key: cord-0271058-5jswci3m
authors: Aouadi, Wahiba; Najburg, Valérie; Legendre, Rachel; Varet, Hugo; Kergoat, Lauriane; Tangy, Frédéric; Larrous, Florence; Komarova, Anastassia V.; Bourhy, Hervé
title: Recognition of copy-back defective interfering rabies virus genomes by RIG-I triggers the antiviral response against vaccine strains
date: 2022-03-24
journal: bioRxiv
DOI: 10.1101/2022.03.24.485606
sha: 3989ed45c0a364e0f5d2694309b390e701f7ddbb
doc_id: 271058
cord_uid: 5jswci3m

Rabies virus (RABV) is a lethal neurotropic virus that causes 60,000 human deaths every year around the world. A typical feature of RABV infection is the suppression of type I and III interferon (IFN)-mediated antiviral response. However, molecular mechanisms leading to RABV sensing by RIG-I-like receptors (RLR) to initiate IFN signaling remain elusive. Here, we showed that RABV RNAs are recognized by RIG-I (retinoic acid-inducible gene I) sensor resulting in an IFN response of the infected cells but that this global feature was differently modulated according to the type of RABV used. RNAs from pathogenic RABV strain, THA, were poorly detected in the cytosol by RIG-I and therefore mediated a weak antiviral response. On the opposite, we revealed a strong interferon activity triggered by the RNAs of the attenuated RABV vaccine SAD strain mediated by RIG-I. Using next-generation sequencing (NGS) combined with bioinformatics tools, we characterized two major 5’copy-back defective interfering (5’cb DI) genomes generated during SAD replication. Furthermore, we identified a specific interaction of 5’cb DI genomes and RIG-I that correlated with a high stimulation of the type I IFN signaling. This study indicates that RNAs from a wild-type RABV poorly activate the RIG-I pathway, while the presence of 5’cb DIs in vaccine SAD strain serves as an intrinsic adjuvant that strengthens its efficiency by enhancing RIG-I detection and therefore strongly stimulates the IFN response. Highlights RABV pathogenic strain replication in vitro is characterized by the absence of defective interfering genomes thus induces a weak RLR-mediated innate immunity antiviral response. RABV vaccine attenuated strain shows a high release of 5’ copy-back defective interfering genomes during replication in vitro and therefore enhances a strong antiviral response upon infection. RIG-I is the main sensor for RABV RNA detection within cells.

• RABV pathogenic strain replication in vitro is characterized by the absence of defective 17 interfering genomes thus induces a weak RLR-mediated innate immunity antiviral 18 response. 19

• RABV vaccine attenuated strain shows a high release of 5' copy-back defective 20 interfering genomes during replication in vitro and therefore enhances a strong antiviral 21 response upon infection. 22

• RIG-I is the main sensor for RABV RNA detection within cells. To explore in more detail the differential involvement of RLR (RIG-I or MDA5) in RABV 160 RNA sensing, the ISRE reporter cell line was treated with non-targeting siRNAs (si-Neg 161 control), targeting RIG-I (si-RIG-I) or MDA5 (si-MDA5). Transient silencing of RIG-I and 162

MDA5 significantly (p ≤0.001, one-way ANOVA using Tukey method) reduced the level of 163 mRNA for RIG-I and MDA5 by ~61% and ~82%, respectively as compared to si-Negative 164 control cells ( Figure 1C ). When the same cells were transfected with RABV RNAs (THA or 165 SAD), only RIG-I silenced reporter cells showed strongly impaired ISRE promoter activation 166 (p <0.0001, one-way ANOVA using Tukey method), while MDA5 silencing did not affect the 167 signaling ( Figure 1D ). 168

Altogether, these results demonstrate that RIG-I is the main cytosolic PRR that detects RABV 169

