Introduction

Critical limb ischemia (CLI) is a serious form of peripheral artery disease (PAD) that leads to ulcers, gangrene, and major amputations [1,2,3]. The mortality rate of CLI is 20% at 6 months and 50% at 5 years after diagnosis, which corresponds to the survival rate of patients with malignant tumors [4]. Endovascular treatment and bypass surgery have been the usual treatments, but have not been effective for some CLI patients (called “no-option” CLI patients) [5, 6]. A variety of therapies, including gene and cell therapy, have been tested in CLI clinical trials [7,8,9,10,11], but no other candidates for CLI indication have been successful in phase III in the USA.

We have recently established a chronic and severe hindlimb ischemia (HLI) mouse model (chronic Type-N) to evaluate blood flow restoration and assess in vivo angiogenesis following therapeutic administration of CLI drug candidates [12]. In this model, therapeutic administration of vasodilator cilostazol, which is used as a standard drug, did not restore blood flow (0.3% cilostazol diet for 1 month starting 6 weeks after surgery), but therapeutic transplantation of both pericytes and vascular endothelial cells, which have the potential to form new blood vessels in vivo, significantly improved ischemic limb blood flow [12]. From this data, it is suggested that angiogenic cell therapies may be useful for the restoration of perfusion. However, in clinical development, combination cell therapy using two types of cells such as human pericytes and endothelial cells are complicated in terms of manufacturing large numbers of cells and quality control under Good Tissue Practice-compliant conditions. Therefore, this study focused on single-cell therapy with pericytes for CLI treatment.

Pericytes are known to enclose and stabilize the endothelial cells that line the capillaries and venules [13, 14]. Once pericytes are damaged, these functions are lost, leading to severe vascular diseases such as diabetic retinopathy [15, 16]. Pericytes affect vascular maturation, endothelial cell differentiation, and angiogenesis by expressing angiogenic factors such as vascular endothelial growth factor (VEGF), angiopoietin-1 (Ang-1), and stromal cell–derived factor-1 (SDF-1) [13, 17,18,19]. Mesoangioblasts (MABs), which have been identified in skeletal muscle as a kind of pericyte [20], have the potential for angiogenesis as well as the ability to differentiate into myoblasts and cartilage [21, 22]. We have confirmed that primary MABs in combination with human umbilical vein endothelial cells (HUVECs) significantly improved blood flow in Type-N mice [12]. Therefore, it was thought that pericytes alone may also be a promising candidate for single-cell therapy for CLI treatment.

Cell therapy requires the preparation of large numbers of cells as a master and working cell bank for therapeutic use [23], but the expandability of primary cells is limited. In addition, the pharmacological function, proliferative activity, and differentiation potential of primary cells vary among donors. Given this difficulty of controlling the characteristics of each donor, we attempted to establish embryonic stem cell (ES)–derived pericytes from a single line in order to ensure consistent function and proliferative characteristics. So far, several protocols for the differentiation of human ES cells (hESC) into pericyte-like cells have been reported [24,25,26,27,28], but it is still difficult to obtain enough cells expressing angiogenic factors due to heterogeneity [25, 26]. In this report, we present a method designed to address these issues and its effectiveness in chronic Type-N mice.

