Neurogenesis in sea urchin embryos and the diversity of deuterostome neurogenic mechanisms

Diverse developmental mechanisms mediate the specification and differentiation of metazoan nervous systems. There are similarities suggesting homology, but we lack a broad understanding of neural development mechanisms in metazoans. In some organisms neurogenic potential is spread throughout the ectoderm and neural precursors arise and differentiate to form a nerve network (Richards and Rentzsch, 2014; Cunningham and Casey, 2014). In many bilaterians, neurogenesis is restricted to specific domains of ectoderm by early patterning (Bier and De Robertis, 2015). A key aspect of neurogenesis is the cell division that marks the transition of pre-mitotic neural progenitors to post-mitotic neural precursors committed to differentiation (Hartenstein and Stollewerk, 2015). In the best-studied models there are various arrangements of asymmetric, self-renewing mitoses within populations of dedicated neural progenitors (Goodman and Doe, 1993; Götz and Huttner, 2005). There are common components and shared gene families that appear to have key roles in the specification and differentiation of neurons (Wei et al., 2009; Sinigaglia et al., 2013; Marlow et al., 2014). In addition, Notch/Delta lateral inhibition of neurogenesis is a common feature, but the details of how it is employed vary widely (Zhong, 2003; Yoon and Gaiano, 2005). It is not clear whether these similarities result from common origin or are features that arose independently because of functional constraints. Overall, we lack details of neurogenic mechanisms for many taxa that would enable a broader understanding of neural evolution (Hartenstein and Stollewerk, 2015).

Bilaterians have central and peripheral components to their nervous system, which are commonly derived from different regions of ectoderm. Typically, the central nervous system (CNS) is derived from a neuroepithelium and peripheral neurons arise in ectoderm and subsequently connect to the CNS (Hartenstein and Stollewerk, 2015). In sea urchin embryos, the animal pole domain (APD) is a specialized region of ectoderm with features of a neuroepithelium (Angerer et al., 2011). The genes expressed in the APD are similar to those expressed in the vertebrate forebrain (Wei et al., 2009) and the APD expands in the absence of Wnt signaling (Yaguchi et al., 2006; Cui et al., 2014). The APD gives rise to the apical organ, the most prominent ganglion of the larval nervous system. The larval nervous system also includes an array of neurons, with features of sensory cells and interneurons, in the ciliary band. Descriptions of the development of the larval nervous system indicate that these cells differentiate in the ciliary band, and neurons extend axons in tracts at the base of the ciliated cells to the apical organ (Burke et al., 2014). The nervous system is thought to function by controlling the direction of ciliary beat by regulating depolarization of ciliated cells in the ciliary band (Mackie et al., 1969; Satterlie and Cameron, 1985; Strathmann, 2007). Thus, a conventional interpretation of the organization of the larval nervous system is that the apical organ constitutes a central, integrating component, and there are peripheral neurons in the ciliary band that respond to stimuli and locally, or coordinately, control ciliary activity.

The apical organ neurons appear to be derived from cells in the APD that differentiate as neurons, but it is not clear which regions of ectoderm give rise to the neurons of the ciliary band. Markers for neural differentiation are expressed in gastrulae by cells within the ciliary band or APD (Bisgrove and Burke, 1986; Nakajima et al., 2004). Thus, the ciliary band has been suggested to be neurogenic (Poustka et al., 2004; Wei et al., 2009; Saudemont et al., 2010). Based on the pattern of expression of Six3 and its requirement for neurogenesis, Wei et al. (2009) suggested that neurons might be specified in the APD or lateral ectoderm and only differentiate in the ciliary band or migrate from the APD to the ciliary band. The number of neurons increases throughout development of the larva, which indicates that neurogenesis is not restricted to embryonic stages and that there must be sites of continuous neurogenesis (Bisgrove and Burke, 1986; Beer et al., 2001). However, the origin of neurons remains unknown because we lack molecular markers for neurogenesis and studies of neurogenesis tracking individual cells.

