Cell migration that orients the dorsoventral axis is coordinated with anteroposterior patterning mediated by Hedgehog signaling in the early spider embryo

The migration of a signaling source is a common strategy for directing the activation or inhibition of a signaling pathway in the appropriate regions during animal development. Such migration must be regulated in concert with ongoing pattern formation to achieve the correct cell-cell interactions (Solnica-Krezel, 2005; Tam et al., 2006). Although the molecular machinery and guiding factors underlying cell migration have been studied using in vitro culture systems and in vivo model systems (Cram et al., 2006; Simpson et al., 2008; Wang et al., 2006), there are few systems that allow simultaneous study of cell migration and pattern formation.

The migration of cells serving as the signaling source is crucial for formation of the embryonic axis in some model organisms. During egg cylinder development in mouse, the distal visceral endoderm cells, which express antagonists of Nodal and Wnt (Kimura-Yoshida et al., 2005; Yamamoto et al., 2004), migrate towards the future anterior side (Beddington and Robertson, 1999). This cell migration is one of the earliest steps in axis formation in mouse, and the importance of the signaling molecules has been established genetically. However, accessibility to the migration and patterning events in mouse egg cylinders is limited.

Early spider embryos, like the mouse embryo, exhibit the migration of a cellular source that secretes signaling molecules. In the house spider Achaearanea tepidariorum, which is emerging as a model organism (Oda and Akiyama-Oda, 2008), a cluster of endoderm cells, termed the cumulus mesenchymal (CM) cells, is internalized from the blastopore and migrates on the basal surface of the static germ disc epithelium (see Fig. 1A) (Akiyama-Oda and Oda, 2003). The migratory CM cells and the associated surface epithelial cells can be easily observed as a distinct white spot that is conventionally called the cumulus. The CM cells express a homolog of decapentaplegic, At-dpp, that encodes the evolutionarily conserved dorsal signal (Akiyama-Oda and Oda, 2003). During migration, the Dpp signal from the CM cells affects the surface epithelium: in these cells, the transcription factor Mothers against dpp (Mad) is phosphorylated and translocated to the nucleus (Fig. 1B).

During spider development, the formation and migration of the CM cells is preceded by the formation of a single embryonic axis termed the embryonic-abembryonic (Em-Ab) axis (Fig. 1A). Centrifugal migration of the CM cells from the Em polar region is the first morphological sign of dorsoventral (D-V) axis formation. No molecular asymmetries that indicate the direction of CM cell migration have yet been identified. As the CM cells arrive at a site on the germ disc rim, the surrounding epithelial cells begin to differentiate into extraembryonic tissue, signifying the dorsal side of the developing embryo. The extraembryonic tissue rapidly expands as the germ disc transforms into a germ band in which the anteroposterior (A-P) and D-V axes become evident. The anterior and caudal regions of the forming germ band appear to be related to the peripheral and central regions, respectively, of the germ disc. Our previous study using parental RNA interference (RNAi) showed that the Dpp signal is essential for extraembryonic tissue formation and D-V axis polarization, although the CM cell migration itself might not require this signal (Akiyama-Oda and Oda, 2006). Because of the easily visible cell migration and its functional importance in axis formation, as well as the availability of parental RNAi, the early spider embryo offers an invaluable opportunity to study cell migration and pattern formation simultaneously.

In this study, we conducted a pilot screen for genes involved in CM cell migration in the Achaearanea embryo. We found that components of the Hedgehog (Hh) signaling pathway are required for CM cell migration, germ disc patterning and axis formation. Based on these results, we propose a model in which CM cell migration is coordinated with pattern formation mediated by Hh signaling.

Laboratory stocks of the house spider Achaearanea tepidariorum were maintained at 25°C. Every mated female was evaluated by monitoring the development of embryos in the first egg sac; only those that produced healthy eggs were used in experiments. Developmental stages were described previously (Akiyama-Oda and Oda, 2003; Yamazaki et al., 2005). Malformed embryos following RNAi treatment were staged based on the time past stage 4. In the experiments in which early embryos were used, the development of sibling embryos derived from the same egg sacs was checked later to confirm that consistent results were observed.

