In vitro neurogenesis: development and functional implications of iPSC technology

Abstract

Neurogenesis is the developmental process regulating cell proliferation of neural stem cells, determining their differentiation into glial and neuronal cells, and orchestrating their organization into finely regulated functional networks. Can this complex process be recapitulated in vitro using induced pluripotent stem cell (iPSC) technology? Can neurodevelopmental and neurodegenerative diseases be modeled using iPSCs? What is the potential of iPSC technology in neurobiology? What are the recent advances in the field of neurological diseases? Since the applications of iPSCs in neurobiology are based on the capacity to regulate in vitro differentiation of human iPSCs into different neuronal subtypes and glial cells, and the possibility of obtaining iPSC-derived neurons and glial cells is based on and hindered by our poor understanding of human embryonic development, we reviewed current knowledge on in vitro neural differentiation from a developmental and cellular biology perspective. We highlight the importance to further advance our understanding on the mechanisms controlling in vivo neurogenesis in order to efficiently guide neurogenesis in vitro for cell modeling and therapeutical applications of iPSCs technology.

In vivo neurogenesis

The development of induced pluripotent stem cell (iPSC) technology stems from developmental biology studies. These studies unveiled some of the developmental logic and sequences of instructive interactions leading to the formation of a functional brain, a process that involves the manifestation, coordination, and integration of several signaling pathways. The aim of many scientists is, in fact, to recognize some of the developmental logic and sequences of instructive interactions leading to neurogenesis. Understanding this logic allows the physiological developmental events generating model systems to be recreated in vitro. Consequently, researchers may perform specific pharmacological treatments in order to further the current knowledge on specific pathways and pathological processes. Thanks to developmental biology, some of the factors involved in mammalian neurogenesis are known, which allows us to speculate that the administration of these factors to stem cells leads to the expression of specific transcription factors that regulate their differentiation into neurons and specific neuronal subtypes according to the concentration and mixture of factors. Recently, Lancaster et al. [1] observed that certain treatment of iPSCs contributes to their self-organization into “cerebral organoids”, which have features of a brain with concentric cell layers expressing markers of different cortical layers, a cavity reminiscent of the brain ventricle, and a structure resembling the choroids plexus. This novel result, not only opens the possibility of developing three-dimensional in vitro models of neurological diseases, but it enlightens the great potential of iPSCs and highlights the importance of fully understanding the mechanisms of in vivo neurogenesis (and in particular brain development). Since the attempts of many scientists to differentiate neuronal cells in vitro consist of unveiling and mimicking what occurs in nature, we decided to discuss the biological (molecular and morphological) events leading to brain development, before discussing in detail the opportunities and challenges of iPSC technology.

The embryonic origin of the brain is ectodermal. Early in development, the neural plate folds over on itself, giving rise to the neural tube. Subsequently, the forebrain, midbrain, and hindbrain vesicles are formed. Eventually, the neural cells start to differentiate and make synaptic contacts. Cellular differentiation in the developing brain is regulated by inducing factors, signaling molecules secreted by neighboring cells, and activated molecules following the instruction of these inducing factors.

A major step forward in understanding the developmental logic of the induction process started in 1924, when Hans Spemann and Hilde Mangold transplanted a group of cells from the dorsal lip of the blastopore, which would normally give rise to the dorsal mesoderm, into the ventral ectoderm of a host amphibian embryo. They discovered that the host had a duplicated body axis with a complete nervous system, thus demonstrating that the nervous system can be induced by non-neural cells [2]. Later experiments showed that the organizer region is a source of follistatin, noggin, and chordin, which are bone morphogenetic proteins (BMPs), a subclass of transforming growth factor (TGF-β-related protein inhibitors). Thus, the fate of neural plate cells depends on at least two signaling systems: one regulating differentiation in the medial–lateral axis and the other regulating the anterior-posterior axis. Ventral patterning is controlled by Shh, and dorsal patterning is controlled by members of the BMP family [3, 4]. Anteriorizing factors, such as Cerberus and Dickkopf, are secreted proteins that inhibit the TGF-β and/or Wnt signaling pathways [5, 6]. Specification of the rostrocaudal axis of the neural tube is also regulated by members of the fibroblast growth factor (Fgf) and Wnt families, retinoids, and activin/nodal-related TGF-β molecules (reviewed by [7, 8]). Therefore, proper brain development depends on regionalization, which is achieved by tightly controlled gene expression patterns. In conclusion, understanding the mechanisms that affect neurogenesis during development will benefit future applications of stem cell therapy, and clues to identifying the differentiation pathways leading to the generation of specific, fully functional subclasses of neurons can be unveiled by studying embryonic neural development in depth.

Embryonic versus neural stem cells and induced pluripotent stem cells

The development of iPSC technology and its potential applications have their experimental and historical bases in studies investigating the differentiating potential of embryonic stem cells (ESCs), firstly into neural stem cells (NSCs) and later into mature neurons. For this reason, we will briefly discuss the features of ESCs and NSCs.

ESCs are pluripotent cells derived from the inner cell mass of blastocyst stage embryos prior to the implantation in the uterus [9, 10]. The blastocyst includes a mass of cells (the inner cell mass or ICM) that will contribute to the developing embryo and it is opposed to an outer layer of cells, from which the fetal portion of the placenta will develop. ESCs are characterized by the ability to self-renew, the lack of contact inhibition, atypical cell cycle regulation, the capacity to form teratocarcinoma when implanted in immunocompromised mice, the ability to contribute to the germline after injection into blastocysts, and the potential to differentiate into ectoderm, mesoderm, and endoderm in vitro and in vivo. ESCs also have the ability to be propagated in culture for years while still retaining their capacity to be pluripotent. Harvesting of ESCs from the ICM of a developing embryo prior to implantation involves mechanical, enzymatic, and immunological methods. ESCs can also be harvested by laser surgery. The ICM is then cultured on feeder cells with complete media. Colonies exhibiting specific features (round boundary, homogeneous tightly packed cells with a high nuclear/cytoplasm ratio, etc.) are selected and further propagated and analyzed. ESCs can be induced to differentiate by altering their culture conditions or generating embryoid bodies by a process of aggregation. Although many modifications have been attempted on methods of differentiation, the efficiency of differentiation toward one specific phenotype is generally in the range of 10–20 % [11, 12].

