A caudal mRNA gradient controls posterior development in the wasp Nasonia

Recent progress in genomic and molecular techniques in different insect species has allowed deep insights into the evolution of developmental regulatory gene networks. As Drosophila provides an unmatched in-depth description of the regulatory network that directs early development,other systems have emerged to take advantage of this knowledge and compare developmental strategies. Studies in the beetle Tribolium, in the cricket Gryllus, in the milkweed bug Oncopeltus and in the grasshopper Schistocerca have highlighted several common and different pathways to pattern the embryo (reviewed by Liu and Kaufman, 2005). We have chosen to study the wasp Nasonia, an important model system with which compare early development that is functionally accessible, through both genetics and parental RNAi (Pultz and Leaf, 2003; Lynch and Desplan,2006).

Much of our understanding of anteroposterior body axis formation has been a result of elegant screens for segmentation defects in Drosophila. The syncytial environment of the Drosophila embryo allows for the generation of morphogenetic gradients of transcription factors, which are established via mRNA localization, protein diffusion and translational repression. Reciprocal gradients are then interpreted by downstream targets in a concentration-dependent manner to establish a complex anteroposterior patterning system that will eventually form the segmented insect body plan. However, much of development in most other insects takes place in a cellularized environment, and thus not all basic mechanisms and principles used in Drosophila patterning can be conserved. Still, many of the genes involved in fly segmentation are well conserved(Tautz, 2004; Liu and Kaufman, 2005).

Although much attention has been focused on anterior patterning in the fly,the major anterior patterning factor bicoid (bcd) is not found outside the dipteran lineage (Dearden and Akam, 1999; Stauber et al., 1999; Stauber et al.,2000; Lynch and Desplan,2003), and this has thus led researchers to investigate the patterning networks of other insects, such as Tribolium(Tautz, 2004; Liu and Kaufman, 2005). Beetles use an ancestral mode of embryogenesis, termed short-germ embryogenesis, in which the embryo develops in the posterior of the egg and only anterior structures are patterned in a syncytial environment. Later,abdominal and posterior structures are formed in a cellularized environment through a region in the posterior of the germ rudiment termed the `growth zone'. This is in contrast to the more derived long-germ mode of patterning found in flies, where the embryo occupies the entire egg, is patterned completely within a syncytial environment and, thus, lacks a posterior growth zone (Davis and Patel, 2002). It has been proposed that an anterior patterning center, such as Bcd in the long-germ Drosophila, would not function well to pattern the anterior of the embryo in short germ insects(Stauber et al., 1999; Stauber et al., 2000):anteriorly localized factors would not be able to reach the germ rudiment at the posterior of the oocyte and would instead pattern the extra-embryonic membranes, which lie at the anterior (Lall and Patel, 2001; van der Zee et al., 2005). Instead, the ancestral system may have exclusively used a posterior patterning center, allowing for posteriorly localized factors to reach the developing embryo easily.

The posterior patterning homeoprotein Caudal (Cad) is conserved throughout evolution from C. elegans to mammals. The Drosophila caudalgene (Dm cad) is involved in posterior embryonic patterning and hindgut formation (Macdonald and Struhl, 1987; Moreno and Morata, 1999; Schulz and Tautz, 1995; Wu and Lengyel, 1998). Dm cad zygotic mutant embryos are not viable and exhibit posterior defects:a lack of anal pad, anal tuft structures and anal sense organs. This rather mild phenotype has been attributed to maternal rescue of the loss of zygotic Dm cad. Maternally mutant embryos that have been paternally rescued also show mild phenotypes with deletions in abdominal segment 8 (A8) and sometimes A4, but are viable. Embryos lacking both maternal and zygotic Dm cad, however, show severe segmentation defects. Although the head and thorax are normal, the body is shortened owing to elimination of all anal structures and disruption of more anterior abdominal segments(Macdonald and Struhl, 1986)(Fig. 3).

The phenotype resulting from loss of cad has also been investigated in Tribolium, Gryllus (cricket) and Artemia(brine shrimp) using RNA interference (RNAi)(Copf et al., 2003; Copf et al., 2004; Shinmyo et al., 2005), and also studied in Sacculina (barnacle)(Rabet et al., 2001). Strikingly, in each organism examined, loss of cad results in embryos where only anterior head structures remain and all thoracic, abdominal and posterior structures fail to form. This phenotype is more severe than the Dm cad^mat+zyg phenotype, and suggests that cadmay play a greater role in patterning ancestral insects than in Drosophila. In the intermediate germ Gryllus embryo, cad plays a major role in thoracic and gnathal patterning by activating transcription of the gap genes hunchback (hb) and Krüppel (Kr). This role in gap gene activation is played by bcd and maternal hb in Drosophila. It has thus been proposed that, in ancestral insects, cad sits at the top of the segmentation cascade and regulates gap gene expression, while bcdhas usurped this role in higher dipterans(Shinmyo et al., 2005).

