brachyenteron is necessary for morphogenesis of the posterior gut but not for anteroposterior axial elongation from the posterior growth zone in the intermediate-germband cricket Gryllus bimaculatus

Molecular mechanisms of anteroposterior (AP) patterning and segmentation are best understood in Drosophila melanogaster, and this knowledge provides a basis for investigating the evolution of these processes in the phylum Arthropoda. In the long-germband insect Drosophila, all body segments and posterior terminal structures, including the posterior gut and anal pads, are specified almost simultaneously at the blastoderm stage. This is in contrast to short- and intermediate-germband embryogenesis, which is thought to be a primitive feature of arthropods (reviewed by Davis and Patel, 2002). In short- and intermediate-germband insects, anterior segments are specified almost simultaneously at the blastoderm stage, whereas posterior segments are sequentially produced from the posterior growth zone, and invagination of the posterior gut starts after AP axial elongation from the growth zone. Thus,because this major difference in the developmental process of posterior patterning is observed among insects, changes in the mechanisms underlying this process are assumed to be key events in the evolutionary transition from short- and intermediate- to long-germband embryogenesis. However, we still do not know how the transition occurred at the molecular level, because the posterior patterning mechanisms in short- and intermediate-germband insects remain poorly understood.

Recent functional studies using RNA interference (RNAi) in short- and intermediate-germband insects have demonstrated that several transcription factors, including Caudal (Cad), Even-skipped (Eve) and Hunchback (Hb), and cellular signaling pathways, including Wingless (Wg)/Armadillo (Arm) and Torso signaling, are involved in elongation and/or segmentation from the posterior growth zone (Copf et al., 2004; Liu and Kaufman, 2004; Liu and Kaufman, 2005; Mito et al., 2005; Miyawaki et al., 2004; Schoppmeier and Schröder,2005; Shinmyo et al.,2005). This has led to two tentative conclusions regarding the evolution of genetic mechanisms directing posterior patterning. First, because a number of homologs of these factors are involved in posterior terminal patterning in Drosophila, the terminal system found in Drosophila may be involved in AP axial specification from the growth zone. Second, because homologs of some of these factors are also involved in AP axial elongation from the primitive streak and tail bud in vertebrates,there may be common mechanisms for AP axial formation between arthropods and vertebrates. These hypotheses prompted us to investigate the role of the Brachyury/brachyenteron (Bra/byn) gene, which is involved in morphogenesis of the posterior gut in Drosophila and AP axial elongation in vertebrates, in short- and intermediate-germband insects.

Bra/byn is the best-characterized T-box gene and has been isolated from many organisms. In vertebrates, Bra is expressed transiently around the blastopore during gastrulation, in the involuting mesoderm and endoderm, and subsequently its expression becomes restricted to the notochord and tailbud (Herrmann, 1991; Kispert and Herrmann, 1994; Kispert et al., 1995; Schulte-Merker et al., 1992; Smith et al., 1991; Wilkinson et al., 1990). Mouse(Wilson and Beddington, 1997; Wilson et al., 1995) and zebrafish (Melby et al., 1996; Schulte-Merker et al., 1994)mutants demonstrate that Bra is necessary for gastrulation, axial specification and caudal morphogenesis. Within arthropods, bynexpression and function have been most extensively investigated in D. melanogaster. Drosophila byn (Dm'byn) is expressed in the posterior terminal region from 0 to ∼20% egg length at the blastoderm stage, where the primordia of the posterior gut and anal pads are located, and continues to be expressed in the hindgut and anal pads throughout embryogenesis (Kispert et al.,1994). In Dm'byn mutant embryos, programmed cell death occurs in primordia of the hindgut and anal pads, resulting in a severe reduction of their structures (Kispert et al., 1994; Singer et al.,1996). Additionally, Dm'byn is known to be involved in the formation of the midgut constrictions, elongation of the Malpighian tubules and specification of the visceral mesoderm in Drosophila embryos (Kusch and Reuter, 1999; Singer et al.,1996). Conserved expression patterns of Byn were reported in the short-germband insects Locusta migratoria and Tribolium castaneum using the anti-TN1-123 antibody that binds specifically to the Byn protein in Drosophila embryos(Kispert et al., 1994). However, the precise expression patterns and functions of byn have not been investigated in short- and intermediate-germband insects.

