Research ArticleImmunology

LGP2 binds to PACT to regulate RIG-I– and MDA5-mediated antiviral responses

See allHide authors and affiliations

Science Signaling  01 Oct 2019:
Vol. 12, Issue 601, eaar3993
DOI: 10.1126/scisignal.aar3993

An antiviral PACT

The RIG-I–like receptors (RLRs) RIG-I and MDA5 sense cytosolic viral RNAs during infection and activate signaling pathways that culminate in the production of type I interferons (IFNs) and proinflammatory cytokines as part of the antiviral immune response. LGP2, another member of the RLR family, inhibits RIG-I–mediated responses while enhancing MDA5 signaling. Sanchez David et al. found that this differential effect of LGP2 depended on its binding to PACT, a cofactor of DICER in the processing of microRNAs. Mutation of a residue in LGP2 required for the LGP2-PACT interaction resulted in loss of the regulatory effect of LGP2 over the other RLRs. Together, these data provide evidence of a connection between the RNA-silencing machinery and the innate immune response.

Abstract

The retinoic acid–inducible gene I (RIG-I)–like receptors (RLRs) RIG-I, MDA5, and LGP2 stimulate inflammatory and antiviral responses by sensing nonself RNA molecules produced during viral replication. Here, we investigated how LGP2 regulates the RIG-I– and MDA5-dependent induction of type I interferon (IFN) signaling and showed that LGP2 interacted with different components of the RNA-silencing machinery. We identified a direct protein-protein interaction between LGP2 and the IFN-inducible, double-stranded RNA binding protein PACT. The LGP2-PACT interaction was mediated by the regulatory C-terminal domain of LGP2 and was necessary for inhibiting RIG-I–dependent responses and for amplifying MDA5-dependent responses. We described a point mutation within LGP2 that disrupted the LGP2-PACT interaction and led to the loss of LGP2-mediated regulation of RIG-I and MDA5 signaling. These results suggest a model in which the LGP2-PACT interaction regulates the inflammatory responses mediated by RIG-I and MDA5 and enables the cellular RNA-silencing machinery to coordinate with the innate immune response.

INTRODUCTION

Progress in our understanding of the cellular pathways that control viral replication suggests that modulating host cell functions early upon viral infection could inhibit a large panel of RNA viruses (15). The retinoic acid–inducible gene I (RIG-I)–like receptors (RLRs) appear to be located at the frontline of the evolutionary race between viruses and the host immune system (6). This family of pattern recognition receptors is a group of cytosolic RNA helicases that can identify viral RNA as nonself through binding to pathogen-associated molecular patterns. To date, three RLRs have been identified: RIG-I, MDA5 (melanoma differentiation–associated gene 5), and LGP2 (laboratory of genetics and physiology 2) (7, 8). RIG-I and MDA5 share a number of structural similarities, including their organization into three distinct domains: (i) a N-terminal caspase activation and recruitment domain (CARD), (ii) a central DExD/H box RNA helicase domain, and (iii) a C-terminal regulatory domain. Upon ligand binding, MDA5 and RIG-I stimulate a signaling cascade through mitochondrial antiviral-signaling protein (MAVS) on mitochondria. MAVS activation leads to the translocation of the transcription factors interferon regulator factor 3 (IRF3) and IRF7 and nuclear factor κB (NF-κB) to the nucleus to coordinate the expression of genes encoding the type I interferons (IFNs) and proinflammatory cytokines, which results in the activation of hundreds of IFN-stimulated genes (ISGs) (7). The RLR response is required to initiate an antiviral state and to orchestrate the inflammatory response in the infected tissue.

Unlike RIG-I and MDA5, LGP2 lacks a CARD domain. It is involved in regulating the activities of MDA5 and RIG-I (9). LGP2 synergizes with MDA5 to increase the specific antiviral response, possibly through increasing the ability of MDA5 to form stable filaments on short double-stranded RNA (dsRNA), thus amplifying its signaling activity (10). In picornavirus and paramyxovirus infections, MDA5-specific RNA ligands have been shown to interact with LGP2 (11, 12). In addition, LGP2 knockout cells have a reduced type I IFN response to picornaviruses ex vivo and in vivo (13).

Ex vivo studies have shown that ectopic expression of LGP2 or of the LGP2 C-terminal domain (CTD) alone leads to loss of the RIG-I–mediated type I IFN response (14, 15). It was first suggested that because LGP2 has a higher affinity for dsRNA than that of RIG-I, it could sequester viral dsRNA, thus preventing RIG-I from recognizing its ligand (16). However, a mutated LGP2 deficient in RNA binding can still inhibit RIG-I signaling (17). Furthermore, it was described that the LGP2 CTD could bind to the RIG-I CTD to inhibit recognition of the viral ligand (15), although this interaction has not been observed by others (18, 19). A third model was proposed in which the interaction between LGP2 and MAVS could be responsible for inhibiting the activation of the signaling cascade (9). However, this hypothesis is contradictory with the fact that LGP2 amplifies the MDA5 response, which shares a common signaling cascade with RIG-I.

In vivo studies have increased the controversy about the mechanism by which LGP2 inhibits the RIG-I response. Consistent with ex vivo models, LGP2 knockout mice developed by Venkataraman et al. (13) exhibit an increased inflammatory response to viruses recognized by RIG-I . However, other in vivo studies with LGP2 knockout mice reported the opposite effect (20, 21). Last, mouse models that overexpressed human LGP2 demonstrated a reduction of the type I IFN response while presenting reduced viral fitness (22, 23). To better understand the mechanism of regulation of the RLR-mediated antiviral response by LGP2, we investigated the LGP2 protein network through a previously validated high-throughput proteomic approach (24). Among the LGP2-binding partners that we identified, we focused on RNA-silencing machinery components, particularly on the IFN-inducible dsRNA-dependent protein kinase activator A (PACT) and its effect on the regulation of RIG-I– and MDA5-dependent antiviral responses by LGP2.

RESULTS

LGP2 overexpression enhances MDA5- and blocks RIG-I–mediated IFN-β promoter signaling

We previously generated a human embryonic kidney–293 (HEK293) cell line stably expressing One-STrEP–tagged LGP2 (hereafter referred to as ST-LGP2; Fig. 1A) and validated the ability of the N-terminally tagged LGP2 to interact with LGP2-specific viral RNA ligands (12). This cell line had 10,000-fold more LGP2 mRNA copies than did standard HEK293 cells (12). First, we analyzed the functional profile of ST-LGP2 cells transfected with synthetic agonists of RIG-I and MDA5. Luciferase (Luc) reporter assays were used to evaluate the IFNB1 (which encodes IFN-β) and NF-κB promoter activities in cells transfected with in vitro–transcribed 5′-triphosphate RNA (5′3P-RNA) and low–molecular weight (LMW) poly(I:C), which are commonly used to stimulate a RIG-I response (25, 26), or with high–molecular weight (HMW) poly(I:C), which specifically triggers MDA5 signaling (26). We observed that, compared to transfection of the parental HEK293 cells, transfection of ST-LGP2 cells with 5′3P-RNA resulted in a statistically significantly reduced IFNB1 promoter activity, whereas transfection with synthetic HMW poly(I:C) statistically significantly increased IFNB1 promoter activity (Fig. 1B). Unexpectedly, transfection of the ST-LGP2 cells with LMW (which is typically a RIG-I agonist) also statistically significantly increased IFNB1 promoter activity (Fig. 1B). This could be explained by the ability of MDA5 to form stable filaments on short dsRNA in the presence of LGP2 (10). A similar profile was observed in ST-LGP2 cells expressing the NF-κB promoter activation upon transfection with either 5′3P-RNA or HMW poly(I:C) (Fig. 1C). These observations confirm previous studies showing that the overexpression of LGP2 synergizes with the type I IFN response upon activation with MDA5-specific agonists, such as HMW poly(I:C) (27), but antagonizes signaling of the RIG-I–specific ligand 5′3P-RNA (14).

