A shared, stochastic pathway mediates exosome protein budding along plasma and endosome membranes

Abstract

Exosomes are small extracellular vesicles of ∼30 to 150 nm that are secreted by all cells, abundant in all biofluids, and play important roles in health and disease. However, details about the mechanism of exosome biogenesis are unclear. Here, we carried out a cargo-based analysis of exosome cargo protein biogenesis in which we identified the most highly enriched exosomal cargo proteins and then followed their biogenesis, trafficking, and exosomal secretion to test different hypotheses for how cells make exosomes. We show that exosome cargo proteins bud from cells (i) in exosome-sized vesicles regardless of whether they are localized to plasma or endosome membranes, (ii) ∼5-fold more efficiently when localized to the plasma membrane, (iii) ∼5-fold less efficiently when targeted to the endosome membrane, (iv) by a stochastic process that leads to ∼100-fold differences in their abundance from one exosome to another, and (v) independently of small GTPase Rab27a, the ESCRT complex-associated protein Alix, or the cargo protein CD63. Taken together, our results demonstrate that cells use a shared, stochastic mechanism to bud exosome cargoes along the spectrum of plasma and endosome membranes and far more efficiently from the plasma membrane than the endosome. Our observations also indicate that the pronounced variation in content between different exosome-sized vesicles is an inevitable consequence of a stochastic mechanism of small vesicle biogenesis, that the origin membrane of exosome-sized extracellular vesicles simply cannot be determined, and that most of what we currently know about exosomes has likely come from studies of plasma membrane-derived vesicles.

Keywords: CD63; CD81; CD9; Rab27a; extracellular vesicle; interferometric reflectance; plasma membrane; protein budding.

Figure 1

CD9, CD63, and CD81 are the most highly enriched proteins of HEK293 exosomes.A, flow diagram of exosome purification procedure. B, anti-CD9 immunoblot of (left to right) the 300g cell pellet (cells), the 5000g pellet (P1), the first 10,000g pellet (P2), the second 10,000g pellet (P3), and exosome pellets obtained by centrifugation at either 80,000g (P4) or 100,000g (P4′). C, immunoblot analysis of HEK293 cell (c) and exosome lysates (ex), probed with antibodies specific for known exosomal proteins. Molecular weights of size markers are listed to the left. Numbers in parentheses refer to the rank on the exocarta.org (150) list of ‘exosome marker proteins’ as of January 1, 2022. CD9, CD63, and CD81 were were ranked as the #1, #7, and #24 most commonly cited exosome marker proteins but were the only proteins that showed obvious enrichment in exosomes. Note also that syntenin (#8), which has recently been suggested to be the most abundant protein in exosomes (151), did not show a level of enrichment that matched that of CD9, CD81, or CD62. These experiments were repeated a minimum of two times for each protein, whereas for other proteins they reflect the results from dozens of independent trials (e.g., CD63, CD9, CD81, and Hsp90).

Figure 2

Plasma membrane–localized exosome cargoes bud more efficiently than an endosome-localized exosome cargo.A–C, confocal fluorescence micrographs of HEK293 cells that had been fixed, permeabilized, stained with DAPI, and processed for immunofluorescence microscopy using mAbs specific for (A) CD63, (B) CD9, and (C) CD81. Bar, 10 μm. These images are representative representations of more than 100 individual cells examined. D, immunoblot of HEK293 cell and exosome fractions probed with antibodies specific for CD63, CD9, and CD81. E, bar graph of the relative budding of CD63, CD9, and CD81, with bar height representing the average and error lines denoting the SEM, normalized to that of CD63. The differences in budding relative to CD63 were 4.9-fold for CD9 (±0.53; p = 0.0053, n = 4) and 15.7-fold for CD81 (±2.9; p = 0.015; n = 4). ∗ and ∗∗ denote p values <0.05 and <0.005, respectively, while n = number of trials. DAPI, 4′,6-diamidino-2-phenylindole.

Figure 3

Redirecting CD63 to the plasma membrane increases rather than decreases its vesicular secretion.A and B, confocal fluorescence micrographs of CD63^−/− cells that had been transfected with plasmids designed to express either (A) WT CD63 or (B) CD63/Y235A, incubated for 2 days, then fixed, permeabilized, stained with DAPI, and then processed for immunofluorescence microscopy using a mAb specific for CD63. Bar, 10 μm. C, anti-CD63 immunoblot of cell and exosome fractions collected from CD63^−/− cells expressing either WT CD63 or CD63/Y235A. D, bar graph of the relative budding of CD63 and CD63/Y235A, with bar height representing the average and error lines denoting the SEM, normalized to that of CD63. The difference in CD63 budding induced by redirecting CD63 to the plasma membrane was 6.1 ± 1.3-fold (p = 0.0038; n = 9). ∗∗ denotes a p value <0.005, while n = number of trials.

