Published June 20, 2018, doi:10.4049/jimmunol.1800456 The Journal of Immunology

The Essential Role of Double-Stranded RNA–Dependent Antiviral Signaling in the Degradation of Nonself Single-Stranded RNA in Nonimmune Cells

Sayaka Kimura,* Tomoh Matsumiya,* Yuko Shiba,* Michi Nakanishi,* Ryo Hayakari,* Shogo Kawaguchi,† Hidemi Yoshida,* and Tadaatsu Imaizumi*

The recognition of nonself dsRNA by retinoic acid–inducible gene-I (RIG-I) leads to the engagement of RIG-I–like receptor signaling. In addition, nonself dsRNA triggers a robust latent RNase (RNase L) activation and leads to the degradation of ribosomal structures and cell death. In contrast, nonself ssRNA is known to be recognized by TLR 7/8 in immune cells such as plasmacytoid dendritic cells and B cells, but little is known regarding the involvement of nonself ssRNA in antiviral signaling in nonimmune cells, including epithelial cells. Moreover, the fate of intracellular nonself ssRNA remains unknown. To address this issue, we developed a quantitative RT-PCR–based approach that monitors the kinetics of nonself ssRNA cleavage following the transfection of HeLa human cervical carcinoma cells, using model nonself ssRNA. We discovered that the degradation of ssRNA is independent of RIG-I and type I IFN signaling because ssRNA did not trigger RIG-I–mediated antiviral signaling. We also found that the kinetics of self (59-capped) and nonself ssRNA decay were unaltered, suggesting that nonself ssRNA is not recognized by nonimmune cells. We further demonstrated that the cleavage of nonself ssRNA is accelerated when nonself dsRNA is also introduced into cells. In addition, the cleavage of nonself ssRNA is completely abolished by knockdown of RNase L. Overall, our data demonstrate the important role of dsRNA–RNase L in nonself ssRNA degradation and may partly explain the positive regulation of the antiviral responses in nonimmune cells. The Journal of Immunology, 2018, 201: 000–000.

ammalian antiviral innate immune responses are the type I IFN–mediated antiviral responses [see (7) for an excellent first line of defense against viruses. Following infec- review]. For example, signals elicited by dsRNA-dependent protein M tion, viruses expose their nucleic acids to the cyto- kinase R (PKR)–eukaryotic initiation factor 2a inhibit initiation of plasm of host cells. In nonimmune cells, such as epithelial cells, protein translation and viral replication (8). retinoic acid–inducible gene-I (RIG-I)–like receptors (RLRs) An additional strategy whereby cells achieve protection against serve as cytoplasmic viral RNA sensors (1). RLRs display distinct infection involves the degradation of viral RNA via latent RNase RNA specificities for individual classes of viruses (2). For example, (RNase L), an enzyme that cleaves both cellular and viral RNAs (9). RIG-I senses relatively short dsRNA (3, 4). In contrast, the RLR RNase L2/2 mice have an increased susceptibility to viral infections melanoma differentiation-associated gene-5 (MDA5) senses long (10), which indicates the fundamental role of RNase L in antiviral dsRNAs. Recognition of nonself RNA is followed by RLR asso- defense responses. Homodimerization of RNase L is required for ciation with the adaptor protein MAVS (mitochondrial antiviral- catalytic activity and is regulated by nonself RNA–mediated acti- signaling protein) (5), which leads to the activation of antiviral vation of OAS (11). A recent study identified OAS3 as the isoform signaling events that result in the production of type I IFNs and the involved in RNase L activation (12). In certain settings, RNase L expression of a variety of IFN-stimulated genes (6), some of which can also deregulate ribosomal function by cleaving rRNA in re- (e.g., 2’,59-oligoadenylate synthase [OAS]) have antiviral proper- sponse to viral infection (13) and can result in translational inhibi- ties. Several groups have investigated the mechanisms involved in tory effects that impact both host and viral protein biosynthesis. These RNase L–mediated effects on host cells may account for the ability of this enzyme to induce apoptosis (14, 15). *Department of Vascular Biology, Institute of Brain Science, Hirosaki University Graduate School of Medicine, Hirosaki 036-8562, Japan; and †Department of Gas- In contrast to dsRNA, little is known about how nonself ssRNA is troenterology and Hematology, Hirosaki University Graduate School of Medicine, controlled in the intracellular innate immune system. In this study, Hirosaki 036-8562, Japan we developed a quantitative PCR–based approach to monitor Received for publication March 26, 2018. Accepted for publication June 1, 2018. nonself ssRNA–mediated antiviral signaling and the kinetics of This work was supported by Grants-in-Aid for Scientific Research (KAKENHI) nonself ssRNA decay using in vitro transcription (IVT)–generated (17K10012 to T.M.), the Takeda Science Foundation (to T.M.), and the Karoji Memorial Fund (to T.M.). non-mammalian ssRNA, followed by gel purification. Our results show that nonself RNA does not trigger intracellular innate im- Address correspondence and reprint requests to Dr. Tomoh Matsumiya, Department of Vascular Biology, Institute of Brain Science, Hirosaki University Graduate School mune responses in nonimmune cells. We also found that the of Medicine, 5 Zaifu-cho, Hirosaki 036-8562, Aomori, Japan. E-mail address: cleavage of nonself ssRNA required dsRNA-mediated activation [email protected] of antiviral responses. Our findings may explain how to execute The online version of this article contains supplemental material. nonself ssRNA in nonimmune cells. Abbreviations used in this article: bla, b-lactamase; cDC, conventional DC; CL, cationic lipid; DC, dendritic cell; GM-DC, GM-CSF–induced DC; IVT, in vitro transcription, in vitro transcribed; luc, luciferase; mbla, capped bla; mluc, capped luc; OAS, 2’,59-oligoadenylate synthase; pDC, plasmacytoid DC; qRT-PCR, quanti- Materials and Methods tative RT-PCR; RIG-I, retinoic acid–inducible gene-I; RLR, RIG-I–like receptor; Cell culture RNase L, latent RNase; siRNA, small interfering RNA; UTR, untranslated region. HeLa cells (Japanese Cancer Resources Bank, Ibaraki, Japan) were Copyright Ó 2018 by The American Association of Immunologists, Inc. 0022-1767/18/$35.00 maintained in a 5% CO2 atmosphere at 37˚C in DMEM (Sigma-Aldrich,

