Non-canonical functions of hunchback in segment patterning of the intermediate germ cricket Gryllus bimaculatus

The major features of the body plan of insects include segmentation and the organization of segments into three distinct regions: head, thorax and abdomen. Although almost all insects have segmented bodies in common, their patterning processes vary substantially from insect to insect. In long germ insects, such as Drosophila, the pattern of all segments is laid out simultaneously in a syncytial environment. In the cellularized embryos of short and intermediate germ insects, the anterior segments form almost simultaneously, and the remaining posterior segmentation occurs sequentially from a posterior growth (proliferative) zone after anterior segmentation is complete (Dearden and Akam,2001). This mode of segmentation is believed to be ancestral(Davis and Patel, 2002; Tautz et al., 1994).

To understand the similarities and differences in the molecular mechanisms of short/intermediate and long germ segmentation, we need to compare the roles of the developmental genes involved in early embryonic patterning. In Drosophila, gap genes play a central role in the early subdivision of the blastoderm into broad regions. Maternal gradients lead to the activation of gap genes a broad domain, and the syncytial environment allows the gap-gene products to diffuse and produce overlapping short-range gradients(Hülskamp and Tautz,1991; Rivera-Pomar and Jäckle, 1996). These short-range gradients define the stripe patterns of the primary pair-rule genes(Small and Levine, 1991). Gap genes are also responsible for providing positional information to regulate the expression of Hox genes, which assign identities to each segment(McGinnis and Krumlauf, 1992). Although gap genes act in a syncytial environment in Drosophila, the orthologs of gap genes in short and intermediate germ insects are expressed in a cellularized environment. The orthologs of the gap genes of these insects cannot act in the same manner as those in Drosophila. Thus,elucidating the functions of the gap genes in short and intermediate germ insects would provide crucial clues to clarify the molecular segmentation mechanisms.

The gap gene hunchback (hb), which codes for a zinc-finger type transcription factor, is crucial for anteroposterior patterning in various insects (Liu and Kaufman, 2004; Lehmann and Nüsslein-Volhard, 1987; Patel et al., 2001; Schröder, 2003; Tautz et al., 1987). For example, in Drosophila, the loss-of-function alleles for hbshow a canonical gap defect: i.e. deletion of the labial through the metathoracic segments. The hb RNAi depletion in Tribolium, a short germ insect, results in deletion of the gnathal and thoracic segments,suggesting that the canonical gap function of hb is conserved(Schröder, 2003). On the other hand, in the intermediate germ insect Oncopeltus, the hb (Of'hb) RNAi depletion results in a non-canonical,gap-like phenotype, i.e. a combination of the transformation of the gnathal and thoracic regions into an abdominal identity and defects in posterior elongation and segmentation (Liu and Kaufman, 2004). This indicates that Of'hb is required to suppress the abdominal identity and for proper germband growth and segmentation (Liu and Kaufman,2004). These results suggest that hb function differs among insects.

To deepen our understanding of the mechanisms of insect segmentation by obtaining data from more phylogenetically basal species, we have focused on the intermediate germ cricket, Gryllus bimaculatus (Gb,Orthoptera), which is more basal than Oncopeltus (Hemiptera). In our study, we isolated Gryllus hb (Gb'hb) and analyzed its functions, using embryonic and parental RNA interference, and found the following: (1) Gb'hb regulates the Hox gene expression to specify a regional identity in the anterior region, as observed in Drosophilaand Oncopeltus; (2) Gb'hb controls germband morphogenesis and segmentation of the anterior region probably through the pair-rule gene, even-skipped at least; (3) Gb'hb may act as a gap gene in a limited region between the posterior of the prothoracic segment and the anterior of the mesothoracic segment; and (4) Gb'hb is involved in forming at least seven abdominal segments, probably through its expression in the posterior growth zone, which is not conserved in Drosophila. These findings suggest that Gb'hb functions in a non-canonical manner in segment patterning. We will discuss both conserved and divergent aspects of Gb'hb functions from an evolutionary point of view.

