Ancestral functions of Delta/Notch signaling in the formation of body and leg segments in the cricket Gryllus bimaculatus

Segmental organization of the anterior-posterior (AP) body axis is observed in diverse animal phyla, including arthropods, annelids and chordates. However, there has been debate regarding whether segmentation among phyla has a common origin and shares common genetic programs (Damen, 2007; Peel et al., 2005; Tautz, 2004). Segmentation mechanisms in Drosophila occur via hierarchical interactions between transcription factor gradients that subdivide embryos into progressively smaller units, eventually resulting in a repetitive pattern of segments along the entire AP axis of the embryo (Ingham, 1988; Nüsslein-Volhard and Wieschaus, 1980; Pankratz and Jackle, 1993; St. Johnston and Nüsslein-Volhard, 1992). An important feature of this segmentation mode, called long-germ segmentation, is that the syncytial state of the early Drosophila embryo allows diffusion of transcription factors to form gradients throughout the egg, enabling simultaneous specification of all segments.

Although the explication of Drosophila segmentation has provided a paradigm for investigating the evolution of the segmentation process, recent studies in other arthropods have demonstrated that long-germ segmentation is evolutionarily derived and not representative of arthropods. Indeed, most arthropods (chelicerates, myriapods, crustaceans and insects) show short- or intermediate-germ segmentation (Davis and Patel, 2002; Liu and Kaufman, 2005; Tautz et al., 1994). Because posterior segments are formed sequentially in an anterior-to-posterior progression from a posterior cellularized region, referred to as a `growth zone' in these arthropods, the free diffusion of transcriptional factors that occurs in Drosophila is not possible in the posterior growth zone. Therefore, it has been speculated that there might have been a drastic change in the molecular mechanisms underlying posterior patterning and segmentation during arthropod evolution (Sander, 1976; Patel, 1994; Davis and Patel, 1999; Peel and Akam, 2003; Tautz, 2004; Peel et al., 2005).

The segmentation of short- and intermediate-germ arthropods is reminiscent of somitogenesis in vertebrates, suggesting involvement of a cell-cell signaling system. In vertebrates, somites are formed sequentially in an anterior-to-posterior progression from the presomitic mesoderm (PSM) (Pourquie, 2001). This process involves a molecular oscillator, called the somitogenesis clock, which drives the periodic expression of genes in the PSM. The formation of each somite is accomplished by a wave of oscillatory gene expression. Several genes of the Notch pathway, including Delta, lunatic fringe, and hairy, exhibit oscillatory expression in the PSM. Therefore, Notch signaling is a crucial component of the somitogenesis clock.

Delta/Notch signaling has been shown to be involved in the segmentation of one short-germ arthropod, the wandering spider Cupiennius salei. In C. salei, Delta is expressed in the posterior growth zone as well as in segmental stripes anterior to the growth zone (Stollewerk et al., 2003). In addition, RNAi silencing of components of the Delta/Notch signaling pathway (i.e. Notch, Delta, Presenilin and Suppressor of Hairless) produces severe segmentation defects (Schoppmeier and Damen, 2005; Stollewerk et al., 2003). Furthermore, in a short-germ insect, the cockroach Periplanetaamericana, Delta is expressed in segmental stripes anterior to the growth zone and knockdown of the Notch gene causes posterior segmentation defects (Pueyo et al., 2008). These observations suggest that the involvement of Delta/Notch signaling in body segmentation might be an ancestral feature of arthropods and that Notch-mediated segmentation might be inherited from a common ancestor of arthropods and vertebrates (Couso, 2009). The apparent lack of involvement of Delta/Notch signaling in body segmentation in Drosophila and Tribolium (Richards et al., 2008; Campos-Ortega and Knust, 1990) suggests that this mechanism might have been lost in the common ancestor of beetles and flies, and possibly of all higher or holometabolous insects. However, such a scenario is controversial because data from paleontology and comparative morphology research seem to suggest that a segmented common ancestor for all bilaterians is unlikely (Chipman, 2010). In addition, the role of Notch signaling in arthropod segmentation process remains ambiguous. Therefore, further exploration of the role of Notch signaling in arthropod segmentation is needed.