Here we describe the expression in sea urchin of several proteins implicated in neurogenesis. We have developed antibodies to putative markers for neural progenitors to determine the sequence in which they are expressed. We restricted our analysis to the first neurons that differentiate, as the time and location of their differentiation are predictable. We used a series of pulse-chase experiments with EdU, a thymidine analog (Chehrehasa et al., 2009), to demonstrate that the last mitotic division of neural progenitors for these neurons coincides with the expression of SoxC. Neurogenesis appears to require the expression of several key neurogenic transcription factors and is also blocked by interfering with FGF signaling or blocking expression of an FGF receptor. These data indicate that, in addition to the neuroepithelium of the APD, neurons arise in restricted regions of ectoderm adjacent to the ciliary bands. For some neurons there is an invariant pattern of appearance of neural progenitors, but most of the ciliary band neurons arise asynchronously and unpredictably. Neural progenitors appear to express SoxC during their last cell cycle and divide once before a single neural precursor expresses Brn1/2/4 and subsequently differentiates. This model of neurogenesis has features that are shared with chordates and is distinct from neurogenesis in other ambulacrarians.

In situ hybridization data suggest that SoxC is expressed by neural progenitors (Howard-Ashby et al., 2006; Poustka et al., 2007). To determine if this is so, we prepared polyclonal rat antiserum and a mouse monoclonal antibody to Strongylocentrotus purpuratus SoxC (Fig. S1). SoxC is expressed initially during cleavage in vegetal lineages (Fig. S2). In the APD, expression of SoxC protein was first detected prior to hatching (blastula, 20 h, 14°C) at the same time that NK2.1 is first detected (Fig. 1, Fig. 2A-I). Initially all cells of the APD express SoxC and levels of fluorescence vary, but by 24 h (hatching) the number of cells expressing SoxC is reduced to one or two cells (Fig. 2A-I). The cells are at the periphery of the APD along the dorsal margin. The number of cells in the APD expressing SoxC increases to eight to ten cells dispersed around the periphery of the apical organ of early plutei (Fig. 2J-N).

Cells in the oral ectoderm also express SoxC. At about 28 h there are two pairs of cells in the positions where postoral neurons differentiate that express SoxC (Fig. 2K). The postoral neurons are identified by their distinctive position at the base of the postoral arms (Fig. 1) (Burke et al., 2014). Cells in these positions continue to express SoxC until late gastrula stages. During gastrulation (36-48 h), cells adjacent to the presumptive ciliary band express SoxC (Fig. 2L-Q). There are 15-25 cells, often in pairs, along the oral or the aboral edge of the ciliary band in prism and pluteus stages that express SoxC (Fig. 2M-Q). In addition, there are scattered cells surrounding the presumptive mouth that express SoxC (Fig. 2M,N,P). Some cells of the ciliary band, identified by expression of Hnf6 (Poustka et al., 2004; Otim et al., 2004) also express SoxC (Fig. 2N-Q). With differential interference contrast or transmitted light imaging, the cells expressing SoxC are morphologically indistinguishable from adjacent cells. Cells expressing SoxC do not co-express synaptotagmin B (SynB) (Fig. 2M). Thus, SoxC is expressed transiently in endomesoderm, in individual cells in the APD, and restricted regions of ectoderm beginning in late blastulae.

