The Proteome and Citrullinome of Hippoglossus hippoglossus Extracellular Vesicles—Novel Insights into Roles of the Serum Secretome in Immune, Gene Regulatory and Metabolic Pathways

Extracellular vesicles (EVs) are lipid bilayer vesicles which are released from cells and play multifaceted roles in cellular communication in health and disease. EVs can be isolated from various body fluids, including serum and plasma, and are usable biomarkers as they can inform health status. Studies on EVs are an emerging research field in teleost fish, with accumulating evidence for important functions in immunity and homeostasis, but remain to be characterised in most fish species, including halibut. Protein deimination is a post-translational modification caused by a conserved family of enzymes, named peptidylarginine deiminases (PADs), and results in changes in protein folding and function via conversion of arginine to citrulline in target proteins. Protein deimination has been recently described in halibut ontogeny and halibut serum. Neither EV profiles, nor total protein or deiminated protein EV cargos have yet been assessed in halibut and are reported in the current study. Halibut serum EVs showed a poly-dispersed population in the size range of 50–600 nm, with modal size of EVs falling at 138 nm, and morphology was further confirmed by transmission electron microscopy. The assessment of EV total protein cargo revealed 124 protein hits and 37 deiminated protein hits, whereof 15 hits were particularly identified in deiminated form only. Protein interaction network analysis showed that deimination hits are involved in a range of gene regulatory, immune, metabolic and developmental processes. The same was found for total EV protein cargo, although a far wider range of pathways was found than for deimination hits only. The expression of complement component C3 and C4, as well as pentraxin-like protein, which were identified by proteomic analysis, was further verified in EVs by western blotting. This showed that C3 is exported in EVs at higher levels than C4 and deiminated C3 was furthermore confirmed to be at high levels in the deimination-enriched EV fractions, while, in comparison, C4 showed very low detection in deimination-enriched EV fractions. Pentraxin was exported in EVs, but not detected in the deimination-enriched fractions. Our findings provide novel insights into EV-mediated communication in halibut serum, via transport of protein cargo, including post-translationally deiminated proteins.


Introduction
Halibut is a teleost flatfish which belongs to the order Heteresomata (Pleuronectiformes). It is one of the largest teleost fish and endangered due to previous overfishing and

The Proteome and Citrullinome of Halibut Serum EVs
Total protein content, as well as F95 enriched protein content, representative of deiminated protein cargo in EVs (the "EV-citrullinome"), was identified by LC-MS/MS analysis. A range of proteins relating to innate and adaptive immunity, as well as gene regulation and cellular function, were identified as deiminated in EV cargo, and are listed in Table 1 (for full details on LC-MS/MS analysis, see Supplementary Table S1). Total EV protein cargo analysis revealed proteins relating to innate and adaptive immunity, nuclear proteins relating to gene regulation, proteins relating to cellular function and metabolism and are listed in Table 2 (for full details on LC-MS/MS analysis, see Supplementary Table S2). Total serum-EV proteins stained by silver staining are shown in Figure 2A, F95 enriched proteins from serum-EVs are shown in Figure 2B and the number of total EV proteins identified, overlapping with deiminated/citrullinated EV proteins identified are presented in the Venn diagram in Figure 2C.

The Proteome and Citrullinome of Halibut Serum EVs
Total protein content, as well as F95 enriched protein content, representative of deiminated protein cargo in EVs (the "EV-citrullinome"), was identified by LC-MS/MS analysis. A range of proteins relating to innate and adaptive immunity, as well as gene regulation and cellular function, were identified as deiminated in EV cargo, and are listed in Table 1 (for full details on LC-MS/MS analysis, see Supplementary Table S1). Total EV protein cargo analysis revealed proteins relating to innate and adaptive immunity, nuclear proteins relating to gene regulation, proteins relating to cellular function and metabolism and are listed in Table 2 (for full details on LC-MS/MS analysis, see Supplementary Table S2). Total serum-EV proteins stained by silver staining are shown in Figure 2A, F95 enriched proteins from serum-EVs are shown in Figure 2B and the number of total EV proteins identified, overlapping with deiminated/citrullinated EV proteins identified are presented in the Venn diagram in Figure 2C. Table 1. Deiminated proteins in serum extracellular vesicles (EVs) of halibut (Hippoglossus hippoglossus L), as identified by F95-enrichment in conjunction with LC-MS/MS analysis. Deiminated proteins were isolated from serum-EVs from a pool of n = 4 fish, using immunoprecipitation with the pan-deimination F95 antibody. The resulting F95-enriched eluate was then analysed by LC-MS/MS and peak list files submitted to Mascot, using the Teleost UniProt database. Peptide sequence hits are listed, showing the number of sequences for protein hits and total score. Species hit names are indicated. In the case of uncharacterised protein ID, proteins matching the same set of peptides are indicated in brackets. Protein hits highlighted in pink (*) are specific to the F95 enriched EV fraction only. Protein names are written in bold. A full list of protein sequence hits and peptides is further provided in Supplementary Table S1             (C) Venn diagram shows the number of candidate protein hits identified as cargo in total serum EVs ("The serum EV proteome") as well as deiminated protein hits in EV cargo (the serum "EV citrullinome").

