Introduction

Teleost fish, the earliest evolutionary group with an immune system exerting both innate and adaptive immunity, is highly diverse, consisting of more than 32 000 species. The innate immune system in fish, like mammals, consists of a variety of molecules and immune cells that provide the first line of defense against microbial attack through recognition of potential pathogens. Recognition and degradation of microbes followed by induction of inflammation are essential processes for clearance of microbes and onset of adaptive immune responses. The innate immune system is triggered by complement factors, antibodies and/ or pattern recognition receptor (PRR) recognition of pathogen-associated molecular patterns (PAMPs) such as nucleic acid structures unique to bacteria and virus (CpG DNA, dsRNA), diverse proteins (flagellin), lipopolysaccharide, lipoteichoic acid and peptidoglycan.

While recognition of potential pathogens by complement factors and antibodies lead to increased phagocytic activity of host cells and degradation of invading microbes, recognition of PAMPs by PRRs ensures, through production of cytokines, that the elicited immune response is tailored to the invading pathogen. The major families of PRRs are the Toll-like receptors (TLRs), Nucleotide binding and oligomerization domain (NOD)- like receptors (NLRs), retinoic acid inducible gene I (RIG-I)- like receptors (RLRs), C-type lectin receptors (CLRs) and absent in melanoma 2 (AIM2)- like receptors (ALRs)1,2. In teleost fish, the TLRs is the most studied family of the PRRs and an enormous diversity has been identified in teleosts (reviewed in1,3,4). This diversity is suggested to be driven by adaptation to specific environments and host-intrinsic factors3. Teleosts possess orthologues to mammalian TLRs, with the exception of TLR6 and TLR10 which have not yet been identified in fish and the existence of a functional TLR4 in fish is subject to discussion. In addition, several TLRs are unique for the teleosteii i.e. TLR18-23, 25–284,5. Fish and amphibians also have a soluble version of TLR5, termed TLR5S6 in addition to a membrane bound TLR5 (TLR5M). TLR5 has been identified in all investigated teleost species, with the exception of the Paracanthopterygii7. From functional studies and functional inference based on sequence homology indicate that fish TLR1, TLR2, TLR5, TLR5S, TLR9, TLR21, TLR285,8,9 recognize bacterial ligands. In general, ligand binding initiates downstream cell signaling mediated via adapter proteins MyD88, MAL, TRIF, TRAM and SARM10, resulting in activation of transcription factors NFκB, IRF3/7, CREB and AP1, finally resulting in production of proinflammatory cytokines like TNFα, IL-12 IL-1β and IL-18 and/or interferons. There is currently little information regarding the downstream cell signaling pathways following activation of the fish-specific TLRs.

As for the TLR family, some NLRs also play a role in antimicrobial immune responses. NOD-like receptors are described in several species of fish including, but not exclusively, zebrafish, channel catfish, Japanese pufferfish and rainbow trout11,12,13,14,15,16. The NLRs described in fish are NOD1, NOD2, NLRC3, NLRC5, NLRCX and NLRC. Importantly, NLRC in fish is different from mammalian NLRC and as many as several hundred genes have been reported from one species13. There are, however, few functional studies of NLRs in fish and there is currently little knowledge of the downstream signaling after activation and how the receptors and signaling are regulated.

The transcriptome of lumpfish, as a representative for Cyclopteridae is highly valuable as this group is poorly characterized genetically and no reference genome or immune gene sequences are available in public databases. Also, it is not clear whether they belong to the suborder Cottoidei within the order Perciformes17 or within the order Scorpaniformes18. In addition to being interesting for comparative studies, mapping of the lumpfish immune system is important for basic immunological studies and for the rational design of immunoprophylactic measures for this species. In recent years, there has been a tremendous increase in the production of farmed lumpfish in Europe and Canada19, due to its ability to eat lice from farmed Atlantic salmon (Salmo salar L.)20. In Norway alone, the number of lumpfish farmed increased from 0.4 million in 2012 to approximately 15 million in 201621.

