The Pathogenesis of Glaucoma in the Interplay with the Immune System

Abstract

Glaucoma was previously thought to be caused only through an elevated intraocular pressure as a sole trigger. Emerging evidence indicates that the pathogenesis of glaucoma depends on several interacting pathogenetic mechanisms, which include mechanical effects by an increased IOP, decreased neutrophine-supply, hypoxia, excitotoxicity, oxidative stress and the involvement of autoimmune processes. These autoimmune processes within the central nervous system are a highly organized response of the innate immunity. Through the recognition of neuronal epitopes, the long-term induction of the innate immune response and its transition to an adaptive form might be central to the pathophysiology of the glaucoma disease. Regardless of the pathogenic mechanism, the consequences are always the establishment of extensive degenerative processes in the optic nerve head, the retinal ganglion cells and the axons of the optic nerve, which will lead in the irreversible destruction of these neurons. This review article summarizes the current knowledge concerning the pathogenesis of glaucoma, with special focus on its interplay with the immune system.

Introduction

Glaucoma, behind cataracts, is the second leading cause of blindness worldwide. For the year 2020 it is expected that approximately 80 million people will suffer from glaucoma, which is anticipated to result in 11.2 million cases of bilateral blindness. ¹  

Nowadays glaucoma is a collective term for a group of neurodegenerative processes affecting the entire visual pathway, characterized by progressive, irreversible destruction and death of retinal ganglion cells. ^2–15 These neurodegenerative processes are associated with a progressive visual field loss, which typically begins with an arcuate Bjerrum scotoma in the central visual field and ends with the total blindness of the eye. ^13,16–23  

Well-marked symptoms are known only in acute angle-closure glaucoma; all other forms of chronic glaucoma are largely asymptomatic. This is the main reason the disease accomplishes its destruction to a far extent unnoticed. As many as 50% of all affected patients live without diagnosis until advanced illness. ¹³ A simplified diagnostic procedure, ideally a rapid glaucoma test that could be carried out by an optician, would be an effective tool to gain time through early diagnosis and treatment. However, to date the pathogenesis of glaucoma is understood only in part; as a result, useful mass screening for glaucoma is lacking with respect to reliable yet affordable and expeditious diagnosis. ²⁴  

Undisputedly, a permanent raised individual ocular pressure above 21 mm Hg is an important pathogenic factor in the disease risk. ^9,10,25–27 Individual ocular pressure constitutes a person-specific variation and follows a circadian rhythm. ²⁸ Following this rhythm, the IOP oscillates between 10 and 21 mm Hg in healthy adults. Newborns have an IOP between 6 and 8 mm Hg, which rises approximately 1 mm Hg every 2 years. ^{25,26,29–34}  

In addition, there are healthy people who have an IOP significantly elevated above normal without ever developing glaucoma. This condition is known as ocular hypertension (OHT). In direct contrast, many patients with an IOP ≤ 21 mm Hg progress to glaucoma. Accordingly, this form of glaucoma is called normal-tension glaucoma (NTG). Important for the pathogenesis of glaucoma therefore seems to be the individual, relative value of IOP. 

Available Treatment Options for Glaucoma

The treatment of glaucoma is almost always aimed at reducing the IOP, which leads, surprisingly, even in patients with NTG, to an alleviation of the disease process. ^{4,13,15,35–39} The reduction of IOP is obtained by eye drops or systemic application of glaucoma medications. These include carbonic anhydrase inhibitors, beta-blockers, cholinergic agonists, α2-adrenoceptor agonists and prostaglandins. ^{25,27,37,40–43} Laser therapy or surgical procedures for ensuring an adequate aqueous humor outflow are seen as a last resort to use. ⁴⁴  

Pathogenesis and Progression of Glaucoma

For decades, a permanent raised IOP over 21 mm Hg was considered the sole trigger for the onset of glaucoma. In fact, in the most common glaucoma type, primary open-angle glaucoma (POAG), the majority of those affected exhibit a raised IOP. Therefore this glaucoma subtype belongs to the group of high-tension glaucomas (HTG). However, approximately one-third of all POAG patients had at no time a pathologically elevated IOP (NTG patients); this raises the question what, if not the IOP, is responsible for the destruction of retinal ganglion cells. ^{25,27,45–47} It is worth mentioning that therapeutic lowering of the IOP leads to a slowing of disease progression in a subgroup of NTG patients. ^43,48–50  

The substantial demise of retinal ganglion cells (RGCs) is accompanied by morphologic changes of the retina, of which the cupping of the optic nerve head (ONH) is the most prominent. Thereby glaucomatous damage is not confined just to the RGCs and their axons, somata, and dendrites. Rather the entire visual pathway from the retina to the visual cortex in the brain is affected. ^{14,26,51–54} However, it is not clear whether the remodeling of the ONH leads to the destruction of RGCs, or conversely, whether the RGC loss in glaucoma is causative for the characteristic cupping of the ONH. ^{8,9,15,30,50,55,56} Beyond controversy, apoptosis is accepted as an important component of glaucomatous neurodegeneration. ^{6,10,12,13,17,19,20,31,36,50,57–64}  

The involvement of apoptotic processes, as well as the correlation between IOP and glaucomatous damage, is rather obvious in pathologic conditions with rapidly raised IOP, as it can be monitored in acute angle-closure glaucomas and acute secondary glaucomas. ^{12,15,20,65–67} As demonstrated in animal models, severe pressure insults may cause axonal degeneration and apoptosis within hours. ^15,31,33,68  

However, in open-angle glaucomas and mild forms of secondary glaucomas, the correlation between IOP and neuronal degeneration is not so apparent. OHT patients with slightly elevated IOP between 21 and 30 mm Hg, without detectable signs of disc cupping and/or visual field defects, do not necessarily progress to glaucoma. ⁶⁹ The Ocular Hypertension Treatment Study demonstrated that over 90% of the untreated control group did not develop any form of glaucoma within the 5-year study period. ^25,27,70  

