Degradation and remodeling of the extracellular matrix (ECM) is a common process for tissue development, growth, and repair. Matrix metalloproteinases (MMPs) are key components involved in degrading the ECM of many tissues in the body, including the oral gingival tissue, which supports the teeth. Research has implicated bacteria, the human immune system, and lipopolysaccharides in the MMP activation process, which leads to breakdown of the ECM in oral gum tissue, leading to gingivitis and periodontitis. MMPs are produced by the surrounding tissue cells and by macrophages and lymphocytes. These enzymes target specific ECM proteins and break them down through peptide bond cleavage. There are several types of these enzymes, and they all have a common basic structure. Regulation of MMPs and their gene expression is accomplished by several different means. MMPs also play a critical role in several inflammatory diseases, including periodontal disease.
The first type of MMP is the collagenase group, which consists of a fibroblast-type that is secreted by interstitial cells and a polymorphonuclear type that is secreted by leukocytes (Overall, 1991; Birkedal-Hansen et al, 1993). The interstitial collagenase breaks down collagen of types I and III, which are found in human dentin (Agematsu et al, 1997), II and X, found in cartilage (Dourado and LuValle, 1998), VII, which is present in the dermal tissue layer of the skin (Jarvikallio et al, 1997), and VIII, found in skin ECM (Shuttleworth, 1997). The leukocyte collagenase degrades collagen of types I, II, and III. Both also break down gelatin.
The second MMP type is the gelatinases, which break down gelatin, collagen, elastin, and fibronectin (Overall, 1991). There are two types called 72-kD and 92-kD gelatinases, which are named so because of their different molecular weights (Birkedal-Hansen, 1993).
The third type of MMP is the stromelysin type. Stromelysins are named so because stromal cells, which produce the reticular connective tissue stroma that supports lymphocytes in the lymph nodes, spleen, and bone marrow, express them (Marieb, 1998). Stromelysins degrade this connective tissue. There are three types of differing molecular weights. They degrade all of the above-mentioned ECM components plus laminin (, a cell adhesion protein that makes up ground substance in the ECM (Overall, 1991). They act like glue to attach cells to the ECM components (Marieb, 1998). Putative metalloproteinase-1 (PUMP-1) is a very short stromelysin-like MMP, and it also breaks down transferrin along with the other ECM components listed above. However, it does not break down laminin (Overall, 1991).
The MMPs are structurally related in that they have conserved homologous regions. There are five general domains that are present in the MMPs. The first is a signal peptide that has 17-29 residues on the amino terminus region (Birkedal-Hansen, 1993). This is a hydrophobic peptide. When this peptide is removed, it signals activation of the enzyme. The propeptide has 77-87 residues, and it serves as the amino terminus of the enzyme after the signal peptide is removed (Birkedal-Hansen, 1993). The region that follows the propeptide is the catalytic domain, which consists of about 110 residues (Birkedal-Hansen, 1993). Here, there are three His residues that serve as ligands for Zn2+, which is necessary for enzyme activity. Following the catalytic domain is a hinge region that is rich in Pro, and it marks the transition to the fifth domain. The fifth domain is called the hemopexin-like domain, which consists of about 200 residues (Birkedal-Hansen, 1993). This domain contributes to substrate specificity. This name was given because the C-terminus shows homology with the blood serum protein hemopexin, which binds to heme (Overall, 1991).
The gelatinases have an extra domain that contributes to their high affinity for gelatin. The domain is located next to the catalytic domain, and it is a 58-amino acid sequence that is repeated three times (Overall, 1991). This sequence is identical to the ECM fibronectin type II motif, which gives the gelatinases their high affinity for binding to and degrading gelatin-denatured collagen a chains.
The three His residues in the catalytic domain serve as the ligands for the Zn2+, but a fourth ligand is required. In the inactive form of the MMP, an unpaired Cys residue serves as the ligand for Zn2+, whereas water binds toZn2+in the active form. There is also a calcium-binding site in the catalytic domain between the hinge region and zinc-binding site that has many Asp and Glu residues (Birkedal-Hansen, 1993).
