Article Figures & Data
Figures
- Figure 1. Synergistic activation of the RARγ and RARβ induces neuronal cell specialization in P19 embryonic stem cells.
(A) Schematic representation of the P19 cell differentiation assay. P19 cells cultured on monolayer are exposed to retinoids during 4 d to induce cell fate commitment; then, they are cultured for six more days on a synthetic medium (Neurobasal, NB) complemented with N2 and B27 (without vitamin A) supplements. (B) Immunofluorescence micrograph of WT P19 cells after 10 d of culture in presence of either ethanol (EtOH: vehicle control), all-trans retinoic acid, the RARα agonist BMS753, the RARβ agonist BSM641, the RARγ agonist BMS961, or the combination of RARβ and RARγ agonists. Cells were stained for the neuronal precursor marker TUBB3 (red) and the marker for mature neurons MAP2 (green). Nuclei were stained with DAPI (blue). (C) Top panel: RT–qPCR revealing the mRNA expression levels of gene markers associated with GABAergic (Gad67), glutamatergic (Glut1), dopaminergic (Th), or cholinergic (Chat) neuronal subtypes in samples treated with the indicated RAR agonists. Bottom panel: RT–qPCR mRNA gene expression levels of the glial fibrillary acidic protein (Gfap), the oligodendrocyte transcription factor 2 (Olig2) and the neuronal precursor marker Tubb3. (D) Immunofluorescence micrograph of P19 Rar-null mutant cells after 10 d of treatment with the aforementioned RAR agonists. (E) RT–qPCR mRNA expression levels of gene markers associated with the aforementioned neuronal subtypes assessed on P19 Rar-null mutant cells. (F) t-Distributed stochastic neighbor embedding analysis of differential gene expression readouts assessed on global transcriptomes performed on WT or Rar-null cells treated with specific agonists (10 d). Differential gene expression has been assessed relative to the ethanol-treated control sample (fold change levels >4). (G) Fraction of up-regulated genes (fold change levels >4) associated with markers corresponding to specialized cells relative to those observed on the gold-standard WT all-trans retinoic acid–treated sample. Fraction levels higher than 95% are only displayed with the heatmap color code (red).
- Figure S1. Generation and validation of P19 cells deficient for each of the RARs.
(A) Gene knockout strategy based on targeted CRISPR/Cas9 DNA cleavage and the incorporation of a puromycin resistance cassette by homologous recombination. (B) Reference of the used plasmids, purchased from Santa Cruz Biotechnology. (C) RT–qPCR assay evaluating the expression of each of the RARs during the first 96 h of P19 cell differentiation driven by all-trans retinoic acid treatment in either WT control cells or the generated Rar-null mutant lines.
- Figure S2. Differential expression response of the neuronal precursor gene Tubb3, the astrocyte-related marker Gfap, and the oligodendrocyte marker Olig2 in the context of the P19 cells deficient for each of the RARs.
P19 Rar mutant lines were treated (10 d) with either the pan-retinoic acid agonist ATRA, or the synthetic agonists BMS753 (Rara-specific), BMS641 (Rarb-specific), and BMS961 (Rarg-specific). Illustrated RT–qPCR assays correspond to the average of four independent experiments.
- Figure S3. Number of differentially expressed genes observed on global transcriptomes and their comparison with genes associated with specialized cells.
(A) Differentially expressed genes (DEGs; relative to ethanol-treated control sample) in the WT and RAR mutant P19 lines and under the listed treatments and at different stages (time-points in days) were computed for a log2 fold change threshold of 2. (B) Number of induced genes in the listed conditions (ID) after 10 d of treatment, and the fraction corresponding to neuronal, astrocyte, or oligodendrocyte precursor cell types, and to the indicated neuronal subtypes. Fractions (in percent) were calculated either relative to the described total number of genes per cell markers listed in Table S1 (*) or relative to the total number of induced genes. Highlighted rows correspond to samples treated with the RARβ + RARγ ligands.
- Figure 2. Temporal gene co-expression analysis during cell specialization driven by retinoid treatment.
(A) Stratification of the temporal transcriptome profiling during WT P19 cell differentiation driven by ATRA treatment. Transcriptomes were assessed on samples collected at 2, 4, and 10 d of treatment. Dashed lines correspond to groups of differentially co-expressed genes (gene co-expression paths; fold change levels >2). The numbers of genes composing each of the co-expression paths are displayed (right). (B) Gene ontology analysis on gene co-expression paths displayed in (A) associated with up-regulated events. (C) Number of genes per co-expression path corresponding to neuronal, astrocyte, or oligodendrocyte precursor cell types assessed during ATRA (left panel), BMS753 (middle panel), and BMS641 + BMS961 (right panel) treatment. (D) Similar to (C) but corresponding to dopaminergic, glutamatergic, and GABAergic neuronal subtypes.