The Dm cad gene is expressed maternally and zygotically in the embryo. Both transcripts share an identical open reading frame and encode a homeodomain protein of 427 amino acids. Maternal Dm cad RNA is first made in the nurse cells and is found evenly distributed throughout the embryo(Mlodzik and Gehring, 1987a)(Fig. 1E). The maternal Dm Cad protein product forms a posterior to anterior gradient via translational repression by Bcd in the anterior. Bcd binding to the cad mRNA is mediated by the Bicoid response element (BRE) in the 3′UTR of the cad transcript(Dubnau and Struhl, 1996;Rivera Pomar et al., 1996). A translationally controlled maternal gradient is also observed in both Bombyx mori (silk moth) and Tribolium,although it is not understood how it is established in these species(Wolff et al., 1998; Xu et al., 1994). Furthermore,the C. elegans cad homolog pal-1 is maternally expressed and its protein product is restricted to cells of the posterior lineage via translational repression by mex-3, a gene that encodes an mRNA-binding protein that shares no homology with Bcd(Hunter and Kenyon, 1996).

As the Drosophila embryo develops, the Cad protein gradient becomes steeper and recedes from the anterior, forming a zygotic abdominal expression domain (Macdonald and Struhl, 1987; Mlodzik and Gehring, 1987a)(Fig. 2H,I). Later, the abdominal expression domain disappears and only a posterior stripe remains(Wu and Lengyel, 1998)(Fig. 2J). In Gryllus,cad is expressed in the early embryo in a posterior-to-anterior gradient,and later is restricted to the posterior growth zone, as in Triboliumand Artemia (Shinmyo et al.,2005; Schulz et al.,1998; Copf et al.,2003).

In order to study evolution of insect patterning, we have chosen the long-germ hymenopteran Nasonia vitripennis (Nvit). This parasitic wasp is a model system where a forwards genetic screen and functional parental RNAi studies have been performed(Pultz et al., 1999; Pultz et al., 2000; Pultz et al., 2005; Lynch et al., 2006a; Lynch et al., 2006b). Nasonia uses a long-germ mode of embryogenesis similar to that of highly derived Drosophila. However, Nasonia does not possess a bcd homolog and might therefore rely on an ancestral patterning system. Thus, Nasonia is an ideal system in which to study the evolution of patterning gene networks in general, as well as to study the specific patterning changes that have occurred during the evolution of long-germ embryogenesis.

In a screen to identify genes involved in embryonic patterning in Nasonia, many mutations in segmentation genes were identified,including a large number that resemble mutations in Drosophila genes of the gap, pair-rule and Polycomb-group(Pultz et al., 1999). In particular, one mutant, head only (ho), has a phenotype very similar to Tc cad RNAi embryos. It is also reminiscent of Dm cad^mat+zyg mutant embryos, although more severe, and was thus hypothesized to be due to a lesion in the Nvit cad locus(Pultz et al., 1999; Pultz et al., 2000). Here, we use parental RNAi to show that ho is most likely to be a zygotic Nvit cad mutant. Using ho together with parental RNAi, we assessed the role of the maternal and zygotic Nvit cad components as compared with that of Dm cad. We find that Nasonia uses mRNA localization to generate a posterior to anterior cad mRNA gradient in the absence of translational regulation of Cad by Bcd. Furthermore, we provide evidence that in Nasonia, cad acts as a crucial posterior patterning center sitting atop the ancestral patterning hierarchy.

Nasonia wild-type and headless stocks were kept at 28°C. head only stocks were raised at 18°C. Embryos were collected and fixed as described in Pultz et al. (Pultz et al., 1999). Embryos were hand peeled on double sided sticky tape in 0.1% Tween 20 in 1 × PBS. Ovaries were fixed in 4% formaldehyde 0.1% Tween 20 in 1 × PBS for 20 minutes, and dehydrated in methanol. Cuticles were mounted in 50% Hoyer's medium and 50% lactic acid. In situ hybridization was performed as previously described(Brent et al., 2003).

Nvit cad parental RNAi was performed as described(Lynch and Desplan, 2006)using forward (5′ TAATACGACTCACTATAGGGAGACCAC-CAGAACCGCCGAGCTAAAGAC 3′ and reverse (5′ TAATACGACTCACTATAGGGAGACCACTCAGCGGCGAGATCAGTTAAA 3′) primers to generate templates via PCR for transcription of double-stranded RNA. (T7 promoters are in bold.)

cad zygotic mutants were generated by crossing pr[1]cad[2]P{ry[+t7.2]=neoFRT}^40A/CyOvirgin females to b[1]pr[1]cad[3]/In(2LR)Gla,wg[Gla-1]/CyO. Maternal cad mutants were generated by crossing pr[1]cad[2]P{ry[+t7.2]=neoFRT}40A/CyOvirgin females to P[ry⁺;hs-FLP]¹²; P[w⁺;Ovo^D1]^2LP[hs-neo;ry⁺FRT]^40A/CyOmales. Third instar larvae and virgin progeny were heat shocked at 37°C for 90 minutes. Virgin females were crossed to b[1]pr[1]cad[3]/In(2LR)Gla,wg[Gla-1]/CyOmales to generate progeny of which half are maternally mutant and half are maternal and zygotic mutants. To generate only maternally mutant progeny,heat-shocked virgin females were crossed to yw males. Ventral misexpression lines (sna>Kr; sna>hb; sna>otd; sna>tll) were gifts from Stephen Small. Males carrying both the snail misexpression transgene and a β2-tubulin-FLP transgene were crossed to ywvirgin females to activate ventral misexpression. The resulting progeny were assayed using in situ hybridization.

