The metameric organization of the insect body plan is initiated with the activation of gap genes, a set of transcription-factor-encoding genes that are zygotically expressed in broad and partially overlapping domains along the anteroposterior (AP) axis of the early embryo. The spatial pattern of gap gene expression domains along the AP axis is generally conserved, but the maternal genes that regulate their expression are not. Building on the comprehensive knowledge of maternal gap gene activation in Drosophila, we used loss- and gain-of-function experiments in the hover fly Episyrphus balteatus (Syrphidae) to address the question of how the maternal regulation of gap genes evolved. We find that, in Episyrphus, a highly diverged bicoid ortholog is solely responsible for the AP polarity of the embryo. Episyrphus bicoid represses anterior zygotic expression of caudal and activates the anterior and central gap genes orthodenticle, hunchback and Krüppel. In bicoid-deficient Episyrphus embryos, nanos is insufficient to generate morphological asymmetry along the AP axis. Furthermore, we find that torso transiently regulates anterior repression of caudal and is required for the activation of orthodenticle, whereas all posterior gap gene domains of knirps, giant, hunchback, tailless and huckebein depend on caudal. We conclude that all maternal coordinate genes have altered their specific functions during the radiation of higher flies (Cyclorrhapha).
In insects, the initiation of differential zygotic gene expression along the anteroposterior (AP) axis of the embryo begins with the activation of gap genes (Davis and Patel, 2002; Pankratz and Jäckle, 1993). The sequence of gap gene expression domains along the embryonic AP axis is generally conserved across insects, but the maternal genes that initiate their expression differ between higher taxa (Brent et al., 2007; Goltsev, 2004; Bucher and Klingler, 2004; Cerny et al., 2008; Liu and Patel, 2010; Lynch et al., 2006a; Lynch et al., 2006b; Marques-Souza et al., 2008; Olesnicky et al., 2006; Pultz et al., 2005; Schröder, 2003; Schröder et al., 2000; Sommer and Tautz, 1991; Stauber et al., 2002; Wolff et al., 1995). How the maternal factors evolve and become exchanged is not well understood.
In Drosophila, the differential expression of gap genes in the early embryo rests on torso, bicoid, hunchback, caudal and nanos. These genes are active in distinct portions of the syncytial embryo and ensure pattern formation at the poles through the terminal gap genes huckebein and tailless, head segmentation through head gap genes such as orthodenticle, empty spiracles and buttonhead, and trunk segmentation through the canonical gap genes hunchback, Krüppel, knirps and giant (Furriols and Casanova, 2003; Pankratz and Jäckle, 1993; Rivera-Pomar and Jäckle, 1996; St Johnston and Nüsslein-Volhard, 1992; Surkova et al., 2008).
The maternal gene products of bicoid, torso, hunchback and caudal provide partially redundant input for the activation of gap genes in their proper domains. The homeodomain transcription factor Bicoid is expressed in an anterior-to-posterior gradient (Berleth et al., 1988; Driever and Nüsslein-Volhard, 1988a; Driever and Nüsslein-Volhard, 1988b; Gregor et al., 2007a; Gregor et al., 2007b; Spirov et al., 2009) and contributes to the activation of all gap genes (Driever et al., 1989; Eldon and Pirrotta, 1991; Finkelstein and Perrimon, 1990; Gao et al., 1996; Gaul and Jäckle, 1987; Hülskamp et al., 1990; Kraut and Levine, 1991; Liaw and Lengyel, 1993; Ochoa-Espinosa et al., 2009; Pignoni et al., 1992; Rivera-Pomar et al., 1995; Rivera-Pomar et al., 1996; Tautz, 1988; Walldorf and Gehring, 1992; Wimmer et al., 1995). The receptor tyrosine kinase Torso signals symmetrically at both poles of the egg and activates the terminal gap genes huckebein and tailless by locally counteracting ubiquitous repressors (Casanova and Struhl, 1993; Jiménez et al., 2000; Paroush et al., 1997; Sprenger and Nüsslein-Volhard, 1992). In the absence of Torso, the posterior domains of huckebein and tailless are missing but, because of the activating input by Bicoid, their anterior domains are only reduced (Brönner and Jäckle, 1991; Pignoni et al., 1992). In the absence of Bicoid, the head gap gene domains are missing and zygotic anterior hunchback expression is substituted by a mirror image duplication of the Torso-dependent posterior domain (Finkelstein and Perrimon, 1990; Tautz, 1988; Walldorf and Gehring, 1992; Wimmer et al., 1995). The central and posterior domains of the trunk gap genes Krüppel, knirps and giant are shifted towards the anterior pole in Bicoid-deficient embryos (Eldon and Pirrotta, 1991; Gaul and Jäckle, 1987; Kraut and Levine, 1991), but they persist because of the maternal activities of hunchback and caudal. The zinc finger transcription factor Hunchback is sufficient to promote the central domain of Krüppel (Hülskamp et al., 1990) and the homeodomain protein Caudal is sufficient to maintain the posterior expression domains of knirps and giant (Rivera-Pomar and Jäckle, 1996; Rivera-Pomar et al., 1995).
