Protein-O-fucosylation of coreceptors may be required for Nodal signaling in Xenopus

Abstract

Nodal-related ligands of TGF-β family play pivotal roles for mesoderm induction and body axis formation during vertebrate early embryogenesis. Nodal ligands are distinct from most other TGF-β ligands family as they require EGF-CFC factors as coreceptors for signaling, in addition to their cognate type I and type II TGF-β receptors. In amphibian Xenopus laevis embryos, 5 Nodal-related genes (Xnr1/2/4/5/6) and 2 EGF-CFC genes (XCR1, XCR3) play roles in mesoderm induction and the accumulation of phosphorylated Smad2, while in mammalian embryos, 1 Nodal gene and 1 EGF-CFC gene (Cripto) play roles during mesoderm induction. Mammalian EGF-CFC factors are reported to be O-fucosylated at a conserved threonine residue of the EGF-like motif by protein-O-fucosyltransferase 1 (Pofut1), but this O-fucose modification is shown to be dispensable for Nodal signaling in mammalian embryos. In this study, we investigated the developmental roles of Xenopus laevis Pofut1 (XPofut1) and its potential function in Nodal signaling. We found that morpholino antisense-mediated knockdown of XPofut1 causes reduction of Smad2 phosphorylation in late blastula and axial truncation in neurula. We also found that the O-fucosyltransferase activity of XPofut1 is important in the marginal zone, but not in the vegetal pole region, of blastula. Interestingly, XPofut1 is necessary for Smad2 phosphorylation induced by Xnr1 or Xnr2, but not by Xnr5 or Xnr6. Among the Nodal signaling components, only EGF-CFC factors are known to be modified by Pofut1. Therefore, based on our current observation, we propose that XPofut1 regulates signaling of a subset of nodal ligands in pregastrulation embryos possibly through modulating the function of EGF-CFC factors.

INTRODUCTION

Nodal ligands are members of the TGF-β family and they play pivotal roles during vertebrate embryogenesis (Hill, 2018, Hill, 2022, Jones and Mullins, 2022). During early embryogenesis, Nodal and nodal-related genes are essential for the specification of dorsal-ventral body axis and the induction of mesoderm. In amphibian Xenopus laevis, 6 Xenopus nodal-related genes (Xnr1∼6) have been identified and, besides Xnr3, they all share mesoderm-inducing activity (Carron and Shi, 2016, Luxardi et al., 2010, Skirkanich et al., 2011, Tadjuidje et al., 2016, Takahashi et al., 2000). Xnr5 and Xnr6 are expressed before the mid blastula transition (MBT) and they are required, after the MBT, for the expression of Xnr1, Xnr2, and Xnr4 and for the accumulation of phosphorylated Smad2, a hallmark of Nodal signaling.

Nodal ligands bind to their cognate type I (Acvr1b and Acvr1c) and type II (Acvr2a and Acvr2b) TGF-β receptors leading to the subsequent phosphorylation of intracellular Smad2 and Smad3 (Hill, 2018, Hill, 2022). Phosphorylated Smad2/3 interact with Smad4 in the cytoplasm, enter the nucleus, form transcriptional complexes with DNA-binding proteins, and induce the expression of target genes.

Nodal ligands, unlike other TGF-β molecules, also require EGF-CFC factors as coreceptors (Calvanese et al., 2015, Dorey and Hill, 2006, Minchiotti et al., 2001, Onuma et al., 2006, Yabe et al., 2003). Mammals have 2 EGF-CFC genes (Cripto, Cryptic) and Xenopus laevis has 3 EGF-CFC genes (XCR1, XCR2, and XCR3) with distinct spatiotemporal expression patterns. EGF-CFC factors contain 2 conserved domains, an EGF-like domain, and a CFC domain, and they are tethered to the plasma membrane by a GPI linkage. There are 2 different hypotheses of how EGF-CFC factors function in Nodal signaling. At first, EGF-CFC factors, acting as coreceptors, were shown to promote the binding of Nodal ligands to the receptor (Calvanese et al., 2015, Minchiotti et al., 2001, Watanabe et al., 2007, Yeo et al., 1999). In an alternative model, EGF-CFC factors were shown to be necessary for the processing and internalization of Nodal ligands into the endosome, where Nodal signaling is thought to predominantly occur (Blanchet et al., 2008a, Blanchet et al., 2008b).

Human and mouse Cripto proteins have been shown to undergo O-linked fucose modification. O-fucosylation is mediated by protein-O-fucosyltransferase 1 and 2 (Pofut1/2), and occurs at conserved Ser/Thr residues in EGF-like repeats (C₂-XXXXS/T-C₃) of a handful of proteins, including plasminogen activators, blood clotting factors, and Notch proteins. In Drosophila, Pofut1 alters the ligand-binding specificity of Notch (Pandey et al., 2019). O-fucosylation of Cripto occurs at Thr88 residue in the ligand-binding EGF-like domain, and substitution of Thr88 to Ala renders Cripto nonfunctional for Nodal signaling (Schiffer et al., 2001). However, a detailed study suggests that Pofut1 is dispensable for Nodal signaling in mammals and that the Thr88 residue itself but not the O-fucose is important for Nodal signaling (Shi et al., 2007).

