Regulation of Easter activity is required for shaping the Dorsal gradient in the Drosophila embryo

Dorsoventral patterning of the Drosophila embryo begins in the ovarian egg chamber, where the developing oocyte is surrounded by an epithelium of somatically derived follicle cells (reviewed byRay and Schüpbach, 1996;Nilson and Schüpbach,1999). During mid-oogenesis, activation of the EGF Receptor (Egfr)in the follicle cells is spatially restricted by the dorsally localized distribution of its ligand encoded by the gurken (grk) gene(Price et al., 1989;Schejter and Shilo, 1989;Neuman-Silberberg and Schüpbach,1993; Sapir et al.,1998; Wasserman and Freeman,1998; Peri et al.,1999; Ghiglione et al.,2002). As a consequence, transcription of pipe, which apparently encodes a glycosaminoglycanmodifying enzyme required for establishing embryonic polarity, is restricted to a broad stripe on the ventral side of the egg chamber (Sen et al., 1998; Sen et al.,2000).

Spatial information originating in the follicular epithelium is later transmitted to the embryo through the perivitelline space, which lies between the vitelline layer of the eggshell and the embryo plasma membrane (reviewed by Morisato and Anderson,1995). The establishment of embryonic polarity is dependent on the ventrally restricted activation of the uniformly distributed receptor Toll(Hashimoto et al., 1988;Hashimoto et al., 1991;Stein et al., 1991). The ligand for Toll is apparently encoded by the spätzle gene, which produces a protein containing a C-terminal cystine knot motif found in many vertebrate growth factors (Morisato and Anderson, 1994). The Spätzle protein is secreted into the perivitelline space as an inactive precursor, and is cleaved into the active ligand (Morisato and Anderson,1994; Schneider et al.,1994) through the activity of a serine protease cascade that includes the products of the genes nudel(Hong and Hashimoto, 1995),gastrulation defective (Konrad et al., 1998), snake(DeLotto and Spierer, 1986)and easter (ea) (Chasan and Anderson, 1989).

Activation of Toll initiates an intracellular signaling pathway that results in the nuclear translocation of the transcription factor encoded bydorsal, a member of the NF-κB/rel family (reviewed byDrier and Steward, 1997). Dorsal is initially present throughout the embryonic cytoplasm, where it is retained by the inhibitory IκB protein encoded by cactus. Signaling on the ventral side leads to the proteolysis of Cactus, thereby releasing Dorsal (Belvin et al.,1995; Bergmann et al.,1996; Reach et al.,1996). Along the dorsoventral axis, high levels of Dorsal protein are present in ventral nuclei, progressively lower levels in lateral nuclei and no detectable protein in dorsal nuclei(Roth et al., 1989;Rushlow et al., 1989;Steward, 1989). The shape of the Dorsal gradient is characterized by the size of the ventral domain(measured by the number of nuclei expressing peak Dorsal) and a distinct slope(assessed by the number of nuclei that lie between highest and lowest nuclear Dorsal). Changing the shape of the Dorsal gradient causes patterning defects that lead to embryonic lethality.

The Dorsal gradient subdivides the axis into distinct domains by setting the expression limits of key zygotic regulatory genes, which are responsible for initiating the differentiation of various tissues. High levels of nuclear Dorsal lead to the transcription of twist in mesodermal precursor cells (Thisse et al., 1988;Jiang et al., 1991;Pan et al., 1991). The Twist protein is itself expressed in a graded fashion in the most ventral 16-18 cells, and this domain can be subdivided into smaller threshold responses(reviewed by Rusch and Levine,1996). Intermediate levels of nuclear Dorsal activate the transcription of short gastrulation (sog) in two lateral stripes flanking the ventral Twist domain, each about 14-16 cells wide(François et al.,1994). The rhomboid (rho) gene is transcribed in a ventral subset of 8-10 cells in each sog domain(Bier et al., 1990;Ip et al., 1992). Thezerknüllt (zen) gene is transcribed in the dorsal∼40% of the embryo circumference, in the region where Dorsal is absent from nuclei (Rushlow et al.,1987). In the experiments described below, we characterize changes in the Dorsal gradient by examining the expression domains of these zygotic genes.

