Antitumor Effects of CD40 Ligand-Expressing Endothelial Progenitor Cells Derived From Human Induced Pluripotent Stem Cells in a Metastatic Breast Cancer Model

Endothelial progenitor cells (EPCs) are attractive as cellular vehicles for targeted cancer gene therapy. An embryoid body formation method was used to derive CD133⁺CD34⁺ EPCs from human induced pluripotent stem (iPS) cells. The findings suggest that human iPS cell-derived EPCs have the potential to serve as tumor-targeted cellular vehicles for anticancer gene therapy.

Abstract

Given their intrinsic ability to home to tumor sites, endothelial progenitor cells (EPCs) are attractive as cellular vehicles for targeted cancer gene therapy. However, collecting sufficient EPCs is one of the challenging issues critical for effective clinical translation of this new approach. In this study, we sought to explore whether human induced pluripotent stem (iPS) cells could be used as a reliable and accessible cell source to generate human EPCs suitable for cancer treatment. We used an embryoid body formation method to derive CD133⁺CD34⁺ EPCs from human iPS cells. The generated EPCs expressed endothelial markers such as CD31, Flk1, and vascular endothelial-cadherin without expression of the CD45 hematopoietic marker. After intravenous injection, the iPS cell-derived EPCs migrated toward orthotopic and lung metastatic tumors in the mouse 4T1 breast cancer model but did not promote tumor growth and metastasis. To investigate their therapeutic potential, the EPCs were transduced with baculovirus encoding the potent T cell costimulatory molecule CD40 ligand. The systemic injection of the CD40 ligand-expressing EPCs stimulated the secretion of both tumor necrosis factor-α and interferon-γ and increased the caspase 3/7 activity in the lungs with metastatic tumors, leading to prolonged survival of the tumor bearing mice. Therefore, our findings suggest that human iPS cell-derived EPCs have the potential to serve as tumor-targeted cellular vehicles for anticancer gene therapy.

Introduction

Endothelial progenitor cells (EPCs) are a group of stem cells that can differentiate into the cells that make up the lining of blood vessels. EPCs are generally characterized by the expression of a set of cell surface markers, for example CD34, CD133, Flk1, von Willebrand factor (vWF), CD31, and vascular endothelial (VE)-cadherin [1–4]. EPCs are also identified by cell morphology, proliferative capacity, and their ability to generate vascular tubes in vitro. EPCs may play roles in vasculogenesis (formation of blood vessels by in situ differentiation of EPCs), angiogenesis (expansion of pre-existing blood vessels), and, possibly, promoting solid tumor growth. During the past decade, the observation that circulating EPCs from the bone marrow can be recruited to tumor neovessels has encouraged many studies to test various sources of endogenous and exogenous EPCs for therapeutic targeting to inhibit tumor neovascularization [3–6]. Since the systemic delivery of therapeutic gene-loaded EPCs can home to tumor microenvironment in multiple metastatic sites through circulation, possibly enabling efficient therapy for metastases, it becomes the most attractive feature in using EPC cellular vehicles for cancer treatment [5, 6].

With the increasing potential of using EPCs as cancer therapeutics, it is desirable to have a reliable and stable supply of human EPCs. Pluripotent stem cells, such as embryonic stem (ES) cells, can be expanded indefinitely in culture and have the remarkable ability to generate different cell types, including endothelial lineage cells [7–11], in virtually unlimited numbers. In particular, mouse embryonic EPCs are able to target lung metastases in an allogeneic setting in mice [6]. Most recently, Su et al. reported that human ES cell-derived CD34⁺/CD133⁺ endothelial cells armed with herpes simplex virus truncated thymidine kinase could inhibit MDA-MB-231 breast cancer growth [12].

Induced pluripotent stem (iPS) cells are another major source of pluripotent stem cells. Human iPS cells appear more attractive for clinical applications, since these cells can be generated relatively easily through reprogramming of differentiated somatic cells with transcription factors, a procedure that circumvents the bioethical controversies associated with the derivation of human ES cells from human embryos [13–16]. We report that tumor tropic EPCs can be derived from human iPS cells, and, after baculoviral transduction to introduce a therapeutic gene and intravenous injection into the body, these cells are able to attenuate tumor growth in the mouse 4T1 breast cancer model.

Materials and Methods

Generation of iPS-EPCs

Human iPS cells were generated using a polycistronic lentiviral vector encoding the Oct4, Klf4, Sox2, and c-Myc genes (Millipore, Bedford, MA, http://www.millipore.com), as stated in our previous study [17]. iPS cells were cultured on Matrigel (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) using mTeSR medium (StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com). To generate EPCs, iPS colonies were treated with 1 mg/ml dispase (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for 5 minutes and detached by scraping the colonies with a pipette tip. The detached colonies were transferred to a low attachment dish with STEMdiff APEL medium (StemCell Technologies) to form embryoid bodies (EBs) for 4 days. This animal product free medium provides good reproducibility of EB formation and high efficiency of differentiation [18]. Afterward, the formed EBs were plated onto Matrigel-coated (BD Biosciences) 10-cm culture dishes and cultured for another 10 days. Cytokines (PeproTech, Rocky Hill, NJ, http://www.peprotech.com) were added to supplement the medium as follows: 20 ng/ml BMP4 (days 0–7), 10 ng/ml Activin A (days 1–4), 8 ng/ml FGF2 (day 2 onward), and 25 ng/ml vascular endothelial growth factor (VEGF) (day 4 onward). In addition, 10 μM SB431542 (Tocris Bioscience, Bristol, U.K., http://www.tocris.com), a transforming growth factor-β (TGF-β) inhibitor, was added on day 7 onward [19]. At day 14, the cells were harvested by treatment with collagenase IV (Invitrogen) for 10 minutes, followed by trypsinization (Trypsin-EDTA; Invitrogen) for another 10 minutes. The harvested cells were then magnetically sorted using CD34 microbeads (Miltenyi Biotec, Cologne, Germany, http://www.miltenyibiotec.com). The sorted cells were seeded on fibronectin-coated plates in endothelial growth medium-2 (Lonza, Basel, Switzerland, http://www.lonza.com) supplemented with 10 μM SB431542 for in vitro expansion. Cells at passages 2 and 3 were used for characterization and animal experiments.

