The Aedes aegypti Aag2 cell line, derived from whole homogenized embryos (Lan and Fallon, 1990), was grown in Roswell Park Memorial Institute (RPMI) medium (Gibco) with 10% decomplemented fetal bovine serum (FBS) (Gibco), 1× nonessential amino acids (Gibco), and 1% penicillin/streptomycin (Gibco) at 28°C with 5% CO₂. Prior EV collection, complete medium was replaced by RPMI medium with 2% EV-depleted FBS, 1× nonessential amino acids, and 1% penicillin/streptomycin. EV depletion from decomplemented FBS was obtained by ultracentrifugation at 100,000 g for 18 h using an S58-A rotor (k-factor 50, Thermo Scientific) in Sorvall MX Plus Series Floor Model Micro-Ultracentrifuge (Thermo Scientific). The EV-depleted supernatant was collected by leaving approximately 1 mL at the bottom of the 8-mL tube, filtered through a 0.22-µm filter (Sartorius), and stored at 4°C.

Aedes aegypti mosquitoes from the “Bora bora” colony collected in French Polynesia in 1980 (Kuno, 2010) were reared in cages (BugDorm) at 27°C ± 1°C, 70% ± 5% relative humidity, with a 12 h:12 h day:night cycle, and ad libitum access to 10% sugar solution. After egg hatching, larvae were reared at 28°C in MilliQ water supplemented with yeast and fish food flakes (TetraMin) until pupation.

Aag2 cells (5 × 10⁶) in T75 flasks (Thermo Scientific) were reared in 8 mL of EV-depleted FBS media, or 1.8 × 10⁷ Aag2 cells were reared in 16 mL of EV-depleted FBS media. As control, unconditioned media (UCM) was subjected to the same procedure without cells. Following 24, 48, or 72 h after changing the media, cell medium was collected and precleared by centrifugation at 1,500 (for NTA, SEC, AFM, and Leprechaun analyses) or 3,500 g (for TEM and cryo-EM analyses) for 10 min at 4°C.

To perform differential centrifugation, the 8-mL precleared cell medium was centrifuged at 10,000 g for 30 min at 4°C, and the pellet was resuspended in 500 µL of PBS. The resulting supernatant was ultracentrifuged at 100,000 g for 3 h at 4°C, and the pellet was resuspended in 7 mL of PBS, followed by performing ultracentrifugation a second time at 100,000 g for 3 h at 4°C. The final pellet was resuspended in 250 µL of PBS for NTA and AFM observations.

EV precipitation by ultracentrifugation of the precleared cell medium was conducted at 100,000 g for 3 h at 4°C using an S50-ST swing-bucket rotor (k-factor 76.6, Thermo Scientific) in Sorvall MX Plus Series Floor Model Micro-Ultracentrifuge. The pellet was washed with 7 mL of PBS (Gibco) and ultracentrifuged again at 100,000 g for 3 h at 4°C. The final pellet was resuspended in 30 µL of PBS for cryo-EM and in 500 µL of PBS for TEM observations.

Ten-day-old female mosquitoes were starved for 4 h and offered to feed on 3 mL of PBS covered with a silicone membrane for 2 h using a membrane feeding system (Hemotek). The solution was then precleared by centrifugation at 1,500 g for 10 min at 4°C and subjected to ultracentrifugation at 100,000 g for 3 h at 4°C. The pellet was resuspended with 30 µL of PBS for TEM analyses.

EV samples from Aag2 cells were diluted to approximately 2 × 10⁸ particles per ml in particle-free PBS and were analyzed using NanoSight NS300 (Malvern) with a 488-nm laser wavelength using the following capture settings: camera type: sCMOS, laser type: Blue488, camera level: 15, slider shutter: 1206, slider gain: 366, FPS: 25.0, number of frames: 1498, temperature: 25.0°C–25.0°C, viscosity (water 0.9 cP), and syringe pump speed: 40. Five 1-min films were taken, and the analysis was performed using NanoSight NTA 3.4 Build 3.4.4 software (Malvern) with a detection threshold set to 5. Only particle sizes found in at least two films were considered. Particles detected in negative UCM control were subtracted from the results for concentration measurements. Analyses were performed at the Institute Charles Gerhardt Montpellier, Montpellier, France.

Seven microliter of the ultracentrifuged EV solution was loaded onto a formvar-coated copper grid 100 mesh (EMS FCF100-Cu-50), left to settle for 2 min, and dried with a blotting paper. The grid was stained with 1% uranyl acetate and analyzed using a TEM F20 operating at 120 KV (Tecnai) with a Veleta numeric camera (Olympus) at the Neuroscience institute of Montpellier, INSERM U1298, France.

