Members of the Wnt subfamilies, 1, 2, 5, 7, 8, A and 16, were previously isolated from Achaearanea or Cupiennius, and subfamilies 1, A and 16 from Glomeris (Additional file 1). Note that Glomeris Wnt16 and WntA were previously erroneously characterised as Wnt7 and Wnt5 orthologues respectively 33. We obtained sequences of a further four Wnt genes from both Achaearanea and Glomeris using degenerate PCR with embryonic cDNA template. The sequences of degenerate primer pairs used to isolate Wnt genes are shown in Additional file 2. Larger fragments of initial PCR fragments were obtained via RACE PCR using the Marathon RACE Kit (Clontech). RNA isolation from spiders and Glomeris, and cDNA synthesis was carried out as described previously 3343.

For Daphnia, known Wnt gene sequences were obtained from GenBank and protein sequences were used to perform tblastn searches of assembled genomic scaffolds, predicted gene models and ESTs (Daphnia pulex v1.1, September 2006; http://www.jgi.doe.gov/Daphnia and http://wFleaBase.org). Segment pairs with an E-value smaller than 10⁵ were selected and the corresponding scaffolds were manually curated with the help of Dappu v1.1 filtered gene models. Predicted gene structures were refined by comparison to Wnt genes from other species (Nematostella vectensis, Drosophila melanogaster, Tribolium castaneum, Apis mellifera, Homo sapiens, Mus musculus and Strongylocentrotus pupuratus) to identify the correct open reading frames. Partial cDNAs were cloned to confirm most intron-exon boundaries. Briefly, TRIzol (Invitrogen) was used to isolate RNA from Daphnia embryos of mixed stages. RNA was reverse transcribed using SuperScriptIII (Invitrogen) and RT-PCR was performed using primers specific to each predicted Wnt open reading frame. Sequence from each gene model was used to search the Daphnia assembly and confirm the presence of twelve Wnt gene sequences and the absence of any additional Wnt family members (Additional file 1). The synteny of Daphnia Wnt genes was inferred from their linkage on the same genomic scaffolds.

Gene models of nine Wnt genes from the tick Ixodes scapularis were retrieved from VectorBase 44 and GenBank deposits (Additional file 1). A fragment of a tenth Ixodes Wnt gene was also identified through tblastn searches (Additional file 1).

Six Platynereis Wnt genes were isolated in a previous study 19. Two more Wnt genes (Pd-Wnt5 and Pd-Wnt8) were found in an EST collection 45. To identify remaining Wnt orthologues a combination of more specific primers were used (Additional file 2). The accession numbers of all Wnt gene sequences used in this study are shown in Additional file 1.

Two data sets were used for the analysis, the first set consisted of 93 amino acid sequences from arthropods, Platynereis and human (Additional file 3: Wnt sequence data set 1), and the larger second set included additional sequences from a nematode, cnidarian and three lophotrochozoans (Additional file 4: Wnt sequence data set 2). Sequences in both data sets were aligned using T-coffee 46 and hand-edited in SeaView 47 to remove poorly aligned amino acid positions (Additional files 3 and 4).

Initially, the best-scoring substitution model was determined among the amino acid models in RAxML 48 as WAG+F+Γ (WAG with empirical base frequencies and the Γ model of rate heterogeneity; Whelan and Goldman 49).

Bayesian phylogenetic analyses were performed with MrBayes 50. The final topology was estimated using 13,000,000 iterations using 3250 burning cycles and sampling every 1000 iteration. Clade support was assessed with posterior probabilities computed with MrBayes and non-parametric bootstrapping implemented in RAxML 48 based on 1000 replicates.

Spiders (Achaearanea tepidariorum and Cupiennius salei) were obtained from laboratory stocks in Cologne and Göttingen 2451. Spider embryos were staged according to Akiyama-Oda and Oda 52. General handling and staging of Glomeris is described in Janssen et al., 33. Tribolium beetles (Ga-1 strain) were obtained from laboratory stocks at Kansas State University. Beetles were reared at 30°C in whole-wheat flour supplemented with 5% dried yeast. Platynereis larvae and juveniles were obtained from a breeding culture established in Gif-sur-Yvette according to the protocols of Fischer and Dorresteijn http://www.platynereis.de.

