Structural and Functional Characterization of a Fish Type I Subgroup d IFN Reveals Its Binding to Receptors

Key Points

• We solve the crystal structure of a fish type I subgroup d IFN, LcIFNd.

• LcIFNd activates the conserved JAK-STAT pathway by binding with CRFB1 and CRFB5.

• We identify key amino acids of LcIFNd that interact with receptors.

Abstract

Teleost fish type I IFNs and the associated receptors from the cytokine receptor family B (CRFB) are characterized by remarkable diversity and complexity. How the fish type I IFNs bind to their receptors is still not fully understood. In this study, we demonstrate that CRFB1 and CRFB5 constitute the receptor pair through which type I subgroup d IFN (IFNd) from large yellow croaker, Larimichthys crocea, activates the conserved JAK-STAT signaling pathway as a part of the antiviral response. Our data suggest that L. crocea IFNd (LcIFNd) has a higher binding affinity with L. crocea CRFB5 (LcCRFB5) than with LcCRFB1. Furthermore, we report the crystal structure of LcIFNd at a 1.49-Å resolution and construct structural models of LcIFNd in binary complexes with predicted structures of extracellular regions of LcCRFB1 and LcCRFB5, respectively. Despite striking similarities in overall architectures of LcIFNd and its ortholog human IFN-ω, the receptor binding patterns between LcIFNd and its receptors show that teleost and mammalian type I IFNs may have differentially selected helices that bind to their homologous receptors. Correspondingly, key residues mediating binding of LcIFNd to LcCRFB1 and LcCRFB5 are largely distinct from the receptor-interacting residues in other fish and mammalian type I IFNs. Our findings reveal a ligand/receptor complex binding mechanism of IFNd in teleost fish, thus providing new insights into the function and evolution of type I IFNs.

Introduction

Type I IFNs are a family of cytokines specialized in coordinating immune responses against viruses and other intracellular pathogens (1–4). In mammals, all type I IFNs bind to the cell surface receptor formed by IFN-α/β receptor (IFNAR) subunits 1 and 2 (IFNAR1 and IFNAR2) (5–7), which activates the JAK-STAT signaling pathway, resulting in the transcription of various IFN-stimulated genes (ISGs) (8, 9). The type I IFN receptor subunits IFNAR1 and IFNAR2 are both single-transmembrane receptors in which the fibronectin III subdomains are essential for the interaction with IFNs. IFNAR1 has an extracellular region composed of four fibronectin III subdomains (SD1–SD4), whereas IFNAR2 has two fibronectin III subdomains (SD1 and SD2). The crystal structures of the IFN-α/ and IFN-ω/IFNAR1/IFNAR2 ternary complexes have provided a comprehensive understanding of the binding and complex assembly mechanisms between mammalian type I IFNs and their receptors (10–14).

Teleost fish possess a complex, evolutionary divergent type I IFN system that mediates defense against viral infection through the conserved JAK-STAT signaling pathway (15–17). Teleost type I IFNs are divided into two groups based on the number of cysteine residues forming disulfide bonds: group I IFNs (subgroups a, d, e, and h) contain two cysteine residues, whereas group II IFNs (subgroups b, c, f, and i) contain four cysteine residues (18). Fish type I IFN receptors belong to the cytokine receptor family B (CRFB) with 17 known orthologs of mammalian IFNARs (19). From an evolutionary perspective, fish CRFB1/CRFB2 and CRFB5 are equivalent to the mammalian IFNAR2 and IFNAR1, respectively. However, unlike mammalian IFNAR1, the extracellular regions of CRFB5 and other known CRFBs possess only two fibronectin III subdomains, implying differences in the receptor binding mechanism. Previous studies have suggested that teleost group I and group II type I IFNs signal through the CRFB1–CRFB5 and CRFB2–CRFB5 receptor complexes, respectively (20–22). Interestingly, a recent study has shown that grass carp (Ctenopharyngodon idella) type I subgroup a IFN (CiIFNa) binds not only to CRFB1 and CRFB5, but also to CRFB2, and that CRFB1 may also form a heterodimeric receptor complex with CRFB2 (23). These findings imply that interactions between teleost type I IFNs and their receptors might be more complex than expected. Nevertheless, to which receptors fish type I IFNs bind and how remain poorly studied.

To date, three-dimensional structures of several fish type I IFNs have been reported, including zebrafish IFN (zIFN)-φ1, zIFN-φ2 (24), large yellow croaker (Larimichthys crocea) type I subgroup i IFN (IFNi) (LcIFNi) (18), and CiIFNa (23), all of which share a similar structure consisting of six α helices (A–F) and a straight helix F. The ternary complex model of LcIFNi and its receptors CRFB2–CRFB5 based on the homologous ternary complex structure has been constructed, shedding light on the mechanisms by which a fish group II type I IFN binds to their receptors (18). Additionally, structural analysis of CiIFNa has identified the key residues involved in its interaction with receptors (23). Structural data are therefore highly desirable, as they can provide detailed information on molecular mechanisms through which fish type I IFN bind to their receptors.

Our previous study identified type I subgroup d IFN (IFNd) from L. crocea (LcIFNd) that showed antiviral activity and regulated the IFN response (25). In this study, we demonstrate that LcIFNd interacts with the receptors LcCRFB1 and LcCRFB5, binding to LcCRFB5 with a higher affinity than binding to LcCRFB1. We also report the crystal structure of LcIFNd, which displays characteristic fish type I group I IFN architecture. Using nuclear magnetic resonance (NMR) titration, homology modeling, and mutational analysis coupled with pull-down assays, we identify key regions and residues of LcIFNd involved in interaction with LcCRFB1 and LcCRFB5. Moreover, using receptor-blocking experiments, we show that LcIFNd triggers the antiviral function through the conserved JAK-STAT signaling pathway by binding to LcCRFB1 and LcCRFB5, and we confirm that the amino acids L18, F43, K113, H128, H135, and M152 of LcIFNd are indispensable for the LcIFNd-mediated JAK-STAT signaling pathway.

Materials and Methods

Primary cells and cell lines

The primary head kidney leukocytes (PKLs) of L. crocea were prepared as described previously (26, 27). The continuous epithelioma papulosum cyprini (EPC) cell line was preserved in our laboratory and cultured at 25°C in L-15 medium supplemented with 10% FBS. Grouper spleen (GS) cells were derived from the spleen of orange-spotted grouper (Epinephelus coioides) and maintained in L-15 medium supplemented with 10% FBS at 25°C. Singapore grouper iridovirus (SGIV) was propagated in GS cells as previous described (28), and the virus stock was stored at −80°C until use. GS cells and SGIV were gifts from Prof. Qiwei Qin in South China Agricultural University. Human embryonic kidney 293T cells were grown in DMEM (Life Technologies) containing 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin (Life Technologies) at 37°C in a 5% CO₂ atmosphere.

Plasmid construction, expression, and purification of LcIFNd

The LcIFNd gene fragment encoding the mature peptide sequence (residues 21–185; API68651) was amplified with primer set (Supplemental Table I) from the L. crocea spleen cDNA using RT-PCR. The LcIFNd gene fragment was cloned into the expression vector pET28a with N-terminal 6× His and Sumo-tag, and the recombinant vector was transformed into BL21(DE3) pLysS-competent cells (EC1002M, WEIDI). Protein expression was induced with 300 mM isopropyl β-d-thiogalactoside at 18°C and 220 rpm for 16 h. Afterward, the cells were collected and lysed in lysis buffer (25 mM Tris-HCl [pH 7.5], 150 mM NaCl, 2 mM β-ME). rLcIFNd was purified using Ni-NTA affinity chromatography, followed by gel-filtration chromatography using a Superdex 75 Increase column (Cytiva) (Supplemental Fig. 1A).

