Adult female P. nigriventer spiders laid eggs in captivity. Eggs were separated from silk and stored at −20°C for later use. These animals were collected under license Permanent Zoological Material no. 11024-3-IBAMA and Special Authorization for Access to Genetic Patrimony no. 001/2008.

Fungal and bacterial strains were obtained from various sources. Escherichia coli SBS363 and M. luteus A270 were from the Pasteur Institute (Paris); Candida albicans (MDM8) was from the Department of Microbiology from the University of São Paulo (Brazil); E. coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853, Serratia marcescens ATCC4112, Staphylococcus aureus ATCC 29213, and Staphylococcus epidermidis ATCC 12228 were from the American Type Culture Collection (ATCC). The following human clinical yeast isolates, which can be agents of candidiasis, obtained from the Oswaldo Cruz Institute (Brazil) were also used: Trichosporon sp. IOC 4569, Candida krusei IOC 4559, Candida glabrata IOC 4565, C. albicans IOC 4558, Candida parapsilosis IOC 4564, Candida tropicalis IOC 4560, and Candida guilliermondii IOC 4557.

Purification of antimicrobials was carried out following the strategies described elsewhere (37, 38). In brief, eggs were suspended in 20 mL of glacial acetic acid and homogenized. The insoluble material was removed by centrifugation at 16,000 g for 30 min. The supernatant was partially purified by applying it in two Sep-Pak C₁₈ (Light tC₁₈—Water Associates) cartridges connected in series equilibrated in 0.05% trifluoroacetic acid, and protein-concentrated fractions were eluted in three steps using 5, 40, and 80% of acetonitrile in acidified water. Only the protein-concentrated fraction eluted in 40% acetonitrile was directly used for HPLC purifications. Fractionation was carried out using a reversed-phase high-performance liquid chromatography (RP-HPLC) semipreparative C₁₈ column (Jupiter, 10 × 250 mm), equilibrated in 2% acetonitrile, and 0.05% trifluoroacetic acid. Elution was successfully performed with a linear 2–60% gradient of solution B [0.10% (v/v) trifluoroacetic acid in acetonitrile] in acidified water {solution A [0.05% (v/v) trifluoroacetic acid in water]} run for 60 min at a flow rate of 1.5 mL/min. Effluent absorbance was monitored at 225 nm. Fractions with antimicrobial activity were further purified using an analytical Jupiter C₁₈ column (250 mm × 4.6 mm) at a flow rate of 1.0 mL/min in 60 min with distinct gradients: from 33.5 to 43.5% of acetonitrile in acidified water for fraction containing the HDP of 1.4 kDa (Figure S1A in Supplementary Material), 22.5 to 32.5% for the oligoventin-enriched fraction (Figure 1B), 27.5 to 37.5% for the anti-M. luteus lysozyme (Figure S1C in Supplementary Material), and 34.5 to 44.5% for the anti-C. albicans lysozyme (Figure S1D in Supplementary Material). A symmetrical peak on the HPLC system, amino acid sequencing, and mass spectrometry analysis ascertained the purity of the peptide or protein. Fractions were lyophilized in a SpeedVac Concentrator.

During the purification procedure, antimicrobial activities were detected or monitored by liquid growth inhibition assays as described in Ref. (38–40), using the Gram-negative bacteria E. coli SBS363 and Gram-positive bacteria M. luteus A270 that were cultured in poor broth nutrient medium (PB: 1.0 g peptone in 100 mL of water containing 86 mM NaCl at pH 7.4; 217 mOsM), and the yeast strain C. albicans MDM8, cultured in poor dextrose broth (1/2 PDB: 1.2 g potato dextrose in 100 mL of H₂O at pH 5.0; 79 mOsM) used at half strength. Determination of antimicrobial activity was performed using fivefold micro titer broth dilution assay in 96-well sterile plates at a final volume of 100 mL. Mid-log phase culture was diluted to a final concentration of 1 × 10⁵ colony forming units/mL. Dried fractions were dissolved in 200 µL of ultrapure water and 20 µL applied in to each well and added to 80 µL of the bacterium/yeast dilution. A total of 100 µL of sterile water and PB or PDB were used as quality controls. Tetracycline and/or amphotericin B were also used as controls for growth inhibition. The microtiter plates were incubated for 18 h at 30°C; growth inhibition was determined by measuring absorbance at 595 nm.

