Abstract
Mutants of the necrotic (nec) gene in Drosophila melanogaster die in the late pupal stage as pharate adults, or hatch as weak, but relatively normal-looking, flies. Adults develop black melanized spots on the body and leg joints, the abdomen swells with hemolymph, and flies die within 3 or 4 days of eclosion. The TOLL-mediated immune response to fungal infections is constitutively activated in nec mutants and pleiotropic phenotypes include melanization and cellular necrosis. These changes are consistent with activation of one or more proteolytic cascades. The nec gene corresponds to Spn43Ac, one of a cluster of three putative serine proteinase inhibitors at 43A1.2, on the right arm of chromosome 2. Although serpins have been implicated in the activation of many diverse pathways, lack of an individual serpin rarely causes a detectable phenotype. Absence of Spn43Ac, however, gives a clear phenotype, which will allow a mutational analysis of critical features of the molecular structure of serpins.
THE serpins (serine proteinase inhibitors) form a divergent group of proteins that are found in plants, animals, and viruses and have been most widely characterized in mammals (Carrell and Travis 1985; Marshall 1993; Potempaet al. 1994; Wright 1996). Serpins bind to the active site of their target proteases as competitive substrates that block the protease activity. The binding of serpins to their cognate protease to form a Michaelis complex occurs via a “bait” region on the exposed loop of the serpin. The protease cleaves the serpin between the P1 and P1′ residues, which releases the loop and covalently attaches the serpin to the protease. The serpin then undergoes a large internal rearrangement in which the unconstrained reactive center loop (RCL) inserts into β-sheet A (Loebermannet al. 1984; Baumann et al. 1991, 1992). These changes trap the protease in an inactive, covalently linked complex with the serpin. Under some conditions, the conformational change within the serpin can occur spontaneously, without cleavage, to give an inactive “latent” conformation (Mottonenet al. 1992; Carrellet al. 1994; Schreuderet al. 1994; Lomaset al. 1995). In the absence of serpins, serine proteases may cleave their normal substrate to give an activated form that can initiate a proteolytic cascade. In mammals, a variety of proteolytic cascades, including blood coagulation, fibrinolysis, complement activation, and the inflammatory response, are regulated in this way (Boswell and Carrell 1988; Potempaet al. 1994).
Invertebrate serpins are less well characterized. Several serpins have been isolated in Manduca sexta (Kanostet al. 1989; Jiang et al. 1994, 1996) and eight in Drosophila melanogaster (Clarket al. 1995; Bayeret al. 1996; Hanet al. 2000). No genetic functions have been identified with serpin transcripts although one of them is known to inactivate trypsin-like proteases in vitro (Hanet al. 2000).
In the mouse, genetic knockout of a large number of serpins has failed to identify mutant phenotypes (D. Lomas, personal communication), with the exception of antithrombin, which causes fetal abortion. By this criterion, most serpins are functionally redundant. The target specificity of serpins tends to be toward a general class of proteases and the serpin/protease balance is actively regulated. As a consequence, lack of an individual serpin usually results in upregulation of similar family members and has limited phenotypic consequences. In humans, a number of pathologies are associated with serpin abnormalities (Lomaset al. 1992; Aulaket al. 1993; Bruceet al. 1994; Daviset al. 1999) and a few of these result from simple lack-of-function mutations (Erdjumentet al. 1988). In general, however, loss of an individual serpin activity causes little or no phenotypic change. Pathological changes, however, can be caused by gain-of-function serpin mutations. One class of mutation gives serpin protein, which forms concatenated polymers in which the RCL of one molecule inserts into β-sheet A of another. These polymers have no protease inhibitory activity and their accumulation leads to a range of disease states. Polymerization of a mutant form of the acute phase serpin, α1-antitrypsin, is associated with liver disease and emphysema in humans (Lomaset al. 1992). The polymerized serpin is retained within hepatic inclusions, leading to reduced levels in the plasma. Serpin polymerization has also been linked to dementia (Daviset al. 1999), while more unusual serpins may suppress tumor growth (Maspin; Zouet al. 1994); inhibit caspases (CrmA; Takahashiet al. 1996); and bind DNA, leading to heterochromatinization in avian blood cells (Ment; Grigoryevet al. 1999). In all these cases, the ability of serpin molecules to undergo conformational changes underlies their biological activity.