Figure 4

Redirecting CD9 to endosomes decreases rather than increases its vesicular secretion.A and B, confocal fluorescence micrographs of CD9^−/−/− cells that had been transfected with plasmids designed to express either (A) WT CD9 or (B) CD9/YEVM, incubated for 2 days, then fixed, permeabilized, stained with DAPI, and then processed for immunofluorescence microscopy using a mAb specific for CD9. Bar, 10 μm. C, anti-CD9 immunoblot of cell and exosome fractions collected from CD9^−/−/− cells expressing either WT CD9 or CD9/YEVM. D, bar graph of the relative budding of CD9 and CD9/YEVM, with bar height representing the average and error lines denoting the SEM, normalized to that of CD9. The magnitude of the decrease in CD9 budding brought on by its targeting to endosomes was 5.5-fold (p = 0.00019; n = 9). ∗∗ denotes a p value <0.005, while n = number of trials.

Figure 5

CD63 and CD9 are found in exosomes of the same size, regardless of whether they are localized to the plasma or endosome membrane.A and B, histograms showing the diameter of (A) CD63-containing exosomes (n = 3686) and (B) CD63/Y235A-containing exosomes (n = 5569), as determined by SPIR imaging, with exosome diameter plotted on the x-axis against the percentage of exosomes on the y-axis that fell within each 5 nm window of sizes. C and D, histograms showing the diameter of (C) CD9-containing exosomes (n = 15,684) and (D) CD9/YEVM-containing exosomes (n = 18,323), as determined by SPIR imaging, with exosome diameter plotted on the x-axis against the percentage of exosomes on the y-axis that fell within each 5 nm window of sizes. This experiment was performed three times. SPIR, single particle interferometric reflectance.

Figure 6

Redirecting CD63 to the plasma membrane leads to its cobudding into the same exosomes as CD9.A and B, scatter plots showing exosome size on the x-axis and anti-CD9 immunostaining intensity on the y-axis, for (A) CD63-containing exosomes and (B) CD63/Y235A-containing exosomes, as determined by SPIR-IFM of exosomes captured on an anti-CD63-functionalized SPIR imaging chip. The black horizontal lines depict the threshold of specific immunostaining, as determined using isotype-specific control antibodies. C, bar graph showing that the percentage of CD63-containing exosomes (20% ± 1%; n = 3) and CD63/Y235A-containing exosomes (86% ± 21%; n = 3) that carried CD9 differed by ∼4-fold (p = 0.0001; n = 3), with ∗∗∗ denoting the p value − 0.0001. D and E, scatter plots showing exosome size on the x-axis and anti-CD63 immunostaining intensity on the y-axis, for (A) CD63-containing exosomes and (B) CD63/Y235A-containing exosomes, as determined by SPIR-IFM of exosomes captured on an anti-CD9-functionalized SPIR imaging chip. The black horizontal lines depict the threshold of specific immunostaining, as determined using isotype-specific control antibodies. F, bar graph showing that the percentage of CD9-containing exosomes that carried CD63 (25% ± 7%; n = 3) was significantly lower than those that carried CD63/Y235A (91% ± 5%; n = 3), a 3.6-fold difference (p = 0.032; n = 3). ∗ denotes a p value <0.05, while n = number of trials. IFM, immunofluorescence microscopy; SPIR, single particle interferometric reflectance.

Figure 7

Redirecting CD9 to endosomes increased its co-budding with CD63 and decreased its cobudding with CD81.A and B, scatter plots showing exosome size on the x-axis and anti-CD9 immunostaining intensity on the. y-axis, for (A) CD9-containing exosomes and (B) CD9/YEVM-containing exosomes, as determined by SPIR-IFM of exosomes captured on an anti-CD63-functionalized SPIR imaging chip. The black horizontal lines depict the threshold of specific immunostaining, as determined using isotype-specific control antibodies. C, bar graph showing that the percentage of CD63-containing exosomes that contained CD9/YEVM (41% ± 2%; n = 9) was significantly higher (p = 0.011) than the percentage of CD63-containing exosomes that contained CD9 (26% ± 3%; n = 9), with ∗∗∗ denoting the p value ≤ 0.0001, while n = number of trials. D and E, scatter plots showing exosome size on the x-axis and anti-CD81 immunostaining intensity on the y-axis, for (A) CD9-containing exosomes and (B) CD9/YEVM-containing exosomes, as determined by SPIR-IFM of exosomes captured on an anti-CD9 functionalized SPIR imaging chip. The black horizontal lines depict the threshold of specific immunostaining, as determined using isotype-specific control antibodies. F, bar graph showing that CD81 was found on a significantly higher percentage of on CD9-containing exosomes (23% ± 4%; n = 9) than it was on CD9/YEVM-containing exosomes (5.4% ± 0.7%; n = 9, p = 0.0037). ∗∗ denoting the p value ≤ 0.005; n = number of trials. IFM, immunofluorescence microscopy; SPIR, single particle interferometric reflectance.

