Oligoadenylate Synthetase 1 displays dual antiviral mechanisms in driving translational shutdown and protecting interferon production

Graphical Abstract

SUMMARY

In response to viral infection, how cells balance translational shutdown to limit viral replication and the induction of antiviral components like interferons (IFNs) is not well understood. Moreover, how distinct isoforms of IFN-induced Oligoadenylate Synthetase 1 (OAS1) contribute to this antiviral response also requires further elucidation. Here, we show that human, but not mouse, OAS1 inhibits SARS-CoV-2 replication through its canonical enzyme activity via RNase L. In contrast, both mouse and human OAS1 protect against West Nile virus infection by a mechanism distinct from canonical RNase L activation. OAS1 binds AU-rich elements (ARE) of specific mRNA, including IFNβ. This binding leads to the sequestration of IFNβ mRNA to the endomembrane regions resulting in prolonged half-life and continued translation. Thus, OAS1 is an ARE binding protein with two mechanisms of antiviral activity: driving inhibition of translation, but also a broader non-canonical function protecting IFN expression from translational shutdown.

ETOC BLURB

How cells induce the production of antiviral molecules while shutting down protein translation to limit viral replication remains unclear. Here, Harioudh et al show a non-canonical function for the interferon-induced antiviral protein, OAS1, which sequesters and protects Ifnb mRNA from degradation to sustain innate antiviral protection against West Nile Virus.

INTRODUCTION

The early defense against viral infection is initiated by the cellular innate immune response and mediated principally by type I interferons (IFNs). Beyond this, type III IFNs also contribute to tissue-specific antiviral immunity against viruses at barrier sites¹. Through induction of IFN-stimulated genes (ISG), both type I and type III IFNs provide the initial protection from viral infections and shape the subsequent adaptive response to resolve infection². However, among hundreds of described ISGs, well-defined molecular mechanisms are known for only a subset of ISGs^1,3–5. One common mechanism for the antiviral function of ISGs is the generalized inhibition of protein translation by dsRNA-activated enzymes (e.g., PKR, OAS, ADAR, etc.)⁴. However, additional modes of antiviral activity are known for some ISGs⁶, including viperin^7,8, IFIT and IFITM proteins⁹, APOBEC¹⁰, TRIM proteins¹¹, IFI16¹², and cGAS¹³. Nevertheless, virus-specific functional mechanisms for the vast majority of ISGs remain unknown.

2’–5’-Oligoadenylate synthetases (OAS) are a family of ISGs (4 human and 12 mouse genes) that belong to a larger family of nucleotidyl-transferases (NTase)^14,15. Canonically, following induction by IFNs, OAS proteins bind viral dsRNA and are activated to synthesize short 2’–5’ oligoadenylates (2–5A) from ATP. These 2–5A molecules bind and activate RNase L, which degrades cellular and viral RNA to inhibit translation. However, work utilizing gene-deficient human cells suggests that only one OAS isoform (OAS3) is necessary and sufficient for RNase L activation^16,17, which has left the functional significance of other enzymatically active (NTase⁺) and enzymatically inactive (NTase⁻) OAS proteins unresolved.

The SNP rs10774671 in human OAS1 gene is associated with susceptibility to infection by several viruses including WNV¹⁸, HCV^19–21 and SARS-CoV-2^22–26. The polymorphism rs10774671 (A/G) is located at the splice acceptor site of the exon 6 on human OAS1 gene. The G allele, which results in the OAS1 transcript variant 1 mRNA and protein P46 is associated with viral resistance^23,24,27,28. The A allele, which codes for the transcript variant 2 and protein P42, correlates with susceptibility to virus infection^18,19,29. Individuals or cell lines carrying G/A heterozygous allele predominantly express P46²⁸, which differs from P42 only in the C-terminal region sequence (54 residues), and contains a conserved CaaX motif^23,30. Prenylation of OAS1 at this CaaX motif leads to its endomembrane localization promoting the binding of viral dsRNA and activation of OAS1. However, this mechanism does not explain the protective effect of mouse Oas1 against WNV.

Mouse chromosome 5 contains 8 Oas1 genes (Oas1a–h). A number of these Oas1-genes lack NTase activity³¹. Genetic studies have mapped one such enzymatically inactive gene, Oas1b as a WNV resistance gene in mice^32–35. Some laboratory strains of mice (e.g., C57BL/6 or BALB/c) have a stop codon in the C-terminal half of the Oas1b gene resulting in a truncated protein, whereas other strains encode the full-length protein conferring WNV resistance. Although the full-length Oas1b lacks NTase activity³⁶, its expression still confers resistance to virus infection³⁷, which indicates that the resistance is likely RNase L independent. Oas1b localizes to the ER and interacts with ATP binding cassette family protein ABCF3³⁸, but a mechanism for the antiviral activity of Oas1b has not yet been established.

In the contest between the infecting virus and the host innate immune response to control cellular protein synthesis, the translational shutdown initiated by OAS/RNase L and PKR creates a conundrum for the host³⁹. Cells need to maintain synthesis of new IFN and ISGs to protect neighboring cells from infection, the so-called antiviral state. Studies have suggested a reprograming of translation following OAS-RNase L pathway activation and escape of specific mRNA, such as IFNβ, from translational shutdown during virus infection or viral mimic polyinosinic:polycytidylic acid (pIC) stimulation^40–42. One possible mechanism of such protection is continued location-specific translation of specific mRNA⁴³. Sequence elements present in the UTR, such as ARE can sequester specific mRNA to ER-associated granules and continue their translation^44,45. However, a specific mechanism of IFN and ISG translation during virus infection is still unclear.

Here we report that OAS1 is an ARE binding protein and binds several cellular mRNA, including IFNβ, through its RNA-binding domain. Specific human and mouse OAS1 isoforms, by the virtue of their endomembrane associated localization, can bind and prolong the half-life of IFNβ mRNA. The resulting enhanced IFN expression and downstream IFN receptor (IFNAR) signaling inhibits WNV replication without requiring OAS1 NTase activity or RNase L. In contrast, human OAS1 through its 2–5A synthetase activity and RNase L activation inhibits SARS-CoV-2. These results establish a paradigm for two distinct mechanisms of antiviral activity for OAS1.

RESULTS

Virus-specific antiviral activity of mouse Oas1b in vivo

Among various OAS1 isoforms in mice, Oas1b is identified as a WNV resistance allele. C57BL/6, BALB/c, and other mouse strains carry a stop codon at the C-terminal of Oas1b coding region resulting in a premature truncation of Oas1b protein and consequent susceptibility to WNV^32,33,37. To investigate the mechanism of Oas1b-mediated WNV resistance in vivo, we created an Oas1b-knockin mouse (Oas1b-KI) restoring the full-length murine Oas1b expression in C57BL/6J background, using CRISPR/Cas9-based gene editing (Figures S1A–C).

Subcutaneous inoculation of Oas1b-KI mice with 10² plaque-forming units (PFU) WNV (New York 1999 strain, WNV-NY) resulted in a 100% survival rate, which contrasted with wild-type (WT) controls that uniformly succumbed to infection (Figure 1A). Primary fibroblasts obtained from the Oas1b-KI mice showed reduced WNV growth compared to fibroblasts from WT C57BL/6J mice (Figure 1B). Analysis of viral burden in different organs of the WNV-NY-infected mice revealed markedly reduced infection in the Oas1b-KI mice than WT mice (Figure 1C). We carried out similar experiments with Powassan virus (POWV), a distantly related, neurotropic tick-transmitted flavivirus, and vesicular stomatitis virus (VSV), a rhabdovirus and found the Oas1b-KI mice were substantially protected from several different viral infections (Figures S1D–E). Given that Oas1b lacks NTase activity, these results confirmed that mouse Oas1b confers resistance to flavivirus infection in vivo and in vitro in an RNase L-independent manner as reported previously⁴⁶. In contrast, human OAS1 inhibits SARS-CoV-2 replication through its NTase activity and RNase L activation^23–25. Thus, to examine antiviral activity of Oas1b against SARS-CoV-2, we inoculated Oas1b-KI and WT mice with the naturally mouse-adapted SARS-CoV-2 B.1.351 variant. However, unlike WNV, Oas1b-KI mice failed to control SARS-CoV-2 viral burden in various organs (Figure 1D). These results show that unlike human OAS1, which can inhibit both WNV¹⁸ and SARS-CoV-2, mouse Oas1b only inhibits WNV. As mouse Oas1b lacks NTase activity, the above observation led us to hypothesize NTase dependent and independent mechanisms of virus-specific antiviral activities of OAS1.

Figure 1. . Mouse Oas1b protects against WNV infection in vivo but not from SARS-CoV-2.

(A) Survival of C57BL/6 WT and Oas1b-KI mice infected with WNV-NY (10² PFU) by footpad infection. Survival was measured for 21 days. The statistical significance of Oas1b-KI survival compared to WT mice was assessed using a Mantel-Cox test. Data is pooled from two independently repeated experiments.

(B) Primary fibroblasts from WT and Oas1b-KI mice were inoculated with WNV-NY at an MOI of 0.01. Virus particles were quantified 48 and 72 h postinfection by foci forming unit assay (FFA) on Vero cells. Mean and SEM were calculated from two dilutions of duplicate infections on Vero cells. The limit of detection in our FFA is 20 FFU/mL. Statistical significance was assessed using a two-way ANOVA with Sidak’s multiple comparison test. Unless otherwise mentioned, in all figures, significance is represented as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

(C) WNV-NY virus burden in the brain, spinal cord, spleen, and serum of WT and Oas1b-KI mice was determined by plaque assay of samples from mice at 2, 4 or 7 days post-infection. The limit of detection in our plaque assay is 20 PFU/ml. Mean and SEM of viral loads from tissue were plotted and the statistical significance of Oas1b-KI compared to WT mouse tissue titers was assessed using a Mann-Whitney test.

