Proteolytic Enzymes in Inflammatory Bowel Disease

Along its length, the intestine is lined by a soft and delicate mucosal layer, ∼1 mm thick, but 400 m2 in area. The mucosa is a highly vascular, loose connective-tissue matrix containing structural collagens, glycoproteins such as laminin and fibronectin, chondroitin sulphate proteoglycans such as versican, hyaluronic acid, and ground substance. These are secreted by lamina propria stromal cells (also known as myofibroblasts or mesenchymal cells), and where the stromal cells abut the epithelium and endothelium, there is synergy between the two cell types to construct the basement membrane, containing laminin, heparan sulphate proteoglycans, entactin, and collagen IV. Of all the tissues in the body, the intestinal lamina propria is the most plastic and the one in which extracellular matrix degradation and repair must be tightly controlled. For example, whereas matrix degradation in the rheumatoid joint causes pain, deformity, and disability, matrix degradation in the gut leads to ulceration, diarrhea, malabsorption, and protein-losing enteropathy. Inflammatory cells such as neutrophils, macrophages, eosinophils, and mast cells are increased in numbers in the mucosa in inflammatory bowel disease (IBD). Although there is a great deal of interest in the cytokines and toxic molecules produced by these cells, they are also a rich source of proteolytic enzymes, such as the serine protease neutrophil elastase, and various acidic cysteine proteases. In addition, endogenous fibroblast-like stromal cells are a potent source of matrix metalloproteinases (MMPs). Although in health, stromal cells secrete matrix and low amounts MMPs and tissue inhibitors of matrix metalloproteinases (TIMPs) as part of normal tissue turnover, in inflammation, when activated by cytokines, they secrete large amounts of MMPs, which can degrade the lamina propria and basement membrane. It is likely that there is subtle tissue-specific control of MMP and TIMP activity in the gut that is important in maintaining normal tissue integrity, causing tissue injury during inflammation, and helping in the repair processes. From the outset we emphasise that whereas proteolytic degradation of lamina propria matrix by the secreted products of extravasating blood-borne cells is undoubtedly important in tissue injury in the gut, we believe that proteolytic degradation by cytokine-activated stromal cells is equally if not more important in chronic gut inflammation. It should also be stated that in contrast to inflammatory conditions of the joint and lung, where proteolytic degradation has long been recognised as an important factor in tissue injury, studies in the gut are only beginning, and the literature is rather small. Therefore in this review, we first briefly give an overview of proteases, even though the role of most of them in the gut is not established, before reviewing the evidence that proteolysis is important in IBD. In fact, however, the deep fissuring ulcers of Crohn's disease, fistulae, fibrosis, and stenosis are prima facie evidence in vivo that proteolytic degradation of the gut wall takes place in IBD. In the past, proteases were classified according to their substrate specificity or molecular size. Now a more rational system is based on a comparison of their active sites, mechanisms of action, and three-dimensional structures. Four classes of proteases are recognized by the International Union of Biochemistry: serine proteases, cysteine proteases, aspartic proteases, and metalloproteases. These classes can be further grouped according to the mechanism by which they catalyse proteolysis; serine and cysteine proteases form covalent enzyme complexes with their substrates, whereas aspartic and metalloproteases do not. This division has important consequences, because the two mechanisms by which catalysis occurs are inhibited in completely different ways. Serine and cysteine proteases have strongly nucleophilic amino acids at their catalytic site, at which hydrogen bond acceptors are aligned to promote the dissociation of the nucleophile in the approach to the transition state and thus increase the fraction in the hyperreactive state. The other two groups, aspartic and metalloproteases, however, rely on general acid/general base catalysis of the attack of a water molecule, and therefore the catalytic residues lack the aggressive nucleophilicity. Otherwise, members of each class are believed to have descended from a common ancestor by divergent evolution. In the next section, the general mechanisms of the four classes of proteolytic enzymes are briefly reviewed; their substrate activities and specific inhibitors are summarised in Table 1. Proteolytic enzymes exist ubiquitously in all biological tissues and fluids; however, practical considerations have made some more amenable to study than others. Extracellular proteases such as pancreatic and gastric protease, mammary plasma protease, and certain bacterial proteases are readily isolated and have therefore been widely studied. Much less is known, by comparison, about intracellular tissue proteases, their enzymatic specificity, and physiological substrates. However, the overall enzymatic process (peptide-bond scission) is identical in all cases, and the differences between the catalytic mechanisms are rather subtle. Substrates and inhibitors of the four major classes of protease Substrates and inhibitors of the