The Wnt and Delta-Notch signalling pathways interact to direct pair-rule gene expression via caudal during segment addition in the spider Parasteatoda tepidariorum

The regulation of arthropod segmentation is best understood in Drosophila melanogaster, which employs a well-characterised cascade of transcription factors to generate its segments almost simultaneously along the antero-posterior axis (reviewed in Pankratz and Jäckle, 1993; Peel et al., 2005). In contrast to this long-germ mode of segmentation, most insects and other arthropods exhibit a short-germ mode of segmentation, during which only a few anterior segments are generated simultaneously and subsequently a species-specific number of posterior segments are added sequentially from a posterior growth zone or segment addition zone (SAZ) (Davis and Patel, 2002; Tautz, 2004; Peel et al., 2005; McGregor et al., 2009). Note, however, that these modes of segmentation may not always be discrete because a mixed mode of segmentation has recently been proposed for Nasonia vitripennis (Rosenberg et al., 2014).

Comparative studies have shown that aspects of the Drosophila segmentation cascade are also found in short-germ arthropods, indicating that these were probably features of the regulation of segmentation in the arthropod common ancestor (Peel et al., 2005). Firstly, there is evidence that hunchback (hb) and Distal-less (Dll) perform gap gene-like functions during formation of the prosomal segments of spiders (Schwager et al., 2009; Pechmann et al., 2011). Secondly, the orthologues of Drosophila pair-rule genes are also expressed in the SAZ and segments of short-germ arthropod embryos, which is consistent with roles in segmentation in these animals, although it is likely that these genes were expressed in single rather than double segmental periodicity ancestrally in arthropods (Frasch et al., 1987; Sommer and Tautz, 1993; Patel et al., 1994; Damen et al., 2000, 2005; Davis et al., 2001; Dearden et al., 2002; Chipman et al., 2004b; Schoppmeier and Damen, 2005b; Choe et al., 2006; Damen, 2007; Mito et al., 2007; Chipman and Akam, 2008; Janssen et al., 2011; Sarrazin et al., 2012; Brena and Akam, 2013; Green and Akam, 2013). Finally, the expression and function of segment polarity genes are highly similar across arthropods (Damen, 2002; Hughes and Kaufman, 2002).

In contrast to the regulation of segment formation in Drosophila and possibly other holometabolous insects, the formation of the SAZ and generation of posterior segments in many short-germ arthropods is regulated by a probably ancestral GRN that includes the Delta-Notch signalling pathway (Stollewerk et al., 2003; Schoppmeier and Damen, 2005a; Oda et al., 2007; Chipman and Akam, 2008; Pueyo et al., 2008), together with Wnt signalling (Bolognesi et al., 2008; McGregor et al., 2008b) and caudal (cad) (McGregor et al., 2009; Chesebro et al., 2013). Further understanding the underlying interactions in this GRN and how it directs segmentation in short-germ arthropods can provide much needed new insights into the evolution of these processes.

The spider Parasteatoda tepidariorum (formerly Achaearanea tepidariorum) has proved to be an excellent chelicerate model for studying segmentation in short-germ arthropods (Schwager et al., 2015). This success has been facilitated by detailed descriptions of its early embryogenesis (Akiyama-Oda and Oda, 2003; Mittmann and Wolff, 2012), the establishment of tools to study gene expression and gene function (Akiyama-Oda and Oda, 2006; McGregor et al., 2008a; Kanayama et al., 2010; Kanayama et al., 2011; Hilbrant et al., 2012), as well as the availability of embryonic transcriptomic resources (Posnien et al., 2014).

In Parasteatoda prosomal and opisthosomal segmentation appear to be regulated by different mechanisms. The formation of prosomal segments requires travelling waves of hedgehog and orthodenticle expression and more posteriorly hb and Dll (Pechmann et al., 2009, 2011; Schwager et al., 2009; Kanayama et al., 2011); whereas the Wnt8 (Pt-Wnt8) and Delta-Notch signalling pathways are required for formation of the SAZ and development of opisthosomal segments (Oda et al., 2007; McGregor et al., 2008b). It was shown previously that knockdown of Pt-Wnt8 or Pt-Delta (Pt-Dl) using parental RNAi results in strongly reduced expression of Pt-cad expression and gives rise to truncated embryos with malformed or even an absence of posterior segments (Oda et al., 2007; McGregor et al., 2008b). This suggested that these signalling pathways might act via Pt-cad during segment addition (Oda et al., 2007; McGregor et al., 2008b). These effects may also be explained, at least in part, by the fact that Pt-Wnt8 is required for the dynamic expression of Pt-Dl, which is associated with the formation of new posterior segments (McGregor et al., 2008b).

