Bioinformatic Analysis of Nematode Migration-Associated Genes Identifies Novel Vertebrate Neural Crest Markers

Neural crest cells are highly motile, yet a limited number of genes governing neural crest migration have been identified by conventional studies. To test the hypothesis that cell migration genes are likely to be conserved over large evolutionary distances and from diverse tissues, we searched for vertebrate homologs of genes important for migration of various cell types in the invertebrate nematode and examined their expression during vertebrate neural crest cell migration. Our systematic analysis utilized a combination of comparative genomic scanning, functional pathway analysis and gene expression profiling to uncover previously unidentified genes expressed by premigratory, emigrating and/or migrating neural crest cells. The results demonstrate that similar gene sets are expressed in migratory cell types across distant animals and different germ layers. Bioinformatics analysis of these factors revealed relationships between these genes within signaling pathways that may be important during neural crest cell migration.


Introduction
Cell migration is essential for tissue and organ formation during embryogenesis and for the regeneration of some adult tissues. Abnormal regulation of cell migration often results in severe developmental defects and can lead to cancer metastasis. The neural crest is a stem cell-like, multipotent, migratory cell population unique to vertebrate embryos. Neural crest progenitors are born at the border of the neural plate and non-neural ectoderm, become localized within the dorsal neural tube and then migrate away from the neural tube as it closes. This emigration process occurs via an epithelial-to-mesenchymal transition (EMT) such that these ectodermally-derived cells delaminate from the dorsal neural tube and invade the surrounding mesenchyme. They then migrate to distant sites in the periphery of embryos and differentiate into various tissues [1][2][3][4].
Although neural crest development has been extensively studied for several decades, few genes functionally involved in neural crest cell migration have been identified. This is mainly because the neural crest is a vertebrate-specific cell type and it remains impractical to carry out large scale genetic screening in vertebrates due to their relatively long generation times and high upkeep costs. In contrast, invertebrate genetic screens using the nematode Caenorhabditis elegans and fruit fly Drosophila melanogaster have identified large number of genes involved in key developmental processes. Mutations disrupting cell migration in different cell types have been identified in C. elegans, many of which have yet to be fully functionally characterized [5][6][7].
Genome-wide sequencing of many organisms has provided insight into the infrastructures of their genomic organization. To date, a tremendous amount of sequence information exists in available databases and is ripe for bioinformatic analysis. Comparative genomics is a powerful tool for the identification of novel genes associated with conserved biological processes even between evolutionarily distant animal model systems. We previously used this approach to show that genes important for the migration of nematode hermaphrodite-specific neurons (HSNs) are also involved during neural crest development (Kee et al., 2007). Of fifteen identified vertebrate homologs of genes required for HSN neuronal cell migration, thirteen were shown to be expressed in premigratory or migratory neural crest cells [8], demonstrating that many cell migration genes are evolutionary conserved and ancient. The remarkable conservation of genes involved in long-range cell migration between worm HSN and chick neural crest emphasizes the utility of exploiting existing sequenced genomes and information from traditional genetics to identify candidates involved in stereotypic cell biological processes like cell migration.
Here, we ask whether such high conservation also exists for regulatory genes functioning in cell migration processes between migratory nematode cells arising from different germ layers and vertebrate neural crest. To address this, we selected twenty-five C. elegans genes required for the migration of various cell types and performed genomic scanning to search for putative vertebrate orthologs; further candidates were identified by bioinformatic analysis of common functional pathway components. This was followed by expression analysis of the putative candidate genes, which revealed previously unidentified regulatory and signaling genes that may play key, conserved roles in controlling cell migration.

Materials and Methods
Computational identification of vertebrate orthologs of worm cell migration genes Twenty-five nematode (C. elegans) genes previously identified in genetic screens as essential for migration of various cell types were selected for this study (Table 1) [6,7]. Putative chicken (Gallus gallus) orthologs of the worm genes were identified by comparative genomic scanning, as shown previously [8]. As the chicken genome has been almost completely sequenced, we were able to identify putative orthologs corresponding to all twenty-five C. elegans cell migration genes (Table 1).

