Grasshopper hunchback expression reveals conserved and novel aspects of axis formation and segmentation

A great deal is known about the molecular mechanisms responsible for patterning and segmentation during Drosophila melanogaster embryogenesis, and this knowledge provides an excellent framework for examining both the conservation and the evolution of these processes in the arthropod phylum. Such comparative data not only provide a more accurate assessment of how these processes vary between insects, but also of how early developmental mechanisms evolve in all animals.

Genetic and molecular analyses have demonstrated that segmentation in D. melanogaster depends on progressive subdivision by a hierarchy of regulatory factors (Rivera-Pomar and Jäckle, 1996; St Johnston and Nüsslein-Volhard, 1992). Initially, maternal gradients of bicoid and nanos provide information that defines position along the length of the syncytial embryo. Translational repression by bicoid and nanos result, respectively, in gradients of caudal and hunchback (Hb) protein. The bicoid, Hb and caudal gradients lead to activation of zygotic gap gene transcription at distinct positions along the anteroposterior (AP) axis. Regulation by maternal and gap gene products then leads to the expression of pair-rule genes in stripes of a two-segment periodicity. The pair-rule genes thus define the initial reiterated pattern within the embryo. Segmental pattern is later maintained and refined by the expression, under pair-rule regulation, of the segment polarity genes.

Both the data summarized above, and the results of various embryonic manipulations, show that in Drosophila the entire body plan is established simultaneously across the length of the blastoderm embryo. This type of development (in which body regions are present at blastoderm in the same proportions as are found in the hatching larva) is known as long germ embryogenesis (Sander, 1976). By contrast, short germ embryos, such as those of the grasshoppers Locusta and Schistocerca, appear to pattern only the head region of the body before gastrulation, more posterior regions being patterned during a subsequent growth phase. Intermediate germ insects fall somewhere between these extremes. The long germ mode of development is seen only in the most phylogenetically derived insect orders, and thus ancestral insect development was probably closer to short or intermediate germ embryogenesis.

Significantly, features of short germ development suggest that the mechanisms of pattern formation in these embryos may be fundamentally different from those seen in Drosophila. For example, during formation of the grasshopper abdomen, segments are defined sequentially along the AP axis, suggesting temporal disparity in the response to positional information along this axis. In addition, cellularization of the grasshopper blastoderm occurs relatively early (Ho et al., 1997), so that positional information must be interpreted in a cellular environment: this differs from the fly segmentation paradigm, where a syncytial environment seems crucial to patterning.

Despite the very different embryogenesis of short germ insects, there is evidence that some molecular aspects of Drosophila segmentation are conserved. In particular the segment polarity genes engrailed and wingless are expressed in a conserved pattern at parasegment boundaries in many insects including the grasshopper, Schistocerca, and beetle, Tribolium castaneum (Brown et al., 1994b; Dearden and Akam, 2001; Nagy and Carroll, 1994; Patel et al., 1989a; Patel et al., 1989b). However, earlier events in Drosophila segmentation appear less conserved in short germ insects. Although genetic and molecular data suggest that pair-rule patterning is conserved in Tribolium (Brown et al., 1994a; Maderspacher et al., 1998; Patel et al., 1992; Patel et al., 1994; Schröder et al., 1999; Sommer and Tautz, 1993), such evidence has proven more elusive in the grasshopper. Recently, a homolog of Drosophila paired (prd) has been found to be expressed with two-segment periodicity during grasshopper segmentation (Davis et al., 2001). Given, however, that the grasshopper eve and ftz orthologs are not expressed in a pair-rule pattern, significant differences in the mechanisms of segmentation between Drosophila and the grasshopper may exist (Dawes et al., 1994; Patel et al., 1992). Reasoning that the source of these differences might lie upstream, we wondered whether the expression of regulators of Drosophila pair-rule genes might also differ significantly in the grasshopper.

In order to better understand the similarities and differences between long and short germ embryogenesis, we have tried to identify orthologs of the earliest factors involved in the Drosophila segmentation hierarchy, and we focus here on hb. The gap gene hb encodes a zinc-finger transcription factor that functions early in development, both as a maternal morphogen and a gap gene. During Drosophila oogenesis, the egg is loaded ubiquitously with maternal hb transcript (Tautz et al., 1987). Upon fertilization, maternal hb transcript is translated only anteriorly, owing to translational repression by a gradient of maternal nanos protein emanating from the posterior end of the embryo (Hülskamp et al., 1989; Irish et al., 1989; Wang and Lehmann, 1991). This translational repression results in the formation of a Hb protein gradient derived from the maternally supplied mRNA (Tautz, 1988). Previous studies in Drosophila reveal that this translational repression depends on the 3′ untranslated region (3′UTR) of the hb transcript. Here, a bipartite sequence (the nanos response element, NRE) is recognized by pumilio protein, which in turn recruits nanos (Murata and Wharton, 1995; Sonoda and Wharton, 1999; Wharton and Struhl, 1991). Together nanos and pumilio act to prevent hb translation in the posterior part of the embryo. The resulting Hb protein gradient is known to behave morphogenetically, regulating its own promoter, as well as other gap genes such as Krüppel, knirps and giant in a concentration-dependent manner (Hülskamp et al., 1990; Simpson-Brose et al., 1994; Struhl et al., 1992).