Crystallization

Purified LcIFNd was concentrated to 10 mg/ml in a buffer containing 20 mM HEPES (pH 7.5) and 150 mM NaCl. Crystals were grown at 16°C using the sitting drop vapor diffusion method with 0.8 μl of recombinant protein and 0.8 μl of reservoir solution (0.1 M Tris-HCl [pH 7.0] and 20% polyethylene glycol 2000). The crystals were harvested in the reservoir solution supplemented with 15–20% glycerol and flash-frozen in liquid nitrogen.

X-ray diffraction data collection and structure determination

Crystal diffraction data for LcIFNd were collected at the Shanghai Synchrotron Radiation Facility (SSRF) BL-02U1 using the single-wavelength anomalous diffraction method. Datasets were processed on-site with the HKL-2000 package. The AutoSol program of the Phenix package was used for single-wavelength anomalous diffraction phasing, followed by iterative manual building using Coot and refinement in Phenix (29, 30). The crystals belong to space group P 1 21 1 with unit-cell dimensions of a = 40.378 Å, b = 38.813 Å, and c = 51.091 Å. The final structure was refined to 1.49-Å resolution (Table I). Structure quality was analyzed during refinement in Phenix and later validated in the PDB validation server. Structural figures were generated using PyMOL (Schrödinger, LLC). The crystal structure has been deposited in the PDB (https://www.rcsb.org/structure/unreleased/8IRQ) under accession number 8IRQ. In the crystal structure of the LcIFNd, 11 residues (aa 29–39) were not observed in the electron density. However, given that structure of LcIFNd predicted by AlphaFold v2.0 is almost entirely consistent with the crystal structure of LcIFNd (root-mean-square deviation of 0.377) (Supplemental Fig. 1B), the missing segment can be surmised to be a flexible loop between structural elements A and B with a high confidence. For the convenience of research, the predicted LcIFNd structure model was used as a research model in this study (Supplemental Fig. 1B, left panel).

Table I. Crystallographic data collection and structure refinement statistics.

Dataset	LcIFNd
Data collection
Beamline	BL-02U1, SSRF
Wavelength (Å)	0.97918
Resolution rangea,b	37.22–1.49 (1.543–1.49)
Space group	P 1 21 1
Cell dimensions
a, b, c (Å)	40.378, 38.813, 51.091
α, β, γ (°)	90.00, 112.817, 90.00
Total reflections	45,866 (4,719)
Unique reflections	23,089 (2,374)
Multiplicity	2.0 (2.0)
Completeness (%)	95.80 (98.67)
Mean I/σ I	16.59 (2.41)
Rmerge	0.02326 (0.2676)
Rmeas	0.03289 (0.3784)
Rpim	0.02326 (0.2676)
CC1/2	0.999 (0.888)
Refinement
Reflections used in refinement	23,060 (2,372)
Reflections used for Rfree	1089 (126)
Rwork	0.1746 (0.2301)
Rfree	0.2242 (0.2824)
Wilson B factor (Å2)	18.98
Number of nonhydrogen atoms	1357
Macromolecules	1206
Protein residues	150
RMS (bonds)	0.017
RMS (angles)	1.99
Ramachandran favored (%)	99.32
Ramachandran allowed (%)	0.68
Ramachandran outliers (%)	0.00
Rotamer outliers (%)	1.53
Clashscore	5.49
Average B factor (Å2)	27.52

CC1/2, Pearson correlation coefficient between two randomly selected halves of the dataset; mean I/σ I, mean average intensity/average intensity error; SSRF, Shanghai Synchrotron Radiation Facility.

^aOne crystal sample was used for structure determination. Values in parentheses are for the highest-resolution shell.

^bCC1/2 in the highest resolution shell was statistically significant at the 0.1% level.

Nano differential scanning fluorimetry

The thermostability of LcIFNd with a cysteine mutation was determined using the nano differential scanning fluorimetry (nanoDSF) method based on intrinsic tryptophan and tyrosine fluorescence. All nanoDSF assays were performed using the Prometheus NT.48 instrument (NanoTemper Technologies). Samples (∼10 μl at 0.5–1 mg/ml) were loaded into standard-grade nanoDSF capillaries, placed on the Prometheus capillary holder, and subjected to a temperature ramping of 1°C/min from 20 to 94.8°C. For a more detailed description of the principle of the nano-DSF assay, please refer to Chen et al. (18).

Luciferase reporter assay

The promoter region (located at 29555747–29557148 bp of the L. crocea genome; GenBank accession no. LT972194.1) (31) of the L. crocea MxA gene was inserted into the Dual-Luciferase reporter plasmid pGL3-basic for the luciferase reporter assay. Sequences encoding CRFB1 (OM864520), CRFB2 (OM864521), CRFB3 (OM864522), CRFB4 (OM864523), and CRFB5 (OM864524) were cloned separately into the pcDNA3.1 vector. EPC cells were seeded at 2 × 10⁴ cells per well into 96-well culture plates and grown to 80% confluence overnight at 28°C in Opti-MEM supplemented with 10% FBS, 100 µg/ml streptomycin, and 200 U/ml penicillin G. The prepared cells were transfected with 100 µl of medium containing 50 ng of pGL3-MxA, 50 ng of one of the pcDNA3.1-CRFB vectors or empty pcDNA3.1 plasmid, and 1 ng of pRL-TK using the FuGENE HD transfection reagent (Promega) according to the manufacturer’s protocol. The cells were stimulated with 250 ng/ml rLcIFNd or control protein (Nus-tag) 48 h after transfection. Luciferase activity was measured 24 h poststimulation using the Dual-Luciferase reporter assay system (Promega). The relative luciferase activity of the MxA promoter was calculated by comparing the normalized luciferase activity in LcIFNd-treated cells with that in control protein–treated cells. All data were obtained from three independent experiments. Values significantly different from all controls are shown as *p < 0.05 and **p < 0.01.

Pull-down assays

Pull-down experiments were used to determine the interaction of LcIFNd with LcCRFB1 and LcCRFB5, respectively, and performed as previously described (18). The GST-tagged LcIFNd was expressed in Escherichia coli. Gene fragments encoding an extracellular region (Ex) of the receptor subunits LcCRFB1 (residues 20–220) and LcCRFB5 (residues 17–234) were subcloned into pFastbac1 vectors containing the hemotin signal peptide. Recombinant LcCRFB1-Ex (C-terminal His-tag) and LcCRFB5-Ex (N-terminal Flag-tag and C-terminal His-tag) were expressed in the insect cell/baculovirus system and purified from the Hi5 cell supernatant with a Ni-NTA affinity chromatography column (Cytiva). The pull-down assays were performed using glutathione Sepharose beads (P2253, Beyotime) by incubating purified GST-tagged LcIFNd protein with His-tagged LcCRFB1-Ex or Flag-tagged LcCRFB5-Ex, respectively. The beads were washed with wash buffer (25 mM HEPES [pH 7.5], 100 mM NaCl, 3 mM β-ME, 0.2% Triton X-100), followed by elution with buffer (20 mM reduced glutathione, 25 mM HEPES [pH 7.5], 100 mM NaCl, 3 mM β-ME, 0.2% Triton X-100). Eluted fractions were used for Western blot analysis. Western blot was performed using GST-tag mouse mAb (T0007, Affinity Biosciences), His-tag mouse mAb (T0009, Affinity Biosciences), or Flag-tag mouse mAb (T0003, Affinity Biosciences) as the primary Abs, and goat anti-mouse IgG (H+L) HRP (S0002, Affinity Biosciences) as the secondary Ab.

