The human colorectal cancer cell line SW48 was purchased from the American Type Culture Collection (Manassas, VA, USA). The SW48LM2 cell subline was established as described below. SW48 and SW48LM2 were maintained in Leibovitz’s L-15 (Sigma) supplemented with antibiotics and 10% fetal bovine serum (HyClone, UT, USA). Prior to being transferred to the mice, the cell lines were incubated in a 100% air (37°C) incubator and passaged on reaching 80% confluence. To culture SW48 and SW48LM2 cells for 7 days under hypoxia, the AnaeroPack Anaero cultivation system with AnaeroPack MicroAero (Mitsubishi Gas Chemical, Japan) was used, providing 8–9% O₂ conditions.

The in vivo experiments were performed in accordance with institutional guidelines and were approved by the Animal Experimentation Committee of the Tokai University, the Keio University and the Central Institute for Experimental Animals (CIEA). NOG mice were bred and used at the age of 7–9 weeks. Liver metastases were experimentally induced by intrasplenic transfer of human cancer cells as described in a previous study (4). NOG mice were euthanized 6 weeks after the intrasplenic transfer of 1×10⁵ SW48 cells, which metastasize poorly to the liver (5). The livers were removed from the mice and macroscopically observed. Hepatic metastatic foci with diameters >2 mm were collected and cut into 1-mm³ cubes. Foci were dispersed into single cells in a solution of trypsin-EDTA (IBL Co., Ltd., Japan). Contamination with mouse-derived cells was examined by PCR using the primers: forward, 5′-TGTAG GTACTAACACTGGCTCGTGTGACAA-3′ and reverse, 5′-GGTGTTGAAGGTCTCAAACATGATCTGTA-3′. PCR with this set of primers can amplify both human and mouse β-actin genes as 247- and 273-bp DNA fragments, respectively. To examine for contamination with other human cell lines, short tandem repeat analysis was carried out using the markers: D5S818, D13S317, D7S820 and DS16S539. Primer information is available at the UniSTS database on the NCBI website (http://www.ncbi.nlm.njh.gov/). To confirm the ability to form liver metastases, the cells derived from the metastatic foci, i.e., SW48LM2 cells, were cultured and intrasplenically transferred at 1×10⁴ cells to a new cohort of NOG mice, followed by macroscopic observation of livers removed 6 weeks after the transfer.

For subcutaneous tumor formation, 1×10⁴ SW48 or SW48LM2 cells suspended in 0.2 ml of serum-free medium were subcutaneously transferred to NOG mice. Palpable tumors were measured weekly, and the mice were sacrificed for histopathological examination 6 weeks after the transfer.

For quantitative evaluation of the invasiveness based on the geometry of tumor tissues, fractal dimension analysis was applied to the histological findings as reported in a previous study (6). Mice were sacrificed 2 weeks after the subcutaneous transfer, followed by harvesting of the formed tumors. Paraffin-embedded samples were sectioned at 5 μm. Immunostaining of the sections with the mouse monoclonal anti-multi-cytokeratin (clone AE1/AE3; Leica Microsystems, Japan) was performed on the Bond Max system (Leica Microsystems). The photographic images of the immunostained sections were prepared by the Imager M1 (Carl Zeiss Microimaging, Japan) and then converted to the gray-scaled images using Photoshop CS3 (Adobe Systems Inc., San Jose, CA, USA). The fractal dimension analysis was carried out using the software ImageJ version 1.33 (http://rsb.info.nih.gov/ij/).

Comparison of the invasiveness of SW48 and SW48LM2 cells in vitro was carried out using the BD BioCoat Matrigel invasion chamber (BD Biosciences, Bedford, MA, USA) as recommended by the manufacturer.

The spheroid culture was carried out using the Sumilon Celltight Spheroid 96U (Sumitomo Bakelite, Japan). The viable cell numbers were analyzed using the Cell Counting Kit-8 (Dojindo Laboratories, Japan) as recommended by the manufacturer.

Total RNA was extracted from collected cells using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and subjected to cDNA synthesis with the PrimeScript RT reagent kit (Takara Bio Inc., Japan). The reactions were prepared with the obtained cDNA and the SYBR Premix Ex Taq II. Gene-specific primers were designed by the Perfect Real Time support system (Takara Bio Inc.) and subjected to the real-time PCR assay using the Thermal Cycler Dice Real Time system (Takara Bio Inc.). The examined molecules were: VEGF-A, VEGF189, TSP1, TSP2, MMP2, MMP7, E-cadherin, S100A4, CA9, CA12, HIF1a, HIF2a, NME1, GLUT1, GLUT3, CYGB, MYC, MCT4, SCO2, COX4-1, COX4-2, TLR3, TLR4, TLR5, TLR7, TLR9, MCL1, LDHA, LDHB, ALDH1L2, ALDH3A1, TFAM, UCP2, TWIST, Slug, Snail, ZEB1, ZEB2, vimentin, CD73, CD47, NT5M, NT5C1A, NT5C1B and NT5C2.

