Chapter 57 Neuroprotection Mechanisms in Glaucoma THOMAS MITTAG Table Of Contents |
The aim of neuroprotection for glaucoma therapy is to use agents that prevent
or delay retinal ganglion cell (RGC) death, as well as
to rescue and promote regeneration of already compromised RGCs. On the
basis of these objectives, “neuroprotection” can include
many classes of agent that protect neurons before onset of a noxious
influence, that rescue neurons already compromised, and that promote
the regeneration of axonal/dendritic connections to restore function. During
the past 15 years or so, neuroscientific research has greatly
increased our understanding of neurodegenerative processes. Much of
this knowledge has been obtained from studies in cell cultures and in
a few acute in vivo model systems, but we still lack good models to mimic
clinically important slow chronic neurodegenerative conditions. Glaucoma is in this latter category. Cultures of mature differentiated RGCs are difficult to maintain, even for short periods, so that most culture studies are done with immature cells from neonatal, or a few days postnatal, rodent retinas. Thus, it is uncertain to what extent RGC cultures reflect adult differentiated RGCs in vivo. The most common acute in vivo animal models that have been used for neuroprotection studies relevant to glaucoma include cutting or crushing of the optic nerve, acute retinal ischemia-reperfusion, and, to a lesser extent, intravitreal injection of an excitotoxic neurotransmitter or analog (e.g., glutamate, N-methyl-D-aspartate [NMDA], and kainic acid). Laser photodamage of the trabecular meshwork in the monkey eye is currently the best available chronic elevated intraocular pressure (IOP) glaucoma model in a primate, but it is seldom available for neuroprotection studies. Considerable progress has been made toward developing induced chronic elevated IOP glaucoma models in rodent eyes, and spontaneous development of elevated IOPs has recently been described in one strain of mouse1,2 (DBA/2) and in a rat.3 However, these elevated IOP models are not yet as well established as the acute rodent models of optic nerve crush or ocular ischemia in which most in vivo neuroprotection studies have been done. Although RGCs appear to be preferentially vulnerable to damage and death in all the rodent models cited, caution must be added about their extrapolation to clinical glaucoma because pathology affects all layers of the retina and is much more rapid and extensive than in primate eyes exposed to chronic elevated IOP. Justification for the optic nerve crush, ischemia, and excitotoxic models is based on theories for the initiation and development of retinal pathology in glaucoma. Each model is thought to mimic at least one component of glaucomatous optic neuropathy, and the evaluation of potential neuroprotective agents for glaucoma in models is based on this premise. A wide range of drug classes shows neuroprotective activity in RGC culture systems, but fewer have been tested and found active in the acute in vivo models of RGC death, and fewer still in chronic high-IOP models, which arguably have the closest relevance to glaucoma. None of these agents have yet reached clinical use for treating glaucoma at the level of the retina or optic nerve head, although one, memantine, is currently in a clinical trial. For the purposes of this volume, this chapter emphasizes the mechanisms leading to RGC death and neuroprotective studies done on in vivo model systems and excludes studies on regeneration. |
INITIATING FACTORS IN GLAUCOMA |
Glaucomatous neuropathy is likely a response to abnormal stressors (primary
risk factors) interacting with several mostly undefined
secondary factors, such as in some cases a genetic predisposition. It
is abundantly clear from numerous clinical and experimental studies that
the level of IOP is a primary risk factor for optic neuropathy in
mammalian retinas. A consistent observation is that a high IOP, however
caused in humans (e.g., trabecular outflow pathology in primary open-angle glaucoma or secondary
glaucomas), or caused by experimental means or that has a genetic
basis in vertebrate animals (rabbit, dog, mouse, rat, and monkey), results
in retinal damage. More recent large-scale human studies
have confirmed that the lower the IOP, the slower the progression
of glaucomatous damage will be, even for IOPs considerably below the
currently accepted range of “normal” IOP (15–18 mm
Hg for most people). Pressure may act as a direct primary stimulus
for apoptosis in neural cells,4 but it is the prevailing hypothesis that secondary signals cause RGCs
to become necrotic or apoptotic. In addition to elevated IOP there is
now also sufficient evidence from studies on patients with glaucoma with
IOPs in the normal range (normal tension glaucoma) to indicate
that chronic hemodynamic changes, such as those associated with
sleep apnea syndrome, and/or localized episodic vasospastic dysfunction
in the ophthalmic microcirculation may be another primary risk factor.5 This would be particularly damaging to RGCs and their axons located in
areas where vascularity is limited and the perfusion safety factor likely
to be small, namely, the retinal and optic nerve head/lamina
cribrosa microvasculatures. Focal dysfunction in the regulation of blood
flow by intrinsic (autoregulatory) or extrinsic mechanisms
at these locations could cause, respectively, ischemic stress to RGC
somata or to their unmyelinated axons in the optic nerve head. There
is some experimental evidence that excessive vasoconstriction by endothelin
may be a contributor to the optic nerve head vascular pathophysiology
in glaucoma.6 At the present time there is scant clinical evidence that dysfunction
of retinal glutamate/aspartate signaling resulting in excitotoxicity
is a primary risk factor in glaucoma. (However, a discussion of
the experimental evidence follows.) There may be an important interaction between vascular dysfunction and elevated IOP.7 One may hypothesize a defective regulation of optic nerve head or retinal blood flow in response to IOP compression of the microcirculation to be the predisposing factor for glaucomatous RGC loss. Subjects with normal microvascular regulation may be able to counteract moderate increases in IOP and thus avoid ischemic damage and glaucomatous retinal pathology (as in patients with ocular hypertension?). Some degree of vascular dysregulation may not be pathologic by itself but may become so with the added stress when IOPs are elevated above normal (as in primary open-angle glaucoma?), whereas a greater degree of vascular dysregulation could be pathologic even with normal IOP levels (as in normal-tension glaucoma?). Any hypothesis for the initiation and progression of glaucomatous neuropathy needs to account for the focal and chronic nature of RGC and nerve fiber layer (nfl) loss. In vascular dysfunction, inability of retinal microvessels in a localized area to autoregulate against IOP compression could lead to hypoperfusion and a focal ischemia that in turn triggers overriding autoregulatory mechanisms that temporarily restore perfusion (e.g., mediated by the localized rise in pCO2 level). This response cycle could result in repetitive transient episodes of ischemia-reperfusion over an extended period of time, causing a gradual focal loss of RGCs and thinning of the nfl leading to that location. A similar postulate can be offered for ischemic-reperfusion episodes occurring in the optic nerve head/lamina cribrosa region when subjected to compression by elevated IOP. Localization of the microvascular pathology to subregions of the optic nerve head could account for the observed focal pattern of RGC death and nfl thinning. These conjectures for a vascular pathology initiating glaucomatous optic neuropathy provide some justification for using ischemia-reperfusion models to develop neuroprotective agents for glaucoma. This vascular hypothesis includes both current general theories about where retinal glaucoma pathology actually begins, namely, axon pathology leading to RGC loss (termed primary, or retrograde cell death) or damage to or death of the ganglion cell body first, followed by axon degeneration (termed secondary, or anterograde death). As discussed subsequently, it is likely that both types of death processes occur in glaucoma. |