Materials and Methods

Maintenance and Differentiation of ES Cells

The hESC lines (SEES5 and SEES6) were obtained from the National Center for Child Health and Development. This was approved by the ethics review committee for human ES cells at Osaka University Graduate School of Medicine, Osaka, Japan. Feeder-free ES cells were maintained in a commercial medium (mTeSR® STEMCELL Technologies, Vancouver, BC, Canada) and were induced to differentiate into mesoderm using an induction medium (STEMdiff® Mesoderm Induction Medium, STEMCELL Technologies) for 2 days. A single-cell suspension of the differentiated cells was prepared at 1.7 × 105 cells/dish on a glass dish (EZSPHERE® SP 100 mm dish, AGC Techno Glass, Ltd., Shizuoka, Japan) in 10 mL of a mixed spheroid-forming medium consisting of 40% (MegaCell medium® Sigma-Aldrich, St. Louis, MO, USA), 25% serum-free expansion medium (StemSpan®, STEMCELL Technologies), 25% human endothelial serum-free medium (Thermo Fisher Scientific, Cleveland, OH, USA), and 10% Bit 9500 Supplement StemSpan® (STEMCELL Technologies), with GlutaMAX (Thermo Fischer Scientific), 0.1% Excyte Supplement® (Millipore, Billerica, MA, USA), 100 µM monothioglycerol (Wako Pure Chemical Industries, Osaka, Japan), 50 µg/mL ascorbic acid (Sigma), and 20 ng/mL basic fibroblast growth factor (bFGF, Proteintech, Chicago, IL, USA) added. Ten more mL of the spheroid-forming medium was added 4 days later, and the medium was changed by half every 2–3 days. At day 12, mesenchymal colonies were collected using 100-µm cell strainers (BD Biosciences, San Jose, CA, USA) and then were plated onto culture dishes coated with human fibronectin (3 µg/mL; Corning, Corning, NY, USA) and human collagen 1 (10 µg/mL; Kyowa Pharma Chemical, Toyama, Japan) in pericyte differentiation medium: 50% Stemline 2® hematopoietic stem cell expansion medium (Sigma), 50% human endothelial serum-free medium (Thermo Fisher Scientific), 1% GlutaMAX, 0.05% Excyte Supplement, 100 µM monothioglycerol, 10 ng/mL bFGF, 50 ng/mL platelet-derived growth factor isoform BB (PDGF-BB) (R&D Systems, Minneapolis, MN, USA). After 3 days, cells were treated with a cell dissociation solution (Accutase®, Innovative Cell Technologies, San Diego, CA, USA) and were plated onto a collagen-coated plate (Corning) in pericyte proliferation medium: 78% MegaCell medium (Sigma), 20% fetal bovine serum, 1% GlutaMAX (Thermo Fischer Scientific), 1% non-essential amino acids (Sigma), 2-mercaptoethanol (Thermo Fischer Scientific), and 5 ng/mL bFGF (Proteintech). When cells reached confluence, the CD56-positive and CD56-negative fractions were sorted using fluorescence-activated flow cytometry (SH800®, Sony, Tokyo, Japan), and both cells were cultured on collagen-coated dishes.

Preparation of Primary Cells

Primary MABs were purified from human skeletal muscle samples. This was approved by the institutional review committee of the Osaka University Graduate School of Medicine, Osaka, Japan.

The purification method has been described previously [12]. BMMSCs and HUVECs were purchased from PromoCell Gmbh. Human PBMCs were purchased from AllCells LLC.

Animals

Male immunodeficient NOD/Shi-scid IL2rγnull (NOG) mice (6 to 8 weeks old) were purchased from In Vivo Sciences International. The experimental protocol was approved by the Institutional Animal Care and Use Committee of Osaka University.

Hindlimb Ischemia (HLI) Model (Chronic Type-N Model)

HLI (Type-N) was generated as previously described [12]. Briefly, the femoral and saphenous arteries and veins and lateral marginal vein of NOG mice were dissected out and severed. Laser Doppler perfusion imaging (LDPI) was performed as previously described [12].

Cell Transplantation into the HLI Model

Type-N mice with both a perfusion rate of < 40% (ischemic/normal) with LDIP and four or five necrotic nails were selected 3 weeks after the operation. The following day, the selected mice were anesthetized, the skin was incised, and the muscles of the lower limbs were exposed. Cells (3 × 106 cells) in plain MegaCell medium containing 5 nM bFGF were intramuscularly transplanted into five sites in the ischemic limb (20 µL/site). LDPI was performed every 2 or 3 weeks.

