A spatiotemporal barrier formed by Follistatin is required for left–right patterning

Significance

The left–right patterning of vertebrate embryos relies on an orderly propagation of the asymmetric Nodal signaling from posterior to anterior. This requires a left–right organizer (LRO) that is located at the posterior tip of the embryo to break the bilateral symmetry, as well as a midline barrier that prevents Nodal from diffusing along the left–right direction. The current study identifies a new barrier, which is set up by Follistatin (Fst) and expressed bilaterally. This barrier gates the propagation of the asymmetric Nodal activity toward the anterior to ensure a proper left–right patterning outcome throughout the entire embryo including the forebrain.

Abstract

How left–right (LR) asymmetry emerges in a patterning field along the anterior–posterior axis remains an unresolved problem in developmental biology. Left-biased Nodal emanating from the LR organizer propagates from posterior to anterior (PA) and establishes the LR pattern of the whole embryo. However, little is known about the regulatory mechanism of the PA spread of Nodal and its asymmetric activation in the forebrain. Here, we identify bilaterally expressed Follistatin (Fst) as a regulator blocking the propagation of the zebrafish Nodal ortholog Southpaw (Spaw) in the right lateral plate mesoderm (LPM), and restricting Spaw transmission in the left LPM to facilitate the establishment of a robust LR asymmetric Nodal patterning. In addition, Fst inhibits the Activin–Nodal signaling pathway in the forebrain thus preventing Nodal activation prior to the arrival, at a later time, of Spaw emanating from the left LPM. This contributes to the orderly propagation of asymmetric Nodal activation along the PA axis. The LR regulation function of Fst is further confirmed in chick and frog embryos. Overall, our results suggest that a robust LR patterning emerges by counteracting a Fst barrier formed along the PA axis.

Left–right (LR) patterning of the embryo is under the regulation of a transient organ, the LR organizer (LRO), (named “node” in mice, “Hensen’s node” in chicks, or “Kupffer’s vesicle” (KV) in zebrafish), which is located at the most posterior end of the midline (1). In either cilia-dependent or cilia-independent manner, the LRO breaks bilateral symmetry and confers a messenger function for LR asymmetry to Nodal, which then begins to be expressed in the lateral plate mesoderm (LPM). Confined by the midline barrier, Nodal then propagates from posterior to anterior (PA) in the left LPM, progressively instructing the asymmetric position of tissues and organs derived from any of the three germ layers, such as the heart (mesoderm), liver/pancreas/gut (endoderm) and the forebrain (neural ectoderm) (2). Thus, the spatiotemporal accuracy of this PA Nodal wave is essential for the correct LR asymmetrical patterning of the developing tissues and organs, especially for the brain, which is located at the anterior most end of the embryo (3–6). However, little is currently known about how the regulation of the propagation and transmission of Nodal signaling occurs to ensure it runs correctly from a PA direction.

LR patterning of the developing brain plays important roles in the anatomical and functional asymmetry of the central nervous system. A mechanism incorporating and integrating the LRO, the midline, and Nodal signaling, is involved in the asymmetric gene expressions in both the LPM and the brain. In zebrafish, activation of lateralized Nodal pathway genes in the forebrain is dependent upon the activity of the nodal ligand Spaw which emanates from the left LPM (7). However, some studies have suggested that the laterality of the LPM and embryonic brain can be uncoupled (4, 8). This has led to the proposal that the expression of Nodal pathway genes in the forebrain is initially repressed by early Nodal signaling and that this repression is then removed later by Nodal emanating from the left LPM (4). Despite this suggestion, mechanistically how early Nodal, left-sided Nodal in the LPM, and related signals in the embryonic brain, all coordinate to establish the laterality of the forebrain remains enigmatic.

Follistatin (Fst), which is a member of the transforming growth factor-β (TGF-β) superfamily, was initially identified as an inhibitor of follicle-stimulating hormone. It plays important roles in many developmental and disease processes such as oogenesis, muscle growth, and tumor development, via its ability to neutralize the activity of multiple TGF-β family members including Activin, Bone Morphogenetic Protein (BMP), and Myostatin (9–11). In this study, we found that the expression of follistatin a (fsta) in zebrafish depends on the activity of Nodal. Either in the morphants of the Nodal ligands nodal related 1 (ndr1) and nodal related 2 (ndr2), or in Nodal coreceptor mutations, the expression of fsta is dramatically decreased. By directly binding and competitively blocking the interaction with its receptors, Fst can neutralize the activity of the LR regulator Spaw, and restricts the propagation and transmission of Spaw on both sides of the embryonic body. Fst loss-of-function results in bilateral expression of spaw in the LPM and precocious bilateral expression of lefty1 in the forebrain. In addition, we found that the LR patterning of the zebrafish forebrain relies on the prior spatiotemporal inhibitory effect of Fst on the Activin-Ndr2-Lefty1 cascade. Furthermore, as Fst also has a similar inhibitory effect on the left and right nodal expression in chick and frog embryos, our finding likely represents a highly conserved mechanism in vertebrates, bridging gaps in our understanding of the roles of Nodal signaling from early embryonic stages that acts toward the establishment of differential LR patterning of the entire embryo.