To determine if the ectodermal cells expressing SoxC are neural progenitors, we developed antibodies to ELAV and Brn1/2/4 (Figs S1 and S3). ELAV is an RNA-binding protein used as a marker for undifferentiated neurons in several organisms (Robinow and White, 1988, 1991; Good, 1995). ELAV protein is expressed in blastulae (24 h) in primary mesenchyme cells and throughout gastrulation in pigment cells (Fig. S3). Differentiated neurons within the ciliary band and APD also express ELAV (Fig. 3B). ELAV protein localizes throughout the cytoplasm and in patches in nuclei of neurons (Fig. 3D-F). Brn1/2/4, a POU family transcription factor with neural specification roles in other organisms (Ryan and Rosenfeld, 1997; Wegner et al., 1993; Brombin et al., 2011), is also expressed by differentiated neurons within the ciliary band, where it localizes to nuclei (Fig. 3A-F). Thus, ELAV and Brn1/2/4 serve as additional markers for neurons and neural progenitors, although ELAV expression in mesodermal lineages limits its utility. Brn1/2/4 is initially expressed at about 48 h in cells in the position of postoral neurons and in the position of the serotonergic neurons along the dorsal margin of the APD (Fig. 1, Fig. 3G-I). Cells in the position of postoral neurons also express SoxC and, in doubly labeled preparations, it is clear that SoxC and Brn1/2/4 are co-expressed by these cells (Fig. 3G-I, insets). Thus, Brn1/2/4 and SoxC appear to be expressed by partially overlapping sets of ectodermal cells. Initially, SoxC is expressed, subsequently SoxC and Brn1/2/4 are co-expressed, then Brn1/2/4 alone is expressed by cells in the position of postoral neurons.

SoxB2 is expressed in all cells beginning in early cleavage. SoxB2 is an HMG-box-containing transcription factor thought to be important in neuronal differentiation (Richards and Rentzsch, 2014; Taguchi et al., 2002). At hatching, SoxB2 is downregulated at the vegetal pole (Fig. 4A,B, inset). In gastrulae it is apparent that the endomesodermal lineages do not express SoxB2 (Fig. 4C). In addition, SoxB2 expression is downregulated in the animal pole; however, all other ectoderm continues to express SoxB2 (Fig. 4C). In embryos prepared with SoxB2 and SoxC antibodies, cells in the APD expressing SoxC are adjacent to the zone in which SoxB2 is cleared. Thus, SoxC and SoxB2 are co-expressed in cells at the periphery of the APD (Fig. 4D). Similarly, cells in the ectoderm adjacent to presumptive ciliary band that express SoxC also express SoxB2 (Fig. 4E-G). Differentiated neurons do not express SoxB2.

As neurons are post-mitotic, we have used pulse-chase experiments with labeled nucleoside to determine the time at which the neurons identified by position are in S-phase of their final cell cycle. Embryos were labeled with EdU for intervals beginning in late cleavage stages, and then prepared for immunofluorescence with anti-SynB as early plutei (72 h, Fig. 5). Nuclei that had incorporated EdU during the labeling interval were distinguishable from unlabeled nuclei (Fig. 5A,B). Differentiated neurons were not labeled with EdU when pulsed at 48 h or later, yet almost all neurons were labeled with EdU when pulsed between 16 h and 20 h (Fig. 5C). The interval during which the postoral neurons could be labeled was between 16 h and 32 h, with the majority of the postoral neurons labeled between 16 h and 28 h (Fig. 5D). Similarly, the first differentiated serotonergic neurons in the apical organ could be EdU labeled between 16 h and 28 h, with the majority being labeled between 16 h and 24 h (Fig. 5E). As cells expressing SoxC are occasionally observed in mitosis (Fig. 5F-H), we attempted to determine whether SoxC-expressing cells are mitotically active by pulsing with EdU for 2 h, then looking for cells labeled with EdU that are expressing SoxC. A small proportion of cells expressing SoxC at the time of fixation were labeled with EdU (Fig. 5I-K). These experiments indicate that neural specification is likely to begin during late cleavage. In addition, SoxC expression begins at almost the same time as neural progenitors, identified by position, are in S-phase of their last cell cycle. These data suggest that neural progenitors express SoxC before completing their last cell cycle and dividing to become neural precursors.

