Fig 2: PAPPA KO-EVs were generated using CRISPR/Cas9.a Schematic depicting hypothesized mechanism of (intra)cellular signalling activated by EV-associated PAPP-A, based on identified proteins and significantly altered phosphosites measured in HMEC-1 upon veh-EV stimulation by (phopho)proteomic analysis. Detected proteins in HMEC-1 are displayed in grey, while significantly changing phosphosites present in cluster C1 (see Fig. 3d) are displayed in brown. b Representative western blot analysis of phosphorylated AKT (pAKT), total AKT (tAKT), phosphorylated ERK1/2 (pERK1/2) and total ERK1/2 (tERK1/2) in HMEC-1 treated with 6 × 1010 or 2 × 1010 CPC-EVs, or with 200 ng/mL free IGF-1 after pre-incubation with different doses of picropodophyllin (PPP). ß-actin (ß-ACT) was included as housekeeping protein (I = phosphorylated protein blot, II = total protein blot). Biological replicates of (b) are displayed in Supplementary Fig. 9g, h. c Sanger sequencing results confirming 1 bp insertion in exon 3 of PAPPA at the CRISPR/Cas9 target site of the PAPPA KO-CPC clone, compared with the NTgRNA polyclonal CPC line. d Western blot analysis showing the absence of PAPP-A in PAPPA KO-EVs, compared with NTgRNA-EVs; the presence of CD81, CD63, Syntenin-1 (SYNT), Flotillin (FLOT1), ß-ACT, and absence of Calnexin (CNX) in both EV populations. FLOT1, ß-ACT and CNX were present in CPC lysate (CL). e Representative NTA plot showing the size distribution and particle concentration of PAPPA KO- and NTgRNA-CPC-EVs. f Protein content per 1 × 1010 PAPPA KO- and NTgRNA-EVs of two representative experiments.

Fig 3: Characterization of PalmGFP- and PalmtdTomato-labeled EVs derived from CPCs. (A) Fluorescent microscopy pictures of CPC cells stably expressing (top) PalmGFP or (bottom) PalmtdTomato. Scale bars represent 400 µm. (B) Representative NTA plots showing the size distribution and particle concentration of EVs after SEC isolation of conditioned medium derived from PalmGFP+- and PalmtdTomato+ CPCs. (C) Western blot analysis showing the presence of CD81, Syntenin-1 (SYNT), AnnexinA1 (ANXA1), GAPDH, PalmGFP and/or PalmtdTomato, and absence of Calnexin (CNX) in (left) PalmGFP+- and (right) PalmtdTomato+ EVs. Cell lysates (CL) derived from (left) PalmGFP+ or (right) (TdTomato-negative) CPCs were included as control. Uncut blots are included in Supplementary Figure S10. (D) Representative TEM images of PalmGFP+ EVs at two different magnifications.

Fig 4: Both co-isolated proteins and EV-associated factors contribute to HMEC-1 activation.a Schematic overview of the EV isolation protocol to obtain pure Opti-EVs. SEC-EVs were loaded in the bottom of a discontinuous OptiprepTM gradient and ultracentrifuged for 16 h. b Resulting fractions (F1–8) were analysed for particle number by NTA (upper panel) and the presence of EV-marker proteins CD81, CD63 and Syntenin-1 (SYNT) by western blotting. Equal volumes of each sample were analysed. Iodixanol concentration was measured in each fraction (lower panel). Fractions 1–5 were pooled and concentrated (Opti-EVs). c Representative NTA plot showing the size distribution and particle concentration of SEC- and purified Opti-EVs. d Protein content per 1 × 1010 SEC- and Opti-EVs of three representative experiments. e Western blot analysis showing presence of CD81, SYNT, ß-actin (ß-ACT) in purified Opti-EV- and protein fractions, compared with crude SEC-EVs. Calnexin (CNX) was only present in CPC lysate (CL). Complete ß-ACT blot is displayed in Supplementary Fig. 10. f Representative TEM image of SEC- and Opti-EVs. g, h Representative western blot analysis of phosphorylated AKT (pAKT), total AKT (tAKT), phosphorylated ERK1/2 (pERK1/2) and total ERK1/2 (tERK1/2) in HMEC-1 treated with SEC-EVs and Opti-EVs normalized on total particle numbers, and with SEC-EVs, Opti-EVs and Opti-protein fraction normalized on total protein content. ß-ACT was included as housekeeping protein (I = phosphorylated protein blot, II = total protein blot). h Quantification of pAKT, tAKT, pERK1/2 and tERK1/2 expression levels using densitometry expressed as pAKT/AKT and pERK/ERK ratios (n = 3). Biological replicates of (g) are displayed in Supplementary Fig. 9e, f. i Wound healing experiment showing effects SEC- and Opti-EVs on HMEC-1 migration normalized both on total particle number and total protein content, analysed both as % wound closure and absolute migration distance. Addition of Opti-protein fraction was normalized on volume (n = 3, technical replicates. Data are representative of three biologically independent experiments). Data are presented as mean ± SD. *p < 0.033.

Fig 5: Generation of EVs expressing targeting peptides for endothelial cells. (A) Amino acid sequence of TP1, TP2, TP-FLAG, which was included as negative control, and TP-9R. (B) Schematic of constructs expressed on EVs consisting of a specific targeting peptide (TPx) connected to transmembrane protein Lamp2b, flanked by a N-terminal signal peptide (SP) and GNSTM glycosylation sequence, and C-terminal HA-tag. (C) Representative western blot analysis showing expression of HA-tag and PalmGFP in cell lysate (CL) harvested from CPC lines stably expressing PalmGFP+ and specific TPs. CPCs transduced with GFP were included as negative control. GAPDH and beta-actin (ß-ACT) were included as house-keeping proteins. (D) Representative NTA plots showing the size distribution and particle concentration of concentrated conditioned medium (CM) derived from PalmGFP+ TP-expressing CPCs. (E) GFP fluorescence per 1010 particles determined in CM derived from PalmGFP+ TP-expressing CPCs (n = 2). Data are displayed as mean ± SD. (F) Representative western blot analysis showing expression of HA-tag, PalmGFP, CD81, Syntenin-1 (SYNT) and AnnexinA1 (ANXA1) in CM. CL of PalmGFP+ CPCs stably expressing TP-FLAG were included as control. CM derived from CPCs transduced with GFP was included as negative control (cntrl). Calnexin (CNX) was only present in CL. Uncut blots are included in Supplementary Figure S10.