Complement Component C3, C4 and Pentraxin-Like Protein verified in Halibut EVs and F95 Enriched EV Protein Cargo Fractions using Western Blotting
Three candidate proteins which were identified as part of EV total protein cargo by LC-MS/MS, namely complement component C3, C4 and pentraxin-like protein, were further assessed by western blotting in halibut serum-EVs ( Figure 3A-C). Both total EV protein cargo as well as the F95 enriched protein cargo were assessed, using halibut-specific C3, C4 and pentraxin-like protein antibodies, respectively, which had previously been generated and validated in our laboratories [13,42]. Here, complement component C3 was verified to be present in total EV protein cargo, where it was strongly detected by western blotting, as well as at lower levels in the deiminated (F95-enriched) protein cargo ( Figure 3A). This confirmed the hits identified by the LC-MS/MS analysis, showing that C3 is exported in EVs both in normal and deiminated form (Tables 1 and 2). Complement component C4 was also confirmed to be exported in total EV cargo by western blotting, albeit at lower levels than C3, in accordance with the LC-MS/MS findings which identified C4 as a hit in total EV cargo. C4 was seen only at very low levels in deiminated form in the F95-enriched EV fraction by Wwestern blotting (Figure 3B), and was not identified as part of the F95-enriched cargo by LC-MS/MS. Pentraxin-like protein was strongly detected in total EV protein cargo by western blotting, but not in the F95-enriched EV protein fractions ( Figure 3C), in accordance with the results from the LC-MS/MS analysis, which only detected pentraxin in total EV cargo ( Table 2). (C) Venn diagram shows the number of candidate protein hits identified as cargo in total serum EVs ("The serum EV proteome") as well as deiminated protein hits in EV cargo (the serum "EV citrullinome").

Complement Component C3, C4 and Pentraxin-Like Protein verified in Halibut EVs and F95 Enriched EV Protein Cargo Fractions Using Western Blotting
Three candidate proteins which were identified as part of EV total protein cargo by LC-MS/MS, namely complement component C3, C4 and pentraxin-like protein, were further assessed by western blotting in halibut serum-EVs ( Figure 3A-C). Both total EV protein cargo as well as the F95 enriched protein cargo were assessed, using halibut-specific C3, C4 and pentraxin-like protein antibodies, respectively, which had previously been generated and validated in our laboratories [13,42]. Here, complement component C3 was verified to be present in total EV protein cargo, where it was strongly detected by western blotting, as well as at lower levels in the deiminated (F95-enriched) protein cargo ( Figure 3A). This confirmed the hits identified by the LC-MS/MS analysis, showing that C3 is exported in EVs both in normal and deiminated form (Tables 1 and 2). Complement component C4 was also confirmed to be exported in total EV cargo by western blotting, albeit at lower levels than C3, in accordance with the LC-MS/MS findings which identified C4 as a hit in total EV cargo. C4 was seen only at very low levels in deiminated form in the F95-enriched EV fraction by Wwestern blotting (Figure 3B), and was not identified as part of the F95-enriched cargo by LC-MS/MS. Pentraxin-like protein was strongly detected in total EV protein cargo by western blotting, but not in the F95-enriched EV protein fractions ( Figure 3C), in accordance with the results from the LC-MS/MS analysis, which only detected pentraxin in total EV cargo ( Table 2). protein cargo of halibut serum-EVs ("EVs") and in F95-enriched protein fractions from serum-EVs ("EVs_F95"), C3 α-and β-chains, as well as α-fragment (α-f) are indicated; (B) complement component C4 detection in total protein cargo of serum-EVs ("EVs") and lower detection observed in F95-enriched EV protein fractions ("EVs_F95"), C4 α-, β-and γ-chains are indicated; (C) pentraxin-like protein detection in total EV protein cargo ("EVs"), which was not detected in the F95-enriched EV protein fractions ("EVs_F95").

