Spontaneous Transformation of Cultured Mouse Bone Marrow–Derived Stromal Cells

Endothelial progenitor cells (1), mesenchymal stem cells (2), hematopoietic stem cells (3), and other bone marrow–derived cells (4) have been shown to contribute to tissue revascularization and repair in animal models of ischemia, and clinical trials are currently under way to test their safety and efficacy in patients (5, 6). At the same time, evidence has grown implicating stem cells in oncogenesis (7). Data from animal models support the origin of epithelial cancers from bone marrow–derived cells (8). Nevertheless, cancer is believed to occur through a multistep process involving the accumulation of mutations in key regulatory genes that promote cell survival and proliferation (9, 10). In this article, we describe how our efforts to investigate the use of mouse bone marrow–derived stromal cells to promote limb revascularization after femoral artery ligation led to the surprising discovery of a tissue culture protocol that results in the rapid generation of cells that have spontaneously undergone malignant transformation.

Isolation and culture of marrow stromal cells. Three 7-week-old C57BL/6 or BALB/c male mice (Jackson Laboratory, Bar Harbor, ME) were anesthetized with an i.m. injection of 1 mg ketamine and 0.5 mg xylazine per animal. Tibiae and femurs were isolated using sterile techniques. Bone marrow cells were collected by flushing the tibiae and femurs with serum-free low-glucose DMEM supplemented with 1 mmol/L EDTA and penicillin/streptomycin (all from Bio Whittaker, Walkersville, MD). Pooled marrow from three animals was first dispersed by gentle pipetting and then separated by gradient centrifugation with lymphocyte separation liquid (density = 1.083 g/mL; Sigma, St. Louis, MO) as follows: 6 mL of the medium containing the marrow cells was layered on top of 3 mL of separation liquid and centrifuged at room temperature at 2,800 rpm for 20 minutes. The mononuclear cells in the middle layer were collected and washed with serum-free DMEM three times by centrifugation, first at 2,000 rpm for 15 minutes then twice at 700 rpm for 10 minutes. Cells (3 × 10⁷) collected after the last wash were resuspended in 10 mL DMEM supplemented with 10% fetal bovine serum (FBS; Gemini BioProducts, Woodland, CA) and penicillin/streptomycin, plated on one uncoated 10-cm tissue culture dish, and incubated at 37°C in 95% air/5% CO₂. Three days later, the nonadherent cells were removed by medium change, and the adherent cells were grown for 2 weeks with media changes thrice per week. At passage 1 (P1), the cells were trypsinized and reseeded in one 6-cm dish. After another 2 weeks, and they were split 1:3 (P2), followed by 1:4 (P3) and then 1:6 (P4 and thereafter), which reflected the increasing rate of cell proliferation. P4 cells were frozen stepwise (30 minutes at −20°C and then overnight at −80°C) and stored in liquid nitrogen until use. Three independent lines of mouse marrow stromal cells (MSC) were generated from C57BL/6 (designated B6.6 and B6.8 cells) and BALB/c (designated Bc.9 cells) mice. Cells were used at P6 for all in vivo studies.

Flow cytometry analysis of cell surface markers. Cells were seeded in a 10-cm dish and cultured to 80% confluence and dispersed by PBS supplemented with 2 mmol/L EDTA and 1% bovine serum albumin. Cells were incubated with Fc blocker at 4°C for 15 minutes and then incubated with various specific phycoerythrin-labeled antibodies or with isotype-matched control antibodies at a concentration of 2 μg/mL for 30 minutes. All antibodies were obtained from BD Biosciences (San Jose, CA) except anti-CD11b, which was from eBioscience (San Diego, CA). After two washes, the cells were analyzed in a FACSCalibur flow cytometer using FACSCan and CellQuest software (BD Biosciences).

Femoral artery ligation and injection of MSCs. Female BALB/c nude (Charles River Laboratories, Wilmington, MA; developed through crosses and backcrosses between BALB/cABon-nu and BALB/cAnNCrj) and scid (National Cancer Institute, Frederick, MD) mice were subjected to general anesthesia with a s.c. injection of 25 mg/kg ketamine and 10 mg/kg xylazine. The skin at the surgical site was shaved and sterilized with disinfectant. The left femoral artery was ligated at mid-thigh level (immediately proximal to the bifurcation of saphenous and posterior tibial branches), first at the proximal end and then the distal end, and excised. Twenty-four hours later, mice were injected along the ligation area with either 1 × 10⁶ MSCs or saline. The mice were divided into three groups (n = 3 mice each): i.m. saline injection, i.m. MSC injection (eight sites around ligation area, 20 μL each site, total of 160 μL PBS containing 1 × 10⁶ cells), and tail vein MSC injection (100 μL PBS containing 1 × 10⁶ cells). All mice were euthanized 35 days after cell injection. Tumors and lungs were excised, fixed in 10% formalin, and embedded in paraffin, and sections were stained with H&E.