RNAs to mediate IFN signaling. Then, RNAs co-purified with RIG-I, MDA5, or CH were extracted and transfected into the  192   ISRE reporter cell line to assess the activation of type-I IFN signaling. First, the ISRE promoter  193 activation was first controlled by transfecting synthetic RNAs. As expected, 5'3P, HMW 194 poly(I:C), and LMW poly(I:C) largely stimulated ISRE expression ( Figure 2B ). These data 195 validated our transfection approach. RIG-I-specific RNAs extracted from THA-infected cells 196 induced a light but significant (p=0.029, single-step method) stimulation of ISRE expression as 197 compared to the negative control CH-specific RNAs ( Figure 2C ). RIG-I-specific SAD RNA 198 induced a strong and significant (p=0.00007, single-step method) ISRE promoter activity in a 199 dose-dependent manner as compared to negative control CH-specific RNAs ( Figure 2D ). For 200 both viral strains, MDA5-specific RNAs showed no significant stimulation of the ISRE 201 promoter activity. These data suggest that RNA extracted from THA and SAD infected cells 202 present molecular patterns that are absent in non-infected cells and that are preferentially 203 recognized by RIG-I. 204

205

To further characterize RLR bound RABV RNAs, ST-RLRs cells were infected with THA or 207 SAD at MOI of 0.5. The cells were harvested after 24h, and RLRs bound RNAs were purified 208 using affinity chromatography. Total RNAs from input cell extract, as well as RLRs-bound 209 RNA output, were extracted. Input and output RNAs were subjected to NGS followed by 210 bioinformatics analysis using the previously described protocol (Chazal et al., 2018). NGS 211 provided high output in the order of 12 to 90 million total reads per sample with around 0.3 and 212 0.29% mapped to THA and SAD genomes, respectively (Table S1 ). The fold change of RLR 213 binding was obtained by normalizing: i) mean abundance of reads (coverage) of either RIG-I 214 or MDA5-ligands by mean read coverage of corresponding input samples, ii) nonspecific RNAs 215 binding were discarded by normalizing RLRs read coverage to reads obtained with cherry 216 negative control. NGS data analysis revealed no visible enrichment of read-abundance in RLR 217 association of negative-and positive-sense viral RNAs upon THA infection in our experimental 218 conditions ( Figure 3A) . Interestingly, the NGS study of the RABV vaccine strain, SAD, showed 219 a significant coverage and much higher enrichment on MDA5 of negative-sense RNA along 220 the whole viral genome than with positive-sense viral RNA. Thus, MDA5 may be engaged in 221 recognition of the SAD negative-sense genome. Conversely, studies analyzing RNA bound to 222 MDA5 in measles virus-infected cells (also negative-sense single-stranded RNA virus) 223

reporting that MDA5 interacts with only positive-strand viral RNA (Runge et al., 2014; 224 Sanchez David et al., 2016) . However, despite viral RNA/MDA5 interaction, we failed to 225 observe any statistically significant stimulation of ISRE promoter in reporter cells in our 226 experimental conditions ( Figure 2D ). 227

Conversely, the analysis of RIG-I-specific RNA ligands purified from SAD-infected cells 228 revealed a significant enrichment of viral negative-and positive-sense RNAs (Figure 3B To further explore the RNA primary sequence motifs of RLR ligands that could explain SAD 237 RNA recognition by RIG-I and or MDA5, we analyzed the AU content of RIG-I and MDA5 238 specific reads as described previously (Sanchez David et al., 2016). We observed that MDA5 239 binding to SAD negative-strand genomic RNAs correlated with high AU content (> AU content 240 of SAD genome 0.55) (p<0.001, Cohen's d=0.53) ( Figure S4D ). However, RIG-I SAD 241

negative-strand ligands were characterized by a lower AU-rich content sequence than that 242 observed for MDA5 ligands suggesting that RIG-I binding did not correlate with RNA primary 243 sequence ( Figure S4B ). For the positive-strand SAD RNA species, we did not detect any 244 preference of binding of AU-rich regions by RIG-I or MDA5 ( Figure S4A and C). In 245 conclusion, our results suggest preferential binding of MDA5 to AU-rich regions of viral RNA 246 but RIG-I ligand interaction to AU-rich sequence was not observed suggesting that RIG-I more 247

presumably recognizes specific RNA secondary or tertiary structures. Table 1 ). 5'cb DI-1668 exhibited statistically significant (p =0.04, one-way ANOVA) read 258 abundance in the RIG-I output samples as compared to the negative control ST-CH. We failed 259 to detect any statistically significant read enrichment for 5'cb DI-2170 in ST-RIG-I output 260 samples comparing to the negative control ST-CH cells (Table 1) . 261

We further studied molecular organization of the detected 5'cb DI-2170 and 5'cb DI-1668 DI 262 viral genomes and juxtaposed them on the Figure 3B 