Global polarity of the gap gene scaffold is provided by the Bicoid gradient, which, in addition to activating gap gene transcription, also represses the translation of the ubiquitous maternal caudal transcript at the anterior pole (Cho et al., 2005; Niessing et al., 1999; Rivera-Pomar et al., 1996). Polarity is additionally provided by Nanos, which is enriched in the posterior embryo (Wang et al., 1994; Wang and Lehmann, 1991). Nanos represses translation of ubiquitous maternal hunchback transcript and thereby allows for the activation of knirps and giant (Hülskamp et al., 1989; Irish et al., 1989; Kraut and Levine, 1991; Sonoda and Wharton, 1999; Struhl, 1989; Wharton and Struhl, 1991). In the absence of bicoid and maternal hunchback activity, global AP polarity is lost, knirps, giant and Torso-dependent domains are expressed symmetrically, and anterior and central gap gene domains are absent (Gavis and Lehmann, 1992; Hülskamp et al., 1990).
As part of our efforts to trace the evolution of the maternal initiators of embryonic pattern formation in dipteran insects, we used the scaffold of gap gene domains to assess the initiation of embryonic pattern formation in the hoverfly Episyrphus balteatus (Syrphidae). Episyrphus belongs to the sister taxon of higher cyclorrhaphan flies (Schizophora, including Drosophila) (Grimaldi and Engel, 2005; Lemke and Schmidt-Ott, 2009; Yeates and Wiegmann, 2005) and is currently one of the most ‘basal’ dipteran species amenable to functional genetic studies in early embryos (Lemke and Schmidt-Ott, 2009; Rafiqi et al., 2008). Like Drosophila, Episyrphus specifies segments simultaneously prior to gastrulation (long-germ development) (Bullock et al., 2004). In a previous study, we have shown that AP patterning in Episyrphus relies more heavily on caudal than embryonic development in Drosophila. Episyrphus caudal (Eba-cad) RNAi disrupts or deletes post-oral segments (Lemke and Schmidt-Ott, 2009), whereas caudal-deficient Drosophila embryos show segmentation in the head, thorax and even parts of the abdomen (Macdonald et al., 1986; Rivera-Pomar and Jäckle, 1996; Olesnicky et al., 2006). Furthermore, we reported that ectopic expression of Episyrphus nanos at the anterior pole suppresses the development of the head, thorax and five abdominal segments (Lemke and Schmidt-Ott, 2009). Here, we provide a comprehensive analysis of gap gene activation in Episyrphus and show that, although the basic scaffold of gap gene expression has remained conserved, maternal regulatory input of all coordinate genes has shifted between Episyrphus and Drosophila.
MATERIALS AND METHODS
Fragments of Episyrphus orthologs were obtained by PCR on cDNA using pairs of degenerate primers. A single primer pair was used to isolate Episyrphus orthologs of knirps and knirps-related (Eba-kni and Eba-knrl; 5′-CCGGCRGCNGGNTTYCAYTTYGG and 5′-ARRCARTGRATYTTRAACCARTTNGA), huckebein (Eba-hkb; 5′-CARACVTAYTCRCGNYTNTTCCG and 5′-CGYTCYWKNGGCAWRTGMGTYTT) and tailless (Eba-tll) (Schröder et al., 2000). Two primer pairs were used in a nested PCR for the isolation of Episyrphus orthologs of Krüppel (Eba-Kr) (Rafiqi et al., 2008), giant [Eba-gt; 5′-TTCAARGCNTWYCCNMRNGAYCC and 5′-GCCCKRATNGCNATYTCRTCYTCYT; nested 5′-GARMGNMGNMGNAARAAYAA (Bucher and Klingler, 2004) and 5′-GCCCKRATNGCNATYTCRTCYTCYT] and torso [Eba-tor; 5′-GTNCAYMGNGAYYTNGCNGC (Schoppmeier and Schröder, 2005) and 5′-TKCCNCCKAGNGTKNNKATCTC; nested 5′-TNYCMGAYTTYGGNCTNAGTCGNGA and 5′-ARNAYNCCRAANSWCCANACRTC (Schoppmeier and Schröder, 2005)]. A homolog of bicoid (Eba-bcd) was identified in 454 transcriptome sequences of normalized cDNA from 0- to 4-hour-old Episyrphus embryos. A detailed description of the transcriptome data will be published elsewhere. The 3′ untranslated region (UTR) of Eba-bcd (0.7 kb including the putative polyadenylation signal), as well as larger cDNA fragments of Eba-kni, Eba-knrl, Eba-Kr, Eba-gt, Eba-tll and Eba-tor, were isolated by rapid amplification of cDNA ends (RACE; see Table S1 in the supplementary material).
