Prolonged Maturation Culture Favors a Reduction in the Tumorigenicity and the Dopaminergic Function of Human ESC-Derived Neural Cells in a Primate Model of Parkinson's Disease§

Authors

  • Daisuke Doi,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Cell Growth and Differentiation, Center for iPS Cell Research and ApplicationInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    3. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Asuka Morizane,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Cell Growth and Differentiation, Center for iPS Cell Research and ApplicationInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Tetsuhiro Kikuchi,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Hirotaka Onoe,

    1. Functional Probe Research Laboratory, RIKEN Center for Molecular Imaging Science, Kobe, Japan
    Search for more papers by this author
  • Takuya Hayashi,

    1. Department of Cell Growth and Differentiation, Center for iPS Cell Research and ApplicationInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Functional Probe Research Laboratory, RIKEN Center for Molecular Imaging Science, Kobe, Japan
    Search for more papers by this author
  • Toshiyuki Kawasaki,

    1. Functional Probe Research Laboratory, RIKEN Center for Molecular Imaging Science, Kobe, Japan
    Search for more papers by this author
  • Makoto Motono,

    1. Department of Cell Growth and Differentiation, Center for iPS Cell Research and ApplicationInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Yoshiki Sasai,

    1. Organogenesis and Neurogenesis Group, RIKEN Center for Developmental Biology, Kobe, Japan
    Search for more papers by this author
  • Hidemoto Saiki,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Masanori Gomi,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Tatsuya Yoshikawa,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Hideki Hayashi,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Mizuya Shinoyama,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Mohamed M. Refaat,

    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Hirofumi Suemori,

    1. Laboratory of Embryonic Stem Cell Research, Stem Cell Research Center, Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Susumu Miyamoto,

    1. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    Search for more papers by this author
  • Jun Takahashi

    Corresponding author
    1. Department of Biological Repair, Institute for Frontier Medical SciencesInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    2. Department of Cell Growth and Differentiation, Center for iPS Cell Research and ApplicationInstitute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    3. Department of Neurosurgery, Clinical Neuroscience, Graduate School of Medicine,Institute for Frontier Medical Sciences, Kyoto University, Kyoto, Japan
    • Department of Biological Repair, Institute for Frontier Medical Sciences, Kyoto University, 53 Shogoin Kawahara-cho, Sakyo-ku, Kyoto 606-8507, Japan
    Search for more papers by this author
    • Telephone: 81-75-751-4840; Fax: 81-75-751-4840


  • Author contributions: D.D. and A.M.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; T.K., H.O., T.H., and T.K.: collection and/or assembly of data, data analysis and interpretation, and manuscript writing; Y.S.: administrative support, data analysis and interpretation, and manuscript writing; M.M., H.S., M.G., T.Y., H.H., M.S., M.M.R., and H.S.: collection and/or assembly of data; S.M.: administrative support and data analysis and interpretation; J.T.: conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript. D.D. and A.M. contributed equally to this article.

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

  • §

    First published online in STEM CELLSEXPRESS February 10, 2012; available online without subscription thorugh the open access option.

Abstract

For the safe clinical application of embryonic stem cells (ESCs) for neurological diseases, it is critical to evaluate the tumorigenicity and function of human ESC (hESC)-derived neural cells in primates. We have herein, for the first time, compared the growth and function of hESC-derived cells with different stages of neural differentiation implanted in the brains of primate models of Parkinson's disease. We herein show that residual undifferentiated cells expressing ESC markers present in the cell preparation can induce tumor formation in the monkey brain. In contrast, a cell preparation matured by 42-day culture with brain-derived neurotrophic factor/glial cell line-derived neurotrophic factor (BDNF/GDNF) treatment did not form tumors and survived as primarily dopaminergic (DA) neurons. In addition, the monkeys with such grafts showed behavioral improvement for at least 12 months. These results support the idea that hESCs, if appropriately matured, can serve as a source for DA neurons without forming any tumors in a primate brain. STEM CELLS 2012;30:935–945

INTRODUCTION

Cell replacement therapy using fetal midbrain cells has been shown to relieve motor symptoms in human Parkinson's disease (PD) patients [1, 2]. One of bottlenecks of this therapy, however, is the limited availability of donor materials. Because of their pluripotency and unlimited proliferative activity, embryonic stem cells (ESCs) are regarded as another candidate for donor cells. Whereas human ESC (hESC)-derived dopaminergic (DA) neurons, when grafted, can improve motor functions in rat PD models (xenograft) [3–6], such cell preparations occasionally form tumors [5, 7, 8]. For the safe clinical application of hESC-derived cells, it is critical to evaluate their tumorigenicity and function in primate PD models, which can be generated by systemic administration of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) [9]. We and others have reported the transplantation of monkey ESC-derived neural progenitor cells (NPCs) [10, 11], monkey fetal tissues [12], or human embryonic NPCs [13] into the brains of monkey models of PD. The tumorigenicity and function of hESC-derived NPCs, however, have never been reported in primate models. Thus, the first aim of this study was to determine whether hESCs are able to form tumors in the primate brain and to examine the characteristics of the tumors that arise. The second aim was to devise a method for reducing the tumorigenicity of the hESC-derived cells in the primate brain. Finally, the third aim was to determine whether hESC-derived cells are able to function as a source for DA neurons in the brains of the PD model monkeys. To answer these questions, we generated NPCs from hESCs by a modified stromal cell-derived inducing activity (SDIA) method [14, 15] using different time periods for differentiation, then grafted the NPCs into the brains of MPTP-treated monkeys, and then compared the growth of the cells, the behavior of the monkeys, and the histological findings for 12 months.

