Lost and Found: The Family of NF-κB Inhibitors Is Larger than Assumed in Salmonid Fish

NF-κB signalling is largely controlled by the family of ‘inhibitors of NF-κB’ (IκB). The relevant databases indicate that the genome of rainbow trout contains multiple gene copies coding for iκbα (nfkbia), iκbε (nfkbie), iκbδ (nkfbid), iκbζ (nfkbiz), and bcl3, but it lacks iκbβ (nfkbib) and iκbη (ankrd42). Strikingly, three nfkbia paralogs are apparently present in salmonid fish, two of which share a high sequence identity, while the third putative nfkbia gene is significantly less like its two paralogs. This particular nfkbia gene product, iκbα, clusters with the human IκBβ in a phylogenetic analysis, while the other two iκbα proteins from trout associate with their human IκBα counterpart. The transcript concentrations were significantly higher for the structurally more closely related nfkbia paralogs than for the structurally less similar paralog, suggesting that iκbβ probably has not been lost from the salmonid genomes but has been incorrectly designated as iκbα. In the present study, two gene variants coding for iκbα (nfkbia) and iκbε (nfkbie) were prominently expressed in the immune tissues and, particularly, in a cell fraction enriched with granulocytes, monocytes/macrophages, and dendritic cells from the head kidney of rainbow trout. Stimulation of salmonid CHSE-214 cells with zymosan significantly upregulated the iκbα-encoding gene while elevating the copy numbers of the inflammatory markers interleukin-1-beta and interleukin-8. Overexpression of iκbα and iκbε in CHSE-214 cells dose-dependently quenched both the basal and stimulated activity of an NF-κB promoter suggesting their involvement in immune-regulatory processes. This study provides the first functional data on iκbε—versus the well-researched iκbα factor—in a non-mammalian model species.


Introduction
The family of NF-κB (nuclear factor kappa-light chain-enhancer of activated B cells)/Rel transcription factors is activated by a broad range of environmental and endogenous cues, including viral and bacterial pathogen-associated molecular patterns (PAMPs) and cytokines [1,2]. The activated NF-κB pathways strongly drive immune and stress responses [3,4], as they direct inflammatory processes and cell growth, differentiation, and survival [5][6][7][8][9].
Several members of the IκB family have been well conserved during evolution, as orthologs have been identified in birds [23], fishes [53], and even insects [54]. However, a detailed characterisation of the iκb proteins from lower vertebrates is still pending. Iκbα is the best researched iκb ortholog and has been characterised in different fish species [53,[55][56][57][58][59][60][61][62][63][64], whereas its iκbβ paralog has apparently been lost in various bony fish families. According to the gene database of the National Center for Biotechnology Information (NCBI), the iκbβ-encoding nfkbib gene is absent from the sequenced genomes of the Salmonidae, Percidae, Gadidae, Carangidae, and Oryziinae, although it is still present in other teleost fish species from the Cyprinidae or Ictaluridae (as of December 2022). Iκbη encoded by the ankrd42 gene is apparently absent from most of the teleostean genomes sequenced so far.
In this report, we provide evidence that iκbβ has not been lost from salmonid genomes, and we characterise this putative iκbβ together with the canonical iκbα and iκbε proteins, the nuclear iκbδ and iκbζ proteins, and bcl3 from the rainbow trout salmonid fish (Oncorhynchus mykiss). The findings of the present study, therefore, provide a comprehensive overview of the structural and functional diversity of nf-κb inhibitors in a non-mammalian model species while also offering starting points for further research into the inflammatory signalling processes in bony fish.

