Sequence-Specific Features of Short Double-Strand, Blunt-End RNAs Have RIG-I- and Type 1 Interferon-Dependent or -Independent Anti-Viral Effects

Abstract

Pathogen-associated molecular patterns, including cytoplasmic DNA and double-strand (ds)RNA trigger the induction of interferon (IFN) and antiviral states protecting cells and organisms from pathogens. Here we discovered that the transfection of human airway cell lines or non-transformed fibroblasts with 24mer dsRNA mimicking the cellular micro-RNA (miR)29b-1* gives strong anti-viral effects against human adenovirus type 5 (AdV-C5), influenza A virus X31 (H3N2), and SARS-CoV-2. These anti-viral effects required blunt-end complementary RNA strands and were not elicited by corresponding single-strand RNAs. dsRNA miR-29b-1* but not randomized miR-29b-1* mimics induced IFN-stimulated gene expression, and downregulated cell adhesion and cell cycle genes, as indicated by transcriptomics and IFN-I responsive Mx1-promoter activity assays. The inhibition of AdV-C5 infection with miR-29b-1* mimic depended on the IFN-alpha receptor 2 (IFNAR2) and the RNA-helicase retinoic acid-inducible gene I (RIG-I) but not cytoplasmic RNA sensors MDA5 and ZNFX1 or MyD88/TRIF adaptors. The antiviral effects of miR29b-1* were independent of a central AUAU-motif inducing dsRNA bending, as mimics with disrupted AUAU-motif were anti-viral in normal but not RIG-I knock-out (KO) or IFNAR2-KO cells. The screening of a library of scrambled short dsRNA sequences identified also anti-viral mimics functioning independently of RIG-I and IFNAR2, thus exemplifying the diverse anti-viral mechanisms of short blunt-end dsRNAs.

Trial registration: ClinicalTrials.gov NCT02735824.

Keywords: DNA virus; RIG-I; RNA therapy; RNA virus; SARS-CoV-2; adenovirus; antiviral agents; influenza virus; interferon; short double-strand blunt-end RNA.

Conflict of interest statement

UFG has been a consultant and stock owner in 3V-Biosciences (now Sagimet Biosciences) and a consultant to F. Hoffmann-La Roche Ltd. and to Union Therapeutics A/S. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Figures

Figure 1

A short blunt-end dsRNA with miR29b-1* sequence reduces AdV-C5 infection efficiency. (A). Qiagen miR29b-1* mimic (29b-1*), but not a control mimic with a randomized miR29b-1* sequence (Rand), reduces AdV-C5 infection efficiency in A549 cells. RNAimax was used for the mimic transfections. No RNA refers to non-transfected cells and v1, v2, and v3 to three different two-fold dilutions of input virus. Infection efficiencies were scored by immunostaining for the late virus protein VI, and infection index (left y-axis, bar plot) refers to the fraction of VI-positive cells over total number of cells analyzed (right y-axis, scatter plot red triangles). The two technical replicates are shown separately. Representative images from the v2-input virus infection are shown on the right-hand side for cells transfected with the Rand and miR29b-1* mimics. Scale bar = 50 µm. (B). A549 cells transfected with miR29b-1* mimics from different vendors display different infection phenotypes. The blunt-end dsRNA miR29b-1* mimics from Qiagen (Qiagen 29b-1*) and Microsynth (Micros. blunt 29b-1*) reduced AdV-C5 infection efficiency, but Dharmacon (Dharm. 29b-1*) or Microsynth (Micros. overh 29b-1*) dsRNA mimics with 3′ overhangs did not significantly deviate from the results obtained with the Rand. Shown are mean values from three technical replicates (infection index bar plot, red squares cell numbers). The error bars represent standard deviation. The left and right panels are from two independent experiments. (C). The blunt-end miR29b-1* mimic potently inhibits AdV-C5 infection in non-transformed human fibroblasts. The two technical replicates are shown separately. 120.03 and 105.03 represent cells from two different donors. Grey bars represent infection index, red symbols cell numbers. (D). The blunt-end miR29b-1* mimic significantly reduces virus progeny production in A549 cells. At the shown time points, cell-associated and extracellular progeny virions were collected and titrated on HeLa cells by staining for the late protein VI. RNAimax refers to mock transfection with transfection reagent but without RNA mimics. The titer of medium-associated progeny from miR29b-1* mimic-transfected cells is highlighted as black squares.

