Teleost-Specific TLR19 Localizes to Endosome, Recognizes dsRNA, Recruits TRIF, Triggers both IFN and NF-κB Pathways, and Protects Cells from Grass Carp Reovirus Infection

Toll-like receptors play essential roles in both the innate immune system and the adaptive immune system by recognizing pathogen-associated molecular patterns (PAMPs) that are the conserved microbial components (1, 2). Those microbial molecules range from bacterial cell surface components to viral genomes (3), including LPS, peptidoglycan (PGN), lipoteichoic acid, mannan, glucan, and nucleic acid (1, 4). TLRs are type I transmembrane proteins, comprising leucine-rich repeat motifs (LRRs) in extracellular region, a transmembrane region, and an intracellular Toll/IL-1 receptor (TIR) domain (5, 6). The LRR domain involves in pathogen recognition (7), whereas the cytoplasmatic TIR domain engages in the downstream signaling pathway(s) as well as in the localization of TLRs (8). Upon stimulation with PAMPs, TLRs initiate signal transduction pathways via the following adaptor molecules: myeloid differentiation primary response gene 88 (MyD88), TIR domain–containing adaptor-inducing IFN-β (TRIF), TIR domain–containing adaptor protein (TIRAP), and TRIF-related adaptor molecule (TRAM). This is followed by activation of a wide range of inducible transcriptional factors such as IFN regulatory factors (IRFs), NF-κB, and AP-1, leading to production of IFNs, inflammatory cytokines, chemokines, and antimicrobial peptides. These activities cause activation of macrophages, recruitment of neutrophils, and induction of IFN-stimulated genes, leading to direct killing of the invading pathogens, and triggering adaptive immunity.

dsRNA, as a pivotal viral PAMP, not only exists in dsRNA viruses, but also generates during viral infection as a replication intermediate for ssRNA viruses or as a by-product of symmetrical transcription in DNA viruses. dsRNA in cell surface or endosome formed by endocytosing viruses is recognized by TLR3, a recruiting TRIF as adaptor molecule in mammals (9). The TIR domain of TRIF binds to the TIR domain of TLR3, which is thought to be the vital event that dictates the downstream signaling process. Upon binding TLR3, TRIF undergoes oligomerization and recruits a signaling complex involving TANK binding kinase 1 (TBK1) and atypical inhibitor of κB kinase ε (IKKε) via TNF receptor-associated factor 3 (TRAF3) for IRF3 phosphorylation and activation (9, 10). Meanwhile, TRIF also recruits TRAF6 and activates TGF-β–activated kinase 1 for NF-κB activation. dsRNA in cytoplasm is sensed by retinoic acid–inducible gene I (RIG-I) and melanoma differentiation-associated protein 5 (11, 12), recruiting adaptor mitochondrial antiviral signaling protein, and activating the downstream molecules mediator of IRF3 activation and TBK1 (13, 14). Then, IRF3 and IRF7 are phosphorylated, followed by transcription of IFNs and NF-κBs (15, 16). In poikilothermal vertebrates, the second TLR member recognizing dsRNA, TLR22, localizes on the cell surface, recognizes extracellular dsRNA, and induces IFN responses to acquire resistance to virus infection via adaptor TRIF (17).

After TLRs recognize corresponding PAMPs, signaling pathways are initiated. Adaptors are recruited. To date, six adaptors of TLRs have been identified in mammals: MyD88, TRIF, TIRAP, TRAM, sterile α and armadillo motif-containing protein 1 (SARM1), and B cell adaptor for phosphoinositide 3-kinase (BCAP), although TRAM is absent in teleost genomes (18, 19). In mammals, TLR7/TLR8, TLR9, TLR11, TLR13, and TLR2/TLR10 recruit MyD88 directly, whereas TLR1, TLR4, TLR5, and TLR6 use TIRAP as the bridging adaptor to recruit MyD88. TLR3 recruits TRIF directly, whereas TLR4 indirectly recruits TRIF via the bridging adaptor TRAM (6, 20). TLR4 is the only receptor employing TIRAP, MyD88, TRAM, and TRIF. In contrast to these four TLR adaptors, the fifth adaptor, SARM1, and the sixth adaptor, BCAP, are the only adaptor proteins for which a negative role in TLR signaling has been proposed (19).

