Phosphoproteomic analyses of kidneys of Atlantic salmon infected with Aeromonas salmonicida

Aeromonas salmonicida (A. salmonicida) is a pathogenic bacterium that causes furunculosis and poses a significant global risk, particularly in economic activities such as Atlantic salmon (Salmo salar) farming. In a previous study, we identified proteins that are significantly upregulated in kidneys of Atlantic salmon challenged with A. salmonicida. Phosphoproteomic analyses were conducted to further clarify the dynamic changes in protein phosphorylation patterns triggered by bacterial infection. To our knowledge, this is the first study to characterize phosphorylation events in proteins from A. salmonicida-infected Atlantic salmon. Overall, we identified over 5635 phosphorylation sites in 3112 proteins, and 1502 up-regulated and 77 down-regulated proteins quantified as a 1.5-fold or greater change relative to control levels. Based on the combined data from proteomic and motif analyses, we hypothesize that five prospective novel kinases (VRK3, GAK, HCK, PKCδ and RSK6) with common functions in inflammatory processes and cellular pathways to regulate apoptosis and the cytoskeleton could serve as potential biomarkers against bacterial propagation in fish. Data from STRING-based functional network analyses indicate that fga is the most central protein. Our collective findings provide new insights into protein phosphorylation patterns, which may serve as effective indicators of A. salmonicida infection in Atlantic salmon.

protein extraction and trypsin digestion. Total protein extraction and purification procedures were performed according to previously described methods 6 . Fish kidney samples were initially transferred to a 5 mL centrifuge tube and sonicated three times on ice using a high intensity ultrasonic processor (Scientz, Ningbo, China) in lysis buffer (8 M urea, 2 mM EDTA, 10 mM DTT and 1% Protease Inhibitor Cocktail). The remaining debris was removed by centrifugation at 20,000 g at 4 °C for 10 min and the supernatant transferred to new tubes. The protein concentration was determined with a 2-D Quant kit (GE, Boston, America) according to the manufacturer's instructions.
Protein solution was reduced with 10 mM dithiothreitol (DTT) for 1 h at 37 °C and alkylated with 20 mM iodoacetamide for 45 min at room temperature in the dark. For trypsin digestion, the protein sample was diluted by adding 100 mM TEAB to urea at a concentration less than 2 M. Finally, trypsin was added at a trypsin-to-protein mass ratio of 1:50 for the first digestion overnight and 1:100 for a second 4 h digestion. Approximately 50 μg protein for each sample was digested with trypsin for subsequent experiments. Acquired peptides were dissolved in solvent A (0.1% FA in 2% ACN) and directly loaded onto a reversed-phase analytical column (Thermo, USA). itRAQ labeling of proteins for proteomic data analysis. The peptide mixture obtained was labeled using the iTRAQ Reagent-8plex Multiplex Kit (AB SCIEX, USA) according to the manufacturer's protocol. The experimental groups, each including three biological replicates, were respectively labeled as 113, C-7 d, 114, C-14 d, 115, L-I-7 d, 116, H-I-7 d, 117, L-I-14 d and 118, H-I-14 (Control: C, Low-infected group: L-I, High-infected group: H-I). All labeled samples were separated with a gradient of 2-60% acetonitrile in 10 mM ammonium bicarbonate, pH 10, over 80 min into 12 fractions with an equal volumes collected via HPLC (Thermo DINOEX Ultimate 3000 BioRS, USA). Fractionated samples were analyzed via liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) performed on an AB SCIEX nanoLC-MS/MS (Thermo Q exactive Orbitrap, USA) system. Subsequent procedures were performed according to previous reports 6 .
