MMPs in the Neuroretina and Optic Nerve: Modulators of Glaucoma Pathogenesis and Repair?

Abstract

Multiple studies in glaucoma patients and in animal models of spontaneous and experimentally-induced glaucoma, reported changes in the expression and activity of several matrix metalloproteinases (MMPs) in the retina, optic nerve, aqueous humor, and trabecular meshwork. These data have led to the hypothesis that MMPs might be involved in glaucoma onset and/or disease progression. However, reports are conflicting and research aiming at providing a clear definition of their causative role is lacking.

In glaucoma, MMPs are thought to act at two different levels. In the trabecular meshwork, they fine-tune the aqueous humor outflow rate and intraocular pressure, in the neuroretina and optic nerve, however, their role during glaucoma disease progression is much less clear. This review provides a comprehensive overview of the research conducted on the expression and function of MMPs in the retina and optic nerve, and on the elucidation of their potential involvement during glaucoma pathogenesis. Additionally, we describe the insecure balance between detrimental and potential beneficial MMP activities during central nervous system recovery and how MMP-based therapies could help to overcome the current pitfalls in the development of retinal ganglion cell neuroprotection and axon regeneration approaches for the treatment of glaucoma.

Matrix Metalloproteinases

At present, more than 20 human matrix metalloproteinases (MMPs) have been identified and divided into collagenases (MMP-1, ‐8, and ‐13), gelatinases (MMP-2 and ‐9), stromelysins (MMP-3, ‐10, and ‐11), and a heterogeneous group of MMPs (MMP-7, ‐12, ‐20, ‐26, and ‐28), based on their preferential substrates. The membrane-type MMPs or MT-MMPs (MT1- to MT6-MMP) are considered a separate class, regardless of their substrate preference. ¹  

All of these zinc (II)-dependent proteases are characterized by a conserved domain structure, including a catalytic domain, which is shielded by an autoinhibitory prodomain to keep the MMP in an inactive state. ^2–4 Besides activation of MMPs via cleavage/disruption of this prodomain, also transcriptional, posttranscriptional, and epigenetic mechanisms contribute to a strict control of MMP activity, as do compartmentalization, substrate availability, and inhibition of the MMPs. ^4,5 The latter is ensured by, among others, the tissue inhibitors of MMPs (TIMPs [TIMP1–4]), that effectuate a local, reversible inhibition of MMPs in tissues. ^2,5  

Matrix metalloproteinases were named after their ability to cleave and remodel the extracellular matrix (ECM), however, they clearly have a much broader degradome, also comprising proteinases, growth factors, cytokines, cell surface receptors, cell adhesion molecules, and even DNA repair enzymes and mediators of apoptosis. Essentially all MMPs have been linked to disease development, including neurodegenerative disorders such as multiple sclerosis, Huntington's disease, Alzheimer's disease, Parkinson's disease, and so on. ^4,6–8 Despite their detrimental impact during central nervous system (CNS) pathology, MMPs are increasingly recognized as essential players during CNS development and in the healthy adult brain. Indeed, a well-balanced MMP/TIMP expression has been shown to confer to neurogenesis, neurite outgrowth, myelinogenesis, angiogenesis, cell survival, and adult neural progenitor migration. ^7–13 Importantly, these processes do not only occur during development, but also after CNS injury, thus promoting a reparative role for MMPs during recovery from CNS injury. Taken together, there is ample evidence corroborating MMPs as fine regulators of CNS physiology and pathology, and any disturbance of their activity may result in profound implications on cell–cell and cell–ECM interactions. 

Glaucoma

Worldwide, more than 60 million people are affected by glaucomatous optic neuropathies, making this multifactorial neurodegenerative disease the second most important cause of irreversible blindness. ¹⁴ Glaucomatous neurodegeneration, characterized by progressive degeneration of the optic nerve and apoptosis of retinal ganglion cell (RGC) somata, is thought to be inflicted by a combination of genetic predisposition and age-related and environmental stressors. ^15,16 The most prevalent and important risk factor, as well as the sole target for clinical intervention, is elevated intraocular pressure (IOP). Although many patients benefit from IOP-lowering therapies, some patients continue to lose vision in spite of all current treatments. ^17,18 Furthermore, despite intensive research efforts, the precise cellular and molecular events translating elevated IOP and other contributing stressors into progressive RGC death, and the time course of glaucoma pathogenesis, still remain largely elusive. Expansion of our knowledge about glaucoma pathogenesis and new insights into contributing molecules/processes, seem of utmost importance for the development of innovative therapies that can preserve or restore vision. 

The complex etiology of the disease is reflected in the divergent array of glaucoma models, both genetic and experimentally induced, that are available to study its pathogenesis in laboratory animals. ^19,20 Although each of these models mimics only a certain aspect of the disease, they do provide valuable insights in the underlying causes and mechanisms leading to RGC degeneration. While ocular hypertension-induced glaucoma models and optic nerve crush/transection are representative for the large cohort of patients with axonal damage inflicted by elevated IOP (or trauma), N-methyl-D-aspartate (NMDA)- and kainic acid (KA)-mediated models and ischemia-reperfusion injury represent excitotoxicity-induced RGC damage and RGC death caused by vascular insufficiency, respectively. Of note, although covering the variety in initial insults contributing to glaucoma pathogenesis, this richness of models might also be (partially) responsible for the seemingly contradicting results that are sometimes emerging from different studies investigating glaucoma pathology. 

Current Knowledge on MMP Expression and Activities in the Posterior Segment of the Glaucomatous Eye

In many cases, glaucomatous damage to the optic nerve and retina is caused by a pathological IOP elevation. Matrix metalloproteinases are the major matrix degrading enzymes and have been described as important modulators of IOP in the healthy human eye, functioning in a feedback mechanism that continuously remodels the trabecular meshwork ECM composition in order to maintain a stable aqueous humor outflow resistance and IOP. ²¹ However, IOP homeostasis is only one aspect of glaucoma in which MMPs are involved. Regardless of the primary cause of the retinal damage, glaucomatous optic neuropathies share a common endpoint involving progressive atrophy of the optic nerve and apoptosis of RGC somata. Both in the retina and optic nerve, MMPs are known to be expressed by various glial as well as neuronal cell types (Tables 1, 2). It is therefore very likely that, besides their role as trabecular ECM remodelers in the anterior segment of the eye, MMPs also contribute to the degenerative events manifesting in the posterior segment. 