We are keenly interested in understanding how the shape of the Dorsal gradient is regulated. Two classes of mutations produce particularly interesting effects on the wild-type shape. In the first class, embryos produced by grk^- and Egfr^- females show two peaks of nuclear Dorsal separated by a shallow ventral minimum(Schüpbach, 1987;Roth and Schüpbach,1994). These ventralized embryos proceed to gastrulate with two ventral furrows instead of the single wild-type ventral furrow. Recent studies showed that this phenotype could be mimicked by overexpression of Spätzle, suggesting that partial axis duplication arises from events in the perivitelline fluid of the embryo(Morisato, 2001). Despite the dramatic reshaping of the ventral domain in these mutant embryos, the slope of the Dorsal gradient remains wild type(Roth and Schüpbach,1994; Morisato,2001).

In the second class, dominant alleles of easter(ea^D) cause a more symmetric distribution of nuclear Dorsal (Steward, 1989). As a consequence, females carrying ea^D mutations produce ventralized embryos, in which ventrolateral structures are expanded at the expense of dorsal structures, or lateralized embryos, in which dorsoventral polarity is largely lost (Chasan and Anderson, 1989; Jin and Anderson, 1990).

The easter gene encodes the final member of the protease cascade required to activate Spätzle. Easter is initially synthesized as an inactive zymogen containing an N-terminal pro-domain and a C-terminal catalytic domain. Proteolytic cleavage at the activation site between these two domains by Snake presumably generates active Easter in vivo(Chasan et al., 1992;Dissing et al., 2001;LeMosy et al., 2001). Yet, in wild-type embryo extracts, active Easter is found in a highM_r complex called Ea-X, which is hypothesized to contain a protease inhibitor of the serpin family(Misra et al., 1998). Easter is proposed to be active only on the ventral side of the embryo. Theea^D mutations, which map to conserved regions within the catalytic domain (Jin and Anderson,1990), somehow cause a loss of this spatial regulation.

We present our analysis of a group of representativeea^D alleles. We have characterized changes in the shape of the Dorsal gradient caused by these mutations, by examining the expression domains of the zygotic genes zen, sog, rho and twist. Within the allelic series, dorsoventral asymmetry was progressively lost and the slope of the Dorsal gradient flattened. When production of activated Easter was examined in ea^D embryo extracts, we detected an Easter form corresponding to the free catalytic domain, which was never observed in wild type. The Ea^D catalytic domain exhibited protease activity, as measured by its ability to generate processed Spätzle in the embryo. In the case of the strongest lateralizing ea^D allele,protease activity was detected several hours after the blastoderm stage in perivitelline fluid transfer experiments. Finally, mutant Ea^Dproteins expressed in cultured Drosophila S2 cells were shown to cleave precursor Spätzle. These data suggest that theea^D mutations interfere with Easter inactivation by the inhibitor X, and support a model in which regulation by X is required for shaping the Dorsal gradient.

The wild-type stock used was Oregon R. The easter allelesea¹, ea⁴, ea^5022rx1, ea^125.3,ea^83l, ea⁵⁰²², ea²⁰ⁿ andea^5.13 have been previously described(Chasan and Anderson, 1989;Jin and Anderson, 1990). Embryos produced by ea⁴/ea^5022rx1 females have no detectable easter mRNA (Chasan et al., 1992), and were used as ea^- embryos;ea^5022rx1 was used as the null allele designatedea^- in the figures and tables. The strongly dorsalizingToll alleles Tl^5BREQ, Tl^9QRE andTl^1-RXH have been previously described(Anderson et al., 1985). The strongly dorsalizing spätzle alleles spz^rm7and spz^D1-RPQ have been previously described; no Spätzle protein is detected in embryos laid byspz^D1-RPQ/Df (3R) Ser^+R82f females(Morisato and Anderson, 1994). The strongly dorsalizing pipe alleles pip³⁸⁶ andpip⁶⁶⁴ have been previously described(Anderson and Nüsslein-Volhard,1984).

Expression of zen, sog and rho RNA in embryos was detected by antisense RNA probes. Hybridization and detection was carried out as described (Tautz and Pfeifle,1989). Embryos were embedded in Spurr resin (Polysciences) and sectioned every 10 μm. The size of each domain was calculated by counting the number of stained cells and dividing by the total number of cells in the embryo circumference at 50% egg length. In order to ensure that only young blastoderm embryos were analyzed, the quantitation presented inTable 1 was restricted to embryo cross-sections that contained 80-100 cells.