The original MS/MS file data were submitted to MaxQuant with the integrated Andromeda search engine (v.1.5.2.8) for data analysis. Raw data files were processed and quantified using Maxquant v1.5.7.4 (21) and searched against the Atlantic salmon database in the NCBI website (97,714 items, updated on December 2015) concatenated with the reverse decoy database. Trypsin/P was specified as the cleavage enzyme allowing up to 2 missing cleavages. Mass error was set to 10 ppm for precursor ions and 0.02 Da for fragment ions. Carbamidomethylation on Cys was specified as fixed modification and oxidation on Met. False discovery rate (FDR) thresholds for protein, peptide and modification sites were specified as 1%. Minimum peptide length was set at 7. We set a 1.5-fold change (which reduced the false positive to a better extent than 2.0-fold change) and p-value < 0.05 as the threshold to distinguish significant changes in data. tandem mass tag (tMt) labeling of proteins for phosphoproteomic data analysis. To identify the protein phosphorylation events in kidney of Atlantic salmon, we applied multiplexed TMT and LC/LC-MS/ MS approaches to quantify the phosphoproteomes of healthy fish and those infected with A. salmonicida for 7 d with biological replicates. The three biological replicates were labeled as 126, K-C-1, 127, K-C-2, 128, K-C-3, 129, K-I-2 and 130, K-I-3 (Control: C, High-infected group: I). TMT-6 plex was selected for quantification. Briefly, one unit of TMT reagent (defined as the amount of reagent required to label 100 μg protein) was thawed and reconstituted in 24 μL ACN. Peptide mixtures were incubated for 2 h at room temperature, pooled, desalted and dried via vacuum centrifugation. The sample was fractionated using high pH reverse-phase HPLC on an Agilent 300 Extend C18 column (5 μm particles, 4.6 mm ID, 250 mm length). The method used was the same as that described in section 2.4. After combining into 8 overall fractions, peptides were dried via vacuum centrifugation.
Peptide mixtures were initially incubated with an IMAC microsphere suspension with vibration. IMAC microspheres with enriched phosphopeptides were collected via centrifugation and the supernatant removed 11 . To eliminate nonspecifically adsorbed peptides, IMAC microspheres were sequentially washed with 50% ACN/6% TFA and 30% ACN/0.1% TFA. To elute enriched phosphopeptides from IMAC microspheres, elution buffer containing 10% NH 4 OH was added and elution performed with vibration. The supernatant containing phosphopeptides was collected and lyophilized for LC-MS/MS analysis.
The resulting MS/MS data on the phosphoproteome were processed using the same search engine as that for proteomic data analysis, except for the processes of phosphorylation at Ser, Thr, Tyr and acetylation at the protein N-terminus, which were specified as variable modifications. The site localization probability set as >0.75. proteome and phosphoproteome analyses of kidneys of atlantic salmon infected with A. salmonicida. Gene ontology (GO) analysis was performed using STRING version 10.5 (https://string-db.org) 12 .
Proteins and phosphoproteins were classified as regulated based on inclusion of at least one regulated phosphosite. The Encyclopedia of Genes and Genomes (KEGG) database was used to identify enriched pathways. For each category, a two-tailed Fisher's exact test was employed to establish enrichment of differentially expressed against all identified proteins. Correction for multiple hypothesis testing was carried out using standard false discovery rate control methods. GO and pathways with corrected p-values < 0.05 were considered significant. All the pathways identified were classified into hierarchical categories according to the KEGG website.

Network analysis.
Proteins were filtered if they were not sampled in at least two of the replicates. All common protein and phosphoprotein candidates were searched against STRING database version 10.5 for protein-protein interactions. We followed these initial filtering steps with STRING analysis 13 . Only interactions between proteins belonging to the searched dataset were selected, thereby excluding external candidates. STRING defines a metric designated "confidence score" to classify interaction confidence. We fetched all interactions with confidence scores ≥0.7, representing high confidence. The interaction network from STRING was visualized in Cytoscape, a graph-theoretical clustering algorithm.
Identification of over-represented kinase substrate motifs. The software tool Motif-x was used to analyze the models of sequences constituted with amino acids in specific positions of modify-21-mers (10 residues upstream and downstream of the site) in all protein sequences 14 . All database protein sequences were used as the background parameter, with other parameters set as default. To further determine whether the identified phosphoproteins activate additional kinases, we investigated the influence of amino acids in the immediate vicinity of each phosphorylation site on phosphorylation of kinase substrates using the Group-based Prediction System (GPS) 15 .

Selection of fish infected with A. salmonicida at different doses and times.
We sampled fish at two stages, specifically, 7 d and 14 d after infection. The two infection times were selected to determine the potential proteins involved in the early and late stages of infection. Based on LD 50 (10 4 CFU/mL), we selected 10 7 and 10 4 CFU/mL as high and low infection doses for challenge. Kidney is the major hematopoiesis organ in fish that performs key roles in immune responsiveness and clearing bacteria from blood 16,17 . Accordingly, kidneys of normal and infected fish (both low and high doses) were isolated at 7 d and 14 d for iTRAQ analysis. We only determined the phosphorylation state of proteins in kidney samples challenged with the high bacterial dose (10 7 CFU/mL) at 7 d, with a view to improving analysis and identifying the molecules involved in the early stages of infection.
