Hypotonic stimulation induced migrasome-like structures

(a-m) The biogenesis of migrasome-like structures in NRK cells stably expressing Tspan4-GFP. In these images, the fluorescence intensity of Tspan4-GFP is shown in a color map scale from purple (low) to yellow (high).

(a) Image series of the biogenesis of migrasome-like structures. Cells were treated with hypotonic Dulbecco’s phosphate-buffered saline (DPBS) with an osmolarity of 76.3 mOsmol and imaged using a confocal microscope. Scale bar, 10 μm.

(b) Representative confocal images of cells treated with DPBS with various osmolarities. 305 mOsmol represents an isotonic condition in which no migrasome-like structures were observed. DPBS diluted to 152.5, 76.3 or 50.8 mOsmol was used to achieve hypotonic stimulations of different magnitude. The size of migrasome-like structures increased as the osmolarity was reduced. Scale bar, 5 μm.

(c) For each migrasome-like structure in (b), the whole lifetime, including the growth and shrinkage, was recorded by time-lapse imaging. The largest diameter reached during the lifetime was measured. The average diameter of migrasome-like structures increased significantly as the osmolarity was reduced. For hypotonic stimulation at 152.5, 76.3 or 50.8 mOsmol, n = 194, 261, 165 migrasome-like structures, respectively.

(d) Illustration of the experimental setup for real-time imaging of the induction of migrasome-like structures.

(e) Image series of cells treated with hypotonic DPBS in two different approaches. The osmolarity of DPBS was reduced to 76.3 mOsmol by either one single step (upper panel) or five steps with 1 min intervals (lower panel). For the step-wise reduction, the osmolarity was lowered by 25% in each step. Imaging time-points were counted from the final stimulation step. Migrasome-like structures induced by the stepwise protocol showed significantly enhanced stability. Scale bar, 5 μm.

(f) Statistical analysis of growth curves of migrasome-like structures in (e). For single step stimulation, n = 16 cells; for step-wise stimulation, n = 13 cells.

(g) Representative confocal images showing the effect of Latrunculin A (LatA) treatment on the biogenesis of migrasome-like structures. Cells were pre-incubated with 0, 0.25, 0.5 or 1 μM LatA for 10 mins and then treated with a five-step hypotonic stimulation as described in (e). LatA enhanced the biogenesis of migrasome-like structures in a dose-dependent manner. The white line indicates the boundary of the cell body. Scale bar, 10 μm.

(h) Statistical analysis of the number of migrasome-like structures per cell in (g). n = 9-14 cells.

(i) Relative mRNA level of SWELL1 analyzed by qPCR. SWELL1 expression was significantly reduced in SWELL1-knockdown (KD) cells compared to control cells.

(j) Representative confocal images of control or SWELL1-KD cells. Cells were treated with a five-step hypotonic stimulation at 2 min intervals. The osmolarity was reduced by 1/6 in each step. Migrasome-like structures are shown in the inserts. Scale bar, 5 μm.

(k) Statistical analysis of the diameter of migrasome-like structures in (j); n=75 for control cells and 104 for SWELL1-KD cells.

(l) Representative confocal images showing the effect of extracellular cations on the biogenesis of migrasome-like structures. Before stimulation, culture medium was replaced by modified DPBS in which either Na+, K+ or Cs+ was the only cation source. Cells were then treated with a five-step hypotonic stimulation at 2 min intervals. The osmolarity was reduced by 1/6 in each step. Scale bar, 5 μm.

(m) Statistical analysis of the diameter of migrasome-like structures in (l); n = 18 for Na+, 119 for K+ and 203 for Cs+.

(n) Multiple primary cell types and cell lines are capable of producing eMigrasomes. Cells were stained with WGA-AF488 (Thermo, W11261) after hypotonic induction of eMigrasome with our protocol. Z-stack image series were captured and sum-slices projections were applied. Scale bar, 10 μm (upper panels) and 5 μm (lower panels).

For all statistical analyses in this figure, P values were calculated using a two-tailed unpaired nonparametric test (Mann-Whitney test). P value<0.05 was considered statistically significant. *** P<0.001. **** P<0.0001.

Mechanistic and morphologic similarity between migrasomes and eMigrasomes

(a) Representative confocal images showing the effect of Tspan4-GFP in the biogenesis of migrasome-like structures. NRK cells were transiently transfected with mCherry vector or Tspan4-mCherry. The two populations of transfected cells were mixed in a 1:1.5 ratio in a test tube and then seeded in a confocal chamber. Cells were pre-incubated with 2 μM LatA for 10 mins and then treated with a three-step hypotonic stimulation with 2 min intervals. In each step, the osmolarity was reduced by 1/6 (16.7%). WGA-AF647 (Thermo, W32466) was then added to stain migrasome-like structures. Z-stack images were captured for further analysis. Scale bar, 5 μm.