Tumor transplantation. Tumors that formed after i.m. injection of B6.6.P6 or B6.8.P6 cells were excised, and a 1 × 1 mm tumor fragment was transplanted into the thigh muscle of an 8-month-old C57BL/6 mouse.

Karyotype analysis. A tumor that formed after i.m. injection of B6.8.P6 cells was excised, and a 4 × 4 mm tumor fragment was minced and transferred to a 15-mL tube containing DMEM and digested with trypsin for 2 hours and then cultured in DMEM/10% FBS. The tumor-derived cells as well as primary cultures of B6.8.P6, Bc.9.P6, B6.8.P3 cells, and P6 rat MSCs were analyzed in the Cytogenetics Core of The Johns Hopkins Cancer Center.

Establishment of mouse MSC cultures. In an effort to investigate the role of MSCs in stimulating vascularization of ischemic tissue, we isolated the mononuclear cell fraction of bone marrow from tibiae and femurs of three male C57BL/6 mice by density gradient centrifugation and pooled these cells on a single uncoated 10-cm tissue culture dish in DMEM supplemented with 10% FBS. The medium was changed after 3 days, adherent cells were cultured for 2 weeks, trypsinized, and replated on a 6-cm dish. After 2 weeks, cells were trypsinized and split 1:3 and 2 weeks later were split 1:4. By passage 3 (P3), the cells (which we designated B6.6) had assumed fibroblastoid morphology and were dividing rapidly in three-dimensional clusters that indicated a loss of contact inhibition (Fig. 1A). Analysis of cell surface marker expression revealed that P7 cells were markedly different from P1 cells, with uniformly high expression of Sca-1 and CD44, increased expression of CXCR4 and CD105, and no expression of the hematopoietic and endothelial cell markers CD45, CD34, CD31, CD11b, c-kit, and Flk-1 (VEGFR2; Fig. 1B; data not shown). These marker profiles are similar to those reported for human MSCs (2–4).

Injected MSCs form tumors in nude and immunocompetent mice. Immunodeficient nude mice were subjected to femoral artery ligation and injected with saline or 1 × 10⁶ C57BL/6 MSCs at P6 (B6.6.P6 cells) into the ischemic thigh muscle or into the tail vein (n = 3 each). Between 3 and 4 weeks after i.m. injection, all three mice had developed rapidly growing tumors at the injection site (Fig. 2A). Histologic evaluation on day 35 revealed soft tissue sarcomas composed of poorly differentiated mesenchymal cells that had invaded into adjacent skeletal muscle and contained large areas of necrosis (Fig. 2B and C). Among three mice that received i.v. injection of B6.6.P6 cells, one mouse developed a large tumor at the base of the tail, and another developed multiple tumors along its back, which probably followed extravasation of the cells during tail vein injection. This latter mouse was also found to have a tumor nodule in the lung (Fig. 2D).

To determine whether the tumorigenicity of B6.6.P6 cells was due to isolation of a rare aberrant clone, we compared the B6.6.P6 cells with a second independent MSC line that we established from bone marrow mononuclear cells of three other C57BL/6 mice 2 months later, which we designated B6.8 cells. Furthermore, to determine whether tumorigenesis was dependent upon the nude genotype, we used recipient scid mice. B6.6.P6 and B6.8.P6 cells were injected i.m. or i.v. into mice that had been subjected to femoral artery ligation or not (n = 3 each experimental condition) to rule out the possibility that tissue hypoxia-ischemia was promoting tumor growth (11). All mice that received i.m. injection of MSCs developed invasive tumors at the injection site (Fig. 3A) that were composed of neoplastic cells with pronounced nuclear atypia (Fig. 3B), regardless of the MSC line injected, or whether the recipient mouse was subjected to femoral artery ligation. Mice that received i.v. injection of MSCs developed tumors in the lung, regardless of the MSC isolate injected, or whether they were subjected to femoral artery ligation. Remarkably, the lung tumors contained areas of bone formation (Fig. 3C) and melanocyte proliferation (Fig. 3D). These results suggest that within a population of MSCs in which there was clear evidence of malignant transformation, some cells retained a degree of plasticity that was sufficient for differentiation.

