A dynamic regulatory network explains ParaHox gene control of gut patterning in the sea urchin Free

Epithelial cell sheet formation has been a crucial step in the evolution of multicellularity. In particular, the development of an internalized gastrointestinal system contributed to the increase of animal morphological complexity conferring the capacity to perform extracellular digestion and releasing organisms from body size constrains, which very likely allowed the development of specialized structures. Accordingly, the molecular system orchestrating digestive tube development can arguably be considered to be one of the first evolved developmental patterning systems and may involve genes that were later co-opted for the development and patterning of new body parts (Nielsen, 2008; Roberts, 2000; Wolpert, 1994), thus providing a model with which to investigate widely employed developmental processes such as differentiation, morphogenesis and axis formation.

The embryonic gut of most deuterostomes forms through common developmental processes during gastrulation: endodermal cells are internalized by invagination through the blastopore of sea urchin and frog embryos and, in a very similar way, ingress with combined involution and migration movements through the primitive streak of the avian and mammalian epiblast. In all cases, the first endodermal cells to exit either the blastopore or the primitive streak will contribute to the anterior definitive endoderm, whereas the later moving cells will contribute to the posterior endoderm. After gastrulation, the primitive gut of most animals seems a tube with no apparent differentiation at the morphological level. However, several distinct or partially overlapping transcription factor and signaling molecule expression territories can be detected along the whole length of the archenteron (Jacobs et al., 2012; Lengyel and Iwaki, 2002; McGhee, 2007; Sherwood et al., 2011; van den Brink, 2007; Zorn and Wells, 2009). The interaction between molecular regulators leads, eventually, to the establishment of fine boundaries of gene expression subdividing the gut into distinct domains, precursors of the different digestive organs.

Two ParaHox transcription factors, Xlox and Cdx, have been proposed as central actors in the gut development in several animals (Brooke et al., 1998; Holland, 2013). Cdx is required for posterior gut development in many cases examined, and Xlox is crucial for more anterior gut domains and pancreas formation. In addition, a number of signaling molecules have been shown to regulate different aspects of vertebrate gut formation (Feng et al., 2012; Jacobs et al., 2012; Spence et al., 2011). However, characterizing the molecular interactions responsible for gut patterning has proved to be a difficult task in vertebrate models such as mouse or chick, in which the generation of transgenic lines is often required and tandem genome duplications occurred (Donoghue and Purnell, 2005).

Most sea urchins are indirect developers and go through a bilateral swimming (and feeding) pelagic larval stage (the pluteus), before metamorphosis leads to the formation of the adult form with its characteristic pentameric body plan. The pluteus has a functional tripartite gut composed of a muscular esophagus, a large spherical stomach and a small tubular intestine (Burke, 1981). A cardiac sphincter between the esophagus and the stomach, and a pyloric sphincter between the stomach and the intestine separate the gut compartments. The small number of cells and the detailed description of cell movements during gastrulation (McClay, 2011), the extensive molecular understanding of the early endoderm specification mechanisms (Croce and McClay, 2010; Davidson et al., 2002; Peter and Davidson, 2010) and the reduced number of gene duplications compared with vertebrates, all make the sea urchin embryo an excellent but so far largely unexploited model to investigate the molecular dynamics of gut development.

ParaHox gene involvement in the sea urchin gut patterning has been already demonstrated (Arnone et al., 2006; Cole et al., 2009). Although Xlox is required for the formation of the pyloric sphincter, Cdx is dedicated to setting up the posterior border of Xlox expression within the intestine.

In this study, we combine the use of functional analysis methodologies with high resolution imaging, RNAseq and qPCR to show that sea urchin larval gut patterning is achieved through the integration of autonomous and conditional mechanisms of cell fate specification. First, we demonstrate that the early active endodermal transcription factors Hox11/13b, FoxA, Bra and Blimp1a (Arenas-Mena et al., 2006; Livi and Davidson, 2006; Oliveri et al., 2006; Rast et al., 2002) synergistically regulate ParaHox gene expression. Then, we uncover a Wnt10 signaling event mediating Cdx-controlled repression of Xlox transcription in posterior gut cells, and provide evidence of a positive auto-regulatory input of Cdx on its own transcription. Moreover, we show that Xlox is required for the activation of stomach terminal differentiation genes and also for the transcription of the muscle-specific myosin heavy chain gene in the pyloric sphincter cells. Finally, we identify a number of additional developmental processes potentially controlled by Xlox and we do that through a genome-wide comparative transcriptome analysis on Xlox morphant gastrulae and plutei. From all the above studies, Xlox clearly emerges as a crucial player of cell differentiation in the sea urchin embryo, whereas Cdx acts as major regulator of intestinal cell identity.

