Long-term Cultures of Bone Marrow–Derived Human Mesenchymal Stem Cells Frequently Undergo Spontaneous Malignant Transformation

Human mesenchymal stem cells (hMSC) aid in tissue maintenance and repair by differentiating into specialized cell types. Due to this ability, hMSC are currently being evaluated for cell-based therapies of tissue injury and degenerative diseases. However, extensive expansion ex vivo is a prerequisite to obtain the cell numbers required for human cell-based therapy protocols. Recent studies indicate that hMSC may contribute to cancer development and progression either by acting as cancer-initiating cells or through interactions with stromal elements. If spontaneous transformation ex vivo occurs, this may jeopardize the use of hMSC as therapeutic tools. Whereas murine MSC readily undergo spontaneous transformation, there are conflicting reports about spontaneous transformation of hMSC. We have addressed this controversy in a two-center study by growing bone marrow–derived hMSC in long-term cultures (5–106 weeks). We report for the first time spontaneous malignant transformation to occur in 45.8% (11 of 24) of these cultures. In comparison with hMSC, the transformed mesenchymal cells (TMC) showed a significantly increased proliferation rate and altered morphology and phenotype. In contrast to hMSC, TMC grew well in soft agar assays and were unable to undergo complete differentiation. Importantly, TMC were highly tumorigenic, causing multiple fast-growing lung deposits when injected into immunodeficient mice. We conclude that spontaneous malignant transformation may represent a biohazard in long-term ex vivo expansion of hMSC. On the other hand, this spontaneous transformation process may represent a unique model for studying molecular pathways initiating malignant transformation of hMSC. [Cancer Res 2009;69(13):5331–9]

Multipotent mesenchymal stem cells may differentiate into various cell lineages including osteoblasts, tenocytes, adipocytes, chondrocytes, myoblasts, and even neurons (1). Thus, these cells are considered to be promising tools for cell-targeted therapies and tissue engineering (2, 3). Tissue-specific human mesenchymal stem cells (hMSC) can be harvested from multiple sources, including placental tissue (4), amniotic fluid (5), cord blood (6), and adipose tissue (7). Yet, bone marrow aspirations remain the hMSC source of choice in most laboratories (8). However, to obtain sufficient cell numbers for clinical protocols, extensive ex vivo expansion is required. Recent controversies about the stability of hMSC (9–12) highlight the need to address hMSC stability in long-term cultures before clinical use.

Although stem cells were identified as precursors of hematologic malignancies more than a decade ago (13), stem cells have only recently been identified as cancer-initiating cells in solid malignancies (14–17). Moreover, whether most solid tumor–initiating cells derive from undifferentiated precursor cells or undifferentiated cells may play an important role in carcinogenesis by interactions with the tumor stroma is not fully understood (18). Albeit MSC are considered precursors of mesenchymal-derived solid tumors, such as sarcomas (19), their role in carcinogenesis may go beyond that. Thus, Houghton and colleagues (20) showed MSC to home from their bone marrow residence, progressing into gastric carcinomas in response to Helicobacter felis infection in mice.

Spontaneous transformation of murine MSC has been observed by others (21, 22), although conflicting evidence suggests this may not be the case for hMSC. Studies indicating no spontaneous transformation in long-term hMSC cultures have been reported (9, 10). However, one group observed spontaneous malignant transformation of adipose-derived hMSC in vitro (11), whereas another group reported a transformed population arising in one culture of bone marrow–derived hMSC/CD133⁺ (12). In accordance with the findings by Rubio and colleagues (11) about genomic instability in transformed mesenchymal cells (TMC), we suggested genomic plasticity of undifferentiated hMSC to permit longevity and thereby render the cells particularly prone to malignant transformation. Accumulation of DNA damage through genomic instability, loss of cell cycle regulation, and deregulated epigenetic signature may arise during long-term culture under standard in vitro conditions, and eventually result in malignant transformation.

The purpose of this study was to evaluate the risk of spontaneous malignant transformation of bone marrow–derived hMSC in vitro. Our findings reveal that 11 of 24 hMSC cultures underwent spontaneous transformation, resulting in highly malignant cells with an aggressive metastatic propensity in nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice. Whereas our findings have significant implications with respect to hMSC ex vivo biohazards, it also reveals a unique model to study spontaneous malignant transformation in stem cells.

