Antagonism of RNA silencing in the yellow fever mosquito, Aedes aegypti, by the nsP2 protein of the prototype alphavirus

Significance

While harmful to humans, mosquito-borne viruses usually establish persistent, nonpathogenic infections in mosquitoes as part of their natural maintenance cycles. Research points to RNA interference (RNAi) as the key antiviral immune response underlying mosquitoes’ resistance to viral damage. Although RNA silencing appears robust enough to protect mosquitoes from viral harm, it remains unclear how pathogens avoid clearance to achieve transmission. Our findings demonstrate that the nsP2 protein of Sindbis virus, the prototype alphavirus, functions as a viral suppressor of RNA silencing in Aedes aegypti. This suggests an evolutionary equilibrium with vector RNAi pathways, requiring viral antagonism sufficient to overcome antiviral immunity without causing debilitating or lethal pathology, a potentially tenuous relationship that could be exploitable for disease intervention.

Abstract

Alphaviruses establish persistent infections in mosquito vectors despite robust antiviral RNA interference (RNAi) pathways, suggesting that they employ mechanisms to counteract host immunity. We demonstrate that the nsP2 protein of Sindbis virus (SINV), the prototype alphavirus, functions as a viral suppressor of RNA silencing in Aedes aegypti mosquitoes. Using a SINV mutant (2V) that prevents cleavage at the nsP2–nsP3 junction, we show that proper proteolytic processing to release mature nsP2 is essential for efficient viral replication in mosquitoes with intact RNAi pathways. Replication defects in the 2V mutant were rescued in Dicer-2 (Dcr-2) null mutant mosquitoes or by expressing the mature nsP2 protein. Biochemical assays revealed that recombinant nsP2 directly binds double-stranded RNA and inhibits Dicer-mediated processing into small interfering RNAs (siRNAs). Furthermore, mosquitoes infected with the 2V mutant exhibited higher ratios of virus-derived siRNAs per viral RNA compared to wild-type infections, confirming that mature nsP2 suppresses the RNAi response. Our findings provide compelling evidence that nsP2 antagonizes RNA silencing in mosquito vectors, representing a critical adaptation that facilitates alphavirus replication.

Mosquito-borne alphaviruses (genus Alphavirus, family Togaviridae) comprise a large group of positive-sense RNA viruses that include several significant human pathogens, such as chikungunya virus (CHIKV), Mayaro virus (MAYV), Ross River virus (RRV), Venezuelan equine encephalitis virus (VEEV), and eastern equine encephalitis virus (EEEV) (1). As arthropod-borne viruses (arboviruses), alphaviruses are maintained in transmission cycles involving mosquitoes and vertebrate reservoir hosts, routinely causing infectious diseases in human populations worldwide (2).

Multiple small RNA pathways, collectively referred to as RNA interference (RNAi), have been identified in insects (3). These include the small interfering RNA (siRNA), microRNA (miRNA), and Piwi-interacting RNA (piRNA) pathways. When viruses infect insects, their replication often generates double-stranded RNA (dsRNA) intermediates that are recognized by the exogenous siRNA pathway and cleaved by the RNase III enzyme Dicer 2 (Dcr-2), producing 21-nucleotide siRNA duplexes (4). These duplexes are incorporated into the RNA-induced silencing complex (RISC) (5). Within the RISC, one strand is eliminated while the guide strand directs Argonaute 2 (Ago-2) to target complementary viral RNAs for specific cleavage (6–8), effectively suppressing viral replication and providing antiviral immunity (9–11). Plants employ similar RNAi-based immune pathways against viruses (12). In response, both plant and insect viruses have independently evolved viral suppressors of RNA silencing (VSRs), indicating the existence of an evolutionary arms race between hosts and pathogens (11, 13–17). Although VSRs are diverse and use various mechanisms to interfere with RNAi pathways, a common theme is binding of dsRNA to prevent RNAi components from accessing replicative intermediates and guide strands (18). The presence of VSRs in plant and insect viruses has led to the hypothesis that arboviruses may utilize analogous approaches to counteract RNAi in mosquito vectors (19).

While harmful to humans, mosquito-borne viruses usually establish persistent, nonpathogenic infections in mosquitoes as part of their natural maintenance cycles (20, 21). Research increasingly points to the siRNA pathway as the key antiviral immune response underlying mosquitoes’ resistance to viral damage (22–24). Although this RNA silencing response appears robust enough to protect mosquitoes from viral harm, it remains unclear how these pathogens avoid clearance to achieve transmission. The presence of VSRs in arboviruses would suggest an evolutionary arms race with vector antiviral immune pathways, fundamentally transforming our understanding of disease transmission (19). Successful virus replication conducive to transmission would require antagonism sufficient to overcome antiviral immunity without causing debilitating or lethal pathology—a tenuous relationship that could potentially be exploited for disease intervention, as it could potentially be manipulated to induce deadly infections in vector insects, preventing human disease transmission (19).

