Conserved and divergent expression aspects of the Drosophila segmentation gene hunchback in the short germ band embryo of the flour beetle Tribolium

Drosophila segmentation is initiated by a cascade of inter-acting transcription factors that are provided maternally and zygotically in the early embryo (Pankratz and Jäckle, 1993). Homologues of these transcription factors can often be found in distantly related species (Tautz and Sommer, 1994). On the basis of the well known function of these transcription factors in Drosophila, it is possible to compare their expression patterns and to infer how far the segmentation pathways are conserved at the molecular level. Comparisons between different insect species are currently of special interest in this respect. When compared to other insects, Drosophila development represents a highly derived form of embryogenesis. In Drosophila, all segments are determined at the syncytial blastoderm stage, before cellularization. This allows the transcription factors to diffuse freely between adjacent nuclei and to excert their function by forming diffusion-controlled gradients. This situation is different for other forms of insect embryogenesis. In the more ancestral species, only the anterior pattern elements are determined at blastoderm stage, while most or all of the abdominal segments are generated in a secondary growth process (Anderson, 1972; Sander, 1976). The embryos of the flour beetle Tribolium castaneum represent such a form of short germ embryogenesis. Though the holometabolous beetles are still more closely related to the Diptera than some older insect orders, it can be argued that the Tribolium embryo may nonetheless represent the archetypical form of insect embryogenesis (Anderson, 1973; Tautz et al., 1994). Tribolium has thus become the focus of a number of com-parative studies in molecular embryology (Sommer and Tautz, 1993, 1994; Brown et al., 1994a,b; Patel et al., 1994; Nagy and Carroll, 1994). These studies have indicated that the segmentation gene hierarchy, as it is known from Drosophila, is much more conserved in these different types of embryos than one would have previously assumed.

We describe here the cloning and the expression pattern of a key gene in the segmentation gene hierachy, namely hb. Drosophila hb codes for a zinc-finger type transcription factor (Tautz et al., 1987) which forms an early morpho-genetic gradient in the embryo that is required to regulate the abdominal gap genes (Hülskamp et al., 1990; Struhl et al., 1992; Schulz and Tautz, 1994, 1995). hb expression is provided maternally and zygotically. The maternal RNA is distributed homogeneously in the embryo, but is under the translational control of the posterior maternal factor nanos (nos). Since NOS protein is distributed in a gradient from posterior to anterior (Gavis and Lehmann, 1992), this trans-lational control results in an opposing gradient of HB protein distribution in the abdominal region of the embryo (Tautz, 1988). The zygotic expression of hb is under the transcrip-tional control of the anterior maternal gene bicoid (bcd), which forms an anterior-posterior gradient of protein distrib-ution (Driever and Nüsslein-Volhard, 1988). This gradient directly activates hb transcription up to a certain threshold concentration of BCD protein (Driever and Nüsslein-Volhard, 1989) whereby hb itself appears to be synergisti-cally involved in this process (Simpson-Brose et al., 1994). Thus hb becomes homogeneously expressed in the whole anterior half of the embryo (Tautz, 1988). HB protein diffuses from the border of this expression domain posteriorly and thus forms a gradient in the abdominal region. It was shown that this zygotic gradient can functionally replace the mater-nally provided gradient (Lehmann and Nüsslein-Volhard, 1987; Hülskamp et al., 1989; Irish et al., 1989; Struhl, 1989). At later stages, there are additional expression domains of hb at blastoderm stage (Tautz et al., 1987; Schröder et al., 1988; Tautz and Pfeifle, 1989). The anterior expression is replaced by two stripes of expression that are not under the direct control of the bcd gradient. One of these stripes, the ‘paraseg-ment 4 stripe’ is under the regulatory control of hb and Krüppel (Kr) (Hülskamp et al., 1994). A posterior expression domain of hb is seen forming first a cap and then a stripe covering parasegments 13/14. This posterior expression domain is under the regulatory control of the terminal gap genes (Casanova, 1990).

We show here that most of this expression pattern of hb is conserved in Tribolium, suggesting that the functional role of hb may also be conserved between Drosophila and Tribolium. However, we describe one novel expression of hb in the prospective serosa cells of the Tribolium embryo that might play a role in determining between extra-embryonic and embryonic tissues. This distinction appears to be one of the first differentiation decisions that is made in the Tribolium embryo.

