Mesenchymal stem cells (MSCs) self-replicate and differentiate into a variety of cell types such as osteoblasts, chondrocytes, adipocytes, and smooth muscle cells [1-5]. These capacities have made MSCs useful in studies of bone and cartilage regeneration [6-8]. One of the sources of human MSCs (hMSCs) is adult bone marrow, although they occur at a rate of one per one-hundred-thousand nucleated cells [6], and the available volume of bone marrow is limited. To secure the numbers of hMSCs required for tissue regeneration, the cell must be expanded ex vivo. Although hMSCs are stable ex vivo, it is possible that they undergo transformation to an unlimited proliferation phenotype during expansion.

Previous studies have demonstrated that Ewing’s sarcoma is derived from MSCs [9-12]. Ewing’s sarcoma is a malignancy that primarily affects children and young adults, with a peak incidence between the ages of 14 and 20 years. It arises mainly in bone and less commonly in soft tissues. The t(11;22)(q24;q12) chromosomal translocation generating EWS-FLI-1 fusion gene is found in 85% of cases [13]. EWS-FLI-1 knockdown inhibits cell proliferation in Ewing’s sarcoma cells [14,15]. Thus, EWS-FLI-1 expression is believed to play a key role in Ewing’s sarcoma development. However, EWS-FLI-1 expression does not transform normal murine and human fibroblasts [16,17], suggesting EWS-FLI-1 promotes malignant transformation in selective cells.

Several reports have demonstrated that EWS-FLI-1 expression transforms murine MSCs; indeed, tumors form when these cells are injected into immunodeficient mice [9,12]. In contrast, EWSFLI- 1 expression in hMSCs does not accelerate cell proliferation and transformation (10). EWS-FLI-1 expression in hMSCs induces a gene expression profile that closely mimics that of Ewing’s sarcoma [9-11] without affecting proliferation. Therefore, MSCs are thought to be the origin of Ewing’s sarcoma, but because EWS-FLI-1 alone cannot transform hMSCs, we believe other factors are required for transformation.

The most important safety concern when using hMSCs in cell- based therapies and tissue engineering is the occurrence of unlimited proliferation during ex vivo culture. To identify the factors required for unlimited hMSC proliferation, we compared the gene expression profiles of hMSCs and Ewing’s sarcoma cell lines and found that Cyclin D2 expression was extremely high in the Ewing’s sarcoma cell lines. Overexpression of Cyclin D2 promotes proliferation of hMSCs, suggesting that Cyclin D2 is a candidate biomaker for hMSC transformation.

Cell culture

hMSCs derived from bone marrow were purchased from Lonza (Walkersville, MD) and cultured in MSCGM BulletKit, a mesenchymal stem cell basal medium with mesenchymal cell growth supplement, L-glutamine, and gentamycin/amphotericin-B (Lonza Walkersville, MD). Ewing’s sarcoma cell lines (Hs 822.T, Hs 863.T, RD-ES, and SK-ES-1) were purchased from American Type Culture Collection (ATCC; Manassas, VA). Hs 822.T and Hs 863.T were cultured in Dulbecco’s Modified Eagle’s medium (DMEM; Gibco) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Intergene). RDES was cultured in RPMI-1640 medium (Gibco) supplemented with 10% FBS. SK-ES-1 was cultured in McCoy’s 5a medium modified (Gibco) supplemented with 15% FBS. 293T (human kidney; ATCC) was cultured in DMEM supplemented with 10% FBS.