The cloning and sequencing of three Drosophila serpin transcripts is described here. Spn43Aa is just proximal to the prickle (pk) transcript (Gubbet al. 1999), while Spn43Ab and Spn43Ac are within the first 5′ intron of pk (Figure 1). Developmental expression profiles and imaginal disc in situ hybridization patterns of the Spn43 transcripts are shown. The deduced amino acid sequences of the serpins are compared and possible reactive centers are identified. The cytogenetic location of nec was reported by Heitzler et al. (1993) and its role in the TOLL-mediated innate immune response to fungal infection by Levashina et al. (1999). The necrotic phenotype is further characterized in this report and its rescue by a genomic DNA fragment containing Spn43Ac is demonstrated in transgenic flies. Both the nec1 and nec2 chromosomes have lesions within the coding region of the Spn43Ac transcript.
MATERIALS AND METHODS
Drosophila stocks: The nec alleles and the Df(2R)sple-D1 and Df(2R)sple-D2 chromosomes used in this study are from Heitzler et al. (1993). Df(2R)pk-78k is from Gubb and Garcia-Bellido (1982), Df(2R)nap-2 from Ringo et al. (1991), and Df(2R)pk-sple-25 and Df(2R)pk-30 from Gubb et al. (1999). The entire nec region is deleted by Df(2R)pk-78k (42E3;43C3), while the molecular boundaries of nec are defined by the overlapping deletions Df(2R)sple-D1 (43A1.2;43B2) and Df(2R)nap-2 (41F4-9;43A1.2). The fluorescent CyO-GFP balancer stock In(2LR)O, Cy dplvI pr cn1 P[w+mC Act:GFP] was constructed by Reichhart and Ferrandon (1998). The white-mottled-4 (wm4) mutation (Muller 1930) was used to test whether the Spn43A transcripts act as dominant suppressors or enhancers of heterochromatic position effect variegation.
Cloning and sequencing: Standard molecular biological techniques were used (Sambrooket al. 1989). Genomic inserts were subcloned from phage in the 43A1.2 region isolated from the EMBL3 library of John Tamkun (Gubbet al.1999). Spn43Aa cDNAs were isolated from the imaginal disc library of Brown and Kafatos (1988) using a 3.25-kb EcoRI fragment (Figure 1) from phage FP11/2 (http://www.gen.cam.ac.uk/dept/gubb.html). Spn43Ab and Spn43Ac cDNAs were isolated from the larval and adult head libraries of Russell and Kaiser (1993) using the 3.2-kb SalI and 2.1 + 6.0-kb SalI fragments, respectively, from the adjacent phage insert (FP10/2, http://www.gen.cam.ac.uk/dept/gubb.html). Putative full-length cDNAs Spn43Aa-NB3, Spn43Ab-SL2, and Spn43Ac-SH8 were subcloned into pBluescript SK+ (Stratagene, La Jolla, CA) and sequenced. Subcloned genomic fragments from the phage walk were sequenced to identify the precise intron/exon structure of these transcripts.
The boundaries of the Df(2R)pk-30 deletion within the pk and Spn43Ab transcripts were defined by sequencing across the deletion end points. Genomic DNA from homozygous Df(2R)pk-30 and wild-type (Canton-S) flies was PCR amplified from 5′ and 3′ primers (CATCGGCACTCGGATCACA and GTTCTCGAGAGATGGTGAC, respectively). The amplification products were 2.9 kb for Canton-S and 1.8 kb for Df(2R)pk-30. The SpeI/XhoI fragment (Figure 1) from the Df(2R)pk-30 PCR product was subcloned into pBluescript SK+ and sequenced from vector primers. The nec1 and nec2 mutations were PCR sequenced using the CTGGCTGCTCAGACCT TCGCC and CATGGGCGTGGGATACTCCAC primers.
Analysis of sequence data: The Wisconsin Package Version 9.1 [Genetics Computer Group (GCG), Madison, Wisconsin] was used for sequence alignment and assembly. DNA sequences for each transcript were compared to database sequences using the Blast program (Altschulet al. 1990) and protein motifs were compared using the ProSite database (Bairochet al. 1997).
Northern hybridization: Total and poly(A)+ RNA extractions and Northern blotting experiments were performed as described in Lemaitre et al. (1996). Probes corresponding to the cDNA of Spn43Aa, Spn43Ab, and Spn43Ac were amplified by PCR using internal-specific primers. A total of 5 μg of poly(A)+ RNA was loaded for each track. A ribosomal protein (Rp49) probe was used as a loading control (O'Connell and Rosbach 1984).