Zygotic expression of Drosophila hb begins at the blastoderm stage (Fig. 1A; Tautz et al., 1987). Drosophila hb mRNA is transcribed from two promoters, P1 and P2, resulting in transcripts of 3.2 and 2.9 kb, respectively (Schröder et al., 1988). The P2 promoter is activated by the anterior morphogen bicoid, while the 3.2 kb transcript is activated across parasegment 4 (PS4) by, among other factors, Hb itself (Schröder et al., 1988; Tautz, 1988). The zygotic Hb protein gradient (Fig. 1A) is similar to its maternal counterpart. Indeed the two gradients are almost redundant: maternal hb is dispensable for AP patterning, and zygotic hb activation by bicoid almost is (Hülskamp et al., 1989; Irish et al., 1989; Wimmer et al., 2000).

Anterior hb expression is crucial for the development of thoracic segments. In addition, a posterior domain of zygotic hb, under independent control, emerges and refines to a domain spanning PS13/14 of the embryo (Fig. 1A)(Tautz, 1988; Tautz et al., 1987). Loss of this expression accounts for the fusion of segments A7 and A8 in the hb mutant. Later in development, hb is expressed in specific mesodermal cells and in the nervous system. Within the nervous system, transient hb expression is seen in neuroblasts and in a sub-population of ganglion mother cells (GMCs) and neurons (Fig. 1B,C,E)(Kambadur et al., 1998; Tautz et al., 1987; Wolf and Schuh, 2000).

hb orthologs have been isolated from other dipterans including D. virilis, Megaselia abdita and the midge, Clogmia albipunctata (Rohr et al., 1999; Sommer et al., 1992; Stauber et al., 2000; Treier et al., 1989). As in D. melanogaster, an anterior domain of hb mRNA is observed for all three species, and may be generated from an initially ubiquitous pool whose maternally repressed transcripts are subsequently degraded. Gap-like expression domains are later observed in both the anterior (gnathal/thoracic) and posterior (abdominal) regions. Thus, the role of hb in anteroposterior patterning is apparently conserved within flies. Indeed, functional data for this has been collected from M. abdita, where RNA interference experiments yield a phenotype that resembles a hypomorphic D. melanogaster hb allele (Stauber et al., 2000).

hb orthologs have also been isolated from insects other than flies, and expression patterns have been described in lepidopterans (Manduca sexta and Bombyx mori) and the beetle, Tribolium castaneum (Kraft and Jäckle, 1994; Wolff et al., 1995; Xu et al., 1997). In all three cases, hb is expressed in the anterior region of the embryo. In the case of Tribolium, this expression domain is thought to be equivalent to the anterior gap domain of Drosophila hb. Also conserved is the abdominal expression domain (in the PS13/14 region) of hb.

In all insect species examined, neural expression of hb is conserved, suggesting that a neural function is ancestral. However, as the expression of the eve and ftz genes during segmentation is not conserved between grasshopper and Drosophila, and these genes lie below gap genes such as hb in the Drosophila segmentation hierarchy, it was unclear whether the role of hb in AP patterning would be conserved in more basal insects.

In order to answer this question, we have isolated full-length orthologs from two grasshopper species, Locusta migratoria and Schistocerca americana, the most basal insect hb homologs yet examined. Sequence analysis of these orthologs compared with hb genes from other phyla enables us to propose a novel ancestral structure for the Hb protein. Additionally, we have examined the expression pattern of hb in the grasshoppers and find that it is expressed both maternally and zygotically during early development. As in other insects, the anterior domain (gnathal/thoracic), abdominal domain and nervous system expression patterns are conserved in the grasshopper. In addition, as we detect at least two different levels of hb expression in the grasshopper gnathal/thoracic segments, we propose that the morphogenetic role of hb is conserved. We also find that grasshopper hb is present anteriorly but absent from the posterior growth zone at early stages, suggesting a conserved role in anterior patterning, despite the very short germ mode of embryogenesis of grasshoppers. Furthermore, we define hb expression domains in the grasshopper mesoderm and serosa. Examination of the Drosophila hb expression pattern shows that it is expressed extra-embryonically, suggesting that the gene may have played an ancestral role in serosal development. Finally, we note that maternally supplied Hb protein is spatially restricted within the newly laid Schistocerca egg, and propose that this initial expression domain distinguishes between the cells that will contribute to the embryo as opposed to extra-embryonic tissue.

Grasshopper eggs were collected from colonies maintained in Berkeley and Chicago, USA (Schistocerca americana) and Canberra, Australia (Locusta migratoria). To stage S. americana embryos from 0 to 10%, timed eggs were collected by taking newly laid eggpods and allowing them to develop at 32°C in a moist environment. At this temperature, S. americana embryos take approximately 20 days to hatch, and thus each day represents 5% of development. Embryos are described either by time after egg lay (0-10%=0-48 hours AEL) or staged (10-40%) using morphological criteria or staining for engrailed (Bentley et al., 1979; Patel et al., 1989a). We also used a set of staging criteria developed by Peter Dearden and Michael Akam for Schistocerca gregaria, to help set our own 0-10% timeline (S. gregaria and S. americana are not identical morphologically, but similar enough to make comparisons reasonable). The development of L. migratoria occurs at a different rate, so these embryos are staged by morphological comparison with S. americana.

