Discussion

Chemokines have traditionally been characterized by their ability to recruit immune cells to sites of injury or inflammation, and extensive studies have focused on their roles in leukocyte trafficking under various pathological conditions22. However, emerging evidence suggests that chemokines can also directly regulate cellular states within the central nervous system23,24, raising the possibility that exogenously administered chemokines may hold therapeutic potential in central nervous system disorders.

This study examined whether rCXCL16 modulates microglial responses and ischemic injury. In primary microglia, rCXCL16 reduced inflammation-associated and increased repair-associated gene expression under OGD/R and LPS stimulation. After MCAO, rCXCL16 reduced TTC-defined infarct volume and mNSS scores, decreased CD86 and increased CD206 immunoreactivity in IBA1 + microglia/macrophages, and lowered iNOS while elevating Arg-1 in ischemic brain tissue. rCXCL16-treated microglia increased HT-22 viability and reduced apoptosis in co-culture; their conditioned medium reproduced the viability benefit, whereas direct rCXCL16 did not. Together, these findings associate rCXCL16-mediated microglial modulation with reduced ischemic injury and improved neurological outcomes.

Although scRNA-seq identified several CXCL16-expressing cell populations, their relative contributions to the post-ischemic increase in tissue CXCL16 remain unresolved. Furthermore, transcript detection does not establish CXCL16 protein production or release in its soluble form, which will require spatial and protein-level validation.

The cellular distribution of CXCR6 also warrants cautious interpretation. In GSE227651, Cxcr6 transcripts were detected predominantly in T/NK lymphocytes and were not detected in the annotated microglial population, whereas sporadic transcript detection was observed in the larger integrated MCAO atlas SCP307825. Such variation may reflect low transcript abundance, sequencing dropout, differences in cellular composition or annotation, or biological heterogeneity among datasets. Although we confirmed CXCR6 protein expression in cultured primary microglia and observed responses to rCXCL16 in vitro, neonatal cultured microglia may not fully reproduce the receptor expression profile of adult microglia in the ischemic brain. Moreover, CXCR6 dependence was not examined using pharmacological blockade or genetic deletion. Thus, the in vivo effects of rCXCL16 cannot be attributed exclusively to microglia and may also involve infiltrating CXCR6-expressing immune populations, particularly T/NK lymphocytes. Tissue-level colocalization and cell-specific inhibition or deletion of CXCR6 will be required to define the relative contribution of these populations.

Microglial responses after ischemic stroke are temporally dynamic and highly heterogeneous and cannot be adequately represented by a binary M1/M2 classification26. Individual microglia may simultaneously exhibit inflammatory, homeostatic, phagocytic, metabolic, and repair-associated features, depending on the stage of injury, anatomical location, and local cellular environment. Therefore, CD16, CD32, iNOS, CD206, Arg-1, and IL-10 should be interpreted as selected state-associated markers rather than definitive indicators of discrete and mutually exclusive M1 or M2 phenotypes. In the present study, rCXCL16 reduced selected inflammation-associated markers and enhanced selected repair-associated markers under both OGD/R and LPS stimulation. These findings indicate modulation of the microglial transcriptional profile but do not establish conversion from one discrete state to another. In addition, our in vitro experiments used neonatal primary microglia, whose transcriptional and functional states may differ from those of adult microglia and may be further altered by isolation and culture27–29. Therefore, the present in vitro findings should not be directly extrapolated to adult microglial responses in the ischemic brain.

Building upon the observed changes in inflammation- and repair-associated gene expression, we next examined whether rCXCL16-treated microglia influenced HT-22 cell survival under OGD/R conditions. In the non-contact co-culture system, untreated microglia did not significantly improve HT-22 cell viability, whereas rCXCL16-pretreated microglia increased cell viability and reduced apoptosis. Further supporting the co-culture findings, conditioned medium from rCXCL16-treated microglia significantly improved HT-22 cell viability, whereas direct rCXCL16 treatment had no significant protective effect across the tested concentration range. Together, these findings support the involvement of soluble factors released by rCXCL16-treated microglia, rather than a prominent direct effect of rCXCL16 on HT-22 cells. However, the specific soluble mediators involved and their dependence on microglial CXCR6 signaling remain to be determined.

In vivo, rCXCL16 treatment was associated with reduced ischemic injury, improved neurological outcomes, and altered inflammation- and repair-associated marker profiles after MCAO. However, because IBA1 does not distinguish resident microglia from infiltrating macrophages and iNOS and Arg-1 were measured in ischemic brain tissue homogenates, these changes cannot be attributed specifically to microglia. Accordingly, the in vivo findings support an association between rCXCL16 treatment and modulation of post-ischemic inflammatory responses rather than establishing a microglia-specific causal mechanism.

Together, these results indicate that rCXCL16 modulates microglial responses and is associated with reduced neuronal injury in vitro and improved outcomes after MCAO.

This study has several limitations. Although the conditioned-medium experiments support the involvement of microglia-derived soluble factors, the responsible mediators remain unidentified. CXCR6 expression was confirmed in cultured primary microglia, but receptor dependence and downstream signaling were not examined using pharmacological blockade or genetic manipulation, and contributions from other CXCR6-expressing cells in vivo cannot be excluded. The cellular specificity and causal contribution of microglia in vivo also remain unresolved because IBA1 does not distinguish resident microglia from infiltrating macrophages, iNOS and Arg-1 were measured in tissue homogenates, and TUNEL staining was not combined with lineage-specific markers. Furthermore, cultured neonatal microglia may not fully recapitulate adult microglial states, and the lack of biological replication at each time point in the single-cell dataset precluded temporal inference and quantitative attribution of CXCL16 sources. Future studies using cell-type-specific CXCR6 manipulation, adult microglial models, and spatial or protein-level analyses are required.