Whole mount in situ hybridisation (WMISH) was performed for spiders as described in the published protocol for Cupiennius embryos 43. For Glomeris, WMISH was performed as described in Prpic and Tautz 53 and Janssen et al. 32. Both spider and Glomeris embryos were counterstained with Sytox Green or DAPI and images were captured with a Leica dissection microscope or a Zeiss Axioplan-2 microscope. For Tribolium and Platynereis, WMISH was performed as described previously 4154555657. All digital images have been subjected to adjustment of brightness, colour values and contrast using Adobe Photoshop CS3.

In Achaearanea and Cupiennius gene expression was investigated in stage 4 to stage 10 embryos, which represent germ disc embryos with radial symmetry (stages 4 to 6), and germ band embryos with axial symmetry (stages 7 to 10) and up to 7 opisthosomal segments 5258. In Glomeris, gene expression was investigated in stage 0 (blastoderm) to stage 6.1 embryos; see Janssen et al. 33 for a detailed description of staging. In Tribolium, gene expression was analysed in embryos at the fully extended germ band stage. In Platynereis, as in many other annelids, the elongation of the body axis continues during post-embryonic development as new segments are added from a sub-terminal SAZ 41. We thus compared the expression of Wnt genes during trunk formation in both embryonic and post-embryonic development.

Combining the findings of database searches, genome annotation, degenerate PCR and Wnt genes identified in previous studies (see Methods), we found a total of eleven Wnt genes in Achaearanea, (with WntA from Cupiennius representing a twelfth spider Wnt gene), seven in Glomeris, twelve in Daphnia, ten in Ixodes and twelve in Platynereis. These sequences were then aligned with the Wnt sequences of Acyrthosiphon pisum, Drosophila, Homo and Tribolium (Additional files 1 and 3). A further alignment was generated using a larger set of Wnt genes containing the Wnt sequences from Caenorhabditis, Capitella, Helobdella, Lottia and Nematostella in addition to the sequences used in the first set of Wnt genes (Additional files 1 and 4). We then carried out phylogenetic analyses using Maximum likelihood approaches (Wnt sequence sets 1 and 2) and additional Bayesian approaches (Wnt sequence set 1) (see Methods) (Figure 1 and Additional files 5, 6, 7).

Our phylogenetic analyses of both sets of Wnt sequences found good support for the thirteen metazoan Wnt gene subfamilies and twelve protostome Wnt subfamilies, which corroborates the findings of several previous studies (Figure 1 and Additional files 5, 6, 7) 17181921. The phylogenetic assignment of Wnt genes from each organism to particular subfamilies is summarised in figure 2.

Our results show that the common ancestor of the arthropods possessed members of all Wnt subfamilies with the exception of Wnt3, supporting the previous suggestion that Wnt3 was lost in the lineage leading to protostomes 17. This is most strikingly evidenced by the identification of members of all the other twelve Wnt subfamilies in both Daphnia and Platynereis (Figures 1, 2 and Additional files 5, 6, 7).

Comparison of insects to other arthropods illustrates that the loss of Wnt genes appears to be more common among the insects, either through loss in the lineage leading to the insects, for example, Wnt2 and Wnt4, or losses in particular clades, for example, Wnt16 in holometabolous insects, Wnt11 in dipterans, and WntA in Drosophila (Figure 2). However there are probably also some cases of Wnt gene loss in non-insect arthropods, for example, Wnt10 may have been lost in chelicerates and myriapods (Figure 2), and we were unable to find a Wnt2 orthologue in Ixodes. In addition, we cannot exclude that there has been more extensive loss of Wnt genes in Glomeris as an alternative explanation to limitations in screening using degenerate PCR in this species.

In contrast to the patterns of Wnt gene loss, the presence of duplicates of Wnt genes appears to be less frequent. While we found duplications of both Wnt7 and Wnt11 in Achaearanea, no other duplications have yet been found in any other arthropod (Figure 2). Furthermore, although duplications of Wnt5, Wnt11 and Wnt16 are found in other lophotrochozoans, we found only single copies of each Wnt gene in Platynereis (Figure 2) 17.

Several Drosophila and Caenorhabditis Wnt genes have previously been described as 'orphan' genes, however, our phylogenetic analysis allowed us to assign these genes to specific subfamilies. We found strong support that Drosophila WntD is the Drosophila orthologue of Wnt8 1959 (Figures 1, 2, and Additional file 5). Moreover, while the Caenorhabditis Wnt genes Cwn-1, Cwn-2 and lin-44 were previously assigned to the Wnt4, Wnt5 and Wnt10 subfamilies respectively, the homology of mom-2 and egl-20 could not be determined 19. Our analysis supports the previous assignments of Cwn-1, Cwn-2 and lin-44, and furthermore indicates that mom-2 and egl-20 are probably Wnt9 and Wnt16 orthologues respectively (Additional file 5).