Isothermal titration calorimetry

The affinity experiments were performed in low-volume Nano isothermal titration calorimetry (ITC) (TA Instruments). LcIFNd, LcCRFB1-Ex, and LcCRFB5-Ex samples were prepared as described above. All samples for the assays were prepared in a buffer containing 20 mM HEPES (pH 7.5) and 150 mM NaCl. To measure the affinity between LcIFNd and LcCRFB1-Ex, 100 μM LcIFNd in the syringe was titrated into a sample cell containing 30 μM LcCRFB1-Ex. To determine the affinity between LcIFNd and LcCRFB5-Ex, 60 μM LcIFNd in a syringe was titrated into a sample cell containing 50 μM LcCRFB5-Ex. In addition, the affinity between LcCRFB1-Ex and LcCRFB5-Ex was also determined. All experiments were carried out at 25°C. Data correction and analysis were performed using NanoAnalyze software (TA Instruments).

NMR analysis

LcIFNd (residues 21–185) was cloned into the pET-28a vector with a 6× His tag. The rLcIFNd protein was then expressed overnight at 18°C in BL21(DE3) cells in M9 minimal media (32) with 1 g/l ¹⁵N-NH₄Cl and 4 g/l ¹³C-glucose as the nitrogen and carbon sources, respectively. The expressed 6× His-tagged protein was purified as described above, buffer exchanged into 20 mM HEPES (pH 7.0), 100 mM NaCl, and 1 mM DTT, and concentrated to 250 μM. Protein backbone resonances were assigned after analysis of HNCA, HNCOCA, HNCACB, HNCO, and ¹⁵N-edited NOESY (nuclear Overhauser effect spectroscopy)–heteronuclear single quantum coherence (HSQC) spectra according to standard protocols (33). To estimate the interaction of the LcIFNd with LcCRFB1-Ex and LcCRFB5-Ex, respectively, LcIFNd samples (250 μM) were titrated into the deuterated LcCRFB1-Ex (250 μM) and LcCRFB5-Ex (250 μM), respectively. LcIFNd+LcCRFB1-Ex and LcIFNd+LcCRFB5-Ex ¹H-¹⁵N HSQC spectra were assigned via the same experiments, respectively. All NMR spectra were recorded at 298 K on a Bruker Avance III 850-MHz spectrometer, acquired using Bruker TopSpin 3.6.2 software, processed using NMRPipe software (34), and analyzed using Sparky software (35).

Construction of the structural model

The structures of LcCRFB1-Ex and LcCRFB5-Ex were predicted in AlphaFold v2.0 (https://www.alphafold.ebi.ac.uk/). The LcIFNd–LcCRFB1-Ex binary complex structure model was spatially docked in PyMOL. The structures of LcIFNd and LcCRFB5-Ex were aligned with those of IFN-ω and IFNAR2, respectively, to construct the model of the LcIFNd–LcCRFB5-Ex binary complex structure. All structural models were energy minimized using Molecular Operating Environment (MOE) software (36).

Mutant protein–protein interaction assays

The following LcIFNd mutants were used for LcIFNd–LcCRFB1-Ex interaction assays: 1) single-residue mutations: K113A, K114A, H119A, K123A, R124A, H128A, K131A, K132A, and H135A; 2) multiple-residue mutations of helix E (MαE): K113/K114/H119/K123/R124/H128/K131/K132/H135A; and 3) multiple-residue mutations of the helices C and D (MαC-D): E67/E68/E74/E75/D76/E85/E89A. For LcIFNd–LcCRFB5-Ex interaction assays, the following single-residue mutations of LcIFNd were used: R11A, E15A, L18A, D19A, E33A, D34A, D39A, F43A, R144A, K145A, R148A, M152A, D155A, and R162A. All mutations were introduced into GST-tagged LcIFNd using the expression vector pGEX-6P-1. The mutant variants of the GST-tagged LcIFNd were expressed in Escherichia coli following the protocol for wild-type LcIFNd described above. His-tagged LcCRFB1-Ex and Flag-tagged LcCRFB5-Ex were expressed in Hi5 cells. Protein interaction was detected by a pull-down assay as described previously (37). Briefly, purified GST-tagged LcIFNd mutants were incubated with His-tagged LcCRFB1-Ex and Flag-tagged LcCRFB5-Ex using His and Flag Sepharose beads (M20008M, Abmart), respectively. The beads were washed with 6 column volumes of wash buffer (25 mM HEPES [pH 7.5], 150 mM NaCl, 3 mM β-ME, 0.2% Triton X-100), after which elution was carried out using elution buffer (as described in Pull-down assays). The eluted fractions were used for Western blot analysis. Western blot was performed by using GST-tag mouse mAb (T0007, Affinity Biosciences), His-tag mouse mAb (T0009, Affinity Biosciences), and Flag-tag mouse mAb (T0003, Affinity Biosciences) as the primary Abs, and goat anti-mouse IgG (H+L) HRP (S0002, Affinity Biosciences) as the secondary Ab.

Preparation of polyclonal Abs

Recombinant LcCRFB1-Ex and LcCRFB5-Ex were expressed in the insect cell/baculovirus system and purified from the Hi5 cell supernatant using Ni-NTA affinity chromatography (Cytiva). Purified LcCRFB1-Ex and LcCRFB5-Ex proteins were used for the preparation of polyclonal Abs (pAbs). To prepare the pAbs, 6-wk-old male New Zealand White rabbits were immunized with 250 mg of purified LcCRFB1-Ex and LcCRFB5-Ex, respectively, in CFA (Sigma-Aldrich, F5881). Rabbits were immunized four times at biweekly intervals. The ELISA validation of pAb titers and further purification steps were described in a previous study (18). Nonspecific rabbit Abs were purified from the sera of unimmunized rabbits and used as controls. The obtained pAbs were stored at −80°C until further use.

Verification of pAb specificity

The specificity of anti–LcCRFB5-Ex pAbs has been verified in our previous studies (18). The specificities of anti–LcCRFB1-Ex pAbs were tested by Western blot against rLcCRFB1-Ex and spleen and head kidney tissues obtained from a fresh L. crocea specimen. To further confirm the specificity of anti–LcCRFB1-Ex pAbs, immunofluorescence labeling of PKLs and HEK-293T cells was performed with pAbs. The pCMV-HA plasmids were overexpressed in HEK-293T cells as control, and pCMV-HA-LcCRFB1 groups containing full-length LcCRFB1 cDNAs were overexpressed in HEK-293T cells as experimental groups. An indirect immunofluorescence assay was performed as previously described (38). Briefly, the isolated PKLs and pCMV-HA– and pCMV-HA-LcCRFB1–transfected HEK-293T cells were incubated with anti–LcCRFB1-Ex pAbs (1:100) for 50 min. After removing the pAbs, cells were incubated with the Alexa Fluor 488–conjugated goat anti-rabbit IgG (1:200, Invitrogen) as the secondary Ab for 50 min. Finally, PKLs and HEK-293T cells were stained with DAPI (1:100, Invitrogen) fluorescent dyes (10 mg/ml) for 5 min in the dark. The stained PKLs and HEK-293T cells were observed under a laser-scanning confocal microscope (Leica Microsystems, Wetzlar, Germany).

