Characterisation of the teleostean κB-Ras family: The two members NKIRAS1 and NKIRAS2 from rainbow trout influence the activity of NF-κB in opposite ways

Two structurally similar NF-κB-inhibitor-interacting Ras-like proteins (NKIRAS) regulate the activity of the transcription factor NF-κB and thereby control several early immune mechanisms in mammals. We identified the orthologous sequences of NKIRAS1 and NKIRAS2 from the rainbow trout Oncorhynchus mykiss. The level of sequence identity was similarly high (≥68%) between the two and in comparison to their mammalian counterparts. Strikingly, NKIRAS2 was present as four transcript variants. These variants differed only in length and in the nucleotide composition of their 5’ termini and were most likely generated by splicing along unconventional splice sites. The shortest NKIRAS2 variant was most strongly expressed in a lymphocyte-enriched population, while NKIRAS1 was most strongly expressed in cells of myeloid origin. Fluorescent-labelled NKIRAS1 and NKIRAS2 proteins from rainbow trout were detected in close association with the p65 subunit of NF-κB in the nucleus and cytoplasm of CHSE-214 cells. Subsequent reporter-gene experiments revealed that NKIRAS1 and a longer NKIRAS2 variant in rainbow trout decreased the level of activated NF-κB, while the two shortest NKIRAS2 variants increased the NF-κB activity. In addition, the overexpression of the shortest NKIRAS2 variant in CHSE214 cells induced a stronger transcription of the genes encoding the pro-inflammatory cytokines TNF, CXCL8, and IL1B compared to non-transfected control cells. This is the first characterisation of NKIRAS orthologues in bony fish and provides additional information to the as yet underexplored inhibition pathways of NF-κB in lower


Introduction
The 'NF-κB-inhibitor-interacting Ras-like proteins' NKIRAS1 and NKIRAS2 (also known as κB-Ras 1 and κB-Ras 2 [1]) are members of the hyperfamily of Ras proteins accounting for more than 150 GTP-binding proteins largely involved in cell-signal-transduction networks [2]. NKIRAS1 and NKIRAS2 contribute to the maintenance of homeostasis in mammalian cells by controlling the activity of NF-κB (nuclear factor 'kappa-light-chain-enhancer' of activated B cells) factors [3].
The NF-κB/Rel factors represent one of the best characterised families of immunorelevant transcription factors [4], which are of particular importance for both innate and adaptive immune responses [5]. Overshooting of immune reactions and pathological responses is prevented by controlling the activity of NF-κB by a range of different signal-responsive mechanisms [6]. Under homeostatic conditions, the 'inhibitors of NF-κB proteins' (IκBs) maintain NF-κB factors in an inactive form in the cytosol [7]. Eight IκB-like family members have been discovered in mammals [8,9], and the alpha member is a prototypic representative in mammals [10] and fish [11]. IκB masks the nuclear localisation signals of NF-κB factors [12], thereby preventing their access to NF-κB-specific response elements in the promoters of relevant genes. NKIRAS proteins interact with the PEST domains of IκB proteins, mainly of the beta type [1,13], to impede their degradation and to restrain the inhibitory complex and its substrate NF-κB within the cytoplasm [1,14]. This inhibition can be removed through activated pattern-recognition receptors and several pro-inflammatory mediators that promote the degradation of IκBα [15].
The NKIRAS proteins are activated through the binding of GTP and Mg 2+ to the 'switch I' and 'switch II' domains [16]. The subsequent conformational rearrangement of NKIRAS1 and NKIRAS2 allows recruitment of downstream effector proteins [17,18]. NKIRAS2 bound to GDP has been reported to suppress the phosphorylation of the NF-κB p65 subunit, thereby interfering with its interaction with the transcriptional coactivator p300 [19].
Comprehensive phylogenetic comparisons of teleostean Ras factors have been published so far [20,21], but functional studies on the role of Ras proteins in fish immunity are rare [22] and nothing is known about the involvement of NKIRAS during the early and presumably NF-κB-driven immune processes in fish. Here, we characterised the two NKIRAS factors in the rainbow trout Oncorhynchus mykiss and investigated their structural and functional properties in terms of interference with the early immune-activating mechanisms of salmonid fish against Aeromonas salmonicida infections.