To our knowledge this is the first study, in a non-chordate deuterostome (with the experimental advantages that this implies), to have investigated the gut patterning process at the functional level and offers a great contribution to our general understanding of the molecular mechanisms involved in the development of a complex and highly organized structure such as the embryonic digestive apparatus.

The first step in the study of the functional regulation of cell differentiation is the detailed analysis in time and space of the actively transcribed factors putatively involved in the process. Since late gastrula stage, the archenteron of the embryo can be subdivided in three regions, foregut, midgut and hindgut that will, respectively, form esophagus, stomach and intestine of the pluteus larva. We analyzed Xlox and Cdx protein localization and gut gene expression relative to Xlox and Cdx transcript localization, from the post gastrular embryo to the pluteus larval stages, including active transcription factors and terminal differentiation genes (Fig. 1).

The expression dynamics of SpLox and SpCdx proteins have been analyzed by immunolocalization, using specific antibodies developed in our laboratory, and the protein versus mRNA spatial localization in the embryos and larvae have been compared. Xlox protein localization at late gastrula stage matched the transcript pattern, being detected in around 30 cells of the posterior gut (Fig. 1A,D,N,Q). However, whereas Xlox transcripts are strictly confined to around 10 cells of the larval pyloric sphincter, we detected the protein in about 30 cells covering the pyloric sphincter and part of the stomach posterior hemisphere, mostly on the aboral side (Fig. 1B,E,O,R). We detected Xlox protein in the stomach domain until the 3-week larval stage. We propose that, late in gastrulation, the most anterior cells expressing Xlox mRNA stop transcribing the gene but retain the protein and develop to form part of the larval stomach. Cdx protein localization exactly matched the corresponding intestinal mRNA expression at all stages (Fig. 1C-I,U,V).

We then analyzed Cdx mRNA spatial expression relative to Xlox, Hox11/13b and Bra at both late gastrula (48 h) and pluteus stages. Experiments at earlier gastrula stage (40-44 h), demonstrate that the first cells expressing Cdx also express Xlox, Hox11/13b and Bra (data not shown), thus suggesting that Cdx, Xlox, Hox11/13b and Bra together define a specific cellular regulatory state, exclusive to this most posterior part of the late gastrula archenteron, and hinting to a possible regulatory interaction among them. Moreover, considering the fact that Xlox, Hox11/13b and Bra expression in those cells starts before Cdx transcription activation (Arenas-Mena et al., 2006; Arnone et al., 2006; Rast et al., 2002), we contemplated and investigated a possible role for Xlox, Hox11/13b and Bra in Cdx regulation. At the late gastrula stage, the domain of Cdx expression expands, overlapping anteriorly only with Xlox and posteriorly only with Hox11/13b and Bra (Fig. 1D-I). The expression domains of Cdx, Xlox, Hox11/13b and Bra in the pluteus show a different relative pattern reflecting the significant changes occurred in the morphology of the gut during the transition from late gastrula to pluteus stages: Cdx is now expressed in the whole intestine, Xlox is transcribed only in the pyloric sphincter cells, and Hox11/13b and Bra in the most posterior cells of the intestine. Although no sign of co-expression is observable at pluteus stage between Xlox and Cdx (Fig. 1E), a partial overlapping between Cdx and Hox11/13b, and between Cdx and Bra is still detectable in the posterior cells of the gut (see arrowheads in Fig. 1G,I).

A very remarkable discovery that we present in this work is the finding of a Wnt10 signal in the posterior gut cells of the embryo. We found that Wnt10 transcription starts a few hours after the activation of Cdx, and is localized in the posterior cells of the gut at late gastrula (Fig. 1J), prism (Fig. 1K) and pluteus stages (Fig. 1L,M). The cells expressing Wnt10 apparently represent a subset of the posterior-most Cdx positive cells. Interestingly, from the moment these cells start expressing Wnt10, Xlox transcripts become undetectable in them (Fig. 1N) and when the gut attains its full length, all the intestinal cells are depleted from Xlox transcripts (Fig. 1O).

Finally, we found that the stomach-specific ManrC1A and ChP genes are already co-expressed at mid gastrula stage in the cells that will give rise to the larval stomach (Fig. 1P) and are never expressed in Xlox- (Fig. 1Q,R), Cdx- (data not shown) or Brn1/2/4-positive cells (Fig. 1S,T). A third stomach marker, Endo-16, transcribed in both midgut and hindgut cells at late gastrula (Ransick et al., 1993), is progressively cleared from the intestinal cells in parallel with the activation of Cdx, and becomes confined to the stomach larval cells. No signs of co-expression of Endo-16 and Cdx have been detected either at prism or at pluteus stages (Fig. 1U,V).