In this study, we investigated the developmental roles of Xenopus laevis Pofut1 (XPofut1) and its potential roles in Xnr signaling. XPofut1 knockdown with an antisense morpholino oligonucleotide (MO) caused aberrant patterns of Smad2 phosphorylation and defects in anterior-posterior patterning. Of interest, we found that XPofut1 differentially regulates Xnr1/2 vs Xnr5/6 signaling and the accumulation of phosphorylated Smad2 in different regions of the embryo. These results suggest that XPofut1 is important for signaling of at least a subset of Xnr ligands during early embryogenesis.

MATERIALS AND METHODS

Plasmids

The open-reading frame of XPofut1 was subcloned into pCS2+ vector. For Sec-XPofut1 (Xenopus laevis Pofut1) and Sec-hPOFUT1 (human POFUT1), the signal peptide sequence of XPofut1 or hPOFUT1 was replaced with the human IgΚ signal peptide sequence. For glycosyltransferase-defective hPOFUT1, Arg127 residue was substituted to Ala.

Morpholino Oligonucleotides

Standard MO, random MO (negative controls), and gene-specific MOs used in this study were purchased from Gene Tools, LLC. The following gene-specific MOs were used:

Name	Sequence (sequence complementary to the start codon is bold)
XPofut1 MO	5′-AAG CCA AAC ACC GCG CTC CAT TCC C-3′
XCR1 MO1	5′-AAA CTG CAT TGT TTT CTG CAA AGG C-3′
XCR1 MO2	5′-ATT TAA TGT GTC CTC AGC AAA AGC C-3′
XCR3 MO1	5′-CAT GGC ACA GTC CTG CTC CAA CTA A-3′
XCR3 MO2	5′-CCA TAC CAT GGC ACA GTC CTG CTC C-3′

Xenopus laevis Embryo Manipulation and Microinjection

Fertilized embryos were dejellied in 3% cysteine/0.1× MMR, microinjected, and cultured at 0.33× MMR. Animal caps were dissected at stage 8.5 and cultured in 0.7× MMR. Staging of embryos was according to Nieuwkoop and Faber (Zahn et al., 2022). Synthetic mRNAs were transcribed with SP6 mMessage mMachine kit (Ambion). For microinjection into 2-cell- or 4-cell-stage embryos, MO and mRNA were injected into each blastomere.

Whole-Mount In Situ Hybridization

In situ hybridization was performed as previously described (Harland, 1991). Digoxigenin (DIG)-labeled RNA probe, containing the antisense sequence of the entire coding region of XPofut1, was synthesized using DIG-UTP (Roche Life Science) and MEGAscript SP6 kit (Ambion). Embryos of selected stages were fixed in MEMFA, dehydrated in methanol, and rehydrated in PTw (1× PBS, 0.1% Tween-20). Embryos were treated with 10 μg/mL of proteinase K (Roche Life Science) for 5 minutes followed by prehybridization. Hybridization was carried out at 62°C overnight. The DIG-labeled probes were visualized using anti–DIG-AP-Fab fragment and BM purple substrate (Roche Life Science).

Quantitative RT-PCR

Embryos and explants of indicated stages were treated with RNAlater (Ambion) overnight at 4°C. Total RNA was isolated using RNeasy Mini Kit (Qiagen) and reverse-transcribed using Superscript III First-strand Synthesis System (Invitrogen). Real-time PCR was performed in Rotor-gene 6 instrument (Corbett Research) using QuantiTect SYBR Green PCR Kit (Qiagen).

A representative result of 3 separate experiments is shown. The following primers were used:

Gene	Direction	Sequence
Eomesodermin (Eomes)	Forward	5′-GCA GAG GTT CTA CCT ATC-3′
	Reverse	5′-CAT TGC TTG AGG TGC TAG G-3′
Goosecoid (Gsc)	Forward	5′-ACA ACT GGA AGC ACT GGA-3′
	Reverse	5′-TCT TAT TCC AGA GGA ACC-3′
EF1α	Forward	5′-CCT GAA CCA CCC AGG CCA GAT TGG TG-3′
	Reverse	5′-GAG GGT AGT CAG AGA AGC TCT CCA CG-3′
VegT	Forward	5′-CAA GTA AAT GTG AGA AAC CGT G-3′
	Reverse	5′-CAA ATACAC ACA CAT TTC CCG-3′
Xbra	Forward	5′-GGA TCG TTA TCA CCT CTG-3′
	Reverse	5′-GTG TAG TCT GTA GCA-3′
Xnr1	Forward	5′-AAC CAT CAC TTA TCA ATA GG-3′
	Reverse	5′-TGT AGG CCA GTA AAA TCA TTA AC-3′
XPofut1	Forward	5′-GAT GCC AGG CGG GTG TCT TGT TT-3′
	Reverse	5′-AGG CCT GAT TTC ATG GAG TCT TTA-3′
XVent1	Forward	5′-TTC CCT TCA GCA TGG TTC AAC-3′
	Reverse	5′-GCA TCT CCT TGG CAT ATT TGG-3′

Luciferase Assay

Firefly luciferase reporter A3-Luc (Yeo et al., 1999), containing 3 tandem repeats of the activin-response element, and control reporter pRL-CMV (50 pg each/embryo) were injected to 2-cell-stage embryos. Luciferase assays were performed using Dual Luciferase Reporter Assay System (Promega). Firefly luciferase activities were normalized to the corresponding Renilla luciferase activities. A representative result of 3 separate experiments is shown.

