Human Embryonic Stem Cell-Derived Dopaminergic Neurons Reverse Functional Deficit in Parkinsonian Rats


  • Dali Yang,

    1. Departments of Anatomy and Neurology, School of Medicine and Public Health, Waisman Center, University of Wisconsin-Madison, Madison, Wisconsin, USA
    Search for more papers by this author
  • Zhi-Jian Zhang,

    1. Departments of Anatomy and Neurology, School of Medicine and Public Health, Waisman Center, University of Wisconsin-Madison, Madison, Wisconsin, USA
    Search for more papers by this author
  • Michael Oldenburg,

    1. Departments of Anatomy and Neurology, School of Medicine and Public Health, Waisman Center, University of Wisconsin-Madison, Madison, Wisconsin, USA
    Search for more papers by this author
  • Melvin Ayala,

    1. Departments of Anatomy and Neurology, School of Medicine and Public Health, Waisman Center, University of Wisconsin-Madison, Madison, Wisconsin, USA
    Search for more papers by this author
  • Su-Chun Zhang M.D., Ph.D.

    Corresponding author
    1. Departments of Anatomy and Neurology, School of Medicine and Public Health, Waisman Center, University of Wisconsin-Madison, Madison, Wisconsin, USA
    • Waisman Center, Room T613, University of Wisconsin, 1500 Highland Avenue, Madison, Wisconsin 53705, USA. Telephone: 608-265-2543; Fax: 608-263-5267
    Search for more papers by this author


We show that human embryonic stem cell-derived dopaminergic neurons survived transplantation to the neurotoxin 6-hydroxydopamine-lesioned rat striatum and, in combination with the cells newly differentiated from their progenitors, contributed to locomotive function recovery at 5 months. The animal behavioral improvement was correlated with the dopamine neurons present in the graft. Although the donor cells contained forebrain and midbrain dopamine neurons, the dopamine neurons present in the graft mainly exhibited a midbrain, or nigra, phenotype, suggesting the importance of midbrain dopamine neurons in functional repair. Furthermore, progenies of grafted cells were neurons and glia with greatly diminished mitotic activity by 5 months. Thus, the in vitro-produced human dopamine neurons can functionally engraft in the brain.

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


Human embryonic stem cells (hESCs), initiated from the inner cell mass of a blastocyst embryo, can be expanded continuously in a stem cell state and, in response to appropriate signals, produce expected differentiated progenies [1, 2]. These two unique characteristics make hESCs a useful tool for exploring novel medical therapies, as well as for developmental studies. In the past several years, strategies have been developed for directing these naïve stem cells to neuroectodermal cells [3, [4]–5] and more differentiated neuronal and glial subtypes [6, [7], [8], [9]–10] (reviewed in [11]). In particular, tremendous efforts have been made in developing hESC-derived cells to replace dopamine (DA) neurons that are degenerated in Parkinson disease [7, 8, 12, [13], [14], [15], [16], [17], [18], [19], [20]–21]. The earliest transplantation study involved the use of hESC-derived neural progenitors by Ben-Hur et al. [12]. They showed that the grafted cells differentiated into DA neurons in vivo, although at a low prevalence, and that the grafted cells induced partial behavioral recovery. Their data suggest a requirement of optimal in vitro dopaminergic neural differentiation prior to transplantation to achieve a complete behavior recovery. Currently, the vast majority of the in vitro DA neuronal differentiation protocols involve the use of cell-cell contact coculture of hESCs with stromal cells, such as PA6 and MS5 cells [8, 13, [14]–15, 18, 20, 21]. Most recently, Roy et al. have reported a robust protocol for differentiating hESCs to DA neurons by coculturing with telomerase-immortalized midbrain astrocytes [16]. We have developed a chemically defined culture system that also allows efficient differentiation of DA neurons from hESCs [7]. A similar approach has been reported recently [17]. Although they have various degrees of efficiency, most of the hESC-derived DA neurons can release dopamine and exhibit some electrophysiological properties of a neuron [7, 8, 14, 16, 18]. However, transplantations of these in vitro-produced human DA neurons, without genetic modification, have not yet produced any meaningful functional contributions in the 6-hydroxydopamine (6-OHDA)-lesioned rodents [13, [14]–15, 17, 21]. One exception is the recent demonstration by Roy et al., which is accompanied by graft overgrowth within 10 weeks [16]. However, teratoma formation, overgrowth, and the presence of primitive neuroepithelia are commonly observed using hESC-derived DA cell preparations even after a relatively short time (8–13 weeks) of survival [15, 16, 20, 21]. In addition, current differentiation approaches produce a mixed population of DA neurons with forebrain and midbrain phenotypes (reviewed in [11]). It remains unknown which type of stem cell-derived DA neural cells contributes to functional recovery (if any) following transplantation into the brain of Parkinson animals. This leaves open the questions of whether hESC-produced DA neural cells can functionally engraft in a safe manner and how to accomplish this.

