|Home | About | Journals | Submit | Contact Us | Français|
We describe a chemically defined protocol for efficient differentiation of human embryonic stem cells (hESCs) to neural epithelial cells and then to functional spinal motor neurons. This protocol comprises four major steps. Human ESCs are differentiated without morphogens into neuroepithelial cells that form neural tube-like rosettes in the first two weeks. The neuroepithelial cells are then specified to Olig2-expressing motoneuorn progenitors in the presence of retinoic acid (RA) and sonic hedgehog (SHH) in the following 2 weeks. These OLIG2 progenitors generate post-mitotic, HB9 expressing motoneurons at the 5th week and mature to functional motor neurons thereafter. The protein factor SHH can be replaced by a small molecule purmorphamine in the entire process, which may facilitate potential clinical applications. This protocol has been shown equally effective in generating motor neurons from human induced pluropotent stem (iPS) cells.
Directing human embryonic stem cells (hESCs) to specific lineages is prerequisite for using hESCs to model early human development and for applying the hESCs derived lineages in clinic. In the past decade various protocols have been presented to differentiate hESCs to neuroectodermal cells (1, 2) including the spinal motor neurons (3–5). These differentiation protocols vary considerably in the starting hESCs, feeder cells, unknown factors (e.g., sera and conditioned media), efficiency, and cell purity (6). We developed a series of neural differentiation protocols, including the one described here, for two objectives: modeling the early human brain development; and producing enriched/pure populations of functional neural cells for therapeutics.
The protocol was devised based on the developmental principle underlying motoneuron development. Spinal motor neurons are differentiated from neuroepithelial (NE) cells in a very narrow band of the ventral neural tube called the pMN domain, where the progenitors express the helix-loop-helix transcription factor Olig2. These Olig2-expressing progenitors are specified in the presence of a particular amount of sonic hedgehog (Shh) that is released from the notochord and subsequently the floor plate (7, 8). Through interaction of Olig2 and neurogenic transcription factors including Ngn2 and Pax6, the Olig2-expressing progenitors differentiate to post-mitotic motor neurons during the neurogenesis phase and express motoneuron-specific transcription factors such as HB9 and Isl1 while downregulating Olig2 (9–12). Thereafter, HB9-expressing motoneuorns mature and express choline acetyltransferase (ChAT), an enzyme that catalizes the synthesis of the transmitter acetylcholine for transmitting signals through the neuromuscular junctions.
Generation of spinal motoneurons from hESCs follows the same basic steps of neuroectoderm induction, motoneuron progenitor specification, differentiation and maturation of post-mitotic motoneurons (Fig. 1). hESCs are removed from the mouse embryonic fibroblast feeder to initiate differentiation. In the serum-free culture condition, these hESCs differentiate to the neuroectoderm fate in two weeks (13). During the neural induction phase, the hESC aggregates are reseeded from day 7 onto a culture surface free of feeder to form individual monolayer colonies, allowing an even exposure to morphogens and a synchronized differentiation of the neuroepithelia. By the end of the 2nd week (day 14–17), NE cells, in the readily identifiable neural tube-like rosettes (1), develop. They express a panel of neuroectoderm transcription factors including PAX6 and SOX1.
NE cells generated in this way bear an anterior phenotype by expressing OTX2 (13). Hence, it is necessary to caudalize and ventralize the NE cells in order to generate spinal motor neurons. We found that early NE cells at day 10, also referred to as primitive NE cells (13), present higher competence to respond to morphogens including RA (3, 14). Therefore, the anterior NE cells are patterned with retinoic acid (RA) and SHH for the subsequent 2 weeks. This treatment results in the induction of OLIG2-expressing ventral spinal progenitors in the 4th week. These OLIG2 cells become post-mitotic in the 5th week and express MN transcription factors like HB9 and ISL1. The MNs, when growing on substrate, extend substantial projections and express distinctive ChAT, indicating gradual maturation. When co-cultured with myoblasts, these hESC-derived MNs form characteristic neuro-muscular junctions. The 5-week in vitro differentiation process coincides with the appearance of motor neurons in the ventral horn of the developing human spinal cord at the 5–6th week.
The protocol is the modification of our previous reports (3, 14). Major modifications include streamlined procedure, simplified media, the use of more potent recombinant SHH (resulting from a mutation at the N-terminus), and application of small molecules capable of activating SHH signaling in human cells. The optimized protocol typically generates about 50% of HB9 expressing motoneruons among the total hESC progenies. This protocol has recently been tested effective for differentiating human iPS cells to spinal motor neurons.
The undifferentiated state of the starting hESCs is a prerequisite for efficient differentiation of neuroepithelial cells and subsequent functional motor neurons. Presence of partially differentiated hESCs or contamination of differentiated hESC colonies will result in unsynchronized neural differentiation and reduce the differentiation efficiency.
In the multiple-step process, we use adherent culture mode except the suspension culture steps in the initial separation of hESCs from MEF and in the purification of neuroepithelial cells. The adherent culture allows direct visualization of neural differentiation, including the neural tube-like rosettes during neuroepithelial induction and neural progenitor migration and neurite outgrowth in the neuronal differentiation phase.
In the neuroepithelial induction phase, we employ a colony culture. Almost all the colonies possess neural tube-like rosettes or at least 90% of the total differentiated cells represent neuroepithelial cells that express PAX6 and SOX1. The colony culture permits readily removal of non-neural colonies. Once non-neural colonies are scraped from the culture, 95–99% among the total population should be PAX6+ cells. This ensures subsequent neural differentiation efficiently.