We have examined byn in the intermediate-germband cricket Gryllus bimaculatus. Gryllus byn (Gb'byn) is expressed in the posterior terminal cells of the embryo during AP axial elongation, and continues to be expressed in the hindgut during late embryogenesis. Reduction of the Gb'byn expression level by RNAi resulted in defects in the posterior gut, but not in the posterior body segments. These results indicate that Gb'byn is not required for AP axial elongation or normal segment formation, but is required for morphogenesis of the posterior gut. We also examined the function of Gryllus caudal (Gb'cad) in posterior patterning by RNAi, and found that Gb'cad is required for internalization of the posterior gut primordium, in addition to AP axial elongation. We compare cad and byn function in Gryllus with their function in other bilaterians, and discuss the evolution of Cdx/cadand Bra/byn function in other animals.

The two-spotted cricket Gryllus bimaculatus was reared at 28-30°C and 70% humidity under a 10-hour light, 14-hour dark photoperiod,as previously described (Niwa et al.,2000). Fertilized eggs were collected with wet kitchen towel and incubated at 28°C in a plastic dish.

Total RNA was extracted from G. bimaculatus at various embryonic stages using Isogen (Nippon-Gene). mRNA was isolated using an Oligotex™-dT30 Super mRNA Purification Kit (TaKaRa). cDNA was synthesized using the Superscript First Strand Synthesis Kit (Invitrogen). To isolate a Gb'byn cDNA fragment by PCR, we used four primers(see Fig. 1 for primer position). The sequences of these guessmers were:

byn-5′A, 5′-ACNAAYGARATGATHGTNAC-3′;

byn-5′B, 5′-GAYCCNRMNGCNATGTAYAC-3′;

byn-3′A, 5′-AANGGRTTRTAYTTDATYTT-3′; and

byn-3′B, 5′-TCRTTYTGGTANGCNGTNAC-3′.

From the short fragment sequence obtained from the degenerate PCRs, we designed gene-specific primers and performed 5′ and 3′ rapid amplification of cDNA ends (RACE) using the SMART RACE cDNA Amplification Kit(Clontech). The following primers were used:

primary 3′RACE PCR, 5′-AACGCC GTCTACGTGCACCCCGAG-3′;

nested 3′RACE PCR, 5′-CAACGGACAGATAATGCTTAACTC-3′;

primary 5′RACE PCR, 5′-GAGTTAAGCATTATCTGTCCGTTG-3′;and

nested 5′RACE PCR, 5′-CCACTCGCCGTTCACGTACTTCCA-3′.

The Gb'byn cDNA sequence has been deposited in the DNA Data Bank of Japan (DDBJ) under Accession Number AB246318.

Embryo fixation and in situ hybridization with a digoxigenin (DIG)-labeled antisense RNA probe were performed as previously described(Niwa et al., 2000; Zhang et al., 2005). The dsRNA used in parental and embryonic RNAi was in vitro transcribed from PCR fragments amplified using primers containing T7 phage promoter sequences. Sense and antisense RNA were synthesized using the MEGAscript Kit (Ambion). The RNA was denatured in boiling water and then annealed at room temperature overnight. The resulting dsRNA was ethanol precipitated and then resuspended in water at a final concentration of 20 μM for Gb'byn(309 bp), Gb'cad (426 bp) and DsRed2 (660 bp). In all RNAi experiments, DsRed2 dsRNA was used as a negative control(Miyawaki et al., 2004). Injections for embryonic and parental RNAi were performed as previously described (Zhang et al., 2002; Mito et al., 2005). For parental RNAi, the injected females were mated with untreated males, and eggs were collected 5-10 days after injection.

In order to clone Gb'byn, we designed degenerate primers to the conserved T domain of byn homologs from other organisms. We then performed RT-PCR using these primers and isolated a short Gb'byn clone. This fragment allowed us to design specific primers for 5′ and 3′ RACE and, thereby, to isolate fragments of the gene. The Gb'byn gene is predicted to encode 357 amino acids, with a highly conserved T domain. The alignment of T domains of Gb'Byn and Bra/Byn proteins of other animals is shown in Fig. 1.