In an effort to better understand the cellular behaviors and functional contribution of hESC-derived DA neural cells in vivo during a longer term, we transplanted in vitro-generated human DA neural cells into the 6-OHDA-lesioned rat striatum and dynamically monitored cell survival, differentiation, and proliferation and functional behavior changes 1, 4, and 20 weeks later. We demonstrated the presence of human DA neurons in the grafts and functional contribution at 5 months. The majority of dopaminergic neurons were differentiated from the grafted progenitors and exhibited some midbrain phenotype. Furthermore, grafted cells exhibited a significantly diminished mitotic activity by 5 months and chiefly differentiated into neurons and glial cells.

Materials and Methods

Culture of hESCs and Differentiation of Midbrain DA Neurons

Human ESC line H9 (NIH Registry, WA09, passages 19–28), expanded in an undifferentiated state [1], was differentiated to DA neurons according to a protocol (supplemental online Fig. 1A) modified from Yan et al. [7]. ESC colonies were differentiated to a nearly pure population of neuroepithelial cells in an adherent colony culture [3, 22]. After approximately 10 days of differentiation from ESCs, differentiating cells exhibited columnar morphology and organized into a rosette form [6, 22]. At this stage, fibroblast growth factor 8b (FGF8b; 50 ng/ml) and sonic hedgehog (SHH; 100 ng/ml) were added to the cultures. In approximately 1 week (day 17–18) the columnar cells organized into neural tube-like rosettes. The rosettes were gently blown off with a serological pipette and grown in suspension culture in the same medium for 1 week. The enriched neural progenitor clusters were dissociated with Accutase (in 0.5 mM EDTA; Innovative Cell Technologies Inc., San Diego, and differentiated on laminin-coated culture Petri dishes for 1 week. The rosette-containing clusters were detached physically using a micropipette and grown in suspension for another week before they were plated on poly-l-ornithine (15 mg/ml)- and laminin (1 μg/ml)-coated Petri dishes for transplant or glass coverslips for immunocytochemical analyses. Wnt3a (expressed in L cells; American Type Culture Collection, Manassas, VA, medium (50%) was added from day 24 to day 44 to promote the proliferation of specified DA progenitors. The neuronal differentiation medium consisted of Dulbecco's modified Eagle's medium/Ham's F-12 medium (2:1), N2 (Gibco), 0.1 mM nonessential amino acids, B27, cAMP (1.0 μM), ascorbic acid (0.2 mM), transforming growth factor type β3 (TGFβ3; 1 ng/ml), brain-derived neurotrophic factor (BDNF; 10 ng/ml), and glial-derived neurotrophic factor (GDNF; 10 ng/ml). FGF8 and SHH were withdrawn at day 44. Cells differentiated for 2 weeks (day 52) were partially digested with Accutase into small clusters for transplantation or fixed for immunocytochemical characterization (supplemental online Fig. 1).

Nigrostriatal Lesions and Cell Transplant

All animal experiments were carried out in accordance with local ethical guidelines and following approval of the University of Wisconsin-Madison animal care and use committee. Female Lewis rats (200–250 g) were anesthetized with 1%–2% isoflurane mixed in oxygen, and 10 μg of 6-OHDA (H4381; Sigma-Aldrich, St. Louis, in a 5 μl mixture (2 mg/ml in 0.9% saline with 0.2 mg/ml ascorbic acid) was stereotaxically (anterior-posterior (AP) = −4.4 mm, lateral (L) = +0.12 mm, vertical (V) = −7.8 mm) injected into the left medial forebrain bundle (nigrostriatal dopaminergic pathway) at a rate of 1 μl/minute. The animals with more than seven asymmetric rotations per minute in response to d-amphetamine challenge 3 weeks postlesion were used as recipients for cell transplant.

Differentiated DA neurons (52 days from hESCs) were partially digested with Accutase at 37°C for approximately 2–3 minutes. An aliquot was fully digested to individual cells to determine the cell number. The remaining partially digested small clusters were then diluted to approximately 50,000 cells per microliter. Four microliters of cell suspension or cell preparation medium (artificial cerebrospinal fluid containing B27, 200 mM ascorbic acid, 1 mM cAMP, 1 ng/ml TGFβ3, 20 ng/ml BDNF, and 20 ng/ml GDNF) was transplanted into the ipsilateral striatum (AP = +0.05 mm, L = +0.3 mm, V = −0.5 to approximately −0.4 mm) of anesthetized animals (4 weeks after successful lesion) using a glass pipette (0.3–0.5 mm in diameter) over a period of 5 minutes. Cyclosporine A (10 mg/kg; Novartis Pharmaceuticals, Annandale, NJ, was injected (i.p.) 24 hours before transplant and every day afterwards until the animals were sacrificed.

