Partial Rescue of the Tbx1 mutant Heart Phenotype by Fgf8: genetic evidence of impaired tissue response to Fgf8

Abstract

Tbx1 is the candidate gene of DiGeorge syndrome and is required in humans and mice for the development of the cardiac outflow tract (OFT) and aortic arch arteries. Loss of function mutants present with reduced cell proliferation and premature differentiation of cardiac progenitor cells of the second heart field (SHF). Tbx1 regulates Fgf8 expression hence the hypothesis that the proliferation impairment may contribute to the heart phenotype of mutants. Here we show that forced Fgf8 expression modifies and partially rescues the OFT septation defects of Tbx1 mutants but only if there is some residual expression of Tbx1. This genetic experiment suggests that Tbx1, directly or indirectly, affects tissue response to Fgf8. Indeed, Tbx1^−/− mouse embryonic fibroblasts were unable to respond to Fgf8 added to the culture media and showed defective response of Erk1/2 and Rsk1. Our data suggest a coordinated pathway modulating Fgf8 ligand expression and tissue response to it in the SHF.

Introduction

Tbx1 is T-box transcription factor required for the development of several organs, including sections of the heart. Haploinsufficiency of TBX1 can cause DiGeorge/Velocardiofacial syndrome, often associated with congenital heart disease. Mouse mutants recapitulate most of the human syndrome phenotype. Tbx1 expression in the mesoderm, and, in particular, in the mesoderm expressing the cardiogenic transcription factor Nkx2.5, is required for the elongation and septation of the cardiac outflow tract as well as for the septation of the ventricles (1). Expression analyses and cell fate mapping established that Tbx1 is expressed in the second heart field (SHF), which is a cardiac progenitor cell (CPC) population that migrates into the cardiac outflow tract (OFT), right ventricle and part of the atria. We have shown that Tbx1 is expressed in tri-potent heart progenitors and, in these cells, it modulates positively cell proliferation and inhibits differentiation (2). Thus, the molecular pathway(s) regulated by Tbx1 in CPCs could be exploited for the expansion of cardiac stem cells in future applications of regenerative medicine. Therefore, it is important to define the mechanisms by which Tbx1 effects its functions in CPCs and thus define the molecular players. There are at least two mechanisms that appear relevant to a role of Tbx1 in cell proliferation. One is the recently described mechanism of Smad1-Tbx1 interaction (3) that leads to inhibition of Smad1 signaling, which inhibits proliferation in the SHF (4). Another is transcriptional regulation of Fgf8, a ligand of the fibroblast growth factor signaling that has, among other functions, mitogenic activity. Although Tbx1 and Fgf8 interact genetically (5), forced expression of Fgf8 in the Tbx1-expression domain is insufficient to rescue the main phenotypic abnormalities of Tbx1^−/− mutants, such as the heart phenotype (6). However, forced expression of Fgf8 in Tbx1-expressing cells is capable of partially rescue the severe thyroid phenotype, which is a cell non-autonomous consequence of Tbx1 loss (7). This finding and the availability of new mutant alleles have encouraged us to revisit the role of Tbx1 in FGF signaling. Here we show that forced expression of Fgf8 in mice that express a small dose of Tbx1 mRNA rescues partially the heart phenotype, thus we hypothesized that Tbx1 is necessary to respond to Fgf8. This hypothesis was successfully tested in a tissue culture model.

Materials and Methods

In this work we used the following mouse lines previously described: Tbx1^Cre (8), Tbx1^neo2 (9), Tbx1^fgf8 (6), Fgf8^fl/fl and Fgf8^+/− (10). Phenotypic analysis was carried out by direct examination of dissected embryos and by histological analyses.

Primary mouse embryonic fibroblasts (MEFs) were isolated from individual Tbx1^−/−, Tbx1^neo2/− and wild-type embryos at E13.5. To this hand, the internal organs, head, tail and limbs were removed. Cells were cultured in Dulbecco's modified Eagle's medium with 20% FBS and 1% NEAA, and used for a maximum of 4 passages. We derived and tested 3 wild type, 5 Tbx1^−/− and 2 Tbx1^neo2/− MEF lines, each derived from individual embryos.

Fgf8 treatment was carried out using recombinant Fgf8b (R&D) at 50μg/ml along with 100ng/ml of heparin (Sigma) for 5, 10 or 15 minutes.