(b) Statistical analysis of the number of migrasome-like structures per cell in NRK cells transiently transfected with mCherry vector or Tspan4-mCherry in (a). n = 53, 53 cells respectively.

(c) Statistical analysis of the number of migrasome-like structures per cell in NRK cells transiently transfected with mCherry vector (n = 26 cells) or CD82-mCherry (n = 44 cells) in Fig S1a.

(d) Statistical analysis of the number of migrasome-like structures per cell in NRK cells transiently transfected with mCherry vector (n = 31 cells) or Tspan1-mCherry (n = 44 cells) in Fig S1b.

(e) Representative confocal images showing the effect of cholesterol extraction on migrasome-like structures. NRK cells stably expressing Tspan4-GFP were stimulated to generate migrasome-like structures as described in (a). Cells were then incubated with 10 mM MβCD or buffer supplied with an equal volume of control solvent (H2O) for 30 min before imaging. Z-stack images were captured for further analysis. Scale bar, 10 μm. Insert scale bar, 5 μm.

(f) Statistical analysis of the number of migrasome-like structures per cell in control cells (n=56) or cells treated with 10 mM MβCD (n=58) in (e).

(g) Representative confocal images showing the effect of sphingomyelin depletion on the biogenesis of eMigrasomes. NRK cells stably expressing Tspan4-GFP were incubated with DMSO or 25 μM SMS2-IN-1 for 16 hrs. Cells were then treated and imaged as described in (a). Scale bar, 5 μm

(h) Statistical analysis of the number of migrasome-like structures per cell in control cells (n=51) or cells treated with 25 μM SMS2-IN-1 (n=46) in (g).

(i) Representative confocal images showing a cell generating natural migrasomes (left) and eMigrasomes (right). Migrasomes and eMigrasomes are morphologically similar. The fluorescence signal of Tspan4-GFP is highly enriched in both migrasomes and eMigrasomes.

Isolation and characterization of eMigrasome

NRK cells stably expressing Tspan4-GFP were used in all experiments in this figure if not otherwise specified.

(a) Schematic illustration showing the process of eMigrasome induction, isolation and purification.

(b) Confocal image (left), threshold edge (middle) and centroid map (right) of purified eMigrasomes. Image processing and analysis were performed using imageJ. The Hough circle transform plugin was applied to recognize and transform thresholded edges into binned objects representing individual eMigrasomes. Scale bar, 5 μm

(c) Statistical analysis of the radius of purified eMigrasomes. Measurement was performed using the map generated by Hough circle transformation analysis. 4725 particles were analyzed and the data were binned to plot the distribution of eMigrasomes radius.

(d) TEM micrograph of negatively stained purified eMigrasomes. Scale bar, 1 μm.

(e) Cryo-EM micrograph of purified eMigrasomes. Scale bar, 200 nm.

(f) Western blot showing the protein level of several markers in cell bodies and eMigrasomes. An equal amount of protein was loaded in each lane.

(g) Representative time-lapse confocal images showing the high permeability of eMigrasomes to Cy5 at 1.5 hrs post purification (left) and the gradual increase in the permeability of eMigrasomes to 40 kDa dextran-TMR (right). Scale bars, 5 μm

(h) Statistical analysis of the percentage of eMigrasomes that were permeable to 40 kDa dextran-TMR at the indicated timepoints. For each time point, eMigrasomes from three different views were analyzed. From left to right, n = 336, 493, 812, 830 and 631.

(i) Representative confocal images of eMigrasomes after sitting at room temperature for 0, 3, 7 or 14 days. Aliquots of eMigrasomes were stored in EP tubes as pellets at room temperature for the indicated time, then resuspended and dropped into a confocal chamber before imaging. Z-stack image series were captured and sum-slices projections were applied. Scale bar, 5 μm.

(j) Number of eMigrasomes after storage at room temperature for 0, 3, 7 or 14 days. Aliquots of eMigrasomes (7 x 106 eMigrasomes per tube in black, 3 x 106 eMigrasomes per tube in green) were stored in EP tubes as pellets at room temperature for the indicated time, then resuspended and stained with WGA561 before counting by FACS.

(k) Time-lapse image series showing purified eMigrasomes treated with 10 mM MβCD or control buffer. A 10 μl drop of concentrated eMigrasomes was settled in a confocal chamber, then sealed and maintained at 37°C during imaging. Buffer containing 10 mM MβCD or control solvent was added to the drop using the equipment illustrated in Fig 1d. Scale bar, 5 μm

eMigrasomes as an antigen carrier for vaccination

(a) Schematic illustration of strategies for loading proteins of interest (POI) onto eMigrasomes. The protocol includes the construction of antigen-encoding plasmids, the construction of engineered cell lines stably expressing the antigens and the production of antigen-loaded eMigrasomes. Membrane proteins, which already carry a transmembrane (TM) domain, are overexpressed in cells (top). Cytosolic proteins are membrane tethered by tagging with the TM sequence and polybasic tail of STX2 (bottom). Cells are then treated with hypotonic buffer to enrich the POI on the surface of eMigrasomes.

