Introduction

Despite more than 600 million years of evolution, the basic components of the bilaterian brain wiring diagram are highly conserved1. One of its signature trait is the presence of two categories of projection neurons: some that connect to target cells located on the same, or ipsilateral, side of the nervous system and others that connect on the opposite, or contralateral, side. The latest are called commissural neurons and they are distributed all along the rostro-caudal axis2. The position and spatio-temporal developmental sequence of a common set of commissural tracts (anterior commissure, posterior commissure, fasciculus retroflexus, optic nerve among others) are highly similar among vertebrates3,4. However, some commissural tracts only exist in some taxa, such as the corpus callosum in placental mammals5, or the Mauthner cells in lampreys, teleosts and amphibians6. It is assumed that the appearance of novel commissural circuits has allowed the acquisition of novel brain functions and behaviours2,7.

Understanding the mechanisms controlling the development and patterning of commissural circuits has represented a daunting challenge to developmental neurobiologists since the end of the nineteenth century8,9. Significant progress has only been made during the past thirty years or so through genetic and biochemical screening. The current model favors a rather simple push-pull mechanism whereby cells at the CNS midline, such as the floor plate in vertebrates, secrete proteins that attract commissural axons and facilitate midline crossing and also repellents that force commissural axons to leave the midline8,10. Netrin-1, the first midline chemoattractant, was simultaneously identified in C. elegans and in chick embryo11. The vertebrate DCC gene (Deleted in Colorectal Carcinoma12), and its homologues in C. elegans13 and Drosophila14, encode a transmembrane receptor, mediating Netrin-1 attraction. In these species, DCC loss-of-function prevents many commissural axons from crossing the midline, thereby supporting DCC pivotal role in midline guidance14,15. In mice and human, mutations in DCC leads to lethality15, movement disorders16 and cancers17. It was proposed that the DCC gene is absent from the chicken genome and that its paralogue, NEOGENIN, mediates NETRIN-1 attraction in this species18. However, multiple in vivo and in vitro studies have shown that in chick embryos, the chemotropic activity of NETRIN-1 on spinal cord commissural axons, enteric neural crest cells and oligodendrocyte precursors is blocked by anti-DCC antibodies19,20. Moreover, the in ovo electroporation of dominant negative constructs of DCC or DCC signaling partners in the chick spinal cord significantly perturbs commissural and motor axon guidance21,22,23. This suggests that a DCC gene might exist in the chick genome, which is known to be fragmented and to contain at least 30 microchromosomes24. Recently, the annotated genomes of 48 bird species were released25 which led us to revisit the evolution history of DCC and NETRIN genes in birds. We have also performed a comparative analysis of the organization and development of commissural circuits in early bird embryos.

Results

DCC gene is present in most sauropsid genomes

We first investigated whether a DCC gene is present in all available sauropsid genomes. Using NCBI and Ensembl genome databases, we found genes annotated as “DCC” in several avian, crocodilian and chelonian genomes (see Supplementary Table S1). We performed a phylogenetic analysis to rule out the possibility that the sauropsid DCC genes would be in fact Neogenin genes. We reconstructed vertebrate phylogeny of DCC and NEOGENIN proteins, using 47 sequences from 27 amniotes genomes, with drosophila Frazzled receptor (the DCC orthologue in flies) as outgroup (Fig. 1a). In this tree, DCC and NEOGENIN sequences cluster in two different well-supported clades, confirming that the two receptors are encoded by two distinct genes. Sauropsid DCC sequences are encompassed in vertebrate DCC group and recapitulate known phylogeny. Moreover, short branch lengths demonstrate that this gene is highly conserved among all vertebrates. Importantly, Neogenin genes could be found in all bird species investigated. The longer branch for neogenin sequences from passerifomes (Pseudopodoces humilis, Geospiza fortis, Fiducela albicolis) reveals a specific sequence divergence in this group. Together, these results confirm that a DCC gene is present in sauropsid including many bird species.

