Antiviral RNAi Response against the Insect-Specific Agua Salud Alphavirus

ABSTRACT Arboviruses transmitted by mosquitoes are responsible for the death of millions of people each year. In addition to arboviruses, many insect-specific viruses (ISVs) have been discovered in mosquitoes in the last decade. ISVs, in contrast to arboviruses transmitted by mosquitoes to vertebrates, cannot replicate in vertebrate cells even when they are evolutionarily closely related to arboviruses. The alphavirus genus includes many arboviruses, although only a few ISVs have been discovered from this genus so far. Here, we investigate the interactions of a recently isolated insect-specific alphavirus, Agua Salud alphavirus (ASALV), with its mosquito host. RNA interference (RNAi) is one of the essential antiviral responses against arboviruses, although there is little knowledge on the interactions of RNAi with ISVs. Through the knockdown of transcripts of the different key RNAi pathway (small interfering RNA [siRNA], microRNA [miRNA], and P-element-induced wimpy testis [PIWI]-interacting RNA [piRNA]) proteins, we show the antiviral role of Ago2 (siRNA), Ago1 (miRNA), and Piwi4 proteins against ASALV in Aedes aegypti-derived cells. ASALV replication was increased in Dicer2 and Ago2 knockout cells, confirming the antiviral role of the siRNA pathway. In infected cells, mainly ASALV-specific siRNAs are produced, while piRNA-like small RNAs, with the characteristic nucleotide bias resulting from ping-pong amplification, are produced only in Dicer2 knockout cells. Taken together, ASALV interactions with the mosquito RNAi response differ from those of arthropod-borne alphaviruses in some aspects, although they also share some commonalities. Further research is needed to understand whether the identified differences can be generalized to other insect-specific alphaviruses. IMPORTANCE Mosquitoes are efficient vectors for many arboviruses that cause emergent infectious diseases in humans. Many insect-specific viruses (ISVs) that can infect mosquitoes but cannot infect vertebrates have been discovered in the last decade. ISVs have attracted great attention due to their potential use in mosquito or arbovirus control, by either decreasing mosquito fitness or restricting arbovirus replication and transmission to humans. However, ISV-mosquito interactions are not well understood. RNA interference (RNAi) is the most important innate immune response against many arboviruses, while it is unknown if it is antiviral against ISVs. Here, we investigate in detail the antiviral effect of the RNAi response in mosquitoes against an ISV for the first time. Using a recently isolated insect-specific alphavirus, we show that the regulation of virus replication was different from that for arthropod-borne alphaviruses despite some similarities. The differences in mosquito-virus interactions could drive the different transmission modes, which could eventually drive the evolution of arboviruses. Hence, an understanding of mosquito-ISV interactions can shed light on the ecology and evolution of both ISVs and the medically important arboviruses.

ABSTRACT Arboviruses transmitted by mosquitoes are responsible for the death of millions of people each year. In addition to arboviruses, many insect-specific viruses (ISVs) have been discovered in mosquitoes in the last decade. ISVs, in contrast to arboviruses transmitted by mosquitoes to vertebrates, cannot replicate in vertebrate cells even when they are evolutionarily closely related to arboviruses. The alphavirus genus includes many arboviruses, although only a few ISVs have been discovered from this genus so far. Here, we investigate the interactions of a recently isolated insect-specific alphavirus, Agua Salud alphavirus (ASALV), with its mosquito host. RNA interference (RNAi) is one of the essential antiviral responses against arboviruses, although there is little knowledge on the interactions of RNAi with ISVs. Through the knockdown of transcripts of the different key RNAi pathway (small interfering RNA [siRNA], microRNA [miRNA], and P-element-induced wimpy testis [PIWI]-interacting RNA [piRNA]) proteins, we show the antiviral role of Ago2 (siRNA), Ago1 (miRNA), and Piwi4 proteins against ASALV in Aedes aegypti-derived cells. ASALV replication was increased in Dicer2 and Ago2 knockout cells, confirming the antiviral role of the siRNA pathway. In