Interferon type I and type II responses in an Atlantic salmon (Salmo salar) SHK-1 cell line by the salmon TRAITS/SGP microarray

S. A. M. Martin, J. B. Taggart, P. Seear, J. E. Bron, R. Talbot, A. J. Teale, G. E. Sweeney, B. Høyheim, D. F. Houlihan, D. R. Tocher, J. Zou, C. J. Secombes


Interferons (IFNs) are cytokines that have proinflammatory, antiviral, and immunomodulatory effects and play a central role during a host response to pathogens. The IFN family contains both type I and type II molecules. While there are a number of type I IFNs, there is only one type II IFN. Recently both type I and type II IFN genes have been cloned in salmonid fish and recombinant proteins produced showing IFN activity. We have stimulated an Atlantic salmon cell line (SHK-1) with both type I and type II recombinant salmonid IFNs and analyzed the transcriptional response by microarray analysis. Cells were exposed to recombinant IFNs for 6 or 24 h or left unexposed as controls. RNA was hybridized to an Atlantic salmon cDNA microarray (salmon 17K feature TRAITS/SGP array) in order to assess differential gene expression in response to IFN exposure. For IFN I and II, 47 and 72 genes were stimulated, respectively; most genes were stimulated by a single IFN type, but some were affected by both IFNs, indicating coregulation of the IFN response in fish. Real-time PCR analysis was employed to confirm the microarray results for selected differentially expressed genes in both a cell line and primary leukocyte cultures.

  • transcriptome
  • immune
  • fish

interferons (IFNs) are cytokines that have key roles in the regulation of both innate and adaptive immune responses. They exhibit a diverse range of activities including antiviral, antitumor, and immunomodulatory roles (4). There are three types of IFN known in mammals (I–III); however, to date only types I and II have been identified in fish. Type I consists of a large number of related proteins including the α-and β-IFNs, which function principally in the antiviral response (55). Type II IFN is a single protein (IFN-γ) that originally was described for its macrophage-activating activity (59). The two families of IFN molecules are not structurally related but share similarities in their receptors and mechanisms of gene activation (41, 55). Type I and II IFNs produce their effect via different heterodimer cell surface receptors. Once IFN is bound to the receptor, signal transduction leads to activation of IFN-responsive genes via conserved signal transduction pathways leading to change of function of the cell (53). It is estimated in mammals that up to 200 or more genes may be affected by IFN stimulation (13). IFN-γ signal transduction is primarily via the Janus-activated kinase (JAK)-signal transducer and activator of transcription (STAT) pathway. Here JAK1 associates with JAK2 on stimulation and causes the phosphorylation of STAT1. Two STAT1 molecules dimerize to form a homodimer, which then binds to specific consensus sequences in the promoter of IFN-γ-responsive genes at specific γ-IFN activation site (GAS) elements, resulting in initiation of transcription. Type I IFN molecules follow a similar pathway, but there are two different kinases that are associated with the receptor. In this case JAK1 and tyrosine kinase 2 (TYK2) are activated and phosphorylate both STAT1 and STAT2, which form a heterodimer that in combination with interferon regulatory factor 9 forms a trimer that binds to interferon-stimulated response elements (IRSE) and subsequently induces transcription (reviewed in Refs. 41, 55). In mammals IFN type I can stimulate genes via both IRSE and GAS elements in the promoter, whereas IFN-γ does not usually act via IRSE (41, 55), although many IFN-responsive genes have both IRSE and GAS sites within their promoters.

The dramatic increase in genome information available for teleost fish, as a result of the whole genome sequencing of several fish species including zebrafish (42), fugu (1), tetraodon (21), medaka, and stickleback (current status reviewed in Ref. 15), as well as large expressed sequence tag (EST) data sets for nonmodel fish such as Atlantic salmon (47), has resulted in considerable gene discovery in the last few years. An example is the identification in fish of many cytokines that help coordinate immune function, such as the proinflammatory cytokines IL-1β (62), IL-6 (3), IL-8 (52), IL-11 (58), transforming growth factors (19), lymphotoxin β (23), and a large number of chemokines (27). In addition both type I and type II IFN molecules have been cloned in a variety of fish species including fugu (64), Atlantic salmon (49), rainbow trout (61, 63), and channel catfish (33). The type I IFN is not clearly an α- or β-IFN molecule but has clear similarity with the type I molecules (48). In addition to the IFN molecules themselves, many of the IFN regulatory factors involved in IFN signaling and regulation have been identified in fish (7), and currently the mechanisms of signaling are being examined in lower vertebrates (8).

