In this study, the expression patterns of matrix metalloproteinases of the collagenase and stromelysin subgroups, and their specific inhibitors TIMPs, were investigated in human cutaneous and intestinal wound samples using in situ hybridization and immunohistochemistry. In situ hybridization is a relevant method for investigating the cellular origin of MMPs and their inhibitors, since most MMPs are transcriptionally regulated i.e. the mRNA accumulates within the cell to detectable levels following a certain stimulus, whereas the protein is often secreted from the cells shortly after synthesis (Rodgers et al, 1994; Fini et al, 1996). However, these methods do not reveal whether these enzymes are activated or not. Due to the general lack of immunohistochemical antibodies detecting only the activated MMP form, in situ zymography on tissues or Westerns on wound fluids or tissue would be the most suitable assays for that purpose.
Figure 5. Collagenase-1, stromelysin-1 and -2, and TIMP-1 and -3 mRNA expression in a normally healing human wound. BM, basement membrane. Modified from Pilcher et al, 1998.
Our results on epithelial collagenase-1 expression, limited to basal keratinocytes bordering normally healing wounds and chronic venous ulcers (II; III), are in agreement with the data in rheumatoid, diabetic, and decubitus ulcers, in ulcerated pyogenic granulomas, and in normally healing and burn wounds obtained by us and others (II; Saarialho-Kere et al, 1992; Saarialho-Kere et al, 1993a; Saarialho-Kere et al, 1993b; Stricklin et al, 1993; Inoue et al, 1995). The expression of collagenase-1 in wound edge epithelium is a constant finding, although the pathobiology of various chronic ulcer types differs from each other. In normally healing wounds collagenase-1 is expressed during the first postoperative day, and in in vitro wounds, as early as 4 hours after wounding (II; III; Inoue et al, 1995). The same applies to cell culture conditions, where migrating keratinocytes in contact with type I collagen express collagenase-1 at least after 24 hours of culturing (Pilcher et al, 1997). Shortly after complete re-epithelialization, no traces of collagenase-1 mRNA are found in keratinocytes (II; Stricklin et al, 1993; Inoue et al, 1995). A similar pattern of expression encountered also in healing rat and mouse skin wounds as well as in porcine burn- and incisional wounds (II; Stricklin et al, 1994b; Okada et al, 1997; Madlener et al, 1998) suggests a major role for this enzyme in dermal wound repair.
An interesting theory on the induction and role of collagenase-1 in keratinocyte migration has recently been presented by Pilcher et al (1997). Contact of keratinocytes to interstitial collagen leads to high affinity binding of integrin α2β1 to collagen. This induces expression of collagenase-1, which degrades interstitial collagen into 1/4 and 3/4 fragments. These fragments denature in physiological temperature to gelatin, losing the ability to bind to integrin α2β1 (Messent et al, 1998). The liberated keratinocyte is then able to extend and rebind to intact collagen further on the wound bed. Once the contact with fibrillar collagen is abolished by completion of re-epithelialization, collagenase-1 expression is shut off (II; III; Saarialho-Kere et al, 1993b; Stricklin et al, 1993; Inoue et al, 1995). While altered cell-ECM contacts may result in the inhibition of collagenase-1 expression, modulation of the intracellular calcium may provide for the signal to stop collagenase-1 secretion in keratinocytes (Sudbeck et al, 1997).
Although in situ hybridization is not a quantitative technique, we can state that the number of both epithelial and stromal collagenase-1 positive cells was higher in chronic than normally healing wounds (II, III). Similarly, elevated collagenase-1 activity has been detected in chronic wounds (Weckroth et al, 1996; Barone et al, 1998). Even though many human studies confirm the important physiological role of collagenase-1 in keratinocyte migration (Saarialho-Kere et al, 1993b; Stricklin et al, 1993; Pilcher et al, 1997), transgenic mice expressing high levels of human collagenase-1 in the epidermis suffer from delayed wound healing due to derangements in re-epithelialization (di Colandrea et al, 1998). Also, the level of active collagenase-1 may vary; tissues and fluids from normally healing wounds contain almost exclusively inactive forms of the enzyme, while fluids from non-healing ulcers possess significant amounts of the active form of collagenase-1 (Nwomeh et al, 1999). These results indicate that the homeostasis of active collagenase-1 is critical for keratinocyte migration to succeed.
