Concise Review: Genetically Engineered Stem Cell Therapy Targeting Angiogenesis and Tumor Stroma in Gastrointestinal Malignancy§


  • Emily Z. Keung,

    1. Department of Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA
    Search for more papers by this author
  • Peter J. Nelson,

    1. Medizinische Klinik und Poliklinik IV, AG Clinical Biochemistry, Ludwig-Maximilians-University, Munich, Germany
    Search for more papers by this author
  • Claudius Conrad

    Corresponding author
    1. Department of Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA
    • Harvard Medical School, Harvard Stem Cell Institute, Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts 02114, USA
    Search for more papers by this author
    • Tel.: +1-617-726-2000; Fax: 617-724-3895

  • Author contributions: E.Z.K.: conception and design, collection and/or assembly of data, manuscript writing, and final approval of manuscript; P.J.N.: collection and/or assembly of data, manuscript writing, and final approval of manuscript; C.C.: conception and design, manuscript writing, and final approval of manuscript.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS November 6, 2012; available online without subscription through the open access option.


Cell-based gene therapy holds considerable promise for the treatment of human malignancy. Genetically engineered cells if delivered to sites of disease could alleviate symptoms or even cure cancer through expression of therapeutic or suicide transgene products. Mesenchymal stem cells (MSCs), nonhematopoietic multipotent cells found primarily in bone marrow, have garnered particular interest as potential tumor-targeting vehicles due to their innate tumortropic homing properties. However, recent strategies go further than simply using MSCs as vehicles and use the stem cell-specific genetic make-up to restrict transgene expression to tumorigenic environments using tumor-tissue specific promoters. This addresses one of the concerns with this novel therapy that nonselective stem cell-based therapy could induce cancer rather than treat it. Even minimal off-target effects can be deleterious, motivating recent strategies to not only enhance MSC homing but also engineer them to make their antitumor effect selective to sites of malignancy. This review will summarize the advances made in the past decade toward developing novel cell-based cancer therapies using genetically engineered MSCs with a focus on strategies to achieve and enhance tumor specificity and their application to targeting gastrointestinal malignancies such as hepatocellular carcinoma and pancreatic adenocarcinoma. STEM CELLS2013;31:227–235


More than 1.5 million patients were diagnosed with and more than 550,000 patients died of malignant disease in the US in 2011 [1]. While our increasing understanding of tumor biology has facilitated the development of new cancer treatments, major limitations remain including short drug half-lives, insufficient delivery to some tumor types, and suboptimal specificity for malignant tissue [2]. The use of transgene protein therapeutics produced continuously at malignant sites would solve half-life and stability issues. To avoid deleterious effects on normal tissue, specificity and targeted delivery of these therapeutics is crucial. Two approaches may improve therapeutic specificity: (a) elucidating and targeting critical molecular pathways specific and essential to tumor growth [3] and (b) imparting or enhancing targeting specificity of cancer therapeutics to desired sites of action (i.e., tumors) [2, 4, 5]. Cell-based anticancer therapy is a novel strategy for targeting solid malignancy, which has garnered growing interest and can affect both approaches. Mesenchymal stem cells (MSCs) in particular have shown promise due to their tumortropic properties [6-14].

This review will discuss (a) methodologies of genetically modifying stem cells to introduce therapeutic genes products, (b) strategies to achieve tumor-targeting specificity, and (c) application of these concepts and technologies to target gastrointestinal malignancies.


MSCs are defined by their ability to self-renew and differentiate into a variety of cell types [15]. They (a) remain plastic adherent when maintained under standard culture conditions, (b) undergo osteogenic, adipogenic, and chondrogenic differentiation, (c) express CD73, CD90, and CD105 markers, and (d) do not express hematopoietic lineage markers [16]. MSCs can be isolated from bone marrow, umbilical cord blood, adipose tissue, and peripheral blood [17]. Important reviews on the differentiation capacity, immunologic features, and homing behavior of MSCs and vectors used to genetically modify MSCs are summarized in Tables 1 [15] and 2 [18].

