Introduction

In spite of the fact that only a small fraction of the genome of complex metazoans consists of coding regions (e.g., 5% in the human genome; International Human Genome Sequencing Consortium 2001), numerous clusters of evolutionarily related genes have been noted. A number of hypotheses have been proposed to explain the evolution and maintenance of these linked genes: (1) Linked genes are expressed under the influence of common enhancers, and in the absence of gene specific repressors should be expressed in a similar spatio-temporal pattern; (2) Linked genes are coordinately regulated by higher-order chromatin organization (Spitz et al. 2003), leading to spatial and/or temporal colinearity of expression, as seen in the expression pattern and genomic organization of the Hox family of transcription factors; and (3) Gene linkage may have no contemporary functional significance—it may simply represent a phylogenetic signature, the outcome of a relatively recent tandem duplication event or the remnant of an ancient gene cluster that has lost its functional significance.

One gene family which exhibits syntenic linkage of unknown functionality and origin is the Wnt family. The Wnt genes comprise an evolutionarily conserved group that is known to affect cell fate decisions during development and oncogenesis (Cadigan and Nusse 1997). Twelve Wnt subfamilies have been identified in deuterostomes (Nusse 2001). A total of 6 of these 12 subfamilies have been identified in Drosophila, whereas 11 have been recovered from the sea anemone Nematostella vectensis, a member of the basal metazoan phylum Cnidaria. This suggests (1) that the Wnt radiation preceded the cnidarian–bilaterian divergence and (2) that several Wnt genes have been lost in the fruit fly lineage (Nusse 2001; Kusserow et al. 2005; Lee et al. 2006).

Several Wnt clusters have been identified in the Bilateria, including the Wnt3–Wnt9B, Wnt3A–Wnt9A, and Wnt2–Wnt16 miniclusters that are found in vertebrates. An evolutionarily conserved cluster of three Wnt genes (Wnt1–Wnt6–Wnt10) is thought to have existed in the last common ancestor of arthropods and deuterostomes (Nusse 2001). In Drosophila, this compact cluster spans ∼70 kb on chromosome 2 at position 27f (Fig. 1a). In human and mouse, Wnt1 is linked closely to Wnt10b, and Wnt6 is linked closely to Wnt10a (Fig. 1b). These two Wnt miniclusters are located on different chromosomes in both species (Nusse 2001).

Fig. 1
figure 1

Wnt1–Wnt6–Wnt10 clusters. a and b reproduced from Nusse (2001)

Full size image

To test the hypothesis that this cluster was present in the cnidarian–bilaterian ancestor, we consulted the recently completed genome sequence of N. vectensis (Nematostella genome sequencing project, Joint Genome Institute, D. Rokhsar, principal investigator; Miller et al. 2005). Additionally, we tested hypotheses of cluster origin and maintenance by searching for additional clusters of Wnt genes (both previously reported and novel) and correlating these with expression patterns of these genes in Nematostella. Recent research regarding gene expression patterns of homeobox clusters has been synthesized into the hypothesis that genes responsible for patterning each of the three germ layers of triploblasts are clustered such that the Hox cluster patterns the neuroectoderm, the ParaHox the endoderm, and the NK the mesoderm (Garcia-Fernàndez 2005). If colinearity is operative for Wnt genes in Nematostella, then those Wnt genes responsible for patterning each germ layer will be clustered, and the order of genes within clusters will mirror the order of Wnt expression territories along the body axis.

Materials and methods

A draft assembly of the Nematostella genome was produced using the Phusion Genome Assembler (Mullikin and Ning 2003) from genomic traces published on the National Center for Biotechnology Information by the Joint Genomes Institute. The assembly, comprising 81,401 contigs spanning ∼3.6 Mb, has been described in detail elsewhere (Ryan et al. 2006). Previously published coding sequences for Nematostella Wnt genes (Kusserow et al. 2005) were blasted against the assembled genome at StellaBase (Altschul et al. 1990; Sullivan et al. 2006). Gene alignments were created using ClustalX (Thompson et al. 1997). The neighbor-joining method (Saitou and Nei 1987) was applied to the PAM–Dayhoff-matrix-based distances using the Phylip software package (Felsenstein 1989). Reliability of internal branches was assessed using the bootstrap method with 1,000 replicates.

