Alternative Splicing Transcripts of Zebrafish LGP2 Gene Differentially Contribute to IFN Antiviral Response

In mammals, host cells possess an innate ability to recognize virus infection and mount a powerful antiviral response. Virus RNA accumulating in the cytoplasm of infected cells can be sensed through cytosolic pattern recognition proteins including retinoic acid–inducible gene-I (RIG-I) and melanoma differentiation–associated gene 5 (MDA5), two well-characterized members of the RIG-I like receptor (RLR) family, both of which harbor three conserved domains by the presence of two N-terminal caspase activation and recruitment domains (CARDs), a central DExD/H-box RNA helicase domain, and a C-terminal regulatory domain (RD) (1, 2).

Extensive studies have suggested a canonical paradigm of RLR signaling. RIG-I undergoes a conformational change upon binding to 5′ppp- or 5′pp-dsRNA through RD and helicase domain, therefore releasing N-terminal CARDs from an autoinhibitory state in an ATP-dependent manner, and subsequently interacting with adaptor mitochondrial antiviral signaling protein (MAVS) for downstream signaling cascade. MDA5 preferentially binds to long dsRNAs to form protein-coated filaments, thus resulting in oligomerization of tandem CARDs to activate MAVS (3). The resultant MAVS activation finally facilitates protein kinases TANK-binding kinase 1 (TBK1) to phosphorylate and activate IFN regulatory factors 3/7 (IRF3/7), leading to the induction of type I IFN and IFN-stimulated genes (ISGs) for the establishment of a broadly effective antiviral state (1, 2). Mediator of IRF3 activation (MITA) has also been reported as a scaffold protein for linking TBK1 and IRF3 to the MAVS complex upon RNA virus infection (4), although it is believed to primarily participate in virus DNA–directed IFN signaling (5).

Laboratory of genetics and physiology 2 (LGP2) is the third member of the RLR family, which shares the helicase domain and RD but lacks the two N-terminal CARDs that are required for signaling (6, 7). The absence of N-terminal CARDs makes it difficult to understand LGP2’s roles in RLR-mediated signaling, and the exact function of LGP2 is puzzling (3). LGP2 is initially identified as a negative regulator of IFN response triggered by Sendai virus and Newcastle disease virus, which are sensed exclusively by RIG-I (7–10), and also by polyinosinic:polycytidylic acid [poly(I:C)], a synthetic dsRNA sensed by either RIG-I or MDA5 (9). Together with the elevated expression of LGP2 by virus infection and IFN treatment, these results support a notion that LGP2 acts as a feedback regulator of RLR signaling (7–9). However, three separate strains of LGP2^−/− mice exhibit disparate phenotypes (11–13). In-depth delineation of the former two strains of deficient mice infected with a similar virus set show a negative or positive role of LGP2 in RLR signaling, respectively (11, 12). Analysis of the third strain reveals that LGP2 is not essential for the induction of innate immune defenses but rather controls CD8⁺ T cell survival and fitness in response to Sendai virus, Dengue virus type 2, or West Nile virus (the latter two are sensed by RIG-I and MDA5) (13).

Despite these differences, the former two LGP2^−/− mice show a common deficiency in resistance to encephalomyocarditis virus infection (11, 12). Because encephalomyocarditis virus RNA is associated with LGP2 as a physiological agonist of MDA5-dependent signaling (14), these data raise a possibility that LGP2 may function as a positive regulator of MDA5 signaling, which is indeed supported by in vitro assays where LGP2 appears to activate MDA5-dependent signaling but repress RIG-I–dependent signaling (15, 16). Mechanistically, LGP2 facilitates MDA5 signal transduction dependent of its RNA binding activity and ATP hydrolysis activity (16–20), and also by modulation of MDA5-RNA interaction (19). Several studies suggest that the cellular LGP2 expression level might be a key factor for the regulation of a switch between positive and negative roles in RLR signal transduction: at lower levels, LGP2 synergizes with MDA5 but not RIG-I to augment IFN signaling; at higher levels, LGP2 acts as an inhibitor of RIG-I and MDA5 signaling (16, 18, 19). However, greater knowledge is required to fully appreciate how LGP2 exerts opposing effects.

To better understand the function of LGP2, we are determined to characterize zebrafish LGP2’s roles in RLR-directed IFN signaling from an evolutionary perspective. The zebrafish model is becoming useful in the study of the vertebrate innate immune response (21). All three RLR receptors, together with the downstream molecules such as MAVS, MITA, TBK1, IRF3, and IRF7, exist in zebrafish genomes (21, 22). Although fish IFNs are not classified into IFN-α/β, but group I and group II IFNs based on cysteine numbers (23–25), and they signal through specific receptors that differ from mammalian type I IFN receptors (24), fish RIG-I and MDA-5 direct IFN expression through a conserved pathway (26–30), indicating that zebrafish possess a highly developed immune system that is remarkably similar to that of mammals. Surprisingly, published documents have shown antithetic biological activities of fish LGP2s (27, 28, 31–34), an interesting and contradictory phenomenon appeared in mammals.

In the current study, we found that three zebrafish LGP2 transcripts, a full-length LGP2 and two truncating variants, LGP2 truncating variant (LGP2v)1 and LGP2v2, are derived from a single LGP2 gene and differentially contribute to host IFN response. Full-length LGP2 alone has the potential to activate IFN signaling, which is particularly significant at limited expression levels of LGP2, but inhibits IFN induction by poly(I:C), a mimic of virus infection, predominantly at high expression levels of LGP2. LGP2v1 or LGP2v2 just show a weak negative regulation in LGP2- or poly(I:C)-directed IFN response. Further studies suggested that LGP2 has bilateral functions and switches its role from a positive to a negative regulator in the presence of poly(I:C) transfection or during spring viremia of carp virus (SVCV) infection. The data are helpful to understand the opposing function of LGP2 in fish and mammals, which is seen in independent experiment systems.

