Developmental expression of COE across the Metazoa supports a conserved role in neuronal cell-type specification and mesodermal development
- First Online:
- Cite this article as:
- Jackson, D.J., Meyer, N.P., Seaver, E. et al. Dev Genes Evol (2010) 220: 221. doi:10.1007/s00427-010-0343-3
- 527 Downloads
The transcription factor COE (collier/olfactory-1/early B cell factor) is an unusual basic helix–loop–helix transcription factor as it lacks a basic domain and is maintained as a single copy gene in the genomes of all currently analysed non-vertebrate Metazoan genomes. Given the unique features of the COE gene, its proposed ancestral role in the specification of chemosensory neurons and the wealth of functional data from vertebrates and Drosophila, the evolutionary history of the COE gene can be readily investigated. We have examined the ways in which COE expression has diversified among the Metazoa by analysing its expression from representatives of four disparate invertebrate phyla: Ctenophora (Mnemiopsis leidyi); Mollusca (Haliotis asinina); Annelida (Capitella teleta and Chaetopterus) and Echinodermata (Strongylocentrotus purpuratus). In addition, we have studied COE function with knockdown experiments in S. purpuratus, which indicate that COE is likely to be involved in repressing serotonergic cell fate in the apical ganglion of dipleurula larvae. These analyses suggest that COE has played an important role in the evolution of ectodermally derived tissues (likely primarily nervous tissues) and mesodermally derived tissues. Our results provide a broad evolutionary foundation from which further studies aimed at the functional characterisation and evolution of COE can be investigated.
Understanding the molecular events that supported the morphological expansion of metazoan body plans is a fundamental aim of evo-devo studies. With the increasing availability of whole genome datasets from a variety of metazoan taxa (Sea Urchin Genome Sequencing Consortium, 2006; Putnam et al. 2007, 2008; Chapman et al. 2010; Srivastava et al. 2010), a growing challenge is to identify common (and truly homologous) genomic outputs and to assess the similarities and differences in their functions across a phylogenetically informative spread of representatives. Such studies hold the potential to provide deep insight into the features that common ancestors were likely to have possessed, and the events that accompanied the genesis and expansion of new clades of animals in the pre- and early Cambrian (Martindale 2005; Wheeler et al. 2009). The identification of deeply conserved developmental patterning mechanisms, such as the Hox code (or at least the precursors to such genes), support the notion that the metazoan ancestor possessed a developmental toolkit of genes that has been modified in lineage-specific ways to generate disparate body plans (Carroll et al. 2001; Larroux et al. 2006; Adamska et al. 2007). Related to the notion of a conserved developmental toolkit is the suggestion that cell types have specific ‘molecular fingerprints’ that can be used to identify cellular homologies across taxa (Arendt 2008). Explicitly, this concept interprets the expression of orthologous genes in similar tissues at similar times as an indication of cellular homology (Arendt 2005); however, it must be pointed out that when comparing the expression of homologous genes between cells, tissues or organs, structures that share homologous gene expression are not necessarily homologous in themselves; for example, orthologs of the gene Distal-less pattern many non-homologous structures across the Metazoa. This important point led Claus Nielsen to coin the term ‘homocracy’ in order to distinguish between non-homologous structures patterned by homologous genes (Nielsen and Martinez 2003). True homology of cells, tissues or organs is more convincingly demonstrated on the molecular level by the synexpression of a demonstrable gene regulatory network (GRN), as such a network is presumably less likely to be co-opted to a non-homologous structure than an individual gene (Davidson 2006). A clear example of this was provided by Dunn et al. (2007) to demonstrate that apical sensory structures of larval urchins and gastropods are unlikely to share homology. They conducted a GRN analysis to test the hypothesis that the apical tufts of larval urchins and gastropods were homologous. In many marine invertebrates, the apical end of the larval body plan consists of what is termed the apical organ, which generally consists of two components: the ciliated apical tuft and the apical ganglion. Because the apical tufts of urchins and gastropods use different GRNs for their specification, Dunn et al. (2007) argued that they were the result of convergent evolution, at least in these two taxa, sometime in the late Precambrian (Peterson 2005).