DNA constructs and RNA probe synthesis
Sequences of putative vertebrate orthologs corresponding to all twenty-five C. elegans cell migration genes were used as queries to search the chicken EST database at www.chick.umist.ac.uk/ ( Table 1). Chicken EST clones of the genes in the predicted FGFR signaling pathway were identified in a similar manner: SOS1 (ChEST793F19), GAB1 (ChEST878L14), FRS2 (ChEST166G11), Shc1 (ChEST352e24), SOS2 (ChEST173l15). PIK3R2 was isolated in our previous screen [9]. Antisense RNA probes for in situ hybridization were prepared as previously described [10].

Embryo collection and whole mount in situ hybridization
White Leghorn chicken eggs were obtained from local farms and were incubated at 38uC for two to three days. Embryos were fixed in 4% paraformaldehyde in PBS (Phosphate Buffered Saline) at 4uC overnight, and subjected to whole mount in situ hybridization, as previously described [11,12]. Stages were determined according to criteria of Hamburger and Hamilton [13]. National Institute of Health guidelines state that for chicken embryos younger than embryonic day 10, no special approval or IACUC documentation is required.

Pathway analysis
Predicted signaling pathways were obtained from our identified putative vertebrate orthologs using Gene Ontology and MetaCore (GeneGo) analysis with the canonical pathways setting. A specific sub-pathway was generated by limiting the connections of the components in each pathway including only genes identified from this study.

Genomic scanning of cell migration genes
Twenty-five genes, identified from genetic screens in C. elegans for their functional importance in cell migration in various tissues [6,7], were selected for computational analysis. Using BLAST search engines against available vertebrate genome sequences and genomic databases [8], we identified vertebrate counterparts, here defined as putative orthologs. The vertebrate gene with highest identity score by computational analysis was used to identify the putative chick ortholog and its available EST (Expressed Sequence Tags) sequence. In this way, we identified orthologs in the chicken genome corresponding to all twenty-five C. elegans cell migration genes (Table 1).

Expression screening of cell migration genes
To test whether the twenty-five vertebrate orthologs of the worm genes were expressed during neural crest development, we performed in situ hybridization at various stages beginning at the premigratory stage (HH8) when cranial neural crest cells are located within the dorsal neural tube and have yet to delaminate from the neuroepithelium by undergoing EMT to initiate migration at HH9. Thirteen chicken transcripts, homologous to the C. elegans cell migration genes ced-2 (Crk), ced-5 (Dock180), , unc-51 (ULK2) and vab-8 (kinesin), were detected in premigratory neural crest cells in the dorsal neural tube ( Figure 1). Neural crest expression was not uniform in every case: the putative vertebrate ortholog of lin-39 (HoxA4) for example, was only expressed in the neural folds of the caudal trunk at later stages ( Figure 1). Furthermore, expression of the fbl-1 ortholog FBLN2 was limited to the tip of the dorsal neural tube and was also expressed in the cranial ectoderm, while the vab8 ortholog (kinesin) was expressed only in the dorsal hindbrain. Thus, we identified homologs of nematode cell migration genes that are expressed in subpopulations of premigratory neural crest cells from different axial levels of the neural tube. All of these genes have known functions in cell migration in the nematode ( Table 1). The fact that they are also expressed by premigratory neural crest cells as they are preparing to initiate EMT raises the intriguing possibility that they may be functionally important in vertebrates as well.

