Mesenchymal stromal cells as cancer promoters.

Mesenchymal stromal cells (MSCs) are multipotent stromal progenitors that can be isolated from various tissues, including bone marrow (BM) [ 1 , 2 ], fat [ 3 ], placenta [ 4 ], and umbilical cord blood [ 5 ]. Evidence that MSCs migrate to tumor sites and contribute to tumor stroma has led to an evaluation of their role in carcinogenesis and tumor progression. Chronic inflammatory stimuli from tumor sites may play a role in the active recruitment of MSCs [ 6 , 7 ] which then modulate tumor growth through cellular contact, release of paracrine factors, and the transfer of biological material via extracellular vesicles (EVs) [ 8 – 13 ]. Across a broad range of tumors, MSCs can promote cancer hallmarks such as increased proliferation, cancer stemness, drug resistance, epithelial to mesenchymal transition (EMT), invasion and metastasis, immune evasion, angiogenesis, and resistance to apoptosis (Fig. 1 ). Fig. 1 MSCs promote cancer hallmarks in various tumors. Across a broad range of cancer cell lines and primary patient-derived tumors, MSCs have been shown to promote cancer stemness, drug resistance, proliferation, invasion and metastasis, immune evasion, angiogenesis, epithelial to mesenchymal transition, and resistance to apoptosis. These cancer hallmarks collectively constitute the pro-tumorigenic features of MSCs. Figure created with biorender.com . Across a broad range of cancer cell lines and primary patient-derived tumors, MSCs have been shown to promote cancer stemness, drug resistance, proliferation, invasion and metastasis, immune evasion, angiogenesis, epithelial to mesenchymal transition, and resistance to apoptosis. These cancer hallmarks collectively constitute the pro-tumorigenic features of MSCs. Figure created with biorender.com . Importantly, MSCs do not always support tumor growth. For example, in vitro co-culture experiments and in vivo co-injection of MSCs with soft tissue sarcomas did not increase cell proliferation, tumor size, metastasis, or recurrence following radiation therapy and surgery, suggesting that MSCs may be safely used in cancer patients [ 14 ]. Under certain circumstances, MSCs may in fact impede tumor growth and they have been shown to inhibit the growth of leukemia, glioma, hepatocellular carcinoma, breast cancer, and pancreatic ductal adenocarcinoma through various mechanisms [ 9 – 13 , 15 – 22 ]. For example, direct co-culture of MSCs with cancer cells appears to inhibit tumor growth while indirect contact through the release of soluble factors tends to promote tumor growth [ 20 , 21 ]. MSCs have also been shown to induce cancer cell apoptosis by downregulating the PI3K/AKT pathway and to suppress tumor growth by inhibiting angiogenesis and releasing soluble ligands such as TRAIL and TGFβ [ 21 , 22 ]. Interestingly, the mutational status of MSCs may influence their anti-tumorigenic properties – for example, tumors containing p53-deficient MSCs grew larger than tumors containing MSCs with wild-type p53 [ 23 ]. The anti-cancer properties of MSCs have inspired numerous clinical trials investigating the therapeutic potential of native and engineered MSCs in the treatment of cancer patients [ 24 ]. While this topic is important, it has been reviewed extensively elsewhere [ 22 , 24 , 25 ] and is outside the scope of this review. Here, we aim to highlight the various mechanisms by which human MSCs can promote tumor growth, survival, and metastasis, and to identify areas that merit further research. MSCs appear to support the growth of various cancers (Fig. 1 ). Co-incubation of MSCs with head and neck cancer cells enhanced cancer cell proliferation and chemotherapy resistance [ 26 ]. In hepatocellular carcinoma, MSCs were shown to accelerate tumor growth and metastasis [ 27 ]. Similarly, adipose-derived MSCs enhanced ovarian cancer cell proliferation, and this effect appeared to depend on MSC-mediated increased expression of the transcription factor PAX8 in ovarian cancer cells [ 28 ]. Human bone marrow MSCs (BM-MSCs) release signals that promote growth and angiogenesis and that inhibit apoptosis and anti-tumor immunity (Fig. 2 ). Growth factors released from BM-MSCs include fibroblast growth factor (FGF), hepatocyte growth factor (HGF), monocyte chemoattractant protein-1 (MCP-1), and interleukin-6 (IL-6); while pro-angiogenic molecules released from BM-MSCs include b-FGF, interleukin 8 (IL-8), and vascular endothelial growth factor (VEGF) [ 29 , 30 ]. BM-MSCs can also inhibit apoptosis through the release of Trail decoy receptors and can suppress anti-tumor immunity by secreting IL-12p40 and soluble IL-2 receptor α [ 29 , 30 ]. In addition, human MSCs express matrix metalloproteinases which can promote tumor invasiveness through the degradation of the extracellular matrix (ECM) and matrix remodeling [ 29 – 32 ]. Fig. 