The developmental basis for the recurrent evolution of deuterostomy and protostomy

Published online:


The mouth opening of bilaterian animals develops either separate from (deuterostomy) or connected to (protostomy) the embryonic blastopore, the site of endomesoderm internalization. Although this distinction preluded the classification of bilaterian animals in Deuterostomia and Protostomia, and has influenced major scenarios of bilaterian evolution, the developmental basis for the appearance of these different embryonic patterns remains unclear. To identify the underlying mechanisms, we compared the development of two brachiopod species that show deuterostomy (Novocrania anomala) and protostomy (Terebratalia transversa), respectively. We show that the differential activity of Wnt signalling, together with the timing and location of mesoderm formation, correlate with the differential behaviour and fate of the blastopore. We further assess these principles in the spiral-cleaving group Annelida, and propose that the developmental relationships of mouth and blastoporal openings are secondary by-products of variations in axial and mesoderm development. This challenges the previous evolutionary emphasis on extant blastoporal behaviours to explain the origin and diversification of bilaterian animals.

In most animals, the precursor cells that form the inner tissues internalize at a specific site of the embryo in a process called gastrulation, which often involves the formation of a transient opening—the blastopore—that can later relate to the mouth and/or the anal openings1. The developmental connection of this transient blastoporal opening to the mouth (Fig. 1a) was typically used to divide bilaterally symmetrical animals (Bilateria) into Deuterostomia and Protostomia2 (Fig. 1b), a node that recent molecular phylogenies strongly support3,4. In the traditional view, the mouth forms independently of the blastopore in Deuterostomia (literally ‘secondary mouth’), but is coupled to the blastoporal opening in Protostomia (‘first mouth’), although tremendous variation in the blastoporal fate is seen in Protostomia5,6 (Fig. 1b; Supplementary Table 1). Nearly all scenarios for bilaterian evolution presume the ancestral correspondence of the bilaterian blastoporal opening to the mouth of anthozoan cnidarians to explain the appearance of a tube-like alimentary canal with mouth and anus7 (Fig. 1c). The amphistomy concept assumes that the ancestral pre-bilaterian condition was a ‘Gastraea’-like organism with a single opening to the gut, which later formed a slit-like blastoporal opening that closed laterally with its ends forming the mouth and anus simultaneously7,​8,​9,​10,​11,​12,​13,​14. Modifications of blastopore closure, for example by convergent extension11, in this ancestral pattern of amphistomic gastrulation7,​8,​9,​10,12,13 would secondarily generate both deutero- and protostomy (Fig. 1c). The planuloid–acoeloid scenario15,​16,​17 deduces protostomy to be the ancestral mode of development in Bilateria, suggesting the secondary evolution of deuterostomy by a precocious formation of the mouth at the anterior area of the embryo and the formation of the anus from the blastopore17,​18,​19 (Fig. 1c). Alternatively, deuterostomy could be the ancestral gastrulation mode in Bilateria20,21, which is supported by its phylogenetic distribution5 (Fig. 1b). If the behaviour and fate of the transient blastoporal opening is important for understanding the evolution of Bilateria, knowledge of the exact developmental mechanisms underlying deuterostomy and protostomy is critical. In this regard, the striking presence of both modes of gastrulation in Protostomia (Fig. 1b), in particular between closely related species, can be used to understand the molecular and developmental basis for the recurrent evolution of deuterostomy and protostomy (Supplementary Table 1).

Figure 1: Fate of the blastopore and the evolution of Bilateria.
Figure 1

a, During deuterostomy, the blastoporal opening (bp; in blue) migrates from the vegetal pole to the posterior end and either closes (as in N. anomala; the anus opens around the red triangle) or forms the anus. The mouth (mo) forms anew. During protostomy, the blastoporal opening migrates anteriorly and becomes the mouth. The anus, if present, opens later. T. transversa lacks an anus. For each developmental stage in a the left panels are lateral views and the right panels are vegetal/ventral views. The asterisk denotes the animal/anterior pole. Scale bars, 50 μm. b, The relation between the blastoporal, mouth and anal openings was traditionally used to split Bilateria into Deuterostomia and Protostomia, which is currently supported by molecular phylogenies. However, several lineages of Protostomia exhibit deuterostomic development, obscuring the ancestral mode of development for Protostomia (question mark). c, The main scenarios for the evolution of Bilateria and the origin of a through gut presume the direct correspondence between the cnidarian mouth and the bilaterian blastopore. In the amphistomy scenario, elongation and mid closure of the blastopore (in blue) of the ancestral bilaterian (bilatero-gastraea) originated the mouth and anus (amphistomy). Subsequent variation during blastopore closure (small arrows) generated deuterostomy and protostomy. In the planuloid–acoeloid scenario, the blastoporal opening (in blue) of a planula-like ancestor formed the mouth (protostomy; curved arrow) of an acoeloid-like bilaterian. The anus evolved later and deuterostomy appeared as a precocious development of the mouth and the retention of the blastopore as anus. Drawings are not to scale.

To identify the developmental basis of different blastoporal fates, we strategically selected two brachiopod species, N. anomala and T. transversa (Supplementary Fig. 1a,j), on which previous experimental embryological work had been conducted22,23 (Supplementary Fig. 2). Brachiopods are marine, sessile, filter-feeding invertebrates that belong to the clade Spiralia3 that, together with the Ecdysozoa and Chaetognatha, form the Protostomia3. N. anomala and T. transversa share a comparable ecology and reproductive strategy (similar yolk content, developmental timing, lecithotrophy, indirect life cycle), but display deuterostomic22,24 and protostomic23 development, respectively (Fig. 1a). After fertilization, the embryos of these two brachiopod species undergo radial cleavage, gastrulation by invagination and form a planktonic larva that eventually metamorphoses into the adult22,23 (Supplementary Fig. 1). This exceptional case study allows us to exclude many developmental variables that might influence gastrulation (for example, yolk, egg size, cell number) and that often become significant when comparing blastoporal behaviours of embryos of distantly related and rapidly evolving lineages.


