Mesenchymal stromal cells, also known as mesenchymal stem cells (MSCs), are multipotent progenitor cells existing in different tissues (1), such as bone marrow, adipose, umbilical cords and brain. In injured tissues, MSCs have the ability to migrate specifically into the damaged tissues and participate in tissue regeneration (2, 3). These stromal cells are known to differentiate into osteocytes, adipocytes, chondrocytes and fibroblasts. Since tumors are considered as “wounds that do not heal”, MSCs can also be recruited specifically into tumor tissues in response to “cues” from tumor cells and the tumor microenvironment (TME) (4).

In the TME of neoplasia, MSCs are one of the key stromal cells, which are reported to play complex and dynamic roles in the pathophysiology of solid tumors (5, 6). Interactions with tumor cells and other components in the TME cause “naïve” MSCs to undergo genetic and functional changes, giving rise to the generation of tumor-promoting cancer-associated MSCs (CA-MSCs) (7, 8). Under the tumor-mediated “education”, CA-MSCs have displayed outstanding roles in the progress of tumor initiation, promotion, progression or metastasis (9, 10).

Generally, CA-MSCs still retain the differentiation capacity and stromal surface markers of the counterpart MSCs, and use the similar mechanisms to support tumor progression (7, 11). Within the confines of the TME, CA-MSCs have been demonstrated to play key roles in the generation of tumor supporting niche (12), in the promotion of neovascularization (13), and in the stimulation of epithelial-mesenchymal transition (EMT) in tumor cells (14, 15). They also contribute to tumor pathophysiology via increasing the proliferation and survival of tumor cells (16), or enriching the population of cancer stem-like cells (CSCs) in the TME (17). However, CA-MSCs possess several unique properties and a more potent pro-tumorigenic function versus their parental MSCs (18, 19). For instance, CA-MSCs display a stronger ability to differentiate into carcinoma-associated fibroblasts (CAFs) than naïve MSCs (20). There is also evidence that CA-MSCs display higher immunosuppressive and angiogenic properties than the parental MSCs (19, 21). However, the origins and generating process of CA-MSCs are still less studied and have not yet been well characterized.

Hitherto, multiple reports have demonstrated that CA-MSCs can be picked up from malignant tissues of patients with several types of solid tumors, including gastric cancer (22), ovarian cancer (9), lung cancer (23), colorectal cancer (24), lymphoma (25) and prostate cancer (26). Generally, CA-MSCs could be separated out from tumor tissues by the following methods, including: (i) direct tissue piece adhesion (15, 22), (ii) tissue piece digestion by protease solution (23, 24), and (iii) immunomagnetic separation of the digested cells (25). The identifications of CA-MSCs depend on the morphology, immunophenotype and multilineage differentiation capacities of the isolated cells, which is similar to the counterpart MSCs. In the TME, CA-MSCs clearly influence the formation and context of TME and consequently play a pro-tumorigenic role in tumor growth and spread (27). Herein, we highlight the potential origins of CA-MSCs, and review the malignant transformation of naïve MSCs into CA-MSCs influenced by tumor cells or non-tumor cells in distinct modes.

MSCs are a heterogeneous population of progenitor cells, which play an active role in tumor progression (28). Upon sensing the “cues” from tumor, MSCs continuously migrate into local tumor sites and incorporate into the integral components of tumor stroma (7, 28). After affected by tumor cells and the surrounding microenvironment, the newly arrived naïve MSCs can be altered in their properties and converted into a pro-tumorigenic population termed as “CA-MSCs” (29). In the TME, CA-MSCs interact with the non-MSCs stromal populations and help to establish a favorable niche for tumorigenesis, metastasis, angiogenesis or drug resistance.

As increasingly demonstrated in the literature, CA-MSCs play a key role in tumor promotion via: (i) influencing the innate and adaptive immune systems to suppress immune response; (ii) promoting the survival and growth of tumor cells by producing growth factors, chemokines and cytokines; (iii) promoting tumor angiogenesis by secreting angiogenic factors or differentiating into endothelial cells; and (iv) promoting tumor cell metastasis by producing chemokines or enhancing EMT in tumor cells (30). However, accumulating evidence has indicated the multiple differences of CA-MSCs from the counterpart naïve MSCs in several aspects (18, 27), which are displayed in details as following:

In inflammatory diseases, MSCs have been unveiled to display tropism for the inflammatory sites and release a broad repertoire of soluble factors to modulate immune response (37). Likewise, similar tropism of MSCs is observed within various solid tumors in response to the “cues” from tumor tissues (38–40). After engaging into the TME, the recruited MSCs can “evolve” into CA-MSCs and participate in the formation of tumoral niche facilitating for tumorigenesis, spread or immune escape. In view of this, as the initial step of MSC-mediated tumor progression, the tropism or migration of MSCs toward tumor sites is a pivotal process and the two main origins have hitherto been discovered for CA-MSC storage in the TME.

