The evolution of ependymin-related proteins
Ependymins were originally defined as fish-specific secreted glycoproteins involved in central nervous system plasticity and memory formation. Subsequent research revealed that these proteins represent a fish-specific lineage of a larger ependymin-related protein family (EPDRs). EPDRs have now been identified in a number of bilaterian animals and have been implicated in diverse non-neural functions. The recent discoveries of putative EPDRs in unicellular holozoans and an expanded EPDR family with potential roles in conspecific communication in crown-of-thorns starfish suggest that the distribution and diversity of EPDRs is significantly broader than currently understood.
We undertook a systematic survey to determine the distribution and evolution of EPDRs in eukaryotes. In addition to Bilateria, EPDR genes were identified in Cnidaria, Placozoa, Porifera, Choanoflagellatea, Filasterea, Apusozoa, Amoebozoa, Charophyta and Percolozoa, and tentatively in Cercozoa and the orphan group Malawimonadidae. EPDRs appear to be absent from prokaryotes and many eukaryote groups including ecdysozoans, fungi, stramenopiles, alveolates, haptistans and cryptistans. The EPDR family can be divided into two major clades and has undergone lineage-specific expansions in a number of metazoan lineages, including in poriferans, molluscs and cephalochordates. Variation in a core set of conserved residues in EPDRs reveals the presence of three distinct protein types; however, 3D modelling predicts overall protein structures to be similar.
Our results reveal an early eukaryotic origin of the EPDR gene family and a dynamic pattern of gene duplication and gene loss in animals. This research provides a phylogenetic framework for the analysis of the functional evolution of this gene family.
KeywordsEpendymin EPDRs MERPs Gene duplication Gene loss
Expressed sequence tags
hidden markov model
mammalian ependymin-related protein
Ependymins are secreted glycoproteins comprised of approximately 200 amino acids with four conserved cysteines. They were first discovered to be associated with the ependyma of the brain in fish where they appear to be required for the formation of long-term memory and neuronal regeneration [1, 2, 3]. Although long believed to be fish-specific, subsequent discoveries of ependymin-related proteins (EPDRs) in mammals [named ‘mammalian ependymin-related proteins’ (MERPS)  or in humans 'upregulated in colorectal cancer gene-1' (UCC1) ], amphibians  and echinoderms  demonstrated that the distribution of the gene family was much broader than originally thought.
These studies also revealed that EPDRs might perform roles outside the central nervous system. Although the specific functions of MERPs are unclear, they are highly expressed in brain, heart, skeletal muscle, kidney and hematopoetic cells [4, 7], and in some tumours . UCC1 is involved in Dupuytren’s disease, a connective tissue disorder, in which it affects the collagen contractility of fibroblasts . In sea cucumbers, EPDR genes are expressed in several tissues and are upregulated during regeneration of the eviscerated gut [6, 9]. The expression of these EPDR genes in non-brain tissues and the implication of their involvement in diverse processes indicated that the functional diversity of this protein family had been underestimated.
It was within this context that Suárez-Castillo and García-Arrarás conducted the first systematic analysis of EPDRs in animals, aided by the growing availability of expressed sequence tag (EST) projects . Not only did they detect EPDRs in non-deuterostome bilaterians, they also identified additional piscine EPDRs that were expressed in non-brain tissues. Phylogenetic analyses of this newly-collated dataset of EPDR sequences revealed the presence of four major clades. The first group, the ‘Fish Brain’ group, contained the original, brain tissue-specific ependymin sequences. Sister to this group was a second fish-specific group, named ‘Fish Tj’, containing other fish EPDRs that were expressed in non-brain tissues. The third major group contained the previously identified ‘MERPS’, as well as echinoderm, amphibian, bird and fish representatives. Finally, the analysis revealed the likely presence of a fourth ‘basal’ group of EPDRs that contained sequences from molluscs, amphioxus, and ascidians. This analysis and the presence of fish sequences in three of the four clades led the authors to propose that gene duplication, likely from an ancestral ‘basal’ EPDR, followed by degeneration and complementation of gene sequences (the DDC model ) resulted in the presence of multiple, functionally-diverse EPDR genes in vertebrates. The placement of some sequences, however, was not entirely congruent with this scenario; namely, the position of echinoderm sequences within the MERP clade, and the grouping of oyster (bivalve mollusc) sequences with that of amphioxus rather than with the other molluscs, which, themselves, appeared as a sister group to the poorly supported ‘basal’ EPDRs.
