Biochemical Defects of Mutant nudel Alleles Causing Early Developmental Arrest or Dorsalization of the Drosophila Embryo

Corresponding author: Ellen K. LeMosy, Department of Cell Biology, Yale University School of Medicine, 333 Cedar St., Rm. C-214, P.O. Box 208002, New Haven, CT 06520., ellen.lemosy{at}att.net (E-mail)

Communicating editor: K. ANDERSON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The nudel gene of Drosophila is maternally required both for structural integrity of the egg and for dorsoventral patterning of the embryo. It encodes a structurally modular protein that is secreted by ovarian follicle cells. Genetic and molecular studies have suggested that the Nudel protein is also functionally modular, with a serine protease domain that is specifically required for ventral development. Here we describe biochemical and immunolocalization studies that provide insight into the molecular basis for the distinct phenotypes produced by nudel mutations and for the interactions between these alleles. Mutations causing loss of embryonic dorsoventral polarity result in a failure to activate the protease domain of Nudel. Our analyses support previous findings that catalytic activity of the protease domain is required for dorsoventral patterning and that the Nudel protease is auto-activated and reveal an important role for a region adjacent to the protease domain in Nudel protease function. Mutations causing egg fragility and early embryonic arrest result in a significant decrease in extracellular Nudel protein, due to defects in post-translational processing, stability, or secretion. On the basis of these and other studies of serine proteases, we suggest potential mechanisms for the complementary and antagonistic interactions between the nudel alleles.

THE eleven dorsal-group genes are maternally expressed genes that act in the establishment of the dorsoventral axis of the Drosophila embryo (MORISATO and ANDERSON 1995 ). The products of these genes cooperate in a transmembrane pathway to generate a signal that is required to specify ventral and lateral cell fates; in the absence of this signal, all cells of the embryo adopt the dorsal cell fate. In this pathway, an extracellular signal is produced within the perivitelline space, a compartment that lies between the embryo plasma membrane and the innermost layer of the eggshell, the vitelline membrane. This signal acts as a ligand for the receptor Toll, ultimately leading to the nuclear translocation of the transcription factor Dorsal in a ventral-to-dorsal morphogenetic gradient. The Toll ligand is generated by a proteolytic processing reaction requiring the activities of three germline-derived serine proteases that appear to function in a sequential cascade as seen for mammalian blood clotting (for recent reviews see ANDERSON 1998 ; ROTH 1998 ; LEMOSY et al. 1999 ). The activities of these proteases may be restricted to the ventral side of the embryo by a localized factor provided by somatic follicle cells that surround the developing oocyte and synthesize the eggshell. Recent studies suggest that the activities of two somatically expressed dorsal-group genes, pipe and windbeutel, are important for generating this localized factor (KONSOLAKI and SCHUPBACH 1998 ; NILSON and SCHUPBACH 1998 ; SEN et al. 1998 ).

We are interested in understanding the function of a third somatically expressed dorsal-group gene, nudel, that is also required for activity of the serine protease cascade. In contrast to the remaining dorsal-group genes, there is an additional requirement for maternal nudel function in early embryonic development (HONG and HASHIMOTO 1996 ). Females lacking detectable nudel mRNA expression produce embryos that have extremely fragile eggshells and arrest during the early cleavage divisions, rarely progressing to the syncytial blastoderm stage when dorsoventral polarity is established. Similar phenotypes are seen for mutations affecting structural proteins of the eggshell, suggesting that the eggshell defect may be the cause of the early developmental arrest in nudel mutants (SAVANT and WARING 1989 ). Mutations in the nudel gene do not form a continuous allelic series, but rather fall into discrete phenotypic classes showing either early developmental arrest or dorsalization. This observation suggests that the nudel gene has distinct functions in structural integrity and in embryonic patterning that may map to distinct domains of the Nudel protein.

Nudel is a large (2616 amino acids) protein that is secreted into the future perivitelline space in mid-oogenesis and associates with the oocyte surface (LEMOSY et al. 1998 ). Consistent with the idea that Nudel is a functionally modular protein, Nudel has a modular structure reminiscent of extracellular matrix proteins (HONG and HASHIMOTO 1995 ). A central serine protease catalytic domain has been shown to be essential for Nudel's function in dorsoventral polarity and might directly activate the first germline-derived protease in the cascade, or, alternatively, might activate a cofactor essential for the proteolytic reactions (HONG and HASHIMOTO 1996 ; LEMOSY et al. 1998 ). Other structural motifsthat might interact with components of the protease cascade include 11 copies of the LDL-receptor ligand-binding repeat (YAMAMOTO et al. 1984 ), a motif implicated in binding to serine proteases and proteaseinhibitor complexes (WILLNOW et al. 1994 ), and an incomplete protease-like region near its C terminus.

Nudel undergoes extensive proteolytic processing that might be required for its structural or patterning functions and that might also separate distinct functional regions of the protein (LEMOSY et al. 1998 ). Additionally, Nudel is predicted from its sequence to be extensively glycosylated, with 23 potential sites for N-linked glycosylation, two serine/threonine-rich regions that might be O-glycosylated, and three potential sites for glycosaminoglycan addition. Supporting this prediction, the 350-kD full-length Nudel protein containing 30 kD of N-linked sugars is processed to generate N-terminal (210 kD) and C-terminal (250 kD) polypeptides that together have ~110 kD of additional mass, likely representing O-linked sugars such as glycosaminoglycans (LEMOSY et al. 1998 ; see Fig 1A). Glycosaminoglycans have been shown to be important for the formation of complexes between serine proteases and other proteins, such as inhibitors (BOURIN and LINDAHL 1993 ), and both N- and O-linked carbohydrates have been shown to be important for interactions of glycoproteins with each other and with the cell surface (WEST 1986 ). Together these structural motifs and post-translational modifications suggest that Nudel might function in the assembly or regulation of multiprotein zymogen activation complexes in which the dorsoventral proteases are activated.

