In the 1970s, a novel subset of non-hematopoietic stem cells with a multilineage potential was described in the marrow stroma by Friedenstein and co-workers.13 They reported an adherent fibroblast-like cell with the ability to generate colonies in vitro (figure 1). These cells, defined as colony-forming unit fibroblasts, demonstrated later the ability to differentiate both in vitro and in vivo into a variety of cell types of mesenchymal origin such as adipocytes, chondrocytes, osteocytes, and tenocytes, and into other cell types such as cardiomyocytes, myocytes, astrocytes, and neurons, among others.14 It was at the end of the 1980s when further studies led to characterize the heterogeneity of the marrow stromal cell pool and to identify a stem cell population with a mesodermal germ layer origin that contributed to the formation of connective tissue, skeletal muscle cells, and vascular system. These cells are known as marrow stroma stem cells or bone marrow-derived MSCs (BM-MSCs).15 16 Of special interest were the experiments performed by several groups showing that BM-MSCs had unique immunomodulatory properties in an allogeneic transplant.17

During the following decades, and thanks to their adherence properties and to new MSCs isolation procedures, an increasing number of human MSCs were identified from diverse organs and tissues such as adipose tissue, muscle, liver, periosteum, dental pulp, perichondrium, menstrual blood, etc.18 MSCs were also described in fetal fluids and tissues, including amniotic fluid, cord blood, placenta, bone marrow, blood, liver, and lung.19 20 Interestingly, although both fetal and adult MSCs share immunological attributes,21 those from fetal origin are present in higher amounts and have shorter cell doubling time and higher proliferative capacity than those from adult tissues.22

To date, there are no unique markers to unequivocally identify MSCs and the presence or the absence of a set of non-specific markers are required to phenotypically characterize them. This challenging phenotypic marker landscape has prevented the proper identification of MSCs isolated from different laboratories.23 In 2006, the International Society for Cellular Therapy defined MSCs as self-renewing multipotent fibroblast-like cells with plastic adherent capacity, capable to differentiate, at least, into adipocytes, osteocytes, and chondrocytes, positive for expression of CD90, CD105, and CD73, and negative for CD45, CD19, CD14, c-kit, CD11b, CD79α, CD34 (except adipose tissue-derived MSCs that express CD34 in short-term cultures), and human leukocyte antigen-DR (HLA-DR).24

MSCs present unique immunological and immunomodulatory functions. Thus, MSCs are considered to have low immunogenicity owing to the lack of expression of HLA class II, CD40, CD80, and CD86 co-stimulatory molecules, and low levels of expression of HLA class I on their surface.25 Although it has been widely considered that this immunological privileged phenotype allows MSCs to evade immune system recognition, recent reports have shown that MSCs can elicit both humoral and cellular immune responses, thus suggesting an immune evasive rather than an immune privileged immunophenotype.26

MSCs also exert immunomodulatory functions over almost all immune cells through cell-to-cell contact and by secreting a wide variety of cytokines, chemokines, and growth factors. Thus, MSCs have been described to regulate the activity of naïve, memory, and effector T cells, B cells, regulatory T cells (Tregs), natural killer cells (NKs), dendritic cells, monocytes/macrophages, and neutrophils.27 28

Finally, and of special interest for systemic cancer treatment, MSCs are known to migrate to sites of injury and inflammation, which are two traits of the TME.29 MSCs respond to inflammation mediators that are highly expressed in the tumor such as growth factors (vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF), granulocyte–macrophage colony-stimulating factor, hematopoietic growth factor, hepatocyte growth factor (HGF), epidermal growth factor (EGF), transforming growth factor beta-1 (TGF-β1), granulocyte colony-stimulating factor or stromal-derived growth factor-1α), chemokines (CCL2, CCL3, and CXCL-12), cytokines (interleukin 6 (IL-6) and IL-8), integrins (vascular cell adhesion molecule 1 (VCAM-1) and intracellular adhesion molecule-1 (ICAM-1)), and selectins (P-selectin and E-selectin).30 In response to these local tumor inflammatory mediators, MSCs initiate a rolling process similar to that described for leukocytes in response also to inflammation, whereby MSCs adhere to the vascular epithelium and begin transmigration into the tumor.31 Evidence of MSCs tumor migration has been reported for several cancer types, pointing out their potential as delivery vectors for antitumor therapies.32

Their easy isolation and expansion, in addition to their immune evading properties and their ability to migrate to the tumor, make MSCs particularly appealing as cell carriers for OVs.