Fish cell lines, including Crucian carp (Carassius auratus L.) blastula embryonic cells (CAB), epithelioma papulosum cyprini cells (EPC), zebrafish liver cells (ZFL), and ZF4, were cultured as described previously (35, 36). SVCV and grass carp reovirus (GCRV) were propagated and titered according to the method of Reed and Muench, by a 50% tissue culture–infective dose (TCID₅₀) assay on EPC cells and CAB cells, respectively. Zebrafish (Danio rerio) strain AB were raised, maintained, reproduced, and staged according to standard protocols. For viral infection, i.p. injection was performed using 50 μl of SVCV (10⁸TCID₅₀ per ml) per fish.

Total RNA was extracted using Trizol Reagent (Invitrogen), and the RNA was treated with RNase-free DNase I (Promega) according to the manufacturer’s protocol. The DNase I-treated RNA was reverse transcribed into cDNA using a first-strand cDNA synthesis kit (Promega) and kept at −20°C for semiquantitative RT-PCR or quantitative real-time PCR (RT-qPCR) analysis. For RT-qPCR, all samples were analyzed in triplicate and the expression values were normalized to β-actin. Primers used for RT-PCR analysis are listed in Supplemental Table I.

For coimmunoprecipitation (Co-IP) assays, expression plasmids LGP2-myc, LGP2v1-myc, and LGP2v2-myc were generated by insertion of the open reading frames (ORF) of zebrafish LGP2, LGP2v1, and LGP2v2 into Not I and BamH I sites of pcDNA3.1(-)-myc vector. For luciferase assays, free-tagged expression plasmids were made by cloning the ORFs of zebrafish LGP2, LGP2v1, and LGP2v2 into BamH I and Not I sites of pcDNA3.1(+) vector. Two dominant negative mutant plasmids, DrRIG-I-DN and DrMDA5-DN, were made by cloning the corresponding sequences (corresponding to 197–937 aa of DrRIG-I and 224–997 aa of DrMDA5) into Nhe I and Kpn I sites of pcDNA3.1(+) vector. All constructs were confirmed by sequencing. Other plasmids, including DrIFN-φ1pro-luc, DrIFN-φ3pro-luc, EPC IFNpro-luc, DrRIG-INter, DrMDA5, DrMAVS, DrMITA, DrTBK1, DrIRF3, DrIRF7, and some dominant negative mutant plasmids including CaRIG-I-DN, CaMDA5-DN, DrMAVS-ΔTM, CaMITA-CT, CaTBK1-K38M, DrIRF1DN, DrIFRF3DN, DrIRF7DN, DrSTAT1a-ΔC, and DrSTAT1b-ΔC were described previously (27, 37, 38).

According to previous reports (26, 27, 35), cells were seeded in 24-well plates, 6-well plates, 3.5 cm dishes, or 10 cm dishes overnight and cotransfected with various constructs at a ratio of 10:10:1 (promoter-driven luciferase plasmid/expression plasmid/Renilla luciferase plasmid pRL-TK) using FuGENE HD Transfection Reagent (Promega). Empty vector pcDNA3.1 was supplemented to ensure equal amounts of the transfected DNA in total among wells. If necessary, the cells were transfected again with poly(I:C), infected with SVCV or GCRV, or transfected with poly(I:C) followed by virus infection. At the indicated time points, the cells were harvested and lysed according to the Dual-Luciferase Reporter Assay System (Promega). Luciferase activities were measured by a Junior LB9509 luminometer (Berthold, Pforzheim, Germany) and normalized to the amounts of Renilla luciferase activities. Unless indicated, the results were representative of more than three independent experiments, each performed in triplicate.

Co-IP assays were performed to determine the levels of phosphorylated fish IRF3. Briefly, cells were extracted in NP-40 lysis buffer (Beyotime), protease inhibitors, and phosphatase inhibitors (Roche) for 30 min on ice. Following removal of nuclei and cellular debris by centrifugation, adequate amounts of supernatants (e.g., four-fifths of a 10 cm dish) were incubated with 50 μl of the designated phosphoprotein Ab that is linked to agarose (Immunechem) for 24 h at 4°C. This phosphoprotein Ab can precipitate all cellular phosphorylated proteins with the phosphorylated serine, threonine, and tyrosine. The agarose-protein complex was washed five times with washing buffer [50 mM Tris-HCl (pH7.5), 150 mM NaCl, 1 mM DTT, 1% NP-40]. By heating the samples for 8 min at 99°C in SDS-PAGE protein loading buffer (1×; Beyotime), the precipitated proteins were released from the immunoprecipitated pellets, separated on SDS-PAGE, transferred to polyvinylidene fluoride membranes (Millipore), and analyzed by immunoblotting with crucian carp C. auratus IRF3 (CaIRF3) polyclonal Ab.

CaIRF3-specific Ab and CaPKR-specific Ab were described previously (26, 35). CaIRF7 Ab was made by immunization of rabbits with the purified recombinant protein of DBD domain of CaIRF7. DrLGP2-specific Ab was generated by immunization of rabbits with a purified peptide corresponding to 192–417 aa of zebrafish LGP2. The recognition specificity of DrLGP2 Ab and no cross-recognition between CaIRF3 Ab and CaIRF7 Ab have been verified by immunoadsorption assays according to our previous study (26).