NEUROPROTECTION OF AXONS |
A commonly held notion suggests that initiation and progression of retinal
damage in IOP-induced primary open-angle glaucoma, and most secondary
glaucomas as well, occur on the unmyelinated axons, particularly at
the optic nerve head/lamina cribrosa. Elevated IOP is thought to
cause structural distortion of the optic nerve head, which alone, or
in conjunction with compression of the microvasculature, results in localized
ischemia. Structural distortion of the cribriform plates, ischemia, or
just the increased pressure itself may be directly deleterious
to unmyelinated nerve fibers and disrupt axoplasmic flow of neurotrophins
to RGCs. However, these stresses could also cause an activation
response in astrocytes or oligodendrocytes of the optic nerve head and
in the nfl of the retina. The activation of astrocytes results in physical
damage to axons by erosion or remodeling of the extracellular matrix (ecm),8 clinically seen as cupping of the optic nerve head, and additional axon
damage caused by the release of toxic mediators, such as tumor necrosis
factor (TNF)-α, interleukin (IL)-1β, and
excess nitric oxide.9 The end result of all these postulated cellular pathologic events is disruption
of a functional axonal connection between RGCs and their target
neurons in the optic tectum of the brain. Interruption of the anterograde
flow of information and materials, but more particularly of the
retrograde axoplasmic flow of neurotrophic proteins to RGC somata, is
thought to be a critical signal for RGC death. However, as noted previously, the
injured axon or its surrounding glial cells may also signal
RGC death. These postulates are the justification for optic nerve cut
or crush as a glaucoma model. On the basis of these hypotheses, potential
neuroprotective agents acting at the level of the optic nerve head
or the nfl can be targeted either to axons or to retinal glial cells (astrocytes, oligodendrocytes, Muller cells, microglia). A
better understanding of the pathophysiology of unmyelinated RGC axons
and of retinal glial cells is needed to develop such protective agents. Axon degeneration can be triggered by a variety of insults, especially physical damage and ischemia, and it appears to be an autodestructive process with similarity to apoptosis. Axonal injury causes the release of cytochrome C from mitochondria but does not appear to follow subsequent apoptotic pathway steps involving caspases,10 as diagrammed for the cell body in Figure 1. In normal mice transected axons distal to their soma fail to conduct an action potential within 1 day of the injury, but in the Wallerian degeneration-Slow (WldS) mouse mutant11 such axons remain structurally intact and can conduct action potentials for up to 3 weeks. It is likely that axon preservation in the WldS mouse applies also to the proximal part of a damaged optic nerve axon and may increase survival of the RGC. The WldS mutation causes overexpression of a chimeric protein composed of a component of the ubiquitin proteosome proteolytic pathway (UPP) and the NAD synthesizing enzyme nicotinamide mononucleotide adenyltransferase-1. Increased NAD synthesis seems to be an important axon-sparing component of the WldS phenotype.12 This recent finding suggests neuroprotective strategies aimed at increasing the cellular NAD supply, and it may explain earlier reports on the beneficial effect of nicotinamide treatment in several neurodegeneration models. However, the UPP may also play a role in the WldS phenotype because this polyfunctional proteolytic system degrades transcription factors and regulatory proteins involved in death signaling, including some that control the release of pro-apoptotic factors from mitochondria (see later discussion and Fig. 1). Further evidence for a clear difference in degenerative pathways for RGC axons and their somata is in a recent report13 on spontaneously glaucomatous DBA/2 mice that were also deficient for the pro-apoptotic protein BCL-2-associated X protein (BAX) (Fig. 1). The optic nerve axons degenerated in these mice without loss of the RGC, suggesting that BAX is essential for completion of the apoptotic program in the soma but is not required for axon degeneration. The damaging initial effect of ischemia on axons that has been most studied is a persistent depolarization that occurs through opening of Na+ and/or Ca2+ entry channels. Sodium channel blockade by some anesthetics, anticonvulsant drugs, and β-adrenergic receptor blockers can significantly reduce axon injury in such models. These agents include, for example, lidocaine, riluzole, flunarizine, betaxolol,14 diazepam, carbamazepine, and phenytoin. The last two agents listed were found to protect against ischemic injury in an in vitro acute rat optic nerve anoxia model at effective drug concentrations below those used to treat epilepsy.15 The wide chronic therapeutic use of these two agents thus provides an opportunity for a retrospective clinical study to assess whether Na+ channel blockade is of value in glaucoma therapy. More specific Ca2+ channel blockers, such as nifedipine and nimodipine, which are primarily used as vasodilators, gave marginally positive results when evaluated in a clinical study.16 The present status in glaucoma therapy of established channel blocking drugs that affect axons, alone or combined with a vascular effect, is that they may have limited value in specific instances,17 but their general use is probably unwarranted. Because of the multiplicity of neural Na+ and Ca2+ channel subtypes, and the state-dependent binding at multiple sites of potential blockers, we may not yet have identified the appropriate channel type or the kind of blocking agent needed for a more effective glaucoma therapeutic agent of this drug class. |
NEUROPROTECTION AIMED AT GLIAL CELLS |
The role of reactive astrocytes in the pathology at the optic nerve head
and the nfl in glaucomatous eyes is far from clear. The general stress/injury
cellular response in astrocytes, both in the optic nerve
head and retina, includes hypertrophy, hyperplasia, increased expression
of intermediate filament proteins (glial fibrillary acidic protein; vimentin), and
changes of ecm proteins, their metabolizing
enzymes, and many other proteins.18 A role for the increases in glial fibrillary acidic protein and vimentin
remains unknown, but the volume reduction, change in composition, and
remodeling or loss of ecm components caused by activated astrocytes
in the optic nerve head and nfl may result in loss of structural support
and a pathologic milieu for unmyelinated RGC axons as well as for
their somata. Some studies indicate that increased nitric oxide caused
by up-regulation of the nitric oxide synthase (NOS) isozyme
NOS-2 in reactive astrocytes at the optic nerve head might directly damage
nearby axons (see later discussion on nitric oxide). However, damage
to the ecm by activated glial cells beyond the optic nerve
head or nitric oxide in the retina itself may also directly affect
RGC axons and their somata.19 Both nfl astrocytes and Muller cells show induction of the ecm modulating
enzyme matrix metalloprotease (MMP)-9 in response to cytokines
like TNF-α. In the mouse optic nerve ligation model (which
combines an ischemia insult and optic nerve crush), there
is a loss of the ecm component laminin in the nfl/RGC layers of the
retina.20 The importance of this cell-matrix interaction for cell survival is indicated
by the finding that neurons grown on a laminin substrate exhibit
sustained anti-apoptotic signaling through the phosphoinositide 3-kinase (PI3-k)/Akt pathway (Fig. 1). In the mouse