The effect of MSCs on tumor progression has been a controversial field for the last decades, with several studies suggesting a role of MSCs in promoting tumor growth and metastasis, while others support their antitumor potential.33 In recent years, the pro-tumorigenic theory has gained force based on studies that demonstrate the effect of MSCs in the TME and its regulation in tumor immunosurveillance.

To better understand the effect of MSCs on the TME, it is important to note that tumors have long been considered as a wound process that fails to resolve.34 35 Thus, it has been described how wound healing and tumor progression share common traits like inflammation, tissue proliferation and remodeling, ECM remodeling, immune cell mobilization and activation, and neovascularization. However, while this is a well-defined process for wound healing, in cancer this process is not properly completed and becomes an unresolved situation.

MSCs contribution to tumor inflammation, tissue proliferation and remodeling steps have been extensively reviewed by Li et al,36 who highlight the key role of tumor-associated MSCs (TA-MSCs) at the first stages of cancer development. TA-MSCs in the tumor inflammatory microenvironment secrete cytokines and chemokines such as IL-6, HGF, CCL-2, CXCL1/2, HGF, or CXCL3, that promote infiltration of myeloid-derived suppressor cells (MDSCs) and monocytes, and differentiation of the latter to M2 macrophages. Additionally, TA-MSCs participate in recruiting new MSCs to tumors where they support tumor proliferation through their transdifferentiation into tumor cancer-associated fibroblasts, which in turn provide survival and proliferative signals to cancer stem cells by secreting growth factors and cytokines such as HGF, TGF-β, FGF, IL-6, IL-8, CXCL1/7, and bone morphogenetic protein. TA-MSCs and recruited MSCs also secrete VEGF and platelet-derived growth factor, which support the neovascularization of the growing tumor. In the last step of tumor progression, the remodeling step, MSCs role is contradictory. On one hand, TA-MSCs release exosomes containing microRNAs, which reduce the tumor growth rate by promoting a dormancy phenotype in tumor cells, and also inhibit angiogenesis. These antitumoral functions are considered a reminiscence of their anti-inflammatory cellular program. On the other hand, TA-MSCs secrete chemokines (CCL-2, CCL-5, CCL9, and CXCL10) and cytokines (IL-8, IL-1β, tumor necrosis factor-alpha (TNFα), and TNFβ) that stimulate the epithelial–mesenchymal transition, increasing tumor cell invasion and metastasis.

Cytokines and exosomes secreted by TA-MSCs, as a result of their crosstalk with tumor cells, modulate the tumor surveillance acting on different players of the innate and the adaptive immune system.37 Regarding the innate immune system, TA-MSCs stimulate recruitment and differentiation of immunosuppressive macrophages M2, via secretion of prostaglandin E2 (PGE2) and exosomes,38 39 and of MDSCs, through the secretion of HGF and CXCL3.40 Moreover, IL-6, PGE2, and microRNAs secreted by MSCs also prevent differentiation of monocytes to dendritic cells (DCs), and maturation of immature into mature DCs.41 TGF-β, IL-6, PGE2, and microRNAs secreted by TA-MSCs are responsible for the impaired function of NK cells, in some cases downregulating the expression of activating NK cell receptors: NKp44, NKp30, NKG2D, DNAX accessory molecule-1 (DNAM-I), and NKG2A.42 43 Finally, MSCs exert a potent γδ-T cell immunosuppression through cyclooxygenase-2 (COX2)-dependent production of PGE2.44

