MDA5 guards against infection by surveying cellular RNA homeostasis

MDA5 guards against infection by surveying cellular RNA homeostasis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article MDA5 guards against infection by surveying cellular RNA homeostasis Natalia G. Sampaio, Linden J. Gearing, Antonio G. Dias Junior, and 8 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6466919/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract MDA5 is an innate immune RNA sensor that senses infection with a range of viruses and other pathogens. MDA5’s RNA agonists are not well defined. We used single-nucleotide resolution crosslinking and immunoprecipitation (iCLIP) to study its ligands. Surprisingly, upon infection with SARS-CoV-2 or encephalomyocarditis virus, MDA5 bound overwhelmingly to cellular RNAs. Many binding sites were intronic and proximal to Alu elements and to potentially base-paired structures. Concomitantly, cytoplasmic levels of intron-containing unspliced transcripts increased in infected cells and displayed enrichment of MDA5 iCLIP peaks. Moreover, overexpression of a splicing factor abrogated MDA5 activation. Finally, when depleted of viral sequences, RNA extracted from infected cells still stimulated MDA5. Taken together, MDA5 surveys RNA processing fidelity and detects infections by sensing perturbations of posttranscriptional events such as splicing, establishing a paradigm of innate immune ‘guarding’ for RNA sensors. Immunology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Pathogens are first detected by pattern recognition receptors (PRRs) that activate immune signal transduction pathways and thereby mediate the host response to infection. Canonically, pathogen-derived molecules known as pathogen-associated molecular patterns (PAMPs) activate PRRs 1 . Viral infections are recognised by a class of PRRs that detect unusual nucleic acids 2 . A crucial response activated by nucleic acid sensors is the production of type I interferons (T1-IFNs) 3 . These cytokines are essential for protection against all viruses. This system is fine-tuned to differentiate infection from the homeostatic state, to elicit protective immunity followed by return to homeostasis. Excessive or uncontrolled T1-IFN responses fail to protect against infection and lead to long-term damage and disease 4 , 5 . Characterising the initiation of T1-IFN responses is therefore fundamental to our ability to understand infectious and other diseases, and to design antiviral therapies. Amongst the PRRs that detect viral infection are the retinoic acid-inducible gene I (RIG-I)-like receptors (RLRs) 6 . This family of RNA sensors includes RIG-I, melanoma differentiation-associated protein 5 (MDA5) and laboratory of genetics and physiology 2 (LGP2). RLRs are primarily located in the cytoplasm. All RLRs have a central helicase domain and a carboxy-terminal domain, which together detect unusual, immunostimulatory RNA molecules. MDA5 and RIG-I also contain two tandem amino-terminal caspase activation and recruitment domains (CARDs), which mediate downstream signalling through interaction with the adaptor mitochondrial antiviral-signalling protein (MAVS). Activated MAVS oligomerises and recruits other factors including the kinase TBK1, that in turn activate interferon regulatory factors (IRFs) and the NF-kB pathway. Ultimately, expression of T1-IFNs and other immune response genes is induced. LGP2 lacks CARDs and modulates MDA5 and RIG-I signalling. RIG-I and MDA5 are activated by different viral infections 6 . For example, influenza A virus infection is sensed by RIG-I; conversely, picornaviruses such as encephalomyocarditis virus (EMCV) or rhinovirus are detected by MDA5 7 . Viruses from other families are recognised by both RIG-I and MDA5, including important human pathogens such as members of the Flaviviridae (e.g. Zika virus, hepatitis C virus), Paramyxoviridae (e.g. measles virus) and Coronaviridae (e.g. SARS-CoV) 8 . SARS-CoV-2 infection has been reported to be partially or exclusively sensed by MDA5 9–12 . Additionally, MDA5 can be activated by infection with non-RNA viruses 13 – 15 and other pathogens such as Plasmodium sp. 16 , 17 , Mycobacterium tuberculosis 18 or Aspergillus fumigatus 19 . The important role of MDA5 in the human immune response is highlighted by cases of inherited MDA5 deficiency leading to increased and/or life-threatening susceptibility to viral infections, for example with Rhinovirus 20 – 23 . An important aspect of RLR signalling is the ability of these receptors to distinguish immunostimulatory RNAs accumulating in infected cells from the RNA content of cells during homeostasis. RIG-I is activated by RNAs with triphosphate or diphosphate groups at the 5’ end 24 – 26 . Additional features of RIG-I-stimulatory RNAs include the lack of methylation marks and base-pairing 27 , 28 . Such RNA species do not occur abundantly in cells, as most cellular RNAs are processed at the 5’ end; for example, mRNAs are capped and methylated. In contrast, some viral RNAs such as the genomes of influenza A virus contain these features, allowing them to be recognised by RIG-I 29 . As such, RNA sensing by RIG-I can be conceptualised by the paradigm of PAMP detection by PRRs. It is noteworthy that some DNA viruses and retroviruses that do not produce viral 5’-(P)PP-RNAs activate RIG-I indirectly by causing the accumulation of unprocessed and/or mislocalised cellular non-coding RNAs with 5’-PPP moieties 30 – 33 . The RNA species that activate MDA5 are less well defined 8 . Early work showed that MDA5 recognises infection with picornaviruses and the synthetic double-stranded (ds) RNA mimic polyriboinosinic:polyribocytidylic acid (Poly(I:C)) 7 , 34 . In particular, MDA5 is essential for innate immune sensing of long Poly(I:C) 35 . Moreover, purified MDA5 forms multimeric filaments on long strands of dsRNA 36 . RNA viruses with positive-sense genomes, including picornaviruses, replicate by producing a negative-sense copy of their genome that serves as a template for synthesis of progeny positive-sense genomes. Annealing of positive and negative sense RNAs can form long dsRNA, known as replicative form dsRNA, which was reported to activate MDA5 37, 38 . Together, these results indicate that, in virally infected cells, MDA5 is activated by long viral dsRNA. However, other findings contradict this view, suggestive of a more complex mechanism. Indeed, we previously demonstrated that high molecular weight complex RNA produced during infection, rather than merely dsRNA, activates MDA5 39 . Moreover, studies analysing the RNAs bound by MDA5 or LGP2 during viral infections suggested that single stranded sections of viral genomes, rather than dsRNA forms, are detected 40 – 42 . However, these studies are constrained by methodological factors, including non-stringent purification of MDA5-RNA complexes, lack of precision in determining MDA5 binding sites, and/or exogenous protein overexpression. Given that MDA5 can sense infection with many different virus families and other pathogens, understanding the nature of MDA5’s RNA agonists is crucial for our comprehension of innate immune responses. We addressed this by employing individual-nucleotide resolution ultraviolet crosslinking and immunoprecipitation (iCLIP) 43 , 44 to stringently and specifically identify RNAs bound by MDA5 during virus infection. We revealed the in situ RNA targets and the precise protein-RNA binding sites of endogenous MDA5 in EMCV- and in SARS-CoV2-infected cells. To our surprise, we found that MDA5 bound mostly to host-derived RNA during virus infection. This host RNA was largely intronic, and close to Alu repetitive elements. EMCV and SARS-CoV-2 infections resulted in the presence of intron-containing pre-mRNA in the cytoplasm and MDA5 binding sites determined with iCLIP were enriched amongst these intronic RNA sequences. Finally, overexpression of the splicing factor SRSF3 during virus infection prevented MDA5 activation, whereas removal of viral RNA from infected cellular RNA did not. This suggests that MDA5 detects imbalances in RNA processing occurring in virus-infected cells by binding to endogenous host intronic RNA accumulating in the cytoplasm. Results Dual IP of endogenous MDA5 isolates MDA5-bound RNA Previous studies of RNAs binding to and/or activating MDA5 relied on recombinant or overexpressed protein 42, 45, 46 . To identify RNA ligands of endogenous MDA5, we screened cell lines for expression of MDA5 and found that monocytic THP1 cells expressed MDA5 protein at baseline (Fig. S1a). EMCV specifically activates MDA5 7, 47 and infects THP1 cells, evident from staining with the J2 monoclonal antibody that detects dsRNA (Fig. S1b), a signature of viral replication 39, 48, 49 , and from accumulation of viral RNA (Fig. S1c). Infection induced IFN b mRNA and interferon-stimulated genes (ISGs) (Fig. S1d). Transfection of total cellular RNA extracted from EMCV-infected THP1 cells activated IFNB1 promoter-driven luciferase expression in HEK293 reporter cells 50 in an MDA5-dependent manner (Fig. S1e). Together, these data show that MDA5-stimulatory RNA was present in EMCV-infected THP1 cells. To obtain an in-depth view of RNAs binding MDA5 during viral infection, we employed iCLIP 43 . This technique allows deep sequencing of RNAs bound by a protein of interest, with resolution of the nucleotide site(s) where the protein binds, and has been successfully used for other dsRNA binding proteins including Dicer 51 and DDX17 52 . As a control, MDA5 knock-out (KO) THP1 cells were generated using CRISPR/Cas9 (Fig. S1a, f). MDA5-KO THP1 did not induce IFNB1 mRNA in response to transfection with total RNA extracted from EMCV-infected HeLa cells, but responded normally to RIG-I stimulation with in vitro transcribed RNA 29 (Fig. S1g). In MDA5-KO cells, T1-IFN responses to EMCV infection were undetectable (Fig. S1d). Furthermore, we surmised that high infectivity levels would be required to detect RNA bound to MDA5. However, EMCV infectivity was low in wild-type (WT) THP1 cells (Fig. S1h). IRF3 signals downstream of multiple nucleic acid sensors including cGAS that is required for baseline T1-IFN responses in THP1 cells 53 . Loss of IRF3 may therefore render cells more susceptible to EMCV without impairing MDA5-RNA interactions. Indeed, infectivity levels increased ten-fold in IRF3-KO THP1 cells 50 (Fig. S1i-k). We therefore used IRF3-KO THP1 cells to investigate MDA5 ligands generated during EMCV infection. To isolate endogenous MDA5, we raised a panel of monoclonal antibodies. We identified two antibodies that precipitated native and denatured MDA5, termed antibody A (clone 16) and antibody B (clone 22), respectively. To increase the stringency and specificity of MDA5 isolation, we used a dual immunoprecipitation (IP) method. Native MDA5 was first isolated using antibody A, and then eluted with a denaturing solution containing high urea and detergent concentrations. Next, the eluate containing denatured protein was used for a second IP with antibody B, which was specific for the denatured protein (Fig. 1a). This dual IP successfully pulled down MDA5 from IRF3-KO cells (Fig. 1b). Next, we performed MDA5 dual IP on lysates from UV-crosslinked cells, and radioactively labelled the bound RNA using Polynucleotide kinase (PNK). Gel electrophoresis and radioblot analysis showed a high molecular weight (MW) smear migrating more slowly than the MW of MDA5 (135 kDa), which was dependent on both UV crosslinking and PNK treatment (Fig. 1c). The intensity of this smear was enhanced by virus infection, indicating increased RNA binding to MDA5 in infected cells (Fig. 1c). RNase A treatment reduced the smear, presumably due to the bound RNA being partially degraded to smaller oligonucleotides protected by the protein (Fig. 1c, d). Extraction and proteinase K digestion of either high MW or low MW sections from the membrane produced RNA of varying lengths, with intermediate and low RNase A concentrations yielding RNA of more than 100 nucleotides (Fig. 1e). We selected the low RNase A treatment condition to capture MDA5-bound RNA in the 100-500 nucleotide length range, which is suitable for iCLIP and sequencing 43 . We then applied our dual IP approach to a modified iCLIP protocol 43 (Fig. 1f). An adaptor for reverse transcription was added to RNA at the 3’ end, and the RNA was radiolabelled at the 5’ end, followed by isolation of high MW RNA by gel electrophoresis and radioblot as described above. An aliquot of pre-IP input material was processed alongside the IP samples and served as a ‘size-matched input’ control for comparison to IP samples 54 . Next, the RNA was reverse transcribed into cDNA using primers containing a sample-specific barcode for multiplexing and a unique molecular identifier (UMI) for identification of PCR duplicates. Of note, reverse transcription is halted at the RNA-protein crosslink site, which allows identification of nucleotide positions bound by MDA5. The cDNA was purified and size-selected by electrophoresis, circularised and cleaved to produce cDNA containing forward and reverse PCR sites for amplification and sequencing (Fig. 1f). MDA5 binds host-derived RNA during EMCV infection We employed our endogenous MDA5 iCLIP method to identify RNAs specifically bound by MDA5 during virus infection. IRF3-KO THP1 cells were left uninfected or were infected with EMCV. As a negative control for IP specificity, MDA5-KO THP1 cells were used. EMCV infectivity was lower in MDA5-KO cells compared to IRF3-KO cells (henceforth referred to as MDA5-WT) (Fig. S1i-k), likely due to baseline T1-IFN production via cGAS-IRF3 53 , priming cells against infection. Nevertheless, these cells still contained viral RNA (Fig. 2a), and thus served as a control for MDA5 IP specificity. Sequencing reads were processed and mapped to combined human and virus genomes to simultaneously identify both viral and host sequences, followed by removal of PCR duplicates and extraction of crosslink sites (Fig. S2a). We obtained 10 4 -10 6 uniquely mapped, deduplicated reads per sample with four independent biological repeats (Fig. S2b, c). To our surprise, we found that the majority (>99%) of RNA sequences detected in iCLIP samples mapped to the human rather than the viral genome (Fig. 2a). There was no enrichment of EMCV RNA in the MDA5 IP compared to the RNA input control samples. We next identified MDA5-RNA crosslink sites, herein referred to as MDA5 binding sites, using PureCLIP 55 that compares input and IP sequencing reads. MDA5 binding sites were only identified in host RNA, and not in viral RNA. The distribution of MDA5 binding sites between intergenic regions, introns, untranslated regions and coding sequences was similar between the three samples, and more than half of all binding sites were in introns (Fig. 2b). We then employed HOMER 56 and MEME-ChIP 57 to identify motifs enriched in MDA5-bound RNA. Based on MDA5’s footprint on dsRNA of 14-15 nt 58 and the notion of MDA5 filament formation 36 , we focussed on 101 nucleotide-length regions containing MDA5 binding sites centrally (i.e., 50 nt upstream and downstream of the crosslinked nucleotide). Both algorithms identified an enrichment of Poly(A) and Poly(U) sequences in MDA5-binding sequences in EMCV-infected cells (Fig. 2c, Fig. S3a). These Poly(A)/(U) motifs were positioned near, but not directly overlapping, the MDA5 crosslinked nucleotide (Fig. 2d). Between 30% and 50% of all MDA5 binding sequences contained either a Poly(A) or a Poly(U) motif, depending on the control and motif algorithm applied (Fig. 2e, f). Importantly, these motifs were enriched compared to uninfected, MDA5-sufficient cells and to infected, MDA5-deficient cells, indicative of specificity. Additionally, motifs that contained GGUU and AACC sequences were specifically enriched in MDA5 binding sites from uninfected WT cells when compared to the infected MDA5-KO control (Fig. 2g, h, Fig S3b). These motifs were detected in 40-50% of binding sites in uninfected samples (Fig. 2i, Fig. S3c). The GGUU/AACC motifs were also detected in MDA5 binding sites from EMCV-infected WT samples (Fig. 2f, Fig. S3c), but were not enriched when compared to MDA5 binding sites from uninfected WT cells. This indicates that the GGUU/AACC motifs present in the infected cells likely represent RNA derived from uninfected cells within the infected cell population; these motifs may thus represent low affinity and/or low abundance MDA5 ligands. Synthetic host RNAs encompassing MDA5-binding sites activate MDA5 in cellulo To validate that host RNAs containing MDA5 binding sites activate MDA5 in cells, we generated by in vitro transcription (IVT) ~500 nucleotide length RNAs from forward and reverse genomic sequences. These contained selected top-scoring sites centrally (Tab. S1), were annealed to form dsRNAs (Fig. S4a, b) and were then transfected into RIG-I-deficient cells to prevent RIG-I stimulation by 5’-triphosphate groups on RNAs produced by IVT. In both Huh7 cells and mouse embryonic fibroblasts (MEFs), a range of T1-IFN and ISG mRNA induction was observed, suggesting that some regions were more stimulatory than others (Fig. S5a-d). For example, regions K and L were largely inert, whereas regions M and N induced strong T1-IFN and ISG responses (Fig. 3a, b). Interestingly, unlike K and L, regions M and N encompassed repetitive sequences (LINE1 and MIR, respectively). Of the 16 regions tested, ten contained repetitive sequences (Tab. S1) and the corresponding IVT-dsRNAs tended to induce stronger T1-IFN responses than the remaining IVT-dsRNAs (Fig. S5c, d). We validated the immunostimulatory potential of M and N dsRNAs at protein level by showing that both induced expression of ISG15 and promoted ISGylation (Fig. 3c). Furthermore, RIG-I/MDA5 double-knockout reporter cells did not respond to any IVT-dsRNA tested, demonstrating that the response was MDA5-dependent (Fig. 3d and S5e-g). Finally, transfection of IVT-ssRNAs did not stimulate cells (Fig. S5h). Taken together, some of the host RNA regions that bound MDA5 upon viral infection, particularly those containing repetitive sequences, induced an MDA5-dependent T1-IFN response when re-introduced as dsRNAs into uninfected cells, validating our iCLIP data. Depletion of viral RNA does not diminish MDA5 activation Transfection of total RNA extracted from cells infected with MDA5-stimulating viruses induces MDA5 activation (Fig. S1e, g and 41, 50 ). To further test the contributions of viral and host RNAs to the activation of MDA5, we extracted total RNA from EMCV-infected THP1 cells and used biotinylated antisense probes to deplete positive sense EMCV RNA by a streptavidin pulldown. This resulted in approximately 1000-fold depletion of viral RNA (Fig. 3e). Input and depleted RNA samples were then transfected into wild-type or MDA5-KO A549 cells 59 . These recipient cells were pre-treated with ribavirin to preclude EMCV replication 39 . As expected, viral RNA was not present in cells transfected with depleted RNA samples (Fig. 3f). Nonetheless, we found that depletion of EMCV positive sense RNA did not prevent the activation of MDA5 in recipient cells, as shown by an equal induction of the T1-IFN response by both input and depleted RNA samples (Fig. 3g-j). This response was dependent on MDA5, since MDA5-KO cells did not respond. Transfection of RNA from uninfected cells did not induce a T1-IFN response. It is noteworthy that our antisense probes were complementary to positive sense EMCV RNA and that the RT-qPCR used to assess depletion detected both negative and positive sense EMCV RNA. Therefore, it is possible that small amounts of negative sense viral RNA, potentially paired with the remaining trace amounts of positive sense EMCV RNA, were present after depletion. Nonetheless, we favour the interpretation that endogenous host RNA generated during virus infection, rather than viral RNA, stimulates MDA5. This endogenous host RNA likely arises from improperly spliced RNA accumulating as a result of virus infection. Cellular RNAs associate with MDA5 during SARS-CoV-2 infection We and others previously reported that SARS-CoV-2 infection activates MDA5 9-12 . To identify MDA5 agonists during SARS-CoV-2 infection, we used Calu-3 cells, an adenocarcinoma-derived lung epithelial cell line. We performed iCLIP with dual MDA5 IP as described above with some methodological improvements to enrich for bound RNA (detailed in methods). We obtained ~10 7 uniquely mapped, deduplicated reads per sample (SARS-CoV-2 infected or uninfected cells; two biological replicates) (Fig. S6a, b). Approximately 10% of RNA sequences in the SARS-CoV-2-infected RNA input samples mapped to the SARS-CoV-2 genome (Fig. 4a). However, there was no enrichment in SARS-CoV-2 RNA in the MDA5 IP samples (Fig. 4a). We did not detect MDA5 binding sites in viral RNA; instead, MDA5 bound only host RNA. Over 50% of binding sites were found in introns (Fig. 4b), followed by intergenic regions (~25%) and protein coding regions (~10%). Given that MDA5 has been previously suggested to bind repetitive RNA 46 , we determined the proportion of binding sites present in repetitive regions. Around 30% of binding sites were located within repetitive regions, which included SINEs ( Alu , MIR), LINEs (L1, L2) and other repeat categories (Fig. 4c). We then extended our analysis to 101 nt regions spanning binding sites and found that ~50% of these sites overlapped with an Alu repeat, without major increases in any of the other repeat categories. Thus, in agreement with the findings for EMCV infection, we observed that during SARS-CoV-2 infection, MDA5 binds host, rather than viral, RNA. This host RNA is mostly intronic and enriched in Alu repeats close to the MDA5-binding site. We next analysed the same 101 nucleotide regions containing MDA5 binding sites using motif-finding algorithms. Poly(U) and Poly(A) motifs were enriched in SARS-CoV-2-infected samples (Fig. 4d-f, Fig. S3d-f). These motifs were approximately 10 nucleotides away from MDA5 binding sites (Fig. 4e). Motifs containing AACC or GGUU sequences were identified in both uninfected and SARS-CoV-2-infected samples at a distance of approximately 10 nucleotides from MDA5 binding sites (Fig. 4f-h, Fig. S3d-f). However, these were not enriched when using uninfected binding sites as a ‘background control’ for binding sites from infected cells. Infected samples contained both infected and uninfected cells (Fig. S6c, d), suggesting that MDA5 binds AACC/GGUU motifs in uninfected cells. In sum, similar sequence motifs were enriched near MDA5 binding sites in both the SARS-CoV-2 and EMCV datasets. Since base-pairing – resulting in the formation of ‘dsRNA-like’ stretches – can occur between Poly(A) and Poly(U) motifs, we analysed the proximity between identified motifs, to determine whether inter- or intra-strand base-pairing was possible. We found that of the 4,579 Poly(A) and Poly(U) motifs identified by MEME-ChIP in the SARS-CoV-2-infected samples, 259 (5.7%) occurred within 200 nucleotides of another complementary motif, either on the same or opposite strand. Of 869 AACC/GGUU motifs found in the uninfected samples, 154 (18%) occurred within 200 nucleotides of another complementary motif on the opposite strand, whereas only 2 were close to a complementary motif on the same strand, suggesting inter-strand complementarity. We also investigated the potential for bidirectional transcription that may have occurred at MDA5 binding sites, potentially generating dsRNA species. We defined ‘overlapping’ binding sites as any two binding sites within 100 nt of each other on opposite strands and compared this to randomly generated sites across the genome. There was no difference in the number of overlapping binding sites in uninfected cells compared to random control, but 0.4% of binding sites from SARS-CoV-2 infected cells were within 100 nt of another site on the opposite strand, compared to 0.1% in randomly generated sites, suggesting a small proportion of MDA5 binding sites in infected cells may have been generated from bidirectional transcription (Fig. 4i). Interestingly, when we further investigated these closely occurring MDA5 binding sites, we found that 41 of 90 (45%) occurred between 10-15 nt from each other (Fig. 4j). Given the ‘footprint’ of MDA5 on dsRNA is estimated to be 14-15 nt 58 , we speculate that these binding sites could represent nucleotides crosslinked at both edges of a single MDA5 protein bound to a segment of dsRNA. Combined, our analysis of the RNAs bound by MDA5 during SARS-CoV-2 infection found enrichment in sequences with the potential to form intra- and inter-strand dsRNA structures, which may constitute cellular ligands for MDA5. Intron-containing pre-mRNAs are increased in the cytoplasm of virus-infected cells Since we found that many MDA5 binding sites were in introns (Figs. 2b and 4b), we next tested if virus infection itself affected the presence of introns in the cytoplasm, where MDA5 is localised. We extracted and sequenced cytoplasmic RNA from THP1 and Calu-3 cells infected with EMCV or SARS-CoV-2, respectively, or uninfected controls. As an additional control, we treated cells with 100 U/ml IFNb. We analysed changes in the abundance of intronic sequences using two methods: IRFinder 60 and index 61 . IRFinder interrogates intron-retained reads within normally spliced mRNA, whereas index examines the overall level of intron reads on a per gene basis, which can cover both intron retention and unspliced pre-mRNA. Analysis with IRFinder showed a modest amount of intron retention after EMCV infection, but not after IFNb treatment, compared to untreated cells (Fig. S7a-b). Index showed that both EMCV and SARS-CoV-2 infection had a significant effect on intron expression, which was much greater than the effect of IFNb treatment (Fig. 5a-i). Indeed, compared to untreated cells, virus infection specifically induced an upregulation in introns in the cytoplasm (Fig. 5b, h), which was not detected in cells treated with IFNb (Fig. 5e). Importantly, there was a significant positive correlation between the introns upregulated/retained in the cytoplasm of virus-infected cells with introns that contained MDA5 binding sites detected by iCLIP (Fig. 5j, k and S6c). Thus, we found that virus infection causes significant changes to the levels of introns in the cytoplasm. These upregulated introns were also associated with MDA5 binding during virus infection, suggesting that they stimulate MDA5. Rescue of mRNA splicing during infection abrogates MDA5 activation Several viruses that activate MDA5 also modulate the subcellular distribution of splicing factors, such as SRSF3 62, 63 , and nuclear ribonucleoproteins, such as hnRNPC 64, 65 , thus significantly affecting host RNA metabolism 66 . Moreover, SARS-CoV-2 infection disrupts host RNA splicing, resulting in intron retention 67 . We therefore hypothesised that virus infections lead to an imbalance in host RNA processing and metabolism, resulting in an increase of intron-containing RNA in the cytoplasm that activates MDA5. To test this idea, we aimed to restore splicing by overexpressing SRSF3 (also known as SRp20), a global regulator of pre-mRNA splicing and mRNA export 68 , which was previously shown to be targeted by poliovirus, a picornavirus that activates MDA5 62 . SRSF3-GFP-overexpressing LIM1215 cells and GFP-overexpressing control cells 69 were equally susceptible to EMCV-induced cell death and expressed equivalent levels of MDA5 and MAVS (Fig. 6a, b). At viral doses that did not result in cell death, the T1-IFN response was blunted in SRSF3-GFP cells compared to GFP control cells (Fig. 6c-j). This was not simply due to lower infection levels in SRSF3-GFP cells, as equivalent or increased viral RNA levels were present in SRSF3-GFP cells compared to GFP control cells (Fig. 6k). These data suggest that – when RNA splicing is “forced” to occur despite virus infection – then MDA5 activation is lost. Discussion MDA5 is an important PRR 8 . Despite its discovery as an RNA sensor of virus infection almost 20 years ago 34 , the PAMP detected by MDA5 during infection has remained obscure and controversial, rendering MDA5 one of the few remaining ‘orphan’ PRRs. Here, we propose that MDA5 monitors