EPC or ZF4 cells cultured in 10 cm dishes were transfected with indicated plasmids. Then 48 h later, the transfected cells were treated for 10 min with 1% formaldehyde at room temperature, collected by cell scraper, lysed, and ultrasonicated to obtain chromatin fragments between 100 and 1000 bp in size. One-tenth of cell lysates were incubated overnight at 4°C, with CaIRF3 Ab and CaIRF7 Ab, respectively, with preimmune serum, or without any Ab as two negative controls, followed by incubation for another 2 h with Dynabeads protein A (Invitrogen) (25 μl per well) at 4°C to form the cross-linked complexes. The cross-linked complexes were added with NaCl (0.2 M) at 95°C for 30 min to reverse formaldehyde cross-links, treated with proteinase K and RNase A sequentially at 62°C for 30 min. Finally, the immunoprecipitated DNA was extracted by the phenol-chloroform method and subsequently quantified using RT-qPCR. The enrichment was determined relative to input following normalization to no-Ab controls.

Using a pair of primers to amplify the entire ORF of zebrafish LGP2 gene from SVCV-infected zebrafish spleen, we cloned three transcripts that are 2040, 1728, and 1644 bp, respectively (Fig. 1A). The zebrafish genome harbors a single LGP2 gene in chromosome 3, which is composed of 12 exons and 11 introns (Fig. 1B), a gene organization similar to that in mouse and flounder (31, 39). These zebrafish transcripts appear to originate from the single LGP2 gene and are generated by alternative splicing, a regulated process that results in three different-size proteins, LGP2 full length, LGP2v1, and LGP2v2 (Fig. 1B). LGP2 represents the full-length transcript, including all 12 exons and encoding a 679 aa-protein; LGP2v1 and LGP2v2 are two shorter ones that lack the exon 9 or the exons 3 and 4, encoding a 575 aa or a 547 aa protein, respectively (Fig. 1C). The full-length LGP2-encoding protein contains an N-terminal DExD/H-box RNA helicase domain predicated with ATP-binding motif and ATPase motif, an intermediate HELICc domain with RNA-binding motif, and a C-terminal RD domain with RNA binding loop and two Zn²⁺-binding motifs; however, LGP2v1 and LGP2v2 have an incomplete DExDc domain and an incomplete HELICc domain, respectively (Fig. 1D).

The three transcripts were detectable in resting cultured zebrafish cells (ZFL), and were significantly upregulated by poly(I:C) transfection showing different expression patterns. The full-length LGP2 was most abundant and similar to Mxb, a typical ISG, was upregulated during the whole inducing course, whereas the expression of LGP2v1 and LGP2v2 peaked at 6 h posttransfection then dropped slightly thereafter (Fig. 2A). Using a pair of universal primers to amplify a shared sequence, RT-qPCR analysis showed that total LGP2 was constitutively expressed in all zebrafish tissues detected (Fig. 2B) and upregulated in response to SVCV infection, displaying a similar expression pattern to Mxb (Fig. 2C). Similar to that in ZFL cells, the three splicing forms were constitutively detected in immune tissues including gill, liver, spleen, head kidney, and body kidney, and were significantly induced by SVCV infection, with stronger expression of LGP2 than of LGP2v1 and LGP2v2 (Fig. 2D). Expression comparison showed a very low transcription level of LGP2, RIG-I, and MDA5 in resting tissues but an upregulated expression in SVCV-infected tissues with RIG-I > LGP2 > MDA5 (Fig. 2E).

To determine the roles of three zebrafish LGP2 transcript variants in IFN antiviral response, EPC cells were transfected with different LGP2 expression constructs together with EPC IFN promoter-driven luciferase constructs followed by detection of luciferase activities. As shown in Fig. 3A, EPC IFN promoter was significantly activated when LGP2 was transfected at low doses (10, 50, 100 ng). Although overexpression of LGP2 at a high dose (200 ng) did not give an obvious stimulatory effect at 28 h posttransfection, extending overexpression time up to 55 h unexpectedly resulted in a definite increase of luciferase activities, comparable to an effect by 10 ng of LGP2 (Fig. 3A). Under the same conditions, neither LGP2v1 nor LGP2v2 showed any stimulatory potential (Fig. 3A). Consistently, overexpression of LGP2 but not LGP2v1 and LGP2v2 in EPC cells for 55 h upregulated the transcription of cellular IFN genes together with four ISGs including IRF3, IRF7, viperin, and Mx, showing a better stimulating effect at low doses (50, 100 ng) than at a high dose (200 ng) (Fig. 3B).

In subsequent experiments, transfection of EPC cells with LGP2 at low doses (10, 50, 100 ng) for 24 h provoked a strong activation of two zebrafish IFN promoters, derived from DrIFN-φ1 and DrIFN-φ3 genes (27), whereas a high dose (200 ng) gave a weak, even undetectable activation; neither LGP2v1 nor LGP2v2 alone had any stimulatory effect (Fig. 3C). Similar results were seen when transfection assays were carried out in CAB cells (Fig. 3D). These results suggest a potential of LGP2 alone rather than LGP2v1 or LGP2v2 to activate IFN response.

Our previous results have confirmed the conservation of RLR-IFN signaling in fish by overexpression of dominant negative mutants of fish RLR signaling molecules, including MAVS, MITA, TBK1, IRF3, and IRF7, to block the activation of fish IFN promoters by poly(I:C) and RIG-I/MDA5 (26, 27, 30, 35, 37). The same strategies were used to investigate the detailed molecular events involved in zebrafish LGP2-dependent signaling. Similarly, 50 and 100 ng of LGP2 markedly activated the promoters of DrIFN-φ1 and DrIFN-φ3, which was stably abolished by cotransfection of the dominant negative mutants of fish MAVS, TBK1, IRF3, or IRF7 (DrMAVS-ΔTM, CaTBK1-K38M, DrIRF3DN, DrIRF7DN) (Fig. 4A, 4B). An obvious blockade was seen in cells that were cotransfected with 50 ng of LGP2 and the dominant negative mutant of fish MITA (CaMITA-CT), although a slight, or no, reduction in both zebrafish IFN promoter activities was detected when 100 ng of LGP2 was used (Fig. 4A, 4B).