Fish
Rainbow trout (O. mykiss) were obtained from a local commercial fish farm (Forellenzucht Uthoff GmbH, Neubrandenburg, Germany) and kept at 15 • C in 1000 L tanks in partially recirculating water systems and a light period of 12 h per day. For the stimulation trial, fish (30-50 g) were intraperitoneally injected with 5 × 10 5 colony-forming units (CFU) of the highly virulent JF5505 strain of A. salmonicida ssp. salmonicida (diluted in 200 μL phosphate buffered saline [PBS]) using a 20G needle. Control fish were injected with 200 μL PBS. At 6, 12 and 24 h post stimulation, four fish per treatment group were euthanised using an overdose of benzocaine (100 mg/L). All experimental procedures complied with the relevant European guidelines on animal welfare (Directive 2010/63/EU on the protection of animals used for scientific purposes) and were approved by the institute's ethics board (approval ID: FLI 28/17).

Cell sorting
Cells were isolated from individual head kidneys, and 4 × 10 4 cells were used for labelling with 50 ng MAb21 [23] per 2 × 10 5 cells at 4 • C for 30 min. The cells were then washed and centrifuged (515×g, 4 • C, 5 min). The cell pellet was resuspended in 180 μL autoMACS Running Buffer (Miltenyi Biotec) before adding 20 μL of anti-mouse secondary antibody conjugated to magnetic beads (Miltenyi Biotec) for each 1× 10 7 cells. After incubation for 20 min at 4 • C, the cells were washed with 700 μL autoMACS Running Buffer and centrifuged for 10 min at 300×g. Cells were resuspended in 500 μL autoMACS Running Buffer and sorted using a single magnetic column and the posselS program on the Auto-MACS Pro Separator (Miltenyi Biotec). The separated positive and negative fractions were centrifuged for 7 min at 515×g and 4 • C and the cell pellets were resuspended in 350 μL RNA RLT lysis buffer (Qiagen) and stored at − 70 • C. RNA extraction and cDNA synthesis were conducted as described previously [24]. Subsequent expression analysis was conducted on MAb21-negative cells (a lymphocyte-enriched population containing mainly B cells, T cells and thrombocytes) and MAb21-positive cells (a population enriched with cells of myeloid origin).

Construction of NKIRAS-expression constructs
We amplified the open-reading frames (ORFs) of NKIRAS1 and three NKIRAS2 transcript variants using the oligonucleotide primers listed in Table 1 and the Platinum Taq High-Fidelity DNA Polymerase (Thermo Fisher Scientific). Amplicons were subcloned in pGEM-T Easy (Promega), retrieved by digestion with the restriction enzymes HindIII and EcoRI and inserted 'in frame' into the mammalian expression vector v280 [25,26] (previously double-digested with the above restriction enzymes). The resulting plasmids (v280_NKIRAS1 and v280_NKIRAS2) were utilised for functional analyses.
We identified the subcellular localisation of the NKIRAS1 (XM_021560247) and NKIRAS2 (XM_021557708) variants in salmonid model cells by inserting a fragment coding for a derivative of the red fluorescent protein (mPlum) and a green fluorescent protein (GFP) at the 3' end of the CDS of our v280_NKIRAS1 and v280_NKIRAS2 plasmids. The mPlum and GFP fragments had previously been amplified from commercial vectors (mPlum: pmPlum, ClonTech/Takara; GFP: pAM 505, NCBI-nucleotide accession code: AF140578) and inserted into the v280 clones using the restriction sites ClaI and HindIII or HindIII and EcoRI, respectively.

