Recognition of 5' triphosphate by RIG-I helicase requires short blunt double-stranded RNA as contained in panhandle of negative-strand virus

Abstract

Antiviral immunity is triggered by immunorecognition of viral nucleic acids. The cytosolic helicase RIG-I is a key sensor of viral infections and is activated by RNA containing a triphosphate at the 5' end. The exact structure of RNA activating RIG-I remains controversial. Here, we established a chemical approach for 5' triphosphate oligoribonucleotide synthesis and found that synthetic single-stranded 5' triphosphate oligoribonucleotides were unable to bind and activate RIG-I. Conversely, the addition of the synthetic complementary strand resulted in optimal binding and activation of RIG-I. Short double-strand conformation with base pairing of the nucleoside carrying the 5' triphosphate was required. RIG-I activation was impaired by a 3' overhang at the 5' triphosphate end. These results define the structure of RNA for full RIG-I activation and explain how RIG-I detects negative-strand RNA viruses that lack long double-stranded RNA but do contain blunt short double-stranded 5' triphosphate RNA in the panhandle region of their single-stranded genome.

Fig. 1. Fully synthetic 5′triphosphate single-stranded RNA is not sufficient to activate RIG-I

A: Reverse Phase HPLC analysis of 3P-A = pppAAC ACA CAC ACA CAC ACA CAC UUU after deprotection under standard ACE deprotection conditions (pH=3.8, 60°C in 30 min). B: MALDI-ToF analysis of 3P-A: calculated molecular weight of 3P-A is 7770; the peak at 3886 represents the double charged peak (z=2). C: MALDI-ToF analysis of 3P-A mixed with HO-A. The difference between the molecular weight of the product 3P-A (1) and the educt HO-A (2) is 240 which corresponds to the molecular weight of an additional triphosphate group (H₃P₃O₉). D: Purified monocytes were stimulated with the indicated single-stranded or double-stranded synthetic or in vitro transcribed RNA oligonucleotides. The questionmark indicates that the 3′end of this in vitro transcribed RNA is not molecularly defined. IFN-α production was analysed 24 hours after stimulation. Data from four independent donors are depicted as mean values ± SEM. E: Indicated RNA stimuli (see suppl. Table 1) were analysed on a denaturing 12 % polyacrylamide gel (containing 50% urea w/v) and stained with methylene blue detecting single-stranded and double-stranded RNA. ivt3P-G without U was generated by in vitro transcription in the absence of the nucleotide UTP. F: Purified monocytes were stimulated with the indicated single-stranded or double-stranded synthetic or in vitro transcribed RNA oligonucleotides. IFN-α production was analysed 24 hours after stimulation. Data from four independent donors are depicted in the bar graph as mean values ± SEM. RNAs were analysed on a denaturing 12 % polyacrylamide gel (containing 50 % urea w/v) and stained with methylene blue.

Fig. 2. RIG-I activation requires a short double strand and prefers 5′-adenosine

Purified monocytes were stimulated with the indicated single-stranded or double-stranded synthetic RNA oligonucleotide. IFN-α production was analysed 24 hours after stimulation. Data from n independent donors are depicted as mean values ± SEM, n as indicated. A: 3P-A was hybridized with synthetic antisense single-stranded RNA with different lengths generating double-stranded RNA with blunt end carrying the triphosphate group. B: 23mer (AS G23 and AS A23) with a blunt triphosphate end and a 3′overhang at the non-triphosphate end. C: Comparison of RIG-I ligand activity of 3P-G to the other three synthetic variants (3P-A, 3P-C, 3P-U) which were hybridized with the corresponding synthetic 24mer (AS-A24, AS-C24, AS-U24) or used as single strands. D: IFN-α inducing activity of 3P-A+AS A24 and 3P-G+AS G24.

Fig. 3. Blunt end at the triphosphate end but not at the non-triphosphate end contributes to RIG-I ligand activity, and 5′monophosphate does not substitute for 5′triphosphate

Purified monocytes were stimulated with the indicated single-stranded or double-stranded synthetic RNA oligonucleotides. IFN-α production was analysed 24 hours after stimulation. Data from four independent donors are depicted as mean values ± SEM. 3P-G and 3P-A were hybridized with corresponding antisense strands with different lengths and positions: A: The use of 25mer (AS G24+A and AS A24+A) and 26mer (AS G24+2A and AS A24+2A) results in a mononucleotide or dinucleotide 5′ overhang at the non-triphosphorylated end. B: The use of 25mer (AS G25 and AS A25) and 26mer (AS G26 and AS A26) results in a mononucleotide or dinucleotide 3′ overhang at the triphosphate end. C: The use of 19mer, 21mer and 23mer single-stranded antisense RNA (AS19, AS21, AS23) results in a 5′overhang at the triphosphate end (-5nt, -3nt, -1nt). D: IFN-α-inducing activity of single-stranded 5′monophosphate RNA (P-A, P-G, P-U, P-C) and synthetic 5′triphosphate single-stranded RNA (3P-A) and combinations with complementary strands of different lengths are compared.

