Self-Contained Induction of Neurons from Human Embryonic Stem Cells

Background Neurons and glial cells can be efficiently induced from mouse embryonic stem (ES) cells in a conditioned medium collected from rat primary-cultured astrocytes (P-ACM). However, the use of rodent primary cells for clinical applications may be hampered by limited supply and risk of contamination with xeno-proteins. Methodology/Principal Findings We have developed an alternative method for unimpeded production of human neurons under xeno-free conditions. Initially, neural stem cells in sphere-like clusters were induced from human ES (hES) cells after being cultured in P-ACM under free-floating conditions. The resultant neural stem cells could circumferentially proliferate under subsequent adhesive culture, and selectively differentiate into neurons or astrocytes by changing the medium to P-ACM or G5, respectively. These hES cell-derived neurons and astrocytes could procure functions similar to those of primary cells. Interestingly, a conditioned medium obtained from the hES cell-derived astrocytes (ES-ACM) could successfully be used to substitute P-ACM for induction of neurons. Neurons made by this method could survive in mice brain after xeno-transplantation. Conclusion/Significance By inducing astrocytes from hES cells in a chemically defined medium, we could produce human neurons without the use of P-ACM. This self-serving method provides an unlimited source of human neural cells and may facilitate clinical applications of hES cells for neurological diseases.


Introduction
Embryonic stem (ES) cells, derived from the inner cell mass of blastocysts, are pluripotent cells that can differentiate into a variety of cell types including neural cells [1,2]. Among the various basic and clinical applications for ES cells, cell transplantation therapy for central nervous diseases is of particular interest because differentiated neurons do not proliferate, and a relatively large number of donor cells are necessary to replace diseased neurons. Several methods have been developed to prepare neural cells from ES cells. Neurons can be obtained indirectly from ES cells via ectodermal cells in embryoid bodies, which are formed from dissociated ES cells, either by induction with retinoic acid or selection [3,4]. Alternatively, neural stem cells and neurons can be directly differentiated from ES cells without forming embryoid bodies by culturing ES cells on mousecultured stroma cells (PA-6) [5], or under chemically defined lowdensity culture conditions [6]. All of these procedures, however, are time consuming and require highly complicated processes to generate many neurons. In addition, their practicality is limited by the possible teratogenicity caused by culture factors, such as retinoic acid, of differentiated cells. We have previously reported an efficient method to prepare transplantable neural cells from mouse ES cells using a conditioned medium collected from rat primary-cultured astrocytes (P-ACM) [7][8][9]. In this study, we applied this method to human ES (hES) cells for induction of neurons and astrocytes. Once the astrocytes were derived from hES cells, they could be substituted for primary astrocytes that induce neurons, thus achieving xeno-free production of neurons.

Neural cell differentiation from hES
Four hES cell-lines stably expressing humanized renilla green fluorescent protein (hrGFP) were obtained. These hES cell-lines were kept in undifferentiated state with positive stem cell markers, such as alkaline phosphatase, Oct-4, and SSEA-4. When cultured in P-ACM containing fibroblast growth factor-2 (FGF-2) under free-floating conditions, colonies of undifferentiated hES cells gave rise to floating spheres composed of neural stem cells and undifferentiated cells, which gradually increased in size during the culture. After 12 days of culture, the spheres were plated onto a poly-L-Lysine/Laminin coated dish and cultivated in neural stem cell medium (NSCM) containing high concentrations of FGF-2 and epidermal growth factor (EGF). Within 24 h, the spheres attached onto the substrate and formed circular clusters of cells. Many of these cells subsequently migrated to the surrounding areas and covered the growth surface of the dish in circular monolayers. After replacing NSCM by P-ACM and culture for 14 days, the spheres differentiated into neurons ( Figures 1A, B) and few astrocytes (Figures 1C). These were identified by the neuronal marker tubulin b III isoform (Tuj1) and the astrocytic marker glial fibrillary acidic protein (GFAP).