Iκb Proteins from Rainbow Trout Are Encoded on 14 Distinct Genes
Our search of the NCBI gene database for NFKBI orthologs in the rainbow trout Oncorhynchus mykiss (assembly USDA_OmykA_1.1) yielded six nfkbia, two nfkbie, two nfkbid, two nfkbiz, and two bcl3 genes (Table 1). By contrast, the orthologs of NFKBIB and ANKRD42 seemed absent not only in salmonids but in many other teleost fishes. The same number of nfkbi genes was present in the closely related Chinook salmon, Oncorhynchus tshawytscha (assembly Otsh_v2.0) ( Table 1, last column). All nfkbi genes from rainbow trout are present as pairs of ohnologs that most likely arose from a whole-genome duplication, which is also reflected by their location on separate chromosomes [65,66]. Additionally, the nfkbia genes clearly exist as three pairs of duplicated paralogs ( Figure 1): (a) one pair of ohnologous nfkbia genes (a1 and a2 on chromosomes 4 and 8, respectively) is located in the immediate vicinity of the genes hsp90b and psma6; (b) a second nfkbia pair (b1 and b2 on chromosome 19 and 25) both neighbour the gene insm2 and lie in the vicinity of the gene fam177a1; and (c) the third nfkbia gene pair (on chromosomes 10 and 12) has been annotated adjacent to the genes gjd2 and zscan21. The ohnologous nfkbie genes a1 and a2 are both in direct proximity to the genes tmem15b and slc35b2, located on the same chromosomes as the nfkbia genes a1 and a2. The ohnologous nfkbid genes a1 and a2 on chromosomes 2 and 3 share fxyd6 in their spatial vicinity, and the bcl3 ohnologs on the same chromosomes are flanked by the genes cbl and tom40. The two nfkbiz ohnologs a1 and a2 on chromosomes 7 and 18 are adjacent to the genes eed, znf, and epr1. The shorter coding sequences of nfkbia and nfbkie (both between 942 bp and 1191 bp in length) are distributed across 5 to 6 exons, while the coding sequences of nfkbid and nfkbiz are significantly longer (between 1461 bp and 1707 bp in length) and are divided over 9 to 12 exons. The bcl3 ohnologs represent the longest nfkbi sequences (>2000 bp) in rainbow trout and are distributed across 12 exons. Based on automated computational analyses, the latest USDA_OmykA_1.1 assembly of the rainbow trout transcriptome includes one shorter nfkbia transcript variant and four shorter nfkbiz transcript variants, two of which are non-sense mRNAs ( Table 2). These variants most likely arose from exon skipping during splice events, but neither this assumption nor the existence of the predicted transcript isoforms has been experimentally validated. The shorter coding sequences of nfkbia and nfbkie (both between 942 bp and 1191 bp in length) are distributed across 5 to 6 exons, while the coding sequences of nfkbid and nfkbiz are significantly longer (between 1461 bp and 1707 bp in length) and are divided over 9 to 12 exons. The bcl3 ohnologs represent the longest nfkbi sequences (>2000 bp) in rainbow trout and are distributed across 12 exons. Based on automated computational analyses, the latest USDA_OmykA_1.1 assembly of the rainbow trout transcriptome includes one shorter nfkbia transcript variant and four shorter nfkbiz transcript variants, two of which are non-sense mRNAs ( Table 2). These variants most likely arose from exon skipping during splice events, but neither this assumption nor the existence of the predicted transcript isoforms has been experimentally validated. The sequence identity of the ohnologous nfkbi-encoded iκb proteins from rainbow trout (including bcl3) ranges from 82% to 100% (Table S1). The two pairs of the iκbα ohnologs a1/a2 and b1/b2 still share about 60% identity. However, comparison of the iκbα paralogs a1/a2 or b1/b2 versus c1/c2 reveals a sequence identity below 30%. This is a similarly low identity to that shared, for instance, by the paralogs iκbα and iκbε.
A phylogenetic analysis across the amino acid sequences of all IκB proteins from humans and fishes revealed that the ohnologous iκbδ, iκbε, and bcl3 sequences from rainbow trout form separate clusters with their counterparts in the other salmonid fishes, including Chinook salmon O. tshawytscha, Atlantic salmon Salmo salar, and brown trout Salmo trutta (Figure 2a), whereas the iκbζ isoforms instead cluster in a species-specific fashion.
Unexpectedly, a pair of ohnologous iκbα sequences (c1 and c2) cluster with the human IκBβ factor (dark blue section on the left side of the dendrogram in Figure 2a), while the other two pairs of iκbα ohnologs (a1 and a2, as well as b1 and b2) cluster-as previously supposed-with the human IκBα factor. The homology of the iκbα ohnologs c1 and c2 with iκbβ from other teleost species is also reflected by 79 amino acid residues between positions 171 and 390 that are shared with the iκbβ sequences from at least two other fish species, but different in one or both iκbα-a and iκbα-b paralogous pairs (Figure 2b).
The overall structural differences of the iκb proteins from rainbow trout are depicted with three-dimensional models highlighting the well-conserved ankyrin-repeat motifs (Figure 3a-g). A defined number of ankyrin-repeat motifs is a characteristic of all IκB proteins. The iκb proteins listed in the NCBI database for rainbow trout contain six ankyrin repeats, except for two iκbα isoforms (nfkbia-b1, nfkbia-b2) and two iκbζ variants (nfkbiz-a1.3, nfkbiz-a1.4) (Figure 3h). The iκbζ variants a1.3 and a1.4 lack the 6th ankyrin repeat, while the amino acid sequence of the 6th ankyrin repeat of the iκbα isoforms b1 and b2 differs from the canonical motif. In addition, the sequences of the 4th ankyrin repeats of both the iκbδ and iκbζ proteins differ from their counterpart sequences in iκbα and iκbε as well as bcl3.   This neighbour-joining tree was constructed using the Poisson-correction distance model; bootstrap values are given at the nodes of each clade. The shaded underlays label the assignment to the IκB subfamilies iκbα/nfkbia (light blue), iκbβ/nfkbib (darker blue) iκbε/nfkbie (yellow), iκbδ/nkfbid (orange), iκbζ/nfkbiz (green), iκbη/ankrd42 (pink), and bcl3 (purple). (b) Alignment of various amino acid sections of the six putative iκbα/nfkbia sequences from rainbow trout and five selected iκbβ/nfkbib orthologs from electric eel (Electrophorus electricus; XP_026884418), northern pike (Esox lucius; XP_010862954), fathead minnow (Pimephales promelas; XP_039505967), channel catfish, and zebrafish. Green underlay marks amino acid residues that are identical across all selected sequences; blue underlay denotes those residues conserved between nfkbia-c sequences and at least two nfkbib sequences. Red arrows denote residues identical in at least one of the nfkbia-c ohnologs from trout and at least two nfkbib sequences from the other fish species, but different in at least two of the nfkbia-a and nfkbia-b sequences. Amino acid positions given above the alignment refer to the nfkbiac1 sequence. One ohnolog is coloured in light grey, and the other one is coloured in dark grey (or black for nfkbiza-2.1) to indicate structural (dis)similarities. The individual ankyrin-repeat domains are each shown in different colours. The N-and C-termini of each protein are indicated by green and red stretches, respectively. (h) Alignment of the ankyrin-repeat domains of the iκb proteins from rainbow trout and their human orthologs. The six ankyrin-repeat domains are coloured according to the above 3D structures (a-g). Bold characters mark amino acid residues that are well conserved across the aligned IκB sequences. A single underline indicates the sequence used for the expression constructs 'nfkbia-AR12' and 'nfkbie-AR12', while a double underline indicates the sequence used for the expression constructs 'nfkbia-AR456' and 'nfkbie-AR456'. The black square indicates the end of the protein sequences of nfkbiza-1.3 and nfkbiz-a1.4. (For the NCBI accession codes see Figure 2).