optic nerve ligation model, loss of laminin and RGCs was associated
with increased MMP-9 activity, which did not occur in MMP-9 knockout
mice20 or when ecm modification was blocked by inhibition of plasminogen activators.21 These findings indicate that, as a consequence of glial activation, the
loss of appropriate interaction with the ecm might be an important signal
within the retina for initiating axon degeneration and apoptosis
in RGC. Similar findings of MMP activity up-regulation and laminin loss
in the nfl/RGC layers of the retina have very recently been found
in the spontaneous glaucoma model in the DBA/2 mouse.22 In contrast with causing damage, reactive glial cells in the retina could potentially have some beneficial effects on the basis of up-regulation of various survival/neurotrophic/growth factors, such as insulin-like growth factor (IGF)-1, IL-10, ciliary neurotrophic factor (CNTF), and members of the chaperone/heat stress proteins (HSPs), such as HSP-27.23 In a similar apparent paradox, nitric oxide and TNF-α can also have protective effects (these two factors are discussed further below). The foregoing discussion indicates several potential mechanisms that could be targeted for developing neuroprotective agents that interfere with destructive actions by activated astrocytes or other glial cells, namely, inhibitors of NOS-2 or enzymes that modify the ecm, and agents or gene therapy that promote local neurotrophin/survival factor synthesis and release to make up for the interruption of axonal supply of such factors from the optic tectum. |
REPLACING AXON-DERIVED SURVIVAL FACTORS |
The most important known consequence of optic nerve axon pathology is loss
of supply to RGCs of the brain-derived neurotrophic factor (BDNF) and
its receptor TrkB. Although BDNF/TrkB signaling is not
essential for RGC survival during development, it is required for maintenance
of RGC in adulthood and appears to be essential for the immediate
survival of RGCs after injury.24 However, several other neurotrophins, growth factors, or cytokines may
also be necessary for the long-term post-injury survival and functional
regeneration of RGCs such as glial cell line-derived neurotrophic factor (GDNF), fibroblast growth factor (FGF), CNTF, NT-4, and
IL-10, all of which can provide some degree of protection
after optic nerve damage.24 High doses of exogenous BDNF delay the death of RGCs in cell culture and
also in vivo in the rat after optic nerve damage,25 particularly when extracellular metabolism is prevented. These findings
on BDNF led to the recent promising study using retinal BDNF gene-therapy
in a rat model of chronic elevated IOP.26 There are some experimental indications that activation of receptors mediating neurotransmission can affect signaling by BDNF and other neurotrophins. Elevated levels of cyclic adenosine monophosphate (cAMP) increase the survival effect of neurotrophins in RGC cultures. As indicated in Figure 1 (green), BDNF receptor activation leads to an intracellular anti-apoptosis signal pathway involving the activation of PI3-k and the serine-threonine kinase Akt, resulting in inactivation of pro-apoptotic proteins, such as BAX or BCL-2-associated death protein (BAD), by phosphorylation (see subsequent discussion). As diagrammed in Figure 1, cAMP, acting through the cAMP response element binding protein-activated transcription factor, can potentiate the BDNF signal cascade either by blocking BAX/BAD pro-apoptotic signals or up-regulating anti-apoptotic BCL-2 expression.27 On this basis one might expect cAMP activators to be beneficial in optic nerve damage models, but there is as yet no evidence for this in vivo. Instead, contrary to this expectation, it has been found that a β-adrenergic antagonist, betaxolol, is beneficial in the ischemia model28 and also up-regulates mRNA for BDNF29 and other neurotrophic/growth factors.30 This response, and other mechanisms of some β-adrenergic antagonists, such as ion-channel blockade, suggests that their neuroprotective effect involves signal systems other than the β-adrenergic receptor/adenylyl cyclase system. Adult RGC survival and regeneration in vivo after optic nerve damage or ischemia can also be promoted by CNTF,31 β-FGF,32 and IGF-1.33 FGF can also be up-regulated in the retina by activation of other receptors, for example, by α2-adrenergic agonists,34 and this may account in part for the neuroprotective effect of this class of drug in the rat optic nerve crush, ischemia, and elevated IOP models.35–37 These findings on β-antagonists and α-agonists affecting growth factors hold out the possibility that conventional pharmaceutical agents acting through well-established receptor systems (such as those linked to cAMP), or have novel drug mechanisms that modulate neurotrophin/growth factor expression, may be a useful therapeutic approach to minimize loss of trophic support caused by damaged axons. For the near term, this approach deserves further investigation because it seems more clinically feasible than the promising but long-term approach of direct replacement of growth factors and neurotrophins by gene therapy. |
ENDOGENOUS NEUROPROTECTION MEDIATED BY PRECONDTIONING |
The normal rat RGC layer has some resident cells that express RNA for BDNF, which
becomes up-regulated and expressed in additional cells for
approximately 2 weeks after an optic nerve crush.38 Up-regulation in the retina itself of BDNF, other neurotrophins/growth
factors, and heat shock proteins thus appear to be components of
a general local protective response against intraocular injury. This important
endogenous protective response occurs with a variety of preconditioning
insults and is not yet understood for all types of insult. For
example, even a saline injection into the vitreous can increase local
neurotrophin/growth factor levels, and an injury to the lens is
effective in promoting survival of RGCs and even axon regeneration after
optic nerve crush in the rat.39,40 This protective injury phenomenon may extend to the primate eye and could account for the finding that laser-induced damage to a focal area of the outer retina in monkeys specifically preserved the RGCs located over the injury area in comparison with RGC loss in areas of undamaged outer retina when glaucoma was subsequently induced by elevated IOP (by laser-scarring of the trabecular meshwork).41 Other examples of this phenomenon include ischemic preconditioning of the retina42 and heat stress, which attenuate RGC death in response to a more severe ischemic episode, excitotoxicity, or optic nerve crush. A mild inflammation of the retina caused by immunization with a peptide derived from a retina-specific protein (interphotoreceptor retinal binding protein) protects RGC against optic nerve crush injury.43 Such a protective immune response can also be elicited in these acute models and in a rat glaucoma model of elevated IOP by immunization with a synthetic peptide capable of targeted auto-antigen mimicry, such as COP-144 (a random copolymer of tyrosine, alanine, lysine, and glutamic acid in specific proportions). The protective action of agents such as COP-1 may be quite general, as is heat stress, and appears to be effective against autoimmune conditions including experimental autoimmune uveitis.45 The protective immune response by agents such as COP-1 appears to be T-cell dependent and may involve up-regulation of BDNF.46 Ischemic and heat shock preconditioning are the best understood cellular mechanisms for endogenous neuroprotection. Responses to hypoxia are mediated by hypoxia-inducible factors (HIF-1,2) that promote the transcription of a specific set of genes with hypoxia-response elements in their regulatory domain. The expression of HIFs can be up-regulated by cytokines, such as IGF, but their activation is controlled posttranslationally. Under normal conditions (normoxia) the oxygen-dependent enzymatic hydroxylation of the low constitutive level of HIF-1 targets the protein for rapid degradation through the