(D) Eight weeks old C57BL/6 WT and Oas1b-KI mice were inoculated intranasally with 10⁶ FFU of SARS-CoV-2 B.1.351 variant. Viral RNA levels were measured at 3 or 7 days post-infection using RT-qPCR. The limit of detection in our assay is 20 PFU/ml. Mean and SEM of viral loads from indicated tissue were plotted and compared for statistical significance using a Mann-Whitney test.

Two distinct mechanisms of antiviral activity of human OAS1

The SNP (rs10774671) in the 3’ end of human OAS1 locus determines the expression of different alternatively spliced isoforms of OAS1 (Figure S2A)^27,28. We genotyped several commonly used human cell lines and found that HEK293T and HeLa cells have homozygous A/A; THP1, HT1080 and BJ-Tert have heterozygous G/A; and Daudi cells have homozygous G/G genotypes (Figure S2B). As described before²⁸, IFN treatment induced OAS1 isoform 2, P42 protein in HeLa cell line (A/A), whereas OAS1 isoform 1, P46 was the primary OAS1 protein induced in G/A cell lines HT1080 and BJ-Tert (Figure S2C). To define isoform-specific functions of OAS1, we created multiple OAS1-deficient (OAS1-KO) cell lines either in homozygous A/A (HeLa) or heterozygous G/A (HT1080 and BJ-Tert) background using CRISPR/Cas9 gene editing (Figure S2C). As the antiviral activity of OAS1 has been shown most consistently against WNV, we used the BSL2 Kunjin strain of this virus (WNV-KUN) as a model. In a pairwise comparison between WT and OAS1-KO cells, WNV-KUN showed higher viral loads in various OAS1-KO heterozygous G/A cell lines (HT1080 and BJ-Tert), but homozygous A/A containing HeLa cells had no significant differences in viral loads between WT and OAS1-KO cells (Figure 2A). This suggests that the antiviral activity of OAS1 against WNV-KUN is associated with the expression of OAS1 P46 isoform. For subsequent investigations, we used HT1080 cells because of the ease of genetic manipulation of these cells compared to BJ-Tert fibroblasts. However, the majority of the experimental results were also confirmed in BJ-Tert cells to enhance rigor. To avoid non-physiological overexpression through transient transfection, we created stable cell lines with inducible expression of P42 or P46 in OAS1-KO HT1080 cells (HT1080 OAS1-KO iP42 and iP46, Figure S2D). Doxycycline (Dox)-induced expression of P46 but not P42 in OAS1-KO HT1080 inhibited WNV-KUN growth at two different time points (Figure 2B and Figure S2E), indicating that the P46 isoform of OAS1 resulting from the rs10774671 G genotype mediates the antiviral activity against this virus.

Figure 2. Human OAS1 inhibits WNV and SARS-CoV-2 infection through two different mechanisms.

(A) WT and OAS1-KO HeLa, HT1080, and BJ-Tert cells were inoculated with WNV-KUN at an MOI of 1. Culture supernatants were collected at indicated hours post-infection (hpi) and quantified for infectious virus particles by FFA on Vero cells. For each cell group, supernatants from two independent infections were used to infect Vero cells in duplicates. Mean and SEM of the calculated FFU/mL were plotted. The statistical significance of OAS1-KO cells compared with WT cells was assessed using a two-way ANOVA with Sidak’s multiple comparison test and represented as before.

(B) HT1080 OAS1-KO, OAS1-KO with inducible OAS1 P42 (OAS1-KO iP42), and OAS1 P46 (OAS1-KO iP46) cells were pre-treated with 2.5 or 1.5 μg/ml Doxycycline (Dox), respectively, for 24 h and then infected with WNV-KUN at an MOI of 1. Culture supernatants from two independent infections were collected 24 h post-infection, and infectious particles were quantified by FFA on Vero cells as described above. The statistical significance of Dox-treated cells compared to untreated cells was assessed as described above.

(C) HT1080 OAS1-KO cells expressing inducible OAS1 P46 and OAS1 P46 D75A/D77A mutant (OAS1-KO iP46 DADA) were pre-treated with 1.5 μg/ml Dox for 24 h and inoculated as indicated followed by FFA on Vero cells. The statistical significance was calculated from two independent infections as above.

(D) HT1080 WT, OAS1-KO, RNase L-KO, OAS1/RNase L-DKO and OAS1/RNase L-DKO inducible OAS1 P46 (pre-treated with 1.5 μg/ml Dox for 24 h) were inoculated with WNV-KUN as indicated, followed by virus titration. Mean and SEM were plotted as described previously, and the statistical significance of all groups was assessed using a one-way ANOVA with Tukey’s multiple comparison test. Representative result is shown from two independently repeated experiments.

(E–F) HT1080 OAS1-KO iP46 (E) and iP46 DADA (F) cells stably expressing human ACE2 (rACE2) were pre-treated with 1.5 μg/ml Dox for 24 h and inoculated with SARS-CoV-2 B.1.351 variant at an MOI of 1. Culture supernatants were collected 24, 48, and 72 h post-infection, followed by FFA on Vero-hACE2-TMPRSS2 cells. The statistical significance of Dox-treated cells compared to untreated cells was calculated using a two-way ANOVA with Sidak’s multiple comparison test. Results are from three (E) and two (F) independently repeated experiments.

(G) HT1080 OAS1-KO iP46 and OAS1/RNase L-DKO iP46 cells with stable ACE2 expression were pre-treated with 1.5 μg/ml Dox for 24 h and infected with SARS-CoV-2 B.1.351 variant at an MOI of 1. Culture supernatants were collected 48, and 72 h post-infection, followed by FFA on Vero-hACE2-TMPRSS2 cells. The statistical significance of Dox-treated cells compared to untreated cells was calculated using a two-way ANOVA with Tukey’s multiple comparison test. Results are from two independently repeated experiments.

Next, we examined whether the NTase activity was necessary for the antiviral activity of P46 against WNV. The active site of OAS enzymes contain critical Asp residues necessary for the NTase activity^47,48. To test this, we induced expression of WT and enzymatically inactive mutant of P46, D75A/D77A (DADA) in OAS1-KO HT1080 cells (iP46 and iP46 DADA, Figure S2G). The P46 DADA mutant suppressed WNV-KUN replication as well as the WT P46 protein at two different time points (Figure 2C and Figure S2F). To validate these results in a different human cell, we used OAS1-KO BJ-Tert cells and induced expression of P42, P46, or P46 DADA mutant proteins (Figure S2H). Again, WT P46 and the P46 DADA mutant inhibited virus growth (Figures S2I–J) confirming the enzyme activity-independent and isoform-specific antiviral activity of P46.

According to the canonical mechanism, RNase L is required for the antiviral activity of OAS proteins⁴⁹. However, the loss of RNase L did not increase overall viral replication for Zika virus, another flavivirus, suggesting virus-specific antiviral roles of RNase L⁵⁰. Our finding that the antiviral activity of P46 was independent of its NTase activity led us to hypothesize that RNase L might not be required for the inhibition of WNV. To test this hypothesis, we generated RNase L-deficient (RNase L-KO), and OAS1/RNase L-double deficient (OAS1/RNase L-DKO) HT1080 cells, in which we inducibly expressed P46 (HT1080 OAS1/RNase L-DKO iP46, Figure S2K). As expected, OAS1-KO cells showed higher viral loads than WT cells (Figure 2D and Figure S2L), whereas RNase L-KO cells had similar virus growth as the WT cells (Figure 2D and Figure S2L). However, OAS1/RNase L-DKO cells showed similar virus growth compared to the OAS1-KO cells, and inducible expression of P46 in OAS1/RNase L-DKO cells restricted virus infection irrespective of RNase L (Figure 2D and Figure S2L). Thus, the antiviral activity of P46 against WNV does not require its enzyme activity or RNase L expression.

To further assess if human OAS1 inhibits SARS-CoV-2 replication through its NTase activity and requires downstream activation of RNase L, we stably expressed human ACE2 in the HT1080 cells (HT1080^rACE2) and quantified SARS-CoV-2 replication. Dox-induced expression of WT, but not the DADA mutant P46, significantly inhibited SARS-CoV-2 replication at two different multiplicities of infection (MOI) (Figures 2E–F and Figures S2M–N). Additionally, we established the downstream requirement of RNase L by showing the loss of SARS-CoV-2 inhibition by P46 in OAS1/RNase L-DKO cells (Figure 2G). Collectively, these results indicate that human OAS1 through its enzyme activity, produces 2–5A that leads to RNase L activation and inhibition of SARS-CoV-2. However, OAS1 can also inhibit WNV through a non-canonical mechanism, which like mouse Oas1b does not require NTase activity.

