A Temporal Window for Signal Activation Dictates the Dimensions of a Nodal Signaling Domain

摘要

class="head no_bottom_margin" id="sec1title">IntroductionIn the development and patterning of embryonic and adult tissues, secreted signaling molecules of the Wnt, Fgf, Hedgehog, and transforming growth factor β (TGF-β) families can act as morphogens to activate different transcriptional programs along a signaling gradient (). Ideas of how morphogens impart spatial information have been dominated by the assumption that these molecules form concentration gradients by diffusion, inducing dose-dependent responses in the receiving field of cells. However, it is becoming increasingly clear that for some ligands, for example, Hedgehog, Wnt, and Fgf, other mechanisms, such as short-range signaling activation, transcriptional feedback, and cellular rearrangements, underlie morphogen function (, , ).Regulation of Nodal signaling in the zebrafish embryo has long served as a paradigm for understanding how morphogens pattern tissues (). Nodals are secreted ligands that belong to the TGF-β superfamily of growth and differentiation factors. During vertebrate development, Nodal is required for stem cell maintenance, specification of mesoderm and endoderm (mesendoderm), and establishment of left-right asymmetry (). Recent work has also suggested that Nodal signaling is reactivated in advanced cancers, where it may be important for self-renewal of cancer stem cells ().Nodal ligands signal through serine/threonine kinase receptor complexes comprising two type I receptors (Acvr1ba [Taram-a]), two type II receptors (Acvr2a/b), and the co-receptor Tdgf1 (Cripto/Oep) (, ). Ligand binding activates the receptors, after which the type I receptor phosphorylates the intracellular signal transducers Smad2 and Smad3, which then bind Smad4 (). These Smad2/3-Smad4 complexes accumulate in the nucleus, where, together with transcription factors such as Foxh1, Mixer, and Oct4, they regulate gene transcription ().At zebrafish blastula stages, two Nodal-related ligands, Ndr1 (Squint) and Ndr2 (Cyclops), specify mesendoderm in marginal cells around the circumference of the embryo by inducing a Smad2-Smad4-Foxh1-dependent transcriptional program (, ). Ndr1/2 are thought to form a signaling gradient by diffusion, extending up to about ten cell tiers from the margin (, , href="#bib42" rid="bib42" class=" bibr popnode">Schier, 2009). Indeed, expression of presumed long-range Nodal target genes such as ta (ntla) and fscn1a suggests low-level signaling up to ten cell tiers from the margin (href="#bib5" rid="bib5" class=" bibr popnode">Bennett et al., 2007). This appears supported by bimolecular fluorescent complementation experiments (href="#bib25" rid="bib25" class=" bibr popnode">Harvey and Smith, 2009). However, other Nodal target genes are expressed in up to five to six cell tiers from the margin, which coincides with nuclear accumulation of Smad2-GFP fusion protein (href="#bib14" rid="bib14" class=" bibr popnode">Dubrulle et al., 2015). Importantly, other signaling pathways, such as Bmp, Wnt, and Fgf, are also active at the margin, which can potentially co-regulate Nodal target genes and thus contribute to their expression domains.Formation of the Nodal signaling domain at the correct time and of appropriate dimensions is thought to be controlled by a reaction-diffusion system (href="#bib34" rid="bib34" class=" bibr popnode">Meinhardt, 2009, href="#bib42" rid="bib42" class=" bibr popnode">Schier, 2009). This model requires positive and negative feedback, which is provided by Nodal-induced expression of both the ligands Ndr1/2 and the antagonists Lefty1 (Lft1) and Lefty2 (Lft2) (href="#bib8" rid="bib8" class=" bibr popnode">Chen and Shen, 2004, href="#bib9" rid="bib9" class=" bibr popnode">Cheng et al., 2004). Besides these feedback mechanisms, the model requires Lft1/2 to be more diffusible than Ndr1/2 (href="#bib36" rid="bib36" class=" bibr popnode">Müller et al., 2012, href="#bib43" rid="bib43" class=" bibr popnode">Schier and Talbot, 2005). These conditions are thought to allow Ndr1/2 to activate signaling at the margin, whereas Lft1/2 proteins would inhibit signaling in more distal cells. Overexpression studies have shown that Ndr1/2 and Lft1/2 can differentially diffuse and that Ndr1, but not Ndr2, can diffuse over a distance to activate signaling (href="#bib6" rid="bib6" class=" bibr popnode">Chen and Schier, 2001, href="#bib7" rid="bib7" class=" bibr popnode">Chen and Schier, 2002, href="#bib36" rid="bib36" class=" bibr popnode">Müller et al., 2012). However, the importance of diffusion of endogenous Ndr1/2 remains unclear, as mesendoderm can develop normally in zygotic ndr1 mutants (href="#bib13" rid="bib13" class=" bibr popnode">Dougan et al., 2003, href="#bib17" rid="bib17" class=" bibr popnode">Feldman et al., 1998, href="#bib30" rid="bib30" class=" bibr popnode">Lim et al., 2013).In addition to the negative feedback provided by Lft1/2, Nodal signaling is regulated by the miR-430/427/302 family of microRNAs (miRNAs) (href="#bib3" rid="bib3" class=" bibr popnode">Bassett et al., 2014, href="#bib10" rid="bib10" class=" bibr popnode">Choi et al., 2007, href="#bib41" rid="bib41" class=" bibr popnode">Rosa et al., 2009). At blastula stages, the miR-430 family is the most abundant family of miRNAs in the zebrafish. Importantly, miR-430 regulates ndr1, lft1, and lft2, but not ndr2, and this is thought to dampen Nodal signaling (href="#bib10" rid="bib10" class=" bibr popnode">Choi et al., 2007). However, to what extent miR-430s contribute to the formation of the Nodal signaling domain is unknown.To develop a specific readout for endogenous Nodal signaling, avoiding overexpression of any pathway components, we generated a transgenic zebrafish Nodal reporter line. Using this line combined with immunofluorescence for phosphorylated Smad2 (P-Smad2), we show that Nodal signals exclusively in cells that express Ndr1/2, up to five to six cell tiers from the margin. This prompted us to revisit the mechanism underlying the formation of the Nodal signaling domain. Our data do not support the reaction-diffusion model, but instead, we propose that Nodal activates signaling in a temporal window that is defined by a miR-430-mediated delay of Lft1/2 translation. In this way, temporal information is converted into spatial information in the developing embryo.