Microarrays have been used successfully in fish for large-scale gene expression studies, in particular for examining responses to environmental variables (17), the immune response to viral pathogens (6), bacterial pathogens (14, 46), and parasitic infections of fish, such as amoebic gill disease (34). Transcriptome analysis has also been used to study the response of fish to vaccination (31, 43), stimulation by lipopolysaccharides (29), stress (25), and cytokine stimulation in trout (30). Although the existing available salmonid microarrays (57) are widely used valuable tools, a novel 16,950 (termed 17K) cDNA microarray generated from different EST clones and supplemented with additional immune-responsive genes by subtractive cloning is described here and may be most appropriate for use in our experiments focused on IFN-induced effects in fish. This array was employed to investigate transcriptomic changes in an Atlantic salmon cell line (SHK-1) exposed to rainbow trout recombinant (r)IFN type I and type II proteins.


TRAITS/SGP cDNA microarray construction.

A cDNA microarray was constructed by the Salmon TRAITS consortium ( funded through the Biotechnology and Biological Sciences Research Council (BBSRC) Exploiting Genomics Initiative in collaboration with the Norwegian Salmon Genome Project funded by the Research Council of Norway. Features were mainly sourced from two existing Atlantic salmon EST collections: the European Union-funded SALGENE project and the Norwegian Salmon Genome Project (SGP; These cDNA libraries (both standard and normalized forms) were constructed with RNA extracted from various tissues of presumed healthy farmed fish fed on normal commercial diets. Both freshwater and seawater life history types were represented. Transcripts were characterized predominantly by 5′ single-pass sequencing, in order to maximize functional annotation. The SALGENE and SGP resources were supplemented with subtracted clones generated by the three UK-based TRAITS partners with suppression subtractive hybridization (SSH) technology (Clontech PCR-Select cDNA subtraction kit). Individual subtracted libraries targeted differential expression resulting from bacterial disease challenge, variation in dietary lipid, shortterm starvation, and “smoltification,” the process of transformation from freshwater parr to seawater-adapted smolt.

Contig assembly of sequence data from the ∼55K available clones, together with ∼57,000 additional Atlantic salmon cDNA sequences available in GenBank (June 2004), was performed with the TGI clustering tools (39), as summarized in Fig. 1. For the ∼9,000 contigs identified, a single clone representative was chosen. Only two criteria were used in clone selection: 1) a standard library clone was preferred to a SSH-derived clone, and 2) a SALGENE clone was selected ahead of an SGP clone (purely for logistical reasons, all SALGENE clones were already curated on site). Additional candidate genes were added that were not found to be present in the available data set. These specific clones were associated with key functions/pathways already being investigated by TRAITS partners. Inserts were PCR amplified from the selected clones with appropriate vector-specific primers. After a quality check (by separation on a 1% agarose gel) and quantification (Pico Green assay, Invitrogen) 16,950 cDNA inserts remained available for printing (Fig. 1). These features were derived from 15 different tissue sources (Table 1), with ∼9% being the result of SSH enrichment (Table 2). Homology searches revealed significant hits (e-value ≤ e−10) for 56% (BLASTN; nr nucleotide database, Dec 2006) and 52% (BLASTX; nr protein database, Dec 2006) of cDNA inserts.

Fig. 1.

Schematic of cDNA feature selection for the TRAITS microarray (see text for further details). EST, expressed sequence tag.

View this table:
Table 1.

Tissue and clone source for cDNA features on TRAITS array

View this table:
Table 2.

