Double-stranded RNA induces antiviral transcriptional response through the Dicer-2/Ampk/FoxO axis in an arthropod

Abstract

Invertebrates mainly rely on sequence-specific RNA interference (RNAi) to resist viral infections. Increasing studies show that double-stranded RNA (dsRNA) can induce sequence-independent protection and that Dicer-2, the key RNAi player that cleaves long dsRNA into small interfering RNA (siRNA), is necessary for this protection. However, how this protection occurs remains unknown. Herein, we report that it is caused by adenosine triphosphate (ATP)-hydrolysis accompanying the dsRNA-cleavage. Dicer-2 helicase domain is ATP-dependent; therefore, the cleavage consumes ATP. ATP depletion activates adenosine monophosphate-activated protein kinase (Ampk) and induces nuclear localization of Fork head box O (FoxO), a key transcriptional factor for dsRNA-induced genes. siRNAs that do not require processing cannot activate the transcriptional response. This study reveals a unique nonspecific antiviral mechanism other than the specific RNAi in shrimp. This mechanism is functionally similar to, but mechanistically different from, the dsRNA-activated antiviral response in vertebrates and suggests an interesting evolution of innate antiviral immunity.

RNA interference (RNAi) is the major antiviral immunity in invertebrates (1, 2). Studies show that double-stranded RNAs (dsRNAs) also induce sequence-independent antiviral protection in arthropods (3, 4). Available information provides two clues: 1) Long but not short dsRNA induces nonspecific antiviral protection (5, 6); 2) dsRNA induces the expression of RNAi-system components and other effectors (6, 7). This dsRNA-induced transcriptional response is distinct from RNAi-system and fits well with the definition of innate antiviral immunity. However, its detailed mechanism is unknown. We aim to determine how the nonspecific antiviral immunity is mounted and clarify the difference between long and short dsRNAs in this mechanism, using shrimp model.

Results

Arbitrary dsRNAs of various sizes transcribed from green fluorescent protein (GFP) gene were injected into shrimp (Fig. 1 A and B) before white spot syndrome virus (WSSV) infection. Long dsRNA (dsGFP, >80 bp) effectively induced a protection, as revealed by the viral load (Fig. 1C). GFP small interfering RNA (siRNA) (siGFP) did not exert the protective effect. dsRNAs transcribed from red fluorescent protein and human immunoglobulin genes also induced antiviral protection, confirming that the effect was sequence-independent (Fig. 1D). Moreover, this immunity was effective against another virus (Fig. 1E) and therefore is a general protection. Using RNA-seq, we identified the key components of RNAi machinery (Dicer-2, Ago2, and Sid-1) as upregulated genes (dsGFP vs. siGFP) (Fig. 1F), consistent with previous findings in shrimp and Drosophila (5, 6). Moreover, the direct antiviral effectors AlfB1 and Draper, which disrupts intact virions and facilities phagocytosis of virions (Fig. 1 G and H), were also induced. These data suggest that arbitrary dsRNAs, but not siRNAs, induce transcription-based antiviral immunity.

Fig. 1.

