Dual transcriptional activities of PAX3 and PAX7 spatially encode spinal cell fates through distinct gene networks

Abstract

Understanding how transcription factors regulate organized cellular diversity in developing tissues remains a major challenge due to their pleiotropic functions. We addressed this by monitoring and genetically modulating the activity of PAX3 and PAX7 during the specification of neural progenitor pools in the embryonic spinal cord. Using mouse models, we show that the balance between the transcriptional activating and repressing functions of these factors is modulated along the dorsoventral axis and is instructive to the patterning of spinal progenitor pools. By combining loss-of-function experiments with functional genomics in spinal organoids, we demonstrate that PAX-mediated repression and activation rely on distinct cis-regulatory genomic modules. This enables both the coexistence of their dual activity in dorsal cell progenitors and the specific control of two major differentiation programs. PAX promote H3K27me3 deposition at silencers to repress ventral identities, while at enhancers, they act as pioneer factors, opening and activating cis-regulatory modules to specify dorsal-most identities. Finally, we show that this pioneer activity is restricted to cells exposed to BMP morphogens, ensuring spatial specificity. These findings reveal how PAX proteins, modulated by morphogen gradients, orchestrate neuronal diversity in the spinal cord, providing a robust framework for neural subtype specification.

Fig 1. BMP signaling-dependent dorso-to-ventral gradient of PAX3/7 transcriptional activity revealed by the P34::tk::LacZ reporter.

(A) DV expression patterns of neural progenitor (NP) and interneuron (IN) TF markers in the developing spinal cord. (B) Immunostaining for β-galactosidase (β-gal; red/grey), GFP (blue/grey) and either OLIG3 (green/grey) or pSMAD1/5/9 (green/grey) on transverse sections of E9.0 and E9.5 P34::tk::LacZ; Pax3^+/GFP spinal cords at brachial level. Black and white panels show magnified views of the boxed region. (C) Quantification of the βgal, GFP and OLIG3 (i) or pSMAD1/5/9 (ii) signal intensity (in arbitrary units, AU) along the DV axis of the neural tube, expressed as the percentage of the neural tube length (bar plots: mean ± s.e.m). (D) Turquoise fluorescence (white) and immunostaining for GFP (blue) and βgal (white) on transverse sections of chick embryos 24 h post-electroporation (hpe) with the indicated constructs. Only the electroporated side of the neural tube is shown (E) Quantification of βgal or Turquoise signal intensity (AU) along the DV axis of the neural tube of chick embryos 24 hpe with the indicated constructs, expressed as a percentage of the neural tube length (mean ± s.e.m). (F) Immunostaining for β-gal (red/grey) and GFP (green) on transverse sections at the brachial level of E9.5 P34::tk::LacZ; Pax3^+/GFP and Pax3^GFP/GFP embryos. (G) Immunostaining for GFP (blue) and βgal (white) on transverse sections of chick embryos 24 hpe with the indicated constructs. Only the electroporated side of the neural tube is shown. In all images, white dotted lines outline neural tubes. Scale bars: 50 µm. The data underlying this figure can be found in S7 Table.

Fig 2. Neuronal diversity within the spinal cord of mouse embryos following modulation of PAX3 and/or PAX7 transcriptional activity.

(A) Immunostaining for LHX2, FOXD3, ISLET1/2, LBX1 and LMX1B (all in white), GFP (green), and DAPI-stained nuclei (blue) on transverse sections of the brachial region of E11.0–E11.5 mouse embryos with the indicated genotypes. Only half of the neural tube is shown. Asterisks indicate loss of a neuronal subtype; arrowheads denote a reduced number of cells of a given identity; arrows point to cells with ectopic dorsal position. “M” denotes a mixed population of INs displaying V1 or dI2-like characteristics. White dotted lines outline neural tubes. Scale bars: 50 µm. (B) Quantification of cells positive for the indicated TFs (dots: values per transverse section, normalized to those of heterozygous littermates to account for stage variations across littermates, dot shapes: independent embryos; bar plots: mean ± s.e.m.; Mann–Whitney U test: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). The data underlying this figure can be found in S7 Table.

Fig 3. Expression, activity, and function of PAX3 and PAX7 during spinal neuronal diversity generation in organoids treated or not with BMP4.