Immunoblotting

Embryos or animal caps at indicated stages were lyzed in a volume of 20 μL/embryo or 2 μL/animal cap with an ice-cold lysis buffer [50 mM Tris HCl (pH 8.0), 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 2 mM EDTA, 10% glycerol, 0.1% SDS, 2 mM β-glycerophosphate, 2 mM imidazole, 10 mM sodium fluoride, 1.15 mM sodium molybdate, 1 mM sodium orthovanadate, 2 mM sodium pyrophosphate, 4 mM sodium tartrate dehydrate, 10 nM calyculin A, 30 nM okadaic acid, 1× Cømplete protease inhibitor cocktail-EDTA (Roche), 1 mM NEM, and 1 mM PMSF]. After centrifugation, supernatants were subjected to SDS-PAGE, and proteins were transferred to nitrocellulose membrane and visualized using appropriate antibodies and ECL reagent. The following antibodies were used: antiphospho-Smad2 (Ser465/467) (Cat. # 3101, Cell Signaling Technology), anti-Smad2/3 (Cat. # 610842, BD Bioscience), and antiactin (Cat. # A4700, Sigma-Aldrich). Total Smad2 and actin were used as loading controls. Each experiment is repeated at least 3 times and the representative results are shown.

RESULTS

Expression Patterns of XPofut1

Xenopus laevis pofut1 L transcript (GenBank accession # NM_001088891) encodes a 380-amino acid long polypeptide (GenBank accession # NP_001082360), and its amino acid sequence is 71.82% (85%) and 73% (85%) identical to human and mouse Pofut1, respectively (numbers in parentheses are similarities) (Supplementary Fig. 1). We named this protein Xenopus laevis protein-O-fucosyltransferase 1 (XPofut1). XPofut1, like its mammalian homologs, contains an N-terminal signal peptide, a glycosyltransferase domain, and a C-terminal KDEL-like ER retention sequence.

We first analyzed the temporal and spatial expression patterns of XPofut1. XPofut1 transcripts are present in unfertilized eggs, and the levels of transcript increase gradually from the late blastula stage (stage 9), peak at the neurula stage (stage 20), and gradually decrease until the tadpole stage (stage 35) (Fig. 1A). In late blastula (stage 9), XPofut1 transcripts are present throughout the animal, marginal, and vegetal regions with slightly higher levels in the vegetal region (Fig. 1B). In early gastrula (stage 10+), XPofut1 transcripts are distributed evenly across the dorsal-ventral axis (Fig. 1C). According to the results of in situ hybridization analyses, XPofut1 expression becomes restricted to specific regions of embryos as development progresses (Fig. 2). XPofut1 transcripts are localized around the blastopore in late gastrula (stage 12) and to the anterior neural plate in neurula (stages 18 and 20). At the tailbud (stages 24-30) and tadpole (stage 37) stages, XPofut1 expression is significantly higher in the anterior-dorsal region than the posterior-ventral region and becomes restricted to the brain, eye, branchial arches, pronephros, and somites.

Fig. 1.

Expression patterns of XPofut1. Total RNAs from embryos or regions of embryos at indicated stages were isolated, the transcript levels of XPofut1 were analyzed by RT-PCR and normalized with those of EF1α. (A) Comparison of the levels of XPofut1 transcripts in unfertilized egg, blastula (stages 7 and 9), gastrula (stages 10+ and 12), neurula (stages 15 and 20), tailbud (stages 25 and 30), and tadpole (stage 35). (B) Stage 9 embryos were divided into the animal (A), marginal (M), and vegetal (V) regions. The levels of VegT (a ventral marker) transcripts are compared to confirm the accuracy of dissection. (C) Stage 10 embryos were divided into the dorsal (D) and vegetal (V) halves. The levels of Gsc (a dorsal marker) and XVent1 (a ventral marker) transcripts are compared to confirm the accuracy of dissection.

Fig. 2.