Characterization of iPS-EPCs

Generated iPS-EPCs were characterized by flow cytometric analysis using anti-CD34, CD31, Flk1, CD144, CD45 antibodies (BD Biosciences) and anti-CD133/1 antibody (Miltenyi Biotec). Quantitative analyses were performed using FACSCalibur flow cytometer (BD Biosciences). iPS-EPCs were also characterized with immunofluorescent staining using primary anti-vWF (Abcam, Cambridge, U.K., http://www.abcam.com), anti-CD31 (Abcam), and secondary fluorescent Alexa Fluor 488 conjugated anti-rabbit antibodies (Invitrogen). Hoechst (Invitrogen) was used to stain the nucleus of cells. Immunofluorescence was visualized, and images were captured using an Olympus image analysis system (Olympus, Tokyo, Japan, http://www.olympus-global.com). The cells labeled with 1,1′-dioctadecyl 3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI-Ac-LDL; Invitrogen) at a concentration of 10 µg/ml were used to test the ability of generated iPS-EPCs to uptake acetylated low-density lipoprotein (LDL). Tubulogenesis assays were performed in 48-well plates coated with Matrigel (150 µl per well; BD Biosciences) and incubated at 37°C overnight.

Baculovirus Preparation and Cell Transduction

Recombinant baculoviral vector expressing the enhanced green fluorescent protein (eGFP) reporter gene (BV-eGFP) was constructed using BAC-to-BAC baculovirus expression system (Invitrogen). BV-eGFP contains the eGFP gene under the control of human cytomegalovirus (CMV) early promoter [20]. Baculoviral vector encoding the CD40 ligand gene (BV-CD40L) was produced by homologous recombination after cotransfection of Sf9 insect cells with a pBacPAK9 transfer vector and the linearized AcMNPV viral DNA (Clontech Laboratories, Mountain View, CA, http://www.clontech.com). The pBacPAK9 transfer vector contains the mouse CD40L gene (InvivoGen, San Diego, CA, http://www.invivogen.com) under the control of the CMV promoter with the R segment and part of the U5 sequence of long terminal repeat from the human T-cell leukemia virus type 1 at 5′ UTR and the woodchuck hepatitis virus post-transcriptional regulatory element at 3′ UTR. Recombinant baculoviruses were amplified in Sf9 cells at a multiplicity of infection (MOI) of 0.1 plaque forming unit (pfu) per cell, and the virus-containing supernatant was collected 3 days after virus infection. Viruses were pelleted down at 28,000g for 1 hour and resuspended in phosphate-buffered saline (PBS). Recombinant DNA research in this study followed the National Institutes of Health guidelines. For transduction, iPS-EPCs were cultured with baculoviral vectors at an MOI of 100 pfu per cell overnight. A full medium change was done the next day to stop the transduction. GFP expression was detected using a fluorescence microscope and by flow cytometric analysis. To confirm CD40L gene expression, flow cytometric analysis was performed using anti-CD40L antibody (BD Biosciences).

Animal Studies

Pathogen-free, 6- to 8-week-old immunocompromised athymic nude (nu/nu) Balb/c mice were used in this study. To generate an orthotopic breast cancer model, mouse 4T1 breast cancer cells that stably express luciferase gene (4T1-luc; PerkinElmer, Hopkinton, MA, http://www.perkinelmer.com) were injected subcutaneously onto the mammary fat pad (1 × 10⁵ cells in 50 µl of PBS). To generate a breast cancer lung metastasis model, 4T1-luc cells were injected intravenously via the tail vein (1 × 10⁴ cells in 200 µl of PBS).

To observe the biodistribution of iPS-EPCs in the mice, a near-infrared (IR) fluorescent dye DiR (Caliper Life Sciences/PerkinElmer) and an orange-red fluorescent dye DiI (Invitrogen) were used to label the cells. Labeled iPS-EPCs (5 × 10⁵ in 200 µl of PBS) were injected into mice through the tail vein. To monitor the bioluminescent signals of 4T1-luc cells in the mice, isoflurane gas-anesthetized animals were injected intraperitoneally with D-luciferin (Promega, Madison, WI, http://www.promega.com) at 100 mg/kg in PBS. 4T1-luc cancer cells and DiR-labeled iPS-EPCs were visualized using the Xenogen IVIS imaging system coupled with a cool charge coupled device camera (PerkinElmer). After images were acquired, luminescent and fluorescent signals were measured with the Xenogen Living Image software, version 2.5. The tissue distribution of DiI-labeled iPS-EPCs was examined under a fluorescent microscope.