DsRNA induces FoxO-dependent antiviral transcriptional response. (A and B) Schematic illustration of experimental procedures and GFP dsRNAs. (C–E) Antiviral protective effect of arbitrary long dsRNAs. Infection was performed 24 h after dsRNA/siRNA stimulation. Viral load was determined another 24 h later. Mean ± SD, n = 4 biological replicates. One-way ANOVA. RFP, red fluorescent protein; hIgG, human immunoglobulin. (F) DEGs between dsGFP and siGFP groups, n = 3 biological replicates. (G) Damage of virions by AlfB1. TEM image of WSSV virions treated with 20 μM of AlfB1 peptide. Scale bar, 200 nm. (H) Affection of hemocytic phagocytosis of virions by Draper-knockdown. Flow-cytometry analysis of hemocytes (10,000) at 2 h after FITC-labeled virions (10⁶) being injected into Draper-pre-silenced shrimp hemocoel. (I–K) Induction of FoxO nuclear-localization by dsGFP. Immunocytochemical analysis of hemocytes at 6 h after stimulation. Scale bar, 10 μm (I). Digitalization of FoxO-nuclei colocalization from eight randomly selected field of view, n > 100 cells (J). Blotting assay of hemocytes nuclear proteins (K). (L and M) Inhibition of dsGFP-induced transcriptional response by FoxO-knockdown. Presilenced shrimp was stimulated with dsGFP/siGFP to evaluate gene transcription 24 h later (L). Infection was performed at 24 h after stimulation to determine viral load another 24 h later (M), n = 4 biological replicates. ^∗∗∗P < 0.001, ns, not significant, Student’s t test. (N) Presence of FoxO-binding sites in the promoters. (O and P) Interaction between FoxO and FoxO-binding sites. Biotin-labeled probes (5 ng) were incubated with recombinant-FoxO or control-tag (2 μg) (O) or dsGFP-stimulated nuclear extracts (5 μg) (P) for EMSA. Excess unlabeled probes and FoxO antibodies were used for competition and supershift assays. (Q) Induction of FoxO–promoter interactions by dsGFP. Immunoprecipitates from dsGFP-stimulated hemocytes were detected using primers for fragments containing FoxO-binding sites. Blank, Protein-A agarose; control, unrelated antibody not cross-reacting with shrimp proteins. Images are representative of three independent replicates.

As FoxO is the key transcription factor for Ago2 and Dicer-2 in Drosophila and for AlfB1 in shrimp (8, 9), we detected the variation in FoxO nuclear levels after dsRNA stimulation. dsGFP significantly up-regulated FoxO nuclear levels, whereas siGFP did not (Fig. 1 I–K). FoxO-knockdown suppressed dsGFP-induced transcription (Fig. 1L) and antiviral protection (Fig. 1M). Electrophoretic mobility shift assay (EMSA) results demonstrated direct binding of both recombinant (Fig. 1O) and native FoxO (Fig. 1P) to predicted FoxO-responsive elements in Dicer-2 and AlfB1 promoters (Fig. 1N). Moreover, chromatin immunoprecipitation (ChIP) assay results illustrated dsGFP-induced FoxO–promoters interaction (Fig. 1Q). Collectively, dsRNA-induced transcriptional response is mediated by FoxO-activation.

As dsRNA, but not siRNA, activated FoxO and antiviral immunity, we attempted to clarify the differences between them. Because dsRNA would be cleaved into siRNA by Dicer-2, we tested whether Dicer-2 is necessary for the function of dsRNA. Dicer-2-knockdown deprived dsGFP-induced protection, whereas Ago2- or Trbp-knockdown did not (Fig. 2A). Moreover, Dicer-2-knockdown suppressed dsGFP-caused FoxO-activation (Fig. 2 B–D) and gene transcription (Fig. 2E). Therefore, the induced antiviral immunity depends on Dicer-2 but not on RNAi. Cleavage may be crucial because this event is unnecessary for siRNA. Because the helicase domain of many invertebrate Dicer-2 requires ATP to process dsRNA into siRNA (10, 11), we assumed that the energy consumption might activate FoxO, the central regulator in energy stress response. dsGFP increased adenosine diphosphate (ADP)/ATP ratio in a dose-dependent manner (Fig. 2F), and Dicer-2-knockdown deprived this increase (Fig. 2G), confirming the necessity of Dicer-2-mediated ATP-hydrolysis for dsGFP-induced FoxO-activation and antiviral immunity.

Fig. 2.

