Actual Invasive Potential of Human Hepatocellular Carcinoma Revealed by in Situ Gelatin Zymography¹

HCC³ is one of the worst prognostic cancers in Japan and develops mostly in chronic liver diseases caused by either hepatitis B or C virus. HCC is frequently encapsulated with ECMs (1, 2) and is surrounded by cirrhotic liver that also contains abundant ECMs. HCC invades into these barriers of ECMs and blood vessels, which eventually causes metastasis in distant organs as well as the liver itself (intrahepatic metastasis; Ref. 3). Intrahepatic metastasis has been observed during the clinical course of many HCCs and it makes the prognosis of the patients poor (3). The intrahepatic metastasis is mostly caused by the portal vein invasion of HCC, which eventually results in the spread of cancer cells into the liver. The portal vein consists of a layer of boundary that contains basement membrane collagens and laminin (4). The capsule invasion and portal vein invasion, the first and the key steps of the invasion and metastasis of HCCs, are initiated by ECM degradation and determine the prognosis of patients with HCC.

The matrix-degrading metalloproteinases have been considered to play an important role in cancer invasion and metastasis (4, 5). We et al. (6, 7) have reported previously a considerable role of both gelatinase B/MMP-9 and the TIMP-1 and -2 (8) in human HCCs, but the actual matrix-degrading activity in situ in HCCs has not been observed. In situ gelatin zymography has been newly developed to detect a tissue gelatinolytic activity at a cellular level. The purpose of the present study was to visualize the cellular localization of the gelatinolytic activity (rather than mRNA or protein expression of proteinases) and to investigate the invasive potential of human HCCs. Here we demonstrate for the first time the in situ gelatinolytic activity in human HCCs and the close association of the activity with HCC invasion.

A total of 30 cases that received a curative surgical resection of HCC were studied. Clinical backgrounds of the patients are shown in Table 1. Informed consent was obtained from each patient. Immediately after partial resection of the liver, a part of the tumor and surrounding noncancerous liver tissue were frozen without fixation in Tissue-Tek OCT compound (Miles, Inc., Elkhart, IN). Serial frozen sections (4 μm) were made for both in situ gelatin zymography and H&E staining. A piece of the tissue was routinely fixed in 10% neutral formalin and was embedded in paraffin for pathological diagnosis. From the other part of the tissue, cancer, including capsule, was separated from noncancerous liver, and these tissues were immediately frozen and stored at −80°C for SDS-gelatin gel zymogram.

Gelatin (denatured collagen)-coated films, developed recently by Fuji Film Co. (Tokyo, Japan), were generous gifts of Dr. Nemori (Fuji Film Co.). For in situ gelatin zymography, a 4-μm frozen section was placed onto the film, incubated at 37°C for 11–13 h in a humid chamber, and then stained with 1% amido black (Nacalai, Kyoto, Japan). Gelatinolysis was detected as the disappearance of amido black staining. To characterize the gelatinolytic proteinases, either 10 mm (pH 8.0) EDTA (metalloproteinase inhibitor; Nacalai), 1 mm phenanthroline (metalloproteinase inhibitor; Nacalai), 1 mm PMSF (serine proteinase inhibitor; Nacalai), or 10 mm N-ethyl maleimide (thiol proteinase inhibitor; Nacalai) were dissolved in saline (0.9% NaCl), and the gelatin films were immersed in each solution for 10 min. The films were thoroughly dried at room temperature and were used for additional experiments. In addition, the in situ gelatin zymography was performed with or without treatment of APMA (gelatinase activator; Nacalai) to confirm whether the gelatinolysis was from the active enzymes. APMA (1 mm) dissolved in saline was put on the sections, and the tissues were each incubated for 6–9 h, then the gelatinolysis was compared with the controls (without APMA treatment). H&E staining was done with the serial sections for the histological examination.

Gelatinolytic activity was also analyzed with the same HCC samples by SDS-gelatin gel zymogram (9). Liver samples were homogenized in 50 mm Tris-HCl (pH 7.4), 0.2 m NaCl, and 10 mm CaCl₂ and then centrifuged at 10,000 × g for 10 min. The protein of the supernatant (10 μg) was separated in a 7.4% SDS-polyacrylamide gel containing 0.1% gelatin under nonreducing condition with electrophoresis buffer [25 mm Tris (pH 8.3), 0.25 m glycine, and 0.1% SDS] at 4°C. The gel was then washed three times in 50 mm Tris-HCl (pH 7.4) and 2% Triton X-100 for 30 min with shaking at room temperature and rinsed three times in 50 mm Tris-HCl (pH 7.4) for 5 min. The gel was incubated in 50 mm Tris-HCl (pH 7.4), 0.2 m NaCl, 5 mm CaCl₂, 0.02% NaN₃, and 1% Triton X-100, at 37°C for 16 h; stained with 0.1% amido black in acetic acid:methanol:distilled water (1:3:6) for 1 h with shaking at room temperature; and then destained in the same mixture without amido black.