Spatial expression patterns of XPofut1. Whole-mount in situ hybridization was performed with embryos of indicated stages. Photographs of representative embryos are shown. (A) Stage 12, vegetal view, dorsal side is toward the top. (B-D) Stage 18, anterior view, dorsal side is toward the top (B), lateral view, dorsal side is toward the top (C), and dorsal view, anterior side is toward the top (D). (E and F) Stage 20, anterior view, dorsal side is toward the top (E) and dorsal view, anterior side is toward the top (F). (G and H) Stage 24, lateral view (G), and dorsal view (H). (I-K) Stage 30, lateral view (I), higher magnification of anterior region (J), and dorsal view (K). (L and M) Stage 37, lateral view (L) and higher magnification of dorsal region (M). ap, auditory placode; ba, branchial arch; bp, blastopore; e, eye; mc, mesencephalon, ng, neural groove; np, neural plate; ns, nephrostome; op, optic placode; ov, optic vesicle; pd, pronephric duct; pn, pronephros; pt, pronephric tubule; rc, rhombencephalon; s, somite.

XPofut1 Is Important for Axial Patterning

The developmental role of XPofut1 is assessed using antisense MO-mediated knockdown. XPofut1 MO blocked in vitro translation of XPofut1 mRNA, which contains the target sequence of MO, but not the translation of mutant XPofut1 mRNA containing alternative codons in the MO target sequence (Supplementary Fig. 2). A negative control MO (XPofut1-sense MO), containing the complementary sequence of XPofut1 MO, had no effect on the translation of wild-type XPofut1 mRNA (Supplementary Fig. 2).

When microinjected into both blastomeres of 2-cell-stage embryos, XPofut1 MO did not cause any discernible defects until the onset of gastrulation (data not shown). However, XPofut1 MO caused apparent defects as development progressed and, at the tadpole and tailbud stages, XPofut1 morphants showed anterior and posterior truncation of the body axis in a dose-dependent manner (Fig. 3A). A control random MO did not cause any discernible defects in tadpoles and tailbud embryos. Defects in XPofut1 morphants could be partially rescued by coinjection of recombinant XPofut1 (Sec-XPofut1) or human POFUT1 (Sec-hPOFUT1) mRNA encoding XPofut1/hPOFUT1 proteins with an XPofut1 MO nontargeted heterologous signal peptide (Fig. 3B). These results suggest that XPofut1 is involved in the axial patterning.

Fig. 3.

XPofut1 knockdown causes axial truncation. (A) Indicated amounts of XPofut1 (2.5-10 nmole/embryo) or control random MO (10 nmole/embryo) were injected into the marginal regions at 2-cell stage and the embryos were cultured until stage 27. Photographs of representative embryos are shown. (B) Indicated combinations of XPofut1 or control random MO (cMO) (5 nmole/embryo), Sec-hPOFUT1 or Sec-XPofut1 mRNA (ng/embryo) were coinjected into the marginal regions at 2-cell stage. The embryos were cultured until stage 27 and scored for phenotypes. Photographs of representative embryos for each category of phenotype are shown, and percentages of embryos to each category are tabulated.

The axial truncation in XPofut1 morphants was resembling those of XCR1 and XCR3 morphants although the defects in XPofut1 morphants were less severe (Supplementary Fig. 3). Knockdown of both XCR1 and XCR3 caused defects at lower doses than knockdown of both XCR1 or XCR3 alone as previously reported (Dorey and Hill, 2006).

XPofut1 Is Important for Xnr1 Signal Transduction

To determine whether XPofut1 is involved in Nodal signaling, we first examine the effects of XPofut1 knockdown on the expression of a Nodal-responsive luciferase reporter A3-Luc. A3-Luc contains 3 tandem repeats of the activin-response element, and activin-response element contains binding sites for the FoxH1/pSmad2/Smad4 complex, a transcriptional complex formed in Xenopus embryos in response to nodal/activin stimuli (Yeo et al., 1999). XPofut1 MO reduced the response of A3-Luc to Xnr1 (Fig. 4A). XPofut1 MO also inhibited the expression of Xnr1 target genes Goosecoid (Gsc), Brachyury (Xbra), and Eomesodermin (Eomes) in ectodermal explant (animal cap) assay (Fig. 4B). These results suggest that XPofut1 is important for Xnr1 signaling.

Fig. 4.

XPofut1 is involved in Xnr1 signal transduction. (A) A3-Luc reporter plasmid (50 pg/embryo) was injected into animal regions at 2-cell stage with indicated combinations of control random or XPofut1 MO (5 nmole/embryo) and Xnr1 mRNA (125 pg/embryo). Animal caps were explanted at stage 8.5, cultured until stage 9.5, and harvested to analyze luciferase activities. (B) Indicated combinations of control random (cMO) or XPofut1 MO (5 nmole/embryo) and Xnr1 mRNA (125 pg/embryo) were injected into animal regions of 2-cell-stage embryos. Animal caps were explanted at stage 8.5, cultured until stage 9.5, and harvested for RT-PCR analysis. EF1α is used as a loading control.