After the high resolution gene expression analysis, we investigated the regulatory mechanisms of Xlox and Cdx transcription. We perturbed protein translation by injecting morpholino antisense oligonucleotides (MOs), then analyzed Xlox and Cdx mRNAs and, in some cases, proteins, in the 72 h morphant larvae.

The results of the gene perturbation experiments are depicted in Fig. 2, and supplementary material Figs S1 and S2. A description of the postulated gene regulatory dynamics is provided below. In control larvae, Cdx transcripts are localized in the intestinal cells, whereas Xlox transcripts are detectable in the cells of the pyloric sphincter. Interestingly, Cdx transcripts were undetectable in the intestinal cells of Hox11/13b, Bra, FoxA, Cdx and Blimp-1a morphant larvae (Fig. 2A-F), suggesting that they are involved in Cdx regulation. Crucially, in Hox11/13b, Bra, FoxA and Cdx knockdown larvae (in which Cdx expression is lost), Xlox transcripts accumulated ectopically in the intestinal cells (Fig. 2G-L), providing additional support to Cdx repressive role on Xlox transcription (Cole et al., 2009). Moreover, Xlox transcripts were not detectable in Blimp-1a morphants (Fig. 2L), suggesting that Blimp-1a might be required for Xlox transcription, which also explained the absence of Cdx from Blimp-1a morphants, as Cdx transcription requires Xlox input (Cole et al., 2009). However, Blimp-1a transcription is active in a broader domain compared with Xlox (Livi and Davidson, 2006), suggesting that other unknown factors are expressed in prospective Xlox-positive cells and recruited for its activation.

An intriguing finding has been the impairment of Cdx transcription when its own protein formation was blocked (Fig. 2E), with the consequent accumulation of Xlox transcripts in the posterior larval gut cells (Fig. 2K). We suggest that Hox11/13b, Bra and FoxA, together with Xlox, all cooperate for the activation of Cdx in the subset of gut cells destined to form the intestine. Subsequently, because in the larval intestine only some Cdx-positive cells still express the factors responsible for its initial activation (see double in situ hybridization in Fig. 1), Cdx becomes recruited for its own transcription, ‘locking-in’ the regulatory state of intestinal cells.

The injection of morpholinos blocking the translation of GataE and TgiF (Howard-Ashby et al., 2006; Lee and Davidson, 2004) did not produce any change in the Xlox and Cdx transcript distribution (supplementary material Fig. S1), allowing us to exclude their involvement in Xlox and Cdx regulation. Most of the results presented in this section and the demonstration of morpholino functionality have been confirmed through the use of anti- Lox, Cdx and Bra antibodies (supplementary material Fig. S2).

Taking all the data into account, we propose a model that places Hox11/13b, Bra, FoxA and Xlox together as members of the regulatory machinery activating Cdx transcription. As expression of the first three genes starts many hours before Cdx activation, we hypothesize that Xlox functions as a switch controlling the exact time at which Cdx transcription is initiated. Finally, Blimp1a results as one of the activators of Xlox transcription: its expression starts many hours before the beginning of gastrulation, thus suggesting the presence of additional factors involved in Xlox transcription activation.

As described above, we found that Cdx activation requires Hox11/13b, FoxA, Bra and Xlox, and that after the translation of Cdx protein, Xlox transcription stops in the developing intestine. In order to explore the possible involvement of intercellular signaling in a Xlox-Cdx positive-negative feedback loop, we screened by qPCR (see Materials and methods for details) a variety of signaling molecules expressed during sea urchin embryonic development. The screening revealed that SpWnt10 transcript number was strongly reduced in Cdx knockdown larvae, persuading us of its involvement in Cdx- regulated processes. We then tested Wnt10 function through the embryonic injection of a morpholino designed to block its translation. Wnt10 knockdown larvae presented an almost normal gut characterized by a significant reduction of the pyloric sphincter constriction when compared with control larvae (for the phenotype of the morphants, see the full projection of confocal z-series in Fig. 3). Interestingly, Xlox and Cdx double in situ hybridization revealed that the two genes were co-expressed in the intestinal cells of Wnt10 knockdown larvae. Immunolocalization experiments confirmed that Xlox protein ectopically accumulates in Wnt10 knockdown larval intestines (supplementary material Fig. S2K). These results significantly contributed to the understanding of Xlox and Cdx regulation, indicating that Cdx-negative control of Xlox transcription is indirect and occurs via a Wnt10 signaling event (Fig. 3A-D). Very importantly, Wnt10 transcripts were not detectable in Cdx knockdown larvae by in situ hybridization (Fig. 3E,F), confirming the role of Cdx in the activation of Wnt10 transcription. We propose that Cdx participates in Wnt10 transcription in the posterior cells of the intestine and that Wnt10 ligands diffuse towards the anterior side of the gut, clearing the intestinal cells of Xlox transcripts through the activation of an effector molecule with repressive functions.