Blastoporal dynamics

Classic embryonic labelling and cutting experiments revealed the opposing blastoporal fates of N. anomala and T. transversa 22,23 (Supplementary Fig. 2). In the deuterostomic N. anomala, the blastoporal opening moves and closes ventroposteriorly (Fig. 1a; Supplementary Fig. 3a). During axial elongation, the ventral ectoderm is a compact, tightly packed cell layer, where cell proliferation (as seen by 5-ethynyl-2′-deoxyuridine (EdU) incorporation assay) is widespread (Supplementary Fig. 3a). In the protostomic T. transversa, the blastoporal opening elongates anteroposteriorly (Supplementary Fig. 3a). The blastoporal opening closes mid-posteriorly, apparently by convergence of the ventral ectoderm23, leaving the anterior end opened. Interestingly, increased EdU incorporation is observed in the anterior blastoporal rim during axial elongation in T. transversa, which is the region that internalizes first23,25 potentially causing the elongation of the blastoporal opening23 (Supplementary Fig. 3a). Importantly, our live-imaging recordings and cell-tracking analyses of gastrulation and axial elongation in N. anomala further exclude the possibility that cells from the blastoporal rim contribute to anterior/oral development (Supplementary Fig. 3b–d; Supplementary Videos 15). Blastopore closure appears to occur by proliferation and convergence of the blastoporal rim cells, and midline convergence does not appear to be a major morphological force at the ventral surface of N. anomala (Supplementary Fig. 3d; Supplementary Videos 3,4). Therefore, morphological and cell-tracking evidences (Supplementary Fig. 3; Supplementary Videos 15), together with previous classic embryological studies22,23 (Supplementary Fig. 2a–d) indicate a blastoporal-independent origin of the mouth in N. anomala, reinforcing the suitability of these two brachiopod species for studying the developmental basis of different blastoporal dynamics.

Axial and mesoderm development

To detect differences in the location and timing of the appearance of the primary embryonic fates in N. anomala and T. transversa that might underlie the different relationships of the blastoporal rim to the formation of adult structures, we compared the expression of evolutionarily conserved molecular markers associated with the development of anterior, posterior and endomesodermal regions (Fig. 2; Supplementary Fig. 4; Supplementary Table 2). Initially, both brachiopod embryos express anterior markers (six3/6, NK2.1, goosecoid (gsc), orthodenticle (otx)) at the animal pole (Fig. 2a), the embryonic region that forms anterior ectodermal structures22,23 (Supplementary Fig. 2). In the deuterostomic species N. anomala, these expression domains are always separate from the blastoporal rim, consistent with fate mapping, morphological and cell-tracking data (Supplementary Fig. 2c; Supplementary Fig. 3; Supplementary Videos 34). In the protostomic species T. transversa, however, otx and brachyury (bra) and later six3/6, NK2.1 and gsc are also expressed on the side of the blastoporal rim that contributes to the mouth after axial elongation occurs (green arrows in Fig. 2a; Supplementary Fig. 4a, b). The posterior marker even-skipped (evx) is initially expressed encircling the vegetal pole of both embryos, as is caudal (cdx) in T. transversa (Fig. 2b). N. anomala retains this expression of evx and expands the expression of cdx around the blastoporal rim during development. However, the expression of these genes becomes restricted to the posterior blastopore lip as gastrulation begins in T. transversa (Fig. 2b). Therefore, the protostomic embryo undergoes an anteroposterior molecular re-patterning of the blastoporal rim before axial elongation and body plan formation.

Figure 2: Gene expression during N. anomala and T. transversa embryogenesis.
Figure 2

ac, Whole-mount in situ hybridization of brachiopod embryos. a, In both brachiopods, six3/6, NK2.1, gsc and otx are expressed in an anterior ectodermal domain. six3/6 and otx also exhibit anterior endomesodermal expression. In T. transversa, otx at gastrula and six3/6, NK2.1 and gsc during early elongation are additionally expressed on the anterior/oral blastoporal rim. b, In N. anomala, evx is expressed around the gastral plate (more intense in the future posterior side), blastoporal opening and posterior end; cdx is expressed on the presumably posterior side of the gastral plate and thereafter as evx. Both genes are expressed around the gastral plate in T. transversa, but become restricted to the posterior blastoporal rim and posterior larval tip after gastrulation. cdx is also expressed in the posterior endoderm. c, twi is a general mesodermal marker in both brachiopods. In N. anomala, foxC and foxF are expressed in the ventral anterior mesoderm. In T. transversa, they are expressed in the anterior mesoderm lining the blastoporal opening, and later in more other putative mesodermal populations. In both brachiopods, foxC exhibits an anterior ectodermal domain. The animal/anterior pole (asterisk) is to the top. The insets are vegetal/ventral views, except for six3/6 and NK2.1 (blastula/gastrula; animal view) and foxC (blastula; animal view) of N. anomala. Upper insets of foxC and foxF in N. anomala are anterior views. Dashed circles in the insets indicate the closing blastoporal opening in N. anomala. Green arrows point towards mouth expression, orange arrowheads towards endo-/mesodermal expression, and black arrowheads to ectodermal domains. bp, blastoporal opening; gp, gastral plate; mo, mouth.

No major differences related to endoderm formation (foxA, GATA456 paralogues) are observed between either species (Supplementary Fig. 4a), but molecular differences coincident with the distinct mesodermal development reported for each brachiopod are evident (Fig. 2c; Supplementary Fig. 4a–c). While N. anomala internalizes the endomesoderm as a unit and later buds the mesoderm, which grows in a posterior-to-anterior direction22,24, T. transversa segregates the mesoderm during gastrulation and the mesoderm grows in an anterodorsal to posterior direction26. Accordingly, we first detect weak mesoderm expression (twist (twi)-expressing cells) in the archenteron wall after gastrulation in N. anomala, and anterior mesodermal markers (foxC, foxF) appear ventrally, spatially separate from the blastoporal opening (Fig. 2c). In contrast, mesoderm (twi-positive cells) is specified already at the blastula stage in T. transversa, with anterior mesodermal markers being expressed at the anterior blastoporal lip, on the side opposite of the domain that expresses posterior genes (Fig. 2c; Supplementary Fig. 4b), and ingresses earlier during axial elongation23. Thus, the anteroposterior re-patterning of the blastoporal rim together with the variation in mesoderm formation and patterning are the two developmental variables that co-vary with the differences in blastoporal fates in these two brachiopod species (Fig. 3; Supplementary Fig. 4d).