View larger version (28K): In this window In a new window Download PPT slide   Figure 1. Nudel structure and processing are disrupted in class I mutant alleles. (A) Top diagram illustrates major structural features of the wild-type Nudel protein, including the serine protease catalytic domain (black), LDL-receptor ligand-binding motifs (gray), serine/threonine-rich regions that may be O-glycosylated (hatched), three potential glycosaminoglycan addition sites (arrowheads), and the putative RGD integrin recognition sequence (asterisk). Also marked are the regions of the Nudel protein recognized by N-terminal (N), protease domain (PD), and C-terminal (C) antibodies. The remainder of A shows the derivation of the wild-type N- and C-terminal Nudel polypeptides recognized by these antibodies. The difference between predicted and apparent molecular masses of the 210-kD and 250-kD polypeptides, apparently derived from a 350-kD precursor by proteolytic cleavage, may be due to glycosylation subsequent to this cleavage event. Nudel undergoes sequential cleavages during development; the last of these are dependent upon the presence of a functional Nudel protease (asterisks; LEMOSY et al. 1998 ). (B) One ovary pair from ndl+/Df (WT) or ndlClass I/Df (allele name) females was analyzed by Western blot, using an antibody to an N-terminal region of Nudel. The wild-type Nudel polypeptides recognized by this antibody include a 350-kD full-length translation product and N-terminal fragments at 210 and 170 kD. Specific but aberrant Nudel polypeptides present in many of the mutant ovaries are indicated with arrowheads. For three of the alleles, ndl16, ndl133, and ndl17, a sixfold longer exposure of the ECL reaction to film revealed the presence of Nudel polypeptides (lanes at right of figure). A pair of nonspecific bands recognized by this antisera is indicated with asterisks in the ndl12 lane; these bands are more prominent in the longer exposure at the right of the figure. Similar Western blots were performed using antibodies recognizing the protease domain and the extreme C terminus of Nudel; polypeptides recognized by these antibodies are indicated in Table 1.

Supporting the idea that the Nudel protein is functionally modular, complementation is observed between mutant nudel alleles (HONG and HASHIMOTO 1996 ). Many alleles demonstrating the early arrest phenotype (class I alleles) are able to complement a hypomorphic dorsalizing (class II) allele, ndl⁰⁴⁶, that has a mutant protease domain. A second hypomorphic class II allele, ndl⁹, that lacks a protease domain mutation, also complements ndl⁰⁴⁶, but is itself complemented by only one class I mutation, ndl¹⁵. One possibility suggested by these observations is that the mutations in ndl⁹ and most of the class I alleles might fall in a common region of Nudel required for both structural and embryonic patterning functions, while the ndl¹⁵ mutation might fall in a distinct region required only for structural integrity (HONG and HASHIMOTO 1996 ). Antagonistic interactions are also observed in which two strongly dorsalizing class II alleles block the activity of the hypomorphic class II alleles. These intriguing interactions suggest that further molecular analysis of the nudel mutant alleles may aid in understanding the relationship between Nudel protein structure and its functions in structural integrity and in dorsoventral patterning.

Toward this goal, we have now performed biochemical and immunolocalization studies of the mutant proteins expressed by the nudel alleles. In contrast to a prediction of the complementation analysis that a common region of Nudel might be affected in most class I mutations, we find that the class I mutations cause the production of defective Nudel proteins that have varied difficulties involving post-translational processing, secretion, stability, and extracellular localization. Protein and sequence analyses of the class II (dorsalizing) mutations reveal an important role for a region adjacent to the protease domain in Nudel protease function and support previous findings that the catalytic activity of the protease domain is required for dorsoventral patterning and that the Nudel protease is auto-activated. On the basis of these and other studies of serine proteases, we suggest potential mechanisms that may underlie the positive and negative interactions between the nudel alleles.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly stocks and genetic analysis:
The wild-type stock was Oregon R. The ndl⁰⁴⁶, ndl⁰⁹³, ndl¹¹¹, ndl¹³³, ndl¹⁶⁹, ndl²⁶⁰, and ndl^RM5 alleles were isolated in screens for female sterility (ANDERSON and NUSSLEIN-VOLHARD 1984 ); the ndl^9-18 series was isolated in screens for mutations that failed to complement the ndl⁰⁴⁶ allele (D. MORISATO and K. V. ANDERSON, personal communication); and the ndl^LP-1 and ndl^LP-2 alleles were isolated in mosaic screens for follicle cell mutations causing embryonic patterning defects (PAI et al. 1998 , PAI et al. 1999 ). A deficiency for the ndl locus, Df(3L)CH12, was used to generate flies hemizygous for the mutant alleles (HONG and HASHIMOTO 1995 ). Genetic characterization of 15 of these alleles has been reported previously (HONG and HASHIMOTO 1996 ). The phenotype of the ndl⁰⁹³, ndl^RM5, ndl^LP-1, and ndl^LP-2 alleles was assayed by examining the embryos laid by hemizygous females under oil and by cuticle preparations (WIESCHAUS and NUSSLEIN-VOLHARD 1986 ). Interactions of these alleles with class II alleles were examined as described previously (HONG and HASHIMOTO 1996 ).

Western blot and immunoprecipitation analysis:
Antibodies to three distinct regions of the Nudel protein were used in Western blot analysis of ovary or embryo proteins as described previously (LEMOSY et al. 1998 ). For immunoprecipitation, ovaries were homogenized in 20 mM Tris-Cl, pH 7.5, 0.15 M NaCl (TBS) and sequentially centrifuged at 40 x g and 16,000 x g in a microfuge; Nudel is selectively enriched in low-speed pellet fractions, especially the 40 x g pellet that contains eggshell and cytoskeletal proteins (FARGNOLI and WARING 1982 ; E. K. LEMOSY, unpublished results). The washed, resuspended 40 x g pellet and 16,000 x g supernatant were boiled in the presence of 1% SDS to solubilize Nudel. The soluble extracts were diluted with 4 volumes of TBS + 1% NP-40, and then Nudel polypeptides were immunoprecipitated by mixing overnight at 4° with N- or C-terminal Nudel antibody and protein A-Sepharose (Sigma, St. Louis). After extensive washing, bound proteins were eluted in reducing sample buffer and were analyzed by Western blotting.

Immunolocalization:
The N-terminal Nudel antibody and rhodamine-phalloidin were used to stain fixed, whole-mount ovary preparations as described (VERHEYEN and COOLEY 1994 ), and the samples were imaged using a LSM-510 confocal microscope (Zeiss, Thornwood, NY). The efficiency of Nudel secretion is best judged at stage 8–9, when most wild-type Nudel is detected extracellularly, and in stage 12 or later, when no wild-type Nudel is present within the follicle cells; at stage 10 there is a burst of wild-type Nudel synthesis leading to high intracellular levels of Nudel that cannot be distinguished from retained Nudel protein. The ndl¹⁵ protein was previously erroneously reported not to be secreted (LEMOSY et al. 1998 ).