Multiplexed quantitative analysis of the whole proteome and phosphoproteome. In a previous study, we established a fish model infected with A. salmonicida 6 . MS-based iTRAQ quantitative proteomics in combination with phosphopeptide enrichment was employed to identify phosphopeptides and phosphoproteins and quantify their expression levels. In total, 5635 phosphorylation sites in 3112 proteins were detected in kidneys of fish, and 1502 and 77 proteins were quantitatively increased and decreased by ≥1.5-fold, respectively (Supplemental Table 1). For proteins possessing multiple phosphorylation sites, GO was used to analyze total phosphorylation in the high-infection group, compared to controls at 7 d. As a result, differentially phosphorylated proteins were sub-categorized into 26 hierarchically structured GO classifications, including 9 biological processes, 5 cellular components and 6 molecular functions ( Fig. 1A-C). In particular, "metabolic process" (17%) and "cellular process" (25%) were highly represented in "biological process", "cell" (37%) and "organelle" (27%) in "cellular component", and "catalytic activity" (20%) and "binding" (67%) in "molecular function" branches of ontology.
All phosphoproteins involved in biological processes, cellular components and molecular functions were differentially phosphorylated as log10 transformations relative to their average Con expression levels during infection in fish kidney (Fig. 1D,E). Most of the upregulated proteins were associated with myosin complex, motor activity, Ras GTPase binding, GTPase binding, small GTPase binding and regulation of GTPase activity while downregulated proteins were predominantly linked with hydrolase, nucleic acid binding and pyrophosphatase activity.  To further determine the differentially phosphorylated and coordinative proteins, we selected significantly altered phosphoproteins with different trends in expression at 7 d (listed in Table 1). Overall, 8 and 11 proteins were identified in the high-dose and low-dose infection groups at 7 d, respectively. Five of the proteins (keratin, type II cytoskeletal cochleal-like, prothrombin-like, membrane-associated progesterone receptor component 1, fibrinogen alpha chain-like, and apolipoprotein A-I precursor) were common between the two groups. Notably, only five (epoxide hydrolase 1, LOW QUALITY PROTEIN: neuroblast differentiation-associated protein AHNAK-like, matrix metalloproteinase-9 precursor, stanniocalcin and lymphocyte-specific protein 1-like) and three (keratin, type I cytoskeletal 18-like, phosphoenolpyruvate carboxykinase, cytosolic [GTP] and hematopoietic cell-specific Lyn substrate 1 isoform X1) proteins were specifically induced in the H-I-7 and L-I-7 treatment groups, respectively, which could potentially serve as indicators of resistance against A. salmonicida infection.

Motif analysis and regulation of protein kinases and phosphatases.
Application of motif-x to determine the amino acid sequence patterns resulted in the identification of 22 highly enriched motifs including Ser and Tyr among all the phosphoproteins (Fig. 4A,B). Proteins were assigned as typical kinases based on an online search of the kinase database (http://www.kinase.com/human/kinome/), mainly including members of STE (17), CAMK (20), AGC (19), CK1(4), CMGC (13), TK (22), TKL (15), RGC (1) and other families (Supplemental Table 2). Additionally, the motif-x tool revealed significant overrepresentation of eight motifs (SxDx, xGSx, RxSx, KxxS, RxxS, RxxSL, RKxS and LXRxxS) among the upregulated A. salmonicida-responsive phosphorylation sites. These clustered proteins are presented in Supplemental Fig. 1. Combined with the protein data presented in Table 1, five proteins, specifically, inactive serine/threonine-protein kinase VRK3-like, cyclin-G-associated kinase isoform X2, tyrosine-protein kinase HCK isoform X1, protein kinase C delta type-like and ribosomal protein S6 kinase alpha-3-like, were identified as potential biomarkers matched three motifs including SxDx, RxSx and KxxS respectively were shown in Table 2. SxDx corresponds to the consensus motif of inactive serine/threonine-protein kinase, VRK3-like, with Ser118 and Ser151 phosphorylation sites, and cyclin-G-associated kinase isoform, X2, with a Ser1049 phosphorylation site and RxSx to the consensus motif for tyrosine-protein kinase HCK isoform X1 with a Ser236 phosphorylation site. The activities of protein kinase C delta type-like with Ser658, Ser297 and Ser344 phosphorylation sites and ribosomal protein S6 kinase alpha-3-like with Ser363 and Ser221 phosphorylation sites are regulated by the KxxS motif. The phosphorylation status of these kinase proteins may serve as an effective biomarker of A. salmonicida infection in Atlantic salmon.

