Transcriptional Regulation and Signaling of Type IV IFN with Identification of the ISG Repertoire in an Amphibian Model, Xenopus laevis

This article is featured in Top Reads, p.1625

Interferons belong to the class II α-helical cytokine family and are involved in antiviral immunity (1). The higher mammalian IFN family is known to be classified into three types, that is, I, II and III, based on the difference of sequence feature, gene locus, phylogeny, crystallographic structure, and receptor complex (1). Mammalian “intronless” type I IFNs consist of at least 10 subtypes, including IFN-α, IFN-β, IFN-δ, IFN-ε, IFN-ζ, IFN-κ, IFN-μ, IFN-ν, IFN-τ, and IFN-ω, which are located on a conserved PTPLAD2/type I IFNs/MTAP gene cluster, except IFN-κ, which has an intron (2, 3). Interestingly, a much more complicated IFN system has been discovered in ectothermic vertebrates, such as fish and amphibians. The identification of conservative intron-containing type I IFNs in fish and amphibians on an alternative locus suggests that amniote type I IFN genes underwent the event of intron loss due to possible retrotransposition (4, 5). In fact, an expansion of the type I IFN gene family exists in Xenopus, which contains both intron-containing and intronless members (6–8), and the origin of intronless type I IFN genes in Xenopus has been explained as an amphibian-specific independent retrotransposition (5). Although the IFN-α and IFN-β receptor genes (IFN-αRs) have been duplicated in fish, vertebrate type I IFNs share the homologous receptor systems, for example, IFN-αR2-1/IFN-αR2-2/IFN-αR1 in fish and IFN-αR1/IFN-αR2 in humans (1, 9). Unlike type I IFNs, most of the vertebrate type II IFN subfamily linked to DYRK2 is located on a highly conserved locus, despite the presence of a single copy of the IFN-γ gene in vertebrates except in fish (10). It has been reported that most vertebrate IFN-γ uses the homologous receptor systems, that is, IFN-γR1/2 (1, 9). Mammal type III IFNs (IFN-λ) with a multiexon structure are situated on a conserved SYCN/type III IFNs/LRFN1 locus and use the receptor system of IFN-λR1 and IL-10R2 (11, 12). IFN-λ genes have also been identified in chickens, green anoles, X. laevis, and cartilaginous fish, but not in teleost fish (13, 14). It is noteworthy that Xenopus has an expanded type III IFN family with both intron-containing and intronless genes (6). Recently, an unannotated IFN subfamily was discovered and designed as IFN-υ in zebrafish and X. laevis (15). The IFN-υ was identified as a new type IFN, that is, a type IV IFN subfamily in vertebrates from fish to primitive mammals, which differs from type I, II, and III IFNs in consideration of sequence feature, gene locus, phylogeny, and receptor complex (15). Although the numbers of both Xenopus type I and III IFNs are expanded, ifnu was found as a single copy gene in Xenopus, as well as in zebrafish (8, 15). IFN-υ and type I IFNs have the similar C-terminal sequences, and multiexon organization of the IFN-υ gene is similar to fish type I IFN and multiexon type III IFNs in vertebrates, but not to type I IFNs in amniotes (15). Moreover, most vertebrate IFN-υ genes are located on the unique and highly conserved locus, being distinct from all other three type IFNs, and IFN-υ is proven as a monophyletic group supported by phylogenetic analysis (15). Importantly, the receptor system of IFN-υ is comprised of IFN-υR1 and IL-10R2, the latter of which is shared by type III IFNs (15). Although antiviral activity of IFN-υ has been reported in zebrafish and X. laevis, it is still not clear how IFN-υ participates in antiviral immunity. Dissection of the transcriptional mechanism and signaling of IFN-υ should be immunologically and evolutionarily important for understanding the functional diversity of IFNs in vertebrates.

In the antiviral immune response, vertebrate type I and III IFNs are rapidly upregulated at the transcription level and serve as the most core IFNs when induced by viruses (1, 11, 16), and induced expression of type I and III IFNs has been reported following the infection of the ranavirus FV3 in amphibians (17, 18). The induced expression of mammalian type I/III IFNs depends mainly on the activation of transcription factors, including IFN regulatory factor (IRF) and NF-κB family members, which are initiated by pattern recognition receptors after recognition of virus components (1, 11, 16). The mammalian IRF family members involved in the induction of type I/III IFNs include mainly IRF3 and IRF7, as well as IRF1 (1, 11, 16). DNA-binding domains at the N-terminal region of these mammalian IRFs are responsible for binding to the positive regulatory domains (PRDs; PRD I or III) and IFN-sensitive responsive element (ISRE) sites in proximal promoter of type I and III IFNs (1, 11, 19). The mammalian p65 molecule encoded by the RELA gene belongs to the NF-κB family and can bind to PRD II/NF-κB sites in promoters of type I/III IFNs through its N-terminal Rel homology domain (RHD) in the form of homodimer or heterodimer with p50 to regulate type I/III IFN expression (11, 20, 21). Evolutionarily, the IRF members (IRF1, IRF3, and IRF7) and p65 have been reported to be involved in the regulation of type I IFN expression in fish and amphibians (4, 9), although little is known about the regulation of type III IFNs in lower vertebrates.