The first Tribolium hb clone was obtained by PCR with degenerate primers in the Zn-finger region as described by Sommer et al. (1992). The PCR fragment was used to screen a genomic library (a gift from S. Brown; Brown et al., 1990). About 7.5 kb of the genomic clone was sequenced using a combination of shot gun and primer walking procedures. To obtain information on the transcript structure, we have constructed an early cDNA library from 0-to 96-hour embryonic RNA in lambda ZAP, using the cDNA synthesis kit from Stratagene (Heidelberg). Several different types of cDNAs were recovered from this library and partially or fully sequenced. The beginning and the end of the different cDNAs are marked in Fig. 1.

To assess which of the different cDNA types are expressed at which time, we have performed RT-PCR experiments using the superscript kit (Life Technologies, Eggenstein) with staged RNA preparations (0-4 and 0-72 hour). The 0-to 4-hour RNA preparation represents the maternal RNA and was staged particularly carefully. To control the staging, a fraction of the eggs was stained with Hoechst 33342 to confirm that only embryos with a maximum of four nuclei were included. The first strand cDNA synthesis was done with a specific internal primer (CW856: 5′-AGAGTGCTGTAAATCCTG-3′, pos. 5638-5520). Aliquots of this cDNA were subjected to PCR using the CW856 primer as well as the following exon specific primers: exon 1 -CW859 (5′-TAAATGATGCGATGGGCG-3′), exon 2 -CW858 (5′-GTGCAAAAATTTCGAACAG-3′; pos. 3088-3106) and exon 3 -CW857 (5′-GTTCGAGCGGGGACGTAG-3′; pos. 5093-5110). The resulting fragments were resolved on agarose gels, blotted and hybridized with a hb cDNA probe that included all exons.

A HincII fragment spanning the amino acid positions 194-524 was expressed in E. coli using the pET vector system (Rosenberg et al., 1987). The protein was partially purified by gel electrophoresis and used to immunize a rabbit. The antiserum was affinity purified against the same protein fragment.

Fixation and staining of the embryos with antibodies or whole-mount in situ hybridization was done using procedures similar to those used for Drosophila (Tautz and Pfeifle, 1989) and employing RNA probes according to the protocol of Klingler and Gergen (1993). In situ double staining was done with digoxigenin- and fluorescein-labelled RNA probes using a similar protocol to that described by Hauptmann and Gerster (1994). All embryos were counterstained with Hoechst 33342 before mounting in glycerol to visualize the distribution of the nuclei in the early embryos. Photographs were taken on a Zeiss axioplan microscope with Kodak Ektachrome 320T film and the slide images were transferred to Kodak photo CD. The final images were edited electronically with Photoshop (Adobe) to balance the sizes of the embryos and the colours, as well as to remove background spots.

Primers from the zinc finger region of the hb gene were employed to amplify and clone a short fragment from Tribolium genomic DNA (Sommer et al., 1992). This fragment was then used to screen genomic and cDNA libraries. One genomic and several cDNA clones were recovered. The coding region as well as a large part of the upstream region was sequenced (Fig. 1). Comparison between the cDNA and the genomic sequences showed that the leaders of two different cDNA types reside within the genomic clone, while the leader of a third cDNA type must lie outside the cloned genomic region (Fig. 2). A potential TATA-box and a consensus transcription start site were found only for the P2 transcript (Fig. 1). Only one large open reading frame is evident which is interrupted by a short intron. The different leaders of the three cDNAs have a common acceptor splice site upstream of the AUG that initiates the open reading frame. The conceptually translated protein encodes the zinc finger domain that was already amplified by PCR and in addition a second zinc finger domain at the C terminus, as is typical for hb in other species (Treier et al., 1989; Sommer, 1992). An additional region of similarity at the amino acid level can be found downstream of the first finger domain (Fig. 1). Furthermore, a specific comparative search with the subregions that are conserved among dipteran hb genes (Treier et al., 1989; Sommer, 1992) reveals several additional very short conserved stretches (Fig. 1). One of these is partially encoded in the first coding exon, suggesting that this exon is indeed translated, though some splicing variants exist that exclude this exon (see below). Together, the amino acid similarities in the Zn-finger domains, as well as in the previously identified A-, C- and D-Boxes (Hülskamp et al., 1994) are sufficient to establish a likely homology of the cloned gene with the Drosophila hb gene.