During neurogenesis, ESCs give rise to neural stem cells (NSCs, or neural progenitors), found in the developing embryonic brain, but also in restricted areas of the adult brain, such as the subventricular zone (SVZ) of the lateral ventricles and the subgranular zone (SGZ) in the hippocampus [13]. NSCs are multipotent, as opposed to the pluripotency of embryonic stem cells, as they can differentiate into neurons and glial cells in response to specific cues [14–16]. The potential of the adult brain to generate new neurons and glial cells has led to studies of the potential application of NSCs in the treatment of neurodegenerative diseases and brain injury. One problem that translational medicine needs to overcome is the limited availability of NSCs, as they cannot be isolated from living adult humans and the use of embryonic NSCs is ethically questionable. One possibility would be to use NSCs from fetuses that are born dead, but their availability may not be sufficient for the therapeutic need.

The work by Uchida at el. [17] is seminal in the field of stem cell neurobiology. They isolated clonogenic human NSCs by combining the use of antibodies against cell surface markers and fluorescence-activated cell sorting. The authors sorted CD133+ CD34-CD45– cells and demonstrated that these cells can generate floating cellular aggregates (neurospheres) in vitro, which can further differentiate into neurons and glial cells [17–20] (Fig. 1b, c). In order to evaluate the potential of in vivo engraftment, migration, and differentiation of CD133+ neurospheres, the cells were injected into the lateral ventricles of neonatal immunodeficient SCID mice. A cluster of human cells was found in the SVZ that continued to proliferate 7 months after transplantation. Uchida et al. [17] affirmed that tumorigenesis was not observed 1 year after transplantation.

Fig. 1.

Maintenance and differentiation of stem cells. a Photos of iPSCs cultured on a feeder layer of murine embryonic fibroblasts (MEF), Matrigel, or in suspension as embryoid bodies. b Bright field images depicting neural stem cells grown in suspension as a neurosphere. c Photograph of a mixed population of neuronal cells. The navy blue arrowheads indicate neural rosettes from which neurites (including axons and dendrites) and neuronal somata mature (lavender arrowhead). The red arrowhead indicates astrocytes. Phase-contrast photographs were taken with a Motic AE21 microscope (Motic Instruments Inc., Canada belonging to OPBG) connected to a Moticam 2300 digital camera using the software Motic Images Plus 2.0. Scale bar 15 μm

Importantly, a revolutionary method to obtain stem cells from adult somatic cells has been developed by Takahashi and Yamanaka [21]. They demonstrated that the retroviral-mediated introduction of four transcription factors (Pou5f1, Klf4, c-Myc, Sox2) into mouse fibroblasts converts these cells into cells that closely resemble pluripotent embryonic stem cells, termed induced pluripotent stem cells (iPSCs). The iPSC technology allowed scientists to speculate about the possibility of performing autologous transplants as opposed to allogenic cellular techniques that require immunosuppressive combined therapies. Notably, the use of iPSCs bypasses the ethical concerns related to embryonic stem cells, as iPSCs have pluripotent potential but come from easily accessible adult tissues.

In addition to fibroblasts, other cell types have been used recently for iPSC derivation, including keratinocytes [22], pancreatic b cells [23], neural cells [24], mature B and T cells [25], melanocytes [26], hepatocytes [27], amniotic cells [28, 29], dental pulp stem cells [30], hair follicles [31], and cells derived from adipose tissue [32, 33]. iPSCs derived from different somatic cells may carry different intrinsic potential to differentiate into specific cell lineages. For example, Tat et al. [34] showed increased expression of the mesodermal and endodermal markers, Brachyury and Foxa2, respectively, in iPSCs derived from adipose cells when compared to fibroblasts-derived iPSCs [34]. This difference in gene expression could compromise the efficiency or the timing of differentiation and makes further studies necessary to fully comprehend the differences among iPSCs derived from different somatic cells. Thus far, fibroblasts have been preferentially used to generate iPSCs from patients suffering from neurological diseases.

Retroviral and lentiviral vectors have been widely used for the delivery of reprogramming factors, but virus-free methods have been explored, such as episomes [35], RNA and protein transfection [36], small molecule carriers (SMoCs) [37], and cell-penetrating peptides (CPPs) [24, 38].

The opportunity offered by iPSC systems in neuronal differentiation

iPSC research is still in its infancy and many laboratories have generated iPSCs from cells of patients with diseases (as listed in [39]), but disease phenotypes have rarely been shown. However, the great potential of iPSC technology has been demonstrated by studies showing that patient-derived iPSCs differentiate into specific cell types, and some cellular deficits have been shown, serving as proof-of-principle for disease modeling [40]. Moving beyond the proof-of-principle stage is a major challenge but necessary to reveal new features of disease biology and new targets for therapeutic intervention. Moreover, iPSCs offer the ability to model monogenic diseases that can manifest cellular phenotypes, but it can also be used for phenotype-based drug screening in complex diseases for which the causative genetic mechanism is still unknown.

Robust and efficient differentiation towards selected cell phenotypes is, without question, the most daunting obstacle to studying diseases. Specific functional cell types have been generated from iPSCs thanks to the identification of developmentally relevant signals in animal models [41]. Being pluripotent, iPSCs can give rise to neuronal lineage cells (neurons, astrocytes, and oligodendrocytes). In order to obtain neural progenitor cells, and subsequently specific neuronal subtypes, the cells are exposed to regulatory factors involved in cell fate choices. During iPSC differentiation, the specific neuronal subtypes required for a particular cell replacement therapy can be generated by altering cell culture conditions. Even though the factors controlling neuronal differentiation are known, they are not easily controlled in vitro. Differentiation is a multistep process that often produces heterogeneous cell populations (i.e., consisting of cells at different stages of maturation) and/or overtly mixed cell populations, presenting difficulties to obtain neurons at the same differentiation stage in the culture dish. Differentiation is a broad concept that includes early, intermediate, and terminal differentiation (Fig. 2) [42].

Fig. 2.

In vitro neurogenesis. a iPSCs during in vitro differentiation into motor neurons or b cerebellar and c cortical telencephalic neurons at different phases (early, intermediate, and terminal). Phase-contrast photographs were taken with a Motic AE21 microscope (Motic Instruments Inc., Canada belonging to OPBG) connected to a Moticam 2300 digital camera using the software Motic Images Plus 2.0. Scale bar 15 μm

Existing methods for the differentiation of mouse ESCs or iPSCs are not always applicable to human cells, and the differentiation of human iPSCs can take much longer due to the time required for human gestation and development. Another limit of in vitro differentiation is the requirement of costly recombinant growth factors and media. The efforts of many scientists are directed towards faithfully achieving directed differentiation of disease-relevant cell types on a large scale.