In addition, the activation of EPC IFN promoter by LGP2 required MAVS, MITA, and TBK1, as evidenced by the results that the activation was severely inhibited by overexpression of DrMAVS-ΔTM, CaMITA-CT, or CaTBK1-K38M (Fig. 4C). LGP2 alone also provoked ISRE-containing promoter-driven luciferase activities, which was diminished by transfection of either or both of zebrafish STAT1a-ΔC and STAT1b-ΔC, two dominant negative mutants of STAT1 (37) (Fig. 4D). These results indicate that zebrafish LGP2 induces ISG expression through the Stat1 pathway. Consistently, overexpression of LGP2 in EPC cells induced an increase in IRF3 and PKR proteins (Fig. 4E), which were markedly inhibited by cotransfection with DrMAVS-ΔTM, CaTBK1-K38M, DrIRF3DN, and DrIRF7DN, respectively (Fig. 4F). These results indicate that LGP2 activates IFN response by upregulation of IFN and ISG expression via a signaling pathway similar to RIG-I/MDA5 signaling.

To strengthen our findings, chromatin immunoprecipitation (ChIP) assays were used to compare the binding affinity of IRF3/7 to IFN promoters under different conditions. As shown in Fig. 4G, the binding of the endogenous IRF3 and IRF7 to DrIFN-φ1 and DrIFN-φ3 promoters or endogenous EPC IFN promoter was enhanced in EPC cells transfected with LGP2 compared with cells transfected with an empty vector; however, the enhanced binding was significantly diminished by cotransfection of TBK1-K38M, MAVS-ΔTM, and MITA-CT, respectively (Fig. 4G). Co-IP assays showed that LGP2, but not LGP2v1 or LGP2v2, significantly promoted the phosphorylation of IRF3 (Fig. 4H). These results indicate that zebrafish LGP2 activates IFN response dependent of IRF3 phosphorylation and IRF3/7 binding to IFN promoters.

To determine a possible role of LGP2, LGP2v1, and LGP2v2 in response to virus infection, EPC cells were transfected with EPC IFN luciferase construct and diverse LGP2 constructs followed by SVCV infection. Luciferase assays showed significant activation of EPC IFN promoter by LGP2, with a better effect at low amounts (10, 50, 100 ng) than large (200 ng) (Fig. 5A), but not by LGP2v1 and LGP2v2 at any an amount followed (or not) by SVCV infection (Supplemental Fig. 1). SVCV alone (1 × 10⁴ TCID₅₀ per ml) failed to induce an obvious activation until 48 h postinfection, and following SVCV infection did not give an enhanced activation of IFN promoter by LGP2, except for an additional effect (Fig. 5A).

Subsequently, the expression of SVCV genes as well as cellular IFN and some ISGs was detected to evaluate the antiviral effects of LGP2, LGP2v1, and LGP2v2 against SVCV infection. The SVCV genome contains five ORFs for nucleoprotein (N), phosphoprotein (P), matrix protein (M), glycoprotein (G) and RNA polymerase (L) (40). RT-qPCR showed that compared with transfection of empty vector, LGP2v1 or LGPv2, overexpression of LGP2 decreased the transcription expression of SVCV genes L, N and G (Fig. 5B) but upregulated the expression of cellular IFN and three representative ISGs, viperin, Mx, and IRF7 (Fig. 5C). Consistently, transfection of LGP2 (100 ng) resulted in an over 3-fold reduction of virus titer relative to the control cells that was transfected with an empty vector, LGP2v1, or LGP2v2 (100 ng); this inhibitory effect was comparable to that by transfection of the same amount of MDA5 and RIG-INter (2.4- and 3.4-fold reduction versus control, respectively), but less than that by transfection of poly(I:C) (100 ng/ml) (18-fold reduction versus control) (Fig. 5D). Moreover, LGP2 exhibited a better stimulatory potential to zebrafish IFN promoters at low doses (<100 ng) but a poorer one at high doses (>100 ng) than MDA5 or RIG-INter at the same doses (Fig. 5E). These data indicate that zebrafish LGP2 rather than LGP2v1 and LGP2v2 confers protection on EPC cells against SVCV infection by induction of IFN response, and at relatively low expression levels, LGP2 alone exerts more significant protection than either RIG-I or MDA5.

Several published studies revealed a negative role of fish LGP2 (26, 32, 33), which is contradictory to the results described above. To clarify this, similar experiments were initially performed by transfection of EPC cells with three LGP2 expression constructs followed by transfection of poly(I:C), an effective IFN stimulator. LGP2 alone but not LGP2v1 or LGP2v2, as expected, displayed a stably positive role in activating DrIFN-φ1 and DrIFN-φ3 promoters, and transfection of poly(I:C) (1 μg/ml) alone stimulated nearly 7-fold more activity of fish IFN promoters than 200 ng of LGP2; however, transfection of EPC cells with LGP2 and subsequently with poly(I:C) resulted in IFN promoter-driven luciferase activities significantly decreased by 23 ∼ 27% relative to poly(I:C) alone, indicating a negative role of LGP2 in poly(I:C)-triggered IFN response (Fig. 6A).