Expression of Twist protein was detected with rabbit anti-Twist antibodies,kindly provided by Siegfried Roth (Universität zu Köln). Primary antibodies were visualized with biotin-conjugated anti-rabbit antibodies and streptavidin-horseradish peroxidase (HRP) using Vectastain ABC (Vector Laboratories) as described (Patel,1994). Embryos were embedded in Spurr resin and sectioned every 10μm.

Cuticles were manually dissected out of their vitelline membrane cases and prepared as described (Wieschaus and Nüsslein-Volhard, 1986).

Embryonic extracts were prepared as previously described(Morisato and Anderson, 1994). Protein samples were separated on a polyacrylamide gel and blotted to PVDF. Rabbit antibodies generated against bacterially produced Spätzle protein were affinity purified against the N- or C-terminal domain of Spätzle protein using previously described methods(Morisato and Anderson, 1994). Rabbit antibodies generated against bacterially produced TrpE-Easter fusion protein (Chasan et al., 1992)were affinity purified against T7 protein 10-Easter fusion protein bound to Affigel 10/15 (BioRad).

The spätzle 1.9 kb cDNA(Morisato and Anderson, 1994)and eaΔN cDNA (Chasan et al., 1992) were expressed under the control of the metallothionein promoter in the pRMHa-3 vector (Bunch et al., 1988). The single base substitutions inea^83l and ea^5.13(Jin and Anderson, 1990) were introduced by the QuikChange (Stratagene) site-directed mutagenesis protocol.

Cultured Drosophila S2 cells were transiently transfected with a total of 4.0 μg DNA (final amount adjusted with vector DNA) using Cellfectin (Invitrogen). After transfection (5 hours), cells were incubated in serum-free medium overnight, and protein production was induced with 0.7 mM cupric sulfate. Conditioned medium was collected 20 hours later and analyzed for the presence of secreted Spätzle and EaΔN.

The embryonic phenotypes produced by many of the ea^Dmutations have been described previously at the level of cuticle patterns and gastrulation behavior (Chasan and Anderson,1989; Jin and Anderson,1990). We focused on five alleles that were classified as being ventralized (ea^125.3, ea^83l andea⁵⁰²²) or lateralized (ea²⁰ⁿ andea^5.13). These alleles are completely penetrant and show a spectrum of phenotypes, in contrast to the dorsalization caused by a loss of maternal easter function (Fig. 1B). Each ea^D mutation is caused by a single amino acid substitution in a conserved region of the Easter catalytic domain(Jin and Anderson, 1990).

Among the ventralizing alleles, embryos produced byea^125.3/+ females exhibited the weakest effects, with only slight expansion of ventral denticle bands(Fig. 1C)(Chasan and Anderson, 1989). Embryos laid by ea^83l/+ and ea⁵⁰²²/+females showed stronger phenotypes, characterized by the expansion of ventral denticle bands, reduction or absence of dorsolaterally derived structures,such as the filzkörper, and near absence of dorsal hairs(Fig. 1E,G)(Chasan and Anderson, 1989). Embryos laid by ea⁵⁰²²/+ females exhibited more disorganized denticles and more severe head deformities than embryos laid byea^83l/+ females. At gastrulation, embryos produced by females carrying all three alleles invaginated an apparently normal ventral furrow, but initiation of the lateral cephalic furrow was shifted to a more dorsal position and fewer dorsal folds were formed.

Embryos produced by females carrying the lateralizing allelesea²⁰ⁿ and ea^5.13 show a marked reduction in dorsoventral asymmetry. Unlike the other ea^Dalleles, ea²⁰ⁿ/+ females laid a mixture of ventralized and lateralized embryos, as determined by analyzing both differentiated cuticles and gastrulation movements. The ventralized embryos(Fig. 1I) showed a stronger phenotype than did embryos laid by ea^83l/+ andea⁵⁰²²/+ females. By comparison, all of the embryos laid by ea^5.13/+ females showed a reduction in both dorsal pattern elements and ventrally derived mesoderm, developing a cuticle with rings of ventral and lateral denticle bands(Fig. 1K)(Chasan and Anderson, 1989). At gastrulation, embryos from ea^5.13/+ females failed to invaginate a ventral furrow and exhibited a widened head fold.