The transcription factor COE was first isolated from vertebrate models and ascribed a role in regulating the expression of olfactory (Wang and Reed 1993) and immune cell fates (Hagman et al. 1993; Travis et al. 1993). A Drosophila homolog of COE was isolated shortly after the vertebrate gene (Crozatier et al. 1996) and has been found to play a variety of roles, including head segmentation (Crozatier et al. 1999), wing patterning (Crozatier et al. 2002), muscle specification (Crozatier and Vincent 1999) and immune cell specification (Crozatier et al. 2004). Vertebrate paralogs of COE have also been reported to play roles in limb development (Mella et al. 2004), and a recent analysis of ascidian development suggests that the last common ancestor of the Chordata may have used COE to specify pharyngeal mesoderm development (Stolfi et al. 2010). Pang et al. (2004) reported that COE expression is restricted to the apical organ (the ciliated and innervated structure at the aboral end of the developing planula larva) of the cnidarian Nematostella vectensis, suggesting that the ancestral eumetazoan function of COE was to specify chemosensory neurons. This last study provides an evo-devo foundation from which the interpretation of various studies demonstrating a role for COE in vertebrate (Wang and Reed 1993; Garel et al. 1997; Dubois et al. 1998; Pozzoli et al. 2001; Corradi et al. 2003; Garcia-Dominguez et al. 2003) and invertebrate (Prasad et al. 1998; Baumgardt et al. 2007) neurogenesis can be placed.
These previous studies, primarily focused on vertebrate models, Drosophila and C. elegans, indicate that COE is highly pleiotropic and plays important roles in the specification of cell types originating from all three embryonic germ layers (Liberg et al. 2002). Here, we have isolated the COE homologues from a variety of invertebrate metazoans and investigated their expression patterns with the goal of uncovering conserved and divergent features of COE expression. COE displays complex expression patterns in all taxa investigated; however, broad commonalities and differences can be identified. Our results suggest that COE may be part of a molecular fingerprint specific to a subset of ectodermally derived cell types, possibly chemosensory, non-serotonergic and located in the anterior ciliated structures. We stress that this proposed cell type homology neither requires nor implies homology of apical organs; apical ganglia may consist of homologous cell types, but apical tufts could be the products of convergent evolution (see “Results and discussion”). Early mesodermal expression of COE in ctenophores, protostomes and vertebrates suggests another shared function may relate to the development of the haematopoietic system and/or connective tissue differentiation (Hagman et al. 1993; Akerblad et al. 2002).
Materials and methods
Gene isolation, whole mount in situ hybridization and MASO injections
For all organisms, previously published protocols for RNA extraction, gene isolation and RACE PCR were employed (for example see Jackson et al. 2005; Matus et al. 2006; Dunn et al. 2007; Thamm and Seaver 2008). Where genomic or EST sequences were available, these were first searched and used to design gene-specific primers for RACE PCR. Organism-specific methods for whole mount in situ hybridizations (WMISH) were employed for the ctenophore Mnemiopsis leidyi (Pang and Martindale 2008), the tropical abalone Haliotis asinina (Jackson et al. 2007), the annelids Capitella teleta (Thamm and Seaver 2008; Blake et al. 2009), previously Capitella sp. I (Blake et al. 2009) and Chaetopterus (Boyle and Seaver 2010), and the sea urchin Strongylocentrotus purpuratus (Dunn et al. 2007). Newly obtained COE sequences have been deposited in GenBank HQ529594 - HQ529597. SpNK2.1 morpholino anti-sense oligonucleotide (MASO) and SpCOE-MASO injections were performed as described in Takacs et al. (2004). The sequences of the SpCOE-MASO and the standard control MASO are, respectively 5’-GATATGACGGTCTCCATCTTACTCC-3’ and 5’-CCTCTTACCTCAGTTACAATTTATA-3’. All primer sequences are available from the authors upon request.