Large scale farmed lumpfish mortalities due to bacterial disease are reported22 and development of vaccines protecting against the most common pathogens is ongoing23. The level of total immunoglobulin M (IgM) in lumpfish sera is lower compared to species like salmon24,25, but it has been shown that lumpfish has the ability to produce specific antibodies upon immunization25 and that vaccination has an effect26. Previous studies have also shown that innate immune functions like phagocytosis and respiratory burst are efficient in lumpfish27 and that IgM+ B-cells display phagocytic ability25. More knowledge about the underlying mechanisms of the immune system of lumpfish at the individual gene level and their immune responses upon bacterial infection is required as this will form the basis for development of immunoprophylactic measures and immune stimulation. Therefore, to characterize the immediate and early induced innate response in this species, lumpfish leukocytes were exposed to the bacterium Vibrio anguillarum serotype O1, a known fish pathogen, for 6 and 24 hours, and RNA sequencing was performed followed by de novo transcriptome assembly and differential gene expression analysis.

Results

Illumina sequencing and de novo transcriptome assembly

Sequencing of RNA isolated from non-treated head kidney leukocytes (HKL) and HKL exposed to Vibrio resulted in 516 million reads. Reads of low quality, low complexity, containing adapter sequence, matching ribosomal or mitochondrial sequences were discarded. The resulting transcriptome consisted of 433 million assembled bases in 346,430 transcripts from 221,659 trinity genes. The median transcript length was 585 bases, mean length 1.25 kb and N50 of 2.5 kb. The RNA sequencing reads after trimming, the differential gene expression data and the assembled transcriptome are submitted to Array Express under accession number E-MTAB-6388.

Annotation of predicted proteins and functional annotation of the Trinity genes

Genes within the assembled transcriptome were annotated using Trinotate. Putative gene functions were identified by Gene ontology (GO) analysis. Of the 221,659 Trinity genes 37,895 were assigned minimum one Gene ontology (GO)-term. GO mapping resulted in 62 GO categories presented in Fig. 1. The GO-terms containing the highest number of genes were binding (23786), organelle (20995), cellular process (25011) and biological regulation (19578). The GO-term ‘immune system processes’ contained 2490 Trinity genes and includes genes involved in the development or function of the immune system e.g. immune response, leukocyte activation, activation of immune response and immune effector process (Fig. 1b). The most abundant immune system process was “innate immune response” which included 956 genes.

Figure 1
figure 1

Gene Ontology (GO) analyses of annotated genes in the lumpfish transcriptome. (a) The annotated genes were divided into the main GO-terms Biological processes, Molecular function and cellular components and further divided into subcategories. (b) Pie chart of the GO term distribution among the annotated genes in the lumpfish transcriptome in the GO term immune system process.

Full size image

Global differential gene expression (DEG) analysis upon bacterial exposure

To gain more information of the early induced innate immune responses in lumpfish, leukocytes were subjected to differential gene expression (DEG) analysis 6 and 24 hrs post bacterial exposure. Principal component analysis (Fig. 2a) revealed a major difference between exposed and non-exposed samples at both time points. This can be seen in the heat map following hierarchical clustering of the DEGs (Fig. 2b). The immune response was stronger and more extensive at 24 hours post exposure (hpe) (Fig. 2c) compared to 6 hpe (Fig. 2d). The number of statistically (p-value < 0.05) and biologically (p-value < 0.05 and fold change >4) significantly regulated genes was higher at 24 hpe compared to 6 hpe (Fig. 2e). The number of genes that were statistically differentially expressed at 24 hpe was 15225 genes (44%) compared to 9033 genes (26%) 6 hpe (Fig. 2e and f). As shown in the Venn diagram, 5389 (16%) genes were significantly differentially expressed at both time points (Fig. 2f).

Figure 2
figure 2

Differential gene expression (DEG) analysis 6 hrs and 24 hrs post bacterial exposure. (a) Principal component analysis. PC1 is time and PC2 is treatment. White circles are non-treated controls 6 hpe, black circles are treated samples 6 hpe, white squares are non-treated controls 24 hpe and black circles are treated samples 24 hpe. (b) Heatmap of transcriptome profiling data of non-treated controls versus bacterial exposed samples 6 and 24 hpe. (c) Volcano plot of DEGs 6 hpe. Significantly regulated genes are shown as black dots. Non-significantly regulated genes are shown as grey dots. (d) Volcano plot of DEGs 24 hpe. Significantly regulated genes are shown as black dots. Non-significantly regulated genes are shown as grey dots. (e) Percentage of DEGs that were significantly regulated (p-value < 0.05) at 6 hpe and 24 hpe are shown in black bars. Percentages of statistically significantly regulated (p-value < 0.05) DEG with an absolute log fold change >2. (f) Venn diagram showing the number of DEGs at the different time point. Only those that were statistically significant are shown. White = 6 hpe, black = 24 hpe and dark grey = genes that were significantly regulated at both time points.