Tissue in situ hybridization: Random-primed digoxygenin (DIG) DNA probes were made with the Boehringer kit and developed with DAB. Template DNAs were gel-purified inserts of the 1.3-kb EcoRI fragment of Spn43Aa-NB3, the 1.3-kb EcoRI fragment of Spn43Ab-SL2, and the 0.6 + 0.7-kb EcoRI fragments of Spn43Ac-SH8.
Transformation of flies: Genomic constructs of each of the three serpins were made using the pWhiteRabbit transformation vector (Dunin-Borkowski and Brown 1995) and genomic fragments of 3.2, 5.2, and 7 kb, respectively (Figure 1). These genomic fragments were cut from phage FP11/3 and FP10/2, in the case of Spn43Aa and Spn43Ac, while the Spn43Ab rescue fragment was cut from a cosmid isolated from the library of J. Tamkun. A total of 1 mg/ml of transformation construct DNA plus 0.25 mg/ml of the helper plasmid ppi25.7wc were microinjected into yellow white embryos (Spradling 1986).
Test for RCL activity in SPN43Ac: To test whether the SPN43Ac protein functions as an active serine protease inhibitor, the putative protease cleavage site (P1 and P1′, Figure 4) was changed from L438-S439 to P438-P439 by PCR-directed mutagenesis. The resulting construct was sequenced, cloned into pUAST, and rescue experiments were made as in Levashina et al. (1999).
Modeling of SPN43Ac structure: A model of the tertiary structure of SPN43Ac was created using the Modeler program (Sali and Blundell 1993). The template structure to which the SPN43Ac sequence was fitted is based on the X-ray crystal structures of M. sexta serpin 1K (Liet al. 1999) and human α1-antitrypsin (P. R. Elliott, X. Y. Pei, T. Dafforn, R. J. Read, R. W. Carrell and D. A. Lomas, unpublished data).
RESULTS
Identification of transcripts: Three short transcripts were identified on developmental Northern blots within a 10-kb region in 43A1.2 (Figure 1). Flies heterozygous for the overlapping deletions that remove the two distal transcripts (Df(2R)sple-D2/Df(2R)nap-2) express amorphic pk and nec mutant phenotypes, but are otherwise wild type (Figure 1). A fourth transcript, expressed at much lower levels, was later identified, which corresponds to the 5′ exon of the pk transcript (Gubbet al. 1999). Spn43Aa maps just proximal to pk, while Spn43Ab and Spn43Ac map within the 5′ pk intron.
—Molecular map of the Spn43A cluster. The Spn43A transcripts map within ∼10 kb, one proximal to pk, on the same strand, and two in the opposite orientation, within the 5′ intron of pk. The Df(2R)sple-D2/Df(2R)nap-2 trans-heterozygotes survive and give a pk nec phenotype. The nec phenotype of Df(2R)sple-D2/Df(2R)nap-2 flies is rescued by a 7.0-kb XB genomic fragment that includes Spn43Ac. Restriction sites: E, S, Spe, B, X, and H (EcoRI, SalI, SpeI, BamHI, XhoI, and HindIII).
Characterization of the nec mutant phenotype: Balanced stocks of nec alleles tend not to give homozygous flies. This is true of our nec1/CyO and nec2/CyO stocks and although trans-heterozygous nec1/nec2 flies survive for several days (see below), the females are sterile. Given that nec females are completely sterile and nec males are weak, any homozygous flies that might hatch in a balanced stock would not give progeny. Under these conditions, accumulation of additional lethal mutations, or genetic modifiers, would not be selected against. In this study, the nec phenotype was characterized in nec1/nec2 mutant flies, which show a nec phenotype indistinguishable from that of the overlapping deletion combination Df(2R)sple-D2/Df(2R)nap-2.
The survival rate of nec1/nec2 larvae was compared with that of heterozygous balancer larvae, using the CyO-GFP balancer (materials and methods). In the progeny of nec1/CyO-GFP × nec2/CyO-GFP, normal and fluorescent first instar larvae were found close to the expected 1:2 ratio (nec1/nec2):(nec/CyO-GFP); 811 of 1850 third instar larvae were scored. These larvae were reared separately until eclosion and both classes were viable. Between 10 and 20% of the nec larvae show brown spots around the posterior spiracles. Both nec and Cy larvae pupated normally, but ∼10% of nec pupae fail to eclose. These results confirm that the major deleterious effect of the nec mutation occurs after the embryonic and larval stages.