Fig. 1 ST-LGP2 cells exhibit inhibition of the type I IFN response to RIG-I–specific ligands but amplify the response to MDA5-specific ligands.

(A) Workflow for generating the ST-LGP2 cell line. (B and C) Measurement of the IFN-β (B) and NF-κB (C) promoter responses by luciferase assay in ST-LGP2 cells (green) and HEK293 cells (black) upon transfection with 100 ng each of the indicated synthetic RLR ligands. HEK293 cells were included as a reference control. Data are means ± SEM of five independent experiments. The asterisk indicates a significant difference. ns, no significant difference. a.u., arbitrary units. (D to F) Measurement of the responsiveness of ST-LGP2 cells to natural RLR ligands. (D) Purification protocol of RLR-specific natural ligands. HeLa cells were infected (at an MOI of 1) with MV, Mengo virus, or Cox virus or were left uninfected (mock). Infection with picornaviruses was performed in the presence of DMSO (as a control) or 1 mM ribavirin (step 1). Total RNAs were then purified and used to transfect ST-LGP2 cells together with the IFN-β reporter plasmid before being subjected to luciferase assays (step 2). (E) Comparative analysis of IFN-β promoter responses in ST-LGP2 cells (green) and HEK293 cells (black) that were transfected with 20 or 100 ng of total RNA isolated from cells infected with the indicated viruses. (F) Comparative analysis of IFN-β promoter responses in the indicated cells after transfection with 100 ng of total RNA isolated from cells infected with the indicated UV-inactivated viruses. Luciferase assays were preformed, and data were analyzed as described for (B). Data are means ± SEM of three (E) and two (F) independent experiments. See table S1 for all details on statistical analysis.

Similarly to synthetic ligands, viral infections directly activate RLRs and trigger a type I IFN response. Paramyxovirus infection generates RIG-I–specific RNA ligands that trigger the type I IFN response (12, 28, 29), whereas picornavirus infection activates LGP2 and MDA5 (11, 13, 20, 3032). Thus, we analyzed the type I IFN response to viral RNA in ST-LGP2 cells (Fig. 1D). HeLa cells were infected at a multiplicity of infection (MOI) of 1 with the negative-sense RNA virus Measles virus (MV; family, Paramyxoviridae) or with the positive-sense RNA viruses Mengovirus (Mengo; family, Picornaviridae; genus, Cardiovirus) and Coxsackie virus (Cox; family, Picornaviridae; genus, Enterovirus). For the two picornaviruses, infections were performed in the presence or absence of the nucleoside analog ribavirin (1 mM), a drug that inhibits viral RNA production (33). Total RNAs were extracted 16 hours after infection for the Mengo and Cox viruses and 24 hours after infection for MV. The purified RNA molecules were then used for the transfection of either HEK293 or ST-LGP2 cells together with a reporter plasmid expressing the luciferase gene (luc) under the control of the IFNB1 promoter sequence. Transfection of the HEK293 cells with total RNAs purified from MV-infected cells induced a dose-dependent activation of the IFNB1 promoter, whereas transfection of the ST-LGP2 cells did not result in the same increase in IFNB1 promoter activity (Fig. 1E), similarly to what we observed in cells treated with 5′3P-RNA (Fig. 1B). Transfection of the ST-LGP2 cells with total RNAs extracted from Mengo- or Cox-infected cells induced statistically significantly greater IFNB1 promoter activity than that observed in transfected HEK293 cells (Fig. 1E). However, increasing amounts of RNAs purified from Mengo-infected cells attenuated the IFNB1 promoter activity in transfected ST-LGP2 cells. Furthermore, when Mengo or Cox infections were performed in the presence of ribavirin, no IFNB1 reporter activity was observed upon transfection of HEK293 cells with the resulting total RNAs (Fig. 1E). However, transfection of ST-LGP2 cells with the total RNA from Mengo-infected cells induced IFNB1 promoter activation, suggesting that 1 mM ribavirin was not sufficient to inhibit Mengo replication in HeLa cells. Last, ultraviolet (UV) inactivation of Mengo and MV (which renders the viruses unable to replicate) before infection of HeLa cells led to the complete elimination of the immunostimulatory activities of the corresponding extracted total RNAs (Fig. 1F). Together, these results confirm that overexpressing ST-LGP2 increases cellular sensitivity to MDA5 ligands and reduces the response to RIG-I agonists.

Identification of LGP2-specific protein partners

Because LGP2 mutants lacking RNA binding activity are still capable of suppressing the RIG-I response (17), we searched for LGP2-specific protein partners that could explain the LGP2-dependent control of RLR signaling. To identify LGP2-specific cellular protein partners, we applied a high-throughput proteomic approach that we previously developed and validated (24). ST-LGP2/protein cocomplexes were purified from ST-LGP2 cells by One-STrEP–tag affinity chromatography (12). A stable cell line expressing the One-STrEP-Cherry protein (hereafter referred to as ST-CH) instead of tagged LGP2 was used as a negative control to enable subtraction of nonspecific binding. Data were obtained from three independent experiments and compared to the negative control CH/protein samples. LGP2- and CH-interacting proteins were resolved by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by silver staining, showing that more proteins copurified with LGP2 than with the negative control CH protein (fig. S1A). LGP2- and CH-protein samples were further analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) using our developed pipeline (Fig. 2A and fig. S1B). Spectra were analyzed using Mascot version 2.3 for the peptide and protein identification steps. A query was performed against a nonredundant database of 20,248 human protein sequences from Swiss-Prot to which corresponding decoy entries were automatically added.

Fig. 2 Proteomic analysis of LGP2-specific host protein partners.

(A) Workflow used to purify LGP2-associated cellular proteins. ST-LGP2 was copurified with interacting cellular partners using STrEP-Tactin beads. Protein complexes were released from the beads and analyzed by direct nano–LC-MS/MS. Nonspecific binding was accounted for by using MS results for ST-CH protein (negative control) and our filtered protocol (fig. S1B). Three biological replicates were obtained and analyzed. (B) LGP2-specific protein cocomplex. Proteins are represented according to their gene symbol and colors depict the corresponding GO term. For the corresponding gene name and protein score for each experimental replicate (see table S2). The P value was obtained with the Fisher exact test, whereas the corrected P value was determined with the Bonferroni test. ncRNA, noncoding RNA.

LGP2-specific protein partners were determined by applying three filtering criteria. First, to be considered as a positive hit, the protein score had to be >60 in at least two of the replicates and not <40 for the third replicate. Second, if a protein was identified at least once in any of the CH-protein triplicates with a score of >20, it was discarded. This filtering protocol limited false positives and enabled the generation of a list of highly specific LGP2 partners (table S2). This list includes well-known partners of LGP2, such as endoribonuclease Dicer (DICER1), interferon-induced, double-stranded RNA-activated protein kinase (PKR), NF-κB–repressing factor (NKRF), double-stranded RNA-binding protein Staufen homolog (STAU1), DEAH box protein 30 (DHX30), and 5´-3´ exoribonuclease 2 (XRN2) (table S2, gray entries) (18), confirming the sensitivity of our proteomics approach.

We then analyzed LGP2 partner proteins for Gene Ontology (GO) enrichment using the DAVID database (34, 35). GO term enrichments with P values of <0.05 from three GO hierarchies (biological process, cellular compartment, and molecular function) were represented using the Cytoscape network analysis tool (36). Proteins from the GO terms “RNA splicing,” “ncRNA metabolic process,” “translation,” “mitochondrion,” and “posttranscriptional gene silencing” were found to be particularly enriched (Fig. 2B). Because of the direct involvement of LGP2 in type I IFN signaling, we also included the GO term “response to stimulus.” We then generated a table with the corresponding GO term references, the number of proteins for each of the GO terms, and the respective P values (Fig. 2B). Our data suggest that LGP2 has the capacity to interact directly and indirectly with numerous proteins that are the components of different cellular functional modules. One or several of these LGP2-specific protein partners could be involved in controlling RLR signaling.

