Paramyxovirus V Proteins Interact with the RIG-I/TRIM25 Regulatory Complex and Inhibit RIG-I Signaling

NiV V interacts with TRIM25.

In the course of investigating NiV V antagonism of the human innate immune response, we examined the colocalization of NiV V with several innate immune signaling proteins. We observed that TRIM25 and NiV V colocalize when coexpressed (Fig. 1A). Furthermore, whereas NiV V typically displays diffuse cytoplasmic localization when expressed alone, we found that when coexpressed with TRIM25, NiV V relocalized into condensed cytoplasmic puncta (Fig. 1A), which corresponded to the subcellular distribution of TRIM25. To explore this further, we examined whether NiV V interacts with TRIM25, as has been described for the influenza A virus NS1 protein (44). 293T cells were transfected with hemagglutinin (HA)-NiV V and V5-GST, TRIM25-V5, or TRIM25 CCD-V5 expression constructs. The TRIM25 CCD (coiled-coil domain) was included because this region is sufficient and necessary for the TRIM25-NS1 interaction. Coimmunoprecipitation (co-IP) against the V5 tag revealed that the NiV V protein could interact with TRIM25-V5 but not with V5-GST or the TRIM25 CCD-V5 domain construct (Fig. 1B). Therefore, it appeared that while NiV V does interact with TRIM25, the TRIM25 domain involved in this interaction is different from the one mediating the TRIM25-NS1 interaction. In the same experiment, we also coexpressed TRIM25-V5 with an HA-tagged construct that expresses only the N-terminal region of NiV V (HA-NiV Vn). HA-NiV Vn was unable to interact with TRIM25, suggesting that the V-specific C-terminal domain (V CTD) is necessary for the V-TRIM25 interaction. As further evidence of this, NiV V but not NiV W (which shares the N-terminal sequence of the V protein but has a different CTD) was precipitated with endogenous TRIM25 (Fig. 1C). To determine whether the V CTD was sufficient for interaction with TRIM25, we cotransfected 293T cells with glutathione S-transferase (GST)-V CTD and either a full-length TRIM25-V5 or a TRIM25 CCD-V5 expression construct. Using an antibody specific for the V CTD, we observed that the NiV V CTD coprecipitated with full-length TRIM25 but not with TRIM25 CCD (Fig. 1D). Therefore, we conclude that the NiV V CTD is both necessary and sufficient for the V-TRIM25 interaction.

NiV V interacts with RIG-I.

As TRIM25 is a known regulatory partner of RIG-I, we next explored whether NiV V and RIG-I can interact. We cotransfected an HA-NiV V expression construct with a Flag–RIG-I, Flag-MDA5, or green fluorescent protein (GFP)-Flag construct in 293T cells and showed that NiV V coprecipitated with both RIG-I and MDA5 but not with GFP (Fig. 2A). We confirmed these results by using endogenous RIG-I. An HA-NiV V or HA-NiV W expression construct was transfected into 293T cells, and immunoprecipitation against the HA tag revealed coprecipitation of endogenous RIG-I with NiV V but not with NiV W (Fig. 2B). This confirmed that the V–RIG-I and V-TRIM25 interactions (Fig. 1C) can take place with endogenous levels of RIG-I and TRIM25 that would be present early in a paramyxovirus infection. Once again, the interaction of NiV V, but not NiV W, with RIG-I indicated that the V CTD was necessary for this interaction as well (Fig. 2B). To establish that the V CTD is sufficient for interaction with RIG-I, we cotransfected 293T cells with an HA-NiV V, GST-NiV V CTD, or HA-NiV Vn expression construct and either a Flag–RIG-I or a GFP-Flag expression construct. Immunoprecipitation revealed that while full-length NiV V and the NiV V CTD alone coprecipitated with RIG-I, the NiV Vn construct did not (Fig. 2C). This indicated that the V CTD was both necessary and sufficient for the V–RIG-I interaction, as had been observed for the V-TRIM25 interaction in Fig. 1.

