The significant role played by proteolytic enzymes in the invasive behaviour of embryonic cells has been known for many years (see Sherbet, 1982,1987). Embryonic trophoblast cells invade the uterine epithelium, the basement membrane, connective tissue and blood vessels in the processes of implantation and the development of the placenta; process which bear considerable analogy with the processes involved in the invasion and metastatic dissemination of cancer cells. The trophoblast cells adhere to the basement membrane and degrade it by the agency of the metalloproteinases, type IV collagenase and interstitial collagenase, and also serine proteases. The metalloproteinase activity associated with the invasive faculty of trophoblasts is controlled by the induction of TIMPs by TGF-/J which is produced by the decidua (reviewed by Lala and Graham, 1990).
A direct correlation between invasive and metastatic ability and the production of MMPs by cancer cells was documented several years ago. Type IV collagenase was shown to be associated with the degradation of basement membrane collagen and with the invasive and metastastic potential of tumour cells (Liotta et al., 1980; Turpeenniemi-Hujanen et al., 1985; Nakajima et al., 1987; Reich et al., 1988; Murphy et al., 1989). Cell transformation produced by exposure to chemical carcinogens has been known to produce alterations of stromelysin-2 gene transcripts (Matrisian et al., 1986a; Ostrowski et al., 1988; Matrisian and Bowden, 1990). Transformation of cells with the H-ras oncogene has similarly been shown to result in the secretion of collagen type IV, together with the gain of metastatic properties (Collier et al., 1988). Oncogenic transformation by ras and myc genes often also generates phenotypes with metastatic ability, but this can be inhibited by the introduction of the E1A gene. Coincidentally, the introduction of the E1A gene also causes an inhibition of MMP (Bernhard et al., 1995). Similar experiments on rat embryo cells transformed by ras and ras+El A oncogenes, were also described by Sreenath et al. (1992) who confirmed a high level of production of stromelysins 1 and 2 by transformed cells with high metastatic potential compared with transformed cells of low metastatic potential. Antibodies raised against type IV collagenase inhibited the invasion in vitro of reconstituted basement membrane by A2058 human melanoma cells (Hoyhtya et al., 1990), emphasising the role played by the MMP. It was found that TIMP-2 could inhibit this invasive and metastatic ability. Ponton et al. (1991) detected reduced expression of TIMP in metastatic variant cell lines derived from SP1 murine mammary adenocarcinoma compared with non-metastatic variant lines. Indeed, alteration of either component of the MMP-TIMP system would be expected to, and does, modulate the invasive and metastatic properties. Antisense RNA-induced TIMP down-regulate in Swiss 3T3 cells which are non-tumorigenic, conferred invasive properties on them (Khokha et al., 1989). The TIMP-1 gene has been transfected into human gastric cancer cells, and upon subcutaneous implantation into nude mice, these transfectant cells have been found to possess greatly reduced ability to metastasise (Watanabe et al., 1996). Albini et al. (1991) showed that alteration of the balance between MMP and TIMP-2, brought about by the addition of extraneous TIMP-2 or antibodies to MMP-2, affected the invasive ability of HT-1080 human fibrosarcoma cells. DeClerck et al. (1992) transfected an invasive cell type of rat embryos cells with TIMP-2 cDNA and demonstrated that TIMP-2 expression in the transfectants suppressed the invasive ability of these cells as well as inhibiting the formation of tumour nodules in the lung upon intravenous injection. Compatible with these observations, we have found a down-regulation of TIMP-2 expression in six invasive glioma cell lines tested compared with non-invasive cell lines and human fetal brain cells (Merzak et al., 1994b). Curiously, in one non-invasive cell line, TIMP-2 was in fact found to be down-regulated, which provides further support to the view that a balance between MMP and TIMP needs to be achieved in the manifestation of a given degree of invasive behaviour. As noted above, the ras oncogene and the adenovirus E1A gene modulate the expression of MMPs. Compatible with this is the observation of the influence of the metastasis-promoting gene 18A2/mtsl (page 191) in relation to invasive behaviour. Merzak et al. (1994b) demonstrated that invasive glioma cells show high levels of 18A2/mtsl transcription and in parallel show a total down-regulation of TLIP-2. In contrast, the non-invasive glioma line and fetal brain cells do not have detectable levels of 18A2/mtsl transcripts but these cells do express TIMP-2. The implications of these observations to the invasive process is significant. We have postulated that 18A2/mtsl genes may confer invasive and metastatic properties by remodelling the ECM and presumably this is achieved through enhancing MMP activity by down-regulating the expression of the TIMPs.