Cell culture, transfections and luciferase assay
The CHSE-214 cell line derived from Oncorhynchus tshawytscha was cultured as described previously [25]. Endotoxin-free preparations (ZymoPure II Plasmid Maxi Prep Kit, ZymoResearch) of the expression constructs were transfected into CHSE-214 cells using the X-tremeGENE HP DNA Transfection Reagent (Roche). For co-transfection assays in six-well plates, we used 50 ng of the modified NF-κB-responsive promoter of the endothelial-leukocyte adhesion molecule gene (ELAM-1-luc) [27] and increasing concentrations (from 100 ng to 750 ng) of the v280_NKIRAS1 and v280_NKIRAS2 plasmids.
Induction experiments were carried out in 24-well plates. Three wells of each row were left as unstimulated controls, while the other three were stimulated for 3 h or 24 h with 100 ng/mL flagellin isolated from the Gram-negative bacterium Salmonella typhimurium (tlrl-stfla; Invivogen), which has been proven to induce inflammatory responses in CHSE-214 cells [25,26,28]. Three hours after incubation, the cells were harvested in ice-cold PBS for RNA isolation and the subsequent profiling of immune-gene expression. Twenty-four hours after stimulation, the luciferase activity of the cell lysates was measured with the Dual-Luciferase Reporter Assay System (Promega) in a Lumat LB9501 luminometer (Berthold). The values were normalised against the protein concentration of the CHSE-214 cell extracts. Each transfection was assayed in triplicate; each transfection experiment was repeated three times.

Quantitative PCR (qPCR) analysis
We profiled the quantity of the various NKIRAS transcripts in rainbow trout leucocytes by establishing a panel of discriminating oligonucleotide primers that allowed a selective quantification of the NKIRAS1 and the four NKIRAS2 transcript variants ( Table 1). The qPCR primers were designed (using Pyrosequencing Assay Design software v.1.0.6; Biotage) to amplify fragments between 108 nt and 201 nt. We further profiled the expression of cytokine-encoding genes in CHSE-214 cells. The Oncorhynchus-specific primer pairs for IL1B, CXCL8, TGFB-2, and IL4/13 were published previously [25]; the primer pairs for TNF-1 and IL10 were optimised for O. tshawytscha (Table 1). EEF1A1 [29] and RPS5 [24] were chosen as reference genes for normalisation of the qPCR data. Quantitative PCR analysis was performed on the LightCycler-96 system (Roche), essentially as previously described [25]. The cDNA input into the individual RT-qPCR assays was equivalent to 2.5 ng total RNA isolated from cells.

Data analysis
A parametric t-test or nonparametric Mann-Whitney U test provided by GraphPad Prim 8 software was used to evaluate the statistical significance of the qPCR data. In all these tests, a two-tailed p value of <0.05 was considered statistically significant. The ClustalW and T-Coffee expresso alignment tools were used to compare multiple NKIRAS nucleotide and amino acid (aa) sequences.
The Molecular Evolutionary Genetics Analysis package MEGA7 [30] was used to construct the phylogenetic tree using the Neighbour-Joining method with a bootstrap test based on 1000 replications. Three-dimensional structures were obtained using PHYRE2 (Protein Homology/analogY Recognition Engine V 2.0) [31]. Signal peptides and disordered protein regions were predicted using the online tools SignaIP-5.0. and PrDOS [32].