Analysis of the arrangement of Wnt genes on the Daphnia genome scaffolds revealed two syntenic clusters of these genes: Wnt9-Wnt1-Wnt6-Wnt10 and Wnt5-Wnt7 (Additional file 8). This is consistent with similar Wnt clusters in other metazoans, including Nematostella, and therefore reflects an ancient arrangement of Wnt genes in animals 1760. Indeed Lottia gigantea and Daphnia exhibit very similar organisation of these Wnt genes (Additional file 8). However, the precise organisation of these clusters can vary between lineages, for example, Wnt6 and Wnt9 are oriented differently in Drosophila and Daphnia (Additional file 8). Interestingly, these Wnt clusters may represent ancient duplications of Wnt genes; a hypothesis supported by the phylogenetic relationships of wg and Wnt6, and Wnt9 and Wnt10 in our study (Figure 1 and Additional file 5) and several previous studies 171819.

To further compare the Wnt genes among arthropods and annelids and to investigate the possible developmental roles of these genes, we characterised the expression of these genes in Achaearanea, Glomeris, and Platynereis and further characterised Wnt genes with segmentally reiterated expression in Tribolium 14. Note that the Drosophila Wnt gene names do not refer to homology with vertebrate Wnt subfamilies, but rather they were mostly named in the order they were discovered (e.g. DWnt2 is actually a Wnt7 orthologue not a Wnt2 orthologue). Therefore, below, we use the gene name with respect to its vertebrate orthologue and where appropriate give the Drosophila name in parenthesis, with the exception of wg (also see Additional file 1).

In Achaearanea, wg is expressed in stripes in the L1 and L2 segments, but only during stage 8, and such stripes are never observed in the other prosomal segments (Additional file 9: panel a). Subsequently, dots of wg expression associated with the developing limb buds are observed in all prosomal segments (Figure 3a, and Additional file 9: panel b). In the opisthosoma, At-wg is only expressed in the dorsal cells of the O2 and O3 segments (Figure 3c, d, and Additional file 9: panel b), and is not observed in the SAZ at any stage (Figure 3b). Later in development, At-wg expression continues in the prosomal appendages, and is also observed in opisthosomal limb buds, the labrum and the hindgut (Figure 3d, and Additional file 9: panel c).

In contrast, in a different spider, Cupiennius, wg is expressed at the posterior of each parasegment and in the SAZ 30 consistent with classic roles in segment addition and boundary formation as described in other arthropods such as Tribolium (Figure 4a, b) 333346162. Remarkably, this suggests that Achaearanea has either lost the expression and associated functions of wg in most segments and the SAZ or there is an additional paralogous wg gene in this spider not found in our PCR screen.

wg expression in Platynereis was previously described in 35. wg is expressed at the posterior boundary of each segment both in the trochophore larva (Figure 5a, b) and during posterior growth (Figure 6a). During annelid posterior growth, wg expression is also observed in the hindgut and in the posterior-most pygidial ectoderm (Figures 6b) 35.

It is likely that the Wnt2 subfamily was lost in the lineage leading to insects (Figure 2), and although we were unable to isolate an orthologue from Glomeris we assayed the expression of the Wnt2 genes from the spider Achaearanea and the annelid Platynereis.

In the spider Achaearanea, Wnt2 is first expressed relatively late in embryogenesis, in the ocular region of the developing cephalic lobes at stage 9, and this expression pattern persists into stage 10 (Figure 3e). We did not observe a distinct expression pattern for Wnt2 in Platynereis, possibly because of a low level of expression.

It is probable that the Wnt4 subfamily was also lost in the lineage leading to insects, but is present in other arthropods and lophotrochozoans (Figure 2). Analysis of Wnt4 expression in Achaearanea and Platynereis shows it is highly divergent between chelicerates and annelids. In Achaearanea, Wnt4 expression is restricted to only few cells at the very posterior of the germ band during the later stages of embryogenesis (Figure 3f). In contrast, in Platynereis, Wnt4 is expressed in stripes in the anterior part of each segment and could therefore be involved in defining segment boundaries (Figures 5c, d, 6l). The stripes are limited to the dorsal and lateral parts of nascent embryonic and post-embryonic segments. Additionally, Pd-Wnt4 is expressed in a longitudinal stripe along the ventral midline in forming segments as well as the SAZ and the ventral pygidial ectoderm (Figures 5c, d, 6l).