Double-stranded RNA (dsRNA) of Eba-bcd spanned nucleotides 1 to 837 of the Eba-bcd open reading frame (ORF; 1 is the first nucleotide of the ORF). Eba-tor dsRNA spanned nucleotides −1237 to −04 (1 is the first nucleotide of the stop codon) in all described analyses. In addition, Eba-cad expression was analyzed after knockdown of Eba-tor by an Eba-tor dsRNA fragment spanning nucleotides −414 to −230, and Eba-otd expression was analyzed after knockdown of Eba-tor by an Eba-tor dsRNA fragment spanning nucleotides −414 to 276 (1 is the first nucleotide of the stop codon). dsRNA for Episyrphus caudal, Episyrphus hunchback and Megaselia bicoid was generated as described (Lemke and Schmidt-Ott, 2009; Lemke et al., 2008). The template for capped Eba-bcd mRNA was amplified by PCR from cDNA with primer pair 5′-CATGCCATGGCGGAAGAACCATGTGTGAC and 5′-ACGCGTCGACTAAACAATTTCTAAAGTATTTGCGTGTTCCG, which introduced an NcoI site at the 5′ end and a SalI site at the 3′ end of the ORF. The PCR product was digested with NcoI and SalI and cloned into pSP35, and capped mRNA was synthesized as described (Lemke and Schmidt-Ott, 2009). Injection and fixation of embryos was carried out as described (Lemke and Schmidt-Ott, 2009; Rafiqi et al., 2008).
In situ hybridization
RNA probes were labeled with digoxigenin, fluorescein or biotin, and wholemount in situ hybridization was performed essentially as described (Kosman et al., 2004; Tautz and Pfeifle, 1989). The Eba-bcd probe comprised nucleotides 1 to 837 of the ORF (1 is the first nucleotide of the ORF), the Eba-kni probe comprised 418 nucleotides of 5′UTR and adjacent nucleotides 1 to 192 of the ORF, the Eba-knrl probe comprised nucleotides 116 to 339 of the presumably truncated ORF plus 630 adjacent nucleotides of potentially intronic sequence, the Eba-Kr probe comprised nucleotides 845 to 1518 of the ORF plus 54 adjacent nucleotides of 3′UTR, the Eba-gt probe comprised 153 nucleotides of 5′UTR and adjacent nucleotides 1 to 1042 of the ORF, the Eba-tll probe comprised nucleotides 8 to 1258 of the ORF and the Eba-tor probe comprised nucleotides −414 to 1 of the ORF (1 is the first nucleotide of the stop codon) plus 276 adjacent nucleotides of 3′UTR. The Eba-hkb probe comprised the 677 nucleotides of the ORF that were amplified by PCR using the indicated degenerate primer pair. Probes for Eba-otd and Eba-hb were prepared as described (Lemke and Schmidt-Ott, 2009). First-instar cuticles were prepared as described (Stern and Sucena, 2000) with a 2:1 mixture of Hoyer's medium and lactic acid.
A diverged bicoid homolog controls global AP polarity of the Episyrphus embryo
A bicoid homolog of Episyrphus (Eba-bcd) was isolated from 454-transcriptome sequences of 0- to 4-hour-old Episyrphus embryos by sequence similarity to bicoid (Fig. 1A). The open reading frame (ORF) of Eba-bcd encodes a homeodomain (with the canonical lysine at position 50) and several conserved Bicoid motifs, including the SLVMRR peptide motif immediately upstream of the homeodomain, the NSEXXEPLTP peptide motif at the N-terminus of the acidic domain and the TPXPLTPXSTP peptide motif close to the C-terminus of the PEST domain, which has been shown to contribute to Bicoid-dependent repression of caudal mRNA translation (Niessing et al., 1999). Eba-bcd does not display sequence conservation in the N-terminal self-inhibiting domain (SID) (Zhao et al., 2002) or in the C-terminal portion of the acidic domain, which influences the ability of Bicoid to function as a transcriptional activator (Schaeffer et al., 1999). Eba-bcd also lacks sequence similarity in the d4EHP binding domain, although two amino acids of the motif might be conserved (Y66 and L73). In the Drosophila protein, these residues are essential for the repression of caudal translation (Cho et al., 2005). Taken together, the sequence data suggest that Eba-bcd is a diverged ortholog of bicoid.
To test whether Eba-bcd functions as maternal anterior determinant during embryogenesis, we analyzed its expression by wholemount in situ hybridization and its function by RNA interference (RNAi) and ectopic mRNA injection experiments. During oogenesis, Eba-bcd transcript was detected in the nurse cells and in anterior and dorsal portions of the oocyte (Fig. 1B). In early embryos, Eba-bcd transcript was localized at the anterior pole (Fig. 1C). Eba-bcd transcripts disappeared prior to the onset of cellularization and zygotic expression was not observed (data not shown).
