Interallelic Complementation at the Drosophila melanogaster gastrulation defective Locus Defines Discrete Functional Domains of the Protein

Corresponding author: Robert DeLotto, Department of Genetics, University of Copenhagen, Copenhagen K, Denmark., rdelotto{at}biobase.dk (E-mail)

Communicating editor: A. J. LOPEZ

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The gastrulation defective (gd) locus encodes a novel serine protease that is involved in specifying the dorsal-ventral axis during embryonic development. Mutant alleles of gd have been classified into three complementation groups, two of which exhibit strong interallelic (intragenic) complementation. To understand the molecular basis of this interallelic complementation, we examined the complementation behavior of additional mutant alleles and sequenced alleles in all complementation groups. The data suggest that there are two discrete functional domains of Gd. A two-domain model of Gd suggesting that it is structurally similar to mammalian complement factors C2 and B has been previously proposed. To test this model we performed SP6 RNA microinjection to assay for activities associated with various domains of Gd. The microinjection data are consistent with the complement factor C2/B-like model. Site-directed mutagenesis suggests that Gd functions as a serine protease. An allele-specific interaction between an autoactivating form of Snake (Snk) and a gd allele altered in the protease domain suggests that Gd directly activates Snk in a protease activation cascade. We propose a model in which Gd is expressed during late oogenesis and bound within the perivitelline space but only becomes catalytically active during embryogenesis.

DORSAL-VENTRAL polarity of the Drosophila embryo is established during embryogenesis by the interpretation of positional information that is placed in the egg during oogenesis. Consequently, the process of specifying dorsal-ventral cell fate can be viewed as occurring in two distinct phases: an early one of producing a stable asymmetric cue in the mature egg and a later one involving interpretation of that positional cue at the syncytial blastoderm stage. Both phases of this process rely extensively upon signal transduction mechanisms (STEIN et al. 1991 ; ROTH et al. 1995 ). The establishment of polarity in the egg involves reciprocal signaling between the somatic follicle cells and the germ-line-derived ooplasm and requires the products of the genes gurken, a TGF- homolog, and torpedo, the Drosophila epidermal growth factor receptor (SCHUPBACH 1987 ; CLIFFORD and SCHUPBACH 1992 ; NEUMAN-SILBERBERG and SCHUPBACH 1993 ). Signaling during oogenesis serves to suppress ventral cell fate on the dorsal side of the developing oocyte. Further downstream in the pathway, the products of the genes nudel, windbeutal, and pipe are required to create a highly stable asymmetric cue that polarizes subsequent patterning during embryo-genesis (reviewed by MORISATO and ANDERSON 1995 ).

The second phase, that of interpretation of the cue during embryogenesis, involves another signal transduction pathway in which a signal originates within the perivitelline space (PVS) of the embryo (STEIN et al. 1991 ). There, a ventrally restricted extracellular signal results in the activation of the plasma membrane receptor Toll on the ventral side of the syncytial blastoderm embryo. This asymmetric activation of Toll directs the formation of a gradient of the transcription factor, Dorsal, along the dorsal-ventral axis of the cellular blastoderm embryo (STEWARD et al. 1988 ; ROTH et al. 1989 ; RUSHLOW et al. 1989 ). As a result, blastoderm nuclei in ventral positions acquire higher concentrations of Dorsal protein while nuclei at more dorsal positions acquire lower concentrations. As a function of their nuclear Dorsal concentrations, blastoderm cells will differentiate with correct dorsal-ventral cell fates according to their relative position.

An outstanding question is how the positional information laid down in the egg is converted into the ventrally restricted signal within the PVS after fertilization (reviewed by ROTH 1994 ). Part of the mechanism appears to involve a proteolytic activation cascade (DELOTTO and SPIERER 1986 ; CHASEN and ANDERSON 1989 ). Several serine proteases sequentially activate each other in a proteolytic activation cascade and ultimately process Spaetzle (Spz), a precursor of an NGF-like growth factor ligand (MORISATO and ANDERSON 1994 ; DELOTTO and DELOTTO 1998 ; DISSING et al. 2001 ; LEMOSY et al. 2001 ). Epistasis data from microinjection experiments place another gene product, that of the gd gene, upstream of snake (snk) in the pathway (SMITH and DELOTTO 1994 ).

The molecular cloning of gd showed it to encode a protein with homology to the serine proteinase superfamily with, however, several structural features that are unusual for members of that superfamily (KONRAD et al. 1998 ). A structural model for Gd, which is consistent with it being a functional serine protease with a structure analogous to the mammalian complement factors C2 and B, has been proposed (DELOTTO 2001 ). Recent biochemical studies indicate proteolytic activity for Gd (DISSING et al. 2001 ; LEMOSY et al. 2001 ). Microinjection of in vitro-synthesized gd wild-type mRNA was shown to rescue the dorsalized phenotype of embryos derived from gd null homozygous females. Furthermore, the dosage of Gd had a primary effect upon the magnitude of the ventralizing signal (DELOTTO 2001 ).