Antiviral activity assays in GS cells

The GS cells were seeded onto six-well plates for 12 h. The cells were pretreated with rLcIFNd at a final concentration of 50 ng/ml or PBS (as a control) for 2 h, after which the cells were infected with SGIV at a multiplicity of infection of 2. At 24 h postinfection, the cells were observed microscopically for a cytopathic effect (CPE) (Leica Microsystems, Wetzlar, Germany). Infected cells were then harvested at 12 and 24 h postinfection for the analysis of viral gene expression by real-time quantitative PCR.

Gene expression analysis by real-time quantitative PCR

To study the effects of LcIFNd on the JAK-STAT signaling pathway in vitro, 5 × 10⁶ PKLs were stimulated with rLcIFNd (250 ng/ml) or BSA (as control; 250 ng/ml) for 6 or 12 h at 28°C. The cultured cells were then sampled for total RNA extraction using the Eastep super total RNA extraction kit (Promega). The cDNA was synthesized using GoScript reverse transcription mix (Promega) with oligo(dT) primers (Supplemental Table I). Then, cDNA samples were diluted to a suitable concentration in nuclease-free water for real-time PCR. The two virus SGIV genes (MCP and VP19) and JAK1, IRF9, STAT1, STAT2, MxA, PKR, and viperin from L. crocea were obtained from the National Center for Biotechnology Information GenBank (https://www.ncbi.nlm.nih.gov/genbank/). Real-time PCR was analyzed using TB Green mix (Takara) and specific primers in a QuantStudio 5 real-time PCR system (Thermo Fisher Scientific). All reactions were performed in triplicate on 96-well plates. Gene expression levels were normalized against the reference gene Ecβ-actin or Lcβ-actin using the 2^−ΔΔCt method (18). The fold change in gene expression levels for each time point was calculated by comparing normalized gene expression levels in the LcIFNd-stimulated and BSA-stimulated groups. All data were obtained from three independent experiments, and each analysis was performed in triplicate.

Western blots for protein expression and phosphorylation analyses

To determine the protein expression patterns and protein phosphorylation of STAT1 and STAT2 after rLcIFNd induction, isolated 2 × 10⁷ PKL cells were incubated with rLcIFNd at a 250 ng/ml final concentration for 0, 15, 30, or 45 min at 28°C. PKL cells treated with control protein (Nus-tag) served as the control. For Western blots, cells were harvested and lysed in RIPA (radioimmunoprecipitation assay) buffer (P0013K, Beyotime). To detect the STAT1 and STAT2 expression patterns and STAT1 phosphorylation, cell lysates were separated using 8% SDS-PAGE gels and blotted onto a polyvinylidene difluoride membrane for immunoblotting with the following Abs: rabbit anti-STAT1 (9172S, CST), rabbit anti-STAT2 (4594S, CST), and rabbit anti–p-STAT1 (7649S, CST) as primary Abs and goat anti-rabbit IgG (H+L) HRP (S0001, Affinity Biosciences) as the secondary Ab. For STAT2 phosphorylation, the cell lysates were separated using 8% Phos-tag SDS-PAGE gels (304-93526, Wako) and blotted onto polyvinylidene difluoride membranes for Western blot analysis with the rabbit anti-STAT2 as the primary Ab and the goat anti-rabbit IgG (H+L) HRP (S0001, Affinity Biosciences) as the secondary Ab. The methods used for Western blots are detailed in Mu et al. (39).

Blocking assays

To determine whether LcIFNd activates the JAK-STAT signaling pathway via a receptor-binding mechanism, LcCRFB1 and/or LcCRFB5 blocking assays were performed using rabbit anti–LcCRFB1-Ex and rabbit anti–LcCRFB5-Ex pAbs. PKL cells (1 × 10⁶ cells per well) were plated in six-well plates and incubated with anti–LcCRFB1-Ex and/or anti-LcCRFB5 pAbs for 3 h. After changing the medium, the cells were treated with rLcIFNd at a final concentration of 250 ng/ml. After 12 h of LcIFNd treatment, the PKL cells were harvested for total RNA extraction and reverse transcribed into cDNA. Expression levels of the key signaling molecules (JAK1, IRF9, STAT1, and STAT2) and antiviral genes (MxA, PKR, and viperin) were determined by real-time PCR, and levels of STAT1 and STAT2 phosphorylation were determined by Western blots as above. PKL cells treated with nonspecific rabbit Abs were used as controls.

Statistical analysis

Statistical analysis was performed using the IBM SPSS Statistics 2.0 software platform with one-way ANOVA, with "p < 0.05" or "p < 0.01" being considered statistically significant.

Results

LcCRFB1 and LcCRFB5 are potential functional receptors of LcIFNd

Previous studies have shown that LcIFNd can upregulate the transcription level of the antiviral gene MxA (25). To investigate which LcCRFB acts as the receptor for LcIFNd, we separately cloned genes encoding five L. crocea CRFB family members, which belong to the orthologs of mammalian type I IFN receptors and contain extracellular and intracellular domains (CRFB1, CRFB2, CRFB3, CRFB4, and CRFB5) into the pcDNA3.1 vector. EPC cells were transfected with a Dual-Luciferase reporter plasmid containing the promoter region of the MxA gene and one of the recombinant pcDNA3.1 vectors carrying L. crocea CRFB1, CRFB2, CRFB3, CRFB4, or CRFB5. The cells were then stimulated with rLcIFNd for 24 h, followed by measurement of luciferase activity. Transfection of CRFB1 or CRFB5 significantly increased MxA response, whereas simultaneous transfection with CRFB1 and CRFB5 led to a >2-fold increase in MxA response compared with cells transfected with the control vector (Fig. 1A), suggesting that both CRFB1 and CRFB5 receptors may respond to LcIFNd stimulation.

FIGURE 1.

LcCRFB1 and LcCRFB5 are potential functional receptors of LcIFNd. (A) Levels of luciferase activity in EPC cells transfected with vectors expressing different CRFB receptors upon stimulation with rLcIFNd and control protein (Nus-tag). The Dual-Luciferase reporter plasmid pGL3-Basic contained the promoter region of the antiviral IFN-stimulated gene MxA from L. crocea. (B) Results of pull-down assays examining the interactions between LcIFNd and extracellular regions of LcCRFB1 and LcCRFB5. (C and D) ITC binding curves for (C) titration of LcIFNd (100 µM) into LcCRFB1-Ex (30 µM) and (D) titration of LcIFNd (60 µM) into LcCRFB5-Ex (50 µM). The K_d and stoichiometry (N) values are expressed as mean ± SD, and all ITC experiments were repeated at least three times. Bars represent the means of three independent experiments ± SD. All experiments were performed in triplicate. *p < 0.05, **p < 0.01.