Previous studies have identified candidate VSRs encoded in arbovirus genomes. These candidates include various capsid, nucleocapsid, nonstructural proteins, and subgenomic flavivirus RNAs (sfRNAs) (19, 25, 26). However, most research provides only partial or correlative evidence linking these candidates to RNAi pathways. Consequently, for arboviruses causing human disease, the rigorous experimental criteria required to classify a protein or RNA element as a bona fide VSR have not yet been met within an insect vector species. In contrast, numerous plant viral proteins (e.g., HC-Pro and P19) and a handful of insect virus proteins (e.g., B2, DCV-1A, and CRPV-1A) are widely accepted as canonical VSRs (11, 15, 27–29). Identification of the flock house virus (FHV) B2 protein represents a gold standard for demonstrating canonical VSR functions in insect hosts (29). This comprehensive approach documented four key experimental results (1) FHV B2 knockout mutants showed poor replication in RNAi-competent Drosophila cells/flies but improved replication in RNAi-deficient systems (9, 30); (2) replication of a B2-deficient virus was rescued through B2 expression in trans (29); (3) B2 was shown to bind both long dsRNA and siRNAs, physically blocking Dicer activity and siRNA loading into RISC (16, 17); and (4) observation of increased accumulation of virus-derived siRNAs in B2-deficient virus infections, while wild-type FHV (expressing B2) generated fewer siRNAs, effectively limiting the antiviral response (9, 29).

Sindbis virus (SINV), the prototype alphavirus, serves as a model system for studying alphavirus biology and virus–host interactions (31). As part of Alsuviricetes (previously called the Alphavirus-like supergroup), SINV shares genomic architecture with other members of this class, which includes plant viruses like tobacco mosaic virus (TMV) (32, 33). TMV’s p126 protein exhibits VSR activity, which has been linked to multiple domains including a functional methyltransferase and helicase, and shares regions of sequence homology with SINV’s nsP2 protein (34, 35). Our research examined whether SINV’s nsP2 functions as a VSR during insect vector infection. Using a SINV mutant (2V) that prevents nsP2 release (36), we demonstrated that proper proteolytic processing to produce mature nsP2 is critical for effective viral replication in mosquitoes with a functional siRNA pathway. However, the mutant’s replication was successfully rescued in wild-type mosquitoes by expressing either FHV B2 or mature nsP2. Our research revealed that nsP2 suppresses siRNA production by directly binding to long double-stranded RNAs and inhibiting Dicer activity. Furthermore, mosquitoes infected with the 2V mutant showed elevated ratios of virus-derived siRNAs per target viral RNA compared to wild-type virus infections. These results strongly support nsP2’s role as an RNA silencing antagonist in mosquito vectors, representing a key adaptation that enables efficient alphavirus replication in the insect host.

Results

Replication of a SINV 2V Mutant Is Rescued in Dcr-2 Null Mosquitoes.

To investigate whether SINV nsP2 functions in limiting the antiviral response of insect vectors, we engineered the well-characterized 2V mutation into a full-length cDNA SINV (TR339) clone (Fig. 1A) (31). This mutation (G806V) prevents proper cleavage at the nsP2–nsP3 junction, resulting in the accumulation of the P23 polyprotein precursor (SI Appendix, Fig. S1) rather than the mature, individual nsP2 and nsP3, proteins (36). While the 2V mutation preserves the amino acid sequences of both proteins (except for the single G806V substitution), it has previously been associated with an attenuated phenotype in invertebrate cells through mechanisms that remain poorly understood (37). To determine if this attenuated phenotype might be related to the host antiviral immune response, we compared the replication of the 2V mutant (SINV/2V) with that of wild-type SINV in Ae. aegypti mosquitoes. SINV/2V exhibited significantly reduced replication compared to wild-type virus in this mosquito host (Fig. 1B). However, when we infected a previously characterized Dcr-2 null mosquito line (22) with these viruses, the attenuated replication of SINV/2V was effectively rescued to levels comparable with that of wild-type SINV (Fig. 1B). These results demonstrate that proper processing of the P23 polyprotein to release mature nsP2 and nsP3 proteins is necessary for efficient viral replication in Ae. aegypti and that the attenuated phenotype of SINV/2V is directly linked to the mosquito’s siRNA-based antiviral immune response.

Fig. 1.

