Membrane-type matrix metalloproteinase-1(MT1-MMP) gene is overexpressed in highly invasive hepatocellular carcinomas

Journal of Hepatology 1998; 28: 231–239 Printed in Denmark ¡ All rights reserved Munksgaard ¡ Copenhagen

Copyright C European Association for the Study of the Liver 1998

Journal of Hepatology ISSN 0168-8278

Membrane-type matrix metalloproteinase-1(MT1-MMP) gene is overexpressed in highly invasive hepatocellular carcinomas Tomika Harada1, Shigeki Arii1, Masahiro Mise1, Takashi Imamura1, Hiroaki Higashitsuji1, Masaharu Furutani1, Mototaka Niwano1, Shun-ichi Ishigami1, Manabu Fukumoto1,2, Motoharu Seiki3, Hiroshi Sato3 and Masayuki Imamura1 1

First Department of Surgery, 2First Department of Pathology, Faculty of Medicine, Kyoto University, Kyoto and 3Department of Molecular Virology and Oncology, Cancer Research Institute, Kanazawa University, Kanazawa, Japan

Background/Aims: The matrix metalloproteinase (MMP) family play important roles in the invasion of cancer cells by degrading the extracellular matrices. The current study was designed to determine the expression pattern of membrane-type matrix metalloproteinase-1 (MT1-MMP) in hepatocellular carcinomas and its participation in invasion potential. Methods: MT1-MMP mRNA expression was examined in 25 human hepatocellular carcinoma specimens using Northern blot, and the correlation to clinicopathological features was evaluated. In situ hybridization and immunohistochemistry were performed to study the localization and the cells responsible for the production. Results: Northern blot analysis revealed high levels of MT1-MMP mRNA expression in tumorous portions in all cases, whereas in non-tumorous portions moderate or faint expression was evident in 22/25 cases. In 21/25 cases, the expression levels in tumorous portion were higher than those in non-tumorous portion. In particular, hepatocellular carcinoma with capsule infiltration demonstrated significantly higher expression

than those without (p∞0.05). In situ hybridization and immunohistochemical study revealed MT1-MMP transcripts and proteins in cancer cells and stromal cells, respectively. MT1-MMP positive cells were preferentially observed in the invading border of tumor nests. The MMP-2 transcript showed a similar pattern to that of MT1-MMP by in situ hybridization. Conclusion: The present study showed that the MT1MMP gene is strongly expressed in hepatocellular carcinoma cells and is involved in the invasion potential of hepatocellular carcinoma, and also that MT1MMP may be one of the key molecules responsible for the invasion potential of hepatocellular carcinoma. Furthermore, the evidence suggests that MT1-MMP and MMP-2 cooperate in the process of cancer invasion.

 of the extracellular matrix (ECM) is thought to be a prerequisite for cancer invasion and metastasis which strongly affects the prognosis of cancer patients (1). Recent investigations have indicated that the the matrix metalloproteinase (MMP) family play important roles in the invasion of cancer cells by digesting the ECM (2), and tumor cells with

highly invasive characteristics have been shown to secrete large quantities of these proteolytic enzymes (3– 12). Tsuchiya et al. demonstrated that MMP–2 and MMP–9 mRNA expression were more positively correlated with liver metastasis, as evaluated in chick embryos, than either MMP-1 or MMP-3 mRNA expression (13). Hepatocellular carcinoma (HCC) is a common malignancy in oriental countries, but there have been no detailed studies of the molecular mechanisms of HCC invasion and metastasis. The invasive ability of HCC can be characterized by the infiltration of cancer cells into the fibrous tumor capsule. Moreover, approxi-

D

Received 6 May; revised 2 September; accepted 15 September 1997

Correspondence: Tomika Harada, First Department of Surgery, Faculty of Medicine, Kyoto University, 54 Shogoin Kawara-cho, Sakyo-ku, Kyoto 606-01, Japan. Tel: 81-75-751-3445. Fax: 81-75-751-3219.

Key words: Hepatocellular carcinoma; Immunohistochemistry; In situ hybridization; MT1-MMP; Northern blot.

