hunchback is required for suppression of abdominal identity, and for proper germband growth and segmentation in the intermediate germband insect Oncopeltus fasciatus

Though all insects possess a highly conserved adult body plan, this morphological conservation belies an underlying developmental diversity that gives rise to this body plan. For example, insects are described as being`short', `intermediate' or `long' germband and show fundamental differences in how their body segments are generated(Davis and Patel, 2002; Krause, 1939; Sander et al., 1985). Long germband segmentation is evolutionarily derived and in this form of embryogenesis, all body segments are specified early and simultaneously during the blastoderm stage. As short and intermediate germ segmentation is found throughout the insects, whereas the long germ type is restricted to the higher insects, it is likely that a form of short or intermediate germband segmentation is evolutionarily ancestral(Davis and Patel, 2002). In this mode of segmentation, only the anterior-most segments are specified during the blastoderm stage, leaving the rest of the body plan to be specified later in embryogenesis, during germband elongation. (Because the short and intermediate forms of segmentation are conceptually so similar, for convenience sake we will refer to both the short and intermediate forms as`short'.)

The fruitfly, Drosophila melanogaster exhibits long germband segmentation. In this fly, embryogenesis begins with the first nuclear divisions occurring without concomitant cellular divisions. Then nuclei fated to form the blastoderm migrate to the egg periphery and persist as a syncytium before cellularization. During the blastoderm stage, action of the segmentation gene cascade serves to subdivide this blastoderm into smaller and smaller regions, producing all the body segments (reviewed by St Johnston and Nüsslein-Volhard,1992). By the end of the blastoderm stage, fate maps show that all of the future body regions including the head, thorax and the entire abdomen are already represented proportionally on the Drosophila blastoderm(Lohs-Schardin et al.,1979).

This is in contrast to short germband segmentation. In this type of segmentation, nuclei fated to contribute to the embryo proper migrate to the egg cortex and cellularize, as in Drosophila. However, it is during the blastoderm phase where the differences between short and long germband segmentation become apparent. In short germband segmentation, only anterior segments – typically the head and thoracic regions – are proportionally represented upon the blastoderm fate map. It is only after the formation of the germband and during germband growth that the rest of the body forms. The posterior segments arise from a disproportionately small region of the posterior of the germ anlagen, termed the `growth zone'. This `growth zone' has not yet been well characterized but is the posterior-most region of the proliferating germband. Growth of this region results in germband elongation during which the rest of the segments are specified. Thus, although long germ segmentation can be thought of as successive spatial subdivision of the early blastoderm, short germ segmentation entails both spatial and temporal aspects, spatial during the blastoderm phase with the temporal aspect occurring later during germband elongation (reviewed by Davis and Patel, 2002).

How is the blastoderm of short germband insects patterned to yield only the anterior-most segments? How does the `growth zone' generate the posterior segments and how are these segments specified? Moreover, how did long germband segmentation evolve from the short form? Although much has been learned about the genetics that regulate long germband segmentation from Drosophila, we unfortunately know little about short germ segmentation. Our understanding of Drosophila segmentation can provide clues, but not answers to these questions. Therefore in order to gain insight into how the evolution from short to long occurred, and to get a better picture of insect segmentation in general, we must address our lack of understanding of short germband segmentation.

The striking embryological differences between short and long germ segmentation imply fundamental differences in patterning at the molecular level. For this reason, the roles of a few early developmental genes involved in anteroposterior axis specification have been the focus for understanding these differences. In Drosophila, the gap genes are responsible for the early subdivision of the blastoderm into broad regions, each of which will eventually encompass several adjacent body segments. As one of the essential differences between short and long germ segmentation lies in the early allocation of the blastoderm fate map, it seems reasonable to compare the action of the gap genes in short and long germ insects. We focus our attention on the gap gene hunchback (hb).

hunchback encodes a zinc-finger-containing transcription factor known to be important for axial patterning in a number of insects(Jürgens et al., 1984; Lehmann and Nüsslein-Volhard,1987; Patel et al.,2001; Schröder,2003; Tautz et al.,1987). Drosophila embryos mutant for hunchbackshow a gap phenotype, with deletions of the labial through third thoracic segments and the eighth abdominal segment.

We have investigated the role of hunchback in an intermediate germband insect, the milkweed bug, Oncopeltus fasciatus(Hemiptera:Lygaeidae). We first examined embryogenesis and segmental specification in this bug using engrailed staining. We then reported the expression pattern of Oncopeltus hunchback (Of'hb)during embryogenesis. A technique for parental RNAi has been previously reported for Tribolium castaneum(Bucher et al., 2002), and we have adapted this technique for use in milkweed bugs in order to determine the function of Of'hb in segmentation. We find that Of'hb is required both for suppressing abdominal identity in the gnathal and thoracic segments, and for proper growth and segmentation of the abdomen.