Genetic rescue of SINV/2V replication in Dcr-2-deficient Ae. aegypti. (A) Schematic illustrating differential processing of SINV and SINV/2V nonstructural polyproteins. The SINV polyprotein undergoes efficient proteolytic cleavage into the individual nonstructural proteins nsP1, nsP2 (green), nsP3 (blue), and nsP4. In contrast, the G806V mutation in SINV/2V disrupts cleavage at the nsP2–nsP3 junction, resulting in accumulation of the uncleaved intermediate polyprotein P23 (gray). (B) Quantification of viral RNA levels in wild-type and Dcr-2 null mutant sibling mosquitoes at 4 d postinfection with either SINV (green) or SINV/2V (gray). Data show mean values ± SEM from three independent biological replicates (n = 5 mosquitoes per replicate). Statistical comparisons were performed using a two-tailed t test.

Expression of B2 Protein Rescues SINV/2V Replication in Ae. aegypti.

To better understand the attenuated phenotype of SINV/2V, we generated a series of double subgenomic Sindbis virus (dsSINV) constructs. Our initial design incorporated the Nodamura virus (NoV) B2 protein together with the 2V mutation (dsSINV/2V-B2; Fig. 2A). To serve as controls, we constructed two additional viruses: dsSINV/2V-B2-fs, a virus with the 2V mutation expressing a nontranslatable B2 containing a frameshift mutation (causing an early premature stop codon), and dsSINV-B2-fs, a virus with normal polyprotein cleavage expressing the identical mutant B2 (Fig. 2A). In Ae. aegypti, dsSINV/2V-B2-fs demonstrated significantly lower replication levels than dsSINV-B2-fs (Fig. 2 B and C). However, the expression of functional B2 rescued the replication of the 2V mutant (Fig. 2 B and C).

Fig. 2.

Expression of a viral suppressor of RNA silencing rescues SINV/2V replication. (A) Schematic representation of SINV and SINV/2V genomes engineered to express either functional NoV B2 protein (orange) or a frame-shifted NoV B2 control (B2-fs, gray) from a duplicated subgenomic promoter (SG-2). (B) Replication kinetics of SINV-B2-fs (green), SINV/2V-B2 (orange), and SINV/2V-B2-fs (gray) viruses in Ae. aegypti. Viral titers were determined by plaque assays performed in triplicate using triturated mosquitoes at the indicated days postinfection. Data represent mean ± SEM from individual mosquitoes (n = 10). Statistical significance was assessed by one-way ANOVA. (C) Quantification of viral RNA levels in Ae. aegypti at indicated days postinfection with SINV-B2-fs (green), SINV/2V-B2 (orange), or SINV/2V-B2-fs (gray). Data represent mean values ± SEM from three independent biological replicates (n = 5 mosquitoes per replicate). Statistical significance was calculated using one-way ANOVA. (D) Distribution by length of small RNAs mapping to the infecting SINV genome in Ae. aegypti. Small RNAs were sequenced from the same mosquito pools used to quantify viral load at 4 d postinfection (C). Green bars represent reads derived from SINV-B2-fs positive strands, orange bars represent sequences from SINV/2V-B2 positive strands, and gray bars represent those from SINV/2V-B2-fs positive strands. Black bars represent sequences derived from negative strands. Results are representative of three independent biological replicates. (E) Comparison of ratios of viral small RNAs to viral mRNAs in mosquitoes infected with SINV-B2-fs (green), SINV/2V-B2 (orange), or SINV/2V-B2-fs (gray). Ratios were calculated from normalized small RNA reads mapping to the viral genome and strand-specific RT-PCR analysis of SINV plus-strands in 1 μg of total RNA at 4 d postinfection. Fold changes are expressed relative to SINV-B2-fs (green), which was assigned a mean value of 1.0. Error bars represent SEM from three biological replicates.

To further investigate this phenomenon, we analyzed small RNAs from infected mosquitoes at 96 h postinfection using small RNA sequencing (Fig. 2D). This analysis revealed two distinct populations of virus-derived small RNAs, consistent with previous observations in alphavirus-infected Aedes mosquitoes and cell lines (19). In dsSINV-B2-fs infections, we detected a prominent 21-nucleotide peak representing equal contributions from positive and negative viral RNA strands (characteristic of siRNA pathway processing), as well as a broader 27 to 29 nucleotide peak indicating piRNA pathway involvement (Fig. 2D). The presence of functional B2 significantly reduced virus-derived small RNAs in the 23 to 30 nucleotide range (Fig. 2D), a previously observed but unexplained effect (38). Although mosquitoes infected with dsSINV/2V-B2-fs accumulated fewer 21-nt siRNAs than those infected with dsSINV-B2-fs, expression of functional B2 (dsSINV/2V-B2) restored siRNAs to levels comparable with those of dsSINV-B2-fs (Fig. 2D), correlating with their similar replication.