To characterize interaction between LcIFNd and the extracellular region of LcCRFB1 and LcCRFB5, we next performed GST pull-down assays. The results showed that both LcCRFB1-Ex and LcCRFB5-Ex could be pulled down by LcIFNd (Fig. 1B). Using ITC, we determined that LcCRFB1-Ex and LcIFNd interact with a K_d of 1.28 μM and stoichiometric coefficient N of 1.24 (Fig. 1C). LcCRFB5-Ex and LcIFNd were found to interact with >6-fold greater affinity (K_d = 203.80 nM, N = 1.14) (Fig. 1D). Meanwhile no interaction between LcCRFB1-Ex and LcCRFB5-Ex was observed (Supplemental Fig. 1C). These results suggest that LcIFNd can interact with both LcCRFB1 and LcCRFB5 and may signal via a receptor complex composed of LcCRFB1 and LcCRFB5.

Crystal structure of LcIFNd

To gain better understanding of LcIFNd at the molecular level, we determined its crystal structure at a 1.49-Å resolution. The crystal belongs to the primitive P1 21 1 space group with one molecule per asymmetric unit. Details of diffraction data, phasing, and refinement statistics are listed in Table I. LcIFNd has a characteristic type I IFN architecture with six α helices, designated A–F, of which helix B is the shortest. A single disulfide bridge C3-105 connecting helix A to helix D is visible in the structure. The helix F of LcIFNd is long and straight, which is the hallmark of type I IFNs (Fig. 2A). Comparing the structure of LcIFNd to available structures of type I IFNs suggests remarkable similarities with other reported fish (Fig. 2B, top panel) and mammalian type I IFNs (Fig. 2B, bottom panel). A major structural distinction is the AB loop, on which no helical structure is present. Such AB loop structure is consistent with zIFN-φ1, CiIFNa, and mouse IFN-β (mIFN-β) (Fig. 2C, left panel), but different from LcIFNi, zIFN-φ2, human IFN (hIFN)-α2, hIFN-ω, and hIFN-β (Fig. 2C, right panel).

FIGURE 2.

Structure analysis of LcIFNd. (A) Structure of LcIFNd in cartoon representation. Secondary structural elements are labeled A–F. The N terminus and C terminus are indicated with N and C, respectively. One disulfide bond is colored yellow and shown in detail with surrounding residues as 2Fo-Fc electron density maps contoured at the 1.0 σ level. (B) Superimposition of LcIFNd (firebrick) with LcIFNi (PDB ID: 7WZ5, marine), zIFN-φ1 (PDB ID: 3PIV, dark violet), zIFN-φ2 (PDB ID: 3PIW, gray 90), CiIFNa (PDB ID: 7WKH, aquamarine), hIFN-α2 (PDB ID: 1RH2, split-pea), hIFN-ω (PDB ID: 3SE4, tv-orange), hIFN-β (PDB ID: 1AU1, cyan-light teal), and mIFN-β (PDB ID: 1IFA, light orange). All structures were superimposed in PyMOL using the “align” command. (C) Left, Superimposition of LcIFNd, zIFN-φ1, CiIFNa, and mIFN-β. Right, Superimposition of LcIFNi, zIFN-φ2, hIFN-α2, hIFN-ω, and hIFN-β. (D) Multiple alignment of mature peptide sequence of LcIFNd with other fish type I IFNs whose crystal structures have been resolved. The conserved cysteine residues involved in forming intramolecular disulfide bonds are indicated by paired lines (C1–C3, C2–C4, and C5–C6) and highlighted in yellow. The cysteine residues C1 and C3 are marked with a star (★)below the alignment. (E) Representation of the C1–C3 disulfide bond in LcIFNd (yellow) superimposed with LcIFNi (a, orange), zIFN-φ1 (b, orange), zIFN-φ2 (c, orange), and CiIFNa (d, orange).

Given that LcIFNd contains two cysteines (C1 and C3) (Fig. 2D), it can be classified as a member of the group I of teleost type I IFNs. We compared the reported structures of fish type I IFNs and found that the C1/C3 conserved cysteine pair can form a disulfide bond linking helix A to helix D in all of the structures (Fig. 2E). Mutating either of the two cysteines resulted in low protein expression levels and significantly reduced the melting temperature value of the mutant protein (Supplemental Fig. 1D), suggesting that the disulfide bond is indispensable for the folding and stability of LcIFNd.

The helix E on LcIFNd is a key region involved in LcCRFB1 binding

To investigate the LcIFNd–LcCRFB1 and LcIFNd–LcCRFB5 interactions in more detail, we next carried out NMR analysis. We first performed ¹H, ¹³C, and ¹⁵N backbone and side-chain resonance assignments of the LcIFNd (residues 21–185) by using heteronuclear multidimensional NMR spectroscopy. Backbone and side-chain resonances of the LcIFNd were assigned based on the recorded heteronuclear two-dimensional/three-dimensional NMR spectra (Supplemental Fig. 1E). In total, ∼70.7% of the ¹H and ¹⁵N resonances of all nonproline residues were assigned unambiguously. For clarity, distribution patterns with the assigned or the unassigned residues are shown in the structure and sequences of LcIFNd (Supplemental Fig. 1F).

To identify specific amino acids in LcIFNd involved in the interaction with LcCRFB1, LcCRFB1-Ex protein samples were titrated with deuterated LcIFNd by NMR titrations. Superimposition of the two-dimensional ¹H-¹⁵N-HSQC spectrum of LcIFNd and LcIFNd+LcCRFB1-Ex showed that the LcCRFB1-Ex induced significant structural changes in the LcIFNd (Fig. 3A), indicating formation of the LcCRFB1-Ex–LcIFNd complex. In contrast, few amino acids underwent significant chemical shifts upon titration of the LcCRFB5-Ex with LcIFNd (Fig. 3B), indicating that chemical exchange processes between LcCRFB5-Ex and LcIFNd are in the slow-exchange regimen. We also observed that chemical shift perturbations for LcIFNd mainly localized to residues within the 112–136 region (Fig. 3Ca). In particular, the residues shifted mainly in helix E (residues 114–133) and E–F loop (residues 134–136) (Fig. 3Cb), with addition of several scattered residues, such as A42, K59, and A138, and R144 (Fig. 3Ca).

FIGURE 3.

The helix E on LcIFNd is a key region involved in LcCRFB1 binding. (A) ¹H-¹⁵N HSQC spectra of LcIFNd (blue) and LcIFNd in complex with LcCRFB1-Ex (pink). Amino acids exhibiting chemical shifts are labeled with their name and residue number. (B) ¹H-¹⁵N HSQC spectra of LcIFNd (blue) and LcIFNd in complex with LcCRFB5-Ex (wine red). Note that few amino acids underwent significant chemical shifts. (C) a, Amino acid sequence of LcIFNd. Rectangles underneath the sequence represent α helices A–F. Amino acid residues with chemical shifts are highlighted in green. b, Structure of LcIFNd with the LcCRFB1-binding region as identified by NMR titration (as in A) colored green. c, Structure of LcIFNd with side chains of positively charged residues K113, K114, H119, K123, R124, H128, K131, K132, and H135 on α helix E shows as blue sticks. (D) Results of pull-down assays examining binding between His-tagged LcCRFB1-Ex and GST-tagged LcIFNd mutant variants. Histograms represent the pull-down ability based on band intensities relative to the value of wild-type GST-LcIFNd pulled down by Flag-LcCRFB1-Ex, which was defined as 1. The band intensities were quantified using ImageJ software. Bars represent the means of three independent experiments ± SD. All experiments were performed in triplicate. *p < 0.05, **p < 0.01.