Total RNA was extracted using a RNeasy minikit (Qiagen). cDNA was generated from 2 g of 403 total RNA by using Superscript IV vilo (#11766050, Invitrogen) and primer 2 (Table S2) The experiments were performed three times and represented as median with 95% confidence 664

interval. P values were calculated using one way ANOVA with Tukey's multiple comparisons. 665

Non-significant (n.s.) were indicated. *p<0.05, **p< 0.01, ***p<0.001, ****P<0.0001 666 667 curves. Black and blue arrows depicted the breakpoints, BP, and reinitiation sites, RI, for 5' 690 cb DI-2170 and 5'cb DI-1668, respectively. 691

The fold change of RLR binding is obtained by i) normalizing the read coverage of output 692

RNAs co-purified with RIG-I, MDA5 or CH by their input extract, ii) normalized read 693 coverage of output samples were divided by the mean of the coverage of negative control CH, 694

at each genomic position. 695 TGATATCCTCTAGGGTGATCCACCTGTTGTACCTATGAAGGTGTCTTTCCACTTCTCT  GGATAGATCCTTGATGCTGCCAACGAGTCTGGTAGTCTTCACTATCTTGTAAATCAA  CCTGATCCAGTGAGATGACAGACTCAGGCTAGAATTACAGGCTACCCTTTTGAACAC  TGAGGTGTCGTTGGACCAACTCCAAGATAGATAAACGTTCCCTCGAATGACTTGCTT  TCTGAAGTAATAAGTTATAGGTCTGTTATACAGGTCGTGAAGTCTTGCGAAGCTAGG  GACGTCACCTAAAGCAGTAGATAGATACATCATAGTACTGCAACATATGTTGAAGT  GCCTCAGGATTTTGGGATCAGACGGTGGATAGAACGAGGGGTCAGTTAGGGGTTTG  GAAACGTTGAAGACTCTGTTGGAGAAAACTATCATGATGGCTATAATGATAGCCAC  TTTAGACAGAGTTCCCCTCTGTAACTCAAGATCATCAACCATCTTGTGGACTAGAAA  AGATTCTACATCACTGTCAATCAGAGTTATGATCATCTCATTGTAAGGATTTGATAT  GATTTCTTCAGGAAATCCTCTCACAAGATCCTGATAGTTCAAGGAACGAGCTCTCTG  CATCTCACTCTTGGGGCTGCTACAATTGAATAACACAAGGCTCATTTCTCGAAGGGT  GGAAGAGGTCAAGTACTCAGCATCTCTGAAAAACTTCCCTCGTTTGGAGAATCGGA  GGTAGAGCTCAGATGAAAAAGACGAAGTTACTTGGGTGATAAACCCTGTGACCGAG  GGGAACGCTCTTGACAGGTGTTGAATAGCCTTGTAGTTTGGATTTACTAGCATAGTC  CCATAAGTTTTGAAGACCAAATAGAGTGGTCCATCTATAGACAATGCAAAATCGGA  CATTAACAGGGTGATCCGGTTGATAGATGCAATGTCAGTAACTTCTGCATCGCAAAT  AATGAGGTCATAGGACATGTTGACCTGCTTTTGGACTGACTGGAAGTATTTCCAGGT  TGCCAAGTTTCTCAAGTCGGACGGTTTTTCCCAGATTGAGTCAAGATCTATCACTCT  GGAGACGATATCATTTCCTCCCCTCATGATTGCTGAAGGAGGCAGTGGATGTGTTCC  GGAAGCCATCAGGTCATTCACCTCTAAAAGACTGTTGAACACAAGCTTGGCATCTG  GAAACATGTTGAGGACTGCCCTTGATATCCCCCCTGACCCGTCCCCAACTACAAGGC  AGAGAGATGGGAAAACATTGAGATCATCTAGAATAGGCTTAAGCTTATAATGAGCA  CCGGTTGCCCACTGAACCACTCTCAAGCCCGAGATCAAAGGGTTCTGGAACCTCTTA  GAGAGGGCCCTTATGTCAAGCTCCGAGACAGGGGCCGGGTTTGCTGAGGTAGAGAC  TGCAACCTGTTGAGCAGAGCAGACCCATTCTGAACATCCTACCTTACGGGACACCTT  CTTGTTGGGGCTGTAATCTCCAGTCATGGTTCTAGCTGCATGGCGCACCTCTTGATCC  ACCCATCTTGTCCTTCGTAAAGAGTCTTTTAGCAGTCGTTGAATGTTGTCGTCTGACT  CTAAGGTATCTTCTCCGTGCCCGCCCAGCACCTGCCTCATCAAAGAACTCAATTGTC  GCAGGTTATCTCTCATACTCTTAGATAGGTTTCTCTCAACCCTCTGGAGTAGAAGAT  GAGACTGGTAAGTAATGAGGGATAGGTACGTCATTTTGGCACTTCTAAAGTCTGAA  AAGATCCATAGCCAGTCATTCTCTGGAGACGCCGTGATTATCTCTCGCTCATAGCGT  AGCACATGTTGGAGATAACACAAGATTGATCTGTTGCCTTCTTTCATAGTGGTTGGA  TAAGCGGCGGGGATTTTCTGAGGGATAGAAAATATCTCTCCTCTAAGAGACGGTTCT   Primers  Sequence  1  GAGGGGAACGCTCTTG  2  CGCTTAACAAATAAACAACAA  3 AGA AAG CAA GTC ATT CGA GGG 4 GAGGGGAACGCTCTCG 5 CGCTTAACAAAAAAACAATAA