As siRNAs direct sequence-specific degradation of viral targets within cells, we next assessed the efficiency of the antiviral response by calculating the ratio of viral siRNAs to positive-sense target viral RNAs. This analysis showed that positive-sense viral RNAs were targeted at comparable levels in mosquitoes infected with dsSINV-B2-fs or dsSINV/2V-B2, explaining their similar levels of replication (Fig. 2 B, C, and E). However, we observed a nearly 70-fold increase in siRNAs targeting positive-sense viral RNAs in mosquitoes infected with dsSINV/2V-B2-fs, suggesting that the presence of the 2V mutation leads to increased degradation of viral RNAs by the immune response (Fig. 2E). Thus, expression of the functional B2 protein appears to rescue 2V mutant replication by suppressing siRNA biogenesis, effectively limiting the antiviral response (Fig. 2 B, C, and E).

Efficient Replication of SINV in Ae. aegypti Requires Release of nsP2.

Next, we examined whether the absence of mature nsP2 affects SINV’s susceptibility to the siRNA pathway by creating additional dsSINV/2V mutants expressing either functional nsP2 (dsSINV/2V-nsP2) or nonfunctional nsP2 with a frameshift mutation (dsSINV/2V-nsP2-fs; Fig. 3A). Consistent with our previous observations using the 2V mutant virus harboring B2-fs, dsSINV/2V-nsP2-fs demonstrated poor replication, which correlated with overall lower levels of siRNA products (Fig. 3 B and C). Notably, functional nsP2 expression restored replication to levels comparable to wild-type virus, accompanied by enhanced siRNA production (Fig. 3 B and C). Analysis of small RNA read pairs mapping to SINV/2V-nsP2-fs and SINV/2V-nsP2 revealed specific enrichment of 21-nt duplexes with 2-nt 3′ overhangs, a canonical signature of Dcr-2 processing (SI Appendix, Fig. S2) (39, 40). This enrichment far exceeded randomized expectations (Z scores = 27,353.25 and 22,896.80, respectively; P < 0.001) for 21-nt reads with 2-nt overhangs, confirming cleavage of double-stranded viral RNA was mediated by Dcr-2.

Fig. 3.

Mature nsP2 suppresses RNA silencing and rescues SINV/2V replication. (A) Schematic showing SINV/2V constructs engineered to express either mature nsP2 protein (green), a frame-shifted nsP2 control (nsP2-fs, gray), a mature nsP3 (blue), or a frame-shifted nsP3 control (nsP3-fs, white) from a second subgenomic promoter (SG-2). (B) Quantification of viral RNA levels in Ae. aegypti 7 d postinfection with SINV/2V-nsP2 (green), SINV/2V-nsP2-fs (gray), SINV/2V-nsP3 or SINV/2V-nsP3-fs (white). Data represent mean values ± SEM from three independent biological replicates (n = 5 mosquitoes per replicate). Statistical significance was calculated using one-way ANOVA. (C) Distribution by length of small RNAs mapping to the infecting SINV genome in Ae. aegypti. Small RNAs were sequenced from the same mosquito pools used to quantify viral load at 7 d postinfection (B). Green bars indicate reads from SINV/2V-nsP2 positive-sense strands, gray bars show reads from SINV/2V-nsP2-fs positive-sense strands, and black bars display negative-sense strand sequences. Data shown are typical of three separate biological replicates. (D) Comparison of ratios of viral small RNAs to viral mRNAs in mosquitoes infected with SINV/2V-nsP2 (green) or SINV/2V-nsP2-fs (gray). Ratios were calculated from normalized small RNA reads mapping to the viral genome and strand-specific RT-PCR analysis of SINV/2V plus-strands in 1 μg of total RNA at 7 d postinfection. Fold changes are expressed relative to SINV/2V-nsP2 (green), which was assigned a mean value of 1.0. Error bars represent SEM from three biological replicates.

Analysis of the ratio between viral siRNAs and positive-sense viral RNAs revealed increased targeting of the 2V-mutant virus when nsP2 expression was absent (Fig. 3D). Conversely, the presence of mature nsP2 decreased this targeting, thus, explaining mechanistically how 2V-mutant replication is restored in the mosquito (Fig. 3 B and D). The rescue of 2V-mutant virus replication was confirmed to be specifically linked to nsP2 expression through comparative experiments. We infected Ae. aegypti with mutants containing either functional nsP3 (dsSINV/2V-nsP3) or nonfunctional nsP3 (dsSINV/2V-nsP3-fs) (Fig. 3A). Results showed that, unlike nsP2, nsP3 expression had no significant effect on 2V-mutant virus replication (Fig. 3B).

SINV nsP2 Binds dsRNAs and Blocks Processing by Dicer.

