Composition and structure of synaptic ectosomes exporting antigen receptor linked to functional CD40 ligand from helper T cells.

Ethics

Leukapheresis products (non-clinical and de-identified) from donor blood were used as a source of human T cells and monocytes. The Non-Clinical Issue division of National Health Service approved the use of leukapheresis reduction (LRS) chambers products at the University of Oxford (REC 11/H0711/7). Clone 35 was isolated from a healthy volunteer where written informed consent was given. Ethical approval was obtained from the University of Oxford Tropical Ethics Committee (OXTREC). T cell lymphoblast, Clone 35 culture and CRISPR gene editing CD4 + T cell lymphoblasts were generated from human peripheral blood CD4 + T cells isolated from healthy donors ( Levine et al., 1997 ). Briefly, CD4 + T cells were isolated by negative selection (RosetteSep Human CD4 + T cell Enrichment Kit, Stemcell technologies) following the manufacturer’s procedure. The CD4 + T cells were activated for 3 days using anti-CD3/anti-CD28 T-cell activation and expansion beads (Dynabeads, ThermoFisher Scientific) in complete medium (RPMI 1640 media supplemented with 10% heat-inactivated fetal bovine serum, 50 U/ml of Penicillin-Streptomycin, 2 mM L-Glutamine, 10 mM HEPES, 1 mM Sodium Pyruvate, and 100 µM non-essential amino acids) with 100 U/ml of recombinant human IL-2 (PeproTech), which was replaced every 2 days keeping the cells at a concentration of 1.5 × 10 6 cells/ml. IL-2 containing media (25 U/mL) was replenished the night before experiments. We refer to these as ‘T cells’ and they were used on day 10 when all division had ceased. The HLA-DRB1*09:01-restricted T cell clone 35 (specific against the influenza H3 HA 338-355 peptide NVPEKQTRGIFGAIAGFI) were expanded using at a ratio of 1 clone: two feeder cells (irradiated, pooled PBMCs from 2 to 3 healthy donors) at a total cell concentration of 3 × 10 6 cells/ml in RPMI 1640 supplemented with 10% heat-inactivated AB human serum and 30 µg/ml of PHA for three days. Then, 100 U/ml of recombinant human IL-2 were added to fresh media, which was replaced every 2 days. For CRISPR gene editing experiments, CD4 + T cells were activated with anti-CD3/anti-CD28 beads as described above. After 3 days activation beads were removed and cells washed three times in Opti-MEM (Gibco). Trans -activating Crispr RNA (Alt-R tracrRNA) and either target or control (CD19) Alt-R CRISPR-Cas9t gRNA were obtained from IDT. For 1.5 × 10 6 T cells Alt-R tracrRNA and Alt-R CRISPR-Cas9 gRNA were mixed in equimolar amounts (150 pmol) prior to incubation at 95°C for 5 min and resultant duplex allowed to cool to room temperature. 150 pmol of ALT-R S.p Cas9 Nuclease V3 (IDT) and duplexed gRNA were mixed in IDT nuclease-free duplex buffer and assembled for 15 min at 37°C. ALT-RCas9 Electroporation Enhancer (IDT) was added (150 pmol) to the resultant ribonucleoprotein and added to 1.5 × 10 6 T cells in 50 µl of Opti-MEM prior to electroporation in an ECM 880 Electro Square Porator (BTX Harvard Apparatus). The cells were expanded for 4 days in recombinant human IL-2 supplemented RPMI as described above. MoDC culture and activation Monocyte-derived dendritic cells (moDC) were generated from the peripheral blood of healthy adults by first isolating monocytes by negative selection (RosetteSep Human Monocyte Enrichment Cocktail, STEMCELL technologies) following the manufacturer’s procedure. Then, 1.5–2 × 10 6 monocytes/ml/cm 2 were stimulated in complete media supplemented with 100 ng/ml of recombinant human GM-CSF and 200 ng/ml of recombinant human IL-4 ( Sallusto and Lanzavecchia, 1994 ). After 5 to 7 days of culture, moDC were used in experiments.