The primers and conditions used for hb amplification are those described by Sommer and Tautz (Sommer and Tautz, 1992). mRNA was extracted from 30% S. americana embryos and used for first strand cDNA synthesis and then PCR. A multimerized copy of this PCR fragment was used to screen a L. migratoria genomic library, and a region of the genomic clone was used to isolate a cDNA from a L. migratoria ovarian cDNA library. Using the sequence of the original S. americana PCR fragment, non-degenerate primers were designed and used for 5′ and 3′ RACE (Gibco BRL) to obtain further cDNA sequence for S. americana. Genomic clones for S. americana were isolated using the GenomeWalker system (Clontech).

For Schistocerca, we have isolated cDNA sequence beginning 199 nucleotides 5′ of the start methionine and extending to a polyA tail 1243 nucleotides 3′ of the stop codon. Our genomic sequence extends 2740 nucleotides 3′ of the stop codon and contains all the cDNA sequence as well as complete introns with the exception of one intron between the N-2 and M-1 zinc fingers, which was hopped over using primers based on the cDNA sequence. The Locusta cDNA sequence begins 152 nucleotides 5′ of the start methionine and extends to a polyA tail 1215 nucleotides 3′ of the stop codon. The Locusta genomic sequence extends from a HindIII site in the last intron to a position 2529 nucleotides 3′ of the stop codon. The S. americana and L. migratoria cDNA and genomic sequences are deposited in GenBank (Accession Numbers, AY040605, AY040606 and AY040607, AY040537, AY040538).

For S. americana embryos, whole-mount in situ hybridization using a digoxigenin-labeled RNA probe corresponding to the hb 3′UTR was performed as previously described (Patel, 1996) with the following changes. Embryos of S. americana were fixed for 15-20 minutes in 3.7% formaldehyde after dissection in 1× PBS (pH=7.4) and were then dehydrated stepwise in methanol. Fixed embryos were not treated with xylene or proteinase K, and the hybridization was carried out overnight at 60°C in sodium dodecyl sulfate (SDS) hybridization solution (SDS-Hyb: 50% formamide, 5× saline sodium citrate (SSC) (pH 4.5), 0.1% Tween-20, 0.3% SDS, 50.0 μg/ml heparin and 100.0 μg/ml sonicated salmon sperm DNA). The embryos were then washed at 60°C in the following solutions of SDS-Hyb/ PTw (1× phosphate-buffered saline (PBS) containing 0.1% Tween-20): 2×20 minutes (80% SDS-Hyb/20% PTw), 2×20 minutes and 2×1 hour (50% SDS-Hyb/50% PTw), 2×20 minutes (20% SDS-Hyb/80% PTw), then 2×20 minutes (100% PTw). Finally, embryos were washed 2×20 minutes in 1×PBS with 0.1% bovine serum albumin (BSA) and 0.1% Triton X-100 (PBT) before adding sheep anti-digoxigenin-AP Fab fragments (Roche), diluted 1:3000 in PBT, and incubating at 4°C overnight. The next day embryos were washed 4×20 minutes and 2×60 minutes in PBT before performing the color reaction.

For L. migratoria in situ hybridization, embryos were dissected in PBS, fixed in 3.7% formaldehyde (Sigma), held in PTw at 4°C and hybridized within 24 hours. The probe used corresponds to a 930 bp XhoI/SalI fragment from the L. migratoria cDNA clone hydrolyzed to an average length of approximately 250 nucleotides. A sense probe was used as a control. Techniques and solutions were as described in Kucharski et al. (Kucharski et al., 2000), except that hybridization was at 55°C.

A TrpE/L. migratoria Hb fusion protein was constructed by cloning a fragment, beginning at the XhoI site within the M-1 finger and continuing through to the end of the coding region, into a pATH vector (Koerner et al., 1991). Rat antisera were produced, and affinity purified, as described previously (Patel et al., 1992). A mouse monoclonal antibody was produced by immunizing with the same TrpE/L. migratoria Hb fusion protein as above. ELISA-positive clones were then screened on S. americana embryos. Both the rat antisera and mouse monoclonal work equally well in S. americana and L. migratoria. We also produced a mouse monoclonal against Drosophila Hb by immunizing with a TrpE/Drosophila Hb fusion protein and screening clones on Drosophila embryos. Embryo staining followed the protocols outlined elsewhere (Patel, 1994). The monoclonal antibody to grasshopper Hb is designated PP 7C11, while that to Drosophila Hb is PP 1G10.

Using a degenerate PCR strategy, a small region of the S. americana hb sequence was isolated from 30% embryos. This S. americana fragment was then used to isolate genomic and cDNA clones of hb from two species of grasshoppers, S. americana and L. migratoria (see Materials and Methods). The predicted protein sequence encoded by both of these genes is shown in Fig. 2. The encoded proteins are approximately 76% identical at the amino acid level, and both contain eight zinc fingers (designated NF1-2, MF1-4 and CF1-2, see Fig. 2). The Hb proteins of Caenorhabditis elegans and Helobdella triserialis (leech) also contain zinc fingers that correspond to these eight (see Fig. 3A-C,F). Alignments of the grasshopper Hb zinc fingers with those of Hb proteins from D. melanogaster, C. elegans and H. triserialis are shown in Fig. 3.