V protein interaction with RIG-I and TRIM25 is conserved across paramyxoviruses and observed in infected cells.

The observation that the V CTD is necessary for both the RIG-I and TRIM25 interactions suggested that these interactions might extend to other paramyxoviruses because of the high conservation of this domain. We first examined the possibility that the V-TRIM25 interaction is conserved. The V5-TRIM25 expression construct was cotransfected into 293T cells with HA-MeV V, HA-NiV V, HA-SeV V, or HA-NiV W. Immunoprecipitation against the V5 tag revealed that the V proteins of MeV, NiV, and SeV all coprecipitated with TRIM25, whereas NiV W did not (Fig. 3A). A similar experiment was also performed with parainfluenza virus type 5 (PIV5) V, and as the V5 tag is derived from the PIV5 V protein, lysates were immunoprecipitated against the HA tag and HA-PIV5 V was shown to coprecipitate with TRIM25 (Fig. 3B). However, we note that this interaction appeared to be less robust in experimental repeats (more than three) than other V-TRIM25 interactions. Taken together, these data indicated that the TRIM25-V protein interaction is conserved across V proteins representing all paramyxovirus genera. To explore the conservation of the V–RIG-I interaction, HA-MeV V and HA-SeV V expression constructs were cotransfected into 293T cells with Flag–RIG-I, GFP-Flag, or an empty vector construct. Immunoprecipitation revealed that both V proteins coprecipitated with RIG-I but not with the empty vector or GFP-Flag (Fig. 3C). A similar experiment performed with HA-NiV V and HA-PIV5 V showed that NiV V and PIV5 V both coprecipitated with RIG-I but not with controls (Fig. 3D). Once again, PIV5 V coprecipitation appeared weaker and less robust in experimental repeats (more than three). These V proteins are representative of different genera in the paramyxovirus family: Henipavirus (NiV V), Respirovirus (SeV V), Morbillivirus (MeV V), and Rubulavirus (PIV5 V), which indicates high-level conservation of the paramyxovirus V protein interaction with RIG-I and TRIM25.

To address whether these interactions occur in the context of paramyxovirus infection, we chose to use MeV as a model since NiV requires BSL4 containment. A549 cells were infected with a recombinant MeV (Edmonston strain) expressing GFP (45) and either fixed or lysed for immunofluorescence or immunoprecipitation assays (Fig. 4). V protein was detected with a specific antibody, and we observed colocalization of MeV V protein with endogenous RIG-I and TRIM25 (Fig. 4A). Interestingly, V protein does not colocalize with endogenous MAVS protein. These results were supported by co-IP of V with endogenous RIG-I in MeV-infected cells, with NS1 from influenza A/PR/8/34 virus serving as a positive control. As shown in Fig. 4B, MeV V protein coimmunoprecipitated with RIG-I, as did TRIM25 and NS1 (left panels). In addition, a reciprocal pulldown assay was done where endogenous RIG-I and TRIM25 both immunoprecipitated with the MeV V protein (Fig. 4C).

Paramyxovirus V proteins interact with distinct RIG-I and MDA5 domains.

Observations in Fig. 1 and 2 indicated that interaction of V proteins with RIG-I and TRIM25 are mediated through the conserved V CTD. Likewise, interaction of paramyxovirus V proteins with MDA5 has been shown to be mediated through the V protein CTD and a region covering residues 676 to 816 of MDA5 that lies within the helicase domain (Fig. 5A) (28, 29, 31). A corresponding region in LGP2 is responsible for interaction of the LGP2 and V proteins, and multiple studies have demonstrated that V proteins do not bind to the corresponding region in RIG-I (28, 29, 31). To define the V–RIG-I interaction domain, we conducted immunoprecipitation experiments and found that both the NiV and MeV V proteins interacted with a truncated RIG-I construct consisting of the N-terminal CARDs of RIG-I (RIG-IN) but not with the truncated version of RIG-I that lacks the CARDs (RIG-IC) (Fig. 5A and B). The RIG-IN construct does not include the region in the helicase domain that is analogous to the V-binding domain in MDA5 (see Fig. 5A), indicating that V interacts with distinct domains in the RIG-I and MDA5 proteins. Interestingly, RIG-IN contains both the site of TRIM25–RIG-I binding and the site of ubiquitination by TRIM25 (lysine 172) (40).

