One major challenge in biomedical research is to recapitulate in vitro the biological events occurring in vivo in normal or diseased organs. There remain serious concerns with the relevance of the most commonly used model systems. For instance, human brain tissue obtained from postmortem samples is subject to numerous artifacts: abnormal brain pH resulting from near death hypoxia, a lengthy postmortem period, residual amounts of medications used. Although they are a major source for primary human neuron cultures, biopsies from the CNS are restricted, owing to the invasiveness of the procedure (Deep-Soboslay et al., 2011). Thus hESC-dN are an attractive alternative to primary neuron culture.

Human embryonic stem cells are derived from the inner cell mass of the 4- to 5-day-old blastocyst. These cells possess two hallmark characteristics: (1) they are able to proliferate in vitro and (2) under controlled culture conditions they are able to differentiate into all three germ layers (ectoderm, mesoderm, endoderm), and thereby represent a potentially inexhaustible source of somatic cells (Thomson et al., 1998). Growing knowledge about differentiation protocols allows the generation of cells found in neural tissue such as neurons and glia. However, the isolation of hESC raises ethical issues due to the destruction of human embryo. The development of hIPS avoids this ethical problem and is a good alternative to hESC.

There are several approaches to generate hIPS from adult somatic cells from various tissues, including nuclear transfer, cell fusion, and direct reprogramming (Hochedlinger and Jaenisch, 2006). The direct reprogramming of differentiated cells (i.e., fibroblasts) into hIPS provides a tractable source of pluripotent cells for regenerative therapy (Figure 1). Direct reprogramming was first realized by the transduction of four transcription factors in fibroblasts (Oct-3/4, Sox2, KLF4, and c-Myc – OSKM factors, Takahashi et al., 2007; Yamanaka, 2008). Cell reprogramming is usually achieved by methods involving viral-derived vectors, but there has been progress toward optimizing security. Several alternatives exist to replace some or all of the OSKM factors: pharmacological molecules, recombinant proteins, signaling factors or use of other transcription factors (Huangfu et al., 2008; Yoshida et al., 2009; Zhou et al., 2009; Gonzalez et al., 2011). More recently, the reprogramming of human somatic cells was driven by the expression of specific miRNA (Anokye-Danso et al., 2011). For therapeutic purposes, hIPS transgene-free were designed and some “safe” non-teratoma-forming cell lines have been identified (Okita et al., 2011). Although still subject to much controversy, hIPS proliferative and differentiation properties resemble hESC (Ohi et al., 2011). Both hESC and hIPS exhibit high intrinsic variability between different cell lines (Bock et al., 2011). Thus, the suitability of each cell line for clinical applications needs to be examined.

For disease modeling purposes, hIPS lines have been generated, for example, from patients affected by spinal muscular atrophy (SMA), familial dysautonomia (FD), Rett syndrome, and down syndrome (Baek et al., 2009; Hotta et al., 2009). Motor neurons derived from SMA or FD patients hIPS exhibited, in vitro, morphological features of the disease (Ebert et al., 2009; Lee et al., 2009). Since hIPS retain a “memory” and potential characteristics of the cells or related tissue they originate from (Tian et al., 2011), it was speculated that this memory could be helpful for modeling of late-onset neurological diseases such as ALS or Parkinson’s disease (PD). Unfortunately, neurons derived from hIPS generated from ALS or PD patients do not readily recapitulate the diseases features (Dimos et al., 2008; Park et al., 2008; Soldner et al., 2009). The reprogramming of an adult cell to a pluripotent state may reset certain epigenetic hallmarks that developed during disease evolution. To avoid this problem, direct transdifferentiation of somatic cells to neural lineages could be considered. It is now possible to use direct reprogramming with human fibroblasts (with specific factors such as Ascl1, Brn2, Myt1l) to generate functional neurons (Vierbuchen et al., 2010; Kim et al., 2011; Pang et al., 2011) and more specifically, dopaminergic neurons (Pfisterer et al., 2011; Figure 1). However, these methods are inconvenient because they generate few cells; in the most recent protocols, about 20% of cells can be directly reprogrammed to functional neurons.

Withdrawing a key factor from the medium or forcing the hPSCs to grow in suspension is enough to induce cell differentiation (Thomson et al., 1998). However, the stochastic nature of differentiating hPSCs generates many different somatic cell types (Martinez et al., 2011). hPSCs-based applications, mainly in the biomedical domain, require specific in vitro differentiation toward the desirable cell population harboring a unique phenotype. Cell preparations containing undifferentiated or insufficiently differentiated hPSCs can lead to cell overgrowth or teratoma formation once transplanted in an organism (Lees et al., 2007; Aubry et al., 2008). For a given neurodegenerative disorder, hPSCs must be differentiated toward the specific neural cell type that could potentially restore the lost functions (Table 2). For example cell replacement therapy to treat PD aims dopaminergic neurons (Marchetto et al., 2010).