To assess the function of Eba-bcd, we used blastoderm gap gene expression domains of Episyrphus orthodenticle (Eba-otd), Episyrphus hunchback (Eba-hb) and Episyrphus caudal (Eba-cad) as molecular markers of early Episyrphus segmentation (Lemke and Schmidt-Ott, 2009). The Eba-otd domain spans the anterior quarter of the syncytial blastoderm and retracts from the anterior pole by the onset of cellularization. In Eba-bcd RNAi embryos, this domain was absent (17/17; Fig. 1D,E). Eba-hb is expressed in a broad anterior domain, which appears prior to cellularization, and in a narrow posterior domain, which appears at the onset of cellularization. In early Eba-bcd RNAi embryos, i.e. prior to the penultimate nuclear division cycle of the blastoderm, Eba-hb expression was absent (8/13; Fig. 1F,G) or strongly reduced (4/13); one embryo showed wild-type expression. In older Eba-bcd RNAi embryos, i.e. at or after the onset of cellularization, Eba-hb was expressed in a narrow domain at the posterior pole (2/10) or symmetrically at both poles (8/10; Fig. 1H,I). Zygotic Eba-cad is activated throughout the posterior three quarters of the early blastoderm and becomes confined to a narrow posterior cap at later blastoderm stages. In Eba-bcd RNAi embryos at early blastoderm stages, Eba-cad was expressed ubiquitously (7/7; Fig. 1J,K). At later blastoderm stages it was expressed in roughly symmetrical narrow caps at both ends (9/9; Fig. 1L,M). Conversely, injection of capped Eba-bcd mRNA into the posterior pole resulted in ubiquitous expression of Eba-otd with higher levels at both poles (Fig. 1N) and in the repression of Eba-cad (Fig. 1O). Cuticles of Eba-bcd RNAi embryos lacked the head, the thorax and three to five abdominal segments. In most of these cuticles, the missing anterior structures were replaced by a mirror-image duplication of the posterior abdomen (21/26; Fig. 1P,Q). However, some cuticles (5/26) lacked posterior structures at the anterior pole and were similar to the strongest phenotypes that we previously obtained following injection of Eba-nos mRNA at the anterior pole (Fig. 1R) (Lemke and Schmidt-Ott, 2009). In Eba-bcd RNAi embryos, this phenotype might reflect the incomplete knockdown, whereas in embryos expressing anterior Nanos, it might reflect incomplete translational suppression through an atypical Nanos response element (see Fig. S1 in the supplementary material). Cuticles of embryos that had been injected with Eba-bcd mRNA at the posterior pole lacked abdominal as well as thoracic segments, and exhibited a mirror-image duplication of a reduced head-skeleton (47/85; Fig. 1S); in a few of these cuticles, remnants of thoracic structures could be discerned between the two head-skeletons (6/47; data not shown). In the remaining cuticles, head structures were only partially duplicated at the assumed posterior pole (34/85; data not shown), and four cuticles were wild type. Taken together, the results suggest that Eba-bcd is necessary and sufficient to determine global AP polarity in Episyrphus embryos.
Eba-bcd and Eba-cad control non-overlapping sets of gap gene expression domains
In Episyrphus, the development of the abdomen, the thorax and parts of the gnathocephalon depends on Eba-cad (Lemke and Schmidt-Ott, 2009). In comparison with the strong RNAi phenotypes of Eba-bcd, these data indicate that at least five abdominal segments, the entire thorax and parts of the gnathocephalon depend on Eba-cad as well as on Eba-bcd. To test whether the overlapping cuticular phenotypes reflect overlapping deletion patterns in the expression domains of gap genes, we cloned Episyrphus homologs of Krüppel (Eba-Kr), knirps (Eba-kni) and knirps-related (Eba-knrl) (see Fig. S2 in the supplementary material), as well as giant (Eba-gt), and examined their expression in wild-type embryos, Eba-bcd RNAi embryos and Eba-cad RNAi embryos.
Like Eba-hb, none of the newly identified genes showed detectable levels of maternal expression in wild-type embryos (data not shown). Eba-Kr expression was initiated in a broad central domain from 40-70% egg length (EL; 0% is the anterior pole; Fig. 2A). At the onset of cellularization, Eba-Kr was activated in a second domain from 0-15% EL (Fig. 2B), which resolved during cellularization into a dorsal horseshoe-like stripe at about 15% EL (Fig. 2C,C′). During gastrulation, the central Eba-Kr domain split into two narrower stripes and new expression domains appeared at the posterior pole around the prospective proctodeal invagination, in the prospective serosa and in each segment (Fig. 2D,E′). Older embryos expressed Eba-Kr in the ventral nerve cord, the amnion and parts of the head (Fig. 2F). Eba-kni was activated in a broad posterior domain spanning 55-85% EL (Fig. 2G). A second domain appeared shortly thereafter in the ventral blastoderm at 0-30% EL. By the onset of cellularization, the posterior boundary of the posterior domain had shifted to 75% EL and the anterior domain had expanded in a narrow transverse stripe, which marked the prospective anterior rim of the cephalic furrow (Fig. 2H). At the onset of gastrulation, the posterior domain had disappeared (Fig. 2I). The anterior domain persisted during germ band extension (Fig. 2J). Expression of Eba-knrl was not observed above background levels in blastoderm embryos (data not shown). Eba-gt expression was initiated in an anterior domain from 5% to 40% EL, and in a posterior stripe spanning 85-90% EL (Fig. 2K). By the onset of cellularization, the posterior domain had expanded, spanning 75-95% EL (Fig. 2L). The anterior domain fragmented during cellularization, first into two and then into three stripes (Fig. 2L,M). During gastrulation, the anterior expression domains were further resolved, while the posterior domain became narrow and faint (Fig. 2N). Following the onset of germ band extension, Eba-gt was also expressed in the serosa (Fig. 2O,P).