For information about the cDNAs used in this study, see Table S1 in the supplementary material. cDNA fragments of At-ptc, At-smo, At-labial (At-lab) and At-gataC were originally isolated by PCR using degenerate primers (see Table S2 in the supplementary material). Longer cDNA clones were then isolated by 5′ and 3′ RACE using a SMART RACE cDNA Amplification Kit (Clontech, Mountain View, CA, USA). Primers used for RACE are shown in Table S2 in the supplementary material. Phylogenetic trees were constructed for Ptc and Smo by the neighbor-joining method (Saitou and Nei, 1987) using PHYLIP 4.0 (see Fig. S1 in the supplementary material). The nucleotide sequences reported here are available in the DDBJ/EMBL/GenBank databases under the accession numbers AB433900-AB433908 and AB524079.

The 709 bp (nt 1-709) and 720 bp (nt 706-1425) regions of At-ptc cDNA, the 1020 bp (nt 1-1020) and 924 bp (nt 1017-1940) regions of At-hh cDNA, the 633 bp (nt 1543-2175) and 823 bp (nt 2176-2998) regions of At-smo cDNA and the entire coding region of green fluorescent protein (gfp) cDNA (Quantum) were used to synthesize At-ptc1, At-ptc2, At-hh1, At-hh2, At-smo1, At-smo2 and gfp double-stranded (ds) RNAs, respectively. gfp dsRNA was used as the control. At-Delta^HH dsRNA was prepared as described previously (Oda et al., 2007). dsRNAs were synthesized as described (Akiyama-Oda and Oda, 2006) and were used at 1.5-2.0 μg/μl for injection. Females were given four injections of dsRNA solution (1-2 μl each) at 2- to 3-day intervals, except for five females: one for At-smo1 and two for At-smo2 were each given three injections; one for At-smo1 and one for At-Delta^HH were given six and five injections, respectively. For simultaneous RNAi against two genes (double RNAi), females were given a total of eight injections (one dsRNA solution after the other) at 2- to 3-day intervals.

Each pool of mRNA was prepared from 30-60 embryos using the QuickPrep Micro mRNA Purification Kit (GE Healthcare, UK). Synthesis of first-strand cDNA was performed using random primers and SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA, USA), followed by PCR using ExTaq polymerase (Takara, Shiga, Japan). The PCR conditions were as follows: 95°C for 30 seconds; 25 or 30 cycles of 95°C for 5 seconds and 60°C for 20 seconds. Controls lacking reverse transcriptase in the cDNA synthesis reaction failed to produce specific products in all cases. Real-time PCR and quantification were performed using SYBR Premix ExTaq with a Thermal Cycler Dice Real-Time System TP800 (Takara). The PCR cycles were as above, followed by a dissociation cycle: 95°C for 15 seconds, 60°C for 30 seconds and 95°C for 15 seconds. The dissociation curves were examined to confirm the presence of the single expected product and the absence of non-specific products or primer dimers. The relative quantities were calculated by the crossing-point standard curve method. All real-time reactions were repeated several times. The nucleotide sequences (5′ to 3′) of the primer pairs were as follows: At-ptc, GCAGTAACTACCAGAGTTACAGC and CTTCTGGATGAGGAACTATCACG; At-hh, TGGTGTAGTAGCTTCCTGCTACG and TGTACCTGAGGATGTGCATAGTC; At-smo, GATCATCTTGTGTGCCTTGC and TGATACATGTGGGCATTTGG; At-ef1α, CTGTACCAGGAGACAATG and ATCTGACCAGGATGGTTC; At-dpp, CCGCATGAGAATTATGGACTGC and CGACTTGATTCCACCTATGAGG; At-α-catenin,ATGCAGCTCGTATATTGG and CACCCTCTTGTAGAGCAA.

Single and dual-color whole-mount in situ hybridization (WISH) and staining with antibody against phosphorylated Mad (pMad) (Persson et al., 1998) were performed as described previously (Akiyama-Oda and Oda, 2003). To detect FITC probes in dual-color WISH, alkaline phosphatase (AP)-conjugated rabbit anti-FITC (DAKO, 1:200 dilution) or sheep anti-fluorescein-AP (Roche, 1:1000) antibody was used. Most samples were counterstained with DAPI. Alexa Fluor 488 phalloidin (Molecular Probes, Eugene, OR, USA) was used for F-actin staining.

Preparation of embryos was as described previously (Oda et al., 2007). Images were acquired every 10 minutes using a stereomicroscope (SZX12, Olympus, Tokyo, Japan) equipped with a color CCD camera (C7780-10, Hamamatsu Photonics, Shizuoka, Japan) controlled by AquaCosmos software (Hamamatsu Photonics). Images were processed using ImageJ 1.42q, MetaMorph 6.1 and Adobe Photoshop CS2 software.