To determine if SoxC is essential for neurogenesis, we used morpholino antisense oligonucleotides (MASO) to interfere with the expression of SoxC (Figs 6 and 7). In embryos injected with the SoxC MASO, fewer cells were detected with anti-SoxC, indicating that any of the three overlapping MASOs that we have used blocks expression of SoxC protein (Fig. 6A-K, Fig. 7A). In 28 h mesenchyme blastulae, the expression of SoxC is completely eliminated in vegetal and APD cells, without any effect on the expression of SoxB2 (Fig. 6A-D, Fig. 7A). In 48 h gastrulae, MASO-injected embryos have fewer cells expressing SoxC (Fig. 6E-G, Fig. 7A). The number of neurons that differentiate by 72 h (SynB expression) in embryos injected with the SoxC MASO is reduced and most of the neurons are abnormal in morphology (Fig. 6H,I, Fig. 7A). Most commonly there are multiple short axons that are not organized into tracts. We concluded that expression of SoxC is necessary for neurogenesis.

Several studies of neurogenesis indicate that there are common features in components mediating embryonic neurogenesis in a wide range of metazoans (Hartenstein and Stollewerk, 2015). In some metazoans, expression of SoxB2 is a necessary condition for neurogenesis (Richards and Rentzsch, 2014). When we suppressed the expression of SoxB2 with a MASO, there was a reduction in the number of cells expressing SoxC (Fig. 6L, Fig. 7A). It is apparent that Six3 is near the top of a regulatory network mediating neurogenesis in sea urchins (Wei et al., 2009). To determine if SoxC expression is dependent on activation of the Six3 gene regulatory network we interfered with expression of Six3 with a MASO. The number of cells expressing SoxC in Six3 MASO-injected embryos was reduced compared with controls (Fig. 6M, Fig. 7A), indicating that SoxC expression is dependent on expression of Six3.

There are also indications from studies of neurogenesis in other metazoans that FGF and FGF receptors are necessary for embryonic neurogenesis (Rentzsch et al., 2008; Altmann and Brivanlou, 2001; Lemaire et al., 2002; Kengaku and Okamoto, 1993). FGF and its role in ectodermal signaling, specifically to pattern mesoderm, have been examined in sea urchins, and there is evidence that FGF is expressed in ectodermal cells adjacent to postoral neural progenitors and the APD (Lapraz et al., 2006; Rottinger et al., 2008; Adomako and Ettensohn, 2013). In addition, FGF receptor 1 (FGFR1) is expressed in cells of the oral half of the embryo, including the APD, presumptive ciliary band, oral ectoderm, and what will become the aboral ectoderm adjacent to the ciliary band (Lapraz et al., 2006; Saudemont et al., 2010). To determine if FGF signaling is required for expression of SoxC we used a selective inhibitor of FGF receptor activation (SU5402) (Mohammadi et al., 1997) and MASO knockdown of FGFR1. There was a decrease in the number of cells expressing SoxC after treatment with SU5402 (Fig. 7B). Treatment with 5 µM SU5402 reduced the number of cells expressing SoxC significantly (Fig. 6U, Fig. 7B). Similarly, interfering with the expression of FGFR1 with a MASO eliminated expression of SoxC (Fig. 6S, Fig. 7B). When we injected embryos with a MASO that interferes with Brn1/2/4 expression, the number of neurons that differentiate by 72 h is reduced (Fig. 6O-Q, Fig. 7A). However, the number of cells expressing SoxC at 48 h and 72 h was unaffected by interfering with Brn1/2/4 expression (Fig. 6E-G,O-Q, Fig. 7A). These experiments indicate that expression of FGFR1, SoxB2, Six3, and FGF receptor activation are necessary for the expression of SoxC and neurogenesis. However, interfering with Brn1/2/4 expression reduces the number of neurons produced, but has no effect on the expression of SoxC, consistent with Brn1/2/4 being downstream. We conclude that these proteins are components of a neurogenic pathway.