Importance of cellular localization and RNA-binding for anti-WNV activity of OAS1

The unique C-terminal tail and the CaaX motif of P46 allows its prenylation and localization at the endomembrane, which is important for its anti-SARS-CoV-2 activity^23,24. To examine the importance of this feature for WNV inhibition, we first characterized the specific localization of OAS1 in our system. Following validation of our staining methods for endogenous and inducibly expressed OAS1 (Figures S3A–B), we characterized the cellular localization of WT P46, P42 and a CaaX motif mutant P46 (C397S) using HT1080 OAS1-KO cells expressing various OAS1 (HT1080 OAS1-KO iP46, iP42 and iP46 C397S, Figures S2D and S3C). As shown in Figure 3A, P46 exhibited unique nucleus-adjacent localization pattern compared to P42, which was abolished upon the CaaX motif mutation (C397S). To define the specific endomembrane compartment that P46 is associated with, we used two marker proteins Calnexin and Golgin-97, respectively associated with ER and Golgi bodies, and quantified OAS1 colocalization. This analysis showed higher correlation of WT P46 expression with ER and Golgi than P42 or C397S mutant P46 (Figure 3B). To further confirm the endomembrane association of P46, we used a different approach, digitonin permeabilization, to fractionate cytosol and endomembrane. This again showed predominant association of P46 with the membrane fraction, which was not the case with P42 and P46 C397S mutant (Figure 3C). Next, we examined the importance of this localization by measuring WNV-KUN replication in P46 C397S mutant expressing cells; this mutant OAS1, which does not localize to the endomembrane, failed to suppress WNV-KUN infection (Figure 3D).

Figure 3. Endomembrane localization and RNA-binding activity of OAS1 P46 is necessary for its antiviral activity against WNV.

(A) Representative confocal micrographs of OAS1 P46, P42 and P46 C397S in HT1080 OAS1-KO cells. HT1080 OAS1-KO iP46, iP42, and iP46 C397S cells were treated with 1.5, 2.5 and 1.5 μg/ml Dox, respectively, for 24 h, followed by staining with OAS1, Golgin-97 and Calnexin antibodies. Scale Bars represent 10μm.

(B) Pearson’s correlation coefficients of individual cell ROI profiles were determined for at least 30 frames of iP46, iP42, and iP46 C397S with Golgin-97 (Golgi body), Calnexin (ER), and DAPI (Nucleus) and plotted. Paired t tests were used for indicated statistical comparisons.

(C) HT1080 OAS1-KO cells with inducible expression of P42 (iP42), P46 (iP46) and P46 C397S mutant (iP46 C397S) were permeabilized with Digitonin to isolate cytosol and membrane fractions followed by immunoblotting of fractions with indicated antibodies.

(D) HT1080 OAS1-KO iP46 or iP46 C397S cells were inoculated with WNV-KUN as indicated after pre-treating with 1.5 μg/ml Dox followed by FFA. The statistical significance was calculated from two independent infections using a two-way ANOVA with Sidak’s multiple comparison test.

(E) HT1080 OAS1-KO cells were transfected overnight with Oas1b WT and various RNA binding mutants (K42E/K57E, K42E/K57E/K60E, K60E, K60E/K191E) for 24 h, followed by WNV-KUN infection at an MOI of 1. Cell supernatants were collected 24 h post-infection, and infectious particles were quantified by FFA on Vero cells. The statistical significance was assessed using a one-way ANOVA with Dunnett’s multiple comparison test. Results are from two independently repeated experiments.

(F) HT1080 OAS1-KO iP46 and iP46 K60E cells were infected as indicated followed by FFA as before. The statistical significance was calculated using a two-way ANOVA with Sidak’s multiple comparison test.

In the context of various mouse Oas1 genes, C57BL/6 mice carry full-length Oas1a as well as another paralog Oas1g genes, both of which do not confer resistance to WNV infection regardless of the presence of C-terminal CaaX motif. However, the full-length Oas1b despite the absence of the CaaX motif protects C57BL/6 mice from lethal WNV infection (Figure 1A). To investigate this isoform-specific anti-WNV activity, we expressed multiple NTase⁺ and NTase⁻ Oas1 cDNAs in OAS1/RNase L-DKO HT1080 cells (Figure S3D) and subsequently measured WNV-KUN infection. Among 4 different Oas1 genes, only Oas1b (NTase⁻) expression robustly inhibited virus growth (Figure S3E). In comparison, Oas1g (NTase⁺) modestly inhibited WNV growth, whereas Oas1a (NTase⁺) did not show any inhibitory activity. Despite multiple efforts, poor protein expression of Oas1h (NTase⁻) cDNA (Figure S3D) precluded us from determining its effects on virus growth. To corroborate and extend our findings, we generated stable HT1080 OAS1-KO cells with inducible expression of FLAG-tagged Oas1a and Oas1b. Immunofluorescence analysis showed a unique subcellular localization of Oas1b that co-localized with Calnexin indicating its endomembrane association, which may be due the presence of a predicted transmembrane region in the C-terminus³⁸. In contrast, Oas1a showed diffuse cytoplasmic, and nuclear localization (Figure S3F). These results suggest that specific endomembrane associated localization is essential for the antiviral activity of OAS1.

Multiple structural and biochemical studies have characterized the dsRNA binding and activation mechanisms of OAS proteins^{14,15,51–54}. The consensus model is that a positively charged surface of OAS1 on the side opposite from its catalytic site is responsible for dsRNA binding. Several positively charged residues on this surface are essential for the RNA binding, confirmed by mutagenesis^48,51. Since Oas1b inhibits WNV replication without requiring enzyme activation but contains conserved residues that are important for RNA binding, we investigated the role of RNA-binding using Oas1b as a template. A series of Oas1b mutants were made targeting these consensus RNA-binding residues and expressed in OAS1-KO HT1080 cells (Figure S3G). Virus growth measurements showed that a single K60E mutation in Oas1b abolished its antiviral activity against WNV-KUN (Figure 3E). The same K60E mutation in human P46 also resulted in a loss of its antiviral activity against WNV-KUN (Figure S3H and Figure 3F) without altering the unique localization of P46 (Figure S3I). These results suggest that in addition to its endomembrane localization, the RNA-binding abilities of P46 and Oas1b are necessary for the protection against WNV infection.

OAS1 binds a specific set of cellular RNA

Although virus-derived dsRNA are thought to be the ligands for the enzyme activation of OAS proteins, multiple studies have shown binding of various cellular and synthetic RNAs to OAS proteins⁵⁴. Given the importance of RNA-binding activity described above, we carried out RNA-immunoprecipitation (RIP) followed by sequencing (RIPseq) to identify cellular RNAs that bind to OAS1. pIC stimulated HT1080 cells expressing WT P46 were subjected to formaldehyde crosslinking, followed by immunoprecipitation with OAS1 antibody or IgG, and sequencing of the recovered RNA. In these experiments, we also performed RIPseq analysis in cells expressing OAS1 K60E mutant (Figure S3H) as a control to identify with high-confidence OAS1 bound cellular RNAs with functional consequences. RIPseq data analysis was restricted to RNAs with at least 1 FPKM (Fragments Per Kilobase of transcript per Million mapped reads) expression in input samples to avoid lowly expressed genes. Following normalization of OAS1 and IgG RIP samples to the input sample, we detected 167 mRNAs that were significantly enriched in OAS1 RIP condition compared to IgG control RIP (log2 fold-change >1 and P < 0.05) (Figure 4A; Table S1). The vast majority of OAS1 bound transcripts, except two, were not significantly enriched in OAS1 K60E mutant suggesting that WT OAS1 binds a specific set of cellular RNAs (Figure 4A). Gene ontology (GO) analysis of OAS1 enriched RNAs indicated enrichment of immunity-related biological pathways in the top-10 enriched GO categories (Figure 4B). Among these, the most significant GO term “cellular response to cytokine” (P-adjusted = 1.82 × 10⁻⁹) includes 26 genes consisting of chemokines/cytokines (CXCL8, CCL2, CCL4, CCL5, CXCL10, IFNL1) and ISGs (IRF1, IFIT2) and others (Figure 4C). IFNB1 and TNF mRNAs, despite showing substantial enrichment in WT OAS1 RIP samples compared to control IgG RIP (Table S1), did not pass the stringent filtering condition (log2 FC >1 and P<0.05).

Figure 4. OAS1 P46 binds multiple cellular mRNAs, including IFNβ.

(A–C) HT1080 OAS1-KO iP46 or iP46 K60E cells were pre-treated with Dox (1.5 μg/ml) for 48 h followed by transfection with LMW pIC (0.75 μg/ml) for 5 h. Cells were crosslinked with 1% formaldehyde and subjected to RIP-Seq analysis using OAS1 or control IgG antibody. (A) Venn diagram showing numbers of significantly enriched cellular RNAs in WT OAS1 and OAS1 K60E mutant RIP conditions (log2 fold-change over IgG RIP >1 and P<0.05). (B) GO analysis of all transcripts that were significantly enriched in WT OAS1 RIP. (C) Heatmap of 167 cellular RNAs enriched in OAS1 RIP compared to IgG RIP. Genes from the top-enriched GO term (GO: 0071345) are indicated.

(D) HT1080 OAS1-KO iP46 cells were transfected with LMW pIC, and subjected to RIP assay as above. Various cytokine and GAPDH mRNAs were quantified in input and RIP samples by RT-qPCR. Percent enrichment of specific mRNA was calculated with respect to the input mRNA. Mean and SEM of the fold enrichment values with respect to the control IgG were plotted from three replicates.

(E) Indicated HT1080 cells were subjected to RIP assay as above.

(F) Immortalized BMDM expressing inducible mouse Oas1b were treated with Dox (1.5 μg/ml) for 48 h, followed by pIC transfection for 5 h. Formaldehyde crosslinked cells were subjected to RIP assay with FLAG-M2 and IgG antibodies. Ifnβ and Tnf mRNAs along with Gapdh mRNA were quantified, and the percent enrichment of specific mRNAs were plotted as in panel D.