the homeostasis of cellular RNA processing and is activated upon perturbations caused by infections (Fig. S8). Conceptually, this finding can be described as ‘guarding’, which involves the detection of pathogen-induced perturbations rather than PAMPs. Several plant immune receptors and the mammalian Pyrin inflammasome employ this mechanism; for example, the latter detects bacterial toxin-induced Rho guanosine triphosphatase inactivation 70 . We found that defects in RNA splicing in virally infected cells triggered MDA5 via recognition of cellular intronic sequences accumulating in the cytoplasm. Our work thus defines innate immune guarding as a principle underpinning cytoplasmic RNA sensing during virus infection. This finding was unexpected as MDA5 is often described as a sensor of viral dsRNA accumulating in infected cells. However, our observations were consistent between two different cell types, monocytic THP1 cells and lung adenocarcinoma Calu-3 cells, and two virus infections, EMCV and SARS-CoV-2. We cannot exclude the possibility that MDA5 detects viral dsRNAs in different settings such as in cells infected with other viruses, in other cell types or at different time points during infection. Nonetheless, our model – in which MDA5 guards posttranscriptional control – provides an attractive explanation as to why MDA5 can detect many different infections. Indeed, SARS-CoV-2 inhibits splicing and induces intron retention 67 . Dengue virus 71 , 72 , poliovirus 62 , 63 , 65 and rhinovirus 73 , 74 also affect host RNA homeostasis by causing cytoplasmic relocalisation of splicing and nuclear RNA binding factors. Furthermore, MDA5 is activated by DNA virus infections, including by herpes simplex virus-1, vaccinia virus and hepatitis B virus 13 , 14 , 39 . DNA viruses do not generate a replicative form viral dsRNA but interfere with splicing 75 . We posit that activation of MDA5 occurs when the fidelity of RNA processing is impacted upon virus infection, either through direct viral antagonists of splicing or indirectly through cellular stress 75 . It will be important to test this in future work for different virus families, as will be the analysis of MDA5’s RNA agonists in non-viral infections 16 – 19 . Interestingly, it was recently suggested that unspliced RNA derived from integrated HIV-1 proviral DNA activates MDA5 76 . Although formally of viral origin, unspliced HIV-1 transcripts are generated by the cell’s transcription machinery; thus, this finding mirrors our observation that MDA5 is activated by intron-containing cellular RNAs. Viral dsRNAs are often associated with viral RNA binding proteins. Moreover, for many viruses, replication takes place in replication factories. For example, SARS-CoV-2 establishes double membrane vesicles, in which its polymerase produces new viral RNA 77 . As such, how MDA5 could gain access to viral dsRNA is questionable, in particular given that in vitro data suggest multiple MDA5 molecules need to bind the same RNA for activation 36 . Moreover, even if MDA5 were to gain access to viral dsRNA in replication factories, it is difficult to envisage how MDA5 would then be able to interact with mitochondrially localised MAVS 78 . Our findings resolve these conundrums: detection of inappropriately processed host RNAs explains how MDA5 can sense viral infections when viral RNAs are shielded by viral proteins or within replication factories. We found specific sequence motifs in host RNA that were associated with MDA5, including Poly(A)/(U)-rich sequences. Structurally, MDA5 binds dsRNA in a sequence-independent manner, due to interactions with the phosphodiester backbone 79 . It is possible that the motifs discovered here are first bound by other RNA-binding proteins that are then replaced by MDA5. Indeed, several RNA-binding proteins specifically bind to Poly(U) tracts 80 , including some known to be relocalised to the cytoplasm during virus infection such as hnRNPC 81 , 82 . Thus, when viruses usurp host RNA factors for their own replication and induce cytoplasmic relocalisation of nuclear factors, this may in fact result in these nuclear factors bringing with them to the cytoplasm unspliced RNAs that then activate MDA5. Although MDA5 was first discovered as a sensor of virus infection, it is also activated in autoinflammatory disease. In the absence of ADAR1, which catalyses the deamination of adenosines to inosines in dsRNA to break base-pairing, MDA5 detects cellular dsRNA species 83 . This leads to sterile induction of T1-IFN. Accordingly, inactivating mutations in ADAR1 cause the T1-IFN-mediated Aicardi-Goutières syndrome, a neurodevelopmental disorder 84 . Host repetitive RNA, in particular inverted Alu repeats, which can form base-paired structures, have been proposed as RNA ligands of MDA5 in this condition 46 , consistent with our observation that many MDA5 binding sites are in proximity of Alu elements. Moreover, MDA5 was also found to be stimulated by RNA transcripts from endogenous retroviruses (ERVs) when cancer cells were treated with demethylating agents (e.g. azacytidine derivatives). These compounds derepress ERVs, resulting in the formation of dsRNAs that stimulate MDA5 85, 86 . Interestingly, treatment of cancer cells with inhibitors of the spliceosome results in intron-retention in cytoplasmic RNA and concomitant activation of T1-IFN via MAVS, potentially explaining the therapeutic efficacy of drugs targeting splicing 87 . Moreover, depletion of the splicing factors hnRNPC and hnRNPM induces MDA5 activation 88 , 89 . These observations provide important parallel lines of evidence to the work presented here and suggest that sensing of aberrant RNA processing is a universal mechanism explaining MDA5 activation during both infections and in sterile settings. It is also noteworthy that in addition to intronic RNA other types of non-coding RNA may contribute to MDA5 activation. For example, in neuronal cell types, 3’UTRs are extended and have been suggested to activate MDA5 due to the presence of inverted repeat Alu elements 90 . An important methodological advance of our work is the use a-MDA5 monoclonal antibodies that allow identification of the RNAs that bind endogenous MDA5 in situ in cells. Whilst here applied to virus infections, our protocol can be used in any human cell type that expresses MDA5, including primary cells and patient samples. The a-MDA5 antibodies and iCLIP protocol are available as an open resource to the research community. We hope they will be applied in future to discover MDA5 agonists in autoinflammation and cancer treatment. Like MDA5, other nucleic acid sensors have also been described to detect viral dsRNA. Protein kinase R (PKR) and members of 2'-5'-oligoadenylate synthetase (OAS) family bind dsRNA and are activated by virus infections 91 , many of which also activate MDA5, and have been associated with similar non-infectious pathologies 92 – 95 . These receptors, rather than inducing a T1-IFN response, largely act to restrict virus spread by inducing global RNA degradation or shutting down mRNA translation. OAS1 was recently found to bind host intronic and repetitive RNA, in addition to viral RNA, during SARS-CoV-2 infection 96 . Whether PKR and OASs are – like MDA5 – activated by host intronic RNAs in the context of different virus infections is worth investigating. Taken together, our work shifts the understanding of MDA5 activation: rather than detecting viral RNA as a PAMP, as is the case with other PRRs, we propose that MDA5 is a sensor of cellular RNA homeostasis. When disturbances in this homeostasis occur as a result of virus infection, MDA5 is triggered by improperly spliced cellular RNAs. These findings have important bearings on treatment approaches for both infectious and non-infectious diseases that involve MDA5 activation. Methods Cell culture and virus infection Cells were cultured at 37°C and 5% CO 2 and routinely screened for mycoplasma contamination. THP1 (gift from Vincenzo Cerundolo) and LIM1215 (gift from Minna-Liisa Änkö) cells were maintained in RPMI (Sigma Aldrich) supplemented with 10% v/v foetal calf serum (FCS) and 2 mM L-glutamine (Gibco). Huh7 (gift from Jane McKeating), A549 (gift from Georg Kocks), BHK-21 (gift from Alain Townsend), p125HEK 50 and MEF (gift from Shizou Akira) cells were maintained in DMEM supplemented with 10% v/v foetal calf serum (FCS) and 2 mM L-glutamine (Gibco). Calu-3 cells (gift from Caroline Goujon) were maintained in MEM (Sigma Aldrich) supplemented with 10% v/v foetal calf serum (FCS), 2 mM L-glutamine (Gibco), 1x sodium pyruvate (Gibco) and 1x non-essential amino acids (Gibco). For generation of MDA5 and RIG-I knockout cells using CRISPR/Cas9 technology, sgRNAs cloned into pX458-Ruby (Addgene 110164, deposited by Dr. Philip Hublitz) and described earlier 50 were used. LTX Transfection kit (Life Technologies) was used to transfect 2 x 10 6 cells THP1 cells with 4 µg of plasmid, or 2 x 10 5 Huh7 and p125HEK cells using 2.5 µg of plasmid according to manufacturer’s instructions. After 24 hr, mRuby-positive cells were single-cell sorted into 96-well plates containing fresh medium for Huh7 and p125HEK cells; or 50% fresh medium and 50% filtered conditioned medium from THP1 parental cells for THP1 cells. Surviving clones were expanded and screened by immunoblotting for ablation of target protein, and Sanger sequenced to ascertain mutation at the gene locus. IRF3 knockout THP1 cells have been previously described 50 . RIG-I knockout MEFs cells were a gift S. Akira. LIM1215 cells stably overexpressing SFSR3-GFP or GFP were described previously 69 . EMCV (gift from C. Reis e Sousa) was produced in BHK-21 cells, and harvested from supernatant by centrifugation at 10,000 g and 0.2 µm filtration before single-aliquot freezing at − 80°C. Viral content was quantified by plaque assay on BHK-21 cells. SARS-CoV-2 Victoria/02/2020 (passage 5) was produced in Vero E6 and titrated by plaque assay as described previously 9 . For cytoplasmic RNA extraction, the BetaCoV/France/IDF0372/2020 isolate was supplied by Sylvie van der Werf and the National Reference Centre for Respiratory Viruses hosted by Institut Pasteur (Paris, France). The patient sample from which strain BetaCoV/France/IDF0372/2020 was isolated was provided by X. Lescure and PY. Yazdanpanah from the Bichat Hospital, Paris, France. Moreover, the strain BetaCoV/France/IDF0372/2020 was supplied through the European Virus Archives goes Global (Evag) platform, a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 653316. Experiments using this strain were performed at the CEMIPAI BSL3 facility (UAR 3725 CNRS Montpellier University). Antibodies and Reagents All chemicals and reagents used were from Merck unless otherwise indicated. MDA5 antibodies were generated by C. Song and B. Jin. After screening, clone 16 (also referred to as Antibody A) and clone 22 (also referred to as Antibody B) were selected for native and denatured IP, respectively, and clone 17 was chosen for immunoblotting. Hybridoma cells expressing clone 16 and clone 17 were grown in bioreactors (CELLline) as per manufacturer’s instructions, and stored at -80°C. For clone 22, the hybridoma cells were non-recoverable, therefore the antibody was de novo sequenced by protein mass spectrometry and synthesized using recombinant protein expression in mammalian cells (Rapid Novor and Absolute Antibody). The J2 antibody against dsRNA was from Scicons, the SRSF3 antibody was from Sigma (WH0006428M8), the MAVS (PA5-17256) and GFP (A21311) antibodies were from Thermo Fisher Scientific, the SARS-CoV-2 Spike antibody (GTX632604) was from GeneTex and the MxA antibody (clone M143, MABF938) was from Merck. The ISG15 antibody was a gift from K. P. Knobeloch. HRP-coupled secondary antibodies were sheep‐α‐mouse and donkey‐α‐rabbit (both GE Healthcare, 1:3000). Confocal Microscopy THP1 cells were grown on sterile poly-L-lysine (0.01%, Merck) treated glass coverslips (5 x 10 5 cells per well in 24-well plates) with phorbol 12-myristate 13-acetate (PMA, 10 ng/ml; Invivogen) for 24 hr. Cells were infected with EMCV at MOI = 10 for 16 hr, washed with PBS, and fixed for 15 min with 4% formaldehyde in cytoskeleton-stabilizing buffer (CSB; 5 mM KCl, 137 mM NaCl, 4 mM NaHCO 3 , 0.4 mM KH 2 PO 4 , 1.1 mM Na 2 HPO 4 , 2 mM MgCl 2 , 5 mM PIPES, 2 mM EGTA and 5.5 mM glucose in water). Cells were washed with PBS, permeabilized with 0.1% Triton-X 100 in CSB for 20 min, and incubated in 0.1M glycine in CSB for 10 min. Cells were washed four times between all subsequent steps with PBS for 5 min each wash. Cells were incubated in blocking buffer (1% BSA, 10% normal goat serum (Abcam) in PBS) for 1 hr, then for 2 hr with anti-dsRNA antibody (J2, 1:200 in blocking buffer), and 1 hr with goat anti-mouse AlexaFluor 488 (1:500 in blocking buffer; Life Technologies, A11029) and AlexaFluor 633 Phalloidin (1:40 in blocking buffer; Thermo Fisher Scientific A22284). Slides were mounted using ProLong Gold antifade mountant with DAPI (Thermo Fisher Scientific, P36931), and imaged using a Zeiss 780 inverted confocal microscope. Immunoblotting Cells were lysed with lysis buffer (10 mM Tris, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 5 uM ZnCl 2 , 0.5% IGEPAL and complete protease inhibitor cocktail (Roche)) and protein was quantified by BCA assay (Pierce). For LIM1215 analysis of cytoplasmic and nuclear protein, cells were lysed in RIPA buffer (50mM Tris-HCl pH 8, 150mM NaCl, 1% IGEPAL, 0.5% Sodium deoxycholate, 0.1% SDS). Samples were denatured using NuPAGE LDS sample loading buffer (Life Technologies) and 10% 2-mercaptoethanol, and heating at 95°C for 5 min. Samples were resolved by electrophoresis on 4–12% Bis-Tris gels with MOPS Running Buffer (Life Technologies NuPAGE system) and transferred to nitrocellulose membrane by electrophoresis at 100V for 2 hr in cold transfer buffer (Life Technologies) with 10% methanol. Membranes were blocked with 0.05% IGEPAL in Tris-buffered saline (TBS-N; 50 mM NaCl, 50 mM Tris-HCl, pH 7.6) containing 5% non-fat milk (5% milk TBS-N) for 1 hr, and probed with primary and HRP-conjugated secondary antibodies diluted in 5% milk TBS-N for 1 hr at room temperature or overnight at 4°C, with rotation. Membranes were washed four times in TBS-N for 5 min each wash after each antibody incubation. Proteins were visualized on iBright (Thermo Fisher Scientific) after exposure to Western LightningPlus-ECL chemiluminescent reagent (PerkinElmer). MDA5 iCLIP Conditions were maintained RNase-free throughout the experiment, conducted as described in Huppertz et al. 43 , with some modifications. IRF3-KO or MDA5-KO THP1 cells (2 x 10 7 cells in 20 cm plates, 10–20 plates per condition) were incubated with PMA overnight, then infected with EMCV (MOI = 5) for 22 hr, or left uninfected (IRF3-KO only). Cells were washed twice with PBS and UV crosslinked using 150 mJ/cm 2 in a Spectrolinker XL-1500 (Spectronics Corp). Cells were scraped, centrifuged at 400 g for 5 min, and PBS supernatant was discarded. Cell pellets were flash frozen and stored at − 80°C until ready for use. Cell pellets were lysed in lysis buffer (10 mM Tris, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 5 µM ZnCl 2 , 1% IGEPAL and complete protease inhibitor cocktail (Roche)) using equivalent of 400 µl of buffer per plate of cells. Samples were incubated on ice for 10 min and clarified by centrifugation at 10,000 g for 10 min. Protein concentration was determined by BCA assay. 5000 µg of clarified cell lysate was used per condition. RNase A (Affymetrix, 70194Y; 20 U/µl) was added at 1:10,000 dilution to samples and incubated for 5 min, followed by cooling and addition of 1:80 dilution of RNAsin Plus RNase inhibitor (Promega; N2611). Anti-MDA5 mAb 16 was covalently coupled to Dynabeads (Invitrogen, 14311D) according to manufacturer’s instructions, using 9 mg Dynabeads and 180 ug of mAb 16 per sample. mAb 16 antibody-coupled Dynabeads were washed with high salt buffer (50 mM Tris pH 7.4, 1M NaCl, 1% IGEPAL, 0.1% SDS, 0.5% deoxycholate) and lysis buffer, and incubated with protein samples for 2 hr at 4°C with rotation. 1% of sample was collected and stored at 4°C, for use as size-matched input control (SMI). Samples were washed twice for 5 min with high salt buffer, and once with PNK buffer (20 mM Tris pH 7.4, 10 mM MgCl 2 , 0.2% Tween-20). Supernatant was removed, 300 µl of urea cracking buffer (50 mM Tris pH 7.4, 6M urea, 1% SDS, 25% PBS) was added, and samples incubated with shaking for 3 min at 65°C, before neutralisation with 3 ml of T-20 IP buffer (50 mM Tris pH 7.4, 150 mM NaCl, 0.5% Tween-20, 0.1 mM EDTA) and addition of RNase and protease inhibitors. A second round of immunoprecipitation was done using anti-MDA5 mAb 22 antibody preincubated with Protein G Dynabeads (Life Technologies; 900 µl beads and 90 µg of antibody) according to manufacturer’s instructions. Samples were incubated with beads for 2 hr at 4°C with rotation, and washed twice for 5 min with high salt buffer, and twice with PNK buffer. RNA in samples was 3’ end dephosphorylated using T4 PNK (New England Biolabs, M0201L) for 20 min, with addition of RNase inhibitor and Turbo DNase I (Life Technologies AM2238, 2U/ul). Samples were washed twice with RNA ligase buffer (5mM Tris HCl pH 7.5, 1 mM MgCl 2 ), and iCLIP 3' linker (Trilink, O-30050-03) was ligated to 3’ end of RNA molecules using T4 RNA Ligase 1 High Concentration (New England Biolabs, M0437M) with the addition of 2.6% DMSO, 40% PEG-8000, and 2% RNase inhibitor, for 2 hr at 25°C with shaking. Samples were washed twice with high salt buffer, changing tubes, and twice with PNK buffer. 10% of sample was set aside for radiolabelling using T4 PNK and g-32P-ATP (Hartmann Analytic; FP-301; 3000Ci/mmol; 10mCi/ml) according to manufacturer’s instructions. IP and SMI samples were denatured using sample loading buffer and DTT, and incubation for 10 min at 70°C. Samples were resolved on 4–12% Bis-Tris gels (Life technologies; NP0322BOX) using MOPS SDS running buffer (Life technologies; NP0001) for 2 hr at 150 V. Samples were transferred to nitrocellulose for 4 hr at 100 V using cold Transfer Buffer (Life technologies; NP0006) with 10% methanol. Membrane was exposed to film for signal visualisation, and the radiograph was used as a guide to excise samples from membrane. Samples of MW > 135 kDa were excised, and membrane was cut into small pieces and placed in a microcentrifuge tube. RNA was released from membrane by protein digestion using Proteinase K (Roche, 03115844001) for 1 hr at 50°C, according to manufacturer’s instructions. RNA was purified by phenol:chloroform:iaa (P3803-100ML) with phaselock gel tubes (5 Prime), and ethanol precipitation. For SMI samples, protocol was done as described 54 . IP and SMI samples were reverse-transcribed with primers shown in Tab. S1, using TGIRT enzyme (InGex; CM0101-50) according to manufacturer’s instructions. Samples were ethanol precipitated, resuspended in TBE-urea sample buffer (Life Technologies; LC8676) and resolved on 6% TBE urea gels (Life Technologies; EC6865BOX) for 40 min at 180V. cDNA sized between 80 and 200 nt was excised, fragmented, and incubated in 400 µl diffusion buffer (0.5 M ammonium acetate, 10 mM magnesium acetate, 1 mM EDTA, 0.1% SDS) for 30 min at 50°C, shaking. Sample was clarified by centrifugation in SpinX columns (Costar) containing two 10 mm glass pre-filters (Whatman, 1823-010), and purified by phenol:chloroform:iaa and ethanol precipitation. cDNA samples were circularized using CircLigase II (Cambio, CL9021K) and BamHI digested (New England Biolabs) as described 43 . Samples were PCR amplified using the P3 and P5 Solexa primers and Accuprime PCR Master Mix (Thermo Fisher Scientific), starting at 20 cycles and increasing in 1–2 cycle steps, until a signal was detectable by Tapestation D1000 High sensitivity kit. Libraries were concentrated using the MiniElute PCR Purification kit (Qiagen) and resolved on a 6% TBE gel (Life Technologies, EC6265BOX) for 30 min at 180V. Gel was stained with SYBRGold (Thermo Fisher Scientific), and fragments between 100 and 250 bp were excised, extracted from gel as above, and purified using QIAQuick gel extraction kit (Qiagen). Samples were quantified by Qubit High Sensitivity DNA kit (Thermo Fisher Scientific, Q32851) and KAPA Library quantification Kit (Illumina, 07960093001). Samples were equalized to 4 nM, and combined at a ratio of 10:1 IP samples to SMI samples, in order to increase sequencing from IP samples. Libraries were sequenced on Illumina NextSeq500 using 75 cycle High Output Kit v2 (TG-160-2105). Each biological repeat was sequenced on a separate chip. For MDA5 iCLIP from SARS-CoV-2-infected and uninfected Calu-3 cells, the protocol above was followed, with some modifications as follows 97 , 98 . 2.5 x 10 7 cells in 20 cm plates (2 plates per condition) were left uninfected or infected with SARS-CoV-2 at an MOI of 0.1 for 48 hr. After dual IP and adaptor ligation, samples were directly treated with proteinase K, without SDS-PAGE size selection. Samples were reverse transcribed with primers shown in Tab. S2 98 , which contained longer UMI and barcode sequences, and carbon spacers to stop the progression of PCR during amplification, removing the need for linearisation after circularisation of libraries. A pre-amplification PCR step of 6 cycles prior to library size selection was performed to increase retention of unique sequences 97 . Libraries were then size-selected using ProNex Chemistry (Promega; NG2001), PCR amplified, and size-selected again as described 97 . Libraries were sequenced on Illumina NextSeq 2000 using 100 cycle P3 Reagents (20040559). All samples were sequenced on a single chip. iCLIP data processing and analysis A diagram of iCLIP data processing is shown in Fig. S2a, and largely followed recommendations outlined in 99 . Fastx-trimmer ( http://hannonlab.cshl.edu/fastx_toolkit/ ) was used to remove low quality barcode regions as described in 99 . Flexbar ( https://github.com/seqan/flexbar ) was used to simultaneously 3’ end trim, demultiplex, and extract UMIs. FastQC was used to determine quality of sequencing reads pre- and post-processing ( https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ ). A custom genome and gtf file containing both the human genome (hg38 assembly) and the EMCV genome (DQ288856.1) were prepared. Samples were mapped against the combined human-EMCV genome using STAR 100 with the following parameters: STAR --runMode alignReads --runThreadN 8 --limitBAMsortRAM 10000000000 --outSAMattributes All --outStd BAM_SortedByCoordinate --outSAMtype BAM SortedByCoordinate --outFilterType BySJout --outReadsUnmapped Fastx --outSAMattrRGline ID:foo --alignEndsType Extend5pOfRead1 --outFilterMismatchNoverReadLmax 0.04 --outFilterMismatchNmax 999 --outFilterMultimapNmax 1 --sjdbOverhang maxReadLength-1 --outSJfilterReads Unique. Samples were indexed using samtools and deduplicated based on UMI using umi_tools dedup 101 . MultiQC was used to quantify rates of mapping 102 . A custom computational pipeline was written using ruffus 103 to automate the sample processing described above. This is available for download at https://github.com/natsampaio/iCLIP_pipeline . To determine crosslinking sites, the PureCLIP peak-calling algorithm was applied 55 . Chromosomes 1–6 were used for the algorithm parameter learning, and CL motifs and SMI controls were incorporated. For motif analysis, FASTA sequences containing 50 bp either side of crosslinking site were obtained using bedtools 104 commands slop and getfasta. Sequences present in the negative control samples (MDA5-KO cells or uninfected cells) and present in rRNA and tRNA (database from SILVA; https://www.arb-silva.de/ ) were subtracted from the sequences from EMCV-infected cells using bedtools slop. The resulting sequences were analysed using the MEME suite (5.0.1) meme-chip algorithm, with -norc and -centrimo-local parameters included 57 . The same sequences were also analysed using HOMER 56 function findMotifsGenome.pl, with -rna. Basic analysis compared test sequences to automatically generated random genetic background. Test sequences were also compared to either MDA5-KO or uninfected sequences using the -bg option. Genomic context of sequences was determined using the bedtools command intersect to quantify number of sequences overlapping with annotated introns, exons, 5’ UTR or 3’ UTR in the genome. Sequences that contained crosslink sites were analysed using RepeatMasker 105 to quantify repetitive regions, and bedtools intersect using a database of annotated repetitive regions in the genome. Integrative Genomics Viewer (IGV 2.8.9) 106 was used to visualize data. Flow cytometry THP1 cells were grown in 12-well plates (1 x 10 6 cells per well) and stimulated for 24 hr with PMA (10 ng/ml). Cells were infected with EMCV using MOIs and incubation times specified in figure legends, prior to addition of cold PBS containing 1% FCS (FACS Buffer). After 10 min incubation at 4°C, cells were lifted by pipetting, and washed with FACS buffer. For all subsequent steps, reagents were diluted in FACS buffer unless stated otherwise, and washed twice with FACS buffer between steps. Cells were stained with Live/Dead Fixable Aqua Cell Stain (1:200 in PBS, Life Technologies, L34957) combined with FcR block (1:200 in PBS, eBioscience), fixed in 4% formaldehyde (10 min), and permeabilised in 0.1% Triton-X (20 min). Cells were stained with J2 dsRNA antibody (1:200, 30 min) and goat anti-mouse AlexaFluor-488 (1:500, 30 min; Life Technologies, A11029), and resuspended in CellFix (1:10 in water; BD, 340181). Cells were analysed by flow cytometry on an Attune NxT Flow Cytometer (Thermo Fisher Scientific) and FlowJo software (BD). Calu-3 cells were grown in 6-well plates (1x10 6 cells per well) and infected with SARS-CoV-2 at an MOI of 0.1 for 48 hr. Cells were washed with PBS, lifted using Trypsin for 15 min at 37°C and fixed using formalin for 20 min at RT. Cells were stained in separate wells with anti-SARS-CoV-2 Spike (1:200) and anti-MxA (1:300) antibodies for 30 min at 4°C, then with goat anti-mouse IgG AlexaFluor-647 (1:500; Life Technologies) and goat anti-mouse IgG2a AlexaFluor-647 (1:500, BioLegend) for 30 min at 4°C using Saponin buffer (PBS, 1% BSA, 0.1% Saponin) for all antibody incubations, and washed twice in the same buffer between steps. Cells were resuspended in FACS buffer and analysed by flow cytometry using a NovoCyte flow cytometer (ACEA Biosciences Inc.) and FlowJo software (BD). In vitro RNA transcription and purification For production of in vitro transcribed RNA (IVT-RNA) from human genome, genomic DNA was extracted from THP1 cells using DNeasy Blood & Tissue Kit (Qiagen, 69504) and used as PCR templates. Herculase II PCR Kit (Agilent, 600677) was used to PCR amplify desired regions with addition of T7 polymerase promoter sequence. MegaScript T7 Kit (Invitrogen, AM1334) was used for production of IVT-RNA according to manufacturer’s instructions. RNA was purified using RNeasy Mini kit (Qiagen, 74104), and quantified by Nanodrop (Thermo Fisher Scientific) and stored at − 80°C in single-use aliquots. RNA was visualised by electrophoresis on denaturing agarose gels and staining with SYBR Gold. To produce dsRNA, forward and reverse complementary IVT-RNA strands were heated to 90°C and cooled at 1°C/30 sec until 25°C, then used immediately for cell transfection. Cell transfection of IVT-RNA Cells were grown in 12-well plates (1.75 x 10 6 cells per well) for 24 hr, and transfected with 175 ng of IVT-RNA, 87.5 ng of high MW Poly(I:C) (Invivogen, tlrl-pic), or 35 ng of Neo 1 – 99 IVT-RNA 29 using 0.7 µl of Lipofectamine 2000 (Thermo Fisher Scientific, 11668027) according to manufacturer’s instructions. Huh7 cells were treated with 30 U/ml IFN-A/D (R&D Systems) for 24 hr prior to transfection. RNA was extracted after 24 hr using RNeasy Plus Mini Kit (Qiagen, 74136) according to manufacturer’s instructions, quantified by Nanodrop, and stored at − 80°C. RT-qPCR RNA (1 µg) was converted into cDNA using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific) and Oligo-dT primers (Invitrogen) according to manufacturer’s instructions. cDNA was diluted to 100 ng/µl and quantitative PCR was performed using TaqMan Real-Time PCR Assays for designated genes and TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) according to manufacturer’s instructions. Alternatively, quantitative PCR was performed using SYBR Green PCR Master Mix (Thermo Fisher Scientific) with gene specific probes detailed in Tab. S2. Assays were performed on QuantStudio 6 Flex Real-Time PCR machines (Thermo Fisher Scientific). Two-way ANOVA was performed with Sidak’s multiple comparisons test using GraphPad Prism software. IFNB1 promoter luciferase assay p125-HEK293 RIG-I-KO 50 and p125-HEK293 RIG-I-KO/MDA5-KO stably expressing a Renilla luciferase downstream of an IFNB1 promoter were plated in 96-well plates (2.5 x 10 4 cells per well) for 24 hr, then treated with 30 U/ml IFN-A/D for a further 24 hr. Cells were transfected with 50 ng of IVT-RNA, 50 ng of cellular RNA, 25 ng of high MW Poly(I:C) (Invivogen, tlrl-pic), or 10 ng of Neo 1 – 99 IVT-RNA using 0.2 µl of Lipofectamine 2000 (Thermo Fisher Scientific, 11668027) according to manufacturer’s instructions. After 16–24 hr, medium was removed and 75 µl of 1:1 solution of cell culture medium and ONE-Glo luciferase assay reagent (Promega, E6120) was added to cells. After 3 min, 50 µl of sample was transferred to a white 96-well plate and luminescence measured on a GloMax Luminometer (Promega). Cytoplasmic RNA extraction and sequencing IRF3-KO THP1 cells were incubated with 10 ng/ml PMA overnight, then left untreated, infected with EMCV (MOI = 2) for 20 hr, or treated with human IFNβ (Rebif) for 6 hr. Calu-3 cells were seeded in 6-well plates (1 x 10 6 cells per well) overnight, then left untreated, infected with SARS-CoV-2 (MOI = 0.1) for 48 hr, or treated with human IFNβ (PBL assay science) for 6 hr. Cytoplasmic RNA was extracted using Cytoplasmic & Nuclear RNA Purification Kit (Norgen) according to manufacturer’s instructions. Total RNA libraries were prepared, starting with 500 ng of total RNA, according to the Illumina Tru-Seq Stranded Total RNA-Seq protocol (protocol 1000000040499) with RiboZero depletion, using unique dual indexes. 