Fig 1.
Characterization of MN differentiation from ESCs.
(A) Scheme outlining the differentiation protocol. ESCs are plated in N2B27 + FGF for 2 days before being exposed to N2B27 + FGF/CHIR, resulting in the production of NMPs at day 3. Cells are subsequently exposed to RA and SAG to promote differentiation into ventral NPs and MNs. (B, C) Expression of NP (Pax6, Olig2, Nkx2.2, Sox1) and MN (Isl1/2) markers between day 4 and day 7 in differentiating ESCs. (D) RT-qPCR analysis of Irx3, Pax6, Nkx6.1, Olig2, and Nkx2.2 expression from day 3 to day 7 reveals progressive ventralization in response to increasing duration of Shh signaling. Underlying data are provided in S1 Data. (E) MN induction after day 5, revealed by RT-qPCR analysis of Sox1, Ngn2, Isl1, and Tubb3. Underlying data are provided in S1 Data. Scale bars = 40 μm. CHIR, Wnt pathway activator CHIR99021; ESC, embryonic stem cell; FGF, fibroblast growth factor 2; MN, motor neuron; NMP, neuromesodermal progenitor; NP, neural progenitor; N2B27, N2 and B27 media supplements; RA, retinoic acid; RT-qPCR, real-time quantitative polymerase chain reaction; SAG, Smoothened/Shh signalling agonist.
Fig 2.
Reconstruction of transcriptional changes during MN differentiation.
(A) Identification of NP cell states using hierarchical clustering of gene expression profiles of the individual cells. (B) Cell state graph constructed from minimum spanning trees, color coded for the cell populations identified in Fig 2A. Stars indicate start and end cells for the reconstruction of transcriptional changes along pseudotime. Shading of edges between cells indicates how often the edge was used in the reconstruction of gene expression along pseudotime (see S1 Text). (C) Cell state graph color coded for expression levels of Irx3, Olig2, Ngn2, Lhx3, Isl1, and Chat. (D) Inferred changes in gene expression over pseudotime from 9,000 shortest paths connecting start and end cells (stars in B). Each shortest path was resampled to a length of 41 pseudo–time points to enable statistical measurements of gene expression. Cell IDs are color coded according to cell states in (A). Quantification of the global rate of change in gene expression identifies three metastable states (light gray) separated by transition states, during which the rate of change in gene expression is increased (dark gray). Transition phases are defined as intervals along the pseudo-temporal timeline at which the second derivative of the global gene variation is negative, while metastable states are characterized by a positive second derivative. (E) Gene expression profiles along pseudo-time for NP TFs (Irx3, Pax6, Nkx6.1, and Olig2), genes associated with the transition to MNs (Ngn2, Lhx3, and Neurod4) and MN markers (Isl1/2, Tubb3, and Chat). (F) Levels of gene expression for Hes1/5, Olig2, and Ngn2 over pseudotime. Note that Olig2 expression appears biphasic, with up-regulation of Olig2 concommitant to Ngn2 induction and repression of Hes1/5 in the transition phase from NP to MN. Chat, choline acetyltransferase; KSP, K shortest paths; MN, motor neuron; NP, neural progenitor; pMN, MN progenitor; p3, V3 interneuron progenitor; Tubb3, neuronal class III beta-tubulin.
Fig 3.
Olig2 expression is higher in Ngn2-expressing progenitors in the pMN domain.
(A–B″) Staining for Ngn2 (A, B) and Olig2 (A′, B′) merged with Mnx1 (green in A″, B″) in spinal cords at e9.5 (A–A″) and e10.5 (B–B″). (C–D′) Higher magnification images of the spinal cords shown in (A–B″). Red arrowheads indicate nuclei with elevated levels of Ngn2 and Olig2. (E, G) Positive correlation between Olig2 and Ngn2 protein levels in individual nuclei at e9.5 (E) (n = 464 nuclei) and e10.5 (G) (n = 1,078 nuclei). Underlying data are provided in S1 Data. (F, H) Levels of Olig2, Mnx1, and Ngn2 in individual nuclei throughout the pMN domain at e9.5 (F) and e10.5 (H). Plotting Olig2 versus Mnx1 protein levels reveals a clear differentiation trajectory from Olig2-positive pMN cells to Mnx1-positive MNs. Note that high levels of Ngn2 are only observed in cells with high levels of Olig2 expression. In addition, Olig2 protein perdures much longer in Mnx1-positive MNs than Ngn2. Underlying data are provided in S1 Data. Scale bars = 50 μm. e, embryonic day; MN, motor neuron; pMN, MN progenitor.