We examined the phenotypes of embryos produced by ea^Dfemales in the absence of wild-type maternal easter activity, in order to study the genetic dominance exerted by these ea^Dmutations. For the ventralizing alleles, embryos produced byea^D/+ and ea^D/ea^-females were virtually indistinguishable(Fig. 1C-H). By contrast, the embryos produced by ea²⁰ⁿ/ea^- andea^5.13/ea^- females were markedly more elongated and had fewer ventral denticle bands than embryos laid byea²⁰ⁿ/+ and ea^5.13/+ females(Fig. 1I-L). Notably, all the embryos laid by ea²⁰ⁿ/ea^- females developed a lateralized head fold during gastrulation, when compared with the mixed population laid by ea²⁰ⁿ/+ females. In summary, the presence of a wild-type dose of easter is able to confer detectable dorsoventral asymmetry to embryos produced by the lateralizingea^D alleles.

Taken together, the analysis of gastrulation behavior and differentiated cuticles suggests that the five ea^D mutations can be ordered in the following allelic series, with the strongest allele exhibiting the greatest loss of dorsoventral polarity: wild type>ea^125.3>ea^83l>ea⁵⁰²²>ea²⁰ⁿ>ea^5.13.

Our primary interest in the ea^D mutations was to understand their effects on the Dorsal gradient. Although we could stain embryos directly for the expression of Dorsal protein, we felt that changes in the shape of the gradient would be too subtle to discern and difficult to quantitate. Therefore, we characterized the expression domains of the zygotic genes zen, sog, rho and twist, each corresponding to a specific concentration range of nuclear Dorsal (Figs2,3,4,Table 1). In order to simplify comparisons between embryos, we expressed domain size as a percentage of the embryo circumference (see Materials and Methods). The major points from these data are summarized below.

The zen gene is transcribed in the dorsal 38% of the wild-type embryo circumference, in those cells that lack nuclear Dorsal(Table 1)(Rushlow et al., 1987). In dorsalized embryos laid by ea^- females, zenexpression expands across the entire dorsoventral axis. By contrast, thezen domain is completely absent in embryos produced byea^D/+ and ea^D/ea^-females for both the ventralizing and lateralizing ea^Dalleles (Table 1). Thus, even in embryos produced by the weakest ventralizing alleleea^125.3, Dorsal protein is present in dorsal nuclei.

The sog gene is transcribed in two ventrolateral stripes that total 31% of the wild-type embryo circumference, with each stripe abutting the ventral domain defined by twist expression(Fig. 2A,B;Table 1)(François et al.,1994). In embryos produced by ea^125.3 females,each domain was expanded, such that the total sog domain occupied 50%of the embryo circumference (Fig. 2C,D; Table 1). In embryos produced by ea^83l and ea⁵⁰²²females, the sog domain was expanded across the dorsal midline(Fig. 2E-H), although in some of these embryos, the staining became weaker on the dorsal side; this region of lower expression might correspond to a concentration of nuclear Dorsal protein normally found in one or two nuclei at the dorsal sogboundary. Because the dorsal boundary was sometimes difficult to determine(see below), only the ventral unstained domain was quantitated in these embryos.

In embryos laid by females carrying the lateralizing alleles, sogstaining was significantly expanded. Consistent with the observations described above, a mixed population was observed among the embryos laid byea²⁰ⁿ/+ females; some embryos showed uniform sogstaining, while others maintained a ventral unstained domain(Fig. 2I). By comparison, all embryos laid by ea²⁰ⁿ/ea^- females showed uniform sog expression(Fig. 2J). Similarly, embryos produced by ea^5.13/+ andea^5.13/ea^- females showed sogstaining across the dorsoventral axis (Fig. 2K,L).

The quantitation of the sog domain showed that the slope of the Dorsal gradient was flattened in all of these mutant embryos. In addition, a small but significant change in the size of the presumptive mesoderm was detected by quantitating the size of the ventral unstained domain. Embryos laid by ea^83l/ea^- andea⁵⁰²²/ea^- females showed a decrease in the size of the ventral domain, when compared with the wild-type size seen in embryos laid by ea^83l/+ andea⁵⁰²²/+ females (Table 1).

