(A) Top: Schematic of 14-day differentiation protocol indicating the timing and use of different small molecules (see for details). Bottom: qRT-PCR analysis for developmental markers. Results represent the averages of 3 independent differentiations and 3 technical replicates per time point. Values are normalized against the time point of highest expression for each of the genes.
(B) Clustering of all regions that exhibit significant differential DNAme during MN differentiation. DMRs were defined based on a posterior probability ≤ 0.05 for an absolute DNAme difference >=0.2, requiring a minimum of 4 CpGs. Numbers and black bars indicate different DMR clusters. See
(C) Subset of gene set enrichment analysis for each cluster in B. See .
(D) WB analysis of 3A and 3B in the parental HUES-64 line (WT), two 3A-KO lines (#138 and #139), one 3B-KO line (#44) and the double 3A/3B-KO line (#21) in stem cell (d0) and MN (d14) stages. The predominat 3B isoform is abolished in 3B-KO cells. GAPDH used as a loading control.
(E) Percentage of propidium iodide (PI)+ apoptotic cells measured by FACS on d14 of MN differentiation presented as means ± SEM; ANOVA (***p<0.001).

(A) Representative ICC images of WT, 3A-KO, 3B-KO and 3A/3B-KO cultures at d16 of MN differentiation with the neuronal marker TUJ1 (red) and the MN markers ISL and ChAT (green). Nuclei stained with DAPI (blue). Scale bar: 50μm.
(B) Quantification of ISL, ChAT and TUJ1+ MN percentages across cell lines and the parental isogenic control at d16 presented as means ± SEM; ANOVA (***p<0.001), n.s: not significant.
(C) WB of ChAT and TUJ1 protein in 3A, 3B-KO and isogenic WT d14 differentiation cultures. GAPDH used as a loading control and fold increase values relative to WT are annotated under each lane.
(D) HPLC quantification of secreted acetylcholine in WT and 3A-KO d25 differentiation cultures presented as means ± SEM; 2-tailed-t-test (*p<0.05).
(E) WB of CHAT protein in 3A-KO and isogenic WT MN cultures over 40 days of differentiation. GAPDH used as a loading control.

(A) Cell lines were differentiated in parallel and samples collected for WGBS and RNA-Seq analysis on d0, 6 and 14.
(B) DMRs (p ≤ 0.05, methylation difference ≥ 0.2) that arise between the WT and 3A-KO during MN differentiation broken down into 4 groups: Differentiation-associated DMRs (diff) that occur between individual time points in the WT differentiation but are impaired in the KO, as well as non-differentiation-associated DMRs (non-diff) that exclusively occur in 3A-KO. These 2 groups are subdivided in DMRs that pre-exist in the pluripotent state and ones that occur only over the differentiation.
(C) DNAme levels based on WGBS of non-preexisting DMRs that arise during WT differentiation and are de-regulated in the 3A-KO (left) and DMRs that arise exclusively in the 3A-KO differentiation (right).
(D) Percentage of differentiation-associated DMRs that fail to gain DNAme in the 3A-KO overlapping with different genomic features: CGI – CpG islands, CGIshore – CpG island shore, H3K27ac tissues, H3K4me3 tissues, DNAse tissues– Union of H3K27ac or H3K4me3 peaks and DNAseq peaks across more than 70 distinct cell and tissue types from the REMC and ENCODE projects, TF binding sites in HUES-64 for OCT4 and NANOG.
(E) Subset of gene sets enriched in the non-preexisting hypomethylated DMRs. See .
(F) Fraction of differentially expressed genes (q-value ≤ 0.01, absolute log₂ fold change ≥ 1, minimum expression ≥ 10 RPKM) arising in the differentiation only or between the WT and KO differentiation stages. See .
(G) Clustering of all genes differentially expressed between the WT and KO across the differentiation time course (z-scores). See .
(H) Subset of gene set enrichment analysis for the MGI database of mouse tissues across development. Results are shown as odds-ratios significant at a BH corrected p-value below 0.05 (fisher’s exact test) for expression clusters 1–3 in G. See .
(I) All DMRs (p ≤ 0.05, methylation difference ≥ 0.2) that arise between the WT and 3B-KO during MN differentiation broken down into 4 groups: Differentiation associated DMRs (diff) that occur between individual time points in the WT differentiation but are impaired in the KO as well as non-differentiation associated DMRs (non-diff) that exclusively occur in 3B-KO. These 2 groups are subdivided in DMRs that pre-exist in the pluripotent state and ones that occur only over the differentiation.
(J) Overlap of all DMRs detected in the WT differentiation (Diff), between the WT and the 3A-KO differentiation (3A-KO) and the WT and 3B-KO differentiation (3B-KO).
(K) Overlap of all differentially expressed genes detected in the WT differentiation (Diff), between the WT and the 3A-KO differentiation (3A-KO) and the WT and the 3B-KO differentiation (3B-KO). See .
(L) Ratio of the number of DMRs and differentially expressed genes detected in WT, 3A-KO, and 3BKO differentiations.
(M) Expression levels (z-scores) of TFs that are differentially expressed in the 3A-KO and 3B-KO differentiation compared to the WT.
(N) Expression levels (z-scores) of all TFs that are differentially expressed between the WT and 3A-KO differentiation

