524

et al., 2000) might represent such a common phenotype.

Retinal ischemia related glaucoma

There is a long history implicating a relationship between ischemia and glaucoma (Posner, 1958). Ocular ischemia can result from a wide variety of systemic (Kaiser et al., 1993; Flammer, 1994; Stroman et al., 1995; Waldmann et al., 1996; Osborne et al., 2004) or local influences, such as trauma, which may then lead to glaucoma, or aggravate an existing glaucomatous condition. Kaiser et al. (1993) noted, for example, that silent myocardial ischemia was correlated with normaltension glaucoma. The particular forms of ischemia that might be relevant to glaucoma require further investigation and documentation. It is known that high IOP-induced ischemia in rats lead to increased levels of intraretinal glutamate (Nucci et al., 2005). This particular experimental paradigm preferentially affects RGCs (Osborne et al., 1999) and may therefore be suitable for studies of glaucoma. It is also of interest to note in this regard that the RGCs of glaucomatous rats with chronic, moderately elevated IOP, are more susceptible to ischemia reperfusion injury (Kawai et al., 2001).

Excitotoxicity and the retina

We have known for 50 years that an excess of the excitatory transmitter, glutamate, can cause cell death in inner layers of the retina (Lucas and Newhouse, 1957). In hindsight, this is an interesting historical parallel to those studies, which have argued that ischemia as an important risk factor for glaucoma. However, the prospective role of glutamate in glaucoma remains controversial, and recent studies have even questioned the sensitivity of RGCs to glutamate meditated excitotoxicity (Ullian et al., 2004). The sensitivity of RGCs to glutamate receptor agonists may be variable and highly dependent on the experimental conditions. For example, RGC death induced by 20 nmol NMDA is enhanced by the addition of glycine, D-serine, or a competitive glycine transporter-1 inhibitor, sarcosine. Thus, the severity of

excitotoxic retinal damage depends on the levels of both glycine and D-serine (Hama et al., 2006). With respect to neurotrophins, it is well established that BDNF can also have profound effects on NMDA receptor currents which may vary depending on the presence of the TrkB and/or the pan-neurotrophin receptor p75 (NTR) (Sandoval et al., 2007). The relationship between the NMDA receptor and BDNF is of particular interest with respect to glaucoma in view of the current clinical trials that are using the NMDA receptor antagonist, memantine (1-amino-3,5-dimethyl-adaman- tane) to treat glaucomatous patients. Recently, this NMDA receptor antagonist has been shown to upregulate the expression of BDNF in the brain (Meisner et al., 2008). Therefore, it is critical to explore further the signal transduction events and to determine if this is relevant to the RGCs.

Signal transduction

NMDA receptor antagonists and CaMKII

CaMKII is known to be present in the retina (Bronstein et al., 1988a, b, 1989, 1993; Ochiishi et al., 1994; Cooper et al., 1995). The CaMKII is present in amacrine cells of the inner nuclear layer (INL) and displaced amacrine cells and ganglion cells in the ganglion cell layer (GCL) (Ochiishi et al., 1994). The specific effect of NMDA on CaMKII in the retina is known (Laabich et al., 2000). The NMDA subtype of glutamate receptor is also a known substrate of CaMKII (Chen and Huang, 1992; Kitamura et al., 1993) and the CaMKII-mediated phosphorylation of glutamate receptors leads to a positive modulation of receptor function and maintenance of synaptic excitability in other neural systems (Fukunaga et al., 1992; McGlade-McCulloh et al., 1993; Tan et al., 1994). However, NMDA is also a known excitotoxin, and its neurotoxicity in the retina has been demonstrated (Siliprandi et al., 1991; Sabel et al., 1995). The early biochemical feature of NMDA-induced excitotoxicity in neurons is the disturbance in ionic balance triggered by calcium and sodium influx through the NMDA receptor/ channel complex (Choi, 1988; Marcoux et al.,

