Identication of transcripts relating to host-parasite interactions and feeding-associated activities through de novo transcriptome sequencing of branchiuran sh lice, Argulus foliaceus (Linnaeus, 1758) and Argulus coregoni Thorell, 1865

The genus Argulus, whose members are widely known as “sh lice”, comprises a group of obligate ectoparasites that cause argulosis. This parasitic disease affects both wild and cultured sh populations and can lead to signicant mortalities. Our previous work conrmed the presence of immunomodulatory or other active products in the secretions of Argulus foliaceus (Linnaeus, 1758) using proteomic analysis conducted on secretory / excretory products (SEPs). Some of the compounds identied have also been described for other host-associated Crustacea, particularly the parasitic copepod Lepeophtheirus salmonis (Krøyer, 1837). The identication and characterisation of such components can contribute to the understanding of the host-parasite relationship for argulids and may also assist in the development of novel control methods, however, due to limitations in existing genomic resources, only 27 proteins could be conrmed in SEPs. In this study, high-throughput sequencing, supported by the use of a range of bioinformatics tools and methodologies were employed on host-associated stages of A. foliaceus and A. coregoni Thorell, 1865 to explore elements of the transcriptome associated with host-parasite interactions and feeding-associated activities.


Background
Argulus is one of the most common crustacean sh ectoparasite genera encountered worldwide [1][2][3], being frequently referred to as " sh lice". Argulus is the largest and most diverse genus of the branchiuran family Argulidae [4]. Members of this genus are responsible for the condition argulosis, a parasitic disease affecting both wild and cultured sh populations [5,6].
Argulus spp., such as Argulus japonicus Thiele, 1900 and Argulus coregoni Thorell, 1864, have been described as causing a skin haemorrhagic response in carp (Cyprinus carpio Linnaeus, 1758) and rainbow trout (Oncorhynchus mykiss Walbaum, 1792) respectively [7,8,9] and it has been suggested that Argulus spp. might secrete substances during feeding to modulate host immunity or to facilitate blood feeding [7]. A general lack of knowledge in this area led von  to call for studies looking into the composition of the expected active secretions originating from branchiuran exocrine glands associated with feeding [10]. Due to limited genomic resources, we recently identi ed only 27 proteins in the secretory / excretory products (SEPs) of Argulus foliaceus [11]. Some of the compounds identi ed have also been described from other host-associated Crustacea, particularly the parasitic copepod Lepeophtheirus salmonis (Krøyer, 1837) [12].
Although argulids affect aquaculture and sheries globally, causing major economic impacts for freshwater aquacultured sh species worldwide, few studies have examined the mechanisms underlying physiological and behavioural aspects of the parasite [11]. In the UK, Argulus spp. have economic impacts for both aquaculture and sports shing industries. To date, control methods employed have targeted few of the biological aspects of the parasite and have had little impact in preventing argulosis, with those controls that do exist, all having critical limitations in their use [14]. Hence new strategies for control are still urgently needed. One avenue for control is the potential for developing vaccines or other immunoprophylactic approaches such as use of immunostimulants / functional feeds. To achieve this, better knowledge of the host response to the parasite and the parasite's mechanisms for combating host immunity are required.
A range of newer techniques, including high-throughput sequencing which allows examination of the parasite's transcriptome, can provide a greater understanding of the biology of the parasite and of key genes involved in particular activities. Such genes and the products they code for can provide potential targets for use in control of argulosis [12,13]. The rst transcriptome analysis for argulids was conducted by  for Argulus siamensis Wilson C.B., 1926, which has a severe impact on the aquaculture industry in India. An Illumina-based sequence analysis, led to the identi cation of 19,290 coding sequences (CDS) including 184 novel CDS and 59,019 open reading frames (ORFs). Using this same transcriptome shotgun assembly (TSA), Christie (2014) was subsequently able to identify 27 transcripts encoding putative neuropeptide precursors, these peptides being known to play a role in physiological/behavioural control across species. This was the rst nding of neuropeptide encoding sequences in any member of the subclass Branchiura [16].
More recently, Pinnow et al. provided Argulus foliaceus (Linnaeus, 1758) transcriptome data (NCBI sequence read archive (SRA) under the Bio-Project number PRJNA293150. 52,725,850). The assembly resulted in 8,424 contigs. These transcriptome data were only used to identify and characterise hemocyanin subunits in A. foliaceus and A. siamensis and provided evidence that hemocyanin was the principal respiratory protein in the stem lineage of the Pancrustacea [15].
Despite the above, genomic resources for Argulus spp. and for the Branchiura more widely, remain extremely impoverished, providing a major constraint to further research in this area. The work reported in this study sought to provide new transcriptomic resources for A. foliaceus and A. coregoni, the two native Argulus species in the UK, and to analyse the new sequences with respect to the previous two transcriptome libraries established for A. siamensis [12] and A. foliaceus [15]. Understanding key biological processes in Argulus spp. will help in searching for candidate immunomodulatory or drug targets, which may assist researchers to develop new drugs or vaccines to provide supplementary control methods. This study also highlights the fact that Argulus spp. might prove productive as a tractable ectoparasitic arthropod model. By lling gaps in the genomic knowledge for this genus, this study provides a useful addition to crustacean genomic resources, particularly for Branchiura and details the rst list of potential host immunomodulators for these three Argulus species.