Behavioral Analyses

Behavioral examinations were conducted before and every month after transplantation until the animals were sacrificed.

Amphetamine-Induced Rotation Behavioral Test.

Amphetamine-induced rotation test was carried out using an automated rotometer. Each rat received amphetamine treatment (2.5 mg/kg, i.p., dissolved in 0.9% sterile saline) and was then placed in the rotometer bowl. The rotation of the rat was recorded by a computer over a 90-minute period. The number of complete (360°) turns was used in this study.

Adjusting Step Test.

Forelimb akinesia was assessed by the adjusting stepping test [23]. The hind limbs and one forepaw were held so that another forepaw was placed on a table; the rat was then passively moved sideways along the table for 0.9m within 5 seconds, first in the forehand direction and then in backhand direction. Stepping numbers over five cycles were then averaged for each forepaw. The results were expressed as a percentage of steps in lesioned side compared with the nonlesioned side.

Cylinder Test.

The cylinder test was used as a motor test of the rat's spontaneous forelimb use asymmetry [24]. The rat was placed in a transparent plastic cylinder and was videotaped until it performed 20 vertical paw placements against the cylinder wall. The data are presented as the percentage of the impaired paw use to the total contacts.

Tissue Preparation and Immunohistochemistry

Details are given in the supplemental online data.

Cell Quantification and Statistical Analyses

For quantification of different populations of cells in the graft, areas covering the graft were captured under a ×20 objective under a fluorescent scope. Cell counts of tyrosine hydroxylase (TH)+ and human nuclei (human nuclear antigen [HNA])+ cells in the graft were performed on every sixth section using the MetaMorph software (Universal Imaging, Downingtown, PA). The graft borders were outlined based on the HNA+ staining. Cell counts from serial sections were corrected and extrapolated for the total size of the graft using the Abercrombie method [25]. Total graft volume was estimated by taking into account the graft area (obtained using the MetaMorph software), the thickness of the section (25 μm), the frequency of sections selected for analysis, and the number of sections obtained through the graft.

Quantitative data are presented as mean ± SE. Comparisons of amphetamine-induced rotation between the groups and over time were performed using the unstructured covariance matrix model for analysis of variance by R, version 2.4.1 ( Independent samples (Mann-Whitney) tests were performed to evaluate the differences of DA neurons, grafted human cells, and Ki67+ cells between the 1-week grafts and 20-week grafts.


DA Neurons Survive Transplantation

We have previously shown that in our chemically defined culture, hESCs differentiate to neuroepithelial cells in 2–3 weeks and TH-expressing neurons in 5 weeks. The TH-expressing cells are DA neurons, as they do not express DβH but secrete dopamine in response to depolarization [7]. Transplantation of neural progenitor cultures (4 weeks of differentiation) often led to formation of neural tube-like rosettes and large grafts in the long term. Grafts with dissociated DA neuron progenitors (5 weeks of differentiation) often resulted in poor survival of DA neurons and variable graft sizes (unpublished observations). These phenomena are similar to recent observations made by other groups [20, 21]. Therefore, we made the following modifications to the cell preparation in the present study. First, neural progenitors, after 4 weeks of differentiation including a 2-week treatment with FGF8 and SHH, were expanded in the presence of Wnt3a-conditioned medium for 20 days. The use of Wnt3a was intended to expand the specified DA progenitors, as Wnt3a exerts a mitogenic effect on DA progenitors [26]. Second, with a micropipette, we physically selected the neural rosettes that re-formed from dissociated progenitors in an adherent culture on day 31 (supplemental online Fig. 1A). This step served as an additional purification procedure. Third, the cell clusters were only briefly treated with Accutase to loosen the cluster but not to dissociate the clusters into individual cells before transplantation. Finally, we chose more mature cells (after 7 weeks of differentiation) for transplantation.

Characterization of the differentiated cells before transplantation indicated that more than 40% of the βIII-tubulin-expressing neurons were positively stained for TH (43.4% ± 10.8% [n = 4; 10,965 cells counted]; supplemental online Fig. 1B). The TH-positive neurons expressed many of the dopaminergic markers, such as Lmx1b, a transcription factor regulating the development of monoamine neurons, including midbrain DA neurons, at an early developmental stage [27]; En1, a mid/hindbrain transcription factor [28, 29]; aldehyde dehydrogenase (ALDH2), an enzyme expressed by midbrain dopaminergic progenitors [30]; and Nurr1 and Ptx3, transcription factors expressed in midbrain DA neurons [31, 32] (supplemental online Fig. 1C–1G). Some TH+ cells were colabeled with Foxg1, a forebrain transcription factor, and no TH-expressing cells were stained for Pax2 (supplemental online Fig. 1H, 1I). Some TH+ neurons were double-stained for GABA (supplemental online Fig. 1J). Most TH+ neurons also expressed synaptophysin (supplemental online Fig. 1K). The DA neuron culture thus contained midbrain, but also some forebrain, DA neurons (supplemental online Fig. 1L). Approximately 31.5% of the total cells were positive for 5-bromo-2′-deoxyuridine (BrdU) (24-hour pulse), and 26.6% of the total cells were positive for Ki67. However, TH-positive cells were negative for Ki67 and BrdU (supplemental online Fig. 1L–1N). Thus, the cell preparation contained both postmitotic DA neurons and dividing neural precursors.