Figure 1: DCC gene has been lost independently twice during bird evolution.
figure 1

(a) Consensus phylogenetic tree of vertebrate DCC and NEOGENIN. Analysis was performed on 47 vertebrate DCC and NEOGENIN amino acid sequences using the Maximum likelihood method, with 1000 bootstrap replicates (for sequence references, see Supplementary Table S1). The tree was rooted using Drosophila FRAZZLED sequence as outgroup and branches displaying bootstrap values below 50 were collapsed. For better visualization, we cut out primary branches of a length equivalent to 1.2. (b) Conserved genomic synteny of amniotes DCC chromosomal region. The Figure displays simplified genomic synteny map comparing positions of DCC and its neighboring genes in different amniotes species (For full analysis, see Supplementary Figure S1). Orthologues of each gene are represented in the same color and displayed in the same column. (c) Proposed scenario for DCC gene loss in the avian lineage. DCC gene loss has occurred twice, independently, in galliformes and in passeriformes.

Full size image

DCC gene was lost twice in aves evolution

In birds, DCC gene is present in most birds major groups26 including paleognathes, anseriformes, strisores, columbaves, gruiformes, aequorlitornithes, accipitriformes, coraciimorphes, falconiformes and psittaciformes (Supplementary Table S1). However, in agreement with a previous report18, we could not find any DCC gene in the presently available chicken Gallus gallus genome. This was also the case for two other galliformes: the turkey Meleagris gallopovo and the quail Coturnix japonicus, suggesting that DCC gene is absent from this group. Another striking result was the absence of DCC gene in a second major avian group, the passeriformes. Indeed we did not detect DCC sequences in any of the eleven passeriformes genomes available on NCBI including the zebra finch Taenopygia guttata. For both galliformes and passeriformes we identified DCC genes in their respective sister group, namely anseriformes and psittaciformes, which supports independent gene losses (Supplementary Table S1).

To understand the evolutionary history of the DCC gene in birds, we investigated the DCC genomic region. We performed a physical co-localization of genetic loci on the same chromosome within an individual or species analysis (or synteny) of this particular region in amniotes genomes, i.e. mammals (human, mouse, platypus), testudinian (painted turtle), crocodilian (alligator) and various birds including paleognathes (ostrich), galloanseres (chicken, turkey, duck, goose) and 15 neoaves (Fig. 1b, Supplementary Fig. S1 and Supplementary Table S3-4-5). When present, DCC is always located on the sexual chromosome Z in birds, in contrast with other sauropsids and mammals in which DCC is on autosomes27,28.

Synteny analysis showed that this region was highly conserved in amniotes and only few gene rearrangements could be observed between sauropsids and mammals, except from two major gene block losses in galliformes and passeriformes (Fig. 1b and Supplementary Fig. S1).

In galliformes, a block of 12 genes, including DCC, was absent from the syntenic region: MAPK4, ME2, ELAC1, SMAD4, MEX3C, DCC, MBD2, POLI, STARD6, DYNAP, RAB27B, and CCDC68 (Supplementary Fig. S1). In addition, a 13th gene, TCF4 appeared to be missing in turkey (Supplementary Fig. S1). Only two of these genes could be detected elsewhere in these galliformes genomes using tblastn algorithm from NCBI: MBD2 gene in chicken and quail and CCDC68 in turkey (Supplementary Fig. S1), but their position in the genome is still undetermined. Moreover, many chromosomal rearrangements could be observed downstream of this region in galliformes, in contrast with the stability of this genomic region in the sister group, anseriformes. In particular, the block of genes nearby this deletion (from TCF4 to CPLX4) is reversed compared to the ancestral sequence. Furthermore, TCF4 gene, that is located at the border of the deletion block, is annotated at 875 bp from the beginning of Z chromosome (Gallus_gallus-4.0; Ch.Z NC_006127.3). Together this suggests that the DCC block was lost in galliformes during a Z chromosome extremity flip as illustrated in Fig. 1c.

In passeriformes, a block of 7 genes, including DCC, is also absent from the locus: DCC, MBD2, POLI, STARD6, DYNAP, RAB27B, and CCDC68. In addition, an 8th gene, TCF4, is missing in the White-throated sparrow and Zebra finch available genomes (Supplementary Fig. S1). As for galliformes, none of these genes could be detected using tblastn algorithm on passerifomes genome sequences. This portion of passeriforme genomes is poorly assembled, and it is impossible to determine how these scaffolds are positioned on Z chromosome. Previous studies demonstrated that the Z chromosome has undergone major gene loss and shortening during bird evolution29. Our observations are compatible with a scenario of two independent Z chromosome rearrangements in galliformes and passeriformes leading to large chromosomic region losses, including the DCC gene block (Fig. 1c).