To better understand the mechanism behind nsP2’s inhibition of the siRNA pathway’s antiviral response, we conducted a comprehensive series of biochemical assays. First, we performed an electrophoretic mobility shift assay (EMSA) to evaluate nsP2’s capacity to bind dsRNAs. Recombinant nsP2 protein expressed in bacteria was incubated with in vitro-synthesized 180-nt dsRNA and analyzed using nondenaturing PAGE with SYBR Green visualization. The results revealed significant electrophoretic mobility shifts when dsRNA was incubated with nsP2 (Fig. 4A; lanes 1 to 7) compared to dsRNA alone (Fig. 4A; lane 10). As a control, we also expressed recombinant nsP3, which like nsP2 is not expressed in its mature form in the SINV 2V mutant. In contrast to nsP2, neither nsP3 protein (Fig. 4A; lane 8) nor His-tag alone (Fig. 4A; lane 9) caused any significant mobility shift of dsRNA. To quantify nsP2’s dsRNA binding affinity, we incubated twofold serial dilutions of the protein with constant amounts of 180-nt dsRNA. By plotting the percentage of bound dsRNA against nsP2 concentration, we generated a binding affinity curve (SI Appendix, Fig. S3). The kinetic analysis demonstrated that approximately 1.0 nmol of nsP2 protein binds to 50% of the 180-nt dsRNA used in our experiments.

Fig. 4.

The SINV nsP2 protein binds dsRNAs and inhibits Dcr-2. (A) EMSA, lane 1 shows nsP2 with dsRNA, lanes 2 to 7 show dsRNA with serial twofold dilutions of nsP2, lane 8 shows dsRNA with recombinant nsP3 control, lane 9 shows dsRNA with His-tag control, and lane 10 shows dsRNA alone. (B) EMSA, lanes 1 to 3 show serial twofold dilutions of nsP2 with in vitro generated siRNA duplexes, lane 4 shows siRNAs with His-tag control, and lane 5 shows siRNAs alone. (C) Dicing assays with recombinant Ae. aegypti or Drosophila melanogaster Dcr-2 enzymes, lane 1 shows dsRNA substrate alone, lane 2 shows dsRNA with Dcr-2, lane 3 shows dsRNA with Dcr-2 and nsP2, lanes 4 to 5 show dsRNA with Dcr-2 and serial twofold dilutions of nsP2, lane 6 shows dsRNA with Dcr-2 and recombinant nsP3 control, and lane 7 shows dsRNA with Dcr-2 and His-tagged MBP control.

Given that several well-characterized VSRs function by sequestering siRNA products generated by Dicer (16, 41), either in addition to or as an alternative mechanism to inhibiting long-dsRNA processing, we investigated whether nsP2 exhibits similar binding of short siRNA duplexes. To test this, we performed EMSAs using twofold serial dilutions of nsP2 incubated with a fixed concentration of siRNA duplexes. The siRNAs were produced through in vitro dicing of a 720-nucleotide dsRNA substrate using recombinant Ae. aegypti Dcr-2. The results revealed mobility shifts in the presence of nsP2 (Fig. 4B; lanes 1 to 3) compared to control reactions containing either His-tag alone or siRNAs alone (lanes 4 and 5, respectively), where no appreciable mobility shift was detected. To further evaluate the binding of dsRNA by nsP2, we conducted chase EMSA experiments to determine whether single-stranded RNA (ssRNA) or DNA could displace the protein from dsRNA (SI Appendix, Fig. S4). Preformed nsP2–dsRNA (120-nucleotides) complexes were incubated with competing nucleic acids, either ssRNA (180-nucleotides) or DNA (60-bp). No significant change in electrophoretic mobility was detected when nsP2-bound dsRNA was challenged with either ssRNA (lane 3) or DNA (lane 5), compared to nsP2–dsRNA alone (lane 2). Controls included unbound dsRNA (lane 1), ssRNA (lane 4), and DNA (lane 6). The nsP2-bound dsRNA was only successfully released from the complex after treatment with proteinase K (lane 7), demonstrating a highly stable nsP2–dsRNA complex.

Based on our EMSA findings, we next examined whether nsP2 interferes with Dcr-2-mediated processing of dsRNA into siRNAs. We conducted in vitro dicing assays using recombinantly expressed Ae. aegypti Dcr-2, with additional parallel assays using D. melanogaster Dcr-2 protein for validation. Results revealed that nsP2 confers significant, although partial, protection of dsRNA against processing by all Dicer proteins tested (Fig. 4C; lanes 3 to 5). In contrast, when identical dicing assays were performed with either recombinant nsP3 (Fig. 4C; lane 6) or His-tag (Fig. 4C; lane 7), long dsRNAs were efficiently processed into siRNAs with no detectable protection observed. The results of our study provide compelling evidence that nsP2 forms direct complexes with long dsRNAs, inhibiting their cleavage by Ae. aegypti Dcr-2. In addition, the ability of nsP2 to also bind short dsRNAs suggests siRNAs that are produced will be less effectively used to silence the virus, thereby compromising multiple essential steps in the host’s immune response against viral infection.