The CM cell cluster starts to migrate from the center of the germ disc at early stage 5 (Fig. 1A). Prior to the start of migration, scattered At-Delta expression appears around the germ disc center; this At-Delta expression is involved in caudal specification (Oda et al., 2007). We examined the relationship between the position of the CM cells and the pattern of At-Delta expression by dual-color WISH. A singed-related gene, 022_P10, was used as a marker of CM cells. As the At-Delta expression domain expanded symmetrically from the germ disc center, the CM cells shifted to an asymmetric position (Fig. 1C). In the majority of mid-stage 5 embryos, the migrating CM cells were positioned close to the border of the At-Delta expression domain. It appeared that CM cell migration was accompanied by dynamic expansion of gene expression starting from the germ disc center. However, the expansion of the At-Delta expression domain stopped at approximately half of the radius of the germ disc, whereas the CM cells continued to migrate until they reached the germ disc rim at late stage 5.

To search for genes required for CM cell migration, we performed a pilot gene-knockdown screen. dsRNAs were prepared from more than 30 genes, most of which were originally isolated as homologs of Drosophila and/or vertebrate patterning genes (see Table S3 in the supplementary material). Each dsRNA was injected into adult females, and the live embryos they produced were observed under a stereomicroscope. In this initial screen, we identified only one candidate gene, At-ptc, a homolog of the Drosophila segment polarity gene patched (ptc). Injection of dsRNA targeting At-ptc (At-ptc1 dsRNA) resulted in embryos in which the cumulus shift was prevented or largely delayed (Fig. 1D, Fig. 2D; see Movie 1 in the supplementary material). The penetrance of this phenotype was high (see Fig. S2 in the supplementary material). Until early stage 5, these embryos were morphologically indistinguishable from wild-type embryos. The prevention of cumulus shift was accompanied by the ectopic formation of extraembryonic tissue around the germ disc center. In extreme cases, the extraembryonic tissue spread over the Em hemisphere, with all embryonic cells shifted to the Ab side, resulting in a pear-shaped form.

To examine RNAi specificity, we prepared another dsRNA (At-ptc2 dsRNA) from a different region of the At-ptc cDNA (see Fig. S2A in the supplementary material). Injection of At-ptc2 dsRNA, but not control gfp dsRNA, yielded the same phenotype as At-ptc1 dsRNA (see Fig. S2B in the supplementary material). Moreover, real-time PCR revealed specific reduction of the At-ptc transcript level, and WISH revealed reduced signals for the cytoplasmic pool of At-ptc transcripts (see Fig. S2C,D in the supplementary material). These results led us to conclude that specific suppression of At-ptc function resulted in prevention of the cumulus shift. Hereafter, embryos derived from females injected with At-ptc or gfp dsRNA are referred to as At-ptc or gfp RNAi embryos, respectively. Because the same results were obtained with At-ptc1 and At-ptc2 dsRNAs, only results obtained with At-ptc1 are shown.

ptc encodes a membrane receptor for the secreted Hh signal (Fuse et al., 1999; Ingham et al., 1991; Marigo et al., 1996; Stone et al., 1996), and Ptc is thought to be a negative regulator of Hh signaling (Ingham et al., 1991; Taipale et al., 2002). In the absence of ligand, Ptc represses the activity of another membrane receptor, Smo (Alcedo et al., 1996; van den Heuvel and Ingham, 1996). When bound by Hh, Ptc no longer inhibits Smo, allowing Smo to transduce the Hh signal intracellularly (Hooper and Scott, 2005; Ingham and McMahon, 2001; Lum and Beachy, 2004). The removal of Ptc also leads to Smo activation even in the absence of Hh (Hooper and Scott, 2005).

To investigate the potential contribution of other components of the Hh signaling pathway to CM cell migration, we performed parental RNAi against the Achaearanea homologs of hh and smo. Two different dsRNAs were prepared for each gene (At-hh1, At-hh2, At-smo1 and At-smo2; see Fig. S3 in the supplementary material). Injection of these four dsRNAs resulted in embryos that had similar phenotypes. This indicated that the phenotypes were specific effects of At-hh and At-smo knockdown and also implied that At-hh and At-smo function in the same pathway. Specific reduction of the target transcripts was also confirmed. The resultant embryos are referred to as At-hh and At-smo RNAi embryos.