We transplanted 6-OHDA-lesioned rats with either the cell preparation medium (as a control; n = 5) or with hESC-derived DA neurons (n = 12), of which both short-term survival (1 week, n = 3; 4 weeks, n = 3) and long-term functional contribution (20 weeks, n = 6) were studied. Reassessment of cell viability at the end of the transplant procedure indicated that 60% of the cells excluded trypan blue. One rat in the 4-week survival group died at 2 weeks postgraft for reasons not related to cell implantation. All other cell-transplanted animals contained a graft in the striatum, as confirmed by labeling for human nuclear protein. A few surviving DA neurons (immunolabeled by TH and human nuclear protein) were present in the graft 24 hours after the transplantation (supplemental online Fig. 2), suggesting that DA neurons can survive the transplantation procedure. Grafts in the 1st week were similar in size among the three animals (Fig. 1A). On average, each graft contained 25,919 ± 4,756 cells. TH+ neurons in the graft, colabeled with human nuclear protein, exhibited an immature neuronal morphology (Fig. 1B). There were approximately 53 ± 6 TH+ cells per graft at this time point. Thus, there was a significant loss of cells including DA neurons or downregulation of TH following transplantation.

Figure Figure 1..

Survival and dopaminergic differentiation of transplanted human cells. The location and size of the grafts were revealed by Ho-labeled nuclei at 1 week (A) or HNA at 4 weeks (C). TH-positive dopamine neurons, double-labeled with human nuclei, were present in all the grafts of 1 (B) and 4 (D) weeks. Scale bars = 200 μm (A, C) and 25 μm (B, D). Abbreviations: HNA, human nuclear antigen; Ho, Hoechst.

Four weeks following transplantation, the graft increased in size (Fig. 1C), with 136,162 cells in one graft and 338,206 cells in another. There was an average of 361 TH-expressing neurons per graft (Fig. 1D). Hence, the grafted cells divided, and new DA neurons appeared either by differentiation from their progenitors or by re-expression of TH.

Locomotor Functional Recovery Correlates with the Presence of DA Neurons in the Long-Term Graft

Amphetamine-induced rotation tests indicated that the rats injected with cell-free medium maintained a stable rotation number, whereas rats engrafted with hESC-derived DA neurons showed a slight increase in the 1st month (not statistically significant), followed by progressively decreased rotations beginning at 8–12 weeks postgraft (Fig. 2A). By 20 weeks after transplantation, DA neuron-grafted animals exhibited significantly reduced rotation numbers compared with the rats injected with the cell preparation medium (p = .0066) (Fig. 2A). Examination of individual transplanted rats indicated that five of the six transplanted rats showed steady decreasing rotations (Fig. 2B), and these animals also exhibited initial contralateral rotation, a sign of potential functional transplantation. One of the rats did not exhibit any initial contralateral rotation, and the ipsilateral rotation increased over time (Fig. 2B), thus decreasing the average reduction of ipsilateral rotation in the cell-transplant group and skewing the progressive trend at 4 months (Fig. 2A). If this rat is excluded, there is a statistically significant difference in the rotation behavior between the transplant and control groups by 4 months. The medium-injected rats did not exhibit any sign of contralateral rotation. The stepping test and cylinder test showed no statistically significant differences between the cell-transplant group and the control group (data not shown).

Figure Figure 2..

Functional contribution of the striatal grafts. (A): Amphetamine-induced rotation analyses show the trend of unilateral rotation numbers per 90 minutes in the control and the cell-transplant groups. **, statistically significant changes in rotation numbers between the control and the cell-transplant group (p < .01). (B): The rotation behavioral changes were plotted over time for individual grafted rats alongside the TH+ dopamine (DA) neuron numbers. (C–E): Typical small (C), large (D), and elongated (E) grafts 20 w post-transplant. The TH+ cells were morphologically mature-looking DA neurons and were present in the graft center (F) or near the graft edge (G). These TH+ neurons were positive for AADC (H) but negative for DBH (I). They were also positive for VMAT (J) and DAT (K). The human nuclei-positive total grafted cells (L) and the number of TH+ DA neurons (M) increased over time. *, significant increase compared with 1-w grafts (p = .02 [L]; p = .02 [M]). Data are represented as mean ± SE. Scale bars = 200 μm (C–E) and 20 μm (F–K). (F–K): Confocal images. Abbreviations: AADC, aromatic amino acid decarboxylase; DAT, dopamine transporter; DBH, dopamine β hydroxylase; HNA, human nuclear antigen; VMAT, vesicular monoamine transporter; w, week(s).