231

T. Harada et al.

mately 90% of HCC cases are associated with chronic liver disease in which the ECM are markedly enriched (14). These observations led us to speculate that the MMP family may be involved in determining the invasion potential of HCC. Indeed, our recent study using Northern blot analysis revealed that the expression level of MMP-9 mRNA was significantly higher in invasive HCC than in non-invasive HCC, as evaluated by cancer infiltration into the capsule (15). However, no correlation between MMP-2 mRNA expression and invasion potential was observed in our study. Although no significant increase in MMP-2 mRNA expression was observed in invasive HCC, it is not possible to preclude a role of MMP-2 because MMP-2 is secreted in a latent form (6,8,16,17). More recently, a new subclass of metalloproteinases, membrane-type matrix metalloproteinase-1, 2, 3 (MT1,2,3-MMP), with trans-membrane domains in their C terminal ends, has been identified (18–21). Of these three MT-MMPs, MT1-MMP which was cloned from human placenta cDNA libraries was shown to activate proMMP-2 on the cell surface of lung carcinomas, although the mechanism of the switch from a latent form to an active one remains to be resolved (18). The present study was designed to determine the expression of MT1-MMP and the role in the invasion in HCC, using Northern blot, and to clarify what kinds of cells are responsible for the production of MT1MMP and their localization using both in situ hybridization and immunohistochemistry. Evidence will be presented indicating that MT1-MMP is one of the most potent factors in the invasion of HCC.

classification (22), histology in the non-tumorous portions of the liver and the survival period after hepatectomy. Invasion potential was evaluated by the microscopic infiltration of cancer cells into the tumor capsule and the macroscopic portal involvement (Table 1).

Materials and Methods

The PCR product of expected size was cloned into the EcoRV site of pBluescript SK(-) plasmid as described by Marchuk et al. (24) and the insert was confirmed by sequencing. The insert was purified and used as a probe for Northern blot analysis.

Surgical specimens Surgical specimens were obtained from 25 patients with hepatocellular carcinoma who underwent hepatic resection at the First Department of Surgery, Kyoto University Hospital, Kyoto, Japan from 1993 to 1995 (Table 1). Of these patients, 20 were men and five were women. Their ages ranged from 39 to 78 years (mean∫SD, 63.20∫9.15). All tissue samples were frozen at ª70æC until subjected to Northern blot analysis, or fixed with 10% neutral formalin for in situ hybridization and immunohistochemical analysis. Each patient gave informed consent before entering the study. Clinicopathological information Clinicopathological features were noted, including the number of tumors, tumor size, histological grade of malignancy according to Edmondson & Steiner’s

232

cDNA probes for Northern blot analysis The probes for human MT1-MMP and MMP-2 were prepared as follows; 5 mg of human liver total RNA was reverse-transcribed with random primers, using a commercial kit (First Strand Synthesis Kit, Pharmacia, Inc., Piscatamay, NJ, USA). The resulting cDNA mixture was subjected to 30 cycles (1 min at 94æC, 1 min at 55æC, 1 min at 72æC ) of polymerase chain reaction (PCR) amplification using a DNA thermal cycler (Astec, Japan), Taq DNA polymerase (Toyobo, Japan), and specific polymerase chain reaction (PCR) primers. PCR was carried out with the following primers, which were based on the cDNA sequence for human MT1MMP (18) and MMP-2 (23). MT1-MMP primers: sense 5ø-GGT CAT CTG CTC CTT TTC CA-3ø primer antisense 5ø-TGC CTT CTC TGC TCC TTT CT- 3ø primer MMP-2 primers: sense 5ø-TGT CTT CCC CTT CAC TTT CC-3ø primer antisense 5ø-ATA CCG CAT CAA TCT TTT CC-3ø primer

Northern Blot analysis Total RNA was isolated from tumorous and non-tumorous liver tissue by the acid guanidinium-thiocyanate/phenol/chloroform method. Poly (A)π RNA was selected by OligotexTM-dT30 (Takara, Japan). A total of 5 mg of poly (A)π RNA from each specimen was electrophoresed on 1.0% agarose/formaldehyde gel and transferred to a Hybond-Nπ nylon membrane filter (Amersham, Bucks, UK). The probes were labeled with [32P] dCTP using the Megaprime Random-Primer DNA Labeling Kit (Amersham Bucks, UK). Hybridization was performed at 65æC for 2 h in Rapid Hybridization Solution (Amersham) with labeled probe. The

