Fz sequences were recovered from published resources: Euperipatoides kanangrensis (Janssen and Budd, 2013); Glomeris marginata (Janssen and Posnien, 2014); Parasteatoda tepidariorum (Posnien et al., 2014); Mesobuthus martensii (Cao et al., 2013); Ixodes scapularis (Lawson et al., 2009): Tribolium castaneum (Richards et al., 2008); Zootermopsis nevadensis (Terrapon et al., 2014); Strigamia maritima (Chipman et al., 2014). Pholcus phalangioides sequences were recovered from a de novo assembled transcriptome, generated from a combination of different embryonic stages and three larval tissues including legs, the prosoma, and the opisthosoma. Total RNA for larval tissue was extracted using the ZR Tissue & Insect RNA MicroPrep™ Kit (Zymo Research) and total RNA extraction of embryonic samples was performed using Trizol. The library for Illumina sequencing was generated using the standard protocol of the TruSeq RNA Sample Preparation Kit (v2, Illumina). All samples were pooled and sequenced on one lane of an Illumina Hiseq2000 sequencer, which resulted in 472,770,758 paired-end reads of 100 bp. The reads were quality trimmed and filtered as previously described (Posnien et al., 2014). All high quality reads were assembled into 444,103 transcripts using the de novo transcriptome assembler Trinity (version r2013_11_10, Grabherr et al., 2011). The GC content of this preliminary de novo assembly is 37.27%. The N50 is 1517 bp with a median transcript length of 284 bp and an average transcript length of 670 bp.

To infer phylogenetic relationships, predicted Fz amino acid sequences were first aligned using MUSCLE (Edgar, 2004; Sievers et al., 2011) after which poorly aligned sites were masked using Gblocks v. 0.91b (Castresana, 2000) with relaxed selection criteria, both as implemented in Seaview v. 4.5.3 (Gouy et al., 2010). Maximum-likelihood analysis was performed using RAxML v. 8 (Stamatakis et al., 2005), specifying an LT model of amino acid replacement and a gamma model of rate heterogeneity (+G) as selected by ProtTest v. 3.2 (Darriba et al., 2011). Nodal support was estimated with 500 bootstrap iterations.

All gene fragments were isolated by means of PCR with gene specific primers. Fragments were cloned into pCR-II-TOPO or pCR4-TOPO vectors (Life Technologies). Sequences are available under accession numbers LN849082 (Ek-fz1), LN849083 (Ek-fz2), LN849084 (Ek-fz3), LN849085 (Ek-fz4), LN849078 (Gm-fz1), LN849079 (Gm-fz2), LN849080 (Gm-fz3), LN849081 (Gm-fz4). The Parasteatoda sequences are all available from the Assembled Searchable Giant Arthropod Read Database (ASGARD) (http://asgard.rc.fas.harvard.edu/) (Zeng and Extavour, 2012): Pt-fz1 (Locus 3605), Pt-fz2 (Locus 1), Pt-fz4-1 (Locus 7239), and Pt-fz4-2 (Locus 2608).

Mature females of Glomeris were collected from February to May in the Reichswald forest (Nordrhein-Westfalen, Germany) near Kranenburg. They were kept in plastic boxes filled with decomposing beech leaves and moist soil at ~20°C. Under these conditions females lay eggs for several weeks. Pregnant females of Euperipatoides were sampled between August and October in the Kanangra Boyd National Park, NSW, Australia, and were then kept at ~8°C in plastic containers filled with moist moss. Under these conditions they survive for at least several months when fed with small crickets once a week. The Parasteatoda culture at Oxford Brookes University was founded by spiders from a strain collected and maintained in Göttingen. Adults were kept separately in plastic vials at 25°C and fed with crickets. Pholcus spiders were obtained from the laboratory stock in Göttingen (Pechmann et al., 2011) and maintained like Parasteatoda, except that adults were fed with large flies like Musca. Embryos were collected, fixed and stored as described in Janssen et al. (2004) for Glomeris, in Janssen et al. (2015b) for Euperipatoides, and in Akiyama-Oda and Oda (2003) for Parasteatoda with slight modifications. Glomeris, Euperipatoides, and Parasteatoda embryos were staged according to Janssen et al. (2004); Janssen and Budd (2013), and Mittmann and Wolff (2012), respectively.