Differential splicing generates four NKIRAS2 variants alongside one NKIRAS1 transcript in rainbow trout
Our search of the NCBI-gene database for NKIRAS-like sequences from rainbow trout revealed one coding sequence (CDS) (accession code: XM_021560247) orthologous to the human NKIRAS1 gene and four transcript variants X1 to X4 (XM_021557705 to XM_021557708) orthologous to the human NKIRAS2 gene (Suppl. Fig. 1). The CDS of NKIRAS1 from rainbow trout comprises 579 nt and shares the same length with the shortest NKIRAS2 variant (X4), which is henceforth referred to as NKIRAS2a. The 579-nt long sequence of NKIRAS2a is also included in the variants X1 to X3, which also bear specific 5' termini of different lengths and nucleotide compositions. The longest NKIRAS2 variant (X1) comprises 693 nt and is henceforth termed NKIRAS2d, followed by NKIRAS2c (X2, 678 nt) and NKIRAS2b (X3, 672 nt). The coding sequences of all five NKIRAS transcripts share an identity of >69% with the human NKIRAS1 and an identity of >77% with the human NKIRAS2. The coding sequences of NKIRAS1 and NKIRAS2a from rainbow trout share an identity of 68% due to 192 exchanged nucleotides (Suppl. Fig. 1). We verified the lengths and nucleotide compositions of NKIRAS1 and NKIRAS2a to NKIRAS2c by cloning and repeated sequencing.
The individual 5' regions of the NKIRAS2 variants b, c and d share a 25-nt sequence immediately upstream of the common start codon (Fig. 1A, orange box). Interestingly, a ≥96% identical sequence fragment resides in other coding gene sections from bacteria (e.g. sequence ID: SIT85884), protozoans (e.g. CDW78563) and vertebrates (e.g. XM_028038946), but they are translated into completely different amino acid (aa) sequences. In variants b and c, a 64-nt long sequence (Fig. 1A, brown box) is located upstream of this 25-nt motif. The 25-nt and 64-nt segments are separated by 4 additional nt in the case of variant b, whereas variant c contains an additional 10-nt long sequence at its 5' end. In variant d, a different 89-nt long sequence is located upstream of the 25-nt motif (Fig. 1A, blue box).
A BLAST search of these three different 5 ′ -UTR segments against the rainbow trout genome [33] identified an approximately 3030-nt long genomic sequence on scaffold 2106 (Fig. 1B). The ORF of NKIRAS2a is distributed across three exons (I, 94 nt; II, 242 nt; III, 243 nt) and has the same organisation seen in mammals. The additional 25-nt, 64-nt and 89-nt long segments, which characterise the individual NKIRAS2 versions -b, -c and -d, respectively, are all encoded in the genomic region upstream of the conventional three NKIRAS2 exons and allow the generation of four different transcript variants via differential splicing.
We also investigated whether undiscovered transcript variants of NKIRAS might exist by screening an RNA-sequence read collection of two rainbow trout strains [34]. We did not identify any further transcript variants of NKIRAS1 and NKIRAS2; however, we observed that both the NKIRAS1 and NKIRAS2 transcripts bore a certain degree of sequence variability. The CDS of NKIRAS1 contained 11 single-nucleotide changes, contributing to a 1.9% variability ( Fig. 2A), which corresponds approximately to the average variability assessed for the rainbow trout genome [35]. The variability of NKIRAS2 was comparably lower (five single nucleotide changes, 0.9% variability; Fig. 2B).
As a last test, we checked the number of NKIRAS transcripts in other salmonid species. Our search in the NCBI gene database revealed only one entry each for the genomic sequences of both NKIRAS1 and NKI-RAS2 in salmonid fishes (O. mykiss, Salmo salar and Salvelinus alpinus). The genes flanking NKIRAS1 (Fig. 2C) and NKIRAS2 (Fig. 2D) are similarly arranged across different vertebrate genomes, although a few of the neighbouring genes are reversed in their arrangement.

NKIRAS1 and NKIRAS2 from rainbow trout share well-conserved structural motifs
The five NKIRAS sequences from trout encode proteins with lengths ranging from 192 to 230 aa. The aa sequences among the vertebrate NKIRAS orthologues are relatively well conserved (>60%; Table 2). In line with this observation, the encoded NKIRAS proteins from trout contain the same motifs and domains that have been characterised in their mammalian NKIRAS counterparts (Fig. 3A). We also found two and three SxDx motifs, which are recognised and modified by A. salmonicida to impede the catalytic activity of Ras proteins from the infected host and control the activity of leukocytes [36].
Despite the numerous structural similarities, 64 aa residues still allowed discrimination between NKIRAS1 and NKIRAS2 (Fig. 3A). Notably, a signal peptide of 17 aa has been predicted only in NKIRAS2d. In our phylogenetic tree, both genes separate, despite the high degree of sequence identity, due to these differing aa residues (Fig. 3B1). The three-dimensional reconstructions of NKIRAS1 (Fig. 3B2) and NKIR-AS2a (Fig. 3B3) from rainbow trout illustrate the well-conserved quaternary structure of particular domains. The characteristics of the active domains, which are critical for the GTPase activity, are located in a similar configuration as observed in their human counterparts [37].