Validation of the interaction between LGP2 and components of the RNA-silencing machinery

Among the GO terms identified by our LC-MS/MS analysis, we focused on proteins of the RNA-silencing machinery because their involvement in the antiviral response in mammals is controversial (37) and because only five of these proteins were sufficient to contribute to a statistically significant P value: DICER1, MOV10, SRRT, RBM4, and PACT (Fig. 2B, light blue).

We first confirmed the specific interaction of LGP2 with DICER1, PKR, and PACT by Western blotting analysis (Fig. 3A). These interactions appeared to be independent of viral infection because binding was observed both in mock- and MV-infected cells. As before, ST-CH protein was used to control for nonspecific protein binding. Transactivation-responsive RNA binding protein (TRBP) was previously found by a yeast two-hybrid assay to interact specifically with LGP2 and to be important for the cardiovirus-dependent IFN response (38). We were unable to confirm the specific binding of TRBP to LGP2 (Fig. 3A). We then focused on the LGP2-PACT interaction because PACT was previously shown to directly interact with the RIG-I CTD and to amplify RIG-I–dependent immunostimulatory activity (39). In addition, other studies have shown that different viruses, including Ebola virus, Middle East respiratory syndrome coronavirus (MERS-CoV), and severe acute respiratory syndrome coronavirus (SARS-CoV), target PACT to subvert the RIG-I–mediated response (4042).

Fig. 3 Validation of the protein-protein interaction between LGP2 and PACT.

(A) LGP2-specific binding to DICER1, PKR, and PACT. ST-LGP2 cells and ST-CH cells were infected with MV at an MOI of 1 (+) or were left uninfected (−). Twenty-four hours after infection, total cell lysates (INPUT) were subjected to affinity purification with STrEP-Tactin beads (OUTPUT). Both sets of samples were then analyzed by Western blotting with specific antibodies against indicated proteins. Three biological replicates were analyzed. Blots from one representative experiment are shown. (B) The LGP2-specific interaction with PACT persists in the presence of RNase. ST-LGP2 cells and ST-CH cells were infected with MV at an MOI of 1 (+) or were left uninfected (−). Twenty-four hours later, total cell lysates (INPUT) were subjected to affinity purification with STrEP-Tactin beads in the absence or presence of either RNaseA or RNaseIII (OUTPUT). Samples were then analyzed by Western blotting with antibodies against PACT, the STrEP tag, and actin. Two biological replicates were analyzed. Blots from one representative experiment are shown.

We first examined whether the interaction between LGP2 and PACT was RNA dependent. To this end, either the ribonuclease A (RNaseA), which specifically hydrolyzes RNA at C and U sites, or the dsRNA-specific RNaseIII were included in our One-STrEP–tag affinity chromatography protocol for the purification of LGP2-protein cocomplexes. Both mock-infected and MV-infected conditions were examined. In both settings, the LGP2-PACT interaction was maintained in the presence of either RNaseA or RNaseIII (Fig. 3B). In parallel, RNA integrity was controlled by RNA purification from each experimental condition (fig. S2A). We performed an additional control to validate the ability of RNaseIII to digest in vitro–transcribed dsRNA under the same conditions (fig. S2B). These results suggest that the LGP2-PACT interaction is mediated by protein-protein binding because it was unaffected by the presence of either RNaseA or RNaseIII.

Validation of the direct protein-protein interaction between LGP2 and PACT

To validate a direct protein-protein interaction between PACT and LGP2 and to examine PACT binding to other RLRs, we used a protein complementation assay (PCA) in which an interaction between prey and bait reconstitutes the luciferase activity (Fig. 4A) (43). RIG-I, MDA5, and LGP2 full-length proteins, as well as RIG-I and LGP2 CTDs, were tested in the PCA to assess their binding with each other and with full-length PACT (Fig. 4B). We validated the interaction between RIG-I CTD and PACT (Fig. 4C), which is consistent with a previous report (39). We found a higher luciferase score for LGP2-PACT than for RIG-I–PACT, and the regulatory CTD of LGP2 was sufficient to reconstitute the LGP2-PACT interaction (Fig. 4C). Our PCA and proteomic approaches failed to identify a direct interaction between RIG-I and LGP2 and between their corresponding CTDs. We also validated the previously reported interaction between MDA5 and LGP2 (44) and the binding of PACT to MDA5 (45) (Fig. 4C). In addition, the PCA confirmed the efficient homo-oligomerization of PACT (46), MDA5 (47, 48), and LGP2 (10). Our results also highlight the property of LGP2 CTD to form quaternary structures independently from the helicase domain.

Fig. 4 Additional validation of the LGP2-PACT interaction.

(A) Schematic representation of the PCA. Protein A (prey) is fused to the N terminus of the Gaussia luciferase (GL1). Protein B (bait) is fused to the C terminus of the Gaussia luciferase (GL2). If protein A interacts with protein B, the luciferase activity is rescued and the normalized luciferase ratio (NLR) is >3.5. (B) Schematic representation of the organization of the RIG-I, RIG-I CTD (RCTD), LGP2, LGP2 CTD (LCTD), and PACT proteins that were used for the PCA. (C) PACT interacts with the CTD of RIG-I and LGP2. HEK293T cells were cotransfected with the plasmids GL1 and GL2 expressing the indicated proteins, and PCA was performed and luciferase measurements were made. Black columns indicate where protein A was fused to GL1 and protein B was fused to GL2, whereas gray columns indicate where protein A was fused to GL2 and protein B was fused to GL1. Data are means ± SEM of three independent experiments. (D) Binding of PACT protein to RLRs. Left: ELISA plates were coated with PACT and incubated with similar concentrations of MDA5, RIG-I, and LGP2. Cherry (CH) was used as a negative control. Bound RLRs were detected with an antibody against the Strep tag. One representative experiment from a series of four is shown. Nonspecific binding of RLRs to the plates was subtracted. Data are means ± SD of technical duplicates. Right: The indicated samples were analyzed by Western blotting with antibody against the STrEP tag to normalize the amounts of RLRs used for the ELISAs. dsRBD, dsRNA binding domain.

We further confirmed the LGP2-PACT interaction with an enzyme-linked immunosorbent assay (ELISA) with purified PACT and LGP2 proteins. More specifically, all three RLRs were compared for their ability to bind to PACT. Double dilutions of each affinity chromatography purified RLR were incubated with plate-bound recombinant PACT protein expressed in Escherichia coli. Affinity-purified CH protein was used as a negative control. To achieve a similar starting concentration for each RLR, the concentration of the affinity-purified proteins was controlled by Western blotting analysis before each experiment (Fig. 4D). We observed that all three RLRs interacted with PACT with similar efficiencies, whereas, as expected, the negative control CH protein failed to bind to PACT (Fig. 4D).

We then performed an additional PCA to test whether LGP2 affected the PACT–RIG-I and PACT-MDA5 interactions. The PACT–RIG-I and PACT-MDA5 pairs were evaluated in the presence of LGP2 protein. PCA scores demonstrated that the PACT–RIG-I interaction was not affected by the addition of LGP2, whereas the addition of LGP2 to the PACT-MDA5 pair decreased the associated PCA scores (fig. S3, A and B). Together, these results validate that LGP2 interacts with PACT and that this interaction is mediated by the regulatory CTD of LGP2. The data also confirm the previously published interactions of RIG-I and MDA5 with PACT (39, 45).