Total RNA was extracted from embryos within 24-72 hours after egg-laying(hAEL) using Isogen (Nippon-Gene). To clone Gb'hb, we first applied a degenerate polymerase chain reaction (PCR). cDNA was synthesized using the Superscript First Strand Synthesis Kit (Invitrogen) with random hexamers. For the PCR, we used hb degenerate primers that targeted GFVAVTK(forward) and GTRRGPK (reverse) motifs in the region coding the central zinc-finger domains. The nucleotide sequences of the primers were 5′-GGC TTC GTB GCN GTN ACN AA-3′ and 5′-TTA GGA CCA CKN CKN GTN CC-3′. To isolate of the Gryllus Krüppel fragment, we used the degenerate primers described by Sommer et al.(Sommer et al., 1992). Using the short-fragments sequences 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 Gb'hb cDNA sequence was deposited in the DNA Data Bank of Japan(DDBJ) (Accession Number, AB120735).

Standard protocols were used for whole-mount in situ hybridization with a digoxigenin (DIG)-labeled antisense RNA probe, as previously described(Niwa et al., 2000). In situ hybridization for double staining was carried out as follows. The antisense RNA probe was labeled with DIG or fluorescein. Hybridization was carried out following the standard protocol. After hybridization, anti-fluorescein-AP Fab fragments (Roche) were added and a color reaction for the fluorescein-labeled probe was performed using NBT/BCIP as the substrate. The samples were then washed in a TNT buffer, before re-fixing using 4% paraformaldehyde in PBS at 4°C overnight. After subsequent washing in TNT, anti-digoxigenin-POD Fab fragments (Roche) were added and then the samples were washed in TNT again. Treatment for signal enhancement was then carried out using the TSA Biotin System (PerkinElmer Life Sciences) following the manufacturer's instructions. The color reaction for the DIG-labeled probe was performed using the Vector Nova Red substrate kit (Vector).

We synthesized double-stranded RNA (dsRNA) using the MEGA-script Kit(Ambion), and used PCR fragments as the template for in vitro transcription. The PCR fragments were amplified using upstream and downstream primers that contained the T7 promoter sequence. The synthesized RNA was extracted using phenol/chloroform and ethanol precipitated. The RNA was denatured in boiled water after being suspended in a Tris-EDTA buffer and annealed at room temperature overnight. The resulting dsRNA was suspended in an appropriate volume of water after ethanol precipitation. The final concentration of dsRNA was adjusted to 20 μM for the Gb'hb dsRNA (395 bp, spanning the four central zinc fingers) and the DsRed2 dsRNA [660 bp, derived from the pDsRed2-N1 (Clontech)]. The DsRed2 dsRNA was used for negative control experiments. For embryonic RNAi, the cricket eggs were collected for 2 hours and used within 1 hour of collection. We microinjected the dsRNA in the posterior end of the egg, as previously described(Zhang et al., 2002). For parental RNAi, we injected adult females with a 0.6 μl dsRNA solution in a ventrolateral position, between segments T3 and A1 (details to be published elsewhere). Fifteen injected females were mated with untreated males, and the eggs were collected from 5 to 10 days after injection.

We isolated cDNA clones of Gb'hb from embryos at 24-72 hAEL using the degenerate PCR and 5′ and 3′ RACE strategy. Nucleotide sequences of three independent clones were analyzed in each RACE experiment and they showed identical structure each other. The Gb'hb cDNA is predicted to encode 572 amino acids. The alignments of conserved domains of Gb'Hb compared with those of other insects are shown in Fig. 1A. Gb'Hb contains six zinc-finger domains, four in the middle region (MF 1-4) and two in the C-terminal region (CF 1-2), which correspond to the MF 1-4 and the CF 1-2 of Drosophila Hb, respectively. In addition, sequences similar to the A-box and C-box of Drosophila Hb(Tautz et al., 1987; Hülskamp et al., 1994)are conserved in Gb'Hb. The overall structure of the Gb'Hb protein is shown in Fig. 1B. In our PCR clone, we did not find the basic box(Hülskamp et al., 1994)or the two N-terminal finger domains found in C. elegans, Oncopeltusand grasshopper Hb (Liu and Kaufman,2004; Patel et al.,2001). From a phylogenetic perspective, there could be an isoform containing these domains in Gryllus.