Numbers of clones derived from subtractive cloning included on TRAITS microarray

Features were printed onto amino-silane-coated glass slides (Corning GAPS II) with a MicroGrid II printer (Genomic Solutions). DNA was resuspended in printing buffer (150 mM sodium phosphate buffer, 0.1% SDS, pH 8) to a concentration of ∼150 ng/ml and spotted with a 48-pin tool (Biorobotics 10K pins). Mean spot diameter was 110 μm. The slide format comprised 48 blocks, each block consisting of 27 columns and 28 rows. Each cDNA was printed in duplicate, with duplicate features always printed within the same block. A number of control DNAs were printed across the array: sonicated Atlantic salmon genomic DNA (96 spots); sheared salmon sperm DNA (Oncorhynchus derived, Sigma; 96 spots); and five different SpotReport (Stratagene) controls, i.e., PCR products 1–3 (Cab, RCA, rbcL genes from Arabidopsis thaliana), human β-actin, and Poly(dA)40–60 oligonucleotide, 20 spots each. In addition, each block had two Cy3 spots located at the upper left corner to aid orientation of the slide during scanning. After printing, DNA spots were fixed by baking at 80°C for 2 h. Before hybridization microarray slides were treated with succinic anhydride and 1-methyl-2-pyrrolidinone (Sigma) to block unbound amino groups [slide manufacturer's (Corning) recommended protocol] and denatured by washing in 95°C MilliQ water for 2 min. Gene Ontology (GO) identifiers were obtained by Blast2GO (10) and manually with the Gene Ontology Consortium facility EBI (2). Further details of the TRAITS/SGP microarray can be found on the ArrayExpress platform ( under accession number A-MEXP-664. Slides can be obtained from ARK Genomics (

Cell culture and exposure to recombinant cytokines.

An Atlantic salmon cell line derived from macrophage-enriched head kidney cells (SHK-1) (11) was used as the cell line of choice in this experiment. The cells were maintained in Leibovitz medium (L-15 medium) containing 10% FCS (Labtech International) and antibiotics (100 μg/ml penicillin and 100 U/ml streptomycin) at 20°C (14) in 80-cm2 flasks. Cells were passaged to fresh flasks at 80% confluence and cultured for 2 days before stimulation in the presence of 10% FCS. The SHK-1 cells were exposed to rainbow trout rIFN type I (63) or rIFN-γ (61) at a dose of 20 ng/ml (sequence accession nos. AJ582754 and AJ616215). This dose was used because no significant additional differences in gene expression responses were observed with doses of between 10 and 100 ng/ml compared with unstimulated cells in previous studies (61, 63). Cells were exposed for both 6 and 24 h, and these times were chosen because IFN responses can occur within 6 h for some genes such as γ-IP, whereas other responses have not been observed until 24 h after exposure (61). Thus, using these times, we anticipated identifying both early and late IFN-responding genes. To initiate the exposure to recombinant cytokines the culture medium was removed from the culture flasks and new medium containing the recombinant proteins was added to the flasks. For control cells the medium was changed but no recombinant proteins were added. For each time point, for both IFN exposures, six replicate flasks of cells were used, with six control unexposed flasks also used at both time points. Before microarray analysis PCR was used to confirm that the stimulated cells had responded to the rIFN molecules. IFN type I increased expression of Mx mRNA and IFN-γ increased γ-IP compared with unstimulated cells (data not shown).

Primary cultures of Atlantic salmon macrophage cells were prepared under sterile conditions from the head kidney. Kidney tissue was gently pushed through a 100-μm nylon mesh (J. Staniar) with ice-cold L-15 medium (Invitrogen Life Technologies) containing 10 U/ml heparin (Sigma-Aldrich). The culture was enriched for macrophages by separation over a 51% Percoll gradient as described by Secombes (54). After being washed with L-15 medium, cells were resuspended in L-15 medium containing antibiotics (100 μg/ml penicillin and 100 U/ml streptomycin). The primary cell culture was left for 24 h before being stimulated with or without rIFNs as described above.

RNA isolation.

SHK-1 and primary head kidney macrophage cells were washed with HBSS before being lysed directly in RNA STAT60 (AMS Biotechnology) according to the manufacturer's instructions for purification of total RNA. Salmon tissues (liver, gill, and kidney) used in microarray testing were homogenized in TRIzol (Invitrogen), following the manufacturer's instructions. After precipitation, RNA was resuspended in diethyl pyrocarbonate (DEPC)-treated water. Further RNA purification was performed with RNeasy cleanup columns (Qiagen) according to the manufacturer's protocol. RNA was eluted from these columns in water, and the concentration was adjusted to 2.5 μg/μl. The integrity of the RNA was determined by electrophoresis (Agilent Bioanalyser 2100), and concentration was measured by spectrophotometry (ND-1000, NanoDrop Technologies). RNA was stored at −80°C until required.

Microarray hybridizations.