The potential to differentiate specific neuronal subtypes

Multiple, different protocols exist for achieving neural induction and the differentiation of iPSCs [43–58], which complicates the generation of tightly controlled and standardized applications of iPSC technology (see Table 1). For example, the starting point for growing pluripotent cells can differ depending on the growth substrate, such as Matrigel versus mouse embryonic fibroblasts (MEFs), and the formulation of the cell media (Fig. 1a). The method of neural induction also varies widely between laboratories, and can include embryoid bodies (EBs), adherent monolayers, rosette formation, stromal feeder layers, or EZ spheres (as explained in [59]). With so much variation, the replication of results can be very difficult. Therefore, any phenotype assessed in models that use iPSCs needs to be robust enough for replication by diverse methods of differentiation.

Table 1.

Methods for in vitro differentiation of neural subtype and glial cells

(A) Methods for differentiating stem cells (SC) into cortical telencephalic glutamatergic neurons

Reference: Bibel et al. (2007) [43]

Farra et al. (2012) [44]

Starting cells: Mouse embryonic stem cells

Media: DMEM

Coating Technique: Poly-ornithine, laminin

Factors and supplements: Retinoic acid, LIF, insulin, transferrin, progesterone, putrescine, sodium selenite, BSA, l-alanine, biotin, l-carnitine, ethanolamine, d-galactose, l-proline, putrescine, Na-pyruvate, Na-selenite, vitamin B12, zinc sulfate, catalase, glutathione, superoxide dismutase, linoleic acid, linolenic acid, progesterone, all-trans retinol, retinylacetate, tocopherol, tocopherolacetate, b-mercaptoethanol, FCS

Obtained cells: Glutamatergic neurons

Efficiency: 90–95 % of cells positive for VGLUT1, bIII-tub, MAP2, and Tau

Duration: ≈15 days

Notes: The neurons differentiate synchronously

Reference: Li et al. 2009 [45]

Starting cells: Human embryonic stem cells

Media: DMEM/F12, Neurobasal

Coating technique: Poly-ornithine, laminin

Factors and supplements: N2, Heparin, SHH, DKK1, WNT3A, BDNF, GDNF, IGF1, B27

Obtained cells: Telencephalic neurons

Efficiency: 79 % of cells positive for bIII-tub, Tbr; 57 % of the neurons expressed VGLUT1

Duration: ≈20 days

Reference: Zeng et al. (2010) [46]

Starting cells: Human iPS

Media: DMEM/F12, Neurobasal medium

Coating technique: Poly-ornithine, laminin

Factors and supplements: Heparin, N2, B27, BDNF, GDNF, IGF1

Obtained cells: Forebrain neural cells

Efficiency: >60 % of TBR1-positive cells

Duration: ≈24 days

Reference: Reyes et al. (2008) [47]

Starting cells: Mouse ES cells

Media: DMEM/F12, Neurobasal medium

Coating technique: Gelatin

Factors and supplements: B27, N2, sodium pyruvate, Dox, BDNF, GDNF, knock-out serum replacement

Obtained cells: Neurog1 transgene expressing cells

Efficiency: 75 % of VGLUT-positive cells

Duration: 5 days

Reference: Espuny-Camacho et al. (2013) [48]

Starting cells: Mouse ES cells, Human ES cells, human iPSC

Media: DMEM/F12, Neurobasal medium

Coating technique: Matrigel, poly-d-lysine/laminin

Factors and supplements: B27, N2, BSA, sodium pyruvate, 2-mercaptoethanol, Noggin, Y27632

Obtained cells: cortical cells

Efficiency: 60 % of VGLUT-positive cells

Duration: 44 days

(B) Methods for differentiating SC into cerebellar glutamatergic neurons

Reference: Erceg et al. (2012) [49]

Starting cells: Human iPSC

Media: DMEM/F12, BME

Coating technique: Laminin, fibronectin

Factors and supplements: FGF2, heparin, N2, Glutamax, FGF8, retinoic acid, ITS, FGF4, WNT1, WNT3A, B27, BMP7, BMP6, GDF7, SHH, NT3, JAG1

Obtained cells: Granule cerebellar neurons

Efficiency: 85 % of the total cells were b-III-tubulin-positive neurons, and ≈75 % of total cells were positive for the cerebellar cell markers ATH1+, ZIC2+, and ZIC3+, which represents the highest efficiency ever achieved

Duration: ≈35 days

Reference: Salero and Hatten (2007) [50]

Starting cells: Murine ES cells

Media: DMEM, DFK5 medium, BME

Coating technique: Fibronectin, laminin

Factors and supplements: FBS, 2-mercaptoethanol, LIF, FGF8B, retinoic acid, ITS, FGF4, bFGF, WNT1, WNT3A, N2, B27, BMP7, GDF7, BMP6, SHH, JAG1, NT3, BDNF

Obtained cells: Cerebellar granule cells

Efficiency: Not specified

Duration: ≈30 days

(C) Methods for differentiating SC into dopaminergic neurons

Reference: Yan et al. (2005) [51]

Starting cells: Human embryonic stem cells

Media: DMEM/F12

Coating technique: Poly-ornithine, laminin

Factors and supplements: Heparin, N2, serum replacer, cAMP, ascorbic acid, BDNF, GDNF, SHH, FGF8

Obtained cells: dopaminergic neurons

Efficiency: ≈30 %

Duration: 30 days

Reference: Emborg et al. (2013) [52]

Starting cells: iPSC from Macaca mulatta monkey

Media: DMEM/F12

Coating technique: Poly-ornithine, laminin

Factors and supplements: SHH, FGF8, ascorbic acid, BDNF, GDNF, TGF-b, cAMP

Obtained cells: Dopaminergic neurons

Efficiency: 16 % of 37 %

Duration: 40 days

(D) Methods for differentiating SC into serotoninergic neurons

Reference: Barberi et al. (2003) [53]

Starting cells: Murine embryonic stem cells

Media: DMEM

Coating technique: Poly-ornithine, laminin

Factors and supplements: Serum replacement, FGF4, SHH, bFGF, N2, FGF8, ascorbic acid, BDNF

Obtained cells: Serotonin neurons

Efficiency: ≈55 %

Duration: ≈14 days

Reference: Shimada et al. (2012) [54]

Starting cells: Murine embryonic stem cells and iPSCs

Media: DMEM

Coating technique: Matrigel

Factors and supplements: Knockout serum replacement, FGF4, FGF8, SHH, Noggin, Chordin

Obtained cells: Serotonin neurons

Efficiency: ≈80 %

Duration: ≈14 days

(E) Methods for differentiating SC into GABAergic neurons.