Results

fsta Is Bilaterally Expressed in the LPM and its Expression Requires the Nodal Activity in Zebrafish.

Follistatin is a conserved extracellular modulator of the TGF-β family among vertebrates (SI Appendix, Fig. S1 A and B). Its expression can be detected by Whole-mount in situ hybridization (WISH) from the early gastrulation stage in zebrafish. During late gastrulation and early somitogenesis, fsta is specifically and bilaterally expressed in the LPM and somites, and it is highly enriched in the head region at organogenesis stages (Fig. 1A and SI Appendix, Fig. S1C). Our previous work has identified fsta as a downstream gene of Nodal signaling (SI Appendix, Fig. S2 A–E) (12). To validate this hierarchical relationship between nodal and fsta, we injected the messenger RNA (mRNA) of a Nodal ligand ndr2, into one blastomere of the animal pole at the 128-cell stage. We found that fsta could be ectopically induced by ndr2. In addition, double knockdown of ndr1 and ndr2 resulted in a dramatic reduction of fsta expression, although the single knockdown of ndr1 or ndr2 had no apparent effect. Similarly, fsta expression was dramatically or completely diminished in the zygotic or maternal zygotic mutants of the Nodal co-receptor, Tdgf1 (Fig. 1B). These results suggested that fsta is downstream of Nodal and its expression relies on the early Nodal activity.

Fig. 1.

fsta is downstream of Nodal signaling and required for LR asymmetry of zebrafish. (A) WISH of fsta in 10 hours post-fertilization (hpf) zebrafish embryos. (B) From left to right: expression of fsta in zebrafish embryos injected with ndr2 mRNA in one blastomere of the animal pole at the 128-cell stage, or ndr1, ndr2, ndr1 and ndr2 MO at the 1-cell stage, and in tdgf1 zygotic mutant or maternal-zygotic mutant zebrafish embryos. (A and B) Numbers in the bottom right of panels indicate the number of embryos with the phenotype shown out of the total number of embryos examined. At least 34 embryos were examined in each experiment. (C) Schematic diagram showing the positions of ATG (translation start site) and splicing MO of fsta; the guide RNA(gRNA) target site; the deletions of two fsta mutant lines generated using CRISPR/Cas9; and the predicted truncated Fsta (Δ5/Δ4) proteins (predicted domains taken from the Uniprot database). (D) The positions of heart and cardiac looping of WT and fsta^Δ5/Δ5 embryos were analyzed using the cardiomyocyte marker myl7. The LR patterning of digestive organs including the liver, pancreas and intestines were analyzed using foxa3. (E) The epithalamic LR patterning of WT and fsta^Δ5/Δ5 embryos was analyzed using the parapineal marker gfi1ab, pineal and parapineal marker otx5, and habenula markers kctd12.1, kctd12.2, and slc18a3b. Red triangles indicate asymmetric expression. Grey stacks indicate absence of expression. (D and E) Statistics are shown on the right of the representative photos, “n” indicates the numbers of examined embryos. At least 49 embryos were examined in each experiment. DV, dorsal view; LV, lateral view. (Scale bar, 200 μm.)

fsta Is Required for the LR Patterning of Zebrafish.

To investigate the function of Fst during early development, we firstly injected fsta mRNA into zebrafish embryos at the 1-cell stage, which resulted in a severe dorsalized phenotype (SI Appendix, Fig. S3A). We then performed loss-of-function studies of fsta by morpholino knockdown and mutations generated by CRISPR/Cas9 technology. Using a single gRNA, we obtained two mutant zebrafish lines that contained a 5bp (base pair) and a 4bp deletions at the same position of the fsta gene, which we named fstaΔ5 and fstaΔ4, respectively. Both mutations resulted in frameshifts (48 amino acids and 10 amino acids, respectively), early stop codons, and predicted truncated proteins that lacked several of the C-terminal domains (Fig. 1C). Meanwhile, we found that fsta mRNA levels in both mutant fish lines were decreased relative to wild-type (WT) (SI Appendix, Fig. S3 B and C), possibly due to nonsense-mediated mRNA decay (13). However, no obvious dorsal-ventral defects were observed in either fsta morphants or mutants (SI Appendix, Fig. S3A).

Decades ago, Dr. Michael Levin proposed that FST may participate in the LR patterning in the chick embryo (14). However, this hypothesis has so far remained untested. Interestingly, in zebrafish fsta^Δ5/Δ5 mutants, some embryos (10 to 20%) showed heterotaxia of visceral organs and heart situs (Fig. 1D), and a large number of embryos (30 to 40%) exhibited complete situs inversus of the central nervous system (Fig. 1E). We next assayed whether the key LR markers were affected by the Fsta mutations. Surprisingly, the Nodal signaling pathway genes, including spaw in the LPM, and ndr2 and lefty1 in the forebrain, became bilaterally expressed in all fsta^Δ5/Δ5 embryos (Fig. 2A). Quantitative analysis showed that, in addition to the ectopic expression in the right LPM, spaw expression in the left LPM was also elevated in fsta^Δ5/Δ5 embryos compared to WT, leading to a bilateral but still LR asymmetric weighted expression of spaw in fsta^Δ5/Δ5 embryos (Fig. 2B). These LR asymmetry defects resulting from Fsta loss-of-function were further verified by the knockdown of fsta using an ATG morpholino and a splicing morpholino and confirmed in another fsta mutant (fsta^Δ4/Δ4), except that only 40% of the fsta^Δ4/Δ4 embryos showed bilateral spaw expression (SI Appendix, Fig. S4 A and B).