Network of phosphoproteins.
Using the STRING database, interactions among 290 phosphoproteins from both the proteome and phosphoproteome were identified, including 111 increased and 2 decreased at the phosphorylation level, with filter criteria of log2 ratios > 2 representing increased proteins and log2 ratios < 2 representing decreased proteins (Supplemental Fig. 2). We had filtered different significantly up or down regulated phosphoproteins with specific motifs in previous analyses. Details of these proteins are summarized in Supplemental Table 2 (Supporting Information). Combined with previous findings, significantly up or down regulated phosphoproteins from both low-infected and high-infected groups, compared to controls, at 7 d were selected (Fig. 5A,B), including fibrinogen alpha chain (fga), prothrombin-like (f2), apolipoprotein A-I precursor (apoa1a), ribosomal protein S6 kinase alpha-3-like (rps6ka3a), hematopoietic cell-specific Lyn substrate 1 isoform X1 (hcls1), membrane-associated progesterone receptor component 1 (pgrmc1) and trifunctional enzyme subunit beta, mitochondrial (HADHB). Among the proteins identified, only fga clustered within the network acted as a central rod.

Discussion
In the current investigation, quantitative proteomics was applied to analyze the phosphoproteome in Atlantic salmon infected with A. salmonicida. To our knowledge, this is the first study to characterize the phosphorylation events in infected Atlantic salmon. Based on proteomic data obtained from our previous experiments 6 in combination with phosphoproteomic data, we identified four proteins (epoxide hydrolase 1, neuroblast differentiation-associated protein AHNAK-like, matrix metalloproteinase-9 precursor, stanniocalcin and lymphocyte-specific protein 1-like) that were significantly altered following infection.  The soluble epoxide hydrolase (sEH), a ubiquitous enzyme in vertebrates that transforms epoxides to their corresponding diols implicated in the inflammatory cascade, plays broad biological roles by virtue of its activity against lipid epoxide substrates and localization in not only liver but also kidneys and vascular tissue [18][19][20][21] . sEH is a bifunctional enzyme with C-terminal hydrolase activity and N-terminal phosphatase activity 22     Matrix metalloproteinase-9 (MMP-9), one of the members of the MMP family, functions in normal and neoplastic invasive processes, cleaves numerous structural components and participates in inflammatory pathways 24,25 . MMP-9 cleaves a diverse range of substrates, such as extracellular matrix (ECM) proteins, ligands and receptors, and is mainly produced by inflammatory cells, including macrophages, leukocytes and monocytes. The protein plays important roles in embryogenesis, innate immune defense and apoptosis 26,27 . In earlier studies, expression of MMP-9 could be induced by bacterial pathogens in fish, suggesting its involvement in fish immune responses [28][29][30] . It is believed that these two proteins contribute to defense against A. salmonicida infection in Atlantic salmon, but further research is required to validate this hypothesis. Few studies have characterized the roles of neuroblast differentiation-associated protein, AHNAK-like and stanniocalcin and lymphocyte-specific protein 1-like during bacterial or viral infection in fish species such as Atlantic salmon. Using earlier results combined with integrated whole proteome and phosphoproteome data, we focused on the potential functions of these four proteins in subsequent experiments.
Motif-X analysis led to the identification of prospective novel kinases, including inactive serine/ threonine-protein kinase VRK3-like, cyclin-G-associated kinase isoform X2, tyrosine-protein kinase HCK isoform X1, protein kinase C delta type-like and ribosomal protein S6 kinase alpha-3-like, which are well-known key modulators and initiators of the global phosphorylation response to A. salmonicida infection in Atlantic salmon.
Two proteins containing the common SxDx motif were identified. Vaccinia-related kinase 3 (VRK3), a member of the novel VRK family is located in the nucleus 31,32 . VRK3 displays alterations in the critical motifs essential for kinase activity, and includes two active vertebrate paralogs (VRK1 and VRK2), one or two orthologs in all metazoans, and an ortholog in most poxviruses, including the founding member, vaccinia virus B1R 31,33-37 . The VRK3 structure may be maintained to provide a binding surface for other proteins, a known function of many pseudokinases 38 . VRK3 also shows putative kinase activity in a cellular context 39 . The protein is highly expressed during development and promotes cell cycle progression by phosphorylating the nuclear envelope protein barrier to autointegration factor (BAF) 40 . Moreover, VRK3 suppresses extracellular signal regulated kinase (ERK) activity through direct binding to the MAP kinase phosphatase (MKP), vaccinia H1-related (VHR) 41 . In our experiments, phosphorylation of VRK3 at Ser118 and Ser151 was increased significantly in A. salmonicida-infected fish at 7 d. However, the underlying mechanisms and specific functions of these phosphorylation events are yet to be defined.