We further validated the binding of specific RNAs using independent RIP experiments followed by RT-qPCR detection of individual mRNA. As shown in Figure 4D, following pIC stimulation several mRNAs, including IFNB1, TNF, IFNL1 and CXCL8, but not GAPDH were substantially enriched in samples with P46 expression (+Dox) compared to the control IgG samples indicating specific binding of these mRNAs to OAS1. Given the importance of early IFNβ (encoded by IFNB1) induction in antiviral protection, we focused on IFNβ and used P46 WT, K60E, and C397S mutant expressing cells to show that the binding of IFNβ mRNA depends on the correct cellular localization and RNA-binding property of OAS1 (Figure 4E). To investigate similar mRNA binding activity of mouse Oas1b, we first used our Oas1b expressing HT1080 cells and confirmed the binding of Oas1b to human IFNβ mRNA (Figure S4A). Next, we used immortalized WT BMDM that inducibly expressed Oas1b (BMDM iOas1b, Figure S4B) to carry out RIP experiments after pIC stimulation. Mouse Ifnb1 and Tnf, but not Gapdh, mRNAs were substantially enriched in Oas1b expressing samples compared to IgG controls (Figure 4F). Collectively, these results suggest that human P46 and mouse Oas1b can bind a specific set of cellular mRNAs, representing a number of virus-induced cytokines including IFNβ.

OAS1 stabilizes and increases steady state expression of IFNβ

Having established the binding of IFNβ mRNA to OAS1, we explored the consequences of this binding on IFNβ expression. pIC stimulation of P46 expressing HT1080, HeLa and BJ-Tert cells or Oas1b-KI BMDM showed increased IFNβ mRNA induction compared to controls (Figures 5A–B and Figures S5A–B). The IFNβ mRNA was also enhanced following infection by two IFN-inducing RNA viruses, Sendai virus (SeV) and WNV-KUN (Figures 5C–D, Figures S5B–C). Similar enhancement of mRNA induction was also confirmed for IFNL1 and TNF (Figure S5D). We confirmed the enhanced IFNβ expression at the protein level using similarly stimulated human HT1080 and mouse BMDMs and performing ELISA of the culture supernatant (Figures 5E–F). However, this increased IFNβ expression was abolished in cells expressing P46 C397S or K60E mutants, which are defective in IFNβ mRNA binding (Figures 5G–H), indicating that a major consequence of IFNβ mRNA binding to OAS1 is the enhancement of IFNβ mRNA and protein expression.

Figure 5. OAS1 increases IFNβ mRNA by altering its degradation without affecting transcription.

(A) Dox (1.5 μg/ml, 24 h) treated (+Dox) or untreated (−Dox) HT1080 OAS1-KO iP46 cells were transfected in duplicates with LMW pIC followed by quantitation of IFNβ and GAPDH mRNA at different time points by RT-qPCR. Each sample was normalized to GAPDH mRNA and expressed as relative expression. Two-way ANOVA with Sidak’s multiple comparison test was used to calculate statistical significance.

(B)Immortalized WT and Oas1b-KI BMDMs were transfected with pIC as above followed by Ifnβ mRNA quantitation by RT-qPCR. The statistical significance was calculated as above.

(C–D) IFNβ mRNA induction in HT1080 OAS1-KO iP46 ±Dox (1.5 μg/ml, 24 h) cells after SeV (C) and WNV-KUN (D) infection. The statistical significance was calculated as above.

(E–F) IFNβ in the culture supernatant was measured by ELISA in pIC stimulated cells as indicated from HT1080 OAS1-KO iP46 ±Dox (1.5 μg/ml, 24 h) (E) and immortalized WT and Oas1b-KI BMDMs (F). Two-way ANOVA with Sidak’s multiple comparison test was used to calculate statistical significance. Results are from two independent experiments.

(G–H) IFNβ mRNA in HT1080 OAS1-KO cells expressing inducible OAS1 P46 C397S (G) and OAS1 P46 K60E (H) treated with Dox (1.5 μg/ml, 24 h), followed by pIC stimulation. The statistical significance of Dox-treated cells compared to untreated cells was calculated from duplicate experiments using a two-way ANOVA with Sidak’s multiple comparison test.

(I) HT1080 OAS1/IFNAR-DKO iP46 cells were cotransfected in duplicates with IFNβ-luciferase and Renilla-luciferase plasmids for 8 h. Following transfection, cells were collected and seeded in a white-walled transparent bottom 96-well plate with or without Dox (1.5 μg/ml) for 24 h, followed by pIC transfection (0.75 μg/ml, 6 h) or WNV-KUN infection (MOI of 1, 6 h). Firefly and Renilla luciferase activities was measured by Dual-Glo luciferase assay. Firefly luciferase activity was normalized to Renilla luciferase activity, and promoter activities were expressed as fold change with respect to no pIC or uninfected controls.

(J) IFNβ mRNA was quantified in BJ-Tert OAS1-KO iP46 cells ±Dox (1.5 μg/ml, 24 h) cells after transfection with polydAdT for 6 h. The statistical significance of Dox-treated cells compared to untreated cells was calculated from duplicate experiments using a two-way ANOVA with Sidak’s multiple comparison test.

(K) IFNβ in the supernatants of HT1080 OAS1/IFNAR-DKO iP46 ±Dox (1.5 μg/ml, 24 h) cells were measured by ELISA after poly(dA:dT) transfection. The statistical significance was calculated as above.

(L–M) IFNβ mRNA decay in HT1080 OAS1-KO iP46 (L) and WT and Oas1b-KI BMDMs (M). Cells were transfected with pIC (0.75 μg/ml, 6 h) followed by treatment with 7 μg/ml Actinomycin D. Total RNA was extracted from cells at indicated time points and IFNβ mRNA was quantified by RT-qPCR. Half-lives were determined by fitting the values with a non-linear one-phase decay equation using GraphPad Prism. Representative result from twice repeated experiments is shown.

Next, we investigated the effects of OAS1 on transcriptional induction of IFNβ through RIG-I-like receptor (RLR) signaling pathway. However, OAS1 expression did not affect the RLR signaling as measured by TBK1 and IRF3 phosphorylation (Figure S5E). Furthermore, we used IFNAR-deficient cells (HT1080 OAS1/IFNAR-DKO iP46, Figure S5F) to avoid any autocrine IFN activity on RLR pathway proteins and measured IFNβ-promoter driven luciferase activity. As shown in Figure 5I, irrespective of P46 expression there were no significant differences in the luciferase activity after pIC stimulation or WNV-KUN infection suggesting OAS1 did not affect the IFNβ promoter activity. Stimulation of P46 expressing cells with a different IFN-inducing stimulus dsDNA (polydAdT) also showed enhanced IFNβ expression (Figures 5J–K) further supporting the hypothesis that OAS1 by binding to IFNβ mRNA, did not affect its transcription. This lack of effect on transcription led us to examine the effect of OAS1 on IFNβ mRNA stability in human and mouse cells. Cells were stimulated with pIC (Figures 5L–M) or SeV for 6 h (Figure S5G), treated with Actinomycin D to inhibit transcription, followed by quantitation of IFNβ mRNA by RT-qPCR. The presence of both human P46 or Oas1b almost doubled the half-life of IFNβ mRNA (Figures 5L–M and Figure S5G), which was not seen in the P46 K60E mutant expressing cells (Figure S5H). Taken together, these results suggested that OAS1 binds to IFNβ mRNA resulting in its stabilization and increased expression.

OAS1 binds to the AU-rich element in the 3’ UTR of IFNβ mRNA

Human IFNβ mRNA is encoded by an intronless gene and has a short 3’ UTR (Figure 6A). We hypothesized that OAS1 likely binds to this UTR region to regulate the decay of IFNβ mRNA. To test this hypothesis, we used a luciferase reporter system where the 3’ UTR region of IFNβ mRNA was cloned at the end of luciferase coding region. Transfection of this construct in HT1080 OAS1-KO iP46 cells showed enhanced luciferase activity compared to the vector control, which was increased by OAS1 expression (Figure 6B). We further mapped the IFNβ 3’ UTR to identify the specific region responsible for this enhancement by OAS1 using various truncations (Figure 6A). As shown in Figure 6C, the loss of the AU-rich element (ARE) abrogated the 3’ UTR-mediated enhancement of luciferase activity indicating the necessity of this region. We established that this ARE is sufficient for stabilizing IFNβ mRNA by using only the ARE region in the same assay (Figure 6D). Finally, to examine the involvement of the ARE in direct binding to OAS1, we transfected luciferase plasmids containing either UTR, d-ARE or ARE in HT1080 OAS1-KO iP46 cells and carried out RIP assay. Substantial enrichment of luciferase mRNA was observed only in presence of UTR or ARE with P46 expression establishing the critical role of ARE region in OAS1 binding. Finally, to examine the spatial distribution of the IFNβ mRNA in cells with or without P46 we carried out fluorescence in situ hybridization (FISH) experiments. As shown in Figure 6F and Figure S6, we found substantial partitioning of IFNβ mRNA in the endomembrane regions only in the presence of P46 expression. These results suggest that the ARE in the IFNβ mRNA allows its binding to OAS1 and prolongs its half-life by partitioning the IFNβ mRNA to the endomembrane regions.

Figure 6. OAS1 binds ARE elements in 3’UTR of IFNβ mRNA.

(A) Various luciferase reporter constructs that were used were schematically shown below a graphical representation of the human IFNβ gene and its 3’ UTR sequence.

(B) HT1080 OAS1-KO iP46 cells were co-transfected in duplicates with empty vector (pIS0) or pIS0 containing full-length IFNβ 3’UTR of (UTR) and Renilla luciferase for 8 h, followed by 24 h Dox (1.5 μg/ml) treatment and dual-luciferase activity measurements. Following normalization with Renilla luciferase activities, fold changes in firefly luciferase activities were calculated with respect to −Dox vector control. The statistical significance of Dox-treated cells compared to untreated cells was calculated using a two-way ANOVA with Sidak’s multiple comparison test.