Table 1. Summary of reviews
  1. Abbreviations: MSC, mesenchymal stem cell; EPC, endothelial progenitor cell.

inline image
Table 2. Gene delivery systems for stem cell engineering
inline image

A critical property of MSCs for cell therapies is their intrinsic homing properties, defined as recruitment, deceleration, and arrest within the vasculature of a tissue followed by transmigration across the endothelium [23]. When transplanted systemically, MSCs home to tumor/metastases and to sites of injury, inflammation, and ischemia [15, 23]. Unlike the well-characterized leukocyte adhesion cascade that defines leukocyte homing, [23] a definitive MSC homing mechanism has yet to be fully elucidated. Studies have implicated the role of cytokines and chemokines vascular endothelial growth factor (VEGF) [26, 27]; CC chemokine ligand 2 (CCL2) [9, 28, 29]; CCL5/RANTES [9, 28-30]) secreted by target tissues and tumors in progenitor cell mobilization and chemotaxis and adhesion molecules and their ligands (integrin α4B1/Very Late Antigen-4 (VLA-4), ligands vascular cell adhesion molecule 1 (VCAM) and cellular fibronectin [5, 31, 32]) in MSC rolling, arrest, and transmigration along endothelium (Fig. 1) [9, 15, 23, 33]. After transmigration, MSCs contribute directly to tumor growth by acting as progenitors for stromal cells. They can acquire endothelial-like characteristics and are involved in regulating vasculogenesis [34]. In addition, they contribute to the generation of carcinoma-associated fibroblasts [35]. The next section will focus on this role of MSCs in cancer initiation and progression.

Figure 1.

MSCs are recruited to tumor and metastases along chemokine gradients. Although a definitive MSC homing mechanism has yet to be fully elucidated, it has been shown that certain chemokines (VEGF, CCL2, and CCL5/RANTES) and ANGPT2 released into circulation guide migration and homing of receptor-positive endothelial progenitor cells and MSCs to tumors. MSC (A) recruitment to sites of tumor, (B) rolling, (C) arrest within the vasculature, and (D) transmigration across endothelium may be mediated by mechanisms and adhesion molecules that resemble leukocyte homing and adhesion cascade. Abbreviations: ANGPT2, angiopoietin 2; CCL, CC chemokine ligand; MSC, mesenchymal stem cell; VEGF, vascular endothelial growth factor.


The importance of angiogenesis and tumor–stromal interactions for solid tumor growth is well recognized [19, 20, 36–38] and are targets for new classes of cancer therapies, including small molecule pharmaceuticals, biologics, and most recently, stem cell-based cell therapy. Recently, a model has emerged linking tumor neoangiogenesis and recruitment of progenitor cells: tumor cells secrete VEGF and increasing VEGF serum levels mobilize progenitor cells from sources such as bone marrow into the bloodstream. Tumor-resident endothelial cells produce a variety of chemokines (CCL-2, CCL-5/RANTES, stromal cell-derived factor-1 (SDF-1)) and angiopoietin 2 (ANGPT2) and their cognate cell surface receptors in turn stimulate their own proliferation [19, 25, 27, 28, 39]. Chemokines and ANGPT2 released into circulation guide migration and homing of receptor-positive endothelial progenitor cells and MSCs to tumors [9, 11, 21, 26, 40]. These cells have been shown to differentiate and become incorporated into tumor stroma and neovasculature, releasing soluble factors that act in autocrine and paracrine fashion to promote further neoangiogenesis (Fig. 2) [19, 34, 40, 41]. A detailed understanding of this homing mechanism and subsequent integration into the tumor microenvironment may facilitate the development of therapies using innate or user-enhanced homing behaviors of progenitor cells to the tumor environment.

Figure 2.

MSCs and their role in cancer initiation and progression. After transmigration, MSCs may contribute directly to tumor growth by (A) acting as progenitors for tumor stroma, (B) neovasculature, and CAFs. (C) Tumor-resident endothelial cells and stromal cells produce chemokines (CCL-2, CCL-5) and angiopoietin 2 which act in autocrine fashion to stimulate their own proliferation and in paracrine fashion to further recruit progenitor cells such as MSCs. Abbreviations: CAF, cancer-associated fibroblast; CCL, CC chemokine ligand; MSC, mesenchymal stem cell.