Netrin genes, except NETRIN-5, are present in all birds

We next investigated whether genes encoding DCC main ligands, the secreted NETRINs30,31,32, were present in known bird genomes. In vertebrates, 5 secreted NETRIN proteins have been described (NETRIN-1 to 5) and are all thought to bind to DCC12,33,34,35. Previous study indicates that NETRIN genes originated prior to vertebrate radiation36. To investigate the evolutionary history of NETRIN genes in birds, we reconstructed their phylogeny using amphioxus NETRIN-1 and NETRIN-4 as outgroups (Supplementary Fig. S2). This phylogenetic tree revealed that NETRIN-2 and NETRIN-3 sequences cluster in a single well-supported clade (see Supplementary Fig. S2), demonstrating that these are actually two annotations of the same gene. This observation was confirmed by synteny analysis (data not shown). No particular divergence could be observed for birds NETRIN-1, NETRIN-2/3, or NETRIN-4 sequences, as compared to the other vertebrates. In particular, these genes were also present and conserved in both galliformes and passeriformes (Supplementary Fig. S2). In contrast, long branches suggested that NETRIN-5 sequences are highly divergent among all vertebrates. Furthermore, it was impossible to find any NETRIN-5 gene in bird genomes (Supplementary Fig. S2). Taken together, these data show that there is no ligand modification counterpart of DCC gene loss in galliformes and passeriformes.

DCC mRNA and protein are not detectable in galliformes and passeriformes

To confirm the loss of DCC in galliformes and passeriformes compared to other bird species, we assessed its mRNA and protein product by performing in situ hybridization and immunostaining. We used two riboprobes, cloned respectively from duck and pigeon cDNA, and two different antibodies, specific for DCC extracellular and intracellular domain (see Methods). We performed experiments on different bird embryos from species with available genomic data: chicken (Gallus gallus), quail (Cortunix japonica), duck (Anas platyrhynchos), pigeon (Columba livia), and zebra finch (Taeniopygia guttata). In addition we used three other galliformes (pheasant, partridge, and quail) assuming that they do not have a DCC gene. Pigeon and duck DCC antisense riboprobes could detect DCC mRNA in their respective species, while sens riboprobes could not (Fig. S3). In addition, the two probes could detect DCC mRNA in both species (Fig. 2e,f). In contrast, no signal was detectable with either probe in galliformes and passeriformes (Fig. 2a–d,g), strongly suggesting an absence of DCC mRNA in these embryos. Similarly, the two anti-DCC antibodies labeled spinal cord commissures in duck and pigeon embryos but not in embryos from galliformes and passeriformes (Fig. 2l and n). Commissural axons were also immunoreactive for DCC in the hindbrain of ostrich embryos (data not shown). This result confirms genomic data of specific DCC gene losses in passeriformes and galliformes. Immunostaining with an antibody against the pan-neuronal marker ßIII-Tubulin showed that the overall organization of axonal tracts was highly similar in spinal cord section from the seven species tested (Fig. 2). Importantly, the ventral spinal cord commissure was present in all birds. We also performed immunostaining with antibodies against the Roundabout 3 (ROBO3) receptor, known to be expressed by growing commissural axons in the spinal cord and hindbrain in developing mammals37,38, zebrafish39 and chicken40. A single ROBO3 gene was detected in all birds (data not shown) and accordingly, ROBO3-immunopositive commissural axons were observed on spinal cord sections from all bird embryos tested (Fig. 2). Thus absence of DCC receptor does not appears to impact bird spinal cord commissural formation.

Figure 2: Both DCC mRNA and protein are expressed in bird spinal cord except in galliformes and passeriformes.
figure 2

Spinal cord sections from different birds are stained: partridge (a,h), pheasant (b,i), quail (c,j), chick (d,k), duck (e,l), pigeon (f,m) and zebra finch (g,n). (a–g) In situ hybridization using DCC antisense riboprobes cloned from duck and pigeon detect strong expression of DCC mRNA expression in the spinal cord (e,f). Duck and pigeon riboprobes cross-react between the two species, but fail to detect DCC mRNA in galliformes (a–d) and passeriformes (g). (h–n) Expression of ßIII-Tubulin and ROBO3 was detected in ventral commissures of all spinal cords. Anti-DCC antibodies against the intracellular and extracellular domains detect strong expression of DCC protein in ventral commissural neurons in duck (l) and pigeon (m) but fail to detect any DCC expression in galliformes (h–k) and passeriformes (n). Arrows indicate the ventral midline. Abbreviations, Drg, dorsal root ganglia; Mr, motor nerve root. Scale bars: 50 μm.