MATERIALS AND METHODS

hESC Culture

hESCs (KhES-1, 2) [16] were used in conformity with “The Guidelines for Derivation and Utilization of Human Embryonic Stem Cells” of the Ministry of Education, Culture, Sports, Science, and Technology of Japan, after approval by the Institutional Review Board. These two lines are listed among the well-characterized hESCs by the International Stem Cell Initiative and have similar characteristics as representative hESC lines [17]. Undifferentiated hESCs were maintained as previously described ([16], Supporting Information Methods). The cells between passage 10 and 40 were subjected to experiments, and we confirmed that the ESCs retained a normal karyotype (Supporting Information Fig. S1).

Induction of Neural Progenitors from hESCs

hESCs were dissociated into single cells after 5 minutes of incubation in Accumax (Innovate Cell Technologies, San Diego, CA, www.innovativecelltech.com) and were plated on 0.1% gelatin-coated dishes to isolate mouse embryonic fibroblasts and hESCs. After 30 minutes of incubation, undifferentiated ESCs were detached by mild tapping of the dishes and were replated on thick growth factor-reduced Matrigel (MG, 4 μl/cm2; Becton Dickinson, Franklin Lakes, NJ, www.bd.com), which covered the dissociated ESCs, then they were cultured in PA6 conditioned medium (CM [15]). To reduce cell death, the ROCK inhibitor, Y-27632 (10 μM; Wako, Osaka, Japan, www.wako-chem.co.jp), was added in the culture medium for the first 3 days [18, 19]. On day 7, the ESC-derived colonies were detached from the MG by incubation with collagenase-trypsin-knockout serum replacement (Supporting Information Methods) dissociation solution and were replated on 6 cm Petri dishes. Next, the colonies were cultured in neurobasal medium-based PA6 CM supplemented with B-27, l-glutamine (2 mM), and fibroblast growth factor 2 (FGF2; 20 ng/ml) as floating spheres for 1 or 3 weeks. To promote DA differentiation, we added human recombinant Shh (200 ng/ml; R&D, Minneapolis, MN, www.rndsystems.com, C24II N terminus) and FGF8b (100 ng/ml; PeproTech, Rocky Hill, NJ, www.peprotech.com) during a floating culture from day 14 until day 28. After that, the spheres were plated on 6 cm plastic dishes coated with poly-L-ornithine and laminin (O/L-coated dishes) and cultured in nonconditioned neurobasal medium supplemented with B-27, L-glutamine (2 mM), BDNF (20 ng/ml), and GDNF (2 ng/ml) without FGF2 for 1–2 weeks. For the in vitro immunofluorescence studies, the spheres were plated on O/L-coated eight-well chamber slides.

Polymerase Chain Reaction Analysis

The total RNA was extracted using an RNeasy Mini kit (QIAGEN, Venlo, Netherlands, www.qiagen.com), and polymerase chain reaction (PCR) analyses were performed as described in Supporting Information Methods.

Immunofluorescence Studies

In vitro and in vivo immunohistochemical analyses were carried out as described in Supporting Information Methods.

Flow Cytometry

Cells were harvested on days 14–28 using Accumax and were subjected to flow cytometry as described in Supporting Information Methods.

Dopamine Release Assay

hESCs were differentiated with a modified SDIA method for 35 days, then the dopamine release assay was performed as previously described ([15], Supporting Information Methods).

Animals and Cell Transplantation

Animals were cared for and handled according to the Guidelines for Animal Experiments of Kyoto University and the Guide for the Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources (Washington, DC, U.S.).

Five-week-old male C.B-17/Icr-scid/scidJcl-severe combined immunodeficiency (SCID) mice (CLEA Japan, Tokyo, Japan, www.clea-japan.com) were used as transplant recipients. Each mouse received a stereotactic injection of 2 μl of cells (KhES-1- and KhES-2-derived cells, 1 × l05 cells per microliter) through a 26-gauge needle into the right side of the striatum (from the Bregma: A +0.5, L +2.0, V +3.0 and +2.5, incisor bar 0). Eight weeks after transplantation, the animals were euthanized with pentobarbital and perfused transcardially with 4% paraformaldehyde. The brains were cut with a cryostat at 40-μm thickness and then were mounted on slides. Every six sections containing the graft region were chosen for either H&E staining or immunofluorescence studies. The graft size was calculated from the area stained with the anti-human nuclear antibody, using the Bioanalyzer software program (Keyence, Osaka, Japan, www.keyence.co.jp).

Monkey Parkinsonian models were generated as previously described ([10], Supporting Information Methods). The coordinates of the targets were obtained from magnetic resonance images (MRIs) of each animal, and hESC-derived cells (KhES-1) were stereotactically transplanted into the bilateral putamen of the MPTP-treated monkeys. The first group included three monkeys (n = 1: No. 52, and n = 2: Nos. 55 and 56 for d14- and d28-spheres, respectively), and the second group included 10 monkeys (n = 4: Nos. 54, 57, 66, and 69, n = 2: Nos. 62 and 63, n = 4: Nos. 65, 67, 68, and 75 for sham surgery, d35-grafts, and d42-grafts, respectively). The floating spheres (d14- and d28-spheres) were harvested directly, the attached spheres (d35- or d42-spheres) were detached by mildly tapping the poly-L-ornithine/laminin-coated dishes, and then they were collected into 500 μl microtubes without dissociation. The size of the spheres was approximately 400 μm or less in diameter. For cell counting, some of the spheres were picked up and dissociated into single cells by 0.25% trypsin. The sphere suspension was prepared at 1 × 105 cells per microliter in Neurobasal medium supplemented with B-27, L-glutamine (2 mM), brain-derived neurotrophic factor (BDNF) (20 ng/ml), glial cell line-derived neurotrophic factor (GDNF) (2 ng/ml), and DNaseI (0.05 mg/ml; Roche, Basel, Switzerland, www.roche.com), then 1 μl of this sphere suspension was injected through a 22-gauge needle along six tracts/side (four injection sites/tract; 4.8 × 106 cells per animal). For the control monkeys, the same volume of the culture medium was injected. After surgery, the monkeys were given antibiotics for 3 days and a daily immunosuppressant (FK506, 0.05 mg/kg, i.m.; Astellas, Tokyo, Japan, www.astellas.com) until sacrifice. The average trough values of circulating FK506 were 20.1 ± 10.5 ng/ml (n = 31, Supporting Information Methods for the values of each animal).