IκBα-Encoding nfkbia-a Transcripts Are Most Strongly Expressed in Immune Tissues and Immune-Cell Fractions
Fifteen genes from rainbow trout produce (at least) seventeen transcript variants coding for iκbα, iκbε, iκbδ, iκbζ, or bcl3 (Tables 1 and 2), as identified in the relevant gene database. We determined the transcript levels of the individual nfkbi genes in nine selected tissues and two sorted immune-cell fractions by designing primer pairs common to ohnologs of the nfkbia-a, nfkbia-b, nfkbia-c, and bcl3 genes as well as primer pairs discriminating between the ohnologs of nfkbie, nfkbid, and nfkbiz (Tables 3 and S2). Across the quantified nfkbia genes, the nfkbia-a transcripts had the highest levels in the spleen (4.1 × 10 6 copies), gills (3. 6 × 10 6 copies), head kidney, and trunk kidney (1.1 to 1.7 × 10 6 copies) and exceeded the levels of the nfkbia-b ohnologs by 2.6-to 3.7-fold and the nfkbia-c ohnologs by 37-to 66-fold in the same four tissues (with p < 0.05) (Figure 4a).  across the aligned IκB sequences. A single underline indicates the sequence used for the expression constructs 'nfkbia-AR12' and 'nfkbie-AR12', while a double underline indicates the sequence used for the expression constructs 'nfkbia-AR456' and 'nfkbie-AR456'. The black square indicates the end of the protein sequences of nfkbiza-1.3 and nfkbiz-a1.4. (For the NCBI accession codes see Figure 2).

IκBα-Encoding nfkbia-a Transcripts Are Most Strongly Expressed in Immune Tissues and Immune-Cell Fractions
Fifteen genes from rainbow trout produce (at least) seventeen transcript variants coding for iκbα, iκbε, iκbδ, iκbζ, or bcl3 (Tables 1 and 2), as identified in the relevant gene database. We determined the transcript levels of the individual nfkbi genes in nine selected tissues and two sorted immune-cell fractions by designing primer pairs common to ohnologs of the nfkbia-a, nfkbia-b, nfkbia-c, and bcl3 genes as well as primer pairs discriminating between the ohnologs of nfkbie, nfkbid, and nfkbiz (Tables 3 and S2). Across the quantified nfkbia genes, the nfkbia-a transcripts had the highest levels in the spleen (4.1 × 10 6 copies), gills (3. 6 × 10 6 copies), head kidney, and trunk kidney (1.1 to 1.7 × 10 6 copies) and exceeded the levels of the nfkbia-b ohnologs by 2.6-to 3.7-fold and the nfkbia-c ohnologs by 37-to 66-fold in the same four tissues (with p < 0.05) (Figure 4a).  In the liver, muscle, intestine, heart, and adipose tissue, the nfkbia expression differences were less pronounced and, in part, not significant. High transcript levels were also recorded for the a1 ohnoloe of the nfkbie gene in the gills (7.6 × 10 5 copies), head kidney (7.1 × 10 5 copies), and spleen (6.3 × 10 5 copies), and these levels exceeded those of the nfkbie-a2 ohnologs by a factor of about 2 to 5 (with p < 0.01) (Figure 4b). The transcripts for the nfkbid and nfkbiz ohnologs ranged between about 430 (nfkbiz-a1.3+1.4 in the muscle) and 1.7 × 10 5 copies (nfkbid-a1 in the gills) but showed few significant differences in transcript levels between the ohnologs (Figure 4c,d). With regard to tissue-specific expression patterns, the transcript levels of nfkbia-a and nfkbie were significantly higher in immune-relevant tissues, including head kidney, gill, and spleen, but (almost) no significant differential expression was observed for nfkbia-c, nfkbid-a1, nfkbid-a2, or nfkbiz-a1.3+1.4 between tissues.
The tissue-wise expression profiles suggested that organs rich in immune cells generally have high levels of nfkbi transcripts. For this reason, we quantified the nfkbi transcripts in (i) a non-myeloid (mAb21N) fraction enriched with T-and B-lymphocytes, natural killer-like cells, and thrombocytes and (ii) a myeloid (mAb21P) fraction enriched with granulocytes, monocytes/macrophages, and dendritic cells from the head kidney of the rainbow trout (O. mykiss) ( Figure 5). also recorded for the a1 ohnoloe of the nfkbie gene in the gills (7.6 × 10 5 copies), head kidney (7.1 × 10 5 copies), and spleen (6.3 × 10 5 copies), and these levels exceeded those of the nfkbie-a2 ohnologs by a factor of about 2 to 5 (with p < 0.01) (Figure 4b). The transcripts for the nfkbid and nfkbiz ohnologs ranged between about 430 (nfkbiz-a1.3+1.4 in the muscle) and 1.7 × 10 5 copies (nfkbid-a1 in the gills) but showed few significant differences in transcript levels between the ohnologs (Figure 4c,d). With regard to tissue-specific expression patterns, the transcript levels of nfkbia-a and nfkbie were significantly higher in immune-relevant tissues, including head kidney, gill, and spleen, but (almost) no significant differential expression was observed for nfkbia-c, nfkbid-a1, nfkbid-a2, or nfkbiz-a1.3+1.4 between tissues.
The tissue-wise expression profiles suggested that organs rich in immune cells generally have high levels of nfkbi transcripts. For this reason, we quantified the nfkbi transcripts in (i) a non-myeloid (mAb21N) fraction enriched with T-and B-lymphocytes, natural killer-like cells, and thrombocytes and (ii) a myeloid (mAb21P) fraction enriched with granulocytes, monocytes/macrophages, and dendritic cells from the head kidney of the rainbow trout (O. mykiss) ( Figure 5). Again, the level of nfkbia-a transcripts was the highest (19 × 10 6 and 4 × 10 6 copies in the mAb21P and mAb21N fractions, respectively) compared to the other nfkbi transcripts, but this difference was only statistically significant for the mAb21P fraction (p < 0.0001). The levels of nfkbiz-a1.3+1.4 were the lowest (3.6 × 10 4 to 4.5 × 10 4 copies in the mAb21P and mAb21N fractions, respectively).