Titration experiments showed a dose-dependent inhibition of LGP2 in the activation of both fish IFN promoters by poly(I:C) and by RIG-I or MDA5, respectively (Fig. 6B, 6C). Intriguingly, larger doses (100 or 200 ng) of LGP2v1 and LGP2v2 also displayed a slight inhibition to DrIFN-φ1 promoter under the same conditions (Fig. 6B, 6C). Similarly, the three zebrafish LGP2 isoforms impeded MAVS-, MITA- and TBK1-mediated IFN signaling, with a more significant effect by LGP2 than that by LGP2v1 and LGP2v2, but did not block IRF3- and IRF7-mediated signaling (Fig. 6D). Interestingly, LGP2 significantly facilitated IRF7 to activate DrIFN-φ1 and DrIFN-φ3 promoters (Fig. 6D). Therefore, three zebrafish LGP2 isoforms negatively regulate poly(I:C)-triggered fish IFN response to different degrees.

The inhibitory role of LGP2 was further determined by transcription detection of ISGs. Semiquantitative RT-PCR showed that transfection of poly(I:C) in ZFL cells strikingly induced, in a time-dependent fashion, the transcription of cellular genes involved in IFN antiviral response, including MDA5, MAVS, IFN-φ1, IFN-φ3, IRF3, and Mxb; however, overexpression of LGP2 decreased the transcription induction of these genes by poly(I:C) (Fig. 7A). These results were further verified by RT-qPCR analysis (Supplemental Fig. 2). In addition, poly(I:C) transfection induced a time-dependent upregulation of LGP2, IRF3, and PKR proteins in ZFL, but overexpression of LGP2 abrogated poly(I:C)-induced protein levels of IRF3 and PKR (Fig. 7B).

In final experiments, Co-IP assays showed that IRF3 phosphorylation was enhanced in ZFL cells transfected with either LGP2 or poly(I:C) alone; however, cotransfection of LGP2 and poly(I:C) resulted in a decreased phosphorylation of IRF3 relative to transfection of poly(I:C) alone (Fig. 7C). Consistently, transfection of either LGP2 or poly(I:C) alone in ZF4 cells increased the binding affinity of endogenous IRF3 and IRF7 to both zebrafish IFN promoters, but compared with transfection of poly(I:C) alone, cotransfection of LGP2 and poly(I:C) attenuated the binding of IRF3 to fish IFN promoters (Fig. 7D). Interestingly, poly(I:C)-mediated IRF7 binding was not diminished but enhanced by overexpression of LGP2, indicating distinct roles of IRF3 and IRF7 in the negative regulation of LGP2 on poly(I:C) signaling.

To determine the antiviral effects of LGP2 in EPC cells in the absence or presence of poly(I:C), in initial experiments, the transfection of poly(I:C) (100 ng/ml) alone showed the best activation on DrIFN-φ1 and DrIFN-φ3 promoters, being 4.3 ∼ 5.4-fold or 3 ∼ 8.4-fold higher than transfection of 100 ng of LGP2 alone or infection of SVCV (5 × 10⁴TCID₅₀ per ml) alone. However, cotransfection of LGP2 and poly(I:C) resulted in ∼2-fold reduction of IFN promoter activation relative to the transfection of poly(I:C) alone (Fig. 8A). The LGP2-mediated suppression was augmented along with the plasmid amount increasing up to 200 ng. Interestingly, the following SVCV infection did not alter the level, or only caused one additional increase, of IFN promoter activation by poly(I:C) and LGP2 individually and collectively (Fig. 8A).

In subsequent experiments, a time-dependent inhibition of LGP2 on poly(I:C)-induced IFN response was shown by luciferase assays using EPC IFN promoters instead of zebrafish IFN promoters (Fig. 8B). Simultaneous detection of virus titers from the supernatants of virally infected EPC cells revealed that transfection of poly(I:C) alone had the best inhibitory effect on SVCV replication, leading to a 5.6-, 42.2-, and 31.6-fold reduction of virus titers relative to control cells at 12, 24, and 36 h postinfection, respectively, but 3.2-, 13.3-, and 17.8-fold reduction was observed for transfection of LGP2 and poly(I:C) collectively, indicating a negative role of LGP2 (Fig. 8C). In this case, transfection of LGP2 alone resulted in 1.8-, 5.6-, and 4.2-fold reduction on SVCV titers relative to control cells (Fig. 8C). These results suggest that LGP2 alone inhibits virus replication by induction of IFN response but in the presence of poly(I:C), it attenuated the inhibition of poly(I:C) on SVCV replication likely through restraining the IFN response.

A continuous titering experiment was carried out to determine the reasons for the antithetic roles of zebrafish LGP2 involved in IFN antiviral response. Titration of poly(I:C) from 0.5 to 80 ng/ml revealed that 2.5 ng of LGP2 did not inhibit or enhance the activation of DrIFN-φ1 promoter by low concentrations of poly(I:C) (<10 ng/ml), whereas it stably inhibited DrIFN-φ1 promoter activation by high concentrations of poly(I:C) (>10 ng/ml) (Fig. 9A). Regarding DrIFN-φ3 promoter activation, 2.5 ng of LGP2 did not make any difference (Fig. 9A, upper panels). Compared to 2.5 ng of LGP2, 200 ng of LGP2 caused a strong inhibition on both zebrafish IFN promoter activation by high concentrations of poly(I:C) (>8 ng/ml), whereas low concentrations of poly(I:C)-mediated signaling were still not influenced (Fig. 9B, upper panel). Under the same conditions, overexpression of LGP2v1 or LGP2v2 with 2.5 ng had no effect on the activation of both fish promoters (Fig. 9A, middle and lower panels); however, overexpression of each with 200 ng resulted in an inhibition of DrIFN-φ1 promoter activation rather than DrIFN-φ3 promoter activation (Fig. 9B, middle and lower panels).