However, in Parasteatoda and other short-germ arthropods that have been shown to employ Wnt/Delta-Notch/Cad, it is not understood how these key factors interact with each other. Furthermore, it is not known how the expression of putatively downstream segmentation genes, such as even-skipped (eve), is regulated compared with other arthropods. In Drosophila, eve is regulated by a combination of maternal and gap factors, whereas in Tribolium, eve expression is regulated by cad and other pair-rule genes that are likely to operate in a circuit (Small et al., 1991, 1992; Fujioka et al., 1996; Choe et al., 2006; El-Sherif et al., 2014), probably downstream of Wnt signalling (Oberhofer et al., 2014). Therefore, to better understand the regulation of segment formation in short-germ arthropods, we further investigated the regulatory interactions between Pt-Dl, Pt-Notch (Pt-N), Pt-Wnt8 and Pt-cad and studied the expression and regulation of the Parasteatoda orthologues of the Drosophila pair-rule genes eve (Pt-eve) and runt (Pt-run-1) during early embryogenesis in this spider.

We found that Delta-Notch signalling is required for activation of Pt-Wnt8 expression in posterior SAZ cells. However, knockdown of Pt-Dl or Pt-N results in increased expression of Pt-Wnt8 in anterior SAZ cells. Therefore, Delta-Notch signalling is required to suppress expression of Pt-Wnt8 in anterior SAZ cells, presumably to facilitate segment formation from the undifferentiated pool of cells maintained by this Wnt gene. We also found that knockdown of Pt-Wnt8 or Pt-Dl results in the loss of Pt-eve and Pt-run-1 expression. This can be explained by the loss of Pt-cad expression when these pathways are perturbed because we show that knockdown of Pt-cad expression alone in the SAZ inhibits Pt-eve and Pt-run-1 expression, although Pt-cad does not appear to be sufficient to activate these genes. In addition, we observed that Pt-eve does not appear to regulate Pt-run-1 expression or vice versa. This finding suggests that while the pair-rule gene orthologues in Parasteatoda may still form a regulatory circuit, it cannot be based on exactly the same regulatory interactions that have been identified in Tribolium (Choe et al., 2006). Therefore, segment addition in Parasteatoda appears to be directed by dynamic interactions between Wnt8 and Delta-Notch signalling in the SAZ that results in the activation of Pt-cad, which is necessary for the regulation of expression of the pair-rule gene orthologues Pt-eve and Pt-run-1.

In Parasteatoda, formation of the SAZ and production of segments from this tissue require both Wnt8 and Delta-Notch signalling (Oda et al., 2007; McGregor et al., 2008b). We previously showed that Pt-Dl expression is established normally in Pt-Wnt8 knockdown embryos, but subsequently fails to successively clear from the posterior (McGregor et al., 2008b). This suggests that Pt-Wnt8 is necessary for dynamic Pt-Dl expression during posterior development in Parasteatoda.

During stage 6, after Pt-Dl expression has cleared from posterior SAZ cells, this gene is expressed in a salt and pepper pattern juxtaposed to a more diffuse domain in anterior SAZ cells (Fig. 1A and Fig. S1A). We noticed that Pt-Wnt8 expression is weaker in anterior SAZ cells, where it overlaps with the diffuse Pt-Dl expression domain, compared with the stronger expression of Pt-Wnt8 detected in posterior SAZ cells (Fig. 1B) (McGregor et al., 2008b). We therefore investigated whether Pt-Dl is involved in the regulation of Pt-Wnt8. We found that knockdown of Pt-Dl using parental RNAi (pRNAi) results in the loss of Pt-Wnt8 expression in the posterior of the SAZ, but conversely gives rise to stronger Pt-Wnt8 expression in the anterior SAZ cells during stage 6 (Fig. 1C).