150 base paired-end sequencing was performed on a NextSeq 2000 P3 chip, loading pooled libraries at a concentration of 1,000 pM (protocol 1000000109376). Approximately 100 million reads were obtained per sample. Base calling was performed using Dragen BCLConvert (v3.7.4). Intron retention and upregulation analysis Read alignment was performed in performed in R (v4.1.2) 107 using the Rsubread package (v2.6.4) 108 . A genome index was built using the custom combined human and EMCV genome using the buildindex function and alignment was performed using sunjunc, with default settings. Exon- and intron-level differential gene expression analyses followed the methods of the index package (v1.0) 108 . A custom GTF file consisting of human and EMCV data was used for mapping. This GTF file was processed using the Annotations scripts from the Intron-reads repository ( https://github.com/charitylaw/Intron-reads ) in order to define exon and gene body regions. Counts were obtained using featureCounts from Rsubread (v2.8.1), with isPairedEnd and useMetaFeatures set to TRUE, allowMultiOverlap set to FALSE and strandSpecific 2. Intron counts were obtained by subtracting exon counts from gene body counts. Two DGEList objects were created using the edgeR package (v3.36.0) 109 using the exon and intron counts respectively, as well as sample annotation and gene annotation, with additional annotation obtained from the biomaRt package (v2.50.1) 110 , 111 . Sample 1C_IFNb, a technical replicate of sample 3C_IFNb, was excluded from further analysis. A design matrix was created incorporating the treatment group, so that the treatments could be compared, as well as the experimental replicate, to control for the experimental batch effect. Total library sizes were obtained by summing the exon and intron counts. Lowly expressed genes, on either an exonic or intronic level, were removed according to the filterByExpr function, incorporating the design matrix. As such, single-exon genes were also excluded from further analysis. Normalisation factors were calculated using the calcNormFactors function with the TMM method 112 . Counts were processed using the voom method 113 and a linear model was fit using the edgeR voomLmFit function, including the design matrix. Comparisons between EMCV-treated, IFNβ-treated and untreated groups were made using the contrasts.fit function from the limma package (v3.50.0) 114 . Empirical Bayes moderated t -tests were performed against a 1.2-fold-change threshold and p -values were obtained using the treat 115 function for differential expression analysis or without a fold-change threshold using the eBayes function 116 for subsequent comparison to iCLIP data. Results of the intron-level analysis were visualised as mean-difference plots, showing the log 2 average expression and log 2 fold changes, highlighting genes significantly differentially expressed at the intron level, with Benjamini-Hochberg adjusted p -values < 0.05. Differential expression results were visualised by plotting exon- versus intron-level log 2 fold changes on a per gene basis, quantifying the number of genes belonging to different differential expression categories, following the plot_index function from the index package. IRFinder (v.1.30) 60 was used to examine differential intron retention. Intron retention was quantified in BAM mode using the BAM files obtained from the Rsubread subjunc alignment, utilising a genome reference built using IRFinder BuildRefProcess on the combined human and EMCV genome and GTF files. Differential intron retention was performed in R using the DESeq2 package (v1.34.0) 117 , following the DESeq2Constructor.R script provided by IRFinder. The resulting DESeqDataSet object was filtered to only include introns that had been designated as clean by IRFinder (excluding known-exon regions). Furthermore, introns were only included if they either had no warnings or were only flagged as NonUniformIntronCover in at least 3 samples (thereby excluding introns that had been categorsed as LowCover, LowSplicing or MinorIsoform in most samples). The design matrix was defined as ~ Group + Group:IRFinder + Replicate and differential expression performed using the DESeq function. The Group:IRFinder interaction term was used for finding differences in intron retention, by comparing GroupEMCV.IRFinderIR and GroupIFNb.IRFinderIR to GroupUnif.IRFinderIR with the DESeq2 results function. Results of the intron-retention analysis were visualised as mean-difference plots, showing log 2 (base mean expression + 1) and log 2 fold changes, highlighting significantly differentially retained introns with Benjamini-Hochberg adjusted p -values < 0.05. To compare the cytoplasmic RNA-seq to the iCLIP results, bedtools intersect was used in stranded mode to find genes from the index analysis and introns from the IRFinder analysis that overlapped with intronic EMCV iCLIP peaks. Results were visualised using the barcodeplot function from the limma package, either using the index EMCV intron-level empirical Bayes moderated t -statistics or IRFinder DESeq2 EMCV intron retention test statistics, showing all genes or introns that overlapped with iCLIP sites with scores higher than the 95% percentile. The iCLIP scores were used for the barcodeplot gene.weights argument; if more than one iCLIP site were overlapping a gene or intron, the higher score was used. In addition, p -values were obtained using the roast 118 function from the edgeR package, using DGEList objects created from the counts from the index and IRFinder analyses. Dispersions were estimated using the estimateDisp function 119 with robust set to TRUE 116 , then the roast tests were performed using the iCLIP score weights with 19,999 rotations and the up-direction p -values were reported. MTT assay LIM1215-GFP and LIM1215-SRSF3-GFP cells were plated at 9 x 10 4 cells/well in 96-well plates, incubated for 24 hr, and infected with EMCV at indicated MOIs for 48 hr. Cell culture medium was replaced with 100 µl of DMEM containing 1.5 mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide), and cells incubated for 3 hr. 150 µl of DMSO (dimethyl sulfoxide) was added to wells, and absorbance at 590 nm measured on OPTIstar plate reader (BMG LabTech). EMCV RNA depletion and transfection into cells THP1 cells were incubated with PMA overnight (10 ng/ml), infected with EMCV (MOI = 5) for 22 hr or left uninfected, and total cellular RNA extracted (RNeasy, Qiagen). A mixture of five dual-biotinylated 35 nucleotide DNA probes (Integrated DNA Technologies) antisense to the EMCV genome (Tab. S2) were prepared at 8 µM and incubated with 250 µg of Hydrophilic Streptavidin Magnetic Beads (S1421, New England Biolabs) for 5 min, followed by washing with Wash/Binding Buffer (0.5 M NaCl, 20 mM Tris-HCl (pH 7.5), 1 mM EDTA). 5 µg of cellular RNA was diluted in 25 µl of Wash/Binding Buffer, heated to 75°C for for 5 min, and chilled on ice for 3 min. The RNA was incubated with probe-coated streptavidin beads for 15 min at 37°C. The EMCV-depleted supernatant was collected and RNA purified with RNeasy kit. A549 wild-type or MDA5-KO 59 were grown in 12- well plates (3 x 10 5 cells/well) in the presence of 30 U/ml IFNβ (Rebif) for 24 hr, pre-treated with 400 µM ribavirin (Sigma, R9644) for 1 hr, and transfected with 200 ng of EMCV-depleted or non-depleted RNA using 0.7 µl of Lipofectamine 2000 (Thermo Fisher Scientific, 11668027) according to manufacturer’s instructions. RNA was extracted after 20 hours using RNeasy Plus Mini Kit (Qiagen, 74136) according to manufacturer’s instructions, quantified by Nanodrop, and stored at − 80°C. Declarations Author contributions (using the CRediT taxonomy) Conceptualisation: N.S. and J.R.; Methodology: N.S. and A.G.D.J..; Software: N.S. and L.J.G.; Validation: N.S. and J.R.; Formal analysis: N.S., L.J.G. and J.R.; Investigation: N.S., A.G.D.J., L.C., V.O., C.C. and A.M.; Resources: M.R. and M.A.; Data curation: N.S.; Writing – Original Draft: N.S. and J.R.; Writing – Review & Editing: all authors; Visualisation: N.S., L.J.G and J.R.; Supervision: J.R. and P.J.H.; Project administration: N.S.; Funding acquisition: N.S., A.G.D.J., P.J.H. and J.R. Acknowledgments The authors thank David Sims, Charlotte George, George Kassiotis, George Young, Vladimir Pena, and members of the Rehwinkel lab for discussion; Chaojun Song and Boquan Jin for raising MDA5 antibodies; and Jurgen Moonen, Antony Mathews, Georgie Wray-McCann, the Monash Translational Health Precinct Medical Genomics Facility and the CEMIPAI BSL3 facility for providing reagents or for technical support. This work was funded by the UK Medical Research Council [MRC core funding of the MRC Human Immunology Unit; J.R.], the Wellcome Trust [grant number 100954; J.R.], the Lister Institute [J.R.], and the Australian National Health and Medical Research Council [P.J.H]. A.G.D.J. was supported by CNPq [grant number 211806/2013-7]. L.C. was supported by the ANRS-MIE [postdoctoral fellowship number ECTZ134113 and project grant number ECTZ134139]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Declaration of interests The authors have declared that no conflict of interest exists. Data availability statement Cytoplasmic RNA sequencing data were uploaded to the NCBI Gene Expression Omnibus (GEO) under the SuperSeries accession GSE214664. Raw FASTQ files, as well as the exon and intron counts, were deposited for EMCV infection of IRF3 KO THP-1 cells (SubSeries GSE211240) and SARS-CoV-2 infection of Calu-3 cells (SubSeries GSE214663).o review GEO accession GSE214664, go to: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE214664 and enter token: gbwtoaialhszpet. Materials availability All unique reagents generated in this study are available from the corresponding authors with a completed Materials Transfer Agreement. References Medzhitov, R. Approaching the asymptote: 20 years later. Immunity 30 ,766-775 (2009). Bartok, E. & Hartmann, G. Immune Sensing Mechanisms that Discriminate Self from Altered Self and Foreign Nucleic Acids. Immunity 53 ,54-77 (2020). McNab, F., Mayer-Barber, K., Sher, A., Wack, A. & O'Garra, A. Type I interferons in infectious disease. Nature reviews. Immunology 15 ,87-103 (2015). Crow, Y.J. & Manel, N. Aicardi-Goutières syndrome and the type I interferonopathies. Nature Reviews Immunology 15 ,429 - 440 (2015). Trinchieri, G. Type I interferon: friend or foe? J Exp Med 207 ,2053-2063 (2010). Rehwinkel, J. & Gack, M.U. RIG-I-like receptors: their regulation and roles in RNA sensing. Nature reviews. Immunology 20 ,537-551 (2020). Kato, H. et al. Differential roles of MDA5 and RIG-I helicases in the recognition of RNA viruses. Nature 441 ,101-105 (2006). Dias Junior, A.G., Sampaio, N.G. & Rehwinkel, J. A Balancing Act: MDA5 in Antiviral Immunity and Autoinflammation. Trends in Microbiology (2018). Sampaio, N.G. et al. The RNA sensor MDA5 detects SARS-CoV-2 infection. Scientific Reports 11 ,13638 (2021). Thorne, L.G. et al. SARS‐CoV‐2 sensing by RIG‐I and MDA5 links epithelial infection to macrophage inflammation. The EMBO Journal ,e107826 (2021). Rebendenne, A. et al. SARS-CoV-2 Triggers an MDA-5-Dependent Interferon Response Which Is Unable To Control Replication in Lung Epithelial Cells. Journal of Virology 95 (2021). Yin, X. et al. MDA5 Governs the Innate Immune Response to SARS-CoV-2 in Lung Epithelial Cells. Cell Reports 34 ,108628 (2021). Melchjorsen, J. et al. Early Innate Recognition of Herpes Simplex Virus in Human Primary Macrophages Is Mediated via the MDA5/MAVS-Dependent and MDA5/MAVS/RNA Polymerase III-Independent Pathways. Journal of Virology 84 ,11350 - 11358 (2010). Lu, H.-L. & Liao, F. Melanoma differentiation-associated gene 5 senses hepatitis B virus and activates innate immune signaling to suppress virus replication. Journal of immunology (Baltimore, Md. : 1950) 191 ,3264 - 3276 (2013). Dou, Y., Yim, H.C., Kirkwood, C.D., Williams, B.R. & Sadler, A.J. The innate immune receptor MDA5 limits rotavirus infection but promotes cell death and pancreatic inflammation. The EMBO journal ,e201696273 (2017). Wu, J. et al. Strain-specific innate immune signaling pathways determine malaria parasitemia dynamics and host mortality. Proceedings of the National Academy of Sciences 111 ,E511-E520 (2014). Liehl, P. et al. Host-cell sensors for Plasmodium activate innate immunity against liver-stage infection. Nature Medicine 20 ,47-53 (2014). Bullen, C.K. et al. MDA5 RNA-sensing pathway activation by Mycobacterium tuberculosis promotes innate immune subversion and pathogen survival. JCI Insight 8 (2023). Wang, X. et al. MDA5 Is an Essential Sensor of a Pathogen-Associated Molecular Pattern Associated with Vitality That Is Necessary for Host Resistance against Aspergillus fumigatus. J Immunol 205 ,3058-3070 (2020). Lamborn, I.T. et al. Recurrent rhinovirus infections in a child with inherited MDA5 deficiency. The Journal of experimental medicine 214 ,1949-1972 (2017). Zaki, M. et al. Recurrent and Prolonged Infections in a Child with a Homozygous IFIH1 Nonsense Mutation. Front Genet 8 ,130 (2017). Asgari, S. et al. Severe viral respiratory infections in children with IFIH1 loss-of-function mutations. Proceedings of the National Academy of Sciences of the United States of America 114 ,8342-8347 (2017). Chen, J. et al. Inborn errors of TLR3- or MDA5-dependent type I IFN immunity in children with enterovirus rhombencephalitis. Journal of Experimental Medicine 218 ,e20211349 (2021). Pichlmair, A. et al. RIG-I-mediated antiviral responses to single-stranded RNA bearing 5'-phosphates. Science 314 ,997-1001 (2006). Hornung, V. et al. 5'-Triphosphate RNA is the ligand for RIG-I. Science 314 ,994-997 (2006). Goubau, D. et al. Antiviral immunity via RIG-I-mediated recognition of RNA bearing 5'-diphosphates. Nature 514 ,372 - 375 (2014). Schlee, M. & Hartmann, G. Discriminating self from non-self in nucleic acid sensing. Nat Rev Immunol 16 ,566-580 (2016). Schuberth-Wagner, C. et al. A Conserved Histidine in the RNA Sensor RIG-I Controls Immune Tolerance to N1-2'O-Methylated Self RNA. Immunity 43 ,41-51 (2015). Rehwinkel, J. et al. RIG-I detects viral genomic RNA during negative-strand RNA virus infection. Cell 140 ,397 - 408 (2010). Chiang, J.J. et al. Viral unmasking of cellular 5S rRNA pseudogene transcripts induces RIG-I-mediated immunity. Nature immunology 19 ,53-62 (2018). Vabret, N. et al. Y RNAs are conserved endogenous RIG-I ligands across RNA virus infection and are targeted by HIV-1. iScience 25 ,104599 (2022). Zhao, Y., Ye, X., Dunker, W., Song, Y. & Karijolich, J. RIG-I like receptor sensing of host RNAs facilitates the cell-intrinsic immune response to KSHV infection. Nature communications 9 ,4841 (2018). Naesens, L. et al. GTF3A mutations predispose to herpes simplex encephalitis by disrupting biogenesis of the host-derived RIG-I ligand RNA5SP141. Science immunology 7 ,eabq4531 (2022). Andrejeva, J. et al. The V proteins of paramyxoviruses bind the IFN-inducible RNA helicase, mda-5, and inhibit its activation of the IFN-beta promoter. Proceedings of the National Academy of Sciences of the United States of America 101 ,17264-17269 (2004). Kato, H. et al. Length-dependent recognition of double-stranded ribonucleic acids by retinoic acid-inducible gene-I and melanoma differentiation-associated gene 5. The Journal of experimental medicine 205 ,1601-1610 (2008). Wu, B. et al. Structural basis for dsRNA recognition, filament formation, and antiviral signal activation by MDA5. Cell 152 ,276 - 289 (2013). Triantafilou, K. et al. Visualisation of direct interaction of MDA5 and the dsRNA replicative intermediate form of positive strand RNA viruses. J Cell Sci 125 ,4761-4769 (2012). Feng, Q. et al. MDA5 detects the double-stranded RNA replicative form in picornavirus-infected cells. Cell Reports 2 ,1187 - 1196 (2012). Pichlmair, A. et al. Activation of MDA5 requires higher-order RNA structures generated during virus infection. Journal of Virology 83 ,10761 - 10769 (2009). Sanchez David, R.Y. et al. Comparative analysis of viral RNA signatures on different RIG-I-like receptors. Elife 5 ,e11275 (2016). Deddouche, S. et al. Identification of an LGP2-associated MDA5 agonist in picornavirus-infected cells. Elife 3 ,e01535 (2014). Runge, S. et al. In vivo ligands of MDA5 and RIG-I in measles virus-infected cells. PLoS pathogens 10 ,e1004081 (2014). Huppertz, I. et al. iCLIP: protein-RNA interactions at nucleotide resolution. Methods (San Diego, Calif.) 65 ,274 - 287 (2014). König, J. et al. iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution. Nature Structural & Molecular Biology 17 ,909 - 915 (2010). David, R.Y.S. et al. Comparative analysis of viral RNA signatures on different RIG-I-like receptors. Elife 5 ,e11275 (2016). Ahmad, S. et al. Breaching Self-Tolerance to Alu Duplex RNA Underlies MDA5-Mediated Inflammation. Cell (2018). Gitlin, L. et al. Essential role of mda-5 in type I IFN responses to polyriboinosinic:polyribocytidylic acid and encephalomyocarditis picornavirus. Proceedings of the National Academy of Sciences 103 ,8459-8464 (2006). Schonborn, J. et al. Monoclonal antibodies to double-stranded RNA as probes of RNA structure in crude nucleic acid extracts. Nucleic acids research 19 ,2993-3000 (1991). Weber, F., Wagner, V., Rasmussen, S.B., Hartmann, R. & Paludan, S.R. Double-stranded RNA is produced by positive-strand RNA viruses and DNA viruses but not in detectable amounts by negative-strand RNA viruses. J Virol 80 ,5059-5064 (2006). Hertzog, J. et al. Infection with a Brazilian isolate of Zika virus generates RIG-I stimulatory RNA and the viral NS5 protein blocks type I IFN induction and signaling. Eur J Immunol 48 ,1120-1136 (2018). Rybak-Wolf, A. et al. A variety of dicer substrates in human and C. elegans. Cell 159 ,1153-1167 (2014). Moy, R.H. et al. Stem-loop recognition by DDX17 facilitates miRNA processing and antiviral defense. Cell 158 ,764-777 (2014). Kazmierski, J. et al. A Baseline Cellular Antiviral State Is Maintained by cGAS and Its Most Frequent Naturally Occurring Variant rs610913. J Immunol (2022). Nostrand, E.L.V. et al. Robust transcriptome-wide discovery of RNA-binding protein binding sites with enhanced CLIP (eCLIP). Nature methods 13 ,508 - 514 (2016). Krakau, S., Richard, H. & Marsico, A. PureCLIP: capturing target-specific protein-RNA interaction footprints from single-nucleotide CLIP-seq data. Genome biology 18 ,240 (2017). Heinz, S. et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell 38 ,576-589 (2010). Machanick, P. & Bailey, T.L. MEME-ChIP: motif analysis of large DNA datasets. Bioinformatics 27 ,1696-1697 (2011). Berke, I.C., Yu, X., Modis, Y. & Egelman, E.H. MDA5 assembles into a polar helical filament on dsRNA. Proceedings of the National Academy of Sciences of the United States of America 109 ,18437 - 18441 (2012). Schilling, M. et al. RIG-I Plays a Dominant Role in the Induction of Transcriptional Changes in Zika Virus-Infected Cells, which Protect from Virus-Induced Cell Death. Cells 9 (2020). Middleton, R. et al. IRFinder: assessing the impact of intron retention on mammalian gene expression. Genome Biology 18 ,51 (2017). Lee, S. et al. Covering all your bases: incorporating intron signal from RNA-seq data. Nar Genom Bioinform 2 (2020). Fitzgerald, K.D. & Semler, B.L. Re-localization of Cellular Protein SRp20 during Poliovirus Infection: Bridging a Viral IRES to the Host Cell Translation Apparatus. PLOS Pathogens 7 ,e1002127 (2011). Álvarez, E., Castelló, A., Carrasco, L. & Izquierdo, J.M. Poliovirus 2A Protease Triggers a Selective Nucleo-Cytoplasmic Redistribution of Splicing Factors to Regulate Alternative Pre-mRNA Splicing. PLOS ONE 8 ,e73723 (2013). Ertel, K.J., Brunner, J.E. & Semler, B.L. Mechanistic consequences of hnRNP C binding to both RNA termini of poliovirus negative-strand RNA intermediates. Journal of virology 84 ,4229-4242 (2010). Gustin, K.E. & Sarnow, P. Effects of poliovirus infection on nucleo-cytoplasmic trafficking and nuclear pore complex composition. The EMBO Journal 20 ,240-249 (2001). Flather, D. & Semler, B.L. Picornaviruses and nuclear functions: targeting a cellular compartment distinct from the replication site of a positive-strand RNA virus. Frontiers in Microbiology 6 (2015). Banerjee, A.K. et al. SARS-CoV-2 Disrupts Splicing, Translation, and Protein Trafficking to Suppress Host Defenses. Cell 183 ,1325-1339.e1321 (2020). Müller-McNicoll, M. et al. SR proteins are NXF1 adaptors that link alternative RNA processing to mRNA export. Genes & Development 30 ,553-566 (2016). Ratnadiwakara, M. et al. SRSF3 confers selective processing of miR-17-92 cluster to promote tumorigenic properties in colorectal cancer. Biorxiv ,667295 (2019). Heilig, R. & Broz, P. Function and mechanism of the pyrin inflammasome. Eur J Immunol 48 ,230-238 (2018). De Maio, F.A. et al. The Dengue Virus NS5 Protein Intrudes in the Cellular Spliceosome and Modulates Splicing. PLoS Pathog 12 ,e1005841 (2016). Brunetti, J.E., Scolaro, L.A. & Castilla, V. The heterogeneous nuclear ribonucleoprotein K (hnRNP K) is a host factor required for dengue virus and Junín virus multiplication. Virus Res 203 ,84-91 (2015). Belov, G.A. et al. Early alteration of nucleocytoplasmic traffic induced by some RNA viruses. Virology 275 ,244-248 (2000). Belov, G.A. et al. Bidirectional Increase in Permeability of Nuclear Envelope upon Poliovirus Infection and Accompanying Alterations of Nuclear Pores. Journal of Virology 78 ,10166-10177 (2004). Ashraf, U., Benoit-Pilven, C., Lacroix, V., Navratil, V. & Naffakh, N. Advances in Analyzing Virus-Induced Alterations of Host Cell Splicing. Trends in microbiology 27 ,268-281 (2019). Guney, M.H. et al. IFIH1 (MDA5) is required for innate immune detection of intron-containing RNA expressed from the HIV-1 provirus. Proceedings of the National Academy of Sciences of the United States of America 121 ,e2404349121 (2024). Cortese, M. et al. Integrative Imaging Reveals SARS-CoV-2-Induced Reshaping of Subcellular Morphologies. Cell Host Microbe 28 ,853-866.e855 (2020). Seth, R.B., Sun, L., Ea, C.K. & Chen, Z.J. Identification and characterization of MAVS, a mitochondrial antiviral signaling protein that activates NF-kappaB and IRF 3. Cell 122 ,669-682 (2005). Wu, B. et al. Structural basis for dsRNA recognition, filament formation, and antiviral signal activation by MDA5. Cell 152 ,276-289 (2013). Dominguez, D. et al. Sequence, Structure, and Context Preferences of Human RNA Binding Proteins. Molecular cell 70 ,854-867 e859 (2018). Phillips, S.L., Soderblom, E.J., Bradrick, S.S. & Garcia-Blanco, M.A. Identification of Proteins Bound to Dengue Viral RNA In Vivo Reveals New Host Proteins Important for Virus Replication. mBio 7 ,e01865-01815 (2016). Ertel, K.J., Brunner, J.E. & Semler, B.L. Mechanistic consequences of hnRNP C binding to both RNA termini of poliovirus negative-strand RNA intermediates. J Virol 84 ,4229-4242 (2010). Rehwinkel, J. & Mehdipour, P. ADAR1: from basic mechanisms to inhibitors. Trends Cell Biol (2024). Rice, G.I. et al. Mutations in ADAR1 cause Aicardi-Goutières syndrome associated with a type I interferon signature. Nat Genet 44 ,1243-1248 (2012). Roulois, D. et al. DNA-Demethylating Agents Target Colorectal Cancer Cells by Inducing Viral Mimicry by Endogenous Transcripts. Cell 162 ,961-973 (2015). Chiappinelli, K.B. et al. Inhibiting DNA Methylation Causes an Interferon Response in Cancer via dsRNA Including Endogenous Retroviruses. Cell 162 ,974-986 (2015). Bowling, E.A. et al. Spliceosome-targeted therapies trigger an antiviral immune response in triple-negative breast cancer. Cell 184 ,384-403 e321 (2021). Herzner, A.-M. et al. ADAR and hnRNPC deficiency synergize in activating endogenous dsRNA-induced type I IFN responses. Journal of Experimental Medicine 218 ,e20201833 (2021). Zheng, R. et al. hnRNPM protects against the dsRNA-mediated interferon response by repressing LINE-associated cryptic splicing. Molecular cell 84 ,2087-2103 e2088 (2024). Dorrity, T.J. et al. Long 3'UTRs predispose neurons to inflammation by promoting immunostimulatory double-stranded RNA formation. Science immunology 8 ,eadg2979 (2023). Peisley, A. & Hur, S. Multi-level regulation of cellular recognition of viral dsRNA. Cell Mol Life Sci 70 ,1949-1963 (2013). Maurano, M. et al. Protein kinase R and the integrated stress response drive immunopathology caused by mutations in the RNA deaminase ADAR1. Immunity (2021). Banerjee, S. et al. OAS-RNase L innate immune pathway mediates the cytotoxicity of a DNA-demethylating drug. Proceedings of the National Academy of Sciences 116 ,201815071 (2019). Liu, C.-X. et al. Structure and Degradation of Circular RNAs Regulate PKR Activation in Innate Immunity. Cell 177 ,865-880.e821 (2019). Li, Y. et al. Ribonuclease L mediates the cell-lethal phenotype of double-stranded RNA editing enzyme ADAR1 deficiency in a human cell line. Elife 6 ,e25687 (2017). Wickenhagen, A. et al. A prenylated dsRNA sensor protects against severe COVID-19. Science ,eabj3624 (2021). Buchbender, A. et al. Improved library preparation with the new iCLIP2 protocol. Methods 178 ,33-48 (2020). Lee, F.C.Y. et al. An improved iCLIP protocol. Biorxiv ,2021.2008.2027.457890 (2021). Busch, A., Bruggemann, M., Ebersberger, S. & Zarnack, K. iCLIP data analysis: A complete pipeline from sequencing reads to RBP binding sites. Methods (San Diego, Calif.) 178 ,49-62 (2020). Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29 ,15-21 (2013). Smith, T., Heger, A. & Sudbery, I. UMI-tools: modeling sequencing errors in Unique Molecular Identifiers to improve quantification accuracy. Genome Res 27 ,491-499 (2017). Ewels, P., Magnusson, M., Lundin, S. & Käller, M. MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32 ,3047-3048 (2016). Goodstadt, L. Ruffus: a lightweight Python library for computational pipelines. Bioinformatics 26 ,2778-2779 (2010). Quinlan, A.R. & Hall, I.M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26 ,841-842 (2010). Smit, A., Hubley, R, Green, P. RepeatMasker Open-4.0. (2013-2015). Robinson, J.T. et al. Integrative genomics viewer. Nat Biotechnol 29 ,24-26 (2011). Team, R.C. R: A language and environment for statistical computing. R Foundation for Statistical Computing. (2021). Liao, Y., Smyth, G.K. & Shi, W. The R package Rsubread is easier, faster, cheaper and better for alignment and quantification of RNA sequencing reads. Nucleic Acids Res 47 ,e47 (2019). Robinson, M.D., McCarthy, D.J. & Smyth, G.K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26 ,139-140 (2010). Durinck, S., Spellman, P.T., Birney, E. & Huber, W. Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. Nat Protoc 4 ,1184-1191 (2009). Durinck, S. et al. BioMart and Bioconductor: a powerful link between biological databases and microarray data analysis. Bioinformatics 21 ,3439-3440 (2005). Robinson, M.D. & Oshlack, A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol 11 ,R25 (2010). Law, C.W., Chen, Y., Shi, W. & Smyth, G.K. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol 15 ,R29 (2014). Ritchie, M.E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic acids research 43 ,e47 (2015). McCarthy, D.J. & Smyth, G.K. Testing significance relative to a fold-change threshold is a TREAT. Bioinformatics 25 ,765-771 (2009). Phipson, B., Lee, S., Majewski, I.J., Alexander, W.S. & Smyth, G.K. Robust hyperparameter estimation protects against hypervariable genes and improves power to detect differential expression. The Annals of Applied Statistics 10 ,946-963, 918 (2016). Love, M.I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology 15 ,550 (2014). Wu, D. et al. ROAST: rotation gene set tests for complex microarray experiments. Bioinformatics 26 ,2176-2182 (2010). Chen, Y., Lun, ATL, and Smyth, GK. Differential expression analysis of complex RNA-seq experiments using edgeR. Statistical Analysis of Next Generation Sequence Data . Somnath Datta and Daniel S Nettleton (eds), Springer, New York, 2014. Additional Declarations The authors declare no competing interests. Supplementary Files supplemental.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6466919","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":444288865,"identity":"2fd30819-ccc2-40ad-b5cb-8224cbd136b0","order_by":0,"name":"Natalia G. Sampaio","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0002-8396-1354","institution":"Hudson Institute of Medical research","correspondingAuthor":true,"prefix":"","firstName":"Natalia","middleName":"G.","lastName":"Sampaio","suffix":""},{"id":444288866,"identity":"7eb7cc89-47c7-490e-a963-45ac85072e20","order_by":1,"name":"Linden J. Gearing","email":"","orcid":"","institution":"Hudson Institute of Medical research","correspondingAuthor":false,"prefix":"","firstName":"Linden","middleName":"J.","lastName":"Gearing","suffix":""},{"id":444288867,"identity":"63e01049-b62c-47ae-b057-7beda6c5dd14","order_by":2,"name":"Antonio G. Dias Junior","email":"","orcid":"","institution":"Medical Research Council Translational Immune Discovery Unit, University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Antonio","middleName":"G. Dias","lastName":"Junior","suffix":""},{"id":444288868,"identity":"da5df1b4-24ab-49d8-b478-cc2f41c6c301","order_by":3,"name":"Lise Chauveau","email":"","orcid":"","institution":"Institut de Recherche en Infectiologie de Montpellier","correspondingAuthor":false,"prefix":"","firstName":"Lise","middleName":"","lastName":"Chauveau","suffix":""},{"id":444288869,"identity":"c9486761-3448-4641-9cf0-397342fc07f3","order_by":4,"name":"Valerie Odon","email":"","orcid":"","institution":"Medical Research Council Translational Immune Discovery Unit, University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Valerie","middleName":"","lastName":"Odon","suffix":""},{"id":444288870,"identity":"15e27354-e91e-4f56-a934-dce6e4a12743","order_by":5,"name":"Chiara Cursi","email":"","orcid":"","institution":"Medical Research Council Translational Immune Discovery Unit, University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Chiara","middleName":"","lastName":"Cursi","suffix":""},{"id":444288871,"identity":"641f7dba-119e-4025-a21a-858fd08bc7b6","order_by":6,"name":"Alice Mayer","email":"","orcid":"","institution":"Medical Research Council Translational Immune Discovery Unit, University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Alice","middleName":"","lastName":"Mayer","suffix":""},{"id":444288872,"identity":"0d773933-f2fa-420a-b7ce-5d5910eb734b","order_by":7,"name":"Madara Ratnadiwakara","email":"","orcid":"","institution":"Hudson Institute of Medical Research","correspondingAuthor":false,"prefix":"","firstName":"Madara","middleName":"","lastName":"Ratnadiwakara","suffix":""},{"id":444288873,"identity":"a30d31d5-8cc9-43d3-b6f6-dff72d3ddd14","order_by":8,"name":"Minna-Liisa Änkö","email":"","orcid":"","institution":"Faculty of Medicine and Health Technology, Tampere University","correspondingAuthor":false,"prefix":"","firstName":"Minna-Liisa","middleName":"","lastName":"Änkö","suffix":""},{"id":444288874,"identity":"6cbe15d4-7b82-4844-851e-7b63c27791bf","order_by":9,"name":"Paul J. Hertzog","email":"","orcid":"","institution":"Hudson Institute of Medical Research","correspondingAuthor":false,"prefix":"","firstName":"Paul","middleName":"J.","lastName":"Hertzog","suffix":""},{"id":444288875,"identity":"77ed7aff-82d0-45fc-92a6-ab504a43151c","order_by":10,"name":"Jan Rehwinkel","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDElEQVRIie2RsWrDMBCGzxg8mc4qbZpXkPGeZ9FhSJe0swdDVQrKUpLVQ+gbdNWsIlAXhb5ABnfJlMFT6RQqu0npYAePGfSJQyDxcf9xAB7PGRLw9lIQuQNVfnxoiIcozA5QDihXTkExQAnnaxXm+QYXVwIrfNmMXnloCBQToFZ1B3u+Z4G1WxTXRlOU21SqKKNgMqBr3q3wGQ0ehb4T5FYQlBqlipMKIgX0oztYsNw5Zd8q829ctQpVsD+hlE0X3ihTA8gPXQKh+oOVO/rGjX5wSkaY0anUbhZcZPFlz/jJcpZ88kKn43Ka1HWhR/L9yZD6a3JzYVm3wn+X8o/QFTuxyHHfh8fj8Xj++AGIRmgq17LuXwAAAABJRU5ErkJggg==","orcid":"","institution":"Medical Research Council Translational Immune Discovery Unit, University of Oxford","correspondingAuthor":true,"prefix":"","firstName":"Jan","middleName":"","lastName":"Rehwinkel","suffix":""}],"badges":[],"createdAt":"2025-04-17 01:13:51","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-6466919/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6466919/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":81258267,"identity":"cba96e72-1cc3-40a4-8135-b6f0387eda13","added_by":"auto","created_at":"2025-04-24 05:34:06","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2393826,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eiCLIP isolates RNA bound by endogenous MDA5 in live cells.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e, MDA5 immunoprecipitation (IP) was performed on WT THP1 cell lysate using antibody A, B, or IgG control, followed by elution under denaturing conditions. Half of the eluted sample was set aside (IP #1), and the remainder was used for a second IP with antibodies as indicated (IP #2). Equivalent volumes of unbound fractions from the first and second IPs were collected. Cell lysate (Lys), IP and unbound fractions were analysed by western blot for MDA5.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, IRF3-KO or MDA5-KO THP1 cells were infected with EMCV (MOI=5) or left uninfected for 22 hr. Cells were lysed, and MDA5 dual IP with antibody A then B was performed. Cell lysates, unbound fractions and IP fractions were analysed by western blot for MDA5. Actin was used as a loading control. The asterisk indicates a nonspecific band.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, Lysates from mock-infected or EMCV-infected (MOI=10) WT THP1 cells, with or without UV crosslinking at 150 mJ/cm\u003csup\u003e2\u003c/sup\u003e, were treated with high or low concentrations of RNase A. MDA5 dual IP was performed, followed by RNA radiolabelling using PNK enzyme, or no PNK as a negative control. Samples were resolved by SDS-PAGE, transferred to nitrocellulose membranes and exposed to radiofilm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed-e\u003c/strong\u003e, Lysates from UV crosslinked, EMCV-infected (MOI=10) WT THP1 cells were treated with increasing concentrations of RNase A, followed by MDA5 dual IP and RNA radiolabelling. Samples were resolved by SDS-PAGE, transferred to nitrocellulose and exposed to radiofilm (\u003cstrong\u003ed\u003c/strong\u003e). High or low molecular weight (MW) sections of membrane from (\u003cstrong\u003ed\u003c/strong\u003e) were cut as indicated, and RNA extracted by proteinase K digestion, followed by TBE-Urea gel electrophoresis and radioblot (\u003cstrong\u003ee\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef\u003c/strong\u003e, Diagram of the MDA5 iCLIP method. IP = immunoprecipitation. SMI = size-matched input; input material run alongside IP sample and excised from membrane at same MW range. UMI = unique molecular identifier; 5 bases added at random to allow removal of PCR duplicates during data processing. NGS = next-generation sequencing.\u003c/p\u003e\n\u003cp\u003eAll data are representative of at least two independent experiments.\u003c/p\u003e","description":"","filename":"Fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/c19a8622b3d1a5033279c4e0.jpg"},{"id":81257331,"identity":"9a6fe5fe-475d-4292-945b-56da6de62f14","added_by":"auto","created_at":"2025-04-24 05:10:06","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1704028,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMDA5 binds cellular RNAs during EMCV infection.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e, Percentage of RNA sequences in RNA input or MDA5 iCLIP samples mapping to the host or EMCV genome.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, Genomic classification of single-nucleotide MDA5 binding sites detected by iCLIP.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec-f\u003c/strong\u003e, Sequences consisting of 50 nucleotides (nt) upstream and downstream of each MDA5 binding site (total length 101 nt) from WT EMCV-infected samples were analysed for \u003cem\u003ede novo\u003c/em\u003e RNA motifs using MEME-ChIP. Specific motifs enriched in WT-EMCV-infected samples compared to WT uninfected samples, and associated \u003cem\u003eE\u003c/em\u003e-values (adjusted \u003cem\u003ep\u003c/em\u003e-value multiplied by the number of motifs in the input file) are shown. Localisation of the motifs discovered by MEME-ChIP relative to the crosslink site (position 0, black arrow) are shown in \u003cstrong\u003ed\u003c/strong\u003e. Proportion of motifs in MDA5 binding sites from EMCV-infected cells is shown in \u003cstrong\u003ef\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee\u003c/strong\u003e, The proportion of binding sites containing Poly(A) and Poly(U) motifs in WT EMCV-infected samples when analysed with either MEME-ChIP or HOMER algorithms as above.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg-i\u003c/strong\u003e, Motifs enriched in WT uninfected samples compared to the MDA5 KO-EMCV-infected negative control were analysed as in (\u003cstrong\u003ec-e\u003c/strong\u003e). M = A or C; W = A or U. Proportion of motifs in MDA5 binding sites from uninfected cells is shown in \u003cstrong\u003ei\u003c/strong\u003e. Data points in (\u003cstrong\u003ea\u003c/strong\u003e) are represent four independent biological repeats, and bars/error bars represent the average and SD. Analyses in (\u003cstrong\u003eb-i\u003c/strong\u003e) were conducted on combined data from four independent biological repeats.\u003c/p\u003e","description":"","filename":"Fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/7c17c7cc54d6b9fc30fab79d.jpg"},{"id":81257337,"identity":"d87c99d8-73ca-483e-8103-514dfe3b320d","added_by":"auto","created_at":"2025-04-24 05:10:06","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2793556,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMDA5 is activated by host RNA during EMCV infection.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-b\u003c/strong\u003e, RIG-I-KO Huh7 cells (\u003cstrong\u003ea\u003c/strong\u003e) and \u003cem\u003eDdx58\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e MEFs (\u003cstrong\u003eb\u003c/strong\u003e) were transfected with the indicated IVT-dsRNAs, Poly(I:C), or Neo\u003csup\u003e1-99 \u003c/sup\u003eIVT-RNA, or were treated with transfection reagent alone (Lipo only). After overnight incubation, RNA was extracted and RT-qPCR performed for \u003cem\u003eIFNB1\u003c/em\u003e and \u003cem\u003eIFIT1\u003c/em\u003e (Huh7), or \u003cem\u003eIfna4\u003c/em\u003e and \u003cem\u003eIfi44\u003c/em\u003e (MEFs). Data are shown relative to \u003cem\u003eGAPDH\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, \u003cem\u003eDdx58\u003c/em\u003e\u003csup\u003e\u003cem\u003e-/-\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e \u003c/em\u003eMEFs transfected as in (\u003cstrong\u003eb\u003c/strong\u003e) were lysed and analysed by western blot for ISG15. Actin served as a loading control.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed\u003c/strong\u003e, IVT-RNAs were transfected into RIG-I-KO or RIG-I/MDA5-DKO p125-HEK \u003cem\u003eIFNB1 \u003c/em\u003ereporter cells. As negative controls, cells were treated with transfection reagent alone (Lipo only). After overnight incubation, luciferase activity in cell lysates was measured. Data from untreated cells were set to 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ee\u003c/strong\u003e, Total RNA extracted from EMCV-infected THP1 cells (22 hours, MOI = 5) was hybridised to biotinylated DNA probes antisense to EMCV positive sense RNA, and EMCV sequences were depleted by streptavidin pulldown. EMCV depleted and nondepleted (input) RNA was analysed for presence of EMCV sequences by RT-qPCR for the indicated EMCV target sequences. Data are shown relative to \u003cem\u003eGAPDH\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ef-j\u003c/strong\u003e, Wild type or MDA5-KO A549 cells were pre-treated with IFNb (30 U/ml) for 24 hours to upregulate MDA5. Ribavirin (400 μM) was added to cells for a further 1 hour. 200 ng of EMCV-depleted and non-depleted (input) RNA from EMCV-infected or uninfected THP1 cells (from \u003cstrong\u003ee\u003c/strong\u003e) was transfected into A549 cells. After 16 hours, RT-qPCR performed for the EMCV 5’ NTR \u003cstrong\u003e(f)\u003c/strong\u003e and the indicated transcripts \u003cstrong\u003e(g-j)\u003c/strong\u003e. Data are shown relative to \u003cem\u003eGAPDH\u003c/em\u003e. Statistical analysis by two-way ANOVA. ** = p \u0026lt; 0.001, *** = p \u0026lt; 0.0005, **** = p \u0026lt; 0.0001, ns = not significant.\u003c/p\u003e\n\u003cp\u003eData points in (\u003cstrong\u003ea-b\u003c/strong\u003e) and (\u003cstrong\u003ec\u003c/strong\u003e) are from two or three independent experiments, and bar graphs represent the average. Data in \u003cstrong\u003e(d)\u003c/strong\u003e is technical replicates from one experiment. Data points in (\u003cstrong\u003ee-j)\u003c/strong\u003e are two independent experiments.\u003cbr\u003e\n\u003c/p\u003e","description":"","filename":"Fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/14f67fdbf6712e43b4b98e0a.jpg"},{"id":81258836,"identity":"19455ca1-44dc-41c4-96f0-96cf36ad2fb5","added_by":"auto","created_at":"2025-04-24 05:42:06","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2485170,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMDA5 binds to host RNA during SARS-CoV-2 virus infection.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e Proportion of RNA input or MDA5 iCLIP samples mapping to the host or the SARS-CoV-2 genome. Symbols represent data from two independent biological repeats.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, Genomic classification of single-nucleotide MDA5 binding sites detected by iCLIP.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec\u003c/strong\u003e, Proportion of single-nucleotide MDA5 binding sites, or 101 nt regions (50 nt upstream and downstream of single nucleotide binding site) that are located within repetitive regions of the human genome.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ed-h\u003c/strong\u003e,101 nt regions around MDA5 binding sites, as described in (\u003cstrong\u003ec\u003c/strong\u003e), were analysed for \u003cem\u003ede novo\u003c/em\u003e RNA motifs using MEME-ChIP. Specific motifs enriched in SARS-CoV-2-infected samples compared to uninfected samples (\u003cstrong\u003ed-e\u003c/strong\u003e), and in uninfected samples (\u003cstrong\u003eg-h\u003c/strong\u003e), as detected by MEME-ChIP. Associated \u003cem\u003eE\u003c/em\u003e-values (adjusted \u003cem\u003ep\u003c/em\u003e-value multiplied by the number of motifs in the input file) are shown. Localisation of the motifs discovered by MEME-ChIP relative to the crosslink site (position 0, black arrow) are shown in (\u003cstrong\u003ee\u003c/strong\u003e and\u003cstrong\u003e h\u003c/strong\u003e). Proportion of motifs in MDA5 binding sites from uninfected or SARS-CoV-2-infected cells is shown in \u003cstrong\u003ef\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ei,\u003c/strong\u003e The distance between all MDA5 binding sites from SARS-CoV-2-infected samples to their nearest binding site in the opposite strand were calculated and compared to randomly generated sites across the genome.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ej\u003c/strong\u003e, As (\u003cstrong\u003ei\u003c/strong\u003e) but depicting the number of sites within 100 nucleotides of each other.\u003c/p\u003e\n\u003cp\u003eData points in (\u003cstrong\u003ea\u003c/strong\u003e) are represent two independent biological repeats, and bars/error bars represent the average and SD. Analyses in (b-j) were conducted on combined data from two independent biological repeats.\u003c/p\u003e","description":"","filename":"Fig4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/7cf0f40396f5ff4ff6192b38.jpg"},{"id":81258835,"identity":"60f1df59-8fa8-45df-8bf9-7f9b8173c22b","added_by":"auto","created_at":"2025-04-24 05:42:06","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":2160520,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVirus infection upregulates introns in the cytoplasm, which are enriched for MDA5 binding sites.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea-f\u003c/strong\u003e, IRF3-KO THP1 cells were left untreated, infected with EMCV (MOI = 2) for 20 hours, or treated with IFNb (100 U/ml) for 6 hours (EMCV and untreated, n=4 independent biological repeats; IFNb, n=3). RNA was extracted from the cytoplasmic fractions, rRNA-depleted and sequenced. Differential expression of genes based on intronic read counts determined by \u003cem\u003eindex\u003c/em\u003eanalysis in EMCV-infected cells (\u003cstrong\u003ea\u003c/strong\u003e) and IFNb-treated cells (\u003cstrong\u003ed\u003c/strong\u003e) compared to untreated cells, with upregulated (red) and downregulated (blue) genes highlighted. Comparison of differential gene expression based on intronic and exonic read counts in EMCV-infected cells (\u003cstrong\u003eb, c\u003c/strong\u003e) and IFNb-treated cells (\u003cstrong\u003ee, f\u003c/strong\u003e) compared to untreated cells. The number of genes up- or down-regulated on an intronic level, an exonic level or both are indicated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eg-i\u003c/strong\u003e, Calu-3 cells were left untreated or infected with SARS-CoV-2 (MOI = 0.1) for 48 hours (n=4 independent biological repeats) and the analysis shown in (\u003cstrong\u003ea-c\u003c/strong\u003e) was repeated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ej-k\u003c/strong\u003e, Barcode plots showing the association of EMCV (\u003cstrong\u003ej\u003c/strong\u003e) or SARS-CoV-2 (\u003cstrong\u003ek\u003c/strong\u003e) iCLIP peaks with intron-level differential gene expression by \u003cem\u003eindex\u003c/em\u003e analysis. Vertical lines indicate introns (or genes) that overlapped with highly scoring intronic iCLIP peaks, weighted by iCLIP score and arranged according to differential expression statistics. The overall enrichment is indicated by the enrichment line above, and significance was calculated using ROAST.\u003c/p\u003e\n\u003cp\u003eAll sequencing data is from four independent biological repeats.\u003c/p\u003e","description":"","filename":"Fig5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/00ddd9a89f1e09aec4895722.jpg"},{"id":81258271,"identity":"7c1ef1c8-2ddb-485c-ba79-43e443105f56","added_by":"auto","created_at":"2025-04-24 05:34:06","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1858319,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverexpression of the splicing factor SRSF3 prevents MDA5 activation during EMCV infection.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea\u003c/strong\u003e, LIM1215 cells stably overexpressing either GFP or SRSF3-GFP were infected with EMCV with the indicated MOIs for 48 hours, and cell viability was assessed by MTT assay.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eb\u003c/strong\u003e, MDA5, MAVS, and GFP protein levels in LIM1215 cells overexpressing either GFP or SRSF3-GFP, untreated or treated with IFNb (100 U/ml) for 24 hours, were analysed by western blot. Tubulin served as a loading control.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ec-k\u003c/strong\u003e, LIM1215 cells overexpressing either GFP or SRSF3-GFP were infected with EMCV at the indicated MOIs for 48 hours. RNA was extracted and RT-qPCR performed for\u003cem\u003e \u003c/em\u003eindicated mRNAs and the EMCV 5’ non-translated region (NTR). Data are shown relative to \u003cem\u003eGAPDH\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eData in (\u003cstrong\u003ea\u003c/strong\u003e) are mean from two independent biological repeats, with error bars representing standard error. Data in (\u003cstrong\u003eb\u003c/strong\u003e) is from a single experiment. Data points in (\u003cstrong\u003ec-k\u003c/strong\u003e) are from two or three independent biological repeats. Statistical analysis by two-way ANOVA. * = p \u0026lt; 0.05, ** = p \u0026lt; 0.001, *** = p \u0026lt; 0.0005, ns = not significant.\u003c/p\u003e","description":"","filename":"Fig6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/fb1de9217db3b6883c64f03b.jpg"},{"id":81258841,"identity":"6f6de9cc-909d-4d4a-871e-140d7dd922b9","added_by":"auto","created_at":"2025-04-24 05:43:15","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":14044704,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/48b911fe-4fc7-4d69-a785-9139914efb37.pdf"},{"id":81257340,"identity":"1db8ae7d-17d8-4f2c-b38e-5e34053b58b3","added_by":"auto","created_at":"2025-04-24 05:10:06","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3359407,"visible":true,"origin":"","legend":"","description":"","filename":"supplemental.docx","url":"https://assets-eu.researchsquare.com/files/rs-6466919/v1/9aa8464e88ba8061a8fd945b.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"MDA5 guards against infection by surveying cellular RNA homeostasis","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePathogens are first detected by pattern recognition receptors (PRRs) that activate immune signal transduction pathways and thereby mediate the host response to infection. Canonically, pathogen-derived molecules known as pathogen-associated molecular patterns (PAMPs) activate PRRs \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Viral infections are recognised by a class of PRRs that detect unusual nucleic acids \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. A crucial response activated by nucleic acid sensors is the production of type I interferons (T1-IFNs) \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. These cytokines are essential for protection against all viruses. This system is fine-tuned to differentiate infection from the homeostatic state, to elicit protective immunity followed by return to homeostasis. Excessive or uncontrolled T1-IFN responses fail to protect against infection and lead to long-term damage and disease \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Characterising the initiation of T1-IFN responses is therefore fundamental to our ability to understand infectious and other diseases, and to design antiviral therapies.\u003c/p\u003e \u003cp\u003eAmongst the PRRs that detect viral infection are the retinoic acid-inducible gene I (RIG-I)-like receptors (RLRs) \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. This family of RNA sensors includes RIG-I, melanoma differentiation-associated protein 5 (MDA5) and laboratory of genetics and physiology 2 (LGP2). RLRs are primarily located in the cytoplasm. All RLRs have a central helicase domain and a carboxy-terminal domain, which together detect unusual, immunostimulatory RNA molecules. MDA5 and RIG-I also contain two tandem amino-terminal caspase activation and recruitment domains (CARDs), which mediate downstream signalling through interaction with the adaptor mitochondrial antiviral-signalling protein (MAVS). Activated MAVS oligomerises and recruits other factors including the kinase TBK1, that in turn activate interferon regulatory factors (IRFs) and the NF-kB pathway. Ultimately, expression of T1-IFNs and other immune response genes is induced. LGP2 lacks CARDs and modulates MDA5 and RIG-I signalling.\u003c/p\u003e \u003cp\u003eRIG-I and MDA5 are activated by different viral infections \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. For example, influenza A virus infection is sensed by RIG-I; conversely, picornaviruses such as encephalomyocarditis virus (EMCV) or rhinovirus are detected by MDA5 \u003csup\u003e7\u003c/sup\u003e. Viruses from other families are recognised by both RIG-I and MDA5, including important human pathogens such as members of the \u003cem\u003eFlaviviridae\u003c/em\u003e (e.g. Zika virus, hepatitis C virus), \u003cem\u003eParamyxoviridae\u003c/em\u003e (e.g. measles virus) and \u003cem\u003eCoronaviridae\u003c/em\u003e (e.g. SARS-CoV) \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. SARS-CoV-2 infection has been reported to be partially or exclusively sensed by MDA5 \u003csup\u003e9\u0026ndash;12\u003c/sup\u003e. Additionally, MDA5 can be activated by infection with non-RNA viruses \u003csup\u003e\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e and other pathogens such as \u003cem\u003ePlasmodium\u003c/em\u003e sp. \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e, \u003cem\u003eMycobacterium tuberculosis\u003c/em\u003e \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e or \u003cem\u003eAspergillus fumigatus\u003c/em\u003e \u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. The important role of MDA5 in the human immune response is highlighted by cases of inherited MDA5 deficiency leading to increased and/or life-threatening susceptibility to viral infections, for example with Rhinovirus \u003csup\u003e\u003cspan additionalcitationids=\"CR21 CR22\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eAn important aspect of RLR signalling is the ability of these receptors to distinguish immunostimulatory RNAs accumulating in infected cells from the RNA content of cells during homeostasis. RIG-I is activated by RNAs with triphosphate or diphosphate groups at the 5\u0026rsquo; end \u003csup\u003e\u003cspan additionalcitationids=\"CR25\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Additional features of RIG-I-stimulatory RNAs include the lack of methylation marks and base-pairing \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Such RNA species do not occur abundantly in cells, as most cellular RNAs are processed at the 5\u0026rsquo; end; for example, mRNAs are capped and methylated. In contrast, some viral RNAs such as the genomes of influenza A virus contain these features, allowing them to be recognised by RIG-I \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. As such, RNA sensing by RIG-I can be conceptualised by the paradigm of PAMP detection by PRRs. It is noteworthy that some DNA viruses and retroviruses that do not produce viral 5\u0026rsquo;-(P)PP-RNAs activate RIG-I indirectly by causing the accumulation of unprocessed and/or mislocalised cellular non-coding RNAs with 5\u0026rsquo;-PPP moieties \u003csup\u003e\u003cspan additionalcitationids=\"CR31 CR32\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe RNA species that activate MDA5 are less well defined \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Early work showed that MDA5 recognises infection with picornaviruses and the synthetic double-stranded (ds) RNA mimic polyriboinosinic:polyribocytidylic acid (Poly(I:C)) \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. In particular, MDA5 is essential for innate immune sensing of long Poly(I:C) \u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Moreover, purified MDA5 forms multimeric filaments on long strands of dsRNA \u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. RNA viruses with positive-sense genomes, including picornaviruses, replicate by producing a negative-sense copy of their genome that serves as a template for synthesis of progeny positive-sense genomes. Annealing of positive and negative sense RNAs can form long dsRNA, known as replicative form dsRNA, which was reported to activate MDA5 \u003csup\u003e37, 38\u003c/sup\u003e. Together, these results indicate that, in virally infected cells, MDA5 is activated by long viral dsRNA.