Sequencing and mutagenesis:
Genomic DNA, prepared from hemizygous flies as described previously (LEVIS et al. 1982 ), was used as template in PCR reactions to amplify the entire nudel sequence (ndl⁹) or two large overlapping fragments (ndl¹⁴) using the Expand Long Template PCR system (Boehringer Mannheim, Indianapolis) or a central region containing the protease domain sequence using Taq polymerase (Perkin-Elmer, Norwalk, CT). Multiple PCR reactions were pooled and sequenced for each allele. Comparison of the sequences of the ndl⁹ and ndl¹⁴ alleles, isolated in a single screen, allowed unambiguous determination of the unique mutations in these alleles. The ndl⁰⁹³, ndl^RM5, and ndl^LP-1 alleles each contained a single point mutation within the protease domain, as well as the previously described polymorphisms relative to the nudel cDNA sequence (HONG and HASHIMOTO 1996 ).

The potential integrin recognition motif, RGD (RUOSLAHTI 1996 ), was mutated to RAA using a PCR-based method as described previously (LEMOSY et al. 1998 ). A SapI fragment containing this mutation was substituted for the wild-type sequence in the genomic nudel rescuing construct, and transformant lines were generated and analyzed in a ndl¹⁴/Df(3L)CH12 background as described previously (SPRADLING 1986 ; LEMOSY et al. 1998 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Class I alleles characterized by reduced or aberrant Nudel protein:
To explore the nature of the defects in the class I nudel alleles that result in egg fragility and early developmental arrest, we first examined Nudel protein expression in mutant ovaries by Western blotting (Fig 1B; summarized in Table 1). For comparison, the processing steps that generate a set of wild-type Nudel polypeptides are illustrated in Fig 1A. The major N- and C-terminal polypeptides appear to be generated rapidly from a 350-kD precursor in mid-oogenesis, while a less-abundant 170-kD N-terminal polypeptide is generated by cleavage within the extracellular space, perhaps similar to the processing of certain vitelline membrane and chorion proteins (NOGUERON and WARING 1995 ; PASCUCCI et al. 1996 ). This wild-type pattern is disrupted in the class I nudel alleles.

Four class I alleles (ndl¹⁰, ndl¹³, ndl¹⁴, and ndl¹³³) were described previously as potential null alleles, because they failed to complement the class II ndl⁰⁴⁶ allele, and three were shown to express little or no nudel mRNA (HONG and HASHIMOTO 1995 , HONG and HASHIMOTO 1996 ). By biochemical analysis, three of these alleles (ndl¹³, ndl¹⁴, and ndl¹³³) were found to contain Nudel proteins detectable with N-terminal but not C-terminal Nudel antibody, suggesting that they encode truncated Nudel proteins (Fig 1B and Table 1). One of these, ndl¹⁴, confirmed to encode a truncation by DNA sequence analysis, contains a point mutation (G3400A) that gives rise to a premature stop codon (W1044Stop) after the potential glycosaminoglycan addition sites and the first block of LDL-receptor ligand-binding repeats and thus does not express the Nudel protease domain. As the ndl¹⁰ allele does not produce any protein detectable with available antibodies, it is the best candidate for a null allele.

Three additional alleles (ndl¹², ndl¹⁷, and ndl¹⁸) appeared to be truncation mutants, based upon the absence of polypeptides recognized by a strong C-terminal Nudel antibody. This interpretation is consistent with the previous finding that ndl¹⁸ contains a 569-bp deletion that causes loss of the C-terminal 402 amino acids (including four LDL-receptor ligand-binding repeats) with replacement by 75 novel amino acids (HONG and HASHIMOTO 1996 ). There appeared to be variability in the proteolytic processing of the truncated proteins, suggested by the observation that the ndl¹⁸ truncation resulted in a single 320-kD polypeptide that did not appear to undergo proteolytic processing to smaller forms, while another apparent truncation, ndl¹², showed distinct N-terminal and protease domain-containing fragments. In no case were additional smaller N-terminal forms detected that might correspond to the extracellular processing of the wild-type 210-kD form to generate a 170-kD polypeptide.

Four class I alleles produced aberrant proteins that were not C-terminal truncations. The ndl¹⁶ allele appeared to express the normal C-terminal polypeptide (250 kD) but had a quite small N-terminal polypeptide (116 kD) that might lack carbohydrate additions or primary sequence, perhaps due to an in-frame deletion. The ndl¹⁵, ndl¹⁶⁹, and ndl^LP-2 alleles showed accumulation of the apparent full-length primary translation product (350 kD) with very reduced amounts of the processed C-terminal polypeptide (250 kD). Many weak bands were detectable with all Nudel antibodies in these mutant ovaries, suggesting that these mutant Nudel proteins are unstable and are degraded.

Aberrant secretion and localization of some class I proteins:
Localization of the class I mutant proteins in ovaries provided further information about the defects in these alleles (Fig 2 and Table 1). The wild-type protein is secreted from follicle cells and associates with the oocyte surface (LEMOSY et al. 1998 ; Fig 2A). The putative null allele, ndl¹⁰, showed no specific staining of the follicle cells or the oocyte surface (Fig 2B). Two class I alleles (ndl¹², ndl¹⁶⁹) expressed proteins that accumulated within the follicle cells and were not secreted. This was particularly striking in ndl¹⁶⁹, which expressed a high level of Nudel protein that persisted long after Nudel synthesis normally ends in stage 10B of oogenesis (Fig 2C). Several others were secreted, including the ndl¹⁵ and ndl^LP-2 proteins that, like ndl¹⁶⁹, largely failed to undergo processing from the 350-kD full-length form (Fig 2D).