Fig. 2.

Loss of or gain of function of fsta affects the LR patterning of the Nodal signaling pathway in zebrafish. (A) LR asymmetric genes spaw in the LPM, lefty1 (lft1) and ndr2 in the forebrain analyzed using WISH in WT and fsta^Δ5/Δ5 embryos. Grey stacks indicate absence of expression. At least 22 embryos were examined in each experiment. (B) Expression areas of spaw in WT and fsta^Δ5/Δ5 embryos were calculated. myod1 was used to indicate the developmental stage. **P < 0.01. At least 10 embryos were examined in each experiment. (A and B) Statistics are shown on the right of the representative photos, “n” indicates the numbers of examined embryos. (C) Schematic diagram of experimental setup for (D) expression of spaw and lefty1 in the light-induced fsta overexpression embryos. (E) Schematic diagram of experimental setup for (F) expression of spaw in zebrafish embryos implanted with Bovine Serum Albumin (BSA)- or rhFST-soaked beads in the left side at the indicated stages. Red triangles indicate the anterior boundary of spaw expression. (D and F) Numbers in the bottom right of panels indicate the number of embryos with the phenotype shown out of the total number of embryos examined. At least a total of 22 embryos were examined in each experiment. (Scale bar, 200 μm.)

The elevated expression of Nodal signaling pathway genes in fsta mutants implied that Fst may function as a negative regulator of Nodal signaling in LR patterning. To test this hypothesis, we constructed a fsta light-inducible transgenic zebrafish line using TAEL-N (TA4-EL222-N), an optogenetic gene expression system (15), to avoid the early effects of fsta overexpression (Fig. 2C). We found that overexpression of fsta from 12 h postfertilization (hpf) led to a dramatic decrease in both spaw expression in the LPM and lefty1 expression in the forebrain (Fig. 2D). In addition, the implantation of a rhFST (recombinant human FST protein) bead in the left LPM resulted in a decrease in spaw expression, suggesting that locally ectopic Fsta could also counteract and restrict the PA propagation of spaw (Fig. 2 E and F).

fsta Mutation Leads to Precocious Expression of spaw in the LPM and of lefty1 in the Forebrain.

Previous work has shown that the temporal control of the PA propagation of spaw is essential for the LR asymmetry of the brain (3). The observations of elevated expression of spaw and lefty1 prompted us to ask whether the timing of the expression of these two genes also changes in fsta mutants. Using myod1 as an indicator of developmental stages, we found that spaw expression could be detected in the LPM by WISH as early as at 12ss stage in some fsta^Δ5/Δ5 embryos, while there was no detectable LPM spaw expression in WT embryos at this stage (Fig. 3A). On average, from stages 13ss to 17ss the anterior boundary of spaw expression in fsta^Δ5/Δ5 preceded in a 1 to 2 somite advanced location to that observed in the WT at the same stages (Fig. 3B). In addition, the lefty1 expression in the forebrain could be detected in more than 20% of the fsta^Δ5/Δ5 embryos at the 17-18ss stage, while in WT lefty1 such expression could be detected in 10% of embryos even at stage 19-20ss (Fig. 3C). At the 21-22ss stage, by which time almost all fsta^Δ5/Δ5 embryos were expressing lefty1 bilaterally in the forebrain, we could still only detect the lefty1 expression by WISH in less than 50% of the WT embryos (Fig. 3C). Overall, the above results suggest that Fsta functions as a negative regulator, not only in the spatial patterning of the LR asymmetric Nodal signaling genes, but also in negatively regulating their temporal expressions.

Fig. 3.

Fsta restricts the transmission of spaw in the LPM and prevents early activation of lefty1 in the forebrain of zebrafish. (A) PA propagation of spaw from 12 to 17-somite stage (ss) in WT and fsta^Δ5/Δ5 embryos. myod1 is used to indicate the development stage. Red triangles indicate the anterior boundary of spaw expression in the left LPM. Blue triangles indicate the anterior boundary of spaw expression in the right LPM. (B) The violin plot shows the statistics of (A), rectangles indicate the middle two quadrants in the distribution and black dots indicate the median. Negative numbers indicate that the anterior boundary of spaw has crossed the first somite. The ratio and numbers of embryos that have spaw expression in the left LPM of WT and fsta^Δ5/Δ5 at 12ss are shown in the histogram. (C) Forebrain expression of lefty1 in WT and fsta^Δ5/Δ5 embryos at the indicated stages. Statistics are shown on the right, “n” indicates the numbers of examined embryos. At least 20 embryos were examined in each experiment. (Scale bar, 200 μm.)

Fsta Is a Secreted Protein That Can Physically Interact with Spaw in Zebrafish.

