Derivation and Expansion Using Only Small Molecules of Human Neural Progenitors for Neurodegenerative Disease Modeling

Phenotypic drug discovery requires billions of cells for high-throughput screening (HTS) campaigns. Because up to several million different small molecules will be tested in a single HTS campaign, even small variability within the cell populations for screening could easily invalidate an entire campaign. Neurodegenerative assays are particularly challenging because neurons are post-mitotic and cannot be expanded for implementation in HTS. Therefore, HTS for neuroprotective compounds requires a cell type that is robustly expandable and able to differentiate into all of the neuronal subtypes involved in disease pathogenesis. Here, we report the derivation and propagation using only small molecules of human neural progenitor cells (small molecule neural precursor cells; smNPCs). smNPCs are robust, exhibit immortal expansion, and do not require cumbersome manual culture and selection steps. We demonstrate that smNPCs have the potential to clonally and efficiently differentiate into neural tube lineages, including motor neurons (MNs) and midbrain dopaminergic neurons (mDANs) as well as neural crest lineages, including peripheral neurons and mesenchymal cells. These properties are so far only matched by pluripotent stem cells. Finally, to demonstrate the usefulness of smNPCs we show that mDANs differentiated from smNPCs with LRRK2 G2019S are more susceptible to apoptosis in the presence of oxidative stress compared to wild-type. Therefore, smNPCs are a powerful biological tool with properties that are optimal for large-scale disease modeling, phenotypic screening, and studies of early human development.


Introduction
Stem cells are positioned to revolutionize drug discovery for neurodegenerative disorders through in vitro disease modeling and HTS. Through reprogramming of primary cells from a patient, induced pluripotent stem cells (iPSCs) can be generated with properties comparable to human embryonic stem cells (hESCs) [1]. For example, two groups have reported that midbrain dopaminergic neurons (mDANs) differentiated from iPSCs generated from patients with Parkinson's disease (PD) with mutations in the gene LRRK2 exhibit disease-associated phenotypes [2,3].
Similarly, it has been demonstrated that motor neurons (MNs) differentiated from hESCs with mutations causing amyotrophic lateral sclerosis (ALS) are susceptible to degeneration [4]. In principle, the phenotypes exhibited by patient-specific cells could be used in high-throughput screening (HTS) campaigns to identify novel neuroprotective compounds for development into new drugs.
Because HTS campaigns can involve up to several million independent wells, the reproducibility of the assay must be extremely high, or else the results will be uninterpretable. Consequently, for the promise of stem cells to become realized at least two hurdles must be overcome. First, HTS using models of neurodegenerative diseases requires cells that are characterized by very robust expansion to produce billions of cells under chemically defined conditions. Second, these cells must be capable of efficient formation of neurons such as mDANs and MNs. However, current stem cell protocols are not robust, require expensive recombinant growth factors, involve manual manipulation, have need of frequent splitting at narrow ratios, need significant time for differentiation, and/or result in inefficient differentiation. As a result, stem cell cultures typically exhibit tremendous variability and, as a result, it is often difficult to reproducibly obtain a statistically significant result using triplicate wells. Therefore, previous cell types including neural stem cells (NSCs) [5], longterm self-renewing rosette-type hESC-derived NSCs (lt-hESNSCs) [6], primitive NSCs (pNSCs) [7], and rosette neural cells (R-NCs) [8], are not easily compatible with HTS. We have identified a novel type of neural progenitor cells capable of robust, immortal expansion and efficient differentiation into both central nervous system (CNS) and neural crest lineages. These properties are so far only matched by pluripotent stem cells. Additionally, our neural precursor cells only require small molecules for self-renewal and expansion, a feature that significantly reduces the cost of largescale disease modeling and, to date, is not possible with any other available cell type.
Although NSCs are competent to differentiate into CNS lineages including neurons, astrocytes and oligodendrocytes, they are not able to efficiently form mDANs or MNs, which makes them unsuitable for neurodegenerative disease modeling [5,9]. Similar to NSCs, lt-hESNSCs, which are differentiated from hESCs, use the same recombinant proteins as NSCs for selfrenewal -FGF2 and EGF -and express markers of ventral hindbrain identity. When treated with the developmental patterning factors Sonic hedgehog (SHH) and Fibroblast growth factor 8 (FGF8) or SHH and retinoic acid (RA), about 70% of differentiated neurons were not mDANs. When cultured with SHH and retinoic acid (RA), about 85% of differentiated neurons were not MNs. These results make large-scale modelling of PD and ALS problematic. In addition, lt-hESNSCs require splitting three or more times per week at very low ratios, which makes expansion to billions of cells for HTS very tedious and cumbersome.
Two cell types have been reported with increased differentiation potential. Primitive NSCs (pNSCs) could efficiently be differentiated into mDANs (about 55%) and MNs (about 54%). However, pNSCs require recombinant Leukemia inhibitory factor (LIF) for self-renewal, which makes them very cost-intensive to culture in large quantities and, therefore, impractical for HTS. Elkabetz et alia reported the derivation of rosette-neural cells (R-NCs) from hESCs [8]. R-NCs could be differentiated into both CNS and neural crest lineages. However, despite the use of high doses of recombinant growth factors, R-NCs were only expandable for up to 4 passages before differentiating. As a result, it is not possible to generate enough R-NCs for even a pilot study to validate an HTS assay let alone an entire HTS campaign. As a result, it is urgently necessary to derive a cell type that is both plastic and can be propagated in a manner compatible with HTS.
We speculated that signals present in the neural plate border region may be sufficient to direct self-renewal of cells in vitro with potential to differentiate into CNS and neural crest lineages including peripheral nervous system (PNS) neurons. During embryogenesis, neuroepithelial cells on the border between the neural plate and neural crest have the developmental potential to form neural tube-derived CNS lineages as well as neural crestderived lineages such as peripheral nervous system (PNS) neurons [10,11]. WNT proteins specify formation of cells at the lateral border of the neural plate and are potent mitogens [12]. Patterning by WNT proteins is antagonized by SHH, which specifies ventral neural tube fates and is also a potent mitogen [12]. It is significant to note that SHH signaling was used by Elkabetz et alia to culture R-NCs [8]. Because R-NCs could not be expanded without committing to the CNS, we hypothesized that WNT signaling in combination with SHH signaling might contribute to the maintenance of precursors with developmental potential for both neural crest and neural plate.
Here we report the derivation of human smNPCs, which have properties uniquely suited to modeling neurodegenerative diseases. We show that smNPCs can be efficiently specified into neural tube and neural crest lineages. This developmental potential is similar to R-NCs, and developmentally upstream of NSCs, lt-hESNSCs, and pNSCs. Unlike R-NCs, we demonstrate that smNPCs are capable of immortal self-renewal using WNT and SHH signals. This combination is also distinct from FGF2 and EGF required for NSCs and lt-hESNSCs. We show that culturing smNPCs with FGF2 results in the formation of rosette-like structures, which have been previously associated with the neural plate-stage of embryogenesis [13]. In addition, smNPCs are easy to handle, do not require manual manipulation, and can be cultured at a wide range of cell densities using only inexpensive small molecules. Finally, we show that mDANs differentiated from smNPCs with the PDassociated mutation LRRK2 G2019S are more sensitive to stress compared to wild-type. Therefore, smNPCs are a robust and affordable tool suitable for disease modeling.