Quantitative real time PCR was performed using reverse-transcribed total RNA from MEFs or whole embryos, using SYBR green and an Applied Biosystem 7900H machine, with the following primers: Etv4 F- cagcaggaagccaccact, R- gggggagtcataggcactg; Etv5 F- gcagtttgtcccagattttca, R- gcagctcccgtttgatctt; Rsk1 F- ttcacacggctctcaaagg, R- ccagctcagccaggtaaaac. Tbx1: E3/E4-F: ctgaccaataacctgctggatga, E3/E4-R-ggctgatatctgtgcatggagtt. Relative quantification was calculated by the ΔΔCt method.

Western blotting analyses were performed using total protein extracts from MEF cells. The antibodies were purchased from Millipore (pRsk1), SantaCruz (Erk1/2, Crkl, Frs2α, Rsk1), Cell Signaling Technology, Inc (p44/42, pErk1/2), Abcam (Actin), and Sigma (Tubulin).

Results and Discussion

Loss of Fgf8 in the Tbx1 domain causes Tbx1^−/−-like OFT septation defects

Using a tamoxifen-inducible Tbx1-cre (Tbx1^mcm) driver, we have shown that Fgf8 dosage reduction in conditional mutants enhances the Tbx1 haploinsufficiency phenotype at the level of aortic arch defects (6), but we could not find OFT defects. In addition, Brown et al. (11) using a Tbx1 promoter-enhancer construct as the transgenic Cre driver, observed various types of OFT defects in Fgf8 conditional mutants, but they did not report truncus arteriosus communis (TAC) the typical OFT defect observed in Tbx1^−/− mutants. Thus, we decided to repeat the experiment using our latest Cre knock-in driver, Tbx1^Cre, which induces robust recombination and is expressed in the endogenous Tbx1 domain (8). RNA in situ hybridization with an Fgf8-deletion specific probe (10) at E9 showed undetectable or very low levels of Fgf8-expression in the pharyngeal region of Tbx1^Cre/+;Fgf8^fl/− embryos compared to controls while other domains such as the fronto-nasal process and limb bud showed robust expression (Fig. 1A,B). Analysis of the OFT phenotype in E18.5 Tbx1^Cre/+;Fgf8^fl/− embryos revealed the presence of truncus arteriosus communis (TAC) in 67% (10/15) of them. In 5 of the 10 embryos with TAC, the defect was partial (Fig. 1E–J), while in the remainder, the defect was complete i.e. total lack of septation, similarly to Tbx1^−/− mutants. In all conditional mutants with TAC, the truncal valve maintained its continuity to the mitral valve, indicating that the alignment of the OFT with the ventricles is less defective in these mutants than in Tbx1^−/− animals. In addition, in all embryos, the TAC was positioned above the inter-ventricular septum straddling a ventricular septal defect (VSD) (Fig. 1H–I). In contrast, in Tbx1^−/− mutants, the TAC is positioned on the right ventricle. Thus, conditional ablation of Fgf8 in the Tbx1 domain interrupts both conal and truncal septation, but rotation and alignment of the OFT are only mildly affected. Overall, these data demonstrate that the loss of expression of Fgf8 in Tbx1-expressing cells causes OFT developmental abnormalities similar to those caused by loss of Tbx1, at least in a Tbx1^+/− background.

Figure 1. Loss of Fgf8 in Tbx1-expressing cells causes Truncus arteriosus Communis.

A–B: Fgf8 in situ hybridization in control and conditional mutants (Tbx1^cre/+;Fgf8^fl/−) at E9.5 shows reduced expression in the pharyngeal region (arrow). O, otocyst; I, 1st pharyngeal arch; fl, forelimb bud.

C–J: OFT and intracardiac phenotype in near term control (C,G) and Tbx1^cre/+; Fgf8^fl/− (D–F; H–J) embryos. Note the interventricular septal defect (*) and truncus arteriosus communis (TAC).

K: Histogram of qRT-PCR data showing the relative expression of Tbx1 E9.5 embryos (3 embryos per genotype).

Fgf8 rescue of OFT septation defects of Tbx1 mutants requires a small amount of functional Tbx1

The Tbx1^Fgf8 allele is null for Tbx1 and expresses an Fgf8 cDNA in the Tbx1 expression domain (6). Despite the predicted importance of the loss of Fgf8 expression in the pathogenesis of OFT septation defects of Tbx1^−/− mutants, this allele did not modify these defects in Tbx1-null embryos (6). Therefore, we considered the hypothesis that the complete absence of Tbx1 may impair the cellular response to Fgf8. This hypothesis would also explain why the Tbx1^Fgf8 allele could partially rescue the thyroid phenotype of Tbx1^−/− mutants (7). Indeed, Tbx1 is not expressed in the thyroid tissue, targeted by Fgf8, thus it cannot have any role in regulating response to Fgf8 in this tissue.