The correct spatiotemporal patterning of LR asymmetric Nodal signaling relies on the LRO (16) and midline barrier (17). However, we were unable to find any obvious defects in the KV or the midline of fsta^Δ5/Δ5embryos, either by morphological analysis, WISH of the key KV gene dand5 (18), immunostaining of the cilia marker acetylated α-tubulin (Ac-tubulin) (19) (Fig. 4A and SI Appendix, Fig. S5 A–C), or the WISH of midline markers lefty1 (20), tbxta (21) or tdgf1 (22) at differing stages (Fig. 4B and SI Appendix, Fig. S5D).

Fig. 4.

Zebrafish Fsta is a secreted protein that can physically interact with spaw. (A) Projection of images from confocal stacks to show cilia (acetylated tubulin, red) in the KV of WT and fsta^Δ5/Δ5 embryos. DAPI staining to show the nuclei of KV cells. (B) WISH of spaw and lefty1 to show the LR patterning and midline in WT and fsta^Δ5/Δ5 embryos. (C) Interaction of Fsta-Flag/FstaΔ5-Flag with Enhanced Green Fluorescent Protein (EGFP) -Spaw in zebrafish embryos as assessed using Co-IP. (D) Schematic diagram of experimental setup for (E) in vivo interaction between Fsta-TurboID or FstaΔ5-TurboID with EGFP-Spaw as analyzed using proximity-dependent biotinylation catalysis. A hollow triangle indicates proprotein, a black triangle indicates mature protein of EGFP-Spaw. (F) Synthesis and secretion analysis of Fsta-Flag and FstaΔ5-Flag in HEK293T cells after transfection with Fsta-Flag or FstaΔ5-Flag plasmid for 48 h. Cells and conditioned medium (CM) were collected for western blot separately. (G) Schematic diagram of experimental setup for (H) projection of images from confocal stacks for Fsta secretion analysis. fsta-mCherry or fstaΔ5-mCherry + H2B-GFP mRNA were injected into one blastomere on the animal pole at the 128-cell stage and then imaged at the sphere stage. Secreted WT Fsta-mCherry can be seen in the extracellular spaces throughout the embryo, while almost no extracellular signaling of FstaΔ5-mCherry could be detected. (A, B, and H) Numbers in the bottom right of panels indicate the number of embryos with the phenotype shown out of the total number of embryos examined. At least 10 embryos were examined in each experiment. (Scale bar, 50 μm.)

To better understand the possible position of Fsta in the hierarchy of LR patterning, we firstly injected the morpholino of myo1d (23), the function of which is essential for the KV, into both WT and fsta^Δ5/Δ5 embryos. We found that knockdown of myo1d resulted in the absence of spaw in both the WT and fsta^Δ5/Δ5 embryos (SI Appendix, Fig. S5E). As Fst often functions as a negative modulator of TGF-β family members (9, 24), we then investigated whether inhibitors of BMP or Nodal, both involved in LR asymmetry patterning, could compensate for the loss-of-Fst function. Consistent with previous reports (25, 26), treatment with a BMP inhibitor resulted in the bilateral expression of spaw in some WT embryos, but without any obvious effects on the bilateral expression of spaw in fsta^Δ5/Δ5 embryos. Interestingly, treatment with the Nodal inhibitor could restore the unilateral spaw patterning for most fsta^Δ5/Δ5 embryos (SI Appendix, Fig. S5F). The above results suggest that Fsta may function downstream of KV, parallel with Spaw, and have possible direct interactions with Nodal.

Spaw propagation and transmission in the LPM rely on its auto-regulatory loop and interaction with its partner Gdf3, also a TGF-β family member (20). Thus, we hypothesized that Fsta may negatively regulate Spaw signaling either by binding to Gdf3 to restrict Spaw transmission, or by binding to Spaw itself to impede its activity. Although gdf3 morpholinos (MO) could restore the unilateral expression of spaw in a few fsta^Δ5/Δ5 embryos (SI Appendix, Fig. S6A), co-immunoprecipitations (Co-IP) assay showed that Fsta did not directly interact with Gdf3 (SI Appendix, Fig. S6 B–E). Instead, both Co-IP and in vivo proximity-dependent catalysis assays (27) showed that Fsta could physically interact with Spaw (Fig. 4 C–E and SI Appendix, Fig. S7 A–C). However, to our surprise, the truncated version of Fsta that exactly mimic FstaΔ5 (Fig. 1C) could still interact with Spaw (Fig. 4 C–E and SI Appendix, Fig. S7A). Interestingly, we found that FstaΔ5 secretion was dramatically diminished in both HEK293T cells and zebrafish embryos, while WT Fsta secretion was prominent supporting a subsequent long-range diffusion away from its source cells (Fig. 4 F–H). As the secreting signal peptides of Fst have been previously reported to be located at its N-terminus (28), it is unlikely that this secretory defect of FstaΔ5 could be caused by C-terminal truncation. Instead, it is more likely that this had occurred due to the extra frameshifted amino acids. To test this possibility, we analyzed another fsta mutant, FstaΔ4, which contains fewer frameshifted amino acids of distinct identities (Fig. 1C). Examined in HEK293T cells and zebrafish embryos, we found that FstaΔ4 could be secreted and interact with Spaw (SI Appendix, Fig. S7 D–F). Structural modeling of the Fsta–Spaw complex using AlphaFold-Multimer revealed a large buried surface area (1,379 Å²) for the WT Fsta and a medium-buried surface area (750 Å²) for the FstaΔ4 (SI Appendix, Fig. S7 G and H), indicating a reduced binding affinity for the truncated Fsta. Collectively, these results suggested that in fsta^Δ5/Δ5 embryos, the defective secretion of FstaΔ5 resulted in a complete loss of the extracellular function of Fsta. This leads to all fsta^Δ5/Δ5 embryos showing a bilateral spaw expression phenotype (Fig. 2A). While in fsta^Δ4/Δ4 embryos, the truncated but secreted FstaΔ4 protein could still interact with Spaw, hence a weaker phenotype of bilateral spaw expression (less than 40% embryos, SI Appendix, Fig. S4B). This weak phenotype was likely due to the impaired affinity of FstaΔ4 to Spaw and decreased fsta mRNA resulted from nonsense-mediated mRNA decay (SI Appendix, Fig. S3 B and C).