To determine if SoxC expression is sufficient to initiate neurogenesis, we injected mRNA encoding full-length SoxC (SoxC FL) or full-length SoxC fused to GFP (SoxC:GFP) into eggs. When SoxC:GFP mRNA is injected, there is a clear GFP localization to nuclei in 24 h blastulae (Fig. 8A). In addition, SoxC:GFP- or SoxC FL-injected blastulae prepared with anti-SoxC indicate that the SoxC protein is expressed and that it localizes to all nuclei, without altering the pattern of expression of SoxB2 (Fig. 8B-D). Beginning at 30 h, the nuclear localization of SoxC:GFP begins to decline and by 48 h SoxC:GFP is no longer detectable (Fig. 8E-G). Similarly, preparations with anti-SoxC at 48 h or 72 h indicate normal control levels of SoxC and SynB expression, with normal patterning when embryos are injected with up to 0.5 pg RNA (Fig. 8I-N, Fig. 7C). Embryos injected with higher doses (2 pg) had low survival and exhibited a range of developmental abnormalities. Expression of SoxB2 and of Brn1/2/4 were unaffected by SoxC mRNA injection (Fig. 8G,J,L). Control embryos expressing equivalent doses of GFP have high levels of fluorescence through at least 96 h with no effect on the expression of SoxC or SynB (Fig. 8H,O,P, Fig. 7C). We conclude from these experiments that overexpression of SoxC protein up to 30 h is not sufficient to initiate neurogenesis.

In sea urchin embryos, the first neurons to develop are in specific locations and they appear in a predictable sequence (Burke et al., 2014). We have made use of serotonergic neurons that appear along the dorsal margin of the APD and the neurons at the base of the postoral arms, which are predictable in the location and time at which they form. Without having a direct marker for cellular lineage, this positional information allows us to interpret data collected from successive stages in several embryos as representative of stages taken from the same embryo. The model has two to four cells in the APD beginning to express SoxC at the time of hatching. Subsequently, there is an interval during gastrulation in which these SoxC-expressing cells co-express Brn1/2/4 and in late gastrulae they differentiate as neurons, co-expressing ELAV, SynB and serotonin (Fig. 9). The distinctive location of postoral cells enables us to identify two to four cells in the oral ectoderm that express SoxC. In gastrulae these cells co-express Brn1/2/4 before downregulating SoxC, and in prism stage embryos there are cells in this location co-expressing ELAV and SynB. Each of the markers is expressed for an interval before subsequent markers are co-expressed. Thus, our model has neural progenitors expressing, in turn, SoxB2, SoxC, and Brn1/2/4, before projecting neurites and expressing ELAV and SynB (Fig. 9). This model is testable, and genetically based lineage-tracing data would provide a continuous record of changes of successive stages of individual cells.

SoxC is expressed in ectodermal cells that are in the location of postoral neurons and in animal pole cells in the location of serotonergic neurons at approximately the same time that neural progenitors are in S-phase of their last cell cycle. In most of the animal pole neural progenitors, the last S-phase occurs between 16 and 24 h and SoxC protein is first expressed at 20 h. In most postoral cells S-phase occurs between 16 and 28 h and SoxC protein is expressed at about 28 h. The fact we see division of cells expressing SoxC and that we are able to label cells expressing SoxC with a pulse of EdU indicates that cells expressing SoxC are not post-mitotic. A parsimonious interpretation of these observations is that SoxC is expressed coincident with neural progenitors initiating a final mitotic division.

Cells expressing SoxC are frequently observed as pairs of adjacent cells. It is reported that SoxC cells co-express Delta (Wei et al., 2016) and that treatment with DAPT, an inhibitor of Notch signaling, or a Delta morpholino results in the formation of clusters of neurons (Materna and Davidson, 2012; Wei et al., 2009; Yaguchi et al., 2011). This suggests that Notch signaling occurs subsequent to the final cell division and that the division is asymmetric, at least in the distribution of molecular components. A model that is consistent with these observations is that SoxC is expressed at the time of the transition of cells from mitotically active neural progenitors to undifferentiated post-mitotic neural precursors. Notch signaling then mediates a process of lateral inhibition in which one cell differentiates as a neuron and the other does not. A key question is the fate of the cell that does not differentiate into a neuron – it could remain as a self-renewing persistent neural progenitor, or it could differentiate to another type of cell. Cells expressing SoxC do not accompany differentiating neurons and there is no consistent patterning of cells expressing SoxC, suggesting that SoxC expression does not persist in the cell presumed to experience elevated Notch signaling. In addition, SoxC-expressing cells that arise within the ciliary band co-express Hnf6, a transcription factor associated with the differentiation of ciliated cells of the ciliary band. This suggests that early larval stages continue to recruit neural progenitors from adjacent ectoderm and ciliary band.