gd, a genetically complex locus at cytological location 11A, was first described in a genetic screen for maternal effect genes affecting embryonic development (MOHLER 1977 ). Twelve mutant alleles were later characterized in an extensive genetic analysis and the data clearly indicated complex interallelic complementation (KONRAD et al. 1988A , KONRAD et al. 1988B ). While all mutant alleles were fully recessive and allelic to gd by virtue of their failure to complement some alleles of gd, genetic complexity was evident in certain heterozygous combinations. For example, both gd²/gd² females and gd¹⁰/gd¹⁰ females produced completely dorsalized eggs and are amorphic by virtue of no change in phenotypic strength when either gd² or gd¹⁰ is placed over Df(1)KA10, a deficiency overlapping gd. However, gd²/gd¹⁰ females produced phenotypically normal first instar larvae, some of which developed to adults. Clonal analysis revealed that gd function is required in the germ line and the analysis of temperature-sensitive (ts) alleles revealed a broad ts period, which includes the last 4–5 hr of oogenesis and extends into the first 1.5–2 hr of embryogenesis. This contrasts with ea, which during embryogenesis has a ts period between 0 and 3 hr after fertilization (ANDERSON and NUSSLEIN-VOLHARD 1984 , ANDERSON and NUSSLEIN-VOLHARD 1986 ). This would suggest that while Gd is needed both during late oogenesis and into early embryogenesis, Ea appears to be required only during embryogenesis. This temporal difference in the requirements of gd and ea is puzzling in light of the biochemical demonstration that Gd and Ea function via Snk in a sequential proteolytic cascade during embryogenesis (DISSING et al. 2001 ).

The structural model for Gd suggests that it is similar to the mammalian complement factors C2 and B, two unusual serine proteases involved in the classical and alternative pathways of activation of the complement cascade (DELOTTO 2001 ). Almost all serine proteases are activated by cleavage at a region called the activation peptide, usually resulting in an isoleucine or valine at the amino terminus of the active catalytic chain (reviewed by STROUD et al. 1977 ). Indeed, Snk and Ea have such a stereotypical structure (DELOTTO and SPIERER 1986 ; CHASEN and ANDERSON 1989 ). Gd lacks significant homology to the serine protease superfamily in the activation peptide region (KONRAD et al. 1998 ). However, it does have an arginine lysine pair followed by homology to a vonWillebrand factor type A (vWF) motif within the polypeptide chain. The vWF domain constitutes a binding site for an activating protease in complement factors C2 and B (OGLESBY et al. 1988 ). In complement factors C2 and B, activation results in a cleavage between the arginine and lysine residues to generate two distinct polypeptide chains (reviewed by ARLAUD et al. 1998 ).

To understand the basis of the interallelic complementation at gd, we extended the complementation analysis to more recently isolated gd alleles and determined the molecular lesions associated with alleles in all complementation groups. To test the C2/B structural model of Gd, we used an SP6 RNA microinjection assay to assay for activities of discrete domains of Gd. Our data support processing of Gd in a manner consistent with the complement C2/B-like model. An allele-specific interaction using RNA microinjection is consistent with biochemical data indicating that Gd directly activates Snk in a proteolytic cascade. Finally, we propose a biological model in which an inactive form of Gd is bound to a membrane within the PVS late during oogenesis but only becomes activated later at syncytial blastoderm stage. Such a model can account for the earlier-starting ts period described for gd as well as accommodate a direct role of Gd in activating the Snk zymogen in the protease cascade.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly strains used:
Wild-type strains were Oregon-R. The genotypes and sources of the alleles are as follows. gd¹ v²⁴/FM3, gd² v/FM3, gd³ v/FM3, gd⁴ v/FM3, gd⁵ v/FM3, gd⁶ v/FM3, gd⁷ v/FM3, and gd¹⁹⁰/FM7, l(1) TW9 were from the Tübingen stock collection. gdⁿ²⁷ y/FM7, gd^p18 y/FM7, gd⁹ v/FM3, and gd¹⁰ v/FM3 were provided by K. Konrad. gd alleles prefixed by V, T, and L, marked with y w P[w+FRT], and balanced over FM7c were isolated and kindly provided by T. Schüpbach. Df(1)KA10 sn³ m/FM7c was from the Bloomington stock collection.

Cuticle preparations and complementation analysis:
The collection of embryos, dechorionation of embryos, and preparation of cuticles were performed as described (WIESCHAUS and NUSSLEIN-VOLHARD 1986 ). Slides of prepared cuticles were examined under darkfield or phase microscopy.

rhomboid in situ hybridization of Drosophila embryos:
In situ hybridizations were performed as described (ROTH 1994 ; ROTH and SCHUPBACH 1994 ). rho cDNA was a generous gift from E. Bier, and the probe was made according to the Boehringer-Mannheim (Roche, Indianapolis) protocol using the SP6/T7 DIG RNA labeling kit. After staining, embryos were cleared and mounted with Permount (Fisher, Pittsburgh).

Cloning of the gd mutant alleles:
Genomic DNA was prepared as follows: 50–100 males of the appropriate stock were homogenized in TE and phenol (pH 8) 1:1. Nucleic acids were isopropanol precipitated, dried, and resuspended in 50–100 ml of TE. DNA (0.5 ml) was added to a 25-ml PCR reaction with the 5' gd genomic and 3' gd genomic flanking primers listed in Table 1. After PCR, the fragments were digested with EcoRI and BglII restriction enzymes, gel purified, and cloned into pGEM3. For the generation of cDNA clones from wild type and the gd mutant alleles, 0- to 2-hr embryos or homozygous females were homogenized in a Dounce with 0.5 ml of 4 M guanidinium isothiocyanate, 5 mM dithiothreitol, and 0.4 ml acid phenol. After two acid phenol/chloroform extractions and one chloroform extraction, nucleic acids were precipitated and poly(A)⁺ RNA was purified as described in SAMBROOK (1989). RNA (3 mg) was used in an RT-PCR reaction (as described in DELOTTO et al. 2001 ) using the 5' gd genomic and 3' gd genomic flanking primers listed in Table 1. The PCR product was digested with EcoRI and BglII and ligated directionally into SP64-T. DNA sequencing reactions were performed on double stranded plasmid DNA using Sequenase Version 2 (USB) according to the manufacturer's protocols using primers listed in Table 1.