Nv caudal was initially cloned by PCR using degenerate forward(5′ CATGAATTCAARACKCGNACKAARGAYAARTA 3′), and the reverse(5′ TGAGTCGACRTTYTGRAACCADATYTTNAC 3′) primers.

Total RNA was isolated from pooled embryos collected 0-4 hours or 4-10 hours after egg laying. SMART PCR cDNA synthesis kit (Clontech) was used for first-strand cDNA synthesis. The 5′ RACE primer (5′GCGGATGGTGATGTACCGGCTAGAGTAG 3′) and 3′ RACE primer (5′AACTCGCCAGCAGCCTCGCCTTGTC 3′) were used to clone the 5′ and 3′ ends of Nasonia caudal mRNA.

PCR was used to characterize Nv caudal genomic sequence. Forward and reverse primer sets included: 5′ CAAGACACGAACGAAGGACAAGTACAG 3′,5′ ACGGTTAGCACTCGGGTATGAACAACT 3′; and 5′GTTGTTGAATTCGCCGAGCTAAAGACCGCAAGCAG 3′, 5′ATTGTTAACGTTGAGCACCGAGTGTTG 3′.

To determine whether head only was linked with Nv caudal,degenerate PCR was used to also clone caudal from the sister species Nasonia giraulti (Ng) for molecular mapping(Pultz et al., 2005). A polymorphism distinguishing the Nv and Ng caudal was detected by Ambion RNAse Mismatch Detect II kit. This allowed testing surviving sons of a Nv head only × Ng cross to determine if they had Ng or Nv caudal. All surviving sons (45/45) had Ng caudal, suggesting that caudal is linked to head only.

We first cloned the Nasonia caudal (Nvit cad) sequence with a degenerate PCR approach using primers directed towards a 120 bp fragment of the conserved homeodomain (see Materials and methods). We next analyzed Nvit cad expression in both the ovary and throughout embryogenesis using whole-mount in situ hybridization. Nvit cad is first observed during oogenesis within the nurse cells; later, Nvit cad mRNA is also found in the oocyte. Unexpectedly, Nvit cadmRNA is localized to the posterior pole of the oocyte in ovarian follicles(Fig. 1A,B). In early embryos,maternal Nvit cad mRNA is localized to a structure containing the germ plasm known as the oosome. Later, Nvit cad mRNA appears to be released and diffuses anteriorly, establishing a graded expression that reaches very far anteriorly (Fig. 1C-F). As this gradient is observed very shortly after embryos are laid, before expression of any of the (zygotic) gap genes, it is unlikely that zygotic transcription has initiated. The resulting posterior to anterior mRNA gradient suggests that a Cad protein gradient exists in Nasonia.

Two to 3 hours after egg laying (hAEL) at 28°C, during pole cell formation, a zygotic gradient of Nvit cad expression forms in the syncytial embryo. Although the exact timing of midblastula transition in Nasonia is not known, it is likely that this expression is zygotic as it coincides with the onset of gap gene expression. The initiation of zygotic Nvit cad transcription overlaps with the maternal expression,extending anteriorly to cover approximately three-quarters of the embryo(Fig. 1E). The expression pattern is complementary to that of Nvit otd, which is expressed at both poles (Lynch et al.,2006a), and cad mRNA is indeed excluded from both the anterior and posterior poles (Fig. 1G, Fig. 3C). By 4 hAEL, after nuclei have migrated towards the surface of the embryo, Nvit cad is absent from a dorsal strip corresponding to the extra-embryonic membranes, which extend along the anteroposterior axis of the embryo (data not shown). Nvit cad early zygotic expression extends further anteriorly than that of Dm cad (compare Fig. 1E and Fig. 2A with Fig. 2H,I), yet it is very reminiscent of cad expression in short-germ insects. At later stages of blastoderm development, Nvit cad zygotic expression begins to clear from the anterior to create a strong abdominal expression domain(Fig. 2A,B). By 5-6 hAEL, at the beginning of cellularization, expression recedes further from the anterior, refining into a posterior stripe(Fig. 2C). Later, a second stripe forms directly posterior to the first stripe, which eventually fades(Fig. 2C-E). This later expression of Nvit cad, just prior to and during gastrulation, is similar to Dm cad expression, where cells expressing this posterior stripe migrate during germband extension and eventually form a cluster of cad-expressing cells corresponding to the hindgut and anal plate primordia (Fig. 2E,F,K,L).

In Dm, hb regulates the abdominal expression of Dm cad in a concentration-dependent manner: High levels of hb repress, whereas low levels activate, Dm cad transcription(Schulz and Tautz, 1995). Additionally, in hb zygotic mutants, the posterior stripe of cad is expanded (Mlodzik and Gehring, 1987b).