RNA was isolated from ovaries quickly dissected under cold Robb's Minimal Saline (Robb, 1969) and flash frozen in liquid nitrogen. Embryonic RNA was isolated from mixed stage embryos. Total RNA was extracted using Trizol (GibcoBRL/Life Technologies) and poly(A⁺) RNA was isolated using the Oligotex mRNA minikit (Qiagen). cDNA was synthesized using the FirstChoice RLM-RACE kit (Ambion). All PCRs were performed using the Advantage2 polymerase mix (BD Biosciences/Clontech). For the degenerate hunchback PCR, the primer pair was:AARCACCAYY-TNGARTAYCA-GTGWGMRTAYTTRCKCARRTG. 5′ and 3′ RACE PCRs were performed using a gene-specific primer and the anchor primers supplied in the FirstChoice RLM-RACE kit. After separation on an agarose gel, candidate PCR products were gel extracted if necessary (Qiagen), and cloned using the PCR-Script Amp Cloning Kit (Stratagene). At least three independent PCRs were performed and several clones sequenced in order to minimize PCR and sequencing artifacts. The Of'hb sequence has been submitted to GenBank under Accession Number AY460341. The Oncopeltus engrailed 5′ RACE fragment used to make the en in situ probe was cloned by designing primers to the previously identified homeodomain region(Peterson et al., 1998) and the sequence has been submitted to GenBank under Accession Number:AY460340.

Total RNA samples used in northern analysis were prepared as for cloning(see above). Probes for both northern analysis and in situ hybridization were synthesized from a 1.1 kb clone containing the 3′ end of the Of'hb ORF (Fig. 2C). Northern probes were prepared with biotin-UTP (Enzo) using the MAXIscript kit(Ambion). Northern blotting was performed using the NorthernMax kit (Ambion)onto BrightStar-Plus membrane (Ambion) and detected using the BrightStar BioDetect kit (Ambion).

The chorions on germband stage embryos were cracked by first shaking in 1:1 heptane:12% paraformaldehyde in PBTw for 20 minutes and then heptane/methanol cracked. After chorion cracking, embryos were fixed for 1 hour in 4%paraformaldehyde in PBT. For blastoderm preparation, it was necessary to first boil the eggs for 1 minute before continuing with the chorion cracking procedure. Ovaries for in situ hybridization were isolated by quick dissection in cold Robb's Minimal Saline, gently rocked in heptane for a few minutes, and fixed in 4% paraformaldehyde in PBTw for 1 hour.

In situ probes were prepared with digoxigenin-UTP, biotin-UTP or fluorescein-UTP (Roche) using the MAXIscript kit (Ambion). The in situ protocol used here was based largely on that of O'Neill and Bier(O'Neill and Bier, 1994) with some modifications. After embryos were fixed, they were soaked for 1 hour in RIPA detergent mix [150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS,1 mM EDTA, 50 mM EDTA, 50 mM Tris-HCl, pH 8.0]. Inclusion or omission of a proteinase K digestion step did not seem to affect the in situ results. In order to inactivate endogenous phosphatases, the embryos were incubated in hybridization buffer [50% formamide, 5× saline sodium citrate (SSC), 100μg/ml heparin, 100 μg/ml sonicated calf thymus DNA, 100 μg/ml yeast RNA, 0.1% Tween-20] at 70°C for 30 minutes. Embryos were then pre-hybridized at 60°C for 30 minutes before hybridization with probe at 60°C for 36 hours. Embryos were washed several times and soaked overnight at 60°C in hybridization buffer. We found that short washes in lower-salt solutions, first [50% formamide, 5× SSC, 0.1% Tween-20], then in [50%formamide, 2× SSC, 0.1% Tween-20] helped reduce background. After washing several times in PBTw, embryos were incubated in antibody hybridization solution [PBTw, 2 mg/ml bovine serum albumin (BSA), 5% normal sheep serum] for at least 1 hour. Alkaline phosphatase conjugated anti-digoxigenin, anti-biotin or anti-fluorescein antibody (Roche) was added to the embryos and allowed to rock overnight at 4°C. The next day, excess antibody was washed off in several changes in PBTw. The final color development step was carried out essentially as described Hauptmann and Gerster (Hauptmann and Gerster,2000) except for two-color in situs where the first AP antibody was inactivated by heating to 70°C for 30 minutes followed by additional fixation for 2 hours before continuing with the second AP antibody.

Template for the in vitro transcription reactions was prepared one of two ways. Plasmids containing the insert of interest were linearized by restriction digest, or template was prepared from a PCR where T3 and T7 phage promoter sequences were added to the primers. Sense and anti-sense RNA was synthesized in two separate reactions using the MEGAscript kit (Ambion). After purification, the sense and anti-sense RNAs were mixed to a final concentration of 2 μg/μl total RNA. The RNA was then annealed in injection buffer (Spradling and Rubin,1982) by heating in a thermocycler to 94°C and held at this temperature for 3 minutes, then slowly cooled to 45°C over the course of 1 hour. Proper annealing of the RNA was confirmed on an agarose gel. Parts of the gene to which these dsRNAs were made are shown in Fig. 2C.

Embryonic RNAi injections were carried out as previously described(Hughes and Kaufman, 2000). For the parental RNAi injections, the dsRNA was loaded into a Hamilton #801 syringe with a 32 gauge point #2 needle. Virgin female Oncopeltuswere anesthetized in CO₂ and injected in the abdomen between the fourth and fifth abdominal sternites with ∼5 μl of dsRNA solution. This volume was necessarily variable due to wound leakage. Injected females were then reared individually with males and allowed to lay eggs. Eggs were allowed to develop at 25°C and were harvested 76-80 hours after egg lay for in situs, while embryos for morphological phenotypic analysis were allowed to develop fully.

