Characterization of the Zika virus induced small RNA response in Aedes aegypti cells

RNA interference (RNAi) controls arbovirus infections in mosquitoes. Two different RNAi pathways are involved in antiviral responses: the PIWI-interacting RNA (piRNA) and exogenous short interfering RNA (exo-siRNA) pathways, which are characterized by the production of virus-derived small RNAs of 25–29 and 21 nucleotides, respectively. The exo-siRNA pathway is considered to be the key mosquito antiviral response mechanism. In Aedes aegypti-derived cells, Zika virus (ZIKV)-specific siRNAs were produced and loaded into the exo-siRNA pathway effector protein Argonaute 2 (Ago2); although the knockdown of Ago2 did not enhance virus replication. Enhanced ZIKV replication was observed in a Dcr2-knockout cell line suggesting that the exo-siRNA pathway is implicated in the antiviral response. Although ZIKV-specific piRNA-sized small RNAs were detected, these lacked the characteristic piRNA ping-pong signature motif and were bound to Ago3 but not Piwi5 or Piwi6. Silencing of PIWI proteins indicated that the knockdown of Ago3, Piwi5 or Piwi6 did not enhance ZIKV replication and only Piwi4 displayed antiviral activity. We also report that the expression of ZIKV capsid (C) protein amplified the replication of a reporter alphavirus; although, unlike yellow fever virus C protein, it does not inhibit the exo-siRNA pathway. Our findings elucidate ZIKV-mosquito RNAi interactions that are important for understanding its spread.


Author summary
The recent outbreak of Zika virus (ZIKV) in the Americas has resulted in a severe threat to public health. ZIKV is transmitted by Aedes aegypti mosquitoes, thus it is important to understand virus-vector interactions. Analysis of ZIKV infection in mosquito cells indicated that two RNA interference pathways are involved during infection: the exogenous short-interfering (si)RNA (exo-siRNA) and PIWI-interacting (pi)RNA pathways. If Dcr2, an enzyme responsible for cleaving dsRNA into siRNAs, is knocked out, ZIKV replication is increased compared to control cells. However, the knockdown of Ago2 expression had no significant enhancing effect on ZIKV replication. In the case of the PIWI pathway, a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
Zika virus (ZIKV) is an arbovirus belonging to the family Flaviviridae, genus Flavivirus. The ZIKV genome is a single-stranded, positive-sense RNA molecule that demonstrates typical flavivirus organization-with a single open reading frame encoding the structural and non-structural proteins, flanked by 5' and 3' non coding regions [1,2]. The emergence of the virus in the Americas at the beginning of 2015, in addition to its links to Guillain-Barré syndrome as well as microcephaly, placed the virus firmly in the spotlight [3][4][5].
Virus-specific piRNAs (vpiRNA) have also been described in arbovirus infected mosquitoes and derived cell lines [20][21][22]24,28,31,32]. These are 25-29 nt in length and are produced through a so-called ''ping-pong" amplification loop which gives the vpiRNAs specific molecular signatures: primary-type piRNAs have a uridine at position 1 [U1] and secondary piRNAs have an adenine at position 10 [A10]. The Ae. aegypti genome encodes seven PIWI proteins (Piwi1-7) and a single Ago3 protein involved in this pathway [24,33,34]. The role of piRNAs in the control of arbovirus infection is enigmatic and although piRNAs have been suggested to be antiviral, this has not been directly demonstrated. The only PIWI protein with antiviral activity in Ae. aegypti is Piwi4. However, Piwi4 does not bind piRNAs, nor is it involved in the production of virus-specific piRNAs [24,34,35]. Previous studies have shown that Ago2 silencing in mosquito-derived cells or mosquitoes increased replication of arboviruses of the Togaviridae (genus Alphavirus), Flaviviridae and Bunyaviridae (genus Orthobunyavirus) families [23,24,31,[35][36][37] following pioneering work on this RNAi effector protein in D. melanogaster [38]. Silencing of Piwi4 has only been shown to result in the upregulation of the replication of the alphavirus Semliki Forest virus (SFV) and the bunyaviruses Bunyamwera virus (BUNV) and Rift Valley fever virus (RVFV) [24,31,39].
