|Home | About | Journals | Submit | Contact Us | Français|
Accidents with the caterpillar Lonomia obliqua are often associated with a coagulation disorder and hemorrhagic syndrome in humans. In the present study, we have constructed cDNA libraries from two venomous structures of the caterpillar, namely the tegument and the bristle. High-throughput sequencing and bioinformatics analyses were performed in parallel. Over one thousand cDNAs were obtained and clustered to produce a database of 538 contigs and singletons (clusters) for the tegument library and 368 for the bristle library. We have thus identified dozens of full-length cDNAs coding for proteins with sequence homology to snake venom prothrombin activator, trypsin-like enzymes, blood coagulation factors and prophenoloxidase cascade activators. We also report cDNA coding for cysteine proteases, Group III phospholipase A2, C-type lectins, lipocalins, in addition to protease inhibitors including serpins, Kazal-type inhibitors, cystatins and trypsin inhibitor-like molecules. Antibacterial proteins and housekeeping genes are also described. A significant number of sequences were devoid of database matches, suggesting that their biologic function remains to be defined. We also report the N-terminus of the most abundant proteins present in the bristle, tegument, hemolymph, and "cryosecretion". Thus, we have created a catalog that contains the predicted molecular weight, isoelectric point, accession number, and putative function for each selected molecule from the venomous structures of L. obliqua. The role of these molecules in the coagulation disorder and hemorrhagic syndrome caused by envenomation with this caterpillar is discussed. All sequence information and the Supplemental Data, including Figures and Tables with hyperlinks to FASTA-formatted files for each contig and the best match to the Databases, are available at http://www.ncbi.nih.gov/projects/omes.
Many toxins that affect hemostasis are produced by venomous animals such as snakes and spiders, which use these compounds to facilitate capture and digestion of prey (Markland, 1997; Aird, 2002). Some lepidopteran larvae, notably those of the family Saturniidae, also produce venoms that adversely impact hemostasis. Unlike snake and spider venoms, caterpillar venoms are used solely for defense against predators, which are envenomed by touching the bristles of the caterpillars (Kelen et al., 1995).
Lonomia sp. caterpillars are known for causing a hemorrhagic syndrome characterized by ecchymoses, hematuria, bleeding from scars and mucous membranes, intracerebral bleeding and acute renal failure (Arocha-Piñango and Guerrero, 2001). In southern Brazil, accidents caused by L. obliqua caterpillars are reportedly increasing (Kelen et al., 1995). Accidents usually occur when the victim, leaning against tree trunks containing dozens or hundreds of caterpillars, comes into contact with the bristles of the caterpillars. Often, the whole animal is smashed in the accident. In the latter case, the insect’s cuticle is broken and many secretions, including hemolymph, penetrate the human skin and enter the circulation (Veiga et al., 2001). While some toxic principles are found in bristle extract, others are present in the hemolymph of L. obliqua (Donato et al., 1998; Veiga et al., 2003). Some of the active principles produced by Lonomia sp. that interfere with the hemostatic system have been characterized: fibrinolytic proteases in the hemolymph of L. achelous (Amarant et al., 1991), prothrombin or factor X activators in L. obliqua bristle (Donato et al., 1998) and α-fibrinogenases found in both bristles and in a secretion obtained after freezing L. obliqua caterpillars (Veiga et al., 2003; Pinto et al., 2004). Other active compounds such as phospholipases were also described in Lonomia sp. (for a review, see Arocha-Piñango and Guerrero, 2001).
Remarkably, structural information on Lonomia venom is almost nonexistent. In fact, only partial amino acid sequences of two fibrinolytic proteases from L. achelous have been reported before (Amarant et al., 1991). Furthermore, a GenBank search for "Lonomia", in November, 2004, revealed only the amino acid sequences of these two fibrinogenases and the sequence of polyhedrin, a multiple nucleopolyhedrovirus from L. obliqua. The need for information on the molecular constituents of Lonomia venom led us to create cDNA libraries from the tegument and bristles of L. obliqua followed by high-throughput sequencing and bioinformatics analysis. In addition, Edman degradation of the most abundant protein has been performed in parallel. This approach allowed us to generate a comprehensive catalog of L. obliqua transcripts (cDNAs) and proteins. The roles of L. obliqua molecular components probably involved in the coagulation disorder and in the hemorrhagic syndrome are also discussed.