In this study, we examined the expression patterns and functions of the Delta and Notch genes in the intermediate-germ cricket Gryllus bimaculatus, focusing on formation of body and leg segments. We found that Delta was expressed transiently in the posterior growth zone and in segmental stripes before the emergence of wingless segmental stripes, whereas Notch was uniformly expressed in early G. bimaculatus embryos. Our functional analyses of Gryllus Delta and Notch genes indicate that Notch signaling is required for proper morphogenesis of segments and formation of posterior segments through regulation of germband elongation, but is not required for the specification of segment boundaries. By contrast, our data demonstrate that Notch signaling is essential for the specification of segment boundaries and normal growth during leg formation.

G. bimaculatus (two-spotted cricket) nymphs and adults were reared at 28-30°C at 70% humidity under 10 hour/14 hour light/dark photoperiod as previously described (Niwa et al., 1997). Fertilized eggs were collected with wet paper towels and incubated at 28°C in a plastic dish. Embryos were staged according to the scheme of Zhang et al. (Zhang et al., 2005).

Total RNA was extracted from cricket embryos at various stages using Isogen (Nippon-Gene). cDNA was synthesized using the Superscript First Strand Synthesis Kit (Invitrogen) with random hexamers. To clone Gb′Delta, we first performed a degenerate polymerase chain reaction (PCR) using primers targeting the VCLKHYQA (forward) and NQDLNYCT (reverse) motifs of a conserved domain including the Delta-Serrate-Lag2 (DSL) domain. 5′ and 3′ RACE PCRs were performed using gene-specific primers and anchor primers supplied in a SMART RACE cDNA Amplification Kit (Clontech). The Gb′Delta cDNA sequences were deposited at the DNA Data Bank of Japan (DDBJ) (accession number: AB635584). To isolate two Gb′Notch PCR fragments, we used two pairs of degenerate primers targeting the CEGDINE (forward)/CDYGCNN (reverse) and AAVAA (forward)/WAAAVNN (reverse) motifs. We then isolated a 2280-bp fragment using specific primers directed against these PCR fragments. The Gb′Notch cDNA sequences were deposited at the DDBJ (accession number: AB635585).

Embryo fixation and whole-mount in situ hybridization with digoxigenin (DIG)-labeled antisense RNA probes were performed as previously described (Niwa et al., 2000; Zhang et al., 2005). Double labeled in situ hybridization was performed as previously described (Mito et al., 2007).

Double-stranded RNA (dsRNA) directed against Gb′Delta, Gb′Notch and DsRed2 was prepared as described previously (Miyawaki et al., 2004). In all RNAi experiments, DsRed2 dsRNA [660 bp, derived from pDsRed2-N1 (Clontech)] was used as a negative control. To exclude potential off-target effects in RNAi experiments, we used two different dsRNAs corresponding to two non-overlapping regions in each of the Gb′Delta and Gb′Notch transcripts and confirmed that both of the dsRNAs derived from the two different regions resulted in the same phenotype. The dsRNAs were designated Gb′Delta1 (508 bp; nucleotide positions 1225-1732), Gb′Delta2 (321 bp; nucleotide positions 901-1221), Gb′Notch1 (677 bp; nucleotide positions 61-737) and Gb′Notch2 (461 bp; nucleotide positions 1678-2138) (Fig. 1A,B). The final concentration of dsRNA was adjusted to 20 μM and dsRNAs were injected as described previously (Mito et al., 2005). Eggs were collected 5 to 10 days after injection.

The ventral nerve cord was visualized by incubating embryos with a goat anti-HRP antibody (Jackson ImmunoResearch) that reacts specifically with the N-linked carbohydrate epitope found on insect neurons (Seppo et al., 2003). After fixation, cricket embryos were washed in KTBT [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM KCl, 0.1% Triton X-100]. The embryos were then blocked with 1.5% blocking reagent (Roche) in KTBT and incubated with FITC-conjugated anti-HRP antibody diluted 1:5 in 1.5% blocking reagent in KTBT for 3 hours at 4°C. Blocking was followed by three washes with KTBT.

The cell proliferation assay was carried out using the Click-iT EdU Alexa Fluor 488 Imaging Kit (Invitrogen) (Salic and Mitchison, 2008). 5-Ethynyl-2′-deoxyuridine (EdU) was injected into the posterior of the cricket egg at the appropriate stages. Embryos were incubated for 4 hours at 28°C after EdU injection and then fixed for 15 minutes. EdU-incorporating cells were detected according the manufacturer's instructions. To counterstain nuclei, embryos were held in a 20 ng/ml solution of Hoechst 33342 in PBT (PBS with 0.1% Tween-20) for 15 minutes at room temperature after EdU incorporation. Embryos were then washed in PBT and mounted on slides for visualization.