XPofut1 Is Important for the Accumulation of Phosphorylated Smad2 During Embryogenesis

Activation of Nodal signaling in embryos leads to the accumulation of Smad2 that is phosphorylated (pSmad2) at the C-terminal SSXS motif (Hill, 2018). To determine whether XPofut1 is involved in Nodal signaling during development, we examined the effects of XPofut1 MO on pSmad2 accumulation in embryos. XPofut1 MO, but not a control random MO, reduced pSmad2 accumulation in stage 9.5 embryos in a dose-dependent manner (Fig. 5A). However, in stage 10 XPofut1 morphants, levels of pSmad2 had recovered to levels comparable to those of sibling wild-type embryos (Fig. 5A). In pregastrula-stage embryos, Xnr signaling, measured by Smad2 phosphorylation, is activated first in the dorsal side, and gradually propagates to the marginal sides and to the ventral sides as development progresses (Carron and Shi, 2016, Faure et al., 2000, Lee et al., 2001). As gastrulation progresses in stage 10 embryos, negative feedback regulation of Xnr signaling, through the expression of Xnr antagonists, reduces Smad2 phosphorylation from the dorsal side first and this reduction in Smad2 phosphorylation also propagates to the marginal and ventral sides as development progresses. Therefore, the recovery of pSmad2 accumulation in stage 10 XPofut1 morphants could result from incomplete XPofut1 knockdown or the absence of negative feedback regulation of Xnr signaling. Nevertheless, the delayed accumulation of pSmad2 was enough to cause axial truncation in XPofut1 morphants in tailbud- and tadpole-stage embryos (Fig. 3A).

Fig. 5.

XPofut1 is important for Smad2 phosphorylation in early embryos. (A) Two-cell-stage embryos were injected into marginal regions with random MO (Co, 10 nmole/embryo) or indicated amounts of XPofut1 MO (nmole/embryo), and cultured until stage 9.5 or 10. The levels of phosphorylated Smad2 (pSmad2) are compared by immunoblotting. Actin is used as loading control. (B-D) Microinjected embryos were cultured until stage 9.5 and harvested to analyze the levels of phosphorylated Smad2 (pSmad2) by immunoblotting. Levels of total Smad2 (Smad2) are also compared and actin is used as a loading control. (B) Random control (Co) or XPofut1 MO (5 nmole/embryo) was injected into marginal regions at 2-cell stage, and then indicated amounts (pg/embryo) of hPOFUT1 mRNA were injected into either 4 animal blastomeres (Animal) or 4 vegetal blastomeres (Vegetal) at 8-cell stage. (C) Random control or XPofut1 MO (5 nmole/embryo) was injected into the marginal region at 2-cell stage, and then mRNA encoding wild-type (WT) or glycosyltransferase-defective mutant form (RA) of hPOFUT1 (100 pg/embryo) was injected into 4 animal blastomeres at 8-cell stage. (D) hPOFUT1 mRNA (100 pg/embryo) was injected into 4 animal blastomeres at 8-cell stage (A), into marginal regions at 2-cell stage (M), or into 4 vegetal blastomeres at 8-cell stage (V).

We examined if the reduction of pSmad2 accumulation in XPofut1 morphants can be relieved with coinjection of MO nontargeted recombinant Xenopus Pofut1 (Sec-XPofut1) or human POFUT1 (hPOFUT1) mRNA. When microinjected to the marginal region of 2-cell-stage XPofut1 morphants, XPofut1 or hPOFUT1 did not reverse the effects of XPofut1 MO on Smad2 phosphorylation (data not shown). mRNAs microinjected to the marginal region of the 2-cell-stage embryos can diffuse throughout the embryo. We examined the effects of restricting the distribution of Pofut1 to the animal or vegetal half of an embryo by microinjection of hPOFUT1 mRNA only in the animal blastomere or vegetal blastomere of 8-cell-stage embryos. mRNA microinjected to animal blastomere of 8-cell-stage embryos would be restricted to animal cap region and marginal zone in stages 8 to 10 embryos (blastula to early gastrula), whereas mRNA microinjected to vegetal blastomere would be distributed to marginal zone and vegetal pole regions in stages 8 to 10 embryos. Microinjection of hPOFUT1 mRNA to the animal half of XPofut1 morphants increased Smad2 phosphorylation in a dose-dependent manner, whereas microinjection of hPOFUT1 mRNA to the vegetal half of XPofut1 morphants did not cause discernible change on Smad2 phosphorylation (Fig. 5B). Sec-XPofut1 mRNA had similar effects (data not shown). Since Smad2 phosphorylation occurs only in the marginal zone (presumptive mesoderm) and the vegetal pole region (presumptive endoderm) but not in the animal cap region (presumptive ectoderm, Xnr ligands not present) in stages 8 to 10 embryos, recovery of Smad2 phosphorylation by expression of hPOFUT1 in the animal half of XPofut1 morphants is likely due to the recovery of Smad2 phosphorylation in the marginal zone. These results suggest that, for the accumulation of Smad2 phosphorylation in blastula/early gastrula, XPofut1 activity is required only in the marginal zone but not in the vegetal pole region even when XPofut1 transcripts are distributed throughout the embryo.