To investigate the effects of iPS-EPCs on metastasis and tumor development in the 4T1 orthotopic breast cancer model, iPS-EPCs, as well as other types of cells used as controls (5 × 10⁵ in 100 µl of PBS), were injected into the tumors 7 days after tumor inoculation. By day 28, the surviving animals were euthanized and the organs collected for ex vivo imaging of 4T1-luc signals. To investigate the effects of iPS-EPCs on animal survival in the 4T1 lung metastasis model, iPS-EPCs, as well as other types of cells used as controls, were injected into mice through the tail vein (5 × 10⁵ in 200 µl of PBS) 3 days after tumor inoculation. The date of each animal's death was recorded.

To evaluate the in vivo immune responses and apoptosis induction, the lungs were collected 28 days after tail vein injection of CD40L-expresssing iPS-EPCs and other control cells into the tumor-bearing mice. Using ProteoJET Mammalian Cell Lysis Reagent (Thermo Fisher Scientific, Waltham, MA, http://www.thermofisher.com) supplemented with pepstatin, leupeptin, and aprotinin (2 ml per gram of tissue), the tissues were homogenized and then sonicated on ice. The supernatants were collected after centrifugation at 13,000g for 20 minutes at 4°C. Interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α) release in lung tissues were determined using the Mouse IFN-γ enzyme-linked immunoassay (ELISA) Kit and Mouse TNF-α ELISA Kit (BD Biosciences). The caspase activity in the lung tissues was measured using Caspase-Glo 3/7 Assay (Promega).

To investigate the therapeutic effects of iPS-EPC-mediated gene delivery, the breast cancer lung metastasis model in nude (nu/nu) Balb/c mice as described in previous paragraphs was used. Mice were randomly divided into 4 groups (n = 10 per group) at day 3 after tail vain injection. BV-CD40L was used for transduction in iPS-EPCs at an MOI of 100 pfu per cell. Seven days after tumor inoculation, baculovirus-transduced iPS-EPCs (5 × 10⁵ in 200 μl of PBS per animal) were injected into the mice through the tail vein. Tumor growth was monitored using whole-animal imaging as described, and the animal survival rates were recorded.

All handling and care of the animals were performed according to the Guidelines on the Care and Use of Animals for Scientific Purposes issued by the National Advisory Committee for Laboratory Animal Research, Singapore.

Statistical Analysis

All data are presented as the mean ± SD. The statistical significance of differences was determined by one-way analysis of variance. The statistical analysis of survival data was performed using the Gehan-Breslow log-rank test followed by the Holm-Sidak method for pairwise multiple comparison tests. A p value of < .05 was considered statistically significant.

Results

Generation of CD34⁺/CD133⁺ EPCs From Human iPS Cells

In human ES cell lines, mesoderm cells in ES cell-derived EBs can be induced to differentiate into endothelial precursors [19]. We tested whether the same protocol could be used to derive endothelial precursors from early-passage human iPS cells (approximately passage 10). EBs formed by human iPS cells cultured on low attachment dishes were plated onto Matrigel-coated culture dishes (BD Biosciences) and treated with BMP4 (days 0–7), Activin A (days 1–4), FGF2 (day 2 onward), VEGF (day 4 onward), and SB431542, a TGF-β inhibitor (day 7 onward) for 2 weeks (Fig. 1A). Flow cytometric analysis revealed that the generated cells were a mixed population of cells containing ≥20% cells positive for CD34, a hematopoietic stem cell marker. The mixed population of cells coexpressed endothelial cell markers, including CD31 (17.60%), Flk1 (8.13%), and VE-cadherin (10.71%) but had a low expression level of the CD45 hematopoietic marker (1.44%) (Fig. 1B).

Figure 1.

Generation of endothelial progenitor cells (EPCs) from human induced pluripotent stem (iPS) cells and EPC characterization. (A): Generation of EPCs from iPS cells. iPS colonies were scraped to generate EBs. After 4 days, EBs were replated onto Matrigel-coated dishes and cultured for an additional 10 days. CD34⁺ cells were then harvested and maintained on fibronectin-coated dishes. (B, C): Flow cytometric analysis before (B) and after (C) magnetic cell sorting to show expression of endothelial lineage cell markers CD34, CD31, Flk1, and VE-cadherin, and the CD133 “stemness” marker, but low-level expression of hematopoietic marker CD45. Human fibroblasts were used as a negative control. Isotype controls were always set at ≤1. The percentages of positive cells are indicated. (D): Immunostaining showed positive expression of vWF (left), Hoechst staining for nucleus counterstain (middle), and merged image of Hoechst and vWF (right). (E): Tubule-like formation generated after iPS-EPCs were seeded onto Matrigel-coated dish. (F): Acetylated low-density lipoprotein (Ac-LDL) uptake assay. Shown are phase contrast (left) and red fluorescence (right) images of Ac-LDL taken up by the generated iPS-EPCs. Abbreviations: EBs, embryoid bodies; iPSC, induced pluripotent stem cell; VE-cad, vascular endothelial-cadherin; vWF, von Willebrand factor.