Reference: Barberi et al. (2003) [53]

Starting cells: Murine embryonic stem cells

Media: DMEM

Coating technique: Poly-ornithine, laminin

Factors and supplements: Serum replacement, SHH, bFGF, N2, FGF8, NT4, BDNF

Obtained cells: GABAergic neurons

Efficiency: ≈70 %

Duration: ≈14 days

Reference: Addae et al. (2012) [55]

Starting cells: Murine embryonic stem cells

Media: Neurobasal medium

Coating technique: Hydrogel, poly-d-lysine, laminin

Factors and supplements: N2, B27, retinoic acid

Obtained cells: GABAergic interneurons

Efficiency: ≈87 %

Duration: ≈14 days

(F) Methods for differentiating SC into motor neurons

Reference: Corti et al. (2012) ([56, see also [63] for a review of motor neuron differentiation])

Starting cells: Human iPSC (19.9)

Media: DMEM/F12

Coating technique: Matrigel

Factors and supplements: N2, heparin, retinoic acid, SHH, BDNF, GDNF, IGF1

Obtained cells: SMI32-positive cells

Efficiency: ≈80 %

Duration: ≈30 days

(G) Methods for differentiating SC into astrocytes

Reference: Barberi et al. (2003) [53]

Starting cells: Murine embryonic stem cells

Media: DMEM

Coating technique: Poly-ornithine, laminin

Factors and supplements: Serum replacement, bFGF, EGF, N2, CNTF, PDGF

Obtained cells: GFAP-positive cells

Efficiency: ≈92 %

Duration: ≈18 days

Reference: Gupta et al. (2012) [57]

Starting cells: Human embryonic stem cells

Media: DMEM

Coating technique: Matrigel

Factors and supplements: B27, BMP2, BMP4, LIF

Obtained cells: GFAP-positive cells

Efficiency: ≈96 %

Duration: ≈55 days

(H) Methods for differentiating SC into oligodendrocytes.

Reference: Hu et al. (2009) [58]

Starting cells: Human embryonic stem cells

Media: DMEM/F12, MEM

Coating technique: Poly-ornithine, laminin

Factors and supplements: N2, N1, cAMP, biotin, heparin, retinoic acid, SHH, Purmorphamine, FGF2, B27, PDGF, IGF, NT3

Obtained cells: Oligodendrocytes

Efficiency: ≈99 %

Duration: ≈3 months

Despite the apparent difficulties in obtaining neurons in vitro, neural specification may be acquired as a default mechanism [60]. Interestingly, Tropepe et al. [60] proposed a default model for neural identity in which the neural state is achieved autonomously after the removal of inhibitory signals. This model is supported by the fact that Noggin and Chordin, which are isolated from mesendodermal tissues and inhibit the activity of ‘epidermal-promoting’ BMP signaling, are sufficient for the induction of a second neural axis in Xenopus. The results obtained by Tropepe et al. [60] suggested that, in isolation at relatively low densities, embryonic stem cells have the autonomous tendency to differentiate into neural cells. Thus, it is paradoxically easier to obtain neurons in a dish than in vivo, where cell density and neural inhibition is maximal and differentiation of the neural stem cell lineage depends on the suppression of neural inhibition. These experiments not only proposed the fascinating ‘neural default model’, but also highlighted the importance of cell density for neural differentiation in vitro.

To achieve efficient neuronal differentiation of pluripotent cells, many protocols exploit multistep strategies that require either co-culture, sphere, or EB formation and/or the addition of specific growth factors (Fig. 1a, c). However, Fico et al. [61] described a simple one-step method for differentiating ES cells into neurons and glial cells as a mixed cellular population, which does not require intermediate cell dissociation and replating steps, but simply consists of a medium supplemented with serum-free Knockout Serum Replacement (KSR) devoid of any other specific neural inducer. Notably, this protocol is in line with the ‘neural default mechanism’ suggested by Tropepe et al. [60]. An additional method for single-step differentiation of iPSCs into neurons was developed by the forced expression of a single transcription factor (Ngn2 or NeuroD1) with lentiviral vectors [62]. The usage of viral vectors by Zhang et al. [62] limits the possibilities of using this method for therapeutic purposes, but it offers a tool to obtain neuronal cultures in 2–3 weeks and with nearly 100 % efficiency. This method is suited to perform drug screenings and deeper analyses of human neurological diseases.

The advantage of obtaining a mixed neuronal population is the ability to perform morphological analysis on different neuronal subtypes in the same experiment from the same starting cell population, but some study designs require the analysis of one particular neuronal subtype. For example, if cerebellar ataxia is the disease of interest, increasing the percentage of cerebellar neurons in the culture is necessary. Many studies concentrate on the differentiation of specific neuronal subtypes and, in these cases, specific growth factors are needed. In particular, differentiation protocols that generate a broad range of neuronal subtypes, but also selectively generate glutamatergic, dopaminergic, serotonergic, or GABAergic neurons, astroglia, or oligodendroglia have been developed (see Table 1).

The use of iPSCs for the treatment of neurological disorders requires the ability of iPSCs to differentiate into the appropriate neuronal subtypes that should be replaced or repaired by the therapy. We report examples of some of the protocols and results for subtype neural specification in Table 1 and Fig. 2. Advanced knowledge of neural development and the understanding of dorso-ventral and medial–lateral patterning influences has provided a good opportunity for generating neural cells from iPSCs, and neurons from different parts of the neural tube have been generated successfully. These studies have revealed that the process of in vitro neurogenesis, much like neural induction during embryonic development, is regulated by the coordinated actions of the BMP, Wnt, and FGF signaling pathways. Neural induction and specification of stem cell-derived neural progenitors follows the same order of signals in vivo, and the proper timing of exposure to these factors can give rise to well-defined neuronal populations.