To test whether Eba-bcd and Eba-cad control in part the same gap gene expression domains, we examined the expression patterns of gap genes in Eba-bcd and Eba-cad RNAi embryos. Following the knockdown of Eba-bcd, Eba-Kr expression was lost in both the anterior and the central domains (22/36; Fig. 2Q) or only in the central domain (9/36; data not shown); a few embryos were indistinguishable from wild type (5/36). Eba-kni was expressed broadly in Eba-bcd RNAi embryos, with a clearance at both poles (26/39; Fig. 2R) or only at the anterior pole (4/39), or it was expressed ubiquitously (9/39). Eba-gt was expressed in Eba-bcd RNAi embryos nearly ubiquitously with clearings at the poles (15/40; Fig. 2S); in the remaining embryos the expression pattern of Eba-gt was highly variable (data not shown).
In Eba-cad RNAi embryos, the anterior and central expression domains of Eba-otd (Lemke and Schmidt-Ott, 2009), Eba-hb (see next section) and Eba-Kr (18/18; Fig. 2T) could not be distinguished from the respective wild-type domains. However, the posterior Eba-kni domain was strongly reduced (2/15) or missing (13/15; Fig. 2U). Similarly, the posterior Eba-gt domain was strongly reduced (16/17) or completely missing (1/17; Fig. 2V). In addition, Eba-cad RNAi affected the refinement of the anterior Eba-gt domain. We conclude from this analysis that Eba-cad is necessary for the activation of posterior trunk gap genes, and that Eba-bcd and Eba-cad are essential for distinct, non-overlapping domains of gap gene expression.
Eba-cad activates terminal gap genes and Eba-hb at the posterior pole
To test whether Eba-cad also controls the posterior expression of terminal gap genes, we isolated Episyrphus homologs of tailless (Eba-tll) and huckebein (Eba-hkb). Eba-tll was activated at both poles of the syncytial blastoderm, spanning 0-20% EL and 80-100% EL, respectively (Fig. 3A). At this stage, we also observed several embryos with weaker, exclusively posterior staining (data not shown), suggesting that Eba-tll activation at the posterior pole might precede Eba-tll activation at the anterior pole. During cellularization, the posterior domain narrowed to a slim cap and the anterior domain retracted from anterior and ventral portions of the blastoderm (Fig. 3B); shortly before the onset of gastrulation, the anterior domain was cleared along the dorsal midline (Fig. 3C). Eba-hkb was activated at both poles of the syncytial blastoderm in comparatively narrow domains (Fig. 3D). During cellularization, expression in the anterior domain appeared weaker than in the posterior domain and shifted ventrally (Fig. 3E). By the onset of gastrulation, Eba-hkb was additionally expressed in two small patches on either side in the presumptive prognathal head (Fig. 3F).
The posterior Eba-tll domain was missing in Eba-cad RNAi embryos at the blastoderm stage (13/13; Fig. 3G); in older embryos, Eba-tll expression at the posterior pole was strongly reduced (data not shown). The posterior domain of Eba-hkb was strongly reduced (41/42); in one embryo it was missing completely (Fig. 3H). In addition, we noticed the absence of the posterior domain of Eba-hb in all but one embryo (12/13; Fig. 3I). Thus, Eba-cad is an essential activator of terminal gap genes at the posterior pole.
Eba-tor shifts positional information in the anterior blastoderm
To assess the regulatory input from the terminal system on Episyrphus gap gene regulation, we isolated a homolog of torso (Eba-tor). Eba-tor transcript was detected in the nurse cells and anterior oocyte of ovarian follicles (Fig. 4A) and throughout early embryos but was absent in the posterior pole plasm (Fig. 4B). After the onset of blastoderm cellularization, Eba-tor transcript was no longer detected (data not shown). Cuticles of Eba-tor RNAi embryos all lacked filzkörper, the abdominal segment A8 was lost and A7 was either reduced or absent, whereas in the anterior we distinguished three main classes of head patterning defects (compare Fig. 4C-F with 4G-N). The most severely affected embryos developed an anterodorsal hole in the cuticle and only rudiments of the cephalopharyngeal skeleton could be identified (4/30; Fig. 4G,H). Less-severely affected embryos developed a complete anterior cuticle but lacked the median tooth and the cephalopharyngeal plates, and the base of the antennal sense organ was fused (3/30; Fig. 4I-K). The majority of cuticles displayed the least severe head phenotype, in which the median tooth was absent and the cephalopharyngeal plates were reduced (23/30; Fig. 4L,M). This weaker phenotype is most similar to the torso phenotype in Drosophila (Schüpbach and Wieschaus, 1986).