Readers are encouraged to contact the authors directly for more detailed protocols.

A description of Oncopeltus embryogenesis using classical histological techniques has been reported previously(Butt, 1947). In order to orient the reader to bug embryogenesis, we augment this previous work with our own observations using a fluorescent nuclear dye and also with stainings for a segment polarity gene, engrailed.

Oncopeltus embryogenesis can be divided into two distinct phases– a blastoderm phase, which in some ways is similar to that of Drosophila, and a germband growth phase which it shares with other short germ insects. Oncopeltus embryogenesis begins with the first nuclear divisions occurring synchronously within the yolk mass without concomitant cellular divisions. After several such divisions, the resulting cleavage nuclei migrate to the egg cortex. By fifteen hours after egg lay,they reach the surface of the egg and after an additional two hours, the formation of cell membranes is complete(Fig. 1A)(Butt, 1947). At this stage,the large ovoid blastoderm superficially appears very Drosophilalike. However, blastoderm cells of a 36- to 40-hour-old embryo are not evenly arranged around the yolk but are concentrated in two broad lateral domains on either side of the yolk mass (Fig. 1B) that are similar to the `lateral plates' seen in Rhodnius (Butt, 1947; Mellanby, 1935).

In addition to observing the cellular movements in the blastoderm, we also wanted to ascertain the number of segments that have been specified at this stage in embryogenesis. As the segment polarity gene engrailed(en) is expressed in the posterior compartments of segments in many arthropods, including Oncopeltus, it serves as a convenient molecular segmental marker (Hughes and Kaufman,2002; Patel et al.,1989; Rogers and Kaufman,1996; Telford and Thomas,1998). In situ hybridization of 36- to 40-hour-old embryos with Oncopeltus engrailed (Of'en) probe revealed a total of six vertical stripes on the blastoderm surface(Fig. 1C). This shows that by this stage, the blastoderm has already been allocated into six segments. We followed the migration of these stripes throughout embryogenesis to deduce their segmental affinities and have determined that these six initial stripes correspond to the mandibular through third thoracic segments. That engrailed is expressed at this stage is somewhat surprising, and shows that anterior patterning has occurred all the way to the segment polarity level long before the posterior body regions even exist. This reinforces the idea that although the Oncopeltus blastoderm may in some ways superficially resemble the Drosophila blastoderm, the milkweed bug blastoderm is subdivided in a distinctly different way.

Oncopeltus embryos are of the `invaginating' type, which refers to the cell movements that give rise to the germband. Shortly after the formation of the blastoderm lateral plates, the germband begins to form when the cells at the posterior end of the blastoderm dive into the center of the yolk mass. The early site of invagination is marked by a small pit at the posterior pole of the late blastoderm (arrowheads in Fig. 1B,C). The cells of the blastoderm surface migrate towards the posterior, while the leading tip of the elongating germband dives into the interior of the yolk mass, towards the anterior pole of the egg. In order to visualize these movements, it is instructive to imagine the blastoderm as an inflated balloon, with invagination occurring as if a finger is poked into the interior of the balloon. Thus, the cells on the outside of the blastoderm move towards the posterior of the egg, dive into the yolk and migrate towards the anterior of the egg. This can easily be seen by comparing Fig. 1C, where six stripes of engrailed expressing cells lie on the blastoderm surface, to Fig. 1D1 and 1D2, where the invaginating germband has pulled the two posterior stripes into the yolk,leaving the four anterior stripes on the surface of the blastoderm. As germband invagination continues, the tip of the germband eventually reaches the anterior pole of the egg and the resulting germband stage embryo ends up with its head at the posterior of the egg (the embryo does eventually right itself through later embryonic movements). As these embryonic movements can potentially lead to confusion, when discussing the blastoderm we will refer to the anteroposterior and dorsoventral axis in regards to the fate maps of the tissues.

During the germband stage, the remaining posterior body segments that were not specified during blastoderm stage are now produced through elongation of the posterior portion of the germband coupled with progressive anterior to posterior segmental specification. First, the abdominal region is generated through rearrangement and growth of the posterior growth zone and then engrailed stripes appear one by one in an anterior to posterior direction (Fig. 1E-H). This is similar to other short germband insects such as Thermobia domestica,Schistocerca americana and Tribolium castaneum(Brown et al., 1994; Patel et al., 1989; Peterson et al., 1998). Thus,it is clear that in Oncopeltus, as in other short and intermediate germband insects, posterior segments arise during a secondary growth phase during which the posterior germband undergoes great elongation with specification of abdominal segments occurring sequentially and in an anterior to posterior direction.

In order to clone Oncopeltus hunchback (Of'hb), we designed degenerate primers to the conserved zinc-finger domain of known hunchback sequences. We then performed PCR using these primers on cDNA made from ovaries or mixed stage embryos and isolated a short initial Of'hb clone. This clone allowed us to then design exact primers for 5′ and 3′ RACE and isolate fragments of Of'hb that together total 2.1kb and encode the entire open reading frame and regions of the 5′ and 3′ UTRs.