As the suppression of SoxC results in a marked reduction in the number of neurons and in some embryos no neurons form at all, we concluded that neurogenesis requires SoxC and that all neurons express SoxC during neurogenesis. However, it is not possible to conclude from our studies that all cells that express SoxC form neurons. The initial expression of SoxC throughout the APD before becoming restricted to a small number of cells suggests that cells expressing SoxC are not immediately committed to a neural precursor fate. Although the proteins we have examined have found use in other organisms as neuron-specific markers, none is expressed in that manner in sea urchins. SoxC is initially expressed in vegetal lineages and is downregulated as the cells differentiate. ELAV is expressed robustly in mesoderm-derived cells throughout early development. Brn1/2/4 protein can be detected in foregut endoderm as previously described (Cole and Arnone, 2009) and SoxB2 is broadly expressed throughout ectoderm. Thus, there are limitations to the use of these proteins as exclusive markers for neural progenitors.

Embryos that develop from eggs injected with full-length SoxC RNA, either fused to GFP or not, express the protein, which then accumulates in all nuclei. The protein can be detected until about 30 h, when it is reduced to background levels in all cells, except those that normally express SoxC. As neural precursors normally form in the APD and the oral ectoderm as early as 24 h, ectopic expression of SoxC by RNA injection has the potential to reveal whether it is sufficient to commit cells to neural precursors. As there is no increase in the number of cells differentiating as neurons through a range of doses that does not cause random abnormalities, we conclude that SoxC is necessary but not sufficient for the formation of neural precursors. The mechanism by which SoxC protein, but not control GFP, is dramatically reduced in early gastrulation merits further investigation.

Our data indicate that peripheral neurons arise in oral ectoderm, ciliary band and adjacent aboral ectoderm and that specification of the first neurons to differentiate is likely to occur in late cleavage. Previous studies have established that ectoderm differentiates in distinct domains in which TGFβ signaling initiates gene regulatory networks that establish cellular identity (Duboc et al., 2004, 2008; Duboc and Lepage, 2008). Nodal signaling during cleavage specifies oral ectoderm and initiates BMP signaling that establishes aboral ectoderm (Duboc et al., 2004; Saudemont et al., 2010; Ben-Tabou de-Leon et al., 2013). The ciliary band appears to form as a consequence of being protected from Nodal and BMP signaling by the antagonists Lefty and Chordin (Duboc et al., 2008; Bradham et al., 2009; Lapraz et al., 2009; Saudemont et al., 2010; Yaguchi et al., 2010). Oral and aboral ectoderm are fields of cells with bilateral subdomains in which distinct regulatory programs appear to operate (Chen et al., 2011; Li et al., 2012; Barsi et al., 2015). Our data indicate that neurons arise within the ciliary bands and adjacent regions of oral and aboral ectoderm. This is consistent with data that indicate Nodal and BMP signaling produce ectoderm that does not support neurogenesis (Yaguchi et al., 2010; Bradham et al., 2009) and suggests that the ciliary band is not the only region of ectoderm that is protected, at least in part, from TGFβ signaling.