Functional pathway analysis
To identify the biochemical pathways in which the identified vertebrate orthologs might act, we performed bioinformatic analysis using MetaCore (GeneGo) software and The Gene Ontology Consortium. This provided candidate pathways in which the orthologous genes had been previously implicated. One of the best hits was in the developmentally-related FGFR signaling pathway, a canonical pathway in the MetaCore database. By analyzing genes common to this pathway, we extracted a predicted pathway including direct connections with the genes identified from our expression studies; FGF, FGFR1, GRB2, CRK, DOCK1 and RAC1 ( Figure 3A, red). The molecular components of this map were previously implicated in EMT and cytoskeletal remodeling during cell migration [14][15][16].
To determine whether the members of this predicted pathway are conserved in migrating neural crest cells, the remaining pathway components ( Figure 3A, black letters) were analyzed by in situ hybridization. Our results show that transcripts of FRS2, GABA1, SHC1, and SOS1 are barely detectable in premigratory neural crest cells ( Figure 3C a, b,d,e), but are present at low levels in migrating neural crest cells ( Figure 3C a9, b9,d9,e9). The guanine nucleotide exchange factors son of sevenless 1 (SOS1) and son of sevenless 2 (SOS2) function in multiple signaling pathways and have been implicated in a wide variety of biological and oncogenic processes [17]. Interestingly, SOS2 is highly up-regulated in the emigrating neural crest cells and maintained in migrating neural crest cells ( Figure 3C, f,f9), while SOS1 is expressed in migrating neural crest cells at low levels ( Figure 3C e9). PIK3R2 (p85b phosphoinositide 3-kinase subunit) is known to regulate tumor progression: p85b expression is elevated in breast and colon carcinomas and genetic alteration of PIK3R2 expression levels modulate tumor progression in vivo [18]. Our result shows PIK3R2 is up-regulated in the neural crest cells emigrating out of neural tube ( Figure 3C, c,c9) and in migrating neural crest cells [9]. Our functional pathway analysis suggests that up-regulation of PIK3R2 and SOS2 correlate with neural crest EMT and that SOS2 appears to be more prevalent than SOS1 in early neural crest development. FGF signaling is known to be required for neural crest induction or specification from neural stem cells, but has not been previously implicated in neural crest EMT or initial cell migration. Thus, our genomic and pathway analysis have revealed previously unidentified genes and pathways that may be critical for normal emigration and/or migration of neural crest cells and are promising candidates for future functional experiments.

Conclusions
Large quantities of genomic sequencing data are available in diverse databases that are ripe for mining. Genetic screens in invertebrates such as worms and fruit flies have identified many molecules important in behavioral, developmental, and physiological processes. Here, we have broadened the utility of cross- Figure 2. Putative chicken orthologs of nematode cell migration genes are conserved in migrating cranial neural crest cells. Chicken embryos were subjected to whole mount in situ hybridization using RNA probes corresponding to the vertebrate orthologs of each nematode gene. (A) Whole mount chicken embryos (upper panel) and tissue sections at the midbrain level of each embryo (lower panel) show that ten vertebrate orthologs are expressed in neural crest cells migrating from the neural tube in chicken embryos at HH stage 9-10; Crk (ced-2), Dock180 (ced-5), p21-Rac1 (ced-10), NR1I3 (daf-12), Fgfr1 (egl-15), Fgf18 (egl-17), Fbln2 (fbl-1), HoxB6 (mab-5), Grb2 (sem-5) and Ulk2 (unc-51). Arrowhead indicates migrating neural crest cells. (B) Five vertebrate orthologs are expressed in migrating neural crest cells at the hindbrain level in HH stage 11-13 embryos; Dock180 (ced-5), p21-Rac1 (ced-10), NR1I3 (daf-12), Fgfr1 (egl-15) and Kinesin (vab-8). Arrows indicate gene expression in migrating cranial neural crest cells in rhombomere 4 and/or rhombomere 6. doi:10.1371/journal.pone.0103024.g002 species comparative genomic screening by combining it with functional pathway analysis to identify additional candidate genes for analysis. Our results have uncovered previously unidentified molecular players up-regulated during the course of neural crest migration. It is interesting to note that many of the genes identified have previously been assumed to be ubiquitously expressed, and would be unlikely to be identified by conventional screens. This proof-of-principle study validates our combinatorial strategy as an effective pipeline for the identification of novel regulatory genes during neural crest development and other complex migratory processes.