Table 1

 

Matrix Metalloproteinase Expression and Activity in The Healthy Versus Glaucomatous Retina

	Expression/Activity in the Retina	Species/Glaucoma Model	Technique(s)	Reference
Human patients
MMP-1	Nuclear and plexiform layers	Human	IHC	24
MMP-2	RCGs and their axons in the NFL	Human	IHC	23
MMP-3	No detectable immunoreactivity	Human	IHC	24
MMP-9	No detectable immunoreactivity	Human	IHC	24
MT1-MMP	Photoreceptor outer segments	Human	WB	25
TIMP-1	No detectable immunoreactivity	Human	IHC	24
	Glaucoma: RGC somata and axons; no immunoreactivity in astrocytes	Human (POAG)	IHC	23
TIMP-2	Inner and outer nuclear layers	Human	IHC	24
Ocular hypertension-induced glaucoma models in rodents
MMP-3	Glaucoma: increased MMP-3 expression	Rat (hypertonic saline episcleral vein injection)	Microarray, qRT-PCR	37
MMP-9	Glaucoma: increased MMP-9 expression in apoptotic cells in GCL, correlating with the degree of IOP exposure	Rat (hypertonic saline episcleral vein injection)	IHC	26
	Glaucoma: decreased MMP-9 expression	Rat (hypertonic saline episcleral vein injection)	Microarray, qRT-PCR	38
TIMP-1	GCL Glaucoma: increased immunoreactivity, correlating with the degree of IOP exposure	Rat (hypertonic saline episcleral vein injection)	IHC	26
	Glaucoma: increased TIMP-1 expression	Rat (hypertonic saline episcleral vein injection)	qRT-PCR	32
	Glaucoma: increased TIMP-1 expression	Rat (hypertonic saline episcleral vein injection)	Microarray, qRT-PCR	37
	Glaucoma: increased TIMP-1 expression	Rat (hypertonic saline episcleral vein injection)	Microarray, qRT-PCR	38
Optic nerve injury-induced glaucoma models in rodents
MMP-2	Glaucoma: MMP-2 expression remains unchanged	Rat (optic nerve transection)	WB, IHC	34
MMP-3	Glaucoma: increased MMP-3 expression	Rat (optic nerve crush, optic nerve transection)	Microarray, qRT-PCR	43
MMP-9	Glaucoma: increased MMP-9 expression in GCL	Rat (optic nerve transection)	WB, IHC	34
Ischemia-induced glaucoma models in rodents
MMP-2	Glaucoma: MMP-2 expression remains unchanged	Mouse (optic nerve ligation)	WB	30
	Glaucoma: increased pro-MMP-2 expression	Rat (ischemia-reperfusion injury)	Zymography	31
MMP-9	Dot-like staining pattern in GCL Glaucoma: increased eGFP+ area and intensity	Mouse (optic nerve ligation)	MMP-9-eGFP reporter mice	27
	Glaucoma: increased MMP-9 expression in RGCs	Mouse (ischemia-reperfusion injury)	IHC, ISZ	28
	Glaucoma: increased MMP-9 expression in reactive astrocytes	Mouse (optic nerve ligation)	WB, IHC	29, 30, 33
	Glaucoma: increased pro-MMP-9 expression in interphotoreceptormatrix	Rat (ischemia-reperfusion injury)	Zymography, IHC	31
TIMP-1	Glaucoma: TIMP-1 expression remains unchanged	Rat (ischemia-reperfusion injury)	Reverse zymography	31
	Glaucoma: TIMP-1 expression remains unchanged	Mouse (optic nerve ligation)	WB	33
TIMP-2	Glaucoma: TIMP-2 expression remains unchanged	Mouse (optic nerve ligation)	WB	33
	Glaucoma: TIMP-2 expression remains unchanged	Rat (ischemia-reperfusion injury)	Reverse zymography	31
Excitotoxicity-induced glaucoma models in rodents
MMP-2	Glaucoma: MMP-2 activity remains unchanged	Rat (intravitreal NMDA injection)	Zymography	32
	RGCs, Müller glia, astrocytes Glaucoma: slightly increased MMP-2 protein levels measured via WB, yet no change in activity levels measured via zymography	Mouse (intravitreal KA injection)	WB, zymography, IHC	33
MMP-9	Glaucoma: increased in expression of pro- and active MMP-9 in RGCs	Rat (intravitreal NMDA injection)	Zymography, ISZ	32
Miscellaneous
MMP-2	RGCs and their axons in the NFL	Monkey	IHC	22
	Photoreceptor inner and outer segments; weak immunoreactivity in OLM	Horse	IHC	36
MMP-9	GCL, IPL, and photoreceptor segments	Horse	IHC	36
MT1-MMP	Photoreceptor inner segments	Horse	IHC	36
	NFL	Mouse (P0)	ISH	50
	Inner retinal layer	Rabbit	IHC	51
TIMP-1	RGC somata and axons in NFL, no immunoreactivity in astrocytes in NFL	Monkey	IHC, ISZ	22
TIMP-2	Photoreceptor inner and outer segments	Horse	IHC	36

 This table provides an overview of reported MMP expression/activity in the healthy and glaucomatous retina, reported from in vivo studies, with corresponding references. Expression data in human patients and data obtained in animal models of glaucoma are categorized under different subheadings, with the latter being subdivided according to the type of initial insult leading to the glaucomatous retinal damage. Of note, optic nerve ligation has been grouped under the subheading ‘ischemia-induced animal models for glaucoma', however, as not only the ophthalmic artery, but also the optic nerve is ligated during this procedure, RGC degeneration in this model is likely due to a combination of ischemia and axonal compression. IHC, immunohistochemistry; ISZ, in situ zymography; ISH, in situ hybridization; WB, western blot; qRT-PCR, quantitative RT-PCR.

