Anisakis larvae were extracted using scissors and tweezers from the body cavity of three female of silver scabbardfish (Lepidopus caudatus) (Mean total length ± SD, 128,6 cm ± 47,25) caught approximately 12 h before from the Adriatic Sea (off San Benedetto del Tronto coast), a fishing area with a known high prevalence of Anisakis infection (Cipriani et al., 2018). After their removal, the larvae were checked for their integrity under a dissecting microscope and the third larval stage was assigned by morphological criteria to Type I larvae (sensu Berland, 1961). Their vitality was evaluated based on their spontaneous movements. Alive and not disrupted larvae were washed in a sterile 1X phosphate-buffered saline solution (PBS, Sigma, St Louis, MO) three times (30 worms/mL) for 1 min each, treated for 1 min with 4% acetic acid (Carlo Erba, Cornaredo, Italy) to inhibit bacterial contamination and rewashed in the sterile PBS for 1 min. Then, the larvae were cultured in filtered sterile PBS (30 larvae/mL/well) with 1% pen-strep in 12 well plates for 24 h, in humified atmosphere at 37°C, 5% CO₂. Three biological replicates were performed.

A representative subsample of 100 Anisakis larvae taken among those cultured, was used for molecular identification. Total genomic DNA from each larva was extracted using the Quick-gDNA Miniprep Kit (ZYMO RESEARCH) following the procedure reported in Irigoitia et al., 2021. The mitochondrial cytochrome c oxidase 2 (mtDNA cox2) gene locus was amplified using the primers 211F (forward; 5′-TTTTCTAGTTATATAGATTGRTTYAT-3′) and 210R (reverse; 5′-CACCAACTCTTAAAATTATC-3′) (Valentini et al., 2006; Mattiucci et al., 2014). The successful PCR products were purified, and Sanger sequenced through an Automated Capillary Electrophoresis Sequencer 3730 DNA Analyzer (Applied Biosystems), using the BigDye^® Terminator v3.1 Cycle Sequencing Kit (Life Technologies). Additionally, a direct genotyping determination of the nuclear metallopeptidase 10 gene locus (nas10 nDNA) was performed by the amplification-refractory mutations system (ARMS) PCR assay at nas10 nDNA by the combined use of OUT-F1 (forward; 5’-TATGGCAAATATTATTATCGTA-3’), OUT-R1 (reverse; 5’-TATTTCCGACAGCAAACAA-3’), INN-F1 (forward; 5’-GCATTGTACACTTCGTATATT-3’), INN-R1 (reverse; 5’-ATTTCTYCAGCAATCGTAAG-3’), following the procedures reported in Palomba et al. (2021b). PCR products were separated by electrophoresis using agarose gel (1.5%) stained with GelRed. The distinct banding patterns were detected using ultraviolet transillumination.

Following the incubation period (24 h, at 37°C, 5% CO₂), L3 were manually removed, and their viability was checked under a stereomicroscope (Leica M205, FCA). The larval culture supernatant was collected, and a protease inhibitor (12,5X/mL) (cOmplete, EDTA-free, Roche) was added, following the standard protocol. Then, EVs were immediately isolated as previously described (Battisti et al., 2017). Briefly, the supernatant was centrifuged twice (4000 rpm, 30 min, 2 times). The cleared supernatant underwent serial ultracentrifugation steps (10,000 g/1 h and 100,000 g/80 min). The pellet was then washed in PBS (100,000 g/80 min). Ultracentrifugation was performed employing Swing 55 rotor and Beckman ultracentrifuge. Finally, the pellet was collected and resuspended in 50 μl PBS and stored at -80°C. Protein concentration, Nanoparticle Tracking analysis (NTA) and proteomic analysis were performed within two weeks from isolation. For Transmission Electron Microscopy, freshly isolated EVs were used. Protein concentration was tested by Bradford assay; purity of the EVs was evaluated as ratio between number of particle and μg of protein (P/μg) (Webber and Clayton, 2013; Théry et al., 2018).

Size determination of the isolated EVs was performed by nanoparticles tracking analysis (NTA) (Dragovic et al., 2011). EVs were thawed on ice and diluted 1:500 in filtered PBS (20 nm filter) and vortexed to achieve the optimal number of EVs/mL ratio. Three videos (30 s each) were recorded for each sample loading, employing the NanoSight NS300 instrument (Malvern Instruments Ltd, Malvern, UK). Measurements were performed employing the NTA 2.3 analytical software. Results were shown as the average of the three recordings.