RNA was reverse transcribed and labeled with either Cy3 or Cy5, using the Fairplay II cDNA labeling kit (Stratagene) according to the manufacturer's instructions. For array analysis six replicates were used for each exposure time point and six control replicates at both 6 and 24 h. Each RNA sample was labeled with both Cy3 and Cy5 to allow for a full dye-swap protocol, giving a total of 48 hybridizations. All reverse transcriptions were carried out at the same time on 96-well plates to reduce technical variation. Briefly, 20 μg of total RNA was reverse transcribed after being primed with oligo(dT). After reverse transcription the RNA template was hydrolyzed with 1 M NaOH for 15 min and then neutralized with 1 M HCl. The cDNA was ethanol precipitated overnight. The cDNA pellets were washed in 80% ethanol and air dried before being resuspended in 5 μl of 2× coupling buffer (Stratagene Fairplay Kit). Once the cDNA had fully dissolved (after at least 30 min) 5 μl of Cy dye was added to each tube and incubated in the dark for 30 min. Prealiquoted Cy3 and Cy5 dyes (GE HealthCare; PA23001, PA25001) were resuspended in 45 μl of DMSO before being added to the coupling buffer. To remove unincorporated dye, the labeled cDNA (total volume 10 μl) was passed through a DyeEx 2.0 spin column (Qiagen). Dye incorporation was checked by spectrophotometry (ND-1000) and by electrophoresis of labeled cDNA on a minigel and visualization by microarray scanner (Perkin Elmer ScanArray 5000XL). For hybridization, 20 μl of labeled sample (10 μl Cy3 and 10 μl Cy5) was added to 85 μl of hybridization buffer (Ambion), 10 μl of poly(A) potassium salt (Sigma; 10 mg/ml), and 5 μl of ultrapure BSA (Ambion; 10 mg/ml). Hybridizations were performed on a Gene TAC Hyb Station (Genomic Solutions) for 16 h at 42°C. After the hybridizations slides were washed with 1× SSC for 10 min at 60°C, 1× SSC + 0.2% SDS for 10 min at 60°C, and 0.1× SSC + 0.2% SDS for 10 min at 42°C. Slides were then rinsed in isopropanol and dried by centrifugation before being scanned. All 48 slides were hybridized at the same time to reduce any technical artifacts that might occur.

Array analysis.

Images of scanned slides (10-μm resolution) were obtained with an Axon 4200A scanner (Axon Instruments). The detected fluorescence was adjusted by altering the photomultiplying tube (this ranged from 700 to 820 arbitrary units) to ensure that few cDNA spots were saturated and that the intensity ratio of the Cy3 and Cy5 signals was close to 1. Image analysis was performed with the GenePix program (version 5.1, Axon Instruments). Abnormal hybridization signals were flagged and not used in subsequent analysis. Edited images were imported into Acuity (version 4, Axon Instruments), and analysis was performed with the following criteria. All spot intensities were adjusted with local background correction and normalized with Lowess transformation (44). The data were filtered to remove signal artifacts and subthreshold signals (<50 units greater than background). Clones were assigned as being differentially expressed if they reached significance of P <0.05 by t-test after Benjamini-Hochberg multiple test correction. For data included in this study only genes that showed greater than twofold difference in expression level are considered. Cluster analysis was performed (Pearson uncentered) on features differentially expressed in at least one experiment with Acuity software. The experimental hybridizations are archived at the European Bioinformatics Institute under accession number E-MEXP-1023. All experimental procedures for array construction, hybridization, and analysis complied with MIAME guidelines (5).

Real-time PCR.

Total RNA for real-time PCR was isolated as described above. Samples for real-time PCR were taken from the same pool of material as used for microarray analysis. RNA (2 μg) was denatured (65°C, 10 min) in the presence of 1 μl of oligo(dT)17 primer (500 ng/μl), left at room temperature for 5 min to allow annealing, and then kept on ice. cDNA was synthesized with 15 U of Bioscript reverse transcriptase (Bioline) in the presence of dNTPs (final concentration 200 μM each) at 42°C for 1 h in a final volume of 20 μl. The cDNA was diluted to 100 μl, and 3 μl was used as template for PCR using primers designed against the Atlantic salmon genes of interest (Table 3). An Opticon qPCR machine was used for monitoring the cDNA amplification with ready-prepared 2× SYBR Green PCR master mix (Bio-Rad). PCR reactions were performed in 25-μl volumes on a white Bio-Rad 96-well PCR plate covered with transparent film. A negative control (no template) reaction was also performed for each primer pair. Efficiency of amplification was determined for each primer pair with 10-fold dilutions (1-, 10-, 100-, and 1,000-fold dilutions). A sample from the serial dilution was separated on an ethidium bromide-stained agarose gel to confirm that a single band of correct size was amplified. Two “control” genes (elongation factor 1α and β-actin) were used for normalization of expression. PCR conditions for all gene assays were 95°C for 5 min followed by 94°C for 15 s, 57°C for 15 s, 72°C for 20 s for 35 cycles. The fluorescence signal output was measured and recorded at 78°C during each cycle for all wells. Melting curves (1°C steps between 55 and 95°C) ensured that only a single product had been amplified in each reaction.