Table 2

 

MMP Expression and Activity in The Healthy Versus Glaucomatous Optic Nerve

	Expression/Activity in the Optic Nerve	Species/Glaucoma Model	Technique(s)	Reference
Human patients
MMP-1	Cytoplasm of few glial cells; faint staining of glial processes around axons and in pial septae Glaucoma: increased number of MMP-1+ glial cells	Human (PAOG, NTG)	IHC	60
	Few astrocytes Glaucoma: increased immunoreactivity in astrocytes and ECM in cribriform plates; axons; small vessels and pial septae	Human (POAG)	IHC	23
MMP-2	Few glial cells; faint staining of glial processes around axons Glaucoma: increased intensity of immunostaining and number of MMP-2+ cells; increased intensity of immunostaining of glial processes around axons and along pial blood vessels Intracytoplasmatic immunostaining of glial cells is more intense in areas with preserved axons	Human (POAG, NTG)	IHC	60
	Astrocytes; axons, higher immunoreactivity in unmyelinated (pre)laminar regions than in myelinated postlaminar regionGlaucoma: MMP-2 expression remains unchanged	Human (POAG)	IHC	23
MMP-3	Few glial cells; faint staining of glial processes around axons Glaucoma: increased immunoreactivity in astroglial cells, glial processes around axons and along pial septae Immunoreactivity is more intense in areas with preserved axons	Human (POAG, NTG)	IHC	60
	Perivascular cells	Human (POAG)	IHC	23
MMP-7	No detectable immunoreactivity	Human (POAG)	IHC	23
MMP-9	No detectable immunoreactivity	Human (POAG)	IHC	23
MMP-12	No detectable immunoreactivity	Human (POAG)	IHC	23
MT1-MMP	Few astrocytes; granular immunostaining associated with blood vessels; no immunoreactivity detected in axons or ECM Glaucoma: increased immunoreactivity in astrocytes	Human (POAG)	IHC	23
TIMP-1	Astrocytes and axons Glaucoma: TIMP-1 expression remains unchanged	Human (POAG)	IHC	23
TIMP-2	Astrocytes and axons Glaucoma: TIMP-2 expression remains unchanged	Human (POAG)	IHC	23
Ocular hypertension-induced glaucoma models in monkey
MMP-1	Few astrocytes; small blood vessels Glaucoma: increased immunoreactivity in reactive astrocytes and ECM; many quiescent astrocytes in myelinated nerve express mRNA	Monkey (laser scarification of trabecular meshwork)	ISH, IHC	22
MMP-2	Few astrocytes; axons Glaucoma: decreased MMP-2 immunoreactivity in nerve bundles due to loss of axons; MMP-2 expression in astrocytes remains unchanged	Monkey (laser scarification of trabecular meshwork)	ISH, IHC	22
MMP-3	Small blood vessels	Monkey (laser scarification of trabecular meshwork)	IHC	22
MMP-9	Some oligodendrocytes in the nerve bundles in the myelinated nerve Glaucoma: MMP-9 expression remains unchanged	Monkey (laser scarification of trabecular meshwork)	IHC	22
MT1-MMP	Low immunoreactivity/mRNA in few astrocytes and around small blood vessels Glaucoma: increased immunoreactivity/mRNA in reactive astrocytes	Monkey (laser scarification of trabecular meshwork)	ISH, IHC	22
TIMP-1	Astrocytes and axons Glaucoma: decreased immunoreactivity due to loss of axons	Monkey (laser scarification of trabecular meshwork)	ISH, IHC	22
TIMP-2	Astrocytes and axons Glaucoma: decreased immunoreactivity due to loss of axons	Monkey (laser scarification of trabecular meshwork)	ISH, IHC	22
Ocular hypertension- and axonal injury-induced glaucoma models in rodents
MMP-2	Glaucoma: increased MMP-2 expression	Rat (hypertonic saline episcleral vein injection)	Microarray, qRT-PCR	62
	Glaucoma: MMP-2 expression remains unchanged	Rat (optic nerve crush)	qRT-PCR, zymography	63
MMP-3	Glaucoma: increased MMP-3 expression, primarily associated with GFAP+ astrocytes	Rat (optic nerve crush)	qRT-PCR, IHC	63
MMP-7	Glaucoma: increased MMP-7 expression, primarily associated with GFAP+ astrocytes	Rat (optic nerve crush)	qRT-PCR, IHC	63
MMP-9	Glaucoma: increased pro-MMP-9 expression, primarily associated with GFAP+ astrocytes	Rat (optic nerve crush)	qRT-PCR, IHC, zymography	63
MMP-12	Glaucoma: increased MMP-12 expression, primarily associated with GFAP+ astrocytes	Rat (optic nerve crush)	qRT-PCR, IHC	63
TIMP-1	Glaucoma: increased TIMP-1 expression	Rat (hypertonic saline episcleral vein injection, optic nerve transection)	Microarray, qRT-PCR	62

 This table provides an overview of MMP expression/activity in the healthy and glaucomatous optic nerve, reported from in vivo studies, with corresponding references. It illustrates the variety in species, glaucoma models, and experimental details that have contributed to a very rich data set about MMP expression/activity, but that have also hampered a straightforward interpretation of these data due to conflicting outcomes. Expression data in human patients and data obtained in animal models of glaucoma are categorized under different subheadings. NTG, normal tension glaucoma.

MMP Expression and Activities in the Glaucomatous Retina

The gelatinases, and notably MMP-9, are by far the best studied MMPs in the retina. This might be largely due to the fact that gelatin zymography allows straightforward quantification of gelatinase activity, as such circumventing sensitivity and specificity issues related to anti-MMP antibodies required for Western blot. Limited research has also been performed on the expression of MMP-1, -3, MT1-MMP, and TIMP-1 in the healthy and glaucomatous retina, but data about other MMPs/TIMPs are virtually absent, as apparent from Table 1. 

MMP Expression in the Retina of Glaucoma Patients.