All water used was of 18-MΩ quality and was produced using a MilliQ apparatus (Millipore, Bedford, MA, USA). Organic compounds were obtained from Sigma Chemical Corporation (St. Louis, MO, USA) or as stated otherwise.
L. obliqua caterpillars were provided by the Health Department of the city of Videira (Santa Catarina, Brazil) after being collected by local inhabitants directly from trees. Bristles were collected from caterpillars by excision at the base of the scoli. Tegument, which included bristles and surrounding tissues, was collected after total dissection of the animals. Both tissues were homogenized in 500 µl of water for preparation of the respective extracts. Hemolymph was collected with a syringe through an incision at the pseudopodia of the caterpillar. Cryosecretion was obtained as described in Pinto et al. (2004). Briefly, caterpillars were placed over Petri dishes and kept at −20°C for 24 h. Frozen caterpillars were then washed with 20 mM HEPES (pH 7.4), and the released protein-rich fluid was collected from the dish. All venomous preparations were then centrifuged to remove debris (12,000 × g for 20 min) and stored at −80°C until use.
Analysis of the protein components of each L. obliqua venomous preparations (bristle, tegument, hemolymph, and cryosecretion) was performed as in Francischetti et al. (2004) and is described in detail in the Supplemental Data (Section 2.7).
Two cDNA libraries were constructed: one using mRNA isolated from L. obliqua bristles and the other from L. obliqua tegument. mRNA was obtained using a Micro-Fast Track mRNA isolation kit (Invitrogen, San Diego, CA, USA) according to manufacturer's instructions. The PCR-based cDNA libraries were constructed using the SMART cDNA library construction kit (Clontech, Palo Alto, CA, USA). Cycle-sequencing reactions used the DTCS labeling kit (Beckman Coulter Inc., Fullerton, CA, USA); nucleotide sequencing was performed in a CEQ 2000 DNA instrument (Beckman Coulter Inc.) as described in Francischetti et al. (2004) and in the Supplemental Data (Sections 2.8 and 2.9).
cDNA sequences analyses and obtaining of full-length sequences were performed as described in detail in Francischetti et al. (2004) and in the Supplemental Data (Sections 2.10 – 2.13). Electronic version of the complete tables (Microsoft Excel format) with hyperlinks to web-based databases and to BLAST results is available at http://www.ncbi.nih.gov/projects/omes.
Statistical tests were performed with SigmaStat version 2.0 (Jandel Software, San Rafael, CA, USA). Kruskal-Wallis ANOVA on ranks was performed, and multiple comparisons were done by the Dunn method. Dual comparisons were made with the Mann-Whitney rank sum test.
Studies on the venom of Lonomia sp. caterpillars have so far characterized two fibrinolytic proteins from L. achelous, named Achelase I and Achelase II (Amarant et al., 1991). The present study attempts to understand the molecular components likely involved in the envenomation by L. obliqua. Accordingly, we have performed SDS-PAGE and constructed cDNA libraries using tissues potentially involved in the envenomation process.
Because accidents with L. obliqua are usually accompanied by smashing of the whole animal, we investigated four types of venomous preparations in the present study: bristle and tegument extracts, hemolymph, and cryosecretion. Figure 1 (left) shows the pattern of protein separation from these preparations applied side-by-side in the same gel followed by staining with Coomassie blue. To identify these proteins, they were transferred to PVDF membrane, and the bands were cut from the membrane and submitted to Edman degradation. Amino-terminal information was successfully obtained for 25 bands. Sequences were compared with the non-redundant (NR) GenBank database and with the cDNAs obtained in the mass sequencing project of L. obliqua (see Section 2). Results are presented in Figure 1 (right). We found many sequences displaying matches to cDNAs from both libraries, coding for diverse proteins such as trypsin, serpin, lipocalin, transferrin, lectin, and others.
Band 19-LOB is the most prominent band in the bristle extract and other samples, and was identified as a lipocalin. A trypsin-like protein (11-LOB) and a lectin (1-LOB) were also identified in the bristle preparation. Other bands that were successfully sequenced in the tegument extract have similarities primarily to proteins involved in cell metabolism or cell structure (housekeeping function). On the other hand, the hemolymph contains a serpin (8-LOH) and a protease inhibitor (23-LOH), among other proteins. The cryosecretion also contains a serpin (8-LOC) and other housekeeping proteins. Finally, in some cases, the N-terminal sequence from a given band matches more than one cDNA; for example, 9-LOB can be assigned to cDNAs coding for hemolin and laminin in the bristle library, while band 16-LOB can be assigned to cDNAs coding for lectin and for a ribosomal protein in the tegument library.
