Ring finger protein 166 potentiates RNA virus-induced interferon-β production via enhancing the ubiquitination of TRAF3 and TRAF6

Host cells orchestrate the production of IFN-β upon detecting invading viral pathogens. Here, we report that Ring finger protein 166 (RNF166) potentiates RNA virus-triggered IFN-β production. Overexpression of RNF166 rather than its homologous proteins RNF114, RNF125, and RNF138, enhanced Sendai virus (SeV)-induced activation of the IFN-β promoter. Knockdown of endogenous RNF166, but not other RNFs, inhibited the IFN-β production induced by SeV and encephalomyocarditis virus. RNF166 interacted with TRAF3 and TRAF6. SeV-induced ubiquitination of TRAF3 and TRAF6 was suppressed when endogenous RNF166 rather than RNF114/138 was knocked down. These findings suggest that RNF166 positively regulates RNA virus-triggered IFN-β production by enhancing the ubiquitination of TRAF3 and TRAF6.

while Ring-finger protein 125 (RNF125) and c-Cbl catalyze the K48-linked ubiquitination of RIG-I and negatively regulate RIG-I-mediated antiviral activity 23,24 . Ubiquitin carboxyl-terminal hydrolase CYLD, a de-ubiquitination enzyme, physically interacts with RIG-I and removes its K63-linked polyubiquitin chains to attenuate antiviral activity 25 . VISA polymers can also recruit ubiquitin ligase family members, multiple TRAFs, through different TRAF-binding motifs to promote K63-linked ubiquitination, thereby recruiting NEMO to the VISA complex, which turns on TBK1 and IKK, resulting in the activation of IRF3 and NF-κ B 13 . In addition, cIAP1/2 acts as a positive regulator by enhancing RNA virus-mediated K63-linked ubiquitination of TRAF3/6, while OTUB1/2 plays an opposite role via deubiquitinating TRAF3/6 26,27 .
In this report, we show that Ring-finger protein 166 (RNF166) potentiates RNA virus-induced IFN-β production via enhancing the ubiquitination of TRAF3 and TRAF6. These findings broaden our understanding of the mechanisms by which RLR signaling is positively regulated upon viral infection.

Results
RNF166 rather than its homologous proteins potentiates RNA virus-induced IFN-β production. RNF166 is closely related to RNF125, which has been reported to negatively regulate RIG-Imediated anti-RNA virus signaling by conjugating ubiquitin chains to RIG-I and leading to the degradation of RIG-I by the proteasome 23 . RNF125 and its homologous proteins RNF114, RNF138, and RNF166 form a subfamily of small C3HC4 RING ubiquitin ligases 28 , so we investigated whether RNF114/138/166 also play a role in RNA virus-induced IFN-β production. We transfected plasmids that encoded RNF114, RNF125, RNF138, and RNF166 into HEK293T cells to perform reporter assays. We found that overexpression of RNF166 but not it's homologous RNF114, 125, and 138, potentiated Sendai virus (SeV)-induced activation of the IFN-β promoter. However, RNF166 had no apparent effect on the overexpression of cGAS and the STING-induced activation of the IFN-β promoter (Fig. 1A), suggesting that RNF166 specifically enhances RNA but not DNA virus-induced IFN-β production. Overexpression of RNF166 can also enhance the transcription of Interferon-stimulated genes (ISGs) like ISG15 and MX1 (Fig. 1B).
To determine whether endogenous RNF166 is involved in anti-RNA virus signaling, we generated stable RNF166-knockdown cell pools using shRNA plasmids that targeted five sites on human RNF166 mRNA. Two shRNA plasmids (#4 and #5) markedly inhibited the expression of endogenous RNF166 mRNA in HEK293T cells, whereas the #1, #2, and #3 shRNA plasmids had little effect on RNF166 mRNA (Fig. 1C). We found that knockdown of RNF166 significantly inhibited the activation of the IFN-β promoter, transcription of IFN-β mRNA, and the secretion of IFN-β triggered by SeV infection (Fig. 1D,E). We also generated stable RNF114-, RNF125-, and RNF138-knockdown HEK293T cell pools to determine their functions (Fig. 1F). Consistent with a previous report that RNF125 acts as a negative regulator of the RIG-I-mediated signaling pathway 23 , knockdown of RNF125 slightly enhanced the SeV-elicited IFN-β production compared to the control cell pool, while knockdown of RNF114 clearly enhanced the SeV-induced IFN-β production, and knockdown of RNF138 had no appreciable effect (Fig. 1G).