The Lamina Cribrosa as the Most Vulnerable Part of the Retinal Ganglion Cell

As mentioned above, an elevated IOP is one of the main risk factors for glaucoma, which may be associated with a deficiency in the cellular nutritional state of the RGCs. This presumption is suggested by an animal model in which an elevated IOP causes the failure of the neurotrophin supply due to the collapse of axonal transport. ^71,72  

RGCs respond to a variety of neurotrophins, but mainly to brain-derived neurotrophic factor (BDNF), ciliary nerve trophic factor (CNTF), glial cell–derived neurotrophic factor (GDNF), and nerve growth factor. ^73–77 In animal models, as well as in cell culture, the neuroprotective effects of these trophic factors are documented. ^78,79 Like other neurotrophins, nerve growth factor (NGF) and BDNF are taken up into the target cell through receptor-mediated endocytosis. Binding partners are the high-affinity receptors TrkA (for NGF) and TrkB (for BDNF), and a low-affinity receptor, p75^NTR, to which all other neurotrophins can bind. The internalization of the receptor–ligand complex followed by the retrograde transport to the soma of the neuron is coordinated by the G-proteins Rabenosyn 1 and 5 (Rab1 and Rab5). ^80–85 Immunohistochemical studies demonstrate that impairing neurotrophin transport from the visual thalamus to the soma of RCGs results in their destruction by apoptosis. ^20,65,72 Similar effects are observed in the optic nerve crush model. ^19,86  

Here the lamina cribrosa sclerae, through which the optic nerve and the central retinal artery and vein enter the bulbus oculi, appears to play a key role, since it is the weakest portion of the sclera. A raised IOP may cause an excavation at the position where the optic nerve (papilla nervi optici) enters the retina. ⁵⁵ By dyeing RGC axons of freshly explanted eyes, Morgan et al. were able to document the path of individual axons through the lamina cribrosa: Most axons take a direct path toward the ONH, while others, approximately 8% to 12%, squeeze through collagenous plates. An increased IOP leads directly to a contusion of these nerve fibers. ⁸⁷ Experiments with primates show that the retrograde and anterograde neuronal transport in glaucomatous eyes is disturbed or even interrupted at just this point. ^{65,71,72,88–91} Additionally, the cerebrospinal fluid pressure influences the degree of mechanical stress, due to a pressure gradient across the tissue of the ONH. ⁹²  

The Demise of Retinal Ganglion Cells as a Central Hallmark of Glaucoma Pathogenesis

In glaucoma, it seems that the nature and position of the injurious trigger determine which path is taken into neurodegeneration. Already documented is the initiation of programmed cell death of RGCs via tumor suppressor protein p53, but also through the direct activation of the “death receptor” CD95 in autoreactive conditions. ^{58,63,93–95}  

In general, the RGC's fate depends on multiple trophic as well as degenerative signaling pathways within the soma and/or signaling from the environment, which has the potential to disrupt neuronal homeostasis. One example is the cytotoxic effect of β-amyloid protein, which causes apoptosis by binding to neurotrophin receptor p75^NTR or through the accumulation of protein aggregates, as in Alzheimer's disease. ^96,97 p75^NTR, another known cause, is involved in light-induced photoreceptor apoptosis. ^98,99  

Furthermore, neurotransmitters such as dopamine, serotonin, and glutamate have the potential to drive RGCs into programmed cell death. As a cause for the triggering of apoptosis, excitatory mechanisms are suggested. ^{36,43,50,100–110} Regarding the potential excitotoxicity of glutamate, it is noteworthy that glutamate transporters like GLAST and EEAC1, as well as glutamate receptors, are downregulated in glaucomatous eyes following astrocyte activation. ^102,111,112 As a consequence, deficient glutamate removal from the extracellular space might represent an interaction by which astrocyte or glial activation can result in RGC death. ^102,112  

Considering the complexity of RGCs, Whitmore et al. postulate the model of “compartmentalized self-destruction” in neurons, suggesting that depending on the nature and location of the initial insult, somal death can be preceded by a number of degenerative processes infesting the axons, dentrites, and synapses of the neuron. ¹⁵ These degenerative processes are phylogenetically highly conserved and known to be involved in a whole range of destructive processes within the vertebral central nervous system (CNS), as well as in axons of Drosophila. ^{10,79,113–125} Since such widespread processes of neuronal degeneration may affect the entire CNS, one can assume that insights into neurons outside the eye are also applicable on RGCs. Current findings support the assumption that self-destruct processes, especially axonal degeneration as seen in non-RGC neurons, also take place during glaucoma. ^10,124,126 One of these self-destruct processes, which is documented across species boundaries, is known as Wallerian degeneration (WD), a common phenomenon in cases of severe axonal damage such as transection or experimental nerve crush (i.e., nerve crush model). WD represents valid evidence that axonal degeneration is a soma-independently organized process that can be studied in mice carrying the Wallerian Degeneration Slow (WldS) gene. ^{15,61,127–131}  

This mutation results in a resistance to axonal degeneration while having no direct effect on somal apoptosis. During WD, the axonal section, which is detached from the soma by the injury, degenerates in a characteristic and autonomous manner, beginning at the point of transection and extending almost synchronously throughout the axon within days. ^124,132,133  

A second frequently observed pathomechanism, so-called dying back (DB), seems to be a variation of WD and follows generalized neuronal stress without acute neuronal trauma. Differing from WD, DB is initiated by a withdrawal of the synapses, followed by an asynchronous progression of the destruction process toward the cell soma. 

In cases of DB, the neuron seems to have a distinct chance to regenerate if the neuronal stress decreases, whereas in WD, somal apoptosis appears to be unavoidable. ^15,134,135 The review article by Whitmore et al. ¹⁵ provides detailed information concerning compartmentalized degeneration. 