The crucial point is how to induce specific hPSCs differentiation toward the desired neural phenotypes. The first step is to obtain neural progenitor cells (NPCs; Figure 1). Essentially, specific differentiation depends on the addition of instructive factors and the removal, or inhibition, of preventive ones (Nat and Hovatta, 2004). To obtain NPCs, many different factors have been tested (Reubinoff et al., 2001; Dhara and Stice, 2008; Suter et al., 2009). The most commonly used are fibroblast growth factor (FGF), EGF, SHH, retinoic acid (RA), and bone morphogenetic protein-antagonists (BMPa); there is also the less well-defined stromal-cells derived inducing activity (SDIA). These factors are known to activate complex pathways such as Hedgehog, mesodermal, BMP, kinase, and WNT signaling but their roles are not entirely elucidated. To inhibit the differentiation toward lineages other than neural and promote neural differentiation, in most protocols, media supplements, such as N2 and B27, are added. N2 contain insulin, transferrin, putrescine, progesterone, and selenium. Insulin promotes proliferation, transferrin promotes proliferation and survival of mature neurons, putrescine is involved in axonal regeneration, and selenium protects against excitotoxicity. B27 contains more than 20 components including vitamins, hormone growth factors, antioxidants, and fatty acids (Suter and Krause, 2008). Ectodermal factors are also used to restrict mesoderm differentiation using P53 pathway (Sasai et al., 2008). Despite the numerous components tested and added, the effective maintenance and stable expansion of NPCs remains complicated, even with the most recently developed protocols (Li et al., 2011). Moreover, no protocol allows obtaining only NPCs; and a selection of cells of interest must be done with techniques like FACS sorting or with inducible suicide gene (Li, 2002; Kawaguchi et al., 2008).

The second step is to drive NPCs toward a specific neural phenotype (Figure 1). Many molecular pathways are involved in this step of differentiation. For example, Wnt/beta-catenin signaling is known to stimulate the formation of dopaminergic neurons (Ding et al., 2011). To get mature neural cell types, the presence of specific factors is necessary (Table 1). Yet, as for NPCs, the purity of neural cell population remains problematic (Pankratz et al., 2007). An additional consideration is that techniques for neural induction depend on the cell line used and the experimental practice (Schwartz et al., 2008; Suter and Krause, 2008; Daadi and Steinberg, 2009).

Two cell culture protocols are commonly used: suspension cultures and adherent cultures. In suspension, hPSCs form a cell mass. The most promising for 3D culture is in suspension. Adherent culture seems to provide better condition to obtain a homogenous cell population. An homogenous individual cell exposition to morphogens is not warranted due to the numerous cell layers. Thus, the concentration gradient can lead to the generation of cells at different developmental stages and subsequently the formation of multilayered structures that contain a heterogenous population of cells, including neural progenitors. The disadvantages of this protocol are: (1) the size of the cell mass varies, even with the same initial cell number and (2) there is variability in the percentage of each cell types generated and in the layer organization. In contrast, the adherent monolayer culture system allows a uniform cell exposition to morphogens and provides a more homogenous cell population. Static monolayer culture model does not mimic the in vivo microenvironment (Wilby et al., 1999) and none of the monolayer protocols used for cell differentiation yield structures and organization similar to those generated in suspension cultures or those with engineered neural tissues (ENTs).

The aim of hPSCs-derived neural tissue culture is to provide models for very early stage of nervous system development (neural tube and post neural tube early stages) and diseases, to provide models for toxicity and drug screening, and to explore the mechanism of action of different molecules. Three-dimensional cultures would allow for the study of interactions between various neural cell types and some intrinsic properties could be more readily compared with CNS physiological properties. To provide a relevant model for CNS modeling, the 3D culture system must adhere to three criteria: (1) to contain most CNS-related cell types (oligodendrocytes, neurons, astrocytes, microglia, endothelial cells, and meningeal fibroblasts); (2) to be biologically relevant (the in vitro system cell components must show similar behavior to those in vivo); (3) to recapitulate some of the developing or mature CNS features, including early neural tube organization. hPSC-derived ENTs have been produced by several laboratories with varying protocols and results (Wang et al., 2011). Amongst them, the use of air–liquid interface cell cultures device allows a 3D organization guided by endogenous developmental cues (Preynat-Seauve et al., 2009). Scaffolds with different materials like cellulose nanofibers, SiO₂, PLGA nanofibers or silicon can also been used with or without coating. Some coatings increase neural differentiation. Some frequently used coating are the laminin to support neural adhesion, the poly-l-lysine, or the alginate gel to induce slow drug release (Leach et al., 2010). All of these in vitro models recapitulate, at least partly, in vivo nervous system development.

