Optimal conditions for in vivo induction of dopaminergic neurons from embryonic stem cells through stromal cell-derived inducing activity



A method of inducing dopamine (DA) neurons from mouse embryonic stem (ES) cells by stromal cell-derived inducing activity (SDIA) was previously reported. When transplanted, SDIA-induced DA neurons integrate into the mouse striatum and remain positive for tyrosine hydroxylase (TH) expression. In the present study, to optimize the transplantation efficiency, we treated mouse ES cells with SDIA for various numbers of days (8–14 days). SDIA-treated ES cell colonies were isolated by papain treatment and then grafted into the 6-hydroxydopamine (6-OHDA)-lesioned mouse striatum. The ratio of the number of surviving TH-positive cells to the total number of grafted cells was highest when ES cells were treated with SDIA for 12 days before transplantation. This ratio revealed that grafting cell colonies was more efficient for obtaining TH-positive cells in vivo than grafting cell suspensions. When we grafted a cell suspension of 2 × 105, 2 × 104, or 2 × 103 cells into the 6-OHDA-lesioned mouse striatum, we observed only a few surviving TH-positive cells. In conclusion, inducing DA neurons from mouse ES cells by SDIA for 12 days and grafting cell colonies into mouse striatum was the most effective method for the survival of TH-positive neurons in vivo. © 2002 Wiley-Liss, Inc.

As cell biotechnology develops, cell transplantation therapy is becoming one of the promising treatments for a number of diseases and conditions, including burns, retinitis pigmentosa, and Parkinson's disease (PD). As a source for transplantation, cells derived from patients, human fetuses, and other species such as pigs have been used clinically.

Among neurological diseases, PD is considered to be a good target for cell transplantation. It is a degenerative disorder characterized by a loss of midbrain dopaminergic neurons (DA neurons), with a subsequent reduction in striatal dopamine content. For the replacement of degenerated DA neurons, ventral midbrain cells of human fetuses have been transplanted into the striatum of PD model animals, and this procedure has resulted in the recovery of function by restoration of dopaminergic neurotransmission in the striatum (Brundin et al., 1988b; Olanow et al., 1996). Furthermore, open-label and double-blind, placebo-controlled clinical trials have demonstrated beneficial effects of bilateral intrastriatal nigral grafts, such as increased [18F]DOPA uptake, improvement of Unified Parkinson's Disease Rating Scale (UPDRS) motor score, and decreased therapeutic dosage of L-DOPA (Freed et al., 2001). These clinical outcomes provide a rationale to cell transplantation therapy for PD. However, because of the limited availability of human fetal tissues and ethical problem regarding their use, other sources for donor tissues have been explored. In this context, neural stem cells and embryonic stem (ES) cells have appeared as possible candidates.

Recently, two groups reported methods for generating DA neurons from ES cells in vitro. Lee et al. (2000) reported multistep induction of DA neurons, i.e., the formation of embryoid bodies and expansion of nestin-positive cells, followed by treatment with laminin, fibroblast growth factor (FGF)-2, FGF8, and Shh-N. Kawasaki et al. (2000) reported an induction of DA neurons by stromal cell-derived inducing activity (SDIA) by culturing ES cells on PA6, a bone marrow-derived stromal cell line. This induction of DA neurons by SDIA has proved also to be applicable to primate ES cells (Kawasaki et al., 2002).

For cell transplantation, it is important to optimize the conditions of cell preparation. For this purpose, we induced DA neurons from mouse ES cells by culturing the latter on PA6 cells for different periods and then transplanting cell colonies that formed into the 6-hydroxydopamine (6-OHDA)-treated mouse striatum. In this report, we describe the optimal differentiation period on PA6 and the most suitable condition of the cells for transplantation.



Male 6–7-week-old C57BL/6CrSlc mice (SLC, Shizuoka, Japan) weighing 18–22 g were used. Animal experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and with the principles presented in the Guidelines for the Use of Animals in Neuroscience Research by the Society for Neuroscience as well as with the institutional guidelines. All surgical procedures described below were performed with animals under sodium pentobarbital anesthesia (30 mg/kg).