Isolation and culture of bone marrow–derived hMSC. After obtaining written informed consent, bone marrow aspirates were collected from 24 healthy donors (14 from Bergen, Norway and 10 from Munich, Germany) and cultured according to two previously reported protocols (23, 24). Age of donors ranged from 18 to 53 y. The mononuclear cell fractions were isolated by Lymphoprep density gradient centrifugation (1.077 g/mL; Axis-Shield) at 800 × g for 25 min. The cells were washed and plated at 1 × 10⁶/25 cm² in tissue culture flasks (Nunc). Marrow stromal cells were selected by plastic adherence. Stromal cell cultures were passaged on reaching subconfluence or within 6 to 8 d using 0.25% trypsin and 1 mmol/L EDTA, and replated at medium densities (6,000 cells/cm²). The cells were cultured at 37°C in 5% CO₂ and 21% O₂ in medium containing α-modified MEM or DMEM supplemented with 2 mmol/L glutamine, 100 units/mL penicillin, 0.1 mg/mL streptomycin, and 10% fetal bovine serum (FBS). Unless otherwise stated, all cell culture products were purchased from Sigma-Aldrich. Phenotypic changes were examined microscopically every second day.

Phenotypic characterization by flow cytometric analysis. Briefly, 5 × 10⁵ cells were analyzed for expression of the following surface markers (see Table 1 for further characteristics): rabbit anti–platelet endothelial cell adhesion molecule-1 (CD31; 1:20; Santa Cruz Biotechnology), mouse anti-CD34-FITC (1:40; Southern Biotech), mouse anti–homing-associated cell adhesion molecule (CD44; 1:20; Sigma-Aldrich), mouse anti-CD45-FITC (1:24; Southern Biotech), mouse anti-SH2/SH3 (CD73; 1:50; BD Biosciences), mouse anti–Thy-1 (CD90)-FITC (1:12; Chemicon), mouse anti-endoglin (CD105)-R-PE (1:40; Southern Biotech), mouse anti–activated leukocyte cell adhesion molecule (CD166; 1:50; BD Biosciences), mouse anti–Stro-1 (1:18; this monoclonal antibody, developed by Beverly Torok-Storb, was obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the National Institute of Child Health and Human Resources and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA), mouse anti–HLA-I FITC (1:10; RayBiotech), and mouse anti–HLA-DR FITC (1:40; Southern Biotech). Cells (10⁶) were analyzed for rabbit anti–α-smooth muscle actin (ASMA; 1:20; Abcam) using LEUCOperm (AbD; Serotec) for intracellular antigen staining. FITC-goat anti-mouse IgG (1:100; Immunotech SAS), FITC-goat anti-rabbit immunoglobulin (1:300; Southern Biotech), FITC-goat anti-mouse IgG1 (1:300; Southern Biotech), and FITC-rat anti-mouse IgM (1:100; eBioscience) were used for detection of unconjugated primary antibodies. Ten thousand cells were analyzed on a FACSCalibur (BD Biosciences) using the proper controls to set gates. Samples before and after transformation were compared using WinMDI software (The Scripps Research Institute, Flow Cytometry Core Facility).

Multilineage differentiation potential of hMSC. Adipogenic, chondrogenic, and osteogenic differentiation assays (Differentiation media Bullet Kits, Cambrex) were done on hMSC and TMC according to the manufacturer's instructions. The extent of adipogenic induction was verified by fixation in 10% buffered formalin and staining of lipid vacuoles in 0.5% Oil-Red-O. Chondrogenic induction was verified by morphologic examination of H&E on paraffin sections and collagen type IV immunohistochemistry (1:250; Sigma-Aldrich), visualized by EnVision-HRP DAB System protocol (DakoCytomation). Osteogenic induction was verified by a calcium deposition assay according to the Stanbio Total Calcium LiquiColor procedure (Stanbio Laboratories).

Proliferation assay. Proliferation assay was done as described elsewhere, with minor variations (25). Briefly, every 7 d, 6,000 cells were seeded in 24-well plates and counted after 12, 24, 48, and 72 h under an inverse microscope using 10 nonoverlapping visual fields per well. Four wells were counted for each donor. Statistics were done using Analyze-it software for Microsoft Excel.

Senescence-associated β-galactosidase staining. Increased endogenous β-galactosidase activity at pH 6 was used to determine the state of senescence according to the β-galactosidase assay protocol (Cell Signaling Technology). Examination by light microscopy confirmed the characteristic blue product in cells possessing high β-galacosidase activity.