Paramyxovirus V proteins interact with the SPRY domain of TRIM25.

We utilized TRIM25 subdomain constructs to determine the domain that is necessary and sufficient for NiV V binding. In immunoprecipitation experiments, we found that the SPRY domain of TRIM25 was sufficient for the interaction with NiV V (Fig. 5C), which is interesting, as the SPRY domain is also reported to be necessary and sufficient for the TRIM25–RIG-I interaction (15, 40). This, along with the V–RIG-I CARD interaction, suggested a potential model for V binding to RIG-I and TRIM25 wherein V binds at the interface of RIG-I and TRIM25 (Fig. 5D) either to prevent RIG-I/TRIM25 complex formation or to arrest RIG-I signaling progression by hyperstabilizing this complex.

V colocalizes with the RIG-I/TRIM25 complex.

To explore the localization of the V/TRIM25 and V/RIG-I complexes, we employed a yellow fluorescent protein (YFP) reconstitution assay called bimolecular fluorescence complementation (BiFC) (46). BiFC, depicted in Fig. 6A using NiV V and RIG-I as an example, uses recombinant constructs of two potential binding partner proteins that have been fused to the C or N terminus of YFP (see Materials and Methods for details). Constructs expressing YN-NiV W and YC–RIG-I, a combination that did not reconstitute YFP, were used as controls (Fig. 6B). Constructs expressing YN-NiV V were tested in HeLa cells with YC–RIG-I (as illustrated in Fig. 6A) and successfully reconstituted YFP, resulting in yellow fluorescence (Fig. 6C). Likewise, when NiV V BiFC constructs were tested with a set of TRIM25 BiFC constructs, we found that YC-NiV V and YN-TRIM25 were able to interact and reconstitute YFP, yielding yellow fluorescence (Fig. 6D). Taken together, these results confirmed the prior immunoprecipitation data showing interactions between V and RIG-I and between V and TRIM25. As seen in Fig. 1A, NiV V relocalized to punctate cytoplasmic foci to colocalize with TRIM25, and these puncta are the sites of YFP reconstitution with V and TRIM25 BiFC constructs (Fig. 6D).

To view all three proteins in the complex, we cotransfected cells with a validated RIG-I–TRIM25 BiFC pair and an HA-tagged NiV V construct (Fig. 6E). YFP reconstitution indicated that NiV V does not disrupt the RIG-I–TRIM25 interaction. Under these conditions, the NiV V protein was relocalized to cytoplasmic puncta that corresponded (enlarged image in Fig. 6E) to RIG-I/TRIM25 complexes, as indicated by the BiFC signal. This suggests that NiV V has an affinity for participation in the RIG-I/TRIM25 complex and that NiV V relocalization into these puncta is likely driven by TRIM25.

To explore effects of NiV V on the RIG-I/TRIM25 complex, we performed a co-IP where 293T cells were cotransfected with expression constructs for HA–RIG-I, HA-NiV V, and TRIM25-V5. Proteins were precipitated with an anti-V5 antibody. The amounts of RIG-I and TRIM25 were titrated such that the interaction between TRIM25 and RIG-I was only faintly visible in the absence of NiV V (Fig. 6F). However, in the presence of NiV V, TRIM25 showed robust precipitation of RIG-I, along with NiV V. As the BiFC data suggested that NiV V is incorporated into RIG-I/TRIM25 complexes, this result suggests that V may stabilize these complexes (Fig. 6F).

V proteins inhibit RIG-I signaling via TRIM25.