Microarray analysis

Total RNA was extracted from hMSCs and Ewing’s sarcoma cell lines with the RNeasy Mini Kit (QIAGEN, Valencia, CA). Total RNA quantity and quality were assessed on an Agilent 2100 Bio-analyzer (Agilent, Santa Clara, CA); 100 ng total RNA was used to generate biotin-modified amplified RNA (aRNA) with GeneChip 3′IVT Express Kit (Affymetrix, Santa Clara, CA). Reverse transcription (RT) of firststrand complementary DNA (cDNA) with the T7 promoter sequence was performed with the T7 oligi(dT) primer. Second-strand cDNA synthesis was used to convert the single-stranded cDNA into a doublestranded DNA template by using DNA polymerase and RNase H to simultaneously degrade the RNA and synthesize second-strand cDNA. In vitro transcription of biotin-modified aRNA with IVT Labeling Master Mix generated multiple copies of biotin-modified aRNA from the double-stranded cDNA templates. The aRNA was purified and quantified; after fragmentation, it was hybridized to GeneChip Human Genome U133 Plus 2.0 Array (Affymetrix). The arrays were stained with phycoerythrin and washed at the GeneChip Fluidics station 450 (Affymetrix). The microarrays were scanned and data extracted using GeneChip scanner 3000 7G (Affymetrix); image analysis was performed using the Affymetrix GeneChip Command Console Software and digitized with the Affymetrix Expression Console.

Data processing and pathway analysis

Data analysis was performed with GeneSpring GX 11.0 software (Agilent Technologies, Santa Clara, CA). Raw data were normalized to the 50^th percentile per chip and the median per gene. Differentially expressed genes were analyzed using Ingenuity Pathway Analysis (IPA) 9.0 (Ingenuity Systems, Redwood City, CA). Fisher’s exact test was used to calculate a P-value. Activation z-score was calculated as a measure of functional and translational activation in Networks and Upstream regulators analysis. An absolute z-score >2 was considered significant.

Real-time RT-PCR

Total RNA was reverse-transcribed with SuperScript III First- Strand Synthesis System for RT-PCR (Life Technologies Co., Carlsbad, CA). Real-time RT-PCR was performed with LightCycler Fast Start DNA Master SYBR Green I (Roche Applied Science, Basel, Switzerland) in a Roche LightCycler instrument (software version 4.0). mRNA expression was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The primers for Cyclin D2 and p16 were 5´-TACTTCAAGTGCGTGCAGAAGGAC-3´ and 5´-TCCCACACTTCCAGTTGCGATCAT- 3´ (Cyclin D2) and 5´-CACTCACGCCCTAAGC- 3´ and 5´-GCAGTGTGACTCAAGAGAA-3´ (p16). The primers for transforming growth factor-b2 (TGF-b2) and GAPDH were from Light Cycler Primer Sets (Search LC GmbH, Heidelberg, Germany).

Cloning and expression of Cyclin D2

Cyclin D2 cDNA was amplified by RT-PCR of mRNA extracted from SK-ES-1 using 5´-GAATTCGCCACCATGGAGCTGCTGTGCCACGAGG-3´ (forward; EcoR I site underlined) and 5´-CTCGAGTCACAGGTCGATATCCCGCACG-3´ (reverse; Xho I site underlined).The amplified products were cloned into pTA2 (ToYoBo, Osaka, Japan) and verified by sequencing. The verified Cyclin D2 cDNA was cloned into the EcoR I and Xho I sites of pLVSIN-CMV Pur (TaKaRa, Shiga, Japan). Lentiviral vector was prepared with the Lenti-XTM Packaging System (TaKaRa) according to manufacturer protocols.

Viral infection

hMSCs were infected with the lentiviral vector containing Cyclin D2 (hMSCs/CyclinD2) or empty vector (hMSCs/Empty) at 37°C for 24 h. Infected cells were selected with 1 mg/mL puromycin for 14 days and the bulk of the resistant cells was used in subsequent experiments.

Western blotting

hMSCs/CyclinD2 and hMSCs/Empty were lysed in RIPA buffer (Wako, Osaka, Japan). Cyclin D2 was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a membrane (Immunobilon-P^SQ; Millipore, Billerica, MA). Blots were blocked and probed overnight at 4°C with a mouse monoclonal antibody against Cyclin D2 (MBL, Nagoya, Japan). Blots were incubated with peroxidase-conjugated anti-mouse IgG (Abcam) and bound antibodies were visualized with Chemi-Lumi One Super Chemiluminescence (Nacalai Tesque, Kyoto, Japan).