Since the publication of this analysis, potential EPDRs have been reported in non-metazoan unicellular and multicellular unikont (Holozoa, Fungi, Apusozoa and Amoebozoa) relatives [12, 13], suggesting EPDRs are more ancient than previously thought. A large duplication of EPDR genes has also recently been reported in echinoderms . The crown-of-thorns starfish has a large EPDR gene family expansion with at least some of the EPDR proteins being secreted outside the animal during aggregation or in response to the presence of a predator, suggesting a role in conspecific signalling . Phylogenetic analysis of echinoderm EPDRs revealed that sea cucumbers and other echinoderms have multiple EPDR genes, suggesting they may perform additional functions in these species. Outside deuterostomes, a number of transcriptome and proteome-based studies, predominately in molluscs, have linked EPDRs with a variety of functions. ‘Sometsuke’, an EPDR described in the abalone (a gastropod mollusc), has a role in shell biomineralisation and pigmentation [15, 16], and EPDRs have also been found in mantle transcriptomes and shell proteomes in other mollusc species [17, 18, 19]. In addition, EPDRs have been implicated in gastropod larval development , are upregulated in response to toxins and metal pollution in bivalves [21, 22, 23, 24], and are differentially regulated during pathogen challenge trials in bivalves [25, 26, 27] and in response to environmental stressors in both bivalves and gastropods [28, 29, 30]. Together, these studies are consistent with EPDRs having a diversity of functions, some of which appear to be lineage-specific.
To further understand the structural and functional evolution of the EPDR family we performed a systematic survey for EPDRs in eukaryotes and conducted phylogenetic analyses to investigate their relationships. We found further support for this gene family evolving before the origin of Metazoa. The EDPRs can be divided into two major clades that can be distinguished by different patterns of conserved residues, notably, different numbers of cysteine residues. This is consistent with these clades being functionally distinct. Mapping these EPDR clades onto an organismal phylogeny suggests these cysteine profiles can arise de novo within lineages, and that convergent evolution is possible. The absence of EPDR genes in many eukaryote clades demonstrates that the gene family has been lost multiple times in protist and metazoan lineages. In contrast, the gene family has undergone extensive independent duplication and divergence within several metazoan groups.
Phylogenetic distribution of EPDRs
The ependymin gene family is represented in the Pfam database (PF00811) , where an estimation of its distribution is provided based upon searches of reference proteomes from the UniProt database . These data suggest that EPDRs are present in metazoans and some unicellular eukaryotic proteomes. However, no EPDRs were reported from bacterial, viral or archaeal reference proteomes (numbering 82,518, 82,616, and 860 proteomes, respectively). We therefore restricted our analysis to eukaryotic taxa.
Putative EPDR transcripts were also detected in the cercozoan Paracercomonas marina (Rhizaria) and Malawimonas jakobiformis (Malawimonadidae), although these designations must be treated with caution until the possibility of contamination can be discounted. The latter most probably accounts for the discovery of an EPDR in a root nodule transcriptome of the tracheophyte Aeschynomene indica given that this sample likely contained an assortment of other organisms, and that EPDR sequences were not found in other tracheophytes including those with whole genome data. This sequence was not investigated further in our analysis. It must be noted that sequence availability for many eukaryotic groups is poor, and sequenced organisms do not represent the entire diversity of these clades . For this reason, it is likely that EPDRs will be detected from additional groups as more data become available. This caveat aside, no EPDRs were detected in a number of sequence-rich groups, including Fungi, Tracheophyta, Chlorophyta, Rhodophyta, Myzozoa, and Euglenozoa.