After the initial stage of tumor development, the tropic naïve MSCs incorporate into tumor mass and their biological properties changed under the influences of tumor cells or non-tumor cells in the TME. Although sharing the similar cell surface markers and plasticity, CA-MSCs display different biology features from naïve MSCs throughout tumor progression. According to emerging evidences, the phenotypes and functions of naïve MSCs can be effectively modulated by tumor cells (8, 52), which give rise to the malignant transition of naïve MSCs into CA-MSCs with striking pro-tumoral property.

A study in prostate cancer reported that tumor cells are important in inducing the transdifferenciation of naïve MSCs into tumor-educated MSCs, which are different from naïve MSCs and can perform potent roles in vascular mimicry and monocyte recruitment (48). Another investigation also showed that naïve MSCs can be “educated” to have pro-metastatic behavior in response to the multiple signals from tumor cells (53). Additionally, a study identified the interactions between primary patient-derived cancer cells and stromal MSCs, and highlighted the specificity of MSC “education/reprogramming” by cancer cells in a tumor-specific manner to support tumor growth (51). In spite of this, understanding the cellular and molecular mechanisms underlying the malignant conversion of tumor tropic MSCs by tumor cells may supply innovative targets for anti-cancer therapy or prognosis in tumor patients.

Possessing important roles in tumor progression, the proportion of CA-MSCs has been demonstrated to dynamically increase throughout the distinct stages of tumor (64). Although tumor cells alone can partly “educate” the naïve MSCs to change their phenotypes and functions, the surrounding microenvironment have also emerged to play a dominant role in the malignant transition of naïve MSCs into tumoral MSCs (65, 66). In various solid tumors, the TME together with tumor cells have been demonstrated to participate in promoting MSC transition into CA-MSCs (67). Tumor cells only cause the recruited naïve MSCs to acquire the biological features of CA-MSCs partially (18, 33). Interactions with microenvironment cells or their factors give rise to both genetic and functional modulations in naïve MSCs (52, 68, 69), which can be effectively transformed into tumor-promoting CA-MSCs.

A study in lung cancer firstly confirmed the role of the TME components in inducing normal tissue-derived MSCs (N-MSCs) to acquire a tumor-associated MSC (T-MSC) expression profile (33). In particular, TGF-β1 and IL-6 generated by the TME components were shown to most likely contribute to the full feature establishment of T-MSCs, which in turn play an important role in tumor metastasis ( Table 2 ). Another report in lymphomas demonstrated that L-MSCs can strikingly enhance tumor growth by recruiting monocytes/macrophages (25), suggesting a critical role of monocytes/macrophages in converting the phenotypes and functions of naïve MSCs into CA-MSCs. In addition, TNF-α-pretreated BM-MSCs were shown to mimic L-MSCs in the chemokine profile and ability to promote tumorigenesis (25), suggesting the involvement of TNF-α from monocytes/macrophages in the malignant transition of MSCs ( Table 2 ). Likewise, a study in gastric cancer also confirmed the necessity of macrophages in the pro-tumor role of GC-MSCs in a mouse xenograft model with macrophage depletion (15). IL-6 and IL-8 were shown to participate in the interactions between GC-MSCs and macrophages ( Table 2 ). Another investigation in gastric cancer focused on tumor-educated neutrophils (TENs) and reported TEN-induced malignant transformation of MSCs by secreting the inflammatory factors including IL-17, IL-23 and TNF-α (70) ( Table 2 ).

In addition, other key factors in the TME such as hypoxia have also been demonstrated to drive the malignant transformation of naïve MSCs into tumor-promoting MSCs in various types of solid tumors (18, 66, 71). Therefore, the TME components together with their derived factors may have a crucial contribution to the “education” of naïve MSCs.