Adult nec1/nec2 flies show a patchy distribution of melanotic spots of variable position and intensity (Figure 2). These spots are restricted to the cuticular surfaces; melanotic masses were never observed within internal tissues. The abdomens of nec adults gradually swell with hemolymph and are extremely distended by 48 hr posteclosion.
Transmission electron microscopy of mutant tissues showed that the epidermal cells undergo necrosis (Figure 3). The sites of necrosis correspond to the sites of extensive melanization of the cuticle, although it is unclear whether the cuticular melanization is caused by the necrosis of the underlying epidermal cells. Interestingly, an additional layer of healthy epidermal cells is seen beneath the necrotic cells. The necrotic mutant phenotype is clearly pleiotropic and it is unclear what causes the lethality of adult nec flies.
—nec1/nec2 adult male. Melanotic patches (arrows) are distributed over the body surface, but preferentially near cuticular joints and sutures. Patches tend to be concentrated at the leg joints, proboscis, ocelli and eyes, the lateral and dorsal thoracic regions, and near the segment boundaries in the abdomen.
Nucleotide and deduced amino acid sequences: Each of the three short transcripts in 43A1.2 (Figure 1) shows homology to the serpin family. The most proximal cDNA, Spn43Aa-NB3, is 1300 nucleotides long, Spn43Ab-SL2 is 1333 nucleotides, and Spn43Ac-SH8 is 1523 nucleotides. These cDNAs encode putative 370-, 394-, and 477-amino-acid peptides, respectively (Figure 4). Spn43Ac has two short introns while Spn43Aa and Spn43Ab each have three (Figure 1). All three proteins contain putative signal peptides (mnhwlsiillgvwisapeg, SPN43Aa; maviisclllllatvsqs, SPN43Ab; and maskvsilllltvhllaaqtfa, SPN43Ac; Nielsenet al. 1997), suggesting that they encode secreted proteins.
The SPN43A serpins are widely diverged from each other, with similar divergences between these serpins and the D. melanogaster Acp76A serpin, M. sexta, Bombyx mori, and mammalian serpins (Table 1). The reactive center loops of SPN43Aa and SPN43Ac contain hinge regions (AAGAS and ASAAS, respectively) typical of inhibitory serpins. The target protease specificity of serpins is strongly influenced by the sequence at the P1 P′1 site on the reactive center loop. The residues in these positions in SPN43Aa and SPN43Ac are MS and LS, respectively (Figure 4). These residues suggest that SPN43Aa is in the antitrypsin inhibitor class, while SPN43Ac is in the antichymotrypsin inhibitor class. The SPN43Ab reactive center loop lacks the typical hinge region and instead contains bulky residues, which would suggest that it is noninhibitory.
The structural homology between the Drosophila SPN43A proteins and serpins for which three-dimensional structures have been solved allows putative three-dimensional structures to be assigned. Any insertions or deletions within the peptide sequences with respect to known serpin scaffolds will be apparent. With all three SPN43A serpins, the only changes are in surface loops that are unlikely to contribute to function, with the exception that SPN43Ac contains a long N-terminal extension of 88 amino acids that includes polyglutamine repeats. Among known serpins, the presence of a polyglutamine repeat is unique to SPN43Ac. Polyglutamine repeats have been described in a number of mammalian proteins, where their function remains unclear, and are also found in Drosophila proteins. Other unusual features of the SPN43A serpins are that SPN43Ab is highly basic (with a predicted isoelectric point of 10); while SPN43Aa and SPN43Ac contain a leucine zipper motif.
Temporal expression patterns: Spn43Aa is expressed predominantly in the early pupae and at much lower levels in the embryo and late larval stages. Spn43Ab is expressed from late embryogenesis onward, with the exception of the early pupal stages. Spn43Ab and Spn43Aa are on opposite DNA strands and their temporal expression patterns are reciprocal. The pk and Spn43Aa transcripts are in the same 5′ to 3′ orientation (Figure 1) and are expressed at similar stages (Gubbet al. 1999). Spn43Ac is expressed at very low levels in the larva and early pupae and at moderate levels in late pupae and adults (Figure 5).