Figure 3: Lateral drawings showing simplified expression domains of the analysed markers shown in Fig. 2.
Figure 3

Only the ectodermal domain of anterior markers is drawn, the expression of endodermal and pan-mesodermal genes is shown just for the earliest stages, and only the anterior mesodermal domain of anterior mesodermal genes is shown in later stages The different green intensities in N. anomala blastula and gastrula drawings reflect the asymmetric expression of cdx and evx at these stages. bp, blastoporal opening; gp, gastral plate; mo, mouth.

Wnt pathway activity

Because endomesoderm specification and anteroposterior development are controlled by the canonical Wnt pathway in many previously studied animals21,27, we hypothesized that differential activity around the blastoporal rim between the two brachiopod species might explain the observed variation in the deployment of anterior genes and the molecular patterning of the mesoderm. We thus treated embryos of N. anomala and T. transversa at different developmental stages with 1-azakenpaullone (Azk) (Supplementary Fig. 5a), a selective inhibitor of GSK3-β28 that has been shown to stabilize β-catenin and overactivate the Wnt pathway in other systems. During cleavage, Azk treatment expands endodermal fates, inhibits ectodermal (both species) and mesodermal markers (T. transversa only), and abolishes gastrulation in both brachiopod species (Supplementary Fig. 5b, c), which is consistent with a conserved early role of β-catenin in endomesoderm specification21.

Azk treatment from blastula stages onwards in both N. anomala and T. transversa expands the expression of posterior genes (evx, cdx), and axin (axn), a read-out of the canonical Wnt pathway29, and causes the reduction of anterior markers (six3/6, NK2.1, gsc, otx, foxF), including the loss of the anterior apical lobe in both embryos, and also the mantle lobe in T. transversa (Fig. 4a–d; Supplementary Figs 6, 7a and 8). These data support a conserved role of the canonical Wnt pathway in inducing posterior fates27. As hypothesized, Azk treatment at the blastula stages inhibits the deployment of anterior genes (six3/6, NK2.1, gsc, bra) and an anterior mesodermal marker (foxF) around the blastoporal rim in the protostomic T. transversa, resulting in the extension of evx and cdx expression to the entire mesodermal and endodermal domains, respectively (Fig. 4c; Supplementary Figs 6a,c and 8a). Therefore, the canonical Wnt pathway influences the fate of different parts of the blastoporal rim by differentially regulating the axial patterning of the mesoderm and the expression of anterior/oral genes in T. transversa.

Figure 4: Effect of Azk on brachiopod embryos at the blastula stage.
Figure 4

a, Whole-mount in situ hybridization against axn. In both brachiopods axn is expressed in the gastral plate, stronger on the posterodorsal side of the blastoporal rim and in posterior territories. b, Analysis of Azk treated embryos of N. anomala. c, Analysis of Azk treated embryos of T. transversa fixed at late elongation. d, Analysis of Azk treated embryos of T. transversa fixed at larval stage. In bd, the first column shows ventral views of control and treated embryos labelled with phalloidin (grey) and Sytox Green (nuclei; yellow). The second column is a schematic representation (lateral view) of control and treated embryos. The other columns are whole-mount in situ hybridizations. Azk induces the loss of anterior structures, the reduction of anterior markers (six3/6, NK2.1, gsc, otx, foxF) and the expansion of posterior fates (axn, evx, cdx) in both brachiopods. In T. transversa, evx and cdx expands into meso- and endoderm, respectively, but not around the whole blastoporal rim, which continues to express otx on one side. In treated embryos, the blastoporal opening elongates and remains open opposite to the posterior tip of the embryo. The dashed line in d indicates probe trapping. The animal/anterior pole (asterisk) is to the top, and the insets are vegetal/ventral views, except in the treated embryo in c (lateral view). Green arrows point towards mouth expression, orange arrowheads towards endo-/mesodermal expression and black arrowheads to ectodermal domains. al, apical lobe; bp, blastoporal opening; en, endoderm; me, mesoderm; ml, mantle lobe; mo, mouth; mp, mesodermal pouches; pl, pedicle lobe. Scale bars, 50 μm.

Surprisingly, treatment with Azk at the blastula stage prevents neither the ectodermal expression of otx on one side of the blastoporal rim nor the restriction of evx and cdx to the opposite ectodermal side in T. transversa (Fig. 4c; Supplementary Figs 6a, 8a). However, the expression of otx in the anterior ectodermal blastoporal lip disappears in early Azk treated gastrulae, suggesting that otx has a late, Wnt-sensitive blastoporal rim expression in T. transversa (Supplementary Fig. 8c). Therefore, our findings indicate that signal(s) acting upstream and/or independently of the Wnt pathway at the blastula stage must trigger the primary asymmetric expression of otx and evx/cdx around the blastoporal rim of the protostomic T. transversa.

Bone morphogenetic protein (BMP) pathway.

We questioned whether these initial cues affecting the protostomic fate of the blastoporal opening in T. transversa could be related to the establishment of the dorsoventral axis, and in particular to BMP signalling, as the blastoporal rim expression of otx in T. transversa occurs in the same region of expression of the BMP antagonist chordin 30 (chd) (Fig. 5a; Supplementary Fig. 4b), and chd also expands in Azk-treated embryos (Supplementary Fig. 6c). Phosphorylation of SMAD1/5, the read-out of the BMP pathway30, is first detected on one side of the blastula, and later, on the dorsal ectodermal side of both brachiopod species (Fig. 5a). Accordingly, chd and the ventral marker netrin (ntr) are expressed first in the gastral plate, and on the ectodermal ventral side after gastrulation (Fig. 5a). Thus, no differences in the expression and activity of BMP signalling core elements are observed between the two brachiopod species.