Figure 2

The blunt-end dsRNA miR29b-1* mimic has broad anti-viral effects in A549 cells and efficiently reduces IAV and SARS-CoV-2 infections. The IAV infection efficiency was scored by immunostaining for the viral nucleoprotein. The SARS-CoV-2 experiment was carried out in A549-ACE2 cells and infection efficiency was scored by immunostaining against the viral nucleocapsid protein. The bars represent infection index (left y-axis) and the cell numbers analyzed (red squares) are shown as a scatter plot (right y-axis). Shown are for IAV mean values from three technical replicates and the error bars represent standard deviation. The two technical replicates in the SARS-CoV-2 experiment are shown separately. No RNA refers to non-transfected cells, Rand to transfection with the randomized miR29b-1* sequence mimic, and 29b-1* to transfection with the miR29b-1* mimic. Rand and miR29b-1* mimics were from Qiagen in the IAV experiment and from Microsynth in the SARS-CoV-2 experiment.

Figure 3

The blunt-end dsRNA miR29b-1* mimic induces IFN response in cells and this response mediates the anti-viral effects of the oligo in A549 cells. (A) Agilent whole genome microarray analysis of gene expression changes in Qiagen miR29b-1*-transfected A549 cells in comparison to control cells, which were transfected with Qiagen non-targeting dsRNA siAllstar (10 nM). Shown is a pathway enrichment analysis of significantly (p-value ≤ 0.05) up-regulated and down-regulated functional pathways, the heatmap illustrating log2 values of gene expression changes for the indicated genes within the pathways. (B) A549 cells transfected with blunt-end miR29b-1* mimic (29b-1*, 2.5 nM), but not those transfected with randomized miR29b-1* sequence mimic (Rand), secrete type I interferons. Clarified culture supernatants were collected from the cells at 24 h, 48 h, and 60 h post transfection and titrated on 293T indicator cells expressing Firefly luciferase under the type I IFN-inducible Mx1 promoter (jetPEI transfections). The y-axis shows normalized Firefly luciferase activities in cell extracts expressed as fold-changes in comparison to values obtained from untreated indicator cells. Indicated amounts of recombinant IFNα2 were used as controls. The bars represent mean values from three technical replicates with standard deviation. No RNA indicates culture supernatant from non-transfected A549 cells. Rand and miR29b-1* mimics were obtained from Qiagen. (C). The anti-viral effect of the blunt-end dsRNA miR29b-1* mimic is ablated in A549 IFNAR2-KO cells. miR29b-1*-transfected A549-IFNAR2 knockout cells were infected with AdV-C5 and infection efficiencies were scored by immunostaining for the late virus protein VI. Infection index (left y-axis, bar plot) refers to the fraction of VI-positive cells over total number of cells analyzed (right y-axis, scatter plot) and v1, v2, and v3 to three different two-fold dilutions of input virus. The two technical replicates are shown separately. Grey bars represent infection index, red symbols cell numbers.