We find that, in addition to an expansion of the posterior Dm cadstripe, hb^zyg mutants show ectopic dorsal expression of Dm cad in the anterior of the embryo(Fig. 3H). Similarly, in the zygotic Nvit hb^hl mutant(Pultz et al., 2005), the posterior Nvit cad stripe is duplicated at the anterior of the embryo. Additionally, faint Nvit cad staining spans the region between the wild-type posterior Nvit cad stripe and the ectopic anterior Nvit cad stripe (Fig. 3D-F). This suggests that Nvit hb prevents Nvit cad expression in the anterior of the embryo. Furthermore, the ectopic anterior Dm and Nvit cad stripes are reminiscent of the duplication of the posterior Dm cad stripe at the anterior of bcd^- mutant embryos(Mlodzik and Gehring,1987b).

As Nvit otd-1 to a large extent plays a role similar to that of Dm bcd (Lynch et al.,2006a), we examined Nvit cad expression in otd-1RNAi embryos. Zygotic Nvit cad becomes derepressed from both poles,resulting in expression throughout the embryo. The area of Nvit cadderepression corresponds to the otd-1 expression domains(Fig. 3A-C).

To assess in more detail the function of hb and otd in flies, and to address the function of Kr and tll in regulating cad, we next used ventral misexpression in Drosophila to examine the effect of the Dm gap genes tll, Kr, otd and hb on Dm cad expression. We used the snail (sna) promoter to drive ectopic expression in a ventral stripe (Andrioli et al.,2002). In sna>tll embryos, cad is activated in the ventral region of the embryo (Fig. 3I). Interestingly, Dm cad is not activated in the anterior ventral region of the embryo, suggesting that it is strongly repressed there. This activation by tll agrees with previous studies that showed a loss of the posterior Dm cad stripe in tllmutant embryos (Mlodzik and Gehring,1987b). Ventral misexpression of the other gap genes does not affect Dm cad expression (data not shown). In the case of hb, one would have expected ventral repression of the posterior Dm cad stripe as it has been shown that high levels of hbrepress Dm cad (Schulz and Tautz,1995). The lack of repression might be due to insufficient levels of ventral hb. The lack of effect of Kr misexpression on Dm cad is consistent with the wild-type expression of Dm cadin Kr mutants (data not shown). These results suggest that the role of hb and other gap genes in regulating cad expression may have changed in the Drosophila gene network, when compared with more ancestral patterning networks.

To examine the function of Nvit cad, we made use of parental RNAi to downregulate Nvit cad function(Lynch and Desplan, 2006). Female pupae were injected with dsRNA and allowed to develop. Embryos from these adult mothers were aged for ∼28 hours at 28°C; cuticles were subsequently examined. Interestingly, high concentrations of dsRNA targeting Nvit cad results in few embryos being laid: embryos derived from these mothers cease developing and do not reach the cuticular stage,suggesting that Nvit cad might play a role in oogenesis that is separate from its role in the early embryo. Females injected with control gfp dsRNA show no cuticular defects and no difference in egg laying from wild-type females (data not shown). Lowering the concentration of Nvit cad dsRNA, however, results in a range of cadphenotypes. The wild-type Nasonia cuticle is composed of mouth hooks at the anterior, three thoracic and ten abdominal denticle belts. The second thoracic, as well as the first three abdominal denticle belts are easily identified by the presence of spiracles(Fig. 4A). Weak Nvit cad RNAi phenotypes show fusion of denticle belts throughout the abdomen but most commonly between segments A2 and A3 (data not shown). Stronger RNAi phenotypes exhibit a combination of fused or missing denticle belts, with progressive loss of segments starting from the posterior. Although, the number of denticle belts missing ranges from 0 to 13, most embryos retain six or seven denticle belts. Embryos exhibiting strong phenotypes, however, typically retain only three or four denticle belts, with T3 or A1 being the most posterior denticle belt remaining (Fig. 4B). These phenotypes are reminiscent of, but more severe than,the Dm cad^mat+zyg phenotype, which also typically shows few abdominal denticle belts and often exhibits fusion of belts. Rarely do Dm cad^mat+zyg phenotypes show loss of denticle belts as far anteriorly as A2 (Fig. 4F).

It has been proposed that the head only (ho) phenotype(Pultz et al., 1999) results from a mutation in Nvit cad. ho cuticles exhibit a loss of posterior denticle belts (compare Fig. 4B with 4C). This phenotype is exacerbated with decreasing temperature. We carried out a meiotic mapping experiment to determine whether the homutation is linked to Nvit cad (see Materials and methods). We tested 50 individuals and recovered no crossovers, placing ho within 2 cM of Nvit cad. Consistent with this hypothesis, the Nvit cad RNAi phenotypic series closely phenocopies the range of ho phenotypes, and affects all of the structures affected by ho, strongly suggesting that ho is a mutation in cad. We also compared the pattern of engrailed (en) mRNA staining in ho and cad RNAi embryos. The wild-type staining pattern for enconsists of five head stripes and 12 trunk stripes(Pultz et al., 1999)(Fig. 4G). en staining in both ho and cad RNAi mutant embryos display variability that reflects their cuticular phenotypes. However, en stripes in the head always form normally. The most typical class of severely affected embryos displays all five normal head stripes but lacks 6 or seven trunk stripes. en trunk stripes also often display fusion, as is seen in the cuticles of both ho and cad RNAi embryos(Fig. 3H,I). These results are consistent with those previously reported for the ho mutant(Pultz et al., 1999). We examined Nvit cad expression in ho embryos using a probe directed against the region encoding the homeodomain and 3′UTR. Nvit cad expression is somewhat reduced at 28°C in ho mutants. However, when females are allowed to lay at 18°C (when the hophenotype is strongest), Nvit cad zygotic expression is almost completely absent (Fig. 2G).