Three microliter of the EV solution was loaded onto a polarized Lacey Carbon-Supported Copper Grid (Electron Microscopy Sciences) in a Leica EMGP2 chamber (Leica Microsystems) at 10°C with a 95% humidity rate, dried for 3 s on a blotting paper, frozen in ethanol, and stored in liquid nitrogen. The grid was observed using the TEM JEOL 2200FS (JEOL, EUROPE) with a K3 camera (Gatan-Ametek, United States) and analyzed using DigitalMicrograph software (GMS 3).

Freshly cleaved muscovite mica sheets (V1 grade, Ted Pella Inc.) were glued on a glass slide, coated overnight at 4°C with 0.01% poly-L-lysine (Sigma), rinsed with 3 mL of MilliQ water, and dried using an N2 flux. A 3D-printed O-ring was then glued on the glass slide to assemble a small liquid cell. The separated EV solution was diluted 10-fold in buffer A (20 mM Tris HCl pH 7.5 and 100 mM NaCl), and 200 μL was deposited on the liquid cell to allow passive particle adsorption. The cell was rinsed with 200 µL of buffer A, and 200 μL of buffer A were added before imaging. Topographic imaging was performed in the quantitative imaging (QI) mode, which is a force-curve-based imaging mode, using Sharp Nitride Lever (SNL) B probes (mean cantilever spring constant kcant = 0.11 N/m, Bruker). The applied force was kept at 450 pN and a constant approach/retract speed of 75 μm/s (z range of 300 nm). Imaging was performed at room temperature using an AFM NanoWizard 4xp instrument (JPK BioAFM, Bruker Nano GmbH, Berlin, Germany) operating in the BSL3 of CEMIPAI facility (UAR3725 CNRS and Montpellier University, France). Picture analyses were performed using JPK Data Processing software (7.0.165 version).

The precleared supernatant was concentrated from 8 to 2 mL by centrifugation at 3,739 g for 5 min at 4°C with a 10-kDa filter (Amicon Ultra-15, PLGC, membrane Ultracel-PL, Merck) and centrifuged at 10,000 g for 10 min at 4°C to remove large particles. An aliquot of the resulting supernatant was analyzed using NTA. The remaining solution was completed to 2 mL with degassed PBS and separated using SEC with a qEV2/35-nm GEN 2 column (IZON) and automatic fraction collector v2 (IZON) using the following parameters: 10 fractions of 1 mL with a default buffer volume. Each fraction was analyzed using NTA after overnight storage at 4°C and concentrated from 500 to ∼50 µL by centrifugation at 10,000 g for 1 h at 4°C with a 3-kDa filter (Amicon Ultra-0,5, Membrane Ultracel-3, PMNL, Merck) before Western blot (WB) analysis.

Sequences from human and Aedes aegypti tetraspanins were obtained from UniprotKB and VectorBase, respectively. Alignment and tree were performed with MEGA11 using the maximum likelihood statistical method.

Target sequences were amplified from Aag2 cDNA using GoTaq Master mix (Promega) and 400 nM of T7 sequence-flanked primer pairs (Supplementary Table S1) designed using the E-RNAi website (Horn and Boutros, 2010). The PCR thermal profile was 95°C for 5 min; 45 cycles of 95°C for 15 s, 55°C for 15 s, and 72°C for 45 s; and a final elongation at 72°C for 10 min. After verifying the size on agarose gel, PCR products were purified using the PCR QIAquick kit (Qiagen) and transcribed overnight using the MegaScript T7 kit (Thermo Fisher). RNA was extracted using the EZNA total RNA kit (Omega), adjusted to 1 μg/μL, and folded by heating at 95°C for 5 min followed by slow cooling. Negative control dsRNA targeting LacZ was similarly produced. A total of 250,000 Aag2 cells plated in P24-well plates were transfected with 0.5 µg of dsRNA using the TransIT-mRNA Transfection Kit (Mirus) for 24 h. After transfection, the media was replaced with 250 µL of RPMI with 2% EV-depleted FBS, 1% penicillin/streptomycin, and 1× nonessential amino acids.

