Colorectal carcinoma is one of the major causes of death from cancer in Western countries.1 The outcome of the disease is usually determined by the occurrence of metastatic dissemination, predominantly to the liver.2 Surgical resection, the only curative therapy for colorectal liver metastases, is possible in fewer than 25% of the patients at the time of diagnosis and the overall prognosis of non-resected colorectal liver metastases is poor.3 A detailed understanding of the biological mechanisms that regulate the establishment and development of colorectal liver metastases may lead to enhancement in non-surgical antitumour management.4
Hepatic stellate cells (HSCs) play a central role in the uptake and storage of vitamin A. During liver injury of any aetiology, HSCs undergo a response known as activation, which is the transition of quiescent cells into proliferative, fibrogenic and contractile myofibroblast-like cells.5 This is widely known to play an essential role in hepatic fibrogenesis, in that they differentiate into myofibroblasts and produce several cytokines (CKs), chemokines and matrix-degrading metalloproteinases as well as their specific inhibitors in their activated state.6,7 Alpha-smooth muscle actin [α-SMA (Abcam, Cambridge, UK)] is a marker of stellate cell activation;7 it increases during the stellate cell activation phase8 and may also serve to increase contractile potential of the cell.9 Moreover, activation of α-SMA-positive HSCs is also observed in the liver parenchyma adjacent to metastases.10 Glial fibrillary acidic protein (GFAP) is a glial-specific protein found in long processes of astrocytes11 and also seems to be expressed in quiescent HSCs (qHSCs). It disappears in normal cells after a week of culture and is not found in the activated phase of HSCs.12 Lipid droplets are the most conspicuous ultrastructural feature of fresh isolated HSCs in the intact liver13 and the loss of lipid droplets is a sign of HSC transdifferentiation into myofibroblast-like cells.14
Shimizu et al.15 showed that a highly metastatic colon carcinoma cell line (LM-H3) retained direct contact with the HSCs that surrounded the sinusoid from which the tumour masses received their nourishment. The authors also showed that peritumoral HSCs were more numerous than HSCs in the surrounding normal parenchyma, implying their local proliferation or migration.15 Furthermore, in a separate study by Illemann et al.,16 the authors showed that, in the desmoplastic growth pattern, the liver cells at the desmoplasia–liver parenchyma interface collapse and disappear as a result of the tumour expansion. In addition, the reticulin fibres, which are made up of various types of collagen, are left behind and generate a collagen-rich fibrous capsule. Furthermore, in the desmoplastic growth of colorectal liver metastases, all three components of the urokinase plasminogen activator (uPA) extracellular protease [urokinase plasminogen activator (uPAR), plasminogen activator inhibitor-1 (PAI-1) and uPA mRNA] showed intense expression whereas little or no upregulation was detected in the pushing growth pattern.16 In other words, liver metastases with desmoplastic growth break down the extracellular matrix and may be a crucial to the ability of metastatic cells to grow and invade the liver.16
In hepatocellular carcinoma and colorectal liver metastases, most investigations are concerned with the activated HSCs that are responsible for the formation of cancer stroma and fibrotic capsule.10,17 In addition, in biliary malignancy and hepatocellular carcinoma, activated stellate cells contribute to the accumulation of tumour stroma.18,19 These results confirm findings from animal studies suggesting that activated stellate cells play a role in hepatic metastases.20,21 In addition, cell culture studies have demonstrated paracrine activation of stellate cells by tumour cells.22,23
In this study, we have characterized qHSCs and compared the expression pattern of α-SMA, Ki-67 and CK18 cleavage products in qHSCs that have been cultured alone and in co-culture with metastatic and non-metastatic colon cancer cell lines and condition media. This has shown changes both in expression patterns of these markers and in rates of proliferation and apoptosis in stellate cells.