gd domain constructs:
gdn_e was made by fusing the signal peptide sequence of Ea to the vWF and serine protease domains of gd as described in HIGUCHI (1990). The outside primers were 5' easter BglII and 3' gd genomic. The inside primers were 5' easter-gd cat and 3' gd cat-easter. The starting templates were ea cDNA (pGEM3) and gdcD7 (DELOTTO 2001 ). The final recombinant PCR fusion was cut with BglII and EcoRI and cloned into pSP64. gdpro and the gdvWF were made by insertion of a stop codon using inverted PCR on the starting plasmids of wild-type full-length gdcD7 and gdn_e, respectively. The primers used in the reverse PCR were gd seq7 and gd3' pro-enz and gd seq7 and 3' gdvWFe, respectively. The gd propolypeptide PCR product was cut with NotI and self-ligated, and the gd vWF domain PCR product was cut with EcoRI and self-ligated.

mRNA microinjection of transcripts into Drosophila embryos:
The mRNA transcripts were synthesized and microinjections performed as previously described (SMITH and DELOTTO 1994 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Analysis of the phenotypic strength of gd alleles:
Screens for maternal effect mutations in various laboratories have generated new alleles of gd (see MATERIALS AND METHODS). We wished to characterize the phenotypic strength of these alleles using the same criteria used in earlier studies (KONRAD et al. 1988A , KONRAD et al. 1988B ). To distinguish between weak hypomorphic and amorphic alleles, we compared the phenotype of embryos produced by females homozygous for an allele to that of females with the allele over Df(1)KA10, a deficiency overlapping gd. The phenotypic strength was quantified by scoring embryos into one of four classes. The classes of phenotypic strength employed here were also used in earlier studies of gd (KONRAD et al. 1988B ). Class I embryos are strongly dorsalized, lacking both ventral denticles and filzkorper. Class II embryos lack ventral denticles but have filzkorper. Class III embryos are distinguished by the presence of ventral denticles and Class H consists of phenotypically normal embryos, which hatch to larvae. Table 2 shows the results of this analysis for all of the gd alleles that we characterized. Overall, our results agree with those previously described for several alleles and the newly isolated alleles fell into the same phenotypic series, varying between amorphic to almost wild type (KONRAD et al. 1988A , KONRAD et al. 1988B ). We conclude that gd², gd⁴, gd⁷–gd¹⁰, gd^VM90, gd^VO27, gd^LF12, gd^Li115, and gdⁿ²⁷ are amorphic alleles; gd¹, gd³, gd⁶, gd^TN124, and gd^LQ4 are moderate hypomorphic alleles; and gd⁵, gd¹⁹⁰, gd^Lu119, and gd^p18 are weak hypomorphic alleles.

Interallelic complementation analysis of gd alleles:
Previous work revealed the existence of three groups of alleles at the gd locus (MOHLER 1977 ; KONRAD et al. 1988A , KONRAD et al. 1988B ). Two groups of alleles exhibited interallelic complementation whereas the third group failed to show any significant interallelic complementation. For the purpose of this analysis, we designated the two groups exhibiting interallelic complementation the gd² group and the gd¹⁰ group, since these two alleles previously exhibited the most dramatic interallelic complementation. The third group, which failed to complement either of the other two groups, we called the noncomplementing group. We performed an interallelic complementation analysis to determine whether the more recently isolated gd alleles could also be assigned to these three complementation groups. This was achieved by crossing each allele to gd² and gd¹, two members of the gd² group, gd¹⁰ and gd⁶, two members of the gd¹⁰ group, and gd⁹, a member of the noncomplementing group. The phenotypic strength of the embryos produced by trans-heterozygous females was compared to the sum of the phenotypic strength of the two alleles in a hemizygous state. If the resulting phenotype was significantly shifted toward wild type, the alleles were scored as demonstrating interallelic complementation. The data are presented in Table 3. We found that all of the more recently isolated alleles of gd could be classified into either the gd² group, the gd¹⁰ group, or the noncomplementing group. For example, while both gd^vo27/Df(1)KA10 and gd²/Df(1)KA10 produce only embryos in class I (Table 2), gd²/gd^VO27 produces 21/100 embryos in class I, 22/100 embryos in class II, and 57/100 embryos in class III (Table 3). Because it complements gd² as well as gd¹, we determined that gd^VO27 is a member of the gd¹⁰ complementation group. On the basis of this type of analysis, we concluded that gd¹–gd³, gd⁵, gd^TN124, gd^Lu119, gd^LQ4, gd^p18, and gd¹⁹⁰ are in the gd² group, gd⁶ and gd^VO27 are in the gd¹⁰ group, and gd⁴, gd⁷, gd⁸, gd⁹, gd^Li115, gd^LF12, gdⁿ²⁷, and gd^VM90 are in the noncomplementing group.