Fsta Regulates Forebrain LR Patterning by Modulating Spaw in the LPM and Activin in the Forebrain.

Inconsistencies in bilateral spaw expression in the LPM (<40% embryos) and lefty1/ndr2 bilateral expression in the forebrain (100% embryos) of fsta^Δ4/Δ4 embryos (SI Appendix, Fig. S4B) suggested an involvement of another unknown factor, the activity of which can be suppressed by Fsta and can induce the expression of Nodal pathway genes independent of Spaw. To test this possibility, we firstly injected spaw MO in WT and fsta^Δ5/Δ5 embryos. Consistent with the previous study (7), loss of Spaw function eliminated the forebrain Nodal pathway gene expression in WT embryos, while in fsta^Δ5/Δ5 embryos, the expression of lefty1 and ndr2 in the forebrain remained bilateral (Fig. 5 A and B). Such a case of uncoupled LPM and forebrain Nodal signaling has been reported previously (4, 8). However, the specific factor that had activated forebrain Nodal signaling remained unclear. As BMP and Activin are two well-established targets of Fst (9, 11), we tested whether these two signaling pathways could respond to the activation of Nodal signaling in the forebrain by Morpholino knockdown. bmp4 MO resulted in the bilateral expression of lefty1 and ndr2 in a small number of the WT embryos. We considered that this might be associated with the previous observations of bilateral spaw expression occurring due to loss of BMP4 function (26). However, injecting bmp4 MO in fsta^Δ5/Δ5 embryos failed to affect the bilateral forebrain lefty1 and ndr2 expression (SI Appendix, Fig. S8 A and B).

Fig. 5.

Fsta regulates LR asymmetry of the zebrafish forebrain by inhibiting Activin. (A and B) Expression of lefty1 (A) and ndr2 (B) in the forebrain in WT and fsta^Δ5/Δ5 embryos injected with spaw, inhbb, spaw + inhbb MO or smad2 MO. Uninjected embryos were used as control. Statistics are shown below, “n” indicates the numbers of examined embryos. At least 34 embryos were examined in each experiment. Yellow stacks indicate right side expression of ndr2. (C and D) Interaction of Fsta-Flag/FstaΔ5-Flag/FstaΔ4-Flag with HA-Inhbb in zebrafish embryos as assessed using Co-IP. (Scale bar, 100 μm.)

For Activin signaling, we tested its two main subunits in zebrafish, inhbaa and inhbb (29), which are expressed in the head region during early development (SI Appendix, Fig. S8C). Interestingly, knockdown of inhbb restored the unilateral expression of forebrain lefty1 and ndr2 in about 50% of the fsta^Δ5/Δ5 embryos, even resulting in the complete absence of the expression for these two genes in the forebrain of ~ 20% of WT or fsta^Δ5/Δ5 embryos (Fig. 5 A and B). Meanwhile, we failed to observe any obvious effect on the forebrain lefty1 and ndr2 expression by inhbaa MO for either WT or fsta^Δ5/Δ5 embryos (SI Appendix, Fig. S8 A and B).

Considering the bilateral expression of inhbb and fsta, it seemed that the unilateral expression of Nodal signaling genes in the forebrain required asymmetric spaw in the LPM. Consistent with this hypothesis, knocking down both spaw and inhbb fully diminished lefty1/ndr2 expression in all WT embryos and most of the fsta^Δ5/Δ5 embryos (Fig. 5 A and B), suggesting a synergistic function of Spaw and Inhbb to overcome the inhibition of Fsta on Nodal signaling in the forebrain. Consistent with this finding, knocking down the common intracellular mediator of the Activin/Nodal signaling, Smad2, in WT and fsta^Δ5/Δ5 embryos also blocked the activation of the Nodal signal in the forebrain (Fig. 5 A and B). We noticed that, even with the combination knockdown of spaw and inhbb, a fraction of fsta^Δ5/Δ5 embryos still showed bilateral expression of lefty1 and ndr2, and this phenotypic heterogeneity was likely due to the efficiency of morpholino injection (30).