The LGP2-PACT interaction is involved in the regulation of the RIG-I signaling pathway and a single C615A mutation in LGP2 abolishes this interaction

To assess the relevance of the LGP2-PACT interaction, we searched for point mutations in LGP2 that resulted in a lack of regulatory activity. Pippig et al. (19) previously observed that a single mutation (C615A) in the zinc-binding motif of LGP2 abolished the LGP2 regulatory activity over RIG-I and MDA5 (Fig. 5A). We thus analyzed by PCA the capacity of this mutant LGP2 to interact with PACT. In cell cotransfected with plasmids encoding PACT and with mutated LGP2 or its mutated CTD, the luciferase activity was statistically significantly reduced compared to that in cells expressing wild-type (WT) LGP2 (Fig. 5B). However, the mutated LGP2 CTD was still able to homo-oligomerize, indicating that it can still form quaternary structures and properly fold. To exclude the effect of the C615A single point mutation on the biochemical and, as a result, on the binding properties of the protein, we used PCA to assess the interactions between LGP2 and LGP2-C615A with five other LGP2 protein partners, either previously known or else identified here, such as MAVS, radical S-adenosyl methionine domain-containing protein 2 (RSAD2), 2ʹ-5ʹ-oligoadenylate synthase 2 (OAS2), RNA-binding protein 4 (RBM4), and serrate RNA effector molecule homolog (SRRT) (fig. S4A). Of these, only PACT protein showed lower binding scores with LGP2-C615A that with WT LGP2. By contrast, interactions between LGP2 and MAVS, RSAD2, OAS2, RBM4, or SRRT were not affected by the single point mutation in LGP2.

Fig. 5 The LGP2-PACT interaction is required for modulating RLR signaling activity.

(A) Structure of the human LGP2 CTD (Protein Data Bank: 2w4r). The domain is colored in yellow. The residues C556, C559, and C612 from the zinc finger domain are colored in blue; C615 is colored in red; the loop between C612 and C615 is colored in cyan; and the zinc atom is in gray. (B) The LGP2-C615A mutant fails to interact with PACT. HEK293T cells were cotransfected with the plasmids GL1 and GL2 expressing the indicated proteins, and PCA was performed and luciferase measurements were made. Black columns indicate where protein A was fused to GL1 and protein B was fused to GL2, whereas gray columns indicate where protein A was fused to GL2 and protein B was fused to GL1. Data are means ± SEM of two independent experiments. (C and D) Characterization of the ST-LGP2–C615A cell line. (C) Analysis of the amounts of LGP2 mRNA relative to that of GAPDH mRNA in the indicated cells. Data are from three biological replicates. (D) Analysis of the binding of LGP2-C615A to PACT. Lysates of the indicated cells (INPUT) were subjected to affinity purification with STrEP-Tactin beads (OUTPUT). The samples were then analyzed by Western blotting with antibodies against STrEP tag, PACT, and β-actin. One representative experiment of three biological replicates is shown. (E and F) Comparative analysis of IFN-β promoter responses to RIG-I ligands. The indicated cells were transfected with 20, 100, and 200 ng of 5′3P-RNA (E) or with RNA isolated from MV-infected HeLa cells (F) before being analyzed by luciferase assay. Data are means ± SEM of three independent experiments. (G and H) Comparative analysis of the relative abundances of IFN-β (G) and IFIT1 (H) mRNAs upon infection of the indicated cell lines with MV. Data are means ± SEM of four independent experiments. See table S1 for all details on statistical analysis. Asterisks indicate statistical significance. ns, not significant.

We also addressed the RNA binding activities of LGP2-C615A. We previously demonstrated that defective interfering (DI) genomes of MV are specific RNA ligands for LGP2 and that this RNA-protein interaction is not dependent on RIG-I (49). Using affinity purification approaches, we compared the ability of either LGP2 or LGP2-C615A to bind to DI genomes. Both LGP2 and LGP2-C615A cell lines were infected with a recombinant MV strain known to express a DI genome of 504 nucleotides (nt) in length. The ST-CH stable cell line was used as a negative control to exclude nonspecific binding to RNA. We observed that the LGP2-C615A mutant retained the ability of bind to the 504-nt-long DI genomes produced during MV replication (fig. S4B), albeit with less affinity than did WT LGP2. These data suggest that LGP2-C615A is still capable of binding to RNA. We then generated a stable cell line overexpressing ST-LGP2–C615A in a similar fashion as the ST-LGP2 cell line. The amounts of LGP2 transcripts and protein were similar in the two cell lines (Fig. 5, C and D). When copurifying LGP2-C615A with its cellular partners from ST-LGP2–C615A cells using our pull-down protocol, PACT no longer coimmunoprecipitated with the mutant LGP protein (Fig. 5D).

To better demonstrate the functional role of the LGP2-PACT interaction for the regulation of the RIG-I–mediated type I IFN response, ST-LGP2 and ST-LGP2–C615A cells were transfected with increasing amounts of 5′3P RNA or with total RNA from MV-infected HeLa cells together with the IFN-β reporter plasmid. We found that the inhibitory effect of LGP2 overexpression on IFN-β promoter activity was lost in the ST-LGP2–C615A cells stimulated either by 5′3P or RNAs purified from MV-infected HeLa cells (Fig. 5, E and F). In addition, type I IFN signaling was analyzed by evaluating the mRNA abundances of IFN-β and IFIT1 (an ISG) in total RNA purified from MV-infected cells. These experiments showed that the ST-LGP2–C615A cells lacked an inhibitory effect on the induction of type I IFN signaling, whereas the ST-LGP2 cells failed to exhibit stimulation of the IFN-β signaling cascade when they were infected (Fig. 5, G and H), findings that are consistent with the results from our earlier IFN-β reporter assays (Fig. 5, E and F). Together, these results confirm that the regulatory CTD of LGP2 mediates the LGP2-PACT interaction and that a single mutation in this domain is sufficient to abolish this interaction. These data also suggest that the LGP2-PACT interaction may play a central role in regulating RIG-I signaling.

The LGP2-PACT interaction is required for the ability of LGP2 to synergize with MDA5

To better understand the effect of the LGP2-C615A mutation on MDA5 signaling, we first analyzed by PCA whether the LGP2-MDA5 interaction was abrogated with the C615A mutation. We observed similar normalized luminescence ratio (NLR) scores with MDA5 in cells expressing either WT LGP2 or LGP2-C615A (Fig. 6A). Thus, that mutation does not appear to interfere with the MDA5-LGP2 interaction. We then analyzed the role of the LGP2-PACT interaction in the synergy between LGP2 and MDA5 to stimulate type I IFN signaling. We transfected ST-LGP2 cells with increasing amounts of a PACT-expressing plasmid together with MDA5-specific synthetic or natural ligands (Fig. 6B). We observed an increased IFN-β promoter activity that correlated with increasing amounts of PACT-expressing plasmid, suggesting a role for PACT in the ability of LGP2 to enhance MDA5 responses. Overexpression of PACT in standard HEK293 cells had little or no effect on the immunostimulatory activity of MDA5-specific ligands. Furthermore, transfection with HMW poly(I:C) did not amplify IFN-β promoter activity in ST-LGP2–C615A cells compared to that in ST-LGP2 cells (Fig. 6C).

Fig. 6 The LGP2-PACT interaction drives the synergy between LGP2 and MDA5.

(A) Both WT LGP2 and the LGP2-C615A mutant interact with MDA5. HEK293T cells were cotransfected with the plasmids GL1 and GL2 expressing the indicated proteins, and PCA was performed and luciferase measurements were made. Black columns indicate where protein A was fused to GL1 and protein B was fused to GL2, whereas gray columns indicate where protein A was fused to GL2 and protein B was fused to GL1. Data are means ± SEM of three independent experiments. Samples were analyzed in triplicate. (B) ST-LGP2 cells were transfected with 0, 40, or 100 ng of pCI-neo encoding PACT and 100 ng of total RNA from Mengo- or Cox-infected HeLa cells or 100 ng of HMW poly(I:C) before being analyzed by luciferase assay. Data are means ± SEM of two independent experiments. (C) Comparative analysis of IFN-β promoter responses to MDA5 ligands. The indicated cells were transfected with 20, 100, or 200 ng of HMW poly(I:C) and then analyzed by luciferase assay. Data are means ± SEM of three independent experiments. (D) An ISRE reporter cell line was transfected with pCI-neo plasmids encoding LGP2 (20 ng), RIG-I (20 ng), or MDA5 (20 ng), alone or in the indicated combinations, before promoter activities were measured. Data are means ± SEM of three independent experiments. (E) An ISRE reporter cell line was cotransfected with pCI-neo plasmids encoding MDA5 (25 ng), either WT LGP2 or the LGP2-C615A mutant (0, 5, 25, 50, or 100 ng), and PACT (0, 5, 50, or 100 ng) before promoter activities were measured. Data are means ± SEM of four independent experiments. See table S1 for all details on statistical analysis. Asterisks indicate statistical significance. ns, not significant.