Molecular alterations associated with gd mutant alleles:
To determine the nature of the changes in mutant alleles of gd, both genomic DNA and cDNA were isolated from each of 20 mutant alleles and subcloned. The nucleotide sequence of both the genomic DNA and cDNA was determined and these results are summarized in Table 4. Most alleles revealed deviations from the wild-type nucleic acid sequence that could be correlated with the mutant phenotype. gd¹⁹⁰, a weak hypomorphic allele, has a single base change in the second intron, which alters splicing site usage and results in the generation of alternatively spliced transcripts at intron 2 (data not shown). For the alleles gd¹, gd⁵, and gd^p18, which are hypomorphic or weakly hypomorphic, no nucleotide changes were detected within either the protein coding sequences or within the introns. The positions of the observed changes relative to the protein structure are summarized in Fig 1. Noncomplementing alleles include lesions mapping throughout the protein coding region and, with the exception of gd⁷, they contain either a premature stop codon or deletion of part of a coding region. Only gd⁷ is predicted to generate a protein with single amino acid substitution. The lesions in the alleles gd^TN124, gd², gd^Lu119, gd³, and gd^LQ4, members of the gd² complementation group, are clustered within the proenzyme polypeptide after the signal peptide sequence and amino proximal to R₁₃₆ and K₁₃₇, which in the complement factor C2/B model correspond to the activation cleavage site. All of these lesions result in single amino acid changes within the predicted propolypeptide domain of Gd. The gd¹⁰ complementation group comprising gd⁶, gd¹⁰, and gd^VO27 has lesions resulting in single amino acid changes located in close proximity to S₄₆₈, the predicted active site serine of Gd. The proximity of these lesions to S₄₆₈ suggests that they could affect Gd function by compromising proteolytic activity of the catalytic chain. Thus alterations in gd² group members map to the presumptive proenzyme polypeptide of the C2/B model and those within the gd¹⁰ group map to the catalytic chain.

View larger version (15K): In this window In a new window Download PPT slide   Figure 1. Location of the molecular lesions in GD. Noncomplementing group alleles map throughout the protein coding region. gd2 group alleles fall within the presumptive propolypeptide region. gd10 group alleles lie close to the active site serine. Both gd2 and gd10 lesions change amino acids while most of the noncomplementing alleles introduce stops or have deletions.

Site-directed mutagenesis of gastrulation defective:
To test whether amino acid residues critical to serine protease catalysis are required for biological activity of Gd, we altered specific amino acid residues within the coding region of the molecule and used SP6 RNA microinjection to assay for changes in phenotypic rescue. Embryos from gd⁹/gd⁹ females are completely dorsalized (Table 5). As previously reported, injection of wild-type gdcD7 RNA into embryos from gd⁹/gd⁹ females locally ventralizes and generates open ventral denticles (DELOTTO 2001 ). We tested whether the presumptive catalytic H₂₉₂ is required by converting it to arginine. As shown, H292R completely eliminated the ability of Gd to restore ventral cuticular features. Alteration of S₄₅₉, which is located near the presumptive catalytic serine, to alanine had little discernible effect upon the ability of Gd to rescue. However, alteration of S₄₆₈, the presumptive catalytic serine, to alanine, abolished the restoration of any ventral pattern elements. This last observation is consistent with the mutagenesis of S₄₆₈ leading to loss of a Gd-derived tryptic fragment binding to protease inhibitors and rescue activity (HAN et al. 2000 ). We conclude that two of the amino acids involved in the serine protease catalytic triad are critical to biological activity and these results are consistent with Gd functioning biochemically as a serine protease.

Analysis of functional domains proposed by the complement factor C2/B model:
If the complement factor C2/B model is correct, a cleavage between R₁₃₆ and K₁₃₇ should produce two polypeptides, an amino-terminally derived propolypeptide and a carboxy-terminally derived catalytic chain. In C2 and B, a domain directly following the cleavage site contains a vonWillebrand factor type A repeat motif, which functions as a binding site for upstream activators. To test the C2/B model, we generated SP6 RNA expressing the full-length Gd protein (wild type), the presumptive catalytic chain (gdn_e), the presumptive proenzyme polypeptide (gdpro), or a polypeptide comprising the vonWillebrand factor type A homology (gdvWF). The constructs are illustrated in Fig 2. RNA was microinjected into embryos from various gd allelic backgrounds and phenotypic rescue was scored. The results are summarized in Table 6.

View larger version (12K): In this window In a new window Download PPT slide   Figure 2. Diagram of constructs for microinjection. Three deletions relative to the wild-type GD structural gene are illustrated. gdne fuses the ea signal peptide to the presumptive catalytic chain. gdpro truncates at the activation cleavage site. gdvWF begins at the activation cleavage site and terminates at the beginning of homology to serine protease catalytic domains. See MATERIALS AND METHODS for details.

Embryos from amorphic gd⁹/gd⁹, gd²/gd², and gd¹⁰/gd¹⁰ females are completely dorsalized and exhibit no filzkorper (see Fig 3A). When wild-type gd RNA is microinjected at >50 µg/ml concentration, the resulting embryos are partially ventralized and exhibit split ventral denticles near the site of injection (Fig 3B). When gdn_e RNA was microinjected into embryos from either gd⁹/gd⁹ or gd²/gd² females, no change in dorsal-ventral cell fate was observed and the embryos remained dorsalized. However, when gdn_e RNA was injected into gd¹⁰/gd¹⁰ embryos, 54% of the embryos exhibited either filzkorper or filzkorper and ventral denticles and the gastrulation pattern exhibited polarity. In many embryos the overall cuticular pattern was very close to wild type (see Fig 3C). gd¹⁰ has a lesion in the presumptive catalytic chain and this result suggests that providing this fragment of Gd as a separate polypeptide can rescue lesions in the catalytic chain.