MSCs also immunosuppress the adaptive immunity in the tumor. MSCs can inhibit naïve and memory T cell responses by upregulating lymphocyte function-associated 3, ICAM-1, and VCAM-1.45 46 Additionally, TA-MSCs impair correct effector lymphocyte activation by secreting inhibiting factors such as TGF-β, PGE2, IL-10, indoleamine 2,3-dioxygenase and nitric oxide.45 Furthermore, TA-MSCs exert a potent TME immunosuppressive effect by increasing the amount of Tregs in the tumor.47

Altogether, these MSC functions in the TME support the role of MSCs in tumor progression and spread. Therefore, it is necessary to carefully modify MSCs prior to their application as antitumor therapies, by knocking down some of these pro-tumoral properties, inserting suicide gene as ‘safety switch’, or combining with OVs that eventually eliminate infected MSC during the treatment. In line with this, modified MSCs have been evaluated in several animal models,30 and their safety and efficacy have been reported in early clinical trials for cancer treatment.48 No malignant or pro-tumoral effects have been reported from any of these studies, indicating that modified MSCs lack the tumor-promoting properties described above.

MYXV expressing green fluorescent protein (GFP) as reporter gene has been combined with human adipose-derived MSCs (ASCs) to treat malignant brain tumor in mice.49 This study demonstrated not only the capability of ASCs to generate new viral particles, but also the tumor-homing properties of infected cells on intracranial administration. As a result, a significant increase in survival was observed in mice treated with the OV-loaded MSCs in comparison with mice treated with MSCs alone.

BM-MSCs have also shown to be permissive to MYXV replication.50 Moreover, using an IL-15-armed MYXV to infect BM-MSCs, the authors demonstrated a superior antitumor efficacy in an immune competent pulmonary melanoma model after intravenous treatment, compared with MYXV monotherapy. An elevated percentage of circulating NK cells was observed only in those animals treated with the virus without cell carriers, indicating that MSCs potentially prevented recognition of the virus by the immune system. Finally, an increase of pro-inflammatory cytokines (interferon (IFN)-γ and TNFα), programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1), and effector T cells infiltration was documented only in tumors from animals treated with MYXV-IL-15-loaded MSCs, indicating a possible antitumor immune response as a consequence of the therapy.

Two different studies have demonstrated the antitumor potential of the combination of HSV and MSCs. Leoni and co-workers infected fetal membrane-derived MSCs (FM-MSCs) with an HSV encoding for a single-chain antibody targeting human epidermal growth factor receptor 2 (HER-2) (R-LM249).51 FM-MSCs viral infection, viral progeny production, and tumor viral delivery were demonstrated. The antitumor efficacy on intravenous administration was evaluated in two different mice models. In a lung metastases model (using the HER-2 positive SK-OV-3 cell line), R-LM249-infected MSCs significantly inhibited the appearance of metastasis, unlike uninfected MSCs. Similar results were observed in a model of brain metastasis (HER2⁺ MDA-MB-453 EGFP cell line), in which R-LM249-infected FM-MSCs prevented the development of brain metastases. In a second study, Du et al reported MSC-mediated delivery of oncolytic-HSV to brain metastases in a murine model of melanoma after a intracarotid single administration.52 This tumor delivery resulted in a significant remission of metastatic tumors and mice survival in comparison to MSCs-treated alone.

MSCs from diverse origins have been combined with MV for cancer treatment. Two groups have evaluated intravenous single administration of MV-loaded BM-MSCs in hepatocellular carcinoma and lymphoblastic leukemia mouse models.53 54 Both studies reported BM-MSCs permissiveness for virus replication and, most importantly, virus delivery to tumors and higher therapeutic efficacy than cell-free MV in the presence of pre-existing neutralizing anti-MV antibodies. MSCs from adipose tissue have also been combined with MV. Mader et al tested MV-infected ASCs antitumor potential in a xenograft murine model of ovarian cancer after a single intraperitoneal administration.55 In line with the studies described above, the use of ASCs as cell carriers not only enhanced the therapeutic potential in comparison with virus alone treatment, but also the capacity to protect the virus from pre-existing antiviral immunity. In a follow-up study, the same authors pointed out the feasibility of using patient-derived MSC as carriers for MV.56

Importantly, these three studies show for the first time the advantage of using MSCs as cell carriers when viral pre-immunity exists.

