Timed Treatment of Sodium Chlorate Can Be Used to Dissect the Role of Fgf Signal During Gastrulation
Sulfated glycosaminoglycans (GAGs) are distributed in the basement membrane of the epiblast and the extracellular matrix of mesoderm cells in gastrulating mouse embryos (Garcia-Garcia and Anderson,2003). Because heparan sulfate (HS), a sulfated GAG, is necessary for Fgf signaling (Bernfield et al.,1999), we used sodium chlorate as an inhibitor of sulfated GAG biosynthesis to examine the role of Fgf signaling during gastrulation (Greve et al.,1988; Ullrich and Huber,2001). Thus, we collected mouse embryos at embryonic day (E) 6.2 or E6.5, and cultured for them for 16–20 hr with 15 mM chlorate, a concentration that we previously showed to inhibit sulfated GAG biosynthesis at early somite stages (Oki et al.,2007). Chlorate-treated embryos were fixed and subjected to morphological and histological analysis. When pre-gastrulation E6.2 embryos were cultured with chlorate, the primitive streak exhibited an irregular shape, with the epiblast layer protruding inwardly into the proamniotic cavity and a mass of cells accumulating in the protrusion, which was not seen in control cultures (Fig. 1A, E, C, G). The cells in the protrusion expressed Lim1, a marker gene for nascent mesoderm (Shawlot and Behringer,1995), but failed to downregulate E-cadherin, suggesting that EMT was impaired (see Fig. S1, which is available online). On the other hand, embryos that were cultured with chlorate from E6.5 (at the onset of gastrulation) onward showed normal ingression of mesodermal cells through the streak and lateral migration (Fig. 1D, H; compare with Fig. 1B, F). In both cases, the synthesis of HS was significantly inhibited in chlorate-treated embryos, whereas HS was normal in control cultures (Fig. 1I–L).
Figure 1. Loss of heparan sulfate causes impaired epithelial-to-mesenchymal transition similar to that in Fgf8 and Fgfr1 mutant embryos. A–D: Lateral views of control (A, B) and chlorate-treated (C, D) embryos, with the anterior to the left. The embryonic stages at which the culture was started are indicated on the left. When the embryos are cultured with chlorate from E6.2 for 16 hr, severe defects are observed in the embryonic portion (C). E–L: Transverse sections of control (E, F, I, J) and chlorate-treated (G, H, K, L) embryos at the level indicated in A–D. Anterior is to the left. F-actin (phalloidin, green) and nuclei (DAPI, magenta) are shown in E–H, and heparan sulfate (anti-HS IgM, red) and nuclei (DAPI, blue) are shown in I–L. Arrowheads in G indicate the protruded ectoderm occupying the amniotic cavity. The amount of heparan sulfate is dramatically reduced in chlorate-treated embryos (K, L). a, anterior; p, posterior.
Download figure to PowerPoint
The irregular streak formation in cultured embryos lacking HS resembles the phenotype of Fgf8−/− and Fgfr1−/− embryos (Deng et al.,1994; Yamaguchi et al.,1994; Sun et al.,1999; Ciruna and Rossant,2001), suggesting that chlorate successfully inhibited Fgf signaling. We next investigated to what extent the chlorate culture system was able to mimic the phenotype of Fgf8 and Fgfr1 mutants. Because Fgf8−/− and Fgfr1−/− embryos show perturbed expression of posterior and axial marker genes (Deng et al.,1994; Yamaguchi et al.,1994; Sun et al.,1999; Ciruna and Rossant,2001), we cultured E6.5 embryos for 16–24 hr with 15 mM chlorate and analyzed gene expression. As expected, chlorate-treated embryos showed the same altered marker gene expression as Fgf8−/− and Fgfr1−/− embryos. T, a marker of the primitive streak and axial mesoderm at E7.5 (Wilkinson et al.,1990; Herrmann,1991), was expressed in the primitive streak, but its expression was not extended anteriorly (Fig. 2A; n=12/12). Tbx6, which is expressed in the primitive streak and nascent mesoderm (Chapman et al.,1996), was lost (Fig. 2B; n=7/7). The expression of Snail, a zinc finger transcription factor that is essential for EMT during gastrulation (Nieto et al.,1992; Smith et al.,1992; Carver et al.,2001), was also absent from the primitive streak (Fig. 2C; n=9/9). FoxA2, which is expressed in the anterior streak at E6.75 and in the node and notochord from E7.5 (Ang et al.,1993; Sasaki and Hogan,1993), was expressed in the anterior streak, although they were at the equivalent stage of E7.5 (Fig. 2D; n=4/5). Nodal expression, which is observed in the posterior embryonic ectoderm and visceral endoderm at E6.5 and in the node at E7.5 (Conlon et al.,1994), persisted in the ectoderm without being confined to the node (Fig. 2E; n=6/6). Shh, which marks the anterior midline structures, such as the notochord, at E7.5 (Echelard et al.,1993; Roelink et al.,1994), was broadly expressed at the distal tip of the embryo without extending anteriorly (Fig. 2F; n=2/6). We also examined the expression of Sef, a probable target gene of Fgf signaling (Lin et al.,2002). In chlorate-treated embryos, the expression of Sef in the streak region was significantly diminished (Fig. 2G; n=7/7). These alterations in gene expression indicate that mesodermal patterning is severely affected in chlorate-treated embryos. Importantly, the shape of the primitive streak and the migration of mesodermal cells were relatively normal in the embryos cultured with chlorate from E6.5 (Fig. 1H). However, the embryos exhibited perturbed marker expression similar to that seen in Fgf8 and Fgfr1 mutants, suggesting that the perturbation of gene expression seen in both mutants is independent of the EMT defects and is directly regulated by Fgf signaling.