infected cells, mainly ASALV-specific siRNAs are produced, while piRNA-like small RNAs, with the characteristic nucleotide bias resulting from ping-pong amplification, are produced only in Dicer2 knockout cells. Taken together, ASALV interactions with the mosquito RNAi response differ from those of arthropod-borne alphaviruses in some aspects, although they also share some commonalities. Further research is needed to understand whether the identified differences can be generalized to other insect-specific alphaviruses. IMPORTANCE Mosquitoes are efficient vectors for many arboviruses that cause emergent infectious diseases in humans. Many insect-specific viruses (ISVs) that can infect mosquitoes but cannot infect vertebrates have been discovered in the last decade. ISVs have attracted great attention due to their potential use in mosquito or arbovirus control, by either decreasing mosquito fitness or restricting arbovirus replication and transmission to humans. However, ISV-mosquito interactions are not well understood. RNA interference (RNAi) is the most important innate immune response against many arboviruses, while it is unknown if it is antiviral against ISVs. Here, we investigate in detail the antiviral effect of the RNAi response in mosquitoes against an ISV for the first time. Using a recently isolated insect-specific alphavirus, we show that the regulation of virus replication was different from that for arthropod-borne alphaviruses despite some similarities. The differences in mosquito-virus interactions could drive the different transmission modes, which could eventually drive the preferentially binds antisense piRNAs derived from endogenous viral elements (EVEs). These EVEs can be integrated into the mosquito genome during RNA virus infection and act as an "adaptive immune response" combined with the produced vpiRNAs and Piwi4 (16). The knockdown of Piwi4 transcripts resulted in increased virus titers, supporting its antiviral role. In contrast, the knockdown of piRNA pathway proteins did not have a strong antiviral role against the arboviruses tested so far (13,(17)(18)(19), except for Rift Valley fever virus (20). Piwi4 has been shown to interact with proteins of the piRNA and siRNA pathways; however, its antiviral activity against an arthropod-borne alphavirus is independent of Dcr2 activity in the Aag2 cells (18).
Compared to arboviruses, ISV interactions with the mosquito RNAi pathways are less studied. Studies on ISVs mainly focused on the detection of virus-specific small RNAs in persistently infected cell lines. Here, the production of vsiRNAs and, in some cases, vpiRNAs was detected for different families, including Flaviviridae, Birnaviridae, and Phenuiviridae (3,21,22) Our knowledge on RNAi-ISV interactions is further limited for insect-specific alphaviruses, as no persistently infected cell lines are known (23,24), and only five insect-specific alphaviruses have been identified in mosquitoes so far: Eilat virus (EILV), Taï Forest alphavirus (TALV), Mwinilunga alphavirus (MWAV), Yada Yada virus, and Agua Salud alphavirus (ASALV) (25)(26)(27)(28)53). So far, only the latter has been studied for its interactions with the RNAi response (28). Indeed, ASALV infection of Aedes albopictusderived cells induces the production of vsiRNAs but lacks the production of vpiRNAs. Moreover, it is unknown whether the siRNA pathway is antiviral against ASALV.
In addition to the mosquitoes' ability to control virus replication through the RNAi pathways, viruses can also suppress the RNAi response. Indeed, some ISVs such as Culex Y virus and Mosinovirus are known to interfere with the RNAi response (29)(30)(31) by encoding an RNAi suppressor protein. However, it is not known whether this is the case for insect-specific alphaviruses.
Here, we investigated the interactions of ASALV with the mosquito RNAi pathways in detail. We show the antiviral role of the exo-siRNA pathway against ASALV by using A. aegypti-derived Dcr2 and Ago2 knockout cell lines. ASALV-specific siRNAs were still produced in the absence of Ago2 but decreased in the Dcr2 knockout cell line. ASALV triggered piRNA-like small RNA production through the ping-pong production pathway only in Dcr2 knockout cells. By knocking down additional key RNAi transcripts, we further show the involvement of Ago1, Ago2, and Piwi4 in the antiviral activity against ASALV in A. aegypti-derived cells.

