Recombinant human full-length RIG-I (1–925) protein harboring a hexa-histidine tag at the N-terminal was expressed in Escherichia coli pET-22b expression vectors (Novagen, Darmstadt, Germany) that were introduced into BL21 (DE3) cells (Novagen). The cells were grown in a shaking incubator at 37℃ in Luria broth medium supplemented with 50 µg/mL ampicillin. Protein expression was induced by the addition of 0.5 mM isopropyl-d-1-thiogalactopyranoside (IPTG). The cells were then further incubated at 16℃ for 24 h. After harvesting, resuspended cells were disrupted by sonication. All recombinant proteins were purified with Ni fast protein liquid chromatography (FPLC) using standard protocols (GE Healthcare, Piscataway, NJ, USA) (17).

The ssRNA and single-stranded DNA (ssDNA) substrates were chemically synthesized (ST pharm Co. Ltd. Seoul, Korea). The 5'-ppp ssRNAs were synthesized by in vitro transcription using T7 polymerase, as described previously (18). For generating duplexes, RNA or DNA oligonucleotides were mixed in annealing buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 1 mM EDTA), boiled at 85℃ for 3 min, and incubated at 25℃ for 20 min. Where indicated, oligonucleotides were 5'-phosphorylated using T4 polynucleotide kinase (TaKaRa, Tokyo, Japan). Yeast tRNA and polyinosinic:polycytidylic acid (poly I:C) nucleotides were purchased from Sigma (St. Louis, MO, USA).

The ATPase assay reaction mixtures contained buffer (20 mM Tris-HCl, pH 7.5, 8 mM DTT, 5 mM MgCl₂, 10 mM KCl), 4% (w/v) glycerol, and the specified RNA (1 µM) plus a trace amount of [γ-³²P]ATP (2 nM, GE Healthcare) mixed with nonradioactive ATP (200 µM) and 200 nM RIG-I. The reactions were initiated by the addition of enzyme and the reactants were incubated at 37℃. At 3 min intervals, the reaction was stopped by the addition of 1 µL of 4 M formic acid to each aliquot (1.5 µL) of reaction mixture. The quenched reaction aliquot (1 µL) was blotted on polyethyleneimine-cellulose for thin-layer chromatography (Macherey-Nagel, Düren, Germany) and developed in 0.4 M potassium phosphate (pH 3.4). Unreacted ATP and product Pi were separated and quantified using a Cyclone Phosphorimager (Packard Instrument Co., Inc., Meriden, CT, USA).

Various amounts (0, 5, 10, 15, 20, 25, 30, 60, and 80 pmol) of recombinant RIG-I protein were mixed with 20 pmol of RNA oligonucleotides in 25 µL of binding buffer (20 mM Tris-HCl, pH 7.5, 1.5 mM MgCl₂, 1.5 mM DTT) for 20 min at 37℃. The reaction mixture was then applied to a 2% agarose gel (containing Tris/borate/EDTA buffer) that was subsequently stained, after which the RNAs were visualized using a UV transilluminator. To detect the proteins, the gel was stained with Coomassie Brilliant Blue.

The carcinoma human alveolar basal epithelial cell line A549 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). The A549 cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Hyclone Laboratories Inc., Logan, UT, USA) supplemented with 100 U of penicillin, 100 mg of streptomycin, and 10% fetal bovine serum (FBS; Gibco-BRL, Grand Island, NY, USA) at 37℃ in 5% CO₂. Transfection of RNA oligonucleotides was conducted using Lipofectamine™ 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions.

A549 cells were seeded onto a 24-well plate (1.0×10⁵) and incubated for 12 h, after which they were transfected with the reporter constructs, pGL-IFN-β (0.2 µg) and pRL-TK (0.1 µg), as used previously (19). The cells were further incubated for 8 h, after which they were transfected with the designated amount of each RNA oligonucleotide. Yeast tRNA and poly I:C RNA were used as negative and positive controls, respectively. After 16 h, the cells were collected and disrupted with Passive Lysis buffer (Promega, Madison, WI, USA) and the cell extracts were subjected to a luciferase assay using a dual-luciferase assay system (Promega). The levels of luminescence from firefly and Renilla luciferases were measured using a VICTOR X3 Multilabel Plate Reader (PerkinElmer, Waltham, MA, USA), and the ratio of firefly luciferase to Renilla luciferase was calculated.

To determine the binding affinity between RNA substrates and RIG-I protein, we carried out an electrophoretic mobility shift assay (EMSA) for monitoring RNA:protein complex formation. We examined several types of duplex oligonucleotides comprising dsRNA with either blunt or sticky ends and RNA:DNA hybrids (Table I). These partial duplex RNA or RNA:DNA hybrids contained 5'-hydroxyl moieties instead of triphosphate moieties at the 5'-terminal. Fixed quantities of the RNA ligands were mixed with varying amounts of RIG-I protein (0, 5, 10, 15, 20, 25, 30, 60, and 80 pmol). The results on 2% agarose gel showed that the 3'-overhang RNA substrate (47R/25R_I) band was more rapidly shifted at low concentrations of RIG-I than the other types of RNA substrate (5'-overhang and double overhang) (Fig. 1A). This indicates that RIG-I proteins have a high affinity for duplex RNA molecules with blunt-ended 5'-termini. In addition, the RNA:DNA hybrid substrates (47R/25D_I and 47R/25D_II) had a lower binding affinity for RIG-I than the dsRNA substrates (47R/25R_I, 47R/25R_II, and 47R/25R_III). Therefore, a duplex RNA with 5'-blunt ends is the favored ligand for RIG-I binding; the RNA:DNA duplex did not support RIG-I binding, even in the presence of 5'-blunt-ended termini.