Figure 2. Chlorate treatment at gastrulation causes mesodermal patterning defects similar to those in Fgf8 and Fgfr1 mutant embryos. A–G: Embryos were cultured from E6.5 for 16 hr (A–E, G) or 24 hr (F) in the absence (top) or presence (bottom) of 15 mM chlorate. Expression of various marker genes was examined by whole-mount in situ hybridization. The genes examined are indicated in each panel. Anterior is to the left (A–E, G) or at the top (F). l, left; r, right.
Download figure to PowerPoint
Interestingly, Fgf8−/− embryos show patterning defects in anterior tissues at E7.5, even though Fgfr1 is not expressed in the visceral endoderm (Yamaguchi et al.,1992; Sun et al.,1999). For example, the expression of anterior visceral endoderm (AVE) markers Hex and Hesx1 (Hermesz et al.,1996; Thomas et al.,1998) is expanded, and anterior neuroectoderm (ANE) markers Six3 and Otx2 (Simeone et al.,1993; Oliver et al.,1995) are broadly expressed. In embryos cultured with chlorate from E6.5, the expression domain of AVE and ANE marker genes was expanded, reaching the distal tip of the embryo (Fig. 3A–D; n=4/4 for each gene), suggesting that the perturbation of these anterior markers is also independent of EMT defects and results indirectly from loss of Fgf signaling during gastrulation (see Discussion section). In summary, the gene expression patterns of chlorate-treated embryos resemble that of Fgf mutants, suggesting that chlorate treatment causes loss of HS, resulting in inhibition of Fgf signaling in gastrulating embryos.
Figure 3. Chlorate treatment at gastrulation causes disturbed AVE and ANE marker gene expression similar to that in Fgf8 and Fgfr1 mutant embryos. A–D: Embryos were cultured from E6.5 for 16 hr in the absence (top) or presence (bottom) of 15 mM chlorate. Expression of AVE and ANE marker genes was examined by whole-mount in situ hybridization. The genes examined are indicated in each panel. Anterior is to the left.
Download figure to PowerPoint
PD173074, But Not SU5402, Is a Potent Inhibitor of Fgf Signaling in Gastrulating Mouse Embryos
Although our data indicate that chlorate causes gastrulation defects, probably by inhibiting HS-dependent Fgf signaling, HS is also known to mediate other signaling molecules, such as BMPs and Wnts (Hacker et al.,2005). HS secreted into the extracellular space interacts with these proteins and serves as a scaffold for establishing molecular gradients, as shown in Drosophila (Belenkaya et al.,2004; Kirkpatrick et al.,2004; Kreuger et al.,2004). Although the gastrulation defects in the chlorate-treated embryos differ from those of BMP and Wnt signaling mutants, we nonetheless examined whether Fgfr-specific inhibitors have the same effect on gastrulation as chlorate.