View this table:
Table 3.

PCR primers used for real-time PCR analysis to assay gene expression of cDNAs found to be stimulated by either rIFN-γ or rIFN type I proteins

To determine the relative expression level of candidate genes the method of Pfaffl (40) was used to obtain relative expression of candidate genes to both β-actin and elongation factor 1α. The efficiency of the PCR reaction for each primer set was measured on the same plate as the experimental samples. The efficiency was calculated as E = 10(−1/s), where s is the slope generated from the serial dilutions when log dilution is plotted against ΔCT (threshold cycle number). For all real-time PCRs triplicate reactions were performed.


Differential gene expression in SHK-1 cells.

There were no visually apparent morphological differences between control SHK-1 cells and those exposed to rIFN for either 6 or 24 h, and cells remained adhered to the flasks. Ten of the 48 slides hybridized had high levels of background or a highly skewed ratio of Cy3 to Cy5 and were omitted from subsequent analyses. A minimum of nine slides were available for each of the four experimental conditions. After stringent quality screening, a mean of 3,287 ± 386 (SE) (19%) features were retained as a data set for analysis. Over the four conditions 168 clones were found to be significantly altered in expression, where the mean of the replicates was altered by at least twofold. Ninety-eight of these were upregulated, of which 72 clones increased after exposure to rIFN-γ and 47 clones increased after rIFN type I exposure. The remaining clones were downregulated, with 37 and 42 showing decreased expression following rIFN-γ and rIFN type I exposure, respectively. Of the upregulated clones, 23 clones were found to be induced by both rIFN molecules (Table 4), demonstrating that there is considerable coregulation of responses. Seven clones were decreased in expression by both rIFNs. For both rIFN molecules there was a greater number of clones stimulated at 24 h after exposure than at 6 h. For rIFN type I, 23 cDNAs were upregulated at 6 h and 38 were upregulated at 24 h, with 14 of these genes increased at both times. For rIFN-γ, 13 genes were found to be stimulated at 6 h compared with 66 genes at 24 h, with 7 upregulated at both times (Table 4). In total, 77% of the clones showing an increase in expression showed nominal homology to a functional protein. Of the clones showing decreased expression, approximately equal numbers were seen at 6 h (19) and 24 h (20) with rIFN-γ, while many more were downregulated at 6 h (34) vs. 24 h (9) with type I rIFN. Tables 5 and 6 show details of the clones identified as being differentially expressed by microarray analysis in this experiment; the full list appears in Supplemental Tables S1 and S2. 1

View this table:
Table 4.

Number of genes demonstrating significantly altered expression in SHK-1 cells after exposure to rIFN molecules, as revealed by microarray analysis

View this table:
Table 5.

Genes identified as upregulated by at least twofold after rIFN-γ or rIFN type I stimulation in Atlantic salmon cell line SHK-1, as determined by microarray analysis

View this table:
Table 6.

Genes identified as downregulated after rIFN-γ or rIFN type I stimulation in Atlantic salmon cell line SHK-1, as determined by microarray analysis

Several of the clones that were upregulated were identified as putative homologs of genes known to be responsive to IFNs in mammals. These included the chemokine γ-IP, which is an IFN-γ-induced molecule in mammals, and the antiviral protein Mx. These two genes were stimulated by both rIFN type I and rIFN-γ in this study. Two IFN-stimulated genes, ISG12 and ISG15 (also termed ubiquitin-like protein), IFN-γ-inducible gene 2 (GIG2) (60), and an IFN-induced transmembrane protein (iip2) were also stimulated. Together these results indicate a substantial IFN response to both rIFN molecules. A number of repeat sequences were identified. It is likely that these are repeats within the noncoding region of the mRNA of regular transcripts; these may be identified in the future when greater sequence coverage of the Atlantic salmon is available.