ACGCTTAACAAATAAACAACAAAAATGAGAAAAACAATCAAACAACCAAAGGTTCAGATTT AGGATCTTGTTTTTTTCAAGATACATCACACAAGAGTCTTAGCATGGGCAGGCTTCCAGGAGT  ATCCGGTTCACAGGCAACTGTAGTCTAGTAGGGATGATCTAGATCTGATATCCTCTAGGGTGA  TCCACCTGTTGTACCTATGAAGGTGTCTTTCCACTTCTCTGGATAGATCCTTGATGCTGCCAAC  GAGTCTGGTAGTCTTCACTATCTTGTAAATCAACCTGATCCAGTGAGATGACAGACTCAGGCT  AGAATTACAGGCTACCCTTTTGAACACTGAGGTGTCGTTGGACCAACTCCAAGATAGATAAA  CGTTCCCTCGAATGACTTGCTTTCTGAAGTAATAAGTTATAGGTCTGTTATACAGGTCGTGAA  GTCTTGCGAAGCTAGGGACGTCACCTAAAGCAGTAGATAGATACATCATAGTACTGCAACAT  ATGTTGAAGTGCCTCAGGATTTTGGGATCAGACGGTGGATAGAACGAGGGGTCAGTTAGGGG  TTTGGAAACGTTGAAGACTCTGTTGGAGAAAACTATCATGATGGCTATAATGATAGCCACTTT  AGACAGAGTTCCCCTCTGTAACTCAAGATCATCAACCATCTTGTGGACTAGAAAAGATTCTAC  ATCACTGTCAATCAGAGTTATGATCATCTCATTGTAAGGATTTGATATGATTTCTTCAGGAAA  TCCTCTCACAAGATCCTGATAGTTCAAGGAACGAGCTCTCTGCATCTCACTCTTGGGGCTGCT  ACAATTGAATAACACAAGGCTCATTTCTCGAAGGGTGGAAGAGGTCAAGTACTCAGCATCTC  TGAAAAACTTCCCTCGTTTGGAGAATCGGAGGTAGAGCTCAGATGAAAAAGACGAAGTTACT  TGGGTGATAAACCCTGTGACCGAGGGGAACGCTCTTGACAGGTGTTGAATAGCCTTGTAGTT  TGGATTTACTAGCATAGTCCCATAAGTTTTGAACTCTGTCTAAAGTGGCTATCATTATAGCCA  TCATGATAGTTTTCTCCAACAGAGTCTTCAACGTTTCCAAACCCCTAACTGACCCCTCGTTCTA  TCCACCGTCTGATCCCAAAATCCTGAGGCACTTCAACATATGTTGCAGTACTATGATGTATCT  ATCTACTGCTTTAGGTGACGTCCCTAGCTTCGCAAGACTTCACGACCTGTATAACAGACCTAT  AACTTATTACTTCAGAAAGCAAGTCATTCGAGGGAACGTTTATCTATCTTGGAGTTGGTCCAA  CGACACCTCAGTGTTCAAAAGGGTAGCCTGTAATTCTAGCCTGAGTCTGTCATCTCACTGGAT  CAGGTTGATTTACAAGATAGTGAAGACTACCAGACTCGTTGGCAGCATCAAGGATCTATCCA  GAGAAGTGGAAAGACACCTTCATAGGTACAACAGGTGGATCACCCTAGAGGATATCAGATCT  AGATCATCCCTACTAGACTACAGTTGCCTGTGAACCGGATACTCCTGGAAGCCTGCCCATGCT 