Embryogenesis in Gryllus has been described previously(Miyawaki et al., 2004; Niwa et al., 1997; Zhang et al., 2005). Briefly,in terms of segmentation, the germ anlage is formed in the ventral side of the posterior quarter of the egg by stage 3.0. Anterior segmentation occurs almost simultaneously by stage 4.0, at least to the level of the segment polarity genes, because Gryllus wingless (Gb'wg) is expressed in five vertical stripes corresponding to the mandibular through second thoracic segments. Then, the remaining posterior segments are sequentially produced through germband elongation from the posterior growth zone. The specification of the posterior segments can be tracked by the appearance of Gb'wg stripes, which appear one by one in the third thoracic segment at stage 4.3, and then in abdominal segment 1 at stage 4.4. At stage 7.5, the posterior-most stripe appears in abdominal segment 10.

We observed the expression pattern of Gb'byn during early embryogenesis by whole-mount in situ hybridization(Fig. 2). We were unable to detect any Gb'byn expression prior to stage 3.0 (data not shown). Gb'byn transcripts were first detected as two spots in the posterior terminal region of the embryo at stage 3.8(Fig. 2A). Gb'byn expression appeared more strongly at stage 4.3(Fig. 2B). During germband elongation, Gb'byn continued to be expressed in the terminal region of the embryo within the ectoderm, where the hindgut primordium is presumably located (Fig. 2C-E). The Gb'byn-expressing cells of the terminal region started to sink inward just after the completion of germband elongation and segmentation (Fig. 2F,G).

Next, we observed the expression pattern of Gb'byn during late embryogenesis, and compared this with the expression of Gb'cad (Fig. 3). At stage 8, when the proctodeum has extended and is positioned parallel to the body axis, Gb'byn was expressed in the developing hindgut (Fig. 3A,B). This Gb'byn expression was maintained during late embryogenesis (Fig. 3C,D). There was no expression of Gb'byn in the mesoderm surrounding the hindgut, posterior midgut and Malpighian tubules.

We have previously reported the expression pattern of Gb'cad during early embryogenesis(Shinmyo et al., 2005). Here,we focused on its expression during late embryogenesis. At stage 9, Gb'cad was expressed in a region adjacent to the hindgut(Fig. 3E). At stage 11, the expression domain was subdivided into two, corresponding to the Malpighian tubules and developing posterior midgut(Fig. 3F). In addition, Gb'cad was expressed throughout late embryogenesis in the cerci and in the region surrounding the orifice of the hindgut(Fig. 3E,F). Double staining indicated that the spatial patterns of Gb'cad and Gb'byn expression were almost complementary to one another in the posterior gut, i.e. in the posterior midgut, Malpighian tubules and hindgut (Fig. 3G,H).

To examine the function of Gb'byn during Gryllusembryogenesis, we used RNAi to deplete Gb'byn transcripts and produce knockdown phenocopies. Two RNAi methods have been established in Gryllus: embryonic RNAi (eRNAi)(Miyawaki et al., 2004), which involves microinjection of dsRNA into the early eggs; and parental RNAi(pRNAi) (Mito et al., 2005),which involves injection of dsRNA into adult virgin females to yield knockdown phenocopies. We confirmed that no qualitative phenocopy differences were produced when using eRNAi or pRNAi, and mainly used pRNAi for our analyses because it does not produce any injection artefacts. As with the wild type,eggs from the Gb'byn RNAi-injected females developed and hatched nymphs 12-13 days after egg laying. No obvious difference was observed in the cuticle patterns of wild-type and Gb'byn RNAi nymphs(Fig. 4A,B). However, most Gb'byn RNAi nymphs (95%, n=118 out of 124)exhibited inhibited growth in the first instar and died before reaching the second instar. To investigate the effects of Gb'byndepletion on gut formation, we compared the morphology of the alimentary canal in first-instar Gb'byn RNAi nymphs(Fig. 4D,E) with that of the wild type (Fig. 4C). The alimentary canal of the wild-type nymph consists of the foregut, including crop and proventriculus, the midgut, including gastric caecum and Malpighian tubules, and the hindgut, including the small and large intestines and rectum sac (Fig. 4C). The majority of Gb'byn RNAi nymphs exhibited severe morphological defects in the posterior gut (95%, n=38 out of 40; Fig. 4D,E), whereas the crop and proventriculus in the foregut, and the anterior region of the midgut,including the gastric caecum, seemed to be formed normally. All affected nymphs shared severe defects in the hindgut and posterior region of the midgut, whereas the severity of the disruption of the Malpighian tubules varied. In most of the affected Gb'byn RNAi nymphs (74%, n=28 out of 38; Fig. 4D), the tubules were much shorter than those of the wild type(Fig. 4C). In the most strongly affected nymphs, the tubules were virtually absent (26%, n=10 out of 38; Fig. 4E). Thus, Gb'byn RNAi nymphs that lack the posterior gut seem unable to absorb food, or to resorb water and ions, resulting in starvation in the first instar. These results indicate that Gb'byn is required for the formation of the posterior gut, but not for posterior elongation and segmentation.