We next explored whether the presence of NiV V in the RIG-I/TRIM25 complex inhibits downstream RIG-I signaling. The first objective was to determine the ability of NiV V to inhibit the activation of RIG-I by a RIG-I-specific ligand. Influenza virus RNA has been reported to potently activate RIG-I without stimulating the MDA5 pathway (13, 47, 48), so this was chosen as the stimulus. 293T cells were transfected with a firefly luciferase reporter under the control of the IFN-β promoter along with a constitutive Renilla luciferase reporter and either the empty vector or an expression construct for NiV V, NiV W, or RIG-IC. RIG-IC, which consists of the RIG-I helicase and regulatory domains only, acts as a dominant negative inhibitor (6), and therefore controls for the RIG-I specificity of IFN-β promoter activation. We have previously shown that NiV W is a broad inhibitor of both RLR- and TLR-mediated IFN induction (49), and here it was used as an additional control. Following plasmid transfection, cells were transfected with RNA collected from influenza virus-infected A549 cells, and 24 h later, luciferase expression was quantified and normalized as a measure of IFN-β promoter activation in response to the RNA stimulus. We observed inhibition by RIG-IC, indicating that the signaling is RIG-I specific, and both the NiV V and NiV W proteins showed inhibition of reporter activation (Fig. 7A). This confirmed that the NiV V protein can inhibit RIG-I activation of IFN-β production.

Given the interaction of NiV V with TRIM25, we hypothesized that the inhibition of RIG-I signaling occurred at the step involving TRIM25 activation of RIG-I. To test this, we used an IFN-β reporter system that requires TRIM25 overexpression to reach full activation. In this system, a small amount of RIG-IN expression plasmid is transfected into cells so as to weakly stimulate the IFN-β reporter. RIG-IN is known to be constantly active in the absence of stimuli since it lacks the repressor domain, and it can be used to activate IFN induction and simulate the effect of RIG-I signaling (6). TRIM25 can further enhance promoter activation by acting on RIG-IN, which allows measurement of the effects of TRIM25 on RIG-I signaling (40). The addition of a small amount of TRIM25 plasmid (50 ng) was required to enhance signaling, and in the presence of increasing amounts of NiV V, the enhancement of RIG-I signaling was reduced to the baseline (Fig. 7B). The same effect of NiV V on TRIM25-dependent enhancement of signaling by full-length RIG-I was observed (Fig. 7C). We extended these results to the MeV V protein, which was coexpressed with RIG-IN in the absence or presence of TRIM25. Like NiV V, MeV V showed inhibition of TRIM25-mediated enhancement of RIG-IN signaling (Fig. 7D). Furthermore, to address the possibility that these effects involve secondary feedback from STAT1 inhibition by NiV V, we also included a NiV V G121E mutant construct that is incapable of binding STAT1 or inhibiting IFN signaling (50). The data show that the mutant construct retains the ability to inhibit RIG-I signaling, indicating that STAT1 binding is not required (Fig. 7D).

V proteins do not require LGP2 or MDA5 to inhibit the RIG-I pathway.

It has been reported that overexpression of LGP2 can facilitate the inhibition of RIG-I signaling by V proteins through inhibition of RIG-I oligomerization (34), but whether this is specific to RIG-I has been questioned (35). Although we did not include exogenous LGP2 in our assays or observe effects of V on RIG-I oligomerization (data not shown), it was nonetheless important to address the potential for a contribution by LGP2, and also MDA5, to our observations. siRNAs targeting LGP2 and MDA5 were used to deplete the expression of these proteins and assess the effects on V-mediated inhibition of RIG-I. To detect LGP2 and MDA5 mRNAs and validate knockdown, 293T cells were pretreated with IFN-α prior to siRNA transfection. Quantitative reverse transcription (qRT)-PCR demonstrated that the specific siRNAs significantly reduced LGP2 and MDA5 expression versus the nontargeting siRNA (Fig. 8A). To examine the effects of MDA5 and LGP2 knockdown on the RIG-I–NiV V interaction, 293T cells were transfected with MDA5, LGP2, or nontargeting siRNA. At 48 h after siRNA transfection, cells were transfected with plasmids expressing Flag–RIG-IN or GFP-Flag, and HA-NiV V and anti-Flag immunoprecipitation was performed. The RIG-IN–NiV V interaction was not affected by any of the siRNAs, suggesting that MDA5 and LGP2 are not required for NiV V to bind RIG-I (Fig. 8B).