Mice with soft tissue sarcomas resulting from i.m. injection of B6.6.P6 or B6.8.P6 cells were euthanized on day 22, and tumor fragments from these mice were transplanted into the thigh muscle of syngeneic C57BL/6 mice. In addition, tumor explants were cultured in DMEM/10% FBS to obtain cells for karyotype analysis. Immunocompetent C57BL/6 mice that received transplants of tumors derived from B6.6.P6 or B6.8.P6 cells developed palpable tumors at 2 weeks and were euthanized at 8 weeks when tumor size was similar to that observed in nude or scid mice at 3 to 5 weeks. Histology again revealed invasive sarcomas (Fig. 3E) with marked nuclear atypia (Fig. 3F). Finally, immunocompetent C57BL/6 mice were injected with B6.6.P6 or B6.8.P6 cells that were passaged only in tissue culture before injection (n = 2 mice per MSC line). Sarcomas developed at the i.m. injection site in all four recipient mice.

Cultured mouse MSCs have an unstable karyotype. The aberrant nuclear morphology that was consistently observed in the tumor histology suggested the possibility of chromosomal abnormalities. Karyotype analysis of a sarcoma that formed after the injection of B6.8.P6 cells revealed that 50 of 50 cells analyzed had greater than the normal mouse number of 40 acrocentric chromosomes, with total chromosome number ranging from 60 to 77 per cell (Fig. 4A). In addition, 12 of 50 tumor cells contained one to seven Robertsonian translocations, which result from the fusion of two acrocentric chromosomes (Fig. 4B).

To determine whether chromosomal aneuploidy was present in the cells before injection into mice, B6.8.P6 cells that had been passaged only in tissue culture were analyzed. Fifty of 50 cells analyzed were aneuploid, with 66 to 82 chromosomes per cell (Fig. 4A), and 14 of 50 cells had one to five Robertsonian translocations (Fig. 4B). Thus, the cells showed chromosomal aneuploidy before tumor formation in vivo.

To determine whether MSCs from C57BL/6 mice were unusual with respect to their chromosomal instability when cultured ex vivo, MSCs were isolated from BALB/c mice. These MSCs, which we designated Bc.9 cells, showed a similar behavior in culture, with a rapid increase in proliferation starting at P3. Cell surface marker expression of Bc.9 cells at P4 revealed that the cells were uniformly positive for CD44 and Sca-1, strongly positive for CD105, and negative for CD45, CD117, and Flk-1 (data not shown). Karyotype analysis of Bc.9 cells at P6 revealed that 50 of 50 cells analyzed were aneuploid, with 66 to 79 chromosomes per cell (Fig. 4A), and each of 50 cells analyzed had at least three translocations (Fig. 4B). Thus, a tumor and two independently derived mouse MSC lines all manifested a markedly aneuploid karyotype and chromosomal instability.

To determine whether chromosome instability arose at the same time as the cells acquired rapid proliferation and lost contact inhibition, we did karyotype analysis on B6.8 cells at P3. Remarkably, these cells also manifested a high degree of aneuploidy and chromosomal instability similar to that which was observed at P6, although in contrast to the P6 cultures, a minority population of cells was present that had maintained the normal chromosome number of 40 per cell (Fig. 4A and B).

In contrast to the behavior of mouse cells, cultures of P6 MSCs derived from rat bone marrow mononuclear cells maintained contact inhibition in culture (Fig. 5A and B) and showed no evidence of lung tumor formation after i.v. injection into syngeneic recipients. Karyotype analysis revealed that the majority of these cells maintained the normal number of 42 rat chromosomes (Fig. 5C), with no translocations, although it is noteworthy that 3 of 50 cells analyzed were tetraploid (84 chromosomes).

These studies show that mouse MSCs are rapidly and spontaneously transformed into malignant cells when cultured under the conditions we have described. These results were obtained and replicated by two different individuals working 1 year apart and are not due to idiosyncratic effects of a particular mouse strain or lot of fetal bovine serum. We have observed that the concentration of the cells at P1 seems to represent a critical aspect of our protocol that is necessary for transformation. The resulting clones showed aneuploidy and chromosomal instability by P3, formed locally invasive sarcomas after i.m. injection, and formed pulmonary nodules after i.v. injection while retaining some characteristics of stem cell plasticity (2, 12, 13). We are not aware of any other adult primary cell type that manifests these unusual properties, which suggests that a small number of genetic alterations are required for their transformation. Delineation of the molecular mechanisms underlying this transformation may provide new insights into the process of oncogenesis and the emerging role of stem cells in cancer (7). Our results provide a method by which sarcomas can be experimentally induced in syngeneic mice of any genotype by altering the cellular microenvironment without forced overexpression of specific oncogenes.