Lesion Surgery

The mice were given unilateral stereotaxic injections of 6-OHDA as described previously (Kaneko et al., 2000). 6-OHDA was dissolved in phosphate-buffered saline (PBS; 8 μg/μl), and 0.5 μl was injected unilaterally with a glass needle at two sites into the striatum. All injection coordinates were according to the atlas of Franklin and Paxions (1997), with the bregma used as a reference, and were as follow: A +0.5, L +2.0, V +3.0 and A +1.2, L +2.0, V +3.0.

Induction of Neural Differentiation of ES Cells

Mouse undifferentiated ES cells (G4-2) were maintained on gelatin-coated dishes in G-MEM supplemented with 1% fetal calf serum (FCS; JRH), 5% KSR (Gibco-BRL), 2 mM glutamine, 0.1 mM nonessential amino acids, 1 mM pyruvate, 0.1 mM 2-mercaptoethanol (2-ME), and 2,000 U/ml LIF (Gibco-BRL). G4-2 cells (a kind gift from Dr. Hitoshi Niwa, Osaka University) carry the blasticidin S-resistance selection marker gene driven by the Oct3/4 promoter (active under differentiated status) and the puromycin-resistance selection marker gene driven by the CAG promoter and were maintained in medium containing 20 μg/ml blasticidin S and 1.5 μg/ml puromycin to eliminate differentiated cells. DA neurons were induced from mouse ES cells by SDIA as previously reported (Kawasaki et al., 2000). Briefly, ES cells were cocultured on PA6 stromal cells in G-MEM supplemented with 5% KSR, 2 mM glutamine, 0.1 mM nonessential amino acids, 1 mM pyruvate, and 0.1 mM 2-ME (“differentiation medium”). ES cell colonies that formed on PA6 monolayers during 8, 10, 12, and 14 days of culture were used as a graft.


ES colonies on PA6 were isolated by incubation with papain for 5 min at room temperature (Papain Dissociation Kit; Worthington, Freehold, NJ). For cell suspensions, the colonies were then treated with papain for an additional 5 min and thereafter pipetted with a 1 ml tip. Cell suspensions of 2 × 105, 2 × 104, or 2 × 103 cells per 2 μl were prepared. To estimate the actual number of TH-positive cells, we replated a small part of the donor colonies on eight-well chamber slides coated with poly-L-ornithine and laminin. By 24 hr, the colonies had attached to the slides, and then they were used for immunofluorescence studies. With deep anesthesia, the mice were placed in a stereotaxic frame 14 days after the 6-OHDA injections. Each animal received an injection of 2 μl (1 μl/min) of ES cell suspension or colonies into the lesioned striatum (from the bregma: A +1.0, L +2.0, V +3.0, incisor bar 0) via a Hamilton microsyringe fitted with a 26-gauge blunt needle.

Immunofluorescence Study

Single- and double-immunofluorescence studies were carried out after permeabilization and blocking for nonspecific binding with 0.3% Triton X-100 and 10% normal donkey serum. Primary rabbit anti-TH antibody (Chemicon International, Inc., Temecula, CA) and mouse anti-β-tubulin-III antibody (Convance Research Products, Richmond, CA) were used at 4°C overnight, and donkey Cy3-labeled and Cy5-labeled secondary antibodies (Jackson Immunoresearch Laboratories, Inc., West Grove, PA) were used at room temperature for 120 min. Green fluorescence protein (GFP), contained in the ES cells, was visualized by its own fluorescence. The percentage of TH-positive cells after papain treatment was evaluated by using a Fluoview FV300 laser confocal microscope (Olympus Optical Co., Tokyo, Japan). For each group, 50 colonies were randomly selected, and thin-slice images of both GFP and TH were evaluated. Fourteen days after transplantation, animals were euthanized with pentobarbital and perfused transcardially with PBS, followed by 4% paraformaldehyde. Their brains were postfixed overnight and equilibrated in graded sucrose. Coronal 40 μm sections were cut serially on a freezing cryostat. Immunostained sections were evaluated by using a confocal microscope. Grafted ES cells were detected by their green fluorescence. The number of TH-positive cells was quantified in every third section throughout the graft and its surroundings. Only the cells with both cell bodies and neurites were counted as TH-positive neurons, and cell number of positive cells in serial sections was corrected by using the Abercrombie (1946) method.