Seventy-two hours post dsRNA transfection, cells were lysed with 350 µL of TRK lysis buffer and RNA was extracted using the EZNA total RNA extraction kit. RNA was DNase-treated, normalized after nanoDrop quantification (ThermoFisher Scientific), and reverse transcribed using the iScript gDNA Clear cDNA Synthesis Kit (Biorad). Gene expression was quantified using Hot firepol SolisGreen qPCR mix (Euromedex) with 300 nM of forward and reverse primers (Supplementary Table S1) in a total volume of 10 µL. The thermal profile was 95°C for 5 min and 40 cycles of 95°C for 15 s, 60°C for 20 s, and 72°C for 20 s, with a final melting curve analysis in LightCycler 96 (Roche). Actin expression was used as a housekeeping gene. Relative expression was quantified using the delta–delta Ct method.

Cell media was centrifuged at 3,000 g for 5 min and kept at −20°C. Cells were lysed with RIPA buffer, incubated on ice for 20 min, and centrifuged at 20,000 g for 20 min before the supernatant was collected. After protein quantification, normalized protein amounts of cell media and lysed cells were loaded on the gel. For EVs concentrated by differential centrifugation and SEC, volumes of pellets and supernatants and volumes of fractions and input were normalized to enable comparison. Cell media, concentrated EV solution, and cell lysate were diluted in 1X reducing Laemmli SDS sample buffer (ThermoScientific), heated at 95°C for 10 min, and loaded on NuPAGE 12%, Bis–Tris gel (Invitrogen) for migration at 120 V for 5 min and then 150 V for 1 h. Protein was transferred onto a nitrocellulose membrane (Bio-Rad) using a Trans-Blot Turbo RTA Mini 0.2 µm Nitrocellulose Transfer Kit (Bio-Rad) and a Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were stained with red ponceau (Sigma) for 5 min. After rinsing in 1× PBS 0.1% Tween, membranes were blocked with 1× PBS 0.1% Tween 20 5% milk (Régilait) for 45 min and incubated with 1:1000 antihuman CD63 (Ab134045, clone EPR5702, Abcam), 1:400 antihuman pan-actin (MA5-11869, clone ACTN05, ThermoScientific), or custom-made anti-Aedes aegypti syntenin antibody at 1:500 for cell media, at 1:200 for separated EV solutions, and at 1:2000 for cell lysate in 1× PBS 0.1% Tween 20, overnight at 4°C. Anti-Aedes aegypti syntenin antibody was generated by Proteogenix (Strasbourg, France) using the CTSFIRGKMDHSVPD peptide sequence (aa 316–330) in rabbit. Secondary antibody staining was performed with 1:2000 HRP-conjugated anti-rabbit IgG (H + L) (7074, Cell Signaling Technology) and 1:2000 HRP-conjugated anti-mouse IgG (H + L) (7076, Cell Signaling Technology) diluted in 1× PBS 0.1% Tween 20 5% milk and incubated for 1 h at room temperature. Blots were developed using SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoScientific) with the ChemiDoc MP Imagin System (Bio-Rad). Images were analyzed using ImageLab Software (Bio-Rad).

Samples were analyzed using the Leprechaun Exosome Human Tetraspanin Flex Kit (Unchained Labs, US) conjugated with hCD63 (Ab134045, clone EPR5702, Abcam) and MIgG control (400101, clone MOPC-21, Biolegend), following manufacturer’s instructions. Precleared cell media was diluted in the manufacturer-supplied incubation solution and incubated for 1 h at RT on the functionalized Tetraspanin Flex Lunis. Lunis were then processed on the Luni Washer (Unchained Labs, US) using the exosome protocol, dried, and imaged using the Leprechaun instrument with Leprechaun Client v2.0. Data were analyzed using Leprechaun Analysis v 2.0 with the automatic fluorescent cutoffs applied.

Gene expression changes were assessed with post hoc multiple comparison Fisher’s LSD tests. Statistical analyses were performed using Prism 9.0 (GraphPad).

To initiate the quantification of EVs secreted from mosquito cells, we evaluated the dynamic of particle secretion using NTA on cell media collected every 24 h at 24, 48, and 72 h post media replacement. We intentionally used the term particles to describe objects detected with NTA as the technology could also quantify non-vesicular objects. Accordingly, in the unconditioned media (UCM), we detected particles ranging from 20 to 380 nm in diameter (Supplementary Figure S1) and subtracted this background from the corresponding size-matched particles in the cell media. Throughout our kinetic, we quantified an increasing number of particles starting from 5.45 × 10⁸ particles/mL at 24 h, 9.83 × 10⁸ particles/mL at 48 h, and up to 1.05 × 10⁹ particles/mL at 72 h (Figure 1A). We then described the particle size distribution at the peak of secretion at 72 h (Figure 1B; Supplementary Figures S2A, B). Particles ranged from 50 to more than 500 nm in diameter, with close to none below 50 nm, 58.4% between 50 and 100 nm, and 25.4% between 100 and 150 nm. We found an average size of 112 ± 7 nm and a mode of 80–90 nm.