The observed increase in chromosome number suggests that either cell fusion or chromosome replication without cytokinesis occurred by P3, resulting in tetraploid cells with 80 chromosomes. Translocation and nondisjunction events in subsequent mitoses altered the chromosome number to the 60 to 82 observed in the three cell lines that were analyzed. The wide variation in chromosome number within each cell line or tumor indicates continued chromosome instability. Aneuploidy and chromosomal/genomic instability appear early in human tumorigenesis and may contribute to cancer initiation (14–16). Fusion of mouse bone marrow–derived cells with host cells in vitro and in vivo has been shown (17–19). Human MSCs manifested a high rate of cell fusion with cocultured pulmonary epithelial cells, but the tumorigenicity of these cells was not investigated (20). The isolation of nontumorigenic MSCs from C57BL/6 mice has been reported, but the protocol involved use of an anti-CD11b antibody to remove adherent granulocytes and monocytes (21). This difference suggests that fusion of these myeloid cells with MSCs may result in transformation. However, our attempts to show fusion by coculturing of bone marrow cells from mice with different cell surface markers have not been successful. Additional studies are required to resolve this issue.

The results presented in this report may have clinical relevance because trials involving the administration of human MSCs to children with genetic disorders and adults with ischemic cardiovascular disease are ongoing. Bone marrow transplantation is associated with an increased risk of cancer in recipients, although there are many potential explanations for this observation (22). Traditional transplantation involves the direct administration of cells to the recipient following harvesting from the donor, whereas protocols using bone marrow progenitors for treatment of ischemic cardiovascular disease in some cases involve expansion of cell populations by culturing. Cell expansion ex vivo, which represents a selection for rapidly dividing cells, may increase the risk of spontaneous transformation.

Several recent reports indicate that MSCs can be transformed in response to a single experimentally induced oncogenic alteration. Transfection of human MSCs with a telomerase expression vector resulted in derivation of cells that were tumorigenic after 256 but not after 95 population doublings, and genetic alterations involving tumor suppressor and oncogene loci were detected in late-passage, tumorigenic cells (23). Similar results were reported for long-term cultures of human MSCs overexpressing telomerase (24). Mouse MSCs transfected with the EWS-FLI-1 oncogene formed tumors that displayed hallmarks of Ewing's sarcoma (25). However, these studies do not directly address the issue of spontaneous transformation.

Mouse MSCs may be particularly predisposed to spontaneous transformation because many differences exist with regard to mechanisms of immortalization and transformation in human and mouse cells (26). In contrast to the striking behavior of mouse MSCs, we have not observed transformation of rat or human MSCs when cultured under the same experimental conditions. However, a transformed clone of MSCs cultured from the mononuclear fraction of a single human bone marrow aspirate was reported recently that was characterized by short doubling time, loss of contact inhibition, abnormal karyotype, and tumors in multiple organs after i.v. or i.p. injection into nod/scid mice (27). Fifty percent of long-term cultures (4-5 months) of MSCs from human adipose tissue were found to undergo spontaneous transformation (28). In another study published while this article was in preparation, mouse bone marrow–derived MSCs were found to form tumors in vivo at P29 but not at P13, which was associated with a gradual increase in telomerase activity and C-MYC expression (29).

In contrast, we have shown that mouse MSCs undergo dramatic chromosomal aneuploidy within three passages, leading to increased rates of proliferation and loss of contact inhibition. The cells form invasive sarcomas when injected into nude or scid mice. Primary tumors from these mice can be transplanted into immunocompetent mice in which they continue to grow, although at a reduced rate relative to immunodeficient mice. The approximate doubling of chromosome number that we observed was not seen in other studies in which MSC transformation only occurred after extended passage (28, 29). The extent to which spontaneous transformation of MSCs and tumorigenesis are affected by species differences, ex vivo culture conditions, alterations in cancer regulatory genes, immune status of the recipient, or other factors remains to be determined. Thus, extensive characterization of genotypic and phenotypic alterations will be required to delineate the molecular pathogenesis and biological significance of the spontaneous transformation of mouse MSCs. Finally, the questions raised by these observations emphasize the need for careful analysis of human bone marrow cells expanded ex vivo for therapeutic administration and the importance of karyotype analysis and other quality control measures to guard against inadvertent administration of transformed cells to patients.

Grant support: The Johns Hopkins Institute for Cell Engineering and NIH grant RR00171.

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 Laura Morsberger and Constance Griffin for cytogenetic analyses and Chi Dang, Stephen Desiderio, Peter Donovan, Ian McNiece, and Bert Vogelstein for helpful discussions.