The mechanism by which MMPs degrade the ECM is not fully understood. However, researchers do know that these enzymes cleave the macromolecules of the ECM into smaller fragments. Indeed, almost every MMP cleaves gelatin and fibronectin, and even some of the collagen types. However, the two collagenases are the only ones that are able to cleave interstitial collagens (Overall, 1991). Cleavage of the ECM macromolecules occurs largely in peptide bonds between hydrophobic residues (Birkedal-Hansen, 1993). The different MMPs discriminate between the various bonds to break because of their selective substrate specificity. It should be clear though that the substrate specificities of the different types of MMPs as a whole are broad and overlapping. So, the different MMPs have different ranges of substrates that they can cleave.
Collagenases have been the best studied of the MMPs. They have high substrate specificity for collagen. Cleavage of collagen occurs along the Gly775-Leu776 and Gly775-Ile776 peptide bond within the collagen alpha helix (Birkedal-Hansen, 1993). This is a unique site within the collagen a helix because it has a tendency to unwrap the triple helical structure. So, collagenase is better able to find and cleave this region.
The two gelatinases cleave several peptide bonds, most of which contain a Gly residue. For example, they can cleave Gly-Val and Gly-Asn peptide bonds. The stromelysins cleave the ECM proteins at a variety of peptide bond sites (Birkedal-Hansen, 1993).
The MMPs are secreted from cells in the inactive form, and must be activated to degrade and remodel the ECM. In vitro, several chemicals can directly activate MMPs (Overall, 1991). Chaotropic salts like potassium bromide (KI) break up hydrophobic interactions and induce conformational changes in the enzyme to activate it (Nina, 2000). Organomercurials like p-aminophenylmercuric acetate, detergents, and trypsin can interact with the Cys-Zn2+ interaction in the catalytic domain to activate MMPs. The proteolytic enzymes like trypsin and plasmin remove part of the propeptide to form the open conformation of the enzyme, which is unstable (Overall, 1991). Then the enzyme undergoes autolytic cleavage in several places to make itself fully active and more stable than the open form. When the Cys residue is dissociated from the Zn2+ with the chemicals mentioned above, the solvent could then interact with the Zn2+ as the fourth ligand. In effect, different chemicals can activate MMPs by catalyzing the removal of the propeptide, but at different peptide bonds.
There has also been evidence that one MMP can super activate another MMP. The interstitial collagenase can be activated by both stromelysins I and II and the PUMP-I (Birkedal-Hansen). For example, stromelysin appears to catalyze the removal of the propeptide by breaking a different peptide bond than is broken by proteolytic cleavage with trypsin. This peptide bond is one residue towards the amino terminal end from the normally cleaved bond, and the active enzyme has a new amino terminus with a Phe residue at the end. This creates an interstitial collagenase with five to eight times more catalytic activity than the collagenase with the normal activation mechanism.
MMPs can also be inhibited from breaking down the ECM. Many synthetic inhibitors have been shown to act on MMPs. Generally, chelating agents like 1,10-phenanthroline and EDTA can remove the Zn2+ from the active domain, but exhibit no selectivity between different MMPs (Birkedal-Hansen, 1993). So, these two agents can be used to inhibit all MMPs.
There have been many studies that examine the substrate specificity of MMPs, and trial-and-error has been used to determine what inhibitors are potent for specific MMPs. For example, researchers have found that sulfur-based inhibitors are highly effective against the two collagenases (Birkedal-Hansen, 1993).
Monoclonal antibodies have also been discovered to have potent inhibitory properties. Monoclonal antibodies are highly specific for their protein substrates. In fact, they can even differentiate between the two similar collagenases (Birkedal-Hansen, 1993). Gelatinase inhibition isn't as effective (about 40-60% inhibition) with these antibodies as collagenase inhibition (about 90% inhibition). Monoclonal antibodies, in effect, prevent degradation of the ECM by inhibiting the action of MMPs.