Using degenerate primers directed against the conserved regions of Drosophila Delta – a conserved DSL domain and nine epidermal growth factor (EGF) repeats (Fig. 1A) – a short Gb′Delta clone was isolated and used to design specific primers for 5′ and 3′ RACE and isolate fragments of the gene. The Gb′Delta gene is predicted to encode 455 amino acids with a conserved DSL domain and five EGF repeats (Fig. 1A).

To clone Gb′Notch, two pairs of degenerate primers were designed against the extracellular and intracellular domains, and two PCR fragments from both domains were obtained. Using specific primers directed against these PCR fragments, we isolated a 2280-bp fragment encoding 760 amino acids and containing several conserved domains, EGF repeats, Notch/Lin-12 repeats, a transmembrane domain, a RAM 23 domain and ankyrin repeats (Fig. 1B).

In Gryllus, the anterior segments, the gnathal to mesothoracic (T2) segments, are specified almost simultaneously in the early embryo, whereas the remaining posterior segments, the metathoracic (T3) and abdominal segments, are formed sequentially through elongation of the posterior growth zone of the germband (Miyawaki et al., 2004). We examined whether Gb′Delta and Gb′Notch are expressed during this segmentation process by whole-mount in situ hybridization (WISH) (Fig. 2).

Intense expression of Gb′Delta was observed in the prospective mandible segment and the posterior region of the embryo 30-34 hours after egg-laying (hAEL) (Fig. 2A,B). As can be observed in sagittal sections (Fig. 2C,C′), the posterior domain was localized to the ectoderm cells in the most posterior region of the embryo, overlapping with the posterior growth zone. Interestingly, Gb′Delta expression receded from the growth zone and remained in the terminal region of the embryo at 36 hAEL (Fig. 2D, arrowhead). Segmental expression of Gb′Delta was observed in the prospective maxilla, labium and prothoracic (T1) segments at 34 hAEL (Fig. 2B), and in the prospective T2 and T3 segments at 36 hAEL (Fig. 2D).

To investigate the location and timing of the Gb′Delta stripes, we used expression of a segment polarity gene Gb′wg as a segmental marker. Gb′wg is expressed in stripes in an anterior-to-posterior sequence during segmentation and seems to be involved in the specification of segment boundaries in Gryllus embryos (Miyawaki et al., 2004). Double staining for Gb′Delta and Gb′wg revealed that the Gb′Delta stripes overlap with Gb′wg stripes (Fig. 2E,F) and precede the corresponding Gb′wg stripe (arrow in Fig. 2G). During subsequent posterior elongation, Gb′Delta stripes in the gnathal and thoracic regions faded out in an AP progression, whereas Gb′Delta stripes appeared sequentially in the prospective abdominal segments (arrows in Fig. 2H,I). Additionally, spots of Gb′Delta expression were observed in the head region (Fig. 2B,D,E,G-I) and trunk (Fig. 2G-I). These spots probably correspond to proneuronal cells, as observed in proneural regions during neural cell fate specification in Drosophila (Baker, 2000) and other arthropods (Dove and Stollewerk, 2003; Kadner and Stollewerk, 2004; Stollewerk, 2002). In contrast to the specific expression pattern of Gb′Delta, Gb′Notch was uniformly expressed during early embryogenesis (compare Fig. 2J,K with 2L,M).

Next, we observed the expression patterns of Gb′Delta and Gb′Notch during late embryogenesis by WISH (Fig. 3). In Drosophila, Delta and Notch are expressed during leg development, where they are required for correct segmentation (Parody and Muskavitch, 1993; Shellenbarger and Mohler, 1975). In stage-8 Gryllus embryo legs, Gb′Delta was expressed in two rings (arrowheads in Fig. 3C), corresponding to the proximal sides of the femur/tibia boundary and the tibia/tarsus boundary (see Fig. 3F), and small spots (Fig. 3C). During stage 9, an additional broad ring appeared in the prospective tarsal region (arrows in Fig. 3D,E). By stage 10, the broad tarsal ring divided into two narrow rings (arrows in Fig. 3F) and additional rings appeared in the prospective trochanter and tarsus (asterisks in Fig. 3F). In addition, faint and broad expression was observed in the prospective femur and tibia (brackets in Fig. 3F).