Viral dsRNA triggers the production of IFNs and NF-κBs. As key components in the innate immune responses and host defense against various pathogens, IFNs exhibit various biological functions, including antiviral activity, antitumor activity, and immunomodulatory effects (6). In mammals, three types of IFNs have been identified according to the receptor complex, namely type I IFN (IFN-I), type II IFN (IFN-II), and type III IFN (IFN-III), respectively (21, 22). In fishes, IFNs just possess IFN-I and IFN-II, and IFN-I is further categorized as group I and II (23, 24). In zebrafish and grass carp, group I contains IFN1 and IFN4, and group II contains IFN2 and IFN3, and IFN-II includes two members: IFN-γ1 and IFN-γ2. IFN-I and IFN-II show different abilities to induce the downstream gene expressions, which are responsible for the expressions of antiviral genes and immune regulatory genes, respectively (23, 25). It is also well established that NF-κB classical signaling pathway plays a crucial role in innate immunity and its activation could induce proinflammatory cytokines in response to infection or injury (26).

So far, at least 13 and 21 TLRs have been identified in mammals and fishes, respectively. Among them, TLR18 to TLR20 and TLR23 to TLR28 are considered to be fish-specific TLRs (27–31). Up to now, TLR19 sequence has been identified in several teleosts such as Danio rerio (32), Ictalurus punctatus (33, 34), Salmo salar (35), Megalobrama amblycephala (36), and Ctenopharyngodon idella (31). All the studies focus on structural analysis and mRNA expression characteristics. However, ligand, adaptor, and signaling pathways for and functions of TLR19 remain unknown. Grass carp (C. idella) is an important aquaculture species in China, but tremendous economic losses are often caused by grass carp reovirus (GCRV) infection. Our previous report demonstrated that mRNA expression levels of grass carp TLR19 were obviously higher in a resistant population to GCRV infection than in susceptible grass carp (37), which implies that TLR19 is a vital molecule in antiviral responses. In the current study, overexpression and knockdown experiments confirmed that TLR19 possesses the antiviral function. Further studies found that TLR19 resides in endosome, recognizes dsRNA, and recruits TRIF to initiate downstream signaling pathways. This is the first attempt to identify the third TLR member recognizing dsRNA in teleost. Better understanding of the immune defense mechanisms contributes to the development of management strategies for reducing diseases in fishes.

C. idella kidney (CIK) cells were provided by the China Center for Type Culture Collection. Cells were grown in DMEM (Life Technologies) supplemented with 10% FBS (Life Technologies), 100 U/ml penicillin and 100 U/ml streptomycin, and maintained at 28°C in a humidified atmosphere of 5% CO₂ incubator (Thermo Scientific). PGN, ultrapure LPS (L4391), polyinosinic:polycytidylic acid (poly(I:C)), and isopropyl-d-1-thi-ogalactopyranoside (IPTG) were purchased from Sigma-Aldrich. dsDNA was prepared and purified from a cDNA template of CIK cells. Hoechst 33342 and ER-Tracker Red dye were from AAT Bioquest and Invitrogen, respectively. FuGENE 6 transfection reagent was purchased from Promega. All the restriction enzymes were purchased from Thermo Scientific. All the primer synthesis and DNA sequencing were carried out in AuGCT Biotechnology, Wuhan, China. We ensured that the experiments followed the ethical guidelines of Huazhong Agricultural University and confirmed that all experimental protocols were approved by Huazhong Agricultural University.

pDsRed1-C1, pCMV-enhanced GFP (-EGFP), and pCMV-EGFP-CMV-SV40 were employed as original plasmids for the constructions of expression vectors. For the subcellular localization studies, the full-length open reading frames of TLR19, ras related in brain 5 (RAB5), RAB7, MyD88, TRIF, TIRAP, and SARM1s2 were amplified with corresponding primers (Supplemental Table I) and were digested with restriction enzymes (EcoRI/KpnI for TLR19; KpnI/BamHI for RAB5, RAB7, MyD88, TRIF, TIRAP, and SARM1s2). Next, TLR19 was ligated into pCMV-EGFP, digested by the corresponding enzyme to construct pTLR19-EGFP fusion vector. The rest of the fragments were ligated into pDsRed1-C1 to construct pDsRed1-C1-RAB5, pDsRed1-C1-RAB7, pDsRed1-C1-MyD88, pDsRed1-C1-TRIF, pDsRed1-C1-TIRAP, and pDsRed1-C1-SARM1s2 fusion vectors, respectively. With the same method, TLR19-HA, MyD88-Flag, TRIF-Flag, SARM1s2-Flag, and SARM1-Flag were ligated into pCMV-EGFP-CMV-SV40 to construct overexpression vectors. For dual-luciferase reporter assays the valid promoters (TBK1, IRF3, IRF7, IFN1, IFN2, IFN3, IFN4, IFN-γ1, IFN-γ2, NF-κB1, and NF-κB2, respectively) were cloned into pGL3-basic luciferase reporter vector (Promega), which had been previously constructed in our laboratory (16, 38).