Of'hb is predicted to encode a 64.3 kDa protein with a total of eight zinc-finger domains. These zinc fingers are clustered with an N-terminal pair, a central cluster of four and a C-terminal pair(Fig. 2C). These eight zinc-finger domains are shared with the hunchback genes from two grasshopper species, Locusta migratoria and Schistocerca americana, while Drosophila melanogaster and Tribolium castaneum hunchback each encode only a total of six zinc fingers(Fig. 2A)(Patel et al., 2001; Tautz et al., 1987). The six fingers from the fly and beetle hunchback correspond to the central four and the C-terminal two fingers from the milkweed bug and the grasshoppers. Alignments of the homologous finger regions of grasshopper and milkweed bug hunchback show that the N-terminal zinc fingers are the most divergent.

As both the Schistocerca(Patel et al., 2001) and Oncopeltus hunchback encoded proteins contain eight zinc fingers,this is likely the ancestral state. However, Tribolium and Drosophila hb each contain only six zinc fingers, which suggests that six fingers is the ancestral state for the holometabola(Patel et al., 2001; Tautz et al., 1987; Wolff et al., 1995). If this is the case, then somewhere in the lineage leading to holometabola, hunchback lost its two N-terminal-most fingers(Fig. 2D). Unfortunately, as the function of the hb encoded protein itself has been studied only in Drosophila (which lacks these fingers), the function of these metal-binding fingers is unknown. It would be fascinating to examine the specific function of these ancient N-terminal fingers and see whether they can be correlated with developmental changes that have evolved in the holometabola.

In addition to the zinc-finger domains, three other domains were found to be conserved. The A-Box, originally identified in Drosophila as a region of similarity between hunchback, Kruppel and the HIV pol gene(Tautz et al., 1987), is also found in Oncopeltus hunchback. Additionally, sequences similar to the Drosophila hunchback Basic Box and the C-Box are also present(Fig. 2B)(Hülskamp et al.,1994).

Northern analysis on both maternally and zygotically derived total RNA revealed a single band of ∼3.2 kb in both samples, showing that Oncopeltus hunchback is expressed in the ovary(Fig. 3A). As our Northern analysis shows the presence of a larger transcript than our cloned fragments(the 5′ and 3′ RACE products together total 2.1 kb), our clones must not represent the entire Of'hb transcript.

In situs on ovaries confirm our northern analysis and show that the Of'hb transcript is expressed in the ovary and developing oocytes. Oncopeltus ovaries are of the telotrophic type, meaning the nurse cells remain in the germarium, while the developing oocytes continue their journey of maturation down each ovariole. The nurse cells supply the oocytes with maternal factors through their connection via nutritive cords. Reflecting this morphology, ovarian in situs show that hb accumulates throughout the germarium and stain especially strongly within zone III (arrowhead in Fig. 3B), where the nurse cells are located (Bonhag and Wick,1953). Transcript continues to accumulate and appears to be evenly distributed within the developing oocytes as they mature. hb becomes undetectable in late oocytes (Fig. 3B), but this absence of staining is probably due to chorion deposition around the mature oocytes as the chorion would present a physical barrier to probe and stain penetration.

Oncopeltus hunchback expression occurs in a distinct blastoderm pattern and a separate germband pattern that reflects the distinct blastoderm and germband phases of milkweed bug embryogenesis. In early embryos before blastoderm formation (12 hours after oviposition at 25°C), hbtranscript accumulates homogeneously throughout the egg (not shown). Shortly after, hb expression appears more strongly in the central region of the blastoderm (Fig. 4A). At 20-24 hours, this central domain becomes more strongly refined(Fig. 4B). By 24-28 hours, the single broad domain of expression begins to contract from the poles and resolve into two bands. The weaker, more anterior band spans 69-84% egg-length(with 0% being the posterior) while the stronger more central band covers 40-64% egg-length (Fig. 4C). As Oncopeltus is an intermediate germ insect, this region of the blastoderm corresponds to different segments than what would be expected in a long germ insect, such as Drosophila. Thus, in order to determine the approximate segmental register of Of'hb expression on the blastoderm,images of milkweed bug embryos separately stained for hunchback and engrailed were juxtaposed. This allowed us to determine that the anterior band spans the region of the blastoderm anterior to the mandibular en stripe, while the posterior band of hunchback appears to span the maxillary and labial segments(Fig. 4E).

Also at this stage, although no overt morphological indentation can yet be seen, a posterior patch of hunchback transcript arises prefiguring the location of future germband invagination (arrow in Fig. 4C1). In late blastoderms and in embryos initiating germband invagination (32-40 hours), the two anterior stripes of hb begin to fade and the blastoderm shows hb staining in a speckled pattern(Fig. 4D). When the layer of cells that express the broad striped hb pattern are peeled from the yolk ball, it is clear that this speckled staining lies underneath the blastoderm layer (not shown). Therefore, this speckled pattern is due to accumulation in the underlying yolk cells. As the germband begins its invagination, this site also shows stain accumulation, which may be a continuation of the earlier posterior patch, or may simply be due to physical capture of stain by the folded tissues of the invagination.