Subsequently, titration of LGP2 showed that although LGP2 alone at low amounts (10, 50, 100 ng) exhibited higher stimulatory potential than at a high amount (200 ng), no amount made a difference on the activation of two zebrafish IFN promoters by 2.5 ng/ml of poly(I:C), but it significantly inhibited the activation by 200 ng/ml of poly(I:C) (Fig. 9C). Similar results were seen for EPC IFN promoter activation by titration of poly(I:C), showing that LGP2 preferentially inhibited IFN promoter activation by high concentrations of poly(I:C) in a dose-dependent manner (Fig. 9D, left panel). In this experiment, the construct LGP2-myc, tagged with myc at the C terminus, lost the stimulatory potential on IFN promoter activation, but still retained the ability of impeding poly(I:C)-triggered IFN signaling (Fig. 9D, right panel). These results indicate that the exact concentration of poly(I:C) is responsible for the positive or negative role of zebrafish LGP2.

Continuous titering experiments were performed in EPC cells transfected with different doses of LGP2 followed by infection with different titers of SVCV. As shown in Fig. 10A, infection of SVCV alone induced a titer-dependent activation of two zebrafish IFN promoters. Regarding DrIFN-φ1 promoter, low titers of SVCV alone (<5 × 10⁴ TCID₅₀ per ml) induced a weak activation (2 ∼ 7.8-fold increase versus control), comparable to that by transfection of LGP2 alone (4 ∼ 10-fold increase versus control), and an additional activation was detected in LGP2-overexpressed cells followed by SVCV infection. High titers of SVCV alone (5 × 10⁴ TCID₅₀ per ml or 1 × 10⁵ TCID₅₀ per ml) stimulated a relatively strong DrIFN-φ1 promoter activity (15.9- and 16.8-fold increase versus control) and, surprisingly, overexpression of 200 ng of LGP2 resulted in an attenuated promoter activation (8.3- and 8.6-fold increase versus control), by an almost 50% reduction. However, overexpression of 50 or 100 ng of LGP2 did not cause an inhibitory effect (Fig. 10A, left panel). Regarding DrIFN-φ3 promoter, the inhibition did not happen, whereas a robust activation was induced by infection of SVCV at high titers (11 ∼ 12.2-fold increase for 5 × 10⁴ TCID₅₀ per ml or 1 × 10⁵ TCID₅₀ per ml) (Fig. 10A, right panel).

A time-course analysis of SVCV infection showed a time-dependent induction of two zebrafish IFN promoter activations by SVCV alone (Fig. 10B). At 36 h postinfection, overexpression of 100 and 200 ng of LGP2 significantly blocked the activation of DrIFN-φ1 promoter by a high titer of SVCV infection (1 × 10⁵ TCID₅₀ per ml) (nearly 45 and 49% reduction, respectively), but not by low titers of SVCV infection (1 × 10³ TCID₅₀ per ml); however, this blockade did not happen at the early phase of SVCV infection (12 and 24 h postinfection) and not for low doses of LGP2 (50 ng) at any time points post–viral infection (Fig. 10B). In another experiment, overexpression of 100 ng of LGP2 significantly impeded the activation of DrIFN-φ1 promoter by SVCV of 5 × 10⁶ TCID₅₀ per ml (∼35% reduction) but not by GCRV of 5 × 10⁶ TCID₅₀ per ml, and no obvious inhibition was detected by transfection of LGP2v1 and LGP2v2 at the same amount (100 ng) (Fig. 10C). Luciferase assays showed that overexpression of the dominant negative mutant of MDA5 (MDA5-DN) attenuated SVCV and GCRV-induced activation of DrIFN-φ1 promoters, and overexpression of MDA5-DN and RIG-I–DN together resulted in much profound inhibition, indicating that SVCV and GCRV activated an IFN response through RLR signaling (Fig. 10D). These results indicate that similar to high concentrations of poly(I:C), high titers of SVCV-mediated signaling were easily inhibited by LGP2, with a much stronger inhibitory effect from high amounts of LGP2.

Continuous titering experiments were next used to determine the combined roles of three LGP2 isoforms in IFN antiviral response. In the absence of poly(I:C), overexpression of LGP2v1 or LGPv2 stably blocked LGP2-mediated activation of DrIFN-φ1 promoter in a dose-dependent manner, but resulted in a compromised activation of DrIFN-φ3 promoter at large amounts (Fig. 11A). The strongest inhibition generally happened when low amounts of LGP2 were transfected, showing a best stimulatory potential, and when two variants were expressed at high levels, displaying the most robust inhibitory effect (Fig. 11A). The function relationship of three LGP2 splicing forms was the same when a low concentration (4 ng/ml) of poly(I:C) was transfected, because at this dose poly(I:C) had a less stimulatory effect than LGP2 (Fig. 11B).

However, when 1 μg/ml of poly(I:C) was used to transfect cells, which showed a better stimulatory potential than any a dose of LGP2 alone, overexpression of LGP2 significantly blocked poly(I:C)-induced IFN promoter activation in a dose-dependent manner. Cotransfection of either LGP2v1 or LGPv2 did not make an additional inhibition, except when they were transfected at large amounts (200 ng) (Fig. 11C). In another experiment by transfection of three LGP2 splicing forms at a high dose (200 ng) and poly(I:C) at a high concentration (1 μg/ml), LGP2 exhibited a strong inhibitory effect on two zebrafish IFN promoter activations by poly(I:C), and LGP2v1 and LGP2v2 displayed a weak one. Moreover, nearly no additional inhibition was observed when both were expressed together (Fig. 11D). These results indicate that LGP2v1 and LGP2v2 have an ability to block the IFN response induced by LGP2 alone, but do not help it make a more severe inhibition on poly(I:C)-mediated IFN signaling.