- Figure 3. Active gene regulatory wire reconstruction during cell specialization driven by retinoids.
(A) Structure of the reconstructed gene regulatory network (GRN) displaying differentially expressed genes stratified into four major groups: neuronal cell markers (582 nodes), astrocytes (161 nodes), oligodendrocyte precursors (OPCs: 133 nodes), and a fourth group composed of genes not retrieved in none of the previous classifications (unassigned: 280 nodes). Nodes associated with the neuronal group have been further stratified on dopaminergic (214 nodes), glutamatergic (111 nodes), GABAergic (87 nodes), or unassigned (170 nodes) neurons. For illustration purposes, all edges were removed and replaced by simplified connectors (blue arrows). The color code associated with nodes reflects the differential gene expression levels in WT P19 cells after 2 d of all-trans retinoic acid (ATRA) treatment. (B, C) Number of edges interconnecting nodes retrieved on each of the aforementioned groups. (D) Scheme illustrating all potential types of node states (active: red; repressed: green; and unresponsive: white) and their inter-relationships defined by the illustrated edges (positive regulation: arrow connector; negative regulation: t-shaped connector). “Active edges” (a) correspond to transcriptionally relevant node/edge relationships and are conserved during the analytical processing of the GRN illustrated in (A). (E) Temporal transcription evolution of the reconstructed GRN during WT P19 cell differentiation. Illustrated barplots correspond to the fraction of active edges (as defined in (D)) relative to the total edges (displayed in (B, C)) issued from the treatment with either the ATRA, the RARα-specific agonist BMS753, or the combination of RARβ and RARγ agonists (BMS641 + BMS961). (F) Barplots corresponding to the fraction of active genes after 10 d of treatment with the RARβ and RARγ agonists (BMS641 + BMS961) of P19 Rar-null mutant lines. (G) Number of total active edges retrieved in GRNs issued from 10 d of treatment with the indicated retinoids and over the different P19 lines. Notice that the number of active edges on the Rara(−/−) line treated with the combination of BMS641 + BMS961 agonists leads to similar levels to those observed on the gold-standard WT line treated with the pan-agonist ATRA. (H) Fraction of active edges (relative to the WT line treated with ATRA) associated with markers corresponding to the classification of specialized cells retrieved in (A) relative to those observed on the gold-standard WT ATRA–treated sample. Fraction levels higher than 90% are only displayed with the heatmap color code (red).
- Figure S4. Gene regulatory network (GRN) displaying the differential gene expression response during the 10 d of treatment.
(A) Temporal gene expression response on WT P19 cells driven by the treatment with either ATRA, the RARα agonist BMS753, or the combination of RARβ + RARγ agonists (BMS961 + BMS641) and visualized within a reconstructed GRN. GRN is composed of 1,156 nodes (genes) and 17,914 relationships (edges) collected from transcription factor–target gene databases (Cholley et al, 2018). Furthermore, nodes composing the GRN are stratified into four major groups based on their correspondence to the following cell specialization markers: neurons, astrocytes, oligodendrocyte precursors, and unassigned nodes when no correspondence to such markers is retrieved. Nodes associated with the class “Neuron” are further stratified on either dopaminergic, glutamatergic, GABAergic, or unassigned neurons. Nodes are colored on the basis of their differential response at the three assessed time-points. (B) Differential gene expression response observed after 10 d of treatment with the combination of RARβ + RARγ agonists (BMS961 + BMS641) on the indicated Rar mutant lines.
- Figure 4. Identification of a subset of active edges remained inhibited by the unliganded RARα receptor in P19 WT cells during the synergistic activation of the RARβ and RARγ.