Our survey also greatly extended the known EPDR gene complement in metazoans. The number of EPDR genes detected in genomes or transcriptomes varies greatly between species, with large expansions (15 EPDR genes or more) detected in the cephalochordate Branchiostoma belcheri, the asteroid echinoderms Patiria miniata and Acanthaster planci, the gastropod molluscs Aplysia californica, Lottia gigantea and Haliotis asinina, the bivalve mollusc Crassostrea gigas, the brachiopod Lingula anatina, and the sponge Amphimedon queenslandica (Additional file 1). Other groups appear to have lost the gene family entirely. For example, no EPDRs could be found in ctenophores or in ecdysozoans.
Determining the evolutionary history of EPDRs within clade 2, which exhibit all three cysteine patterns, is similarly problematic. Within this clade the type 1 profile is restricted to two amoebozoan sequences (the remaining amoebozoan sequences are type 2, Fig. 3 and Additional file 6). Likewise, the type 3 profile is not common, being restricted to charophyte (Archaeplastida) and excavate taxa, as well as to a small subset of molluscan sequences. Given that these molluscan sequences are the only example of the type 3 profile in unikonts, and that the clades containing them are nested within type 2 containing clades (mostly with good support, Fig. 3 and Additional file 6), it is likely that this pattern arose convergently in molluscs from a type 2 ancestral gene. A similar situation exists in the charophyte taxa, where all three EPDR sequences form a monophyletic group (in maximum likelihood but not Bayesian analyses), but two of the sequences are type 3 and the third is type 2 (Fig. 3). Unless this discrepancy is due to contamination, this may represent an additional case of convergent evolution.
EPDR duplications in metazoan lineages
The phylogenetic analysis of EPDR protein sequences revealed independent large gene family expansions in a number of metazoan taxa, as outlined above. Seventeen EPDR genes were identified from the demosponge Amphimedon queenslandica genome and formed a well-supported group within clade 1 in the phylogenetic analysis (Fig. 3, Additional file 6 clade A). Only two EPDRs were identified from ESTs of a second demosponge, Suberites domuncula. These sequences were also placed within clade 1, but clustered with two of the 8 EPDRs from the calcareous sponge Sycon ciliatum (Additional file 6 clade B). A fourth sponge species, the homoscleromorph Oscarella carmela, has both clade 1 and clade 2 EPDR representatives (Additional file 6 clades C, D). One of the clade 1 sequences clusters with a S. ciliatum sequence within the MERPs, while the positions of the remaining clade 1 sequences are poorly resolved. These results indicate that at least one ancestral clade 1 gene was present in the sponge ancestor, and strongly suggest that the A. queenslandica genes are the result of taxon-specific duplications, possibly functioning in a different manner to those in other sponge classes.
Numerous EPDR duplications have also occurred in lophotrochozoan lineages. Molluscan EPDR representatives form numerous clades throughout the tree, in both clade 1 and clade 2. Some groups contain representatives from most of the surveyed gastropod and bivalve taxa, indicating that some diversification occurred prior to their divergence (e.g., clade E, Additional file 6). Others are restricted to bivalves, gastropods, or to individual species (e.g., clade F, Additional file 6), indicating post-divergence diversification. All cephalopod sequences fall into a single group within clade 2 and do not cluster with other molluscan sequences, reflecting different patterns of evolution within this class (Additional file 6 clade G). Another large expansion has occurred within the brachiopod Lingula anatina, and representative EPDRs from this species are likewise distributed broadly throughout the tree, generally as well-supported clusters of 2–5 sequences. Both molluscs and brachiopods have representative sequences placed within the extended MERPs, elsewhere in clade 1, and in various (poorly-supported) groups within clade 2, indicating the presence of at least two clade 1 genes and one (probably more) clade 2 gene in the last common ancestor of lophotrochozoans.
A number of expansions have also occurred within deuterostome lineages. As observed previously , the EPDR repertoire of Echinodermata is the result of multiple duplication events throughout the evolution of this lineage, featuring large duplications of clade 2 EPDRs, particularly within asteroid lineages (e.g., clade H, Additional file 6). Within cephalochordates, two separate gene expansions appear to have occurred within the Branchiostoma lineage. With the exception of one divergent sequence, the EPDR genes are split between two major clades, one that falls within clade 1 (Additional file 6 clade I), and the other within clade 2 (Additional file 6 clade J, not monophyletic). A number of close 1:1 orthologues can be observed between the species; therefore these expansions took place, at least in part, prior to speciation. No MERP candidate can be found within the two Branchiostoma species; given the presence of echinoderm and vertebrate sequences within this group, we can infer that this is due to loss. Based on the distribution of deuterostome EPDRs, it can be inferred that the last common ancestor possessed one MERP, a second clade 1 EPDR, and at least one clade 2 EPDR gene.