Fig. 4. RIG-I is not activated by highly structured 3P-ssRNA and tolerates central but not distal mismatches of 3P-dsRNA

Purified monocytes were stimulated with the indicated single-stranded or double-stranded synthetic RNA oligonucleotides. IFN-β production was analysed 24 hours after stimulation. Data from four independent donors are depicted as mean values ± SEM. Analysis of A) related 3P-RNAs (3P-GFP1 (Hornung et al., 2006), and variations thereof), with gradual stability of secondary structure, B) 3P-GFP2 hybridized with corresponding antisense strands with different lengths and positions as indicated, C) 3P-GFP2 hybridized with corresponding antisense strands with single or multiple mismatched nucleotides at different positions as indicated, D) PAGE-purified in vitro transcripts corresponding to the predicted panhandle structure of Rabies virus genome with blunt end (match) and with 3′overhang of a poly A tail and a mismatched (2 nt) triphosphorylated end (mismatch).

Fig. 5. Synthetic 3P-dsRNA is recognized by RIG-I

Mouse embryonic fibroblasts (MEFs) with indicated genotype were transfected with indicated 3P-dsRNA. Ifih1(+/+)and Mavs (+/+)represent wild type MEFs with the same passage number as the corresponding mutant MEFs Ifih1 (-/-) and Mavs (-/-), respectively. Triplicate data of a representative experiment are depicted as mean values ± SEM.

Fig. 6. IFN-α inducing activity of RIG-I RNA ligands correlates with RIG-I ATPase activity and with RIG-I binding affinity

A: For the ATPase assay purified RIG-I protein was incubated with increasing amounts of indicated RNA molecules (from 10^-9 nM to 1800 nM) and the release of ADP was analysed after 30 min at 37°C by a FRET-based competitive immunoassay. The percentage of ADP release is plotted against the decadic logarithm of the concentration of indicated RNAs. Triplicate data of a representative experiment are depicted as mean values ± SEM. Half effective concentration (EC50) was determined by statistical analysis (non-linear regression). Low EC50 represents high RIG-I ATPase activity. B: The EC50 of the synthetic 3P-A hybridized with the indicated antisense RNAs are compared. C: The EC50 of 5′triphosphate double-stranded RNA with different 5′bases (A, G, U, C) are compared. D-G: Alphascreen: Purified (His6)-tagged RIG-I was incubated with different RNA molecules bearing a biotin (*) on the 5′end of the antisense strand (at the non-triphosphate end of the double-stranded RNA; D-F) or at the 3′ end of the 3P-bearing strand (G). (His6)-tagged RIG-I protein was bound to Ni-chelate beads (donor); biotinylated RNA was bound to streptavidine beads (acceptor). The resulting fluorescence correlates with the number and proximity of interacting donor - acceptor pairs. Concentration of indicated RNAs is plotted against the percentage of maximum binding to RIG-I. The dissociation constant Kd(app) is calculated by statistical analysis (non-linear regression). E, F and G: Kd(app) of non-modified, monophosphate and triphosphate RNAs hybridized to indicated antisense RNAs of different lengths are compared. F: CIAP 3P-A + AS A24 was incubated with active alkaline phosphatase; the other stimuli were incubated with heat-inactivated active alkaline phosphatase. H: Purified (His6)-tagged RIG-I protein was analysed by SDS-PAGE and Coomassie Blue staining.

Fig. 7. Molecular structure of RIG-I ligand

Optimal RIG-I ligand activity requires short double-stranded RNA containing at least one 5′triphosphate. The nucleoside carrying the 5′triphosphate needs to be part of the Watson-Crick base pairing (GU wobble base pairing less efficient) and pppA, pppG and pppU are preferred over pppC. Although overhangs at the non-triphosphorylated end of the double-stranded RNA have no major impact on RIG-I activity, a 3′overhang at the 5′triphosphorylated end impairs RIG-I activity.