Introduction

snail genes encode zinc–finger DNA-binding proteins that act as transcriptional repressors and serve as regulators of cell shape and movement, specifically epithelial–mesenchymal transitions (reviewed in Nieto 2002). In gnathostome vertebrates, which have three distinct snail paralogous groups (Manzanares et al. 2004), snail genes are expressed in multiple embryonic tissues (Locascio et al. 2002; Manzanares et al. 2004; Nieto 2002). snail expression is first detected during early gastrulation before becoming restricted to presomitic mesoderm and the tailbud; mice homozygous for snail knockout alleles die early in gestation with defects in gastrulation and mesoderm formation (Carver et al. 2001). In addition to this early mesodermal expression, another common feature of snail gene expression in gnathostomes is in premigratory neural crest (pnc) and its pharyngeal arch derivatives (Nieto 2002); inhibition of snail2 paralogs produces defects in neural crest development (e.g., Carl et al. 1999; Nieto et al. 1994). Since neural crest is a key vertebrate innovation (Gans and Northcutt 1983), the evolution of snail neural crest function may have been important to the origin of vertebrates.

Comparative studies in nonvertebrate chordates provide important clues for understanding the evolution of vertebrate snail function. In tunicates, which do not form somites, snail is expressed in trunk muscle precursors (Corbo et al. 1997) and in cells along the dorsal neural tube (Wada and Saiga 1999). Similarly, in amphioxus, snail is expressed in presomitic mesoderm and in the lateral neural plate (Langeland et al. 1998). These findings indicate the ancestral role of snail in marking mesoderm and also indicate that a snail-expressing pnc-like population of cells was present prior to the origin of vertebrates.

Lampreys lie at the base of the vertebrate clade. They are both an important emerging system for comparative analysis of vertebrate embryonic development (reviewed in Osorio and Retaux 2008), as well as a critical reference point for determining the timing of gene duplication events (e.g., Neidert et al. 2001). In a comprehensive study of neural crest regulatory genes, Sauka-Spengler et al. (2007) briefly described early snail expression in the lamprey Petromyzon marinus that does not correlate with gnathostome snail pnc patterns. Here, we provide a more complete expression analysis of P. marinus snail including later embryonic stages and additional tissues, as well as a phylogenetic analysis of lamprey snail. Our results indicate that this single lamprey snail gene retains ancestral snail patterning domains present in nonvertebrate chordates, but is also expressed in tissues that are broadly equivalent to the combined sites of expression of all three gnathostome snail paralogy groups. Intriguingly, our results indicate that snail may be involved in neural crest development, but not in early pnc specification as it is in gnathostomes.

Materials and methods

Isolation of Petromyzon marinus snail

We screened our lamprey embryonic cDNA library (Tomsa and Langeland 1999) at low stringency with a zebrafish Snail1a probe (Thisse et al. 1993). A positive clone was excised into phagemids and sequenced. In an attempt to identify any additional snail genes in the library, we employed a degenerate primer-mediated polymerase chain reaction (PCR) strategy to conserved sequence motifs.

Phylogenetic analysis

The deduced PmSnail amino acid sequence was aligned with other chordate snail homologs using Clustal X2 (http://www.clustal.org/), and the alignment was manually edited with JalView (http://www.jalview.org/) to remove poorly conserved regions. The resulting alignment (supplemental material) was used to construct a neighbor-joining tree implemented by Clustal X2 with default settings. The resulting tree was rooted on the single snail homolog from the cephalochordate Branchiostoma floridae (accession AAC35351). Other snail protein sequences included are those from hagfish (Eptatretus burgeri; AB288229), shark (Scyliorhinus canicula; AAN62356, AAN62355), zebrafish (NP_001008581, NP_571141, NP_571064, NP_001070853) Xenopus tropicalis (NP_989267, NP_984424), chick (NP_990473, I50738), mouse (NP_035557, AAI00727, NP_035545), and human (NP_005976, NP_003059, NP_840101).

