Mesenchymal stem/stromal cells as next-generation drug delivery vehicles for cancer therapeutics

ABSTRACT

Introduction

Drug delivery to solid tumors remains a significant therapeutic challenge. Mesenchymal stem/stromal cells (MSCs) home to tumor tissues and can be employed as tumor targeted drug/gene delivery vehicles. Reportedly, therapeutic gene- or anti-cancer drug-loaded MSCs have shown remarkable anti-tumor effects in preclinical studies, and some clinical trials for assessing therapeutic MSCs in patients with cancer have been registered.

Areas covered

In the present review, we first discuss the source and interdonor heterogeneity of MSCs, their tumor-homing mechanism, and the route of MSC administration in MSC-based cancer therapy. We then summarize the therapeutic applications of MSCs as a drug delivery vehicle for therapeutic genes or anti-cancer drugs and the drug delivery mechanism from drug-loaded MSCs to cancer cells.

Expert opinion

Although numerous preclinical studies have revealed significant anti-tumor effects, several clinical trials assessing MSC-based cancer gene therapy have failed to demonstrate corroborative results, documenting limited therapeutic effects. Notably, a successful clinical outcome with MSC-based cancer therapy would require the interdonor heterogeneity of administered MSCs to be resolved, along with improved tumor-homing efficiency and optimized drug delivery efficiency from MSCs to cancer cells.

KEYWORDS:

• Anti-cancer drug
• cancer therapy
• cell-based drug delivery
• cell engineering
• mesenchymal stem/stromal cells
• targeting
• tumor homing

1. Introduction

Ongoing medical research is persistently striving to overcome various types of cancers. However, cancer remains one of the leading causes of death worldwide, with an estimated 19.3 million new cases and 10.0 million cancer-related deaths reported in 2020 [1]. Chemotherapy and immunotherapy play a vital role in cancer therapy, along with surgical intervention and radiation therapy. Various types of anti-cancer agents, including chemotherapeutics, hormones and molecular targeted agents, have been approved for clinical use [2]. However, most of these anti-cancer agents are systemically administered and distribute not only to tumor tissues but also to normal tissues, where they may induce side effects. In addition, some types of cancers do not exhibit any response to anti-cancer agents because the delivery efficiency of the agents is very low [3,4]. Therefore, tumor-targeted drug delivery is extremely important to improve the safety and efficacy of anti-cancer agents. Several cancer-targeting methods have been developed, including passively targeted nanocarriers (e.g. liposomes and micelles) and actively targeted therapeutics (e.g. antibody-drug conjugates [ADCs]). Although the tumor accumulation of nanocarriers and ADCs is basically based on the enhanced permeability and retention (EPR) effect [5], effective cancer therapy has been achieved only in some cancer types. Several unsolved issues limit the therapeutic effects of nanocarriers and ADCs. For example, the high interstitial fluid pressure in tumor tissues significantly limited the accumulation of nanocarriers in inner tumor tissue regions [3]. In addition, the abundant stroma in some tumor types potently suppressed the uptake of nanocarriers and the binding of ADCs to surface receptors of cancer cells [6]. Therefore, cancer-targeting methods that do not depend on the EPR effect are currently in demand.

Cells with tumor-homing ability have recently been recognized as a drug delivery vehicle for cancer therapy. Some types of immune cells, stem cells, and progenitor cells are known to possess tumor-homing abilities [7,8]. Endogenous monocytes, neutrophils, and lymphocytes are recruited from the bone marrow (BM) or lymphoid tissues to tumor tissues, where they contribute to immune reactions. In addition, exogenously transplanted leukocytes and monocytes migrate to tumor tissues [7–9]. In 2000, Aboody et al. reported that neural stem cells injected into the frontoparietal lobe of the brain migrated to brain tumor tissues in a mouse model of glioma [10]. Furthermore, systemically administrated hematopoietic progenitor cells and fibroblast-derived neural stem cells were detected in the tumors of tumor-bearing mice [11,12]. These cells are guided by chemokines secreted from solid tumor tissues and spontaneously invade tumors. Therefore, some cell types with tumor-homing ability can be developed as drug delivery vehicles independent of the EPR effect.

Among the various types of tumor-homing cells, mesenchymal stem/stromal cells (MSCs) are the most attractive candidates for drug delivery vehicles for cancer therapy, as they demonstrate tumor tropism, a self-renewing/proliferative capacity, low immunogenicity, and low risk of tumorigenesis (Figure 1A). According to the minimal criteria proposed by the International Society for Cell & Gene Therapy (ISCT), MSCs are plastic adherent cells that express CD73, CD90, and CD105, but do not express CD34, CD45, CD14, CD11b, CD79a, CD19, and HLA-DR, and at least differentiate into adipocytes, osteoblasts, and chondrocytes [13,14]. MSCs can be isolated from various tissues, including BM, adipose tissue (AD), and umbilical cord (UC; UC blood and Wharton’s jelly) [15] or induced from pluripotent stem cells [16], and rapidly expanded on a clinical scale using a standard monolayer culture method [15]. The risk of tumor formation with MSCs is considered extremely low owing to their genetic stability [17], and several in vivo tumorigenicity tests [18] and long-term follow-up of patients with MSC transplantation [19] revealed no tumor formation. Moreover, low immunogenicity is another unique characteristic of MSCs, and allogeneic MSC transplantation rarely causes rejection because MSCs express low levels of major histocompatibility complex (MHC) class I and exceptionally low levels of MHC class II and costimulatory molecules (CD40, CD80, and CD86) [20,21]. Furthermore, in animal models, MSCs homed to various primary tumors and metastatic foci, including Kaposi’s sarcoma, Ewing sarcoma, fibrosarcoma, glioma, melanoma, lung cancer, hepatocellular cancer, stomach cancer, colon cancer, pancreatic cancer, renal cell carcinoma, breast cancer, and ovarian cancer [8,14,15,22–24]. Accordingly, the use of MSCs would be useful for improving the safety and efficacy of anti-cancer agents through the enhanced delivery efficiency of anti-cancer agents to tumor tissues. Many researchers have evaluated the application of MSCs as a drug delivery vehicle for anti-cancer drugs and therapeutic genes (Figure 1B).

Figure 1. The properties and therapeutic applications of MSCs as a vehicle for anti-cancer agents. (a) MSCs can be isolated from various tissues or induced from pluripotent stem cells. The International Society for Cell & Gene Therapy (ISCT) has defined minimum criteria for MSCs: adhesion to plastic plates, expression of CD73, CD90, CD105, no expression of CD34, CD45, CD14, CD11b, CD79a, CD19, and HLA-DR, and differentiation into adipocyte, osteoblast, and chondrocyte. MSCs are useful as a source of tumor-targeted drug delivery vehicles owing to their high self-renewing or proliferation potency, low immunogenicity, low risk of tumorigenesis, and tumor tropism. (b) Various engineered MSCs. MSCs are employed as a cancer-targeting vehicle for therapeutic genes, oncolytic virus, anti-cancer drugs, and drug-encapsulated nanoparticles. MSCs, mesenchymal stem/stromal cells

Based on their self-renewal and proliferative capacity, low immunogenicity, low risk of tumorigenesis, and tumor tropism, MSCs are considered a next-generation drug delivery vehicle for cancer therapeutics. In the present review, we describe the properties and therapeutic applications of MSCs as drug delivery vehicles for anti-cancer drugs during cancer therapy.

