Organization of serpin gene-1 from Manduca sexta. Evolution of a family of alternate exons encoding the reactive site loop.

Manduca sexta serpin gene-1 encodes a family of serpins whose amino acid sequences are identical in their amino-terminal 336 residues but variable in their carboxyl-terminal 39-46 residues, which includes the reactive site loop (Jiang, H., Wang, Y., and Kanost, M. R. (1994) J. Biol. Chem. 269, 55-58). Here, we report the gene's complete nucleotide sequence and exon-intron structure. A unique characteristic of this gene is its exon 9, which is present in 12 alternate forms between exons 8 and 10. Isolation and characterization of cDNA clones containing exons 9C, 9H, and 9I, which were not found previously, indicate that all 12 alternate forms of exon 9 can be utilized to generate 12 different serpins. The splicing pathway apparently allows inclusion of only one exon 9 per molecule of mature serpin-1 mRNA. Analysis of exon-intron border sequences reveals unique features that may be involved in regulation of RNA splicing. The exon 9 region has apparently evolved through rounds of exon duplication and sequence divergence. The exons near the center of the region may have evolved recently, whereas the outermost exons are the most ancient. Exons 9G and 9H were duplicated as a pair from exons 9E and 9F, an event that may have occurred more than once in the history of this gene.

The serpin 1 superfamily contains a large number of proteins that function as inhibitors of serine proteinases as well as proteins related in sequence which are not inhibitors (1). Serpins are typically 370 -390 amino acid residues long, with a reactive site loop 30 -40 residues from the carboxyl terminus. This loop, exposed at the surface of the protein, is the site of interaction between serpins and the serine proteinases they inhibit. The serpin reactive site loop binds to the active site of a target proteinase in a manner similar to the binding of a substrate, forming a very stable serpin-proteinase complex. Formation of this complex involves a specific peptide bond in the reactive site loop, the scissile bond (designated P 1 -P 1 Ј). The amino acid sequence of the reactive site loop determines an inhibitor's selectivity. Altering the reactive site loop sequence, particularly at the P 1 position, can cause dramatic changes in the proteinase selectivity of a serpin (1). Comparisons of serpin sequences have demonstrated that the reactive site loop and adjacent sequence is the least conserved region of the proteins. It has been suggested that after duplications of serpin genes, rapid evolutionary change of the reactive site loop region provides new inhibitor selectivities that may have value during natural selection (2,3).
At least nine different serpins are present in mammalian plasma. They regulate the activity of serine proteinases involved in diverse physiological functions such as blood coagulation, fibrinolysis, complement activation, and inflammatory responses (1). Serpins have also been found in the hemolymph of invertebrates, including three groups of arthropods: insects, crayfish, and horseshoe crabs (4). These arthropod serpins have 12-30% amino acid sequence identity with various mammalian serpins. Like their mammalian counterparts, invertebrate serpins may regulate proteinases released from blood cells during inflammation-like processes, and they may regulate pathways of proteinase cascades that ultimately activate enzymes involved in blood coagulation and melanization (4 -6).
Study of serpins from two species of lepidopteran insects, Manduca sexta (tobacco hornworm) and Bombyx mori (silkworm), has revealed a novel genetic mechanism for generating diversity of serpin reactive site loops. Amino acid sequences of two serpins isolated from B. mori are identical in their first 336 amino acid residues but differ in sequence from residue 337 to the carboxyl terminus (7). We isolated a large number of serpin cDNA clones from M. sexta and found the same phenomenon (8). In 38 cDNA clones, the DNA sequences of the 5Ј-untranslated region, the sequence encoding a secretion signal peptide, and the first 336 amino acid residues of the mature protein were identical. Beyond this point the sequences differed from clone to clone, with 10 distinct sequence variants encoding the carboxyl-terminal 39 -45 residues, which includes the reactive site loop. The 3Ј-untranslated sequences following the translation stop codon were again identical in all of these serpin cDNAs.
The occurrence in M. sexta of serpin cDNAs with a variable region encoding the reactive site loop between two constant regions appears to be due to mutually exclusive alternate exon use (8). The variable sequence is encoded by multiple versions of the ninth exon of M. sexta serpin gene-1. We present here the complete nucleotide sequence of M. sexta serpin gene-1, in which we have identified 12 alternate versions of exon 9. We have also isolated cDNAs corresponding to all of these exon 9 variants, indicating that they are all expressed. An analysis of the region of the gene containing the exon 9 variants suggests a pathway for the duplication and divergence of the alternate exons during evolution of the gene.