In Tribolium, Wnt5 is expressed in ventral stripes at the posterior of each parasegment and curiously in at least one row of cells in the anterior of each parasegment overlapping with en expression (Figure 4c, d). Tc-Wnt5 expression is also observed in the SAZ, distal tips of developing appendages, in the region of the labrum/stomodeum, and the ocular region of the head lobes (Figure 4c) 14.

In Achaearanea, Wnt5 is first expressed in a broad anterior domain (Additional file 9: panel d), and subsequently, in the cephalic lobes, throughout the SAZ, and segmentally in the developing neuroectoderm on either side the ventral midline, juxtaposed to en expressing cells (Figure 7a, b). At-Wnt5 transcripts can also be detected in a medial ring in the appendages, the labrum and the heart (Figure 7a). Similar expression patterns have been described for Wnt5 in Cupiennius 3063.

In Platynereis, Wnt5 is also expressed in clear segmental stripes. However, in contrast to Wnt5 expression observed at the posterior region of parasegments in arthropods, Pd-Wnt5 is only expressed in the anterior part of segments (Figures 5e, f, 6m-o, and Additional file 10: panel f). These stripes encompass both the ectoderm and the underlying mesoderm (Figure 5f). Pd-Wnt5 is also expressed weakly in the hindgut during posterior growth and in a complex pattern in forming appendages.

In Tribolium, Wnt6 is expressed in the developing brain, appendages, and in segmental stripes that overlap with en expressing cells (i.e. posterior to wg expression) (Figure 4e, f). Tc-Wnt6 is also expressed in a sub-terminal region of the SAZ (Figure 4e) 14. In Glomeris, Wnt6 is expressed in reiterated stripes in completed segments, directly anterior to en expression, and at later stages is observed in dorsal patches in each segment (Figure 8a, b) similar to Wnt6 expression in older Tribolium embryos. In addition, Gm-Wnt6 is expressed in specific domains in the developing brain, in and at the posterior of the germ band, including expression in the anal valves (Figure 8a). Diffuse expression of Gm-Wnt6 is also observed in the gut later in embryogenesis (not shown).

In Platynereis, Wnt6 is expressed in the mesoderm of trochophore larvae, and in the mesodermal layer of the intestine in the growing juvenile (Additional file 10: panels a and g).

In Tribolium, Wnt7 is expressed segmentally in two clusters of cells either side of the ventral midline abutting en expressing cells, essentially in a similar pattern to Tc-wg (Figure 4a, b, g, h) 14. Tc-Wnt7 is later expressed in the dorsal of the developing limbs and in the developing brain (Figure 4g) 14.

The two Achaearanea Wnt7 paralogues exhibit non-overlapping expression patterns similar to subsets of Tc-Wnt7 expression: At-Wnt7-1 is expressed only in the SAZ (Figure 3g), and At-Wnt7-2 is expressed in the proximal dorsal region of the developing appendages and in the developing brain (Figure 3h). However, neither of the spider Wnt7 genes or Glomeris Wnt7 is expressed in a segmental pattern like Tc-Wnt7. Gm-Wnt7 is expressed in older embryos at the posterior of the germ band in the anal valves, the brain, the heart, the midgut, the labrum, the mandibles, and possibly also weakly in the other developing appendages (Figure 8g). In Platynereis, Wnt7 is expressed in the mesoderm of the larva (not shown) and later during juvenile growth (Figure 6p). Pd-Wnt7 is also strongly expressed in the hindgut (Figure 6p).

In Achaearanea embryos, Wnt8 is expressed in the posterior most cells of the SAZ, the cephalic lobes, the developing stomodeum, the appendages, and in ectodermal stripes anterior to en in each segment (Figure 7c, d) 24. Although Wnt8 is also expressed in the SAZ of Tribolium embryos, it is not expressed segmentally in this beetle 1423.

In Glomeris, Wnt8 is expressed in two anterior domains and in the putative SAZ (albeit quite weakly), however, expression was only found in early embryos (Figure 8h) and no segmentally reiterated expression was observed.

Platynereis Wnt8 is expressed strongly in the future brain of the larva (Additional file 10: panel b). Faint ventral stripes are also detected in late stage trochophore larvae (Additional file 10: panel b), but no corresponding pattern is detected during posterior growth.