Immunofluorescence and imaging
S. purpuratus embryos were fixed in 4% paraformaldehyde in filtered sea water for 15 min at room temperature and then post-fixed for 1 min in −20°C methanol. Fixed embryos were rinsed in phosphate-buffered saline (PBS) then incubated in PBS containing 5% normal lamb serum and 0.01% Tween 20 for 30 min at room temperature to reduce non-specific binding. Specimens were incubated in primary antibody for 16 h at 4°C with rabbit anti-serotonin (Chemicon, CA) (1:500) and mouse or rat anti-synaptogamin B (1:800) according to Nakajima et al. (2004). Embryos were rinsed in PBS and incubated in goat anti-rabbit Alexa 488 (1:800) and goat anti-rat Alexa 568 (1:1500) (Molecular Probes). Preparations were examined and photographed with a Leica DM 6000 B epifluorescence microscope (Leica Microsystems) and Hamamatsu Orca-ER camera (Hamamatsu Photonics) using Openlab 4.0.4. To determine the number of immunoreactive cells present, a double-blinded procedure was utilised such that näive observers examined specimens on a monitor focusing through the specimen to identify all immunoreactive cells. Statistical analyses were conducted using GraphPad Prism (4.03).
COE homologues from a variety of metazoan taxa were retrieved from public databases following BLAST searches against GenBank and from whole genome draft assemblies of the placozoan Trichoplax adhaerens (see Supplementary Online Material for all accession numbers and genomic locations). These sequences were combined with full-length fragments isolated by RACE PCR and aligned using MUSCLE (Edgar 2004). Following manual adjustment (especially of the HLH region), the alignment was used to schematically represent conserved COE domains and was then processed by GBlocks (Talavera and Castresana 2007) to retrieve highly conserved positions for phylogenetic analysis. A Bayesian analysis of the resulting dataset was carried out with the parallel version of MrBayes (MrBayes v. 3.1.2p) on a Linux cluster at the Gesellschaft für wissenschaftliche Datenverarbeitung, Göttingen. The following settings were used: prset aamodelpr = mixed nchains = 4 nruns = 4 temp = 0.2 printfreq = 1000 samplefreq = 1000 ngen = 1800000. A burnin value of 25% was set to exclude non-convergent tree topologies.
Results and discussion
COE spatial expression
Besides mesodermal expression in H. asinina, C. teleta and Chaetopterus, COE transcripts also are detected in the developing central nervous system and in other putative neural elements. In the mollusc H. asinina, COE is expressed in the supra- and subesophageal ganglia (labelled as spg and sbg in Fig. 6c, d, g, h, k, l) while in Chaetopterus and C. teleta, COE is expressed in the developing brain (labelled as br in Figs. 5, 6 and 7) and ventral nerve cord (labelled as vnc in Figs. 5, 6 and 7). These results indicate that COE likely plays a role in neural specification in these animals. Outside of the developing central nervous system, COE is detected in the apical tuft of H. asinina (labelled as at in Fig. 6). In Chaetopterus, ChCOE is expressed in dorso–apical ectoderm (labelled as de in Fig. 7) but not in apical tuft cells (labelled with a red asterisk in Fig. 7). In C. teleta, CtCOE is expressed in a small number of cells in the central, anterior ectoderm (Fig. 4h, arrows in Fig. 4j, n, n’, p). Among these three lophotrochozoans, C. teleta is the only larva that lacks an apical tuft. In addition to apical ectodermal expression, strong COE expression is detected in the posterior ectodermal cells of H. asinina (pc in Fig. 6a, e, m–o) and Chaetopterus (Fig. 7a–h, m–o), but not in C. teleta (Fig. 4a–j, r). Interestingly, in both H. asinina (Hinman et al. 2003) and Chaetopterus (NPM, unpublished data), but not in C. teleta (NPM, unpublished data), neurons are present in this posterior region. Thus, the posterior ectodermal domain of COE expression may also be involved in neural specification.
Sea urchin expression
In the sea urchin, SpCOE is first detected at around 24 h post-fertilisation with transcripts diffusely localised to the apical end of the embryo (data not shown). By 44 h, SpCOE is expressed in three clear loci in the apical organ (Fig. 8a). SpCOE expression is not detected around the blastopore (Fig. 8a) as in M. leidyi, nor is it detected in mesodermal cells as in M. leidyi and the lophotrochozoans analysed.
