A limited role for autophagy during arbovirus infection of mosquito cells

Macroautophagy is an evolutionarily conserved cellular process critical for maintaining cellular homeostasis. It can additionally function as an innate immune response to viral infection as has been demonstrated for a number of arthropod-borne (arbo-) viruses. Arboviruses are maintained in a transmission cycle between vertebrate hosts and invertebrate vectors yet the majority of studies assessing autophagy-arbovirus interactions have been limited to the mammalian host. Therefore we evaluated the role of autophagy during arbovirus infection of the invertebrate vector using the tractable Aag2 Aedes aegypti mosquito cell culture system. Our data demonstrates that autophagy is significantly induced in mosquito cells upon infection with two flaviviruses, dengue virus (DENV) and Zika virus (ZIKV), as well as an unrelated mosquito-borne virus, chikungunya virus (CHIKV; Togaviridae). While assessing the role of autophagy during arbovirus infection, we observed a somewhat paradoxical outcome. Both induction and suppression of autophagy via torin 1 and spautin-1, respectively, resulted in increased viral titers for all three viruses, yet suppression of autophagy-related genes had no effect. Interestingly, chemical modulators of autophagy had either no effect or opposite effects in another widely used mosquito cell line, C6/36 Aedes albopictus cells. Together, our data reveals a limited role for autophagy during arbovirus infection of mosquito cells. Further, our findings suggest that commonly used chemical modulators of autophagy alter mosquito cells in such a way as to promote viral replication; however, it is unclear if this occurs directly through autophagic manipulation or other means. Author Summary Arthropod-borne (arbo) viruses, specifically those transmitted by Aedes aegypti mosquitoes, cause significant morbidity and mortality and pose a continued public health threat worldwide. Many of these viruses lack vaccines or therapeutics and current mosquito control strategies are underperforming. For these reasons, identifying vulnerabilities within the transmission cycle that can be targeted will be critical to the development of novel control interventions. Autophagy is a highly conserved cellular pathway and previous studies manipulating this pathway have shown promise in minimizing viral infections in mammalian hosts. In this study we examined arbovirus-autophagy interactions within vector mosquitoes. The goal was to elucidate the role of autophagy during infection of mosquitoes in hopes of identifying critical interactions that can be targeted by novel approaches to block infection of and transmission by vector mosquitoes.

Author Summary: 1 Arthropod-borne (arbo) viruses, specifically those transmitted by Aedes aegypti mosquitoes, 2 cause significant morbidity and mortality and pose a continued public health threat worldwide. 3 Many of these viruses lack vaccines or therapeutics and current mosquito control strategies are 4 underperforming. For these reasons, identifying vulnerabilities within the transmission cycle that 5 can be targeted will be critical to the development of novel control interventions. Autophagy is a 6 highly conserved cellular pathway and previous studies manipulating this pathway have shown 7 promise in minimizing viral infections in mammalian hosts. In this study we examined 8 arbovirus-autophagy interactions within vector mosquitoes. The goal was to elucidate the role of 9 autophagy during infection of mosquitoes in hopes of identifying critical interactions that can be 10 targeted by novel approaches to block infection of and transmission by vector mosquitoes.  However, as insecticide resistance is becoming more commonplace it is critical that novel 2 approaches for preventing and treating arboviral diseases, targeting either vector or host, be 3 developed. 4 Macroautophagy (herein referred to as autophagy) is an essential cellular process 5 required for maintaining homeostasis and plays a crucial role in development, cell differentiation 6 and immunity (1, 2). Consequently, abnormal autophagic activity has been linked to a number of 7 pathologies, including cancer and neurodegenerative diseases (3). This evolutionarily conserved 8 pathway mediates the degradation and recycling of cellular components. Upon induction, protein 9 aggregates, damaged or dysfunctional organelles and foreign bodies are sequestered into cup-10 shaped double-membrane vesicles termed phagophores or isolation membranes (4).