A member of the serine proteases, urokinase plasminogen activator (uPA), is expressed by migrating keratinocytes in both normally healing and chronic wounds (II; Grøndahl-Hansen et al, 1988). The main task for plasminogen activation cascade during wound healing is probably fibrin degradation, since wounds in plasminogen-deficient mice have impaired fibrin resolution and heal poorly, while wounds in mice deficient in both plasminogen and fibrin heal as quickly as wounds in normal mice (Bugge et al, 1996; Rømer et al, 1996). Most probably, plasmin also activates MMPs such as collagenase-1 and stromelysin-2 in the wound edge (see Woessner, 1991). Although members of both MMP- and serine protease families appear essential to keratinocyte migration, the migration proceeds, although slowly, with either plasminogen or MMP-activity excluded (Lund et al, 1999). Keratinocyte locomotion is ceased only if mice lack both MMP activity and plasminogen (Lund et al, 1999). These findings suggest that MMPs and serine proteases can perform each others` tasks during wound repair.
Collagenase-1 expression is not limited to the epithelium. Stromal fibroblast-like as well as macrophage-like cells express collagenase-1 in all types of wounds examined (II; III; Saarialho-Kere et al, 1992; Saarialho-Kere et al, 1993b; Stricklin et al, 1993). It is tempting to speculate that moderate expression of collagenase-1 in normally healing wounds is a part of the normal, reparative process. The prolonged presence of high amounts of collagenase-1 in chronic dermal wounds, and even in conditions preceding chronic ulcers, may implicate excess proteolysis leading to disturbed healing of the lesions (II; III; Weckroth et al, 1996; Herouy et al, 1998). A recent study by Nwomeh et al (1999) suggests that at least in chronic venous ulcer fluids, neutrophil collagenase is the predominant collagenase. We have not assessed the expression of this enzyme in wounds, but its origin most likely are the polymorphonuclear leukocytes, which are more or less abundant in chronic wounds (Herrick et al, 1992). During the inflammatory phase of acute wound healing, the neutrophils are the first cells to arrive to the injured area (see Clark, 1995), which makes it probable that collagenase-2 liberated from their granules would be responsible for initiating collagenolysis in wounds. This neutrophilic reaction may persist too long in chronic wounds, and together with the lack of activated macrophages (Moore et al, 1997), may lead to altered cytokine profiles and excessive MMP expression.
The third MMP capable of degrading fibrillar collagens, collagenase-3 (Freije et al, 1994; Knäuper et al, 1996), has a different pattern of expression in wounds than collagenase-1. Its expression was never detected in the epithelium of any wound type, nor in the stroma of normally healing wounds (III). Contrasting acute dermal wounds, stromal fibroblasts are able to express MMP-13, when acute oral wounds heal (Ravanti et al, 1999b). This may reflect the different cellular environments of dermal and mucosal fibroblasts, and imply that studies on mucosal healing cannot always be directly extrapolated to dermal wound healing. Interestingly, collagenase-3 mRNA is expressed by inflamed gingival keratinocytes protruding towards underlying connective tissue. These MMP-13 positive cells colocalize with laminin-5 (Uitto et al, 1998), and are detected in a localization similar to gelatinase A in inflamed oral mucosa (Mäkelä et al, 1999). Thus, gelatinase A and collagenase-3 may be operational in oral wound healing and in pathological epithelial migration, while collagenase-1 assists in keratinocyte migration during dermal wound repair (Mäkelä et al, 1999). This is supported by our findings on collagenase-3 in invasive carcinomas of the skin and the mucosa, but not in the epithelium of healing cutaneous wounds (III; Airola et al, 1997; Johansson et al, 1997c; Johansson et al, 1999).