Full size image

GO enrichment analysis showed that among the upregulated transcripts at 24 hpe, GO-terms with lowest p-value were; response to stimulus (log10 p-value −19.7), defense response (log10 p-value −18.9), response to stress (log10 p-value −17.1), positive regulation of immune system processes (log10 p-value −16.6) and regulation of intracellular signal transduction (log10 p-value −16.1) (Fig. 3a). Among downregulated transcripts at 24 hpe, the GO-terms with lowest p-value were; “small molecule biosynthetic process” (log10 p-value −11.3), “single-organism process” (log10 p-value −9.8), “response to interleukin 4” (log10 p-value −7.9), “cytokinesis” (log10 p-value −7.3) and “defense response“(log10 p-value −7.0) (Fig. 3b). For upregulated transcripts at 6 hpe, the GO-terms with lowest p-values were; “response to lipopolysaccharide” (log10 p-value −10.4), “inflammatory response” (log10 p-value −9.7), “response to biotic stimulus” (log10 p-value −9.4), “regulation of intracellular signal transduction” (log10 p-value −8.1) and “response to external stimulus” (log10 p-value −8.1) (Fig. 3c). For downregulated transcripts at 6 hpe, the p-values were not as low as at 24 hpe (Fig. 3d).

Figure 3
figure 3

Enrich GO-analysis 6 and 24 hours post bacterial exposure. Semantic plots of up and down regulated (log fold change >2 and p-value < 0.001) enriched GO terms at 6 and 24 hours post exposure, generated through REVIGO. Enrichment p-value is plotted in red, through yellow and green to blue; where blue is the smallest p-value and red the biggest p-value. Size of the circles correlates to the semantic size of the GO terms.

Full size image

Analyses of KEGG pathways belonging to the immune system were performed (Table 1). Several genes were identified for each KEGG ID, and thus, the number of lumpfish genes in DEG was higher than the number of KEGG IDs in DEG (Table 1). Further, the 20 most significantly regulated genes at 24 and 6 hpe (based on p-values) were identified (Supplemental Table 1). At 24 hours, the most significantly regulated gene was TLR5S, followed by interleukin 8 (IL-8) which is also known as neutrophil chemotactic factor and an uncharacterized protein. From blast search the uncharacterized protein likely belongs to the interleukin 6 (IL-6) family, most closely related to Leukemia Inhibitory factor (LIF) (Supplemental Table 1).

Table 1 Overview of identified lumpfish genes in immune system pathways*.
Full size table

The 50 most up- and down-regulated genes at each time point were identified (Supplemental Tables 2–5). Many of the upregulated immune genes at 24 hpe were cytokines such as IL-1β, IL-6, IL-8 and IL-17, or belonged to either the complement cascade (CFH, CFB, C8a, C8b and C5) or the TLR pathway (TLR5s). Other studies have shown that members of the NLR family of pattern recognition receptors also recognize bacterial antigens and regulation of genes encoding these receptors was investigated. The response of NOD1, NOD2 and other NLRs were very weakly regulated or non-responsive (data not shown). Since the most regulated genes belonged to the complement cascade and TLR signaling, these pathways were investigated at the individual gene level.

Complement cascade

The complement system can be activated by three biochemical pathways; the classical complement pathway, the alternative complement pathway and the lectin pathway. Many genes encoding complement proteins were identified in lumpfish (shown in Fig. 4a and listed in Table 1), including components such as C3, C6 and C7. The differential gene expression analyses showed that upon exposure to V. anguillarum complement factor responses were higher at 24 hpe compared to 6hpe (Fig. 4b). The most upregulated genes were the regulatory factors complement factor H (CFH) and complement factor B (CFB), complement components C5, vitronectin (VTN) and complement factors 8a and 8b. The latter are subunits of the membrane attack complex responsible for lysis of microbes. Also, complement factor P, which is a positive regulator for C3 and C5 convertases was also upregulated at 24 hpe. The most highly downregulated genes were complement C1q subcomponent subunit A (C1QA) and subunit C (C1QC) which are part of the classical pathway, in addition to complement components C2 (Fig. 4b). Lumpfish genes verified (by blast) as belonging to the complement cascade are given in Supplemental Table 6.