UPP, resulting in a half-life of only a few minutes in the cytoplasm. However, during hypoxic stress the UPP-mediated destruction is attenuated and the HIFs translocate to the nucleus. Genes that are up-regulated by HIFs include those required by cells to adapt to decreased oxygen by shifting their metabolism to glycolysis; genes for erythropoietin, which is neuroprotective in many model systems including RGC after optic nerve axotomy47; and genes for vascular endothelial growth factor, which is involved in neovascularization that is often a consequence of ischemia.48 Heat stress in particular (but other stresses as well) induces a set of protective heat shock proteins in most cell types termed HSPs, which are members of the larger chaperonin family. Constitutive expression of HSPs in retinal neurons and glia is enhanced in glaucomatous primate retinas, especially in the RGC and nfl layers,49 but also in the optic nerve head.50 Induction of some HSP proteins by systemic treatment with Zn2+ or geranylgeranylacetone is reported to be neuroprotective of RGC in a high IOP glaucoma model in the rat,51,52 as is HSP-27 applied to the eye by electroporation in the rat ischemia-reperfusion model.53 The expression of HSPs is controlled in part by a transcription factor, HSF-1. Similar to the HIF described previously, HSF-1 does not translocate to the nucleus in the absence of a stress. A new agent, bimoclomol, has recently been described that binds to HSF-1, prolonging its activation, DNA-binding, and transcriptional activity.54 Bimoclomol is a hydroxylamine derivative and the co-inducing or potentiating effect of stress-induced HSP expression by this class of drug has already shown promise in one model of severe neurodegeneration.55 Agents, such as bimoclomol, that act on stress-activated transcription of protective genes represent an important lead-compound toward the goal of harnessing endogenous neuroprotection for future therapies. This endogenous response is likely to be more effective than the mostly single-action agents currently being studied because multiple protective pathways are activated. |
NEUROPROTECTION OF RETINAL GANGLION CELLS |
Investigations of the time course and location of RGC death in the optic nerve crush and ischemia-reperfusion models in the rat and partial transection of the optic nerve in the monkey56 have led to the idea that there is a two-stage loss of RGCs in glaucoma; a primary retrograde cell death consequent to axon pathology, followed by more extensive anterograde damage causing apoptosis of RGCs even though their axons may initially be undamaged. This secondary anterograde death is proposed to be caused by local toxic mediators released in the ganglion cell layer of the retina by other cells (astrocytes, Muller cells), or factors resulting from the primary death of RGCs that induce apoptosis in neighboring healthy RGCs. At present, no direct evidence confirms that secondary or collateral killing of RGCs occurs in clinical glaucoma. However, the observations that glaucoma progression is more often by enlargement of visual field defects than by development of new foci suggest that more RGCs may be lost by secondary mediators generated in the retina than by a primary insult to axons. |
NEUROPROTECTION AIMED AT INTRACELLULAR APOPTOSIS SIGNALING |
The predominant mode of death of RGCs in the acute glaucoma model systems
is through a genetic preprogrammed cell death mechanism termed apoptosis. Studies in primate glaucoma, in which RGC loss has a much slower rate, suggest
that the loss also occurs by apoptotic death.57 In neural model systems, ischemic/hypoxic insults appear to activate
apoptotic signaling pathways different from those activated by trophic
insufficiency or by excitotoxin exposure, and there appear to be alternative
default pathways for some insults, such as excitotoxicity. Thus, the
possibility of developing the most effective therapeutic agent
targeted to intracellular apoptotic signals in RGCs ultimately depends
on knowledge of the nature of the initiating signals to the cell as
well as the intracellular signal cascade involved in the neurodegeneration. Apoptotic pathways involve complex multiply linked signaling steps that can be subdivided into three phases: initiation, decision, and degradation. Figure 1 is a schematic representation of these phases that depicts some of the known signal pathways and proteins, identified mostly from studies on cultured neurons, and it includes both pro-apoptotic and anti-apoptotic pathways. Studies on cell culture systems have also established the proof-of-principle that blocking one or more steps in pro-apoptotic signaling can delay the demise of a cell if not prevent it. It is evident that the earlier the pro-apoptotic signal cascade is interrupted the greater the likelihood will be of reversing it and thus preventing cell death. At present, most agents studied in model systems to establish proof-of-principle have been targeted to the post-mitochondrial degradation signals. These include drugs that are caspase inhibitors, inhibitors of caspase-activated DNAses, and inhibitors of endonucleases (Fig. 1). These agents have been shown to delay neural cell death in model systems, but there is some question whether blocking components mediating the late decisional and degradative phases may not be too late a step for a clinically useful drug to prevent, as opposed to delay, the death of that cell. However, on the basis of the emerging role of activated caspases generating toxic peptides from normal neuronal proteins such as amyloid precursor protein (APP), blockade of caspases and other proteases associated with apoptosis could be an important way to prevent signals from dying RGC causing secondary RGC death (see later discussion on APP). Mitochondria are most often viewed for their primary metabolic role of oxidative phosphorylation, but they also contribute to the control of intracellular Ca2+ levels and generate superoxide, both of which can influence the survival of neurons. However, their role in apoptosis signaling is emerging as a major target area for neuroprotection. There is an increasing body of evidence that mitochondria play a key decisional role in apoptosis initiated by the insults of ischemia, excitotoxins, and trophic insufficiency (shown in red) thought to be responsible for RGC death in glaucoma.58 The scheme in Figure 1 illustrates a pro-apoptotic pathway for trophic insufficiency (shown in yellow), acting through the enzyme sphingomyelinase, ceramide, and a phosphatase, which interrupts the anti-apoptotic signal pathway for BDNF and other trophic/growth factors (shown in green) by preventing activation of Akt, a key step in the neurotrophin anti-apoptotic signal pathway. The active (phosphorylated) form of Akt, which is itself a kinase, converts a pro-apoptotic protein such as BAD or BAX into phosphorylated form (P-BAD/BAX), thus inactivating it. Also shown is a pathway for local ischemia and overactivation of glutamate receptors (ionotropic glutamate receptor [iGluR]) acting through p53 and pro-apoptotic protein BAX, or related proteins such as BIM. These pathways also affect the anti-apoptotic interaction of BCL-2–like proteins that have specific binding sites on mitochondria. (Not shown is a more direct alternative pathway for iGluR overactivation and ischemia, which by induced cell depolarization, elevation of intracellular Ca2+ levels, and metabolic stress may directly cause mitochondrial depolarization and activate mitochondrial apoptotic signaling.) Thus, the major noxious initiating stimuli thought to be relevant to glaucoma (shown in red) seem to converge on the mitochondrion. APOPTOSIS SIGNALING BY MITOCHONDRIA The mitochondrion is a double-membrane organelle with the components for oxidative phosphorylation (e.g., the electron transport chain, cytochrome C, adenosine triphosphate (ATP) synthase, and adenine nucleotide transporter) located within or associated with the inner membrane. Energy