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1800456 2 dsRNA CLEAVES NONSELF ssRNA IN NONIMMUNE CELLS

St. Louis, MO) supplemented with 10% FBS (Perbio Science, Switzerland) Immunoblot analyses and antibiotics. For the immunoblot analyses, we washed the cells twice with PBS and then Preparation of dendritic cells lysed them in hypotonic lysis buffer (10 mM Tris [pH 7.4], 100 mM NaCl, 1.5 mM MgCl2, and 0.5% NP-40] containing 0.2% protease inhibitors Preparation of bone marrow dendritic cells (DCs) from female ICR mouse (Sigma-Aldrich). The lysates were cleared by centrifugation at 6000 rpm femora (8 wk old) was performed as Kato et al. (16) reported. Briefly, bone for 15 min at 4˚C. Aliquots of the supernatants (10 mg) were subjected to marrow cells were cultured in RPMI 1640 (Sigma-Aldrich) supplemented electrophoresis on 10% SDS-polyacrylamide gels. The proteins were m with 10% FBS, 100 M 2-ME (Sigma-Aldrich), and 10 ng/ml recombinant transferred to polyvinylidene fluoride membranes (Millipore, Billerica, murine GM-CSF (PeproTech, Rocky Hill, NJ) or 100 ng/ml recombinant MA), which were then blocked for 1 h at room temperature in TBST buffer human Flt3 ligand (PeproTech). After 8 d, the cells were collected and (20 mM Tris [pH 7.4], 150 mM NaCl, 0.1% Tween-20) containing 5% used as DCs (Flt3 ligand–induced DCs [Flt-DCs]) or GM-CSF–induced nonfat dry milk (blocking buffer). The membranes were incubated over- DCs (GM-DCs), respectively. To culture GM-DCs, the medium was night at 4˚C with one of the following primary Abs: mouse anti-RIG-I replaced with fresh medium containing GM-CSF every 2 d. All the animal (Enzo Life Sciences, Miami, FL), mouse anti-RNase L (BD Biosciences, experiments in this study were approved by the Animal Research Com- San Jose, CA), or rabbit anti-b-actin (Sigma-Aldrich). After five washes mittee at Hirosaki University and were conducted according to the with TBST, the membranes were further incubated for 1 h at room tem- Guidelines for Animal Experimentation, Hirosaki University, Japan. perature with HRP-labeled bovine anti-rabbit (Santa Cruz Biotechnology, Generation of in vitro-transcribed RNAs Santa Cruz, CA) or goat anti-mouse IgG Abs (Thermo Fisher Scientific) in blocking buffer. The washes were repeated using TBST, and the immu- The sense and antisense ssRNAs for b-lactamase (bla) and bla dsRNA were noreactive proteins were then visualized using Luminata Crescendo generated as described previously (17). Firefly luciferase (luc) ssRNA was Western HRP Substrate (Millipore). synthesized as a template using a T7 RiboMAX Large Scale RNA Pro- duction kit and contained the linearized full-length luc gene harboring the Blue native PAGE T7 promoter (Promega Life Sciences, Madison, WI). The length of the HeLa cells were harvested in NativePAGE sample buffer (Thermo Fisher ∼ 9 IVT-ssRNAs was 900 (bla) and 1800 bp (luc). The 5 -capped bla (mbla) Scientific) containing 1% n-dodecyl-D-maltoside. The lysate was then and luc RNAs were synthesized using a mMESSAGE mMACHINE kit centrifuged at 20,000 3 g at 4 ˚C for 30 min to collect the soluble fraction. (Thermo Fisher Scientific, Waltham, MA). The samples were loaded onto 4–16% Novex Bis-Tris protein gels Transfection (Thermo Fisher Scientific). NativeMark unstained protein standards (Thermo Fisher Scientific) were used as molecular mass markers. The Transient transfections were performed as previously reported (18). Briefly, details of the blue native (BN) PAGE protocols were those provided in the cells were seeded into 12-well culture plates at a density of 5 3 104 cells instruction manual from Thermo Fisher Scientific. Before protein transfer, per well. After 16 h, the cells reached 30–50% confluence and were used the gels were washed with transfer buffer containing 0.05% SDS for for the studies requiring transfection with small interfering RNA (siRNA). 20 min. The transfers of all protein electrophoresis gels were done onto The experiments involving transfection with nonself ssRNA required a polyvinylidene fluoride membranes. Proteins on the membranes were fixed longer (20 h) incubation after plating to reach a 90% cellular confluency. in 8% acetic acid for 15 min. After membrane blocking in TBST-5% milk To silence RNase L and RIG-I expression, we used Silencer siRNAs for 1 h, immunoblotting was performed as described above. (Thermo Fisher Scientific) against RNase L (s12065) and RIG-I (s223616). The siRNAs were delivered using Lipofectamine RNAiMAX (Thermo Half-life analysis Fisher Scientific). To introduce nonself ssRNAs into HeLa cells, we used The corrected data were fit to single rate equations using KaleidaGraph TransFectin transfection reagent (Bio-Rad, Hercules, CA). (Synergy Software, Reading, PA), and t1/2 were calculated using these Measurement of IFN-a production equations. The reported t1/2 are the averages of at least three independent experiments. DCs (1 3 106 per ml) were transfected with nonself ssRNA (100 ng) or dsRNA (100 ng) using TransFectin as a cationic lipid (CL) for 24 h. Statistics a Culture supernatants were collected and analyzed for mouse IFN- pro- The statistical significance was analyzed using one-way ANOVA followed duction by ELISA (R&D Systems, Minneapolis, MN). by a post hoc Fisher protected least significant difference test. All proba- , Total RNA extraction and quantitative RT-PCR bility values were based on two-tailed tests, and p 0.05 was considered to be significant. To extract the total RNA, we used an illustra RNAspin Mini Kit (GE Healthcare Life Sciences, Buckinghamshire, U.K.). Equal amounts of Results total RNA were then used as templates for the sscDNA synthesis using PrimeScript RT Master Mix (Takara Bio, Kusatsu, Japan), following the ssRNA does not stimulate RLR signaling manufacturer’s recommendations. A CFX96 Real-Time PCR Detection The initial aim of the study was to assess the impact of IVT- System (Bio-Rad) was used for the quantitative assessment of IFN-B and synthesized ssRNA stimulation on cellular responses. We se- GAPDH mRNA levels and bla and luc RNA levels. The sequences of primers used in our studies are as follows: IFN-b lected bla, a nonmammalian gene encoded by bacteria resistant forward, 59-ACTGCCTCAAGGACAGGATG-39, IFN-b reverse, 59-AG- to b-lactam antibiotics, as our model nonself ssRNA. Our pre- CCAGGAGGTTCTCAACAA-39; bla forward, 59-GGCACCTATCTCA- liminary experiments showed activation of RLR signaling in 9 9 9 GCGATCT-3 , bla reverse, 5 -GATAAATCTGGAGCCGGTGA-3 ; luc response to IVT-generated ssRNA (data not shown). Because forward, 59-ACGGATTACCAGGGATTTCAGTC-39; luc reverse, 59-AG- GCTCCTCAGAAACAGCTCTTC-39; GAPDH forward, 59-CCACCCAT- IVT-generated ssRNA may contain dsRNA contaminants (3, 19), GGCAAATTCCATGGCA-39; and GAPDH reverse, 59-TCTAGACGGCA we confirmed the quality of the generated bla ssRNA. Agarose GGTCAGGTCCACC-39. gel electrophoresis of the IVT-generated ssRNA revealed the To carry out DNA amplification, we used SsoAdvanced Universal SYBR unexpected product of bla dsRNAinadditiontothatofbla Green Supermix (Bio-Rad), following the manufacturer’s specifications. ssRNA (Fig. 1A). Therefore, we extracted the single band cor- The amplification conditions were the following: 30 s at 95˚C followed by 45 cycles at 95˚C for 10 s and 58˚C for 30 s. Melting curve DNA am- responding to bla ssRNA from the gel and purified it by gel plification analyses were performed by slowly heating the samples from 65 purification. Transfection of either the IVT-generated bla ssRNA to 95˚C at 0.1˚C increments per second and continuously monitoring the (“crude”) or the purified bla ssRNA (“purified”) into HeLa cells fluorescence. The melting curve and quantitative analyses were performed showed that the purified ssRNA did not activate the expression of using a CFX manager (Bio-Rad). IFN-b mRNA, which reflects the activation of antiviral signal- RNA electrophoresis ing. In contrast, the crude bla ssRNA induced the expression of b Total RNA samples (1 mg) were subjected to electrophoresis on 1% aga- IFN- mRNA in a concentration-dependent manner (Fig. 1B, rose gels containing 3.7% formaldehyde. We used 20 mM MOPS (pH 7) as 1C). We confirmed that neither sense nor antisense bla ssRNA the running buffer and detected rRNA using ultraviolet transillumination. activated the RLR signaling (data not shown). The Journal of Immunology 3

