Matrix metalloproteinase 2 (MMP-2) is a member of the MMPs family, a class of enzymes with proteolytic activity, secreted in a latent form and activated at the cellular surface by the complex membrane type 1-MMP (MT1-MMP) and tissue inhibitor of MMPs (TIMP-2).1, 2 When activated, MMP-2 degrades the extracellular matrix (ECM) components of the basement membrane (BM) removing tissue boundaries and facilitating epithelial cell motility.3–6 Its activity, however, is inhibited by TIMP-2 so that a too extensive degradation is avoided.7, 8 For these reasons, the role of TIMP-2 is intriguing, not only in view of its involvement in the MMP-2 activation process, but also because of its inhibitory function of MMP-2 proteolytic activity. In the human body, the expression of MMP-2 and TIMP-2 overlaps in the various different tissues and proteolysis occurs as a result of an imbalance between MMP-2 and TIMP-2 activity.9, 10
MMP-2 has been implicated in a number of situations such as development, morphogenesis and tumor progression, where remodeling of the tissue architecture and alteration of ECM turn-over occur.11–14 Several studies have reported that MMP-2 is up-regulated in the course of different malignancies and is involved in tumor metastasis.15, 16 Other studies have described a downregulation of TIMP-2 in cancer tissues together with essentially normal levels of MMP-2.17 Metastatic dissemination, however, is facilitated by increased levels of MMP-2 or decreased levels of TIMP-2, either of which can shift the balance in favor of proteolysis and cause BM degradation.4, 18 This is the key point for the beginning of tumor spread and metastasis formation, which have a positive correlation with MMP-2 proteolytic activity.
The occurrence of metastasis is one of the most serious factors in the course of cancers such as hepatocellular carcinoma (HCC). HCC is the seventh most frequent malignancy in the world and its frequency is constantly increasing in the Mediterranean area because of the spread of HCV infection.19 Although many improvements have been made in terms of diagnosis and treatment, the long-term survival of HCC patients remains unsatisfactory. Tumor recurrence, as well as intra-hepatic and vessel metastasis, are the worst problems in patients with HCC, even in those undergoing surgical therapy, because these severely affect prognosis and survival.19, 20 Furthermore, no therapy is currently available to prevent or block HCC cell invasion, mainly due to the poor current knowledge of the mechanisms that regulate cancer cell motility and invasion. Recently, we have reported that human HCC cell lines with a motile and invasive phenotype produce and activate high levels of MMP-2, that are used to migrate on the ECM substrate and to invade through a reconstituted BM.21 A synthetic inhibitor of MMPs blocks migration and invasion in vitro and the same inhibitor showed a cytostatic therapeutic effect in an orthotopic metastatic HCC model in nude mice.22
In the Mediterranean area as well as in North American countries, HCC develops only in chronically injured livers, where altered turn-over and increased deposition of ECM proteins has been reported. In such an environment, upregulation of MMPs and TIMP-2 has been reported and it is possible they have a role in the rearrangement of the liver tissue architecture.23, 24 There is no BM in the liver, but HCC cancer cells grow surrounded by ECM proteins secreted as a consequence of cirrhosis and therefore proteolytic activity is required to allow HCC cells to penetrate and cross over such tissue boundaries.21 The goal of our study is to investigate the expression and the clinical role of MMP-2, TIMP-2 and MT1-MMP in patients with HCC.
MATERIAL AND METHODS
Patients and specimens
We have studied serum and tissue samples of 40 Italian patients with HCC (35 men and 5 women age 40–82 years) that underwent surgery or fine needle biopsy. All the patients were clinically monitored for a 2-year period in the Department of Internal Medicine, University of Bari Medical School. As reported in Table I, all patients were affected by underlying cirrhosis, which was HBV related in 10/40 (25%), HCV related in 25/40 (62.5%), due to a coinfection by HBV-HDV or HBV-HCV in 4/40 (10%), of unknown etiology in 1/40 (2.5%). As control tissue, we also studied 5 subjects with normal liver function who underwent surgery for accidental trauma. The presence of the hepatitis B surface antigen (HBsAg) and of anti-delta antibodies was detected by an enzyme monoclonal immunoassay (Auszyme, Abbott Laboratories, Abbott Park, IL); anti-HCV antibodies were determined by a third-generation enzyme-linked immunosorbent assay (Ortho Diagnostic Systems, Raritan, NY).