Enhancing Homing to Target Tissues

The current model of MSC homing to tumor sites has informed several therapeutic strategies. Several studies suggest that local irradiation increases recruitment of circulating MSCs to the tumor microenvironment [7, 35, 42]. Total body irradiation and/or local irradiation followed by human MSC (hMSC) transplantation in a nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model increased homing and engraftment at sites exposed to higher levels of irradiation [42]. In a murine mammary carcinoma (4T1) model, MSCs preferentially home and engraft to the irradiated hind limb compared to the nonirradiated contralateral hind limb and were identified in tumor stroma, intravascular structures, and tumor parenchyma [7].

The enhanced distribution of MSCs within irradiated tumors has been suggested to result from enhanced vascular permeability following irradiation [43]. Others have found enhanced MSC tropism and engraftment as a result of soluble gradients of paracrine factors. Tumor cells secrete cytokines in response to radiation leading to chemokine receptor upregulation on MSCs and enhanced chemotaxis toward chemokine ligand-bearing tumors [7]. Cytokines released by irradiated tumor cells (transforming growth factor-B1, VEGF, platelet-derived growth factor-BB) enhance the migratory properties of MSCs [7]. Furthermore, in the presence of irradiated tumor cells, MSCs upregulate CC chemokine receptor 2, CCR2. CCR2 plays a critical role in migratory cell mobilization and recruitment to inflammatory sites and its ligand (CCL2) is expressed by tumor types, including gastrointestinal cancers.

Another strategy to enhance MSC homing and engraftment has been studied in models of cardiac ischemia. In vitro “priming” of MSCs (culturing and preconditioning MSCs [23, 44–48] in the presence of soluble factors such as insulin-like growth factor or hypoxia) leads to upregulation of cell surface chemotactic receptors and migratory capacity. MSCs have also been genetically engineered to overexpress molecules involved in homing, including C-X-C chemokine receptor type 4, CXCR4, prior to their transplantation [49–53].

Control of Transgene Expression Using Tissue-Specific Promoters

While therapeutic strategies using genetically modified MSCs have generally used a constitutively expressed transgene and relied on MSC homing mechanisms alone for tumor targeting, additional specificity can be achieved using tissue-specific promoters/enhancers to drive transgene expression once therapeutic MSCs have reached their target tissues. Early examples used the carcinoembryonic antigen promoter/enhancer to drive transgene expression [54, 55]. More recently, selective targeting of therapeutic gene expression using the tumor environment to activate transgene expression has been shown to be is feasible and may limit side effects should MSCs migrate to other tissue niches [56–59].

Angiogenesis is essential for tumor initiation and progression. This process is dependent on the ANGPT-TIE system, which is necessary for the angiogenic switch in tumors [20]. ANGPT2 is secreted by endothelial cells at sites of active vascular remodeling, as occurs in tumors, and functions in an autocrine manner via TIE2 (Tyrosine kinase with immunoglobulin-like and EGF-like domains 2). TIE2 is predominantly expressed on endothelial cells but is found on cancer cells, tumor-associated macrophages, and TIE2-expressing monocytes (TEMs). Hypoxia-induced ANGPT2 expression mediated by TIE2 may be responsible for recruiting TEMs into tumors and plays a critical role in angiogenesis as their depletion in mice inhibits tumor angiogenesis and growth [36]. Agents targeting this pathway are under development with encouraging anticancer activity observed in early clinical studies [20, 60].

Elucidation of this signaling pathway has contributed to the development of genetically modified stem cells that express a transgene product of interest under the control of TIE2 promoter/enhancer elements upon reaching their target tissue/tumor and in the presence of ANGPT2 ligand. Multiple transgene products, including Herpes simplex virus thymidine kinase (HSV-tk) and interferon-α (IFN-α) (used as adjuvant therapy for melanoma) have been targeted to tumors using this approach. We have engineered MSCs to express the HSV-tk gene under the control of Tie2 regulatory elements, enabling selective expression of this therapeutic gene only after MSCs have homed to and, more importantly, initiated differentiation in the tumor in support of tumor neoangiogenesis. HSV-tk gene therapy with the prodrug ganciclovir (GCV) forms the basis of a widely used strategy for suicide gene therapy (discussed later). Monophosphorylation of GCV is the first step of GCV conversion to toxic metabolites including GCV-triphosphate which inhibits cellular DNA polymerases. Using a syngeneic orthotopic mouse pancreatic carcinoma model and a spontaneous breast tumor model, we showed that murine MSCs genetically engineered to express HSV-tk under the Tie2 promoter/enhancer (a) homed to growing tumor vasculature, (b) activated the Tie2 promoter in the tumor environment, and (c) significantly decreased tumor volume (orthotopic pancreatic carcinoma model) or delayed tumor growth (breast carcinoma model) following GCV prodrug treatment [56]. In mammary and Lewis lung carcinoma mouse models, bone marrow-derived progenitor cells transduced with lentiviral vectors expressing the suicide gene HSV-tk under the control of Tie2 transcription regulatory elements not only homed preferentially to tumor angiogenesis and were closely associated with vascular endothelial cells but also achieved substantial inhibition of angiogenesis and slower tumor growth without systemic toxicity in the presence of GCV [57, 58].