(A-C) Single-cell RNA-Seq showing HES1+, PAX6+ cells and HES1/PAX6 mRNA ratio along the transcriptionally-segregated WT (circled) and 3A-KO cultures at d6 of MN differentiation.
(D) WB of HES1 and PAX6 protein in 3A-KO and WT cultures. GAPDH used as a loading control. The ratio of the signal was quantified using Fiji.
(E-F) Percentage of PAX6+ cells in 3A-KO and WT cultures quantified by ICC over MN differentiation presented as means ± SEM, 2-tailed t-test (*p<0.05, **p<0.01, ***p<0.001). Scale bar: 50μm.
(G) tSNE analysis of single cell transcriptional profiles of WT and 3A-KO d14 cultures grouped into 10 distinct cell sub-populations. WT (circled) segregated from 3A-KO cells.
(H) Dendrogram of hierarchical clustering of the 10 existing subpopulations in WT and 3A-KO d14 cultures. (I) Distribution of neuronal (ASCL1, SYP) and non-neuronal (FOXA2, SOX2) genes in the tSNE space accross WT and 3A-KO cultures.
(J) Single-cell expression levels of key neuronal markers among the WT and 3A-KO subpopulations identified by scRNA-seq shown as z-scores.
(K) Fraction of cells classified as neuronal and non-neuronal in the WT and 3A-KO condition at d14.
(L) Expression pattern (z-scores) of floor plate and glial genes across the subpopulations defined in G.
(M) tSNE plot showing the distribution of WT (circled) and 3A-KO cells expressing high levels of ARX, SHH or both at d14.
(N) Pie-charts indicating the percentages of ARX+/SHH+/FOXA2+ floor plate cells in 3A-KO and WT cultures.
(O) Representative ICC images of WT and 3A-KO cultures for FOXA2+ (green) and TUJ1+ (red) cells. Nuclei stained with DAPI (blue).
(P) Percentages of TUJ1+ neurons and FOXA2+ floor plate cells in WT and 3A-KOs presented as means ± SEM, ANOVA p-values 3.565e-41 (% FOXA2+ cells), 1.462e-22 (%TUJ1+ cells).
(Q) Ratio of ChAT/FOXA2 protein levels analyzed by WB. GAPDH used as a loading control.
(R) Schematic showing the combination of TFs guiding distinct progenitor domains in the developing spinal cord. The dorso-ventral axis (D-V) is shown on the left. Sonic hedgehog (Shh) and BMP/TGFb signaling drive this patterning. FP: floor plate.

(A) Integrated analysis results of DMRs located within 100kb of a gene differentially expressed between WT and 3A-KO cells. Heatmaps of DNAme (left) and gene expression (middle) levels shown as z-scores. Selected TFs are highlighted on the left. If multiple DMRs are associated with the same gene, the average DNAme level was plotted. Right: Associated gene expression and genomic feature annotation, black bars indicate DMR overlapping with an H3K27ac, H3K4me3 or DNAseq peak in at least one of 70 distinct cell and tissue types profiled by the REMC or ENCODE consortium.
(B-C) High-resolution view of the PAX6 (chr11 31,751,210–31,898,098) and the ARX locus (chrX 25,001,926–25,040,681) with DNAme levels shown as black (WT) and red (KO) dots. Each dot represents one CpG, with the position on the y-axis indicating the DNAme level (bottom: 0%, top, 100%). Black boxes indicate DMRs arising during the WT differentiation process (diffDMRs), red boxes indicate DMRs identified DMRs between WT and KO. Grey bars in the EZH2 track indicate EZH2 binding sites in ESC, grey bars in sgRNA track indicate binding sites of sgRNAs. Grey box and heatmap shows the DNAme level of all CpGs covered with at least 5 reads in the indicated DMR across all conditions.