We further generated stable RNF166-knockdown cell pools with HeLa cells and obtained similar results; the production of IFN-β induced by SeV, encephalomyocarditis virus (EMCV) and human influenza A virus infection, as well as poly (I:C)-transfection notably decreased when endogenous RNF166 expression was knocked down (Fig. 1H-K). However, knockdown of RNF166 had no effect on DNA analog-induced activation of IFN-β (Fig. 1L). These results suggested that RNF166 rather than its homologous proteins physiologically potentiates RNA virus-induced IFN-β production.
RNF166 targets TRAF3 and TRAF6 to potentiate VISA-mediated antiviral signaling. We next determined which molecules are targets of RNF166 in the anti-RNA virus signaling pathway. In reporter assays, overexpression of RNF166 potentiated VISA-, but not the downstream kinase TBK1-mediated activation of the IFN-β promoter ( Fig. 2A). Repoter Assay and Bioassay results showed that RNF166 but not the other RNFs potentiated VISA-mediated transcription and secretion of IFN-β , while the overexpression of RNF114 or RNF125 had inhibitory effects (Fig. 2B). Accordingly, knockdown of RNF166 inhibited VISA-but not TBK1-mediated activation of the IFN-β promoter (Fig. 2C). These data suggested that RNF166 acts on signaling components that are downstream of VISA and upstream of TBK1.
We next used co-immunoprecipitation to determine whether RNF166 interacts with VISA and its downstream components. Overexpressed RNF166 markedly associated with TRAF3 and TRAF6 in 293T cells, and also interacted with VISA, but had no detectable interactions with RIG-I, MDA5, TBK1, and IRF3 (Fig. 2D). RNF166 was subsequently shown to co-localize with TRAF3 and TRAF6 in co-transfected HeLa cells (Fig. 2E). Overexpressed RNF166 localized predominantly in the cytosol as dots which were probably aggregates of RNF166 protein. Double immunofluorescent staining showed that overexpressed TRAF6/3, especially TRAF6, had a similar distribution pattern and overlapped with RNF166, while we did not detect co-localization between RNF166 and VISA.
To define these interactions under physiological conditions, we set out to determine associations between endogenous RNF166 and the targets. As we could not obtain an effective antibody to detect endogenous RNF166 through preparation or commercial purchase, we performed immunoprecipitation between overexpressed RNF166 and endogenous TRAF3 and TRAF6 with or without SeV infection. Overexpressed RNF166 weakly interacted with endogenous TRAF3 and TRAF6, and SeV infection greatly enhanced these interactions (Fig. 2F). These data suggested that RNF166 associates with TRAF3 and TRAF6 during viral infection.
To further clarify whether RNF166 targets TRAF3 and TRAF6, we used shRNA to knockdown endogenous TRAF3 or TRAF6, and found that RNF166 no longer potentiated VISA-induced IFN-β activation when TRAF3 or TRAF6 expression was suppressed (Fig. 2G). These data further supported the idea that RNF166 targets TRAF3 and TRAF6 to potentiate RNA virus-induced IFN-β production.
