Ethics statement

Our research complies with all relevant ethical regulations, and all procedures were approved by the ethics committee of Ulm University (88/17, 89/16, 89/17, 337/18 and 131/16). Body fluids were donated by healthy individuals after informed consent by the respective donors. Vaginal tissue was obtained from the Clinic for Gynecology and Obstetrics, Ulm University Medical Center, from surgical removal in patients with pelvic organ prolapse after informed patient consent. Pooled human male AB blood group serum was obtained from Sigma Aldrich (cat. no. H4522).

Cells

Human embryonic kidney cells (HEK293T; ATCC #CRL3216), HeLa cells expressing CCR5, CXCR4 and HIV-1-tat-responsive β-galactosidase (TZM-bl; NIBSC #ARP5011), human hepatocellular carcinoma cells (Huh-7; provided by R. Bartenschlager, University of Heidelberg), primary human foreskin fibroblasts (HFFs; kindly provided by J. von Einem; Ulm University Medical Center), baby hamster kidney (BHK-21; ATCC #CCL-10) cells, BHK cells expressing HSV-responsive β-galactosidase (ELVIS61), and human lung epithelial cells (Calu-3; ATCC #HTB-55; kindly provided by M. Frick, Ulm University) were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% (v/v) foetal bovine serum (FBS), 100 U ml−1 penicillin–streptomycin (Pen-Strep) and 2 mM l-glutamine. African green monkey kidney epithelial cells (Vero E6; ATCC #CRL-1586) were cultured in DMEM supplemented with 2.5% (v/v) FBS, 100 U ml−1 Pen-Strep, 2 mM l-glutamine, 1 mM sodium pyruvate and 1% (v/v) non-essential amino acids. All commercially obtained cells were authenticated by the vendor (ATCC or National Institutes of Health AIDS Reagent Program) and not further validated by our laboratory. Cell lines obtained from other groups, who published these cells, were not further authenticated. Ex vivo vaginal tissue was obtained as excess material during surgical procedures and cut into 1 × 1 × 2 cm3 tissue blocks, which were each maintained in 200 μl RPMI1640 medium containing 15% (v/v) FBS, 100 U ml−1 Pen-Strep, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 μg ml−1 gentamicin and 25 μg ml−1 amphotericin B. All cells and ex vivo vaginal tissue blocks were incubated in humidified (95%) incubators at 37 °C and 5% CO2.

Viruses

To propagate ZIKV, HSV-1, HSV-2 and SARS-CoV-2, subconfluent Vero E6 cells were inoculated with virus in 5 ml culture medium (for ZIKV, HSV-1 and HSV-2) or 3.5 ml serum-free medium containing 1 μg ml−1 trypsin (for SARS-CoV-2) and incubated for 2 h. Medium containing 10–25 mM HEPES was then added to 40 ml total volume and cytopathic effects were monitored for 3–5 days. To generate HIV-1 virus stocks, subconfluent HEK293T cells were transfected with plasmids encoding full-length viruses using Transit LT-1 (Mirus) at a 3:1 ratio of reagent to DNA. Cells were incubated for 2 days before supernatants were collected. Virus-containing supernatants were generally centrifuged for 3 min at 325g to remove cellular debris, aliquoted and stored at −80 °C as virus stocks. Infectious virus titre was determined by endpoint titration. DENV R2A16, WNV16 and transcription- and replication-competent Ebola virus-like particles (EBOV-trVLP)62,63 were generated as described previously. HCV was produced as described before64,65, and viral titres were determined by infecting Huh7.5 cells stably expressing the HCV RFP-NLS-IPS reporter65,66,67. VSV was amplified as described before68,69. Enhanced green fluorescent protein (EGFP)-expressing CHIKV70 was produced by in vitro-transcription of and subsequent electroporation of RNA into BHK-21 cells. Virus-containing supernatant was collected, passaged once on BHK-21 cells and viral titres were determined by titration on HEK293T cells.

EV purification

To purify EVs from non-vesicular proteins and impurities in body fluids, TFF with subsequent BE-SEC was applied as described previously17,71. This methodology may be classified as ‘intermediate recovery, intermediate specificity’ according to MISEV2018 (ref. 72). Source fluids (Supplementary Table 2) were subjected to pre-processing as described in the following. As a standard protocol, source fluids were pre-clarified by low-speed centrifugation in 50 ml polypropylene tubes (Sarstedt) at 700g for 5 min, followed by 3,000g for 5–15 min to remove larger particles and debris.

For purification of EVs from breast milk, a special pre-clearance protocol was followed to sequentially remove the fat-containing fraction, as previously described73. Fresh, non-cooled/non-frozen whole human breast milk (stored in Breastmilk Storage Bags, Lansinoh, <4 h after collection at room temperature) was first centrifuged at 3,000g for 10 min at room temperature, and the cream layer was removed. The residual fluid was again centrifuged first at 5,000g (30 min 4 °C), then at 10,000g (30 min 4 °C), each time removing cream after centrifugation, before applying 0.22 μm vacuum filtration and proceeding with EV isolation. For semen, spermatocytes were pelleted by centrifuging at 21,000g for 15–30 min to obtain seminal plasma before proceeding with purification. For saliva, urine and blood, the standard centrifugal protocol was applied.