The embryogenesis of Gryllus begins with mitoses, without cellular divisions. The nuclei then migrate to the egg periphery, followed by cellularization (Sarashina et al.,2005). During the blastoderm stage, the germ anlage forms in the posteroventral region of the egg around 30 hAEL. At 30 hAEL, Gb'hbwas expressed weakly as a band in the middle of the embryo and also expressed in a stripe along the midline, probably in the invaginating mesoderm(Fig. 2A). As Gryllus wingless (Gb'wg) is expressed in each segment(Miyawaki et al., 2004), we used it as a segment-marker gene. The middle Gb'hb domain was located between the anterior and posterior expression domains of Gb'wg(Fig. 2B). The middle Gb'hb domain, corresponding to the prospective gnathal region, then incompletely resolved into two bands at 32 hAEL(Fig. 2C), and the anterior band was further resolved into two stripes at 36 hAEL(Fig. 2D). Double staining for Gb'hb and Gb'wg at 40 hAEL indicated that the gap-like middle domain of Gb'hb expression spanned from the prospective mandibular to labial segments (Fig. 2E). By 42 hAEL, the expression domain in the gnathal region became faint and eventually faded out (Fig. 2F).

In the later stages, Gb'hb was expressed as spots in the head region (Fig. 2I), and then in the trunk (Fig. 2H-J). Because we know that hb is expressed in neuroblasts as spots in Drosophila (Isshiki et al.,2001), and is probably expressed the same in Tribolium(Wolff et al., 1995) and in the grasshopper (Patel et al.,2001), each spot may correspond to a neuroblast. As in Drosophila, Gb'hb may be also involved in neural patterning.

An additional Gb'hb expression domain appeared near the posterior end of the abdomen (Fig. 2G)and became more intense as the posterior region elongated(Fig. 2H). The posterior band then split into two stripes, and a new band appeared in the posterior region(Fig. 2J). Double staining for Gb'hb and Gb'wg indicated that the two intense stripes were located between the A7- and A9-Gb'wg stripes(Fig. 2K). In Drosophila, the posterior hb expression domain is located in the region of parasegment (PS) 13/14 and is required to form the abdominal segments A7 and A8 (Tautz et al.,1987; Lehmann and Nüsslein-Volhard, 1987; Bender et al., 1987). In the grasshopper, the hb posterior expression domain spans a region from the posterior compartment of A7 to the anterior compartment of A9(Patel et al., 2001).

We detected Gb'hb transcripts in the ovary and early eggs using RT-PCR (data not shown), which indicated the presence of maternal transcripts. It is unclear, though, when the maternal transcripts are replaced by the zygotic ones.

To investigate the Gb'hb function, we applied both embryonic RNAi(eRNAi) and parental RNAi (pRNAi) to deplete the transcript(Bucher et al., 2002; Miyawaki et al., 2004). We confirmed that no qualitative phenotype differences were produced by either the eRNAi or the pRNAi. We used pRNAi for further analyses using in situ hybridization, because it does not produce injection artifacts and all of the developed embryos showed effects of RNAi by our pRNAi for Gb'hb(Table 1). We were able to categorize the resulting embryos into three phenotypic classes, based on the severity, as shown in the embryos just before hatching and in the embryos stained with a segment marker gene Gb'wg at stage 9(Fig. 3). The most severe class I embryos (Fig. 3J-M) showed a gap-like phenotype, in which the head with the mandible was followed by several segments without appendages. In the class I embryos, the antenna,mandible and cercus formed almost normally. In milder phenotypes, the class II(Fig. 3G-I) and III(Fig. 3D-F), a part of the appendages was suppressed. Of the affected appendages, the T3 leg was the most resistant to Gb'hb depletion. In the class II embryos, all gnathal and thoracic appendages were suppressed, except for vestigial legs in the T3 segment, and the embryos were shortened as a result of the reduced number of segments (Fig. 3G). Segmentation disturbances were frequently observed in all of the phenotypic classes (Fig. 3F,H,I). Such phenotypes are reminiscent of those resulting from hb depletion in another intermediate germ insect, Oncopeltus, in which the gnathal and thoracic regions are transformed into an abdominal identity, forming a small abdomen with defective segments (Liu and Kaufman, 2004). Even in the most severe class I embryos,vestiges of T3 legs were observed in 70% of the RNAi embryos (n=28 out of 40, Fig. 3L). Most of these embryos had only three or four segments posterior to the leg vestiges,indicating that seven abdominal segments had been deleted in the severe case. This number may not be maximal because it was often difficult to ascertain the precise number of deleted segments because of segmentation disturbances. The class I embryos without leg vestiges showed no crucial differences from embryos with leg vestiges in their external morphology, indicating that all of the class I embryos were resulted from suppressing the gnathal and thoracic identities and growth and segmentation defects in the posterior region.