Results and discussion

Wnt6 and Wnt10 are more closely linked in Nematostella than in the fruit fly or human—the intergenic distance is only 4,474 nucleotides. The relative orientation of these two genes is reversed in Nematostella (5′-Wnt10-Wnt6-3′) versus bilaterian animals (5′-Wnt6-Wnt10-3′). Unlike in the fruit fly, Wnt1 is not closely linked to Wnt6 or Wnt10 in Nematostella (Fig. 1c). To help resolve the evolutionary events leading to this organization in human, fruit fly, and Nematostella, we searched for a conserved syntenic linkage of protein-coding regions adjacent to the Wnt clusters. The two protein-coding regions in closest proximity to each cluster in humans and Drosophila, both upstream and downstream, were used to query the Nematostella genome (tblastn) to determine if any of these protein-coding regions are present on the same assembled contig as Nv–Wnt1 or Nv–Wnt6–Nv–Wnt10 (Table 1). No genes which are neighbors to Wnt clusters in Drosophila or human are closely linked to either Wnt6–Wnt10 or Wnt1 in Nematostella. This clustering is suggestive of one of three evolutionary scenarios (Fig. 2). Assuming Nv–Wnt1 was not lost from a primordial Wnt1–Wnt6–Wnt10 cluster (Fig. 2a), the cnidarian–bilaterian ancestor likely possessed a cluster comprised of Wnt6–Wnt10. Based on the assembly, Wnt10 and Wnt1 can be no closer than 59 kb in Nematostella (Fig. 1).

Table 1 Genes linked to select Wnt clusters in Homo sapiens and D. melanogaster
Full size table
Fig. 2
figure 2

Three plausible scenarios for the evolution of the Wnt1–Wnt6–Wnt10 cluster. Wnt1/wg is represented by green triangles, Wnt6 by red triangles, and Wnt10 by blue triangles. Potential ancestral states are shown at the nodes representing the bilaterian ancestor and the cnidarian–bilaterian ancestor. In scenario a, Drosophila retains the condition present in the cnidarian–bilaterian ancestor. The lineage leading to humans experiences (1) a cluster duplication, (2) the loss of Wnt1 from one cluster, (3) the loss of Wnt6 from the other cluster, and (4) an inversion of Wnt1. In the lineage leading to Nematostella, (5) Wnt1 becomes dispersed from the cluster, and (6) the relative order of Wnt6 and Wnt10 becomes reversed. In scenario b, only Wnt6 and Wnt10 are clustered in the cnidarian–bilaterian ancestor. In the line leading to bilateria, (1) Wnt1 joins the Wnt6–Wnt10 cluster. Subsequently, the line leading to humans undergoes the same changes as in scenario a. In the line leading to Nematostella, (6) the relative order of Wnt6 and Wnt10 becomes reversed. In scenario c, Nematostella retains the condition found in the cnidarian–bilaterian ancestor. In the line leading to bilateria, (1) the relative order of Wnt6 and Wnt10 becomes reversed, and (2) Wnt1 joins the Wnt6–Wnt10 cluster. The line leading to humans experiences the same changes as in scenarios a and b

Full size image

We noted a novel cluster comprised of Wnt5 and Wnt7 in Nematostella. The 3′ end of Wnt5 is 9,458 nucleotides from the 3′ end of Wnt7 (Fig. 3). The published Wnt7 and Wnt7b transcripts (Kusserow et al. 2005) map to a single locus in Nematostella and share 533 nucleotides (Fig. 3). Only a single Wnt7 transcript is known from Drosophila, but diverse bilaterians (e.g., human and common house spider, Achaearanea tepidariorum) express at least two distinct Wnt7 transcripts. In human and mouse, these are encoded by separate loci. It is possible that the ancestral bilaterian possessed two Wnt7 loci and that one of these has been lost in the fruit fly lineage.