With the vast amount of data available on the association between the MMP/TIMP system and the invasive and metastatic phenotype may be regarded by some as beyond debate. Nevertheless, there are important questions which have remained unanswered over the past decade. Ostrowski et al. (1988) noted no differences in the levels of MMP secretion between some primary tumours and their metastatic deposits. This suggests that the expression of MMP may be heterogeneous and further that the subpopulation of cells which metastasise may not produce MMPs. The expression of MMP-2 and MMP-9 has been reported in the plasma membrane of breast carcinoma cells, but a considerable variability in expression has also been described (Visscher et al., 1994). These authors found no relationship between MMP expression and the clinical course of the disease, but the expression of TIMP-2 did relate to progression. Furthermore, they found that TIMP-2 activity was detectable in the stromal component of the tumours. The contribution of stromal cells to the proteolytic pool has been documented by others. Basset et al. (1990, 1993) found high expression of stromelysin-3 in stromal cells of breast cancer and stromelysin is expressed in this way in other tumours such as head and neck cancers (Muller et al., 1993) and basal cell carcinomas (Wolf et al., 1992). Kawami et al. (1993) found the expression of stromelysin-3, rather than type IV collagenase to correlate strongly with metastasis of breast cancer to the lymph nodes. They suggested that the MMP might be produced by stromal cells surrounding the cancer cell. MMP expression occurs at higher levels in the connective tissue stroma of colorectal carcinomas compared with adenomas and normal mucosal tissue and the most intense staining was found in the stromal component associated with neoplastic glands. The pattern of expression of TIMP was similar to that of MMP (Hewitt et al., 1991). MMP-2 and also MMP-9 were expressed at significantly higher levels in transitional cell carcinoma of the bladder than in normal bladder tissue. Both MMPs were associated with invasive rather than with superficial tumours and were expressed chiefly in stromal tissue rather than in tumour cells (Davies B et al., 1993). Presumably, the secretion of the proteinases by stromal cells might aid the dissemination of the metastasising subpopulation. In other words, the faculty of invasion and metastasis may be a paracrine phenomenon. Metastatic deposits themselves can disseminate and presumably, again, the paracrine phenomenon might account for this.
It is obviously of great significance to determine the intratumoral distribution of the proteinases, and whether the absence of MMP in metastatic deposits reported by Ostrowski et al. (1988) is a one-off event. It may be recalled, however, that the fraction of the tumour cell population producing MMPs, in this case type IV collagenase, has been found to correlate with nodal spread as well as with the presence of distant metastases (Grigioni et al., 1986; Cioce et al., 1991). A study of 186 cases of node-negative breast cancers revealed that the occurrence of a high proportion (>80%) of type IV collagenase (MMP-2) positive cells is associated with greatly increased local invasion and recurrence, but tumours with far lower MMP-2 positivity showed significantly higher incidence of distant metastases (Daidone et al., 1991). These data seem to imply essentially that metastatic dissemination occurs as an event distinguishable from local invasion and recurrence. A further inference from this study is that local dissolution of basement membrane might be occurring around only a small proportion of tumour cells. This generates a further conundrum that a small subpopulation of cells might be responding to paracrine signals and begin to express proteolytic enzymes that will enable them to access the vascular compartment for dissemination.
The metalloproteinases, MMP-3, MMP-7 (pump-1), and MMP-10 have been studied in human gastric and colon carcinomas. Of these, only MMP-7 has been found to be expressed in a majority of these tumours and this expression appears to be restricted to tumour cells and is not found in stromal cells or lymphocytes (McDonnell et al., 1991). After an investigation of the expression of MMP-2 in lung cancer cell lines, Zucker et al. (1992) concluded that factors other than MMP alone might be involved in the invasive and metastatic process. This is based on two important observations. They found that high levels of MMP-2 were associated with high tumorigenicity and invasive and metastatic potential of lung cancer cell lines. They then transfected these cells with K-rev-1 which reverts them to a less malignant phenotype. Zucker et al. (1992) noted that the revertants produced MMP-2 at the same levels as the parent cell line, but possessed slightly lower invasive ability and a vastly reduced metastatic potential.
Even after more than a decade of studies, there is much uncertainty about many aspects of the expression of the MMP/TIMP system in relation to cancer invasion and metastasis. Undoubtedly, there has emerged a clear correlation between the expression and regulation of MMP activity and the invasive process, with MMP expression often prominent in the invasive zones of a tumour, or at the interface between tumour and stromal component of the tumour. Apparently, no stringent substrate specificity is exhibited by MMPs. Nevertheless, some MMPs have been claimed to be more tumour-associated or related to progression than others. The significance of these perceived specificities of association is not understood at present. No solid evidence is yet available about whether MMP expression constitutes an autocrine or paracrine function of the cancer cell and this is a very important question that needs to be addressed. The influence of other genes, such as those coding for growth factors and metastasis suppressor and dominant metastasis genes, on the possible modulation of MMP/TIMP, together with possible repercussions for the invasive and metastatic properties of the cancer cell, also needs to be studied in depth. Here again we have correlative conclusions rather than incisive experiments providing insights into the mechanisms of gene interactions. Not surprisingly, therefore, there have not been any significant studies of the importance of MMP/TIMP expression as a prognostic aid and presumably this reflects these uncertainties.
Was this article helpful?