During the germband phase of embryogenesis, Oncopeltus hunchbackis expressed in a very different pattern than during blastoderm phase. If the broad blastoderm domains persisted through germband invagination, one would expect that in germbands, hunchback would likewise stain in a gap-like pattern in the gnathal/prothoracic region. Interestingly, this is not the case. From very early through very late germband stage embryos, Of'hb is never detected in an anterior gap gene-like domain(Fig. 5). Early germbands still undergoing invagination show very weak hb staining in a symmetrical chevron pattern which is segmentally reiterated and is excluded from the tip of the germband (Fig. 5A,B). This blocky segmental expression is most likely not a continuation of the earlier blastoderm domain but rather represents de novo expression in this new pattern. This expression is similar to a segmentally reiterated pattern also seen in Musca and Tribolium and may represent mesodermal staining as it does in Schistocerca(Patel et al., 2001; Sommer and Tautz, 1991; Wolff et al., 1995). As the broad bands of Of'hb expression in the blastoderm are already beginning to weaken by late blastoderm, it is likely that expression in these segments has already faded below detection by the time the germband completes invagination. This is in contrast to the hunchback expression patterns in both Schistocerca and in Tribolium, where it is expressed in a broad gap-like domain encompassing the gnathal to anterior first thoracic segments in early germbands(Patel et al., 2001; Wolff et al., 1995).

During germband elongation, Oncopeltus hunchback is expressed strongly in a patch at the very posterior of the germband. This posterior domain of hb expression begins as two small dots at the tip of the elongating germband (Fig. 5C). Double staining with engrailed shows that these dots only appear after the first abdominal segment has already formed. These two spots quickly expand to form a half-moon shape (Fig. 5D) that persists continuously throughout germband extension and can be detected until the formation of the tenth and final abdominal engrailed stripe (Fig. 5F). This continuous expression suggests that hunchbackmay be required in the growth zone throughout germband elongation.

Earlier, just before germband invagination, there is a small patch of hunchback expression in the posterior of the blastoderm (arrow in Fig. 4C1). As hb is expressed in the posterior growth zone during abdomen formation, these two domains may at first glance seem to represent continuous hbexpression. However, this is not the case as early germbands, which are just beginning the invagination process, do not express hb at the posterior tip (Fig. 5A,B). It is only after invagination is complete and during the formation of the abdominal segments that hb expression reappears in the growth zone. Therefore, these two domains cannot represent a continuous expression of hb. Rather, the patch of expression in the `growth zone' must represent a de novo initiation of hunchback transcription.

hunchback is also expressed in a pattern that may represent the developing nervous system. The punctate pattern in late germbands are reminiscent of neural expression patterns and are consistent with hbneural expression in other insects and its function in Drosophila(Isshiki et al., 2001; Patel et al., 2001; Wolff et al., 1995). This neural-like expression suggests that hunchback is also required for neurogenesis in Oncopeltus.

Recently, it has been reported that female Tribolium pupa which had been injected with double-stranded RNA (dsRNA) produce progeny showing an RNAi knockdown phenotype (Bucher et al.,2002). In order to test this method in Oncopeltus, we injected dsRNA corresponding to a region of Oncopeltus Sex combs reduced (Of'Scr), a homeotic gene, into the abdomens of adult virgin females and scored their progeny for defects. These progeny had their labial appendages transformed into a pair of appendages of mixed leg/antennal-like identity (Fig. 6B) – a phenotype identical to that already published for Of'Scr by direct injection of dsRNA into early embryos(Hughes and Kaufman, 2000). We achieved much higher penetrance and a similar range of phenotypes using parental RNAi (pRNAi) when compared with embryonic RNAi (eRNAi). All surviving injected females eventually produced clutches of embryos with the Scrphenotype and these animals showed the complete range of defects as reported earlier using embryonic RNAi. Sometimes, the first clutch laid contained wild-type hatchlings – most likely because these eggs had already completed oogenesis and laid down their chorions that would be impenetrable to dsRNA. These same females would later go on to lay clutches that did show the Scr knockdown phenotype. We also found that over the span of about 3 weeks, the severity of defect would first increase, peaking around at 10 days post-injection and then gradually decrease after that(Fig. 6C). As pRNAi showed identical phenotypes as eRNAi without any injection artifacts and showed the full range of severity, this technique should prove to be a highly specific and convenient method for studying gene function in Oncopeltus.

In order to determine the functional role of hunchback during Oncopeltus embryogenesis, we used RNAi to deplete the hbtranscript and produce knockdown phenotypes. As parental RNAi does not yield any injection artifacts, we preferred to use pRNAi but had to first confirm that for hunchback, pRNAi and eRNAi produced equivalent phenotypes. We found that for Of'hb, as for Of'Scr, no qualitative difference in the phenotypes were produced by either technique(Table 1). In order to test the effect of dsRNA length, we also injected several overlapping hbdsRNAs of varying length – 300 bp, 500 bp and 1.1 kb(Fig. 2C), and found that the resulting phenotypes from all the fragments were again qualitatively identical. However, we did find that the shorter fragments gave slightly more severe phenotypes (Table 1– compare the frequency of classes IV and V for the 300 bp fragment and the 1.1 kb fragment). Moreover, a different 300 bp dsRNA fragment was tested individually and also pooled with the first 300 bp fragment and each gave the same qualitative phenotype as the other longer fragments. As both pRNAi and eRNAi gave identical phenotypes and all the differently sized dsRNA fragments tested also yielded the same defects, we are confident that they truly represent depletion of the hunchback gene product.