View larger version (40K): In this window In a new window Download PPT slide   Figure 2. Defects in localization of some class I polypeptides. Whole-mount ovary preparations were stained with N-terminal Nudel antibody (green) and rhodamine phalloidin to label the cortical actin cytoskeleton of the oocyte and follicle cells (red). Shown are portions of stage 10 egg chambers containing the oocyte and an overlying layer of somatic follicle cells (fc); occasionally germline nurse cells (nc) are shown. In each panel, the double arrows are within the oocyte and point to rhodamine phalloidin staining at the border of the oocyte. In B and E, detector gain was increased to allow visualization of specific staining, if present, over the background level seen in germline nurse cells. (A) ndl+/Df egg chamber shows staining within follicle cells and at the oocyte surface. (B) ndl10/Df egg chamber shows no specific follicle staining or staining in the extracellular space between the follicle cells and oocyte. (C) ndl169/Df egg chamber shows strong follicle cell expression with no secretion of the protein. This mutant protein persists in follicle cells of late-stage egg chambers (arrowheads). (D) ndl15/Df egg chamber shows strong follicle cell expression and extracellular protein localization at the oocyte surface. In earlier stages, it is clear that the majority of ndl15 protein is secreted (see MATERIALS AND METHODS). (E) ndl16/Df egg chamber shows diffuse staining in the extracellular space without clear localization to the oocyte surface. (F) Egg chamber expressing a nudel transgene in which the RGD sequence has been mutated to RAA. Proper localization of this protein at the oocyte surface suggests that Nudel is unlikely to bind to an integrin via the putative RGD integrin recognition sequence. Bar, 40 µm.

In most of the alleles demonstrating secretion of Nudel, this secreted protein was localized at the oocyte surface (Fig 2 and Table 1). However, we observed that the ndl¹⁶ allele showed secretion of Nudel protein without clear localization to the oocyte surface, instead showing diffuse staining within the extracellular space (Fig 2E); similar staining was seen for the ndl¹³³ allele (not shown). These alleles express N-terminal Nudel polypeptides that migrate rapidly in SDS-PAGE gels (Fig 1) and might lack modifications or primary sequence required for association with the oocyte surface. Protein sequence elements required for this surface association are not known, but apparently do not include the putative RGD integrin recognition motif present in the N-terminal portion of the Nudel protein (HONG and HASHIMOTO 1995 ). Expression of a transgenic nudel construct in which the RGD had been mutated to RAA, a change predicted to eliminate functional interactions with integrins (PRIETO et al. 1993 ), resulted in a protein that was functional in both egg integrity and dorsoventral polarity (not shown) and was normally localized to the oocyte surface (Fig 2F).

Stability of class I proteins within the extracellular space:
The ndl¹⁵ and ndl^LP-2 alleles demonstrate an interesting difference in phenotype, although they produce similarly secreted but incompletely processed proteins. The ndl^LP-2 allele is unusual among the class I alleles in that a significant proportion (up to 50%) of embryos derived from hemizygous females progress past early development and either are dorsalized or hatch. This phenotype indicates that ndl^LP-2 is hypomorphic for both structural and patterning functions of Nudel. In contrast, the ndl¹⁵ allele is completely defective in structural function, although, like ndl^LP-2, ndl¹⁵ is able to strongly complement class II hypomorphic alleles (Table 1; HONG and HASHIMOTO 1996 ). This difference in phenotype between ndl^LP-2 and ndl¹⁵ may be due to a difference in stability within the extracellular space of the proteins encoded by these alleles. While the ndl^LP-2 protein persisted and could be detected in the embryo, although in reduced amounts relative to wild type (not shown), the ndl¹⁵ protein was degraded over the course of oogenesis (Fig 3A) and could not be detected in early embryos (not shown). These observations suggest that stability of Nudel over the course of oogenesis may be important to the maintenance of structural integrity.

View larger version (22K): In this window In a new window Download PPT slide   Figure 3. Aberrant Nudel polypeptides might be stabilized by assembly into a disulfide-bonded matrix. (A) Western blot of class I ndl15 protein, here probed with N-terminal antibody, present in extracts of ndl15/Df ovaries. While the 350-kD full-length protein is prominent in total preparations of ndl15/Df ovaries containing all developmental stages (lane 1), this protein is absent in extracts of late-stage (stages 11–14) egg chambers and is replaced by progressively smaller and aberrantly sized polypeptides (lanes 2 and 3), consistent with degradation of the ndl15 protein. In mutant laid eggs (not shown), none of the Nudel antibodies recognize specific polypeptides, suggesting that degradation is complete before the onset of embryogenesis. (B) Immunoprecipitation of Nudel from 40 x g pellet and 16,000 x g supernatant fractions of wild-type (WT) ovaries (preparation schematized in C; also see MATERIALS AND METHODS). While N- and C-terminal polypeptides are not coprecipitated in denaturing immunoprecipitations of the 16,000 x g supernatant (lanes 1 and 2), they show strong coprecipitation from extracts of the 40 x g pellet (lanes 3–6). Intermediate fractions, such as the 16,000 x g pellet, show weaker coprecipitation of the Nudel polypeptides (not shown). (C) Schematic showing derivation of the fractions used for immunoprecipitation. Below, the 16,000 x g supernatant appears to contain uncrosslinked N- and C-terminal polypeptides, while in the 40 x g pellet, these polypeptides, including the 170-kD cleavage product generated extracellularly, appear to be linked by disulfide bonds and thus coprecipitate under denaturing conditions.

It is, perhaps, surprising that the unstable ndl¹⁵ protein is able to strongly complement class II alleles to give 50–90% hatching larvae, including the ndl⁹ allele that is complemented by only two class I alleles, ndl¹⁵ and ndl^LP-2. One possible explanation for this complementation is that the unstable ndl¹⁵ protein is stabilized by interaction with a normally processed Nudel protein. While direct examination of this possibility is difficult, observations from biochemical studies of the wild-type Nudel protein suggest a mechanism for such stabilization. During oogenesis, most Nudel is associated with a very low-speed (40 x g) pellet fraction that is highly enriched in eggshell proteins, but some is present in supernatant fractions (FARGNOLI and WARING 1982 ; see MATERIALS AND METHODS). Under denaturing conditions, immunoprecipitation of Nudel from extracts of the 40 x g pellet gave a strikingly different result when compared to immunoprecipitation of Nudel from supernatant fractions (Fig 3B). Antibodies to the N-terminal fragment of Nudel were able to coprecipitate the C-terminal fragment from extracts of the 40 x g pellet, and vice versa, suggesting that these fragments are at least partially associated by disulfide bonds (schematized in Fig 3C, though these bonds need not be direct between the two fragments). Within supernatant fractions, however, we detected little or no coprecipitation of the N- and C-terminal fragments, indicating that these fragments are not obligatorily linked. The 170-kD N-terminal Nudel form, generated within the extracellular space in the last third of oogenesis, was exclusively present within the 40 x g pellet fraction and could be precipitated with the C-terminal antibody. Together, these observations suggest that secreted Nudel is progressively stabilized within an extracellular matrix layer around the oocyte by disulfide bonds and perhaps by N-terminal cleavage. In the case of class I proteins, it is possible that coexpression with a Nudel protein competent for matrix assembly allows stabilization of the class I protein (SOTTILE and WILEY 1994 ; LAMANDE et al. 1998 ).

