The present study has confirmed that the vasorelaxing effect of NMDA depends on the perivascular retinal tissue, whereas this tissue does not influence the vasorelaxing effect of adenosine.
8 9 In addition, the study has shown that these mechanisms are coupled so that NMDA-induced vasorelaxation of porcine retinal arterioles in vitro is mediated through adenosine.
Glutamate is a widely distributed excitatory neurotransmitter in the vertebrate retina and has specific activity on two major classes of receptors (i.e., metabotropic receptors and ionotropic counterparts), such as receptors selectively activated by NMDA. The NMDA receptor has been suggested to be an important contributor to the neurotoxic effects of glutamate observed during ischemia and cell death in retinal disease.
11 12 13 14 15 In the perivascular retinal tissue, NMDA receptors have been identified on retinal ganglion cells, amacrine cells, and cone photoreceptors,
11 as well as in Müller cells,
16 but not on astrocytes.
17 On the other hand, NMDA receptors have neither been identified on vascular smooth muscle cells nor on endothelial cells of the retinal vascular walls.
11 15 This anatomic evidence is in accordance with the finding that NMDA-induced vasorelaxation depends on the presence of perivascular retinal tissue. However, it remains to be shown how the different NMDA receptor-containing retinal cell type(s) participate in this response.
Adenosine is synthesized by 5′-nucleotidase dephosphorylation of AMP, which results from increased turnover of ATP in retinal neurons and glial cells. Both adenosine and adenosine A1 and A2 receptors have been demonstrated in all layers of the human retina.
18 In the present study, inhibition of the NMDA receptor did not affect adenosine-induced vasorelaxation. However, inhibition of both adenosine receptors and hydrolysis of ATP significantly reduced the vasorelaxing effect of NMDA. Combined with evidence from a previous study showing that the vasorelaxing effect of ATP is abolished by blocking its hydrolysis to adenosine,
9 it can be hypothesized that the vasorelaxing effect of adenosine is due to an action on the vascular wall, either directly on the vascular smooth muscle cells or via the endothelial cells. Furthermore, the formation of adenosine is by hydrolysis of ATP which is released after stimulation of NMDA receptors in the perivascular retinal tissue. These data are consistent with data from pial arterioles, where administration of adenosine receptor antagonists have been shown to prevent NMDA-induced vasorelaxation.
19
The interaction between NMDA and adenosine in the dilation of retinal arterioles includes components that have still not been elucidated in detail. Thus, previous studies have shown that vasodilation induced by NMDA but not by adenosine can be blocked by the prostaglandin synthesis inhibitor ibuprofen.
8 9 Prostaglandins are synthesized from arachidonic acid by the cyclooxygenase (COX) enzyme, of which two subtypes, COX-1 and -2, have been identified.
20 21 22 COX-1 is ubiquitously expressed in the retinal cells and is required for cell homeostasis, whereas COX-2 is abundantly present in synaptic regions of the retina.
23 COX-2 has been shown to be upregulated in response to oxidative stress in ischemia–reperfusion studies,
24 25 which implies that prostaglandins are involved in changing retinal perfusion during pathologic conditions. However, prostaglandins may have both vasorelaxing and vasoconstricting effects,
3 4 26 27 and most prostaglandin receptors have been found in the retinal vascular walls.
28 Therefore, it is an important challenge to identify which of the prostaglandins and corresponding receptors are involved when ibuprofen abolishes NMDA-induced vasorelaxation and to explain why adenosine-induced vasorelaxation is unaffected by prostaglandin synthesis inhibition.
In summary, the present findings suggest that the vasorelaxing effect of adenosine is due to an action on the vascular wall and that adenosine is synthesized by hydrolysis of ATP, which is released after stimulation of NMDA receptors in the perivascular retinal tissue. A more detailed elucidation of the mechanisms involved in this reaction pattern is needed, but the reactions may be part of a pathway that is especially active during ischemia or high metabolic activity where a relative lack of nutrients and accumulation of metabolites would induce a compensatory vasorelaxation and a consequent regulation of the blood flow. The present findings may therefore contribute to a deeper understanding of the mechanisms underlying the regulation of vascular tone in the response to metabolic changes in the retina.