RNAi with Oncopeltus hunchback resulted in phenotypes that ranged in severity from wild-type to very severe, and we were able to categorize the resulting embryos into five phenotypic classes based on this range of phenotypic severity. This allowed us to arrange a hypomorphic series of hunchback function that aided our interpretation of the phenotype. The Of'hb RNAi phenotype consisted of two aspects, which we will describe in turn.

The first aspect of the hunchback depletion phenotype is a homeosis of the gnathal and thoracic regions towards abdominal identity. Class I embryos (the weakest phenotypic class) made up 8.3% of the embryos(Table 1). These embryos showed strong suppression of the labium and mildly defective first thoracic legs(Fig. 7B1,B2). The fact that the labial segment is most strongly affected demonstrates that this segment is at the epicenter of defect and is consistent with the strong band of hunchback expression spanning the maxillary through anterior prothoracic segments in the blastoderm.

Embryos with a moderate RNAi defect constituted classes II and III and together totaled 43.3% of all the embryos(Table 1). In these embryos, in addition to the labial defects, it appears that the region of defect has expanded to include the anterior thoracic regions. Appendages on these segments are suppressed and deformed while the body segments begin to adopt abdominal features. For example, in class II embryos, the pigmentation of the gnathal region and anterior thorax begins to resemble that of the abdomen(compare Fig. 7C1 with Fig. 7A1) and spiracles of abdominal identity and position are apparent in the gnathal region (Fig. 7C2,C3). Furthermore, a ventral spiracle that in wild-type animals is present on the fourth abdominal sternite is ectopically produced in the transformed thoracic region of affected animals(Fig. 7C5,C6). In the more severe class III embryos, the region of defect expands to encompass more of the thoracic region, resulting in increased suppression of the first and second legs often leaving only the third leg(Fig. 7D2). In addition,ventral segmentation appears abdominal like(Fig. 7D2) showing that abdominal transformation of these segments is nearly complete. Additionally in these animals, dorsal segmental defects are occasionally seen(Fig. 7D1). Thus, at the morphological level, these hunchback hypomorphs show a transformation of the labial and thoracic segments towards abdominal identity.

We reasoned that transformation of the gnathal and thoracic regions towards abdomen should be accompanied by expression of abdominal genes in these transformed segments. In wild-type animals, the homeotic gene abdominal-A (abd-A) is expressed in the abdomen from the second through eighth abdominal segments(Fig. 7E). Therefore, we examined the expression domains of abd-A in embryos that were depleted for hunchback. Although we did not detect abd-Aexpression during the blastoderm stage in hb RNAi embryos (data not shown), we did find ectopic expression of abd-A during the germband stage. In hb-depleted germband stage embryos, abd-A was ectopically expressed in the gnathal and anterior thoracic segments,coincident with suppression of appendages on these segments(Fig. 7F,G). In weaker hypomorphs, the ectopic domain spanned the posterior of the maxillary segment through the anterior of the first thoracic segment and leg(Fig. 7F). In more strongly affected RNAi embryos, the ectopic domain of abd-A was more expanded– into the first and second thoracic segments and was associated with suppression of legs on these segments (Fig. 7G).

The ectopic expression of abd-A shows that hb RNAi embryos at both the morphological and molecular level show a homeosis of the labium and thorax towards abdominal identity. Moreover the hypomorphic series shows that the domain of transformation starts at the labial and first thoracic segments in weakly affected animals, expands posteriorly to include the second thoracic segment in moderately affected ones, and incorporates the third thoracic segment in strong hypomorphs, which suggests that the labial and first thoracic segments are the most sensitive to hunchbackdepletion.

In addition to the anterior homeotic transformations described above, hunchback depletion revealed another defect apparent in the abdomen. Decreasing hb activity results in increased posterior compaction. This defect is most evident through the hypomorphic series where a trend towards smaller and more defective abdominal segments in stronger phenotypic classes is apparent. Whereas class I embryos show a largely wild-type abdomen(Fig. 7B1), class II embryos show mild compaction of the abdomen (compare Fig. 7C1 with 7A1). This compaction increases and is often associated with segmental defects in class III embryos (Fig. 7D1). Embryos of stronger phenotypic classes are smaller than normal overall, and the abdomen shows much stronger compaction (compare the hb RNAi embryos in Fig. 8A,C2,D to the uninjected animals in Fig. 7A1,A2). In class IV embryos, a tiny and deformed posterior leg is often situated near the extreme posterior of the animal(Fig. 8B,C1). As the third thoracic segment is the last to be transformed and its leg is most resistant to suppression, this posterior appendage is most likely the partially suppressed remnants of the third thoracic leg. In these animals, the gnathal and anterior thoracic segments are strongly transformed towards abdomen,resulting in suppression of their appendages, while the third thoracic segment is only partially transformed, leaving a remnant of the third leg. Therefore,the posterior position of these stubby legs underscores the extreme degree of posterior compaction in these animals. Coupled with the homeosis described above, strongly affected animals would thus have a normal head followed by several abdominal-like segments that consist of a transformed labium and thorax and a tiny compacted abdomen.