A Cluster analysis heat map (Fig. 2) was performed on all the data from replicate slides on the genes in Tables 5 and 6 without filtering to give a visualization of the gene expression patterns. Those genes that are highly induced at 6 h generally remain induced at 24 h, and often, as seen with γ-IP, serum amyloid A, and ubiquitin-like protein, these genes also appear to be coregulated by both recombinant IFNs and may represent immediate-early responses. At 24 h there is more differentiation in expression responses following exposure to the two IFNs. For downregulated genes there appears to be more downregulation at 6 h than at 24 h, which may be an immediate cellular transcriptional response.

Fig. 2.

Cluster analysis (Pearson) of all genes found altered in expression after exposure to recombinant interferons (IFNs), visualized as a heat map. Means of all replicate slides have been included, and no filtering has been applied. The scale of induction or repression is shown. γ, IFN-γ; type I, IFN type I.

To assess the cellular processes that were affected by the rIFN molecules, the upregulated genes were classified according to biological process GO identifiers (Fig. 3). Only 51% of the differentially expressed proteins had GO identifiers, reflecting the fact that the current annotation of salmon sequences is still limited; however, the GO identifiers do give insights into the cellular processes that are altered by these recombinant cytokines. For example, the GO identifiers indicate that a large proportion of genes can be assigned as immune-response genes for both IFNs. In addition, genes encoding molecules involved with protein metabolic processes are also upregulated; this may reflect the reprogramming of the cells, with translation and protein modification being altered. The type I IFN molecule stimulates genes related to responses to biotic stimuli, which may be linked to the innate immune functions of the type I IFNs.

Fig. 3.

cDNAs that were upregulated by recombinant IFN molecules were assigned high-order Gene Ontology (GO) classifications. All GOs were for biological process, and the GO identifiers are also given.

Seven genes that were upregulated were selected for real-time PCR confirmation. Six of the genes are believed, from mammalian studies, to be stimulated by either IFN-γ or IFN type I. These candidate genes comprise two chemokines, γ-IP and a CXCL2-like chemokine; Mx, a characteristically IFN type I-responsive gene; the IFN-γ-inducible gene GIG2; β2-microglobulin; and a ubiquitin-like protein. One further gene encoding serum amyloid A was included, which may be described as an acute-phase response protein. β-Actin was used for normalization of gene expression. This gene was not found to be modulated by the recombinant cytokines in the microarray analysis. Another commonly reported reference gene, ELF-1α (36), was found to be reduced in expression by IFN-γ at the 6-h sample and was therefore not considered for the present study. In general all genes examined by real-time PCR reflect the expression found by microarray analysis (Table 7), and there was a significant correlation between microarray data and real-time PCR fold changes between control and exposed RNA samples (Pearson correlation = 0.812, P < 0.001; Fig. 4).

Fig. 4.

Comparison of expression data obtained by microarray analysis with that obtained through real-time PCR for candidate genes found to be differentially expressed after recombinant IFN exposure. Fold increase relates to the fold difference in gene expression between the exposed and control cells obtained by either microarray analysis (x-axis) or real-time PCR (y-axis). Data were log transformed to improve normality before correlation (Pearson correlation 0.812, P < 0.001).

View this table:
Table 7.

Candidate gene expression in Atlantic salmon SHK-1 cells exposed to rIFNs, as determined by real-time PCR and microarray analyses

An additional gene expression analysis using the same set of genes for real-time PCR was performed with primary head kidney leukocytes to assess how the cell line compared with cells more akin to fresh macrophages (Table 8). The genes found to be upregulated in the SHK-1 cells were similarly found to be increased in the primary cultures; however, the magnitude of the response varied considerably, with some genes (such as γ-IP) showing a reduced response in primary cultures and others (Mx and GIG2) showing a greater increase in primary cultures than in the SHK-1 cell line.

View this table:
Table 8.

Gene expression of candidate genes in Atlantic salmon primary leukocyte cultures exposed to rIFN molecules


The regulation and function of IFN molecules in lower vertebrates is not yet fully understood. To explore this further in fish, we have used functional rainbow trout rIFNs (61) to stimulate an Atlantic salmon cell line. We have analyzed the transcriptional response of mRNAs with a recently developed Atlantic salmon cDNA microarray that includes cDNAs for immunoresponse genes identified through subtractive cloning (31). We chose to use a cell line in order to remove fish-to-fish variation found in in vivo experiments and confounding factors such as stress- and behavior-induced gene expression changes.