After filtration, fluids were diluted fourfold in sterile PBS and filtered through a 0.22-μm syringe filter (Millipore Millex-GP, PES) or vacuum filtration units (Millipore SteriCup, PES). The filtrate was diafiltrated with two volumes of sterile PBS and concentrated to ~20–40 ml using a KR2i TFF system (SpectrumLabs) or an Äkta Flux S (Cytiva) with 300 kDa cut-off hollow fibres at a flow rate of 100 ml min−1 (Repligen, transmembrane pressure 3.0 psi; shear rate at 3,700 s−1, mPES). The pre-concentrated sample was subsequently loaded onto two daisy-chained BE-SEC columns (bind-elute size exclusion chromatography; HiScreen Capto Core 700 column, Cytiva) connected to an ÄKTAstart chromatography system (Cytiva) with UNICORN start 1.2.0.164, which was primed (3 ml min−1 flow rate) with 500–1,000 ml sterile PBS beforehand. After loading the entire sample volume, PBS was pumped through the columns (1.5 ml min−1 flow rate) and the EV sample collected as the first peak in A280 absorbance. The collected fraction was then concentrated using a 100 kDa molecular weight cut-off spin-filter (Amicon (RC) or Vivaspin (PES), 3,220g for 60 min). Bovine serum albumin (BSA) was added to 0.1% (v/v) and samples aliquoted to 1.5-ml polypropylene reaction tubes and stored at −80 °C for further downstream analysis. Systems were cleaned with 1 M NaOH in ddH2O (TFF) or 1 M NaOH in 30% isopropanol (BE-SEC) and rinsed with ddH2O until the pH of the flowthrough was ~7.

Nanoparticle tracking analysis

NTA was used to determine concentration, size distribution and surface charge (zeta potential) of particles in body fluids, EV isolations or synthetic liposomes using a ZetaView TWIN (Particle Metrix) as previously described17. Samples were diluted in particle-free PBS (for concentration and size distribution), and videos of the scattering particles were recorded with the following settings: 25 °C fixed temperature, 11 positions, 1 cycle, sensitivity 85–90, shutter 100, 15 fps, 2 s videos per position, 3–5 measurements. Videos were evaluated by ZetaView Analyze 08.05.05 SP2 and 08.05.12 SP1. Between the samples, the chamber was thoroughly flushed with particle-free PBS.

Metabolic activity measurement

Effects of substances or purified EVs on cellular metabolic activity were evaluated using CellTiter-Glo Luminescent Cell Viability Assay (Promega), quantifying intracellular ATP levels. Assays for cell viability determination were conducted in parallel to antiviral assays, with medium only being added instead of virus. CellTiter-Glo assay was performed according to the manufacturer’s instructions; medium was removed from the culture, and 50% substrate in PBS was added. After 10 min, luminescence of the samples was measured in an Orion II Microplate Luminometer (Titertek Berthold) using Simplicity 4.2 software. Untreated controls were set to 100% viability.

Infection assaysZIKV and WNV

ZIKV and WNV infection and its quantification by immunodetection were performed as previously described16,17. Briefly, target cells (6,000 Vero E6, A549, HeLa or primary fibroblasts) were seeded in 96-well plates in 100 µl medium per well one day prior and inoculated with virus (multiplicity of infection (MOI) 0.15–0.2) and compounds/vesicles in a total volume of 100 µl. After 2 h, the medium was replaced. Infection was quantified after 2 days by detecting flavivirus E protein with a horseradish peroxidase (HRP)-conjugated secondary antibody. For this, medium was removed and cells were fixed with 4% (para)formaldehyde in PBS (PFA) for 20 min at room temperature. Cells were then permeabilized with 100% ice-cold methanol for 5 min and washed with PBS, before adding 1:10,000 (ZIKV) or 1:5,000 (WNV) diluted mouse anti-flavivirus protein E antibody 4G2 (Absolute Antibody #Ab00230-2.0) in antibody buffer. After 1 h incubation at 37 °C, cells were washed three times with washing buffer before a secondary anti-mouse antibody conjugated to HRP (Thermo Fisher Scientific #31460; 1:10,000) was added and incubated for 1 h at 37 °C. Following four washing steps, TMB peroxidase substrate (Medac) was added and the reaction stopped after 5 min light-protected incubation at room temperature by adding an equal volume of 0.5 M H2SO4. The optical density was recorded at 450 nm and baseline corrected for 650 nm using the vMax Kinetic ELISA microplate reader (Molecular Devices) and SoftMax Pro 7.0.3 software. Values were corrected for the background signal derived from uninfected cells, and untreated controls were set to 100% infection. For immunofluorescent imaging of ZIKV infection, cells were fixed and permeabilized in the same manner, but a secondary antibody with fluorescent label was used to detect ZIKV-E (Thermo Fisher Scientific #A11001, diluted 1:400). After washing, nuclei were stained with 1:2,000 diluted Hoechst 33342 (10 μg ml−1 stock in H2O, Thermo Fisher Scientific #H1399) and incubated for 10 min at room temperature. Imaging was then performed in a Cytation 3 Microplate Reader/Microscope using Gen5 software (3.04.17, Biotek).