DsRNA activates Dicer-2/Ampk/FoxO axis for transcriptional response. (A) Effect of RNAi-components knockdown on dsGFP-induced antiviral immunity. Presilenced shrimp were stimulated with dsGFP/siGFP. Infection was performed 24 h later to determine viral load another 24 h later, n = 4 biological replicates. (B–D) Inhibition of dsGFP-induced FoxO-nuclear localization by Dicer-2-knockdown. Presilenced shrimp were stimulated with dsGFP/siGFP. Hemocytes were sampled 6 h later to extract nuclear protein for blotting assay (B), and immunocytochemical assay. Scale bar, 10 μm (C). FoxO-nuclei colocalization was digitalized from eight randomly selected field of view, n > 100 cells (D). (E) Inhibition of dsGFP-induced gene transcription by Dicer-2-knockdown. Gene expression was detected at 24 h after stimulation, n = 4 biological replicates. (F) Induction of ADP/ATP ratio by dsGFP. Hemocytes ADP/ATP ratio was determined at 3 h after dsGFP (1, 2, 5 μg/g) or siGFP (5 μg/g) stimulation, n = 4 biological replicates. (G) Inhibition of dsGFP-induced increase in ADP/ATP ratio by Dicer-2-knockdown. Presilenced shrimp were stimulated with dsGFP/siGFP. ADP/ATP ratio was determined 3 h later, n = 4 biological replicates. (H) Induction of hemocytes Ampkα-phosphorylation by dsGFP. Botting assay was performed at 6 h after stimulation. (I) Inhibition of dsGFP-induced Ampkα-phosphorylation by Dicer-2 knockdown. Presilenced shrimp were stimulated with dsGFP/siGFP. Botting assay was performed 6 h later. (J–M) Inhibition of dsGFP-induced FoxO-nuclear localization and gene transcription by Ampkα-knockdown. Experimental procedures are similar to (B–E). (N) Working model. Energy consumption caused by Dicer-2-mediated dsRNA-cleavage activates Ampk/FoxO axis to induce RNAi components and antiviral effectors. Images are representative of three independent replicates. ^∗∗∗P < 0.001, ns, not significant, Student’s t test.

As Ampk is a crucial sensor for ATP-decrease and principal kinase for FoxO-activation (12), we hypothesized that shrimp Ampk is essential for signal transduction. dsGFP induced Ampkα T172-phosphorylation (Fig. 2H), the principal modification for Ampk-activation (12). Moreover, Dicer-2-knockdown suppressed dsGFP-induced Ampkα-phosphorylation, suggesting that Ampk functions downstream of Dicer-2 (Fig. 2I). Interestingly, Ampkα-knockdown deprived dsGFP-induced FoxO-activation (Fig. 2 J–L) and transcription (Fig. 2M). These data demonstrate the decisive role of Ampk for dsGFP-induced antiviral immunity and suggest the dsRNA/Dicer-2/Ampk/FoxO sequential signaling.

Discussion

Our data provide explanation for the sequence-independent general antiviral immunity in shrimp. This mechanism, together with the classical RNAi-dependent specific antiviral immunity, is initiated by Dicer-2-mediated cleavage of dsRNA into siRNA. siRNAs cause the degradation of viral mRNAs via complementary base-pairing, while the energy consumption during the cleavage process activates Ampk/FoxO-mediated transcriptional response (Fig. 2N). This mechanism may be analogous to mammalian innate immunity, in which dsRNA is recognized by RIG-I-like receptors (RLRs) and initiates the interferon response (13). Interestingly, Dicer-2 is phylogenetically closely related to RLRs. Recognition of dsRNA by Drosophila Dicer-2 can elicit antiviral transcription, and mosquito Dicer-2 is able to function as pattern recognition rector in an RNAi-independent manner (14, 15). However, the detailed mechanisms involved are completely different. Interferon response utilizes a protein-based mechanism by which RLRs interact with adaptors to activate downstream signaling (13), whereas the Dicer-2-mediated transcriptional response is achieved by altering the intracellular energy status. This alteration is related to Dicer-2 helicase domain which binds dsRNA and catalyzes ATP hydrolysis (10). As Dicer ATPase activity is retained in Ecdysozoa invertebrates and lost in vertebrates (11), this helicase-dependent transcriptional response may be restricted in some invertebrates and has been replaced by the RLRs-interferon response in vertebrates. Nevertheless, the similarity in function and difference in mechanism between two responses suggest an interesting evolution of innate antiviral immunity.

Materials and Methods

dsGFP/siGFP was injected into kuruma shrimp Marsupenaeus japonicus (approximately 5 g) hemocoel (5 μg/g body-weight) for stimulation. WSSV infection (5 × 10⁵ virions) or gene-transcription evaluation was performed 24 h later. Transcriptional regulatory mechanism was analyzed using EMSA and ChIP assays. Detailed experimental methods are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.