Characterization of dengue virus 3’UTR RNA binding proteins in mosquitoes reveals that AeStaufen reduces subgenomic flaviviral RNA in saliva

Dengue viruses (DENV) are expanding global pathogens that are transmitted through the bite of mosquitoes, mostly Aedes aegypti. As RNA viruses, DENV rely on RNA-binding proteins (RBPs) to complete their life cycle. Alternatively, RBPs can act as restriction factors that prevent DENV multiplication. While the importance of RBPs is well-supported in humans, there is a dearth of information about their influence on DENV transmission by mosquitoes. Such knowledge could be harnessed to design novel, effective interventions against DENV. Here, we successfully adapted RNA-affinity chromatography coupled with mass spectrometry – a technique initially developed in mammalian cells – to identify RBPs in Ae. aegypti cells. We identified fourteen RBPs interacting with DENV serotype 2 3’UTR, which is involved in the viral multiplication and produces subgenomic flaviviral RNA (sfRNA). We validated the RNA affinity results for two RBPs by confirming that AePur binds the 3’UTR, whereas AeStaufen interacts with both 3’UTR and sfRNA. Using in vivo functional evaluation, we determined that RBPs like AeRan, AeExoRNase, and AeRNase have pro-viral functions, whereas AeGTPase, AeAtu, and AePur have anti-viral functions in mosquitoes. Furthermore, we showed that human and mosquito Pur homologs have a shared affinity to DENV2 RNA, although the anti-viral effect is specific to the mosquito protein. Importantly, we revealed that AeStaufen mediates a reduction of gRNA and sfRNA copies in several mosquito tissues, including the salivary glands and that AeStaufen-mediated sfRNA reduction diminishes the concentration of transmission-enhancing sfRNA in saliva, thereby revealing AeStaufen’s role in DENV transmission. By characterizing the first RBPs that bind to DENV2 3’UTR in mosquitoes, our study unravels new pro- and anti-viral targets for the design of novel therapeutic interventions as well as provides foundation for studying the role of RBPs in virus-vector interactions. Author abstract Dengue viruses are important human pathogens transmitted by mosquitoes. Currently, there are no effective control measures for dengue. The RNA-binding proteins (RBPs) in mosquitoes, which bind to the dengue virus genome to regulate viral multiplication, could serve as new targets for developing therapeutic interventions. In this study, we pioneered the use of RNA-affinity chromatography – a technique that identifies proteins binding to specific RNA sequences – in mosquito cells. This led to the detection of fourteen RBPs that bind to the 3’UTR of dengue virus serotype 2. We validated our results using immunoprecipitation method. Furthermore, we demonstrated that 6 of the 14 RBPs influence viral multiplication in mosquitoes. Among these six RBPs, we showed that the AePur mosquito and human homologs share an affinity to dengue virus RNA, whereas the anti-viral function is specific to the mosquito homolog. Importantly, we revealed that AeStaufen mediates a reduction of genomic and subgenomic flaviviral RNAs in multiple mosquito tissues. We also showed that the reduction of subgenomic flaviviral RNA in salivary glands diminishes the secretion of salivary subgenomic RNA, which facilitates infection at the bite site, thereby unveiling the function of AeStaufen in the virus transmission. By characterizing the first mosquito RBPs that bind to dengue virus genome, our study paves the way for leveraging these proteins as potential targets to block virus transmission.