Cell proliferation

Proliferation of hMSCs/CyclinD2 and hMSCs/Empty was measured with the TetraColor ONE reagent (Seikagaku Co., Tokyo, Japan). Cultures were incubated for 2 h in medium containing the reagent. Absorbance was read at 450 nm (reference at 600 nm) on a plate reader (SH-9000, Corona Electric Co., Ibaraki, Japan).

Cyclin D2 expression in Ewing’s sarcoma cell lines versus hMSCs

To identify the factors required for proliferation of hMSCs, we compared the gene expression profiles of hMSCs and four Ewing’s sarcoma cell lines (Hs 822.T, Hs 863.T, RD-ES, and SK-ES-1) (Figure 1a). Hs 822.T and Hs 863.T had similar expression profiles, as did RD-ES and SK-ES-1. The expression profiles of Hs 822.T and Hs 863.T were more similar to those of hMSCs than to those of the other Ewing’s sarcoma (RD-ES and SK-ES-1). Therefore, we first compared the expression profiles of hMSCs, Hs 822.T, and Hs 863.T. We identified 44 genes that differed by at least 10-fold between hMSCs and Ewing’s sarcoma cell lines (data not shown). These were narrowed to 9 genes by selecting genes that also differed from hMSCs by more than 10-fold in RD-ES and SK-ES-1 (Table 1). CCND2 (Cyclin D2) stood out in this group of 9 genes, because it represents a family of key cell-cycle regulators. Indeed, aberrant expression of Cyclin D2 has been associated with tumor progression in many tumor types [18-21]. We measured Cyclin D2 mRNA expression in hMSCs and Ewing’s sarcoma cell lines by real-time PCR and confirmed its extreme induction in Ewing’s sarcoma (Figure 1b). Hs 822.T and Hs 863.T exhibited similar Cyclin D2 expression levels; expression was higher in RD-ES and SK-ES-1.

Table 1: Expression of Cyclin D2 was higher in Ewing’s sarcoma cell lines than in hMSCs. The nine genes differentially expressed by at least 10-fold between hMSCs and the four Ewing’s sarcoma cell lines.

Figure 1a: Expression of Cyclin D2 was higher in Ewing’s sarcoma cell lines than in hMSCs. Global gene expression was measured in hMSCs and Ewing’s sarcoma cell lines. The whole-genome expression pattern is displayed in hierarchical cluster format. The color key shown at the bottom is in Log₂ scale.

Figure 1b: Expression of Cyclin D2 was higher in Ewing’s sarcoma cell lines than in hMSCs. Cyclin D2 expression in hMSCs and Ewing’s sarcoma cell lines was determined by real-time RT-PCR with values normalized to GAPDH. Results are plotted relative to hMSCs. Data from triplicate samples (means ± SD) are shown.

Overexpression of Cyclin D2 promoted hMSC proliferation

We transduced the Cyclin D2 gene into hMSCs by using a lentiviral vector; an empty vector served as a negative control. At 14 days after infection when puromycin selection was completed, Cyclin D2 mRNA expression in hMSCs infected with lentiviral vector containing Cyclin D2 (hMSCs/CyclinD2) was about 10,000-fold higher than in hMSCs infected with empty vector (hMSCs/Empty) (Figure 2a). Cyclin D2 was stably expressed in hMSCs at least 56 days after infection (Figure 2b). Next, we observed the cell morphology of the hMSCs/CyclinD2. Phase contrast microscopy revealed the normal fibroblast-like morphology of hMSCs/Empty (Figure 2c) and smaller spread areas in hMSCs/ CyclinD2 (Figures 2c and 2d). Furthermore, proliferation of hMSCs/ CyclinD2 was faster than that of hMSCs/Empty, indicating that overexpression of Cyclin D2 promoted hMSC proliferation. However, the proliferation slowed over time and did not result in unlimited proliferation (data not shown).

Figure 2a: Over expression of Cyclin D2 promoted hMSC proliferation. At 14 days after infection when puromycin selection was completed, Cyclin D2 mRNA in hMSCs/CyclinD2 and hMSCs/Empty was determined by real-time RT-PCR and normalized to GAPDH. Results are plotted relative to hMSCs/ Empty.