2. Properties of MSCs

Understanding the properties of MSCs is valuable for the application of MSCs as a drug delivery vehicle. In this chapter, we first classify the literature according to the MSC source, tumor homing mechanisms, and MSC administration route. Next, we summarize a few recent studies on the influence of exogenously administered MSCs on the tumor microenvironment.

2.1. MSC source

2.1.1. Tissue sources of MSCs

MSCs can be isolated from various tissues, with BM, AD, and UC considered major sources. In regenerative medicine, MSCs have been used to treat autoimmune diseases, cardiovascular diseases, neurological disorders, and severe inflammatory diseases [15,25]. More than a thousand clinical trials have been registered. BM-, AD-, or UC-derived MSCs have been employed in over 90% of clinical trials [25]. Furthermore, BM- and AD-derived MSCs have been approved for clinical use to treat acute graft versus host disease (GVHD), osteoarthritis clone disease, and spinal cord injury in the European Union and Japan (Alofisel®, Stemirac®, Temcell®). In research assessing MSC-based drug delivery, BM-, AD-, and UC-derived MSCs have been frequently utilized.

BM is the most common source of MSCs, and BM-derived MSCs have been used for over 20 years in the field of regenerative medicine [25]. However, BM-derived MSCs present some unfavorable aspects and disadvantages when compared with AD and UC as sources of MSCs. For example, BM aspiration is highly invasive, and the volume of collectible BM is extremely low (15 mL/kg from a healthy donor). Moreover, BM contains a relatively low number of MSCs (0.001‒0.01% of total BM cells) [25–27]. In comparison, the collection of AD tissue is much less invasive, and the yield of MSCs can be 100‒500-fold greater than that of BM [27–29]. UC can be obtained as perinatal medical waste with a high yield (3 × 10⁵ cells/cm of fresh Wharton’s jelly) [30]. All MSCs prepared from these tissues share typical characteristics, such as tumor-homing, tissue repair, and immune regulation, and AD- and UC-derived MSCs have been increasingly used in recent years [25]. Conversely, some comparative studies have reported that BM-, AD-, and UC-derived MSCs present unique characteristics in terms of the proliferation rate, the expression level of immunosuppressive cytokines, and tumor tropism [27–31]. However, it is challenging to systematically characterize MSCs depending only on their source, as MSCs from any source are highly heterogeneous. In addition, the method employed for isolating MSCs can affect the collectible cell population, and standard isolation protocols are yet to be established [30,32]. Furthermore, the donors’ age and health condition [21], cell culture conditions, and cell passage numbers [21,33] greatly impact the MSC characteristics. Therefore, selecting the tissue source of MSCs may be insufficient to govern their use as a drug delivery vehicle for cancer. Instead, it is important to select MSCs with high tumor-homing ability or to optimize the cell culture protocol to improve or maintain the tumor-homing ability for the therapeutic application of MSCs in cancer therapy.

2.1.2. Preparation of homogenous MSCs

As described above, the interdonor heterogeneity of MSCs could reduce the drug delivery efficiency of MSC-based drug delivery and could hinder clinical applications. Zyrafete et al. succeeded in reducing interdonor heterogeneity and established a cell bank of clinical-grade MSCs by mixing and pooling cells isolated from the BM of eight healthy donors [34]. This cell bank can supply 500 doses of MSCs for the treatment of GVHD. Pluripotent stem cells can also be a source of homogenous MSCs on a larger scale. Several protocols to induce MSCs from embryonic stem (ES) cells and induced pluripotent stem (iPS) cells have been developed [16], and MSCs differentiated from these pluripotent cells reportedly present characteristics similar to those of primary MSCs [35]. Adrian et al. succeeded in establishing an MSC supply system using iPS cells, which could generate 29 million clinical doses (2 × 10⁶ cells/kg) of good manufacturing practice (GMP)-grade MSCs, CYP-001. In a phase I clinical trial, the safety of CYP-001 was demonstrated in patients with GVHD [35]. Therefore, pluripotent stem cell is an attractive source to prepare homogenous MSC-products from a perspective of scalability, and pluripotent stem cell-derived MSCs will be becoming more widely used in the near future.

2.1.3. Autologous vs. allogeneic transplantation of MSCs

Both autologous and allogeneic MSCs have been assessed in clinical trials for MSC-based cancer gene therapy (Table 1). Autologous transplantation carries no risk of immune rejection. Several studies have reported that the survival time of syngeneic MSCs transplanted into mice was longer than that of allogeneic MSCs [20]. Zangi et al. reported that transplantation of allogeneic MSCs induced a memory T cell response [36]. However, some drawbacks associated with the use of autologous MSCs for transplantation tend to persist. First, the processes of an expanded culture or engineering of MSCs and validation of the final product require a prolonged period of several weeks and high cost. It is difficult to obtain sufficient BM or AD tissues in patients with low body weight and high or low age. Furthermore, MSCs isolated from patients with diabetes, rheumatoid arthritis, or systemic lupus erythematosus have shown impaired function [20]. Conversely, allogeneic MSCs are available as ‘off-the-shelf’ cell products prepared on a large scale. In addition, they can reduce the interdonor heterogeneity, as it is possible to prepare clinical doses of MSCs from one tissue source and avoid off-specification products. Recently, several cell banks of allogeneic MSCs have been established, and GMP-grade MSCs have been easily obtained [30,34,35].

Table 1. Clinical trials of MSC-related cancer gene therapy registered in Clinical Trials.gov. The data were searched in April 2021

No.	Phase/ States	Tumor type	Agents	MSC sources	Doses	Routes	References
NCT01129739	P2 Unknown	MDS	Unmodified	UC-MSC	1 × 10^6 cells/kg, 2 cycles	i.v.	-
NCT01844661	P1/1b Completed	Metastatic and refractory tumors	Oncolytic virus	Autologous BM-derived MSC	2 × 10^6 cells/kg (children) or 0.5‒1 × 10^6 cells/kg (adults)	i.v.	[37]
NCT01983709	P2 Terminated	Prostate tumor (evaluate tumor homing)	Unmodified	Allogeneic BM- derived MSC	2 × 10^6 cells/kg (maximum dose of 2 × 10^8 cells)	i.v.	[99]
NCT02008539	P1/2 Terminated	Advanced gastrointestinal cancer	HSV-tk	Autologous BM-derived MSC	1.5‒3.0 × 10^6 cells/kg	i.v.	[85]
NCT02068794	P1/2 Recruiting	Recurrent ovarian cancer	MV-NIS	AD-derived MSC	-	i.p.	[50]
NCT02079324	P1 Unknown	Head and neck cancer	IL-12	-	-	i.t.	-
NCT02530047	P1 Completed	Advanced ovarian cancer	IFN-γ	-	Starting dose: 1 × 10^5 cells/kg once a week for 4 treatments. Up to 4 dose levels tested	i.p.	-
NCT03184935	P1/2 Suspended	Myelodysplastic syndrome	Unmodified MSCs	UC-MSC	-	i.v.	-
NCT03298763	P1/2 Recruiting	Metastatic non-small-cell lung cancer	TRAIL	Allogeneic BM-derived MSC	Day 1: Cisplatin and pemetrexed Day 2: 4 × 10^8 MSCs This schedule is repeated every 21 days until 3 cycles	i.v.	-
NCT03608631	P1 Recruiting	Metastatic pancreas cancer	Exosomes loaded with siRNA against KrasG12D	-	-	i.v.	-
NCT03896568	P1 Recruiting	High-grade glioma	Oncolytic virus	Allogeneic BM-derived MSC	-	i.c.	-