This posterior compaction is perhaps most clearly seen during embryogenesis, with staining for engrailed. By using the degree of gnathal and thoracic appendage suppression as an indicator for phenotypic severity, we can see (as judged by disruptions in Of'en expression)that abdominal segmentation becomes more defective in more severely affected animals. Putative class II animals show mild defects in abdominal segmentation(Fig. 7H), while class IV(Fig. 8F) and class IV/V(Fig. 8G) animals show more strongly affected abdominal patterning. These defects may be the cause of the posterior compaction seen at the morphological level. Although the abdomens of hunchback-depleted embryos can be severely compacted, it does not seem that this compaction is merely due to a precocious termination of segmentation. If precocious termination were the case, we would expect fewer but relatively normal abdominal segments. Therefore, it is possible that hunchback is somehow required for normal posterior growth with the segmentation defect occurring as a consequence of this. Importantly, while segmentation in the abdomen is defective, it is apparent that segmentation is normal in the gnathal and thoracic segments (see Fig. 7H, Fig. 8F,G). Although the appendages of these segments are suppressed, their morphology and engrailed expression appear otherwise normal.

The most severe class of animals comprised 11.3% of the total(Table 1) and the interpretation of this class is based somewhat on an extrapolation of the hypomorphic series. These embryos are extremely small and are found buried in the yolk mass. In normal embryogenesis, the germband undergoes dorsal closure to encircle the yolk so in the wild-type case, the yolk mass ends up inside the embryo. In class V embryos, it appears instead that the embryo develops without enveloping the bulk of the yolk and thus much of the egg space is filled with unused yolk. Upon closer examination of these embryos it seems that some internal organs may have actually developed on the outside of the embryo, similar to the `everted' embryos described by Sander(Sander, 1976). In this class of animals, only the anterior-most structures are clearly identifiable. Eyes and antenna develop, but the rest of the body is highly reduced and composed of fewer and smaller segments (Fig. 8E). It is important to note that although this body region is small, it is apparent that some segmentation has occurred. In a putative class V germband, segmentation of the anterior is apparent, with a few anterior segmental grooves clearly forming (arrows in Fig. 8H). This region of somewhat normal segmentation is followed by a highly defective region. abd-A in situs show that aside from the head and mandibular segments,the entire trunk expresses abd-A. Therefore, in this extreme class of hb depletion, posterior segmentation is highly defective and all of the post-maxillary body has adopted an abdominal fate. This phenotypic class may represent the most severe combination of the two aspects of hunchback depletion where the labial and thoracic segments are completely transformed towards abdominal identity and the abdomen has been severely compacted. This results in an animal in which the head is followed by a small number of segments, all of which have abdominal identity.

Our complete model including both aspects of hunchback function in Oncopeltus is presented in Fig. 9. We propose that as Of'hb function is depleted, first the labium and then the thorax is transformed towards abdominal identity. A second requirement is evident in the abdomen. In increasingly severe hb depletions, growth and segmentation in the abdomen becomes increasingly defective resulting in posterior compaction. In addition,apparent is a lack of clear segmental defects in the gnathal and thoracic regions, even in strongly affected animals. Therefore, in the most severe cases, a normal head is followed by a few segments of abdominal identity. Although this would at first glance appear to mimic the hunchback gap phenotype in Drosophila and Tribolium, we conclude that the strongest phenotype is due to the combination of gnathal and thoracic transformation towards abdomen coupled with severe posterior compaction.

At first glance, the Oncopeltus blastoderm seems to resemble that of Drosophila. It is easy to imagine that developmental mechanisms instructing embryogenesis would be conserved between these species. However,since milkweed bugs undergo intermediate germband segmentation, there are fundamental differences between patterning in these two species. Oncopeltus embryogenesis consists of two distinct phases – a blastoderm phase and a germband phase and is reflected in the expression of engrailed. Our engrailed staining reveals that in Oncopeltus, the blastoderm has only been allocated into six anterior segments and confirms the milkweed bug's status as an intermediate germband insect. Moreover, the fact that engrailed transcript can be detected at all during the blastoderm stage shows that the degree of anterior patterning has progressed all the way to level of the segment polarity genes. This means that molecular patterning of the anterior segments is essentially complete at a stage where the posterior body regions do not yet even exist. Thus, milkweed bugs show a severe heterochronic discontinuity between anterior and posterior segmental patterning. This anteroposterior discontinuity is in marked contrast to Drosophila, where by the completion of the blastoderm stage, the entire body region is proportionally represented and segmental patterning occurs more or less simultaneously across the entire blastoderm (Lohs-Schardin et al.,1979).