Digestive functions in the stomach are inhibited in Xlox knockdown larvae (Cole et al., 2009). We explored the expression of stomach terminal differentiation genes, ManrC1A, ChP and Endo-16, in Xlox and Cdx knockdown larvae. The expression of all three genes in Xlox morphants was severely affected (Fig. 4). In particular, ChP and ManrC1A transcription was dramatically reduced, to the point that no transcripts could be detected by either colorimetric (Fig. 4A,B,D,E) or fluorescent in situ hybridization (Fig. 4G-H). Xlox is thus upstream of a cascade of regulatory events responsible for the differentiation of the larval stomach. However, the localization of Xlox mRNA and protein relative to ChP and ManrC1A transcripts during gut development (see Fig. 1) advocates the involvement of a signaling event in the induction of stomach cell differentiation. Although Cdx knockdown did not have any noticeable effect on ChP and ManrC1A transcription (Fig. 4C,F,I), Endo-16 ectopic expression was observed in the intestine of both Xlox and Cdx morphants (Fig. 4J-L). We propose that Cdx is responsible for Endo-16 repression from the intestinal cells, acting either directly on its transcription or through the repression in the intestine of the Endo-16 midgut-hindgut activators, and that Xlox executes a repressive function on Endo16 transcription (Cole et al., 2009) in the cells where Cdx is never active and that will eventually form the pyloric sphincter.

The most obvious morphological defect in Xlox knockdown larvae is the absence of the pyloric sphincter (e.g. compare confocal images in Fig. 4G,H), the cells of which express the muscle-specific terminal differentiation gene, Myosin heavy chain (SpMHC) (Andrikou et al., 2013; Venuti et al., 1993). In control larvae, MHC is expressed in the four muscular structures of the gut, the esophageal muscles and the cardiac, pyloric and anal sphincters. The esophageal muscles are of mesodermal origin whereas sphincter muscles are endodermal (Gustafson and Wolpert, 1967), suggesting the existence of different regulatory mechanisms involved in the acquisition of muscle cell identity in each type. We found that MHC transcripts were specifically absent from the cells forming the pyloric sphincter in Xlox morphants (Fig. 4M-O) and suggest that a MHC-specific cis-regulatory module active only in these cells is responsive, directly or indirectly, to an Xlox regulatory input. This regulatory mechanism drives the differentiation into muscles of a small population of endodermal gut cells, allowing the formation of the constriction between stomach and intestine.

In order to obtain a global vision of transcriptional responses to Xlox perturbation, we performed a genome-wide comparative transcriptome analysis (RNAseq) on Xlox knockdown versus control 48 and 72 h embryonic RNA (Fig. 5 and Table 1). The MA plots in Fig. 5 clearly show that the effect of Xlox perturbation on gene expression was much stronger at 72 h, thus potentially supporting our hypothesis of a signaling event depending on Xlox and directing stomach differentiation. In particular, we found 5767 differentially expressed transcripts, 23.2% of them upregulated and 28.8% downregulated (by at least 1.5 fold). In Table 1 we provide a list of genes whose expression was significantly affected in the differential RNAseq analysis, grouping them based on their functional category and providing the expression pattern in the sea urchin embryo, when available. Among these, there are genes, such as SpCdx, SpLox and Endo16, that have been previously identified as SpLox targets through qPCR and in situ hybridization analysis by Cole et al. (2009). Among the most strongly newly identified downregulated genes are the enzymes involved in the metabolism of proteins, carbohydrates and lipids, possibly supporting a role for Xlox in the differentiation of a functional digestive system. In addition, we found a substantial decrease in the expression of Ache-13 (an acetylcholinesterase that functions in neuromuscular junctions) that we correlated with the absence of pyloric sphincter muscular fibers in Xlox morphant larvae (Fig. 4M-O). Likewise, three transcription factors of the Forkhead family (FoxY, FoxD and FoxP), mainly expressed in foregut and midgut territories, were significantly downregulated, thus suggesting their potential involvement in Xlox gut patterning control. NeuroD1, a regulator of insulin production in vertebrates (Babu et al., 2008; Kaneto et al., 2009), was also strongly downregulated, indicating a possible evolutionary conserved cooperative function for Xlox and NeuroD1 in the sea urchin larva. Moreover, a GABA receptor and a cholinergic receptor were highly downregulated and we associated these results with the potential role of Xlox in the specification of some neural cells of the ciliary band from a group of ectodermal cells in which Xlox transcripts have been revealed (Cole and Arnone, 2009).