Class II alleles characterized by defects in Nudel protease activation:
In contrast to the class I alleles, the class II alleles demonstrated normal synthesis and processing of Nudel during oogenesis (not shown). It was shown previously that three class II alleles, ndl⁰⁴⁶, ndl¹¹¹, and ndl²⁶⁰, have point mutations within the protease catalytic domain that might be responsible for the loss of activity in dorsoventral polarity establishment (HONG and HASHIMOTO 1996 ; Table 2). Supporting a role for a catalytically active Nudel protease in dorsoventral polarity, site-directed mutations of the catalytic serine residue or the zymogen activation site of Nudel's protease domain lead to dorsalization of the embryo (LEMOSY et al. 1998 ). In these mutants, Nudel protease appears not to undergo auto-activation in early embryogenesis, as assayed by the absence of a putative active Nudel protease form (33 kD) and C-terminal Nudel fragments (50–60 kD) that are dependent on the presence of functional Nudel protease; instead, incompletely processed Nudel polypeptides accumulate (Fig 1A).

To further explore the basis of the defects leading to dorsalization of nudel mutant embryos, we examined Nudel protease activation in the dorsalizing nudel alleles, including three previously uncharacterized alleles (ndl⁰⁹³, ndl^RM5, and ndl^LP-1). Identical to the results seen with the site-directed mutations, five alleles that were inactive in dorsoventral polarity establishment (ndl⁰⁹³, ndl¹¹¹, ndl²⁶⁰, ndl^LP-1, and ndl^RM5) showed complete absence of the 33-kD Nudel protease and 50–60-kD C-terminal fragments, with accumulation of intermediate forms (illustrated with ndl¹¹¹, Fig 4). Two alleles having partial activity in dorsoventral polarity establishment (ndl⁹ and ndl⁰⁴⁶) exhibited some processing of the C-terminal region of Nudel to 50–60-kD fragments, an event previously shown to require active Nudel protease (LEMOSY et al. 1998 ), but the 33-kD Nudel protease was undetectable, and the incompletely processed protease domain and C-terminal forms were present at high levels as seen for the completely inactive class II alleles (illustrated with ndl⁹, Fig 4). We conclude that the hypomorphic alleles, including the ndl⁹ allele that lacks a protease domain mutation, demonstrate significantly impaired Nudel protease activation.

View larger version (50K): In this window In a new window Download PPT slide   Figure 4. Class II mutations exhibit defects in Nudel protease activation. Western blots of total proteins from 0–2 hr OR (WT), ndl9 (9), or ndl111 (111) fertilized eggs were reacted with protease domain antibody (lanes 1–3) or C-terminal antibody (lanes 4–6). The mutant alleles show accumulation of a 38-kD Nudel protease form thought to have an uncleaved zymogen cleavage site (LEMOSY et al. 1998 ), while the putative active Nudel protease (33 kD) is undetectable (lanes 2 and 3). The mutant alleles similarly show accumulation of incompletely processed C-terminal polypeptides (110–130 kD; lanes 5 and 6). However, ndl9 is able to generate some completely processed C-terminal forms (50–60 kD; lane 5), an event requiring functional Nudel protease. All Nudel polypeptides are indicated; those associated with the presence of functional Nudel protease are marked with an asterisk.

Molecular characterization of the class II mutations:
To determine the molecular basis for the failure of Nudel protease activation in the ndl⁹, ndl⁰⁹³, ndl^RM5, and ndl^LP-1 alleles, we sequenced the entire ndl⁹ allele and the catalytic domain sequences of the three remaining alleles. In each case we found a unique point mutation that could explain the mutant phenotype (Table 2). In the ndl^RM5 and ndl^LP-1 alleles, the mutations are G-A transitions as expected for EMS mutations (LIM and SNYDER 1968 ); the ndl⁰⁹³ and ndl⁹ alleles contain A-T transversions that could represent spontaneous mutations picked up in these screens.

The ndl⁹ mutation causes the substitution of a serine for a cysteine that lies outside, but is closest to, the N terminus of the catalytic domain (Fig 5). In the regulated serine proteases of blood clotting, the nearest cysteine N-terminal to the catalytic domain is disulfide-bonded to a conserved cysteine within the catalytic domain, causing the N-terminal regulatory and C-terminal catalytic domains to remain together after zymogen activation (NEMERSON and FURIE 1980 ). The corresponding catalytic domain cysteine is conserved in Nudel, suggesting that it would normally be disulfide-bonded to the cysteine missing in the ndl⁹ protein. The absence of this cysteine does not appear to impair the folding of the catalytic domain within the Nudel precursor, as the ndl⁹ protein is processed and secreted normally during oogenesis (not shown).

View larger version (48K): In this window In a new window Download PPT slide   Figure 5. Alignment of Nudel protease with human plasma kallikrein and human Factor IX, showing positions of class II nudel mutations in relationship to protease structural features. The positions of the class II mutations are indicated by arrows extending from the mutated residue to its replacement, and the allele names are indicated above in parentheses. A conserved cysteine within the catalytic domain (asterisk) is predicted to form a disulfide bond with the cysteine mutated in the ndl9 allele. Highly conserved regions known as CR1–CR7 (FURIE et al. 1982 ; boxed) form the core of the protease domain including the catalytic site and the substrate-binding pocket. To show proximity to residues important for protease function, the catalytic triad residues are indicated by bold letters and overhead arrowheads, while three residues critical for determining the substrate cleavage specificity are indicated by gray shading. Dots below the alignment indicate residues in Factor IX that are known to mutate to give the hemophilia Bm phenotype (see DISCUSSION).