Taken together, the linkage analysis, paternal RNAi phenotypic series, as well as loss of Nvit cad expression in ho strongly suggest that ho is due to a lesion in the Nvit cad locus. Therefore, ho mutant embryos will be used here to examine the effects of zygotic lack of Nvit cad expression, in contrast to parental Nvit cad RNAi, which knocks down both maternal and zygotic Nvt cadexpression.

To test whether maternal Nvit cad contributes to embryonic patterning and to decipher its place in the Nasonia patterning hierarchy, we examined the effects on gap gene expression of knocking down maternal and zygotic Nvit cad and compared our results with those obtained in ho mutants. We also compared the function of cadin Nasonia with its role in Drosophila.

Nvit hunchback (Nvit hb) is expressed maternally and zygotically in the Nasonia embryo(Pultz et al., 2005; Lynch et al., 2006a)(Fig. 5A). Maternal Nvit hb is first distributed throughout the embryo and remains unaffected in ho, as well as in Nvit cad RNAi embryos. Later zygotic Nvit hb appears as an anterior cap and as a broad stripe in the posterior of the embryo (Pultz et al.,2005). In ho mutants, the anterior expression domain of Nvit hb expands toward the posterior. The same effect is seen in embryos derived from females injected with Nvit cad dsRNA, suggesting that zygotic, but not maternal Nvit cad, positions the posterior boundary of the Nvit hb anterior zygotic expression domain(Fig. 5B,C). As cad is generally thought of as a transcriptional activator, we examined whether Nvit cad might activate a repressor of hb. A candidate for this repressor is Krüppel (Kr).

Nvit Kr is expressed in a broad stripe in the center of the Nasonia embryo resembling Dm Kr expression(Fig. 5G). In hoembryos, there is a clear reduction in Nvit Kr transcription, leaving only a thin stripe of expression (Fig. 5I). In Nvit cad RNAi embryos, the central broad expression domain of Nvit Kr is absent(Fig. 5H). Therefore, both maternal and zygotic Nvit cad components are required to activate Nvit Kr expression. Anterior Nvit hb expression might therefore expand towards the posterior in Nvit cad mutants owing to the absence of Nvit Kr expression(Fig. 5G-I). Indeed, in the absence of Nvit Kr, Nvit hb shows a similar posterior expansion(A.E.B. and C.D., unpublished). We next examined whether loss of Dm cad also affects Dm Kr expression. In sharp contrast to the role of Nvit cad in Nvit Kr activation, we find that Dm Kr is not affected in zygotic, maternal or maternal + zygotic Dm cad mutant embryos (Fig. 6I,J). Furthermore, there is no effect on Dm hbexpression in the same mutant genotypes(Fig. 6K,L).

cad has been shown to activate hb expression in Gryllus embryos where cad parental RNAi appears to cause a posterior shift in Gb hb expression, suggesting that it not only activates, but also sets the position of Gb hb(Shinmyo et al., 2005). This observation suggests that cad plays an ancestral role in activating hb expression. Our results, however, suggest that Nvit cadrepresses anterior Nvit hb through Nvit Kr, whereas it is Nvit otd that activates posterior Nvit hb(Lynch et al., 2006a). Surprisingly, the posterior Nvit hb stripe remains unaffected in Nvit cad RNAi (Fig. 5A-C).

Dm cad acts as a transcriptional activator of Dm giant(gt) and Dm knirps (kni)(Rivera-Pomar et al., 1995; Schulz and Tautz, 1995). We verified that removing either zygotic or maternal Dm cad alone shows no effect on Dm gt and Dm kni expression (data not shown). However, removing both maternal and zygotic Dm cad causes a reduction in the expression of the posterior stripe of Dm gt(Fig. 6A-D), while the posterior stripe of Dm kni is reduced in intensity and expanded posteriorly (Fig. 6E-H). This expansion is probably due to a reduction in Dm gt, which acts as a repressor of Dm kni (Rivera-Pomar et al., 1995). Nevertheless, these phenotypes are fairly mild.

Nvit kni and Nvit gt are expressed zygotically in a similar pattern to their fly counterparts(Fig. 5D,J; Fig. 6A,C,E,G). Zygotic Nvit cad appears to be necessary to activate the posterior stripes of both Nvit gt and Nvit kni, since they are missing in ho mutants (Fig. 5F,L). In Nvit cad RNAi embryos, the same effect is observed at the posterior, while the anterior expression domains of Nvit kni and Nvit gt are also affected. Although positioned properly,the anterior Nvit kni domain is dramatically reduced(Fig. 5E), while the anterior domain of Nvit gt is expanded posteriorly(Fig. 5K). This expansion is not only due to loss of Kr as there is no dramatic posterior expansion of Nvit gt in Nvit Kr RNAi embryos (A.E.B. and C.D., unpublished). Therefore maternal Nvit cad probably represses anterior Nvit gt directly or activates another repressor of Nvit gt, thereby establishing its posterior border of expression.