To determine whether secreted particles are EVs and further describe their size distribution, we applied TEM and cryo-EM to EVs ultracentrifuged from cell media collected at 72 h. As for NTA, we first analyzed UCM with TEM and did not observe EVs but only protein aggregates (Supplementary Figure S3), providing a rationale for the detection of non-vesicular objects with NTA (Supplementary Figure S1). In cell media, TEM revealed enclosed cup-shaped structures typical of EVs (Figure 1C; Supplementary Figure S4A). EV size distribution as observed with TEM ranged from 20 to 350 nm in diameter, with 21.5% below 50 nm, 55.6% between 50 and 100 nm, and 13% between 100 and 150 nm. The average EV size was 82 ± 3 nm, and the mode was 60–70 nm (Figure 1D). Using cryo-EM, we observed vesicles delimited by a lipid bilayer (Figure 1E), which is a characteristic feature of EVs. Of note, we found EVs varying in shapes from round to more complex ovoid forms (Supplementary Figure S4B), and we could not determine whether this is a particularity of mosquito EVs or an artefact inherent to our experimental design. Size distribution of EVs observed with cryo-EM varied from 20 to 400 nm in diameter, with 13.9% below 50 nm, 48.1% between 50 and 100 nm, and 17.7% between 100 and 150 nm (Figure 1F). The average EV size was 115 ± 9 nm, and the mode was 80–90 nm. Of note, in both TEM and cryo-EM images, we observed non-vesicular aggregates and a few 70–80 nm hexagonal particles, which resemble virions (Figures 1C, E; Supplementary Figures S4C, D). Overall, we established that mosquito cells secrete EVs of variable sizes, mostly ranging from 50 to 100 nm, as determined by NTA and two microscopic technologies.

To describe EVs in mosquito saliva, we collected saliva pools, concentrated EVs by ultracentrifugation, and applied TEM. Images showed EVs (Figure 2A) ranging from 20 to 150 nm in diameter, with 80% below 50 nm, 17.1% between 50 and 100 nm, and 2.9% between 100 and 150 nm for an average size of 43 ± 1 nm and a mode of 30–40 nm (Figure 2B). Our TEM analysis confirmed the secretion of EVs in mosquito saliva and indicated a relatively small size of salivary EVs.

We first tested whether centrifugation at 10,000 g and 100,000 g pelleted different size populations of mosquito cell EVs. EVs from mosquito cell media were separated by differential centrifugation, and the resulting pellets were analyzed by NTA. Centrifugation at 10,000 g isolated larger EVs (mean size of 243 ± 25 nm and mode of 150–200 nm) with no particles detected below 100 nm (Figure 3A). Subsequent 100,000 g ultracentrifugation concentrated smaller EVs, with 75.4% of the particles ranging from 50 to 100 nm, with a mean size of 106 ± 32 nm and a mode of 60–70 nm (Figure 3B). We further characterized the size distribution of the EVs pelleted at 100,000 g by using AFM. Topographic cartography of EVs enabled the measurement of width and height of each vesicle (Figure 3C; Supplementary Figures S5A–C), which were combined to calculate an estimated 3D diameter (Kanno et al., 2002). Size distribution of the pelleted EVs as measured with AFM showed that most of the particles were smaller than 150 nm, with 22.7% ranging from 30 to 50 nm, 61.9% ranging from 50 to 100 nm, and 16% ranging from 100 to 150 nm, for an overall mean size of 75 ± 3 nm (Figure 3D).

Second, we assessed the capacity of SEC to separate EVs by size. Ultrafiltrated EVs from mosquito cell media were fractionated in 10 SEC fractions, in which we analyzed the EV size distribution by NTA. Although higher quantities of particles were detected in the first fractions, the size distribution was not altered across the fractions (Figures 3E, F). The most abundant 50- to 100-nm size range of EVs (Figure 1B) represented the majority in all fractions, and there was neither depletion in a smaller size range nor enrichment in larger EVs with the increasing number of fractions (Figure 3F). Although slightly variable, the trends were similar in a biological repeat (Supplementary Figures S6A, B).

