Highly efficient induction of functionally mature excitatory neurons from feeder-free human ES/iPS cells

Cortical excitatory neurons (Cx neurons) are the most dominant neuronal cell type in the cerebral cortex, which play a central role in cognition, perception, intellectual behavior and emotional processing. Robust in vitro induction of Cx neurons may facilitate as a tool for the elucidation of brain development and pathomechanism of the intractable neurodevelopmental and neurodegenerative disorders including Alzheimer’s disease, and thus potentially contribute to drug development. Here, we report a defined method for efficient induction of Cx neurons from the feeder-free-conditioned human embryonic stem cells (ES cells) and induced pluripotent stem cells (iPS cells). By using this method, human ES/iPS cells could be differentiated into ~99% MAP2-positive neurons by three weeks, and these induced neurons, within five weeks, presented various characteristics of mature excitatory neurons such as strong expression of glutamatergic neuron-specific markers (subunits of AMPA and NDMA receptors and CAMKIIα), highly synchronized spontaneous firing and excitatory postsynaptic current (EPSC). Moreover, the Cx neurons showed susceptibility to the toxicity of Aβ42 oligomers and excitotoxicity of excessive glutamates, which is another advantage in terms of toxicity test and searching for the therapeutic agents. Taken together, this study provides a novel research platform for the study of neural development and degeneration based on the feeder-free human ES/iPS cell system.


Introduction
Cortical excitatory neurons (Cx neurons) account for approximately 80% of all neuronal cells in the cerebral cortex, in which they play a central role in cognition, perception, intellectual behavior and emotional processing [1]. Using glutamate as their neurotransmitter, Cx neurons, also known as projection neurons, are incorporated in high-order neuronal networks, and their activity is locally Moreover, previous studies demonstrated that concentrating neurons with a region-specific identity made it possible to recapitulate disease-associated in vitro phenotypes [2][3][4]. Thus, a robust induction of Cx neurons in vitro holds a great promise for high-throughput screening of therapeutic candidates for these diseases.

Human embryonic stem cells (ES cells) and induced pluripotent stem cells (iPS cells) have potentials
for unlimited proliferation and differentiation into all the three germ layers [5,6]. Based on pioneer studies on neural differentiation of mouse and human ES/iPS cells [7,8], induction of Cx neurons from human ES/iPS cells have been reported from various groups [2,[9][10][11][12][13][14][15][16]. However, there remain several technical and biological limitations for these methods. For example, first, the reported methods used on-feeder human ES/iPS cells as a starting material, in which the maintenance of on-feeder human ES/iPS cells requires heterogenous feeder cells such as primary mouse embryonic fibroblasts (MEFs) or STO feeder cells [17]. Feeder cell preparation including a large-scale expansion and a mitotic inactivation necessitates significant effort. Moreover, due to the difficulty in completely elimination of the feeder cells, remaining feeders may hamper successful differentiation of a directed linage(s).
Second, several reported methods utilized 3D culture system. Direct floating culture of ES/iPS cells, so called embryoid body (EB) [18,19], neurosphere [2] or organoid [14,15], usually results in a population of differentiated cells with developmental heterogeneity (neural stem cells located in the center, and neurons in the edge). Moreover, the cell-autonomous formation of anterior-posterior and dorsal-ventral axes in floating cell aggregates during 3D culture is frequently observed. Importantly, these 3D methods necessitate a long-term differentiation culture (over 2~3 months) in order to obtain functional mature neurons, which may be costly and not favorable for the high-throughput screening.
To circumvent these limitations, we here report a defined method for the direct differentiation of Cx neurons from feeder-free human ES/iPS cells. By the step-by-step optimization of the differentiation procedures, we demonstrated a successful induction of Cx neurons with high efficiency (~ 100%).

qRT-PCR
RNA extraction, reverse transcription, and qRT-PCR was performed as described previously [26]. The level of ACTB expression was used as an internal control for normalization. As a control of cerebral cortex, we used total cDNA obtained from an adult human sample.

Immunocytochemistry
Immunocytochemistry was performed as described previously [26]. For high-content quantitative immunocytochemical analysis, we used IN Cell Analyzer 6000 (GE Healthcare) as described previously [3]. The detailed analysis protocol using IN Cell Analyzer 6000 is available upon request.
Information of primary and secondary antibodies used in this study are available from the corresponding authors.