Discussion

Viral pathogens transmitted by mosquitoes have coevolved with their vectors, establishing relationships that minimize harmful effects on mosquito viability (19). Only recently have researchers begun to understand how mosquitoes resist the cytopathic effects of viruses, primarily through RNAi-based immunity (22–24). Nevertheless, these viruses continue to circulate in nature, resulting in considerable morbidity and mortality among humans (42). The mechanisms through which arboviruses regulate antiviral small RNA pathways within their vector hosts are not well understood. However, given the well-characterized examples of plant and insect virus proteins that disrupt small RNA pathways (11, 13–17), arboviruses probably encode similar proteins that counteract host defense mechanisms (19).

The nonstructural proteins of SINV are derived from polyproteins, P123 and P1234. The protease activity of nsP2 is essential for cleaving these polyproteins into the individual nonstructural proteins, nsP1, nsP2, nsP3, and nsP4 (31). This sequential processing creates three distinct replicase complexes with unique template specificities that regulate alphavirus genome and mRNA synthesis (SI Appendix, Fig. S1) (31). In addition to its protease function, nsP2 serves as an RNA helicase and nucleoside triphosphatase (31, 43, 44). However, biochemical studies indicate low helicase processivity, suggesting that it is unlikely the protein’s role is to unwind long dsRNAs, such as replicative intermediates (43, 44). The centrally located protease domain in the mature nsP2 is flanked by the amino-terminal RNA helicase domain, and an inactive RNA methyltransferase-like (MT-like) domain of unknown function at the carboxy-terminus (45). While essential for viral RNA synthesis, only a portion of nsP2 associates with the other nonstructural proteins in viral replicative complexes, with a significant fraction existing freely in the cytoplasm and nucleus (46, 47). Interestingly, SINV nsP2 triggers degradation of Rpb1 (RNA polymerase II’s catalytic subunit) in vertebrate cells, globally inhibiting cellular transcription and indirectly interfering with antiviral responses. However, Rpb1 is not degraded during replication in mosquito cells, where the virus establishes a noncytopathic, persistent infection (48).

Rather, our research indicates that mature nsP2 forms direct complexes with long dsRNAs, thereby inhibiting their cleavage by Ae. aegypti Dcr-2 (Fig. 4 A and C). Moreover, nsP2’s capacity to bind short dsRNA duplexes suggests that any siRNAs produced are similarly sequestered (Fig. 4B). This dual binding activity suggests that nsP2 compromises multiple critical steps in the host’s RNAi-mediated antiviral defense, i.e., both the generation of siRNAs from viral dsRNA precursors and the utilization of siRNA effectors for sequence-specific silencing of viral transcripts. The functional significance of these biochemical experiments was confirmed through rescue of the 2V-mutant’s replication-deficient phenotype (Figs. 1–3). The G806V mutation (2V) prevents cleavage at the nsP2–nsP3 junction (49), providing critical insights into how proteolytic polyprotein processing and the release of mature nsP2 affects this function. This mutation provides an elegant experimental system to study nsP2’s role as a VSR, as it specifically prevents the production of mature nsP2 without affecting key functions necessary for replication (37, 50). By comparing wild-type virus with the 2V mutant, differences in RNAi susceptibility can be attributed to VSR activity in the mature nsP2.

When the P23 polyprotein remains unprocessed, VSR activity is undetectable (Figs. 1–3), likely because the critical dsRNA binding domain is inaccessible within the larger fusion protein. Structural analysis indicates that the MT-like domain, rich in positively charged amino acids, binds RNA nonspecifically (45, 51). In the P23 polyprotein structure, nsP3 wraps around nsP2’s MT-like domain (51). Opposite charges between the nsP2 (basic/positively charged) and nsP3 (acidic/negatively charged) interfaces likely create electrostatic attractions that help stabilize the P23 polyprotein complex before cleavage occurs. It has been hypothesized that RNA binding to the MT-like domain may facilitate the separation of nsP2 and nsP3 after cleavage (51). Presumably, structural constraints in the unprocessed P23 polyprotein prevent effective interaction with dsRNA substrates and inhibition of the RNAi machinery. Consequently, host Dicer enzymes can more efficiently convert viral dsRNA replication intermediates into siRNAs, resulting in enhanced targeting of viral RNAs by RISCs (Figs. 1–4). The more robust antiviral response appears to contribute substantially to the weakened phenotype of the 2V mutant in insect hosts, establishing a clear connection between viral protein processing and immune evasion.

Overall, our results suggest that the SINV nsP2 protein functions as a VSR in Ae. aegypti mosquitoes, representing a key adaptation that enables alphavirus infection. Identification of nsP2 as a VSR in alphaviruses also provides further evidence for an evolutionary arms race between arboviruses and their mosquito vectors’ antiviral immune systems. The nsP2 protein’s RNA binding affinity appears precisely calibrated to overcome mosquito immunity without triggering lethal pathology. The presumably tenuous nature of this relationship presents a promising target for disease intervention strategies. By understanding how nsP2 suppresses RNAi pathways in Ae. aegypti, researchers could potentially develop approaches to disrupt this balance, either enhancing vector immunity to eliminate the virus or increasing viral antagonism to cause lethal infections in mosquitoes. Such interventions could prevent these insects from transmitting alphavirus pathogens to humans, effectively breaking the transmission cycle of diseases like chikungunya, Mayaro, Ross River, eastern equine encephalitis, and Venezuelan equine encephalitis.