We observed the internal structure of the Gb'hb RNAi embryos using sagittal sections. The digestive tract of the wild-type embryos consisted of the fore-, mid- and hindgut with characteristic constriction patterns(Fig. 3N). The midgut contained yolk granules. In the class II embryos, the fore-, mid- and hindgut were formed, although the fore- and hindgut were compressed in the shortened Gb'hb RNAi embryos (Fig. 3O). The class I embryos had a sac-like midgut, and the fore- and hindgut were further compressed (Fig. 3P). Our observation of the segmental musculature revealed that anterior segments of the Gb'hb RNAi embryos were associated with the segmental musculature of the abdominal character, even in severe phenotypes(Fig. 3R,S). However, these segments were wider than the abdominal segments of the wild type and were not arranged in an orderly reiterated pattern, suggesting incomplete transformation to the abdomen.

To further investigate the effect of Gb'hb depletion on segment identity, we examined the expression patterns of the Hox genes in the Gb'hb RNAi embryos. In the wild-type embryo at stage 9(Niwa et al., 1997), Gryllus abdominal-A (Gb'abdA) is expressed in the posterior compartment of A1 and the remaining abdominal segments(Fig. 4A)(Miyawaki et al., 2004; Zhang et al., 2005), thus making it an abdominal-marker gene. We observed ectopic expression of Gb'abdA in the gnathal and thoracic regions in the Gb'hbRNAi embryos (Fig. 4B,C). This supports our interpretation based on the morphological observation that suppression of appendages in the Gb'hb RNAi embryos is a result of the homeosis of the gnathal and thoracic regions towards an abdominal identity, even in the most severe phenotype(Fig. 4D). This is essentially consistent with the results of an RNAi analysis of Oncopeltus hb(Liu and Kaufman, 2004).

Transformation of the gnathal and thoracic regions should be accompanied by a change in the expression patterns of the Hox genes, other than Gb'abdA. Gryllus Sex combs reduced (Gb'Scr) is intensely expressed in the labial and T1 segments in the wild-type embryos(Fig. 4E)(Miyawaki et al., 2004; Zhang et al., 2005). In the Gb'hb RNAi embryos, the Gb'Scr expression in the labial and T1 segments was greatly reduced (Fig. 4F,G) and disappeared in the most severe case(Fig. 4H). This is concordant with suppression of the gnathal and thoracic identity in the Gb'hbRNAi embryos. Gryllus Antennapedia (Gb'Antp) is expressed in all of the thoracic and abdominal segments in the wild-type embryos(Fig. 4I)(Zhang et al., 2005). In the abdominal region, strong expression of Gb'Antp is observed along the midline. In the Gb'hb RNAi embryos, the anterior expression boundary was shifted to the labial segment in all phenotype classes(Fig. 4J-L). The Gb'Antp expression in the anterior region of the Gb'hb RNAi embryos was strong in midline, suggesting that Gb'Antp expression in abdomen-like patterns was ectopically induced in the anterior region. Gryllus Ultrabithorax (Gb'Ubx) is strongly expressed in T3 legs and A1 prolegs, and weakly in the abdominal segments(Fig. 4M)(Zhang et al., 2005). We used Gb'Ubx as a marker gene for the T3 and A1 segments in the RNAi embryos. In the Gb'hb RNAi embryos, Gb'Ubx was ectopically expressed in the anterior region, in the maxillary to T2 segments(Fig. 4N,O). In weak phenotypes, the Gb'Ubx expression in the T3 leg bud was retained(Fig. 4N), indicating that the T3 segment was the most resistant to Gb'hb RNAi depletion. In more severe cases, it was reduced (Fig. 4O). In the most severe case, Gb'Ubx was expressed from the maxillary segments towards the posterior end in a similar pattern to the abdominal segments of wild type (Fig. 4P). Thus, expression patterns of the Hox genes suggest that in the Gb'hb RNAi embryos, gnathal and thoracic identities were suppressed and transformed towards an abdominal identity.

