Combinatorial Wnt signaling landscape during brachiopod anteroposterior patterning

Abstract

Background

Wnt signaling pathways play crucial roles in animal development. They establish embryonic axes, specify cell fates, and regulate tissue morphogenesis from the early embryo to organogenesis. It is becoming increasingly recognized that these distinct developmental outcomes depend upon dynamic interactions between multiple ligands, receptors, antagonists, and other pathway modulators, consolidating the view that a combinatorial “code” controls the output of Wnt signaling. However, due to the lack of comprehensive analyses of Wnt components in several animal groups, it remains unclear if specific combinations always give rise to specific outcomes, and if these combinatorial patterns are conserved throughout evolution.

Results

In this work, we investigate the combinatorial expression of Wnt signaling components during the axial patterning of the brachiopod Terebratalia transversa. We find that T. transversa has a conserved repertoire of ligands, receptors, and antagonists. These genes are expressed throughout embryogenesis but undergo significant upregulation during axial elongation. At this stage, Frizzled domains occupy broad regions across the body while Wnt domains are narrower and distributed in partially overlapping patches; antagonists are mostly restricted to the anterior end. Based on their combinatorial expression, we identify a series of unique transcriptional subregions along the anteroposterior axis that coincide with the different morphological subdivisions of the brachiopod larval body. When comparing these data across the animal phylogeny, we find that the expression of Frizzled genes is relatively conserved, whereas the expression of Wnt genes is more variable.

Conclusions

Our results suggest that the differential activation of Wnt signaling pathways may play a role in regionalizing the anteroposterior axis of brachiopod larvae. More generally, our analyses suggest that changes in the receptor context of Wnt ligands may act as a mechanism for the evolution and diversification of the metazoan body axis.

Background

Wnt genes play multiple roles during embryogenesis [1, 2]. They encode secreted glycoproteins with a conserved series of cysteine residues that often act as a symmetry-breaking signal [3]. Wnt activity can establish embryonic axes [4, 5], mediate cell fate decisions in early embryos [6,7,8], and specify endomesodermal tissues during gastrulation [9,10,11,12,13]. Moreover, they can also control morphogenetic processes, such as apical constriction, convergent extension, and cell migration [14,15,16,17,18,19,20]. This multitude of roles is consistent with the sheer complexity of Wnt signaling pathways, which involves 13 subfamilies of Wnt ligands, five subfamilies of Frizzled receptors, additional co-receptors, and different agonists, antagonists, downstream players, and effector molecules. In general terms, there are at least three interconnected Wnt pathways. The Wnt/beta-catenin (canonical) pathway regulates cell fate specification through the activity of beta-catenin, the Wnt/PCP (planar cell polarity) pathway controls cell polarity during tissue morphogenesis, and the Wnt/calcium pathway regulates intracellular calcium levels for convergent extension movements—although these functions are not exclusive, as the Wnt/PCP and Wnt/calcium pathways can also control cell fate specification through the activity of other transcriptional effectors like ATF2 and NFAT, respectively [21, 22]. Understanding how these intricate signaling networks regulate embryogenesis and influences developmental evolution remains a significant challenge.

The discovery of staggered Wnt expression domains in the tail bud of amphioxus embryos [23] and along the body axis of sea anemone embryos [24] raised the hypothesis that different combinations of Wnt genes can pattern different body regions. This idea, commonly referred to as the Wnt code or landscape [25,26,27], is an analogy to the Hox code, where the combinatorial expression of nested Hox domains determines the positional identities of tissues along the body axis [28]. Over the years, however, it has been increasingly recognized that the output of Wnt signaling does not depend solely on the expression of Wnt ligands. In fact, Wnt pathways operate via an intricate network of dynamic protein interactions where the downstream response depends on the local availability of receptors, the presence of different antagonists, and the activity of pathway modulators [29, 30]. That means that, depending on the receptors present in the tissue, one Wnt ligand can activate or inhibit a different Wnt pathway and thus determine processes as diverse as fate specification, cellular organization, and tissue morphogenesis [31,32,33]. This complexity makes the Wnt code particularly challenging to elucidate.

A necessary step to untangle this combinatorial code is to extend the analyses of ligand–receptor contexts of Wnt genes to other animal groups using a comparative approach [34]. The comparison between flies and other animals was crucial to reveal the broader importance of the Hox genes as a high-level axial patterning system and not merely an arthropod-specific feature linked to body segmentation [35,36,37]. But while the expression of some Wnt genes is conserved in different animals (e.g., wnt8 is expressed in the neuroectoderm of vertebrates [38, 39] and in spiders [27], annelids [40], and hemichordates [41]), it remains unclear if the combinatorial expression of Wnt signaling components along the embryonic body axis is conserved across the animal phylogeny.

The developmental Wnt landscape has been widely studied in several animal groups, but fewer works have analyzed the receptor context information in the Spiralia, a major branch of bilaterian animals with diverse body patterns [42, 43]. While ecdysozoans have lost several Wnt genes [44,45,46], spiralians have retained the ancestral Wnt complement [27, 47, 48], indicating that they can be an informative group to understand the role of Wnt signaling in axial patterning evolution. However, most analyses about Wnt genes in Spiralia were performed in annelids and mollusks, and the expression data in other spiralian lineages is still lacking.

Brachiopods are sessile spiralians with bivalve shells [49]. They have a reduced adult morphology, but complex embryogenesis where a radially symmetric gastrula undergoes a series of morphogenetic changes to form a larval body subdivided into a series of distinct lobes along the anteroposterior axis [50, 51]. In previous studies using the rhynchonelliform brachiopod Terebratalia transversa (Sowerby, 1846)—a species with a well-characterized embryonic development [52,53,54,55,56,57,58,59,60,61,62]—we found that Wnt signaling plays a role in the specification of the endomesoderm and posterior fates [63], and in the patterning of the head–trunk boundary [64]. Moreover, we found that over-activation of the Wnt/beta-catenin pathway disrupts the molecular and morphological organization of the larval subdivisions [63, 64], suggesting that Wnt activity contributes to the axial patterning of the larva. However, a full characterization of the Wnt signaling components and their developmental expression, including the ligand–receptor contexts is lacking for T. transversa, and for brachiopods in general.

In this study, we characterize the Wnt complement of the brachiopod T. transversa and investigate the spatiotemporal expression of Wnt signaling components throughout embryogenesis. We find that during axial elongation, the expression of ligands, receptors, and antagonists show an anteroposterior organization forming regionalized transcriptional territories, each expressing a unique combination of transcripts. These territories precede and coincide with the morphological subdivisions of the larval body, indicating that the differential activation of Wnt signaling may contribute to pattern the brachiopod larval body. We identified differences in receptor-context that may be involved in patterning an evolutionary novelty of lecithotrophic brachiopod larvae, the reversible mantle lobe. A comparative analysis reveals that while the expression of Frizzled receptors is evolutionarily conserved, the expression of Wnt ligands is more variable. This suggests that the evolutionary shuffling in the expression of Wnt ligands may be a mechanism underlying the evolution of anteroposterior diversification in bilaterians.