Methods
Transcriptome sequencing parasites. The latter comprised tissue taken from the expected position of the pre-oral spine gland and proboscis glands as established from histological sections [18]. Equal amounts of RNA from the four sources were used.
Library preparation from supplied total RNA and transcriptome sequencing, using Illumina based protocols, were outsourced to commercial suppliers. Different companies were used for two species, dictated by service availability at the time. For A. coregoni, the Edinburgh Genomics (University of Edinburgh, UK) provided 150 base paired-end reads generated on the MiSeq platform, while for A. foliaceus the Theragen BioInstitute, Republic of Korea supplied 100 base paired-end reads generated on a HiSeq 2500 platform).
To assist assembly, two existing Argulus transcriptome datasets were used. These comprised an A. siamensis dataset [12] and an A. foliaceus dataset [15] lodged in the NCBI Bioprojects database under accession numbers PRJNA293150 and PRJNA167720, respectively.
Sequence data processing and transcriptome assembly for A. foliaceus and A. coregoni The transcriptome reads obtained for A. coregoni and A. foliaceus were assembled using the following bioinformatics pipeline, illustrated in Figure 1.

Quality Control Processing
Quality control of the raw data was conducted using FastQC v0.11.4 [16]. The data were then trimmed using the Trimmomatics v0.35 program [17]. This program was run using the parameters option "ILLUMINACLIP". Deconseq v0.4.3 program [18] was used to remove any sequence contamination. For this, reads were mapped against different databases.

De novo assembly and annotation
De novo assembly was performed on the cleaned RNA-seq raw data using Trinity v2.1.1 [19]; the K-25 default parameter of the Trinity software was used for the assembly. The selection of the Trinity tool for nal assembly was decided upon following a trial including other assembler software i.e. Velvet and ABySS. Trinity was found to give higher numbers of more consistent reads. All the sample reads were merged into a single dataset for each species for nal assembly. Then Transdecoder v2.0.1 [23] was used to nd the coding region within identi ed transcripts. Identi cation of the coding region transcripts gave ORFs and amino acid sequences, to prepare the assembled dataset for annotation. Annotation was achieved by 1) BLAST v2.2.31 software [21], BLAST+ [22] using the uniprot/trembl-invertebrate database as query [23] and 2) Annotation HMMER v3.1b2 [24] using the Pfam A v29.0 as query [25].

Gene Ontology annotations
Gene Ontology (GO) annotations were retrieved from the UniProt GO annotation project GO annotation was conducted for both A. foliaceus and A. coregoni transcriptomes as well as for the pre-existing Argulus transcriptome libraries (A. siamensis with Accession: PRJNA167720 and A. foliaceus with Accession: PRJNA293150) curated by the National Centre for Biotechnology Information (NCBI).