The following sections provide an overview of how, on the one hand, the immune system eliminates pathogens and cell debris to maintain the homeostasis of the CNS and, on the other hand, observations of well-studied diseases such as Alzheimer's or multiple sclerosis (MS). The latter show what terrible damage the immune system is able to provoke in situations of cellular stress and unphysiological conditions. Many of the well-known neurodegenerative incidents that are caused by misdirected immune processes within the CNS (e.g., reactive astrogliosis) are also detectable in glaucoma. ¹³⁶ This illustrates how homeostasis and survival of the RGC depend on a well-balanced function of the immune system. 

Autoimmune Mechanisms in the Onset of Glaucoma

The first hint for an involvement of the immune system in glaucoma was delivered by Wax in 1998 with the detection of antibodies against endogenous antigens such as heat shock protein 60 (HSP60) in the serum of NTG patients. ¹³⁷ Heat shock proteins (HSPs) are components of cellular defense mechanisms and are upregulated under pathophysiological conditions. ¹³⁸ According to their molecular weight, HSPs are divided into two main families: the major HSPs (HSP90, HSP70, and HSP60) and the small HSPs, including HSP27. HSP60 and HSP90 are constitutively expressed in the CNS, whereas HSP27 and HSP70 are highly inducible in the retinal glia, the Müller cells, and neurons by stressors such as ischemia and hyperthermia. ^139–142 However, HSP27 expression in the intact, undamaged retina has been found to be limited. ¹⁴³ In states of cellular stress, HSP27 and HSP70 act in an antiapoptotic manner, whereas HSP60 promotes the induction of apoptosis. ^144,145 Direct evidence for the involvement of HSP27 in glaucomatous processes comes from the detection of this protein in RGC axons and ONH of affected human donor eyes, considering that the axons and ONH are mainly battered under glaucoma. ¹⁴⁶  

In an animal model, HSP27 develops its neuroprotective effects after axotomy and ischemia, possibly through the inhibition of caspase-3. ^147–149 Furthermore, HSP70 also shows neuroprotective effects after ischemic conditions. ^141,150  

Since it has been documented that antibodies are able to enter the cytoplasm of intact cells, the effect of autoreactive immunoglobulins, in particular on cellular processes, is taken into account. ^151,152 Rats immunized with HSP27 show an elevated rate of apoptosis in the RGC, where the loss is focused near the area centralis. NTG patients display the worst damage in the same retinal area. ^153,154  

In the case of HSP27, either α-HSP27 autoantibodies could protect HSP27 from premature degradation by matrix metalloproteinases (MMPs), considering that neuroprotective HSP27 is among the MMP substrates, or this antibody may inhibit the neuroprotective activity of HSP27 and thus lead directly to increased apoptosis rates of RGC in glaucoma. ^142,155,156 Significantly increased levels of MMPs in glaucomatous retinae support a hypothetical protective role of α-HSP27 autoantibodies. ¹⁵⁷  

On the other hand, the accumulation of MMPs correlates with the release of TNF-α particularly in patients with NTG, in which MMPs are involved in the processing of the membrane-bound TNF-α precursor. ¹⁵⁸  

But it's not just about antibodies against HSPs: Documented peculiarities in the antibody repertoire of glaucoma patients have been reported by Maruyama et al. These authors illustrated that the sera of approximately 20% of all glaucoma patients contain antibodies against neuron-specific γ-enolase, a key enzyme for glycolysis. In the healthy control group, only 10% of those examined showed these antibodies. However, in glaucoma patients with increased γ-enolase antibody titers, none of the elevated autoantibody titers against rhodopsin, α-crystallin, and HSP60 otherwise commonly detected in glaucoma patients could be demonstrated. ^{8,146,153,159,160} Further studies, published by Grus et al. and Joachim et al., also revealed significant differences in the antibody profiles of glaucoma patients. ^161–163  

Furthermore, a significantly increased antibody reactivity against ONH glycosaminoglycans (GAGs), glutathione S-transferase, alpha-fodrin, and neurofilament protein has been found in the sera of patients with glaucoma. ^163–166 In the case of anti-GAG antibodies, a direct physiological impact is postulated due to the change of organization and physical characteristics of GAGs located in the ONH. This may increase the susceptibility of the ONH and the lamina cribrosa to tissue damage and subsequently additional insult to the remaining axons. ¹⁶⁴ However, in most studies, antibody titers in NTG patients are elevated, although some patients with elevated IOP show decreased autoantibody reactivity as well. ¹⁶¹  

An explanation for these measurable deviations in the antibody repertoire of patients with glaucoma could be found in our own evolutionary history: Some proteins, such as histones or HSP, are strongly enough conserved phylogenetically that molecular similarities between host and pathogenic germs occasionally result in cross-reactivity. ^{137,167–169}  

Additionally or independently thereof, epitope spreading could be another source of autoantibodies in glaucoma: Typically, a humoral response begins with the attack on the epitope that elicited the initial immune response, spreading to other epitopes on the same antigen (intramolecular epitope spread) or to similar epitopes on other antigens (intermolecular epitope spread). Autoimmunity may arise from epitope spreading on inflammation-induced posttranslational modified self-antigens or from self-antigens that result from immune-mediated apoptosis. ^170–172  

In many cases, the autoantigen that triggered the initial immune response has been identified and in fact shares sequence homology with a disease-causing antigenic pathogen component. In this context, the application of epitope mapping has allowed detailed characterization of those epitopes whose autoantibodies bind in diseases such as systemic lupus erythematosus (SLE) (antispliceosome antibodies), Devic's syndrome (anti-aquaporin-4 antibodies), and NTG (antirhodopsin antibodies). ^173–175 Intriguingly, antirhodopsin antibodies found in sera of NTG patients also bind to viral and bacterial epitopes. ¹⁷⁴  