\u003c/p\u003e \u003cp\u003eHowever, other findings contradict this view, suggestive of a more complex mechanism. Indeed, we previously demonstrated that high molecular weight complex RNA produced during infection, rather than merely dsRNA, activates MDA5 \u003csup\u003e39\u003c/sup\u003e. Moreover, studies analysing the RNAs bound by MDA5 or LGP2 during viral infections suggested that single stranded sections of viral genomes, rather than dsRNA forms, are detected \u003csup\u003e\u003cspan additionalcitationids=\"CR41\" citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. However, these studies are constrained by methodological factors, including non-stringent purification of MDA5-RNA complexes, lack of precision in determining MDA5 binding sites, and/or exogenous protein overexpression.\u003c/p\u003e \u003cp\u003eGiven that MDA5 can sense infection with many different virus families and other pathogens, understanding the nature of MDA5\u0026rsquo;s RNA agonists is crucial for our comprehension of innate immune responses. We addressed this by employing individual-nucleotide resolution ultraviolet crosslinking and immunoprecipitation (iCLIP) \u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e to stringently and specifically identify RNAs bound by MDA5 during virus infection. We revealed the \u003cem\u003ein situ\u003c/em\u003e RNA targets and the precise protein-RNA binding sites of endogenous MDA5 in EMCV- and in SARS-CoV2-infected cells. To our surprise, we found that MDA5 bound mostly to host-derived RNA during virus infection. This host RNA was largely intronic, and close to \u003cem\u003eAlu\u003c/em\u003e repetitive elements. EMCV and SARS-CoV-2 infections resulted in the presence of intron-containing pre-mRNA in the cytoplasm and MDA5 binding sites determined with iCLIP were enriched amongst these intronic RNA sequences. Finally, overexpression of the splicing factor SRSF3 during virus infection prevented MDA5 activation, whereas removal of viral RNA from infected cellular RNA did not. This suggests that MDA5 detects imbalances in RNA processing occurring in virus-infected cells by binding to endogenous host intronic RNA accumulating in the cytoplasm.\u003c/p\u003e"},{"header":"Results","content":"\u003ch2\u003eDual IP of endogenous MDA5 isolates MDA5-bound RNA\u003c/h2\u003e\n\u003cp\u003ePrevious studies of RNAs binding to and/or activating MDA5 relied on recombinant or overexpressed protein \u003csup\u003e42, 45, 46\u003c/sup\u003e. To identify RNA ligands of endogenous MDA5, we screened cell lines for expression of MDA5 and found that monocytic THP1 cells expressed MDA5 protein at baseline (Fig. S1a). EMCV specifically activates MDA5 \u003csup\u003e7, 47\u003c/sup\u003e and infects THP1 cells, evident from staining with the J2 monoclonal antibody that detects dsRNA (Fig. S1b), a signature of viral replication \u003csup\u003e39, 48, 49\u003c/sup\u003e, and from accumulation of viral RNA (Fig. S1c). Infection induced \u003cem\u003eIFN\u003c/em\u003e\u003cem\u003eb\u003c/em\u003e mRNA and interferon-stimulated genes (ISGs) (Fig. S1d). Transfection of total cellular RNA extracted from EMCV-infected THP1 cells activated \u003cem\u003eIFNB1\u003c/em\u003e promoter-driven luciferase expression in HEK293 reporter cells \u003csup\u003e50\u003c/sup\u003e in an MDA5-dependent manner (Fig. S1e). Together, these data show that MDA5-stimulatory RNA was present in EMCV-infected THP1 cells.\u003c/p\u003e\n\u003cp\u003eTo obtain an in-depth view of RNAs binding MDA5 during viral infection, we employed iCLIP \u003csup\u003e43\u003c/sup\u003e. This technique allows deep sequencing of RNAs bound by a protein of interest, with resolution of the nucleotide site(s) where the protein binds, and has been successfully used for other dsRNA binding proteins including Dicer \u003csup\u003e51\u003c/sup\u003e and DDX17 \u003csup\u003e52\u003c/sup\u003e. As a control, MDA5 knock-out (KO) THP1 cells were generated using CRISPR/Cas9 (Fig. S1a, f). MDA5-KO THP1 did not induce \u003cem\u003eIFNB1\u003c/em\u003e mRNA in response to transfection with total RNA extracted from EMCV-infected HeLa cells, but responded normally to RIG-I stimulation with \u003cem\u003ein vitro\u003c/em\u003e transcribed RNA \u003csup\u003e29\u003c/sup\u003e (Fig. S1g). In MDA5-KO cells, T1-IFN responses to EMCV infection were undetectable (Fig. S1d). Furthermore, we surmised that high infectivity levels would be required to detect RNA bound to MDA5. However, EMCV infectivity was low in wild-type (WT) THP1 cells (Fig. S1h). IRF3 signals downstream of multiple nucleic acid sensors including cGAS that is required for baseline T1-IFN responses in THP1 cells \u003csup\u003e53\u003c/sup\u003e. Loss of IRF3 may therefore render cells more susceptible to EMCV without impairing MDA5-RNA interactions. Indeed, infectivity levels increased ten-fold in IRF3-KO THP1 cells \u003csup\u003e50\u003c/sup\u003e (Fig. S1i-k). We therefore used IRF3-KO THP1 cells to investigate MDA5 ligands generated during EMCV infection.\u003c/p\u003e\n\u003cp\u003eTo isolate endogenous MDA5, we raised a panel of monoclonal antibodies. We identified two antibodies that precipitated native and denatured MDA5, termed antibody A (clone 16) and antibody B (clone 22), respectively. To increase the stringency and specificity of MDA5 isolation, we used a dual immunoprecipitation (IP) method. Native MDA5 was first isolated using antibody A, and then eluted with a denaturing solution containing high urea and detergent concentrations. Next, the eluate containing denatured protein was used for a second IP with antibody B, which was specific for the denatured protein (Fig. 1a). This dual IP successfully pulled down MDA5 from IRF3-KO cells (Fig. 1b).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNext, we performed MDA5 dual IP on lysates from UV-crosslinked cells, and radioactively labelled the bound RNA using Polynucleotide kinase (PNK). Gel electrophoresis and radioblot analysis showed a high molecular weight (MW) smear migrating more slowly than the MW of MDA5 (135 kDa), which was dependent on both UV crosslinking and PNK treatment (Fig. 1c). The intensity of this smear was enhanced by virus infection, indicating increased RNA binding to MDA5 in infected cells (Fig. 1c). RNase A treatment reduced the smear, presumably due to the bound RNA being partially degraded to smaller oligonucleotides protected by the protein (Fig. 1c, d). Extraction and proteinase K digestion of either high MW or low MW sections from the membrane produced RNA of varying lengths, with intermediate and low RNase A concentrations yielding RNA of more than 100 nucleotides (Fig. 1e). We selected the low RNase A treatment condition to capture MDA5-bound RNA in the 100-500 nucleotide length range, which is suitable for iCLIP and sequencing \u003csup\u003e43\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eWe then applied our dual IP approach to a modified iCLIP protocol \u003csup\u003e43\u003c/sup\u003e (Fig. 1f). An adaptor for reverse transcription was added to RNA at the 3\u0026rsquo; end, and the RNA was radiolabelled at the 5\u0026rsquo; end, followed by isolation of high MW RNA by gel electrophoresis and radioblot as described above. An aliquot of pre-IP input material was processed alongside the IP samples and served as a \u0026lsquo;size-matched input\u0026rsquo; control for comparison to IP samples \u003csup\u003e54\u003c/sup\u003e. Next, the RNA was reverse transcribed into cDNA using primers containing a sample-specific barcode for multiplexing and a unique molecular identifier (UMI) for identification of PCR duplicates. Of note, reverse transcription is halted at the RNA-protein crosslink site, which allows identification of nucleotide positions bound by MDA5. The cDNA was purified and size-selected by electrophoresis, circularised and cleaved to produce cDNA containing forward and reverse PCR sites for amplification and sequencing (Fig. 1f).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eMDA5 binds host-derived RNA during EMCV infection\u003c/h2\u003e\n\u003cp\u003eWe employed our endogenous MDA5 iCLIP method to identify RNAs specifically bound by MDA5 during virus infection. IRF3-KO THP1 cells were left uninfected or were infected with EMCV. As a negative control for IP specificity, MDA5-KO THP1 cells were used. EMCV infectivity was lower in MDA5-KO cells compared to IRF3-KO cells (henceforth referred to as MDA5-WT) (Fig. S1i-k), likely due to baseline T1-IFN production via cGAS-IRF3 \u003csup\u003e53\u003c/sup\u003e, priming cells against infection. Nevertheless, these cells still contained viral RNA (Fig. 2a), and thus served as a control for MDA5 IP specificity. Sequencing reads were processed and mapped to combined human and virus genomes to simultaneously identify both viral and host sequences, followed by removal of PCR duplicates and extraction of crosslink sites (Fig. S2a). We obtained 10\u003csup\u003e4\u003c/sup\u003e-10\u003csup\u003e6\u003c/sup\u003e uniquely mapped, deduplicated reads per sample with four independent biological repeats (Fig. S2b, c).\u003c/p\u003e\n\u003cp\u003eTo our surprise, we found that the majority (\u0026gt;99%) of RNA sequences detected in iCLIP samples mapped to the human rather than the viral genome (Fig. 2a). There was no enrichment of EMCV RNA in the MDA5 IP compared to the RNA input control samples. We next identified MDA5-RNA crosslink sites, herein referred to as MDA5 binding sites, using PureCLIP \u003csup\u003e55\u003c/sup\u003e that compares input and IP sequencing reads. MDA5 binding sites were only identified in host RNA, and not in viral RNA. The distribution of MDA5 binding sites between intergenic regions, introns, untranslated regions and coding sequences was similar between the three samples, and more than half of all binding sites were in introns (Fig. 2b).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe then employed HOMER \u003csup\u003e56\u003c/sup\u003e and MEME-ChIP \u003csup\u003e57\u003c/sup\u003e to identify motifs enriched in MDA5-bound RNA. Based on MDA5\u0026rsquo;s footprint on dsRNA of 14-15 nt \u003csup\u003e58\u003c/sup\u003e and the notion of MDA5 filament formation \u003csup\u003e36\u003c/sup\u003e, we focussed on 101 nucleotide-length regions containing MDA5 binding sites centrally (i.e., 50 nt upstream and downstream of the crosslinked nucleotide). Both algorithms identified an enrichment of Poly(A) and Poly(U) sequences in MDA5-binding sequences in EMCV-infected cells (Fig. 2c, Fig. S3a). These Poly(A)/(U) motifs were positioned near, but not directly overlapping, the MDA5 crosslinked nucleotide (Fig. 2d). Between 30% and 50% of all MDA5 binding sequences contained either a Poly(A) or a Poly(U) motif, depending on the control and motif algorithm applied (Fig. 2e, f). Importantly, these motifs were enriched compared to uninfected, MDA5-sufficient cells and to infected, MDA5-deficient cells, indicative of specificity. Additionally, motifs that contained GGUU and AACC sequences were specifically enriched in MDA5 binding sites from uninfected WT cells when compared to the infected MDA5-KO control (Fig. 2g, h, Fig S3b). These motifs were detected in 40-50% of binding sites in uninfected samples (Fig. 2i, Fig. S3c). The GGUU/AACC motifs were also detected in MDA5 binding sites from EMCV-infected WT samples (Fig. 2f, Fig. S3c), but were not enriched when compared to MDA5 binding sites from uninfected WT cells. This indicates that the GGUU/AACC motifs present in the infected cells likely represent RNA derived from uninfected cells within the infected cell population; these motifs may thus represent low affinity and/or low abundance MDA5 ligands.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eSynthetic host RNAs encompassing MDA5-binding sites activate MDA5 \u003cem\u003ein cellulo\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eTo validate that host RNAs containing MDA5 binding sites activate MDA5 in cells, we generated by \u003cem\u003ein vitro\u0026nbsp;\u003c/em\u003etranscription (IVT) ~500 nucleotide length RNAs from forward and reverse genomic sequences. These contained selected top-scoring sites centrally (Tab. S1), were annealed to form dsRNAs (Fig. S4a, b) and were then transfected into RIG-I-deficient cells to prevent RIG-I stimulation by 5\u0026rsquo;-triphosphate groups on RNAs produced by IVT. In both Huh7 cells and mouse embryonic fibroblasts (MEFs), a range of T1-IFN and ISG mRNA induction was observed, suggesting that some regions were more stimulatory than others (Fig. S5a-d). For example, regions K and L were largely inert, whereas regions M and N induced strong T1-IFN and ISG responses (Fig. 3a, b). Interestingly, unlike K and L, regions M and N encompassed repetitive sequences (LINE1 and MIR, respectively). Of the 16 regions tested, ten contained repetitive sequences (Tab. S1) and the corresponding IVT-dsRNAs tended to induce stronger T1-IFN responses than the remaining IVT-dsRNAs (Fig. S5c, d). We validated the immunostimulatory potential of M and N dsRNAs at protein level by showing that both induced expression of ISG15 and promoted ISGylation (Fig. 3c). Furthermore, RIG-I/MDA5 double-knockout reporter cells did not respond to any IVT-dsRNA tested, demonstrating that the response was MDA5-dependent (Fig. 3d and S5e-g). Finally, transfection of IVT-ssRNAs did not stimulate cells (Fig. S5h). Taken together, some of the host RNA regions that bound MDA5 upon viral infection, particularly those containing repetitive sequences, induced an MDA5-dependent T1-IFN response when re-introduced as dsRNAs into uninfected cells, validating our iCLIP data.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eDepletion of viral RNA does not diminish MDA5 activation\u003c/h2\u003e\n\u003cp\u003eTransfection of total RNA extracted from cells infected with MDA5-stimulating viruses induces MDA5 activation (Fig. S1e, g and \u003csup\u003e41, 50\u003c/sup\u003e). To further test the contributions of viral and host RNAs to the activation of MDA5, we extracted total RNA from EMCV-infected THP1 cells and used biotinylated antisense probes to deplete positive sense EMCV RNA by a streptavidin pulldown. This resulted in approximately 1000-fold depletion of viral RNA (Fig. 3e). Input and depleted RNA samples were then transfected into wild-type or MDA5-KO A549 cells \u003csup\u003e59\u003c/sup\u003e. These recipient cells were pre-treated with ribavirin to preclude EMCV replication \u003csup\u003e39\u003c/sup\u003e. As expected, viral RNA was not present in cells transfected with depleted RNA samples (Fig. 3f). Nonetheless, we found that depletion of EMCV positive sense RNA did not prevent the activation of MDA5 in recipient cells, as shown by an equal induction of the T1-IFN response by both input and depleted RNA samples (Fig. 3g-j). This response was dependent on MDA5, since MDA5-KO cells did not respond. Transfection of RNA from uninfected cells did not induce a T1-IFN response. It is noteworthy that our antisense probes were complementary to positive sense EMCV RNA and that the RT-qPCR used to assess depletion detected both negative and positive sense EMCV RNA. Therefore, it is possible that small amounts of negative sense viral RNA, potentially paired with the remaining trace amounts of positive sense EMCV RNA, were present after depletion. Nonetheless, we favour the interpretation that endogenous host RNA generated during virus infection, rather than viral RNA, stimulates MDA5. This endogenous host RNA likely arises from improperly spliced RNA accumulating as a result of virus infection.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eCellular RNAs associate with MDA5 during SARS-CoV-2 infection\u003c/h2\u003e\n\u003cp\u003eWe and others previously reported that SARS-CoV-2 infection activates MDA5 \u003csup\u003e9-12\u003c/sup\u003e. To identify MDA5 agonists during SARS-CoV-2 infection, we used Calu-3 cells, an adenocarcinoma-derived lung epithelial cell line. We performed iCLIP with dual MDA5 IP as described above with some methodological improvements to enrich for bound RNA (detailed in methods). We obtained ~10\u003csup\u003e7\u003c/sup\u003e uniquely mapped, deduplicated reads per sample (SARS-CoV-2 infected or uninfected cells; two biological replicates) (Fig. S6a, b).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eApproximately 10% of RNA sequences in the SARS-CoV-2-infected RNA input samples mapped to the SARS-CoV-2 genome (Fig. 4a). However, there was no enrichment in SARS-CoV-2 RNA in the MDA5 IP samples (Fig. 4a). We did not detect MDA5 binding sites in viral RNA; instead, MDA5 bound only host RNA. Over 50% of binding sites were found in introns (Fig. 4b), followed by intergenic regions (~25%) and protein coding regions (~10%). Given that MDA5 has been previously suggested to bind repetitive RNA \u003csup\u003e46\u003c/sup\u003e, we determined the proportion of binding sites present in repetitive regions. Around 30% of binding sites were located within repetitive regions, which included SINEs (\u003cem\u003eAlu\u003c/em\u003e, MIR), LINEs (L1, L2) and other repeat categories (Fig. 4c). We then extended our analysis to 101 nt regions spanning binding sites and found that ~50% of these sites overlapped with an \u003cem\u003eAlu\u003c/em\u003e repeat, without major increases in any of the other repeat categories. Thus, in agreement with the findings for EMCV infection, we observed that during SARS-CoV-2 infection, MDA5 binds host, rather than viral, RNA. This host RNA is mostly intronic and enriched in \u003cem\u003eAlu\u003c/em\u003e repeats close to the MDA5-binding site.\u003c/p\u003e\n\u003cp\u003eWe next analysed the same 101 nucleotide regions containing MDA5 binding sites using motif-finding algorithms. Poly(U) and Poly(A) motifs were enriched in SARS-CoV-2-infected samples (Fig. 4d-f, Fig. S3d-f). These motifs were approximately 10 nucleotides away from MDA5 binding sites (Fig. 4e). Motifs containing AACC or GGUU sequences were identified in both uninfected and SARS-CoV-2-infected samples at a distance of approximately 10 nucleotides from MDA5 binding sites (Fig. 4f-h, Fig. S3d-f). However, these were not enriched when using uninfected binding sites as a \u0026lsquo;background control\u0026rsquo; for binding sites from infected cells. Infected samples contained both infected and uninfected cells (Fig. S6c, d), suggesting that MDA5 binds AACC/GGUU motifs in uninfected cells. In sum, similar sequence motifs were enriched near MDA5 binding sites in both the SARS-CoV-2 and EMCV datasets.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSince base-pairing \u0026ndash; resulting in the formation of \u0026lsquo;dsRNA-like\u0026rsquo; stretches \u0026ndash; can occur between Poly(A) and Poly(U) motifs, we analysed the proximity between identified motifs, to determine whether inter- or intra-strand base-pairing was possible. We found that of the 4,579 Poly(A) and Poly(U) motifs identified by MEME-ChIP in the SARS-CoV-2-infected samples, 259 (5.7%) occurred within 200 nucleotides of another complementary motif, either on the same or opposite strand. Of 869 AACC/GGUU motifs found in the uninfected samples, 154 (18%) occurred within 200 nucleotides of another complementary motif on the opposite strand, whereas only 2 were close to a complementary motif on the same strand, suggesting inter-strand complementarity. We also investigated the potential for bidirectional transcription that may have occurred at MDA5 binding sites, potentially generating dsRNA species. We defined \u0026lsquo;overlapping\u0026rsquo; binding sites as any two binding sites within 100 nt of each other on opposite strands and compared this to randomly generated sites across the genome. There was no difference in the number of overlapping binding sites in uninfected cells compared to random control, but 0.4% of binding sites from SARS-CoV-2 infected cells were within 100 nt of another site on the opposite strand, compared to 0.1% in randomly generated sites, suggesting a small proportion of MDA5 binding sites in infected cells may have been generated from bidirectional transcription (Fig. 4i). Interestingly, when we further investigated these closely occurring MDA5 binding sites, we found that 41 of 90 (45%) occurred between 10-15 nt from each other (Fig. 4j). Given the \u0026lsquo;footprint\u0026rsquo; of MDA5 on dsRNA is estimated to be 14-15 nt \u003csup\u003e58\u003c/sup\u003e, we speculate that these binding sites could represent nucleotides crosslinked at both edges of a single MDA5 protein bound to a segment of dsRNA. Combined, our analysis of the RNAs bound by MDA5 during SARS-CoV-2 infection found enrichment in sequences with the potential to form intra- and inter-strand dsRNA structures, which may constitute cellular ligands for MDA5.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eIntron-containing pre-mRNAs are increased in the cytoplasm of virus-infected cells\u003c/h2\u003e\n\u003cp\u003eSince we found that many MDA5 binding sites were in introns (Figs. 2b and 4b), we next tested if virus infection itself affected the presence of introns in the cytoplasm, where MDA5 is localised. We extracted and sequenced cytoplasmic RNA from THP1 and Calu-3 cells infected with EMCV or SARS-CoV-2, respectively, or uninfected controls. As an additional control, we treated cells with 100 U/ml IFNb. We analysed changes in the abundance of intronic sequences using two methods: IRFinder \u003csup\u003e60\u003c/sup\u003e and \u003cem\u003eindex\u0026nbsp;\u003c/em\u003e\u003csup\u003e61\u003c/sup\u003e. IRFinder interrogates intron-retained reads within normally spliced mRNA, whereas \u003cem\u003eindex\u003c/em\u003e examines the overall level of intron reads on a per gene basis, which can cover both intron retention and unspliced pre-mRNA. Analysis with IRFinder showed a modest amount of intron retention after EMCV infection, but not after IFNb treatment, compared to untreated cells (Fig. S7a-b). \u003cem\u003eIndex\u003c/em\u003e showed that both EMCV and SARS-CoV-2 infection had a significant effect on intron expression, which was much greater than the effect of IFNb treatment (Fig. 5a-i). Indeed, compared to untreated cells, virus infection specifically induced an upregulation in introns in the cytoplasm (Fig. 5b, h), which was not detected in cells treated with IFNb (Fig. 5e). Importantly, there was a significant positive correlation between the introns upregulated/retained in the cytoplasm of virus-infected cells with introns that contained MDA5 binding sites detected by iCLIP (Fig. 5j, k and S6c). Thus, we found that virus infection causes significant changes to the levels of introns in the cytoplasm. These upregulated introns were also associated with MDA5 binding during virus infection, suggesting that they stimulate MDA5. \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eRescue of mRNA splicing during infection abrogates MDA5 activation\u003c/h2\u003e\n\u003cp\u003eSeveral viruses that activate MDA5 also modulate the subcellular distribution of splicing factors, such as SRSF3 \u003csup\u003e62, 63\u003c/sup\u003e, and nuclear ribonucleoproteins, such as hnRNPC \u003csup\u003e64, 65\u003c/sup\u003e, thus significantly affecting host RNA metabolism \u003csup\u003e66\u003c/sup\u003e. Moreover, SARS-CoV-2 infection disrupts host RNA splicing, resulting in intron retention \u003csup\u003e67\u003c/sup\u003e. We therefore hypothesised that virus infections lead to an imbalance in host RNA processing and metabolism, resulting in an increase of intron-containing RNA in the cytoplasm that activates MDA5. To test this idea, we aimed to restore splicing by overexpressing SRSF3 (also known as SRp20), a global regulator of pre-mRNA splicing and mRNA export \u003csup\u003e68\u003c/sup\u003e, which was previously shown to be targeted by poliovirus, a picornavirus that activates MDA5 \u003csup\u003e62\u003c/sup\u003e. SRSF3-GFP-overexpressing LIM1215 cells and GFP-overexpressing control cells\u0026nbsp;\u003csup\u003e69\u003c/sup\u003e were equally susceptible to EMCV-induced cell death and expressed equivalent levels of MDA5 and MAVS (Fig. 6a, b). At viral doses that did not result in cell death, the T1-IFN response was blunted in SRSF3-GFP cells compared to GFP control cells (Fig. 6c-j). This was not simply due to lower infection levels in SRSF3-GFP cells, as equivalent or increased viral RNA levels were present in SRSF3-GFP cells compared to GFP control cells (Fig. 6k). These data suggest that \u0026ndash; when RNA splicing is \u0026ldquo;forced\u0026rdquo; to occur despite virus infection \u0026ndash; then MDA5 activation is lost. \u0026nbsp;\u003cbr\u003e \u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eMDA5 is an important PRR \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Despite its discovery as an RNA sensor of virus infection almost 20 years ago \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e, the PAMP detected by MDA5 during infection has remained obscure and controversial, rendering MDA5 one of the few remaining \u0026lsquo;orphan\u0026rsquo; PRRs. Here, we propose that MDA5 monitors the homeostasis of cellular RNA processing and is activated upon perturbations caused by infections (Fig. S8). Conceptually, this finding can be described as \u0026lsquo;guarding\u0026rsquo;, which involves the detection of pathogen-induced perturbations rather than PAMPs. Several plant immune receptors and the mammalian Pyrin inflammasome employ this mechanism; for example, the latter detects bacterial toxin-induced Rho guanosine triphosphatase inactivation \u003csup\u003e\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e\u003c/sup\u003e. We found that defects in RNA splicing in virally infected cells triggered MDA5 via recognition of cellular intronic sequences accumulating in the cytoplasm. Our work thus defines innate immune guarding as a principle underpinning cytoplasmic RNA sensing during virus infection.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThis finding was unexpected as MDA5 is often described as a sensor of viral dsRNA accumulating in infected cells. However, our observations were consistent between two different cell types, monocytic THP1 cells and lung adenocarcinoma Calu-3 cells, and two virus infections, EMCV and SARS-CoV-2. We cannot exclude the possibility that MDA5 detects viral dsRNAs in different settings such as in cells infected with other viruses, in other cell types or at different time points during infection. Nonetheless, our model \u0026ndash; in which MDA5 guards posttranscriptional control \u0026ndash; provides an attractive explanation as to why MDA5 can detect many different infections. Indeed, SARS-CoV-2 inhibits splicing and induces intron retention \u003csup\u003e\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e. Dengue virus \u003csup\u003e\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e, \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e\u003c/sup\u003e, poliovirus \u003csup\u003e\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e, \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e\u003c/sup\u003e and rhinovirus \u003csup\u003e\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e, \u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e\u003c/sup\u003e also affect host RNA homeostasis by causing cytoplasmic relocalisation of splicing and nuclear RNA binding factors. Furthermore, MDA5 is activated by DNA virus infections, including by herpes simplex virus-1, vaccinia virus and hepatitis B virus \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. DNA viruses do not generate a replicative form viral dsRNA but interfere with splicing \u003csup\u003e\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e\u003c/sup\u003e. We posit that activation of MDA5 occurs when the fidelity of RNA processing is impacted upon virus infection, either through direct viral antagonists of splicing or indirectly through cellular stress \u003csup\u003e\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e\u003c/sup\u003e. It will be important to test this in future work for different virus families, as will be the analysis of MDA5\u0026rsquo;s RNA agonists in non-viral infections \u003csup\u003e\u003cspan additionalcitationids=\"CR17 CR18\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Interestingly, it was recently suggested that unspliced RNA derived from integrated HIV-1 proviral DNA activates MDA5 \u003csup\u003e76\u003c/sup\u003e. Although formally of viral origin, unspliced HIV-1 transcripts are generated by the cell\u0026rsquo;s transcription machinery; thus, this finding mirrors our observation that MDA5 is activated by intron-containing cellular RNAs.