After immunomagnetic separation to enrich CD34⁺ cells and cell expansion in an endothelial growth medium, ≥95% of cells at passage 2 were CD34⁺. Most of the purified CD34⁺ cells were also positive for CD31 (81.29%). Approximately 21% of the CD34⁺ cells coexpressed Flk1, 48% expressed VE-cadherin, and only 8.98% of these cells were CD45⁺ (Fig. 1C). Since CD34 is expressed on both immature EPCs and mature circulating endothelial cells, we also examined the expression of CD133, a stem cell marker restricted to immature EPCs, on the sorted cells and found that 30% of the cells were CD133⁺ (Fig. 1C). As a negative control, human fibroblasts used to generate iPS cells were tested and showed almost no CD133 expression (0.76%). Furthermore, many of the iPS cell-derived cells were positive for vWF staining (Fig. 1D) and CD31 and CD34 staining (supplemental online Fig. 1). Biological and functional characterization demonstrated that the derived cells readily formed tubule-like structures on seeding onto Matrigel (BD Biosciences) (Fig. 1E) and were able to incorporate the DiI-labeled acetylated LDL (Fig. 1F). As such, we used the term of iPS-EPCs to refer to all endothelial lineage cells derived from iPS cells in the following sections.

Tumor Tropism of iPS-EPCs

To examine the in vivo tumor tropism of the generated iPS-EPCs, we established two mouse breast cancer models in athymic Balb/c nude mice: an orthotopic breast cancer model generated by subcutaneous injection of luciferase-expressing mouse 4T1 breast cancer (4T1-luc) cells onto the mammary fat pad and a breast cancer lung metastasis model generated by intravenous injection of 4T1-luc cells via the tail vein.

We first tested in vivo tumor tropism of iPS-EPCs in the orthotopic breast cancer model. After labeling iPS-EPCs with the near-IR fluorescent dye DiR, we injected the cells into the tumor-bearing mice through the tail vein at day 7 after tumor inoculation. The Xenogen IVIS Imaging System was used for noninvasive dual-color in vivo imaging to longitudinally monitor the distribution of the labeled iPS-EPCs at days 0, 1, 3, 7, and 14 post-EPC injection in the same sets of animals. Whole-body bioluminescence imaging for 4T1-luc cells showed that the tumors grew quickly around the inoculated mammary fat pad during the 21-day observation period, and luciferase signals outside the primary tumor region were observed at day 21 after tumor inoculation, indicating metastasis (Fig. 2A). Whole-body DiR fluorescence imaging for iPS-EPCs showed that tail vein-injected stem cells initially distributed predominantly to the lung region. One day after tail vein injection, the size of the DiR signal-positive area in the thoracic region significantly increased. We also observed significantly increased DiR signals in the tumor inoculation site in tumor-bearing mice (Fig. 2A). The DiR signals in this region were significantly greater than those in mice without tumors and remained high for at least 14 days, although the overall DiR signal intensity reduced from day 3 onward (Fig. 2A). On average, approximately 12% of the injected iPS-EPCs homed in the mammary pad region inoculated with 4T1-luc tumor cells as calculated by the DiR signal in the region over the whole body signal (Fig. 2B).

Figure 2.

Tumor tropism of induced pluripotent stem (iPS) cell-derived EPCs in the 4T1 orthotopic mouse model of breast cancer. The tumor model was established using orthotopic injection of mouse 4T1-luc breast cancer cells into the mammary fat pad in Balb/c nude mice. At day 7 after tumor inoculation, DiR-labeled iPS-EPCs were injected through the tail vein into 4T1 tumor-bearing mice and Balb/c nude mice without tumor, two animals per group. (A): Whole-animal imaging. Mice were imaged ventrally with the IVIS imaging system on the indicated days. Imaging at day 0 was performed 5 minutes after iPS-EPC tail vein injection. Dual-color imaging was performed in 4T1 tumor-bearing mice to show both DiR-labeled iPS-EPCs and 4T1-luc cells. (B): Percentage distribution of iPS-EPCs in the primary tumor region in the mammary fat pad over the whole body. ∗, ∗∗, p = .03, p = .026, p = .006, and p = .021 for days 1, 3, 7, and 14, respectively, by one-way analysis of variance. The data are representative of three independent experiments. Abbreviation: EPC, endothelial progenitor cell.

We then tested in vivo tumor tropism of iPS-EPCs in the breast cancer lung metastasis model. We injected the DiR-labeled iPS-EPCs into the tumor-bearing mice through the tail vein at day 3 after tumor inoculation. Whole-body bioluminescence imaging showed that tumor growth became detectable at day 10 after tumor inoculation (day 7 after EPC injection) (Fig. 3A). By day 17 after tumor inoculation, cancer cell dissemination to other organs was observed. Whole-body DiR fluorescence imaging showed that tail vein-injected iPS-EPCs were distributed predominantly to the lung region and decreased over time from day 1 to day 14 after EPC injection (Fig. 3A). Compared with the DiR signals in normal Balb/c mice injected with the labeled iPS-EPCs, only without 4T1 tumor cell inoculation, the DiR signals in the tumor-bearing mice were relatively stronger in the lung region, and a statistically significant difference between the two groups of mice was observed on day 14 (Fig. 3B).

Figure 3.

Tumor tropism of induced pluripotent stem (iPS) cell-derived EPCs in the 4T1 breast cancer lung metastasis model. The tumor module was established by intravenous injection of 4T1-luc mouse breast tumor cells into Balb/c nude mice. At day 3 after tumor inoculation, DiR-labeled iPS-EPCs were injected through the tail vein into 4T1 tumor-bearing mice and Balb/c nude mice without tumor, three animals per group. (A): Whole-animal imaging. Mice were imaged ventrally with the IVIS imaging system on the indicated days. Imaging at day 0 was performed 5 minutes after iPS-EPC tail vein injection. Dual-color imaging was performed in 4T1 tumor-bearing mice to show both DiR-labeled iPS-EPCs and 4T1-luc cells. (B): The DiR signal over time. ∗, p = .036 and p = .022 on days 7 and 14, respectively, by one-way analysis of variance. The data are representative of three independent experiments. Abbreviation: EPC, endothelial progenitor cell.