Next, we investigated whether the glycosyltransferase activity of Pofut1 is necessary for the reversion of pSmad2 reduction caused by XPofut1 knockdown. When overexpressed only in the animal half of an embryo, hPOFUT1 RA mutant, with an Arg-to-Ala substitution of Arg127 residue important for the glycosyltransferase activity, was unable to increase Smad2 phosphorylation in XPofut1 morphants (Fig. 5C). We also examined what effects does Pofut1 overexpression in different regions of the embryo have on Smad2 phosphorylation. When microinjected to the animal half, hPOFUT1 mRNA did not cause any discernible change in Smad2 phosphorylation at stage 9.5 (Fig. 5D). However, microinjection of hPOFUT1 mRNA to the marginal region at 2-cell stage (mRNA is distributed throughout the embryo) or to vegetal half slightly, but reproducibly, decreased Smad2 phosphorylation at stage 9.5 (Fig. 5D). These results suggest that the glycosyltransferase activity of XPofut1 is important for the accumulation of pSmad2 in the marginal zone, while XPofut1 is unnecessary or inhibitory to the accumulation of pSmad2 in the vegetal pole region at least prior to the onset of gastrulation.

We also examined if activation of other signaling pathways, that are important for axial patterning in Xenopus embryos, can reverse the effect of XPofut1 MO on Xnr1-induced Smad2 phosphorylation. FGF and Wnt signaling pathways coordinate with Nodal signaling pathway in the control of embryonic body axis formation, germ layer specification, and gastrulation movement (Jones and Mullins, 2022). We also examined the effects of Notch signaling pathway activation in XPofut1 morphants as Pofut1 has been shown to modulate Notch signaling in mammalian embryos and as Notch signaling pathway plays important roles in early embryogenesis (Pandey et al., 2019). As expected, src-hALK4(TD), a constitutively active recombinant form of Acvr1b (type I receptor for Nodal ligands), significantly increased Smad2 phosphorylation in XPofut1 morphants (Supplementary Fig. 4). However, neither the activation of FGF signaling pathway with overexpression of constitutively active Ras (H-RasV12), Wnt signaling pathway with overexpression of β-catenin, nor Notch signaling pathway with overexpression of Notch intracellular domain caused any discernible change in the reduction of Smad2 phosphorylation caused by XPofut1 knockdown (Supplementary Fig. 4).

XPofut1 Is Necessary for Xnr1/2 Signaling but Not for Xnr5/6 Signaling

Xnr5/6 transcripts are accumulated and restricted to the vegetal region of pre-MBT embryos, and Xnr5/6 induce the expression of Xnr1/2 at the onset of MBT. Xnr1/2 transcripts become restricted to the marginal zone as development progresses (Luxardi et al., 2010, Takahashi et al., 2000). Since overexpressing Pofut1 in animal vs vegetal half of an embryo resulted in different outcomes, we examined the effects of XPofut1 knockdown on the signal transduction of Xnr1, Xnr2, Xnr5, and Xnr6. In animal cap assays, XPofut1 MO blocked Xnr1-induced Smad2 phosphorylation and coinjection of hPOFUT1 mRNA reversed the effect of XPofut1 MO on Smad2 phosphorylation (Fig. 6A). However, neither XPofut1 MO nor coinjection of hPOFUT1 mRNA caused any discernible change in Xnr5 or Xnr6-induced Smad2 phosphorylation (Fig. 6A). Next, we examined the effects of XPofut1 MO on Xnr-induced Smad2 phosphorylation in the presence of XCR1. XPofut1 MO reduced Xnr1 or Xnr2-induced Smad2 phosphorylation when XCR1 is coexpressed, whereas XPofut1 MO had no effect on Xnr5 or Xnr6-induced Smad2 phosphorylation (Fig. 6B). These results suggest that XPofut1 is important for Xnr1 and Xnr2 signal transduction, whereas it is not required for Xnr5 or Xnr6 signal transduction.

Fig. 6.

XPofut1 regulates Xnr1 and Xnr5/6 signaling differently. Embryos were injected with indicated combinations of MO and mRNA. Animal caps were explanted at stage 8.5, cultured until stage 9.5, and harvested to analyze the levels of phosphorylated Smad2 (pSmad2) by immunoblotting. Levels of total Smad2 (Smad2) are also compared and actin is used as a loading control. (A) Embryos were injected into marginal regions with indicated combinations of XPofut1 MO (5 nmole/embryo) and mRNAs encoding Xnr1 (125 pg/embryo), Xnr5 (40 pg/embryo), or Xnr6 (300 pg/embryo) at 2-cell stage, and then injected into 4 animal blastomeres with hPOFUT1 mRNA (100 pg/embryo) at 8-cell stage. (B) Embryos were injected into marginal regions with indicated combinations of XPofut1 MO (5 nmole/embryo) and mRNAs encoding Xnr1 (125 pg/embryo), Xnr2 (20 pg/embryo), Xnr4 (100 pg/embryo), Xnr5 (40 pg/embryo), Xnr6 (300 pg/embryo), or XCR1 (250 pg/embryo) at 2-cell stage.