ZIKV virion attachment assay

Virion attachment assay was conducted as previously described16,17. Vero E6 (150,000 cells per well) were seeded in eight-well μ-slides (Ibidi) 1 day prior, and cells were then inoculated with ZIKV MR766 (MOI 7.4–35) in the presence of putative attachment inhibitors and incubated for 2 h at 4 °C in 300 μl total volume. Cells were then fixed with 4% PFA for 10 min at 4 °C, washed with PBS and unspecific binding sites blocked by 30 min incubation with attachment assay blocking buffer, 5% (v/v) FBS and 1% (v/v) BSA in PBS. Virions were then detected with 1:10,000 diluted mouse anti-flavivirus protein E antibody 4G2 (Absolute Antibody #Ab00230-2.0) in attachment assay binding buffer, PBS with 1% (v/v) BSA for 45 min. After three washing steps with PBS, samples were incubated with 1:1,000 diluted goat anti-mouse secondary antibody conjugated to A647 (Thermo Fisher Scientific #A27029) in binding buffer for 45 min. Finally, cell nuclei were stained with 1:2,000 diluted Hoechst 33342 (Thermo Fisher Scientific #H1399) for 20 min. Attached virus particles were imaged as z-stacks of 25 slices of 0.55 µm by confocal microscopy using a Zeiss LSM 710. Images were processed and combined to maximum intensity projections using the ZEN 2.3 (blue edition) software. Attached ZIKV virions were quantified using a custom ImageJ (Fiji Version 2.3.0) macro (Supplementary Code 1) that automatically identifies and counts local fluorescence maxima of a set of 1,024 × 1,024 px confocal images, and co-localization was evaluated using Huygens Professional software (Version 19.10) using Gaussian minimums for thresholding. Pearson’s co-localization coefficients are reported as a result, with a value of 0.6 or greater considered a positive correlation.

ZIKV infection of vaginal tissue blocks

Vaginal tissues obtained from surgical procedures were cut to 2 × 2 × 1 mm3 tissue blocks, and 32 blocks were incubated with compounds and virus for 2.5 h at 37 °C. After 3× washing with 50 ml PBS, blocks were sorted to individual wells of a 96-well culture plate with 250 μl ex vivo medium16, 100 μl aas taken as a wash control and replaced by fresh medium. Supernatants were centrifuged at 325g for 3 min, transferred to new plates and stored at −80 °C. To study replication, 100 μl supernatant was further collected and replaced by fresh medium on days 2, 4 and 6 post infection and replication finally evaluated by quantitative reverse-transcription polymerase chain reaction as described17 using StepOne Software 2.3. Primers: RKI-ZIKV-forward 5′-ACGGCYCTYGCTGGAGC-3′; RKI-ZIKV-reverse 5′-GGAATATGACACRCCCTTCAAYCTAAG-3′; ZIKV-probe FAM- AGGCTGAGATGGATGGTGCAAAGGG-BNQ535 (biomers.net)

DENV

DENV R2A (Thailand/16681/84)74,75 infection was performed as previously described16,74. Vero E6 cells were seeded 1 day before infection in 96-well plates (6,000 cells per well) and inoculated with virus (MOI 0.15) and compounds/vesicles in a total volume of 200 µl the next day. Infection was determined 3 days post-infection by measuring Renilla luciferase activity.

EBOV

EBOV-trVLPs were produced and used in infection assay as previously described62,63. Briefly, for EBOV-trVLP production, HEK293T cells (six-well format, 40,000 cells per well) were transfected with expression plasmids for EBOV-NP, EBOV-VP35, EBOV-VP30, EBOV-L, T7-polymerase and the EBOV-trVLP minigenome (p4cis-vRNARLuc). Medium was replaced after 16 h post, and cells were incubated for an additional 48 h, before EBOV-trVLP-containing supernatants were collected and clarified from debris by centrifugation (4,000g, 10 min). For infection, HEK293T target cells (96-well plate format, 12,000 cells per well), which were transfected to express EBOV-NP, EBOV-VP30, EBOV-VP35, EBOV-L and DC-SIGN, were seeded 1 day before 96-well plates (12,000 cells per well), inoculated with undiluted EBOV-trVLP supernatant from producer cells and compounds/vesicles in a total volume of 200 µl at 24 h post transfection. Medium was replaced after 6 h and infection was determined at 72 h post inoculation by measuring Renilla luciferase activity in cell lysates. To this end, medium was removed and cells were lysed by incubation with Cell Culture Lysis Reagent (Promega) followed by incubation for 30 min at room temperature. Lysates were then transferred to white 96-well plates, before Renilla luciferase substrate was added (coelenterazine, 2 μM) and luminescence was measured after 60 s incubation. Values were corrected for the background signal derived from uninfected cells, and untreated controls were set to 100% infection.