Abbreviations: AD, adipose tissue; BM, bone marrow; HSV-tk, herpes simplex virus-thymidine kinase; IFN, interferon; IL, interleukin; i.p., intraperitoneal injection; i.t., intratumoral injection; i.c., intracranial injection; i.v., intravenous injection; MDS, myelodysplastic syndromes; MSCs, mesenchymal stem/stromal cells; MV-NIS, oncolytic measles virus encoding thyroidal sodium iodide symporter; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand; UC, umbilical cord.

2.2. Tumor-homing mechanism of MSCs

2.2.1. Tumor-homing mechanism of systemically administered MSCs

The homing of MSCs to tumor tissues demonstrates some similarities to leukocyte migration in inflamed tissues. Systemically administered MSCs are guided by chemokines secreted from solid tumors and activated following interaction with the endothelium [14,15,38,39]. Some chemokines in the blood flow and cell adhesion molecules on the endothelial surface of tumor blood vessels induce the rolling of MSCs, as well as their adherence to the endothelium [15,38,39]. After adhesion, MSCs migrate into tumor tissues through the vessel wall [15,39,40]. Reportedly, several types of chemokine receptors exist on MSCs, such as C-C chemokine receptor (CCR)-1, CCR-2, CCR-4, CCR-7, CCR-9, and C-X-C chemokine receptor (CXCR)-1, CXCR-4, CXCR-6, and CXCR-7 [15,38,39]. CXCR-4, the stromal cell-derived factor-1 (SDF-1) receptor, has been well investigated and is considered to play a major role in MSC homing [15,39]. However, SDF-1 is constantly secreted from the BM and various other tissues. Lourenco et al. reported that the macrophage migration inhibitory factor (MIF)-CXCR-4 axis plays a crucial role in the tumor-homing of MSCs [41]. MIF is overexpressed in several tumor types, and the authors demonstrated that the MIF-CXCR4 axis induced MSC homing to tumor tissues by activating the mitogen-activated protein kinase (MAPK) pathway. Both galectin-1 and very late antigen-4 (VLA-4, integrin α4β1) are known to play key roles during rolling and adhesion [39]. Galectin-1 and VLA-4 on MSCs bind to P-selectin and vascular cell adhesion molecule-1 expressed on the surface of vascular endothelial cells, respectively, and mediate MSC rolling and endothelial adhesion. After adhesion to the endothelium, MSCs express matrix metalloproteinase (MMP)-2 and transmigrate across the vessel wall [39,40]. Typically, the transmigration of leukocytes requires 10‒20 min, whereas that of MSCs takes 60‒120 min [39].

2.2.2. Improvement of MSC tumor-homing

In addition to understanding the tumor-homing mechanism of MSCs, attempts have been made to improve the tumor-homing property of MSCs by employing gene and cell engineering techniques. Several studies have reported that the overexpression of chemokine receptors, including CCR-1, CXCR-4, and CXCR-7, enhanced the homing ability of MSCs to target sites [38,40]. Modification of the cell surface with SLeX, a P-selectin ligand, E-selectin-binding peptide, and antibodies against P-selectin and VCAM1 facilitated the rolling and adhesion of MSCs to the vascular endothelium at the target site, resulting in an improved homing rate [38,40]. Priming methods can upregulate the expression of migration-related genes without gene transfer. Feng et al. revealed that hypoxia priming (1.5% O₂) improved the tumor-homing of MSCs in a glioblastoma mouse model [42]. Under hypoxic conditions, the disassembly of hypoxia-inducible factor-1α was inhibited, and the expression of target genes relevant to cell proliferation, migration, and invasion was increased. Smith et al. reported that pre-exposure of MSCs to a glioma-conditioned medium containing fibronectin and laminin enhanced their tumor-homing in a glioblastoma mouse model [43]. Loading iron oxide NPs to MSCs is another method to improve the homing efficacy of MSCs to the target site via magnetic guidance. Several studies demonstrated that the number of MSCs homing to the target site was increased by the loading of iron oxide NPs to MSCs and by applying the external magnetic field to the target site [35,44–47]. In addition, several studies demonstrated that the expression of migration-related genes of MSCs were upregulated by only iron oxide NPs loading [44,48] or by the combination of the loading and application of the external magnetic field [49].

2.2.3. Tumor-homing of non-systemically administered MSCs

Several studies have reported that non-systemically administered MSCs migrate to tumor tissues. Intraperitoneally injected MSCs reportedly migrate to tumors in orthotopic ovarian tumor model mice [50], peritoneal dissemination model mice [24,51], and TH-MYCN transgenic mice that spontaneously develop glioblastoma in the superior mesenteric ganglion [52]. Furthermore, Rincón et al. reported that intraperitoneally administered MSCs homed to the subcutaneous tumor in a non-small-cell lung carcinoma cell line (CMT64) xenograft mouse model [53]. However, compared with the tumor-homing of systemically administered MSCs, little is known regarding the cascade and molecular mechanism underlying tumor homing of non-systemically administered MSCs.

2.3. Administration methods and fate of MSCs after administration

2.3.1. Transplantation forms of MSCs

MSC transplantation can be classified into three categories: cell suspensions, multicellular spheroids, and cell sheets. A cell suspension is the most basic form of drug administration. Single or glued cells, constructed from two or multiple cells suspended in saline or ringer’s solutions, are administered. Cell spheroids or multicellular spheroids are three-dimensionally cultured cell aggregates, ranging between 100‒500 μm in size. Multicellular spheroids can present innate cellular functions by resembling in vivo conditions, including high cell-cell interactions and the deposition of extracellular matrices in the three-dimensional structure [54]. Smruthi et al. demonstrated that spheroid formation of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-expressing MSCs improved TRAIL expression and the anti-tumor effect in a glioblastoma xenograft mouse model [55]. In addition, cell sheets can improve the innate cellular function and survival rate of transplanted MSCs. However, to our knowledge, few studies have been reported on MSC-based cancer therapy using multicellular spheroids or cell sheets. Therefore, we focus on suspended MSCs and discuss the administration route and biodistribution after administration in the following section.

