Generation of Tyrosine Hydroxylase Positive Neurons from Human Embryonic Stem Cells after Coculture with Cellular Substrates and Exposure to GDNF


  • Kimberley A. Buytaert-Hoefen,

    1. Division of Clinical Pharmacology, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado, USA
    Search for more papers by this author
  • Enrique Alvarez,

    1. Division of Clinical Pharmacology, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado, USA
    Search for more papers by this author
  • Curt R. Freed M.D.

    Corresponding author
    1. Division of Clinical Pharmacology, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado, USA
    • University of Colorado Health Sciences Center, 4200 East Ninth Avenue, C237, Denver, Colorado 80262, USA. Telephone: 303-315-8455; Fax: 303-315-3272
    Search for more papers by this author


Tyrosine hydroxylase (TH)–positive neurons were generated from human embryonic stem (hES) cells by coculturing on astrocytes or PA6 stromal cells. After 3 to 4 weeks in culture, TH-positive cells with neuronal morphology developed. Coculture with astrocytes from the embryonic striatum produced a larger number of TH-positive cells than did coculture with astrocytes from embryonic mesencephalon (329 ± 149 versus 33 ± 16 TH-positive cells per well, p < .05). In other experiments using PA6 cells as a substrate, glial-derived neurotrophic factor (GDNF) was added to the media of differentiating hES cells, and this led to a doubling of the number of TH-positive cells (PA6: 443 ± 105 TH-positive cells per well versus PA6 + GDNF: 934 ± 136, p < .05). We conclude that substrates of striatal astrocytes and PA6 cells can promote differentiation of human embryonic stem cells to a TH-positive phenotype and that GDNF can increase the number of cells expressing that phenotype.


Parkinson's disease is characterized by the loss of midbrain neurons that synthesize dopamine. Pharmacological treatment with L-3,4-dihydroxyphenylalanine (L-dopa) works initially, but over time there is reduced efficacy along with motor complications [1]. We and others have reported that transplantation of dopamine cells from the human embryonic mesencephalon, 7–8 weeks after conception, can improve motor function in people with advanced Parkinson's disease [2, 3, 4]. Limitations of this procedure include difficulty in obtaining embryonic brain tissue, as well as poor survival of the transplanted dopamine neurons. Human embryonic stem (hES) cells may provide an unlimited source of neural cells for transplantation if they can be differentiated into authentic dopamine neurons [5].

ES cells are isolated from the inner cell mass of the blastocyst [6, 7]. ES cells have the ability to remain undifferentiated and to proliferate indefinitely in vitro. When culture conditions are changed to allow for ES cell differentiation, derivatives of all three embryonic germ layers are produced [6, 8]. Transplantation of ES cells or their partially differentiated progeny, embryoid bodies, leads to the development of some dopamine neurons and teratomas [9, 10]. For hES cells to be used for transplantation into patients with Parkinson's disease, they must be differentiated into dopamine neurons with no residual ES cells. The differentiation pattern of ES cells can be influenced by factors such as stromal cell–derived inducing activity (SDIA), which has been associated with PA6 cells [11]. Coculturing PA6 cells with ES cells has resulted in the induction of neural progenitor cells and, subsequently, tyrosine hydroxylase (TH)–positive neurons in both mouse and nonhuman primate cells [11, 12]. We have sought to extend this work to human embryonic stem cells by differentiating them in cocultures with PA6 cells and astrocytes, as well as with the addition of glial-derived neurotrophic factor (GDNF).


hES cells (BG01) were generously provided by BresaGen Limited (Adelaide, Australia, and Athens, GA). Cells were maintained in an undifferentiated state on mouse fibroblast feeder layers with hES cell media (Dulbecco's Modified Eagle Medium/nutrient mixture 1:1), 15% defined fetal bovine serum (HyClone, Logan, UT), 5% knockout serum replacement (Gibco, Carlsbad, CA), 100 units/ml recombinant human leukemia inhibitory factor (Chemicon, Temecula, CA), 0.1 mM 2-mercaptoethanol, 2 mM L-glutamine, 1% penicillin-streptomycin, 0.1 mM nonessential amino acids, and 4 ng/ml basic fibroblast growth factor–human recombinant (Sigma Chemical Corp., St. Louis).

For differentiation, a finely drawn, heat-sealed sterile Pasteur pipet was used to manually dissect (divide) colonies into several smaller clumps of approximately 50 cells each. These clumps (small colonies) of cells were plated onto a variety of cell substrates. We chose to coculture hES cells on PA6 cells with or without hanging baskets of embryonic striatal tissue, on embryonic striatal and embryonic mesencephalic astrocytes, or alone on gelatin.