Effects of iPS-EPCs on Metastasis and Tumor Development

The potential role of EPCs in promoting tumor growth is a concern when these cells are used as delivery vehicles for cancer therapy [21, 22]. To investigate whether iPS-EPCs alone affect tumor progression and metastasis in the orthotopic breast cancer model, we injected iPS-EPCs intratumorally at day 7 after inoculation of 4T1-luc breast cancer cells. Intratumor injections of iPS cells, human foreskin fibroblasts (HFFs), and PBS were included as controls. By day 28, there were 9 surviving animals in the iPS-EPC-injected group, 7 each in the PBS and HFF group, and 5 in the iPS cell-injected group. The surviving animals were sacrificed, and the organs were collected for ex vivo bioluminescence imaging to examine for metastases (Fig. 4A). The 4T1-luc signal intensities in the examined organs were comparable in all groups (Fig. 4B). Although the tumor signal in the intestine was higher in the iPS-EPC group than in the other groups, statistical analysis revealed no significant differences. Weight measurement of the tumors collected from the surviving animals also revealed no statistically significant difference (Fig. 4C).

Figure 4.

Effects of induced iPS-EPCs on 4T1 tumor development and metastasis in the 4T1 orthotopic mouse model of breast cancer. The tumor model was established using orthotopic injection of mouse 4T1-Luc breast cancer cells into the mammary fat pad in Balb/c nude mice. At day 7 after tumor inoculation, PBS, HFF, iPS cells, and iPS-EPCs were injected directly into the tumors, 10 animals per group. (A): Ex vivo images of different organs collected at day 28, three representatives per group. Left to right: Lung, liver, spleen, kidneys, heart, brain, intestine, spinal cord, femur, and primary tumor mass. (B): Quantitative comparison of Luc signals in the organs collected from surviving animals at day 28; n = 7, 7, 5, and 9 in the PBS, HFF, iPS cell and iPS-EPC group, respectively. (C): Tumor weight comparison. The weights of the tumors collected from the surviving animals at day 28 were measured. Analysis of variance statistical analysis revealed no statistically significant difference among the four groups in (B, C). Abbreviations: EPC, endothelial progenitor cell; HFF, human foreskin fibroblast; iPS, induced pluripotent stem; PBS, phosphate-buffered saline.

The possible effects of iPS-EPCs on tumor development were further investigated in the breast cancer lung metastasis model. Taking advantage of the quick animal death in this model, we focused on the effects on survival of tumor-bearing animals in this experiment. iPS-EPCs were injected intravenously via the tail vein 3 days after intravenous injection of 4T1-luc cells. Tumor growth was monitored by whole-body bioluminescent imaging of 4T1-luc cells. Figure 5A shows representative images of mice in each group at 4 different points. The bioluminescence intensities, indicative of tumor volume, demonstrated that there were no significant differences among the PBS group, iPS cell group, and iPS-EPC group, although a significantly higher level of 4T1-luc signal was observed in the HFF group at day 21 (Fig. 5B). Mice started to die at day 17 in all groups. The median survival time was 26.2, 26.8, 25.6, and 27.6 days for the PBS, fibroblast, iPS cell, and iPS-EPC groups, respectively. There was no statistically significant difference among the four groups in the survival rate (p = .926) (Fig. 5C). Our findings from the two tested tumor models indicate that iPS-EPCs did not obviously promote tumor development.

Figure 5.

Effects of iPS-EPCs on 4T1 tumor development and animal survival in the 4T1 breast cancer lung metastasis model. The tumor model was established by intravenous injection of 4T1-luc mouse breast tumor cells into Balb/c nude mice. At day 3 after tumor inoculation, PBS, HFF, iPS cells, and iPS-EPCs were injected into the tumor-bearing mice through the tail vein, 10 animals per group. (A): Bioluminescent images of tumor growth in representative animals from each group at days 1, 7, 14 and 21 after iPS-EPC injection. (B): Quantitative analysis of bioluminescent signal changes over time. Bars indicate SD. ∗, p < .05 versus the group of mice injected with PBS or iPS-EPCs by analysis of variance. (C): There was no difference in survival rate in different groups of 4T1 tumor-bearing mice. The statistical analysis was performed using the log-rank test. Abbreviations: EPC, endothelial progenitor cell; HFF, human foreskin fibroblast; iPS, induced pluripotent stem; PBS, phosphate-buffered saline.

iPS-EPCs Expressing CD40L Impede Tumor Development in a Breast Cancer Lung Metastasis Model