EXPERIMENTAL PROCEDURES
Isolation of Genomic Clones-A M. sexta genomic DNA library in phage EMBL3 (9) was screened by hybridization with a 32 P-labeled M. sexta serpin-1B (previously called alaserpin) cDNA clone (10) by methods described by Sambrook et al. (11). A total of 10 5 recombinant plaques (three genomic equivalents) from the library was screened. Positive plaques were purified, and DNA was prepared from plate lysates (12). phage DNA from clone E1 was digested with SalI, and the resulting fragments were subcloned into plasmid pTZ18U. A 2.7-kb SalI-EcoR I fragment from clone A2 was subcloned into plasmid pBluescript(KS).
A polymerase chain reaction (PCR) was used to obtain a 5-kb DNA fragment of the serpin gene-1 which was not represented in the clones. Oligonucleotide primers were designed, based on sequence data obtained from regions near the ends of clone E1 (primer E: 5Ј-CT-T ACG TGG GAT CCA TAG AC-3Ј) and clone A2 (primer A: 5Ј-CGA T-TG AGG TCT AGA AGG A-3Ј). The PCR was carried out using a 1:1 mixture of Taq and Pfu DNA polymerases (Stratagene). Thermal cycling conditions were: 95°C, 20 s; 53°C, 60 s; 72°C, 4.5 min for 30 cycles. The resulting DNA product was isolated by agarose gel electrophoresis and cloned into pGem-T vector (Promega).
Southern Blot Analysis of Genomic DNA-Genomic DNA was isolated from a single M. sexta larva (13). Samples of the DNA (10 g) were digested with restriction enzymes at 37°C for 6 h. The resulting fragments were separated by electrophoresis in a 1% agarose gel, transferred to GeneScreen Plus membrane (DuPont NEN), and hybridized with 32 P-labeled serpin-1B cDNA (10), all according to methods described by Sambrook et al. (11).
Isolation of Serpin cDNA Clones by PCR-DNA isolated from an amplified M. sexta hemocyte cDNA library (8) was used as template for a PCR to obtain serpin cDNA fragments containing sequences encoded by exon 9, using a primer sequence from exon 7 (5Ј-GGA TCC TA-A AGC TCT TTC-3Ј), and a vector-specific primer (5Ј-GTT TTC-CCA GTC ACG ACG T-3Ј) near the 3Ј end of the cDNA inserts. The PCR products were digested with HindIII and KpnI. The resulting 0.3-kb fragments were cloned into plasmid pBluescript(KS), creating a minilibrary of serpin cDNA clones containing the region from the Hin-dIII site in exon 8 to the 3Ј end. The recombinants that contained exons 9H and 9I were identified by DNA sequence analysis of randomly selected clones. To identify clones containing exon 9C, a genomic fragment containing this exon was amplified by PCR and used to screen the serpin minilibrary by colony hybridization (11). The positive clones were confirmed by DNA sequencing to contain exon 9C.
DNA Sequencing-cDNA and genomic DNA sequences were determined from double-stranded plasmid DNA templates by the dideoxynucleotide method (14) using modified T7 DNA polymerase (U. S. Biochemical Corp.). cDNA inserts were sequenced using T7 or T3 primers and a set of additional primers derived from the serpin cDNA sequence (10). The DNA fragments containing the serpin gene were sequenced using subcloned restriction enzyme fragments, subclones obtained through exonuclease III deletion (15), and oligonucleotide primers derived from previously determined sequences.
Computer-assisted Analysis of Sequence Data-Sequence analysis was performed using the GCG Sequence Analysis Software Package version 7.3.1 (16) and IBI Pustell programs.