Two different cDNA libraries were constructed: one from a section of the tegument of L. obliqua, which includes the bristle, surrounding tissues, and chitinous processes, and a second from homogenized bristle only. A total of 1,152 independent clones from the tegument library and 960 independent clones from the bristle library were 5’ sequenced, yielding 938 and 730 sequences, respectively, of good quality for analysis. Next, these sequences were followed by bioinformatics analysis as described in Francischetti et al. (2004) and in the Supplemental Data, and included: i) clustering at high stringency levels (98% identity over 100 nucleotides), ii) BLAST search against the non-redundant and protein motifs databases, iii) comparison with the PFAM, SMART and Kog databases using RPSBlast and iv) submission of the translated sequences to the Signal P server (Nielsen et al., 1997) (see Section 2). This initial approach allowed us to obtain a fingerprint of the protein families or "clusters" present in these tissues. Several sequences were then selected based on novelty or the protein family it assigns for and extension of their corresponding cDNAs were performed until the poly A was reached.
In the cDNA library of L. obliqua tegument, 592 sequences (63.1%) showed database hits and were organized into 278 clusters, while the other 346 sequences showed no similarity to any known sequences and formed 260 clusters (complete supplemental table available for download as Excel spreadsheet). As far as the absolute number of sequences is concerned, we found that 16.1% have a potential signal peptide, 62.6% code for housekeeping proteins, and 21.3% are indeterminate. After clustering the sequences, 12% of the clusters were found to have a potential signal peptide, 58% code for housekeeping proteins, and 30% are indeterminate. Moreover, 477 sequences matched database sequences obtained from a variety of insects, including the lepidopterans Bombyx mori, Manduca sexta, Heliothis virescens, Antheraea pernyi, Galleria mellonella and others, as well as to the mosquito Anopheles gambiae and to the fruit fly Drosophila spp..
In the cDNA library of L. obliqua bristles, 408 sequences (55.9%) showed database hits and were organized into 176 clusters, while the other 322 unknown sequences formed 192 clusters (complete supplemental table available for download as Excel spreadsheet). Forty percent of all sequences code for proteins with a potential signal peptide, 43.3% for housekeeping proteins, and 16.7% are indeterminate. Considering the number of clusters, 17% had a potential signal peptide, 53% code for housekeeping proteins, and 30% were indeterminate. A total of 311 sequences had hits to sequences from such insects as M. sexta, B. mori, Samia cynthia, Spodoptera frugiperda, and Apis mellifera.
Comparing these results with those from other studies describing transcriptomes of venomous or blood-sucking animals (Valenzuela et al., 2002; Francischetti et al., 2002; Francischetti et al., 2004), the L. obliqua display fewer secreted peptides, especially in the tegument. This can be attributed to the fact that those studies are based on cDNA libraries constructed with mRNAs extracted from venom/salivary glands, which are rich in transcripts and proteins with specific roles in envenomation/blood-feeding. An estimate of the percentage of sequences and clusters of putative toxins for both bristle and tegument is shown in Figure 2, a–d. Concerning the tegument library, the most abundant sequences are those coding for serpins (25.8%), followed by serine proteases (16.1%) and for lipocalin (16.1%) (Fig. 2a). A similar distribution was found when analyzing the percentage of clusters (Fig. 2b).
In the bristle library, over 50% of cDNAs that were sequenced code for a lipocalin, followed by kininogen (16.5%) and serine proteases (14.7%) (Fig. 2c). However, the number of clusters found for serine protease is the most abundant (25%), followed by lectins (15%) and serpins (10%). Such a distribution of transcripts in the bristle library reveals that sequences from the same family (e.g. serine proteases) and found in low copy number may display significant differences in their primary sequence. Therefore, these sequences were independently clustered (Fig. 2d) suggesting functional diversity in the same protein family. On the other hand, lipocalin is the most abundant cDNA in the bristle, but forming only one cluster, indicative of high similarity in their primary sequences. Finally, in contrast to other venoms, no metalloproteases or disintegrins (Fox et al., 2002) were found in our libraries.