Pt-N is expressed in a similar pattern to Pt-Dl in the SAZ at stage 6, but Pt-N expression is then maintained in a more diffuse pattern during stage 7, with slightly stronger expression observed in the newly forming segment (Fig. S2). Note that Pt-Dl and Pt-N may initially be expressed in different cells in the SAZ since knockdown of one leads to more diffuse expression of the other at stage 6, suggesting they inhibit each other's expression at this stage (Fig. S2; Oda et al., 2007).

We then tested whether Pt-Wnt8 expression also requires Pt-N and found that knockdown of Pt-N using parental RNAi had a very similar effect to knockdown of Pt-Dl on the expression of Pt-Wnt8 (Fig. S3). This suggests that Delta-Notch signalling is required to first activate Pt-Wnt8 expression in posterior SAZ cells during stage 5, but subsequently downregulates Pt-Wnt8 in anterior SAZ cells, possibly to facilitate the formation of segments from this tissue. It is also possible that Pt-N, but not Pt-Dl helps to maintain Pt-Wnt8 expression in the SAZ, because the expression of Pt-N persists in the SAZ whereas Pt-Dl expression is cyclical (Fig. S2).

It was previously shown that Pt-Dl and Pt-Wnt8 are required for the establishment of Pt-cad expression in the SAZ (Oda et al., 2007; McGregor et al., 2008b). Like Pt-Dl, Pt-cad also exhibits dynamic expression in the SAZ, forming stripes of expression in each nascent segment (Fig. S4) (Oda et al., 2007). This expression of Pt-cad is slightly out of phase with Pt-Dl expression, but expression of these two genes overlaps in some cells (Fig. S1A).

We then asked if Pt-cad regulates Pt-Dl in the SAZ. Since Pt-cad pRNAi has no discernible effect on morphology or gene expression despite repeated attempts (data not shown), we instead performed embryonic RNAi (eRNAi), where dsRNA is injected into single blastomeres at the 8- to 16-cell stage, leading to RNAi effects in clones of the injected blastomere (Kanayama et al., 2010, 2011). We found that eRNAi using Pt-cad results in strongly reduced levels of Pt-cad transcripts in eRNAi clones of SAZ cells compared with Pt-cad expression in SAZ cells neighbouring the clone (n=5) (Fig. S5A). However, the level of Pt-Dl expression was unaffected in Pt-cad knockdown clones in the SAZ that overlapped with cells that express both Pt-Dl and Pt-cad in wild-type embryos (n=14) (Fig. S1B). This suggests that Pt-cad is not involved in the regulation of Pt-Dl during posterior development in Parasteatoda.

To better understand segment addition in Parasteatoda, we next characterised expression of Pt-eve during early embryogenesis. We observed that Pt-eve is initially expressed in a small oval domain of approximately 20 cells in the SAZ at stage 6 (Fig. S6A). This expression domain then increases in size (Fig. 2A), but concomitantly, the centre clears to form a transient ring of expression (Fig. 2B). This ring of Pt-eve expression is broken by the apparent loss of transcripts in the most posterior cells (Fig. 2C), giving rise to a stripe of expression, approximately 3 to 5 cells wide, in the nascent O1 segment during stage 7 (Fig. 2D). At this stage, expression of Pt-eve is again observed in a circular domain in the most posterior cells of the SAZ (Fig. 2D), which again clears centrally (Fig. 2E) to form a second stripe in the presumptive O2 segment. At the same time, the older stripe of Pt-eve expression in O1 begins to narrow and expression decreases (Fig. 2F).

Subsequently, Pt-eve expression undergoes similar dynamic cycles of strong expression in the posterior SAZ cells followed by the clearance of expression from this region and expression in the forming segments in anterior SAZ cells. As Pt-eve stripes form in nascent segments, the expression in the older, more anterior, segments fades. For example, during formation of O3 (Fig. 2G), Pt-eve expression is observed in O2 and O3 and the SAZ but is no longer detected in O1. De novo Pt-eve expression is seen in the developing central nervous system in older anterior segments (Fig. 2H). In summary, the expression of Pt-eve is consistent with the involvement of this gene in regulating formation of all segments posterior to and including O1, as well as differentiation of the nervous system.