Magnetic Resonance Imaging

Detailed methods are described in Supplementary Information. Briefly, blood flow in the feet of mice was assessed by using dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) 2 to 3 weeks after cell transplantation. Prior to DCE-MRI, LDPI was performed in the same week. All MRI examinations were performed on an 11.7-T small animal MRI scanner (Bruker BioSpin, Ettlingen, Germany). Mice were anesthetized with an isoflurane during pretreatment and MRI scans. In DCE-MRI, the whole of both left and right feet was scanned using a fast low-angle shot sequence, and 210 frames of images were acquired for 7 min at 2-s intervals. A bolus of 0.1 mmol/kg gadolinium-based contrast agent (Omniscan; GE Healthcare, Milwaukee, WI, USA) was administered intravenously through the left external jugular vein at 2 min after starting the serial image acquisition. DCE-MRI data were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA). For each foot, region of interest (ROI) was manually drawn over the whole foot, and the signal intensity-time curve in the ROI was obtained. The initial area under the signal intensity-time curve over the first 60 s after Omniscan injection (IAUC60) was calculated as an index for blood flow. The left/right (ischemic/normal) ratio of the IAUC60 was then calculated.

Statistical Analysis

Statistical analysis was performed by Student’s t-test or one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test (GraphPad Prism ®, Graph Pad Software, San Diego CA, USA). P < 0.05 was considered significant.

Results

Differentiation Induction from ES Cells into Pericyte-Like Cells

In order to generate pericyte-like cells from ES cells, we performed a partially modified differentiation method via mesenchymoangioblasts (Fig. 1a) [26, 29, 30]. First, feeder-free ES cells were differentiated into primitive posterior mesoderm, and the appearance of apelin receptor (APLNR) + PGFDRα + cells was confirmed by flow cytometry on day 2 (Fig. 1b). Next, the formation of mesenchymal colonies was induced using spheroid-forming dishes in basic fibroblast growth factor (FGF2)–supplemented media instead of methylcellulose culture (Fig. 1c). After 12 days, colonies larger than the 100-µm mesh were collected and cultured on dishes coated with fibronectin and collagen for 3 days in platelet-derived growth factor isoform BB (PDGF-BB) and FGF2-supplemented media to differentiate into pericyte monolayers. Flow cytometry was performed to examine whether these cells exhibited the characteristics of pericytes. Primary MABs, which show features of pericytes, were used for comparison. Both ES (SEES5, SEES6) cell-derived monolayer cells and primary MABs-derived monolayer cells expressed mesenchymal stem cell markers such as CD73 and CD105 and expressed pericyte markers such as CD146, NG2, and PDGFRb, but expressed no CD31. The expression of mesenchymal stem cell markers and pericyte markers was confirmed for two ES (SEES5, SEES6) cell-derived monolayer cells. Thus, these data suggest that the monolayer cells have a pericyte-like phenotype (Fig. 1d).

Fig. 1
figure 1

Differentiation induction from ES cells into pericyte-like cells. a Schematic diagram of the differentiation protocol to generate pericyte-like cells from hESCs. Following mesoderm induction (Day 0–2, cells were transferred into spheroid-forming dishes in basic fibroblast growth factor–supplemented media to induce colony formation). Colonies larger than 100 µm were collected on day 14 and plated on fibronectin and collagen-coated plates to induce pericyte differentiation. b Flow cytometry of the primitive mesodermal population after 2 days of hESC culture. c Photograph shows a colony in the spheroid-forming dish (scale bar, 100 µm). d Flow cytometry analysis of differentiated monolayer cells derived from two different ES cells and primary mesoangioblasts. Histograms represent surface staining (black line) compared to the isotype control (shaded gray)

Full size image

Angiogenic Potential of the CD56-Negative Fraction of the Differentiated Monolayer Cells

Flow cytometry analysis showed that the differentiated cells expressed CD56 extensively (Fig. 1d), but the expression of CD56 by primary MABs was negative [31]. Therefore, we sorted CD56-negative and CD56-positive cells to compare the potential for angiogenesis (Fig. 2a). Since pericytes have been reported to secrete VEGF [19], we measured VEGF protein in CD56-negative and CD56-positive cell cultures by ELISA and found that the CD56-negative cells secreted significantly higher amounts of VEGF than did the CD56-positive cells (Fig. 2b).