We were unable to isolate a Wnt9 gene from either spider species or Glomeris, but this may reflect a limitation of degenerate PCR rather than a loss in these lineages because a Wnt9 orthologue is found in the tick Ixodes (Figure 2). In Platynereis, Wnt9 is first expressed at the posterior pole in the trochophore larva (Additional file 10: panel c). During juvenile posterior growth, it is just observed in a few cells scattered in the gut endoderm (Additional file 10: panel h).

Again we were unable to isolate a Wnt10 gene from either Achaearanea or Glomeris and in addition no Wnt10 orthologue was found in the Ixodes gene models (Figure 2). In Tribolium, Wnt10 is expressed in a similar pattern to wg in the cephalic lobes, appendages and at the posterior parasegmental boundaries abutting en expression (Figure 4i, j) 14. Similar to wg/Wnt1 expression in Platynereis, Pd-Wnt10 is expressed at the posterior boundary of each segment and in the hindgut during posterior growth (Figure 6c, d). Surprisingly, we did not observe a similar expression pattern in the trochophore larva perhaps due to probe detection limitations. Instead, two pairs of cells were stained presumably in the anterior larval mesoderm (Additional file 10: panel d).

There are two Wnt11 genes in Achaearanea. While we did not detect any embryonic expression of the Wnt11-1 paralogue, At-Wnt11-2 is expressed in the SAZ starting at stage 6 (Additional file 9: panel e) and then throughout segmentation (Figure 3i). At-Wnt11-2 is also expressed in the developing appendages in an anterior domain along the proximo-distal axis of the prosomal appendages and in a distal domain in the buds of opisthosomal appendages (Figure 3i). In Glomeris, Wnt11 is first expressed at the posterior of the germ band (Figure 8i), and later in the anal valves, and at the tips of each appendage (Figure 8j). In the maxillae three spots of expression are also observed that resemble the expression of wg in Glomeris (Figure 8j).

In Platynereis, Wnt11 is strongly expressed in segmental stripes in the posterior part of each segment in a similar position to wg in the larva (Figure 5g, h) and during juvenile growth (Figure 6e-g). Pd-Wnt11 is also expressed in the brain (not shown), the stomodeum and the presumptive pygidium (Figure 5h). During posterior growth, it is also strongly expressed posteriorly, but in the ectoderm covering the pygidium at the base of the tentacular cirri rather than in the hindgut like other Wnt genes (Figure 6e).

Investigation of Wnt16 expression in Achaearanea, Glomeris and Platynereis showed that in all three of these animals Wnt16 is expressed in segmental stripes directly anterior to en (Figures 5i, j, 6h-j, 7e, f, 8c, d). Thus like wg, Wnt16 might be involved in the generation of segmental and parasegmental boundaries in annelids and arthropods respectively (perhaps with the exception of holometabolous insects, see figure 2). In nascent segments of Glomeris and Achaearanea embryos, Wnt16 is observed in ventral restricted stripes (Figures 7e, f, 8c). However, in older segments Wnt16 is expressed in stripes either side of the ventral midline (Figures 7e, f, 8c, d). Wnt16 expression is also observed in the cephalic lobes and the distal tips of the appendages in the spider and millipede (Figures 7e, 8c).

In Platynereis, Wnt16 is expressed in segmental stripes just at the posterior border of segments (Figure 5i, j, 6h-k). Interestingly, the trochophore larvae show five stripes of Wnt16 of unequal strength, in addition to the three parapodia bearing larval trunk segments delineated by the other Wnt genes. Pd-Wnt16 is also expressed in the peristomium (the "ring" that carries the mouth just below the prototroch in annelids) and in a transient segmental anlage just posterior to it. Both segment-like structures fuse with the head at metamorphosis. During posterior growth, Pd-Wnt16 is also expressed in the pygidium mesoderm, but not in the hindgut or pygidial ectoderm like other Wnt genes (Figure 6k).

Analysis of the expression of WntA orthologues in Tribolium, Cupiennius and Platynereis again revealed quite different patterns for this Wnt subfamily across protostomes. In Tribolium, WntA is expressed in the head lobes, appendages, SAZ and segmental stripes (Figure 4k, l) 14. The segmental expression of WntA in Tribolium is again found anterior to en in a similar domain to wg (Figure 4l). Glomeris WntA is expressed in clusters of cells in the ventral neuroectoderm posterior to en expressing cells, at the posterior end of the germ band (weakly) and developing heart (Figure 8e, f) 33. Expression of WntA is also observed in the SAZ of the spider Cupiennius (Additional file 9: panel g), and although we also observed expression in a distal domain in the spinnerets and a lateral spot in the cheliceres, WntA is not expressed segmentally in this spider (Additional file 9: panels g-i). Thus WntA expression is rather different between mandibulates and chelicerates.