The expression of several gut transcription factors, e.g. SpGataE (Lee and Davidson, 2004), SpHnf1 and SpTgiF (Howard-Ashby et al., 2006), and SpPtf1a (Annunziata et al., 2013b), was not significantly affected in Xlox knockdown larvae. Finally, this analysis did not reveal any significantly affected signaling molecule-encoding gene, thus leaving unresolved the mechanism mediating Xlox function in the activation of stomach differentiation genes and introducing new possible hypotheses, such as Xlox requirement in the control of post-translational events, something not testable with the approaches followed so far.

Interestingly, the validation by qPCR of one of the most affected genes, SpPpglcp (a phosphoglycolate phosphatase), with a 3 h time resolution, from 44 h to 50 h (supplementary material Fig. S3), has revealed a strong decrease in its expression since the first analyzed stage (44 h), hinting a possible direct role for Xlox in the activation of Ppglcp transcription. However, information about Ppglcp spatial expression would be needed for a more complete interpretation of the strong effect of Xlox knockdown on its transcription.

In the present study, we show that the Xlox transcription factor is required to induce ChP and ManrC1A transcription in the stomach cells. This, together with the reduction of digestive function and the alteration of the stomach morphology in Xlox morphants (Cole et al., 2009), as well as the strong reduction in expression of digestive enzymes observed in our transcriptomic analysis, provide a strong support to Xlox key role in the development of a functional stomach. However, the detailed expression dynamics analyzed in this work imply the existence of a signaling event (S1 in Fig. 6) under the control of Xlox that induces the expression of the two stomach-specific genes. Moreover, the sharp boundary of gene expression between stomach and esophagus suggests that the stomach precursor cells are already committed to their future fate, but require an Xlox-induced signal in order to differentiate. Additionally, Xlox is upstream of the entire cascade that regulates intestine formation by acting as a decisive activator of Cdx transcription in the intestine precursor cells (Cole et al., 2009). Xlox also drives the formation of the pyloric sphincter constriction, triggering the differentiation of muscles from a small subset of endodermal cells, via the activation of MHC transcription. Interestingly, the murine Xlox homolog Pdx-1 has a crucial role in pancreas formation and differentiation, and is also involved in pyloric sphincter morphogenesis (Offield et al., 1996). Tissue boundaries keep physically separated neighboring groups of cells with distinct differentiation fates and often act as organizing centers for compartment patterning (for a review, see Dahmann et al., 2011). We propose that Xlox-positive cells located at the stomach-intestine boundary may function as an organizer of gut domains in both the anterior and posterior sides. Thus, the pyloric sphincter represents a very powerful model for investigating the gene regulatory interactions acting at the level of morphological boundaries, and the sea urchin embryo is well suited to this kind of study. A powerful approach we employed in this context was the genome-wide differential transcriptomic analysis on Xlox morphants. This type of study has provided a large amount of information and has high potential in resolving Xlox patterning functions. Among the most strongly affected genes were the ones encoding metabolic enzymes, muscle-associated proteins and neuronal factors, plus some transcription factors expressed in the gut, strongly suggesting that Xlox is upstream of a number of events leading to the gut AP differentiation.

Finally, we detected Xlox proteins in cells where the mRNA was untraceable. Analyses of gene expression through transcript detection are widely used in the construction of gene regulatory networks. Our study highlights the importance of taking protein in addition to transcript localization into account when defining cellular regulatory states.