Derivation of a Population of Expandable Human Neural Epithelial Cells
We tested the effects of introducing both WNT and SHH signals to cultures of differentiating pluripotent stem cells. hESCs and hiPSCs were differentiated via human embryoid bodies (hEBs). Neural induction was initiated through inhibition of both BMP and TGFb signaling using the small molecules dorsomorphin (DM) and SB43152 (SB) [14,15]. The small molecule CHIR99021 (CHIR), a GSK3b inhibitor, was added to stimulate the canonical WNT signaling pathway. The SHH pathway was stimulated using the small molecule purmorphamine (PMA). Differentiating hEBs exposed to CHIR and PMA were marked by the formation and expansion of an epithelium ( Figure 1A). These epithelial cells expressed markers of neural progenitors including SOX1, SOX2, NESTIN, and PAX6, but not mesodermal marker T ( Figure S1A). When disaggregated and plated on Matrigel, homogeneous colonies of epithelial cells formed ( Figure 1B). These neural epithelial cells could be split enzymatically without manual selection at a 1:10 to 1:20 ratio and expanded as cell lines for more than 150 population doublings and exhibited a diploid karyotype ( Figure S1B). An analysis of doubling time indicated that neural epithelial cells divided approximately once per day, which was consistent over multiple passages as well as cultures derived from multiple pluripotent stem cell lines ( Figure S1C). Immunostaining of neural epithelial cell colonies demonstrated the uniform expression of the neural progenitor markers SOX1, SOX2, NESTIN, and PAX6 ( Figures 1C). Immunostaining also showed the expression of FORSE1, which has been previously associated with early neural progenitors ( Figure 1C) [7]. Upon spontaneous differentiation by withdrawal of the small molecules used for expansion, smNPC differentiated into cells expressing NEUN/ TUBBIII, GFAP/S100beta, and O4/OLIG2, which mark neurons, astrocytes, and oligodendrocytes ( Figures 1D and S2). Therefore, neural epithelial cells express markers that are characteristic of early neural progenitors. Interestingly, the cellular morphology and culture conditions of the neural epithelial cells are not typical of NSCs [5].
Neural epithelial cells after 20 passages were further characterized by microarray expression analysis. These cells did not express the pluripotency markers OCT4 and NANOG, the mesendodermal markers AFP, T, and SOX17, and the trophoblast marker EOMES ( Figure S3A). In contrast, neural epithelial cells showed high expression of neural precursor markers, including SOX2, PAX6, HES5, and ASCL1 ( Figure S3A). qRT-PCR analysis confirmed that neural epithelial cells express markers of neural progenitors, including SOX2, SOX1, PAX6, and PAX3, which were stably expressed beginning at about passage 5 (Table S1 and Figure  S3B). This suggests that neural epithelial cells stably expressed markers of neural fate commitment after extensive cell culture starting at about passage 4. qRT-PCR also confirmed that neural epithelial cells did not express non-neural markers, including OCT4, AFP, SOX17, CK8, CK18, and T, after 4 to 5 passages ( Figure S3B). This was achieved without any manual manipulation, just by replating the cells under specific culture conditions that strongly favored the expansion of these cells. These results demonstrate that cultures of neural epithelial cells do not contain subpopulations of cells expressing genetic markers of mesodermal, endodermal, trophoblastic, or pluripotent cells.
Interestingly, although the neural epithelial cells do not morphologically resemble neural rosettes, microarray analysis demonstrated strong expression of the (pre-) neural rosette genetic markers DACH1, PLZF, LMO3, NR2F1, PLAGL1, LIX1, and EVI1 ( Figure S3A). This suggested that neural epithelial cells might have the ability to form neural rosettes. To test this, we cultured neural epithelial cell colonies in the presence of FGF2, which has previously been reported to induce hEBs to form neural rosettes [7,16]. After 2 days of culturing neural epithelial cell colonies in the presence of FGF2, numerous areas were found to exhibit the morphology of neural rosettes ( Figure S4A). For further characterization, we immunostained the neural epithelial cells for N-CADHERIN and ZO-1, which is expressed by neural rosettes but spatially localized to the apical surface [8]. Although N-CADHERIN and ZO-1 expression were readily detected in colonies of neural epithelial cells, they were not spatially localized within the colonies ( Figure S4B). In contrast, after FGF2 treatment, N-CADHERIN and ZO-1 expression were found at the apical surface of the rosettes ( Figure S4B). Therefore, neural epithelial cells express early rosette markers and are capable of forming neural rosettes when cultured under appropriate conditions.