2.3.2. Administration routes and biodistributions of MSCs

The MSC administration routes are broadly classified into systemic (intravascular) and local administration. Systemic (intravascular) administration route includes intravenous, intra-arterial, and intraportal administration. Most intravenously administered MSCs are temporarily entrapped in the lung vasculature and gradually transferred to the systemic circulation [24,39]. The ex vivo culturing of MSCs increases the cell size and expression of cell adhesion molecules [56], and these morphological and biological changes are considered major causes for the lung entrapment of intravenously administered MSCs [39,56]. Accordingly, the intravenous injection of MSCs can efficiently target lung tumors. Conversely, several studies have demonstrated that reducing the lung entrapment of MSCs improved the homing rate to the targets outside the lungs. They include intravenous administration of MSCs by cell surface coating with heparin [57], avoidance of cell hypertrophy by employing a modified culture method [38,56], and blocking of cell adhesion proteins with antibodies [56]. Intra-arterially injected MSCs are predominantly distributed to first-pass organs, and intraportally administered MSCs are mainly distributed in the liver. Therefore, local intra-arterial and intraportal administrations are effective in targeting tumors in specific organs or tissues. However, both intra-arterial and intraportal administration require invasive surgical intervention, associated with a high risk of bleeding complications. Intravenous administration is the least invasive and useful for repeated administrations. In clinical trials of regenerative medicine, the intravenous administration route is the most frequently employed [25], and many clinical trials have shown the safety of intravenous injection of MSCs (1‒2 × 10⁶ cells/kg) [58].

In a preclinical study of MSC-based drug delivery for cancer therapy, local (non-intravascular) administration routes, including intracranial, nasal, peritoneal, peritumoral, thecal, tumoral, and subcutaneous administration routes, have been reported. Intra-tumoral and peritumoral administration routes are conceptually the simplest and can efficiently deliver MSCs to target tumors. These routes are suitable for easily detectable tumors, such as skin tumors. Intracranial administration is attractive for targeting brain tumors such as glioblastoma, as it can bypass the blood-brain barrier (BBB). Among primary brain tumors, glioblastoma is the most malignant tumor and is not curable with surgery alone, as glioblastoma cells invade surrounding brain tissues. Temozolomide is the only standard anti-cancer drug for glioblastoma, but the five-year survival rate with standard therapy composed of radiotherapy plus temozolomide is only 9.8% [59]. Thus, MSC-based drug delivery is expected to be a novel strategy for improving the efficacy of glioblastoma therapy. Although several studies have reported that systemically administered MSCs cross the BBB and are homed to the brain tumor [15,60], a comparative study by Nakamizo et al. revealed that the intracranial administration of interferon (IFN)-β-expressing MSCs significantly prolonged the survival of glioblastoma model mice when compared with intravenous administration [61]. Intranasal administration of MSCs, which has been reported to bypass the BBB [62], may also present an alternative route for targeting brain tumors.

2.3.3. Disappearance of MSCs from the body

It has been reported that most MSCs disappear from the body by undergoing necrotic cell death within one week of administration [63,64], in which shear stress, lack of nutrients, hypoxia, and detachment-induced apoptosis have been observed [63,65]. Although MSCs can generally escape T cell and natural killer (NK) cell-mediated destruction, MSCs may be phagocytosed by monocytes or macrophages, because some studies demonstrated that MSCs were phagocytosed by monocytes/macrophages under in vitro culture conditions [66,67].This rapid disappearance could avoid unexpected adverse effects attributed to MSCs and may be beneficial to establish the safety of MSCs as a drug delivery vehicle. However, regulating the survival duration of administered MSCs is necessary to deliver therapeutic MSCs to tumor sites efficiently.

2.4. The influence of administered MSC on tumor tissues

2.4.1. Differentiation capacity of MSCs to cancer-associated fibroblasts

Although MSCs are expected to be developed as drug delivery vehicles, several studies have reported that MSCs support the construction of the tumor microenvironment by differentiation into cancer (carcinoma-, tumor-)-associated fibroblasts (CAFs) [23,68], as well as accelerate tumor growth and metastasis by stimulating tumor cells via direct cell-cell interactions and tumor cell-derived extracellular vesicles (EVs) [68–71]. CAFs closely resemble myofibroblasts and express α-smooth muscle actin (α-SMA) and fibroblast activation protein (FAP) [72]. CAFs reportedly promote tumor growth and suppress tumor immunity through matrix remodeling and immune crosstalk [72]. Several studies have demonstrated that CAFs originate from MSCs. Quante et al. and Raz et al. revealed that MSCs recruited from the BM differentiated into CAFs [23,68]. In the tumor tissue, transforming growth factor (TGF)-β1 and SDF-1 induce the differentiation of MSCs into CAFs. These reports suggest that endogenous MSCs naturally possess tumor tropism and pro-tumor activity. In addition, several studies have reported that other tissue-derived MSCs, such as AD-derived MSCs, possess the ability to differentiate into CAF-like cells under in vitro culture conditions [20]. However, it remains unclear whether MSCs can be recruited from tissues other than the BM for differentiation into CAFs.

Although endogenous MSCs may act as pro-tumorigenic cells, MSC administration might minimally impact tumor growth. While induction from MSCs to CAFs required 4‒30 days [73–75], most MSCs disappeared from the body within a week [63,64]. In addition, the differentiation of MSCs increased the expression of MHC molecules [21], which could result in immune rejection. Therefore, the risk that exogenously administered MSCs can be a primary source of CAFs could be low, especially when allogeneic MSCs are employed.

2.4.2. Influence of MSC-derived EVs on tumor progression

EVs, including exosomes, microvesicles, and microparticles, are cell-released and lipid bilayer-delimited nanoparticles [76]. EVs contain various proteins, lipids, and nucleic acids and can mediate cell-cell interactions. Several studies have suggested that EVs secreted from MSCs in tumor tissues promote tumor growth and metastasis. It has been reported that microRNA-15a, 21, 34a, 191, 125a, 140, and 222/223 in MSC-derived EVs improved the resistance to anti-cancer drugs, proliferation, migration of cancer cells, and angiogenesis by tumor stromal cells [69]. However, studies have also reported that MSC-derived EVs possess anti-tumor effects [69]. Therefore, it remains unclear whether MSC-derived EVs exhibit pro- or anti-tumorigenic effects.

Although endogenous MSC-derived EVs may act as pro-tumorigenic vesicles, MSC administration might have little effect on tumor growth. The safe dose of MSCs ranges between 1‒2 × 10⁶ cells/kg [58], whereas MSC-derived EVs were administered at doses equivalent to EVs that can be collected from 4‒7 × 10⁷ cells/kg MSCs [70,71]. Based on these numbers, it can be stated that EVs secreted from administered MSCs have little impact on tumor progression. Therefore, at safe doses of MSCs, pro-tumorigenicity of MSCs is considered extremely low. However, undesired or unexpected adverse effects cannot be completely ruled out, and the effect of MSC transplantation prior to its clinical application should be carefully tested.