Both libraries also contained sequences coding for housekeeping proteins apparently devoid of toxic properties. Accordingly, we report cDNAs coding for proteins involved in transcription and translation (elongation factors, ribosomal proteins), metabolism (juvenile hormone metabolism and transport, NADH-dehydrogenase, cytochrome c oxidase, heme oxygenase, ATP synthase, glutathione S-transferase, aldehyde dehydrogenase, glyceraldehyde 3-phosphate dehydrogenase), protein processing (chaperones, proteasome), and other functions. Sequences related to cell structure, such as cuticle proteins, myosin, actin, tubulin and tropomyosin, were found primarily in the cDNA library of tegument.
In an attempt to maximize the identification of putative toxins in L. obliqua tissues, we constructed separate cDNA libraries from both bristle and tegument mRNAs. Table 1 presents an overview of sequences coding for secreted proteins that are potentially associated with toxic functions, obtained after the first round of sequence and bioinformatics analysis. The number of sequences found in a given cluster is also available as an approximate estimate of cluster representation and complexity. Matches to the NR and protein motifs databases with their corresponding e value are also reported. Table 1 also shows, when available, the corresponding N-terminus obtained by Edman degradation of specific proteins.
In the tegument cDNA library (Table 1), we found 34 sequences forming 21 clusters coding for putative toxins, including protease inhibitors, trypsin-like serine proteases, lectins, esterases, and others. Among the putative secreted proteins, cluster 1, with 36 sequences, codes for a cuticule potein. Other clusters are less represented and code for proteins with sequence homology to ferritin and attacin, as well as to unknown proteins.
In the bristle cDNA library, 116 sequences may be involved in the envenomation process, and they were grouped in 23 clusters. Among these, the most abundant is cluster 1, coding for a lipocalin with 54 sequences, followed by cluster 8, which codes for a kininogen precursor. Other clusters code for proteins of diverse families, including serine proteases, lectins, serpins, cystatins, trypsin-like inhibitors, and cysteine-rich molecules. Among the putative secreted proteins, we found sequences coding for cecropin, chemosensory protein, and others.
Table 1 also describes clusters that contain sequences where an ORF with or without signal peptide could be identified, but for which there were no databases hits. Accordingly, clusters 330, 349, 387, and others in the tegument library assign for proteins of unknown function. Some of these sequences were identified by SDS-PAGE from both the bristle and tegument libraries, demonstrating that these cDNAs code for actual proteins (see Fig. 1 and Section 3.1). The spreadsheet containing the complete list of the sequences coding for proteins with secretory, housekeeping, or indeterminate functions, with or without database hits, can be downloaded at http://www.ncbi.nlm.nih.gov/projects/omes/ and extracted to a personal computer for viewing. The spreadsheets also include i) hyperlinks to the best match of the NR database, ii) links to NR matches found for the cluster, iii) matches to the conserved domain database, iv) FASTA-formatted files for each cluster, and v) CAP assembled alignments of each cluster having two or more sequences.
To maximize the information for the proteins presented in Table 1, selected molecules were re-sequenced and extended to obtain, when applicable and possible, their full-length cDNAs. The full coding sequences were then searched against the NR protein database and SignalP server to confirm, respectively, sequence similarity and the presence of a signal peptide. When a signal peptide was predicted to exist, the molecular weight and the isoelectric point of the mature protein were also calculated and the putative function annotated. The summary of our findings is presented in Table 2, which also includes the accession numbers in the NCBI databases, the length of the sequences (partial or full-length), the tissue source of the transcript (bristle or tegument), and the putative function. A brief discussion of selected sequences from tegument and bristle and their possible participation on the envenomation by L. obliqua is presented below.
Serine proteases can be found in microorganisms, plants and animals, being involved in many physiological processes, and can be classified as trypsin-like, subtilisin-like, and carboxypeptidase-like serine proteinases. Human serine proteinases are involved in hemostasis, digestion, cellular differentiation, and many other processes; those involved in hemostasis are typically trypsin-like proteases. In insects, various serine proteases participate in hemolymph coagulation and prophenoloxidase-activating cascades in response to wounding and infection (Kanost et al., 2004). Both processes involve serine protease activation of zymogens and culminate in formation of the coagulum and phenoloxidase activation, respectively. These enzymes display high sequence and structural similarity and have evolutionarily diverged into two main groups. One group, whose members include prothrombin, FIX, X, FVII, and protein C, characteristically contains a Gla domain, an EGF motif, and eventually a Kringle domain. The other group is represented by prekallikrein, FXI, plasminogen, urokinase, T-PA, FXII, and trypsin. These proteins have no Gla domain but have selected other modules such as Apple, Kringle, EGF, and fibronectin-like domains (Larson and Katherine, 1998). Notably, our bristle library contains full-length sequences matching both groups of serine proteases, which are briefly discussed below.