The rho gene is transcribed in two ventrolateral stripes that total 18.5% of the wild-type embryo circumference(Fig. 3A,B;Table 1)(Bier et al., 1990). In contrast to the weak sog expression observed across the dorsal side in some mutant embryos, dorsal boundaries for the rho domain were well defined. The size of each rho domain was modestly expanded in embryos produced by females carrying the ventralizing allelesea^125.3, ea^83l andea⁵⁰²² (Fig. 3C-H; Table 1). The size of each rho domain was further increased in embryos laid byea²⁰ⁿ/+ females, while the two stripes were fused into a single domain in embryos produced byea²⁰ⁿ/ea^- females(Fig. 3I,J). Embryos produced by ea^5.13/+ andea^5.13/ea^- females expressedrho in all cells along the dorsoventral axis(Fig. 3K,L). As noted above for the quantitation of sog expression, the expansion of rhoexpression observed in embryos produced byea^83l/ea^- andea⁵⁰²²/ea^- females were each accompanied by a decrease in the ventral unstained domain.

The twist gene is transcribed in cells that give rise to mesoderm. The Twist protein, visualized by anti-Twist antibodies, is expressed in the ventral 21.5% of the wild-type embryo circumference(Table 1)(Thisse et al., 1988). The size of the Twist domain was nearly the same as in wild type in embryos produced by females carrying the ventralizing allelesea^125.3, ea^83l andea⁵⁰²² (Table 1). The Twist domain was slightly reduced in embryos laid byea²⁰ⁿ/+ females, while no Twist expression was detected in embryos laid by ea²⁰ⁿ/ea^- females(Table 1). A very faint,reduced Twist domain was observed among some embryos laid byea^5.13/+ females (data not shown), while no Twist expression was detected in embryos laid byea^5.13/ea^- females. In the wild-type embryo, the size of the Twist domain was in good agreement with the ventral domain also defined by the absence of sog and rhoexpression. In the ea^D mutant embryos, the Twist domain appeared slightly larger than the ventral domain when determined by examiningsog and rho transcription; some cells could be expressing low levels of both rho and Twist, as a consequence of a reduction in the slope of the Dorsal gradient. In no case was the ventral domain expanded in any of the embryos produced by the ea^D alleles.

A comparison of changes in the Dorsal gradient shape, as inferred from the expression domains of marker genes, is depicted inFig. 4. First, analysis ofzen expression showed that the dorsal domain, defined by the absence of nuclear Dorsal, was lost in all of the mutant embryos. Second, the domain of low nuclear Dorsal, reflected by the activation of sogtranscription, was expanded dorsally in all mutant embryos. This decrease in the slope of the Dorsal gradient was further characterized by the analysis ofrho expression, a marker corresponding to intermediate levels of nuclear Dorsal. Finally, a reduction in the size of the ventral domain of high nuclear Dorsal, inferred from Twist staining and the absence of sogand rho transcription, was observed in all mutant embryos, except for the weakest ventralizing allele ea^125.3. In moderately ventralized embryos, this reduction was small but significant; in lateralized embryos, the ventral domain was completely absent.

In order to address the mechanism underlying these changes in the Dorsal gradient, we asked if the ea^D mutations were affecting the production or inhibition of active Easter. Easter zymogen activation requires cleavage at a site between an N-terminal pro-domain and a C-terminal catalytic domain (Chasan et al., 1992). However, the cleaved protease domain is never detected in wild-type embryos. Instead, activated Easter is found in a stable complex called Ea-X that migrates as a 80-85 kDa band (Misra et al., 1998).

We prepared extracts from embryos produced byea^83l/ea^- andea^5.13/ea^- females, as representative of the ventralizing and lateralizing alleles. In order to generate a size marker for the C-terminal catalytic domain, we prepared extracts from embryos laid by eaΔN/+ andea⁸/ea^- females. In the N-terminal deletion mutant eaΔN, the pro-domain is deleted and the signal sequence is fused directly to the catalytic domain(Chasan et al., 1992). The embryos laid by females carrying a P[eaΔN] transgene are weakly ventralized. In embryos carrying the recessiveea⁸ allele, production of the Easter catalytic domain is observed rather than the Ea-X complex(Misra et al., 1998).