To further investigate the effects of Gb'byn depletion on posterior terminal patterning, we examined the expression patterns of Gb'wg, Gryllus hedgehog (Gb'hh) and Gb'cad during Gb'byn RNAi embryogenesis. First, we confirmed that Gb'byn expression in the terminal region was reduced in the Gb'byn RNAi embryos at stage 5.2(Fig. 5A,B). In late stages, Gb'byn expression in the hindgut(Fig. 5C) provides a useful marker for characterizing defects. In Gb'byn RNAi embryos,the Gb'byn expression domain was greatly reduced in the hindgut (92%, n=23 out of 25; Fig. 5D). Although this indicated a severe reduction of the hindgut,there was still a hindgut remnant that expressed Gb'byn in all Gb'byn RNAi embryos. This suggested that the hindgut primordium invaginated normally in Gb'byn RNAi embryos, but that subsequent development of the hindgut did not occur normally, thereby implying a requirement for Gb'byn in hindgut development post-invagination. However, we cannot rule out the possibility that Gb'byn is required for the hindgut invagination itself,because almost all Gb'byn RNAi embryos might show hypomorphic phenocopies, as judged by the fact that weak Gb'byn expression was detected in almost all Gb'byn RNAi embryos (Fig. 5B). In this case, hindgut development after invagination would be more sensitive to Gb'byn reduction than development before invagination.

Although Gb'hh expression is observed in the terminal region during germband elongation(Miyawaki et al., 2004),overlapping with Gb'byn expression, Gb'hhexpression patterns were unaffected in the Gb'byn RNAi embryos (data not shown). During invagination of the proctodeum in wild-type embryos, Gb'hh is expressed in the developing hindgut(Inoue et al., 2002). At stage 11-12, the expression domain became subdivided into three regions: strong expression in the small intestine and rectum sac, and weak expression in the large intestine (Fig. 5E)(Inoue et al., 2002). In the Gb'byn RNAi embryos, abnormal expression of Gb'hh was observed in the hindgut remnant, probably as a combined pattern of the small intestine and rectum sac expression domains,with reductions in both (100%, n=25; Fig. 5F). This indicates a dramatic defect in the large intestine. In addition, Gb'hhexpression in the Malpighian tubules was also disrupted in the Gb'byn RNAi embryos (Fig. 5F).

Gb'wg is expressed in the posterior growth zone during germband elongation (Miyawaki et al.,2004). This expression pattern was unaffected in the Gb'byn RNAi embryos (data not shown). In the wild-type embryos at stage 11-12, Gb'wg expression was detected in two regions, the anterior region of the small intestine and the posterior rectum of the hindgut (Fig. 5G)(Inoue et al., 2002). In the Gb'byn RNAi embryos, Gb'wg expression was detected in both anterior and posterior regions of the severely reduced hindgut, with reduced expression domains (92%, n=23 out of 25; Fig. 5H). This result indicates a dramatic defect in the large intestine of Gb'byn RNAi embryos, as well as relatively mild defects in the small intestine and rectum,consistent with the pattern of Gb'hh expression in the Gb'byn RNAi embryos.