MRI Studies

T1- and T2-weighted MRIs were obtained using a 3-T MRI scanner (GE Signa Horizon Lx VH3; GE Healthcare, Little Chalfont, Buckinghamshire, U.K., www3.gehealthcare.com) and an eight-channel receiving coil (Supporting Information Methods for sequence parameters used for the MRI scans). The graft volume was analyzed by using functional MRIs of the brain (FMRIB) Software Libraries [20]. In T1- and T2-weighted images, non-brain structures were removed using a brain extraction tool [21] and were coregistered with each other by rigid body transformation using the FMRIB's linear registration tool [22]. Next, to identify a graft region, we applied automated segmentation to both T1- and T2-weighted images using a fast automated segmentation tool [23], which is based on a method with a hidden Markov random field model and an associated expectation-maximization algorithm. We applied a multichannel set of images of four different types of tissues (both T1- and T2-weighted images) to be segmented and searched. This process reproducibly generated four types of tissues: white matter, gray matter, cerebrospinal fluid (CSF) space (plus graft region if any), and the putamen. Next, the partial volume images for the segment of the CSF plus graft were thresholded at 30%, binarized, and manually edited to remove the CSF region. The resultant segment image for the graft was used to calculate the volume of the graft.

Positron Emission Tomography Studies

All positron emission tomography (PET) scans were performed as previously described in accordance with the Principles of Laboratory Animal Care (NIH publication 85-23, revised 1985) [24]. Briefly, monkeys were anesthetized with a continuous i.v. infusion of propofol (10–20 mg/kg per hour). During anesthesia, oxygen gas was supplied through a nasal cannula throughout the study, and the heart rate, electrocardiogram, body temperature, and oxygen saturation were monitored. Before the emission scan, the animal's head was placed near the center of the PET camera's field of view, and the transmission scan was performed for 30 minutes with a rotating 68Ge-68Ga pin source to be used for attenuation correction. At the start of the emission scan, one of the 6-[18F]fluoro-L-3,4-dihydroxyphenylalanine([18F]DOPA), 3′-[18F]fluoro-3′-deoxythymidine, or 2-[18F]fluoro-2-deoxy-glucose ([18F]FDG) solutions (approximately 37 MBq/kg in 2.0 ml saline) was intravenously injected as a bolus (within a period of 10 seconds) via a cannula implanted into the saphenous vein (Supporting Information Methods for the preparation of these solutions). For [18F]DOPA PET scans, a solution of carbidopa (10 mg/kg, i.v.) was administered intravenously 5 minutes prior to the scan, in order to minimize peripheral decarboxylation of [18F]DOPA. The PET images were acquired in a three-dimensional list mode by using an animal PET scanner, microPET Focus-220 (Siemens Healthcare, Malvern, PA, www.medical.siemens.com), with a scan duration of 120 minutes for [18F]DOPA and 90 minutes for [18F]FLT and [18F]FDG, respectively. The axial and transaxial resolutions of the PET scanner were 1.35 mm at full-width-half-maximum. The [18F]DOPA uptake rate (Ki, minute−1), an index of dopamine biosynthesis, was calculated using the Patlak graphical analysis system [25], in which the cerebellum was used as the reference region [26] (Supporting Information Methods for the analysis of PET data).

Behavioral Analysis

The behavior of each monkey was evaluated according to a previously described rating scale for monkey PD models [10]. The normal and minimum score is 0, whereas the maximum total score is 24. The evaluation was performed by a trained examiner who was not involved in the cell transplantation. Spontaneous movements were measured as previously described [27]. The video records during 15–30 minutes were analyzed using the Vigie Primate video-based analysis system (View Point, Lyon, France, www.vplsi.com), and changes in pixels from one image to the next were counted every 66.67 microseconds. Regarding L-dopa tests, L-dopa+Benserazide (20 + 5 mg/kg) was intravenously injected into each monkey, then the amount of spontaneous movements during 15–30 and 45–60 minutes was analyzed.

Statistical Analysis

The statistical analyses were performed using a commercially available software package (GraphPad Prism, GraphPad Software Inc., La Jolla, CA, www.graphpad.com). Data from experiments about histological evaluations were analyzed by a one-factor single-factor analysis of variance (ANOVA) and a Tukey's post hoc analysis (Figs. 1E, 2C, 2D and Supporting Information Figs. S2A, S2E, S6A, S6C, S6H). The behavioral data were analyzed by a repeated ANOVA with group and repeated measurements of the neurological score (Fig. 6A) or spontaneous movements (Fig. 6C) included as factors. The Huyn-Feldt adjustment was applied to the degree of freedom of the repeated ANOVA if needed. A post hoc analysis with the Bonferroni correction was applied to the time-dependent change in neurological scores when needed (Fig. 6A). The differences were considered statistically significant when p < .05, and the significance level was shown by any one of *, p < .05; **, p < .01; or ***, p < .001. The data are presented as the mean ± SD. All in vitro results were derived from at least three independent experiments.

Figure 1.