Full size image

To further study DCC expression pattern in birds, we performed whole-mount anti-DCC immunostaining, 3DISCO clearing and 3D imaging with light sheet microscopy on chick, pheasant, duck, pigeon and zebra finch embryos at HH21-22. In the chick, early axonal tracts41 and peripheral nerves could be labeled with anti-ßIII Tubulin immunostaining but none expressed DCC (Fig. 3a,b and Supplementary Movie S1). We also failed to detect any DCC expression in pheasant and zebra finch embryos (Supplementary Fig. S4), confirming what was found in the spinal cord. In contrast, many axons were immunoreactive for DCC in duck and pigeon embryos (Fig. 3c,d). As previously shown in rodents and xenopus, DCC was broadly expressed by commissural axons in the mesencephalon, diencephalon and rhombencephalon. DCC was also detected in retinal ganglion cells, in the olfactory nerves, motor axons and fasciculus retroflexus (Fig. 3e–j and Supplementary Movie S1). Together, these data confirm that DCC was selectively lost in the galliforme and passeriforme lineages despite its highly conserved expression pattern in other tetrapods.

Figure 3: 3DISCO analysis of DCC receptor expression in chicken, pigeon and duck embryos.
figure 3

DCC (a) and ßIII-Tubulin (b) whole-mount immunostaining on HH22 chicken embryos after 3DISCO clearing. No DCC expression is detected in chicken embryos whereas many axonal tracts are strongly labeled with anti-ßIII-Tubulin. (c–j), By contrast, DCC is strongly expressed in pigeon (c,e–g) and duck (d,h–j) embryos at stages equivalent to HH22 or HH28. (h–j) in pigeon embryo, DCC is found in commissural axons crossing the floor plate (arrowhead in e), in retinal ganglion cells (Rgc) of the retina and the optic nerve (On; f), in the habenula nucleus (Ha) and fasciculus retroflexus (Fr; g). (h–j) in duck embryo, DCC is found in spinal cord (Sc) commissural axons, motor nerve roots (Mr), Rgc in the eye (i), optic nerve (On) diencephalon (Di) and olfactory nerve (Olf). Scale bars are 200 μm except (e,j), 100 μm and f, 50 μm. Abbreviations, Di: diencephalon; Hb: hindbrain; Tec, tectum; Olf: olfactory nerve; Tel, telencephalon; Tg, trigeminal ganglion.

Full size image

Homogenous organization of early commissural tracts in bird embryos

To determine if the lack of DCC in passerifomes and galliformes might have resulted in a different organization of their commissural projections, we compared the early development of commissural tracts in the chick with that of pigeon and duck embryos. We used anti-ROBO3 immunostaining to specifically label all posterior commissural tracts and reconstruct their 3D organization. In addition, we used anti-ßIII Tubulin to investigate rostral commissural tracts development. In all embryos, the position, developmental sequence and density of commissural axons were comparable (Fig. 4). As expected from previous work in the mouse, when present, DCC homogeneously stained all spinal cord and hindbrain commissural axons, as does ROBO3 (Fig. 4, Supplementary Fig. S2 and Supplementary Movies S2-S3). The fasciculus retroflexus was also labeled, and no major differences were observed (data not shown). At more rostral levels where ROBO3 was not expressed, commissural tracts such as the posterior commissure or the post optic commissure tract (Supplementary Fig. S4) could be observed with anti-ßIII Tubulin, and no noticeable difference was detected between chick and duck or pigeon embryos.