Figure 4
figure 4

An overview of the complement cascade in lumpfish (a) The molecules in the complement cascade identified in lumpfish are shown with red boxes, those that are not yet identified are shown in grey. The figure is modified from KEGG map0461063. (b) Differential gene expression analyses of members of the complement cascade 6 hrs and 24 hrs post exposure (hpe). Only those that are statistically significant regulated (p-value < 0.05) are shown. The color gradient represents highly upregulated (dark brown) to highly downregulated (dark blue) genes. The exact values are given for each gene. The genes are sorted by fold regulation at 24 hpe.

Full size image

TLRs and TLR signaling

The TLR family of signaling PRRs plays an essential role in the early innate immune response against both bacteria and viruses. In the lumpfish transcriptome, 13 TLRs were identified; TLR1, 2, 3, 5 M, 5 S, 7, 8, 9, 13, 14, 21, 22 and 28 (Fig. 5, Table 2). Activation of TLRs initiates intracellular signaling resulting in production of inflammatory cytokines and co-stimulatory molecules important in early pro-inflammatory responses, chemotaxis and activation of T cells. Many of the molecules involved in the TLR signaling pathway were identified in lumpfish (Fig. 6), including the adaptor proteins MyD88, TRIF (also known as TICAM1) and TIRAP (also known as MAL). TICAM 2 (TRAM) was not identified. All transcripts listed in Table 2 were annotated following a BLAST search against NCBI’s non-redundant database, for which the hit with highest total score is included in the Table. MyD88, TRIF and TIRAP were full-length, but for SARM only two short non-overlapping fragments were identified.

Figure 5
figure 5

Phylogenetic tree of TLRs. Full-length TLR sequences in public databases were included in the phylogenetic analyses. The TLRs are divided into families and subtypes. The TLRs identified in the lumpfish transcriptome is shown by red letters, including TLR1, −2, −3, −5 (membrane-bound and soluble), −7, −8, −9, −13, −14, −21, −22 and −28. The full-length name of the species and accession numbers of the sequences in the Figure is given in Supplemental Tables 7 and 8.

Full size image
Table 2 Verified TLRs in lumpfish and genes in TLR signaling pathway.
Full size table
Figure 6
figure 6

An overview of the Toll-like receptor signaling pathway in lumpfish (a) The molecules in the TLR signaling pathway identified in lumpfish are shown with red boxes, those that are not yet identified are shown in grey. The figure is modified from KEGG map0462063. (b) Differential gene expression analyses of members of the TLR pathway 6 hrs and 24 hrs post exposure (hpe). Only those that are statistically significant regulated (p-value < 0.05) are shown. The color gradient represents highly upregulated (dark brown) to highly downregulated (dark blue) genes. The exact values are given for each gene. The genes are sorted by fold regulation at 24 hpe.

Full size image

Members of the tumor necrosis factor receptor (TNFR)-associated factor (TRAF) family are important mediators of various signaling pathways, including the TLR signaling pathway. The TRAFs identified in lumpfish were TRAF2-6. Further, IRAK1, 3 and 4 were identified. Also, main components of the two downstream signaling routes, NF-κB signaling; NEMO, IKKA, IKKB, IKB, p50 and p65 and MAPK-signaling pathways; MKKs, ERK, JNK, p38, c-fos Jun were identified (Fig. 6 and Table 2). Components that were not mapped through batch mapping in the KEGG pathway database were searched for manually in the lumpfish transcriptome using synonyms or sequences from related species. Activation of NF-κB induces production of the pro-inflammatory cytokines, while activation of MAPK has impact on several immune functions including proliferation, differentiation, survival, apoptosis, chemoattraction and production of inflammatory mediators. TNFα, IL-1β, IL-6 and IL-12 were among the cytokines identified in the lumpfish transcriptome. Also, the chemokines IL-8 and MIP1 β (macrophage inflammatory protein, also known as CCL4) were identified. The genes most upregulated at both 6 hpe and 24 hpe included proinflammatory cytokines (IL-1β, IL-6, TNFα), a homologue of IL-17 (IL-17C1), IL-8 and the soluble form of TLR5 (TLR5S) (Supplemental Tables 2 and 4). Members of the NFκβ pathway, but not the MAPK pathway were upregulated (Fig. 6b). Interestingly, another IL-17 homologue (IL17A/F3) was one of the most down-regulated immune genes at 24 hpe (Supplemental Table 3). TLR13 (logFC −5.21) and TLR2 (logFC −2.74) were down-regulated at 24 hpe and 6 hpe, respectively.

