Immunohistochemical detection of membrane-type-1-matrix metalloproteinase in colorectal carcinoma

We investigated whether the expression of membrane-type-1 matrix metalloproteinase (MT1-MMP), matrix metalloproteinase-2 (MMP-2) and tissue inhibitor of metalloproteinase-2 (TIMP-2) was consistent with the proposed roles of these proteins in promoting metastasis in colorectal cancer. The expression of MT1-MMP was significantly more frequent in deeply invasive carcinomas (P = 0.007) and in cases of vascular invasion (P = 0.02). The frequency of detection of MMP-2 in the stroma was much greater in vascular invasion-positive cases (42%) than in negative cases (20%;P = 0.02). The rate of detection of TIMP-2 in tumour cell cytoplasm increased with the depth of invasion (P = 0.03). TIMP-2 in the stroma was found more frequently in tumours with lymphatic invasion and lymph node metastasis (P< 0.05). Significant correlations were found between detection of MT1-MMP and MMP-2 in tumour cell cytoplasm (P< 0.05), of MT1-MMP and TIMP-2 in tumour cell cytoplasm (P< 0.01), and of MMP-2 and TIMP-2 in tumour cell cytoplasm (P< 0.01). Immunohistochemical detection of MT1-MMP and TIMP-2 might be useful for monitoring infiltration in colorectal carcinoma but is not correlated with distant metastases. © 2000 Cancer Research Campaign

consistent with the proposed roles of these proteins in the promotion of the metastatic behavior of tumours.

Patients and tumour samples
A total of 92 adenocarcinomas of the colon and rectum were studied. Tumours were obtained surgically between 1988 and 1993 at the Department of Surgery II, Oita Medical University. There were 11 cases of simultaneous distant metastasis and nine cases of allochronic distant metastasis. All specimens were fixed in 10% buffered formalin and embedded in paraffin. The tumours were staged according to the standard TNM classification (Sobin and Wittekind, 1997).

Immunohistochemistry
For immunohistochemical analysis, 4 µm-thick sections were cut from formalin-fixed, paraffin-embedded blocks and placed on silan-coated slides. After deparaffinization, the sections were incubated in 3% hydrogen peroxide for 20 min in order for devitalization of the peroxidase. Deparaffinized and rehydrated specimens were heated in 10 mM citrate buffer, pH 6.0, for 10 min in an autoclave at 121°C. After cooling to room temperature (RT) for 30 min, the specimens were incubated with normal rabbit serum for 15 min at RT. Then they were incubated with various primary antibodies, namely, monoclonal antibody against MT1-MMP (114-6G6, 1:25; Fuji Chemical Industries, Japan), monoclonal antibody against MMP-2 (42-5D11, 1:100; Fuji Chemical Industries) or monoclonal antibody against TIMP-2 (67-4H11, 1:100; Fuji Chemical Industries) for 16 h at 4°C. After incubation, immunohistochemical staining was performed by the standard Immunohistochemical detection of membrane-type-1matrix metalloproteinase in colorectal carcinoma avidin-biotin-peroxidase complex (ABC) technique with an LSAB kit (Nichirei, Tokyo, Japan) and 3,3′-diaminobenzidine as the chromogen. Nuclei were counterstained with haematoxylin. For evaluation of immunohistochemical staining, specimens were divided into two groups as follows. The immunopositive cell area was used for the evaluation of the immunohistochemical staining of MT1-MMP, MMP-2 and TIMP-2 monoclonal antibody: negative, 0-10%; positive, > 10%. Expression of MMP-2 and TIMP-2 was also evaluated in terms of immunostaining of the tumour cell cytoplasm and the stroma. A clinicopathologic study was performed by reference to the depth of invasion, lymphatic invasion, venous invasion, lymph node metastasis, and distant metastasis.

Statistical analysis
Correlations between the expression of each antigen and the various clinicopathologic factors were examined by the χ-squared test, Fisher's exact probability test and Mann-Whitney's U-test. Furthermore, correlations between the expression of pairs of antigens were studied by the χ-squared test.

Immunohistochemical staining of MT1-MMP, MMP-2 and TIMP-2
Immunohistochemical staining indicated that MT1-MMP was localized predominantly in the tumour cell cytoplasm and it was weakly or not expressed in normal tissue ( Figure 1A). The frequency of samples positive for MT1-MMP was 36% (33/92). MMP-2 ( Figure 1B) and TIMP-2 ( Figure 1C) were localized by immunostaining in the tumour cell cytoplasm and the stroma. The frequency of samples positive for MMP-2 in the tumour cell cytoplasm was 20% (18/92) and in the stroma it was 30% (27/92). Thirteen samples (14%) in 92 were positive for MMP-2 in both the tumour cell cytoplasm and the stroma. The frequency for samples positive for TIMP-2 in the tumour cell cytoplasm was 32% (29/92) and in the stroma it was 47% (43/92). Eighteen samples (20%) in 92 were positive for TIMP-2 in both the tumour cell cytoplasm and the stroma. Table 1 shows the correlations between the expression of each antigen and the various clinicopathologic factors. The frequency of immunodetection of MT1-MMP increased with increases in the depth of invasion (P = 0.007; Mann-Whitney U-Test). The percentage of MT1-MMP-positive cases was significantly higher in vascular invasion-positive cases (54%) than in invasionnegative cases (29%; P = 0.02; χ-squared test). The percentage of cases positive for MMP-2 in the stroma was significantly higher in vascular invasion-positive cases (42%) than in invasion-negative cases (20%; P = 0.02; χ-squared test). The frequency of detection of TIMP-2 in the tumour cell cytoplasm increased with increases in the depth of invasion (P = 0.03; Mann-Whitney U-test). TIMP-2 was detected in the stroma more frequently in tumours with lymphatic invasion and lymph node metastasis (P < 0.05; χ-squared test) than in tumours without such features.