Regarding these facts, we have to keep in mind that autoreactive antibodies can be not only destructive but also protective. Recent findings support the hypothesis that these antibodies contribute to the clearance of cellular damage and promote repair. So, the humoral immune system can be seen as a double-edged sword, with the potential both to destroy and to heal. A repertoire of low-affine, so called natural autoantibodies is present in every healthy human and seems to be of great significance during tissue injury when these autoantibodies contribute to the removal of cellular debris and support tissue healing. ^176,177 Furthermore, humoral autoimmunity seems to have a protective role. Under discussion, among other functions, are the masking of self-recognition against pathogenic autoantibodies, the neutralization of inflammatory cytokines like TNF-α through anticytokine antibodies, and an immune-regulatory function. ^178–181 A decreased reactivity of naturally occurring and perhaps protective autoantibodies may therefore lead to a loss of immune protection and consequently an increased risk of developing glaucoma. ¹⁸²  

Microglia Coordinate the Immune System of the Central Nervous System

Today, three retinal glial cell types are known: The microglia, subdivided into star-shaped astrocytes and specialized retinal Müller cells, face the macroglia. In the retina, astrocytes are localized in the ganglion cell layer, whereas Müller cells span the entire retina. Microglia, ovoid or amoeboid shaped, can be found in the ganglion cell and inner plexiform layer. The macroglia provide the supply of the ganglion cells with neurotrophins, regulate extracellular ion concentration, and support neuronal metabolism. ^183–186 In contrast, microglia act as phagocytes that eliminate apoptotic neurons, cell debris, and pathogens. ¹⁸⁷ Microglia recognize these via Toll-like receptors (TLR) and react by releasing proinflammatory cytokines such as TNF-α and nitric oxide synthase-2 (NOS-2). Apoptotic cells release anti-inflammatory mediators such as TGF-β prior to phagocytosis by microglia. ¹⁸⁸ Microglia represent the first line of immune defense and are therefore equipped with a finely structured arsenal of defense mechanisms and a high degree of flexibility to act at the appropriate place. They destroy microbes by releasing cytotoxic substances, remove cell debris by phagocytosis, and act as antigen-presenting cells. ¹⁸⁹  

Apart from the normal motility behavior under physiological conditions, in the event of activation the cells migrate collectively to the site of injury. Chipped neurons usually set free nucleotides such as adenosine triphosphate (ATP), which may activate microglia. ^190,191 After a few days, this event is followed by glial proliferation that can end up in a reactive gliosis. ¹⁹² The involvement of a gliosis in neurodegenerative processes has been demonstrated in an animal model for Alzheimer's disease as well as in glaucoma. ^{68,105,136,193–195}  

An indication of the positive chemotactic motility of microglia is provided by experiments with CXCR3 knockout mice. The chemokine CXCL10 expressed under degenerative conditions by neurons cannot bind the microglial CXCR3 receptor. This results in an accumulation of damaged and dead cells that would have been gone after a few days under normal circumstances. ¹⁹⁶ In an animal model of amyotrophic lateral sclerosis (ALS) and Parkinson's disease, another chemokine, CX3CL1, has been identified that is released from damaged neurons as a chemoattractant for glial cells. CX3CR1^−/− mice whose microglia possess no functional chemokine receptor exhibit increased neuronal apoptosis, which indicates the neuroprotective role of microglia. ^197,198 Apart from the supportive and protective function of the glia, an injury caused by degenerative processes or glial activation in glaucomatous eyes does damage. ^32,187,199 At least in the eye, the activation correlates with the accumulation of mitogen-activated protein kinase within the glial cells. ²⁰⁰ Such activation occurs in the CNS as a general response to any form of injury or illness and is considered a cellular attempt to maintain the impaired mechanisms. 

The Role of Innate Immunity in the Manifestation of Central Nervous System Autoimmune Diseases

The activation of microglia, an unmistakable sign of inflammatory processes within the CNS, is documented in patients with Alzheimer's disease, Parkinson's disease, Huntington's disease, MS, and ALS. ^201–203 Found in serum and cerebrospinal fluid of these patients are elevated titers of IL-6, IL-1, and TNF-α, chemokines of innate immunity that are potent initiators of the extrinsic apoptosis of neurons. ^201–203 Especially for IL-1 and TNF-α, the activation of caspases has been shown, whereupon TNF-α increases apoptosis of neurons via the inhibition of insulin-like growth factor-1 (IGF-1). ^204,205 In addition to the secretion of chemokines, activated microglial cells set free toxins such as nitric oxide (NO) and prostaglandins, which in turn initiate neurodegenerative processes. ^203,206 As professional antigen-presenting cells, activated microglia control the transition from innate immunity to an adaptive immune response in the CNS. ^206–208  

However, in glaucoma, TNF-α is of special interest: TNF-α and its receptors TNFR1a and TNFR1b have been shown to be upregulated in the retinae of glaucoma patients, in whom increased expression of TNF receptors is predominant in the ONH. ^158,209,210 In severely damaged ONHs, the axons of the RGCs express TNFR1 and may be the direct target for TNF-α–mediated neurodegeneration. ²¹⁰ Thereby the increased TNF-α expression seems to be specifically controlled. TrkC.T1, a truncated neurotrophin receptor isoform lacking the kinase domain, has been shown to be rapidly upregulated in glaucomatous retinae. On closer examination, TrkC.T1 seems to be part of a fatal loop: Elevated IOP upregulates glial TrkC.T1 expression in glia; TrkC.T1 controls glial TNF-α production; and TNF-α causes RGC death. ²¹¹  

Another interesting aspect in this context is the occurrence of single-nucleotide polymorphisms of the TNF-α gene, found to be significantly higher in high-tension glaucoma patients. ²¹² Ambiguous so far are the effects of those polymorphisms on the pathogenesis of glaucoma. While studies from Pakistan and Iran have shown a strong connection of POAG and pseudoexfoliative glaucoma with TNF-α polymorphisms, others could not confirm this correlation. ^213–216  

