It is also diYcult to explain how elevated IOP leads to elevated levels extracellular glutamate. Initial reports suggested that retinal glutamate transporters are decreased in glaucoma (Naskar et al., 2000; Martin et al., 2002). However, more recent evidence suggests the opposite may occur, with levels of certain transporters elevated in response to increased IOP (WoldeMussie et al., 2004; Hartwick et al., 2005; Sullivan et al., 2006).

Although these inconsistencies leave many issues about the role of glutamate in glaucoma unresolved, these observations do suggest the elevation of intracellular calcium by glutamate acting on NMDA receptors may be detrimental to retinal ganglion cells under some conditions. This response may contribute to glaucomatous loss of retinal ganglion cells, but it is unclear if and how elevated pressure is linked to this process.

3. Purines

Recent work from our laboratory has suggested that purines may connect increased IOP with changes in ganglion cell health and eventually even death. This hypothesis provides for a source of excess extracellular ATP, involvement of a toxic receptor, and a negative feedback system that may preserve ganglion cell health. Preliminary findings may also clarify a role of glutamate in the process. The basic hypothesis is modeled in Fig. 1.

IOP

Ganglion cell

+−

Cell death

FIGURE 1 Hypothesis of eVect of purines on ganglion cell health in glaucoma. Elevated pressure leads to a release of ATP. Although astrocytes are indicated, release from other sources such as Mu¨ller cells or endothelial cells is also possible. This excess extracellular ATP can stimulate P2X7 receptors on retinal ganglion cells, leading to elevation of calcium, injury and eventually perhaps cell death. Alternatively, the excess ATP can be converted into to adenosine by ecto nucleotidases (E) and prevent calcium overload and cellular damage.

a. Pressure and ATP release. Although originally known as a cellular energy source, ATP has been recognized as an extracellular transmitter for several decades (Burnstock et al., 1970). It is now accepted that the extracellular signaling role of ATP has a negligible impact on cell energetics under most circumstances. ATP can be released from both neuronal and non neuronal cells. A particularly eVective trigger for ATP release from non neural cells involves mechanical distention in the form of pressure, flow or stretch (Burnstock, 1999). We have found that ocular tissues also release ATP in response to mechanical perturbation (Mitchell et al., 1998; Fleischhauer et al., 2001; Mitchell, 2001). This led to the hypothesis that excess ATP might also be released in response to the changing IOP found in glaucoma.

Several preliminary lines of evidence suggest excess ATP may be released into the extracellular space in glaucoma. ATP was increased in bovine eyecups exposed to elevated hydrostatic pressure. This eVect that was not due to a change in oxygen partial pressure or cell lysis (Zhang et al., 2006c). Whether the observed release reflected actual hydrostatic pressure or a relative movement of the retina with respect to the sclera remains to be determined, but the response was robust and proportional to the change in pressure. The ectoATPase NTPDase1 acts as a marker for excess extracellular ATP in retinal cells (Lu et al., 2007), and initial trials indicate that NTPDase1 levels are higher in retina from primate eyes with experimental glaucoma than in contralateral controls (Lu et al., 2007). ATP levels are elevated fivefold in the aqueous humor of patients with acute angle closure glaucoma (Zhang et al., 2007); this excess ATP could act as a precursor for the increased adenosine found in the anterior chamber of glaucomatous eyes (Daines et al., 2003). The source of excess extracellular ATP in the glaucomatous posterior eye is unknown, but mechanical stimulation of astrocytes on the outer retinal layer has led to release of ATP from the Mu¨ller cells in areas adjacent to the ganglion cells, implicating glial cells in the response (Newman, 2001). The propensity of astrocytes to release ATP throughout the nervous system also suggests they could contribute to excess extracellular ATP in the retina or optic nerve head as well (Joseph et al., 2003).

b. P2X7 Receptors, NMDA Receptors, and Ganglion Cell Death. The eVects of any excess extracellular ATP will be primarily determined by the particular receptors expressed on adjacent membranes. Several studies have demonstrated that P2X7 receptors are expressed in adult retinal ganglion cells (Brandle et al., 1998; Ishii et al., 2003; Puthussery and Fletcher, 2004). The P2X7 receptor is distinguished from P2X1–6 receptors by its elongated C terminus and its tendency to kill peripheral cells (Surprenant et al., 1996). This finding suggested that stimulation of the P2X7 receptor on ganglion cells might also be lethal. Activation of the P2X7 receptor leads to an elevation of

intracellular calcium that shows little inactivation, and to the death of ganglion cells (Zhang et al., 2005). Although the agonist BzATP can stimulate other P2 receptors in addition to the P2X7 receptor, the involvement of the P2X7 was confirmed with multiple assays including inhibition by 100 nM of antagonist Brilliant Blue G. This death involved activation of caspase 3, but was not associated with the increased permeability to large fluorescent dyes associated with the activation of the P2X7 receptor in peripheral cells (Surprenant et al., 1996; Zhang et al., 2005). Several preliminary observations suggest that the NMDA receptor may be involved in the death of retinal ganglion cells accompanying P2X7 receptor stimulation (Mitchell et al., 2006; Mitchell, 2008).