Cyclin G-associated kinase (GAK) is a protein containing highly conserved serine/threonine kinases with multiple functional domains 42 . GAK was initially reported to be associated with cyclin G and shown to play an important role in uncoating clathrin-coated vesicles (CCVs) in non-neuronal cells 43 . Another study demonstrated a role of GAK in clathrin-mediated endocytosis/vesicle trafficking in the cytoplasm and nucleus 44 . In addition, GAK has been implicated as an androgen receptor-interacting transcriptional coactivator 45 . GAK interacts specifically with interleukin 12 receptor β2 (IL-12Rβ2) 46 that plays a central role in the initiation and control of cell-mediated immune responses to suppress IL-12-induced production of IFN-γ. Moreover, GAK is one of the two active kinases present in clathrin-coated vesicles and its Ser/Thr kinase activity is directed towards the l2 components of CCVs 47 . Based on integrated data from the current study and earlier experiments, we propose that GAK containing a phosphorylation site at Ser1049 has the potential to modulate immune responses in Atlantic salmon. The mechanisms linking GAK activity with immune pathways for development of a defense system against A. salmonicida in Atlantic salmon remain to be resolved.
Another protein, hematopoietic cell kinase (hck) containing the RxSx motif, is a member of the src family of tyrosine kinases and primarily expressed in myeloid cells. Expression of the gene is principally limited to cells of monocyte/ macrophage and granulocyte lineages [48][49][50][51] . Levels of the gene increase with differentiation along either the macrophage or granulocyte pathways [52][53][54] and are further augmented by activation stimuli, including LPS, in mature monocytes and macrophages 51,53,55 . A number of studies support a role of hck in the terminal differentiation of macrophages and/ or signal transduction pathways of mature monocytes, macrophages lymphocytes, mast cells, and basophils [56][57][58][59] . We observed increased phosphorylation of this protein only at the Ser236 site with bacterial treatment.
The two other selected proteins had a common KxxS motif. The first was protein kinase C delta (PKCδ), which is involved in cytoskeletal regulation and apoptosis as a member of the PKC family 60,61 . Proteolytic activation of PKCδ is considered responsible for apoptotic execution 62,63 . Previous studies have focused on the mechanisms underlying PKCδ involvement in regulating cellular apoptosis and the cytoskeleton 60,63,64 . Notably, we observed significantly increased phosphorylation of this protein at the Ser297, Ser344 and Ser658 sites.
The second protein, ribosomal S6 protein kinase (RSK), phosphorylates serum response factors in vitro at an in vivo serum-stimulated phosphorylation site 65 . The protein is also regulated by extracellular signal regulated kinase (ERK) to inhibit the proapoptic Bcl2-family member, BCL2, an antagonist of cell death and caspase 66 . In this study, phosphorylation of RSK at Ser221 and Ser363 was increased significantly in infected fish at 7 d.
In conclusion, the five prospective novel kinases identified in our analyses (VRK3, GAK, HCK, PKCδ and RSK6) potentially contribute to resistance against bacterial propagation in fish. These proteins have common functions in regulation of apoptosis and the cytoskeleton in inflammatory and cellular pathways. However, further research is required to clarify whether these proteins participate in the defense mechanism against microbial pathogens and the underlying mechanisms.
Our findings, in conjunction with previous data, show that fga is the most central protein in the pathogenic network and may serve a potential biomarker of A. salmonicida infection in Atlantic salmon. In keeping with this proposal, earlier proteomic profiling studies support the utility of fga as a diagnostic marker of lung squamous cell carcinoma 67 . We therefore hypothesize the protein with well-established key functions than other proteins obtained from STRING-based functional network analysis in fish against bacteria.
In summary, we have presented an integrated analysis of the whole proteome and phosphoproteome showing that specific proteins are overexpressed and collectively drive the immune response against high-dose infection of A. salmonicida in kidneys of Atlantic salmon at 7 d. These data provide a framework for further investigation of Atlantic salmon resistance to A. salmonicida, and selection of indicators to improve resistance.