Time-lapse recording of At-smo RNAi embryos revealed that in the most severely affected egg sacs, more than 60% of embryos showed defects associated with CM cell migration. The defects were categorized into three types (Fig. 2B,C; see Movie 1 in the supplementary material): the first and the second defects were a prevention and a delay, respectively, of the cumulus shift observed around the center of the germ disc, whereas the third defect was a cumulus shift that continued around the periphery of the germ disc. These three types of defect were also found in At-hh RNAi embryos (Fig. 2E,F; see Movie 1 in the supplementary material). However, more than 70% of the embryos showed a normal, or an almost normal, cumulus shift, even in severely affected egg sacs. Owing to the low penetrance of the cumulus-shift phenotypes, we initially failed to identify At-hh as positive in the screen described above (see Table S3 in the supplementary material).

In our analyses of serial egg sacs from females injected with At-smo dsRNA (Fig. 2B,C; see also Fig. 8H), we noticed that prevention of cumulus shift was observed only around the time that the maximum effect of RNAi was expected (around the twentieth day from the start of injection, when four injections were performed) (Akiyama-Oda and Oda, 2006; Oda et al., 2007). By contrast, the delayed and continued cumulus-shift defects were observed in a wider time window, and the proportion of the latter decreased around the twentieth day. Based on these observations, we concluded that the prevention of cumulus shift reflects a severe effect of At-smo RNAi, whereas the continued shift reflects a mild effect, and the delayed shift an intermediate effect.

We next examined whether the CM cells formed normally in At-ptc, At-hh and At-smo RNAi embryos. As revealed by phalloidin staining, the CM cells showed normal internalization and clustering and a normal distribution of F-actin (Fig. 3A). No abnormalities were observed in the configuration of the surface epithelium around the closed blastopore (data not shown). Staining for the CM cell markers 022_P10 and At-dpp and for phosphorylated (p) Mad showed that the CM cells differentiated normally in these RNAi embryos and that they acted as the source of the Dpp signal (Fig. 3B-D).

We examined the consequences of the arrest of the Dpp signal source at the germ disc center in At-ptc RNAi embryos. The D-V pattern elements are arranged in the order: the extraembryonic tissue, the At-gataC expression domain and the At-short gastrulation (At-sog) expression domain (Fig. 4A,B). Observation of these D-V pattern elements in At-ptc RNAi embryos showed that although a D-V axis developed, it was oriented in parallel to the Em-Ab axis (Fig. 4C,D). Normally, expression of At-sog, which encodes a Dpp antagonist, is confined to the ventral midline region of the elongating germ band, where At-single-minded (At-sim) is subsequently expressed (Akiyama-Oda and Oda, 2006). In the At-ptc RNAi embryo, this process appeared to take place in the apical region where expression of the anterior marker At-orthodenticle (At-otd) was observed (Fig. 4E-G). These results indicated that the At-ptc RNAi embryo failed to orthogonalize the major embryonic axes. This was consistent with the expression pattern of the segment marker At-engrailed (At-en), which was expressed in concentric circles (Fig. 4H). Taken together, these data suggested that the cumulus shift is a crucial step in the formation of the bilaterally symmetric body pattern.

In At-hh and At-smo RNAi embryos, regardless of CM cell migration, the germ disc gradually shrank during and after stage 6 and failed to form a germ band (see Movie 1 in the supplementary material). Although the relative quantity of At-dpp transcripts was not much changed in these RNAi embryos (see Fig. S3D in the supplementary material), the D-V pattern did not develop (Fig. 4I). At-sog expression decreased, and the At-gataC expression domain failed to form. This condition was in contrast to the clear D-V pattern that developed in At-ptc RNAi embryos (Fig. 4C,D). These data suggested that At-hh and At-smo, but not At-ptc, are necessary for the progression of D-V axis development.

To further investigate the contribution of Hh signaling to axis formation, we examined late stage 7 embryos. At this stage, there is normally expression of the anterior (At-otd) and caudal [At-caudal (At-cad)] markers on the opposite sides of the extending germ bands (Fig. 5A). At-ptc RNAi embryos lacked At-cad but not At-otd expression. Conversely, At-hh and At-smo RNAi embryos lacked At-otd but not At-cad expression. Furthermore, the majority of At-hh and At-smo RNAi embryos had a greatly expanded At-cad expression domain. These data suggested that At-ptc is required for posterior patterning and that At-hh and At-smo are required for anterior patterning. The data also suggested that At-hh and At-smo are involved in the repression of caudal fate.

