Tbx5-hedgehog molecular networks are essential in the second heart field for atrial septation

Abstract

The developmental mechanisms underlying human congenital heart disease (CHD) are poorly understood. Atrial septal defects (ASDs) can result from haploinsufficiency of cardiogenic transcription factors including TBX5. We demonstrated that Tbx5 is required in the second heart field (SHF) for atrial septation in mice. Conditional Tbx5 haploinsufficiency in the SHF but not the myocardium or endocardium caused ASDs. Tbx5 SHF knockout embryos lacked atrial septum progenitors. We found that Tbx5 mutant SHF progenitors demonstrated cell-cycle progression defects and that Tbx5 regulated cell-cycle progression genes including Cdk6. Activated hedgehog (Hh) signaling rescued ASDs in Tbx5 mutant embryos, placing Tbx5 upstream or parallel to Hh in cardiac progenitors. Tbx5 regulated SHF Gas1 and Osr1 expression, supporting both pathways. These results describe a SHF Tbx5-Hh network required for atrial septation. A paradigm defining molecular requirements in SHF cardiac progenitors for cardiac septum morphogenesis has implications for the ontogeny of CHD.

Figure 1. Tbx5 Is Expressed in the Posterior Second Heart Field

Expression of Tbx5 and Gli1 analyzed by in situ hybridization on wild-type mouse embryos in sagittal sections at E9.0 (A–D) and transverse sections at E10.0 (E–L). Tbx5 (C and G) and Gli1 (D and H) were both expressed in the DMP and the splanchnic mesoderm. Tbx5 (I and K) but not Gli1 (J and L) was also expressed in the heart. Magnification: (A), (B), (E), (F), (I), (J) = 100×; (C), (D), (G), (H), (K), (L) = 400×.

Figure 2. Tbx5 Is Not Required in the Myocardium or Endocardium but Is Required in SHF Hh-Receiving Cells for Atrial Septation

(A–H) Analysis of atrial septation in myocardial or endocardial Tbx5 mutant embryos. Atrial septation was normal in Tbx5^fll/+ (A, C, E, G; 7/7), Tbx5^Tnt-Cre/+ embryos (B, D; 7/7; p = 1), and Tbx5^Tie2-Cre/+ (F,H; 7/7; p = 1) embryos at E13.5. (I and J) Cre-dependent somatic recombination of conditional Tbx5 locus. At both E9.5 and E10.5, hearts from Tbx5^Tnt-Cre/+ (I) and Tbx5^Tie2-Cre/+ (J) demonstrated a Tbx5 null allele (blue arrow, 480 bp band) and Cre expression (green arrow head, 220 bp band) in the heart (lane 2, 3) but not the tail (lane 1). Both the heart and tail DNA showed Tbx5 wild-type (black arrow: 158 bp band) and Tbx5-flox (red arrow: 198 bp band) alleles. (K–R) Atrial septal defects in germline and SHF Tbx5 mutant embryos. Primum ASDs were observed in 40% of Tbx5^−/+ embryos, consistent with previous reports (Bruneau et al., 2001), (L and N) (5/12; p = 0.0466) and 40% of Tbx5^{Gli1Cre-ERT2/+} embryos (P and R) (5/12) compared to 0% of Tbx5^fl/+ embryos (O and Q) (0/7; p = 0.0466) at E13.5 in transverse sections. Magnification: (A), (B), (E), (F), (K), (L), (O), (P) = 40×; (C), (D), (G), (H), (M), (N), (Q), (R) = 100×.

Figure 3. Tbx5 Is Required in SHF Hh-Receiving Cells for Atrial Septation

(A–E) Hh-receiving cardiac fate map in Tbx5 mutant embryos. Hh-receiving cells were marked by β-galactosidase expression in R26R^{Gli1Cre-ERT2/+} embryos at E7.5 and E8.5 by tamoxifen administration and analyzed by at E10.5. All β-galactosidase positive cells in the dorsal mesenchymal protrusion (DMP) (red arrows in C and D) were counted. Significantly fewer β-gal positive DMP cells were present in Tbx5^−/+; R26R^{Gli1Cre-ERT2/+} embryos (187.4 ±11.5) Data are presented as mean ± SEM, n = 3–4. (B and D) compared to R26R^{Gli1Cre-ERT2/+} embryos (92.4 ± 24.1, p = 0.0265) (A and C). (F–I) Tbx5 expression in SHF Tbx5 mutant embryos. Tbx5 expression is extinguished specifically in the DMP and SHF of Tbx5^Mef2c-Cre but not Mef2c^Cre/+ embryos at E9.5 by in situ hybridization. Myocardial Tbx5 expression is maintained in both genotypes (F and G red arrows). (J–M) SHF Tbx5 expression is required for DMP development. The DMP was a well-defined structure in Tbx5^fl/fl embryos (L, black arrow; 5/5) but absent in Tbx5^Mef2c-Cre embryos (K and M; 0/5; p = 0.0005). (N–Q) Absence of SHF fate map in Tbx5 SHF mutant embryos. β-galactosidase expressing cells are absent from the pSHF including the DMP region in Tbx5^Mef2c-Cre;R26R^Mef2c-Cre/+ embryos (Q, black arrow) but present in control R26R^Mef2c-Cre/+ embryos (P, black arrow) at E9.5. β-galactosidase positive cells are observed in the anterior SHF, outflow tract, and right ventricles in both genotypes (N and P versus O and Q). Data are presented as mean ± SEM, n = 3–4. Magnification: (A), (B), (F), (G), (J), (K), (N), (O) = 100×; (C), (D), (H), (I), (L), (M), (P), (Q) = 400×.