To address this hypothesis in vivo, we carried out rescue experiments in mice that express a low amount of Tbx1 (per se insufficient to support normal development of the OFT), but perhaps sufficient to partially restore cellular response to Fgf8.

We used the Tbx1^neo2 allele that expresses 15–20% of the WT Tbx1 mRNA level (12). Tbx1^neo2/− animals show very similar cardiovascular defects as Tbx1^−/− mutants, indicating that this low level of Tbx1 expression is insufficient to modify significantly the OFT null phenotype (9, 12). We maintain the Tbx1^Fgf8 allele in the Tbx1^Fgf8/Dp1 background, where Dp1 is a segmental duplication of the Tbx1 locus which rescues the high penetrance of aortic arch defects associated with the Tbx1^Fgf8 allele (6). Tbx1^Fgf8/Dp1 mice were crossed with Tbx1^neo2/+ mice, and the offspring examined atE18.5. In total, we have analyzed 97 embryos (18 Tbx1^Fgf8/neo2, 29 Tbx1^Fgf8/+, 19 Tbx1^neo2/Dp1, and 31 Tbx1^+/Dp1), of these, only Tbx1^Fgf8/neo2 and Tbx1^Fgf8/+ embryos presented with cardiovascular abnormalities. In particular, 19 Tbx1^Fgf8/+ embryos (66%) exhibited aortic arch artery abnormalities but no OFT defects (consistent with previously reported data (6)). All the 18 Tbx1^Fgf8/neo2 embryos presented with OFT defects (Fig 2A–D”). These included TAC (10 embryos, 56%), double outlet right ventricle (DORV) (7 embryos, 39%) and transposition of the great arteries (TGA, 1 embryo or 6%). This phenotype represents a considerable improvement compared to the one observed in Tbx1^neo2/− embryos. Indeed, the latter genotype is associated with 95% incidence of TAC (9, 12), thus almost all embryos lack OFT septation. In contrast, only 56% of Tbx1^Fgf8/neo2 embryos lack OFT septation. We could no find any other phenotypic changes compared to the Tbx1^neo2/− phenotype. To evaluate whether the partial phenotypic rescue may be caused by up-regulation of the Tbx1^neo2 allele by Fgf8, we performed real time quantitative (qRT) PCR analysis on E9 embryos (Fig. 1K). The average relative expression level was not significantly different (19% for Tbx1^Fgf8/neo2 vs. 14% for Tbx1^neo2/−; p=0.07 with 2-tailed T-test). These data support the hypothesis that reduced Fgf8 expression in Tbx1^−/− embryos contributes to the OFT septation phenotype and it also suggest that Tbx1 is required for tissue response to Fgf8.

Figure 2. Forced Fgf8 expression in the Tbx1 domain partially rescues cardiovascular defects of Tbx1-hypomorphs.

A–D: Isolated hearts from near-term control Tbx1^dp1/neo2 (A) and Tbx1^fgf8/neo2 (B–D) embryos show a range of phenotypes, including alignment and OFT defects. Sections of the corresponding hearts show DORV (B–B”), TAC type A4 (C'–C”, distal septation of aorta and pulmonary trunk with unseptated truncal valve overriding the interventricular septum) or TAC type A2 (D'–D”). Note the continuity between truncal and mitral valves in C” which is lost in D” (asterisk in D”). Ao, aorta; PT, pulmonary trunk; DORV, double outlet right ventricle; TAC, truncus; LA, left atrium; A2, A4 according to van Praagh's classification. Note VSD in C” and D” (arrowheads).

E. qRT-PCR of Fgf8 target genes Etv4 and Etv5 expression in mouse embryonic fibroblasts (MEFs) from Tbx1^+/+ and Tbx1^−/− embryos (two independent lines per genotype) exposed to recombinant FGF8 in culture. Mutant cells do not respond properly to FGF8 stimulation.

F–G. Western blots of proteins extracted from the same cells. Mutant cells express very little or undetectable level of Rsk1 and Erk1/2 phosphorylation.

H: Western blot of proteins extracted from wild type and Tbx1^neo2/− embryos. Low levels of Tbx1 are sufficient to rescue Rsk1 phosphorylation response to FGF8.

