The MSC cell line was obtained from ScienCell Research Laboratories (Carlsbad, CA, USA), which was isolated from human bone marrow, and characterized by immunofluorescence with CD44 and CD90 antibodies, and lipid staining after differentiation. The fifth passage MSCs did not express the surface marker, CD34; they expressed low levels of fetal liver kinase-1 (Flk-1) and higher levels of CD29 and CD105. Quantitative RT-PCR showed that the MSCs expressed octamer-binding transcription factor-4 (OCT-4) and Flk-1. They were cultured in Alpha Minimum Essential Medium (α-MEM, Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco) and 100 U/ml penicillin/streptomycin solution. The fifth to eighth passage cells were used in the following experiments.

MHCC97-H is human HCC cell line with a higher metastatic potential (18,19). These cells were cultured in high glucose Dulbecco’s modified Eagle’s medium (H-DMEM, Gibco), supplemented with 10% FBS at 37°C in a humidified incubator containing 5% CO₂.

MHCC97-H cells were labeled with green fluorescence protein (GFP) by transfection with the plasma vector, pEGFP-N1 (Clontech). Lipofectamine 2000 which mediated the highest transfection rates was used as a transfection agent. After 2 weeks of selection with G418 (800 μg/ml), the individual G418-resistant clones were picked up and subcultured. Finally, a stable eukaryotic cell line with the highest fraction of EGFP expression was obtained (GFP-MHCC97-H).

MSCs were labeled with red fluorescence protein (RFP) and GFP by transfection with the plasma vector, pERFP-N1 (Clontech), and after 2 weeks of selection with G418 (400 μg/ml), stable RFP-MSCs and GFP-MSCs were then acquired, respectively.

MSCs were labeled with 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories) according to the manufacturer’s instructions.

MSCs were also labeled with 5-bromodeoxyuridine (BrdU, Sigma) according to the manufacturer’s intuductions. For BrdU labeling, MSCs were cultured for 15 min in the presence of 2 mM BrdU, washed and expanded for 2 additional passages before being injected into the mice. This procedure labeled 70–80% of the MSCs.

To evaluate the in vitro cell fusion between MSCs and HCC cells, the GFP-MHCC97-H cells were co-cultured with DAPI-MSCs at a ratio of 5:1 in a 6-well plate. After 4 days, the culture was observed under a fluorescence microscope. Similarly, GFP-MHCC97-H cells and RFP-MSCs were also co-cultured at a ratio of 5:1 and observed under a microscope.

An Axioplan Epifluorescence microscope (Carl Zeiss, Oberkochen, Germany) was used and images were obtained using a DC300 digital video camera (Leica). Optical images were acquired using a DMR microscope connected to a DC300 video camera (Leica).

RFP-MSCs and GFP-MSCs were co-cultured at a ratio of 1:1 in a 6-well plate. The culture was observed under a fluorescence microscope to evaluate the in vitro cell fusion of the MSCs.

MSCs (2×10⁵) were plated and cultured in 6-well plate. When the cells had reached 60% confluency, they were fixed with 100% methanol, permeabilized with 0.5% Triton X-100, and sequentially incubated with primary anti-matrix metalloproteinase (MMP)2 or primary anti-MMP9 monoclonal antibodies and anti-mouse immunoglobulin (Ig) coupled with horseradish peroxidase (HRP). The cells were then stained with 3,3′-diaminobenzidine (DAB) and counterstained with hematoxylin.

In vitro invasion assay was performed as follows: briefly, 80 μl of serum-free α-MEM-diluted Matrigel (0.8 mg/ml) was added to the Transwell filters (8.0μm pore size) of a Boyden chamber (Costar, MA, USA) and incubated at 37°C for 2 h to form matrix gel. MSCs (1×10⁶) were cultured in FBS-free α-MEM for 24 h, and subsequently the cells were collected and counted. Cells (2×10⁵) were re-suspended with α-MEM and seeded in the upper well of Transwell chamber, a mixture of 600 μl of α-MEM with 10% fetal calf serum (FCS) was added to the lower chamber, serving as the chemoattractant. After incubation at 37°C for 48 h, the cells that had invaded across the Matrigel and passed through the Transwell filter were stained and observed under a light microscope.

Equal amounts of protein from the MSCs, MHCC97-H and MHCC97-L cells (HCC cells with lower metastasis) were mixed with SDS buffer and incubated for 20 min at 37°C. After incubation, samples (30 μg/lane) were added onto a 4.5% (w/v) stacking polyacrylamide gel and separated on a 7.5% (w/v) polyacrylamide gel containing 1 mg/ml gelatin for the detection of MMP2 and MMP9 activities. After electrophoresis, the gels were soaked in 2.5% Triton X-100 for 1 h to remove SDS and incubated for 16 h at 37°C in 50 mM Tris-HCl (pH 7.6) containing 150 mM NaCl, 10 mM CaCl₂ and 0.02% NaN₃. Finally, the gels were stained for 1 h in 45% methanol/10% acetic acid containing 0.5% Coomassie brilliant blue G250. Proteolytic activity was detected as clear bands on a blue background of the Coomassie blue staining gel.