In situ hybridizations

Lamprey embryos were staged and fixed as previously described (Tomsa and Langeland 1999). Antisense digoxigenin dUTP riboprobes were synthesized from the 3′UTR of our PmSnail cDNA. In situ hybridization was carried out as described (Tomsa and Langeland 1999), and embryos were mounted whole in Permount. Selected embryos were embedded in agarose and sectioned on a Vibratome as described (McCauley and Bronner-Fraser 2006). Sectioned tissues were visualized by either DIC optics or apotome optical sectioning utilizing the near infrared fluorescence emitted by the NBT/BCIP chromogenic precipitates (Trinh et al. 2007).

Results and discussion

We have cloned and characterized a snail homolog from the lamprey P. marinus. Comparative analysis of this gene with other vertebrate and nonvertebrate chordate snail genes indicate that lamprey snail is broadly characteristic of the ancestral vertebrate snail gene, both in terms of its phylogenetic position and its expression pattern.

Gnathostome-specific snail gene duplications

BLAST searching confirmed that our P. marinus hybridization clone is a bona fide snail homolog, which we have named PmSnail and submitted to Genbank under accession number FJ372630. Sequence alignments indicate that this clone is missing approximately 10–15 codons from the 5′ end of the cDNA which encompasses the SNAG domain (Nieto 2002). Blasting of lamprey genomic traces (http://genome.wustl.edu/tools/blast/) yielded no sequences identical to our clone and did not recover the 5′ SNAG domain. Our degenerate PCR screen resulted in four additional independent isolations of the same gene and no other snail variants.

Our molecular phylogeny of snail genes (Fig. 1) shows strong support (NJ bootstrap value = 82.6%) for placing PmSnail as an outgroup to all gnathostome snail genes. Within gnathostomes, snail genes form well-supported snail2 and snail3 paralogy groups, and a paraphyletic grouping of snail1 genes. This overall snail family topography, minus lamprey and hagfish snail, is very similar to that reported by Manzanares et al. (2004).

Fig. 1
figure 1

Lamprey snail is an outgroup to all gnathostome snail genes. Neighbor-joining tree of vertebrate snail genes, rooted on amphioxus snail. The tree indicates robust support (bootstrap value 862/1,000) for grouping lamprey snail independently of gnathostome snail1, snail2, and snail 3 paralogy groups. This would indicate that the gene duplication events giving rise to these groups are gnathostome-specific and not shared by lampreys

Full size image

Our phylogenetic analysis indicates that the single lamprey snail gene is equally orthologous to all gnathostome snail genes. The simplest model to explain this tree topology is that the duplications that gave rise to the various gnathostome snail paralogs are not shared by lampreys and thus took place after the lamprey–gnathostome divergence. Within gnathostome snail genes, our tree indicates an early duplication giving rise to the snail1/2 and snail3 clades, followed by a second duplication to form the snail1 and snail2 clades. Note that snail1 genes do not form a monophyletic group, suggesting that they have evolved quite differently in different lineages. Note also that some lineages have apparently lost snail3 paralogs.

Interestingly, our tree shows PmSnail branching out before the SnailA homolog from the hagfish Eptatretus burgeri (Ota et al. 2007). This topology does not agree with either of two competing hypotheses for hagfish phylogeny that place them either as basal to the lamprey + gnathostome clade (Janvier 1996) or as the sister group to lampreys (e.g. Mallat and Sullivan 1998; Stock and Whitt 1992).

It is formally possible that there are additional undiscovered snail genes in the P. marinus genome. However, we repeatedly cloned the same gene by degenerate PCR of embryonic cDNA, suggesting that any additional snail genes are expressed at virtually undetectable levels. Furthermore, low-stringency Southern blots on P. marinus genomic DNA indicate only a single snail gene (data not shown), and BLAST searching the as-yet incomplete P. marinus genomic traces yielded no additional hits. While only complete genomic sequence will provide an unequivocal answer, all evidence to date points to a single lamprey snail gene. Even if there are additional lamprey snail genes, however, our phylogenetic analysis suggests that they would not align within any of the gnathostome snail paralogy groups. Instead, we claim they would have to represent independent gene duplication events in the lamprey lineage, similar to that we have previously reported for the Dlx gene family (Neidert et al. 2001).