Materials and Methods

Molecular Cloning and Vector Construction.

All viral constructs were derived from the TR339 cDNA clone (52, 53). The 2V mutation was introduced into the TR339 backbone with the QuikChange Lightning Multi Site-Directed Mutagenesis Kit from Agilent Technologies (Santa Clara, CA). Expression systems were constructed by amplifying the duplicated 26S promoter and multiple cloning site (MCS) from pTE/3’2 J and incorporating into pTR339 using the BsiWI and XhoI restriction sites (54, 55). Sequences encoding the Nodamura virus B2, nsP2, or nsP3 proteins, and nontranslatable control elements were inserted downstream of the duplicated 26S promoter by utilizing the AscI and PacI sites within the MCS. For protein expression studies, we inserted 6xHis-tagged nsP2 or nsP3 cistrons within the NotI and BamHI sites of pMAL-c5x. The sequence of all constructs was verified using an Applied Biosystems 3730xl DNA Analyzer at Genewiz (South Plainfield, NJ).

Viruses and Cells.

Infectious RNA transcripts were generated from plasmid templates linearized with XhoI in a standard in vitro transcription reaction (56, 57). The transcripts were electrotransfected into BHK-21 cells using protocols that we have published previously (55). Culture supernatant was harvested, clarified by centrifugation, and stored at −80 °C. Virus titers were quantified in triplicate using plaque assays on Vero cells. All cells used in these experiments were sourced from ATCC and maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), nonessential amino acids (NEAA), L-glutamine, and antibiotics.

Mosquito Infection Protocol.

One- to two-day-old adult Liverpool (Lvp) strain mosquitoes and Dcr-2 null mutants were infected by thoracic injection with 0.5 µL of SINV at a concentration of 10⁶ pfu/mL, diluted in DMEM supplemented with 10% FBS, NEAA, L-glutamine, and antibiotics. Control mosquitoes received mock injections with an equal volume of virus-free DMEM. Wild-type siblings served as additional controls for experiments involving Dcr-2 null mutants. Mosquitoes were maintained at 28 °C, with 80% relative humidity and a 14-h day/10-h night light cycle. Viral loads in mosquito samples were measured using plaque assays conducted in triplicate, as well as strand-specific qRT-PCR.

RNA Collection, Extraction, and Quantification.

Mosquito specimens were collected at designated intervals, flash frozen using liquid nitrogen, and stored at −80 °C until processing. RNA extraction was performed on homogenized mosquito samples (n = 5) using TRI Reagent RT (Molecular Research Center, Cincinnati, OH) following the manufacturer’s protocol. Total RNA yield was determined using a Qubit fluorometer (Thermo Scientific, Waltham, MA). Viral RNA levels were assessed through strand-specific quantitative RT-PCR as previously described by Plaskon et al. (58). The reactions employed TaqMan technology (Applied Biosystems, Waltham, MA) with the following oligonucleotides:

• Forward primer: 5′-ATCACAATTGGCAACGAGAAGAG-3′

• Reverse primer: 5′-CTGTGGGTTCGGAGAATAGTGG-3′

• Probe: 5′-CTAAAAGCAGCCGAACTC-3′

To ensure reproducibility, we performed each experiment with a minimum three independent biological replicates. The resulting data were evaluated for statistical significance using a one-way ANOVA.

Small RNA Library Preparation and Analysis.

Libraries were prepared from the same total RNA isolated for RT-PCR. Small RNAs (18 to 35 nucleotides) were isolated using polyacrylamide gel electrophoresis (PAGE) and prepared for sequencing with the Illumina (San Diego, CA) TruSeq small RNA sample prep kit following the manufacturer’s instructions. All biological replicates within each cohort were multiplexed and sequenced on a single HiSeq (Illumina) lane at The Scripps Research Institute’s Next Generation Sequencing and Microarray Core Facility in La Jolla, CA. Bioinformatic analysis followed previously described protocols (38). Briefly, we trimmed 3′ adapter sequences using Fastx_clipper (https://hannonlab.cshl.edu/fastx_toolkit/) and removed identifiable noncoding RNA sequences (rRNAs, tRNAs, snRNAs, snoRNAs) using Bowtie (59). Remaining reads were mapped to the TR339 reference genome using Bowtie (v-mode with one mismatch permitted). For differential expression analysis between replicate datasets, we used the edgeR Bioconductor package (https://bioconductor.org/packages/release/bioc/html/edgeR.html). Reads with fewer than five copies per million were excluded, and all identifiable reads (vsRNAs, miRNAs, TE-derived small RNAs, endo-siRNAs) were used to normalize datasets through the TMM method (60). For figure presentation, all datasets were normalized to the smallest library by using a custom script to randomly select equal numbers of small RNA reads from each larger dataset; thus, creating datasets of equal size for comparison. Representative replicates were then selected for figure presentation.