Both TRAF3 and TRAF6 play a critical role in SeV-induced interferon-β production in HEK293T cells. That TRAF3 acts as a critical adaptor in RIG-I-mediated antiviral signaling has been demonstrated by several studies 29,30 , but its role has been doubted recently 13 . So we then determined whether TRAF3 or TRAF6 plays a critical role in SeV-induced IFN-β production by using shRNA to stably knock down endogenous TRAF3 or TRAF6 in HEK293T cells (Fig. 3A). We found that VISA-mediated activation of the IFN-β promoter, SeV-induced transcription, and the secretion of IFN-β were markedly reduced (Figs 2G and 3B), and SeV-induced phosphorylation and dimerization of IRF3 was apparently inhibited when TRAF3 or TRAF6 expression was suppressed (Fig. 3C). (C) HEK293T cell pools with stable knockdown RNF166 were generated by shRNA. Knockdown efficiency was determined by RT-PCR (left) and Q-PCR (right). (D) SeV-mediated activation of the IFN-β reporter was inhibited when the endogenous RNF166 was stably knocked down. Transfection and luciferase assays were performed as in (A). (E) SeV-induced IFN-β production by 293T-shRNF166-#4 and #5 cells was lower than by 293T-shGFP cells. Cells (1 × 10 6 ) were infected with SeV. 12 h after infection, cells were analyzed by Q-PCR (left) and the supernatants were collected for IFN-β bioassays (right). (F) HEK293T cell pools with stable knockdown of RNF114, RNF125, and RNF166 were generated using shRNA. Knockdown efficiency was determined by RT-PCR (left) and Q-PCR (right). (G) SeV-induced IFN-β production by cell pools with stable knockdown of RNF114, RNF125, and RNF138. Cells were infected and analyzed as in (E). (H) The knockdown efficiency of RNF166 in HeLa cells was determined by RT-PCR. (I-L) SeV-, EMCV-infection, poly (I:C)-transfection(4 ug) and Influenza A-infection, rather than dsVACV-transfection(4 ug) induced production of IFN-β by HeLa-shRNF166-#4 and #5 cells were lower than by HeLa-shGFP cells. Q-PCR and bioassays were performed at 12 h after treatment as in (E). Each graph represents the mean ± SD of three independent experiments done in triplicate. ***indicates P < 0.001; **indicates P < 0.01; ns (not significant) indicates P > 0.05.
To further determine the function of TRAF3 and TRAF6, we generated TRAF3 or TRAF6 knockout HEK293T cell lines using the CRISPR/Cas9 system (Fig. 3D). Phosphorylation of IRF3 triggered by SeV-infection was greatly inhibited and the production of IFN-β was blocked when TRAF3 or TRAF6 was deleted (Fig. 3E,F). These results suggested that both TRAF3 and TRAF6 play a critical role in SeV-induced IFN-β production in HEK293T cells.
Functional domain mapping of RNF166. RNF166 contains a RING domain, a zinc finger domain, and an ubiquitin-binding domain (UIM) (Fig. 4A). To dissect the functional role of these domains, we generated three deletion mutants and assessed their ability to up-regulate SeV-induced IFN-β production. We found that the presence of an intact RING domain is essential for RNF166 function, as overexpression of a RING deletion mutant (RNF166 Δ RING) failed to enhance, but rather inhibited the activation of the IFN-β promoter induced by SeV and the overexpression of VISA, while the UIM    (1 μ g). At 12 h after transfection, cells were infected with SeV, and at 24 h after infection, mRNA of P protein was determined by Q-PCR. (F) RNF166 interacted with TRAF3 and TRAF6 via its zinc-finger domain. Transfection and immunoprecipitation (IP) were performed as in Fig. 2 (D). Each graph represents the mean ± SD of three independent experiments done in triplicate. ***indicates P < 0.001; **indicates P < 0.01; ns (not significant) indicates P > 0.05. domain deletion mutant (RNF166 ∆UIM) had no apparent effect (Fig. 4B,C). Consistently, on infection with NDV-eGFP (Newcastle disease virus-enhanced green fluorescent protein), we found that overexpression of RNF166 in HEK293T cells rendered them remarkably resistant to NDV infection and reduced the levels of NDV-eGFP-positive cells, while the RING deletion mutant had the opposite effect (Fig. 4D); and the same results were obtained when RNF166 or RNF166 Δ RING were co-expressed with VISA. Also we found overexpression of RNF166 can inhibit the replication of SeV (measured by mRNA level of SeV P protein), while the RING delete mutant enhanced the proliferation of SeV (Fig. 4E). These data indicated that the RING domain is indispensable for the ability of RNF166 to up-regulate cellular anti-RNA virus activity.