Here we studied ZIKV interactions with the RNAi response of Ae. aegypti mosquito cells. We found that ZIKV induced the production of both virus-specific siRNAs and piRNAs, although vpiRNAs lack the specific signature expected of such small RNAs. Furthermore, our findings indicate that although vsiRNAs were loaded into Ago2, silencing of this protein did not enhance virus replication. Indeed, increased replication in Dcr2 knockout cell lines was the main indicator for an antiviral role of the exo-siRNA pathway. With the exception of Piwi4, the knockdown of PIWI proteins also did not enhance virus replication. Moreover, ZIKV capsid protein (C) amplified the replication of an alphavirus-based reporter system but did not display exo-siRNA pathway inhibition, unlike that of another flavivirus, yellow fever virus (YFV) [40]. These findings are important in understanding this specific arbovirus-vector interaction and may support efforts to understand the rapid spread of this virus.

Protein immunoprecipitation
For small RNA capture assays, 1x10 7 cells stably expressing V5-eGFP, V5-Ago2 or V5-Piwi4 were infected with ZIKV PE243 at MOI 1. At 48 hpi the cells were scraped and washed with PBS. A detailed protocol has been described previously [41]. In short, cells were resuspended in lysis buffer (150mM NaCl, 5mM MgCl2, 20mM Hepes (pH 7.4), 0.5% Triton X-100, protease inhibitor cocktail [Roche]) and kept on ice for 20 min, followed by centrifugation at 15000 x g at 4˚C for 25 min. The supernatant was then transferred into fresh tubes on ice and mouse anti-V5 antibody (at 1:500; ab27671 Abcam) added. Tubes were rotated for 2 h at 4˚C. Following this, protein G magnetic beads (Dynabeads Protein G, Life Technologies) equilibrated with cold washing buffer (150mM NaCl, 5mM MgCl 2 , 20mM Hepes (pH 7.4), 0.5% Triton X-100) were added. Tubes were again rotated for 1 h at 4˚C. The beads were then washed 4 times with washing buffer. Washed beads were finally resuspended in 100 μl washing buffer and 1/20 of the volume was subjected to western blot analysis while the remainder was used for RNA extraction.
Extraction of protein-bound small RNA from beads 5 μl of proteinase K (20mg/ml) was added to magnetic beads resuspended in washing buffer and samples were placed into a water bath at 37˚C for 30 min. Following this, 1 ml of Trizol reagent (Life Technologies) was added to the sample and processed as per the manufacturer's instructions.
cDNA synthesis 250 μl Trizol was added to 1.8x10 5 AF5 or AF319 cells per well of a 24-well plate. Material from two wells was then pooled and the total cellular RNA extracted as per the manufacturer's instructions. 1.5 μg of total RNA was used for cDNA synthesis using Superscript III (Life Technologies) and oligo(dT)15 (Promega), for quantification of mRNAs, or random primers (Promega), for quantification of ZIKV genomic RNA, according to manufacturer's instructions.

Small RNA sequencing and sequence analysis
Trizol was added to 1x10 6 cells or immunoprecipitation samples and RNA was extracted according to the manufacturer's protocol. To increase small RNA precipitation efficiency, glycogen was added as a carrier. Small RNAs of 15-40 nt in length were gel purified and sequenced on an Illumina HiSeq 4000 at BGI Tech. Sequence reads were aligned to the ZIKV reference genome (Genbank accession number: KX197192.1) using in-house BLAST guided bioinformatics pipeline. Maximum of one mismatch was allowed in the alignments. Reads that matched the reference genome with alignment lengths from 18bp to 36bp were extracted for further analysis. They were separated into two groups, positive and negative based on the reads' alignment to genome and antigenome respectively. Size distribution of aligned small RNAs is shown in Fig 1. See S1 Table for additional information. Small RNA sequencing data available at Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra) under accession PRJNA396680.