Importantly, upon viral induction, mammalian IFNs can be secreted extracellularly to act on infected cells to kill viruses, and on uninfected cells to establish defense through the activation of classical JAK-STAT signaling (22, 23). Both mammalian type I and III IFNs are able to phosphorylate STAT1 and STAT2, which can interact with IRF9 to form IFN-stimulated gene (ISG) factor 3 (ISGF3) (11). The ISGF3 complex can be transferred into nucleus, where it binds to ISRE in promoters of ISGs to regulate their expression (11). About 300 ISGs have been identified in mammals, many of which possess a variety of functions ranging from virus inhibition to immune cell activation (24, 25). In zebrafish, due to the expansion of gene families, >400 ISGs have been identified, which contain, in comparison with mammalian ISGs, 72 orthologous groups, including RIG-like receptors (RLRs), MX dynamin-like GTPase (MX), radical S-adenosyl methionine domain containing 2 (RSAD2/Viperin), tripartite motif (TRIM) containing 25 (TRIM25), IRF7, transporter for Ag processing 1 (TAP1), IFN-induced protein with tetratricopeptide repeats (IFITs), and others, and contain 14 fish-specific gene families, such as fish novel TRIMs (finTRIMs, FTRs), grass carp reovirus (GCRV)–induced genes (GIG2s), and fish-specific NOD-like receptors (NLRs) (26). These pluripotential fish ISGs are involved in pathogen recognition, enzyme activation or catalysis, transcriptional regulation, Ag presentation, and other biological processes (26). For example, the finTRIM family has expanded dramatically to form distinct subgroups (27). Although most finTRIM family members, similar to TRIMs, possess RING finger, B-boxes, coiled-coil, and the PRY-SPRY domain, fish FTR83 and FTRCA1, belonging to different subgroups, play a positive or a negative regulatory role in antiviral immunity through the promotion or abrogation of type I IFN expression, respectively (28, 29). In amphibians, the well-known Mx and Viperin genes have been identified as type I– and type IV–induced ISGs (5, 15). However, the repertoire of ISGs is largely unknown in amphibians, especially in relation to the effect of type IV IFN. On the basis of a previous report on type IV IFN receptor subunits, IFN-υR1 and IL-10R2 in amphibians (15), the signaling pathway of IFN-υ was investigated in this study. Furthermore, promoter elements of IFN-υ were analyzed, and the transcriptional mechanism was revealed in the amphibian model, X. laevis. The repertoire of ISGs was then characterized in the X. laevis A6 cell line with the identification of expanded ISG families and amphibian novel TRIM protein (AMNTR) family. Using AMNTR50 as an example, its role in negatively regulating antiviral immunity was further investigated in this study.

The epithelioma papulosum cyprini (EPC, GDC174) (30) and human embryonic kidney 293T (HEK293T, GDC0187) cells were purchased from the China Center for Type Culture Collection and maintained in MEM (Life Technologies) and DMEM (Life Technologies) supplemented with 10% FBS (Life Technologies) in a CO₂ (5%) incubator (Heracell 150i, Thermo Fisher Scientific) at 28 and 37°C, respectively. The X. laevis kidney epithelium A6 cell line (CCL-102) was purchased from American Type Culture Collection and cultured in 75% NCTC-109 medium (Life Technologies) supplemented with 10% FBS (Life Technologies) and 15% distilled water at 26°C in a CO₂ (5%) incubator (WCI-180, Wiggens). FV3 (VR-567) was also obtained from American Type Culture Collection and propagated in EPC cells, and viral titers were detected by using a plaque assay as previously described (4, 5). All experiments were approved by the Institute of Hydrobiology, Chinese Academy of Sciences.

Total RNA extraction and cDNA synthesis were described in previous studies (4, 15). Briefly, the UNIQ-10 column TRIzol total RNA isolation kit (Sangon Biotech) was used to purify the total RNA from cells, which was extracted by using TRIzol reagent (Ambion) according to the user guides. The DNA remnants in total RNA were digested by using DNase I (RNase-free, Thermo Fisher Scientific), and the first-strand cDNA was synthesized by using a RevertAid first-strand cDNA synthesis kit (Thermo Fisher Scientific) and a GeneRacer kit (Invitrogen) following the user guides. Phanta max super-fidelity DNA polymerase (Vazyme Biotech) was used to amplify the gene fragments with specific primers (Supplemental Table I), which were designed based on the X. laevis genome and transcriptome data.

The locus and sequence data were retrieved from available genome assemblies and annotations, including Homo sapiens (human, version GRCh38. p14, GCF_000001405.40, https://www.ncbi.nlm.nih.gov/assembly/GCF_ 000001405.40/), Mus musculus (house mouse, version GRCm39, GCF_000001635.27, https://www.ncbi.nlm.nih.gov/assembly/GCF_ 000001635.27/), Gallus gallus (chicken, version bGalGal1.mat.broiler.GRCg7b, GCF_016699485.2, https://www.ncbi.nlm.nih.gov/assembly/GCF_016699485.2/), Xenopus tropicalis (tropical clawed frog, version UCB_Xtro_10.0, GCF_000004195.4, https://www.ncbi.nlm.nih.gov/assembly/GCF_000004195.4/), X. laevis (African clawed frog, version Xenopus_laevis_v10.1, GCF_017654675.1, https://www.ncbi.nlm.nih.gov/assembly/GCF_017654675.1/), Rana temporaria (common frog, version aRanTem1.1, GCF_905171775.1, https://www.ncbi.nlm.nih.gov/assembly/GCF_905171775.1/), Nanorana parkeri (Tibetan frog, version ASM93562v1, GCF_000935625.1, https://www.ncbi.nlm.nih.gov/assembly/GCF_000935625.1/), Bufo gargarizans (Asiatic toad, version ASM1485885v1, GCF_014858855.1, https://www.ncbi.nlm.nih.gov/assembly/GCF_014858855.1/), Bufo bufo (common toad, version aBufBuf1.1, GCF_905171765.1, https://www.ncbi.nlm.nih.gov/assembly/GCF_905171765.1/), and Danio rerio (zebrafish, version GRCz11, GCF_000002035.6, https://www.ncbi.nlm.nih.gov/assembly/GCF_000002035.6/), in the National Center for Biotechnology Information (NCBI) database.

Protein sequences were predicted by using the Translate software online (https://web.expasy.org/translate/). Conserved protein domains were predicted through sequence alignment in the Conserved Domain Database (CDD, NCBI) and online Simple Modular Architecture Research Tool (SMART). Multiple sequence alignments were performed by using the Clustal X program. Phylogenetic analyses were carried out with neighbor-joining method in MEGA 6.0 and the MEGA-X software package. Protein sequences were predicted by using the Translate software online (https://web.expasy.org/translate/). Conserved protein domains were predicted through sequence alignment in the Conserved Domain Database (NCBI) and online Simple Modular Architecture Research Tool. Multiple sequence alignments were performed by using the Clustal X program. Phylogenetic analyses were carried out with neighbor-joining method in MEGA 6.0 and the MEGA-X software package.