Tbx1^−/− cells do not respond normally to Fgf8

To support the above in vivo findings that suggest a requirement of Tbx1 for normal response to Fgf8, we evaluated the response of cultured mouse embryonic fibroblasts (MEFs) from wild type and Tbx1^−/− embryos, to recombinant Fgf8b protein added to the culture media. We measured the expression of two genes known to respond positively to Fgf8 treatment, Etv4 (also known as Pea3) and Etv5 (also known as Erm) (13). qRT-PCR analysis on two independent wild type MEF lines revealed a robust response to Fgf8, while two independent Tbx1^−/− lines did not show any significant response (Fig. 2E). Next, we tested the phosphorylation of Rsk1, a downstream component of FGF and other growth factors signaling pathways, responsible for relaying the signal to the nucleus. Results showed a robust response to Fgf8 in wild type MEF cell lines, but very low signal in mutant cells (Fig. 2F). Rsk1 is a phosphorylation target of the MAP kinase Erk1/2. Therefore, we tested the phosphorylation response of Erk1/2 to FGF8. Analogously to the results with Rsk1, we did not find any difference in Erk activation in mutant cells (Fig. 2G). We checked whether loss of Tbx1 causes any difference in the expression of Rsk1 or Erk1/2. For the latter, we could not demonstrate any difference in expression (Fig. 2G). However, for Rsk1 we saw reduced expression in most cell lines analyzed (each from an individual embryo) at the protein level (Supplementary Fig. 1) and RNA level (Supplementary Fig. 2). However, even in the presence of near normal levels of Rsk1 protein, we could not detect any phosphorylation response to FGF8 in mutant MEFs (Supplementary Fig. 1, lanes 3–4). The observed variability may be due to the heterogeneous nature of the primary MEF cultures. We also tested whether Frs2α (14) and Crkl (15) proteins are expressed differentially in mutant vs. wild type cells, but we could not detect any difference (Supplementary Fig. 3).

Next, we tested whether the small amount of Tbx1 expression in Tbx1^neo2/− MEFs is sufficient to rescue the response to FGF8 in culture. Results showed that indeed Tbx1^neo2/− MEFs are capable to respond to FGF8 as revealed by Rsk1 phosphorylation (Fig. 2H).

Taken together, our in vivo and in vitro data strongly suggest an impairment of tissue response to Fgf8 associated with loss of Tbx1. Indeed, in vivo, replacement of Fgf8 expression in Tbx1 expressing cells is capable of partially rescuing the OFT phenotype only in the presence of a small amount of Tbx1 expression. Analogously, in vitro, MEF cells require a minimal level of Tbx1 expression to respond to Fgf8. Further work is necessary to address the mechanism by which loss of Tbx1 impairs response to Fgf8. One can speculate that this may involve direct, cell autonomous interaction with the FGF signal transduction pathway, or an indirect interaction. Mesp1-fated mesodermal cells (which include cardiac progenitors) require Fgfr1 and Fgfr2 to support normal OFT development (16), thus suggesting that cardiac progenitors are targeted by FGF signaling. Because Tbx1 is expressed in these cells, it is possible that it plays a cell-autonomous role in modulating their response to Fgf8. It is also possible that the loss of Tbx1, by affecting differentiation, makes cells unresponsive to Fgf8.

Our data open a new window into the regulation of the response of cardiac progenitor cells to growth factors. Further work will be necessary to fully understand the significance of this mechanism for the proliferation and growth of these cells.

Supplementary Material

01

Figure S1 Western blot analysis of proteins extracted from four independent MEF cultures, one derived from a wild type (lanes 1– 2) and three from Tbx1^−/− embryos (lanes 3–8). Each culture was starved for 24 hours (lane 1,3,5,7) and exposed to recombinant FGF8 for 15 minutes (lane 2,4,6,8). Mutant cells express low levels of phosphorylated Rsk1 (p-Rsk1) and in some cases also low levels of total Rsk1 protein.

02

Figure S2 Relative expression of Rsk1 in 8 independent wild type and Tbx1^−/− MEF cultures. For each point is indicated the ratio WT vs. mutant expression. Note that in all cases Rsk1 was expressed at a higher level in WT cells compared to mutant cells.

03

Figure S3 Western Blot analysis of proteins extracted from four independent MEF cultures, two from wild type embryos (lanes 1–2), and two from Tbx1^−/− embryos (lanes 3–4). Tbx1 deletion does not affect Crkl and Frs2α protein expression.