The Hb protein of D. melanogaster contains six zinc fingers, four in the middle region of the protein and two in the C-terminal region (Fig. 3F). The four grasshopper Hb zinc fingers MF1-4 and CF1-2 correspond to the four central fingers and the two C-terminal fingers of D. melanogaster, respectively (see Fig. 3B-C,F). The two grasshopper fingers NF1 and NF2, however, do not have corresponding fingers in the D. melanogaster Hb protein, although these two fingers are present in C. elegans and H. triserialis Hb (see Fig. 3A,F). In addition, our analysis of the hb sequences of C. elegans and H. triserialis reveal that they also contain an additional zinc finger (ExF) located between the MF1-4 and CF1-2 (Fig. 3D,F). Beyond their relative position however, the C. elegans and H. triserialis ExFs bear little similarity to one another (see Fig. 3D). In addition to the zinc-finger regions, there are other conserved domains defined initially by comparative studies with other insect hb sequences, as well as functional analysis in D. melanogaster (Hülskamp et. al., 1994; Wolff at el., 1995). These domains are conserved in the grasshopper sequences and include the so called A box, Basic box and C box (see Fig. 2).

Another notably conserved region between grasshopper hb and its orthologs in other insects is a putative nanos response element (NRE) within the 3′ UTR (see Fig. 3E). D. melanogaster and D. virilis hb contain two NREs in their 3′ UTRs, while the 3′UTRs of Musca domestica (house fly) and T. castaneum (beetle) hb each contain a single NRE-like sequence. Like Musca and Tribolium, both S. americana and L. migratoria contain single NRE-like sequences of 540 and 460 nucleotides, respectively, after the stop codon (Fig. 3E).

Northern analysis of L. migratoria tissue indicates that hb mRNA is expressed during both oogenesis and embryogenesis (Fig. 4A). As in Drosophila, a single transcript is detected during oogenesis, and a second transcript of different size is additionally expressed during embryogenesis (Fig. 4A). In L. migratoria, the maternal transcript is 4.2 kb in length, and the zygotic transcripts are 4.2 kb and 6.0 kb in length. RT-PCR analysis reveals that hb is also transcribed both maternally and zygotically in S. americana (data not shown). The two Drosophila hb transcripts contain the same coding regions and 3′ UTRs, and differ only in their 5′ UTR regions. This is due to the sole use of the P1 promoter during oogenesis, and use of both P1 and P2 zygotically. Our analyses of cDNA clones and RT-PCR products also suggest that the grasshopper zygotic and maternal mRNAs do not differ in their coding regions or in their 3′ UTRs. This, therefore, resembles the situation in Drosophila, although we have yet to analyze maternal versus zygotic 5′ UTR regions to determine if the two different transcript sizes are due to differences in transcription start sites.

To study the expression pattern of hb during grasshopper development, we carried out whole-mount in situ hybridization and generated both polyclonal antisera and a monoclonal antibody for immunohistochemical localization. In our descriptions below, we will simply refer to the expression patterns seen as the ‘grasshopper pattern’, given that we obtained consistent results in both L. migratoria and S. americana from 12% of development onwards. The results presented for earlier stages are for S. americana.

Consistent with our northern hybridization and RT-PCR results, in situ analysis shows that hb mRNA accumulates in developing oocytes. Grasshopper hb transcript is detectable in oocytes as they leave the germarium (Fig. 4B). By mid-oogenesis hb mRNA levels have risen (Fig. 4C), but then drop and become undetectable during later oogenesis when the oocyte nucleus begins to move posteriorly (Fig. 4D). Grasshopper hb transcript is not detected in cells within the germarium or in any somatic cells of the ovary. One important difference between Drosophila and grasshoppers is that Drosophila possesses meroistic ovaries, in which each oocyte is part of a 16 cell cyst including 15 nurse cells (which transcribe maternal factors). Grasshoppers, however, have panoistic ovaries, in which the oocyte develops in the absence of nurse cells. This means that hb mRNA observed in developing grasshopper oocytes is likely to be transcribed by the oocyte nucleus.

Surprisingly, in addition to mRNA, Hb protein is also present in the oocytes of grasshoppers: this contrasts with Drosophila oocytes, which contain hb mRNA that is not translated until after fertilization (see Discussion). In grasshoppers, Hb protein is detected in oocytes shortly after they have exited the germarium, the protein being localized to the oocyte nucleus (the germinal vesicle, Fig. 4F,G). Initially the oocyte nucleus is positioned centrally within the developing oocyte, but as the oocyte matures and slowly enlarges, the germinal vesicle comes to lie at the extreme posterior end of the egg (Fig. 4F,H-J). At about this time the oocyte begins a phase of rapid enlargement and yolk accumulation, and Hb protein leaves the oocyte nucleus and enters the surrounding cytoplasm (Fig. 4J). Shortly afterwards, germinal vesicle breakdown occurs.

In freshly laid eggs (30 minutes after egg lay; AEL) we are unable to detect hb mRNA by in situ hybridization (data not shown). However, Hb protein (presumably maternally provided), is localized to small islands of cytoplasm in the posterior 15% of the egg length, with the boundary of the domain being slightly graded (Fig. 5A,B). At this time, embryonic nuclei (energids) are dividing and moving up towards the cortex of the egg, but none have yet reached the surface. During the next few hours, these energids reach the surface of the egg and rapidly cellularize as they do so (Fig. 5C-E; for cellularization timing in the related species S. gregaria see Ho et al. (Ho et al., 1997)). At this point, we observe that some of the Hb protein in the egg cytoplasm is now within the cytoplasm of newly formed cells (Fig. 5D,E). Consistent with maternal Hb protein localization to the posterior 15% or so of the egg, only those cells forming in this posterior region contain Hb protein. This maternally provided Hb protein appears to enter the nucleus, and at about the time that this occurs we begin to detect the first accumulation of zygotic hb mRNA in the cytoplasm of these same cells (data not shown). At ∼22 hours AEL, the nuclei that are condensing at the posterior end of the egg to form the embryonic blastodisc contain Hb protein (Fig. 5F, G), while anteriorly located extra-embryonic nuclei do not (Fig. 5H,I). By 32 hours AEL, the cells at the posterior end have condensed, forming a defined circular blastodisc. At this stage, the egg is composed of two cell types: those that contain hb mRNA and protein, and are part of the blastodisc (embryonic), and those that are not part of the disc (extra-embryonic) and contain little or no hb mRNA or protein (mRNA pattern shown in Fig. 5J,K).