PA6 cells were plated at 100,000 cells per well in a gelatin-coated 12-well plate. Twenty-four hours later, hES cells were added to the PA6 cell substrate. hES cells were plated (15 colonies per well) in differentiation media (Glasgow minimum essential medium, 10% knockout serum replacement, 0.1 mM 2-mercaptoethanol, 1% sodium pyruvate, 1% penicillin-streptomycin, and 0.1 mM nonessential amino acids). In an effort to increase the number of TH-positive cells, a coarse suspension of about 75,000 human embryonic striatal cells (7–8 weeks postconception) was placed into each hanging basket (0.4 μm pore membrane, Falcon #353180) and added to the culture 24 hours after the addition of hES cells.

Human embryonic tissue was acquired from women undergoing elective abortions. Specific informed consent was obtained from each donor, according to regulations of the institutional review board of the University of Colorado (COMIRB) and state and federal laws. hES cell differentiation was carried out for 4 weeks. Cells were fixed with 4% paraformaldehyde and then processed with immunocytochemistry for TH (using a rabbit polyclonal antibody from Pel-Freez, Rogers, AR), as previously described [13]. Only TH-positive cells with definitive cell bodies were counted.

To test astrocytes as a substrate for differentiation, we isolated astrocytes from two brain regions involved in the development of dopamine neurons: the striatum and the mesencephalon. Striatal and mesencephalic astrocytes were purified from embryonic rat tissue (E14.5) by agitating cultures grown to confluency (approximately 1 week) at 250 rpm, 36.2°C for 18 hours. Other cell types were rinsed away, and adherent astrocytes were grown confluent again. For hES cell differentiation, astrocytes were plated at 100,000 cells per well in a gelatin-coated 12-well plate. hES cells were plated (15 colonies per well) on astrocytes originating from either embryonic striatum or embryonic rat ventral mesencephalon. Cells were cultured in differentiation media for 3 weeks. Cells were fixed with 4% paraformaldehyde and then processed with immunocytochemistry for TH.

To determine if GDNF could mimic the factors released from embryonic astrocytes, another experiment was performed. hES cells (15 colonies per well) were cultured on a gelatin-coated 12-well plate alone or in coculture with PA6 stromal cells with or without the addition of GDNF (10 ng/ml) in differentiation media for 3 weeks. Cells were fixed with 4% paraformaldehyde and then processed with immunocytochemistry for TH.

Reverse transcription polymerase chain reaction (RT-PCR) was performed on the third experiment. Total RNA from cocultured ES cells was isolated using the Trizol reagent, a monophasic solution of phenol and guanidine isothiocyanate, according to the manufacturer's recommended protocol (Invitrogen, Carlsbad, CA). RNA samples were reverse-transcribed by MuLV reverse transcriptase (Applied Biosystems, Foster City, CA) and random hexamers (Applied Biosystems). PCR amplifications of the cDNAs for human engrailed-1, ptx-3, TH, DAT, and the housekeeping gene β-actin were carried out. The primers that were used yielded the following products:





Primers were designed using the following sequences: NM_001426.2, NM_005029.3, NM_199292.1, and M95167.1 (National Center for Biotechnology Information nucleotide database). The primers used for amplification of human β-actin had the sequences 5′-CCTCGCCTTTGCC-GATCC-3′ and 5′-GGATCTTCATGAGGTAGTCAGTC-3′, yielding a 626-bp product [14]. Amplification of β-actin cDNA served as control of intactness of RNA and for RNA level equilibration. Furthermore, this primer pair spans an intron in order to detect any DNA contamination in the PCR reaction. The primers were purchased from Invitrogen. The PCR conditions were as follows: 94°C for 4 minutes, and then varying number of cycles of 94°C/30 seconds; 60°C for β-actin, DAT, and ptx-3 or 62°C for TH or 65°C for engrailed-1/30 seconds; and 72°C/30 seconds to reach the exponential phase of the PCR reaction. cDNA products and a 100-bp DNA ladder (New England Biolabs, Beverly, MA) to provide size standards were electrophoresed on 1% agarose gels and visualized with ethidium bromide incorporation under UV light.


In our earliest experiments with hES cells plated on PA6 cells, we observed relatively poor survival with an average of only one colony surviving out of 15 colonies plated. Nonetheless, we found that hES cells grown on PA6 cells could differentiate into TH-positive cells when they were cocultured for 4 weeks. Because the number of TH-positive cells was low under these conditions, we sought to enrich the differentiation environment by exposing the coculture to human embryonic striatal tissue. To prevent contamination of hES/PA6 cocultures with TH-positive cells from the striatum, the striatal cells were cultured in hanging baskets. This system permitted soluble factors released from the striatum to influence the hES cells. As shown in Figure 1, the combination of culture on PA6 cells plus soluble factors from human embryonic striatum led to differentiated colonies with a large TH-positive cell yield.