We found that the glycosyltransferase activity of XPofut1 is important for the accumulation of pSmad2 in embryo (Fig. 5C). Among the components of Nodal signal transduction pathway, only EGF-CFC factors are shown to be O-fucosylated so far (Schiffer et al., 2001, Shi et al., 2007). Of 3 EGF-CFC factors in Xenopus laevis, XCR1 and XCR3 are shown to be expressed and functional in pregastrulation embryos (Dorey and Hill, 2006). We examined if XPofut1 interacts with XCR proteins. XPofut1 interacted with all 3 XCR proteins, with significantly weaker interaction for XCR2 than XCR1 or XCR3 (Supplementary Fig. 5).

DISCUSSION

We propose that XPofut1 is important for signaling by a subset Xnr ligands during early embryogenesis of Xenopus laevis, based on the fact that XPofut1 knockdown causes axial truncation, the reduction of Smad2 phosphorylation, and the decrease in the expression of Nodal target genes and a Nodal-responsive reporter. Interestingly, we found that XPofut1 is required for the accumulation of phosphorylated Smad2 (pSmad2) in the marginal zone in pregastrulation embryos, whereas XPofut1 is unnecessary or inhibitory for the accumulation of pSmad2 in the vegetal pole region.

Roles of XPofut1 During Embryogenesis

Our results indicate that XPofut1 is important for the accumulation of pSmad2 in pregastrulation embryos and axial patterning. During early development in Xenopus laevis, Xnr signaling is necessary for the accumulation of pSmad2 (Luxardi et al., 2010, Skirkanich et al., 2011, Tadjuidje et al., 2016, Takahashi et al., 2000). These results suggest that XPofut1 plays important roles in Xnr signal transduction.

In Xenopus laevis embryos, Smad2 phosphorylation is initiated by Xnr5/6 and Vg1 and then increased by Xnr1/2 that are target genes of Xnr5/6 and Vg1 (Luxardi et al., 2010, Skirkanich et al., 2011, Takahashi et al., 2000). Xnr5/6 transcripts are first expressed in the vegetal pole region of pre-MBT embryos, restricted to the vegetal pole region, and disappear before the onset of gastrulation. Once the expression of Xnr1/2 is induced at the onset of MBT, Xnr1/2 transcripts are distributed in the marginal zone and the vegetal pole region of blastula and early gastrula. Double knockdown of Xnr5/6 causes defects in mesoderm formation and gastrulation movement. Double knockdown of Xnr1/2 also causes defects in gastrulation movement, but the effects of Xnr1/2 knockdown on mesodermal gene expression are less severe than Xnr5/6 knockdown. These results led to a hypothesis that Xnr5/6 are required for mesoderm induction, and Xnr1/2 are additionally required for proper gastrulation movement (Luxardi et al., 2010). Our results suggest that XPofut1 knockdown does not abolish Smad2 phosphorylation but it does cause reduction and aberrant temporal patterns of Smad2 phosphorylation. XPofut1 morphants do not show pronounced gastrulation defects, but it does display axial truncation in neurula, which is an indication of reduced mesoderm formation and/or reduced gastrulation movement. Our results indicate that XPofut1 is required in the marginal zone, where Xnr1/2 transcripts are localized, and for Xnr1/2-induced Smad2 phosphorylation. However, XPofut1 is not necessary in the vegetal pole region, where Xnr5/6 transcripts are localized, and for Xnr5/6-induced Smad2 phosphorylation. Taken together, we propose that XPofut1 controls axial patterning by regulating Xnr1 and Xnr2 signaling, but XPofut1 is not necessary for Xnr5/6 signaling (Fig. 7A). More detailed analyses of the effects of XPofut1 knockdown in different regions of the embryo on the spatial and temporal patterns of Smad2 phosphorylation are needed to understand the roles of XPofut1 during embryogenesis.

Fig. 7.

A schematic model for XPofut1 functional in Xnr signaling. (A) XPofut1 is necessary for Xnr1/2-induced Smad2 phosphorylation. XPofut1 may control Xnr1/2 signaling by O-fucosylation (dashed arrow) of XCRs, the Xenopus laevis EGF-CFC factors. XPofut1 is not necessary for Xnr5/6-induced Smad2 phosphorylation. (B) Spatial pattern of XPofut1 activity and XCR expression in stages 8 to 10 Xenopus laevis embryos. In the vegetal pole region (Vegetal), XPofut1 transcripts are present but XPofut1 function is not necessary for Smad2 phosphorylation. In the marginal zone (Marginal), XPofut1 function is necessary for Smad2 phosphorylation. In the animal cap region (Animal), XPofut1 is functional as shown in animal cap assays (Fig. 6), yet there are no Xnr ligands or Smad2 phosphorylation in this region. See “Discussion” for details.