In the skin, collagenase-3 mRNA expression was detected only in fibroblast- and macrophage-like cells in fibrotic areas of chronic venous ulcers deeper in the wound bed than collagenase-1 mRNA, suggesting distinct roles for these enzymes (III). At least part of the collagenase-3 positive cells were fibroblasts, as assessed by staining for procollagen-1 (III). These fibroblasts may be stimulated to express collagenase-3 by contact with surrounding collagenous matrix mediated by integrin receptors α1 β1 and α2 β1, based on the following findings: 1) normal human skin fibroblasts cultured inside a three-dimensional collagen-gel, but not when grown in monolayer, express collagenase-3 (III) and 2) the induction generated by three-dimensional collagen is mediated by integrin receptors α1 β1 and α2 β1 (Ravanti et al, 1999a). Since collagenase-3 has a wider variety of substrates than collagenase-1, being also able to degrade e.g. gelatin (Knäuper et al, 1996), it may participate in the remodeling of the immature collagenous matrix by degrading fibrillar collagens and their cleavage products, as well as other connective tissue components. Furthermore, collagenase-3 activates progelatinase B in a two-step cleavage mechanism in vitro (Wysocki et al, 1993; Bullen et al, 1995; Knäuper et al, 1997), and may activate this enzyme also in chronic ulcers in vivo. MT1-MMP activates both collagenase-3 and gelatinase B, yet by distinct mechanisms (see Murphy et al, 1999). MT1-MMP, gelatinase B and collagenase-3 are expressed concomitantly at least in rat wounds (Okada et al, 1997), and vulvar squamous cell carcinomas (Johansson et al, 1999), and all these enzymes can be involved in complex activation cascades of the wound environment.
In our dermal wound samples, stromelysin-1 was expressed by keratinocytes distinct from but adjacent to the migrating front of keratinocytes, in part of both normally healing and chronic wounds (II). Due to its location stromelysin-1 is not likely to participate in keratinocyte migration. Rather, it may participate in the remodeling of the newly formed basement membrane. Stromelysin-1 is expressed in proliferating keratinocytes that already have a BM beneath them, composed of the stromelysin-1 substrates laminin and type IV collagen (II; Juhasz et al, 1993; Wilhelm et al, 1987). Also tenascin, a glycoprotein abundant in healing wounds, is readily degraded by stromelysin (see Imper & van Wart, 1998). During wound healing, tenascin mRNA is expressed by basal keratinocytes, resulting in an intense staining of this glycoprotein in the papillary dermis underneath the incomplete BM, as assessed by staining of heparan sulphate and laminin (Aukhil et al, 1995; Latjinhouwers et al, 1996; Vaalamo et al, unpublished results). Stromelysin-1 may also act in removing excess tenascin after completion of migration, when its cell-spreading activities are no longer needed. The activity of stromelysin-1 may be controlled by TIMP-1 and -3 which in acute wounds are expressed in areas partly colocalizing with stromelysin-1 expression (II; V).
As shown in this study for the first time, stromelysin-2 mRNA is expressed in migrating keratinocytes in both acute and chronic skin wounds, as well as in full-thickness murine wounds (II; Madlener et al, 1996). Wounding of the skin results in transition of stationary keratinocytes into migrating cells, and these cells start to actively produce laminin-5 (Larjava et al, 1993a; Pyke et al, 1994). Production of this protein partly co-localizes with stromelysin-2 expression (II; Vaalamo, unpublished results; Larjava et al, 1993a). Laminin-5 has been shown either to stimulate (Zhang & Kramer, 1996), or to inhibit epithelial cell movement (O´Toole et al, 1997). Furthermore, specific cleaveage of laminin-5 by gelatinase A induces migration of breast epithelial cells (Giannelli et al, 1997). In dermal wounds, the expression and activity of gelatinase A is rather weak (Salo et al, 1994; Young & Grinnell, 1994). Since wide-spectrum proteinase stromelysin-2 is expressed in approximity or co-inciding with laminin-5 production, it is tempting to speculate that it participates in degradation of this BM component in healing wounds, and possibly accelerates re-epithelialization. However, the capacity of stromelysin-2 to degrade laminin-5 remains uninvestigated to our knowledge, and needs to be determined. Fibronectin, which is abundant in the granulation tissue, provides another candidate target for stromelysin-2 (Clark et al, 1992; Juhasz et al, 1993). However, recent studies suggest that at least in chronic wounds, neutrophil elastase is the proteinase responsible for excessive degradation of fibronectin (Rao et al 1995; Grinnell & Zhu, 1996), and may perform this task in normally healing wounds as well.