A number of Gb′Delta rings were also observed in the developing antenna of the late stage embryos (Fig. 3A,B,G-J). In stage-8 antennae, three intense rings of Gb′Delta expression (arrowheads in Fig. 3G) were observed in the middle region and a few weak rings were observed in the proximal and distal regions (Fig. 3G). By stage 10, additional rings were intercalated between the intense rings and also appeared in the proximal and distal regions (Fig. 3H-J).

Gb′Notch was expressed throughout the whole embryo, with upgraded expression in rings in the legs and antennae during late embryogenesis (Fig. 3K-Q). In stage-8 legs, intense expression of Gb′Notch was observed as a broad ring in the prospective femur (bracket in Fig. 3M) and as a smaller domain in a more distal region (arrowhead in Fig. 3M). By stage 10, these intense domains became restricted to the femur/tibia boundary and the tibia/tarsus boundary, consequently forming intense ring-like domains in these segment boundaries (Fig. 3N,O, arrowheads). In addition, a new intense ring appeared in the prospective tarsus (arrow in Fig. 3N) and divided into two narrow rings (Fig. 3O, arrows), similar to the Gb′Delta tarsal ring formation. In antennae, narrow rings were intercalated between three intense rings (arrowheads in Fig. 3P,Q) and were also added in the distal region by stage 10 (Fig. 3P,Q), similar to the rings of Gb′Delta expression observed in the antenna.

It is known that fringe, a negative regulator of Notch signaling, is expressed in the epidermis of the trunk segments after segmentation in the short-germ insect Schistocerca gregaria, suggesting the involvement of Notch signaling in segment morphogenesis, but not in the establishment of segmental boundaries (Dearden and Akam, 2000). Similarly, Gb′Delta was expressed in the trunk epidermis of embryos during late embryogenesis (Fig. 3B, arrows).

To examine the functions of Gb′Delta and Gb′Notch in the formation of body and leg segments, we injected dsRNA directed against these genes into adult females to determine knockdown phenotypes (parental RNAi) (Bucher et al., 2002; Mito et al., 2005). We categorized Gb′Delta and Gb′Notch dsRNA-injected embryos into three classes based on the phenotype severity of embryos after segmentation (Table 1), and observed a higher degree of `undeveloped embryos' in Gb′Delta and Gb′Notch RNAi-treated embryos. The impeded development of RNAi-treated embryos could be due to an effect on oogenesis as these embryos were more numerous and the number of eggs laid decreased by the day in RNAi-treated embryos. In the present study, we focused on effects on embryonic morphology. Compared with late-stage control embryos (Fig. 4A,A′), most Gb′Delta RNAi-treated embryos were classified as class III, showing various morphological abnormalities in the formation or size and/or shape of their segments (52.6%, n=154/293, Fig. 4B,B′). In most cases, the size of segments was enlarged in the thoracic region and reduced in the abdominal region (Fig. 4B,B′). Segmental boundaries were irregularly arranged and not well defined in some class III embryos (Fig. 4D,D′). We also observed a shortening of the distance between the right and left appendages or proximal fusion of the right and left appendages, suggesting that the development of midline structures was defective in Gb′Delta RNAi-treated embryos (Fig. 4B′, arrow).

Phenotype class II embryos, which had more severe morphological defects than those observed in class III embryos, showed a loss of the posterior terminal structures and some abdominal segments (8.5%, n=25/293, Fig. 4C,C′). In class I, the most severe phenotypic category, embryos were missing some thoracic segments and all abdominal segments (2.7%, n=8/293, Fig. 4E). The appendages in Gb′Delta RNAi-treated embryos were frequently curved and somewhat thin, but maintained correctly formed segment junctions, unlike those in the Notch RNAi experiments (see below).

Gb′Notch RNAi-treated embryos showed similar body segment abnormalities (53.8%, n=119/221, Fig. 4F,F′). In these embryos, the segments were irregularly shaped with an abnormal (reduced or enlarged) size. In addition, segmental boundaries were not properly arranged and were less well defined (Fig. 4F,F′) than in control embryos (Fig. 4A,A′), and the cuticle frequently failed to form (Fig. 4F-H,F′-H′). We also observed fusion of the left and right appendages, as observed in Gb′Delta RNAi-treated embryos (Fig. 4F′, arrow). Moreover, more severely affected embryos, class II embryos (10.0%, n=22/221, Fig. 4G,G′) and class I embryos (2.3%, n=5/221, Fig. 4H,H′), were present. In addition to body segment malformations, Gb′Notch RNAi-treated embryos exhibited shortening of all appendages (compare Fig. 4A′ with 4F′) with fusion of segments (see below).