So far, there are a number of findings of elevated cytokine titers and activated glial cells within the CNS, but it is not clear what is causative and what is classified as an epiphenomenon for the manifestation of an autoimmune disease. What seems to be clear is that cellular stressors such as oxidative stress, mitochondrial dysfunction, or excitotoxicity by glutamate are not sufficient as singular stressors to trigger the demise of neurons. Animal models for ALS, Alzheimer's disease, Parkinson's, and Huntington prove this assumption. ^203,217,218  

Similarly, the presence of antibodies directed against neuronal antigens affects the cellular function of the epitope, but these are not able to trigger a neurodegenerative autoimmune disease. Tezel et al. succeeded in inducing apoptosis of neurons and cells of the retinal vascular system in a retinal cell culture by the application of antibodies against α-A- and α-B-crystallin, as well as against HSP27. ¹⁵³ This, however, is not directly comparable with the induction of a complex autoimmune disease in vivo. 

In the case of paraneoplastic neurologic disease (PND), which can occur as a secondary symptom of a tumor disease, epitopes generated by tumor cells in the body correspond to those of neurons in the CNS. This leads to the establishment of a neurodegenerative process within the CNS. The therapeutic reduction of autoantibody titer by plasmapheresis provides no persistent relief of the clinical course. ²¹⁹ Conversely, it is not possible to induce a PND in animal models by injecting antibodies from infected individuals. Autoantibodies alone are therefore not sufficient to initiate a neurodegenerative disease. ^219,220 However, it is conceivable that autoantibodies directed against ocular structures have the potential to activate the complement system via binding C1q of the classical pathway. ^{165,221–223}  

The discovery of activated T cells and macrophages in the brain parenchyma of neurodegenerative disease patients confirms the participation of the cellular immune response directed against the body's own neural tissue within the CNS. ^224–232 The cause of inflammatory processes in the CNS may therefore be T cells and autoantibodies from the periphery that find access to the CNS, where they are again reactivated by local antigen-presenting cells (APC) or the microglia, or, in the case of antibodies, bind directly to their neuronal epitopes. The reactivation of T cells triggers the release of inflammatory cytokines and thereby the maintenance of the immune response. Phagocytes float to the site of inflammation, take up cell debris by endocytosis, present fragments thereof in peptide size on the major histocompatibility complex (MHC)-II on their surface, and thus again activate the adaptive immune response. The adaptive immunity leads to antigen-specific CD4⁺ T-helper cell activation and differentiation of naive B cells into antibody-producing plasma cells in the periphery. These antibodies cross the blood–brain barrier, particularly under pathologic circumstances, but even at elevated IOP. ^233,234 These immunoglobulins not only bind to epitopes on the surface but also enter the cell by receptor-mediated endocytosis, as in the case of α-HSP27 antibody. In the cytoplasm of neurons, these immunoglobulins may cause apoptosis. ^142,151 The resulting cellular debris is taken up by the phagocytes of the CNS, the microglia as antigen, and also peripheral T cells. This process is hosting an adaptive immune response to neural targets in the CNS. 

Through recurring cytokine release and antigen presentation, a self-perpetuating cycle is under way. The result is apoptosis of neurons, increased permeability of the blood–brain barrier, and upregulation of neuronal MHC-I, causing a chronic activation of innate immunity and establishment of an adaptive autoimmune response in the CNS. The induction of intercellular adhesion molecule 1 (ICAM 1) and vascular endothelial growth factor (VEGF) by proinflammatory cytokines of innate and adaptive immunity leads to further impairment of the barrier function of the blood–brain barrier, since they allow the passage of activated T lymphocytes ^235,236 (Fig. 1). 

Figure 1

 

Hypothetical mechanism of neurodegeneration in the CNS involving innate and adaptive immunity. Damage of peripheral organs leads to the activation of immunological processes involving antigen-presenting cells (APC). The APC float to local lymph nodes where activation of the adaptive immune response takes place. (3, 4) The following clonal expansion of B cells leads to the production of antigen-specific antibodies. (5) These antibodies pass through the permeable blood–brain barrier. (6) In the CNS, the antibodies bind to neuronal antigens, interfering with their function, and cause cellular stress and finally apoptosis of the neuron. (7) The apoptosis of neurons leads to activation of microglia. (8) Activated microglia phagocytose the apoptotic cells and present their antigens on their surface. (9) Microglia secrete proinflammatory cytokines. Increased titers of vascular endothelial growth factor (VEGF) and the intercellular adhesion molecule 1 (ICAM 1) lead to increased permeability of the blood–brain barrier. (11) This leads to further influx of T lymphocytes. (12, 13) T cells and neuronal antigens activate microglia, which also leads to the destruction of neurons. (14) Alternatively, cellular damage or pathogens can cause chronic immune responses of the nonadaptive immune defense. Reprinted with permission from Nguyen MD, Julien JP, Rivest S. Innate immunity: the missing link in neuroprotection and neurodegeneration? Nat Rev Neurosci. 2002;3:216–227. Copyright 2002 Macmillan Publishers Ltd: Nature Reviews Neuroscience.