CHIR and PMA Modulate Expression of Markers of Neural Crest and Ventral Neural Tube Identity in Human Neural Epithelial Cells
Using microarray transcription profiling, we assessed the expression of markers of regional identity by cultures neural epithelial cells. PAX3, IRX3 and MSX1, which mark marginal neural plate [17] or neural plate border cells [12], were readily detectable in neural epithelial cells ( Figure S5A). SOX9, which marks neural crest cells, was also present, albeit to a lesser extent. NKX6.1, OLIG2, NKX2.2, and FOXA2, which mark medial neural plate and ventral neural tube, were not detected. GSH2, which marks dorsal neural tube progenitors, was also not detected. Of the tested rostrocaudal markers, only HOXA2 and HOXB2, which mark anterior hindbrain identity, were highly expressed ( Figure   S5B). This pattern is consistent with the neural plate border and marginal neural plate region, which could suggest competence to differentiate into CNS and PNS lineages [10,11,12,18]. However, we are unaware of cells in vivo that express the same combination of markers as smNPCs.
Next, we asked if altering the concentrations of CHIR and PMA effected expression of lineage markers by cultures of neural epithelial cells. To answer this, neural epithelial cells at passages 15-20 were cultured with different concentrations of CHIR and PMA alone or in combination for 6 days. qRT-PCR demonstrated that 3 mM CHIR alone induced expression of PAX3 slightly more than MSX1 compared with expansion conditions (3 mM CHIR plus 0.5 mM PMA; Figure 2A). When the concentration of CHIR was reduced from 3 mM to 1.5 mM, MSX1 and PAX3 were downregulated by about the same amount ( Figure 2A). When neural epithelial cells were cultured with PMA alone, MSX1 and PAX3 were down-regulated compared to expansion conditions. In addition, PMA alone resulted in increased the expression of NKX6-1, NKX2-1, OLIG2, and FOXA2 compared to expansion conditions ( Figure 2A). Of interest, we found that expression of the neural progenitor marker SOX1 increased in cultures treated with only PMA compared to cultures grown under expansion conditions ( Figure 2A). These results are consistent with the developmental roles of WNT and SHH signaling [12,19].
To confirm the qRT-PCR results, we performed immunostaining on neural epithelial cells cultured for 6 days with PMA, CHIR, or both. PMA alone resulted in decreased PAX3 expression as well as increased expression of NKX6.1, NKX2.2, and FOXA2 compared to expansion conditions as well as CHIR alone ( Figure 2B). In addition, more cells expressed NKX6.1, NKX2.2, and FOXA2 at 1 mM PMA compared to 0.5 mM PMA. In contrast, fewer PAX3-positive cells were observed at 1 mM PMA than at 0.5 mM PMA. There was increased SOX1staining intensity in cells treated with only PMA compared to cells cultivated under expansion conditions ( Figure 2B). More cells were SOX1-positive with PMA compared to cells treated with CHIR alone ( Figure 2B). These data are consistent with the qRT-PCR results and suggest that cultures of neural epithelial cells can be directed to differentiate into ventral neural tube-and neural crestderived lineages using CHIR and PMA, respectively. However, these differences may reflect different subpopulations within the culture. Experiments below discriminate between these alternatives.
Next, we sought to determine if expression of markers of rostrocaudal identity by cultures of neural epithelial cells could be modulated by developmental factors. Retinoids are produced in vivo by somites and specify spinal cord fate, which can be mimicked in vitro with all-trans retinoic acid (RA) [20]. Neural epithelial cells at passage 20 treated with RA for 1 or 8 days exhibited lower levels of HOXA2 and increased levels of HOXA4, HOXB4, but not HOXA9 ( Figure 2C). Therefore, RA directs cultures of neural epithelial cells to express caudal patterning markers, consistent with the role of retinoids during embryonic development. In contrast, under none of the tested conditions were forebrain markers, such as FOXG1, induced, suggesting that neural epithelial cells are committed to expressing markers of caudal fates, consistent with the role of WNT factors in caudal fate specification.

Directed Differentiation of Neural Epithelial Cells into Neural Crest Lineages: Peripheral Neurons and Mesenchymal Cells
When cultured with only CHIR, cultures of neural epithelial cells expressed markers that might indicate potential to differen- Immunostaining of spontaneously differentiated neural epithelial cells for TUBBIII and NEUN, for GFAP and S100-beta after astrocyte differentiation, as well as O4 and OLIG2 after spontaneous differentiation, indicating oligodendrocyte formation. Scale bars are 100 mm. See also Figure S1. doi:10.1371/journal.pone.0059252.g001 tiate into neural crest cells. For this reason, we tested the ability of neural epithelial cells to differentiate into neural crest cells, by culturing cells at passages 15-25 in the presence of CHIR for 2 days followed by BMP4 without CHIR or 10% fetal calf serum (FCS; Figure 3A). After four days, cells exhibited upregulation of PAX7, SLUG, and the neural crest surface marker HNK-1, but downregulation of PAX6 ( Figures 3B and C). qRT-PCR demonstrated that PAX7, SOX9 and TFAP2A were induced by BMP4 ( Figure S6A). Immunostaining showed that 2 days of CHIR and BMP4 induced expression of TFAP2A ( Figure S6B). This combination of markers is consistent with neural crest progenitors.
After maturation for two weeks, we observed greater than 80% of neurons expressed PERIPHERIN ( Figure 3D). Some of the PERIPHERIN-positive neurons expressed the neural crest marker TFAP2A ( Figure S5C). Immunostaining and confocal microscopy demonstrated that a subset of cells expressed both PERIPHERIN and BRN3A, which is a combination that specifically marks PNS sensory neurons ( Figure 3D and Figure S6D). After differentiation and maturation, qRT-PCR analyses confirmed that peripheral sensory neuron markers PERIPHERIN and BRN3A were upregulated by BMP4 ( Figure 3E). As expected, 8 days of treatment with PMA instead of BMP4 essentially abolished expression of these markers ( Figure 3F). Overall, the efficiency of directing differentiation of neural epithelial cells into PERIPHERIN and TUBBIII double-positive cells was about 40 to 50% ( Figure 3F). We conclude that cultures of neural epithelial cells are capable of forming cells expressing markers of peripheral neurons, including sensory neurons.
Neural crest cells can form non-neural cells including mesenchymal cells. To test whether neural epithelial cells can also form non-neural cells, the cells at passages 20-25 were differentiated using first CHIR only for 2 days and then serum-containing medium for 14-21 days ( Figure 3A). Under these conditions, we observed the formation of mesenchymal cells that could be cultured for more than 10 passages and resembled primary human fibroblasts. Human fibroblasts, which are mesenchymal cells, as well as the differentiated mesenchymal cells expressed the markers VIMENTIN, CD9, SMA, NESTIN, and alkaline phosphatase ( Figure S6E). Expression of SMA and alkaline phosphatase may indicate spontaneous differentiation of cells into smooth muscle and osteoblastic cells. Finally, using mesenchymal stem cell protocols, we were able to differentiate the neural epithelial cellderived mesenchymal cells into cells expressing markers of adipocytes and osteocytes ( Figure S6F). Based on these data, we conclude that cultures of neural epithelial cells are capable of forming cells expressing markers of peripheral neurons as well as mesenchymal neural crest cell derivatives.