Because Delta/Notch signaling is known to be involved in neurogenesis in other species, such as Drosophila and Cupiennius (Baker, 2000; Stollewerk, 2002), and both Gb′Delta and Gb′Notch RNAi-treated embryos showed defects in the midline region (Fig. 4B,F), we examined the role of Delta/Notch signaling in the formation of the central nervous system (CNS). Anti-HRP staining (Fig. 5) revealed disorganization of the CNS with partial loss or hyperplasia of neurons in RNAi-treated embryos (Fig. 5B′,C′), suggesting that the neurogenic role for Delta/Notch signaling is conserved in Gryllus.

Compared with control legs with distinguishable segments, including the coxa, trochanter, femur, tibia and tarsus, with the tibial spurs in the distal end of the tibia (arrow in Fig. 6A) and the tarsal spur in the distal end of tarsal segment 1 (arrowhead in Fig. 6A), Gb′Notch RNAi-treated embryos displayed shortened legs and a loss of joints, resulting in fusion of the femoral and tibial segments (Fig. 6B,C). In the RNAi-affected legs, tibial spur- and tarsal spur-like structures were observed (arrow and arrowhead, respectively, in Fig. 6C).

In Drosophila legs, activated epidermal growth factor receptor (EGFR) signaling is present in all segments and is necessary for joint formation (Galindo et al., 2005). A similar periodic pattern is observed for Gryllus Egfr (Gb′Egfr) expression during leg development (Nakamura et al., 2008). To evaluate the effects of Gb′Notch depletion on leg formation, we investigated Gb′Egfr expression in Gb′Notch RNAi-treated legs at stage 11. In control T3 legs, Gb′Egfr was expressed in the femur/tibia and tibia/tarsus joints (Fig. 6D, arrows) and as three tarsal rings (Fig. 6D, arrowheads). Expression of Gb′Egfr was severely reduced in Gb′Notch RNAi-treated T3 legs (Fig. 6E). In the RNAi legs, the most distal ring was retained (Fig. 6E, red arrowhead), and another tarsal ring (Fig. 6E, black arrowhead) and a more proximal domain (Fig. 6E, white arrowhead) were observed, probably corresponding to a region of spur-like structures in later embryos. Hence, Gb′Notch depletion appeared to result in loss of leg joints, accompanied by a reduction of EGFR signaling activity. By contrast, the Gb′Egfr rings were not affected in Gb′Delta RNAi-treated embryos at this stage (n>20, data not shown).

In addition, we also investigated the expression patterns of orthologs of the Drosophila leg patterning genes, dachshund and Distal-less, in Gb′Notch RNAi-treated legs at stage 11. In T3 legs of Gryllus embryos, two broad rings of the Gb′dachshund (Gb′dac) expression were apparent in the distal femur and distal tibia/proximal tarsus regions at this stage (Inoue et al., 2002) (Fig. 6F, arrowheads). In Gb′Notch RNAi-treated legs, only a single broad expression domain was observed (Fig. 6G, bracket), consistent with the observed morphological defects of reduced segment size and fusion of femoral to tarsal segments. In wild-type embryo legs, Gb′Distal-less (Gb′Dll) was strongly expressed in the distal tip and more proximal tarsal domains (Fig. 6H, arrowheads), as well as in the region from the proximal tarsus to the distal tibia, but weakly expressed in the femur to the distal trochanter (Inoue et al., 2002) (Fig. 6H). Gb′Dll was expressed strongly in the distal region of the legs of Gb′Notch RNAi-treated embryos (Fig. 6I, bracket) and weakly in a more proximal region adjacent to the strong expression domain (Fig. 6I, arrowhead). Such an alteration of the Gb′Dll expression pattern is consistent with the morphological fusion and shortening of the leg segments in RNAi-treated embryo legs. These results indicate that, in Gryllus leg development, Notch signaling acts downstream of the proximodistal axis patterning genes to form the leg joints and regulate leg growth.