Antibodies

Primary monoclonal antibodies (mAb) used for dSTORM were anti-TCR-Alexa Fluor (AF) 488 (clone IP26; BioLegend), anti-CD40L-AF647 (clone 24–31; BioLegend), anti-ICOS-AF647 (clone C398.4A; BioLegend), anti-BST2-AF647 (clone RS38E; BioLegend), anti-HLA-DR-AF488 (clone L243; BioLegend), anti-CD81-AF647 (clone 5A6; BioLegend) anti-CD83-AF647 (clone HB15e; BioLegend) and Wheat Germ Agglutinin WGA-CF568 (Biotum) to label the surface of the SEs. All antibody clones used to assess relative or absolute quantification of protein transfer from cells to BSLB are listed in Supplementary file 2A . Isotype controls matching the relevant fluorescent dyes were used for background correction and gating. Other mAb or affinity purified antibodies are described with specific methods below. Small unilamellar vesicles (SUVs) SUV are defined as vesicles in the 20–100 nm range. SUV were formed by extrusion as described using the Avanti Miniextruder with a 100 nm filter ( Crites et al., 2015 ). When SUV were used to mimic SE, all lipids were combined prior to SUV formation, whereas BSLB and PSLB composition could be determined by mixing different proportions of stock SUVs as the final bilayer composition is determined by the average of the input SUV. NTA-SUVs for attachment of His tagged proteins were composed of 85.5 mol% DOPC, 2 mol% head group labeled ATTO-390-DOPE, and 12.5 mol% DOGS-NTA at a total lipid concentration of 4 mM. Plain SUVs that were not able to bind His tagged proteins, were composed of 98 mol% DOPC and 2% ATTO-390-DOPE at a total lipid concentration of 4 mM. Stock SUV for formation of BSLB or PSLB were composed of 0.4 mM solution of lipids in PBS with 100 mol% DOPC; 75 mol% DOPC and 25 mol% DOGS-NTA; 98 mol% DOPC and 2 mol% DOPE-CAP-Biotin; or 98% DOPC; 2 mol% ATTO-(390 or 488)-DOPE. These stocks could be mixed in different ratios prior to formation of BSLB or PSLB to generate mobile bilayers of the desired final composition. All lipids were purchased from Avanti Polar Lipids, Inc (Alabaster, AL). Nanoparticle Tracking Analysis A 10 µL aliquot of SUVs or eluted SE preparation was re-suspended in PBS in a 1:100 dilution and kept on ice for Nanoparticle Tracking Analysis (NTA). The instrument used for NTA was Nanosight NS300 (Malvern Instruments Ltd) set on light scattering mode and instrument sensitivity of 15. Measurements were taken with the aid of a syringe pump to improve reproducibility. Three sequential recordings of 60 s each were obtained per sample and NTA 3.2 software was used to process and average the three recordings to determine the mean size. moDC activation by CD40L on SUV. His-tagged recombinant soluble CD40L (sCD40L, BioLegend) was incubated with NTA-SUV or plain SUV, at ratios designed to match CD40L densities found on SE for 20 min at 24°C prior to addition to the moDCs. After 24 hr, moDCs were recovered by spinning down plates at 1500 rpm for 5 min and resuspended in flow cytometry staining buffer (10% Heat-Inactivated Goat Serum, 0.04% sodium azide in PBS pH 7.4) and incubated for 30 min at 4°C. A final concentration of 10–30 nM of each mAb was used. The multicolor panel included anti-HLA-DR PerCP (clone L243), anti-CD40 AF647 (clone 5C3), anti-ICAM-1 Brilliant Violet 510/Brilliant Violet 785 (clone HA58), anti-CD80 PE (clone 2D10), anti-CD86 Brilliant Violet 785 (clone IT2.2) and anti-ICOSL PE-Cy7 (clone 2D3). Isotype control antibodies clones MOPC-21 (IgG1, κ), MOPC-173 (IgG2a, κ) and MPC-11 (IgG2b, κ) were used matching the relevant fluorescent dyes. Staining was performed for 30 min at 4°C in the dark and constant agitation after which cells were washed twice and single cell fluorescence measurements were made by flow cytometry. Bead Supported Lipid Bilayers Silica beads (5.0 µm diameter, Bangs Laboratories, Inc) were washed extensively with PBS in a 1.5 ml conical microcentrifuge tubes. BSLBs were formed by incubation with mixtures of SUVs to generate a final lipid composition of 0.2 mol% ATTO 488-DOPE; 12.5 mol% DOGS -NTA and a mol% of DOPE-CAP-Biotin to yield 10–5000 molecules/µm 2 UCHT1-Fab in DOPC at a total lipid concentration of 0.4 mM. The resultant BSLB were washed with 1% human serum albumin (HSA)-supplemented HEPES-buffered saline (HBS), subsequently referred to as HBS/HSA. After blocking with 5% casein in PBS containing 100 μM NiSO 4 , to saturate NTA sites, 50 µg/mL unlabelled streptavidin was then coupled to biotin head groups by incubation with concentrations of streptavidin determined to yield 10–5,000 molec. /µm 2 site densities. After 20 min, the BSLB were washed 2x with HBS-HSA and biotinylated UCHT1-Fab (variable density as indicated), His-tagged ICAM-1 (200 molec. /µm 2 ), CD40 (500 molec./µm 2 ), and ICOSL (100 molec./µm 2 ) were then incubated with the bilayers at concentrations to achieve the indicated site densities (in range of 1–100 nM). Excess proteins were removed by washing with HBS/HSA after 20 min. T cells (5 × 10 5 /well) were incubated with BSLB at 1:1 ratio in a V-bottomed 96 well plate (Corning) for 1 hr at 37°C in 100 µl HBS/HSA. BSLB: cell conjugates were pelleted at 500 x g for 1 min prior to resuspension in 50 mM EDTA in PBS at 4°C to release His-tagged proteins from the BSLB, while leaving the UCHT1-Fab attached, thus selectively retaining TCR + SE. The single BSLB and cells were gently resuspended prior to staining for flow cytometry analysis or sorting.