Neural induction from human embryonic stem cells (hESCs) using a modified stromal cell-derived inducing activity method. (A): The culture protocol using Matrigel and PA6-conditioned medium, followed by the addition of Shh/FGF8 and BDNF/GDNF. (B): A phase-contrast image (day 7) and immunofluorescence images of the spheres. Oct3/4 (green), Nestin (red), and DAPI (blue) for d7- and d14-spheres; Tuj1 (green), TH (red), and DAPI (blue) for d28-spheres. Bars = 100 μm. (C): The time course analysis of the expression of pluripotent (Oct3/4, Nanog), neural progenitor (Sox1), neuronal (Map2ab), DA progenitor (Lmx1a, En1), immature DA neuron (Nurr1), and mature DA neuron (TH) markers (upper graph); forebrain (Six3), forebrain and midbrain (Otx2), midbrain/hind brain boundary (Pax2), hind brain (Gbx2), and floor plate (Corin, Foxa2) markers (lower graph) by quantitative polymerase chain reaction. The expression levels in hESCs before differentiation (d0) were set to 1, and the relative expression levels (log) are presented as the mean ± SD. (D): The double-labeled immunofluorescence images of the spheres on days 28, 35, and 42 for Tuj1 (green) and TH (red). Bars = 50 μm. (E): The percentages of TH+ cells per Tuj1+ cells at each time point. The data are presented as the mean ± SD (*, p < .05 and ***, p < .001). (F): Immunofluorescence images of differentiated ESCs on O/L-coated slides for TH (red) and Tuj1, AADC, Nurr1, Pitx3, or Girk2 (green). Bars = 100 μm. (G): An high performance liquid chromatography (HPLC) analysis showing dopamine release in the culture medium of differentiated ESCs in response to KCl evoked depolarization. Abbreviations: BDNF, brain-derived neurotrophic factor; bFGF, basic fibroblast growth factor; DA, dopaminergic; DAPI, 4′,6-diamidino-2-phenylindole; GDNF, glial cell line-derived neurotrophic factor.

Figure 2.

Growth of d14-, d28-, d35-, and d42-grafts in the brains of stromal cell-derived inducing activity (SCID) mice. (A): The H&E staining of the brains of SCID mice at 8 weeks after transplantation. Bars = 1 mm (upper) and 100 μm (lower). (B): Immunofluorescence images of the grafts for human-specific nestin (green) and Ki67 (red). Bars = 1 mm (upper) and 100 μm (lower). Insets show magnified images of Ki67+ cells. (C): A comparison of the volumes among d14- (•, n = 8 and 10 for KhES-1 and KhES-2, respectively), d28- (▪, n = 14 and 9), d35 (▴, n = 19 and 15), and d42-grafts (▾, n = 17 and 20). (D): The percentages of the Ki67+ cells among the total cells (DAPI+) in each graft. The data are presented as the mean ± SD (*, p < .05; **, p < .01; ***, p < .001). Abbreviation: DAPI, 4′,6-diamidino-2-phenylindole.

RESULTS

Generation of DA Neurons from hESCs Using a Modified SDIA Method

Previous studies have shown that hESCs are able to generate DA neurons when treated using the SDIA method [5, 8, 28–30]. In this study, we initiated neural induction by culturing hESCs (KhES-1 and -2) on MG-coated dish in serum-free CM from PA6 cells, followed by their expansion as floating spheres in the presence of FGF2 (modified SDIA method; Fig. 1A) [15]. The cells formed spheroidal colonies on thick MG and contained a large proportion of Nestin+ NPCs (>80%) on day 7 (Fig. 1B). The spheres contained both undifferentiated ESCs (Oct3/4+) and NPCs (Nestin+) on day 14, whereas the ESCs (Oct3/4+ and Nanog+) disappeared by day 28 (data not shown). To promote DA differentiation, we added FGF8 and Shh during a floating culture from day 14 until day 28. Then, the spheres were replated on poly-L-ornithine/laminin (O/L)-coated dishes and cultured in neurobasal medium supplemented with BDNF, GDNF, ascorbic acid, and dibutyryl cyclic AMP for 1 or 2 weeks (Fig. 1A). This method was more efficient for neural induction and DA differentiation than the original SDIA method (Supporting Information Fig. S2).

The quantitative reverse transcription (RT) analyses revealed the expression of the undifferentiated cell markers, Oct3/4 and Nanog, to decrease by less than 5% on day 14 compared to day 0, and then to gradually decrease thereafter (Fig. 1C). The expression of a neural cell marker, Sox1, reached a plateau on day 14, while that of a postmitotic neuronal marker, Map2ab, gradually increased. The expression of markers for DA neuron progenitors [31, 32], Lmx1a and En1, decreased after day 28, whereas that of an immature DA neuron marker, Nurr1, and a DA neuron marker, TH, kept increasing, thus indicating that maturation toward DA neurons proceeded during the course of the culture. The analyses also revealed that both forebrain and midbrain cells were generated in our culture method. The expression of a floor plate marker, Corin [33], was mildly increased, but another one, Foxa2 [34], did not increased. An immunofluorescence study revealed the ratio of TH+ cells per Tuj1+ cells on day 28 to be 11.7% ± 2.2%, which thereafter gradually increased to 25.1% ± 6.6% and 34.7% ± 7.3% on days 35 and 42, respectively (Fig. 1D, 1E).

To confirm the differentiation toward midbrain DA neurons, we continued an attached culture on O/L-coated dishes for an additional 2 weeks (day 56). The TH+ cells extended neurites, and they expressed midbrain DA neuron markers, such as AADC, Nurr1, Pitx3, and Girk2 (Fig. 1F), and the ratio of Nurr1+ cells per TH+ cells was 44.7% ± 6.6%. A high performance liquid chromatography (HPLC) analysis revealed that the cells could secrete dopamine in response to high potassium stimulation (Fig. 1G), thus suggesting that the hESCs could generate functional midbrain DA neurons in vitro.