We further characterized the effect of nonself ssRNA and dsRNA on antiviral signaling. To exclude the sequence-specific effect on the antiviral signaling, we next generated luc ssRNA, another non-mammalian gene, from fireflies. Transfection of dsRNA in- duced the degradation of both 28S and 18S rRNA in a dsRNA concentration-dependent manner. In contrast, bla or luc ssRNA induced neither the cleavage of rRNA (Fig. 2A) nor the expression of IFN-b (Fig. 2B–D). GU- or U-rich RNA ssRNA oligonucleotides were derived from HIV-1-stimulated immune cells to secrete IFN-a via the recog- nition of murine TLR7 and human TLR8 (20), which are pattern recognition receptors expressed in DCs. Sequence analysis of bla and luc revealed that there were GU- or U-rich sequences (e.g., 59- UGCUUUUCUGUGACUGGUG-39 in bla;59-GGAGGAGUU- GUGUUUGUGG-39 in luc). In addition, prediction of secondary structures of ssRNA using the RNAfold prediction tool (21) revealed stem-loop structures in bla and luc RNAs (Supplemental Fig. 1). DCs are a type of APC that initiates immune responses, in- cluding antiviral innate immunity (22). DCs are roughly divided into two subsets: conventional DCs (cDCs) and plasmacytoid DCs (pDCs) (23). Kato et al. (16) showed that in cDCs, RIG is essential for antiviral signaling, including production of type I IFNs, whereas TLR is required to induce type I IFNs in pDCs. We ex- amined whether the ssRNA and dsRNA that we generated are detected by DCs. We introduced RNA in the presence of CL in DCs. CL are known to facilitate the uptake of RNA by DC (24). In fact, dsRNA (100 ng) was unable to induce IFN-a in the absence of CL in both subtypes of DCs (Fig. 3). dsRNA-CL induced IFN-a production both in GM-DCs (cDCs) and in Flt3-DCs (pDCs) (Fig. 3), suggesting that dsRNA was sensed by RIG-I in cDCs and by TLR3 in pDCs. Nonself ssRNA-CL did not trigger IFN-a production in cDCs (Fig. 3A), indicating that RIG-I does not sense ssRNA. TLR7 is known to be a critical receptor for murine pDC responses to sense nonself ssRNA (25). Indeed, ssRNA triggered the marked production of IFN-a in pDCs (Fig. 3B), suggesting that the nonself ssRNA used in this study is immunologically nonself ssRNA. These results suggest that nonself ssRNAs are not sensed by the cellular innate immune system and do not trigger antiviral signaling in nonimmune cells. Decay of nonself ssRNA versus mRNA Cellular dsRNA is cleaved by RNase L (26). In contrast, the fate of nonself ssRNA has remained unclear. We therefore investigated the stability of nonself ssRNA. The 59-terminal m7GppGm (cap) structure is known to stabilize mRNA and to recognize self RNA (27). To compare the stability of nonself and self ssRNA, we generated 59 capped-ssRNA (mRNA). Initial studies showed that transfection of both mbla and capped luc (mluc) did not activate antiviral signaling (data not shown). After the transfection of bla, mbla, luc,ormluc, we traced the levels of those RNAs by using quantitative RT-PCR (qRT-PCR) for up to 8 h. The t1/2 of bla and mbla were calculated to be 3.41 and 3.33 h, respectively (Fig. 4A), whereas those of luc and mluc were 4.74 and 4.59 h, respectively (Fig. 4B). Collectively, no significant differences in decay were observed between the nonself ssRNA and its capped form of RNA. We noted that capping of the ssRNA markedly enhanced the rates of protein translation by the rabbit reticulocyte lysate system,