1990; Goldberg and Choi, 1993; Terashima et al., 1994). One consequence of such ionic imbalance is the activation/overactivation of many vital cellular enzymes including protein kinases and phosphatases, phospholipases, and proteases. This is said to lead to a cascade of both biochemical and physical changes (cytoskeletal breakdown) leading to neuronal death (Saido et al., 1994). RGCs express multiple subtypes of ionotropic and metabotropic glutamate receptors (HamassakiBritto et al., 1993; Hartveit et al., 1995), yet excitotoxic loss of RGCs is thought to result primarily from glutamate interacting with the NMDA receptor subtype. Administration of NMDA receptor antagonists is an effective method of preventing RGCs loss in various models of RGC death in which excitotoxicity is implicated, such as intraocular injection of glutamate or NMDA (Vorverk et al., 1996; Laabich et al., 2000). If NMDA receptor antagonists similarly reduce RGC loss in models of chronic hypertension, then a role for excitotoxicity in glaucoma would be more acceptable. The first study to address this issue was by Chaudhary et al. (1998), who elevated IOP in rats by cauterizing the episcleral veins, and administered the prototypical, non-competitive NMDA receptor antagonist MK-801. They found a marked preservation of RGC numbers in animals treated with the

525

compound. However, MK-801 is unsuitable for use clinically because it blocks normal glutaminergic neurotransmission and would, therefore give rise to potentially severe adverse effects. Accordingly, other studies have made use of the non-competitive NMDA receptor antagonist, memantine. This compound has recently been shown to be clinically useful in the treatment of moderate to severe Alzheimer’s disease (Areosa et al., 2005). A phase III clinical trial is currently in progress to determine the effect of memantine on visual field loss in glaucoma patients. If memantine proves to be clinically useful, this will provide further support for the postulate that excitotoxicity plays a role in the pathogenesis of glaucoma. However, the mechanism by which NMDA receptor antagonists could attenuate glaucoma remains unclear. Therefore, there is an urgent need to study the NMDA-signal transduction pathway in RGCs.

Alteration of CaMKII-aB and neuroprotective effect of m-AIP in retinal neurons exposed to NMDA

Excessive activation of glutamate receptors mediates neuronal death, but the intracellular signaling pathways that mediate this type of neuronal death are only partly understood. We have demonstrated that the level of the CaMKII-aB transcript, which contains a nuclear localizing signal (Diagram 1), is

526

altered in retinal neurons exposed to NMDA, whereas the level of the CaMKII-a transcript is not affected by such treatment (Laabich et al., 2000). The CaMKII-aB gene has been cloned and sequenced from a variety of sources including the human (Li et al., 2001). There is a temporal correlation between the early appearance of the glutamate-stimulated CaMKII-aB transcript and the later appearance of apoptosis in the retina (Laabich and Cooper, 2000). In these latter studies, the terminal deoxyribonucleotidyl transferase (TdT)-mediated biotin-16-dUTP nick-end labeling (TUNEL) method was used to detect fragmented DNA in fixed tissue sections of rat retina. The TUNEL assay confirmed that cell death is observed in the inner nuclear and GCLs at 24 h following injection of 4 mM NMDA. Previous reports have shown these TUNEL-labeled cells to be ganglion cells and displaced amacrine cells in the GCL, and amacrine cells in the INL, especially cholinergic neurons (Sisk and Kuwabara, 1985; Siliprandi et al., 1991; Gavrielli et al., 1992).

It is suggested that NMDA-induced neuronal cell death is the result of a sustained increase in the intracellular concentration of Ca2+, and overactivation of vital Ca2+-dependent cellular enzymes such as CaMKII (Fukunaga and Soderling, 1990; Fukunaga et al., 1992). While the aB subunit of CaMKII was shown to be elevated following a single application of NMDA, this response is indicative of a temporal correlation but not necessarily a cause and effect relationship, between the level of such CaMKII transcripts and NMDAinduced cell death, in the retina (Laabich et al., 2000). The use of kinase inhibitors has helped to determine the role of CaMKII. KN-62, an inhibitor of CaMKII, does not provide a complete inhibition of NMDA-mediated neurotoxicity in vitro (Hajimohammadreza et al., 1995). This may be because KN-62 is a competitive inhibitor for calmodulin or that pre-existing CaMKII that is already auto-phosphorylated is not inhibited by this compound. Investigations with m-AIP (myristoylated autocamtide-2-related inhibitory peptide), a non-competitive inhibitor for CaMKII, and reported to be more specific than the KNseries of inhibitors, was evaluated for its putative neuroprotective effect on NMDA-mediated cell

death. It is known that m-AIP is specific, and that it completely inhibits CaMKII-mediated phosphorylation of an exogenous substrate in vitro, and does not affect cyclic AMP-dependent protein kinase, protein kinase C, or calmodulin-dependent protein kinase IV (Ishida et al., 1995, 1998).