Thus far, no more than four studies have investigated MMP localization in the human retina ^22–25 (Table 1). These failed to detect any MMP-3 or -9 expression in the healthy retina, ²⁴ yet found MMP-2 to be expressed in RGC somata and their axons. ²³ Matrix metalloproteinase 1 was observed both in nuclear and plexiform layers, ²⁴ while MT1-MMP was only present in the photoreceptor outer segments ²⁵ and TIMP-2 expression was observed within the inner and outer nuclear layers of the healthy retina. ²⁴ In contrast, TIMP-1 expression, which was undetectable in the healthy retina, appeared to be upregulated in RGC somata and axons of primary open-angle glaucoma (POAG) patients. ²³  

MMP Expression in the Retina in Animal Models of Glaucoma.

In the retina, like in any other tissue, the status and composition of the ECM is believed to modulate the synthesis and release of MMPs, and vice versa. This MMP-ECM interplay affects cell–cell and cell–ECM interactions, which may ultimately determine cell survival versus death. ²⁶ Several studies in rodent models of glaucoma, induced by ischemia-reperfusion injury, ^27–31 NMDA- and KA-mediated excitotoxicity, ^32,33 optic nerve transection, ³⁴ or ocular hypertension, ²⁶ revealed a positive correlation between RGC death and MMP-9 activity in the RGC layer (Table 1), whereas MMP-9 activity negatively correlated with laminin immunostaining in the inner limiting membrane. Increased MMP-9 activity in the RGC layer plays a key role in the promotion of RGC death, as it induces degradation of laminin, resulting in abrogation of integrin-mediated survival signaling pathways and detachment-induced apoptosis of RGCs (“anoikis”) ^26,28,35 (Fig. 1). Indeed, MMP-9–deficient mice subjected to an ischemia-reperfusion injury model are protected from RGC death and laminin degradation, ²⁷ providing the ultimate evidence for MMP-9 as a crucial regulator of RGC death. In contrast, MMP-2 deficiency did not protect from RGC death after ischemia-reperfusion injury ²⁷ and MMP-2 activity/expression was reported to remain unchanged after excitotoxic injury, ischemia-reperfusion injury, or optic nerve transection. ^{22,30,32–34} However, the latter is still under debate, as others revealed increased MMP-2 activity within the first hours post ischemia-reperfusion or excitotoxic injury ^31,33 (Table 1). 

Figure 1

 

Schematic representation of known MMP and TIMP activities in the glaucomatous retina. Proposed MMP-9 working mechanisms (blue): various insults, including elevated IOP and ischemia, result in the elevation of retinal glutamate levels and stimulate the production of IL-1 by astrocytes, inducing an upregulation of MMP-9 synthesis in astrocytes and/or RGCs. In addition, IL-1 also increases the production of NO, which can activate extracellular pro-MMP-9. Finally, MMP-9 expression increases as elevated IOP results in a diminished deposition of TGF-β2, a known MMP inhibitor, in the ECM. This increased MMP-9 activity in the RGC layer abrogates laminin-integrin signaling, mediated by focal adhesion kinase (FAK), phosphatidylinositide 3-kinase (PI3K), and protein kinase B (PKB) kinases, resulting in decreased expression of the antiapoptotic protein B-cell lymphoma-extra-large (Bcl-xL), and detachment-induced apoptosis of RGCs. Proposed TIMP-1 working mechanisms (red): Ocular hypertension–induced elevation of MMP-9 activity in the RGC layer is accompanied by elevated TIMP-1 expression in RGCs. This TIMP-1 elevation is likely to act in a dual way. First, TIMP-1 will inhibit MMP-9 and as such keep pro-apoptotic MMP-9 activity within limits, second, TIMP-1 might also promote RGC survival by directly suppressing pro-apoptosis signaling in an MMP-independent manner. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; OPL, outer plexiform layer; PRL, photoreceptor layer; PIP2, phosphatidylinositol 4,5-bisphosphate; PIP3, phosphatidylinositol 3,4,5-triphosphate.

Colocalization studies by means of immunohistochemistry confirmed reactive astrocytes as the major source of MMP-9 expression after retinal ischemia-reperfusion injury or KA-mediated excitotoxicity in mice, ^29,33 and excluded endothelial cells, microglia, and RGCs. ²⁹ In contrast, in situ zymography pointed out that RGCs exhibit increased MMP-9 activity after ischemia-reperfusion injury or NMDA-mediated excitotoxicity in rats, ^28,32 while the equine retina revealed MMP-9 expression localizing to the RGC layer, inner plexiform layer (IPL), and photoreceptor segments ^24,36 (Table 1). The cellular source(s) of the observed MMP-9 elevation in the glaucomatous retina thus remain obscure and more research with carefully validated anti-MMP-9 antibodies is required. For MMP-2, constitutive expression has been located to many cell types. In mice, MMP-2 was found in RGCs, Müller cells, and astrocytes. ³³ Also in the monkey retina, MMP-2 has been detected in RGC somata and their axons, whereas MMP-2–positive astrocytes were only found in the optic nerve head (ONH). ²² In the equine retina, MMP-2 expression was observed in both the inner and outer segments of the photoreceptors and in the adjacent outer limiting membrane (OLM) ³⁶ (Table 1). 

Notably, while MMP-9 expression is low in the healthy retina, prominent expression of its major inhibitor, TIMP-1, is seen in RGC somata and their axons in the nerve fiber layer (NFL) ^22,23,26 (Table 1). Moreover, the increase in MMP-9 activity in response to elevated IOP was repeatedly reported to be accompanied by an increased TIMP-1 expression, which was found to display a positive correlation with the degree of IOP exposure. ^26,32,37,38 While constitutive TIMP-1 expression was suggested to support synaptic plasticity at the RGC terminals and maintenance of an intact ECM (i.e., by keeping MMP activity within limits), ²² the increase in TIMP-1 levels might facilitate its neuroprotective effects on RGCs, either via inhibition of detrimental MMP-9 activity, or via MMP-independent antiapoptotic actions, ²⁶ as shown elsewhere in the CNS ^39,40 (Fig. 1). Expression of TIMP-2, which seems rather confined to the outer retina, was reported to remain unchanged ^24,31,33,36 (Table 1). 