Table I. Clinical Characteristics of Patients with HCC
Number of patients
Age (yrs, mean value)
61.3 ± 10.9
66.9 ± 7.8
Multiple viral infection
Tumor size (cm, mean value)
5.4 ± 2.2
4.9 ± 4.3
HCC histological grading
Presence of metastasis
7 pts (25%)
25 pts (89.2%)
2 pts (7.1%)
HCC tissue specimens were collected from the primary and the metastatic nodules of HCC and from the peritumoral area. Tissue samples were fixed with 3.7% formaldehyde and processed for routine histology, whereas adjacent specimens were immediately snap-frozen in liquid nitrogen and processed for immunohistochemistry. All the serum samples were collected before patients underwent any kind of treatment. As control we tested the sera of 15 healthy volunteer subjects and of 20 patients with liver cirrhosis (LC). Our study was performed in accordance with the Helsinki Declaration and informed written consent was obtained from all patients before surgery or biopsy.
The tissues were included in Optimal Cutting Temperature 4583 (OCT) embedding compound (Miles Laboratories, Inc, Naperville, IL) and 5 μm thick sections were serially cut with a microtome (Microtom, HM 505E, Carl Zeiss Oberkochen, Germany), collected on appropriate glass slides (Sigma, St. Louis, MO) and processed for indirect alkaline phosphatase.
Briefly, sections were fixed in a cold chloroform/acetone mixture for 10 min, air dried, incubated with the following primary monoclonal antibodies: anti-TIMP-2 was purchased from Calbiochem (San Diego, CA) and anti-MT1-MMP from Chemicon (Temecula, CA). A polyclonal antibody anti-MMP-2 was purchased from Chemicon. All the primary antibodies were diluted in RPMI medium with 10% added fetal calf serum (FCS). After gentle washing, sections were then incubated with a proper secondary antibody (Dako, Denmark) for 30 min in a humidified chamber. Sections were washed and incubated with alkaline phosphatase-conjugated anti-secondary antibody immunoglobulins. Staining was developed with red fuchsin chromogen and abundantly washed for 20 min. Finally, sections were mounted with glycerol and examined with a Nikon Eclipse photomicroscope (Nikon Corp., Tokyo, Japan).
Immunohistochemistry staining was quantified by counting the number of positive cells per microscopic field. Ten randomly chosen microscopic fields were evaluated for each tissue specimen and the mean and standard deviation (SD) were calculated.
Detection of MMP-2 and TIMP-2 by ELISA
In all patients, serum samples were collected before surgery and stored at −20°C. Serum concentrations of MMP-2 and TIMP-2 were determined according to the manufacturers' instructions (Amersham Pharmacia Biotech, Biochrom Ltd, Cambridge UK). Briefly, both kits are based on a solid ELISA sandwich format. Serum samples were incubated in a microwell plate coated with anti-MMP-2 or anti-TIMP-2 antibody. After extensive washes, captured MMP-2 or TIMP-2 were detected by a peroxidase labeled Fab antibody. Reactions were stopped by adding acid solution and the optical density was read at 450 nm in a microtiter plate spectrophotometer. Serum concentrations were obtained by interpolation from a standard curve.
Student's t-test was used to determine the 99% confidence intervals (CI) for the MMP-2 and TIMP-2 serum levels and immunostaining quantification between HCC patients with and without metastasis. The correlation between serum and tissue levels of TIMP-2 in HCC patients with and without metastasis was studied with the Pearson correlation coefficient.
The tissue specimens were divided and processed for routine histology and for immunohistochemical studies. We investigated the pattern of MMP-2, MT1-MMP and TIMP-2 expression in each single patient and serial sections were prepared to compare more homogeneous and consistent microscopic fields. HCC primary and metastatic nodules, as well as peritumoral tissues were examined at the same time under the same experimental conditions. No staining was observed in sections incubated with the secondary antibody alone as negative control (data not shown).