In this study, we found TIMP-1 and TIMP-3 mRNA expression only in the epithelium of normally healing wounds (II; V). Contrasting this, TIMP-3 protein was widely expressed by proliferating keratinocytes in both wound types. This discrepancy may be explained by mRNA levels that are too low in chronic wounds to be detected by in situ hybridization. TIMP-1 and -3 are regulated in response to cell-ECM interactions and by various cytokines; e.g. laminin-1 inhibits TIMP-1 expression while collagen type I stimulates it in cultured keratinocytes (Petersen et al, 1990). Furthermore, TGF-β stimulates the expression of TIMP-3 at least in primary human keratinocytes (Airola et al, 1998). TIMP-1 and -3 expression induced in wound edge keratinocytes may be needed in 1) the protection of the newly formed basement membrane from degradation by epithelial MMPs, e.g. stromelysin-1 (II), 2) in the regulation of keratinocyte proliferation due to their growth promoting activities (see Gomez et al, 1997), or 3) the protein released from the cells into the matrix may be needed to inactivate excess MMPs produced by the leading edge of keratinocytes. TIMP-2 protein locates beneath the migrating front of keratinocytes, and it is found in greater amounts in acute wounds than in chronic ones (V). Whereas TIMPs-1 and -3 may act chiefly in neutralizing secreted collagenase-1, TIMP-2 may neutralize the gelatinases, according to their in vitro preferences (Howard et al, 1991; Apte et al, 1995).
Expression of TIMPs is not limited to epithelial structures. In addition to TIMP-2 protein, expression for TIMP-1 mRNA, and TIMP-3 mRNA and protein was detected in the stromal compartment of healing wounds. Stromal expression of these enzymes may be needed in the inactivation of stromal MMPs such as collagenase-1 and stromelysin-1 (II; III; Stricklin et al, 1993), and collagenase-3 in chronic ulcers (III). Localization of TIMPs-1 and -3 surrounding blood vessels suggests a role in regulating wound angiogenesis (Johnson et al, 1994; Anand-Apte et al, 1997). TIMP-4 is the most recently cloned of the TIMPs (Greene et al, 1996). It exhibits a restricted pattern of moderate expression in human tissues such as the heart, kidney, placenta, colon and testes (Greene et al, 1996). Accordingly, we found TIMP-4 protein only in sporadic activated fibroblast-like cells of chronic wound stroma. Thus, the role of TIMP-4 appears a minor one in dermal wounds, compared to other TIMPs (II; V).
Tissue inhibitors of metalloproteinases are considered to be the major inhibitors for MMPs. However, unspecific inhibitors such as alpha 2-macroglobulin may also participate in the regulation of MMPs in wound environment, since collagenase-1-alpha 2-macroglobulin complexes have been detected in both normally healing burn wounds and chronic venous ulcers (Grinnell et al, 1998). Our studies indicate that the relative lack of local inhibitors of MMPs, TIMPs-1, -2, and -3, in chronic ulcers may lead to inadequate activation of MMPs, with subsequent digestion of the ECM, and perhaps, delayed migration and disturbed wound healing.
