We examined whether damage to the glia limitans (GL), via exposure to the gliotoxin l-alpha-aminoadipic acid (l-alphaAAA), alters hypercapnia-induced pial arteriolar dilation in vivo. Anesthetized female rats were prepared with closed cranial windows. Pial arteriolar diameters were measured using intravital microscopy. l-alphaAAA (2 mM) was injected into the space under the cranial windows 24 h before the study, and injury to the GL was confirmed by light microscopy. l-alphaAAA was associated with a reduction in pial arteriolar CO(2) reactivity to 40-50% of the level seen in vehicle-treated controls, with no further reduction in the CO(2) response after nitric oxide (NO) synthase (NOS) inhibition via N(omega)-nitro-l-arginine (l-NNA). Subsequent blockade of prostanoid synthesis, via indomethacin (Indo), reduced CO(2) reactivity to 10-15% of normal. In vehicle-treated controls, l-NNA, followed by Indo, reduced the response to approximately 50% and then to 15-20% of the normocapnic value, respectively. On the other hand, l-alphaAAA had no effect on vascular responses to the endothelium-dependent vasodilator acetylcholine or the NO donor SNAP and did not alter cortical somatosensory evoked responses. This indicates an absence of any direct l-alphaAAA actions on pial arterioles or influence on neuronal transmission. Furthermore, l-alphaAAA did not alter the vasodilation elicited by topical application of an acidic artificial cerebrospinal fluid solution, suggesting that the GL influences the pial arteriolar relaxation elicited by hypercapnic, but not local extracellular (EC), acidosis. That differences exist in the mechanisms mediating hypercapnia- versus EC acidosis-induced pial arteriolar dilations was further exemplified by the finding that topical application of a neuronal NOS (nNOS)-selective blocker (ARR-17477) reduced the response to hypercapnia (by approximately 65%) but not the response to EC acidosis. Disruption of GL gap junctional communication, using an antisense oligodeoxynucleotide (ODN) connexin43 knockdown approach, was accompanied by a 33% lower CO(2) reactivity versus missense ODN-treated controls. These results suggest that the GL contribution to the hypercapnic vascular response appears to involve the NO-dependent component rather than the prostanoid-dependent component and may involve gap junctional communication. We speculate that the GL may act to facilitate the spread, to pial vessels, of hypercapnia-induced vasodilating signals arising in the comparatively few scattered nNOS neurons that lie well beneath the GL.