Figure 2b: Overexpression of Cyclin D2 promoted hMSC proliferation. Cyclin D2 protein expression was visualized by western blotting at the indicated days after infection. Equal amounts of protein were loaded in each lane.

Figure 2c: Overexpression of Cyclin D2 promoted hMSC proliferation. At 14 days after infection, cells were observed by phase contrast microscopy.

Figure 2d: Overexpression of Cyclin D2 promoted hMSC proliferation. Fifteen random hMSCs/Empty and hMSCs/CyclinD2 cells were measured with Image J 1.43t software (NIH, USA). Values were plotted relative to the averaged area of hMSCs/Empty.

Figure 2e: Overexpression of Cyclin D2 promoted hMSC proliferation. At 14 days after infection, equal numbers of cells were seeded in 24-well plates and proliferation was determined with Tetra Color One 1, 6, and 12 days after seeding. The optical density (Cyclin D2/Empty) of triplicate samples is shown.

TGF-b2 expression was down-regulated in hMSCs/CyclinD2, but p16 expression was not

To investigate the effects of overexpression of Cyclin D2 on the cell cycle, we examined the change in cell cycle-associated gene expression over time (p16, p21, Bmi1, TGF-b1, and TGF-b2). We did not detect significant differences in expression of p21, Bmi1, and TGF-b1 between hMSCs/CyclinD2 and hMSCs/Empty (data not shown). However, TGF-b2 expression was lower in hMSCs/CyclinD2 than in hMSCs/ Empty (Figure 3a). In addition, the increase in TGF-b2 expression during culture was suppressed in hMSCs/CyclinD2 compared with hMSCs/Empty. In contrast, the increasing rate of p16 expression in hMSCs/CyclinD2 was higher than in hMSCs/Empty, although expression in both cell types was comparable 14 days after infection (Figure 3b).

Figure 3a: TGF-b2 expression over time. TGF-b2 mRNA expression was determined by real-time RT-PCR at 14, 28, and 42 days after infection. Results are plotted relative to hMSCs/Empty at 14 days after infection. A representative of three independent experiments is shown

Figure 3b: p16 expression over time. p16 mRNA expression was determined by real-time RT-PCR at 14, 28, and 42 days after infection. Results are plotted relative to hMSCs/Empty at 14 days after infection. A representative of three independent experiments is shown.

Overexpression of Cyclin D2 altered the expression of genes associated with cell proliferation and interphase

Total RNA was extracted from hMSCs/CyclinD2 and hMSCs/ Empty 14 days after infection and analyzed by DNA microarray, which identified 690 genes that were differentially expressed by at least 2-fold between hMSCs/CyclinD2 and hMSCs/Empty (Figure 4a). Gene ontology (GO) analysis revealed these genes are associated with movement, development, growth and proliferation, cell cycle, and intercellular signaling and interactions (Table 2). Specific predictions indicated that proliferation and interphase are activated in hMSCs/CyclinD2. The induced genes that are associated with proliferation and interphase are listed in tables 3 and 4; in summary, 94 of 186 genes and 19 of 50 genes exhibited expression shifts consistent with increased in proliferation and interphase, respectively.

Figure 4a: Global gene expression in hMSCs/CyclinD2. Global gene expression patterns were compared in hMSCs/CyclinD2 and hMSCs/ Empty. Genes differentially expressed by at least 2-fold are shown (690 genes). The color key shown at the bottom is in Log₂ scale

Table 2: Global gene expression in hMSCs/CyclinD2. Gene ontology (GO) analysis of the 690 genes was performed with Ingenuity Pathway Analysis (IPA) 9.0. Top five functional categories and the specified categories are listed. An absolute z-score >2 was considered as significant.

Table 3: ‘Proliferation of cells’ genes differentially expressed by at least 2-fold between hMSCs/CyclinD2 and hMSCs/Empty.