Calibration of flow cytometry data

T cells and BSLB were analyzed using antibodies with known AF647:Ab ratio ( Supplementary file 2A ) in parallel with the Quantum AF647 Molecules of Equivalent Soluble Fluorescent dye (MESF) beads, allowing the calculation of the absolute number of mAb bound per T cell and per BSLB after subtraction of unspecific signals given by isotype control antibodies.

Airyscan microscopy

Airyscan imaging of BSLB-cell conjugates was performed on a confocal laser-scanning microscope Zeiss LSM 880 equipped with Airyscan detection module (Zeiss, Oberkochen, Germany) using the Plan-Apochromat 63×/1.46 Oil objective (Zeiss, Oberkochen, Germany). The Argon laser at 488 nm and diode laser at 561 nm were used as excitation sources, with power setting of ~ 1% and~6%, respectively, which is equivalent to 1 mW and 10 mW. The powers were set in this range in order to achieve the comparable strength of fluorescent signal for both channels. Fluorescence emission was collected at around 515 nm and 653 nm for the green and magenta channels, respectively, with the following filters BP420-480+BP495-550 (green) and BP555-620+LP645 (magenta). The emission signals were collected on the 32 channel GaAsP-PMT Airy detector. The datasets were acquired as Z-stacks with 43.5 nm pixel size and 185 nm axial steps, which correspond to ~ 50–55 slices per 3D data set.