Materials and methods
Isolation and culture of quiescent hepatic stellate cells
Quiescent HSCs were isolated from the liver of a male Wistar rat, as reported in a previous study.13 The HSC isolation and purification was based on the enzymatic digestion of the rat liver by collagenase followed by the centrifugation of crude cell suspension through a density gradient containing 28.7% (w/v) Histodenz (Sigma-Aldrich, St. Louis, MO, USA).
The cell pellet was suspended and cultured in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal calf serum (FCS), 5 ml penicillin/streptomycin, 5 ml fungizone® (Life Technologies Limited, Paisley, UK) and 5 ml glutamine and then counted with a haemocytometer. Mean yields of 6 × 106 mean HSCs were obtained per animal. The purity of the HSC preparations obtained from the density centrifugation ranged from 75% to 80%, as shown by vitamin A autofluorescence and fat droplet staining. The viability of the cell preparations was > 80%, as estimated by trypan blue exclusion.
Vitamin A autofluorescence, staining of the lipids and immunofluorescent detection of glial fibrillary acidic protein
To identify freshly isolated rat HSCs, vitamin A autofluorescence was detected by seeding the freshly isolated HSCs at a density of 20 000 cells per ml of DMEM supplemented with 10% FCS, 5 ml penicillin/streptomycin, 5 ml fungisone and 5 ml glutamine in a 48-well plate. The cells were then incubated for 2 hours at 37°C in a 5% carbon dioxide (CO2) atmosphere and 100% humidity. Subsequently, the HSCs were examined in a dark room under a fluorescent microscope (Leica AF 6000 LX microscope system v. 2.1.1, Leica Microsystems GmbH, Wetzlar, Germany) at approximately 330 nm. Then, in order to detect lipid droplets, the HSCs were incubated for 3 hours in a 48-well plate in which the cells were fixed with 10% formol saline for 10 minutes and then washed in distilled water. The cells were incubated with Oil red O for 15 minutes at room temperature and then rinsed in distilled water. Lipid droplets were then visualized under the light microscope and appeared as red vesicles. Ultimately, the fresh isolated HSCs were also identified with the mouse monoclonal primary antibody against GFAP (Abcam, Cambridge, UK). Astrocyte primary cells were seeded in an additional two-chamber slide for GFAP expression control. One chamber was used as a positive control (treated by a primary antibody) and the other as a negative control. The slides were then visualized under the fluorescence microscope.
Immunofluorescent detection of alpha-smooth muscle actin
To determine the duration of qHSC transdifferentiation, α-SMA was sequentially stained on days 1, 3 and 5. The cells were then plated in a chamber slide with a cover at a density of 2 × 104 cells/ml for each chamber. This chamber slide was labelled with details of the α-SMA and the day of staining. The LX-2 cell line was seeded in a further two-chamber slide for α-SMA expression control. One chamber was used as a positive control (treated by primary antibody) and the other as a negative control. Stellate cells with controls were grown in a 10% FCS-containing medium and incubated then stained on days 1, 3 and 5. The cells were then washed in 1 × phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde in PBS for 15 minutes. After fixation, the cells were washed twice in ice-cold PBS and then permeabilized with PBS containing 0.25% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA) for 10 minutes. They were then washed three times in PBS (for 5 minutes each time) and incubated with blocking solution (1% bovine serum albumin in PBS with Tween-20) for 30 minutes. Mouse monoclonal primary antibodies for α-SMA (diluted to a 1:100 ratio in blocking solution) were incubated with the cells overnight at 4 C. After three washes in PBS, the cells were incubated with fluorescein-labelled anti-mouse immunoglobulin G (IgG) (Abcam, Cambridge, UK) (diluted to a 1 : 4000 ratio in blocking buffer) for 1 hour. The cells were washed three times in PBS and then stained with 4′,6-diamidino-2-phenylindole (DAPI) for 1 minute and then rinsed with PBS. The cells were then viewed through a fluorescence microscope. The lipid droplets staining of qHSCs using Oil red O was also used in a sequential manner to determine how long the cells remain quiescent throughout days 1, 3 and 5.