FIGURE 1. IVT-generated ssRNA contains dsRNA, which can activate antiviral signaling. (A) Following IVT, 1 mg of bla ssRNA and 1 mg of bla expressions of IFN-b (B) and bla (C) were measured using qRT-PCR. Data dsRNA were subjected to electrophoresis on a 1% agarose gel. HeLa cells represent the average of three measurements. Mean 6 SD of three ex- were transfected with the indicated ssRNAs or dsRNA for 5 h. The periments is shown. *p , 0.01 versus “0.” 4 dsRNA CLEAVES NONSELF ssRNA IN NONIMMUNE CELLS

FIGURE 2. Nonself ssRNA does not stimulate antiviral signaling. HeLa cells were transfected with the indicated amounts of bla dsRNA, bla ssRNA, or luc ssRNA, and, after culturing the cells for 5 h, total RNA was extracted. (A) One-microgram aliquots were then subjected to electrophoresis on a 1% agarose gel. The arrows indicate typical degradation products generated by the activation of antiviral signaling. The expression levels of IFN-b (B), bla (C), and luc (D) were measured using qRT-PCR. Data represent the average of three measurements. Mean 6 SD of three experiments is shown. *p , 0.01 versus “0.” indicating that this capping was functionally effective (data not dsRNA-dependent antiviral activities, including IFN-b (Fig. 6A) shown). These results further support the evasive ability of nonself and rRNA degradation (Fig. 6B). These results suggest that the ssRNA from intracellular innate immune recognition. elimination of nonself ssRNA requires the activation of the in- tracellular innate immune system by nonself dsRNA in nonim- dsRNA accelerates the decay of nonself ssRNA mune cells. TLR3 and RLR can recognize viral dsRNA, which is generated during viral infection as a replication intermediate of ssRNA Effect of RNase L on nonself ssRNA decay viruses (2, 28). In addition to previous reports, our data clearly RNase L is an essential endoribonuclease in antiviral innate im- revealed that nonself dsRNA is recognized and degraded by the munity. Activation of RNase L induces degradation of rRNA even cellular innate immune system. Next, we investigated whether in nonimmune cells (13). Our previous results showed that dsRNA nonself dsRNA could affect the decay of nonself ssRNA. We first induced the cleavage of rRNA, whereas ssRNA did not (Fig. 2A). performed ssRNA transfection, followed by the transfection of Therefore, we investigated whether ssRNA or dsRNA can activate dsRNA to mimic the viral life cycle (Fig. 5A). To specifically RNase L. RNase L was detected as a 185 kDa complex by gel assess the ssRNA in the cell extract, which contained cellular total filtration analysis (29) and as a 78–80 kDa protein under gel de- RNA and dsRNA, HeLa cells were transfected with luc ssRNA naturing condition (30, 31). Consistent with these early studies, and bla dsRNA. In the absence of bla dsRNA, the t1/2 of luc blue native PAGE analysis showed that in the resting state, the ssRNA was calculated to be 4.85 h (Fig. 5B). In contrast, the t1/2 of molecular mass of RNase L was ∼180 kDa under nondenaturing luc ssRNA was calculated to be 1.19 h when the cells were ad- conditions (Fig. 7, arrow head). ssRNA transfection did not alter ditionally transfected with dsRNA. This result demonstrated an the molecular mass of RNase L under nondenaturing condi- ∼2.6-fold faster decay of ssRNA in the presence of dsRNA. We tions. In contrast, when the cells were transfected with dsRNA, next examined whether ssRNA alters dsRNA-dependent antiviral RNase L with a high molecular mass was detected (Fig. 7, dot- signaling. We did not observe a synergistic effect of ssRNA on arrow). Although whether this high molecular mass indicates the The Journal of Immunology 5

FIGURE 4. Model self and nonself ssRNAs are metabolized with sim- ilar kinetics. HeLa cells were transfected with 10 ng of (A) bla ssRNA or capped-bla ssRNA (mbla)or(B) luc ssRNA or capped-luc ssRNA (mluc). FIGURE 3. Nonself ssRNA triggers IFN-a production in pDCs but not in The stability of each ssRNA was determined by qRT-PCR at different time cDCs. ssRNA or dsRNA was introduced into GM-DCs used as cDCs (A)or points starting 5 h after ssRNA transfection. The median of the results Flt3 ligand–induced DCs (FLT-DCs) as pDCs (B) in the presence or absence of from three independent experiments is shown. CL for 24 h. IFN-a productioninthesupernatantswasmeasuredbyELISA. Data are shown as the mean 6 SD of triplicate samples of a repre- thought to be essential for RNase L–mediated antiviral signaling. sentative from three independent experiments. *p , 0.01 versus untransfected Indeed, hepatitis C virus genotypes with a lower frequency of UU control. and UA dinucleotides may contribute to increased resistance to IFN therapy (33). The frequencies of UU and UA dinucleotides in oligomerization of RNase L or the formation of a complex with bla and luc are shown in Supplemental Fig. 2. Although UU and other proteins will be clarified in our future work, these results UA dinucleotides are in all mammalian RNAs, they are generally suggest that dsRNA but not ssRNA is able to activate RNase L in underrepresented (34). Calculation of the dinucleotide frequency nonimmune cells. showed that the ratio of UA was less than expected (0.81 to bla; RNase L prefers to cleave at UU and UA dinucleotides, pro- 0.72 to luc) and that the ratio of UU was greater than expected ducing RNA fragments with 59-hydroxyl and 2’, 39-cyclin (1.10 to bla; 1.21 to luc), indicating that both bla and luc RNAs phosphate termini (32); therefore, the dinucleotide frequency is are targeted by RNase L. 6 dsRNA CLEAVES NONSELF ssRNA IN NONIMMUNE CELLS