Even in the absence of inflammatory events in the CNS, the transfer of T cells in the subarachnoid space (SAS) between blood and cerebrospinal fluid is feasible. ²³⁷ Undoubtedly the permeability of the blood–brain barrier could play a major role in the pathogenesis of neurodegenerative processes in the CNS such as Alzheimer's disease, MS, and human immunodeficiency virus (HIV)-associated encephalitis. ^202,238 Recently, Howell et al. documented the leukocyte transendothelial migration pathway as activated in the DBA/2J glaucoma model. This results in the inflow of proinflammatory monocytes (and/or monocyte-derived cells) into the optic nerve prior to detectable neuronal damage. ²³⁹  

Induction of Innate Immunity as a Pathogenic Factor in the CNS

Today, we know that innate immune responses are responsible for many chronic neurodegenerative processes within the CNS. ²³⁶ The retina in this case has its own immune system, including immune cells: microglia, dendritic cells, and even their own perivascular macrophages, all components of the innate immunity. ²⁴⁰ An essential component of innate immunity is receptor-mediated pathogen recognition by the Toll-like receptor family (TLR) and triggering receptor expressed on myeloid cells (TREM). These two classes of receptors are found on the microglia, neutrophils, dendritic cells, and macrophages in the CNS and the periphery. ^241–246 Since TREM was identified only in 2000, it is still not really clear what their ligands are. Nevertheless it seems that TREM modulates the activity of TLR; that is, TREM-1 has an activating and TREM-2 an inhibitory influence on TLR. ^247,248  

In the case of pathogen recognition, TLR will induce the coordinated activation of transcription factor NF-κB (nuclear factor kappa-light-chain-enhancer of activated B-cells), together with the increased expression of cytokines IL-1b and IL-6, chemokine IL-8, and the costimulatory proteins CD80 and CD86 (also known as B7.1 and B7.2) for targeted regulation of innate immunity. ^225,249 In general, TLR are dimer-organized, mostly outward-directed receptors, activated by pathogen-associated molecular patterns (PAMPs). The PAMPS include substances such as lipopolysaccharide or bacterial CpG DNA, which is characteristic for Gram-negative bacteria. ^249–251 Of the total of 12 isotypes so far identified, TLR 3, 7, 8, and 9 monitor the internal cellular compartments. ^{242,252–254}  

In animal models, experimental elevation of IOP leads to increased expression of TLR 2, 3, and 4 and of HSP27, HSP60, and HSP72, as well as the characteristic TLR signaling cascade adapter proteins and kinases. These findings were confirmed by proteomic analysis of glaucomatous donor eyes. This supports the hypothesis that TLR contribute to the activation of the innate immune system in glaucoma. The increased HSP expression seems to stimulate the immune system even further. ²⁵⁵  

The latter may be due to the cross-reactivity between HSP and TLR, but it can also be explained by the ability of HSPs to present antigens. ^256,257 Thus, TLR 4 reacts on HSP27 and HSP60, and TLR 2 recognizes glutathione S-transferase as PAMP. ^258–260 HSP72 interacts with TLR 2 and TLR 4. ²⁶¹ This capability of HSPs to cause cross-reactivity can be explained by their distinctive antigenicity. Because HSPs are among the most highly conserved and abundant proteins in nature, as a logical consequence they are also among the major antigenic proteins for pathogens like bacteria, mycobacteria, and parasites, which share HSP epitopes with human HSPs. ^169,262,263 Via the stimulation of innate immunity by a cross-reaction, which is not in these cases caused by molecular mimicry, autoreactive mechanisms are set in place. ^264–266 Molecular mimicry with a pathogen initially includes “appearance” to the surface properties of host cells to evade the surveillance of the immune defense. ^267,268 Whether this is caused by cross-reactivity observed in phylogenetic conservation or molecular mimicry is obviously dependent on the relevant antigen. 

Either way, the consequences are the same. As seen in the cross-reactivity caused by conserved amino acid sequences, mimicry may also establish autoreactive processes. ^269–271 The correlation between the initiation of aberrant immune responses due to molecular mimicry between pathogen and host tissue has been unveiled in the development of numerous autoimmune diseases. The simultaneous occurrence of enteric bacteria or virus infection and the onset of autoimmune processes speaks for itself. ^272,273 For glaucoma, an infection with Helicobacter pylori is discussed as a disease-causing trigger. ^269,270,274  

Where Do the Neuronal Antigens in the Periphery Originate?

Besides the already mentioned cross-reactions, other theories could help to explain the existence of CNS-specific antigens and autoantibodies in the periphery. Proteins, fragmented or oxidatively modified by reactive oxygen species (ROS) or metalloproteinases (MMPs), could serve as autoantigens. ^{155,156,275,276} Examples include HSP27, HSP70, and HSP90, which are known to be substrates for MMP-9. 

Autoantibodies directed against these HSPs exhibit increased reactivity in glaucoma and other neurodegenerative diseases such as MS or Guillain-Barré syndrome. ^{153,155,277,278}  

The Janus-Headed Complement System

In addition to the cellular components of immunological surveillance, the complement system represents the main component of the innate immune response that can be activated by the so-called classical path, the lectin pathway, and the alternative pathway. 

C1q represents the first component of the classical pathway, which can lead, either via antibody mediation or via direct binding, to complement activation. 

In the case of C1q binding to an apoptotic cell, a pathogen, or cell debris, the activation of a protease cascade directs the synthesis of complement component C3 (opsonization). ²⁷⁹ Opsonization with C3 fragments C3b and iCb heads the elimination of debris by direct activation of C3 receptors on macrophages and microglial phagocytosis. Activated C3 has the ability to activate the terminal components of the complement system. 

This allows the formation of membrane attack complex (MAC), the direct elimination of the target cell by a lytic process. These are the well-known facts about the function of the complement system. But how can one interpret recent findings regarding increased transcription rates and accumulation of complement protein C1q and downstream complement protein C3 within the adult glaucomatous retina? ^280,281  

As described above, the production of autoantibodies is a hallmark of many autoimmune diseases, as well as in autoimmune glaucoma. ^{7,8,47,161–163,165,172,179,282–292} These autoantibodies may bind antigens on cell surfaces or form immune complexes after catching circulating antigens, which may, if not rendered harmless in time, accumulate in lymph nodes and diverse organs such as the kidney glomeruli. These immune complexes have the potential to activate the complement cascade via the classical pathway. ^171,292,293  

However, in the OHT animal model (DBA/2 mice) and donor eyes of OHT patients, transcription rates of C1q and C3 are significantly increased. Twenty-eight days after experimental OHT induction, an accumulation of C1q and C3 transcripts is detected in the RGC layer and the ONH ²⁹ (Fig. 2). 