Trans-differentiation

An alternative to iPSC-derived neurons comes from Vierbuchen et al. [64], who demonstrated that neurons can be generated from murine fibroblasts by directed reprogramming without an intermediary ESC-like state via viral infection of five transcription factors: Ascl1, Brn2, Myt1l, Zic1, and Olig2. Using patch-clamp recordings, the authors demonstrated that the generated neurons had functional membrane properties and were able to assume mature neuronal morphologies and functional synapses. Further investigations by the authors showed that Ascl1 was sufficient to induce neuronal conversion, but the other transcription factors were important for facilitating the conversion and maturation. The neurons obtained in this manner were excitatory and expressed markers of cortical identity. A small proportion of the induced neurons (iN) expressed markers of GABAergic neurons, but no other neurotransmitter phenotypes were detected. Importantly, the generation of iN cells was fast, efficient, and devoid of tumorigenic pluripotent stem cells [64]. Therefore, iN cells could provide a novel and potent system for studying cellular identity and plasticity, neurological disease modeling, drug discovery, and regenerative medicine. However, a better understanding of the mechanisms controlling the reprogramming of somatic cells, such as trans-differentiation, is necessary, and the production of specific neuronal subtypes is also required for application in cell therapy.

The direct conversion of fibroblasts has also been tested in humans by direct conversion into multi-lineage blood progenitors and neuronal cells [65, 66]. Trans-differentiation has the advantage of bypassing the time-consuming and potentially mutagenic iPSC reprogramming process, but has the disadvantage that it does not generate a self-renewing, stable progenitor population.

Proof-of-principle for disease modeling with iPSCs

The development of animal models for neurological diseases is difficult and often does not fully recapitulate the human phenotype. Animal studies often cannot unravel the complexities of the human brain. For example, Down’s syndrome is caused by trisomy of chromosome 21, but mice do not have a chromosome 21. Therefore, the generation of an animal model depends on targeting homologous regions in different chromosomes. Another example of the difficulties found in recapitulating human disease is spinal muscular atrophy (SMA). In SMA patients, the survival motor neuron (SMN) protein concentration is decreased in all tissues, but only motor neurons undergo degeneration. Because the affected cell type (motor neurons) cannot be harvested without harm to the patients, the in vitro model uses patient fibroblasts, which also exhibit the decrease in SMN protein (as in all other tissues) and are used to screen drugs for the ability to increase SMN expression. However, these cells are not affected by degeneration, and phenotypic rescue cannot be evaluated. Thus, the pathology is not fully represented and the effects in human beings might not be represented accurately.

The generation of iPSC disease models is recommended because the iPSCs carry the mutations associated with the disease, and they can differentiate into many types of neurons and neuronal support cells found in the brain and spinal cord.

Although potentially possible, modeling diseases that are genetically complex [such as sporadic Alzheimer’s disease or Parkinson’s disease (PD)] is a difficult challenge that requires the identification of cellular phenotypes that correlate with a known aspect of disease pathology, which is feasible in iPSC systems. Disease phenotypes have been demonstrated in several monogenic diseases, including SMA, fragile X syndrome, Rett’s syndrome, and PD [67–69].

Human neurological diseases, including stroke, neurodegenerative disorders, multiple sclerosis, neuro-trauma, and neurodevelopmental disorders, may be caused by a loss of neurons and/or glial cells in the brain or spinal cord. Although the adult brain contains NSCs in restricted areas and acute neurological insults stimulate a basal rate of neural progenitor/precursor proliferation and differentiation, they do not contribute significantly to functional recovery. We can speculate that, because the environmental factors to repair the insult are present, the injury can be repaired by increasing the pool of neuroprogenitors. Thus, studies on iPSC-derived neuroprogenitors are setting the basis for cell replacement strategies in a broad spectrum of human neurological diseases. Many candidate drugs with promising results in animal models fail in clinical trials. Therefore, the use of patient-specific iPSCs has the significant advantage of capturing the unique genetic background of the patient, the affected cell type, and the developmental time, as well as the possibility of performing toxicology tests on several cell types (i.e. hepatocytes and cardiomyocytes) [69–71].

Thus, patient-specific iPSC-based modeling of neurogenetic and neurodegenerative diseases could become an efficient and widely used tool for in vitro disease modeling, drug screening, and gene therapy. In fact, genetic correction of patient-specific iPSCs may be required before transplantation.

Neurodevelopmental disorders

Down’s syndrome

Down’s syndrome is the most common chromosomal abnormality among live born infants with an incidence of up to 1 in 700 births [72]. Because the murine genome is not endowed with chromosome 21, a mouse model that faithfully recapitulates the disease is difficult to obtain. Thus, human iPSC models of Down’s syndrome might be very useful. Human NSCs have been used to develop models of Down’s syndrome [73, 74]. These studies demonstrate that in vitro neurogenesis decreases over time, probably due to the fact that trisomy 21 alters chromatin formation [75], which is essential in cell differentiation and tissue development [74]. Thus, the difficulty with this model is that although cortical development can be obtained, NSC senescence strongly limits the possibility of performing further studies. The development of Down’s syndrome iPSC models may allow the continuous availability of cells for several differentiation studies. iPSCs are currently obtained from Down’s syndrome patients [76], but their differentiation potential and neuronal phenotypes have not yet been investigated (reviewed in [77]).

Fragile X-associated tremor/ataxia syndrome

Fragile X-associated tremor/ataxia syndrome is a neurodegenerative disorder that affects adult males who are carriers of pre-mutation expansions (55–200 CGG repeats) in the 5′-untranslated region (UTR) of the Fragile X Mental Retardation-1 (FMR1) gene [78–80]. The phenotype of FXTA patients consists of progressive intention tremor and gait ataxia, which is sometimes associated with parkinsonism, dysautonomia, anxiety, peripheral neuropathy, and cognitive decline [81]. Fragile-X syndrome belongs to the autism spectrum of disorders and is the most common cause of inherited mental retardation, with a prevalence of 1/3600 [82]. The pathology is caused by silencing of the FMR1 gene due to abnormal CGG repeat expansions in the 5′-UTR region of the gene (leading to transcriptional silencing), which is developmentally dependent [83]. In undifferentiated human FXTA embryonic stem cells, FMR1 is expressed, and gene silencing occurs only upon differentiation [83]. Urbach et al. [68] demonstrated that in FXTA-iPSC lines from three different patients, the gene is transcriptionally silent in both the pluripotent and differentiated states [68]. Thus, FXTA-iPSCs do not recapitulate the developmentally dependent silencing of FMR1, however, they represent a unique genetic model system that harbors a completely silent FMR1 locus for studying disease phenotypes and offer an excellent tool for drug screening. Thus, this system has been used by the same scientists to investigate the ability of several chromatin remodeling drugs, such as 5-azacytidine, to reactivate the expression of FMR1 [84]. Notably, the treatment results in the reactivation of FMR1 expression, which persists even after drug withdrawal. This system shows how an apparent scientific failure can lead to an unexpected potential therapeutic success.