Figure 8

Exosome biogenesis by HEK293 cells is unaffected by inactivating mutations in the Rab27a gene.A, amino acid sequence of (upper line) WT Rab27a and (lower lines) predicted protein products of Rab27a alleles #1 and #2 in the HEK293 Rab27a^−/− cell line Rab27a_ko_SCC12. Allele #1 is the result of an ∼10,000 bp deletion of the DNA between the Cas9 target sites in exon 2 and exon 4, as well as an insertion of undefined length in its place. This large insertion/deletion (indel) mutation removed the splice donor site at the 5′ end of intron 2. Sequence analysis of Rab27a cDNAs from this cell line revealed that this allele expresses an mRNA derived from a cryptic splice donor site within exon 2 spiced to various a splice acceptor sites within intron 3, deleting all of exon 3 in the process. The resulting mRNAs are incapable of expressing more than the first 31 amino acids of the 221 amino acid-long Rab27a protein. Allele #2 carries two mutations, a 12 bp deletion in exon 2 and a frameshift mutation in exon 4. The exon 4 mutation deletes the C-terminal 70 amino acids of the protein, including its GTP-binding pocket (152, 153) and C-terminal cysteine residues required for its prenylation, membrane localization, and function (154), while the in-frame exon 2 deletion eliminates the unique peptide insertion that defines the Rab27 protein family. B, immunoblot analysis of whole cell lysates of Rab27^−/− cells and WT HEK293 cells, interrogated with antibodies specific for (upper panel) Rab27a and (lower panel) CD63. Arrowheads denote the detected proteins. Molecular weight markers in kDa are shown to the right. C, immunoblot of cell and exosome fractions collected from cultures of WT and Rab27a^−/− cell lines, probed with antibodies specific for CD63, CD9, and CD81. Molecular weight markers are shown in kDa. D, bar graphs of exosome yield of triplicate cultures of WT and Rab27a^−/− cells, as determined by (upper graph) nanoparticle tracking analysis and (lower graph) resistive pulse sensing. Bar height denotes the average, error bars represent the SEM, and individual data points are shown. These experiments were performed three times. cDNA, complementary DNA.

Figure 9

Exosome biogenesis by HEK293 cells is unaffected by inactivating mutations in the Alix gene.A, amino acid sequence of (upper line) the WT Alix gene and (lower lines) the predicted protein products of Alix alleles #1 and #2 in the HEK293 Alix^−/− cell line 1J. Allele #1 carries a stop codon at position 69 of the 868 amino acid-long ORF due to an 85 nucleotide-long insertion into exon 1. Allele #2 carries a ∼38,000 bp deletion between exon 1 and exon 8 that shifts the reading frame after codon 68, leading to a stop codon nine codons later. B, immunoblot analysis of whole cell lysates of WT HEK293 cells and the Alix^−/− cell line 1J, interrogated with antibodies specific for (upper panel) Alix and (lower panel) actin. Arrowheads denote the detected proteins. Molecular weight markers are shown in kDa. C, immunoblot of cell and exosome fractions collected from cultures of WT HEK293 cells and the Alix^−/− cell line 1J, probed with antibodies specific for CD63, CD81, and CD9. D, bar graphs of exosome yield of triplicate cultures of WT and Alix^−/− cells, as determined by nanoparticle tracking analysis. Bar height denotes the average, error bars represent the SEM, and individual data points are shown. These experiments were performed three times.

Figure 10

No decrease in exosome yield by CD63^−/−293F cells.A, immunoblot analysis of whole cell lysates of WT 293F cells and the F/CD63^−/− cell line described previously (137), interrogated with antibodies specific for (left panel) CD63 and (right panel) HSP90. Molecular weight markers are shown in kDa. B, exosome concentration in CTCS samples collected from cultures of WT 293F cells (left data, in blue) and F/CD63^−/− cells (right data, in coral), as determined by NTA and presented as box plots. ∗ denotes p <0.05. C, exosome concentration in exosome fractions collected from cultures of WT 293F cells (left data, in blue) and F/CD63^{−/−−/−} cells (right data, in coral), as determined by NTA, and presented as a box plots. Dots represent individual data points. These experiments were performed three times. CTCS, clarified tissue culture supernatant; NTA, nanoparticle tracking analysis.