Whereas the foregut and midgut domains appear definitively established at the late gastrula stage, undergoing mainly morphological changes during the subsequent stages of development, the hindgut domain remains a dynamic site of cell fate specification throughout the late gastrula and early prism stages (see Fig. 6). A crucial event taking place during hindgut development is the activation of Cdx transcription in the most posterior cells of the late gastrula gut; we found a cassette of regulatory factors, namely Xlox, Hox11/13b, FoxA and Bra, involved in Cdx activation. These four transcription factors are all necessary for Cdx transcription, although we do not know whether they function directly on Cdx activation or recruit intermediate factors. One possibility, as they are all expressed in the cells where Cdx transcription starts, is that they might function in an ‘AND’ logic (Davidson, 2010), probably together with other transcription factors and/or co-factors, to drive the specification of intestinal cells. Cdx, however, has an ‘exclusion effect’ (Oliveri and Davidson, 2007), repressing a potential alternative cell regulatory state available to the intestinal cells (its effect on Xlox and Endo-16). We discovered that the role of Cdx in repressing Xlox in the intestine is mediated by a Wnt10 ligand. In addition, we showed that Wnt10 ligands are actively transcribed in the most posterior cells of the gut at the late gastrula-prism stage and propose that they diffuse towards the anterior side of the gut, progressively clearing the intestinal cells from Xlox (and probably other) transcripts. We do not know whether the newly identified Wnt signal works through β-catenin or activates a non-canonical Wnt pathway. The identity of the receptor mediating this Wnt10 signaling event is also unknown. Four Frizzled receptor genes have been identified so far in the Strongylocentrotus purpuratus genome but no information is available about the expression dynamics of these genes during embryogenesis. The expression of these receptors has been deeply investigated in another sea urchin species, Paracentrotus lividus, indicating Fzd9/10 as possible candidate for mediating Wnt10 signaling in the hindgut (Croce et al., 2006; Lhomond et al., 2012; Robert et al., 2014). It would be interesting to study the dynamics of expression of the four Frizzled receptor genes in S. purpuratus in order to identify a possible candidate for the transduction of the Wnt10 signal to the intestinal cells. Interestingly, mechanisms of function for Cdx transcription factors involving activation of Wnt signaling have been previously observed in mouse, precisely in the uro-rectal mesoderm development (van de Ven et al., 2011). Moreover, Cdx appears, together with Bra, FoxA and Wnt, as part of a conserved cassette of factors regulating posterior gut development. These genes are all expressed in the blastopore or blastopore equivalent of frog, zebrafish, Drosophila and mouse (Lengyel and Iwaki, 2002). Furthermore, Cdx genes, Wnt signaling, Bra and posterior Hox genes act together to control posterior morphogenesis in the different murine embryonic germ layers (van de Ven et al., 2011). We found Cdx, Wnt10, FoxA, Bra and Hox11/13b actively involved in the sea urchin posterior gut development, suggesting conservation of a gene network functioning for the differentiation of posterior structures. The strong reduction of Bra expression in Cdx morphant embryos observed by qPCR (supplementary material Fig. S4), together with the very similar expression pattern observed for Bra and Wnt10 from late gastrula stage until pluteus, allow us to propose two possible scenarios: (1) Bra receives a fundamental regulatory input from Cdx and participates, together with Cdx, in Wnt10 activation in the very posterior cells of the post gastrular embryo; (2) Cdx activates Wnt10 expression indirectly, by activating Bra. Both scenarios would provide an explanation for the fact that Wnt10 is expressed only in a subset of Cdx-expressing cells, although we have no data so far to support a role for Bra in Wnt10 activation. This would not be the first case in which Bra activates Wnt signals: in chordates, Bra recruits canonical Wnt signaling to sustain the posterior mesodermal progenitors during the outgrowth of the body and to regulate the mechanism of somitogenesis (Martin and Kimelman, 2009).

We also showed that Endo-16 transcripts accumulate ectopically in the intestine of Cdx morphant larvae and that Hox11/13b is required for activating Cdx transcription, thus providing a candidate for the factor hypothesized to mediate the Hox11/13b knockdown phenotype in the posterior gut (Arenas-Mena et al., 2006). Interestingly, Cdx also functions as a repressor in the intestine of vertebrates: Cdx2 conditional ablation from early murine endoderm results in a posterior-to-anterior gut transformation, through the replacement of the posterior intestinal epithelium with esophageal epithelia (Gao et al., 2009). Furthermore, ectopic expression of Pdx1 (the ortholog of the sea urchin Xlox gene) has been demonstrated in Cdx2 mouse mutant intestine (Grainger et al., 2010). As already suggested by Grainger et al. (2010), this would hint at conservation of the cross-regulatory loop described in sea urchin between the two genes. However, two phases of Cdx activation have been observed in chordates (Chawengsaksophak et al., 2004; Osborne et al., 2009; Reece-Hoyes et al., 2002), the first in the blastopore-primitive streak regions, where it is required for axial elongation, at least in mouse (Chawengsaksophak et al., 2004), and the second in the posterior gut, where it is involved in intestinal patterning. In the sea urchin embryo, Cdx transcription starts towards the end of gastrulation. Thus, in the sea urchin, Cdx function in the early phases of gastrulation has been lost and only its function in posterior gut patterning has been retained. This condition is more likely to be an echinoid peculiarity rather than a feature common to all ambulacrarians or to all echinoderms, as Cdx ‘biphasic’ transcription has been described in sea stars and hemichordates (Annunziata et al., 2013a; Ikuta et al., 2013). Very importantly, the fact that Cdx is not required in the early steps of sea urchin gastrulation offers a great advantage for the study of its role in the intestine differentiation, facilitating the interpretation of functional studies.