In Schistocerca, the blastodisc at this point is oriented more or less perpendicular to the long axis of the egg. In Locusta, the disc is slightly offset down the side of the egg with the future embryonic tail pointing anteriorly with respect to the egg. By 34-36 hours AEL (Fig. 5L,M), the disc has become asymmetric, taking on a shallow ‘V’ shape that can be observed with DAPI staining (as cells are more densely packed embryonically than extra-embryonically), and the beginning of a gastrulation furrow is visible in the center of the embryo. The future head lobes are located at the top of the shallow ‘V’ and the future posterior region of the embryo at the base of the ‘V’. At this stage, Hb protein is still expressed throughout the embryo (Fig. 5M).

Just after the embryo takes on this ‘V’ shape (∼44 hours AEL), the pattern of hb expression undergoes two major alterations: expression of hb begins in extra-embryonic cells and expression is lost from within a circular domain at the posterior end of the embryo (Fig. 6A,B). At this stage, the extra-embryonic cells have a very flattened appearance and large nuclei relative to the embryonic cells, and express markers such as zen. The zen and later hb-positive extra-embryonic nuclei go on to form the serosa (Dearden et al., 2000). The amnion, an extra-embryonic membrane that comes to lie on top of the embryo, does not express hb when the tissue becomes morphologically obvious later in Schistocerca development (by the stage shown in Fig. 6I).

Interestingly, the loss of hb expression in the posterior circular domain is first seen at the level of protein expression (Fig. 6A), and at 44 hours AEL, hb mRNA is still seen throughout the embryo (data not shown). By about 48 hours, the discrepancy between the pattern of mRNA and protein disappears. Now both hb mRNA and protein are absent from the posterior region of the embryo (protein Fig. 6C,D; mRNA Fig. 6E,F), but present throughout the rest of the embryo and in the extra-embryonic cells surrounding the embryo. The hb expression domain is clearly modulated so that expression is relatively weak in the head lobes, but strong within an arc-shaped region just posterior to them (Fig. 6C-F). The posterior hb-negative region rapidly expands, apparently through a combination of cell proliferation and rearrangement. At ∼52 hours, the posterior of the embryo starts to grow, and within the embryo, hb continues to be expressed weakly in the head lobes and strongly in an arc across the middle of the embryo (Fig. 6G,H). As the embryo grows (60 hours AEL=15% of development), this arc takes on the shape of a straight band of expression (Fig. 6I,J). This band is clearly not uniform, however, as there are two discrete sub-domains with different levels of Hb protein expression. This ‘step’ pattern at the posterior margin of the stripe is very noticeable at 17% of development (Fig. 6K,L). In situ analysis of hb mRNA shows that this step pattern of expression seen for the protein is also seen for hb mRNA (Fig. 6M,N).

The position of this hb domain was mapped at 17-20% of development by double labeling for both Hb and engrailed protein expression (Fig. 7A,B). At this stage, the engrailed stripes of T1 and T2 are just forming and the posterior boundary of hb expression is at the anterior edge of the T1 engrailed stripe. At 20% of development (Fig. 7C,D), the engrailed stripes of the gnathal segments have appeared, and reveal that the band of Hb protein extends from the posterior compartment of the mandibular (pMn) segment through to the anterior compartment of T1 (aT1), and thus spans parasegments 1-3. The more strongly stained subdomain spans pMn to pLa and the weaker subdomain is confined to the anterior compartment of T1 (Fig. 7C,D). This domain can be more precisely mapped using the pax3/7 homolog pby1 (Davis et al., 2001).

At 22% of development, a new domain of hb expression appears in the abdomen (Fig. 7E). This hb domain fades before engrailed stripes appear in this part of the abdomen, so we are not able to map its position precisely, but can place it roughly in the region of A4/A5. At about 24% of development, in situ hybridization shows that this A4/A5 hb domain is fading, and a new domain is appearing in a more posterior part of the abdomen, around A7/A9 (Fig. 7F). By 26% of development, the A4/A5 domain has disappeared, and the posterior (A7/A9) abdominal domain is intensifying (Fig. 7G). At this time, the A6 engrailed stripe is just appearing, and the second abdominal hb domain is clearly posterior to this. At 27% of development (Fig. 7I; engrailed in black and Hb in brown), the A7 engrailed stripe appears, and now we can see that this posterior hb domain begins at the anterior edge of this engrailed stripe, and thus has its anterior boundary in the posterior compartment of A7. This posterior hb expression domain fades before the next engrailed stripe forms, so we cannot precisely place its posterior boundary, but it is roughly within the anterior compartment of A9 (data not shown).