Figure 5: Dorsoventral development and effect of DMH1 in brachiopod embryos.
Figure 5

a, The first line shows lateral and dorsal views of immunostaining against pSMAD1/5 (grey) and Sytox Green (nuclei; red). The two lines below are whole-mount in situ hybridizations. pSMAD1/5 appears first on one side of the blastulae, and later on the dorsal side of both brachiopod embryos (white arrowheads). Accordingly, chd and ntr are first expressed on the gastral plate (black arrowheads) and later on the ventral side of the embryo (black arrowheads). b, Analysis of DMH1 treated embryos of N. anomala. c, Analysis of DMH1 treated embryos of T. transversa. In b,c, the first column shows ventral views of embryos labelled with phalloidin (grey) and Sytox Green (nuclei; yellow). The second column is a schematic representation (lateral view) of control and treated embryos. The third column shows dorsal views (except in treated N. anomala; lateral view) of immunostaining against pSMAD1/5 (grey; the staining in treated embryos in the surface epidermis is background) with Sytox Green (nuclei; red). The other columns are whole-mount in situ hybridizations. DMH1 treatment causes the loss of pSMAD1/5, the expansion of ventral ectodermal markers (NK2.1, gsc) and anterior markers (six3/6, NK2.1, gsc, otx, foxF) and the reduction of posterior fates (evx, cdx). The impaired posterior development affects blastopore closure in both brachiopods. The animal/anterior pole (asterisk) is to the top, and the insets are vegetal/ventral views, except in b control pSMAD1/5 (dorsal); six3/6, NK2.1, gsc treated N. anomala (anterior views); and the upper inset in gsc treated T. transversa (anterior view). Orange arrowheads point to endo-/mesodermal expression and black arrowheads to ectodermal domains. al, apical lobe; bp, blastoporal opening; dg, gut; en, endoderm; em, endomesoderm; gl, gut lumen; gp, gastral plate; me, mesoderm; ml, mantle lobe; mo, mouth; pl, pedicle lobe. Scale bars, 50 μm.

To discriminate the role of the BMP signalling during early brachiopod development and its impact on the different blastoporal fates of N. anomala and T. transversa, we treated embryos with the drug dorsomorphin homologue 1 (DMH1), a selective inhibitor of SMAD1/5 phosphorylation31 and BMP activity (Supplementary Fig. 5a). In line with a conserved role of the BMP pathway in dorso­ventral axis specification30, DMH1 treatment induces the dorsal expansion of ventral ectodermal genes (chd, ntr) and ventrally expressed anterior ectodermal genes (NK2.1, gsc) in both brachiopod species (Fig. 5b,c; Supplementary Figs 7b and 9). However, DMH1 treatment does not eliminate the re-patterning of the blastoporal rim in T. transversa (Supplementary Fig. 9a), although anterior ectodermal and mesodermal markers (six3/6, NK2.1, gsc, otx, foxF, foxA) are expanded and posterior fates (evx, cdx) are reduced in both brachiopods (Fig. 5b,c; Supplementary Fig. 9). Thus, the early expression of otx and evx/cdx on opposing sides of the blastoporal rim in the protostomic T. transversa occurs independently of the BMP pathway, which excludes a role of the dorsoventral axis in regulating the different blastoporal fates of the two brachiopod embryos, and suggests that the anteroposterior repatterning of the vegetal pole in the protostomic species is regulated by earlier-acting signals. Because the impairment of dorsoventral development affects anteroposterior patterning (Fig. 5b,c; Supplementary Figs 7b and 9), and vice versa (Supplementary Fig. 6c), our findings additionally suggest interplay between the specification of the anteroposterior (canonical Wnt pathway) and dorsoventral (BMP pathway) axes to regulate cell fates and body elongation along the primary axis in brachiopod embryos, as also observed in other bilaterian embryos32,​33,​34, and in the specification of the oral–aboral and directive axes of cnidarians35,​36,​37.

Blastoporal diversity in Annelida

To test whether the developmental principles we observed in brachiopods may explain the variation in the fate of the blastoporal opening in another bilaterian lineage, we examined the development of the annelid Owenia fusiformis and compared its development with available data from other annelids (Supplementary Fig. 10a; Supplementary Table 9). O. fusiformis is a member of the Oweniidae, the potential sister group of all remaining Annelida38, whose members usually exhibit a highly stereotypical cleavage program referred to as quartet spiral cleavage39. In the closely related species O. collaris 40, the blastoporal fate has been described as being deuterostomic. Our morphological analysis demonstrates that O. fusiformis forms a vegetal, round blastoporal opening during gastrulation, which moves only slightly along the animal–vegetal axis to end up in the anterior mouth. Therefore, O. fusiformis shows protostomous formation of the mouth (Supplementary Fig. 10d–h). Further analyses will be required to identify the exact origin of the anus.

We predicted based on our brachiopod data that anterior and posterior genes would be deployed in different sides of the blastoporal rim during gastrulation in the protostomic O. fusiformis, and that the way mesoderm develops may relate to the behaviour of the blastoporal opening. The anterior ectodermal genes six3/6, NK2.1, gsc, otx, the foregut marker foxA and the posterior endo- and ectodermal genes evx and cdx are expressed on opposing sides of the blastoporal rim before axial elongation (Fig. 6a,b; Supplementary Fig. 11a). Mesodermal markers (twi, bra, foxC, foxF and FoxL1) suggest two distinct mesodermal populations (one anterior to the blastoporal rim and one posterior), which apparently expand as the embryo elongates anteroposteriorly (Fig. 6c). The short growth of the paired posterior mesodermal bands40 (Supplementary Fig. 10f,g), probably related to a delayed formation of the definitive trunk in O. fusiformis 41, may explain the short axial elongation of the embryo, and thus the slight anterior move of the blastoporal opening (Supplementary Fig. 10e–g).

Figure 6: Gene expression during development of O. fusiformis.
Figure 6

Whole-mount in situ hybridization. a, six3/6, NK2.1, gsc and otx are expressed at the anterior blastoporal lip and in the mouth of the mitraria larva. six3/6 is also expressed in the animal pole, NK2.1 transiently in an animal/anterior domain during gastrulation and otx in the ciliated bands. b, evx and cdx are expressed on the posterior side of the gastrula, first in the posterior endoderm, and then evx in the posterior larval ectoderm as elongation begins. cdx is also expressed in the hindgut and anus. c, Mesodermal genes are not detected during gastrulation. twi is expressed in two lateral bands extending from posterior to anterior, and in one antero-dorsal cell. bra is detected in the mesodermal cells that putatively form the trunk in the mitraria. foxC and foxF are expressed in the most anterior-lateral mesodermal cells and in a few cells in the anterior dorsal side of the embryo. In the mitraria, foxC and foxF are expressed in the posterior pharyngeal mesoderm. foxL1 is expressed in the anterior portion of the mesodermal bands and in the lateral–posterior mesoderm of the pharynx. Mesodermal expression was assigned based on embryo and larval morphology (Supplementary Fig. 10) and previous studies40,41. In all panels, the animal/anterior pole (asterisk) is to the top and the ventral side is to the left-bottom in gastrula and larval stages. Insets are vegetal/ventral views, except in blastula stages (animal views). Green arrows point towards mouth expression, orange arrowheads towards endo-/mesodermal expression and black arrowheads to ectodermal domains. an, anus; bp, blastoporal opening; gp, gastral plate; mo, mouth.