Selective differentiation of hES cells into neurons and astrocytes
By culture of the spheres in NSCM, neural stem cells were able to migrate from the attached spheres to the surrounding area and subsequently form a circular cluster. NSCM containing FGF-2 and EGF promotes neural stem cells proliferation, while repressing their differentiation into any type of neural cells. After removing the core of the attached spheres mechanically, the remaining neural stem cells could proliferate in NSCM and selectively differentiate into neurons and astrocytes by subculture in an appropriate medium. To differentiate into neurons, neural stem cells were subcultured using 0.05% Trypsin/EDTA in P-ACM for 14 days (Figures 2A, B). After these 14 days of subculture, a large number of cells expressed Tuj1 (84.065.1%, n = 3). On the other hand, to differentiate into astrocytes, neural stem cells were subcultured in G5 medium for 14 days ( Figure 2C). After this subculture, a large number of cells expressed Xeno-free induction of astrocytes using a chemically defined medium For collection of xeno-free astrocytes derived from hES cells, we used a chemically defined N2 medium for neural induction. When cultured in N2 medium containing FGF-2 and EGF under freefloating conditions, colonies of undifferentiated hES cells gave rise to floating spheres. As N2 was less efficient than P-ACM for obtaining astrocytes, we prepared hES cells in large scale (.10 8 cells). After differentiation, millions of astrocytes were gained under xeno-free condition ( Figure 3A).

Neuronal induction of hES cell-derived astrocytes
Next, we investigated whether the astrocytes derived from hES cells can be substituted for primary astrocytes to differentiate hES cells into neural cells. A conditioned medium of hES cell-derived astrocytes was collected after two days culture and used as ES-ACM by adding an equal amount of N2 medium. As in the case of P-ACM, hES cells cultured in ES-ACM differentiated into neural cells via formation of spheres. After switching the cells from NSCM to ES-ACM, many cells had neuronal-like appearance with long neurites. By 6 weeks of culture in ES-ACM, most cells had neural morphology and expressed microtubule-associated protein 2 (MAP2) ( Figure 3B). During our procedure for differentiating hES cells using ES-ACM, expression of several markers was analyzed by RT-PCR ( Figure 4A). By 8 weeks culture Neural stem cells were cultured for 14 days after removal of the core of spheres with a glass pipette and change of medium to G5 medium. The proliferated cells were subcultured onto PLL/LAM coated plate and cultured for 14 days in G5 medium. Immunostaining with antibody to GFAP showed that the subcultured cells had differentiated into astrocytes. Expression of hrGFP (green) and GFAP (red) staining profiles. Bar = 100 mm. doi:10.1371/journal.pone.0006318.g002  in ES-ACM, some cells showed tyrosine hydroxylase (TH)immunoreactivity ( Figure 4B). Furthermore, when cultured in ES-ACM again, the cells could differentiate into astrocytes via formation of spheres. To induce differentiation of neural stem cells into astrocytes, the culture medium was changed from NSCM to G5 medium. After medium change, most neural stem cells had the appearance of typical astrocytes. By 2 weeks culture in G5 medium, the majority (82.461.8%, n = 3) of cells expressed GFAP, and few (,1%) expressed MAP2.

Electrophysiological analysis of neurons differentiated from hES cells
For electrophysiological study, hES-derived neurons were cultured on coverslips for 4-6 weeks. The coverslips were transferred to a recording chamber before use. Neurons were selected based on their appearance (spherical shape with long neurites). The resting membrane potential of the neurons were 262.0,211.1 mV (n = 26). Among 26 cells examined, action potentials were elicited in 22 cells ( Figure 5A, Control). Application of 1 mM tetrodotoxin (TTX) completely suppressed the overshoot ( Figure 5A, right panel). The action potentials were evoked only when resting membrane potentials were set to 270 mV by current injection in 11 cells. The rest of the cells did not possess membrane excitability (n = 4 out of 26, closed circle in Figure 5B).

Survival of hES cell-derived neurons in mice brain
To examine differentiation of hES cell-derived cells in vivo, we transplanted neural stem cells induced with ES-ACM into mice brain (n = 7). Before transplantation, the majority (85.165.1%, n = 4) of donor cells expressed Nestin, a marker for neural stem cell ( Figure 6A), and no cells expressed octamer transcription factor-3 (Oct-3) and stage-specific embryonic antigen (SSEA-4), two makers for undifferentiated cells (data not shown). Four weeks after engraftment, many (2-3610 2 ) hrGFP-positive cells were recognized ( Figure 6B, C). Some of these cells (,10%) were also Tuj1-immunoreactive ( Figure 6D). In the vicinity of the grafts, few cells were immunoreactive against Ki-67, a marker for proliferation. However, none of these Ki-67-positive cells were positive for hrGFP ( Figure 6E). Teratoma was not detected in any of the transplanted mice.