Fig 4.
Quantification of a fluorescent Olig2 reporter reveals a marked up-regulation of Olig2 prior to MN differentiation.
(A) Design of the Olig2-mKate2 reporter. A 3xNLS-FLAG-mKate2 reporter was fused to the C-terminus of endogenous Olig2 via a T2A self-cleaving peptide. (B) Western blot analysis reveals that the targeted allele shows the same expression dynamics and levels as endogenous Olig2. The targeted allele runs at slightly increased molecular weight due to addition of the T2A peptide (see A). Note that both alleles are targeted in this cell line and, consequently, no protein of wild-type size was detected. (C-F″) Immunofluorescence for mKate2 with Olig2 (C–C″), Isl1/2 (D–D″), Sox1 (E–E″), and Nkx2.2 (F–F″) at day 6 of the differentiations. (G–J) Quantification of protein levels of mKate2 and Olig2 (G, n = 2,851 nuclei), Isl1/2 (H, same dataset as G), Sox1 (I, n = 2,049 nuclei) and Nkx2.2 (J, n = 2,034 nuclei) in individual nuclei. Note the positive correlation between mKate2 and Olig2 and Isl1/2, and the negative correlation between mKate2 and Sox1 and Nkx2.2. Underlying data are provided in S1 Data. (K) Inhibition of Notch signaling using 10 ng/μl DBZ causes an increase of neurogenesis. Immunofluorescent staining for Olig2, mKate2, and Tubb3 in control or after 24 hours DBZ treatment at day 6 of the differentiation. (L) Frequency plots of mKate2 fluorescence intensity obtained by flow cytometry reveal a strong increase in the number of mKate2HIGH cells after 24 hours DBZ treatment. Scale bars = 25 μm. DBZ, dibenzazepine; FLAG, FLAG epitope tag; mKate2, monomeric far-red fluorescent protein Katushka-2; MN, motor neuron; Tubb3, neuronal class III beta-tubulin; T2A, Thosea asigna virus 2A peptide sequence; 3xNLS, 3 copies of a nuclear localization sequence peptide.
Fig 5.
Olig2 and Hes are dynamically expressed in the mouse neural tube.
(A–D) Expression patterns of Ngn2 (green in A), Olig2 (red in A, C, D), Hes1 (red in B, green in C), and Hes5 (green in B, D) in the neural tube at e10.5. Note the low expression levels of Hes1/5 and high expression levels of Ngn2 in the pMN domain (compare A, B). (E) Hes5 (green) expression coincides with the expression of high levels of Pax6 (red) in the intermediate neural tube. (F, G) Hes1 expression (green) is readily detected in both Nkx2.2+ p3 progenitors (red in F) and floor plate cells labelled by Foxa2 expression (red in G). (H–Q′) Time course of Olig2 (blue), Hes1 (red), Hes5 (red), and Ngn2 (green) expression in neural tubes between e8.5 and e10.5. Multiple panels shown for e9.5 reflect developmental progression from caudal to rostral positions along the neuraxis. Hes1 expression appears to recede from the ventral neural tube upon the onset of Olig2 expression at e8.5 (H) and is thereafter absent from most Olig2+ cells (I–L). Olig2 and Hes5 are initially coexpressed (M, N). Over time, Hes5 expression progressively disappears from the pMN domain (N–Q), and Ngn2 concomitantly increases (N′–Q′). Insets show single channel images of the outlined area for the respective markers. Scale bars = 50 μm. e, embryonic day; pMN, MN progenitor; p3, V3 interneuron progenitor.