Easter forms were detected on an immunoblot probed with anti-Easter antibodies (Fig. 5). The amount of Easter zymogen correlated with easter dose; a higher level was observed in wild type compared with embryos laid by +/ea^-females (data not shown). As expected, the Ea-X band was present in the wild-type extract, appeared more prominent in the eaΔN/+ embryo extract and was absent in the ea⁸/ea^- embryo extract. In embryos produced by the ea^D mutations, the Ea-X band was reduced in the ea^83l/ea^- extract and virtually absent in the ea^5.13/ea^- extract. Significantly, there was a corresponding increase in the level of C-terminal catalytic domain. The amount of the catalytic domain in theea^5.13/ea^- extract appeared comparable with the level in the ea⁸/ea^- extract. These findings suggest that in the ea^D mutations, zymogen activation produces a catalytic domain that fails to be or is only partially inactivated by the inhibitor X.

The production of the C-terminal catalytic domain had previously been observed only with catalytically inactive easter mutants, exemplified by the ea⁸ allele, a missense mutation located in the presumptive substrate binding pocket, and the ea^S338Aallele, in which the active site serine-338 was replaced with alanine(Misra et al., 1998). We hypothesized that the C-terminal catalytic domain in theea^D alleles, in contrast to the recessive mutations,retained protease activity after zymogen activation. In order to test this model, we assessed Ea^D protease activity in two different experiments.

First, we determined the level of processed Spätzle inea^D embryo extracts, because Easter protease activity generates processed Spätzle (Morisato and Anderson, 1994; DeLotto and DeLotto, 1998). No processed Spätzle is produced inea^- extracts; a higher level of processed Spätzle is observed in a transformant line carrying the eaΔN mutation(Morisato and Anderson, 1994). Extracts were prepared from embryos laid by ea^D/+ andea^D/ea^- females. Precursor and processed forms of Spätzle were detected on an immunoblot probed with antibodies specific to the Spätzle C-terminal domain (Fig. 6). As expected, a lower level of processed Spätzle was observed in the +/ea^- embryo extract compared with the wild-type embryo extract. Similarly, the level of processed Spätzle was lower in the ea^D/ea^- embryo extract compared with the corresponding ea^D/+ embryo extract. In general,the amount of processed Spätzle in each of theea^D/ea^- lanes appeared roughly comparable with the level in the +/ea^- lane. The exception to this generalization was observed with the lateralizing ea^5.13allele. The amount of processed Spätzle observed in extracts prepared from embryos laid by ea^5.13/+ andea^5.13/ea^- females was reproducibly higher than in wild type.

Second, we used an injection assay to ask whether we could detect wild-type and ea^D activated Easter in the embryo. By transferring perivitelline fluid from one embryo to another, Stein et al.(Stein et al., 1991)characterized a ventralizing activity exhibiting properties expected of the Toll ligand. When this `polarizing activity', later identified to be processed Spätzle, was injected into the perivitelline space of recipient embryos laid by pipe^- females, the site of injection defined the ventral pole of a new axis (Stein et al.,1991; Stein and Nüsslein-Volhard, 1992;Schneider et al., 1994). We reasoned that by using donor embryos laid by spz^- females(and thereby removing the presumptive Toll ligand), we might be able to detect the activity responsible for generating processed Spätzle, i.e. active Easter.

We carried out perivitelline fluid transfer experiments from gastrulating donor embryos laid by different mutant females into stage 4 recipient embryos laid by pipe^- females. We did not detect axis-inducing activity from donor embryos laid by either spz^- orspz^- Toll^- females(Table 2), suggesting that active Easter is either rapidly inactivated or sequestered in a non-transplantable complex. Furthermore, we did not detect activity from donor embryos laid by ea^83l spz^- females. By contrast, we observed axis-inducing activity from donor embryos laid byea^5.13 spz^- females. This observation not only provided functional evidence for Ea^5.13 protease activity, but also demonstrated that the activity remained stable until gastrulation, many hours after zymogen activation.