To assess the prospective of iPS-EPCs as a cellular vehicle to deliver therapeutic genes, we transduced the cells with baculovirus expressing CD40L, a member of the TNF gene family [23, 24]. Flow cytometric analysis indicated that baculovirus-transduced iPS-EPCs still expressed CD34 (Fig. 6A), and up to 82.23% of iPS-EPCs could be transduced to express detectable levels of CD40L (Fig. 6B). To examine the in vivo migration capacity of these transduced cells toward the tumors in the lung, the cells were prelabeled with the orange-red fluorescent dye DiI and injected into the lateral tail vein of athymic Balb/c nude mice bearing 4T1-luc breast cancer lung metastases. The labeled cells were found mostly within and surrounding the tumor nodules, with a high density of Hoechst-stained nuclei and only a few labeled cells in the normal lung tissues (Fig. 6C). Some of these labeled cells were observed in the neighborhood of the blood vessels (supplemental online Fig. 2). To evaluate in vivo immune responses, we used ELISA kits to determine the TNF-α and IFN-γ concentrations in tissue homogenates of the lungs collected 28 days after tail vein injection of CD40L-expresssing iPS-EPCs into the tumor-bearing mice. We detected significantly increased levels of the two proapoptotic cytokines (Fig. 6D). Moreover, we also detected significantly increased caspase 3/7 activities in the lung tissue homogenates (Fig. 6E), indicating the induction of apoptosis after delivery of CD40L into the lungs by the iPS-EPCs.

Figure 6.

CD40L-expressing iPS-EPCs provide in vivo immunostimulatory effects in the 4T1 breast cancer lung metastasis model. (A): Baculoviral transduction efficiency in iPS-EPCs. A baculoviral vector with the enhanced green fluorescent protein gene under the control of the cytomegalovirus (CMV) promoter was used. Left: Flow cytometry analysis of the transgene expression. Right: Fluorescence imaging of EGFP expression. (B): Flow cytometry analysis of CD40L expression in iPS-EPCs after transduction with a baculoviral vector containing CD40L gene under the control of the CMV promoter. (C): In vivo tumor tropism of CD40L-expressing iPS-EPCs in the lung. The tumor model was established by intravenous injection of 4T1-luc mouse breast tumor cells into Balb/c nude mice. The lungs were collected 3 days after the injection of iPS-EPCs prelabeled with DiI, a red-fluorescent dye, for sectioning and microscope examination. Note the accumulation of DiI-positive cells within a tumor nodule. (D): TNF-α and INF-γ concentrations in lung tissue homogenates. The lungs were collected 28 days after injection of iPS-EPCs into tumor-bearing mice to prepare tissue homogenates. TNF-α and INF-γ concentrations were measured using enzyme-linked immunosorbent assay kits. (E): Caspase 3/7 activity assay. Caspase 3/7 activity levels in the lung tissue homogenates were measured using a luminogenic caspase-3/7 substrate, n = 5 per group in (D, E). Bars indicate SD. ∗, ∗∗, ∗∗∗, p < .05, p < .01, and p < .001, respectively, versus the group of mice injected with iPS-EPC/BV-CD40L by analysis of variance. The data are representative of two independent experiments. Abbreviations: BV, baculoviral vector; DAPI, 4′,6-diamidino-2-phenylindole; EGFP, enhanced green fluorescent protein; IFN, interferon; iPS-EPC, induced pluripotent stem-endothelial progenitor cell; TNF, tumor necrosis factor.

Tumor growth in the tumor-bearing mice was monitored by whole-body bioluminescent imaging of 4T1-luc cells. Figure 7A shows representative 4T1 cell-bearing mice from different groups at 4 different points. The bioluminescence intensities, indicative of tumor volume, demonstrated that the inhibitory effect of CD40L-expressing iPS-EPCs on tumor growth was visible on day 14. A statistically significantly lower level of 4T1 cell-related luciferase signals was observed in the treated mice by day 21 after tumor inoculation compared with that in the control groups (Fig. 7B). Attributed to the inhibitory effect of CD40L-expressing iPS-EPCs, survival of the tumor-bearing mice in this group was significantly prolonged (Fig. 7C). On day 34, although all mice in the three control groups had died, 60% of the animals in the treated group were still alive. The median survival time was 25 to 29 days in the control groups but 34 days in the treated group (Fig. 7D). The difference was statistically significant (p = .004, Holm-Sidak test). There was no statistically significant difference among the control groups in median survival time (p > .05).

Figure 7.

Therapeutic effects of CD40L-expressing iPS-EPCs in the 4T1 breast cancer lung metastasis model. The tumor model was established by intravenous injection of 4T1-luc mouse breast tumor cells into Balb/c nude mice, 10 animals per group. (A): Bioluminescent images of tumor growth in representative animals from each group at days 1, 7, 14, and 21 after iPS-EPC injection. (B): Quantitative analysis of bioluminescent signals. Bars indicate SD. ∗, p < .05 versus the group of mice injected with iPS-EPC/BV-CD40L by analysis of variance. (C): CD40L expression prolongs the life of 4T1 tumor-bearing mice. The statistical analysis was performed using the log-rank test. (D): Median survival days. Statistical analysis by the log-rank test reveals the significantly improved survival in the group of mice injected with iPS-EPC/BV-CD40L compared with the other groups (∗, p < .05). Abbreviations: BV, baculoviral vector; EPC, endothelial progenitor cell; iPS, induced pluripotent stem; PBS, phosphate-buffered saline.