Discussion

The innate immune system is of major importance for fish as aquatic vertebrates are generally more heavily exposed to pathogens than terrestrial vertebrates and the adaptive defenses are less efficient in aquatic vertebrates. The major humoral components essential for innate defense in vertebrates are antibodies and the complement system which tag and kill invading microbes and promote inflammatory responses28. Furthermore, conserved structures on potential pathogenic organisms such as flagellin are recognized by the host’s PRRs and trigger intracellular signaling pathways which results in production of inflammatory cytokines and initiation of adaptive immune responses tailored to the infecting agent.

To obtain information of the gene repertoire in lumpfish head kidney leukocytes and early anti-bacterial immune responses, leukocytes were exposed to the pathogenic bacterium V.anguillarum O1 and RNA was isolated 6 and 24 hpe. De novo transcriptome assembly and global differential gene expression revealed that the complement system and TLR signaling pathway were the most highly upregulated innate immune processes. Another family of PRRs involved in bacterial recognition is NLRs. In vivo challenge experiments in other teleost species have shown that expression of NOD1 and NOD2 is upregulated in several tissues after bacterial infection29,30. Lumpfish NLRs were either non-regulated or weakly downregulated. The functions and roles of NLRs in lumpfish upon bacterial infection should, therefore, be further explored. The ligand specificity of the expanded fish-specific NLRC family reported from several fish species is currently unknown and it will be exciting to elucidate their role and importance in fish immunity.

Several components of the complement cascade were identified within the lumpfish transcriptome as shown in Fig. 4. Most genes belonging to the classical and alternative pathway were identified, but not the mannose-binding lectin (MBL) involved in the lectin pathway. Of the complement receptors, CR1, CR3, CR4 and C5AR1 were identified, but not complement receptor CR2. This is similar to other fish species (summarized in28). In humans it is known that the complement system cross-talks with other pathways and modulates adaptive immune responses31,32. Information regarding cross-talk between pathways and involvement of B and T cells in fish are scarce, and discrimination of some components, in example C1r/C1s, requires functional analyses at the protein level.

In the lumpfish transcriptome, TLR1, −2, −3, −5 (membrane-bound and soluble), −7, −8, −9, −13, −14, −21, −22 and −28 were identified. All lumpfish TLR transcripts, with the exception of TLR22, encoded full-length sequences. Phylogenetic analyses (Fig. 5) show that lumpfish TLRs group together with the order Perciformes, most closely with orange-spotted grouper (Epinephelus coioides). While some teleost TLRs are orthologues of mammalian counterparts, equivalents to human TLR6 and TLR10 have not yet been found in fish. Many of the TLRs in fish are not present in mammals. These include TLR5S, −14, −18, −19, −20, −21, −22, −23, −24, −25, −26, −27 and −28, and of these, some are fish-specific (TLR18-23, 25–28). The soluble variant of TLR5 is widely present in teleosts and has been identified in several species such as rainbow trout33, catfish34,35, gilthead seabream36, flounder37 and orange spotted grouper38. Since V. anguillarum is a flagellated bacterium, it was not unexpected that TLR5S was highly upregulated during early immune responses. Actually, it was the most significantly regulated gene at 24 hpe and among the most significantly upregulated genes at 6 hpe (Supplemental Table 1). In lumpfish leukocytes, TLR5M was not significantly regulated either at 6 hpe or 24 hpe. This is similar to the situation in rainbow trout where expression of TLR5S, but not TLR5M, was induced by V.anguillarum and purified recombinant V.anguillarum flagellin33. Upregulation of TLR5S transcripts during bacterial exposure is reported in other fish species35,36. Humans do not have TLR5S, but the innate immune response to flagellin mediated by human TLR5M is similar to that of teleost fishes39. Interestingly, a study of Tsujita and colleagues showed that TLR5S from rainbow trout amplifies the human TLR5 response via physical binding to flagellin40. How TLR5S initiates downstream signaling is not yet known, but a hypothetical mechanism has been suggested in which TLR5S binds circulating flagellin and transports it to TLR5M. In this way danger signals are amplified in a similar manner to LPS recognition by human TLR4 and the soluble factors LBP and CD1441. It is known that activation of TLR5 in mammals results in activation of NF-kappa-B and production of proinflammatory cytokines. The DEG analyses of lumpfish leukocytes indicated that the NF-κB signaling pathway, not the MAPK signaling pathway, was activated, as inhibitors of both nuclear factor kappa-B kinase alpha (NFKBIA, also known as IκBα and IKKA) and NFκB were highly upregulated. DEG analysis showed that IKKA was upregulated at both 6 hpe and 24 hpe, while NFkB was most highly upregulated at 24 hpe. Gene expressions of transcripts involved in the MAPK signaling pathway, such as MP2K3 and MP2K6, showed little change (after 6 hpe) or were downregulated (after 24 hpe). It will be interesting to investigate whether regulation of the TLR5 signaling pathway is conserved, or whether teleosts have developed another regulatory mechanism than mammals.