Colony-forming assay. Colony-forming assay was done as previously described (26). Briefly, 1 × 10⁵ single cells in suspension were seeded on an interface of two different Matrigel concentrations. Single-cell suspension was confirmed by microscopy. After 3 weeks of incubation, colony formations (diameter >60 μm) in the different specimens were compared. During long-term experiments, this setup was done repeatedly every week.

PCR. Total RNA was isolated from cultured cells using the RNeasy kit (Qiagen). cDNA synthesis was done with a cDNA kit (Boehringer) according to the manufacturer's protocol. PCR was done as previously described (1). The PCR products were analyzed using 2% agarose gel electrophoresis and visualized using ethidium bromide staining. The following oligonucleotide sequences were used: glyceraldehyde-3-phosphate dehydrogenase primers, 5′-TCCAAAATCAAGTGGGGCGATGCT-3′ (upper), 5′-ACCACCTGGTGCTCAGTGTAGCCC-3′ (lower); hTERT primers; 5′-ACCAAGCATTCCTGCTCAAGCTG-3′ (upper), 5′-CGGCAGGTGTGCTGGACACTC-3′ (lower).

Telomeric repeat amplification protocol (TRAP). Telomerase activity was measured using the Quantitative Telomerase Detection Kit (Allied Biotech) according to the manufacturer's protocol. Cell lysates were prepared using the TRAPEZE 1× CHAPS Lysis Buffer (Chemicon). The protein concentration for each lysate was determined using the Bio-Rad Protein Assay Kit. The samples were analyzed using an iCycler (Bio-Rad) and compared with a control template standard curve for final quantification of telomerase activity.

Lentiviral vector production and transduction. The lentiviral transfer vectors pWPXL-eGFP and pWPXL-LacZ, packaging plasmid psPAX.2, and envelope plasmid pMD2.G were kindly provided by Dr. Didier Trono's laboratory (CMU, Geneva, Switzerland). Lentiviral particles were produced according to the protocols described by Dr. Trono's lab.¹⁰

Briefly, Superfect (Invitrogen) was used to triple-transfect psPAX.2, pMD2.G, and the respective transfer vectors into 293FT producer cells. Supernatant containing lentiviral particles was harvested 48 h posttransfection. TMC were transduced with lentiviral particles in the presence of 10 μg/mL polybrene (Sigma-Aldrich).

Fluorescence-activated cell sorting. For use in optical imaging experiments, lentivirally transduced cells harboring high enhanced green fluorescent protein (eGFP) expression were sorted using FACSAria Special Order Cell Sorting Instrument (Becton Dickinson) on the basis of single-cell viability and high eGFP expression. The fluorescence-activated cell sorting experiments were done at the Molecular Imaging Center (Fuge, Norwegian Research Council), University of Bergen.

In vivo experiments. Immunodeficient NOD.CB17.Prkdc^scid mice (abbreviated NOD/SCID) from the Jackson Laboratory were used. The mice were anesthetized by 1% isofluorane and injected i.v. with 3.2 × 10⁶ TMC or TMC-LacZ, or s.c. with 4.5 × 10⁶ TMC in 250 μL physiologic saline. Control animals were injected with equal numbers of hMSC. Each group consisted of seven animals. The mice were monitored and weighed daily throughout the experiment. As soon as the animals lost 10% body weight or showed any signs of tumor burden, they were sacrificed and perfusion fixed with 3.7% paraformaldehyde before the tumors were removed and postfixed in 3.7% paraformaldehyde for 48 h. Tumors were embedded in paraffin, sectioned, and stained with H&E before examination by microscopy. All experiments were approved by The Norwegian Animal Research Authority and conducted according to The European Convention for the Protection of Vertebrates Used for Scientific Purposes.

In vivo optical imaging. NOD/LtSz-Prkdc^scid/B2m^null (27) mice (abbreviated as NOD/SCID/β2m^null; originally obtained from Dr. Leonard Schultz, The Jackson Laboratory, Bar Harbor, ME) were expanded and maintained under defined flora conditions in individually ventilated (HEPA-filtered air) sterile microisolator cages (Techniplast) at the University of Bergen animal facility. Before imaging, anesthetized mice were depilated and moved to the heated translational stage of the eXpore Optix (ART/GE Healthcare). Animals were maintained under gas anesthesia during scanning.

TMC-eGFP in vivo model. NOD/SCID/β2m^null mice, 6 to 8 wk old (n = 7 per group), were irradiated from a photon radiation source (BCC Dynaray CH4, 4-MV photon irradiation source) with a sublethal dose of 2.5 Gy (60 cGy/min) 24 h before injection. TMC-eGFP or hMSC (3 × 10⁶) were suspended in 200 μL physiologic saline before injection via the dorsal tail vein. Recipient mice were monitored every second day by time-domain optical imaging for eGFP and sacrificed following institutional guidelines when moribund, as defined by weight loss, lethargy, and/or paralysis/respiratory distress.