Minimal inhibitory concentrations (MICs) were determined using the purified peptide against Gram-negative, Gram-positive, fungal, and yeast strains. MIC determination was performed using a fivefold microtiter broth dilution assay of stock solution, and serial dilution in 96-well sterile plates at a final volume of 100 µL where 20 µL of stock solution was applied in to each well at serial dilution twofold microtiter broth dilution and added to 80 µL of the bacterium/yeast dilution. MIC is defined as the minimal concentration of peptide that caused 100% growth inhibition (33–35, 38, 39). In an attempt to test how rich broth medium affects oligoventin activity, growth inhibition assays were carried out using RPMI-1640 (Sigma-Aldrich) with MOPS 0.165 mol/L (RPMI without bicarbonate 10.4 g/L; MOPS [3-(n-morpholino)propanesulfonic acid] 34.53 g/L) at pHs 7.0 and 5.0 against C. albicans MDM8.

Synergy was measured by checkerboard titration assays using a fivefold microtiter broth dilution assay of stock solution and serial dilution in 96-well sterile plates at a final volume of 100 µL. Oligoventin was diluted along the rows of a microtiter tray and the avian lysozyme was diluted along the columns. A total of 80 µL of M. luteus was added to each well, diluted to a final concentration of 1 × 10⁵ colony forming units/mL. The fractional inhibitory concentration (FIC) was determined after 18 h of incubation of the plates at 30°C. Synergy was defined as an FIC index of 0.5 or less as it represents at least fourfold decrease in the MIC of each compound (41), calculated according to the following formula: FIC index = [A]/MIC_A + [B]/MIC_B, where [A] was the concentration of drug A in a well that represented the lowest inhibitory concentration in its row, MIC_A was the MIC of drug A alone, [B] was the concentration of drug B in a well that represented the lowest inhibitory concentration in its row, and MIC_B was the MIC of drug B alone. Growth inhibition was determined by measuring absorbance at 595 nm. Bioassays were done in triplicate.

The hemolytic activity of oligoventin was tested in duplicate using human erythrocytes. A 2.5% (v/v) suspension of erythrocytes washed in PBS was incubated with oligoventin ranging from 0.4 to 188.8 µM in a 96-well plate for 3 h with intermittent shaking. The absorbance in the supernatant was measured at 415 nm. Hemolysis caused by PBS and 1% (v/v) Triton X-100 were used as 0 and 100% controls, respectively.

The fractions enriched for peptides were spotted (0.5 µL) onto the sample slide, dried on the bench, and crystallized with 0.5 µL of matrix solution [5 mg/mL (w/v) CHCA (α-cyano-4-hydroxycinnamic acid), in 50% acetonitrile and 0.1% TFA] (Sigma). The samples were analyzed on an Ettan MALDI-ToF/Pro spectrometer (Amersham Biosciences) operating in reflectron and positive ion mode. To determine the amino acid sequence of peptides, Edman degradation was performed in a PPSq 21 Automated Protein Sequencer (Shimadzu Co., Japan). Lysozymes were analyzed by SDS polyacrylamide gel electrophoresis (12.5% SDS-PAGE). In-gel lysozymes were destained, dehydrated in 100% acetronitrile for 10 min, and lyophilized. Freeze-dried purified protein was dissolved (1 mg/mL) in denaturant buffer [6 M GdmCl (guanidinium chloride), 0.25 M Tris/HCl, and 1 mM EDTA, pH 8.5]. A total of 20 µL of 2-mercaptoethanol (Sigma) was added to the mixture, followed by vortex-mixing and incubating at 37°C for 2 h. After incubation, 100 µL of 4-vinylpyridine was added to the solution, followed by incubation at room temperature (26°C) for 2 h. The reduction and alkylation of the protein were confirmed by mass spectrometry. Reduced and alkylated proteins were digested with trypsin (Boehringer Mannheim) and tryptic peptides were analyzed by tandem mass spectrometry (MS/MS) in a Q-TOF Ultima API (Micromass) spectrometer operating in positive ion mode. In the mass spectrometer, doubly charged ions of sufficient abundance were selected for MS/MS fragmentation. MS/MS peak list files were submitted to an in-house version of MASCOT server (Matrix Science, USA) and screened against the Uniprot database. Representative resulting spectra and corresponding tables are provided in Table S1 in Supplementary Material.

BLAST searches were done against the Ixodes scapularis IscaW1 genome (42), a well-annotated arachnid genome resource, using oligoventin sequence as the query. Peptide–protein matching was adjusted with the following stringent settings: word size: 2; filter off; e-value 20,000; composition-based statistic off; PAM30 scoring matrix.