derived from the electron transport chain pumps protons and other ions from the mitochondrial matrix across the inner membrane, maintaining a pH gradient and an electrochemical potential difference, termed the mitochondrial potential (MP). The recent development of potentiometric dyes that accumulate in and provide a measure of MP has shown that metabolic stress as well as noxious stimuli, such as trophic withdrawal, causes a decrease in MP that precedes apoptotic degeneration of a cell. The decrease in MP under metabolic stress, which depletes ATP, is caused in part by closure of ATP-activated K+ channels. As important as metabolism to maintain MP is the regulation of MP by a large complex of proteins that associate together at contact points of the mitochondrial inner and outer membranes, termed the mitochondrial permeability transition pore complex (PTPC). The PTPC normally functions as a regulated pore that can effect highly selective transfer of small molecules, including small proteins, between the matrix (the mitochondrial intermembrane space) and the host cell's cytoplasm. However, a sustained abnormal nonselective high-conductance opening of the pore results in dissipation of MP and is a critical event in the release of pro-apoptotic degradative signaling proteins. In the rat chronic elevated IOP glaucoma model, a significant downward shift in the distribution of MP level in RGC mitochondria was found compared with the retina of the contralateral normotensive eye.59 This preliminary result indicates that conclusions about the major role of mitochondria in apoptotic cell death obtained with cultured neurons may apply to apoptosis of RGCs in glaucoma. The release of signaling molecules from the mitochondria (pink, Fig. 1) includes cytochrome C and other caspase activators (Apaf-1), procaspase 9, and activators of the endonucleases that degrade DNA (apoptosis inducing factor [AIF]). Several ligands have been found that maintain regulation of the PTPC in the molecule-selective and low-conductive or closed position, and prevent the release of apoptotic degradative signal proteins. These drugs include cyclosporine and its analogue FK506, which have anti-apoptotic activity in neural culture systems. Systemic treatment with FK506 is somewhat effective in reducing RGC loss in the in vivo optic nerve crush model in the rat.60 Another agent that maintains MP, but by activating ATP-regulated K+ channels (gabapentin-lactam), has also recently been shown to prolong RGC survival in culture.61 Proof-of-principle experiments have yet to be done on chronic elevated IOP models to determine whether MP stabilizing agents such as cyclosporine, FK506, or K (ATP) channel activators can prevent or delay RGC death in these clinically more relevant cases. BCL-FAMILY PROTEINS IN APOPTOSIS Members of the BCL family of proteins are the major participants in regulation of PTPC opening. The association of the anti-apoptotic protein BCL-2 and/or BCL-XL with the PTPC appears to facilitate closure and/or to prevent the conformational change of the pore into the high-conductance state, thus maintaining the MP and blocking release of apoptotic signal proteins. Conversely, PTPC binding of another member of this protein family, BAX dimer, increases nonselective mitochondrial permeability and is pro-apoptotic. As indicated in Figure 1, up-regulation of BAX, and/or a related protein BIM,62 appears to be a major signal for apoptosis initiated by lack of neurotrophic anti-apoptotic signaling in the retina or by excitotoxicity. A fifth member of this family, BAD, can bind to BCL-2/BCL-XL and deactivate its antiapoptotic action in mitochondria. However, when BAD is phosphorylated through the neurotrophin receptor/PI3–kinase/Akt signal pathway (to P-BAD), its ability to block the anti-apoptotic actions of BCL-2/BCL-XL is removed. These findings from studies in neural cultures indicate that members of the BCL protein family function as essential “gatekeepers” of PTPC pore opening and can control the decisional phase of apoptosis. Almost all neural cells in the rat retina express BCL-XL and some also express BCL-2.63 BAX and BIM are also expressed and up-regulated in RGCs after optic nerve lesions.64 Studies done mostly in mice have established the important role of some members of the BCL protein family in RGC survival/death signaling. The effect of gene knockout and overexpression of BCL-2 and for BAX have been studied in two types of RGC apoptosis: developmental programmed cell death (PCD) in the neonate and pathologic cell death after optic nerve axotomy, ischemia, or NMDA excitotoxity.25 Developmental PCD in the mouse neonate is complete by 10 to 14 days after birth with more than 60% loss of RGC. BCL-2 deficiency had no effect on the rate and extent of RGC death in developmental PCD or after optic nerve axotomy in the adult. However, adult BCL-2 null mutants had 29% fewer RGC axons65 and showed no change in the 10% to 15% of RGC surviving axotomy,66 suggesting one minority adult subpopulation of RGC may require BCL-2 and a smaller subpopulation that is independent of it for long-term survival. In contrast, in mice overexpressing BCL-2 90% of the RGC in the newborn are maintained through the developmental PCD phase into adulthood. Axotomy in adult mice overexpressing BCL-2 also showed a major difference; 65% of RGC showed long-term survival compared with approximately 10% in normal mice.67 With regard to BAX deficiency, developmental PCD was largely blocked with more than double the normal number of RGC surviving in the adult. BAX deficiency also prevented RGC death caused by optic nerve crush,68 as well as in the glaucoma model with chronic high IOP in the DBA/2J mouse,13 but did not protect against NMDA-mediated excitotoxic death. BAX overexpression might be expected to potentiate apoptosis, but surprisingly was found not to affect developmental PCD of RGC or change the extent of RGC loss caused by optic nerve crush, axotomy, or excitotoxicity. Overall these findings emphasize the important role that BCL-family proteins play in survival/death signaling in RGC and indicate the complexity of their interactions in apoptosis. The role of these proteins appears to differ in the apoptosis pathways for neonatal RGC in developmental PCD and for apoptotic RGC death in the adult caused by injury or other insult. Also, different experimental damaging stimuli appear to activate different pathways of cell death. In particular, there is a significant difference in pathways thought to be of major importance in glaucoma, namely, between optic nerve damage or ischemia, and experimental excitotoxicity with NMDA. As indicated in Figure 1, one of the pathways for excitotoxicity by overactivity of iGluR and ischemic stress in the retina involves the p53 tumor suppressor protein, which is a transcription factor. The p53 protein is not required for developmental PCD, but in ischemia or oxyradical exposure, p53 expression increases in neurons.69 Mice deficient in p53 are reported to be resistant to kainate neuronal damage,70 but NMDA excitotoxicity in the retina of such mice still proceeds, but by an alternative pathway.71 Up-regulation of p53 increases expression of a number of genes involved in apoptosis, for example, BAX (see earlier discussion) and especially glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Fig. 1). GAPDH has important metabolic functions and converts NAD+ to NADH, but it also has other activities, in particular a role in apoptosis signaling that is not completely understood.72 GAPDH is the major source of NADH to support mitochondrial oxidative phosphorylation. It occurs in the neural cytoplasm as a proenzyme in the form of a tetramer bound to RNA. An increase in cytoplasmic NAD+/NADH ratio signals metabolic demand and frees the inactive bound form to enzymatically active GAPDH. However, modification of GAPDH, for example, by oxidation or nitrosylation by nitric oxide, releases it from RNA binding in an enzymatically