(C–D) HT1080 OAS1-KO iP46 cells were transfected with indicated pIS0 vector or pIS0 containing various fragments of 3’UTR of IFNβ and analyzed for luciferase activity as above.

(E) HT1080 OAS1-KO iP46 cells transfected with indicated pIS0 constructs followed by Dox treatment (1.5 μg/ml, 48 h). Formaldehyde crosslinked cell lysates were subjected to RIP assay as before. GAPDH and firefly luciferase mRNA were quantified in input and RIP samples by RT-qPCR. Percent enrichment of specific mRNA was calculated with respect to the input mRNA. Mean and SEM of the fold enrichment values with respect to the control IgG were plotted from three replicates.

(F) HT1080 OAS1-KO iP46 cells treated with Dox (1.5 μg/ml, 48 h) followed by pIC stimulation (0.75 μg/ml, 6 h) were subjected to simultaneous immunofluorescence and fluorescence in situ hybridization analysis for respective visualization and partitioning of OAS1 and IFNβ mRNA.

IFN signaling is necessary for OAS1-mediated protection from WNV infection

To establish a cause-and-effect relationship between anti-WNV activity of OAS1 and its effect on IFN production, we used HT1080 OAS1/IFNAR-DKO iP46 cells (Figure S5F) and examined WNV-KUN replication after P46 expression (+Dox). As shown in Figure 7A and Figure S7A, the suppression of virus titer by P46 expression was abrogated in the IFNAR-deficient cells establishing the requirement of intact IFNAR signaling in the anti-WNV activity of P46. We also used mouse BMDM from WT and Oas1b-KI mice and used neutralizing antibody to block IFNAR signaling and measured WNV replication. As expected, WNV replication was reduced in Oas1b-KI BMDM compared to the control WT (Figure 7B and Figure S7B). However, blocking IFNAR signaling eliminated this protection and resulted in similar virus titers from the WT and Oas1b-KI BMDM. Using a similar approach, we examined the role of IFNAR signaling in Oas1b-mediated WNV protection in vivo. As shown in Figure 7C, isotype control antibody (IgG) treatment resulted in reduced viral loads in various organs of Oas1b-KI mice compared to the WT mice. However, IFNAR antibody treatment reverted this phenotype, showing similar WNV burdens between two groups; thus established the necessity of IFN response in the anti-WNV activity of Oas1b in vivo. Taken together, these results suggest that IFN enhancement is a common and important mechanism for human P46 and mouse Oas1b to protect from WNV infection. Finally, we also used our HT1080 OAS1/IFNAR-DKO iP46 cells, expressed ACE2 and measured SARS-CoV-2 replication (Figure S7C). Despite the absence of IFNAR signaling, expression of P46 still inhibited SARS-CoV-2 replication confirming an IFN-independent mechanism. Collectively, these results establish that OAS1 inhibits viral replication through two different mechanisms, which are virus-specific (Figure 7D).

Figure 7. OAS1 provides antiviral activity against WNV through IFNβ-IFNAR signaling.

(A) HT1080 WT, OAS1-KO, OAS1/IFNAR1-DKO and OAS1/IFNAR1-DKO iP46 cells (pre-treated with 1.5 μg/ml Dox for 24 h) were inoculated with WNV-KUN at an MOI of 1. Culture supernatants were collected at 48 h post-infection, and infectious particles were quantified by FFA on Vero cells. Mean and SEM values were plotted as described before. The statistical significance of KO cells compared to WT cells was assessed using a one-way ANOVA with Dunnett’s multiple comparison test. Results are from two independent experiments.

(B) WT and Oas1b-KI BMDMs were inoculated with WNV-NY at an MOI of 0.1 for 1 h (virus absorption). After 1 h, the virus-containing media were replaced with complete DMEM containing IFNAR Ab (clone MAR1-5A3, 10 and 20 μg/ml). Culture supernatants were harvested 72 h post-infection and infectious particles were quantified by FFA on Vero cells. Two-way ANOVA with Sidak’s multiple comparison test was used to calculate statistical significance.

(C) WT and Oas1b-KI mice were treated either with isotype control IgG or IFNAR Ab (clone MAR1-5A3, 2 mg/mouse, intraperitoneal injection) a day before inoculation with WNV-NY (10² PFU) as in Figure 1. WNV load in various organs were determined by RT-qPCR at 6 dpi. Mean and SEM of viral loads from tissues were plotted and the statistical significance of Oas1b-KI compared to WT mouse tissue titers was assessed using a Mann-Whitney test.

(D) Schematic representation of the model for two distinct antiviral mechanisms of OAS1 against WNV and SARS-CoV-2.

DISCUSSION

We describe a model for OAS1 antiviral function that does not require its NTase activity. OAS1, through its ability to bind ARE in IFN mRNA, prolongs the half-life and enhances IFN expression, which through autocrine and paracrine signaling provides protection from WNV infection. This mechanism is shared between human P46 and mouse Oas1b, requires endomembrane localization of OAS1, and is independent of its enzyme activity and RNase L. However, human OAS1 through its enzyme activity produces 2–5A, leading to RNase L activation, which causes inhibition of SARS-CoV-2 (Figure 7D). Our data address a long appreciated conundrum in which enzymatically inactive Oas1b and OAS1 genes exhibit antiviral activity^14,55 and provide a common mechanism of NTase activity-independent antiviral function of OAS1 proteins. Using multiple OAS1-deficient human or mouse cells and inducible expression systems, we demonstrate that OAS1 proteins, specifically human P46 and mouse Oas1b, can associate with specific mRNA to enhance their expression. The enhanced expression of one such mRNA, IFNβ subsequently confers protection from WNV infection through canonical IFNAR signaling. However, we also find that this is not the predominant mechanism of SARS-CoV-2 inhibition by human OAS1 (Figure S7C), which might be due to multiple and robust IFN-evasion mechanisms present in SARS-CoV-2⁵⁶. SARS-CoV-2 is inhibited specifically by human OAS1 through its NTase activity and downstream activation of RNase L independent of IFNAR signaling.

Since the discovery of OAS family proteins, their NTase activity has been a primary focus of research . Although molecular evolution studies indicate positive selection of a number of these genes, the significance of multiple paralogs with similar enzyme activity has been unclear^57–59. Similarly, the functional significance of OAS-like protein OASL remained an enigma as it lacks NTase activity^60,61. We showed that human OASL and mouse Oasl2, through their interactions with RNA sensor RIG-I and DNA sensor cGAS, can differentially modulate RNA and DNA virus replications^62–64. Results presented here showed similar virus-specific antiviral mechanism for OAS1. Irrespective of their NTase activity, the RNA binding function of OAS proteins are critical determinants of overall cellular functions of OAS proteins. OAS1 proteins are known to bind various ssRNA including cellular mRNAs and miRNA⁵⁴. However, these properties have been studied in the context of OAS1 enzyme activity. Here, we show how this RNA-binding property of OAS1 influences other aspects of cellular physiology such as mRNA stability. Indeed, regulation of IRF7 mRNA translation by mouse Oasl1 has been reported before⁶⁵. Our study shows another example of a broader modulation of cellular RNA metabolism by OAS-family proteins.

AREs are well-known sequence elements in mRNA that can modulate cellular protein expression^66–68. Although initially characterized to destabilize mRNA, it is now established that rather than the cis-element itself, the binding of specific trans-acting factors determines the function of a particular ARE⁶⁹. Well established examples of such complex ARE-mediated regulation have been characterized for TNF and IFNγ^67,70. In the case of type I IFN, AREs have been known, but the context-dependent effect on expression has been unclear^71,72. Results presented here not only identified OAS1 as an ARE-binding protein but also showed functional consequences of ARE in IFNβ and IFNλ1 3’ UTR. While the OAS1 enzyme activity might promote degradation of the vast majority of cellular mRNA through RNase L, specific mRNAs coding for antiviral proteins might be sequestered and protected from this degradation^40–42. AREs also have been previously implicated in mRNA localization to membrane-associated granules and translation⁴⁴. Our model unifies these two observations and suggests that OAS1-mediated sequestration of specific cytokine mRNAs to endosomal membranes, which enables their continued translation during virus infection. This model also provide a framework of antiviral activity of OAS1 in uninfected cells³⁹, where in the absence of activating viral dsRNA, OAS1 can function to enhance antiviral protein expression.

Our results also clarify previously observed cell-line specific variabilities of OAS1 antiviral function, which is further complicated by the presence of multiple Oas1 genes in mice making it challenging to create human-relevant mouse models for in vivo studies. For example, several commonly used human cell lines such as HEK293T, HeLa, A549 have the rs10774671: A/A genotype, which lacks P46 expression. This resulted in reports of variable contribution of OAS1 in RNase L activation^16,26. Our results establish that human OAS1 P46 isoform, generated from the G haplotype of SNP rs10774671, is the canonical OAS1 that protects against WNV and SARS-CoV-2 and explains the mechanism of previously observed viral resistance phenotype¹⁸. We also demonstrate that an analogous mechanism applies to mouse Oas1b and explains its antiviral activity in vivo.

Beyond antiviral activity, OAS1-binding mRNAs that were identified in our RIPseq assay may modulate other cellular processes. Besides resistance to multiple viruses, the G genotype of SNP rs10774671 is associated with autoimmune disease including Sjorgen’s syndrome⁷³. Given the increased IFN and other inflammatory cytokine expression by P46, it is possible that P46, in addition to its ability to provide host protection, could predispose the host to autoimmunity. Our study, therefore, suggests a possible broader function of the OAS-family proteins in modulating host immunity.