Rett’s syndrome

Rett’s syndrome, a disorder in the autism spectrum, is mostly caused by mutations in methyl-CpG-binding protein 2 (MECP2), a protein involved in DNA methylation that regulates an array of different genes [85]. Mecp2 mutant mice show the importance of the protein in neuronal maturation [86]. MECP2 is an X-linked gene, and the authors who obtained iPSCs from Rett’s syndrome patients examined the ability of their clones to reset the X chromosome and whether X-inactivation occurs again after neural differentiation in vivo. Importantly, the authors showed that iPSCs have two active X chromosomes and that X inactivation occurs upon neural differentiation, producing mosaic neuronal cultures with different ratios of cells expressing normal MECP2 levels, mimicking what is observed in the patients’ brains. iPSCs derived from healthy controls and patients with Rett’s syndrome were differentiated into glutamatergic and GABAergic neurons and their neurogenetic potential assessed, as well as synapse number and their neuronal morphology. Although no changes in neurogenesis were observed, a substantial reduction in synapse number and the number of spines was reported [69]. The functionality of the obtained neurons was disrupted using electrophysiology and calcium imaging. This study represents a proof-of-principle for modeling monogenic diseases involving X-linked genes and reflects the capacity to derive functionally active neurons from control and patient-derived neurons, carefully analyzing the morphological and electrophysiological properties of control and Rett’s syndrome patient-derived neurons [69].

Neurodegenerative disorders

Spinal muscular atrophy

With an incidence of 1 in 10,000 live births and a carrier frequency of 1 in 50, SMA is an autosomal recessive disease characterized by progressive neurodegeneration of the α-spinal motor neurons, which leads to muscle weakness and death [87]. SMA is caused by loss of function mutations in the SMN1 gene, which encodes the ubiquitously expressed SMN protein. Affected and healthy individuals also carry a nearly identical gene obtained during evolution by gene duplication, named SMN2. The etiology of SMA is still unclear, as SMN is part of a multiprotein complex involved in the assembly of spliceosomal small nuclear ribonucleoprotein complexes. Why reduced levels of this ubiquitously expressed protein specifically cause motoneuron degeneration is still a matter of debate. Human SMN2 has the same open reading sequence as SMN1, but it differs from SMN1 by a C to T nucleotide transition in exon 7, which causes the production of two alternative isoforms: one devoid of exon 7 that produces an unstable and non-functional SMN protein, and one that includes exon 7 and encodes for the functional SMN protein [88–91]. Unfortunately, the functional isoform accounts for only 20 % of the total transcripts translated from SMN2, but modulation of the alternative splicing of exon 7 rescues the long isoform to levels comparable to the healthy condition. Different levels of severity can be distinguished in SMA patients according to the abundance of SMN encoded from the full-length transcripts of SMN2. Patients affected by SMA type I begin to display features of the disease (i.e., muscle weakness) around the sixth month of life, and death occurs as a result of respiratory failure before the age of 2 years [69]. Type II SMA patients experience muscle weakness between 6 and 18 months. Type III SMA patients can walk unaided but have muscle weakness and are unable to run; disease onset is after 18 months of age and they have a normal lifespan. Several genetic murine models of SMA [92–95] that mimic the genetic basis and phenotype of human SMA are available, offering the possibility to evaluate correction of the SMN2 splicing pattern and the rescue of the SMA neurodegenerative features in vivo, but their genetic background does not faithfully recreate the patients’ condition, as the murine genome is endowed with only one Smn gene.

Therefore, the results obtained in SMA mice can only offer speculation on the therapeutic possibilities of increasing SMN transcripts obtained from human SMN2, as their genome, being devoid of SMN2, does not include the endogenous biological machinery to regulate SMN2 splicing. In this context the utility of applying iPSC research to SMA is clear. The studies by Ebert et al. [67] provide a proof-of-principle for the use of iPSCs as an in vitro disease model system with possible applications in drug screening by using the SMA patients’ iPSCs. Ebert et al. [67] successfully differentiated the patient-derived iPSCs into motor neurons, and over time the cell body size was reduced and the cells underwent degeneration, demonstrating that the process of reprogramming and differentiation faithfully mimics the disease phenotype. Moreover, the iPSC-derived motor neurons were treated with valproic acid or tobramycin, two drugs that increase the expression of both full-length and truncated versions of the SMN protein, and these treatments prevented the death of SMA motor neurons [70].

The challenge with this approach to treatment, is engineering the patients’ cells so that they encode a sufficient amount of SMN protein to prevent α-motor neuron degeneration. Thus, engineering the patients’ fibroblasts, which can be reprogrammed into iPSCs, would allow re-implantation into the patients, avoiding the problems associated with allograft immune rejection.

Major advancements in stem cell regenerative medicine have been made by Dr Stefania Corti, whose group achieved improved neuromuscular function and increased lifespan in SMA mice intrathecally grafted with NSCs [96]. Because NSC derivation from the spinal cord is a limit for future pre-clinical studies due to scarcity, Corti’s group moved to experimentation with NSCs derived from embryonic stem cells, demonstrating that transplantation into SMA mice improves their phenotype and survival [97]. Corti et al. [56] recently published a study in which iPSCs were successfully engineered. iPSCs obtained from SMA patients were genetically corrected with single-stranded oligonucleotides (converting SMN2 into an SMN1-like gene), and they exhibited ameliorated features compared to uncorrected SMA iPSCs [56]. The corrected SMA iPSCs were differentiated into motor neurons and transplanted into SMA mice, improving the disease phenotype and extending the lifespan of the animals. These seminal studies hold great promise for autologous transplantation of genetically corrected SMA motor neurons derived from patients’ iPSCs.