The effect of MDA5 or LGP2 siRNA on the ability of NiV V to inhibit IFN induction by RIG-I/TRIM25 was also examined. 293T cells received no siRNA; nontargeting siRNA; or siRNAs specific to MDA5, LGP2, or the MAVS protein. MAVS protein siRNA served as a positive control for disruption of the RIG-I pathway. At 48 h after siRNA transfection, cells were transfected with the IFN-β reporter plasmid system together with RIG-IN, TRIM25, and NiV V, as indicated in Fig. 8C. The reporter assay data were analyzed by two-way analysis of variance (ANOVA) to determine if the siRNA treatment had any global effects on signaling (in the absence of NiV V) and then to compare the NiV V inhibition of TRIM25–RIG-I-mediated IFN induction under different siRNA conditions. The results for no siRNA, nontargeting siRNA, and MDA5 siRNA treatment were not statistically distinguishable from one another, indicating that MDA5 does not contribute to this signaling event. MAVS protein siRNA, as expected, had a significant effect on RIG-I signaling. siRNA against LGP2 also had a significant effect on the RIG-I pathway; however, within the LGP2 siRNA-treated group, NiV V still appeared to attenuate the level of IFN induction caused by the coexpression of RIG-IN and TRIM25 (Fig. 8C).

To further verify the lack of a requirement for LGP2 and MDA5, we expressed a mutant form of NiV V (R409A) that was reported to be deficient in MDA5 binding (30). No significant difference was observed between the abilities of WT NiV V and NiV V R409A to inhibit RIG-I signaling (Fig. 8D). As a control for inhibition of RIG-I signaling, we included the NiV W protein, which is known to inhibit IFN induction through inhibition of IFN regulatory factor 3 (49). We conclude that neither MDA5 nor LGP2 is necessary for V to inhibit the TRIM25/RIG-I signaling complex.

V proteins inhibit TRIM25-mediated RIG-I ubiquitination.

The data thus far indicated that V proteins interact with RIG-I and TRIM25 in a manner that inhibits TRIM25-mediated enhancement of downstream RIG-I signaling. We hypothesized that V may be interfering with TRIM25 ubiquitination of RIG-I, so to investigate this, we performed a GST pulldown assay of lysates from 293T cells transfected with GST or GST–RIG-IN in the absence or presence of V5-TRIM25. Ubiquitin was also overexpressed in this system. Western blot analysis of eluants with an anti-GST antibody showed evidence of covalent modification of GST–RIG-IN with ubiquitin, which was enhanced under conditions where TRIM25 was overexpressed (Fig. 9A). When NiV V was expressed, RIG-IN displayed loss of this ubiquitination pattern (Fig. 9A), indicating that NiV V inhibits RIG-I ubiquitination by TRIM25. We extended these results to the MeV and PIV5 V proteins (Fig. 9B) and added an additional control where HA-ubiquitin was not coexpressed with RIG-IN and TRIM25. This control confirmed that the level of ubiquitination observed was directly related to the coexpression of all three components needed for TRIM25-mediated ubiquitination of RIG-I (Fig. 9B). As in Fig. 9A, we observed robust ubiquitination of RIG-IN in the presence of TRIM25 and ubiquitin. However, when any of the paramyxovirus V proteins were present, the ubiquitination of RIG-IN was ablated. This result indicates that multiple paramyxovirus V proteins can inhibit RIG-I ubiquitination, and thus, this is a conserved mechanism.