Meanwhile, we did not observe any obvious differences in inhbb expression levels between the WT and the fsta^Δ5/Δ5 mutants (SI Appendix, Fig. S8C), and the knocking down of inhbb had no measurable effect upon the expression of spaw in the LPM (SI Appendix, Fig. S8D). This suggested that Fsta regulated Inhbb activity may occur through physical interactions at the protein level. As examined by Co-IP assay, we did find that Fsta, but not FstaΔ5 or FstaΔ4, could interact with Inhbb (Fig. 5 C and D and SI Appendix, Fig. S8E). This could potentially explain the 100% bilateral lefty1/ndr2 forebrain expression phenotypes in both fsta^Δ5/Δ5 and fsta^Δ4/Δ4 embryos.

Extracellular Fsta Can Robustly Inhibit Nodal/Activin Activity.

The preserved interaction of FstaΔ5 and FstaΔ4 with Spaw, but not with Inhbb, prompted us to explore if there could be a binding preference for the N- or C-terminal sequences of Fsta for either Spaw or Inhbb. To address this question, we constructed another truncated version of Fsta in which the TGF-β binding domain had been deleted (FstaΔTB, SI Appendix, Fig. S9A). We found that removal of the TGF-β binding domain did not affect the secretion or diffusion of FstaΔTB in either HEK293T cells or zebrafish embryos (SI Appendix, Fig. S9 B and C). Interestingly, Co-IP analysis revealed that FstaΔTB had lost its interaction with Spaw, but preserved its interaction ability with Inhbb (SI Appendix, Fig. S9 D and E). In summary, these results reveal that the N terminal (TGF-β binding domains) of Fsta were essential for its interaction with Spaw, but it is the C terminal domains that were important for its interaction with Inhbb (SI Appendix, Fig. S9F).

We then investigated whether the interaction of Fsta with Spaw or Inhbb could impede their activities. Using lefty1 and ndr2 as readouts of Nodal/Activin activity, we found that both spaw and inhbb mRNA could induce ectopic expression of lefty1 and ndr2 by injection, either at the 1-cell stage or into one blastomere at the 128-cell stage. This induced expression of lefty1 and ndr2 could be dramatically reduced by fsta mRNA injection at either 128-cell stage or1-cell stage, respectively (SI Appendix, Fig. S10 A and B). To further confirm the extracellular effect of Fsta on Nodal/Activin, we carried out a cell grafting assay of zebrafish embryos (Fig. 6A). We found that donor cells that contained WT Fsta could robustly inhibit the ectopic lefty1 expression induced by Spaw or Inhbb in the host embryos, compared to EGFP controls (Fig. 6B). Consistent with previous Co-IP results, FstaΔ5 lost this inhibitory activity on Spaw and Inhbb, while FstaΔ4 retained partial inhibitory activity on Spaw, but not on Inhbb. Conversely, FstaΔTB could still dramatically reduce the lefty1 expression induced by Inhbb, but had no such effect upon Spaw (Fig. 6B). In summary, these results suggested that Fsta could block Nodal and Inhbb activities by its physical interaction with each, in a cell nonautonomous manner.

Fig. 6.

Fsta antagonizes the activity of Spaw and Activin by blocking binding to their receptors. (A) Schematic diagram of experimental setup for (B) WISH of lefty1 is used for a readout of Nodal/Activin activity. Immunostaining of EGFP shows grafted cells. Donor cells from embryos injected with EGFP mRNA or + fsta or truncated fsta mRNA (Δ5/Δ4/ΔTB) were grafted into host embryos that were overexpressed with spaw or inhbb. WT embryos were used as control. Numbers in the bottom right of panels indicate the number of embryos with the phenotype shown out of the total number of embryos examined. At least 20 embryos were examined in each experiment. (C) Interaction of Fsta–Flag, EGFP–Spaw, and HA–Tdgf1 in zebrafish embryos as assessed using competitive protein binding assay. (D) Interaction of EGFP–Spaw with its candidate receptors in zebrafish embryos as assessed using Co-IP. (E) Interaction of Fsta–Flag, EGFP–Spaw, and Acvr2ab–HA in zebrafish embryos as assessed using competitive protein binding assays. (F) Interaction of Fsta–Flag, EGFP–Spaw, and Acvr1bb–HA in zebrafish embryos as assessed using competitive protein binding assays. (Scale bar, 100 μm.)

Fsta Antagonizes Nodal Activity by Blocking Binding to its Receptor.

A previous study has shown that Fst inhibits Activin activity by blocking the binding to its receptors (31). To investigate how Fsta inhibited Nodal activity, we performed a competitive protein binding assay in zebrafish embryos. We found that Fsta could block the interaction between Spaw and Tdgf1 (Fig. 6C), a co-receptor of Nodal signaling (32), in a concentration-dependent manner, and that this blocking was not dependent on the physical interaction between Fsta and Tdgf1 (SI Appendix, Fig. S10C). In addition to the co-receptor Tdgf1, Nodal signaling transduction also requires the ligands to bind with its type II and I receptors. There are several paralogue receptors of Nodal signaling in zebrafish (33, 34). As it was unknown which of those receptors can robustly bind with Spaw, we firstly conducted a screening using Co-IP assays in zebrafish embryos. We found that the type II receptor Acvr2ab and type I receptor Acvr1bb showed high affinities to Spaw (Fig. 6D). Then through the use of competitive protein binding assay, we found that Fsta also could block Spaw’s interaction with Acvr2ab and Acvr1bb in a concentration-dependent manner (Fig. 6 E and F), while Fsta could not directly interact with Acvr2ab or Acvr1bb (SI Appendix, Fig. S10 D and E). These data provided evidence that Fsta antagonizes Nodal activity by blocking the interaction between the Nodal ligand and its receptors (SI Appendix, Fig. S10F). Further analysis showed that Fsta could also block the interaction between Spaw and Gdf3 (SI Appendix, Fig. S11A), which could lead to restricted Spaw diffusion (SI Appendix, Fig. S11 B and C).