Extracts from the pellets of cultured cells were measured by the Agilent CE Capillary Electrophoresis system equipped with an air pressure pump, an Agilent 1100 series MSD mass spectrometer and an Agilent 1100 series isocratic high performance liquid chromatography pump, a G1603A Agilent CE-MS adaptor kit and G1607A Agilent CE-MS sprayer kit (Agilent Technologies) as previously described (7).

Cells were stained with PE conjugated mouse anti-human CD73 (clone AD2; BD Biosciences, San Diego, CA, USA). The flow cytometric analysis was performed using MoFlo (Beckman Coulter, Miami, FL, USA).

Statistical analysis was carried out using the two-sample t-test or Fisher’s exact test. Values of p<0.05 were considered to be statistically significant.

Poorly formed metastatic foci were noted in the liver following the intrasplenic transfer of SW48 cells to NOG mice (Fig. 1A). Cells were isolated from the visible foci and were then subjected to the short-term culture. The intrasplenic transfer of the cultured cells to NOG mice led to the rapid formation of hepatic metastatic foci (Fig. 1A). This established cell subline was termed SW48LM2. The growth under monolayer culture conditions was similar between the SW48 and SW48LM2 cells (Fig. 1A).

The growth features of the subcutaneous tumors were similar between the SW48 and SW48LM2 cells (Fig. 1B). In the subcutaneous transfer, no metastases were macroscopically observed in the liver or other organs, except thoracic involvement and skin ulcerations. The latter metastases tended to occur more frequently in the subcutaneous transfer of SW48LM2 cells, although the differences were not statistically significant (Fig. 1B). For evaluation of the invasiveness of the SW48LM2 cells based on tumor shape profiles, subcutaneous tumors, formed 2 weeks after the transfer, were subjected to fractal dimension analysis. The mean fractal dimensions showed that the two cell lines were similar in morphological invasiveness (Fig. 2A). In addition, the in vitro invasion assay detected no significant differences between SW48 and SW48LM2 cells (data not shown). Similar changes in the viable cell numbers were observed throughout the time period of the in vitro spheroid formation between the SW48 and SW48LM2 cells, whereas the spheroids were morphologically different (Fig. 2B). As internal hypoxia acutely affects spheroid formation via tumor cells, such morphological differences appeared to reflect the distinct features of the cells with respect to hypoxic adaptation (8). We therefore carried out the analysis of monolayer-cultured cells under normoxia and hypoxia.

Gene transcripts of 41 molecules reportedly associated positively or negatively with tumor growth, invasion or metastases were analyzed by real-time RT-PCR (2,9–16). Four overexpressed and four underexpressed genes were detected in the SW48LM2 cells (Fig. 3). Notably, an ~50-fold reduction under normoxia and an ~100-fold reduction under hypoxia were observed with respect to the ecto-5′-nucleotidase CD73 (also known as NT5E) gene transcripts. Of note was that the gene transcripts of CD73, the monocarboxylate transporter MCT4, and S100A4 were reduced in the liver-metastatic cell line, conflicting with previous reports showing a positive association with metastatic cancer phenotypes (2,9–12). In contrast, a lower expression of the TLR3 gene as well as a higher expression of aldehyde dehydrogenase ALDH1L2 and ALDH3A1 genes and a splicing variant of VEGF-A, VEGF189, corroborated reported findings (13–15). It appeared plausible that an epithelial mesenchymal transition-associated molecule, vimentin, was highly expressed in the SW48LM2 cells (16). Differential expression of genes encoding metabolic enzymes and transporter prompted us to examine the metabolomic profiles of the SW48 and SW48LM2 cells.

Metabolites of glycolysis, the pentose phosphate pathway, TCA cycle and anabolic and catabolic reactions of amino acids, as well as nucleotides in the SW48 and SW48LM2 cells were analyzed by CE-MS. As a result, a decrease in purine but not pyrimidine nucleosides and a reciprocal increase in purine nucleotides were observed in the SW48LM2 cells under normoxia and hypoxia, suggesting a possible association with a reduced CD73 expression (Fig. 4A).

To rule out the contribution of four purine nucleotidases other than CD73 to the altered metabolism, the expression levels of the nucleotisades were analyzed by real-time RT-PCR (Fig. 4B). No significant differences were noted in the expression levels of either the NT5M or NT5C2 gene between the SW48 and SW48LM2 cells cultured under normoxia and hypoxia (Fig 4B). Gene transcript levels of NT5CA1 and NT5CB1 were undetectable within the evaluable cycle numbers of real-time RT-PCR. The flow cytometric analysis showed an evident reduction in CD73 expression in the SW48LM2 cells (Figs. 3 and 4C). Purine nucleosides generated extracellularly by CD73 can be transferred into tumor cells via their nucleoside transporters, such as ENT1 and ENT2 (17). Collectively, decreased purine nucleosides and reciprocally increased purine nucleotides in the SW48LM2 cells appeared to be associated with a reduced CD73 expression.