V protein inhibition of RIG-I ubiquitination prevents the downstream RIG-I–MAVS protein interaction.

To investigate the consequences of inhibiting RIG-I ubiquitination, we examined the effect of NiV V on the ability of RIG-I to interact with MAVS protein. We performed a GST pulldown assay of lysates of 293T cells transfected with GST or GST–RIG-IN, TRIM25, HA-ubiquitin, and NiV V as indicated in Fig. 10A. In addition to examining RIG-IN ubiquitination, a MAVS protein antibody was used to detect interaction between RIG-IN and the MAVS protein. In the absence of V protein expression, GST–RIG-IN interacted with endogenous MAVS protein, and when TRIM25 was coexpressed with RIG-IN, this interaction was enhanced, which correlated with increased RIG-IN ubiquitination (Fig. 10A). However, when the NiV, MeV, or PIV5 V protein was present, the RIG-IN interaction with the MAVS protein was substantially diminished (Fig. 10A and B). Therefore, NiV V, MeV V, and PIV5 V are all able to prevent the RIG-I–MAVS protein interaction and this correlates with their inhibition of TRIM25-mediated RIG-I ubiquitination.

Plasmids and siRNAs.

All expression constructs were in the pCAGGS vector unless otherwise stated (52). The NiV V and W expression plasmids were described previously (53). The PIV5 V protein expression construct was amplified by RT-PCR from cells infected with PIV5 purchased from ATCC, tagged with HA on the N terminus during amplification, and then cloned into the pCAGGS vector between SacI and XhoI restriction sites. The MeV V expression construct was amplified from a plasmid encoding the V protein of the MeV Edmonston strain, tagged with HA on the N terminus during amplification, and cloned into the pCAGGS vector between EcoRI and XhoI restriction sites. The SeV V expression construct was amplified from a V plasmid prepared from SeV Cantell by Carolina López, tagged with HA on the N terminus during amplification, and then cloned into the pCAGGS vector between SacI and XhoI restriction sites. The G121E mutant NiV V protein was generously provided by Michael Ciancanelli and Christopher Basler as previously described (50). RIG-IN, RIG-IC, and full-length RIG-I constructs were described elsewhere (6, 54–56). GST, HA-ubiquitin (HA-Ub), and TRIM25 constructs were generated in previous work (40, 44). TRIM25 subdomain constructs were kindly provided by Michaela Gack and subcloned into the pCAGGS vector by Clontech Infusion Cloning. The IFN-β firefly luciferase reporter plasmid and the constitutive Renilla luciferase reporter plasmid were reported in prior work (57). NiV V R409A was produced from the NiV V plasmid by site-directed mutagenesis with the Agilent QuikChange II site-directed mutagenesis kit (catalog no. 200521) in accordance with the manufacturer's instructions. Sequences encoding YFP amino acids 1 to 154 (yn) and 155 to the end (yc) were subcloned from pEYN-NXF1 and pEYC-Y14 (kindly provided by Peter Lichter, German Cancer Research Center, Heidelberg, Germany) (46) into mammalian expression vector pCAGGS (52). NiV V and W were subcloned into these BiFC-pCAGGS vectors either between ClaI and XhoI (X-yn and X-yc constructs) or between XhoI and NheI (yn-X and yc-X constructs). RIG-I, TRIM25, and the MAVS protein were amplified from template DNA described in previous work (40, 56) and subcloned into the BiFC pCAGGS vectors either between ClaI and XhoI (X-yn and X-yc constructs) or between XhoI and NheI (yn-X and yc-X constructs). The eight possible combinations of yn and yc constructs for every protein-protein interaction of interest were assayed for BiFC complementation, and those combinations showing the strongest BiFC signal were selected for further studies.

siRNA pools against the MAVS protein, LGP2, and MDA5, as well as nontargeting controls, were obtained from Dharmacon (MDA5, catalog no. L-013041-00-0005; LGP2, catalog no. L-010582-00-0005; MAVS protein, catalog no. L-024237-00-0005; nontargeting, catalog no. D-001810-10-05).

