|Home | About | Journals | Submit | Contact Us | Français|
Normal homeostatic adjustment of elevated intraocular pressure (IOP) involves remodeling the extracellular matrix (ECM) of the trabecular meshwork (TM). This entails sensing elevated IOP, releasing numerous activated proteinases to degrade existing ECM and concurrent biosynthesis of replacement ECM components. To increase or decrease IOP, the quantity, physical properties and/or organization of new components should be somewhat different from those replaced in order to modify outflow resistance. ECM degradation and replacement biosynthesis in the outflow pathway must be tightly controlled and focused to retain the complex structural organization of the tissue. Recently identified podosome- or invadopodia-like structures (PILS) may aid in the focal degradation of ECM and organization of replacement components.
Increased resistance to aqueous humor flow through the trabecular meshwork (TM) results in elevated intraocular pressure (IOP), which is a primary risk factor for primary open-angle glaucoma (POAG)(Boland and Quigley, 2007). Normal homeostatic adjustment of outflow resistance appears to be triggered by cellular sensing of changes in stretch or distortion of the cribriform or juxtacanalicular region of the TM (Bradley et al., 2001; Acott and Kelley, 2008). When these signals are received, extracellular matrix (ECM) turnover is adjusted to shift the balance toward higher or lower resistance. This homeostatic adjustment appears to begin with selective proteinase secretion and activation, targeted ECM cleavage and fragment uptake, and biosynthesis of replacement ECM components (Acott and Kelley, 2008). Since mRNA levels for many ECM genes in the TM are relatively high (Vittal et al., 2005), ongoing ECM turnover must be a regular maintenance event, even when the resistance does not need to be modified (Bradley et al., 1998).
Various proteinases and their inhibitors are expressed in the TM in a variety of instances. These include matrix metalloproteinases (MMPs), membrane type (MT) MMPs, ADAMs (A Disintegrin-like And Metalloproteinase), ADAMTSs (ADAM with ThromboSpondin type 1 motifs), plasminogen activators (PA), tissue inhibitors of metalloproteinases (TIMPs), PA inhibitors, and serine proteinase inhibitors (serpins) (Table 1 and Table 2)(Alexander et al., 1991; Borras, 2003; Fuchshofer et al., 2003; Lo et al., 2003; Vittal et al., 2005; Oh et al., 2006). The significance of MMPs in outflow resistance was shown when treatment of anterior segments in perfusion culture with MMPs was found to increase outflow, while specific inhibition of MMP activity decreased outflow facility (Bradley et al., 1998). Also, when pressure was increased in perfusion culture there was a concomitant increase in outflow facility, which stabilized over time. This homeostatic pressure response likely involves MMPs, since adjustment of outflow facility was concurrent with increased MMP activity and ECM turnover. The proteolytic degradation of ECM components probably affects structural organization and alters biological interactions, leading to a modified outflow resistance. However, despite many years of research, the ECM components that contribute to outflow resistance remain elusive (Acott and Kelley, 2008). Glycosaminoglycans (GAGs) are thought to be a major contributing factor since treatment with enzymes that specifically degrade one or more type of GAG, or with agents that modify GAG chain biosynthesis, increase outflow facility (Barany and Scotchbrook, 1954; Francois, 1975; Knepper et al., 1984; Sawaguchi et al., 1992; Sawaguchi et al., 1993; Hubbard et al., 1997; Johnson and Bahler, 1999; Keller et al., 2008). However, GAG-specific enzymatic changes in outflow of humans or higher primates remain controversial (Acott and Kelley, 2008).
In addition to their degradation activities, many of these proteinases participate in cell-cell and cell-matrix interactions, process latent proteins to active forms, convert structural ECM proteins to signaling molecules, change tissue architecture, and provide cues for cell proliferation and survival (Page-McCaw et al., 2007). Thus, degradation of structural components alone may not directly result in IOP adjustment, but that proteolytic alterations could cause secondary effects that modify outflow resistance. In order to facilitate these adjustments, replacement ECM is likely to differ from what was degraded. Biosynthesis of replacement molecules is therefore a crucial component for normal homeostatic adjustment of outflow resistance.