qRT-PCR
Quantitative RT-PCR for ZIKV and the housekeeping gene S7 was performed using specific primers (S2 Table), SYBR green Mastermix (Abi) and an ABI7500 Fast according to manufacturer's protocol. Results were analyzed using the ΔΔCt method.

Transfection of nucleic acids
For transfections, 2 μl of Dharmafect 2 (GE Healthcare) per well of a 24 well plate was used. For RNAi reporter assays in SFV-infected cells, 1 ng of dsRNA (Fluc specific or LacZ specific) or 1 ng siRNA (targeting Fluc or Hygromycin B resistance gene) were co-transfected with 100 ng of pIZ-Fluc. For assessing silencing efficiency in ZIKV-infected cells, 200 ng of pIZ-Fluc and 100 ng of pAc1-Rluc plasmids were co-transfected with 0.25 ng of dsRNA (Fluc specific or eGFP specific) or 0.25 ng of siRNA (targeting Fluc or Hygromycin B resistance gene). To induce mosquito host gene silencing, 300 ng of gene-specific dsRNA was used.
dsRNA production T7 promoter-flanked PCR products were used for in vitro transcription. These were treated with DNase1 and RNaseA followed by column purification of the dsRNA using the RNAi Megascript kit (Life Technologies) as per the manufacturer's instructions. For silencing purposes, previously described and verified dsRNAs were used [24,31,36].

Protein sequence comparison
To determine the similarity between the capsid protein of different flaviviruses, Needleman-Wunsch algorithm was used (http://www.ebi.ac.uk/Tools/psa/emboss_needle/). The Genbank access numbers are as follows:

ZIKV infection induces the production of virus-derived small RNAs in Ae. aegypti-derived mosquito cells
To assess the production of virus-derived small RNAs Aag2 cells were infected with a Brazilian ZIKV isolate, PE243 [45] at a multiplicity of infection (MOI) of 1. RNA was isolated at 48 hours post-infection (hpi) and small RNA pools were sequenced and analyzed. ZIKV-specific 21 nt long siRNAs (vsiRNAs) were found to be produced in infected cells (Fig 1A). Furthermore, the number of vsiRNAs mapping to the ZIKV genome were found to be roughly equal to those mapping to the antigenome with the vsiRNAs being distributed across the genome and antigenome (Fig 1B). Small RNAs in the size range of ZIKV-specific vpiRNAs (25-29 nts) were also detected. Positional mapping indicated that one region near the 3' end of the genome gave rise to the majority of vpiRNAs as their 5' ends mapped to positions 10094-10098 ( Fig  1C), a region in the NS5-encoding region. Furthermore, these vpiRNAs were exclusively of positive polarity. Additional sequencing analysis of these putative vpiRNAs molecules showed a lack of the typical piRNA ping-pong signature, i.e. the absence of U1/ A10 predominance in the small RNA sequences. Intriguingly, DENV-specific vpiRNAs also mapped to a small number of locations on the genome [35]. Comparative analysis showed that the most predominant piRNA sequences of DENV (piRNA 5' ends mapping to position 9985-9990) overlap with this ZIKV-specific, piRNA-like small RNA generating region (Fig 1D).