Alpha macroglobulins are yet another set of inhibitors of the MMPs. They are hormone-transporting glycoproteins, with a molecular weight range of 620-680 kD (worldmedicus website, 2001). The mechanism of inhibition is complex but unique in that the MMP activates the inhibitor to inhibit the MMP (Birkedal-Hansen, 1993). First, the MMP cleaves a 40-amino acid bait region on the a macroglobulin. This causes a conformational change on the inhibitor, which leads to hydrolysis of a thioester bond. Upon hydrolysis, a reactive Glu residue is produced that reacts with and binds to a Lys residue on the MMP. The binding of the inhibitor to the MMP causes steric hindrance of the large MMP, resulting in no catalytic activity. When the a macroglobulins are present with MMPs, they immediately capture and inhibit the MMPs, especially collagenase. They are better substrates for collagenases than even type I collagen. So, they are very strong inhibitors.
The most well studied inhibitors are the tissue inhibitors of MMPs (TIMPs). A variety of tissue cells synthesize TIMPs. They are effective in both inhibiting the active MMPs and blocking the activation of the latent or inactive MMPs (Overall, 1991). There are two types that have been identified: TIMP-1 and TIMP-2.
TIMP-1 is a glycoprotein that has a 23-residue signal region and a 184-residue inhibitory region (Birkedal-Hansen, 1993). It has six disulfide bonds that lend the protein its high stability. TIMP-1 is effective in inhibiting the interstitial collagenase. TIMP-2 does not have any sugars attached and is thus unglycosylated. It has a 26-residue signal region and a 194-residue inhibitory region (Birkedal-Hansen, 1993). TIMP-2 targets the two gelatinases in both their latent and active forms. When these inhibitors form a 1:1 complex with specific MMPs, the MMPs are not cleaved, unlike like a macroglobulins. After dissociation from the MMPs, they are also fully functional. Site-directed mutagenesis has shown that the inhibitory function of the TIMP-1 is most likely within the first 134 residues of the protein (Birkedal-Hansen, 1993).
Studies have shown that the TIMPs irreversibly bind noncovalently to the activated MMPs at the active site. On the other hand, the inhibitors bind noncovalently with inactive MMPs outside of the MMP active site domain (Overall, 1991). This is evident by the fact that gelatinases are activated by organomercurials while still bound to the TIMP-2. The organomercurials activate by interacting at the active site of the MMP (Overall, 1991).
Recent research has revealed the TIMP-2 binding site on the 72kD human gelatinase (Overall et al, 1999). Using site-directed mutagenesis, it was determined that the TIMP-2 binds to the hemopexin-like region (the fifth domain) on the carboxyl end of gelatinase A. Another TIMP-2 can also bind to the active site to inhibit the enzyme. So, it appears that two TIMPs must bind to gelatinase in order for complete inhibition to occur.
To add to the complexity of activation and repression of MMPs, researchers have recently tested the effects of addition of IL-1 and retinoic acid in combination to proteoglycan fragments of bovine nasal cartilage (Shingleton et al, 2000). They found that adding retinoic acid stimulated production of TIMP in cartilage tissue culture by day seven. IL-1 stimulated secretion of MMP-13, a collagenase, which accumulated to exceed the TIMP concentration, leading to proteoglycan breakdown by day fourteen. Thus, retinoic acid and IL-1 work together to promote degradation of connective tissue.
Regulation of MMP expression is complex and can occur through several different mechanisms. Growth factors and cytokines are strong activators of MMP expression (Birkedal-Hansen, 1993). Interleukin-1 (IL-1), platelet-derived growth factor (PDGF), epidermal growth factor (EGF), and tumor necrosis factor a (TNFa) are several growth factors that stimulate synthesis of collagenases and stromelysins (Borghaei et al, 1999). Each growth factor binds to specific areas of the DNA to stimulate transcription of MMP mRNA. It is complex in that different cells types respond differently to the same growth factor by activating expression of different MMPs. Some of the growth factors also overlap in that they stimulate expression of not only several MMPs, but also TIMPs. Viruses, antibiotics like mitomycin, and ultraviolet light can also stimulate expression. However, dexamethasone, retinoic acid, and gamma interferon can repress MMP gene expression (Overall, 1991).