SARS-CoV-2

SARS-CoV-2 infection and its quantification by immunodetection was performed as previously described17,76. Calu-3 (15,000 cells per well) target cells were seeded in 96-well plates in 100 µl and the next day replaced with 70 μl of fresh medium. Virus and compounds/vesicles were added in a total volume of 100 µl and 80 μl fresh medium added after 1 h. Infection was quantified after 2 days by detecting SARS-CoV-2 N protein with an HRP-conjugated secondary antibody. Following fixation by adding PFA to a final concentration of 4% and incubating at room temperature for 30 min, medium was discarded and cells permeabilized for 5 min at room temperature by adding 100 µl of 0.1% Triton in PBS. After washing with PBS, 1:5,000 diluted mouse anti-SARS-CoV-2 N protein antibody 40143-MM05 (SinoBiological #40143-MM05) was added in antibody buffer at 37 °C. After 1 h, the cells were washed three times with washing buffer before a secondary anti-mouse antibody conjugated with HRP (Thermo Fisher Scientific #31460) was added (1:15,000) and incubated for 1 h at 37 °C. Following four times of washing, the TMB peroxidase substrate (Medac) was added. After 5 min light-protected incubation at room temperature, reaction was stopped using 0.5 M H2SO4. The optical density was recorded at 450 nm and baseline corrected for 620 nm using the Asys Expert 96 UV microplate reader (Biochrom). Values were corrected for the background signal derived from uninfected cells, and untreated controls were set to 100% infection.

HSV-1 and HSV-2

HSV-1 and HSV-2 infection was determined by measuring green fluorescent protein (GFP) in Vero E6 cells infected with reporter viruses. Vero E6 (6,000 cells per well) were seeded 1 day prior and incubated 1–2 days after addition of virus and test compounds the next day. Medium was then removed and cells were washed once with PBS before adding 0.25% trypsin–ethylenediaminetetraacetic acid for detachment of cells. After incubation at 37 °C and visual confirmation of detachment, an equal volume of medium (containing 10% FBS) was added and cells from wells infected in triplicates were pooled in a 96-well V-bottom plate. After centrifuging 325g for 3 min, supernatants were aspirated and cells fixed by adding 4% PFA in PBS and incubating 1 h at room temperature. An equal volume of fluorescence-activated cell sorting buffer was then added, and cells were analysed using a CytoFLEX flow cytometer and CytExpert 2.3 software. Uninfected cells were used as a reference to set gating for the GFP channel, and the percentage of GFP+ cells was used as a quantitative readout for relative infection. Alternatively, infection was determined by ELVIS reporter cells as previously described77,78.

HIV-1

HIV-1 infection was determined using TZM-bl cells, which are HeLa cells engineered to express β-galactosidase under an HIV-responsive promoter. A total of 10,000 TZM-bl cells were seeded 1 day before 96-well plates, virus and test compounds were added the next day, medium was replaced after 2 h and infection was determined at 2 days post-infection. For this, supernatants were removed, Gal-Screen reagent was added and transferred to white well plates following incubation at room temperature, and luminescence was recorded for 0.1 s using a Berthold Microplate reader using Simplicity 4.2 software. Values were background-subtracted (uninfected cells) and normalized to cells in the absence of test compounds.

HCMV

HCMV infection was studied on primary HFFs. TB40/E HCMV stocks were kindly provided by Jens von Einem, Institute of Virology, Ulm University Medical Center. HFFs (10,000 cells per well) were seeded 1 day prior, virus and test compounds were added the next day and infection was then determined at 1 day post-infection by immunodetection of HCMV-IE protein in fixed cells as previously described79.

VSV

VSV–EGFP (kindly provided by KK Conzelmann LMU Munich)69 infection was performed in Vero E6 cells. A total of 10,000 cells were seeded 1 day before infection in 96-well plates in 100 μl medium per well and inoculated with virus (MOI 0.02) and compounds/vesicles the next day. Medium was replaced after 2 h. One day post-infection, the supernatant was removed, and cells were washed with PBS and detached using trypsin (0.25%) before being transferred to V-bottom plates for fixation (4% PFA, 1 h, 4 °C). Cells were then analysed for GFP fluorescence by flow cytometry (CytoFLEX) using uninfected cells as a control for gating.