We further performed co-immunoprecipitation to detect which domain of RNF166 is required for interactions with TRAF3 and TRAF6. The results showed that the RING deletion and UIM deletion mutants interacted with TRAF3/6 just like full-length RNF166, while the mutant carrying the RING domain only did not (Fig. 4F), suggesting RNF166 interacts with TRAF3 and TRAF6 via its zinc finger domain, and the RING domain is necessary for its positive regulatory function.

RNF166 enhances RNA virus-induced ubiquitination of TRAF3 and TRAF6. It is known that
the RING domain is the critical functional domain for E3 ubiquitin ligase, and polyubiquitination of TRAF3 and TRAF6 are important for RLR signaling 13,27 , so we suggested that RNF166 could affect the ubiquitination of TRAF3 and TRAF6. We found overexpressed RNF166 enhanced the ubiquitination of TRAF3 and TRAF6 in co-transfection experiments, consistent with previous results, and RNF166 lost this effect when the RING domain was deleted (Fig. 5A).
We then found that the SeV-induced ubiquitination of endogenous TRAF3 and TRAF6 was notably inhibited when RNF166 expression was suppressed (Fig. 5B,C). We further found that K63-linked rather than K48-linked ubiquitination of TRAF3 and TRAF6 was decreased upon SeV infection when RNF166 was knocked down (Fig. 5B,C). However, we did not find an apparent effect of knockdown of endogenous RNF114 and RNF138 on SeV-induced ubiquitination of TRAF3 and TRAF6 (Fig. 5D,E). These data suggested that endogenous RNF166 rather than its homologous proteins RNF114 and RNF138 specifically enhances the SeV-induced ubiquitination of TRAF3 and TRAF6.

Discussion
Virus-triggered production of IFN-β is critical for the antiviral immune response and is delicately regulated in space and time by various molecules and distinct mechanisms. Ubiquitination has emerged as a critical role in forming signaling complexes and transducing signals downstream 13 .
RNF125 has been reported to negatively regulate RIG-I-mediated antiviral activity via conjugating ubiquitin chains to RIG-I and MDA5, leading to their degradation by the proteasome. However, the functions of its homologous proteins RNF114, RNF125, and RNF138 in innate immune signaling pathways remain elusive. Here, we report that RNF166 rather than RNF114, RNF138, and RNF125 potentiates the RNA virus-induced production of IFN-β . Several lines of evidence support this finding. First, overexpressed RNF166 potentiated the SeV-and VISA-mediated induction of IFN-β . Second, knockdown of endogenous RNF166 by shRNA reduced the RNA virus-and dsRNA analog-induced production of IFN-β . Third, RNF166 interacted with endogenous TRAF3 and TRAF6, and these interactions were enhanced upon viral infection. Furthermore, we found that the potentiation of the antiviral effect of RNF166 was mediated by enhancing the virus-induced ubiquitination of TRAF3 and TRAF6. Notably, endogenous RNF114 and RNF138 had no apparent effect on the virus-induced ubiquitination of TRAF3 and TRAF6, and this finding may explain why RNF114 and RNF138 do not enhance SeV-induced antiviral activity while RNF166 did.
TRAF3 and TRAF6 play a vital role in the RIG-I-mediated signaling pathway [29][30][31] . Therefore, both molecules are functionally regulated by cellular components to control the immune response and viral proteins to evade the host immune system. These regulatory mechanisms include interfering with the interaction between VISA and TRAF3/6 32-34 , deubiquitinating TRAF3/6 26,[32][33][34][35] , and disrupting the binding of downstream molecules to TRAF3/6 [36][37][38][39] . Although a recent study has shown that TRAF3 is dispensable for the RNA virus-induced activation of TBK1 and IRF3 13 , our data suggest that TRAF3 is critical for the SeV-elicited production of IFN-β in HEK293T cells. We cannot exclude the possibility that the differences in the cell lines and experimental methods we used explains the different conclusions. However, we agree that TRAF6 is a critical component downstream of VISA 13 .