The 3D model of the HECT domain was generated using the online server HHPred (43) for template identification and Modeller 9v15 (44) for the model construction. PDB entry 2ONI was used as template (71% similarity). The quality of the final structure was accessed using MolProbity (45) showing just one residue out of the Ramachandran allowed region (Ala 61) and 99.02% of the residues placed on the favorable region. Rate of amino acid evolution among Nedd4 HECT E3s was calculated from 50 homologs using maximum-likelihood phylogenetic analysis and mapped onto the protein structure using default parameters with Consurf (46).

Human proteasome cleavage predictions were simulated with a stringent 0.7 threshold for four sequences (Table S3 in Supplementary Material) using the neural network algorithm Netchop3.1 online server “http://www.cbs.dtu.dk/services/NetChop/.” We used the C-term 3.0 network, which is trained on 1,260 naturally occurring MHC class I ligands (8, 47).

Bayesian phylogenetic inference was carried out with modifications from Ref. (48). Orthologous sequences containing the HECT domain (PF00632) were collected from a large number of eukaryotes. A complete list of Uniprot identifiers was used for the acquisition of 4,392 HECT-containing protein sequences. Redundancy in our dataset was minimized by employing a clustering methodology using the CD-HIT software (49) with a threshold of >90%. One representative protein from each cluster was retrieved for further analysis. This yielded 318 protein sequences relative to 370 eukaryotic species (Table S2 in Supplementary Material), which were aligned with T-coffee (50). The final alignment was manually edited in GeneDoc (51) resulting in 2,470 sites. The choice of the best-fit model of evolution was performed with ProtTest3 (52) using the Akaike Information Criterion, which led us to choose the WAG model. The phylogenetic reconstruction was inferred by the Bayesian inference method implemented in the Beast v1.7.0 software (53, 54). The starting tree was randomly generated, and the proportion of invariable sites and g-distributed rate variation across sites were estimated. The substitution rate categories were set in four categories, and we modeled the molecular clock accordingly to relaxed clock model available (55). The clades were supported by posterior probabilities obtained by Bayesian analysis. For Bayesian method generations, the burn-in was determined in Tracer (54) through log-likelihood scores, and data were summarized in TreeAnnotator (54) after trees that were out of the convergence area had been discarded. A total of 10,000,000 trees were generated, from which 25,000 were burned out of the final tree. The visualization and the final tree edition were performed using FigTree v1.3.1 “http://tree.bio.ed.ac.uk/software/figtree/.” Finally, proteins belonging to each phylogenetic cluster were dissected for revealing the oligoventin-orthologous sequence, from which sequence logos belonging to each clade were generated in Weblogo (56).

The Immune Epitope Database (57) was screened using Nedd4 E3 HECT proteins as queries and yielded nine MHC ligands.

We combined solid-phase purification with assay-guided RP-HPLC runs to isolate host defense effectors from an acidic extract from P. nigriventer non-infected eggs. Four fractions with antimicrobial activity were found (Figure 1A). Mass analysis indicates that they were enriched for innate immune defense effectors, namely, HDPs and lysozymes (Figure S1 in Supplementary Material). To isolate these host defense effectors, analytical RP-HPLC runs coupled with bioassays yielded two fractions with molecules ranging in size from 0.8 to 1.7 kDa (Figures S1A–C in Supplementary Material), which were active against the Gram-positive bacteria M. luteus A270, and two lysozyme-like molecules ranging in size from 14 to 16 kDa (Figures S1D–F in Supplementary Material). A peptide with a mass of 1.4 kDa (Figure S1A in Supplementary Material) was purified to homogeneity, and Edman degradation showed that this peptide was N-terminally blocked. Further investigation by tandem mass spectrometry is needed to sequence this putative HDP. The fraction enriched for oligoventin (Figure 1A) was further purified. N-terminal sequencing by Edman degradation revealed an oligopeptide with eight residues: QPFSLERW, which we named oligoventin. Matrix-assisted laser desorption/ionization—time of flight mass spectrometry (MALDI-ToF-MS) analysis of the resulting fraction shows that the oligoventin molecular weight (MW) observed 1,061.4 Da (M + H⁺ = 1,062.4, Figure 1B) corresponds to the MW calculated (1,061.5 Da).