NKIRAS copy number is slightly modulated after infection with A. salmonicida
The human and murine NKIRAS1 and NKIRAS2 genes are widely expressed in various tissues, including immune cells [3]. Our qPCR measurements (Fig. 4A) revealed that NKIRAS2a (~3000 copies/μg RNA) was the most strongly expressed NKIRAS gene in the MAb 21-negative cell population from rainbow trout enriched with lymphocytes, whereas NKIRAS2b was the least expressed variant (~200 copies/μg RNA). In the MAb21-positive cells of myeloid origin, NKI-RAS1 was most strongly expressed (~670 copies/μg RNA), whereas all other NKIRAS transcripts were expressed at a similarly low level (<100 copies/μg RNA).
Infection with a highly virulent strain of A. salmonicida significantly upregulated the levels of the NKIRAS2 variants a, b, and c (p < 0.05) in the MAb 21-negative cell population by about twofold at 12 or 24 h post infection (hpi) compared to PBS controls (Fig. 4B). In the MAb21positive cell population, the NKIRAS1 level was slightly downregulated within 24 h, while the NKIRAS2a level was 2.7-fold and 4.0fold upregulated after 12 h and 7 d, respectively (Fig. 4C). However, the fourfold increase resulted only in about 230 copies/μg RNA. The concentrations of the other NKIRAS transcripts at 7 days post infection returned to values similar to those measured in the cells of naïve trout.

Overexpressed NKIRAS1 and NKIRAS2 are localised in the cytoplasm and nucleus of salmonid cells
We overexpressed fluorescent-labelled NKIRAS1 and NKIRAS2a from rainbow trout in CHSE-214 cells to investigate the localisation of the salmonid NKIRAS factors. Confocal microscopy revealed that NKIRAS1 (flagged with mPlum) and NKIRAS2 (flagged with GFP) apparently shared the same subcellular localisation in the nucleus and cytoplasm (Fig. 5A1, A2) as is observed in humans [38]. We also stained native NF-κB p65 factors using a monoclonal anti-RELA antibody from rabbit. In unstimulated cells, NKIRAS1 and NKIRAS2 seem to restrain p65 within the cytoplasm (Fig. 5B1, B2), but we could not detect transfer of p65 to the nucleus after stimulation with flagellin for 1 h (data not shown).

NKIRAS1 and NKIRAS2a modulate the activity of NF-κB in opposite directions in CHSE-214 cells
Stimulation of CHSE-214 cells with flagellin increased the level of active NF-κB by an average of 2.8-fold (Fig. 5C1-C4, 2nd column). We transfected the CHSE-214 cells with increasing concentrations (250 ng to 750 ng) of the plasmids expressing NKIRAS1 to drop the flagellininduced NF-κB activation down to the level of the control cells (Fig. 5C1). Transfection with the same amounts of the NKIRAS2cencoding plasmid had a similar dampening effect on the level of activated NF-κB, even though the lowest concentration (250 ng) of transfected NKIRAS2c had a less pronounced effect compared with the NKIRAS1-encoding plasmid (Fig. 5C2). By contrast, increasing concentrations of NKIRAS2a (Fig. 5C3) and NKIRAS2b (Fig. 5C4) raised the level of activated NF-κB by 21-fold (with a quantity of 750 ng of transfected plasmid).