Since Wnt8 and Delta-Notch signalling are required for the formation of the SAZ and the generation of posterior segments, we tested whether these pathways are required for Pt-eve expression. Compared with wild-type embryos (Fig. 3A), we did not detect Pt-eve expression in Pt-Dl pRNAi embryos (Fig. 3B). Similarly, we found that Pt-eve expression was strongly reduced in Pt-Wnt8 pRNAi embryos (Fig. 3C). Therefore both Pt-Wnt8 and Pt-Dl are required for Pt-eve expression in the SAZ. However, since knockdown of Pt-Dl and Pt-Wnt8 also results in the loss of Pt-cad expression (Oda et al., 2007; McGregor et al., 2008b), the effect of knocking down these pathways on Pt-eve expression might be an indirect effect caused by loss of Pt-cad expression.

To investigate if Pt-cad could regulate Pt-eve expression and/or vice versa, we first carried out double in situ hybridisations to compare the expression of these two genes relative to each other during posterior development (Fig. 4A,A′,C,C′ and Fig. S6). Pt-cad and Pt-eve expression are initially detected at early stage 6 when cells appear to first express Pt-cad alone and then express both genes in a small oval-shaped domain (Fig. S6A,A′). Subsequently, Pt-eve and Pt-cad expression expand into an overlapping circular domain in the SAZ, but Pt-cad expression appears to persist in the more posterior cells from which expression of Pt-eve has cleared (Fig. 4A,A′). These two genes continue to be expressed in a similar fashion during the subsequent addition of segments. For example, at stage 7, both Pt-eve and Pt-cad are expressed in overlapping stripes in the nascent O1 segment: Pt-eve is expressed exclusively in the anterior-most row of cells, whereas Pt-cad is also expressed solely in approximately two rows of the most posterior cells (Fig. 4C,C′). At this stage, a new domain of overlapping expression of Pt-eve and Pt-cad can also be observed in posterior SAZ cells (Fig. 4C).

The relative expression patterns of Pt-cad and Pt-eve suggest that there might be a regulatory interaction between these two genes during posterior development. We therefore performed eRNAi and generated 12 independent Pt-cad eRNAi clones in cells that overlapped with the normal expression domains of Pt-eve in stage 6 and 7 embryos. In all embryos, in which the Pt-cad eRNAi clone overlapped with the circular Pt-eve expression domain at stage 6, we observed that Pt-eve expression was completely lost or very strongly reduced (Fig. 4B). Later, at stage 7, when Pt-eve expression is observed in posterior SAZ cells and in more anterior cells in a stripe corresponding to the nascent O1 segment, expression of Pt-eve was reduced in both expression domains that overlapped with Pt-cad eRNAi clones (Fig. 4D). These results suggest that Pt-cad is probably required for the activation and maintenance of Pt-eve expression during posterior development in Parasteatoda; however, it is not clear whether this regulation is direct or indirect.

We then tested if Pt-cad expression is sufficient to activate Pt-eve expression by injecting capped Pt-cad-eGFP mRNA into blastomeres at the 16-cell stage and allowing them to develop until stage 5 (i.e. before Pt-cad and Pt-eve are normally expressed). We were able to detect clones of cells with nuclear GFP expression (Fig. S7), demonstrating that Pt-cad was expressed and able to localise to the nuclei (n=5). However, we did not observe expression of Pt-eve in any of these cells, indicating that although Pt-cad expression is required for Pt-eve expression, it is not sufficient in these conditions (Fig. S7). Indeed, since some of these cells near the pole of the germ disc at this stage are likely to express Wnt8 and Dl, this implies that an additional factor or factors are required to activate Pt-eve (Fig. S7).

Next, we tested whether Pt-eve regulates Pt-cad. As is the case for Pt-cad, pRNAi against Pt-eve had no discernible effects on morphology or gene expression, suggesting that this approach does not work for this gene (data not shown). However, we were able to knock down Pt-eve expression in clones of cells that overlapped with the normal expression of this gene using eRNAi (n=4) (Fig. S5) (Kanayama et al., 2010). Using this approach, we then assessed Pt-cad expression in Pt-eve eRNAi clones, and found that Pt-cad expression in both posterior SAZ cells and in the nascent O1 segment was unaffected in overlapping Pt-eve eRNAi cell clones (n=16) (Fig. S5C,D), suggesting that Pt-eve does not regulate Pt-cad.

In the beetle Tribolium, pair-rule genes have been shown to function in a regulatory circuit (Choe et al., 2006). Central to this model is the production of dynamic stripes of eve expression that rely on a negative feedback loop whereby eve activates runt (run), which activates odd-skipped, which then represses eve (Choe et al., 2006).