Analysis of the interactions between virus-derived RNAs and RNAi effector proteins
We analyzed whether virus-derived small RNAs are loaded into RNAi effector proteins and determined which of these would be crucial for mediating antiviral activities. Ago2-bound small RNAs were investigated using a previously developed Aag2 cell line which stably expresses V5-tagged Ago2 [41]. Ago2 was immunoprecipitated (S1 Fig) at 48 hpi following infection with ZIKV (MOI 1) and the captured small RNAs were sequenced. Samples from cells expressing V5-tagged eGFP were used as control. Small RNA analysis showed that Ago2 bound 21 nt small RNAs (Fig 2) and these ZIKV-specific vsiRNAs were distributed along both genomic and antigenome strands. This indicates that dsRNA-derived vsiRNAs were successfully loaded into Ago2. Similar experiments were conducted to determine which RNAs were bound by Ago3, Piwi4, Piwi5 or Piwi6 proteins (Fig 2, S2 Fig), which are involved in the piRNA ping-pong amplification loop in Ae. aegypti [34,46]. Sequence analysis revealed Ago3-bound ZIKV genome vpiRNAs mapped specifically to positions 10094-10098 of the viral genome (Fig 2B), while the read numbers for other potential vpiRNAs were too low for further analysis. However, no ZIKV-specific piRNA molecules were found to be associated with either Piwi5 or Piwi6 (S2 Fig), although small amounts of 18 nt and 19 nt small ZIKV-specific RNAs were found to be associated with these proteins. In the case of Piwi4, relatively low amounts of ZIKV-specific vsiRNA and vpiRNA-like small RNAs were detected (S2 Fig). However, these may not be bound by Piwi4 directly as other RNAi pathway proteins (Ago2, Dcr2, Ago3, Piwi5 and Piwi6) can interact with Piwi4, as previously shown [34,41], and this will require further investigations. Several flaviviruses are known to produce miRNA-like small RNAs (22-24 nt) and thereby regulate host gene expression [47]. However, our data indicated that the majority of small RNAs in the size range of 22-24 nt could be mapped to positions 10094-10098 (S3 Fig) and these molecules were bound by Ago3, suggesting that these small RNAs were related to vpiRNAs rather than miRNAs.

The effect of RNAi effector knockdown on ZIKV replication
We previously showed that the knockdown of Ago2 and Piwi4 enhanced replication of the alphavirus SFV and the orthobunyavirus BUNV in Aag2 cells [24,31]. Similarly, others have shown that silencing of Ago2 in Ae. aegypti-derived cells or Ae. aegypti mosquitoes enhances Sindbis virus (SINV; genus Alphavirus) and dengue virus (DENV, a flavivirus related to ZIKV) replication [23,35,48]. However, in the case of DENV this was not found to be statistically significant. In order to assess the antiviral roles of these known RNAi effectors on ZIKV, cells were first transfected with previously validated dsRNAs (see Materials and Methods) targeting RNAi effector proteins (Fig 3A) followed by infection with ZIKV at MOI 0.1 24 h later. For validation of the assay, Firefly luciferase (FFLuc) expressing SFV vector SFV(3H)-FFLuc was used. Surprisingly, we found that the knockdown of Ago2 had no effect on ZIKV titer at 72 hpi or 120 hpi (Fig 3B and 3C). However, the knockdown of Piwi4 significantly increased virus titers as measured at 72 hpi and 120 hpi time points. Additionally, silencing of Ago3 also resulted in decreased ZIKV titers at 120 hpi while silencing of Piwi5 or Piwi6 had no major effect on ZIKV production. However, in the case of SFV(3H)-FFLuc luciferase readings at 48 hpi indicated that Ago2 and Piwi4 were antiviral, as luciferase expression was 25 and 8 fold higher respectively (Fig 3D). Similarly, at 72 hpi, where the reads were 12 and 6 fold higher for Ago2 and Piwi4 respectively (Fig 3E). Silencing of Ago3, Piwi5 or Piwi6 had no major effect on luciferase levels, as expected [24] (Fig 3D and 3E).