View larger version (94K): In this window In a new window Download PPT slide   Figure 3. Microinjection phenotypes of embryos. (A) An embryo from a gd9/gd9 female. (B) An embryo from a gd9/gd9 female injected with wild-type gd SP6 RNA. (C) An embryo from a gd10/gd10 female injected with gdne RNA. (D) An embryo from a gd10/gd10 female injected with XaSnk RNA showing restoration of filzkorper, ventral denticles, and head skeleton. (E) An embryo from a gd10/gd10 female injected with XaSnk RNA showing uniform width of ventral denticles along the anterior-posterior axis.

When gdpro was injected into embryos from either gd²/gd² or gd¹⁰/gd¹⁰ females, no rescue of the dorsalized phenotype was observed. Since we might have expected that gdpro would rescue embryos from gd²/gd² females, we tested for activity of gdpro in wild-type embryos. When gdpro was injected into wild-type embryos, none of the embryos hatched and all of the embryos were partially dorsalized, revealing a dominant negative effect of this polypeptide (see Table 6). This suggests that, while gdpro cannot rescue lesions within the proenzyme polypeptide region, nevertheless, it can affect the patterning system in wild-type embryos. This may occur by binding and competing with some component of the patterning system to downregulate the ventralizing signal. To test whether the activity of full-length Gd could be reconstituted when both the propolypeptide and catalytic chain are provided in trans, gdn_e and gdpro were microinjected simultaneously into gd⁹/gd⁹ embryos. In these injections no phenotypic rescue was observed, suggesting that they cannot be supplied as two separate polypeptides to rescue gd⁹.

To test the role of the von Willebrand type A motif domain, gdvWF was microinjected into wild-type embryos. No embryos hatched and all of the embryos were to some extent dorsalized, indicating a dominant negative effect of the vonWillebrand factor type A homology domain. To determine how gdvWF alters the Dorsal gradient, we microinjected gdvWF RNA into wild-type embryos and examined the expression of rhomboid (rho), a marker for the slope and position of the Dorsal gradient, using in situ hybridization (Fig 4). In wild-type embryos rho is expressed bilaterally as two stripes 6–8 cells wide in a region corresponding to ventral neurogenic ectoderm (Fig 4A; BIER et al. 1990 ). Since it is expressed only in response to intermediate levels of Dorsal, it is a sensitive indicator of the position and slope of the Dorsal gradient (ROTH and SCHUPBACH 1994 ). When gdvWF RNA is injected into the posterior of a wild-type embryo, the rho stripe is widened and shifted ventrally (Fig 4B). This reveals that expression of the vWF domain has the effect of dorsalizing and also making the Dorsal gradient shallower. We conclude that the gdvWF domain competes with components of the cascade that are normally required for the regulation of Gd and proper formation of the Dorsal gradient.

View larger version (138K): In this window In a new window Download PPT slide   Figure 4. Effect of gdvWF upon wild-type embryos. (A) Rhomboid in situ hybridization showing ventrolateral rhomboid stripes in a wild-type embryo. (B) Rhomboid in situ hybridization of a wild-type embryo microinjected in the posterior pole with gdvWF RNA showing broadening and ventral displacement of rhomboid stripe. In both A and B, posterior is to the right and ventral is at the bottom.

Allele-specific interaction with an autoactivating form of Snake:
To determine whether Gd directly activates the zymogen form of Snk, we tested for an allele-specific interaction between Gd and Snk. We have previously described a novel form of Snk called XaSnake, which has been mutated in its activation peptide region in such a fashion that the activation peptide sequence resembles its own substrate specificity (SMITH et al. 1995 ). The consequence of this mutation is to generate an autoactivating form of Snk. However, microinjection of XaSnake does not result in a dominant ventralizing effect and does not bypass the requirement for upstream functions as was observed with a different construct called snkn_e (SMITH and DELOTTO 1994 ). Thus XaSnake, while autoactivating, still requires interaction with upstream components to properly exert its function. If Gd is the direct upstream activator of Snk, we reasoned that microinjection of XaSnake into an embryo from a gd²/gd² female would not rescue since a presumptive regulatory function of Gd should be affected in this case. However, XaSnake might rescue embryos from gd¹⁰/gd¹⁰ females since only the activity of the catalytic chain is predicted to be compromised in this case, while the regulatory function of the proenzyme polypeptide should be normal. As expected, when XaSnake RNA was injected into embryos from gd²/gd² females, no phenotypic rescue was observed (Table 6). However, when XaSnake was injected into embryos from gd¹⁰/gd¹⁰ females, 54% of the cuticles exhibited signs of restoration of ventralizing signal and the gastrulation pattern was normally polarized (Fig 3E). This allele-specific interaction between XaSnake and gd¹⁰ is consistent with biochemical evidence demonstrating GD to be a direct activator of Snk in the protease cascade (DISSING et al. 2001 ; LEMOSY et al. 2001 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The data presented here suggest an explanation for the interallelic complementation observed within the gd² and the gd¹⁰ complementation groups. This interpretation is also consistent with the complement factor C2/B-like structural model. This C2/B model proposes that Gd, in the process of becoming activated, is cleaved into two separate polypeptide chains. Since all mutations in alleles of the gd² group map to the presumptive proenzyme polypeptide, these alterations are expected to specifically alter the activity of the propolypeptide chain. In most serine proteases, this polypeptide is involved in regulating the activity of the catalytic chain by modulating interactions with cofactors and other components of activation complexes in protease activation cascades. The alleles comprising the gd¹⁰ complementation group have lesions in close proximity to the putative active site serine residue, predicting that they will disrupt the activity of the catalytic chain. Grouping of the alterations to these two regions of Gd would suggest that each part of the Gd protein has an independent biochemical activity.