Differential recognition of viral RNA by RIG-I

Antiviral Response

Importance of rabies virus nucleoprotein in viral evasion of interferon response in 575 the brain

MDA-5 recognition of a murine norovirus

Nonencapsidated 5'Copy-Back Defective Interfering Genomes

Recombinant Measles Viruses Are Recognized by RIG-I and LGP2 but Not MDA5

Measles Virus C Protein

Impairs Production of Defective Copyback Double-Stranded Viral RNA and Activation of

Activation of MDA5 Requires Higher-Order

RNA Structures Generated during Virus Infection

BEDTools: A flexible suite of utilities for comparing 588 genomic features

RIG-I-like receptors: their regulation and roles in RNA 590 sensing

RIG-I detects viral genomic RNA during negative-strand 593 RNA virus infection

Dissection of Interferon-Antagonistic Functions of Rabies Virus Phosphoprotein : Inhibition of 596

Interferon Regulatory Factor 3 Activation Is Important for Pathogenicity

In Vivo Ligands of MDA5 and RIG-I in Measles 599

Comparative analysis of viral RNA 602 signatures on different RIG-I-like receptors

LGP2 binds to PACT to 605 regulate RIG-I-and MDA5-mediated antiviral responses

Recognition of 5′ Triphosphate by RIG-I

Helicase Requires Short Blunt Double-Stranded RNA as Contained in Panhandle of Negative-609

A conserved histidine in 612 the RNA sensor RIG-I controls immune tolerance to N1-2'O-methylated self RNA

Differential Type I IFN-Inducing Abilities of Wild

Type versus Vaccine Strains of Measles Virus

Lyssavirus matrix protein cooperates with phosphoprotein to modulate the Jak-Stat 619 pathway

Sendai virus defective-interfering genomes 621 and the activation of interferon-beta

Immunostimulatory Defective Viral Genomes from 624

Respiratory Syncytial Virus Promote a Strong Innate Antiviral Response during Infection in 625 Mice and Humans

Reduced accumulation of 628 defective viral genomes contributes to severe outcome in influenza virus

Attenuated Rabies Virus Activates, while Pathogenic Rabies Virus Evades, the Host

Innate Immune Responses in the Central Nervous System

Conservation of a Unique Mechanism of Immune 635 Evasion across the

Structural basis for dsRNA recognition, filament formation, and antiviral signal 638 activation by MDA5

MDA5 Governs the Innate Immune Response to

Defective Interfering Particles of Negative-Strand RNA

Ribose 2′-O-methylation provides 646 a molecular signature for the distinction of self and non-self mRNA dependent on the RNA 647 sensor Mda5

Detection of 5'cb DI-1668, 5'cb DI-2170 (left panel), and SAD genome

ST-RIG-I, and ST-699 MDA5 cells (MOI of 0.5, 24h post-infection). RT-PCR was performed using universal 700 5' cb primers (1 and 2) and full-length genome primers

SAD and THA genomes (right 702 panel) by RT-PCR on total RNAs extracted from infected neuroblastoma cells 703 (SK.N.SH) at MOI of 0.5 for 24h using universal 5' cb primers (1 and 2) and full-length 704 genome primers

on 2 µg of total RNAs extracted from SAD or THA infected neuroblastoma 706 cells (SK.N.SH) at MOI of 0.5 for 24h using THA-optimized 5' cb DI primers (4 and 707 5) (left panel) and THA-optimized full-length genome primers

In silico analysis of RLR-specific SAD ligands: 767 AU content of specific RLR RNAs partners (A, B) for RIG-I and (C, D) for MDA5. Relation 768 between AU content and mean count for each position

We acknowledge all the members of the Lyssavirus, Epidemiology and Neuropathology unit 489 and the Vaccines-innovation laboratory at Pasteur Institute for their help and useful discussions.