By comparing two brachiopod species with opposing blastoporal fates, our findings help to illuminate the developmental basis for the recurrent evolution of deuterostomy and protostomy (Fig. 7). Although the nature of the primary developmental input(s) remains elusive, our results indicate that early signals repress Wnt activity on the ventral ectodermal side of the blastoporal rim in the protostomic T. transversa (Fig. 7). This allows the expression of anterior ecto- and mesodermal genes in the ventral blastopore lip, which together with distinct dynamics of mesodermal morphogenesis22,​23,​24,26, correlate with a behaviour of the blastoporal opening different from that of N. anomala (Fig. 7; Supplementary Animation). Therefore, the different blastoporal fates observed in N. anomala and T. transversa may be caused by the differences in the modes and timing of axis specification and mesoderm development. Remarkably, our findings do not support a premature development of the mouth in N. anomala (Fig. 2a), as the planuloid-acoeloid scenario would imply15,​16,​17,​18. Although the amphistomy concept7,​8,​9,​10,​11,​12,​13 could conceptually explain the different blastoporal fates of T. transversa and N. anomala, our data do not show the proposed similar, but inverted, dynamics of blastopore closure between the two brachiopods (Supplementary Fig. 3; Supplementary Videos 15). Therefore, our study dissents on a mechanistic level from current scenarios for the evolution of deuterostomy and protostomy (Fig. 1c).

Figure 7: The developmental basis of different blastoporal fates in N. anomala and T. transversa.
Figure 7

Proposed model of the developmental control of fate of the blastoporal opening (green circle) in the studied brachiopods. N. anomala specifies only posterior mesodermal fates around the blastoporal rim, which resultsin the coordinated movement of this opening to the posterior larval end. In the protostomic T. transversa, inferred early developmental inputs (Fx, in red) seem to restrict the activity of the Wnt pathway to the future posterior blastoporal lip, releasing the opposite side to deploy anterior ecto- and mesodermal fates. During axial growth, the blastoporal opening elongates and becomes co-opted into the mouth. The striped mesodermal areas in the early gastrula of N. anomala are prospective fates. The animal/anterior pole (asterisk) is to the top. Drawings are not to scale. al, apical lobe; BMP, bone morphogenic protein pathway; bp, blastoporal opening; mo, mouth; pl, pedicle lobe.

We further demonstrate that comparable developmental events to those observed in the protostomic brachiopod T. transversa may act in the protostomic annelid O. fusiformis (Fig. 6). Our data on this annelid are consistent with the reported expression of some of these genes in other protostomic polychaetes (Supplementary Table 9), but differ from Capitella teleta, whose blastoporal opening closes completely, and whose mouth forms later at a position anterior to the site of blastopore closure42. As it is also observed in the deuterostomous brachiopod N. anomala, the annelid C. teleta does not exhibit an overt anteroposterior patterning of the blastoporal rim43,​44,​45 (Supplementary Table 9). Additionally, C. teleta displays a distinct mode of axial and precocious mesodermal development46,47 with respect to other studied annelids and spiral-cleaving embryos39 (Supplementary Table 9). Therefore, variation in axial and mesoderm development also correlates with modifications in the behaviour of the blastoporal opening in annelids, which solidifies the conclusions derived from the observations on brachiopod embryos. Our results together indicate that changes in the establishment of the axial identities and fate maps alone can explain the vast diversity of blastoporal behaviours observed in Spiralia.

Although the formation of a blastoporal opening during gastrulation is probably a homologous feature of cnidarian and bilaterian embryogenesis21,48, its cellular composition, shape and later destination is dynamic during development and in evolution. Our study shows that distinct behaviours of the blastoporal opening appear to be influenced by changes in the fates of the cells that move over the blastoporal rim on their way to their final embryonic locations. Consequently, we propose that the shape and behaviour of the blastoporal opening is a secondary effect of the embryonic architecture during gastrulation and axial elongation—commonly referred to as a ‘spandrel’ in evolutionary biology49. The co-option of the blastoporal opening—if still present—to a deuterostomic or protostomic fate in specific animal groups would thus correspond to secondary adaptations of this transient opening for the development of a digestive opening (the anus and the mouth, respectively)20 that have occurred independently multiple times during evolution, as the distribution of these characters in bilaterian phylogeny reveals (Fig. 1b). This scenario challenges the assumed value of extant blastoporal behaviours for explaining the evolutionary origin and diversification of Bilateria that has been presumed for over 100 years4,​5,​6,​7,9,​10,​11. Freeing the constraint that the mouth and anus have a necessary association with the embryonic blastopore will help in understanding the developmental events underlying the evolution of an alimentary canal5,20,21,50, and ultimately the appearance and diversification of bilaterian body plans.


Obtaining animals and embryos

Gravid adults were collected from the coast near Friday Harbor Laboratories, USA (T. transversa), Espeland Marine Biological Station, Norway (N. anomala) and Station Biologique de Roscoff, France (O. fusiformis) during their reproductive season (T. transversa: January; N. anomala: September; O. fusiformis: June and July), and spawned as previously described22,23,40. Embryos were collected at various stages of development up to the late larval stage.

Drug treatments

Embryos of N. anomala and T. transversa were treated with either 0.5–10 μM Azk or 1–10 μM DMH1 diluted in seawater at different developmental stages (see Supplementary Fig. 5a for a detailed experimental set-up). Solutions were changed every 24 h, and control conditions were performed with 0.1–1% dimethylsulfoxide. Treatments that started after fertilization were initiated on two-cell-stage embryos (approximately 2–3 h after sperm addition in both brachiopod species), to ensure correct fertilization of the oocytes. Treatments on early blastula stages of T. transversa were performed at approx. 11 h postfertilization, when the embryos were hollow spheres of blastomeres. Treatments on blastula stages (approx. 20 h postfertilization in both brachiopod species) were conducted on mature blastulae, characterized by the presence of columnar cells with cilia, and the display of spinning/swimming behaviour. Treatments on gastrula stages were initiated as soon as the first signs of gastral plate invagination were evident (approx. 24 h postfertilization in T. transversa). Once control embryos reached the desired developmental stage, control and treated embryos were fixed in 4% paraformaldehyde for 1 h at room temperature, washed several times with 0.1% Tween-20 phosphate-buffered saline (PTw), and stored in methanol at −20 °C (for gene expression analyses and phospho-SMAD1/5 antibody labelling) or PTw at 4 °C (for actin labelling).