11
Appropriation of cargo can occur in a non-selective manner known as bulk autophagy where 12 cytoplasmic material is indiscriminately engulfed or through selective autophagy, which 13 specifically targets poly-ubiquitin tagged proteins (5). Through a series of kinase signaling 14 cascades and recruitment events the phagophore elongates, eventually encompassing the cargo 15 and forming an autophagosome (4). Subsequently, the mature autophagosomes fuse with 16 lysosomes resulting in the acidification of the newly formed autophagolysosome and degradation 17 of the cargo (4). Targeting aspects of the autophagy pathway has become a promising strategy 18 for combatting a number of diseases (6, 7). 19 Numerous stimuli have the ability to activate autophagy including innate immunity and 20 cellular stress. It is therefore not surprising that viral infections often induce an autophagic titers and improves clinical outcomes in mice. This process is mediated by the autophagy cargo 4 receptor p62, which recognizes poly-ubiquitin tagged viral capsid proteins and transports them to 5 maturing autophagosomes for degradation (10,13,(15)(16)(17)(18). In comparison, flavivirus-autophagy 6 interactions are much more nuanced (19). For instance, during DENV infection, p62 actively 7 targets DENV proteins for degradation and replication complexes associated with the ER 8 (endoplasmic reticulum) are targeted by reticulophagy; however, DENV benefits from an 9 energetically favorable environment as a result of virus-induced lipophagy and efficient 10 processing of mature virions (9,(19)(20)(21). While it is clear that arboviruses closely interact with 11 autophagy during infection, these studies have been limited to either non-vector organisms or conversion of Atg8 to Atg8-PE. Further, we found that chemical inhibition and induction of 20 autophagy both increased viral titers, while suppression of autophagy related genes had no effect.

21
Interestingly, chemical modulators had minimal effects on DENV titers in another commonly 22 used mosquito cell line, Aedes albopictus C6/36 cells, and spautin-1 inhibition of autophagy 23 significantly decreased in DENV titers in mammalian cells as previously reported (21). Together, 1 these data reveal a limited role for autophagy during DENV, ZIKV and CHIKV infection of 2 mosquito cells and highlight differences in autophagy-virus interactions between cell culture 3 systems. Further, our data suggest that outcomes associated with commonly used chemical 4 modulators of autophagy are cell-dependent and may result from cell-specific interactions with 5 the chemicals.  To assess the effect of pharmacological modulators on autophagy, the following drugs 4 were used in cell culture infections of C6/36, Aag2, and BHK21 cells: torin 1, an autophagy 5 inducer that works by inhibiting mTOR kinase, and spautin 1 (Tocris Bioscience), which inhibits 6 autophagy by interfering with USP10 and USP13, two ubiquitin-specific peptidases, resulting in 7 the degradation of class III phosphatidylinositol-3 kinase complexes (23). Bafilomycin A1 8 (InvivoGen), an inhibitor of V-ATPase, was also used in inhibition of late stage autophagy in 9 infectious and non-infectious immunoblotting of Aag2 cells. All drugs were prepared in dimethyl 10 sulfoxide (DMSO) and stored at -20°C.

11
Immunoblots 12 Two days after Aag2 cells were seeded at 1.5x10 6 cells/well in a 12-well plate, cells were 13 treated with 1% DMSO, 1 µM bafilomycin A1, 1 µM torin-1 or 10 µM spautin-1. Cells were 14 harvested 24 hours later using RIPA buffer. Total protein in cell lysates was quantified using a 15 Pierce BCA Protein Assay kit (ThermoFisher) on a microplate reader. An equal amount of 16 protein from each sample was separated on a 16% Tris-glycine gel and transferred to a 17 polyvinylidene difluoride (PVDF) membrane. Autophagy was detected using a rabbit anti-Atg8 18 antibody at a 1:8,000 dilution in PBS + 5% BSA + 1% Tween. Polyclonal anti-Atg8 antibody 19 was generated by ProSci Inc. as previously described (24). The blot was then probed with 20 horseradish peroxidase (HRP) goat anti-rabbit antibody at a 1:20,000 dilution. Actin was 21 detected using a rabbit anti-β-Actin antibody (Abcam) in PBS + 5% BSA + 1% Tween at a 22 1:8,000 dilution. The secondary probe was performed using the same secondary antibody as 1 above. Membranes were developed in SuperSignal West Pico chemiluminescent substrate 2 (Thermo Fisher) for 5 minutes at room temperature. Band intensities were quantified using 3 ImageJ.

4
The effect of virus infection in combination with pharmacological modulators on 5 autophagy was also assayed via western blot. Aag2 cells were prepared as above and infected Aedes aegypti Atg5, Atg14, Atg8 and luciferase genes were amplified using primer sets 16 containing the T7 promotor (Table S1). Amplicons were PCR purified and a T7 Megascript kit 17 was used to synthesize dsRNA molecules for transfection per the manufacturer's 18 recommendations (Ambion). Two days prior to transfection, Aag2 cells were seed at 1.5x10 6 19 cells/well in 12-well plates. 1 µg/well of dsRNA was combined with OPTI-MEM and 20 Lipofectamine 2000 transfection reagent (Invitrogen) and allowed to incubate at room 21 temperature for 20 minutes. The media on the Aag2 cells was discarded and replaced with 1 mL 22 of Opti-MEM. The lipid-nucleic complexes were then added to the wells containing OPTI-MEM and were incubated at 28°C for four hours, after which the media was discarded and again 1 replaced with the standard 10% FBS media. Cells were harvested 48 hours after transfection by 2 manually scraping the cell monolayer using a 1 ml pipette tip. The cells were pelleted and re-3 suspended in QVL Lysis buffer from an Omega Bio-tek Viral RNA Extraction kit or RIPA 4 buffer. Total RNA was then extracted following the manufacturer's instructions. Extractions 5 were DNase treated and purified via phenol/chloroform extraction.