Similarly, hematopoietic stem/progenitor cells from CD1 athymic mice transduced with a lentiviral vector expressing Ifna1 under the control of Tie2 promoter/enhancer elements homed to angiogenic blood vessels in tumors preferentially activated the Tie2 promoter/enhancer in the tumor microenvironment and achieved substantial antitumor response and near complete abrogation of metastasis in orthotopic human glioma and spontaneous mouse mammary carcinoma models [57]. This has clinical importance as systemic IFN-α is poorly tolerated but local expression at sites of tumor and metastases would be highly beneficial in treating cancers such as melanoma.

Using the CCL5 Promoter to Target Tumor Stroma

Recent work increasingly supports the concept that stromal fibroblasts are not simply “enablers” of cancer but play an active role in tumorigenesis. In the prevailing tumorigenesis model, tumor-associated stroma is activated, expressing myofibroblastic markers (α-smooth muscle actin), extracellular matrix proteins (vimentin), growth factors (stromal-derived factor-1), and chemokines (CCL5/RANTES). These act in autocrine and paracrine fashion to potentiate and support tumor growth and survival and to actively recruit cells which home to the tumor environment [39, 61]. Stromal fibroblasts may also promote cancer cell invasion and metastasis [41]. The emerging role of cancer-associated fibroblasts (CAFs) in epithelial cancer is described in several excellent reviews [39, 62, 63]. While the origin of myofibroblasts and CAFs remains to be fully elucidated, increasing evidence suggest that bone marrow-derived progenitor cells, including MSCs, may be sources of tumor stromal fibroblasts [24, 62, 64]. MSCs are actively recruited to the tumor environment and contribute to the diverse cell types that comprise tumor stroma, including cells of the tumor vasculature and stromal fibroblast-like cells [65]. MSC recruitment to the tumor niche induces expression of the chemokine CCL5 (RANTES), which acts as a chemoattractant of progenitor cells including MSCs. This process is associated with increased tumor neovascularization, cancer growth and metastasis by autocrine activation of the tumor and recruitment of stromal cell types to sites of primary tumor growth [28, 29, 65].

Taking advantage of the homing behavior of MSCs to the tumor environment in response to CCL5, we genetically engineered MSCs to express the suicide gene HSV-tk under control of the CCL5 promoter, resulting in selective activation of TK-gene expression only in the tumor stromal environment. In a mouse orthotopic model of pancreatic carcinoma, these MSCs homed to primary pancreatic tumor stroma and activated the CCL5 promoter. Treatment with these MSCs and GCV prodrug resulted in significant reduction of primary pancreatic tumor growth and incidence of metastases [65].


Transgenes Encoding Proteins with Antitumor Activity

MSCs can be targeted spatially and transgene expression can be regulated by signals from a tumor environment. The identity of the therapeutic transgene is also clearly important for antitumor activity and therapeutic specificity and depends on tumor biology. MSCs have been genetically engineered to express gene products with direct antitumor activity, including IFNs, proapoptotic proteins, and antiangiogenic agents (Fig. 3) [2, 16]. Both IFN-B and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) have potent antitumor activity. The therapeutic efficacy of systemically administered recombinant IFN and TRAIL is limited by insufficient bioavailability at the site of tumor growth due to short protein half-life in circulation and dose-limiting side effects. Genetically engineered MSCs can be used to produce proteins with antitumor activity locally in the tumor environment, overcoming many challenges of conventional protein-based treatments with respect to delivery and specificity [66].

Figure 3.