inactive form that interrupts the NADH supply. Thus the signaling by this non-enzyme form of GAPDH to participate in apoptosis is mediated in part by metabolic stress to mitochondria caused by NADH deficiency (e.g., interruption of electron transport, oxidative phosphorylation, and loss of MP). This signaling can be initiated by ischemia or elevated levels of oxyradicals and/or nitric oxide. Additional properties of enzymatically inactive GAPDH in neurons undergoing apoptosis is interference with the anti-apoptotic BCL-2–like proteins indicated in Figure 1, and nuclear accumulation. The role of localization of GAPDH to the nucleus as a possible transcription factor for proteins involved in apoptotic signaling is not known. Deprenyl and related propargylamines, which were originally developed as inhibitors of monamine oxidase enzymes, have anti-apoptotic properties related to their interaction with GAPDH. The mechanism of this activity is thought to be that these drugs bind to GAPDH, which causes dissociation of the native tetramer into dimers. The dimeric forms of GAPDH remain active in glycolysis and NADPH synthesis, but it appears that their capacity to signal apoptosis and for nuclear binding is prevented by deprenyl binding. Whatever the mechanism for anti-apoptotic activity, the finding that deprenyl increases survival of RGCs in the rat optic nerve crush model may indicate a future role for propargylamines in glaucoma therapy.73 |
MEDIATORS OF SECONDARY RETINAL GANGLION CELL DEATH |
GLUTAMATE EXCITOTOXICITY An obvious candidate for a toxic endogenous secondary mediator is excessive glutamate—the major neurotransmitter input to RGCs.74 Prolonged exposure to a high level of glutamate has been found to be toxic to virtually all neurons and results from the persistent activation of glutamate receptor-coupled ion channels. This causes prolonged depolarization of the cell by Na+ entry/K+ efflux through the open glutamate receptor-coupled channels, as well as excessive Ca2+ entry into the cell. Excitotoxic death of RGC uses the p53-dependent apoptotic pathway shown in Figure 1, but not exclusively so.71 Among the multiplicity of such glutamate receptors, the NMDA and kainate subtypes seem to be the most important in causing toxicity in the retina, but these excitotoxins kill other retinal neurons in addition to RGC. Each NMDA/kainate receptor complex has multiple sites for binding of ligands besides glutamate. These include an accessory binding site for glycine or D-serine acting as a co-agonist, a modulatory site that binds polyamines, and binding sites within the channel for drugs such as MK801 or memantine. In addition to these ligand-binding sites, the receptor/channel activity is controlled by divalent cations (Mg2+, Zn2+), by the redox state or nitrosylation of proteins composing the channel, and by phosphorylation/dephosphorylation sites regulated by protein kinase/phosphoprotein phosphatase enzymes. All of these are potential sites for drugs that might be useful to control excitotoxicity if it is definitively shown to be involved in glaucomatous retinal pathology. Several such agents that have been tested for proof-of-principle in model systems may have application to block or delay RGC death, and one of them (memantine) is in clinical trial. The loss of ionic homeostasis by prolonged iGluR (Fig. 1) overactivation leads to depletion of ATP, which is consumed in the effort to restore ionic balance, and to a severe metabolic stress to the cell. Metabolic stress and Ca2+ overload, whether caused by glutamate/aspartate excitotoxicity or by ischemia, can initiate production of additional toxic mediators as well as activate intracellular mitochondrial apoptotic signal transduction pathways (as discussed previously). A specific example of this is the association of NMDA receptors with postsynaptic density protein-95, which couples NMDA receptor activation to potential nitric oxide toxicity by acting as a cosignal with Ca2+ for activation of neuronal NOS.75 This may account for the finding that knockout mice that do not express neuronal NOS were found resistant to RGC death by NMDA excitotoxicity.76 This indicates that it is actually nitric oxide that mediates experimental excitotoxic death of RGC in the mouse retina (nitric oxide toxic and protective actions are discussed further next). The evidence that glutamate excitotoxicity could play a role in clinical glaucoma and anterograde death of RGCs is circumstantial and also controversial. Elevated IOP in the eyes of some animal models decreases glutamate transporters77 and in some subjects with glaucoma is associated with increased glutamate levels relative to other amino acids in the vitreous.78 Elevation of glutamate levels was also found in the aqueous humor after optic nerve crush in the rat.79 Antagonists of iGluR, particularly NMDA receptor blockers, such as MK801 and memantine, appear to reduce the loss of RGCs in rodent optic nerve crush, ischemia-reperfusion, and chronic high IOP models.80–82 Despite these promising results from model systems, many significant questions relating to a role for glutamate excitotoxicity in clinical glaucoma remain. For example, it is unclear whether the modest elevations of vitreous glutamate reported in some studies reflect toxic concentrations at the level of RGCs or whether this is merely an epiphenomenon resulting from a stress-induced minor perturbation or adjustment of the rapid turnover rate of the glutamate/glutamine system mediating normal visual function. If one accepts that excitotoxic levels of glutamate do in fact cause secondary RGC death in the glaucomatous retina, this raises two important questions. The first is the cause for the excessive glutamate level, that is, whether there is an abnormally high release from nerve terminals synapsing on RGCs or whether the release is at normal levels, but the uptake and metabolism of glutamate by cells surrounding RGCs are impaired. If either of these suppositions is correct, then it implies that glaucoma initiates pathology in retinal cells other than RGCs, either excessive glutamate signaling by neurons such as the bipolar cells and/or especially in glial cells that take up and metabolize glutamate, for example, in Muller cells. Muller cells respond to almost any retinal stress by changes in gene expression and functional activities possibly involved in glutamate metabolism, but the significance of this in glaucoma is still largely unknown. Neuronally released glutamate is cleared through high-affinity glutamate transporters by rapid uptake, mainly into glial cells and their processes that surround every glutamatergic synapse, as Muller cells do with respect to RGCs, but also into neurons. The transporters are activated by extracellular glutamate and use the electrochemical gradient of Na+ as the driving force for uptake, resulting in tight coupling of the uptake of one glutamate with approximately three Na+ ions and the efflux of one K+ ion. Glial cells maintain this ability to rapidly terminate glutamatergic transmission by metabolizing the glutamate to glutamine and restoring the Na+ and K+ gradients through the Na+/K+-ATPase; enzyme activities that together consume the two ATP molecules provided by glycolysis of one glucose molecule to lactate. Thus, any impairment of glucose uptake or metabolism that depletes ATP in glial cells, such as ischemia, has the potential to facilitate excitotoxicity to neurons at normal levels of glutamate neurotransmission by decreasing the efficiency of glutamate uptake, thus raising and prolonging its extracellular level. The possibility that elevated endogenous glutamate can cause excitotoxic loss of RGC has been demonstrated in the rat eye by intravitreal injection of inhibitors of the transporter.83 Additional support for this may be a study done in the primate eye,41 which showed that RGCs in glaucomatous monkey eyes were preserved in focal areas where visual input to the RGCs was removed by prior