2.4.3. Suppression of anti-tumor immunity by MSC administration

The influence of MSCs on tumor immunity needs to be cautiously monitored. MSCs potently suppress the activity of lymphocytes, promote the polarization of monocytes/macrophages toward an anti-inflammatory phenotype, suppress the migration and maturation of dendritic cells, and induce regulatory T cells. The immunomodulatory properties of MSCs are beneficial for treating inflammatory and autoimmune diseases, but may be unbeneficial for cancer treatment. Chen et al. reported that intravenously administered MSCs suppressed T cell immunity via the programmed cell death-1 (PD-1)/PD-L1 pathway and reduced the survival of a multiple myeloma cell line (5TGM1) multiple myeloma model mice [77]. Furthermore, another group reported that MSCs suppressed cell proliferation and IL-6 expression by NK cells in a human hepatocellular carcinoma cell line (BEL7404) and LM3 xenograft model mice [78]. These reports indicate that MSC administration may inhibit the cancer immunity cycle [79] and the cancer-NK cell immunity cycle [80]. Several studies have suggested that MSCs promote the polarization of monocytes/macrophages toward a different subpopulation called tumor-associated macrophages (TAMs), which promote angiogenesis, tumor cell growth and invasion, and suppress cytolytic T cell responses [20,21]. In contrast, the undesirable effect of MSC administration on tumor immunity can be suppressed by gene engineering of MSCs. Chen et al. reported that administration of MSCs suppressed T cell responses and promoted tumor growth; this effect was prevented by knocking down PD-L1 using shRNA [77]. Yu et al. reported that co-transplantation of MSCs and mouse breast cancer cell line 4T1 cells promoted tumor growth, although administration of MSCs overexpressing Sirt1, the closest mammalian homolog of the yeast enzyme Sir2, suppressed tumor growth by recruiting NK cells [81]. Li et al. demonstrated that FAPα is a critical molecule for the recruitment of monocytes/macrophages and polarization into TAM by MSCs [82]. The elucidation of molecular mechanisms underlying interactions between MSCs and immune cells will provide strategies for preventing tumor immune suppression mediated by MSCs.

3. MSC-based drug delivery for cancer therapy

Several studies on MSC-based drug delivery for cancer therapy have been reported. In the following subsections, we describe strategies and issues involved in the application of MSCs as a vehicle for therapeutic genes and anti-cancer drugs.

3.1. Cancer gene therapy using gene-engineered MSCs

3.1.1. Preclinical study of MSC-based cancer gene therapy

Cancer gene therapy involves the administration of vectors carrying a therapeutic gene encoding an anti-tumor protein or the administration of genetically modified cells expressing an anti-tumor protein. As efficient gene delivery to the tumor tissue is a key step in cancer gene therapy, tumor-homing MSCs have been considered a vehicle for the therapeutic gene (Table 2, Figure 2A). In this subsection, we describe some therapeutic genes, as well as their therapeutic mechanisms in MSC-based cancer gene therapy. In 2002, Studeny et al. demonstrated that the administration of IFN-β-expressing MSCs suppressed tumor growth [83]. Elevated tumor concentrations of IFN-β can suppress tumor cell growth by inducing apoptosis, cell cycle arrest, and activation of the immune response. In general, the systemic administration of IFN-β has shown limited therapeutic effects owing to its short half-life in blood and low delivery efficiency to the tumor site. The authors demonstrated that subcutaneous co-injection of IFN-β-expressing MSCs markedly reduced the volume of the human melanoma cell line A375SM tumor in mice. This result suggests that MSCs can sustainably release therapeutic proteins at the tumor site. Furthermore, intravenously injected IFN-β-expressing MSCs were detected inside the tumor tissue in a lung metastasis mouse model, and intravenous cell injections prolonged the survival of lung metastasis model mice. These results demonstrate the potential of cancer gene therapy using MSCs. TRAIL is a cytokine of the tumor necrosis factor (TNF) family that initiates apoptosis by binding to its death receptors and selectively induces apoptosis of cancer cells. Previous studies have demonstrated that TRAIL-expressing MSCs show considerable anti-tumor effects in tumor-bearing mice [15,20]. Reportedly, herpes simplex virus-thymidine kinase (HSV-tk), a prodrug-activating suicide gene, phosphorylates ganciclovir, a prodrug antiviral agent, in HSV-tk-expressing cells and induces cellular apoptosis. This suicide system is widely applied to eliminate transplanted cells from the host to enhance the safety of cell-based therapy and regulate the function of transplanted cells [8,84]. Phosphorylated ganciclovir in HSV-tk-expressing cells is transferred to neighboring cells via gap junctions and induces apoptosis of neighboring cells [8]. Phosphorylated ganciclovir reportedly competes with deoxyguanosine triphosphate and thus can selectively induce apoptosis in actively dividing cells such as tumor cells without significantly impacting normal tissues/cells. The regulation of gene expression by specific promoters enables avoiding the ‘off-target effect’ and improving the safety of the MSC-based cancer gene therapy. Tie-2, CCL5 (RANTES), and alpha-fetoprotein promoter have been employed in preclinical and clinical studies [85–87]. Oncolytic viruses are defined as viruses that selectively replicate in and kill cancer cells and are harmless to normal tissues/cells. Once oncolytic virus infection occurs, an anti-tumor immune response is activated, and immunogenic cell death of the infected cells and neighboring tumor cells is induced. Thus, oncolytic virus therapy can be expected to afford a vaccination effect [88]. MSCs can be developed as a suitable vehicle for oncolytic viruses, as several oncolytic viruses are immunogenic and lack tumor tropism [88]. However, many genetically engineered oncolytic viruses are designed to be activated in cancer cells. Some oncolytic viruses have been modified to efficiently infect and replicate MSCs [89].

Figure 2. Manufacture of engineered MSCs for anti-cancer therapy. Autologous MSCs are isolated from the patient’s tissue, including BM and AD. The number of MSCs that can be collected differ between source tissues. Several weeks are needed for MSC expansion and gene engineering. After drug loading and quality check, therapeutic gene-modified or drug-loaded MSCs are administered to the patient. Allogeneic MSCs are induced from allogeneic iPS cells or isolated from healthy donor tissue such as BM, AD, and UC. Although allogeneic iPS cell-derived MSCs were previously employed, a project for establishing cell banks for autologous iPS cells is ongoing [98,139]. iPS cells can supply greater doses of MSCs from a single cell bank. Allogeneic tissue-derived- and iPS cell-derived MSCs are processed to the final product in the factory. The stored allogeneic MSC-product can be supplied on demand. AD, adipose tissue; BM, bone marrow; iPS, induced pluripotent stem; MSCs, mesenchymal stem/stromal cells; UC, umbilical cord

Table 2. Gene types and their therapeutic mechanisms in MSC-based cancer gene therapy