To investigate whether a similar circuit based on the regulation of Pt-run-1 by Pt-eve could be involved in segment addition in Parasteatoda, we first assayed the expression of Pt-run-1 compared with Pt-eve. Note that we identified one other runt-like gene in the transcriptome of Parasteatoda, but it is not expressed in a pattern consistent with a role in segmentation (data not shown).

We found that the expression of Pt-run-1 commences during stage 6 (Fig. S8A), at approximately the same time that Pt-eve expression is first detected (Fig. 2A). Moreover, Pt-run-1 and Pt-eve expression partially overlap in posterior and anterior SAZ cells at all stages assayed (Fig. S8). However, Pt-eve is expressed approximately three cell rows anterior to Pt-run-1 in stripes in the anterior of the SAZ (Fig. S8G-I).

We then tested whether Pt-eve is required for the expression of Pt-run-1 using eRNAi. However, Pt-run-1 expression appeared to be normal where it overlapped with Pt-eve RNAi cell clones in the SAZ compared with neighbouring cells (n=12) (Fig. 5). This suggests that, in contrast to the pair-rule circuit in Tribolium, in Parasteatoda, Pt-eve does not regulate Pt-run-1 during segment addition. We also investigated if Pt-run-1 is required for activation of Pt-eve; however, knockdown of Pt-run-1 using eRNAi had no effect on Pt-eve expression in the SAZ (n=10) (Fig. S9). Indeed, it appears that Pt-run-1 is actually regulated in parallel to Pt-eve because RNAi knockdown of Pt-Wnt8 and Pt-Dl by pRNAi, and Pt-cad by eRNAi all greatly reduced Pt-run-1 expression (Fig. S10), which is similar to the effect on Pt-eve expression when these genes are knocked down (Figs 3 and 4).

We have found that Pt-eve and Pt-run-1 expression in the SAZ requires Pt-cad expression, which explains the loss of their expression in Pt-Wnt8 or Pt-Dl knockdown embryos, because these signalling pathways are also required for Pt-cad expression (Oda et al., 2007; McGregor et al., 2008b). However, we also found that Pt-cad expression alone is probably not sufficient to activate Pt-eve expression. This supports the interpretation that Pt-cad expression is regulated by Wnt8 and Delta-Notch signalling and that these factors together activate the expression of Pt-eve and Pt-run-1 (Fig. 6), although it remains possible that other currently unknown transcription factors that may or may not be regulated by Wnt8 and Delta-Notch signalling are also required.

Furthermore, since it does not appear that Pt-eve feeds back to regulate Pt-cad or that Pt-Dl expression requires Pt-cad, it is likely that the cyclical expression of segmentation genes, such as Pt-eve and Pt-run-1, and the production of segments from the spider SAZ, is driven by dynamic interplay between the Wnt8 and Delta-Notch signalling pathways (Fig. 6). This model is supported by our finding that Delta-Notch signalling is required to activate Pt-Wnt8 in posterior SAZ cells and that Pt-Wnt8 then facilitates dynamic expression of Pt-Dl in the SAZ (Fig. 6). Our results suggest that Delta-Notch signalling then suppresses Pt-Wnt8 expression in anterior SAZ cells possibly to allow the formation of segments from a pool of undifferentiated cells maintained by this Wnt ligand (McGregor et al., 2008b). However, it is still unclear how dynamic expression of Pt-Dl and Pt-N is generated and how Delta-Notch signalling activates Pt-Wnt8 in posterior SAZ cells, but suppresses the expression of this Wnt ligand gene in anterior SAZ cells. Based on the effects of the reciprocal knockdown of Pt-N and Pt-Dl on the expression of the other in the Parasteatoda SAZ (this study and Oda et al., 2007), we hypothesise that this could involve auto-inhibitory regulation of genes in the Delta-Notch pathway (Kageyama et al., 2007) as well as other, still undiscovered, genes expressed in the SAZ of Parasteatoda.