The luciferase reporter plasmid of X. laevis IFN1 (a member of intron-containing type I IFNs, GenBank accession no. XM_041577517.1, https://www.ncbi.nlm.nih.gov/nuccore/XM_041577517.1/) and expression plasmids of X. laevis IRF3, IRF7, and p65 were described and verified functionally in a previous study (4). The X. laevis STAT1-Flag (STAT1, with GenBank accession no. NP_001082256.1, https://www.ncbi.nlm.nih.gov/protein/NP_001082256.1/), 2HA-STAT2 (STAT2, OQ260016, https://www.ncbi.nlm.nih.gov/nuccore/OQ260016/), IRF9-GFP (IRF9, OQ260017, https://www.ncbi.nlm.nih.gov/nuccore/OQ260017/), IFNL5 (NM_001402970.1, an intron-containing type III IFN gene, https://www.ncbi.nlm.nih.gov/nuccore/NM_001402970.1/), IRF1 (NM_001089781.1, https://www.ncbi.nlm.nih.gov/nuccore/NM_001089781.1/), and AMNTR50 (OQ260013, https://www.ncbi.nlm.nih.gov/nuccore/OQ260013/) sequences were inserted into p3XFLAG-CMV^TM-14, pEGFP-C1, or pTurbo-GFP-N expression vectors by using the ClonExpress II one step cloning kit (Vazyme Biotech). Proximal promoter sequences of X. laevis amntr50, mx1 (NM_001256769.1, https://www.ncbi.nlm.nih.gov/nuccore/NM_001256769.1/), viperin (XM_018263999.2, https://www.ncbi.nlm.nih.gov/nuccore/XM_018263999.2/), ifnu (MW924834.1, https://www.ncbi.nlm.nih.gov/nuccore/MW924834.1/), and ifnl were inserted into pGL3-basic vector. The endotoxin-free plasmids were extracted and purified by using an E.Z.N.A. endo-free plasmid mini kit II (Omega Bio-tek) following a standard protocol.

The luciferase reporter assay was performed as described previously (4). Briefly, A6 cells (∼5 × 10⁴ cells/well) were seeded into 48-well plates and transfected with pRL-TK plasmid (Promega) and luciferase reporter vectors of IFN or ISG promoters (wild-type or mutants) by using FuGENE HD transfection reagent (Promega). Next, the transfected cells were incubated with the recombinant IFN proteins (X. laevis IFN1, X. laevis IFNL5, and X. laevis IFNU) in the medium or cotransfected separately with the IRF1, IRF3, IRF7, and p65 expression plasmids. Twenty-four hours later, the cells were collected and detected by using the Dual-Luciferase reporter assay system (Promega) and GloMax-Multi Jr detection system (Promega).

The standard curve was established as described previously (14). Quantitative real-time PCRs (qRT-PCRs) were performed by using the QuantStudio 3 real-time PCR system (Applied Biosystems). The PCR reaction system with a total volume of 20 μl contained 10 μl of PowerUp SYBR Green master mix (Applied Biosystems), 7 μl of sterile water, 1 μl of each specific primer (Supplemental Table I), and 1 μl of cDNA template, with the protocol as follows: one cycle of 50 and 95°C for a total of 4 min, followed by 45 cycles of 95°C for 15 s and 60°C for 30 s. Data analysis was performed using the 2^−ΔΔCt method, and the gene expression for each sample was normalized against β-actin.

For gene expression analyses, sterile 12-well plates were used to culture A6 cells (∼1.5 × 10⁵ cells/well), which were stimulated with polyinosinic-polycytidylic acid (poly(I:C); Sigma-Aldrich) at a final concentration of 10 μg/ml for 0, 2, 4, 6, and 10 h or were infected with FV3 virus at a multiplicity of infection (MOI) of 0.3 for 0, 10, 26, 39, and 49 h. Then, the cells were collected to detect the expression of target genes by using qRT-PCR.

RNA interference was performed to decrease the expression of target genes, including irf1, irf3, irf7, and rela (p65), in A6 cells. All gene-specific small interfering RNA (siRNA) and negative control siRNA reagents were designed and synthesized by Tsingke Biotechnology and were transfected into the A6 cells by using X-tremeGENE siRNA transfection reagent (Roche). Next, the cells were stimulated using poly(I:C) at a final concentration of 10 μg/ml and were collected to detect the knockdown effect of the target genes by using qRT-PCR with the specific primers (Supplemental Table I).

The experimental methods for EMSA and the preparation of prokaryotic/eukaryotic recombinant proteins were described previously (4). Briefly, 2 μg of empty vectors (as control) and 2 μg of each eukaryotic expression plasmid (p3XFLAG-CMV-14-xl-IFN1, p3XFLAG-CMV-14-xl-IFNL5, and p3XFLAG-CMV-14-xl-IFNU) were separately transfected into HEK293T cells (seeded in six-well plates) to produce eukaryotic recombinant IFN proteins in an equal volume of medium. The volume of supernatant medium from the HEK293T cells transfected with empty vectors or IFN expression plasmids was also equivalent in all experiments in relation to treatments of A6 cells by recombinant IFNs. For prokaryotic recombinant protein production, the coding sequence of IRF1 (Met¹–Arg¹²⁴), IRF3 (Met¹–Thr¹²⁴), IRF7 (Met¹–Pro¹²⁷), and p65 (Met¹–Glu¹⁹³) were inserted into the PATX-SUMO expression vectors (AtaGenix), respectively, which were then transformed into the Escherichia coli T7E strain. After induction by isopropyl β-d-thiogalactoside (IPTG), the soluble fusion proteins were purified through affinity method against His-Tag on Ni-NTA resin (Millipore). The recombinant proteins were determined by Bradford’s method and SDS-PAGE.

For EMSA, the nuclear protein was extracted using Minute cytoplasmic and nuclear extraction kits (Invent Biotechnologies). An EMSA probe biotin labeling kit (Beyotime) and EMSA/gel-shift kit (Beyotime) were used to perform probe labeling and binding with probe/recombinant or nuclear protein according to the manufacturer’s instructions. After the separation by native PAGE, the probes were transferred to the nylon membrane and were detected by autoradiography using the LightShift chemiluminescent EMSA kit (Pierce) and ChemiDoc MP imaging system (Bio-Rad).