The therapeutic possibilities offered by iPSCs may seem to be far from clinical trials, but much effort is being made to differentiate iPSCs into specific neuronal subtypes. Further elucidation of the molecular factors regulating neurogenesis is ongoing and this allows the generation of methods closer to the bed of the patients. As SMA is a severe disease with high incidence and fast progression of the degeneration of the spinal motor neurons, the more therapeutic approaches that are developed, the greater the possibility of offering a valid therapeutic benefit.

Parkinson’s disease

Parkinson’s disease is caused by the progressive loss of midbrain dopaminergic neurons in the nigrostriatal tract and is the second most common neurodegenerative disorder. Over 75 % of the dopaminergic nigrostriatal neurons are lost before the classical clinical features of idiopathic PD manifest: slowness to move (bradykinesia, difficulty initiating voluntary movement), increased muscle tone (rigidity as a result of simultaneous contraction of flexor and extensor muscles), and tremor at rest. In some cases, clear causative agents are identified, such as vascular lesions in the region of the nigrostriatal pathway or the administration of anti-dopaminergic drugs in schizophrenia [98]. In fact, antipsychotic drugs used to treat schizophrenia act mainly by blocking the stimulation of dopamine receptors by the neurotransmitter dopamine and for this reason their use may lead to the functional disruption of the same dopaminergic neurons affected in PD. Although more than 90 % of cases seem to be sporadic, many genes have been linked to PD, including LRRK2, PARK2, SNCA, UCHL1, PARK7, PINK1, GBA, and SNCAIP [99]. The most common therapy is dopamine replacement with levodopa (l-DOPA), the precursor to dopamine, as dopamine itself does not pass the blood–brain-barrier. The problem with l-DOPA is that it produces adverse effects (i.e., nausea, vomiting, hypotension, confusion, hallucinations) due to the over-stimulation of dopamine receptors. After many years of therapy (i.e., 5–15 years), l-DOPA becomes less effective and patients experience the ‘ON–OFF phenomenon’. One of the reasons l-DOPA becomes less effective is that it cannot be taken up by the depleted dopaminergic neurons. In this context, the design of new therapeutics may need to deal with the issue of trying to recreate synaptic networks and preserve them from degeneration [100]. Thus, cell replacement therapy using iPSCs seems particularly suitable for PD, because it aims to restore synaptic networks.

Transplantation of fetal midbrain cells has been shown to restore dopamine function in animal models and human patients [101–106]. Also, dopamine neurons derived from embryonic stem cells have been shown to function when grafted into parkinsonian rats [107, 108]. To evaluate the therapeutic potential of iPSCs, Wernig et al. [109] tested whether functional neurons can be generated from reprogrammed fibroblasts in vitro and showed that iPSC-derived neurons synaptically integrate and are functional after transplantation into mouse models of PD [109]. Using stringent cell sorting for cells negative for SSEA1, a pluripotency marker, the authors succeeded in reducing the number of contaminating undifferentiated iPSCs in the neural cell cultures. The elimination of undifferentiated cells led to a substantially reduced risk of teratoma formation after transplantation; they failed to detect teratomas in any of the transplanted rats 8 weeks after transplantation. One notable potential risk of therapeutic reprogramming is that unknown genetic factors that cause the patient’s disease can potentially lead to degeneration of the reprogrammed and transplanted cells. In light of the late onset of this neurodegenerative disease, this possible complication might not have a critical effect during the lifetime of a treated patient.

An iPSC-based model of PD was produced by Nguyen et al. [110], who derived iPSCs from a patient with a mutation in the gene encoding leucine-rich repeat kinase 2 (LRRK2). Mutations in LRRK2 are responsible for the most common cause of familial PD. Nguyen et al. [110] observed that patient-derived dopaminergic neurons exhibit increased vulnerability to stress by hydrogen peroxide, 6-hydroxydopamine. In fact, 6-hydroxydopamine is a catecholamine-selective neurotoxin that selectively accumulates in dopaminergic neurons because of its similarity with endogenous catecholamine, but 6-hydroxydopamine rapidly oxidizes at physiologic pH and forms reactive oxygen species (ROS), such as 6-hydroxydopamine-quinone and hydrogen peroxide that damage proteins, lipids and DNA, causing a selective neuronal degeneration that mimics the situation in Parkinson’ disease patients. The 6-hydroxydopamine toxicity is caused by oxidative stress, but also by direct interactions with cellular regulatory processes, as it inhibits the mitochondrial respiratory chain complexes I and IV [111]. These results are consistent with the notion that both genetic and environmental factors contribute to the development of PD, and confirming the fidelity of iPSCs in recapitulating the patients’ disease.

In conclusion, despite the opportunities offered by iPSC technology in the context of neurological diseases, many challenges still have to be faced (i.e., differences among iPSC lines in DNA methylation, maintenance of X-inactivation in female cells, and spontaneous differentiation leading to heterogenous cultures), but the technologies to overcome these difficulties are already available as recently explained by Sandoe and Eggan [112].

Therapeutic possibilities for translational medicine in neurobiology

iPSCs have been differentiated into a variety of phenotypes, including neurons [43–56], astrocytes [53, 57], and oligodendrocytes [58]. Myelin diseases can occur as inflammatory demyelinating disorders of adulthood or childhood leukodystrophies and cerebral palsy. The intracerebral delivery of neuroprogenitors into brains with demyelinated neurons may offer a feasible strategy for myelin repair. Thus, glial disorders are particularly promising initial targets for cell-based therapy of neurological disease. Using a common strategy for neuroprogenitor implantation, a broad set of pediatric and adult disorders of the brain and spinal cord may be subjected to structural repair (as reviewed by [113]).

Recently, Emborg et al. [52] demonstrated that autologous iPSC-derived neural cells transplanted in the primate brain are not rejected. Neural progenitors survive and produce mature neurons, astrocytes, and oligodendrocytes in the monkey brain. Importantly, the grafted neural cells appear to structurally integrate into the host brain, the differentiated neurons extend long processes, and the oligodendrocytes participate in myelination [52].