EMSAs.

SINV nsP2 and nsP3 sequences were cloned into the pMAL-cX5 expression vector for recombinant protein production. His-tagged proteins were expressed in Escherichia coli and subsequently purified using affinity chromatography following standard manufacturer protocols (New England Biolabs Inc., Ipswich, MA). Final protein concentrations were quantified using a Qubit fluorometer (Thermo Scientific, Waltham, MA). Multiple nucleic acid substrates were prepared for binding assays. Long double-stranded RNA (dsRNA) substrates of 180-nt were synthesized from a DNA template PCR-amplified from GFP using a forward (5′-ATGGCTAGCAAAGGAGAAGAACTTTTCACTG-3′) and reverse primer (5′-GTATGTAGCATCACCTTCACCCTCTCCAC-3′), as previously described (61). A shorter 120-nt dsRNA substrate was generated using the same methodology. Small interfering RNA duplexes were produced through in vitro dicing of long dsRNA using recombinant Ae. aegypti Dcr-2, followed by purification with TRI Reagent RT (Molecular Research Center, Cincinnati, OH) and LiCl precipitation. For control nucleic acids, a 60-base pair DNA sequence was PCR-amplified from GFP using a forward (5′-GGAAAGCTTACCCTTAAATTTATTTGC-3′) and reverse primer (5′-TGTTGGCCATGGAACAGGTAG-3′), then gel-purified. Single-stranded RNA (180-nt) was synthesized by in vitro transcription using SP6 RNA Polymerase (Thermo Scientific, Waltham, MA) from a DNA template amplified from GFP with forward (5′-ATGGCTAGCAAAGGAGAAGAACTTTTCACTG-3′) and reverse primers (5′-TGTTGGCCATGGAACAGGTAGTTTTCCAGTAG-3′). The ssRNA was subsequently purified with TRI Reagent RT and LiCl precipitation. Binding reactions were performed using 100 ng of nucleic acid substrate (180-nt dsRNA, 120-nt dsRNA, or 21-nt siRNA duplexes) and 0.45 nmol of purified recombinant nsP2 or nsP3 protein in binding buffer containing 10 mM Tris (pH 8.0), 50 mM KCl, 3.5 mM DTT, 2.5% glycerol, and 0.25% Tween-20. Following a 20-min incubation at room temperature, samples were loaded onto prechilled nondenaturing polyacrylamide gels for electrophoretic separation. Ribonucleoprotein complexes were subsequently visualized using SYBR Green staining (Thermo Scientific, Waltham, MA).

Dicer Protection Assays.

The cDNA sequences encoding AaDcr-2 (GenBank accession number XM_021850223.1) and DmDcr-2 (GenBank accession number NM_079054.5) were cloned into the pMAL-c5x vector with His-tag fusion. The recombinant proteins were expressed and affinity-purified following the manufacturer’s protocol. To optimize enzymatic activity, the purification buffer was exchanged with a reaction-compatible buffer (20 mM HEPES-KOH, pH 8.0, 100 mM NaCl, and 1 mM TCEP) using Econo-Pac 10DG Columns (Bio-Rad, Hercules, CA). In vitro dicing assays were performed according to the protocol described by Cenik et al. (62) with only minor modifications. To establish baseline Dicer activity, control reactions containing either AaDcr-2 or DmDcr-2 protein (1.0 nmol) were incubated with a 720-nt dsRNA (500 ng) substrate. To assess viral protein protection, experimental reactions contained identical components supplemented with either nsP2 or nsP3 (1.8 nmol). All reactions were incubated overnight at room temperature in reaction buffer (18 mM HEPES-KOH, pH 7.4, 100 mM potassium acetate, 3.3 mM magnesium acetate, 7.5 mM DTT, 5% glycerol, 0.25% Tween-20), and an ATP regeneration system (2.25 µg/mL creatine kinase, 15 mM creatine phosphate, and 1 mM ATP). Reaction products were separated by 15% urea-PAGE under denaturing conditions and visualized using SYBR Green staining (Thermo Scientific, Waltham, MA).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All small RNA datasets generated by this study are available for download through the National Center for Biotechnology Information, Sequence Read Archive, BioProject ID PRJNA1295959 (63). All custom code and data analysis scripts used in this study are publicly available at GitHub (https://doi.org/10.5281/zenodo.17955948) (64). All other data are included in the manuscript and/or SI Appendix.