2 Mechanisms of MSC-mediated pro-tumorigenic signaling. MSCs can release a host of soluble ligands that promote phenotypes such as tumor growth, angiogenesis, extracellular matrix degradation, tumor invasion, cancer stem cell properties, and immune evasion. MSCs can also release decoy receptors which shield cancer cells from anti-tumor signaling, ultimately conferring resistance to apoptosis and evasion of immune-mediated clearance. MCP-1 monocyte chemoattractant protein-1, FGF fibroblast growth factor, HGF hepatocyte growth factor, IL- interleukin, VEGF vascular endothelial growth factor, MMP matrix metalloproteinase, CCL5 chemokine ligand 5, PDGF platelet-derived growth factor, TRAIL TNF-related apoptosis inducing ligand. Figure created with biorender.com. MSCs can release a host of soluble ligands that promote phenotypes such as tumor growth, angiogenesis, extracellular matrix degradation, tumor invasion, cancer stem cell properties, and immune evasion. MSCs can also release decoy receptors which shield cancer cells from anti-tumor signaling, ultimately conferring resistance to apoptosis and evasion of immune-mediated clearance. MCP-1 monocyte chemoattractant protein-1, FGF fibroblast growth factor, HGF hepatocyte growth factor, IL- interleukin, VEGF vascular endothelial growth factor, MMP matrix metalloproteinase, CCL5 chemokine ligand 5, PDGF platelet-derived growth factor, TRAIL TNF-related apoptosis inducing ligand. Figure created with biorender.com. Release of IL-6 from MSCs is a particularly important mechanism by which MSCs promote the progression of various cancers. Co-culture experiments demonstrated that IL-6 from BM-MSCs stimulates the proliferation of estrogen-receptor α (ERα) positive breast cancer cells through activation of the STAT3 pathway [ 15 , 33 ]. Similarly, IL-6 from BM-MSCs activates the Erk 1/2 pathway and enhances proliferation of neuroblastoma cells [ 34 ]. In colorectal cancer, IL-6 from BM-MSCs supports angiogenesis by stimulating the release of endothelin-1 (ET-1) from colorectal cancer cells, leading to Akt and ERK activation in endothelial cells and ultimately enhancing their recruitment and angiogenic capacity [ 35 ]. IL-6 secreted by MSCs also enhanced the growth of the ovarian cancer cell line, SKOV3 [ 36 ]. Using a model of human BM-MSCs and pancreatic cancer cell lines, our group demonstrated that BM-MSCs promoted tumor growth in vitro, and in heterotopic pancreatic cancer xenografts in nude mice, through MSC-derived IL-6 and activation of the STAT3 signaling pathway [ 37 ]. We showed that knockdown of IL-6 in MSCs by siRNA, or inhibition of IL-6 signaling by blocking IL-6 receptors with tocilizumab, or its downstream effector, STAT3, with a small molecule inhibitor, largely inhibited MSC mediated-tumor promotion and delayed tumor formation in vitro and in vivo [ 37 ]. Extracellular vesicles (EVs) from MSCs may also enhance tumor growth. EVs from BM-MSCs promoted the growth of gastric and colon cancer in mice by activating ERK1/2 and p38 MAPK signaling and upregulating VEGF [ 38 ]. MSC-derived EVs have also been shown to promote the growth of osteosarcoma, gastric cancer, and breast cancer cell in a hedgehog-dependent manner [ 39 , 40 ]. The growth-promoting effects of MSC EVs appear to depend on the state of MSCs and the cargo carried by these EVs. For example, one group demonstrated that EVs from MSCs that were pre-challenged with hypoxia promoted the proliferation, survival, invasiveness and EMT of lung cancer cell lines and enhanced macrophage M2 polarization [ 41 ]. These effects were believed to be driven by EV-mediated transfer of miR-21-5p [ 41 ]. Similarly, another group noted that EV-mediated transfer of miR-193a-3p, miR-210-3p and miR-5100 from hypoxic BM-MSCs to lung cancer cells could promote invasion of lung cancer cells by activating STAT3 signaling-induced EMT [ 42 ]. MSC EV-mediated transfer of miR-410 to lung adenocarcinoma cells was also reportedly associated with reduced PTEN expression and increased tumor growth in a xenograft model [ 43 ]. In addition to miRNAs, EVs may also transfer proteins from MSCs to cancer cells. When MSCs were cultured with hepatocellular carcinoma cells (HCC), MSC EVs appeared to transfer Transmembrane BAX Inhibitor Motif Containing 6 (TMBIM6) to HCC cells, resulting in increased proliferation, invasion, and sphere formation, and inhibition of apoptosis [ 44 ]. In AML, BM-MSCs are important components of the bone marrow microenvironment and EVs from BM-MSCs appear to increase resistance to the chemotherapeutic agent cytarabine [ 45 ]. Given the reported increased levels of miR-10a in BM-MSCs and, after co-culture, in AML cells, it is possible that this increased chemotherapy resistance is mediated by EV transfer of miR-10a from BM-MSCs to AML cells [ 45 ]. MSCs can also confer tumor-initiating potential