Transmission Electron Microscopy (TEM) of EVs was performed according to Borrelli et al. (2018). Briefly, freshly isolated EVs were fixed in 2% paraformaldehyde and adsorbed on formvar-carbon-coated copper grids. The grids were then incubated in 1% glutaraldehyde for 5 min, washed with deionized water eight times, and then negatively stained with 2% uranyl oxalate (pH 7.0) for 5 min and methyl cellulose/uranyl for 10 min at 4°C. Excess methyl cellulose/uranyl was blotted off, and the grids were air dried and observed with a TEM (Philips Morgagni 268D) at an accelerating voltage of 80 kV within 48 h from staining. Digital images were taken with Mega View imaging software.

Protein fraction was extracted from the EV preparation (Abramowicz et al., 2018). Briefly, the samples were mixed with acetonitrile to the final concentration of 50% (v/v), and after 45 min of incubation at RT with occasional mixing cycles, acetonitrile was evaporated using a centrifugal vacuum concentrator. Protein concentration was determined by Bradford assay (Biorad). A shotgun proteomic strategy was employed on the protein content of L3 A. pegreffii EVs. Briefly, approximately 7 μg of the sample was mixed with SDS and DTT, boiled, cooled to room temperature, and then alkylated with iodoacetamide in the dark for 30 min. Proteolysis was carried out in an S-Trap filter (ProtiFi; Huntington, NY) following the manufacturer’s procedure. Phosphoric acid (1.2% final concentration) and binding buffer (six volumes) were added. After gentle mixing, the protein solution was loaded to the S-Trap filter, spun at 2000 rpm, and the flow-through was collected and reloaded onto the filter. This step was repeated three times, followed by three times washing with binding buffer. Digestion buffer containing trypsin at 1:10 (w:w) was added into the filter and proteolysis was carried out. The final proteolytic peptide mixture was pooled, lyophilized, resuspended in 0.2% formic acid, and then split into three equal technical replicates, which were then analysed by liquid chromatography-mass spectrometry (LC-MS/MS) using LTQ Orbitrap XL (ThermoScientific, Waltha, MA, USA) coupled to a nanoHPLC system (nanoEasy II, ThermoScientific, Waltha, MA, USA). The three samples were loaded, concentrated, and desalted on a C18 Easy-Column (L = 2 cm, ID = 100 μm; cat. no. 03-052-619, ThermoScientific SC001). Fractionation online with the nanospray ESI source was then achieved on a C18 reverse-phase capillary column (L = 20 cm, ID = 7.5 μm; cat. no. NS-AC-12, NanoSeparations, Niewkoop, Netherlands) at a flow rate of 250 nl/min in a gradient from 5% to 95% of eluent solvent B (eluent B: 0.2% formic acid in 95% acetonitrile; eluent A: 0.2% formic acid and 2% acetonitrile in ultrapure water) over 285 min. The MS/MS acquisition method was set up in a data-dependent acquisition mode, with a full scan ranging from 400 to 1800 m/z range, followed by fragmentation in CID modality of the top 10 ions (MS/MS scan) selected based on intensity and charge state (+2, +3 charges). In the selection, an exclusion time of 40 seconds was applied.

The EVs protein content was profiled through the quantitative proteomics software package MaxQuant (Max Planck Institute of Biochemistry, Martinsried, DE) (Tyanova et al., 2016), employing the Andromeda algorithm against the query database, its reverse decoy database, and a database of common contaminant proteins integrated into the MaxQuant package v. 1.6.0.16. In particular, a protein identity searching process was carried out on the LC-MS/MS spectra, collected from the three replicates, against a customized database achieved by the de novo transcriptome assembly of A. pegreffii L3 (97,480 peptide sequences) (Palomba et al., 2022). The following search parameters were used: trypsin as proteolytic enzyme; 2 as a maximum allowed missed cleavages; carbamidomethyl cysteine as fixed modification; oxidation of methionine and pyroglutamic acid at the peptide N-terminus as variable modifications; 7 as minimum peptide length considered in protein identification; 1% FDR both for peptide spectrum matching and for protein identification. The minimum number of peptides for protein identification was set to 4, with at least 3 unique peptides. Alignment between contiguous HPLC runs was activated. The validation of protein identification was based on the q-value. All further identification and quantification parameters were set as default. Only proteins that were identified by MS/MS analysis in all three replicates were accepted in the final protein list. To assess whether certain classes of proteins were enriched in EV proteome, the gene ontology (GO) analysis and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways analysis for cellular components, molecular functions, and biological processes were carried out using eggNOG-mapper v. 2 (Cantalapiedra et al., 2021) and OMA browser (Altenhoff et al., 2021). InterPro protein family classification and enzyme identification were performed using OmicsBox with the Blast2GO algorithm (Götz et al., 2008). ClusterProfler and AnnotationHub74 were subsequently employed to obtain the enrichment analysis of proteins clusters. The predicted EV proteins were blasted against the known and available EV-associated proteins of nematodes, i.e., Anisakis spp. (Boysen et al., 2020), Ascaris suum (Hansen et al., 2019), Nippostrongylus brasiliensis (Eichenberger et al., 2018a), Brugia malayi (Harischandra et al., 2018) and Trichuris muris (Eichenberger et al., 2018b). The predicted EV proteins were blasted against seven databases. In detail, proteases and protease inhibitors, essential proteins for life, and potential allergic proteins were identified using the BLASTp search against MEROPS (The peptidase database) (Rawlings et al., 2011), DEG (Database of Essential Genes) (Luo et al., 2021) and FARRP (Food Allergy Research and Resource Program) databases, respectively. Potential allergens were also confirmed by the AllerCatPro 2.0 server (Maurer-Stroh et al., 2019). Putative pathogenicity-related proteins were detected using a BLASTp search against the protVirDB (Database of Protozoan Virulent Proteins), VICTORS (Virulence Factors database), and VFDB (Virulence Factors of Pathogenic Bacteria) databases. Host-parasite interactions were predicted using the HPIDB 3.0 (Kumar and Nanduri, 2010; Ammari et al., 2016), which was run with the default setting using EVs searched against human (Homo sapiens, UniProt proteome ID: UP000005640; # of entries 79,052) and bottlenose dolphin (Tursiops truncatus, UniProt proteome ID: UP000245320; # of entries 45,130) proteomes. Analyses were performed on the high-performance computing platforms provided by ELIXIR-IT HPC@CINECA (Castrignanò et al., 2020).