In situ hybridization of Euperipatoides and Glomeris embryos was performed as described in Janssen et al. (2015b) and Prpic and Tautz (2003), respectively. Embryos were hybridized with Digoxigenin-labeled RNA probes at 62°C for at least 16 h. Parasteatoda in situ hybridization was carried out after the whole-mount protocol for spiders (Prpic et al., 2008) with minor modifications. Primers used to generate probes are available on request.

Nuclear staining was performed by incubation in 1 μg/ml 4-6-diamidin-2-phenylindol (DAPI) in phosphate-buffered saline with 0.1% Tween-20 for 30 min.

Embryos were analyzed under a Leica dissection microscope equipped with a Leica DC100 (Glomeris and Euperipatoides) or with a Jenoptik ProgRes C3 (Parasteatoda) digital camera. Linear corrections of brightness, contrast, and color values were performed with Adobe Photoshop.

Coding sequences of all putative Fz receptors recovered from genomic or transcriptomic resources for 10 panarthropod species were mostly complete. Exceptions include the Mm-fz1 sequence, which appears to be truncated after transmembrane domain 5. Gm-fz2, Zn-fz3, and Mm-fz4-1 sequences are possibly incomplete at the N-terminus (Figure S1).

Phylogenetic reconstruction using maximum likelihood recovered four distinct orthology groups: FzI, FzII, FzIII, and FzIV (Figure S2). This is in concordance with previous findings in other metazoans (Schenkelaars et al., 2015). To avoid insect overrepresentation and potential disrupting effects of the derived Dm-fz3 and Dm-fz4 sequences, Drosophila sequences were omitted from our final analysis (Figure 1). This had little influence on the topology of FzI, II, and IV orthology groups, but led to clustering of the Glomeris Gm-fz3 sequence with the other non-onychophoran sequences within the FzIII orthology group.

In all investigated species, a single FzI ortholog was identified. FzII orthologs are present in the genomic resources of all species, except for the genome of the scorpion Mesobuthus. FzIV orthologs were also found in all investigated species. Moreover, it appears that a duplication of FzIV can either be traced to the base of the Arachnopulmonata (Scorpiones and Tetrapulmonata—including spiders) (Sharma et al., 2014), or happened independently in spiders and scorpions (Figures 1, 2). The only orthology group that showed gene loss in representatives of both the Chelicerata and Mandibulata was FzIII (Figures 1, 2).

Fz receptors contain a conserved amino-terminal extracellular region that mediates binding to Wnt ligands: the cysteine-rich domain (CRD) (Figure 3 and Figure S1). The highest sequence similarity among panarthropod Fz CRD domains was found in the FzI orthology group (only nine sites with < 60% consensus), whereas the FzIII CRD sequences are most divergent (30 sites with < 60% consensus) (Figure 3). Moreover, the Drosophila Dm-fz1, Dm-fz3 and Dm-fz4 sequences stand out due to their relatively large number of deviations from the consensus of their respective orthology groups (Figure 3). This is consistent with long branches observed for these Drosophila sequences in our phylogenetic reconstruction (Figure S2), and is an indication that insights gained from functional studies of Fz receptors in Drosophila might not be representative of their function in other panarthropods.

A second functional domain is the KTxxxW motif, which is typically found directly after the seventh transmembrane domain (Huang and Klein, 2004). In Drosophila and vertebrates, this intracellular motif has been associated with binding to proteins of the Disheveled family. Alignment of the KTxxxW motif from panarthropods (Figure 3, positions 548–566) shows high sequence similarity in the FzI/II orthology groups, and less similarity for the FzIII/IV sequences. Interestingly, as with Drosophila Dm-fz3 and Dm-fz4, the tryptophan of this motif has been lost in both of the spider Parasteatoda Pt-fz4 paralogs, which could indicate a change or loss of the binding affinity of this motif in these two proteins. The tryptophan residue has, however, been retained in most other panarthropods FzIII/IV sequences, including the spider Pholcus (Figure 3 and Figure S1).

We identified orthologs of the FzI subfamily in Euperipatoides as well as the two arthropods, Glomeris, and Parasteatoda (Figures 1, 2).

In early embryonic stages of Euperipatoides, Ek-fz1 is expressed ubiquitously except for the posterior segment addition zone (SAZ), but is then upregulated in a segmental pattern in transverse stripes (Figure 4A). In the head lobes, expression is observed in a patch-like domain at the posterior margin of each of the two hemispheres (Figures 4A,B). Later, expression disappears from the tip of the outgrowing appendages (Figures 4C,D) and cells in the nervous system express Ek-fz1 at higher levels (Figure 4C). This expression profile persists through to later developmental stages (Figure 4E).