Fractions enriched for lysozymes (Figures S1D–F in Supplementary Material) were active against the Gram-positive bacteria M. luteus A270 and the yeast C. albicans MDM8, respectively, and were further fractionated by C₁₈ RP-HPLC. Reduction, alkylation, and trypsinization of these fractions followed by comparison of the resulting tryptic peptides by LC-ESI-MS/MS suggest that both of these antimicrobial factors are C-type lysozymes (Table S1 in Supplementary Material). Oligoventin and the putative HDP with 1.4 kDa show similar relative abundances. Altogether, these four antimicrobials represent less than 4% of the total protein content (Figure S2 in Supplementary Material).

Oligoventin presents antimicrobial activity and MICs in the micromolar range (Table 1), markedly against Gram-positive bacteria: M. luteus A270 and the multi-resistant S. aureus ATCC 29213 and S. epidermidis ATCC 12228 strains. It is also active against the yeast C. albicans MDM8 and the Gram-negative bacteria S. marcescens ATCC 4112. However, it has antimicrobial activity in concentrations higher than those of rondonin (35), a cryptic peptide derived from the oxygen-carrier protein hemocyanin (19, 35). Oligoventin is active at concentrations ranging from 47 to 188.9 µM, while rondonin is active from 16.5 to 33.5 µM, except for the Tricosporon sp. IOC 4569 fungi strain, which is 2.1 µM (35). In contrast, oligoventin inhibits growth of Gram-positive, Gram-negative, and yeast strains, thus having a broader spectrum of antimicrobial activity compared to that of rondonin.

C-type lysozymes, classical players of innate immunity, are ubiquitous and constitutively expressed in leukocytes, providing an immediate defensive barrier to invading pathogens (20, 41, 58). Because context-specific co-expression of lysozymes with other host defense elements such as HDPs suggests synergy between these factors (41, 59), we investigated whether oligoventin has a mutually potentiating effect on the G. gallus C-lysozyme, as C-type lysozymes from arachnids are highly similar to avian lysozymes (Figure 1D). We determined the MIC of avian C-lysozyme as 0.01–0.02 µM against M. luteus A270. When both factors were tested together against this strain, we found a 0.37 FIC index, indicating strong synergy (41) between these molecules (Table S2 in Supplementary Material). MIC of oligoventin is potentiated 8-fold by lysozymes, exerting its effect at a concentration as low as 11.81 µM, whereas oligoventin potentiates lysozymes by decreasing its MIC 3.5-fold, causing growth inhibition at 0.006 µM. Moreover, oligoventin exhibits 4.3% hemolytic activity against human erythrocytes at 188.9 µM and 2% at 94.5 µM (Figure S3 in Supplementary Material). However, both oligoventin and lysozymes did not display antimicrobial activity against C. albicans MDM8 when cultured in RPMI-1640 (Sigma-Aldrich) at pH 5.0 or 7.0, even when oligoventin concentration was increased twofold relative to its MIC (376 µM). Similarly, although we determined lysozymes as having a MIC of 1.3–2.6 µM against C. albicans MDM8 in poor dextrose broth medium, this classical HDP also lacked antimicrobial activity in RPMI-1640, even when lysozyme concentration was increased more than 50-fold relative to its MIC (data not shown).

To determine the oligoventin precursor protein, we BLAST screened the blacklegged tick I. scapularis genome IscaW1.4 from Vectorbase (42), a high-quality genomic resource for arachnids. Results showed high homology (88%) between oligoventin and the C-terminal HECT (Homologous to the E6-AP Carboxyl Terminus) domain sequence between residues 920–927 of Nedd4 (neural precursor cell expressed, developmentally downregulated 4-like) E3 ubiquitin ligases (Figure 1C; Figures S4 and S5 in Supplementary Material). Consistently, BLAST screening the Uniprot database with adjustments for peptide–protein matching revealed similar results for orthologous Nedd4 from two arachnids (Ixodes ricinus, Uniprot ID: V5IJJ8, and Rhipicephalus pulchellus, Uniprot IDs: L7ML19, L7ME55) and one crustacean (Daphnia pulex, Uniprot ID: E9GKW9). These results suggest oligoventin is a cryptic peptide released by the proteolysis of a P. nigriventer Nedd4 protein ortholog. Alternative hypotheses of oligoventin precursor proteins are indicated in Table S3 in Supplementary Material.