Identifying shared transcript expression among the three Argulus species
OrthoVenn is a web-based genomics tool used for comparative and visualisation analyses of orthologous protein clusters from different species. Use of cluster information helps in identifying protein function and evolution across multiple species [29]. OrthoVenn was used to check for shared expressed genes / proteins. The tool was used to identify gene clusters enriched in the four transcriptome libraries available for A. foliaceus, A. coregoni and A. siamensis. To achieve this, a list of proteins was rst prepared for each Argulus species and then the ClusterVenn tool used to generate Venn diagrams by clustering proteins of similar sequence and function, i.e. orthologous and paralogous genes, among the species. The tool compares between clusters only and ignores singletons where there are no orthologues between species and no paralogues within species. The default OrthoVenn parameters were used i.e. e-value cutoff of 1e-5 for all-to-all protein similarity comparisons and an in ation value (-I) of 1.5 for the generation of orthologous clusters.
In addition, OrthoVenn allows searching for speci c proteins using the BLAST feature. In this study, it was used to search for some known immunomodulatory candidates to check if any existed within the annotated shared clusters and related functions. The selected targets were: trypsin, serpin (serine protease inhibitor), cathepsin-L, aspartic protease, ferritin, cysteine protease, enolase, phospholipase, adenosine deaminase, apyrase, metalloprotease, thrombin inhibitor and venom serine protease.

Phylogenetic reconstruction
The longest coding sequence (CDS) for selected proteins of A. foliaceus, A. coregoni and A. siamensis were translated into amino acid sequences and a BLASTP was performed against all oligostracan sequences present in the NCBI NR database (accessed 2016-11-09). A total of 10 species were represented in the top 100 shared proteins (Additional le 1). Each shared protein sequence was retrieved and aligned individually using ClustalO v1.2.3. Each alignment was concatenated before running RaxML v8.0.0 using partitioned models (one model per protein, based on a GAMMA model of rate heterogeneity with an estimate of proportion of invariable sites and using the BLOSUM62 substitution rates and maximum-likelihood estimate of the base frequencies) with 1000 bootstrap [-# 1000 -m PROTGAMMAIBLOSUM62X].

Results
De novo transcriptome sequencing of Argulus coregoni and Argulus foliaceus Quality Control Processing All samples shipped passed the commercial companies external initial QC (Bioanalyser), with RIN values exceeding 0.8. Transcriptome libraries were prepared by the rms using Illumina based kits and platforms. One library sequenced by Edinburgh Genomics (MiSeq platform; 150 base paired-end reads) the other by Thetagen (HiSeq2500 platform, 100 base paired-end reads).

De novo assembly and annotation
The MiSeq transcriptome sequencing for A. coregoni generated 11,265,959 150 base paired-end raw sequence reads (Table 1). Of these, 10,840,092 reads passed quality control and ltering, and were then merged into a single dataset, assembled using Trinity v2.1.1, and blasted to generate a de novo transcriptome assembly comprising 40,954 transcripts (contigs). The maximum transcript contig length was 9,791 bp, the mean contig length was 1,787 and the N50 value was 2,339 bp.
The A. foliaceus sequencing (HiSeq2500 platform) produced 88,255,979 150 base paired-end raw sequence reads (Table 1) with 84,256,934 passing quality control and ltering. The high-quality reads were merged into a single dataset, assembled with Trinity v2.1.1, and blasted to get total assembled contiguous sequences of 66,940 reads. The maximum transcript contig length was 17,078 bp, the mean transcript contig length was 1,842 bp and the N50 value was 2,573 bp.
The pre-existing TSA for A. siamensis (Accession PRJNA167720) and A. foliaceus (Accession: PRJNA293150) were also examined and the total number of reads, number of assembled contigs and N50 are summarised in (Table 1).

