Human malignant melanoma cell line A-375, human lung adenocarcinoma cell line A549 and human colorectal adenocarcinoma cell line HT-29 were purchased from ATCC (Manassas, VA, USA). Human epidermoid carcinoma cell line A-431, human gastric carcinoma cell line NCI-N87 and human pancreatic adenocarcinoma cell line Capan-1 were obtained from the Korean Cell Line Bank (Seoul, Korea). The cell lines were cultured in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) and 0.03% antibiotic-antimycotic (all from Gibco, Grand Island, NY, USA) at 37°C in a 5% CO₂ humidified chamber.

hATMSCs were prepared from surplus banked stem cells under good manufacturing practice conditions (RNL Bio, Seoul, Korea) and used as anonymized materials. In brief, human abdominal subcutaneous fat tissue was obtained by simple liposuction following informed consent from donors, digested with collagenase I (1 mg/ml), and filtered through a 100-μm nylon sieve to remove cellular debris and centrifuged at 470 g for 5 min. The pellet was resuspended in RCME cell attachment medium (RNL Bio) containing 10% FBS and cultured at 37°C. After 24 h, the cell medium was changed to RKCM cell growth medium (RNL Bio) containing 5% FBS. When the cells reached 80–90% confluency, they were subcultured and expanded in RKCM medium until passage 3. The immunophenotypes of the hATMSCs were analyzed using flow cytometric analysis. Trypsinized hATMSCs were suspended [1×10⁶ hATMSCs/25 μl of phosphate-buffered saline (PBS)] and stained with FITC-conjugated CD31, CD34 and CD45 antibodies and PE-conjugated CD73 and CD90 antibodies (BD Pharmingen, San Diego, CA, USA).

Severe combined immunodeficient (SCID) and nude mice can be used as human tumor xenograft models (25). Six-week old female BALB/c-nude mice (Orient Bio Inc., Seongnam, Korea) and hairless SCID mice (Charles River, Wilmington, MA, USA) were housed in an environmentally controlled room (22±2°C, 40–60% humidity and a 12-h light cycle). The present study was approved by the Institutional Review Board (IRB) and the Institutional Animal Care and Use Committee (IACUC) of the Biomedical Research Institute at Seoul National University Hospital. Cancer cells (5×10⁶) were subcutaneously inoculated into the back of mice. The injection was made through the subcutaneous layer of the cervicodorsal part of the animals. When the tumor volume reached close to 100 mm³, the mice were intratumorally injected with 1×10⁵ hATMSCs suspended in PBS or PBS once a week for 8 weeks (n=3–6/group). The tumor size was measured twice a week using a caliper (Mitutoyo, Tokyo, Japan), and the tumor volumes were calculated as π/6 × (volume × width × height) according to a previously described method (26). After sacrifice at 8 weeks, the tumor masses were removed, weighed and fixed in 10% neutral buffered formaldehyde solution for histological analysis or preserved at −70°C.

A549 and HT-29 cancer cells (1×10⁵) were co-cultured with hATMSCs in RPMI-1640 medium with 10% FBS in 6-well plates for 72 h. Indirect co-culture was conducted using Transwell inserts (0.4-μm pore size, polycarbonate membrane; SPL, Pocheon, Korea) in 6-well plates to prevent cell-to-cell contact. All of the experiments were conducted in triplicate. Total cell number was assessed by a trypan blue exclusion assay using a hemocytometer under a microscope.

The proteins were extracted using RIPA lysis buffer (Millipore, Bedford, MA, USA) containing 0.1% phosphatase inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA) and protease inhibitor (Roche, Indianapolis, IN, USA). The protein concentrations were determined using a BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA). Protein samples were loaded for immunoblotting using antibodies against phosphorylated nuclear factor κB (NF-κB) p65, β-actin (Cell Signaling Technology, Beverly, MA, USA), p-JAK3, p-STAT3 and β-catenin (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Densitometry of the target bands was analyzed using ImageJ [National Center for Biotechnology Information (NCBI)].

Total RNA from the tumors was isolated using Ambion mirVana miRNA Isolation kit (Ambion, Austin, TX, USA). RNA quality was assessed using an Agilent 2100 Bioanalyzer using the RNA 6000 Nano Chip (Agilent Technologies, Amstelveen, The Netherlands), and the quantity was determined by the ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Affymetrix GeneChip^® Human Gene 2.0 ST Array (Affymetrix, Santa Clara, CA, USA) was used to analyze differential gene expression profiles as previously described (27). In order to determine whether genes were differentially expressed among the groups, the genes that showed >1.5-fold difference between the average signal values of the control and test groups were selected. Additionally, genes with a p-value <0.05 were extracted by Student's t-test (two-tailed, unpaired). After significant genes were identified, the web-based tool Database for Annotation, Visualization and Integrated Discovery (DAVID) was used to perform the biological interpretation of the differentially expressed genes. The genes were classified based on the information of gene function in Gene Ontology (GO) (http://david.abc.ncifcrf.gov/home.jsp).