Time-domain fluorescence imaging. For eGFP fluorescence imaging, we used an eXplore Optix time-domain imager configured for eGFP imaging experiments as previously described (28). For in vivo studies, whole-body images of mice were taken before TMC inoculation and used as representative background autofluorescence for every animal. Subsequently, raw fluorescence data were analyzed by fitting of measured fluorescence lifetime decay signal by the Levenberg Marquet least squares method (28) and fluorescence lifetime gating for eGFP (2.65 ns) using Optiview software (ART, Advanced Research Technologies, Inc.).

Immunohistochemistry. Immunohistochemical staining was done on 5-μm paraffin sections of tumor tissue. Sections on coated glass slides were dried 1 h at 60°C; all subsequent steps were done at room temperature. Retrieval method, buffer solutions, antibody detection method, and concentration of the primary antibodies were varied according to the protocols described for each primary antibody. The following antibodies were used: pan-cytokeratin AE1/AE3 (1:100; DakoCytomation), CK18 DC10 [1:100; Lab Vision (Neomarkers)], CK13 KS-1A3 (1:400; Novocastra), vimentin V9 (1:100; DakoCytomation), actin 1A4 (1:600; DakoCytomation), and CD99 12E7 (1:400; DakoCytomation). The slides were incubated with primary antibody for 60 min at room temperature. Unspecific peroxidase activity was blocked by 3% H₂O₂ treatment for 5 min. The signal was developed with diaminobenzidine DAB⁺ (DakoCytomation) for 5 min. Between each step, there were two 1-min washing steps in washing buffer (15 mmol/L Tris, 5 mmol/L NaCl, 0.05% Tween 20; pH 7.5). Finally, the slides were counterstained with hematoxylin for 1 min, dehydrated, and mounted in Entellan (Merck). For microscopy, a Leica DMLB microscope was used. Pictures were taken with ColorView soft imaging system (Olympus).

Fluorescence in situ hybridization on paraffin-embedded TMC tumors in mice. Pretreatment and hybridization: Tissue sections of s.c. TMC tumors were deparaffinized and rehydrated. The samples were pretreated using a Histology FISH Accessory Kit (DakoCytomation) following the manufacturer's description. The slides were treated with 0.2 mg/mL RNase A (Sigma-Aldrich) in 2× SSC (3 mol/L NaCl, 0.3 mol/L sodium citrate, pH 7.0) for 15 min at 37°C and washed 3 × 2 min in PBS/0.2% Tween. The slides were dehydrated and air-dried. A mixture of probes, Star*FISH Human Chromosome Pan-Centromeric Probes FITC labeled and Star*FISH Mouse Pan-Centromeric Probes Cy3 labeled (Cambio), was preheated at 70°C for 5 min and cooled to 37°C. The probe was applied to the tumor sections and sealed using CISH UnderCover-Slips (Invitrogen). The slides were hybridized for 1 d on a Techne Flexigene Thermal Cycler at 90°C 8 min, 80°C 1 min, 60°C 1 min, 45°C 4 h, 42°C 4 h, and 37°C overnight. Stringency wash and mounting: Probe was rinsed off using distilled water and PBS. The slides were immersed and washed 2 × 10 min in 2× SSC, 2 × 5 min in PBS/0.2% Tween, and mounted with Vectashield with 4′,6-diamidino-2-phenylindole DAPI (Vector Labs). The samples were examined by a Leica DM RXA fluorescence microscope (Leica Microsystems).

Rate of spontaneous malignant transformation in hMSC cultures. To address the current controversies about spontaneous malignant transformation of hMSC cultures, bone marrow–derived hMSC samples were harvested and expanded in two different laboratories. Following long-term hMSC culture in vitro, we observed spontaneous malignant transformation in 11 of 24 (45.8%) cultures. The rate of transformation in the two separate laboratories was 50% (7 of 14) and 40% (4 of 10), respectively. The transformation was characterized by dramatically altered morphology, from uniform spindle-shaped cells (hMSC; Fig. 1A,, left) to round epithelial-like cells displaying an increased nucleus-to-cytoplasma ratio (Fig. 1A,, right). Transformed cells arose in colonies (Fig. 1A,, middle), indicating clonal origin of the transformed cells. Notably, 8% (2 of 24) of the cultures transformed without preceding signs of senescence. In the remaining samples (41%, 9 of 22), the changes in morphology occurred after a phase of senescence, as determined by decreased proliferation and senescence-associated β-galactosidase staining at pH 6 (Fig. 1B). These findings indicate that escape from senescence is a crucial step in the malignant transformation for the majority of samples.