RESULTS
ASALV efficiently replicates in AF5 cells. The successful replication of ASALV was previously shown in A. albopictus-derived C6/36 and U4.4 cells (28). To verify that ASALV could replicate in A. aegypti-derived AF5 cells, a cumulative growth curve was performed by collecting the supernatant every 24 h until 72 h postinfection (hpi). The growth curves showed that ASALV efficiently replicates in AF5 cells (Fig. 1), without any visible cytopathic effect.
piRNA-sized small RNAs with ping-pong characteristics are produced only in Dcr2 KO cells. To investigate the production of ASALV-specific small RNAs in the different cells, small RNA sequencing of ASALV-infected cells was performed in A. aegyptiderived AF5, AF319, and AF525 cells. Cells were infected with ASALV (MOI of 0.5), and total RNA was isolated at 48 hpi, followed by small RNA sequencing and bioinformatics analysis. Two independent replicates per cell line were performed, resulting in similar findings (Table 1 and Fig. 3; see also Fig. S1 and S2 in the supplemental material).
In AF5 cells, ASALV-specific siRNAs (Fig. 3A) are produced and map across the genome (sense) and antigenome (antisense) (Fig. 3B), similar to the results previously observed in U4.4 cells (28). Similarly, in Ago2 KO AF525 cells, the majority of ASALVspecific small RNAs are 21-nt-long vsiRNAs (Fig. 3A). They also map across the whole genome and antigenome although with a bias to the 59 and 39 ends (Fig. 3B), which is not observed in AF5 cells. In Dcr2 KO AF319 cells, ASALV-specific siRNAs are strongly decreased, and a majority of them map to the 39 end of the ASALV genome.
piRNA-sized small RNAs were observed at a low concentration in both AF5 and AF525 cells ( Fig. 3A to C) and did not show the "ping-pong" amplification characteristics (Fig. 4). In contrast, AF319 cells produce ASALV-specific piRNA-sized small RNAs ( Fig. 3A to C) with the ping-pong amplification characteristics (Fig. 4). Antisense and sense piRNA-sized small RNAs showed a clear 10-nucleotide overlap. Adenine was the most frequent nucleotide at the 10th position of the sense piRNA-sized small RNA sequence, although the bias was not very strong. In antisense piRNA-sized small RNA sequences, uridine was the most frequent nucleotide at the first position (Fig. 4).
In all cells, piRNA-sized small RNAs were mapped around the subgenomic promoter and 59 end of the subgenomic RNA, encoding the capsid protein, similar to vpiRNAs produced by arthropod-borne alphaviruses (14,17). However, in AF319 cells, some piRNA-sized small RNAs also map to the 59 end of the genome (Fig. 3C).
Taken together, vsiRNAs are the main small RNA species produced against ASALV infection under normal circumstances. In the absence of Dcr2, ASALV can induce piRNAsized small RNAs with sequence characteristics indicative of the ping-pong amplification pathway.
siRNA pathway, miRNA pathway, and Piwi4 are involved in the antiviral RNAi response against ASALV. The increased ASALV infection in the knockout cell lines supports the involvement of the siRNA pathway in antiviral defense against ASALV. To investigate the involvement of the other RNAi pathway proteins against ASALV in A. aegypti-derived AF5 cells, transcripts of different RNAi proteins were silenced by transfecting cells with sequence-specific dsRNAs (Ago1, Ago2, Ago3, Piwi4, Piwi5, and Piwi6) prior to ASALV infection (MOI of 0.5) (Fig. 5A). Successful silencing was verified (Fig. 5A), and viral RNA was quantified in the cells at 48 hpi and compared to the levels in control cells (transfected with dsRNA specific for LacZ [dsLacZ]). The increase in viral replication was statistically significant in cells where Ago1 (t = 2.817; df = 4.665; P = 0.040), Ago2 (t = 6.437; df = 5.039; P = 0.001), or Piwi4 (t = 2.628; df = 8.543; P = 0.029) transcripts were silenced (Fig. 5B). The ASALV RNA fold change was more pronounced in Ago2-silenced cells than in Ago1-and Piwi4silenced cells (Table S2). Furthermore, when Piwi4 silencing was conducted using Piwi4-specific siRNAs instead of dsRNAs, ASALV replication increased in both AF5 and AF319 cells, although this increase was not significant (Fig. S3).