In order to study the Spätzle cleavage reaction, we co-expressed precursor Spätzle and wild-type EaΔN proteins in culturedDrosophila S2 cells. We observed production of processed Spätzle in the conditioned medium, with the level of cleavage dependent on the amount of EaΔN expressed (Fig. 7A, lanes 3-5; Fig. 7B, lanes 2-4). We did not detect the formation of the inhibited EaΔN-X form in these transfected S2 cells.

We analyzed the effect of the ea^83l andea^5.13 mutations on the cleavage reaction. The level of Spätzle processing carried out by Ea^83lΔN was indistinguishable from wild-type EaΔN(Fig. 7A, lanes 6-8),suggesting that the ea^83l mutation does not affect Easter catalytic activity. By comparison, Ea^5.13ΔN exhibited weaker protease activity (Fig. 7A,lanes 9-11), although it appeared to be expressed at higher levels(Fig. 7B, lanes 8-10). This result suggests that the lateralized phenotype produced by theea^5.13 mutation arises from two separate effects: (1) lack of proper inhibition following zymogen activation leads to loss of the dorsalzen domain, and (2) reduced Easter catalytic activity prevents formation of the ventral twist domain (see Discussion).

We show that dominant ventralizing and lateralizing eastermutations cause profound changes in the shape of the Dorsal gradient, as visualized by the expression of zygotic marker genes. The increasing severity of ea^D phenotypes, initially determined by examining cuticle patterns and gastrulation movements, are correlated with a decrease in the slope of the Dorsal gradient. In both ventralized and lateralized embryos,the dorsal zen domain is absent, replaced by an expanded lateralsog domain. In lateralized embryos, the ventral twist domain is also absent, with the lateral sog domain expanded along the entire dorsoventral axis.

Ventralized embryos produced by grk^- andEgfr^- females exhibit an expansion of the ventral Twist domain at the expense of the dorsal domain, while maintaining a wild-type slope of the Dorsal gradient, as assessed by the size of the rho andsog domains (Schüpbach,1987; Roth and Schüpbach,1994; Morisato,2001). By contrast, in the ventralized embryos produced byea^83l/ea^- andea⁵⁰²²/ea^- females, the slope of the Dorsal gradient is flattened, leading to broader domains of rho andsog expression. This change is accompanied by a decrease, rather than an increase, in the size of the ventral Twist domain. Theea^D ventralized phenotype thus appears to arise from a redistribution of the ventral signal. This change in the shape of the Dorsal gradient is even more dramatic in lateralized embryos, leading to a loss of detectable dorsoventral polarity.

Monitoring the expression of target genes enabled the analysis of the Dorsal gradient, but also placed a limit on resolution. For example, the phenotypes of embryos laid by ea²⁰ⁿ/ea^- andea^5.13/ea^- females initially appeared identical when assessed by sog RNA expression(Fig. 2J,L). A more refined image of the shape of the Dorsal gradient emerged after monitoringrho RNA expression (Fig. 3J,L), which responds to a narrower concentration range of nuclear Dorsal. Although embryos laid by ea^5.13/ea^-females appear symmetric, it remains formally possible that residual polarity could be detected if a marker corresponding to an even narrower range of nuclear Dorsal were available.

Embryos produced by ea^125.3/ea^-,ea^83l/ea^-, ea⁵⁰²²/ea^- andea²⁰ⁿ/ea^- females show varying degrees of dorsal-ventral asymmetry. The presence of a wild-type dose of eastercauses a slight expansion of the Twist domain and a slight reduction in therho domains, thereby producing a shift towards the normal shape of a Dorsal gradient. The Dorsal gradient in embryos laid byea^D/+ females reflects a partial contribution from eacheaster allele, rather than a simple superimposition of the gradient shapes observed in embryos laid by ea^D/ea^- and+/ea^- females. This behavior could be explained if only a set amount of Easter zymogen (wild type and mutant combined) could be cleaved by a limiting amount of activated Snake.

Although embryos laid by ea^5.13/+ andea^5.13/ea^- females can be distinguished by their cuticles (Fig. 1K,L), the expression of marker genes in these embryos is quite similar. In particular,formation of the Twist domain is largely inhibited in embryos laid byea^5.13/+ females. It remains to be determined whether the stronger dominance exhibited by ea^5.13 can be explained by the simple dose argument presented above, or whether Ea^5.13 is interfering with the proper formation of the Dorsal gradient by a more active mechanism.