To determine the influence of soluble factors in various serum, we have tested fetal bovine serum purchased from Biochrom, Sigma-Aldrich, and Biowest. The different sera were not found to influence the frequency of transformation (data not shown). Commercially available media for hMSC culture, Mesencult (Stem Cell Technologies) and MSCBM (Cambrex), were also tested, and malignant transformation was found to occur in samples cultured with these media at a frequency similar to noncommercial media. To determine whether TMC-derived paracrine factors may favor transformation, we also grew hMSC in conditioned medium obtained from TMC. We did not observe any induction of transformation (data not shown). Taken together, the results from our study indicate that malignant transformation is not triggered solely by contents of the cell culture medium or cell-derived paracrine factors, but rather suggest that intrinsic properties of hMSC may drive the process of malignant transformation.

Phenotypic alterations of TMC. To characterize the phenotypic differences between hMSC and TMC, the cultures were assayed by immunocytochemistry and analyzed by multicolor flow cytometry. Bone marrow–derived hMSC expressed high levels of the cell surface markers CD73, CD90, CD105, and HLA-1 and moderate levels of CD44, CD166, ASMA, and Stro-1. In addition, bone marrow–derived hMSC did not express the following cell surface markers: CD31, CD34, CD45, and HLA-DR (Fig. 1C,, red). These observations are in line with recently published phenotypic expression patterns of hMSC (29, 30). Compared with hMSC, TMC showed increased expression of CD44 and CD166, whereas CD105 was slightly down-regulated. Some markers used to identify hMSC, such as CD73, CD90, ASMA, and Stro-1, were undetectable in TMC (Fig. 1C , blue).

Limited differentiation potential of TMC. Human MSC cultured in the appropriate differentiation media caused differentiation into adipocytes, chondrocytes, and osteocytes (Fig. 1D). After transformation, the potential of cells to differentiate into adipocytes and chodrocytes was dramatically reduced (Fig. 1D,, left inset; and data not shown). However, differentiation into the osteogenic lineage induced elevated calcium production in TMC similar to hMSC (Fig. 1D , right), indicating intact osteogenic differentiation potential in TMC. The possible application of differentiation induction in a therapeutic context of sarcomas should therefore be further explored.

Senescence precedes malignant transformation in the majority of cultures. With exception of the two hMSC samples where no signs of senescence were detected before malignant transformation, all hMSC cultures entered senescence 25 to 71 days after harvest and initial seeding, represented by β-galactosidase staining (Fig. 1B), as well as decreased proliferation rate (Fig. 2). In contrast to the fast proliferating hMSC cultures with a mean cell doubling time of 60 hours during the first 6 to 10 weeks after harvest (Fig. 2B,, left), the senescent cultures had a mean cell doubling time of 214 hours (Fig. 2B,, middle). Among the senescent cultures, 41% (9 of 22) escaped senescence and underwent malignant transformation. TMC displayed a mean doubling time of only 12 hours (Fig. 2B , right). On escape from senescence, the TMC cultures were immortalized, and at present, the TMC are still in continuous culture and have exceeded more than 100 passages.

Transformation causes an increase in telomerase activity. Telomerase is activated in 70% to 90% of malignant tissues (31) and in many immortal cell lines, but the majority of normal somatic cells do not present any detectable telomerase activity. In accordance with published literature on hMSC (32–34), our telomerase activity assay confirmed the low endogenous levels of telomerase activity in hMSC (Fig. 3). Following malignant transformation, the telomerase activity increased significantly (P = 0.00003), in agreement with previous reports on malignant transformation (11, 22).

Enhanced anchorage-independent growth of TMC. Parallel to the observation of altered morphology and proliferation rate, we observed enhanced anchorage-independent growth and colony formation of TMC in soft agar assays, indicating malignant transformation (Fig. 4). The hMSC did not show colony formation in soft agar assays.