Protein Interaction Networks Enriched for Halibut Serum-EV Deiminated/Citrullinated Protein Cargo
For the generation of protein-protein interaction networks to further understanding of putative protein pathways regulated by deimination, deiminated (F95-enriched) protein hits from halibut EVs were assessed by STRING analysis. The protein hits were assessed using the general teleost STRING database, selecting the zebrafish (Danio rerio) database as a model database, as no specific database for halibut is available in STRING and zebrafish showed the highest identity with the teleost protein hits identified as deiminated in halibut serum-EVs. The protein-protein interaction networks showed a PPI enrichment p-value of 5.15 × 10 −5 , indicating significantly more interactions than expected from a random set of proteins ( Figure 4). protein cargo of halibut serum-EVs ("EVs") and in F95-enriched protein fractions from serum-EVs ("EVs_F95"), C3 αand β-chains, as well as α-fragment (α-f) are indicated; (B) complement component C4 detection in total protein cargo of serum-EVs ("EVs") and lower detection observed in F95-enriched EV protein fractions ("EVs_F95"), C4 α-, βand γ-chains are indicated; (C) pentraxinlike protein detection in total EV protein cargo ("EVs"), which was not detected in the F95-enriched EV protein fractions ("EVs_F95").

Protein Interaction Networks Enriched for Halibut Serum-EV Deiminated/Citrullinated Protein Cargo
For the generation of protein-protein interaction networks to further understanding of putative protein pathways regulated by deimination, deiminated (F95-enriched) protein hits from halibut EVs were assessed by STRING analysis. The protein hits were assessed using the general teleost STRING database, selecting the zebrafish (Danio rerio) database as a model database, as no specific database for halibut is available in STRING and zebrafish showed the highest identity with the teleost protein hits identified as deiminated in halibut serum-EVs. The protein-protein interaction networks showed a PPI enrichment p-value of 5.15 × 10 −5 , indicating significantly more interactions than expected from a random set of proteins ( Figure 4).

Protein Interaction Networks Enriched for Halibut Serum-EV Total Protein Cargo
The same approach for the generation of protein-protein interaction networks, selecting the zebrafish (D. rerio) STRING database as a representative database for teleost fish, was also applied for total protein EV cargo identified in halibut, showing a PPI enrichment p-value: <1.0 × 10 −16 for the protein networks generated, indicating significantly more interactions than expected from a random set of proteins ( Figure 5).
Reactome pathways for total EV protein cargo included LDL remodelling, plasma lipoprotein assembly, remodelling and clearance, innate immune system, Toll-like receptor cascades, neutrophil degranulation, the regulation of complement cascade, terminal complement pathway, the activation of C3 and C5, the initial triggering of complement, platelet degranulation, platelet activation, GRB2:SOS linkage to MAPK, integrin signalling, integrin cell surface interactions, p130Cas linkage to MAPK signalling for integrins, MAP2K and MAPK activation, the activation of matrix metalloproteinases, chylomicron assembly, the common pathway of fibrin clot formation, the intrinsic pathway of fibrin clot formation, the formation of fibrin clot, clotting cascade, platelet aggregation (plug formation), the formation of the cornified envelope, the regulation of IGF transport, the binding and uptake of ligands by scavenger receptors, collagen degradation, the metabolism of vitamins and cofactors, retinoid metabolism and transport, clathrin-mediated endocytosis, peptide ligand-binding receptors, extracellular matrix organisation, G alpha signalling events, hemostasis, signalling and aggregation, developmental biology, GPCR downstream signalling, post-translational protein modification, signal transduction, metabolism of proteins ( Figure 5B).  UniProt keywords for total EV protein content included methylation, kringle, nucleosome core, serine protease, secreted, chromosome, protease, disulphide bond, signalling ( Figure 5C). PFAM protein domains for total EV protein cargo included fibrinogen alpha/beta chain family, anaphylatoxin-like domain, vault protein inter-alpha-trypsin domain, MG2 domain, alpha-2-macroglobulin family, alpha-2-macroglobulin complement component, lipoprotein amino terminal region, UNC-6/NTR/C345C module, kringle domain, domain of unknown function (DUF1943), vWf type A domain, vWf type D domain, hemopexin, FYVE zinc finger, fibrinogen beta, gamma chains, C-terminal globular domain, trypsin-like peptidase domain, CUB domain, core histone H2A/H2B/H3/H4, and trypsin ( Figure 5D).