Several theories exist, which have in common the inflammatory cytokine IL-1, about how MMP-9 activity is induced in experimental models of RGC degeneration (Fig. 1). Glaucomatous injury in the retina stimulates the production of IL-1β by astrocytes, which could upregulate MMP-9 synthesis in astrocytes (and/or RGCs). ³⁰ In addition, IL-1 also increases nitric oxide (NO) production, ^30,33,41 which may activate extracellular pro-MMP-9 via S-nitrosylation and oxidation of the enzyme's cysteine switch. ^32,33,42 Alternatively, MMP-9 expression might be augmented directly or indirectly, via IL-1 signaling cascades, by increased retinal glutamate levels, which can be induced by various stimuli, including injury, ischemia, and elevated IOP. ^26,33 Finally, in addition to IL-1, other cytokines such as TNF-α and TGF-β2 are likely to contribute to the upregulation of MMP expression in the glaucomatous retinas as well. In rodent models for glaucoma, as well as in human patients, elevated levels of TNF-α and decreased levels of TGF-β2 have been repeatedly noted in the glaucomatous retina, and these cytokines are well known as inducer and inhibitor of MMP transcription, respectively. ^26,43–46 However, in contrast to reported observations in the optic nerve head (see below), a causal relationship between altered TNF-α or TGF-β2 expression and increased MMP transcription has not yet been described in the glaucomatous retina. 

Of note, alternative/additional mechanisms of action of MMP-9, besides anoikis, have never been excluded. These could comprise a direct contribution of MMP-9 to apoptotic signaling cascades, MMP-9-dependent cleavage of precursors of neurotoxic proteins, MMP-9–mediated increases in Ca²⁺ influx or dysfunctional retrograde transport of cell-survival factors secondary to the loss of ECM attachment. ^27,32,47  

Limited attention has been devoted to expression of other MMPs (i.e., MMP-3 and MT1-MMP) in the healthy and glaucomatous retina so far. A microarray analysis on the retina of rats subjected to an ocular hypertension glaucoma model, revealed 3.5-fold upregulated MMP-3 mRNA levels at 35 days post induction of ocular hypertension. ³⁷ Likewise, MMP-3 mRNA levels were strongly increased after optic nerve crush and optic nerve transection in rats (i.e., over 50-fold and over 250-fold, respectively). ⁴³ The cellular origin of MMP-3 expression has not yet been discovered. Nevertheless, these data, suggesting MMP-3 to be an important player during glaucomatous damage, together with the described pro- and antiapoptotic actions of MMP-3 in the CNS, ^48,49 indicate that MMP-3 is worth investigating. For MT1-MMP, knowledge is restricted to expression studies in the healthy retina, where MT1-MMP has been reported in the NFL of newborn (P0) mice and adult rabbits ^50,51 and in the photoreceptor segments of equine retina. ³⁶  

Overall, the expression of MMP-2, -3, and -9 in the healthy and glaucomatous retina hints that MMPs might be important players in retinal health and disease. However, largely due to a lack of functional studies, it is currently impossible to define their exact or causative role. Whereas MMP-9 has a clear negative impact on RGC survival, the role of MMP-2 in glaucomatous damage in the retina remains obscure. Although MT1-MMP, MMP-1, and -3 were shown to be expressed in the retina, the latter even being upregulated upon axonal damage, we can only speculate about their function. For TIMP-1, a neuroprotective role in the CNS has been suggested, both via modulation of MMP activity as well as via MMP-unrelated pathways, but whether a similar function exists in the retina remains currently elusive. 

MMP Expression and Activities in the Glaucomatous Optic Nerve (Head)

As RGC axons exit the primate (or canine) eye to the optic nerve, their sole support and protection is the lamina cribrosa, making the ONH a fragile site in an otherwise rigid corneoscleral shell ⁵² and the major site of impact of the mechanical stress generated by elevated IOP. The lamina cribrosa is composed of cribriform plates that are aligned to form channels for the RGC axons, and an ECM with collagenous columns, to which glial cells (glial fibrillary acidic protein⁺ [GFAP⁺] astrocytes and GFAP^- lamina cribrosa cells) are anchored by a basement membrane. ⁵³ Upon exposure to excessive mechanical stress, characteristic changes take place in the lamina cribrosa: the cribriform plates collapse and the glial cells shift their production of ECM components to those that are characteristic for the glaucomatous phenotype and increase their secretion of MMPs. ^22,53–57 As a consequence, the ECM is remodeled, which adversely affects the capacity of the lamina cribrosa to support RGC axons and predisposes RGCs to axonal compression, arrest of axoplasmic flow, and apoptosis. ^53,55,56  

Although the rodent ONH only has a rudimentary lamina cribrosa, it does possess a glial lamina with a collagenous composition and an ultrastructural organization that closely resembles that of the primate. Indeed, the ONH still appears to be the site of early pressure-induced optic nerve injury in rodents, and the above mentioned mechanisms of pathological remodeling seem conserved. ⁵⁸  

MMP Expression in the ONH of Glaucoma Patients.

Despite the restricted number of studies investigating in vivo MMP/TIMP expression and activities in the human glaucomatous optic nerve, we do have some basic, indirect insights into the actions of MMPs that contribute to optic nerve axonal degeneration in glaucoma patients. 