Figure 4. Tbx5 Is Required for Cell-Cycle Progress of Posterior SHF Cardiac Progenitors

(A–F) PSHF proliferation requires Tbx5. Fewer pSHF BrdU incorporated cells were observed in Tbx5^−/+ (B, D, E) than wild-type embryos (A, C, E) (61.4% ± 4.3% versus 34% ± 5.7%, p = 0.001). Fewer BrdU and Hh-receiving β-galactosidase pSHF comarked cells were observed in Tbx5^Gli1Cre-ERT2;R26R^{Gli1Cre-ERT2/+} embryos than in R26R^{Gli1Cre-ERT2/+} embryos (34.9% ± 8.4% versus 21.6% ± 5.6%, p = 0.0492). Data are presented as mean ± SEM, n = 3–4. (G–J) Distinct Tbx5-dependent transcripts in the pSHF and heart at E9.5 by real-time PCR. Tbx5 was downregulated in the heart and pSHF (G and H). Nppa and Gja5 were downregulated in the heart but not the pSHF (G and H). Gata6 and Mef2c were downregulated in the posterior SHF but not in the heart. (I and J) Genes that promote G1-S progress (Cdk2, Cdk6, Cdk25a, PCNA, Tfbp2) were downregulated whereas a gene that represses G1-S progress (Cdkn2a) was upregulated. Data are presented as mean ± SEM, n = 3–4. (K–Q) Tbx5 is required for cell-cycle progression in Hh-receiving pSHF cells. Significantly fewer β-galactosidase marked pSHF Hh-receiving cells were observed in M-phase by anti-Histone-H3 (H3S10) immunohistochemistry in R26R^{Gli1Cre-ERT2/+}; Tbx5^Gli1Cre-ERT2 embryos than in R26R^{Gli1Cre-ERT2/+} embryos (N and M, white arrowheads mark cells in mitosis diffuse nuclear staining) (7.73 ± 0.51% versus 4.25 ± 0.71%, p = 0.0037). There was no difference in the number of late G2 cells (M and N, red arrowhead marking cell in G2 phase with punctate nuclear staining) (15.31 ±2.23% versus 14.60± 1.94%, p = 0.9237). (Q) Quantification of H3S10 and β-galactosidase costained cells in the pSHF and DMP region. Magnification: (A), (B), (K), (L) = 100×; (C), (D), (M)–(P) = 400×. (R) Schematic of mouse Cdk6 including Tbx5 binding region (He et al., 2011) adjacent to Cdk6 promoter and cloned genomic fragment tested for Tbx5-dependent expression (green bar). Data are presented as mean ± SEM, n = 3–4. (S) Tbx5 significantly stimulated firefly luciferase expression from the Cdk6 promoter region compared to pGL3-vector control (10.55 ± 4.27 versus 1.28 ± 0.49, p < 0.009). Data are presented as mean ±SEM, n = 3. See also Figure S1 and Tables S1 and S2.