Methodology for applying MSCs in gastrointestinal cancer therapy. (A): MSCs are isolated from each patient. (B): MSCs are cultured and expanded ex vivo. Therapeutic transgenes under the control of tumor/tissue-specific regulatory elements are introduced. (C): Genetically engineered MSCs are administered to patients; following systemic infusion, they home to target/tumor tissue where they differentiate and express therapeutic gene. (D): MSCs have been engineered to express multiple classes of therapeutic genes. Abbreviation: MSCs, mesenchymal stem cells.

In a mouse melanoma xenograft model, hMSCs genetically engineered to express IFN-B demonstrated engraftment preferentially at tumor sites and inhibited growth of malignant cells in vivo. This effect could not be achieved by systemically delivered IFN-B or IFN-B produced by genetically engineered MSCs located at sites distant from tumor [10, 11]. Similarly, hMSCs genetically engineered to express secreted recombinant TRAIL induced caspase-mediated apoptosis in glioma cell lines and primary glioma cells and had profound antitumor effects in orthotopic glioma tumor mouse models [67–69]. hMSCs genetically modified to express soluble TRAIL (sTRAIL) induced apoptosis when cocultured with human colorectal cancer cell lines including DLD-1. When systemically administered in a xenograft model, these MSCs failed to affect DLD-1 xenograft growth secondary to pulmonary entrapment and low rate of tumor homing [70], highlighting the strong potential but also a current limitation of therapeutic MSC transplantation.

While TRAIL is a potent and specific inducer of apoptosis in many cancer cells, pancreatic carcinoma cells exhibit intrinsic resistance toward TRAIL. Several molecular mechanisms have been implicated in TRAIL resistance, including expression of antiapoptotic proteins like X-linked inhibitor of apoptosis protein (XIAP) [71]. XIAP inhibition using RNA interference (RNAi) enhances TRAIL-induced apoptosis in pancreatic cancer both in vitro and in vivo [72, 73]. Others have extended on this concept of combination therapy by genetically engineering hMSCs to express sTRAIL (hMSC.sTRAIL). In a human pancreatic cancer mouse xenograft model these engineered hMSCs slowed tumor growth. In combination with XIAP inhibition by RNAi, treatment with hMSC.TRAIL resulted in not only slowing of tumor growth but also tumor remission and inhibition of metastatic growth [66].

Transgenes Encoding Enzymes Which Locally Activate Systemically Administered Prodrugs

Gene-directed enzyme prodrug therapy using MSCs is another elegant approach in which genetically engineered MSCs not only serve as carriers of the transgene but also drive transgene expression within the tumor environment through their differentiation potential after homing to target tissues. In this strategy, which has been reviewed elsewhere [2, 6, 16], the transgene encodes a foreign enzyme that converts a less toxic systemically administered prodrug into its active cytotoxic form only in the tumor environment. The cytotoxic drug exerts its effect not only on cells in which it is formed, the subset of tumor and tumor-associated stromal cells derived from MSCs expressing the foreign enzyme, but also on neighboring tumor cells that do not express the enzyme via a process known as the “bystander effect” [8, 16, 74].

A number of enzyme–prodrug systems have been developed, including the previously described HSV-tk/GCV system, cytosine deaminase (CD)/5-fluorocytosine (5-FC), cytochrome P450 isoforms/cyclophosphamide (CPA) or isofosfamide (IFO), and tomato thymidine kinase 1 (toTK1)/zidovudine (AZT) systems [2, 6, 16, 74, 75]. The basis for the CD/5-FC system is the ability of bacterial or yeast CD to convert the nontoxic prodrug 5-FC to the active drug 5-FU. The P450 isoforms have been used to catalyze the conversion of IFO and CPA to unstable, membrane diffusible DNA-alkylating metabolites. toTK1 phosphorylates AZT to AZT disphosphate, which has greater efficiency in killing glioblastoma cells than the HSV-tk/GCV system [8, 65]. The latter two systems described have predominantly been applied to glioblastoma models. The choice of enzyme–prodrug system should therefore be based on the characteristics and biology of the tumor of interest.

Several examples using the HSV-tk/GCV system were previously described in this review. The CD/5-FC suicide gene/prodrug system has also been applied to the field of genetically engineered MSCs for gastrointestinal malignancy therapy. Human adipose tissue-derived MSCs transduced with the yeast CD gene exhibited directed migration toward human colon cancer cells HT-29 in vitro and when administered in immunocompromised mice treated with 5-FC resulted in significant inhibition of tumor xenograft growth [76].