Fig 6.
Repression of Hes1/5 in the pMN domain depends on Olig2 activity.
(A–D) Expression of Cre (green in A–D), Olig2 (red in A), Ngn2 (red in B), Hes1 (red in C, grey in C′), and Hes5 (red in D, grey in D′) in e10.5 Olig2Cre heterozygous embryos. (E–H) In Olig2Cre/Cre homozygous mutants, Hes1 expands dorsally (G, G′) and Hes5 ventrally (H, H′) into the pMN domain, marked by Cre expressed from the Olig2 locus. The expansion of Hes1/5 coincides with a loss of the high levels of Ngn2 normally seen in the pMN domain. (I) Quantification of Hes1, Hes5, and Ngn2 expression in Olig2Cre heterozygous and homozygous embryos. The overlap between Cre and Hes1/5 significantly increases in Olig2Cre homozygotes, while overlap between Ngn2 and Cre is strongly reduced. Plot shows the mean ±SEM from multiple sections collected from 3–5 embryos for each group. Each section is represented by a single dot, with n = 8–11 for each group. Underlying data are provided in S1 Data. **** p < 0.0001, unpaired t test. (J–S) Electroporation of myc-tagged OLIG2 and an OLIG2-bHLH-Engrailed repressor domain fusion protein in chick neural tubes represses expression of the Hes5 homologues HES5-1–HES5-3 (K–M; Q–R) and the Hes1 homologue HAIRY1 (N, S). “+” indicates transfected side of the spinal cords. Results are representative of >5 successfully transfected embryos collected from two or more experiments. Scale bars = 50 μm. bHLH, basic helix-loop-helix DNA binding domain; Cre, bacteriophage P1 Cre recombinase; e, embryonic day; EnR, Engrailed transcriptional repression domain; EP, electroporation; pMN, MN progenitor.
Fig 7.
Olig2 binds to an evolutionarily conserved element near Hes5.
(A) Identification of an evolutionarily conserved element containing an E-box in the vicinity of the Hes5 genomic locus in chick, mouse, and human (Hes5(e1)). (B) Analysis of Olig2 Chip-Seq data from [63] reveals Olig2 binding sites in the vicinity of the Hes1 and Hes5 genes. The peak corresponding to the Hes5(e1) element is highlighted in red. (C) Electrophoretic mobility shift assays show that both Olig2 and E12 homodimers can individually bind to the Hes5(e1) E-box and do not form any heterodimeric complexes (lanes 1–4). Positions of the different protein complexes are indicated by colored arrows. Binding depends on the E-box, as both proteins fail to bind probes containing an E-box mutation (Hes5(e1ΔE)) (lanes 5–7). Olig2 binding to Hes5(e1) can be abolished by the addition of unlabelled Hes5(e1) probes, but not those containing the E-box mutation (lanes 8–14). (D) Id1 inhibits binding of E12, but not of Olig2 or Ngn2, to the Hes5(e1) element. Olig2, E12, and Ngn2 alone or Ngn2/E12 heterodimers can bind the Hes5(e1) element. Mixing Olig2 or Ngn2 with Id1 does not inhibit their homodimeric binding activities (lanes 2, 5, 8, and 10). In contrast, Id1 strongly inhibits binding of both E12/E12 and Ngn2/E12 complexes (lanes 6 and 10). The addition of E12 without and with Id1 does not affect Olig2 binding efficiency (lanes 2, 4, and 7). ATG, translational initation codon; Chip-Seq, chromatin immunoprecipitation-sequence; E-box, bHLH transcription factor binding site; N2, Ngn2 protein; O2, Olig2 protein.
Fig 8.
The Hes5(e1) element is required for repression of reporter genes in the pMN domain.