LIMITATIONS OF THE STUDY

We showed OAS1 binds to the ARE of the IFNβ mRNA, and based on our RIPseq analysis potentially other mRNAs. However, given that ARE sequences are degenerate one limitation of our study is to define the preferred ARE sequences that leads to OAS1 binding and increased expression. Furthermore, the RNA binding property is thought to be conserved in all OAS-family proteins irrespective of their NTase activity. It will be interesting to determine whether other OAS-family proteins regulate cellular mRNA and protein expression through similar mechanisms. A comprehensive RIPseq analysis with other OAS-family proteins can define the extent of their involvement in cellular protein expression and consequent disease pathologies.

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to the Lead Contact, Saumendra N. Sarkar (saumen@pitt.edu).

Materials availability.

All requests for resources and reagents should be directed to the Lead Contact author. This includes viruses, proteins, and cells. All reagents will be made available on request after completion of a Materials Transfer Agreement (MTA).

Data and code availability.

All data supporting the findings of this study are available within the paper and are available from the corresponding author upon request. This paper does not include original code or structures.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

All mice experiments were carried out according to the protocol approved by the University of Pittsburgh and Washington University IACUC committees. All mice were in C57BL/6J background and maintained in specific pathogen-free facilities. WT type mice were purchased from the Jackson Laboratory. Oas1b-KI mice were generated by replacing the stop codon with Arg residue (X253R) using CRISPR/Cas9/homology-directed repair by the Transgenic and Gene Targeting Core, University of Pittsburgh. Additionally, a new BspEI restriction site was inserted to allow genotyping of Oas1b-KI mice (Figure S1A). Virus infection experiments were carried out in approximately equal proportions of male and female mice.

Cells and Viruses

HEK293T (Source: Human embryonic kidney, Age/Sex: unknown), Vero (Source: African green monkey kidney, Age/Sex: unknown), Vero-hACE2-TMPRSS2⁷⁴, BJ-Tert (also called BJ-5ta, Source: Human foreskin, Age/Sex: neonatal male) and THP1 (Source: Human acute monocytic leukemia, Age/Sex: 1 year male), HT1080 (Source: Human fibrosarcoma, Age/Sex: 35 year male), Daudi (Source: Human B lymphoblast Burkitt’s lymphoma, Age/Sex: 16 year male) cells were obtained from ATCC. VSV, SeV, WNV-NY, WNV-KUN, SARS-CoV-2 and POWV have been described before^64,74–77.

Plasmids and recombinant DNA

Human OAS1 P42 and P46 cDNA were cloned in pENTR-D-TOPO (Thermo Fisher) vector using appropriate primers from IFN-treated THP1 cells, followed by sequence verification. Mouse Oas1 cDNAs with C-terminal FLAG-MYC tag in pCMV6-Entry were purchased from Origene. For stable inducible expression, all the cDNAs were cloned in pLenti-TRE-DEST-EFpuro-2A-rTA vector, a gift from Dr. Masahiro Shuda⁷⁸ using Gateway Cloning (Thermo Fisher). All the mutants were generated using the Q5^® Site-Directed Mutagenesis Kit (New England Biolabs) as per manufacturer’s instruction.

METHOD DETAILS

Cells and Cell line generation

HEK293T, HEK293FT (Invitrogen), BJ-Tert, HeLa, HT1080, mouse fibroblasts, BMDM and Vero cells were cultured with DMEM (Fisher Scientific) supplemented with 10% FBS, 1% Pen-Strep (Fisher Scientific) and 5 μg/ml Plasmocin (Invivogen). D1-4G2-4-15, Daudi and THP1 cells were cultured with RPMI (Fisher Scientific) supplemented with 10% FBS, 1% Pen-Strep and 5 μg/ml Plasmocin. BMDMs were cultured with DMEM supplemented 10% FBS, 1% Pen-Strep, 5 μg/ml Plasmocin and 20% L929 conditioned media.

CRISPR-Cas9-mediated genome editing to generate OAS1 deficient HeLa, OAS1, RNase L and IFNAR1 deficient HT1080 and OAS1 deficient BJ-Tert cells have been described before^62,79. Briefly, plasmids carrying Cas9-mCherry and gRNA expression cassettes were transiently transfected to the respective cell lines followed by single-cell isolation and colony screening by immunoblotting. To avoid clonal variation reported phenotypes were validated in multiple knockout clones of every single or double deficient cell lines.

Mouse fibroblasts were obtained from respective strains of adult mice (4–6 weeks old) tails by dissociating small pieces of tissue with collagenase (1000 U/mL) and trypsin. Mouse BMDMs were obtained from respective strains of adult mice (4–6 weeks old) by flushing cells from femurs with DMEM followed by culture with DMEM supplemented with 20% L929 conditioning media. Cells were kept for 3–4 days for differentiation into macrophages before use. Differentiated BMDMs were immortalized using CreJ2 virus (retroviral infection) as described previously⁸⁰.

Lentiviral transductions were used to generate all inducible OAS1 expressing cell lines. Lentiviral particles were generated by transfecting 293FT cells with psPAX2 and pMD2.G-VSV-G packaging plasmids (Addgene) along with the transgene in pLenti-TRE-DEST-EFpuro-2A-rTA. Viral particles were harvested 72 h post transfection and used to infect target cells followed by puromycin selection. Target protein expressions were validated by immunoblotting after Doxycycline stimulation. To quantify SARS-CoV-2, various HT1080 cells were transduced with lentivirus, packaged with hACE2 plasmid (Addgene). Then selected with 400 μg/ml Hygromycin to generate cells with stable expression of hACE2.

Genotyping of OAS1 rs10774671

Genomic DNA of several cell lines were genotyped for the single nucleotide polymorphism rs10774671 of OAS1 by restriction length polymorphism assay using AluI, adapted from²⁷, PCR and gel electrophoresis on a 3% agarose gel.

Virus infection and quantitation

For in vivo WNV-NY and POWV infection 6–8 weeks old mice were anesthetized by isoflurane and infected by footpad injection with 100 PFU/mice WNV-NY and 100 FFU/mice POWV. Survival of mice was monitored over a course of 3 weeks. Additionally, viral burden from serum, brain, spinal cord, and spleen was measured by PFU or FFU assay on Vero cells after harvesting tissue or serum 2-, 4- or 7-days post WNV-NY infection. Plaques were visualized after 3 days by staining the monolayers with 1% crystal violet and manually counted⁸¹. For in vitro WNV-KUN infection cells were infected at various MOI with DMEM without FBS and left to adsorb for 1 h, followed by incubation with 2% FBS DMEM for 24–48 h. Viral titers were measured by foci forming assay (FFU) on Vero cells with 4G2 antibody. SARS-CoV-2 infections and quantitation were carried out as described before⁷⁴.

Immunoblotting

Transfected and/or treated and un-treated cells were lysed in lysis buffer (Sodium deoxycholate 0.1%, Triton X-100 1%, HEPES (pH 7.4) 20 mM, NaCl 150 mM, MgCl2 1.5 mM, EGTA 2 mM, DTT 2 mM, NaF 10 mM, β-Glycerophosphate 12.5 mM, Na3VO4 1 mM, PMSF 1 mM, and Protease Inhibitor). Cell lysates were boiled in 1 X SDS–PAGE loading buffer and subjected to SDS-PAGE electrophoresis. Following transfer to Polyvinylidene difluoride membranes, the blots were incubated with target antibody followed by appropriate HRP-conjugated secondary antibody and visualized by ECL detection.

Immunofluorescence and RNA FISH staining

For immunofluorescence (IF) cells were seeded overnight on fibronectin (5 μg/ml) coated coverslips in a 12-well plate. Appropriately treated cells were rinsed with PBS, fixed with 4% paraformaldehyde diluted in PBS for 15 minutes at RT, and rinsed 3 times with PBS followed by permeabilization with 1% Triton X-100 (in PBS) for 10 minutes at RT. PBS rinsed coverslips were blocked for 1 h at RT with 5% FBS (diluted in PBS), and incubated with primary antibodies (1:100–200) in antibody dilution buffer (1% BSA in PBS) for overnight at 4 °C. Cells were rinsed 3 times with PBS and incubated with filtered fluorochrome-conjugated secondary antibodies diluted 1:500 in antibody dilution buffer for 1–2 hours in the dark at RT. Coverslips were rinsed 3 times with PBS and mounted on slides in DAPI containing mounting medium (Vectashield).

Stellaris RNA FISH probes were purchased from Biosearch Technologies and used following manufacturers protocol for sequential IF and FISH. In brief, HT1080 OAS1-KO iP46 cells were seeded on fibronectin (5 μg/ml) coated coverslips for 24 h and treated with ± Doxycycline (1.5 μg/ml) in a 12-well plate. Following transfection with p(I:C)(1 μg/ml, 4 h) cells were rinsed with PBS, fixed in 4% paraformaldehyde diluted in PBS for 15 minutes at RT, and rinsed 3 times with PBS followed by permeabilization with 1% Triton X-100 (in PBS) for 10 minutes at RT. Coverslips were incubated with OAS1 primary antibody (1:100) overnight at 4 °C. Cells were rinsed 3 times with PBS and incubated with filtered fluorochrome-conjugated secondary antibody diluted at 1:500 for 1–2 hours in the dark at RT. Coverslips were rinsed 3 times with PBS, fixed in 4% paraformaldehyde for 10 minutes at RT and washed twice with PBS before incubating with Wash Buffer A (Stellaris RNA FISH Wash Buffer A from Biosearch Technologies) for 2–5 minutes at RT. In each well, 200 μl of Hybridization Buffer containing IFNβ probe was added and hybridized overnight in the dark at 37°C in a humidified chamber. Next day, hybridization mix was removed, coverslips were washed with Wash Buffer A for 30 minutes in the dark at 37°C followed by incubation with Wash Buffer B for 5 minutes at RT. Coverslips were mounted on slides in DAPI containing mounting medium (Vectashield). Images were acquired using a confocal Olympus IX81 equipped with a spinning-disk confocal head using Olympus CellSense software. For colocalization, Pearson’s correlation coefficient R values of ROI profiles were calculated with CellSense.