Total RNA was collected from A6 cells treated for 24 h with control medium or an equal volume of medium containing recombinant X. laevis IFN-υ, respectively. The total amount and integrity of RNA were assessed using the RNA Nano 6000 Assay kit of the Bioanalyzer 2100 system (Agilent Technologies). For library preparation, mRNA was purified from the total RNA by using poly-T oligo-attached magnetic beads (Novogene). Fragmentation was performed using divalent cations under elevated temperature in first strand synthesis reaction buffer (NEB). Next, the NEBNext Ultra II DNA library prep kit for Illumina (NEB) was used to generate libraries. The first-strand cDNA was synthesized by using Moloney murine leukemia virus reverse transcriptase (NEB) with random hexamer primer, and redundant RNA was degraded by RNase H (NEB). Subsequently, the second-strand cDNA synthesis was performed using DNA polymerase I (NEB) and 2′-deoxynucleoside 5′-triphosphate. The remaining overhangs were converted into blunt ends through exonuclease/polymerase activities. After 3′-end adenylation and sequencing adaptor ligating of the cDNA fragments, PCR was carried out and the products were purified by the AMPure XP system (Beckman Coulter) to generate the libraries, which were quantified by a Qubit 2.0 fluorometer (Thermo) and qRT-PCR. The insert size of the libraries was detected by an Agilent 2100 bioanalyzer. Sequencing was performed on the NovaSeq 6000 system (Illumina), and the end reading of 150-base pairing was generated.

For the quality control of data, clean data (clean reads) were obtained by removing the reads containing adapters or N base and the low-quality reads (the number of bases with Qphred ≤20 accounted for 50% of the total read length) from the raw data (raw reads). Next, the clean reads were aligned to the X. laevis reference genome (version Xenopus_laevis_v10.1, GCF_017654675.1) using HISAT2 v2.0.5. The featureCounts v1.5.0-p3 was used to count the reads number mapped to each gene. Differential expression analysis was performed using the DESeq2 package (1.20.0). The clusterProfiler (3.8.1) package was used to implement the statistical enrichment of differential expression genes in Kyoto Encyclopedia of Genes and Genomes pathways.

Coimmunoprecipitation (Co-IP) experiments were carried out as in a previous study (31). Briefly, HEK293T cells (1 × 10⁷ cells) were seeded in sterile six-well plates and were cotransfected with eukaryotic expression plasmids of STAT1-Flag/2HA-STAT2/IRF9-GFP, 2HA-AMNTR50/IRF3-Flag, or empty vectors for 48 h. Next, the cells were collected for Co-IP using anti-FLAG M2 affinity gel (Millipore) or anti-hemagglutinin (HA) immunomagnetic beads (MIP0063, Dia-An Biotechnology).

The synthetic STAT2/STAT2p peptides, including CKPELTLQYEQYLQRK and CKPELTLQYEQY(p)LQRK, were cross-linked with keyhole limpet hemocyanin to immunize rabbits to raise polyclonal antiserum (anti-STAT2 and anti-STAT2p, respectively) by AtaGenix Laboratories. The protein samples were detected by Western blotting as previously described (31). The SimplePAGE 4–12% Bis-Tris SDS-PAGE gel (Sangon Biotech) was used to separate protein samples, which were then transferred to a polyvinylidene difluoride membrane (0.45 μm; Millipore). The membranes were incubated with the primary Abs, including anti-Flag (F1804, mouse monoclonal M2, Sigma-Aldrich), anti-HA (H3663, mouse monoclonal HA-7, Sigma-Aldrich), anti–enhanced GFP (Z200943, mouse monoclonal 3F11, Dia-An Biotech), anti-STAT1p (SAB4300032, phosphorylation site of tyrosine 701, rabbit polyclonal, Sigma-Aldrich), anti-STAT1 (14994, rabbit monoclonal D1K9Y, CST), anti-GAPDH (D110016, rabbit polyclonal, Sangon Biotech), and anti-STAT2/STAT2p serum, in TBST buffer containing 1% nonfat dry milk at 4°C overnight. After washing with TBST buffer, the blots were incubated with secondary Abs and detected by using a ChemiDoc MP imaging system (Bio-Rad).

Results data were analyzed statistically with a Student t test or ANOVA with a post hoc Tukey test (significant level parameter, 0.01) in SPSS 16.0 software. Significant difference is indicated for p values (*p < 0.05, **p < 0.01) or by different letters (a > b > c > d). The data from at least two independent experiments are presented as mean ± SEM.

For understanding the expression pattern of amphibian ifnu in response to virus infection, the DNA virus, FV3, and poly(I:C) were used to infect or stimulate the X. laevis A6 cells to detect the expression of type I, III, and IV IFNs, including X. laevis IFN1, IFN-λ5, and IFN-υ genes, as well as some IRF members (IRF1, IRF3, and IRF7) and RELA (p65) genes (Fig. 1). As expected, IFN1, IFN-λ, and IFN-υ and Viperin (an antiviral ISG as positive control) genes were strongly induced by FV3 and poly(I:C) (Fig. 1A, 1C), and the transcription factors irf1, irf3, irf7, and rela were also significantly upregulated by FV3 and poly(I:C), respectively (Fig. 1B, 1D).

Promoters of type I IFNs from zebrafish to humans, including zebrafish IFNφ1, X. laevis IFN1, and human IFN-β, and of type III IFNs (X. laevis IFN-λ5, human IFN-λ1 and λ3) and type IV IFNs (X. laevis IFN-υ, zebrafish IFN-υ) were analyzed in relation to the composition of transcriptional elements and expression (Fig. 2). Similar to type I and III IFN genes in humans and X. laevis, X. laevis IFN-υ and zebrafish IFN-υ all possess potential ISRE and NF-κB sites (Fig. 2A). Sequence alignment results indicated that sequences of ISRE and NF-κB sites in X. laevis IFN-υ and zebrafish IFN-υ have a conserved sequence feature when compared with ISRE/PRD I/III and NF-κB/PRD II from type I and III IFN genes, respectively (Fig. 2B, 2C). In fact, schematic analysis of promoters in type IV IFNs and in type III IFNs showed that the transcriptional elements in the proximal promoter of type IV IFN are similar to those of type III IFN genes on the basis of relative positions of ISRE and NF-κB sites (Fig. 2A, 2D, 2E). To verify the functional transcription sites, the proximal promoters with wild-type or mutant ISRE and NF-κB sites of X. laevis IFN-υ and X. laevis IFN-λ5 were constructed to analyze the promoter activities (Fig. 2D, 2E). As shown in Fig. 2F and 2G, the activities of wild-type X. laevis IFN-υ and X. laevis IFN-λ5 promoter plasmids increased significantly following poly(I:C) stimulation, but the mutants of ISRE or NF-κB sites in the promoters lost responsive activities to the stimulation. In addition, as a positive control, the X. laevis IFN-λ5 promoter showed similar results (Fig. 2H).