The BLAST analysis of the 100 sequences of Anisakis obtained at the mtDNA cox2 gene locus (~600 bp) retrieved a percentage of identity of 99-100% with the sequences of A. pegreffii previously deposited (KY565564-KY565562). Additionally, the ARMS-PCR analysis enabled us to genotype the same individuals belonging to the species A. pegreffii. Briefly, the use of the nas10 primers generated a specific PCR product of 117 bp, amplifying the C-allele (Palomba et al., 2021b).

Purified EVs, released during the 24h culture of L3 (L3-EVs) in PBS, were characterized by Nanoparticles Tracking Analysis (NTA). Results indicated that L3-EVs had an estimated vesicle size of 65-295 nm and peaked at a mean diameter of 132,3 ± 0.7 nm, which had the prototypical size characteristic of both microvesicles and exosomes (Figure 1A). The concentration was 1,54 x 10¹¹ particles/mL, corresponding 5 x 10⁹ particles/worm and a protein content of 0,104 μg/mL. The P/μg was 1.48 x 10¹². The morphology of the L3-EVs, investigated by TEM, showed that L3-EVs displayed a typical rounded-shaped structure, with lipid bilayer-bound membrane structures, approximately 80-240 nm in diameter (Figure 1B), fully in agreement with the NTA measurements.

The characterization of the protein landscape of the L3-EVs, obtained starting from the de novo transcriptomic data available for A. pegreffii (Palomba et al., 2022) inferred from the analysis of the so far transcriptomes published, allowed. to detect a total of 1083 transcripts. They matched with sequences determined by MS/MS spectra, permitting to identify 153 protein groups. Of those, 5 were unidentified proteins. With respect to the transcriptome used (i.e., 97480 contigs), the percentage of identified proteins was 0.2%. The results of the protein identification search are summarized in the Supplementary File 1.

The proteomic components of L3-EVs were classified by GO annotation according to putative molecular function, biological process, and cellular compartments (Figure 2). Around 95% of the proteins were annotated with GO terms. In the molecular function category, there was a high prevalence of “DNA-binding transcription factor activity” (101 proteins); in the cellular component category the most abundant GO term was “cellular component organization” (98 proteins); finally, in the biological process category, the most frequent GO term was “response to organic substance” (97 proteins). A strong enrichment for 26 terms of the biological process category was found (Supplementary Figure 1). In the molecular category, a strong enrichment was found only for two terms “aminopeptidase activity, carboxypeptidase activity, metalloexopeptidase activity, aldehyde dehydrogenase (NAD+) activity” (Supplementary Figure 1). While no enrichment terms were found for the “cellular component” category. The KEGG analysis revealed the presence of proteins involved in various pathways (25 pathways) (Figure 3). The most frequent was the “metabolic pathway” (57 proteins), followed by “biosynthesis of secondary metabolites” (30 proteins) and “microbial metabolism in diverse environments” (27 proteins). A strong enrichment was observed for 22 pathways (Supplementary Figure 2).