A search for sequence regions conserved between the upstream regions of hb in Drosophila and Tribolium by dot plot comparisons did not yield any clear results. It was previ-ously shown that some highly conserved regions can be detected in the upstream region between the D. melanogaster and D. virilis hb genes (Treier et al., 1989; Lukowitz et al., 1994). Given that the evolutionary time of separation between Drosophila and Tribolium is only about four times longer, one might have expected that at least some of these motifs are still conserved, but even specific searches for these motifs were unsuccessful. In spite of this, we found that some regulatory elements of the Tribolium hb upstream region are still functionally recognized in Drosophila (Wolff et al., unpublished data).

The recovery of different types of cDNA clones suggested that there are different transcripts from the gene. To study these further and to analyse whether at least one is tran-scribed maternally, we have employed PCR experiments with reverse transcribed RNA from maternal and zygotic stages of embryoge-nesis. Three primer pairs were used which can differ-entiate between the three different transcripts (Fig. 2). The results show that the RNAs from two of the promoters, namely the most distal one (P1) and the proximal one (P3) are already transcribed mater-nally, while the promoter between these (P2) is only zygotically active. Half of the P3 transcripts still contain a small intron, while a fraction of the P1 and P2 transcripts lack the first coding exon (Fig. 2). Though this transcription pattern is more complicated than in Drosophila, it is nonetheless reminiscent of the situation there. In Drosophila, the distal P1 promoter yields both maternal and zygotic tran-scripts, while the proximal P2 promoter yields only early zygotic transcripts (Tautz et al., 1987; Schröder et al., 1988; Treier et al., 1989; Lukowitz et al., 1994). The equivalent of an even more proximal P3 promoter was not found in Drosophila.

To study the spatial and temporal expression of hb in Tribolium, we have utilized whole-mount in situ hybridization. In these experiments, early embryos always stain strongly and homogeneously with the hb probe (not shown). It seems likely therefore that the maternal hb mRNA that was identified in the RT-PCR experiments is distributed homogeneously in the early embryo, as is the maternal hb RNA in Drosophila (Tautz and Pfeifle, 1989). In Drosophila, the translation of this RNA is under the negative regulatory control of the posterior maternal system and HB protein from this RNA is therefore expressed mainly in the anterior half of the embryo, while it forms a gradient extending into the posterior half (Tautz, 1988). To study the protein distribution of hb in Tribolium, we have raised an antiserum against a large part of the coding portion of the gene. This antiserum was affinity purified and then used to stain embryos. The HB protein is first detected during the last nuclear division cycles shortly before blasto-derm stage. All nuclei are homogeneously stained at this stage, only those that are in mitosis show no staining (Fig. 3A-C). Shortly after the completion of the last nuclear division cycle, the posterior terminus becomes cleared from protein expression (Fig. 3D-G). At this stage, the RNA is still present in this area (Fig. 4A), suggesting that it is translational control that causes this effect. This suggests that the maternal hb RNA in Tribolium may also be under the regulatory control of a posterior factor. Indeed, a short nucleotide motif that is remi-niscent of the nanos response elements (NRE) that mediate the translational control in Drosophila (Wharton and Struhl, 1991; Murata and Wharton, 1995) can be found in the 3′-region of the mRNA (Fig. 1). Furthermore, functional homologs of nanos were recovered from very diverse dipteran species, indi-cating that its function is evolutionarily conserved (Curtis et al., 1995). The fact that at early stages all nuclei express HB protein might be explained by the fact that nanos itself is under translational regulatory control (Gavis and Lehmann, 1994) and that the first NOS protein is produced only somewhat later in Tribolium, albeit still before the onset of the first zygotic expression (see below).

The first differential distribution of transcripts can be seen at the early blastoderm stage, after the posterior pole has become cleared of HB protein (see above). At this stage, one observes an anterior cap and a large posterior domain (Fig. 4A). It seems likely that these two domains are the result of a zygotic expression, since their level of expression is higher than that of the maternal RNA expression. The anterior cap expands dorsally towards the posterior at later stages (Fig. 4B), while the posterior domain recedes from the posterior tip and even-tually forms a broad band in the central region of the embryo (Fig. 4B). This band splits into two subdomains at later stages (Fig. 4C). Concomitant with the dorsal expansion of the anterior domain, the posterior one recedes to the ventral side of the embryo (Fig. 4B,C).