Whereas interallelic complementation is often due to dimerization or multimerization of a protein, we do not favor this interpretation to explain the interallelic complementation observed at the gd locus. First, we have looked for dimerization of Gd using recombinant forms of the protein expressed using the baculovirus system and have found no evidence of either covalent or noncovalent dimers to date (M. DISSING and R. DELOTTO, unpublished results). Second, if proteolytic cleavage at the arginine-lysine pair is part of the normal activation mechanism, then two separate polypeptides would be generated. Complementation would then arise by each allele providing a functional polypeptide consisting of either the catalytic or propolypeptide chain. These two polypeptides appear to have independent biochemical functions and might function sequentially. Alternatively, they may be involved in formation of a larger multiprotein activation complex, something for which there is strong precedent among the coagulation and complement proteases (MANN et al. 1988 ). In this case we might expect that after activation and cleavage two functional polypeptides may associate in an activation complex that can initiate the protease cascade and direct processing of Ea and Spz.

The genetic data indicate that functions mapping to each putative polypeptide chain can in some cases be provided in trans to restore normal function to the system. Microinjection of gdn_e into embryos from gd¹⁰ females restores the dorsal-ventral pattern with correct polarity with respect to the asymmetry of the egg. This indicates that the function of the carboxy-terminal catalytic chain may be provided as late as stage 2 of embryonic development. Microinjection of the propolypeptide into embryos from gd²/gd² females did not result in rescue. This may be due to the inability of gdpro to displace a nonfunctional form of the Gd propolypeptide from an activation complex within the PVS. Transplantation experiments with perivitelline fluid revealed activities for Snk, Ea, and Spz, although they failed to find Gd activity (STEIN and NUSSLEIN-VOLHARD 1992 ). This result can be interpreted to indicate that Gd is bound to receptors within the PVS at early times during embryonic development. Perhaps gdpro contains binding sites for such a receptor. gdpro can also dorsalize a wild-type embryo when introduced at stage 2. This dominant negative effect indicates that the Gd propolypeptide can interfere with the biochemical pathway in such a way as to reduce the level of the ventralizing signal. However, this competitive effect might be due to interaction with something other than a membrane receptor for Gd and might exert its effect by complexing with and rendering inaccessible the proenzyme of Snk. Similarly, gdvWF, which comprises the vonWillebrand type A homology region, is also able to produce a dominant negative effect when introduced into wild-type embryos. We take these data to indicate that other components of the biochemical pathway interact with this region of the presumptive catalytic chain of Gd. The RNA microinjection result is consistent with the idea that this region might be involved in binding of a positive modulator of activation. However, biochemical aspects of the regulation of Gd are clearly complex and further biochemical analysis of the protein will be necessary.

The allele-specific interaction between gd¹⁰ and XaSnake is consistent with Gd directly activating the Snk zymogen in a protease cascade. When injected into embryos from gd⁹/gd⁹ females, the catalytic chain of Gd does not rescue or exhibit a dominant lateralizing or ventralizing effect as do analogous constructs for Snk (snkN_e) and Ea (ean; CHASAN et al. 1992 ; SMITH and DELOTTO 1994 ). Therefore, it would appear that both the propolypeptide as well as the catalytic chain are absolutely required for Gd to efficiently activate the downstream components of the pathway.

A role of the propolypeptide in the spatial regulation of Gd activity is supported by the fact that microinjection of gdn_e into the posterior pole of an embryo from a gd¹⁰/gd¹⁰ female results in rescue that is uniform along the anterior-posterior axis of the embryo. This contrasts with injection of wild-type gd RNA, which produces phenotypic rescue that is greater near the site of injection and less extreme away from the site of injection. The result of injection of gdn_e into embryos from gd¹⁰/gd¹⁰ females suggests that the Gd catalytic chain is capable of freely diffusing within the PVS, while full-length Gd is not. This would argue for a localizing or binding function for the Gd propolypeptide.

A model for how Gd functions that fits the ts period data and the genetic and molecular data is as follows. In this model the two separable functions, that of the propolypeptide chain and that of the catalytic chain, are required at two distinct times. Gd protein would be expressed from maternal mRNA late during oogenesis, secreted, and localized to the plasma membrane surface within the PVS via binding sites within the proenzyme polypeptide. Gd is uniformly distributed relative to the dorsal-ventral axis. It remains bound to the plasma membrane and remains inactive until the syncytial blastoderm stage. At this time, Gd becomes autocatalytically active only on the ventral side and initiates a proteolytic cascade resulting in the ventrally restricted production of a processed form of Spaetzle. This "localization during oogenesis/activation during embryogenesis" model explains the ability of Gd to restore ventrolateral pattern elements as late as stage 2 of embryogenesis by microinjection of RNA and its failure to normalize the pattern in embryos from gd⁹/gd⁹ females. This would be due to the failure to establish the normal distribution of bound Gd within the PVS during embryogenesis because of the nonuniform secretion of Gd into the PVS. Aspects of this model might be tested by generating a heat-shock-inducible form of Gd and a P-element-mediated transformed line.