TMC are tumorigenic in vivo. To evaluate the tumorigenicity of TMC in vivo, cells were transduced with lentiviral eGFP vectors and injected i.v. into NOD/SCID/β2m^null mice. Engraftment of i.v. injected TMC-eGFP and subsequent disease progression were monitored longitudinally by time-domain optical imaging of eGFP fluorescence. We observed initial colonization by TMC-eGFP of the lungs from day 9, a finding subsequently confirmed by ex vivo optical imaging of the different organs at necropsy (Fig. 5A). All of the animals injected with TMC were euthanized within 39 days postinjection due to symptoms related to tumor burden (Fig. 5B, Kaplan-Meier curve). Control animals injected with hMSC were sacrificed 3 months postinjection without signs of tumor development.

For histolopathologic examination, TMC or TMC-LacZ were injected i.v. or s.c. into NOD/SCID mice. TMC-LacZ injected i.v. initiated tumor formation in the lungs (Fig. 5B,, bottom, H&E sections). The TMC tumor tissue exhibited epitheloid characteristics by means of polygonal tumor cells arranged in solid clusters and high vascularity (Fig. 5B,, top, H&E sections). By immunohistochemistry, we have characterized the s.c. tumors using three epithelial specific markers, AE1/AE3 (a cytokeratin cocktail), CK18, and CK13 (Fig. 5C,, left), as well as the primitive/mesenchymal marker vimentin, a marker of myogenous differentiation (actin 1A4), and a marker of desmoplastic small round blue cell tumors and Ewing sarcoma (CD99; Fig. 5C,, right). Taken together, these results indicate that the TMC give rise to morphologically epitheloid tumors with expression of epithelial and mesenchymal markers of differentiation. Staining with human and mouse specific centromere probes verified the human origin of the tumor (Fig. 5D), and showed involvement of host endothelial cells in neovascularization in developing s.c. tumors (Fig. 5D , bottom).

Recent publications present conflicting evidence about the transformation propensity of hMSC. Whereas Rubio and colleagues (11) observed spontaneous malignant transformation of adipose-derived hMSC in vitro, other investigators reported hMSC to remain stable with no transformation in long-term cultures of bone marrow– and adipose tissue–derived hMSC (9, 10). In line with the findings of Rubio and colleagues, we have shown that TMC depend on high hTERT expression to sustain the telomerase activity required for their rapid proliferation. According to several reports, the telomerase hTERT gene alone does not induce malignant transformation (33, 35). However, in a recent study, Christensen and colleagues (36) show that hTERT-transduced hMSC are prone to extensive radiation-induced genomic instability and subsequent malignant transformation. One possibility is that in vitro cell culture conditions provide stress-induced genomic instability, contributing to the malignant phenotype. Notably, mutation propensity has been related to oxygen tension (37), and matrix elasticity reported to direct lineage specificity of MSC in culture (38, 39). Thus, optimalization of ex vivo culture conditions is a prerequisite for expansion of hMSC for use in clinical protocols. Whatever the complete mechanism of spontaneous malignant transformation may be, the findings of our laboratories highlight the unstable nature of undifferentiated bone marrow–derived hMSC and the need to determine the molecular pathways involved in spontaneous transformation before safe use in clinical protocols.

In a larger context, our data fit into a scenario in which hMSC in vivo home from their areas of residence into areas of inflammation or extensive tissue damage to contribute to tissue repair and restoration of tissue homeostasis. Areas of recurrent or continuous repair are prone to malignant transformation, and hMSC may ultimately contribute to the development of cancer in situ as shown by others. Although such processes most likely are tightly controlled, there may be instances where an inflammatory microenvironment as a stress factor could trigger genetic events that favor malignant transformation.

Our results show that a significant fraction of bone marrow–derived hMSC samples undergo spontaneous malignant transformation following in vitro culture. In our opinion, this has a tremendous effect on the biosafety issues of future cell-based therapies and regenerative medicine regimens.

No potential conflicts of interest were disclosed.

Grant support: Norwegian Cancer Society, The Norwegian Research Council, The Innovest Program on Quality Research, Helse-Vest, Haukeland University Hospital, The Bergen Translational Research Program, the Centre de Recherche de Public de la Santé (CRP-Santé) through a grant from the Research Ministry in Luxembourg, the European Commission 6th Framework Programme (contract 504743), and the Binational SYSTHER-INREMOS Virtual Institute, funded by the German and Slovenian Federal Ministries of Education and Research. The funding by the Wilhelm-Toennis Grant of the German Society of Neurosurgery is appreciated.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Christine Eriksen and Synnøve Yndestad for technical assistance, and Ole Johnny Steffensen for immunohistochemical staining. Furthermore, we would like to thank the volunteer bone marrow donors, who made this study possible.