A Comparative Proteomic Analysis of Extracellular Vesicles Associated With Lipotoxicity

Abstract

Extracellular vesicles (EVs) are emerging mediators of intercellular communication in nonalcoholic steatohepatitis (NASH). Palmitate, a lipotoxic saturated fatty acid, activates hepatocellular endoplasmic reticulum stress, which has been demonstrated to be important in NASH pathogenesis, including in the release of EVs. We have previously demonstrated that the release of palmitate-stimulated EVs is dependent on the de novo synthesis of ceramide, which is trafficked by the ceramide transport protein, STARD11. The trafficking of ceramide is a critical step in the release of lipotoxic EVs, as cells deficient in STARD11 do not release palmitate-stimulated EVs. Here, we examined the hypothesis that protein cargoes are trafficked to lipotoxic EVs in a ceramide-dependent manner. We performed quantitative proteomic analysis of palmitate-stimulated EVs in control and STARD11 knockout hepatocyte cell lines. Proteomics was performed on EVs isolated by size exclusion chromatography, ultracentrifugation, and density gradient separation, and EV proteins were measured by mass spectrometry. We also performed human EV proteomics from a control and a NASH plasma sample, for comparative analyses with hepatocyte-derived lipotoxic EVs. Size exclusion chromatography yielded most unique EV proteins. Ceramide-dependent lipotoxic EVs contain damage-associated molecular patterns and adhesion molecules. Haptoglobin, vascular non-inflammatory molecule-1, and insulin-like growth factor-binding protein complex acid labile subunit were commonly detected in NASH and hepatocyte-derived ceramide-dependent EVs. Lipotoxic EV proteomics provides novel candidate proteins to investigate in NASH pathogenesis and as diagnostic biomarkers for hepatocyte-derived EVs in NASH patients.

Keywords: DAMP; StAR-related lipid transfer domain 11; exosome; hepatocyte; microvesicle.

FIGURE 1

Overview of EV proteomics methods. (A) Schema represents three extracellular vesicle (EV) isolation methods for EV proteomics. In size exclusion chromatography (SEC) methods, fractions 6.5 to 10.5 were combined and pelleted by ultracentrifugation (UTC). For the UTC sample, the 100,000 × g fraction was utilized. For Iodixanol, density gradient (DG) fractions 1–2, 3–6, and 7–10 were collected and combined. (B) The number of identified proteins for each of the three methods, SEC, UTC, and DG. The gray bars display the total number of identified proteins, the orange bars display the number of the total protein IDs that were found in the ExoCarta database, and the blue bars display the number of proteins from each method that was found in the ExoCarta top 100 EV proteins. (C) Venn diagrams depicting the number of unique proteins that were detected in vehicle (Veh) or palmitate (PA) stimulated EVs by each method. (D) Heatmap shows common 537 proteins that were detected by each method. Heatmap color represents Log2 protein abundances. The red box encloses the cluster of proteins with higher abundance in EVs isolated by SEC.

FIGURE 2

Comparison of EV proteomics. (A) Heatmap depicting the abundance 83 common proteins (Log2 protein abundance) of ExoCarta top 100 EV proteins detected by each method. (B) Differential expression of proteins by volcano plot where the x-axis represents Log2 (DG PA/UTC PA) and the y-axis represents Log2 (UTC PA/UTC Veh). (C) Differential expression of proteins by volcano plots where the x-axis represents Log2 (DG PA/SEC PA) and the y-axis represents Log2 (UTC PA/UTC Veh). (D) Differential expression of proteins by volcano plots where the x-axis represents Log2 (SEC PA/UTC PA) and the y-axis represents Log2 (UTC PA/UTC Veh). For (B–D), red dots indicate proteins from the top 100 ExoCarta EV proteins. (E) tSNE dimension reduction analysis was performed by Rtsne R package in all the data including missing values from PA stimulated EVs to depict the commonly and uniquely detected proteins by each of the three methods.

FIGURE 3

Comparative proteomics of lipotoxic EVs from WT or STARD11-/- cells. (A) Principal component analysis of proteins detected in wild-type (WT) or STARD11 knockout (STARD11-/-) extracellular vesicles (EVs) from palmitate (PA)- or vehicle (Veh)-treated cells. (B) The heatmap depicts top 50 significant proteins between four different groups. (C) Volcano plot shows comparison between WT Veh and WT PA; orange circle represents p < 0.05 and Log2 Fold change > 1; red circle represents p < 0.05 and Log2 Fold change < −1. (D) Volcano plot shows comparison between STARD11-/- PA and WT PA; red circle represents p < 0.05 and Log2 Fold change > 1; blue circle represents p < 0.05 and Log2 fold change < −1. (E) Figure shows ingenuity pathway analysis based on WT PA/STARD11-/- PA data.