Overexpression of NKIRAS1 and NKIRAS2 increases the in vitro expression of specific cytokines
We also investigated the influence of the three NKIRAS factors NKIRAS1, NKIRAS2a and NKIRAS2c, which had modulated the NF-κB level in different directions, on the expression of cytokine-encoding genes in CHSE-214 cells. These genes included those for the pro- inflammatory cytokines IL1B, TNF and CXCL8 and the antiinflammatory cytokines TGFB2, IL10 and IL4/13 (Fig. 6). Stimulation of the non-transfected control cells with flagellin for 3 h significantly increased the copy numbers of the pro-inflammatory cytokine genes IL1B by fourfold, and CXCL8 and TNF by approximately twofold, whereas the copy numbers of the genes TGFB2, IL10 and IL4/13 remained almost at the same levels under all the examined conditions. We detected almost no changes in cells that were stimulated and transfected with NKIRAS1 versus stimulated non-transfected cells. The stimulation and transfection with NKIRAS2c increased the level of TNF and IL1B by 125-150%. The stimulation and transfection with NKIRAS2a provoked an increase in the transcript abundance of TNF (254%; p = 0.03), CXCL8 (162%; p = 0.09), and IL1B (133%; p = 0.74) over the stimulated non-transfected cells.

Discussion
NKIRAS1 and NKIRAS2 constitute a small Ras subfamily which has been widely investigated with regard to the regulation of NF-κB in mammals [3]. In all extant vertebrate species, the genes encoding NKIRAS1 and NKIRAS2 are most likely paralogues [39] that descended from a whole-genome duplication (WGD) occurring in the ancestral vertebrate [40]. Although salmonid fishes underwent two additional WGDs compared with mammals [33,41,42], NKIRAS1 and NKIRAS2 each appear to be present as only one gene copy. Nevertheless, differential splicing of the original NKIRAS2 mRNAs from rainbow trout most likely allows the generation of at least four different NKIRAS2 variants with differentially long 5 ′ -ends. This splicing mechanism does not affect the succession of the exon order or the length of particular exons, as observed previously in rainbow trout [24,25], but it elongates the N-terminus of the NKIRAS2 variants b, c, and d compared to their paralogue NKIRAS2a in rainbow trout and their orthologues in all other vertebrate species. From a structural point of view, the shortest variant a therefore seems to be the prototypic NKIRAS2 version. In the Atlantic salmon S. salar, NKIRAS1 (but not NKIRAS2) is present as two different coding sequences (XM_014177875 and XM_014177876), one of which extends beyond the 5 ′ -end, similar to the longer NKIRAS2 variants from rainbow trout. Accordingly, the mechanism of the N-terminal NKIRAS-protein extension does not seem to be restricted to rainbow trout; however, the functional consequence associated with this structural change remains unknown.
The present gene profiling of the five NKIRAS transcripts in the lymphoid and myeloid cells of rainbow trout confirmed that the expression of NKIRAS either remained at a generally low transcript level or was only slightly altered after bacterial infection. A revision of ⁓50 transcriptome studies on salmonid fish after exposure to various immune stimuli [43] confirmed that the copy numbers of NKIRAS1 and NKIRAS2 rarely exceed those of the transcript level detected in the control fish. This indicates that the transcription of NKIRAS genes is subject to low variability. The permanent presence of NKIRAS seems to be a crucial mechanism for maintaining homeostasis, as mammalian NKIRAS has been proven to control NF-κB [13] and, by extension, numerous immune-gene programs [5,6]. Our structural analysis of the NKIRAS sequences from rainbow trout suggested that all variants may function similarly to their mammalian counterpart because the essential aa residues and motifs of the NKIRAS orthologues were well preserved during vertebrate evolution.
The microscopy examination of NKIRAS-overexpressing CHSE cells did not provide evidence of a direct influence of NKIRAS on the translocation of NF-κB. This may reflect the artificial overexpression of NKIRAS or the stress to which the cells were exposed by the previous transfection. Our reporter-gene experiments in CHSE-214 cells revealed that expression of rainbow trout NKIRAS1 and NKIRAS2c reduced the activity of NF-κB, in agreement with previous reports on mammalian NKIRAS proteins [1]. By contrast, expression of NKIRAS2a and NKIR-AS2b had an inducing effect on the activity of NF-κB. We found evidence to support an increase in the transcript amount of pro-inflammatory cytokine genes, at least in response to NKIRAS2a expression, in stimulated CHSE-214 cells compared to cells without overexpressed NKIRAS2a.