To examine the role of Eba-tor in blastoderm patterning, we analyzed the expression of the terminal gap genes in Eba-tor RNAi embryos. In most Eba-tor RNAi embryos, posterior Eba-tll expression was lost, whereas a reduced expression domain could still be observed at the anterior pole (18/28; Fig. 5A), suggesting that Eba-bcd might promote the activation of Eba-tll at the anterior pole. In the remaining embryos, anterior expression of Eba-tll expression was reduced even further (6/28) or lost (4/28; Fig. 5B). As in the case of Eba-tll, remnants of the anterior Eba-hkb domain were observed in several Eba-tor RNAi embryos (6/24; Fig. 5C), which suggests that Eba-bcd also promotes the expression of Eba-hkb. However, in most Eba-tor RNAi embryos, Eba-hkb expression was absent entirely (18/24; Fig. 5D).
Eba-cad expression prior to the onset of cellularization was expanded (25/65; Fig. 1J; Fig. 5E), ubiquitous (11/65; Fig. 5F) or indistinguishable from wild type (29/65) in Eba-tor RNAi embryos. A similar result, albeit with lower penetrance, was obtained with a non-overlapping smaller fragment of Eba-tor dsRNA (see Materials and Methods). In embryos injected with the smaller dsRNA fragment, most embryos showed wild-type Eba-cad expression (17/23), some embryos displayed anteriorly expanded expression (5/23) and a single embryo displayed ubiquitous Eba-cad expression. Anterior expansion or ubiquitous caudal expression was not observed in wild-type embryos or embryos injected with Megaselia bicoid dsRNA (Lemke et al., 2008) stained in parallel with Eba-tor RNAi embryos (data not shown). The narrow posterior expression domain of Eba-cad during late cellularization (Fig. 1L) was typically lost (9/15) or reduced (4/15; data not shown) following Eba-tor RNAi, whereas the two remaining embryos showed wild-type expression. Thus, the activity of Eba-tor transiently promotes the repression of Eba-cad at the anterior pole and is required to maintain late blastoderm expression of Eba-cad at the posterior pole.
Expression of the remaining head and trunk gap genes was affected mostly at the termini in Eba-tor RNAi embryos. Eba-otd was strongly reduced (25/33) or absent (8/33; Fig. 5G), which we confirmed using a non-overlapping fragment of Eba-tor dsRNA (see Materials and Methods). In embryos injected with the alternative dsRNA fragment, Eba-otd expression was reduced (29/56), absent (2/56) or indistinguishable from wild-type (25/56). The anterior domain of Eba-Kr was missing (14/16) or strongly reduced (2/16) in Eba-tor RNAi embryos, whereas the central domain of Eba-Kr was present, although slightly shifted towards the anterior pole (Fig. 5H). The anterior domain of Eba-kni was moderately (7/37) or strongly (26/37) compressed and shifted towards the anterior pole (Fig. 5I,J), whereas in the remaining embryos, anterior expression was indistinguishable from wild type. The posterior Eba-kni domain was present, although slightly expanded, and its boundaries appeared less well-defined. Like the anterior domain of Eba-kni, the anterior domain of Eba-gt was typically shifted towards the anterior pole and lacked the two anterior-most head stripes (13/14), whereas the posterior domain failed to fully retract from the posterior pole (Fig. 5K). In embryos at the onset of gastrulation, this anterior shift of Eba-kni and Eba-gt expression in the presumptive head region coincided with a parallel shift of the cephalic furrow (data not shown). Finally, the posterior hunchback domain was strongly reduced (36/39; Fig. 5L), absent (1/39) or indistinguishable from wild type (2/39); the early anterior expression domain of Eba-hb appeared to be unaffected (9/9; data not shown). Taken together, our analysis of head and trunk gap gene expression domains in Eba-tor RNAi embryos suggests that the knockdown of Eba-tor shifts positional information in the anterior blastoderm by about 10-20% EL towards the anterior pole.
Eba-bcd substitutes for bicoid and maternal hunchback
Two independent protein gradients contribute to global AP polarity in Drosophila – Bicoid and maternal Hunchback (Tautz, 1988). Although maternal hunchback is not required to establish global AP polarity (Lehmann and Nüsslein-Volhard, 1987), embryos from bicoid-deficient mothers are asymmetric as they duplicate only the ‘telson’ (i.e. filzkörper and anal plates) (Frohnhöfer and Nüsslein-Volhard, 1986). The Nanos-dependent gradient of maternal hunchback activity ensures that Krüppel, knirps and giant are still expressed in distinct, albeit anteriorly shifted, domains (Fig. 6A) (Eldon and Pirrotta, 1991; Hülskamp et al., 1990; Kraut and Levine, 1991). Embryos without bicoid and maternal hunchback activity lack the central Krüppel domain, exhibit symmetrical knirps and giant expression and develop a mirror-image duplication of the posterior abdomen with a symmetry plane in the sixth abdominal segment (Hülskamp et al., 1990). Here, we have shown that global AP polarity of the Episyrphus embryo depends entirely on Eba-bcd. Early Eba-bcd RNAi embryos lacked Eba-hb and Eba-otd expression, and Eba-cad was derepressed at the anterior pole (Fig. 1E,G,K). Eba-bcd RNAi embryos also lacked the central domain of Eba-Kr (Fig. 2Q; Fig. 6A), exhibited symmetrical Eba-kni and Eba-gt expression (Fig. 2R,S; Fig. 6A) and developed a mirror-image duplication of the posterior abdomen with a symmetry plane in the sixth abdominal segment (Fig. 1Q). Thus, in contrast to our previous model (which postulated distinct regulators for anterior Eba-cad repression and anterior gap gene activation) (Lemke and Schmidt-Ott, 2009), head-to-tail polarity of the Episyrphus embryo appears to rely on a single gene, Eba-bcd.