The SHK-1 Atlantic salmon cell line used in this study was derived from primary cultures of salmon head kidney, a major hematopoietic tissue in fish (12). It was developed initially for the culture of a salmon viral pathogen, infectious salmon anemia virus (ISAV), and characterized as being adherent with typical macrophage-like morphology. The cells are known to phagocytose the fish bacterial pathogen Aeromonas salmonicida but show minimal antibacterial activity (11). Further characterization of the cell line in relation to major histocompatibility complex (MHC) I and II expression has shown that both MHC I heavy chain and MHC II β-chain mRNAs are induced after infection by two different viruses (ISAV and infectious pancreatic necrosis virus), indicating that the cells respond to viral challenge possibly through an IFN signaling mechanism (24).

There have been significant advances in fish IFN research in the past few years, with both type I and type II IFNs being sequenced from a variety of fish species (33, 49, 61). In the case of IFN-γ, protein sequence identity between teleost and mammalian molecules is low, in the range of 16–23% (61). Fish type I IFN molecules show 27–30% sequence identity with mammalian IFN-α and 13–18% with mammalian IFN-β proteins (49). The fish type I IFN is believed to be a more ancestral form, with IFN-α and IFN-β resulting from a duplication event, which occurred after the divergence of fish from other vertebrates.

There have been a number of recent studies in salmonid fish in which genes such as Mx and γ-IP have been used as indicators of either type I IFN or IFN-γ production in vivo following viral infection in salmon (32) with real-time PCR. Others have used subtractive cloning with virus-infected tissues in an attempt to clone IFN-induced genes in salmonids (35) from which a number of virus-induced transcripts were identified. More recently microarray analysis (43) has been used to investigate gene expression responses following DNA vaccination to infectious hematopoietic necrosis virus (IHNV), where a number of genes in common with this study (ISG12, β2-microglobulin, and Mx) were shown to be increased in expression, suggesting that the DNA vaccine elicits an IFN response.

The pattern of gene expression stimulated by the different rIFNs showed many similarities. In both cases there was a greater number of genes found to be upregulated at 24 h than at 6 h. In addition, where genes were upregulated at both time points, the magnitude of increase was higher at 24 h. For the downregulated genes there were fewer genes altered by both IFN molecules and no clear pattern shared between the two exposures. Many more genes were downregulated by type I IFN at 6 h than at 24 h, which may indicate early transcription-inhibition responses.

Both type I and II IFNs act through IFN receptors, as described above, so the responsive genes identified in this study may have been directly stimulated by the IFN signal transduction pathway. Many of these genes might be expected to contain either a GAS or an ISRE (or both) motif within the promoter sequence (41). Alternatively, they may be secondary genes, where their expression is induced indirectly by other proteins turned on by the IFN stimulation. Few promoter elements of Atlantic salmon genes have been cloned and characterized. However, within the genes identified here several have promoter sequences available for Atlantic salmon, or the closely related rainbow trout. The Mx protein is characteristically an IFN type I-responsive gene involved in antiviral activity and has been identified in many teleost species including rainbow trout (56) and Atlantic salmon (50). The promoter sequence of rainbow trout Mx has an ISRE site (9) but no apparent GAS site, suggesting this might be specifically induced by type I IFN alone. In this study we found by microarray analysis that Mx was stimulated at 24 h by both IFN types, but the magnitude of the response to IFN type I was three times that of the response to IFN-γ. These results, together with the real-time PCR data, which also indicated upregulation at 6 h, suggest that in fish there is significant cross talk between the two pathways.

Another gene found to be highly induced by both IFNs at both 6 h and 24 h was γ-IP (26), a chemokine with greatest homology to CXCL10 in mammals that is strongly induced by IFN-γ (28). The present work has demonstrated that this gene is highly induced by both IFN-γ and IFN type I, which was unexpected. Examination of the promoter elements of this gene revealed both ISRE and GAS elements present (data not shown), so that in fish this gene may be under the control of both type I IFN and IFN-γ. The coregulated gene, ISG12, has not to date had its promoter region characterized but is clearly induced by type I IFN and IFN-γ in both fish and mammals (16). It has also been shown to be stimulated by a DNA vaccine against IHNV in rainbow trout (43).

