A non-enveloped arbovirus released in lysosome-derived extracellular vesicles induces super-infection exclusion

Recent developments on extracellular vesicles (EVs) containing multiple virus particles challenge the rigid definition of non-enveloped viruses. However, how non-enveloped viruses hijack cell machinery to promote non-lytic release in EVs, and their functional roles, remain to be clarified. Here we used Bluetongue virus (BTV) as a model of a non-enveloped arthropod-borne virus and discovered that the majority of viruses are released in EVs. Based on the cellular proteins detected in these EVs, and use of inhibitors targeting the cellular degradation process, we demonstrated that these extracellular vesicles are derived from secretory lysosomes, in which the acidic pH is neutralized upon the infection. Moreover, we report that secreted EVs are more efficient than free-viruses for initiating infections, but that they trigger super-infection exclusion that only free-viruses can overcome.

We also confirmed this phenomenon in vivo in bovine blood samples positive for BTV-8 (S1B Fig)  139 collected during BTV outbreaks in France in 2019 (Fig 2A). Viral titrations of the EVs and free virus 140 particles fractions separated by differential centrifugation revealed that virus particles were mainly 141 contained within EVs (Fig 2B). However, no significant difference in infectivity between free virus To determine the origin of EVs containing BTV virus particles, we tested the presence of specific 155 cellular markers. Among the different markers of EVs assessed, we determined the presence of 156 Annexin A2, the tumour susceptibility gene 101 (TSG101), LC3-I and LC3-II, and the lysosomal 157 associated protein LAMP1 by western blot (Fig 3B and S2A Fig). Note that TSG101 is a MVBs 158 marker and LC3B an autophagy related protein. To identify the cellular mechanisms involved in virus 159 packaging in EVs, we first investigated the potential role of autophagy, using the 3-methyladenin 160 (3-MA) that inhibits class III PI3K activity. We observed that 3-MA failed to significantly decrease the secretion of infectious EVs from sheep cells infected at a MOI of 10 ( Fig 3B), suggesting that the 162 induction of autophagy is not essential for the secretion of infectious EVs. However, inhibition of 163 autophagosome -lysosome fusion with chloroquine (CQ), led to a significant reduction of infectious 164 EVs released as compared to the control (Fig 3B), indicating that the late steps of autophagy are 165 necessary for infectious EVs. In addition, inhibition of MVBs regulator protein HSP90 using 166 geldanamycin in BTV-infected cells (MOI=10) also led to a significant reduction of infectivity 167 measured in the EVs fraction, as compared to the control, indicating a possible role for MVBs in the 168 release of infectious EVs. In contrast, GW4869, a drug that inhibits the release of exosomes (small 169 vesicles ~200nm) derived from MVBs, did not affect the secretion levels of EVs containing BTV 170 ( Fig 3B). As expected, except for GW4869, these drugs also had an inhibitory effect on the virus 171 However, these EVs cannot be exosomes, based on the results obtained with the GW4869 inhibitor. 175 Following recent description that autophagy and MVBs share common molecular machinery and 176 organelles such as amphisomes [18], we asked if EVs could be derived from secreted lysosomes after 177 the fusion with amphisomes. Therefore, we purified the EVs using an isopycnic ultracentrifugation 178 and found that most of the infectivity was associated with the low-density fractions (~1.05 to 1.08 179 g/mL), along with the cathepsin enzymes, a lysosomal marker ( Fig 3C). Interestingly, the ratio of cell 180 surface LAMP1 to cytosolic LAMP1 was also significantly higher in infected cells (Fig 3D), 181 indicating an increase of LAMP1 located at the plasma membrane in infected cells. The presence of 182 viral proteins was then investigated in the lysosomes of BTV infected cells. 183 Using super resolution microscopy, we observed that in infected cells only, outer capsid protein VP5 184 as well as LAMP1 co-localised with the MVBs marker HSP90 (Fig 3E), indicating a fusion between 185 MVBs and the autophagy compartments. Supporting these results, we also observed the co-186 localisation of LAMP1 and VP5 with TSG101, another MVBs marker (S2D Fig),   NS3 is the only viral membrane protein and it is likely to be involved in the mechanism of EV 210 secretion. Therefore, we tested the presence of virus in EVs ( Fig 4A) following infection with 211 available NS3 mutants [14,21,22]. Two of these mutants impair the synthesis of the two 212 NS3 isoforms, respectively NS3 (mutant NS3M1) and NS3A (mutant NS3M14), while another with a 213 modified late domain motif (NS3GAAP) cannot bind TSG101 and one, unable to bind the outer capsid 214 protein VP2 (NS3stop211). Of the four mutant viruses, only BTV NS3M14 significantly reduced virus 215 particle release in EVs (Fig 4A). To understand this phenotype, we analysed the localisation of the 216 mutant virus particles in infected cells and observed that the outer capsid protein VP5, and NS3, were 217 still co-localised with the lysosomal markers LAMP1 and cathepsin (Fig 4B, S3 Fig), suggesting that 218 the absence of BTV NS3M14 in EVs was not due to a defect in virus trafficking. We then examined 219 lysosomes in cells infected with wild type BTV (BTVWT) and non-infected cells expressing a 220 LAMP1-GFP fusion protein in the presence of a fluorescent lysotracker specific to acidic organelles 221 ( Fig 4C). Confocal microscopy revealed a significant decrease of lysotracker fluorescence intensity 222 in the lysosomes of BTVWT infected cells, when compared to non-infected cells, suggesting that 223 BTVWT neutralises the acidic pH of lysosomes ( Fig 4D). The number of lysosomes detected in 224 infected cells was also significantly higher than in non-infected cells (Fig 3E and 3G). In contrast, 225 cells infected with BTV NS3M14 showed comparable levels of lysotracker fluorescence to non-226 infected cells, suggesting that BTV NS3M14 was unable to counter lysosomes acidification. To provide 227 further confirmation, cells infected with BTV NS3M14 and non-infected cells were examined in the 228 presence of the lysosomal V-ATPase inhibitory drug bafilomycin A1 (Fig 3E and 3G). Interestingly, 229 bafilomycin A1 significantly increased the levels of BTV NS3M14 in EVs (Fig 4H), while not affecting 230 the intracellular replication of BTV NS3M14, whereas the drug had a strong inhibitory effect on the 231 replication of BTVWT (S2B Fig). Consistent with this model, we observed that the intracellular 232 calcium levels of cells infected with BTVWT were significantly higher than the calcium levels in BTV 233 NS3M14 and non-infected cells (Fig 4I and 4J). Together, our data strongly support that newly 234 synthesised BTV particles are released via EVs derived from secretory lysosomes after pH 235 neutralisation.