laser ablation of the photoreceptors in that area. One interpretation of this finding is that removing the source of glutamate neurotransmission to RGCs prevents RGC death that results from chronic high IOP. However an alternative interpretation discussed previously is that the laser damage induces a local endogenous protective response analogous to that for other forms of preconditioning. A second question arising from the hypothesis that glutamate excitotoxicity causes a secondary apoptotic death in apparently normal RGCs that are unaffected by the primary pathology relates to whether such cells are in fact normal or have an abnormal sensitivity to glutamate excitotoxicity. The latter supposition implies that RGCs undergoing the primary glaucomatous pathology or their associated glial cells may release diffusible mediators that make neighboring RGCs susceptible to damage by normal glutamate levels. Such nonprotein mediators could include peroxynitrite formed from nitric oxide, or high levels of reactive oxyradicals resulting from the mitochondrial response to metabolic stress (see earlier discussions). PROTEIN MEDIATORS OF RETINAL GANGLION CELL DEATH Toxic mediators of secondary RGC death can also be derived from endogenous proteins or peptides. One such factor secreted by injured RGC has recently been identified as semaphorin-3, which was thought to function only in development as a chemorepellent guidance molecule involved in directing growing axons to their target. Semaphorin-3 is up-regulated in the RGC layer in ocular hypertensive rabbits84 and is transiently highly induced throughout the rat retina shortly after optic nerve transection. A peptide derived from semaphorin-3A causes more than 50% loss of RGC when injected into normal rat eyes, and the loss of RGCs after optic nerve axotomy can be completely prevented by intravitreal injection of a semaphorin-3 neutralizing antibody.85 Clinical and experimental studies indicate that β-amyloid might also play a role in RGC death, as is thought to be the case in the death of neurons in Alzheimer's disease.86 The β-amyloid peptide is produced from a normal neuronal precursor protein, APP, which is expressed in RGCs.87 The β-amyloid peptide can be generated by specific cleavage of APP by the protease β-secretase (also termed presenilin-1) and further cleaved to additional pro-apoptotic peptides by activated caspases—the same proteases involved in apoptotic degradation. After optic nerve section or chronic high IOP in the rat, RGCs undergoing apoptosis show activation of caspase-3 and/or 8 and also generate the diffusible β-amyloid that is itself a potent inducer of secondary neuron death.88 Some APP is also expressed in Muller cells that stain for presenilin-1 as well.89 Therefore, stressed Muller cells could also be a source of β-amyloid in a glaucomatous retina. Stressed retinas express additional extracellular protein mediators that can cause secondary apoptotic cell death. TNF-α is a signal protein of this type (Fig. 1) that like nitric oxide can have both destructive and protective actions (see below). The preceding brief description of some of the possible mediators of secondary RGC death indicates various drug classes that may have potential therapeutic value for glaucoma. These would include glutamate receptor antagonists or agonists at inhibitory γ-aminobutyric acid (GABA) receptors to counteract excitotoxicity; NOS inhibitors; antioxidants and free radical traps to minimize damage by oxyradicals; and protease inhibitors to block generation of neurotoxic peptides. MEDIATORS WITH DESTRUCTIVE AND PROTECTIVE ACTIONS TNF-α is an important cytokine present in many pathologic conditions, and it may play several roles in the glaucomatous retina. TNF-α is produced mainly by glial cells, particularly when stress-activated, and has two known receptors: TNF-R1 directly associated with pro-apoptotic signaling and TNF-R2 associated with anti-apoptotic activity. In addition, TNF-α can activate cells to release further potentially neurotoxic substances, such as nitric oxide, in the human glaucomatous optic nerve head.90 Up-regulation of TNF-α in the glial cells of the inner retinal layer and of TNF-R1 in the RGC was found in glaucomatous human retinas,91 suggesting that TNF-α may be involved in secondary death of RGC by acting directly on the cell body.92 Figure 1 indicates the direct pro-apoptotic pathway for TNF-R1 and caspase-8, which does not involve mitochondrial signaling. However, in another experimental study TNF-α was reported to prevent secondary death of RGC after axotomy of the optic nerve in vivo.93 Experiments on mixed glial/RGC cultures showed that TNF-α decreased cellular outward K+ currents, which was associated with increased anti-apoptosis signaling by the PI3-k/Akt pathway in the RGC. These findings suggest neuroprotection of RGC by TNF-R2 receptors. The availability of mice deficient in TNF-α, or either of its receptors, has added further complexity to the role(s) of this cytokine in the retina. In the mouse ischemia-reperfusion model, TNF-α deficiency did not affect the extent of RGC loss, whereas absence of TNF-R1 greatly reduced loss.94 A similar decrease in RGC loss was also found for the optic nerve crush model in TNF-R1–deficient mice.95 Absence of TNF-R2 led to enhancement of RGC loss. These experimental findings on TNF-α leave open the question on the role this cytokine plays at the RGC level in clinical glaucoma. As discussed previously, nitric oxide (NO) is thought to be destructive in the glaucomatous retina, but it also has the potential to be neuroprotective at the RGC level. On the one hand it appears to participate in RGC death caused by excitotoxic activation of NMDA receptors, but it can also block excitotoxicity by S-nitrosylation of iGluR receptors. Other experiments indicate additional neuroprotective actions of NO,96 including inactivation of caspases, protein kinase G activation of PI3-k, and increases in expression of genes for protective proteins, heme oxygenase, BCL-2, and HSP-70. Understanding the role(s) of NO in glaucoma is further complicated by a recent report on DBA/2 glaucomatous mice in which the absence of NOS-2 gene expression did not affect the progression of glaucomatous damage to the optic nerve.97 On the other hand NOS-2 inhibitors are reported to protect RGC in the chronic high IOP and ischemia models in the rat.98 These conflicting findings might be explained by the hypotheses that in the absence of NOS-2 the death of RGC in glaucoma proceeds by an alternative pathway or that NOS-2 inhibitors act by mechanisms other than NOS-2 inhibition. The immune system can also have different roles in glaucoma. As discussed in the section on neuroprotective responses to preconditioning, induction of protective immunity by auto-antigen reactive T cells may protect the retina against secondary degeneration of RGC. The protective response seems to require immunization by an antigen that is also present at or near the site of damage, but this also has the risk of causing a destructive autoimmune response at the same locus. The possibility that autoimmune mechanisms have a pathogenic role in some glaucoma cases arose from initial observations that many patients with normal tension glaucoma had elevated serum titers of antibodies cross-reactive to optic nerve and retina antigens. Among these are autoantibodies to neuroprotective HSPs, such as HSP-60 and HSP-27, which are up-regulated in glaucomatous eyes.99 The antibodies to HSP-27 are toxic to RGC in culture, but it is not known whether this direct pro-apoptotic effect on RGC occurs in vivo. Immune-mediated retinal pathology may also be caused by resident microglial cells that are capable of an immune response. Activated microglia, similar to the other glial cells in the retina, can release cytokines, such as TNF-α and other factors that are potentially neurotoxic. |
PROSPECTS FOR CLINICAL NEUROPROTECTION IN GLAUCOMA |