The mRNA for the ubiquitin-like protein UBL, also termed ISG15, was also found to be highly induced in this study. It is one of the earliest characterized genes found to be responsive to type I IFN molecules (18). This 15-kDa protein has sequence and functional properties similar to ubiquitin and has recently been characterized in Atlantic salmon (51). As with ubiquitin, which targets proteins by ubiquitination for destruction by the proteasome, the UBL protein ISG15 has similar mechanisms for targeting proteins. The exact fate of these targeted proteins remains unclear (12). It is known that the JAK/STAT signal transduction pathway is affected by ISG15, but this mechanism is not, as yet, fully determined (22). Many of the modifying enzymes associated with the UBL protein ISG15 are also induced by IFN, suggesting that it may have a role in regulation of the IFN signal transduction pathway (38). However, it remains unclear whether this UBL protein is involved in generation of peptides for antigen presentation on MHC I molecules.

One of the key functions of IFN-γ is the stimulation of antigen presentation via MHC I molecules. In this experiment we found that β2-microglobulin and the MHC class I heavy chain were upregulated by exposure of the cell line to rIFN-γ. For efficient antigen presentation many other proteins are involved in the production and transport of antigenic peptides to the cell surface (4). These proteins include proteasome subunits LMP2, LMP7, and TAP molecules, among others. However, although present on the microarray, none of these molecules was found to be significantly modulated by stimulation of SHK-1 cells with either rIFN in this study.

To assess the modulation of genes by the rIFNs in primary Atlantic salmon enriched macrophage cultures compared with the SHK-1 cell line, real-time PCR was performed on the same set of genes. Interestingly, although gene expression was in most cases increased as found in the cell line, the expression pattern varied between the two cell types. The increase of γ-IP expression was less in primary cultures when stimulated with both rIFNs; however, as with the cell line the response was greater when the cells were stimulated with type I IFN. In the primary cell cultures there was a greater increase in Mx expression after exposure to type I IFN. These differences in expression most likely reflect the more heterogeneous nature of the primary cells.

Fewer genes were downregulated by exposure to rIFNs than upregulated. Type I rIFN caused a decrease in expression of mRNAs for several structural proteins including tubulin, keratin, and collagen. It has been reported that the expressions of genes encoding structural proteins were altered in salmon macrophages exposed to a Piscirickettsia salmonis bacterial infection, with both keratin and α-collagen showing reduced expression (46). Similarly, in liver tissue, collagen expression was reduced after an A. salmonicida infection (31). In mammals both IFN-γ and IFN-α decrease the expression of collagen genes (20, 37). The most abundant group of mRNAs found to be downregulated were those associated with transposable elements, especially at the 6-h timing for both IFN-γ and type I IFN. The genomes of salmonid fish contain large numbers of transposable elements, many of which are actively transcribed (25). Their transcription is known to be modulated by a variety of stimuli in rainbow trout, including bacterial infection, stress, and toxin exposure (25), but the function and mechanism of this regulation remain unclear. It should be noted that no transposable elements were upregulated.

There is growing evidence that there is more than one signal transduction pathway that mammals use to elicit the IFN response. For example, some IFN-γ responsive genes may have a STAT1-independent mechanism of stimulation (45). Thus IFN-mediated signaling pathways are yet to be fully described (41, 55). In this paper we show that many genes are regulated by both type I and type II IFN in fish, and that there is much more coregulation of genes than expected, suggesting that there are both similar and novel aspects of the fish response to IFNs. Caution should always be taken when interpreting results from cell line experiments, however, as there may be certain phenotypic features that differ from those observed in vivo. This might be one explanation for the lack of more widespread upregulation of genes involved in the MHC I pathway of antigen presentation, with only some genes responsive to IFN exposure in SHK-1 cells. Examining the lower vertebrate response to IFNs will not only help extend our knowledge of the evolution of IFN signaling but may also help in the control and understanding of how fish respond to viral attack.


The TRAITS/SGP microarray was constructed by ARK Genomics and funded by BBSRC Grants 98/EGA17674 (University of Stirling), 98/EGA17675 (University of Aberdeen), and 72/EGA17676 (Cardiff University) and by Grant 139617/140 “Salmon Genome Project” of the Research Council of Norway.


The TRAITS/SGP cDNA array development effort was coordinated by the University of Stirling.


  • 1 The online version of this article contains supplemental material.

  • Address for reprint requests and other correspondence: S. A. M. Martin, School of Biological Sciences, Zoology Dept., Univ. of Aberdeen, Tillydrone Ave., Aberdeen AB24 2TZ, UK.

    Article published online before print. See web site for date of publication (


View Abstract