Duplicated genes are found in clusters within the genomes of a number of the investigated species, including A. queenslandica, B. belcheri, and A. planci (Additional file 8), and clustered genes tended to group together in the phylogenetic analysis (e.g., A. planci genes within clade K, Additional file 6), suggesting that expansion of this gene family has occurred via tandem duplication in several lineages. Intra-class synteny analysis was performed between species with whole genome information where possible to determine whether EPDR genes are conserved within larger gene clusters, and whether this information could be used to aid in evolutionary reconstructions. No synteny was observed among any of the species compared (data not shown).
Sequence characteristics of EPDR proteins
Mapping the conserved cysteine residues on to each predicted structure indicates that some are likely to participate in cysteine bonds. Although I-TASSER does not explicitly model disulphide bridges, close localisation of cysteine residues may indicate that bonds occur. In cysteine profile 1, disulphide bridges may form between cysteines 94 and 186, and also between cysteines 23 and 148 (Fig. 4e, h). The former bond would anchor the C-terminal tail of the protein, whereas the latter would occur at the bottom of the protein pocket. Cysteine profile 2 contains two additional cysteines at position 24 (also at the bottom of the pocket), and at position 107, which appears to be within the pocket itself (Fig. 4f, i). There is a larger distance between these additional cysteines, therefore they may not be involved in cysteine bond formation. Profile 3 is similar to profile 1, except that it has a cysteine at position 107 (within the pocket) rather than at position 148 (at the bottom of the pocket) (Fig. 4g, j). It is unlikely that these two residues would form a bond, based on their predicted localisations. Other, non-cysteine residues that are highly conserved within EPDRs may also occupy important locations within the protein. One example is the highly conserved proline residue at position 150; this proline is situated in the loop that forms the floor of the pocket, and may influence ligand binding (yellow residue in Fig. 4e-j), although this residue is not absolutely conserved (i.e. there is no conserved proline in this position in D. rerio 134034, Fig. 4h; despite other ‘Fish Brain’ ependymins possessing the conserved residue).
Previous authors have reported the presence of conserved glycosylation sites 3 and 27 amino acids upstream of the second conserved cysteine in fish ependymin proteins, and many of their functional properties have been attributed to their N-linked carbohydrates [2, 10, 34, 37]. Putative glycosylation sites have also been identified in mammalian and echinoderm EPDRs; however, the positions of these sites are not conserved with those of piscine ependymins [6, 7]. We found predicted N-glycosylation sites in all but 30 of the EPDR sequences surveyed (Additional file 6, sequences indicated by a star); five of these were from ESTs that appeared incomplete. All taxa possessing EPDR members that do not have predicted N-glycosylation sites have additional EPDR genes that do, except for D. discoideum. Although the number and position of predicted glycosylation sites is not conserved (red regions in Fig. 4e-j), their presence in the majority of EPDR proteins investigated suggests that N-linked carbohydrates are functionally important.
The discovery of EPDRs in non-metazoan eukaryote taxa reveals that the gene family arose much earlier than previously thought. Given that relationships among major eukaryotic groups are still uncertain [33, 38], it is difficult to reconstruct the earliest evolutionary history of EPDR genes, i.e., whether the patchy distribution of EPDRs in these groups is the result of gene loss or the evolution of the gene family after the divergence of the major clades. In addition, the presence of single EPDR sequences in transcriptomes from single rhizarian and malawimonad species is problematic; it is unclear if they are contamination. This distinction is critical for reconstruction of the evolutionary history of this gene family, especially given that they possess sequence characters otherwise found only in unikont taxa. Horizontal gene transfer could have occurred, and careful analysis of whole genome data from these protist lineages will be required when they become available.