The Gb'hb RNAi effects extend outside the expression domain of the gap pattern in the early stages. Effects outside of this expression domain may be caused through regulation of other segmentation genes and/or Hox genes during early embryogenesis. To investigate the effects of Gb'hbdepletion on early embryogenesis, we examined the expression patterns of the Gryllus orthologs of Drosophila segmentation genes(Krüppel, even-skipped and wingless) and Hox genes(Antennapedia and abdominal-A).

We isolated a fragment of the gap-gene ortholog Gryllus Krüppel (Gb'Kr) from the degenerate PCR and 5′ RACE,using cDNA from embryos within 24-72 hAEL. This fragment contained the Type 1 to 3 zinc fingers and part of the Type 4 zinc finger of the five zinc fingers(Type 1-5) in Drosophila Kr (Fig. 5) (Rosenberg et al.,1986). In the wild-type embryos at 40-42 hAEL, Gb'Kr is expressed in a broad domain in the thoracic region and spot-like domains in the gnathal region (Fig. 6A,details to be published elsewhere). In Gb'hb RNAi embryos, Gb'Kr expression was severely affected. The Gb'Kr expression domain in the thoracic region was reduced in the RNAi embryos(Fig. 6B,C). In severe cases,the thoracic broad domain was suppressed, except in the peripheral region(Fig. 6C). This suggests that Gb'hb is involved in transcriptional regulation of Gb'Kr in the thoracic region directly or indirectly. The reduction of the Gb'Kr expression domain could be partially due to the defect in embryonic growth, because the Gb'hb RNAi embryos in the early stages were shorter and wider than wild-type embryos at the same time after egg-laying. Such a morphological change in the Gb'hb RNAi embryos suggests that Gb'hb is required for proper development of the germband.

Next, we observed the expression patterns of pair-rule and segment polarity gene orthologs in Gb'hb RNAi embryos. In wild-type embryos at 38-40 hAEL, Gryllus even-skipped (Gb'eve) is expressed in the five clearly resolved stripes in the gnathal and thoracic regions, and an additional stripe or two are resolved from the broad expression domain in the posterior growth zone (Fig. 6D). The five anterior stripes first appear almost simultaneously,as the most anterior segmental stripe and two pair-rule-like broad stripes,each of which resolves into two segmental stripes (data not shown). The Gb'hb expression domain covers the gnathal region(Fig. 2E). However, we found that the first to third Gb'eve stripes in the gnathal region were not affected in the Gb'hb RNAi embryos, even though the total number of stripes was reduced at this stage (Fig. 6E,F). In the most severe case, only three stripes were observed(Fig. 6F). This suggests retardation in the timing of the sequential formation of the stripes, or suppression of the formation of segmental stripes 4 and 5 (derived from the secondary pair-rule-like stripe). Gb'wg is expressed in seven segmental stripes in the gnathal/thoracic and abdominal regions in the wild-type embryos at 45-47 hAEL (Fig. 6G). In the Gb'hb RNAi embryos, the number of stripes was reduced to five, and the shapes of thoracic stripes became irregular(Fig. 6H,I). The T1 stripe of Gb'wg was the most severely affected, indicating fusion with the T2 stripe (Fig. 6H). In severe cases, the labial stripe was also affected, while the first and second gnathal stripes appeared normal (Fig. 6I). In addition to defects in the anterior region, the shape of the elongating posterior region in the Gb'hb RNAi embryos was abnormal at this stage (Fig. 6H,I).

To see the effects of Gb'hb depletion in the homeotic segment specification in early embryos, we examined the expression patterns of the homeotic genes in the Gb'hb RNAi embryos. Gb'Antp is strongly expressed in the thoracic region of the wild-type embryos at 38-40 hAEL (Fig. 6J)(Zhang et al., 2005). The Gb'Antp domain was reduced in the Gb'hb RNAi embryos(Fig. 6K). The expression of Gb'abdA begins at the boundary between the thoracic region and the abdominal region in the wild-type embryos by 45-47 hAEL(Fig. 6L)(Zhang et al., 2005). The Gb'hb RNAi resulted in ectopic expression of Gb'abdA in the gnathal and thoracic regions (Fig. 6M). These results indicate that Gb'hb is involved in regulating the Hox genes in the gnathal and thoracic regions during early embryogenesis.