Discussion
We have shown in this study that neurons and astrocytes can be produced efficiently from hES cells using a conditioned medium collected from either rat primary-cultured astrocytes or hES cellderived astrocytes. Astrocytes derived from hES cells can be used for continuous generation of neurons. Although a number of media including serum-free media supplemented with various cytokines and/or growth factors have been developed [10,11] to keep a long-term culture of neuronal cells, synthetic culture systems can usually maintain neural cells stable for only few weeks. Although a conditioned medium of primary-cultured astrocytes can be effective in culturing neurons for a longer period of time, the use of primary astrocytes may not be practical due to a number of limitations, including restricted availability of neural tissues as source of astrocytes, and extensive time and effort to obtain astrocytes from living tissues. Additionally, it is very difficult to maintain a stable culture of primary cells in a culture vessel, and subculture of these cells is limited within few passages.
The properties of primary-cultured astrocytes vary depending on the maturation stage of the living body and the region of the living tissue from which the astrocytes are derived. In addition, when astrocytes are obtained from a living body, contamination with cells other than the desired astrocytes is inevitable. Thus, it is difficult to prepare a stable astrocyte-conditioned medium having substantially uniform quality. With our method, on the other hand, ES-ACM can efficiently induce differentiation of hES cells into neural cells. Moreover, large amounts of ES-ACM can be produced stably and readily. ES-ACM, like P-ACM, can keep neuron cultures stable for more than eight weeks until mature neuronal phenotypes are apparent. In addition, completely xenofree ES-ACM can be generated from immature hES cells by culture in chemically defined medium. With this completely xenofree ES-ACM, xeno-free neurons and astrocytes can repeatedly be produced. In our transplantation experiment, donor cells did not express undifferentiated markers, such as Oct-3 and SSEA-4. In addition, only few Ki-67 positive cells found in the vicinity of the grafts were hrGFP-negative. These cells were unlikely to be derived from donor cells. It is important to exclude tumorigenicity of neuronal cells derived with this xeno-free method in future study for application to the cell therapy. Further studies are necessary to identify the specific molecules that induce neural cells in P-and ES-ACMs.
Recently, two methods adopting human tissue-derived cells have been reported as appropriate for clinical applications. One is an improved stromal cell-induced method that uses an amniotic membrane matrix [12]. The other uses telomerase-immortalized midbrain astrocytes [13]. Although both methods are xeno-free, they still need primary human tissues. On the other hand, with our method, neural cells can be induced from ES cells themselves. This self-serving method can supply donor cells consistently and may have an advantage for clinical applications.

ES cell culture
All experiments using hES cells were performed in conformity with ''The Guidelines for Derivation and Utilization of Human Embryonic Stem Cells'' of the Japanese government after approval by the institutional review board of Mitsubishi Tanabe Pharma Corporation. Two hES cell lines, SA002 and SA181, were obtained from Cellartis AB (Goteborg, Sweden) [14] and maintained on a mitotically inactivated mouse embryonic fibroblast feeder layer in a culture medium (vitroHES, Vitrolife AB, Goteborg, Sweden), supplemented with 4 ng/ml FGF-2 (Invitrogen, Carlsbad, CA). For passaging, the hES cells were treated with collagenase type IV (200 U/ml; Invitrogen) for 5 min, gently scraped from the culture dish, and then split 1:2-1:4 onto a feeder layer of mouse embryonic fibroblasts inactivated with 10 mg/ml mitomycin C.

Electroporation
All recombinant DNA experiments conformed to National Institute of Health (NIH) guidelines. First, a pGFP plasmid, in which hrGFP (Stratagene, La Jolla, CA) was expressed under the control of a CAG promoter (a gift from J. Miyazaki) [15] was constructed. Ten micrograms of the linearized plasmid was then electroporated into a suspension of hES cells (10 7 cells) in 0.8 mL of PBS using a Gene Pulser (500 mF, 250 V, Bio-Rad, Hercules, CA). The cells were next incubated on ice for 10 minutes, plated, and allowed to recover for 24 hours before selection with G418 (200 mg/mL). The cells were daily fed with the culture medium containing G418 for 12 days, after which the resulting ten G418-