Excluding the suction blisters, the wound samples we used in our studies were from elderly people (II; III; V). Aging alters the inflammatory response to injury; wounds of the aged show an early marked increase in the neutrophil response, and delayed monocyte/macrophage and lymphocyte appearance together with alterations in the inflammatory and endothelial cell adhesion molecule profiles (Ashcroft et al, 1998). The age-related increase in gelatinases A and B with gelatin degradation and proteinase activity is detected in acute wounds, while collagenase-1 and stromelysin-1 profiles do not differ in the young and the aged (Ashcroft et al, 1997a). TIMP-1 and -2 mRNA levels in wounds of the young are elevated between 3-21 days, while the aged have a constant basal level expression of TIMP-1 and -2 (Ashcroft et al, 1997b). Our results on MMP-expression in normally healing wounds represent the events in wounds of the aged, and may differ in some aspects from normally healing wounds in the young. However, the fact that both the acute and the chronic wound samples examined were from people over 50 years of age, makes comparing of these wound types more reliable.
Figure 6. Collagenases-1 and -3, stromelysins-1 and -2, matrilysin, macrophage metalloelastase, and TIMPs-1 and -3 mRNA expression in IBD with intestinal ulceration.
Contrasting the only previous study demonstrating sporadic collagenase-1 protein expression in IBD (Bailey et al, 1994), we detected abundant expression of collagenase-1 mRNA in the stroma of IBD lesions (I). The signal located beneath the ulcers in plump cells, which were most probably activated fibroblasts, since double labeling showed distinct location of collagenase-1 mRNA and the macrophage marker CD-68 (I). Not only was collagenase-1 mRNA expressed in IBD; a majority of chronic intestinal ulcer samples showed expression for collagenase-3 mRNA in the granulation tissue, yet in fewer cells than collagenase-1 (I; IV). Unlike in skin, the signal was not confined to fibrotic areas or cells surrounding blood vessels. Rather, deeply penetrating ulcers showed diffuse signal in fibroblast and macrophage-like cells of the granulation tissue (IV). In both dermal and intestinal ulcers, at least part of the positive cells were shown to be fibroblasts, while we could not demonstrate any macrophage being responsible for the expression of collagenase-3 mRNA (III; IV). The epithelium remained negative for both collagenase-1 and collagenase-3 in the gut (I; IV).
Inflammation in the intestine may be responsible for the induction of MMPs in IBD. Chronic immune activation outside as well as inside the lamina propria has been suggested in these diseases (Pallone et al, 1987). Strong evidence for T-cell mediated damage of the intestine is obtained from studies with fetal ileal organ explants (Pender et al, 1996; Pender et al, 1997); collagenase-1, gelatinase A, and stromelysin-1 are produced in response to T-cell activation, and they are upregulated in the mesenchymal cells by IL-1β and TNF-α. Elevated levels of IL-1β and TNF-α have been shown in IBD in vivo as well (Ligumsky et al, 1990; Cappello et al, 1992; Murch et al,1993). Furthermore, recombinant stromelysin-1 added to mucosal explants results in tissue destruction within 24 hours, while synthetic inhibitor of MMPs reduces mucosal damage caused by T-cell activation (Pender et al, 1997). This provides evidence that activation of lamina propria T-cells responding to luminal antigens may lead to tissue damage by MMPs. T-cells may not only mediate tissue damage by altered cytokine/growth factor profiles, but are also able to express MMPs such as stromelysin-2 and gelatinase B (Conca & Wilmroth, 1994; Johnatty et al, 1997). In our studies, IBD specimens showed intense mRNA expression for collagenase-1, stromelysin-1, and HME, and moderate expression for collagenase-3 by the stromal cells (I; IV). While the collagenases may be responsible for the digestion of fibrillar collagens, stromelysin-1 and HME could mediate the loss of lamina propria glycosaminoclycans such as chondroitin and dermatan sulphates (Murch et al, 1993; Pender et al, 1996; see Chandler et al, 1997; see Wilson 1998). Altogether, these results indicate that T-cells may have an important role in mediating tissue damage by MMPs in IBD in vivo, too. Mast cells, which are particularly numerous in the intestinal wall (see Pope, 1998), release serine proteinases chymase and tryptase and may contribute to the intestinal damage by activating stromal MMPs collagenase-1 and stromelysin-1 (Lees et al, 1994).