Fst Is Required for the LR Patterning in the Chick and Frog.

The highly conserved protein sequence and function of Fst prompted us to ask whether the LR asymmetry regulation function of Fst is also conserved in vertebrates other than zebrafish. WISH results showed that chick FST was also expressed bilaterally in the LPM at stage 6, then explicitly expressed in the somites (Fig. 7A). Electroporation of FST MO in chick embryos resulted in bilateral expression of NODAL (Fig. 7B). The implantation of rhFST soaked beads at the left side of developing chick embryos at stage 6 could eliminate NODAL expression in the LPM (Fig. 7 C and D). We then checked the expression pattern of fst in the Xenopus tropicalis. WISH results showed that Xenopus fst was expressed in the dorsal organizer at stage 10.5, and then enriched in the anterior notochord, midbrain, hindbrain, and somites at stage 18 (Fig. 7E) (35). One advantage of the Xenopus is that owing to holoblastic development, Xenopus permits injection at the 2–4 cell stage to assess the assignment of gene function in left versus right lineages (36). We performed gain- and loss-of-function studies of Fst by unilaterally injecting fst CRISPR gRNAs plus Cas9 protein or fst mRNA together with FITC-dextran into X. tropicalis embryos at the two-cell stage, and then distinguished by fluorescence on left or right at stage 20 to 24 (Fig. 7F). The editing efficiency of fst gRNAs was confirmed using the Synthego ICE software tool (SI Appendix, Fig. S12). We found that injection of fst gRNAs and Cas9 on the right side of the embryos resulted in the bilateral expression of nodal1. However, nodal1 expression did not change significantly when fst was downregulated specifically on the left side (Fig. 7 G and H). In addition, overexpressing fst mRNA on the left side led to the absence of nodal1, while almost no differences were observed when fst mRNA was injected on the right side (Fig. 7 I and J). Altogether, these results document an evolutionarily conserved function of Fst in governing LR asymmetry across these different vertebrates (Fig. 8).

Fig. 7.

Fst is required for the LR patterning of Chick and Xenopus. (A) Expression pattern of FST in chick embryos at indicated stages. (B) Expression of NODAL in WT and FST MO transfected chick embryos at the indicated stages. (C) Schematic diagram of experimental setup for (D) expression of NODAL in chick embryos implanted with BSA- or rhFst-soaked beads on the left side of the embryos at the indicated stages. (E) Expression patterns of fst in Xenopus embryos at indicated stages. (A and B, D and E) Numbers in the bottom right of panels indicate the number of embryos with the phenotype shown out of the total number of embryos examined. At least four embryos were examined in each experiment. (F) Schematic diagram of experimental setup for (G) expression of nodal1 in Xenopus tropicalis embryos injected with fst gRNAs + Cas9 or (I) with fst mRNA into one cell at the two-cell stage. Uninjected embryos were used as control. Red triangles indicate the expression of nodal1. (H and J) Statistics of (G) and (I), “n” indicates the numbers of examined embryo. At least 41 embryos were examined in each experiment. (Scale bar, 200 μm.)

Fig. 8.

Working model of Fst function as a molecular barrier of LR asymmetry in the LPM (Zebrafish, Chick, and Frog) and Epithalamus (zebrafish). Briefly, Nodal induced fst to express bilaterally on both sides of the developing embryos. Under the function of LRO, Nodal signaling is weaker in the right LPM where it is completely blocked by Fst, while stronger Nodal in the left LPM can overcome the inhibition of Fst and spread from PA regions, though still restricted as it does so by Fst. In zebrafish, Activin could induce ndr2/lefty1 expression in the forebrain, but this is then suppressed by Fst at an early stage. When Spaw arrives, as emanated from the left LPM, this helps overcome the suppression of Fst so that ndr2/lefty1 can be turned on in the left side of the forebrain.

Discussion

In this study, we initially found that the expression of fsta was revealed to depend on early Nodal activity and was located in the relatively anterior part of both left and right LPM, on the path of Nodal propagation from a posterior to anterior direction. The removal of fsta resulted in both spatially ectopic expression of spaw in the right LPM and temporally precocious expression of spaw in left LPM and lefty1 in the forebrain. Overexpression of fsta, either globally or locally, both caused propagation defects of spaw. Over all, our work suggests that, in addition to the midline barrier which separates the left and the right regions, Fst is shown to have an integral role in the setting up of another barrier that runs from posterior to anterior (Fig. 8) and this barrier is required for the generation of a robust LR asymmetric pattern in vertebrates.