CHIKV

The CHIKV LR2006-OPY 5′GFP infectious clone expressing EGFP under the control of a subgenomic promotor (hereafter referred to as CHIKV) has been described previously70. A total of 10,000 HEK293T cells were seeded in 96-well plates in 100 µl medium per well and inoculated with CHIKV (MOI 10) and compounds/vesicles on the following day. Viral inoculum was removed and replaced with fresh medium 2 h post infection. One day post-infection, the supernatant was removed, and cells were washed with PBS and detached using trypsin (0.25%) before fixation (2% PFA, 1.5 h, room temperature). Cells were then analysed for the percentage of cells expressing GFP by flow cytometry (FACSLyric) using uninfected cells as a control for gating.

HCV

HCV infection rates were analysed by using a Jc1p7-Gluc-2A-NS2 reporter strain65,80 and measuring the gaussia luciferase activity in the supernatant. Briefly, 5,000 Huh-7 cells were seeded in 100 µl medium per 96 wells (DMEM (high glucose, sodium pyruvate) supplemented with 10% FBS, 1% glutamine and 1% Pen-Strep) in 96-well plates. The next day, medium was changed to DMEM additionally supplemented with gentamicin (100 µg ml−1) and compounds/vesicles and virus inoculum (MOI 0.4) were added. After 2 h, cells were washed once in PBS and further cultivated in fresh DMEM containing gentamicin (200 µl per well). Two days post-infection supernatants were collected and lysed in 1% Triton X-100 (4 °C, 1 h). Secreted gaussia luciferase activity was determined using coelenterazine (10 µM, Carl Roth) and a Centro LB 960 luminometer (Berthold Technologies), and mock values were subtracted.

Virion and EV lipid modificationsPSD reaction and tracking by DSB-3

Viruses were treated with PS decarboxylase (referred to as PSD), specifically, a modified, water-soluble and His-tagged version of Plasmodium knowlesi also known as His6-Δ34PkPSD for converting surface-exposed PS to PE. The enzyme was produced and purified as previously described81. For treatment, PSD was spiked directly into virus stocks at up to 100 μg ml−1 final concentration and reactions incubated at 30 °C while shaking at 450 rpm on an orbital shaker. Reaction kinetics were tracked by a distyrylbenzene-bis-aldehyde (DSB-3) that fluoresces upon reaction with ethanolamine as previously described with slight modification82. Briefly, samples of the running PSD reaction were taken at 0 (before addition of enzyme), 5, 15, 30 and 60 min, adding 50 μl sample to a mixture of 37 μl decarboxylation buffer and 12.5 μl of a tetraborate buffer (100 mM sodium tetraborate, pH 9) to stop the enzyme reaction, storing the samples at 4 °C until samples from all timepoints were collected. Afterwards, Triton X-100 was added to 1.3 mM final concentration followed by 10 μl of a 100 μM DSB-3 solution. After incubation for 60 min at room temperature, fluorescence was measured at λex = 403 nm and λem = 508 using a microplate fluorescence reader (BioTek Synergy) with Gen5 3.08.01. PSD removal by Ni-NTA bead treatment results in substantial reduced signal in infection inhibition experiments; therefore, PSD-containing samples were diluted tenfold on cells to achieve maximally 10 μg ml−1 on-cell concentrations, which were not cytotoxic (Extended Data Fig. 1b).

Phospholipase treatment and blockade of surface-exposed PS

To abrogate the antiviral activity of EVs, surface-exposed PS (and other phospholipids) were first ‘shaved’ by digestion with phospholipase D (PLD from Streptomyces chromofuscus, Sigma Aldrich #P0065). For this, EVs purified using TFF/BE-SEC as described in Tris buffer (50 mM Tris base and 100 mM NaCl, adjusted to pH 7.4 with HCl) were digested with 500 U PLD in the presence of 10 mM CaCl2 on a tube rotator at 40 °C. PLD was then removed by CaptoCore700 BE-SEC and vesicles concentrated by 100 kDa ultrafiltration as for initial purification. To block residual surface-exposed PS, vesicles were additionally incubated with bovine LA (Cellsystems, #BLAC-1200) at 50 μg ml−1 overnight. Antiviral activity of ‘shaved & blocked’ EVs was then tested directly or after incubation for 1 week at 4 °C followed by 1 day at 37 °C (‘+ Incubated’).

Cyclodextrin-mediated lipid exchange

PS-depleted vesicles were re-supplied with PS to restore antiviral activity by using cyclodextrin-mediated lipid exchange. Methyl‐α‐cyclodextrin (AraChem #CdexA-076/BR) was incubated with donor liposomes consisting of 100 mol% DOPS at a ratio of 1.5 mM lipid to 40 mM cyclodextrin for 2 h while on a tube rotator at 40 °C. Lipid-loaded cyclodextrins were then separated from the donor liposomes by ultrafiltration with a 10 kDa MWCO (Amicon 0.5, Merck Millipore). Loaded cyclodextrins were then mixed with PS-depleted vesicles (350 μl loaded cyclodextrin filtration flowthrough + 50 μl depleted vesicles + 100 μl PBS) and samples incubated for 2 h while on a tube rotator at 40 °C. Cyclodextrins were then again separated by ultrafiltration with 10 kDa MWCO as before, and the retentate containing lipid-exchanged vesicles was eluted.