(b) Representative confocal image of a cell expressing PD1-mCherry and Tspan4-GFP. The transmembrane protein PD1-mCherry was loaded onto eMigrasomes as shown in the top part of (a). Scale bar, 5 μm

(c) Representative confocal image of a cell expressing membrane-tethered OVA-mCherry (mOVA-mc) and Tspan4-GFP. To create an extracellular membrane-tethered form of OVA (mOVA), the sequence of OVA was fused to the C-terminus of a truncated form of mouse STX2, in which only the transmembrane region and a polybasic tail remained. The mCherry tag was fused to the C-terminus of OVA to trace the localization of this fusion protein. The soluble protein OVA was loaded onto the membrane of eMigrasomes, as shown in the bottom part of (a). Scale bar, 5 μm.

(d) Representative confocal image of eMigrasomes isolated from MCA-205 cells stably expressing Tspan4-GFP and mOVA-mCherry (eM-OVA). Scale bar, 5 μm.

(e) Western blot showing the amount of full-length mOVA-mCherry protein in host cell and eM-OVA. Cell lysate (C) containing 1 μg total protein and purified eM-OVA samples containing 0.1, 0.3 or 1 μg total protein were loaded. The protein-immobilized PVDF membrane was firstly incubated with anti-OVA antibody and then stripped and re-blotted with anti-mCherry antibody. The antigen mOVA-mCherry was highly enriched in eMigrasomes compared to host cells.

MCA-205 cells stably expressing Tspan4-GFP and mOVA-mCherry were used for all experiments in the rest of this figure if not otherwise specified.

(f) Amount of OVA-specific IgG in mouse serum on day 14 after intravenous (i.v), nasal or subcutaneous (s.c) immunization with eM-OVA (20 µg/mouse). OVA-specific IgG was quantified by ELISA.

(g) ELISA quantification of OVA-specific IgG in sera from wild-type (WT) mice on day 14 after tail intravenous immunization with eM-OVA at the indicated dose.

(h) ELISA quantification of OVA-specific IgG in sera from WT mice on day 14 after tail intravenous immunization with eM-OVA (20 µg/mouse) or intraperitoneal immunization with Alum/OVA.

(i) Illustration of the experimental setup for assaying the stability of eM-OVA.

(j) Immunoblotting analysis of the amount of OVA protein in samples of eM-OVA which were left at room temperature for 0, 3, 7 or 14 days. 2 µg protein was loaded in each lane.

(k) ELISA quantification of OVA-specific IgG in sera from WT mice on day 14 after tail intravenous immunization with eM-OVA stored at room temperature for 0 days (D0), 3 days (D3), 7 days (D7) or 14 days (D14). 20 µg eM-OVA was injected per mouse.

An eMigrasome-based vaccine induces a strong humoral protective response against SARS-CoV-2

(a) Representative confocal images showing the presence of Spike-mCherry (S-mc) in engineered cells and isolated eMigrasomes. Scale bar, 5 μm.

(b) Schematic diagram of the experimental procedure for immunization with Spike-loaded eMigrasomes (eM-S) and collection of serum.

(c) Illustration of the rVSV-venus-SARS-CoV-2 system. VSV-Venus-SARS-CoV2 was mixed with vaccinated mice sera. Vero-TMPRSS2 cells were infected with the reporter virus/serum mixture with an MOI of 0.01. 40 hrs post infection, the Venus-positive infected cells were quantified to estimate the NT50 value for each serum.

(d) Spike-specific IgG was quantified in WT mice immunized with eM-S (20 µg/mouse, i.v.) at different time points. Each symbol represents one individual animal.

(e) Neutralization curves are presented for sera from primary-vaccination and boost-vaccination. Dots along the lines represent means from individual serum samples. Nonlinear regression was performed using the equation for the normalized response versus the inhibitor, incorporating a variable slope.

(f) NT50 of individual mouse in vaccinated groups were compared by p-value (Paired t-test) are indicated. Dotted lines represent assay limits of detection. Each line represents an individual mouse.

The effect of tetraspanin expression on the formation of migrasome-like structures

(a) NRK cells were transiently transfected with mCherry vector or CD82-mCherry. The cells were then treated and imaged as described in Fig 2a.

(b) NRK cells were transiently transfected with mCherry vector or Tspan1-mCherry. The cells were then treated and imaged as described in Fig 2a.