Gm-fz1 is expressed ubiquitously and uniformly at all investigated stages of embryogenesis in Glomeris (Figures 5A–D). In Parasteatoda, Pt-fz1 is also probably ubiquitously expressed at low levels, but the most prominent expression of Pt-fz1 expression starts at stage 9.1 in the ventral neuroectoderm (Figure 6), which then persists through subsequent stages. However, no specific spatial expression of Pt-fz1 was detected before or after this stage of embryogenesis in Parasteatoda.

In the early embryos of both Drosophila and Tribolium, the fz1 genes are initially uniformly expressed (Park et al., 1994; Müller et al., 1999; Beermann et al., 2011). While this ubiquitous expression persists in Tribolium embryos (Beermann et al., 2011), the expression of Drosophila Fz resolves into broad stripes between the Engrailed (En) expressing cells (Müller et al., 1999). In both of these insects, Fz appears to function together with the respective Fz2 receptor to regulate segment formation via the Wingless (Wg) ligand (Chen and Struhl, 1999; Müller et al., 1999; Bolognesi et al., 2008; Beermann et al., 2011). The expression of fz1 genes in Euperipatoides, Glomeris and Parasteatoda is consistent with a role for fz1 genes in segmentation across the panarthropods.

Drosophila fz is also expressed in other tissues including the imaginal discs and central nervous system (Park et al., 1994). However, fz mutants can produce viable adult flies that only exhibit defects in planar cell polarity (reviewed in Strutt, 2003). This suggests that, like during segmentation (see above), this receptor functions in combination with other Fz proteins in a range of roles in Drosophila, and given its widely distributed expression in Euperipatoides, Glomeris, and Parasteatoda, this is probably the case in other panarthropods.

We identified orthologs of the FzII subfamily in all three of the panarthropods that we investigated (Figures 1, 2).

In Euperipatoides, Ek-fz2 is strongly expressed during embryogenesis in the ventral region of the head lobes, including the mouth that forms from the anterior of the mouth–anus furrow (Janssen et al., 2015a). Interestingly, tissue anterior to the head lobes proper also expresses Ek-fz2 (Figures 4F,F′). This remains the only detectable expression (Figures 4G,H) until stage 20 when dot-like domains appear in an anterior to posterior order at the base of the walking limbs, and in a small domain ventrally at the base of the third and fourth walking limb (Figure 4I). The latter expression may be correlated with the development of the nephridia, which are more pronounced in these two segments (Mayer, 2006).

In Glomeris, Gm-fz2 is expressed in an anterior domain covering ~30% of the early post-blastoderm stage embryo (Figure 5E). At later stages this domain includes the anterior and median region of the developing head lobes including the stomodaeum (Figures S3A–C). At approximately stage 0.5, a faint transverse stripe of expression appears (Figure S3A). This stripe likely corresponds to expression in the mandibular segment as is obvious in later stages (cf. Figures S3A,B). Within the mandibular segment, expression is restricted to the anterior of the developing mandibles (Figure 5F and Figures S3B,C). The dorsal segmental tissue that forms at stage 3 (Janssen et al., 2004) expresses Gm-fz2 in a complex pattern (Figures 5F–H and Figure S3C), and Gm-fz2 is also expressed in the midgut from stage 5 onwards (Figure 5G).

In Parasteatoda, Pt-fz2 expression commences at stage 5 in a broad ring around the germ disc (Figure 6B). After radial symmetry is broken (Akiyama-Oda and Oda, 2003), Pt-fz2 continues to be expressed in an anterior stripe (Figure 6C), which widens at a later stage, but the expression intensity remains comparable. At stage 8.1 Pt-fz2 appears to be expressed at the anterior border of the germ band and in the forming L1–L4 segments, although the expression in the L2–L4 segments is much broader than in L1 (Figure 6D). At stage 8.2, when prosomal segmentation has become morphologically visible and the first opisthosomal segment (O1) has formed, Pt-fz2 is expressed in the anterior compartment of each segment and in the segmental grooves (see Figure 6E). There is also a narrow Pt-fz2 expression domain at the anterior margin of the germ band (see Figure 6E). However, no expression is observed in the SAZ. Subsequently, Pt-fz2 expression in each segment contracts to the segmental grooves, and the anterior-most band expands to a wedge shaped domain in the anterior precheliceral lobes, partially encompassing the stomodaeum (Figure 6F). Strong expression is also observed in the ventral neuroectoderm and the dorsal periphery of each segment (see Figure 6F).