Graft Growth in the Mouse Brain Is Reduced by Prolonged Maturation Culture

Based on the above findings, we next compared the in vivo growth of the spheres on days 14, 28, 35, and 42 (d14-, 28-, 35-, and 42-spheres, respectively) in the mouse brain. The d14-, d28-, d35-, and d42-spheres derived from KhES-1 and KhES-2 cells were grafted into the right putamen of SCID mice (2 × 105 cells per animal). H&E staining of the brain performed 8 weeks later showed no teratomatous or malignant findings to be present. In the d14- and d28-grafts, rosette formation suggesting a neural epithelial nature was observed (Fig. 2A, 2B). A double-labeled immunofluorescence study for nestin, a NPC marker, and Ki67, a marker for proliferating cells, revealed that nestin cells (undifferentiated or non-neural cells) were also proliferating in the d14-grafts (Fig. 2B), while the majority of the cells were nestin+ in other grafts, and some of them were still proliferating. In both cell lines, the d42-grafts were significantly smaller in volume than the other grafts (Fig. 2C). The percentage of Ki67+ cells at 8 weeks post-transplantation was significantly larger in the d14-grafts, thus suggesting that they had a greater potential for tumorigenicity (Fig. 2D). These results suggest a tumorigenic potential of the d14-grafts.

Undifferentiated ESCs Contribute to Tumor Formation in the Brain of MPTP-Treated Monkeys

To investigate their survival and tumorigenicity in a primate brain, the d14- and d28-spheres (KhES-1) were grafted into the bilateral posterior putamen of MPTP-treated cynomolgus monkeys (n = 1 and 2, respectively). In this experiment, we did not add Shh/FGF8 during the culture on days 14–28, thus resulting in a ratio of TH+ cells per Tuj1+ cells of 3.1% ± 1.6% in the d28-spheres. Because the monkeys had already been treated with MPTP and implantation of such immature cells could be painful for them, we minimized the number of the monkeys under the guidance of the animal research committee of our institute. To determine whether the results were reproducible, we injected the cells through multiple tracts bilaterally (4.8 × 106 cells total in 12 tracts per animal). An MRI study performed 6 months after implantation revealed large lesions with perifocal edema in the monkey injected with d14-spheres (Fig. 3C), and the grafts derived from d14-spheres (d14-grafts) were substantially larger in volume than the d28-grafts (Fig. 3A). Between 1 and 6 months, the estimated doubling time for the d14-grafts (10.58 ± 0.24 days) was shorter than for d28-grafts (29.18 ± 6.63 days; Fig. 3B). Between 6 and 9 months, however, the doubling times of both grafts became much longer, suggesting that their growth speed was reduced in both cases.

Figure 3.

The histological analysis of a d14-sphere-derived tumor and the d28-graft. (A): The changes in the graft volumes in each side after transplantation of d14-spheres (No. 52) and d28-spheres (Nos. 55 and 56). (B): The doubling times of the grafts between 1–6 and 6–9 months after transplantation. (C–E): The T2-weighted magnetic resonance images (MRIs) (upper, left), FDG-positron emission tomography (PET) images (lower, left), FLT-PET images (upper, right), and combined MR and FLT-PET images (lower, right) of the monkeys implanted with d14-spheres (C), d28-spheres (D), and culture medium only (E, control) at 9 months. (F): The ratio of the FLT uptake (SUV) in the graft area of the monkeys (Nos. 52, 55, and 56) compared to that in the cerebellum (control area) at 9 months. The gross appearance of the brain section (G) and H&E-stained pictures of the d14-graft (H–K), showing pigmentation and epithelial-like structures (I), the accumulation of undifferentiated cells (J), and neural histology with a clear border (K). The dotted line in (H) indicates the border of the graft. The scale bars in G and H = 1 mm, I, J, and K (left) = 200 μm, (right) = 100 μm. (L): H&E-stained pictures of the d28-graft showing the neural appearance with a clear border. Bars = 100 μm. Abbreviations: FDG, fluoro-2-deoxy-glucose; FLT, fluoro-thymidine; SUV, standard uptake value.

At 6 months, PET for [18F]FDG [35] revealed an increase in the local glucose metabolism in the d14-grafts (Fig. 3C left). In contrast, the metabolism in d28-grafts was low (Fig. 3D left), comparable to the level in a control monkey treated with injections of culture medium (Fig. 3E left). 3′-[18F]fluoro-3′-deoxythymidine [18F]FLT, a fluorinated thymidine analog, is a PET tracer for evaluating cell proliferation activity and has been applied to the diagnosis of various malignant brain tumors [36]. FLT-PET showed a high focal uptake only in the monkey that received d14-spheres (Fig. 3C–3E right). When compared to the cerebellum as a control, the average standard uptake values (SUVs) were higher in the d14-grafts (Fig. 3F), especially in the hotspots (> 3.0).

At 9 months, the monkeys were sacrificed and their tissues were subjected to histological studies. The d14-grafts were well-demarcated, with no malignant findings such as necrosis or pleomorphism (Fig. 3G–3K). Pigmentation and epithelial-like structures were observed (Fig. 3I), but there were no hairs, cartilage, or muscle fibers. The undifferentiated cells with a high nuclear/cytoplasm ratio were tightly accumulated in small patches (Fig. 3J), where a hotspot in the FLT-PET image was overlapped (Fig. 3C). In a major part of the graft, however, the cell density was much lower and more homogeneous, and the majority of the cells were neural cell-like (Fig. 3K). In the two monkeys implanted with d28-spheres, the grafts were also well-demarcated and showed no malignancy or teratoma-like findings (Fig. 3L). In contrast to d14-grafts, however, no accumulation of undifferentiated cells was observed, and most of the cells were homogeneous and neural cell-like.