Tumour cells, primary astrocytes and the LX-2 cell line
Two types of tumour cells were used, a highly metastatic colon carcinoma cell line, HT-29, and a non-metastatic colorectal carcinoma, Colo-741. Primary astrocytes were provided by Professor Mimoun Azzouz of the Medical School at the University of Sheffield, UK. The LX-2 cell line, which was the activated phase of human HSCs, was provided by Professor Scott Friedman of the Mount Sinai School of Medicine in New York, NY, USA.
Hepatic stellate cell co-culture with colon cancer cell lines of HT-29 and/or Colo-741
A MitoTracker probe (Invitrogen, Paisley, UK) was used to label the tumour cell lines to distinguish them from the primary HSCs. In a comparative approach, α-SMA expression was investigated in the HSC co-culture with (1) a metastatic colorectal adenocarcinoma cell line, HT-29 and (2) a non-metastatic colon carcinoma cell line, Colo-741. Using immunocytochemistry, the α-SMA was measured on days 1, 2, 3, 5, 7 and 10 in a sequential manner. The labelled tumour cells were plated and incubated with the HSCs in a four-well chamber slide with cover 2 hours after the HSCs began the culture. The tumour cells were plated at a density of 1500 cells/ml and the primary HSCs at a density of 7000 cells/ml. Primary qHSCs were used as a control and the cells were viewed through a fluorescence microscope. Actin expressing cells were counted in five typical fields of view per well. Photographs were then imported to the Image J program, v. 1.45 (developed at the National Institutes of Health, Bethesda, MD, USA) and the cells expressing actin were counted. The cell numbers were expressed as a percentage per field of view and the results reported as a mean ± standard error of the mean (SEM) of triplicates.
Proliferation assessment of hepatic stellate cells co-cultured with colon cancer cell lines conditioned medium using Ki67 essay
Ki67 was used to clarify the proliferation effects of the conditioned medium of tumour cell lines on qHSCs on day 1 of the co-culture experiment. The HT-29 tumour cell line was used as a positive control because it was proliferating, while the qHSCs were used as a negative control because the qHSCs at day 1 were not proliferating.
Hepatic stellate cells were plated and incubated in a four-well chamber slide with cover (Nunc™, Fisher Scientific, Leicestershire, UK) at a density of 7000 cell/ml/well in DMEM supplemented with 10% FCS. The incubated stellate cells were co-cultured with HT-29 cell lines or Colo-741 cell lines conditioned medium after 3 hours of seeding overnight. By adapting immunocytochemistry, Ki67 was investigated on day 1 of the co-culture. The cells were washed in 1× PBS and then fixed in 1:1 methanol–acetone for 10 minutes. After fixation, the endogenous peroxidise activity was quenched by incubating the chamber slides with 3% H2O2 for 30 minutes. The peroxidise was washed off with distilled water for 5 minutes and the slides were incubated with diluted horse serum to a 2% final concentration for 20 minutes.
The blocking serum was drained off and the primary antibody monoclonal mouse anti-human (Dako Denmark A/S, Glostrup, Denmark) (diluted to a 1:100 ratio) was prepared in a tris-buffered saline with Tween-20 (TBST) solution and incubated for an hour at room temperature. After three washes with TBST for 5 minutes each, the cells were incubated for 30 minutes with diluted biotinylated secondary antibody (to a 1:200 ratio in 2% serum TBST). An avidin–biotin complex (ABC) reagent was prepared at the same time with the secondary antibody, as it has to stand for 30 minutes before use. The cells were washed twice in PBS (for 5 minutes each time) and then incubated with ABC reagent for 30 minutes at room temperature. Peroxidase substrate 3,3′-diaminobenzidine (DAB) solution was made up in 5 ml distilled water by adding two drops of buffer, four drops of DAB and two drops of peroxide, which were prepared immediately before use. The solution was incubated with the cells until the desired stain intensity had developed, usually within 10 minutes, and then rinsed in tap water. After that, the slides were counterstained in haematoxylin for 30 seconds and rinsed in tap water. Finally, the slides were dehydrated and cleared by immersion in 70%, 95% and 100% ethanol, for 3 minutes at each concentration, followed by xylene 1 for 3 minutes and then mounting xylene 3 for 3 minutes also. The slides were kept wet during the DePex mount under a fume hood by leaving them in xylene and then viewed using a light microscope.