EdU labelling

Embryos of N. anomala and T. transversa at the required developmental stage were incubated in 100 μM EdU diluted in sea water for 20 min and subsequently fixed in 4% paraformaldehyde in sea water for 1 h at 4 °C. Identification of EdU-labelled nuclei was performed following manufacturer recommendations (Life Technologies). Embryos and larvae were cleared in benzyl benzoate/benzyl alcohol (2:1) before being scanned in a Leica SP5 confocal laser-scanning microscope. Image stacks were analysed with Fiji and Photoshop CS6 (Adobe). Brightness/contrast and colour balance adjustments were applied to the whole image, not parts.

Whole-mount in situ hybridization

Genes were identified from RNAseq data and gene orthologies were inferred by maximum likelihood analyses51 (Supplementary Fig. 12). Single colorimetric in situ hybridization of brachiopod and annelid embryos and larvae were performed following an established protocol52. Stained embryos and larvae were cleared in 70% glycerol and imaged with a Zeiss Axiocam HRc connected to a Zeiss Axioscope Ax10 using bright-field Nomarski optics. Double-fluorescent whole-mount in situ hybridization of T. transversa early gastrula was performed until first antibody incubation following the same protocol as for single colorimetric in situ hybridization. After the blocking step, samples were first incubated overnight at 4 °C with an antibody anti-DNP (dinitrophenyl) POD (peroxidase)-conjugated (Perkin Elmer) diluted 1:100 in a blocking solution. Samples were then washed in PTw and developed with TSA-Cy3 following the manufacturer’s recommendations (Perkin Elmer). Before developing the second probe, remaining POD activity from the antibody anti-DNP was quenched by incubating embryos with 1% oxygen peroxide diluted in PTw for 1 h at room temperature, followed by 10 min incubation at 67 °C in 50% formamide, 2 × sodium salt citrate, 1% sodium dodecyl sulfate and 0.1% Tween-20. After POD inactivation, samples were blocked in a blocking solution for 1 h and incubated with antibody anti-DIG (digoxigenin) POD conjugated diluted 1:100 in blocking solution overnight at 4 °C. After antibody washes with PTw, signal detection was performed with TSA-Cy5 as recommended by the manufacturer (Perkin Elmer). Fluorescently labelled embryos were cleared in benzyl benzoate/benzyl alcohol (2:1) before being scanned in a Leica SP5 confocal laser-scanning microscope. Image stacks were analysed with Fiji and Photoshop CS6 (Adobe). Brightness/contrast and colour balance adjustments were always applied to the whole image, not parts. Detailed in situ hybridization protocols are available upon request.

Actin labelling and immunohistochemistry

Brachiopod and annelid embryos were incubated with either BODIPY FL or Alexa647-conjugated phalloidin/phallacidin (Life Technologies) and Sytox Green or DAPI (4’,6-diamidino-2-phenylindole) (Life Technologies) diluted in 1% bovine serum albumin (BSA) in 0.1% Triton X-100 phosphate-buffered saline for 1 h at room temperature to detect actin filaments and nuclei. Phosphorylated SMAD1/5 (pSMAD1/5) was detected on N. anomala and T. transversa embryos fixed and stored as for gene expression analyses. Before staining, embryos were gradually rehydrated to PTw, permeabilized in 0.2% Tween-20, 0.2% Triton X-100 phosphate-buffered saline (PTwTx) and blocked in 1% BSA PTwTx for 1 h. Samples were incubated with antibody anti-pSMAD1/5 (Cell Signaling; ref. 9516) diluted 1:50 in 5% normal goat serum in PTwTx overnight at 4 °C, washed in 1% BSA PTwTx for 4 h, and the signal detected with an anti-rabbit Alexa647-conjugated antibody (Life Technologies) diluted 1:250 in 5% normal goat serum in PTwTx. In all cases embryos and larvae were cleared in benzyl benzoate/benzyl alcohol (2:1) before being scanned in a Leica SP5 confocal laser-scanning microscope. Image stacks were analysed with Fiji and Photoshop CS6 (Adobe). Brightness/contrast and colour balance adjustments were applied to the whole image, not parts.

4-Dimensional microscopy

N. anomala embryos at the desired developmental stage were mounted under a 22 × 22 mm coverslide with clay feet on each corner. Gentle pressure was applied to the coverslide to immobilize the embryos, while still leaving enough space to develop normally. A Zeiss Ax10 Imager.M2 microscope with an internal focus drive was used to move the temperature-controlled stage to record the z series. Pictures were captured with a SensiCam camera (PCO.Imaging), and compressed tenfold with a wavelet function (Lurawave). All recordings were performed at 12 °C with a ×40 lens. Raw data is available upon request. The embryos recorded during blastula stage (Supplementary Video 1) and blastopore closure (Supplementary Videos 4 and 5) were then removed from the microscope slide and placed in clean seawater to allow them recover. They were then fixed in 4% paraformaldehyde as described above, and their morphology assessed by actin and nuclear labelling (Supplementary Fig. 3c). The embryo recorded during gastrulation (Supplementary Video 2) failed to proceed to normal axial elongation due to the mounting position. The embryo recorded during axial elongation (Supplementary Video 3) swam away during the recording and could not be recovered. Recordings were analysed as described elsewhere53, using SIMIºBioCell software (SIMI, Germany). Time-lapse images were assembled into video recordings using Fiji, iMovie (Apple) and Photoshop CS6 (Adobe). The number of frames taken, the time between frames and the number of focal levels per frame for each recording were as follows:

  • Supplementary Video 1: 365 frames; 180 s between frames; 30 focal levels; 2 μm increment. Total recorded time: 18 h and 15 min of development.