Figure 4: Commissural projections labeled with ROBO3 appear similar in birds with or without DCC.
figure 4

(a–o) 3D light sheet images of whole-mount bird embryos labeled with anti-Robo3 antibodies. (a–c) in H21-22 chick embryos, Robo3 is expressed by commissural axons in the tectum (Tec), ventral midbrain (Mb) and hindbrain (Hb) but not in the telencephalon (Te). The floor plate is indicated by an arrowhead in (b). Commissural axon growth cones (arrow) are seen approaching the midline in (c). At equivalent developmental stages, the spatial pattern of Robo3+ commissural projections is similar in pheasant (d–f), duck (g–i), pigeon (j–l) and zebra finch (m–o) embryos. Scale bars, 300 μm in (a,d,g,j,m); 150 μm, in (b,e,h,k,n); 100 μm in (c,f,i,l,o).

Full size image

Discussion

We found DCC genes in representative species of saurians, chelonians, crocodilians, as well as in many bird species, including paleognathes, anseriformes and numerous neoaves. In contrast, DCC gene is missing in chicken, in agreement with previous observations18, as well as in other galliformes, the turkey and the quail. Furthermore, DCC is also absent in passeriformes, such as crow, fly catcher and zebra finch. DCC gene was also not found in some bird genomes outside from passerifomes and galliformes (Table S1), however these bird genomes belong to low-coverage genome sequencing groups42. Moreover, in these cases, other birds from the same groups have a DCC gene, suggesting that its absence is more likely due to incomplete genome sequences, than to a real absence of DCC in these species. Phylogeny analysis clearly clustered bird DCC sequences with the DCC of the other sauropsids, and together with mammalian and actinoterpygian sequences in a single DCC clade. By contrast, NEOGENIN, the DCC paralogue, appears to exist in all birds.

This supports that a DCC gene was present in bird ancestor, conserved in various avian groups, but lost in two distinct groups, galliformes and passeriformes. According to current bird phylogeny, these two groups are not closely related26,43, suggesting that two independent events of DCC loss occurred during bird radiation. One is aware that the absence of some genes in current bird genomes could be related to incomplete sequencing of some genomic regions44. We show here that the presence or absence of DCC mRNA and DCC protein in the brain matched with the presence or absence of DCC gene in the corresponding bird genomes. In the phylogeny tree, the short lengths of bird DCC branches indicated that bird DCC sequences did not diverge as compared to the other amniote sequences. In addition DCC expression pattern when present is similar to what has been described in the mouse14,45. This indicates that the loss of DCC in some bird groups was not preceded by any major change of DCC structure and function in the bird lineage. Moreover, synteny analyses showed that both in galliformes and passeriformes, many genes are lost together with DCC. This shows that the loss of DCC in these birds is not targeted specifically on DCC gene, but concerns a whole genomic region. This differs from other recently reported gene loss in birds, such as the loss of KISS gene46. In this case, a degenerated sequence of KISS could be observed in some bird species, such as duck, zebra finch, and rock pigeon, and a specific loss of this gene in many other birds, including falcon and chicken, suggesting that accumulation of mutations and alteration of function preceded the loss of the gene46. Here, the lack of DCC would result from independent accidental loss events in both galliformes and passeriformes during Z chromosome recombination. This result has been confirmed in an independent study by Patthey et al.47

In vertebrates, DCC binds to NETRIN-112,48, NETRIN2/334, NETRIN-433 and possibly NETRIN-535. Our phylogeny analyses showed that all birds possess NETRIN-1, 2/3, and 4, as the other osteichthyans with no special divergence. Interestingly, we could not retrieve any NETRIN-5 sequence in birds, while this gene is present in other amniotes, as well as in teleost fish. This suggests an early loss of NETRIN-5 in the bird lineage. In mammals, NETRIN-5 is expressed in the developing brain, notably in neurogenic regions32 but its function is still largely unknown. Besides the loss of DCC in galliformes and passeriformes, the lack of NETRIN-5 in extant avian species represents another striking specificity of the DCC/NETRIN system in birds.