The cytokines that were most differentially regulated were IL-1β, IL-8, IL-6, TNF-α, one of the IL-17A/F3 and IL17C1. All were highly upregulated, except IL17A/F3 which was barely differentially regulated at 6 hpe and highly downregulated at 24 hpe. IL-1β has diverse functions including being a major regulator of inflammatory processes. It is a chemoattractant for fish leukocytes, it stimulates chemokine production in cells following infection and is known to induce expression of TNF-α. Further, IL-1β also modulates differentiation of T helper 17 cells (Th17) and expression of IL-17 family members42,43. Th17 cells are a subset of activated CD4 + T cells and are known to play a role in mucosal immunity and tissue inflammation. In mice, in addition to IL-1β, IL-6 and the transcription factor RORγt, transforming growth factor β1 (TGF-β1) is required for differentiation of Th17 cells. In humans, RORγt and Th17 polarization was induced by IL-1β and enhanced by IL-6, but suppressed by TGF-β1 and IL-1244. Although the exact regulation of Th17 cells in fish is not yet understood, it is widely accepted that fish have Th17 cells as all the major components of mammalian Th17 cell development are present in fish, including Th17 driver cytokines (IL-6, TGF- β1, IL-21 and IL-23), transcription factor (RORγ) and effector cytokines (IL17A/F, IL-22)42,45. Since some of the Th17 components in fish have multiple isoforms, it has been suggested that an even more complex Th17 type responses and regulation are present in fish compared to mammals45.

The most highly regulated cytokine in lumpfish leukocytes following bacterial exposure belonged to the IL-17 family. IL-17A/F3 was highly downregulated 24 hpe, while one of the IL-17C proteins, IL-17C1, was the most upregulated transcript, at 6 hpe and 24 hpe. IL-17 cytokines are central mediators of inflammatory responses and have been functionally characterized in jawed and jawless vertebrates and in invertebrates such as molluscs, nematodes and arthropods46,47,48,49,50,51. Teleost fish have several IL-17 molecules including IL17A and IL-17F, termed IL17A/F1-3, IL-17B, IL17C and IL17D42,52. One IL-17 originally termed IL-17N53 is likely to represent a fourth IL-17A/F member. An IL17E equivalent has thus far not been identified in fish, but two IL17C genes have been reported in rainbow trout48 and Japanese pufferfish54. Two IL-17C-like genes were also identified in lumpfish, but no IL-17E. It has been suggested that an ancient IL17C may have diverged into IL-17C and IL-17E in early mammals, whereas two IL-17C genes can be present in teleosts. Although relatively few studies have reported bioactivity of the IL-17 molecules in fish, studies from different species suggest that while IL-17 proteins play a role in innate immunity, they may have evolved specialized roles. Recombinant IL-17A/F from grass carp and trout can increase expression of proinflammatory cytokines in isolated head kidney leukocytes55 and splenocytes56, respectively, while IL-17D in grass carp increase expression of IL-1β, IL-8, TNF-α but not IL-6 (reviewed in42).