As Delta and Notch RNAi phenotypes exhibited segmentation defects, we examined further the effects of RNAi treatment during the segment patterning process. We observed that Gb′Delta was expressed in segmental stripes earlier than Gb′wg stripes (Fig. 2F), suggesting that Gb′Delta/Notch signaling could function upstream of segment polarity genes, such as Gb′wg, in segmentation.

If the body segment malformations observed in Gb′Delta or Gb′Notch RNAi-treated embryos were caused by defects in the segmentation process, silencing of Gb′Delta or Gb′Notch should affect the Gb′wg striped pattern. To examine this possibility, we analyzed expression of Gb′wg in Gb′Delta and Gb′Notch RNAi-treated embryos during early germband stages (Fig. 7). In control embryos at 36 hAEL, Gb′wg was expressed as six anterior (gnathal and thoracic) segmental stripes with specific domains in the anterior head and posterior terminal region (Miyawaki et al., 2004) (Fig. 7A). We detected a reduction of the number of stripes in Gb′Delta RNAi-treated embryos (61.9%, n=13/21) and Gb′Notch RNAi-treated embryos (63.6%, n=7/11) (Fig. 7B-E). In the most severely affected Gb′Delta RNAi (28.6%, n=6/21) and Gb′Notch RNAi-treated embryos (27.3%, n=3/11), only two segmental stripes were observed at this stage and the posterior terminal domain was retained (Fig. 7C,E). Given that at least four segmental stripes were observed in subsequent stages (Fig. 7G-J), these reductions in stripe number might be due, at least in part, to a segment patterning delay. In more mature embryos, at 45 hAEL, which normally have eight segmental stripes of Gb′wg in the anterior head and posterior terminal domains (Fig. 7F), Gb′Delta and Gb′Notch RNAi treatment resulted in a reduction of stripe number, disorganization of the stripe pattern, loss or severe reduction of the posterior terminal domain and loss of the midline separation between the left and right stripes (Fig. 7G-J). In addition, severely affected embryos had reduced posterior growth zone and arrested AP axis elongation at this stage (Fig. 7H,J, arrows). These effects appear to be related to defects observed in later stage Gb′Delta and Gb′Notch RNAi-treated embryos (Fig. 4E,H), namely the loss of thoracic and abdominal segments, fusion of left and right legs, and segment malformation. Thus, inhibition of Notch signaling affects segmentation gene expression in early germbands producing a defective segment morphology phenotype.

To explore further the germband elongation defect, we analyzed expression of Gb′caudal (Gb′cad) in Gb′Delta and Gb′Notch RNAi-treated embryos during posterior elongation. Given that Gb′cad is required for formation of the posterior region and Gb′cad RNAi-treated embryos show a severe defect of posterior segments including all thoracic segments (Shinmyo et al., 2005), we predicted that Gb′cad expression would be affected in Gb′Delta and Gb′Notch RNAi-treated embryos during posterior elongation. Contrary to our expectations, the posterior domain of Gb′cad was retained in RNAi-treated embryos (100%, n=22/22, Gb′Delta RNAi; 100%, n=18/18, Gb′Notch RNAi) (Fig. 8A-F). Conversely, we also examined Gb′Delta expression in Gb′cad RNAi-treated embryos. In Gb′cad RNAi-treated embryos, Gb′Delta stripes were reduced whereas the posterior expression of Gb′Delta was retained in a reduced domain compared with that in control embryos (100%, n=6/6) (Fig. 8G,H), suggesting that Gb′cad is necessary for expression of Gb′Delta in anterior segments, but not in the posterior region.

To investigate the cause of the posterior elongation defect, we examined proliferation and cell death in Gb′Notch RNAi-treated embryos. An EdU incorporation assay for cell proliferation assessment (Salic and Mitchison, 2008) showed a pronounced reduction of mitotic cells in the posterior region of Gb′Notch RNAi-treated embryos (33%; Fig. 8J,K) relative to control embryos (70%; Fig. 8I,K). There was no significant difference in the rate of cell death between control and Gb′Notch RNAi-treated embryos (data not shown). Therefore, the defect in posterior elongation in Gb′Notch RNAi-treated embryos might be caused by inhibition of posterior cell proliferation. We were unable to assess the RNAi effects on cell proliferation in anterior regions owing to the presence of significant changes in anterior morphology in RNAi-affected embryos that prevented a precise comparison of cell proliferation rates in specific regions relative to wild-type embryos. Thus, it remains unclear whether cell proliferation is inhibited specifically in the posterior region. Taken together, these results suggest that Delta/Notch signaling might be involved in AP axis elongation during early germband development via control of cell proliferation, independently of the Wg/Cad pathway (Shinmyo et al., 2005).