Results

Terebratalia transversa has a conserved repertoire of Wnt genes

Metazoans have a large Wnt repertoire with 13 subfamilies [24, 48, 65]. To characterize the Wnt complement of the brachiopod Terebratalia transversa, we surveyed a reference transcriptome of this species for Wnt genes using similarity searches with known Wnt genes from other animals. We identified 13 Wnt genes with representatives of 12 of the 13 Wnt subfamilies (Fig. 1). T. transversa is missing wnt3, a gene known to have been lost in Protostomia [27, 48], and has two copies of wnt1. One of the wnt1 paralogs—named hereafter wnt1t—has a conserved Wnt domain, but is highly divergent at the sequence level compared to other wnt1 orthologs across bilaterians (Additional file 1: Fig. S1). Our phylogenetic analysis suggests that this paralog originated via a lineage-specific duplication within T. transversa or rhynchonelliform brachiopods (Additional file 2: Fig. S2). Besides the loss of wnt3 and duplication of wnt1, T. transversa shows a single representative ortholog for the remaining subfamilies, suggesting that the ancestral repertoire of metazoan Wnt genes remained largely conserved.

Fig. 1

Orthology assignment of Terebratalia transversa Wnt genes. Best-scoring tree of a maximum likelihood phylogenetic analysis using the amino acid sequences of known metazoan Wnt genes. The color-coding represents different Wnt subfamilies and the numbers show the support values of individual branches. Terebratalia transversa (Tt) orthologs are outlined by a box. The other species are Branchiostoma floridae (Bf), Capitella teleta (Ct), Drosophila melanogaster (Dm), Homo sapiens (Hs), Lingula anatina (La), Lottia gigantea (Lg), Platynereis dumerilii (Pd), Saccoglossus kowalevskii (Sk), and Tribolium castaneum (Tc)

Wnt genes are upregulated in concert during axial elongation

To uncover the developmental dynamics of Wnt expression in T. transversa, we analyzed stage-specific RNA-Seq data from the unfertilized egg to the post-metamorphic juveniles (Fig. 2 and Table 1). We detect no Wnt transcripts expressed in the oocyte or mid blastula stages (the high levels of wnt4 and wntA in early stages is due to a bias in the expression quantification, see “Methods” for a detailed explanation). Wnt expression shifts significantly at the late blastula stage (19 h), when a concerted upregulation of wnt1, wnt1t, wnt8, wnt10, and wnt16 occurs (Fig. 2). Throughout gastrulation, Wnt genes continue to be upregulated with wnt1 and wnt5 in the early gastrula (26 h); wnt6, wnt7, and wnt11 in the mid gastrula (37 h); and wnt2, wnt9, and wnt10 in the late gastrula (51 h). Between the late gastrula and early larva, all Wnt genes are expressed, but some are downregulated after gastrulation (wnt6 and wnt10) and after metamorphosis (wnt7 and wnt16) (Fig. 2). Therefore, Wnt expression is dynamic throughout development but peaks late in gastrulation, when the body elongates along the anteroposterior axis, and at the onset of the morphological differentiation of the larval lobes in T. transversa.

Fig. 2

Expression of Wnt signaling components during Terebratalia transversa development. The heatmap represents the log-normalized transcript counts for ligands, receptors, and antagonists calculated from stage-specific RNA-Seq data. Each cell shows the average value between two replicates. Asterisks in wnt4 and wntA denote samples where the expression levels were overestimated due to the expression of an antisense gene present in the same transcript (see “Methods” for details). The black outline marks the late gastrula stage (51 h), when all Wnt genes are expressed. The illustrations depict T. transversa developmental stages from the oocyte until the post-metamorphic juvenile. The stages we analyzed using in situ hybridization (early gastrula to late larva) are highlighted in magenta

Table 1 Developmental stages sampled for the stage-specific transcriptome of Terebratalia transversa

Wnt expression domains partially overlap along the anteroposterior axis

To uncover the spatial localization of Wnt ligands during brachiopod development, we performed in situ hybridization in T. transversa embryos from late blastula to competent larva (Figs. 3, 4, 5, Additional file 3: Fig. S3, Additional file 4: Fig. S4, and Additional file 5: Fig. S5).

Fig. 3

Whole-mount colorimetric in situ hybridization of Terebratalia transversa wnt1, wnt1t, wnt2, wnt4, wnt5, wnt6, and wnt7. The panels show representative expression patterns for the developmental stages between late blastula and late larva. The samples are oriented with a blastoporal/ventral view and anterior end to the top. Black arrowheads indicate the apical–mantle boundary. White arrowheads indicate the mantle–pedicle boundary. Panels for wnt1 originally published under a Creative Commons Attribution License in [64] and reprinted here for completion. Scale bars = 20 µm

Fig. 4

Whole-mount colorimetric in situ hybridization of Terebratalia transversa wnt8, wnt9, wnt10, wnt11, wnt16, and wntA. The panels show representative expression patterns for the developmental stages between late blastula and late larva. The samples are oriented with a blastoporal/ventral view and anterior end to the top. Black arrowheads indicate the apical–mantle boundary. White arrowheads demarcate the mantle–pedicle boundary. Scale bars = 20 µm

Fig. 5

Whole-mount double-fluorescent in situ hybridization of Terebratalia transversa Wnt genes. A–D Expression of wnt1 (magenta) and wnt5 (green) in the early gastrula (A), late gastrula (B) and early larva in ventral (C) and dorsal (D) views. E–F Expression of wnt5 (green) and fz5/8 (magenta) in the mid gastrula (E) and late gastrula (F) in ventral views. G,H Expression of wnt7 (green) engrailed (en) (magenta) in the early larva in ventral (G) and dorsal (H) views. I–L Expression of wnt1 (magenta) and wnt8 (green) in the mid gastrula (I) and late gastrula (J) in ventral views, late gastrula in an midbody optical section (K), and early larva in dorsal view (L). Green and magenta lines highlight the extension and overlap between domains. Areas in the tissue where the expression overlaps appear in white. Samples oriented with anterior end to the top. Black arrowheads indicate the apical–mantle boundary. White arrowheads demarcate the mantle–pedicle boundary. The panels show representative expression patterns for each sample. Scale bars = 20 µm

wnt1 is expressed in the ectoderm and invaginating endomesoderm of the posterior blastopore lip in the late blastula (Fig. 3, Additional file 3: Fig. S3, and [64]). This domain expands laterally, forming a ventral ectodermal band at the anterior most portion of the pedicle lobe in the early larva, a region that gives rise to the ventral shell primordium in the late larva. From the late gastrula stage, wnt1 is also expressed in a narrow ectodermal stripe around the posterior region of the apical lobe and in the dorsal shell primordium. The apical lobe expression fades, and a new wnt1 domain appears encircling the posterior subdivision of the pedicle lobe in the late larva.

