Reduced expression of the Nodal co-receptor Oep causes loss of mesendodermal competence in zebrafish

During embryogenesis, the activation of specific gene expression programs results in the generation of functionally differentiated cells. In vertebrates, this process is largely driven by cell signaling (Gilbert, 2010). During the formation of the three germ layers (ectoderm, mesoderm and endoderm), Nodal signaling plays an important role in inducing pluripotent blastomeres to become mesendoderm (Dubrulle et al., 2015; Feldman et al., 2000, 1998; Kunwar et al., 2003; Thisse et al., 2000). Although all cells of the embryo are initially competent to respond to Nodal, blastomeres of the prospective ectoderm lose mesendodermal competence over time (Grainger and Gurdon, 1989; Ho and Kimmel, 1993). The mechanisms underlying this loss of mesendodermal competence are not fully understood.

Given the structure of the Nodal signaling pathway, the loss of competence in the prospective ectoderm could be caused by extracellular, cytoplasmic or nuclear changes. Nodal morphogens (reviewed by Schier, 2009) are extracellular ligands of the TGFβ signaling pathway that act through type I and type II Activin receptors. Upon ligand binding, the receptors phosphorylate cytoplasmic Smad2, which in turn recruits Smad4 and enters the nucleus. Here, the Smad2/4 complex associates with additional transcription factors such as FoxHI and Mixer (Kunwar et al., 2003; Slagle et al., 2011), as well as chromatin modifiers (Dahle et al., 2010; Morikawa et al., 2013; Xi et al., 2011), to regulate the transcription of Nodal target genes. Experiments performed in Xenopus animal caps have suggested that changes in the cytoplasm and nucleus play a role in the loss of mesendodermal competence. Phosphorylation of the linker region of Smad2 results in its nuclear exclusion and, consequently, in the loss of competence (Grimm and Gurdon, 2002). Furthermore, the biological activity of Smad4 in the ectoderm is attenuated by ubiquitylation by Ectodermin (Dupont et al., 2005). In the nucleus, accumulation of the somatic linker histone H1 selectively silences genes required for mesoderm differentiation (Steinbach et al., 1997), an effect that depends on the ratio between histone H3.3 and H1 (Lim et al., 2013). Similarly, trimethylation of histone H3 lysine 27 has been suggested to inhibit mesendoderm differentiation in zebrafish embryos (Shiomi et al., 2017). A role for extracellular factors, however, was not addressed in these studies, and it remains unclear whether these also play a role in the loss of competence.

The extracellular factors required to respond to Nodal signaling include the Activin receptors, as well as Nodal co-receptors from the EGF-CFC family (Cheng et al., 2003; Chu and Shen, 2010; Dorey and Hill, 2006; Fleming et al., 2013; Gritsman et al., 1999; Schier et al., 1997). For example, the zebrafish Nodal co-receptor One-eyed pinhead (Oep; also known as Tdgf1) is essential to mediate the interaction between Nodal ligand and receptor (Gritsman et al., 1999; Zhang et al., 1998), and oep mutants show severe defects in mesendoderm specification (Schier et al., 1997). The expression of oep in the prospective ectoderm decreases during gastrulation (van Boxtel et al., 2015; Zhang et al., 1998), suggesting that a reduction in oep expression might play a role in the loss of mesendodermal competence. In previous studies, this aspect would not have been detected because either Activin or constitutively active Smad2 was used to activate the Nodal signaling pathway. In contrast to Nodal protein-mediated induction, these strategies do not require EGF-CFC co-receptors to activate Nodal signaling (Gritsman et al., 1999). Thus, it remains unclear whether the reduced expression of oep is involved in the loss of mesendodermal competence.

Here we analyzed how mesendodermal competence in zebrafish prospective ectoderm is lost during gastrulation, and show that it is due to a decreasing expression level of the Nodal co-receptor Oep.