To characterize the gelatinolytic proteinases, either 10 mm EDTA, 1 mm phenanthroline, 1 mm PMSF, or 10 mm N-ethyl maleimide was added in the buffer and reacted during the gel incubation after electrophoresis.

To examine whether the gelatinolytic enzymes in the tissue exist in latent (inactive) or active form, samples were treated with or without 1 mm APMA at 37°C for 1 h before SDS-gelatin gel separation.

χ² test was used for statistical analysis.

In situ gelatin zymography revealed the gelatinolytic activity in cancer cells of HCCs, as well as in stromal cells of noncancerous tumor-bearing livers (Fig. 1,A). In HCC nodules, the gelatinolysis was observed heterogeneously. Because the quantitation of this activity was difficult, we compared the gelatinolytic intensity between HCC and the corresponding noncancerous liver. The gelatinolytic intensity was stronger in 15 HCC nodules than the corresponding noncancerous livers (Table 1). An intense gelatinolytic activity was detected in the cancer cells in the front of tumor invasion (Fig. 1,A; compare with H&E staining of the serial section shown in Fig. 1,B). Almost equal intensity was found in 10 cases, and weaker intensity was observed in 5 HCCs compared with the noncancerous livers. The relative intensity of gelatinolysis had no obvious association with either the size or the differentiation grades of HCCs. EDTA and phenanthroline, both metalloproteinases inhibitors, completely abolished the gelatinolytic activity (Fig. 1 C), but neither serine proteinase inhibitor (PMSF) nor thiol proteinase inhibitor (N-ethyl maleimide) did (data not shown), indicating that the gelatinolytic enzymes were metalloproteinases.

To confirm whether the positive signals (gelatinolysis) are from the active enzymes but not from their latent forms, the in situ gelatin zymography was performed with APMA treatment. On activation with APMA treatment, the gelatinolysis was observed as early as 8-h incubation with APMA, and the signals became very strong (Fig. 2,A), compared with the serial section without APMA treatment (Fig. 2,B). Without APMA (standard method), the gelatinolysis was usually not observed before 10 h, and it becomes recognizable after 11–13-h incubation. It is now clearly demonstrated that the APMA treatment converted the latent enzymes into active ones. This, together with the negative control (Fig. 1 C; experiments with the treatment of EDTA or Phenanthroline), proves that the positive signals (gelatinolysis) are from the active enzymes but not from their latent forms.

The presence or absence of invasion to the capsule of HCC and/or to the portal vein was histologically examined under a light microscope and then compared with the relative gelatinolytic intensity (Tables 1 and 2). The capsule invasion was observed in 12 cases in the group that showed greater gelatinolytic intensity in HCCs than the surrounding noncancerous basal livers. Only two cases with capsule invasion were found in the group that the gelatinolytic intensity was almost equal or less in HCC than the basal liver. The relative gelatinolytic activity was significantly associated with the capsule invasion of HCC (P < 0.0003). Portal vein invasion was observed in six cases in the group that had greater gelatinolytic intensity in HCC than the basal liver, but none was found in the other groups (P < 0.006).

In contrast to the in situ gelatin zymography, which reveals actual gelatinolytic enzyme activity at a cellular level, the SDS-gelatin gel zymogram detects the amounts of both active gelatinolytic enzymes and proenzymes (latent forms). This is because SDS in a gel electrophoresis causes activation of the latent enzymes without proteolytic cleavage of the inhibitory sequence (10). SDS-gelatin gel zymogram demonstrated three or four gelatinolytic bands, including gelatinase B (M_r 92,000 gelatinase/MMP9) and gelatinase A (M_r 72,000 gelatinase/MMP2; Fig. 3). These enzymes were present mostly in latent forms. There was no consistent difference in the amount of the enzymes between HCCs and corresponding noncancerous livers. Some HCC nodules had active gelatinase A, but the gelatinolytic intensity of the bands did not associate with the HCC invasion to the capsule or portal veins. The levels of gelatinolytic enzymes analyzed by the SDS-gelatin gel zymogram correlated neither with those found by the in situ gelatin zymography nor the invasive states of HCCs.

Consistent with the in situ gelatin zymography, EDTA and phenanthroline completely abrogated the gelatinolytic activity, but either serine or thiol proteinase inhibitors did not (Fig. 4).

The gelatin gel zymogram performed with APMA activation revealed that the gelatinolytic enzymes in both HCC and noncancerous liver tissue existed mostly in the latent (inactive) form and were activated by APMA (Fig. 5).

Using in situ gelatin zymography, we have succeeded in revealing the cellular localization of the gelatinolytic activity in liver tissues for the first time. HCC cells, as well as stromal cells of the noncancerous liver, possessed the gelatinolytic activity. A half of the HCCs studied had strong gelatinolytic activity that degrades ECMs in the surrounding fibrotic tissues and blood vessels. The actual gelatinolytic intensity of HCC was significantly and closely associated with cancer invasion to the capsule and also to the portal veins. It is now clearly demonstrated that HCC has an invasive potential with the gelatin (matrix)-degrading metalloproteinases.