None of the models of specification of oral, aboral or ciliary band ectoderm predicts the appearance of individual cells expressing a different regulatory program within these domains (Saudemont et al., 2010; Chen et al., 2011; Li et al., 2012; Barsi et al., 2015). Yet our data suggest that a gene regulatory network that specifies a neural fate controls the differentiation of individual neural progenitors within specific domains of ectoderm. FGF signaling and FGFR1 are necessary for the differentiation of neurons and are expressed such that they potentially initiate neurogenesis. A component of our hypothesis is that FGFR1 signaling diverts cells from oral ectoderm, ciliary band or aboral ectodermal fates by inducing expression of Six3, which initiates a neurogenic regulatory network. FGFR1 is expressed throughout the tissues that appear to give rise to neurons beginning in late cleavage stages (Lapraz et al., 2006). We envision a localized effect modulated by other factors, as all these cells do not become neural. FGFA is expressed in restricted regions of the blastoderm (Rottinger et al., 2008) and FGFR2 is expressed in mesodermal lineages. To date, analysis of FGF function in sea urchin embryos has focused on roles in patterning mesoderm (Lapraz et al., 2006; Rottinger et al., 2008; Adomako and Ettensohn, 2013). However, a role for FGF in initiating neurogenesis in other metazoans suggests that this is a conserved function (Rentzsch et al., 2008; Sinigaglia et al., 2013). In ascidians, FGF is essential for the specification of peripheral neurons that are arrayed laterally along the tail, a process that is initiated during cleavage (Ohtsuka et al., 2014). In vertebrates, FGF has an essential role in the specification of sensory placodes (Ariza-Cosano et al., 2015; Kengaku and Okamoto, 1993).

Deuterostomes share close kinship at a molecular level, yet body plans and neural organization are bafflingly diverse. Chordates restrict most neurogenesis to one region of the embryo and employ dedicated neural progenitors extensively in the generation of neurons. The other extreme appears in hemichordates, where neurogenesis occurs throughout the embryonic ectoderm and gives rise to a subepidermal network of nerves (Cunningham and Casey, 2014). Similarly, in asteroids, neural precursors arise throughout the ectoderm and migrate extensively before differentiating in the ciliary bands (Yankura et al., 2013; Nakajima et al., 2004). Broad comparisons within the deuterostomes, when echinoid neurogenesis is considered, indicate that the differences between ambulacrarians and chordates are less extreme than initial comparisons suggest, and that echinoid neurogenesis is distinct from that of hemichordates and asteroids (Table 1). Apparently, deuterostomes have in common the potential for restricting neurogenesis to specific domains of ectoderm, the ability to pattern neural progenitors to produce invariant positioning of neurons, and asymmetric division of neural progenitors. Detailed analyses of the neurogenic mechanisms of a complete range of taxa have the potential to distinguish features shared by common origin from features derived independently because of functional limits.

Gametes were collected from S. purpuratus induced to spawn by shaking. Sperm was diluted 1:1000 in sea water prior to fertilization and embryos were grown at 12-14°C. Eggs were prepared for microinjection as described previously (Krupke et al., 2014). Injection solutions contained water, 120 mM KCl, MASO or RNA and 1 µg/ml Rhodamine Green or Rhodamine B. Some injection preparations included 22% glycerol. MASOs were obtained from GeneTools, diluted to a suitable stock concentration (1 mM) and further diluted and filtered before filling injection needles. MASO sequences (5′-3′) and concentrations used: SoxC I, TTTTGAGGAACCATCTTGAAGTCAG (400 µM); SoxC II, CTTGAAGTCAGCATTCACTTTCGTG (1 mM); SoxC III, GAACCATCTTGAAGTCAGCATTCAC (400 nM); Six3-MO2, CCCCGGTCGCTGGGCGATGTTTCTG (1 mM) (Wei et al., 2009); FGFR1, TCCTCGGACAACGCGGCAGACTCAT (400 µM); SoxB2, GATATGACGGTCTCCATCTTACTCC (500 µM); Brn1/2/4, TCAATGAATCCTGGATCCCGAACAT (100 µM); StdControl1, CCTCTTACCTCAGTTACAATTTATA (300 µM); and Control2, TTTCAACCGTTTCCAAAGAACCAGG (300 µM). Knockdowns were confirmed with depletion of antibody binding where applicable (Fig. S1). Ectopic expression of SoxC was achieved using synthetic, capped mRNAs derived from the full-length S. purpuratus SoxC cDNA, or a construct in which the SoxC cDNA lacks a stop codon and is fused in frame to eGFP, cloned into a pCS2+ vector and transcribed using the SP6 mMessage mMachine Kit (Ambion). FGF inhibitor was diluted in DMSO and this stock further diluted into embryo cultures immediately after fertilization, and embryos were treated for the duration of the experiment. All experiments had at least three replicates.