Immunocytochemistry detection the cleavage of cytokeratin-18 in the co-cultured hepatic stellate cells
In order to determine CK18 cleavage product in HSCs co-culture with the (1) metastatic colorectal adenocarcinoma cell line tumour, HT-29, and (2) non-metastatic colon carcinoma cell line tumour, Colo-741, for 3 days co-culture, M30 CytoDEATH™ monoclonal antibody (Roche Diagnostics Ltd., West Sussex, UK) was used to detect the apoptotic HSCs.
The HSCs were plated and incubated in an eight-well chamber slide with cover at a density of 7000 cells/ml and co-cultured with HT-29 or Colo-741 labelled with a MitoTraker red CMXRos at a density of 1500 cells/ml for 3 days. The HSC co-culture was divided into positive and negative treatment groups. The cells were then washed in 1 × PBS and fixed in ice-cold methanol at −20 C for 30 minutes. After fixation, the cells were washed in washing buffer (PBS containing 0.1% Tween-20) twice, for 5 minutes each time. The washing buffer was then removed and the primary antibodies (M30 CytoDEATH™) diluted to a 1:50 ratio in blocking solution and incubated with the cells for 60 minutes at room temperature. After two washes with washing buffer, the cells were incubated with fluorescein-labelled anti-mouse IgG (diluted to a 1:4000 ratio in blocking buffer) for 1 hour. The cells were then washed twice in washing buffer, stained with DAPI for 1 minute and rinsed with PBS. The cells were then viewed through a fluorescence microscope.
Cells expressing CK18 were counted in five typical fields of view per well. As an additional counting method, photographs were imported to the Image J program and the cells expressing CK18 were counted. The cell numbers were expressed as a percentage per field of view and the results reported as a mean ± SEM of triplicates.
Results were expressed as mean ± SEM and the data were analysed by one-way analysis of variance (ANOVA), followed by a Bonferroni post hoc test using SPSS v. 16.0 (SPSS Inc., Chicago, IL, USA). However, in the case of the CK18 cleavage product experiment, the data were analysed using Student's t-test. Differences were considered statistically significant at P < 0.05 and each experiment was performed three times.
Hepatic stellate cell identification with vitamin A autofluorescence staining of lipid and immunocytochemistry detection of glial fibrillary acidic protein
In the normal rat liver, the HSCs that had a star-like appearance were identified by vitamin A autofluorescence. They exhibited a striking, rapidly fading blue-green autofluorescence when excited with a light of approximately 330 nm (Figure 1A). Under the light microscope examination, 4Oil red O staining was shown to have multivesicular red bodies of fat droplets distributed in the periphery of the cytoplasm (Figure 1B). GFAP is an intermediate protein found in quiescent HSCs that disappears 2 days after isolation. It is a cell type marker for qHSCs, which may allow a distinction between HSCs and other fibroblastic liver cells (Figure 1C).
Sequential staining of Oil Red O and alpha-smooth muscle actin onto plastic culture of quiescent hepatic stellate cells
Hepatic stellate cells in the first 3 days of uncoated plastic culture remain quiescent (Figure 2A and B) and change to myofibroblast-like cells after day 3, as identified by α-SMA and Oil red-O staining. HSCs revealed a loss of lipid droplets after day 3 of culture. The cells progressively spread and flattened after day 3 and had the typical myofibroblast appearance of activated HSCs by day 5 (Figure 2C). Conversely, α-SMA was absent from the early plastic uncoated culture that was represented throughout days 1 and 3 of the sequential staining (Figure 2D and E); however, α-SMA expression was detected on day 5 (Figure 2F).