Figure 5. Hh-Signaling Is Epistatic to Tbx5 in Atrial Septation

(A–D) Genetic interaction between Tbx5 and Smo in atrial septation. Tbx5^−/+; Smo^−/+ embryos (A and D) demonstrated significantly more ASDs (7/8) than Tbx5^−/+ embryos (5/12; p = 0.0404) or Smo^−/+ embryos (A and C)(0/7; p = 0.0007) at E14.5. (E–P) Rescue of ASDs in Tbx5 SHF mutant embryos by constitutive Hh-signaling. Primum ASDs were observed in 40% of Tbx5^{Gli1Cre-ERT2/+} (I and k; 5/12) but 0% of R26-SmoM2 ^{Gli1Cre-ERT2/+};Tbx5 ^{Gli1Cre-ERT2/+} embryos (J and L) (p = 0.0098). Primum ASDs were observed in 100% of Tbx5^Gli1Cre-ERT2 (M and O; 5/5) but 20% of R26-SmoM2 ^{Gli1Cre-ERT2/+}; Tbx5 ^Gli1Cre-ERT2 embryos (N and P) (1/5; p = 0.0109). No heart defects were observed in Gli1Cre^ERT2/+ or R26-SmoM2 ^{Gli1Cre-ERT2/+} embryos. (Q) The decreased percentage of phosphorylated H3-histone positive β-galactosidase positive pSHF cells in Tbx5 ^{Gli1 Cre-ERT2} R26R ^{Gli1Cre-ERT2/+} embryos compared to Gli1Cre^ERT2/+ or R26-SmoM2^{Gli1Cre-ERT2/+}, R26R^{Gli1Cre-ERT2/+} embryos was normalized in R26-SmM2 ^{Gli1Cre-ERT2/+} Tbx5 ^Gli1Cre-ERT2 R26R ^{Gli1Cre-ERT2/+} embryos at E9.5. Data are presented as mean ± SEM, n = 3–4.

Figure 6. Tbx5 Act Upstream and Parallel to Hh-signaling in Atrial Septation

(A) Decreased expression of Hh-dependent genes in the SHF of Tbx5 mutant embryos. Ptch, Gli1, and Gas1 expression were significantly decreased in Tbx5^−/+ embryos at E9.5 by real-time PCR. Data are presented as mean ± SEM, n = 3–4. (B) Tbx5 significantly stimulated firefly luciferase expression from the Gas1 promoter region compared to pGL3-vector control (4.288 ± 1.43 versus 1.424 ± 0.357, p < 0.05). Data are presented as mean ± SEM, n = 3. (C) Schematic of mouse Gas1 including Tbx5 binding region (He et al., 2011) adjacent to Gas1 promoter and the cloned genomic fragment tested for Tbx5-dependent expression in vitro (green bar). See also Table S2. (D–G) Expression of Osr1 analyzed by in situ hybridization on wild-type mouse embryos by whole mount at E9 (D) and E10 (E) and sagittal sections at E9.5 (F and G). Osr1 was expressed robustly in the pSHF and DMP (black arrow, DMP reflection) and splanchnic mesothelium (red arrow, dorsal mesocardium). At, atrium; OFT, outflow tract; SHF, second heart field. Magnification: (D) and (E) = 100×; (F) and (G) = 400 ×. (H) Expression of Osr1 is decreased in pSHF of Tbx5^−/+ (fold change, 0.41 ± 0.2, p < 0.05) and Shh^−/− embryos 0.5 ± 0.04, p < 0.05) compared to wild-type embryos at E9.5 by real-time-PCR. Data are presented as mean ± SEM, n = 3. (I) Tbx5 significantly stimulated firefly luciferase expression from Osr1 genomic fragments Osr1-F1 (5.82 ± 1.42; p = 0.0009) and Osr1-F4 (5.6 ± 1.02; p = 0.0009) compared to pGL3 empty vector (1.28 ± 0.49). Tbx5 did not stimulate expression from Osr1 genomic fragments Osr1-F2 (1.4 ± 0.59; p = 0.61; p = 0.28) or Osr1-F3 (1.5 ± 0.41). Data are presented as mean ± SEM, n = 3. See also Table S2. (J) Tbx5 binds to Tbx5 responsive Osr1 genomic fragment in the SHF by ChIP-PCR. The Tbx5-responsive DNA fragments were immunoprecipated by Tbx5 antibody and were amplified by real-time PCR. Osr1-F1 (98.14 ± 45.72) and Osr1-F4 (82.54 ± 54.06) were significantly enriched compared to input DNA. Data are presented as mean ± SEM, n = 3. See also Table S2. (K) Schematic of mouse Osr1 including Tbx5 binding region (He et al., 2011) adjacent to Osr1 and the cloned genomic fragments Osr1-F1-F4 tested for Tbx5-dependent expression in-vitro (green bar). See also Table S2. (L) Model of Tbx5 transcriptional regulation in atrial septation. Tbx5 acts upstream of Hh-signaling, directly activating Gas1 expression. Tbx5 acts parallel to Hh-signaling, directly activating Osr1 expression. Tbx5 also directly activates Cdk6 in regulating atrial progenitor proliferation. (M and N) Model for combinatorial role of Tbx5 and Hh-signaling in atrial septation. Tbx5 promotes proliferation of pSHF inflow cardiac progenitors and Hh-signaling. PSHF Hh-signal receiving Tbx5⁺Gli1⁺ cells are specified as atrial septum progenitors and generate the atrial septum.