four major classes of protease This is the most widely studied class of enzymes in the protease field and includes two distinct families: the mammalian serine proteases such as plasminogen activators, chymotrypsin, trypsin, elastase, cathepsin G, and kallikrein and the bacterial serine proteases (e.g., subtitlisin). They differ from each other in amino acid sequence and three-dimensional structure but share a common active site with the catalytic triad of Asp 102, His 57, Ser 195 (chymotrypsin numbering), and an enzymatic mechanism that proceeds by a tetrahedral transition state intermediate during both the acylation and deacylation steps of catalysis (1). Serine proteases include the plasminogen activators and are thought to be key regulators of connective tissue turnover (2), acting by specific catalysis of the generation of plasmin from its plasma zymogen, plasminogen. Two major plasminogen activators have been described, a tissue plasminogen activator (t-PA) and a urokinasetype plasminogen activator (u-PA). Both cleave plasminogen at an Arg-Val bond near the carboxy terminus, resulting in an autocatalytic cleavage near the amino terminus of the enzyme. Considerable attention has been paid to another group of serine proteases, which are expressed by granulocytes, leukocytes, and mast cells. This group includes leukocyte cathepsin G, neutrophil elastase (3), and chymases, the chymotrypsin-like enzymes of mast cells. These enzymes are stored in granules as inactive precursors and are released in response to inflammatory or allergic reactions in an active form (4). These enzymes characteristically lack a disulphide bond near the active serine site that is present in all other known mammalian serine proteases. The serine protease inhibitors (serpins) includes α1-antitrypsin, α1-antichymotrypsin, antithrombin III, and nexin. They can inhibit most serine proteinases. They are important regulators of neutrophil elastase activity and may also act against plasmin. Other inhibitors such as plasminogen activator inhibitors (PAIs) are potential regulators of connective tissue degradation and inhibit both u-PA and t-PA. Synthetic inhibitors include diisopropyl fluorophosphate (DPF) and phenylmethane-sulphonyl fluoride (PMSF), which react with the activesite serine, and chloromethyl ketone derivatives of amino acids and peptides, which react with the histidine of the catalytic triad. α2-Macroglobulin, the universal protease inhibitor, also inhibits proteases of this class. It is a 772-kDa protein confined to the blood and is composed of four nearly identical, disulphide-bonding polypeptide chains. It displays very broad specificity, traps the protease irreversibly in the so-called bait region, and covalently anchors the trapped enzyme by a process of transacylation. This class of proteases bears great similarity to the serine protease group because a covalent intermediate is formed, the attacking nucleophile is the sulphur atom of a cysteine side chain (cysteine 25 in the papain numbering system, acting like serine 195 in chymotrypsin). This class of proteinases includes several mammalian lysosomal cathepsins, cytosolic calcium activated proteases (calpains) and the plant proteases, papain and actinidin. Cathepsins B, L, N, and S are secreted by stimulated connective tissue cells and macrophages in liver and spleen. They have been shown to degrade connective tissue proteins in vitro (5). Both cathepsins B and L are generally stored in lysosomal granules, often in a latent form requiring activation by proteolysis. They are stable at alkaline pH and are metabolically active at low pH values (6,7). These proteinases have been directly implicated in bone resorption (8), the degradation of cartilage proteoglycan and collagen (9,–11), and experimentally induce kidney disease. In contrast to the low-molecular-weight lysosomal proteases (20-40 kDa), the cytosolic cysteine proteases (350-750 kDa), which have been isolated from skeletal muscle, liver, adrenals, and reticulocytes, are adenosine triphosphate (ATP) dependent, do not require Mg, and are not inhibited by vanadate. Cystatins are a superfamily of cysteine proteinase inhibitors that have been divided into three families, type I (stefins), type 2 (cystatin C and S), and type 3 (kininogens), based on primary amino acid sequence homology and domain structure. The cystatins are tight-binding reversible inhibitors of many cysteine proteases and are not known to inhibit proteinases from other catalytic classes. The aspartic proteases include bacterial penicillopepsin, mammalian pepsin, cathepsin D, renin, chymosin, and certain fungal proteases. This group of enzymes is believed to catalyse the cleavage of peptide bonds without the use of nucleophilic attack by a functional group of the enzyme (12). The aspartic peptidases have two aspartic acid side chains (residues 32 and 215 of the porcine pepsin numbering system) in close proximity to each other. The enzymes are active in the pH range 2-3. Mice deficient for cathepsin D exhibit progressive intestinal mucosa atrophy and extensive destruction of lymphoid cells, suggesting that apart from proteolytic functions, this enzyme may regulate cell growth and tissue homeostasis (13). Aspartic proteases are inactivated by diazoacetyl compounds as well as by the acetylated pentapeptide, pepstatin A. This group of enzymes is believed to catalyse the cleavage of peptide bonds without the use of nucleophilic