Discussion

Both autologous sources, such as peripheral blood and bone marrow, and allogeneic sources, such as umbilical cord blood, fetal liver, and embryonic stem cells, can be used to isolate/generate endothelial lineage cells [4]. iPS cells are a unique type of biological materials that can be used to generate either autologous or allogeneic cells. If the cells to be used for treatment are the differentiated progeny of iPS cells reprogrammed from the patient’s own cells, these cells are viewed as “autologous” cells. The likelihood of immune rejection toward them after transplantation will be significantly reduced. In the case that a human iPS bank consisting of various human leukocyte antigen (HLA) types is established, iPS cells selected from the cell bank can be alternatively used to generate semiallogeneic differentiated cells sharing some of the HLA alleles with recipients. The third method of using iPS cells is to generate allogeneic differentiated cells suitable for applications that do not require long-term survival of the transplanted cells, for example using stem cells as cellular vehicles for targeted cancer therapy. Although long-term immunosuppressive treatment is required for many regenerative medicine applications of allogeneic stem/progenitor cells, it will not be necessary when iPS cell-derived cells are used as exogenous cellular vehicles for cancer treatment, since the application does not really require stem cell engraftment.

Using allogeneic cells derived from iPS cells as cellular vehicles for cancer therapy can be justified for several reasons. First, genetic engineering is possible to introduce a therapeutic gene into iPS cells to generate a master cell line for cell therapeutic production. Second, the allogeneic cell approach allows large-scale mass production of cell therapeutics, a prerequisite for widespread application of cell therapy suitable for repeated patient treatments. Large-scale mass production will increase the cost-effectiveness by reducing the laboriousness and simplifying the logistics of cell culture operations. Mass production will also make it possible to prepare cryopreserved derived cells as commercial products in a “ready-to-go” format. Third, the allogeneic cell approach allows standardized manufacture, which should be helpful in eliminating variability in the quality of cellular products, thus facilitating reliable comparative analysis of clinical outcomes.

The iPS-EPCs generated in the present study display phenotypic and functional traits consistent with previous findings [8, 9, 11, 19, 25, 26]. CD34⁺/CD133⁺/Flk1⁺ cells are defined as EPCs [3, 4]. CD133 (AC133, prominin-1) is a transmembrane domain cell surface glycoprotein that has been widely used as a marker for stem cells, including EPCs [1, 2]. Endothelial and hematopoietic progenitors share common markers such as CD34, CD31, and Flk1. Hematopoietic cells were then distinguished from their endothelial counterpart by the expression of CD45, a common leukocyte antigen not expressed in endothelial cells [25]. Approximately 30% of iPS cell-derived cells in this study express the CD133⁺ stem cell marker (Fig. 1C), indicating that they are still in an early phase of endothelial differentiation. Moreover, these iPS cell-derived cells are positive for the previously mentioned endothelial progenitor markers and display other characteristics associated with EPCs, including VE-cadherin, vWF, uptake of Ac-LDL, and formation of tubule structure on seeding onto Matrigel (BD Biosciences) (Fig. 1C–1F) [3, 27]. Although the term “iPS-EPC” was used for these iPS cell-derived cells, we do realize that the population could contain cells at different levels of differentiation along the endothelial lineage. A further purification step (e.g., isolation of CD133⁺ cells) could be added to remove relatively mature cells. For clinical applications, more robust and cost-effective cell expansion technologies will be required. In this aspect, expansion methods such as automated bioreactors and rotary culture machines hold promise for generating iPS cell derivatives in bulk [28–31].

Our biodistribution experiments have clearly demonstrated the tumor tropism of the generated iPS-EPCs. Other types of adult stem cells, such as mesenchymal stem cells (MSCs) and neural stem cells (NSCs), as well as MSCs and NSCs derived from pluripotent stem cells, also show the ability to home to tumors [17, 32–35]. However, it has been noted that MSCs that reach tumor stroma can contribute to tumor development through the mechanisms of creating a niche to support cancer stem cells, promoting angiogenesis, and/or promoting cancer metastasis [36–38]. Both MSCs and NSCs display immunosuppressive effects [39–42], which may promote tumor growth by inhibiting the “tumor surveillance” functions of the immune system, thus compromising the effectiveness of using these cellular vehicles for cancer treatment. Unlike MSCs and NSCs, there is no report of immunosuppressive effects for EPCs.

Our study also showed that most intravenously injected iPS-EPCs were trapped in the lungs in the orthotopic breast cancer model, most likely because of the narrow diameters of lung capillaries. After initial pulmonary trapping, many intravenously injected cells will usually redistribute from the lungs to the liver and spleens. Comparable results were shown by Wei et al., who found that tail vein injection of mouse embryonic EPCs into C3H mice resulted in the injected cells that were sequestered in the lung and spleen predominantly [6]. Our previous biodistribution studies using human NSCs also showed a similar cell distribution pattern [17, 34]. This distribution pattern is useful for eliminating metastatic cancer cells in the lungs, liver, and spleen and has provided a rationale for us to test the therapeutic effects of intravenously injected iPS-EPCs in the breast cancer lung metastasis model in the present study. This cell distribution pattern, however, also emphasizes the importance of minimizing off-target transgene expression if the cell delivery aims to treat primary breast cancer. Several strategies can be used to enhance the tumor tissue-specific transgene expression for cancer gene therapy. The popular one is the use of tumor environment-specific promoters, including the hypoxia response elements, proendothelin-1 promoter, and survivin promoter [43, 44]. Given that the vesicular stomatitis virus glycoprotein (VSV-G) can promote the formation of multinucleated syncytia to kill cells in a pH-dependent manner, we have generated a VSV-G mutant that efficiently promotes syncytium formation at the tumor extracellular pH but not at pH 7.4. Using transduced NSCs derived from iPS cells to deliver the VSV-G mutant into mice with metastatic breast cancer in the lung through tail vein injection, we observed tumor-selective killing without obvious toxicity to normal nontargeted organs [34].