Notably, At-ptc RNAi embryos showed only two At-en stripes at late stage 8; this contrasts with wild-type embryos at the same stage, which had more than seven stripes (Fig. 4H), and with At-hh and At-smo RNAi embryos, which showed no At-en expression (Fig. 5B). These results suggested that Hh signaling is involved in segmentation of the spider embryo.

To better understand the phenotypic data, we examined the expression of At-ptc, At-hh and At-smo transcripts during early stages of embryonic development. WISH could not be used in stage 1 and 2 embryos owing to technical problems with the fixation process. Using RT-PCR, At-ptc and At-smo transcripts were detected at considerable levels in stage 1 embryos, but At-hh transcripts were not detected (see Fig. S4 in the supplementary material). Stage 2 was the earliest stage at which At-hh transcripts could be detected by RT-PCR. These data suggested that At-ptc and At-smo transcripts, but not At-hh transcripts, might be maternally loaded in the egg.

At-hh expression was observed on the Ab side of the embryo at stage 3 (Fig. 6A), and expression was restricted to the rim of the germ disc by early stage 5 (Fig. 6B). At-ptc transcripts were detected in cells located on the Em side at stage 3, with the highest levels in cells near the blastopore region (Fig. 6E). In the blastopore region, we observed cells that expressed both At-ptc and the CM-cell marker (Fig. 6N,N′), although it was unclear whether these cells were future CM cells. Ubiquitous expression of At-ptc transcripts was observed in the germ disc epithelium at stage 4 (Fig. 6F). However, this expression pattern gradually changed during stage 5, when expression was reduced in the central region of the germ disc and was enhanced in the peripheral region adjacent to the At-hh-expressing circular domain (Fig. 6I,O-Q). The reduction of At-ptc expression preceded the expansion of the At-Delta expression domain from the center (Fig. 6S,T). The expression patterns of At-hh and At-ptc in the germ disc did not show any asymmetry with respect to the position of the migrating CM cells during stage 5 (Fig. 6L,M,O-R), although a few At-ptc-positive cells were irregularly scattered near the germ disc center in about half of the stage 5 embryos examined (Fig. 6I).

During stage 6, the forming caudal lobe initiated At-hh and At-ptc expression, whereas the anterior region of the forming germ band, which is related to the germ disc peripheral region, showed a stripe of At-hh expression flanked with stripes of At-ptc expression (Fig. 6C,G,J). The position of the At-hh stripe relative to the anterior rim was shifted several cell widths to the posterior after stage 5 (Fig. 6B,C). Later, the formed germ band exhibited segmental stripes of expression of these genes (Fig. 6D,H,K), as is seen in Drosophila (Hooper and Scott, 1989; Lee et al., 1992; Nakano et al., 1989; Tabata et al., 1992). Although At-smo transcripts were detected during the early stages by RT-PCR (see Fig. S4 in the supplementary material), a specific At-smo expression pattern was not detected by WISH.

The expression pattern of At-hh was examined in At-Delta and At-smo RNAi embryos. Like At-hh, At-Delta is initially expressed on the Ab side of the embryo (Oda et al., 2007). The initial At-hh expression on the Ab side of the stage 3 embryo (Fig. 6A) was independent of At-Delta and At-smo (Fig. 6U,Y), but the later At-hh expression at the rim of the germ disc (Fig. 6B,I,L) was dependent on At-Delta and At-smo (Fig. 6V,Z; see Fig. S3 in the supplementary material). At-ptc expression was also examined in At-smo RNAi embryos. The At-ptc expression at the blastopore region (Fig. 6E,F) was independent of At-smo (Fig. 6W), whereas the At-ptc expression in the germ disc epithelium (Fig. 6O-Q) was dependent on At-smo (Fig. 6W,X). This result is consistent with ptc being a known target of Hh signaling (Alexandre et al., 1996; Goodrich et al., 1996; Hidalgo and Ingham, 1990).