Fig. 2
figure 2figure 2

Angiogenesis potential of ES-derived pericyte-like cells. a Flow cytometry analysis showing gating (upper) on CD56 sorting and CD56 expression in CD56-negative and CD56-positive cells after sorting (lower). b The secretion of VEGF by CD56-negative and CD56-positive cells. Data are expressed as means ± SEM (n = 3 per group). ***P < 0.001. c Photographs of the Matrigel plug in each condition: HUVECs alone (left), CD56( −) ES-PLCs plus HUVECs (center), CD56( +) cells plus HUVECs (right). d Quantification of the hemoglobin content of Matrigel plugs. Data are expressed as mean ± SEM; n = 3 for each Matrigel. *P = 0.028 CD56( −) ES-PLCs/HUVEC vs CD56( +) cells/HUVECs. e H&E-stained section (left) and immunohistochemically co-stained section (right) of the Matrigel plug with ES-PLCs plus HUVECs. Arrows indicate blood vessels. Human lamin A/C (brown) and smooth muscle actin (red) are shown in the co-stained section (scale bar = 20 µm)

Full size image

Next, Matrigel plug assay with HUVECs alone, CD56-negative cells plus HUVECs, and CD56-positive cells plus HUVECs were performed, and no clear neovascularization was observed with HUVECs alone. On the other hand, photographic data and quantification of the hemoglobin content showed that angiogenesis was clearly induced in Matrigel with CD56-negative and CD56-positive cells combined with HUVECs (Fig. 2c, d). The amount of hemoglobin in the Matrigel of CD56-negative cells with HUVECS was significantly higher than that of CD56-positive cells with HUVECS (568 µg/mL and 242 µg/mL, respectively). In addition, IHC analyses of the Matrigel showed that SMA-positive CD56-negative cells were present around blood vessels (Fig. 2e). Thus, the established CD56-negative cells were found to have more pericyte-like properties than the CD56-positive cells and were named “ES-derived pericyte-like cells” (ES-PLCs).

High Proliferative Potential and Angiogenic Cytokine Production of ES-PLCs

Since a large number of cells are required for clinical settings of CLI, we assessed the proliferation capacity of ES-PLCs. Primary MABs with pericyte function were also assessed as a comparison. As a result, ES-PLCs from different ES cells showed a higher proliferative capacity than primary MABs (Fig. 3a). Next, we examined the paracrine factors related to angiogenesis and found that ES-PLCs secreted angiogenic factors such as Ang-1, SDF-1, and IL-8 (SEES6-derived ES-PLCs, 2848 pg/mL, 2729 pg/mL, 218 pg/mL, and SEES5-derived ES-PLCs, 2452 pg/mL, 5974 pg/mL, 86 pg/mL) in addition to VEGF (Fig. 3b). Figures 2 and 3 showed that ES-PLCs had high proliferation capacity, secreted angiogenic factors to form new blood vessels, and played a role as pericytes in the process of angiogenesis.

Fig. 3
figure 3

Expandability and cytokine production of ES-PLCs. a Growth curves of SEES6-derived ES-PLCs (filled circle), SEES5-derived ES-PLCs (filled square), and primary mesoangioblasts (filled triangle). This growth curve was obtained by a single experiment. b Proangiogenic proteins in ES-PLCs. ES-PLCs were cultured for 72 h in a humidified incubator, with 5% CO2 at 37 °C, under hypoxic conditions (5% O2). Supernatants were collected, and the protein concentration was determined by ELISA. Data are expressed as mean ± SEM, n = 3

Full size image

Immunomodulation Potential of ES-PLCs

Immune rejection is an unavoidable issue when allogeneic cells are administrated in clinical settings. We examined the expression of HLA molecules and PD-1 ligand in ES-PLCs activated with or without IFNγ. Figure 4a revealed that the expression of HLA class II molecules was not confirmed in unstimulated ES-PLCs but was induced in activated ES-PLCs. The level of increased expression of HLA class II molecules in ES-PLCs was comparable to that in BMMSCs. On the other hand, the expression of PD-L1 and PD-L2, which inhibit T-cell proliferation via PD-1, was also upregulated in activated ES-PLCs (Fig. 4a). Next, since it has been reported that mesenchymal stem cells inhibit the proliferation of T cells [32, 33], we examined whether ES-PLCs would have the same effect. As a result, Fig. 4b showed that ES-PLCs strongly suppressed T-cell proliferation in a cell-number-dependent manner.