LR patterning, and the corresponding placement of the relative asymmetric positions of different organs, depends on Nodal activity occurring sequentially and at precisely the correct times and locations. During embryonic development, morphogens and their antagonists often form opposing gradients. Examples of this include BMP and Chordin/Noggin in their roles in dorsal–ventral patterning (37), or the self-enhancement lateral inhibition (SELI) system represented by the roles of Nodal and Lefty in mesoendoderm determination (38), along with many other similar processes. These act to ensure the robustness and spatiotemporal accuracy of developmental patterning. It has been proposed that, small LR differences at the posterior notochord are converted into a robust LR asymmetry primarily via the coordination of LPM Lefty and Nodal, which function together as a SELI system (39). Our work suggests that, at least in zebrafish, chick and frogs, Fst, functioning along the AP axis, is an integral requirement for the regulation on LR asymmetric Nodal signaling. In the right LPM, where Nodal is low due to the function of Dand5 (20), Fst completely blocks the propagation of Nodal within the right LPM, thereby ensuring the spatial accuracy of the LR-biased Nodal signaling. In the left LPM, due to the function of LRO and fluid flow, the higher levels of Nodal signaling begin to overcome the inhibition of Fst and propagate from a PA direction, despite such a process still being restricted, balanced and controlled by Fst. These events ensure the temporal accuracy of the Nodal signaling transmission.

Previous work has shown that the directional laterality of the zebrafish epithalamus is imposed by the Nodal signaling in the LPM (7). Decades ago, the works of Concha et al., proposed a model where epithalamic Nodal signaling is normally bilaterally repressed by a Nodal downstream source, and then later Nodal signaling from the LPM unilaterally alleviates this repression (4). In this study, we show that Activin is required for the activation of the Nodal signaling in the zebrafish epithalamus, but that this activation is repressed by Fst, the expression of which requires early Nodal activity. Spaw from the LPM overcomes this repression of Fst at a later stage, resulting in unilateral Nodal signaling in the epithalamus. Our work has therefore discovered this previously unrecognized function of Activin in the LR patterning of the zebrafish forebrain and identified Fsta as a Nodal downstream factor that represses forebrain Nodal signaling. This regulation in the forebrain as constituted by the collective coordination of Fsta, Activin, and Nodal signaling, provides a robust system that can swiftly respond to the later Nodal signals transferred from the LPM.

TGF-β family members and their receptors and antagonists form a complex network that establishes the vertebrate body plan (40). FST has been reported to bind and antagonize multiple TGF-β family members including Activin, BMPs, and Growth Differentiation Factors (GDFs) (9, 11). Here we identified Nodal as a new target of Fst, which fills a gap in our understanding of the buffering function of Fst in the TGF-β networks. As also noted in previous studies, here we found that the N-terminal and C-terminal sections of Fst have different affinities to Activin and Nodal. Further investigations on how such differing affinities and inhibition abilities of Fst for its multiple targets are achieved, together with the biological outcomes resulting from these differences, are still required to provide a more complete understanding of this “buffering” function of Fst.

Despite differences in LR symmetry breaking mechanisms between species, Nodal as an asymmetric messenger appears to be highly conserved across vertebrates as well as some invertebrates (1, 41). Interestingly, our work found that it was not only the expression patterns of fst in zebrafish, chick, and frog embryos that shared high similarity, but its barrier function in LR asymmetry was also highly conserved across these species. These results suggest that the transcriptional regulation of Fst, and its inhibition function to Nodal, may have evolved together with Nodal in achieving LR asymmetry patterning. However, confirmation of whether Fst also regulates LR asymmetry in mammals awaits further investigation.

In conclusion, in the zebrafish, frog and chick, we identified Fst as a new LR regulator downstream of LRO and parallel to the midline barrier. We also discovered an active-repression and repression–alleviation system collectively constituted by interactions between Activin, Fst and Spaw in the LR patterning of the zebrafish forebrain. Whether this system also operates in other vertebrates’ forebrain awaits further investigation (Fig. 8).

Materials and Methods

The full Materials and Methods is in the SI Appendix, a brief summary follows.

The WGCNA (Weighted correlation network analysis) package (https://CRAN.R-project.org/package=WGCNA) was used to analyze the correlation between module eigengenes (genes function as a robust unit) and sample traits (Nodal concentration and Nodal response time). All animal procedures were performed per the requirements of the “Regulation for the Use of Experimental Animals in Zhejiang Province,” with the approval by the Zhejiang University Animal Care and Use Committee. Specific experimental details of morpholino knockdown, gene editing, mRNA overexpression, optogenetics system for fsta expression, bead implantation, and embryonic cell transplantation in zebrafish embryos; FST morpholino knockdown and beads implantation in chick embryos; fst knockdown and overexpression in Xenopus embryos; are described in supporting information. WISH, immunohistochemical, immunofluorescence, immunoblotting, Co-IP and qPCR analyses were performed using standard methods. The MOs and primers used in this study are listed in SI Appendix, Tables S1 and S2 (SI Appendix, Tables S1 and S2).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Expression matrix from bulk RNA-seq datasets of Nodal explants was obtained from our previous study (with GEO accession number GSE165654) (12). All other study data are included in the article and/or SI Appendix.