The data described above suggested that Hh signaling is involved in the formation of the A-P axis, which is first evident in the germ band (Fig. 5). To investigate the possibility that Hh signaling regulates patterning events in the germ disc, we examined the expression of four distinct regional markers in At-ptc, At-hh and At-smo RNAi germ discs at late stage 5: At-Delta, At-Deformed (At-Dfd), At-otd and At-lab. At-Delta expression, which is normally present in the central region of the germ disc, was expanded to the rim in the At-hh and At-smo RNAi germ discs; conversely, it was missing entirely in the At-ptc RNAi germ disc (Fig. 7A). Early stage 5 At-ptc RNAi embryos also lacked the At-Delta expression (not shown). These results indicate that At-ptc, or negative regulation of Hh signaling, is required for the initiation of At-Delta expression around the germ disc center, whereas activation of Hh signaling represses the central fate specification in the peripheral region. Similarly, At-Dfd expression, which is normally present in the broad region of the germ disc, except within several cell widths of the periphery, was expanded to the rim in the At-hh and At-smo RNAi germ discs, although it was normal in the At-ptc RNAi germ disc (Fig. 7B,B′). At-otd and At-lab expression normally appear at and near the rim. The domain of the latter was broader than that of the former, and both overlapped the At-hh expression domain (for details, see Fig. S5 in the supplementary material). The At-otd expression was missing entirely in the At-hh and At-smo RNAi germ discs, but was ectopically induced around the central region, with unaffected rim expression, in the At-ptc RNAi germ disc (Fig. 7C). The At-lab expression was greatly reduced in the At-hh and At-smo RNAi germ discs and, conversely, was expanded towards the central region in the At-ptc RNAi germ disc (Fig. 7D). These results indicated that activation of Hh signaling promotes the specification of peripheral fate. Thus, Hh signaling specifies the central-peripheral pattern in the germ disc, which is related to the caudal-anterior pattern of the germ band. Similarly, the specification of the central and peripheral mesoderm was dependent on Hh signaling, although the specification of central and peripheral endoderm was not (see Fig. S6 in the supplementary material).

Double RNAi for At-ptc and At-smo yielded phenotypes similar to those of single RNAi for At-smo, whereas double RNAi for At-ptc and At-hh yielded phenotypes similar to those of single RNAi for At-ptc (Fig. 7). These epistatic relationships between At-hh, At-ptc and At-smo (At-hh → At-ptc → At-smo) are consistent with established findings in Drosophila and vertebrates (Hooper and Scott, 2005; Ingham and McMahon, 2001; Lum and Beachy, 2004).

In RNAi-mediated knockdown experiments, leaky expression of target gene products might be unavoidable. In the case of At-ptc single RNAi, there might be residual At-Ptc that is inhibited by At-Hh. In double RNAi for At-ptc and At-hh, such residual At-Ptc might contribute to the inhibition of At-Smo. With this in mind, one might expect that phenotypes for At-hh and At-ptc double RNAi are milder than those for At-ptc single RNAi and are more like those observed for At-hh single RNAi. This is consistent with the absence of central At-otd expression and the slight reduction in peripheral At-otd expression observed in the double-RNAi germ discs (Fig. 7C). Similar logic may be applied to the At-ptc and At-smo double RNAi. Thus, the central-peripheral pattern in the germ disc may depend on the activities (and the balance of activities) of the Hh signaling components.

To understand the process by which a reduction in Hh signaling activity results in greatly expanded At-Delta expression in late stage 5 germ discs (Fig. 7A), we examined At-smo RNAi embryos at earlier stages. In severe cases, At-Delta expression predominated in the germ disc from early stage 4 (Fig. 8A,B). This situation contrasts sharply with that of a normal germ disc, in which little At-Delta expression can be observed at early stage 4 (Fig. 1C). These observations suggested that Hh signaling activity is required in the entire germ disc prior to stage 5 for repression of central gene expression.

At early stage 5, when the CM cells begin migration in the normal embryo, At-smo RNAi germ discs showed various degrees of expansion of the At-Delta expression domain (Fig. 8D-G). To investigate the relationship between the expansion phenotype and the cumulus-shift phenotype (Fig. 2B,C), we analyzed serial egg sacs derived from individual females injected with At-smo1 dsRNA or gfp dsRNA (Fig. 8H). Embryos from each egg sac were divided into two pools. One pool was examined for At-Delta expression at early stage 5 and the expansion levels categorized into four classes: class I, comparable to normal; class II, slightly expanded; class III, largely, but not fully, expanded (some part of the expression domain reaching the rim); class IV, fully expanded (Fig. 8D-G). The second pool of embryos was examined for the cumulus shift. These analyses revealed that the prevention and delay of cumulus shifts correlated significantly with the expression of the class III and class IV phenotypes (Fig. 8I). These results suggested that the predominant central (or caudal) fate, which is evident at the molecular level prior to stage 5, correlates with a failure to initiate CM cell migration.

