Cancer stem cell‐like SP cells have a high adhesion ability to the peritoneum in gastric carcinoma

Side population analysis using flow cytometry.  The cells were suspended at 1 × 10⁶ cells/mL in PBS supplemented with 4% FBS. These cells were then incubated at 37°C for 60 min with 5 µg/mL Hoechst 33342 (Sigma Chemicals, St Louis, MO, USA), either alone or in the presence of 500 µmol/L verapamil (Sigma). After incubation, 1 µg/mL propidium iodide was added and then filtered through a 40‐µm cell strainer (Becton Dickinson, San Diego, CA, USA) to obtain single‐suspension cells. Analyses and sorting were performed using BD LSR (B.D.) or FACS Vantage SE (Becton Dickinson). Hoechst 33342 was excited with the UV laser at 350 nm and fluorescence emission was measured with 405/BP30 (Hoechst blue) and 570/BP20 (Hoechst red) optical filters.

Animal models.  Female BALB/c nude mice (Nihon CLEA, Tokyo, Japan) were used for the in vivo study. They were housed in specific pathogen‐free conditions and fed standard chow pellets and water ad libitum. All experiments were performed according to the standard guidelines for animal experiments of Osaka City University Medical School. We examined the tumorigenicity in each cell line by intraperitoneal or subcutaneous inoculation. OCUM‐2M, OCUM‐2D, OCUM‐2M/SP, OCUM‐2D/SP, and OCUM‐2MD3 cells in various numbers were injected into the abdominal cavity or subcutaneously into the back area of the mice, and monitored for tumor development.

Adhesion assay.  The adhesion ability of cancer cells to peritoneal components was examined using the adhesion assay. The binding of cancer cells to peritoneal components, fibronectin (Mallinckrodt Specialty Chemicals, St. Louis, MO, USA), matrigel (Collaborative Research, Bedford, MA, USA), and laminin (Mallinckrodt Specialty Chemicals), were investigated as follows. Briefly, a 96‐well microtiter plate was coated with fibronectin (4 µg/well), matrigel (2 µg/well), or laminin (4 µg/well). Gastric cancer cells and SP cells (4 × 10⁵ cells) were seeded onto these components in 96‐well microtiter plate (Falcon, Franklin Lakes, NJ). They were allowed to adhere to each well for 30 min at 37°C and then were gently washed twice in PBS. The adhesion cancer cells were quantified by the 3‐(4,5‐dimethylthiazol‐2‐yl)‐2,5‐diphenyl‐tetrazolium bromide (MTT; Sigma) colorimetric assay. The percentage of cells adhering was calculated as follows: % bindings = (OD of treated surface – only ECM component)/OD of total surface × 100. Total surface means total cancer cells, 4 × 10⁵ cells, seeded on the microtiter plates with matrigel, fibronectin, and laminin.

Real‐time reverse transcription‐PCR.  mRNA expression of CD44, the integrin family, Sox2, Oct3/4, and CD133 was examined as follows. Total cellular RNA was extracted from these cells with Trizol (Life Technologies) according to the manufacturer's protocol. cDNAs were synthesized from 1 µg of the RNA with a Moloney murine leukemia virus‐reverse transcription kit (Life Technologies) using random hexamers. To determine fold changes in each gene, real‐time reverse transcription‐PCR was done on the ABI Prism 7000 (Applied Biosystems, Foster City, CA, USA) using a commercially available gene expression assay for α1‐integrin, α2‐integrin, α3‐integrin, α5‐integrin, β1‐integrin, β3‐integrin, β5‐integrin, CD44, Oct3/4, Sox2, and CD133 (Hs00235030, Hs00158148, Hs00233722, Hs00233732, Hs01127543, Hs00173978, Hs00609896, Hs00174139, Hs00999632, Hs00415716, and Hs01009250, respectively). PCR was carried out at 95°C for 15 s and 60°C for 60 s for 40 cycles. As an internal standard to normalize mRNA levels for differences in sample concentration and loading, amplification of glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) was used. The threshold cycle values were used to calculate the relative expression ratios between control and treated cells using the formula described by Pfaffl.^{( 20 )}

Immunohistochemical techniques.  Hematoxylin–eosin staining and the immunohistochemical determination of Sox2, Oct3/4, and CD133 were examined using subcutaneous tumors. Briefly, paraffin‐embedded sections were cut at 4 µm. Following deparaffinization and heating in Target Retrieval Solution (Dako, Carpinteria, CA, USA), the sections were incubated with a 1:50 dilution of a Sox2 antibody (MAB2018; R&D Systems, Minneapolis, MN, USA), with a 1:300 dilution of an Oct3/4 antibody (sc‐5279; Santa Cruz Biotechnology, CA, USA), with a CD133 antibody (Miltenyi Biotec, Auburn, CA, USA), and were treated with a secondary antibody. The mean percentage of positive tumor cells for the expression of Oct3/4, Sox2, and CD133 was determined at 400‐fold magnification. Staining intensity was marked by ‘+++’, ‘++’, ‘+’, and ‘–’, followed by the percentage of positive tissue samples. Further, cases with more than 50% positively stained tumor cells were defined as ‘+++’, with more than 20% and less than 50% were defined as ‘++’, with few cells were defined as ‘+’, and with no positive cells were defined as ‘–’.

Statistical analysis.  Comparisons among the data sets were made by Student's t‐test or Fisher's exact test. Differences were considered to be statistically significant when the P‐value was 0.05 or less.