RESULTS
Nucleotide Sequence of M. sexta Serpin Gene-1-We screened a M. sexta genomic DNA library with a serpin-1B cDNA probe (previously called alaserpin) (10). We isolated three overlapping clones (C2, A1, E1) and another clone (A2) that did not overlap with this group (Fig. 1). Restriction mapping and Southern blot analysis demonstrated that the 5Ј end of the cDNA hybridized with clone E1 and that clone A2 hybridized with the 3Ј end of the cDNA (data not shown). We used sequence from the ends of E1 and A2 to design primers for a PCR to obtain the missing DNA fragment linking these two clones. Primer E was from a sequence 187 nucleotides from the 3Ј end of E1; primer A was from a sequence 291 nucleotides from the 5Ј end of A2. Using these primers and M. sexta genomic DNA as template in a PCR, we obtained a 5.6-kb product (PCR-1) that overlapped the sequences of E1 and A2 and filled in the missing genomic DNA sequence. The entire region of the serpin gene-1 locus that was subcloned and sequenced is 21,536 bp long (Fig. 2).
Southern blot analysis ( Fig. 3) showed that the number and size of restriction enzyme digestion fragments from M. sexta genomic DNA which hybridize with serpin-1B cDNA probe are consistent with the restriction enzyme map derived from the cloned inserts that were sequenced (Fig. 1). This suggests that there is a single copy of serpin gene-1 in the M. sexta genome. We have identified a second serpin gene, M. sexta serpin gene-2, which is only distantly related to gene-1 and does not cross-hybridize even under low stringency conditions. 2 Transcription Initiation and Upstream Sequences-The transcription initiation site was determined previously by primer extension analysis to be 15 residues in the 5Ј direction from the end of the serpin-1B (alaserpin) cDNA clone (17). This corresponds to a G residue located 38 residues downstream from a TATA sequence (Fig. 2). The five residues preceding the initiation site (TTAGT) coincide well with an arthropod initiator consensus sequence (TCAGT) (18).
Expression of serpin gene-1 in fat body of M. sexta appears to be negatively regulated by ecdysteroids (19). In computer searching to identify putative ecdysteroid receptor-binding sequences in serpin gene-1 we located a sequence beginning at position Ϫ327 (AGGTCAGGAACTC) which matches in 11 out of 13 positions with a consensus ecdysone receptor binding sequence (RGG/TTC/GANTGC/AC/ACTY) (20). Perhaps when ecdysteroid levels are elevated at periods such as the larvallarval molts and prepupal stage, binding of an ecdysteroid receptor to this sequence interferes with transcription of the serpin gene, resulting in the observed decrease in serpin-1 mRNA.
Organization of the Serpin Gene-1-Comparison of the genomic sequence with the sequence of cloned serpin-1 cDNAs indicated that serpin gene-1 is divided into 10 exons, with 12 alternate forms of exon 9 ( Figs. 1 and 2). From the transcription initiation site to the end of exon 10, the gene is 20,581 bp long. There is a clear difference in GC content between the exons (39% GϩC) and the very AϩT-rich introns (30% GϩC). Exon 1 encodes the translation start site and the first six amino acid residues of the signal peptide. Exon 2 encodes the remainder of the signal peptide and the beginning of the mature protein sequence. Exons 3-8 encode the remainder of the serpin-1 sequence up to amino acid residue 336. Twelve alternate versions of exon 9 then occur within a stretch of approximately 11.6 kb. These are followed by exon 10, which encodes the 3Ј-untranslated sequence common to all of the serpin-1 cDNAs.
Analysis of Intron Sequences-We used the FASTA program (GCG) to search the GenBank DNA sequence data base with sequences of the introns from serpin gene-1 and found several high scoring matches with sequences from other M. sexta genes. A 28-nucleotide region within the first intron (between exons 1 and 2) beginning at position 853 was identical at 26 positions with a sequence in the upstream region of the M. sexta microvitellogenin gene (21) (Fig. 4A). The very high sequence similarity and the location of these sequences near the 5Ј ends of two genes expressed in fat body of M. sexta may indicate this sequence as a potential gene regulatory element.
A 173-bp region in the intron between exons 9I and 9J had significant similarity with noncoding sequences from three other M. sexta genes (Fig. 4B): a sequence from intron D of arylphorin gene A (22), a sequence from the intron of the gene for eclosion hormone (23), and a sequence upstream from the gene for larval cuticle protein 14 (24). This sequence may represent a repetitive element in the M. sexta genome.
We also discovered a 201-bp perfect inverted repeat sequence in the intron between exons 9J and 9K from nucleotide 16912 to 17112 (Fig. 5). This sequence could potentially form a large hairpin structure in the serpin-1 pre-mRNA or a cruciform structure in the double-stranded DNA of the serpin gene-1. Such a structure could conceivably influence RNA splicing and alternate exon use in the serpin gene-1. The lack of mismatches in this long inverted repeat suggests that there may be some selective pressure to preserve its structure and function.