Cancer cells were analyzed by a flow cytometer (FACSCalibur; Becton-Dickinson, San Jose, CA, USA) to assess levels of proliferation, apoptosis and NF-κB expression. Apoptosis and proliferation were evaluated using Annexin V-FITC Early Apoptosis Detection kit (Cell Signaling Technology) and BrdU Flow kit (BD Pharmingen), respectively, according to the manufacturer's instructions. Harvested tumor cells were centrifuged, resuspended in PBS and fixed in 2–4% formaldehyde. Phospho-NF-κB p65-Alexa Fluor 647 conjugate (Cell Signaling Technology) antibody was used for NF-κB analysis.

Statistical analysis of the results was performed using one-way ANOVA (SPSS, Inc., Chicago, IL, USA). p-values of <0.05 were considered to indicate statistically significant results. If the variance was significant, the data were analyzed by the multiple comparison procedure of Dunnett's test.

For immunophenotypic characterization of the hATMSCs, surface marker expression was examined by flow cytometry. The hATMSCs expressed high levels of MSC markers CD73 and CD90, whereas the hATMSCs were negative for cell adhesion molecule marker CD31 and hematopoietic stem cell markers CD34 and CD45 (Fig. 1A).

In Korea, the stomach, colon and lung have been reported to be the five leading primary sites of cancer both in males and females in 2013 (28). In addition to xenograft models for these cancers, three well-known xenografts, A-375 melanoma, skin squamous cell carcinoma A-431 and Capan-1 pancreatic xenografts (29–31), were used in order to investigate the dual effects of the hATMSCs on tumor growth. The tumor size was measured for 8 weeks after the first subcutaneous hATMSC injection in the BALB/c-nude mice subcutaneously inoculated with the cancer cells. The rates of growth of the A431, Capan-1 and A549 tumors were reduced in the hATMSC-injected groups compared with the respective control groups, although the differences did not reach statistical significance (Fig. 1B). In contrast, NCI-N87 and A375 tumors showed the tendency toward increased volume in the hATMSC-injected groups in comparison to the respective control groups. Particularly, the tumor weight of the HT-29 tumors in the hATMSC group was significantly greater than that in the control group. The differences in tumor weights between the hATMSC and the respective control groups showed consistent patterns with the changes in tumor volume (Fig. 1C and D).

Based on the changes in mean volumes and weights by the hATMSC injection in the BALB/c-nude mice, we selected the A549 and the HT-29 tumors for additional in vivo experiments using hairless SCID mice, and confirmed similar dual effects of the hATMSCs in these two types of tumors. The volume and weight of the A549 tumors were significantly reduced in the hATMSC group compared with the parameters in the control group, whereas the tumor volume and weight of the hATMSC-treated HT-29 tumors were significantly greater than the control group (Fig. 2).

Tian et al (8) reported opposite effects of MSCs in vivo and in vitro; human BM-MSCs exerted a tumor inhibitory effect in vitro, yet surprisingly had a promoting effect on tumor growth in vivo. To confirm the effects of the hATMSCs on tumor growth in vivo, the A549 and HT-29 cells were co-cultured with the hATMSCs at different ratios of 10:0.5, 10:1 and 10:2 (Fig. 3A). The direct co-culture of the A549 cells with the hATMSCs at the ratios of 10:1 and 10:2 attenuated the proliferation rates of the cells by 9.78 and 19.82%, respectively, compared to the control group (10:0), whereas the direct co-culture of the HT-29 cells with the hATMSCs at the ratios of 10:1 and 10:2 increased the proliferation rates of the cells by 17.22 and 29.78%, respectively, compared to the control group (10:0). This probably resulted from the proliferation rates of the cancer cells changed by the hATMSCs. However, further evidence that excluded the confounding effects associated with changes in the proliferation rates of the hATMSCs by the cancer cells was needed. Thus, indirect co-culture of the cancer cells with the hATMSCs using cell inserts was conducted in order to prevent cell-to-cell contact. The same trends were observed in the indirect co-culture. The proliferation rates of the A549 cells were reduced by 9.72 and 18.57%, respectively, whereas those of the HT-29 cells were increased by 19.90 and 25.79%, respectively, compared to the control group (10:0).