To understand whether the transcription of X. laevis IFN-υ can be induced by IRFs and NF-κB, a luciferase reporter assay was performed to detect the activity of the X. laevis IFN-υ promoter in A6 cells through the overexpression of IRF1, IRF3, IRF7, and p65. As shown in Fig. 3A, activities of promoter plasmids of X. laevis IFN1, IFN-λ5, and IFN-υ were significantly activated by wild-type IRF1-, IRF3-, IRF7-, and p65-expressing plasmids, but not the N terminus mutants. Subsequently, it was found that the mutants of ISRE and NF-κB sites of X. laevis IFN-λ5 (Fig. 3B) and X. laevis IFN-υ (Fig. 3C) had significantly reduced activities mediated by IRF1, IRF3, IRF7, and p65, respectively, except for a nonsignificant reduction of X. laevis IFN-λ5 ISRE2-mut in response to IRF3 (Fig. 3B, 3C).

Moreover, the knockdown of the IRFs and p65 in A6 cells was carried out to validate their effect on the expression of X. laevis IFN1, IFN-λ5, and IFN-υ in response to poly(I:C) stimulation. The transcription level of IRF1, IRF3, and p65 decreased significantly by some synthetic siRNAs, such as siRNA-IRF1-2, siRNA-IRF3-2, and siRNA-p65-1 (Fig. 3D–F). With the use of these three siRNA sequences and the siRNA against IRF7, siRNA-IRF7-289, as reported previously (4), the expression of X. laevis IFN1, IFN-λ5, and IFN-υ was clearly reduced when compared with the respective negative controls (siRNA-NC) (Fig. 3G).

It was further confirmed through EMSA that purified recombinant DNA-binding domains at the N-terminal regions of IRF1/IRF3/IRF7 and p65 were separately associated with ISRE and NF-κB sites in the X. laevis IFN-υ promoter (Fig. 4). The recombinant transcription factors revealed binding activity to promoter probes labeled with biotin, which could be blocked by the unlabeled probes with wild-type elements, but not by the mutant probes, through competitive binding.

One important step of antiviral mechanism for type I and III IFNs is to establish a defense status in uninfected cells, which depends on signal transduction of IFNs and induction of antiviral ISGs. Both mx and viperin are well-known antiviral ISGs, and proximal promoters with wild-type or mutant ISRE sites of the two genes were constructed into luciferase reporter plasmids to understand the signaling of type I, III, and IV IFNs in frogs (Fig. 5A, 5B). It was observed that X. laevis IFN1, IFN-λ5, and IFN-υ promoted significantly the activity of mx and viperin reporter plasmids with wild-type ISRE, but not the mutants (Fig. 5C, 5D). The standard ISRE reporter plasmids were also activated by X. laevis IFN1, IFN-λ5, and IFN-υ (Fig. 5E). In fact, the three IFNs could induce the phosphorylation of STAT1 and STAT2 in A6 cells (Fig. 5F). In mammals, ISGF3 is composed of STAT1, STAT2, and IRF9, and the Co-IP assay revealed that X. laevis STAT1 interacts with STAT2 and IRF9 (Fig. 5G). Finally, X. laevis IFN-υ was capable of inducing ISGF3 in A6 cells through EMSA (Fig. 5H).

For understanding the repertoire of ISGs induced by the type IV IFN, A6 cells were incubated with recombinant X. laevis IFN-υ in the medium or equivoluminal negative control medium from HEK293T cells transfected with X. laevis IFN-υ–expressing plasmid or empty vector, respectively. It was reported that the antiviral state of A6 cells can be enhanced markedly by 24-h stimulation of recombinant X. laevis IFN-υ (15), and thus this time point was chosen to perform the expression analysis of ISGs. The cell RNA was then sequenced by the NovaSeq 6000 system (Illumina) and further mapped to the assembly of X. laevis (version Xenopus_laevis_v10.1). The transcription data were deposited in the NCBI database with accession no. PRJNA929021. A total of 447 genes were observed to be significantly upregulated, based on cutoff values of an adjusted p <0.05 and |log₂ fold change| >1 (Supplemental Fig. 1, Supplemental Table II). As indicated through the Kyoto Encyclopedia of Genes and Genomes, X. laevis IFN-υ–induced ISGs were enriched in immune-related pathways, including RIG-I–like receptor, NOD-like receptor, cytosolic DNA sensing, HSV1 infection, necroptosis, C-type lectin receptor, TLR, and proteasome-related signaling (Supplemental Fig. 1).

Among these X. laevis IFN-υ–induced ISGs, 66 groups/families were identified as human homologous ISGs, among which at least 268 genes (including the predicted pseudogenes) surprisingly did not correspond to the repertoire from human ISGs and IFNφ1-induced ISGs in zebrafish, although orthologs of some of these specific genes/families could be found in humans (Supplemental Table II). For further understanding the repertoire of ISGs, these conserved genes were classified according to known function models described as previously reported (26).

In general, enzyme genes or families induced by X. laevis IFN-υ contained Mx, RSAD2, cytidine/uridine monophosphate kinase 2 (CMPK2), guanylate-binding protein (GBP), GTPase very large IFN inducible (GVIN), proteasome 20S subunit β (PSMB), ring finger protein 213 (RNF213), HECT and RLD domain containing E3 ubiquitin protein ligase family (HERC), PYD and CARD domain containing (PYCARD), DExD/H-box helicase 60 (DDX60), Mov10 RISC complex RNA helicase (MOV10), eukaryotic translation initiation factor 2 α kinase 2 (EIF2AK2, also known as PKR), and TRIM25. Surprisingly, except for the TRIM family members, the X. laevis IFN-υ–induced ISGs included a very large family of TRIM-like molecules, which was identified in this study as amphibian-specific genes and was designed as AMNTR (Fig. 6A).

The group of transcription factors as ISGs identified in X. laevis comprised IRFs, STATs, PHD finger protein 11 (PHF11), poly(ADP-ribose) polymerase family (PARP), and deltex E3 ubiquitin ligase 3L (DTX3L) (Fig. 6B). In Fig. 6C, RLR pathway molecules, Z-DNA binding protein 1 (ZBP1), and stimulator of IFN response cGAMP interactor (STING, also known as MITA), which are pathogen sensors, are identified. The membrane proteins mainly consisted of receptor transporter protein (RTP), transmembrane protein 140 (TMEM140), and MHC families (Fig. 6D).