(A) Schematic of the differentiation protocol used to generate spinal organoids from mESCs. BMP4-free organoids (∅ condition) are enriched in dp4–dp6 progenitors by day 5 and their associated interneurons (INs) by day 7. BMP4 treatment from day 3 to day 4 shifts differentiation toward dp1–dp3 progenitors and corresponding INs. Only a few ventral V0 and V1 lineage cells are present in untreated organoids. (B) Immunostaining for PAX7 or PAX3 (white) and DAPI-stained nuclei (blue) on sections of day 5 WT spinal organoids, treated or not with BMP4. (C) Quantification of the percentage of PAX7⁺ or PAX3⁺ cells per organoid (dots: individual organoid values; bar plots: mean ± s.e.m.) (D) Quantification of PAX7 (i) and PAX3 (ii) nuclear intensity levels (dots: individual nuclear values in arbitrary units (AU); bar plots: mean ± s.e.m.; Mann–Whitney U test: **p < 0.01; ****p < 0.0001). (E) Images: Immunostaining for β-galactosidase (βgal; red), OLIG3 (green) and DAPI-stained nuclei (blue) on sections of day 5 P34::tk::LacZ spinal organoids treated or not with BMP4 (arrows points to neural crest cells). Right panels are magnified views of the squares shown in the left panels. Graphs: Quantification of the percentage of βgal⁺ cells per day 5 organoids treated or not with BMP4 (left) (dots: organoids and bar plots: mean ± s.e.m.) and of the percentage of βgal⁺ cells within the OLIG3⁺ population in organoid treated with BMP4 (right) (dots: organoids and bar plots: mean ± s.e.m.). (F) Immunostaining for LHX2, LBX1, and FOXD3 (all in white) and DAPI stained nuclei (blue) on sections of day 7 spinal organoids with the indicated genotypes, treated or not BMP4. (G) Quantification of the percentage of cells expressing LHX2 (i), LBX1 (ii), or FOXD3 (iii) in day 7 organoids of the indicated genotypes (3 independent clones per genotype). (dots: values per transverse section, dots shapes: independent clones; bar plots: mean ± s.e.m.; Two-way ANOVA: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). WT: Wild-type; DKO: Pax3; Pax7 double mutant. In all images, white dotted lines outline organoids. Scale bars: 50 µm. The data underlying this figure can be found in S7 Table.

Fig 4. Transcriptional programmes regulated by PAX3 and PAX7 activity in organoids treated or not to BMP4.

(A) Principal component analysis of day 6 organoids with the specified genotype, treated or not with BMP4. (B) Volcano plots comparing gene expression between WT and DKO day 6 organoids, treated or not with BMP4. The y-axis shows statistical significance (−log₁₀(q-value)) and the x-axis represents log₂(fold change). White dots: non-significant differentially expressed genes (DEG), black/colored dots: significant DEGs. Blue, pink, and green dots highlight master TFs for ventral, dI1–dI3, and dI4–dI6 IN lineages, respectively. (C–D) Normalized mean-centered expression levels of genes significantly downregulated or upregulated in DKO versus WT day6 organoids, treated (C) or not (D) with BMP4, shown for WT, Pax7^−/−, Pax3^−/−, and DKO organoids (dots/lines: individual genes; bars: mean ± s.e.m.; n = 3 clones/genotype). (E) Enrichment of genes encoding specifiers of dp1–dp3, dp4–dp6 and p0–p1 progenitors (NPs) and their corresponding INs, among genes downregulated (act.) or upregulated (rep.) in DKO organoids compared to WT day 6 organoids, with or without BMP4. (F) Heatmaps showing normalized, mean-centered mRNA expression of TFs marking dp1–dp3, dp4–dp6 and p0–p1 NPs and their associated INs assessed by RNA-seq in day 6 WT or DKO organoids, treated or not with BMP4, in 3 independent experiments. Fold change across samples are color-coded in blue (lower levels) to yellow (higher levels). (G) Immunostaining for OLIG3 (white) and DAPI-stained nuclei (blue) on sections of day 5 spinal organoids with the indicated genotype, treated or not with BMP4. White dotted lines outline organoids. Scale bar: 50 µm. Graph: Quantification of the percentage of OLIG3⁺ cells in these organoids (dots: values per transverse section, dots shapes: independent clones; bar plots: mean ± s.e.m; Mann–Whitney U test: ***p < 0.001; ****p < 0.0001; ns: non-significant). (H) Immunostaining for OLIG3 (green), PAX6 (red) and DAPI stained nuclei (blue) on transverse sections of E9.5 mouse embryos with the indicated genotypes. White dotted lines outline neural tubes. Scale bar: 50 µm. Graph: quantification of the number of OLIG3⁺ cells in such embryos (dots: values per transverse section, dots shapes: independent embryos; bar plots: mean ± s.e.m; Mann–Whitney U test: ****: p < 0.0001). The data underlying this figure can be found in S1 and S7 Tables.

Fig 5. Dynamics of chromatin states at PAX3/7-bound CRMs following PAX3/7 loss and BMP4 exposure.