To dissect the molecular changes underlying the loss of responsiveness to Nodal signals, we investigated the time window in which competence is lost. It has previously been shown that injection of Nodal protein into the intercellular space can induce Smad2 phosphorylation and mesendodermal gene expression in the prospective ectoderm (Dubrulle et al., 2015). We delivered recombinant human NODAL protein (rhNodal) at 50% epiboly [5.3 hours post fertilization (hpf)] and analyzed the expression of Nodal target genes by in situ hybridization 1.5 h later (Fig. 1A). As expected, rhNodal injection led to ectopic expression of mesendodermal markers no tail (ntl; also known as T or tbxta/b) (Schulte-Merker et al., 1994), sebox (Poulain and Lepage, 2002) and bhikhari (bhik) (Vogel and Gerster, 1999) (Fig. 1B, Fig. S1). In addition to activating Nodal target genes, rhNodal inhibited the expression of the ectodermal markers sox2 and sox3 (Fig. 1C). This is consistent with the known inhibitory effect of Nodal signaling on ectodermal fate (Bennett et al., 2007).

Next, to identify the time point when competence is lost, we injected rhNodal at later developmental stages and analyzed ectopic ntl expression. rhNodal was able to induce ntl expression in animal cap cells until late shield stage, but this ability was lost at 75% epiboly (Fig. 1D). This might explain a previous observation that prospective ectodermal cells are committed to an ectodermal fate at 8 hpf (75% epiboly) (Ho and Kimmel, 1993). In summary, zebrafish prospective ectoderm cells lose their responsiveness to Nodal between shield stage (6 hpf) and 75% epiboly (8 hpf).

Next, we asked whether the loss of Nodal responsiveness at 75% epiboly is caused by extracellular or intracellular changes. We used photoswitchable Activin receptors I and II, which activate the Nodal signaling pathway upon blue light illumination (OptoAcvRs) (Sako et al., 2016) (Fig. 2A). Owing to the lack of extracellular domains, the activation of OptoAcvRs is uncoupled from all extracellular influences. We injected mRNAs encoding the OptoAcvR receptors at the 1-cell stage, bypassing the need for endogenous Activin receptors. Thus, the ability of OptoAcvR-injected embryos to induce transcription of mesendodermal genes depends solely on the cytoplasmic portion of the Nodal signaling pathway and the nuclear context. Upon OptoAcvR injection, we illuminated embryos for 1.5 h starting at 50% epiboly, shield or 75% epiboly stage (see Fig. S2 for a detailed description of this approach). Light stimulation induced ectopic ntl expression at all stages examined, including 75% epiboly, whereas control embryos that were kept in the dark did not show any ectopic ntl expression (Fig. 2B). The ability of OptoAcvR, but not rhNodal, to induce ntl at 75% epiboly suggests that the loss of Nodal responsiveness is caused by changes at the extracellular level.

The Nodal co-receptor Oep is an important competence factor for Nodal signaling. The reduced levels of oep mRNA observed during gastrulation in the prospective ectoderm (van Boxtel et al., 2015; Zhang et al., 1998) suggest that a decrease in the oep expression level might be involved in the loss of competence. To test this hypothesis, we carefully analyzed the spatiotemporal expression pattern of oep during gastrulation (Fig. 3A). This revealed that the decline in oep expression levels within the prospective ectoderm temporally coincides with the loss of responsiveness to rhNodal injection.

To test whether the reduced expression of oep is responsible for the loss of responsiveness to Nodal at this stage, we attempted to induce expression of Nodal target genes by intercellular injection of Activin A, a Nodal-related ligand that does not require Oep to activate the Nodal pathway. We injected Activin A at 50%, shield or 75% epiboly and analyzed embryos for expression of ntl (Fig. 3B), bhik and gsc (Fig. S3) after 1.5 h. Injection of Activin A led to ectopic induction of all genes within the prospective ectoderm at early stages as well as at 75% epiboly. Taken together, these data suggest that loss of Nodal responsiveness in the prospective ectoderm at 75% epiboly is caused by the reduced expression of oep.