To knockdown the expression of TLR19, RNA interference assay was performed by transfecting siRNA targeting TLR19 mRNA. Three siRNA sequences (s1: 5′-UUGAGUUUCUUAUAGUGCC-3′, s2: 5′-UCUAGUGAGGUUAAGGACC-3′, s3: 5′-UUGAAGAUGAAGGUUGGAG-3′) were designed and synthesized by RiboBio. The silencing effects of the three TLR19 siRNA candidates were evaluated by real-time quantitative RT-PCR (qRT-PCR) as well as a negative control siRNA provided by the supplier.

All the vectors and siRNAs were transfected into CIK cells by FuGENE 6 transfection reagent (Promega) according to the manufacturer’s instructions.

For the acquisition of anti-VP56 polyclonal antiserum, VP56 gene (GenBank accession number KU161132; https://www.ncbi.nlm.nih.gov/nuccore/KU161132) of GCRV was amplified with corresponding primers (Supplemental Table I) and was cloned into pGEX-4T1 vector (Novagen). The plasmid pGEX-4T1-VP56 was transformed into the Escherichia coli BL21 cells (Novagen) for prokaryotic expression. The fusion protein was induced by isopropyl β-D-1-thiogalactopyranoside and purified by GST-Bind Resin Chromatography (Novagen). The purified protein was applied to immunize BALB/c mice to acquire the polyclonal anti-VP56 antiserum according to a previous report (39). The specificity was tested by Western blotting (WB) assay and preabsorption experiment according to the previous report (38). Anti-IRF3 rabbit polyclonal antiserum was previously prepared and presented by Prof. Yibing Zhang, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, China (40). The anti-IRF7 rabbit polyclonal antiserum was previously prepared in our laboratory (38). Anti-Flag tag (ab45766), anti-HA tag (ab18181), anti-His tag (ab18184) mouse mAbs, and anti–β-tubulin primary rabbit polyclonal Ab (ab6046) were purchased from Abcam. IRDye 800CW donkey anti-rabbit IgG (H + L) (P/N 925-32213) and anti-mouse IgG (H + L) (P/N 925-32212) secondary Abs were purchased from LI-COR. Goat-anti-mouse Ig-HRP conjugate secondary Ab (A0216) was purchased from Beyotime.

Total RNAs were extracted using RNAiso Plus (Takara, Japan) and first-strand cDNAs were synthesized using random hexamer primers with reverse transcription kit (Invitrogen). The mRNA expressions of target genes were quantified using SYBR Premix Ex Taq II reagent (Takara) and a LightCycler 480 II real-time PCR system (Roche, Switzerland). Primers are listed in Supplemental Table II. The mRNA expression levels were normalized to the expression level of EF1α, and the data were analyzed using the 2^−ΔΔCT method as described previously (23).

CIK cells were seeded in 24-well plates at a density of 5 × 10⁵ cells/ml for 24 h. Then cotransfection was performed with the overexpression plasmid, target promoter luciferase plasmid, and pRL-TK (internal control reporter vector). After 24 h transfection, cells were stimulated with GCRV or poly(I:C). After 24 h incubation, cells were washed with PBS, lysed with passive lysis buffer (Promega), and assayed for luciferase activities in a luminometer by the Dual-Luciferase Reporter Assay System (Promega). The luciferase reading of each sample was first normalized against those in the pRL-TK levels, and the relative light unit intensity was presented as the ratio of firefly luciferase to renilla luciferase. All the experiments were performed in triplicate and repeated at least three times.

Samples (1 × 10⁶ cells per well) were infected with GCRV 097 strain (multiplicity of infection = 1). After 24 h, supernatants were serially diluted in 10-fold and incubated with CIK cells in a flat 96-well plate to determine the 50% tissue culture infective dose. Cells were incubated at 28°C for 7 d. On day 7, the plates were examined for the presence of viral cytopathic effect under the microscope.