Gb'cad is expressed in the posterior growth zone during germband elongation (Shinmyo et al.,2005). This expression pattern was unaffected in Gb'byn RNAi embryos (data not shown). In wild-type embryos at stage 9, Gb'cad was expressed in the region surrounding the orifice of the hindgut and in the region adjacent to the hindgut(Fig. 3E, Fig. 5I). Gb'cad expression in both domains was greatly reduced in the Gb'byn RNAi embryos (100%, n=10; Fig. 5J). Additional domains of Gb'cad expression in the cerci were not affected in the Gb'byn RNAi embryos (Fig. 5, compare I with J). At stages 11-12, Gb'cadexpression was detected in the region surrounding the orifice of the hindgut,the Malpighian tubules and the posterior midgut in wild-type embryos(Fig. 3F, Fig. 5K). Gb'cad expression in the region surrounding the orifice of the hindgut and posterior midgut was greatly reduced in the Gb'byn RNAi embryos (88%, n=22 out of 25; Fig. 5L), indicating a reduction in these structures. We also found that Gb'cad was weakly expressed in the very small remnant of the Malpighian tubules seen in all Gb'byn RNAi embryos (100%, n=25; Fig. 5L), indicating that the primordium of the Malpighian tubules was formed in Gb'bynRNAi embryos. This suggests that the disruption of the Malpighian tubules observed in the Gb'byn RNAi nymphs(Fig. 4D,E) resulted from an inhibition of tubule elongation. This interpretation is supported by the fact that the shortened Malpighian tubules were formed in most Gb'byn RNAi nymphs (Fig. 4D). Thus, in Gb'byn RNAi embryos, the expression patterns of the marker genes for the posterior gut suggest that Gb'byn is necessary for differentiation of the posterior midgut and hindgut, and for elongation of the Malpighian tubules.

In Drosophila, cad is essential for invagination and maintenance of the hindgut primordium (Wu and Lengyel,1998). Although it has been shown that cad is required for the formation of all trunk segments in short- and intermediate-germband insects (Copf et al., 2004; Shinmyo et al., 2005), the role of cad in posterior terminal patterning in these insects has not been investigated. To determine Gb'cad function in Gryllus, we generated embryos depleted of Gb'cad by pRNAi, and examined the expression patterns of Gb'byn during Gb'cad RNAi embryogenesis. First, we confirmed that most Gb'cad RNAi embryos obtained by pRNAi exhibited severe defects in the trunk segments (Fig. 6A-C), as described previously in Gb'cad eRNAi experiments (Shinmyo et al.,2005). In wild-type embryos at stage 4, Gb'bynwas expressed in the posterior terminal region(Fig. 6D), and this remained unaffected in Gb'cad RNAi embryos (100%, n=10; Fig. 6E). This suggests that Gb'cad is not involved in establishing the hindgut primordium. In wild-type embryos, Gb'byn-expressing cells in the terminal region sink inwards at stage 7.5 and are completely internalized by stage 9 (Fig. 3C, Fig. 6F). In the Gb'cad RNAi embryos, the Gb'byn-expressing cells failed to invaginate at stage 9, remaining on the outside of the embryo(Fig. 6G,H).

We also examined the expression pattern of Gb'hh, which is also used as a marker gene for the hindgut(Fig. 6I). In the Gb'cad RNAi embryos, Gb'hh expression was observed in the external hindgut remnant(Fig. 6J,K). These observations indicate that the invagination of the hindgut primordium did not occur in the Gb'cad RNAi embryos, suggesting that Gb'cad is not necessary to establish the hindgut primordium,but is required for internalization of the primordium. However, we cannot rule out the possibility that the hindgut primordium, in which Gb'byn is normally expressed, is not correctly specified in the Gb'cad RNAi embryos. In this case, Gb'cad and Gb'byn would be activated independently in the hindgut primordium, and both genes would be necessary for the establishment of the hindgut primordium. Further expression analyses of hindgut markers in Gb'cad RNAi embryos will be required to determine Gb'cad function in specification of the hindgut primordium.

We have isolated the Gb'byn gene from the intermediate-germband cricket Gryllus bimaculatus and investigated its developmental function using RNAi. We found that Gb'bynis not required for AP axial elongation or normal segment formation, but is required for the specification of the posterior gut. We also found that Gb'cad is required for internalization of the hindgut primordium, in addition to AP axial elongation. Here, we discuss the functions of Gb'cad and Gb'byn in Gryllusembryogenesis, and compare them with their functions in other bilaterians.