Hepatic stellate cells co-cultured with colon cancer cell lines of HT-29 and Colo-741
Hepatic stellate cell co-cultures with a tumour metastatic cell line, HT-29, and non-tumour metastatic cell line, Colo-741, exerted different influences on the qHSCs, as judged by the expression of α-SMA. For example, on day 1 of the HSC HT-29 co-culture, α-SMA was revealed and increased continuously throughout the duration of the HSCs co-culture (Figure 3 and Table 1). In contrast, the HSC Colo-741 co-cultures expressed α-SMA only on days 7 and 10, as identified by staining with primary antibody (Abcam, Cambridge, UK), diluted to a 1:100 ratio in blocking solution (Figure 4 and Table 1), and the α-SMA count was low when compared with control HSCs, maintained in single culture (Figure 5 and Table 1). α-SMA expression in HSCs control was expressed after day 3, which is concurrent with the result of the qHSCs in Figure 2F and the expression of α-SMA continued until day 10 (Figure 5).
Proliferation assessment of hepatic stellate cells co-cultured with colon cancer cell lines conditioned medium using Ki67
The HT-29 conditioned medium enhanced qHSC proliferation on day 1 of co-culture rather than express actin as in the cell–cell contact. However, qHSCs co-cultured with Colo-741 conditioned medium did not develop any changes (Figure 6).
Cytokeratin-18 cleavage staining in hepatic stellate cells co-cultured with colon cancer cell lines
The HSC apoptosis in culture was quantified in vitro by M30 CytoDEATH™ mouse monoclonal antibody staining. Apoptotic cells were detected and counted under fluorescence as demonstrated in (Figure 7 and Table 2). The Colo-741 non-metastatic cancer cell line showed a significant apoptotic effect (P < 0.05) on the qHSCs in co-culture. Compared with this, the metastatic cancer cell line, HT-29, showed a low apoptotic effect on the qHSCs in co-culture.
In this study, HSCs were isolated and identified using three different markers, vitamin A autofluorescence, Oil red O staining of lipid droplets and GFAP staining. These markers are significant in characterizing the fresh qHSCs. For example, GFAP expression is present in qHSCs and expression is high after isolation but reduces with the age of the culture.2425 Vitamin A autofluorescence and lipid droplets are lost in the active phase of HSCs and synthesize extracellular matrix instead.7,26,27 However, according to previous studies,12,2829 induction of α-SMA is a marker of stellate cell activation.
The loss of lipid droplets is considered by some authors a sign of HSC transdifferentiation into myofibroblast-like cells14,30 as well as α-SMA expression.7 Therefore, in order to determine the duration of qHSC transdifferention, the present in vitro study using serial immunocytochemistry staining showed that the qHSCs expressed α-SMA at day 5, whereas, sequential staining of lipid or no fat droplets was noted on day 3 and completely disappeared by day 5, suggesting that activation had started by day 3. This alteration of HSCs to the active phase occurs gradually as in α-SMA sequential staining, the expression on the first days (1 and 3) were absent, which has also been reported also by other studies.12,28,29
In the present study, α-SMA expression was compared in HSCs co-cultured with (1) a metastatic colorectal adenocarcinoma cell line tumour, HT-29 and (2) a non-metastatic colon carcinoma cell line tumour, Colo-741. The results have shown that following the co-culture of HSCs with HT-29, α-SMA expression occurred after 24 hours, which marked rapid activation and transdifferentiation to myofibroblast-like cells. However, this is not the case in the HSCs sequential staining of α-SMA and co-cultured experiment control. Furthermore, the percentage myofibroblast-like cells that expressed α-SMA increased progressively throughout the experiment with HT-29 (P < 0.05).