Human ONH glia were reported to increase their secretion of cytokines, such as TGF-β1 ^53,57,59 and TNF-α, ^44,45,54,60 in response to glaucomatous damage, which may act in an auto- or paracrine way to promote other stretch-induced reactions, including stimulation of MMP-2 secretion, leading to progressive ECM remodeling (Fig. 2). Indeed, increased immunoreactivity for MMP-2 has been observed in the lamina cribrosa of glaucomatous patients and in vitro in ONH glial cells undergoing mechanical stretching, corroborating ONH glial cells, next to the RGC axons, as the source of MMP-2 ^53,54,57,60 (Table 2). In addition to MMP-2, also MT1-MMP and TIMP-3 were reported to be upregulated in an in vitro culture of human lamina cribrosa cells exposed to cyclic mechanical stretching. ⁵⁷  

Figure 2

 

Schematic representation of suggested MMP and TIMP activities in the glaucomatous optic nerve. In response to the mechanical stress generated by elevated IOP, lamina cribrosa cells and astrocytes increase their secretion of TGF-β1 and TNF-α, which act in an auto- or paracrine manner to induce MMP-2 expression and ECM remodeling in the optic nerve head (1). Moreover, astrocytes also express MMP-1 and MT1-MMP, while undergoing a transition from a quiescent to a reactive phenotype. This transition involves detachment from the basement membrane and relocation throughout the axon bundles, processes that are both facilitated via proteolysis of the ECM and cell-surface adhesion molecules by MMP-1 and MT1-MMP (2). If not counterbalanced by TIMP-1, expressed in RGC axons and astrocytes, MMP-1 will continue to degrade the ECM, resulting in a lamina cribrosa extracellular environment that is no longer able to support axonal/neuronal survival (3).

In the healthy human optic nerve, quiescent astrocytes express low levels of MMP-1, -2, and MT1-MMP, as well as TIMP-1 and -2. ^23,60 In the glaucomatous optic nerve, astrocytes undergo a transition from a quiescent to a reactive phenotype, which involves increased expression of MMP-1 and MT1-MMP, changes in cell shape and relocation in the nerve bundles. ^23,60 The increased expression of MT1-MMP on their plasma membrane was suggested to allow astrocytes to detach and migrate from the underlying basement membrane, by cleavage of ECM and cell-surface adhesion molecules at their migration front and/or by altering their cytoskeleton (Fig. 2). Of note, the membrane-bound MT1-MMP can not only directly exert these actions, but can also indirectly activate MMP-2 expressed by RGC axons and ONH glial cells, assuring that MMP-2 is activated at the cell–ECM interface. ²³ Subsequently, MMP-1 is believed to permit migration of astrocytes throughout the ECM of the lamina cribrosa into the nerve bundles, where MMP-1, if not counterbalanced by TIMP-1, will continue to degrade the ECM around the axons and consequently interfere with neuronal survival ²³ (Fig. 2). Notably, a similar neurotoxic effect of MMP-1, released by astrocytes, was observed in organotypic rat spinal cord and human brain astrocyte cultures, via destruction of the ECM and/or activation of cytokines. ⁶¹  

MMP Expression in the ONH in Animal Models of Glaucoma.

Several lines of evidence, obtained from research in various animal models of glaucoma, support the above-mentioned modes of action of MMPs in the glaucomatous ONH. The TGF-β1 receptor, MMP-2 and TIMP-1 were shown to be augmented in the optic nerve during the onset stage in an IOP-induced glaucoma model in rats, ⁶² and in the healthy monkey optic nerve, low levels of MMP-1, -2, and MT1-MMP, as well as TIMP-1 and -2, expressed by quiescent astrocytes, were revealed. ²² In the latter study, Agapova et al. ⁶³ confirmed their previous work on human samples. In addition, by comparing MMP expression in the optic nerve of monkeys with ocular hypertension-induced glaucoma to monkeys subjected to optic nerve transection, they were able to prove that the observed ECM remodeling and elevated MMP-1 and MT1-MMP expression in the optic nerve during glaucoma, occur in response to elevated IOP and are not secondary to axonal loss. ²²  

In addition to these animal studies, Hughes et al. ⁶⁴ investigated the involvement of MMPs in Wallerian degeneration induced by ONC in rats and found MMP-9 expression in optic nerve astrocytes, which is in contrast to the expression pattern found in the glaucomatous monkey ONH, where MMP-9 was expressed by oligodendrocytes. ²² Matrix metalloproteinase-3, in accordance to expression studies in monkey, was observed in glial cells and blood vessels and upregulated in glia upon glaucomatous damage to the optic nerve. ^22,64 Matrix metalloproteinase-7 and -12, finally, are both expressed by astrocytes upon optic nerve crush in rats ⁶³ (Table 2). 

Overall, even when integrating the findings from human and animal studies, the available data about the expression of MMPs in the glaucomatous optic nerve are sparse and puzzling (Table 2). There is a general consensus that elevated IOP induces an increase in MMP-1 and MT1-MMP expression levels, which might in turn negatively impact on the optic nerve microenvironment. However, reports disagree on the effect of ocular hypertension–induced mechanical stretching on MMP-2 and TIMP expression in the ONH: While some describe an increase in MMP-2 and TIMP-1 levels, ^53,57,60,62 others note constitutive expression of MMP-2, TIMP-1, and -2, ^22,23,64 or even decreased levels of TIMP-1 and -2 due to axon loss. ²²  

Can MMP Modulation Facilitate Neuroprotective and Regenerative Glaucoma Therapies?

Current glaucoma therapies are all directed toward a sustained reduction of IOP, and slow down or halt glaucomatous disease progression in many but not all patients. ^17,18 Nonetheless, glaucoma typically leads to irreversible loss of vision because of the inability of adult mammals to repair or regenerate damaged RGCs. An effective treatment for glaucoma, able to restore vision, would involve neuroprotection and/or replacement of damaged RGCs in the retina, as well as induction of their axonal regenerative capacity in the optic nerve. ⁶⁵ Here, we review current evidence for a carefully balanced MMP activity as crucial modulator of the receptiveness of the retinal and optic nerve extracellular environment to neuroprotection, stimulation of axon regeneration and cell transplantation therapy. 

Can MMPs Modulate RGC Neuroprotection in the Retina?