wnt1t expression domains differ significantly from its paralog wnt1. We first detect wnt1t transcripts in a single ectodermal spot at the animal pole of the early gastrula (Fig. 3 and Additional file 3: Fig. S3). This domain localizes to the anterior end of the embryo and is expressed until the early larva, when only subsets of cells continue expressing wnt1t. At this stage, a central patch of ventral ectoderm posterior to the mouth also begins expressing wnt1t. Finally, in the late larva, wnt1t is upregulated in an ectodermal ring beneath the mantle lobe, and at the terminal tip of the pedicle lobe ectoderm.

wnt2 is only expressed in the late gastrula and early larva stages in bilateral ectodermal bands that encircle the posterior portion of the apical lobe almost entirely, except for the ventral and dorsal midlines (Fig. 3 and Additional file 3: Fig. S3).

wnt4 is expressed at the posterodorsal ectoderm from late blastula to late gastrula (Fig. 3 and Additional file 3: Fig. S3). The pattern is similar to wnt1, but wnt4 transcripts are localized more dorsally (Additional file 5: Fig. S5A–C). In the early larva, the expression at the posterior end and dorsal portion fades, the domain becomes narrower, and acquires a subterminal position within the ventral ectoderm of the pedicle lobe. This domain is still present in the late larva, when wnt4 begins to be expressed in the posterior endoderm.

wnt5 is expressed in three distinct ectodermal domains—in the apical, mantle, and pedicle lobes, respectively. We first detect expression in the early gastrula with transcripts at the posterior blastopore lip and anterolateral ectoderm (Fig. 3 and Additional file 3: Fig. S3). The posterior ectodermal domain is narrower than the wnt1 domain (Figs. 3 and 5A) and maintains a terminal position until the late larva stage, when the tip of the pedicle lobe is cleared from expression (Fig. 3 and Additional file 3: Fig. S3). The anterolateral domains expand in the mid gastrula to encircle the posterior portion of the apical lobe ectoderm, and fade in the late larva. wnt5 is also expressed in the leading edge of the growing mantle lobe ectoderm from mid gastrula to late larva. The ectodermal expression domains of wnt5 and wnt1 occupy distinct regions along the anteroposterior axis that coincide with the subdivisions of the larval lobes (Fig. 5B,C and Additional file 5: Fig. S5G).

wnt6 transcripts localize to the posterior blastopore lip, similarly to wnt1 and wnt4, from the early to the late gastrula (Fig. 3 and Additional file 3: Fig. S3). This ectodermal domain is cleared in the early larva, when wnt6 is activated in a midbody section of the endoderm. In the late larva, we also detect wnt6 domains in the ectoderm of the apical and pedicle lobes.

wnt7 initiates as a lateral pair of anterior ectodermal stripes that progressively extend around the entire embryo circumference until the early larva (Fig. 3 and Additional file 3: Fig. S3). This wnt7 stripe demarcates the apical–mantle boundary, partially overlapping with wnt1- and engrailed-expressing cells at the anteriormost region of the mantle lobe (Fig. 5G,H and Additional file 5: Fig. S5H; see also [64]). In the early larva, the anterior wnt7 stripe disappears, and a posterior ectodermal stripe appears demarcating the boundary between the subterminal and terminal regions of the pedicle lobe, between the posterior territories of wnt1 and wnt5.

wnt8 is expressed in a ring of cells in the invaginating endomesoderm and in a broad ectodermal band encircling the late blastula (Fig. 4 and Additional file 4: Fig. S4). In the early and mid gastrula, wnt8 transcripts are cleared from the endomesoderm and from the anterior ectoderm, and two distinct ectodermal domains remain: a pair of broad lateral territories in the apical lobe, and a wide posterodorsal domain in the pedicle lobe. The lateral territories expand ventrally and dorsally, encircling the apical lobe ectoderm, while the posterior ectodermal domain fades in the late gastrula. The anterior wnt8 territories partially overlap with the anterior expression of wnt1 in the apical lobe ectoderm (Fig. 5I–L and Additional file 5: Fig. S5G).

wnt9 transcripts are first expressed in the invaginated endomesoderm of late gastrula embryos at a subterminal position (Fig. 4 and Additional file 4: Fig. S4). The domain forms a contiguous patch of mesodermal and endodermal cells expressing wnt9 in the early larva, which differentiates into two distinct territories, one endodermal in the central portion of the gut and another mesodermal in a bilateral pair of anterior domains near the apical–mantle boundary.

wnt10 is expressed from the mid gastrula stage in a posterior ectodermal domain, which acquires a subterminal position within the pedicle lobe in the early larva (Fig. 4 and Additional file 4: Fig. S4). Additionally, we detect wnt10 transcripts in the late gastrula at a dorsal ectodermal patch of the apical lobe, similar to the dorsal domain of wnt1t, and in the late larva at the posterior mesoderm.

wnt11 is expressed in a posterodorsal ectodermal domain in the early gastrula, similar to wnt4 (Fig. 4 and Additional file 4: Fig. S4). The domain encircles the pedicle lobe ectoderm in the early larva and becomes reduced to a narrow ectodermal stripe on the ventral portion of the pedicle lobe in the late larva. In the early larva, wnt11 is also expressed in the ventral ectoderm at the mantle–pedicle boundary, and in the posterior endoderm of the larval gut (Fig. 4 and Additional file 4: Fig. S4).

wnt16 is expressed in the invaginating endomesoderm and vegetal ectoderm around the blastopore in the late blastula (Fig. 4 and Additional file 4: Fig. S4). During gastrulation, the endomesodermal expression clears, and only the ectodermal domain remains as lateral patches near the blastopore lip. With the blastopore closure, wnt16 forms a heart-shaped domain in the ectoderm and presumably mesoderm at the ventral midline of the mantle lobe in the early larva.

wntA appears in the mid gastrula as paired, ventral ectodermal domains located at the posterior portion of the apical lobe (Fig. 4 and Additional file 4: Fig. S4). In the late larva, these anterior ectodermal domains are cleared, and wntA expression is activated in paired, ventral mesodermal bands adjacent to the mouth.

Overall, Wnt genes are primarily expressed in the ectoderm, in diverse partially overlapping domains along the anteroposterior axis (Additional file 6: Fig. S6 and Additional file 7: Fig. S7).