Sequence Analysis of the Exon 9 Region-When we searched the sequence between exon 8 and exon 10 for regions that could encode the sequence Pro-Phe-Xaa-Phe, which is conserved in all of the serpin amino acid sequences (see Fig. 7A), we detected 12 open reading frames whose length and sequence are consistent with their identification as alternate versions of exon 9. These are named alphabetically A-K, and then Z. We named the most 3Ј exon 9Z when we did not yet know the total number of exon 9 variants (8). We previously cloned cDNAs that contained regions corresponding with exon 9A, B, D, E, F, G, and Z (8). We also previously obtained cDNA clones for which we did not yet have genomic exon 9 sequence. These have now been located (in the gene fragment obtained by PCR) as exon 9J (previously named 9), and exon 9K (previously exon 9). For another cDNA clone with exon 9, differing from 9J() only by a 6-bp deletion, we did not identify a comparable exon 9 in the genomic sequence. Thus, 9 may be an allelic variant of 9J().
To determine whether all of the exon 9 variants detected in the gene sequence are actually expressed in spliced serpin-1 mRNAs, we used a gene-specific primer and a vector-specific primer to amplify DNA from a M. sexta hemocyte cDNA library and cloned the resulting PCR fragments into plasmid vectors. In this way we obtained a large number of cDNA clones covering the 3Ј part of exon 8 and all of exons 9 and 10. When these clones were sequenced we identified correctly spliced versions of all of the exon 9 variants, including exon 9C, H, and I, which were not obtained in previous library screening. These results indicate that all 12 variants of exon 9 can be correctly spliced and are present in serpin-1 mRNA populations from hemocytes.  When we compared DNA sequences of exon 9 variants in the gene with the corresponding cDNAs, five minor discrepancies were identified: ATG (Met, at 9013) to TTG (Leu) in exon 9D; ATT (Ile, at 10089) to ATC (Ile) in exon 9E; GAT (Asp, at 10964) to GAC (Asp) in exon 9F; CTC (Leu, at 15646) to TTC (Phe) in exon 9J; TTT (Phe, at 17409 -17410) to ACT (Thr) in exon 9K. These differences may result from allelic variation.
We analyzed the sequences at the junctions between the exons and introns of serpin gene-1 (Fig. 6). We speculate that differences in the intron sequences bordering the exon 9 variants compared with other exons in the gene might be involved in regulation of the mutually exclusive use of exon 9 variants. There are some differences in the consensus sequences at the intron borders of the exon 9 variants compared with the consensus sequence for the other exons in the gene and with a consensus sequence derived from many Drosophila genes (25). At position Ϫ4 at the 3Ј end of the introns, T is present in 9 out of 12 of the exon 9 variants. There are no A residues at this position in the exon 9 group. However, in the other group of exons, the consensus is A or T, with no G residues present at position Ϫ4. At the first residue in the exons, the exon 9 group has a consensus for C (9/12) with no G residues. In the exon 1-8, 10 group the consensus is A or G, and C is not represented at this position. In the 5Ј intron splice site at position 5, the exon 9 group has A, C, or T, with no G residues, whereas the consensus for the remainder of the gene, and in Drosophila genes, is G at this position. Also, a strong preference for T at Ϫ3 and A at ϩ3 in the exon 9 group is in contrast to the exon 1-8, 10 group and the Drosophila consensus.
Evolution of the Exon 9 Region-The amino acid sequence encoded by the exon 9 variants includes the reactive site loop, which is the site for interaction of a serpin with the active site of a serine proteinase. Changes in the amino acid sequence of this region can alter a serpin's selectivity for proteinase inhibition. It appears that multiple copies of exon 9 have arisen in serpin gene-1 through repeated exon 9 duplications. We have investigated the pathway by which such duplications might have occurred during evolution of M. sexta serpin gene-1.