Considering that the aforementioned chemical shift perturbation occurred upon adding LcCRFB1-Ex to LcIFNd, we hypothesized that the shifted residues may be involved in LcCRFB1 binding. To verify our hypothesis, we performed pull-down assays with LcIFNd in which shifted residues were mutated to alanine. The shifted residues were selected based on their side chain properties or spatial distribution (Fig. 3Cc), which resulted in preparation of nine single-residue mutations in helix E (K113A, K114A, H119A, K123A, R124A, H128A, K131A, K132A, and H135A) and a multiple-residue mutation in which all nine residues were mutated to alanine (referred to as MαE). The results of pull-down assays showed that mutants MαE, K113A, K114, H119, K123A, R124A, H128A, and H135A significantly reduce the ability of LcIFNd to bind LcCRFB1-Ex (Fig. 3D). These findings indicate that helix E of LcIFNd (particularly residues K113, K114, H119, K123, R124, H128, and H135) is the key structural element involved in LcCRFB1 binding.

The interface formed by helices C–D–E of LcIFNd mediates binding to LcCRFB1

Although the aforementioned results of NMR titration and pull-down assays show that the helix E of LcIFNd is crucial for LcCRFB1 binding, we speculate that the binding interface is likely to encompass additional regions of LcIFNd. Moreover, NMR titration with LcIFNd and LcCRFB5-Ex did not produce any significant chemical shifts in LcIFNd, leaving the interaction uncharacterized. Therefore, to further characterize the interactions between LcIFNd and LcCRFB1 and LcCRFB5, we predicted the structures of LcCRFB1-Ex and LcCRFB5-Ex using AlphaFold v2.0 and constructed structural models of LcIFNd–LcCRFB1-Ex and LcIFNd–LcCRFB5-Ex complexes (described in Materials and Methods). The structural models strongly suggest the involvement of the “elbow” regions formed between two bent FNIII domains (D1 and D2) of LcCRFB1-Ex and LcCRFB5-Ex in LcIFNd recognition (Supplemental Fig. 1G, 1H) (40, 41). This subsection focuses on the analysis of LcIFNd–LcCRFB1 interactions, whereas the LcIFNd–LcCRFB5-Ex interactions are analyzed in detail in the following subsection.

Electrostatic surface potential analysis showed that the surface of helix E in LcIFNd is formed mainly by positively charged amino acids, which is in line with data from the NMR titration experiments (Fig. 4A, left panel). Conversely, parts of helix C, loop CD, and helix D (residues E67, E68, E74, E75, D76, E85, and E89) form a continuous negatively charged surface area (Fig. 4A, left panel). The surface charge on one face of LcIFNd is thus distributed to form a positively charged region (PCR) and a negatively charged region (NCR). Notably, we also found that the elbow region of LcCRFB1-Ex contains prominent positively and negatively charged surfaces (Fig. 4A, right panel). These observations suggest that binding of LcIFNd to LcCRFB1-Ex may rely on paired interactions with the surfaces of opposite charge on LcCRFB1-Ex, which would form two binding pockets (site 1 and site 2) (Fig. 4B, 4C). In site 1, the intermolecular contacts are established between the helix E of LcIFNd and the D2 domain of LcCRFB1-Ex, with salt bridges or polar interactions formed between residue pairs K113_IFNd-D125_CRFB1, K114_IFNd-E116_CRFB1, H119_IFNd-E122_CRFB1, K123_IFNd-D124_CRFB1, H128_IFNd-D153_CRFB1, and H135_IFNd-E152_CRFB1 (Fig. 4Ca). In site 2, the intermolecular contacts are established between the helices C–D of LcIFNd and the D1 domain of LcCRFB1-Ex, with salt bridges or polar interactions formed between residue pairs E67_IFNd-K45_CRFB1, E74_IFNd-R37_CRFB1, E74_IFNd-K39_CRFB1, D76_IFNd-R63_CRFB1, E89_IFNd-R40_CRFB1, and E85_IFNd-K41_CRFB1 (Fig. 4Cb). Using isothermal calorimetry, we checked LcIFNd mutants MαE and MαC-D (all negatively charged residues in the NCR of LcIFNd substituted to alanine; more details are provided in Materials and Methods) for their ability to bind LcCRFB1-Ex in vitro. Both mutants lost their ability to bind LcCRFB1-Ex (Supplemental Fig. 1I), suggesting that helices C–D and E of LcIFNd form key interfaces for binding to LcCRFB1. Noted that although both LcIFNd and hIFN-ω interact respectively with LcCRFB1-Ex and hIFNAR1 through interface formed by helices C–D and E, the amino acid residues involved in receptor binding are generally not conserved between LcIFNd and hIFN-ω (Fig. 4Cc, d).

FIGURE 4.

The helix C–D–E interface of LcIFNd is involved in LcCRFB1 binding. (A) Left, Surface charge distribution of the helix C–D–E interface of LcIFNd. The negatively charged region (NCR) and the positively charged region (PCR) on the interface are indicated in red and blue, respectively. Residues predicted to be involved in interaction with LcCRFB1 are labeled. Right, Surface charge distribution of LcCRFB1-Ex. (B) Surface charge distribution of binding model between LcIFNd and LcCRFB1-Ex. Site 1 represents the binding region between the PCR of LcIFNd and the NCR of LcCRFB1-Ex, whereas site 2 represents the binding region between the NCR of LcIFNd and the PCR of LcCRFB1-Ex. (C) Predicted structural model of the LcIFNd–LcCRFB1-Ex complex with relative locations of the two interacting sites. The first (a) and the second (b) interacting sites are shown in a detailed view. Interacting residues of LcIFNd and LcCRFB1 are colored firebrick and green, respectively. A detailed view of superimposed LcIFNd and hIFN-ω regions mediating interactions with LcCRFB1 and hIFNAR1 (c and d, respectively) is shown. Interacting residues are shown as sticks and colored firebrick and light orange, respectively.

Identification of key LcIFNd regions and residues involved in the interaction with LcCRFB5

The crystal structure of LcIFNd and the predicted structure of LcCRFB5-Ex display high similarity to the structures of hIFN-ω and the hIFNAR2 extracellular region (18), respectively (Fig. 2B, bottom panel, Supplemental Fig. 1H, left). Considering these similarities, we superimposed the crystal structure of LcIFNd and the predicted structure of LcCRFB5-Ex with the previously reported structure of the hIFN-ω–IFNAR2 binary complex (PBD ID: 3SE4) to construct the structural model of the LcIFNd–LcCRFB5-Ex complex. LcIFNd and LcCRFB5-Ex aligned well with the hIFN-ω–IFNAR2 binary complex (Supplemental Fig. 1H, right), implying conservation of receptor binding modes. To identify the key amino acids involved in the binding of LcIFNd to LcCRFB5, we analyzed the charge distribution at the predicted LcIFNd–LcCRFB5-Ex interaction interface. The binding interface between the two proteins encompasses predominantly hydrophobic interactions, with the addition of some polar interactions (Fig. 5A, left and middle panels). Two main intermolecular interface sites can be observed in the predicted model of the LcIFNd–LcCRFB5-Ex complex (Fig. 5B). In the first site, a large hydrophobic pocket is formed between residues L18, F43, and M152 of LcIFNd and P51, I77, V79, and L95 of LcCRFB5-Ex. In addition, residues E33 and D34 of LcIFNd and residues R81 and H90 of LcCRFB5-Ex may form polar interactions (Fig. 5Ba). The second interacting site mainly comprises polar interactions or salt bridges between residues R11, E15, and D155 of LcIFNd and R140, and E141 of LcCRFB5-Ex, respectively (Fig. 5Bb). Through further comparison, we found that the amino acid residues F29, L32, and R35 of hIFN-ω involved in receptor binding via the AB loop are different from these amino acids, E33, D34, and D39 of LcIFNd (Fig. 5Bc). Similarly, the hydrophobic region consisting of L18, F43, and M150 in LcIFNd is located in the same region as the hydrophobic interaction region consisting of L17, V18, M151, F155, and L156 in hIFN-ω. In addition, L18 of LcIFNd and L17 of hIFN-ω are highly conserved in this region (Fig. 5Bd). Results of pull-down assays demonstrated that GST-tagged LcIFNd mutants L18A and F43A almost entirely abolished binding to Flag-tagged LcCRFB5-Ex, whereas mutants E15A, K145A, M152A, and D155A could weaken the binding (Fig. 5C). Altogether, our results show that the residues E15, L18, F43, K145A, M152A, and D155 of LcIFNd are important for binding to the extracellular region of LcCRFB5.