There is functional evidence that similar genetic interactions to those that we have identified in Parasteatoda probably also regulate segmentation in other arthropods. In embryos of the cockroach Periplaneta, there is feedback between Wnt1 and Dl in the SAZ, and Wnt1 also activates cad to generate a signalling centre responsible for the generation of posterior segments (Chesebro et al., 2013). Unlike in Parasteatoda, however, cad represses Dl in the Periplaneta SAZ (Chesebro et al., 2013). Furthermore, it is likely that Wnt1 is required for cad expression during posterior development in Tribolium as in Gryllus (Shinmyo et al., 2005; McGregor, 2006; Oberhofer et al., 2014) and a recent study has shown that the graded expression of Tc-cad is required for the dynamic expression of Tc-eve (El-Sherif et al., 2014). This suggests that although there are differences in the regulation of segment addition among short-germ arthropods, the regulation of eve by cad, probably directed by upstream signalling pathways, may have been used ancestrally in arthropods.

In Parasteatoda, Pt-eve and Pt-run-1 are expressed in the SAZ and subsequently in stripes associated with the formation of all of the segments that are generated from this tissue i.e. O1 and all of the following posterior segments. This result is consistent with previous analysis of these and other pair-rule orthologues in the Central American wandering spider Cupiennius salei (Damen et al., 2000, 2005). Our data provide further evidence for differences in the regulation of prosomal and opisthosomal segments in spiders, whereby gap and pair-rule gene orthologues respectively direct the formation of segments in these tagmata (Damen et al., 2000, 2005; Pechmann et al., 2009, 2011; Schwager et al., 2009). Interestingly, this also indicates that the roles of eve and run-1 in spiders is restricted to formation of more posterior segments than, for example, in the insects D. melanogaster and Tribolium, and the myriapods Strigamia maritima and Glomeris marginata, in which eve is expressed in a segmental pattern in four segments more anterior to O1/T2 (Frasch et al., 1987; Brown et al., 1997; Janssen et al., 2011; Brena and Akam, 2013). Moreover, while there is a hierarchy of primary and secondary pair-rule gene orthologues in arthropods (Damen, 2007), our study further exemplifies that the register of the expression of these genes has diverged among these animals: expression of eve and run overlaps in forming segments in Glomeris as in Parasteatoda, but they are out of phase in Strigamia and Drosophila (Green and Akam, 2013).

The requirement of eve for run expression in Tribolium is a key regulatory step in the pair-rule circuit that underlies segmentation in this beetle (Choe et al., 2006). However, in Parasteatoda, we did not find a requirement of Pt-eve for the expression of Pt-run-1 (or vice versa), which suggests that if this spider, like Tribolium, also employs a pair-rule circuit, it is composed of different genetic interactions. It also remains possible that a pair-rule circuit may not be a feature of spider segmentation, possibly because these arthropods employ Delta-Notch in combination with Wnt signalling instead. Thus a pair-rule circuit might be a derived feature of segmentation in arthropods that no longer depends on Wnt with Delta-Notch signalling. Indeed, there is no evidence that Delta-Notch signalling regulates segmentation in Tribolium although some possible Delta-Notch pathway genes, including hairy, are expressed in a pattern consistent with a role in segmentation (Tautz, 2004; Aranda et al., 2008).

Although it is difficult to infer the role of genes and pathways and architecture of gene regulatory networks that may have existed in a common ancestor from the expression and function of a few genes and pathways in divergent extant organisms, an understanding is emerging of how segment addition may have been regulated in the arthropod common ancestor. Studies of segmentation in a range of arthropods, including spiders, the myriapod Strigamia and Periplaneta suggest that a Wnt/Delta-Notch/Cad-based system acting upstream or possibly in parallel to pair-rule gene orthologues was probably the ancestral mechanism in arthropods (Chipman et al., 2004a; Chipman and Akam, 2008; McGregor et al., 2008b, 2009; Pueyo et al., 2008; Brena and Akam, 2013; Chesebro et al., 2013). Furthermore, the work on spiders and Strigamia indicates that in this ancestor, segments were probably added with single segment periodicity (e.g. this work; Brena and Akam, 2013).

We have found that, during segment addition in Parasteatoda, the dynamic expression of Pt-eve and Pt-run-1 requires Pt-cad, whose expression is a read-out of a dynamic interplay between the Wnt and Delta-Notch signalling pathways. This provides new insights into the transition between the formation of the SAZ and the production of segments from this tissue in the spider, and the evolution of this key developmental process among arthropods. Future work in this spider will allow us to determine whether Pt-eve and Pt-run-1 are direct targets of Pt-cad, precisely how Delta-Notch and Wnt signalling interact at the cellular and molecular level, and investigate the possible involvement of other genes during segment addition.