The proteins associated with L3-EVs appeared to belong to distinct families. A total of 224 protein families were detected. Most of them (196 families) were represented by only two and one protein (37 and 159 families, respectively). The most represented are the alpha/beta hydrolase fold family (9 proteins), NAD(P)-binding domain superfamily (8 proteins), concanavalin A-like lectin/glucanase domain superfamily (7 proteins) and thioredoxin-like superfamily (6 proteins) (Figure 4). The most common type of proteins in L3-EVs are enzymes (73,2%). The most abundant class is represented by hydrolases (40), transferases (25), oxidoreductases (20), lyases (8), isomerases (8), translocases (8) and ligases (3) (Figure 5). In particular, the hydrolases class is represented by 3 glycosylases and enzymes acting on peptide bonds (peptidases, 27), ester bonds (4), acid anhydrides (3), ether bonds irata (1), hydrolase (1), and carbon-nitrogen bonds (1).

The protein repertoire associated with L3-EVs, when compared with that previously obtained from unidentified L3 of Anisakis spp. by Boysen et al. (2020), revealed a total of 11 proteins shared by the two data sets, showing a blast similarity ranging from 32.0 up to 100% (Table 1). In particular, they were: the chloride intracellular channel exc-4, the actin 2, the pepsin-I3 domain-containing protein, the p-type domain-containing protein, glutamate dehydrogenase (NAD(P) (+)), heat shock protein 70, 14-3-3 zeta, ras-related protein Rab-11B and three unnamed proteins (Table 1). The three unnamed proteins showed a blast similarity of 96.58%, 95.41% and 78.00% with two histidine acid phosphatases and prostatic acid phosphatase, respectively (Table 1).

The carried proteins of L3-EVs, when compared with those available for other ascaridoid nematodes, i.e. A. suum, N. brasiliensis B. malayi and T. muris revealed that top ten proteins showed a blast similarity ranging from 93.8 up to 100% (Table 2). In particular, the actin 2, galectin, adenylate kinase isoenzyme 1, and the PPlase cyclophilin-type domain-containing protein were also detected in the EVs proteome of A. suum. The actin 2 was also revealed in the EVs proteome of both N. brasiliensis and B. malayi. The ras-related protein (rab-11B) was recognized in the EVs proteome of B. malayi, while the heat shock protein 70 (HSP70) was also in common with N. brasiliensis (Table 2). No EVs proteins of A. pegreffii showed high similarity with EVs of T. muris.

A total of 123 proteins were found to be “essential” for the survival of the parasite by DEG (Database of Essential Genes database). The top ten proteins with the highest similarity (ranging from 97 to 76,16%) against the proteins of this database are listed in Table 3. Similarly, 130 proteases/protease inhibitors were identified in the A. pegreffii EVs, by interpolating data with the peptidase of the MEROPS database. The top 10 proteins with the highest similarity (ranging from 100 to 98,91%), were listed in Table 4. Results were also blasted against the databases reporting potential pathogenic proteins, i.e. protVirDB (Protozoan Virulent Proteins Database), VICTORS (Virulence Factors Database) and (Virulence Factors of Pathogenic Bacteria). Four putative pathogenicity-related proteins with the highest similarity against these databases were identified. Results are reported in Table 5. Interestingly, the heat shock protein 70 was detected both in VICTORS and VIRDB databases. When the FARRP (Food Allergy Research and Resource Program) database was interrogated, a total of 20 proteins (Table 6) were identified as potential allergens. The AllerCatPro predicts that 14 proteins have a high possible known allergenic potential (Table 6).

The interaction network among parasite proteins versus proteins of the definitive (cetaceans) and accidental (human) hosts was investigated. As prototype of a cetacean host, we have considered the bottlenose dolphin (Tursiops truncatus), which represents a natural definitive host of A. pegreffii across its range of distribution (Cipriani et al., 2022). When predicting the interaction between A. pegreffii and the bottlenose dolphin, a total of 27 EVs parasite proteins and 36 dolphin proteins were identified (Figure 6A; Supplementary File 2). L3-EV proteins that showed the highest number of potential interactions with the natural definitive host proteins were the heat shock protein 90 (HSP90) (8 interactions), the rab GDP dissociation inhibitor (rab GDI) (6 interactions), the elongation factor 2 (EF2) (6 interactions). Among the bottlenose dolphin’s proteins, the polyubiquitin-B isoform (UBB) showed potential interactions with the highest number of Anisakis proteins (20 interactions) (Figure 6A; Supplementary File 2).

When predicting the interaction between A. pegreffii and the human host, a total of 27 EV parasite proteins and 29 human proteins were identified (Figure 6B; Supplementary File 2). The L3-EV proteins that showed the highest number of potential interactions with human proteins were the elongation factor 2 (EF2) (6 interactions), the heat shock protein 90 (HSP90) (5 interactions), the rab GDP dissociation inhibitor (rab GDI) (4 interactions), the actin (ACT2) (3 interactions). Among the human proteins, the polyubiquitin-C (UBB) showed potential interactions with the highest number of Anisakis’ proteins (21 interactions) (Figure 6B; Supplementary File 2).