Discussion
This is the first study to assess EV profile signatures in halibut biofluids, identifying both total serum-EV protein cargo as well as deiminated protein cargo in serum-EVs. The size profiling of halibut serum-EVs by NTA showed vesicles in the range of 50-600 nm, which indicates a higher amount of larger EVs compared with human EVs, which typically fall in the size range of 30-300 nm. In comparison, while few teleost fish have been profiled for EVs, cod (Gadus morhua), serum-EVs were found to be in the size range of mainly 50-300 nm [33,38], while cod mucus-EVs are in the size range of 50-500 nm [32]. In other taxa across the phylogeny tree, differences in plasma or serum EV size profiles have indeed been reported. In elasmobranches (nurse shark Ginglymostoma cirratum) a higher abundance of small EVs in the 10-200 nm size range was observed [14]; in a group of eight pelagic seabird species, some species-specific differences were reported showing plasma-EVs at 50-200 nm size range for some birds and others showing larger EVs at 250-500 nm [37], while in reptile (alligator-Alligator mississippiesis), plasma EVs were in the size range of 50-400 nm [16]. In llama (Lama glama), plasma-EVs were reported at 40-400 nm [34], while Bos taurus plasma-EV showed size profiles of 70-500 nm [35]. Naked mole-rat (Heterocephelus glaber) plasma shows similar EV size profiles as human plasma at 50-300 nm [38], as does rat (Rattus norvegicus) plasma at 50-250 nm [43]. In sea mammals, such as pinnipeds and cetaceans, serum-EVs were observed at 50-600 nm in seals [40], similar to as observed in halibut in the current study. In four species of whale, EV profiles were seen in the ranges of 50-500 (minke whale Balaenoptera acutorostrata), 50-400 (fin whale Balaenoptera physalus), 80-300 (humpback whale Megaptera novaeangliae) and 90-300 nm (Cuvier's beaked whale Ziphius cavirostris), respectively, while orca serum-EVs (Orcinus orca; dolphin family) were reported at 30-500 nm [39]. Reports of EV profiling of haemolymph from species lower in the phylogeny tree include Crustacea (lobster Homarus americanus) with EVs in the 10-500 nm size range (with the majority of EVs being small in the 22-115 nm size range) [22]; Mollusca haemolymph EVs at 50-300 nm (blue mussel, Mytilus edulis), 30-300 nm (soft shell clam Mya arenaria), 90-500 nm (Eastern oyster Crassostrea virginica) and 20-300 nm (Atlantic jacknife clam Ensis leei), respectively [24]; Arthropoda (horseshoe crab Limulus polyphemus) EVs at 20-400 nm (with the majority of EVs falling within 40-123 nm) [23]. In the protozoa Giardia intestinalis, two distinct size populations of EVs have been described (20-80 nm and 100-400 nm, respectively), which display different functions in host-pathogen interactions [21]. In Gram-negative and Gram-positive bacteria, with EV profiles described at 10-600 nm and 60-400 nm, respectively, EV profiles were shown to change in response to drug-treatment both with respect to size profile and EV cargo content [19,44]. This does indicate that EV size profiles differ between taxa and this may, amongst others, also have effects on EV cargo content, including proteomic, post-translationally modified proteomic cargo, as well as other genomic and non-coding RNA and mitochondrial-derived cargo [45]. Indeed, in teleost, it has been reported that changes in EV numbers and EV deimination protein and microRNA cargo can be a biomarker for environmental temperature factors [33] and, in response to other stressors, teleost plasma EVs have been found enriched with Hsp70 [46] and selected micro-RNAs [47]. In human parasitic disease, EV profiles can also be indicative of infection status [48]. Therefore, the characterisation of EVs across a wide range of taxa further highlights their potential for biomarker application or "EV-fingerprinting" for the assessment of animal health.
Analysing both whole proteomic and the deiminated protein content of halibut serum-EVs in the current study, some differences were found in protein-interaction pathways, while overall both the whole proteome and the EV-citrullinome involved a number of immune, metabolic and gene regulatory pathways.