Figure 2

 

Distribution of C1q transcripts in the retina. Compared to the control (A) strong signals for C1q can be detected in OHT patients in the ganglion cell layer (B) and the optic nerve head (ONH, [C]) strong signals for C1q (arrow). Reprinted with permission from Kuehn MH, Kim CY, Ostojic J, et al. Retinal synthesis and deposition of complement components induced by ocular hypertension. Exp Eye Res. 2006;83:620–628. Copyright 2006 Elsevier.

This expression of C1q and C3 correlates with the activation of glial cells by the complement system. The activated glia hereupon responds with the secretion of NO, TNF-α, superoxide (O^2–), and other effector substances. This involves glia in the reorganization of the ONH and the destruction of RGCs. ^56,294–297 Additionally, the classical complement cascade contributes to development and function refinement of the CNS and visual system via pruning of inappropriate synapses. ^{281,298–300} This process of coordinated synapse elimination may be facilitated through the ability of developing CNS neurons to bind C1q directly. This allows activation of the classical complement cascade independently of antibodies. ³⁰¹ However, the involvement of the complement system in synapse elimination is strictly limited to the development phase of the CNS and visual system under normal conditions, but seems to be reactivated in glaucoma. Recent studies point in the direction that components of the adaptive immune system also participate in the developmental synaptic refinement. In particular, class I major histocompatibily complex (MHC-I), which is not expressed by neurons under physiological conditions, is detectable on developing neurons and on neurons under inflammatory conditions. ^302,303 It is intriguing to speculate that components of the complement pathway may be interacting with MHC-I, mediating synapse elimination. To date, such an interaction is elusive. However, the expression of the MHC-I complex on neuronal surfaces makes axons vulnerable to attacks of CD8⁺ T cells by perforin-mediated cytotoxicity and the release of granzyme ²²⁷ (Fig. 3). 

Figure 3

 

Cytotoxic T cells use three different routes to destroy their target. Initially the CD8⁺ T cell recognizes the target via the interaction of its T-cell receptor (TCR) with the target cell's MHC-I complex, including presented peptide (square). The destruction of the target cell is performed by either (i) release of cytotoxic granules and the resulting perforation of the cell membrane, (ii) activation of the target cell's Fas/CD95 receptor by Fas ligand/CD95 ligand, or (iii) the release of cytokines such as TNF-α. Reprinted with permission from Neumann H, Medana IM, Bauer J, Lassmann H. Cytotoxic T lymphocytes in autoimmune and degenerative CNS diseases. Trends Neurosci. 2002;25:313–319. Copyright 2002 Elsevier.

Reactive gliosis, release of inflammatory cytokines, and upregulation of the complement cascade as hallmarks of developmental synapse elimination in early stages of glaucoma reoccur in many disease models, as well as in eyes of human donors. ^{29,280,281,304} Recent work investigating the first steps of glaucoma genesis in the DBA/2J mouse points to the activation of astrocytes as one of the earliest events in glaucoma pathogenesis. This awakening of the innate immunity seems to be triggered by a contusion of RGC axons traversing the sclera through the lamina cribrosa. ^10,126,305 Since the lamina is the weakest part of the sclera, elevated IOP inevitably leads to mechanical stress on RGC axons at this point. ^2,87,305,306 In the course of glaucoma progression, microglial proliferation and activation are accompanied by the secretion of TNF-α and IL-1β. ^{57,210,295,307–309} As a consequence, this release of inflammatory cytokines have the potential to impair the integrity of the blood–brain barrier. ^{209,233,234,310,311} The reactivation of the complement cascade is noticeable by an increased transcription rate of C1q, C3, and C4 and the subsequent accumulation of these proteins within the inner plexiform layer (IPL). ^{29,280,281,304,312} Synapses, “tagged” with C1q and/or C3b, are phagocytized by microglia; downstream formation of the MAC results in synapse loss without participation of glial cells. Apart from the neurotoxic effects of a misdirected complement cascade, the decease of RGCs is thought to be caused by proapoptotic TNF-α and the excitotoxic effect of glutamate, which is supported through the downregulation of glutamate transporters during reactive gliosis. What's left of the RGCs after their demise is thought to be opsonized and removed by the complement system. ²⁹⁹ The ongoing elimination of apoptotic bodies, cells, and debris has a high priority, because in the case of aggregation they endanger the metabolism and offer an autologous immune epitope. ³¹³  

Where the C1q-controlled neutralization of endogenous antigens is defective, a misallocation of the immune system may result in the establishment of autoimmune processes such as SLE or glomerulonephritis. ³¹⁴  

Generally, it appears that a deviation of the complement system is responsible for a number of neurodegenerative processes, such as ischemic conditions within the CNS, MS, Alzheimer's, and Parkinson's disease. ^315–318 Whereas C1q has a role as an eliminator of superfluous synaptic connections during a short phase of embryonic and postnatal maturation of the CNS, its expression in the adult CNS is suppressed under physiological conditions. However, the expression of C1q by neurons and microglia under ischemic, traumatic, and other pathogenic influences is raised very quickly. Apart from glaucoma, the massive upregulation of C1q expression is also detectable in Alzheimer's disease and ALS. ^{32,304,312,319–323}  