For MTT assay, cells that overexpressed, knockdown TLR19 or vector, were severally seeded in 96-well plates (1 × 10⁵ cells per well) overnight, then infected with GCRV. At the scheduled time, 20 μl of MTT (5 mg/ml in PBS) was added to each well. After 4 h incubation at 28°C, DMSO (100 μl per well) was added at 28°C for 10 min. The OD₄₉₀ was measured by a microplate reader (Infinite F200; Tecan). Data were expressed as a viability index, which was the ratio of the mean OD value of quartic wells at a corresponding time point to that at 0 h postinfection.

To determine whether TLR19 binds the downstream adaptor, CIK cells in 10-cm² dishes were cotransfected with the indicated plasmids for 24 h. The cells were lysed in immunoprecipitation lysis buffer (20 mM Tris [pH 7.4], 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM Na₃VO₄, 0.5 μg/ml leupeptin, 2.5 mM sodium pyrophosphate) (Beyotime) added with 1 mM PMSF for 30 min on ice, and the cellular debris was removed by centrifugation at 12,000 × g for 30 min at 4°C. The supernatant was transferred to a fresh tube and incubated with 1 μg of Ab with gentle shaking overnight at 4°C. Protein A+G Sepharose beads (30 μl) (Beyotime) were added to the mixture and incubated for 2 h at 4°C. After centrifugation at 3000 × g for 5 min, the beads were collected and washed four times with lysis buffer. Subsequently, the beads were suspended in 20 μl 2 × SDS loading buffer and denatured at 95°C for 10 min, followed by WB detection.

For WB analysis, protein extracts were separated by 8–12% SDS-PAGE gels and transferred onto nitrocellulose membranes (Millipore). The membranes were blocked in fresh 3% nonfat dry milk dissolved in TBST buffer at 4°C overnight, then incubated with appropriate primary Abs for 2 h at room temperature: anti-HA (monoclonal, 1:1000), anti-Flag (monoclonal, 1:1000), anti–β-tubulin (polyclonal, 1:5000), anti-IRF3 (polyclonal, 1:1000), anti-IRF7 (polyclonal, 1:1000), and anti-VP56 (polyclonal, 1:1000), respectively. They were then washed thrice with TBST buffer and incubated with secondary Ab for 1 h at room temperature. After washing three times with TBST buffer, the nitrocellulose membranes were scanned and imaged by an Odyssey CLx Imaging System (LI-COR). The results were obtained from three independent experiments.

The fragment encoding the extracellular region (LRR domain) of TLR19 was obtained by PCR amplification with primers listed in Supplemental Table I. Afterwards, the purified PCR product was digested with restriction enzymes KpnI and BamHI, and then ligated into the pET-32a (+) expression vector. Recombinant protein was expressed in E. coli BL21 (DE3) cells, denatured in 8 M urea and purified using Ni-NTA beads.

For PAMP-binding assays, the activities of LRR protein binding PAMPs (LPS, PGN, dsDNA, and poly(I:C)) were measured by ELISA method as previously described and revised by ourselves (41). Briefly, 96-well microtiter plates (Immulon 2H; Dynex Technologies, Ashford, U.K.) were coated with PAMPs (10 μg per well), washed with TBST and blocked with 1% BSA. One hundred microliters of LRR proteins with different concentrations were added to the corresponding wells. The protein (rVector) expressed by empty vector was employed as control. After incubation at 18°C for 3 h, the plates were washed three times with PBS with Tween 20, and 100 μl of anti-His (monoclonal, 1:1000) was added as the first Ab. After incubation at 37°C for 1 h, the plates were washed again and incubated with 100 μl of goat-anti-mouse Ig-HRP conjugate (polyclonal, 1:3000) as the second Ab for 1 h. The plates were washed four times with TBST for 5 min, and 100 μl of 3,3′,5,5′-tetramethylbenzidine solution (PA107; Tiangen, China) was added to each well, and then incubated at room temperature in the dark for 15 min. The reaction was stopped by adding 50 μl of 2 M H₂SO₄. The absorbance was measured on Synergy 4 Hybrid Microplate Reader (BioTek, Winooski, VT) at 450 nm. The wells filled with 100 μl TBS were used as negative control. The examination was performed by three independent experiments.