Categories	Agents	Anti-tumor mechanisms	Experimental types	Ref.
IFN	IFN-α	Activate immune systems and directly inhibit tumor cell growth	In vitro In vivo	[15,22]
	IFN-β			[15,22, 38,83]
	IFN-γ			[90]
IL	IL-2	Activate immune systems	In vitro In vivo	[38]
	IL-7		In vivo	[22]
	IL-7 + 12		In vivo	[91]
	IL-12		In vivo	[15,22]
	IL-18		In vitro	[22]
	IL-18 + IFN-β		In vivo	[92]
	IL-21		In vivo	[21]
	IL-24		In vitro In vivo	[93]
TNF superfamily	TRAIL	Activate immune systems and directly inhibit tumor cell growth	In vitro In vivo	[15,22, 38,55]
Chemokine	CX3CL	Activate immune systems	In vivo	[38]
	CCL9		In vivo	[94]
Apoptosis-inducing protein	Apoptin	Inhibit angiogenesis	In vitro In vivo	[95]
Anti-angiogenic factor	Pigment epithelium-derived factor	Inhibit angiogenesis	In vivo	[15]
Antagonist of HGF	NK4	Inhibit HGF-mediated cell proliferation, migration, invasion of tumor cells	In vitro In vivo	[38]
Tissue-specific transcription factor	Hepatocyte nuclear factor 4α	Inhibit the growth of hepatocellular carcinoma	In vitro In vivo	[15]
Growth factor	Bone morphogenetic protein 4	Decrease the tumorigenicity of brain tumor initiating cells	In vivo	[96]
Suicide gene	HSV-tk	Activate ganciclovir and induce apoptosis of HSV-tk expressing cells and neighboring cancer cells	In vitro In vivo	[15]
	Cytosine deaminase	Convert 5-fluorocytosine to 5-fluorouracil	In vitro In vivo	[38]
Prodrug activating gene	Carboxylesterase	Convert the prodrug irinotecan into the active drug SN-38	In vitro In vivo	[51]
Oncolytic virus	Adenovirus	Lysis of tumor cells and induce immunogenic cell death	In vitro In vivo	[88]
	HSV
	MV
	MYXV
microRNA	miR 124, 145	Deliver via EVs	In vitro	[97]

Abbreviations: CCL, CC chemokine ligand; EV, extracellular vesicle; HGF, hepatocyte growth factor; HSV, herpes simplex virus; HSV-tk, herpes simplex virus-thymidine kinase; IFN, interferon; IL, interleukin; MSCs, mesenchymal stem/stromal cells; MV, measles virus; MYXV, myxoma virus; NK4, a four-kringle antagonist of hepatocyte growth factor; TNF, tumor necrosis factor; CX3CL, C-X3-C motif chemokine ligand.

3.1.2. Clinical trials of MSC-based cancer gene therapy

The study of MSC-based cancer gene therapy has entered the clinical stage. According to Clinical Trials.gov, 11 clinical trials on MSC-based cancer gene therapy have been registered (Table 1, the data were searched in April 2021.). All studies in phases I and II trials evaluated the safety and tolerance of the administered gene-modified MSCs. A phase I/II clinical trial by von Einem et al. demonstrated the safety of intravenous administered HSV-tk-expressing autologous MSCs and ganciclovir (NCT02008539). Another phase I trial revealed the safety of oncolytic virus-loaded autologous MSCs (NCT01844661). However, no therapeutic effects were observed in these studies. Although several preclinical studies have demonstrated significant anti-tumor effects, these effects have not been corroborated in most clinical trials. This can be attributed to numerous possible reasons, one of which is the MSC dose employed. Niess’s group administered HSV-tk expressing MSCs at a dose of 5 × 10⁵ cells/mouse to hepatocellular carcinoma xenograft model mice, which is equivalent to 2.5 × 10⁷ cells/kg [85]; however, the mean dose in clinical trials was 2.65 × 10⁶ cells/kg [85]. The low homing efficiency and interdonor heterogeneity of MSCs could affect their therapeutic potential. Schweizer et al. reported that intravenously administered MSCs were not detected at the tumor site in patients with prostate cancer [99]. This result indicates the necessity for improving the homing ability of MSCs. Melen et al. reported that clinical outcomes were influenced by the expression of chemokine receptors related to migration. Oncolytic virus-loaded autologous MSCs were intravenously administered to 13 pediatric patients. Clinical responses were observed in 5 patients, with no response recorded in 8 patients. The authors compared the gene and protein expression profiles of cells between responders and non-responders and reported that the expression levels of CCR1, CXCR1, and CXCR4 were higher in the response group than that in non-responders [100].

3.2. Application of MSCs as a vehicle for anti-cancer drugs and functionalized nanoparticles

3.2.1. MSC loading of anti-cancer drugs

MSCs have been employed not only as gene vehicles but also as vehicles for anti-cancer drugs such as doxorubicin (DOX), paclitaxel (PTX), and gemcitabine (Table 3). In most studies, free anti-cancer drugs or drug-encapsulated nanoparticles (NPs) were loaded into the cytoplasm of MSCs through passive diffusion or endocytosis (Figure 3A). Free anti-cancer drugs can be easily loaded into MSCs by simple incubation with anti-cancer drugs for several hours [101]. Then, loaded drugs are released from MSCs via passive diffusion, drug efflux transporters such as p-glycoprotein (P-gp) [14,102], EVs [101], or tunneling nanotubes [103] (Figure 3B). The expression of P-gp reportedly varies (Table 4), suggesting substantial interdonor heterogeneity in expression. Some studies have revealed that anti-cancer drugs were released from MSCs via P-gp; however, P-gp expression was scarcely examined in most of these studies. Free anti-cancer drug-loaded MSCs suppressed tumor cell growth following co-culture with tumor cells and local administration to tumor-bearing mice [101,102,104–107]. However, few studies have succeeded in suppressing tumor growth following systemic administration, and it may be challenging to efficiently deliver drugs using free drug-loaded MSCs owing to the short retention time of the loaded drug. Although drugs are rapidly released from drug-loaded MSCs, systemically administered MSCs reach the tumor tissue approximately 4‒6 h after administration [24,108]. The use of drug-encapsulated NPs that sustainably release the drug can improve the retention time of drug loading in the cell. Zhang et al. reported that the loading of polyamidoamine dendrimer-DOX (DOX-PAMAM) could delay DOX release from MSCs by 12‒48 h when compared with free DOX [109]. Some NPs, such as PAMAM dendrimers, silica NPs, and liposomes, may be suitable for improving drug retention in MSCs due to their low initial burst release. NPs loaded to MSCs are released extracellularly in an intact form via exocytosis and EVs [48,110–112] or may only release the drug after disassembling the NPs in the cellular endosome/lysosome [112,113] (Figure 3). However, the drug delivery efficiency of NP-loaded MSCs to tumor cells was not high. The in vitro anti-tumor effect of NP-loaded MSCs was lower than that of NPs, possibly due to poor drug release from MSCs. For example, the IC₅₀ values of PTX-PLGA-NPs and PTX-PLGA-NP-loaded MSCs against the human lung adenocarcinoma cell line A549 were 2.1 nM and 22 nM, respectively [114]. The stimulus-triggered drug release system is useful for improving drug release from NP-loaded MSCs. Huang et al. reported that PTX release from SPION-PTX-NP-loaded MSCs was improved by high-frequency magnetic field treatment [60]. This drug release has been attributed to magnetothermally induced structural disruption of SPION-PTX-NP and necrosis of NP-loaded MSCs. Apart from the low in vitro anti-tumor effects, several studies have successfully suppressed tumor growth in tumor-bearing mice by systemic administration of NP-loaded MSCs; the observed anti-tumor effect was superior to that of NPs owing to the improved delivery of anti-cancer drugs to the tumor tissue.