Cellular fractionation

Sequential detergent extraction method was used to fractionate cytosol and endomembrane⁸². Doxycycline-treated HT1080 OAS1-KO iP42, iP46, and iP46 C397S cells were grown in a 100 mm dish. After washing the monolayer with ice-cold PBS, 1 ml permeabilization buffer (110 mM Potassium acetate, 25 mM K-HEPES, 2.5 mM Magnesium acetate, 1 mM EGTA, 0.025% Digitonin, 1 mM DTT, and protease inhibitors) was added and incubated for 20 minutes with occasional swirling. The soluble cytosolic fractions were collected in a pre-cooled microcentrifuge tube followed by washing with 5 ml wash buffer (110 mM Potassium acetate, 25 mM K-HEPES, 2.5 mM Magnesium acetate, 1 mM EGTA, 0.004% Digitonin, 1 mM DTT). Lysis buffer (01.% Sodium deoxycholate, 1% Triton X-100, 20 mM K-HEPES, 150 mM NaCl, 1.5 mM MgCl2, 2 mM EGTA, 2 mM DTT, 10 mM NaF, 12.5 mM β-Glycerophosphate, 1 mM Na3VO4, 1 mM PMSF, and protease inhibitors) was added to the remaining cells, incubated on ice for 15 minutes to collect the membrane fraction. Both cytosol and membrane fractions were clarified at 10000 rpm for 10 minutes followed by SDS-PAGE electrophoresis and immunoblotting.

Transfection

Lipofectamine 2000 (Invitrogen) was used for all the transfections of pIC, polydAdT, and plasmids in HT1080, BMDM, and BJ-Tert cells as per manufacturer’s instructions.

Quantitative RT-PCR analysis

Total RNA was isolated from appropriately stimulated cells using Trizol (Life Technologies) or Zymo’s Quick-RNA MicroPrep Kit, and reverse transcribed using iScript cDNA Synthesis Kit (Bio-Rad). cDNA synthesized from RNA was subjected to qPCR using either Fast EvaGreen Supermix or SsoAdvanced Universal Probes Supermix in a CFX96 Real Time System (Bio-Rad) according to the manufacturer’s instructions. Custom primers or commercially available TaqMan probes are listed in the Key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-4G2	Cultures from D1-4G2-4-15 hybridoma cells from ATCC	Cat# ATCC HB-112
Rabbit monoclonal anti-OAS1	Cell Signaling Technology	Cat# 14498S
Rabbit monoclonal anti-RNase L	Cell Signaling Technology	Cat# 27281S
Mouse monoclonal anti-β-Actin	Santa Cruz Biotechnology	Cat# Sc-47778
Mouse monoclonal anti-α-Tubulin	Cell Signaling Technology	Cat# 3873S
Mouse monoclonal anti-MYC	OriGene	Cat# TA100010
Rabbit monoclonal anti-IFNAR1	Abcam	Cat# Ab124764
Mouse monoclonal ANTI-FLAG M2	Sigma	Cat# F3165
Rabbit polyclonal anti-FLAG	Cell Signaling Technology	Cat# 2368S
Rabbit monoclonal anti-pTBK1 (Ser172)	Cell Signaling Technology	Cat# 5483
Rabbit monoclonal anti-pIRF3 (Ser396)	Cell Signaling Technology	Cat# 4947
Rabbit monoclonal anti-IRF3	Cell Signaling Technology	Cat# 4302
Anti-rabbit IgG HRP-linked Secondary Ab	Cell Signaling Technology	Cat# 7074
Anti-rabbit IgG HRP-linked Secondary Ab	Rockland	Cat# 611-1302
Anti-mouse IgG HRP-linked Secondary Ab	Rockland	Cat# 610-103-121
Normal Rabbit IgG	Cell Signaling Technology	Cat# 2729
Mouse IgG1 Isotype Control	R&D Systems	Cat# MAB002
Mouse monoclonal anti-Golgin-97 (CDF4)	Invitrogen	Cat# 14-9767-82
Goat polyclonal anti-Calnexin	Invitrogen	Cat# PA5-19169
F(ab’)2-Goat anti-Mouse IgG Secondary Ab, Alexa Fluor 488	Thermo Fisher Scientific	Cat# A-11017
Goat anti-Rabbit IgG Secondary Ab, Alexa Fluor 488	Thermo Fisher Scientific	Cat# A-11034
Goat anti-Mouse IgG Secondary Ab, Alexa Fluor 594	Thermo Fisher Scientific	Cat# A-11005
Donkey anti-Goat IgG Secondary Ab, Alexa Fluor 647	Thermo Fisher Scientific	Cat# A-21447
Mouse monoclonal anti-IFNβ (HDβ-4A7) (for mouse IFNP neutralization)	Leinco Technology	Cat# I-1182
Rat monoclonal anti-IFNβ (ELISA Capture Ab)	Santa Cruz Biotechnology	Cat# sc-57201
Rabbit polyclonal anti-IFNβ (ELISA Detection Ab)	R&D Systems	Cat# 32400-1
Bacteria and Virus Strains
E. coli (DH5α, TOP10, STBL3)	Thermo Fisher Scientific	Cat# 18265017, C404010, C737303
West Nile Virus New York 1999 Strain (WNVNY)	Diamond Laboratory	Daffis et al., 2011
West Nile Virus Kunjin Strain (WNV-KUN)	Diamond Laboratory	Daffis et al., 2011
Powassan virus (POWV)	Diamond Laboratory	VanBlargan et al., 2018
Vesicular stomatitis Indiana Virus (VSV)	Sarkar Laboratory	Ghosh et al., 2019
Sendai virus (SeV, Cantell strain)	Charles River Laboratories	Cat#10100774
Chemicals and Recombinant Proteins
Recombinant Human IFN-alpha 2 (alpha 2b)	R&D Systems	Cat# 11105-1
Lipofectamine 2000	Thermo Fisher Scientific	Cat# 11668027
Poly(I:C) LMW	InvivoGen	Cat# tlrl-picw
Poly(dA:dT)	InvivoGen	Cat# tlrl-patn
Recombinant Mouse IFN-β (ELISA Std.)	BioLegend	Cat# 581309
TMB Substrate Reagent Set	BD Biosciences	Cat# 555214
Clarity and Clarity Max ECL Western Blotting Substrates	Bio-Rad	Cat# 1705061, 1705062
Actinomycin D	Sigma	Cat# A1410
Puromycin	InvivoGen	Cat# ant-pr-1
Hygromycin B Gold	InvivoGen	Cat# ant-hg-1
Triton X-100	Bio-Rad	Cat# 161-0407
Bovine Serum Albumin (BSA)	Sigma	Cat# A4503
32% Paraformaldehyde	Electron Microscopy Sciences	Cat# 15714-S
16% Formaldehyde	Thermo Fisher Scientific	Cat# 28908
VECTASHIELD Antifade Mounting Medium with DAPI	Fisher Scientific	Cat# NC1695563
Sodium chloride	Sigma	Cat# S5886
Glycine	Bio-Rad	Cat# 161-0718
Nonidet P-40	Sigma	Cat# 74385
Formamide	Sigma	Cat# F9037
Digitonin	Fisher Scientific	Cat# 30-041-0250MG
Potassium acetate	Sigma	Cat# P1190
HEPES	Sigma	Cat# H4034
Magnesium acetate tetrahydrate	Sigma	Cat# M5661
0.5 M EGTA	Fisher Scientific	Cat# AAJ60767AD
DL-Dithiothreitol solution (DTT)	Sigma	Cat# 43816
Fibronectin	Fisher Scientific	Cat# CB-40008A
Sodium deoxycholate	Sigma	Cat# 30970
Sodium Dodecyl Sulfate (SDS)	Thermo Fisher Scientific	Cat# J75819.A1
0.5 M EDTA	Quality Biological	Cat# 351-027-721
1 M Tris pH 8.0	VWR	Cat# E199
Roche Protease Inhibitor Cocktail Tablets	Sigma	Cat# 5892970001
SUPERase•In RNase Inhibitor	Thermo Fisher Scientific	Cat# AM2694
SureBeads Protein A Magnetic Beads	Bio-Rad	Cat# 161-4013
SureBeads Protein G Magnetic Beads	Bio-Rad	Cat# 161-4023
Magnesium Chloride	Sigma	Cat# M8266-100G
Proteinase K	Sigma	Cat# P2308
2-Mercaptoethanol	Sigma	Cat# M7154
4X Laemmli Sample Buffer	Bio-Rad	Cat# 161-0747
Critical Commercial Assays
pENTR/D-TOPO Cloning Kit	Thermo Fisher Scientific	Cat# K240020
Gateway LR Clonase II Enzyme mix	Thermo Fisher Scientific	Cat# 11791020
SsoFast EvaGreen Supermix	Bio-Rad	Cat# 172-5203
SsoAdvanced Universal Probes Supermix	Bio-Rad	Cat# 172-5281
iScript cDNA Synthesis Kit	Bio-Rad	Cat# 1708891
Q5 Site-Directed Mutagenesis Kit	New England Biolabs	Cat# E0554S
Q5 Hot Start High-Fidelity 2X Master Mix	New England Biolabs	Cat# M0494S
OneTaq Quick-Load 2X Master Mix with Std Buffer	New England Biolabs	Cat# M0486S
Wizard Genomic DNA Purification Kit	Promega	Cat# A1120
QIAprep Spin Miniprep Kit	Qiagen	Cat# 27106
QIAGEN Plasmid Midi Kit	Qiagen	Cat# 12143
Macherey-Nagel NucleoSpin RNA Kit	Fisher Scientific	Cat# NC9581114
Zymo’s Quick-RNA MicroPrep Kit	Fisher Scientific	Cat# 50-197-7424
Dual-Glo Luciferase Assay System	Promega	Cat# E2920
R&D Systems Human IFNβ DuoSet ELISA Kit	Fisher Scientific	Cat# DY81405
Stellaris® FISH Probes, Human IFNB1 with CAL Fluor® Red 610 Dye	Biosearch Technologies	Cat# VSMF-2953-5
Stellaris RNA FISH Hybridization Buffer	Biosearch Technologies	Cat# SMF-HB1-10
Stellaris RNA FISH Wash Buffer A	Biosearch Technologies	Cat# SMF-WA1-60
Stellaris RNA FISH Wash Buffer B	Biosearch Technologies	Cat# SMF-WB1-20
Experimental Models: Cell Lines
HEK293FT	Invitrogen	Cat# R70007
HeLa	ATCC (Sarkar Laboratory)
HT1080	ATCC (Sarkar Laboratory)
BJ-Tert (BJ-5ta)	ATCC (Sarkar Laboratory)
Primary Mouse fibroblast	Prepared in Sarkar Laboratory
Primary Mouse BMDM	Prepared in Sarkar Laboratory
Experimental Models: Organisms/Strains
Mouse: C57BL/6J	Jackson Laboratory	Cat# 000664
Mouse: C57BL/6 Oas1b-KI	University of Pittsburgh Transgenic Core	This paper
Oligonucleotides
OAS1 rs10774671 genotyping primers: F: TCACAGTGTCTACCGTAAATGCTC R: AGAAAGTCAAGGCTGGAATTTCAT	Integrated DNA Technologies	This paper
IFNB1 TaqMan probe	Integrated DNA Technologies	Unique Assay ID# Hs.PT.58.39481063.g
GAPDH TaqMan probe	Integrated DNA Technologies	Unique Assay ID# qHsaCEP0041396
IFNB1 qPCR primers: F: TGGGAGGATTCTGCATTACC R: CAGCATCTGCTGGTTGAAGA	Integrated DNA Technologies	Zhu et al., 2014
GAPDH qPCR primers: F: TGCACCACCAACTGCTTAGC R: GGCATGGACTGTGGTCATGAG	Integrated DNA Technologies	Ghosh et al., 2019
IFNL1 qPCR primers: F: CGCCTTGGAAGAGTCACTCA R: GAAGCCTCAGGTCCCAATTC	Integrated DNA Technologies	This paper
TNF qPCR primers: F: TGGCCCAGGCAGTCAGA R: GGTTTGCTACAACATGGGCTACA	Integrated DNA Technologies	This paper
CXCL8 (IL-8) qPCR primers: F: GTTTTTGAAGAGGGCTGAGAATTC R: CATGAAGTGTTGAAGTAGATTTGCTTG	Integrated DNA Technologies	Zhu et al., 2010
Firefly Luc qPCR primers: F: TTCGTCCCAGTAAGCTATGT R: GAAGGTTGTGGATCTGGATA	Integrated DNA Technologies	This paper
Ifnβ qPCR primers: F: AGCTCCAAGAAAGGACGAACA R: GCCCTGTAGGTGAGGTTGAT	Integrated DNA Technologies	Ghosh et al., 2019
Gapdh qPCR primers: F: TCACCACCATGGAGAAGGC R: GCTAAGCAGTTGGTGGTGCA	Integrated DNA Technologies	Ghosh et al., 2019
Tnfα qPCR primers: F: CATCTTCTCAAAATTCGAGTGACAA R: TGGGAGTAGACAAGGTACAACCC	Integrated DNA Technologies	This paper
Recombinant DNA
pENTR-D-TOPO	Thermo Fisher	Cat# K240020
Mouse Oas1a in pCMV6-Entry	Origene	Cat# MR220171
Mouse Oas1b in pCMV6-Entry	Origene	Cat# MR220810
Mouse Oas1g in pCMV6-Entry	Origene	Cat# MR205646
Mouse Oas1h in pCMV6-Entry	Origene	Cat# MR212260
pLenti-TRE-DEST-EFpuro-2A-rTA	Dr. Masahiro Shuda	Velasquez et al., 2018
pIS0	Addgene	Plasmid #12178
Software and Algorithms
SnapGene V7.03	SnapGene Software	https://www.snapgene.com
GraphPad Prism V10	GraphPad Software	https://www.graphpad.com