Iκbα and iκbε Localise to the Cytoplasm as well as the Nucleus of Salmonid Model CHSE-214 Cells
Their prominent expression in the immune tissues of rainbow trout suggested that nfkbia-a2 and nfkbie-a2 were appropriate iκb factors to seek the first insights into the regulation of nf-κb pathways in salmonid fish. Again, the level of nfkbia-a transcripts was the highest (19 × 10 6 and 4 × 10 6 copies in the mAb21P and mAb21N fractions, respectively) compared to the other nfkbi transcripts, but this difference was only statistically significant for the mAb21P fraction (p < 0.0001). The levels of nfkbiz-a1.3+1.4 were the lowest (3.6 × 10 4 to 4.5 × 10 4 copies in the mAb21P and mAb21N fractions, respectively).

Iκbα and iκbε Localise to the Cytoplasm as well as the Nucleus of Salmonid Model CHSE-214 Cells
Their prominent expression in the immune tissues of rainbow trout suggested that nfkbia-a2 and nfkbie-a2 were appropriate iκb factors to seek the first insights into the regulation of nf-κb pathways in salmonid fish.
The ankyrin repeat is the signature motif of all IκB proteins (cf. Figure 3). We transiently overexpressed each of the three construct variants of iκbα or iκbε in CHSE-214 cells, including (i) the full sequence as well as truncated variants comprising (ii) the two N-terminal ankyrin repeats and (iii) the three C-terminal ankyrin repeats (Figure 3h). Confocal imaging indicated a differential localisation of the different iκbα or iκbε constructs ( Figure 6). The concentration of the full-length iκbα was higher in the cytoplasm than in the nucleus (Figure 6a), while both iκbα-AR12 and iκbα-AR456 proteins were localised to a greater extent in the nucleus than in the cytoplasm (Figure 6b,c). The full-length iκbε factor and its derivative iκbε-AR12 and iκbε-AR456 proteins seemed to be evenly distributed in both the cytoplasm and nucleus (Figure 6d-f). The ankyrin repeat is the signature motif of all IκB proteins (cf. Figure 3). We transiently overexpressed each of the three construct variants of iκbα or iκbε in CHSE-214 cells, including (i) the full sequence as well as truncated variants comprising (ii) the two N-terminal ankyrin repeats and (iii) the three C-terminal ankyrin repeats (Figure 3h). Confocal imaging indicated a differential localisation of the different iκbα or iκbε constructs ( Figure 6). The concentration of the full-length iκbα was higher in the cytoplasm than in the nucleus (Figure 6a), while both iκbα-AR12 and iκbα-AR456 proteins were localised to a greater extent in the nucleus than in the cytoplasm (Figure 6b,c). The full-length iκbε factor and its derivative iκbε-AR12 and iκbε-AR456 proteins seemed to be evenly distributed in both the cytoplasm and nucleus (Figure 6d-f).

Iκbα and iκbε Reduce the Basal and Stimulated nf-κb Activity
We used the six iκbα or iκbε expression constructs described in Sections 2.3 and 4.2 to assess their impact on the nf-κb activity in CHSE-214 cells. The overexpression of the full-length iκbα factor (1000 ng) robustly and significantly reduced the basal nf-κb activity down to 0.09-fold (p = 0.0004) (Figure 7a) compared to the non-transfected controls. Similarly, the overexpression of full-length iκbε (1000 ng) resulted in only 0.06-fold basal nf-κb activity (with p < 0.0001) compared to the controls (Figure 7b). In contrast to the full-length constructs, the two N-terminal ankyrin repeats or the three C-terminal ankyrin repeats of iκbα enhanced the basal nf-κb activity by 11.8-fold (p = 0.08) or 35.0-fold (p < 0.00001), respectively (Figure 7c). The pattern for the truncated iκbε constructs differed in the stronger nf-κb activation (5.2-fold; p = 0.23) caused by the N-terminal ankyrin repeats than by the C-terminal domains (2.3-fold; p = 0.79) ( Figure  7d).
Stimulation of the non-transfected CHSE-214 cells with the fungal cell wall component zymosan doubled the nf-κb activity (2.0-fold; p < 0.0001) compared to the basal state (Figure 7a,b). Increasing the amounts of overexpressed iκbα factor from rainbow trout