The proliferation of the cancer cells indirectly co-cultured with the hATMSCs was measured by BrdU staining and flow cytometry. The percentage of BrdU-positive A549 cells was reduced by the hATMSC co-culture by 34 and 23% at the ratios of 10:1 and 10:2, respectively, whereas there was a significant induction of the percentage of BrdU-positive HT-29 cells co-cultured with the hATMSCs at the ratios of 10:2 (Fig. 3B). The apoptotic rates of the cancer cells co-cultured with the hATMSCs were analyzed by Annexin V/PI double staining and flow cytometry. While the percentage of live A549 cells was significantly diminished by co-culture with the hATMSCs, the percentage of apoptotic A549 cells was elevated by 40, 85 and 95% at the ratios of 10:0.5, 10:1 and 10:2, respectively (Fig. 3C). However, the hATMSCs had opposite effects on the HT-29 cells; there was an increase in live HT-29 cells and a decrease in apoptotic HT-29 cells of 23 and 26% at the ratios of 10:1 and 10:2, respectively. Collectively, the hATMSCs inhibited the proliferation, yet elevated the apoptosis in the A549 cancer cells, whereas it promoted the proliferation and reduced apoptosis in the HT-29 cancer cells.

In the microarray experiment with the A549 tumors of the hairless SCID mice, 4,022 probe sets were used to identify significant differences between the hATMSC and the control group in regards to expression levels of genes (fold-change >1.5) (Fig. 4A and B). Particularly, we found that the expression levels of 1,725 probes were increased and those of 2,297 probes were decreased in the hATMSC group. In contrast, in the microarray experiment with the HT-29 tumors of the hairless SCID mice, 164 probes were found expressed more highly in the hATMSC group, whereas 135 probes were found to be more lowly expressed in the hATMSC group. Based on more stringent statistical criteria (fold-change >1.5 and p<0.05), 568 genes were expressed at a higher level and 1,191 genes were expressed at a lower level in the hATMSC group compared with the control group in the A549 tumors (Table I). In contrast, 9 genes were expressed at a higher level and 14 genes were expressed at a lower level in the hATMSC group when compared with the control group in the HT-29 tumors (Table II).

Among the genes differentially expressed in the xenograft tumors after the hATMSC injection, histone cluster 1, H2aj (HIST1H2AJ) was expressed in direct proportion to the xenograft tumor size. In other words, it was expressed at lower level in the A549 tumors and was expressed at a higher level in the HT-29 tumors after the hATMSC injection in comparison to the respective control groups. In addition, neuropeptide Y receptor Y4 (NPY4R) and LOC100505817 were expressed in inverse proportion to the xenograft tumor size. They were expressed at a higher level in the A549 tumors and were expressed at a lower level in the HT-29 tumors after the hATMSC injection in comparison to the respective control groups (Fig. 4A and C).

In order to identify the biological function of the genes differentially expressed following the hATMSC injection in the A549 and HT-29 tumor models using hairless SCID mice, GO analysis was performed. The statistically significant categories of GO terms [p<0.01; false discovery rate (FDR) <0.05] are shown in Table III. No enriched GO term was altered following treatment with the hATMSCs in the HT-29 tumors, whereas the genes involved in biological processes, such as 'Regulation of growth' and 'Regulation of cell growth' were upregulated in the hATMSC-treated A549 tumors. The genes involved in biological processes, such as 'Nucleosome assembly', 'Nucleosome organization', 'Protein-DNA complex assembly', 'Chromatin assembly or disassembly', 'Cell-cell adhesion' and 'Regulation of transcription (DNA-dependent)' were downregulated in the hATMSC-treated A549 tumors.

The proteins extracted from the xenograft tumor masses of the hairless SCID mice were analyzed by western blot analysis to investigate whether there is a relationship between certain cell signaling pathways and the hATMSC-associated modulation of tumor growth. Among several proteins related to cell signaling pathways, the expression level of phosphorylated NF-κB p65 was reduced by 31% in the hATMSC-injected A549 tumor, whereas it was increased by 77% in the hATMSC-injected HT-29 tumors (Fig. 5A and B). The levels of other cell signaling-related proteins, such as p-JAK3 and p-STAT3, were not affected by the hATMSCs in the A549 and HT-29 tumors, and the β-catenin protein level was elevated by the hATMSCs in the HT-29 tumors, yet not affected in the A549 tumors, providing a possible correlation between the NF-κB pathway and hATMSC-related tumor modulation, at least in the A549 and the HT-29 tumors.

Changes in phosphorylated NF-κB p65 in the tumor masses in vivo were confirmed using flow cytometric analysis in the cancer cells indirectly co-cultured with the hATMSCs, which revealed significant reductions in phosphorylated NF-κB p65 by 12, 25 and 32% in the co-culture of the A549 cancer cells with the hATMSCs at the ratios of 10:0.5, 10:1 and 10:2, respectively, compared with the control (10:0) (Fig. 5C). However, there was an inverse effect in the HT-29 cancer cells, showing a profound increase in phosphorylated NF-κB p65 by 15% in the co-culture of HT-29 cancer cells with the hATMSCs at the ratio of 10:2.