Histological analyses revealed that the 5-month grafts had various sizes, with two rats having small grafts (2–4 mm3; Fig. 2C), two having large grafts (9–15 mm3; Fig. 2D), and two having an elongated shape (4.9 mm3; Fig. 2E). Accordingly, the number of grafted cells, identified by positive staining for human nuclear protein, varied in individual rats, with an average of 570,559 ± 263,416 donor-derived HNA+ cells per graft (Fig. 2L), significantly higher than in 1-week grafts (p = .02).

TH-expressing cells were observed in every graft. An average of 1,273 ± 340 TH+ neurons survived per graft, a significant increase compared with the TH+ neurons in the 1-week grafts (p = .02) (Fig. 2M). TH+ neurons were distributed throughout the graft, albeit not evenly (Fig. 2F, 2G). In contrast to the preferential localization to the periphery of the graft described in primate ESC transplants [16, 33], the graft center (Fig. 2F) had at least the same density of TH+ neurons as the periphery (Fig. 2G). These TH+ cells were positive for aromatic amino acid decarboxylase (Fig. 2H) but negative for dopamine β hydroxylase (Fig. 2I), indicating that these TH+ cells are DA neurons. They were also positively stained for vesicular monoamine transporter (VMAT)2. Of note, grafts also included VMAT2+ cells that were TH− (Fig. 2J). Immunostaining with three different sources of antibodies against dopamine transporter (DAT) indicated that discrete DAT-expressing neurons were present in the graft, mainly in the TH+ neurites and rarely in TH+ neuronal cell bodies (Fig. 2K). The lack of clear DAT staining in the TH+ cell body might be related to the slightly overfixed tissue, as immunostaining for DAT requires very mild fixation.

The number of TH+ DA neurons in each graft was not proportional to the graft size or total graft cell number. When plotted against the rotational behavioral changes, it showed that the rats with a reduction in asymmetrical rotation had at least several hundred DA neurons in the graft (Fig. 2B). The animal with no obvious rotation reduction had fewer (81) DA cells (Fig. 2B, green line).

Grafted Human Cells Interact with Host Tissues

Since the grafted cells contained both midbrain and forebrain DA neurons, we asked what the phenotypes of DA neurons are in the 5-month graft. Immunostaining for TH and GABA indicated that the TH+ cells in the graft were not colabeled for GABA, whereas a few GABA+ neurons were observed in the graft (Fig. 3A). Serotonin neurons, often generated alongside midbrain DA neurons [34], were rare in the graft (Fig. 3B). Additional analyses indicated that a large population of human cells (68.34% ± 30.2%) was positively stained for Foxg1 (Fig. 3C), whereas no human cells in the graft were stained for En1 (Fig. 3D). En1 was undetectable in grafts from week 1, indicating rapid downregulation following transplantation. However, most TH+ neurons were not colabeled for Foxg1 (Fig. 3C). These findings suggest that the DA neurons in the graft do not resemble those in the forebrain, particularly olfactory DA neurons. Olfactory DA neurons coexpress GABA [35]. Other transcription factors, such as Pitx3, Lmx1b, and Nurr1, were not detectable in the graft at 20 weeks after transplantation using the antibodies available. Girk2, a potassium channel that is expressed relatively specific to dopamine neurons in the substantia nigra of the midbrain [36, 37], was seen in many of the TH-expressing neurons in the graft (Fig. 3E). This suggests that the DA neurons in the graft exhibit a midbrain phenotype, despite downregulation of midbrain transcription factors En1, Pitx3, Lmx1b, and Nurr1.

Figure Figure 3..

Interactions between grafted human cells and the host tissues. (A): The TH+ cells were negative for GABA, although a few GABA+ cells were present in the graft. (B): There were some 5-HT-positive neurons in the graft. (C): Many cells in the graft were positive for Foxg1, but most of the TH+ cells were negative for Foxg1. Note that at this low magnification, TH+ cells were shown to distribute throughout the graft relatively evenly, with long TH+ fibers at the periphery. (D): TH+ cells in the graft were not labeled for En1. (E): TH-expressing neurons were Girk2-positive, although Girk2 was also expressed by a subset of non-TH-positive grafted cells. (F): Bundles of TH+ and synaptophysin+ fibers extended out of the graft. The inset shows a graft on the left side and bundles of TH+ (green) and synaptophysin+ (red) fibers extending out to the striatum at a low magnification. Blue indicates Hoechst-stained nuclei. (G): Punctate synaptophysin staining on TH+ neurons were shown by z-sections. (H): A TH+ neuron on the edge of the graft extended a process into the area occupied by DARPP32+ cells. (I): DARPP32+ cells were densely labeled by punctate human synaptophysin staining, confirmed by z-sections. Scale bar = 20 μm. Abbreviations: g, graft; HNA, human nuclear antigen.