The purpose of this work was to investigate the gut patterning process in its entirety, exploring the dynamics of gene expression and regulation in space and time, a feasible objective using the sea urchin embryo as model system. The data we present here clearly show that while the initial patterning of the gut tube relies mainly on the A-P distribution of transcription factor territories of expression, the subsequent compartmentalization depends, at least partially, on signaling events under the control of the two ParaHox genes, Xlox and Cdx (see Fig. 6). It is important to note that none of the interactions found in this study has been demonstrated to be direct and that the wiring diagram presented in Fig. 6 remains far from being complete, missing many other levels of regulation such as the involvement of small non coding RNAs, post translational modifications, protein-protein interactions and chromatin remodeling.

Our findings suggest that Xlox ancestral function in the endoderm was probably to pattern the digestive tube directing stomach cell differentiation and pyloric sphincter formation, as supported by conservation of its gut domain of expression in several protostomes and non-vertebrate deuterostomes. The vertebrate homolog Pdx-1 has been possibly coopted for the development of the pancreas, maintaining its role in the formation of the stomach-intestine constriction. Similarly, the crossregulatory loop between Xlox and Cdx firstly described in sea urchin (Cole et al., 2009) appears as a conserved gut patterning mechanism among deuterostomes (Grainger et al., 2010). We predict that many of the hereby demonstrated regulatory interactions shaping the sea urchin gut, and the many others we will find by the ongoing ChIPSeq and RNASeq analyses downstream of SpLox and SpCdx, will be instrumental for the discovery of yet unknown patterning mechanisms of the vertebrate gut.

Adult Strongylocentrotus purpuratus were obtained from Patrick Leahy (Kerchoff Marine Laboratory, California Institute of Technology, Pasadena, USA) and housed in circulating sea water aquaria in the Stazione Zoologica Anton Dohrn of Naples, Italy. Gametes were obtained by standard methods and embryos were cultured at 15°C in filtered sea water diluted 9:1 with de-ionized water.

For single gene expression, we followed the protocol outlined previously (Minokawa et al., 2004). Fluorescent double WM in situ hybridization were performed as described previously (Cole et al., 2009). RNA probe sequences for SpLox, SpCdx, SpBrn1/2/4, SpBra and SpEndo16 are as previously published: SpLox and SpCdx (Arnone et al., 2006); SpBrn1/2/4 (Cole and Arnone, 2009); SpBra (Rast et al., 2002); and SpEndo16 (Ransick et al., 1993). SpHox11/13b, SpChP, SpManrC1A, SpMHC bacterial clones were picked from the S. purpuratus cDNA library available in the laboratory (http://goblet.molgen.mpg.de/cgi-bin/seaurchin-database.cgi). SpWnt10 cDNA was obtained by RT-PCR from 65 h embryo total RNA (using the following primers: SpWnt10-F: 5′-AGACGATGGAATTGCTCCAG-3′; SpWnt10-R: 5′-GGTTAACCCATTGCGAGCTA-3′) and then cloned into the Topo-TA cloning vector (Invitrogen).

For acetylated tubulin staining combined with in situ hybridization, a 1:250 dilution of the mouse monoclonal anti-acetylated tubulin antibody (T7451; Sigma-Aldrich) was added to the blocking solution containing the peroxidase-conjugated antibody and incubated overnight at 4°C. The next day the embryos were processed for in situ hybridization, then incubated for 1 h in blocking solution with a dilution 1:1000 of the AlexaFluor 488 goat anti-mouse immunoglobulin G (IgG) (A21202; Molecular Probes), washed in MOPS (3-morpholinopropane-1-sulfonic acid) buffer and mounted for imaging. For Xlox and Cdx immunostaining, larvae were fixed in 2% PFA in phosphate-buffered saline (PBS) for 10 min at room temperature, washed multiple times in PBS with 0.1% Tween-20 (PBST), blocked in 5% goat serum in PBST and incubated overnight at 4°C with the custom primary antibodies diluted at working concentration (1:500; PRIMM) in 5% goat serum in PBST. For Bra immunostaining, embryos and larvae were fixed in 2% PFA in PBS for 10 min, then for 1 min in methanol, blocked in 5% goat serum in PBST and incubated overnight with the custom antibody at working concentration (1:200; A11008; PRIMM). Following primary antibody incubation, embryos and larvae were washed several times in PBST, incubated for 1 h at room temperature with the secondary antibody Alexa Fluor 488 goat anti-rabbit IgG (Molecular Probes, Invitrogen) diluted 1:1000 in 5% goat serum in PBST, washed in PBST and mounted for imaging with a confocal microscope (Zeiss 510Meta).