Iκbα and iκbε Reduce the Basal and Stimulated nf-κb Activity
We used the six iκbα or iκbε expression constructs described in Sections 2.3 and 4.2 to assess their impact on the nf-κb activity in CHSE-214 cells. The overexpression of the fulllength iκbα factor (1000 ng) robustly and significantly reduced the basal nf-κb activity down to 0.09-fold (p = 0.0004) (Figure 7a) compared to the non-transfected controls. Similarly, the overexpression of full-length iκbε (1000 ng) resulted in only 0.06-fold basal nf-κb activity (with p < 0.0001) compared to the controls (Figure 7b). In contrast to the full-length constructs, the two N-terminal ankyrin repeats or the three C-terminal ankyrin repeats of iκbα enhanced the basal nf-κb activity by 11.8-fold (p = 0.08) or 35.0-fold (p < 0.00001), respectively (Figure 7c). The pattern for the truncated iκbε constructs differed in the stronger nf-κb activation (5.2-fold; p = 0.23) caused by the N-terminal ankyrin repeats than by the C-terminal domains (2.3-fold; p = 0.79) (Figure 7d). dose-dependently lowered this stimulated nf-κb activity down to 0.66-fold (20 ng expression vector; p < 0.05) and 0.05-fold (1000 ng expression vector; p < 0.001) compared to the non-transfected cells (Figure 7a). Again, the overexpression of iκbε had a very similar effect on the stimulated nf-κb activity, as observed for iκbα (p < 0.0001) (Figure 7b). Cells overexpressing either the N-terminal or the C-terminal ankyrin repeats of iκbα or iκbε showed an already enhanced basal nf-κb activity, and stimulation induced a further increase in active nf-κb. Figure 7. Overexpression of (a,b) gfp-tagged iκbα (green fluorescence) or (c,d) plum-tagged iκbε constructs (red fluorescence) in salmonid CHSE-214 cells. The luciferase activity of the ELAM-reporter vector was quantified in CHSE-214 cells co-expressing one of the six iκb constructs expressing (a,b) full-length iκbα and its truncated derivatives iκbα-AR12 and iκbα-AR456 and (c,d) full-length iκbε and its truncated derivatives iκbε-AR12 and iκbε-AR456. The concentrations of the iκb-expressing vector used for the transfection of the cells are indicated on the abscissa. Bars denote the mean values ± SEM. Statistical significance was assessed using two-way ANOVA (*, p < 0.05; ***, p < 0.001; ****, p < 0.0001). Figure 7. Overexpression of (a,b) gfp-tagged iκbα (green fluorescence) or (c,d) plum-tagged iκbε constructs (red fluorescence) in salmonid CHSE-214 cells. The luciferase activity of the ELAMreporter vector was quantified in CHSE-214 cells co-expressing one of the six iκb constructs expressing (a,b) full-length iκbα and its truncated derivatives iκbα-AR12 and iκbα-AR456 and (c,d) fulllength iκbε and its truncated derivatives iκbε-AR12 and iκbε-AR456. The concentrations of the iκb-expressing vector used for the transfection of the cells are indicated on the abscissa. Bars denote the mean values ± SEM. Statistical significance was assessed using two-way ANOVA (*, p < 0.05; ***, p < 0.001; ****, p < 0.0001).
Stimulation of the non-transfected CHSE-214 cells with the fungal cell wall component zymosan doubled the nf-κb activity (2.0-fold; p < 0.0001) compared to the basal state (Figure 7a,b). Increasing the amounts of overexpressed iκbα factor from rainbow trout dosedependently lowered this stimulated nf-κb activity down to 0.66-fold (20 ng expression vector; p < 0.05) and 0.05-fold (1000 ng expression vector; p < 0.001) compared to the non-transfected cells (Figure 7a). Again, the overexpression of iκbε had a very similar effect on the stimulated nf-κb activity, as observed for iκbα (p < 0.0001) (Figure 7b). Cells overexpressing either the N-terminal or the C-terminal ankyrin repeats of iκbα or iκbε showed an already enhanced basal nf-κb activity, and stimulation induced a further increase in active nf-κb.
Admittedly, the expression constructs used for the above overexpression studies did not only encode distinct iκb factors but also a fluorescent protein (gfp or plum). To exclude the possibility that the fluorescent protein had an additional effect on the nf-κb activity, we verified that expression vectors encoding either iκbα or iκbε coupled to a fluorescent protein (gfp or plum) and an expression plasmid encoding either iκbα or iκbε without fluorescent protein had a similar effect on the nf-κb cellular activity (Figure 8). Admittedly, the expression constructs used for the above overexpression studies did not only encode distinct iκb factors but also a fluorescent protein (gfp or plum). To exclude the possibility that the fluorescent protein had an additional effect on the nf-κb activity, we verified that expression vectors encoding either iκbα or iκbε coupled to a fluorescent protein (gfp or plum) and an expression plasmid encoding either iκbα or iκbε without fluorescent protein had a similar effect on the nf-κb cellular activity (Figure 8). Notably, zymosan stimulation induced a significantly different nf-κb activity in CHSE-214 cells transfected with 200 ng iκbε tagged with plum versus iκbε without plum, but this differential activity pattern was not consistent across the other concentrations. Therefore, we do not assume any significant influence of the fluorescent tag on the performance of the expressed iκb factor.
Having established that both iκbα and iκbε factors from rainbow trout significantly reduced the nf-κb activity in vitro, we used qPCR to test whether the overexpression of both factors would modulate the transcription of a panel of nf-κb-dependent immune genes in the same CHSE-214 model cells. The transcript levels of the seven selected immune genes were similar between untransfected cells and cells overexpressing iκbα or iκbε (Figure 9). Notably, zymosan stimulation induced a significantly different nf-κb activity in CHSE-214 cells transfected with 200 ng iκbε tagged with plum versus iκbε without plum, but this differential activity pattern was not consistent across the other concentrations. Therefore, we do not assume any significant influence of the fluorescent tag on the performance of the expressed iκb factor.
Having established that both iκbα and iκbε factors from rainbow trout significantly reduced the nf-κb activity in vitro, we used qPCR to test whether the overexpression of both factors would modulate the transcription of a panel of nf-κb-dependent immune genes in the same CHSE-214 model cells. The transcript levels of the seven selected immune genes were similar between untransfected cells and cells overexpressing iκbα or iκbε (Figure 9). Stimulation with zymosan for four hours significantly (p < 0.01) increased the transcript levels of characteristic inflammatory markers, such as il1b and cxcl8, but also of nfkbia (LOC112249975). Nevertheless, the differences in transcript levels of the induced immune genes in non-transfected cells versus cells expressing iκbα or iκbε were not statistically significant after stimulation with zymosan.