In addition, X. laevis IFN-υ also strongly upregulated other antiviral immune-related or functional unknown gene families, such as IFIT, sacsin molecular chaperone (SACS), phospholipase A2 inhibitor and LY6/PLAUR domain containing (PINLYPL) and transporter 1/2, ATP binding cassette subfamily B member (TAP1/2), and tudor domain containing 7 (TDRD7) (Fig. 6E).

Furthermore, the induction of identified ISGs was verified by a qRT-PCR assay, and all detected genes belonging to the above five groups of gene families, including enzyme genes (Fig. 7A), membrane protein genes (Fig. 7B), transfection factor genes (Fig. 7C), pathogen sensor genes (Fig. 7D), and other genes (Fig. 7E), were significantly upregulated by X. laevis IFN-υ.

In fish lower vertebrates, some gene families, such as fintrim, gig2, and fish-specific NLRs, have been reported to have undergone significant expansion (26). In X. laevis, five ISG families were observed with notable expansion based on phylogenetic and gene locus analyses, including AMNTR, GVIN, GBP, RTP, and IFIT. Similar to vertebrate TRIM proteins, AMNTR family members in X. laevis possess an N-terminal RING finger domain, two B-boxes, and a C-terminal PRY-SPRY domain (Fig. 8A). However, most AMNTRs were found to be encoded by intronless genes, which is different from intron-containing TRIMs in other vertebrates or finTRIM genes in fish (Fig. 8A). The X. laevis AMNTR family contains at least 100 members (Fig. 8B, Supplemental Fig. 2), which are distributed on different loci, such as on chromosomes 1L, 2L, 3L, 6L, 8L, 9_10L, 1S, 2S, 6S, and 9_10S (Supplemental Fig. 2). Among these AMNTRs, 61 members were identified as induced by X. laevis IFN-υ (Supplemental Table II). In fact, the conserved AMNTR genes were also found in other amphibian species, such as the common toad (Bufo bufo), Asiatic toad (Bufo gargarizans), common frog (Rana temporaria), and Tibetan frog (Nanorana parkeri). Phylogenetic analysis indicated that amphibian AMNTRs and vertebrate TRIMs/finTRIMs were separated into different clades, and evolutionary divergence was also observed within the AMNTR family (Fig. 8B).

Nineteen GVIN genes in A6 cells were upregulated by X. laevis IFN-υ (Supplemental Table II). In fact, the X. laevis GVIN gene family contained at least 40 members that were clustered and located on four different loci. (Fig. 9A). Based on results from the phylogenetic tree (Fig. 9B), members in the X. laevis GVIN family were separated into two major groups, groups I and II, and were designed as GVIN in Xenopus (GVINX), as they were not always clustered with mouse GVIN molecules, nor as orthologs of mouse GVIN1, GVIN2, or GVIN3. GVINXs in group I contained six genes, GVINX34–39, which seemed to be more closely related with the amniote GVIN family (Fig. 9B). It was also supported by comparative analysis of protein sequences that the length and conserved domain (RHD3) of group I members, such as GVINX39, were similar to mouse GVIN1, but group II molecules possessed shorter protein sequences or additional domains, such as KISc (Fig. 9C).

Similar to the GBP family in humans, the duplication of multicopy genes was observed in the GBP family in X. laevis, with the presence of conserved tandem domains of GBP/GBP_C. However, it was difficult or impossible to determine a one-to-one correspondence in GBP members between X. laevis and humans, as shown in gene locus and phylogenetic analyses (Fig. 10A–C). Accordingly, GBP genes in X. laevis were named as gbp-a, gbp-b, gbp-c and gbp-d (Fig. 10A, 10B). Unlike the GBP family, the X. laevis PARP14 gene family also had an expansion in comparison with the single copy PARP14 in humans (Fig. 10D). The X. laevis PARP14 genes, except parp14.2 and parp14.3, were clustered on chromosome 9_10L and linked with parp9 and dtx3l.L, which were highly conserved in humans, and were also induced by X. laevis IFN-υ in A6 cells (Figs. 6B, 10D). However, the X. laevis PARP15s were not clustered in a clade with human PARP15 in the phylogenetic tree (Fig. 10E), although the sequence of PARP14 and PARP15 in the frog all contained conserved features, such as 3× A1pp/1× PARP and 2× A1pp/1× PARP domains, as PARP14 in humans (Fig. 10F).

The human RTP genes are scattered across two loci, that is, chromosomes 2 and 3. However, the X. laevis RTP genes were clustered between masp1.L/S and pdcd1.L/S on chromosome 5L/S, which are two conserved orthologous genes on human chromosomes 3 and 32, respectively (Fig. 11A). Similar to GBP and GVIN family genes, RTP genes in X. laevis could not be identified also accordingly as one-to-one orthologous genes in humans (Fig. 11A, 11B). To avoid any confusion, RTP genes in X. laevis were designed as rtp-a, rtp-b, rtp-c, rtp-d, and rtp-e. However, X. laevis RTPs possess conserved zf-3CxxC and transmembrane domains (Fig. 11C).

The IFIT family as the well-known ISGs are found to be duplicated in many vertebrate species, as observed in X. laevis. The frog IFIT loci had conserved collinearity with humans (Fig. 11D). Phylogenetic analysis indicated that X. laevis IFIT members could be grouped into IFITA and IFITB, although the two groups were clustered into a large clade (Fig. 11E). Interestingly, X. laevis IFITA1 and IFITB1 sequence features, such as the number of TPR domains, were similar to human IFIT1 and IFIT2, respectively (Fig. 11F).

To understand whether the amphibian-specific AMNTR family is involved in antiviral immunity, the function of an AMNTR family member, amntr50, induced strongly by X. laevis IFN-υ, was investigated. The cDNA sequence of amntr50 was cloned from A6 cells and was deposited in GenBank under accession no. OQ260013. The full-length open reading frame of amntr50 contains 1548 nucleobase pairs, which were predicted to encode 515 aa. AMNTR50 had high protein identity (71.7%) with X. tropicalis AMNTR50, but it showed low identity with zebrafish FTR83 (18.1%), X. laevis TRIM25 (20.1%), human TRIM16 (13.6%), TRIM25 (17.0%), and TRIM39 (17.3%) (Supplemental Fig. 2).