If the disappearance of Oep from the prospective ectoderm were causing the loss of responsiveness to Nodal, extending the expression of oep would be expected to prolong the window of competence. We injected mRNA encoding Oep into 1-cell stage embryos and observed elevated levels of oep mRNA throughout gastrulation stages (Fig. S3A, compare with Fig. 3A). We then injected rhNodal into the intracellular space of the animal pole of oep-injected embryos at 50% epiboly, shield or 75% epiboly. As predicted, the oep mRNA-injected embryos ectopically expressed ntl, gsc and bhik at 75% epiboly (Fig. 3C,D, Fig. S3). Control embryos that were injected with egfp instead of oep mRNA did not show ectopic ntl, gsc and bhik induction at 75% epiboly (Fig. 3C,D, Fig. S3). The observation that extending oep expression is sufficient to prolong the window of competence provides a functional demonstration of the central role of Oep levels in determining when mesendodermal competence is lost.

Next, we analyzed how oep is lost in the prospective ectoderm. Our data show that oep expression levels decrease in the prospective ectoderm but remain high in the mesendoderm (Fig. 3A). This suggests two potential mechanisms for the decrease in oep expression in the prospective ectoderm. Either maternal oep mRNA is specifically degraded in the ectoderm, or maternal oep transcripts are degraded throughout the entire embryo and the gene is zygotically transcribed in the mesendoderm. In the latter case, inhibiting zygotic transcription would result in the loss of oep expression in the entire embryo. To test this, we inhibited transcription between dome stage and 60% epiboly using flavopiridol, a drug that inhibits RNA polymerase II elongation (Bensaude, 2011; Yang et al., 2016) (Fig. 4A). We selected this time window because zygotic transcription of oep has started at dome, and at 60% epiboly oep is differentially expressed. No oep expression was detected in flavopiridol-treated embryos, whereas control embryos showed oep expression in mesendoderm, as observed in wild-type embryos (Fig. 4B). To confirm this result, we analyzed the levels of nascent oep transcripts by RT-qPCR at 60% epiboly and observed a significant decrease in flavopiridol-treated compared with untreated embryos (Fig. 4C). These results are in agreement with a previous study showing that zygotic oep deletion mutants completely lack mesendodermal oep expression (Zhang et al., 1998). Thus, the decrease of oep expression in the prospective ectoderm is a consequence of embryo-wide degradation of maternal oep transcripts.

The embryo-wide degradation of maternally loaded oep mRNA raises the question of how this is regulated. The stability of maternally loaded mRNAs in zebrafish is influenced by the presence of miR-430 target sites, RNA methylation, and codon composition (Kontur and Giraldez, 2017). A search for miR-430 target sites in the 3′UTR of oep resulted in only one hit (Fig. S4A), which is much less than generally found in transcripts that depend on miR-430 for their degradation (Giraldez et al., 2006). In agreement with this, our analysis of a previously published dataset (Giraldez et al., 2006) suggests that oep transcripts are not destabilized by miR-430 (Fig. S4B). Similarly, m6A-dependent maternal mRNA clearance does not seem to play a role in the degradation of maternally loaded oep mRNA, as depletion of the m6A-binding protein Ythdf2 does not lead to any increase in oep mRNA (Fig. S4C) (Zhao et al., 2017). The codon composition of the oep coding sequence, however, is among the most destabilizing found in zebrafish genes (Fig. 4D). This suggests that the degradation of oep during early zebrafish embryogenesis is due to the use of codons that destabilize mRNA. Together, these data show that the decrease of oep expression in the prospective ectoderm is a consequence of embryo-wide degradation of oep transcripts, at least partially mediated by an abundance of destabilizing codons.