Table 4: ‘Interphase’ genes differentially expressed by at least 2-fold between hMSCs/CyclinD2 and hMSCs/Empty

Although EWS-FLI-1 expression transformed murine MSCs, expression in hMSCs did not promote cell proliferation. In this study, we found that Cyclin D2 expression was extremely high in the Ewing’s sarcoma cell lines and overexpression of Cyclin D2 in hMSCs promoted cell proliferation. GO analysis also predicted that cell proliferation and interphase were activated by overexpression of Cyclin D2.

Cyclin D2 is a member of the family of D-type cyclins that mediate cell cycle regulation, differentiation, and oncogenic transformation [22,23]. D-type cyclins inactivate retinoblastoma (Rb) by phosphorylation, inducing release of E2F. Free E2F activates genes involved in the activation and maintenance of DNA synthesis. Thus, overexpression of Cyclin D2 generally has growth-promoting effects. Consistent with this notion, overexpression of Cyclin D2 in HeLa cells, in which Rb is inactivated by human papillomavirus E6 and E7 proteins [24], did not promote cell proliferation (data not shown). On the other hand, increased expression of Cyclin D2 inhibits proliferation of primary human fibroblasts [25], indicating that Cyclin D2 has both positive and negative roles in the cell cycle, depending on cell type. We found that Cyclin D2 in hMSCs has a positive role in the cell cycle (Figure 2e).

TGF-b2 expression was suppressed in hMSCs/CyclinD2 compared with hMSC/Empty during culture (Figure 3a). We previously demonstrated that hMSC growth is reduced and TGF-b2 expression increases during long-term culture [26]. We also reported that fibroblast growth factor-2 (FGF-2) stimulates hMSC growth by suppressing the up-regulation of TGF-b2 [27]. It is unclear how overexpression of Cyclin D2 suppresses the TGF-b2 increase, but this suppression may be involved in the acceleration of hMSCs/CyclinD2. In contrast, the expression of p16, which is up-regulated with aging [28], was increased in both cell types during culture, indicating that not only hMSCs/Empty but also hMSCs/CyclinD2 were aging normally. The rate of increase in hMSCs/Cyclin D2 was higher than in hMSCs/Empty (Figure 3b), suggesting that the promotion of cell proliferation in hMSCs/CyclinD2 induced cellular senescence and enhanced p16 expression. p16 is a tumor suppressor gene [29]; thus, this increase in p16 expression probably prevented unlimited proliferation. Consistent with this notion, some Ewing’s sarcomas contain a homozygous deletion of the p16 locus [16], possibly facilitating subsequent transformation.

In this study, overexpression of Cyclin D2 promoted proliferation of hMSCs but did not lead to unlimited proliferation. Other factors are required for the unlimited proliferation of hMSCs. IGF2BP1 was aberrantly expressed in Ewing’s sarcoma (Table 1), consistent with a previous report of an association between increased IGF2BP1 expression and tumor progression in patients with lung cancer [30]. Thus, we attempted to transduce the IGF2BP1 gene into hMSCs, but IGF2BP1 expression was up-regulated by only 2-fold and transduction efficiency was low (data not shown). The cause for this inefficiency is unclear. Because the growth kinetics of IGF2BP1-transformed E. coli is quite slow (data not shown), it is likely that overexpression of IGF2BP1 is deleterious for hMSCs.

We did not tested whether the other genes listed in Table 1 affect proliferation of hMSCs, because these genes were not thought to directly affect the proliferation. Furthermore, not all Ewing’s sarcomas express EWS-FLI-1: indeed, EWS-FLI-1 mRNA was not detected in Hs 822.T and Hs 863.T (data not shown). Thus, we did not transduce the EWSFLI- 1 gene into hMSCs. However, it is possible that the cooperation of these proteins is important for the development of Ewing’s sarcoma. Thus, it would be interesting to transduce these genes into hMSCs in addition to Cyclin D2.

Cyclin D2 promotes hMSC proliferation and is a candidate biomaker for hMSC transformation.

The authors would like to thank Atsuko Matsuoka for helpful discussions. This work was supported by the Health and Labor Sciences Research Grants for Research on Regulatory Science of Pharmaceuticals and Medical Devices (H23- IYAKU-SHITEI-027) from the Ministry of Health, Labor and Welfare of Japan.