Discussion
Previous reports have suggested that rainbow trout possess one [53] or four [59] functional iκbα-encoding genes, but our research at the NCBI gene database revealed six nfkbia gene copies on different chromosomes of O. mykiss. Only one NFKBIA gene is present in mammals and two nfkbia paralogs have been characterised in several fish species, including zebrafish D. rerio [58], rock bream Oplegnathus fasciatus [55], orange-spotted grouper Epinephelus coioides [57], and blunt snout bream Megalobrama amblycephala [59]. It is rather unlikely that the additional teleost-specific whole-genome duplication in fish yielded three and not two nfkbia paralogs in salmonids, which then underwent an additional genome duplication [67] that eventually produced three pairs of nfkbia ohnologs. Several mechanisms could explain why certain genes are present in more copies than expected. For instance, the tandem duplication of genes arises from the un- Stimulation with zymosan for four hours significantly (p < 0.01) increased the transcript levels of characteristic inflammatory markers, such as il1b and cxcl8, but also of nfkbia (LOC112249975). Nevertheless, the differences in transcript levels of the induced immune genes in non-transfected cells versus cells expressing iκbα or iκbε were not statistically significant after stimulation with zymosan.

Discussion
Previous reports have suggested that rainbow trout possess one [53] or four [59] functional iκbα-encoding genes, but our research at the NCBI gene database revealed six nfkbia gene copies on different chromosomes of O. mykiss. Only one NFKBIA gene is present in mammals and two nfkbia paralogs have been characterised in several fish species, including zebrafish D. rerio [58], rock bream Oplegnathus fasciatus [55], orange-spotted grouper Epinephelus coioides [57], and blunt snout bream Megalobrama amblycephala [59]. It is rather unlikely that the additional teleost-specific whole-genome duplication in fish yielded three and not two nfkbia paralogs in salmonids, which then underwent an additional genome duplication [67] that eventually produced three pairs of nfkbia ohnologs. Several mechanisms could explain why certain genes are present in more copies than expected. For instance, the tandem duplication of genes arises from the unequal exchange between sister chromo-somes [68]. Our structural and phylogenetic analyses suggest that one pair of the putative nfkbia ohnologs is the supposedly lost nfkbib gene. The sequence identity between the iκbα ohnologs a1/a2 or b1/b2 versus c1/c2 is comparably low, as reflected by our phylogenetic analysis that assigned the a1/a2 and b1/b2 pairs to the α-subfamily of IκB factors, while the c1/c2 pair was assigned to the β-subfamily of IκB factors. However, the genes flanking the human NFKBIB gene (including SIRT2, RINL, SARS2, and CCR2) [69] are either not present in rainbow trout or they are located on different chromosomes where they flank different genes. Conversely, the genes flanking nfkbia-c (presumably nfkbib) in rainbow trout (such as gjd2, rab6a and zscan21) are located on different chromosomes in the human genome, where they do not flank the same genes as they do in the trout genome. However, nfkbib genes are present in various teleost fishes, and nfkbib from the northern pike Esox lucius, for instance, is adjacent to the same genes as nfkbia-c (presumably nfkbib) in the rainbow trout genome (such as rab6a or gjd1a). The NCBI gene database (accessed on 14 December 2022) lists one nfkbib copy in 24 species, including 6 representatives of the Cypriniformes (carp fishes), 4 representatives of the Siluriformes (catfishes), 3 representatives each of the Characiformes (characins), and Clupeiformes (herring) and 2 representatives of the Osteoglossiformes (elephantfishes). In stark contrast, nfkbia and nfkbie are encoded in the genomes of more than 120 sequenced fish species, and nfkbiz is present in 115 species. The number of fish species possessing nfkbid is significantly lower at 63. We conclude from these indications that iκbβ has certainly been lost in many of the fish species sequenced thus far but not in salmonid fishes.
Three previous studies provide kaleidoscopic insights into the distinct characteristics and functions of teleostean iκbα paralogs: (a) iκbα-a is downregulated in the liver of the rock bream O. fasciatus a few hours after stimulation with flagellin, while its paralog iκbα-b is upregulated [55]; (b) the iκbα-a protein from the orange-spotted grouper Epinephelus coioides is distributed across the cytoplasm and nucleus, while its paralog iκbα-b mainly localises to the cytoplasm [57]; (c) the tnf-stimulated resynthesis of iκbα-b from the zebrafish D. rerio takes twice as long as the production of its paralog iκbα-a [58]. Our data on the iκbα paralogs and ohnologs of rainbow trout reveal that the expression of the iκbα-a-encoding gene is significantly higher in immune organs and in a head-kidney cell fraction enriched with granulocytes, monocytes/macrophages, and dendritic cells compared to its paralogs. Similar to what is observed in mammals, the iκb-encoding nfkbi genes from rainbow trout are constitutively expressed, albeit in a tissue-specific fashion [17,18,41]. These tissuespecific expression patterns obviously vary in different fish species, as demonstrated at least for nfkbia. The prominent nfkbia transcript level in the spleen of rainbow trout is in line with findings in the Japanese eel Anguilla japonica [63] and the mandarin fish Siniperca chuatsi [64], but it contrasts with the rather low splenic expression in the rock bream O. fasciatus [55], the blunt snout bream M. amblycephala [59] or the half-smooth tongue sole Cynoglossus semilaevis [60]. No comparable datasets are presently available for the expression profiles of the other nfkbi transcripts. In addition to the tissue-specific expression patterns, differential nfkbia transcript levels have been identified as indicative parameters for immune stimulation [70][71][72], exposure to toxic substances [73], or consumption of different diets [74,75] in diverse fish species, including rainbow trout [75,76]. The present study also confirms the significant upregulation of nfkbia, but not nfkbie, after in vitro fungal stimulation.
Iκbα and iκbε from rainbow trout localise to the cytoplasm and nuclei of unstimulated cells, as observed for IκBα orthologs from mammals [28] and bony fish [63]. Mammalian IκBε has been detected in the cytoplasm as well as in the nucleus [77], while the spatial distribution of iκbε in teleostean cells has not yet been analysed. In contrast to the full iκbα-protein from rainbow trout, the truncated iκbα versions were mostly located in the nucleus.
The overexpression of iκbα from rainbow trout blocked the basal and stimulated nf-κb activity in vitro, which is in line with many other reports on similar reporter-gene experiments in fish cells [56,57,63]. This again proves that the function of IκBα as an efficient regulator of NF-κB signalling is well preserved across vertebrates. The effectiveness of iκbε from rainbow trout was similar to that of iκbα in terms of restricting nf-κb activity, but no comparative data on the biological activity of its orthologue in other teleost species are presently available.
Our expression constructs encoding either the first two or the last three ankyrin repeat motifs of iκbα and iκbε from rainbow trout increased the nf-κb activity, whereas the fulllength iκbα and iκbε proteins did not. Moreover, we observed that the truncated iκbα variants had higher concentrations in the nucleus than in the cytoplasm, whereas the opposite was apparent for their full-length counterparts. The two N-terminal ankyrin repeat domains of the mammalian IκBα are known to contact the nuclear localisation signal of rela/nf-κB p65 [35,78,79]. Thus, the N-terminal domains of iκbα from rainbow trout quite plausibly have significant involvements in the oscillations of nf-κB. The three C-terminal ankyrin repeat domains in the mammalian IκB ortholog interact with the Nterminal domain of the Rel homology region or the PEST region of rela/nf-κB p65 and/or nfkb1/nf-κB p50 [35,78,79]. Accordingly, the ankyrin repeat domains seem to fulfil specific functions in fish, as they do in mammals, and their number and position may be crucial for their ultimate function.
The number of ankyrin-repeat motifs is probably not a criterion that discriminates iκbα from iκbβ. Iκbαa from orange-spotted grouper E. coioides contains five ankyrin repeat motifs, while iκbαb contains six ankyrin repeats [57]. In rainbow trout, the paralog pairs iκbα-a and iκbα-c/iκbβ possess six ankyrin-repeat motifs, while iκbα-b contains only five prototypical ankyrin-repeat motifs. Although the ankyrin-repeat domain is the most conserved feature of IκB proteins, it is found in many other proteins. Therefore, we can assume that more ankyrin repeat-containing proteins affect the activity of NF-κB than are currently known. In mammals, the ankyrin repeat-containing proteins INK4 and myotrophin, for instance, have been proven to modulate the efficacy of NF-κB functions [80,81], but their role in inflammatory processes in fish is not yet known.
In summary, the intensive structural comparisons presented here demonstrate that iκbβ does exist in salmonid fish, but its expression is significantly lower than that of the paraloguous iκbα proteins. Our comprehensive overexpression studies in trout cells provide insights into the regulation potential of a set of nf-κb inhibitors from rainbow trout, thereby providing the first functional results for iκbε in lower vertebrates. In ongoing studies, we are investigating the interplay between nf-κb inhibitors and nf-κb/rel factors from trout under defined environmental conditions.