Our study shows that reduced expression of the Nodal co-receptor oep in the prospective ectoderm results in the loss of mesendodermal competence. We note that our analysis of oep mRNA expression might not reflect the functionally relevant level of Oep protein. However, the observation that the competence window is extended upon oep mRNA overexpression suggests that at 75% epiboly, Oep protein is normally reduced to levels that do not support functional signaling. The initial competence of all blastomeres is provided by the translation of maternal oep mRNA. The mRNA is later degraded throughout the entire embryo, probably owing to the use of codons that reduce mRNA stability. This results in the loss of mesendodermal competence in the prospective ectoderm. Although the localized production and limited reach of Nodal signals, together with the action of Nodal inhibitors, restrict the range of Nodal signaling such that cells in the prospective ectoderm do not usually receive Nodal signal (Bell et al., 2003; Chen and Schier, 2002; Muller et al., 2012; van Boxtel et al., 2015; Wang et al., 2016), the loss of oep might confer robustness to the system. Indeed, although oep overexpression does not result in a dramatic developmental phenotype (Zhang et al., 1998), extending the window of oep expression does extend the expression domain of the direct Nodal target bhik (Fig. S5).

The loss of mesendodermal competence caused by reduced expression of the Nodal co-receptor constitutes a different mechanism to those observed in previous studies, in which changes in cytoplasm and nucleus were shown to play a role in the loss of competence (Dupont et al., 2005; Grimm and Gurdon, 2002; Lim et al., 2013; Shiomi et al., 2017; Steinbach et al., 1997). Our observation that mesendodermal genes can still be activated when Oep is bypassed suggests that, at least in zebrafish, the cytoplasmic and nuclear components of the pathway are still intact when competence to respond to Nodal is lost. Interestingly, we did observe a decrease in the strength of ectopic ntl induction at 75% epiboly, which might suggest that, at this stage, changes in cytoplasmic and nuclear components of the pathway, or the chromatin structure of target genes, also affect the response to Nodal signaling.

Taken together, we show that reduced expression of the Nodal co-receptor oep in the prospective ectoderm results in the loss of mesendodermal competence. The loss of competence as a result of reduced levels of a receptor represents a simple, cell-autonomous mechanism. Because of the very upstream position of receptors in signaling pathways, such a mechanism ensures a global effect on the responsiveness to signals during development. Loss of a receptor has previously been shown to regulate the competence of embryonic cardiomyocytes to undergo Purkinje fiber differentiation during avian cardiogenesis (Kanzawa et al., 2002). Future studies will reveal how widely the loss of a receptor is used to modulate competence during development.

Wild-type (TLAB) Danio rerio embryos were maintained and raised under standard conditions, in accordance with the declaration of Helsinki and the ethical standards of European and German animal welfare legislation.

mRNAs encoding OptoAcvRI and II, Oep and EGFP were synthesized in vitro, purified, and injected into 1-cell stage embryos. Nodal and Activin A protein were obtained commercially and injected into the intercellular space at the indicated stages. For details, including analysis of the oep overexpression phenotype, see the supplementary Materials and Methods.

Embryos were placed in a custom-made illumination chamber (Fig. S2) providing 470 nm LED light (10 µW mm⁻²) in 20 s on/20 s off light pulses for 1.5 h, after which they were fixed. Control siblings were kept in a light-tight box next to the light-induced embryos.

Whole-mount mRNA in situ hybridization was performed as described (Thisse and Thisse, 2008). For details, see the supplementary Materials and Methods.

Transcription was inhibited by soaking embryos in Danieu's medium containing 5 µg/ml flavopiridol (Enzo Life Sciences, ALX-430-161) in 0.1% DMSO. Control embryos were soaked in Danieu's medium containing 0.1% DMSO.

Total RNA was extracted from 25 embryos, genomic DNA removed and cDNA synthesized from 1 µg total RNA. RT-PCRs were performed in technical duplicate. For details, see the supplementary Materials and Methods.

We thank three anonymous reviewers for their valuable input which significantly improved the manuscript, the C.-P. Heisenberg lab (IST Vienna, Austria) for sharing the OptoAcvR constructs before publication, members of the N.L.V. lab for general support and discussions, and Lennart Hilbert, Máté Pálfy, Iain Patten and Carine Stapel for critically reading the manuscript.