In summary, our transcriptomic data suggests that the complement system recognized the pathogenic bacterium and activated subunits of the membrane attack complex (MAC) which is a prerequisite for formation of a MAC complex at the surface of the microbe and thereafter cell lysis. Also, complement receptors involved in phagocytosis, degranulation and chemotaxis were upregulated which is related to the need to recruit host phagocytic cells for clearance of the bacterium. One of the most highly upregulated genes was IL-8 which is a chemokine involved in chemotaxis and attraction of neutrophilic cells. Another immediate innate immune response essential to prevent infection is promotion of inflammation and production of cytokines that ensures the immune response is tailored to the infecting microbe. Our study suggests that TLR5S recognized flagellin and triggered downstream signaling through the NFk-B signaling pathway resulting in production of pro-inflammatory cytokines (IL-1β, TNFα, IL-6, IL-12 and IL-17). IL-12 is needed for activation of naïve T-cells and IL-17 induces production of chemokines. Our transcriptomic data adds valuable information about the immune responses in lumpfish during the early stages of a bacterial infection. Functional analysis of the proteins involved in the signaling pathways is however necessary to gain further insight into the role of specific proteins and the interaction between them.

The lumpfish transcriptome presented provides a valuable base for comparative and phylogenetic analyses as lumpfish is a representative of the infraorder Cottoidea, a phylogenetic group which is poorly characterized immunologically and genetically. Furthermore, the lumpfish is a novel and a very important species for aquaculture since it is used for sea-lice control in salmon farming19. Although production of lumpfish has generally been successful, there have been challenges with large-scale mortality due to bacterial infections22. Vaccines against selected lumpfish pathogens are in use23,26, but more knowledge of the lumpfish immune system and responses to bacterial exposure at the individual gene level is important. Thus, the identification of immune genes, transcriptome–wide mapping of signaling pathways and early immune responses presented here are highly valuable as they provide a basis for development of more efficient immune prophylactic measures and provide important tools for evaluation of the efficacy of different prophylactic measures.

Materials and Methods

The work in the presented manuscript was performed on cells isolated from dead fish. The fish were sacrificed with a sharp blow to the head which is an appropriate procedure under Norwegian law. All experiments were performed in accordance with relevant guidelines and regulations. Rearing of fish under normal, optimal conditions does not require ethical approval under Norwegian law (FOR 1996- 01- 15 no. 23)

Fish

Farmed lumpfish (C. lumpus L.) were provided from Fjord Forsk Sogn AS, a commercial breeder in Sogn & Fjordane County, Norway. The fish were kept in a 500 L tank at the Aquatic and Industrial Laboratory (ILAB) within the High-Technology Centre in Bergen under normal rearing conditions with a light regime 12 h light: 12 h dark. The water temperature was 8 °C, salinity 34 PSU and a minimum of 77% oxygen saturation in the outlet water. The fish were fed with the commercial dry feed Amber Neptune (1.5 mm).

Bacterial culture

Vibrio anguillarum serotype O1 (8752) isolated from moribund lumpfish after a disease-outbreak in 2012 in Møre & Romsdal county in Norway was cultured in tryptic soy broth containing 2% NaCl at 20 °C, 200 rpm until late log phase. The bacterium was washed once in PBS and re-suspended in L-15 + medium without antibiotics.

Isolation of leukocytes and in vitro bacterial exposure

Head kidney leukocytes were isolated as described previously using discontinuous Percoll gradients27. Both left and right kidney lobes from 15 fish were included. Cell number, viability and aggregation factor was determined using a CASY Cell Counter™ (Innovatis AG). For in vitro bacterial exposure, 5 × 106 cells in L-15 + medium without antibiotics were added to each well in a 24-well plate (Nunc) and mixed with the bacterium V. anguillarum O1 (MOI 1:10) in a total volume of 0.5 mL. In wells with non-exposed cells, medium was added instead of bacterial cells. The plates were incubated at 15 °C. After 1.5 hour, pencillin/streptomycin was added to each well and the plates were further incubated until 6 hrs and 24 hours post bacterial exposure. In order to obtain an as comprehensive transcriptome as possible, a sample with leukocytes exposed with infectious pancreatic necrosis virus for 24 hrs was also included. This sample was used for the de novo transcriptome assembly, but was not part of the DEG analysis. Following incubation, the plates were centrifuged for 10 min at 200 × g. The supernatants were removed and lysis buffer was added directly to the wells. The lysates were stored at −80 °C prior to RNA isolation.