In Drosophila, all segments and posterior terminal structures are specified by the blastoderm stage. By contrast, in Gryllus, posterior segments are specified in an anterior to posterior direction through elongation of the posterior growth zone, and invagination of the posterior gut starts after the specification of the posterior segments. It is not clear how and when the posterior segments and terminal structures are specified, or how developmental timing of their structures is controlled during Gryllusembryogenesis. We found that Gb'byn is expressed as two spots in the posterior terminal cells of embryos, where the hindgut primordium is presumably located, before AP axial elongation(Fig. 2A). This suggests that the specification of the hindgut primordium occurs before AP axial elongation,and independently of posterior segment specification. This interpretation is supported by our previous data indicating that the terminal structures, such as the hindgut and cerci, are formed in Gb'hb or Gb'krüppel RNAi embryos, in which posterior segments generated from the growth zone are severely defective(Mito et al., 2005; Mito et al., 2006). Further support comes from studies showing that in Gryllus embryos subjected to lethal doses of radiation, the most posterior segment carrying the cerci is always present, even if many other segments are missing(Sander, 1976).

The interpretation may also be supported by the observation that Gb'byn expression appears normal in early Gb'cad RNAi embryos, which presumably lack all trunk segments at late stages (Fig. 6E).

It is important to note that tailless, which acts upstream of byn in Drosophila terminal patterning, is already expressed at the blastoderm stage at the posterior pole of Tribolium embryos. This suggests that there is a group of cells within the posterior growth zone that is determined at the blastoderm stage to produce the terminal structures in Tribolium (Schröder et al., 2000). This conceivably might also apply to Gryllusembryogenesis.

We found that the posterior gut, consisting of the posterior midgut,Malpighian tubules and hindgut, was severely reduced in Gb'byn RNAi nymphs (Fig. 4, Fig. 7A). Furthermore, detailed analysis of the expression patterns of tissue-specific markers revealed that Gb'byn is necessary for differentiation of the midgut and hindgut, and for elongation of the Malpighian tubules (Fig. 5, Fig. 7A). In Drosophila byn mutants, the posterior gut is severely reduced as a consequence of massive apoptosis in the gut primordia(Kispert et al., 1994; Singer et al., 1996). It remains unclear whether apoptosis contributes to the reduced posterior gut in Gb'byn RNAi embryos because of a technical problem associated with the TUNEL staining. However, the similarities in phenotype suggest that byn function during embryogenesis is highly conserved between long- and intermediate-germband insects. Bra is not reported to be involved in gut formation in vertebrates, but it is expressed in the posterior gut endoderm of hemichordates(Peterson et al., 1999) and echinoderms (Gross and McClay,2001; Shoguchi et al.,1999). Although the posterior gut endoderm of these animals is substantially different from the hindgut ectoderm of insects, these similarities suggest that the involvement of Bra/byn in specification of the posterior gut might be ancestral to bilaterians. A similar presumption might also extend to the role of Cdx/cad in gut development. In Gryllus, Gb'cad is expressed in the Malpighian tubules and posterior midgut endoderm during late embryogenesis(Fig. 2). Our RNAi analysis shows that Gb'cad is necessary for internalization of the posterior gut primordium (Fig. 6). In Drosophila, Dm'cad is known to be expressed in the Malpighian tubules and posterior midgut endoderm of older embryos (Macdonald and Struhl,1986; Mlodzik et al.,1985), and to be essential for internalization and maintenance of the posterior gut primordium (Wu and Lengyel, 1998). Thus, the expression pattern and function of cad in posterior gut development are highly conserved between Gryllus and Drosophila. In vertebrates, Cdx genes are expressed in the gut endoderm during late embryogenesis (reviewed by Freund et al., 1998), and Cdx2 mutant mice develop intestinal tumors(Chawengsaksophak et al.,1997). In Caenorhabditis elegans, the cadhomolog pal-1 is expressed zygotically in mesoderm cells of the posterior gut (Edgar et al.,2001). On the basis of these data, we hypothesize that the involvement of Cdx/cad and Bra/byn in the specification of the posterior gut might be an ancestral feature of bilaterians.