Fig. 4
figure 4

Immunomodulatory functions of ES-PLCs. a Flow cytometry analysis of activated or non-activated BMMSCs and ES-PLCs. b T-cell proliferation assays. Assays were performed using carboxyfluorescein succinimidyl ester (CFSE)–labeled human peripheral blood mononuclear cells (PBMC) activated with CD3 and CD28-coated beads and co-cultured with or without human BMMSCs or ES-PLCs or for 6 days. Histograms showing number of CFSE-labeled activated CD3 positive cells. **P < 0.01, ***P < 0.001

Full size image

Enhanced Angiogenic Potential of bFGF-ES-PLCs

Although bFGF has been reported to be a potent angiogenic factor, ES-PLCs did not secrete bFGF (Fig. 3b). For this reason, we introduced bFGF gene into ES-PLCs to enhance angiogenic potential. Firstly, to promote bFGF secretion in cells, the secretory sequence of bone morphogenetic protein 2/4 was mutated to two cysteine locations [34] and linked to the bFGF gene, and then ES-PLCs were infected with lentivirus encoding bFGF, and ES-PLCs expressing bFGF (bFGF-ES-PLCs) were established. Consequently, ELISA data showed that bFGF was abundantly expressed in culture medium of bFGF-ES-PLCs (64.3 ng/mL) (Fig. 5a). Next, we investigated whether bFGF-ES-PLCs enhance angiogenic function in vivo compared with ES-PLCs. In Matrigel plug assay, bFGF-ES-PLCs showed significantly higher angiogenesis than ES-PLCs (Fig. 5b, c). Moreover, we examined whether bFGF-ES-PLCs restore blood flow in hindlimb ischemia (HLI) mice better than ES-PLCs. In severe ischemic conditions, where therapeutic transplantation of BMMSCs did not improve blood flow (Supplemental Fig. 1), therapeutic implantation of SEES6-derived ES-PLC did not restore blood flow compared to medium control to a significant degree. However, SEES6-derived bFGF-ES-PLCs therapeutic transplantation markedly and significantly improved blood flow rate compared to medium control and ES-PLCs (P < 0.001 and P < 0.01, respectively). Given the superior performance of bFGF-ES-PLCs, we focused our subsequent experiments on bFGF-ES-PLCs and furthered probed their in vivo effects in more detail. A separate experiment was performed to quantify the microvessel density in the footpads of vehicle vs bFGF-ES-PLC-treated HLI animals. CD31 staining in the foot was performed to confirm the increase in neovascularization. The number of CD31-expressing cells significantly increased in bFGF-ES-PLCs group, which may reflect the improvement of blood flow (P < 0.001) (Fig. 5d, e). DCE-MRI enables noninvasively measurement of blood circulation in deep tissue. DCE-MRI also showed that bFGF-ES-PLCs transplantation increased blood flow in the ischemic feet (Fig. 6a). There was a relative correlation (r = 0.7442) between the blood flow ratios obtained by DCE-MRI and those observed in the LDPI (Fig. 6b). In this experiment, we also observed increased CD31 expression at the site of bFGF-ES-PLCs transplantation (Supplemental Fig. 2). This data was separate from the experiment in Fig. 5 and shows that angiogenesis was observed 2–3 weeks after bFGF-ES-PLCs administration.