Frizzled genes exhibit gradual expression changes throughout embryogenesis

Frizzled genes encode seven-pass transmembrane proteins with an extracellular cystein-rich domain and act as receptors in Wnt signaling pathways [66]. There are five Frizzled subfamilies in metazoans [67], but the subfamily fz3/6 is only found in tunicates and vertebrates [68]. In the brachiopod T. transversa, we identified a total of four Frizzled genes with a single ortholog for the fz1/2/7, fz5/8, fz9/10, and fz4 subfamilies, respectively (Additional file 8: Fig. S8).

Frizzled receptors are expressed throughout T. transversa development. In the unfertilized oocyte, fz1/2/7 and fz5/8 are highly expressed (Fig. 2). The expression of fz1/2/7 remains high from the oocyte to juvenile stages, while the expression of fz5/8 peaks before gastrulation and decays over time. fz4, which is initially expressed at lower levels, peaks late in development, at the larval stages, an expression profile complementary to the one of fz5/8 (Fig. 2). In contrast, fz9/10 expression increases during gastrulation and remains relatively constant in subsequent stages.

Overall, each Frizzled shows a unique expression profile but in contrast to Wnt dynamic changes, the levels of Frizzled transcripts change more gradually during development.

Frizzled expression domains occupy broad but distinct body regions

We carried out in situ hybridization for all Frizzled genes of T. transversa to reveal their domains of expression during axial elongation (Fig. 6).

Fig. 6

Whole-mount colorimetric in situ hybridization of Terebratalia transversa Frizzled and Wnt antagonist genes. Developmental stages between late blastula and late larva for fz1/2/7, fz4, fz5/8, fz9/10, sfrp1/2/5, dkk5, and wif. The panels show representative expression patterns for each sample. The samples are oriented with a blastoporal/ventral view and anterior end to the top, except for sfrp1/2/5 early gastrula showing the animal pole. Black arrowheads indicate the apical–mantle boundary. White arrowheads demarcate the mantle–pedicle boundary. Scale bars = 20 µm

fz1/2/7 expression is mostly ubiquitous (Fig. 6 and Additional file 9: Fig. S9). It is expressed in all tissues of the late blastula, with strong signal in the animal pole ectoderm and invaginating endomesoderm. During gastrulation, the anterior and middle portions of the body continue to express fz1/2/7 across all germ layers, but the posterior transcripts get progressively cleared from the pedicle lobe tissues. In larval stages, fz1/2/7 is upregulated in the terminal portion of the pedicle lobe ectoderm, and becomes nearly ubiquitous again in the late larva.

fz4 is first expressed in the animal pole ectoderm of the late blastula (Fig. 6 and Additional file 9: Fig. S9). These anterior transcripts form a subapical ectodermal ring encircling the animal pole of the early gastrula that localizes to the posterior portion of the apical lobe in subsequent stages. fz4 is also expressed in the anterior mesoderm from the early gastrula. In the late gastrula, we detect fz4 transcripts in the dorsal ectoderm between the mantle and pedicle lobes, a domain that becomes stronger in the late larva as it expands around the dorsal ectoderm of the pedicle lobe as well as in the infolded ectodermal and mesodermal tissues of the growing mantle lobe. Additionally, a fz4 domain appears at the posterior tip of the pedicle lobe ectoderm in the late larva.

fz5/8 is mainly expressed at the anteriormost region of the embryo’s ectoderm and mesoderm (Fig. 6 and Additional file 9: Fig. S9). We first detect transcripts in the animal pole ectoderm of the late blastula, these transcripts become restricted to the anterodorsal portion of the ectoderm in the early gastrula, and finally expand to the anteroventral ectoderm from mid gastrula to early larva. This ectodermal territory of fz5/8 is complementary to the ectodermal domain of fz4 in the apical lobe without overlapping with the apical lobe domain of wnt5 (Fig. 5E,F). fz5/8 is also expressed in the anterior mesoderm from the early gastrula and in the chaetae sacs of the late larva.

fz9/10 transcripts are limited to the middle portion of the body throughout development (Fig. 6 and Additional file 9: Fig. S9). In the late blastula, fz9/10 is initially expressed in the ectoderm posterior to the blastopore, but this domain expands to cover almost the entire ectoderm around the blastopore of the early gastrula; it is only absent from a narrow anterior portion. With gastrulation, fz9/10 begins to be expressed in the entire mesoderm, as well as in the ectoderm of the apical–mantle boundary, and in the anterior portion of the pedicle lobe ectoderm. Expression in the lateral mantle lobe ectoderm is weaker, and the terminal portion of the pedicle lobe is cleared from fz9/10 transcripts. Interestingly, the anterior limit of fz9/10 expression abuts the posterior limit of fz4 expression in the apical lobe. fz9/10 expression in the late larva fades, except in the posterior apical lobe ectoderm, and in the anterior and posterior region of the mesoderm.

Taken together, the expression of most Frizzled genes extend over broad but distinct domains along the body. Except for fz1/2/7, which is expressed ubiquitously, the ectodermal territories of fz5/8, fz4, and fz9/10 are sequentially arranged from anterior to posterior, respectively, without overlap until the late gastrula stage and the onset of larval morphogenesis.

Wnt antagonist expression is mostly limited to the anterior end

To obtain a more comprehensive picture of the Wnt signaling landscape in T. transversa, we also analyzed the expression of three Wnt antagonist genes: a Secreted Frizzled-Related Protein (sfrp), a Dickkopf protein (dkk), and a Wnt Inhibitory Factor (wif).

sFRP is a soluble protein that antagonizes Wnt activity by directly binding to Wnt ligands or to Frizzled receptors [69]. It has a cysteine-rich domain with high affinity to Wnt proteins. The sFRP family can be divided into two subfamilies, sfrp1/2/5 and sfrp3/4 [69, 70]. In T. transversa, we only identified a sfrp1/2/5 ortholog (Additional file 10: Fig. S10), which is highly expressed throughout development (Fig. 2). The transcripts locate to the animal pole ectoderm in the late blastula and forms a strong anterior ectodermal domain in subsequent stages, in a pattern similar to the expression of fz5/8 (Fig. 6 and Additional file 11: Fig. S11). sfrp1/2/5 is also expressed in a narrow domain at the anterior mesoderm throughout development, and in a paired domain in the mantle lobe mesoderm restricted to the early larva stage. In the late larva, the anterior domain becomes limited to dorsal patches on the dorsal ectoderm of the apical lobe.