6
Silencing efficiency of Atg5 and Atg14 was determined through RT-qPCR with 7 Universal SYBR Green master mix (Bio-Rad), using luciferase samples as the non-targeting 8 control group and GAPDH as a reference gene (Table S1). Silencing efficiency of Atg8 was 9 determined by Western blotting. Two days after dsAtg8 was transfected into Aag2 cells, cells 10 were infected with DENV as above. Cells were harvested in RIPA buffer 24 hpi.

11
Immunoblotting was performed as previously described.

17
Aag2 cells were seeded and transfected with dsLuciferase, dsAtg5, and dsAtg14 as previously  The effect of cell line on DENV virus titers was determined by following the above 21 protocol in a mammalian cell line (BHK21 c15) and a mosquito cell line with a dysfunctional 22 RNA interference pathway (C6/36). C6/36 cells were seeded at 1.5x10 6 cells/well and BHK cells 1 were seeded at 5x10 5 cells/ well in 12 well plates. The following day, both cell lines were 2 infected with DENV and treated with 1% DMSO, 1 µM torin-1 or 10 µM spautin-1. Titers were 3 harvested at 48 hpi. Each experiment was completed in triplicate with two biological replicates 4 per experimental replicate.

5
For titration of DENV and ZIKV samples, BHK21 c15 cells were grown to a confluent 6 monolayer in 12-well plates and infected with 10-fold serial dilutions of virus for 1 hour at room 7 temperature and overlaid with a mixture of the standard media and 5% methyl cellulose. After 8 incubation at 37°C for four days, the cells were fixed with 7.4% formaldehyde and stained with 9 gentian violet. Viral titers were determined by counting plaques. For titration of CHIKV, the 10 above protocol was followed in Vero cells.

12
In order to track autophagy during virus infection we quantified Atg8+ puncta by 13 confocal microscopy. While we possess a polyclonal anti-Atg8 antibody, it did not produce 14 reliable and consistent results. Therefore, we generated an expression vector to provide EGFP 15 tagged Atg8 in trans. This was achieved by cloning the full length Ae. aegypti Atg8 gene 16 (AAEL007162) into the pIEx-EGFP vector 3' to EGFP by Gibson Assembly. The generation of 17 pIEx-EGFP has been described previously (25). Aag2 cells were seeded on coverslips at a 18 density of 1.5x10 6 cells/well in a 12 well plate and 24 hrs. later transfected with 1.5 µg/ well of Immunoblot fold-change differences and puncta abundance were analyzed by one-way ANOVA 9 with a Sidak's multiple comparisons test. All viral titer data was analyzed by one-way ANOVA 10 with a Dunnett's multiple comparisons test. Suppression of Atg5 and Atg14 was calculated with 11 a two-tailed t-test. alone had no discernible effects on the ratio of Atg8-PE to β-actin compared to the DMSO 21 control cells as determined by immunoblot (Fig. S1). However, the combination of torin-1 and bafilomycin A1 resulted in a significant increase in Atg8-PE to β-actin ratios. By inhibiting 1 autophagolysosome acidification we 2 were able to observe signs of 3 autophagic activity suggesting that 4 Aag2 cells have a high rate of 5 autophagic flux and providing 6 evidence that torin-1 and bafilomycin A1 are functioning as expected with regards to autophagy.