The experiments described above suggest how the spectrum of phenotypes observed in ventralized and lateralized embryos can be explained by separately considering two distinct properties of the Easter protein: (1) inactivation by inhibitor X; and (2) Easter protease activity.

In both ventralized and lateralized embryos, the shape of the Dorsal gradient is altered by the absence of the dorsal zen domain, which is replaced by an expanded sog domain. Our studies suggest that this phenotype arises from the failure of activated Easter to be properly regulated after zymogen cleavage: the protease domain fails to form a complex with X and remains active (Fig. 5). This interpretation is consistent with earlier studies that characterized the effects of changing ea^D dose. Injection ofea^83l and ea^125.3 RNA intoea^- embryos produced a ventralized phenotype, while lower levels of the same RNA rescued to hatching, suggesting that the Easter produced by these ventralizing alleles were defective in negative regulation(Jin and Anderson, 1990).

The stability of the Ea-X complex suggested that X might be a serpin,reacting with the active site serine of Easter(Misra et al., 1998). Most mutations that map in the Easter catalytic domain would be expected to affect both protease activity and the interaction with inhibitor X. We suggest that the ventralizing ea^D alleles (as exemplified byea^83l) form a special class of mutations that retain catalytic activity, but affect inhibition by X. These studies imply that regulation of Easter following zymogen activation is required for maintaining polarity during formation of the Dorsal gradient. If activated Easter were capable of diffusion, X may primarily play a kinetic role to maintain the initial asymmetry of zymogen activation, by inhibiting activated Easter before its diffusion to the dorsal side.

In lateralized embryos, the Dorsal gradient appears even less polar. Both the dorsal zen and the ventral twist domains are absent,with the lateral sog domain expanded along the entire dorsoventral axis. The experiments described above suggest that in addition to their failure to be inactivated by X, the lateralizing alleles also partially reduce Easter protease activity. As the ea^D mutations map to conserved regions within the protease domain, it is not surprising that some of these alleles show effects on catalytic activity. This point is most clearly observed for the case of the Ea^5.13ΔN protein, which exhibits less Spätzle processing activity than wild-type EaΔN upon expression in cultured S2 cells (Fig. 7). In the embryo, this weaker Ea^5.13 protease is apparently unable to generate the level of processed Spätzle required for nuclear translocation of Dorsal that leads to twisttranscription.

The final amount of processed Spätzle generated in the embryo depends on both Easter specific activity and the length of time the enzyme remains active. In the case of embryos produced by ea^5.13 females,perivitelline transfer experiments detected stable Ea^5.13 activity many hours after cellularization (Table 2). Despite the lower specific activity of Ea^5.13suggested by the S2 experiments (Fig. 7), the prolonged time of Easter action results in a higher level of processed Spätzle in embryos (Fig. 6). Similarly, the ea²⁰ⁿ mutation appears to cause a significant decrease in specific activity, as suggested by a reduction in the combined size of the domains expressing rho RNA and Twist in embryos produced by ea²⁰ⁿ/ea^- females compared with embryos laid by other ea^D/ea^- females(Fig. 4 andTable 1). Yet, nearly wild-type amounts of processed Spätzle are observed in ea²⁰ⁿembryo extracts (Fig. 6),probably because Ea²⁰ⁿ fails to be quickly inactivated by inhibitor X.

At a broader level, the analysis of ea^D mutations underscores the important relationship between timing of signal production and generation of spatial pattern. Previous studies demonstrated that cleavage of Spätzle is required by Easter for Toll activation. The experiments described here refine this model by showing that the wild-type shape of the Dorsal gradient requires high levels of Easter protease activity during a brief period of time. Failure to properly inactivate Easter leads to a loss of the dorsal domain of the axis, thereby leading to a ventralized phenotype. When this defect is coupled with a reduction in Easter protease function,wild-type levels of processed Spätzle are eventually produced in the embryo, but the shape of the Dorsal gradient becomes more symmetric.

We gratefully acknowledge D. Goff for her generosity in providing scientific advice and sharing reagents. We thank M. Byrne, S. Alvares and D. King for experimental assistance, and G. Denton, D. Goff, C. Hashimoto and S. Misra for helpful comments on the manuscript. This work was supported by grant GM 52084 from the National Institutes of Health.