126
Fourteen A. aegypti proteins interact with DENV2 3'UTR 127 To identify proteins interacting with DENV2 3'UTR, we adapted our previously-described 128 approach of RNA-affinity chromatography coupled with mass spectrometry (MS) using 129 mosquito cell lysates (Fig. 1A). We generated a construct containing the T7 promotor, a 130 tobramycin adapter sequence, and DENV2 3'UTR from the NGC strain. For control, we 131 replaced the 3'UTR sequence with a size-matched NS2A gene sequence from DENV2, as 132 previously described [18]. The constructs were transcribed in vitro and the RNA sequences 133 were allowed to form secondary structures prior to being incubated with tobramycin- responses [29]. Eventually, we eluted RNA-bound proteins by adding an excess of tobramycin 137 and quantified them using quantitative MS. 138 Among the 385 proteins detected in the eluates (Table S1), we defined DENV2 3'UTR-139 interacting proteins as those enriched more than 1.5-fold over the NS2A RNA control, with an 140 adjusted p-value < 0.1. We have identified 14 proteins that met these criteria ( Fig. 1B; Table   141 1), which include (in decreasing order of affinity with the viral 3'UTR): AeMaleless, an ATP-  of this protein could be conserved in both mosquitoes and humans. In the case of AeStaufen, 160 its putative function in RNA decay [32] could modulate the quantity of viral RNA fragments 161 [24]. For these reasons, we validated our RNA-affinity chromatography results with RNA 162 immunoprecipitation (RIP) for AePur and AeStaufen. Pur proteins are a family of single-163 stranded nucleic acid-binding proteins that are highly conserved from bacteria through humans 164 [38]. Accordingly, we were able to use a commercially available antibody that targets human 165 PurB, to pull down AePur from lysates of orally-infected mosquitoes ( Fig. 2A). Using primers 166 that target the envelope gene located in the single open reading frame of DENV gRNA, we 167 observed an enrichment for gRNA (4.59 ± 0.43 fold change) in AePur immunoprecipitates as 168 compared to the IgG control (Fig. 2B). To determine whether AePur binds to gRNA 3'UTR or 169 to sfRNA, both of which have the same sequence, we also quantified sfRNA enrichment as 170 previously described [27]. SfRNA was not enriched in AePur immunoprecipitates as compared 171 to the IgG control (Fig. 2B). Together with the RNA-affinity chromatography results, these 172 data validate AePur interaction with the 3'UTR sequence in DENV2 gRNA. 173 We next tried to confirm AeStaufen binding to DENV2 3'UTR in infected mosquitoes, 174 using the same approach as for AePur. However, attempts to immuno-precipitate AeStaufen 175 using commercially available antibodies developed against its human homolog were 176 unsuccessful. Therefore, we designed an alternative approach that uses C6/36 cells derived 177 from Aedes albopictus mosquitoes, which are highly susceptible to DENV infection [39].  (Fig. 2D). As done in AePur immunoprecipitation, we also determined the 184 affinity of AeStaufen-V5 to sfRNA and observed a lower but significant enrichment (2.69 ± 185 0.12) for sfRNA in AeStaufen-V5 immuno-precipitates. Together, these results suggest that 186 AeStaufen has a high affinity to gRNA 3'UTR and also interacts with sfRNA. Overall, using 187 RIP experiments, we have validated the 3'UTR-protein interactions identified with the RNA-188 affinity chromatography and determined the affinity of AePur and AeStaufen for DENV2  To test whether the mosquito proteins that bind DENV2 3'UTR influence mosquito infection, 193 we performed in vivo dsRNA-mediated silencing for each of the 14 proteins. As control, we blood meal containing 10 6 DENV pfu/ml, which is within the range of inoculum concentration 197 observed in patient serums [40]. The blood feeding rate was lower only in mosquitoes whose 198 AeMaleless and AePur genes were silenced, while other silencing had no effect (Fig. S3A).

199
Seven days post oral infection, we quantified DENV titers in whole mosquitoes using  The infection rate provides information about the ability of the virus to initiate infection, while 206 infection intensity is a measure of virus multiplication. Nonetheless, these two biological 207 parameters are interdependent; for instance, reduced viral multiplication (i.e. infection 208 intensity) may lead to viral elimination that would then lower infection rate.

209
In terms of infection rate, we observed that silencing of AeRan and AeExoRNase reduced   We next tested if human Pur proteins influenced DENV2 infection. We separately 236 depleted either HsPurA or HsPurB expression using siRNA-mediated silencing in human cells.

237
To prevent untargeted silencing effects, we used three different siRNA sequences for each of 238 the two Pur proteins and observed specific depletion of either HsPurA or HsPurB (Fig. 4C).

239
The depleted cells were infected with DENV2 and viral titer was quantified using plaque 240 forming unit (PFU) assay 24h later. There was no effect of either HsPURA or HsPURB 241 silencing on viral titer as compared to non-transfected (NT) and siRNA-transfected controls 242 (siNC) (Fig. 4D). To rule out the possibility that background levels of the HsPur proteins and/or 243 functional redundancy between HsPurA and HsPurB could be responsible for unaffected 244 DENV titers, we generated cells devoid of both HsPurA and HsPurB via CRISPR/Cas9 editing.