DISCUSSION
MMP-2, which is a type IV collagen-degrading enzyme, is a very important factor in the infiltration and metastasis of several carcinomas. TIMPs are intrinsic inhibitors of MMPs and have been studied clinically as potential carcinostatic agents (Watson et al, 1995;An et al, 1997). Theoretically, MMP-2 and TIMP-2 should be immunolocalized only in fibroblasts and monocytes at the sites that produce MMP-2 and TIMP-2 (Poulsom et al, 1992;Liabakk et al, 1996;Pyke et al, 1993;Ito et al, 1995). However, they have also been immunolocalized in the cytoplasm and cell membranes of cancer cells (Tomita and Iwata, 1996;Nomura et al, 1996;Höyhtyä et al. 1994). In our study, we found that MMP-2 and TIMP-2 were immunostained not only in the stroma of tumours but also in the cytoplasm of cancer cells, as well as there being elevated frequencies of expression of MT1-MMP, MMP-2, and TIMP-2 in the cytoplasm of cancer cells. One reason for this is that TIMP-2 and proMMP-2 might have been anchored to the cell membrane by MT1-MMP Imai et al, 1996).
Expression of MMP-2 was not strongly correlated with factors related to infiltration apart from vascular invasion in our study. One explanation of our results is that the immunostaining with the MMP-2-specific antibody used in this study did not reflect the activity of MMP-2 since the antibody recognized both MMP-2 and proMMP-2 (Fujimoto et al, 1993). Furthermore, Liabakk et al (1996) reported that less-advanced tumours at Dukes' stage A have higher levels of active MMP-2 than do more invasive tumours at Dukes' stage B. These results raise the possibility that MMP-2 might be necessary while the tumour is in the process of penetrating the bowel wall in tumours at Dukes' stage A but might be less important when tumours have spread beyond muscle into the surrounding adipose tissue, as at stage B. Consequently, the immunohistochemical detection of MMP-2 appears not to be an appropriate indicator of infiltration and metastasis in a clinical setting.
The expression of TIMP-2 has been reported to be closely correlated with the progression of human colorectal cancer, a proposal that conflicts with the original function of TIMP-2 as an inhibitor of MMPs (Murashige et al, 1996;Tomita and Iwata, 1996). In our study, expression of TIMP-2 in the tumour cell cytoplasm was correlated with depth of invasion and expression of TIMP-2 in the stroma was correlated with lymphatic invasion and lymph node metastasis. These results indicate that TIMP-2 is a reliable indicator of the progression of human colorectal carcinoma. These results are in conflict with the original proposed functions of inhibitors of MMPs for the following reasons. TIMPs counteract the proteolytic functions of MMPs in a stoichiometric manner, at a ratio of 1:1 (Stetler-Stevenson et al, 1993b). However, the presence of excess TIMPs, as compared to MMPs, in tumour tissue might indicate that an abnormal ratio in tumour tissue was associated with growth and metastasis, and the binding and biological actions of MMPs and TIMPs might be altered in the presence of excess TIMPs (Tomita and Iwata, 1996;Kossakowska et al, 1996). In addition, the TIMP-2-specific antibody used in our study also recognized the proMMP-2/TIMP-2 complex . TIMP-2 in this complex functions as an activator of proMMP-2. Accordingly, under difficult circumstances, such as when quantifying TIMP-2 and MMP-2, immunohistochemical detection of TIMP-2 might indicate that TIMP-2 is essential for the activation of proMMP-2.
Recently, MT1-MMP was identified in the cell membranes of transfected cells that expressed MT-MMP and the expression of MT1-MMP induced the activation of the precursor to MMP-2, proMMP-2 (Sato et al, 1994). There were few previous reports of immunohistochemical detection of MT1-MMP in colorectal carcinoma. In tumour cells of invasive lung carcinomas, MT1-MMP was immunolocalized in the carcinoma cells but not in the parenchymal or stromal cells of the surrounding normal tissue (Sato et al, 1994). In gastric carcinoma, MT1-MMP was predominantly immunolocalized in the carcinoma cells, and some carcinoma cells showed immunostaining on the cell membrane (Nomura et al, 1995). We found that MT1-MMP was immunolocalized predominantly in colorectal carcinoma cells, while its levels were low or it was absent in normal tissue. Moreover, expression of MT1-MMP was correlated with vascular invasion in cases of gastric carcinoma (Nomura et al, 1995) and with lymph node metastases in cases of lung carcinoma (Tokuraku et al, 1995) in previous studies. The present study showed that expression of MT1-MMP was correlated with vascular invasion and the depth of invasion. These results suggest that the expression of MT1-MMP might reflect infiltration in colorectal carcinoma. Thus, immunohistochemical detection of MT1-MMP and TIMP-2 provides appropriate indices of infiltration but expression is not correlated with distant metastases in colorectal carcinoma.
Our results indicate that not only MMPs but also other factors are required for the development of distant metastases.