Isolation of total RNA

Total RNA was isolated using GeneElute Mammalian Total RNA miniprep kit (Sigma) according to the manufacturer’s instructions. Samples were treated with DNase I (Sigma) to removed traces of genomic DNA and the concentration of total RNA determined in a Nanodrop®ND-1000 UV-Vis spectrophotometer (Nanodrop Technologies). Total RNA extracts from three-five fish were pooled, in total 5 µg per pooled sample. For each time point three parallels were prepared for RNA sequencing. The pooled RNA (5 µg) was cleaned using RNA clean & concentrator-5 (zymo research) according to the manufacturer’s instructions and the quality of the RNA were determined in an Agilent 2100 bioanalyzer. RNA isolated from virus infected leukocytes was kept separately. The RQI values were in the range 6.3–9.3.

Transcriptome sequencing, assembly and annotation

The Norwegian High Throughput Sequencing Centre prepared sequencing libraries using TruSeq™RNA sample Preparation kit (Illumina®) according to the manufacturer’s protocol and performed paired-end strand-specific sequencing on the Illumina HiSeq platform with a 125 bp read length, resulting in a total of 516 million reads. Read quality was first assessed using FastQC, and Trinity’s option for read trimming by quality was included during assembly (trimmomatic). Reads of low quality, low complexity, containing adapter sequence, matching ribosomal or mitochondrial sequences were discarded. Transcripts were assembled using Trinity v2.0.657 with read normalization enabled and library type specified, otherwise keeping default settings. Known contaminants (Vibrio and IPNV) were removed from the assembly using blast. During the analyses, other non-eukaryotic sequences were discovered and additionally removed from the expression value matrices, with a more generic contaminant removal procedure58. More information on all steps of the sequencing data processing is given in Supplemental methods. The resulting transcriptome consisted of 433 million assembled bases in 346,430 transcripts from 221,659 “genes”. The median transcript length was 585 bases, mean length 1.25 kb and N50 of 2.5 kb. Following assembly transcripts were annotated with BLAST matches, protein domains and GO terms using the Trinotate toolkit (https://trinotate.github.io).

Bioinformatical analyses

Gene ontology mapping was performed in J-express Gene expression analysis software. Detailed information about the gene included in each category was obtained using Quick GO, which is a fast browser for Gene Ontology terms and annotation (http://www.ebi.ac.uk/QuickGO/GTerm?id=GO:0006954#term=annotation). Verification of the annotation of the transcripts was performed with BLAST search (https://blast.ncbi.nlm.nih.gov/Blast.cgi), multiple sequence alignment (MSA) using PAGAN v.0.6151. The phylogenetic tree was constructed from MSA by maximum likelihood with IQ-TREE 1.5.459 using automatic model selection60 followed by 100,000 ultrafast bootstraps61. An overview of the species and accession numbers included in the phylogenetic analyses are given in Supplemental Tables 7 and 8, respectively. Pathway analyses were performed using KEGG61,62,63. KEGG pathways analysis64 was performed by annotating the transcripts using BLAST against KO genes in KEGG, downloaded 08.02.2017. Transcripts with a BLAST score of 300 and above against KO genes in KEGG were mapped to the KEGG pathways as described in the KEGG Mapper tool. Transcript abundances for three biological replicates for treatment and control at 6 and 24 hpe were estimated using RSEM as part of the Trinity pipeline (Supplementary results of Trinity RSEM). The read count estimates were used as a basis for differential expression analysis using the Limma R-package65. Only genes with at least 10 reads in at least three samples were considered for differential expression analysis (34280 of 221659 assembled genes). Fold changes between groups and adjusted p-values (BH correction for multiple testing) were exported for downstream analyses. The DEG analyses were visualized in Graph-Pad prism 5. GO enrichment was calculated using GO-seq.66 and visualized in REVIGO.

The datasets generated during the current study are available in Array Express repository.