Donor Population

To characterize the in vitro differentiation of mouse ES cells, we examined the percentage of TH-positive colonies cultured on PA6 cells for different periods. As reported previously (Kawasaki et al., 2000), the percentage increased until day 12, when it reached a plateau at 80% (Fig. 1A). Next, to estimate the survival of TH-positive colonies after colony isolation, we plated papain-treated colonies onto poly-L-ornithine/laminin-coated slides. Immunofluorescence study of these colonies revealed that the percentage of the TH-positive colonies decreased by half after isolation by the papain treatment (Fig. 1A). This demonstrates that the actual percentage of TH-positive colonies for grafting was about 40% at days 12–14.

Figure 1.

In vitro analysis of SDIA-induced formation of TH-positive neurons. A: Percentage of TH-positive colonies before and after isolation from PA6 monolayers. At day 12, the percentage reached a plateau. B: Percentage of TH-positive cells in isolated colonies. The percentage increased day by day, especially between days 12 and 14. C: Double-label immunostaining of a TH-positive colony at day 14. TH-positive neurons (red) were induced from mouse ES cells expressing GFP (green).

TH-positive cells occupied 20–30% of the ES cell-derived neurons, which represented roughly 40–50% of SDIA-treated ES cells. Thus, TH-positive cells were approximately 10% of the total cells (Kawasaki et al., 2000; data not shown). However, after papain treatment, the number of TH-positive neurons decreased significantly (Fig. 1B). The percentage of the TH-positive cells in the colonies isolated by papain treatment increased during the culture period on PA6, especially between days 12 and 14 (Fig. 1B). The actual percentage of the total grafted cells that were TH positive was about 4% at day 14. In this study, we used ES cells that expressed GFP, so that it could be used as a lineage tracer. The Double-labeling immunofluorescence study of a colony at day 14 showed the existence of TH-positive neurons extending neurites in a GFP-positive colony (Fig. 1C). TH-positive neurons tended to form clusters in colonies. In early colonies, TH-positive neurons were fewer and sparsely distributed (data not shown). These observations suggest that DA neuron progenitors proliferated and then differentiated into TH-positive neurons in colonies.


To examine the differentiation and survival of ES cell-derived TH-positive cells in vivo, we detached the colonies from PA6 monolayers by papain treatment and grafted them into the 6-OHDA-lesioned mouse striatum. Endogenous nigrostriatal DA fibers in the striatum were depleted by the injection of 6-OHDA into the ipsilateral side. Treatment with 6-OHDA was performed 14 days before grafting. We tested SDIA-treated ES cells that had been cultured on PA6 for 8 (n = 4), 10 (n = 4), 12 (n = 5), or 14 (n = 3) days. The recipient mice were sacrificed and fixed by perfusion 14 days after the transplantation, then subjected to immunofluorescence analysis.

In a low-magnification view of the double-labeled immunostained brain, evidence of depletion of TH-positive fibers in the lesioned striatum and a mass of grafted cells were observed (Fig. 2A). At higher magnification, TH-positive cells were observed in the graft (Fig. 2C). Coexpression of the marker GFP (Fig. 2B) and β-tubulin-III (Fig. 2D) indicated that these cells were neurons derived from the grafted ES cells.

Figure 2.

Integration of SDIA-induced TH-positive neurons in the mouse striatum. A: Double-label immunostaining of 6-OHDA-lesioned striatum with graft. On the lesioned side, TH-positive fibers (red) were depleted, and ES cells (green) survived after 2 weeks. B–E: Confocal microscopy images with high magnification of the grafted cells. TH-positive neurons (C; arrowheads) coexpressing GFP (B) and β-tubulin-III (D), merged in E, and showing extended neurites in the striatum.

Optimal Period

To determine the best period for culture on PA6 cells before transplantation, we counted the number of TH-positive cells in the striatum 14 days after transplantation of each group of ES cells (Table I). By the same method of preparation, the highest number of TH-positive cells in vivo was found in the group in which ES cells had been cultured for 12 days on PA6 before transplantation. Because the total number of grafted cells differed among groups, we next compared the ratio of the number of surviving TH-positive cells to the total number of grafted cells (Fig. 3A). The results were analyzed by one-way ANOVA (F = 7.033, P < 0.01). Multiple comparisons of the above-mentioned ratio between groups revealed a statistical difference between days 8 and 12 (P < 0.01) and between days 10 and 12 (P < 0.05) by the post hoc Tukey-Kramer test. These results suggest that, judged from the survival rate 2 weeks after transplantation, culturing ES cells for 12 days on PA6 was the most efficient for yielding the greatest number of TH-positive cells in the graft.