The previously uncharacterized class II alleles were found to have point mutations within the catalytic domain. In these, as well as the previously described ndl¹¹¹ and ndl²⁶⁰ alleles, the mutations are in regions of this domain that are predicted to be crucial for formation of the functional catalytic and substrate-binding sites (FURIE et al. 1982 ; THOMPSON 1991 ; Fig 5). In the ndl^LP-1 allele, a glycine that is absolutely conserved among trypsin-like serine proteases has been altered to a serine; this residue is buried within the protein very near the ion pair formed after zymogen cleavage between the new N-terminal -amino group and an aspartate adjacent to the active site serine. Mutations of this glycine in Factor IX are believed to interfere with formation of this ion pair, which is essential for conformational changes involved in forming both the functional catalytic site and the substrate-binding pocket (THOMPSON 1991 ). The ndl⁰⁹³ mutation alters a histidine residue to a leucine in a conserved region (CR7) important for forming the extended substrate-binding site. This histidine lies between two residues critical for formation of the substrate-binding pocket, and the corresponding residue in other proteases is believed to directly contact the P2 residue of the substrate. It should be noted that the presence of a histidine in this site in wild-type Nudel is quite unusual. In vertebrate serine proteases (GREER 1990 ), this residue is almost always a tryptophan or, less frequently, a phenylalanine, although in invertebrate proteases phenylalanine seems to predominate (DELOTTO and SPIERER 1986 ; CHASAN and ANDERSON 1989 ). The ndl^RM5 mutation alters an alanine to a threonine in the CR7 region (SPITZER et al. 1988 ).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have characterized the biochemical defects associated with nudel mutations giving rise to the class I (early arrest) and class II (dorsalized) phenotypes and have provided genetic and biochemical information about four previously uncharacterized alleles. These studies reveal unifying themes that characterize the distinct mutant classes. We discuss these themes and suggest mechanisms involving properties of serine proteases that may underlie the previously described allelic interactions.

Class I mutations result in relative loss of Nudel from the extracellular space:
Although the class I mutations cause a variety of defects in Nudel protein structure and processing, they share the common feature that there is a substantial decrease in the total amount of Nudel present within the extracellular space where it is presumably required. This quantitative defect is likely to be the root cause of the class I phenotypes of egg fragility and early embryonic arrest. Similar phenotypes of egg flaccidity and early developmental arrest can arise from such varied defects as decreased yolk uptake and loss of structural proteins of the eggshell (WARING et al. 1983 ; SAVANT and WARING 1989 ). The localization of Nudel at the oocyte surface is consistent with involvement in such functions as vitelline membrane biogenesis or oocyte adhesion to an extracellular matrix.

The C-terminal 402 amino acids of Nudel appear not to be essential for association with the oocyte surface (ndl¹⁸ protein), while protein sequences or post-translational modifications in the N-terminal half of Nudel may be required for this association (ndl¹⁶ protein). The oocyte surface localization of mutant Nudel proteins (ndl¹⁵, ndl¹⁷, ndl¹⁸, ndl^LP-2) that appear not to have undergone proteolytic processing or extensive carbohydrate addition suggests that some post-translational modifications are not essential for secretion or surface binding. However, these defective proteins may be unstable in the extracellular space (ndl¹⁵ protein), perhaps due to the absence of carbohydrates that may protect glycoproteins from proteolysis or due to failure to assemble into an extracellular matrix (WEST 1986 ; LAMANDE et al. 1998 ).

Perhaps the biggest surprise is that, despite these gross defects in Nudel expression and biogenesis, most of the class I alleles are able to complement the class II ndl⁰⁴⁶ allele (HONG and HASHIMOTO 1996 ). With the exception of ndl¹¹, where no Nudel could be detected, each of the class I alleles that is able to complement the class II ndl⁰⁴⁶ allele exhibits either the presence of a polypeptide containing the Nudel protease domain or, in the case of the very weakly expressing ndl¹⁶ and ndl¹⁷ alleles (where our weak protease domain antibody failed to detect a specific polypeptide), the presence of a Nudel polypeptide large enough to contain the protease domain (Table 1). Among these complementing alleles, the degree of complementation correlates with the secretion level of the class I protein. Together, these findings are consistent with the idea that the complementing class I alleles are able to deliver Nudel protease to the extracellular space. It appears that a very small amount of functional Nudel protease, present within a variety of mutant Nudel precursors, is sufficient to complement the defective Nudel protease made by the ndl⁰⁴⁶ allele.

Class II mutations compromise Nudel protease function:
Consistent with previous studies suggesting that catalytic activity of the Nudel protease is essential for the establishment of dorsoventral polarity and that the Nudel protease is auto-activated (LEMOSY et al. 1998 ), all of the class II mutations were found to have defects in Nudel protease activation, and all but one could be ascribed to mutations within the serine protease catalytic domain itself. The exception, ndl⁹, which affects a cysteine N-terminal to the protease domain, is particularly intriguing. This mutation is predicted to disrupt a disulfide bond linking the protease catalytic domain to a potential N-terminal regulatory domain and might also affect the structure of this prodomain. The steep temperature dependence of the function of the ndl⁹ protein (HONG and HASHIMOTO 1996 ) would be consistent with impaired thermal stability of interactions between the protease domain and an N-terminal domain that is not covalently linked in the ndl⁹ protein. This N-terminal domain could be required for Nudel protease binding to cofactors or substrates, similar to the N-terminal regulatory domain of enterokinase that is essential for interactions of this serine protease with its macromolecular substrate, trypsinogen (LU et al. 1997 ).

Potential basis for complementary allelic interactions:
The complementation of ndl⁰⁴⁶ by class I alleles appears to rely on the combination of a small amount of functional Nudel protease, derived from the class I mutant protein, with the mutant ndl⁰⁴⁶ protease, since only two of the class I alleles (ndl¹⁵ and ndl^LP-2) are able to complement the other hypomorphic class II allele, ndl⁹. The ndl⁰⁴⁶ and ndl⁹ proteases might differ in their ability to be complemented by class I alleles solely because ndl⁰⁴⁶ has greater activity in dorsoventral patterning than ndl⁹ at the temperature where this complementation was defined (22°), presumably reflecting greater catalytic activity (E. K. LEMOSY, unpublished results). Arguing against this, however, we have observed that at 18°, a temperature at which the ndl⁹ allele shows greater activity in dorsoventral patterning than does ndl⁰⁴⁶, ndl⁹ is still not complemented by the class I ndl¹⁸ allele. We suggest that the ndl⁹ and ndl⁰⁴⁶ proteases may have qualitatively distinct impairments that explain their differing behaviors in complementation experiments.