Uchida et al. [114] and Gupta et al. [115] also investigated the therapeutic use of NSCs, which can differentiate into myelin-producing oligodendrocytes. Uchida et al. [114] demonstrated that expanded human NSCs transplanted into three sites in the brains of neonatal or juvenile hypomyelinated mice can differentiate into oligodendrocytes. Transplanted NSCs generated functional myelin, even in animals with severe symptomatic hypomyelination, suggesting that this strategy may be useful for treating dysmyelinating diseases [114]. Furthermore, NSCs were capable of protecting host neurons from cell death under conditions of progressive central nervous system neurodegeneration (as observed in a mouse model of infantile neuronal ceroid lipofuscinosis) or injury (as observed in ischemic cerebral cortices of rats after distal middle cerebral artery occlusion, spinal cord-injured mice, and in early chronic spinal cord injury NOD-SCID mouse model), without evidence of tumor formation or microglial cell activation [116–120]. In a rodent model of acquired spinal cord contusion injury, NSCs restored some locomotor function, and cell fate analysis revealed that a high proportion of transplanted NSCs differentiated into oligodendrocytes or neurons without any evidence of glial scar formation, suggesting the potential of these cells for treating inherited or acquired demyelinating disorders [117, 120, 121].

In a sister study performed in humans, Gupta et al. [115] transplanted NSCs into patients with Pelizaeus-Merzbacher disease (PMD), a severe progressive dysmyelinating disorder. MRI studies demonstrated consistent durable and progressive myelin production after transplantation, and modest gains in neurological function in three of the four subjects analyzed. The clinical outcomes indicate that the intervention is safe and tolerated by PMD patients. The biological properties of the NSCs and the radiological findings in this study suggest donor-derived myelination in the region of cell transplantation, and suggest the potential for applying this approach to other myelination disorder [115]. Overall, these results support further studies on the clinical benefits of NSC transplantation in dysmyelination disorders.

The potential of human iPSC therapy in stroke has been evaluated by Chen et al. [122], and Kawai et al. [123], who transplanted undifferentiated murine iPSC in cerebral ischemic rats and mice, respectively. Chen et al. [122] demonstrated improvement of motor function, reduction of infarct size and neuroprotection, while Kawai et al. [123] showed tridermal teratoma formation following transplantation. These studies suggest that iPSC technology can offer potential therapeutic efficacy, but tumorigenesis events should be carefully controlled and avoided. A possible way to overcome this problem is to investigate the use of iPSC-derived cells instead of undifferentiated iPSCs. Importantly Polentes et al. [124] used iPSC-derived neural progenitors in a stroke model and demonstrated that these cells transplanted in postischemic striata of Sprague–Dawley rats exert trophic effects on the host brain before their integration. Another demonstration that iPSCs can be considered for therapeutic applications comes from the study of Espuny-Camacho et al. [48], who showed that human ESC-derived cortical neurons transplanted into the neonatal brain of CD1 and NOD/SCID mice establish functional synapses with the host circuitry. These results suggest that ESC/iPSC-derived neurons are successfully integrated in the brain in vivo.

Therapeutic difficulties for clinical applications of iPSC transplantation

The great potential of iPSCs consist in providing cell sources for better understanding various diseases, for developing drugs, and for performing cell transplantation therapies. But before iPSC technology can be considered as an effective tool for translational medicine, many challenges should be overcome. iPSC technology solves the obstacles of immune rejection and ethical concerns related to human ESCs, but it is facing problem related to teratoma formation after transplantation into patients. Anomalous reprogramming may lead to iPSCs with aberrant differentiating potential and the risk is the development of teratoma [125]. iPSCs are, in fact, able to self-renew and differentiate into cells from all three germ layers. Any undifferentiated cells present in differentiated cell cultures used for regenerative medicine can lead to teratoma formation, therefore any of these undifferentiated cells should be carefully removed before any clinical use. Hence, effective methods for removal of undifferentiated cells is currently under investigation in several laboratories, and for this reason, compliance of Good Manufacturing Practices (GMPs) is fundamental for the use of iPSCs in cell therapy [11]. Thus, the refinement of the methodologies for efficient iPSCs reprogramming and the development of methods to identify aberrant iPSC is fundamental prior to therapeutical application. For example many laboratories are currently engaged in developing somatic reprogramming using only small-molecule compounds [126]. An additional challenge for iPSC technology is avoiding chromosomal abnormalities. Somatic cells (i.e., fibroblasts) may have mutations that will be propagated when they are reprogrammed into iPSCs. For this reason it is highly recommended to confirm the results obtained using iPSCs from patients and control subjects on at least three independently obtained colonies. In addition, the reprogramming itself may increase the probability of an altered genome from those of the individual from whom the iPSCs were derived. Thus, karyotypic abnormalities may occur in iPSCs and compromise the differentiation potential, leading to augmented tumorigenicity [127–129]. Furthermore, simply propagating cells in culture may lead to karyotypic abnormalities. Therefore, monitoring iPSC karyotypes should be considered compulsory immediately following reprogramming and during propagation.

Conclusions

iPSC research has potential applications in: (1) establishing disease models, (2) performing drug screening, and (3) cell transplantation-based therapy.

In neurobiology, most of the current knowledge about disease-related neuronal phenotypes has been gathered from postmortem studies in humans because the ability to obtain live brain tissue is limited. Thus, establishing disease models would offer many advantages. The crucial experiments performed in monogenic diseases exemplify the power of iPSCs in illuminating human disease at the molecular, cellular, and functional levels, and highlight the potential of reducing the risk of drug discovery programs at the preclinical stage.

The ability to differentiate iPSCs from one patient into cell types that enable testing for cardiac, hepatic, and neuronal toxicity at the preclinical stages has the potential to lower the risk of compounds at an earlier stage of drug development. This approach can reduce the cost burden of failed drug development programs and enable the enrichment of programs with the highest probability of success [130]. To meet its therapeutic potential, the iPSC-based drug discovery platform will have to achieve more efficient reprogramming programs, more efficient directed differentiation, and more robust disease phenotypes, especially for complex and common diseases.

One major advantage of the development of iPSCs is that it overcomes the ethical controversies and rejection issues of autologous stem cells, though tumor formation is a challenge for cell-replacement therapy. Many scientists are currently dedicated to avoiding genomic integration in the derivation of safe iPSCs for cell replacement therapy by improving transgene-free or non-viral methods. Despite the positive results obtained from studies using iPSCs, the field is still in its infancy and numerous challenges must still be overcome. Because these challenges are demanding, we hope that future collaboration between academic groups and industry will occur in order to provide the quickest route to the development of novel therapeutics for patients. In conclusion, we wish that in the near future the progress in gene therapy and iPSC technology will allow reprogramming of somatic cells from a patient into iPSCs, which can than be genetically corrected and differentiated into the cell type of interest before safely transplanting them back into the patient without the risk of tumorigenesis.