Several non-ectodermal patterns of hb expression are also visible beginning at about 20% of development. These include two phases of mesodermal expression. First, within the gap domains (Mn-T1, A4-A5 and A7-A9), hb is expressed in the underlying mesoderm at the same time that is expressed in the ectoderm. Second, hb is later expressed in the mesoderm of every segment at about the time that this germ layer condenses to take on a wedge shaped configuration within each segment. This occurs at roughly the same time that engrailed expression begins in the ectoderm of these segments (Fig. 7G,H, Fig. 8E). Mesodermal condensation causes hb to appear to form a striped pattern with almost all of the mesodermal wedge located under the anterior compartment ectoderm (Fig. 8E). As the mesoderm migrates laterally, expression is eliminated from some mesodermal cells. In the abdomen, expression remains in the more distal cells within each body somite (Fig. 8F). In the gnathal and thoracic segments, expression is retained in the more proximal cells (Fig. 8G), which eventually form a ring of staining at the base of each appendage (Fig. 8H,I,K). At 42% of development, expression of hb can be seen in the nuclei of longitudinal muscles 3 and 4, and what appear to be muscles at the base of the limbs (Fig. 8J,M)

As in Drosophila, there is also dynamic expression of hb in the nervous system (Fig. 1B,C,E,F, Fig. 8A-D). In grasshoppers, hb expression is seen throughout the neuroepithelium just before neuroblast delamination begins. Higher levels of Hb protein accumulate in the cells that will be the neuroblasts (Fig. 8A). Once delamination occurs, all neuroblasts appear to express hb at least transiently (Fig. 8B). As these neuroblasts begin to produce ganglion mother cell (GMC) progeny, expression is lost, but retained in the GMCs. This is clearly illustrated by following expression in neuroblasts 6-1 and 6-2. At 27% of development, NB 6-1 and 6-2 express both En and Hb proteins (Fig. 8C). By 29% of development, these neuroblasts express engrailed but not Hb, although the latter is detected in their initial GMC progeny (Fig. 8D). Thus, as in Drosophila, it appears that hb expression is transiently seen in all neuroblasts, but subsequently restricted to the first one or two of their GMC progeny. By 42% of development, Hb protein expression is seen in a number of neurons, with the overall ‘bowtie-shaped’ pattern of staining in each neuromere being quite similar to that seen in Drosophila at stage 15 (compare Fig. 8J,L with Fig. 1F). By 45% of development, Hb protein is found in a small set of neurons including aCC and pCC (Fig. 8N).

Expression of hb is also seen in a small number of peripheral neurons, most notably in specific neurons that form within the developing appendages. At 26% of development, the first hb positive neural precursors appear in the epithelium of the limbs (Fig. 8G). By 29%, these neurons have delaminated into the limb (Fig. 8H), and by 42% of development, several neurons are clearly visible within the gnathal and thoracic appendages.

In order to further understand the early patterning of grasshopper embryos and make inferences about the ancestral role of hunchback (hb) in the insects, we have isolated and analyzed the expression pattern of the S. americana and L. migratoria hb orthologs. The comparison of the grasshopper hb sequences with previously isolated orthologs allows us to propose an ancestral structure for the Hb protein. Examination of the expression pattern of the grasshopper hb gene has raised questions about its regulation and function in these insects, as well as its ancestral role in arthropods.

Alignment of the Hb proteins highlights a highly stereotypical arrangement of putative zinc fingers that is apparently conserved across phyla (Fig. 3). Not only is the clustering pattern of the metal binding fingers conserved, but the spacing of structural residues within homologous fingers appears to be consistent across phyla (Fig. 3). From the evolutionary distribution of zinc fingers, we can infer that the common ancestor of the annelids, nematodes and arthropods possessed a hb gene encoding at least eight zinc fingers. Although the M fingers are known to be capable of binding DNA, the function of the N- and C-terminal zinc-finger domains remains unclear: they may, for example, be important for the recruitment of transcriptional co-factors. Whatever the case, the N terminal fingers are not found in the fly and the flour beetle (Fig. 3F), and may thus play some ancestral role that has been lost in the lineage leading to Drosophila and Tribolium. Other regions of conservation such as the A, B and basic domain are present in both Drosophila and grasshoppers (Fig. 2), and may play a conserved role in interactions with co-factors such as dMi-2 (Kehle et al., 1998).

Further evidence for a conserved role for the hb gene comes from the neural expression pattern in grasshopper. Grasshopper hb is transiently expressed in neuroblasts, and in a specific subset of GMCs and neurons, just as it is in Drosophila (Goodman and Doe, 1993; Kambadur et al., 1998; Tautz et al., 1987). The conservation of multiple protein domains across species, together with the highly conserved neural expression domains, shows that we have isolated true hb orthologs from L. migratoria and S. americana.

Although we have not identified a clear vertebrate hb ortholog in the databases, the family of Ikaros/Helios zinc-finger proteins have a similar structure to hb (Georgopoulos et al., 1992). In particular, they contain clusters of zinc fingers analogous to the M and C fingers in the grasshopper Hb protein. Interestingly, Ikaros feeds into gene regulation through chromatin remodeling complexes and interacts with the mouse ortholog of dMi-2 (Kim et al., 1999; Sun et al., 1996). This is reminiscent of the interaction of Drosophila Hb protein with dMi-2 and subsequent maintenance of Hox gene repression through the action of the polycomb group genes.

The expression pattern of grasshopper hb allows us to make inferences about potential functions of the gene, as well as the transcriptional and post-transcriptional regulation of hb expression, both maternally and zygotically.