IFNβ quantification in cell supernatant

Enzyme-linked immunosorbent assays (ELISAs) were performed in triplicates to quantify the secreted IFNβ in human and mouse cells. Human IFNβ ELISA was assayed using R&D Systems Human IFNβ DuoSet ELISA Kit (Fisher Scientific). Mouse Ifnβ was assayed by a custom ELISA protocol using the IFN-β capture antibody (Clone 7F-D3, Santa Cruz Biotechnology), polyclonal detection antibody (R&D Systems) and secondary antibody (Cell Signaling Technology).

RNA immunoprecipitation and analysis

Appropriately treated cells were subjected to crosslinking for 10 min at room temperature using 1% formaldehyde, followed by neutralization with 1M glycine 10 min. Crosslinked samples were re-suspended and lysed in buffer B (0.5% NP-40, 0.25% sodium deoxycholate, 0.1% SDS, 10 mM EDTA, 50 mM Tris-HCL pH 8, 1X protease inhibitor, 40 U/ml RNase inhibitor) and incubated for 45 min with rotation at 4°C. Lysates were cleared by centrifugation and diluted to 1 ml with RIP buffer (0.5% NP-40, 1.2 mM EDTA, 16.7 mM Tris-HCL pH 8, 167 mM NaCl, 1X protease inhibitor, 40 U/ml RNase inhibitor). Following pre-clearing with Protein A/G magnetic beads for 60 min at 4°C, each sample was divided in three parts (1:4.5:4.5). The smaller part was used for the analysis of input RNA, while the others were subjected to immunoprecipitation with control IgG or OAS1/FLAG antibody. Antibodies were captured using magnetic beads, washed with RIP Buffer, High Salt Buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCL pH 8 and 500 mM NaCl) and finally with TE buffer (1 mM EDTA and 10 mM Tris-HCL pH 8). RNA was eluted twice with elution buffer (5 mM Tris-HCL pH 8, 150 mM NaCl, 1 mM MgCl₂, 1% SDS, 40 U/ml RNase inhibitor and 200 μg/ml Proteinase K) for 5 min at 55°C. Eluted RNAs were de-crosslinked by incubation for 2 h at 65°C in presence of 200 mM NaCl, followed by RNA purification using NucleoSpin RNA Kit (Macherey-Nagel) according to manufacturer’s instructions. Purified RNAs were either subjected to RNAseq (Novogene) or RT-qPCR analysis Each sample was normalized and expressed as fold enrichment as described before^83,84 and detailed in figure legends. RNA-seq analysis was performed using CLC genomics workbench (Qiagen) for the quality control, genome alignment and gene level quantification. Low-expressed genes (FPKM < 1 in input sample) were first discarded, followed by normalization of transcript abundance in RIP samples (IgG, WT and K60E mutant) to respective inputs. Statistical significance was calculated using Welch’s t-test by comparing the mean of normalized RNA levels in IgG and OAS1 RIP samples. Transcripts with P<0.05 and log2 fold-change over IgG RIP conditions were considered significant.

Reporter assays

HT1080 OAS1/IFNAR-DKO iP46 cells were transfected with 4 μg IFNβ-luciferase and 120 ng null-Renilla luciferase reporter plasmids using lipofectamine 2000. After 8h, cells were trypsinized and seeded into a 96-well plate with or without Dox (1.5 μg/ml). 24h later, cells were stimulated with pIC (0.75 μg/ml, 6h) or infected with WNV-KUN (MOI 1, 24h), and luciferase activities were measured using Dual-Glo Luciferase assay system (Promega). The results were expressed as fold induction of firefly luciferase relative to un-stimulated or uninfected control cells after normalizing to Renilla luciferase.

IFNβ mRNA decay

For the mRNA decay assay, cells were stimulated with pIC (0.75 μg/ml) or infected with SeV (80 HAU/ml) for 6h. After stimulation, cells were washed with fresh media, and Actinomycin D-containing media (7 μg/ml) was added. Samples were harvested at 0, 15, 30, 60, 120, and 180 min time intervals, total RNA was extracted, and IFNβ mRNA was quantified by RT-qPCR.

Half-lives were determined by fitting the values with a non-linear one-phase decay equation using GraphPad Prism.

Quantification and Statistical Analysis

For in vitro and in vivo experiments, statistical significance was calculated as indicated in Figure Legends and represented as * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001 using GraphPad Prism 8.0.

Supplementary Material

2

Table S1: List of transcripts significantly enriched in WT OAS1 RIP relative to IgG control RIP. Related to Figure 4.

HIGHLIGHTS.

1. Human and mouse OAS1 protect against WNV through a non-canonical mechanism.

2. OAS1 binds multiple cellular mRNAs through its unique endomembrane localization.

3. OAS1 binds to ARE region of IFN mRNA, which prolongs half-life and expression.

4. Increased IFN expression by OAS1 leads to WNV inhibition via IFNAR signaling.