The effect of the knockdown of RNAi effector proteins on ZIKV replication was also assessed by quantifying the levels of viral genomic RNA present in infected cells. This was achieved by qRT-PCR which indicated that, surprisingly, silencing of Ago2 resulted in lower levels of viral genomic RNA (Fig 3F). Knockdown of Piwi4 was found to enhance virus replication while silencing of Ago3 had an inhibitory effect, suggesting that Ago3 has proviral activity. These data correlated with virus titers (Fig 3). The unexpected failure of Ago2 silencing to permit increased ZIKV replication led us to verify the role of the exo-siRNA pathway through the use of Dcr2 knockout cells. AF5 cells are derived from a single clone of Aag2 cells which possess functional Dcr2 activity and subsequent siRNA production. Conversely, CRISPR-Cas9 technology was used to produce the Dcr2-deficient AF5 derivative cell line, AF319, which does not generate siRNAs, as described previously [41]. A comparison between the replication of ZIKV in both AF5 and AF319 cell lines showed higher virus production in AF319 at 72 and 120 hpi (Fig 4A). Moreover, the results of the qRT-PCR experiment to measure viral genomic RNA levels showed that the lack of Dcr2 benefits virus replication (Fig 4B). This indicated that Dcr2 can recognize and cleave ZIKV-derived dsRNA and the exo-siRNA pathway does play an antiviral role against ZIKV.
Since insect and plant viruses can inhibit RNAi pathways [49], we assessed if ZIKV can interfere with siRNA-based silencing. For this, pIZ-Fluc and pAc1-Rluc plasmids were cotransfected together with dsRNAs targeting firefly luciferase or eGFP (control) into mock-or ZIKV-infected (MOI 1) Aag2 cells at 96 hpi. In mock-infected and ZIKV-infected cells the silencing was 5-fold, which indicates that ZIKV infection did not reduce silencing efficiency (Fig 5A). Similarly, by using siRNAs targeting firefly luciferase or Hygromycin B resistance gene (negative control), we observed the same outcome as siRNAs targeting Fluc functioned as efficiently in mock and ZIKV-infected cells (Fig 5B). These data indicated that ZIKV did not inhibit the siRNA pathway.

Effect of ZIKV capsid (C) protein on antiviral RNAi
A recent report has determined that the expression of the C protein of the related mosquitoborne flavivirus, YFV, can enhance the replication of SINV. It was shown to bind viral dsRNA and protect nucleic acid from cleavage by human dicer [40]. The replication-enhancing effect was found to be common for the C protein of other flaviviruses, including ZIKV. However, no further studies were conducted with these. The sequence similarity between ZIKV and YFV capsid is only 47.2% (identity 24.8%), as determined by the Needleman-Wunsch algorithm.
Despite that we could not detect that ZIKV infection affects dsRNA or siRNA-mediated silencing in a plasmid-based assay (see above, Fig 5), we decided to assess the hypothetical possibility that ZIKV C protein could also, similar to YFV, inhibit the antiviral siRNA response. We started by cloning the full length ZIKV C (amino acid residues 1-122) under the duplicated subgenomic promoter of SFV. Recombinant SFV expressing either tombusvirus siRNAbinding p19 [50] or eGFP were used as positive and negative controls, respectively. In addition, these viruses expressed Renilla luciferase reporter inserted into duplicated nsP2 cleavage sites situated between non-structural nsP3 and nsP4 (Fig 6A). Infection of AF5 or (Dcr2 KO)   AF319 cell lines at low MOI (0.01) was carried out to monitor virus replication and/or spread by determining Renilla luciferase activity. Although the replication of SFV expressing ZIKV C was enhanced in AF5 cells as expected (Fig 6A), surprisingly, enhanced replication was also observed in Dcr2 deficient AF319 cells. This contrasts to results with p19 where its enhancing effect in AF319 cells compared to AF5 cells is lost (Fig 6B). The effect of the expression of ZIKV C was found to be even more prominent and evident at lower a MOI (0.001) (Fig 6C). Both ZIKV C and p19 benefitted SFV replication in AF5 cells, with the former promoting greater virus replication compared to p19. Similarly, in AF319, the effect of Zika C on SFV replication was strong and Rluc levels were again significantly higher compared to viruses expressing p19 or eGFP. These results indicated that the ZIKV C protein could help to overcome inhibitory antiviral processes and enhance SFV replication and/or spread more efficiently than p19, even in the absence of a functioning exo-siRNA pathway.