In grasshoppers, maternal hb is translated during oogenesis, and the protein initially accumulates in the oocyte nucleus (Fig. 4F-I). The nucleus is centrally placed during early stages of oogenesis, but moves posteriorly later in oocyte development (Fig. 4F). The Hb protein then appears to leave the nucleus, but remains in the posterior 15% of the egg (Fig. 4J, Fig. 5A). This expression pattern suggests two novelties in our understanding of the regulation of grasshopper maternal hb. First, in contrast to Drosophila (see Tautz, 1988), hb transcript is translated during grasshopper oogenesis and maternally provided in the form of protein. This indicates that maternal Drosophila hb may be regulated by a translational control mechanism, so that the protein is not produced during oogenesis or very early embryogenesis (Wreden et al., 1997), and that Hb protein might play a novel function during grasshopper oogenesis or early embryogenesis. Second, spatial localization of maternal grasshopper Hb protein may be mediated by the oocyte nucleus.

This pattern of maternal Hb protein expression is particularly interesting because the position of the oocyte nucleus in short germ insects often correlates with the position where the germ anlagen (embryonic blastodisc) forms. Furthermore, irradiation of very early embryonic nuclei at what would normally be the position of the germ anlagen leads to the destruction of these nuclei. Adjacent nuclei, which would normally have taken on an extra-embryonic fate, enter the gap and go on to form a normal embryonic primordium (Patel, 2000). Such experiments imply that a non-nucleic acid factor is localized at the site occupied by the oocyte nucleus during oogenesis and that this factor plays an instructive role in determining the position at which the embryonic primordium will form. We propose that this factor may be maternally provided Hb protein. In this model, the position of the grasshopper oocyte nucleus at the end of oogenesis would specify the embryonic primordium by leaving Hb protein at a specific location. This maternally provided Hb protein is subsequently incorporated into cells that form in the posterior 15% of the egg. Relevant to this discussion is the observation that maternal grasshopper zen protein is initially found in the embryonic primordium, but becomes restricted to extra-embryonic cells (Dearden et al., 2000). Thus, early nuclei that express both zen and Hb proteins may be fated as embryonic, while those that express only zen may be fated to become extra-embryonic.

From about 44 to 96 hours AEL, hb expression in the early grasshopper embryo resembles the early fly pattern in that it is found throughout the anterior (head and anterior thorax) part of the embryo. Grasshopper hb is not expressed in the region of the embryo that will eventually contribute to the posterior thorax and abdomen. This suggests a conserved role for hb in setting up the initial embryonic axis. However, while grasshopper eggs inherit maternal Hb protein, we have been unable to detect maternal transcript in newly laid eggs. The hb expression seen during this 44-96 hour period would seem to be mainly derived from zygotically transcribed hb. This suggests that maternally supplied hb may not play a role in AP patterning, and that this role is played by zygotic hb. This contrasts with Drosophila where both maternal and zygotic hb participate in axis formation. In grasshoppers then, the importance of maternal hb may be to help distinguish between embryonic and extra-embryonic fate.

A zygotic patterning bias during axis formation outside of the Dipterans is also suggested by genetic evidence from Hymenoptera. The existence of a protein carrying out a hb like role in A/P patterning of the wasp Nasonia vitripennis has been suggested by the close resemblance of the headless phenotype to that of Drosophila hb (Pultz et al., 1999). The headless phenotype, though zygotic, is stronger than that of the zygotic Drosophila hb mutant. As with our evidence from grasshopper, this suggests that although maternal AP patterning mechanisms exist in multiple insect orders, flies show an extreme bias towards maternal patterning.

At about 44 hours AEL (Fig. 5M), presence of hb transcript but lack of protein in cells of the posterior growth zone, implies translational control. The downregulation of hb in the posterior embryonic primordium may possibly be instigated by a factor such as nanos, as in Drosophila. Indeed, the sequences of the grasshopper hb 3′UTRs contain a putative nanos response element (NRE), the sequence required to recruit translational control proteins to the Drosophila hb transcript (Fig. 3E; Wharton and Struhl, 1991). By 17%, regulation probably occurs at the transcriptional level. Such regulation may either involve repression by a protein such as Krüppel, or the decreased concentration of a transcriptional activator (just as bicoid concentration regulates Drosophila hb transcription) (Driever and Nüsslein-Volhard, 1989; Jäckle et al., 1986; Struhl et al., 1989). The posterior boundary of the weak Hb sub-domain in T1 would not overlap with Krüppel, if the location of the latter matches its exact position in flies (PS4 to PS7; Harding and Levine, 1988; Knipple et al., 1985). Thus, it is possible that the anterior Krüppel boundary is set via repression by grasshopper hb. Alternatively, Krüppel may overlap with hb and be responsible for the weak hb subdomain in T1 region (see above). These scenarios are reminiscent of the cross-regulatory interactions between Krüppel and hb in Drosophila. Cloning of grasshopper Krüppel and analysis of its expression, as well as analysis of the hb promoter, will shed light on the transcriptional control of hb.

The presence and positioning of the gnathal/thoracic domain of grasshopper hb suggests that the functions of this domain, which in Drosophila include control of gap, pair-rule and Hox genes, may be conserved in grasshopper. Anterior hb expression spans the gnathal segments, and seems well positioned for the control of possible head gap genes. The hb regulation of pair-rule stripes may also be conserved in the grasshopper although eve, for example, is not expressed in a pair-rule pattern in the grasshopper but rather in a posterior domain (Patel et al., 1992). This fact does not preclude hb regulation of the anterior eve border early in development, allowing for possible conservation of regulatory relationships in the face of clearly divergent spatial pattern. Grasshopper pby1 (Davis et al., 2001), a Pax 3/7 family member related to the Drosophila pair-rule gene paired, is expressed in a pair-rule pattern that may be regulated by hb expression, thus providing a pair-rule link between the gap patterning of hb and the segmental patterns of engrailed and wingless.