FIGURE 5. dsRNA accelerates exogenous ssRNA degradation. (A) Il- lustration of the experimental approach used in this study. HeLa cells were transfected with luc ssRNA (10 ng/well). Five hours following transfec- tion, we washed the cells twice with PBS, added fresh growth medium, and transfected the cells with bla dsRNA for up to 8 h. (B) The stability of luc ssRNA was determined by qRT-PCR at different time points starting 5 h after bla dsRNA transfection. The median of the results from three inde- pendent experiments is shown.

We next investigated the impact of RNase L, a critical molecule in dsRNA degradation and nonself RNA degradation. We also examined the effect of RIG-I to confirm the role of RLR signaling in ssRNA degradation. Sufficient gene silencing was observed with RNase L– or RIG-I–specific siRNA (Fig. 8A). dsRNA-dependent FIGURE 6. ssRNA does not alter dsRNA-dependent IFN-b expression rRNA degradation was completely abolished by the knockdown or RNase L activation. HeLa cells were transfected ssRNA (100 ng) for of RNase L (Fig. 8B). In contrast, silencing of RIG-I showed no 5 h. The cells were then washed, the medium was replaced, and the cells such effect, suggesting that RLR signaling, including type I IFN– were further transfected with dsRNA (100 ng) for 4 h to extract total RNA. mediated antiviral signaling, does not participate in nonself (A) The level of IFN-b was determined by qRT-PCR. Data are shown as dsRNA degradation. Because we found that dsRNA could have an the mean 6 SD of triplicate samples of a representative from three inde- effect on ssRNA decay (Fig. 5B), we also examined the impact of pendent experiments. *p , 0.01 versus untransfected control. (B)RNA electrophoresis was performed as described in Fig. 2A. RNase L on nonself ssRNA degradation. The t1/2 of luc ssRNA in the absence of bla dsRNA was calculated to be 5.40 h, whereas it was 1.90 h in the presence of dsRNA (Fig. 8C). The knockdown of the IVT-generated ssRNA used was 59-triphosphate-ssRNA. The RNase L had a negligible impact on ssRNA decay (t1/2 = 6.46 h) in latter set of studies reported that the IVT-generated ssRNA may the absence of dsRNA. Additionally, the silencing of RNase L contain unexpected byproducts, such as a hairpin-containing abolished the ssRNA decay that was accelerated by dsRNA (t1/2 = byproduct. We found that IVT generated an unexpected byprod- 6.56 h). These results suggest that RNase L is required for dsRNA- uct, which corresponded to the electrophoretic mobility of dsRNA dependent ssRNA degradation. on an agarose gel. The removal of this byproduct via gel extrac- tion did not influence the activation of antiviral signaling, showing Discussion that ssRNA does not stimulate RLR signaling in nonimmune cells. In the current study, we initially tested the ability of ssRNA to have ssRNA is known to be recognized by TLR7 (20). TLR7 is con- an impact on RLR signaling activity. Previous studies of RIG-I stitutively expressed in human and murine pDCs and B cells (37). showed that ssRNA is a potent RIG-I ligand (35, 36), whereas The basal expression level of TLR7 in nonimmune cells is very other studies showed that chemically synthesized ssRNA is unable low to undetectable. The induction of TLR7 is observed in he- to be recognized by RIG-I (3, 4). In the former group of studies, patocytes that have been infected with the hepatitis C virus (38). The Journal of Immunology 7

FIGURE 7. Nonself dsRNA activates RNase L. HeLa cells were trans- fected with ssRNA (100 ng) or dsRNA (100 ng) for 5 h. After the cells were harvested in NativePAGE sample buffer, the lysates were subjected to SDS-PAGE (upper panel) or blue native PAGE (lower panel). Immunoblot analyses using anti-RNase L and anti-actin were performed as described in Materials and Methods. The arrowhead indicates ∼180 kDa RNase L. The dotted arrow indicates a high-molecular-mass complex containing RNase L.