For the subcellular localization of TLR19 in resting state, cells transiently transfected with TLR19-EGFP were plated on microscopic coverglasses in 12-well plates overnight to achieve 50% confluency. The following day, cells were stained with ER-Tracker dye. Briefly, the cells were rinsed with HBSS and added ER-Tracker staining solution, and incubated for ∼15–30 min at 37°C. For the subcellular location of TLR19 post-poly(I:C) stimulation and downstream adaptors, transiently cotransfected cells were plated on microscopic coverglasses. These cells did not stain by ER-Tracker dye. All the cells on microscopic coverglasses were fixed using 4% formaldehyde for 2 min at 37°C before incubation in 1 μg/ml Hoechst 33342 for 10 min in the dark to stain the nuclear; afterward, the stained cells were rinsed with PBS. Images were taken with an UltraVIEW VoX 3D Live Cell Imaging System (PerkinElmer).

Statistical analysis and presentation graphics were carried out using the SPSS 10.0 and GraphPad Prism 6.0 software. Results were presented as mean ± SD for at least three independent experiments. All data were subjected to one-way ANOVA, followed by an unpaired, two-tailed t test. A p value <0.05 was considered to be a statistically significant difference (*p < 0.05, **p < 0.01, ***p < 0.001).

The open reading frame of grass carp TLR19 encodes a protein of 957 aa, with a molecular mass of 106.3 kDa and an isoelectric point of 5.67 theoretically. TLR19 protein contains nine LRRs, a transmembrane region, and a TIR domain. The GCRV-VP56 gene encodes a protein of 512 aa, with a predicted m.w. of 56.6 kDa and an isoelectric point of 5.05. The specificity of the polyclonal anti-VP56 antiserum was verified and is shown in Supplemental Fig. 1.

To investigate the role of TLR19 in host defense against GCRV infection, mRNA expression levels of TLR19 were measured, which were significantly upregulated at 48 h post-GCRV infection (Fig. 1A), implying that TLR19 took part in antiviral immune responses. To clarify positive or negative regulatory mechanisms of TLR19 in GCRV infection, knockdown and overexpression of TLR19 were performed. Among the three candidate siRNA sequences, s3 showed the highest interference efficiency (Fig. 1B). The mRNA and protein levels of TLR19 were markedly overexpressed (Fig. 1C). GCRV was added to cell culture medium after overexpression or knockdown of TLR19. The results showed that mRNA and protein levels of VP56 (minor outer capsid) of GCRV decreased markedly in the overexpression group (Fig. 1D). Meanwhile, VP56 expressions in the knockdown group were distinctly higher than those in the control group (Fig. 1E). GCRV titer was significantly reduced in TLR19 overexpressing cells and markedly increased in TLR19 knockdown cells (Fig. 1F). These results indicated that TLR19 played a positive role in anti-GCRV immune responses.

To further confirm the result, flow cytometry and MTT assay were conducted. Flow cytometry could differentiate live cells from virions and cell debris, and more than ten thousand cells were analyzed in each group. The results demonstrated that overexpression of TLR19 reduced cell mortality rate (Fig. 2A–C), and knockdown of TLR19 increased cell mortality rate (Fig. 2D–F). In MTT assay, cell proliferation was observed in TLR19 overexpressing cells, and cell death in TLR19 knockdown was more serious than that in control post-GCRV infection (Fig. 2G). These results indicated that TLR19 restricted the replication of GCRV and protected cells from GCRV infection.

IFNs and NF-κBs are recognized as important molecules involving in virus-triggered innate immune responses. To identify the role of TLR19 in GCRV-mediated IFN and NF-κB induction, luciferase reporter assays were performed to examine the promoter activities of IFNs and NF-κBs upon TLR19 overexpression or knockdown. As shown in Fig. 3A and 3B, except for IFN2, IFN4, and IFN-γ1, the promoter activities of all the examined IFN and NF-κB genes were significantly increased in TLR19 overexpressed cells under mock condition, and this increase was more pronounced in the case of GCRV-infected condition. In addition, the promoter activities of TBK1, IRF3, and IRF7 were also examined, which may mediate IFN production. We found that the promoter activity of IRF3 but not IRF7 and TBK1 was markedly upregulated. To confirm these results, IRF3, IFN1, IFN-γ2, and NF-κB1, as representations for IRFs, IFNs, and NF-κBs respectively, were examined by luciferase reporter assays in TLR19 knockdown cells. As shown in Fig. 3C, knockdown of TLR19 remarkably reduced the basal activities of these gene promoters.