Dkk is a secreted glycoprotein containing two cysteine-rich domains that antagonizes Wnt signaling by inhibiting lrp5/6 co-receptors [71, 72]. These proteins are generally divided into two subfamilies, dkk1/2/4 and dkk3 [71]. In T. transversa, however, we identified a single dkk ortholog that groups with a previously unidentified Dkk subfamily, named hereafter dkk5 (Additional file 12: Fig. S12). Our phylogenetic analysis reveals that non-vertebrate deuterostomes, such as hemichordates and cephalochordates, have orthologs for dkk1/2/4, dkk3, and dkk5, suggesting this was the ancestral Dkk repertoire of bilaterians, and that dkk1/2/4 and dkk5 were subsequently lost in protostomes and vertebrates, respectively (Additional file 12: Fig. S12). The expression of dkk5 in T. transversa is upregulated in the late blastula and downregulated in the juvenile (Fig. 2). It localizes to the animal pole ectoderm in the late blastula, and anterior ectoderm in subsequent stages similar to the expression of sfrp1/2/5, except that dkk5 becomes limited to the ventral ectoderm and is not expressed in the mesoderm (Fig. 6 and Additional file 11: Fig. S11).

Wif is another protein that inhibits Wnt activity by direct binding to Wnt proteins [73]. The protein has five EGF repeats and a typical WIF domain which is shared with RYK receptor tyrosine kinases [72, 73]. In T. transversa, we identified one wif ortholog (Additional file 13: Fig. S13). The expression levels are relatively low and somewhat stable throughout development, with a peak in the late gastrula (Fig. 2). Unlike sfrp1/2/5 and dkk5, wif is mainly expressed in mesodermal tissues throughout the analyzed developmental stages; it is also broadly but faintly expressed in the ectoderm until the early larva, and it is not expressed in the endoderm (Fig. 6 and Additional file 11: Fig. S11).

Overall, the expression of the analyzed Wnt antagonist genes is restricted to the anterior portion of the ectoderm (sfrp1/2/5 and dkk5), and to the mesoderm (wif), regions which coincide with the absence or limited expression of Wnt ligands.

Planar cell polarity genes show patched expression during axial elongation

Proper regulation of planar cell polarity (PCP) is crucial to guide morphogenetic processes, such as convergent extension, and to orient the formation of structures during development [74, 75]. We identified several core components of the PCP pathway in T. transversa. These include orthologs for dishevelled (dsh, also known as dvl), diego (dgo, also known as ankrd6 or diversin), prickle (pk), flamingo (fmi, also known as stan or celsr), strabismus (stbm, also known as vang or vangl), and the downstream transducer c-jun n-terminal kinase (jnk, also known as mapk8). Then, we analyzed their expression between the early and late gastrula stages.

Dsh is a central regulator of the Wnt/beta-catenin, Wnt/PCP, and Wnt/calcium pathways [76]. The protein has three conserved domains (DIX, PDZ, and DEP domains), and two conserved regions before and after the PDZ domain [77]. In T. transversa, we identified a single copy of dsh (Additional file 14: Fig. S14) which is highly expressed in every developmental stage (Additional file 15: Fig. S15). The expression is stronger in a narrow dorsal domain of the anterior ectoderm and in the anterior portion of the mesoderm (Fig. 7), but dsh transcripts are also expressed at lower levels in all embryonic tissues (Additional file 16: Fig. S16).

Fig. 7

Whole-mount colorimetric in situ hybridization of Terebratalia transversa Wnt/PCP pathway components. Developmental stages between early gastrula and late gastrula for dsh, dgo, pk, fmi, stbm, and jnk. The panels show representative expression patterns for each sample. The stainings for dsh are underdeveloped (see Additional file 16: Fig. S16). The samples are oriented in a blastoporal/ventral view (left) and in a lateral view (right). Black arrowheads indicate the apical–mantle boundary. White arrowheads demarcate the mantle–pedicle boundary. Scale bars = 20 µm

Dgo is a cytoplasmic protein containing 6–8 ankyrin repeat domains that suppresses Wnt/beta-catenin signaling and activates the Wnt/PCP pathway [78, 79]. We found a single dgo ortholog in T. transversa with six ankyrin repeats (Additional file 17: Fig. S17). dgo transcripts are deposited maternally, quickly degrade, and only recover higher levels of expression in the late larva (Additional file 15: Fig. S15). However, we still detect two pairs of dorsal ectodermal domains in the apical and pedicle lobes of the late gastrula (Fig. 7).

Pk is a protein that contains a PET domain and three LIM domains [80] and competes with Dgo for Dsh binding [81]. We identified a single pk ortholog in T. transversa (Additional file 18: Fig. S18), which is highly expressed throughout development (Additional file 15: Fig. S15). pk transcripts are present in a small patch of ectoderm posterior to the blastopore in the early gastrula (Fig. 7). In the mid gastrula, pk is upregulated in the mesoderm and forms paired ventral domains within the mantle lobe mesoderm of the late gastrula, when paired ventral domains also appear in the apical lobe ectoderm. Given the high expression levels of pk in our RNA-Seq data, we cannot exclude the possibility that it is more broadly expressed than we could detect in our in situ hybridization.

Fmi is a seven-pass transmembrane cadherin that regulates cell polarity [82, 83]. In T. transversa, we identified one ortholog of fmi (Additional file 19: Fig. S19). In contrast to other polarity genes, it is not expressed maternally; fmi expression peaks around the late gastrula (Additional file 15: Fig. S15). fmi transcripts are present in most ectodermal tissues but show stronger signal on bilateral patches present in the apical lobe ectoderm of the late gastrula (Fig. 7).

Stbm is a four-pass transmembrane protein that forms a signaling complex with FMI [84, 85]. Terebratalia transversa stbm ortholog (Additional file 20: Fig. S20) is initially expressed in high levels, which gradually decay during development (Additional file 15: Fig. S15). Accordingly, stbm is ubiquitously expressed in embryonic tissues during gastrulation (Fig. 7).

Jnk is a kinase that regulates epithelial metamorphosis and is a downstream transducer of the PCP pathway [86]. The jnk ortholog in T. transversa (Additional file 21: Fig. S21) is highly expressed throughout the development (Additional file 15: Fig. S15) and ubiquitously expressed in the late gastrula, except for broad bilateral regions in the apical lobe ectoderm (Fig. 7).

In conclusion, while fmi, stbm, and dsh are expressed ubiquitously, the other cell polarity genes dgo, pk, and jnk are expressed in non-overlapping patches at different regions of the late gastrula.

Distinct Wnt subregions coincide with larval body subdivisions

Given the importance of specific ligand–receptor contexts for the outcome of Wnt signaling [29, 30], we compiled the data above to describe the combination of Wnt ligands, Frizzled receptors, and antagonist genes being expressed in the different tissues of T. transversa embryos throughout ontogeny.