When assessing protein-protein interaction networks for EVs enriched in deiminated proteins, these related to local network clusters for deiminated proteins in serum-EVs included histone H3/CENP-A, core histone H2A/H2B/H3/H4 network, post-translational protein phosphorylation and the regulation of IGF transport. In relation to such networks, UniProt keywords for deiminated proteins identified in serum-EVs included methylation, cytoskeleton, disulphide bond, cytoplasm and signalling. Reactome pathways enriched in deiminated proteins in the serum EVs included GRB2:SOS linkage to MAPK signalling for integrins, p130Cas linkage to MAPK signalling for integrins, MAP2K and MAPK activation, integrin signalling, initial triggering of complement, L1CAM interactions, posttranslational protein phosphorylation, regulation of actin dynamics for phagocytic cup formation, regulation of IGF transport, platelet degranulation, integrin cell surface interactions, cell junction organisation, clathrin-mediated endocytosis, VEGFA-VEGFR2 pathway, extracellular matrix organisation, developmental biology, innate immune system and neutrophil degranulation. Correspondingly, PFAM protein domains for deiminated proteins identified in the serum EVs included alpha-macro-globulin tiolester bond-forming region, alpha-2-macroglobulin family N-terminal region, MG2 domain, alpha-2 macroglobulin family, alpha-macroglobulin complement component, alpha-macroglobulin receptor, UNC-6/NTR/C345C module, core histone H2A/H2B/H3/H4 and trypsin. SMART protein domains for deiminated EV proteins included alpha-macroglobulin family, alpha-2macroglobulin, alpha-macroglobulin receptor, kazal-type serine protease inhibitors, domains found in plexins, semaphorins and integrins and trypsin-like serine. Protein domains and features (InterPro) for deiminated proteins in serum-EVs included macroglobulin domain MG4 and MG3, the tissue inhibitor of metalloproteinases-like OB-fold, netrin module, non-TIMP type, netrin domain, PSI domain, peptidase S1A, chymotrypsin family, peptidase S1 PA clan, serine proteases trypsin family, histidine active site and serine active site.
In comparison with deiminated EV protein content, more pathways were revealed for serum-EV total protein content, as would be expected due to only some of the proteins in the EV cargo being candidates for post-translational deimination and exported in EVs in deiminated form. Assessing protein interaction networks for total protein EV content showed local network clusters for fibrinogen family, fibrinolysis, the common pathway of fibrin clot formation, clotting cascade, ApoM domain, selenoprotein P, histone H3/CENP-A, histone H4, the terminal pathway of complement, alternative complement activation, MG2 domain, terminal complement pathway, lectin pathway, plakophilin/delta catenin desmosomal, the regulation of IGF transport, adherens junctions interactions, MHC class II antigen, MHC class II antigen presentation, post-translational protein phosphorylation, Histone H4, and core histone H2A/H2B/H3/H4.
The reactome pathways for total EV protein cargo included LDL remodelling, plasma lipoprotein assembly, remodelling and clearance, innate immune system, Toll-like receptor cascades, neutrophil degranulation, the regulation of the complement cascade, the terminal complement pathway, the activation of C3 and C5, the initial triggering of complement, platelet degranulation, platelet activation, GRB2:SOS linkage to MAPK, integrin signalling, integrin cell surface interactions, p130Cas linkage to MAPK signalling for integrins, MAP2K and MAPK activation, the activation of matrix metalloproteinases, chylomicron assembly, the common pathway of fibrin clot formation, the intrinsic pathway of fibrin clot formation, the formation of fibrin clot, clotting cascade, platelet aggregation (plug formation), the formation of the cornified envelope, the regulation of IGF transport, the binding and uptake of ligands by scavenger receptors, collagen degradation, the metabolism of vitamins and cofactors, retinoid metabolism and transport, clathrin-mediated endocytosis, peptide ligand-binding receptors, extracellular matrix organisation, G alpha signalling events, hemostasis, signalling and aggregation, developmental biology, GPCR downstream signalling, post-translational protein modification, signal transduction, and the metabolism of proteins.