LOqua-SP1 (cluster 220, tegument library). This protein is homologous to a serine protease involved in the prophenoloxidase cascade in M. sexta (gi|39655053) as well as to human coagulation factors IXa (gi|67596), VII (gi|10518503), II (gi|4503635), and Xa (gi|538554) (Fig. 3). It is also similar to trocarin (gi|54402099), a prothrombin activator from Tropidechis carinatus (Joseph et al., 1999). Of interest, unlike the coagulation factors, LOqua-SP1 lacks a Gla domain and the EGF-1 and EGF-2 motifs but contains a conserved serine proteinase domain. In addition, the molecular mass of mature LOqua-SP1 (55.2 kDa) is very similar to those of coagulation factors IX (56 kDa), VII (50 kDa), and X (58.8 kDa). A cladogram for LOqua-SP1 and these serine proteases is shown in Figure 3b, available for download and in the Supplemental Data. Consistent with a pro-hemostatic function for these sequences, previous studies have described a prothrombin activator from bristle extract of L. obliqua (Donato et al., 1998). Cloning and expression of LOqua-SP1 may help to elucidate its substrate specificity and role in envenomation.
LOqua-SP6 (cluster 12, bristle library). This serine protease of ~30 kDa is homologous to serine proteases from the mosquito A. gambiae (gi|31238311) and from the flea Ctenocephalides felis (gi|4530052). It is also similar to a plasminogen activator from the centipede Scolopendra subspinipes (gi|4098568) (Fig. 8, available for download and in the Supplemental Data). Consistent with these sequences, α-fibrinogenase activity was recently reported in L. obliqua crude bristle extract (Veiga et al., 2003). This protein, with an apparent molecular mass of 35 kDa, was purified from the cryosecretion (Pinto et al., 2004). LOqua-SP6 is also similar to a venom proteinase from the bumblebee Bombus pennsylvanicus (gi|1079195) and to human plasma kallikrein (gi|4504877). At present, the specificity of this enzyme is unknown, but an α-fibrinogenase (crotalase) displaying kallikrein-like activity has been described in C. adamanteus venom (Markland et al., 1982).
Cysteine proteinases are endopeptidases that rely on the presence of a cysteinyl thiol group for their catalytic activity. Most of the well characterized cysteine proteinases of the papain family, which includes the cathepsins, have the active site motif QGCGSCWAF (Chapman et al., 1997). Sequences in the cDNA libraries of bristles and tegument of L. obliqua showed homology to cathepsins.
LOqua-CysPep2 (cluster 55, bristle library). This is a 34.7-kDa cathepsin-like cysteine proteinase containing the motif QGSCGSCWAF and homologous to cathepsins from H Helicoverpa armigera (gi|7537454), Aedes aegypti (gi|5031250), Triatoma infestans (gi|38147393), and Araneus ventricosus (gi|31872149) (Fig. 9, available for download and in the Supplemental Data). While in some hematophagous arthropods cathepsin-like enzymes have an important role in yolk degradation during embryogenesis (Sappington and Raikhel, 1998), in others, such as Rhodnius prolixus, they have a role in digestion of the blood meal (Lopez-Ordoñez et al., 2001). Considering that the present study is on a larval phase of L. obliqua, it is not possible to assign an oogenic function for a cathepsin-like protease. On the contrary, these enzymes may be somehow involved in envenomation or in cellular lysosomal function.