The previously mentioned study (Laabich et al., 2000) showed an in vivo inhibition of CaMKII activity following an intravitreal injection of NMDA together with m-AIP. A concentration of 500 mM m-AIP resulted in a 52% reduction in CaMKII activity when compared with shamcontrols. This reduction in CaMKII activity below basal levels did not reach the zero-threshold, but it was clearly sufficient to inhibit the NMDAinduced cell death. Whereas, a lower concentration of m-AIP (100 mM) decreased the number of TUNEL-labeled cells in the GCL and INL, it did not completely block cell death. At this lower concentration, the NMDA-stimulated enzyme activity appeared to be somewhat reduced, but this reduction did not meet the test of significance, and presumably some residual NMDA-stimulated activity remained. These results compliment other reports of glutamate or NMDA-stimulated CaMKII activity (Fukunaga et al., 1992; Morioka et al., 1995) leading to neuronal cell death. The fact that the CaMKII enzyme activity can be blocked to at least the 50% basal level without causing cell death, indicates that either the increase in CaM- KII-aB level and/or its phosphorylation and/or an increase in the phosphorylation of cytoplasmic CaMKII-a are critical steps in the cell signaling machinery leading to cell death. This is further discussed below.

There is an interesting parallel application of protein kinase inhibitors which have been demonstrated to lower IOP in animal (rabbits and monkeys) studies (Tian et al., 1999; Honjo et al., 2001). Several laboratories have independently discovered that an inhibitor of a new class of kinases, the rho-associated coiled coil-forming kinase (ROCK), is efficacious in lowering IOP. The selective ROCK inhibitor, Y-27632, has been observed to increase outflow facility of aqueous humor in enucleated porcine eyes (Rao et al., 2001) and in the rabbit (Honjo et al., 2001; Waki et al., 2001). It induces change in cell shape of cultured

human trabecular meshwork (TM) and Schlemm’s canal cells and decreases actin stress fibers and myosin light-chain phosphorylation in these cells, which is associated with widening of the extracellular spaces in the TM, especially of the juxtacanalicular tissue (Honjo et al., 2001; Rao et al., 2001). It is not yet clear if any of these inhibitors affect RGC death or survival.

Caspase-3 activation in NMDA-induced retinal cell death and its inhibition by m-AIP

A number of gene-products may be modulators of neuronal apoptosis resulting from NMDA insult. Caspase-3, a key member of the ICE protease family is implicated in the pathway leading to apoptosis (Chen et al., 1998; Namura et al., 1998). It has been suggested that caspase-3 is activated in neuronal cells that undergo apoptosis but it is not activated during necrosis (Armstrong et al., 1997). In a previous report (Laabich and Cooper, 2000), caspase-3 was found to be expressed at low levels in the adult rat retina However, after 30 min and

527

2 h post-injection with NMDA (Fig. 1A), the expression of caspase-3 protein was markedly increased. Furthermore, the caspase-3p32 was proteolytically activated, and the caspase-3 active fragment, p17, was markedly increased in the retina. TUNEL-labeled cells started to become apparent at some point after 2 h, but prior to 24 h (Laabich and Cooper, 2000). The increase in caspase-3/p32 protein may be due to an increase in gene expression. The precursor caspase-3 and its activated product returned to baseline 24 h after the intraocular injection with NMDA.

These results demonstrating NMDA-activated caspase in the retina are consistent with other published reports. For example, caspase-3/p32 and the active form p17 were increased in hippocampal neurons following transient global ischemia (Chen et al., 1998). Also, caspase-3 is activated and cleaved in hippocampal neurons 30–60 min of exposure to glutamate (Mattson et al., 1998). Furthermore, caspase-3 protease activation was detected within a similar time frame following the induction of cerebral ischemia (Namura et al., 1998).

2 h postinjection / IDV

A

sham-control NMDA

caspase-3

B

β-actin

32 kDa

17 kDa

sham-control NMDA + AIP

β-actin 32 kDa

17 kDa

sham-control NMDA + AIP

32 kDa 17 kDa β-actin

caspase-3