Modulation of Xnr Signaling by XPofut1

Pofut1 catalyzes O-fucosylation of conserved Ser/Thr residues (C₂-XXXXS/T-C₃) in EGF-like repeats (Ajima et al., 2017, Pandey et al., 2019). Human and mouse Cripto, which are EGF-CFC factors, have been shown to be O-fucosylated (Schiffer et al., 2001, Shi et al., 2007), and XCR1/2/3 all have potential O-fucosylation target sequence (C₂-XNGGT-C₃) in their EGF-like repeats (Dorey and Hill, 2006). For human and mouse Cripto, substitution of the conserved Thr residue, to which O-fucose is attached, to Ala renders Cripto nonfunctional for Nodal signaling (Schiffer et al., 2001). However, mouse Pofut1 null embryos do not show typical phenotypes of Nodal or Cripto null mutant but display phenotypes consistent with deficiencies in the canonical Notch signaling (Shi et al., 2007). The results also indicate that the target Thr residue itself, but not O-fucose, is important for Nodal signaling in mammals (Shi et al., 2007).

In contrast to mammalian Pofut1, we found evidences that XPofut1 modulates Smad2 phosphorylation induced by a subset of Xnr ligands. Why do mammalian and Xenopus laevis Pofut1 show different function in the regulation of Nodal signaling? During the period of germ layer formation and gastrulation, mammalian embryos express only 1 Nodal ligand and only 1 EGF-CFC factor (Cripto), whereas Xenopus laevis embryos express 5 nodal-related ligands (Xnr1/2/4/5/6) and 2 EGF-CFC factors (XCR1 and XCR3) (Hill, 2022). It is intriguing to postulate that amphibian Pofut1 may have gained a function to control/distinguish signaling by multiple nodal-related ligands or that mammalian Pofut1 may have lost the ability to control signaling by single Nodal ligand. Zebrafish embryos express 2 nodal-related ligands (Cyclops and Squint) and only 1 EGF-CFC factor (Oep) during germ layer formation and gastrulation (Hill, 2022). Examining the function of Pofut1 on Nodal signaling in zebrafish and other vertebrate embryos may shed light on understanding the basis of discrepancy between the function of mammalian and Xenopus Pofut1 in the control of Nodal signaling.

How does XPofut1 modulate signaling for only a subset of Xnr ligands? O-fucosylation has been shown to modulate the ligand-binding ability of at least one receptor protein. O-fucosylation of Notch in Drosophila alters ligand-binding affinity of Notch (Pandey et al., 2019). Can O-fucosylation modulate Nodal binding to its receptors or coreceptors? Each Xnr ligand may have different requirements, such as different types of XCR proteins (XCR1 vs XCR3) or O-fucosylation of XCR proteins, for binding to their receptor/coreceptor complexes.

Our results indicate that XPofut1 can modulate Smad2 phosphorylation with its O-fucosyltransferase activity. Among the components of Nodal signal transduction pathway, only EGF-CFC factors have been shown to be O-fucosylated, and only XCRs have O-fucosylation target sequences among the Xnr signaling components. Among XCRs, XCR1 and XCR3 are expressed during the period of mesoderm induction (Dorey and Hill, 2006) (Fig. 7B). We found that XPofut1 interacts with XCR proteins (Supplementary Fig. 5). Knockdown studies also show XCR1 and XCR3 are required for signaling by a partially overlapping subset of Xnr ligands (Dorey and Hill, 2006). Xnr1, Xnr2, and Xnr6 require both XCR1 and XCR3 to signal efficiently, while Xnr5 specifically requires XCR1 for signaling. Both XCR1 and XCR3 are necessary for Smad2 phosphorylation in the marginal zone, whereas only XCR3 is necessary for Smad2 phosphorylation in the vegetal pole region. We found that XPofut1 is required for Smad2 phosphorylation in the marginal zone, but not in the vegetal pole region (Fig. 5). Interestingly, XPofut1 morphants display axial truncation (Fig. 3 and Supplementary Fig. 3) similar to XCR1 morphants that show delay in blastopore lip formation (Dorey and Hill, 2006). XCR3 morphants suffer more severe gastrulation defects. Taken together, we postulate that XPofut1 may only modulate the function of XCR1 or XPofut1 may modulate the function of XCR1 and XCR3 only in the marginal zone.

Our results suggest that XPofut1 plays important roles in Xnr1 and Xnr2 signaling and in the marginal zone of pregastrulation embryo, where Xnr1/2 and XCR1/3 are required for proper development (Dorey and Hill, 2006, Luxardi et al., 2010) (Fig. 7). We also found that XPofut1 is dispensable for Xnr5/6 signaling and in the vegetal pole region of pregastrulation embryos, where Xnr5/6 and XCR3 is required for proper development. Therefore, we propose that XPofut1 regulates the signaling by a subset of Xnr ligands, Xnr1/2 but not Xnr5/6, by modulating the function of one or more XCRs through O-fucosylation (Fig. 7). Examining the effects of XPofut1 to the binding affinity of each Xnr ligand to its receptor/coreceptor complexes would shed light on the mechanisms of how XPofut1 regulates Xnr signaling and development.

Author Contributions

Yeon-Jin Kim: Writing – original draft, Formal analysis, Data curation. Soo Young Lee: Writing – review & editing, supervision. Chang-Yeol Yeo: Writing – review & editing, Writing – original draft, Supervision. Seung-Joo Nho: Formal analysis, Data curation.

Declaration of Competing Interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplemental material

Supplementary material

.