Matrix metalloproteinase-9 deficiency was already mentioned earlier in this review for its neuroprotective effects in a rat ischemia-reperfusion model for glaucoma ²⁷ (Fig. 1). In retinal excitotoxicity models in rats and mice, the broad-spectrum MMP inhibitor GM6001 prevented pathological remodeling of the inner limiting membrane and detachment-induced apoptosis of the RGCs, confirming that MMPs, most likely MMP-9, indeed contribute to RGC death. ^28,32,33 Specific MMP-9 inhibitors, or agents acting on downstream targets in the MMP-9–induced RGC apoptotic pathway, might thus serve as neuroprotective agents in retinal neurodegenerative diseases, including glaucoma, retinal artery/vein occlusion, and ischemic optic neuropathy. ³² Also TIMP-1 expression has been shown to be upregulated in the retina of rats exposed to ocular hypertension. ^26,37,38 This TIMP-1 elevation is likely to reflect the canonical coregulation of MMPs and TIMPs, designed to keep pro-apoptotic MMP-9 activity within limits, ²⁶ but, in addition, TIMP-1 might also promote RGC survival by directly suppressing apoptosis signaling pathways, in an MMP-independent manner ^26,39,40 (Fig. 1). 

Can MMPs Modulate Axonal Regeneration in the Optic Nerve?

Upon injury in the mammalian CNS, the ECM transforms from a growth-permissive to an inhibitory environment, thereby largely preventing CNS neurons from regenerating axons to their former targets. This failure of the mammalian CNS to regenerate axonal projections is multifactorial. In addition to the diminished intrinsic growth capacity of adult CNS neurons, ⁶⁶ the injured CNS becomes deprived of neurotrophic factors and/or their receptors, ⁶⁷ and glial scar- and myelin-derived inhibitory molecules form a physicochemical barrier for regenerating axons. ^68,69  

Matrix metalloproteinases have been suggested as key facilitators of successful axonal regeneration for several reasons (Fig. 3). First of all, MMPs add to the clearance of cellular and matrix debris at the site of injury. Second, all major constituents of the glial scar are substrates of at least one MMP ^69,70 and increased MMP activity reduces glial scarring and associated inhibitory molecules. In particular, the chondroitin sulphate proteoglycans (CSPGs) have been denoted for their strong inhibitory impact on axonal regeneration ⁶⁸ and are subject to cleavage by MMPs. Indeed, the repulsion of neurite growth cones and “masking” of neurite-promoting laminin by CSPGs, can be (partially) abrogated by MMPs. ^69,71,72 Third, MMPs are also able to degrade myelin-derived inhibitory ligands released by degenerating CNS axons, such as neurite outgrowth inhibitor (Nogo), myelin-associated glycoprotein (MAG), oligodendrocyte myelin glycoprotein (OMgp), and myelin basic protein (MBP). ^{8,68,69,73,74} Fourth, besides disarming inhibitory ECM and signaling molecules, MMPs can also unmask, activate, or release others that have a beneficial effect on CNS repair. As such, they can indirectly provide neurotrophic support to regenerating axons via the release of sequestered growth factors (e.g., release of bFGF from ECM heparin sulphate proteoglycans) or the conversion of inactive progrowth factors to their active forms (e.g., nerve growth factor [NGF], brain-derived neurotrophic factor [BDNF]). ^8,69 Finally, as the directed outgrowth of axons requires extensive motility and infiltration within the nervous tissue, MMPs can invoke ‘focalized' proteolysis at the growth cone, ⁷⁵ thereby reorganizing the ECM to facilitate attachment and motility of the growing axon ⁷⁶ (Fig. 3). 

Figure 3

 

Schematic representation of suggested MMP and TIMP activities in the regenerating optic nerve. Matrix metalloproteinase-2, -9, and to a lesser extent MMP-1 and -3, are upregulated at the site of injury and in the proximal stump of the optic nerve, presumably by (an) unknown factor(s), released by the numerous collateral axonal sprouts that stimulate astrocytes to increase their MMP expression. Initially, MMP-2 and -9 contribute to the removal of tissue debris during early postinjury inflammation (1). Later on during the recovery phase, a well-balanced MMP activity might promote axonal regeneration via degradation of glial scar–inhibitory ligands, such as CSPGs (2); via degradation of myelin-derived inhibitory ligands, such as Nogo, MAG, OMgp and MBP (3); via unmasking, release and/or activation of growth-promoting molecules such as bFGF, pro-NGF, and pro-BDNF (4); and/or via focalized ECM proteolysis at the growth cones of regenerating axons (5).

Although thus far only described in the brain and spinal cord, it can be assumed that these MMP actions can be extrapolated to the entire CNS, including the retina and optic nerve. Indeed, stimulation of axonal regeneration after optic nerve transection in rats, by intravitreal transplantation of a peripheral nerve segment, coincides with significantly enhanced MMP activity in the regenerating optic nerve as compared with the nonregenerating optic nerve. ⁶⁸ More specifically, MMP-2 and -9, and to a lesser extent also MMP-1 and -3, are upregulated in the proximal optic nerve stump and at the site of the optic nerve injury, while TIMP-1 and -2 are suppressed. This immediate local upregulation (coinciding with post injury inflammation) of MMP-2 and -9 by reactive astrocytes is thought to clear tissue debris, in concert with invading macrophages (Fig. 3). The later burst in gelatinase activity (i.e., 8 days post injury), which colocalizes with ECM compounds of the glial scar, might serve to break down and dissolve the glial scar, thereby clearing the path for regenerating axons ^68,77 (Fig. 3). Remarkably, gelatinase activity only rarely colocalizes with regenerating optic nerve axons, in contrast to regenerating peripheral nerves, where growth cones secrete MMP-2 (and MMP-3) to facilitate the progression of dorsal root ganglia cell axons on the peripheral nerve basal lamina. ^75,78 Instead, reactive astrocytes at the site of injury appear to be the major source of gelatinases in the CNS. ^68,77 Ahmed et al. ⁶⁸ hypothesized that the numerous collateral sprouts in the proximal stump of the transected optic nerve and the growth cones at the front of the regenerating axons, release a, thus far unknown, factor that conditions the surrounding astrocytes to increase MMP and downregulate TIMP expression (Fig. 3). The result is a resolution of glial- and myelin-derived growth inhibitory ligands, and RGC axon regeneration is less hampered. ⁶⁸ Also, other experimental approaches that succeeded to induce axonal regeneration in the optic nerve, attributed at least part of their success to MMP-related activities. Transplantation of neural tube-derived chicken embryonic stem cells at the site of optic nerve transection in rats induced the activation of MMP-2 and MT1-MMP in optic nerve astrocytes and resulted in regrowth of axons across the lesion site due to degradation of CSPGs. ⁷⁹ Similarly, transplantation of olfactory ensheating cells was reported to promote neurite outgrowth in vitro in a rat retinal cell culture, via secretion of MMP-2 and degradation of CSPGs. ⁸⁰ Moreover, endogenous MMP-2, produced by retinal astrocytes, seems to be essential to neurite outgrowth in an in vitro mouse retinal explant model. ⁸¹  