No RNAi suppressor effect of ASALV was detected in AF5 cells. Several insect viruses have been reported to encode proteins that interfere with the antiviral RNAi pathway, named viral suppressors of RNAi (VSRs). VSRs can interfere at different steps of the RNAi pathways by interacting with key molecules (e.g., dsRNA or siRNAs) or proteins (e.g., Ago2 or Dcr2), mostly of the exo-siRNA pathway. To determine if ASALV can suppress the exo-siRNA response in mosquito-derived cells, a previously used luciferase-based RNAi suppressor assay was performed (18,32). AF5 cells were either infected with ASALV (MOI of 10) or mock infected. After 24 hpi, cells were cotransfected with firefly luciferase (FFluc) and Renilla luciferase (internal control) expression constructs as well as dsRNA (FFluc or LacZ as a control) or siRNA (siRNA against FFluc [siFFluc] or against hygromycin B (siHyg) as a control) to induce silencing. Luciferase activity was measured 24 h posttransfection, and the levels of sequence-specific silencing of firefly luciferase in ASALV-and mock-infected cells were compared.

DISCUSSION
RNA interference (RNAi) is an important antiviral response in insects, including mosquitoes. The interaction between the mosquito RNAi pathways and a variety of viruses can be identified by detecting virus-specific small RNAs and increased viral infection in the case of the silencing of key proteins of the different RNAi pathways. RNAi has been shown to act antivirally in mosquitoes against all viruses tested so far, although differences regarding the importance of specific pathways or proteins have been reported (8). Our knowledge about the antiviral RNAi response in mosquitoes comes from arbovirus studies, although mosquitoes often harbor insect-specific viruses (ISVs). Small RNAs specific to a variety of ISVs were found in infected cells and mosquitoes. However, the antiviral role of the RNAi pathway against ISVs is not known (3). Here, we identified the antiviral function of the mosquito RNAi pathways against an insect-specific alphavirus for the first time.
The only previous study investigating an RNAi response specific to an insect-specific alphavirus showed the production of ASALV-specific 21-nt vsiRNAs in A. albopictusderived (U4.4) cells, although no vpiRNAs were observed (28). Our results confirm this previously reported lack of ASALV-specific piRNA production in A. aegypti-derived RNAi-competent AF5 cells (Fig. 4). In contrast, arthropod-borne alphaviruses induce both vsiRNAs and vpiRNAs in vitro in A. aegypti-and A. albopictus-derived cell lines as well as in mosquitoes (17,(33)(34)(35)(36)(37). Despite the difference in the small RNAs that are produced during infection, the mapping of ASALV-specific siRNAs (in both AF5 and U4.4 cells) was very similar to the mapping of arthropod-borne alphaviruses. Both map along the genome and antigenome, more or less equally, with some cold and hot spots (17,35,36). This suggests that similar to arthropod-borne alphaviruses, ASALV also mainly induces vsiRNA production through dsRNA replicative intermediates.
ASALV replication is increased in both Ago2-silenced ( Fig. 5B; see also Table S2 in the supplemental material) and Ago2 or Dcr2 knockout (Fig. 2) cells, highlighting the antiviral role of the exo-siRNA pathway against ASALV. Similarly, silencing or knockout of Ago2 or Dcr2 induced an increase in infection by the tested arthropod-borne alphaviruses (15,17,19,38,39). Furthermore, similar results have been found for arboviruses belonging to other virus families or orders (8), except for Zika virus (ZIKV), where no antiviral activity was reported for Ago2 in silenced cells (9,19). For the arthropod-borne alphavirus Semliki Forest virus (SFV), the magnitudes of the increase in infection were similar in Dcr2 and Ago2 knockout cells (9). In contrast, for ASALV, the differences  between Ago2 and Dcr2 knockout cells suggest an additional role of Dcr2 in the antiviral response against ASALV independent of Ago2. For instance, Dcr2 can detect viral RNA and induce an antiviral protein, Vago, which activates the Jak-STAT pathway, leading to an antiviral effect in Culex quinquefasciatus (Hsu)-derived cells (40,41). Notably, however, Vago does not seem to be induced in infected A. aegypti-derived Aag2 cells (42). Alternatively, this increased antiviral effect of Dcr2 against ASALV might be linked to another as-yet-unknown antiviral pathway related to Dcr2 activity. ASALV-specific piRNA-sized small RNAs with ping-pong amplification characteristics were produced only in Dcr2 knockout cells. Similar to the results with arthropod-borne alphaviruses, ASALV-specific piRNA-sized small RNAs were mainly produced from the positive (genomic) strand, in contrast to vsiRNAs, which have