For each experiment and for each morpholino oligonucleotide (MO), around 400 embryos were injected with approximately 2-4 pl of oligonucleotide solution at 0.1 mM and each experiment was repeated three times. In all experiments, as a negative control, embryos were injected with 0.1 mM of the standard control morpholino and compared side by side with uninjected and knockdown embryos. The injection of the standard control morpholino (GeneTools) did not have any effect on the development of embryos. MOs against SpLox, SpHox11/13b, SpFoxA, SpBlimp1a, SpGataE, SpTgiF and SpBra protein translation and the morpholino targeting the donor splice site between the first and second SpCdx exons, were already available (http://sugp.caltech.edu/endomes/) (Arenas-Mena et al., 2006; Cole et al., 2009; Livi and Davidson, 2006; Oliveri et al., 2006; Rast et al., 2002). MOs against SpWnt10 and SpCdx translation were newly designed and acquired from Gene Tools (Corvallis) (SpWnt10, 5′-AACTGCATCTGCTTACGATTCATAC-3′; SpCdxM, 5′-TGGGTGCAGATACTCTAGCGTCATC-3′).

In order to identify the signaling molecule responsible for the midgut differentiation process, 15 signaling ligand encoding genes have been analyzed. The genes included in the analysis were selected from the Strongylocentrotus purpuratus Signaling Ligand Page (available at http://140.109.48.251/ICOBUserfile/SuYuLab/Su_and_Yu_Lab/Home.html), a database created by Yi-Hsien Su's laboratory (Institute of Cellular and Organismic Biology, Academia Sinica, Taiwan), in which the temporal expression pattern of all signaling ligands expressed during sea urchin embryonic development is provided. The signaling ligand database was screened using as selective criteria the windows of activation of the genes. In particular, the ligands whose expression started or significantly increased a few hours after 44 hpf (the developmental time when Cdx expression begins) were considered to be eligible signaling molecules and screened by qPCR. The differential quantitative expression of the selected genes in control and Cdx knockdown embryos (data not shown) was analyzed. The 15 signaling molecules included in the analysis were: SpWnt8, SpIGF2, SpHGF, SpAgrin3, SpVegF, SpBMP3, SpBMP5/8, SpNotch-lik2, SpNotch-lig3, SpNotch-lid5, SpEphrin, SpSHH, SpWnt10, SpWnt4 and SpFGF9/16/20. For the qPCR, the primers were designed based on the sequences available in the signaling ligand database.

Three biological replicas of 500 injected with SpLox MO and control (KCl-injected) embryos were collected at 48 and 72 h. About 1 μg of RNA was extracted from each sample using the RNAqueous-Micro Kit (Ambion, Life Technologies-Invitrogen) and used for RNA-seq. cDNA libraries were prepared with 1 µg of starting total RNA and using the Illumina TruSeq RNA Sample Preparation Kit (Illumina), according to TruSeq protocol. Library size and integrity were determined using the Agilent Bioanalyzer 2100. Each library was diluted to 2 nM and denaturated, 8 pM of each library was loaded onto cBot (Illumina) for cluster generation with cBot Paired End Cluster Generation Kit (Illumina) and sequenced using the Illumina HiSeq 1500 with 100 bp paired-end reads in triplicate, obtaining ∼31-38 million reads for replicate. The sequencing service was provided by the Laboratory of Molecular Medicine and Genomics (http://www.labmedmolge.unisa.it) at the University of Salerno, Italy. Sequence read quality was controlled using FastQC program (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc/). Filtered reads were mapped to the Sea Urchin genome Spur_3_1.LinearScaffold downloaded from SpBase (www.spbase.org) using TopHat v2.0.8b (Trapnell et al., 2009), that runs on Bowtie 2 version 2.1.0 (Langmead et al., 2009). The reads were counted using the HTSeq-package (Anders and Huber, 2010). Normalization of read numbers between samples and differential expression analysis was performed using DEseq (Anders and Huber, 2010). The transcripts with an adjusted P<0.05 were considered to be differentially expressed. Differential expression is represented with MA-plots generated with the DESeq software.

We acknowledge Patrick Leahy and the Southern California Sea Urchin Company for animal supply; Davide Caramiello for taking care of the sea urchins; the SZN Molecular Biology Service for technical assistance; the Laboratorio di Medicina Molecolare e Genomica, Università degli Studi di Salerno, Italy, for the RNA-seq experiment; Giorgio Giurato and Francesca Rizzo for the MA-plots; Eric Davidson for Bra, FoxA, GataE and TgiF MOs; Yi-Hsien Su for the signaling molecule database; Evelyn Houliston and Pedro Martinez for revising our manuscript; and Claudia Cuomo for the kind experimental help.