We evaluated syntenin and the mosquito homologs of human CD63 (hCD63) as EV markers. Although A. aegypti syntenin (AAEL0005391) is functionally uncharacterized, comparison with the human protein sequence showed a moderate homology with 46% identities and 65% similarities, and a good conservation of the functional PDZ domains and LYPX(n)L motifs (Table 1). Furthermore, we found putative homologs of the syntenin partners such as syndecan and ALIX in the A. aegypti genome (Supplementary Table S2). To identify hCD63 homologs in A. Aegypti, we built a maximum likelihood amino acid tree (Supplementary Figure S7) and selected Tsp29Fa (AAEL003210) and Tsp29Fb (AAEL005536) as putative homologs. Tsp29Fb was previously identified as the hCD63 homolog (Vora et al., 2018); however, the AAEL annotation has changed ever since.

To validate antibodies for these two mosquito EV markers, we separately depleted mosquito syntenin, Tsp29Fa, and Tsp29Fb through RNAi in mosquito cells and quantified the effect on protein levels using WB. To detect syntenin, we used a custom-made antibody targeting A. aegypti syntenin, whereas we used a human-targeted commercial CD63 antibody to tentatively detect Tsp29Fa and/or Tsp29Fb. DsRNA-mediated depletion significantly reduced the expression of the targeted genes (Figure 4A) and did not alter cell survival (Supplementary Figure S8). Of note, dsRNA against Tsp29Fb also decreased mRNA levels for Tsp29Fa (Figure 4A). With the mosquito syntenin antibody, we observed a moderate reduction in cellular levels and a strong decrease in secreted quantity for syntenin following syntenin RNAi depletion (Figures 4B–E; Supplementary Figures S9A, B). Although the hCD63 antibody detected a band around the expected size (27 kDa for Tsp29Fa and 32 kDa for Tsp29Fb), depletion of neither Tsp29Fa nor Tsp29Fb reduced the amounts of cellular and secreted proteins detected with the hCD63 antibody (Figures 4B–E; Supplementary Figures S9A, B). Of note, we did not detect the hCD63 homolog in naïve media. Our results indicate two potential markers for mosquito EVs.

To further characterize mosquito syntenin and hCD63 homologs as EV markers, we implemented differential centrifugation and SEC technologies, for which we previously described the capacity to separate mosquito EVs according to their size (Figure 3). The analyzed cell media was collected 72 h post initial media replacement. WB detection of the two markers across the different steps of differential centrifugation (Figure 5A) showed that mosquito syntenin and hCD63 homologs were not pelleted at 10,000 g (see 10 k pellet) but remained in the resulting supernatant (see 10 k supernatant) together with most of the total protein content (Figure 5B). In contrast, a first ultracentrifugation at 100,000 g pelleted both markers (see ultra A pellet) and clearly depleted the markers from the resulting supernatant (see Ultra A supernatant), where most of the total protein content remained (Figure 5B). A second ultracentrifugation on the resuspended pellet from the first ultracentrifugation was less efficient at pelleting hCD63 homologs and syntenin (see ultra B pellet); however, most of the total protein remained in the resulting supernatant (see ultra B supernatant) (Figure 3B). To further support the isolation of the putative EV markers by ultracentrifugation compared to non-vesicular proteins, we repeated WB with the anti-pan-actin antibody in pellets and supernatants from the successive ultracentrifugations. Whereas hCD63 homologs were enriched in ultracentrifugation pellets, pan-actin remained in the supernatants (Figure 5C), showing selective concentration in EV markers.

We analyze the distribution of the two markers across the 10 SEC fractions. As mosquito syntenin and hCD63 homologs were not found in pellets resulting from 10,000 g centrifugation, we performed the SEC on cell media depleted of the 10,000-g pellet. High numbers of particles were eluted with each fraction, as determined by NTA (Figure 5D). Both syntenin and hCD63 homologs were detected in each of the fractions (Figure 5E). Interestingly, the EV markers were not associated with total protein levels, which were depleted in the SEC fractions compared to the input (see SEC input) (Figure 5D).

To further characterize hCD63 homologs and the associated EVs, we applied a novel technology combining EV immunocapture and interferometry-based size analysis, encompassed in the Leprechaun system (Unchained lab). First, we found that the hCD63 antibody captured more vesicles than IgG control (Figure 6A), indicating both the ability of the antibody to recognize the native form of the mosquito homolog and the outer position of the targeted epitope. Second, we observed that hCD63-captured vesicles ranged from 30 to 200 nm and were enriched in EVs smaller than 50 nm, which comprised 77.7% of captured EVs (Figure 6B). The Leprechaun technology improved the characterization of the hCD63 homolog marker.