In this study, we took advantage of the availability of parental RNAi in the spider Achaearanea tepidariorum to carry out a function-based screen for genes involved in cell migration during embryogenesis. The knockdown of some genes affected CM cell migration but had no apparent effects on earlier events, such as blastoderm, germ disc and CM cell formation, a finding that validates the use of this unique model system for studying cell migration. A large-scale screen will most likely identify more genes involved in the regulation of CM cell migration. As with border cell migration in the Drosophila egg chamber (Montell, 2003; Rørth, 2002), the simplicity of this model system is advantageous, allowing the study of molecular mechanisms that regulate the initiation, direction and termination of cell migration.

Our data clearly show that Hh signaling is involved in specifying the peripheral-central pattern of the germ disc, similar to the role of Hh signaling in determining the D-V pattern of the vertebrate neural tube (Dessaud et al., 2008). The cell migration defects of At-ptc and At-smo RNAi embryos correlated with patterning defects in the germ disc. In the central region of the At-ptc RNAi germ disc, prevention of CM cell migration was accompanied by a lack of central gene expression and by ectopic expression of peripheral genes. In the peripheral region of the moderately affected At-smo RNAi germ disc, continued CM cell migration was accompanied by a lack of peripheral gene expression and by an expansion of central gene expression. Moreover, severely affected At-smo RNAi embryos exhibited prevention or delay of CM cell migration. This defect was concurrent with an early predominance of central gene expression over the germ disc. The simultaneous observation of cell migration defects and patterning defects might indicate that CM cell movement is tied to some aspect of positional information. We propose a model that explains the defects caused by perturbations in Hh signaling (Fig. 9).

Hh signaling mediates long-range patterning in a variety of vertebrate and invertebrate tissues (Briscoe et al., 2001; Dessaud et al., 2008; Ingham and McMahon, 2001; Ingham and Placzek, 2006). The predominance of central gene expression in the severely affected At-smo RNAi germ disc prior to stage 5 suggests that At-Hh initially travels from its source (the outside and rim of the forming and formed germ disc) up to the central region to repress central gene expression in the normal germ disc (Fig. 9A). Considering that At-ptc is a putative target of Hh signaling, this idea is also supported by the observation that At-ptc transcripts were ubiquitously distributed in the forming and formed germ discs at stages 3 and 4. The travelling distance of sonic hedgehog (Shh), a vertebrate homolog of Hh, was estimated to be ∼300 μm (∼30 cell diameters) in the mouse limb bud (Lewis et al., 2001). This distance is comparable to the radius of the Achaearanea germ disc. Initial rapid diffusion of the Hh ligand over a long distance has been supported by a simulation study of Shh signaling dynamics in the vertebrate neural tube (Saha and Schaffer, 2006). Our data obtained with At-Delta and At-smo RNAi embryos implies that the maintenance of the Hh signal source is regulated by Notch signaling and by Hh signaling itself.

Through At-Smo-mediated signal transduction, high levels of At-Hh (short-range signal) promote peripheral gene expression, whereas moderate and low levels of At-Hh (long-range signal) are probably able to repress central gene expression in the germ disc. This repression might be followed by a derepression process that involves At-ptc (Fig. 9B). The ptc gene is a target of Hh signaling (Alexandre et al., 1996; Goodrich et al., 1996; Hidalgo and Ingham, 1990), and the Ptc protein serves to inhibit Hh transport (Chen and Struhl, 1996; Jeong and McMahon, 2005). These two regulatory mechanisms might account for the progressive derepression of central gene expression from the germ disc center. At-ptc transcription is enhanced in more peripheral regions of the germ disc, depending on the distance from the At-Hh signal source. The peripheral At-Ptc inhibits the movement of At-Hh towards the germ disc center, and, accordingly, the central At-Ptc blocks At-Smo activity in progressively more distant regions from the germ disc center, leading to the derepression of central gene expression. This dynamic sequence of signaling events might contribute to determining the shape of the positional value gradient that reflects the future A-P axis (Fig. 9C).