\u003c/p\u003e \u003cp\u003eViral dsRNAs are often associated with viral RNA binding proteins. Moreover, for many viruses, replication takes place in replication factories. For example, SARS-CoV-2 establishes double membrane vesicles, in which its polymerase produces new viral RNA \u003csup\u003e\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e\u003c/sup\u003e. As such, how MDA5 could gain access to viral dsRNA is questionable, in particular given that \u003cem\u003ein vitro\u003c/em\u003e data suggest multiple MDA5 molecules need to bind the same RNA for activation \u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Moreover, even if MDA5 were to gain access to viral dsRNA in replication factories, it is difficult to envisage how MDA5 would then be able to interact with mitochondrially localised MAVS \u003csup\u003e\u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e78\u003c/span\u003e\u003c/sup\u003e. Our findings resolve these conundrums: detection of inappropriately processed host RNAs explains how MDA5 can sense viral infections when viral RNAs are shielded by viral proteins or within replication factories.\u003c/p\u003e \u003cp\u003eWe found specific sequence motifs in host RNA that were associated with MDA5, including Poly(A)/(U)-rich sequences. Structurally, MDA5 binds dsRNA in a sequence-independent manner, due to interactions with the phosphodiester backbone \u003csup\u003e\u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e79\u003c/span\u003e\u003c/sup\u003e. It is possible that the motifs discovered here are first bound by other RNA-binding proteins that are then replaced by MDA5. Indeed, several RNA-binding proteins specifically bind to Poly(U) tracts \u003csup\u003e\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e80\u003c/span\u003e\u003c/sup\u003e, including some known to be relocalised to the cytoplasm during virus infection such as hnRNPC \u003csup\u003e\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e, \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e82\u003c/span\u003e\u003c/sup\u003e. Thus, when viruses usurp host RNA factors for their own replication and induce cytoplasmic relocalisation of nuclear factors, this may in fact result in these nuclear factors bringing with them to the cytoplasm unspliced RNAs that then activate MDA5.\u003c/p\u003e \u003cp\u003eAlthough MDA5 was first discovered as a sensor of virus infection, it is also activated in autoinflammatory disease. In the absence of ADAR1, which catalyses the deamination of adenosines to inosines in dsRNA to break base-pairing, MDA5 detects cellular dsRNA species \u003csup\u003e\u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e83\u003c/span\u003e\u003c/sup\u003e. This leads to sterile induction of T1-IFN. Accordingly, inactivating mutations in \u003cem\u003eADAR1\u003c/em\u003e cause the T1-IFN-mediated Aicardi-Gouti\u0026egrave;res syndrome, a neurodevelopmental disorder \u003csup\u003e\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e\u003c/sup\u003e. Host repetitive RNA, in particular inverted \u003cem\u003eAlu\u003c/em\u003e repeats, which can form base-paired structures, have been proposed as RNA ligands of MDA5 in this condition \u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e, consistent with our observation that many MDA5 binding sites are in proximity of \u003cem\u003eAlu\u003c/em\u003e elements. Moreover, MDA5 was also found to be stimulated by RNA transcripts from endogenous retroviruses (ERVs) when cancer cells were treated with demethylating agents (e.g. azacytidine derivatives). These compounds derepress ERVs, resulting in the formation of dsRNAs that stimulate MDA5 \u003csup\u003e85, 86\u003c/sup\u003e. Interestingly, treatment of cancer cells with inhibitors of the spliceosome results in intron-retention in cytoplasmic RNA and concomitant activation of T1-IFN \u003cem\u003evia\u003c/em\u003e MAVS, potentially explaining the therapeutic efficacy of drugs targeting splicing \u003csup\u003e\u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e87\u003c/span\u003e\u003c/sup\u003e. Moreover, depletion of the splicing factors hnRNPC and hnRNPM induces MDA5 activation \u003csup\u003e\u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e88\u003c/span\u003e, \u003cspan citationid=\"CR89\" class=\"CitationRef\"\u003e89\u003c/span\u003e\u003c/sup\u003e. These observations provide important parallel lines of evidence to the work presented here and suggest that sensing of aberrant RNA processing is a universal mechanism explaining MDA5 activation during both infections and in sterile settings. It is also noteworthy that in addition to intronic RNA other types of non-coding RNA may contribute to MDA5 activation. For example, in neuronal cell types, 3\u0026rsquo;UTRs are extended and have been suggested to activate MDA5 due to the presence of inverted repeat \u003cem\u003eAlu\u003c/em\u003e elements \u003csup\u003e\u003cspan citationid=\"CR90\" class=\"CitationRef\"\u003e90\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eAn important methodological advance of our work is the use a-MDA5 monoclonal antibodies that allow identification of the RNAs that bind endogenous MDA5 \u003cem\u003ein situ\u003c/em\u003e in cells. Whilst here applied to virus infections, our protocol can be used in any human cell type that expresses MDA5, including primary cells and patient samples. The a-MDA5 antibodies and iCLIP protocol are available as an open resource to the research community. We hope they will be applied in future to discover MDA5 agonists in autoinflammation and cancer treatment.\u003c/p\u003e \u003cp\u003eLike MDA5, other nucleic acid sensors have also been described to detect viral dsRNA. Protein kinase R (PKR) and members of 2'-5'-oligoadenylate synthetase (OAS) family bind dsRNA and are activated by virus infections \u003csup\u003e\u003cspan citationid=\"CR91\" class=\"CitationRef\"\u003e91\u003c/span\u003e\u003c/sup\u003e, many of which also activate MDA5, and have been associated with similar non-infectious pathologies \u003csup\u003e\u003cspan additionalcitationids=\"CR93 CR94\" citationid=\"CR92\" class=\"CitationRef\"\u003e92\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e95\u003c/span\u003e\u003c/sup\u003e. These receptors, rather than inducing a T1-IFN response, largely act to restrict virus spread by inducing global RNA degradation or shutting down mRNA translation. OAS1 was recently found to bind host intronic and repetitive RNA, in addition to viral RNA, during SARS-CoV-2 infection \u003csup\u003e\u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e96\u003c/span\u003e\u003c/sup\u003e. Whether PKR and OASs are \u0026ndash; like MDA5 \u0026ndash; activated by host intronic RNAs in the context of different virus infections is worth investigating.\u003c/p\u003e \u003cp\u003eTaken together, our work shifts the understanding of MDA5 activation: rather than detecting viral RNA as a PAMP, as is the case with other PRRs, we propose that MDA5 is a sensor of cellular RNA homeostasis. When disturbances in this homeostasis occur as a result of virus infection, MDA5 is triggered by improperly spliced cellular RNAs. These findings have important bearings on treatment approaches for both infectious and non-infectious diseases that involve MDA5 activation.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCell culture and virus infection\u003c/h2\u003e \u003cp\u003eCells were cultured at 37\u0026deg;C and 5% CO\u003csub\u003e2\u003c/sub\u003e and routinely screened for mycoplasma contamination. THP1 (gift from Vincenzo Cerundolo) and LIM1215 (gift from Minna-Liisa \u0026Auml;nk\u0026ouml;) cells were maintained in RPMI (Sigma Aldrich) supplemented with 10% v/v foetal calf serum (FCS) and 2 mM L-glutamine (Gibco). Huh7 (gift from Jane McKeating), A549 (gift from Georg Kocks), BHK-21 (gift from Alain Townsend), p125HEK \u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e and MEF (gift from Shizou Akira) cells were maintained in DMEM supplemented with 10% v/v foetal calf serum (FCS) and 2 mM L-glutamine (Gibco). Calu-3 cells (gift from Caroline Goujon) were maintained in MEM (Sigma Aldrich) supplemented with 10% v/v foetal calf serum (FCS), 2 mM L-glutamine (Gibco), 1x sodium pyruvate (Gibco) and 1x non-essential amino acids (Gibco). For generation of MDA5 and RIG-I knockout cells using CRISPR/Cas9 technology, sgRNAs cloned into pX458-Ruby (Addgene 110164, deposited by Dr. Philip Hublitz) and described earlier \u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e were used. LTX Transfection kit (Life Technologies) was used to transfect 2 x 10\u003csup\u003e6\u003c/sup\u003e cells THP1 cells with 4 \u0026micro;g of plasmid, or 2 x 10\u003csup\u003e5\u003c/sup\u003e Huh7 and p125HEK cells using 2.5 \u0026micro;g of plasmid according to manufacturer\u0026rsquo;s instructions. After 24 hr, mRuby-positive cells were single-cell sorted into 96-well plates containing fresh medium for Huh7 and p125HEK cells; or 50% fresh medium and 50% filtered conditioned medium from THP1 parental cells for THP1 cells. Surviving clones were expanded and screened by immunoblotting for ablation of target protein, and Sanger sequenced to ascertain mutation at the gene locus. IRF3 knockout THP1 cells have been previously described \u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e. RIG-I knockout MEFs cells were a gift S. Akira. LIM1215 cells stably overexpressing SFSR3-GFP or GFP were described previously\u003csup\u003e\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e\u003c/sup\u003e. EMCV (gift from C. Reis e Sousa) was produced in BHK-21 cells, and harvested from supernatant by centrifugation at 10,000 g and 0.2 \u0026micro;m filtration before single-aliquot freezing at \u0026minus;\u0026thinsp;80\u0026deg;C. Viral content was quantified by plaque assay on BHK-21 cells. SARS-CoV-2 Victoria/02/2020 (passage 5) was produced in Vero E6 and titrated by plaque assay as described previously\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. For cytoplasmic RNA extraction, the BetaCoV/France/IDF0372/2020 isolate was supplied by Sylvie van der Werf and the National Reference Centre for Respiratory Viruses hosted by Institut Pasteur (Paris, France). The patient sample from which strain BetaCoV/France/IDF0372/2020 was isolated was provided by X. Lescure and PY. Yazdanpanah from the Bichat Hospital, Paris, France. Moreover, the strain BetaCoV/France/IDF0372/2020 was supplied through the European Virus Archives goes Global (Evag) platform, a project that has received funding from the European Union\u0026rsquo;s Horizon 2020 research and innovation programme under grant agreement No 653316. Experiments using this strain were performed at the CEMIPAI BSL3 facility (UAR 3725 CNRS Montpellier University).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eAntibodies and Reagents\u003c/h2\u003e \u003cp\u003eAll chemicals and reagents used were from Merck unless otherwise indicated. MDA5 antibodies were generated by C. Song and B. Jin. After screening, clone 16 (also referred to as Antibody A) and clone 22 (also referred to as Antibody B) were selected for native and denatured IP, respectively, and clone 17 was chosen for immunoblotting. Hybridoma cells expressing clone 16 and clone 17 were grown in bioreactors (CELLline) as per manufacturer\u0026rsquo;s instructions, and stored at -80\u0026deg;C. For clone 22, the hybridoma cells were non-recoverable, therefore the antibody was \u003cem\u003ede novo\u003c/em\u003e sequenced by protein mass spectrometry and synthesized using recombinant protein expression in mammalian cells (Rapid Novor and Absolute Antibody). The J2 antibody against dsRNA was from Scicons, the SRSF3 antibody was from Sigma (WH0006428M8), the MAVS (PA5-17256) and GFP (A21311) antibodies were from Thermo Fisher Scientific, the SARS-CoV-2 Spike antibody (GTX632604) was from GeneTex and the MxA antibody (clone M143, MABF938) was from Merck. The ISG15 antibody was a gift from K. P. Knobeloch. HRP-coupled secondary antibodies were sheep‐α‐mouse and donkey‐α‐rabbit (both GE Healthcare, 1:3000).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eConfocal Microscopy\u003c/h2\u003e \u003cp\u003eTHP1 cells were grown on sterile poly-L-lysine (0.01%, Merck) treated glass coverslips (5 x 10\u003csup\u003e5\u003c/sup\u003e cells per well in 24-well plates) with phorbol 12-myristate 13-acetate (PMA, 10 ng/ml; Invivogen) for 24 hr. Cells were infected with EMCV at MOI\u0026thinsp;=\u0026thinsp;10 for 16 hr, washed with PBS, and fixed for 15 min with 4% formaldehyde in cytoskeleton-stabilizing buffer (CSB; 5 mM KCl, 137 mM NaCl, 4 mM NaHCO\u003csub\u003e3\u003c/sub\u003e, 0.4 mM KH\u003csub\u003e2\u003c/sub\u003ePO\u003csub\u003e4\u003c/sub\u003e, 1.1 mM Na\u003csub\u003e2\u003c/sub\u003eHPO\u003csub\u003e4\u003c/sub\u003e, 2 mM MgCl\u003csub\u003e2\u003c/sub\u003e, 5 mM PIPES, 2 mM EGTA and 5.5 mM glucose in water). Cells were washed with PBS, permeabilized with 0.1% Triton-X 100 in CSB for 20 min, and incubated in 0.1M glycine in CSB for 10 min. Cells were washed four times between all subsequent steps with PBS for 5 min each wash. Cells were incubated in blocking buffer (1% BSA, 10% normal goat serum (Abcam) in PBS) for 1 hr, then for 2 hr with anti-dsRNA antibody (J2, 1:200 in blocking buffer), and 1 hr with goat anti-mouse AlexaFluor 488 (1:500 in blocking buffer; Life Technologies, A11029) and AlexaFluor 633 Phalloidin (1:40 in blocking buffer; Thermo Fisher Scientific A22284). Slides were mounted using ProLong Gold antifade mountant with DAPI (Thermo Fisher Scientific, P36931), and imaged using a Zeiss 780 inverted confocal microscope.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eImmunoblotting\u003c/h2\u003e \u003cp\u003eCells were lysed with lysis buffer (10 mM Tris, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 5 uM ZnCl\u003csub\u003e2\u003c/sub\u003e, 0.5% IGEPAL and complete protease inhibitor cocktail (Roche)) and protein was quantified by BCA assay (Pierce). For LIM1215 analysis of cytoplasmic and nuclear protein, cells were lysed in RIPA buffer (50mM Tris-HCl pH 8, 150mM NaCl, 1% IGEPAL, 0.5% Sodium deoxycholate, 0.1% SDS). Samples were denatured using NuPAGE LDS sample loading buffer (Life Technologies) and 10% 2-mercaptoethanol, and heating at 95\u0026deg;C for 5 min. Samples were resolved by electrophoresis on 4\u0026ndash;12% Bis-Tris gels with MOPS Running Buffer (Life Technologies NuPAGE system) and transferred to nitrocellulose membrane by electrophoresis at 100V for 2 hr in cold transfer buffer (Life Technologies) with 10% methanol. Membranes were blocked with 0.05% IGEPAL in Tris-buffered saline (TBS-N; 50 mM NaCl, 50 mM Tris-HCl, pH 7.6) containing 5% non-fat milk (5% milk TBS-N) for 1 hr, and probed with primary and HRP-conjugated secondary antibodies diluted in 5% milk TBS-N for 1 hr at room temperature or overnight at 4\u0026deg;C, with rotation. Membranes were washed four times in TBS-N for 5 min each wash after each antibody incubation. Proteins were visualized on iBright (Thermo Fisher Scientific) after exposure to Western LightningPlus-ECL chemiluminescent reagent (PerkinElmer).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eMDA5 iCLIP\u003c/h2\u003e \u003cp\u003eConditions were maintained RNase-free throughout the experiment, conducted as described in Huppertz et al.\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e, with some modifications. IRF3-KO or MDA5-KO THP1 cells (2 x 10\u003csup\u003e7\u003c/sup\u003e cells in 20 cm plates, 10\u0026ndash;20 plates per condition) were incubated with PMA overnight, then infected with EMCV (MOI\u0026thinsp;=\u0026thinsp;5) for 22 hr, or left uninfected (IRF3-KO only). Cells were washed twice with PBS and UV crosslinked using 150 mJ/cm\u003csup\u003e2\u003c/sup\u003e in a Spectrolinker XL-1500 (Spectronics Corp). Cells were scraped, centrifuged at 400 g for 5 min, and PBS supernatant was discarded. Cell pellets were flash frozen and stored at \u0026minus;\u0026thinsp;80\u0026deg;C until ready for use. Cell pellets were lysed in lysis buffer (10 mM Tris, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 5 \u0026micro;M ZnCl\u003csub\u003e2\u003c/sub\u003e, 1% IGEPAL and complete protease inhibitor cocktail (Roche)) using equivalent of 400 \u0026micro;l of buffer per plate of cells. Samples were incubated on ice for 10 min and clarified by centrifugation at 10,000 g for 10 min. Protein concentration was determined by BCA assay. 5000 \u0026micro;g of clarified cell lysate was used per condition. RNase A (Affymetrix, 70194Y; 20 U/\u0026micro;l) was added at 1:10,000 dilution to samples and incubated for 5 min, followed by cooling and addition of 1:80 dilution of RNAsin Plus RNase inhibitor (Promega; N2611). Anti-MDA5 mAb 16 was covalently coupled to Dynabeads (Invitrogen, 14311D) according to manufacturer\u0026rsquo;s instructions, using 9 mg Dynabeads and 180 ug of mAb 16 per sample. mAb 16 antibody-coupled Dynabeads were washed with high salt buffer (50 mM Tris pH 7.4, 1M NaCl, 1% IGEPAL, 0.1% SDS, 0.5% deoxycholate) and lysis buffer, and incubated with protein samples for 2 hr at 4\u0026deg;C with rotation. 1% of sample was collected and stored at 4\u0026deg;C, for use as size-matched input control (SMI). Samples were washed twice for 5 min with high salt buffer, and once with PNK buffer (20 mM Tris pH 7.4, 10 mM MgCl\u003csub\u003e2\u003c/sub\u003e, 0.2% Tween-20). Supernatant was removed, 300 \u0026micro;l of urea cracking buffer (50 mM Tris pH 7.4, 6M urea, 1% SDS, 25% PBS) was added, and samples incubated with shaking for 3 min at 65\u0026deg;C, before neutralisation with 3 ml of T-20 IP buffer (50 mM Tris pH 7.4, 150 mM NaCl, 0.5% Tween-20, 0.1 mM EDTA) and addition of RNase and protease inhibitors. A second round of immunoprecipitation was done using anti-MDA5 mAb 22 antibody preincubated with Protein G Dynabeads (Life Technologies; 900 \u0026micro;l beads and 90 \u0026micro;g of antibody) according to manufacturer\u0026rsquo;s instructions. Samples were incubated with beads for 2 hr at 4\u0026deg;C with rotation, and washed twice for 5 min with high salt buffer, and twice with PNK buffer. RNA in samples was 3\u0026rsquo; end dephosphorylated using T4 PNK (New England Biolabs, M0201L) for 20 min, with addition of RNase inhibitor and Turbo DNase I (Life Technologies AM2238, 2U/ul). Samples were washed twice with RNA ligase buffer (5mM Tris HCl pH 7.5, 1 mM MgCl\u003csub\u003e2\u003c/sub\u003e), and iCLIP 3' linker (Trilink, O-30050-03) was ligated to 3\u0026rsquo; end of RNA molecules using T4 RNA Ligase 1 High Concentration (New England Biolabs, M0437M) with the addition of 2.6% DMSO, 40% PEG-8000, and 2% RNase inhibitor, for 2 hr at 25\u0026deg;C with shaking. Samples were washed twice with high salt buffer, changing tubes, and twice with PNK buffer. 10% of sample was set aside for radiolabelling using T4 PNK and g-32P-ATP (Hartmann Analytic; FP-301; 3000Ci/mmol; 10mCi/ml) according to manufacturer\u0026rsquo;s instructions. IP and SMI samples were denatured using sample loading buffer and DTT, and incubation for 10 min at 70\u0026deg;C. Samples were resolved on 4\u0026ndash;12% Bis-Tris gels (Life technologies; NP0322BOX) using MOPS SDS running buffer (Life technologies; NP0001) for 2 hr at 150 V. Samples were transferred to nitrocellulose for 4 hr at 100 V using cold Transfer Buffer (Life technologies; NP0006) with 10% methanol. Membrane was exposed to film for signal visualisation, and the radiograph was used as a guide to excise samples from membrane.\u003c/p\u003e \u003cp\u003eSamples of MW\u0026thinsp;\u0026gt;\u0026thinsp;135 kDa were excised, and membrane was cut into small pieces and placed in a microcentrifuge tube. RNA was released from membrane by protein digestion using Proteinase K (Roche, 03115844001) for 1 hr at 50\u0026deg;C, according to manufacturer\u0026rsquo;s instructions. RNA was purified by phenol:chloroform:iaa (P3803-100ML) with phaselock gel tubes (5 Prime), and ethanol precipitation. For SMI samples, protocol was done as described \u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e. IP and SMI samples were reverse-transcribed with primers shown in Tab. S1, using TGIRT enzyme (InGex; CM0101-50) according to manufacturer\u0026rsquo;s instructions. Samples were ethanol precipitated, resuspended in TBE-urea sample buffer (Life Technologies; LC8676) and resolved on 6% TBE urea gels (Life Technologies; EC6865BOX) for 40 min at 180V. cDNA sized between 80 and 200 nt was excised, fragmented, and incubated in 400 \u0026micro;l diffusion buffer (0.5 M ammonium acetate, 10 mM magnesium acetate, 1 mM EDTA, 0.1% SDS) for 30 min at 50\u0026deg;C, shaking. Sample was clarified by centrifugation in SpinX columns (Costar) containing two 10 mm glass pre-filters (Whatman, 1823-010), and purified by phenol:chloroform:iaa and ethanol precipitation. cDNA samples were circularized using CircLigase II (Cambio, CL9021K) and BamHI digested (New England Biolabs) as described \u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. Samples were PCR amplified using the P3 and P5 Solexa primers and Accuprime PCR Master Mix (Thermo Fisher Scientific), starting at 20 cycles and increasing in 1\u0026ndash;2 cycle steps, until a signal was detectable by Tapestation D1000 High sensitivity kit. Libraries were concentrated using the MiniElute PCR Purification kit (Qiagen) and resolved on a 6% TBE gel (Life Technologies, EC6265BOX) for 30 min at 180V. Gel was stained with SYBRGold (Thermo Fisher Scientific), and fragments between 100 and 250 bp were excised, extracted from gel as above, and purified using QIAQuick gel extraction kit (Qiagen). Samples were quantified by Qubit High Sensitivity DNA kit (Thermo Fisher Scientific, Q32851) and KAPA Library quantification Kit (Illumina, 07960093001). Samples were equalized to 4 nM, and combined at a ratio of 10:1 IP samples to SMI samples, in order to increase sequencing from IP samples. Libraries were sequenced on Illumina NextSeq500 using 75 cycle High Output Kit v2 (TG-160-2105). Each biological repeat was sequenced on a separate chip.\u003c/p\u003e \u003cp\u003eFor MDA5 iCLIP from SARS-CoV-2-infected and uninfected Calu-3 cells, the protocol above was followed, with some modifications as follows \u003csup\u003e\u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e97\u003c/span\u003e, \u003cspan citationid=\"CR98\" class=\"CitationRef\"\u003e98\u003c/span\u003e\u003c/sup\u003e. 2.5 x 10\u003csup\u003e7\u003c/sup\u003e cells in 20 cm plates (2 plates per condition) were left uninfected or infected with SARS-CoV-2 at an MOI of 0.1 for 48 hr. After dual IP and adaptor ligation, samples were directly treated with proteinase K, without SDS-PAGE size selection. Samples were reverse transcribed with primers shown in Tab. S2 \u003csup\u003e98\u003c/sup\u003e, which contained longer UMI and barcode sequences, and carbon spacers to stop the progression of PCR during amplification, removing the need for linearisation after circularisation of libraries. A pre-amplification PCR step of 6 cycles prior to library size selection was performed to increase retention of unique sequences \u003csup\u003e\u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e97\u003c/span\u003e\u003c/sup\u003e. Libraries were then size-selected using ProNex Chemistry (Promega; NG2001), PCR amplified, and size-selected again as described \u003csup\u003e\u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e97\u003c/span\u003e\u003c/sup\u003e. Libraries were sequenced on Illumina NextSeq 2000 using 100 cycle P3 Reagents (20040559). All samples were sequenced on a single chip.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eiCLIP data processing and analysis\u003c/h2\u003e \u003cp\u003eA diagram of iCLIP data processing is shown in Fig. S2a, and largely followed recommendations outlined in \u003csup\u003e\u003cspan citationid=\"CR99\" class=\"CitationRef\"\u003e99\u003c/span\u003e\u003c/sup\u003e. Fastx-trimmer (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://hannonlab.cshl.edu/fastx_toolkit/\u003c/span\u003e\u003cspan address=\"http://hannonlab.cshl.edu/fastx_toolkit/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to remove low quality barcode regions as described in \u003csup\u003e\u003cspan citationid=\"CR99\" class=\"CitationRef\"\u003e99\u003c/span\u003e\u003c/sup\u003e. Flexbar (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/seqan/flexbar\u003c/span\u003e\u003cspan address=\"https://github.com/seqan/flexbar\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to simultaneously 3\u0026rsquo; end trim, demultiplex, and extract UMIs. FastQC was used to determine quality of sequencing reads pre- and post-processing (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.bioinformatics.babraham.ac.uk/projects/fastqc/\u003c/span\u003e\u003cspan address=\"https://www.bioinformatics.babraham.ac.uk/projects/fastqc/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). A custom genome and gtf file containing both the human genome (hg38 assembly) and the EMCV genome (DQ288856.1) were prepared. Samples were mapped against the combined human-EMCV genome using STAR \u003csup\u003e\u003cspan citationid=\"CR100\" class=\"CitationRef\"\u003e100\u003c/span\u003e\u003c/sup\u003e with the following parameters: STAR --runMode alignReads --runThreadN 8 --limitBAMsortRAM 10000000000 --outSAMattributes All --outStd BAM_SortedByCoordinate --outSAMtype BAM SortedByCoordinate --outFilterType BySJout --outReadsUnmapped Fastx --outSAMattrRGline ID:foo --alignEndsType Extend5pOfRead1 --outFilterMismatchNoverReadLmax 0.04 --outFilterMismatchNmax 999 --outFilterMultimapNmax 1 --sjdbOverhang maxReadLength-1 --outSJfilterReads Unique. Samples were indexed using samtools and deduplicated based on UMI using umi_tools dedup \u003csup\u003e\u003cspan citationid=\"CR101\" class=\"CitationRef\"\u003e101\u003c/span\u003e\u003c/sup\u003e. MultiQC was used to quantify rates of mapping \u003csup\u003e\u003cspan citationid=\"CR102\" class=\"CitationRef\"\u003e102\u003c/span\u003e\u003c/sup\u003e. A custom computational pipeline was written using ruffus \u003csup\u003e\u003cspan citationid=\"CR103\" class=\"CitationRef\"\u003e103\u003c/span\u003e\u003c/sup\u003e to automate the sample processing described above. This is available for download at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/natsampaio/iCLIP_pipeline\u003c/span\u003e\u003cspan address=\"https://github.com/natsampaio/iCLIP_pipeline\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eTo determine crosslinking sites, the PureCLIP peak-calling algorithm was applied \u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e. Chromosomes 1\u0026ndash;6 were used for the algorithm parameter learning, and CL motifs and SMI controls were incorporated. For motif analysis, FASTA sequences containing 50 bp either side of crosslinking site were obtained using bedtools \u003csup\u003e\u003cspan citationid=\"CR104\" class=\"CitationRef\"\u003e104\u003c/span\u003e\u003c/sup\u003e commands slop and getfasta. Sequences present in the negative control samples (MDA5-KO cells or uninfected cells) and present in rRNA and tRNA (database from SILVA; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.arb-silva.de/\u003c/span\u003e\u003cspan address=\"https://www.arb-silva.de/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) were subtracted from the sequences from EMCV-infected cells using bedtools slop. The resulting sequences were analysed using the MEME suite (5.0.1) meme-chip algorithm, with -norc and -centrimo-local parameters included \u003csup\u003e\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e. The same sequences were also analysed using HOMER \u003csup\u003e\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e function findMotifsGenome.pl, with -rna. Basic analysis compared test sequences to automatically generated random genetic background. Test sequences were also compared to either MDA5-KO or uninfected sequences using the -bg option. Genomic context of sequences was determined using the bedtools command intersect to quantify number of sequences overlapping with annotated introns, exons, 5\u0026rsquo; UTR or 3\u0026rsquo; UTR in the genome. Sequences that contained crosslink sites were analysed using RepeatMasker \u003csup\u003e\u003cspan citationid=\"CR105\" class=\"CitationRef\"\u003e105\u003c/span\u003e\u003c/sup\u003e to quantify repetitive regions, and bedtools intersect using a database of annotated repetitive regions in the genome. Integrative Genomics Viewer (IGV 2.8.9) \u003csup\u003e\u003cspan citationid=\"CR106\" class=\"CitationRef\"\u003e106\u003c/span\u003e\u003c/sup\u003e was used to visualize data.