*3.3. Proteomic Analysis of EVs from mHSC*

Mass spectrometry for mHSC EVs revealed striking qualitative and quantitative differences of proteins in EVs from D4 mHSC versus EVs from P1 mHSC (Supplemental Tables S2 and S3). Analysis of five separate EV samples from D4 mHSC or three separate EV samples from P1 mHSC resulted in the identification of between, respectively, 31 and 58 proteins or 426 and 504 proteins (Figure 3A). Subsequent analysis of these data was focused on 46 proteins that were present in at least three EV samples from D4 mHSC and on 337 proteins that were common to all 3 EV samples from P1 mHSC. Of these, 19 proteins were unique to D4 mHSC EVs, 310 proteins were unique to P1 mHSC EVs and 27 proteins were shared between D4 and P1 mHSC EVs (Figure 3B). With respect to the 19 D4-specific proteins, the most abundant (quantitative value ~100) included three histones (H15, H10, H11) and ezrin (EZR1), a membrane-bound cytoskeleton linker protein (Figure 3C). Of the shared proteins, 11 proteins were present at significantly higher levels in EVs from D4 mHSC compared to EVs from P1 mHSC and six of these (H4, H13, H2B1F, H14, K2C1, H2A1B) were the most abundant proteins (quantitative value ~100–1000) in EVs from D4 mHSC (Figure 3D). No D4 mHSC-specific EV protein had this level of abundance (Figure 3C) but three proteins (FN1 (also termed FINC), FBLN2, PGBM) reached a comparable abundance level among the proteins specific to P1 mHSC EVs (Figure 3E). Western blot analysis was used to confirm the high abundance of keratin in EVs from D4 mHSC and of FN1 in EVs from P1 mHSC (Figure 4A). Western blot and MS analysis demonstrated the presence of EDA and EDB sequences in the FN1 protein showing that it was the cell-associated form as opposed to the plasma form which lacks these domains (Supplemental Figure S1). For the 21 proteins that were differentially expressed in EVs from D4 versus P1 cells, analysis of the producer cells by RT-PCR showed that their corresponding cellular transcripts were comparably differentially expressed for only eight candidates (H10, K2C8, K2C1, ACTG, EZR1, TSPAN8, RAI3, AQP1) with the rest showing no

significant difference in expression except H2B1F which showed inverse cellular mRNA expression (P1 > D4) as compared to EV protein level (D4 > P1) (Figure 4B).

**Figure 3.** Proteomic composition of mHSC EVs. (**A**) Summary of quantitative features of EV proteins analyzed from five D4 mHSC EV samples or three P1 mHSC EV samples. (**B**) Venn diagram showing distribution of proteins between EVs from D4 versus P1 mHSC. The figure also shows the identities and quantifications of (**C**) the 19 proteins specific for D4 mHSC EVs; (**D**) the 27 proteins shared by both groups for which 11 proteins were significantly enriched in EVs from D4 mHSC (\*, *p* < 0.05, \*\*, *p* < 0.01) and (**E**) the 20 most abundant proteins specific for EVs from P1 mHSC.

When all D4 mHSC EV proteins or all P1 mHSC EV proteins were analyzed using GO/Funrich, the 20 most highly represented components were generally differentially expressed, with proteasome complex and collagen trimer being unique to P1 mHSC EVs (Figure 5A). The 27 proteins common to EVs from both D4 and P1 mHSC shared enrichment for components that included exosomes, membranes, cytoplasm, extracellular space, focal adhesion, ECM, microparticles, vesicles, cytoskeleton and cell-cell adherins (Figure 5B), many of which were shared with proteins that were P1-specific (Figure 5C). Proteins in D4-specific EVs included some of the same components (exosomes, vesicles) but in there was an absence of extracellular or adhesion components and instead a concentration of membraneous (apical, aipicolateral or microvillus membrane, sarcolemma), nuclear (chromosome, euchromatin) and structural-functional (dystrophin glycoprotein complex, costamere) components (Figure 5D).

**Figure 4.** EV Western blot and cellular transcript levels for differentially expressed EV proteins. (**A**) Western blots of EVs from D4 mHSC or P1 mHSC to verify differential levels of representative proteins identified by MS. (**B**) qRT-PCR was performed on RNA from D4 or P1 mHSC using primers designed to amplify mRNA corresponding to proteins that were more highly expressed in EVs from D4 mHSC than in EVs from P1 mHSC (see Figure 3C,D). Data are mean ± S.E.M. for duplicate determinations performed twice individually. \*\*, *p* < 0.01, \*\*\*, *p* < 0.005.

**Figure 5.** Cellular component analysis of proteins in EVs from D4 and/or P1 mHSC. GO analysis was performed using the Funrich database to identify EV proteins with significant enrichment versus the Uniprot database. The figure shows the 20 most highly ranked cellular components for proteins that were (**A**) in the entire mHSC EV proteome from D4 cells compared to those from P1 cells; (**B**) shared between EVs from D4 and P1 mHSC; (**C**) specific to P1 mHSC EVs. (**D**) Components specific to D4 mHSC EVs. Only cellular components with significant enrichment (\* *p* < 0.05) are shown.

KEGG pathway analysis of all 46 proteins in D4 mHSC EVs and all 337 proteins in P1 mHSC EVs revealed one unique pathway for D4 mHSC EVs (alcoholism, involving 8% of the proteins), four shared pathways for D4 and P1 mHSC EVs (regulation of actin cytoskeleton, leucocyte transendothelial migration, systemic lupus erythematosus, and proteoglycans in cancer, each involving 3–11% of the proteins) and 33 unique pathways for P1 mHSC EVs (for which focal adhesion, PI3K-Akt signaling, proteasome, ECM-receptor interaction and pathways in cancer each involved the highest proportion (8–10%) of the proteins) (Figure 6A). While no KEGG pathway was identified for the 19 D4 mHSC-specific EV proteins, most of the these pathways were, respectively, prominent for either the 27 proteins that were shared between D4 and P1 mHSC EVs (regulation of actin cytoskeleton, leucocyte transendothelial migration, systemic lupus erythematosus, alcoholism) (Figure 6B) or the P1 mHSC EV-specific proteins (focal adhesion, PI3K-Akt signaling, proteasome, ECM-interactions, pathways in cancer) (Figure 6C).

**Figure 6.** KEGG pathway analysis of proteins in EVs from D4 and/or P1 mHSC. EV proteins were analyzed online using DAVID v6.8 software for KEGG pathway analysis. The figure shows pathways that were (**A**) in the entire mHSC EV proteome from D4 mHSC versus P1 mHSC; (**B**) shared between EVs from D4 and P1 mHSC; or (**C**) specific to P1 mHSC EVs (top 20 pathways shown). Only pathways with significant enrichment (*p* < 0.05) are shown.

STRING analysis of the proteomic data to identify principal protein interactions and functions revealed striking differences between each type of EV. Whereas the proteins in D4 mHSC EVs were organized into a simple network comprising nodes that included keratins and histones (Supplemental Figure S2), those in P1 mHSC EVs demonstrated much more complex interactions with principal nodes containing proteins associated with collagens, ECM, vesicular transport, metabolic enzymes, proteasomes, ribosomes, chaperones and tRNA ligase (Supplemental Figure S3). Representative proteins from key nodes in each network were verified by Western blot as being specific for D4 mHSC EVs (keratin) or P1 mHSC EVs (PSMA6, RPS27A) (Figure 4A).