To further characterize the involvement of PACT in the synergy between LGP2 and MDA5, we measured type I IFN stimulatory activity upon transfection of STING-37 cells that stably express the luc gene under the control of a promoter sequence containing five IFN-stimulated response elements (ISREs) with plasmids encoding RIG-I, MDA5, and LGP2 (Fig. 6D) (4). No stimulation with MDA5-specific ligands was required to activate ISRE promoter activity when the cells coexpressed MDA5 and LGP2, thus validating an experimental model for analyzing cooperativity between MDA5 and LGP2 for endogenous dsRNA (Fig. 6D) (10). Last, we observed that the MDA5 signaling activity triggered by the coexpression of MDA5 and LGP2 was additionally amplified in a PACT-dependent manner (Fig. 6E). However, the synergy between MDA5 and LGP2 was lost when the cells expressed the LGP2-C615A mutant, which removed any dependence on PACT for this response. These experiments highlight the importance of the direct protein-protein interaction between LGP2 and PACT for the enhancement of MDA5-mediated immune responses to self and nonself RNAs.

DISCUSSION

The RLR family of innate immune receptors is responsible for triggering the signaling cascade involving the IRF and NF-κB pathways upon viral infection. In contrast to RIG-I and MDA5, LGP2 is unable to bind to the downstream adaptor MAVS through CARD homotypic interactions. LGP2 acts as an inhibitor of RIG-I signaling cascade and an activator of MDA5 signaling (1012, 14, 15, 17). To further understand the mechanism underlying the modulation of these two RLRs by LGP2, we used a cell line overexpressing LGP2 fused to a N-terminal STrEP-tag (Fig. 1A) (12). We observed that transfection of MDA5-specific ligands resulted in an increased activity in ST-LGP2 cells, whereas RIG-I–specific ligands lost their immunostimulatory activity (Fig. 1, B to E). Although LMW poly(I:C) specifically triggers RIG-I, but not MDA5 (26), when it was used to transfect ST-LGP2 cells, the result was an increased immunostimulatory activity. We speculate that upon transfection of LMW poly(I:C) in our cellular model, LGP2 enhanced signaling through MDA5 by increasing its affinity for shorter dsRNA structures (10).

We then identified a set of LGP2 protein partners by affinity chromatography purification of LGP2 in association with specific macromolecular complexes from ST-LGP2 cells, which was followed by LC-MS/MS analysis (Fig. 2A). Upon proteomic data analysis, we focused our efforts on members of the RNA silencing machinery that coimmunoprecipitated with ST-LGP2. Our choice was motivated by several arguments: (i) that the RLR helicase domain is a direct paralog of the DICER helicase domain (50), (ii) that RLR orthologs in nematodes sense viral dsRNA to initiate the antiviral small interfering RNA (siRNA) response (51, 52), and (iii) that the involvement of the RNA-silencing machinery in the mammalian RLR response remains unclear (37).

PACT is the direct paralog of TRBP. Both proteins contain two dsRNA binding domains and serve as cofactors for DICER in the processing and biogenesis of microRNAs (53). The LGP2-TRBP interaction enhances viral RNA sensing by LGP2-MDA5 upon infection with the picornavirus cardiovirus (38). Note that in our study, TRBP-LGP2 binding had no negative effect on RIG-I signaling. Only trace amounts of TRBP were detected by our MS approach, and the protein failed to pass our MS data filtering criteria. Moreover, Western blotting analysis failed to validate an LGP2-specific interaction with TRBP (Fig. 3A). It would be interesting in future studies to compare the relative strengths of the LGP2-PACT and LGP2-TRBP interactions and to understand whether LGP2 can selectively bind to PACT, TRBP, or both to modulate the RLR response.

Increasing evidence links PACT to the host antiviral response. First, overexpression of PACT leads to PKR activation (54). Second, PACT interacts with the CTD of RIG-I and activates its immunostimulatory activity (39). Third, PACT facilitates the RNA-induced formation of MDA5 oligomers (45). Last, different viral proteins have been described to antagonize PACT to subvert the RLR signaling cascade (4042). Therefore, PACT appears as an interesting candidate to be involved in the regulatory effect of LGP2 on the RLR response. Two independent studies have described the LGP2-PACT interaction (55, 56). To validate the protein-binding nature of the LGP2-PACT interaction, we performed affinity purification of LGP2-specific protein complexes in the presence of the ribonucleases RNaseA and RNaseIII (Fig. 3B). The efficiency of RNA cleavage was controlled in a parallel experiment by RNA extraction from the corresponding samples (fig. S2). We observed that the LGP2-PACT interaction was resistant to RNAses. These data are consistent with the findings reported by van der Veen et al. (55).

We used PCA as a second approach to validate the LGP2-PACT protein-protein interaction (43). This method enabled us to determine that PACT interacts with LGP2 through the LGP2 CTD (Fig. 4C). We also confirmed that PACT interacts with the RIG-I CTD (39). We observed high PCA scores for the LGP2-PACT and LGP2 CTD–PACT pairs, suggesting that there is an efficient interaction between LGP2 and PACT. The binding of PACT to LGP2 was further validated by ELISA (Fig. 4D). In the same assay, we observed that all three RLRs bound to PACT with similar efficiencies. Additional PCA experiments showed that LGP2 was unable to perturb the PACT–RIG-I interaction (fig. S3A), whereas the MDA5-PACT interaction was less stable in the presence of LGP2 (fig. S3B). Furthermore, we found that a single mutation (C615A) disrupted LGP2-specific binding to PACT without modifying its ability to bind to MDA5 (Fig. 6A) and only slightly decreased the scores of binding with five other cellular protein partners of LGP2: MAVS, RASD2, OAS2, RBM4, and SRRT (fig. S4A). We also observed that the LGP2-C615A mutant maintained (albeit a decreased) specific RNA-protein interaction with MV DI genomes (fig. S4B). The LGP2-C615A mutant was studied through a panel of approaches to understand the extent of the regulatory activity that the LGP2-PACT interaction exerts over the type I IFN signaling cascade. We next showed that disrupting the LGP2-PACT interaction disrupted the regulatory effect that LGP2 has over RIG-I and MDA5 signaling (Figs. 5 and 6).

We failed to detect an interaction between LGP2 and RIG-I by LC-MS/MS (Fig. 2B) or PCA (Fig. 4C), in contrast with previous reports (9, 15) but in accordance with others (18, 19). Thus, the previously hypothesized mechanism of RIG-I regulation by LGP2 through direct protein-protein interaction is questioned. We propose a model in which the LGP2-PACT complex regulates RIG-I through a PACT–RIG-I interaction. More studies should be performed to uncover the detailed molecular mechanism by which RIG-I is regulated by the LGP2-PACT complex.

Different genetic studies have identified a connection between MDA5 and different autoimmune disorders (57). There is a direct link between MDA5 filament formation on short dsRNA and the MDA5 genotypes linked to these diseases (58). Previous studies showed that cotransfection of cells with plasmid encoding MDA5 and low amounts of LGP2-encoding plasmid can trigger MDA5 signaling activity by increasing its affinity for short dsRNA (10, 27). We observed the same phenotype when we cotransfected the ISRE reporter cell line STING-37 with plasmids encoding MDA5 and LGP2 (Fig. 6D). Note that the presence of increasing amounts of PACT exacerbated this phenomenon only when WT LGP2 was overexpressed and not when we overexpressed the LGP2-C615A mutant, which is unable to interact with PACT (Fig. 6E). Hence, these data suggest that the ability of LGP2 to synergize with the MDA5-specific immune response requires PACT. Our work also suggests the possible involvement of PACT in the sensing of self-RNAs through MDA5.