ZEN Airyscan software

(Zeiss) was used to process the acquired data sets. This software processes each of the 32 Airy detector channels separately by performing filtering, deconvolution and pixel reassignment in order to obtain images with enhanced resolution and improved signal to noise ratio. The value of Wiener filter in ZEN software was chosen in accordance with the value in ‘auto’ reconstruction modality and was set around 7, to ensure the absence of deconvolution artefacts ( Korobchevskaya et al., 2017 ). Drift was corrected using the MultiStackReg plug-in of ImageJ (National Institute of Health). Rendering was performed in Imaris software (Bitplane). Planar Supported Lipid Bilayers (PSLB) SUV mixtures were injected into flow chambers formed by sealing acid piranha cleaned glass coverslips to adhesive backed plastic manifolds with six flow channels (StickySlide VI 0.4; Ibidi) ( Papa et al., 2017 ). After 20 min the channels were flushed with HBS-HSA without introducing air bubbles to remove excess SUVs. After blocking for 20 min with 5% casein supplemented with 100 µM NiCl 2 , to saturate NTA sites, followed by 15 min incubation with streptavidin (Sigma Aldrich), washing and then monobiotinyated or His-tagged proteins were incubated on bilayers for additional 20 min. Protein concentrations required to achieve desired densities on bilayers were calculated from calibration curves constructed from flow-cytometric measurements of BSLB, compared with reference beads containing known numbers of the appropriate fluorescent dyes (Bangs Laboratories). Bilayers were continuous liquid disordered phase as determined by fluorescence recovery after photobleaching with a 10 µm bleach spot on an FV1200 confocal microscope (Olympus). T cell immunological synapse formation on PSLB CD4 + T cells were incubated at 37°C on SLB containing either ICAM-1 alone, ICAM-1 and UCHT1-Fab or ICAM-1 UCHT1-Fab, CD40 and ICOSL. After 20–90 min of incubation the cells either fixed with 4% electron microscopy grade formaldehyde in PHEM buffer (10 mM EGTA, 2 mM MgCl 2 , 60 mM Pipes, 25 mM HEPES, pH 7.0), permeabilized with 0.1% Triton X-100 (if necessary for access to intracellular spaces) and stained with primary conjugated antibodies and imaged. Alternatively, the cells were washed off with cold PBS and the SE left behind were stained with directly conjugated antibodies and fixed with 4% formaldehyde in PHEM buffer. Prior the labeling of moDCs with mAbs on SLB for TIRF imaging, the cells were blocked for Fc receptors with 5% HSA and 5% goat or donkey serum for 1 hr at 24°C. Total internal reflection fluorescence microscopy (TIRFM) TIRFM was performed on an Olympus IX83 inverted microscope equipped with a 4-line (405 nm, 488 nm, 561 nm, and 640 nm laser) illumination system. The system was fitted with an Olympus UApON 150 × 1.45 numerical aperture objective, and a Photomertrics Evolve delta EMCCD camera to provide Nyquist sampling. Live experiments were performed with an incubator box maintaining 37°C and a continuous autofocus mechanism. Quantification of fluorescence intensity was performed with ImageJ (National Institute of Health). dSTORM imaging and data analysis. For three colour dSTORM imaging extracellular vesicles were stained using either wheat germ agglutinin (WGA) directly conjugated with CF568 (Biotium) or anti-CD81-AF647. First, 640 nm laser light was used for exciting the AF647 dye and switching it to the dark state. Second, 488 nm laser light was used for exciting the AF488 dye and switching it to the dark state. Third, 560 nm laser light was used for exciting the CF568 dye and switching it to the dark state. An additional 405 nm laser light was used for reactivating the AF647, AF488 and CF568 fluorescence. The emitted light from all dyes was collected by the same objective and imaged onto the electron-multiplying charge-coupled device camera with an effective exposure time of 10 ms. A maximum of 5000 frames for antibodies conjugated with AF647, CF568 and AF488 condition were acquired. For visualizing the WGA labelled extracellular vesicles minimum of 80,000 frames were acquired. For each receptor, the specificity of the labeling was confirmed by staining the vesicles with isotype-matched control antibodies (data not shown). Because multicolour dSTORM imaging is performed in sequential mode by using three different optical detection paths (same dichroic but different emission filters), an image registration is required to generate the final three-color dSTORM image ( Bálint et al., 2013 ; Bates et al., 2012 ; Lopes et al., 2017 ). Therefore, fiducial markers (TetraSpek Fluorescent Microspheres; Invitrogen) of 100 nm, which were visible in 488 nm, 561 nm and 640 nm channels, were used to align the 488 nm channel to 640 nm channel. The difference between 561 nm channel and 640 nm channel was negligible and therefore transformation was not performed for 561 nm channel. The images of the beads in both channels were used to calculate a polynomial transformation function that maps the 488 nm channel onto the 640 nm channel, using the MultiStackReg plug-in of ImageJ (National Institute of Health) to account for differences in magnification and rotation, for example. The transformation was applied to each frame of the 488 nm channel. dSTORM images were analyzed and rendered as previously described ( Bates et al., 2007 ; Huang et al., 2008 ) using custom-written software (Insight3, provided by B. Huang, University of California, San Francisco). In brief, peaks in single-molecule images were identified based on a threshold and fit to a simple Gaussian to determine the x and y positions. Only localizations with photon count ≥ 2000 photons were included, and localizations that appeared within one pixel in five consecutive frames were merged together and fitted as one localization. The final images were rendered by representing the x and y positions of the localizations as a Gaussian with a width that corresponds to the determined localization precision. Sample drift during acquisition was calculated and subtracted by reconstructing dSTORM images from subsets of frames (500 frames) and correlating these images to a reference frame (the initial time segment). ImageJ was used to merge rendered high-resolution images (National Institute of Health).