Gd occupies a pivotal role in the process of specifying the dorsal-ventral axis of the embryo. Our data are consistent with Gd directly activating Snk and therefore suggest that Gd is the earliest acting of the germ-line-derived proteases in the PVS. Gd appears to be a molecule intimately involved in the interpretation of the ventral prepattern of the egg (R. DELOTTO, unpublished results). Biochemical data from coexpression experiments indicate that Gd activates Snk and triggers a proteolytic cascade comprising Snk, Ea, and Spz and in the process Gd undergoes rather complex proteolytic processing (DISSING et al. 2001 ). Some of the sizes of these fragments are consistent with predicted sizes for an active form of Gd from the C2/B model (DELOTTO 2001 ). Taken together, the available data suggest that Gd is responding to spatial prepattern information and to temporal cues for activation and receiving feedback from the cascade that modulates its activity. As Gd is the focus for these multiple inputs, it may constitute a nexus for regulating the shape of the Dorsal gradient. From this perspective, it may not be surprising that complex genetic interactions might be attributed to the gd locus.

	ACKNOWLEDGMENTS

We thank Boaz Rosenblat for assistance with the DNA sequencing, Marianne Dissing for helpful discussions, Ken Konrad and Anthony Mahowald for providing stocks and information in advance of publication, Trudi Schüpbach for providing stocks, and Gerd Jurgens for helpful discussions. Funding for this work was provided by the U.S. National Science Foundation, the Danish Cancer Fund, the Novo Nordisk Fund, the Vera and Karl Johan Michaelsens Legacy, and the Danish Natural Sciences Research Councils.

Manuscript received February 28, 2001; Accepted for publication July 11, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSON, K., and C. NÜSSLEIN-VOLHARD, 1984 Genetic analysis of dorsal-ventral embryonic pattern in Drosophila, pp. 269–289 in Pattern Formation: A Primer in Developmental Biology, edited by G. MALAKINSKI and P. BRYANT. Macmillan, New York.

ANDERSON, K., and C. NÜSSLEIN-VOLHARD, 1986 Dorsal group genes of Drosophila, pp. 177–194 in Gametogenesis and the Early Embryo, edited by J. GALL. A. R. Liss, New York.

ARLAUD, G. J., J. E. VOLANAKIS, N. M. THIELENS, S. V. NARAYANA, and V. ROSSI et al., 1998  The atypical serine proteases of the complement system. Adv. Immunol. 69:249-307[Medline].

BIER, E., L. JAN, and Y.-N. JAN, 1990  rhomboid, a gene required for dorsoventral axis establishment and peripheral nervous system development in Drosophila melanogaster. Genes Dev. 4:190-203[Abstract/Free Full Text].

CHASEN, R. and K. V. ANDERSON, 1989  The role of easter, an apparent serine protease, in organizing the dorsal-ventral axis of the Drosophila embryo. Cell 56:391-400[Medline].

CHASAN, R., Y. JIN, and K. V. ANDERSON, 1992  Activation of the easter zymogen is regulated by five other genes to define dorsal-ventral polarity in the Drosophila embryo. Development 115:607-616[Abstract].

CLIFFORD, R. and T. SCHÜPBACH, 1992  The torpedo (DER) receptor tyrosine kinase is required at multiple times during Drosophila embryogenesis. Development 115:853-872[Abstract].

DELOTTO, R., 2001  Gastrulation defective, a complement factor C2/B-like serine proteose, interprets a ventral prepattern in Drosophila. EMBO Rep. 2:721-726[Medline].

DELOTTO, R. and P. SPIERER, 1986  A gene required for the specification of dorsal-ventral pattern in Drosophila appears to encode a serine protease. Nature 323:688-692[Medline].

DELOTTO, Y. and R. DELOTTO, 1998  Proteolytic processing of the Drosophila Spaetzle protein by Easter generates a dimeric NGF-like molecule with ventralizing activity. Mech. Dev. 72:141-148[Medline].

DELOTTO, Y., C. SMITH, and R. DELOTTO, 2001  Multiple isoforms of the NGF-like growth factor ligand are encoded as alternatively spliced mRNAs in precellular blastoderm embryos. Mol. Gen. Genet. 264:643-652[Medline].

DISSING, M., H. GIORDANO, and R. DELOTTO, 2001  Autoproteolysis and feedback in a protease cascade directing Drosophila dorsal-ventral cell fate. EMBO J. 20:2387-2393[Medline].

HAN, J. H., S. H. LEE, Y. Q. TAN, E. K. LEMOSY, and C. HASHIMOTO, 2000  Gastrulation defective is a serine protease involved in activating the receptor toll to polarize the Drosophila embryo. Proc. Natl. Acad. Sci. USA 97:9093-9097[Abstract/Free Full Text].

HIGUCHI, R., 1990 Recombinant PCR, pp. 177–183 in PCR Protocols: A Guide to Methods and Applications, edited by M. INNIS, D. GELFAND, J. SNINSKY and T. WHITE. Academic Press, San Diego.

KONRAD, K., T. GORALSKI, and A. MAHOWALD, 1988a  Developmental analysis of fs(1)gastrulation defective, a dorsal group gene of Drosophila melanogaster. Wilhelm Roux's Arch. Dev. Biol. 197:75-91.

KONRAD, K., T. GORALSKI, and A. MAHOWALD, 1988b  Developmental genetics of the gastrulation defective locus in Drosophila melanogaster. Dev. Biol. 127:133-142[Medline].