similar amounts of sense and antisense RNA molecules. A likely explanation for this is the Dcr2-dependent production mechanism of vsiRNAs from dsRNA molecules. In contrast, the production of vpiRNAs is Dcr independent and not dependent on dsRNA inducer molecules. Previous reports have also shown an increase of SFV-specific vpiRNAs in cells lacking the Dcr2 protein (18). It is possible that the increase in vpiRNA production is a result of (i) the increased viral replication due to the lack of the antiviral Dcr2 protein, (ii) the high concentration of ASALV RNA in the cytoplasm that is not cut into vsiRNAs, or (iii) a combination of both. Although ASALV replication was increased in Ago2 knockout cells, no ping-pong-specific vpiRNAs were detected. While this could mean that increased viral replication is not solely sufficient for ASALV-specific vpiRNA production, it has to be noted that the increase in ASALV replication in Ago2 knockout cells was still lower than that in Dcr2 knockout cells. Therefore, it could be that the increased ASALV RNA concentration in Ago2 KO cells is not sufficient to trigger vpiRNA production, in contrast to Dcr2 KO cells. In addition, it is likely that in Dcr2 knockout cells specifically, the amount of viral dsRNA molecules would increase. As the precise trigger for vpiRNA production in mosquitoes is not yet known, it could be that the concentration of ASALV dsRNA in Dcr2 knockout cells could play a role in triggering vpiRNA production. On the other hand, the putative essential proteins for the biogenesis of vpiRNAs, Piwi5 and Ago3, were not antiviral against ASALV (Fig. 5B), consistent with findings from arthropodborne alphaviruses (14,15,19).
The silencing of Piwi4 resulted in a small but statistically significant increase in ASALV replication, as was previously shown for other arboviruses, including alphaviruses (18,43). The general antiviral role of Piwi4 is still not clear. Piwi4 is not required for the production of SFV-or SINV-specific vpiRNAs, but it was recently shown to bind DENV-specific piRNAs derived from viral cDNA in infected A. aegypti (16). While an interaction between Piwi4 and piRNA as well as siRNA pathway proteins, including Dcr2, has previously been shown, Piwi4 antiviral activity is independent of Dcr2 in SFVinfected cells (18,44). To check this for ASALV, we silenced Piwi4 by siRNAs in both Dcr2-competent and Dcr2 knockout cell lines. While the silencing of Piwi4 through siRNA increased ASALV replication, the increase was not significant in either of the cell lines (Fig. S3). Whether this is due to the lower silencing efficiency of Piwi4 with siRNAs than with dsRNA is not known. In addition, in the knockout cell lines, it is possible that the high viral replication does not leave much room for additional significant increase in the ASALV replication upon siRNA-based silencing. Hence, it was not possible to conclude whether the effect of Piwi4 is Dcr2 independent.
Our results suggest an antiviral effect of Ago1, which is primarily involved in the miRNA pathway (Fig. 5). Although the mosquito miRNA response has been shown to interact with viruses through either mosquito-or virus-encoded miRNAs (6), the silencing of Ago1 has not resulted in changes in arboviral alphavirus replication (17,45,46). Similar increases in virus infection upon Ago1 silencing have been reported for midgeborne orthobunyaviruses in A. aegypti-derived cells, in contrast to mosquito-borne orthobunyaviruses (43). Additional experiments are needed to determine if the FIG 5 Ago1, Ago2, and Piwi4 silencing increases ASALV replication in A. aegypti-derived AF5 cells. Cells were transfected with either gene-specific dsRNAs or control dsRNA (LacZ specific). The following day, cells were infected with ASALV (MOI of 0.5), and total RNA was isolated at 48 h postinfection. (A) mRNA targets were quantified using gene-specific primers and ribosomal protein S7 RNA as the housekeeping transcript. The 2 2DDCT values of mRNA targets were calculated with the mean normalized RNA expression of a given transcript in the control cells, within the same replicate, as a control. The resulting mean fold changes and standard errors of the means are shown. (B) ASALV RNA was quantified using ASALV-specific primers and ribosomal protein S7 RNA as the housekeeping transcript. ASALV RNA fold changes were calculated using the 2 2DDCT method with the mean normalized expression of ASALV RNA, of all replicates, in the control cells as a control. Bar plots represent the mean fold changes calculated for each group. At least five independent replicates were performed (*, P , 0.05; **, P , 0.01). difference in Ago1 activity against arthropod-borne alphaviruses compared to insectspecific alphaviruses can be generalized.