and support cancer stem cells (Fig. 2 ). For example, human MSCs can stimulate cancer cells to acquire stem cell-like characteristics, via paracrine signaling involving IL-6, IL-8 and CCL5 [ 46 – 49 ]. MSCs introduced into the tibia of NOD/SCID mice traveled to breast tumor xenografts and increased the breast cancer stem cell population through cytokine loops involving IL-6 and CXCL7 [ 46 ]. Similarly, human BM-MSCs promoted cancer stem cell features of lung cancer via the JAK2/STAT3 pathway [ 47 ] and in colorectal cancer through secretion of IL-6 and activation of the STAT3 pathway [ 49 ]. BM-MSCs also promoted stemness features in hypopharyngeal cancer cells, increasing their clonogenic and sphere-forming capacities [ 50 ]. In the setting of B cell precursor-acute lymphoblastic leukemia cultured long-term with MSCs, leukemia cells acquired phenotypic changes that resemble leukemia stem cells, such as upregulation of CD34 and downregulation of CD38 [ 51 ]. Another group demonstrated that PDGF release from ovarian carcinoma-associated MSCs significantly increased stemness, metastatic potential, and chemoresistance of ovarian cancer stem cells [ 52 ]. Adipose-derived MSCs were also shown to activate NF-kappa B signaling in cervical cancer cells and to confer EMT and stemness properties [ 53 ]. Non-coding RNAs have been shown to play important roles in the regulation of normal and cancer stem cells and one group found that in response to MSC-secreted TGF-β1, the long non-coding RNA (lncRNA) MACC1-AS1 is induced and antagonizes the tumor suppressive microRNA miR-145 in gastric cancer cells, ultimately resulting in fatty acid oxidation-mediated stemness and chemoresistance [ 54 ]. HCP5 is another lncRNA that is upregulated in gastric cancer cells after co-culture with MSCs and it was shown to drive fatty acid oxidation, stemness, and chemoresistance by sequestering miR-3619 [ 55 ]. Further supporting a role for MSCs in promoting gastric cancer stemness, co-culture experiments demonstrated that MSCs induced an upregulation of natriuretic peptide receptor A (NPRA) in gastric cancer cells, and this consequently facilitated cancer stemness and chemoresistance through fatty acid oxidation [ 56 ]. Cancer cells may play a role in inducing these stem cell-supporting properties of cancer. For example, gastric cancer cells reprogrammed naïve MSCs to support cancer stemness by activating the R-spondin/Lgr5 axis and WNT/β-catenin signaling pathway [ 57 ]. The migration of MSCs to tumor sites facilitates their tumor-supporting functions and resembles the normal process of MSC mobilization and homing to damaged tissue [ 58 ]. In nude mice, a significant increase in MSC homing was observed in response to tissue injury induced by total body irradiation (TBI) with or without additional local irradiation [ 58 ]. Tumors are sites of chronic inflammation or “wounds that do not heal” and their inflammatory signals appear to be involved in the active recruitment of MSCs [ 6 , 7 , 59 ]. Inflammatory signals can enhance MSC migration by upregulating surface expression of chemokine receptors [ 60 , 61 ]. Tumor-necrosis factor alpha (TNF-α) upregulates the chemokine receptors CCR2, CCR3 and CCR4 on MSCs and ultimately enhances their migration toward chemokines [ 60 ]. TNF-α and IL-1β further facilitate MSC extravasation across endothelial tissue in a VCAM-1 dependent manner [ 62 , 63 ]. In murine human colon adenocarcinoma xenograft models, TNF-α activated NF-κB signaling and induced VCAM-1 expression on MSCs, thereby facilitating the accumulation of MSCs at tumor sites [ 64 ]. Growth factors and chemokines can also promote MSC migration to various other tumors [ 65 – 68 ]. For example, stromal-derived factor 1 alpha (SDF-1α) induced BM-MSC migration towards osteosarcoma [ 68 ], colon cancer [ 69 ], and breast cancer cells [ 70 ]. CCL2, IL-6, and TGF-β stimulated the migration of BM-MSCs towards breast carcinomas [ 46 , 71 – 73 ] while VEGF promoted BM-MSC migration towards glioma [ 74 ] and pancreatic adenocarcinoma [ 75 ]. CCL25 is an important chemoattractant for MSCs towards multiple myeloma (MM), and the knockdown of the CCL25 receptor CCR9 in human BM-MSCs significantly reduced MSC tropism towards MM cells in vitro [ 76 ]. Interestingly, highly aggressive and metastatic phenotypes show a greater ability to attract MSCs [ 71 ]. For example, advanced renal cell carcinomas were shown to induce MSC migration via the secretion of amphiregulin, fibronectin, and hyaluronic acid [ 77 ]. MSCs that migrate to tumor sites can further promote tumor growth. While it is well-established that MSCs can be found in the stroma of many tumors, MSCs do not home to all tumors and it is difficult to discern whether cancer-associated MSCs have migrated from the bone marrow to tumor sites or if they instead arise from adjacent normal