In the mouse, DCC has been reported to be essential for NETRIN-1 mediated attraction in many different systems15 and DCC knockouts are not viable14. Previous in vitro studies have shown that in the chick as well, NETRIN-1 attracts sensory and motor neurons, spinal cord and hindbrain interneurons, GnRH neurons, enteric neural crest cells and oligodendrocyte precursors20,21,23,49,50,51,52,53. However, the conclusion of several previously published studies deserve to be reconsidered in the light of our results. For instance, the specificity of anti-DCC antibodies used to block NETRIN-1 activity on chick cells19,20 is highly questionable as there is no DCC in chick genome. Likewise, the axon guidance defects observed after electroporation of truncated or mutated human DCC receptors, or of a ROBO1 ectodomain in chick spinal cord cannot be attributed to a dominant negative effect such as dimerization with endogenous DCC22,23. Other models, not requiring DCC, should be proposed to explain these results. For instance, the exogenous DCC receptors might bind to and perturb the function of Robo1/Robo2/Robo3 receptors, which are all expressed by chick commissural axons54,55 and are known DCC partners in mouse neurons56,57. They might also trap NETRIN-1 thereby titrating it from its other receptors.

Our analysis of early bird embryos failed to reveal any major differences in the development of commissural tracts, regardless of the presence of DCC. Although such differences in commissural systems might exist at later developmental stages, the puzzling question remains on how some bird species coped with the accidental losses of DCC and in particular on the identity of the receptor(s) mediating NETRIN-1 chemoattractive activity in these species.

A first obvious candidate is the DCC paralogue NEOGENIN, as previously proposed18. In the mouse, DCC and NEOGENIN cooperate to attract spinal cord commissural axons to the floor plate, and NEOGENIN can partially compensate for DCC absence in rodents58. Phylogeny analysis reveals a divergence in NEOGENIN sequences in passeriformes, which suggests a neofunctionalization of NEOGENIN in this group. As this bird group has lost DCC, we may raise the hypothesis that this neofunctionalization could possibly be related to NEOGENIN taking on DCC function. However, no such NEOGENIN sequence divergence was observed in galliformes, which have also lost DCC. Therefore such a scenario of NEOGENIN neofunctionalization could not apply in this later group. Although NEOGENIN is present in at least some chick spinal cord commissural neurons18, it has not yet been shown to be expressed by all NETRIN-1 responsive neurons in this species, and unlike DCC, NEOGENIN has other ligands59.

In chick embryo, NETRIN-1 was also shown to bind Down syndrome cell-adhesion molecule (DSCAM) and silencing DSCAM expression in chick spinal cord neurons impairs NETRIN-1 attraction60. In rodents, recent in vivo studies using knockout mice have challenged this model and suggest that DSCAM61 is dispensable for NETRIN-1 attraction of commissural axons. It remains to confirm that this is also the case in passeriformes and galliformes.

Interestingly, DCC was also shown to bind dorsal repulsive axon guidance protein (DRAXIN) a secreted molecule which repels various classes of commissural axons62. DRAXIN and NETRIN-1 bind each other and compete for DCC binding63. How DRAXIN, which was first isolated in chick embryo62, might function in birds that have no DCC is also a mystery, but NEOGENIN could also be involved. Importantly, preliminary analysis of bird genomes indicates that, DSCAM, UNC5A-D and DRAXIN genes exist in all birds (data not shown).

In mammals, DCC plays a role in commissural axon guidance but also in the migration of neural crest cells and of GnRH neurons from the olfactory epithelium20,64. DCC influences the development of the autonomic innervation of arteries65. Moreover, DCC is a dependence receptor that can induce apoptosis in absence of NETRIN-166, a mechanism that might explain its anti-tumorigenic properties17. Therefore, one would expect DCC loss to have important consequences on the development or function of many organs in the corresponding bird species, if not fully compensated by other receptors.

Possible loss of SMAD4, also located in the DCC genomic region, is particularly interesting as SMAD4 knockout mice are embryonic lethal67. However, we could find a SMAD4-like gene on chicken chromosome 25. This gene is conserved and its locus is unchanged in all birds and many vertebrates except mammals, and has been already characterized in xenopus68. It will be important to understand how some birds can cope with the absence of SMAD4, or if SMAD4-like could replace SMAD4.

Importantly, cancer has been described in all vertebrates including birds and most of the missing genes such as SMAD467, SKA169, MEX3C70 and DCC have been linked to tumorigenesis. Interestingly, there is in chick a high incidence of spontaneous ovarian cancer with a prevalence reaching up to 35% after 3.5 years of age71,72. This correlates well with the downregulation of DCC expression described in human ovarian tumors73,74.