We next investigated the effect of the 5'-ppp moiety on the binding efficacy of RIG-I to the 5'-blunt-ended duplex of RNA or RNA:DNA. As shown in Fig. 1B, the presence of 5'-ppp in the duplex RNA or RNA:DNA significantly affected binding affinity to RIG-I; the amount of duplex RNA or RNA:DNA that were bound to RIG-I was apparently enhanced in the presence of 5'-ppp in the duplex nucleic acid ligands. These results suggest that RIG-I prefers duplex RNA with blunt-ended 5'-ppp termini for effective binding.

We hypothesized that the binding affinity of RIG-I for RNA substrates is correlated with ATP hydrolysis activity. Therefore, we next examined the RNA ligand structure preference of RIG-I for ATP hydrolysis stimulation. RIG-I protein was incubated at room temperature for various periods with ATP spiked with [γ-³²P]ATP after the addition of several RNA substrates. The reaction mixture was subjected to polyethyleneimine-cellulose thin-layer chromatography, and the reaction products (ADP and Pi) were resolved. ATP hydrolyzed by RIG-I was quantified by measuring the radioactivity of the products and the signal intensity was plotted (Fig. 2A). The results revealed that the ATP hydrolysis activity of RIG-I was stimulated by dsRNA ligands. The dsRNA ligands (47R/25R_I and 47/25R_II) that contained a 25-bp duplex region and a 22-nt single-stranded tail stimulated ATPase activity equally, regardless of the position of the single-stranded tail. However, the presence of blunt ends in the dsRNA was important for the stimulation of ATPase activity; the dsRNA ligand (47R/25R_III) containing the 25-bp duplex region and 11-nt tails at both ends showed lower ATPase activity than the other dsRNAs.

To determine whether the presence of 5'-ppp in RNA affects the ATPase activity of RIG-I, time-dependent hydrolysis of ATP by RIG-I was monitored in the presence of ssRNAs with either 5'-ppp or 5'-OH moieties at the termini (Fig. 2B). ATP hydrolysis by RIG-I was significantly stimulated by ssRNA with 5'-ppp, whereas ssRNA with a hydroxyl group at the 5' terminus showed slower ATPase activity compared with 5'-ppp ssRNA. Neither ssDNA without 5'-ppp nor shorter ssRNA stimulated ATP hydrolysis activity to the degree observed in ssRNA with 5'-ppp. When the ATP hydrolysis kinetics were compared, ATP hydrolysis activity was stimulated to a greater extent by blunt-ended dsRNA than by 5'-ppp ssRNA. Taken together, these results suggest that RIG-I prefers double-stranded RNA with a blunt end and/or 5'-ppp at the terminus for ATP hydrolysis.

The results of RNA binding and ATP hydrolysis by RIG-I prompted us to hypothesize that interferon (IFN) production via RIG-I would be strongly enhanced with the exogenous RNA ligands preferred by RIG-I. To test the hypothesis, various types of dsRNA and RNA:DNA hybrids were transfected into lung epithelial cells (A549 cells), and IFN expression levels were quantified using the dual luciferase assay (Fig. 3). Previously, we observed that RNAs containing the 5'-ppp moiety significantly increased the expression of type I IFN compared with their RNA counterparts without the 5'-ppp moiety in A549 cells through a RIG-I-mediated innate immune response (7). In addition, it has been recently reported that the presence of 5'-triphosphate in duplex RNA facilitate conformational change of RIG-I to form dimers by exposing CARD domain through interaction with positively charged residues of RIG-I such as lysine or histidine, which can stimulate IFN-β induction pathway in cells. However, absence of 5'-ppp in duplex RNA ligands is unable to cause swiveling of CARD domain, resulting in the absence of IFN-β induction (20). Therefore, we also prepared duplex nucleic acid ligands containing RNA strands with 5'-ppp moieties. We first examined the induction of IFN production with dsRNAs and RNA:DNA hybrids, which are devoid of the 5'-ppp moiety. None of the ligands stimulated IFN promoter reporter gene activity, regardless of the ligand transfection dosage. In contrast, RNAs containing the 5'-ppp moiety significantly increased the expression of type I IFN compared with their RNA counterparts without the 5'-ppp moiety in the A549 cells. Thus, the presence of ppp at the 5' terminus in the nucleic acid ligand is a major determining factor for IFN induction of the innate immune response via the RIG-I pathway. Of the 5'-ppp RNA ligands that stimulated IFN expression, the 47R/25R_I double-stranded RNA showed greater IFN expression than any other type of nucleic acid ligand. Taken together, these results suggest that RIG-I favors a blunt-ended dsRNA with a 5'-ppp moiety at the terminus as a nucleic acid ligand for innate immune response.

In conclusion, the presence of 5'-ppp and a 5' blunt end in the dsRNA substrate enhances the binding affinity of RIG-I to the dsRNA ligand as well as the ATPase activity of RIG-I. These structural motifs provide a basis for IFN expression as an innate immune response via the RIG-I pathway. Furthermore, the presence of 5'-ppp in the dsRNA strongly stimulates IFN induction via RIG-I in lung epithelial cells, in which intracellular RIG-I is abundant. Thus, the present study suggests that IFN induction as an innate immune response via the RIG-I pathway is positively correlated with structural motifs in dsRNA that dictate both high affinity for RIG-I and stimulation of ATPase activity by RIG-I.