tissue-resident MSCs. An important criticism of the model in which bone marrow MSCs migrate through the circulation to reach tumor sites is that circulating MSCs are exceedingly rare under normal conditions [ 78 ]. Adjacent tissue-resident MSCs are therefore believed to be at least an important, or possibly the primary, source of cancer-associated MSCs [ 79 ]. Cancer-associated MSCs (CA-MSCs), which are MSCs found in tumor stroma, possess unique pro-tumorigenic features and functional characteristics that are not shared by MSCs derived from healthy tissues [ 80 – 86 ] (Fig. 3 ). Co-implantation of ovarian cancer cells with CA-MSCs in mice promoted tumor xenograft growth more effectively than co-implantation with control MSCs. Similarly, breast cancer-derived CA-MSCs significantly enhanced in vitro mammosphere formation and the in vivo growth of tumors from the breast cancer cell line, MCF7, compared with normal tissue MSCs; this was mediated partially via the EGF/EGFR/Akt pathway [ 83 ]. Several studies also showed that gastric carcinoma-derived CA-MSCs support the growth of gastric cancer cells through HGF secretion, activation of c-MET signaling, production of IL-8, and the release of EVs containing miR-22 which has been linked to invasion and metastasis [ 84 , 85 , 87 ]. Gastric cancer-derived MSCs were also shown to promote immune evasion and chemotherapy resistance by upregulating the immune inhibitory ligand programmed death-ligand 1 (PD-L1), the DNA repair protein Rad51, and the drug resistance protein multi-drug resistance 1 (MDR1) in gastric cancer cells. Similarly, AML cells were shown to physically interact with MSCs and to induce a reprogrammed transcriptome that facilitates aberrant cell proliferation and differentiation and severely compromises their immunomodulatory capability [ 88 ]. Interestingly, selective inhibition of AML-associated MSCs with a selective CaV1.2 channel blocker drug impaired leukemia progression in vitro and in vivo [ 88 ]. AML-associated MSCs appear to also promote drug resistance and post-chemotherapy relapse. Interactions between fibronectin on BM-MSCs and VLA-4 expressed on leukemic cells activates PI3K/AKT/Bcl-2 signaling and results in the acquisition of drug resistance [ 89 ]. Even after chemotherapy clearance of the bulk of AML cells, a subset of BM-MSCs, namely Nestin+ MSCs, persist in the bone marrow niche and further support the survival and post-chemotherapy relapse of AML through increased oxidative phosphorylation, tricarboxylic acid cycle activity, and glutathione-mediated antioxidant defense [ 90 ]. Fig. 3 MSC-derived CAFs release pro-tumorigenic signals. TGF-β signaling mediates the conversion of MSCs to CAF-like MSCs and progression to MSC-derived cancer associated fibroblasts which subsequently release pro-tumorigenic signaling ligands. a-SMA alpha-smooth muscle actin, TGF-β transforming growth factor-β, CAF cancer-associated fibroblasts, MMP matrix metalloproteinase, VEGF vascular endothelial growth factor, ICAM-1 intercellular adhesion molecule 1, HGF hepatocyte growth factor. Figure created with biorender.com. TGF-β signaling mediates the conversion of MSCs to CAF-like MSCs and progression to MSC-derived cancer associated fibroblasts which subsequently release pro-tumorigenic signaling ligands. a-SMA alpha-smooth muscle actin, TGF-β transforming growth factor-β, CAF cancer-associated fibroblasts, MMP matrix metalloproteinase, VEGF vascular endothelial growth factor, ICAM-1 intercellular adhesion molecule 1, HGF hepatocyte growth factor. Figure created with biorender.com. Cancer-supporting phenotypes of MSCs may arise in a premalignant niche. Endometriosis is an example of a premalignant condition which is believed to be a precursor of ovarian cancer, especially clear cell carcinoma (OCCC) and endometrial carcinoma [ 91 ]. A subset of endometriosis-derived MSCs (enMSC), characterized by loss of CD10 expression, was found to specifically support OCCC growth by donating growth-promoting iron, via their elevated levels of the iron export proteins hephaestin and ferroportin, to OCCCs [ 92 ]. An interesting observation of this study was that OCCC cells appear to be particularly vulnerable to the iron-dependent form of cell death known as ferroptosis [ 92 ]. Additionally, CA-MSCs display superior immunosuppressive capabilities compared with normal tissue MSCs [ 93 – 96 ]. MSCs derived from human breast cancer tissue have higher levels of IL-10 and TGF-β production and greater potential to increase the numbers of regulatory T cells compared with MSCs from normal breast tissue [ 95 ]. Similarly, MSCs isolated from human cervical tumors significantly inhibited T cell recognition of cervical cancer cells by cytotoxic T lymphocytes (CTLs) through IL-10 - mediated downregulation of human leukocyte antigen (HLA) class I molecules on cancer cells [ 96 ]. Mouse cancer models highlight major differences in the immunomodulatory capabilities between CA-MSCs and normal tissue MSCs, including their ability to control monocyte/macrophage trafficking and function in vivo [ 93 , 94 ]. MSCs isolated from pancreatic tumors expressed significantly higher levels of immunomodulatory factors and exhibited increased tumor-promoting abilities [ 94 ]. Similarly, MSCs derived from murine spontaneous lymphoma tumors supported tumor growth through CCL2-mediated recruitment of monocytes and M2 macrophages to the TME. This effect was mimicked by priming BM-MSCs with TNF-α, further supporting the role of inflammation in modulating MSC activity in the TME. Although the mechanisms by which cancer cells may reprogram MSCs towards tumor-supporting phenotypes are yet to be fully elucidated, one group demonstrated that ovarian cancer cells can induce MSCs to undergo a mesenchymal-to-epithelial transition (MET). MET in ovarian cancer-associated MSCs is mediated by WT1 and EZH2 and is characterized by enhancer-enriched DNA hypermethylation, altered chromatin accessibility, and differential histone modifications [ 97 ]. Another interesting recent observation is that MSCs may fuse with cancer cells, including glioma cells and breast cancer cells in two separate studies, and that these hybrid cells can increase tumor growth, angiogenesis, treatment resistance, and metastasis [ 98 , 99 ]. The formation of breast cancer-MSC hybrid cells is driven by WNT5A signaling and depends on breast cancer cell-derived IL-6 and MSC-derived CCL2 [ 99 ]. Tumor resident MSCs may be a potential cellular source of cancer associated fibroblasts (CAFs) (Fig. 3 ) [ 100 ]. Unlike normal stroma, the reactive stroma of tumors is rich in vascular supply, collagen, fibrin deposition, and actively proliferating fibroblasts [ 101 – 103 ]. CAFs are key members of the tumor stroma, providing structural and functional support to tumor cells [ 101 , 104 – 107 ]. CAFs support tumor development, growth, and metastasis through the release of pro-tumorigenic growth factors, chemokines, and matrix metalloproteinases (MMPs) [ 101 , 102 , 108 ]. In addition, CAFs enhance cancer cell stemness by activation of the Wnt and Notch signaling pathways [ 109 ]. CAFs have a distinct morphology, and a characteristic surface marker profile including α-SMA, fibroblast activation protein (FAP), Thy-1, desmin, and S100A4 protein [ 110 ]. CAFs also tend to have higher proliferation rates and produce higher levels of VEGF and immunosuppressive factors than MSCs [ 100 ]. CAF-like MSCs appear to be an intermediate subset of MSCs that share features of MSCs and CAFs and may be precursors to CAFs. CAF-like MSCs have higher α-SMA expression than naïve MSCs [ 111 ], while, unlike CAFs, also expressing MSC cellular markers such as CD44, CD90, and CD105 [ 76 , 111 ]. Through EMT, cancer cells can transform into myoepithelial cells and transdifferentiate into CAFs to generate their own stroma [ 103 , 112 , 113 ]. It appears that MSCs can also serve as a source of cancer-associated myofibroblasts. Convincing evidence of the transformation of bone marrow-derived cells to CAFs came from a study which identified bone marrow-derived cancer-associated myofibroblasts in rectal and gastric tumors following allogeneic bone marrow transplantation [ 114 ]. The recruitment of BM-derived myofibroblasts to cancer stroma is additionally supported by evidence from syngeneic tumor mouse models [ 115 , 116 ]. Using BM reconstitution studies in a mouse model of inflammation-induced gastric cancer, Quante et al. showed that 20% of α-SMA positive myofibroblasts in gastric tumor dysplasia are most likely bone marrow-derived MSCs. Similarly, in a xenograft model of human pancreatic cancer, 40% of the CAF population in tumor stroma originated from mouse α-SMA positive bone-marrow-derived cells [ 117 ]. Following BM transplantation in transgenic mice, a BM-derived subpopulation of α-SMA positive CAFs was identified in the stroma of primary and metastatic mammary tumors, which was distinguishable from resident CAFs by the lack of PDGFRα expression [ 118 ]. These BM-derived CAFs did not co-express CD45, which supported their mesenchymal origin [ 118 ]. Similarly, CAFs expressing αFAP- and fibroblast specific protein (FSP-1) isolated from primary neuroblastoma cells exhibited phenotypic and functional characteristics of BM-MSCs [ 119 ]. Further support for the notion that MSCs can differentiate into CAFs came from a 2021 study in which pancreatic ductal adenocarcinoma cells were co-transplanted with adipose-derived MSCs and subsequently, single cell sequencing of CAFs from this xenograft model revealed that they were derived from the transplanted MSCs [ 120 ]. Another group fluorescently labeled BM-MSCs that were transplanted into the subserosal layers of the stomach in