Gene ontology annotation
Gene ontology (GO) terms were retrieved for the annotated transcripts, with the longest sequence for each transcript selected. Following the use of BLASTp against the UniProt_Trembl (invertebrate) database, the maximum number of GO terms were identi ed. The GO analysis was applied both to the current datasets (A. foliaceus and A. coregoni) and to the pre-existing reference datasets from NCBI (A. siamensis and A. foliaceus). Table 2 summarises the GO distribution for the three Argulus species from the four transcriptome datasets. The GO terms were associated with transcripts for molecular function (MF), biological process (BP) and cellular component (CC). The GO category terms for A. coregoni were the same as those obtained for A. foliaceus. In the MF category, binding activity and catalytic activity were the most heavily represented. For BP the most represented terms were "transport" and "metabolic process" and for CC the most represented terms were "membrane" and "integral component of membrane" Overall, the GO distributions for the four transcriptomes of the three species (A. coregoni, A. foliaceus and A. siamensis) were highly similar (Table 2), and in the GO distribution charts of the rst 10 clustered annotated proteins for the Argulus species (Fig. 2 & 3) Identifying shared genes expressed in Argulus species Venn diagrams ( Fig. 4A and B) [15] were combined to check for shared proteins/genes. Out of 7955 clusters over both datasets, 6522 orthologous clusters and 6425 single-copy gene clusters were found (Fig. 4A). OrthoVenn analysis of the three Argulus species showed that 13,324 orthologous clusters were formed based on the protein sequences from the three species. The diagram shows that 6,674 gene clusters were shared by all three species (Fig. 4B).
Blasting speci cally selected targets (trypsin, serpin, serine protease, cathepsin-L and aspartic protease, ferritin, cysteine protease, enolase, phospholipase, adenosine deaminase, apyrase, metalloprotease, thrombin inhibitor, venom serine protease) against the resulting shared clusters (6,674) gave a number of hits. For trypsin, 25 annotated clusters were identi ed and classi ed into six groups according to their functions. Serpin resulted in four clusters described as alaserpin (Swiss-Prot Hit database), all having the same GO annotation. Eleven clusters were found for serine protease and grouped into eight GO annotations. Only a single cluster each was identi ed for cathepsin (cathepsin-L), apyrase, aspartic protease (Aspartic protease 6), cysteine protease, enolase and thrombin inhibitor. Two types of ferritin were found with one being yolk ferritin. Two cluster proteins comprised phospholipase A2, enzymes that hydrolyse phospholipids into fatty acids and other lipophilic substances, combining 11 GO annotation functions. There were 17 clusters of adenosine deaminase as determined using Swiss-Prot Hit; each one having a different GO annotation function. By checking for matches to metalloprotease, 5 clusters distributed between three GO annotation functions were identi ed. Finally, searching for matches to venom serine protease resulted in 6 gene (protein) clusters with 3 different GO annotations (Table 3).

Phylogenetic reconstruction
The ndings of the phylogenetic analysis (Fig. 5) re ected those of Regier et al. [45] and con rms the position of the sequenced Argulus species with respect to the superclass Oligostraca.

Discussion
In this study we have generated sequence resources for the poorly characterised genus Argulus (A. foliaceus, A. coregoni and A. siamensis) and for the Branchiura more generally. While there were two existing genomic datasets for Argulus species (A. siamensis and A. foliaceus), these were limited in terms of their coverage. The new transcriptomes comprise larger data sets for Argulus with 84,256,934 reads assembled into 66,940 contigs for A. foliaceus and a rst transcriptome for A. coregoni, 10,840,092 reads, assembled to give 40,954 contigs.
As there is no reference genome for Argulus, the transcriptomes of both A. foliaceus and A. coregoni were assembled de novo using Trinity v2.1.1 software. The assembled transcriptomes are important for both genomic and proteomic studies [11], especially in terms of providing the complete ORF information for genes coding for proteins that may play important roles in biological processes and might therefore provide potentially suitable vaccine candidates [12] or targets for anti-parasiticides.