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eFlow cytometry\u003c/h2\u003e \u003cp\u003eTHP1 cells were grown in 12-well plates (1 x 10\u003csup\u003e6\u003c/sup\u003e cells per well) and stimulated for 24 hr with PMA (10 ng/ml). Cells were infected with EMCV using MOIs and incubation times specified in figure legends, prior to addition of cold PBS containing 1% FCS (FACS Buffer). After 10 min incubation at 4\u0026deg;C, cells were lifted by pipetting, and washed with FACS buffer. For all subsequent steps, reagents were diluted in FACS buffer unless stated otherwise, and washed twice with FACS buffer between steps. Cells were stained with Live/Dead Fixable Aqua Cell Stain (1:200 in PBS, Life Technologies, L34957) combined with FcR block (1:200 in PBS, eBioscience), fixed in 4% formaldehyde (10 min), and permeabilised in 0.1% Triton-X (20 min). Cells were stained with J2 dsRNA antibody (1:200, 30 min) and goat anti-mouse AlexaFluor-488 (1:500, 30 min; Life Technologies, A11029), and resuspended in CellFix (1:10 in water; BD, 340181). Cells were analysed by flow cytometry on an Attune NxT Flow Cytometer (Thermo Fisher Scientific) and FlowJo software (BD).\u003c/p\u003e \u003cp\u003eCalu-3 cells were grown in 6-well plates (1x10\u003csup\u003e6\u003c/sup\u003e cells per well) and infected with SARS-CoV-2 at an MOI of 0.1 for 48 hr. Cells were washed with PBS, lifted using Trypsin for 15 min at 37\u0026deg;C and fixed using formalin for 20 min at RT. Cells were stained in separate wells with anti-SARS-CoV-2 Spike (1:200) and anti-MxA (1:300) antibodies for 30 min at 4\u0026deg;C, then with goat anti-mouse IgG AlexaFluor-647 (1:500; Life Technologies) and goat anti-mouse IgG2a AlexaFluor-647 (1:500, BioLegend) for 30 min at 4\u0026deg;C using Saponin buffer (PBS, 1% BSA, 0.1% Saponin) for all antibody incubations, and washed twice in the same buffer between steps. Cells were resuspended in FACS buffer and analysed by flow cytometry using a NovoCyte flow cytometer (ACEA Biosciences Inc.) and FlowJo software (BD).\u003c/p\u003e \u003cp\u003e \u003cspan type=\"ItalicUnderline\" class=\"ItalicUnderline\" name=\"Emphasis\"\u003eIn vitro\u003c/span\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eRNA transcription and purification\u003c/span\u003e\u003c/p\u003e \u003cp\u003eFor production of \u003cem\u003ein vitro\u003c/em\u003e transcribed RNA (IVT-RNA) from human genome, genomic DNA was extracted from THP1 cells using DNeasy Blood \u0026amp; Tissue Kit (Qiagen, 69504) and used as PCR templates. Herculase II PCR Kit (Agilent, 600677) was used to PCR amplify desired regions with addition of T7 polymerase promoter sequence. MegaScript T7 Kit (Invitrogen, AM1334) was used for production of IVT-RNA according to manufacturer\u0026rsquo;s instructions. RNA was purified using RNeasy Mini kit (Qiagen, 74104), and quantified by Nanodrop (Thermo Fisher Scientific) and stored at \u0026minus;\u0026thinsp;80\u0026deg;C in single-use aliquots. RNA was visualised by electrophoresis on denaturing agarose gels and staining with SYBR Gold. To produce dsRNA, forward and reverse complementary IVT-RNA strands were heated to 90\u0026deg;C and cooled at 1\u0026deg;C/30 sec until 25\u0026deg;C, then used immediately for cell transfection.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eCell transfection of IVT-RNA\u003c/h2\u003e \u003cp\u003eCells were grown in 12-well plates (1.75 x 10\u003csup\u003e6\u003c/sup\u003e cells per well) for 24 hr, and transfected with 175 ng of IVT-RNA, 87.5 ng of high MW Poly(I:C) (Invivogen, tlrl-pic), or 35 ng of Neo\u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5 CR6 CR7 CR8 CR9 CR10 CR11 CR12 CR13 CR14 CR15 CR16 CR17 CR18 CR19 CR20 CR21 CR22 CR23 CR24 CR25 CR26 CR27 CR28 CR29 CR30 CR31 CR32 CR33 CR34 CR35 CR36 CR37 CR38 CR39 CR40 CR41 CR42 CR43 CR44 CR45 CR46 CR47 CR48 CR49 CR50 CR51 CR52 CR53 CR54 CR55 CR56 CR57 CR58 CR59 CR60 CR61 CR62 CR63 CR64 CR65 CR66 CR67 CR68 CR69 CR70 CR71 CR72 CR73 CR74 CR75 CR76 CR77 CR78 CR79 CR80 CR81 CR82 CR83 CR84 CR85 CR86 CR87 CR88 CR89 CR90 CR91 CR92 CR93 CR94 CR95 CR96 CR97 CR98\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR99\" class=\"CitationRef\"\u003e99\u003c/span\u003e\u003c/sup\u003e IVT-RNA \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e using 0.7 \u0026micro;l of Lipofectamine 2000 (Thermo Fisher Scientific, 11668027) according to manufacturer\u0026rsquo;s instructions. Huh7 cells were treated with 30 U/ml IFN-A/D (R\u0026amp;D Systems) for 24 hr prior to transfection. RNA was extracted after 24 hr using RNeasy Plus Mini Kit (Qiagen, 74136) according to manufacturer\u0026rsquo;s instructions, quantified by Nanodrop, and stored at \u0026minus;\u0026thinsp;80\u0026deg;C.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eRT-qPCR\u003c/h2\u003e \u003cp\u003eRNA (1 \u0026micro;g) was converted into cDNA using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific) and Oligo-dT primers (Invitrogen) according to manufacturer\u0026rsquo;s instructions. cDNA was diluted to 100 ng/\u0026micro;l and quantitative PCR was performed using TaqMan Real-Time PCR Assays for designated genes and TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) according to manufacturer\u0026rsquo;s instructions. Alternatively, quantitative PCR was performed using SYBR Green PCR Master Mix (Thermo Fisher Scientific) with gene specific probes detailed in Tab. S2. Assays were performed on QuantStudio 6 Flex Real-Time PCR machines (Thermo Fisher Scientific). Two-way ANOVA was performed with Sidak\u0026rsquo;s multiple comparisons test using GraphPad Prism software.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"ItalicUnderline\" class=\"ItalicUnderline\" name=\"Emphasis\"\u003eIFNB1\u003c/span\u003e \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003epromoter luciferase assay\u003c/span\u003e\u003c/p\u003e \u003cp\u003ep125-HEK293 RIG-I-KO \u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e and p125-HEK293 RIG-I-KO/MDA5-KO stably expressing a Renilla luciferase downstream of an \u003cem\u003eIFNB1\u003c/em\u003e promoter were plated in 96-well plates (2.5 x 10\u003csup\u003e4\u003c/sup\u003e cells per well) for 24 hr, then treated with 30 U/ml IFN-A/D for a further 24 hr. Cells were transfected with 50 ng of IVT-RNA, 50 ng of cellular RNA, 25 ng of high MW Poly(I:C) (Invivogen, tlrl-pic), or 10 ng of Neo\u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5 CR6 CR7 CR8 CR9 CR10 CR11 CR12 CR13 CR14 CR15 CR16 CR17 CR18 CR19 CR20 CR21 CR22 CR23 CR24 CR25 CR26 CR27 CR28 CR29 CR30 CR31 CR32 CR33 CR34 CR35 CR36 CR37 CR38 CR39 CR40 CR41 CR42 CR43 CR44 CR45 CR46 CR47 CR48 CR49 CR50 CR51 CR52 CR53 CR54 CR55 CR56 CR57 CR58 CR59 CR60 CR61 CR62 CR63 CR64 CR65 CR66 CR67 CR68 CR69 CR70 CR71 CR72 CR73 CR74 CR75 CR76 CR77 CR78 CR79 CR80 CR81 CR82 CR83 CR84 CR85 CR86 CR87 CR88 CR89 CR90 CR91 CR92 CR93 CR94 CR95 CR96 CR97 CR98\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR99\" class=\"CitationRef\"\u003e99\u003c/span\u003e\u003c/sup\u003e IVT-RNA using 0.2 \u0026micro;l of Lipofectamine 2000 (Thermo Fisher Scientific, 11668027) according to manufacturer\u0026rsquo;s instructions. After 16\u0026ndash;24 hr, medium was removed and 75 \u0026micro;l of 1:1 solution of cell culture medium and ONE-Glo luciferase assay reagent (Promega, E6120) was added to cells. After 3 min, 50 \u0026micro;l of sample was transferred to a white 96-well plate and luminescence measured on a GloMax Luminometer (Promega).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eCytoplasmic RNA extraction and sequencing\u003c/h2\u003e \u003cp\u003eIRF3-KO THP1 cells were incubated with 10 ng/ml PMA overnight, then left untreated, infected with EMCV (MOI\u0026thinsp;=\u0026thinsp;2) for 20 hr, or treated with human IFNβ (Rebif) for 6 hr. Calu-3 cells were seeded in 6-well plates (1 x 10\u003csup\u003e6\u003c/sup\u003e cells per well) overnight, then left untreated, infected with SARS-CoV-2 (MOI\u0026thinsp;=\u0026thinsp;0.1) for 48 hr, or treated with human IFNβ (PBL assay science) for 6 hr. Cytoplasmic RNA was extracted using Cytoplasmic \u0026amp; Nuclear RNA Purification Kit (Norgen) according to manufacturer\u0026rsquo;s instructions. Total RNA libraries were prepared, starting with 500 ng of total RNA, according to the Illumina Tru-Seq Stranded Total RNA-Seq protocol (protocol 1000000040499) with RiboZero depletion, using unique dual indexes. 150 base paired-end sequencing was performed on a NextSeq 2000 P3 chip, loading pooled libraries at a concentration of 1,000 pM (protocol 1000000109376). Approximately 100\u0026nbsp;million reads were obtained per sample. Base calling was performed using Dragen BCLConvert (v3.7.4).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eIntron retention and upregulation analysis\u003c/h2\u003e \u003cp\u003eRead alignment was performed in performed in R (v4.1.2) \u003csup\u003e\u003cspan citationid=\"CR107\" class=\"CitationRef\"\u003e107\u003c/span\u003e\u003c/sup\u003e using the Rsubread package (v2.6.4) \u003csup\u003e\u003cspan citationid=\"CR108\" class=\"CitationRef\"\u003e108\u003c/span\u003e\u003c/sup\u003e. A genome index was built using the custom combined human and EMCV genome using the buildindex function and alignment was performed using sunjunc, with default settings. Exon- and intron-level differential gene expression analyses followed the methods of the \u003cem\u003eindex\u003c/em\u003e package (v1.0) \u003csup\u003e\u003cspan citationid=\"CR108\" class=\"CitationRef\"\u003e108\u003c/span\u003e\u003c/sup\u003e. A custom GTF file consisting of human and EMCV data was used for mapping. This GTF file was processed using the Annotations scripts from the Intron-reads repository (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/charitylaw/Intron-reads\u003c/span\u003e\u003cspan address=\"https://github.com/charitylaw/Intron-reads\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) in order to define exon and gene body regions. Counts were obtained using featureCounts from Rsubread (v2.8.1), with isPairedEnd and useMetaFeatures set to TRUE, allowMultiOverlap set to FALSE and strandSpecific 2. Intron counts were obtained by subtracting exon counts from gene body counts.\u003c/p\u003e \u003cp\u003eTwo DGEList objects were created using the edgeR package (v3.36.0) \u003csup\u003e\u003cspan citationid=\"CR109\" class=\"CitationRef\"\u003e109\u003c/span\u003e\u003c/sup\u003e using the exon and intron counts respectively, as well as sample annotation and gene annotation, with additional annotation obtained from the biomaRt package (v2.50.1) \u003csup\u003e\u003cspan citationid=\"CR110\" class=\"CitationRef\"\u003e110\u003c/span\u003e, \u003cspan citationid=\"CR111\" class=\"CitationRef\"\u003e111\u003c/span\u003e\u003c/sup\u003e. Sample 1C_IFNb, a technical replicate of sample 3C_IFNb, was excluded from further analysis. A design matrix was created incorporating the treatment group, so that the treatments could be compared, as well as the experimental replicate, to control for the experimental batch effect. Total library sizes were obtained by summing the exon and intron counts. Lowly expressed genes, on either an exonic or intronic level, were removed according to the filterByExpr function, incorporating the design matrix. As such, single-exon genes were also excluded from further analysis. Normalisation factors were calculated using the calcNormFactors function with the TMM method \u003csup\u003e\u003cspan citationid=\"CR112\" class=\"CitationRef\"\u003e112\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eCounts were processed using the voom method \u003csup\u003e\u003cspan citationid=\"CR113\" class=\"CitationRef\"\u003e113\u003c/span\u003e\u003c/sup\u003e and a linear model was fit using the edgeR voomLmFit function, including the design matrix. Comparisons between EMCV-treated, IFNβ-treated and untreated groups were made using the contrasts.fit function from the limma package (v3.50.0) \u003csup\u003e\u003cspan citationid=\"CR114\" class=\"CitationRef\"\u003e114\u003c/span\u003e\u003c/sup\u003e. Empirical Bayes moderated \u003cem\u003et\u003c/em\u003e-tests were performed against a 1.2-fold-change threshold and \u003cem\u003ep\u003c/em\u003e-values were obtained using the treat \u003csup\u003e\u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e115\u003c/span\u003e\u003c/sup\u003e function for differential expression analysis or without a fold-change threshold using the eBayes function \u003csup\u003e\u003cspan citationid=\"CR116\" class=\"CitationRef\"\u003e116\u003c/span\u003e\u003c/sup\u003e for subsequent comparison to iCLIP data.\u003c/p\u003e \u003cp\u003eResults of the intron-level analysis were visualised as mean-difference plots, showing the log\u003csub\u003e2\u003c/sub\u003e average expression and log\u003csub\u003e2\u003c/sub\u003e fold changes, highlighting genes significantly differentially expressed at the intron level, with Benjamini-Hochberg adjusted \u003cem\u003ep\u003c/em\u003e-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Differential expression results were visualised by plotting exon- versus intron-level log\u003csub\u003e2\u003c/sub\u003e fold changes on a per gene basis, quantifying the number of genes belonging to different differential expression categories, following the plot_index function from the \u003cem\u003eindex\u003c/em\u003e package.\u003c/p\u003e \u003cp\u003eIRFinder (v.1.30) \u003csup\u003e\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e was used to examine differential intron retention. Intron retention was quantified in BAM mode using the BAM files obtained from the Rsubread subjunc alignment, utilising a genome reference built using IRFinder BuildRefProcess on the combined human and EMCV genome and GTF files. Differential intron retention was performed in R using the DESeq2 package (v1.34.0) \u003csup\u003e\u003cspan citationid=\"CR117\" class=\"CitationRef\"\u003e117\u003c/span\u003e\u003c/sup\u003e, following the DESeq2Constructor.R script provided by IRFinder. The resulting DESeqDataSet object was filtered to only include introns that had been designated as clean by IRFinder (excluding known-exon regions). Furthermore, introns were only included if they either had no warnings or were only flagged as NonUniformIntronCover in at least 3 samples (thereby excluding introns that had been categorsed as LowCover, LowSplicing or MinorIsoform in most samples). The design matrix was defined as ~\u0026thinsp;Group\u0026thinsp;+\u0026thinsp;Group:IRFinder\u0026thinsp;+\u0026thinsp;Replicate and differential expression performed using the DESeq function. The Group:IRFinder interaction term was used for finding differences in intron retention, by comparing GroupEMCV.IRFinderIR and GroupIFNb.IRFinderIR to GroupUnif.IRFinderIR with the DESeq2 results function.\u003c/p\u003e \u003cp\u003eResults of the intron-retention analysis were visualised as mean-difference plots, showing log\u003csub\u003e2\u003c/sub\u003e (base mean expression\u0026thinsp;+\u0026thinsp;1) and log\u003csub\u003e2\u003c/sub\u003e fold changes, highlighting significantly differentially retained introns with Benjamini-Hochberg adjusted \u003cem\u003ep\u003c/em\u003e-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05. To compare the cytoplasmic RNA-seq to the iCLIP results, bedtools intersect was used in stranded mode to find genes from the \u003cem\u003eindex\u003c/em\u003e analysis and introns from the IRFinder analysis that overlapped with intronic EMCV iCLIP peaks. Results were visualised using the barcodeplot function from the limma package, either using the \u003cem\u003eindex\u003c/em\u003e EMCV intron-level empirical Bayes moderated \u003cem\u003et\u003c/em\u003e-statistics or IRFinder DESeq2 EMCV intron retention test statistics, showing all genes or introns that overlapped with iCLIP sites with scores higher than the 95% percentile. The iCLIP scores were used for the barcodeplot gene.weights argument; if more than one iCLIP site were overlapping a gene or intron, the higher score was used. In addition, \u003cem\u003ep\u003c/em\u003e-values were obtained using the roast \u003csup\u003e\u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e118\u003c/span\u003e\u003c/sup\u003e function from the edgeR package, using DGEList objects created from the counts from the index and IRFinder analyses. Dispersions were estimated using the estimateDisp function \u003csup\u003e\u003cspan citationid=\"CR119\" class=\"CitationRef\"\u003e119\u003c/span\u003e\u003c/sup\u003e with robust set to TRUE \u003csup\u003e\u003cspan citationid=\"CR116\" class=\"CitationRef\"\u003e116\u003c/span\u003e\u003c/sup\u003e, then the roast tests were performed using the iCLIP score weights with 19,999 rotations and the up-direction \u003cem\u003ep\u003c/em\u003e-values were reported.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eMTT assay\u003c/h2\u003e \u003cp\u003eLIM1215-GFP and LIM1215-SRSF3-GFP cells were plated at 9 x 10\u003csup\u003e4\u003c/sup\u003e cells/well in 96-well plates, incubated for 24 hr, and infected with EMCV at indicated MOIs for 48 hr. Cell culture medium was replaced with 100 \u0026micro;l of DMEM containing 1.5 mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide), and cells incubated for 3 hr. 150 \u0026micro;l of DMSO (dimethyl sulfoxide) was added to wells, and absorbance at 590 nm measured on OPTIstar plate reader (BMG LabTech).\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eEMCV RNA depletion and transfection into cells\u003c/h2\u003e \u003cp\u003eTHP1 cells were incubated with PMA overnight (10 ng/ml), infected with EMCV (MOI\u0026thinsp;=\u0026thinsp;5) for 22 hr or left uninfected, and total cellular RNA extracted (RNeasy, Qiagen). A mixture of five dual-biotinylated 35 nucleotide DNA probes (Integrated DNA Technologies) antisense to the EMCV genome (Tab. S2) were prepared at 8 \u0026micro;M and incubated with 250 \u0026micro;g of Hydrophilic Streptavidin Magnetic Beads (S1421, New England Biolabs) for 5 min, followed by washing with Wash/Binding Buffer (0.5 M NaCl, 20 mM Tris-HCl (pH 7.5), 1 mM EDTA). 5 \u0026micro;g of cellular RNA was diluted in 25 \u0026micro;l of Wash/Binding Buffer, heated to 75\u0026deg;C for for 5 min, and chilled on ice for 3 min. The RNA was incubated with probe-coated streptavidin beads for 15 min at 37\u0026deg;C. The EMCV-depleted supernatant was collected and RNA purified with RNeasy kit.\u003c/p\u003e \u003cp\u003eA549 wild-type or MDA5-KO \u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e were grown in 12- well plates (3 x 10\u003csup\u003e5\u003c/sup\u003e cells/well) in the presence of 30 U/ml IFNβ (Rebif) for 24 hr, pre-treated with 400 \u0026micro;M ribavirin (Sigma, R9644) for 1 hr, and transfected with 200 ng of EMCV-depleted or non-depleted RNA using 0.7 \u0026micro;l of Lipofectamine 2000 (Thermo Fisher Scientific, 11668027) according to manufacturer\u0026rsquo;s instructions. RNA was extracted after 20 hours using RNeasy Plus Mini Kit (Qiagen, 74136) according to manufacturer\u0026rsquo;s instructions, quantified by Nanodrop, and stored at \u0026minus;\u0026thinsp;80\u0026deg;C.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor contributions\u0026nbsp;\u003c/strong\u003e(using the CRediT taxonomy)\u003c/p\u003e\n\u003cp\u003eConceptualisation: N.S. and J.R.; Methodology: N.S. and A.G.D.J..; Software: N.S. and L.J.G.; Validation: N.S. and J.R.; Formal analysis: N.S., L.J.G. and J.R.; Investigation: N.S., A.G.D.J., L.C., V.O., C.C. and A.M.; Resources: M.R. and M.A.; Data curation: N.S.; Writing \u0026ndash; Original Draft: N.S. and J.R.; Writing \u0026ndash; Review \u0026amp; Editing: all authors; Visualisation: N.S., L.J.G and J.R.; Supervision: J.R. and P.J.H.; Project administration: N.S.; Funding acquisition: N.S., A.G.D.J., P.J.H. and J.R.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank David Sims, Charlotte George, George Kassiotis, George Young, Vladimir Pena, and members of the Rehwinkel lab for discussion; Chaojun Song and Boquan Jin for raising MDA5 antibodies; and Jurgen Moonen, Antony Mathews, Georgie Wray-McCann, the Monash Translational Health Precinct Medical Genomics Facility and the CEMIPAI BSL3 facility for providing reagents or for technical support. This work was funded by the UK Medical Research Council [MRC core funding of the MRC Human Immunology Unit; J.R.], the Wellcome Trust [grant number 100954; J.R.], the Lister Institute [J.R.], and the Australian National Health and Medical Research Council [P.J.H]. A.G.D.J. was supported by CNPq [grant number 211806/2013-7]. L.C. was supported by the ANRS-MIE [postdoctoral fellowship number ECTZ134113 and project grant number ECTZ134139]. \u0026nbsp;The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have declared that no conflict of interest exists.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCytoplasmic RNA sequencing data were uploaded to the NCBI Gene Expression Omnibus (GEO) under the SuperSeries accession GSE214664. Raw FASTQ files, as well as the exon and intron counts, were deposited for EMCV infection of IRF3 KO THP-1 cells (SubSeries GSE211240) and SARS-CoV-2 infection of Calu-3 cells (SubSeries GSE214663).o review GEO accession GSE214664, go to: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE214664 and enter token: gbwtoaialhszpet.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterials availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll unique reagents generated in this study are available from the corresponding authors with a completed Materials Transfer Agreement.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMedzhitov, R. Approaching the asymptote: 20 years later. \u003cem\u003eImmunity\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e,766-775 (2009).\u003c/li\u003e\n\u003cli\u003eBartok, E. \u0026amp; Hartmann, G. Immune Sensing Mechanisms that Discriminate Self from Altered Self and Foreign Nucleic Acids. \u003cem\u003eImmunity\u003c/em\u003e \u003cstrong\u003e53\u003c/strong\u003e,54-77 (2020).\u003c/li\u003e\n\u003cli\u003eMcNab, F., Mayer-Barber, K., Sher, A., Wack, A. \u0026amp; O\u0026apos;Garra, A. Type I interferons in infectious disease. \u003cem\u003eNature reviews. Immunology\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e,87-103 (2015).\u003c/li\u003e\n\u003cli\u003eCrow, Y.J. \u0026amp; Manel, N. Aicardi-Gouti\u0026egrave;res syndrome and the type I interferonopathies. \u003cem\u003eNature Reviews Immunology\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e,429 - 440 (2015).\u003c/li\u003e\n\u003cli\u003eTrinchieri, G. Type I interferon: friend or foe? \u003cem\u003eJ Exp Med\u003c/em\u003e \u003cstrong\u003e207\u003c/strong\u003e,2053-2063 (2010).\u003c/li\u003e\n\u003cli\u003eRehwinkel, J. \u0026amp; Gack, M.U. RIG-I-like receptors: their regulation and roles in RNA sensing. \u003cem\u003eNature reviews. Immunology\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e,537-551 (2020).\u003c/li\u003e\n\u003cli\u003eKato, H.\u003cem\u003e et al.\u003c/em\u003e Differential roles of MDA5 and RIG-I helicases in the recognition of RNA viruses. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e441\u003c/strong\u003e,101-105 (2006).\u003c/li\u003e\n\u003cli\u003eDias Junior, A.G., Sampaio, N.G. \u0026amp; Rehwinkel, J. A Balancing Act: MDA5 in Antiviral Immunity and Autoinflammation. \u003cem\u003eTrends in Microbiology\u003c/em\u003e (2018).\u003c/li\u003e\n\u003cli\u003eSampaio, N.G.\u003cem\u003e et al.\u003c/em\u003e The RNA sensor MDA5 detects SARS-CoV-2 infection. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e,13638 (2021).\u003c/li\u003e\n\u003cli\u003eThorne, L.G.\u003cem\u003e et al.\u003c/em\u003e SARS‐CoV‐2 sensing by RIG‐I and MDA5 links epithelial infection to macrophage inflammation. \u003cem\u003eThe EMBO Journal\u003c/em\u003e,e107826 (2021).\u003c/li\u003e\n\u003cli\u003eRebendenne, A.\u003cem\u003e et al.\u003c/em\u003e SARS-CoV-2 Triggers an MDA-5-Dependent Interferon Response Which Is Unable To Control Replication in Lung Epithelial Cells. \u003cem\u003eJournal of Virology\u003c/em\u003e \u003cstrong\u003e95\u003c/strong\u003e (2021).\u003c/li\u003e\n\u003cli\u003eYin, X.\u003cem\u003e et al.\u003c/em\u003e MDA5 Governs the Innate Immune Response to SARS-CoV-2 in Lung Epithelial Cells. \u003cem\u003eCell Reports\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e,108628 (2021).\u003c/li\u003e\n\u003cli\u003eMelchjorsen, J.\u003cem\u003e et al.\u003c/em\u003e Early Innate Recognition of Herpes Simplex Virus in Human Primary Macrophages Is Mediated via the MDA5/MAVS-Dependent and MDA5/MAVS/RNA Polymerase III-Independent Pathways. \u003cem\u003eJournal of Virology\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e,11350 - 11358 (2010).\u003c/li\u003e\n\u003cli\u003eLu, H.-L. \u0026amp; Liao, F. Melanoma differentiation-associated gene 5 senses hepatitis B virus and activates innate immune signaling to suppress virus replication. \u003cem\u003eJournal of immunology (Baltimore, Md. : 1950)\u003c/em\u003e \u003cstrong\u003e191\u003c/strong\u003e,3264 - 3276 (2013).\u003c/li\u003e\n\u003cli\u003eDou, Y., Yim, H.C., Kirkwood, C.D., Williams, B.R. \u0026amp; Sadler, A.J. The innate immune receptor MDA5 limits rotavirus infection but promotes cell death and pancreatic inflammation. \u003cem\u003eThe EMBO journal\u003c/em\u003e,e201696273 (2017).\u003c/li\u003e\n\u003cli\u003eWu, J.\u003cem\u003e et al.\u003c/em\u003e Strain-specific innate immune signaling pathways determine malaria parasitemia dynamics and host mortality. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e \u003cstrong\u003e111\u003c/strong\u003e,E511-E520 (2014).\u003c/li\u003e\n\u003cli\u003eLiehl, P.\u003cem\u003e et al.\u003c/em\u003e Host-cell sensors for Plasmodium activate innate immunity against liver-stage infection. \u003cem\u003eNature Medicine\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e,47-53 (2014).\u003c/li\u003e\n\u003cli\u003eBullen, C.K.\u003cem\u003e et al.\u003c/em\u003e MDA5 RNA-sensing pathway activation by Mycobacterium tuberculosis promotes innate immune subversion and pathogen survival. \u003cem\u003eJCI Insight\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2023).\u003c/li\u003e\n\u003cli\u003eWang, X.\u003cem\u003e et al.\u003c/em\u003e MDA5 Is an Essential Sensor of a Pathogen-Associated Molecular Pattern Associated with Vitality That Is Necessary for Host Resistance against Aspergillus fumigatus. \u003cem\u003eJ Immunol\u003c/em\u003e \u003cstrong\u003e205\u003c/strong\u003e,3058-3070 (2020).\u003c/li\u003e\n\u003cli\u003eLamborn, I.T.\u003cem\u003e et al.