In both cases, according to the neurological scores [10], there were no apparent changes in the animals' behavior (Supporting Information Fig. S3A, S3B). Nine months after grafting, the d28-grafts contained a small percentage of TH+ neurons (0.62% of total cells), whereas the d14-grafts contained very few of them (<0.01%; Supporting Information Fig. S3C).

These results suggest that d14-spheres, which contained undifferentiated ESCs, retained their tumorigenicity in the monkey brain. The d28-grafts showed milder growth but may lead to an expansion of the grafts with neural components.

Prolonged Maturation of hESCs Reduces Their Tumorigenicity in the Brain of MPTP-Treated Monkeys

To determine whether the maturation culture after day 28 affected the tumorigenicity and DA function in the primate brain, we grafted d35- and d42-spheres (KhES-1) into the bilateral posterior putamen of MPTP-treated monkeys (n = 2 and 5, respectively; 4.8 × 106 cells total in 12 tracts per animal). MRI evaluations at 12 months after transplantation revealed that the grafted cells survived in six out of seven monkeys (Fig. 4A). In animal No.74, we observed high signal masses of the grafts at 1 month, but they had disappeared by 3 months (Supporting Information Fig. S4). Importantly, no sign of tumor formation causing a mass effect or edema was observed in any of these cases. According to the estimated volumes obtained by MRI, both the d35-grafts (Nos. 62 and 63) showed continuous growth through the 12-month period (Fig. 4B). In contrast, three of the four d42-grafts (Nos. 65, 67, 68, and 75) ceased growing after 6 months. In each case, the doubling time became gradually longer, indicating that the grafts were expanding more slowly (Fig. 4C). In PET studies at 12 months, no focal uptake of [18F]FLT was observed (Fig. 4D), and the SUVs in the grafts were almost the same as or less than that in the cerebellum (Fig. 4E), thus indicating that these values were very low in all the cases.

Figure 4.

Reduced tumorigenicity resulted from prolonged differentiation of the human embryonic stem cells. (A): T2-weighted magnetic resonance images (MRIs) of the monkeys implanted with d35- (62, 63) or d42- (65, 67, 68, 75) spheres at 12 months post-transplantation. The arrowheads indicate the grafts. (B): The changes in the graft volumes of each monkey after cell transplantation. (C): The changes in the doubling times of the grafts between 1–3, 3–6, and 6–12 months post-transplantation. (D): A T2-weighted MRI (upper, left), FLT-positron emission tomography (PET) image (lower, left), and combined MRI and FLT-PET images (upper, right) of a monkey implanted with d35-spheres (62) at 12 months. (E): The ratio of the FLT uptake (standard uptake value) in the graft area of the monkeys compared to that in the cerebellum (control area) at 12 months. (F): Double-labeled immunofluorescence images of the d42-graft for Nestin (green), Ki67 (red), and nuclei (DAPI, blue). The dotted lines in F (left) indicate the graft-host border. Bars = 500 μm (left) and 50 μm (right). (G): The percentages of the Ki67+ cells among the total cells (DAPI+) in the grafts on each side at 12 months after implantation. Abbreviations: FLT, fluoro-thymidine; DAPI, 4′,6-diamidino-2-phenylindole.

Some cell-dense clusters were found in the grafts, and they mainly contained Nestin+ NPCs, rather than NeuN+ postmitotic neurons (Figs. 4F, 5A). Of note, no Oct3/4+ or Nanog+ cells were found in these clusters. In these areas, some of the Nestin+ cells expressed Ki67 (Fig. 4F), and the average ratio of Ki67+ cells to the total cells varied between 0.2% and 2.3% (Fig. 4G). These results suggest that d42-spheres retain a very low potential for tumor formation.

Figure 5.

The dopaminergic differentiation of the human embryonic stem cell derived neural progenitor cells in the monkey brain. (A): The gross appearance of the d42-graft with DAB staining for TH. Bar = 1 mm. (B): The image of the boxed area in A. Bar = 100 μm (upper). A magnified image of the boxed area in the upper image is shown in the lower panel. Bar = 50 μm. (C–G): Immunofluorescence images of the d42-graft for TH (green) and VMAT2 (red) (C), for TH (green) and AADC (red) (D), for TH (green) and Pitx3 (red) (E), for TH (green), GAD65/67 (red) and NeuN (blue) (F), for NeuN (green), TH (red), and nuclei (DAPI, blue) (G). Bars = 1 mm (G left), 500 μm (G middle), 100 μm (C–G right). The dotted lines in G (left) indicate graft-host borders. The middle panel of G shows the area in the box in G (left), and G (right) shows the area in the box in the middle panel of G, respectively. The inset shows the magnified image of NeuN+/TH+ cells. (H): The number of TH+ cells in the grafts on each side at 12 months after implantation. (I, J): T2-weighted magnetic resonance images (MRIs) (upper), DOPA-positron emission tomography (PET) images (middle), and combined MR and DOPA-PET images (bottom) of the monkeys injected with culture medium only (I, control) or with d42-spheres (J) at 12 months. (K): Increased DOPA uptake (Ki value) was observed in the graft area compared to that in the other areas of the putamen at 12 months (n = 2 and 4 for d35- and d42-grafts, respectively). (L): The correlation between increased DOPA uptake and the behavior score change at 12 months after implantation (R2 = 0.7099, p = .035). The blue and red dots indicate the d35- and d42-grafts, respectively. Abbreviations: DAB, diaminobenzine; DAPI, 4′,6-diamidino-2-phenylindole.