Proteomic analysis using LC-MS/MS, identified a range of innate and adaptive immune proteins to be exported in serum-EVs, including in deiminated form, as listed above. This also included a range of complement components, whereof C3 and C5 were detected as deiminated in serum EVs, while in total EV cargo, C1, C3, C4, C5, C6, C7, C8 and C9 were also identified as hits, as well as factor B and factor H. This correlates with previous findings reporting C3 to be deiminated in teleost fish, both in halibut and cod [13,32,33]. Furthermore, a proteomic analysis of deiminated target proteins in halibut serum identified C5, C7, C8 C9 and C1-inhibitor to be deiminated in whole halibut serum [13]. These findings, and the current study, indicate that not all complement components are exported in EVs in deiminated form, and some are found in deiminated form only in whole serum, while being exported in non-deiminated form in serum-EVs. Recent studies assessing protein deimination across the phylogeny tree have indeed identified various complement components as deimination candidates in a range of taxa [14,16,[32][33][34][35]37,39,40]. Furthermore, C5 has been verified to be a deimination candidate by bacterial arginine deiminase, allowing for immune modulation of the host and bacterial immune evasion [18].
In the current study we furthermore evaluated by western blotting some key complement proteins identified by LC/MS-MS in EV total protein cargo and deimination-enriched protein cargo. For this purpose, we used halibut-specific antibodies against C3, C4 and pentraxin-like protein, previously developed and described by our group [13,42]. Using western blotting analysis, we verified the presence of C3, C4 and pentraxin-like protein in halibut serum-EVs, showing that these are indeed exported in EVs, as also identified by LC-MS/MS anlaysis. The C3 antibody also reacted strongly with the F95 enriched protein eluate from the serum EVs, while a lower signal was seen for C4, indicating that C3 is present at higher levels in deiminated form in serum-EVs, compared with C4. Pentraxinlike protein was only observed in total protein cargo of serum-EVs, but not the F95 enriched serum-EV eluate and this corresponds with the LC-MS/MS analysis which revealed hits with a pentaxin for the total protein cargo analysis of serum-EVs, but not the F95-enriched fraction. Our current findings in halibut serum-EV cargo also correspond to our previous analysis on serum EVs and mucus EVs in Atlantic cod, where C3 was detected at higher levels in serum-EVs than C4, both for total protein as well as in the F95-enriched eluate for a putative deiminated form [32,33,38]. Furthermore, cod serum and mucus EVs were also found to contain pentraxin-like protein (CRP-like), which was not detected in deiminated form in the cod EVs, similar to as observed for pentraxin-like protein in halibut serum-EVs in the current study.
Overall, our and others' findings indicate that the complement system can be modulated by deimination both by the host and by pathogen interactions. Understanding of post-translational regulation of complement components via deimination is still in its infancy and requires in depth investigation as deimination may facilitate multifaceted functions of complement proteins in immunity and tissue remodelling in health and disease, also across phylogeny. Such regulation via deimination may furthermore allow for targeted modulation in relation to a range of pathological processes, including infection and autoimmune diseases, where PADs, EVs and the complement system all play important roles.
Furthermore, some protein candidates were only detected in the F95 eluate, indicating that they were exported in deiminated form only in the serum EVs. This included cytoplasmic 2 actin, tubulin alpha chain, keratin 93, trypsin-3-like, centrosomal protein of 162 kDa, 2-phospho-D-glycerate hydro-lyase, integrin beta and myosin_tail_1 domain-containing protein. Compared with a previous analysis from our group on deiminated proteins in whole halibut serum [13], deiminated candidates found here to be exported specifically in EVs are cytoplasmic 2 actin, tubulin alpha chain, centrosomal protein of 162 kDa, 2phospho-D-glycerate hydro-lyase, integrin beta and myosin_tail_1 domain-containing protein. This indicates that there are differences in deiminated protein cargo in serum-EVs compared with whole serum, and this corresponds to findings from other comparative studies analysing differences in KEGG (Kyoto encyclopedia of genes and genomes) and GO (gene ontology) enrichment pathways for deiminated proteins in whole serum/plasma versus EVs in diverse taxa, including in cow, camelid, alligator, rat, naked mole-rat, shark and cod [14,16,[33][34][35][36]43]. Furthermore, differences in deimination signatures in whole serum versus EV cargo have been reported to relate to immune/growth trade off in response to environmental temperature in teleost cod [33]. Such findings, including the findings reported in our current study, emphasise that variations in EV cargo, including via the transport of deiminated proteins, may play hitherto under-recognized and important roles in cellular communication in health and disease across the phylogeny tree.