306
We next considered if differences between EVs and free virus particles are due to cell defence 307 mechanisms, such as super-infection exclusion. Although not always understood, this cell defence 308 mechanism in which infected cells become resistant to a second infection has been observed for 309 several viruses. We first evaluated the presence of a super-infection exclusion mechanism in sheep 310 cells infected with BTV, using two different viruses, a non-modified BTV (BTVWT) and a fluorescent 311 BTV (BTVUnaG) encoding a fluorescent fusion protein VP6-UnaG. After controlling that BTVUnaG 312 virus particles are also mainly released in EVs (Fig 6A), we first infected sheep cells with BTVWT, 313 followed by a second infection with BTVUnaG immediately (0 hpi) or at 1, 2, 3, 4 or 5 hpi (Fig 6B). Supporting this mechanism, we observed the presence of the lysosomal cathepsin proteins in the same 393 low-density fractions as the viral particles after isopycnic density centrifugation, as well as co-394 localisation of the lysosomal markers LAMP1 and cathepsin with the viral proteins. Interestingly, the 395 BTV NS3M14 virus, which has a mutated first start codon and is only able to synthesis an NS3A 396 isoform of NS3 (lacking the N-terminal 13 amino-acids) [21], released significantly less virus 397 particles in EVs than BTVWT. Conversely, the BTV NS3M1 mutant (with a mutation in the second 398 start codon preventing the synthesis of the NS3A isoform) showed comparable levels of infectious 399 EVs to the BTVWT. We did not observe a change in the co-localisation of VP5 or NS3 with the 400 lysosomal markers in cells infected with BTV NS3M14 compared with BTVWT. In contrast, we found 401 that lysosome acidity was higher in non-infected cells and cells infected with BTV NS3M14, compared 402 with BTVWT. This suggests that NS3M14 is unable to counter lysosomal acidification as shown for 403 WT NS3. We were able to revert this phenotype using bafilomycin A1, an inhibitor of the vacuolar 404 type H+-ATPase, and showed that lysosomal pH disruption in BTV NS3M14 infected cells restored Interestingly, it has been proposed that the endoplasmic reticulum, in close spatial association with 411 lysosomes, regulates the calcium storage in these organelle [29,30]. We previously showed that NS3 412 transits though the endoplasmic reticulum (ER), and that disrupting NS3 trafficking through the ER 413 causes release of immature particles. It would be thus interesting to explore the interplay between 414 NS3, the ER and lysosomes in order to determine how NS3 could regulate the secretion of 415 BTV particles in EVs. Altogether, our data support an original model of a virus particles release 416 mechanism involving MVBs and secretory lysosomes. Interestingly, this phenomenon is not 417 restricted to viruses, as it has also been described for intracellular uropathogenic E. coli [20] EVs on BTV induced pathogenicity, as infected ruminants variably respond to BTV infection, with cattle being mainly asymptomatic and sheep exhibiting more severe clinical signs, such as 437 haemorrhage and ulcer in the gastro-intestinal tract or muscle necrosis [35]. In this study, we were 438 able to identify infectious EVs in the blood of bovines infected in 2019, but it was not possible to 439 obtain any evidence of infectious EVs in blood samples of infected sheep. However, we note that 440 extracting EVs from frozen whole blood could potentially alter the integrity and the shape of the 441 observed membranes structure, and further experiments will be needed to explore these aspects in 442 more detail. It is important to note that BTV pathogenesis is markedly different in sheep and cattle 443 and is also dependent on the virus serotype by centrifugation (500xg -10 min), and infectivity was quantified by a TCID50 method. PT infected 505 cells were lysed by three cycle of freeze thawing. Lysed cells were resuspended in PBS and cell debris 506 was spun down at 500 g for 5 min. Viral RNA extraction was then performed on the harvested 507 supernatant as indicated below. 508