Directed Differentiation of Neural Epithelial Cells into Ventral Neural Tube Lineages: mDANs and MNs
When cultured with only PMA, cultures of neural epithelial cells expressed markers that might indicate potential to differentiate into ventral neural tube cell lineages. For this reason, we tested the ability of neural epithelial cells to differentiate into mDANs and MNs, which are derived from ventral neural tube progenitors in vivo. First, we exposed neural epithelial cells at passages 15-25 to PMA and FGF8 for 8 days ( Figure 4A), which specify formation of ventral midbrain cells including mDANs [21]. After maturation for 2 weeks, immunostaining of neural epithelial cells demonstrated that a large proportion had differentiated into TH-, FOXA2-, and TUBBIII-positive neurons, an expression pattern that specifically marks mDANs ( Figures 4B and C). qRT-PCR showed upregulation of markers of mDAN differentiation, including EN-1, LMX1A, LMX1B, NURR1, FOXA2, and AADC ( Figure 4D). The overall efficiency (up to ,35% of total cells and up to 70% of neurons) of differentiation of neural epithelial cells into cells expressing markers of mDAN identity was consistent among 3 different neural epithelial cell cultures at multiple different passage numbers from three different pluripotent stem cell lines ( Figure 4E). Thus, cultures of neural epithelial cells have the potential to differentiate into cells expressing markers indicative of mDAN identity.
SHH and RA signaling in combination specify the formation of MNs [22]. As the expression of markers of neural fate by neural epithelial cells is modulated by both SHH and RA, we asked whether both factors together can direct differentiation along the MN lineage. Neural epithelial cells at passages 15-25 were treated for 2 days with 1 mM PMA and then for 8 days with 1 mM PMA and 1 mM RA to induce ventralization and caudalization, with subsequent maturation for 2 weeks without PMA or RA ( Figure 5A). Immunostaining of the cells demonstrated the presence of ISLET1 and CHAT double-positive cells that also expressed MAP2 and SMI32, consistent with MN identity (Figures 5B and C). MN identity is supported by the presence of HB9 and TUBBIII double-positive cells (immunostaining, see Figure 5D). qRT-PCR analysis showed significant upregulation of markers of MN differentiation including HB9, ISLET1, CHAT, and HOXB4 (qRT-PCR, see Figure 5E). Counting of single cells after maturation showed that neural epithelial cells formed cells expressing markers of MN identity with an efficiency of approximately 50% ( Figure 5F).

Cultures of Neural Epithelial Cells Contain Multipotent Stem Cells Capable of Forming both Neural Tube and Neural Crest Cell Lineages
Although these results indicate that cultures of neural epithelial cells have the potential to differentiate into both neural crest and neural tube cell derivates, it is possible that these derivatives arise from multiple subpopulations of cells in these cultures separately committed to either neural tube or neural crest cell fates. We therefore asked whether single neural epithelial cells are multipotent and capable of forming both neural tube and neural crest cell lineages. To answer this question, we generated 3 clonal neural epithelial cells lines from a single hESC-derived culture of neural epithelial cells ( Figure S7A). These clonal lines expressed the neural progenitor makers NESTIN, SOX2, SOX1, and PAX6 ( Figure S7B). Each of these three lines could be differentiated into mDANs, MNs, and PNS neurons ( Figure S7C). Finally, we demonstrated that each of these three lines could be directed to differentiate into cells expressing markers of astrocytes and olidodendrocytes ( Figure S7D). Therefore, we concluded that cultures of neural epithelial cells contain cells that self-renew and are competent to differentiate into cells expressing markers consistent with caudal derivatives of both neural tube and neural   crest cells. Because of their differentiation capacity as well as their ability to self-renew with only small molecules, we termed these cells small molecule neural progenitor cells (smNPCs).
Neurons Formed from smNPCs are Electrophysiologically Functional, Integrate and Mature in vivo Our next objective was to evaluate the electrophysiological function of smNPC-derived mDANs using patch clamping after two weeks of maturation. Stepping the membrane holding potential from 270 to +20 mV with 10 mV increments elicited a fast-activating, fast-inactivating inward current followed by a slower activating, slowly deactivating outward current (Figure 6A). The I-V curves of both currents are typical for sodium inward and potassium outward currents through voltage-gated channels ( Figure 6B) [23,24]. Current-clamp recordings demonstrated the presence of neurons that spontaneously fired action potentials (APs) with frequencies of up to 2.1 Hz (mean 1.0060.28 Hz, n = 12; Figure 6C). One critical test of neuronal identity is whether smNPC-derived neurons can form functional synaptic connections using spontaneous miniature events [25]. To this end, we used an in voltage clamp whole-cell configuration with a holding potential of 270 mV and with a frequency 0.3560.11 Hz. The average amplitude of miniature spontaneous postsynaptic currents was 21.1862.47 pA (peak value; n = 7 cells, 360 events analyzed). Representative trace and offline analysis results are shown in Figure S8A-F. The offline analysis revealed that recorded miniature spontaneous postsynaptic currents have the amplitude or kinetic parameters comparable to those of human neurons [26,27,28]. These results demonstrate that smNPC-derived neurons have acquired the electrical properties of excitable neurons and developed synaptic contacts between neurons.
Our final test was to evaluate the survival and differentiation potential of smNPCs in vivo. 1.5610 5 smNPCs and smNPCs differentiated with PMA and FGF8 for 8 days were stereotactically transplanted into the midbrain of adult mice. Two weeks and eight weeks after transplantation, transplanted cells were identifiable with species-specific antibodies against human Nuclei and human NCAM ( Figure 6D Figures 6F, 6G and S9). Only when smNPCs were predifferentiated with PMA and FGF8 and transplanted into the substantia nigra, did they continue to differentiate towards the dopaminergic subtype in vivo, as demonstrated by the presence of TH and FOXA2 double-positive cells (Figures 6H and S9F). Therefore, smNPCs are capable of differentiating into neurons, including dopaminergic neurons, in vivo after transplantation. smNPCs were also derived from iPSCs. iPSC-derived smNPCs expressed similar markers and had the same differentiation potential as hESC-derived smNPCs ( Figure S10).
smNPC-derived mDANs with LRRK2 G2019S are Susceptible to Degeneration Next, we assessed the ability of smNPCs to recapitulate neurodegeneration-associated pathology in vitro. For this experiment, smNPCs were derived from iPSCs generated from patients with PD carrying the mutation LRRK2 G2019S as well as from age-and gender-matched controls. These smNPCs were then matured into cultures of mDANs for two weeks. First, we determined the affects of oxidative stress by culturing replated neurons in N2 medium alone to eliminate the antioxidants present in the B27 supplement and the neurotrophins used in the standard culture medium. After 48 hours, we observed an increase of 37% in cultures with LRRK2 G2019S in the number of TH and cleaved CASPASE3 double-positive cells, which marks dopaminergic neurons undergoing apoptosis (Figure 7 and Figure S11). Addition of 6-hydroxydopamine or rotenone resulted in a significant increase in the number of TH and cleaved CASPASE3 doublepositive cells (Figure 7 and Figure S11). These results are in agreement with the previously published phenotype [3]. Under all stressing conditions, more than 80% of the cells that were cleaved CASPASE3-positive also expressed TH ( Figure S12). Therefore, apoptosis is preferentially induced in mDA neurons under the tested conditions. However, in contrast to the previous publica-tion, this protocol is rapid, efficient, uses robustly expandable cells, and involves no manual manipulation. These characteristics should make smNPC-derived disease models more easily amenable to HTS compared to previous cell types.