c. Neuroprotection by Adenosine. Extracellular nucleotidases rapidly dephosphorylate released ATP into adenosine (Robson et al., 2006). This close relationship between levels of extracellular ATP and adenosine, combined with the ability of these agonists to stimulate distinct receptors, makes the study of purinergic transmission particularly complex. These contrasting eVects are especially acute in retinal ganglion cells, as activation of P2X7 receptors leads to cell death, while stimulation of certain adenosine receptors is neuroprotective.

In keeping with these opposing principals, a brief application of ATP clearly raised calcium levels, while prolonged exposure to ATP did not kill cells and in some cases was actually protective (Zhang et al., 2006b). These protective actions of ATP were not mimicked by analogue ATPgS; levels of cell loss were similar with ATPgS and BzATP. As the bond of the terminal phosphate group in ATPgS makes it much more resistant to hydrolysis, this implied that the hydrolysis product could confer protection. The most likely candidate was adenosine, and adenosine itself was indeed able to prevent the calcium rise and death triggered by BzATP in retinal ganglion cells, consistent with the protection seen with ATP but not ATPgS (Zhang et al., 2006b). This is also consistent with the general recognition of adenosine as a neuroprotective agent. Adenosine makes a major contribution to the response to retinal ischemia (Ghiardi et al., 1999), while levels of adenosine rise in the ischemic retina and limit the neuronal damage (Roth et al., 1997a,b).

The protective actions of adenosine following P2X7 receptor activation in ganglion cells is response mediated, at least in part, by the A3 adenosine receptor. The A3 receptor antagonist MRS1191 prevented the ability of adenosine to block the calcium rise triggered by BzATP (Zhang et al., 2006b). Cl IB MECA, a relatively specific agonist for the A3 receptor, mimicked the ability of adenosine to inhibit the calcium rise triggered by BzATP. Cl IB MECA and another A3 receptor agonist IB MECA also reduced the cell death triggered by BzATP.

Although these pharmacological experiments provided clear functional evidence that the A3 receptor is neuroprotective, molecular identification of the receptor was required as a previous in situ hybridization study was unable to find any message for the A3 receptor in the rat eye (Kvanta et al., 1997). However, traditional and quantitative PCR did identify the A3 receptor in material from the ganglion cell layer of the rat retina using laser capture microdissection (Zhang et al., 2006a). To ensure the message did not come from other cell types also present in the ganglion cell layer, analysis was repeated on ganglion cells isolated using the immunopanning technique. The A3 receptor identified in ganglion cells was cloned and found to be >99% identical to that found elsewhere.

It is likely that the A3 receptor acts with the A1 receptor to protect ganglion cells. Agonists for the A1 adenosine receptor are known to protect retinal ganglion cells against ischemic challenge (Larsen and Osborne, 1996), and are also known to block calcium channels on ganglion cells of amphibians (Sun et al., 2002) and rats (Hartwick et al., 2004). The hyperpolarization of ganglion cells by adenosine was largely inhibited by an A1 antagonist (Newman, 2003). In other systems, the A1 adenosine receptor inhibits voltage dependent calcium channels, following direct block of the CaVa subunit by Gbg coupled to the A1 receptor (Clapham, 1994; Dolphin, 2003). Although the mechanisms linking the A3 receptor with neuroprotection on ganglion cells have not yet been fully resolved, it is likely that both A1 and A3 receptors share a pathway as both receptors activate PTX sensitive Gi/Go proteins and can activate Gbg (Schulte and Fredholm, 2002).

The propensity of ATP to transduce mechanical stimuli into neurochemical signals, combined with the lethal eVects of the P2X7 receptor and the protective eVects of adenosine, suggest that ATP and the P2X7 receptor could provide a link between elevated IOP and the death of retinal ganglion cells. These observations also suggest that conversion of extracellular ATP into adenosine could simultaneously remove a toxic agent and produce a protective one. Upregulation of nucleotidases in response to released ATP may thus be a key adaptation that breaks the link between elevated IOP and cell death. It may also provide an entry point for intervention in the treatment of glaucoma.

V. CONCLUSION

Multiple factors are likely to compromise the function of retinal ganglion cells in glaucoma and eventually lead to their death. In so far as IOP contributes to this pathology, the regulation of aqueous humor dynamics provides the major approach currently used to protect retinal ganglion cells.