—Electron microscope sections of adult cuticle and epithelium. (A) Wild type showing lightly stained cuticle (dashed line). The underlying epithelial cells show irregular light nuclei containing discrete nucleoli and a microvillar layer (V) that is presumably secreting the basal layer of cuticle (arrow). Necrotic cells are shown in B–D with electron-dense cuticle, from a melanized region in an adult nec1/nec2 fly. This epithelial layer shows no microvilli and the nuclei are rounded (arrows) with irregular inclusions, with the appearance of fragmenting chromatin. An additional layer of healthy proliferating cells (dashed line) underlies the primary epithelium. In D, the necrotic cell nucleus (arrow) shows blebbing of nuclear membrane and is surrounded by vesiculated cellular debris. In contrast, the underlying cells show lightly stained nuclei, normal looking mitochondria (M), and microvilli on the apical surface (V).
—Sequence alignment. The C-terminal region of the SPN43A serpins compared with the M. sexta Serpin-1 exon 9B variant (alaserpin SwissProt P14754); human α1-antichymotrypsin (EMBL X68733) and B. mori antichymotrypsin II (SwissProt P80034); and D. melanogaster serpins SP1, 2, and 6 (EMBL AJ251744, 251745, and 251749). Conserved amino acids are in boldface type and the L-zipper motifs in SPN43Aa and SPN43Ac are shown in bold italics. The reactive center loop is underlined (---), the hinge region is indicated (^^^^^), and the hinge sequences of SPN43Aa and SPN43Ac are shown in boldface. The protease cleavage sites (P1 and P1′) are marked with **.
Spatial expression patterns: In general, the Spn43A transcripts are not expressed at high levels in imaginal discs. Localized expression of Spn43Aa occurs at the sites of innervated bristles on the notum and wing and both Spn43Aa and Spn43Ab are expressed weakly in the eye disc. Spn43Ab gives concentric rings in the leg disc with a central dot at the position of the presumptive tarsal claw and is expressed after the morphogenetic furrow in the eye (Figure 6). Spn43Ac expression was not detected in imaginal discs. SPN43Ac protein, however, is present in adult fat body, but not detected in epidermis or blood cells (J.-M. Reichhart, unpublished results).
Percentage amino acid identity (similarity) of serpins
Heterochromatic inactivation: To test whether any of the 43A serpins might stabilize heterochromatin, similar to the avian MENT serpin (Grigoryevet al. 1999), deletions in the 43A1.2 region were crossed with wm4 to see if they acted as suppressors or enhancers of position effect variegation (for review, see Henikoff 1990). In a wm4 background, Df(2R)pk-sple-51/+, Df(2R)nap-2/+, Df(2R)sple-D2/+, Df(2R)sple-D1/+, Df(2R)pk-30/+ and nec1/+ all retain the normal level of variegated w expression (data not shown). In addition, the homozygous deletion of Spn43Ab, in wm4; Df(2R)pk-30/Df(2R)pk-30 flies, does not modify the wm4 phenotype.
—Developmental Northerns. Repeated probings of a single filter with (A) a Spn43Aa probe, (B) a Spn43Ab probe, and (C) Spn43Ac + Rp49 (arrow, and at equivalent positions in A and B) probes. The Spn43Aa transcript is expressed at high levels during the prepupal and early pupal stages. Spn43Ab and Spn43Ac are expressed from late embryo to adult stages except that Spn43Ab is inactive during the prepupal and early pupal stages, while Spn43Aa is being actively transcribed. Developmental stages: embryonic, (E1) 0–3 hr, (E2) 3–6 hr, (E3) 6–12 hr, (E4) 12–24 hr; larval, (L1, 2, and 3) first, second, and third instars; pupal, (PP) white prepupae, (P1) 12–48-hr pupae, (P2) 48–96-hr pupae; adult male (M); and female (F).