In this study, ssRNA did not induce type I IFN in HeLa nonim- mune epithelial cells, indicating no functional TLR7-mediated pathway in these cells or in many types of nonimmune cells. Because nonself ssRNAwas not recognized in nonimmune cells, we next explored whether ssRNA was eliminated from such cells. The features required for recognition and cleavage of nonself ssRNA remain undefined; however, some issues merit consider- ation. Key regulators of signaling events are encoded by short-lived host mRNAs (39) that can be stabilized through mechanisms in- volving specific sequences in the untranslated region (UTR) (40). FIGURE 8. dsRNA-dependent RNase L activation is required for ac- For example, the 59 ends of all eukaryotic cellular mRNAs possess celeration of nonself ssRNA degradation. (A) HeLa cells were transfected a cap structure, which consists of an N-7 methylguanosine moiety. with siRNAs against RNase L (RNL), RIG-I (RIG), or control siRNA (c) Capping is essential for the prevention of triphosphate-triggered for 48 h. We harvested the cells and subjected the lysates (5 mg) to im- innate immune activation (41). In the current study, our model munoblot analyses using anti-RNase L, anti-RIG-I, and anti-actin Abs to confirm the knockdown efficiency by each siRNA. (B) Following ssRNA harbored neither a 59 nor 39 UTR, thus ruling out the the siRNA transfection for 48 h, the cells were transfected with dsRNA mediation of regulatory effects by UTR-dependent mechanisms. (100 ng) for 5 h, and total RNA was extracted. RNA electrophoresis was Capping, a feature of self ssRNAs, has been reported to induce performed as described in Fig. 2A. (C) The stability of luc ssRNA was resistance to RNase L–mediated cleavage (42). Using our exper- determined by qRT-PCR at different time points starting 5 h after bla imental system, we compared the kinetics of self and ssRNA dsRNA transfection. The median of the results from three independent decay and found that these species were metabolized with identical experiments is shown. 8 dsRNA CLEAVES NONSELF ssRNA IN NONIMMUNE CELLS kinetics, most likely requiring the same cellular machinery. These Disclosures observations suggest that in the absence of dsRNA, RNase L is not The authors have no financial conflicts of interest. involved in the degradation of self and ssRNAs. Recently, RNase L was shown to be involved in selective mRNA destabilization by microRNA (43). In that study, the authors activated RNase L by References synthetic 2-5A stimulation. Cells were exposed to IFN as a result of 1. Matsumiya, T., and D. M. Stafforini. 2010. Function and regulation of retinoic acid-inducible gene-I. Crit. Rev. Immunol. 30: 489–513. adopting the IFN-induced antiviral state (44); however, the trimetric 2. Kato, H., O. Takeuchi, S. Sato, M. Yoneyama, M. Yamamoto, K. Matsui, and tetrameric forms of 2-5A, which is essential for RNase L ac- S. Uematsu, A. Jung, T. Kawai, K. J. Ishii, et al. 2006. Differential roles of tivation, are produced in IFN-induced virus-infected cells (45). In MDA5 and RIG-I helicases in the recognition of RNA viruses. Nature 441: 101–105. fact, what pathway besides dsRNA can induce endogenous 2-5A to 3. Schlee, M., A. Roth, V. Hornung, C. A. Hagmann, V. Wimmenauer, W. Barchet, activate RNase L has remained unclear. Elucidation of RNase L C. Coch, M. Janke, A. Mihailovic, G. Wardle, et al. 2009. Recognition of 59 activation in the absence of dsRNA will reveal RNase L–dependent triphosphate by RIG-I helicase requires short blunt double-stranded RNA as contained in panhandle of negative-strand virus. Immunity 31: 25–34. selective mRNA decay. 4. Schmidt, A., T. Schwerd, W. Hamm, J. C. Hellmuth, S. Cui, M. Wenzel, In contrast to ssRNA, the fate of dsRNA has been well studied. F. S. Hoffmann, M. C. Michallet, R. Besch, K. P. Hopfner, et al. 2009. 59- triphosphate RNA requires base-paired structures to activate antiviral signaling The constitutive expression of low levels of RNase L in most via RIG-I. Proc. Natl. Acad. Sci. USA 106: 12067–12072. mammalian tissues may represent a cellular readiness to respond to 5. Seth, R. B., L. Sun, C. K. Ea, and Z. J. Chen. 2005. Identification and charac- viral infections (46). The activation of RNase L requires homo- terization of MAVS, a mitochondrial antiviral signaling protein that activates NF-kappaB and IRF 3. Cell 122: 669–682. dimerization, a response that involves type I IFN–mediated OAS 6. Der, S. D., A. Zhou, B. R. Williams, and R. H. Silverman. 1998. Identification of induction, dsRNA-dependent OAS activation, and generation of genes differentially regulated by interferon alpha, beta, or gamma using oligo- 2-5A (47). Although most RNA viruses are single-stranded, viral nucleotide arrays. Proc. Natl. Acad. Sci. USA 95: 15623–15628. 7. Sadler, A. J., and B. R. Williams. 2008. Interferon-inducible antiviral effectors. dsRNA replicative intermediates can activate RNase L (e.g., Nat. Rev. Immunol. 8: 559–568. during encephalomyocarditis viral infections) (13). In our study, 8. Hershey, J. W. 1991. Translational control in mammalian cells. Annu. Rev. Biochem. 60: 717–755. we therefore carried out an initial ssRNA introduction, followed 9. Wreschner, D. H., J. W. McCauley, J. J. Skehel, and I. M. Kerr. 1981. Interferon by dsRNA introduction, into the cells as a model for this viral action–sequence specificity of the ppp(A2’p)nA-dependent ribonuclease. Nature replication process. RNase L has been shown to regulate RLR 289: 414–417. 10. Zhou, A., J. Paranjape, T. L. Brown, H. Nie, S. Naik, B. Dong, A. Chang, signaling in diverse and opposite manners depending on cell type. B. Trapp, R. Fairchild, C. Colmenares, and R. H. Silverman. 