Many viruses infecting insects encode proteins to suppress the RNAi pathway, such as flock house virus or Culex Y virus (47). Several arboviruses, such as dengue and West Nile viruses, have also been shown to interfere with the RNAi response by employing competitive substrates for Dcr2, derived from their nucleic acids (47). Furthermore, recent work has identified the nonstructural protein NS2A of flaviviruses as a potent suppressor of RNAi (48). In our experimental system, we did not observe any RNAi suppressor activity by ASALV at either 24 or 48 h posttransfection (Fig. S4).
ISVs belonging to some of the arbovirus families and orders, such as Bunyavirales (49) and Flaviviridae (50), are thought to be ancestral to arboviruses, suggesting that dual-host (invertebrate-vertebrate) tropism evolved from invertebrate-specific viruses. As not many insect-specific alphaviruses have been discovered so far, it is difficult to identify whether ISVs or the arthropod-borne alphaviruses are the ancestors in the alphavirus genus (51). Nevertheless, like other insect-specific alphaviruses so far, ASALV is basal to the Western equine encephalitis virus complex clade, suggesting that arthropod-borne alphaviruses in this clade could have evolved from an ancestral insect-specific virus (28,51). It is also possible that the changes in the mosquito-virus interactions drive their evolution, resulting in their ability to transmit to vertebrates. In this context, differences between arboviral and insect-specific alphavirus interactions with mosquito RNAi pathways could be one of the reasons why ISVs were restricted to invertebrate hosts. In contrast to the arthropod-borne alphaviruses studied so far, we showed that ASALV-specific vpiRNAs are not produced in Dcr2-competent cells, and Ago1 was antiviral against ASALV. However, to be able to generalize this observation to other insect-specific alphaviruses, more studies describing their interactions with mosquito hosts are needed. Further studies taking both the persistent nature of ISVs and the tissue specificity of the RNAi response into account could determine whether the interactions of insect-specific alphaviruses with the RNAi pathways restrict ISVs to their mosquito hosts.
ASALV stock. Previously isolated and plaque-purified ASALV was used for all experiments (28). Virus stocks were produced by inoculating C6/36 cells. The supernatant was harvested upon the observation of morphological changes and was cleared from the cell debris by centrifugation. For 50% tissue culture infective dose (TCID 50 ) virus quantification, 4 Â 10 4 C6/36 cells per well were seeded into 96-well plates 2 h before infection. Serial dilutions were performed in L15 complete medium.
dsRNA synthesis. Primers specific for A. aegypti Ago1, Ago2, Ago3, Piwi4, Piwi5, and Piwi6 (17) and LacZ (Aedes-T7-BGal F/R) (52) flanked by T7 RNA polymerase promoter sequences were used to amplify gene-specific fragments. Amplified fragments were validated by Sanger sequencing. PCR products were used for in vitro transcription and subsequent column-based purification using the MEGAscript RNAi kit (Thermo Fisher Scientific) according to the manufacturer's instructions.
Growth kinetics. A total of 4 Â 10 5 AF5 cells per well were seeded into 12-well plates a day prior to infection and kept at 28°C overnight. Cells were infected with ASALV at a multiplicity of infection (MOI) of 0.1. After 1 h of incubation, the infectious medium was replaced with 1 mL of fresh L15 medium with supplements. Samples were taken at different time points (0, 24, 48, and 72 h postinfection [hpi]). Infection and negative controls were performed in triplicates, and three independent experiments were performed. The amount of viral RNA in the supernatant was quantified using RNA isolated from supernatant samples with TRIzol LS (Invitrogen) according to the manufacturer's protocol. A QuantiTect SYBR green quantitative real-time PCR (qRT-PCR) one-step kit (Qiagen) was used to quantify ASALV using previously established primers (28). Samples were run in technical triplicates. An in-run calibrator and an external standard curve were used to perform absolute quantification using a Roche LightCycler 480 II instrument.