In Platynereis, Pd-WntA is strongly expressed in the parapodial anlagen in larvae and during posterior growth (Additional file 10: panels e and i). Pd-WntA expression is later observed at the distal extremities of growing parapodia (Additional file 10: panel i). A striped expression in the mesoderm during posterior growth has probably no connection to segment formation as high magnification shows that these stripes correspond to the walls of lateral blood vessel branching from the dorsal and ventral blood vessels (Additional file 10: panel j).

It has been shown that the thirteen subfamilies of Wnt genes found in metazoans appeared before the evolution of bilaterians, and that thirteen and twelve subfamilies are represented in extant deuterostomes and protostomes respectively 171819202122 (Figure 2). Strikingly we have now found twelve Wnt subfamilies in both an arthropod, the crustacean Daphnia, and in the annelid Platynereis confirming that the common ancestor of protostomes contained all Wnt subfamilies except Wnt3. Furthermore, our study, the first broad survey of Wnt gene diversity across arthropods, shows that the common ancestors of arthropods and ecdysozoans also contained representatives of all twelve Wnt subfamilies found in protostomes (Figures 1, 2, and Additional file 5).

In insects there has been extensive loss of Wnt genes, for example, only seven and six Wnt genes are found in Drosophila and Acyrthosiphon respectively 1416. This reflects the absence of Wnt2 and Wnt4 in all insects and lineage specific patterns of loss such as Wnt11 in dipterans (Figure 2). Moreover, this suggests that while the loss of Wnt genes has been common in insects and the nematode Caenorhabditis, most ecdysozoans may actually have retained a larger repertoire of these genes (Figure 2). Similarly, the leech, Helobdella, also appears to have lost a number of Wnt genes with respect to other lophotrochozoans like Capitella 17 and Platynereis. However the reasons for retention of a large repertoire of Wnt genes in some lineages and extensive loss in others is currently unknown.

Curiously duplications of individual Wnt genes (i.e. apart from those generated by whole genome duplications in deuterostomes) are rather rare (Figure 2). The reason for this could be that the concentration of individual Wnt ligands is important for the overall combinatorial output of Wnt signalling in particular tissues (see below). Indeed, in animals with Wnt duplications, the paralogues appear to have been subject to subfunctionalisation, as evidenced by distinct expression patterns of the Wnt7 and Wnt11 paralogues in the spider, and duplicated Wnt genes in lophotrochozoans 17. Our data also support previous phylogenetic studies of Wnt genes suggesting ancient duplications may have given rise to clusters of Wnt genes, such as the Wnt9-wg-Wnt6-Wnt10 cluster found in Daphnia and other metazoans (Additional files 5 and 8).

Our present study of Wnt gene expression in a range of arthropods and an annelid, and previous studies in other metazoans 141718216465, show that numerous Wnt genes are often expressed in the same cells or tissues; for example, various Wnt genes are expressed in the SAZ and at the same position within segments (Figure 9). Does this imply that Wnt ligands are essentially redundant? The lack of obvious phenotypic effects associated with the loss of expression of some Wnt genes in particular tissues suggests that they may be functionally interchangeable in certain contexts [e.g. 2366]. However, there are also several arguments against the general functional redundancy of these ligands. First, the fact that twelve or thirteen Wnt genes are retained in many animals argues against redundancy. Second, since Wnt ligands diffuse from source cells and thus can act on a range of different target cells, expression of multiple Wnts in the same cell does not necessarily mean they have the same function. Third, studies directly comparing the function of different Wnts have provided direct experimental evidence that these ligands are functionally distinct. In Drosophila wg and Wnt9 (DWnt4) have similar expression in segmental stripes, but they play different roles in ectodermal patterning 567, while over-expression of the other five Drosophila Wnt genes has no affect on cuticular patterns 6468. Furthermore, Llimargas and Lawrence 64 found that wg and Wnt7 (DWnt2) act together during Drosophila tracheal development, but none of the five other Drosophila Wnts could perform the same roles. These results, as well as those of studies in Caenorhabditis [e.g. 6566], reflect increasing evidence that Wnt signalling is more complex than simple linear signalling pathways, and that Wnt ligands expressed in similar patterns may work agonistically and antagonistically to fine tune cellular responses 42. Indeed, it is perhaps even more realistic to think of an overall Wnt ligand landscape or code rather than the function of individual Wnts 69.