Fig. 5
figure 5figure 5

Evaluation of bFGF-transduced ES-PLCs for angiogenesis. a bFGF transduced ES-PLCs secreted bFGF. Data are expressed as mean ± SEM; n = 3 for each. b Photographs of the Matrigel plug in each condition: HUVECs alone (left), ES-PLCs plus HUVECs (center), bFGF-ES-PLCs plus HUVECs (right). c Quantification of the hemoglobin content of Matrigel plugs. Data are expressed as mean ± SEM; n = 3 for each Matrigel. ***P < 0.001. d Quantification of laser Doppler perfusion images (upper) and area under the curve in three treatment arms (lower) in chronic Type-N mice. Blood flow was measured in selected Type-N mice. Data are expressed as mean ± SEM; n = 9 for medium, n = 7 for ES-PLCs, n = 9 for bFGF-ES-PLCs. **P < 0.01, ***P < 0.001. e Quantification of microvessel density in the footpads by CD31 staining. The microvessel counts were counted in three random fields (upper). The representative images of immunohistochemical staining are shown (lower). Data are expressed as mean ± SEM; n = 11 for medium, n = 11 for bFGF-ES-PLCs. ***P < 0.001

Full size image
Fig. 6
figure 6

Comparative study of assessing the effects of bFGF-transduced ES-PLCs on improving blood flow by LPDI and DCE-MRI. a Quantification of laser Doppler perfusion images (left) and the temporal change of the IAUC60 ratio (right) in the foot region. Data are expressed as mean ± SEM; n = 8 for medium, n = 11 for bFGF-ES-PLCs. *P < 0.05, **P < 0.01. b Correlation of the foot blood flow ratios (ischemic/normal) obtained by LDPI with those by DCE-MRI. Correlation was analyzed using Pearson’s rank correlation coefficient (r = 0.7442, P < 0.001)

Full size image

Discussion

We established ES-derived pericyte-like cells (ES-PLCs) with angiogenic properties by selecting CD56-negative population for the treatment of ischemic disease and then investigated the pharmacological effects in a chronic Type-N murine model. Until now, there have been several protocols for differentiating human ES cells into pericyte-like cells, but they have not been proliferative enough for clinical use, nor have they expressed enough angiogenic factors. We addressed these problems and succeeded in obtaining proliferative pericyte-like cells that express VEGF. ES-PLCs were highly proliferative in vitro; expressed VEGF, Ang-1, and SDF-1; and generated new blood vessels in collaboration with HUVECs in Matrigel assays. Most experiments were performed at PDL = 10–20, but ES-PLCs at PDL = 30 also showed angiogenesis in the Matrigel plug assay (data not shown). To further enhance angiogenic function, we introduced bFGF gene into ES-PLCs with lentivirus, which induced more angiogenesis in Matrigel and significantly restored blood flow in chronic HLI. Consequently, ES-PLCs with bFGF overexpression may be promising for CLI therapy.

Mesenchymoangioblasts were reported to be formed in semisolid culture, but its viscosity makes it difficult to obtain colonies of the same size. Therefore, we performed a liquid suspension culture with spheroid-forming plates to stably form a large number of mesenchymoangioblast-like colonies. In addition, we have improved a differentiation protocol by CD56-negative cell selection to obtain ES-PLCs with high angiogenic properties. When surface markers of ES-PLCs were examined in Fig. 1, pericyte markers such as CD146, CD105, and NG2 were confirmed, but the expression pattern of CD56 in ES-PLCs was different from that of primary MABs. CD56 is known as one of the early mesodermal markers, and CD326-CD56 + cells can have the potential to differentiate into multiple mesodermal lineages such as blood, endothelium, cardiomyocytes, mesenchymal stem cells, and smooth muscle cells [35]. From this report, it was considered that ES-PLCs with CD56 expression may not completely differentiate from mesoderm into mature pericytes. In fact, ES-PLCs in CD56-negative fractions showed higher angiogenic potential in Matrigel and VEGF expression assays than those in CD56-positive fractions. Therefore, CD56-negative cells are suggested to be closer to mature pericytes.

In clinical trials, avoiding immune rejection of ES-derived cells is an important issue in allogeneic cell therapy. Allogeneic BMMSCs have been clinically evaluated for various diseases including CLI, and no immune rejection has been observed due to their immunomodulatory function [36, 37]. In fact, ES-PLCs had immunomodulatory effects similar to those of BMMSCs. Flow cytometry analysis showed that the expression of HLA class II molecules was upregulated in activated ES-PLCs as well as in activated BMMSCs, while the expression of PD-L1 and PD-L2, which are thought to be part of the mechanism of immunosuppression, was increased, suggesting that they contribute to avoiding immune rejection in CLI patients.