attack by a functional group of the enzyme. The active-site components include zinc and its three ligands of two glutamic acids and one histidine. Another glutamic acid side chain acts as the nucleophile, directly or with the participation of a water molecule. All MMPs have a putative Zn2+-binding site HEXXH, and all require Ca2+ for stability and exhibit a preferred cleavage specificity for the N-terminal side of hydrophobic residues (14,15). There are ∼19 MMPs recognised to date, and within this class, four subclasses of proteases can be recognized according to their substrates: collagenases, gelatinases, stromelysins, and membrane-type (MT-) MMPs. They are secreted by various cell types including mesenchymal cells, monocytes, macrophages, neutrophils, keratinocytes, and tumor cells. They are either cell membrane bound or membrane associated and are metal binding. Within this class, four MT-MMPs have been identified and had their genes mapped (16). Overexpression of MT-MMPs specifically results in the activation of both endogenous and exogenous gelatinase A, which is involved in both connective tissue remodeling and tumor invasions (17). MMPs are functionally active at neutral pH, secreted in proenzyme forms requiring extracellular activation. They can be activated by various reagents but usually initiate self-cleavage, probably because of the imbalance of natural tissue inhibitor TIMP and after proteolysis by other enzymes such as plasmin or mercurial compounds in vitro (18). The activated protease of this family can degrade all classes of extracellular matrix (19,20) and are often associated with tumor invasion and cartilage degradation. More recently, findings suggested that MMPs are secreted by and interact with immune cells and play an important role in immune responses (21). Four TIMPs have been identified so far. The activity of MMPs is tightly regulated by TIMPs both at the level of the rate of activation and the rate of subsequent substrate degradation. TIMPs form high-affinity, irreversible, noncovalent complexes with the active forms of MMPs and are generally expressed in resorbing tissue systems (22). EDTA, EGTA, phenanthroline, phosphoramidon, and a group of low-molecular-weight synthetic inhibitors such as hydroxamic acid peptides also show specific inhibiting ability for MMPs. Although it seems self-evident, there is actually very little documentation that proteases are increased in IBD mucosa. Neutrophil elastasc is markedly increased in stools of patients with IBD (23), and it would seem fair to conclude that degranulating and apoptotic neutrophils in the mucosa release elastase locally. Likewise it is also likely that other leukocyte serine and acidic proteinases and neutrophil collagenase will be increased in the mucosa in IBD, although this has never been shown. In an immunohistologic study with polyclonal antisera, Bailey et al. (24) examined the expression of gelatinase A and B, stromelysin, and interstitial collagenase in Crohn's disease and ulcerative colitis (UC). Gelatinase B was seen in infiltrating neutrophils. Collagenase was only rarely seen in a few cells in normal and diseased intestine. Stromelysin-1 was seen in a few mononuclear cells in normal intestine, and there was only a slight increase in diseased bowel. However, extracellular stromelysin was seen in areas of tissue injury. The most compelling evidence that MMPs are important in IBD comes from Saarialho-Kere et al. (25), who used in situ hybridisation to examine MMP and TIMP-1 expression around gastric and IBD ulcers. Very high expression of transcripts for interstitial collagenase and stromelysin-1 were seen in ulcer beds and for collagenase; at least, the cells containing the transcripts appeared to be fibroblasts. TIMP-1 also was overexpressed in the ulcer bed. Epithelium adjacent to ulcers also strongly expressed matrilysin. Mucosal mast cells contain chymases that are rapidly secreted after mast cell activation. Human mast cell chymase can cleave procollagen and result in fibril formation, which may be important in fibrosis (26). Crohn's disease is classically associated with strictures, and there is enhanced collagen III secretion by lamina propria fibroblast/smooth-muscle cells activated with transforming growth factor β (TGF-β) using cells from strictured areas (27). A different effect is seen with platelet-derived growth factor (PDGF), which actually decreases collagen III synthesis in cells from strictured areas. Interleukin 1β (IL-1β) is also a potent stimulator of collagenase messenger RNA (mRNA) and protein secretion in intestinal smooth-muscle cells but decreases procollagen synthesis (28). We have been interested in examining MMPs and TIMPs in inflamed areas of mucosa distant from ulcers. By quantitative competitive reverse transcriptase-polymerase chain reaction (RT-PCR), stromelysin-1 and collagenase transcripts are undetectable in normal bowel, but TIMP-1 is highly expressed. In both Crohn's and UC, stromelysin transcripts are markedly elevated, but TIMP-1 remains constant. Surprisingly, collagenase transcripts also were undetectable. The increased amounts of stromelysin-1 in IBD are consistent with our earlier observations that by histology there was a reduction in the density of charged residues in the lamina propria in IBD, which is probably due to a reduction in lamina propria proteoglycans (29). This is also an underinvestigated area in which there is great