Discussion
HTS on stem cell-based phenotypic assays have the potential to discover revolutionary new drugs to treat neurodegenerative diseases. However, the scale of HTS campaigns requires a source of cells capable of robust and immortal expansion without costly growth factors or cumbersome manual steps. In addition, the cell source needs to be able to efficiently differentiate into lineages such as MNs and mDANs for disease modeling. Here, we have demonstrated that smNPCs possess these properties (Table 1 and Figure 8). No other reported cell type such as NSCs, lt-hESNSCs, pNSCs, R-NCs or even direct differentiation of hPSCs can meet these requirements. Although one report demonstrates the use of NSCs derived from hESCs for HTS, it is interesting to note that the screen was for small molecules that were selectively toxic to cells that were not dopaminergic neurons [29]. While this does, indeed, establish proof-of-principle for HTS using human NSCs, we argue that using inefficient differentiation protocols together with chemicals that are toxic to most of the resulting cells is likely to introduce many artifacts and is clearly not an optimal approach.
The relationship of cell types derived in vitro such as lt-hESNSCs and pNSCs, with specific cell populations in developing embryos is of significant interest [13]. Unfortunately, such a comparison is not possible at present because data directly comparing the gene expression of specific embryonic cell populations with cells generated in vitro is lacking. Nevertheless, it is possible to speculate about the order of these cells based upon their reported differentiation potential. Pluripotent stem cells, of course, have the greatest differentiation potential of any cell type than can be cultured in vitro. Next, smNPCs and R-NCs have the competence to form both neural tube and neural crest lineages. In contrast, pNSCs are restricted to the CNS and unable to differentiate into PERIPHERIN-positive neurons [7]. However, pNSCs are able to efficiently form both mDANs and MNs. Finally, NSCs are restricted to the CNS and are unable to efficiently form mDANs and MNs. Although the differentiation potential of lt-hESNSCs for neural crest has not been robustly tested, treatment of lt-hESNSCs with BMP4 after pre-treatment with valproic acid and 59-aza-29deoxycytidine resulted in only 5% of cells expressed PAX3 [30]. This strongly suggests that lt-hESNSCs are restricted to CNS fates. Like NSCs, they require FGF2 and EGF for self-renewal, and their ability to form mDANs and MNs is significantly reduced compared to pNSCs. For these reasons, we would order the cell types with decreasing differentiation potential as follows: hPSCs.smNPCs and R-NCs .pNSCs.lt-hESNSCs.NSCs.

Ethics Statement
Informed consent was obtained for all patients donating samples to this study prior to the donation using a written form and protocol that was prior approved by the instutional review board: Ethik-Kommission der Medizinischen Fakultä t am Universitä tsklinikum Tübingen. All experiments involving animals (e.g. cell transplantation) were carried out in accordance with local institutional guidelines under the protocol 87-51.04.2011.A057, which was approved by Landesamt für Natur, Umwelt und Verbraucherschutz of the state of North Rhine-Westphalia, Germany. In vitro experiments were carried out with existing cell lines obtained from previous studies except where noted below. The appropriate citations are given next to each cell line in the Materials and Methods.

Generation of iPSCs
The iPSCs used in this study were newly generated. Informed consent was obtained from all patients involved in our study prior to cell donation as described in the ethics section above. Dermal fibroblasts, obtained from skin biopsies of patients with PD and healthy controls, were cultured in fibroblast medium, which consisted of DMEM supplemented with 10% fetal calf serum, 1% penicillin/streptomycin/glutamine, 1% nonessential amino acids, 1% sodium pyruvate (all PAA), and 0.5 mM beta-mercaptoethanol (Invitrogen).
The reprogramming of human dermal fibroblasts was adapted from Takahashi et alia [1]. Retroviral vectors containing OCT4 (Addgene 17217), SOX2 (Addgene 17218), KLF4 (Addgene 17219), and, when indicated, c-MYC (Addgene 17220) were co-transfected using Fugene 6 (Roche) into 293 T cells (purchased from ATCC) together with the appropriate packaging plasmids (Addgene 8454    and 8449). After 48 hours, supernatants containing viral particles were applied to the patients' fibroblasts in the presence of 6 mg/ mL protamine sulphate (Sigma Aldrich). Two to four infections were performed for each fibroblast sample. One day later, fibroblasts were reseeded on mouse embryo fibroblast (MEF) feeder cells or on gelatin-coated cell culture dishes. The mouse embryonic fibroblasts (MEFs) used in this study were derived in the laboratory of Prof. Dr. Hans Schöler and have been reported previously [31]. The following day, hESC medium supplemented with 1 mM valproic acid (Sigma Aldrich) was added, and the culture medium was changed daily thereafter. After 10-14 days, iPSC-like colonies were observed and valproic acid was discontinued. Individual colonies were isolated and clonally expanded. In total, two iPSC lines, designated Control 1 and Control 2, were derived from healthy patients and two iPSC lines, designated LRRK2 1 and LRRK2 2, were derived from patients with PD harboring the mutation G2019S.