Nvit otd has recently been shown to act as a morphogen involved in anterior patterning. Moreover, Nvit otd is involved in posterior patterning and its posterior cuticular phenotype partly overlaps with that of Nvit cad (Lynch et al.,2006a). We therefore examined whether Nvit cad regulates expression of posterior Nvit otd and Nvit tailless (Nvit tll) (Lynch et al.,2006b), which is involved in terminal patterning. Nvit otd is expressed maternally at both poles in the early embryo. Zygotic expression later forms caps at both poles of the embryo(Lynch et al., 2006a). ho and Nvit cad RNAi embryos show normal maternal and zygotic expression of Nvit otd, consistent with the model that Nvit otd is a maternal morphogen that regulates its own expression(data not shown) (Lynch et al.,2006a).

Nvit tll is expressed zygotically and resembles Dm tllexpression (compare Fig. 4Mwith Fig. 5M)(Lynch et al., 2006b). Nvit tll is affected identically in both ho and Nvit cad RNAi embryos, where both the anterior and the posterior expression domains of Nvit tll are reduced. Later, however, the anterior expression domain is restored, while the posterior domain remains absent. Zygotic Nvit cad therefore activates both Nvit tllexpression domains, but is not necessary for later activation of anterior Nvit tll expression (Fig. 5M-O). In Drosophila, otd (data not shown) and tll remain unaffected in the three different classes of Dm cad mutants (Fig. 6M,N).

Our results indicate that Nvit cad plays a major role in gap gene regulation. As we see a much weaker regulatory contribution of Dm cadat the level of the gap genes, it is likely that this role has been taken over by bcd in Drosophila or become redundant with other patterning factors. However, although Dm cad might have become obsolete at the level of gap gene regulation, the Dm cad^mat+zyg phenotype does show severe segmentation defects. We therefore examined the role of Dm cad in regulating pair-rule gene expression as a possible explanation for the severe cuticular phenotype resulting from the complete loss of Dm cad.

Dm cad activates the pair rule gene fushi tarazu(ftz) through direct binding to the `zebra stripe' promoter element(Dearolf et al., 1989). Maternal Dm cad mutant embryos show an expansion of stripes 2, 4 and 7, while stripes 3, 5 and 6 are narrower than in wild-type embryos(Macdonald and Struhl, 1986). Dm cad^mat+zyg mutant embryos have loss of up to four posterior ftz stripes (MacDonald and Struhl, 1986) (Fig. 7D).

As Dm cad-binding sites have been identified in the 3+7 and 4+6 enhancer elements of the pair-rule gene Drosophila even skipped(eve) (Hader et al.,1998; Schroeder et al.,2004), we looked at the expression of Dm eve in the different Dm cad mutant backgrounds. In a small number of Dm cad^zyg mutant embryos, the posterior stripes of Dm eve are weakly reduced. In Dm cad^mat mutants, Dm eve stripes 4-7 are expressed weakly with stripes 5 and 6 not well resolved in some cases. In Dm cad^mat+zyg mutants, however,there is a loss of stripes 4, 6 and 7, a posterior expansion of stripe 5, as well as a reduction in stripe 3 expression(Fig. 7B). These results validate the presence of Cad-binding sites in the stripes 3/7 and 4/6 enhancer elements and further support the idea that the severity of the Dm cad^mat+zyg phenotype is a result of aberrant pair-rule gene expression rather than of defects in gap gene expression.

We find that the expression of Nvit eve and Nvit ftz is also severely affected in ho mutants, with few stripes remaining. In Nvit cad RNAi-treated embryos, both Nvit ftz and Nvit eve are more severely affected, often with only one or two pair-rule stripes remaining (data not shown). These results may simply reflect the effect on gap genes in the ho and RNAi treated embryos. Alternatively, as in Drosophila, Nvit cad may also directly regulate pair rule gene expression.

Much of our understanding of body plan formation comes from studies in Drosophila where Bcd, a factor that is not present outside the dipteran lineage, is a major organizer of the anteroposterior axis and is required for all anterior fates. However, as Bcd is not a conserved feature of anterior patterning, developmental biologists have sought out comparative analyses of anteroposterior development in insects such as Triboliumand Gryllus that lack Bcd (reviewed by Liu and Kaufman, 2005). However, making direct comparisons in short and intermediate germ insects with the long-germ insect Drosophila is complicated by the fact that they represent different modes of embryonic development. Here, we investigate the role of caudal in posterior patterning in Nasonia, a long germ insect that lacks bcd. We find that: (1) a maternal gradient of Nvit cad is achieved through mRNA localization rather than through translational repression by Bcd as in Drosophila; (2) Nvit cad plays a greater role in patterning the embryo than does Dm cad, and this role expands far anteriorly; and (3) Nvit cad is an activator of gap gene expression, in contrast to its role as a pair-rule gene activator in Drosophila.