FIGURE 4

Predicted protein–protein interactions among WT EV proteins. Protein–protein interaction analysis was performed by STRING software (A) on WT Veh top 100 expressed proteins, and (B) on WT PA top 100 expressed proteins. The edges indicate both functional and physical protein associations; line thickness indicates the strength of data support; we used minimum required interaction score with high confidence (0.700); disconnected nodes in the network were excluded; red color depicts the S100 family proteins; blue color represents the annexin family; green color represents ribosomal proteins; dark green represents cell adhesion proteins; yellow color represents stress response proteins; and pink color represents glycolysis proteins. These proteins were annotated by using UniProt and InterPro.

FIGURE 5

Predicted protein–protein interactions among STARD11-/- EV proteins. (A) Protein–protein interaction analysis was performed by STRING software on STARD11-/- PA top 100 expressed proteins; red color shows annexin family proteins; blue color represents ribosomal proteins; green color represents stress response proteins; yellow color represents glycolysis proteins; and pink color represents cell adhesion proteins. (B) The Venn diagram represents the number of common and unique proteins between STARD11-/- PA and WT PA EV proteins.

FIGURE 6

Overview of plasma EV proteomics methods. (A) Schema represents two different EV isolation methods. In SEC, fractions 6.5 to 10.5 were combined and pelleted by UTC. For DG fractions 1–2, 3–6, and 7–10 were collected and combined. (B) Venn diagram depicting the number of common and unique proteins in control and NASH plasma EVs isolated by SEC compared to DG fractions 3–6. (C) Venn diagram depicting the number of common and unique proteins in control plasma EVs isolated by SEC and NASH plasma EVs isolated by SEC. (D) Venn diagram depicting the number of common and unique proteins in control plasma EVs isolated by DG fractions 3–6 and NASH plasma EVs isolated by DG fractions 3–6. (E) Protein–protein interaction analysis was performed by STRING software on unique NASH EV proteins; the edges indicate both functional and physical protein associations; line thickness indicates the strength of data support; we used minimum required interaction score with medium confidence (0.400); disconnected nodes in the network were excluded; red color represents immune system process. The annotation of these proteins was done by using Gene Ontology.

FIGURE 7

Differentially expressed proteins among plasma EVs. Heatmap representing proteins with greater than 1.5-fold change in NASH plasma EVs, which were isolated (A) by SEC and (B) by DG. (C) Protein–protein interaction analysis was performed by STRING software on significantly enriched proteins in NASH EVs; the edges indicate both functional and physical protein associations; line thickness indicates the strength of data support; we used minimum required interaction score with medium confidence (0.400); disconnected nodes in the network were excluded; blue color represents cell adhesion proteins; green color represents plasma lipoprotein particle protein; yellow color represents vesicle-mediated transport; these proteins annotation was done by using Gene Ontology and UniProt.

FIGURE 8

Comparison of palmitate-stimulated EV proteins with NASH plasma EV proteins. (A) Venn diagram compared 357 proteins of human NASH plasma EVs and 1,866 proteins of mouse hepatocyte-derived, palmitate-stimulated EVs. (B) Protein–protein interaction analysis was performed by STRING software; the edges indicate both functional and physical protein associations; line thickness indicates the strength of data support; we used minimum required interaction score with high confidence (0.700); disconnected nodes in the network were excluded; red color represents focal adhesion proteins; blue color represents RAB subfamily of small GTPases proteins; these proteins annotation was done by using KEGG and SMART. (C) Bar graph depicts EV proteins detected in ceramide-dependent palmitate-stimulated lipotoxic EVs and NASH plasma EVs; proteins are shown which had p < 0.05; x-axis and color represents Log2 fold change.

FIGURE 9

Sankey diagram linking palmitate-stimulated EV proteins with NASH plasma EV proteins. Sankey diagram linking top 8 commonly detected gene ontology (GO) terms from 1,766 hepatocyte-derived palmitate-stimulated EV proteins and 357 human NASH plasma EV proteins, (A) by biological process analysis, (B) by molecular function analysis, and (C) by cellular component analysis. A ribbon’s thickness indicates −log10FDR for each cluster of GO terms in each GO analysis.