We isolated Gb′Delta and Gb′Notch genes from the intermediate-germ cricket Gryllus bimaculatus and investigated their expression and function during embryogenesis. Our data suggest that Notch signaling is involved in segment morphogenesis, posterior elongation, neural differentiation and leg segmentation.

Gb′Delta and Gb′Notch were expressed in developing legs in a periodic pattern of concentric rings. RNAi knockdown of Gb′Notch expression markedly reduced leg length with a loss of joints and disrupted leg patterning gene expression. These results suggest that Notch signaling is essential for the correct specification of segment boundaries and normal leg growth in Gryllus.

In flies and spiders, Notch and its ligands are expressed in developing legs, as observed in Gryllus, and depletion of Notch causes segment fusion and leg shortening (de Celis et al., 1998; Bishop et al., 1999; Rauskolb and Irvine, 1999; Prpic and Damen, 2009). de Celis et al. (de Celis et al., 1998) have proposed that Notch activity is required for the establishment of boundaries that act as organizing centers for subsequent growth within leg segments, based on their mosaic analyses of developing legs lacking Notch, as well as data from experiments involving several mutants that affect leg segment lengths and/or the formation of intervening joints. Notch activity might also play such a role in Gryllus leg development.

Compared with Drosophila Notch mutants, Drosophila mutants of the Notch ligands Delta and Serrate show mild phenotypes in leg formation (Bishop et al., 1999). Drosophila wild-type legs contain five tarsal segments in a distal portion, whereas Delta mutant legs are shortened with fused tarsal segments 2-4. Ultimately, the Notch mutant phenotype looks like a composition of the Drosophila Delta and Serrate mutant phenotypes (Bishop et al., 1999). We could not detect obvious abnormalities in leg formation in Gb′Delta RNAi-treated embryos by our morphological analysis or WISH analysis with a Gb′Egfr probe (data not shown), implying that clearly discernible trochanter/femur, femur/tibia, and tibia/tarsus joints were formed normally in these embryos. Thus, it is possible that there are other Notch ligand(s) that function redundantly with Gb′Delta in developing legs. However, it is impossible to evaluate correctly whether more distal regions (i.e. the tarsus and pre-tarsus) of the legs of Gb′Delta RNAi-treated animals are normally segmented in these analyses because most Gb′Delta RNAi-treated embryos ceased to develop prior to showing distinguishable morphologies or Gb′Egfr expression in these structures. Nevertheless, though it is unclear whether Gb′Delta is necessary for tarsal segmentation, the role of Delta/Notch signaling in leg development seems to be highly conserved among Drosophila, Cupiennius and Gryllus species.

In the present experiments, RNAi knockdown of either Gb′Delta or Gb′Notch RNAi disrupted trunk segment formation. The RNAi effects on segmentation are a composite of segmentation disturbance/malformation and posterior segment loss. Because Gb′wg stripes were less disorganized in early RNAi-treated embryos than at later stages (Figs 4, 7), Gb′Delta/Notch signaling might not be necessary for specification of segment boundaries but rather might be critical for segment morphogenesis. Administration of RNAi directed against components of the Delta/Notch signaling pathway, including Notch, Delta, Presenilin and Suppressor of Hairless, in Cupiennius spiders also resulted in severe segment malformation accompanied by disorganization of segmentation gene expression (Schoppmeier and Damen, 2005; Stollewerk et al., 2003). Hence, the function of Notch signaling in segment morphogenesis might be conserved between Gryllus and Cupiennius, though the causes of segment malformation in these species remain to be explored further for more detailed comparisons.

The segmentation disturbance observed in the RNAi-treated Gryllus embryos is also reminiscent of a defect seen in Drosophila Notch and Delta mutant embryos, in which a segmentation disturbance was attributed to abnormal determination of the mesectoderm (ventral midline cells) (Menne and Klambt, 1994; Martin-Bermudo et al., 1995). Whether Notch signaling plays similar roles in mesoderm patterning and ventral midline specification in Gryllus as in Drosophila should be the focus of future studies.