CBC analysis

Coordinate-based colocalization (CBC) mediated analysis between two proteins was performed using an ImageJ (National Institute of Health) plug-in ( Ovesný et al., 2014 ) based on an algorithm described previously ( Malkusch et al., 2012 ). To assess the correlation function for each localization, the x-y coordinate list from 488 nm and 640 nm dSTORM channels was used. For each localization from the 488 nm channel, the correlation function to each localization from the 640 nm channel was calculated. This parameter can vary from −1 (perfectly segregated) to 0 (uncorrelated distributions) to +1 (perfectly colocalized). The correlation coefficients were plotted as a histogram of occurrences with a 0.1 binning. The Nearest-neighbor distance (NND) between each localization from the 488 nm channel and its closest localization from the 640 nm channel was measured and plotted as the median NND between localizations per cell.

Cross-correlation analysis

Cross correlation analysis is independent of the number of localizations and is not susceptible to over-counting artifacts related to fluorescent dye re-blinking and the complements other approaches ( Stone et al., 2017 ). Cross-correlation analysis between two proteins was performed using MATLAB software provided by Sarah Shelby and Sarah Veatch from University of Michigan. Regions containing cells were masked by region of interest and the cross-correlation function from x-y coordinate list from 488 nm and 640 nm dSTORM channels was computed from these regions using an algorithm described previously ( Stone et al., 2017 ; Shelby et al., 2013 ; Veatch et al., 2012 ). Cross-correlation functions, C(r,q), were firstly tabulated by computing the distances between pairs of localized molecules, then C(r) is obtained by averaging over angles. Generally, C(r) is tabulated from ungrouped images, meaning that localizations detected within a small radius in sequential frames are counted independently. Finally, a normalized histogram with these distances was constructed into discrete bins covering radial distances up to 1000 nm. Cross-correlation functions only indicate significant correlations when the spatial distribution of the first probe influences the spatial distribution of the second probe, even when one or both of the probes are clustered themselves. Error bars are estimated using the variance within the radial average of the two dimensional C(r, q), the average lateral resolution of the measurement, and the numbers of probes imaged in each channel. The cross-correlation function tabulated from the images indicates that molecules are highly colocalized, where the magnitude of the cross-correlation yield (C(r)>1) is higher than randomly co-distributed molecules (C(r)=1). Cytokine array Primary moDCs were incubated on the extracellular vesicles der(ived from CD4 + T cells on SLB containing either only ICAM-1 (200 molec/µm 2 ), ICAM-1 and UCHT1-Fab (300 molec/µm 2 ), or ICAM-1, UCHT1-Fab, CD40 (500 molec/µm 2 ) and ICOSL (100 molec/µm 2 ), at 37°C for 24 hr. Cell supernatants were recovered and centrifuged at 350 g for 5 min at RT to remove cells and cell debris. Cytokine production was quantified in the supernatants by Human XL Cytokine Array kit (ARY022B; R and D Systems), according to manufacturer’s instructions. The positive signal from cytokines was determined by measuring the average signal of the pair of duplicate spots by using ImageJ (National Institute of Health). Differences between arrays were corrected by using the average intensity of positive spots within the array. Fold change of the cytokine production between conditions was determined by normalizing the data to SLB containing only ICAM-1.