Tissue engineering may provide advanced in vitro models for drug testing in combination with non-destructive techniques for long-term studies. Cell proliferation, migration, differentiation, and synaptogenesis could be followed in ENTs and give precious information. ENTs could reduce time, cost, and number of animals necessary for pre-clinical studies. However, the tissue thickness and variety of cell types found in the hPSC-derived culture may be challenging. It is difficult to monitor cell morphology and phenotype during cell differentiation process in ENTs. Compared to cell derived in a monolayer, hPSCs derived in 3D cultures could provide a more elaborate system for developmental neurotoxicity testing. The research aim will determine the choice between the two culture methods (Figure 2).

First described in 1817 by James Parkinson, this degenerative disorder results from the death of dopaminergic neurons in the ventral midbrain substantia nigra (Goto et al., 1990). The prevalence of Parkinson’s disease (PD) is about 1–2% of the population over 65 years (Alves et al., 2008). Symptoms are severe motor deficits like muscle rigidity, tremors, and unstable gait and posture. Current treatments consists of the administration of drug levodopa (l-dopa), a dopamine precursor able to cross the blood–brain barrier and be metabolized into dopamine (Sethi, 2010). Deep brain stimulations are also used (Tuszynski, 2007). However, these treatments only alleviate symptoms; they do not correct deficits and are progressively ineffective with PD progression. Furthermore, long-term use of l-dopa induces dyskinesia (Calabresi et al., 2010). Research for more efficient alternative treatments are currently being investigated. Transplantation of neurons from fetal ventral midbrain to replace lost dopamine neurons shows varied and sometimes no benefit for the patients in clinical trials (Freed et al., 2001; Olanow et al., 2003). Moreover, due to ethical concerns and the difficulties in obtaining adequate tissue, this alternative will likely remain marginal. There has been progress in other areas though; hPSCs derived to dopaminergic neurons and then transplanted into a rat model of Parkinson’s disease produced improvements in motor function (Ben-Hur et al., 2004; Roy et al., 2006; Chiba et al., 2008). As techniques have progressed to the point that researchers can obtain pure dopaminergic neurons from hPSCs (Cho et al., 2008; Swistowski et al., 2010; Kim, 2011). Moreover, derivation of specific dopaminergic neurons from patient IPS has been achieved and transplantation of these cells into a rodent PD model showed an alleviation of motor deficits (Cooper et al., 2010; Hargus et al., 2010). All together, these studies show that hPSCs are promising candidates for cell replacement therapy.

The most advanced hPSC-derived therapy aims to treat SCI. It is the first treatment to be evaluated in clinical trials (Geron Corporation, 2009). This trial has been halted for economic reasons by Geron enterprise but continue to be monitored. In United States, incidence of SCI is estimated to be about 12,000 cases each year (Qin et al., 2010). After a spinal cord trauma, symptoms can vary depending on the localization of the damages as well as various internal and external factors (Jagatsinh, 2009). To treat motor deficit related to SCI the connection between motor cortex and muscles must be restored. For this purpose, the transplantation of motor neurons and oligodendrocytes can be considered. These two cells types can be derived from hPSCs (Kerr et al., 2010) and hPSCs induced to motor neurons promote functional recovery after SCI in a rat model (Rossi et al., 2010). Tissue engineering approaches has been tested to treat SCI. They combine hPSCs with collagen or fibrin-based scaffold. These scaffolds are able to deliver growth factors promoting hPSCs differentiation into oligodendrocytes and neurons. (Hatami et al., 2009; Johnson et al., 2010). These studies showed that implanted cells increase locomotor functions and enhance functional recovery in a rat model of SCI (Kerr et al., 2010; Niapour et al., 2011; Lee et al., 2012).

Another promising trial is for Huntington’s disease (HD), a neurodegenerative genetic disorder that causes dementia and affects muscle coordination. Prevalence of this disease is about 0.01% of the population (Warren and Yellowlees, 1990). As for PD, some studies have investigated the potential of fetal tissue transplantation as treatment and show more encouraging results for HD treatment than for PD treatment (Frank and Biglan, 2007; Gallina et al., 2010). Another experiment involved differentiation hIPS into neural progenitors and transplanting them into a rat model of HD; grafted animals had better performance than controls (Song et al., 2007). Unfortunately, in these tests, the mechanism of recovery was not clear: was it due to factors released by the graft or by the host tissue?

Human pluripotent stem cells were also occasionally used for traumatic brain injury and Alzheimer’s disease (Molcanyi et al., 2007; Moghadam et al., 2009). Cell replacement therapy could be also investigated in some case of severe epilepsies by implantation of specific GABAergic neurons directly into affected areas. Most of these studies use mouse models and embryonic stem cells (mESCs), so much work would need to be repeated with hESC in pre-clinical testing to determine the viability of such therapies (Wang et al., 2006; Riess et al., 2007).