1997. Interferon For example, the ability of RNase L to catalyze the cleavage of action and apoptosis are defective in mice devoid of 29,59-oligoadenylate- self RNAs leads to the generation of small degradation products dependent RNase L. EMBO J. 16: 6355–6363. 11. Dong, B., and R. H. Silverman. 1995. 2-5A-dependent RNase molecules di- that can have diverse consequences depending on the cellular merize during activation by 2-5A. J. Biol. Chem. 270: 4133–4137. context (48, 49). Aside from rRNA, little is known regarding the 12. Li, Y., S. Banerjee, Y. Wang, S. A. Goldstein, B. Dong, C. Gaughan, R. H. Silverman, and S. R. Weiss. 2016. Activation of RNase L is dependent on identities of natural RNase L substrates (47). RNase L, activated OAS3 expression during infection with diverse human viruses. Proc. Natl. Acad. in response to viral infection, catalyzes rRNA cleavage and leads Sci. USA 113: 2241–2246. to cellular stress responses that induce apoptotic death (50), the 13. Silverman, R. H., J. J. Skehel, T. C. James, D. H. Wreschner, and I. M. Kerr. 1983. rRNA cleavage as an index of ppp(A2’p)nA activity in interferon-treated ultimate antiviral consequence of infection (44). However, addi- encephalomyocarditis virus-infected cells. J. Virol. 46: 1051–1055. tional RNase L–mediated responses have been described. A lack 14. Castelli, J. C., B. A. Hassel, K. A. Wood, X. L. Li, K. Amemiya, M. C. Dalakas, of RNase L activity increases resistance to staurosporine-induced P. F. Torrence, and R. J. Youle. 1997. A study of the interferon antiviral mechanism: apoptosis activation by the 2-5A system. J. Exp. Med. 186: 967– apoptosis (51), suggesting that RNase L is a central component of 972. viral infection–independent stress responses. Moreover, RNase L 15. Li, G., Y. Xiang, K. Sabapathy, and R. H. Silverman. 2004. An apoptotic sig- naling pathway in the interferon antiviral response mediated by RNase L and c- has been shown to protect mice from viral-induced encephalo- Jun NH2-terminal kinase. J. Biol. Chem. 279: 1123–1131. myelitis (52), and this effect was independent of IFN-a/b sig- 16. Kato, H., S. Sato, M. Yoneyama, M. Yamamoto, S. Uematsu, K. Matsui, naling and was not associated with overall decreases in viral load. T. Tsujimura, K. Takeda, T. Fujita, O. Takeuchi, and S. Akira. 2005. Cell type- specific involvement of RIG-I in antiviral response. Immunity 23: 19–28. Instead, RNase L restricted viral replication in region- and cell 17. Xing, F., T. Matsumiya, K. Onomoto, R. Hayakari, T. Imaizumi, H. Yoshida, type–specific manners in gray matter, and it limited neuro- M. Yoneyama, T. Fujita, and K. Satoh. 2012. Foreign RNA induces the degra- degeneration and demyelination. These examples emphasize that, dation of mitochondrial antiviral signaling protein (MAVS): the role of intra- cellular antiviral factors. PLoS One 7: e45136. in addition to its effects on viral eradication and RLR-signaling, 18. Matsumiya, T., F. Xing, M. Ebina, R. Hayakari, T. Imaizumi, H. Yoshida, RNase L plays additional, diverse roles in antiviral defense, cell H. Kikuchi, M. K. Topham, K. Satoh, and D. M. Stafforini. 2013. Novel role for molecular transporter importin 9 in posttranscriptional regulation of IFN-ε ex- survival, and apoptosis, and is induced by diverse stimuli. In fact, pression. J. Immunol. 191: 1907–1915. we found that dsRNA-dependent RNase L activation is essential 19. Mellits, K. H., T. Pe’ery, L. Manche, H. D. Robertson, and M. B. Mathews. 1990. for nonself ssRNA cleavage. We were unable to show whether Removal of double-stranded contaminants from RNA transcripts: synthesis of adenovirus VA RNAI from a T7 vector. Nucleic Acids Res. 18: 5401–5406. RNase L is directly involved in nonself ssRNA degradation. In this 20. Heil, F., H. Hemmi, H. Hochrein, F. Ampenberger, C. Kirschning, regard, a previously uncharacterized cellular RNase(s) may be S. Akira, G. Lipford, H. Wagner, and S. Bauer. 2004. Species-specific involved in the cleavage of nonself ssRNA. Future studies will recognition of single-stranded RNA via toll-like receptor 7 and 8. Science 303: 1526–1529. elucidate the mechanisms involved in these metabolic events. 21. Gruber, A. R., R. Lorenz, S. H. Bernhart, R. Neubo¨ck, and I. L. Hofacker. 2008. In summary, we established the role of dsRNA-mediated anti- The Vienna RNA websuite. Nucleic Acids Res. 36: W70–W74. 22. Steinman, R. M. 1991. The dendritic cell system and its role in immunogenicity. viral response in the elimination of nonself ssRNA. Importantly, Annu. Rev. Immunol. 9: 271–296. nonself ssRNA is not recognized by intracellular RNA sensors in 23. Villadangos, J. A., and L. Young. 2008. Antigen-presentation properties of nonimmune cells; therefore, it does not trigger antiviral signaling. plasmacytoid dendritic cells. Immunity 29: 352–361. 24. Boczkowski, D., S. K. Nair, D. Snyder, and E. Gilboa. 1996. Dendritic cells Furthermore, the kinetics of self (capped) and nonself ssRNA decay pulsed with RNA are potent antigen-presenting cells in vitro and in vivo. J. Exp. are unaltered. Nonself dsRNA is widely known to trigger robust Med. 184: 465–472. RNase L activation and antiviral signaling. Our data highlight an 25. Diebold, S. S., T. Kaisho, H. Hemmi, S. Akira, and C. Reis e Sousa. 2004. Innate antiviral responses by means of TLR7-mediated recognition of single-stranded important role for dsRNA and subsequent RNase L activation in the RNA. Science 303: 1529–1531. degradation of nonself ssRNA. This system of nonself ssRNA 26. Chakrabarti, A., B. K. Jha, and R. H. Silverman. 2011. New insights into the role of RNase L in innate immunity. J. Interferon Cytokine Res. 31: 49–57. cleavage may serve to protect the host from infection of ssRNA 27. Furuichi, Y., A. LaFiandra, and A. J. Shatkin. 1977. 59-Terminal structure and viruses in nonimmune cells. mRNA stability. Nature 266: 235–239. The Journal of Immunology 9