3D molecular modeling of the blacklegged tick Nedd4 E3 (Uniprot ID: B7Q5Q0) HECT domain (Figure 2A) shows the oligoventin-homologous epitope folding as a β-sheet and its close position to the catalytic cysteine site and the PY motif. PY motifs are internal regulatory motifs that are recognized by WW domains. Notice that the internal PY motif from HECT domains is different from the PPxY motifs found in substrates targeted also by the WW domains of catalytic HECT E3 ligases [reviewed in Ref. (60)] and herein we refer only to the regulatory PY motif from HECT domains.

Surface mapping projection of orthologous sequences onto the 3D model shows high-sequence conservation in the oligoventin-homologous, the catalytic site, and the regulatory PY motif C-terminal sequences (Figures 2B,C). Such high-sequence conservation suggests that the oligoventin-encrypted site is functionally important across different taxa.

We used bioinformatics approaches to verify if oligoventin generation could occur by HECT E3s proteasome-mediated proteolysis. A small dataset (Figure S5 in Supplementary Material) consisting of only the C-terminal Nedd4 HECT domain of ubiquitin ligase sequences for two invertebrates (the arachnid I. scapularis and the crustacean D. pulex) and two vertebrates (mouse and human) were used as input in NetChop3.1, a neural network algorithm trained to predict 26S proteasomal cleavage sites both for constitutive and immunoproteasomes (8, 47). This approach is appropriate because tissue-specific proteasomes, namely, constitutive, immune-, or thymus-specific proteasomes, are structurally rearranged in ways that combine different regulatory and catalytic domains, thus yielding different products (3–6, 8, 9, 47, 61). Therefore, simulating several proteasomes in a single prediction method magnifies the possibility of accurately mapping multiple cleavage sites onto a template protein (8, 47). Figure 2D shows the site positioned within the amino acid residues 920–927 in which oligoventin shows marked homology (Figure 1C) is enriched for cleavage sites in all sequences evaluated. In contrast, the flanking residues lack cleavage sites (Figure S6 and Table S4 in Supplementary Material). These results suggest high conservation of cleavage sites within metazoan E3s. Therefore, our approach indicates that Nedd4 26S proteasome-mediated proteolysis might explain oligoventin production from an E3 ubiquitin ligase precursor protein.

In spite of the HECT domain deep conservation (Figures 2B,C), HECT E3s are pervasive within eukaryotic genomes, comprising more than 33 protein families highly diversified in animals, which have undergone wide architectural rearrangement (1–7). Therefore, our structural conservation data do not inform to what extent the different HECT-containing E3 families may contribute to oligoventin production. To address this issue, we conducted a comprehensive Bayesian phylogenetic analysis on 318 orthologous sequences recovered from all major eukaryotic clades (Table S5 in Supplementary Material), which yielded a tree topology (Figure 3A) consistent with previous findings (2). Our results indicate that Nedd4 proteins are enriched for the oligoventin motif, but not WWP, Itchy, Smurf, and fungal Nedd4 HECT ligases. Figure 3B illustrates sequence logos for each corresponding encrypted site within sampled orthologous sequences. We found that Nedd4s are enriched for a motif composed of Q(P/M/L)F(S/T)(L/I)E(R/K/Q)W. In arthropods positioned near the base of extant ecdysozoans, namely, crustaceans and arachnids, the motif is more pronounced, with two conservative amino acid substitutions, one at the fourth position (S/T) and another at the seventh (K/R). The oligoventin sequence signature is also encrypted to some extent in Nedd4 from insects and vertebrates, with a single non-conservative change at the seventh position in chordates.

Because all nucleated cells from jawed vertebrates present their own antigens derived from cytosolic proteins to cytotoxic T cells through MHC class I (3, 4, 8, 9, 16, 47, 57), we investigated if human and mouse Nedd4 proteins (7, 60) were involved in antigen presentation as a means of comparing the contribution of HECT E3s to the adaptive and innate immune systems. Screening the Immune Epitope Database (57) for Nedd4-derived antigens yielded nine peptides, ranging in size from 8 to 16 residues (MWs between 0.9 and 1.2 kDa), indeed involved in antigen presentation as revealed by MHC ligand assays (Table 2). These results strongly support the hypothesis of convergent evolution of the UPP function in immunity. Therefore, it seems that the UPP was co-opted multiple times in immune systems. Whereas in mammals, Nedd4 E3s play a role in the adaptive immune system as a core member of the UPP (2, 7, 9, 16, 47, 57, 60, 62) and as precursors for MHC class I antigens in mammals (e.g., Table 2), it is possible that a function for E3s has also evolved in the innate defense as precursors of HDPs in a pathway likely dependent on the ubiquitin–proteasome system, at least in arachnids (Figure 4).