a mouse model of chronic H pylori infection and found that these MSCs migrated to the mucosal layer and promoted gastric cancer progression by differentiating into both gastric epithelial cells and CAFs [ 121 ]. MSCs cultured with colon cancer cells also differentiated into CAFs that subsequently released ICAM-1 and promoted the proliferation, migration, and invasion of colon cancer cells via STAT3 and AKT signaling [ 122 ]. EVs may have a role in the transformation of MSCs into CAF-like cells. EVs from ovarian cancer cells were shown to induce the transformation of adipose-tissue-MSCs into a CAF phenotype primarily through TGF-β receptor upregulation and downstream activation of SMAD-2 signaling [ 123 ]. Consistent with these results, TGF-β-bearing EVs derived from prostate cancer cells, but not soluble TGF-β, stimulated the differentiation of MSCs into tumor-promoting myofibroblasts capable of secreting higher levels of VEGF, HGF and MPPs [ 123 ]. Whether this effect is indeed mediated by the delivery of EV cargo, rather than receptor-mediated signaling remains unclear, however, with one group reporting that exposure of BM-MSCs to soluble TGF-β upregulates the expression of CAF markers in BM-MSCs [ 124 ]. Besides being a potential CAF source, MSCs can also actively modulate the tumor immune microenvironment. MSCs interact with immune cells in the TME and can suppress T-cell and B cell function, inhibit antigen-presenting cell maturation, suppress natural killer (NK) cell proliferation and cytotoxicity, and induce regulatory T cells (Treg) generation [ 125 – 132 ]. Factors such as TGF-β, HGF [ 133 ], IL-2, and IL-10 mediate many of the immune suppressive effects of MSCs [ 125 ] (Fig. 2 ). MSCs also induce indoleamine 2,3 dioxygenase (IDO), which reduces T-cell proliferation [ 134 ], and secrete nitric oxide (NO) in response to pro-inflammatory cytokines in mice [ 128 , 134 – 137 ]. MSCs were also shown to induce the expression of the immune suppressing programmed death ligand 1 (PD-L1) on breast cancer cells through the secretion of CCL5 [ 138 ]. Thus, by suppressing both innate and adaptive anti-tumor immune responses, MSCs can create a tumor permissive microenvironment that allows cancer cells to evade immune detection. In addition to inducing Tregs and inhibiting NK cell function, MSCs can further suppress anti-tumor immune responses by inducing immunosuppressive M2 macrophage polarization [ 95 , 139 – 141 ]. Gastric cancer-derived MSCs promote M2 macrophage polarization through substantial secretion of IL-6 and IL-8 [ 133 ]. Chemokine secretion, including secretion of Ccl2 [ 134 ], Cx3cl1, and Tgf-β1 from ovarian cancer-associated MSCs recruited immune-suppressive CD14+Ly6C+Cx3cr1+ monocytic cells and polarized macrophages to an immune suppressive Ccr2hiF4/80+Cx3cr1+CD206+ phenotype [ 142 ]. Breast cancer-associated MSC exosomes were also reported to drive M2 macrophage polarization [ 143 ]. Treg induction by MSCs further suppresses anti-tumor immunity. B16 melanoma tumors formed in allogeneic recipient mice only when murine BM-MSCs were co-injected with the tumor cells partly due to the generation of CD8+ Tregs [ 139 ]. Similarly, in co-culture experiments with peripheral blood mononuclear cells and breast cancer cell lines, MSCs released TGF-β, IL-10, and IL-4 which helped induce Tregs and reduced cytotoxic T lymphocytes (CTLs) and NK cell-mediated cell-killing [ 140 ]. In vivo, MSC-mediated Treg induction and upregulation of anti-inflammatory mediators enhanced tumor growth and metastasis when human peripheral blood derived MSCs were systemically co-administered with murine mammary carcinoma in an immunocompetent mouse model [ 141 ]. NK cells serve as important anti-tumor effector cells and their function can also be inhibited by certain MSCs. Gastric cancer-derived MSCs were shown to inhibit the degranulation capacity, perforin production, and cytotoxicity of NK cells by upregulating fructose-1,6-bisphosphatase (FBP1) expression and consequently impairing NK cell glycolysis and viability [ 144 , 145 ]. Gastric cancer-derived MSCs were further shown to inhibit mTOR signaling in NK cells leading to NK cell dysfunction [ 146 ]. Pro-inflammatory mediators also play a role in MSC-mediated immune suppression. Syngeneic mouse models suggest that inflammation can promote the immunosuppressive properties of MSCs. Studies of mouse MSCs show that inflammatory molecules can improve cell-to-cell contact between MSCs and T-cells through induction of ICAM-1 and VCAM-1 on the surface of MSCs, thereby facilitating MSC-induced suppression of T cell function [ 147 ]. In addition to increasing cell-to-cell contact, another mechanism by which inflammation can promote MSC-based immunosuppression is by inducing changes to MSC gene expression and phenotype in a way that ultimately facilitates immunosuppression. In