(A) Heatmaps of normalized FLAG CUT&Tag signals for all PAX3/7-bound CRMs identified in organoids expressing FLAG-PAX3 (PAX3), FLAG-PAX7 (PAX7), or not (CTRL), treated or untreated with BMP4 at day 5 of differentiation. Signals are shown over a 3 kb window centered on CRM midpoints (n = 3,297). Three CRM groups are highlighted: those specific to BMP4-treated organoids, specific to untreated organoids, and shared between both conditions. (B) UCSC Genome Browser screenshots showing normalised FLAG CUT&Tag read distributions in wild-type (CTRL) and FLAG-PAX3 (PAX3) or FLAG-PAX7 (PAX7) organoids, with or without BMP4 treatment, at day 5 of differentiation. Signal is shown in counts per million (CPM). Coordinates: chr5:37816721-37837786 (left, scale bar: 5 kb), chr2:31628811-31658766 (middle, scale bar: 3 kb), and chr5:98159967-98180223 (right, scale bar: 4 kb). Shared binding sites are highlighted in grey, BMP4-specific in green, and untreated-specific in salmon pink. (C) UCSC Genome Browser screenshots showing normalised RNA-seq (RNA), FLAG, or H3K27me3 CUT&Tag read distributions in wild-type (CTRL), FLAG-PAX3 (PAX3) or FLAG-PAX7 (PAX7) expressing, and Pax3; Pax7 double-knockout (DKO) organoids, at days 3, 5 or 6 (D3, D5, D6) of differentiation, treated or not BMP4. Signal is shown in CPM. Coordinates: chr2:31627589-31,663,216, scale bar: 5 kb. (D) Metaplots showing H3K27me3 enrichment across a 30 kb region centered on PAX3/7-bound CRMs with significant differential H3K27me3 deposition between WT and DKO organoids. Data are shown for WT and DKO organoids at days 3 and 6 of differentiation. (E) UCSC Genome Browser screenshots showing normalised RNA-seq (RNA), FLAG CUT&Tag, ATAC-seq (chromatin opening), or H3K4me2 CUT&Tag read distributions in wild-type (WT), FLAG-PAX3 (PAX3), FLAG-PAX7 (PAX7) expressing, and Pax3; Pax7 double-knockout (DKO) organoids, treated or untreated with BMP4, at days 3, 5, or 6 (D3, D5, D6) of differentiation. Signals are shown in CPM. Coordinates: chr10:19355317–19416668, scale bar: 10 kb. (F, G) Metaplots showing ATAC-seq (F) and H3K4me2 CUT&Tag (G) signal enrichment over a 3 kb region centered on PAX3/7-bound CRMs displaying significant differences in chromatin accessibility or H3K4me2 deposition between CTRL and DKO organoids at day 6 of differentiation. Data are shown for WT and DKO organoids at days 3 and 6 of differentiation. (H) Enrichment analysis of PAX3/7-activated and -repressed genes (from Fig 4) among genes located near PAX3/7-bound CRMs displaying significant differential ATAC-seq signals (blue), H3K4me2 deposition (green), or H3K27me3 deposition (dark red) between CTL and DKO organoids at day 6 of differentiation (bars: −log₁₀(p-value)). The data underlying this figure can be found in S2 Table.

Fig 6. Silencer or enhancer activity of PAX3/7-bound CRMs nearby DV patterning genes.

(A) Table indicating the presence of PAX3/7-bound CRMs (black) near DV patterning TFs driving dI1–dI3, dI4–dI6, and V0/V1 INs differentiation, along with their chromatin state: opened by PAX (Open, blue), PAX-mediated H3K4me2 deposition (Activ. green), or PAX -mediated H3K27me3 deposition (Rep. dark red). (B) Fluorescence of mCherry (red) and Turquoise (white) on transverse sections of the chick neural tube, 24 h post-electroporation with the reporter plasmids from EF1α (i) or MLP (ii) promoters’ activity. (C, D) Panel i: UCSC Genome Browser screenshots showing normalised FLAG (black tracks) CUT&Tag read distributions in control (CTRL), FLAG-PAX3 (PAX3), FLAG-PAX7 (PAX7) at day 5 of differentiation and normalized H3K27me3 (dark red tracks) or H3K4me2 (green tracks) CUT&Tag read distributions in wild type (WT) and Pax3; Pax7 double-knockout (DKO) organoids at day 6 of differentiation, either treated or not BMP4. Read scales are shown in CPM. Grey bands highlight the position of Dbx1CRM4 chr7:49697106-49701929 (C) and Olig3CRM1 CRMs (chr10:19412350-19416446 (D). Panel ii Position of the shortened (s), or shortened and mutated (sm) version of either Dbx1CRM4 and Olig3CRM1, with the following color scheme: black for PrD, light blue for PrD-HD. Panels iii–v: Images: Fluorescence of mCherry (red) and Turquoise (white) on transverse sections of the chick neural tube, 24 h post-electroporation with pCAG-mCherry and the indicated constructs. Graphs: Quantification of the ratio between Turquoise and mCherry signal intensities along the dorsoventral (DV) axis of the neural tube (NT) expressed as a percentage of NT length (bar plots represent mean ± s.e.m.; n ≥ 4 embryos). Salmon-colored rectangles indicate the positions of PAX3/7-expressing progenitors. In all images, white dotted lines outline neural tubes. Scale bars: 50 µm. The data underlying this figure can be found in S2 and S7 Tables.