The reason for these contrary NF-κB regulation mechanisms may be the specificity of the NKIRAS variants for the various combinations of NF-κB/IκB complexes. This would require, as a first step, that the five NKIRAS variants interact specifically with the multiple IκBα proteins in rainbow trout that have been identified apart from the prototype [11]. No IκBβ proteins have yet been isolated from salmonid fish; however, IκBα from rainbow trout contains six ankyrin repeats, as occurs in the human IκBβ, whereas the human IκBα has only five ankyrin domains [44]. Particular IκBα variants may therefore compensate for the loss of IκBβ in rainbow trout. A second point is that these IκB proteins, in turn, interact intimately with the NF-κB/Rel factors. However, few of those factors contain a transcription activation domain, such as p65 (RelA); others such as p50 (NFKB1) or p52 (NFKB2) lack this domain entirely and require interactions with further transcription factors to induce transcription [5,6]. The different NKIRAS-mediated mechanisms controlling the activity of certain NF-κB/Rel factors could explain the difference in the expression of the reporter construct and the inflammatory genes after overexpression of the individual NKIRAS variants.
The finding that transcript variants of NKIRAS2 exert different impacts on the activity of NF-κB could be viewed as unexpected, especially in light of the fact that these variants differ only in their N-termini, which account for only about one sixth to one seventh of the shortest variant 2a. Nevertheless, the N-terminus has been reported to determine several key properties of a protein [45], such as the half-life. In this context, we refer to the predicted 17 aa signal peptide of variant d of NKIRAS2. This signal peptide could guide the insertion of NKIRAS2d into cellular membranes along the secretory pathway, and its cleavage would ensure that the mature NKIRAS2 variant d factor starts with a glutamic acid residue at the N-terminus. The presence of this residue decreases the half-life of a protein to 1 h [46]. NKIRAS2d would therefore be a rather fragile factor compared to its NKIRAS2 isoforms that start with a methionine residue and have an expected half-life of 30 h [46]. Alternatively, the preservation of signal peptides in the mature protein could provide additional features, such as selective coupling to G proteins or regulative functions [47]. Regardless of the presence of signal peptides, the N-terminus is well established to determine the affinity of a protein for involvement in different cell signalling cascades [48]. The different N-termini probably also affect the interaction of NKIRAS with PEST domain-containing factors other than IκB, such as hormone receptors. This hypothesis might be addressed in future analyses.  The HeatMap illustrates the averaged fold-change values (according to the legend on the right) of the mRNA concentrations measured in nontransfected cells (2nd row) and transfected cells (either with NKIRAS1-, NKIRAS2a-or NKIRAS2c-expression plasmids, 3rd to 5th row) after a 3 h stimulation with 100 ng/mL flagellin and compared to unstimulated control cells (set as 1.0, 1st row). The quantified transcripts are listed as gene symbols above the scheme. All expression values were normalised against the geometric mean of two reference genes.

Conclusions
We describe here the structure of five NKIRAS variants in rainbow trout which also modulate the activity of NF-κB and, thus, the proinflammatory immune responses in a teleostean model cell. Beyond NF-κB signal transduction, NKIRAS proteins may conceivably control further signalling cascades. Unfortunately, insufficient knowledge is available regarding the NF-κB network in fish; therefore, the obtained data cannot be integrated and adequately evaluated. Our future research will address the interaction of NF-κB and its associated factors in bony fish. All authors discussed the results and commented on the manuscript.