(A) Top panel: Venn diagram revealing the RARα-specific programs corresponding to the common active edges retrieved in WT, Rarg(−/−), and Rarb(−/−) P19 lines treated with the RARα agonist BMS753 (9,328 active edges). Bottom panel: Venn diagram revealing the “common programs” (3,806 active edges) driven by the RARα agonist and those responding to the synergistic activation of the RARβ + RARγ receptors; and a subset of active edges specifically driven by the RARα agonist BMS753 (3,830). This last subset is defined herein as “inhibited programs by the unliganded RARα,” because they remain unresponsive on WT cells treated with the combination of RARβ + RARγ ligands (BMS641 + BMS961), but they are reactivated on the Rara(−/−) line. (B) Top Venn diagram: comparison between the number of genes retrieved in the “inhibited” and the “common” programs highlighted in (A). Bottom Venn diagram: comparison between the number of transcription factors retrieved in the “inhibited” and the “common” programs highlighted in (A). (C) Heatmap illustrating the promoter epigenetic status (repressive mark H3K27me3 and the active mark H3K4me3), the chromatin accessibility (FAIRE), and the transcriptional response (RNA Polymerase II) of genes specific to the “inhibited” or the “common” program, and those shared between these two programs after 2 d of ATRA treatment. (D) Number of genes presenting the indicated promoter epigenetic combinatorial status in the conditions illustrated in (C). (E) Heatmap displaying the differential expression levels for genes associated with the “inhibited”-specific program at different time-points and retinoid treatment. Notice that although most of these genes remained unresponsive when treated with the combination of RARβ + RARγ ligands (BMS641 + BMS961) in WT cells, they are up-regulated on the Rara(−/−) line. (F) Open-chromatin FAIRE sites retrieved on the promoters of the “inhibited programs by the unliganded RARα.” (G) Motif analysis performed on the FAIRE sites presented in (F), revealing the enrichment of the RXRα primary motif. (H) Binding site enrichment analysis performed on the aforementioned FAIRE sites, by comparing with 71 RXR or RAR ChIP-seq publicly available profiles (NGS-QC Generator database: https://ngsqc.org/). Blue bars correspond to the mean fraction of binding sites, and orange bars correspond to the highest fraction assessed over the indicated number of studies.
- Figure 5. Gene regulatory view of the master players inhibited by the unliganded RARα receptor during neuronal cell specialization.
(A) Top 50 transcription factors retrieved within the “inhibited program” ranked on the basis of the fraction of downstream controlled genes within the reconstructed GRN (Fig 4A) (blue line). The orange line corresponds to the fraction of downstream controlled genes predicted on a randomized GRN (master regulatory index [MRI] computed as described in TETRAMER: Cholley et al, 2018). (B) Confidence associated with the TFs’ ranking. Notice that a MRI > 70% presents the most confident P-values. (C) Transcription factors (22) presenting a MRI > 70% and colored on the basis of their promoter epigenetic combinatorial status. (D) Gene co-regulatory view of the 22 TFs, illustrating their most relevant (co-)regulated players and including their known relationships with RXRα and RAR nuclear receptors.
- Figure 6. RARγ- and RARβ-driven cell specialization programs retrieved in P19 ECs are also reactivated during differentiation of mouse embryonic stem cells.
(A) Schematic representation of mouse ES (E14) cell differentiation assay. mES cells cultured on monolayer are exposed to retinoids during 8 d to induce cell fate commitment; then, they are cultured for seven more days on a synthetic medium (Neurobasal, NB) complemented with B27 (without vitamin A) supplements. (B) RT–qPCR after 8 d of differentiation, revealing the mRNA expression levels of gene markers associated with the stem cell markers (Nanog and Sox2), neuronal precursors (Ascl1, Nestin, and Tubb3), and glial cells (astrocyte-related: Gfap; oligodendrocyte-related: Olig2). (C) RT–qPCR after 15 d of differentiation, revealing the mRNA expression levels of gene markers associated with the neuronal marker Tubb3, the glial cell–related markers Gfap and Olig2, and the neuronal subtype markers Gad67 (GABAergic), Th (dopaminergic), and Tph2 (serotonergic). (D) Immunofluorescence micrograph of mES cells after 15 d of culture in the presence of either ethanol (EtOH: vehicle control), all-trans retinoic acid, the RARα agonist BMS753, or the combination of RARβ and RARγ agonists. Cells were stained for the neuronal precursor marker TUBB3 (red) and the marker for mature neurons MAP2 (green). Nuclei were stained with DAPI (blue). (E) RT–qPCR mRNA expression levels measured in mES cells (15 d of differentiation) corresponding to the top 22 master TFs identified in P19 cells (Fig 5). Differential gene expression is expressed relative to the expression levels observed in the presence of the ethanol control sample after 15 d of differentiation. The dashed red line demarcates a fold change threshold value of 3. (F) Number of overexpressed TFs under the various retinoid treatments computed at three different fold change thresholds (Log2). (G) Subset of the “inhibited program” retrieved in P19 cells, revealing the gene co-regulatory network associated with Prdm8. (H) RT–qPCR mRNA expression levels measured in mES cells (15 d of differentiation) corresponding to the different downstream targets of Prm8 revealed in P19 cells. The dashed red line demarcates a fold change threshold value of 3.
Supplementary Materials
In this Issue
Related Articles
Cited By...
- No citing articles found.