Although apparently absent in Episyrphus, maternal hunchback expression has been observed in a wide range of dipterans, including higher (Drosophila, Musca), lower (Megaselia) and non-cyclorrhaphan flies (Clogmia) (Rohr et al., 1999; Sommer and Tautz, 1991; Stauber et al., 2000). As the maternal expression and posterior localization of the nanos transcript is also widely conserved in early dipteran embryos (Calvo et al., 2005; Curtis et al., 1995; Goltsev et al., 2004; Lemke and Schmidt-Ott, 2009), and as nanos homologs from cyclorrhaphan and non-cyclorrhaphan dipterans exhibit rescue activity in nanos-deficient Drosophila embryos (Curtis et al., 1995), the absence of maternal hunchback input as a second source of AP polarity in Episyrphus might simply reflect plasticity in early dipteran development [for other examples see Schetelig et al. (Schetelig et al., 2008), Stauber et al. (Stauber et al., 2008)]. However, direct functional evidence for a conserved role of nanos and maternal hunchback expression in providing head-to-tail polarity in dipteran embryos is currently limited to higher cyclorrhaphan flies. Musca bicoid RNAi embryos display head defects as well as deletions in the abdomen similar to those seen in hypomorphic bicoid mutants but the AP polarity of cuticles is not disrupted (Shaw et al., 2001). This observation is consistent with the activity of a second source of AP polarity that might be provided by a nanos-dependent maternal Hunchback gradient. Yet in Megaselia, a cyclorrhaphan outgroup of the Episyrphus–Musca–Drosophila clade, strong bicoid RNAi induces the formation of a symmetrical double abdomen (Lemke et al., 2008; Stauber et al., 2000), and nanos RNAi does not cause a segmentation phenotype (S.L. and U.S.-O., unpublished). Hence, in Megaselia, maternal hunchback and nanos are not sufficient for generating global AP polarity of the segmented body plan, just like in Episyrphus. Accordingly, this contribution has been either lost independently in the lineages leading to Megaselia and Episyrphus, or a morphologically sizable input of maternal hunchback and nanos activities in specifying AP polarity is characteristic of higher cyclorrhaphan flies and generally absent in lower dipterans. Functional studies in lower dipterans will be necessary to distinguish between these possibilities.
The terminal system is required for Bicoid-dependent gene regulation in Episyrphus
We found that Eba-tor, in addition to regulating its canonical targets Eba-tll and Eba-hkb, transiently regulates anterior repression of Eba-cad and contributes to the activation of Eba-otd (Fig. 5A-G; Fig. 6A,C). This contribution of Eba-tor to anterior patterning differs from Drosophila, where bicoid is the only known repressor of caudal, and where torso contributes to orthodenticle regulation mainly by repressing it at the anterior tip of late blastoderm embryos (Finkelstein and Perrimon, 1990; Gao et al., 1996). Consistent with a stronger torso input on blastoderm patterning in Episyrphus, we observed more-severe head defects in cuticles of Eba-tor RNAi embryos than have been reported for torso-mutant Drosophila embryos (Schüpbach and Wieschaus, 1986).
Regulation of Eba-cad by Eba-tor appears to be independent of Eba-otd because Eba-otd RNAi does not cause an expansion of the Eba-cad domain (Lemke and Schmidt-Ott, 2009). Repression of Eba-cad is also likely to be independent of the canonical torso targets tailless and huckebein, which are co-expressed with Eba-cad at the posterior pole. It is possible, however, that Eba-tor interacts directly with Eba-bcd to provide anterior repression of Eba-cad. In Drosophila, Bicoid is believed to be phosphorylated in a Torso-dependent manner, possibly indicating a direct modification of Bcd activity by the Torso pathway (Ronchi et al., 1993). The functional significance of these modifications on Drosophila AP patterning is not clear, but, in Episyrphus, similar post-translational modifications of Bicoid might be required to repress zygotic transcription of Eba-cad at the anterior pole. Alternatively, an unidentified repressor, jointly regulated by Eba-tor and Eba-bcd, might account for the repression of early zygotic Eba-cad expression in the anterior quarter of the Episyrphus embryo.