We aligned the amino acid sequences encoded by the exon 9 variants (Fig. 7A) as well as the nucleotide sequences of the exon 9 variants and flanking intron sequences (Fig. 7B) and used these alignments to produce evolutionary trees of the exon 9 family. The tree shown in Fig. 8 (derived from the nucleotide sequence alignment in Fig. 7B) is typical of trees produced with a number of slightly different alignments of both the nucleotide and amino acid sequences, although there were small differences in branching order in some trees. All of the trees we produced are consistent with the notion that exons 9A and 9Z and exons 9B and 9K form two exon pairs, each with a member at opposite ends of the exon 9 region, and that these exons diverged earlier than the rest of the exon 9 sequences. The remaining exon 9 sequences are apparently from a single lineage and differ from 9A, 9B, 9K, and 9Z in having conserved cysteine and proline residues near their carboxyl termini (Fig.  7A). The most recent exon duplication events occurred in the center of the exon 9 region, giving rise to exons 9C to 9J, of which exons 9E to 9H are the latest to appear.
The terminal branches of this tree group exon 9E with 9G and 9F with 9H. This suggests that the most recent rearrangement of the exon 9 region was the duplication of a pair of exons, FIG. 6. Sequences at the exon-intron splice junctions in M. sexta serpin gene-1. Consensus sequences were determined at each position when a particular type of residue was present in at least five of eight positions (for borders of exons 1-8 and 10) or 9/12 positions (for borders of exon 9 variants). The Drosophila consensus sequence is from (25). Positions at which the consensus sequence for the exon 9 variants differs from both the consensus for the other serpin gene-1 exons and the Drosophila consensus are underlined. M ϭ A or C; R ϭ A or G; Y ϭ C or T; W ϭ A or T.

FIG. 5. A 201-bp perfect inverted repeat sequence in M. sexta serpin gene-1.
9E and 9F to produce 9G and 9H (or vice versa). This hypothesis is supported by sequence similarity detected in intron sequences. A region of similarity was found in a 363-bp region near the 3Ј ends of introns H e and H g , adjacent to exons 9E and 9G. The Bestfit score for these aligned sequences is 9 standard deviations greater than the mean score for 50 alignments from the randomized sequences of the aligned regions. An even more significant similarity exists in alignment of 392-bp sequences near the 5Ј ends of introns H g and H i , adjacent to exons 9F and 9H (the Bestfit score is 36 standard deviations greater than the mean for 50 alignments of the randomized intron sequences). DISCUSSION We have isolated DNA clones including the locus for M. sexta serpin gene-1 and have determined the nucleotide sequence of a 21.5-kb region that includes the serpin gene-1. The gene is composed of 10 exons, with 12 alternate versions of exon 9. Analysis of intron positions in many genes from the serpin superfamily has shown that there are a large number of sites at which introns are present, and it is likely that new introns have been acquired numerous times during divergence of the superfamily (26). M. sexta serpin gene-1 has introns at a few positions identical with some vertebrate serpin genes. The intron between exons 5 and 6 is at the same position as introns in genes for human ␣ 1 -antichymotrypsin, ␣ 1 -proteinase inhibitor, protein C inhibitor, antithrombin III, C1 inhibitor, leuserpin, ␣ 2 -antiplasmin, angiotensinogen, and mouse serpin J6. This position must therefore be an ancient location for an intron in serpin genes. The introns between exons 6 and 7 and between exons 7 and 8 are at the same positions as introns in the human ␣ 2 -antiplasmin and C1 inhibitor genes. This suggests that an intron at this position may have been present in a common ancestor of M. sexta serpin gene-1 and these two human genes, prior to the evolutionary divergence of the invertebrates and vertebrates, and that this intron may have been lost from other vertebrate serpin genes.
The intron between exons 8 and 9 in M. sexta serpin gene-1 is most intriguing because its presence allows for the alternate exon splicing to produce multiple serpin mRNAs that encode proteins with a constant protein framework and variable reactive site sequences. Among other serpin genes that have been sequenced, an intron at this position has been identified only in the genes for human and rat plasminogen activator inhibitor-1 (27,28). Although no alternate exons have been observed in these genes, the potential exists for alternate exon use in other serpin genes that contain an intron at this position.
Serpin gene duplication and sequence divergence have resulted in the presence of families of serpin genes in mammals such as humans, mice, and cattle (29 -31). Sequence divergence in these genes is greatest in the region that encodes the reactive site, resulting in rapid evolution of serpins with new inhibitor selectivities (4,30). Evolution of the capacity to produce serpins with a variety of reactive site sequences has followed a different route in M. sexta serpin gene-1. Instead of duplication of the entire gene, only the exon encoding the reactive site loop has been duplicated. Evolutionary sequence divergence of these exons has led to the ability to synthesize serpins with 12 different reactive site sequences from a single gene through use of mutually exclusive alternate exon splicing. Similar alternate exon use probably occurs in another lepidopteran insect, the silkworm B. mori (7). It will be of interest to determine how widespread this type of serpin gene structure may be in other insects and arthropods.