ELISA
Enzyme-linked immunosorbent assay (ELISA) was performed as described previously [27]. In brief, we collected culture media of Cx neurons. The collected media were briefly centrifuged to remove insoluble material and can be kept at -80℃ until the analysis. The remaining cells were lysed in RIPA buffer (Wako) and protein concentration was measured by BCA Protein assay (Pierce). Aβ40 and Aβ42 levels in the media were measured using commercial kits, Human

Electrophysiology and Ca imaging
Microelectrode array (MEA) recording was performed using the Maestro system (Axion Biosystems) as described previously [28]. In brief, 12-well MEA plates were pre-coated using Poly-L-Lysine and Laminin, and neural progenitors (on 15 div) were subsequently plated onto the electrode area in the MEA plate. Data were acquired using a sampling rate of 12.5 kHz and filtered using a 200-3000 Hz Butterworth band-pass filter. Detection threshold was set to +6.0 × SD of the baseline electrode noise.
Spike raster plots were analyzed using Neural Metric Tool (Axion Biosystems). The spike count files generated from the recordings were used to calculate the number of active electrodes (defined as an electrode which has an average of more than 5 spikes/min) in each well, the average per-active electrode mean firing rate (MFR or spikes/min) and the standard deviation of the average per-active electrode MFR. The data from the initial 3 min in each data file were omitted to enable the activity to stabilize in the Maestro, and 10-15 min of activity was subsequently recorded.

Statistical Analysis
All data were expressed as means ± SD. Statistical significance of differences was analyzed with the Welch's t-test. Differences of P < 0.05 were expressed as * , P < 0.01 as ** , and P < 0.001 were as *** , which were considered statistically significant.

Efficient induction of neural progenitors with a cerebrocortical identity from feeder-free human iPS cells
Initially, we sought to devise an optimized induction method of SOX1(+) incipient ectodermal cells from 201B7 healthy-control human iPS cells [6] cultured in a feeder-free condition using AK02N as a culture media and iMatrix-511 as a coating materials [20,24]. Based on the principle methodology described in previous studies [2,[9][10][11][12][13]29,30], we testified a 7-days culture of iPS cells using three media, such as N2B27 [25], MHM/B27 [2] and GMEM/KSR [31] without supplementation of any chemical compounds (see Material and Methods). FACS analysis using cell-permeable antibodies revealed that, among the culture media, GMEM/KSR medium resulted in the highest yield of SOX1(+) cells, compared to those of N2B27 and MHM/B27 media (Supplementary Figure 1A).
Next, we performed simultaneous quantification of SOX1(+) cells and OCT4(+) pluripotent cells since the remaining pluripotent cells may hamper the subsequent neuronal differentiation due to their high proliferation rate and unpredictable differentiation into other lineages. In addition, we tested supplementation of a BMP4 inhibitor, DMH1, and TGF-β inhibitor, SB431542, to prevent the cells from the non-ectodermal differentiation [7]. When using the N2B27 medium, although DMH1 and SB431542 treatment ameliorated the SOX1(+)/OCT4(-) rates, we observed approximately 30-60% remaining OCT4(+) cells (Supplementary Figure 1B). On the other hand, when using the GMEM/KSR medium, we found higher rates of SOX1(+)/OCT4(-) cells and very low rates of OCT4(+) cells (~ 5%) in the combination with DMH1 and SB431542 treatment (Supplementary Figure 1C). Therefore, we decided to focus on the method using the GMEM/KSR medium supplemented with DMH1 and SB431542 during the initial phase of induction (Left part of Figure 1A) for the further analyses. Using total RNA derived from the differentiated cells at 12 div (days in vitro) and the undifferentiated iPS cells, we compared the gene expressions of pluripotency (NANOG and  Figure 2B). Collectively, these data demonstrated that our optimized method enabled a robust induction of neural progenitors with an identity of cerebral cortex.