It should be noted that morphogenesis of the Malpighian tubules and posterior midgut is blocked in Gb'byn RNAi nymphs, and that Gb'cad expression in these tissues is reduced in Gb'byn RNAi embryos, even though Gb'bynexpression is not detected in these tissues in Gryllus embryos. There are two possible explanations for this phenomenon. First, it might be that very low levels of Gb'byn expression, which are below detectable levels, are sufficient for the development of these tissues. In this case, Gb'byn would be required for the formation of the Malpighian tubules and posterior midgut through the direct or indirect regulation of Gb'cad expression in these structures. Second,morphogenesis of the Malpighian tubules and posterior midgut depends upon signaling from the contiguous hindgut, where Gb'byn is expressed. In this case, the reduction in Gb'cad expression in the Malpighian tubules and posterior midgut of the Gb'bynRNAi embryos would result from the inhibition of hindgut development. A similar phenomenon is also observed in Drosophila byn mutants(Singer et al., 1996),suggesting conservation in the mechanisms of terminal patterning.

The progressive growth of AP axial structures from a posterior region is observed in such diverse animals as chordates, short- and intermediate-germband arthropods, annelids and molluscs. In short- and intermediate-germband arthropods, posterior segments are sequentially produced from the posterior growth zone, where cad is expressed(Copf et al., 2003; Dearden and Akam, 2001; Schulz et al., 1998; Shinmyo et al., 2005)(Fig. 7B) and required for AP axial elongation from the growth zone (Copf et al., 2004; Shinmyo et al.,2005). cad expression in the growth zone is likely to be regulated by Wg/Arm signaling in Gryllus embryos(Shinmyo et al., 2005)(Fig. 7B). Segmentation in short- and intermediate-germband arthropods resembles somitogenesis in vertebrates, in which somites are generated progressively from a posteriorly located presomitic zone (reviewed by Peel et al., 2005). In addition, the Cdx genes, which are regulated by Wnt signaling, are expressed in the nascent mesoderm of the primitive streak (Ikeya and Takada,2001; Marom et al.,1997; Meyer and Gruss,1993) (Fig. 7B),and are involved in axial elongation and somitogenesis(Epstein et al., 1997; Subramanian et al., 1995; van den Akker et al., 2002). These similarities suggest that the molecular mechanisms underlying these processes are conserved between short- and intermediate-germband arthropods and vertebrates. Recently, it has been shown that even-skipped(eve) is expressed in the posterior growth zone and is required for AP axial elongation in the intermediate-germband insect Oncopeltus fasciatus (Liu and Kaufman,2005). This fact may also suggest conserved mechanisms for these processes because, in vertebrates, Evx1 (the eve homolog) is known to be expressed in the primitive streak and tail bud, although its function has not been investigated (Dush and Martin, 1992). These data suggest that AP axial formation from the posterior growth zone is ancestral to bilaterians. A similar hypothesis has been proposed, based on a comparison of Bra expression patterns in molluscs and vertebrates. In vertebrates, Bra is also expressed in the nascent mesoderm of the primitive streak and tail bud(Kispert and Herrmann, 1994; Knezevic et al., 1997; Wilkinson et al., 1990)(Fig. 7B), and is necessary for AP axial formation (Wilson and Beddington,1997). Because Bra expression in the posterior pole of the AP axis, up to the end of mollusc larval development, is similar to that in vertebrates, Lartillot et al.(Lartillot et al., 2002) have proposed that Bra might have a conserved role in the regulation of AP patterning among bilaterians, through maintenance of the posterior growth zone. This hypothesis implies that the role of Bra/byn in AP axial elongation might be ancestral to bilaterians. Importantly, we found that Gb'byn is expressed exclusively in the posterior terminal region(Fig. 7B), and is not involved in AP axial elongation from the growth zone. Therefore, if the hypothesis is correct, our results suggest that the function of Bra/byn in AP axial elongation might have been lost in insects. More data from a wider range of protostomes will be required to confirm this.

We thank T. Matsushita, H. Okamoto and T. Watanabe for their technical assistance. We also thank two anonymous reviewers for suggestions that improved the manuscript. 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.