To further investigate the impact of TLR19 on IFN and NF-κB pathways in response to GCRV infection, TLR19 overexpression cells were infected with GCRV. qRT-PCR showed that TLR19 overexpression markedly increased the transcription of IFN1, IFN3, and NF-κBs (Fig. 4). Furthermore, we analyzed the expressions of TBK1, IRF3, and IRF7, and the results showed that IRF3 was significantly upregulated at 48 h post-GCRV infection. In addition, the expression levels of effector molecules of IFN and NF-κB pathways, including GCRV-induced gene 1 (Gig1) (42), myxovirus resistance 2 (Mx2) (43), TNF-α, and IL-1β were also examined. As shown in Fig. 5, TLR19 overexpression significantly induced the expressions of these genes under GCRV infection. All of these results indicated that TLR19 played a positive role in IFN and NF-κB pathways, which was mainly regulated by IRF3.

According to mRNA expression profiles of IRF3 and IRF7, we performed WB at 48 h poststimulation. There was a weak blotting band above the corresponding objective band, which has previously been confirmed as the phosphorylation form of IRF3 or IRF7 (16, 38). The results indicated that both protein and phosphorylation levels of IRF3 were boosted by TLR19; as for IRF7, TLR19 restrained phosphorylation and did not affect protein expression (Fig. 6).

To screen ligand(s) of TLR19, four representative PAMPs (LPS, PGN, poly(I:C), and dsDNA) were employed. First, CIK cells were stimulated and mRNA expressions were examined by qRT-PCR. As shown in Fig. 7A, mRNA expression of TLR19 gene was extremely significantly upregulated post-poly(I:C) stimulation, suggesting that TLR19 responds to poly(I:C). Second, recombinant protein of LRR domain was expressed in vitro and purified. As shown in Fig. 7B, molecular mass of recombinant LRR was ∼98 kDa by SDS-PAGE, whereas the remodeled pET-32a (+) vector (assigned as rVector) produced an obvious protein band of ∼26 kDa. The binding ability between LRR and PAMPs was examined by ELISA. In comparison, poly(I:C) had a strong binding effect with LRR, followed, in turn, by LPS and PGN. dsDNA did not exhibit any binding character (Fig. 7C). Third, luciferase reporter assays were performed to test these results. On the one hand, it has previously been reported that IRF3 overexpression significantly activated the promoter of IFN1 (38). Thus, it is reasonable that the activity of IFN1 promoter is chosen to confirm that TLR19 recognizes poly(I:C). On the other hand, LPS can provoke a strong inflammatory reaction in mammals (44), so the activity of NF-κB2 promoter is chosen to examine whether LPS can activate NF-κB2 promoter via TLR19. As shown in Fig. 7D and 7E, LPS, PGN, and dsDNA activated neither IFN1 nor NF-κB2 promoters. In contrast, poly(I:C) significantly promoted IFN1 and NF-κB2 promoters. All the results indicated TLR19 recognizes dsRNA analog.

The subcellular localization of TLR19 was predicted by online software (http://mu2py.biocomp.unibo.it/mempype/default/predict). The predicted results showed that TLR19 localized on internal membrane. For precise subcellular localization, we transfected CIK cells with TLR19-EGFP fusion vector, followed by staining with ER-Tracker Red dye. Next, localization was visualized under a confocal microscopy. As shown in Fig. 8A, TLR19 does not colocalize with endoplasmic reticulum (ER), implying that TLR19 is synthesized in ribosome not binding on ER. After that, TLR19 moves to early endosome (Fig. 8B, 8C).