MT1-MMP expression in HCC primary tumor, peritumoral and healthy tissue
MT1-MMP was localized on serial sections with a monoclonal antibody. In HCC primary nodules, no substantial differences were observed between patients with and without metastasis. In all cases, MT1-MMP was distributed mainly in the parenchyma, cancer cells exhibited positive staining around the cellular membrane, whereas almost no staining was observed in the cytoplasm. In the cirrhotic peritumoral tissue, MT1-MMP was concentrated mainly in the parenchyma rather than in the portal space. In healthy tissue MT1-MMP staining was very weak and distributed in the parenchyma (Fig. 1).
MMP-2 expression in HCC primary tumor, peritumoral and healthy tissue
MMP-2 was present in all the HCC samples from primary nodules, although some differences in terms of staining intensity were evident in different specimens. In primary nodule tissues, MMP-2 staining seems to be slightly more abundant in patients with metastasis than without although the difference is not statistically significant (Fig. 1). In these tissues, the distribution pattern was similar in both metastatic and non metastatic samples (Fig. 1). In particular, MMP-2 was strongly expressed by HCC malignant cells, diffuse throughout the tumor area, mostly distributed intracellularly, with a granular aspect, but also present in the extracellular environment. Positive staining was observed at cell-cell contact as well as around all the cells. Occasionally, staining was also present in the portal space, with a diffuse pattern. In peritumoral tissue, MMP-2 was constantly expressed in all patients. In all the peritumoral tissues, MMP-2 was strongly concentrated in the portal space but also in the parenchyma where it showed a pattern of distribution similar to that observed in the HCC tissue (not shown). In control tissue, MMP-2 was localized in the parenchyma, mainly intracellularly and was less abundant than in HCC patients (Fig. 1).
TIMP-2 expression in HCC primary tumor, peritumoral and healthy tissue
TIMP-2 was evaluated on serial sections using a monoclonal antibody. In HCC primary nodules, it was expressed in all the samples examined, although some differences in terms of staining intensity were observed in the tissues of different patients. TIMP-2 staining was stronger in the primary nodule tissues of patients without metastasis than in those with (Fig. 1). No differences in terms of localization, however, were evident among patients with HCC regardless of the presence of metastasis. TIMP-2 was present mostly in the parenchyma, in the cytoplasm of cancer cells, but also in the extracellular space, showing a codistribution with MMP-2 staining. In conclusion, the intensity of MMP-2 and TIMP-2 staining was inversely correlated and the proteolytic balance shifted in favor of ECM degradation. In peritumoral tissues, TIMP-2 was expressed in the portal space, as well as in the parenchyma (not shown). In healthy control tissue, it was present in the cytoplasm of hepatocytes distributed in the parenchyma mainly around the centrolobular vein (Fig. 1).
MT1-MMP MMP-2 and TIMP-2 imbalance in HCC patients with or without metastasis
MT1-MMP, MMP-2 and TIMP-2 staining was quantified in HCC primary nodules and in healthy tissues (control) by counting the number of positive cells per microscopic field. Ten randomly chosen microscopic fields were counted for each staining and the mean is reported in Figure 2. As shown in the figure, the number of MT1-MMP and MMP-2 positive cells in HCC tissue sections of patients with and without metastasis did not show any significant difference between the 2 groups of patients. Finally, we counted the cells positively stained with a monoclonal antibody against TIMP-2 and, as shown, a significant difference was here observed between sections of HCC patients with and without metastasis (p < 0.0001).
MT1-MMP, MMP-2 and TIMP-2 in intra-hepatic HCC metastatic tissue
MT1-MMP, MMP-2 and TIMP-2 expression were investigated in intra-hepatic HCC metastatic lesions, defined according to standard clinical and pathological criteria.25 In Figure 3, we show low magnification of staining of the cirrhotic peritumoral lesion infiltrated by HCC metastatic cells. MT1-MMP was expressed in the cirrhotic peritumoral tissue and had a similar pattern but stronger staining in the HCC metastatic nodules.
Although the difference in terms of MMP-2 staining was very slight in the primary nodules of HCC patients with and without metastasis, in the metastatic tissues it became more evident. MMP-2 was concentrated along the invasive edge mostly in contact with the fibrotic tissue surrounding the HCC cancer cells. The main difference, however, was observed with regards to TIMP-2, that strongly stained the cirrhotic peritumoral tissue, whereas it was weakly expressed in the HCC metastatic tissue in Figure 3. In conclusion, we observed an upregulation of MMP-2 and a downregulation of TIMP-2 in the HCC metastatic nodules.