Taken together, MMPs can potentially play part in many different aspects of successful axonal regeneration. In the optic nerve, it is already evident that they can add to the conversion of the nerve environment from a repressive to a facilitative substratum for axon growth. However, many other aspects of RGC axon regeneration to which MMPs could theoretically contribute, remain unexplored. Recently, substantial progress was made in initiating robust axonal regeneration of RGCs in rodents via induction of controlled ocular inflammation and phosphatase and tensin homolog (PTEN) deletion. ^82–86 Given these new insights, it might be a good time to re-assess the potential involvement of MMPs in these models of RGC axonal regeneration. 

Can MMPs Modulate Cell Therapy in the Retina?

Besides the prevention of neurite outgrowth, glial scar–derived growth-inhibitory molecules also hamper successful cell transplantation therapies by blocking donor–host integration in the retina. ^87–90 However, increased MMP expression and subsequent decreased deposition of inhibitory ECM molecules in the retina, was reported to create a more permissive environment for regeneration and cell integration. Indeed, elevated expression of MMP-2, -9 and MT1-MMP in the mouse retina, decreases glial barrier formation at the outer limiting membrane and facilitates incorporation of grafted photoreceptor sheets. ⁹⁰ Accordingly, controlled release of MMP-2 directly at the site of injury, enhances progenitor cell integration and retinal repopulation after subretinal transplantation in rodents. ^87,91–94 Unfortunately, focus has been on cell therapies for the outer retina (i.e., for photoreceptor degeneration), ^95–99 and little evidence has been provided thus far that intravitreally transplanted stem/progenitor cells can survive and integrate in the inner retina. 

Of note, transplantation of progenitor/stem cells in the inner retina could be used to replace lost RGCs and ultimately, but rather unlikely, to rewire the entire complex retinotopic circuit. But, alternatively, progenitor/stem cells could also be delivered to provide neuroprotection to the remainder RGCs or to transform the host retina into a state that is more permissive to endogenous attempts of repair. 

Anyway, regardless of the envisioned strategy, the improvement of the integration efficiency of the transplanted cells is one of the many challenges for the successful development of cell therapy in the glaucomatous retina. ⁸⁸ Nonetheless, given the modest successes with human Müller stem cell and rat oligodendrocyte precursor cell transplantations in rat glaucoma models, ^100,101 the recent insights into the nature of the barriers to inner retinal engraftment ^88,89 and the lessons learned from research in the outer retina, it might be worthwhile to explore whether MMPs could be facilitators of host–donor integration in the inner retina. 

Concluding Remarks and Future Directions

Overall, our knowledge on the involvement of MMPs during the onset and disease progression of glaucoma is limited. While MMPs are recognized beneficial players in IOP homeostasis in the anterior segment of the eye, ²¹ we are still in the dark about their nature and the processes they contribute to, in the posterior segment. In the optic nerve, MMP-1 and MT1-MMP are believed to be involved in the onset of glaucoma, and upregulated MMP-3 and -9 expression in the retina hints that MMPs might be important players in glaucoma pathogenesis as well. However, because of the shortage of functional studies, it is currently impossible to define their exact or causative role. 

Nevertheless, the scientific community progresses to understand the complex spatial and temporal regulation of beneficial and detrimental effects of MMPs during and after CNS injury and is starting to explore the potential of MMP-based approaches to facilitate CNS repair. In the retina and optic nerve, in particular the gelatinases are emerging as modulators of the receptiveness to neuroprotection, stimulation of axon regeneration, and cell transplantation therapy. However, more in-depth studies of MMP activity in time and space are imperative to disentangle the complex ‘protease web' of which they are part and to identify the myriad of functions fulfilled by MMPs. ¹⁰² Such studies should consist of system-wide, in vivo approaches, including conditional and/or cell-specific MMP gene deletion, MMP downregulation via RNAi, highly specific inhibitors or blocking antibodies/nanobodies, or alternatively, MMP upregulation via cell-specific viral vector-mediated gene delivery or recombinant MMP, all in combination with high-content proteomic profiling techniques. Indeed, essential to the design of future MMP-based therapies, is the identification of direct MMP substrates and downstream pathways underlying time and space restricted MMP functions. This can be achieved by means of degradomics studies after genetic or pharmacologic MMP inhibition, mice deficient in particular MMP substrates, in vitro cleavage, cleavage site protection/mutagenesis or yeast two-hybrid assays. These approaches should point out MMP drug targets exacerbating pathology as well as antitargets that provide protection or resolution, and will not only allow the development of a novel generation of highly specific MMP inhibitors and MMP inducers/activators, but more likely, they will result in alternative therapies targeting upstream or downstream factors of MMP-modulated networks. 

Acknowledgments

The authors thank the Research Foundation Flanders (FWO-Vlaanderen, Belgium, G.05311.10) and the Katholieke Universiteit (KU) Leuven Research Council (KU Leuven, Belgium, BOF-OT/10/033). Lies De Groef is a research fellow of the Flemish government agency for Innovation by Science and Technology (IWT-Vlaanderen, Belgium), Eline Dekeyster is a research fellow and Ingeborg Stalmans is a senior clinical research fellow of FWO-Vlaanderen. 

Disclosure: L. De Groef, None; I. Van Hove, None; E. Dekeyster, None; I. Stalmans, None; L. Moons, None