hES cell differentiation
Whole colonies of undifferentiated hES cells, 800-1000 mm in diameter, were picked up from the feeder layer using a glass capillary and transferred into non-adhesive bacteriological dishes each containing P-ACM supplemented with 20 ng/ml FGF-2 (R&D Systems Inc., Minneapolis). P-ACM was prepared as described previously [7]. The colonies were then cultured for 12 days, giving rise to spheres, which were next plated onto poly-L-Lysine/Laminin (Sigma-Aldrich, St. Louis) coated dishes and cultivated for seven days in NSCM (Neurobasal medium supplied with B27 supplement, both from Invitrogen, 20 ng/ml FGF-2, and 20 ng/ml recombinant EGF [R&D systems]). At this stage, the spheres gave rise to circular clusters of cells, many of which migrated from the clusters to the surrounding areas. After replacing the NSCM with P-ACM and culture for 14 days, the spheres differentiated into neurons and few astrocytes. To obtain more and purer neurons, the centers of the clusters containing undifferentiated ES cells were removed with a glass capillary, and the rest of the clusters were cultured for seven days in NSCM. Neuronal differentiation was then induced by subculture of neural stem cells using 0.05% Trypsin/EDTA in P-ACM for 14 days. To induce astrocytic differentiation, the neural stem cells were subcultured in G5 medium (Neurobasal medium supplemented with G5 supplement, both from Invitrogen, 10 ng/ml FGF-2, and 20 ng/ml EGF) for 14 days.
To induce astrocytic differentiation under xeno-free conditions, the colonies of hES cells were transferred into non-adhesive bacteriological dishes each containing N2 medium (Neurobasal medium supplied with N2 supplement, Invitrogen) supplemented with 20 ng/ml of FGF-2 and EGF. After the colonies were cultured for 12 days, few of them gave rise to spheres containing neural stem cells, which were subsequently plated onto poly-L-Lysine/Laminin in G5 medium. Centers of the spheres containing undifferentiated hES cells were removed with a glass capillary, and the rest of the clusters were cultured for seven days in G5 medium. By repeating over this cycle eight times, the spheres were purified to obtain pure neural stem cells. These neural stem cells were subcultured in G5 medium for 14 days to induce astrocytes. For collection of ES-ACM, hES cells derived-astrocytes were cultured in N2 medium for two days.
Cell density of neural lineages (neurons and astrocytes) was determined by counting the numbers of DAPI, Tuj + and GFAP + cells per field at a magnification of 200 times using an inverted microscope. Five visual fields were randomly selected and counted for each sample. Numbers presented in figures represent the average percentage and SEM of positive cells over DAPI from three samples per each examination.

Transplantation Experiment
Neural stem cells derived from hES cells using ES-ACM were implanted into the mouse striatum. All animal experimental protocols were approved by the Animal Ethics Committee of Mitsubishi Tanabe Pharma Corporation. 8-week-old C57BL/6 Cr Slc mice (SLC, Shizuoka, Japan) were anesthetized with pentobarbital and fixed on a stereotactic device (Narishige, Tokyo, Japan). By using a glass pipette with an inner diameter of 100 mm, 1610 5 cells/5 ml were slowly (0.3 ml/min) injected into the striatum (AP60 mm, ML +2.0 mm, DV 23.0 mm from bregma) of an adult male mouse. Four weeks after the transplantation, the recipient mouse was anesthetized with pentobarbital and perfused with ice-cold 4% paraformaldehyde in PBS. The brains of each mouse were postfixed in the same solution, cryoprotected with 30% sucrose in PBS for 48 h, and frozen. Coronal sections (thickness 40 mm) were cut on a microtome with freezing unit, collected in PBS (pH 7.4), and divided into series. Brain sections were incubated overnight with primary antibodies at 4uC. The primary antibodies used for immunohistochemistry ware mouse anti-Tuj1 (1:800, Covance, USA) and rabbit anti-Ki67 (1:25, abcam, UK). For detection of the primary antibodies, Alexa Fluor 594 goat anti-mouse IgG (1:1000; Molecular Probes) and Alexa Fluor 594 goat anti-rabbit IgG (1:1000; Molecular Probes) were incubated with the samples. Immunoreactivity was assessed and viewed under confocal laser scanning microscopy (FV10i; Olympus, Tokyo).