Figure Figure 1..

Photomicrograph of a human embryonic stem cell colony plated on PA6 cells and exposed to embryonic human striatum for 4 weeks, then immunocytochemically stained for tyrosine hydroxylase (TH). This combined treatment produced a dense population of TH-positive cells (insets). Scale bar: 100 μm.

Figure 2 presents quantitative results from cocultures of hES cells with PA6 cells and astrocytes. As shown in panel I, substrates and other treatments had a major effect on the number of TH-positive cells generated. In Experiment A, hES cells cocultured for 4 weeks on PA6 cells produced few TH-positive cells (13 ± 11 TH-positive cells per well). When cells were exposed to embryonic striatum in hanging baskets, cell number increased substantially (269 ± 89 TH-positive cells per well; p < .05, compared with growth on PA6 alone). In Experiment B, hES cells differentiated on striatal astrocytes for 3 weeks produced significantly more TH-positive cells then when hES cells were plated on mesencephalic astrocytes (329 ± 149 versus 33 ± 16 TH-positive cells per well, p < .05).

Figure Figure 2..

Effects of differentiation conditions on (I) TH cell number, (II) colony number, and (III) colony circumference are shown after three separate experimental strategies. Experiment A: hES cells cultured for 4 weeks on PA6 cells (A1) or PA6 cells plus human embryonic striatum suspended in hanging baskets (A2). Experiment B: hES cells cultured for 3 weeks on rat embryonic mesencephalic astrocytes (B1) or on rat embryonic striatal astrocytes (B2). Experiment C: hES cells cultured for 3 weeks on a gelatin-coated substrate (C1), PA6 cells (C2), or PA6 cells with 10 ng/ml GDNF (C3). Using a gelatin-coated substrate produced no TH-positive cells. GDNF treatment at least doubled the TH-positive cell yield from differentiated hES cells (F [2,15] = 22, p < .05) compared with other treatments in Experiment C. The coculture conditions of PA6 with human embryonic striatum suspended in hanging baskets for 4 weeks, striatal astrocytes for 3 weeks, and PA6 for 3 weeks produced similar yields of TH-positive cells. The difference in TH-positive cell yield between cocultures of PA6 cells for 3 and 4 weeks may be due to a downregulation of the TH phenotype. Plating hES cells for 3 weeks on PA6 cells with or without GDNF treatment increased colony size derived from differentiated hES cells (F [2,15] = 10.05, p < .05) compared with the gelatin-coated substrate treatment in Experiment C. Three weeks appears to be the optimal time to produce large colonies. After this time, cell death may produce shrinking of colony size. Plating hES cells for 3 weeks on PA6 cells with GDNF treatment increased the number of surviving colonies (F [2,15] = 3.9, p < .07) compared with other conditions in Experiment C. Abbreviations: GDNF, glial-derived neurotrophic factor; hES, human embryonic stem; TH, tyrosine hydroxylase; * and # refer to p < .05 for the indicated comparisons.

In Experiment C, we studied the effects of GDNF on the differentiation of TH-positive cells. Because GDNF is a factor produced by embryonic astrocytes and can promote survival and differentiation of dopamine neurons, we added GDNF to hES/PA6 cell cocultures and compared results with hES/PA6 alone or with hES cells on a gelatin-coated substrate after 3 weeks in culture. We found that GDNF doubled the number of TH-positive cells in cocultures with PA6 cells (PA6 + GDNF: 934 ± 136; PA6 alone: 443 ± 105, or, on a gelatin-coated substrate: 0 ± 0 TH-positive cells per well). We also noted that the increase in TH-positive cell number was associated with greater colony survival. Of the 15 colonies placed in each well, PA6 + GDNF led to survival of 13 ± 2 colonies per well, compared with 9 ± 1 colonies per well with PA6 alone or 2 ± 0 colonies per well for no cell substrate, p < .07). The size of the colonies was also increased by GDNF (PA6 + GDNF: 807 ± 109 μm circumference per well, compared with PA6 alone: 550 ± 146 μm circumference per well or on a gelatin-coated substrate: 368 ± 132 μm circumference per well, p < .05). Thus, GDNF increased the number of ES cell colonies that survived, increased the size of the colonies, and increased the number of TH-positive cells in these colonies.

As shown in Figure 3, RT-PCR was performed to examine the effects of PA6 and GDNF on the transcription of genes linked to the dopaminergic neuronal phenotype. Results showed that gene transcription during differentiation on PA6 cells was increased by exposure to GDNF. Somewhat surprisingly, there was expression of engrailed-1, ptx3, and TH in the hES cells grown on a gelatin-coated plastic substrate, despite the fact that no TH-positive cells were seen. Because no TH protein was seen in cells differentiated without substrate, the PCR results show that while mRNA was generated, there was little or no protein translation.