In conclusion, our results suggest that commissural axon guidance mechanisms are not conserved between bird species but that overall this does not seem to have a major impact on brain patterning. This illustrates the great plasticity of axon guidance mechanism, and how diverse this system can be among vertebrates. Another example of this diversity was recently reported in mammals, where mutations of few amino acids in mammalian ROBO3 receptor have completely modified its mechanism of action in commissural neurons56. To fully appreciate this diversity, it will be essential to reconstruct the phylogenic history of commissural guidance receptors and ligands in vertebrates.

Methods

Genomic databases analysis

Protein sequences from annotated genes were extracted from Ensembl or NCBI genome browsers (http://www.ensembl.org/index.html and http://www.ncbi.nlm.nih.gov/nuccore/). The TBLASTN algorithm of the NCBI website (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used on the genomic databases available when genes where not previously annotated. See Supplementary Table S1 and S2 for complete genome and sequences information.

Phylogenetic analysis

DCC-NEOGENIN Analysis

47 sequences composed of predicted mature netrin-receptor (DCC-NEOGENIN) with N-terminal signal peptide were first aligned using ClustalW75, then manually adjusted. The JTT (Jones, Taylor and Thornton) protein substitution matrix of the resulting alignment was determined using ProTest software76. Phylogenetic analysis of the NEOGENIN-DCC receptors alignment was performed using the Maximum Likelihood method with 1,000 bootstrap replicates (RaxML software, https://www.phylo.org/portal2). DCC-NEOGENIN homologous Drosophila melanogaster FRAZZLED receptor was used as outgroup.

NETRIN-1/2/3/5 and NETRIN-4 Analysis

52 sequences for NETRIN-1/2/3/5 and 19 sequences for NETRIN-4, each one composed of a predicted mature NETRIN protein with N-terminal signal peptide, were aligned using ClustalW. The JTT protein substitution matrix of the resulting alignment was determined using ProTest software. Phylogenetic analysis of the NETRIN sequences alignment was performed using the Maximum Likelihood method with 1,000 bootstrap replicates (RaxML software) using Branchiostoma floridae NETRIN-1 and NETRIN-4 as outgroups, respectively.

DCC synteny analysis

Synteny maps of the DCC conserved genomic region were reconstructed for mammals (human, mouse, platypus), chelonian (painted turtle), crocodilians (alligator) and birds: paleognathae (ostrich), galloansers (duck, goose, chicken, turkey), and neoaves (falcon, bald eagle, royal eagle, adeli penguin, emperor penguin, pigeon, ibis, egret, cuckoo, chimney swift, hoatzin, zebra finch, sparrow, flycatcher and crow). Analyses of DCC neighbouring genes were performed manually using complete or preliminary annotated genome sequences from NCBI genome browser (http://www.ncbi.nlm.nih.gov/gene/), including numerous unplaced genomic scaffolds (see Supplementary Table S3 for references and locations of the genes used in the synteny analysis, Table S4 for a complete list of the genes used in this analysis, and Table S5 for a complete list of species used in the analysis). To complete this analysis we used TBLASTN algorithm on NCBI database to identify non-annotated DCC neighbouring genes and confirm gene absence (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Animal sampling

Eggs from chicken Gallus gallus, duck Anas platyrhynchos, zebra finch Taenopygia guttata, pigeon Columba livia, quail Coturnix japonica, pheasant Phasianus colchicus and partridge Perdrix perdrix were incubated at 37 °C in humid conditions. Embryos were collected at different time points depending on their embryological stage. All procedures were performed in accordance to the guidelines approved by French Ministry of Agriculture and UPMC University ethic committee. Stage determination was done according to literature77,78,79. Exact number of embryos collected per stage is presented in Supplementary Table S6. Embryos were first transferred to ice-cold PBS 1X; from E8, the nervous system were dissected and all embryos were then fixed by immersion in 4% paraformaldehyde overnight at 4 °C. Samples were transferred to PBS 1X and kept at 4 °C until use.

For whole-mount immunostaining, samples were dehydrated in methanol (MeOH 50%in PBS 1X - MeOH 80% in PBS 1X - MeOH 100%) and incubated overnight in MeOH with 5%H2O2 to suppress blood auto-fluorescence. Samples were then rehydrated (MeOH 100%- MeOH 80% in PBS 1X - MeOH 50%in PBS 1X - PBS 1X) and kept in PBS 1X at 4 °C until use.