In Drosophila embryos, Dfz2 is also initially expressed in a broad domain but is absent from the anterior and posterior poles (Bhanot et al., 1996). Dfz2 expression then develops into broad segmental stripes posterior to En expressing cells, but unlike Fz, this expression does not quite extend as far as the wg expressing cells (Bhanot et al., 1996; Müller et al., 1999).

In Tribolium, like Drosophila, Glomeris, and Parasteatoda, Tc-fz2 is also initially expressed in a broad presegmental region that resolves into segmental stripes anterior to wg expression (with the exception of the more complex pattern of Gm-fz2), and like in Parasteatoda and Glomeris, expression of Tc-fz2 in Tribolium is absent from the SAZ (Beermann et al., 2011).

Taken together, comparison of the expression of fz2 genes across arthropods suggests that this member of the Fz family was employed in segmentation in the common ancestor of these animals. In developing insect segments it is most likely that Fz2 is bound by Wg, however, in Parasteatoda the most probable candidates are Wnt5, Wnt8, and Wnt16 because wg is not expressed in an obvious segmental pattern in this spider (Janssen et al., 2010). Moreover, given the lack of a segmental pattern of expression of fz2 in Euperipatoides (Figure 4), it remains unclear if this gene played a similar role in the ancestor of panarthropods. It is also evident from our results that the anterior expression of fz2 in the median head region is conserved in Glomeris, Euperipatoides, Tribolium, and Parasteatoda and therefore probably represents a conserved ancestral expression domain of this gene.

We identified orthologs of the FzIII subfamily in both Euperipatoides and Glomeris, but like Tribolium, Parasteatoda lacks a fz3 gene (Figures 1, 2).

Ek-fz3 is expressed segmentally in the form of transverse stripes (Figures 7A,C). These stripes are level with the position of the limb primordia (Figure 7A). The tips of outgrowing limbs, however, are free from expression (Figures 7D,E). Transcription in the head lobes is restricted to lateral tissue (Figures 7A,B,E) and complements that of Ek-fz2. Expression in the trunk segments complements that of Ek-fz4 (see below). The SAZ does not express Ek-fz3, but a faint signal is visible in the anus (Figures 7D,E). At later developmental stages, expression is seen in the mouth (Figure 7E).

In Glomeris we did not detect any specific staining for fz3 in embryos younger than stage 3. From this stage onwards, Gm-fz3 is expressed in the form of two dots in each eye field (Figures 5I,J). At later developmental stages, expression appears in the lateral region of the anal valves (Figures 5K,L). Detection of Gm-fz3 transcripts requires prolonged staining time, therefore, it is unclear whether this gene is expressed ubiquitously at a low level, or if this weak signal represents background staining. It is possible that Gm-fz3 is also expressed in the developing gut (Figure 5).

The expression of fz3 has been investigated previously in Drosophila, but it is not completely straightforward to relate these patterns to those we have observed in Glomeris and Euperipatoides. Drosophila fz3 is expressed in segmentally reiterated stripes during later embryogenesis like that observed in Euperipatoides, as well as, for example, in the brain, eye and leg discs, and anal tissue like Glomeris and/or Euperipatoides (Sato et al., 1999; Sivasankaran et al., 2000). This may indicate that fz3 played an ancestral role in tissues like the nervous system, eye, and appendage development, however, a survey of a wider range of panarthropod species that still possess a member of the FzIII subfamily is required to address this more fully.

We identified single copies of FzIV subfamily genes in Euperipatoides and Glomeris, and two fz4 paralogs in Parasteatoda (Figures 1, 2).

The early expression of Ek-fz4 is comparable to that of Ek-fz1. Ek-fz4 is expressed in a segmental fashion. Within the head lobes, the primordia of the frontal appendages, however, are free from expression as are the eyes and a wedge-shaped domain ventral to the eyes (Figure 7F). At later stages it becomes clear that segmental expression is initially located between the outgrowing limbs (Figure 7G). Expression in the limb buds is restricted to anterior and posterior proximal tissue (Figures 7G–I). Tissue ventral to the base of the limbs no longer expresses Ek-fz4 at these later stages, or only weakly (cf. Figures 7H,I). As the ventral nervous system grows out in the process of ventral closure, tissue ventral to the bases of the limbs remains free of transcripts (Figure 7J).