Table I. Total Number of Injected Cells and TH-Positive Neurons/Graft
 Day 8Day 10Day 12Day 14
Total number of grafted cells (×104)64.5104.527.5 ± 3.340.5
TH-positive neurons/graft44 ± 19 n = 475 ± 35 n = 4110 ± 26 n = 550 ± 17 n = 3
Figure 3.

Comparison of the different cellular conditions for transplantation. A: Determination of the optimal differentiation period. Cells cultured for 12 days on PA6 yielded the highest ratio of the number of surviving TH-positive cells to the total number of grafted cells. Analysis using StatView software (SAS Institute, Cary, NC) revealed a statistical difference between days 8 and 12 (P < 0.01) and between days 10 and 12 (P < 0.05). Values are means ± SEM. B: Comparison of cell colonies with cell suspensions as the donor condition. Grafting cell colonies yielded many more TH-positive cells in vivo than grafting cell suspensions, and the difference was statistically significant (P < 0.01). Values are means ± SEM.

Optimal Cell Conditions

In the experiment described above, we grafted cells in the form of colonies. To optimize cell conditions for transplantation, we compared the efficiency of grafting cell colonies (n = 5) with that of grafting a cell suspension of 2 × 105 (n = 5), 2 × 104 (n = 5), or 2 × 103 (n = 5) cells. In both cases, the cells were grafted after they had been cultured on PA6 for 12 days. We grafted these cells 14 days after 6-OHDA injections and fixed the mice by perfusion 14 days after the transplantation. Although the cells grafted as cell suspensions also survived in the 6-OHDA-lesioned striatum, the number of surviving GFP-positive cells was much lower than when the cells were grafted as colonies, indicating the colony condition to be more efficient for implanting TH-positive cells in vivo than the cell-suspension condition (Fig. 3B). The results were analyzed by ANOVA (F = 13.742, P < 0.01). Multiple comparisons of the survival ratio between groups revealed a statistical difference between colony group and each of the suspension groups (P < 0.01) by the post hoc Tukey-Kramer test.


In the present study, we examined the optimal conditions for inducing DA neurons from mouse ES cells by SDIA and grafting these cells into the 6-OHDA-lesioned mouse striatum. In those processes, the best period of culturing for ES cells on PA6 monolayers to yield the highest number of TH-positive cells in vivo was 12 days.

Theoretically, TH-positive neurons could have been derived from the grafted cells by two mechanisms. One is that postmitotic TH-positive neurons in the graft survived as they are. The second mechanism is that progenitors of DA neurons in the graft proliferated and differentiated in the brain. In vitro analysis in the present study revealed that the percentage of TH-positive “colonies” reached a plateau by day 12. Nevertheless, the percentage of TH-positive “cells” increased mainly between days 12 and 14. During the process in which the colonies expanded from a few cells, cells committed to the DA neuron lineage may have proliferated in the colony and become postmitotic TH-positive neurons. We infer that commitment to the DA neuron lineage was complete by day 12 but that most of these cells were not mature enough to express TH at that time. In this view, the maturation of the progenitors of DA neurons may have proceeded between days 12 and 14 in each colony.

According to previous reports on the transplantation of ventral mesencephalon from rat embryos in a PD rat model, the optimal donor age for in vivo survival was E11–15 (Björklund et al., 1983; Brundin et al., 1988a). After that period, the survival rate decreased dramatically. This timing is consistent with the time when the first DA neurons of the locus ceruleus and substantia nigra are generated in the rat (Brundin et al., 1988a). This is also the time for DA neuron progenitors in the substantia nigra to undergo their last mitosis (Brundin et al., 1988a; Björklund et al., 1983). In mice, the first DA neurons emerge in the ventral mesencephalon during E10–13 (Di Porzio et al., 1990). It has been shown that the SDIA method mimics the time course of early development of the midbrain (Kawasaki et al., 2000). With all of these facts taken into account, it is reasonable to estimate that ES cell colonies cultured on PA6 for 12 days contain not only mature DA neurons but also a considerable number of DA neuron progenitors ready for their last mitosis. The latter population (DA neuron progenitors) are likely to make good donor cells for transplantation.