The expression of amntr50 was strongly upregulated by X. laevis IFN-υ, poly(I:C), and FV3 infection (Figs. 7A, 12A, 12B). To verify whether amntr50 is an ISG, the promoter reporter plasmids with potential ISRE sites (wild-type and mutant) were constructed to detect luciferase activity. As shown in Fig. 12C and 12D, the wild-type promoter, AMNTR50 (wild-type) had significantly higher activity than did the mutants of ISRE1 and ISRE2, when induced by X. laevis IFN1, IFN-λ5, and IFN-υ. Unexpectedly, the overexpression of AMNTR50 resulted in significant decrease in the expression of antiviral ISGs Mx and Viperin and caused increased FV3 titers, all in A6 cells (Fig. 12E, 12F). In fact, the increased activity of ISGs Mx and Viperin and IFNs X. laevis IFN1, IFN-λ5, and IFN-υ, induced by IRF3, was reduced significantly by AMNTR50 (Fig. 12G), and it was found that AMNTR50 interacted with IRF3 and consistently attenuated its protein level in a dose-dependent manner (Fig. 12H, 12I).

It is thus demonstrated that IFN-υ in amphibians can be transcriptionally regulated by IRF1, IRF3, IRF7, and p65 (Fig. 13), and it can bind with IFN-υR1 and IL-10R2 (15), then signaling through ISFG3 to generate the expression of a large repertoire of ISGs (Fig. 13).

In addition to the well-characterized three types of IFNs in vertebrates (1, 9), IFN-υ, which is considered to be type IV IFN, has been reported as an evolutionarily conserved gene in vertebrates from fish to primitive mammals (15). With the identification of its receptors, IFN-υR2 and IL-10R2, X. laevis and zebrafish IFN-υ were found to be able to induce the expression of immune-related genes (15). In this study, the proximal promoter of IFN-υ in the amphibian model X. laevis was identified to contain ISRE and NF-κB sites, which can be activated by critical members of IRF and the NF-κB family. It was further found that X. laevis IFN-υ can signal through ISGF3 to induce the expression of >400 ISGs, including conserved ISG groups/families as observed in mammals (24, 25), and amphibian-specific gene families, such as AMNTRs. It is considered that the type IV IFN signaling and its role in antiviral immunity were further understood in amphibians.

In the evolution of vertebrates against viruses, IFNs are considered as core molecules, and they have evolved with conserved regulatory elements in promoters (4, 12, 14). PRD I/III/ISRE and PRD II/NF-κB are critical sites to respond to transcription factors, including IRF and NF-κB family molecules, to control the expression of mammalian type I and III IFN genes (12, 19). Interestingly, the presence of conserved ISRE and NF-κB sites in the proximal promoter of the IFN-υ gene in amphibians is much more comparatively similar to the position of these sites in type III IFN genes (11, 12). In fact, the upregulated transcription of X. laevis IFN-υ by IRF1/3/7 and p65 through the functional ISRE and NF-κB sites, as revealed in the current study, may suggest that the mechanism involved in the induction of IFN-υ gene is very much similar to type I and III IFNs in response to viral infections.

Upon viral infection, mammalian type I and III IFNs are secreted extracellularly to induce the expression of antiviral ISGs through the activation of ISGF3 (1, 12, 22). In the current study, the finding that X. laevis IFN-υ can promote the phosphorylation of STAT1/2 and also the activation of ISGF3 may suggest that type I, III, and IV IFNs have similar signaling, and these three types of IFNs may have a common ancestor, as suggested previously (15). However, it is difficult to estimate whether type IV IFN is evolutionarily related more closely with type I or III IFNs, because type I−IV IFNs have been identified in cartilaginous fish (15, 32), which may imply that the four types of IFNs must have diverged at least >450 millions years ago before the divergence of chondrichthyans (33). In contrast, the type I IFN family is expanded in most vertebrates, especially in Xenopus and some reptiles, which also have an expansion of type III IFN genes (6, 14). However, the type IV IFN family seems to be contractive or redundant in some vertebrates and reveals different evolution characteristics when compared with type I and III IFNs (8, 15).

Although some recombinant IFN proteins, such as vertebrate type I IFNs, are reported to have a direct antimicrobial effect (34), the pleiotropic activity of mammal IFNs is largely dependent on the induction of ISGs with a wide range of biological functions (24). As a newly established type of IFN, a wide spectrum of ISGs induced by IFN-υ in A6 cells may provide a solid base for further understanding its antiviral roles in amphibians and also in comparative studies in other lineages of vertebrates. In this study, 66 families induced by X. laevis IFN-υ were identified as homologous ISGs in mammal, but at least 268 genes cannot be correspondingly related to the repertoire of ISGs in humans, nor IFN-φ1–induced ISGs in zebrafish, and these ISGs can be considered at least as X. laevis–specific ISGs (XLSISGs), suggesting that a large degree of specificity may be present in the repertoire of IFN-υ–induced ISGs in amphibian, and it would be of interest to comparatively investigate the repertoire of ISGs as induced by type I and III IFNs, as well as type IV IFN, in amphibians.

Virus sensor genes in RLR signaling and transcription factors in IFN signaling have been identified as conservative ISGs induced by X. laevis IFN-υ in the current study, as induced by type I and III IFNs (11, 22, 35). The three mammalian RLR members, RIG-I (DDX58), MDA5 (IFIH1), and LGP2 (DHX58), which can recognize viral RNA in cytoplasm (36), are reported to have upregulation after exposure to IFN and after viral infection (37, 38). Another two ISGs, ZNFX1 and ZBP1, identified in the current study were reported recently in humans, with the former identified as a mitochondria-related, helicase-recognizing dsRNA virus involved in the early stage of RNA virus infection to regulate positive RLR signaling (39), and the latter confirmed as an intracellular DNA virus sensor to promote IFN expression (40, 41). However, ZBP1 as an ISG was not found in teleost fish (26), suggesting that ZBP1 signaling may have evolved likely after fish diversification.