MT1-MMP expression in hepatocellular carcinoma

membranes were rehybridized with S26 ribosomal protein (S26r) DNA probe as an internal control (15,25). In order to quantify mRNA expression, the level of mRNA expression was measured by densitometric scanning of the autoradiographs. The ratio of the MT1-MMP mRNA-band densitometric value to that of the S26r mRNA-band for each of the tumorous portions (T) and non-tumorous portions (N) was studied, and the ratio of the normalized mRNA levels in the tumorous portions to those in the non-tumorous portions was calculated (T/N ratio). The amounts of mRNA transferred to different filters were corrected by using the ratio of densitometric values of T98G (a glioblastoma cell line used as a positive control) to that of S26r or overlapped cases. As shown later in the results, a single band of 4.5 kb corresponding to the MT1-MMP transcript was observed, as previously described by Sato et al. (18). This band was distinct from that for MT2-MMP (3.5 kb) and that for MT3-MMP (12 kb) (19,20). The primer used for the RT-PCR was derived from the 3ø non-coding region specific to MT1-MMP. Furthermore, to confirm that the cDNA obtained with the method described above was that for MT1-MMP, we made sure that the nucleotide sequence of the cDNA for MT1MMP showed low homology with those for MT2MMP and MT3-MMP (data not shown). Accordingly, the MT1-MMP cDNA probe used in this study would not be expected to cross-react with either MT2-MMP or MT3-MMP. RNA probes for in situ hybridization Sense and antisense RNA probes for MT1-MMP and MMP-2, which were used for the in situ hybridization, were synthesized from the cDNA that were used in the Northern blot analysis. The plasmids were linearized with appropriate restriction enzymes, then transcribed and labeled with digoxigenin (DIG)-UTP using the DIG RNA labeling kit (Boehringer Mannheim Biochemica, Mannheim, FRG) following the manufacturer’s instructions. Transcribed products were recognized by gel-electrophoresis and the digoxigeninlabeled probes were quantified by direct colorimetric detection, using a Labeling Detection Kit (Boehringer Mannheim ) (data not shown ) (26). In situ hybridization In situ hybridization was carried out using specimens selected as described below by the previously reported method, with some modifications (27,28). Surgical specimens were fixed with 10% neutral formalin and embedded in 5 mm-thick paraffin sections. After de-

waxing, the sections were digested in Proteinase K solutions (Workstation BBS-2 R&D Systems Europe, UK) in Tris-HCl at a concentration of 100 mg/ml and incubated at 37æC for 15 min. Prehybridization was performed in a solution containing 50% formamide, 400 mM NaCl, 100 mg/ml of salmon sperm sonicated DNA (Workstation BBS-2), 250 mg/ml of yeast tRNA (Gibco BRL Products, Life Technologies Inc, USA) and 1*Denhardt’s solution (Amresco Inc., USA) at 37æC for 1 h. Then the sections were covered with the hybridization buffer (pre-hybridization buffer plus pre-denatured labeled anti-sense or sense probes at a concentration of 100 ng/ml) at 37æC for 16 h. The sections were incubated with sheep anti-digoxigenin alkaline phosphate conjugate antibody, diluted 1:600 in Tris buffered saline at room temperature for 30 min. After washing, alkaline phosphate activity was demonstrated using revealing reagents containing 5bromo-4-chloro-3-indolyl phosphate (BCIP enzyme substrate) and nitro-blue tetrazolium (NBT chromogen) at 37æC for 2 h (Digoxigenin Detection Kit BBS–8, R&D Systems Inc., USA). For the control study, sense and anti-sense RNA probes were run on sequential sections of each specimen to check for nonspecific hybridization. Additionally, RNase (100 mg/ml) treatment of tissue sections prior to hybridization with antisense probe further confirmed the absence of nonspecific signal. Negative control also included selective deletion of digoxigenin-labeled probe. In order to preserve the mRNA, all solutions were treated with diethylpyrocarbonate (DEPC) and autoclaved prior to use, and equipment used for this study was treated to remove nucleases, most notably RNase. Immunohistochemistry To immunolocalize MT1-MMP products, mouse monoclonal antibody against human MT1-MMP (113–5B7) (18) was applied to 4-mm-thick paraffin embedded sections in the same specimens as those of in situ hybridization. They were stained by the avidin-biotin procedure with Vectastain Elite ABC kit (Vector Laboratories, Burlingame, California, USA). The concentration of the anti-MT1-MMP antibody was 5 mg/ ml. Non-immunized mouse IgG was substituted for the primary antibody as a negative control. Analysis Data were expressed as mean∫SEM, and statistical analysis was carried out with the Wilcoxon-MannWhitney test. A p∞0.05 was considered to be statistically significant.