We previously reported that the combination of primary MABs and HUVECs significantly restored blood flow in chronic Type-N mice in which cilostazol was not effective. In most HLI models, compounds or cells are administered immediately after model creation, whereas Type-N is a therapeutic HLI model in which cells are administered several weeks after ischemia induction. Type-N may be more clinically predictive than the conventional HLI model [12]. In this study, we evaluated the angiogenic potential of ES-PLCs expressing bFGF in chronic Type-N mice. As shown in Fig. 5, ES-PLCs with bFGF overexpression significantly restored blood flow in chronic Type-N mice. In contrast, BM-MSCs did not restore it. These data suggest that ES-PLCs expressing bFGF have stronger angiogenesis than the BM-MSC administration [12]. We also attempted to overexpress bFGF in BM-MSCs to determine their effect on angiogenesis, but unfortunately, when bFGF was expressed in these cells, their morphology changed and they stopped functioning (data not shown). Since ES-PLCs secreted Ang-1, SDF-1, IL-8, and VEGF protein, the paracrine effect including bFGF greatly contributes to the restoration of blood flow through therapeutic angiogenesis. The function of ES-PLC as a pericyte may be also crucial for angiogenesis.

We have shown by not only LDPI but also DCE-MRI that bFGF-ES-PLCs improved blood flow. The increased IAUC60 in ischemic feet transplanted with bFGF-ES-PLCs indicated that contrast agent administered intravenously has reached new blood vessels generated by the bFGF-ES-PLCs, suggesting improved blood flow in ischemic feet. LDPI detected improved blood flow in the skin surface of foot, while DCE-MRI did it in the entire foot including deep tissues. Currently, ankle-brachial index (ABI) and transcutaneous oxygen pressure (TcPO2) are used in clinical practice as indicators of improved perfusion in critical limb ischemia [38], but it is difficult to accurately measure these parameters in animal models. On the other hand, DCE-MRI has already been used in both clinical practice [39, 40] and preclinical settings. In addition, DCE-MRI in contrast with LDPI can assess blood flow in the entire foot noninvasively. Our data suggest that DCE-MRI can be a useful evaluation system from the perspective of translational research.

As shown in Fig. 5, the concentration of bFGF protein in ES-PLCs infected with MOI20 lentivirus was approximately 60 ng/mL after 3 days of culture. Because endogenous expression of VEGF, SDF-1, and Ang-1 in ES-PLCs was approximately 1–5 ng/mL, bFGF expression level by lentivirus was thought to be too high. Since overexpression of bFGF was reported to cause dysplasia [41], bFGF expression levels and exposure should be minimized for clinical trials. Because bFGF exhibits metastasis-promoting and growth-promoting effects on tumor cells [42, 43], bFGF sprays used in clinical practice are not administered to patients with malignant tumors. It may be useful to control the expression of bFGF. In terms of tumorigenicity, the methods of gene transfer and the control of gene expression need to be optimized for clinical trials. In addition, another safety test is necessary to ensure that the bFGF-ES-PLCs are not contaminated with original ES cells. Teratoma formation by residual undifferentiated ES cells is a major concern in the development of cell therapy using ES-derived differentiated cells. This possibility needs to be confirmed in a tumorigenesis study in the future using immunocompromised NSG or NOG mice.

In conclusion, it was possible to purify pericyte-like cells with high proliferative potential and angiogenic function from ES cells. bFGF-ES-PLCs significantly enhanced angiogenesis and improved blood flow in a chronic HLI model, more potently than BMMSCs. In addition, bFGF-ES-PLCs showed strong immunomodulatory function. Given the noted growth-promoting effects of bFGF, there may be an increased risk of tumor exacerbation. Therefore, the method of gene transfer and the control of gene expression should be carefully considered and optimized for clinical trials.