The term neuroprotective, when used in its broadest sense, includes any agent that counteracts
the two major known risk factors in glaucoma, IOP, and vascular dysregulation, because
such entities are also ultimately protective for optic
nerve axons and RGCs. Various drugs and surgical procedures are currently
available to effectively control elevated IOP, but a great need
remains to better understand the vascular pathology that seems associated
with some glaucomas and to develop topically active agents that specifically
increase microvascular perfusion in the posterior segment of
the eye. Neuroprotective therapies aimed at the glaucomatous retina will most likely focus on agents that protect RGCs and optic nerve axons from the initiating and extracellular signals for RGC death, and secondarily on agents that interrupt the intracellular signal cascade for RGC apoptosis at the decisional or earlier phase (Fig. 1). The main focus should be to intercept, block, or counteract signals leading to damage of RGC axons or somata because current knowledge about regeneration is not sufficiently advanced for development of drugs targeted to this component of neuroprotection. In addition to agents already mentioned in the foregoing part of this chapter, recent studies on retinal neurons in culture or in acute in vivo models have provided additional leads for potential neuroprotective drugs in glaucoma. These include citicholine,100 the tetracycline antibiotic minocycline,101 lithium chloride,102 arachidonic acid,103 inhibitors of cyclo-oxygenase-2,104 calpain and other protease inhibitors,105 gingko biloba extract,106 and endogenous antioxidants thioredoxin,107 hemopexin,108 and metallothionein.109 Despite all these possibilities, most of which are only partially effective in preserving RGC for a short period, there is slow progress in moving forward from proof-of-principle to a clinically useful neuroprotective agent for glaucoma therapy. There are several reasons for this, including the difficulties of performing in vivo experimental drug studies in chronic elevated IOP animal models and also in devising clinical studies for retinal neuroprotection. Another reason for slow development of clinical neuroprotectants may relate to the “one agent, one effect” concept for experimental drug studies. This research approach has greatly increased our understanding of cellular responses and intracellular signal cascades relating to neuron death or survival. However, for a process as complex as neuron apoptosis, which has multiple signals and alternative/default pathways, no single agent is likely to be highly effective. It may be that the path to a truly effective neuroprotective glaucoma therapy is to study a cocktail of several agents, each of which affects a different signal cascade leading to cell survival or to apoptosis. The same considerations probably will apply to the gene-therapy approach. With regard to drugs in current ophthalmic clinical use that, on the basis of model system studies, already appear to have some neuroprotective activity on the retina/optic nerve, namely, brimonidine and betaxolol, it will be a challenge to demonstrate in a clinical study that these agents actually have a direct neuroprotective component that is independent of their IOP-lowering effect. Furthermore, for these established drugs and any new agents that are to be used topically, it will be important to determine whether a sufficient quantity of drug can reach the retina or optic nerve head to exert a measurable protective effect on RGCs after administration to the surface of the eye. Unfortunately, almost all the pharmacokinetic factors affecting topically applied ophthalmic drugs work against penetration to the posterior pole. These include losses caused by the dynamics of the tear film, penetration through the cornea, drug dilution and clearance from the anterior chamber, absorption into and clearance by the vasculature of the anterior uvea, diffusion into and dilution by the vitreous, and vascular clearance from the retina. Getting a topically applied drug to the retina will be difficult to achieve for any combined ocular hypotensive + neuroprotective agent developed in the future. Thus, a great need exists for innovative approaches to increase access of topical agents to the retina, because the alternative approach of systemic chronic treatment for glaucoma brings with it another set of problems. Among these new approaches are a means to greatly increase drug penetration across barriers of cornea, conjunctiva, and sclera, and a means to improve the amount of drug retained by tissues in the vicinity of the retina. Possible examples of such strategies are to link drugs to peptoid molecular transporters modeled on part of the TAT sequence of the acquired immune deficiency syndrome virus110 and/or increasing the binding of drug to melanin in the retinal pigment epithelium. Pharmacodynamic factors relating to topical application will therefore play a decisive role in determining whether any future single drug or a cocktail that is purely neuroprotective (based on proof-of-principle in model systems and no effect on IOP) can become an effective sole therapy for glaucoma. Chronic systemic therapy is the alternative to topical administration of neuroprotective agents for glaucoma. Proof-of-principle for some degree of neuroprotection in model systems has already been established for five such drugs that are in clinical use for other conditions and that have no significant IOP effect, namely, phenytoin for epilepsy, memantine and riluzole for Alzheimer's disease, lithium for bipolar disorder, and deprenyl for Parkinson's disease. It may be worthwhile to study these patient groups to establish the usefulness of these agents for glaucoma by evaluating their effectiveness, each presumably with a different neuroprotective mechanism, to lower the incidence of glaucomatous neuropathy. An added benefit of such studies could be to firmly establish whether the neuroprotective effects found in model systems of RGC death extrapolate to clinical glaucoma. However, even in such retrospective studies it might be difficult to determine whether these agents can slow progression of glaucoma, because their effect might be quite small when given singly. Also when individuals in the study population develop elevated IOP or show visual field loss or optic nerve cupping, they will have to be treated with a currently accepted agent to lower IOP and be excluded from the study. For the same reasons, a long-term clinical trial of a novel purely neuroprotective agent administered to patients with glaucoma on a chronic systemic basis will require extensive justification, and this has the additional difficulty of assessing the incidence and severity of side effects of a new systemic drug in relation to the ophthalmic benefit of treatment. On the basis of these considerations, for the near term, neuroprotective agents for glaucoma therapy will most probably be drugs that either lower IOP or improve vascular perfusion in the retina/optic nerve head, combined with a proven neuroprotective action. The drugs should be topically active and will have been shown to reach the retina and optic nerve head in amounts sufficient to exert some neuroprotective effect, even if less than they have in model systems of RGC loss. It may be possible to develop agents to be given on a chronic systemic basis that are purely neuroprotective, that are more effective than current treatments for glaucoma, and that have an acceptably low incidence of side effects. However, this should be considered a long-term future goal, along with gene therapies for the retina introduced into the eye. It may take a long time to realize the goal of direct retinal neuroprotective therapy for glaucoma based on the disappointing progress, despite an enormous effort by basic science researchers and the pharmaceutical industry during the past decade to develop truly effective therapies for the other major neurodegenerative diseases. |