Fish and Sampling
Blood was collected from four adult halibut (Hippoglossus hippoglossus L.; weight 4.5-5.0 kg), which were obtained from the experimental fish farm Fiskeldi Eyjafjardar hf, Thorlakshofn, Iceland (under licence from the Institute for Experimental Pathology, University of Iceland, number #0002 kt-650269-4549, approved by the central animal ethics committee in Iceland (Icelandic Food Regulation Authority, MAST Matvaelastofnun). Following 1-3 mL blood collection from a gill vessel, the blot was left to clot overnight at 4 • C, and thereafter serum collection was performed by centrifugation at 750 g for 10 min. Serum aliquots of 200 µL were stored at −20 • C until used. The health status of the fish at the fish farm was routinely examined at regular 3 monthly intervals by the Fish Disease Laboratory, Institute for Experimental Pathology, Keldur, Iceland, declaring the fish healthy and disease free.

EV Isolation and Nanoparticle Tracking (NTA) Analysis
EVs were isolated from halibut serum of four individual fish by step-wise centrifugation, according to previously established methods in our laboratory [14,22,39] and according to the guidelines of the International Society for Extracellular Vesicles (ISEV) [49]. Total EV isolates were prepared from the individual 100 µL serum aliquots (n = 4), which were diluted 1:5 in Dulbecco's PBS (DPBS, which had previously been ultrafiltered using a 0.22 µm filter, before use) and then centrifuged at 4000× g for 30 min at 4 • C, to ensure the removal of aggregates and apoptotic bodies. The supernatants containing the EVs were collected and ultracentrifuged at 100,000× g for 1 h at 4 • C. The EV-enriched pellets were then resuspended in 1 mL DPBS ("washing step") and ultracentrifuged again at 100,000× g for 1 h at 4 • C. The final EV-enriched pellets were then resuspended in 100 µL DPBS and analysed by NTA for size distribution profiles, using the NanoSight NS300 system (Malvern Panalytical Ltd, Malvern, UK), recording five 60 s videos for each sample. The number of particles per frame was kept in-between 40 to 60 and replicate histograms were generated from the videos, using the NanoSight software 3.0 (Malvern), representing mean and confidence intervals of the five recordings for each sample.