Pluripotent Stem Cell Culture
The human ESC line HUES6 was used in this study and was purchased from the hESC Collection (Harvard University). The derivation of this line has been reported previously [32]. HUES6 and iPSCs were cultured on a layer of mitotically inactivated (with mitomycin C (Tocris)) mouse embryo fibroblasts (MEFs) in hESC medium. The mouse embryonic fibroblasts (MEFs) used in this study were derived in the laboratory of Prof. Dr. Hans Schöler and have been reported previously [31]. hESC medium consisted of Knockout DMEM (Invitrogen) with 20% Knockout Serum Replacement (Invitrogen), 1 mM beta-mercaptoethanol (Invitrogen), 1% nonessential amino acids (NEAA, Invitrogen), 1% penicillin/streptomycin/glutamine (PAA), freshly supplemented with 5 ng/mL FGF2 (Peprotech). Pluripotent stem cells were split 1:5 to 1:8 every 5-7 days. Colonies were mechanically disaggregated with 1 mg/mL collagenase IV (Invitrogen). 10 mM ROCK Inhibitor (Ascent Scientific) was added for 24 hours after splitting.

smNPC Derivation
For generation of smNPCs from pluripotent stem cells, colonies were detached from the MEFs 3-4 days after splitting, using 2 mg/mL collagenase IV. Pieces of colonies were collected by sedimentation and resuspended in hESC medium (without FGF2) supplemented with 10 mM SB-431542 (Ascent Scientific), 1 mM dorsomorphin (Tocris) for neural induction, as well as 3 mM CHIR 99021 (Axon Medchem) and 0.5 mM PMA (Alexis), and cultured in petri dishes. Medium was replaced on day 2 by N2B27 medium supplemented with the same small molecule supplements. N2B27 medium consisted of DMEM-F12 (Invitrogen)/Neurobasal (Invitrogen) 50:50 with 1:200 N2 supplement (Invitrogen), 1:100 B27 supplement lacking vitamin A (Invitrogen) with 1% penicillin/streptomycin/glutamine (PAA). On day 4, SB-431542 and dorsomorphin were withdrawn and 150 mM Ascorbic Acid (AA; Sigma) was added to the medium. On day 6, the EBs, which showed intensive neuroepithelial outgrowth, were triturated with a 1,000 mL pipette into smaller pieces and plated on Matrigelcoated (Matrigel, growth factor reduced, high concentration; BD Biosciences) 12-well plates at a density of about 10-15 per well in smNPC expansion medium (N2B27 with CHIR, PMA, and AA). For coating, Matrigel was diluted to a final dilution of 1:100 in Knockout DMEM (Invitrogen) prior to coating 500 mL per well of a 12-ell plate overnight. Coated plates were wrapped with parafilm and kept in the fridge for up to 1 month. The first split was performed at a 1:5 to 1:10 ratio on days 2 to 4 after plating. All the remaining splitting ratios were at least 1:10. The higher splitting ratios selected better for smNPC colonies and led to a high purity with fewer splits. After a maximum of 5 splits, cultures were virtually free of contaminating non-smNPCs.
smNPC Culture smNPC were cultured on Matrigel-coated 12-well (Nunc) cellculture plates. smNPC expansion medium consisted of N2B27 freshly supplemented with CHIR, PMA, and AA, with a medium change every other day. Typically, cells were split 1:10 to 1:15 every 5 or 6 days. For splitting, cells were digested into single cells for about 15 minutes at 37uC with prewarmed accutase (PAA). Cells were diluted in DMEM (PAA) for centrifugation at 2006g for 5 minutes. The cell pellet was resuspended in fresh smNPC expansion medium and plated on Matrigel-coated cell culture dishes.

Differentiation of smNPCs
All differentiation experiments were conducted with smNPCs of passage 13 and above. For undirected differentiation, including neurons, astrocytes and oligodendrocytes, it was sufficient to change smNPC expansion medium to N2B27 medium without supplements. For obtaining more homogenously plated cultures, cells were digested to single cells with Accutase after two weeks of differentiation, replated on fresh matrigel-coated plates and further differentiated for at least one week. For better survival, 50 mM dbcAMP (Sigma Aldrich) was added after replating.
For generation of more ventral CNS neurons, including mDANs, smNPC expansion medium was changed 2 days after splitting to N2B27 medium with 100 ng/mL FGF8 (Peprotech), 1 mM PMA, and 200 mM AA. After 8 days in this medium, maturation medium-N2B27 with 10 ng/mL BDNF (Peprotech), 10 ng/mL GDNF (Peprotech), 1 ng/mL TGF-b3 (Peprotech), 200 mM AA, and 500 mM dbcAMP-was used for the maturation of neurons. 0.5 mM PMA was added to this medium for 2 more days. One day after changing to maturation medium, the cultures were split at a 1:3 ratio as small clumps, or single cells after Accutase treatment, or earlier when cultures became overconfluent. Cultures were analyzed after 2 weeks in maturation conditions unless otherwise indicated.
For induction of posterior cells, including MNs, smNPC expansion medium was changed to N2B27 with 1 mM PMA 3 days after splitting. Two days later, 1 mM retinoic acid (RA, Sigma) and 1 mM PMA were added for 8 days. Following one day in maturation medium (N2B27 with BDNF, GDNF, and dbcAMP), cultures were also split as clumps or single cells after Accutase treatment at a ratio of 1:2 to 1:3. Cells were cultured in maturation medium for 2 weeks.
For generation of PNS neurons, smNPCs 2 days after splitting were switched to N2B27 with only CHIR for 2 days. Afterward, 10 ng/mL BMP4 (R&D Systems) was added for 8 days. Splitting and maturation was performed as described for the generation of MNs.
For directed astrocyte differentiation, smNPCs were cultured with 10 ng/ml FGF2 and 10 ng/ml EGF (Peprotech) for 2 days and later switched to N2 medium with 4% FCS (PAA) supplemented with 10 ng/ml CNTF (Peprotech) for at least 2 weeks. Cultures were split using accutase when confluent and replated on fresh Matrigel-coated plates. After withdrawal of CNTF, cells were treated with 500 mM dbcAMP in N2 medium with 4% FCS for at least one week, or could be expanded for several weeks in 4% FCS containing N2 medium using 10 ng/ml EGF before being treated with dbcAMP. After dbcAMP treatment, cells were kept in 4% FCS in N2 for at least one more week.
For mesenchymal neural crest differentiation, smNPCs were cultured with CHIR only for 2 days after splitting and subsequently changed to DMEM (PAA) with 10% FCS and 1% penicillin/streptomycin/glutamine. Cultures were split at a 1:3 ratio when confluent using trypsin (Invitrogen) and cultured on cell culture-treated plastic dishes. Mesenchymal cells derived from smNPCs were differentiated into osteocytes and adipocytes for 14 days, using the Human MSC Functional Identification Kit (R&D Systems). The supplied reagents were used according to the manufacturer's instructions.