Embryos were collected from adult female Parasteatoda tepidariorum from our laboratory culture in Oxford that was founded with spiders from Göttingen (Germany). The spider culture was kept at 25°C and embryos of stages 5 to 9 fixed as described in Akiyama-Oda and Oda (2003). Embryos were staged according to Mittmann and Wolff (2012). Note that these stages were chosen for this study because the SAZ develops from the caudal lobe, which is formed during stages 5 and 6. The first segment (O1) is added from the SAZ during stage 7, and subsequent segments are added from stage 8 onwards.

Total RNA was extracted from a mixture of embryonic stages 5 to 9 using the RNeasy Lipid Tissue Mini Kit (Qiagen). cDNA was synthesised from total RNA with the QuantiTect Reverse Transcription Kit (Qiagen). Pt-cad (AB096075), Pt-eve (locus_7056), Pt-Dl (AB287420), Pt-N (AB287421), Pt-Wnt8 (FJ013049), Pt-run-1 (locus_15496) and Pt-run-2 (locus_12769) sequences were obtained from GenBank or the Parasteatoda transcriptome (Posnien et al., 2014). Gene-specific cDNA fragments were amplified with primers designed with Primer3 (http://primer3.ut.ee) and cloned into pCR4-TOPO vector (Invitrogen, Life Technologies): Pt-eve (731 bp), GCAGGGTCTTCGAACTTCAG and GTTGGAAGAGTTGCGTCGTT; Pt-cad (1005 bp), TGTTGATGGGAGATGGTTCC and AAAGCCCCTTTCGAAGATGT; Pt-cad F1 (456 bp), ATGTATTCCCCTACAGCTAGAC and ATCGCTGGAAACTGCAACAATAG; Pt-cad F2 (429 bp), GGTATGAGTGGTACTGAATCACC and TCAGTAGATACTAATATTTGCTATATTTAGAG; Pt-run-1 F1 (741 bp), ATGCATTTACCAGCAGATTCAGTGA and AACAGCGAGAGTGACATCCAAATTATA; Pt-run-1 F2 (792 bp), TCTCCAACATCTCAAGATTCATGTTC and TCAGTATGGCCTCCATAGACCT; Pt-Dl (967 bp), ACAAACCACACGGCTTTTTC and GCTTGGTCAAGCAGTCATCA; Pt-N F1 (701 bp), TGCAGCACATTCGAGACATG and CCGAGCCATTGTCTTCATCG; Pt-N F2 (675 bp), GTTCTCCTGGGCTAATGGGT and TCTTCTGGTGATGAGCTGCA; Pt-Wnt8, see McGregor et al. (2008b).

RNA probes were labelled with digoxigenin (Roche) and detected with an alkaline phosphatase-conjugated anti-DIG antibody (Fab fragments, Roche) using the substrate Nitro Blue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP) (Roche), resulting in purple/blue staining. For double in situ hybridisation, an additional probe was labelled with fluorescein (Roche) and detected with an alkaline phosphatase-conjugated anti-fluorescein antibody (Fab fragments, Roche) and with INT (2-[4-iodophenyl]-3-[4-nitrophenyl]-5-phenyltetrazolium chloride)/BCIP (Roche), resulting in orange staining. In situ hybridisations were carried out according to the whole-mount protocol for spiders (Prpic et al., 2008), with minor modifications. The anti-DIG and anti-fluorescein antibodies were pre-absorbed overnight at 4°C with embryos from stages 6 to 8.2. Note that in situ hybridisation staining reactions on control and experimental (RNAi) embryos were carried out for the same lengths of time. For double in situ hybridisations, the first staining reaction was stopped by incubating the samples at 65°C with inactivation buffer (50 ml hybridisation buffer B, 0.1 ml 10% Tween-20, 1.5 ml 10% SDS). The embryos were then washed twice with PBS-T for 15 min and twice for 20 min. Subsequently, the embryos were incubated in blocking solution for 30 min as for the regular in situ hybridisation staining and then with the anti-fluorescein antibody at a dilution of 1:2000 in blocking solution for 3 h. Nuclear staining was performed by incubation of embryos in 1 μg/ml 4-6-diamidino-2-phenylindol (DAPI) in PBS with 0.1% Tween-20 for 30 min. Segmental identity in stained embryos was assigned from morphological markers ascertained from images of DAPI staining.