Mutually exclusive alternate exon use is known in other genes, most notably tropomyosin genes from mammals, chicken, and Drosophila (33,34). In such genes there are two to four alternate exons, usually with a constitutive, default exon choice and an alternate selected under regulation of cell-specific splicing machinery. Correct splicing of the M. sexta serpin-1 pre-mRNA requires precise selection of only one exon 9 out of 12 choices. This must involve regulation by splicing factors at the level of recognition of exon 9 sequences and exclusion of more than one exon 9 sequence between exons 8 and 10. Sequences of serpin-1 cDNA clones indicate that all of the alternate exon 9 variants in the gene are present in serpin-1 mRNAs and that they are spliced in mature mRNA in a mutually exclusive manner. With only one exception, in which a cDNA contained exon 9B spliced to exon 9C followed by exon 10 (8), we have found that serpin-1 mRNAs contain only a single exon 9 sequence. Differences in the sequences at exon 9-intron borders compared with other exons in the gene (Fig. 6) might be signals for regulation of the splicing process.
From the analysis of the alignments of nucleic acid sequences of the exon 9 variants and alignments of the amino acid sequences they encode (Figs. 7 and 8) we propose that an ancestor of serpin gene-1 contained a single exon 9 that was duplicated. These may have been the precursors of exons 9A and 9Z. Further duplication events led to exons 9B and 9K and insertion of an additional codon for proline at the carboxyl terminus. From an ancestor of 9K diverged the remaining exons, all with a four-codon insertion at the carboxyl-terminal end of the reactive site loop. The tree then forms two branches, one with exon 9C and another with 9J as the last common ancestor. The most recent duplications involve exons 9D, 9E, 9F, 9G, 9H, and 9I, all near the center of the exon 9 region, with duplication of 9E and 9F as a pair to produce 9G and 9H.
Of course this speculation about evolution is based only on the observable structure of the gene from a single specimen of a contemporary M. sexta. The actual path by which the exon 9 variants arose may actually have been more complex. It is likely that duplication of these exons has occurred through unequal crossing over. At each such event one daughter chromosome would have more copies of exon 9, and the other would have fewer. Furthermore, once multiple copies of the exon were present, several copies could be gained or lost at each instance of unequal crossing over. It seems likely that change in the number of copies of exon 9 in serpin gene-1 is a continuing process and that individuals from different populations could have a serpin gene-1 structure distinct from the one we have characterized. It will be necessary to examine the structure of serpin gene-1 from different M. sexta strains and from other Manduca species to understand more fully the evolution and population genetics of this gene.
Why does M. sexta need so many different proteinase inhibitors? The physiological roles of serpins in insects have not yet been established. We have expressed the serpin-1 variants as recombinant proteins and found proteinase inhibitor activity for 11 of the 12 reactive sites. 3 The individual variants can inhibit enzymes with chymotrypsin-, elastase-, or trypsin-like specificities. Serpins in insect hemolymph might inhibit proteinases produced by pathogenic microorganisms. This role has been postulated for families of serpins in mice (2,4). Another potential function for these inhibitors is regulation of endogenous serine proteinases in hemolymph. Proteinase zymogens are present in plasma, and others may be released from hemocytes in inflammation-like responses. We have cloned cDNAs for five different serine proteinases expressed by M. sexta hemocytes, and we have identified biochemically several different proteinase activities in hemolymph plasma. 4 Some of the serine proteinases in insect hemolymph are involved in a cascade that leads to activation of hemolymph prophenoloxidase (6). Phenoloxidase activation occurs in response to wounding and bacterial or fungal infection. The quinones and melanin resulting from phenoloxidase activity appear to be components of a protective response to microbial infection in insects. Regulating this pathway may be an important role of some of the serpin-1 variants. Thus, the inhibitors produced from M. sexta serpin gene-1 may participate in several physiological functions related to wound healing and antimicrobial defense. It is also quite likely that these serpins inhibit endogenous proteinases involved in processes in insect physiology which have not yet been discovered.