  • Supplementary Video 2: 429 frames; 180 s between frames; 25 focal levels; 1.7 μm increment. Total recorded time: 21 h and 27 min of development.

  • Supplementary Video 3: 266 frames; 180 s between frames; 30 focal levels; 1.3 μm increment. Total recorded time: 13 h and 18 min of development.

  • Supplementary Video 4: 800 frames; 180 s between frames; 25 focal levels; 1.8 μm increment. Total recorded time: 40 h of development.

  • Supplementary Video 5: 373 frames; 180 s between frames 1 and 239, and 360 s between frames 361 and 373; 30 focal levels; 1.1 μm increment. Total recorded time: 25 h and 21 min of development.

Data availability

T. transversa, N. anomala and O. fusiformis sequence data have been deposited in GenBank with the primary accession numbers KF946061 to KF946084 and KR232531 to KR232552). The original image stacks generated during 4D microscopy are available from the authors on reasonable request.

Additional information

How to cite this article: Martín-Durán, J. M., Passamaneck, Y. J., Martindale, M. Q. & Hejnol, A. The developmental basis for the recurrent evolution of deuterostomy and protostomy. Nat. Ecol. Evol. 1, 0005 (2016).


  1. 1.

    & Embryology, Constructing the Organism (Sinauer Associates, 1997).

  2. 2.

    Die systematische Einteilung des Tierreichs. Verh. Zool. Bot. Ges. Wien 58, 491–511 (1908).

  3. 3.

    , , & Animal phylogeny and its evolutionary implications. Annu. Rev. Ecol. Evol. Syst. 45, 371–395 (2014).

  4. 4.

    et al. Xenacoelomorpha is the sister group to Nephrozoa. Nature 530, 89–93 (2016).

  5. 5.

    , , , & Deuterostomic development in the protostome Priapulus caudatus . Curr. Biol. 22, 2161–2166 (2012).

  6. 6.

    & in Animal Evolution: Genes, Genomes, Fossils and Trees (eds Telford M. J.& Littlewood D. T. J.) 33–40 (Oxford Univ. Press, 2009).

  7. 7.

    On the origin of metameric segmentation and some other morphological questions. Q. J. Microsc. Sci. 24, 43–82 (1884).

  8. 8.

    On the early phylogeny of the Metazoa: the bilatero-gastrea theory. Zool. Bidr. Uppsala 30, 321–354 (1955).

  9. 9.

    Die entstehung der metamerie der wirbellosen. Zool. Anz. 14, 18–23 (1950).

  10. 10.

    & Dorsal or ventral: similarities in fate maps and gastrulation patterns in annelids, arthropods and chordates. Mech. Dev. 61, 7–21 (1997).

  11. 11.

    , , , & Polychaete trunk neuroectoderm converges and extends by mediolateral cell intercalation. Proc. Natl Acad. Sci. USA 104, 2727–2732 (2007).

  12. 12.

    , & Evolution of the bilaterian larval foregut. Nature 409, 81–85 (2001).

  13. 13.

    Evolution of deuterostomy – and origin of the chordates. Biol. Rev. Camb. Philos. Soc. (2015).

  14. 14.

    New ideas on the origin of bilateral animals. Russ. J. Mar. Biol. 30, S22–S33 (2004).

  15. 15.

    Die Organisation der Turbellaria Acoela (von Wilhelm Engelmann, 1891).

  16. 16.

    The Invertebrates. Vol II: Platyhelminthes and Rhynchocoela. (McGraw-Hill, 1951).

  17. 17.

    On the origin and evolution of the lower Metazoa. Zeitschr. Zool. Syst. Evol.-Forsch. 16, 40–88 (1978).

  18. 18.

    Principles of Comparative Anatomy of Invertebrates (Univ. Chicago Press, 1969).

  19. 19.

    Phylogenetischer status und bedeutung der mesenchymaten Bilateria. Zool. Jahrb. Anat. 103, 354–373 (1980).

  20. 20.

    Notes on the embryology and classification of the animal kingdom: comprising a revision of speculations relative to the origin and significance of the germ-layers. Q. J. Microsc. Sci. s2-s17, 399–454 (1877).

  21. 21.

    & A developmental perspective: changes in the position of the blastopore during bilaterian evolution. Dev. Cell 17, 162–174 (2009).

  22. 22.

    Regional specification during embryogenesis in the craniiform brachiopod Crania anomala. Dev. Biol. 227, 219–238 (2000).

  23. 23.

    Regional specification during embryogenesis in the articulate brachiopod Terebratalia. Dev. Biol. 160, 196–213 (1993).

  24. 24.

    The development of the brachiopod Crania (Neocrania) anomala (O. F. Müller) and its phylogenetic significance. Acta Zool. 72, 7–28 (1991).

  25. 25.

    Regional specification during embryogenesis in Rhynchonelliform brachiopods. Dev. Biol. 261, 268–287 (2003).

  26. 26.

    & in Reproduction of Marine Invertebrates (eds Giese, A. C., Pearse, J. S. & Pearse V. B. ) 47–84 (Boxwood Press, 1991).

  27. 27.

    & Wnt signaling and the polarity of the primary body axis. Cell 139, 1056–1068 (2009).

  28. 28.

    , , , & 1-Azakenpaullone is a selective inhibitor of glycogen synthase kinase-3 beta. Bioorg. Med. Chem. Lett. 14, 413–416 (2004).

  29. 29.

    et al. Wnt/beta-catenin/Tcf signaling induces the transcription of Axin2, a negative regulator of the signaling pathway. Mol. Cell. Biol. 22, 1172–1183 (2002).

  30. 30.

    Evo–devo: variations on ancestral themes. Cell 132, 185–195 (2008).

  31. 31.

    et al. In vivo structure–activity relationship study of dorsomorphin analogues identifies selective VEGF and BMP inhibitors. ACS Chem. Biol. 5, 245–253 (2010).

  32. 32.

    et al. Integrating patterning signals: Wnt/GSK3 regulates the duration of the BMP/Smad1 signal. Cell 131, 980–993 (2007).

  33. 33.

    & Anteroposterior and dorsoventral patterning are coordinated by an identical patterning clock. Development 140, 1970–1980 (2013).

  34. 34.

    , , & Axial patterning interactions in the sea urchin embryo: suppression of nodal by Wnt1 signaling. Development 139, 1662–1669 (2012).