In this study, we examined the expression patterns and functions of an ortholog of the Drosophila gap gene hb in an orthopteran Gryllus bimaculatus, which phylogenetically is the most basal species of insects ever examined for hb functions. So far, the functions of insect hb genes have been studied in a dipteran Drosophila,coleopteran Tribolium and hemipteran Oncopeltus(Of'hb). Interestingly, the Of'hb functions were found to differ from the canonical functions of Drosophila or Tribolium hb (Liu and Kaufman,2004). Furthermore, in this study we found that the non-canonical functions of Gb'hb partly differ from those of Of'hb. Here,in comparison with the functions of hb in other species, we discuss the Gb'hb functions in the anterior region at first(Fig. 7A-D), the regulatory network among Gryllus segmentation genes(Fig. 7E) and the evolution of hb functions.

In the early stages of Gryllus embryos, we observed a large morphological change, in which the germ anlage becomes thinner and longer to form the germband (Miyawaki et al.,2004), probably caused by cell rearrangement. This morphological change is reminiscent of the germband extension observed in Drosophila embryogenesis. In the early stages of Gb'hb RNAi embryos, the germband was shorter and wider than in the wild type, and the Gb'wg stripes corresponding to the prospective thoracic segments became irregular (Fig. 7B), suggesting that the cell rearrangement defects occurred during the morphogenesis of the germband(Fig. 7A). In Drosophila,hb is required for proper cell rearrangement in the germband extension through regulation of expression of pair-rule genes(Irvine and Wieschaus, 1994). Therefore, this fact and our results suggest that Gb'hb may be involved in the cell rearrangement for morphogenesis in the germband.

In Drosophila hb mutants, the region between the labial and metathoracic segments is known to be deleted. In Gb'hb RNAi embryos,we were not able to find such a large deletion in the anterior region. However, precise analysis of the expression patterns of Gb'wg and Gb'eve stripes in Gb'hb RNAi embryos revealed that a narrow region was deleted. In wild-type Gryllus embryos, the Gb'wgstripes are located at the anteroposterior boundary of each segment(Miyawaki et al., 2004; Niwa et al., 2000), as shown in Fig. 7B; in the Gb'hb RNAi embryos the T1 and T2 stripes of Gb'wg were fused, indicating that the region from the posterior T1 to the anterior T2 was deleted (indicated by a bracket in Fig. 7B). This is supported by the fact that segmental stripes 4 and 5 of the Gb'eve were not observed in the Gb'hb RNAi embryos. These results indicated that Gb'hb functions as a gap gene in a limited region from the posterior T1 to the anterior T2(Fig. 7D). We can not deny that more segments would be deleted in more severe Gb'hb phenotype,because in our RNAi experiments we are unable to confirm a complete knockdown of Gb'Hb activity. However, the most severe class of the Gb'hb phenotypes that we observed is likely to result from the strongest depletion of Gb'hb, because appearance frequency of the most severe phenotype increased with injected amount of dsRNA for Gb'hb without change in the phenotype itself (data not shown). This presumption is supported by the fact that the anterior deleted region was not extended in Gb'hb RNAi embryos, when more severe effects of RNAi on abdominal regions appeared.

It is very mysterious that the segmental deletion appears in the thoracic region where Gb'hb transcripts were not detected. Similar phenomena were reported for gap gene orthologs of other short/intermediate germ insects such as Oncopeltus (Liu and Kaufman, 2004) and Tribolium(Bucher and Klingler, 2004). The following two possibilities might explain the phenomenon: one is that Gb'hb indirectly regulates the expression of pair-rule genes, such as Gb'eve in the thoracic region through Gb'Kr and/or other downstream genes; the other is that very low levels of Gb'hbexpression could directly regulate the expression of pair-rule genes in the thoracic region. The latter possibility may be supported by the fact that in grasshopper embryos, transcripts of hb and its proteins exhibit distinct stepped expression levels in the gnathal/thoracic region, i.e. high levels in the gnathal region and low levels in the T1 region(Patel et al., 2001).