Whether mammals are capable of eliciting a proper siRNA-mediated antiviral response is a matter of controversy (59). In the nematode Caenorhabditis elegans, RLR orthologs are directly involved in the antiviral siRNA response by recognizing viral dsRNA (51, 52). A previous study described antiviral siRNA in mammals when MAVS signaling is inactive (60). Furthermore, another study suggested that the inefficiency of the RNA interference (RNAi) machinery in antiviral defense in mammalian somatic cells could be, in part, attributed to the inhibition of DICER by LGP2. The exact role of the LGP2-PACT interaction in inhibiting the RNAi machinery remains to be addressed. Rabies virus and influenza A virus infections of transgenic mice overexpressing human LGP2 leads to a reduced inflammatory response while keeping virus replication lower than that in control mice (22, 23). All of these data suggest that LGP2 has an additional role in the antiviral response in mammals independently of IFN signaling. An RLR effector-like function independent of MAVS signaling has also been previously described (61). Given that the RLR helicase domain is a direct paralog of the DICER helicase domain (50) and that there is an interaction between LGP2 and the RNAi machinery, it would be interesting to address whether the effector-like function of RLRs could be orchestrated by an antiviral siRNA response. In conclusion, our study suggests that the LGP2-PACT interaction is fundamental for the regulation of RLR signaling and provides a previously uncharacterized connection between the cellular RNA-silencing machinery and the innate immune response. Our findings reveal the involvement of PACT in the recognition of self-RNAs and thus contribute to the increasing knowledge on the mechanisms leading to autoimmune disorders.

MATERIALS AND METHODS

Cells, plasmids, and recombinant viruses

HEK293 [American Type Culture Collection (ATCC), CRL-1573)], HeLa (human cervix adenocarcinoma epithelial; ATCC, CCL-2), and Vero (fibroblast-like kidney cell from African green monkey; ATCC, CCL-81) cells were maintained in Dulbecco’s modified Eagle medium (DMEM; #61965-026, Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS) ((#A15-101, GE Healthcare) and penicillin (100 U/ml/) and streptomycin (100 μg/ml) (#15140122, Life Technologies). The absence of mycoplasma was regularly checked (#30-1012K, ATCC) in all cell lines. To establish the ST-LGP2–C615A clonal stable cell line, we used a modified pEXPR-IBA105 plasmid carrying a Gateway (GW) cassette (pEXPR-IBA105GW) as previously described (Fig. 1A) (12). The STING-37 cell line corresponded to HEK293 cells stably transfected with an ISRE-luc reporter gene (4). The ST-LGP2 and ST-CH cell lines have been previously described (12). The STING-37, ST-LGP2, ST-CH, and ST-LGP2–C615A cell lines were maintained in DMEM GlutaMAX, supplemented with 10% heat-inactivated FCS, penicillin (100 U/ml), streptomycin (100 μg/ml), and G418 (400 μg/ml; #G8168, Sigma). The sequences encoding LGP2, LCTD, RIG-I, RCTD, MDA5, and PACT were amplified by standard polymerase chain reaction (PCR) from a splenic complementary DNA library (Invitrogen) using specific primers with AttB1 and AttB2 sequences included (table S3). The corresponding DNA fragments were cloned by in vitro recombination into the pDONR207 entry vector (BP reaction, Gateway, Invitrogen). The pDON207 and pDON223 entry vectors encompassing the sequences encoding MAVS, RSAD2, OAS2, RBM4, SRRT, and MX1 originated from the Center for Cancer Systems Biology (CCSB) Human ORFeome Collection Dana-Farber Cancer Institute (DFCI). The attenuated MV Schwarz vaccine strain (MV) has been previously described (62). Recombinant MV-CHIKV (Chikungunya virus), expressing the CHIKV structural proteins (E1, E2, E3, C, and 6K) in additional transcription unit (ATU2) and known to produce a 504-nt DI genome, was used to test LGP2- and LGP2-C615A–specific RNA ligands (49, 63). Mengo [WT Mengo virus, C50UC10 as previously described (64)] and Cox virus B3 (Nancy strain) were provided by N. Escriou (Unité de Génomique Virale et Vaccination, Pasteur Institute, Paris) and G. Moratorio (Institut Pasteur Montevideo, Uruguay), respectively. Virus stocks were produced on Vero cells and titrated by TCID50 (median tissue culture infectious dose).

Directional mutagenesis

The LGP2-C615A and LCTD-C615A mutants were generated by directional mutagenesis using the plasmids pDONR207-LGP2 or pDONR207-LCTD. The forward primer (5′- TGTCATCAGCTGCAGGAACGCGGGGGAGGTCTGGGG-3′) and reverse primer (5′-CCCCAGACCTCCCCCGCGTTCCTGCAGCTGATGACA-3′) were used in accordance with the QuikChange Lightning Multi Site-Directed Mutagenesis Kit protocol (#210518-5 StrataGene). The LGP2-C615A mutant was recombined from pDONR207 to One-STrEP–tag pEXPR-IBA105GW by in vitro recombination (LR reaction).

Synthetic RLR ligands

Short, 5′3P-bearing RNA molecules were obtained from the pCI-neo plasmid linearized by Xba I before transcription. T7 transcription reactions were performed with a T7 RiboMAX Express Large-Scale RNA Production System (#P1320, Promega). Poly(I:C) ligands are LMW (#tlrl-picw, Invivogen) and HMW (#tlrl-pic, Invivogen).

Viral RLR ligands

HeLa cells (5 × 107) were infected at an MOI of 1 for 16 hours with Mengo or CoxB3 viruses and for 24 hours with MV. Cells were washed twice with cold phosphate-buffered saline (PBS) and lysed in 2 ml of lysis buffer [20 mM Mops-KOH (pH 7.4), 120 mM KCl, 0.5% Igepal, 2 mM β-mercaptoethanol]. RNA purification was performed using TRI Reagent LS (#T3934, Sigma). RNA was dissolved in 80 μl of deoxyribonuclease-free and RNase-free ultrapure water. Extracted RNAs were analyzed using NanoVue (GE Healthcare). For picornavirus infection, cells were pretreated with 1 mM ribavirin (#R9644, Sigma).

IFN-β and NF-κB reporter assays

To determine the responsiveness of ST-LGP2 cells to synthetic ligands, the expression of IFN-β and NF-κB was determined by transient transfection of the reporter plasmids pIFNβ-Fluc and pNFκB-Fluc containing the Fluc gene under the control of the IFN-β promoter [IFN-b-pGL3 (65)] or the NF-κB promoter (StrataGene). For RNA transfection analysis, cells were plated in 24-well plates (2 × 105 cells per well). After 24 hours, the cells were transfected with pIFNβ-Fluc or pNFκB-Fluc (250 ng per well), pTK-Rluc plasmid (25 ng per well), which contains a thymidine kinase promoter just upstream of the Rluc gene (Promega), and synthetic/natural RLR RNA ligands using jetPRIME reagent (#114-07, Polyplus). After 24 hours, the cells were lysed, and the Fluc and Rluc activities were measured in cell lysates using the Dual-luciferase Reporter Assay System (#2920, Promega) according to the manufacturer’s instructions. Reporter activity was calculated as a triplicate of the ratio of Fluc activity to the reference Rluc activity.