Transgene Encoding the Sodium Iodide Symporter: Applications to Real-Time Imaging and Cancer Therapy

The sodium iodide symporter (NIS) mediates active transport of iodide into the thyroid gland, provides the molecular basis for the diagnostic and therapeutic application of radioiodine in the treatment of thyroid cancer, and underlies one of the most effective and specific forms of systemic anticancer radiotherapy available. NIS has been recently developed as a reporter/therapeutic gene in nonthyroid cancers [77, 78]. It is used as a reporter gene for 123I-scintigraphy and 124I-positron emission tomography (PET) imaging. For therapeutic applications, it represents a powerful cytoreductive gene therapy strategy by transporting radionuclides such as 131I, 123I, 125I, 124I, 99mTc, 188Re, or 211At.

Engineering MSCs to express the NIS gene allows their tracking in vivo and provides expanded therapeutic strategies. This approach has been studied in murine models of breast cancer [77] and hepatocellular carcinoma (HCC) [78]. In the HCC model, the distribution of adoptively applied NIS-MSCs to HCC xenografts was first characterized by 123I-scintigraphy and 124I-PET imaging. For therapy, MSC-mediated NIS gene delivery followed by 131I application resulted in a significant delay in grafted HCC growth. Thus, local NIS gene transfer using engineered MSCs allows selective accumulation of a therapeutically effective dose of 131I in tumors, and importantly, as this strategy appears to be more effective than standard suicide genes, opens the door for potentially curative treatment options in solid tumors [78].

Choice of Targeting Strategy Must Be Based on Thorough Understanding of Tumor Biology

Genetically engineered MSCs can selectively target and deliver cancer therapies to tumors and metastases. The success of such a strategy is predicated on thorough understanding of the biology of the tumor target of interest. This was demonstrated by our group [59] in which our goal was to target the unique tumor biology of HCC with its high rate of angiogenesis and profound tumor stroma to attract engineered MSCs and trigger suicide gene expression in a tumor-tissue specific manner. MSCs isolated from bone marrow of C57/B16p53-/- mice were stably transfected with HSV-TK gene under the control of the Tie2 promoter/enhancer (targeting tumor angiogenesis) or the CCL5 promoter (targeting fibroblast-like tumor stromal compartment). Nontherapeutic control MSCs and engineered MSCs (Tie2/HSV-TK+ MSC, CCL5/HSV-TK+ MSC) were injected intravenously into mice with orthotopically growing HCC xenografts and subsequently treated systemically with GCV. Ex vivo examination of hepatic tumors following nontherapeutic MSC injection showed tumor-specific recruitment of MSCs, enhanced tumor growth, and increased microvascular density. Application of CCL5/HSV-TK+ MSC with GCV prodrug resulted in significantly reduced tumor growth by 56.4% compared with control group and 71.6% compared with nontherapeutic MSC injections. We also demonstrated that treatment with CCL5/HSV-TK+ MSC lead to higher tumor inhibition compared with control group, treatment with nontherapeutic MSCs, and treatment with Tie2/HSV-TK+ MSC. Interestingly, treatment with Tie2/HSV-TK+ MSC failed to show significant reduction in tumor growth compared with control group. While it is not well understood why treatment with Tie2/HSV-TK+ MSC proved less effective compared with CCL5/HSV-TK+ MSC, one hypothesis is that tumors treated in this study were established tumors with significant necrotic volume and in which angiogenesis may no longer be the driving force for tumor growth [59].

Current Limitations of Genetically Engineered Stem Cell Cancer Therapy

While genetically engineered MSCs have considerable therapeutic potential in a wide range of human diseases including cancer, their clinical application in humans may be hampered by concerns of both feasibility and safety. MSC isolation and expansion continues to be a challenge; the population of hMSCs obtained from sources can be heterogenous with significant variability in reported chemokine receptor repertoire despite similar isolation and culture conditions [22]. While this may reflect the heterogeneous nature of a typical MSC population and their potential to home to different tissues to enhance tissue repair or dampen inflammation [15], it raises the question whether differences in methods used to culture, expand, and study MSCs might influence their phenotype and functional properties. Passage number is important, for example, as MSCs have been shown to gain or lose certain cell surface receptors during culture [23]. There are also concerns that MSCs may support and propagate tumor growth and metastasis with evidence suggesting that at least in some circumstances MSCs are immunosuppressive and favor tumor growth [61, 79, 80].