Multiple sites of lamprey snail expression

snail transcripts can be detected by RT-PCR in unfertilized lamprey eggs and early cleavage stage embryos (data not shown). snail transcripts can be detected by whole mount in situ hybridization at the neural plate stage (Tahara 1988 stage 18), where they are expressed diffusely across the entire neural plate and into the flanking tissue (Fig. 2a). As somitogenesis begins (stages 20 and 22), snail expression becomes upregulated more discretely in both the presomitic mesoderm, the tailbud, and the definitive somites (Fig. 2b,c), which is characteristic of many gnathostome snail genes. Interestingly, snail expression also resolves to bilateral stripes in the neural tube (Fig. 3a,b), a pattern heretofore undescribed for any vertebrate snail genes. snail expression is also detected in the notochord (Fig. 3a,b), similar to that reported for vertebrate snail3 genes (Manzanares et al. 2004). We also detect snail expression in cells just lateral to the neural tube in stages 20 and 22 embryos (Fig. 3b,c). In postsomite formation (stage 25), snail expression is detected primarily in the pharyngeal arches (Fig. 4a), and sections reveal that the expression is in the ectomesenchyme (Fig. 4b). Expression is also detected in the developing lens (Fig. 4a).

Fig. 2
figure 2

Early snail expression in the neural plate, neural tube and somites. a snail is first detected at the neural plate stage (stage 18) in a broad pattern encompassing both the neural plate (np) and flanking mesoderm. b By early neural tube stage (stage 20), snail expression has resolved to the somites (s) and neural tube (nt) in more anterior regions, but remains diffuse throughout the tailbud (tb). c This pattern persists throughout somitogenesis (e.g., stage 22)

Full size image
Fig. 3
figure 3

Transverse sections reveal additional aspects of snail expression. a Apotome optical section of anterior transverse tissue slice of stage 20 embryo showing snail expression in bilateral stripes in the neural tube (nt) as well as in the somites (s) and notochord (n). b Apotome optical section of more posterior tissue slice of stage 20 embryo highlighting snail-expressing cells that may be neural crest cells (nc) delaminating from the neural tube. c Apotome optical section of mid-body transverse slice of stage 22 embryo showing snail expression in the neural tube (nt) and somites (s) as well as in cells that may be neural crest cells (nc) delaminating bilaterally from the neural tube

Full size image
Fig. 4
figure 4

Later snail expression. Post-somitogenesis (e.g., stage 25, a), snail expression becomes limited to the pharyngeal arches (pa) and the developing lens (l). b Horizontal section of stage 25 embryo shows snail expression in the crest-derived ectomesenchyme (em) of the pharyngeal arches

Full size image

Lamprey snail exhibits ancestral patterning roles

Consistent with its basal phylogenetic position among vertebrate snail genes, lamprey snail is expressed in patterns that broadly encompass the expression patterns of all gnathostome snail genes. In addition to the overall similarity, there are important differences between lamprey and gnathostome snail expression patterns.

Mesoderm and derivatives

In all gnathostomes examined, snail genes are expressed in presomitic mesoderm. We have previously shown in amphioxus that snail expression correlates with presomitic mesoderm (Langeland et al. 1998). Given this context, our finding of lamprey snail expression in presomitic mesoderm is expected. Most but not all gnathostome snail genes that are expressed in the presomitic mesoderm persist in their expression during somite formation (Hammerschmidt and Nusslein-Volhard 1993; Isaac et al. 1997; Locascio et al. 2002; Nieto et al. 1992; Nieto et al. 1994; Smith et al. 1992; Thisse et al. 1995; Thisse et al. 1993). Our finding that lamprey snail expression persists during somite formation suggests that this pattern is ancestral in vertebrates, and the lack of such persistence in amphioxus suggests it is a vertebrate innovation.