Regulation of Eba-otd by Eba-tor could be caused by ectopic Eba-cad expression at the anterior pole. However, we found that Eba-cad was eventually repressed at the anterior pole in Eba-tor RNAi embryos but we did not observe a corresponding delayed onset of Eba-otd expression, which suggests that ectopic Eba-cad expression is not the cause of Eba-otd suppression. Instead, Torso activity might be required for Eba-otd expression by suppressing the activity of a ubiquitous repressor. This mechanism would be comparable with Torso-dependent huckebein and tailless expression at the posterior pole of Drosophila embryos, where torso suppresses the activity of two ubiquitous repressors, Capicua and Groucho (Jiménez et al., 2000; Paroush et al., 1997). A similar mechanism might also contribute to the regulation of Eba-kni and Eba-gt, which, unlike in Drosophila, are also strongly affected by the loss of torso activity (Fig. 5I-K; Fig. 6A). Although this particular model might be difficult to test if Capicua and Groucho are provided as maternal proteins (in preliminary experiments we did not observe derepression of tailless, huckebein or orthodenticle in Episyrphus capicua RNAi embryos) (S.L. and U.S.-O., unpublished), our results suggest that Eba-tor is crucial for anterior gap gene expression in general. Thus, despite being the sole anterior determinant, Eba-bcd might not have the same activation potential as bicoid, which in Drosophila can overcome the repressing input of capicua or groucho in the absence of torso activity (Schaeffer et al., 2000). Consistent with a weak activation potential of Eba-bcd, we found that in Episyrphus posterior gap gene expression domains depend on the activating input of caudal.
Independent regulation of gap and pair-rule genes by caudal
In Drosophila, gap genes are required for regulating the expression of pair-rule segmentation genes in partially overlapping sets of seven transverse stripes (Pankratz and Jäckle, 1993). Independently of the gap genes, caudal appears to contribute directly to the activation of pair-rule genes (Li et al., 2008; Schroeder et al., 2004). In Caudal-deficient embryos, up to four pair-rule stripes are disrupted or missing (Macdonald and Struhl, 1986; Olesnicky et al., 2006) and cuticles of Caudal-deficient Drosophila embryos tend to show pair-rule segmentation defects in the abdomen and thorax (Macdonald and Struhl, 1986). Reduced posterior domains of knirps and giant in caudal-deficient Drosophila embryos can only partly account for this phenotype (Olesnicky et al., 2006; Rivera-Pomar and Jäckle, 1996; Rivera-Pomar et al., 1995). In other insects, it has been more difficult to distinguish the role of caudal in the regulation of gap and pair-rule genes, as the expression patterns of both groups of genes are strongly affected by caudal RNAi. In Nasonia, caudal RNAi embryos retain only one or two anterior stripes of even-skipped, but they also lack the anterior domain of knirps, the central domain of Krüppel and all posterior gap gene domains (Olesnicky et al., 2006). Similarly, caudal RNAi suppresses all stripes of even-skipped in the cricket Gryllus bimaculatus, but it also suppresses hunchback and Krüppel (other gap genes have not been analyzed) (Shinmyo et al., 2005).
In Episyrphus, knockdown of caudal suppresses all but the first stripe of even-skipped (Lemke and Schmidt-Ott, 2009) but only the posterior gap gene domains of knirps, giant and tailless (Fig. 2U,V; Fig. 3G-I; Fig. 6A,D,E). Compared with Drosophila, the Episyrphus data reveal an expanded role of caudal in the activation of gap genes similar to, but not as strong as, in Nasonia and Gryllus. However, the presence of the central domain of Krüppel and the anterior domain of hunchback in Eba-cad RNAi embryos suggest that the loss of gap gene domains does not fully account for missing pair-rule stripes. Taken together with the Drosophila data, our observations therefore suggest that, independent of its role in gap gene activation, caudal is required for pair-rule gene expression in Episyrphus. This gap-gene-independent regulation of pair-rule genes by caudal might be an ancient heritage of insects.
Episyrphus and other lower cyclorrhaphan flies establish global AP polarity only through bicoid and lack sizable input of nanos, although endogenous nanos activity in these species might stabilize the AP axis by repressing anterior development. Despite the absence of a redundant maternal system to generate global AP polarity, Eba-bcd appears to be a less potent transcriptional activator than Bicoid. In contrast to Drosophila, gap gene activation at the anterior pole of the Episyrphus embryo requires a strong contribution of the terminal system, whereas the posterior domains of knirps and giant are strictly dependent on caudal and do not appear to receive a significant activating input by Eba-bcd. Thus, rather than a strong activation potential, the exclusive control of the central Eba-Kr domain by Eba-bcd appears to be the crucial difference to Drosophila, which renders AP polarity in the Episyrphus embryo entirely dependent on bicoid.
We thank Ab. Matteen Rafiqi for comments on the manuscript, Annette Lemke for help with the Episyrphus culture and two anonymous reviewers for their helpful comments. Funding was provided by NSF grants 0719445 and 0840687 to U.S.-O.
Competing interests statement
The authors declare no competing financial interests.