To detect the distribution of MSCs in vivo, we subcutaneously injected 6×10⁶ GFP- MHCC97-H cells into nude mice (n=4). When the nodular tumors were formulated, we injected 5×10⁵ of human DAPI-MSCs into the tail veins of the mice. Four days after the injection, the mice were sacrificed, the subcutaneous tumor tissues, livers and lungs were removed and embedded with 1:4 dilution of optimum cutting temperature (OCT) compound in phosphate-buffered saline (PBS). Fresh-frozen tumor sections (5-μm thick) were mounted on glass slides, and the distribution of DAPI-MSCs was observed under a fluorescence microscope. These experiments were approved by the Shanghai Medical Experimental Animal Care Commission.

We also orthotopically implanted tissues of subcutaneous tumor into the livers of 10 nude mice, and 15 days after the implantation, 5 out of the 10 mice were intravenously injected with 5×10⁵ of human BrdU-MSCs 3 times per week. The other 5 mice were injected with PBS as the controls. After 20 days, the tumors, livers and lungs were removed and fixed in paraformaldehyde and embedded in paraffin wax. Paraffin sections (5-μm-thick) were mounted on glass slides.

The tumor and liver slides were deparaffinized and rehydrated over 10 min through a graded alcohol series to deionized water. Subsequently, 1% Antigen Unmasking Solution (Vector Laboratories) was added and the slides were microwaved to enhance antigen retrieval, followed by immunostaining with primary mouse anti-human antibody against BrdU (Sigma). Goat anti-mouse IgG-peroxidase (A9917, Sigma) or Cy3-Goat anti-mouse IgG were used as the secondary antibodies.

Lung samples were sliced into 20 sections of 5 μm thickness, and a 50-μm interval between 2 successive sections. After staining with hematoxylin and eosin (H&E), the sections were independently observed under a microscope by 2 pathologists to evaluate pulmonary metastasis.

The data were analyzed using SPSS 11.5 software (SPSS Corp., Chicago, IL). The Student’s t-test was used to analyze the differences in tumor weight. Fisher’s exact test was used for the comparison of the ratio involved. All statistical tests were two-sided and P<0.05 was considered to indicate a statistically significant difference.

After the co-culture of GFP-MHCC97-H with RFP-MSCs, and the co-culture of GFP-MHCC97-H with DAPI-MSCs, the culture formation mimicked the in vivo tumor structures. MSCs gradually circled and partitioned cancer cells (Fig. 1). Cell fusion was observed in the cultures on the 4th day, and binucleated or yellow fluorescent hybrid cells had formed. However, the cells fused at a very low frequency (approximately 4–5 GFP⁺ RFP⁺ cells and 4–5 binucleated cells were observed per microscopic sight; ×20), and the cell fusion did not obviously increase in the following days (Fig. 2A and B).

When GFP-MSCs and RFP-MSCs were co-cultured for 4 days, cell fusion was also observed in the culture; however, the cell fusion was also generated at a low frequency and did not increase in the following time-periods (Fig. 2C).

We found that MSCs expressed MMP2 at a low level (Fig. 3A), but highly expressed MMP9 (Fig. 3B), as shown by immunocytochemistry staining with MMP2 and MMP9 monoclonal antibodies. MSCs infiltrated through the Matrigel, as shown by in vitro Transwell assay (Fig. 3C). The zymographic pattern showed that the MSCs had a higher activity of MMP9 than the MHCC97-L cells (HCC cells with a lower metastatic potential) (Fig. 3D) (17,22).

Four days after injecting DAPI-MSCs into the mice, we found that fewer MSCs had engrafted the heart, spleen, kidneys and bone marrow, (Fig. 4A) while, more DAPI-MSCs were present in the tumor site compared with the normal liver tissues (Fig. 4B). More DAPI-MSCs were present in the site of pulmonary metastasis compared with the normal lung tissues (Fig. 4C).

Twenty days after the BrdU-MSC injection, we did not find any more binucleated cells in the tumor tissuses and the cells were mainly distributed in the tumor stroma. Moreover, we found that the components of the tumor stroma had increased compared to the tumors not injected with MSCs, as shown by immunostaining with Brdu (Fig. 5A–D).

After the in vivo injection of MSCs, we also found that the invasive capability of the HCC cells significantly decreased. The rate of pulmonary metastasis was 20% in the group injected with MSCs and 100% in the group not injected with MSCs (P=0.01). The total number of metastatic cells, as well as the total number of metastatic lesions in the lungs had decreased (Table I).