Two papers in this issue of Gutfocus on RNA and protein levels of matrix metalloproteinases (MMPs) in diagnostic biopsies of patients with inflammatory bowel disease (see pages 57 and 63).1 ,2 Both reports are congruent in that in ulcerated lesions, MMP-3 mRNA, which encodes a key enzyme in matrix degradation, is highly increased (up to 15- and 100-fold, depending on the methodology used), paralleled by a more moderate increase in MMP-3 protein levels, whereas there is little change in levels of the physiological inhibitor of most MMPs, tissue inhibitor of metalloproteinases 1 (TIMP-1). Thus the elevated ratio of MMP-3 over TIMP-1 would favour matrix degradation. Similar data were obtained for MMP-1/TIMP-1 (degradation of fibrillar collagens), MMP-2/TIMP-2 (degradation of basement membrane collagen and denatured collagens), and for membrane-type MMP-1 (MMP-14) which can activate MMP-2.2

Whereas matrix dissolution is plausible for ulcerative colitis (UC), it comes as a surprise that comparable levels of expression for MMPs and MMP/TIMP-1 ratios were found in ulcers of Crohn's disease (CD) which rather leads to intestinal fibrosis. As expression of several collagens which are the major components of scar tissue was also not different between lesions of CD and UC,3 the factors that determine fibrogenesis (that is, enhanced matrix deposition) and fibrolysis (that is, stimulated matrix removal) in the gut remain to be determined.

What can we learn from these studies in terms of pathophysiology, and what diagnostic and therapeutic consequences can be derived? Firstly, reliable micromethods were established that allowed both RNA and protein quantification of several MMPs and TIMPs from a single diagnostic biopsy. Secondly, expression of various MMPs and TIMPs correlated with the histological degree of intestinal inflammation, but did not depend on its aetiology.

However, there are many explanations for the obviously quite different matrix metabolism in UC and CD. mRNA and protein quantification do not allow us to draw conclusions as to the biological activity of the MMPs, the majority of which are secreted as inactive proenzymes that have to undergo a complex proteolytic processing to become fully active. MMP processing is brought about by plasmin, furin-like proteases, or MMPs themselves (with MMP-3 and MMP-14 playing prominent roles as proactivators of other MMPs), with the proactivators again being tightly regulated.4 Furthermore, matrix degradation is usually restricted to small cell membrane associated compartments whereas the great mass of MMPs remains inactive and complexed to TIMPs. Thus mere quantification does not tell us anything about the temporospatial expression of the various MMPs or TIMPs.

Fortunately, the cell types responsible for MMP expression in normal and inflamed intestine in vivo are fairly well defined, thanks to several studies using in situ hybridisation, in part in combination with cell type specific markers.5-9 Although this knowledge may allow some extrapolation, determination of the focal proteolytic activity of individual MMPs in vivo defies current technology. The picture is further complicated by prominent sequestration of most MMP precursors in the matrix where many bind to certain collagens; this explains the relatively large amounts of pro-MMPs which are usually not stored intracellularly that can be extracted from tissues.1 ,2

Despite the complexity of MMP regulation in intestinal inflammation, some in vitro and in vivo data allow us to draw more definite conclusions. Thus mice deficient in MMP-7 and MMP-11, which in the gut are almost exclusively produced by intestinal epithelial cells (but which were not measured in the two present reports), are less susceptible to genetic or chemical intestinal carcinogenesis, and deletion of the macrophage specific metalloelastase MMP-12 prevents macrophage transmigration through basement membranes in vivo (reviewed by Nagase and Woessner4). In vitro stimulation of T lymphocytes in organ cultures of fetal intestine caused activation of several MMPs but only MMP-3 proved to be a key enzyme for degradation of the lamina propria extracellular matrix.10 As intestinal fibroblasts are the prime sources of this protease,7 ,8 and as tumour necrosis factor α (TNF-α), released by the T lymphocytes, is a powerful inducer of fibroblast MMP-3, this creates a plausible link between mucosal inflammation and destruction of the subepithelial matrix (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Metalloproteinases (MMPs) and their cellular sources in the intestine. Minor or still debated MMP expression is shown in italics. Intestinal fibroblasts appear to play the most prominent role in MMP release during mucosal inflammation and subsequent transformation. Tumour necrosis factor α (TNF-α) has been identified as a major MMP inducing cytokine. Several currently used drugs affect MMP expression and activity indirectly. MMP inhibitors may prove useful in modulating mucosal transformation.

What determines the partly divergent evolution of the lesions in UC and CD? Are there differences in the overall or localised expression of other MMPs, MMP proactivators, or of other classes of matrix degrading proteases that were not determined in the present investigations? Such differences should be expected due to the divergent cytokine profile in UC and CD which resembles a TH2 and a TH1 pattern, respectively, as cytokines are potent modulators of MMP expression and activity.

The usefulness of corticosteroids and immune suppressants such as azathioprine which act mainly on lymphocytes underscores the relevance of activated T cells, and the often dramatic improvement in complicated CD with TNF-α blockade may derive, at least in part, from inhibition of MMP activation favouring for example, closure of fistulas. TNF-α blockade does not appear to be equally effective in UC where in contrast, interferon alpha (IFN-α), a TH1-like cytokine, may be promising. The domain for quantification of MMPs, as exemplified in the two present reports, may be to monitor such therapies and, by extension of the spectrum of the molecules analysed, to find better and more specific predictors of disease activity in UC and CD. A novel therapeutic approach could be blockade of certain MMPs, such as MMP-3, by local or systemic application of synthetic MMP inhibitors. Much industrial research effort has been invested in the development of such compounds for treatment of tumours or osteoarthritic joint destruction.

Therefore, these two papers have introduced novel mucosa derived parameters which may prove useful to assess prognosis, disease activity, and treatment response in inflammatory bowel disease. By using this methodology, other genes involved in mucosal remodelling may be found that more specifically reflect the underlying aetiology. Their use in future therapeutic trials is awaited with much interest.