We detected maternal transcripts in Gryllus eggs using RT-PCR. However, in Schistocerca, which belongs to Orthoptera along with Gryllus, the Hb protein rather than the hb transcript appears to be maternally provided (Patel et al., 2001). It remains unclear whether maternal transcripts or proteins of Gb'hb functions in the early stages. Because our embryonic and parental RNAi experiments exhibited no qualitative differences and the embryonic RNAi yielded the severest class of phenotypes at a higher rate than the parental RNAi, the Gb'hb RNAi phenotypes we obtained seem to be a result of the depletion of zygotic transcripts. Thus, we could not estimate the contribution of the maternal transcripts of Gb'hb or its proteins to the RNAi phenotypes.

We found that the maxillary, labial and thoracic segments were homeotically transformed into abdominal identity in the late stages of Gb'hb RNAi embryos, while the anterior head and mandibular segment remained intact(Fig. 7C,D). A similar transformation has been reported in Of'hb RNAi embryos(Liu and Kaufman, 2004). As we observed that the ectopic expression of Gb'abdA in the maxillary,labial and thoracic segments is induced in early Gb'hb RNAi embryos not later than the stage when normal expression of Gb'abdA begins in the abdominal region (Fig. 7B), Gb'hb probably suppresses the abdominal identity in the maxillary,labial and thoracic segments by repressing the Gb'abdA expression during early embryogenesis, which is consistent with a report for Of'hb.

As the homeotic transformation occurs in a much larger region than the hb expression domain, it is also mysterious how Gb'hbregulates the expression of homeotic genes. In Oncopeltus, Of'hb was speculated to indirectly regulate the expression of Of'abdA, because ectopic expression of Of'abdA was not detected in the blastoderm stage (Liu and Kaufman, 2004). This may be true in the thoracic region of the Gryllus embryos,because the T3 segment is the most resistant to the homeotic transformation into abdominal identity caused by Gb'hb RNAi depletion. Because the expression of Gb'Kr was reduced in Gb'hb RNAi embryos, and we know that both hb and Kr act as abdA repressors in Drosophila (Casares and Sánchez-Herrero, 1995; Shimell et al., 2000), Gb'Kr appears to be a candidate of Gb'abdA repressor in the thoracic region, although the precise mechanism for induction of the Gb'Kr expression by Gb'hb remains to be clarified.

We also observed that Gb'Antp was expressed in the prospective labial and thoracic segments in the early stages(Fig. 7A). In the shortened and widened Gb'hb RNAi embryos, the Gb'Antp expression domain was severely reduced (Fig. 7A). Thus, Gb'hb regulates expression of Gb'Antp, directly or indirectly. The reduced Gb'Antp expression domain may correspond to the region most resistant to the homeotic transformation into abdominal identity caused by Gb'hb RNAi depletion, probably overlapping the Gb'Ubx expression and leading to T3 segment identity(Zhang et al., 2005). This should be closely related to the formation of the residual leg-like structures observed in Gb'hb RNAi embryos(Fig. 7C).

In the late stages, Gb'hb RNAi depletion also affected the expression patterns of Gb'Scr, Gb'Antp and Gb'Ubx. We found that Gb'hb is required for the expression of Gb'Scr in the labial and T1 segments, for the suppression of Gb'Antp expression in the maxillary segment, and for the suppression of Gb'Ubx expression in the region from the maxillary to T2 segments. Our results suggest that Gb'hb determines the anterior segment identities through regulation of expression of the Hox genes, as found in Drosophila.

Fig. 7E shows a regulatory network involving hb in Drosophila and a putative one in Gryllus. In Drosophila, the prolonged syncytial stage allows transcription factors to diffuse freely between the adjacent nuclei and exert their functions by forming diffusion-controlled gradients. Bicoid (Bcd)activates hb and Kr in the anterior region of the early Drosophila embryo with its morphogenetic gradient(Driever and Nüsslein-Volhard,1989; Hoch et al.,1991). In turn, the morphogenetic gradient of the Hb protein organizes the expression of the other gap genes Kr and knirps (kni)(Hülskamp et al., 1990). Hb acts as an activator of Kr and a repressor of kni at low levels, while high levels of Hb repress the Kr expression(Hülskamp et al., 1990). These gap genes define the expression domains of pair-rule genes, such as eve, to generate their periodic expression patterns. The functions of hb and Kr as an activator and a repressor, respectively, for the second pair-rule stripe (stripe 2) formation of eve have been extensively investigated (Small et al.,1991; Stanojevic et al.,1991). Additionally, hb regulates other pair-rule stripes of eve (stripe 3) and other pair-rule genes, e.g. runt and paired (Gutjahr et al.,1993; Klingler et al.,1996; Small et al.,1996).