MMP-2 and TIMP-2 levels in the serum of HCC patients
Serum concentrations of MMP-2 and TIMP-2 were measured by an ELISA kit in the same patients and the results are reported in Figure 4. Patients with HCC showed higher levels of MMP-2 but also of TIMP-2 compared to controls. In normal subjects, the MMP-2 concentration was 2,421 ± 379 ng/ml, in patients with LC it was 2,142 ± 160 ng/ml (p = 0.19) whereas it was more concentrated in HCC patients with metastasis 2,984.3 ± 729.1 ng/ml compared to patients without metastasis 2,514.2 ± 344.2 ng/ml, although this difference was not statistically significant (p = 0.08).
On the contrary, the TIMP-2 concentration was very low in normal subjects 6.6 ± 7 ng/ml, but more elevated in LC patients 173.6 ± 83 ng/ml. In patients with HCC it was more concentrated in the samples of patients without metastasis 171.2 ± 61 ng/ml compared to those with metastasis 72.4 ± 30.7 ng/ml and this difference was statistically significant (p < 0.0001).
In conclusion, the MMP-2 concentration was similar between healthy subjects and LC patients. In HCC patients the difference in MMP-2 levels between patients with and without metastasis was not statistically significant. On the contrary, TIMP-2 was upregulated in LC, as well as in HCC patients without metastasis, whereas it was markedly downregulated in HCC patients with metastasis.
Clinical outcome of HCC patients with and without metastasis
All the patients were followed over a 24-month period with periodical 3-monthly clinical examinations, routine blood tests and ultrasound tomography, while radiological examinations were performed every 6 months. In Table II we calculated the number of the HCC patients with and without metastasis showing high and low levels of TIMP-2 in both serum and tissue, based on the median value of the serum and tissue TIMP-2 levels respectively. As shown in Table II, the percentage of the metastatic patients with low levels of TIMP-2 in the primary nodule tissue was 100% and in the serum was 72%; whereas in the nonmetastatic patients the percentage of patients with high levels of TIMP-2 in the primary nodule tissue and in the serum was 100%. In Figure 5, we reported the percentage of survival in patients with and without metastasis. After 12 months, such a percentage was 90% in patients without and 65% in patients with metastasis. This difference was even more evident after 18 and 24 months, when 90% of the patients without metastasis were still alive versus 40% and less than 30% of the patients with metastasis, respectively.
Table II. Distribution of the HCC Patients Based on Serum and Tissue Level of TIMP-2
Patients with Mtx
Patients w/o Mtx
Finally, we investigated the correlation between serum and tissue levels of TIMP-2 in each HCC patient with or without metastasis; and as shown in Figure 6, we found that it was r = 0.47 (p < 0.01) and r = 0.38 (p < 0.2) respectively. In conclusion, these data suggest that low levels of TIMP-2 correlate with the occurrence of metastasis, affecting the longterm survival of patients with HCC. In addition, they also suggest that TIMP-2 detection might be used as a prognostic factor.
Proteolysis of ECM components has been widely documented to be an essential step in the tumor metastasis cascade. We investigated the expression of MT1-MMP, MMP-2 and TIMP-2 in patients with HCC and we found that an imbalance between MMP-2 and TIMP-2 in patients with HCC correlates with cancer spread and metastasis formation. We based this conclusion on the following observations: (i) in patients with metastasis, MMP-2 was mostly concentrated in the metastatic nodule, along the advancing edge; (ii) in the serum, TIMP-2 levels were more concentrated in patients without metastasis than in those with; (iii) the tissue levels of TIMP-2 were higher in the primary HCC nodule of non metastatic patients than in that of patients with metastasis; and 4) patients with metastasis showed a significant correlation between tissue and serum levels of TIMP-2.