233

T. Harada et al.

Fig. 1. Northern blot analysis for MT1-MMP mRNA expression in tumorous and non-tumorous liver tissue from patients with HCC. Five micrograms of poly (A)π RNA from each human tissue and 10 mg of total RNA from T98G (a glioblastoma cell line as a positive control) and human placenta were used in Northern blot analysis. Hybridization was performed with human MT1-MMP cDNA as described in Materials and Methods. S26r mRNA is shown as the corresponding internal control in the lower panel. a: cases from 1 to 5; b: cases from 6 to 9. N: non-tumorous portion; T: tumorous portion; C: the tumorous portion adjacent to the capsular portion; T98G: a glioblastoma cell line; Pla: human placenta. T98G and the human placenta were used for positive controls. Bands of 4.5 kb bands corresponding to the MT1-MMP transcript were observed. MT1-MMP mRNA is highly expressed in the tumorous portions.

MMP mRNA in the tumorous portion adjacent to the capsule, that is at the invading border area of the tumor, was higher than that in the central portion of the tumor (case 4). Fig. 1b shows MT1-MMP mRNA expression in four other cases. MT1-MMP mRNA expression was detected in the tumorous portions in all cases. Conversely, non-tumorous tissue showed only a faint signal for MT1-MMP. The expression of MT1-MMP mRNA in the tumorous portions adjacent to the capsule was of almost the same degree as that in the tumorous portions. Table 1 summarizes the normalized MT1-MMP mRNA levels and the T/N ratio for each patient, as well as summarizing several clinicopathological features. In 21 of 25 cases the MT1-MMP mRNA levels in the tumorous portions were higher than those in the non-tumorous portions, and in 17 cases the T/N ratio was more than 2-fold higher.

Results MT1-MMP mRNA expression in tumorous and nontumorous liver tissue from patients with HCC Northern blot analysis was performed to examine MT1-MMP mRNA expression in both tumorous and non-tumorous liver portions from 25 HCC patients. Representative cases are shown in Fig. 1. Bands of 4.5 kb, corresponding to the MT1-MMP transcript, were observed. Fig. 1a shows MT1-MMP mRNA expression in liver tissue from five patients with HCC. All cases showed higher levels of MT1-MMP mRNA expression in the tumorous portions than in the corresponding non-tumorous portions. In four cases (cases 1, 3, 4 and 5), MT1-MMP mRNA expression in the tumorous portions was exclusively elevated. In contrast, non-tumorous portions showed very faint signals for MT1-MMP. The expression level of MT1-

234

Fig. 2. Correlation between the T/N ratio of MT1-MMP mRNA expression and the microscopic infiltration of cancer cells into the tumor capsule. T and N represent the ratio of the MT1-MMP mRNA densitometric value to those of S26r mRNA for the tumorous and non-tumorous portions, respectively. T/N ratio, patients with capsular infiltration: 6.01∫1.23, patients with absence of capsular infiltration: 0.85∫0.13 (p∞0.05, nΩ25).

65 67 66 71 39 66 60 67 55 66 65 66 68 78 51 66 70 65 52 76 58 74 61 45 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

m m m m m m m m m m m f f m m m m f m f m m m f m

Sex

25 4 35 18 40 47 30 47 39 48 50 52 13 5 43 4 43 41 37 15 4 1 7 20 50

dead dead dead dead alive alive dead dead dead dead alive alive alive dead alive dead alive alive alive alive dead dead dead dead alive

Survival period after hepatectomy (months)

2.6 10.2 4.5 8.5 12.0 4.5 3.0 6.1 16.0 3.1 5.9 4.0 5.5 8.5 5.1 5.0 6.1 5.5 1.8 2.0 4.5 7.0 6.0 12.0 3.8

1 1 2 3 1 1 1 1 1 1 1 1 1 1 1 1 2 3 3 1 10 10 2 2 1

π ª π π π π π π π π π ª π π π π ª π π π ª π π π ª

Tumor size Number of Capsular (cm) tumors infiltration

π π ª ª π ª ª π ª ª π ª ª π ª ª π ª π ª ª π π π ª

Portal involvement

3 3 2 3 3 2 3 2 2 3 3 2 2 3 2 3 2 2 2 2 1 2 3 2 2

CH CH LC CH CH LC LC LC CH LC CH LC CH CH CH LC CH LC LC LC LC LC LC LC LC

0.32 0.29 0.28 0.32 0.26 0.00 0.23 0.03 0.06 0.25 0.24 0.28 0.22 0.00 0.08 0.00 0.14 0.34 0.72 0.58 0.57 0.23 0.77 0.57 0.45