As in dermal wounds, stromelysin-2 mRNA was detected in migrating intestinal epithelium bordering the ulcers, erosions and ruptured crypt abscesses, partly co-localizing with laminin-5 (I; IV). KGF is a cytokine capable of inducing stromelysin-2 at least in cultured HaCat keratinocytes, and its mRNA and protein expression is upregulated in both murine dermal wounds and, up to 150-fold, in IBD (Werner et al, 1992; Brauchle et al, 1996). This potent mitogen for both keratinocytes and gastrointestinal epithelial cells may contribute to stromelysin-2 induction in both dermal and intestinal wounds in humans. In IBD, we detected stromelysin-2 in areas of disrupted epithelium (IV), and analogously, expression of KGF was particularly high in areas characterized by epithelial damage in IBD (Brauchle et al, 1996).
Like collagenase-1 in dermal wounds, matrilysin was detected in migrating intestinal epithelial cells (I; IV). Both matrilysin mRNA and protein were expressed by the wound edge intestinal epithelium, while in the skin, only normal glandular epithelium expresses matrilysin, with no induction during wound healing (II; Saarialho-Kere et al, 1995). This may reflect differences in the composition of the ECM in these organs. While dermis is dense and rich in fibrillar collagens, the lamina propria of the intestine is loose and cellular, and there is no such need for degradation of fibrillar collagen as in the skin. Our results also support the previous findings of matrilysin mainly in cells and tissues of secretory epithelial origin (see Wilson et al, 1998). In contrast to findings in mice (Wilson et al, 1995), matrilysin is not expressed in intact intestinal mucosa (I; IV). In mice, matrilysin co-localizes in the Paneth cell granules with bactericidal cryptidins, and in vitro it is able to process cryptidin precursors (Wilson et al, 1999). Thus, it may function in intestinal mucosal defense in mice. We did not assess matrilysin in normally healing intestinal wounds due to the lack of proper samples. This makes it difficult to judge whether matrilysin really is needed for normal wound repair, without underlying disease and inflammation. Matrilysin may also play a role in the pathogenesis of inflammatory bowel disease, since it is able to process proTNF-α, an inflammatory mediator found in IBD tissues (Cappello et al, 1992; Murch et al, 1993; Chandler et al, 1996).
Our study provides the first evidence of matrilysin in human intestinal epithelium without malignant progression (I). Earlier studies focused on the role of matrilysin in both premalignant adenomas as well as colon and gastric carcinomas, with no expression in normal or inflamed intestine (McDonnel et al, 1991; Newell et al, 1994; Yamamoto et al, 1994). It has been suggested, that gastric carcinomas use integrin α6β4, laminin-5, and laminin-1 as the machinery of adhesion and invasion (Tani et al, 1996). Among these, β4-integrin subunit is degraded by matrilysin in vitro (von Bredow et al, 1997). The anchoring complex components α6 and β4 integrin subunits are a part of normal intestinal epithelium as well (Leivo et al, 1996). Since expression of matrilysin by migrating cells of the intestinal epithelium was a constant finding (I; IV), it is tempting to speculate that both benign and malignant intestinal epithelial cells use matrilysin to detach from the anchoring complex by degrading β4-integrin subunit, when migrating.