TLR19 recognizes dsRNA. But what is the downstream adaptor(s) of TLR19? In grass carp, SARM1 has three splice variants (SARM1, SARM1s1, SARM1s2). SARM1 and SARM1s1 reside in mitochondria, whereas SARM1s2 distributes throughout the whole cell (45). Thus, MyD88, TRIF, TIRAP, and SARM1s2 possess the possibility of colocalization with TLR19. We constructed red fluorescent protein (RFP) fusion vectors to visualize their localizations in CIK cells by confocal microscope. The results clearly showed that TLR19 colocalizes with TRIF, not other adaptors (Fig. 9A). The result indicated that TRIF is the potential adaptor of TLR19. Generally, TLR adaptors are TIR domain-containing proteins that interact with TLRs. Although SARM1 resides in mitochondrion, it has TIR domain. To further test whether TLR19 utilizes TRIF as adaptor, HA-labeled TLR19 and Flag-labeled TRIF, MyD88, TIRAP, SARM1, and SARM1s2 vectors were constructed. Coimmunoprecipitations were carried out. The results verified the interaction between TLR19 and TRIF (Fig. 9C). No other interactions were detectable (Fig. 9B, 9D–F). Notably, overexpression of just the TIR domain of TLR19 is unreliable to study the interaction between TLR19 and adaptors (Supplemental Fig. 2). All the data above demonstrated that TRIF is recruited as adaptor of TLR19.

In mammals, TLR3 recognizes dsRNA and recruits TRIF to initiate downstream signaling, which has been verified in teleost (17). Both TLR3 and TLR19 localize to endosome. To explore the differential roles between TLR3 and TLR19, mRNA expression patterns of IFNs and NF-κBs in TLR3 overexpression cells (Fig. 10A) were investigated. We found that TLR3 significantly upregulated the expression of IFN1 (Fig. 10D) and barely changed the expression of IFN3 (Fig. 10E) post-GCRV infection. With regard to NF-κBs, TLR3 overexpression promoted the expression of NF-κB1, not NF-κB2, but it was independent of GCRV infection (Fig. 10F, 10G). Additionally, we also examined mRNA expressions of IRF3 and IRF7. We found that TLR3 overexpression increased the production of IRF3 post-GCRV infection (Fig. 10B). Although the expression of IRF7 was also upregulated, it did not rely on TLR3 (Figs. 10C, 11). Collectively, these data clearly illustrated that TLR19 induced stronger IFN and NF-κB responses than TLR3 in CIK cells post-GCRV infection.

TLRs have three different domains. The LRR domain recognizes different PAMPs. Previous studies showed that the intracellular TLRs, including TLR3, TLR7, TLR8, and TLR9, could detect nucleic acids (46). TLR3 depends on the electrostatic surface potentials of LRRs to recognize dsRNA, and the ascending lateral surface has obviously positively charged regions (47). Similar to TLR3, TLR19 is also strongly charged on the ascending lateral surface of LRRs (30), implying that TLR19 may bind to dsRNA in connection with the electrostatic potential of the ascending lateral surface. In the current study, dsRNA analog poly(I:C) upregulates TLR19 expression, which means that TLR19 responds to poly(I:C). LRR-PAMP–binding assay validates the interaction between LRRs and poly(I:C). In addition, poly(I:C) markedly activates IFN1 and NF-κB2 promoters in TLR19 overexpression cells. The results indicated that poly(I:C) is a ligand for TLR19.

So far, why do teleosts possess two intracellular dsRNA recognition receptors: TLR3 and TLR19? In antiviral responses, the roles of TLR3 appear to be dependent on the viral genome structure, entry route into the cells, viral affecting sites, and the host antiviral effector functions. For example, in TLR3-deficient (TLR3^−/−) mice, the immune response to reovirus was unaffected compared with wild-type mice; however, TLR3^−/− mice are hypersusceptible to murine cytomegalovirus infection (48). Cytokine (IFN-I, IL-12p40, and IFN-γ) production, and NK cell and NKT cell activation are impaired in TLR3-deficient mice (49). Fishes live in an aquatic ecosystem and are exposed to many kinds of RNA viruses (41, 50, 51). In the evolutionary process, fishes have developed a set of RNA-sensing TLR systems. In our study, in response to GCRV, TLR3 overexpression significantly induces the expression of IFN1, whereas TLR19 overexpression not only significantly induces IFN1 and IFN3, but also markedly upregulates the expression of NF-κBs. Compared with TLR3, TLR19 may exert a stronger role in inhibition of GCRV.