In summary, these and previously reported results demonstrate both similarities and differences in COE expression patterns from various metazoan taxa. We detected a clear COE expression in the mesoderm of the ctenophore (Fig. 3a–g) and all three lophotrochozoan species: H. asinina (Fig. 6a, b, j); Chaetopterus (Fig. 7a, b, e, f, i, j); C. teleta (Fig. 4a, b, e, f) but not in S. purpuratus (Fig. 8a). Furthermore, COE is expressed in the developing central nervous system of all three lophotrochozoan species examined (Fig. 4, 5, 6 and 7) and in putative neural elements in the apical organs of both H. asinina and S. purpuratus (Figs. 6a–c; 8a) but not Chaetopterus (Fig. 7a–g, k, l).
Functional characterization of COE in S. purpuratus
To determine whether SpCOE functions in neuronal and/or ciliary tuft specification in the apical ganglion of sea urchin larvae, we knocked down SpCOE translation with morpholino anti-sense oligonucleotides (MASOs) and assayed the resulting phenotypes by WMISH for SpCOE, SpNK2.1, and the ciliary marker Sptektin3 and by immunofluorescence labelling of serotonin and synaptogamin B. Previously, Dunn et al. (2007) showed that SpNK2.1 is upstream of several ciliary markers and that abrogation of SpNK2.1 function resulted in the absence of ciliary marker transcripts (α2tubulin, radial spoke 3, tektin3, RSH p63 and dynein p33) and the apical tuft itself. Here, we show that injection of an SpCOE-MASO had no affect on its own expression (Fig. 8a; cf. 8b). This suggests that SpCOE is not autoregulated in the sea urchin apical plate, unlike Drosophila-COE which autoregulates after gastrulation in cells of the posterior intercalary and anterior mandibular segments (Crozatier et al. 1999). Furthermore, like SpNK2.1 (Fig. 8f; cf. 8d), SpCOE does not regulate SpNK2.1 expression (Fig. 8e; cf. 8d). Conversely, injection of an SpNK2.1-MASO resulted in the abrogation of SpCOE transcripts (Fig. 8a; cf. 8c) suggesting that SpCOE acts downstream of SpNK2.1. Unlike the SpNK2.1-MASO phenotype (Dunn et al. 2007), knockdown of COE resulted in the apparently normal expression of the cilia marker Sptektin3 (Fig. 8h; cf. 8g), the presence of an apparently normal apical tuft and larvae with normal swimming behaviour.
The expression data from lophotrochozoans and the functional results in S. purpuratus suggest that COE may have functioned ancestrally to repress the specification of serotonergic neurons. In H. asinina, serotonin-positive cells are absent from the posterior end of the larvae (Hinman et al. 2003) where strong COE expression is detected. FMRF+ cells are present in this region of Aplysia californica (Dickinson et al. 2000) Ilyanassa obsoleta (Dickinson and Croll 2003) and H. asinina (Cummins, Croll and Degnan unpublished data) trochophores. The expression of the pan-neuronal marker HasElav (Perrone-Bizzozero and Bolognani 2002; Benito-Gutiérrez et al. 2005; Marlow et al. 2009) in this region of H. asinina trochophores (Fig. 6p) further supports the hypothesis that non-serotonergic neurons will be born here. In Chaetopterus, a single, large ‘horn-shaped’ posterior serotonergic neuron is visible by the end of gastrulation (~9 hpf); however, there are likely other, non-serotonergic neurons that form in this region at similar times (NPM, unpublished data). C. teleta larvae do not posses posterior serotonergic neurons and do not express COE posteriorly. It will be interesting to functionally determine whether COE represses serotonergic cell fate in annelids and molluscs as it does in S. purpuratus.