The phylogenetic analysis presented here, although weakly supported, is congruent with that conducted by Suárez-Castillo and García-Arrarás . Our larger dataset, facilitated by increased availability of sequences, reveals that the discrepancies detected in this earlier study (particularly within the ‘basal’ group) are the result of incomplete sampling. Clade 2 sequences were not included by Suárez-Castillo and García-Arrarás in their analysis. This may reflect the absence of these sequences in the NCBI database at the time and/or differences in search strategy (i.e., HMM searches as performed here, in contrast to BLAST performed in the previous study). We argue that the overall similarity in sequence length, the presence of shared highly conserved residues in comparable positions (including the three ultra-conserved cysteines), and the similar predicted 3D structures suggest that clade 1 and clade 2 EPDRs arose from a single ancestral gene.
Our analysis also reveals the dynamic nature of evolution of EPDRs in metazoans. Independent gene duplication events have occurred in numerous lineages, including sponges, molluscs, echinoderms and cephalochordates, and multiple tandem gene duplication events appear to have occurred at different points during molluscan and echinoderm evolution.
The underlying reason for these duplication events, and for the retention of multiple EPDR genes, is unknown. It is possible that these gene duplicates have undergone divergence and functional complementation (DDC), as has been suggested for vertebrate EPDRs , although confirmation awaits better understanding of their functions. Equally curious is the apparent loss of the EPDR gene family entirely from ecdysozoans and ctenophores. These losses indicate that EPDR genes are not performing functions that are essential for the survival of eukaryotic organisms, or that these functions can be compensated for by other mechanisms.
Based on prior reports in the literature, the range of functions performed by EPDRs is extraordinarily diverse. This may be attributable to particular biochemical or structural properties of EPDRs that are applicable in many different functional contexts. For example, the involvement of these proteins in fish memory consolidation , fish optic nerve regeneration , sea cucumber intestinal regeneration , and human fibroblast contractility  may be due to the interaction of EPDRs with components of the extracellular matrix (including collagen ) in a calcium-dependent manner . Similar properties may underlie the co-option of EPDRs into the molluscan shell  (McDougall and Degnan, unpublished). Additionally, structural features of EPDRs can activate signalling pathways, as peptide fragments with different amino acid sequences that originate from the same region of fish and human EPDRs can activate AP1 and the JNK signalling pathway in cell culture via an unknown mechanism [42, 43]. A signalling role is also likely for the crown-of-thorns starfish A. planci EPDRs that are released into the water column by aggregating or alarmed animals . Therefore, EPDRs may act as ligands themselves, in addition to being able to bind other ligands within their central pocket. Additional biochemical properties found for fish ependymins are also likely to be important for EPDR function, including the ability to form disulphide-bound dimers  and higher molecular weight aggregates [45, 46], the presence of N-linked carbohydrates that confer Ca2+ or cell-cell/cell-matrix binding ability [2, 3], and the ability to change structural conformation in the presence of Ca2+ . Discovering whether these properties are associated with the conserved cysteine residues characteristic of the EPDRs, and how they relate to each of the three conserved profiles identified here, will be an interesting avenue of future research.
The analyses presented here have revealed that the EPDR gene family arose early in eukaryotic evolution and is substantially larger than previously thought. We also identify a previously unrecognised clade of EPDR sequences with distinct patterns of conserved cysteine residues. Despite significant primary sequence divergence outside the core conserved residues, predicted 3D structures of EPDRs are similar, even among the three profile types. EPDRs are evolutionarily dynamic, having undergone independent expansions in multiple animal lineages, while being entirely lost from others. Although widely distributed, the structure and function of these proteins are largely unknown. Understanding the structure-function relationship of EPDRs will likely provide an explanation for the apparent proclivity of this family to be co-opted into diverse and lineage-specific roles in the extracellular environment, ranging from memory formation to biomineralization to conspecific communication.