Quantitative PCR (qPCR) Analysis
We recorded the expression of nfkbia-a, nfkbia-b, nfkbi-c, nfkbie, nfkbid, and nfkbiz in nine tissues (adipose tissue, gills, head kidney, heart, intestine, liver, muscle, spleen and trunk kidney) and sorted cells from rainbow trout. All analyses were conducted using animal materials left over from previous analyses [82]. We used the monoclonal antibody mAb21 to separate an mAb21-positive head-kidney cell fraction consisting of >95% myeloid cells from a mAb21-negative fraction consisting mostly of B-and T-lymphocytes, as well as thrombocytes. RNA was isolated from tissues and sorted cells using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and the ISOLATE II RNA Micro Kit (Bioline/Meridian Bioscience, Luckenwalde, Germany), respectively, including an in-column DNase treatment. We also profiled the expression of various Chinook salmon-specific immune genes (il1b, tnf, cxcl8a, cxcl8b, tgfb, il10, nfkbia and nfkbie) in transfected, stimulated CHSE-214 cells (derived from Chinook salmon [O. tshawytscha] embryos). After stimulation, the transfected CHSE-214 cells were washed twice with phosphate-buffered saline and then harvested by a 20-min incubation in lysis buffer (RNeasy Mini Kit; Qiagen, Hilden, Germany).
After isolation, RNA was reverse transcribed into cDNA using the SensiFAST cDNA Synthesis Kit (Bioline/Meridian Bioscience). Subsequently, the quantity of the nfkbi transcripts was selectively recorded by a panel of exon-skipping oligonucleotide primers specific for rainbow trout (Pyrosequencing Assay Design software v.1.0.6; Biotage, Uppsala, Sweden; Table 2). These primers were either common for both nfkbi ohnologs (nfkbia-a, nfkbia-b nfkbi-c, and bcl3-a) or discriminated between nfkbi ohnologs (nfkbie, nfkbid, and nfkbiz). Of particular note, no discriminating primers could be designed for the individual nfkbiz gene variants a1 and a2 due to the high sequence identity (99-100%) (Table S1); instead, we derived common primers for each of the two similar nfkbiz transcript variants nfkbiz-a1.1/a2.1 and nfkbiz-a1.3/a1.4. The primer pairs listed in Table 3 amplified fragments between 86 and 191 nucleotides in length. Rps5 (ribosomal protein S5) and eef1a1 (eukaryotic translation elongation factor) were used as reference genes to normalise the expression data. The qPCR analyses were conducted using the LightCycler-96 system (Roche, Basel, Switzerland) according to the following programme: initial denaturation at 95 • C for 5 min, followed by 40 cycles of denaturation at 95 • C for 30 s, primer annealing at 60 • C for 15 s, elongation at 72 • C for 15 s, and the fluorescence measurement at 72 • C for 10 s. The amplicon quality was assessed by gel electrophoresis and melting-curve analysis. In addition, we checked the primer specificity by sequencing the amplicons (Azenta Life Sciences, Griesheim, Germany). The qPCR data were extracted using the LightCycler-96 analysis software v. 1.1.0.1320 (Roche).