Confocal microscopy for hybrid EV receptor occupation and uptake assay

For analysing EV receptor occupation, Vero E6 (70,000 cells per well) were seeded in eight-well μ-slides (Ibidi) 1 day prior. Cy5-labelled hybrid semen EVs were then added at indicated concentrations and slides incubated 2 h at 4 °C. Supernatants were then removed and cells fixed by addition of 4% PFA for 10 min at room temperature followed by addition of anti-Axl-PE (Thermo Fisher Scientific #12-1087-42, diluted in PBS + 1 vol% FBS to 120 ng ml−1) and incubation at 37 °C for 1 h. Cells were then washed 3× with PBS + 1 vol% FBS before staining nuclei with 1:2,000 diluted Hoechst 33342 (Thermo Fisher Scientific #H1399). After two more washing steps, cells were imaged as z-stacks of 20 slices of 0.55 µm by confocal microscopy using a Zeiss LSM 710 and ZEN 2.3 (blue edition) software. z-Stacks were analysed by Huygens Professional software (Version 19.10) using Gaussian minimums for thresholding. Mander’s M2 overlap coefficient (reporting the fraction of PE/Axl-positive pixels that are also Cy5/SemenEV positive) is reported as a result.

For hybrid EV uptake assay, 40,000 Vero E6 cells were seeded to a 35 μ-dish (Ibidi) in 2 ml medium 1 day prior. The next day, LysoTracker Green DND-26 (Thermo Fisher, #L7526) was added at 75 nM in fresh medium and cells were incubated for 30 min before replacing the medium with phenol-red free DMEM (supplemented as the normal growth medium), adding hybrid semen EVs at 5 × 1010 particles ml−1 and imaging the entire cell volume as z-stack immediately using a Leica SP8 confocal microscope in resonance mode. Cells were incubated between imaging timepoints, and the gain corresponding to LysoTracker Green DND-26 was increased to compensate for fading during long-term imaging.

Phospholipid HPTLC with iodine vapor staining

To analyse lipid content of liposomes and EVs by thin-layer chromatography, total lipid was isolated using the Folch method. For this, 25 μl sample was mixed with 500 μl solvent consisting of a 2:1 (v/v) ratio of chloroform and methanol, followed by 5 min of sonication in a bath sonicator and 2,000 rpm shaking for 20 min at 4 °C. Then 125 μl water was added to induce phase separation, and samples were centrifuged at 2,000g for 5 min. The upper aqueous layer was removed, and the lower phase was transferred to new vials. Solvent was then evaporated under nitrogen, and 25 μl chloroform was added before shaking for 5 min at 2,000 rpm and sonicating for 5 min. High-performance thin-layer chromatography (HPTLC) plates (Sigma-Aldrich) were activated by heating to 110 °C for 30 min, and samples were then spotted in 3 μl volume using graduated glass pipettes (Hirschmann Ringcaps). The running chamber was filled with mobile phase, consisting of methyl acetate, isopropanol, chloroform, methanol and aqueous KCl (0.25) at a ratio of 25:25:25:10:9, as described previously83. Plates were then inserted and allowed to run until the solvent front had reached within 0.5 cm of the upper plate limit. Plates were then air dried for 10 min and developed in an iodine vapour chamber. After 30 min, plates were immediately imaged in a GelDoc (Bio-Rad).