The key to understanding these degenerative events is to be found in the extensive cross talk between complement and TLR signaling pathways. This complement–TLR interplay, based on synergistic or antagonistic interactions, reinforces innate immunity or regulates excessive inflammation. The interaction of complement on TLR signaling is primarily executed via the C5a receptor (C5aR), and to a much lesser extent via the C3a receptor (C3aR). This cross talk appears to involve the mitogen-activated protein kinase (MAPK) extracellular signal-regulated kinase (ERK1/2), and the c-Jun N-terminal kinase (JNK). ³²⁴ While the mutual reinforcement of complement cascade and TLR with its increased TNF-α, IL-1β, and IL-6 responses might be useful for rapid infection fighting, such an enhanced inflammatory response can slip out of control. Several membrane-associated proteins should help to control complement action to prevent the host from unintended injury. These proteins are decay-accelerating factor (DAF), membrane cofactor protein (MCP), complement receptor type 1 (CR1), and CD59. Although the complement cascade is accepted as participating in autoimmunity, little is known concerning the degree to which membrane complement regulatory proteins are capable of modulating complement action. In any case, knockout mice lacking these regulatory proteins display a higher susceptibility to autoimmune injury than wild-type mice. ³²⁵  

Increasingly, the complement system, traditionally known as a peripheral complemental function, migrates to the center of the immune system as a global mediator of immune surveillance, cell homeostasis, and tissue development and repair. The review article by Ricklin et al. ²⁹³ offers additional insights into recent findings concerning the complement system. 

Clinical Diagnosis of Autoimmune Glaucoma

To diagnose glaucoma caused by autoimmunity, the attending physician has no choice but first to exclude all other causes of glaucoma. These include elevated IOP in high-pressure glaucoma. In the case of NTG, pathogenic factors like ischemia, migraine, systemic nocturnal hypotension, or sleep apnea must be excluded as a cause for glaucoma. In summary, it can be said that the diagnosis of autoimmune glaucoma is a diagnosis of exclusion. ³²⁶  

Based on the available data on autoreactivities, one can presume a tissue specificity of the immune response in glaucoma. This specificity derives from the autoantigen that has been initially recognized. ^327,328 A complicating factor is that any given autoimmune disease is heterologous, varying from patient to patient with regard to the course of disease, the severity, and the underlying dysfunction of the immune system. 

To minimize these uncertainties, the creation of a personalized “autoantibody signature” is proposed by some authors. Using the antigen microarray technique, this tool would provide insight into the course of disease, particularly with reference to the patient's response to a therapeutic agent. ¹⁷² A recent study, performed with patients suffering from rheumatoid arthritis (RA), investigated the antibody profile in response to the anti-TNF therapeutic agent etanercept. In summary, a panel of proteomic markers was shown to be able to distinguish responders from nonresponders. ²⁹¹ Suggesting that TNF-α may also be an important factor in the neurodegenerative process of glaucoma, a OHT mouse model was used to investigate the effect of etanercept on glaucoma. In fact, the therapeutic anti-TNF-α antibody etanercept is effective in rescuing RGCs from OHT-induced death. ³²⁹ An investigation of the neuroprotective effect of etanercept could enable the discovery of proteomic factors that are involved in the pathogenesis of glaucoma. A panel of proteomic markers that allows the distinction of healthy from those suffering from glaucoma is an enticing perspective. 

While the investigation of autoantibody profiles as a methodical approach might help to increase understanding of the mechanisms of aberrant immunity in glaucoma, the identification of diagnostic biomarkers would be a great benefit. Proteomic biomarkers are established tools in early diagnosis and are well suited to monitoring the course of numerous diseases. As an example, superoxide dismutase is used in prenatal diagnostics as a biomarker for Down syndrome because of the excessive expression of genes encoded on chromosome 21. ^330,331  

In glaucoma, distinct immunglobulins, such as autoantibodies to alpha-fodrin ¹⁶³ or glutathione S-transferase, ¹⁶⁵ may serve as biomarkers as well as proteins that are significantly linked to cytotoxic conditions in glaucoma. Actually, transthyretin has been identified as a biomarker for glaucoma in aqueous humor of patients with POAG. ³³² Another biomarker for POAG, brain-derived neurotrophic factor (BDNF), could be detected in the sera of glaucoma patients. ³³³  

Markers that correspond with oxidative stress are superoxide dismutase, ^334–339 glutathione peroxidase, ^334,337,340 ascorbic acid, ^334,335,340 α-tocopherol, ^340,341 retinol, ^228,335 nitric oxide, ^{43,57,157,210,221,295,296,337,342–346} 8-hydroxy-20-deoxyguanosine, ³⁴⁷ and malondialdehyde. ^337,348 Detailed information concerning all glaucoma biomarkers identified to date is provided in the review article of Pinazo-Duran et al. ³⁴⁹  

However, as of now, no neuroprotective treatments that directly target pathogenic mechanisms in the retina or optic nerve and no laboratory diagnostics are approved for clinical use against glaucoma. But the progress that glaucoma research has made in recent years gives reason to hope that this will change soon. 

Summary

According to current knowledge, a sole trigger for glaucoma can be excluded. Apparently the lamina cribrosa as the most vulnerable site of the RGCs is of special interest. This may be the position at which impact-induced stress triggers the activation of one or more self-destruct programs, depending on the severity of the impact, WD, DB, or the initiation of a gliosis—or none of these. While it is now understood that many cell types and mechanisms respond to cellular stress, it is not clear where the earliest pathologic events take place and where the key trigger comes from. In this context, the CNS-specific immune system is of particular importance. Under physiological conditions, this ensures a perfect homeostasis. However, this state can be destroyed in the next moment under conditions of cellular stress. As a trigger for cellular stress, a variety of possible factors are under consideration. Among these factors are reactive oxygen species (ROS), protein aggregation, ischemia, vasospasm, deficient neurotrophin supply, and even light. However, the early events in glial activation seem to be crucial for the manifestation of glaucoma. These events need further investigation, because they may be a key to an effective therapy. 

Acknowledgments

The author thanks Barbara Pullmann, Annette Weber, and Ulrich Hoeger for valuable support. The author thanks Paul L. Kaufman for detailed and helpful comments. 

Footnotes

 Disclosure: J. Rieck, None