KONRAD, K. D., T. J. GORALSKI, A. P. MAHOWALD, and J. L. MARSH, 1998  The gastrulation defective gene of Drosophila melanogaster is a member of the serine protease superfamily. Proc. Natl. Acad. Sci. USA 95:6819-6824[Abstract/Free Full Text].

LEMOSY, E. K., Y. Q. TAN, and C. HASHIMOTO, 2001  Activation of a protease cascade involved in patterning the Drosophila embryo. Proc. Natl. Acad. Sci. USA 98:5055-5060[Abstract/Free Full Text].

MANN, K., R. JENNY, and S. KRISNASWAMY, 1988  Cofactor proteins in the assembly and expression of blood clotting enzyme complexes. Annu. Rev. Biochem. 57:915-956[Medline].

MOHLER, J.-D., 1977  Developmental genetics of the Drosophila egg. I. Identification of 59(sex-linked cistrons with maternal effects on embryonic development. Genetics 85):259-272.

MORISATO, D. and K. ANDERSON, 1994  The spaetzle gene encodes a component of the extracellular signalling pathway establishing the dorsal-ventral pattern of the Drosophila embryo. Cell 76:677-688[Medline].

MORISATO, D. and K. V. ANDERSON, 1995  Signaling pathways that establish the dorsal-ventral pattern of the Drosophila embryo. Annu. Rev. Genet. 29:371-399[Medline].

NEUMAN-SILBERBERG, F. S. and T. SCHÜPBACH, 1993  The Drosophila dorsoventral patterning gene gurken produces a dorsally localized RNA and encodes a TGF alpha-like protein. Cell 75:165-174[Medline].

OGLESBY, T. J., M. A. ACCAVITTI, and J. E. VOLANAKIS, 1988  Evidence for a C4b binding site on the C2b domain of C2. J. Immunol. 141:926-931[Abstract].

ROTH, S., 1994  Proteolytic generation of a morphogen. Curr. Biol. 4:755-757[Medline].

ROTH, S. and T. SCHÜPBACH, 1994  The relationship between ovarian and embryonic dorsoventral patterning in Drosophila. Development 120:2245-2257[Abstract].

ROTH, S., D. STEIN, and C. NÜSSLEIN-VOLHARD, 1989  A gradient of nuclear localization of the dorsal protein determines dorsoventral pattern in the Drosophila embryo. Cell 59:1189-1202[Medline].

ROTH, S., F. S. NEUMAN-SILBERBERG, G. BARCELO, and T. SCHÜPBACH, 1995  cornichon and the EGF receptor signaling process are necessary for both anterior-posterior and dorsal-ventral pattern formation in Drosophila. Cell 81:967-978[Medline].

RUSHLOW, C., K. HAN, J. MANLEY, and M. LEVINE, 1989  The graded distribution of the dorsal morphogen is initiated by selective nuclear transport in Drosophila. Cell 59:1165-1177[Medline].

SCHÜPBACH, T., 1987  Germ line and soma cooperate during oogenesis to establish the dorsoventral pattern of egg shell and embryo in Drosophila melanogaster. Cell 49:699-707[Medline].

SMITH, C. and R. DELOTTO, 1994  Ventralizing signal determined by protease activation in Drosophila embryogenesis. Nature 368:548-551[Medline].

SMITH, C., H. GIORDANO, M. SCHWARTZ, and R. DELOTTO, 1995  Spatial regulation of Drosophila Snake protease activity in the generation of dorsal-ventral polarity. Development 121:4127-4135[Abstract].

STEIN, D. and C. NÜSSLEIN-VOLHARD, 1992  Multiple extracellular activities in Drosophila egg perivitelline fluid are required for establishment of embryonic dorsal-ventral polarity. Cell 68:429-440[Medline].

STEIN, D., S. ROTH, E. VOGELSANG, and C. NÜSSLEIN-VOLHARD, 1991  The polarity of the dorsoventral axis in the Drosophila embryo is defined by an extracellular signal. Cell 65:725-735[Medline].

STEWARD, R., S. ZUSMAN, L. HUANG, and P. SCHEDL, 1988  The dorsal protein is distributed in a gradient in early Drosophila embryos. Cell 55:487-495[Medline].

STROUD, R., A. KOSSIAKOFF, and J. CHAMBERS, 1977  Mechanisms of zymogen activation. Ann. Rev. Biophys. Bioeng. 6:177-193[Medline].

WIESCHAUS, E., and C. NÜSSLEIN-VOLHARD, 1986 Looking at embryos, pp. 199–228 in Drosophila: A Practical Approach, edited by D. B. ROBERTS. IRL Press, Oxford/Washington, DC.

This article has been cited by other articles:

				A. Goldraij, L. J. Beamer, and J. C. Polacco Interallelic Complementation at the Ubiquitous Urease Coding Locus of Soybean Plant Physiology, August 1, 2003; 132(4): 1801 - 1810. [Abstract] [Full Text] [PDF]

				T. Rose, E. K. LeMosy, A. M. Cantwell, D. Banerjee-Roy, J. B. Skeath, and E. Di Cera Three-dimensional Models of Proteases Involved in Patterning of the Drosophila Embryo. CRUCIAL ROLE OF PREDICTED CATION BINDING SITES J. Biol. Chem., March 21, 2003; 278(13): 11320 - 11330. [Abstract] [Full Text] [PDF]