MMPs are a group of 23 related proteinases that degrade ECM (Page-McCaw et al., 2007). All MMPs have a similar domain structure and in their simplest form are composed of a prodomain and a catalytic domain (Fig. 1)(Chakraborti et al., 2003; Ra and Parks, 2007). MMPs may also contain C-terminal hemopexin-like (PEX) domains, which function in substrate recognition or proteinase localization, that are separated from the N-terminal catalytic domain by a flexible hinge region. Some MMPs also have a transmembrane or a glycosylphosphatidylinositol anchor, which anchors them to the cell surface. MMPs are synthesized as zymogens, inactive proenzymes, where the prodomain confers latency to the active enzymes. A free thiol of a conserved cysteine residue in the prodomain electrostatically interacts with a Zn2+ ion that is histidine-ligated into the active site of the catalytic domain in the sequence HEXXHXXGXXH (Fig. 1). Extracellular posttranslational cleavage, usually by another MMP or by a furin proprotein convertase, disrupts the thiol-Zn2+ interaction and releases the biologically active proteinase, a process known as the cysteine switch (Chakraborti et al., 2003). Serine proteinases, such as plasmin and thrombin, can activate proMMPs, while interaction with αV integrin has been implicated for MMP2 (Brooks et al., 1996; Ra and Parks, 2007). However, for most MMPs, the actual physiologic activation mechanisms are uncertain.
Numerous MMPs have been detected in TM cells and tissue (Table 1) and MMP mRNA and protein levels were upregulated in response to mechanical stretching, elevated pressure and various cytokines or growth factors (Alexander et al., 1991; Samples et al., 1993; Alexander et al., 1998; Bradley et al., 1998; Bradley et al., 2000; Gonzalez et al., 2000; WuDunn, 2001; Fleenor et al., 2003; Pang et al., 2003; Vittitow and Borras, 2004; Oh et al., 2006; Kelley et al., 2007). MMPs -2, -14, -15 and -16 are constitutively expressed at relatively high levels in the TM (Alexander et al., 1991; Bradley et al., 2001; Vittal et al., 2005). MMPs -1, -3, and -9 are normally expressed at low levels but are massively upregulated and activated in response to various stimulations. Agents that increase outflow facility in perfusion culture, or laser trabeculoplasty surgery and medications that lower IOP in POAG patients, also affect expression and/or activation of certain MMPs (Parshley et al., 1996; Johnson, 1997; Peterson et al., 1999; Shearer and Crosson, 2002; Fleenor et al., 2003; Pang et al., 2003; Oh et al., 2006; Sanka et al., 2007; Bahler et al., 2008). Addition of purified MMPs to human anterior segment perfusion cultures increased outflow facility, while inhibiting their endogenous activity by TIMP2 or synthetic MMP inhibitors reduced outflow facility (Bradley et al., 1998). This argues strongly that controlled ECM turnover is required for maintenance of outflow resistance and that MMP regulation seems to be a major component for adjusting outflow resistance. Immunofluorescence studies showed that MMP2 strongly stained normal human TM in the uveal, corneoscleral and juxtacanalicular regions and the inner wall of Schlemm’s canal (Ronkko et al., 2007). In POAG TM, immunostaining intensity for MMPs -1 and -3 was greatly increased, while by Western blot, MMPs -2 and -9 were reduced compared to age-matched control eyes (Ronkko et al., 2007; Govindarajan et al., 2008). The increase in MMPs -1 and -3 levels suggests that POAG TM cells continually strive to correct outflow resistance, but instead the persistent action of these MMPs likely causes an overproduction of replacement ECM and/or cause cleavage of ECM components that are not usually accessible to degradation. TGFβ2, a cytokine whose levels are increased in POAG aqueous humor and decreases outflow facility in perfusion culture, increased expression, but not activation, of pro-MMP2 (Fuchshofer et al., 2003; Gottanka et al., 2004; Fleenor et al., 2006).
Compartmentalization of MMPs has been noted in many cell types and tissues, but especially in invadopodia and podosomes of invasive cells (Chen and Wang, 1999). The term compartmentalization refers to the location of the inactive enzyme in the pericellular matrix, and its site of activation (Ra and Parks, 2007). Due to the potency of these enzymes in degrading ECM, MMPs are targeted to specific locations where they become highly concentrated. This regulates activation of latent MMPs and targets their catalytic activity to certain substrates in the pericellular environment. For instance, MMP14 (MT1-MMP) cleaves pro-MMP2 and pro-MMP13 to their active forms on the cell surface, which triggers potent pericellular proteolytic activity (Itoh and Seiki, 2006). Recent identification of clusters of podosome- or invadopodia-like structures (PILS) in TM cells show that MMPs -2 and -14 are localized to regions of the cell that appear to function in both adhesion and matrix degradation (Fig. 2)(Linder, 2007; Aga et al., 2008). Focusing MMP14 at such structures could increase its local concentration to allow homodimerization of its PEX domains. This in turn would facilitate MMP2 activation and further augment targeted proteolytic activity at the cell surface. PILS have been found in TM tissue suggesting that these focal regions of regulated and targeted ECM turnover may be important in regulating outflow facility in response to elevated IOP (Aga et al., 2008).