FIGURE 5.

Identification of key LcIFNd regions and residues involved in the interaction with LcCRFB5. (A) Left, Surface charge distribution of LcCRFB5-Ex. Middle, Surface charge distribution of the helix A–B–F interface of LcIFNd. Right, LcIFNd residues predicted to be involved in interaction with LcCRFB5 are labeled in the structure of LcIFNd. Positively charged, hydrophobic, and negatively charged residues are shown as blue, aquamarine, and yellow sticks, respectively. (B) Predicted structural model of the LcIFNd–LcCRFB5-Ex complex with relative locations of the two interacting sites. The first (a) and the second (b) interacting sites are shown in a detailed view. Interacting residues of LcIFNd and LcCRFB5 are colored firebrick and olive, respectively. A detailed view of superimposed LcIFNd and hIFN-ω regions mediating interactions with LcCRFB1 and hIFNAR2 (c and d, respectively) is shown. Interacting residues are shown as sticks and are colored firebrick and light orange, respectively. Residues L17 and L18 are indicated with a black arrow. (C) Results of pull-down assays examining binding between Flag-tagged LcCRFB5-Ex and GST-tagged LcIFNd and its mutant variants. Histograms represent the pull-down ability based on band intensities relative to the value of wild-type GST-LcIFNd pulled down by Flag-LcCRFB5-Ex, which was defined as 1. The band intensities were quantified using ImageJ software. Bars represent the means of three independent experiments ± SD. All experiments were performed in triplicate. *p < 0.05, **p < 0.01.

LcIFNd activates JAK-STAT signaling

Previous studies have demonstrated that L. crocea possesses an evolutionarily conserved JAK-STAT pathway that is likely to perform regulatory functions similar to those in higher vertebrates (18). To investigate LcIFNd-mediated signal transduction, we stimulated PKL cells with rLcIFNd and monitored the activation of the JAK-STAT pathway. The mRNA expression levels of JAK1, IRF9, STAT1, and STAT2 (Fig. 6Aa) and IFN-stimulated antiviral genes MxA, PKR, and viperin (Fig. 6Ab) were significantly increased at 6 or 12 h after LcIFNd stimulation. Meanwhile, rLcIFNd significantly increased phosphorylation levels of both STAT1 (Fig. 6B) and STAT2 (Fig. 6C). The antiviral activity of rLcIFNd was examined using the SIGV-infected GS cells model as described previously (25). We found that pretreatment of GS cells with rLcIFNd 2 h prior to SIGV infection resulted in significant inhibition of CPE compared with the control cells (Fig. 6Da), which was accompanied by a reduced expression of viral genes at 12 or 24 h (Fig. 6Db). These results showed that LcIFNd activates the JAK-STAT signaling pathway in a conserved manner, as found in higher vertebrates, leading to transcription of IFN-stimulated antiviral genes such as MxA, PKR, and viperin.

FIGURE 6.

Activation of JAK-STAT signaling and antiviral activity of rLcIFNd. (A) a, Expression levels of JAK-STAT pathway signaling effectors JAK1, IRF9, STAT1, and STAT2, and (b) IFN-stimulated antiviral genes MxA, PKR, and viperin in PKL cells determined by real-time PCR at 6 and 12 h after stimulation with rLcIFNd (n = 5). (B and C) Left, Western blot results showing STAT1 and STAT2 expression and phosphorylation levels in PKL cells at different time points during LcIFNd stimulation. Right, Histograms represent relative protein expression and phosphorylation levels of STAT1 and STAT2. (D) GS cells were pretreated with rLcIFNd at a final concentration of 50 ng/ml or PBS (control) for 2 h; then, the cells were infected with SGIV at a multiplicity of infection of 2. a, At 24 h postinfection, GS cells were observed for CPE using microscopy. b, At 12 and 24 h postinfection, the expression levels of MCP and VP19 genes were detected by real-time PCR and normalized to that of Ecβ-actin. Bars represent the means of three independent experiments ± SD. All experiments were performed in triplicate. *p < 0.05, **p < 0.01 (unpaired Student t test).

LcIFNd signaling requires LcCRFB1 and LcCRFB5

The findings of this study have demonstrated that LcCRFB1 and LcCRFB5 serve as receptors of LcIFNd. To further verify whether LcCRFB1 and LcCRFB5 are required for LcIFNd signaling, we prepared anti–LcCRFB1-Ex and anti–LcCRFB5-Ex pAbs and performed an Ab blockade assay. Anti–LcCRFB1-Ex pAb specificity was validated by Western blot and immunofluorescence assays (Fig. 7A), and anti–LcCRFB5-Ex pAb specificity was validated as described previously (18). PKL cells stimulated with rLcIFNd in the presence of anti–LcCRFB1-Ex, anti–LcCRFB5-Ex, or both pAbs exhibited a dramatic decrease in expression levels of JAK1, IRF9, STAT1, and STAT2 (Fig. 7B), as well as MxA, PKR, and viperin (Fig. 7C). Similarly, the phosphorylation of STAT1 and STAT2 were significantly weakened by blocking LcCRFB1 and LcCRFB5 (Fig. 7D). These results confirm that both LcCRFB1 and LcCRFB5 are required for LcIFNd signaling, and blocking either of the receptor subunits is sufficient to inhibit the activation of JAK-STAT signaling pathway, which in turn inhibit the transcription of ISGs (Fig. 8).

FIGURE 7.

LcCRFB1 and LcCRFB5 receptor blockade impairs LcIFNd signal transduction. (A) a, Left, SDS-PAGE gel showing purified rLcCRFB1-Ex. Right, Detection of endogenous full-length LcCRFB1 from PKL cells (lane 1) and spleen cells (lane 2) and rLcCRFB1-Ex (lane 3) and control (lane 4). The theoretical molecular masses of full-length LcCRFB1 and LcCRFB1-Ex are 59.62 and 22.76 kDa, respectively. b, Verification of anti–LcCRFB1-Ex pAb specificity by immunofluorescence labeling in PKL and HEK-293T cells. The pCMV-HA plasmid was overexpressed in HEK-293T cells as control, and pCMV-HA-LcCRFB1 containing full-length LcCRFB1 cDNA was overexpressed in HEK-293T cells as the experimental group. (B) JAK1, IRF9, STAT1, and STAT2 and (C) MxA, PKR, and viperin expression levels upon receptor blockade with anti–LcCRFB1-Ex, anti–LcCRFB5-Ex, or both pAbs in LcIFNd-stimulated PKLs. (D) STAT1 and STAT2 phosphorylation were inhibited by anti–LcCRFB1-Ex, anti–LcCRFB5-Ex, or both Abs in LcIFNd-stimulated PKLs. Bars represent the means of three independent experiments ± SD. All experiments were performed in triplicate. Different lowercase letters indicate significant differences. **p < 0.01.