The loss of sequentially formed segments in Gb′Delta and Gb′Notch RNAi-treated Gryllus embryos is likely to be at least partially due to a reduction in cell proliferation in the growth zone. Expression of Gb′Delta in the posterior growth zone in a specific domain at early stages might be essential for control of posterior elongation and regulation of the region of Notch signaling activation. As the early posterior domains of Gb′cad and Gb′wg are retained in Gb′Delta and Gb′Notch RNAi-treated embryos, Notch signaling might not be important for establishing the posterior growth zone in early germbands, but rather might be involved in maintaining a functional growth zone during subsequent posterior elongation processes. Furthermore, it should be noted that neither expression of Gb′cad in the terminal region of Gb′Delta and Gb′Notch RNAi-treated embryos nor expression of Gb′Delta in the terminal region of Gb′cad RNAi-treated embryos was eliminated. We showed previously that Gb′Wg/Arm signaling and Gb′cad are essential for posterior elongation (Miyawaki et al., 2004; Shinmyo et al., 2005). Therefore, the function of Gb′Delta/Notch signaling in posterior elongation might be parallel to the cascade for posterior elongation mediating Gb′wg and Gb′cad.

In another basal, short-germ insect, the cockroach Periplaneta americana, Notch RNAi treatment has been shown to result in a loss of abdominal segments and reduced cell proliferation in the terminal region (Pueyo et al., 2008), though RNAi effects on segment formation appeared to be weaker than those observed for Gryllus. Thus, the function of Notch signaling in posterior elongation might be conserved among short/intermediate-germ insects. In the cockroach, RNAi directed against Notch affects hairy segmental expression, which was thought to be a consequence of disruption of a Notch-mediated segmentation mechanism. However, RNAi effects on posterior segmentation could include secondary defects as a result of a failure to maintain a functional growth zone. Thus, the existence of a Notch-mediated segmentation mechanism remains ambiguous in insects. However, the situation for the spider seems to be different from that of the cockroach. That is, RNAi directed against Delta and Notch was shown to affect hairy segmental expression in Cupiennius, without any observable reduction in cell proliferation in the growth zone (Stollewerk et al., 2003). Nevertheless, malformation of the posterior growth zone was observed. Thus, the cause of the segment defect remains unclear.

Previous work in the spider Achaearanea tepidariorum revealed that Delta/Notch signaling is essential for caudal lobe formation preceding initiation of segmentation (Oda et al., 2007). In Achaearanea embryos, Delta and Notch RNAi treatment results in a severe reduction of the posterior region (opisthosoma) due to a failure to generate cad-expressing caudal ectoderm cells, suggesting that Delta/Notch signaling promotes caudal expression and caudal ectoderm fate specification (Oda et al., 2007). Considering the findings of Oda et al. in Achaearanea, our findings in Gryllus suggest that Notch signaling involvement in the posterior elongation process might be conserved between Gryllus and Achaearanea, though the genetic interactions between Notch signaling and cad might differ between them. In Cupiennius, however, the posterior region was not reduced in Delta and Notch RNAi-treated embryos, though it was malformed without a reduction in cell proliferation in the growth zone (Stollewerk et al., 2003). Hence, the function of Notch signaling in posterior patterning seems to be divergent between these arthropod species. However, it should be noted that in Cupiennius, dsRNA was administered to embryos at a relatively later stage than in Gryllus and Achaearanea, possibly after segmentation had already been initiated (Schoppmeier and Damen, 2001). Thus, the phenotypic discrepancy between these species might arise from differences in RNAi methods and/or RNAi efficiency. Intriguingly, RNAi silencing of Presenilin and Suppressor of Hairless causes a defect in posterior growth and a breakdown of posterior segmentation in Cupiennius (Schoppmeier and Damen, 2005), implying that Notch signaling in Cupiennius is also involved in the posterior elongation process.

In light of analogous findings in other arthropod species, the present findings suggest that involvement of Delta/Notch signaling in posterior elongation might be an ancestral mode among arthropods. Further analyses of the posterior growth zone are needed to fully understand Delta/Notch involvement in arthropod embryogenesis.

We thank Takashi Matsushita, Tomohiro Uda and Chieko Ishifune for their technical assistance. This work was supported by a grant from the Japanese Ministry of Education, Culture and Sports, Science and Technology to T.M., H.O. and S.N.