Figure 4

RIG-I mediates anti-viral effects of the blunt-end dsRNA miR29b-1* mimic. (A) Anti-viral effects of the blunt-end miR29b-1* mimic (29b-1*) require dsRNA structure since the individual ssRNA oligos of the mimic (ss29b-1*_S and ss29b-1*_AS) do not reduce AdV-C5 infection efficiency. A549 cells were infected with AdV-C5 and infection efficiencies were scored by immunostaining for the late virus protein VI. The left y-axis (bar plot) shows the infection index, the right y-axis (scatter plot, red triangles) shows the number of cells analyzed, and v1 and v2 refer to two different two-fold dilutions of the input virus. The two technical replicates are shown separately. No RNA indicates cells treated with the RNAimax transfection reagent alone and Rand refers to cells transfected with the mimic with randomized miR29b-1* sequence. NI is the non-infected control. (B) siRNA-mediated knockdown of RIG-I or MAVS equalized AdV-C5 infection in miR29b-1* and Rand mimic-transfected A549 cells. Infection efficiencies were scored as in (A) and the values shown represent the mean from three technical replicates with standard deviation (grey bars infection index, red squares cell numbers). siRNA-mediated knockdown efficiencies were scored by RT-qPCR. siNT refers to a control, non-targeting siRNA. Rand and miR29b-1* mimics were obtained from Qiagen. (C) The anti-viral effect of the blunt-end dsRNA miR29b-1* mimic (10 nM final concentration) is ablated in A549-RIG-I knockout cells. Results from two different knockout clones created by two different guide RNAs (sg1 and sg2) are shown. The AdV-C5 infection efficiency was scored as in (A). v1, v2, and v3 represent three different input virus amounts with two-fold dilutions. (D) No significant IFN secretion from miR29b-1*-transfected A549-RIG-I KO (sg1) cells. Clarified culture supernatants were collected from the transfected cells at 48 h post-transfection and scored for IFN on 293T indicator cells as described in legend to Figure 3B. The values were normalized to that of mock transfection (RNAimax only) and the two technical replicates in the assay are shown separately.

Figure 5

Altering the centrally positioned AUAU-tract in miR29b-1* does not affect the anti-viral potency of the mimics. (A,B) The blunt-end dsRNA miR29b-1* (29b-1*) contains a centrally positioned AUAU-tract that induces a bend into the RNA molecule. Modified blunt-end dsRNA miR29b-1* mimics with a disrupted central AUAU-tract (29b-1*_sAU2 and 29b-1*_sAU3) still retain anti-viral effects against AdV-C5 in parental but not in RIG-I knockout A549 cells. Infection efficiencies were scored by immunostaining for the late virus protein VI. Infection index (left y-axis, bar plot) refers to the fraction of VI-positive cells over total number of cells analyzed (right y-axis, scatter plot, red triangles). The two technical replicates are shown separately, and v1 and v2 refer to two different two-fold dilutions of the input virus. Rand refers to cells transfected with the mimic with randomized miR29b-1* sequence and NI is the non-infected control. Sequences of the mimics are shown in the 5′ to 3′ orientation leaving the reverse complement strand out. (C) Central AUAU-tract motif is not important for IFN induction. The experiment was carried out as described in legend to Figure 3B. dsRNA were used at 10 nM final concentration. No RNA refers to medium from cells treated with the transfection reagent RNAimax alone. (D) Significant infection recovery in miR29b-1*_sAU2- or miR29b-1*_sAU3-transfected IFNAR2 KO A549 cells. Infection efficiencies were scored as described in panel (A).

Figure 6

Short blunt-end dsRNAs with randomized sequences commonly show anti-viral effects but not always via RIG-I and IFN. (A) AdV-C5 infection efficiency in wild type A549 cells transfected with scrambled, blunt-end 24-nucleotide long dsRNAs. The sequences of the mimics used are listed in Supplementary Table S1. Infection efficiencies were scored by immunostaining for the late virus protein VI and infection index (left y-axis, bar plot) refers to the fraction of VI-positive cells over total number of cells analyzed (right y-axis, scatter plot red triangles). The two technical replicates are shown separately. (B,C) Test of the potent anti-viral scrambled dsRNAs in A549 RIG-I KO (B) and IFNAR2 KO (C) cells. Analyses were carried out as in (A), except that two different two-fold dilutions of the input virus were used. The experiments shown in Figure 5D and Figure 6C are from the same 96-well plate and therefore the Rand and 29b-1* controls are the same in these two figures. Grey bars represent infection index, red symbols cell numbers. (D) List of sequences of the guide strands (5′ to 3′-orientation) of dsRNAs miR29b-1*, Rand, and scr 20, 24, 32, 33, where numbers are in italics indicate dsRNAs with a base composition distinct from miR29b-1*, Rand, scr 20, and 24.