For example, the ndl⁰⁴⁶ protease might be defective in self-cleavage at its zymogen activation site (ZUR and NEMERSON 1978 ) but, in the presence of a small amount of wild-type protease (class I proteins) or a weakly functional protease (ndl⁹ protein) capable of cleaving this site, could be converted to a form having greater catalytic activity against downstream substrates. The ndl⁹ protease might be more generally defective, lacking interactions with an essential prodomain. The complementation of ndl⁹ by the relatively high-expressing class I alleles ndl¹⁵ and ndl^LP-2 might involve stabilization of the class I protein during oogenesis by the ndl⁹ protein, as well as the class I and ndl⁹ proteases acting in parallel to propel the total Nudel protease activity over a threshold required for activity of the dorsoventral protease cascade (HONG and HASHIMOTO 1996 ).

Potential basis for antagonistic interactions:
A potential mechanism that may underlie the inhibition of ndl⁹ and ndl⁰⁴⁶ activity by each of the completely inactive class II alleles is suggested by biochemical studies of mutations in the human coagulation protease, Factor IX, that cause severe hemophilia. A subset of these mutations, known as hemophilia B_m mutations, overlaps closely with those found in the strong class II ndl mutations (GIANNELLI et al. 1998 ; indicated on Fig 5). The mutant Factor IX proteins produced by these alleles have the unusual characteristic of inhibiting Factor VII-mediated coagulation in laboratory assays. These proteins bind and are cleaved by the activating protease Factor VIIa, but are not readily released to allow Factor VIIa to cleave another substrate molecule, apparently because these proteins fail to undergo the conformational changes associated with formation of the catalytic site and the substrate-binding pocket (OSTERUD et al. 1981 ; BERTINA et al. 1990 ). These proteins remain in a zymogen-like conformation, able to strongly bind their activating protease but not substrates or inhibitors.

Because the Nudel protease undergoes auto-catalytic activation (LEMOSY et al. 1998 ), the similar mutations in the inactive class II ndl alleles would likely result in stable binding of the inactive class II protein to the partially active ndl⁰⁴⁶ or ndl⁹ protease, thus preventing activation of other ndl⁰⁴⁶ or ndl⁹ zymogen molecules. These alleles do not have detectable dominant-negative effects against the wild-type Nudel protease, perhaps because the wild-type enzyme is present in excess over what is required for activating the dorsoventral protease cascade.

	ACKNOWLEDGMENTS

We are grateful to Li-Mei Pai and Trudi Schüpbach for providing the ndl^LP-1 and ndl^LP-2 alleles; Elizabeth Johnston for assistance in the RGD mutagenesis; and Chris Borland, Cathy Branda, and Michael Tiemeyer for helpful discussions of the manuscript. E.K.L. was supported by postdoctoral fellowships from the American Heart Association, Heritage Affiliate, and the National Institutes of Health (HD-08041). This work was further supported by National Institutes of Health grant GM-49370 (C.H.).

Manuscript received August 4, 1999; Accepted for publication September 13, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSON, K. V., 1998  Pinning down positional information: dorsal-ventral polarity in the Drosophila embryo. Cell 95:439-442[Medline].

ANDERSON, K. V. and C. NÜSSLEIN-VOLHARD, 1984  Information for the dorsal-ventral pattern of the Drosophila embryo is stored as maternal mRNA. Nature 311:223-227[Medline].

BERTINA, R. M., I. K. VAN DER LINDEN, P. M. MANNUCCI, H. H. REINALDA-POOT, and R. CUPERS et al., 1990  Mutations in hemophilia B_m occur at the Arg180-Val activation site or in the catalytic domain of Factor IX. J. Biol. Chem. 265:10876-10883[Abstract/Free Full Text].

BOURIN, M.-C. and U. LINDAHL, 1993  Glycosaminoglycans and the regulation of blood coagulation. Biochem. J. 289:313-330.

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

FARGNOLI, J. and G. WARING, 1982  Identification of vitelline membrane proteins in Drosophila melanogaster.. Dev. Biol. 92:306-314[Medline].

FURIE, B., D. H. BING, R. J. FELDMANN, D. J. ROBISON, and J. P. BURNIER et al., 1982  Computer-generated models of blood coagulation Factor Xa, Factor IXa and thrombin based upon structural homology with other serine proteases. J. Biol. Chem. 257:3875-3882[Abstract/Free Full Text].

GIANNELLI, F., P. M. GREEN, S. S. SOMMER, M. POON, and M. LUDWIG et al., 1998  Haemophilia B: database of point mutations and short additions and deletions—eighth edition. Nucleic Acids Res. 26:265-268[Abstract/Free Full Text].

GREER, J., 1990  Comparative modeling methods: applications to the family of the mammalian serine proteases. Proteins 7:317-334[Medline].

HONG, C. C. and C. HASHIMOTO, 1995  An unusual mosaic protein with a protease domain, encoded by the nudel gene, is involved in defining embryonic dorsoventral polarity in Drosophila. Cell 82:785-794[Medline].

HONG, C. C. and C. HASHIMOTO, 1996  The maternal nudel protein of Drosophila has two distinct roles important for embryogenesis. Genetics 143:1653-1661[Abstract].

KONSOLAKI, M. and T. SCHÜPBACH, 1998  Windbeutel, a gene required for dorso-ventral patterning in Drosophila, encodes a protein that has homologies to vertebrate proteins of the endoplasmic reticulum. Genes Dev. 12:120-131[Abstract/Free Full Text].

LAMANDE, S. R., E. SIGALAS, T. C. PAN, M. L. CHU, and M. DZIADEK et al., 1998  The role of the alpha3(VI) chain in collagen VI assembly. Expression of an alpha3(VI) chain lacking N-terminal modules N10-N7 restores collagen VI assembly, secretion, and matrix deposition in an alpha3(VI)-deficient cell line. J. Biol. Chem. 273:7423-7430[Abstract/Free Full Text].

LEMOSY, E. K., D. KEMLER, and C. HASHIMOTO, 1998  Role of Nudel protease activation in triggering dorsoventral polarization of the Drosophila embryo. Development 125:4045-4053[Abstract].

LEMOSY, E. K., C. C. HONG, and C. HASHIMOTO, 1999  Signal transduction by a protease cascade. Trends Cell Biol. 9:102-107[Medline].