We first tested a synthetic Fgfr1 inhibitor, SU5402 (Mohammadi et al.,1997). This compound is widely used to inhibit Fgf signaling in gastrulating Xenopus and zebrafish embryos and causes severe defects in mesoderm patterning (e.g., Griffin and Kimelman,2003; Fletcher and Harland,2008). We cultured E6.5 embryos with 40 μM SU5402 for 16 hr and found that Tbx6 expression was normal (data not shown). We also tested a higher dose of SU5402 (100 μM), but the expression of T, Nodal, and Snail was still unaffected (data not shown). It is unlikely that the SU5402 we used had decreased inhibitory activity, since we were able to reproduce the published data with this compound (Corson et al.,2003; Maretto et al.,2008). Erk phosphorylation was diminished in the extra-embryonic ectoderm when E6.2 embryos were cultured in 40 μM SU5402 for 1.5 hr, and Tbx6 expression was downregulated when E7.2 embryos were cultured in the same concentration of SU5402 for 20 hr (data not shown). These results suggest that the action of SU5402 is context-dependent, and this compound is not able to perturb gene expression in the same ways as chlorate treatment.
We next tested another compound, PD173074, which inhibits auto-phosphorylation of Fgfr1 with higher specificity and affinity than does SU5402 (Mohammadi et al.,1998). Although this compound is widely used in cell biological experiments and is able to substitute for SU5402 at lower doses (e.g., Ying et al.,2008), to our knowledge there has been no report of PD173074 treatment in early vertebrate development. We cultured E6.5 embryos for 16 hr in various concentration of PD173074 (0.2–5 μM) and examined the expression of Tbx6. Based on the staining level, Tbx6 transcripts were reduced by about half with 0.5 μM PD173074, and were completely abolished at more than 1 μM PD173074 (Fig. 4B; see also Fig. 6A, D, and data not shown). We, therefore, used 1 μM of the compound to test whether it could recapitulate the perturbed gene expression observed in Fgf8 and Fgfr1 mutants. Notably, PD173074 treatment of E6.5 embryos resulted in expression patterns that were indistinguishable from those of chlorate-treated and Fgf8 and Fgfr1 mutant embryos (Fig. 4A–J). In addition, embryos cultured under these conditions did not exhibit EMT defects, consistent with chlorate-treated E6.5 embryos (data not shown). These results indicate that unlike SU5402, PD173074 is able to inhibit Fgfr1 in gastrulating embryos.
Figure 4. PD173074 treatment at gastrulation phenocopies Fgf8 and Fgfr1 mutant embryos. A–J: Embryos were cultured from E6.5 for 16 hr (A–E, G–J) or 24 hr (F) with DMSO (top) or 1 μM PD173074 (bottom). Expression of various marker genes was examined by whole-mount in situ hybridization. The genes examined are indicated in each panel. Anterior is to the left (A–E, G–J) or at the top (F). The number of embryos with perturbed marker gene expression is as follows: 3/3 (A), 10/10 (B), 3/3 (C), 3/5 (D), 3/3 (E), 4/4 (F), 7/7 (G), 6/6 (H), 3/3 (I), 3/3 (J).
Download figure to PowerPoint
Fgf Signaling Is Required for the Expression of Nodal and Its Upstream Genes
It has been reported that Nodal expression in the node and LPM disappears in hypomorphic Fgf8 mutants (Fgf8neo/−) (Meyers and Martin,1999). Because Nodal expression in the node is required for its subsequent expression in the LPM, it is likely that loss of Fgf8 principally affects Nodal expression in the node. However, it was unclear why reduced Fgf8 expression resulted in impaired expression of Nodal in the node. To address this question, we first examined whether Nodal expression in the node was diminished when cultured with Fgf signaling inhibitors. We collected embryos at the late-bud stage, just prior to the generation of nodal flow and Nodal expression in the node, and cultured them with PD173074 or chlorate. In control cultures, Nodal expression became apparent in the node when embryos were cultured until the two- to three-somite stage (for 12–14 hr; Fig. 5A; n=4/4). When the embryos were cultured further, Nodal expression also appeared in the left LPM, and eventually expanded throughout the LPM by the five- to six-somite stage (for 16–20 hr; data not shown and Fig. 5B; n=8/10). In contrast, embryos cultured with 1 μM PD173074 failed to express Nodal in the node at the two- to three-somite stage or in the LPM at the five- to six-somite stage (Fig. 5E, F; n=4/4 and 10/10, respectively). Similar results were obtained by culturing until the two- to three-somite stage with 50 mM chlorate, although culturing at this concentration until the five- to six-somite stage caused severe growth retardation (Fig. 5I; n=6/6 and data not shown). Altogether, these results are consistent with the impaired Nodal expression observed in Fgf8 hypomorphic mutants.