The timing of the derepression of central gene expression is close to that of the initiation of CM cell migration. An important aspect of our model is that the CM cells move down along the emerging positional value gradient. The shape of the positional value gradient might be affected by depletion of components of the Hh signaling network, which may thereby affect the behavior of the CM cells (Fig. 9D-F). In At-ptc RNAi embryos, At-Smo is activated throughout the entire germ disc and the derepression mechanism fails to function, as the peak of the gradient is not formed at the germ disc center. The ectopic At-otd expression implies that the slope of the gradient may be reversed around the germ disc center and also that the terminal patterning system might function as in Drosophila (Finkelstein and Perrimon, 1990). In the At-smo RNAi embryos, the slope of the gradient may be reduced depending on the level of residual At-Smo activities. The reversed polarity and reduced slope of the positional value gradient around the germ disc center might result in the delay or prevention of CM cell migration. The defect in the termination of CM cell migration in the At-smo RNAi embryo might be attributable to a lack of a ‘stop signal’ associated with peripheral cell fate or to local positional value gradients potentially formed along the circumference of the germ disc.

More studies are needed to elucidate the molecular basis of the proposed positional value gradient. It is possible that concentrations of extracellular Hh or of other components of the Hh signaling network directly relate to the positional values. Previous studies proposed that Hh and Shh induce and restrict cell migration (Bijlsma et al., 2007; Deshpande et al., 2001; Fu et al., 2004). In particular, it was shown in vitro that mouse mesenchymal fibroblasts migrate towards an Shh gradient in a transcription-independent manner (Bijlsma et al., 2007). However, our data provide no evidence that At-Hh serves as a guiding cue for CM cell migration. Alternatively, secreted or cell-surface proteins expressed under the control of Hh signaling might be involved in the positional information that affects the CM cells. Although expression of At-Delta is controlled by Hh signaling, its knockdown does not affect CM cell migration (Oda et al., 2007). To test our hypotheses, technical advances, as well as the identification of more of the genes involved in the cell migration, are needed.

This study provides the first evidence that Hh signaling mediates the formation of the two major embryonic axes, the A-P and D-V axes, in a bilaterally symmetric animal. However, the importance of Hh signaling has been documented for tissue-level axis formation in Drosophila and vertebrates (Ingham and McMahon, 2001; Tabata, 2001). The roles that the Achaearanea Hh system plays in germ disc patterning are similar to those played by Drosophila Bicoid in A-P patterning (Driever and Nüsslein-Volhard, 1988). The former is based on the diffusion of extracellular signals in a cell-based embryo, and the latter is based on the diffusion of transcription factors in a syncytial embryo. This striking contrast indicates that drastic changes in developmental programs can occur without disrupting the basic arthropod body plan.

There is another important difference between Drosophila and Achaearanea in the cellular and molecular mechanisms that orthogonalize the two embryonic axes. In Drosophila, migration of the nucleus in the oocyte localizes the source of the Gurken signal, a TGFα-like protein (Neuman-Silberberg and Schüpbach, 1993), to an asymmetric position with respect to the primary axis, and this specifies the orientation of the D-V axis (Roth, 2003; van Eeden and St Johnston, 1999). Through many subsequent steps, the domain of dpp transcription is determined on the dorsal side of the blastoderm embryo (Morisato and Anderson, 1995). The corresponding symmetry-breaking event in Achaearanea appears to be the migration of the CM cells, which express the evolutionarily conserved dorsal signal Dpp, in the germ disc stage embryo (Oda and Akiyama-Oda, 2008). Thus far, no genes have been found that are expressed asymmetrically in the germ disc stage embryo prior to CM cell migration. Our model (Fig. 9) predicts that the initial direction of migration of the CM cells is determined through stochastic processes related to Hh signaling network dynamics. Stochastic processes have also been proposed to determine the direction of migration of the Drosophila oocyte nucleus (Roth et al., 1999). Despite the differences in the components, the basic principles of axis formation seem to be similar between even these phylogenetically distant arthropods. As in spiders, cell migration plays a key role in mammalian embryonic axis formation (Beddington and Robertson, 1999; Kimura-Yoshida et al., 2005; Yamamoto et al., 2004). Recent studies in planaria show that Hh signaling functions in establishing A-P polarity in regenerating tissues (Rink et al., 2009; Yazawa et al., 2009). Studies in the simple spider model could contribute to a better understanding of common aspects of axis formation in bilateria and of early evolution of the bilaterian body plan.

We thank T. Tabata for antibody, K. Agata and H. Tarui for EST clones, M. Kanayama for Delta RNAi samples, A. Noda for technical assistance and K. Agata for critical reading of the manuscript. This work was partly supported by JSPS and MEXT KAKENHI to Y.A. and H.O. Y.A. is a JSPS Research Fellow.