Mesenchymal stem cell transformation and sarcoma genesis

Mouse MSCs have been consistently demonstrated to spontaneously undergo tumorigenic transformation after long term ex vivo culture [4, 6]. This transformation process can also be induced by certain manipulations, including both gene targeting and drug or chemical treatment to affect crucial pathways (Table 1) [11‐14]. In contrast, human MSCs do not spontaneously transform in vitro, even after long term culturing, which will be discussed later in more detail [9, 13, 15].

Mouse MCSs are reported to spontaneously undergo changes in morphology, proliferation rate, migration ability, cell surface marker profile, genomic constitution and most importantly tumorigenicity after long term in vitro culture [4, 9, 20, 21]. Meanwhile, one study has also revealed that mMSCs could transform even after short term in vitro culture. Injection of passage 3 mMSCs into mice resulted in formation of tumours comparable with soft tissue sarcomas [19]. Transformed mouse MSCs always show a higher proliferation rate than the native cells [3, 4]. These transformed cells exhibit tumorigenicity, as shown by anchorage-independent growth assay and xeno-transplantation in mice and zebrafish, while this is not observed with low passage mouse MSCs before their transformation [3, 4, 6, 22]. Interestingly, the readiness of in vitro tumorigenic transformation seems to be a unique property of mouse MSCs since it is absent in most other mouse stem cells, including hematopoietic stem cells and embryonic stem cells [23]. This readiness can be probably ascribed to the genetic instability already shown in mouse MSCs very shortly after isolation from bone marrow, although the cytogenetic abnormalities in low passage mouse MSCs are considerably less in number than in transformed mouse MSCs [23]. Interestingly, MSC spontaneous transformation happens much less frequently in vivo, as shown by the low incidence of spontaneous sarcomagenesis in mice. This can be possibly explained by the different microenvironment of in vitro and in vivo conditions for MSCs. Solid research on the role of the in vivo niche of BMMSCs in guarding its genomic stability is needed to answer this question more exactly.

Transformation of mouse MSCs has been induced by an array of manipulations, including knockout of tumour suppressor genes, overexpression of oncogenes and drug administration to affect signaling pathways. The pathways targeted by these manipulations are mostly involved in cell cycle checkpoint control, cell survival, proliferation and apoptosis (Table 2) [14]. In one study, loss of tumour suppressor P21 and Tp53 in mouse adipose derived MSCs (AMSC) induced in vitro transformation and in vivo so-called fibrosarcoma formation after transplantation [24]. In another study, both Tp53−/− Rb−/− and Tp53−/− mouse AMSCs were generated through Cre mediated excision of loxP flanked loci. Leiomyosarcoma-like tumours were developed in the in vivo tumorigenicity assays of these 2 types of mouse AMSCs [8]. The combination of Cdkn2a loss and C-myc overexpression in mouse BMMSCs gave rise to osteosarcomas accompanied by the loss of adipogenic differentiation capacity in transformed mouse BMMSCs [16]. Besides directly targeting in vitro cultured MSCs, several genetically engineered mouse models have been developed to investigate the effects of genes on transformation process. A conditional mouse model with Tp53 homozygous deletion has been created by crossing Prx1-Cre transgenic mice to mice bearing alleles of Tp53 flanked by loxP. Prx1 is specifically expressed in the early mesenchymal tissues of embryonic limb buds [17]. In these P53-deficient mice many types of sarcomas occurred in the mesenchymal cells of limb buds and osteosarcoma was the most common type. A mouse model with loss of RB generated also through Cre-loxP system was not found to display tumorigenesis. However, loss of RB accelerated tumorigenesis in P53-deficient mice [17]. These induced transformation studies established the importance of the P53 pathway in preventing mouse MSC transformation. Besides, in spontaneous transformation studies of mouse MSCs, defects in Tp53 or Cdkn2a genes were frequently found [18]. P53 and P14, proteins encoded by these two genes, are both important members of P53 pathway, further corroborating the crucial role of P53 pathway in mouse MSC transformation [4, 25]. Upregulated oncogenic pathways have also been shown to induce or potentiate mouse MSC transformation. Fos is an oncogene encoding a transcription factor downstream of many growth factor pathways. The Fos overexpression transgenic mice resulted in the development of bone tumours, with chondrosarcomas as the main type [26]. This is puzzling as the driver mutation in human central chondrosarcoma is IDH1 or IDH2 [27], while in peripheral chondrosarcomas this is not known [27, 28], but no indication for involvement of Fos is found [28, 29]. The PI3K-AKT pathway is crucially involved in apoptosis and proliferation. In a study a mouse model with homozygous loss of Pten, a negative regulator of the PI3K-AKT pathway in smooth muscle lineage cells developed leiomyosarcomas [30, 31]. The MAPK pathway is principally responsible for mitosis. Overexpression of K-ras, a component of the MARK pathway in addition to P53 loss induced sarcoma formation in mice more efficiently than in mice with P53 loss alone [32, 33]. These studies underscore the role of different oncogenic pathways to promote mouse MSC transformation.

Human MSCs have not been shown to undergo spontaneous transformation in vitro[9, 15, 43]. There have been few reports on spontaneous human MSC in vitro transformation, of which two turned out to be caused by contamination by tumour cell lines and were retracted afterwards [34, 35, 44]. Meanwhile, there are several studies demonstrating that human MSCs did not go through transformation in spite of long term in vitro culturing [12, 15]. For the possibility of in vivo spontaneous transformation, there have been few cases of osteosarcoma genesis in patients infused with bone marrow MSCs for other diseases [45‐47]. The majority studies of human MSC transformation are based on genetic approaches to knock out important tumour suppressor genes and overexpress certain oncogenes Table 3) [14]. In contrast to mouse MSC studies, four of the induced human MSC transformation studies consist of the exogenous expression of hTERT in human cells [38, 48‐50]. This may be attributed to the much shorter telomeres in human MSCs than their mouse counterparts, the much shorter life span of mice than human and the difference in telomere damage signaling pathways between mouse and human [41, 50, 51]. Consistent with mouse MSC studies, the disruption of cell cycle control machineries, exemplified by P53 and RB pathways are also important for human MSC transformation. For instance, the introduction of SV40-LT, which perturbs both P53 and RB proteins potently promoted human MSC transformation [38]. Furthermore, the overexpression of some oncogenes has also been shown to contribute to the transformation, such as H-RAS[5‐7]. Although the definite spontaneous transformation capacity of mouse MSCs is not a mimicry of human MSCs, the signaling pathways underlying their tumorigenic transformation show high consistency, including the P53 pathway, RB pathway, PI3K-AKT pathway and MAPK pathway and so on.