Antibodies and Western blotting analysis

Protein extracts were resolved by SDS-PAGE on 4 to 12% Criterion gels ((#345-0123, Bio-Rad) with Mops running buffer, and the samples were then transferred to cellulose membranes (GE Healthcare) with the Criterion Blotter system (Bio-Rad). The following antibodies were used: anti–STrEP-tag (#34850, Qiagen), anti-LGP2 (#HPA019570, Sigma), anti-PACT (#sc377103, Santa Cruz Biotechnology), anti-DICER1 (#ab14601, Abcam), anti-TRBP (#ab42018, Abcam), and monoclonal anti–β-actin antibody (A5441, Sigma). Horseradish peroxidase (HRP)–coupled anti-mouse (#NA9310V, GE Healthcare) or anti-rabbit (#RPN4301, GE Healthcare) antibodies were used as secondary antibodies. Peroxidase activity was visualized with an ECL Plus Western Blotting Detection System (#RPN2132, GE Healthcare).

Affinity chromatography of ST-LGP2 and subsequent protein purification

ST-CH, ST-LGP2, and ST-LGP2–C615A cells (5 × 107) were plated and harvested after 24 hours. The cells were washed twice with cold PBS and lysed in 6 ml of lysis buffer [20 mM Mops-KOH (pH 7.4), 120 mM KCl, 0.5% Igepal, and 2 mM β-mercaptoethanol] supplemented with RNasin (200 U/ml; N2515, Promega) and cOmplete Protease Inhibitor Cocktail (#11873580001, Roche). Cell lysates were incubated on ice for 20 min with gentle mixing every 5 min and then clarified by centrifugation at 16,000g for 15 min at 4°C. Cell lysates were incubated for 2 hours on a spinning wheel at 4°C with 200 μl of StrepTactin Sepharose High Performance beads (#28-9355-99, GE Healthcare). For the RNAse treatment assay, RNaseA (#EN0531, Thermo Fisher Scientific) and RNaseIII (#M0245 L, NEB) were used at final concentrations of 0.1 mg/ml and 5 U/ml, respectively. Beads were collected by centrifugation (1600g for 5 min at 4°C) and washed twice for 5 min on a spinning wheel with 5 ml of washing buffer [20 mM Mops-KOH (pH 7.4), 120 mM KCl, and 2 mM β-mercaptoethanol] supplemented with cOmplete Protease Inhibitor Cocktail. Precipitates were eluted using biotin elution buffer (#2-1019-025, IBA). For the MS analysis, protein complexes were precipitated in 12% of trichloroacetic acid (TCA) (#T0699, Sigma) after 16 hours at 4°C. The dsRNA control for RNaseIII digestion encompassing 425 nt of MV genome sequence (L2) was obtained as previously described (66).

LC-MS/MS analysis

Precipitated proteins (1 μg) from ST-LGP2/protein and ST-CH/protein complexes were used in each nano–LC-MS/MS experiment. Proteins were digested overnight at 37°C with 20 μl of a trypsin solution prepared as follows: high-performance LC (HPLC) water, trypsin-sequencing grade (12.5 μg/ml; #PRV5111, Promega), 10% HPLC grade acetonitrile (ACN), and 25 mM ammonium carbonate (#09830, Sigma Aldrich). The resulting peptide extracts were speed-vac–dried and dissolved in 12 μl of HPLC grade water and 0.1% formic acid. Samples (5 μl) were then used for nano–LC-MS/MS analysis using a nanochromatography system (Easy nLC, Proxeon) connected online to a LTQ Velos Orbitrap (Thermo Fisher Scientific) mass spectrometer. A 2-cm-long and 5-μm particle size C18 Easy column (Proxeon) was used for peptide trapping and desalting. A 10-cm-long and 3-μm particle size C18 Easy column (Proxeon) was used for peptide separation. The peptide elution gradient was from 100% buffer A (HPLC grade water and 0.1% formic acid) to 35% buffer B (HPLC grade ACN and 0.1% formic acid) in 60 min using a constant flow of 300 nl/min. MS spectra were acquired on the Orbitrap analyzer at resolution R = 30,000. After each MS spectrum, an automatic selection of the 20 most intense precursor ions was activated with a 15-s dynamic exclusion delay to acquire MS/MS spectra on the LTQ Velos analyzer using CID fragmentation mode at 35% relative resonant activation energy for 40 msec. Raw data were preprocessed using ProteomeDiscoverer version 1.4 (Thermo Fisher Scientific). Preprocessing consisted of MS/MS spectrum averaging of spectra from equal precursor molecular weight, with a 5 parts per million (ppm) mass tolerance and a 90-s time window. No threshold was applied either to MS ions intensities or to MS/MS fragment ion intensities. All spectra were analyzed using Mascot version 2.3 (Matrix Science). Queries were performed against a nonredundant database of 20,352 human protein sequences from Swiss-Prot (release 2013-02). Mascot was run in MS/MS Ion search mode with the following parameter settings: no fixed modification, variable modification [oxidation (Met), phosphorylation (Ser, Thr, and Tyr), acetylation (Lys and N-term), deamidation (Asn and Gln)], precursor mass tolerance of 7 ppm, fragment ions mass tolerance of 0.5 Da, two missed cleavages, and trypsin as the digestion enzyme. False discovery rates were calculated for each peptide by the percolator node for validation of peptide identification. To reduce the number of nonspecific hits, if the protein hit presented a score of >20 in any of the three ST-CH replicates, the protein was immediately excluded from the list. A second filter was added whereby the protein had to have a score of >60 for two of the three replicates and could have a score of >40 for the third replicate (fig. S1A).

LGP2 interaction network visualization

For GO term enrichment analysis of biological process, cellular compartment, and molecular function, the list was analyzed in DAVID (34, 35). GO terms with P values of 0.01 or less were selected. Furthermore, the LGP2 interaction network was represented on Cytoscape (36).

Protein complementation assay

Open reading frame (ORF) sequences corresponding to LGP2, LCTD, RIG-I, RCTD, and PACT were recombined from the corresponding pDONR207 plasmids into pGL1 and pGL2 vectors so that they could be expressed in fusions with fragments 1 to 93 (GL1) and 94 to 169 (GL2) of Gaussia princeps luciferase as previously described (43). HEK293T cells were seeded in 24-well plates at a concentration of 2 × 105 cells per well. After 24 hours, cells were transfected with 100 ng of each of the Gluc1 and Gluc2 plasmid constructs. At 24 hours after transfection, the cells were harvested with 100 μl of Renilla lysis buffer (#E2820, Promega) for 30 min. Gaussia princeps luciferase enzymatic activity was measured using a Berthold Centro XS LB960 luminometer by injecting 50 μl of Renilla luciferase assay reagent (#E2820, Promega) into 10 μl of clarified cell lysates and counting the luminescence for 10 s. For PCA analysis in the presence of LGP2, 200 ng corresponded to the 1:1:2 mass concentration ratio (RIG-I or MDA5:PACT:LGP2) that was used for each transfection. The pCI-neo empty vector was used as a control. Results were represented as the fold-change normalized over the sum of the controls, specified herein as the NLR. For a given protein pair A-B, the luminescence activity of cells transfected with the GL1-A and GL2-B vectors (AB) was divided by the sum of the luminescence activity for control wells transfected with GL1-A and empty vector expressing the GL2 fragment alone (CtrlA) and with empty vector expressing the GL1 fragment alone and GL2-B (CtrlB). Thus, NLR = AB/(CtrlA + CtrlB). For each interaction, the final result was calculated from the mean NLR of at least triplicate experiments.