As in Drosophila, grasshopper hb may also be involved in Hox gene regulation, for example, by acting as a repressor to prevent anterior Ubx expression. However, the posterior boundary of grasshopper hb is shifted one segment anteriorly relative to Drosophila, while Ubx protein is expressed with the same anterior border in both insects (Kelsh et al., 1994). This highlights the absence of PS4 hb expression in grasshopper, and suggests the possibility of an additional level of Ubx regulation in this insect. It is also important to note that in Tribolium, Ubx transcripts appear more anteriorly than Ubx protein (Bennett et al., 1999). If this is also the case in the grasshopper, then the direct transcriptional control of Ubx by Hb protein may be conserved across insects, but with subsequent translational control setting the Ubx protein boundary in some species.

Some domains of grasshopper hb expression, such as the gnathal/thoracic, A7-A9 gap domains and CNS patterns, are presumably homologous to those observed in Drosophila, and thus probably ancestral. Other domains such as the A4-5 gap domain, and serosal, PNS and mesodermal expression domains are less obviously so.

The expression of hb in extra-embryonic primordia seems conserved in many species outside the higher flies, and it has even been suggested that this domain has been lost in Drosophila as a result of fusion and reduction of the amnion and serosa (Rohr et al., 1999). The reduction of extra-embryonic material and its movement into the DV axis of the egg maps phylogenetically to higher dipterans (the cyclorhapphan flies), and has been proposed to reflect profound changes in the specification of anterior tissue (Schmidt-Ott, 2000; Stauber et al., 1999). The cumulative data from insects suggests that serosal hb expression is ancestral, especially as there is extra-embryonic expression in D. melanogaster (Fig. 1C,D). Thus, it is formally possible that hb is involved in serosal patterning, whichever body axis this tissue or its evolutionary derivatives lie in.

The grasshopper hb mesodermal stripes at first sight seem unlike any domains observed in other species. The striped appearance of mesodermal hb is actually due to compaction of this tissue during morphological segmentation. Segmental stripes have also been observed in the trunk of Musca domestica and Tribolium castaneum (Sommer and Tautz, 1991; Wolff et al., 1995). If hb stripes in these insects are also found mesodermally, then mesodermal hb expression may be ancestral, the striped appearance being simply due to mesodermal condensation. Such early mesodermal expression has not been seen in Drosophila, but hb has recently been shown to be expressed later in specific mesodermal cells that play a role in tracheal guidance in D. melanogaster (Wolf and Schuh, 2000). Thus, while early hb expression in the mesoderm may differ between grasshoppers and flies, it may have a conserved expression in mesoderm later in development, and play a role in events such as tracheal guidance.

We have found that expression of hb in varying concentrations across the gnathal/T1 primordium is conserved through to basal insects such as the grasshopper. This suggests that hb may behave morphogenetically, despite lack of a prolonged syncytial stage in the grasshopper. Grasshopper hb does not, however, form a smooth gradient, as might arise by protein diffusion, but instead is expressed in a distinct stepped pattern that is seen at both the protein and mRNA level. This implies that morphogenetic activity may be set up or maintained by a fundamentally different mechanism from the bicoid diffusion found in Drosophila. Orthologs of hb have been cloned from annelids, nematodes and a mollusk (Fay et al., 1999; Savage and Shankland, 1996; Sommer et al., 1992), and expression patterns have been described in C. elegans and the leech H. triserialis (Fay et al., 1999; Iwasa et al., 2000; Savage and Shankland, 1996). In neither organism is an early A/P gradient observed (although the C. elegans data are from a GFP construct and not from observation of endogenous mRNA or protein) (Fay et al., 1999), suggesting that hb is not required for anteroposterior patterning in C. elegans and H. triserialis. These data suggest that hb may have been co-opted into a role in anterior patterning within the arthropods, or that such a role has been lost in the lineages leading to C. elegans and H. triserialis.

Cumulative evidence from insects seems to indicate that the role of hb as a gap gene is highly conserved. Thus, despite the fact that the expression of pair-rule orthologs may vary across insects, expression of an earlier component of the segmentation hierarchy is conserved. However, it is less certain that maternal hb is vital to patterning of the grasshopper anteroposterior axis. Instead, this role may be mediated by the regulation of zygotic hb, suggesting a bias toward zygotic axial patterning in more basal insects. Nevertheless, the maternal Hb protein product in grasshoppers may play a role in determining embryonic from extra-embryonic cells within the early egg.

We thank David Bentley for access to the Berkeley grasshopper colony, Bridget Lear for her unwavering commitment to keeping our Chicago grasshopper colony flourishing, and Greg Davis and Bridget Lear for helpful discussions and critical reading of the manuscript. We also thank Peter Dearden and Michael Akam for communication of unpublished results and for providing their timeline of early embryonic development in S. gregaria, Corey Goodman, David Bentley and Joyce West for help and support in the early stages of this project, Danny Brower for playing devil’s advocate on Locusta development, and Kirsten Balding for help with sequencing. N. H. P. is an HHMI investigator, and S. L. is supported by an EMBO fellowship.