Identification of the nec transcript: The genetic limits for the nec transcript are within the overlap between Df(2R)sple-D2 and Df(2R)nap-2 (Figure 1), an interval that extends for 17–20 kb within the pk 5′ intron (Gubbet al. 1999). The Df(2R)sple-D1 breakpoint is proximal to Spn43Aa (Figure 1) by at least 8 kb (data not shown), but the phenotype of the Df(2R)spleD1/Df(2R)nap-2 heterozygote is indistinguishable from that of Df(2R) spleD2/Df(2R)nap-2; the nec and pk phenotypes do not become more extreme, implying that Spn43Aa cannot be nec and that there is no additional phenotype that could be attributed to the Spn43Aa transcript. Similarly, the Spn43Ab transcript appears to be redundant as the Df(2R)pk-30 deletion gives only an amorphic pk phenotype (Df(2R)pk-30 flies can be maintained as a fertile homozygous stock). Sequencing across the Df(2R)pk-30 deletion endpoints confirms that it corresponds to a microdeletion of 1306 bp, extending from bp 27 of the pk 5′ exon to 7 bp 5′ of the second intron of Spn43Ab, including the putative active site loop. These results imply that Spn43Ac might correspond to nec, but do not preclude an unidentified transcript, or a requirement for more than one of the Spn43A serpins.
—Imaginal disc tissue in situ. (A) Spn43Aa hybridizes to presumptive sites of sensory bristles and sensillae in the wing, either side of the anterior D/V boundary and the presumptive vein 3 campaniform sensillae (arrow). Spn43Aa does not hybridize to the leg disc or eye disc, except weakly behind the morphogenetic furrow. Spn43Ab does not hybridize to the wing disc, but the leg (B) disc shows rings of staining close to segment boundaries and at the center of the disc at the site of the presumptive tarsal claws, and the eye disc (C) shows staining along the lateral boundaries and the morphogenetic furrow (arrow). nec transcripts were not detected in imaginal discs.
Rescue of phenotype: Transgenic rescue constructs containing genomic fragments spanning the three Spn43A transcripts (Figure 1) were tested in a nec1 bwD/Df(2R)pk-78k background (Table 2). For the P[Spn43Aa+] and P[Spn43Ab+] crosses, nec flies hatched, although at the reduced frequencies compared to nec+ siblings. These flies developed necrotic patches within 24 hr and died within 3 days of eclosion. In contrast, nec1 bwD Df(2R)pk-78k; P[Spn43Ac]/+ transformants eclosed at the expected frequency. When reexamined 10 days later, the nec1 bwD/Df(2R)pk-78k; P[Spn43Ac]/+ flies remained wild type for nec, indicating that the P[Spn43Ac] insert rescues the necrotic phenotype completely. The viability of nec1 bwD/Df(2R)pk-78k; P[Spn43Ac]/+ flies under these conditions is indistinguishable from their nec1 bwD/CyO; P[Spn43Ac]/+ siblings, with >98% surviving for 10 days. At 29°, 114/118 adult nec1/nec2; UAS-p[Spn43Ac]/Gal4-da survive for 7 days (Figure 8), with 113 remaining alive after 9 days.
Having established rigorously that the Spn43Ac corresponds to nec, we henceforth refer to this transcript as nec.
Characterization of the nec1 and nec2 mutations: PCR sequencing of DNA from nec1/Df(2R)pk-78k and nec2/Df(2R)pk-78k flies identified a 6-bp deletion in nec1, resulting in deletion of two isoleucine residues at positions 118 and 119. In nec2, Q37 is replaced by a stop codon, giving a 5′ truncation of the peptide within the polyglutamine repeat.
Suppression of adult lethality: nec has recently been shown to control expression of the antifungal peptide Drosomycin via the TOLL pathway (Levashinaet al. 1999). To investigate whether the lethal nec phenotype is associated with activation of the immune response, double mutant flies were constructed with a loss-of-function allele of spaetzle (spz), the gene coding for the Toll-ligand. As shown in Figure 7, nec; spz flies survive significantly longer than their nec siblings (from the cross of nec1/CyO; spz/TM3 × nec2/CyO; spz/TM3). Half of the nec flies are dead within 2 days, while the partially rescued nec; spz flies only start to die after day 2 and it is 5.5 days before half are dead.
Test crosses for rescue of nec phenotype
—The survival of adult nec vs. nec; spz flies at 25°. A total of 222 nec1/nec2 (♦) and 81 nec1/nec2; spzrm7 (▪) flies were counted and transferred daily. Confidence intervals of 95% are ≤0.1 for each point.
