglia, whereas vehicle-treated microglial conditioned medium did not differ significantly from the OGD/R group (Supplementary Fig. 2I). These findings support the involvement of soluble factors released by rCXCL16-treated microglia rather than a prominent direct protective effect of rCXCL16 on HT-22 cells.

Fig. 4: rCXCL16-pretreated microglia protect HT-22 cells against OGD/R-induced viability loss and apoptosis in co-culture. (A) Schematic of the co-culture model. MG, microglia. (B) MTT assay determining the optimal OGD duration in HT-22 cells. Co-culture with rCXCL16-pretreated microglia significantly improved neuronal viability (C) and reduced neuronal apoptosis (E)following OGD/R exposure. Representative FACS plots (D) show neuronal apoptosis in each group under OGD/R conditions. Data in (B) and (C) are from six technical replicates, typical results from three independent replicates. Data in (E) are from three independent experiments (each point represents one experiment). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

rCXCL16 attenuates infarct volume and enhances neurological outcomes following MCAO

To evaluate its neuroprotective effects in vivo, rCXCL16 was administered intracerebroventricularly at doses of 10, 30, or 60 µg/kg at 1 h after MCAO. TTC staining revealed that treatment with rCXCL16 at doses of 30 and 60 µg/kg, but not 10 µg/kg, markedly decreased infarct volume compared with the vehicle group. Notably, this effect exhibited a bell-shaped dose–response pattern, with the intermediate dose showing the most pronounced reduction (Fig. 5A-B). Consistent with the TTC results, neurological function was also significantly improved in rCXCL16-treated mice, with the greatest improvement observed at the intermediate dose (Fig. 5C). We therefore chose 30 µg/kg as the optimal dose for further in vivo testing.

Fig. 5: rCXCL16 reduces infarct volume and neurological deficits in MCAO mice. Mice were treated with vehicle or rCXCL16 (10, 30, or 60 µg/kg). At 24 h post-MCAO, TTC staining (A-B) showed that rCXCL16 at 30 and 60 µg/kg significantly reduced infarct volume versus vehicle, with a bell-shaped dose–response and maximal effect at 30 µg/kg (n = 6 mice per group). Neurological scores (C) showed parallel improvements (n = 10–12 mice per group). ns, not significant; *p < 0.05, **p < 0.01, ****p < 0.0001.

rCXCL16 alters inflammation- and repair-associated marker profiles in microglia/macrophages after MCAO

To examine whether rCXCL16 altered microglia/macrophage-associated responses in vivo, we assessed CD86 and CD206 immunoreactivity in Iba-1+ cells, together with iNOS and Arg-1 protein expression in ischemic brain tissue. MCAO increased the number of CD86+/Iba-1+ cells in the peri-infarct region (Fig. 6A and C), whereas rCXCL16 treatment reduced this population and increased the number of CD206+/Iba-1+ cells (Fig. 6B and D). Consistently, rCXCL16 reduced iNOS protein expression and increased Arg-1 expression in ischemic brain tissue (Fig. 6E-G). These findings indicate that rCXCL16 treatment is associated with a shift toward reduced inflammatory and enhanced repair-associated marker expression after MCAO (see Fig. 6).

Fig. 6: rCXCL16 modulates inflammation- and repair-associated marker profiles after MCAO. Immunofluorescence analysis showed that rCXCL16 reduced CD86+/Iba-1+ microglia/macrophages and increased CD206+/Iba-1+ microglia/macrophages in the peri-infarct region (n = 5). Western blotting showed reduced iNOS and increased Arg-1 protein expression in ischemic brain tissue (n = 6). Scale bar = 100 μm. *p < 0.05, ****p < 0.0001.

rCXCL16 reduces apoptosis in vivo

To determine whether rCXCL16 treatment attenuates apoptotic cell death, we next examined cell apoptosis in the ischemic cortex. TUNEL staining displayed an elevation of apoptotic cells in the vehicle-treated MCAO group. However, rCXCL16 administration markedly lowered the count of TUNEL-positive cells (Fig. 7A-B). Western blotting was further performed to confirm anti-apoptotic effects. Compared to the vehicle group, rCXCL16 significantly downregulated Bax expression and up-regulated Bcl-2 levels, leading to a markedly reduced Bax/Bcl-2 ratio (Fig. 7C-D). These data demonstrate that rCXCL16 effectively suppresses apoptosis, contributing to its neuroprotective role in ischemic stroke.

Fig. 7: rCXCL16 reduces apoptosis in the ischemic cortex following MCAO. TUNEL staining (A-B) and Western blot analysis of Bax and Bcl-2 (C, D) showed that rCXCL16 treatment substantially decreased apoptotic cell numbers and enhanced the Bcl-2/Bax ratio (n = 6). *p < 0.05, ***p < 0.001, ****p < 0.0001.