In this study, we identify three zebrafish LGP2 splicing transcripts from a single LGP2 gene. They are differentially expressed in resting cells and upregulated in zebrafish tissues infected with SVCV. Among them, full-length LGP2 is the most abundant in the infected tissues, but at the early time of infection or in certain resting tissues, (e.g., gill, spleen), LGP2v1 mRNA appears to be the most abundant (Fig. 2), indicating their differential roles in host immune response. Actually, full-length LGP2 not only has the potential to activate IFN antiviral response but also to block poly(I:C)- and SVCV-triggered IFN signaling under given conditions; however, the two shorter splicing variants LGP2v1 and LGP2v2 lose their immunoactive properties but retain an inhibitory role in RLR signaling. Accordingly, the three zebrafish LGP2 isoforms differentially contribute to IFN antiviral response.

The data presented in this study may solve the puzzle of the controversial function of fish LGP2. Most studies have suggested a positive role of fish LGP2 by ectopic expression of LGP2 in fish cells followed by virus infection (28, 31, 41, 42). However, our previous results have found that overexpression of crucian carp LGP2 or its RD domain significantly blocks the activation of fish IFN promoter by cytosolic poly(I:C) or by RIG-I and MDA5, respectively (27). In the above-mentioned studies, fish LGP2 is expressed from expression vectors; therefore, the expression level of LGP2 is high, as is the transfected poly(I:C) (27, 28, 31–33, 41, 42). In the current study, we can replicate the positive role of the full-length LGP2 in the absence of poly(I:C) and the negative one in cells transfected with poly(I:C) at large concentrations (Figs. 3, 6), which is subsequently verified by comprehensive analyses of gene expression, signaling transduction, and antiviral effects (Figs. 3–8). Moreover, when fish cells are transfected with poly(I:C) at limited concentrations, zebrafish LGP2 is still a potent stimulator of IFN signaling (Fig. 9). These results indicate that zebrafish LGP2 is actually a bifunctional protein and might play opposing functions under different conditions. Notably, zebrafish LGP2 exerts its positive function not in a dose-dependent fashion, because the most stimulating potential of LGP2 is observed at low levels rather than at high levels (Fig. 3). On the contrary, poly(I:C) stimulates IFN expression dependent of its concentration (25, 26) (Fig. 9). When the poly(I:C) concentration reaches a certain point, which results in much more IFN expression than zebrafish LGP2 alone, the functional role of zebrafish LGP2 is switched from being a positive regulator to a negative one (Fig. 9). Thus, the stimulatory potential of cytosolic dsRNA, closely related to its concentration in cells, is crucial for the determination of zebrafish LGP2’s function in a positive or negative role.

This notion is further verified by overexpression of LGP2 in fish cells followed by infection with SVCV, a negative-stranded RNA virus (40). Similar to the transfected poly(I:C) (27, 30), SVCV infection stimulates IFN response through both RIG-I and MDA5 (Fig. 10D). The activation of fish IFN promoters by high titers of SVCV infection is easily inhibited by high doses of LGP2 (Fig. 10A), because the inhibitory effect of LGP2, contrary to its stimulatory potential, is proportional to its expression levels (Fig. 6C). Moreover, time-course analyses showed that in the early phase of virus infection, SVCV is a poor inducer due to its low replication levels and, concomitantly, zebrafish LGP2 functions as a positive regulator because together with SVCV, it induces an additional activation of fish IFN promoters; however, in the late phase of SVCV infection, the inhibition occurs when SVCV replicates in cells up to a relatively high titer that causes a far more significant induction of IFN expression than zebrafish LGP2 alone (Fig. 10B). These results mean that the initial role of zebrafish LGP2 in vivo should be positive at the beginning of SVCV infection, and its negative effect may only occur in the late phase.

Therefore, whether zebrafish LGP2 exerts a positive or a negative role depends on SVCV titers, exactly on the balance of immunostimulatory potential between LGP2 and SVCV at this moment. In mammals, several studies have suggested that the cellular LGP2 expression level is a key factor for regulation of a switch between positive and negative roles in RLR signal transduction (15, 16, 18, 19). This conclusion is derived from function analysis of LGP2 on MDA5-mediated IFN signaling in the presence of poly(I:C), particularly at limited levels (16, 18, 19). Notably, these studies have also shown that human LGP2 significantly promotes IFN-β promoter activation in the presence of low concentrations of poly(I:C) (16), but inhibits promoter activation in the presence of large concentrations of poly(I:C) (9), which is consistent with the function analysis of zebrafish LGP2 in the current study (Fig. 9). Considering the conserved structure of vertebrate LGP2, its function is likely conserved in fish and mammals. If this is true in mice, it is easy to understand why LGP2-deficient mice are more susceptible to virus infection (12, 13), because they have lost the initially positive regulation of LGP2 in the early phase of virus infection. Our results might provide a basis for further study of mammalian LGP2 function in vitro.