Oncolytic Ads (OAds) have been the most frequently used viruses in combination with MSCs for cancer treatment. Pioneer studies in 2007 compared Ad infectivity and viral production in MSCs using chimeric or modified adenoviral fibers.57 58 These studies also highlighted the capacity of OAd-loaded MSCs to reach the tumor after single intravenous administration, and to reduce tumor growth in lung and breast tumor murine models.

The group of Frederick Lang evaluated the combination of BM-MSCs with the OAd delta-24/RGD. In a first work using two different glioma xenograft murine models (U87 and U215-V121), the authors detected brain tumor tropism and antitumor activity of infected MSCs after single intracarotid administration.59 A second study demonstrated a main role of TGF-β secreted by glioma stem cells in the BM-MSCs glioma tropism.60

Bolontrade et al described a subpopulation of BM-MSCs, overexpressing integrins α2, α3, and α5, with and enhanced tumor migration capacity.61 Loading these cells with an OAd led to a significant therapeutic efficacy compared with unloaded MSCs and virus as monotherapies in a melanoma xenograft model after a single retro orbital administration.

Hypoxic conditions have been also reported to favor MSCs tumor-homing properties by inducing the upregulation of CXCR4 and CXCR1 receptors.62 However, the combination of an E1B-55KD-deleted OAd with BM-MSCs cultured under hypoxic conditions did not control tumor growth after a single intraperitoneal administration in a xenograft model of colon cancer. Authors attributed the low potency of the virus as a possible explanation for this low efficacy, although no control group with virus as monotherapy was included in the study. The combination with a more potent OAd would have been desirable to demonstrate the benefit of culturing BM-MSCs in hypoxia for this type of cancer therapy.

Hepatocellular carcinoma murine models have been used to evaluate the efficacy of umbilical cord-derived MSCs (HU-MSCs) and BM-MSCs loaded with different OAds. Combining an OAd expressing IL-24 with HU-MSCs engineered to express the E1A adenoviral protein, Li and colleagues reported successful virus replication and virus production, and efficient migration of virus-loaded HU-MSCs to tumors after a single intravenous administration.63 Furthermore, addition of a low-dose 5-fluorouracil to the OAdv-infected HU-MSCs combination resulted in effective tumor growth inhibition. HU-MSCs were also combined with an OAd expressing a microRNA-122 target sequence to avoid virus replication in normal liver cells.64 Significant antitumor efficacy was reported after two intravenous administrations using orthotopic and subcutaneous hepatocarcinoma tumor models, although direct comparison with virus alone was not performed. Finally, BM-MSCs as cell carriers for an OAd encoding a WNT-inhibiting (WNTi) decoy receptor have been tested in an orthotopic hepatocellular carcinoma. After optimizing the OAd dose for loading into BM-MSCS, two intravenous administrations of OAd-WNTi-loaded BM-MSC showed remarkable therapeutic efficacy in vivo65 . Interestingly, direct comparison with OAd-WNTi without cell carriers showed 8.1-fold greater therapeutic efficacy for cell-protected viruses compared with the ‘naked’ virus group. Moreover, MSCs efficiently extended OAd blood circulation time while preventing hepatic damage in comparison with direct virus administration. In a follow-up study, the same group used a dynamic in silico TME model interaction network and a predictive mathematical model to identify two parameters to improve the outcome: the multiplicity of infection used to infect MSCs and the number of OAd-MSCs administered.66

Ad/RLX, an OAd coding for relaxin, a peptide hormone with ECM collagen degradation properties, has been combined with BM-MSCs for preclinical pancreatic cancer treatment.67 Nude mice carrying AsPC-1 pancreatic xenograft tumors received three intravenously administration of naked Ad/RLX, Ad/RLX-loaded BM-MSCs or BM-MSCs infected with Ad/RLX previously treated with poly(ethyleneimine)-conjugated poly(CBA-DAH) (PCDP), a cationic polymer that increases MSCs viral infectivity. Significant antitumor efficacy was observed only for the last group, probably due to the higher MSCs infectivity and viral production of PCDP-treated Ad/RLX.