These apparent movements of the expression domains reflect the movements of the nuclei in the early embryo. The early germ anlage forms ventrally in the posterior two thirds of the egg. Accordingly, the embryonic nuclei move into this region, mainly from the dorsal side, but partly also from the anterior region. This movement is compensated by the movement of the serosa nuclei that migrate dorsally from anterior to posterior. The serosa nuclei become enlarged (probably due to polyploidization) and are more scattered at later stages, which makes them clearly distinguishable from the embryonic nuclei (Fig. 5B,C). The border of the anterior hb domain conicides exactly with the border between these prospective extra-embryonic and embryonic fields. This border becomes established very early and quickly forms a very sharp boundary, as is evident both from RNA and protein distribu-tion patterns (Fig. 5A-C). None of the other expression domains of segmentation genes that have so far been studied in the early Tribolium embryo show such a distinct border at such an early stage. This suggests that the developmental decision for the committment of extra-embryonic versus embryonic tissue is made at a very early stage of embryogen-esis, possibly even before any subdivision of the anterior-posterior axis becomes established. hb continues to be expressed in the extra-embryonic cells until very late in devel-opment. RNA expression is seen exactly at the rim of the closing serosa (Fig. 4D) and protein expression is seen in all serosa cells almost until the end of embryogenesis (Fig. 5D). Interestingly, since hb RNA seems to be absent in the serosa at these later stages (Fig. 4E), one has to conclude that the HB protein in these cells has a much longer halflife than in other tissues where the protein staining pattern always matches the RNA staining pattern very closely. Given the very early and also persistent expression of hb in the serosa cells, one could speculate that hb is causally involved in the formation of these cells.

On the basis of these considerations, it is evident that the anterior hb expression domain in Tribolium cannot be homol-ogous to the anterior hb expression domain in Drosophila, since extra-embryonic cells are not specified at such an early stage in Drosophila. Instead, it appears that the posterior hb expression domain in Tribolium must be homologous to the anterior domain in Drosophila. This is evident from the fact that it is expressed in the prospective head and thoracic region in the developing embryo (Fig. 4C,E). Furthermore, it lies in the same region where the anterior pair rule and segment polarity stripes are formed in Tribolium (Sommer and Tautz, 1993; Patel et al., 1994; Nagy and Carroll, 1994; Brown et al., 1994a,b) and anteriorly adjacent to the expression domain of Tribolium Krüppel (Sommer and Tautz, 1993), similar to the situation in Drosophila.

The posterior blastoderm expression domain of hb can still be seen during gastrulation and in the early germ band (Fig. 4E). At this stage it becomes clear that it is expressed in the region of the prospective gnathal and first thoracic segments. In Tribolium, double staining with the hairy probe (Fig. 6) shows that the more posterior hb stripe overlaps with the second hairy stripe, which demarcates PS 2/3 in Drosophila. The border of the early zygotic expression of hb in Drosophila is in a similar position (Tautz et al., 1987), though its exact determination is difficult since hb forms a short range gradient at this stage. Fur-thermore, a secondary blastoderm expression stripe of hb occurs in PS4 a little later, overlapping the border of the previous domain (Tautz et al., 1987; Schröder et al., 1988; Hülskamp et al., 1994). Thus, within the limits of reasonable resolution, we conclude that the posterior border of the Tribolium hb domain appears to be at a homologous position to that of the early Drosophila hb domain.

Halfway through germ band extension, additional expression patterns are seen. At first there is a very weak expression that appears to occur segmentally (Fig. 4F,G). Such an expression pattern is not known from Drosophila. However, intriguingly, it occurs very similarly in the distantly related dipteran Musca domestica (Sommer and Tautz, 1991) as well as in the lepidopteran Manduca sexta (Kraft and Jäckle, 1994). Another expression of hb is visible at the posterior end at this stage (Fig. 4F-I). This expression of hb is reminiscent of the posterior hb expression seen in Drosophila at blastoderm stage. In Drosophila the posterior hb domain is located in the region of PS 12-13 and is required for the formation of abdominal segments A7 and A8 (Tautz et al., 1987; Lehmann and Nüsslein-Volhard, 1987; Bender et al., 1987). To map the Tribolium domain more precisely, we have again used double staining with the Tribolium hairy probe. The results show that the anterior border of this hb domain is posterior to hairy stripe no. 6 and overlaps with hairy stripe no. 7 (Fig. 6D), which is in line with the situation in Drosophila. Tribolium forms two more abdominal segments than Drosophila and accordingly, an eighth hairy stripe develops. This stripe lies within the posterior hb domain (Fig. 6E,F). On the basis of these results we suggest that the posterior hb domain expression in Drosophila and Tribolium is at a homologous position.