\u003c/em\u003e Recurrent rhinovirus infections in a child with inherited MDA5 deficiency. \u003cem\u003eThe Journal of experimental medicine\u003c/em\u003e \u003cstrong\u003e214\u003c/strong\u003e,1949-1972 (2017).\u003c/li\u003e\n\u003cli\u003eZaki, M.\u003cem\u003e et al.\u003c/em\u003e Recurrent and Prolonged Infections in a Child with a Homozygous IFIH1 Nonsense Mutation. \u003cem\u003eFront Genet\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e,130 (2017).\u003c/li\u003e\n\u003cli\u003eAsgari, S.\u003cem\u003e et al.\u003c/em\u003e Severe viral respiratory infections in children with IFIH1 loss-of-function mutations. \u003cem\u003eProceedings of the National Academy of Sciences of the United States of America\u003c/em\u003e \u003cstrong\u003e114\u003c/strong\u003e,8342-8347 (2017).\u003c/li\u003e\n\u003cli\u003eChen, J.\u003cem\u003e et al.\u003c/em\u003e Inborn errors of TLR3- or MDA5-dependent type I IFN immunity in children with enterovirus rhombencephalitis. \u003cem\u003eJournal of Experimental Medicine\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e,e20211349 (2021).\u003c/li\u003e\n\u003cli\u003ePichlmair, A.\u003cem\u003e et al.\u003c/em\u003e RIG-I-mediated antiviral responses to single-stranded RNA bearing 5\u0026apos;-phosphates. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e314\u003c/strong\u003e,997-1001 (2006).\u003c/li\u003e\n\u003cli\u003eHornung, V.\u003cem\u003e et al.\u003c/em\u003e 5\u0026apos;-Triphosphate RNA is the ligand for RIG-I. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e314\u003c/strong\u003e,994-997 (2006).\u003c/li\u003e\n\u003cli\u003eGoubau, D.\u003cem\u003e et al.\u003c/em\u003e Antiviral immunity via RIG-I-mediated recognition of RNA bearing 5\u0026apos;-diphosphates. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e514\u003c/strong\u003e,372 - 375 (2014).\u003c/li\u003e\n\u003cli\u003eSchlee, M. \u0026amp; Hartmann, G. Discriminating self from non-self in nucleic acid sensing. \u003cem\u003eNat Rev Immunol\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e,566-580 (2016).\u003c/li\u003e\n\u003cli\u003eSchuberth-Wagner, C.\u003cem\u003e et al.\u003c/em\u003e A Conserved Histidine in the RNA Sensor RIG-I Controls Immune Tolerance to N1-2\u0026apos;O-Methylated Self RNA. \u003cem\u003eImmunity\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e,41-51 (2015).\u003c/li\u003e\n\u003cli\u003eRehwinkel, J.\u003cem\u003e et al.\u003c/em\u003e RIG-I detects viral genomic RNA during negative-strand RNA virus infection. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e140\u003c/strong\u003e,397 - 408 (2010).\u003c/li\u003e\n\u003cli\u003eChiang, J.J.\u003cem\u003e et al.\u003c/em\u003e Viral unmasking of cellular 5S rRNA pseudogene transcripts induces RIG-I-mediated immunity. \u003cem\u003eNature immunology\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e,53-62 (2018).\u003c/li\u003e\n\u003cli\u003eVabret, N.\u003cem\u003e et al.\u003c/em\u003e Y RNAs are conserved endogenous RIG-I ligands across RNA virus infection and are targeted by HIV-1. \u003cem\u003eiScience\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e,104599 (2022).\u003c/li\u003e\n\u003cli\u003eZhao, Y., Ye, X., Dunker, W., Song, Y. \u0026amp; Karijolich, J. RIG-I like receptor sensing of host RNAs facilitates the cell-intrinsic immune response to KSHV infection. \u003cem\u003eNature communications\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e,4841 (2018).\u003c/li\u003e\n\u003cli\u003eNaesens, L.\u003cem\u003e et al.\u003c/em\u003e GTF3A mutations predispose to herpes simplex encephalitis by disrupting biogenesis of the host-derived RIG-I ligand RNA5SP141. \u003cem\u003eScience immunology\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e,eabq4531 (2022).\u003c/li\u003e\n\u003cli\u003eAndrejeva, J.\u003cem\u003e et al.\u003c/em\u003e The V proteins of paramyxoviruses bind the IFN-inducible RNA helicase, mda-5, and inhibit its activation of the IFN-beta promoter. \u003cem\u003eProceedings of the National Academy of Sciences of the United States of America\u003c/em\u003e \u003cstrong\u003e101\u003c/strong\u003e,17264-17269 (2004).\u003c/li\u003e\n\u003cli\u003eKato, H.\u003cem\u003e et al.\u003c/em\u003e Length-dependent recognition of double-stranded ribonucleic acids by retinoic acid-inducible gene-I and melanoma differentiation-associated gene 5. \u003cem\u003eThe Journal of experimental medicine\u003c/em\u003e \u003cstrong\u003e205\u003c/strong\u003e,1601-1610 (2008).\u003c/li\u003e\n\u003cli\u003eWu, B.\u003cem\u003e et al.\u003c/em\u003e Structural basis for dsRNA recognition, filament formation, and antiviral signal activation by MDA5. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e152\u003c/strong\u003e,276 - 289 (2013).\u003c/li\u003e\n\u003cli\u003eTriantafilou, K.\u003cem\u003e et al.\u003c/em\u003e Visualisation of direct interaction of MDA5 and the dsRNA replicative intermediate form of positive strand RNA viruses. \u003cem\u003eJ Cell Sci\u003c/em\u003e \u003cstrong\u003e125\u003c/strong\u003e,4761-4769 (2012).\u003c/li\u003e\n\u003cli\u003eFeng, Q.\u003cem\u003e et al.\u003c/em\u003e MDA5 detects the double-stranded RNA replicative form in picornavirus-infected cells. \u003cem\u003eCell Reports\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e,1187 - 1196 (2012).\u003c/li\u003e\n\u003cli\u003ePichlmair, A.\u003cem\u003e et al.\u003c/em\u003e Activation of MDA5 requires higher-order RNA structures generated during virus infection. \u003cem\u003eJournal of Virology\u003c/em\u003e \u003cstrong\u003e83\u003c/strong\u003e,10761 - 10769 (2009).\u003c/li\u003e\n\u003cli\u003eSanchez David, R.Y.\u003cem\u003e et al.\u003c/em\u003e Comparative analysis of viral RNA signatures on different RIG-I-like receptors. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e,e11275 (2016).\u003c/li\u003e\n\u003cli\u003eDeddouche, S.\u003cem\u003e et al.\u003c/em\u003e Identification of an LGP2-associated MDA5 agonist in picornavirus-infected cells. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e,e01535 (2014).\u003c/li\u003e\n\u003cli\u003eRunge, S.\u003cem\u003e et al.\u003c/em\u003e In vivo ligands of MDA5 and RIG-I in measles virus-infected cells. \u003cem\u003ePLoS pathogens\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e,e1004081 (2014).\u003c/li\u003e\n\u003cli\u003eHuppertz, I.\u003cem\u003e et al.\u003c/em\u003e iCLIP: protein-RNA interactions at nucleotide resolution. \u003cem\u003eMethods (San Diego, Calif.)\u003c/em\u003e \u003cstrong\u003e65\u003c/strong\u003e,274 - 287 (2014).\u003c/li\u003e\n\u003cli\u003eK\u0026ouml;nig, J.\u003cem\u003e et al.\u003c/em\u003e iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution. \u003cem\u003eNature Structural \u0026amp; Molecular Biology\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e,909 - 915 (2010).\u003c/li\u003e\n\u003cli\u003eDavid, R.Y.S.\u003cem\u003e et al.\u003c/em\u003e Comparative analysis of viral RNA signatures on different RIG-I-like receptors. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e,e11275 (2016).\u003c/li\u003e\n\u003cli\u003eAhmad, S.\u003cem\u003e et al.\u003c/em\u003e Breaching Self-Tolerance to Alu Duplex RNA Underlies MDA5-Mediated Inflammation. \u003cem\u003eCell\u003c/em\u003e (2018).\u003c/li\u003e\n\u003cli\u003eGitlin, L.\u003cem\u003e et al.\u003c/em\u003e Essential role of mda-5 in type I IFN responses to polyriboinosinic:polyribocytidylic acid and encephalomyocarditis picornavirus. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e \u003cstrong\u003e103\u003c/strong\u003e,8459-8464 (2006).\u003c/li\u003e\n\u003cli\u003eSchonborn, J.\u003cem\u003e et al.\u003c/em\u003e Monoclonal antibodies to double-stranded RNA as probes of RNA structure in crude nucleic acid extracts. \u003cem\u003eNucleic acids research\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e,2993-3000 (1991).\u003c/li\u003e\n\u003cli\u003eWeber, F., Wagner, V., Rasmussen, S.B., Hartmann, R. \u0026amp; Paludan, S.R. Double-stranded RNA is produced by positive-strand RNA viruses and DNA viruses but not in detectable amounts by negative-strand RNA viruses. \u003cem\u003eJ Virol\u003c/em\u003e \u003cstrong\u003e80\u003c/strong\u003e,5059-5064 (2006).\u003c/li\u003e\n\u003cli\u003eHertzog, J.\u003cem\u003e et al.\u003c/em\u003e Infection with a Brazilian isolate of Zika virus generates RIG-I stimulatory RNA and the viral NS5 protein blocks type I IFN induction and signaling. \u003cem\u003eEur J Immunol\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e,1120-1136 (2018).\u003c/li\u003e\n\u003cli\u003eRybak-Wolf, A.\u003cem\u003e et al.\u003c/em\u003e A variety of dicer substrates in human and C. elegans. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e159\u003c/strong\u003e,1153-1167 (2014).\u003c/li\u003e\n\u003cli\u003eMoy, R.H.\u003cem\u003e et al.\u003c/em\u003e Stem-loop recognition by DDX17 facilitates miRNA processing and antiviral defense. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e158\u003c/strong\u003e,764-777 (2014).\u003c/li\u003e\n\u003cli\u003eKazmierski, J.\u003cem\u003e et al.\u003c/em\u003e A Baseline Cellular Antiviral State Is Maintained by cGAS and Its Most Frequent Naturally Occurring Variant rs610913. \u003cem\u003eJ Immunol\u003c/em\u003e (2022).\u003c/li\u003e\n\u003cli\u003eNostrand, E.L.V.\u003cem\u003e et al.\u003c/em\u003e Robust transcriptome-wide discovery of RNA-binding protein binding sites with enhanced CLIP (eCLIP). \u003cem\u003eNature methods\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e,508 - 514 (2016).\u003c/li\u003e\n\u003cli\u003eKrakau, S., Richard, H. \u0026amp; Marsico, A. PureCLIP: capturing target-specific protein-RNA interaction footprints from single-nucleotide CLIP-seq data. \u003cem\u003eGenome biology\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e,240 (2017).\u003c/li\u003e\n\u003cli\u003eHeinz, S.\u003cem\u003e et al.\u003c/em\u003e Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. \u003cem\u003eMol Cell\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e,576-589 (2010).\u003c/li\u003e\n\u003cli\u003eMachanick, P. \u0026amp; Bailey, T.L. MEME-ChIP: motif analysis of large DNA datasets. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e,1696-1697 (2011).\u003c/li\u003e\n\u003cli\u003eBerke, I.C., Yu, X., Modis, Y. \u0026amp; Egelman, E.H. MDA5 assembles into a polar helical filament on dsRNA. \u003cem\u003eProceedings of the National Academy of Sciences of the United States of America\u003c/em\u003e \u003cstrong\u003e109\u003c/strong\u003e,18437 - 18441 (2012).\u003c/li\u003e\n\u003cli\u003eSchilling, M.\u003cem\u003e et al.\u003c/em\u003e RIG-I Plays a Dominant Role in the Induction of Transcriptional Changes in Zika Virus-Infected Cells, which Protect from Virus-Induced Cell Death. \u003cem\u003eCells\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e (2020).\u003c/li\u003e\n\u003cli\u003eMiddleton, R.\u003cem\u003e et al.\u003c/em\u003e IRFinder: assessing the impact of intron retention on mammalian gene expression. \u003cem\u003eGenome Biology\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e,51 (2017).\u003c/li\u003e\n\u003cli\u003eLee, S.\u003cem\u003e et al.\u003c/em\u003e Covering all your bases: incorporating intron signal from RNA-seq data. \u003cem\u003eNar Genom Bioinform\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e (2020).\u003c/li\u003e\n\u003cli\u003eFitzgerald, K.D. \u0026amp; Semler, B.L. Re-localization of Cellular Protein SRp20 during Poliovirus Infection: Bridging a Viral IRES to the Host Cell Translation Apparatus. \u003cem\u003ePLOS Pathogens\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e,e1002127 (2011).\u003c/li\u003e\n\u003cli\u003e\u0026Aacute;lvarez, E., Castell\u0026oacute;, A., Carrasco, L. \u0026amp; Izquierdo, J.M. Poliovirus 2A Protease Triggers a Selective Nucleo-Cytoplasmic Redistribution of Splicing Factors to Regulate Alternative Pre-mRNA Splicing. \u003cem\u003ePLOS ONE\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e,e73723 (2013).\u003c/li\u003e\n\u003cli\u003eErtel, K.J., Brunner, J.E. \u0026amp; Semler, B.L. Mechanistic consequences of hnRNP C binding to both RNA termini of poliovirus negative-strand RNA intermediates. \u003cem\u003eJournal of virology\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e,4229-4242 (2010).\u003c/li\u003e\n\u003cli\u003eGustin, K.E. \u0026amp; Sarnow, P. Effects of poliovirus infection on nucleo-cytoplasmic trafficking and nuclear pore complex composition. \u003cem\u003eThe EMBO Journal\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e,240-249 (2001).\u003c/li\u003e\n\u003cli\u003eFlather, D. \u0026amp; Semler, B.L. Picornaviruses and nuclear functions: targeting a cellular compartment distinct from the replication site of a positive-strand RNA virus. \u003cem\u003eFrontiers in Microbiology\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e (2015).\u003c/li\u003e\n\u003cli\u003eBanerjee, A.K.\u003cem\u003e et al.\u003c/em\u003e SARS-CoV-2 Disrupts Splicing, Translation, and Protein Trafficking to Suppress Host Defenses. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e183\u003c/strong\u003e,1325-1339.e1321 (2020).\u003c/li\u003e\n\u003cli\u003eM\u0026uuml;ller-McNicoll, M.\u003cem\u003e et al.\u003c/em\u003e SR proteins are NXF1 adaptors that link alternative RNA processing to mRNA export. \u003cem\u003eGenes \u0026amp; Development\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e,553-566 (2016).\u003c/li\u003e\n\u003cli\u003eRatnadiwakara, M.\u003cem\u003e et al.\u003c/em\u003e SRSF3 confers selective processing of miR-17-92 cluster to promote tumorigenic properties in colorectal cancer. \u003cem\u003eBiorxiv\u003c/em\u003e,667295 (2019).\u003c/li\u003e\n\u003cli\u003eHeilig, R. \u0026amp; Broz, P. Function and mechanism of the pyrin inflammasome. \u003cem\u003eEur J Immunol\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e,230-238 (2018).\u003c/li\u003e\n\u003cli\u003eDe Maio, F.A.\u003cem\u003e et al.\u003c/em\u003e The Dengue Virus NS5 Protein Intrudes in the Cellular Spliceosome and Modulates Splicing. \u003cem\u003ePLoS Pathog\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e,e1005841 (2016).\u003c/li\u003e\n\u003cli\u003eBrunetti, J.E., Scolaro, L.A. \u0026amp; Castilla, V. The heterogeneous nuclear ribonucleoprotein K (hnRNP K) is a host factor required for dengue virus and Jun\u0026iacute;n virus multiplication. \u003cem\u003eVirus Res\u003c/em\u003e \u003cstrong\u003e203\u003c/strong\u003e,84-91 (2015).\u003c/li\u003e\n\u003cli\u003eBelov, G.A.\u003cem\u003e et al.\u003c/em\u003e Early alteration of nucleocytoplasmic traffic induced by some RNA viruses. \u003cem\u003eVirology\u003c/em\u003e \u003cstrong\u003e275\u003c/strong\u003e,244-248 (2000).\u003c/li\u003e\n\u003cli\u003eBelov, G.A.\u003cem\u003e et al.\u003c/em\u003e Bidirectional Increase in Permeability of Nuclear Envelope upon Poliovirus Infection and Accompanying Alterations of Nuclear Pores. \u003cem\u003eJournal of Virology\u003c/em\u003e \u003cstrong\u003e78\u003c/strong\u003e,10166-10177 (2004).\u003c/li\u003e\n\u003cli\u003eAshraf, U., Benoit-Pilven, C., Lacroix, V., Navratil, V. \u0026amp; Naffakh, N. Advances in Analyzing Virus-Induced Alterations of Host Cell Splicing. \u003cem\u003eTrends in microbiology\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e,268-281 (2019).\u003c/li\u003e\n\u003cli\u003eGuney, M.H.\u003cem\u003e et al.\u003c/em\u003e IFIH1 (MDA5) is required for innate immune detection of intron-containing RNA expressed from the HIV-1 provirus. \u003cem\u003eProceedings of the National Academy of Sciences of the United States of America\u003c/em\u003e \u003cstrong\u003e121\u003c/strong\u003e,e2404349121 (2024).\u003c/li\u003e\n\u003cli\u003eCortese, M.\u003cem\u003e et al.\u003c/em\u003e Integrative Imaging Reveals SARS-CoV-2-Induced Reshaping of Subcellular Morphologies. \u003cem\u003eCell Host Microbe\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e,853-866.e855 (2020).\u003c/li\u003e\n\u003cli\u003eSeth, R.B., Sun, L., Ea, C.K. \u0026amp; Chen, Z.J. Identification and characterization of MAVS, a mitochondrial antiviral signaling protein that activates NF-kappaB and IRF 3. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e122\u003c/strong\u003e,669-682 (2005).\u003c/li\u003e\n\u003cli\u003eWu, B.\u003cem\u003e et al.\u003c/em\u003e Structural basis for dsRNA recognition, filament formation, and antiviral signal activation by MDA5. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e152\u003c/strong\u003e,276-289 (2013).\u003c/li\u003e\n\u003cli\u003eDominguez, D.\u003cem\u003e et al.\u003c/em\u003e Sequence, Structure, and Context Preferences of Human RNA Binding Proteins. \u003cem\u003eMolecular cell\u003c/em\u003e \u003cstrong\u003e70\u003c/strong\u003e,854-867 e859 (2018).\u003c/li\u003e\n\u003cli\u003ePhillips, S.L., Soderblom, E.J., Bradrick, S.S. \u0026amp; Garcia-Blanco, M.A. Identification of Proteins Bound to Dengue Viral RNA In Vivo Reveals New Host Proteins Important for Virus Replication. \u003cem\u003emBio\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e,e01865-01815 (2016).\u003c/li\u003e\n\u003cli\u003eErtel, K.J., Brunner, J.E. \u0026amp; Semler, B.L. Mechanistic consequences of hnRNP C binding to both RNA termini of poliovirus negative-strand RNA intermediates. \u003cem\u003eJ Virol\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e,4229-4242 (2010).\u003c/li\u003e\n\u003cli\u003eRehwinkel, J. \u0026amp; Mehdipour, P. ADAR1: from basic mechanisms to inhibitors. \u003cem\u003eTrends Cell Biol\u003c/em\u003e (2024).\u003c/li\u003e\n\u003cli\u003eRice, G.I.\u003cem\u003e et al.\u003c/em\u003e Mutations in ADAR1 cause Aicardi-Gouti\u0026egrave;res syndrome associated with a type I interferon signature. \u003cem\u003eNat Genet\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e,1243-1248 (2012).\u003c/li\u003e\n\u003cli\u003eRoulois, D.\u003cem\u003e et al.\u003c/em\u003e DNA-Demethylating Agents Target Colorectal Cancer Cells by Inducing Viral Mimicry by Endogenous Transcripts. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e162\u003c/strong\u003e,961-973 (2015).\u003c/li\u003e\n\u003cli\u003eChiappinelli, K.B.\u003cem\u003e et al.\u003c/em\u003e Inhibiting DNA Methylation Causes an Interferon Response in Cancer via dsRNA Including Endogenous Retroviruses. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e162\u003c/strong\u003e,974-986 (2015).\u003c/li\u003e\n\u003cli\u003eBowling, E.A.\u003cem\u003e et al.\u003c/em\u003e Spliceosome-targeted therapies trigger an antiviral immune response in triple-negative breast cancer. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e,384-403 e321 (2021).\u003c/li\u003e\n\u003cli\u003eHerzner, A.-M.\u003cem\u003e et al.\u003c/em\u003e ADAR and hnRNPC deficiency synergize in activating endogenous dsRNA-induced type I IFN responses. \u003cem\u003eJournal of Experimental Medicine\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e,e20201833 (2021).\u003c/li\u003e\n\u003cli\u003eZheng, R.\u003cem\u003e et al.\u003c/em\u003e hnRNPM protects against the dsRNA-mediated interferon response by repressing LINE-associated cryptic splicing. \u003cem\u003eMolecular cell\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e,2087-2103 e2088 (2024).\u003c/li\u003e\n\u003cli\u003eDorrity, T.J.\u003cem\u003e et al.\u003c/em\u003e Long 3\u0026apos;UTRs predispose neurons to inflammation by promoting immunostimulatory double-stranded RNA formation. \u003cem\u003eScience immunology\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e,eadg2979 (2023).\u003c/li\u003e\n\u003cli\u003ePeisley, A. \u0026amp; Hur, S. Multi-level regulation of cellular recognition of viral dsRNA. \u003cem\u003eCell Mol Life Sci\u003c/em\u003e \u003cstrong\u003e70\u003c/strong\u003e,1949-1963 (2013).\u003c/li\u003e\n\u003cli\u003eMaurano, M.\u003cem\u003e et al.\u003c/em\u003e Protein kinase R and the integrated stress response drive immunopathology caused by mutations in the RNA deaminase ADAR1. \u003cem\u003eImmunity\u003c/em\u003e (2021).\u003c/li\u003e\n\u003cli\u003eBanerjee, S.\u003cem\u003e et al.\u003c/em\u003e OAS-RNase L innate immune pathway mediates the cytotoxicity of a DNA-demethylating drug. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e \u003cstrong\u003e116\u003c/strong\u003e,201815071 (2019).\u003c/li\u003e\n\u003cli\u003eLiu, C.-X.\u003cem\u003e et al.\u003c/em\u003e Structure and Degradation of Circular RNAs Regulate PKR Activation in Innate Immunity. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e177\u003c/strong\u003e,865-880.e821 (2019).\u003c/li\u003e\n\u003cli\u003eLi, Y.\u003cem\u003e et al.\u003c/em\u003e Ribonuclease L mediates the cell-lethal phenotype of double-stranded RNA editing enzyme ADAR1 deficiency in a human cell line. \u003cem\u003eElife\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e,e25687 (2017).\u003c/li\u003e\n\u003cli\u003eWickenhagen, A.\u003cem\u003e et al.\u003c/em\u003e A prenylated dsRNA sensor protects against severe COVID-19. \u003cem\u003eScience\u003c/em\u003e,eabj3624 (2021).\u003c/li\u003e\n\u003cli\u003eBuchbender, A.\u003cem\u003e et al.\u003c/em\u003e Improved library preparation with the new iCLIP2 protocol. \u003cem\u003eMethods\u003c/em\u003e \u003cstrong\u003e178\u003c/strong\u003e,33-48 (2020).\u003c/li\u003e\n\u003cli\u003eLee, F.C.Y.\u003cem\u003e et al.\u003c/em\u003e An improved iCLIP protocol. \u003cem\u003eBiorxiv\u003c/em\u003e,2021.2008.2027.457890 (2021).\u003c/li\u003e\n\u003cli\u003eBusch, A., Bruggemann, M., Ebersberger, S. \u0026amp; Zarnack, K. iCLIP data analysis: A complete pipeline from sequencing reads to RBP binding sites. \u003cem\u003eMethods (San Diego, Calif.)\u003c/em\u003e \u003cstrong\u003e178\u003c/strong\u003e,49-62 (2020).\u003c/li\u003e\n\u003cli\u003eDobin, A.\u003cem\u003e et al.\u003c/em\u003e STAR: ultrafast universal RNA-seq aligner. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e,15-21 (2013).\u003c/li\u003e\n\u003cli\u003eSmith, T., Heger, A. \u0026amp; Sudbery, I. UMI-tools: modeling sequencing errors in Unique Molecular Identifiers to improve quantification accuracy. \u003cem\u003eGenome Res\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e,491-499 (2017).\u003c/li\u003e\n\u003cli\u003eEwels, P., Magnusson, M., Lundin, S. \u0026amp; K\u0026auml;ller, M. MultiQC: summarize analysis results for multiple tools and samples in a single report. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e,3047-3048 (2016).\u003c/li\u003e\n\u003cli\u003eGoodstadt, L. Ruffus: a lightweight Python library for computational pipelines. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e,2778-2779 (2010).\u003c/li\u003e\n\u003cli\u003eQuinlan, A.R. \u0026amp; Hall, I.M. BEDTools: a flexible suite of utilities for comparing genomic features. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e,841-842 (2010).\u003c/li\u003e\n\u003cli\u003eSmit, A., Hubley, R, Green, P. RepeatMasker Open-4.0. (2013-2015).\u003c/li\u003e\n\u003cli\u003eRobinson, J.T.\u003cem\u003e et al.\u003c/em\u003e Integrative genomics viewer. \u003cem\u003eNat Biotechnol\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e,24-26 (2011).\u003c/li\u003e\n\u003cli\u003eTeam, R.C. R: A language and environment for statistical computing. R Foundation for Statistical Computing. (2021).\u003c/li\u003e\n\u003cli\u003eLiao, Y., Smyth, G.K. \u0026amp; Shi, W. The R package Rsubread is easier, faster, cheaper and better for alignment and quantification of RNA sequencing reads. \u003cem\u003eNucleic Acids Res\u003c/em\u003e \u003cstrong\u003e47\u003c/strong\u003e,e47 (2019).\u003c/li\u003e\n\u003cli\u003eRobinson, M.D., McCarthy, D.J. \u0026amp; Smyth, G.K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e,139-140 (2010).\u003c/li\u003e\n\u003cli\u003eDurinck, S., Spellman, P.T., Birney, E. \u0026amp; Huber, W. Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. \u003cem\u003eNat Protoc\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e,1184-1191 (2009).\u003c/li\u003e\n\u003cli\u003eDurinck, S.\u003cem\u003e et al.\u003c/em\u003e BioMart and Bioconductor: a powerful link between biological databases and microarray data analysis. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e,3439-3440 (2005).\u003c/li\u003e\n\u003cli\u003eRobinson, M.D. \u0026amp; Oshlack, A. A scaling normalization method for differential expression analysis of RNA-seq data. \u003cem\u003eGenome Biol\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e,R25 (2010).\u003c/li\u003e\n\u003cli\u003eLaw, C.W., Chen, Y., Shi, W. \u0026amp; Smyth, G.K. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. \u003cem\u003eGenome Biol\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e,R29 (2014).\u003c/li\u003e\n\u003cli\u003eRitchie, M.E.\u003cem\u003e et al.\u003c/em\u003e limma powers differential expression analyses for RNA-sequencing and microarray studies. \u003cem\u003eNucleic acids research\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e,e47 (2015).\u003c/li\u003e\n\u003cli\u003eMcCarthy, D.J. \u0026amp; Smyth, G.K. Testing significance relative to a fold-change threshold is a TREAT. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e,765-771 (2009).\u003c/li\u003e\n\u003cli\u003ePhipson, B., Lee, S., Majewski, I.J., Alexander, W.S. \u0026amp; Smyth, G.K. Robust hyperparameter estimation protects against hypervariable genes and improves power to detect differential expression. \u003cem\u003eThe Annals of Applied Statistics\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e,946-963, 918 (2016).\u003c/li\u003e\n\u003cli\u003eLove, M.I., Huber, W. \u0026amp; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. \u003cem\u003eGenome Biology\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e,550 (2014).\u003c/li\u003e\n\u003cli\u003eWu, D.\u003cem\u003e et al.\u003c/em\u003e ROAST: rotation gene set tests for complex microarray experiments. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e,2176-2182 (2010).\u003c/li\u003e\n\u003cli\u003eChen, Y., Lun, ATL, and Smyth, GK. Differential expression analysis of complex RNA-seq experiments using edgeR. \u003cem\u003eStatistical Analysis of Next Generation Sequence Data\u003c/em\u003e. Somnath Datta and Daniel S Nettleton (eds), Springer, New York, 2014.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"University of Oxford","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6466919/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6466919/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eMDA5 is an innate immune RNA sensor that senses infection with a range of viruses and other pathogens. MDA5\u0026rsquo;s RNA agonists are not well defined. We used single-nucleotide resolution crosslinking and immunoprecipitation (iCLIP) to study its ligands. Surprisingly, upon infection with SARS-CoV-2 or encephalomyocarditis virus, MDA5 bound overwhelmingly to cellular RNAs. Many binding sites were intronic and proximal to \u003cem\u003eAlu\u003c/em\u003e elements and to potentially base-paired structures. Concomitantly, cytoplasmic levels of intron-containing unspliced transcripts increased in infected cells and displayed enrichment of MDA5 iCLIP peaks. Moreover, overexpression of a splicing factor abrogated MDA5 activation. Finally, when depleted of viral sequences, RNA extracted from infected cells still stimulated MDA5. Taken together, MDA5 surveys RNA processing fidelity and detects infections by sensing perturbations of posttranscriptional events such as splicing, establishing a paradigm of innate immune \u0026lsquo;guarding\u0026rsquo; for RNA sensors.\u003c/p\u003e","manuscriptTitle":"MDA5 guards against infection by surveying cellular RNA homeostasis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-24 05:10:01","doi":"10.21203/rs.3.rs-6466919/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e2df9829-6542-4db6-9fd1-0f304b65c594","owner":[],"postedDate":"April 24th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":47300951,"name":"Immunology"}],"tags":[],"updatedAt":"2025-04-24T05:10:01+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-24 05:10:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6466919","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6466919","identity":"rs-6466919","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}