Active MMPs degrade many different ECM molecules (Table 1)(Chakraborti et al., 2003), most of which are also expressed in the TM (Acott and Kelley, 2008). Rather than degrading all ECM, it is likely that only certain molecules or parts of molecules are degraded. This focal, targeted degradation is probably facilitated by PILS. For instance, collagen IV α3 is cleaved by MMP9 and collagen XVIII by other MMPs to release tumstatin or endostatin, respectively, but the structural triple-helical portion is not degraded (Page-McCaw et al., 2007). Both these collagen types are expressed in the TM (Hann et al., 2001; Ohlmann et al., 2005) and both statins bind integrins, which launch a signaling cascade that disrupts the actin cytoskeleton (Bix and Iozzo, 2005). Therefore, in the TM, proteolytically-released fragments could impact outflow resistance via cytoskeletal adjustments.
Another class of zinc-dependent metalloproteinases is the adamalysins, which are categorized into two subfamilies known as the ADAMs and the ADAMTSs (Table 1). Both these proteinase groups are involved in ectodomain shedding, activation of cell surface receptors and growth factors and ECM protein cleavage (Apte, 2004; Huovila et al., 2005; Porter et al., 2005). The major difference between the two subfamilies is that ADAMs are mostly transmembrane proteins while ADAMTSs are secreted molecules that bind to the ECM.
There are currently 29 ADAM family members that cleave a wide range of substrates (Huovila et al., 2005). Similar to MMPs, ADAMs have a multi-domain structure and are synthesized as proenzymes that are activated by a cysteine switch (Fig. 1). A terminal aspartate residue in the sequence HEXXHXXGXXHD distinguishes the adamalysins from the MMPs. Lying C-terminal to the catalytic domain is the disintegrin domain, which contains an integrin-binding motif, a cysteine-rich region, an EGF-like domain, a transmembrane region and a C-terminal cytosolic tail. The tail contains multiple motifs which interact with other cytosolic proteins and can be phosphorylated on Ser/Thr and Tyr residues (Huovila et al., 2005).
There are 19 ADAMTSs and 3 ADAMTS-like enzymes. Like the ADAMs, ADAMTSs have a wide range of substrates including collagen, versican, and aggrecan (Porter et al., 2005). The domain structure of the ADAMTSs is similar to that of ADAMs and MMPs, but there are some differences that distinguish them (Fig. 1). There is some debate as to whether the prodomains confer latency to the ADAMTS enzyme, but the prodomains are removed by furin-cleavage in the secretory pathway or by MMP17 on the cell surface (Apte, 2004). The ancillary region is comprised of a disintegrin-like domain, a central thrombospondin-like (TS) domain, a cysteine-rich domain, a spacer domain, and a varying number (0–14) of C-terminal TS domains. The ancillary domain can be proteolytically cleaved, which either augments or inhibits activity of the ADAMTS enzyme (Gao et al., 2002).
Although ADAMs and ADAMTS are expressed in other eye tissues (Wride et al., 2006), little is known about the action of these proteinases in the TM. Our studies have shown that TNFα and/or IL-1α treatment of TM cells stimulate mRNA expression of ADAMs -9 and -10 and ADAMTSs -4, -5, -7 and -13 (Chen et al., 2005; Acott and Kelley, 2008). TGFβ2 treatment of glaucomatous TM cells increased ADAMTS5 mRNA expression (Fleenor et al., 2006). A neo-epitope of versican that is exposed following ADAMTS4 cleavage was found to localize to PILS (Fig. 2)(Aga et al., 2008) and in areas of the TM that experience high segmental flow, as assessed by fluorescent Qdot nanoparticle labeling of anterior segments (Bradley et al., 2008). Immunolocalization of ADAMTS4 showed that its expression was highly increased in the JCT region in human anterior segments that were subjected to elevated pressure (Keller et al., unpublished observations). Moreover, treatment of human and pig anterior segments in perfusion culture with ADAMTS4 increased outflow facility. Further studies are required to clarify the possible roles of ADAMs and ADAMTSs in outflow resistance.