In light of our study, RNF166 plays an important role in RNA virus-triggered IFN-β production by enhancing the ubiquitination of TRAF3 and TRAF6. We noted that expression of a mutant with the RING domain deleted, RNF166-∆RING, markedly suppressed VISA-mediated activation of the IFN-β promoter and anti-NDV activity. These data indicate that, as a dominant-negative mutant, RNF166∆RING competes with endogenous RNF166 to bind with TRAF3 or TRAF6, so they cannot be efficiently ubiquitinated and blocks VISA signaling. However, STING has no apparent TRAF-binding motifs that are critical for recruiting TRAFs and can directly recruit IRF3 and TBK1 for activation via its carboxyl terminal region 40 ; this may explain why RNF166 does not potentiate the cGAS-and STING-induced activation of the IFN-β promoter.
RNF114 was first identified as a psoriasis-susceptibility gene 41,42 . Later study revealed that overexpressed RNF114 enhances NF-Κ B and IRF3 reporter activity and increases type I IFN mRNA levels 43 . However, the analysis of cells with RNF114 knockdown yielded a heterogeneous set of results that may have been due to the presence of different populations within the polyclonal cell lines or a redundant role of RNF114 43 . Recently, another independent study revealed that RNF114 acts as a negative regulator of NF-κ B-dependent transcription by stabilizing the A20 protein and Iκ Bα 44 . Our data from overexpression and knockdown experiments indicated that RNF114 acts as a negative regulator of SeV-induced IFN-β signaling. A20 has been demonstrated to negatively regulate RIG-I-mediated signaling in several independent studies [45][46][47] , so it is possible that RNF114 depends on A20 to play negative regulatory roles. However, further investigations are needed to confirm the function of RNF114 in the innate immunity signaling pathway.
Although RNF114, RNF125, RNF138, and RNF166 have similar domain structures, their functions in SeV-induced signaling are quite diverse, including both positive and negative effects. Future investigations into their structures are expected to elucidate the molecular mechanisms underlying their different functions in the innate immunity signaling pathway.

Constructs. Mammalian expression plasmids for
Transfection and Luciferase Assay. HEK293T cells were seeded onto 24-well dishes and transfected the next day with polyethylenimine (Electron Microscopy Sciences). To normalize for transfection efficiency, 50 ng of pRL-Tk Renilla luciferase reporter plasmid was added to each transfection. About 18 h after transfection, assays were performed using a dual-specific luciferase assay kit (Promega). Firefly luciferase activity was normalized based on Renilla luciferase activity. All reporter assays were repeated at least three times.

Co-immunoprecipitation and western blot analysis. For transient transfection and immunopre-
cipitation experiments, HEK293T cells (2 × 10 6 ) were transfected with the indicated plasmids for 20 h. The transfected cells were lysed in 0.5 ml lysis buffer (20 mM Tris [pH 7.5], 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 10 mg/ml aprotinin, 10 mg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). For each immunoprecipitation, a 0.4-ml aliquot of lysate was incubated with 0.5 μ g of the indicated antibody and 25 μ l of a 1:1 slurry of protein A-Sepharose (GE Healthcare) for 4 h. The Sepharose beads were washed three times with 1 ml lysis buffer. The precipitates were analyzed by western blotting with the indicated antibodies and visualized by incubation with IRDye800-conjugated secondary antibodies using an Odyssey infrared imaging system (Licor Inc.).
Statistical analysis. Two-way ANOVA analysis were used to analyze data. Experiments were repeated at least three times. Results were considered significant at p < 0.05.