Conserved mesodermal expression of COE
A principle feature of the COE expression patterns we have described here is expression in mesodermal tissues in ctenophore, annelid and molluscan representatives. While the fates of these tissues are not yet known, it is of note that one of the first two papers to isolate and characterise the function of COE (from a vertebrate) was to describe its role in B cell specification, a mesoderm derivative and a type of immune cell primarily responsible for antibody production (Hagman et al. 1993). Recently, a role for COE in blood cell homeostasis was also reported for Drosophila where it is involved in the differentiation of immune cells (lamellocytes) following parasitisation (Crozatier et al. 2004; Krzemien et al. 2007). Activity in haematopoietic precursors is therefore likely to be a deeply conserved feature of COE, with the possibility of it having had a specific role in the evolution of the metazoan immune system (Hartenstein 2006). If COE also specifies haematopoietic cells in spiralians, then this function likely arose shortly after the split of cnidarians from the bilateria. Alternatively, COE may function in distinct mesodermal cell types in spiralians. Interpreting the expression of COE in early mesodermal derivatives of M. leidyi is challenging because the phylogenetic position of ctenophores is currently unstable (Dunn et al. 2008; Pick et al. 2010), and ctenophores do not have a circulatory system (Brusca and Brusca 2002).
A COE cell type molecular fingerprint in ectodermally derived apical ciliated structures
Most marine invertebrates develop via a planktonic larval period (here defined as indirect development) during which either active (by means of swimming) or passive (current driven) dispersal is achieved (Brusca and Brusca 2002). Prior to settlement and metamorphosis, many species must identify a suitable substrate upon which to attach (Jackson et al. 2002). It is believed that apical organs are involved in recognising metamorphosis-inducing cues via chemoreception (Hadfield et al. 2000; Leise et al. 2001; Thavaradhara and Leise 2001). Elucidating the evolutionary origins of these structures across a variety of metazoan phyla has the potential to address questions of larval homology and more broadly whether the ancestor of the Metazoa developed indirectly or directly. Previously, Dunn et al. (2007) conducted a comparative GRN analysis of two transcription factors (NK2.1 and HNF6) and their regulation of genes required for apical tuft ciliation in the sea urchin S. purpuratus and the abalone Haliotis rufescens. They found significant differences in the synexpression of these genes and concluded that the apical tufts of urchins and abalone are specified in different ways and are therefore unlikely to have the same evolutionary origins. Given its clear ecological role and restriction to primary marine larvae, the apical tuft is an appropriate structure to evaluate the homology of the apical organ; however, a difficulty with the Dunn et al. (2007) study is that although the dissimilarity between the trochophore (Haliotis) and dipleurula (S. purpuratus) was clear, the primary gene under investigation, NK2.1, does not have a clear homolog in cnidarians (Ryan et al. 2006), and hence evolutionary polarity could not be unambiguously established. Expression of COE in the apical organ of cnidarians could suggest potential homology and indeed a parsimonious interpretation of our COE data per se would be that the apical ganglia of urchins, abalone and cnidarians are homologous; however an alternative hypothesis that COE functions in the apical ganglion to specify chemosensory neurons and that this cell type was independently recruited to the apical region in multiple metazoan lineages, must also be considered. Such a scenario would fall under Nielsen’s definition of homocracy—expression of homologous genes in non-homologous structures. While our COE data alone cannot distinguish between these hypotheses, our functional data clearly indicate that at least in the sea urchin it is not involved in apical tuft specification, but in the specification of proper neuronal cell fate within the ganglion itself.
We anticipate that more detailed analyses of COE regulation and function and the inclusion of more metazoan representatives that posses larval apical tufts, will significantly contribute to the resolution of the larval homology problem. Homologous genomic outputs such as COE, which represent the genetic material upon which evolution acted to diversify cell types during the early history of metazoan evolution, are well suited for such studies. Our data has generated testable hypotheses and will serve as a platform for future studies aimed at resolving the evolutionary histories of larval and adult metazoan body plans. Our findings also highlight the need for functional assays to be developed for a wider range of metazoan model systems, which will complement the flood of sequence data the field of evo-devo is now experiencing.
This work was supported by the Australian Research Council through funding to BMD and DJJ, and Deutsche Forschungsgemeinschaft funding to DJJ and the Courant Research Centre for Geobiology, Göttingen, through the German Excellence Initiative. RDB is supported by a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada. KJP and ECS are supported by the National Science Foundation (IOS09-23754 to ECS).
All experiments described here comply with the current laws and regulations of the respective countries they were conducted in.
Conflict of interest
The authors declare no conflicts of interest.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.