TBX3 is essential for establishment of the posterior boundary of anterior genes and upregulation of posterior genes together with HAND2 during the onset of limb bud development

Abstract

During limb bud formation, axis polarities are established as evidenced by the spatially restricted expression of key regulator genes. In particular, the mutually antagonistic interaction between the GLI3 repressor and HAND2 results in distinct and non-overlapping anterior-distal Gli3 and posterior Hand2 expression domains. This is a hallmark of the establishment of antero-posterior limb axis polarity, together with spatially restricted expression of homeodomain and other transcriptional regulators. Here, we show that TBX3 is required for establishment of the posterior expression boundary of anterior genes in mouse limb buds. ChIP-seq and differential gene expression analysis of wild-type and mutant limb buds identifies TBX3-specific and shared TBX3-HAND2 target genes. High sensitivity fluorescent whole-mount in situ hybridisation shows that the posterior expression boundaries of anterior genes are positioned by TBX3-mediated repression, which excludes anterior genes such as Gli3, Alx4, Hand1 and Irx3/5 from the posterior limb bud mesenchyme. This exclusion delineates the posterior mesenchymal territory competent to establish the Shh-expressing limb bud organiser. In turn, HAND2 is required for Shh activation and cooperates with TBX3 to upregulate shared posterior identity target genes in early limb buds.

Keywords: Gli3; Enhancer activities; Gene expression boundary; Irx genes; Limb bud; Mouse; TBX3 target genes.

Fig. 1.

Generation and characterisation of the Tbx3^3xFLAG (Tbx3^3xF)-tagged mouse allele. (A) Scheme summarising the strategy to generate the Tbx3^3xFLAG allele. Green and red arrowheads indicate the genotyping primers. (B) Western blot analysis of whole embryos (WE) and forelimb bud (FL) protein extracts (E10.5). After transfer, the western blot membrane was cut into three pieces prior to detection of TBX3 proteins and GAPDH as a loading control. Anti-TBX3 antibodies (exposure 30 sec; see Fig. S1B) detect the wild-type (Wt) and 3xFLAG proteins (TBX3^3xF/3xF; left panel); anti-FLAG antibodies (exposure 4 sec; see Fig. S1A) specifically detect TBX3^3xFLAG proteins (right panel, n=3). Detection of the GAPDH protein serves as a loading control (exposure 4 sec; see Fig. S1A). Solid lines indicate the cropped areas. For even longer exposure, see Fig. S1C (90 sec). (C) WISH analysis of Wt and Tbx3^3xF/3xF mouse embryos (n=3; 28-30 somites, E9.75). Scale bar: 250 µm. (D-F) Immunofluorescence analysis of limb bud sections (n=3; 34-36 somites, E10.5). Scale bars: 20 µm in D; 200 µm in E,F. All forelimb buds are oriented with anterior to the top and posterior to the bottom.

Fig. 2.

Identification of TBX3^3xF target genes in early mouse forelimb buds. (A) Overlapping regions of open chromatin (ATAC-seq) and TBX3^3xF ChIP-seq peaks identifies 3057 regions encoding potential CRMs. (B) Top 10 de novo motif analysis. (C) Intersection of TBX3-bound regions (E9.75-E10.25), ATAC-seq (E9.75) and DEGs (E9.75-E10.0) in wild-type and Tbx3^ΔΔ forelimb buds identifies 141 TBX3 candidate target genes. (D) Heatmap of the candidate targets genes in wild-type and Tbx3^ΔΔ forelimb buds. The z-score scale represents mean-subtracted regularised log-transformed read counts.

Fig. 3.

TBX3 candidate target gene regulatory network in early limb development. (A,B) Heatmaps illustrating relative gene expression of developmental regulators in wild-type (Wt) and Tbx3-deficient samples (n=3 biological replicates, E9.75-E10.0). The z-score scale represents mean-subtracted regularised log-transformed read counts. Shown are genes (n=53) that have essential functions in limb buds (Table S5). (C) TBX3 target GRN for early limb buds. In addition, target genes expressed in the AER and ectoderm (ECT), functioning in limb identity (ID) or chondrogenesis and skeletal development are shown. Arrows point to genes for which expression is positively regulated by TBX3 (white boxes), and inhibitory lines indicate genes for which expression is repressed (grey boxes). Asterisks indicate TFs. (D,E) WISH analysis of select genes in wild-type and Tbx3^Δ/Δc mouse forelimb buds (E9.75-E10.25, 29-32 somites). Asterisks indicate limb bud margins for some limb buds (D). Arrowheads indicate the posterior expression boundaries (Alx4, Irx3, Irx5) or upregulation (Hand1) in wild-type (white) and mutant (grey) limb buds. Forelimb buds are oriented with anterior to the top and posterior to the bottom. Scale bars: 100 µm.

Fig. 4.