Isopycnic gradient centrifugation of EVs 545
EVs purification by isopycnic gradient centrifugation was performed at 150,000xg for 4 hours at 4°C, 546 using a SW41 TI swinging bucket rotor (Beckman Coulter). EVs resuspended in 900 µL D-MEM 547 and diluted with 100 µL iodixanol (45%) to a final 5% concentration. EVs were layered on top of a 548 10-45% iodixanol discontinue gradient and harvested in 9 fractions, from the bottom of the tube using 549 a syringe. Fraction density was calculated by measuring the absorbance at 244nm after a 1:10,000 550 dilution and using a standard curve obtained by serial dilution of iodixanol. 551

Specific infectivity 552
Specific infectivity of BTV in EVs and free virus fractions correspond to the number of particles 553 required to infect a cell. We approximated the particles/mL as the genome copy number/mL measured 554 by qPCR using S10 specific primers after extraction and reverse transcription of the viral genome. 555 The number of pfu/mL was obtained by virus titration using a plaque assay.

Western Blot 557
Proteins were detected from either resuspended EVs or cell lysates. Cell monolayers were washed 558 twice with a PBS solution and lysed for 30 min at 4°C with a RIPA lysis buffer, 1X protease inhibitor 559 cocktail, and 5 mM EDTA (78438, Thermofisher scientific). Cell lysates, or EVs, were diluted in 4X 560 NuPAGE™ LDS Sample Buffer (Thermofisher scientific) to a final 1x concentration and incubated 561 at 95°C for 5 min. Proteins were then separated on a 10% SDS gel and blotted on a PVDF membrane. 562 When indicated, the membranes were cut horizontally for incubation with different antibodies. The Biotechnology), were incubated overnight at 4°C, followed by a 2 hours incubation with respective 567 Horseradish peroxidase-coupled secondary antibody (anti-mouse IgG ab97023, anti-guinea pig IgG 568 ab97155 and anti-rabbit IgG ab97051, Abcam). Luminescence was then detected using SuperSignal 569 West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific). The ImageJ software was 570 used to perform linear contrast enhancement. 571

Transmission electron microscopy 572
To image EVs without sectioning, purified EVs were resuspended in D-MEM, adsorbed onto a 573 carbon-coated copper grid, and stained with a 1% solution of uranyl acetate. For imaging of EVs 574 sections, purified EVs were resuspended in 30 µL D-MEM (0% FCS) and mixed with a low volume 575 of Glutaraldehyde 25% (final concentration = 2.5%) (16220, Electron Microscopy Sciences). After 576 10 min of fixation at RT, EVs were briefly warmed to 37°C, and 16 µL of EVs were dropped on 4 577 µL of pre-melted low-melting agar (5% in water) and kept at 42°C before the addition of EVs. 578 so that drugs did not interfere with viral entry. These concentrations were selected because they were 599 already validated by previous studies on BTV from us and others [43,44]. At 24 hpi, cell supernatants 600 were harvested for isolation of EVs and analysis of infectivity. For intracellular virus particle 601 quantification in presence or absence of the different drugs, the media culture of PT infected cells 602 was removed at 24 hpi, and cell monolayers were rinsed twice with PBS. Cells were then lysed by 3 603 successive cycles of freeze/thaw at -80°C/35°C. Cell lysates were then resuspended in 500 µL DMEM 604 and virus particles were quantified using a TCID50 method. 605

Quantification of cell surface proteins 606
Cell surface protein expression of LAMP1 was measured by immunofluorescence using flow 607 cytometry. PT cells were seeded in 6-well plates for 24 h before infection with BTV-1 (MOI=10).

Data availability 708
All relevant data are within the manuscript and its Supporting Information files. 709