In the Colo-741 co-culture, α-SMA expression was detected at days 7 and 10 only. The non-metastatic cell line, Colo-741, had a significant apoptotic effect on the fresh isolated HSCs (P < 0.05). This result may explain why the α-SMA expression consistency increased in the HT-29 co-culture, whereas the Colo-741 co-culture expressed α-SMA at a low percentage on days 7 and 10 compared with the control. In previous studies, activation shown by α-SMA-positive HSCs was observed in liver parenchyma adjacent to metastases.10,15 The uPA mRNA-positive cells were identified by combined immunohistochemistry and in situ hybridization to be primarily α-SMA-positive myofibroblasts.31 In colon cancer liver metastases, uPA mRNA was also expressed by stromal α-SMA-positive myofibroblasts, at the periphery of the desmoplastic growth pattern metastases, which may indicate that they have a role in growth and invasion of the liver by tumour cells.16 Additionally, metastasized tumour cells alter the supporting mesenchymal tissue in which they grow. This phenomenon, known as the stromal reaction, includes the activation of fibroblasts or myofibroblast-like cells.32 The stromal reaction plays a role in colorectal carcinoma either at the primary or metastatic site. For example, stromal fibroblasts express activation markers such as α-SMA33 and fibroblast-activating protein-α;34 these markers are associated with tumour progression and, in addition, tumour-associated myofibroblasts stimulate tumour invasiveness.35 Moreover, instead of cell-to-cell contact co-culture, qHSCs co-cultured with HT-29 conditioned medium show proliferation by Ki-67 rather than α-SMA expression, whereas the qHSCs co-cultured with Colo-741 did not reveal any change.
The CK18 cleavage product liberates a neoepitope that is specifically recognized by the M30 CytoDEATH™ monoclonal antibody. Specific proteolytic cleavage of CK18 is an event that takes place before disruption of membrane asymmetry and induction of DNA strand breaks. Numerous studies confirm that the M30 CytoDEATH™ antibody detects only apoptotic but not viable or necrotic cells.36–38 The CK18 intermediate filament of epithelial cells are cleaved in vitro by caspase-6, 3 and 7.39,40 The cleavage of CK18 by caspase marks an early event in the apoptotic process.41 Colo-741 and HT-29 tumour cell lines differ in terms of the occurrence of apoptosis as detected by CK18 cleavage products. Colo-741 had a significant apoptotic effect on the fresh isolated HSCs (P < 0.05), which may explain the finding of the HT-29 co-culture that had increased the α-SMA expression consistently, whereas Colo-741 expressed α-SMA in low percentage at days 7 and 10 compared with the control. A malignancy may be required for successful establishment because the sequential accumulation of mutations may be necessary to block apoptosis. For example, dysregulation of apoptosis may be necessary to promote growth, prevent elimination by cytotoxic lymphocytes and allow survival.42 A novel mechanism to escape immune recognition by neoplastic cells is loss of Fas-receptor expression and the development of Fas ligand expression by the cancer cells. Expression of the Fas ligand results in apoptosis of Fas-receptor-expressing cytotoxic lymphocytes as they attempt to attack the neoplastic cell; loss of Fas receptors by the neoplastic cell ensures its survival despite recognition by the cytotoxic lymphocytes.43,44 These observations may have a relation with the HT-29 co-culture finding, as the desmoplastic growth pattern did not induce apoptosis despite the dense infiltration of lymphocytes at the tumour periphery.45 The mechanism underlying the proapoptotic Colo-741 co-culture may relate to Fas and Fas ligand as detected in the fully activated HSCs (on day 7), whereas HSCs in resting and transitional phase show a higher resistance to CD95-mediated apoptosis.46 However, in the current experiment, the HSC apoptosis was detected on day 4 and this observation may relate to Fas ligand-expressing tumour cells.47
In conclusion, the metastatic colon cancer cell line, HT-29, stimulates HSCs to transform into myofibroblast-like cells. Low α-SMA expression with the non-metastatic colon cancer cell line, Colo-741, co-culture may reflect the level of HSC apoptosis. This result may be linked with the non-metastatic behaviour of Colo-741. These data provide a critical link between metastatic and non-metastatic colon cancer cell lines and HSC activation, which ultimately needs further investigation to clarify the molecular mechanism by which the HSCs or tumour cells promote tumorigenesis.