To determine whether the ZIKV C protein could inhibit the exo-siRNA pathway, Aag2 cells were infected (MOI 1) with recombinant SFV expressing ZIKV C, p19 or eGFP as described above. At 24 hpi cells were transfected with a firefly luciferase (Fluc)-expressing plasmid along with Fluc-specific (or negative control) dsRNAs/siRNAs. When dsRNAs were transfected, determination of Fluc activity at 24 h post-transfection showed that neither the expression of ZIKV C nor p19 affected the silencing efficiency, as compared to eGFP (Fig 6D). Conversely, p19 expression led to reduced siRNA-mediated silencing activity, as expected, as approximately 20% luciferase activity in p19 was observed versus 5% activity in eGFP expressing cells. However, the expression of the ZIKV C protein had only a minor effect with 7% activity of Fluc remaining. Thus, this suggested that the ZIKV C protein has no effect on small RNA pathway-mediated gene silencing, which correlates well with the notion that in ZIKV infected cells no general siRNA pathway inhibition occurs (Fig 5).

Discussion
The production of virus-specific siRNAs and piRNA-like molecules in infected mosquitoes or mosquito-derived cells has been demonstrated for viruses from all of the major arbovirus families: Flaviviridae, Togaviridae and Bunyaviridae [6,7]. Here we demonstrated that ZIKV also induces the production of 21 nt vsiRNAs as well as small RNAs of the size expected for piR-NAs; although these do not display the molecular signature of typical piRNAs. We found that ZIKV-specific vsiRNAs were loaded into Ago2 and thus, in principle, can target viral RNA and mediate degradation. Surprisingly however, the knockdown of Ago2 was not sufficient to increase ZIKV replication and only the knockout of Dcr2 led to an increase in virus replication. This suggests that the exo-siRNA pathway does mediate antiviral activity against this virus; however, its replication may be protected from Ago2 mediated activity through a yet unknown virus-mediated resistance mechanism. This may not be so exceptional as it has previously been observed that the knockdown of Ago2 only resulted in an approximately 2-fold (statistically non-significant) increase in DENV genomic RNA in Aag2 cells [35]. Similarly, Ago2 silencing was only found to be marginally beneficial (20%) for BUNV replication [31], but another orthobunyavirus -Schmallenberg virus-showed a 7-fold increase in viral RNA levels. Perhaps direct dicing of viral dsRNA is more relevant in this context than RISC-mediated antiviral effects. It is also known that flavivirus replication is coupled to viral packaging [51]. Our data could suggest that the packaging of ZIKV RNA occurs rapidly which prevents the recognition of genomic RNA by Ago2 loaded with specific siRNAs. ZIKV is known to heavily modify the ER and form replication factories in mammalian cells [52]. Importantly in the context of RNAi, West Nile Virus (WNV) was shown to become resistant to siRNAs after infection of vertebrate cells following transfection but not electroporation, suggesting that evasion of silencing was taking place [53]. Thus, it could be expected that similar membrane modifications occur in insect cells and such replication factories could shield ZIKV genome from Dcr2 and/or Ago2 detection. This is also relevant from a comparative point of view, as alphavirus replication (despite also occurring in cytoplasmic, membranous structures) is generally increased by Ago2 silencing. How RNAi effectors target viral replication complexes is therefore an intriguing question and localization studies are required to shed light on these processes and differences between virus families. Thus, there are common points and differences between arbovirus families with regards to RNAi responses in mosquito vectors, and the functional relevance of these need to be further assessed in comparative studies.