In Drosophila, hb is expressed also in the developing nervous system, most notably in the early neuroblasts and later in the CNS. A similar expression of hb can be detected in Tribolium (Fig. 4H-K), though a more detailed comparison with the Drosophila expression needs still to be done.

Morphologically, Drosophila and Tribolium represent two very different types of early development, namely long and short germ band embryogenesis. While the early pattern formation process in long germ band embryos occurs essen-tially in a syncitial situation, patterning in short germ band embryos occurs at least partially in a cellular environment. This had raised the question of whether the same molecular mechanisms could govern early development, or whether there would be substantial adaptations to the special mode of syncitial development seen in Drosophila. However, the analysis of the expression of the gap gene Krüppel as well as of pair-rule gene and segment-polarity gene homologues in Tribolium has suggested that the molecular basis of early development might not be so different after all (reviewed by Tautz and Sommer, 1994). The expression analysis of the gap gene hb now provides additional evidence suggesting that the early molecular aspects of development are indeed much more conserved than the morphological comparisons had suggested. The first conserved aspect relates to the maternal expression of hb. Both in Tribolium and in Drosophila there is a posterior factor that excerts a translational control on the early hb RNA which is initially homogeneously distributed in the egg. In Drosophila, this factor is nos, in conjunction with pumilio and an as yet unidentified protein (Wang and Lehmann, 1991; Gavis and Lehmann, 1992; Murata and Wharton, 1995). In Tribolium we find a conserved motif in the 3′ end of the mRNA where these gene products bind (Murata and Wharton, 1995). It is therefore reasonable to assume that the same factors are also active in Tribolium, even though homologues have not yet been identified in this species.

The earliest zygotic expression of hb in Tribolium consists of two separate domains, one demarcating the region of the prospective serosa cells and one in the region of the develop-ing head and thoracic segments. This latter domain can be con-sidered to represent a conserved expression between Drosophila and Tribolium, despite its different location within the egg. In Drosophila, the equivalent expression domain covers the whole anterior half of the embryo, while in Tribolium it is expressed in the posterior two thirds. Still, it seems possible that both domains could be under a similar reg-ulatory maternal control. In Drosophila, hb is activated by the anterior BCD protein gradient, its posterior border being set by a threshold concentration below which the BCD protein cannot function as an activator. Since such a posterior boundary is not evident at the earliest stages in Tribolium, one would have to postulate that an activating effect by a potential Trib-bcd homologue would at first be active throughout the embryo. This would still leave unexplained how the anterior border of the posterior hb domain is determined. Either the bcd gradient alone, or in conjunction with a factor from the terminal system might be responsible for this. Such an interaction between bcd and the terminal system exists in Drosophila (Ronchi et al., 1993). However, a completely different regulatory control of this anterior border seems still possible.

The regulatory control of the posterior border of the zygotic hb domain might be more relevant for the segmentation process. A short range gradient of hb emanating from this border controls the expression of the more posterior gap genes in Drosophila (Hülskamp and Tautz, 1991). In Tribolium, the maternal translational control (see above) produces such a protein gradient. Interestingly, the zygotic RNA expression domain, which is seen a little later, extends at first to the posterior tip of the embryo. However, since the posterior pole remains free of HB protein even at this stage, we have to conclude that both the maternal and the zygotic RNAs are under the negative translational control of the posterior maternal factors. This is in line with the fact that both types of RNA have the same 3′ untranslated region and thus both carry the NRE motif. The apparent retraction of the zygotic domain from the pole at later stages might then be solely due to a reduced stability of the untranslated RNA. However, an addi-tional regulatory input from the terminal system might also exist.

Independently of how this posterior border of hb expression might be regulated, we note that it lies eventually at a similar position to that in Drosophila, at least when compared on the basis of the expression in the respective segment anlagen. Most importantly, it lies anteriorly to the expression domain of Tribolium Krüppel (Sommer and Tautz, 1993) and could thus be involved in its regulation in a similar manner to that in Drosophila, namely by repressing Krüppel at high concentra-tions and activating it at lower concentrations (Hülskamp et al., 1990; Struhl et al., 1992; Schulz and Tautz, 1994).