Regardless of the presence of DA neurons in the graft 1 week and 4 weeks following transplantation, rotation behavioral changes were not observed until 3 months postgraft. This suggests that the simple presence of DA neurons alone is not sufficient for functional contribution. Immunohistochemical analyses indicated that some TH+ fibers extended to the striatum, but they were relatively short in comparison with the extensive human neural fibers that were labeled for human-specific synaptophysin (Fig. 3F). The TH+ neuronal fibers colabeled with human-specific synaptophysin, signifying that they emanated from the graft (Fig. 3F). DA neurons in the graft were intensely stained for synaptophysin on the cell membrane and on the processes in a punctate form (Fig. 3G), suggesting that the DA neurons may form synaptic connections with surrounding grafted neurons. Some TH+ neurons on the edge of the graft extended processes to the host striatal DARPP32-positive GABA neurons (Fig. 3H). These DARPP32+ cells were always negative for human nuclear protein, demonstrating that they are endogenous GABA neurons. Double immunostaining for DARPP32 and the human-specific synaptophysin indicated that human neuronal processes formed extensive contacts with the DARPP32+ cells (Fig. 3I). Thus, the grafted neurons may also form connections with the host neurons, especially the dopaminergic target, striatal DARPP32-expressing GABA neurons.

Grafted Human Neural Progenitors Largely Cease Proliferation and Become Neurons and Glia

Graft size and total cell numbers increased substantially over the 5-month period, indicating that some of the grafted progenitor cells must have proliferated. Immunostaining for Ki67, a protein expressed by proliferating cells, showed that approximately 22% of human cells in the 1-week grafts were positive for Ki67 (Fig. 4A, 4C). The proliferation rate had decreased to approximately 11% at 4 weeks and significantly (p = .025) further to approximately 1% by 5 months compared with 1-week grafts (Fig. 4B, 4C). The trend mirrors that of Ki67-expressing cells following transplantation of human fetal neural progenitors in the 6-OHDA-lesioned Parkinson rats [38].

Figure Figure 4..

Proliferation and nondopaminergic differentiation of grafted human cells. Approximately 22% of grafted cells expressed Ki67 at 1 w post-transplant (A), whereas very few of them were positive for Ki67 by 20 w (B). TH+ neurons were not colabeled with Ki67 (B). (C): Quantification of the proportion of Ki67 cells indicated a steady decrease from 1–4 and 20 w post-transplant. *, significant difference between the 20-w and the 1-w grafts (p = .025). (D): No cells in the graft were positive for Pax6 in the graft, whereas Pax6+ cells could be detected at normal adult forebrain (inset). (E–G): Nestin and Sox2 immunostaining in 1-, 4-, and 20-w grafts, respectively. (H): Endogenous expression pattern of Sox2 in the adult rat brain. (I): Many of the grafted cells were positive for nestin but they exhibited process-bearing neuronal-like morphology. A large proportion of human cells were stained for MAP2 (J), and fewer cells were positive for GFAP (K). The proportion of nestin, MAP2, and GFAP+ cells in the 20-w graft is shown in (L). Quantitative data are given as mean ± SE. Scale bar = 20 μm. Abbreviations: GFAP, glial fibrillary acidic protein; HNA, human nuclear antigen; Ho, Hoechst; LV, lateral ventricle; MAP, microtubule-associated protein; w, week(s).

Alongside the increased total cell number, TH-positive neurons in the graft also increased over the 5-month period (Fig. 2M). Double immunostaining showed that no TH-positive neurons were colabeled with Ki67 in any of the 1-, 4-, and 20-week grafts (Fig. 4B).

To identify whether there were primitive stem cells in the graft, we performed immunostaining for primitive stem cell markers such as Oct4, a transcription factor expressed by pluripotent stem cells and primary intracranial germinomas [39], and Pax6, a primitive neuroepithelial marker in humans [6]. No cells in the grafts were positively stained for Oct4 (not shown) or Pax6, although the Pax6 antibody positively stained endogenous (forebrain) cells (Fig. 4D). No cells were stained for an early epidermal marker, cytokeratin, similar to our previous observation [40].