Histochemistry

In situ hybridization

Tissue sectioning and in situ hybridization were performed as previously described37. Pigeon DCC probe was designed in highly conserved domain of DCC gene coding from Fibronectin 5 to P1 domain. The sequence was amplified from pigeon embryos cDNA using following primers (Forward: 5′- CAGTAGGTGTCCAGGCTGTTG - 3′; Reverse: 5′- CCCGTTGGCTTCTCCATGTTC - 3′) and cloned into pCRII-TOPO plasmid (ThermoFisher). The Duck DCC cDNA was kindly provided by Dr Sara Wilson.

Sections immunostaining

Immunostaining were performed as previously described37. The following primary antibodies were used: mouse anti-βIII Tubulin (1:1000, MMS435P-Covance), goat anti-ROBO3 (1:500, AF3076-R&D), goat anti-DCC (1:400, Sc-6535-Santacruz, raised against intracellular C-terminal domain of human DCC), mouse anti-DCC (1:300, AF-OP45-Calbiochem, raised against extracellular domain of human DCC). Corresponding secondary antibodies were used: donkey anti-rabbit Alexa488 (1:500, 711-545-152-Jackson), bovine anti-goat Cy3 (1:500, 805-165-180, Jackson), donkey anti-mouse cy3 (1:500, 715-165-150-Jackson), goat anti-mouse DL649 (1:500, 115-495-205-Jackson). Sections were counterstained with Hoechst and examined with a fluorescent microscope (DM6000, Leica) coupled to a CoolSnapHQ camera (Roper Scientific).

Whole-mount Immunostaining

Samples were incubated at room temperature (RT) in a solution (PBSGT) of PBS 1X containing 0.2% gelatin (Prolabo), 0.5% Triton X-100 (Sigma-Aldrich) and 0.01% thimerosal (Sigma-Aldrich) for 3 h (E4-E6) or 8 h. Samples were next transferred to PBSGT containing the primary antibodies and placed at 37 °C, with rotation at 70 rpm, for 4 days (E4-E6) or 7 days. Primary antibodies used were the following: mouse anti-βIII Tubulin (1:1000, MMS435P-Covance), goat anti-ROBO3 (1:400, AF3076 R&D), goat anti-DCC (1:400, Sc-6535 Santacruz) and mouse anti-DCC (1:300, AF-OP45-Calbiochem). Samples were then washed 3 times in PBSGT for 2 h at RT and incubated for 24 h at 37 °C in secondary antibody diluted in PBSGT. Secondary antibodies used were the following: donkey anti-rabbit Alexa647 (1:500, 711-605-152-Jackson), bovine anti-goat Cy3 (1:500, 805-165-180, Jackson) and donkey anti-mouse Alexa488 (1:500, A21202-Lifetechnolgie). After 4 washes of 2 h in PBSGT at RT, samples were stored at 4 °C in PBS until clearing.

Small samples (E4) were included in agarose 1.5% prior tissue clearing for better positioning in the ultramicroscope chamber.

Tissue clearing was performed using 3DISCO-clearing procedure as previously described80. Samples are stored in dibenzylether (DBE) in light protected glass vials at RT.

Ultramicroscopy

3D imaging was performed with an ultramicroscope (LaVision BioTec) using ImspectorPro software (LaVision BioTec). The light sheet was generated by a laser (wavelength 488 and 561 nm, Coherent Sapphire Laser and 640 nm, Coherent OBIS 640–100LX laser, LaVision BioTec) and two cylindrical lenses. A binocular stereomicroscope (MXV10, Olympus) with a 2X objective (MVPLAPO, Olympus) was used at different magnifications (1.25×, 1.6×, 2×, 2.5×, 3.2× and 4×). Samples were placed in an imaging reservoir made of 100% quartz (LaVision BioTec) filled with DBE and illuminated from the side by the laser light. Images were acquired with a PCO Edge SCMOS CCD camera (LaVision BioTec).

Image processing was performed using Imaris software (Bitmap), as described previously80.

Additional Information

How to cite this article: Friocourt, F. et al. Recurrent DCC gene losses during bird evolution. Sci. Rep. 7, 37569; doi: 10.1038/srep37569 (2017).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.