We have investigated the function of Nvit cad in early embryogenesis using both parental RNAi and the ho mutation, which probably results from the loss of zygotic Nvit cad. This allowed us to distinguish maternal and zygotic functions for a gene in a species other than Drosophila. The fact that the Nvit cad RNAi phenotype is much more severe than total loss of Dm cad is not surprising given the fact that both maternal and zygotic expression patterns of Nvit cad reach much further towards the anterior of the embryo than Dm cad. Similarly, in Gryllus, the Gb cad RNAi phenotype includes a complete loss of thoracic, abdominal and posterior structures. This is reflected in the wild-type expression of Gb cad, which is expressed in the presumptive gnathal and thoracic regions, as well as in the posterior growth zone (Shinmyo et al.,2005).

The fact that the severe cad phenotype is conserved in arthropods suggests that ancestrally, cad played a greater role in embryonic development but has lost some of its importance in Drosophila. We discuss these roles of cad, and what function it has retained in Drosophila for pair-rule gene regulation.

We have shown that Nasonia establishes a maternal mRNA gradient in the early embryo using mRNA localization and diffusion. Maternal Nvit cad mRNA is tightly localized to the posterior of the oocyte. After the embryo is laid, however, the mRNA diffuses far towards the anterior creating an mRNA gradient. Nasonia has thus devised a new mechanism for establishing a posterior-to-anterior gradient of cad mRNA, which probably forms a similar gradient at the protein level. In Drosophila,cad maternal transcripts are homogenously distributed throughout the early embryo and the Cad protein gradient is produced later through translational repression by Bcd (Dubnau and Struhl, 1996; Rivera-Pomar et al., 1996) (Fig. 1D). A redundant translational repression system may exist in Nasonia to ensure that no Cad is produced at the anterior.

The mechanism that establishes the Cad gradient in Nasonia is of particular importance as bcd is a new addition to the developmental network and is found only in higher dipterans. Consequently, Bcd cannot be responsible for establishing the Cad gradient in more ancestral species(Lynch and Desplan, 2003). Nonetheless the Cad gradient is conserved among insects. In Tribolium, Cad protein is first expressed homogenously throughout the embryo. Later, however, a posterior to anterior protein gradient forms but nothing is known about the mechanisms leading to the formation of this gradient. Interestingly, however, when a transgene encoding the Tc cad mRNA is placed in Drosophila, it leads to the formation of a translational gradient that is dependent on bcd(Wolff et al., 1998). This argues that a common underlying mechanism may be responsible for establishing the protein gradient in Tribolium and in Drosophila. It is likely that bcd took over the function of a translational repressor present in ancestral insects, perhaps including Nasonia. The mRNA gradient might therefore be specific to the wasp. Interestingly, Nvit otd mRNA is also localized to both the anterior and posterior poles of the embryo, which has not been reported in any other species(Lynch et al., 2006a). This suggests that Nasonia may extensively use RNA localization mechanisms for setting up the anteroposterior axes in the embryo. Moreover, maternal mRNA localization may be a common feature of long germ development. Studies performed in other Hymenopterans, which undergo extremely diverse modes of embryogenesis, ranging from long-germ embryogenesis in Apis mellifera(Davis and Patel, 2002) to the polyembryonic development of Copidosoma floridanum(Grbic, 2003) will aid in identifying the conserved mechanisms among these diverse modes of embryogenesis. In Copidosoma, up to 2000 embryos may be produced clonally from a single egg, showing that maternal determinants cannot play similar axial patterning roles in this insect as seen in long and short germ insects (Zhurov et al., 2004). However, work in the long germ Apis mellifera might elucidate whether maternal mRNA localization is a common feature of long-germ embryogenesis.

In Nasonia, cad functions as an activator of gap gene expression,placing it at the top of the segmentation network similar to bcd in Drosophila. However, we find no evidence that Cad acts as a morphogenetic gradient. Gap genes are primary interpreters of anteroposterior cues and serve to divide the early embryo into broad expression domains. Among gap genes, Dm Kr is a particularly important player that acts as a potent repressor of other gap and pair rule genes. Positioning the Krdomain is therefore crucial and Bcd is involved in Dm Kr regulation in addition to activating a large number of anterior patterning genes such as Dm hb. bcd is therefore considered a master patterning gene(Hoch et al., 1991). In Gryllus, Gb Kr and Gr hb are activated by cad and it was hypothesized that this represents the ancestral function of cad, placing it at the top of the segmentation hierarchy(Shinmyo et al., 2005). Cad-binding sites have been identified in Dm Kr regulatory region,which may be vestiges that had once functioned in an ancestral patterning system (Schroeder et al.,2004). Our results in Nasonia confirm that the role of cad to activate Kr is conserved and supports the notion that this role has been usurped by bcd in Drosophila. We also find that Nvit cad activates tll expression. This role is not conserved in the fly, despite the presence of cad-binding sites in the regulatory region of Dm tll(Schroeder et al., 2004).

kni and gt, which are only weakly affected in Drosophila cad mutants, absolutely require cad in Nasonia. It is likely that kni and gt rely instead on bcd for activation in Drosophila. It should be noted that the anterior patterning gene bcd is involved in activating gap gene expression in the posterior of the embryo. This is also true of the posterior-most stripe of the pair-rule gene hairy, which relies on the combined activity of Bcd and Cad for activation(La Rosee et al., 1997). Similarly, although cad is typically thought to regulate expression in the posterior of the embryo, maternal Nvit cad is involved in regulating the anterior expression domains of both kni and gt.