Insects and other arthropods are a rich source of protease inhibitors such as serine protease inhibitors (serpins), Kunitz-type, and Kazal-type inhibitors (Kanost et al., 2004). These protease inhibitors found in arthropod hemolymph are likely to function in protecting their hosts from infection by pathogens or parasites and in regulating physiological processes. Some may inhibit fungal or bacterial proteinases, while others probably have roles in regulating endogenous proteinases involved in coagulation, prophenoloxidase activation, or cytokine activation (Kanost, 1999). Further, it has been shown that some Kazal-type inhibitors found in arthropods play a role in blood-feeding processes by inhibiting thrombin and other coagulation factors (Friedrich et al., 1993; Campos et al., 2004). Here we report the full-length sequences for eight protease inhibitors, including Kazal-type, trypsin inhibitor-like, cysteine protease inhibitors (cystatin), and serpins (Table 2).
LOqua-Serp2 (cluster 131, tegument library). This is a 41.6-kDa secreted serpin similar to other members of this family. Alignments are shown in Figure 4 and the phylogenetic tree is shown as Figure 4b, available for download and in the Supplemental Data. LOqua-Serp2 is homologous to serpin-2 from B. mori (gi|7341330) and to serpins from two hematophagous insects, the cat flea C. felis (gi|25527494) and the mosquito A. gambiae (gi|6759386). Of interest, this serpin also showed homology to bovine plasminogen activator inhibitor (PAI-1, gi|27806497), to antithrombin from zebrafish (AT, gi|33504509) and human (ATIII, gi|113936), and to serpins from the ticks R. appendiculatus (gi|17223662) and Ixodes ricinus (gi|14140097).
LOqua-Serp3 (cluster 42, tegument library). This serpin showed homology to a coagulation inhibitor from horseshoe crab (gi|31872149), to ATIII (gi|113936), and to PAI-1 (gi|27806497), as well as to a serpin from the tick Haemaphysalis longicornis (gi|42662201) (alignments not shown). Consistent with the sequence in cluster 42 of the tegument library, we found by Edman degradation the sequence TETDLQRILRdgNDqFtaKM for bands 8-LOB, 8-LOH, and 8-LOC. The sequence FHYFGHEYDRTALGDAIdKA was also obtained for band 8-LOH and matched the sequence of cluster 247 of the same library, which also codes for a serpin (LOqua-Serp6) (Fig. 1).
Considering the participation of serpins in many physiological processes in arthropods, such as hemolymph coagulation and immunity, it is possible that one or more L. obliqua serpins may affect human blood coagulation by inhibiting coagulation factors. However, the function and specificity of L. obliqua serpins and their participation in envenomation remain to be determined.
LOqua-Cyst1 (cluster 125, tegument library). This sequence displayed homology to cysteine proteinase inhibitors, mostly of the cystatin family, including a cystatin found in hemocytes of the horseshoe crab Tachypleus tridentatus (gi|47115697) and to cystatins from the puff adders Bitis arietans (gi|118194) and B. gabonica (gi|38570038), as well as to murine kininogen (gi|12963497) (alignments not shown).
LOqua-PI3 (cluster 374, tegument library). This is a secreted trypsin inhibitor-like protein with homology to an immune-induced metalloproteinase inhibitor from the moth G. mellonella (gi|34485736) and also similar to an anticoagulant from the venom of the scorpion Buthus martensii (gi|14388599) (Fig. 10, available for download and in the Supplemental Data); however, while the Lonomia full-length sequence has 398 amino acids, the metalloproteinase inhibitor and the scorpion toxins consist of 170 and 89 residues, respectively. The function of this trypsin inhibitor-like molecule deserves further study.
Phospholipases are almost universally present in animal venoms. PLA2 promotes hydrolysis of phospholipids with generation of free fatty acids and lysophospholipids, thus producing indirect hemolysis. PLA2 may inhibit blood coagulation by direct interaction with coagulation factors or through degradation of phospholipids involved in coagulation complexes, in addition to antiplatelet effects, neurotoxicity, cardiotoxicity, and myotoxicity (Kini, 2003).