The specificity of Wnt signalling is also facilitated by the great complexity of transduction mechanisms employed 42. Wnt ligands are capable of binding to several different receptors, including 7-pass Frizzled receptors and the receptor tyrosine kinases Ryk and Ror, which in turn are capable of activating several cross-talking cytoplasmic pathways. It has been proposed that these transduction mechanisms allow a combinatorial action of Wnt ligands in particular tissues 42. Interestingly, however, this also opens the possibility that a given cellular response might be achieved with several different Wnt ligand combinations. Therefore, the expression of alternative Wnt combinations in a given tissue in different taxa could still generate the same intracellular signalling outcome. This may partly explain the diversification Wnt ligand gene expression across metazoans.

Our analysis allows the first broad comparison of Wnt expression patterns across the arthropods, and our characterisation of Wnt expression in an annelid extends this comparison to other segmented protostomes.

Arguably, one of the most interesting observations emerging from this comparison is the high proportion of Wnt genes expressed in segmental stripes reminiscent of segment polarity gene expression in Drosophila (Figure 9). In fact, no less than six Wnt genes show this kind of pattern in Platynereis and Tribolium, and at least five in Glomeris and four in Achaearanea (Figure 9). It is particularly striking that eleven out of twelve protostome Wnt genes (the exception being Wnt2) exhibit a striped pattern in at least one species, and no less than nine (Wnt11 and Wnt4 are the exceptions) in at least one arthropod (Figure 9). Generally, these stripes appear before the morphological appearance of segments, suggesting that some of these genes may play roles in segment formation, although others may only be involved in the ontogenesis of segmental organs rather than segmental patterning. The last common ancestor of all arthropods was undoubtedly a metameric animal, and our study suggests that a number of Wnt genes probably played a role in the patterning of its segments. However, only wg is expressed in similar stripes across all arthropod species considered here, and even the function of this gene may have changed somewhat in Achaearanea. Therefore, some Wnt genes have lost their segmental expression in some lineages, and indeed, Wnt16 was lost altogether in holometabolous insects (Figure 2). Conversely, some Wnt genes may have evolved segmental patterning functions, for example, Wnt7 in Tribolium.

Experimental approaches have also revealed differences among Wnt genes with respect to their role in segmental patterning. In Drosophila, only wg and Wnt9 (DWnt4) appear to regulate the establishment of the metameric pattern 57646870. In Tribolium, while wg RNAi produces segmentation defects 23, RNAi against other segmentally expressed Wnt genes did not affect segmentation. Clearly functional data on the other Wnt genes in arthropods, particularly non-insect arthropods, is required to investigate the roles of these genes in segmental patterning further.

Despite differences in the expression and probably the function of Wnt genes across taxa, there are nevertheless some noticeable similarities: Wnt5 is expressed in ventral stripes in Tribolium, Achaearanea and Platynereis; Wnt16 is expressed in reiterated stripes in Achaearanea, Glomeris and Platynereis; Wnt10 forms stripes in Tribolium and Platynereis (Figure 9). Furthermore, with a few exceptions, segmental expression of Wnt genes nearly always anteriorly abuts en expression in arthropods (Figure 9). Together with the fact that no Wnt gene demarcates the segmental boundary in arthropods (with the possible exception of Tribolium Wnt6), this vindicates the view that parasegment boundaries are the essential organizers of segmental patterning in these animals 3031.

It was previously proposed that the ancestral protostome was an annelid-like segmented worm, and that arthropod cuticular segmentation evolved out of frame with the ancestral segmentation 71. This is supported by en, wg 35 and hedgehog 36 expression patterns in Platynereis. In this view, arthropod parasegments are an embryonic recapitulation of ancestral segmentation. In Platynereis, although incomplete stripes of Wnt4 and Wnt5 are found in the anterior region of segments, wg, Wnt10, Wnt11 and Wnt16 are all expressed in circular stripes at the posterior segmental boundaries, anterior to en (Figure 9), thus supporting the hypothesis that arthropod parasegments and annelid segments are homologous. It is noteworthy, however, that the analysis of the expression patterns of a complete set of Wnt genes in another annelid, the leech Helobdella, led the authors of this study to very different interpretations 17. In the leech, the duplicated genes Wnt11a, Wnt11c, Wnt16a and Wnt16b also give striped segmental patterns but only in the late germ band stage well after the segmental pattern is already laid down, whereas in Capitella, Wnt5, Wnt11 and Wnt16 are not expressed in ectodermal stripes but rather in segmentally iterated patterns in the mesoderm 17. Only Capitella Wnt11 is expressed transiently in the ectoderm of the SAZ. These discrepancies show that the actual role of Wnt signalling in segment formation will have to be tested in detail in non-insect arthropod and annelid models before reaching conclusions on its evolution.