Identification, phylogenetic analysis and characterisation of EPDR genes
Predicted protein datasets were downloaded from a range of species for which whole genome data are available. This dataset was supplemented by transcriptomes of key taxa; sequence reads were downloaded from NCBI nucleotide, protein, TSA or SRA databases, assembled if necessary using Trinity vr20131110 , and translated where required using TransDecoder (within the Trinity suite) with a minimum peptide length of 50 amino acids. Details of all data sources are provided in Additional file 10. These datasets were then searched for the ependymin domain using HMMER3.1  and the ependymin pHMM (Pfam  accession PF00811.14) with an e-value cutoff of 1e-05. For each species, all identified sequences (Additional file 11) were aligned using COBALT, a constraint-based tool that performs alignments aided by information about protein domains , with a gap penalty of 13 (all other settings were default). Alignments were viewed and manually adjusted within AliView . Identical or very similar sequences were removed, as were incomplete sequences that were missing a large portion of the conserved EPDR domain (Additional file 3). The alignment was then trimmed to remove apparent sequence-specific insertions and the divergent 3′ portion of the sequences (Additional file 4).
Maximum likelihood trees were produced using RAxML 7.7.6 , with automatic model selection (PMB model selected) and 100 replicates of nonparametric bootstrapping. Bayesian analysis was performed using MrBayes 3.2.6 [53, 54], using automatic (mixed) model selection (Blosum) with sampling every 10,000 generations. The analysis was run for 30 million generations and did not reach convergence (standard deviation of split frequencies 0.121180). Topologies of the resulting phylogenetic trees were largely congruent.
Sequences were analysed using SignalP 4.0  to predict signal sequences, and NetNGlyc 1.0  to predict glycosylation sites. Sequence logos were created for each class of EPDR using WebLogo . Predictions of 3D structure and potential ligand binding sites were performed using I-TASSER , and resulting PDB model viewing and alignment was performed in UCSF Chimera . Violin plots of protein sequence lengths were produced in R  using the program ggplot2 .
Synteny analysis was performed either using JGI’s inbuilt synteny analysis tool (for JGI hosted genomes) , or by reciprocal BLAST of neighbouring gene models using the BLAST function provided by species-specific genome browsers.
Identification and characterisation of Branchiostoma belcheri ependymin-related gene sequences
Initial BLASTP searches of the B. belcheri ‘v18h27.r3’ reference predicted protein set  with EPDR proteins identified several EPDR-like protein models. However, when the genomic loci associated with these were searched using TBLASTN, it was evident that some exons had been omitted or misattributed. In one instance, two distinct EPDR-like loci were conflated into a single protein model, and conversely, some models were partial sequences originating from a single locus. Similar issues were previously encountered with A. planci EPDR gene models . The proteins used here therefore are built from manually identified exons, but take their name from the original model(s). The EST database associated with this genome project (‘B.bel_xiamen_beihai.merged454EST’) provided at least partial and in most instances full-length support for these curated EPDR proteins. The manually curated protein sequences are available in Additional file 11. For a note on incorrectly predicted EPDR sequences see Additional file 2.
Identification and characterisation of Amphimedon queenslandica ependymin-related gene sequences
As for B. belcheri, several A. queenslandica EPDR gene models appeared to be incorrect based upon the presence of multiple ependymin domains within a model, the poor alignment of these sequences against other EPDRs, and/or on the poor alignment of sequenced transcripts to the models when viewed on the genome browser [62, 63]. Where possible, transcript information was used to identify the correct intron/exon architecture of the genes in the genome assembly; where no transcripts were present, translations of all three frames were investigated for alternative architectures and selected if they possessed canonical splice donor/acceptor sites and their use resulted in a better alignment. The manually curated protein sequences are available in Additional file 11.
Identification and characterisation of Haliotis asinina ependymin-related gene sequences
H. asinina EPDR sequences were identified by HMM searches on predicted open reading frames from an in-house Trinity assembly of reads from pooled mantle tissue of six juvenile abalone. The library was sequenced on 1/6th of a lane on an Illumina HiSeq 2000 (2 X 100 bp reads). Raw reads have been deposited into the NCBI Sequence Read Archive, BioProject ID PRJNA386701. For this transcriptome, assembly was performed without normalization. Assembled H. asinina EPDR transcripts are available in Additional file 11.
The authors would like to thank Edna Holman (nee Suárez-Castillo) for providing the alignments used in her previous analysis.