Of note, L. obliqua bristle extract induces in vitro hemolysis on rat and human erythrocytes as well as intravascular hemolysis in rats, probably due to a PLA2 (Seibert et al., 2004). Both cDNA libraries contain sequences homologous to PLA2 and other proteins that may have hemolytic activity. Of interest, the full-length sequence of cluster 154 of the bristle library (LOqua-PLA1) did not display homology to any snake Group I and Group II PLA2 known to cause coagulation disturbances and myotoxicity (Lambeau and Lazdunski, 1999). In fact, no Asp-49 or Lys-49 PLA2 were found in our libraries. On the other hand, as depicted in Figure 5, this sequence matched Group III PLA2s from the insects D. melanogaster (gi|21627771) and A. mellifera (gi|16904372), from the venom of the gila monster Heloderma suspectum (gi|104338), and from the scorpions Anuroctonus phaiodactylus (gi|46484897) and Pandinus imperator (gi|37079101), as well as a Group IX PLA2 toxin from the marine medusa Rhopilema nomadica (gi|999133). Furthermore, LOqua-PLA1 also displays homology to the carboxy domain of human Group III PLA2 (gi|7657126) whose function is related to arachidonate release from cells (Murakami et al., 2003). At present, the substrate specificity and function for this putative secreted PLA2 is a matter of speculation, but it may be somehow involved in the hemolysis observed after envenomation (Seibert et al., 2004).
The lipocalin family is a large group of proteins that exhibit great structural and functional variation. They are typically small extracellular proteins (~20 kDa) sharing common molecular recognition properties: the binding of small, mainly hydrophobic molecules; binding to specific cell-surface receptors; and formation of covalent and non-covalent complexes with other soluble macromolecules (Åkerstrom et al., 2000). In arthropods, the most studied proteins of this family are the lobster colorant crustacyanin and the colorant bilin-binding protein and insecticyanin from butterfly larvae. On the other hand, lipocalins with antihemostatic functions have been isolated from the saliva of blood-sucking arthropods. These proteins are known to inhibit platelet aggregation and reduce blood coagulation in addition to reducing inflammation and promoting vasodilatation (Ribeiro and Francischetti, 2003; Ribeiro et al., 2004).
Both cDNA libraries of L. obliqua contain sequences with homology to lipocalins. The tegument library has one cluster formed by five sequences, while the bristle library has three clusters, two of them containing only one sequence and the third, LOqua-Lipcl1, formed by 54 sequences. The function of this lipocalin may be related to its ligand-binding property.
Lectins are molecules that usually bind sugars. We found a total of ten sequences with homology to lectins, including c-type lectins, which have a typical carbohydrate recognition domain (Harrison et al., 2003). These sequences formed clusters 120 and 129 in the tegument library, and clusters 33, 140, and 184 in the bristle library (Table 1). We obtained the full-length sequences of all except cluster 129 of the tegument library; however, matching this truncated cluster we found, by Edman degradation, the sequence TVLKDLAK for band 16-LOB. Below is a brief discussion on the major L. obliqua lectins.
LOqua-Lect5 (cluster 33, bristle library). Figure 6 shows that this protein has homology to lectins found in snake venoms. Typically, they are 30-kDa proteins consisting of the covalent association of two identical or homologous polypeptides, and they play significant roles in envenomation by binding to coagulation factors and to platelet receptors, as well as through fibrinogen clotting inhibition (Braud et al., 2000). Considering the predicted molecular weight of mature LOqua-Lect5 (16.3 kDa), we suggest that it is a subunit of a dimeric lectin.
LOqua-Lect1 and LOqua-Lect3 (clusters 120, tegument library, and 140, bristle library, respectively). These two lectins, with molecular masses of 33.6 and 33.9 kDa, respectively, displayed homology to a lipopolysaccharide-binding protein from B. mori (gi|4469126), H. sapiens chondroitin sulfate proteoglycan 2, also known as versican (gi|21361116), and to a lectin that appears to be involved in the envenomation by the fish Thalassophryne nattereri (gi|51863395) (Lopes-Ferreira et al., 1998) (Fig. 11, available for download and in the Supplemental Data).
Bacterial infection in lepidopteran insects induces production of antibacterial factors in the hemolymph such as attacin and cecropin (Kanost et al., 2004). In the present study, we found three sequences with homology to attacin, one in the tegument library (LOqua-Def2, truncated) and the other two forming one cluster in the bristle library (LOqua-Def3). Furthermore, we found five sequences in the bristle library forming three clusters with homology to cecropin, one of which had its full-length sequence determined (LOqua-Def4). Even though these peptides may have an immunity function in Lonomia, the possibility exists that they may also affect vertebrate physiology.
Full-length sequences of other proteins, secreted or not, were obtained, such as cuticle proteins, sensory proteins, proteins related to cell metabolism, and ribosomal proteins, among others. Results for these proteins are in Table 3, available for download and in the Supplemental Data.