EGF is involved in protection as well as healing of gastrointestinal mucosa (Konturek et al, 1991). It may, together with TGF-α, stimulate cell proliferation and DNA synthesis in the healing of gastric lesions, and protect the intestine against ethanol or stress induced damage in rats (Konturek et al, 1992). Other cytokines that have been shown to accelerate healing of GI-lesions include PDGF and TGF-β (Mustoe et al, 1990; see Jones et al, 1999). Culturing enterocytes on collagen stimulates migration more than laminin or fibronectin (Basson et al, 1992a; Basson et al, 1992b). However, EGF and TGF-β stimulate cultured enterocytes migration only on laminin (Basson et al, 1992a; Basson et al, 1992b). Different effects on migration may reflect specific repair processes of the intestine; namely restitution and repair by reepithelialization and proliferation. These growth factors may also influence the expression of matrilysin and stromelysin-2 in the migrating cells. It would be interesting to know whether these enzymes are involved in both repair processes. Unfortunately, we were not able to determine any areas with only restitution going on. Therefore we can only suggest that in the intestine, cells migrating over the connective tissue express matrilysin and stromelysin-2, and require these enzymes probably for degradation of ECM/BM components (I; IV).
No expression of matrilysin was detected in the connective tissue compartments of dermal or intestinal ulcers (I; II). However, macrophage metalloelastase was expressed intensively by macrophage-like cells in IBD (IV). All the samples of IBD were positive for metalloelastase, and the positive cells located in two distinct areas: 1) underneath the shedding epithelium and 2) within the macrophage-rich zone of the inflamed connective tissue. HME has a broad substrate specificity with the ability to degrade BM components such as type IV collagen, laminin and fibronectin (see Chandler et al, 1997). We suggest a role for this macrophage-derived enzyme in the shedding of intestinal epithelium, possibly preceding formation of erosion/ulcer during the course of IBD (IV). Deeper in the stroma, this enzyme may also aid in macrophage migration, a process in which it has previously been implicated (Shipley et al, 1996).
Like in chronic cutaneous wounds (II; V), no TIMP-1 nor TIMP-3 mRNAs were detected in the epithelium of chronic intestinal ulcers (I; IV). No TIMP transcripts were detected in the epithelium even in intestinal samples with histologically active re-epithelialization (IV). When comparing the expression of TIMP-3 in cutaneous and intestinal wounds, one must bear in mind its tightly scheduled expression; only 3-5 days old skin wounds exhibited epithelial transcripts for TIMPs (I; V). Determining the duration of intestinal ulcers and erosions is practically impossible, and clinical course of the IBD suggests rather long-lasting injuries.
The inflammatory infiltrate of the chronic intestinal ulcers and erosions showed strong signal for both TIMP-1 and TIMP-3 (I; IV). The expression of these inhibitors co-localized partly with that of MMPs such as collagenases-1 and -3 and stromelysin-1 (I; IV). TIMPs may be upregulated in the intestine to prevent damage caused by the MMPs. Various cytokines and growth factors of the inflamed intestine may trigger TIMP production, and it is tempting to speculate that factors beneficial to intestinal wound healing, such as IL-10 and EGF (see Anand-Apte et al, 1996; Pender et al, 1998a; see Jones et al, 1999) would mediate the upregulation of TIMPs, and in this way limit tissue destruction by MMPs.
On one hand, the location of TIMP-3 mRNA around blood vessels in IBD proposes a role for it in stabilizing vessel walls, since it has antiangiogenic properties (Anand-Apte et al, 1997). On the other hand, overexpression of TIMP-3 in rat vascular smooth muscle cells promotes apoptosis (Baker et al, 1998) which does not exclude an alternate role for TIMP-3 in mediating apoptosis of vascular cells with subsequent vessel regression.
Inflammatory bowel disease lesions have elevated levels of TNF-α (Murch et al, 1993; Ligumsky et al, 1990), which may mediate tissue damage by accelerating the inflammatory response. The p55 TNF receptor immunoadhesin prevents TNF-α-mediated MMP-production and tissue damage of the intestine (Pender et al, 1998b). Conversion of this cytokine from proform into active form can be prevented by TIMP-3, an inhibitor of TACE (TNF-α converting enzyme) (Amour et al, 1998). Thus, TIMP-3 may inhibit proteolysis not only by local inhibitory acitivity, but also by preventing MMP-production.