Cytotoxicity Experiments
For assessing sensitivity for cytotoxicity of wild type or LRRK2 G2019S mDANs, patient-specific iPSC -derived smNPCs at passage 15 were differentiated as mentioned above. All splitting procedures were performed as single cells using Accutase treatment. After 14 days of differentiation, mDAN cultures were digested to single cells using Accutase and reseeded in maturation medium on Matrigel-coated 48well or 96well plates (Nunc) at 70,000 or 35,000 cells per well. Two days later, medium was changed against N2 medium (DMEM/F12 with 1% N2 supplement and 1% penicillin/streptomycin/glutamine) for six hours to remove antioxidants and enzymes present in B27 supplement. Medium was changed against fresh N2 medium or N2 medium supplemented with 5 mM 6-OHDA, or 10 mM 6-OHDA (Tocris), or 100 nM Rotenone (Sigma). Two days later, cells were fixed and stained for TH and cleaved CASPASE3, as mentioned below.

Transplantation
For analyzing the in vivo differentiation potential of smNPCs, smNPC and mDAN progenitors were transplanted into the midbrain of male NOD.CB17-Prkdc scid /NCrHsd mice (purchased from Harlan; 8 weeks, ,25 g). The latter were differentiated towards mDANs for 8 days as described previously. Before transplantation, the cells were dissociated to single cells for about 15 minutes at 37uC with pre-warmed accutase and resuspended in medium at a density of 5610 4 cells per microliter. For stereotactical transplantation, animals were deeply anesthetized by intraperitoneal injection of 0.017 ml of 2.5% Avertin per gram of body weight and positioned into a stereotatic frame (David Kopf Instruments, model 940). Injection of 3 ml of the cell suspension was performed using a Hamilton 7005KH 5 ml syringe. Franklin & Paxinos mouse brain atlas was consulted for assessing the stereotactic coordinates of the midbrain in relation to bregma (anteroposterior: 23 mm, mediolateral: 61,5 mm, dorsoventral: 24,4 mm below skull).

Immunocytochemistry
For confocal microscopy, cells were plated on Matrigel-coated glass coverslips. Cultures were fixed for 20 minutes with 4% paraformaldehyde (Electron Microscopy Sciences) in PBS (Invitrogen) and washed twice with PBS. Permeabilization and blocking was done in one step using 0.1% Triton X-100 (Sigma Aldrich), 10% FCS, and 1% BSA in PBS for 45 minutes. Plates or coverslips were washed once with 0.1% BSA in PBS and the primary antibodies were applied overnight at 4uC in 1% BSA in PBS. The next day, following one washing step with 0.1% BSA in PBS, secondary antibodies were applied for one hour at room temperature in 1% BSA in PBS. Finally, cells were washed three times with 0.1% BSA in PBS-T (0.05% Tween-20), including a Hoechst counterstaining for nuclei in the second washing step. Cells were mounted in Vectashield mounting medium (Vector Labs) and imaged on a Zeiss PALM/Axiovert fluorescence microscope or a Zeiss LSM700 confocal microscope. If necessary, images were merged using ImageJ and Adobe Photoshop.
To determine the efficiency of differentiation into specific neurons, after 2 weeks in maturation medium, cells were disaggregated and seeded at a density of 5610 4 cells per well in maturation medium on Matrigel-coated 48-well plates. The next day, the cells were fixed and stained, as mentioned above. Cell counting and evaluation of the differentiation efficiency was performed using Cellomics ArrayScan high content imager with the supplied software. 25 fields were taken from each well with a 10X magnification, and the total number of cells was determined by counting the Hoechst-positive nuclei. Three independently differentiated cultures were evaluated for each iPSC line. MN cultures were counted manually from 5 randomly taken pictures at 10X magnification from each well. The

Perfusion, Sectioning and Immunohistochemistry
Two and eight weeks after transplantation, anesthetized animals were intracardially perfused with 50 ml 16PBS following 50 ml 4% PFA/1 PBS solution. The brains were isolated and post-fixed with 4% PFA/1 PBS solution over night at 4uC. A vibratome (Leica VT 1200 S) was used to prepare 40 mm sagittal sections. Permeabilization was performed by using TBS+/+/+ (TBS 0.1 M Tris, 150 mM NaCl, pH 7.4/0.5% Triton-X 100/0.1% Na-Azide/0.1% Na-Citrate/5% normal goat serum) for at least 1 h. Free floating sections were then incubated in TBS+/+/+ containing primary antibodies for 48 h on a shaker at 4uC, followed by 2 h incubation with Alexa-fluorophore conjugated secondary antibodies (Invitrogen) and Hoechst 33342 (Invitrogen) in TBS+/+/+ at room temperature. Quantitative RT-PCR (qRT-PCR) Total RNA was isolated from cultured cells using RNeasy columns (QIAGEN), according to manufacturer instructions, including an on-column DNase digest. Isolated RNA was reverse-transcribed using M-MLV Reverse Transcriptase (USB) with oligo-dT 16 primers (Metabion) for 1 h at 42uC. qRT-PCR was performed on an Applied Biosystems 7500 Real-Time PCR system with SYBR green PCR master mix (ABI) and 56 ng of original RNA equivalents per 20 mL PCR reaction. Cycling conditions were 40 cycles of 15 s, 95uC/60 s 60uC. Relative expression levels were calculated using the 2 22D method, normalized to biological reference samples and using GAPDH and ACTB as housekeeping genes.