Using xenograft chick embryo model of pancreatic cancer, Kaczorowski et al showed the capacity of OAd-infected BM-MSCs to invade tumor spheroids generating new viral particles in the tumor.68 Moreover, using a chimeric OAd (fiber 5/3) expressing TNF-related apoptosis-inducing ligand as load for BM-MSCs, increased tumor reduction was determined compared with the virus monotherapy.

MSCs isolated from menstrual blood (MenSCs) have been also used as cell carriers for OAd. Guo et al used MenSCs for tumor delivery of a chimeric OAd in murine models of colorectal cancer.69 After intravenous or intraperitoneal administration, correct MSCs tumor migration was determined. Moreover, two intravenous administrations of Ad5/F11-infected MenSCs resulted in significant antitumor efficacy although viral treatment alone was not included in the study as a control.

ICOVIR15, an E1a/delta24-modified promoter and RGD fiber OAd, has been used in combination with MSCs for cancer treatment. The group of M Brenner used BM-MSCs infected with an ICOVIR15 engineered to express an inducible Caspase 9 suicide gene (ICOVIR15-Ad.iC91) for preclinical lung cancer treatment.70 After one single intravenous administration, MSCs could be detected in tumor tissues and higher tumor control was observed in animals treated with ICOVIR15-Ad.iC9-MSCs.

We have focused on using MenSCs as an alternative ICOVIR15 cell carrier. In a first study, viral capsid modifications were compared to select the most efficient in terms of MSCs infection and viral progeny production.71 Using a lung xenograft murine model, tumor homing potential of uninfected and infected MenSCs was demonstrated after intraperitoneal administration. A subsequent study reported the benefit of using allogeneic instead of autologous MenSCs, as a consequence of the higher pro-inflammatory environment generated when OAd-loaded MenSCs interacted with allogeneic peripheral blood mononuclear cells (PBMCs).72 Activation of both T cells and NK cells was associated with this pro-inflammatory MSCs phenotype indicating the potential capability of Ad-infected MSCs to favor tumor immune recognition. The therapeutic potential of OAd-MenSCs was demonstrated in tumor-bearing NSG mice who received intravenous injections of allogeneic human PBMCs. Monocytes and NK cells contributed to efficacy. Finally, the therapeutic potential of OAd-MenSCs has been further improved by loading MenSCs with an ICOVIR15 oncolytic vector engineered to express an EGF receptor-targeting bispecific T cell engager (cBiTE). Antitumor efficacy of ICOVIR15-cBiTE MenSCs was superior to both monotherapy controls. Altogether, these results suggest that to improve viral-loaded MSCs therapy OAd should be optimized.73

Finally, two different groups have demonstrated the preclinical benefit of using OAd-loaded MSCs to treat brain tumors after local or intratumoral administration. Sonabend and co-workers reported that injecting OAd-MSCs 5 mm away from brain lesions, promote cell migration to the tumor which results in local virus production. No efficacy experiment was performed in this study. More recently, Martinez-Quintanilla et al employed ICOVIR17, an ICOVIR15 modified to encode for soluble hyaluronidase PH20 that mediates ECM hyaluronic acid degradation, in combination with ASCs to treat glioblastoma (GBM) intratumorally.74 A significant antitumor efficacy was observed in mice bearing intracranial U87 tumors or patient-derived GBM xenografts.

In summary, MSCs from diverse origins have demonstrated potential as cell carriers for RNA or DNA OVs with different degrees of lytic properties (ie, low or high),75 highlighting the versatility of these cells to combine with OVs.