We were able to identify distinct transcriptional subregions, each expressing a unique combination of ligands, receptors, and antagonists, along the brachiopod embryonic axes. At the onset of gastrulation, late blastula embryos exhibit two ectodermal Wnt subregions corresponding to the animal and vegetal tissues (Additional file 22: Fig. S22A). Tissues in the animal pole express Wnt antagonists and all Frizzled genes (except fz9/10) but no Wnt genes, while the tissues in the vegetal pole express four ligands (wnt8 and wnt16 more broadly) within a fz1/2/7 and fz9/10 receptor context. At this stage, the endomesoderm expresses the same ligands as vegetal tissues, except for wnt4, but fz1/2/7 is the only receptor expressed in the invaginating archenteron. From early to mid gastrula, animal and vegetal Wnt subregions subdivide due to changes in the relative position between receptor domains and the upregulation of other ligands. Notably, a subapical Wnt subregion differentiates in the early gastrula, characterized by the expression of fz4, wnt8 and wnt5, when fz5/8 and fz4 become no longer coexpressed at the animal pole (Additional file 22: Fig. S22B). Likewise, two Wnt subregions emerge in the mid gastrula from the initial vegetal landscape when fz9/10 expression becomes subterminal—one midbody, continuing to express fz1/2/7, fz9/10, wnt7, and wnt16, and one posterior, expressing only fz1/2/7 and several Wnt genes (Additional file 22: Fig. S22C). Finally, at the late gastrula stage, fz9/10 is cleared from a portion of the midbody and fz1/2/7 is cleared from the posterior end, giving rise to an additional midbody subregion expressing fz1/2/7, wnt5 and wnt16, and to another subregion at the posterior end expressing only Wnt but no Frizzled genes (Fig. 8A). Therefore, from late blastula to the elongated late gastrula, we observe a progressive differentiation of ectodermal Wnt subregions, from the initial animal and vegetal ones, to the six distinct subregions along the anteroposterior axis of the embryo (Fig. 8A, Table 2).

Fig. 8

Summary of Terebratalia transversa Wnt signaling landscape during axial elongation. A Late gastrula in blastoporal/ventral view. Schematic drawings of Wnt genes colored by subfamilies, Frizzled genes by lighter shades of gray, and antagonists by darker shades of gray. The spatial localization of expression domains is superimposed on the embryo (left) and projected to highlight the individualized Wnt genes within the different transcriptional subregions grouped by germ layer (right). The gray boxes show the pattern of individual genes mapped to the embryo for clearer visualization of overlapping domains. B Late gastrula and early larva embryos showing the correspondence between Wnt subregions (dashed lines) and the morphological boundaries of the larva. The samples were stained for F-actin (cell membranes) to highlight the cell shape differences between body regions

Table 2 Wnt signaling transcriptional subregions in the ectoderm of the brachiopod Terebratalia transversa

Remarkably, the Wnt subregions established in the late gastrula stage coincide with the morphological subdivisions of the larval body (Fig. 8B). The anteriormost subregion expressing antagonist genes and fz5/8 but no Wnt ligands (except wnt1t), gives rise to the larval apical organ and neuropile at the narrower portion of the apical lobe [87]. In turn, the adjacent subapical subregion expressing Wnt genes and fz4 but no antagonists, undergoes intense cell proliferation [63] and gives rise to the wider portion of the larval apical lobe and adult lophophore, the crown of tentacles that brachiopods use for filter-feeding (Fig. 8). The subregion posterior to the apical lobe, which expresses fz1/2/7, fz9/10, wnt1, and wnt7, gives rise to a deep epithelial invagination that demarcates the apical–mantle boundary [64]. The abutting subregion expressing solely fz1/2/7, wnt5, and wnt16, gives rise to a prominent, skirt-like epithelial outgrowth that forms the larval mantle lobe, a structure that is then reversed anteriorly during metamorphosis [53] (Fig. 8). This reversible mantle lobe is considered an evolutionary innovation of lecithotrophic brachiopod larvae [88]. Tissues posterior to the mantle lobe, which express fz9/10, wnt1, wnt4, and wnt10, become the posterior end of the adult body, while the posteriormost subregion expressing wnt5, wnt6, and wnt11, but no Frizzled genes, gives rise to the posterior end of the larval pedicle lobe, which becomes the structure that attaches adult brachiopods to the substrate [54] (Fig. 8). Overall, a combinatorial analysis of the expression of Wnt signaling components in T. transversa reveals that the distinct transcriptional subregions established during gastrulation correlate with the distinct morphogenetic outcomes along the anteroposterior axis of larval brachiopods.

Discussion

Our work characterizes the expression of Wnt signaling components during the embryonic development of T. transversa, a brachiopod species that has largely retained the ancestral repertoire of Wnt genes. We find that Wnt genes are upregulated at the onset of gastrulation and during axial elongation, and that the combinatorial expression of ligands, receptors, and antagonists, forms distinct transcriptional subregions along the primary body axis. The boundaries between these subregions coincide with the morphological subdivisions of the larval body, suggesting that a combinatorial Wnt signaling landscape—or Wnt code—may play a role in the regionalization of the brachiopod anteroposterior body axis.

We observe, for example, that the expression of Wnt inhibitors in T. transversa correlates with the anteriormost fates of the apical lobe in the brachiopod larva (Fig. 8). In addition to sfrp and dkk5, the anterior subregion of the brachiopod expresses fz5/8, a gene known to activate sFRP and Dkk antagonists in sea urchins [89, 90], and six3/6 [58], a transcription factor that antagonizes Wnt/beta-catenin signaling in different animals [91,92,93,94], and expresses almost no Wnt genes. This could indicate that Wnt/beta-catenin signaling is being inhibited at the anteriormost brachiopod subregion, a hypothesis consistent with the observation that early over-activation of this pathway in T. transversa dramatically reduces the expression of anterior markers and results in the complete loss of anterior structures (apical and mantle lobes) (see Supplementary Figure S9 in [64], and Fig. 4d and Supplementary Fig. 6b in [63]). Interestingly, the adjacent subregion in the brachiopod expresses wnt8 and other ligands (Additional file 22: Fig. S22), a pattern comparable to the expression of Wnt components in the anterior neuroectoderm of sea urchin embryos [95,96,97]. In echinoderms, this pattern emerges from a negative feedback loop between the anterior Wnt inhibitors (activated by fz5/8) and the posterior wnt8-mediated Wnt/beta-catenin signaling, which helps to consolidate the anterior and posterior neuroectoderm identities [89, 90]. Whether a comparable regulatory logic could also be contributing to the patterning of the brachiopod neuroectoderm and apical lobe subdivision remains a speculation at this point, and will need to be experimentally tested in future studies.