Transmission Electron Microscopy (TEM)
EVs were further characterised by Transmission Electron Microscopy (TEM) as follows: A pool of EVs, isolated from serum of the four individual animals as described above, was used for morphological analysis using TEM according to previously described methods [16,23,34]. In brief, 100 mM sodium cacodylate buffer (pH 7.4) was used to resuspend the EVs, which were then placed onto a glow discharged carbon support film on a grid and fixed at room temperature for 1 min in 2.5% glutaraldehyde in 100 mM sodium cacodylate buffer (pH 7.0). For the staining of EVs, 2% aqueous Uranyl Acetate (Sigma, Gillingham, UK) was used for 1 min, thereafter removing the excess stain. EV imaging was performed using a JEOL JEM 1400 transmission electron microscope (JEOL, Tokyo, Japan) operated at 80 kV at a magnification of 30,000× to 60,000×. Digital images were recorded using an AMT XR60 CCD camera (Deben, Bury St. Edmunds, UK).

Proteomic Analysis and Protein Identification
The isolation of deiminated/citrullinated proteins from serum-EVs was carried out by immunoprecipitation, using the Catch and Release ® v2.0 Reversible Immunoprecipitation System (Merck, Feltham, UK) according to the manufacturer's instructions in conjunction with the pan-deimination F95 antibody (MABN328, Merck), which specifically detects proteins modified by citrullination [50]. F95 enrichment was performed overnight at 4 • C on a rotating platform from a pool of sera (n = 4 individuals), followed by the elution of the F95 bound proteins under reducing conditions, according to the manufacturer's instructions (Merck). The F95 eluate was diluted in 2 × Laemmli sample buffer for subsequent SDS-PAGE and western blotting analysis. The total F95 bound protein eluate, as well as total protein from serum-EVs, were also analysed by liquid chromatography-mass spectrometry (LC-MS/MS) (performed by Cambridge Centre for Proteomics, Cambridge, UK), by in-gel digestion, as previously described [32,34]. For the identification of deiminated protein hits, the files were submitted to the Mascot search algorithm (Matrix Science, London, UK) and searched against the UniProt database for Teleostei (CCP_Teleostei Teleostei_20201009; 4085639 sequences; 2121030378 residues). A search was also conducted against a common contaminant database (cRAP 20190401; 125 sequences; 41,129 residues). A significance threshold value of p < 0.05 and a peptide cut-off score of 53 were also applied (carried out by Cambridge Proteomics, Cambridge, UK).
In addition to the LC-MS/MS analysis, both total EV proteins and F95 enriched EV proteins were assessed specifically for halibut C3, C4 and pentraxin-like protein content, using mono-specific antibodies, which were previously prepared against these proteins by our group [13,42] (see Section 4.5).

Silver Staining
Total proteins isolated from serum EVs and the F95-enriched protein eluates from halibut serum EVs were assessed by silver staining following SDS-PAGE in 4-20% gradient TGX gels (BioRad) under reducing conditions. The BioRad Silver Stain Plus Kit (1610449, BioRad) was used to visualise the protein bands according to the manufacturer's instructions (BioRad).

Protein-Protein Interaction Network Analysis
For the construction of protein-protein interaction networks for deiminated proteins identified in halibut serum-EVs and for total protein content from serum EVs, respectively, STRING analysis (Search Tool for the Retrieval of Interacting Genes/Proteins; https://string-db.org/) was applied. The protein networks were built based on the protein names, using the teleost fish STRING database and using the function of "search multiple proteins". Settings were as "basic" and "medium confidence". Colour lines connecting the nodes represent the following evidence-based interactions for the network edges: "known interactions" (this is based on experimentally determined data or curated databases); "predicted interactions" (this is based on gene co-occurrence, gene neighbourhood or gene fusion); "others" (this is based on co-expression, text mining or protein homology (see colour key for lines in Figure 4A). Networks were assessed for local network clusters, reactome pathways, PFAM and SMART protein domains and UniProt keywords. The zebrafish (Danio rerio) STRING database was used as representative for Teleostei for the creation of the networks as no specific halibut STRING database is available (due to lack of annotation available for halibut), and D. rerio showed the most hit number identity with the proteins identified in halibut EVs.

Statistical Analysis
The Nanosight 3.0 software (Malvern) was used for the generation of NTA curves, which represent mean and standard error of mean (SEM), indicated by confidence intervals. Significance for protein network analysis generated in STRING (https://string-db.org/) was considered as p ≤ 0.05.

Conclusions
This study is the first report of EV profile signatures in halibut, analysing total protein and specifically also deiminated protein cargo in serum-EVs. Halibut serum EVs showed a poly-dispersed population with EVs in the size range of 50-600 nm, positive for phylogenetically conserved EV markers. Proteomic analysis of EV total protein cargo revealed 124 protein hits and 37 deiminated protein hits, whereof 15 hits were particularly identified in deiminated form only. Protein interaction network analysis revealed GO pathways for EV mediated protein cargo transport, relating to a range of gene regulatory, immune, metabolic and developmental processes, some of which were enriched for deiminated proteins. Further assessment of key immune related proteins-complement components C3, C4 and pentraxin-identified that C3 is exported in serum-EVs at higher levels than C4, also in deiminated form, while pentraxin was found in whole protein EV content only, but not in deiminated form. Our findings emphasize the putative differences in cell communication mediated by EV protein versus post-translationally deiminated protein cargo (the "EV-citrullinome"), providing novel insights into EV-mediated communication in halibut serum. Our findings furthermore contribute to current understanding of EV signatures across the phylogeny tree, with the potential for biomarker development and EV "fingerprinting" for the assessment of animal health.