Drosophila undergoes a long germ band mode of development, where all segments are formed simultaneously. In contrast most insects and other arthropods develop through variations of the short germ band mode, which is more ancestral. In the short germ band mode of development, only the anterior-most segments are initially specified and subsequently the posterior segments are added sequentially from unsegmented posterior tissue, which is often called a posterior growth zone 7273. However, even within arthropods, the term "growth zone" encompasses a diversity of tissue types that use different combinations of cell proliferation, movement and differentiation to generate new segments 40. Therefore, the 'growth zone' may be more appropriately named a segment addition zone (SAZ) because sequential addition of segments is truly the key common process involved 314041. Despite differences in the process of segment addition among arthropods, it has also been argued that this is an ancestral character of bilaterians 41.

A large proportion of Wnt genes in Tribolium, Achaearanea and Glomeris embryos are expressed in the SAZ (Figure 9). The crucial role played by Wnt signalling during segment addition has been functionally demonstrated in a few arthropods. Wnt8 knockdown in both Tribolium and in Achaearanea resulted in a posterior truncation of the body 2324. A similar phenotype is obtained in Oncopeltus with wg RNAi 25, but not in Tribolium 23 or Gryllus 74, despite expression of wg in the SAZ of this beetle. This suggests that the respective roles of Wnt ligands during segment addition have evolved differentially among arthropod lineages, and is consistent with differences in the expression of Wnt genes in this region (Figure 9).

Axis truncations produced by depletions of armadillo/β-catenin, pangolin/TCF and arrow/LRP5/6 in Gryllus, Tribolium and Oncopeltus 23257475 further evidence the crucial role played by the β-catenin pathway in segment addition. Nevertheless, given the multiplicity of ligands involved, it will be important to investigate whether posterior addition of segments in arthropods is regulated by Wnt ligands through combinatorial transduction pathways 42.

Analysis of posterior expression of Wnt genes in the annelid Platynereis brings some valuable insight to understanding segment addition in protostomes. No less than six Wnt ligands are expressed in the terminal region of the annelid body, the pygidium, during axis elongation. However the annelid SAZ is located anterior to the pygidium and is represented by a thin ring of cells in which even-skipped and caudal (cad) are involved in regulating the synchronous mitotic cycles that produce new segments 41. The posterior expression domains of Platynereis wg, Wnt5, Wnt7, Wnt10, Wnt11 and Wnt16 cannot completely be superimposed because they cover the hindgut, the external pygidial ectoderm, and the pygidium mesoderm. However none of these Platynereis Wnt genes is actually expressed in the SAZ sensu stricto, suggesting that they act from a posterior signalling centre located in the mitotically quiescent pygidium and separate from the proliferating cells that are the source of the new segments 41.

In the short germ band arthropods considered in this work, the detailed organization of the SAZ is largely unknown and therefore it is not known if there is a separate segment founder cell zone and putative signalling centre that differentially express Wnt genes. Clearly some arthropod Wnt genes are expressed in the proctodeum towards the end of embryogenesis, and thus in a location homologous to the annelid hindgut. Interestingly, the posterior expression of wg in an arthropod with anamorphic development (segments are added during larval development), the crustacean Triops, shows two separate domains: a complete ring near or in the SAZ and the hindgut 62.

It has been shown that knockdown of the posteriorly expressed Wnt8 in a spider perturbs the posterior expression of cad and Delta/Notch pathway components 24. Given similar observations in several vertebrates 76777879, a Wnt signalling centre acting upstream of cad and the Delta/Notch pathway may have regulated posterior development in the last common ancestor of bilaterian animals (Urbilateria) 8081. This interpretation is further strengthened by the arthropod expression data in our study. Moreover, we also found evidence for a posterior Wnt signalling centre in a distantly related protostome group, the annelids, in which cad and Delta/Notch are also involved in posterior addition 418283. However, the evolution of posterior Wnt signalling has likely been complex in bilaterians, for example, Wnt8 is not expressed at this location in annelids and therefore its role must be played by one or several other Wnts ligands in these animals.