Notably, the communications between MSCs and non-tumor cells in the TME are complex and the mechanisms deserve to be further investigated in-depth for better understanding tumor progression. A study in gastric cancer demonstrated that miR-155-5p down-regulation induces a phenotypical and functional transition of BM-MSCs into GC-MSC-like cells depending on NF-κB p65 activation (72), suggesting a novel mechanism underlying the CA-MSC “remodeling”. Another report further demonstrated the partial involvement of miR-155-5p in the BM-MSCs “education” by gastric cancer cells (73). However, whether the NF-κB p65/miR-155-5p axis participates in the process of MSC transition, which is modulated by non-tumor cells or their derived factors, still needs to be further investigated. On the other hand, TGF-β secreted by tumor-educated non-tumor cells has been reported by various studies to be responsible for MSC transdifferenciation (45, 74). In spite of this, there is still insufficient information regarding the role of TGF-β signalling pathway in the mechanisms underlying MSC “education” by the non-tumor cells.

Collectively, although non-tumor cells in the TME have been confirmed to play a dominant role in the malignant transition of naïve MSCs, the related cellular and molecular mechanisms are specific and remain ambiguous, which is urgently needed to be illuminated for improving the existing anti-cancer treatments and prognosis.

In response to the “cues” from tumor, the recruited MSCs can acquire biological characteristics of CA-MSCs via a direct cell-to-cell contact, paracrine or exosome-mediated mode. Although the generating process of tumor-promoting CA-MSCs has not yet been clearly elucidated, multiple reports have demonstrated the specificities of MSC transition, by which naïve MSCs can be “re-programmed” (i) in a tumor type-dependent manner; (ii) with a relatively continued stable phenotype and (iii) with different kinetics throughout tumor progression. The specificities of CA-MSC generation further advance our understanding of kinetic changes on the genetic and phenotypic properties of tumor resident MSCs.

As one of the key stromal cells within tumor niche, CA-MSCs have been isolated from various types of solid tumors and identified to play important roles in tumorigenesis, metastasis and drug resistance (27, 76, 77). Compared with naïve MSCs, CA-MSCs display several unique phenotypes and more potent pro-tumorigenic functions, which are considered as a key regulator of tumor fate (77–79). Hitherto, the origins and generating progress of CA-MSCs still remain ambiguous.

In response to the “cues” from tumor cells or the TME, naïve MSCs are recruited from both the adjacent normal tissues (16) and the distant circulation (25) into tumor site. Moreover, the nearby non-cancerous stroma is positioned as the primarily source of naïve MSCs. Although the origins and tumor-tropic progress of MSCs have been clarified to some extent for our understanding, detailed mechanisms underlying the tropism and migration of MSCs towards tumor should be further investigated. After arrived into the TME, the phenotypes and functions of naïve MSCs are “shaped” by the tumor cells and other components in the microenvironment in a direct cell-to-cell contact, paracrine or exosome-mediated mode ( Figure 1 ) (52, 80). In addition, the conversion of naïve MSCs into CA-MSCs has been demonstrated to be a specific progression, which is tumor-type dependent, malignantly stable, and dynamic at distinct stages of tumorigenesis and progression (18, 33). However, more in-depth studies are still needed for discovering the generating process or transition of CA-MSCs in solid tumors, which may provide biomarkers or potential targets such as CA-MSC upstream or downstream modulators, for cancer prognosis and treatments.

As one of the key components in tumor stroma, CA-MSCs display a potent role in the development and progression of solid tumors, such as gastric cancer, ovarian cancer and lung cancer. Understandings of the origins and generation of CA-MSCs will facilitate the development of novel targets for anti-cancer treatments and prognosis. After tropism into tumor site from the circulation or adjacent normal tissues, the infiltrated naïve MSCs can be “educated” into CA-MSCs by tumor cells and other components in the TME. However, more efforts still need to be devoted to better elucidate the cellular and molecular mechanisms underlying MSC migration and malignant transition, which may provide more promising targets for tumor therapy and prognosis

WL, JY, PZ and HL collected the related paper and drafted the manuscript. WL and SZ participated in the design of the review. All authors contributed to the article and approved the submitted version.

This review was supported by the National Natural Science Foundation of China (Grant No. 81402280), the Foundation of Open Project from Jiangsu Key Laboratory (Grant No. XZSYSKF20200001), the Jiangsu Postdoctoral Research Foundation (Grant No. 1501079A) and the Doctoral Foundation from the First People’s Hospital of Lianyungang (Grant No. BS1503).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.