Antibody production was as previously described (Krupke and Burke, 2014). Antigens were made using a pET28b(+) plasmid (Novagen) for expression of 6×His-tagged proteins. SoxC, SoxB2, ELAV and Brn1/2/4 constructs, encoding partial or full-length proteins, were produced with high-fidelity PCR from cDNA isolated from S. purpuratus embryos and cloned using the pGEM-T Easy system (Promega) (Fig. S1). Protein expression was induced in E. coli (BL21). Bacterial lysate was prepared and protein was solubilized in binding buffer (6 M guanidine HCl, 0.5 M NaCl, 100 mM Na₂HPO₄, 100 mM NaH₂PO₄, 10 mM imidazole, 10 mM Tris, 1 mM 2-mercaptoethanol, pH 8.0) prior to affinity purification by immobilized metal ion affinity chromatography (IMAC) using Chelex 100 resin (Bio-Rad). Following IMAC, proteins were further purified by size exclusion on Hi Load 16/60 Superdex 75 prep grade column (GE Healthcare) using the ÄKTA Prime Plus chromatography system (GE Healthcare). Purified protein in PBS was mixed 1:1 with Freund's complete adjuvant for immunization or with Freund's incomplete adjuvant for booster injections. Rats, rabbits or mice were immunized by subcutaneous injection of 100 µg antigen in 250 μl adjuvant, and booster injections were performed 21 days and 42 days after the initial immunization. Terminal bleed via cardiac puncture was performed after 52-56 days. Blood was incubated at 37°C for 45 min and then 4°C overnight. Samples were centrifuged at 1000 g and the serum collected.

Embryos were collected and fixed for 15 min in PEM fixative (Krupke et al., 2014). To determine the stages at which specific markers were expressed, embryos from a batch of eggs were fixed at intervals. Initially, intervals were 12 to 24 h, but to determine times at which expression began or ended, intervals of 2 h between fixations were used. Embryos were rinsed with PBS, blocked for 1 h in SuperBlock (Thermo) before incubation overnight in primary antibody diluted in SuperBlock. Embryos were rinsed three times with PBS and then Alexa Fluor-conjugated fluorescent secondary antibodies (Invitrogen) were used to visualize antibody labeling on a Zeiss LSM 700 confocal microscope. Imaging and analysis were conducted using ZEN software (Zeiss). Adobe Photoshop was used to prepare figures and adjust image contrast and brightness. Cell counts were made by sequential examination of individual images from stacks that focused through the entire thickness of the specimen. Statistics were calculated using Sigma Plot version 5.0 (Systat Software). At least three biological replicates of all experiments were performed. For details of primary and secondary antibodies used, see Table S1.

A 2 ml aliquot of embryos that were at the desired time to initiate labeling was incubated with 0.2 µl 10 mM stock solution of ethynyl deoxyuridine (EdU). After the interval of labeling, embryos were rinsed twice with sea water and incubated until 72 h, at which time they were fixed. Click-iT reaction mixture (Click-iT EdU Alexa 647, Life Technologies) was added directly to the primary antibody labeling solution and embryos were prepared as described above. For each labeling interval, five embryos were optically sectioned and labeled neuronal nuclei were counted manually by two observers. There are 15-20 neurons expressing SynB in early plutei (72 h).

We thank Zheng Wei, Lynne Angerer and Bob Angerer for providing reagents and unpublished data. Gary Wessel, Isabelle Peter and Eric Davidson generously provided reagents.