Figure Figure 3..

Reverse transcription polymerase chain reaction reveals the expression of several transcription factors involved in the development of dopamine neurons. The PA6 + glial-derived neurotrophic factor group displayed an increased expression of all transcription factors measured. Tyrosine hydroxylase appears to be transcribed but not translated in the gelatin-coated substrate group.


We have produced TH-positive cells from hES cells by coculturing them with PA6 cells, thereby revealing that PA6 cells can cause differentiation of hES cells just as they did mouse and nonhuman primate ES cells [11, 12]. With that result as our starting point, we found that TH-positive cell yield could be increased by the addition of soluble factors produced by human embryonic striatal tissue that was placed in coculture but separated by a semipermeable membrane. Because we felt that astrocytes might be the striatal cell type responsible for producing factors that led to greater differentiation of the hES cells into TH-positive neurons, we cocultured hES cells with rat embryonic striatal astrocytes. Interestingly, we found that culturing hES cells on a bed of embryonic striatal astrocytes produced about the same number of TH-positive cells as culturing on PA6 cells with the added presence of the hanging baskets of embryonic striatum. Therefore, we conclude that embryonic striatal astrocytes can support hES cell survival and subsequent differentiation to a neural and then TH-positive phenotype. These effects seemed to be specific for striatal astrocytes since striatal astrocytes were 30-fold better at generating TH-positive cells than were embryonic mesencephalic astrocytes.

We next went on to evaluate factors that might account for the survival and differentiation effects of striatal astrocytes. GDNF has been reported to stimulate the differentiation of mesencephalic neurons and is a potent survival factor for dopamine neurons [15, 16]. Furthermore, GDNF reduces apoptosis in dopamine neurons [13]. In the postnatal rat, GDNF is expressed at low levels in substantia nigra type 1 astrocytes, while high levels are expressed in the striatum. In early development, GDNF appears to have a role in both local and target-derived support of dopamine neurons [17]. High levels of GDNF are reported in the adult striatum with no detectable mRNA message in the substantia nigra [18], suggesting that GDNF is a target-derived factor for dopamine neurons in the adult brain.

In our experiments with PA6 cell coculture, we found that GDNF increased the number and size of differentiating hES cell colonies while approximately doubling the number of TH-positive cells. Because all elements of the cultures appeared to be enhanced, these results suggest that GDNF is a global survival factor for differentiated hES cells under these culture conditions. Furthermore, GDNF increased the expression of transcription factors involved in the development of dopamine neurons.

The mechanism by which stromal cell cocultures promote differentiation of hES cells to TH-positive neurons is uncertain. As noted in the introduction, Kawasaki et al. [11, 12] reported the generation of dopamine neurons from both murine and nonhuman primate ES cells by plating them onto PA6 cells, either live cells or cells previously fixed with paraformaldehyde, in serum-free media. The differentiation stimulus was described as SDIA and has not been further characterized. PA6 cells are a clonal preadipose stromal cell line isolated from newborn mouse calvaria [19]. Barberi et al. [20] reported that a second stromal cell line, MS-5, is also capable of inducing TH-positive cells from mouse ES cells in a similar coculture system. In their experiments, as with Kawasaki et al., transplantation of the TH-positive cells into rodent models of Parkinson's disease led to behavioral effects, indicating that the cells had dopaminergic effects in vivo. Understanding which factors are produced by the two different stromal cell lines could lead to better tools for controlled differentiation of ES cells.

An alternate strategy for differentiation of hES cells to neural phenotypes has used initial differentiation to embryoid bodies, as described by Zhang et al. [21] and Schultz et al. [22]. Both groups found neural markers in colonies after 7–10 days in serum-free differentiation medium. The time required for the optimal development of human dopamine neurons from hES cells is unclear. In the intact human embryo, dopamine neurons are fully differentiated at about day 45 and extend neurites at about day 56 [23]. Our experiment with hES cells plated on embryonic rat striatal astrocytes displayed high yields of TH-positive cells after as little as 3 weeks of differentiation. Because our in vitro conditions do not replicate the environment of the embryo, the differentiation of hES cells may be accelerated.

Future experiments will be directed at showing that TH-positive cells produced from hES cells are authentic midbrain dopamine neurons. These tests will include transplantation into a rat model of Parkinson's disease.


Embryonic striatal astrocytes and PA6 stromal cells provide efficient substrates for differentiation of human embryonic stem cells into TH-positive neurons. The astrocyte-derived factor GDNF increased the overall number of TH-positive cells derived from human embryonic stem cells.


This research was supported by the American Parkinson's Disease Association, the Program to End Parkinson's Disease, and the Mitchell Family Foundation.