In order to regulate these potent proteinases, MMP activity needs to be tightly controlled. TM cells can downregulate MMP mRNA or protein levels, change the activation state of these enzymes or release inhibitors that inactivate the enzymes. There are four TIMPs all of which are expressed in human TM cells (Table 2). Factors known to upregulate MMP expression also affect TIMP mRNA expression and therefore a delicate balance likely exists in vivo between MMP activation and inhibition by TIMPs (Samples et al., 1993; Pang et al., 2003; Oh et al., 2006). TIMP2 not only acts as an inhibitor of MMP activity, but it is also involved in activation of pro-MMP2 (Wang et al., 2000). The C-terminal domain of TIMP2 simultaneously binds to the C-terminus of pro-MMP2 and the N-terminus of activated MMP14 (Fig. 3). This trimeric complex recruits an adjacent MMP14 molecule, which is responsible for the resultant cleavage of pro-MMP2. The relative concentration and binding of TIMP2 influences whether MMP14 will participate in cleavage of MMP2 and whether it will cleave other substrates (Kudo et al., 2007). In perfused anterior segment organ culture, increased pressure resulted in more activated MMP2 (Bradley et al., 2001). In cell culture, mechanical stretch, which is presumable mimicking pressure increases, increased MMPs -2 and 14 while decreasing TIMP2 levels (Bradley et al., 2001). This reduction in TIMP2 appears to bring it to the correct level to facilitate activation of MMP2 without completely blocking the process. TIMP1 inhibits ADAM10 but not ADAM17, whereas TIMP3 inhibits both enzymes (Rapti et al., 2008). TIMP3 inhibits ADAMTSs -4 and -5 activity (Kashiwagi et al., 2001).
Many other proteinases are expressed in the TM and may function directly or indirectly to modulate outflow resistance (Table 1). Tissue plasminogen activator (tPA), a serine proteinase that converts plasminogen to plasmin, is expressed in the TM (Park et al., 1987; Shuman et al., 1988; Chen et al., 2005) and by Schlemm’s canal endothelial cells (Stamer et al., 1998). Since plasmin can activate proMMPs, the action of tPA may be a secondary effect rather than direct ECM proteolysis. tPA activity was reduced in response to dexamethasone, a corticosteroid that can induce glaucoma and decrease outflow facility in human anterior perfusion culture (Snyder et al., 1993; Clark et al., 1995). Urokinase-type PA (uPA) distribution overlaps with that of tPA and may also be involved in ECM turnover (Tripathi et al., 1990; Chen et al., 2005; Vittal et al., 2005).
Serpins are a class of serine proteinase inhibitors (Table 2) (Gettins, 2002). There are 16 members of this family including antithrombin, α1-proteinase inhibitor (PAI-1), and antitrypsin and they function in a wide range of physiological processes. Serpin expression is modulated by various stimulations in TM cells (Zhao et al., 2004; Chen et al., 2005; Rozsa et al., 2006; Fan et al., 2008). SerpinA3 (α1-chymotrypsin) was upregulated, while serpinB2 (PAI-2) was downregulated by corticosteroid treatment of human TM cells (Lo et al., 2003; Rozsa et al., 2006; Fan et al., 2008). SerpinE1 (PAI-1) was upregulated 6.8-fold by TGFβ2 treatment of glaucomatous TM cells and by TGFβ1 in HTM cells (Zhao et al., 2004; Fleenor et al., 2006). Furthermore, TGFβ2-induced expression and secretion of PAI-1 reduced MMP2 activity (Fuchshofer et al., 2003). Thus, modulation of serpin expression in response to various factors may affect MMP activity. Aqueous humor contains significant amounts of serpinC1 (antithrombin III), serpinA1 (α-1-antitrypsin), α-2-macroglobulin, and tPA (Tripathi et al., 1988; Ando et al., 1993; Rao et al., 2000). Interestingly, α-2-macroglobulin is an endogenous inhibitor of ADAMTSs -4 and-5 (Tortorella et al., 2004). Increased concentrations of these inhibitors in aqueous humor may alter MMP activity in active flow areas.
MMPs and other proteinases in the TM may bind GAGs, which could limit diffusion of the active enzyme in tissue (Page-McCaw et al., 2007). GAGs are long sugar chains that are usually attached to a core protein. Since GAGs are thought to be a significant source of outflow resistance, GAG binding may directly affect expression and/or activation of MMPs. The types of GAGs are chondroitin/dermatan sulfate (CS/DS), heparin/heparan sulfate (HS), keratan sulfate (KS) and hyaluronan (HA). Chondroitin-4-sulfate (C-4-S) concomitantly binds MMP2 and MMP16 to facilitate the conversion of latent pro-MMP2 to its active form (Iida et al., 2007). It is feasible that MMP2 located in PILS could, in part, be activated by a similar mechanism. MMP2 can also bind tightly to HS chains of syndecan-2, which suppresses its activation (Munesue et al., 2007). Hyaluronan, applied exogenously to corneal explants in organ culture, resulted in increased expression and activation of MMPs -2 and -9 (Isnard et al., 2001), while treatment of cartilage chondrocytes with low molecular weight HA fragments increased MMP3 mRNA expression (Ohno et al., 2005). Together, these studies suggest that GAG chains, or GAG chain fragments, may facilitate expression and activation of pro-MMPs and therefore play a role in targeted degradation of pericellular matrix. TM GAG content varies with age and POAG disease progression (Gong et al., 1992; Knepper et al., 1996; Knepper et al., 1996), so minor changes may profoundly affect proteinase activity and/or inhibition and potentially disrupt normal ECM turnover during homeostatic adjustment of IOP.