Circulating, bone-marrow-derived EPCs might follow gradients of growth factors and cytokines that are released into the circulation by tumors and contribute to tumor neovascularization, tumor growth, and metastasis [21, 22]. However, this contribution remains controversial, since the existence of these EPCs during tumor progression vascularization has been established in certain models and disproved in others, depending on tested tumor models and methods [45]. We did not observe an obvious contribution of iPS-EPCs to breast cancer growth and metastasis in both orthotopic and metastatic tumor models. In the lung metastatic cancer model, injection of fibroblasts alone resulted in an increased tumor volume at day 21 compared with the other groups, although there was no significant difference in animal survival by the end of the experiment. Carcinoma-associated fibroblasts (CAFs), the activated fibroblasts, are the primary type of host cells in tumor microenvironment. The origin of CAFs has not been conclusively established and remains controversial [46]. Whether the systemically injected fibroblasts tested in the present study were recruited by the tumor and activated in situ to become CAFs is worth pursuing in future research.

CD40L was used as a model therapeutic agent to test iPS-EPC-mediated gene delivery and cancer therapy. CD40L, also called CD154, is a 33-kDa type II membrane protein belonging to the TNF gene family and is mainly expressed on activated CD4⁺ T cells. It interacts with CD40 expressed on a wide range of antigen-presenting cells (APCs), including dendritic cells, macrophages, and B lymphocytes, as well as nonimmune cells, such as epithelial cells, endothelial cells, and malignant cells [23, 24]. CD40L–CD40 interaction induces activation in APCs in association with T-cell receptor stimulation by MHC molecules on the APCs, ensuring the generation of antigen-specific cytotoxic T lymphocytes. Increased antigen presentation by APCs can lead to increased cytokine production, expression of other costimulatory molecules, and induction of cytotoxicity. CD40L can also exert antitumor effects by inducing tumor cell apoptosis via autocrine/paracrine induction of death ligands such as TNF-α, FasL, and TNF-related apoptosis-inducing ligand [24]. Local CD40L gene delivery can directly transduce tumor cells and effectively inhibit tumor growth by mediating multiple antitumor effects, including induction of T-cell responses, upregulation of TH1 cytokines, and induction of apoptosis [23, 47–49].

Our results showed that CD40L delivered by systemically injected iPS-EPCs could also reduce the tumor burden and eventually prolong the survival of tumor-bearing animals. To test human EPCs, immunocompromised Balb/c nude mice were used to produce tumor models in the present study. This strain of mice lacks T cells but still acquires B cells and innate immune cells, such as natural killer cells and complement immune system. Thus, upregulation of TH1 cytokines and induction of apoptosis, but not induction of T-cell responses, might play more important roles in the observed antitumor action in this study. Different from the strategy of direct transduction of tumor cells with the CD40L gene, our approach uses baculoviral CD40L transduction of iPS-EPCs and the tumor tropism property of the cells to activate local CD40L-CD40 interactions in the tumor, likely through a bystander mechanism of intercellular transfer of the CD40L protein from donor iPS-EPCs to tumor cells and tumor-infiltrating immune cells. It has been established that CD40L protein transfer to tumor cells occurs in transduced fibroblasts, epithelial cells, and bone marrow stromal cells [50].

In the present study, we demonstrated for the first time that baculoviral vectors can be used for genetic modification of human EPCs to provide a window of transgene expression suitable for cancer therapy. The large cloning capacity, ease of virus production, and low cytotoxicity to transduced cells are the obvious advantages associated with the emergence of baculoviral vectors in the gene delivery field [51]. Insect baculovirus also bypasses the risk of virus replication and potential viral infection in the host cells, risks borne by conventionally used animal viruses, such as adenovirus, retrovirus, and adeno-associated virus. Pre-existing host immune response against adenoviral vectors, a type of vectors commonly used for transient transgene expression, is another concern restricting the use of these vectors [52, 53]. Unlike adenovirus, baculovirus is not targeted by pre-existing immunity in humans [54]. The transient gene expression mediated by baculovirus of nonintegrating nature, however, limits its efficacy in studies that require long-term transgene expression. To overcome the limitation and also to avoid random integration, we have developed several baculoviral transduction-based approaches using either recombinase-mediated cassette exchange, zinc finger nuclease technology, or transcription activator-like effector nuclease technology for site-specific integration of a transgene in human pluripotent cells [55–58]. We are in the midst of establishing iPS cell lines to achieve stable expression of transgenes in the differentiated progeny of the iPS cells, including EPCs.

Conclusion

We have demonstrated the feasibility of generating EPCs from early-passage human iPS cells. We further demonstrated the tumor tropism of the EPCs and their ability as a new type of cellular vehicle to deliver immune stimulatory molecules to inhibit metastatic breast cancer growth in the lung in a mouse tumor model. Since this approach may use patient's own cells to prepare cellular therapeutics and is based on nontoxic immunotherapy, it holds potential for translation to clinical application and may be particularly valuable as a new type of antimetastatic cancer therapy.

Author Contributions

Y.I.P.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; C.C., D.H.L., C.W., and J.Z.: collection and assembly of data; W.F.: conception and design; S.W.: conception and design, financial support, manuscript writing, final approval of manuscript.

Disclosure of Potential Conflicts of Interest

D.H.L. has compensated employment and intellectual property rights.