Immunostaining for nestin and Sox2 (another neural-precursor marker) showed that nestin-positive cells and Sox2-expressing cells existed in the graft throughout the 5-month period (Fig. 4E–4G). The distribution of Sox2-positive cells was more dispersed in the 5-month graft than 1-week and 4-week grafts, similar to the pattern in the host brain tissue (Fig. 4E–4H). Characterization of the 20-week grafts indicated that many cells were positively labeled for nestin (33.4% ± 9.5%; Fig. 4I, 4L). These nestin-positive cells exhibited neuronal or glial morphology (Fig. 4I) and were sometimes double-stained for tubulin or glial fibrillary acidic protein (GFAP; not shown). The available nestin antibody often stains more mature neural cells in addition to neural progenitors [40]. Microtubule-associated protein 2+ cells were the main population of grafted cells (36.9% ± 24%; Fig. 4J, 4L). GFAP+ cells with typical astrocyte morphology were observed, both on the edge of and within the graft (16.9% ± 8.0%; Fig. 4K, 4L). Immature oligodendrocytes labeled by glutathione S-transferase-π, a marker for oligodendrocyte progenitors in vivo, were observed mostly in the corpus callosum area, where some of the grafted cells had migrated. Mature oligodendrocytes, as indicated by myelin basic protein staining, were also observed (not shown).


Our data indicate that the in vitro-produced human dopaminergic neurons contribute to partial locomotive functional recovery in a hemi-Parkinsonian rat. Although some grafted dopamine neurons survive the transplant procedure, the majority of the DA neurons in the 5-month-old graft appears to differentiate from the grafted progenitors. The majority of DA neurons in the graft exhibited a midbrain phenotype despite the mixed population of midbrain and forebrain donor DA neurons. Although the cell preparation contained large numbers of dividing cells, cells in the graft exhibited greatly diminished proliferation and differentiated to neurons and glia by 5 months. In contrast to tumor formation by hESC-derived dopaminergic cultures within 8–13 weeks reported to date, our extensive analyses showed no obvious tumor formation after 5 months of survival, the longest time so far achieved for hESC-derived dopaminergic cell transplants in the Parkinson rat model. Therefore, hESC-produced dopamine neurons can functionally engraft in the brain.

Several groups have reported that hESC-produced DA neurons are present in the graft in the 6-OHDA-lesioned rodents in the short term (2–13 weeks), sometimes in impressive numbers [12, 13, 15, 17, 20, 21]. However, functional contribution has not been reported, with a few exceptions. In a report by Roy et al., recovery of rotation behavior is observed as early as 4 weeks post-transplantation [16]. In our present study, functional recovery did not occur until approximately 3 months following transplant and began to exhibit a statistically significant difference, compared with the control transplants, by the 5th month. This is generally consistent with reports of behavioral recovery 2–3 months after transplantation with fetal human mesencephalic tissues as a donor [41, 42]. Sonntag et al. also reported that 1 of the 20 grafted animals exhibited significant reduction in amphetamine-induced rotation at 12 weeks postgraft, and this rat had more than 500 TH+ cells [20]. The delayed functional effect in our present study does not appear to be due to the lack of DA neurons. DA neurons are present in the graft shortly after transplantation (1st week), and there are significant numbers of them (>300 per graft) in the 4th week of transplantation. This exceeds the minimum numbers that are thought to be sufficient to produce locomotive behavioral changes in the 6-OHDA-lesioned rat [43, [44]–45]. This implies that a simple secretion of DA is not sufficient and that the integration of DA neurons into the striatal tissue may be necessary to achieve a stable functional recovery [46, 47]. There are signs of potential integration of hESC-derived cells into the host striatum in our present study, including the extension of TH+ fibers into the striatum and synaptic-like connections between grafted human neurons and the host DARPP32-expressing GABAergic neurons. Synaptic-like connections between grafted primate ESC-derived TH- and DARPP32-positive cells have also been reported lately [48], although the DARPP32-positive cells in our present study are always endogenous. The requirement of an extended period for human DA neurons to integrate into the host animal brain may be related to the intrinsic property of human neurons that take at least 10 weeks of hESC differentiation to exhibit functional synaptic transduction [49]. This is also consistent with previous observations that hESC-derived neural progenitors take months after transplantation into the rodent brain to become mature neurons [50], fire action potentials [51], and form synapses [40]. The relatively low degree of DA neuron fiber output to the striatum may be another factor. These factors may explain why the cylinder test and the stepping test had not begun to show obvious improvement. The initial contralateral rotation following amphetamine challenge in the cell-grafted but not the medium-injected rats suggests that the reduction in amphetamine-induced rotation is caused by functional implantation, even in the absence of improvement for the stepping test.