FIGURE 8.

LcIFNd signaling and ligand–receptor binding models. (A) The teleost fish type I group I IFNd binds to the surface receptor complex consisting of the receptor subunit CRFB5 and CRFB1, triggering the recruitment and activation of JAK1 and TYK2, which then phosphorylate STAT1 and STAT2, respectively. Phosphorylated STAT1 and STAT2 dimerize and assemble with IRF9 to form a transcription factor complex known as IFN-stimulated gene (ISG) factor 3 (ISGF3). ISGF3 then translocates to the nucleus and binds IFN-stimulated response elements (ISREs), activating the transcription of ISGs. (B) Cartoon representation of the structural model of the LcIFNd–LcCRFB1-Ex–LcCRFB5-Ex complex.

To investigate the importance of the aforementioned key LcIFNd residues involved in receptor binding on LcIFNd signaling, PKL cells were stimulated with the relevant LcIFNd mutants, and expression levels of key signaling molecules (JAK1 and STAT1) and antiviral effectors (MxA and PKR) were quantified at 6 h poststimulation. Compared to wild-type LcIFNd, the mRNA levels of the four genes were significantly reduced when PKL cells were stimulated with the LcCRFB1-Ex–binding mutants MαE, K113A, H119A, K123A, R124A, H128A, H135A, and MαC-D (Supplemental Fig. 2A). Similarly, the mRNA levels of the same genes were significantly reduced when stimulating cells with LcCRFB5-Ex–binding mutants E15A, L18A, F43A, K145A, M152A, and D155A (Supplemental Fig. 2B). Taken together with the results of pull-down assays and the LcIFNd mutants described above (Figs. 3D, 5C), these data suggest that the LcIFNd residues K113, H119, K123, R124, H128, and H135 are critical for binding to LcCRFB1, and that LcIFNd residues E15, L18, F43, K145, M152, and D155 are critical for binding to LcCRFB5.

Discussion

In mammals, type I IFNs are the main orchestrators of the antiviral innate immune response. All 16 human type I IFNs initiate signaling by binding to the same receptors composed of the IFNAR1 and IFNAR2 subunits. In zebrafish and pufferfish, however, genome-wide sequence analyses demonstrated the presence of at least 17 CRFB members (19, 42). Recent studies on type I IFN receptors in teleost fish have focused on signaling through CRFB1, CRFB2, and CRFB5. Thus far, research suggests that fish group I IFNs signal through the receptor complex consisting of CRFB1–CRFB5 and that fish group II IFNs signal through CRFB2–CRFB5 (18, 20). However, CiIFNa in grass carp (a group I IFN) was found to signal through CRFB1, CRFB2, and CRFB5 (23). Here, our findings unambiguously demonstrate that the IFNd from L. crocea, a group I IFN, binds the CRFB1/CRFB5 receptor pair (Fig. 1A, 1B).

The ITC results in this study show that LcIFNd binds LcCRFB5-Ex with a higher affinity (K_d = 203.80 nM) than LcCRFB1-Ex (K_d = 1.28 μM) (Fig. 1C, 1D). Furthermore, we did not observe significant ¹H and ¹⁵N chemical shifts in LcIFNd upon LcCRFB5-Ex titration (Fig. 3B), indicating that LcCRFB5-Ex binding to LcIFNd is in a slow exchange on the NMR time scale. These findings are also in agreement with our previous observation in L. crocea where LcCRFB5 is the higher affinity receptor (K_{d =} 74.64 nM) than LcCRFB2 (K_d = 996.90 nM) for LcIFNi (18). The grass carp IFNa also appears to have a higher binding affinity for CRFB5 than for CRFB1 and CRFB2 (23), further suggesting that CRFB5 may act as a higher affinity receptor for teleost type I IFNs, whereas CRFB1 and CRFB2 may act as lower affinity receptors. Notably, IFNAR1, the ortholog of CRFB5 in mammals, serves as the high-affinity receptor for IFN-β in mice, but as the low-affinity receptor for IFN-β and IFN-α2 in humans (11, 43). These findings demonstrate that the affinity of type I IFN for their receptors may be rather varied among species.

Our structural data and analysis may also explain why fish group I and group II type I IFNs bind to CRFB1–CRFB5 and CRFB2–CRFB5 receptor complexes, respectively. The helical element in the AB loop, which is considered to be a key site involved in receptor interaction in group II IFNs such as LcIFNi and zIFN-φ2 (14, 18), is absent in the group I IFNs LcIFNd, zIFN-φ1 (24), and CiIFNa (23) (Fig. 2C). Furthermore, the LcCRFB5/LcIFNd interface is formed by residues of the D1 and D2 subdomains of LcCRFB5 and by helices A, B, and F of LcIFNd, whereas the LcCRFB5/LcIFNi interface is formed by residues of the D1 and D2 subdomains of LcCRFB5 and by helices C, D, and E in LcIFNi (18). The differences in AB loop structure and the composition of the CRFB5-interacting interface could therefore play an important role in conferring affinity for different receptor complexes to fish group I and group II type I IFNs.

Recent studies have confirmed that the overall structure of fish type I IFNs is similar to that of mammalian type I IFNs (44, 45), as is the case with LcIFNd reported in our study. Nevertheless, we also observed notable differences between IFNd receptor and IFN-ω receptor binding modes, including the helical domain and amino acid species. Interestingly, the region encompassing L18, F43, and M152 in LcIFNd and the region encompassing L17, V18, M151, F155, and L156 in hIFN-ω form a hydrophobic pocket in corresponding locations (Fig. 5Bd). Moreover, L18 of LcIFNd and L17 of hIFN-ω are fully conserved in this region (Fig. 5Bd), suggesting a certain degree of conservation of residues involved in receptor binding despite overall low sequence similarity between LcIFNd and hIFN-ω.

The ancient and powerful host first-line antiviral defense strategy of the type I IFN system is conserved from mammals to teleosts (46). Zebrafish IFN-φ1–4 (20, 24), Atlantic salmon IFNa1 (47), grass carp IFNa (23, 48), and large yellow croaker IFNi (18, 25) have all been shown to induce antiviral response through the JAK-STAT signaling pathway. Our findings show that LcIFNd also signals through this conserved pathway (Fig. 6). We further found that the LcIFNd residues involved in receptor interaction (Supplemental Fig. 2C) are highly conserved in the members of fish type I subgroup d among other fish species, suggesting that the manner by which fish type I subgroup d IFNs bind to their receptors CRFB1 and CRFB5 is conserved. The mechanisms by which LcIFNd binds to the receptors LcCRFB1 and LcCRFB5 and subsequently activates antiviral response through JAK-STAT signaling pathway are summarized in Fig. 8.

In summary, this study demonstrates that LcCRFB1 and LcCRFB5 are the functional receptors of LcIFNd, reveals a ligand/receptor complex binding mechanism of IFNd in teleost fish, and suggests that the receptor binding mechanism is determined by structural differences between group I and group II IFNs of teleost type I IFNs. Our results thus provide new insights into function and evolution of type I IFNs.

Disclosures

The authors have no financial conflicts of interest.