1.52 0.33 1.17 0.84 0.76 1.01 0.19 0.41 1.26 0.79 0.74 0.12 0.52 0.65 1.23 0.95 0.15 1.60 1.12 1.41 0.48 0.76 0.92 1.75 0.35

T

4.75 1.14 4.18 2.63 2.92 N.D. 0.83 13.67 21.00 3.16 3.08 0.43 2.36 N.D. 15.38 N.D. 1.07 4.71 1.56 2.43 0.84 3.30 1.19 3.07 0.78

T/N ratio

N

Grade of differentiationa (tumorous portion)

Underlying liver diseasea (non-tumorous portion)

MT-1 MMP mRNA expressionc

Histology

a Histologic grade of malignancy was determined according to Edmondson & Steiner’s classification (22). b LC, liver cirrhosis; CH, chronic hepatitis. c N and T represent the ratio of the MT1MMP mRNA densitometric value to that of S26r mRNA in the non-tumorous portion and the tumorous portion, respectively. The amounts of mRNA transferred to different filters were corrected by using the ratio of densitometric values for T98G mRNA (a glioblastoma cell line used as a positive control) to values for S26r mRNA or overlapped cases.

Age

Case no

TABLE 1 Clinicopathological features of each patient and normalized MT1-MMP mRNA levels

MT1-MMP expression in hepatocellular carcinoma

235

T. Harada et al.

Fig. 3. In situ hybridization study for MT1-MMP and MMP-2 in hepatocellular carcinoma tissues (Case 1). In situ hybridization was performed with anti-sense and sense probes for MT1-MMP and MMP-2 in hepatocellular carcinoma tissues. a: anti-sense probe for MT1-MMP (¿200); b: high magnification of a (¿400); c: anti-sense probe as a negative control of a (¿200); d: anti-sense probe for MMP-2 (¿200); e: high magnification of d (¿400); f: anti-sense probe as a negative control of d (¿200). N: non-tumorous portion; T: tumorous portion. MT1-MMP mRNA transcript was observed in both cancer cells (long arrows) and fibroblasts in the stroma (arrow heads). Note that the intensity of MT1-MMP mRNA expression was stronger at the invading border area of the tumor than in the center of the tumors, and MT1-MMP and MMP-2 show a similar pattern of mRNA expression.

Correlation between MT1-MMP expression and clinicopathological features Fig. 2 shows that the levels of expression of MT1-MMP mRNA in the tumorous portions of HCC patients with

236

capsular infiltration were significantly higher than those in patients without (T/N ratio, presence of capsular infiltration, 6.01∫1.23 vs absence of capsular infiltration, 0.85∫0.13, p∞0.05, nΩ25). There was no apparent cor-

MT1-MMP expression in hepatocellular carcinoma

Fig. 4. Immunolocalization of MT1-MMP in hepatocellular carcinoma tissues (Case 1). Immunostaining was performed as described in Materials and Methods. a: mouse monoclonal antibody to human MT1-MMP (5 mg/ml) (¿200); b: high magnification (¿400); c: non-immunized mouse IgG as a negative control (¿200). N: non-tumorous portion; T: tumorous portion. Positive staining was observed in both cancer cells (long arrows) and fibroblasts in the stroma (arrow heads).

relation between MT1-MMP mRNA expression and the presence of portal involvement, tumor size, histological grade according to Edmondson & Steiner’s classification and the survival period after operation. In situ hybridization In five cases exhibiting relatively high expression of MT1-MMP mRNA at different levels (cases 1, 5, 9, 18 and 19), in situ hybridization was performed to identify the cells expressing MT1-MMP mRNA and to determine the localization of the gene transcript. In all cases examined, MT1-MMP mRNA transcript was observed in both cancer cells and fibroblasts in the stroma. Fig. 3a and 3b show the representative case (case 1). The intensity of expression was stronger at the invading border of the tumor than at the central portion. Immunohistochemistry The result of immunohistochemical study using antiMT1-MMP antibody was compatible with that of in situ hybridization. The positive staining was observed in both cancer cells and fibroblasts in the stroma with a preferential distribution in the marginal areas of the

tumor nests. The finding in case 1 is demonstrated in Fig. 4a and 4b as representative. Localization of MT1-MMP and MMP-2 transcripts We also examined the localization of MMP-2 mRNA expression that is thought to be activated by MT1MMP from a latent to an active form in the same cases (Fig. 3d and 3e) by in situ hybridization. The distribution of MMP-2 transcript showed a pattern similar to that of MT1-MMP.