Figure 3. Drug-loading methods of MSCs and the drug delivery route from drug-loaded MSCs to cancer cells. (a) Drug-loading methods for the application of MSCs as a drug delivery vehicle. Free anti-cancer drugs and drug-encapsulated NPs can be loaded on MSCs through passive diffusion, specific transporters, and endocytosis. Drug-encapsulated NPs can be modified to the surface of MCSs using cell surface modification methods. (b) The intercellular drug delivery mechanism. NPs modified on the surface of MSCs are taken up by neighboring cancer cells through endocytosis. Drugs loaded within MSCs are directly delivered to neighboring cancer cells through gap-junction and tunneling nanotubes (direct delivery). MSCs release loaded drugs to the outside of cells through passive diffusion, specific transporters, exocytosis, and EVs, and released drugs are then taken up by cancer cells (indirect delivery). EVs, extracellular vesicles; MSCs, mesenchymal stem/stromal cells; NPs, nanoparticles

Table 3. Loading methods and therapeutic applications of drug-loaded MSCs

Loading method	Loaded agents	Experimental models	Administration route	References (year)
Internalization	Coumarin-6-loaded PLA NPs	Glioma model (cell tracking)	i.t.	[115] 2010
	PTX	Subcutaneous tumor model	Co-injection with cancer cells (s.c.)	[102] 2011
	PTX-loaded chitosan NPs	In vitro	-	[116] 2013
	PTX	In vitro	-	[101] 2014
	PTX-loaded PLGA NPs	In vitro	-	[113] 2014
	A photosensitizer-loaded silica NPs, producing ROS by light irradiation	Subcutaneous tumor model	Co-injection with tumor cells (s.c.)	[117] 2014
	Gemcitabine	In vitro	-	[104] 2015
	PTX	In vitro	-	[105] 2015
	DOX-loaded, RGD-modified PAMAM	Glioma model	i.c. and i.t.	[109] 2015
	Au NPs, increasing their temperature by NIR irradiation	Subcutaneous tumor model	i.v.	[118] 2015
	G144-loaded PLGA microparticles	Subcutaneous tumor model	i.v.	[119] 2016
	PTX and gold nanorod-loaded organosilica NPs	Subcutaneous tumor model	i.t.	[120] 2016
	PTX, DOX or gemcitabine	In vitro	-	[106] 2017
	Sorafenib	Glioblastoma model	i.n.	[107] 2017
	DOX-loaded PLGA NPs	Lung cancer model	i.v.	[110] 2017
	DOX-SPION-loaded NPs	Glioblastoma model	i.v.	[60] 2017
	DOX-loaded PLGA NPs	Lung cancer model	i.v.	[111] 2018
	PTX-loaded PLGA NPs	Lung cancer model	i.v.	[121] 2018
	DOX-loaded NPs composed of transferrin and albumin	Subcutaneous tumor model	i.v.	[112] 2018
	Plasmonic-magnetic hybrid NPs loaded with DOX, gold nanorods, and iron oxide nanocluster	Subcutaneous tumor model	i.t. and i.v.	[48] 2018
	Vincristine-loaded silica NPs	In vitro	-	[122] 2019
	Pirarubicin-loaded PLGA NPs	Subcutaneous tumor model	i.t. or peritumoral	[123] 2019
	TAT peptide-functionalized PTX-loaded PLGA NPs	Lung tumor model	i.v.	[114] 2019
	Au NPs increased their temperature by NIR irradiation	Subcutaneous tumor model	i.v.	[124] 2019
	A photosensitizer-loaded PMMA NPs	Subcutaneous tumor model	i.t.	[125] 2020
Cell surface modification	DOX-loaded silica NPs	Subcutaneous tumor model	i.t.	[126] 2011
	Curcumin-loaded chitosan NPs	Lung tumor model	i.v.	[127] 2019
	PTX-loaded PLGA NPs	Peritoneal dissemination model	i.p.	[134] 2019
	DOX-loaded liposomes	Subcutaneous and lung tumor model	i.t. and i.v.	[138] 2021
Hybrid	iRGD-modified DOX-PAMAM	Lung tumor model	i.v.	[128] 2017

Abbreviations: DOX, doxorubicin; G144, an activated prodrug activated prodrug; i.c., intracranial injection; iRGD, a tumor homing peptide (CRGDKGPDC); MSCs, mesenchymal stem/stromal cells; NPs, nanoparticles; NIR, near-infrared; PLA NPs, poly lactic acid nanoparticles; PLGA, poly(lactic-co-glycolide); PMMA NP, poly methyl methacrylate core-shell fluorescent nanoparticles; PTX, paclitaxel; ROS, reactive oxygen species; s.c., subcutaneous injection; SPION, superparamagnetic iron oxide nanoparticles; TAT, trans-activator of transcription protein.

Table 4. Heterogeneity of P-gp expression in MSCs

P-gp expression	Detection methods	MSC sources	Surface markers	Ref.
+	Flow cytometric analysis	Mouse BM	CD5^+, 44^+, 45 R^+, 73^+, sIG^+	[129]
+	Flow cytometric analysis	Human BM	CD13^+, 44^+, 73^+, 90^+, 105^+, HLA-I^+ CD14^−, 31^−, 33^−, 34^−, 45^−, 80^−, HLA-II^−	[102]
+	Western blotting	Rat BM	CD90^+, 29^+ CD34^−, 45^−	[116]
+	Western blotting	Rat BM	CD44^+, 29^+ CD34^−	[130]
+	Flow cytometric analysis and RT-PCR	Human BM	Positive cell ratio: CD34: 4.5%, CD45: 10.7%, CD29: 81.7%, CD105: 69.2%	[131]
±*	Flow cytometric analysis	Human AD	CD90^+, 73^+, 44^+, 13^+, 105^+ CD45^−, HLA-DR^−	[105]
-	Western blotting	-	-	[132]

* Analyzed MSCs from the seven donors; P-gp expression was observed in six out of seven donors.

Abbreviations: AD, adipose tissue; BM, bone marrow; CD, cluster of differentiation; HLA, human leukocyte antigen; MSCs, mesenchymal stem/stromal cells; P-gp, p-glycoprotein; RT-PCR, reverse transcription-polymerase chain reaction; sIG, surface immunoglobulin.

3.2.2. Surface modification of MSCs with anti-cancer drugs

Internalization of free anti-cancer drugs and anti-cancer drug-encapsulated NPs is cytotoxic to MSCs and affects the viability and tumor tropism of MSCs [60,102,105,107,109,110]. Some studies have successfully modified the surface of MSCs with anti-cancer drug-encapsulated NPs and reduced the cytotoxicity of loaded drugs toward MSCs (Table 3, Figure 2C). Layek et al. successfully modified the cell surface of MSCs with PTX-PLGA NPs without inducing cytotoxicity by employing a strain-promoted [3 + 2] azide-alkyne cycloaddition (SPAAC) reaction [133], a biorthogonal reaction [134]. Cell surface modification with PTX-PLGA NPs marginally affected the migratory ability of the cells to the tumor cell-conditioned medium. Furthermore, intraperitoneal administration of PTX-NP-loaded MSCs demonstrated a higher therapeutic effect than that of NP administration in peritoneal dissemination model mice, established by intraperitoneal injection of a human ovarian cancer cell line, MA148 cells. Takayama et al. demonstrated that cell surface modification using the avidin-biotin complex method could quickly modify the cell surface with DOX liposomes without significantly influencing the viability, migration, and tumor-homing ability of MSCs [135]. Cell surface modification using the avidin-biotin complex method could stably modify cells with avidinated or biotinylated compounds; over 80% of modified DOX liposomes remained loaded in MSCs at 24 h. Furthermore, DOX liposomes on the surface of MSCs were directly taken up by neighboring murine colon adenocarcinoma colon26 cells via endocytosis (Figure 3), efficiently suppressing the growth of colon26 cells in subcutaneous xenograft and lung metastasis mouse models. Although few studies have focused on cell surface modification-based drug loading when compared with loading inside cells, these studies indicate that cell surface modification with anti-cancer drugs can be valuable for efficient cancer-targeted drug delivery.