Our study also reveals a correlation between wnt5 and the formation of the mantle lobe in T. transversa. The Wnt subregion that gives rise to the mantle lobe is characterized by the expression of wnt5 and fz1/2/7 and emerges only in the late gastrula with the clearance of fz9/10 transcripts from embryo’s midbody (Fig. 8). wnt16 is also expressed at this subregion but limited to the midline around the closing blastopore. During larval morphogenesis, wnt5 remains expressed at the growing edges of the elongating mantle lobe (Fig. Additional file 3: Fig. S3). In other animals, wnt5 regulates convergent extension movements during tissue morphogenesis via the Wnt/PCP pathway [3, 16, 17, 98] and is commonly expressed in tissue outgrowths such as the vertebrate tail and limb buds [23, 99, 100]. This contextual similarity of wnt5 expression suggests that convergent extension could be a possible mechanism for the elongation of the mantle lobe in brachiopod larvae. One piece of evidence consistent with this hypothesis is that a late over-activation of the Wnt/beta-catenin pathway in T. transversa inhibits the elongation of the mantle lobe (see Supplementary Figure S8 in [64]). However, whether wnt5 or the Wnt/PCP pathway plays a role in this process, and whether the specific receptor context in the brachiopod mantle lobe is important to control the signaling output, remains to be determined by targeted functional approaches.

Altogether, these observations suggest the possibility that the different transcriptional subregions could contribute to the regional specification and morphogenetic control of tissues along the anteroposterior axis of brachiopods. It is important to note, however, that these combinatorial patterns are solely based on coexpression data, which does not guarantee actual signaling, and that long-range interactions might still occur due to the secreted nature of Wnt proteins and their highly promiscuous ligand–receptor binding. Nevertheless, our brachiopod data hints at the idea that changes in these Wnt subregions may have been important for developmental innovations in the primary body axis of animals.

A broad phylogenetic survey comparing our brachiopod data to other animals suggests that the combinatorial expression of Wnt signaling components along the body emerged with the Cnidaria–Bilateria expansion of Wnt and Frizzled gene families (Fig. 9A). Ctenophores have only four Wnt, two Frizzled, and one sFRP and their expression is not staggered along the embryonic ectoderm [101]. Sponges exhibit Wnt territories along the larval body axis [102,103,104], but the expression data is scarce and some species underwent large Wnt expansions which have no clear orthologs with the bilaterian Wnt genes [103, 104]. While the early evolution of Wnt genes remains poorly understood, the Wnt repertoire between cnidarians and bilaterians is well conserved with clear orthologs, enabling a more informative comparison about the evolution of the Wnt signaling landscape (Fig. 9A).

Fig. 9

Comparative overview of the Wnt signaling landscape in animals. A Phylogenetic tree showing the repertoire of Wnt ligands, Frizzled receptors, and Wnt antagonists in different animal groups, and their presumed gene expansions and gene losses during evolution. Ctenophores have few Wnt genes while poriferans, due to lineage-specific expansions, have several Wnt genes; their orthology to bilaterian Wnt genes, however, remains uncertain. For this reason, it is assumed that Wnt and Frizzled genes expanded to 13 and 4 subfamilies, respectively, in the last common ancestor of cnidarians and bilaterians. Alternatively, ancestral orthologs of Wnt subfamilies may have been lost or significantly modified in ctenophores and sponges. The main losses in bilaterians were wntA in vertebrates, wnt2 and wnt11 in echinoderms, wnt3 in protostomes, and multiple Wnt genes in arthropods—but, generally, they show a well-conserved Wnt repertoire. Asterisks indicate Wnt genes with uncertain orthology. The tree topology is based on [105, 106]. B Schematic drawings illustrating the expression of Wnt signaling components at the late gastrula stage of different metazoan species. The receptor (lighter shades of gray) and antagonist (darker shades of gray) subregions are superimposed on the embryo (left). The region of expression of individual Frizzled genes (black lines) and Wnt genes (colored lines) is shown for each species depicted by germ layer. Embryos are oriented with aboral/anterior end up and oral/posterior end bottom. Asterisk indicates the blastopore position. The gene complement, orthology assignment, and expression patterns were compiled from previous works on Ctenophora [101, 107], Porifera [102,103,104], Cnidaria [24, 25, 108, 109], Xenacoelomorpha (based on Xenoturbella transcriptome [110] as the Wnt genes in acoels are highly derived [111]), Vertebrata [1], Cephalochordata [23, 112,113,114,115,116,117,118,119], Echinodermata [68, 89, 90, 95,96,97, 120,121,122], Hemichordata [41], Brachiopoda (based on [64] and this study), Annelida [27, 40, 47, 48, 123], and Arthropoda [44]

Despite the great morphological differences, most both cnidarian and bilaterian embryos exhibit a common spatial organization of their Wnt landscape with antagonist genes expressed anteriorly, Wnt genes expressed predominantly posteriorly, and Frizzled genes expressed more broadly along the body (Fig. 9B). Frizzled expression is especially well conserved across groups. While most animals provide maternal fz1/2/7 and fz5/8 transcripts in the egg (Table 3), embryos of every group investigated so far express fz5/8 at the aboral/anterior ectoderm, often coexpressed with sfrp and rarely with Wnt ligands (Fig. 9B). Expression of fz5/8 in the anterior mesoderm is also common. fz9/10 expression is usually complementary to fz5/8, localizing to the midbody/trunk portion of the body, and fz1/2/7 is normally expressed broadly, even though the anterior and posterior domain limits can vary (Fig. 9B). A closer comparison between T. transversa and the hemichordate Saccoglossus kowalevskii reveals that the spatial organization of their Frizzled domains is strikingly similar (Additional file 23: Fig. S23A). The exception to this conservation is fz4, which is subapical in brachiopods and hemichordates, but shows a more variable expression in other groups. Overall, these similarities indicate that cnidarians and bilaterians share an ancient axial organization of Frizzled domains that remained conserved during evolution.

Table 3 Maternal load of Wnt and Frizzled transcripts in metazoan eggs

In contrast, the expression of Wnt genes is more variable. The maternal load of Wnt transcripts is diverse and can even differ within a group (Table 3), while the embryonic expression shows great variability across the body axis and germ layers between groups (Fig. 9B). wnt2 is the least variable gene. Except for echinoderms which lack the gene, wnt2 is the only Wnt to be solely expressed in a single body region and germ layer across groups—the anterior ectoderm. wnt1 has the most consistent expression domain at the posterior ectoderm, but it is also expressed at the anterior ectoderm (Fig. 9B and Additional file 23: Fig. S23A). The other frequent ectodermal patterns are wnt5 and wnt8 in the anterior, and wnt4, wnt3, wnt6, and wnt11 at the posterior end. However, these genes are also expressed in other positions and germ layers. The most diverse patterns in this sense are from the genes wnt11, wnt9, and wnt5 (Fig. 9B). wnt9 and wnt11, for example, are expressed at entirely different anterior and posterior receptor contexts between brachiopod and hemichordate embryos (Additional file 23: Fig. S23A). In fact, based on our expression data, even the consistent genes wnt1 and wnt2 are expressed in different receptor contexts between groups. A more detailed comparison between the brachiopod and hemichordate landscapes reveals only a few relatively conserved Wnt–Frizzled combinations (Additional file 23: Fig. S23B). Altogether, this comparative analysis suggests that changes in Wnt expression, like the addition or re-deployment of domains to different tissues or receptor contexts, may have occurred frequently during evolution.