Identification of the enzymes for the gelatinolytic activity was the next issue after the observation with the in situ gelatin zymography. Various molecules, including MMPs (4), serine proteinases (11), and thiol proteinases (12) have been reported to be involved in the process of cancer invasion and metastasis. The gelatinolytic activity observed in the present in situ gelatin zymography and SDS-gelatin gel zymogram was completely inhibited by metalloproteinase inhibitors but not by either serine or thiol proteinase inhibitors. It is therefore clearly identified that the enzymes are MMPs, and the majority are gelatinase B and gelatinase A. Indeed, we reported previously the overexpression of gelatinase B in HCCs and its correlation with capsular invasion (6). In addition, active gelatinase A has been suggested to play a role in tumor spread of human HCCs (13). Gelatinases degrade basement membrane type IV collagen and also other ECMs that are abundant in liver; those include types I, V, and VI collagens, fibronectin, laminin, elastin, proteoglycans, and entactin (14, 15). The present study proves that gelatinase B and gelatinase A are certainly important enzymes for the invasion of HCC.

The gelatinases in liver tissue exist mostly in the latent form, and the differences in the amounts of the enzymes were not consistent between HCCs and corresponding noncancerous livers. In contrast, a half of the HCC cases had stronger actual gelatinolytic activity than the tumor-bearing noncancerous liver, as was demonstrated by the present in situ gelatin zymography. The difference of the gelatinolytic activity between in situ gelatin zymography and SDS-gelatin gel zymogram seems to be intriguing. It suggests that other MMPs besides gelatinases A and B, such as MMP-7 (13), may be responsible for the degradation. The discrepancies also suggest an important role of modulators of MMPs, such as activators and inhibitors. The two HCC samples (4C and 7C in Fig. 3) showing the active form of gelatinase A in SDS-gelatin gel zymogram actually did not provide stronger signals than the others in the in situ gelatin zymography. The gelatinolytic intensity was similar to that of others. The reason why the results from in situ gelatin zymography did not correlate with those from SDS-gelatin zymogram may well be explained as follows: in situ gelatin zymography demonstrates the fine local activity of the active enzymes at a cellular level. It reveals the final activity of the localized enzymes that resulted from the in situ interaction with various activators and inhibitors. On the other hand, the SDS-gel zymogram reveals the kinds and amounts of the enzymes. It also shows whether the enzymes are in active or in latent forms. However, because the tissue must be homogenized for the SDS-gel zymogram, the enzymes in the tissue homogenate interact with a large pool of activators and inhibitors extracted from a huge number of cells of various types and, therefore, do not keep the original forms (latent, active, or inactive) any longer as they were in situ. Thereby, the in situ gelatin zymography can represent the enzyme activity at a cellular level, but the SDS-gel zymogram cannot. This is the fundamental difference between these two methods, indicating the importance and significance of the in situ gelatin zymography.

Activation of the latent MMPs appears to be the most important event in the HCC invasion. As was shown in the present in situ study, stronger gelatinolytic activity was found in many HCCs than the corresponding noncancerous livers, suggesting that HCC activates MMPs more than the noncancerous liver does. It has been found that MT-MMP activates progelatinase A on the invasive cancer cell surface (16). This may fit to the HCCs, because MT1-MMP immunostaining has been observed in both HCC cells and fibroblasts in the hepatic stroma (17). Progelatinase A could be activated by MT1-MMP on the surface of the HCC, and this may be a reason why the gelatinolytic activity is high in the cancer cell. Progelatinase B has been reported to be activated by gelatinase A (15), MMP-3 (18), plasmin (19), and tissue kallikrein (20). On the other hand, we have found in HCCs previously an overexpression of both TIMP-1 and -2 (8) that inhibit the activity of the gelatinases. Furthermore, transforming growth factor-β also regulates the expression of MMPs (21) and TIMPs (22). The regulation of these various molecules appears to play a key role in the activation of MMPs. In addition, the intense gelatinolytic activity at the invasion front of HCCs suggests interaction between HCC cells and host stromal cells (or stroma) for the activation of MMPs. This interaction may directly influence the growth and invasion of HCC.

In situ gelatin zymography is presently the most rational and powerful technique for addressing the question as to whether the matrix-degrading activity is up- or down-regulated at a cellular level. This provides meaningful information, particularly for the invasion and metastasis of tumors, tissue fibrogenesis (e.g., liver cirrhosis and lung fibrosis), wound healing, and remodeling of stroma. The in situ gelatin zymography may also be useful to test new drugs that inhibit gelatinolytic activity aiming for preventing cancer invasion and metastasis.

We thank Dr. K. Morii and Himeji Red Cross Hospital for providing tissues.