The primary symptoms characterizing accidental contact with Lonomia obliqua caterpillars are severe hemorrhage and acute renal failure (Kelen et al., 1995). The coagulation disorder presented by these patients appears to be related to the action of many caterpillar toxins acting in a redundant manner on the hemostatic system. Accordingly, molecules that activate factor X and prothrombin (Donato et al., 1998) and likely coded by sequences such as LOqua-SP1 and LOqua-SP6, among others, may play an important role in envenomation. The action of these enzymes leads ultimately to thrombin formation that, among other effects, induces platelet aggregation and fibrin formation. Thrombin also induces compensatory fibrinolysis by a mechanism involving release of tissue plasminogen activator (t-PA) by endothelial cells. Plasmin, thus formed, acts on FXIIIa-crosslinked fibrin and generates D-dimers (Williams, 1998).
D-dimers are found at remarkably high levels in the blood of caterpillar-envenomed patients (Zannin et al., 2003). However, these levels appear to be disproportionate to the activation of the blood coagulation cascade. Actually, the laboratorial profile of these patients resembles "primary fibrinolysis", a term that has been applied to cases of atypical disseminated intravascular coagulation in which the laboratorial and clinical manifestations are dominated by the effects of fibrinolysis (Williams, 1998). These cases, as reported for L. obliqua envenomation (Zannin et al., 2003) or after infusion of snake venom α-fibrinogenases such as Ancrod (Bell, 1997; Markland, 1997), are characterized by a marked hypofibrinogenemia and very high titers D-dimers, but normal FII and FX levels, and platelet counts; therefore, a consumption coagulopathy did not take place. Of note, Ancrod induces a thrombin-independent, FXIII proenzyme-dependent formation of crosslinked fibrin oligomers that act as cofactors in the activation of plasminogen by t-PA, with plasmin generation and appearance of D-dimers (Dempfle et al., 2000; Dempfle et al., 2001). In this regard, L. obliqua cryosecretion, hemolymph (Pinto et al., 2004) and, to a lesser extent the bristle extract (Veiga et al., 2003), present marked fibrinogenolytic activity, attributed to an enzyme named lonofibrase, which degrades the α-chain of fibrinogen (Pinto et al., 2004). We speculate that, in addition to prothrombin activator and other enzymes, lonofibrase may also contribute to the consumption of fibrinogen by directly degrading the molecule and indirectly, and perhaps most critically, through activation of the fibrinolytic system, as described for Ancrod (Dempfle et al., 2000; Dempfle et al., 2001). Molecular identification, cloning, and expression of lonofibrase, in addition to other molecules, will be an important step in our understanding of the envenomation process. A diagram containing the potential targets for some L. obliqua toxins is presented in Figure 7.
Finally, manipulation of the fibrinolytic system appears to be the rule rather than the exception regarding accidents with Lonomia sp. caterpillars. In fact, L. achelous venom contains enzymes that affect fibrinolysis in a distinct manner when compared with L. obliqua: a urokinase-like protease that degrades FXIII, and two plasmin-like activities (Arocha-Piñango and Guerrero, 2001). Other components such as PLA2, serpins, and lectins, in addition to the host humoral response to envenomation, may also participate in the symptoms manifested by most patients, including hemolysis, bleeding, and acute renal failure (Kellen et al., 1995). It appears that our attempt to generate a catalog containing the putative toxic proteins of the caterpillar is an effective approach to generate testable hypotheses on the molecular basis of envenomation. It may help to find candidate proteins for development of a reliable diagnostic kit for this envenomation and for the improvement of serum production (Rocha-Campos et al., 2001; Theakston et al., 2003).
We thank Drs. Thomas E. Wellems, Robert W. Gwadz, and Thomas J. Kindt (NIAID/NIH) for encouragement and support and the Brazilian agency CAPES (Ministry of Science and Technology Department) for the fellowship to A.B.G. Veiga. We thank Van My Phan (LMVR/NIAID) for technical assistance and Dr. Mark K. Garfield (Research Technologies Branch/NIAID) for protein sequencing. We acknowledge Brenda Rae Marshall (NIAID) for editorial assistance. We express special thanks to the Health Department of the city of Videira (Santa Catarina, Brazil) for providing L. obliqua caterpillars. The authors are grateful to both reviewers for their time, comments and valuable suggestions in the manuscript.