Mass Spectrometry AF488+

BSLB were sorted on a FACS ARIA III and lysed by sonication (Bioruptor Pico) in 0.5% NP-40 in 50 mM ammonium bicarbonate and 6 M urea. Cysteines were reduced and alkylated by addition of first 5 µl of 200 mM dithiothreitol (30 min at 24°C) and 10 µl of 200 mM iodoacetamide (60 min at RT in dark). The protein solution was then precipitated with chloroform and methanol ( Wessel and Flügge, 1984 ), and resuspended in 6 M Urea. For digest the protein solution was diluted in 50 mM ammonium bicarbonate, pH 7, and 0.6 µg trypsin was added for digest at 37°C overnight. Peptides were desalted with a C18 solid phase extraction cartridge (SOLA, Thermo Fisher Scientific) and resuspended in 15 µl 2% acetonitrile and 0.1% trifluoroacetic acid in water. Samples were analyzed on a LC-MS/MS platform consisting of Orbitrap Fusion Lumos coupled to a UPLC ultimate 3000 RSLCnano (both Thermo Fisher Scientific). Samples were loaded in 1% acetonitrile and 0.1% trifluoroacetic acid in water and eluted with a gradient from 2% to 35% acetonitrile, 0.1% formic acid and 5% dimethylsulfoxide in water in 60 min with a flow rate of 250 nl/min on an EASY-Spray column (ES803, Thermo Fisher Scientific). The survey scan was acquired at a resolution of 120.000 between 380–1500 m/z and an automatic gain control target of 4E5. Selected precursor ions were isolated in the quadrupole with a mass isolation window of 1.6 Th and analyzed after CID fragmentation at 35% normalized collision energy in the linear ion trap in rapid scan mode. The duty cycle was fixed at 3 s with a maximum injection time of 300 ms, AGC target of 4000 and parallelization enabled. Selected precursor masses were excluded for the following 60 s. Proteomic data was analyzed in Maxquant (V1.5.7.4, ref) using default parameters and Label Free Quantitation. The data was searched against the mouse canonical Uniprot database (29/07/2015) and the human Uniprot database (15/10/2014). FDR on peptide and protein level were set to 1%. Second peptide and ‘match between runs’ options were enabled. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE ( Vizcaíno et al., 2016 ) partner repository with the dataset identifier PXD007988 (https://www.ebi.ac.uk/pride/archive/projects/PXD007988) .

Statistical analysis

All statistical analyses were performed using SigmaPlot 13.0 (Systat Software Inc), OriginPro 2017 software (OriginLab) or GraphPad Prism v 7.0 and 8.0 (GraphPad Software, Inc). Statistical analyses are detailed in each figure legend.

Additional files 10.7554/eLife.47528.028 Supplementary file 1. Legend. Proteins Identified by MS/MS analysis and represented in Figure 5 . Fold and Log Fold two represent fold change of BSLB coated with ICAM-1, UCHT1-Fab, CD40, ICOSL over BSLB coated with ICAM-1, CD40 and ICOSL. Colors represent proteins are found for particular pathways than is predicted by chance as determined by reactome pathway 10.7554/eLife.47528.029 Supplementary file 2. Supplementary tables. ( A ) Summary of antibodies used for determination of relative and absolute enrichment of bead-transferred proteins. ( B ) Surface CD40L levels and densities on T cells exposed to different BSLB substrates (either No UCHT1-Fab ± CD40 or UCHT1-Fab ± CD40). ( C ) Estimated number of CD40L molecules per SE and estimated CD40L densities on SE using parameters obtained by dSTORM and quantitative FCM. ( D ) Size distribution of SUVs as determined by Nanoparticle Tracking Analyses using light scattering and Brownian motion. 10.7554/eLife.47528.030 Transparent reporting form