In mammals, TRIF alters the distribution profile in response to dsRNA; overexpressed TRIF is localized to the speckle-like structures in cytosol and colocalized with TLR3 under poly(I:C) stimulation (52). Accordingly, we observed that TRIF diffusely distributed in the cytosol in the CIK cell line (data not shown). Surprisingly, postcotransfection with TRIF-RFP and TLR19-EGFP plasmids, TRIF appears as speckle-like structures, concentrating together and colocalizing with TLR19 without any stimulation (Fig. 9A). Compared with other adaptors, TRIF harbors specific structural motifs in its N and C termini: N-terminal region is crucial for TRIF-mediated IRF3 activation, and C-terminal region involves in NF-κB activation (9). Our findings revealed that the expressions of IRF3 and NF-κBs are significantly upregulated upon TLR19 overexpression via TRIF. These results support that TLR19 recruits TRIF to initiate downstream signaling. We then confirmed that TLR19 interacts with TRIF through coimmunoprecipitation assays. Although the interaction between TLR19 and TRIF was verified, the interaction mode between them needs to be explored, as TLR3 recruits TRIF directly, whereas TLR4 recruits TRIF via the bridging adaptor TRAM (6, 20). Teleosts lose TRAM in genomes. Future studies can be considered by mass spectrometry method.

As important transcription factors, IRF3 and IRF7 had different expression profiles upon TLR19 overexpression. The expression of IRF3 but not IRF7 was significantly upregulated. The consequence was in line with the luciferase reporter assay. Additionally, TLR19 overexpression promoted protein and phosphorylation levels of IRF3, whereas it inhibited the phosphorylation of IRF7. Efficient and self-limiting innate immune response to viral infection relies on a tight balance between IRF3 and IRF7 (53). TLR3 also enhanced mRNA expression of IRF3, not IRF7. However, the RIG-I–like receptor pathway more depends on IRF7, not IRF3 (38, 54). The results showed that TLR and RIG-I–like receptor pathways cooperate and have different emphasis on IRF3/7 to defend viral infection.

As for IFNs and NF-κBs, IFN1 and IFN3 are the predominant IFN-Is in CIK cells (38), which are significantly induced upon TLR19 overexpression post-GCRV infection. The expression patterns in CIK cells are similar to those in CAB cells, in which IFN1 and IFN3 are ubiquitously produced and are inducible in response to viral infection (55, 56), indicating that IFN1 and IFN3 are involved in the antiviral response regulated by TLR19. mRNA expressions of IFN2 and IFN4 were markedly increased after GCRV infection, but the increase did not rely on TLR19. With regard to IFN-II, TLR19 overexpression significantly enhanced the promoter activity of IFN-γ2 under stimulation of GCRV. IFN-γ2 shares more structural similarities with higher vertebrate IFN-γ gene (57), which can inhibit viral replication directly, and most importantly possesses the immunostimulatory and immunomodulatory effects (58). The results indicated that TLR19 regulated the immune system by IFN-γ2. NF-κB is a major transcription factor that regulates genes responsible for both the innate and adaptive immune response (59). NF-κB controls many genes involved in inflammation. There are five proteins in the NF-κB family: NF-κB1, NF-κB2, RelA, RelB, c-Rel. Both NF-κB1 and NF-κB2 were significantly upregulated at promoter and mRNA levels mediated by TLR19, which meant that TLR19 facilitates inflammation. To confirm the activation of IFN and NF-κB pathways, dual representative effector genes in the two pathways were employed to examine mRNA expressions. The results demonstrated that TLR19 overexpression can significantly improve mRNA expressions of all the four genes post-GCRV infection. Collectively, TLR19 triggers both IFN and NF-κB pathways indeed.

In conclusion, teleost-specific TLR19 localizes to endosome and recognizes dsRNA analog with LRR domain. TLR19 recruits TRIF but not other adaptor molecules by TIR domain, enhances the protein and phosphorylation levels of IRF3, and inhibits phosphorylation of IRF7. TLR19 facilitates both IFN and NF-κB pathways; however, TLR3 localizing to endosome and recognizing dsRNA analog only promotes IFN and NF-κB1 expressions independent of GCRV infection (Figs. 10, 11). TLR19 plays a positive role in antiviral infection. This is the third TLR family member that can recognize dsRNA analog, which will serve the antiviral immune mechanisms in teleosts and evolutionary immunology. It will also provide ideas and methods to systematically explore signaling pathways and functions of species-specific TLRs.

We highly appreciate Prof. Yibing Zhang from Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, China, for kindly providing rabbit anti-IRF3 antiserum. We thank Yanqi Zhang, Hang Su, Xun Xiao, Jiacheng Zhang, Bo Liang, and Cong Li for helpful discussions and assistance in experiments.

The authors have no financial conflicts of interest.