CD44, the major hyaluronan receptor, also plays an important role in MMP regulation. CD44 can attenuate the secretion and activation of MMP2 (Takahashi et al., 1999). CD44 is also a receptor for active MMP9 in keratinocytes and this interaction subsequently activates latent TGFβ (Yu and Stamenkovic, 1999). MMP14 can bind directly to CD44 via its PEX domains, which results in shedding of a soluble 70kDa extracellular fragment (sCD44) (Kajita et al., 2001; Mori et al., 2002). sCD44 concentrations were higher in aqueous humor of POAG patients than in age-matched control samples (Knepper et al., 2002; Nolan et al., 2007). Thus, CD44 not only regulates MMP activity, but is itself a substrate for MMP cleavage.
Other proteinases and inhibitors also bind GAGs and ECM molecules, e.g. ADAM12 binds to syndecan-4 and ADAM-9 binds α6α1 and αvα5 integrins (Huovila et al., 2005). Integrin-ADAM association facilitates cell-cell interactions and sequesters the enzyme in an inactive state at the cell surface (Bridges and Bowditch, 2005). There is no evidence to suggest that the disintegrin domains of ADAMTSs also bind integrins (Porter et al., 2005). ADAMTSs binds sulfated GAGs in the ECM or on the cell surface, or to HS and CS chains of syndecan-1 (Apte, 2004; Gao et al., 2004). The spacer region of ADAMTS4 binds the C-terminal region of fibronectin, which may inhibit its enzymatic activity (Fig. 1)(Hashimoto et al., 2004). MMP inhibitors also bind GAGs. The N-terminal region of TIMP3 interacts with HS and possibly CS GAG chains (Yu et al., 2000). Serpin activity is also profoundly affected by the presence of GAGs, such as HS and DS, which accelerate interaction with their target proteinases (Pike et al., 2005).
An integral component of any normal ongoing or homeostatic resistance adjustment involving ECM turnover is the biosynthetic replacement of degraded components (Acott and Kelley, 2008). Modifications in outflow resistance will likely include changes in levels or organization of ECM components. TM cells alter mRNA expression of many ECM genes in response to mechanical stretching, exposure to exogenous soluble factors and synthetic agents or by treatment with IOP-lowering drugs and laser trabeculoplasty (Lo et al., 2003; Zhao et al., 2004; Vittal et al., 2005; Fleenor et al., 2006; Rozsa et al., 2006; Fan et al., 2008). ECM gene mRNA levels also vary between non-glaucomatous and glaucomatous eyes (Diskin et al., 2006; Liton et al., 2006). PILS could potentially coordinate the biosynthetic replacement and organization of ECM components in degraded areas. Some of the ECM molecules that are synthesized are different from their predecessors and can have extra exons included, or excluded, which may contain additional ECM or cell binding sites (Vittal et al., 2005; Zhao and Russell, 2005; Keller et al., 2007). Continued exposure of TM cells to elevated IOP, or from many years of normal homeostatic adjustment, could conceivably cause an excess of cell-cell, cell-protein and/or protein-protein interactions that would contribute to the observed increased rigidity of the TM in POAG or with age (Gabelt and Kaufman, 2005; Schlunck et al., 2008). In addition, stiffness could be exacerbated by an increase in ECM cross-linking. For instance, POAG TM cells have increased protein levels and activity of the enzyme tissue transglutaminase (Tovar-Vidales et al., 2008). Increased stiffness may hinder detection of mechanical stretch/distortion signals by cells in the JCT region. Consequently, the normal signals that are required to initiate ECM turnover may not be received and the typical mechanisms to adjust outflow resistance would not be activated. Alternatively, such ECM cross-linking could change susceptibility of ECM substrates to proteinase cleavage and turnover.
Support was provided by NIH EY003279, EY008247, EY010572, the Glaucoma Research Foundation (to KEK), and by an unrestricted grant to Casey Eye Institute from Research to Prevent Blindness, New York, NY.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.