It was reported by Clark & Greenwood [31] that 50 % of crustacean transcripts from transcriptomic studies
were orthologous genes to a model organism and most of these genes belong to biological processes involving metabolism, development, or regulation of a biological process. This concurs with the Argulus GO outcomes, as the predominant genes from the biological process analysis were involved in response to metabolic processes. This was a predictable result as shown from other organisms, due to the enormous number of GO terms that cover basic processes required to maintain a living organism [47]. One of the most important processes, and one that has some shared properties between invertebrates and vertebrates is the innate immune system [48]. Knowledge of crustacean immune and stress responses are key to understanding crustacean physiology [46]. Although, the annotation method applied in this study is not su cient to provide detailed information about the immune pathways, the GO molecular functions associated with the Argulus transcripts highlighted speci c processes that can also be involved in host-parasite interactions such as hydrolase, transferase and metallopeptidase activities.
Hydrolases have been identi ed from different parasite venoms which are known to include proteases, peptidases and glycosidase enzymes [34,35]. As we have not only identi ed hydrolase transcripts here in Argulus tissues, but also recently in Argulus SEPs through proteomic analysis [11], the presence of hydrolases might re ect their contribution to host-parasite interactions in Argulus species. The identi ed metalloproteases from different arthropods are assumed to be related to different biological processes such as digestion, cuticle degradation and host immune suppression [47]. In general, the GO annotations from the three-species identi ed a high representation of genes involved in metabolism, protein binding and nucleotide binding. These are linked with high basal metabolic activity [51], and although the precise reason is not known, various studies on the effects of parasitism on infected mammals have reported that parasites increase the basal metabolic rate of the host [52]. The patterns of the GO analysis from three Argulus species were similar. Further characterisation of the transcripts involved in immune pathway functions and their potential contribution to successful parasitism can provide a valuable addition to understanding of Argulus biology and physiology.
The construction of a limited phylogenetic tree based on the top 100 shared genes for the four transcriptome datasets from Argulus, available from both this study and from previous data for other arthropods, con rms the position of the sequenced Argulus species with respect to the superclass Oligostraca, as previously suggested by Regier et al. [45]. These results suggest that A. coregoni is more closely related to A. foliaceus than to A. siamensis. This probably re ects the fact that A. foliaceus and A. coregoni are indigenous Argulus species in northern Europe and UK with closer and more recent evolutionary links than A. siamensis, an invasive species from Asia. Long association in similar environments is also re ected by similar host speci cities, or lack of them, allowing them to infect such disparate species as salmonids e.g. rainbow trout (O. mykiss) and cyprinids e.g. common carp (C. carpio). This may also mean that they share more similar mechanisms for successful parasite-host interaction.
Comparative functional annotation of overlapping clusters among the Argulus species identi ed 6,674 shared gene clusters, suggesting continued conservation after speciation. Identi cation of conserved genes can help in the search for more universal control targets in these species, for other Argulus species and perhaps for Branchiuran parasites more widely.
The saliva of almost all studied blood feeding arthropods has been shown to contain a range of immunomodulating proteins. Speci c components have active functions in altering the host immune response to help blood sucking [53] and exploiting aspects of these powerful substances can provide an effective way to control disease distribution [54]. A range of proteins / genes that have been previously characterised as being important immune mediators for blood-feeding insects and other haematophagous parasites, and identi ed recently by our group in SEPs of the parasitic crustaceans, L. salmonis and A. foliaceus [11,12], were here discovered in Argulus spp. transcriptome for the rst time and indeed for the Branchiura more widely. Proteins identi ed as trypsin, serpin, serine protease, cathepsin-L, aspartic protease, ferritin, cysteine protease, enolase, phospholipase, adenosine deaminase, apyrase, metalloprotease, thrombin inhibitor, and venom serine proteases in shared transcript clusters for the three Argulus species provide potential candidates for investigation of Argulus controls in the future.
Transcripts of a trypsin-like protease were detected in Argulus. Skin mucus of Atlantic salmon (Salmo salar L.) infected by sea lice L. salmonis contains trypsin-like proteases derived from the parasite, which suggests that the secretion of this protease during feeding is a mechanism to impair the host immune response [28, 29 ,54-58]. Seven types of L. salmonis trypsin-like proteases have been inferred to play roles in digestion and other host-parasite interactions [59]. Valenzuela-Munoz et al. similarly identi ed the up-regulation of trypsin-1 gene in the sea louse Caligus rogercresseyi (Boxshall & Bravo, 2000) posttreatment with an organophosphate parasiticide, suggesting the gene to be associated with resistance / susceptibility to organophosphates [56], while Braden et al [60] found signi cant upregulation of trypsin-1 in L. salmonis after 2 days post infection of lice feeding on Atlantic salmon (S. salar). Serine protease inhibitors (serpins), identi ed for the rst time here in Argulus, are recognised as the largest protein group in tick saliva and were suggested to play an important role during tick feeding by modulating host immunity [33,[61][62][63][64][65]. Serpins characterised from Ixodes scapularis saliva were found to inhibit thrombin and trypsin activity and lower platelet aggregation [66]. Serpins are also recognised from insects, for example salivary serpin from Aedes aegypti mosquitoes, serves as an active anti-coagulant due to its ability to inhibit Factor Xa, which is the activated form of the coagulation factor thrombo-kinase (Stark & James 1998). Serpins have also been noted to be differentially expressed in salivary glands of the mosquito Anopheles culicifacies, suggesting involvement in the blood feeding process [40]. We have previously identi ed serpins in A. foliaceus SEPs suggesting their role in facilitating parasitism and modulating host immune responses in argulids, and the presence of trypsins suggests their role in parasite-host interactions and digestion [11].