This research was funded by Australian Research Council grants to BMD and SFC (DP130102543). IMLS gratefully acknowledges start-up funding for her lab from MASTS (Marine Alliance for Science and Technology Scotland) and seedcorn funding through the Wellcome Trust ISSF3 grant number 204821/Z/16/Z. The funders had no role in the design of the study, in the collection, analysis and interpretation of data, or in writing the manuscript.
Availability of data and materials
Datasets used in this article are included within this published article and its supplementary information files. Raw sequencing data for Haliotis asinina have been deposited into the NCBI Sequence Read Archive, BioProject ID PRJNA386701.
BMD, CM and SFC conceived the study. MJH and SFC, and SCD and IMLS identified and analysed A. queenslandica and B. belcheri sequences, respectively. CM performed all other analyses. CM and BMD drafted the original manuscript. All authors edited, read and approved the final manuscript.
Ethics approval and consent to participate
No ethics approval was required for this study.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 4.Apostolopoulos J, Sparrow RL, McLeod JL, Collier FM, Darcy PK, Slater HR, Ngu C, Gregorio-King CC, Kirkland MA. Identification and characterization of a novel family of mammalian ependymin-related proteins (MERPs) in hematopoietic, nonhematopoietic, and malignant tissues. DNA Cell Biol. 2001;20(10):625–35.CrossRefPubMedGoogle Scholar
- 9.Zheng F-X, Sun X-Q, Fang B-H, Hong X-G, Zhang J-X. Comparative analysis of genes expressed in regenerating intestine and non-eviscerated intestine of Apostichopus japonicus Selenka (Aspidochirotida: Stichopodidae) and cloning of ependymin gene. Hydrobiologia. 2006;571(1):109–22.CrossRefGoogle Scholar
- 20.Heyland A, Vue Z, Voolstra CR, Medina M, Moroz LL. Developmental transcriptome of Aplysia californica. J Exp Biol. 2010;316B(2):113–34.Google Scholar
- 27.Romero A, Forn-Cuní G, Moreira R, Milan M, Bargelloni L, Figueras A, Novoa B. An immune-enriched oligo-microarray analysis of gene expression in Manila clam (Venerupis philippinarum) haemocytes after a Perkinsus olseni challenge. Fish Shellfish Immunol. 2015;43(1):275–86.CrossRefPubMedGoogle Scholar
- 38.Burki F, Kaplan M, Tikhonenkov DV, Zlatogursky V, Minh BQ, Radaykina LV, Smirnov A, Mylnikov AP, Keeling PJ. Untangling the early diversification of eukaryotes: a phylogenomic study of the evolutionary origins of Centrohelida, Haptophyta and Cryptista. Proc R Soc B. 2016;283(1823):20152802–10.CrossRefPubMedGoogle Scholar
- 42.Adams DS, Hasson B, Boyer-Boiteau A, El-Khishin A, Shashoua VE. A peptide fragment of ependymin neurotrophic factor uses protein kinase C and the mitogen-activated protein kinase pathway to activate c-Jun N-terminal kinase and a functional AP-1 containing c-Jun and c-Fos proteins in mouse NB2a cells. J Neurosci Res. 2003;72(3):405–16.CrossRefPubMedGoogle Scholar
- 43.Saif S. AP-1 is required for CMX-8933-induced SOD upregulation and is translocated in response to a human EPN mimetic. PhD Thesis. Worcester: Polytechnic Institute; 2004.Google Scholar
- 56.Gupta R, Jung E, Brunak S. Prediction of N-glycosylation sites in human proteins. In preparation; 2004. http://www.cbs.dtu.dk/services/NetNGlyc/. Accessed 24 May 2018.
- 58.R Core Team. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing; 2014.Google Scholar
- 64.Horton T, Kroh A, Ahyong S, Bailly N, Boury-Esnault N, Brandão SN, Costello MJ, Gofas S, Hernandez F, Mees J, et al. World register of marine species (WoRMS): WoRMS Editorial Board; 2018. http://www.marinespecies.org. Accessed 24 May 2018.
- 68.Maddison DR, Schultz KS. The tree of life web project; 2007.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.