Motoneuron progenitor population reaches a peak in the 4th week. If purmorphamine replaces SHH in the protocol, it increases the proportion of OLIG2-expressing cells from 50% to 60–80% of the total cells.
Differentiation of OLIG2-expressing motoneuron progenitors to HB9-expressing post-mitotic motor neurons takes another week. By the end of the 5th week, the HB9-expressing cells account for half of the total population. The HB9-expressing cells rarely migrate away from the cluster; rather, they stay in the cluster or immediate periphery of the cluster and extend extremely long processes (axons) that often travel throughout the entire 11-mm diameter coverslip. Dissociating the OLIG2-expressing progenitor spheres often results in a significant motor neuron loss, thus we use small clusters of MNs for final differentiation.
After the 5th week, the motor neurons can be further cultured for several weeks or months depending on the applications. Other mature motoneuron markers, e.g., ChAT and VaChAT, will appear over time.
The cell clusters in suspension grow big over time. When the spheres are larger than 300 um in diameter, they should get dissociated to smaller ones for continued growth and expansion. We usually split the big clusters using two simple procedures.
Alternatively, the bigger clusters are dissociated with ACCUTASE. The enzymatic effect of ACCUTASE is not as powerful as trypsin and no enzyme inhibitor is needed after dissociation. Store the ACCUTASE in aliquots of 5 ml or 10 ml at −20 °C. Thaw the frozen aliquot in a refrigerator overnight before using.
1Do not leave hESCs in dispase longer than 15 min. Longer incubation in dispasee may result in poor survival of the lifted hESCs.
2The cell density significantly affects the neural differentiation. High density significantly compromises the efficiency of neural specification.
3hESC aggregates (embryoid bodies) free of feeder cells usually float and do not attach to the culture surface. Feeder fibroblasts around the hESCs may re-form feeder, which results in EB attachment. Gently tapping the bottom of the flask will release the loosely attached EBs. Briefly pipette the EBs to remove the dead cells and feeder cells, and then transfer the EBs to a new flask.
4Do not seed the colonies in a high density. The ideal density is that after 7 days of growth attached clusters remain as individual colonies without merging to eachother. Incubate the culture at 37 °C overnight.
5Well differentiated hESC aggregates tend to attach to plastic surface after a week in suspension. Dead cells around the clusters may interfere with the attachment of the aggregates Wash the aggregates with neural differentiation medium and plate them again onto a new plate coated with laminin (20 ug/ml). Alternatively, addition of 10% FBS into the culture overnight will promote the attachment of the aggregates. The FBS should be removed right after the aggregate attachment. Presence of FBS will inhibit neural differentiation.
6We refer to these columnar epithelial cells as early neural rosettes. Surrounding the neural epithelial rosettes are flat round cells which are likely of the neural crest lineage. The neural epithelial cells at this state express Pax6 and many other neural transcription factors but not Sox1. We refer to cells at this stage as primitive neural epithelial cells. These primitive neural epithelial cells are responsive to morphogens like RA and SHH for regional patterning. Therefore, we will start the process of motoneuron specification at day 10.
7Within the clusters, multiple neural tube-like rosettes appear. Immunostaining will indicate that these cells express both PAX6 and SOX1. The cells in the form of neural tube-like rosettes attach to the substrate loosely whereas the flat cells in the surrounding attach more tightly. We refer to these cells as definitive neural epithelial cells. Thus, it takes about 2 weeks for hESCs to differentiate to neural epithelial cells. The readily identifiable rosettes formation is a valuable parameter to judge the quality of differentiation. Partially differentiated hESCs, overly damaged EBs or early RA-treated culture may result in poor rosette formation. If there are colonies that do not possess rosettes, these are usually non-neural colonies. Scrap those colonies with a pipette tip after marking them using an objective marker that is mounted in a phase contrast scope. This step will minimize, if not eliminate, the contamination of non-neural cells.
8Clusters of non-neural lineage may be present in the culture if the non-neural colonies are not scraped before lifting. Instead of forming bright round spheres, those clusters are usually grey or dark with irregular shapes. Should there be any non-neural cell contamination in culture, the partially differentiated hESCs are inevitably the source. These partially differentiated hESCs usually generate “bad colonies” which can be easily recognized by direct observation. Mark the “bad colonies” and manually remove them in the step of “rosettes” formation.
9Motor neuron progenitors represent a vulnerable population in culture. Enzymatic disaggregation of neuroepithelial spheres can damage the population thus resulting in very few motor neurons. Mild dissociation of the progenitor clusters with accutase (for 3–5 min) can facilitate monolayer formation after attachment. Plating cells at a higher density (30,000 cells/11-mm coverslip), or seeding small clusters (100–200 µm) will help cell survival. Addition of B27 and low concentration of SHH/RA in culture will also help minimize cell death.
10We’ve noticed that the HB9 antibodies from different sources vary significantly in terms of specificity. The MNR2 (HB9, monoclonal antibody, DSHB 81.5C10) is a reliable antibody for staining motoneurons from various species including human. The Chemicon Inc is releasing a new polyclonal anti-HB9 that is developed in goat to replace its previous less-specific rabbit HB9 antibody.
11When using antibodies against ChAT to label mature motor neurons, the available ChAT antibody may present strong background in cultured cells (though it stains ChAT-expressing cells in vivo very well). This is usually because of an inappropriate fixation of the enzyme. Using picric acid buffer for fixation and diluting the antibody will reduce the background. Try to use this antibody against ChAT at 1:500 dilution (goat IgG, Chemicon AB144P).