It has been proposed that midbrain dopamine neurons are necessary for functional engraftment [52]. In our grafts, very few DA neurons were present in the 1st week. Most of the DA neurons in the 5-month graft were presumably differentiated from their progenitors, although re-expression of TH was not excluded. Differentiation of DA neurons from grafted progenitors has been reported by Ben-Hur et al., in a study in which they transplanted hESC-derived neural progenitors instead of DA neurons [12]. The facts that the DA neurons in our grafts did not colabel with GABA and the forebrain transcription factor Foxg1, whereas GABA- and Foxg1-expressing cells were present in the grafts, indicate that the DA neurons in the grafts do not resemble forebrain DA neurons. This could be due to a selective loss of the forebrain DA neurons that normally do not form synapses with striatal GABA neurons. However, the midbrain transcription factor En1 was not detected in the TH+ cells by immunohistochemistry either. Since many DA neurons and neural progenitors carry a midbrain phenotype before transplantation, it is likely that the expression of developmentally regulated transcription factor En1 is suppressed in the striatal environment, thus making it undetectable. Developmentally, mid/hindbrain and forebrain transcription factors reciprocally antagonize each other [53]. Phenotypic instability of in vitro-generated human neurons might be another possibility. The expression of Girk2 in DA neurons in the graft supports the likelihood of a midbrain phenotype, especially the nigra phenotype, for the functionally grafted DA neurons.

Tumorigenic potential is a major concern if hESC-derived cells are going to be applied clinically, especially in light of the reports that hESC-derived dopaminergic transplants result in overgrowth or even teratoma formation within a short period (8–13 weeks) of survival [16, 20, 21]. The lack of rosette structures in the graft, the presence of scant Ki67+ cells in the 5-month-old graft, and the lack of expression of primitive stem cell markers, such as Oct4 and Pax6, indicate the absence of tumors. We believe that the lack of tumorigenic potential of our dopaminergic cultures may be related to the following factors. First, our neuroepithelial differentiation protocol yields a nearly pure population of synchronized neuroectodermal cells, with undetectable expression of mesodermal and endodermal markers [22]. This step essentially eliminates undifferentiated ESCs and cells of a non-neural lineage. Most other dopaminergic differentiation cultures to date use stromal cell coculture or suspension aggregation cultures [8, 13, [14]–15, 18, 21], which either carry over tumorigenic stromal cells or inefficiently exclude undifferentiated cells in the aggregates. Second, we differentiated the hESCs to DA neurons for a total of 7 weeks. This period is equivalent to 8 weeks of human gestation, in which most fetal mesencephalic cells are chosen for clinical transplant therapy [54, 55]. Reports to date, including our own unpublished experiments, indicate that cells differentiated for a shorter period often result in large grafts in the rodent brain. These cultures usually contain more progenitors, which explains why most TH+ DA neurons are present in the graft periphery [21, 48]. In our present transplants, DA neurons are distributed throughout the graft, similar to the distribution pattern seen with the primary mesencephalic transplant [41]. Third, our cells are largely restricted to the midbrain fate. However, large grafts were observed in two of six rats by 5 months after transplantation, indicating neural overgrowth. This may be attributed to the injection of more cells in these two animals, as our cell preparation is not completely dissociated into individual cells. It may also be attributed to the presence of neural progenitors with forebrain phenotypes in our cell preparation. Forebrain progenitors intrinsically undergo many more cell divisions than midbrain progenitors before exiting the cell cycle. We have recently shown that hESC-derived neuroepithelial cells (without FGF8 and SHH treatment and displaying forebrain phenotypes), following transplantation to postnatal immune-deficient mice, divided in the first 2–3 months before differentiating to postmitotic neurons and glia [40]. Indeed, we have observed a large population of Foxg1-expressing cells in the larger grafts. Minimizing the proportion of forebrain progenitors by more completely patterning the cells to the midbrain phenotype will likely reduce the initial cell division, which will be a critical area to be further explored.

Although our transplant work is performed mainly using the H9 cell line, we and others have previously shown that many different hESC lines exhibit a similar behavior in dopaminergic differentiation with similar phenotypes [7, 8, 12, [13], [14], [15], [16], [17]–18, 21]. The demonstration of functional engraftment of the hESC-generated dopaminergic cells for 5 months without tumor formation, similar to what has been observed using human fetal tissue transplant [41, 56], raises hopes for potential uses of stem cell-produced progenies for therapy.


We have demonstrated that hESC-derived dopaminergic neurons and their progenitors survive for 5 months in the brain of an experimental rat model of Parkinson disease and contribute to locomotor functional recovery. Although the transplanted cell populations carried both forebrain and midbrain characteristics, the dopaminergic neurons in the graft exhibited a midbrain phenotype. The hESC-derived DA cells display greatly diminished proliferative activity and do not result in obvious tumor formation in the long-term graft, suggesting their potential therapeutic application in the treatment of Parkinson disease.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank Dr. Clive Svendsen for critical reading of the manuscript. The project was supported by the National Institute of Neurological Disorders and Stroke, NIH (U01-NS046587, R01-NS045926); by The Michael J. Fox Foundation; and partly by a core grant to the Waisman Center from the National Institute of Child Health and Human Development (P30-HD03352). D.Y. and Z.-J.Z. contributed equally to this article. D.Y. is currently affiliated with the Division of Cell and Gene Therapy, Department of Neurology, University of Rochester Medical Center, Rochester, NY.