(A) Experimental schematic: PAX6::GFP was re-expressed from d0 to d6 (mimicking PAX6 in WT cells) using a Dox-inducible system in 3A-KO cultures.
(B) Top panels show green fluorescent live cell images and bottom panels show brightfield images of d6 and d8 MN cultures infected with PAX6:GFP. Scale bar: 50μm.
(C) WB of PAX6 protein in 3A-KO cells infected with TETi-PAX6::GFP and rtTA lentivirus.
(D) Quantification of ISL+, ChAT+, and FOXA2+ cells over (DAPI+) in PAX6::GFP transduced and non-transduced 3A-KO cells on d16. Red lines indicate the average percentage in WT cultures. Data are presented as means ± SEM; ANOVA (***p<0.001).
(E) Top, illustration of the targeted DNAme editing strategy based on a catalytically inactive dCas9 fused to 3A. Bottom, experimental schematic.
(F) WB analysis of PAX6 in WT and 3A-KO lines non-transduced and transduced with dCas9–3A-BFP and PAX6 or ARX-sgRNA-RFP viruses. GAPDH used as a loading control.
(G) Quantification of ISL+, ChAT+ and FOXA2+ cells over (DAPI+). Red lines indicate the average percentage in WT cultures. Data are represented as means ± SEM; ANOVA (*p<0.05, **p<0.01, ***p<0.001), n.s: not significant.
(H) PAX6 gene. Top: Grey bars indicate regions targeted by amplicon sequencing; red bars indicate binding sites of sgRNAs. Bottom: Heatmaps indicate average DNAme level of each targeted DMR based on amplicon sequencing (AMP) and WGBS for the WT, 3A-KO and 3A-KO with dCas9–3A and two sgRNAs against the PAX6 and ARX DMRs at d6 and d14.
(I) Dendrogram based on hierarchical clustering of DNAme levels across all DMRs in WT and 3A-KO differentiating cultures.
(J) DNAme levels based on WGBS of non-preexisting DMRs that arise during WT differentiation and are de-regulated in the 3A-KO (left) and DMRs that arise exclusively in the 3A-KO differentiation (right) for the WT, 3A-KO and the 3A-KO together dCas9–3A dual locus targeting (sgPAX6+sgARX).

(A) 2-photon microscopy images of d40 MNs derived from 3A-KO and isogenic control ESCs. MNs are labeled by Dextran Texas Red for morphological and electrophysiological studies. Scale bar: 50μm.
(B) Analysis of average neurite length in d40 MNs. Data are means of independent differentiations (n=4, m=31 ± SEM), 2-tail t test (p=0.0298).
(C) Sholl analysis of d40 MNs. Data are mean of independent differentiations (n=4, m=31 ± SEM), Mann Whitney test (p<0.05).
(D) Western blot analysis of PSD95 in d0 to d40 3A-KO and WT MN cultures. GAPDH used as a loading control.
(E-F) MEA-based activity analysis of the mean firing rate and the average burst duration in 3A-KO and isogenic control cultures over 40 days in vitro. Data are mean of independent differentiations (n=3, m=12 ± SEM); 2-way ANOVA (E, p=0.0008; F, p<0.0001).
(G) Activity of a typical WT (left panel, blue trace) and 3A-KO (right panel, red trace) MN recorded in current clamp are shown below depolarizing current steps that evoked activity (top traces) and the frequency plot of the evoked action potentials (middle traces). Scale bar in control current trace: 400pA. Scale bar in KO current trace: 200pA. Scale bars for all voltage traces: horizontal: 1s; vertical: 50mV.
(H) Current at firing onset in WT and 3A-KO MNs. Typical voltage threshold of action potentials and corresponding I-on values are shown for a WT (blue trace, left panel) and 3A-KO (red trace, right panel) MNs. Horizontal scale bar: 0.5s; vertical: 50mV apply to both traces.
(I) Persistent inward current in KO cell line (red trace) was evoked at a significantly hyperpolarized voltage compared to the WT (blue trace). The average onset voltage (voltage at which the downward deflection in the current traces is first initiated) was −55mV in KO and -51 in WT, similar to the examples shown.
(J) The AP height (size) was greater in KO respect to WT MNs (example trace in red). This was driven by a faster rate of rise in the upswing. The rate of fall was unchanged. Scale bar vertical: 10mV; horizontal: 0.1s. See for measurements and statistics.