Fragments of the coding regions of Pt-Dl (967 bp), Pt-Wnt8 (718 bp), Pt-cad (1005 bp completely and as two non-overlapping fragments), Pt-eve (713 bp), run-1 (672 bp as two non-overlapping fragments), Pt-N (1190 bp as two non-overlapping fragments) and GFP were amplified from plasmids using universal primers, which both contained a 5′ T7 promoter binding site (Fwd T7, 5′-TAATACGACTCACTATAGGG-3′; Rev T7/T3, 5′-TAATACGACTCACTATAGGGAATTAACCCTCACTAAAGGGA-3′). The introduction of the T7 promoter sequence on the antisense strand, using the Rev T7/T3 primer, allows the in vitro transcription of both strands in one reaction with the MegaScript T7 transcription kit (Invitrogen). Double-stranded (ds) RNA was then generated by annealing the transcripts in a water bath starting at 95°C and slowly cooled down to room temperature. The dsRNA was then adjusted to a concentration of 1.5-2.0 μg/μl for injections.

For each gene, at least three adult female spiders were injected according to the protocol by Akiyama-Oda and Oda (2006). dsRNA was injected into the opisthosoma of spiders at concentrations of 1.5-2.0 μg/μl every 2-3 days up to a total of five injections. The injected spiders were mated after the second injection. Embryos from injected spiders were fixed for gene expression and phenotypic analyses 2 and 4 days after egg laying. Embryos from GFP-injected control females were generated and treated as described above.

Embryonic injections were carried out as described in Kanayama et al. (2010) with minor changes (GC100F-10 capillaries, Harvard Apparatus; needle puller PC-10, Narishige). Embryos were injected at the 8- or 16-cell stage with an injection mix composed of 10 μl fluorescein isothiocyanate (FITC)-dextran (2 μg/μl, MW 10,000, Sigma), 10 µl biotin-dextran (2 μg/μl, MW 10,000, Sigma) and 5 μl dsRNA (1.5 to 2.0 μg/μl) and fixed when they reached developmental stages 6 and 7. In order to visualise the clones of eRNAi cells, the co-injected biotin-dextran was detected with the Vectastain ABC-AP kit, which was carried out according to the manufacturer's protocol (Vector Laboratories) following in situ hybridisation. At least 200 embryos were injected for each gene of interest to ensure that multiple independent clones were generated in the SAZ.

An 885 bp fragment of the Pt-cad CDS was isolated and cloned using the pENTR Directional TOPO Cloning Kit (Life Technologies). The fragment was shuttled into pAWG, upstream of the enhanced GFP (eGFP), with the Gateway LR Clonase II Enzyme Mix (Invitrogen). The cad-eGFP construct was amplified from pAWG-cad-eGFP with 5′ extensions containing PstI and BamHI for forward and reverse primers, respectively. This construct was ligated into the pSP64 Poly(A) vector (Promega) after double digestion with PstI and BamHI. The pSP64-cad-eGFP-PolyA was then linearised with NheI and the resulting template was used for the SP6 transcription reaction with mMESSAGE mMACHINE SP6 Transcription Kit (Ambion) following the manufacturer's instructions. Capped eGFP-NLS mRNA was prepared from plasmid pSP64-NLS-tdEosFP-polyA-NotI+ (a gift from Hiroki Oda and Yasuko Akiyama-Oda) as described previously (Kanayama et al., 2010). Capped mRNAs were injected as described by Kanayama et al. (2010).

Embryos were imaged using a Leica fluorescence stereomicroscope equipped with a Jenoptik ProgRes C3 digital camera. Bright-field and UV channel images were merged using Adobe Photoshop CS6, which was also used for linear corrections of brightness, contrast and colour values. Images for the Pt-cad overexpression experiment (Fig. S7) were taken with a Zeiss Axio Zoom V16 stereomicroscope, equipped with an Axiocam 506 mono and a colour digital camera.

We thank Hiroki Oda and Yasuko Akiyama-Oda for providing training on spider eRNAi experiments and providing us with plasmid pSP64-NLS-tdEosFP-polyA-NotI+. We thank Sebastian Kittelmann for comments on the manuscript and Christina Jahn for discussions about pair-rule gene expression. We also thank the three reviewers for their very constructive comments and suggestions.