Activity of NEC reactive center loop: To test whether the nec phenotype is linked to a serine protease inhibitory function of NEC, the P1 P1′ residues (L438 and S439) of the putative reactive center loop were altered to proline residues (see materials and methods), which are never found at these positions within an active inhibitor. The resulting construct (P[UAS-necPP]) was tested for complementation of the nec phenotype in a simple survival test. The ubiquitously expressed Gal4-da strain was used to drive expression of P[UAS-necPP] using the expression system of Brand and Perrimon (1993). The nec1/nec2; P[UAS-necPP]/Gal4-da flies expressed a nec phenotype indistinguishable from their nec1/nec2; P[UAS-necPP]/MKRS siblings, with an adult half-life of ∼1.5 days posteclosion. In contrast, the wild-type P[UAS-nec] construct (Levashinaet al. 1999) rescues the nec phenotype completely to give a healthy fertile stock in nec1/nec2; Gal4-da/P[UAS-necPP] flies (Table 3, Figure 8). Western blots of hemolymph proteins from nec2/Df(2R)pk78k; Spn43AcPP/+ flies show a strong SPN43Ac-PP protein band, confirming that the transgenic protein is expressed and is stable (N. Pelte and J.-M. Reichhart, unpublished data).
Test crosses for activity of necPP
—The survival of adult nec vs. nec-PP flies at 29°. A total of 153 nec1/nec2 (▪); 81 nec1/nec2; UAS-necPP/Gal4-da (♦); and 237 nec1/nec2; UAS-necPP/Gal4-da (μ) flies were counted and transferred daily. Confidence intervals of 95% are ≤0.1 for each point.
DISCUSSION
We show here that the Spn43Ac serpin transcript corresponds to the nec gene and that it is not functionally redundant. Deletions that include all three serpins of the Spn43A cluster do not enhance the necrotic phenotype compared to that of the nec1/nec2 mutant combination. The other two serpins, Spn43Aa and Spn43Ab, are more typical of serpins in that a deletion gives no detectable phenotype. The nec1 and nec2 mutations both map within the Spn43Ac transcript and a genomic fragment that includes this transcript completely rescues the nec phenotype. The nec1 mutation deletes two isoleucine residues (I118 and I119; Figures 4 and 9) within the putative helix-A of the serpin, leaving the remainder of the nucleotide sequence in frame. This alteration would probably disrupt helix-A and the underlying β-sheet B, which could cause misfolding of the protein. The nec2 mutation causes a Q37 to stop codon transition within the N-terminal polyglutamine repeat that deletes the entire serpin domain. The nec mutant phenotype is not rescued by a transgenic construct carrying L438P + S439P transitions within the putative RCL of NEC, confirming that the NEC protein acts as an active serine protease inhibitor.
In arthropods, injury initiates proteolytic cascades, leading to rapid blood clotting and melanization at the site of injury. The black patches in nec mutants may result from melanization via activation of a phenol oxidase cascade that is under the direct control of NEC. Alternatively, the melanization might be a secondary reaction to cellular damage caused by the activation of a distinct protease cascade in the absence of functional NEC. The additional layer of epidermal cells underlying necrotic patches is reminiscent of the wound healing response seen in mammalian systems.
The demonstration of activation of the TOLL-mediated immune response by NEC (Levashinaet al. 1999) prompted us to construct the double mutant nec; spz strain. In this double mutant strain the nec melanotic patches underlying the cuticle are delayed in appearance and reduced in size. In addition, the survival of adult nec; spz flies is extended relative to nec flies (Figure 7). Interestingly, constitutive activation of the TOLL pathway in the dominant mutant Toll10B does not cause a necrotic phenotype or adult lethality. We propose that the NEC target protease(s) is upregulated by activation of the TOLL pathway. This response would be masked in Toll10B flies, as the serpin activity of the wild-type nec allele would be sufficient to block excess protease. In a nec mutant, however, activation of the TOLL pathway would lead to expression of active protease(s) without the inhibitory NEC serpin. In this case, the active protease(s) would cause cellular and tissue damage and lead to reduced viability. In the nec; spz double mutant, levels of the NEC target protease would remain at the basal level found in unchallenged wild-type flies. This basal level of protease activity could still cause limited damage, given that the nec; spz double mutant flies lack functional NEC inhibitor.
—A homology model of SPN43Ac based on the X-ray crystal structures of M. sexta serpin 1K and human α1-antitrypsin. Views from front and back are shown. The A-β-sheet is green, the reactive center loop is red, and the A-helix is cyan. The unique N-terminal extension of SPN43Ac is omitted. The position of the nec1 deletion is indicated on the back view by a ball-and-stick marker. The picture was produced using Molscript (Kraulis 1991).