response to inflammatory signals, MSCs tend to upregulate inducible nitric oxide synthase (iNOS), TGF-β, and indoleamine 2,3-dioxygenase (IDO), which subsequently contribute to immunosuppression. Pro-inflammatory cytokines in the MSC local environment can regulate MSC-mediated immune suppression via iNOS induction [ 148 ]. In a syngeneic mouse model of melanoma, silencing iNOS in MSCs inhibited tumor growth compared with tumors with wild type MSCs [ 148 ]. Similarly, pre-incubation of mouse MSCs with the inflammatory molecules, IFN-γ, TNF-α, or IL-1α prior to co-transplantation of MSCs with tumor cells in mice can facilitate tumor formation in allogeneic recipients through the induction of iNOS [ 137 ] and the upregulation of TGF-β expression by MSCs [ 149 ]. In response to IL-1β from monocytes, MSCs secrete TGF-β which then inhibits alloreactive T cells [ 150 ]. The induction of IDO expression is another important mechanism by which inflammatory cytokines enhance the immunosuppressive capacity of BM-MSCs [ 151 , 152 ]. This was particularly well-demonstrated in experiments that demonstrated a phenotypic switch in MSCs, from immunosuppressive to immune enhancing, when IDO was knocked down [ 148 ]. Bi-directional communication between cells of the TME and cancer cells can induce new neoplastic capabilities that ultimately promote tumor invasion and metastasis [ 153 ]. Human MSCs can enhance cancer metastasis by supporting tumor cell dissociation, dissemination, and engraftment into distant tissues. In vitro evidence suggests that human BM-MSCs promote the detachment of tumor cells from the primary tumor through the downregulation of E-cadherin expression and the subsequent disruption of cell-cell adhesion [ 154 – 156 ]. This effect is partly mediated by activation of a disintegrin and metalloprotease domain-containing protein 10 (ADAM 10) [ 154 ]. One group compared tumor-associated and normal lung tissue MSCs in the setting of lung cancer and found that tumor-associated MSCs acquired the capacity to promote lung cancer metastasis [ 157 ]. To further characterize these tumor-associated MSCs, this group demonstrated that four genes, GREM1, LOXL2, ADAMTS12 and ITGA11 contributed to this metastasis-promoting phenotype [ 157 ]. In vivo evidence also supports a significant role for MSCs in promoting metastasis. One such study demonstrated that co-transplantation of human BM-MSCs with a colon cancer cell line in mouse spleen enhanced the number of metastatic lesions in the liver almost five-fold relative to tumor alone [ 158 ]. MSCs can further facilitate tumor cell motility by degrading and remodeling the ECM through matrix metalloproteinase (MMP) release. Human MSCs express MMP-2 and MMP-9 [ 29 – 32 ] and these are upregulated by inflammatory cytokines in the TME [ 31 ]. The engagement of TLR ligands can also enhance expression of MMP-13 by MSCs [ 159 ]. However, it is important to note that a matrix-protective role of MSCs though tissue inhibitor of metalloproteinases (TIMP) mediated MMP inhibition has also been reported [ 160 ]. Cancer cells seem to play an active role in modifying MSCs in a way that supports tumor progression. For example, EVs from renal cancer stem cells can induce a significant upregulation of MMP1, MMP2 and MMP3 expression by human BM-MSCs [ 161 ]. Growing evidence implicates MSCs with re-activation of EMT and enhanced invasive progression of tumors [ 124 , 162 – 164 ]. EMT is a mechanism by which cancer cells transition from an epithelial phenotype into a migratory mesenchymal phenotype to enable dissociation, invasion, dissemination, and engraftment [ 153 , 165 – 167 ]. The co-culture of human BM-MSCs with tumor cells can trigger EMT transition in tumor cells through upregulation of EMT regulators including Twist, Snail and Zeb1, the downregulation of E-cadherin, and the upregulation of mesenchymal markers [ 168 , 169 ]. TGF-β appears to play a key role in controlling this process [ 168 , 170 , 171 ]. BM-MSC membrane-bound TGF-β induced by cell-to-cell contact with cancer cells can promote the expression of EMT-related genes in colorectal tumor cells [ 168 ]. Similarly, adipose-derived MSCs promoted EMT of cervical cancer cells and enhanced their migration, invasion, angiogenesis, and tumorigenesis [ 53 ]. MSCs also promote tumor metastasis by releasing trafficking molecules within the tumor niche. For example, in a humanized model of breast cancer metastasis to bone, IL-17B produced by human BM-MSCs stimulated the migration and metastasis of breast cancer cells [ 71 ]. Similarly, MSC secretion of chemokine receptor type 5 (CCR5) ligands including CCL3, CCL4, and CCL5 appear to enhance metastasis. Human BM-MSCs enhanced breast cancer metastasis through paracrine signaling involving CCL5 which induces Akt-mediated migration and metastatic spread of cancer cells into