Despite the bilateral function of zebrafish LGP2 on SVCV infection, we cannot exclude other possible mechanisms of fish LGP2 against different viruses. For example, human LGP2 selectively downregulates IFN response by seasonal influenza A viruses that can activate IRF3 and IFN transcription (43). Interestingly, solid evidence has shown that grouper LGP2 and grass carp LGP2 negatively regulate IFN response against virus infection in vitro (32, 33). However, it should be pointed out that in the current study overexpression of zebrafish LGP2-myc, a tag-fused LGP2 plasmid, cannot potentiate IFN promoter activation, but its inhibitory function retains intact (Fig. 9D). The same happens to the MITA gene, as evidenced by the findings that tag-fused fish and mouse MITA proteins lose their antiviral activity but display a dominant negative effect (27, 44). In addition, although GCRV, a dsRNA virus (45), induces IFN response through RIG-I and MDA5 (Fig. 10D), we failed to detect an obvious inhibitory effect of zebrafish LGP2 on IFN promoter activation by GCRV (Fig. 10C). This is probably not the case, because in some separate assays we have indeed observed the inhibitory effect of zebrafish LGP2 (data not shown). A reasonable explanation is that it is easy to titer poly(I:C) to handle its immunostimulatory levels, but due to the complex interaction between fish cells and a given virus, it is hard to test an appropriate point of virus titer or infection duration, at which zebrafish LGP2 always plays a negative role. The same is seen for LGP2v1 and LGP2v2 on activation both fish IFN promoters under SVCV infection (Fig. 10C) and for LGP2 on DrIFN-φ3 promoter activation by SVCV (Fig. 10A). Considering the differential sensitivity of DrIFN-φ1 and DrIFN-φ3 promoters to LGP2 and a relatively poor inhibitory effect of LGP2v1 and LGP2v2 (Figs. 9, 10), this inhibition would occur if the expression levels of LGP2, LGP2v1, and LGP2v2 were very high, although it is hard to operate due to limited capacity of transfection. If this is true, the in vivo inhibitory effects of LGP2v1 and LGP2v2 may not be relevant due to their weak expression levels under virus infection.

The unique structure of LGP2 proteins, which lack N-terminal CARD domains, makes it difficult to understand how they trigger IFN antiviral response as RIG-I and MDA5 do. In mammals, whereas initial experiments do not reveal an ability of LGP2 to activate IFNβ promoter and inhibit virus replication (7, 8), subsequent studies have obtained the opposite results (12, 46). In these cases, LGP2-directed ISRE stimulation in 293T cells might be attributable to its association with endogenous RLRs that contain CARDs (46). Direct evidence shows that LGP2 enhancement of poly(I:C) signaling is dependent on MDA5 (16). Interestingly, in vitro siRNA-based depletion of PUM1, a suppressor of LGP2 expression, directly upregulates LGP2, which in turn induces a host cell antiviral state through the accumulation of mRNAs encoding IFN-β and some ISGs including RIG-I, MDA5, and MITA (47). This result indicates that LGP2 is a master activator of innate immunity genes under certain conditions (47). The effect of depletion of PUM1 is similar to that of overexpression of zebrafish LGP2 alone in the current study, which results in an upsurge of IFN and ISGs, therefore conferring protection on EPC cells against SVCV infection (Fig. 5). Mechanistically, zebrafish LGP2 induces an IFN response through MAVS signaling and MITA signaling, which depends upon LGP2-enhanced phosphorylation of IRF3 and IRF3/7 binding to IFN promoters (Fig. 4). To ensure the accuracy of the results, we use several cell lines to replicate experiments and similar results are obtained, e.g., for IFN promoter activation in CAB and EPC, for upregulation of IFN and ISGs in EPC and ZFL, and for enhanced phosphorylation of IRF3, and enhanced binding of IRF3/7 to IFN promoters in EPC, ZFL, and ZF4.

Intriguingly, zebrafish LGP2 negatively regulates a poly(I:C)-triggered IFN response through the blockade of important signaling factors including RIG-I, MDA5, MAVS, MITA, and TBK1, but not of IRF3/7 (Fig. 6). In mammals, three hypotheses have been addressed to interpret how mammalian LGP2 interacts with RIG-I and MDA5 to inhibit IFN signaling (3, 15, 20); moreover, LGP2 can associate with adaptor MAVS, thus competing with the downstream kinase IKKi/ε and inhibiting IRF3 activation (9). Considering the cytosolic localization of fish LGP2 (32), it is easy to understand that zebrafish LGP2 cannot inhibit IRF3/7-induced IFN expression (Fig. 6D). A puzzling question is how LGP2 utilizes a set of RLR signaling molecules to exert a disparate function in the different phases of virus infection. Similar to the above-described mechanism for function switch of zebrafish LGP2 on transfected poly(I:C)- or SVCV-triggered IFN signaling, it is likely that the balance of stimulatory potential between zebrafish LGP2 and a given signaling factor in a specific cell state ingeniously determines which role zebrafish LGP2 has to play.

In summary, the data in the current study seemingly resolve the puzzles in fish LGP2. It is likely that zebrafish LGP2 is expressed at low levels in the beginning of virus infection, thus functioning as a positive regulator of IFN signaling, but in the late phase of virus infection it switches to a negative role. At low expression levels, zebrafish LGP2 has more stimulatory potential than MDA5 and RIG-I (Fig. 5D, 5E); therefore, LGP2 is likely a master activator of IFN innate antiviral response in the early time of virus infection, because at this time, zebrafish RIG-I and MDA5 are also expressed at low levels (Fig. 2E) and have a weaker stimulatory potential than LGP2 (Fig. 5E). Nevertheless, the maximum stimulatory effect of LGP2 is less than that of MDA5 and RIG-I along with the increase of their expression levels, and in addition to the role of LGP2 in IFN response, LGP2 has the potential to control CD8⁺ T cell survival and fitness against divergent RNA viruses (13). Therefore, the antiviral effect of LGP2 observed in knockout mice should be attributable to its multiple immunoactive properties in innate immunity and adaptive immunity.

The authors have no financial conflicts of interest.