hESC-Derived Cells Function As DA Neurons in the Brains of MPTP-Treated Monkeys

We next examined the in vivo DA differentiation of the grafted cells. A histological study at 12 months showed clusters of TH+ neurons that extended neurites within the graft and into the surrounding host brain, thus suggesting reinnervation to the putamen by the grafted cells (Fig. 5A, 5B). The number of TH+ cells in each side of the brain varied from 1.3 to 18.6 × 103, and the highest number was observed in the right side of animal No. 75 (Fig. 5H). Many of these TH+ cells were also positive for midbrain DA neuron markers, such as VMAT2, AADC, and Pitx3 (Fig. 5C–5E). The ratio of NeuN+ cells to total cells (4′,6-diamidino-2-phenylindole (DAPI) count) in the grafts was 39.2% ± 3.3% (Fig. 5G). While, the ratio of TH+ cells to NeuN+ cells was 9.7% ± 4.5% (Fig. 5G) and most of the other neurons were GABAergic (Fig. 5F).

DA synthesis can be visualized by PET using 6-[18F]fluoro-L-3,4-dihydroxyphenylalanine [18F]DOPA [37]. At 12 months, DOPA-PET revealed increased uptake of [18F]DOPA in the grafts (Fig. 5I, 5J). The increase in the Ki value of each graft compared to that of anterior parts of the putamen was higher for the d42-grafts (Fig. 5K). As the posterior putamen was more sensitive to MPTP treatment [37], the differences in the Ki values between posterior (estimated graft area) and anterior putamen of the control monkeys were zero or even negative. The increase in these values significantly correlated (R2 = 0.7099, p = .035) with the change in the neurological scores in each monkey, suggesting that dopamine production by the grafted cells contributed to their behavioral improvements (Fig. 5L).

To evaluate the neurological symptoms of the MPTP-treated monkeys before and after cell transplantation, we used two different methods: a neurological rating scale as used previously [10] and a video-based analysis of spontaneous movements [27]. Neurological scores were rated in a blinded manner for each monkey treated with control injections (n = 4), d35-spheres (n = 2), and d42-spheres (n = 4), revealing that only the monkeys implanted with d42-spheres showed significant behavioral recovery after 3 months (Fig. 6A and Supporting Information S5A; F(12, 36) = 25.8, p < .001). The amount of spontaneous movement of each monkey was quantified based on pixel changes in video frames during a 15-minute period (Supporting Information Fig. S5B, S5C). When these values were adjusted as the ratio to the pretransplantation state, a remarkable increase was observed in animal No. 75 (Fig. 6B), and a mild increase was observed in animals No. 66 (control) and No. 65 (d42-spheres). However, we did not find any significant difference in the time course change of the spontaneous movements among the three groups (Fig. 6C).

Figure 6.

The behavioral changes of the monkeys after transplantation of d35- or d42-spheres. (A): The time course of changes in the neurological scores of the monkeys injected with culture medium only (control, blue, n = 4), or implanted with d35- (red, n = 2) or d42-spheres (green, n = 4). The data are presented as the mean ± SD. A repeated ANOVA including group and repeated rating of scores as factors disclosed that there was a significant interaction between the group and repeated rating of scores (F(24, 84) = 4.17, p < .001), indicating that the time course changes of the scores were different among the three groups. Furthermore, a repeated ANOVA for the rated scores in each group revealed that a significant effect of time was found in the rated scores only in the group implanted with d42-grafts (F(12, 36) = 25.8, p < .001). A post hoc analysis in this group, by a pairwise comparison of the scores between pregraft and each time point postgraft, revealed a significant decrease in the scores at 3–9, 11, and 12 months post-transplantation (Bonferroni corrected *, p < .05). (B, C): The relative fold change (log) of the amount of spontaneous movement of each animal (B) or each group (C): the monkeys injected with culture medium only (control, blue, n = 2) or implanted with d35- (red, n = 2) or d42-spheres (green, n = 4). No statistically significant difference was observed among the three groups (repeated analysis of variance (ANOVA), F(3.03, 7.58) = 0.422, p = .744, Huynh-Feldt adjusted). (D): The relative fold changes (log) in the amount of spontaneous movement in response to L-dopa administration.

Taken together, these results indicate that the d42-spheres functioned as DA neurons in the MPTP-treated monkeys. To determine whether the host condition affected the behavioral changes, we examined the change in the spontaneous movements in response to L-dopa administration before cell transplantation. Three (Nos. 65, 69, and 75) out of six monkeys tested showed a good response, two (Nos. 66 and 68) showed a mild response, and one (No. 67) showed no response (Fig. 6D). It is worth noting that the good responders (No. 65 and especially No. 75) also showed an increase in the spontaneous movements after cell transplantation.

DISCUSSION

This preclinical study using monkey models of PD provides valuable information about the potential application of hESCs for patients with PD. With regard to safety issues, our study has uncovered three important points. First, if undifferentiated ESCs were contaminated, the cell preparation can be tumorigenic in a monkey brain. Second, the grafts were well-demarcated and showed no malignant or even no teratomatous findings; the grafts were mainly composed of immature neural cells. Third, and most importantly, a prolonged differentiation/maturation of the cells led to favorable results with respect to reduced tumorigenic growth of the grafts. Regarding the function of the grafts, hESC-derived NPCs were able to survive and function as DA neurons in MPTP-treated monkeys. In addition, FLT- and FDOPA-PET studies were useful for monitoring cell proliferation and DA function in the brain.