(A, B) Co-electroporation of CMV/β-actin::nLacZ and Hes5(e1) reporter plasmids into chick spinal cord. Although electroporation (revealed by β-Gal antibody staining, magenta in [A]) is uniform along the dorsal-ventral axis, expression of the EGFP reporter is confined to intermediate parts of the neural tube (A, B), and little coexpression of Olig2 and EGFP was detected (B). (C) Design of Hes5(e1) and Hes5(e1ΔE) reporters. The Hes5(e1) element was cloned in front of β-globin minimal promoter to drive EGFP reporter gene expression. To test the importance of the E-box in the Hes5(e1) element, critical base pairs for Olig2 binding were mutated (red). (D, E) Co-electroporation of CMV/β-actin::-nLacZ and Hes5(e1ΔE) reporter plasmids into chick spinal cord. In contrast to the Hes5(e1) reporter plasmid, significant coexpression of Olig2 and GFP in the pMN domain is detected (E). Note that E-box mutation reduced the basal activity of the reporter such that longer exposure times were needed to achieve the signal levels seen in the intermediate spinal cord with the nonmutated Hes5(e1) reporter (S8B and S8C Fig). (F) Scatter dot plots display the dorsal-ventral positions (distance from the roof plate) of individual cells expressing the Hes5(e1) and Hes5(e1ΔE) reporters, relative to CMV/β-actin::-nLacZ and Olig2. Results are aggregated from five representative sections taken from five well-electroporated and stage-matched spinal cords. The Hes5(e1ΔE) reporter exhibits a significant ventral shift in its activity and considerable overlap with Olig2 expression (blue dotted box). Box plots include the median and whiskers represent 5th and 95th percentiles. Data points that lay outside the DV scale used to assess these experiments were excluded from this analysis. ** p = 0.0005, Mann-Whitney test; p = 0.6649. Underlying data are provided in S1 Data. (G) EGFP expression in Hes5(e1)-nEGFP whole mount embryos at e10.5. (H–H″) Cryosections of Hes5(e1)-nEGFP embryos at e10.5 assayed for GFP, Olig2, and Hes5. EGFP expression colocalizes with Hes5 expression (H″) but not with Olig2 (H). (I–N) Hes5(e1)-nEGFP expression in Olig2 heterozygous (I, K, L) and homozygous mutants (J, M, N). In Olig2 heterozygotes, little nEGFP expression can be detected in the Olig2 expression domain, resulting in a pronounced gap between the expression domains of EGFP, Nkx2.2, and Hes1 (K, L). By contrast, the EGFP, Nkx2.2, and Hes1 expression domains directly abut each other in Olig2 homozygous mutants (M, N). β-Gal, beta-galactosidase; βGlob, beta-globin; CMV/β-actin::nLacZ, cytomegalovirus/chick beta-actin promoter driving nuclear LacZ gene expression; ΔE, E-box deletion; E-box, bHLH protein binding site; EGFP, enhanced green fluorescent protein; FP, floor plate; GFP, green fluorescent protein; H5(e1), Hes5(e1) genomic element; ns, not significant; pMN, motor neuron progenitor; WT, wild-type.
Fig 9.
Olig2 coordinates patterning and neuronal differentiation.
(A) Proposed model of the Olig2-controlled gene regulatory network. Olig2 not only acts as central organizer for dorsal-ventral patterning in the spinal cord but also controls the rate of MN differentiation through direct repression of Hes TFs. This leads to a higher levels of Ngn2 expression and, consequently, a higher rate of neuronal differentiation in the pMN domain, compared to adjacent progenitor domains. (B) Olig2 is a core component of the Shh-controlled gene regulatory network that patterns the ventral spinal cord [6,67]. (C) Olig2-mediated down-regulation of the Notch effectors Hes1/5 relieves repression of Ngn2 in the pMN domain. (D) Consolidated activities of Ngn2 and Olig2 cause differentiation of NPs to MNs. Olig2 promotes differentiation of MNs through repression of alternative IN cell fates. bHLH, basic helix-loop-helix; IN, interneuron; MN, motor neuron; NP, neural progenitor; pMN, MN progenitor; p2, V2 interneuron progenitor; p3, V3 interneuron progenitor; RA, retinoic acid; Shh, sonic hedgehog; TF, transcription factor; V2, V2 interneuron; V3, V3 interneuron.