ASALV infection of knockout cells and small RNA sequencing. A total of 3 Â 10 5 cells/well (AF5, AF525, and AF319) were seeded into 24-well plates and infected with ASALV at an MOI of 0.5 the following day. Total RNA of infected cells was isolated at 48 hpi with TRIzol according to the manufacturer's protocol, using glycogen as a carrier. A QuantiTect SYBR green qRT-PCR one-step kit (Qiagen) was used to quantify ASALV using previously established primers (28). ASALV RNA fold changes were calculated using the 2 2DDCT method with ribosomal protein S7 RNA as the housekeeping gene and AF5 cells as the control group.
To investigate the production of ASALV-specific small RNAs in AF5, AF525, and AF319 cells, 8 Â 10 5 cells were seeded into a 6-well plate and infected with ASALV (MOI of 1). Total RNA was isolated at 48 hpi with TRIzol (Ambion), according to the manufacturer's protocol, with glycogen as a carrier. Small RNAs of 1 mg total RNA were sequenced using BGISEQ-500 at BGI Tech (Hong Kong, China) as previously described (9). For one of the AF525 samples (see Fig. S1 in the supplemental material), total RNA was sequenced at IKMB (Kiel, Germany), using 100 ng total RNA for library preparation with the Nextflex small RNA-Seq kit v3 (PerkinElmer Inc., Waltham, MA, USA), followed by library sequencing on one lane of the NovaSeq6000 SP v1.0 platform (2 by 50 bp). Data analyses were performed as previously described (15). The ASALV genome sequence was used as the template (GenBank accession number MK959115).
Knockdown experiments. A total of 2.5 Â 10 5 AF5 cells/well were seeded into 24-well plates the day before transfection with 200 ng of gene-specific dsRNAs or control dsRNA (dsLacZ) per well and transfected using 1 mL of Dharmafect2 reagent (GE Dharmacon). For siRNA knockdowns in knockout cells, 20 nM either Piwi4-specific siRNAs or control siRNA (Horizon Discovery) was transfected using 2 mL Dharmafect2 reagent (GE Dharmacon), as previously described (18). The following day, ASALV infection (MOI of 0.5) was performed. At 48 hpi, total RNA was isolated from cells using TRIzol (Ambion). cDNA of 1.5 mg RNA was produced using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega) and oligo(dT) 15 primers (Thermo Fisher Scientific) according to the manufacturers' protocols. SYBR green qRT-PCR for mRNA targets was performed using gene-specific primers (Table S1) and ribosomal protein S7 RNA as the housekeeping gene transcript. Results were analyzed using the 2 2DDCT method with LacZ dsRNA samples as the control. All qPCRs were performed in technical triplicates.
RNA silencing suppressor assay. To assess whether the presence of ASALV in cells could suppress the RNA silencing response, AF5 cells were seeded into 24-well plates (1.8 Â 10 5 cells/well) 1 day prior to ASALV infection (MOI of 10). The day after infection, cells (ASALV or mock infected) were transfected with firefly and Renilla luciferase expression constructs, pIZ-Fluc and pAcIE1-Rluc (18,32), and either 0.5 ng dsRNA (either dsFluc or dsLacZ as a negative control) or 0.1 ng siRNA (either siFluc or siHyg as a negative control) using 1 mL of Dharmafect2. At 24 h posttransfection, the cells were lysed, and luciferase was measured with a dual-luciferase assay (dual-luciferase reporter assay system; Promega) according to the manufacturer's protocol on a Glomax luminometer.
Statistical analyses. R (version 3.5.2) was used for statistical analyses. First, the normality (Shapiro-Wilk test) and variance (F test) of the data were tested. Student's t test was used for normally distributed homoscedastic data, or the Welch t test was used for normally distributed heteroscedastic data. A P value of ,0.05 was considered statistically significant.
Data availability. Small RNA sequencing data are available in the NCBI Sequence Read Archive under BioProject accession number PRJNA725665.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.