Maturation culture of neural progenitors with a CDK inhibitor
By single-cell dissociation and subsequent replating of neural progenitors on 15 div, we next attempted further differentiation of these cells into functional mature neurons (Right part of Figure   1A). Based on our preliminary results that the BrainPhys-based medium, not Neurobasal-based medium, enhanced neuronal survival and spontaneous firing activity of the differentiated cells evaluated by Ca imaging (data not shown), we utilized the BrainPhys medium supplemented with GDNF, BDNF, ascorbic acid (AA) and dbcAMP for neuronal survival and maturation. We also supplemented a γ -secretase inhibitor DAPT and CDK4/6 inhibitor PD0332991 for further neuronal differentiation through the promotion of cell cycle exit [16]. Since prolonged γ -secretase inhibition may impair neuronal function [32], DAPT was withdrawn on 21 div. For the ease of further analyses, we initially compared three coating conditions before replating of neuronal progenitors as follows: Poly-L-Lysine/Laminin, Matrigel and Laminin only. Six days after replating, we found the least number of cell aggregations in the Poly-L-Lysine/Laminin condition (Supplementary Figure 3A).
Using the optimized BrainPhys medium and coating condition, we observed a time-dependent increase of SYN1, DLG4 (PSD-95) and CAMK2A gene expression upon 5-6 or 7-8 weeks after differentiation of iPS cells ( Figure 1C). We also confirmed the increasing of glutamate ionotropic receptor genes essential for functional maturation of glutamatergic neurons (Supplementary Figure   3B), and Alzheimer's disease-associated marker Tau (MAPT) and its postnatal-specific 4R isoform (Supplementary Figure 3C). In addition, immunocytochemical analysis revealed that most of the differentiated cells (5 weeks after differentiation) were positive for MAP2, Foxg1 and NeuN ( Figure   1D). Given these results, the differentiated cells for 5 weeks or more are hereinafter referred to as the Cx neurons. Moreover, we discovered that the Cx neurons (10 weeks after differentiation) showed dot-like immunoreactivity of synaptic markers including SynI, vGluT1, vGluT2 and Homer1, implying the formation of functional synapses ( Figure 1E).

Electrophysiological analyses of the Cx neurons
To ascertain the formation of functional synapses and local circuits, we performed Calcium (Ca) imaging analysis of the Cx neurons using a fluorescent Ca indicator Fluo-8 (Material and Methods).

Susceptibility to toxic agents
To explore the glutamatergic neuron-specific susceptibility of the Cx neurons to the known neurotoxic agents, we initially tested the supplementation of excessive L-glutamate (L-Glu) inducing cell death by excessive excitotoxicity through glutamate receptors [33]. As shown in Figure 3A Next, we addressed the applicability of the Cx neurons for the in vitro modeling of Alzheimer's disease. Supplementation of Aβ42 oligomer in the Cx neuron culture ( Figure 4A) resulted in a dose-dependent decrease in neurite length per cell ( Figure 4B) and WST-8 bioreduction activity ( Figure 4C). Furthermore, we assessed endogenous secretion of Aβ40 and Aβ42 from the Cx neurons.
Supplementation of DAPT dramatically decreased the secretion of Aβ40 ( Figure 4D). Using the Cx neurons differentiated from a familial Alzheimer's patient-derived iPS cell line, PS1 (A246E) [22], we found that disease-specific phenotypes such as decreased Aβ40 secretion ( Figure 4E, left), increased Aβ42 secretion ( Figure 4E, center) and thus increased ratio of Aβ42/Aβ40 ( Figure 4E, right) were recapitulated. Thus, we demonstrated that utilizing the Cx neurons were advantageous for recapitulating region-specific neurological pathology.

Reproducibility of the induction method in multiple human ES/iPS cell lines
Lastly, we assessed the reproducibility of the induction method using other human ES/iPS cell lines with different origins, such as 1210B2 and KhES1 (Supplementary Figure 5A). We showed that these two lines were successfully differentiated into MAP2(+) positive neurons at high efficiencies comparable to that of 201B7 (data not shown). Moreover, using the electrophysiology analysis, we demonstrated that the Cx neurons derived from these lines exhibited a maturity which was susceptible to the toxicity test of L-Glu (Supplementary Figure 5B) and Aβ42 oligomer (Supplementary Figure   5C).

Discussion
In the present study, we report the defined induction method of Cx neurons from the feeder-free human ES/iPS cells. We highlight three advantages of this method as following: (i) High efficiency of neuronal induction with a region (cerebral cortex)-specific identity, which is important for recapitulation of disease-specific phenotypes [2][3][4]. (ii) High reproducibility in multiple cell lines, which is essential for unbiased comparison and robust applicability for disease-specific lines. In particular, since a major part of underlying genetic/epigenetic mechanisms of sporadic diseases remains unclear, we believe that in vitro systems with high purity and reproducibility could tackle this issue. (iii) Ease of the method. Our initial motivation was to explore the ease-of-use for high-throughput screening using human ES/iPS cells. To this end, we exploited feeder-free human ES/iPS cells, and also, succeeded in shortening the induction period required for obtaining functional mature neurons (upon 5 weeks). This makes it possible to perform cost-effective screening of therapeutic candidates using Cx neurons.
Amyloid beta, including Aβ40 and Aβ42, is the major components of senile plaques found in the brain of Alzheimer's disease patients. Whether and how amyloid beta causes neurodegeneration in patients' brain are yet unclear and under debate, our data suggested that Cx neurons induced by our method could be directly applicable for the elucidation of this mechanism in vitro as the endogenous Aβ