Discussion Previous studies have demonstrated that MMP expression is involved in the invasion and metastatic behavior of carcinomas (2–12). It is commonly accepted that most MMP are secreted in latent forms and are activated by various factors, including several serine proteinases such as trypsin, plasmin, and elastases in addition to acid exposure (29). These serine proteinases can convert proMMP-1, proMMP-3, and proMMP-9 from inactive latent to active forms (30–32). However, proMMP-2 cannot be activated by serine proteinases and it has been suggested that proMMP-2 may be activated on the sur-

237

T. Harada et al.

face of the cancer cells by some membrane-associated activator (8,16,33). More recently, the molecule contributing to the activation of MMP-2 was identified by Sato et al. They designated this molecule as a new type of MMP, membrane-type MMP-1 (MT1-MMP), and it is thought to be the membrane-associated activator of MMP-2 (18). They showed that cells transfected with MT1-MMP cDNA acquired the ability to activate proMMP-2 on the cell membrane, and that cell membrane fractions isolated from MT1-MMP expressing cells activated purified proMMP–2 during co-incubation. Recent studies using Northern blot analysis in our laboratory have indicated that the invasion potential of HCC is closely correlated with the expression of MMP9 mRNA but not with that of MMP-2 mRNA (15). However, this observation does not necessarily exclude the possibility of MMP-2 involvement, because Northern blot analysis cannot discriminate a latent form from an active one. Furthermore, we obtained suggestive evidence that MMP-2 was preferentially observed at the invading border of the tumorous tissue. Based on these observations, the present study was designed to investigate the expression pattern of MT1-MMP in HCC and its role in the invasion potential of HCC. The present study showed that MT1-MMP mRNA was expressed in the tumorous portions of HCC in all cases, while expression in the non-tumorous portions was minimal. With respect to clinicopathological variables, the intensity of MT1-MMP mRNA expression in the tumor tissue was positively correlated with cancer infiltration into the tumor capsule, but was not correlated with portal involvement, patient age, patient sex, distribution, size of the tumor, the grade of cell differentiation, nor the survival period after hepatectomy. Yamamoto et al. reported that MT1-MMP mRNA expression correlated with capsular infiltration in HCC using Northern blot and gelatin zymography, which is compatible with the results of our study (34). Collectively, these observations suggest that MT1MMP is involved in the ability to invade, most notably the capsule infiltration of cancer cells. However, MT1MMP is not exclusively responsible for invasion potential, because even tumors without a high expression of MT1-MMP sometimes infiltrate the tumor capsules. Previously we found that MMP-9 also participated in the invasion of the tumor capsule (15). Indeed, tumors with capsular invasion which did not have high expression of MT1-MMP mRNA exhibited high expression of MMP-9 mRNA. Northern blot analysis thus was more reliable for quantitative evaluation than either in situ hybridization or immunohistochemistry, but could not clarify which

238

cells produced MT1-MMP. We next performed in situ hybridization and immunohistochemistry, although the specimens studied were limited. Okada et al. indicated that MT1-MMP transcript was observed only in the fibroblastic cells in several cancer tissues such as colon, breast, and head and neck carcinomas (35). In contrast Nomura et al. demonstrated the expression of MT1MMP in both cancer cells and stromal cells in stomach carcinomas (36). The in situ hybridization study presented here similarly demonstrated that MT1-MMP mRNA was present not only in cancer cells but also in fibroblasts in the stroma, which was also confirmed by the immunohistochemical study. Furthermore, in situ hybridization and immunohistochemistry revealed that MT1-MMP was preferentially observed at the invading border area rather than in the central portion of the tumors. Interestingly, the localization of MT1-MMP mRNA expression was very similar to that for MMP-2 mRNA. This observation strongly suggests mutual avidity between these molecules. Moreover, in situ hybridization and immunohistochemistry revealed that both MT1MMP and MMP-2 were expressed more intensely in the invading border than in the central portion of the tumors. Taken together, these results imply the close participation of MT1-MMP in the invasion ability of HCC through MMP-2. We have previously reported that using Northern blot, the expression level of TIMP-2 (Tissue inhibitor of metalloproteinase-2), the corresponding endogenous inhibitor of MMP-2, corresponded closely to that of MMP-2 in HCC tissue, which suggests that these two molecules form a complex (15). Further study will clarify how MT1-MMP, MMP-2 and TIMP-2 interact with one another, and lead to the activation of MMP-2. In conclusion, the present study has shown the high expression of the MT1-MMP gene in HCC cells and suggests the possible involvement of MT1-MMP in the invasion potential of HCC. The results also suggest that MT1-MMP and MMP-2 may cooperate in the process of cancer cell invasion, although the exact mechanism of MMP-2 activation by MT1-MMP remains to be resolved. This evidence of MT1-MMP involvement may provide a novel strategy against the invasion and metastasis of HCC.