The insect serpin most closely related to the cluster of Spn43A transcripts is the M. sexta serpin1-B (Jianget al. 1996) with 25–30% conservation at the amino acid level (Kanostet al. 1989; Jiang et al. 1994, 1996). This level of conservation is characteristic within the serpin family, with the C-terminal half of the protein tending to show the greatest conservation (Sommeret al. 1987). The protease-binding specificity of a particular serpin depends on the amino acid sequence of its reactive center (Boswell and Carrell 1988; Huber and Carrell 1989; Carrell and Evans 1992). The critical importance of the reactive center loop is underlined by the genomic organization of M. sexta serpin-1 (Kanostet al. 1989). This transcription unit contains 12 variants of the exon that encodes the reactive center loop, exon 9. Mutually exclusive splicing of exon 9 generates multiple serpin isoforms (Jianget al. 1994). These variants show differing proteinase affinities, which has been interpreted as suggesting that the protein isoforms of M. sexta serpin-1 gene regulate different aspects of the wound healing and antimicrobial defense (Jianget al. 1996). While this is an attractive possibility, most serpins will inhibit a broad range of proteinases in vitro, so that biochemical assays of serpin/proteinase specificity give a poor indication of physiologically significant interactions. The Drosophila Spn43A cluster of serpins shows no trace of the genomic organization of the Manduca serpin-1, but could regulate some of the activities suggested for the hornworm protein. In particular, the nec phenotype is suggestive of activation of a phenol oxidase cascade, causing epithelial necrosis with subsequent regeneration.
The localized patterns of expression of the Spn43Aa and Spn43Ab serpins in imaginal discs are hard to interpret given that the transcripts appear to encode secreted proteins. To some extent, these transcription patterns may reflect ectopic regulation by enhancer elements in the adjacent pk promotor and be irrelevant to the function of the serpins. The expression of Spn43Aa transcripts in the presumptive vein 3 sensillae, however, is not seen with pk (Gubbet al. 1999), although the expression pattern of these two transcripts along the dorsoventral boundary of the wing is similar. Spn43Ab expression in the wing disc is absent during stages when the Spn43Aa and pk transcripts, on the opposite DNA strand, are being expressed. The expression of Spn43Ab at the segmental boundaries in the leg and at the site of the presumptive tarsal claw in the center of the disc is distinctive and may mark a critical region for the control of leg disc growth. Like MENT, SPN43Ab is extremely basic and might have a similar function in stabilizing heterochromatic domains. This possibility was not confirmed by genetic tests, as none of the Spn43A serpins show the expected dominant suppression of position effect variegation. Even when homozygous, the Df(2R)pk-30 deletion, including the pk 5′ exon and Spn43Ab, does not modify the wm4 phenotype. In addition, the putative signal peptide in SPN43Ab would not be expected in a nuclear protein.
The lack of phenotypes other than nec and pk with deletions that remove all three serpins implies that the Spn43Aa and Spn43Ab serpins are redundant, with other genetic functions mapping elsewhere in the genome. Within the Spn43A cluster itself, however, the three serpin genes do not complement each other. The visible nec phenotype implies an essential role for the NEC serpin and will allow direct mutagenesis screens to identify critical regions within serpin molecules.
Acknowledgments
We thank Reine Klock for the northern blot experiments; Daniel Zachary for the electron microscopy; Darin Coulson, Annie Meunier, and Glynnis Johnson for skilled assistance in fly pushing; Robin Carrell and David Lomas for discussion and critical reading of the manuscript; John Tamkun, Nick Brown, and Steve Russell for cDNA and genomic libraries; and the FlyBase consortium, particularly Rachel Drysdale,for rulings on the Spn nomenclature. This work was funded by Medical Research Council programme grants to Michael Ashburner, David Gubb, and Steven Russell, with E.L. and J.-M.R. being supported by the CNRS and grants from the Marie Curie Research training Program (EU) and the NATO Scientific Research Program.
The EMBL accession numbers for the SPN43A serpins are SPN43Aa (AJ245442), SPN43Ab (AJ245443), and NEC (SPN43Ac) (AJ245444).
Footnotes
-
Communicating editor: T. C. Kaufman
- Received February 24, 2000.
- Accepted July 3, 2000.
- Copyright © 2000 by the Genetics Society of America