Shotgun lipidomics

Mass spectrometry (MS)-based lipid analysis was performed by CRO Lipotype GmbH as described84. Lipids were extracted using a two-step chloroform/methanol procedure85. Samples were spiked with internal lipid standard mixture containing: cardiolipin 14:0/14:0/14:0/14:0 (CL), ceramide 18:1;2/17:0 (Cer), diacylglycerol 17:0/17:0 (DAG), hexosylceramide 18:1;2/12:0 (HexCer), lyso-phosphatidate 17:0 (LPA), lyso-phosphatidylcholine 12:0 (LPC), lyso-phosphatidylethanolamine 17:1 (LPE), lyso-phosphatidylglycerol 17:1 (LPG), lyso-phosphatidylinositol 17:1 (LPI), lyso-phosphatidylserine 17:1 (LPS), phosphatidate 17:0/17:0 (PA), phosphatidylcholine 17:0/17:0 (PC), phosphatidylethanolamine 17:0/17:0 (PE), phosphatidylglycerol 17:0/17:0 (PG), phosphatidylinositol 16:0/16:0 (PI), phosphatidylserine 17:0/17:0 (PS), cholesterol ester 20:0 (CE), sphingomyelin 18:1;2/12:0;0 (SM) and triacylglycerol 17:0/17:0/17:0 (TAG). After extraction, the organic phase was transferred to an infusion plate and dried in a speed vacuum concentrator. First-step dry extract was resuspended in 7.5 mM ammonium acetate in chloroform/methanol/propanol (1:2:4, v-v:v) and second-step dry extract in 33% ethanol solution of methylamine in chloroform/methanol (0.003:5:1; v-v:v). All liquid handling steps were performed using Hamilton Robotics STARlet robotic platform with the Anti Droplet Control feature for organic solvents pipetting. Samples were analysed by direct infusion on a Qexactive mass spectrometer (Thermo Scientific) equipped with a TriVersa NanoMate ion source (Advion Biosciences). Samples were analysed in both positive and negative ion modes with a resolution of Rm/z = 200 = 280,000 for MS and Rm/z = 200 = 17,500 for MS/MS experiments, in a single acquisition. MS/MS was triggered by an inclusion list encompassing corresponding MS mass ranges scanned in 1 Da increments86. Both MS and MS/MS data were combined to monitor CE, DAG and TAG ions as ammonium adducts; PC and PC O-, as acetate adducts; and CL, PA, PE, PE O-, PG, PI and PS as deprotonated anions. MS only was used to monitor LPA, LPE, LPE O-, LPI and LPS as deprotonated anions; Cer, HexCer, SM, LPC and LPC O- as acetate adducts. Data presented as “mol% lipid” herein indicate the proportion of specific lipid species amongst all analysed. Cholesterol was not included in our analysis. Data were analysed with in-house developed lipid identification software based on LipidXplorer87,88. Data post-processing and normalization were performed using an in-house developed data management system. Only lipid identifications with a signal-to-noise ratio >5 and a signal intensity fivefold higher than in corresponding blank samples were considered for further data analysis.

Multiplex bead-based EV surface protein profiling by flow cytometry

Purified EVs were subjected to bead-based multiplex EV analysis (MACSPlex Exosome Kit, human, Miltenyi Biotec) as previously described17,89 with adaptations. Briefly, EVs purified by TFF/BE-SEC were diluted at input doses of 109 or 1010 NTA-quantified particles per assay in MACSplex buffer and incubated overnight with MACSPlex Exosome Capture Beads on an orbital shaker at 450 rpm at room temperature. Beads were washed with MACSPlex buffer, and most of the supernatant was aspirated. For staining of captured EVs, a cocktail of APC-conjugated anti-CD9, anti-CD63 or anti-CD81 detection antibodies (Tetraspanins (TSPN); Miltenyi Biotec; #130-108-813; 5 µl each) or alternatively for detection of PS, 0.5 μg LA-A647 (Haematologic Technologies, #BLAC-ALEXA647) were added tube followed by incubation at 450 rpm for 1 h at room temperature in the dark. Next, the samples were washed twice with MACSplex buffer and liquid removed before resuspension in 150 µl MACSPlex buffer. Samples were then transferred to a V-bottom 96-well microtitre plate (Thermo Scientific) and analysed by flow cytometry using a Cytoflex LX (Beckman Coulter). CytExpert 2.3 (Beckman Coulter) was used to analyse flow cytometric data (see Supplementary Information for gating strategy). Median fluorescence intensities (MFIs) for all 39 capture bead subsets were background-corrected by subtracting respective MFI values from matched non-EV containing buffer controls that were treated exactly like EV samples (buffer + capture beads + antibodies). Raw MFI values below those obtained for isotype controls of each respective sample are reported as ‘not detected’. For experiments directly comparing samples not measured on the same day (Fig. 4g), MFI was converted to molecules of equivalent soluble fluorochrome (MESF) by calibration beads (Bangs Laboratories Quantum #647 for calibrating LA A647 data; #823 for calibrating TSPN-APC data) to ensure consistency.

Nano-flow cytometry

For nano-flow cytometry, a Flow NanoAnalyzer (NanoFCM) equipped with a 488 nm and a 638 nm laser, was calibrated using 200 nm polystyrene beads (NanoFCM) with a defined concentration of 2.08 × 108 particles ml−1, which were also used as a reference for particle concentration. In addition, monodisperse silica beads (NanoFCM) of four different diameters (68 nm, 91 nm, 113 nm and 155 nm) served as size reference standards. Freshly filtered (0.1 μm) 1× TE (Tris-EDTA) buffer pH 7.4 (Lonza) was analysed to define the background signal, which was subtracted from all other measurements. Each distribution histogram or dot plot was derived from data collected for 1 min with a sample pressure of 1.0 kPa. The EV samples were diluted with filtered (0.1 μm) 1× TE buffer, resulting in a particle count in the optimal range of 2,500–12,000 events. Particle concentration and size distribution were calculated using NanoFCM software (NF Profession V2.0). For immunofluorescent staining, 12.5 μM of corresponding antibodies in 50 μl TE was used: fluorescein isothiocyanate (FITC)-conjugated mouse anti-human CD9 antibody (clone HI9a; BioLegend #312104), FITC-conjugated mouse anti-human CD81 antibody (clone TAPA-1; BioLegend #349504) and FITC-conjugated mouse anti-human CD63 antibody (clone H5C6; BioLegend #353005); as isotype controls, FITC-conjugated mouse IgG1, κ (clone MOCP-21; BioLegend #400109) and FITC-conjugated mouse IgG2a, κ (MOPC-173; BioLegend #400207), 2 ng µl−1 of each antibody in 50 µl 1× TE buffer. PS-staining was done using 5 µg ml−1 LA-A647 (Haematologic Technologies, #BLAC-ALEXA647). After removing antibody aggregates by centrifugation at 12,000g for 10 min, the supernatant was added to 2 × 108 particles, followed by incubation for 12 h at 37 °C under constant shaking and washing with 1 ml 1× TE buffer by ultracentrifugation at 110,000g for 45 min at 4 °C (Beckman Coulter MAX-XP centrifuge, TLA-45 rotor; Beckman Coulter). The pellet was resuspended in 50 μl 1× TE buffer for nano-flow cytometric analysis.