28. Akira, S., S. Uematsu, and O. Takeuchi. 2006. Pathogen recognition and innate and IFN-gamma in vascular endothelial cells: an essential role for TNF-alpha in immunity. Cell 124: 783–801. post-transcriptional regulation of CX3CL1. J. Immunol. 184: 4205–4214. 29. Slattery, E., N. Ghosh, H. Samanta, and P. Lengyel. 1979. Interferon, double- 41. Dong, H., K. Fink, R. Zu¨st, S. P. Lim, C. F. Qin, and P. Y. Shi. 2014. Flavivirus stranded RNA, and RNA degradation: activation of an endonuclease by (29-59) RNA methylation. J. Gen. Virol. 95: 763–778. An. Proc. Natl. Acad. Sci. USA 76: 4778–4782. 42. Anderson, B. R., H. Muramatsu, B. K. Jha, R. H. Silverman, D. Weissman, and 30. Silverman, R. H., D. D. Jung, N. L. Nolan-Sorden, C. W. Dieffenbach, K. Kariko´. 2011. Nucleoside modifications in RNA limit activation of 29-59- V. P. Kedar, and D. N. SenGupta. 1988. Purification and analysis of murine 2-5A- oligoadenylate synthetase and increase resistance to cleavage by RNase L. dependent RNase. J. Biol. Chem. 263: 7336–7341. Nucleic Acids Res. 39: 9329–9338. 31. Bisbal, C., and R. H. Silverman. 2007. Diverse functions of RNase L and im- 43. Rath, S., J. Donovan, G. Whitney, A. Chitrakar, W. Wang, and A. Korennykh. plications in pathology. Biochimie 89: 789–798. 2015. Human RNase L tunes gene expression by selectively destabilizing the 32. Cooper, D. A., B. K. Jha, R. H. Silverman, J. R. Hesselberth, and D. J. Barton. microRNA-regulated transcriptome. Proc. Natl. Acad. Sci. USA 112: 15916– 2014. Ribonuclease L and metal-ion-independent endoribonuclease cleavage 15921. sites in host and viral RNAs. Nucleic Acids Res. 42: 5202–5216. 44. Silverman, R. H. 2007. Viral encounters with 29,59-oligoadenylate synthetase 33. Washenberger, C. L., J. Q. Han, K. J. Kechris, B. K. Jha, R. H. Silverman, and and RNase L during the interferon antiviral response. J. Virol. 81: 12720–12729. D. J. Barton. 2007. Hepatitis C virus RNA: dinucleotide frequencies and 45. Wreschner, D. H., T. C. James, R. H. Silverman, and I. M. Kerr. 1981. Ribosomal cleavage by RNase L. Virus Res. 130: 85–95. RNA cleavage, nuclease activation and 2-5A(ppp(A2’p)nA) in interferon-treated 34. Al-Saif, M., and K. S. Khabar. 2012. UU/UA dinucleotide frequency reduction in cells. Nucleic Acids Res. 9: 1571–1581. coding regions results in increased mRNA stability and protein expression. Mol. 46. Zhou, A., R. J. Molinaro, K. Malathi, and R. H. Silverman. 2005. Mapping of Ther. 20: 954–959. the human RNASEL promoter and expression in cancer and normal cells. 35. Hornung, V., J. Ellegast, S. Kim, K. Brzo´zka, A. Jung, H. Kato, H. Poeck, J. Interferon Cytokine Res. 25: 595–603. S. Akira, K. K. Conzelmann, M. Schlee, et al. 2006. 59-Triphosphate RNA is the 47. Brennan-Laun, S. E., H. J. Ezelle, X. L. Li, and B. A. Hassel. 2014. RNase-L ligand for RIG-I. Science 314: 994–997. control of cellular mRNAs: roles in biologic functions and mechanisms of 36. Pichlmair, A., O. Schulz, C. P. Tan, J. Rehwinkel, H. Kato, O. Takeuchi, substrate targeting. J. Interferon Cytokine Res. 34: 275–288. S. Akira, M. Way, G. Schiavo, and C. Reis e Sousa. 2009. Activation of MDA5 48. Malathi, K., B. Dong, M. Gale, Jr., and R. H. Silverman. 2007. Small self-RNA requires higher-order RNA structures generated during virus infection. J. Virol. generated by RNase L amplifies antiviral innate immunity. Nature 448: 816–819. 83: 10761–10769. 49. Banerjee, S., A. Chakrabarti, B. K. Jha, S. R. Weiss, and R. H. Silverman. 2014. 37. Petes, C., N. Odoardi, and K. Gee. 2017. The toll for trafficking: toll-like re- Cell-type-specific effects of RNase L on viral induction of beta interferon. MBio ceptor 7 delivery to the endosome. Front. Immunol. 8: 1075. 5: e00856–14. 38. Lee, J., C. C. Wu, K. J. Lee, T. H. Chuang, K. Katakura, Y. T. Liu, M. Chan, 50. Liang, S. L., D. Quirk, and A. Zhou. 2006. RNase L: its biological roles and R. Tawatao, M. Chung, C. Shen, et al. 2006. Activation of anti-hepatitis C virus regulation. IUBMB Life 58: 508–514. responses via Toll-like receptor 7. Proc. Natl. Acad. Sci. USA 103: 1828–1833. 51. Castelli, J. C., B. A. Hassel, A. Maran, J. Paranjape, J. A. Hewitt, X. L. Li, 39. Dixon, D. A., C. D. Kaplan, T. M. McIntyre, G. A. Zimmerman, and Y. T. Hsu, R. H. Silverman, and R. J. Youle. 1998. The role of 29-59 S. M. Prescott. 2000. Post-transcriptional control of cyclooxygenase-2 gene oligoadenylate-activated ribonuclease L in apoptosis. Cell Death Differ. 5: 313– expression. The role of the 39-untranslated region. J. Biol. Chem. 275: 11750– 320. 11757. 52. Ireland, D. D., S. A. Stohlman, D. R. Hinton, P. Kapil, R. H. Silverman, 40. Matsumiya, T., K. Ota, T. Imaizumi, H. Yoshida, H. Kimura, and K. Satoh. 2010. R. A. Atkinson, and C. C. Bergmann. 2009. RNase L mediated protection from Characterization of synergistic induction of CX3CL1/fractalkine by TNF-alpha virus induced demyelination. PLoS Pathog. 5: e1000602.