Transcripts for cathepsin-L identi ed from Argulus in the present study may signify a role for this protease in digestion and / or modulation of the host immune system. In Hysterothylacium aduncum (Rudolphi, 1802), a parasitic anisakid nematode of sh, cathepsin L was proposed to be a vital protein involved in digestion processes [68]. Cathepsin L was also identi ed as a potential anticoagulant in SEPs of adult Haemonchus contortus, a parasitic nematode of sheep and other ruminants [38]. Franta et al. suggested that this could provide a novel target for control of ticks and tick-borne pathogens [69]. Transcripts for cathepsin-L were also reported to be expressed by the copepodid life stages of sea lice, C. rogercresseyi [34,70], and were upregulated in feeding adult L. salmonis [60] and showed activity in SEPs of pre-adult and adult L. salmonis [12,71], suggesting a likely role in modulating the host immune system to facilitate attachment and feeding [71].
Transcripts for aspartic protease 6 were also identi ed from Argulus in this study, this being a protein common to the three species examined. Aspartic proteases have been proposed to have a role during C. rogercresseyi feeding through haemoglobin proteolysis, where aspartic protease D was highly expressed in the chalimus and adult stages [34]. The salivary glands of blood feeding arthropods frequently contain anticoagulation inhibitors [32] and in this study we found evidence for the existence of the thrombin inhibitor rhodniin in all three Argulus species studied. It has previously been reported to prevent blood clotting by inhibiting thrombin [72] to facilitate blood acquisition. Apyrase is another immunomodulatory molecule, which we found in Argulus. This enzyme is believed to have anti-pain; anti-in ammatory and anti-haemostatic activity, thus potentially playing a role during feeding to prevent blood clotting and host behavioural responses [32]. Ribeiro et al. described the apyrase from R. prolixus to act as a platelet aggregation inhibitor [73]. A number of metalloproteases were also identi ed in Argulus, these being a common enzyme group in other blood-feeding arthropods as well, although it is important to note that metalloproteases also perform a wide range of other roles in arthropods more generally. Transcriptome analysis in conjunction with KEGG pathway analysis conducted by Sahoo et al. on A. siamensis resulted in the identi cation of proteases including serine proteases and metalloproteases which have been described to have antigenic properties in some ectoparasites [14]. Apyrase and metalloprotease found in tick salivary secretions were suggested to affect antiplatelet aggregation and coagulation activity [44] and may serve a similar function for processing blood meals in Argulus. Adenosine deaminase, which can act as a vasodilator and antiplatelet aggregation factor for arthropod feeding [32,43] was also found in Argulus in the present study. By searching for adenosine deaminase using OrthoVenn BLAST feature, two contigs of adenosine deaminase transcripts resulted from Argulus species, adenosine deaminase and RNA-speci c adenosine deaminase-1. In the sand y Lutzomyia longipalpis [54] and the mosquito Aedes aegypti [74] adenosine deaminase can assist feeding and modulate host immune responses by metabolize the purine 2-deoxyadenosine and adenosine to 2deoxyinosine and inosine, correspondingly. The mechanism of immune evasion of salivary adenosine deaminase of Lu. longipalpis and Ae. Aegypti, is regulated by the removal of adenosine, a molecule associated with pain perception and mast cell degranulation. Then, the resulted product, inosine, functions as immunomodulator by inhibiting the production of pro in ammatory cytokines [53].
The exact role of the cysteine protease transcript found here in all three Argulus species is not known but it was suggested to have roles in nutrition and / or host immune evasion in the nematode H. contortus (Rudolphi, 1803) Cobb, 1898 [38]. It was also suggested to have a critical function in the infection process of the trophont parasitic stages of the ciliate, Ichthyophthirius multi liis and blood feeding parasitic copepod Their roles include catalysing turnover of parasite peptides and growth factors and regulation of proteins, host tissue penetration and invasion, digestion and modulation of host defence systems [75,76]. Phospholipases were also represented in the shared Argulus clusters, with 14 different clusters recognised in Swiss-Prot homology blasting. One of the most notable of these is phospholipase A2 (PLA2), a platelet-activating factor acetyl hydrolase, which is known to deactivate platelet-activating factor by Leishmania parasites [42] and known to facilitate blood feeding [77]. Transcripts for two types of ferritin were also found in Argulus in the present study, Ferritin-1 heavy chain and yolk ferritin, possibly involved in processing and storing iron after blood meals. Recently, ferritin was reported up-regulated in the salivary gland of the mosquito A. culicifaciens, after blood feeding [40], which prevents iron overload by assisting in iron homeostasis through iron transport and storage [78]. Hajdusek et al., working on Ixodes ricinus and Rhipicephalus annulatus (Say, 1821), illustrated the success of using the secreted form of ferritin 2 protein (FER2) as an antigen in an anti-tick vaccine [79]. Transcripts for enolase B, another protein suggested to be involved in blood digestion / feeding and defensive activity in ticks and other blood feeing parasites [80], were also found in Argulus spp. here as well as in sea lice, L. salmonis [70]. Verónica Díaz-Martín et al [81], reported that, enolase protein collected from the argasid tick Ornithodoros moubata saliva, known to bind with plasminogen, and hence may play a role in tick feeding by maintain blood uidity. Investigations for both the cestode Echinococcus granulosus, (Batsch, 1786) [82] and the nematode Ascaris suum, (Goeze, 1782) have suggested enolase to be a promising vaccine candidate [41,82].
Future work is needed to determine whether the identi ed compounds are secreted / expressed from Argulus feeding glands (i.e. pre-oral spine and proboscis glands), which we have recently characterised using histology and transmission electron microscopy [18]. This study substantially extends genomic resources for Argulus and is the rst study to explore shared genes expressed across species, providing a preliminary list of potential feeding enhancers / immunomodulators / defensins, thus potential broad scope vaccine / drug targets.