Viruses and cells.

Co-IP and reporter assays were conducted with 293T cells (ATCC CRL-3216). Influenza virus infections were conducted with A549 cells (ATCC CCL-185). Influenza A/WSN/1933 virus stocks were grown in MDCK cells, and their titers were determined. BiFC experiments were conducted with HeLa cells (ATCC CCL-2).

GFP-expressing MeV was generously provided by Benhur Lee and grown as described in reference 45. Briefly, Vero cells were infected with MeV, supernatants were collected after 10 days postinfection, and the titers were determined. For the immunoprecipitation and immunofluorescence assays, A549 cells were infected with MeV and cells were lysed or fixed, respectively, at 24 h postinfection.

Infection and RNA purification.

For collection of RNA from influenza virus-infected cells, A549 cells were infected for 1 h with influenza A/WSN/1933 virus at a multiplicity of infection (MOI) of 0.01. At 24 h postinfection, total RNA was harvested with the Qiagen RNeasy MIDI kit in accordance with the manufacturer's protocol. Cell culture supernatants were collected, and virus titers were measured by hemagglutination assay. RNA was utilized to activate RIG-I only if titers were measured at ≥16 hemagglutination units, which typically corresponds to a titer of ≥10⁷ PFU/ml (58). RNA for quantitative PCR (qPCR) was purified with the Qiagen RNeasy minikit.

DNA transfection conditions.

DNA transfections were performed in suspension, except for transfections for immunofluorescence experiments, where HeLa cells were seeded 24 h before transfection. When necessary, the empty vector was added to produce equal amounts of total transfected DNA. Lipofectamine 2000 (Life Technologies SKU no. 11668027) was used at a DNA-to-reagent ratio of 1:1 or 1:2. The DNA transfection procedure was performed as described above; however, at 12 to 20 h posttransfection, the medium was changed to allow cells to recover before RNA transfection. At 24 h after DNA transfection, cells were transfected with RNA collected from influenza virus-infected cells. Lipofectamine 2000 was used at a 1:2 ratio. Cells were harvested 12 to 16 h after RNA transfection and processed for reporter assays. siRNA was reverse transfected 48 h prior to DNA transfection with the Lipofectamine RNAiMAX reagent (Life Technologies SKU no. 13778-150). DNA transfection was then performed as already described, and cells were harvested 72 h after siRNA treatment. All assays were performed at least three times and in triplicate.

Immunoprecipitation or GST pulldown assay.

Cells were lysed at 4°C in 1% Triton X-100 buffer containing protease inhibitor cocktail (eComplete mini; Roche product no. 11836170001). Ten percent of the lysate was saved as whole-cell extract (WCE), and the remainder was incubated overnight at 4°C with a specific antibody. The antibodies used in this work were anti-HA (Sigma-Aldrich; rabbit, product no. H-6908; mouse, product no. H-9658), anti-Flag (Sigma-Aldrich; rabbit, product no. F7425; mouse Flag-M2, product no. F3165), anti-MeV V (59), or anti-V5 (Life Technologies; mouse, product no. R960-25). Protein A agarose beads (Roche product no. 11719408001) were then added for 2 to 8 h of incubation at 4°C. Beads were washed three or more times in lysis buffer or phosphate-buffered saline (PBS)-Tween and then eluted in SDS running buffer by boiling for 5 min. For GST pulldown assays, GST beads (Pierce product no. 16100) were added to the lysates for at least 2 h of incubation at 4°C and then treated as described above. Immunoprecipitation experiments were performed at least three times.

IFN treatment.

Human IFN-α (Calbiochem/Millipore catalog no. 407294) was added to cells for 24 h of incubation at a concentration of 100 IU/ml.

Immunoblot assay detection.