Whole Genome Expression Analysis
DNA-free total RNA samples (500 ng) to be hybridized on Illumina human-12 V3 expression BeadChips were processed using a linear amplification kit (Ambion) generating biotin-labeled cRNA (IVT duration: 14 h). This was quality-checked on a 2100 Bioanalyzer (Agilent) and hybridized as recommended and using materials and reagents provided by the manufacturer. In BeadStudio, raw data were background-subtracted and normalized using the ''cubic spline'' algorithm. Differential gene expression was assessed on the basis of thresholds for both expression ratios and statistical significance employing the ''Illumina custom'' algorithm considering standard deviations from replicate beads within each array. Signal intensities below 50% of the detection threshold were arbitrarily trimmed to the value corresponding to 50% of detection. This procedure underestimates expression changes for genes undetectable in the reference sample (or vice versa) but avoids nonsense ratios, such as negative or unrealistically high values.
Karyotype Analysis smNPCs at passage 25 were cultured until confluent. Three hours before chromosome preparation, colcimid (KaryoMAX; Invitrogen) was added to a final concentration of 0.3 mg/mL. After this incubation, the colcemid-containing solution was removed, the cells washed with PBS, and digested to a single-cell suspension with prewarmed Trypsin-EDTA, diluted in DMEM, and collected by centrifugation. The cell pellet was resuspended in 37uC prewarmed 75 mM KCl solution and incubated at room temperature for 7 minutes. The pellet was resuspended in icecold fixation solution (3:1 methanol/acetic acid) while carefully shaking the cell suspension. Once fixed, the cells were collected by centrifugation and carefully resuspended in fresh fixative and incubated for 20 minutes at 4uC. Cells were spread by dropping different dilutions of cells in fixative on glass slides (Menzel Glä ser, Thermo Scientific). The chromosomes were GTG-banded using standard procedures. Metaphase spreads were analyzed on a Zeiss AxioScop. 10 metaphases were analyzed from each line using the Cytovision software (Applied Imaging Corporation).

Generation of Single-cell Clonal Lines
For the generation of single-cell clones, smNPCs were infected with a pLenti CMV -SV40-Blasticidine construct based on the pLenti6/V5 expression system (Invitrogen), which includes a blasticidin-resistance cassette. Virus production was performed in 293T cells using the ViraPower packaging mix (Invitrogen). One 6-cm plate of 293T cells were transfected using FuGENE 6 (Roche) according to the manufacturer's instructions with 2 mg packaging mix and 1 mg expression construct. One day after transfection, medium was changed for N2B27 medium. The following day, the medium supernatant was filtered to remove 293T cells, supplemented with 6 mg/mL protamine sulfate (Sigma), 3 mM CHIR 99021, 0.5 mM PMA, 150 mM AA, and directly used for infection of freshly plated smNPC. The next day, infected smNPCs were washed 4 times with PBS and fed with fresh smNPC expansion medium. Selection with 5 mg/mL blasticidine (PAA) in smNPC expansion medium started 2 days later and was maintained for 2 more weeks.
Blasticidin-resistant smNPCs were digested and triturated to single cells using accutase for 30 minutes and filtered using a 40mm cell strainer (BD Biosciences) to remove remaining cell aggregates. Single cells were counted and seeded at a density of 10 cells per well on a Matrigel-coated well of a 6-well plate, together with approximately 200,000 uninfected smNPCs in expansion medium. Four days later, cells were again selected with 5 mg/mL blasticidin, until only resistant, single colonies remained on the plate that were spotted and marked. Selection was maintained for 1 more week, single colonies were picked, replated on 4-well plates, and expanded under standard smNPC conditions, and blasticidin resistance was continued for 1 more week to exclude surviving non-resistant cells. Once sufficiently expanded, single cell-derived clones were differentiated as described above.

Evaluation of Electrophysiological Function
The transmembrane current and spontaneous activity were recorded from smNPC-derived neurons, after 3 weeks of differentiation according to the mDAN protocol, using the whole-cell configuration of the patch-clamp technique [33]. The patch pipettes were fabricated from borosilicate glass on a PIP-6 pipette puller (HEKA Elektronik, Lambrecht, Germany). When filled with pipette solution, they had tip resistances of 5-7 MV. Recordings were done using a HEKA EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany) and Pulse 8.61 Aqusition Software (HEKA Elektronik, Lambrecht, Germany). Series resistance and pipette and whole-cell capacitance were cancelled electronically. Cells were perfused with a bath solution containing    Figure S11 LRRK2 G2019S increases sensitivity of dopaminergic neurons derived from smNPCs to stress compared with controls. smNPCs were derived from two patient-specific LRRK2 mutant iPSCs, alongside with two ageand sex-matched controls. smNPCs were differentiated into mDANs, replated as single cells and incubated with N2 medium only, or supplemented with 5 mM 6-Hydroxydopamine (6-OHDA), or 10 mM 6-OHDA, or 100 nM rotenone. After two days, apoptotic mDANs were identified by immunostaining for TH and cleaved CASPASE3 (CASP3). Error bars represent variation between two independently stressed wells. (A) LRRK2 mutant mDANs show a higher degree of apoptosis in TH+ cells, as compared to healthy controls. (B) Additional stressors separate the mDAN cytotoxicity phenotype better between LRRK2 mutant and wild-type neurons than withdrawal of antioxidants and neurotrophins alone. *indicates p,0.05 according to the Student's t-test. (JPG) Figure S12 smNPC-derived mDANs are specifically susceptible to oxidative stress. (A) Overview image of a stressed dopaminergic neuron culture after differentiation from smNPCs stressed with 6 OH-Dopa for two days before fixation. Cultures were stained with the indicated markers. Note that most of the cells positive for cleaved CASPASE3 (CASP3) as an indicator for apoptosis are also positive for TH. Only few CASP3+ cells are negative for TH (some indicated by arrowheads). The scale bar indicates 100 mm. (B) Stressed cultures were stained for TH and CASP3 and counted. The graph indicates a high specificity of stress-induced apoptosis for dopaminergic neurons as shown by CASP3 and TH double-positive cells. Under the tested conditions, more than about 80% of the CASP3-positive cells also express TH. The experiment was performed in duplicates. Error bars indicate variance. (TIF)