At early stages, Glomeris Gm-fz4 is expressed in all tissues except for a distinct region in the anterior head, possibly the eye-field (Figure 5M). Expression is then upregulated in segmental stripes in the SAZ and newly formed segments, but the anal valves remain free of expression throughout development (Figures 5M–P).

Interestingly, Gm-fz4 is also expressed in the dorsal extraembryonic tissue in form of an anterior cap (Figures S3H–J). At stage 3, the anterior of the head lobes are free of expression. With the exception of the labrum, there is no detectable expression in the distal part of all appendages (Figures S3D–G). At all later stages, Gm-fz4 is strongly expressed in the midline (Figures 5N–P), but only at around stage 3 do longitudinal stripes of expression appear on either side of the midline (Figure 5N). In dorsal tissue, two domains per segmental unit exhibit recognizably higher expression (Figures 5N–P).

In Parasteatoda, Pt-fz4-1 expression commences at stage 8.2 in a stripe at the posterior of the O2 segment and in a stripe at the anterior portion of the SAZ (see Figure 8A). Faint Pt-fz4-1 expression can also be detected in the mesoderm of the forming limbs and in a distinct domain on each of the precheliceral lobes, at stage 9.1 (Figure 8B). Subsequently, Pt-fz4-1 expression becomes stronger and more broadly expressed in the limb mesoderm and expands also in the ventral neuroectoderm (see Figure 8C). The expression in the precheliceral lobes continues during stage 9.2 (see Figure 8C). Then during stage 12, Pt-fz4-1 is strongly expressed in the limb mesoderm, in the labrum and in the mesoderm of the opisthosomal segments (see Figure 8D). As the brain differentiates, expression refines to the anterior border of the precheliceral lobes: the anlage of the prosomal shield that overgrows the brain in the following stages (Wolff and Hilbrant, 2011; Mittmann and Wolff, 2012) (Figure 8D).

Pt-fz4-2 expression arises in an anterior stripe at stage 6 (Figure 9A). Compared to the anterior Pt-fz2 stripe at a similar stage (see Figure 6C), this Pt-fz4-2 expression domain is narrower (Figures 9A,B). Later, at stage 8.2, Pt-fz4-2 is expressed in the segmental grooves in the prosomal and opisthosomal segments and in a ring around the future labrum (Figure 9C). The expression of Pt-fz4-2 retracts to the dorsal periphery of each segment at stage 9.2 and the domain around the labrum becomes more defined (Figure 9D). Pt-fz4-2 is also strongly expressed in the limb and opisthosomal mesoderm at stage 12 (Figure 9E). Anteriorly, Pt-fz4-2 expression continues in the labrum, and also appears in a domain directly anterior to it (Figure 9E).

The two fz4 paralogs in Parasteatoda exhibit both similarities and differences in their expression. For example, both are expressed in the labrum and in similar patterns in the extended walking legs, but fz4-1 is expressed earlier in the nervous system and head lobes, while fz4-2 is expressed segmentally including in anterior segments (Figures 8, 9). Moreover, fz4-2 expression appears in restricted domains at the segmental borders and the forming limbs compared to fz4-1, which is more broadly expressed within the segments, limbs and the developing nervous system. These results may suggest that these two paralogs have been subject to subfunctionalization (Force et al., 1999; Lynch and Force, 2000).

It is clear from our comparative analysis and from previous work in Drosophila and Tribolium (Janson et al., 2001; Beermann et al., 2011) that genes of the FzIV family likely play a diverse range of roles during the development of panarthropods. However, comparable patterns emerge among the arthropods, which suggest that these genes were involved in nervous system development, and possibly segment formation, as well as perhaps limb development more widely among panarthropods.

In Tribolium, Tc-fz4 appears to function together with Tc-fz1 to regulate leg development most likely by binding Wg, although other Wnt ligands could be involved (Bolognesi et al., 2008; Grossmann et al., 2009; Beermann et al., 2011). Similarly, Wg is a likely candidate for binding the Fz4 receptors in the developing appendages of Parasteatoda, although Wnt5, Wnt8, Wnt11-2, and Wnt16 are also expressed during limb bud formation and/or appendage elongation in this spider (Janssen et al., 2010).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.