Fig. 3
figure 3

A novel Wnt cluster of Wnt5/Wnt7 is observed in N. vectensis with alternative splicing resulting in distinct Wnt7 and Wnt7b transcripts. Wnt7 is composed from exons 1b, 2, 3, and 6 and Wnt7b from exons 1, 1b, 2, 3, 4, and 5

Full size image

Each Wnt7 transcript in Nematostella could be more closely related to a different Wnt7 locus in bilaterians if the Wnt7 duplication in Bilateria occurred by retroposition of a Wnt7 splice variant present in the cnidarian–bilaterian ancestor. To test this possibility, we performed a phylogenetic analysis of Wnt7 genes from Nematostella and various bilaterian taxa. Unlike a previously published Wnt phylogeny (Kusserow et al. 2005), we excluded from the alignment those amino acid positions aligned with the 533-nucleotide stretch that is shared by the Wnt7 and Wnt7b transcripts of Nematostella (Fig. 4a). Our results suggest that the Wnt7 locus underwent independent gene duplications in the ecdysozoa and the deuterostomia and that these duplicate genes bear no direct relationship to the alternate splice variants of Nematostella (Fig. 4b).

Fig. 4
figure 4

Evolutionary relationships of Wnt7 variants from diverse taxa. a An alignment of Wnt7 genes was generated using ClustalX (Thompson et al. 1997). The nucleotides aligned to the exons shared by Nematostella Wnt7 and Wnt7b are highlighted in gray. b A neighbor-joining tree generated from a PAM–Dayhoff matrix comparison of Wnt7, excluding those portions of each sequence which align to the shared exons of NvWnt7 and NvWnt7b (Felsenstein 1989). Labels in the nodes represent the percentage of times each branch occurred in 1,000 trees. The scale bar represents the number of substitutions per site. At, Achaearanea tepidariorum (AtWnt7a—AB167809, AtWnt7b—AB167811); Bb, Branchiostoma belcheri (AF206499); Bf, Branchiostoma floridae (BfWnt7a—AF100739, BfWnt7b—AF061975); Dm, Drosophila melanogaster (NM_023653); He, Heliocidaris erythrogramma (AY532158); Hs, Homo sapiens (HsWnt7a—NM_004625, HsWnt7b—NM_058238); and Nv, Nematostella vectensis (NvWnt5—AY725202, NvWnt7a—AY687350, NvWnt7b—AY725204)

Full size image

No other clusters of Wnt genes were observed. Nonetheless, the widely conserved clustering of Wnt6 and Wnt10 and the presence of a Wnt5–Wnt7 cluster in Nematostella suggest that selection may be acting to maintain a close linkage. Neighboring Wnt genes could be under the influence of common cis-regulatory elements, as has been observed in other developmentally important genes such as the Hox family of transcription factors whose expression is collinear with their genomic organization (Kondo and Duboule 1999). In Nematostella, Wnt genes are expressed in distinct domains along the anterior–posterior axis of the developing larva, and they are presumably critical for patterning this axis (Kusserow et al. 2005). Similarities in expression suggest possible common regulation for Wnt5 and Wnt7 but not for Wnt6 and Wnt10. Wnt6 is expressed in body wall endoderm in the center of the body column, whereas Wnt10 is expressed in scattered pharyngeal cells. These expression territories should not be regarded as adjacent—the pharynx is an inversion of the body wall at the oral end of the animal and as such should be regarded as the animal’s oral extremity (Stephenson 1926). Wnt5 and Wnt7/7b are expressed in adjacent axial locations, at the junction of the mouth and the pharynx. However, Wnt5 is expressed in the endoderm, whereas Wnt7/7b is expressed in the ectoderm (Kusserow et al. 2005; Fig. 5). It is conceivable that one or more shared cis-regulatory elements impact the axial expression of both genes, but support for collinear germ-specific gene clusters is not provided.

Fig. 5
figure 5

Wnt5 (blue) and Wnt7/7b (black) are expressed at the oral pole of developing Nematostella planulae in the endoderm and ectoderm, respectively (Kusserow et al. 2005)

Full size image