In Gryllus (Fig. 7E), it was shown that Gryllus caudal (Gb'cad),instead of bcd, organizes the gap domains of Gb'hb and Gb'Kr (Shinmyo et al.,2005). Our Gb'hb RNAi analysis suggested that Gb'hb activates Gb'Kr, directly or indirectly, although more data are needed to establish this regulatory interaction. Furthermore, we found that Gb'hb directly or indirectly regulates the Gb'eveexpression in the prospective thoracic region of the embryo, suggesting that the hierarchical relationship between hb and eve is conserved between Gryllus and Drosophila. Our results also suggested that the Gb'abdA expression is suppressed by Gb'hband Gb'Kr, directly or indirectly, in the anterior region, as in Drosophila. Thus, although the maternal morphogenetic organizer in the regulatory network of segmentation may have switched from cad to bcd during insect evolution, its downstream relationship in the segmentation and Hox genes appears to have been mostly conserved between Gryllus and Drosophila(Fig. 7E).

In spite of these conserved aspects of the putative regulatory network, we found that Gb'hb functions differ considerably from those of Drosophila hb in segment formation(Fig. 7D). Although we cannot deny that more segments might be deleted in Gryllus, as observed in the Drosophila hb mutant, it is likely that Gb'hb plays the role of a gap gene that patterns fewer segments than in Drosophila. Interestingly, in the hb RNAi phenotypes of Oncopeltus, no segmentation gap was observed in the gnathal and thoracic regions, suggesting that the requirement of the hb gap function in this species is minimal, if at all present (Liu and Kaufman, 2004). On the contrary, Tribolium hb was reported to have the anterior gap function similar to Drosophila(Schröder, 2003). In Tribolium, (a short germ insect) some of the anterior segments are patterned under syncytial conditions(Sommer and Tautz, 1993). These lines of evidence suggest that the number of anterior segments regulated by hb increased during evolution from cellular to syncytial segmentation, or that the canonical function of hb may have evolved from the non-canonical functions of an ancestral hb during insect evolution. Although the anterior hb-expression domain appears to be fundamentally conserved in insects, the number of pair-rule stripes of the pair-rule genes regulated by hb in the anterior region may increase during evolution, probably owing to modification of the cis-regulatory elements of the pair-rule genes. Comparative analyses of the cis-regulatory elements of the pair-rule genes would shed some light on this issue.

In short/intermediate germ insects, the posterior segments form sequentially from the posterior growth zone during germband elongation. We found that Gb'hb is required in the formation of at least seven posterior segments. In addition to our finding, Of'hb was found to be expressed in the posterior growth zone and involved in growth and segmentation(Liu and Kaufman, 2004). On the other hand, in Drosophila, hb is reported to be required only for the formation of the A7 and A8 segments(Lehmann and Nüsslein-Volhard,1987) (Fig. 7D). These facts suggest that the hb function in the posterior region may have been reduced during evolution from the short/intermediate to long germ embryogenesis. As hb is similarly expressed in the prospective A7 to A9 segments in Gryllus and Drosophila, the hbfunctions in these segments may have been conserved in both short/intermediate and long germ embryogenesis. Recently, Peel and Akam(Peel and Akam, 2003) proposed an assumption that the posterior sequential segmentation in short/intermediate germ insects is controlled by a Notch signaling-dependent segmentation clock. If it is the case, Gb'hb might be involved in regulating segmentation clock. It is also possible that Gb'hb controls posterior growth through regulation of the morphogenesis of the germband, because the shape of the elongating posterior region was affected by Gb'hb RNAi depletion(Fig. 7B). Further precise analyses of posterior segmentation in short/intermediate germ insects will be required to elucidate its mechanisms.

We thank K. Matsushima, A. Takeuchi, C. Kobayashi and W. Shinahara for their technical assistance. We also thank the anonymous reviewers whose comments were very helpful. This work was supported by a grant from the Ministry of Education, Culture, Sports, Science and Technology of Japan to T.M., I.S., O.H. and S.N.