Conclusion
In the absence of a fully annotated Argulus genome, and in view of the previously limited genomic resources available for argulids and branchiurans more widely, this study has made a key contribution to argulid genomics and sheds further light on crustacean sh parasites more widely. High-throughput sequencing technologies have allowed the discovery of a number of conserved proteins across the species studied and have shed light on possible immunomodulator targets. The data produced from this study for A. foliaceus and A. coregoni adds considerably to that available from previous transcriptome studies.
The current study has provided evidence, for the rst time to our knowledge, of potential effectors of host immunomodulation, potentially providing future studies with targets for Argulus drug or vaccine development. There are 145 described species of Argulus worldwide [83] and it is therefore useful to nd conserved secretory proteins that may be used for control across different Argulus species. Thus, a universal vaccine or drug treatment strategy could be applied for different Argulus species, as has been previously suggested with respect to the development of a universal vaccine for ticks [84]. The rst step for developing any new control treatment is to search for potential targets [85] and the outcomes of the research described in this study provide important initial steps in this direction.

Declarations Funding
This work was generously supported by Sultan Qaboos University, Oman and by the Schlumberger Foundation Faculty for the Future programme.

Availability of data and materials
The sequence data for the transcriptome were submitted to the EBI (https://www.ebi.ac.uk/ena/browser/view/PRJEB34947) with a project number PRJEB34947 under the title "De novo transcriptome sequencing of branchiuran sh lice, Argulus foliaceus (Linnaeus, 1758) and Argulus coregoni Thorell, 1865").
Ethics approval and consent to participate The project was ethically approved by the University of Stirling Ethics Committee (Animal Welfare and Ethics Review Board; AWERB).

Consent for publication
Not applicable.