Tumorigenicity of SP cells in vivo.  Tumorigenicity was examined by the intraperitoneal inoculation or subcutaneous inoculation of cancer cells (Table 1). The intraperitoneal inoculation of 5 × 10⁵, 1 × 10⁶, and 5 × 10⁶ OCUM‐2M/SP cells resulted in peritoneal metastasis in one (20%), two (40%), and five (100%) of five treated mice, respectively. In OCUM‐2MD3 cells, the intraperitoneal inoculation of 5 × 10⁵, 1 × 10⁶, and 5 × 10⁶ cells resulted peritoneal metastasis in none of five mice, two (25%) of eight mice, and 11 (100%) of 11 mice, respectively. In OCUM‐2D/SP cells, the intraperitoneal inoculation of 1 × 10⁶ and 5 × 10⁶ cells resulted peritoneal metastasis in none of four mice and three (75%) of four mice, respectively. In contrast, peritoneal metastasis after intraperitoneal inoculation of 5 × 10⁵, 1 × 10⁶, 5 × 10⁶, 2 × 10⁷, and 5 × 10⁷ of OCUM‐2M cells and 1 × 10⁶ and 5 × 10⁶ cells of OCUM‐2D cells was not found in any mice. Tumorigenicity of OCUM‐2M/SP and OCUM‐2MD3 cells by intraperitoneal inoculation was significantly greater (P < 0.01, Fisher's exact test) than that of OCUM‐2M cells. Tumorigenicity of OCUM‐2D/SP cells by intraperitoneal inoculation was also significantly greater than that of OCUM‐2D cells. The subcutaneous injection of only 5 × 10⁴ OCUM‐2M/SP cells resulted in xenografted tumor formation in four (80%) of five mice, while 1 × 10⁵, 1 × 10⁶, and 5 × 10⁶ cells of OCUM‐2M resulted in xenografted tumor formation in none of nine mice, two (20%) of 10 mice, and three (25%) of 14 mice, respectively. In OCUM‐2D/SP cells, subcutaneous injection of 1 × 10⁵ and 1 × 10⁶ cells resulted xenografted tumor formation in three (27%) of 11 mice and eight (100%) of eight mice, respectively. Tumorigenicity by subcutaneous injection of OCUM‐2M/SP and OCUM‐2D/SP cells was also significantly greater (P < 0.05, Fisher's exact test) than that of OCUM‐2M and OCUM‐2D cells (Table 1).

The adhesion ability of SP cells to peritoneal components.  Figure 2a shows cancer cells binding to fibronectin, matrigel, and laminin. The number of adhesion cells of OCUM‐2MD3 and OCUM‐2M/SP was greater compared to that of OCUM‐2M and the number of adhesion cells of OCUM‐2D/SP was also greater compared to that of OCUM‐2D. The adhesion ability to fibronectin, matrigel, and laminin of OCUM‐2M/SP cells (20.0%, 26.3%, 19.3%) and OCUM‐2MD3 cells (25.5%, 35.7%, 31.6%) was significantly increased compared with that of OCUM‐2M cells (1.2%, 4.0%, 11.7%). Taken together, the adhesion ability of OCUM‐2D/SP cells (69.0%, 16.7%, 27.1%) was significantly increased compared with that of OCUM‐2D cells (24.6%, 8.3%, 10.5%) (Fig. 2b).

mRNA expression of adhesion molecules in SP cells.  The expression of α2‐, α3‐, α5‐, β1‐, β3‐, and β5‐integrin, CD44, Oct3/4, and Sox2 mRNA was significantly increased 11.1‐, 4.9‐, 29‐, 2.7‐, 10.8‐, 7.6‐, 5.8‐, 1.4‐, and 37.6‐fold in OCUM‐2M/SP, and 13.9‐, 7.7‐, 5.2‐, 1.9‐, 1.1‐, 6.8‐, 5.4‐, 3.7‐, and 1.9‐fold in OCUM‐2MD3 cells, compared with that in OCUM‐2M cells. In contrast, the expression of α1‐integrin and CD133 mRNA in OCUM‐2M/SP and OCUM‐2MD3 cells was significantly decreased compared with that of OCUM‐2M cells (Fig. 3a). In OCUM‐2D/SP cells, the expression of α2‐, α5‐, β3‐, and β5‐integrin, CD44, Oct3/4, and Sox2 mRNA was significantly increased 2.3‐, 1.6‐, 2.3‐, 3.6‐, 9.8‐, 3.4‐, and 10.4‐fold in comparison with that of OCUM‐2D cells, while the expression of α1‐, α3‐, and β1‐integrin, and CD133, mRNA was significantly decreased (Fig. 3b).

Stem cell marker expression in tumors by SP cells.  The evaluation of Oct3/4 staining of xenografted tumor was 3+ positive by OCUM‐2M/SP, OCUM‐2D/SP, and OCUM‐2MD3 cells, while that by OCUM‐2M and OCUM‐2D cells was negative. The evaluation of Sox2 expression of xenografted tumor by OCUM‐2M/SP and OCUM‐2D/SP cells was 3+ positive, while those by OCUM‐2M, OCUM‐2D, and OCUM‐2MD3 were ‘+’, negative, and negative, respectively. The evaluation of CD133 staining of xenografted tumor by OCUM‐2M/SP and OCUM‐2D/SP cells was negative, while those by parent cells were 2+ (Fig. 4).