RT-qPCR analysis of gene expression

The analysis was performed using Applied Biosystem’s StepOnePlusTM technology. The mRNA expression profiles in ST-LGP2 cells and ST-LGP2–C615A cells were determined by quantitative reverse transcription polymerase chain reaction (RT-qPCR) analysis of total RNA isolated with the RNeasy Mini Kit (#72012, Qiagen). Reactions were performed with 100 ng of RNA using TaqMan RNA-to-Ct 1-Step Kit (#4392938, Thermo Fisher Scientific) and 1 μl of custom TaqMan Gene Expression Assays (Hs00225561_m1 for LGP2 and Hs99999905_m1 for GAPDH, Life Technologies). For IFN-β or IFIT1 gene expression assays, total RNA was purified 24 hours after infection with MV. The relative amounts of IFN-β or IFIT1 mRNAs were quantified by analogous one-step real-time PCR using GAPDH mRNA abundance as the internal control (IFN-β #Hs01077958_s1 and IFIT1 #Hs00356631_s1, Life Technologies). All of the measurements were performed in triplicate and analyzed as generated by the StepOnePlusTM software system. Results were normalized relative to the amount of GAPDH mRNA, and the amount of RLR mRNA in HEK293 cells was set to 1. All of the measurements were performed in triplicate and analyzed with StepOnePlusTM software.

ISRE reporter cells and MDA5/LGP2 cooperativity

To determine the immunostimulatory activity when cotransfecting cells with plasmids encoding LGP2 and MDA5, ISRE reporter cell lines (STING-37 cells) were plated 1 day before transfection in 24-well plates (2 × 105 cells per well). Plasmids (pCI-neo) encoding LGP2, or LGP2-C615A, or MDA5, or PACT were used in transfections separately or simultaneously using jetPRIME. All experiments were performed with a total of 250 ng of DNA. Twenty-four hours after transfection, the cells were lysed with passive lysis buffer (#E1941, Promega), and the Fluc activity was measured with the Bright-Glo Luciferase Assay System (#E2638, Promega).

ELISA analysis of the binding of MDA5, RIG-I, and LGP2 to PACT

BL21-CodonPlus (DE3) cells transformed with pET15b (His-PACT) were used to express and purify PACT protein as previously described (67). Purified PACT (1 μg/ml) protein in 1X Dulbecco’s PBS buffer (DPBS, Gibco) was bound to 96-well ELISA plates (200 ng per well) at 4°C overnight. The wells were washed with PBS containing 0.1% Tween 20 (PBS-T) three times at room temperature. The wells were then blocked with PBS-T containing 3% nonfat dry milk for 1 hour at room temperature and washed three times with PBS-T. Double dilutions of STrEP-tag affinity purified MDA5, RIG-I, LGP2, and the negative control CH proteins were prepared in DPBS. The quantities of purified RLRs were controlled by Western blotting analysis. Each dilution of the RLR (50 μl) was bound for 1 hour at 37°C to the wells coated with PACT and to the control uncoated wells. The latter were used to measure the nonspecific binding of each RLR. After the wells were washed, bound RLRs were detected with mouse anti–STrEP-tag (#34850, Qiagen), followed by HRP-conjugated goat anti-mouse IgG and (#115-035-146, Jackson ImmunoResearch). Colorimetric HRP substrate was added, and the amounts of bound RLRs were quantified by absorbance at 450 nm normalized to 620 nm.

Affinity chromatography of LGP2 and LGP2-C615A ribonucleoprotein complexes, subsequent RNA purification, and RT-qPCR analysis

A previously described and validated protocol for the infection of ST-LGP2 and ST-LGP2–C615A cells with recombinant MV-CHIKV followed by STrEP-protein purification was applied (49). RNA extraction was performed directly from RNA-protein complexes eluted upon StrepTactin Sepharose affinity purification using TRI Reagent LS. RNA was extracted with the TRI Reagent LS before or after affinity chromatography purification on StrepTactin Sepharose. The 504-nt MV DI genome probe and primers were previously described (49). The extracted RNAs were analyzed using the nanodrop and Bioanalyzer RNA nano Kit before qPCR analysis. MV DI RNA RT-qPCR analysis was performed using Applied Biosystem StepOnePlu technology. Reactions were performed with 10 ng of RNA using a TaqMan RNA-to-Ct 1-Step Kit for one-step RT-qPCR analyses. Reactions were performed in a final volume of 20 μl in the presence of 100 nM TaqMan MV DI-RNA probe and 250 nM DI-RNA forward (and reverse) primers.

Statistical analysis

The R environment was used for all of the analyses (68). Triplicate or quadruplicate values in each experimental condition were averaged. The data were then adjusted such that each independent experiment presented the same mean for HEK293 Mock (Fig. 1, B and C), HEK293 Mock (20 ng of RNA) (Fig. 1E), MV UV (Fig. 1F), RIG-I–PACT A/B (Fig. 4C), LGP2 A/B (Fig. 5B), HEK293 (20 ng RNA) (Fig. 5, E and F, and Fig. 6C), HEK293 Mock (Fig. 5, G and H), mCherry (Fig. 6D), PACT (0 ng), LGP2 (0 ng) (Fig. 6E), RIG-I–PACT Mock A/B (fig.S3, A and B), and SRRT LGP2 WT (fig. S4A). Data were also log2 converted before statistical analysis was performed. Statistical significance was set at P ≤ 0.05. In each figure, type I error was controlled for by correcting the P values according to the Benjamini & Hochberg method [“BH” option in the p.adjust() function of R]. Data were analyzed using a linear model to take into account any potential interaction effects between the covariates. Two-by-two comparisons were performed using the contrast() function of the contrast package. The results are summarized in table S1.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/12/601/eaar3993/DC1

Fig. S1. Proteomics approach to study LGP2-specific protein partners.

Fig. S2. Assessment of RNA integrity within LGP2-protein complexes upon purification in the presence of RNaseA and RNaseIII.

Fig. S3. PCA analysis of the binding of RIG-I and MDA5 to PACT in the presence of LGP2.

Fig. S4. PCA analysis of the binding of the LGP2-C615A mutant with a known protein and with RNA binding partners of LGP2.

Table S1. Statistical analyses.

Table S2. List of direct and indirect LGP2-specific cellular partners identified by MS.

Table S3. List of primers.

REFERENCES AND NOTES

Acknowledgments: We thank A. Zhukova (Evolutionary Bioinformatics/Bioinformatics and Biostatistics Hub, C3BI, Institut Pasteur) and F. Gwinner (Systems Biology Laboratory, Institut Pasteur, Paris) for their help with bioinformatics analysis of the MS data. We thank N. Escriou (Unité de Génomique Virale et Vaccination) for the mengovirus stock, E. Meurs (Unité de Hépacivirus et immunité innée, Institut Pasteur) for the antibody against PKR, G. Moratorio (Institut Pasteur Montevideo) for the coxsackie virus stock, the CCSB DFCI for providing clones from the Human ORFeome Collection, and M. Nowakowski (Platform of Recombinant Proteins in Prokaryotic Cells, Institut Pasteur) for PACT protein purification. We thank N. Vabret for critical reading of the manuscript. We acknowledge all the members of the Unité de Génomique Virale et Vaccination at Pasteur Institute, and especially N. Escriou, for their help and useful discussions. Funding: This study was supported by the Institut Pasteur and the CNRS. R.Y.S.D. was supported by the Fondation pour la Recherche Médicale (FDT20140931129). We thank the Region Ile-de-France (SESAME), the Paris-Diderot University (ARS), and CNRS for funding part of the LC-MS/MS equipment of the proteomics core facility at the Institut Jacques-Monod. Author contributions: R.Y.S.D., C.C., and A.V.K. designed the experiments. R.Y.S.D., C.C., and V.N. performed the experiments under the supervision of A.V.K. and F.T. G.A.M. performed independent statistical analysis of all the data. T.L. and J.-M.C. performed MS analysis. B.S. helped with Cytoscape and Y.J. helped with the PCA analyses. J.B. supervised the expression and purification of PACT protein. G.B. and N.J. contributed to the ELISA and PCA data analyses and discussion of the results. R.Y.S.D., F.T., and A.V.K. conceived and wrote the manuscript, whereas all authors endorse its content. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The complete MS datasets are available in the PRIDE partner repository (68) under the identification no. PXD013191 as .raw files and associated pep.xml and .xlsx files. All other data needed to evaluate the conclusions in the paper are present in the paper and the Supplementary Materials.
View Abstract

Navigate This Article