Liposome synthesis

Liposomes were prepared by thin-film hydration and extrusion as described17,90. For this, lipids in chloroform were added to a glass round-bottom flask and the solvent was then evaporated by slowly applying a vacuum at a Schlenk line. The vacuum was held for 2 h and then purged with argon. Alternatively, solvent was evaporated under a nitrogen gas stream for at least 1 h at room temperature and the lipid film then overlaid with argon. Next, the lipid film was hydrated by adding PBS and the flasks were shaken at 180 rpm at a temperature dependent on the transition temperature of the lipids utilized. Small unilamellar vesicles were then prepared by at least 25× extrusion through 0.2 μm polycarbonate membranes (Nuclepore Track-Etched Membrane, Whatman) in a Mini Extruder (Avanti Polar Lipids); for lipid compositions with transition temperature above room temperature this was done on a heating block set to the respective temperature. Liposomes were quantified by NTA using a ZetaView (ParticleMetrix).

Hybrid EV-liposome preparation

To label TFF/BE-SEC-purified EVs with a synthetic dye-conjugated lipid, a lipid mixture containing 90 mol% dioleoyl PC and 10 mol% Cy5-dioleoylPC was added to glass vials and the solvent was evaporated as described above. To 100 μM dried lipid mixture, 2.5 × 1012 EVs (or unlabelled liposomes) were then added and the mixtures were shaken at 45 °C, 180 rpm for 1 h. Serial extrusion was then done 5× through 1 μm, 10× through 0.2 μm and 20× through 0.1 μm membranes. Vesicles were then stored at −80 °C under argon until use.

Determination of protein content by BCA assay

The protein content of samples was quantified using Pierce Rapid Gold BCA Protein Assay Kit as described by the manufacturer (Thermo Fisher) using the microwell procedure using using the vMax Kinetic ELISA microplate reader (Molecular Devices) and SoftMax Pro 7.0.3 software.

SDS-PAGE

Sodium dodecyl sulfate polyacrylamide gel electrophoresis and total protein staining were performed as previously described17. Protein samples were mixed with protein loading buffer (LI-COR) and tris(2-carboxyethyl)phosphine as reducing agent (50 mM final concentration) and heated to 70 °C for 10 min. Proteins were then separated on NuPAGE 4–12% BisTris gels (200 V, 30 min). For total protein staining, gels were fixed with a 50% methanol:7% acetic acid solution and stained with GelCode Blue (Thermo Fisher) for 1 h at room temperature. After destaining with ultrapure water, the gel was imaged on a LI-COR Odyssey near-infrared imager or a ChemiDoc Imaging Systems.

Western blot

Western blot analysis of EV and non-EV marker proteins was performed as previously described17. Briefly, protein concentrations of samples were first measured by bicinchoninic acid assay (BCA) as described above and then adjusted to that of the lowest-concentrated sample with PBS. Concentration-adjusted samples were mixed with protein loading buffer (LI-COR) and tris(2-carboxyethyl)phosphine (Sigma Aldrich, 50 mM final concentration) and heated to 70 °C for 10 min. Proteins were then separated on NuPAGE 4–12% BisTris gels, blotted onto Immobilon-FL polyvinylidene fluoride membranes via semi-dry transfer and blocked with 0.25% casein in PBS-T 0.05% (Thermo Scientific). Membranes were then stained with primary antibodies (CD9, 1:1,000, Cell Signaling Technology #13174; flotillin-1, 1:1,000, Cell Signaling Technology, #18634; HSP70, 1:1,000, Cell Signaling Technology, #4876) in 0.025% casein in PBS-T 0.05% (staining buffer) overnight a 4 °C. The next day, membranes were washed 3× with PBS-T 0.05% and incubated with secondary antibody (1:2,500, Bio-Rad #12005870; #12004162 or 1:10,000 Thermo Fisher Scientific #31460