This variable nature of Wnt expression might be an important factor influencing axial evolution. Evolutionary changes in the ligand–receptor context could effectively alter cell fates and morphogenetic events, generating tissue and shape diversity and providing a basis for developmental innovations along the body [3, 25, 26]. Such Wnt function shuffling is indeed associated with novelties in the chordate lineage [112]. Investigating the combinatorial landscape of Wnt signaling components across the phylogeny will be crucial to uncovering the role of Wnt shuffling as a potential mechanism contributing to the diversification of the metazoan body axis.

Conclusions

Our data reveals a correlation between distinct Wnt transcriptional subregions and the morphological subdivisions in the larval body of a brachiopod. We hypothesize that the underlying combinatorial landscape may play a role in the patterning and morphogenesis of the different regions along the anteroposterior axis, and that changes in this landscape could be associated with the evolution of the reversible mantle lobe, a morphogenetic innovation of brachiopod larvae. As Wnt developmental expression is variable across animal groups, we propose that evolutionary changes in ligand–receptor context may have been important to axial evolution in animals.

Methods

Sample collection

T. transversa (Sowerby, 1846) adult specimens were collected by dredging the rocky seabeds around Friday Harbor, San Juan Islands, USA, and were kept in a seawater table with running seawater at the Friday Harbor Laboratories (University of Washington). To obtain embryos, we dissected the gonads of ripe individuals and fertilized the gametes in vitro as previously described [56, 132]. We cultured the embryos in E-ware glass bowls (i.e., glassware never exposed to chemicals) with filtered seawater and temperature around 7.6 °C. Water in culturing bowls was exchanged daily. Using a glass pipette, we collected samples for RNA sequencing and for in situ hybridization at representative developmental stages (Table 1). For the RNA-Seq samples, we collected two biological replicates, each containing the eggs of a single female fertilized with mixed sperm from three different males. We preserved the embryos directly in RNAlater at room temperature. For the in situ hybridization samples, we fixed the embryos for 1 h in 4% paraformaldehyde at room temperature, washed thoroughly in 1 × PBS with 0.1% Tween-20, and stored them in 100% methanol at − 20 °C.

RNA sequencing and analyses

We extracted the total RNA using Trizol. Library preparation and sequencing was performed at the EMBL Genomic Core Facilities (GENECORE). The samples were randomized and multiplexed on four lanes of a Illumina HighSeq 2000 system, and sequenced to an average of 24 ± 5 million 50 bp of unstranded single-end reads. To quantify transcript abundances, we used Kallisto v0.46.0 [133] to pseudoalign the reads to a reference transcriptome of T. transversa. This reference transcriptome was originally assembled with Trinity [134] from a deeply sequenced, unstranded paired-end RNA-Seq dataset of mixed developmental stages [135]. Next, we imported the estimated counts from Kallisto to DESeq2 [136] to estimate the library size factors and data dispersion, homogenize the variance across expression ranks, and apply a variance-stabilizing transformation to the data before the expression analyses. To visualize the normalized expression data, we generated heatmaps using pheatmap [137] and ggplot2 [138] in R [139] running in RStudio Desktop [140].

Due to the unstranded nature of our sequencing data, we analyzed the coverage of mapped reads to identify potential biases in the quantification of expression levels. For that, we mapped the RNA-Seq reads to the transcripts of Wnt signaling components of T. transversa using Salmon v1.10.1 [141], and then created read coverage plots for each gene using the ggcoverage package [142]. While the majority of genes show uniform read coverage profiles, we identified two cases of uneven coverage that significantly overestimates the expression levels of wnt4 and wntA in early developmental stages (Additional file 24: Fig. S24). In these samples, reads predominantly mapped to the 3′ region of the transcript, while the Wnt coding sequence region had a low mapping rate. This pattern could be explained by the presence of an isoform lacking the Wnt domain, or by the expression of another gene in the opposite strand at the same locus, potentially assembled in the same contig due to the unstranded reads. Since both wnt4 and wntA transcripts have long open reading frames in the antisense direction, and these transcripts fully map to a single scaffold in a draft genome assembly of T. transversa, the latter hypothesis is more likely. This suggests that, in these two cases, the high expression values result from the contiguous antisense genes, rather than from wnt4 or wntA.

The code and pipeline for the RNA-Seq analysis and the mapping and coverage files are available in the paper’s repository [143].

Gene orthology

We searched for Wnt signaling orthologs in T. transversa by querying known Wnt genes against the available transcriptome using BLAST + [144]. To resolve the orthology of the obtained putative orthologs, we aligned the protein sequences of T. transversa with well-annotated proteins from other metazoans using MAFFT 7.310 [84], removed non-informative sections using GBlocks 0.91b [145], and inspected the multiple sequence alignment using UGENE [146]. Using the blocked alignments as input, we ran a maximum likelihood phylogenetic analysis using the automatic model recognition and rapid bootstrap options of RAxML 8.2.12 [147], and rendered the resulting trees using the Interactive Tree Of Life web application [148]. The gene orthology pipeline is available in the paper’s repository [143].

Cloning and in situ hybridization

We synthesized cDNA from a total RNA extraction of mixed developmental stages of T. transversa using the SMARTer RACE cDNA Amplification kit (Clontech). For each transcript, we designed gene-specific primer pairs within the coding sequence using Primer3 [149] to obtain product sizes between 800 and 1200 bp (Additional file 25: Table S1). We then cloned each fragment into a pGEM-T Easy Vector, amplified the antisense sequences using T7 or SP6 polymerase, and synthesized DIG-labeled riboprobes using the MEGAscript kit (Ambion). Finally, to visualize gene expression, we followed the established protocols in T. transversa for single colorimetric in situ hybridization [58, 150], and double fluorescent in situ hybridization [63, 64]. Cloning details are available in the paper’s repository [143].

Microscopy and image processing

We mounted the embryos between a glass slide and a coverslip, supported by clay feet, in 70% glycerol in PBS, and imaged the representative expression patterns for each sample. For colorimetric in situ data, we acquired differential interference contrast (DIC or Nomarski) images using a Zeiss AxioCam HRc attached to a Zeiss Axioscope A1. For fluorescent samples, we scanned volumetric stacks in a Leica TCS SP5 confocal microscope and generated maximum intensity projections using Fiji [151]. We adjusted intensity levels to improve contrast—without clipping signal from high or low ranges—using Fiji or GIMP, and assembled the illustrations and figure plates using Inkscape. Images at the original resolution are available in the paper’s repository [143].