  35. 35.

    et al. Axis patterning by BMPs: cnidarian network reveals evolutionary constraints. Cell Rep. 10, 1646–1654 (2015).

  36. 36.

    & RGM regulates BMP-mediated secondary axis formation in the sea anemone Nematostella vectensis . Cell Rep. 9, 1921–1930 (2014).

  37. 37.

    , , & Pre-bilaterian origin of the blastoporal axial organizer. Nat. Commun. 7, 11694 (2016).

  38. 38.

    et al. Illuminating the base of the annelid tree using transcriptomics. Mol. Biol. Evol. 31, 1391–1401 (2014).

  39. 39.

    Variation in spiralian development: insights from polychaetes. Int. J. Dev. Biol. 58, 457–467 (2014).

  40. 40.

    & Unusual development of the mitraria larva in the polychaete Owenia collaris . Biol. Bull. 217, 253–268 (2009).

  41. 41.

    On the mitraria larva of Owenia fusiformis Delle Chiaje. Phil. Trans. R. Soc. Lond. B 221, 231–334 (1932).

  42. 42.

    Zur entwicklungsgeschichte der Capitelliden. Mitt. Aus Der Zool. Station Zu Neapel 13, 1–292 (1899).

  43. 43.

    , , & Expression of the pair-rule gene homologs runt, Pax3/7, even-skipped-1 and even-skipped-2 during larval and juvenile development of the polychaete annelid Capitella teleta does not support a role in segmentation. EvoDevo 3, 8 (2012).

  44. 44.

    , & Molecular conservation of metazoan gut formation: evidence from expression of endomesoderm genes in Capitella teleta (Annelida). EvoDevo 5, 39 (2014).

  45. 45.

    & ParaHox gene expression in the polychaete annelid Capitella sp. I. Dev. Genes Evol. 216, 81–88 (2006).

  46. 46.

    , & An organizing activity is required for head patterning and cell fate specification in the polychaete annelid Capitella teleta: new insights into cell-cell signaling in Lophotrochozoa. Dev. Biol. 379, 107–122 (2013).

  47. 47.

    , , & A comprehensive fate map by intracellular injection of identified blastomeres in the marine polychaete Capitella teleta . EvoDevo 1, 8 (2010).

  48. 48.

    , , & The blastoporal organiser of a sea anemone. Curr. Biol. 17, R874–R876 (2007).

  49. 49.

    & The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc. R. Soc. Lond. B 205, 581–598 (1979).

  50. 50.

    et al. Evolutionary modification of mouth position in deuterostomes. Semin. Cell Dev. Biol. 18, 502–511 (2007).

  51. 51.

    RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).

  52. 52.

    & Acoel development indicates the independent evolution of the bilaterian mouth and anus. Nature 456, 382–386 (2008).

  53. 53.

    & What a couple of dimensions can do for you: Comparative developmental studies using 4D microscopy—examples from tardigrade development. Integr. Comp. Biol. 46, 151–161 (2006).

Download references


We thank H. Hausen and O. Voecking for sharing the RNAseq data of O. fusiformis and expertise with the spawnings, B. C. Vellutini for help with collections and drug treatments, and G. S. Richards, F. Rentzsch, M. Iglesias and the members of the Hejnol laboratory for their comments on the manuscript. We also thank the staff at Friday Harbor Laboratories, Espeland Marine Biological Station and Station Biologique de Roscoff for assistance with animal collections. The study was funded by the core budget of the Sars Centre and supported by The European Research Council Community’s Framework Program Horizon 2020 (2014–2020) ERC grant agreement 648861 and an L. Meltzers Høyskolefond grant to A.H. J.M.M.-D. was supported by Marie Curie fellowship IEF 329024.

Author information


  1. Sars International Centre for Marine Molecular Biology, University of Bergen, Thormøhlensgate 55, Bergen 5006, Norway

    • José M. Martín-Durán
    •  & Andreas Hejnol
  2. The Whitney Laboratory for Marine Bioscience, University of Florida, 9505 Ocean Shore Boulevard, St Augustine, Florida 32080, USA

    • Yale J. Passamaneck
    •  & Mark Q. Martindale
  3. Kewalo Marine Laboratory, PBRC, University of Hawaii, 41 Ahui Street, Honolulu, Hawaii 96813, USA.

    • Yale J. Passamaneck
    •  & Mark Q. Martindale


  1. Search for José M. Martín-Durán in:

  2. Search for Yale J. Passamaneck in:

  3. Search for Mark Q. Martindale in:

  4. Search for Andreas Hejnol in:


J.M.M.-D. and A.H. conceived the project. J.M.M.-D., Y.J.P. and A.H. performed animal collections and cloned genes, J.M.M.-D conducted the experiments, and J.M.M.-D. and A.H. performed the four-dimensional recordings. Y.J.P. carried out the EdU analysis in T. transversa. J.M.M.-D. and A.H. analysed the data and wrote the manuscript, and Y.J.P. and M.Q.M. edited the paper. All authors discussed and commented on the data.

Competing interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to José M. Martín-Durán or Andreas Hejnol.

Supplementary information

PDF files

  1. 1.

    Supplementary information

    Supplementary figures 1–12; Supplementary Tables 1–9; Supplementary References and legends for Supplementary Videos

HTML files

  1. 1.

    Supplementary Animation

    Animated summary of the main findings of the manuscript regarding brachiopod gastrulation.


  1. 1.

    Supplementary Video 1

    Description: Time-lapse recording of an embryo of N. anomala from the 5 early blastula stage to gastrulation, viewed from the animal hemisphere.

  2. 2.

    Supplementary Video 2

    Time-lapse recording of N. anomala from the early gastrula stage to the onset of axial elongation (shift of the blastopore to a ventral-posterior position), viewed from the vegetal pole.

  3. 3.

    Supplementary Video 3

    Time-lapse recording of N. anomala during early axial elongation, viewed from the ventral side.

  4. 4.

    Supplementary Video 4

    Time-lapse recording of N. anomala during axial elongation and blastopore closure, viewed from the ventral side.

  5. 5.

    Supplementary Video 5

    Time-lapse recording of N. anomala during late axial elongation and early larva differentiation (apical lobe-mantle lobe boundary formation; closure of the blastopore), viewed from the ventral side.