References 1. Liotta LA, Steeg PS, Stetler-Stevenson WG. Cancer metastasis and angiogenesis: an imbalance of positive and negative regulation. Cell 1991; 64: 327–36. 2. Stetler-Stevenson WG, Aznavoorian S, Liotta LA. Tumor cell interactions with the extracellular matrix during invasion and metastasis. Annu Rev Cell Biol 1993; 9: 541–73. 3. Yamagata S, Yoshii Y, Suh JG, Tanaka R, Shimizu S. Occur-

MT1-MMP expression in hepatocellular carcinoma

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

rence of an active form of gelatinase in human gastric and colorectal carcinoma tissues. Cancer Lett 1991; 59: 51–5. Sato S, Kida Y, Mai M, Endo Y, Sasaki T, Seiki M, et al. Expression of genes encoding typeIV collagen-degrading metalloproteinases and tissue inhibitors of metalloproteinases in various human tumor cells. Oncogene 1992; 7: 77–83. Davies B, Miles DW, Happerfield LC, Naylor MS, Bobrow LG, Rubens RD, et al. Activity of type IV collagenases in benign and malignant breast disease. Br J Cancer 1993; 67: 1126–31. Brown PD, Bloxidge RE, Stuart NS, Gatter KC, Carmichael J. Association between expression of activated 72-kilodalton gelatinase and tumor spread in non-small cell lung carcinoma. J Natl Cancer Inst 1993; 85: 574–8. Rao JS, Steck PA, Mohanam S, Stetler-Stevenson WG, Liotta LA, Sawaya R. Elevated levels of M(r) 92000 type IV collagenase in human brain tumors. Cancer Res 1993; 53: 2208– 11. Azzam HS, Arand G, Lippman ME, Thompson EW. Association of MMP-2 activation potential with metastatic progression in human breast cancer cell lines independent of MMP-2 production. J Natl Cancer Inst 1993; 85: 1758–64. Davies B, Waxman J, Wasan H, Abel P, Williams G, Krausz T, et al. Levels of matrix metalloproteinases in bladder cancer correlate with tumor grade and invasion. Cancer Res 1993; 53: 5365–9. Nakajima M, Welch DR, Wynn DM, Tsuruo T, Nicolson GL. Serum and plasma M(r) 92000 progelatinase levels correlate with spontaneous metastasis of rat 13762NF mammary adenocarcinoma. Cancer Res 1993; 53: 5802–7. Hamdy FC, Fadlon EJ, Cottam D, Lawry J, Thurrel W, Silcocks PB, et al. Matrix metalloproteinase 9 expression in primary human prostatic adenocarcinoma and benign prostatic hyperplasia. Br J Cancer 1994; 69: 177–82. Naylor MS, Stamp GW, Davies BD, Balkwill FR. Expression and activity of MMPs and their regulators in ovarian cancer. Int J Cancer 1994; 58: 50–6. Tsuchiya Y, Endo Y, Sato H, Okada Y, Mai M, Seiki H, et al. Expression of type-IV collagenases in human tumor cell lines that can form liver colonies in chick embryos. Int J Cancer 1994; 56: 46–51. The Liver Cancer Study Group of Japan. Predictive factors for long-term prognosis after partial hepatectomy for patients with hepatocellular carcinoma in Japan. Cancer 1994; 74: 2772–80. Arii S, Mise M, Harada T, Furutani M, Ishigami S, Niwano M, et al. Overexpression of matrix metalloproteinase-9 gene in hepatocellular carcinoma with invasive potential. Hepatology 1996; 24: 316–22. Brown PD, Kleiner DE, Unsworth EJ, Stetler-Stevenson WG.Cellular activation of the 72 kDa typeIV procollagenase/ TIMP-2 complex. Kidney Int 1993; 43: 163–70. Brown PD, Bloxidge RE, Anderson E, Howell A. Expression of activated gelatinase in human invasive breast carcinoma. Clin Exp Metastasis 1993; 11: 183–9. Sato H, Takino T, Okada Y, Cao J, Shinagawa A, Seiki M, et al. A matrix metalloproteinase expressed on the surface of invasive tumor cells. Nature (Lond) 1994; 370: 61–5. Will H, Hinzmann B. cDNA sequence and mRNA tissue distribution of a novel human matrix metalloproteinase with a potential transmembrane segment. Eur J Biochem 1995; 231: 602–8.