LEVIS, R., P. M. BINGHAM, and G. M. RUBIN, 1982  Physical map of the white locus of Drosophila melanogaster.. Proc. Natl. Acad. Sci. USA 79:564-568[Abstract/Free Full Text].

LIM, J. K. and L. A. SNYDER, 1968  The mutagenic effects of two monofunctional alkylating chemicals on mature spermatozoa of Drosophila. Mutat. Res. 6:129-137[Medline].

LU, D., X. YUAN, X. ZHENG, and J. E. SADLER, 1997  Bovine enteropeptidase is activated by trypsin, and the specificity of enteropeptidase depends on the heavy chain. J. Biol. Chem. 272:31293-31300[Abstract/Free Full Text].

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].

NEMERSON, Y. and B. FURIE, 1980  Zymogens and cofactors of blood coagulation. CRC Crit. Rev. Biochem. 9:45-85[Medline].

NILSON, L. A. and T. SCHÜPBACH, 1998  Localized requirements for windbeutel and pipe reveal a dorsoventral prepattern within the follicular epithelium of the Drosophila ovary. Cell 93:253-262[Medline].

NOGUERÓN, M. I. and G. L. WARING, 1995  Regulated processing of dec-1 eggshell proteins in Drosophila. Dev. Biol. 172:272-279[Medline].

OSTERUD, B., C. K. KASPER, K. K. LAVINE, C. PRODANOS, and S. I. RAPAPORT, 1981  Purification and properties of an abnormal blood coagulation factor IX (factor IXBm)/kinetics of its inhibition of factor X activation by factor VII and bovine tissue factor. Thromb. Hemost. 45:55-59[Medline].

PAI, L.-M., C. PEREZ, and T. SCHÜPBACH, 1998  Screens for genes involved in oogenesis in follicle cells. Ann. Dros. Res. Conf. 39:752C.

PAI, L.-M., M. PEKAREV, and T. SCHÜPBACH, 1999  A screen for genes involved in axial patterning in follicle cells. Ann. Dros. Res. Conf. 40:771B.

PASCUCCI, T., J. PERRINO, A. P. MAHOWALD, and G. L. WARING, 1996  Eggshell assembly in Drosophila: processing and localization of vitelline membrane and chorion proteins. Dev. Biol. 177:590-598[Medline].

PRIETO, A. L., G. M. EDELMAN, and K. L. CROSSIN, 1993  Multiple integrins mediate cell attachment to cytotactin/tenascin. Proc. Natl. Acad. Sci. USA 90:10154-10158[Abstract/Free Full Text].

ROTH, S., 1998  Drosophila development: the secrets of delayed induction. Curr. Biol. 8:R906-R910[Medline].

RUOSLAHTI, E., 1996  RGD and other recognition sequences for integrins. Annu. Rev. Cell Dev. Biol. 12:697-715[Medline].

SAVANT, S. S. and G. L. WARING, 1989  Molecular analysis and rescue of a vitelline membrane mutant in Drosophila. Dev. Biol. 135:43-52[Medline].

SEN, J., J. S. GOLTZ, L. STEVENS, and D. STEIN, 1998  Spatially-restricted expression of pipe in the Drosophila egg chamber defines embryonic dorsal-ventral polarity. Cell 95:471-481[Medline].

SOTTILE, J. and S. WILEY, 1994  Assembly of amino-terminal fibronectin dimers into the extracellular matrix. J. Biol. Chem. 269:17192-17198[Abstract/Free Full Text].

SPITZER, S. G., U. R. PENDURTHI, C. K. KASPER, and S. P. BAJAJ, 1988  Molecular defect in factor IXBm Lake Elsinore. Substitution of Ala390 by Val in the catalytic domain. J. Biol. Chem. 263:10545-10548[Abstract/Free Full Text].

SPRADLING, A. C., 1986 P element-mediated transformation, pp. 175–197 in Drosophila: A Practical Approach, edited by D. B. ROBERTS. IRL Press, Oxford.

THOMPSON, A. R., 1991  Molecular biology of the hemophilias. Prog. Hemost. Thromb. 10:175-214[Medline].

VERHEYEN, E., and L. COOLEY, 1994 Looking at oogenesis, pp. 545–564 in Drosophila melanogaster: Practical Uses in Cell and Molecular Biology, edited by L. S. B. GOLDSTEIN and E. A. FYRBERG. Academic Press, New York.

WARING, G. L., J. P. DIORIO, and S. HENNEN, 1983  Isolation of germ line-dependent female-sterile mutation that affects yolk specific sequestration and chorion formation in Drosophila. Dev. Biol. 100:452-463[Medline].

WEST, C. M., 1986  Current ideas on the significance of protein glycosylation. Mol. Cell. Biochem. 72:3-20[Medline].

WIESCHAUS, E., and C. NÜSSLEIN-VOLHARD, 1986 Looking at embryos, pp. 199–227 in Drosophila: A Practical Approach, edited by D. B. ROBERTS. IRL Press, Washington, DC.

WILLNOW, T. E., K. ORTH, and J. HERZ, 1994  Molecular dissection of ligand binding sites on the low density lipoprotein receptor-related protein. J. Biol. Chem. 269:15827-15832[Abstract/Free Full Text].

YAMAMOTO, T., C. G. DAVIS, M. S. BROWN, W. J. SCHNEIDER, and M. L. CASEY et al., 1984  The human LDL receptor: a cysteine-rich protein with multiple Alu sequence in its mRNA. Cell 39:27-38[Medline].

ZUR, M. and Y. NEMERSON, 1978  The esterase activity of coagulation factor VII. J. Biol. Chem. 253:2203-2209[Free Full Text].

This article has been cited by other articles:

				W. Tadros, S. A. Houston, A. Bashirullah, R. L. Cooperstock, J. L. Semotok, B. H. Reed, and H. D. Lipshitz Regulation of Maternal Transcript Destabilization During Egg Activation in Drosophila Genetics, July 1, 2003; 164(3): 989 - 1001. [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]

				J.-H. Han, S. H. Lee, Y.-Q. Tan, E. K. LeMosy, and C. Hashimoto Gastrulation Defective is a serine protease involved in activating the receptor Toll to polarize the Drosophila embryo PNAS, August 1, 2000; 97(16): 9093 - 9097. [Abstract] [Full Text] [PDF]