To our knowledge, no study has shown that vpiRNAs directly inhibit virus replication and the role of PIWI proteins remains enigmatic. Indeed, only the Piwi4 protein has consistently shown antiviral activity against bunyaviruses and alphaviruses [24,31]. However, the antiviral effector mechanism of Piwi4 is unknown. It may interact with piRNA-like small RNAs or 21 nt small RNAs directly, or this association may occur via Piwi4 interaction partners Ago3, Piwi5, Piwi6 and Ago2. Moreover, Piwi4 is not required for the production of virus-specific piRNAs, at least in the case of alphaviruses and DENV [24,35]. Although we could detect the production of ZIKV-specific piRNA-sized small RNAs, these could be mapped to a single site on the viral genomic strand and lacked the characteristic piRNA signature. Those piRNA-like molecules or vpiRNAs were found to be bound by Ago3; although, intriguingly the knockdown of Ago3 decreased ZIKV replication slightly. Similarly in the case of DENV, vpiRNAs mapped to a few discrete sites on the genome and Ago3 was also found to be proviral [35]. In another study, Piwi5, a key protein involved in piRNA production, was found to positively affect BUNV virus [31], which further questions the antiviral role of piRNAs, at least during the acute phase of infection. Intriguingly, a recent study [54] assessed the induction of virusderived small RNAs following infection of Ae. aegypti by ZIKV up to 14 days post-infection and, similarly, virus-specific piRNA-like molecules did not display the ping-pong signature. However, over time piRNAs could be mapped on more locations on the ZIKV genome. The relevance of these observations needs to be experimentally assessed, especially as a study in Aag2 cells infected with mosquito-borne RVFV also showed an increase in vpiRNAs (with the ping-pong signature) especially from the S and M genome segments over time [21]. The role and activity of the piRNA pathway may be delayed compared to the exo-siRNA pathway.
Since flaviviruses encode miRNAs to control host cell gene expression [47], it is possible that ZIKV, or flaviviruses in general, use piRNAs for this purpose, although this needs further experimental verification. Regardless, the direct role of vpiRNAs remains to be elucidated and the function of Piwi4 in this or other pathways remains perplexing, despite its consistent antiviral activity.
The ability of the flavivirus C protein to inhibit small RNA-based antiviral responses is intriguing. The YFV C protein was found to bind both single-stranded RNA and dsRNA, and prevented cleavage of the latter [40] in an in vitro experiment with human Dicer. Furthermore, expression of the C protein from other flaviviruses (WNV, Rio Bravo virus, ZIKV, DENV) enhanced SINV replication. However, the lengths (or specific amino acid residues) of the expressed flavivirus C proteins were not indicated in the published study and so it was not possible to determine if longer, membrane-bound or shorter, cleaved C proteins were mediating the observed effects [55]. As shown here, the full length ZIKV C protein could also enhance the replication of another alphavirus, SFV. However, this also occurred in cells lacking Dcr2, which suggests that the ZIKV C protein inhibits antiviral processes other than the exo-siRNA pathway (Fig 5). YFV is phylogenetically more distant than DENV to ZIKV, and although the general topology of the C protein is conserved among the three viruses [55], the sequence similarity between YFV and ZIKV C proteins is low (24.8%). Thus, it is likely that their proviral mode of action is different. Moreover, in ZIKV-infected cells no reduction in the efficiency of siRNAs or dsRNAs to silence gene expression could be detected, which also suggests that the ZIKV C protein acts in a manner that does not involve the counteraction of Dcr2.
Understanding of virus-vector relations provides further insights into the spread of disease. Recent studies by us and others indicate a complex interplay between arboviruses and mosquito host responses. Viruses have adopted countermeasures against the immune system, which in turn results in selective pressures on the vector. It is highly possible that relative resistance to the antiviral effect of Ago2 is a result of this process. Future work on flaviviruses, such as ZIKV, will need to identify whether any direct targeting of this effector takes place, but also why vpiRNA characteristics are different to other arbovirus families. This is likely to give important clues relating to how different arboviruses are spread by mosquitoes and whether family-specific weaknesses can be exploited.
Supporting information S1 Table. Information regarding small RNA sequencing data. Number of ZIKV-specific sequencing reads obtained by analyzing small RNAs in total cellular RNA samples or in samples that were captured by pulldown of V5-tagged eGFP, Ago2, Ago3, Piwi5, Piwi6 and Piwi4 from Aag2 cells. (DOCX) S2 Table. PCR primers used in the study.