4. Conclusion

MSCs are a useful drug delivery vehicle for cancer therapy owing to their innate properties, including tumor tropism, self-renewal or proliferation potency, low immunogenicity, and lack of tumorigenicity. Several studies have demonstrated that MSCs can deliver therapeutic genes to tumor sites and present excellent anti-tumor effects in preclinical trials [14,15,83]. In addition, recent studies have demonstrated that anti-cancer drug-loaded MSCs possess remarkable anti-tumor effects in various in vivo experiments. Although some issues need to be addressed, MSC-based drug delivery will afford an effective cancer-targeting therapy in the future.

5. Expert opinion

MSCs are expected to be developed as a useful drug delivery vehicle for cancer-targeting therapy, especially for treating cancers with a poor prognosis, such as glioblastoma, multi-organ metastases, peritoneal dissemination, and relapsed/refractory cancers. To obtain reliable therapeutic outcomes with MSCs, improving some aspects of current MSC-based therapy remains crucial.

First, the heterogeneity of MSCs needs to be addressed. Donor-dependent variations in tumor tropism and self-renewal or proliferation potency of MSCs can markedly affect the therapeutic effect of MSC-based therapy. This is because obtaining MSCs with high tumor tropism on a clinical scale from every patient is challenging. Therefore, enhancing the migration ability and improving the self-renewal or proliferation potency of MSCs by employing gene or cell engineering techniques should be considered. Furthermore, the use of allogeneic tissue-derived MSCs can reduce interdonor heterogeneity and allow the preparation of adequate clinical doses of MSCs from a single tissue source after confirming their properties. iPS cells are also an attractive cell source for MSCs, and the therapeutic applications of iPS-derived homogenous MSCs prepared on a large scale will be expanded. In addition, as several previous studies have reported [136,137], the quality control of ex vivo expansion process is one of the keys in the manufacturing process of homogenous MSC products.

Second, the tumor delivery efficiency of MSCs should be improved by regulating their biodistribution. Although the safety of intravenous MSC administration has been demonstrated in some clinical trials, the homing rate of intravenously administered MSCs to the tumor site is as low as 1% or less, except for lung tumors, mainly due to their entrapment in the lung vasculature. Accordingly, some completed clinical trials assessing MSC-based cancer gene therapy failed to demonstrate significant therapeutic effects. In several preclinical studies, intravenously administered gene-modified and drug-loaded MSCs showed significant anti-tumor effects at high doses (0.5‒2 × 10⁶ cells/mouse, equal to 1.5‒6 × 10⁷ cells/kg [138]); however, the safe dose in humans is 1‒2 × 10⁶ cells/kg. Therefore, it may be difficult to considerably increase the MSC dose to avoid the risk of developing adverse effects, such as thrombosis and pulmonary embolism. In order to achieve therapeutic efficacy within the safe dose range of MSCs, it is crucial to regulate the biodistribution and improve the tumor-homing efficiency of administered MSCs. Several studies have succeeded in avoiding lung entrapment and improving the tumor tropism of MSCs by using gene or cell engineering techniques or optimizing MSC culture conditions. The development of these techniques will be one of the keys to achieving successful clinical outcomes of therapeutic gene-modified and drug-loaded MSCs for targeted cancer therapy.

Third, the MSC-based drug delivery should be optimized. Over the last ten years, several methods for loading anti-cancer drugs to MSCs have been developed, and some studies have achieved high loading efficiency of anti-cancer drugs without affecting the MSC characteristics, along with efficient delivery of anti-cancer drugs to the tumor site. However, the efficiency of anti-cancer drug delivery from MSCs to tumor cells in the tumor tissue is extremely low and needs to be improved. Recently, some studies, including our previous report, have elucidated the drug delivery mechanism from drug-loaded MSCs to cancer cells [101,111,135]. Understanding the drug delivery mechanisms and optimizing drug delivery will significantly improve the clinical translation of MSC-based drug delivery.

Article highlights

• Cells with tumor-homing ability are expected to act as a tumor-targeted drug delivery vehicle.

• Mesenchymal stem/stromal cells (MSCs) are considered the most attractive candidate drug delivery vehicles for cancer therapy owing to their tumor tropism.

• Many preclinical studies have demonstrated that gene- or anti-cancer drug-loaded MSCs suppress tumor growth.

• Current clinical trials assessing MSCs have reported limited therapeutic effects.

• The key to achieving a reliable outcome with MSC-based cancer therapy is to adjust the heterogeneity of MSCs and optimize the drug delivery efficiency from MSCs to cancer cells with the regulation of their biodistribution.

Abbreviations

Table

ADCs	Antibody-drug conjugates
α-SMA	α-Smooth muscle actin
AD	Adipose tissue
BBB	Blood-brain barrier
BM	Bone marrow
CAFs	Cancer-associated fibroblasts
CCR	C-C chemokine receptor
CXCR	C-X-C chemokine receptor
DOX	Doxorubicin
EPR	Enhanced permeability and retention
ES	Embryonic stem
EVs	Extracellular vesicles
FAP	Fibroblast activation protein
GMP	Good manufacturing practice
GVHD	Graft versus host disease
HSV-tk	Herpes simplex virus-thymidine kinase
IC50	Half maximal inhibitory concentration
IFN	Interferon
iPS	Induced pluripotent stem
ISCT	International Society for Cell & Gene Therapy
MHC	Major histocompatibility complex
MMP	Matrix metalloproteinase
MSCs	Mesenchymal stem/stromal cells
NK	Natural killer
NPs	Nanoparticles
PAMAM	Polyamidoamine
PLGA	Poly(lactic-co-glycolide)
PD-1	Programmed cell death-1
P-gp	P-glycoprotein
PTX	Paclitaxel
SDF-1	Stromal cell-derived factor-1
SPAAC	Strain-promoted [3 + 2] azide-alkyne cycloaddition
TAMs	Tumor-associated macrophages
TGF	Transforming growth factor
TRAIL	Tumor necrosis factor-related apoptosis-inducing ligand
UC	Umbilical cord
VLA-4	Very late antigen-4

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This work was funded by a Grant-in-Aid for Young Scientists (B) (No. 15K18850 and 17K15437) and a Grant-in-Aid for JSPS Fellows (20J14253) from the Japan Society for the Promotion of Science (JSPS).