The transcription factor ISGF3, containing STAT1, STAT2, and IRF9, is activated by mammalian type I and III IFNs (11, 22, 42), and also by type IV IFN as identified in amphibians in the current study. Interestingly, X. laevis IRF1 and IRF7 are involved in the regulation of type IV IFN, as in type I/III IFNs, and it is reasonable to speculate that the induction of IRF1 and IRF7 by X. laevis type IV IFN may have a positive effect on the IFN signaling loop. In addition, mammalian IRF1 may have many other functions, including pathogen clearance, antiproliferation/antiapoptosis, immune cell development, Ag presentation, and particularly an IFN-dependent inflammation response (1, 43). Other transcription factors, such as ATF3 and PHF11, have received less attention regarding their possible function. It is known that mammalian ATF3, which belongs to the ATF/cAMP response element-binding (ATF/CREB) family, is involved in neuronal regeneration, oncogenesis regulation, and negative regulation of TLR signaling (44, 45), and that PHF11 is likely associated with allergy and T cell activation in mammals (46).

Members of the mammalian PARP family are involved in multiple biological processes, including transcriptional regulation, DNA damage response, metabolism, and homeostasis (47, 48). It has been confirmed that mammalian PARP9, PARP11, PARP12, PARP13, and PARP14 possess antiviral activities (49). PARP9 interacts with DTX3L to form a complex, which is then associated with STAT1 to promote transcriptional activity in mammals. Moreover, mammalian PARP12 and PARP14 were found to inhibit viral replication, and PARP14 is involved in the induction of type I IFNs in response to virus infection (49). Unlike the single copy presence of PARP14 in humans, the X. laevis PARP14 gene family has an expansion of two phylogenetic clades, in which most members were induced by X. laevis IFN-υ as identified in the current study. Surprisingly, whether PARP15 is an ISG or not in humans is not verified, but the PARP15 subfamily members, as an XLSISG, are strongly upregulated by X. laevis IFN-υ. It should be very valuable to investigate the function of amphibian PARP14 and PARP15 members in host defense in the future.

As a type I IFN-induced transmembrane ISG, mammal RTP4 can interact with TANK-binding kinase 1 (TBK1) to negatively regulate the type I IFN response and increase virus replication by interfering with phosphorylation and expression of both TBK1 and IRF3 (50). Although gene duplication occurred in the human RTP family, only RTP4 has been identified as an ISG. Expectedly, the X. laevis RTP family also contains multiple genes, and the location of the two conserved genes (masp1 and pdcd1) around RTP genes indicated that the human RTP locus may have undergone rearrangement during evolution. Unexpectedly, among the X. laevis RTP family, 10 members were observed to be strongly upregulated by X. laevis IFN-υ. Another X. laevis IFN-υ–induced transmembrane ISG is TMEM140, the homolog of which has antiviral activity in mammals (51). In addition, some vertebrate type I/II IFN-induced immune pathway–related transmembrane molecules, such as MHCs, B2M, CD74, TAP1/2, and TAPBP, as well as other CIITA, NLRC5, PSME2, and the 20S proteasome subunits (PSMB7, PSMB8, and PSMB9), were found to be upregulated by X. laevis IFN-υ, implying that type IV IFN signaling is likely associated with these pathways, although their biological function is unknown in A6 cells and requires further investigation in amphibians.

The X. laevis type IV IFN induces broad-spectrum enzyme genes. GBP family homologs, which are also induced by type IV IFN in amphibians, are reported to play important roles in host defense against bacterial, viral, and parasitic pathogens, and they may be activated by type I and III IFNs, and perhaps even by type II IFN in mammals (52). The well-known antiviral enzyme genes induced by mammalian IFNs also include Mx, RSAD2, PKR, DDX60, TRIM25, MOV10, RNF213, and ZC3HAV1, which suppress virus proliferation through the inhibition of protein and RNA synthesis, promotion of viral RNA degradation, or activation of the antiviral pattern recognition receptor (such as RIG-I signaling) pathway (23, 53–60). The GVIN (also known as very large IFN-inducible GTPases [VLIGs]) family has clear expansion in X. laevis and fish, and X. laevis GVINXs were separated into two phylogenetic groups (groups I and II). However, the human GVIN family contains only two pseudogenes, and the current understanding of their function is very limited (61). Further discussion on this aspect goes beyond the topic of the present research.

Surprisingly, a large TRIM-like XLSISG family, AMNTR, was identified as the type IV IFN-induced genes, with the inclusion of at least 100 members. In teleost fish, a large fish-specific TRIM family (finTRIM) was also discovered, and some members in the finTRIM family showed opposite functions in relation to antiviral immunity; for example, FTR83 is antiviral, but FTRCA1 is a negative regulator (28, 29). In the current study, AMNTR50, an AMNTR family member, can be induced by type I, III, and IV IFNs through the ISRE sites in the proximal promoter. Surprisingly, AMNTR50 was found to negatively regulate the expression of type I, III, and IV IFNs, resulting in the increased virus replication in A6 cells. It is thus confirmed that AMNTR50, together with other AMNTR family members, can be recognized as functional ISGs, but may contribute to negative feedback of IFN expression in amphibians.

Except for the immune-related typical ISGs, that is, USP18, TDRD7, SOCS1, PYCARD, OGFR, IFI35, LGALS9B, IL4I1, IFIT, HERC, HELZ2, EPSTI1, CMPK2, APOL1, and ADAR in vertebrates (23, 24, 26), the strongly induced XLSISGs by IFN-υ contain the gene groups/families, including ALPK2, C4A, FCRLB, IL18BP, LY6A, MORC3, NUGGC, OR51E2, PLAAT3, PHYHDLC.1, PINLYPL, RBM18, RNF112, RNF135, RNPEP, SACS, and TRIM65, some of which have human orthologs and are involved in the immune response (62, 63). For example, mammal IL18BP as a secretory receptor negatively regulated IL-18 signaling through the neutralization of IL-18 (62), and TRIM65 has antiviral activity via ubiquitination of the MDA5 signaling pathway in mammals (63). However, the biological function of most ISGs induced by IFN-υ may be of interest for further investigation from an evolutionary point of view.

In conclusion, the proximal promoter of X. laevis IFN-υ was found to contain functional ISRE and NF-κB sites, and this IFN can signal through ISGF3 to activate a wide spectrum of ISGs with the expansion of some gene groups/families in amphibians.

The authors have no financial conflicts of interest.

We acknowledge laboratory members for discussions and arguments.