Construction of Nfkbi-Expression Constructs
Three different vectors were used to express the complete ORFs of selected nfkbi genes (nfkbia and nfkbie) or distinct fragments of those genes (i.e., ankyrin repeat 1 and 2 or ankyrin repeat 4, 5, and 6) from rainbow trout. These vectors included the mammalian expression vector v280 [83] and two v280 derivatives attaching the red fluorescent protein mPlum or a green fluorescent protein (gfp) at the 3 end of the inserted target fragments. These fragments were generated using Platinum Taq High-Fidelity DNA polymerase (Thermo Fisher Scientific, Bremen, Germany) and oligonucleotide primers linked with specific restriction sites (Table 4). The Biometra TAdvanced cycler (Analytik Jena, Jena, Germany) was used to amplify the gene fragments according to the following programme: initial denaturation at 94 • C for 2 min, followed by 40 cycles of denaturation at 94 • C for 30 s, primer annealing at 60 • C for 30 s, elongation at 72 • C for 2 min, and a final extension step at 68 • C for 5 min. Amplicons were inserted into the above expression vectors by double digestion with HindIII and EcoRI (nfkbia) or BglII (nfkbie). All expression vectors were sequenced and checked for correct assembly before use.

Cell Transfection, Luciferase Assay and Confocal Microscopy
The salmonid cell line CHSE-214 was cultured as described previously [83]. The CHSE-214 cells were transfected in six-well plates with a total of 2050 ng endotoxin-free prepared DNA (ZymoPure II Plasmid Maxi Prep Kit, ZymoResearch, Freiburg, Germany) using X-tremeGENE HP DNA Transfection Reagent (Roche, Mannheim, Germany). The co-transfection assays contained 50 ng of the NF-κB-responsive promoter (endothelialleukocyte adhesion molecule)-reporter (luciferase) construct ELAM-1-luc, defined concentrations (20 ng-1000 ng) of nfkbi expression vectors, and varying amounts of empty-vector DNA to ensure that the total DNA amount per assay remained constant.
For the stimulation experiments, co-transfected cells were split into 24-well plates: Three wells per row remained unstimulated, while the other three wells were challenged with 1 mg/mL zymosan from Saccharomyces cerevisiae (tlrl-zyn; Invivogen, Toulouse, France) for 4 h or 24 h. After incubation, the cell lysates were collected, and the luciferase activity of each assay was measured using the Dual-Luciferase Reporter Assay System (Promega, Mannheim, Germany) at the Lumat LB9501 luminometer (Berthold, Bad Wildbad, Germany). The resulting relative light units were normalised by the protein concentrations of the CHSE-214 cell extracts. Each transfection experiment was measured in triplicate and conducted at least twice.

Data Analysis
The qPCR data was normalised against the reference genes and based on gene-specific standard curves, and the individual copy numbers were calculated (R 2 > 0.99; 10 7 -10 3 copies per 5 µL). The GraphPad Prism software (v9.1.0) was used for the statistical analysis of the normalised qPCR data. Significant differences between the different tissues/cell fractions were assessed using two-way analysis of variance (ANOVA) followed by a Holm-Šídák's post-hoc test to correct for multiple comparisons. A parametric t-test conducted using GraphPad Prism software v.9.5.1 was run to evaluate the statistical significance of the reporter-gene measurements. p-values less than 0.05 were considered statistically significant.
A phylogenetic dendrogram was reconstructed with the neighbour-joining method based on log-corrected distances and optimised manually. Node robustness was evaluated on a bootstrap analysis based on 1000 iterations. SMART (Simple Modular Architecture Research Tool) [86] was applied to identify motifs and domains of the IκB proteins. The three-dimensional iκb protein structures were predicted using I-TASSER (Iterative Threading ASSEmbly Refinement) [87] in complement with UCSF ChimeraX v.1.1 [88].  Institutional Review Board Statement: All BioFia experiments were approved by the animal welfare officer of the 'Fraunhofer IMTE Büsum' and the local authority of Schleswig-Holstein, according to the German animal welfare law (approval ID: NTP-ID 00043858-1-0). Note: the present study exclusively utilised samples from previous experiments [51,82] following the 3R principles in science.

Data Availability Statement:
The qPCR and reporter-gene data generated during the current study are not publicly available but are available on request. The nucleotide and amino acid sequences of the analysed IκB factors are available in the NCBI database; the respective accession numbers are provided in the manuscript.