Between one and 6 months after implantation, the doubling time for the d14-grafts was 10.5 days. Since the doubling time of medulloblastoma, a malignant embryonal brain tumor, estimated from serial MRIs is 20–24 days [38], the initial growth of the d14-graft can be considered rapid. Between 6 and 9 months, however, their doubling time became much longer, ultimately reaching 155 days. The average doubling times of benign, atypical, and anaplastic meningiomas are reported to be 415, 178, and 205 days, respectively [39]. Considering these data, the d14-grafts behaved like atypical or anaplastic tumors, even if they showed no signs of malignancy. These grafts did not cause any behavioral impairment at the time point of 9 months, but they might cause deterioration over a period of several years. In contrast, three out of four d42-grafts almost ceased growth between 6 and 12 months (doubling time >27.4 years), and the percentages of Ki67+ cells in these grafts were less than 1%. Considering that the percentages in benign, atypical, and anaplastic meningiomas have been reported to be 0.7%, 2.1%, and 11%, respectively [40], d42-grafts behaved like benign masses.

The elimination of undifferentiated ESCs from cell preparations may be useful for preventing the formation of large tumors [41, 42]. Our study, however, shows that such elimination cannot completely prevent the formation of slow growing neural masses. The combination of surface antigens for NPC markers such as CD133, CD15, CD24, and CD29 is presented to select suitable cells for cell replacement therapy [43, 44]. d35- or d42-grafts contained no Oct3/4+ cells, while a small number of Nestin+ cells continued to proliferate even in the d42-grafts at 12 months post-transplantation. Furthermore, our preliminary study showed that positive selection of NPCs by sorting polysialylated-neural cell adhesion molecule+ (PSA-NCAM)+ cells was not sufficient to avoid the formation of neural masses in vivo, because they still exhibited continuous growth (Supporting Information Fig. S6). These results indicate that the simple elimination of undifferentiated ESCs or selection of NPCs cannot prevent the proliferation of Nestin+ cells in the graft and suggest the need for selection of more mature neuronal cells.

The evaluation of neurological scores revealed that the monkeys implanted with d42-spheres showed significant behavioral improvement during 3–12 months post-transplantation. We previously reported a short-term observation, which demonstrated an improvement by 3 months after monkey ESC transplantation [10]. Transplantation of human fetal NPCs [13] or monkey fetal DA neurons [12] also resulted in behavioral improvement of MPTP-treated monkeys at 2–4, and 6–9 months, respectively. In human double-blind clinical trials, an initial improvement was first observed by 4 or 6 months [1, 2]. Considering the maturation and neurite extension of the grafted cells, however, human patients may require a longer time period before experiencing a substantial benefit [45, 46]. Although the improvement lasted for 12 months in the present study, the degree of improvement peaked at 5 or 6 months post-transplantation. In a human clinical trial, the neurological scores worsened after withdrawal of immunosuppressants at 6 months [2]. Therefore, one possible reason for the worsening of behavior is that chronic inflammation might have affected the function of the grafts despite continuous administration of FK506. It is also possible that the TH+ cells generated by our method are heterogeneous and still contain olfactory bulb-type or diencephalic DA neurons, which thus may resulted in an insufficient reinnervation of the putamen. Recently, a floor plate-based differentiation protocol for ventral midbrain DA neurons has been reported [6]. Floor plate-derived midbrain DA neurons expressing Foxa2 well functioned without neural overgrowth in animal models. In our study, differentiated TH+ cells showed mild expression of Corin but no Foxa2. It is possible that insufficient floor plate induction might result in lack of functional synapse formation by the grafted DA neurons. Another potential reason is that the formation of functional synapses might be difficult in this xeno-transplantation model.

The monkeys implanted with d42-spheres showed behavioral improvement, while those implanted with d35-spheres did not. There might be two reasons for this difference. First, the d35-grafts were still growing at 12 months, indicating that these grafts still contain a substantial number of premature NPCs, which may interfere the function of the DA neurons. Second, the d42-grafts contained more TH+ cells (5,382.0 ± 5,535.2 cells per side; n = 8) than the d35-grafts (2,764.8 ± 897.9 cells per side; n = 4), although the difference was not statistically significant (p = .230). Furthermore, the Ki values in DOPA-PET of the d42-grafts were higher than those of the d35-grafts. Considering that there was a significant correlation between the increase in the Ki values and the degrees of the behavioral changes, the larger number of functional DA neurons in the d42-grafts can account for the better result.

Another factor that may affect the neurological improvement is the responsiveness to L-dopa. Interestingly, among the monkeys implanted with d42-spheres, the good responder, animal No.75, showed a remarkable increase in the spontaneous movements after cell transplantation. Our results may support the idea that the responsiveness to L-dopa is an important criterion for selecting patients who are likely to respond to cell replacement therapy, but a larger scale analysis will be needed to confirm whether this is the case.

CONCLUSIONS

The present study demonstrated that hESCs, if appropriately matured, can serve as a source for DA neurons without forming any tumors in a primate brain. Hopefully, the same strategy can be applied using human induced pluripotent stem cells [47]. Further studies with more MPTP-treated and control monkeys will help to clarify the effects of the cell transplantation.

Acknowledgements

We thank Dr. N. Nakatsuji (Kyoto University, Institute for Integrated Cell-Material Sciences) for providing hESCs, Dr. Y. Ono (KAN Research Institute) for an anti-Nurr1 antibody, Drs. T. Yamamoto and H. Magotani (Shin Nippon Biomedical Laboratories, Ltd.) for their help with the monkey, Astellas Pharma Inc. for FK506, and K. Kubota, M. Katsukawa, and E. Yamasaki in our laboratory for their technical assistance. This study was supported by the following grants: a Grant-in-Aid for Scientific Research from the JSPS; a grant to the Kobe Cluster, and a grant from the Project for Realization of Regenerative Medicine from the MEXT; a Health and Labor Sciences Research Grant for Research on Regenerative Medicine for Clinical Application; a grant from the Society of Catecholamine and Nervous Disease; and a grant from the Michael J. Fox Foundation for Parkinson's Research.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.

Ancillary