245
To prevent untargeted effects of the knock-out approach, we produced four different cell lines 246 in which both HsPurA and HsPurB expressions were abrogated (Fig. 4E). Similar to the 247 silencing approach (Fig. 4D), DENV titers remained unaffected in the absence of both HsPur   (Table S2). Similar to FFU quantification described above, we calculated infection rate as the silencing, suggesting that AeStaufen does not influence infection onset (Fig. 5B). However, 275 infection intensity increased 3.02-fold and 5.67-fold upon AeStaufen depletion in the carcass 276 and salivary glands, respectively (Fig. 5B). It is interesting to note that these tissues exhibited 277 highest AeStaufen expression (Fig. 5A). In the midgut, AeStaufen depletion led to a 3.61-fold 278 increase in infection intensity (Fig. 5B), although the difference was not statistically significant 279 (p = 0.12, as determined by t-test). These results reveal that AeStaufen mediates reduction in 280 DENV gRNA copies in multiple organs of mosquitoes.

281
Since AeStaufen also interacts with DENV2 sfRNA ( AeStaufen mediates decrease in both gRNA and sfRNA copies in all tissues, but that its effect 292 on sfRNA is more pronounced in the midgut and salivary glands. 293 We have previously shown that DENV2 strains producing higher sfRNA copies in the  (Table S2). Both infection rate and infection intensity were unaffected by AeStaufen depletion 300 in the salivary glands and saliva (Fig. 5E, F). Together with the lack of effect on infection in 301 AeStaufen-depleted mosquitoes (Fig. 3), these results indicate that AeStaufen does not alter 302 the production of infectious viral particles. in AeStaufen-depleted mosquito saliva at 10 dpi. gRNA detection rate and gRNA copies per 310 infected saliva were very similar in AeStaufen-depleted and control mosquitoes (Fig. 6A).

311
Inversely, sfRNA:gRNA ratio increased 1.89-fold upon AeStaufen depletion, although sfRNA 312 detection rate was unaltered (Fig. 6B). 313 We next tested whether the increase in sfRNA:gRNA ratio in the saliva was caused by 314 AeStaufen-mediated sfRNA degradation in the salivary glands. To infect salivary glands 315 without going through midgut infection, we inoculated AeStaufen-depleted mosquitoes and 316 quantified gRNA and sfRNA in salivary glands at 7 days post inoculation. gRNA detection 317 rate (p = 0.10, as determined by Z-test) and gRNA copies per infected salivary glands were not 318 altered (Fig. 6C). However, similarly to what we observed in salivary glands from orally-319 infected mosquitoes, sfRNA:gRNA ratio was increased 1.77-fold in salivary glands from 16 320 inoculated mosquitoes (Fig. 6D). Altogether, these results indicate that AeStaufen-mediated 321 reduction of sfRNA quantity in salivary glands influences the ratio of sfRNA:gRNA in saliva.  Interestingly, the binding affinity to DENV RNA is conserved in the human homologs of 341 eight of the 14 RBPs we identified in mosquitoes (Table 1). A Comprehensive identification 342 of RNA-binding proteins by mass spectrometry (ChIRP-MS) study in human cells found that 17 343 DHX9, DHX29, DHX36, and DHX57 (homologs of AeMaleless) and ELAVL1 (AeSex lethal 344 homolog) interact with DENV2 and Zika virus genomes [11]. However, an RNA-affinity 345 chromatography approach (similar to the one used in this study) found that DHX9 has less 346 affinity to the 3'UTR of DENV1-4 than to a control RNA fragment, while DHX36 only 347 marginally (1.30-fold) interacts with DENV2 3'UTR [43]. These contrasting results for 348 genomic RNA and 3'UTR binding suggest that the RBPs preferentially interact with sequences that the virus has evolved to exploit the biological similarities between its two hosts. 357 We have identified six RBPs that influence DENV infection in mosquitoes. We found that  AeStaufen. To inform about a proportional relationship between sfRNA and gRNA quantities, 415 we calculated sfRNA:gRNA ratio. In the carcass, the ratio was unchanged by AeStaufen 416 depletion, suggesting that a higher gRNA copies led to increased production of sfRNA.     They were maintained at 28°C and 50% relative humidity in a 12h:12h light: dark cycle. AeStaufen. Blood viral titers were validated by plaque assay as described above. Mosquitoes 506 were let to blood-feed for 1.5 h in the Hemotek membrane feeder system (Discovery 507 Workshops) covered with porcine intestine (sausage casing). Fully engorged mosquitoes were 508 selected and maintained in similar conditions as for the colony, with ad libitum access to water 509 and 10 % sugar solution. Blood-feeding rate was calculated by dividing the number of 510 engorged mosquitoes by the total number of mosquitoes that were offered the blood meals.   The DNA templates used to generate dsRNA against the 14 3'UTR-interacting proteins were 588 synthesized using T7-tagged primer pairs as detailed in