distant tissues [ 8 ]. In line with these experimental findings of the role of MSC-mediated paracrine signaling in cancer progression, clinical evidence suggests that patients with colorectal cancer cells that overexpress CCR5, and patients with high serum levels of CCL3 and CCL4 exhibited poorer prognoses [ 172 ]. MSCs may also enhance tumor metastasis at secondary sites via the release of trafficking molecules such as CCL2 and SDF-1 that attract circulating tumor cells and facilitate their entry into the parenchyma of distant organs [ 173 – 175 ]. Partly through Tac1-mediated regulation of SDF-1α and CXCR4, MSCs support the dissemination of breast cancer cells to the bone marrow [ 175 ]. In addition, after cancer cells migrate to secondary sites, MSCs can provide them with structural support and enable their engraftment through the release of ECM components [ 176 ]. MSCs therefore contribute to the creation of tumor-favorable environments.

MSCs are increasingly recognized as important supporters of tumorigenesis. While MSC-tumor cell interactions are complex and appear to be specific to tumor type, several common mechanisms by which MSCs support tumor progression were highlighted in this review. MSCs are actively recruited into the stroma of developing tumors via inflammatory signals or tumor-derived factors to stimulate healing and repair. These cancer-associated MSCs may arise from the bone marrow and migrate to tumors or, perhaps more likely given the rarity of circulating MSCs, they could arise from adjacent normal tissue. After recruitment, they interact with tumor cells via paracrine loops and/or direct cellular contacts to modulate tumor progression. Naïve MSCs are co-opted and reprogrammed by tumor cells to support the tumorigenic process through their contribution to CAF-like cells, ECM remodeling, the production of growth stimulating and pro-angiogenic factors such as IL-6, IL-8, and VEGF, and the activation of anti-apoptotic pathways and immune evasion. Furthermore, MSCs can lead to more malignant and invasive tumor phenotypes by re-activation of the EMT program, induction of cancer cell stemness, release of trafficking molecules in the TME such as CCl-5, SDF-1 and IL-17B, acquisition of drug resistance, and activation of signaling pathways associated with tumor growth and invasiveness including STAT3, PI3K/Akt, ERK1/2, and Notch. Pro-inflammatory cytokines at tumor sites play a role in altering the pro-tumorigenic activity of MSCs and stimulate them to express anti-inflammatory mediators, matrix modulating enzymes, and pro-metastatic molecules. Tumor-secreted factors such as TGF-β and Wnt3a stimulate MSCs to acquire a CAF-like phenotype which is a more potent driver of cancer growth than naïve MSCs. These tumor-secreted factors contribute to tumor stroma formation and angiogenesis, and they supply the paracrine factors necessary for tumor growth and progression. Indeed, in contrast to their counterparts residing in normal tissue, CA-MSCs display phenotypic and functional characteristics that are important for tumor propagation, including higher immunosuppressive abilities, skewing toward Treg and Th-2 immune responses, and increased chemokine secretion. In the bone marrow, MSCs create a favorable environment for the engraftment, homing, and progression of disseminated malignant cells. An important challenge that confounds the generalizability of the pro-tumorigenic properties reviewed here is the heterogeneity of MSCs. In addition to differences based on the tissue source of MSCs (i.e. bone marrow vs cord blood vs adipose tissue), MSCs can also exhibit differences in proliferative capacity, differentiation patterns, and profile of secreted cytokines even across colonies that arise from the same original batch [ 177 ]. Differences in individuals from whom MSCs are derived – such as age, sex, genetic makeup, and health conditions – further add to the complexity and heterogeneity of MSCs [ 177 ]. While a goal of this review has been to comprehensively outline the various tumorigenic properties of MSCs and the range of factors that these cells can release, it is unlikely that the MSCs studied in any individual laboratory will contain all these properties or release all the factors discussed in this review. This review has focused exclusively on the tumor-supporting features of MSCs. However, it is important to remember that numerous studies have reported a role for MSCs in slowing the progression of various malignancies and many groups are working to harness MSCs for therapeutic applications in the treatment of cancer patients. Nonetheless, the body of evidence supporting the tumor-promoting effects of some MSC populations is convincing and further work to delineate the mechanisms by which MSCs support tumorigenesis is important and may ultimately lead to novel targeted therapies.