7
As a result of the high autophagic flux, the addition of spautin-1 had no effect on baseline 8 autophagy levels. However, when added to cells treated with torin-1 and bafilomycin A1, 9 spautin-1 reduced the ratio of Atg8-PE to β-actin thereby validating its ability to inhibit 10 autophagy in Aag2 cells (Fig. S1).  As previously determined, our inability to observe autophagy induction may be due to the high 19 rate of autophagic flux associated with these cells. Consequently, addition of bafilomycin A1 to 20 the DENV-2 infections significantly increased Atg8-PE to β-actin ratios (Fig. 1A-B). Similar 21 increases in the ratio could be observed for CHIKV infections, although these results were not 22 significant ( Fig. 3A-B). Interestingly this did not increase ratios during ZIKV infection compared to the bafilomycin A1 control group ( Fig. 2A-B). To further validate these findings, Aag2 cells 1 were transfected with a plasmid constitutively expressing Atg8-EGFP and infected in the 2 presence or absence of bafilomycin A1. As with the immunoblot, DENV infections in the 3 absence of bafilomycin A1 resulted in a marginal increase in autophagic activity, while the 4 addition of bafilomycin A1 significantly increased Atg8+ puncta/ cell compared to the controls 5 ( Fig. 1C-D). Similarly, increases in Atg8+ puncta/ cell were observed during ZIKV and CHIKV  (Fig. 4). Interestingly, induction of autophagy 1 with torin-1 also resulted in significantly increased titers for all three viruses at all time points 2 (Fig. 4). Further, blockage of V-ATPase activity with bafilomycin A1 resulting in an increase in 3 autophagolysosomal pH levels significantly increased DENV-2 and ZIKV titers at all time 4 points, but not those of CHIKV (Fig. 4). Due to the unexpected and paradoxical findings, we 5 examined the effects of autophagy through suppression of three genes within the autophagy 6 pathway: Atg5 (required for phagophore elongation), Atg8 (required for phagophore elongation 7 and membrane curvature) and Atg14 (essential autophagy regulator). Treatment of Aag2 cells 8 with gene specific dsRNA resulted in significant suppression of Atg5 and Atg14 as determined 9 by RT-qPCR as well as depletion of Atg8 as determined by immunoblot (Fig. S2). Suppression  (21). In an effort to recapitulate these findings we treated DENV-2 2 infected BHK-21 cells with spautin-1 and observed a significant reduction in DENV-2 titers 3 (Fig. 6). Induction of autophagy with torin-1 had no effect (Fig. 6). Next we wanted to determine 4 if our results in Aag2 cells were similar in another commonly used mosquito cell line, C6/36 5 Aedes albopictus cells. As before, treatment of Aag2 cells with both torin-1 and spautin-1 6 resulted in significantly increased DENV-2 titers; however, we observed a slight, but significant 7 decrease in titers upon torin-1 treatment and no effect upon spautin-1 treatment in C6/36 cells 8 (Fig. 7). This data suggests that these commonly used chemical compounds can have profoundly 9 different effects on host cell-virus interactions.  Comparing our results to what has been reported in mammalian systems reveals that at 7 least for DENV and ZIKV the general patterns are consistent. The increase in titers upon TOR-8 inhibition with torin-1 suggests that both DENV-2 and ZIKV benefit from downstream pathways 9 regulated by TOR, one of which is autophagy. We observed increased flavivirus titers in Aag2 10 cells upon torin-1 treatment suggesting that flaviviruses benefit from an autophagic state and is 11 consistent with what has been found in mammalian cells (9,21). However, increased titers were 12 also observed upon suppression of autophagy by spautin-1 suggesting an antiviral role. While 13 these outcomes seem contradictory, they are consistent to the current model for DENV-14 autophagy interactions. In mammalian cells, DENV and DENV replication complexes are 15 targeted for degradation by autophagy while simultaneously benefitting from more efficient 16 mature virion processing and increased energy availability as a result of virus-induced lipophagy 17 (9,(19)(20)(21). While the flavivirus data was consistent with what has been observed in mammalian 18 systems, the CHIKV data was not. Numerous studies have found that autophagy functions in an 19 antiviral manner during infection with members of the family Togaviridae, specifically 20 alphaviruses (10,13,15,17). While inhibition of autophagy did result in increased CHIKV titers 21 as would be expected, induction also increased titers. This discrepancy may reflect a difference 22 in the role of autophagy during CHIKV infection of mosquito and mammalian cells.

23
This study provides an important first-step in understanding autophagy-arbovirus 1 interactions within mosquitoes and highlights differences between the mammalian and mosquito 2 autophagic response to infection. However, future work assessing these interactions within 3 mosquitoes will be needed due to the limitations of our cell culture system. As previously 4 mentioned, the origin of mosquito cell culture is unknown and therefore, it is difficult to   AaeAtg5  T7-Forward  TAA TAC GAC TCA CTA TAG GGT CCG ATG AAA CCG ATG TC  XM_001661191   T7-Reverse  TAA TAC GAC TCA CTA TAG GGC ATC GAA CTG GAA TGT G   RT-qPCR-Forward  AGT TCG ATG TTA TGC CGA GG   RT-qPCR-Reverse  GAT AGC TGA GGT GTT CCG AG   AaeAtg14  T7-Forward  TAA TAC GAC TCA CTA TAG GGT TTC TCG  targeting control group and GAPDH as a reference gene. Data was analyzed with a two-tailed t-7 test. C) Silencing efficiency of Atg8 was determined by immunoblot. Two days after dsAtg8 was 8 transfected into Aag2 cells, cells were infected with DENV as above. Cells were harvested in 9 RIPA buffer 24 hpi. Immunoblotting was performed as previously described.