Oncopeltus hunchback expression and function reflect the biphasic nature of milkweed bug embryogenesis. hb is expressed in two broad stripes during the blastoderm stage. The stronger band spans the posterior maxillary, labial and anterior first thoracic segments. As hb is not expressed in these segments during the germband stage (except in the mesoderm and a neural-like domain), we attribute its region-specifying function in these segments to its expression domain in the blastoderm. This abdomen-repression function in the labial and thoracic segments may occur either through direct suppression of abd-A in the anterior, or may occur indirectly through regulation of a downstream gene responsible for specifying abdominal regional identity. As the ectopic domain of abd-A in hb RNAi animals was not detected at the blastoderm stage but only later during the germband stage and the region of homeosis in hb RNAi animals is much larger than the hb blastoderm expression domain, it is most likely that hb indirectly regulates abd-A. Indeed, normal abd-A expression in the abdomen appears long after hb expression in the growth zone has already faded. Our observation that hunchback in milkweed bugs serves to repress abdominal identity is not without precedence. In Drosophila,certain hypomorphic alleles of hunchback, class V alleles, also produce homeotic transformations of the gnathal or thoracic segments(Jürgens et al., 1984; Lehmann and Nüsslein-Volhard,1987). However, unlike Oncopeltus, these transformations are superimposed on a deletion phenotype.

As in flies, hunchback RNAi depletion in the red flour beetle, Tribolium castaneum, also produces the gnathal and thoracic gap phenotype (Schröder,2003). In this light, it is interesting that Oncopeltus hunchback knockdowns do not show the canonical gap phenotype but rather a transformation. This may reflect either incomplete RNAi knockdown of the Of'hb gene product or may reflect differences in anterior-posterior patterning in milkweed bugs. It is possible that if further depletion were possible, the anterior homeotic phenotype would be replaced by deletion of those segments. However, this would be in contrast to the case in Drosophila as weak alleles of hunchback in flies are not associated with transformations but rather yield small deletions while stronger alleles serve to increase the size of the deleted region(Lehmann and Nüsslein-Volhard,1987). The Oncopeltus hunchback phenotype reported here seems to reflect a strong depletion of the hunchback gene product because of the large domain of the homeotic defect from the labium to the third thoracic segment in the severe phenotypic classes. Transformation of the third thoracic segment is indicative of strong hb RNAi yet in these same animals, the labial segment is present and without segmental defects. If deletion of the labial segment were merely an issue of sensitivity, we would expect that animals with transformed third thoracic segments also show segmental defects in the labial segment. Therefore, if hunchback has a gap function in this animal, its requirement must be minimal.

We have shown that hunchback transcript is provided maternally in Oncopeltus and it is a formal possibility that the protein is as well. If this were the case, it is possible that maternal hunchbackserves to specify the presence of the gnathal and thoracic segments, while zygotic activity functions to suppress abdominal identity in these regions. Although maternal loading of hunchback-encoded protein has not been reported in either flies or beetles, the protein is provided maternally in grasshoppers. In grasshoppers however, axial patterning by hunchbackappears to be performed entirely by zygotic function whereas maternal hunchback activity in this animal may serve to distinguish embryonic from extra-embryonic cells (Patel et al.,2001).

In strongly affected Oncopeltus hunchback RNAi animals, the abdomen is severely compacted and segmentation is defective. hunchback function in the developing germband probably reflects its expression in the posterior `growth zone'. Therefore we propose that Of'hb is required for proper growth and segmentation of the posterior germband. At this time we can only speculate on the nature of this requirement. It may be that Of'hb is directly involved in the generation of segments as the posterior germband grows. However, it may also be that Of'hb is merely required for posterior elongation of the germband while the actual patterning of segments occurs relatively independently of growth. Thus, the segmental defects seen in developing germbands may be a consequence of improper elongation. Alternatively, hunchback may be required for proper functioning of the growth zone itself.

hunchback has also been examined in Schistocerca and Tribolium, two other short germband insects. Schistocerca hunchback is not expressed continuously in the growth zone as it is in both milkweed bugs and beetles but rather arises in abdominal patches corresponding to the A4/A5 and A7-A9 segments(Patel et al., 2001). Therefore the continuous expression in the growth zone may represent a derived pattern in the insects. As noted, Tribolium and Oncopeltus hunchback are expressed in identical patterns in the growth zone. However, it has not yet been reported that hb RNAi in Tribolium leads to posterior compaction, rather the phenotype reported is the canonical hunchback gap phenotype. With such disparate expression and functional data from these three insects, it is difficult to determine the ancestral function of hunchback in the insect growth zone and it is clearly imperative that wider taxonomic sampling needs to be done.

This brings us to examinations of the very nature of the insect growth zone itself. Almost nothing is known about how this special region of the germband develops and ultimately gives rise to the posterior segments. There are no overt morphological features that distinguish it. However, the growth zone must be special as several segmentation genes such as even-skipped,caudal and hunchback are expressed there(Dearden and Akam, 2001; Patel et al., 1994; Wolff et al., 1995) (P.Z.L.,unpublished). Given that some form of short germband development is ancestral in insects, we must understand this mode of development in order to understand the evolutionary transition from short to long germ segmentation. Moreover, as other arthropods undergo embryogenesis in a manner similar to insect short germband segmentation, functional studies in Oncopeltus and other short germband insects may in fact shed light on all the arthropods.

We thank Cynthia L. Hughes and David R. Angelini for critical reading of the manuscript. We also thank two anonymous reviewers whose comments and expertise were very helpful. Equipment and facilities used in this work were partially supplied by INGEN funding. P.Z.L. was supported by an NIH Genetics Training Grant GMO7757 and an NSF Integrative Graduate Education and Research Traineeship (IGERT) 2039L1A-A1. T.C.K. is an HHMI Investigator.