Lysates and eluants were subjected to SDS-PAGE on Tris glycine gels in the corresponding buffer. Gels were transferred to nitrocellulose membranes by semidry transfer, blocked, and incubated with the appropriate antibody. The antibodies used were anti-MeV V protein (59), anti-HA (Sigma-Aldrich; rabbit, product no. H-6908; mouse, product no. H-9658), anti-Flag (Sigma-Aldrich; rabbit, product no. F7425; mouse Flag-M2, product no. F3165), anti-V5 (Life Technologies; mouse, product no. R960-25), anti-NiV V (60), anti-RIG-I (36), anti-TRIM25 (BD Transduction Laboratories product no. 610570) anti-MAVS protein (Bethyl Laboratories; rabbit, product no. A300-782A), or anti-GST (horseradish peroxidase [HRP] conjugated, raised in a goat, from GE Healthcare Life Sciences; product no. RPN1236). Rabbit and mouse HRP-conjugated secondary antibodies were obtained from Invitrogen/Life Technologies (goat anti-rabbit, product no. G21234; goat anti-mouse, product no. G21040) and Sigma (rabbit anti-mouse, product no. A9044). Western blot assay quantifications were performed with Bio-Rad ImageLab software. All experiments were performed at least three times.

Reporter assays.

Cells were transfected with an IFN-β promoter-driven firefly luciferase reporter plasmid and with a thymidine kinase promoter-driven Renilla luciferase plasmid for normalization. Cells were harvested and lysed in passive lysis buffer from the Promega Dual Luciferase kit (product no. E1960). Cells were disrupted by freeze-thawing in a dry ice-ethanol bath, and lysates were analyzed for luciferase expression on a Promega GloMax luminometer. All assays were performed in triplicate.

qPCR.

RNA was reverse transcribed with random hexamer primers by using the Superscript III First Strand Synthesis kit from Invitrogen/Life Technologies. qPCR primers were designed for MDA5 and LGP2, and primers for α-tubulin were used as a normalization control. Samples were prepared with the Roche 480 SYBR green I master mix (Roche product no. 04707516001), and qPCR was performed on a Roche 480 LightCycler. Knockdown efficiency was calculated by the ΔΔC_T method recommended by Dharmacon/Thermo Scientific for biological replicates, with the gene for α-tubulin as a reference (normalization). Experiments were performed in triplicate.

Immunofluorescence assay.

HeLa cells were cultured in glass-bottom 12-well plates (MaTtek, Ashland, MA) and transfected with the plasmids indicated. At 24 h posttransfection, plates were fixed in ice-cold absolute methanol for 30 min and blocked with 1% bovine serum albumin in PBS. For BiFC experiments, plates were incubated for 3 h at 30°C and then fixed and blocked by same procedure. Samples were incubated with the primary antibodies (anti-HA, anti-Flag, anti-TRIM25, or anti-RIG-I, as appropriate—see Immunoblotting) for 1 h, washed, and incubated for 1 h with secondary antibodies. Nuclei were stained with 1 μg/ml DAPI (4′,6-diamidino-2-phenylindole; Invitrogen). Experiments were performed at least three times.

Microscopic image acquisition.

Confocal laser scanning microscopy was performed with a Zeiss LSM 510 Meta (Carl Zeiss Microimaging, Thornwood, NY) fitted with a Plan Apochromat 63×/1.4 or 40×/1.4 oil objective. One-micrometer Z-stack slices were taken for the images indicated. Images were collected at 8 bits and at a resolution of 1,024 by 1,024 pixels. Images were processed for three-dimensional projection with LSM Image Browse software. For Fig. 4A, images were acquired with a Zeiss LSM880 confocal laser scanning microscope (Carl Zeiss Microimaging, Thornwood, NY). The intensity of fluorescence histograms was analyzed with Fiji software (https://fiji.sc/). In all cases, images are representative of more than five fields analyzed per condition, with analysis of more than 100 cells per field. Experiments were repeated at least three times.