From our experience the main limitations standing between the progress made in the laboratory developing MSC-based antitumor therapies and their clinical application are similar to those of other phase I drugs and revolve around efficacy and safety. Manufactured stem cells for therapeutic oncologic application need to meet the standards of Good Manufacturing Practice regulations. Therapeutic stem cells are biological products that would ideally be isolated from each patient to receive stem cell-based therapy and, as such, issues regarding uniformity and purity of isolated MSCs and stability during the expansion process are more complicated than with nonbiological agents. As with any therapy involving genetic manipulations, carcinogenic transformation of the therapeutic stem cells leading to secondary tumors is another concern. “Insertional genotoxicity” is an important factor to consider when choosing a vector type and design for cell therapy. Insertions may result in dominant gain-of-function mutations (such as activation of proto-oncogenes flanking an insertion site) mediated either by enhancer and/or promoter elements in the vector or by aberrant splicing from the vector transcript. This is favored by the genetic structure of retroviruses, which are efficient and frequently used transfection agents. Additionally, as experiments monitoring for insertional mutagenesis are often performed in rodents with relatively short life spans and the true mutagenic risk cannot be determined on the basis of vector choice and the total integration load in the transplanted cells alone, the true risk remains ill-defined. Using primate animal models able to tolerate larger transplanted MSCs quantities and with longer life spans and as well as transient transfections have been proposed. One way to balance the ethical dilemma of inducing secondary malignancies and with-holding potentially lifesaving therapy is to initially select patients with tumors who under best possible care have a short life expectancy such as pancreas or primary liver cancer.

Another current limitation to MSC-based therapy is pulmonary sequestration, limiting efficiency of MSC homing to tumor. As one advantage of genetically engineered stem cell-based therapeutics over direct tumor transfection is that the therapy can be administered systemically, tumor sequestration in the lungs or secondary lymphatic organs needs to be considered during the development of stem cell-based agents. Factors that determine which MSCs undergo sequestration and to what degree should be investigated; monitoring for this using stem cells transiently or stably labeled with a reporter construct is certainly feasible with currently available technologies.

We have observed in several in vivo gastrointestinal tumor models that naïve MSCs that do not bear the therapeutic construct lead to enhanced tumor formation. This is not unexpected as MSC recruitment to the tumor microenvironment has been shown to enhance various aspects of tumor progression and this homing behavior is fundamental to the strategy of using engineered MSCs in a therapeutic fashion. Therefore, efficacy of MSC transfection needs to be confirmed during the development of therapeutic MSCs to treat cancers. Standard techniques such as transfecting the MSCs with a resistance gene against a cytotoxic agent (e.g., blasticidine resistance gene under the control of cytomegalovirus promoter) has proven to be helpful in our hands to ensure purity of MSCs bearing the genetically engineered therapeutic construct.


Genetically engineered MSCs are promising potential cancer therapies, achieving tumor selectivity over normal tissues by three complementary mechanisms. First, MSCs preferentially home to sites of inflammation, ischemia, and malignancy. Second, using target tissue/tumor-specific promoters, genetically engineered MSCs only express therapeutic gene products in the appropriate biological (i.e., malignant) context. Third, biologic agents encoded by transgenes may themselves have targeted and differential effects on tumor versus normal cells. The successful application of genetically engineered MSCs toward cancer treatment requires a thorough understanding of the tumor biology of interest. Although a number of phase I and II clinical trials of ex vivo hematopoietic stem cell gene therapy for hematologic diseases have been completed or are ongoing [81], challenges remain which need to be overcome before genetically engineered MSCs can be successfully applied to treatment of solid malignancies, including gastrointestinal malignancies. In our opinion, these include concerns regarding the uniformity and full characterization, efficiency of production and delivery, and potential tumorigenicity of native and genetically engineered MSCs. While much work remains ahead, genetically engineered stem cells have tremendous potential as future targeted cancer therapy.


We thank Dr. Albert J. Keung for helpful discussions and manuscript review.


The authors indicate no potential conflicts of interest.