The percentage of TH-positive colonies decreased by half after colony isolation. One of the reasons is the damage to the neurites. Interestingly, the decrease in the percentage of MAP2ab-positive or serotonin-positive colonies was much less (about 10%; data not shown). This may indicate that DA neurons are more sensitive to chemical or physical injury than other types of neurons. In focusing on the culture period on PA6 cells, we did not use any cytokines that have neuroprotective or neurotrophic effects. Cytokines such as GDNF (Caldwell et al., 2001; Akerud et al., 2001) or FGF20 (Ohmachi et al., 2000) are reported to have such effects. Their inclusion might help to prevent loss of DA neurons during colony isolation and transplantation.

Another concern in the present study was the cell conditions on transplantation. Grafting cells as colonies resulted in better survival of TH-positive cells in vivo than grafting cells in suspension. One possible reason is a difference in the degree of axotomy of TH-positive neurons between the two conditions. Another possibility is that DA neurons or their progenitors are, as discussed above, damaged by protease treatment. Early studies using human embryonic nigral grafts compared the outcomes of solid grafts and suspension grafts. There was little difference between the two in the survival of TH-positive neurons 4–6 weeks after transplantation into the 6-OHDA-lesioned rat striatum, but the optimal “window” for donor age was different in each case (Freeman et al., 1995; Olanow et al., 1996). Suspension grafts survived best when the donor age was between postconception days 34 and 56, whereas 44 and 65 days were optimal for solid grafts. Therefore, it is possible that more TH-positive neurons would have been observed if we grafted suspensions of cells cultured on PA6 for fewer days, corresponding to earlier embryonic days.

Although we observed no survival of TH-positive neurons from the transplantation of 2,000 suspended cells, Bj¨orklund et al. (2002) reported that transplanting a low density (1,000–2,000 cells) of undifferentiated mouse ES cells into the rat striatum resulted in a proliferation of ES cells and their differentiation into DA neurons. The authors claimed that ES cells in a high concentration often developed into cells derived from all germ layers and that those in a low concentration had decreased cell-to-cell contact and were influenced more by host signals. It seems that undifferentiated ES cells and SDIA-treated ES cells may have different requirements for their response to low-density transplantation.

In this study, 110 TH-positive neurons survived after the grafting of 2.8 × 105 cells. This ratio is lower than that in a previous study by Kawasaki et al. (2000). A possible reason for fewer TH-positive neurons in the present study might be the timing of transplantation after 6-OHDA treatment (3 days vs. 14 days).

If progenitors of DA neurons exist in the graft, as judged from the time course of normal development, a period of 14 days after transplantation seems to be long enough for them to differentiate into TH-positive neurons. However, it is possible that DA neuron progenitors proliferate and remain undifferentiated in the striatum for a longer time. In the present study, we focused on only survival and differentiation of TH-positive cells, but cell migration and neurite extension are also important for an improvement of motor disorders. Although there were no differences of cell migration and neurite extension between cell colonies and cell suspensions 14 days after transplantation in the present study (data not shown), it is possible that cell suspensions have the advantage of better migration throughout the lesioned striatum after a longer period. Furthermore, it would take a longer time for the grafted cells to form a new circuit with host neurons. In consideration of all of this, cell survival, differentiation, tumor formation, and effects on motor function should be examined for longer periods of time to accumulate data for preclinical studies. Now that ES cell lines from monkey (Suemori et al., 2001; Kawasaki et al., 2002) and human (Schuldiner et al., 2001; Zhang et al., 2001; Reubinoff et al., 2001; Xu et al., 2001) sources have been established, similar studies using these primate ES cells and monkeys as a host will be crucial before future clinical studies are performed.


We thank Hitoshi Niwa for providing the G4-2 cells. We are grateful to Hiroshi Kawasaki and Kenji Mizuseki for technical advice and discussion. This work was supported by grants from the following programs: Special Coordination Funds for Promoting Science and Technology from the MECSST; and Health Sciences Research Grants in Research on Human Genome, Tissue Engineering, and Food Biotechnology from the MHLW of Japan.