20. Takino T, Sato H, Shinagawa, Seiki M. Identification of the second membrane-type matrix metalloproteinase (MT-MMP2) gene from a human placenta cDNA library. J Biol Chem 1995; 270: 23013–20. 21. Gordon Research Conference on matrix metalloproteinases. Andover, NH, USA. 16–21 July1, 1995. 22. Edmondson HA, Steiner PE. Primary carcinoma of the liver: a study of 100 cases among 48 900 necropsies. Cancer 1954; 7: 462–503. 23. Collier IE, Wilhelm SM, Eisen AZ, Marmer BL, Grant GA, Seltzer JL, et al. Ha-ras oncogene-transformed human bronchial epithelial cells (TBE-1) secrete a single metalloproteinase capable of degrading basement membrane collagen. J Biol Chem 1988; 263: 6579–87. 24. Marchuk D, Drumm M, Saulino A, Collins FS. Construction of T-vectors, a rapid and general system for direct cloning of unmodified PCR products. Nucleic Acids Res 1991; 19: 1154. 25. Vincent S, Marty L, Fort P. S26 ribosomal protein RNA: an invariant control for gene regulation experiments in eucaryotic cells and tissues. Nucleic Acids Res 1993; 21: 1498. 26. Holtke HJ, Kessler C. Non-radioactive labeling of RNA transcripts in vitro with hapten digoxigenin (DIG) hybridization and ELISA based detection. Nucleic Acids Res 1990; 18: 5843–51. 27. Koji T, Nakane PK. Localization in situ of specific mRNA using himine-tymine dimerized DNA probes: sensitive and reliable non-radioactive in situ hybridization. Acta Pathol 1990; 40: 793–807. 28. Shorrock K, Roberts P, Pringle JH, Lauder I. Demonstration of insulin and glucagon mRNA in routinely fixed and processed pancreatic tissue by in situ hybridization. J Pathol 1991; 165: 105–10. 29. Birkedal-Hansen H, Moore WGI, Bodden MK, Windsor LJ, Birkedal-Hansen B, DeCarlo A, et al. Matrix metalloproteinases: a review. Crit Rev Oral Biol Med 1993; 4: 197–250. 30. Suzuki K, Enghild J J, Morodomi T, Slavesen G, Nagase H. Mechanism of activation of tissue procollagenase by matrix metalloproteinase-3 (stromelysin). Biochemistry 1990; 29: 10261–70. 31. Okada Y, Harris ED Jr, Nagase H. The precursor of metalloendopeptidase from human rheumatoid synovial fibroblasts. Biochem J 1988; 254: 731–41. 32. Okada Y, Gonoji Y, Naka K, Tomita K, Nakanishi I, Iwata K, et al. Matrix metalloproteinase 9 from HT1080 human fibrosarcoma cells. J Biol Chem 1992; 267: 21712–9. 33. Okada Y, Morodomi T, Enghild J J, Suzuki K, Yasui K, Nakanishi I, et al. Matrix metalloproteinase 2 from human rheumatoid synovial fibroblasts. Eur J Biochem 1990; 194: 721–30. 34. Yamamoto H, Itoh F, Adachi Y, Sakamoto H, Adachi M, Hinoda Y, et al. Relation of enhanced secretion of active matrix metalloproteinases with tumor spread in human hepatocellular carcinoma. Gastroenterology 1997; 112: 1290–6. 35. Okada A, Bellocq J P, Rouyer N, Chenard MP, Rio MC, Chambon P, et al. Membrane-type matrix metalloproteinase (MT-MMP ) gene is expressed in stromal cells of human colon, breast, and head and neck carcinomas. Proc Natl Acad Sci USA 1995; 92: 2730–4. 36. Nomura H, Sato H, Seiki M, Mai M, Okada Y. Expression of membrane-type matrix metalloproteinase in human gastric carcinomas. Cancer Res 1995; 55: 3263–6.

239