For example, although elastase inhibitors have been well in injury, to our they have not been in of gut injury. Likewise although MMP inhibitors are for rheumatoid and and are in they have not been in of IBD. has been used to examine gelatinase A, stromelysin, and TIMP-1 expression in the mucosa of with colitis Gelatinase A is seen in neutrophils after injury. In the first after injury, collagenase and stromelysin were and the the next few In a of gastric by of acid into the and collagenase activity is in the tissue disease has major and it was shown that as well as the of the disease, are by a matrix inhibitor The mucosal mast cell chymase when into to the gut but result in of plasma proteins into the gut The substrate for this activity has not been but may be a protein important in the We have been the and molecular for the mucosal destruction seen when the is used to the cells in the lamina propria of gut This response is cell dependent, because injury is not seen if is to from with few cells and is also by A and The first that proteases be important in this system from the that the injury was associated with the of from the lamina propria and their in the These were identified as sulphate suggesting that they were associated with tissue injury was inhibited by suggesting that proteases were In further injury was associated with markedly increased interstitial collagenase and stromelysin-1 transcripts in the tissues and increased inactive and active stromelysin and collagenase in the Gelatinase A was not were TIMP-1 and amounts of stromelysin-1 can mucosal destruction and when directly to the evidence for a major role for stromelysin in this system from the that mucosal degradation was inhibited by a stromelysin inhibitor and not a collagenase The stromelysin inhibitor was if after the of the response with and not the of the immune response in the that mucosal degradation was an stromal cells secreted large amounts of stromelysin and collagenase when activated with cytokines such as tumor factor or evidence that the injury was not by the and cytokines but were the result of increased of stromelysin by mucosal In further and more it was shown that mucosal degradation in this is inhibited by a protein with the of the tissue injury was a reduction in the of stromelysin-1 in the but not collagenase or We believe that these results are highly the use of in Crohn's disease and we would that one of the in which these is the result of their of MMP by stromal cells. This is important in the of by this is by of or is not Although vivo in gut we believe that these results are to the in Crohn's disease. There is evidence that there is an immune response in the mucosa in Crohn's Likewise in the gut the response by is strongly In however, increased of and are seen in a of gut including IBD and colitis We that the ulcerative seen in the mucosa in all these conditions are due to activation of stromelysin-1 probably other by stromal cells. It is also likely that activated cells can proteinase by stromal cells cell A for this would be an between and (21). In other the gut mucosa because of an imbalance between MMP matrix and TIMP this of stromal cell MMP by cells and of stromal cell MMP by cells and of the most important of MMPs is that they are secreted as inactive that have to be by other proteases to the active form of the enzyme. In proteases are tightly regulated by endogenous proteinase inhibitors such as the TIMPs for MMPs, for serine proteinases and cystatins for the proteinases. The major inhibits the activity of all of the classes of proteases. There is evidence that between different proteases is important in their is a potent activator of matrix metalloproteinases and is probably very important in mucosal mast cell chymase can latent interstitial collagenase and stromelysin-1 elastase can the of stromelysin-1 It is well that activated stromelysin-1 is a potent activator of interstitial collagenase It was by et al. that it is to tissue injury as due to the activity of a proteinase and matrix degradation is the result of a proteinase with the potential to be regulated at many In the gut, this also is likely to be and will also probably with disease and within the the are more important in disease in mucosal ulceration, and may be important in remodeling the mucosal during the as occurs in in chronic disease, interstitial may be important in matrix degradation in the layer, resulting in and In UC, in which there is of the mucosa by neutrophils, tissue injury may be by activation of produced by cells in response to increased cytokines, with the MMPs activated by neutrophil serine proteinases. and mechanisms the immune system to regulate proteinase by stromal cells in the In addition, proteinases produced by infiltrating inflammatory cells can tissue injury and the proteinases produced by stromal cells. We that there is evidence to that proteinases are the molecules of mucosal in IBD and are important in and fibrosis in Crohn's disease. for example, has a broad substrate range including type III, type and type and laminin and is thus to degrade major components of the mucosal extracellular the activity of MMPs should tissue the large numbers of inhibitors in the it would seem that some of these should be in a more subtle proteinases also probably regulate the of the molecules that control because in these molecules are bound to extracellular This was by the the of and the Crohn's in