Although the role of cad in activating the gap genes seems to have been taken over by bcd in Drosophila, complete loss of Dm cad does result in severe segmentation defects. We have shown that Dm cad acts at the level of pair-rule genes instead and that it is a strong transcriptional regulator of Dm eve expression.

bcd is believed to have evolved as a duplication of zen(Dearden and Akam, 1999; Stauber et al., 1999; Stauber et al., 2000) that later acquired a K₅₀ residue within its homeodomain, giving it the same binding specificity as Otd (Treisman et al., 1989). It has thus been proposed that otd is a major ancestral anterior patterning gene, the role of which has been taken over by bcd (reviewed by Lynch and Desplan, 2003). Interestingly, in Nasonia, otd is expressed maternally and zygotically at both poles. Loss of Nvit otdresults in the loss of both anterior and posterior structures(Lynch et al., 2006a). This phenotype is somewhat overlapping in the posterior with Nvit cad. Thus, it is likely that Nvit cad and Nvit otd work in concert to regulate posterior genes, as seen in fly with bcd and cad activating posterior hairy and kni stripes. The presence of Bcd (K₅₀) binding sites in promoter elements of genes expressed in the posterior of the Drosophila embryo may thus reflect an ancestral role of otd in activating posterior genes, although Nasonia remains the sole example so far of posterior otdexpression.

Nvit otd is necessary to repress Nvit cad from both poles of the embryo. Additionally, Nvit hb represses later Nvit cad expression in the anterior of the embryo. This suggests that zygotic Nvit cad is first activated throughout the embryo, and that a strong repression system is required to prevent Nvit cad from specifying posterior fates in the anterior. In Drosophila, the absence of Bcd leads to the expansion of maternal Cad to the anterior of the egg. This results in the duplication at the anterior of the embryo of a telson, a structure that requires cad. However, the mere presence of Cad at the anterior is not sufficient to induce the formation of a telson. In embryos where Bcd is present but unable to bind the cad 3′UTR, Cad is expanded anteriorly, yet only head involution defects are seen but no telson forms at the anterior (Mlodzik et al.,1990; Neissing et al.,1999; Neissing et al.,2002). This is probably due to the presence of bcd-dependent Hb at the anterior, which might inhibit Cad protein function. Like bcd, Nvit otd also acts, probably in concert with Nvit hb, in repressing posterior development in the anteriormost region of the embryo by repressing Nvit cad. However, Nvit otd controls Nvit cad at the transcriptional level, whereas bcd represses Dm cad at the translational level.

We propose that Nvit cad and Nvit otd function together in patterning the posteriormost segments(Lynch et al., 2006a). Nvit cad acts as the ancestral posterior patterning center responsible for activating the gap genes in the thoracic, abdominal and posterior regions of the long-germ wasp embryo, but it is Nvit otdthat functions as a morphogen by setting the positions of the gap genes.

In conclusion, the posterior-to-anterior gradient of Nasoniamaternal cad is established in a novel way through the formation of an mRNA gradient. Moreover, maternal Nvit cad plays a distinct role from its zygotic counterpart. Together, maternal and zygotic Nvit cadregulate gap gene expression in a non-redundant manner, placing cadat the top of the segmentation network. In Drosophila, it seems that cad has lost, to bcd, its ability to activate gap genes and instead its role in the patterning network is to regulate pair rule genes. The combinatorial activation of posterior kni by cad and bcd in Drosophila may be a remnant of the ancestral role of cad as the key transcriptional activator of gap genes. We thus propose that cad is the ancestral patterning center in short-germ embryogenesis and that this role is retained in Nasonia long-germ development but largely lost in Drosophila.

Lori Westendorf, Sam Gale and Jason Pitt made significant contributions in cloning and characterizing Nvit caudal. We thank Carol Trent for the Nasonia lambda genomic library. We are grateful to S. J. Small for the generous gift of the ventral misexpression Drosophila lines. We are grateful to Darrell Killian, Steve Small and Jeremy Lynch for their helpful comments on this article. We thank the Desplan laboratory and Flynet for their constant support. This work was supported by grants from NIH GM-64864 to C.D. and NSF IBN-9808769 to M.A.P. and D.L. E.C.O. was supported by NIH Training Grant 5 T32 HD007520. This investigation was conducted in a facility constructed with support from Research Facilities Improvement Grant C06 RR-15518-01 from NCRR, NIH. E.C.O. and C.D. conceived and designed the experiments. E.C.O. performed the experiments in flies and Nasonia,and generated the data for all figures. M.A.P. and D.L. initiated work on the cad locus; M.W. cloned and characterized the 5′ region of wild-type Nvit cad; L.T. mapped the Nvit cad locus to the vicinity of ho and performed the initial search for the ho mutation;M.A.P. and D.L. supervised the cloning and analysis of cad cDNA and genomics sequences. E.C.O. and C.D. analyzed the data and generated the figures. E.C.O. wrote the paper.