Lamprey snail is also expressed in the notochord during somitogenesis. This is unlike snail expression in nonvertebrate chordates. Amphioxus does not express snail in the notochord (Langeland et al. 1998) and Ciona has only weak snail expression in notochord precursors (Corbo et al. 1997). However, the notochord expression is very similar to the expression of snail3 paralogs described by Manzanares et al. (2004). Our finding suggests that snail expression in the differentiated notochord is a vertebrate innovation that has been retained by lamprey snail but has been lost by gnathostome snail1 and snail2 paralogs.

Neural tube, neural crest, and derivatives

Neural crest is a key vertebrate innovation (Gans and Northcutt 1983; Sauka-Spengler et al. 2007). Neural crest cells delaminate from the invaginating neural tube, migrate extensively, and differentiate into a variety of tissues including sensory ganglia, pigment cells, cartilage and, in some species, bone. While most are well characterized in gnathostomes, lampreys also have definitive neural crest characterized both by vital dye labeling (McCauley and Bronner-Fraser 2003) and gene expression (Sauka-Spengler et al. 2007). In all gnathostomes examined, one or more snail genes are expressed in cells along the lateral edge of the neural plate and on the crest of the invaginating neural tube and thus provide a consistent marker of presumptive neural crest.

Curiously, as described by Sauka-Spengler et al. (2007) and confirmed here, snail is expressed across the neural plate in early (stage 18) lamprey embryos and does not seem to be specific to neural crest. This is a clear and important difference from gnathostomes. Another important difference is our finding that lamprey snail expression is persistently upregulated in the neural tube during somitogenesis stages. No pattern like this has been observed in gnathostomes, although it is similar to the pattern described for amphioxus snail (Langeland et al. 1998). While the extent of this neural tube expression is inconsistent with all snail-expressing cells being premigratory neural crest, we did detect snail expression in cells lateral to the neural tube of stages 20 and 22 embryos, during which crest is known to migrate (McCauley and Bronner-Fraser 2003). We also found that lamprey snail is later expressed in the ectomesenchyme of the pharyngeal arches. This is similar to gnathostome snail expression and is consistent with neural crest origin of the snail-expressing cells (McCauley and Bronner-Fraser 2003). Whether these lateral snail-expressing cells are indeed of neural crest origin will require further examination including lineage analysis of snail-expressing cells. Lampreys share a vast cascade of neural crest regulatory genes with gnathostomes (Sauka-Spengler et al. 2007). While snail may play a role in later neural crest development in lampreys, the recruitment of snail into the early neural crest cascade appears to be gnathostome-specific.

Vertebrate snail gene regulatory evolution

Vertebrates have clearly undergone widespread gene duplications if not whole genome duplications that are thought to have facilitated the large-scale morphological innovations that accompanied vertebrate origins. The traditional view (Ohno 1970) is that duplication releases selective constraints on one paralog, offering the possibility of new functions evolving (neofunctionalization). The consequences of such neofunctionalization would be a net increase in expression complexity following duplication. Alternatively, according to the duplication–degeneration–complementation (DDC) model (Force et al. 1999), another fate of duplicate paralogs with pleiotropic expression is subfunctionalization where mutations disable enhancers of the paralogs in complementary fashion. This would result in each paralog having less complex expression patterns than the nonduplicated ancestral gene.

Our data suggest a modest level of regulatory neofunctionalization resulting in novel expression in the somites, notochord, lens, and crest-derived pharyngeal arch tissue, occurring in a single unduplicated snail gene prior to the vertebrate radiation. At least two gene duplication events occurred during the early gnathostome radiation. During this time, snail was recruited into the early neural crest regulatory network and into limb bud formation (both neofunctionalization events), and the multiple sites of snail expression were partitioned among the three snail paralogs (DDC). Locascio et al. (2002) have documented significant shuffling of expression domains between snail1 and snail2 paralogs in different gnathostome lineages indicating additional levels of complexity in snail regulatory evolution after the lamprey–gnathostome divergence.