The most anterior zygotic hb domain does not seem to relate to any hb expression domain in Drosophila. This may be explained by the fact that it demarcates a structure that is not present in Drosophila, namely the serosa. The serosa is an extra-embryonic membrane that eventually surrounds the whole developing embryo and that is not directly connected to it at later stages. This is different for the other extra-embryonic tissue, the amnion, which remains attached to the embryo until dorsal closure is complete (Anderson, 1972). The extra-embryonic tissue of Drosophila, the amnio-serosa, remains attached to the embryo until dorsal closure. In this respect it resembles the amnion, suggesting that an equivalent of a serosa might be absent in Drosophila, which would explain the absence of hb expression in the respective tissue in Drosophila. However, we have obtained conflicting data on this point. We have studied the expression pattern of an apparent homologue of zerknüllt (zen) in Tribolium and found that it is expressed in the same anterior domain as hb and thus in the serosa cells, but not in the amnion which forms at later stages (unpublished data). In Drosophila, the zen gene is expressed in the amnio-serosa (Rushlow et al., 1987), which would suggest a closer relationship to the serosa cells. Thus, the question of the homology status of the extra-embryonic tissues must remain open until more comparative data from other species are obtained.

However, setting this question aside, one can simply take the anterior hb (or zen) expression domain as a marker for the prospective extra-embryonic versus embryonic tissues. It then becomes clear that only a very small portion of the whole embryo is dedicated, at blastoderm stage, to become extra-embryonic tissue, while most of the nuclei/cells will eventu-ally end up in the embryo or the amnion precursor cells which at this stage are located in a small rim around the embryonic cells (Anderson, 1972). This observation suggests that mater-nally derived gradients could act more or less in the same way as in Drosophila, namely throughout the egg (St. Johnston and Nüsslein-Volhard, 1992). Such gradients could instruct the nuclei before they move to the posterior ventral side where they form the germ anlage. Interestingly, this movement of the nuclei is characteristic for all short germ band embryos, but does not occur in long germ band embryos (Anderson, 1972). Thus, long germ band embryos must have lost the respective developmental pathway and are therefore neotenic to short germ band embryos in this respect.

Drosophila hb also shows a posterior expression domain at blastoderm stage which is regulated by the terminal system. Apparently the same expression domain is present in Tribolium, though not at the blastoderm stage, but towards the end of germ band elongation. This raises the interesting question of how this is regulated in Tribolium. The terminal maternal information in Drosophila is transmitted via torso, which codes for a receptor tyrosine kinase (Sprenger et al., 1992). Signalling from these receptor molecules (Casanova and Struhl, 1993) eventually activates the terminal gap genes tailless and huckebein which code for transcription factors that form short range gradients in the embryo (St. Johnston and Nüsslein-Volhard, 1992; Pankratz and Jäckle, 1993). tailless then acts as an activator for the terminal hb expression and huckebein as a repressor (Casanova, 1990). Evidently, this system, if conserved in Tribolium, must act somewhat differ-ently. If a terminal signal exists and if this signal activates terminal gap genes, these would exert their regulatory function on hb only much later, namely when the germ band has elongated and when it is in fact physically not linked to the posterior pole any more, but has moved up to the dorsal side. This would imply that the cells that have received the terminal signal at blastoderm stage are already committed to a terminal fate, even though the relevant genes are expressed only at later stages. Interestingly, early fate mapping experiments in short germ embryos have indicated that the proctodaeum, which depends on the terminal system in Drosophila, can be fate mapped already at blastoderm stage (Anderson, 1972), indi-cating that the respective cells are indeed already committed. There is even some molecular evidence for an early terminal signal in Tribolium, namely from the expression pattern of the segment polarity gene wingless (Nagy and Carroll, 1994). Apart of its segmental expression, wingless also has a posterior terminal expression domain, both in Drosophila and in Tribolium. Though the function and exact regulation of this domain is not yet known, it can nonetheless be observed in Tribolium from blastoderm stage onwards until the end of germband extension.

We conclude that although the hb expression between Drosophila and Tribolium differs in some details, the general pattern is nonetheless surprisingly well conserved. Most importantly, since hb in Drosophila is more or less directly regulated by three maternal systems (bcd, nos and tor) and since the respectively regulated expression domains can be seen in Tribolium, we can speculate that these maternal systems may also be conserved in Tribolium.

We thank Susan Brown for providing the genomic Tribolium library, Rebecca Oberhauser for help with sequencing, Robert Wheeler for antibody staining experiments, Claire Fazakerley and Gabby Dover for making unpublished Musca hunchback sequences available and Martin Klingler for critically reading the manuscript. This work was supported by grants from the Deutsche Forschungs-gemeinschaft and the Human Frontier of Science Program.