Studies have linked second messengers to help signal synthesis of MMPs. It has been shown that protein kinase C (PKC) is involved in signaling gene expression, along with other protein kinases (Overall, 1991). Cyclic adenosine monophosphate (cAMP) also plays a role in initiating and repressing MMP expression (Birkedal-Hansen, 1993). For example, in rat osteosarcoma cells, cAMP stimulates MMP expression, and in human fibroblasts, cAMP represses MMP expression, indicating possible tissue specificity. Physical stimuli can also trigger events leading to MMP synthesis. Phagocytosis, heat shock, and treatment with agents that disrupt the cytoskeleton like cytochalasin B can initiate expression (Birkedal-Hansen, 1993). The rearrangement of actin polymers is associated with cell shape change, and can trigger expression as well. Cellular substrate-receptor interaction may also trigger gene expression, like the interaction between a monoclonal antibody and its integrin receptor.
Gingival collagenases are involved in the onset and progression of periodontal disease, which is an inflammation and degradation of the tooth supporting gingival tissue (Ashley et al, 1999). The normal microflora of the mouth can become highly populated as a result of poor dental hygiene. The bacteria thrive on the carbohydrate-based food morsels and they produce acids that dissolve the calcium salts of the teeth. Lipolysaccharides are also secreted and are the culprit in activating transcription of the collagenase genes (Ashley et al, 1999). The bacteria aren't the only players to wreak havoc on the gums. The body's own immune system may also play a role. The immune system perpetually fights off these oral bacteria. So, there can accumulate a large number of mediators like IL-1a,ß and TNF-a that are also able to activate collagenase gene expression (Ashley et al, 1999). So, the pathological response of the body to the bacterial oral infection is overproduction of collagenases that break down the gingival collagen supporting the teeth.
The ratio of active to latent MMPs increases in gingival fluid, and as a result, collagenases break down the gingival collagen (Ashley et al, 1999). This causes recession of the gum line and detachment of the supportive gingival tissue from the root of the teeth. Pockets form and more bacteria can seep deep into the crevices between the gums and teeth. This can activate more collagenases to break down the ECM, which leads to more detachment and decay.
This ongoing cycle can be stopped with the aid of antibiotics that control the oral microflora. A subantimicrobial dose doxycycline can be applied to the affected region of the mouth, which helps to control the bacterial population and reverse the ratio of latent to active MMPs by inhibiting the active form of collagenase (Southard and Godowski, 1998). The symptoms of periodontitis have been shown to reverse using this product called Periostat: the clinical parameters of the disease improve, including the relative attachment levels of the gingiva to the tooth root.
By controlling microbial population and the production of MMPs in the gingival crevicular fluid, other homeostatic imbalances of the body may be avoided, like coronary heart disease, premature births, and atherosclerosis, which have been linked to periodontitis. A study showed that bacteria found in dental plaque caused clotting in arteries of rabbits (Atrix Labs Website, 1996). So, the same bacteria that cause periodontal disease may cause the deposits found in patients with atherosclerosis. Researchers believe that periodontal infections impair growth of the fetus because release of cytokines is enhanced (Marwick, 2000). In another study, researchers found a significant relation between periodontal disease and C-reactive protein and fibrinogen levels, which are indicative of risk for cardiovascular disease (Wu et al, 2000).
Much of the ongoing research of matrix metalloproteinases is directed at determining the exact substrate specificity of the enzymes, fully understanding the known inhibitors, and finding new ones. Degradation and remodeling of the ECM is a natural process that occurs in many tissue types. However, when this process gets out of hand, like in periodontal disease, it is of great benefit to be able to control or influence this process to improve the health of the patient experiencing a particular imbalance of MMP activity.
Copyright © 2001 Joe Kolody and Koni Stone
Back to the Stanislaus Journal of Biochemistry 2001 Table of Contents