Elevated levels of MMP-2 have been implicated in the course of several physiological and pathological conditions when rearrangement of the original tissue architecture occurs.11–13 In our study MMP-2 serum levels were increased in patients with HCC respect to patients with LC although the difference was not statistically significant. Furthermore, also the differences in terms of MMP-2 levels in the serum and in the primary nodule tissue between HCC patients with and without metastasis were very slight and not statistically significant. Despite this, there was a remarkable difference in terms of localization of MMP-2 in the metastatic tissue, where it was mostly concentrated along the advancing edge, suggesting involvement in the invasive and metastatic process as described for other malignancies.13, 26, 27 The slight differences in MMP-2 and MT1-MMP in the serum or in the primary nodule cancer tissues between HCC patients with and without metastasis seems to be in contrast with other studies reporting a significant increase of MT1-MMP or MMP-2 in patients with HCC and their positive correlation with the occurrence of metastasis.28–30 The latter studies, however, have been performed in Asian countries where HCC originates in healthy liver tissues, whereas in Mediterranean countries it commonly develops in cirrhotic livers31 where an over-expression of MT1-MMP and MMP-2 has been reported23, 24 and this could therefore mask differences in MMP-2 or MT1-MMP expression between patients with or without metastasis.
Other studies have reported that the imbalance between MMP-2 and TIMP-2 observed in several different malignancies is mainly caused by lower levels of TIMP-2 rather than higher levels of MMP-2.17 Furthermore, TIMP-2 has been found to be upregulated in the course of liver cirrhosis, where it could play a role in regulating the remodeling of ECM proteins.24 In agreement with these data, we detected increased levels of TIMP-2 in patients with liver cirrhosis and with HCC without metastasis, but significantly lower levels in patients with metastasis. Our findings are consistent with other studies reporting a downregulation of TIMP-2 in the most aggressive and metastatic tumors.32, 33 This is also supported by our staining data on tissues, showing decreased expression of TIMP-2 in HCC tissues, in particular at the invasive edge of HCC border where it infiltrates the surrounding cirrhotic parenchyma. It remains unclear what may determine a downregulation in patients with metastatic HCC and further studies are required to clarify this issue, although a possible explanation may have to do with the transforming growth factor (TGF) β1. TGF/β1 has been widely reported to be up-regulated in HCC and it also seems to be correlated with prognosis and survival.34 TIMP-2 production has been shown to be downregulated by TGF/β1 and therefore we hypothesize that it might decrease TIMP-2 expression in patients with HCC,35 thus facilitating invasion and metastasis of the cancer cells. Further studies are needed to confirm this hypothesis that if true, might have important clinical implications.
The peculiar architecture of the liver and the absence of a BM tissue boundary facilitate the spread of HCC cells in the surrounding parenchyma through sinusoidal ducts.36 In our study, however, the occurrence of metastasis in the course of HCC was not related to the degree of cellular differentiation nor to the size of the original tumor, but rather to the MMP-2/TIMP-2 ratio, suggesting a potential role of proteolytic activity in the spread of HCC cells. In this situation, TIMP-2 is likely to play a key role, because patients with lower levels had shorter long-term survival, probably because of metastasis occurrence. This is consistent with our recent in vitro data suggesting that HCC cell invasion through a reconstituted BM requires gelatinase activity because the MMP inhibitor, BB-94, blocks migration and invasion.21 This same inhibitor BB-94 has been successfully used in a nude mice model, where it blocks HCC growth and invasion.22 Furthermore, in colon carcinoma, a reduction of liver metastasis development has been induced by TIMP-2 carried into the hepatic tissue by an adenovirus.37
In conclusion, our results suggest that an imbalance between MMP-2 and TIMP-2 in HCC primary nodule tissues could play a role in the spread of HCC cells. This finding was quantified by measurement of MMP-2 and TIMP-2 concentrations in the serum of the same patients. Furthermore, the weak expression of TIMP-2 along the advancing edge of infiltrating metastatic HCC nodules suggests its involvement in the invasive process. More studies are needed to confirm the relevance of the proteolytic balance in patients with HCC, to devise target therapies that could prevent or reduce the metastatic ability of HCC cells by modulating their proteolytic activity. Finally, because a significant correlation was found between the serum and the tissue levels of TIMP-2 in metastatic patients with HCC we suggest that TIMP-2 may be used as a prognostic marker during follow-up of patients with HCC, to individuate early patients that could develop cancer metastasis.
This work was supported by the Ministry for University Technological and Scientific Research (MURST; grant to O.S.) and by the Ministry for Public Health (grant to G.G. and M.Q.).