Tbx3 and Hand2 cooperatively control posterior limb skeletal identities and co-regulate shared target genes in early limb buds. (A-C) Skeletal analysis of wild-type (Wt), Hand2-deficient (A), Tbx3^Δ/Δc (B) and Tbx3^Δ/ΔcHand2^Δ/+ (C) mouse forelimbs (E14.5). The genetic interaction of Tbx3 and Hand2 regulates posterior limb skeletal elements (ulna and posterior digits; C). The number of embryos exhibiting the illustrated phenotype is indicated at the bottom of each panel. Digit identities are indicated from anterior to posterior (1-5). Asterisks indicate digit malformations. Scale bars: 500 µm. (D) RNA-FISH analysis of forelimb buds (n=4 biological replicates per genotype; E10.0; 29-32 somites). All limb buds are oriented with anterior to the top and posterior to the bottom. Scale bar: 200 µm. (E) TBX3 and HAND2 share a small subset of their target genes (n=31). (F) Heatmap of the target genes co-regulated by TBX3 and HAND2. Genes in bold have known functions in early limb buds. Asterisks indicate that Shh and Gli3 are manually curated target genes.

Fig. 5.

Hand2 and Tbx3 requirement for activation and upregulation of key regulators involved in establishment of posterior identities. (A) RNA-FISH analysis of the shared targets Ptch1 and Shh in wild-type (Wt), Tbx3^Δ/Δc and Hand2^Δ/Δc forelimb buds (E10.0; 29-32 somites). (B,C) RNA-FISH analysis of Hoxd11 (B), Hoxd13 and Ets2 (C) in wild-type and mutant forelimb buds. Hoxd13 and Ets2 were colocalised; Hoxd11 was analysed using different limb buds. All forelimb buds are oriented with anterior to the top and posterior to the bottom. For all genes, n=4 biological replicates per genotype were analysed. Dashed lines delineate the shape of the limb bud. Scale bars: 200 µm.

Fig. 6.

The essential function of TBX3 in posterior expression boundary formation is independent of HAND2. (A) Colocalisation of Tbx3 (green), Irx3 (red) and Sall3 (blue) in wild-type (Wt), Tbx3^Δ/Δc and Hand2^Δ/Δc forelimb buds (E10.0, 29-32 somites). n=4 biological replicates were analysed per gene and genotype. White arrowhead points to the region of overlap between Tbx3 (green) and Sall3 (blue) expression. (B) Colocalisation of Tbx3 (green), Irx5 (red) and Gli3 (blue) expression in forelimb buds (E10.0, 29-32 somites). n=3 biological replicates were analysed per gene and genotype. Right-most panels in A and B show enlargements of the posterior-proximal regions. All limb buds are oriented with anterior to the top and posterior to the bottom. Dashed lines delineate the shape of the limb bud. Scale bars: 200 µm (main panels); 67 µm (enlargements).

Fig. 7.

Tbx3 is required to restrict the activity of the Gli3 limb enhancers and Gli3 expression from the posterior limb bud mesenchyme. (A) UCSC browser view of parts of the Gli3 genomic landscape with regions of accessible and active chromatin tracks (ATAC-seq and H3K27ac peaks; Malkmus et al., 2021). The two limb enhancers mm1179 and mm-hs1586 are highlighted. The enlargements on the right show the genomic regions tested for Gli3 enhancer activity (red bar) and the HAND2 and TBX3 ChIP-seq peaks. The green bars indicate the called peaks. Both replicates for the TBX3 ChIP-seq are shown; the HAND2 ChIP-seq data is from Osterwalder et al. (2014). (B-K) RNA-FISH analysis of wild-type (Wt) and Tbx3^Δ/Δc mouse limb buds (E10.5, 35-37 somites) of endogenous Tbx3 (yellow in E,J), transgenic enhancer-lacZ reporter RNA expression (blue in E,F,J,K; mm1179 in C,E,F and mm-hs1586 in H,J,K) and endogenous Gli3 expression (green in F,K). n=5 biological replicates were analysed per probe and genotype. Note: low and variable levels of remaining non-functional Tbx3 ^Δ transcripts are detected in Tbx3^Δ/Δc limb buds by the HCR^TM probe set. All forelimb buds are oriented with anterior to the top and posterior to the bottom. Dashed lines delineate the shape of the limb bud. Scale bars: 200 µm.

Fig. 8.

TBX3 restricts the posterior boundary of anteriorly expressed genes and TBX3 interacts with HAND2 to upregulate posterior gene expression. Scheme summarising the major findings of this study. TBX3 functions in AP boundary formation by restricting anterior genes from being expressed in the posterior limb bud mesenchyme (red lines). The four limb bud enhancers for Gli3 are indicated (mm1179 and mm-hs1586, Fig. 7A; mm682, Fig. S8; RII: Bastida et al., 2020). Two of these are directly repressed by TBX3 (mm1179 and mm-hs1586). In the posterior limb bud mesenchyme, TBX3 interacts with HAND2 to upregulate the expression of key regulatory genes (green arrows). HAND2 is required for posterior Tbx3 expression, and TBX3 in turn reinforces posterior Hand2 expression. The predominant role of HAND2 in positive regulation of posterior genes is indicated by thicker green arrows. All transcription units are shown in grey; CRMs and enhancers are indicated by white boxes.