Even though MMPs have been implicated in developmental and disease processes, their actual role remains unraveled. Many in vitro studies have elucidated their substrate specificities and activities, but none of these results have been proven in vivo (see Woessner, 1998). Therefore, it can only be suggested that they degrade ECM molecules also in vivo. To resolve the actual consequences of MMP activity in tissues, various MMP-knockout mice have been generated (see Shapiro, 1998). Indeed, different in vivo tasks for MMPs can be suggested according to these studies. For example, macrophages from HME-deficient mice fail to penetrate basement membranes (Shipley et al, 1996), and gelatinase A and matrilysin deficiencies inhibit tumor growth (Itoh et al, 1998; Wilson et al, 1997), indicating roles for these MMPs in macrophage migration and tumor cell invasion, respectively. However, wound healing has only been investigated in mice deficient in stromelysin-1 (Bullard et al, 1999). These mice display no disorders in migration or reepithelialization, but suffer from impaired wound contraction (Bullard et al, 1999). The viability and rather normal development of many MMP-knockout mice suggest that many of the tasks of a single enzyme can be performed by other proteinases, due to overlapping degradative activities. This functional overlap is demonstrated by impaired wound healing in mice with either plasminogen deficiency or galardin-blocked MMP-activity, but a total failure to heal in mice with both plasminogen deficiency and galardin-treatment (Lund et al, 1999).
Many of the studies cited here were performed with mice, rats, rabbits or pigs (Stricklin et al, 1994b; Fini et al, 1996; Madlener et al, 1996; Okada et al, 1997; Madlener et al, 1998). We too performed part of our work with animals: collagenase-1 and TIMP-1 mRNA expression in incisional pig wounds confirmed the findings in humans (II), as was the case in burn wounds of pigs and humans (Stricklin et al, 1993; Stricklin et al, 1994b). In normally healing rat wounds, MMPs such as collagenase-3, gelatinase A, MT1-MMP and stromelysin-1 were coordinately expressed, and MT1-MMP was shown to activate progelatinase A in the wound stroma (Okada et al, 1997). These results partly correspond to those of humans. Of interest was the expression of collagenase-3 in the migrating epithelium, which did not occur in normally healing human skin wounds (Okada et al, 1997; III). This may be explained by rat collagenase-3 considered as the counterpart of human collagenase-1 (Knäuper et al, 1996). Similarly as in humans, no matrilysin was found during rat skin wound healing (Okada et al, 1997). Expression of stromelysin-2 in wound healing in mice, and lack of the enzyme in rat wounds, as well as different MMP-expression patterns in rat and rabbit corneal wound healing, provide examples of MMP variations between rodents (Madlener et al, 1996; Fini et al, 1996; Okada et al, 1997). In order to determine stromelysin-2 and collagenase-3 expression in normally healing intestinal wounds, we used samples of rat ileal anastomosis (IV). These samples provided somewhat unexpected results: stromelysin-2 mRNA expression did not localize in the migrating epithelium, but in the granulation tissue, as the expression for collagenase-3 mRNA (IV). In mice, matrilysin was constitutively expressed by Paneth cells in crypts of normal small intestine (Wilson et al, 1995), which was not supported by our work in human intestine (I; IV). Rats with acetic acid induced gastric ulcers displayed a rapid elevation in collagenase-1 and gelatinase B levels, with gradual return to control values during the healing phase (Baragi et al, 1997). Colonic anastomoses resulted in elevated MMP and TIMP-1 expression confined to the suture line, while anastomoses combined with ischemia or colonic obstruction resulted in a more widespread expression of MMPs and TIMP-1 (Savage et al, 1997; Savage et al, 1998). Data on MMPs in acute human intestinal lesions would be valuable, but such samples are difficult to obtain. Altogether, these slightly varying, species dependent results indicate that even though work that has been done with animal models provides valuable information, direct conclusions to healing human wounds cannot be drawn.