Figure 5. Inhibition of Fgf signaling affects the Tbx6-Dll1 cascade and results in diminished Nodal expression in the node. A–K: Expression of Nodal (A, B, E, F, I), Dll1 (C, G, J), and Tbx6 (D, H, K) in embryos cultured with DMSO (A–D), 1 μM PD173074 (E–H), or 50 mM chlorate (I–K). Embryos collected at the late-bud stage were cultured until the two- to three-somite stage, except in B and F, which show embryos cultured until the five- to six-somite stage. Anterior is at the top.
Download figure to PowerPoint
We next investigated the expression of genes upstream of Nodal. Previous studies suggested that Nodal expression in the node requires the Notch ligand Dll1, which activates the node-specific enhancer of Nodal via the Notch signal-dependent transcription factor Rbp-Jκ (Krebs et al.,2003; Raya et al.,2003). Furthermore, genetic and biochemical analyses have revealed that Dll1 is a direct target of Tbx6 (White and Chapman,2005; Hadjantonakis et al.,2008). We, therefore, investigated the expression of Dll1 and Tbx6 in PD173074- or chlorate-treated embryos. In control cultured embryos at the late-bud stage to the 2- to 3-somite stage, Dll1 and Tbx6 are expressed in the pre-somitic mesoderm encircling the node crown cells (Fig. 5C, D; n=4 for each gene). However, in embryos cultured with PD173074 or chlorate, the expression of both genes was greatly reduced in this region (Fig. 5G, H, J, K; n=5 for each case). Because the reduction in Dll1 expression seems to be more severe than observed in Tbx6−/− embryos (Hadjantonakis et al.,2008), Dll1 may be regulated by other transcription factors in addition to Tbx6. Nonetheless, these results suggest that Fgf signaling induces Nodal expression in the node by the Tbx6-Dll1 cascade. In addition, this cascade probably still functions even after Nodal expression in the node has been established, because the expression of Nodal in the node, as well as that of Tbx6 and Dll1, is significantly reduced when embryos are cultured with PD173074 to the five- to six-somite stage from head-fold stage, when Nodal expression is already apparent in the node (data not shown). Thus, Fgf signaling is necessary for both initiating and maintaining Nodal expression in the node through the Tbx6-Dll1 cascade.
Fgf Signaling Regulates Tbx6 Expression Through the Ras/MAPK Cascade
Fgfr signals mainly through three pathways: the Ras/MAPK, PLCγ, and PI3K cascades (reviewed in Eswarakumar et al.,2005; Thisse and Thisse,2005). While Fgf signaling is necessary for gastrulation and mesoderm patterning, it is largely unknown which cascade Fgf signaling uses in these developmental processes. To investigate this question, we tested available chemical compounds that specifically inhibit these cascades. E6.5 embryos were cultured for 16 hr with inhibitors for Mek (PD98059 at 100 μM or U0126 at 100 μM), PLCγ (U73122 at 8 μM), or PI3K (LY294002 at 100 μM) and were examined for Tbx6 expression. However, no alterations in Tbx6 expression were observed with either compound (Fig. 6A–C; n=5 for each compound; and data not shown). We next tested the combination of these compounds with a low dose of Fgfr inhibitor that partially impairs Fgf signaling. Although no effects were observed by either inhibitor in conjunction with 40 μM SU5402 (data not shown), culturing with PD173074 alone at 0.5 μM decreased Tbx6 expression to about half that of controls (Fig. 6A, D; n=5). Under these sensitized conditions (0.5 μM PD173074), treatment with the Mek inhibitors (100 μM PD98059 or 100 μM U0126) completely eliminated Tbx6 expression (Fig. 6E, F; n=8 for each compound). In contrast, the PLCγ or PI3K inhibitor (20 μM U73122 or 100 μM LY294002) had no effect on Tbx6 expression (data not shown). This is the first evidence suggesting that Fgf signaling exerts its effects through the Ras/MAPK cascade during gastrulation.
Figure 6. Synergistic effect of Fgfr and Mek inhibitors on Tbx6 expression. A–F: E6.5 embryos were cultured for 16 hr with DMSO (A, D), 100 μM PD98059 (B,E) or 100 μM U0126 (C,F) in the absence (A–C) or presence (D–F) of 100 μM PD173074. Anterior is to the left. PD98059 (B) or U0126 (C) alone does not affect Tbx6 expression, but in combination with a low dose of PD173074, they eliminate Tbx6 expression (E, F).
Download figure to PowerPoint