Src family kinases engage differential pathways for encapsulation into extracellular vesicles

Abstract

Extracellular vesicles (EVs) are heterogeneous biological nanoparticles secreted by all cell types. Identifying the proteins preferentially encapsulated in secreted EVs will help understand their heterogeneity. Src family kinases including Src and Fyn are a group of tyrosine kinases with fatty acylation modifications and/or multiple lysine residues (contributing charge interaction) at their N‐terminus. Here, we demonstrate that Src and Fyn kinases were preferentially encapsulated in EVs and fatty acylation including myristoylation and palmitoylation facilitated their encapsulation. Genetic loss or pharmacological inhibition of myristoylation suppressed Src and/or Fyn kinase levels in EVs. Similarly, loss of palmitoylation reduced Fyn levels in EVs. Additionally, mutation of lysine at sites 5, 7, and 9 of Src kinase also inhibited the encapsulation of myristoylated Src into EVs. Knockdown of TSG101, which is a protein involved in the endosomal sorting complexes required for transport (ESCRT) protein complex mediated EVs biogenesis and led to a reduction of Src levels in EVs. In contrast, filipin III treatment, which disturbed the lipid raft structure, reduced Fyn kinase levels, but not Src kinase levels in EVs. Finally, elevated levels of Src protein were detected in the serum EVs of host mice carrying constitutively active Src‐mediated prostate tumours in vivo. Collectively, the data suggest that different EVs biogenesis pathways exist and can regulate the encapsulation of specific proteins into EVs. This study provides an understanding of the EVs heterogeneity created by different EVs biogenesis pathways.

1. INTRODUCTION

Extracellular vesicles (EVs) are secreted from almost all cell types. EVs contain proteins, DNA, mRNA, microRNAs and other biological components (Kalluri & LeBleu, 2020; Kowal et al., 2014), which mediate cell‐to‐cell communication (Simons & Raposo, 2009; Villarroya‐Beltri et al., 2014). EVs are comprised of microvesicles through the budding of the cell membrane, exosomes through cell exocytosis originated from the fusion of multivesicular bodies with the plasma membrane, and apoptotic bodies when cells undergo apoptosis (Colombo et al., 2014; Keller et al., 2006; Thery et al., 2002). EVs regulate the cell microenvironment and facilitate communication networks for the dissemination of cancer cells (Kahlert & Kalluri, 2013). Tumour‐derived EVs also promote tumorigenesis through transferring EVs contents and/or escape from immune surveillance (Abusamra et al., 2005; Skog et al., 2008).

One of the EVs biogenesis pathways is regulated by the Endosomal Sorting Complex Required for Transport (ESCRT) protein complex. The ESCRT‐dependent pathway consists of four protein complexes: ESCRT‐0/I/II/III, which drive multiple vesicle body formation and scission (Kalluri & LeBleu, 2020). Among a complexity of the ESCRT family proteins, TSG101 belonging to the ESCRT‐I protein complex regulates the EVs biogenesis process and encapsulates itself into EVs as an EVs biomarker (Henne et al., 2011; Villarroya‐Beltri et al., 2014). Lipid rafts are cholesterol‐rich cytoplasmic membrane domains and participate in intraluminal vesicles formation, and thereby the EVs biogenesis process. Proteins associated with lipid rafts such as EGFR and PDGFR, and others have a preference to be encapsulated into EVs through this platform (Skryabin et al., 2020).

Src family kinases (SFKs) including Src and Fyn are a group of non‐receptor tyrosine kinases that are fundamental regulators of signal transduction involving cellular differentiation, adhesion, cell‐cycle progression, invasion, and migration (Erpel & Courtneidge, 1995; Fizazi, 2007; Parsons & Parsons, 1997). All SFKs members are composed of an N‐terminal Src Homology (SH) four domain controlling membrane association, SH3, SH2, and tyrosine kinase SH1 domains bearing an autophosphorylation site (e.g., Tyrosine 416 in Src in chicken), and a short C‐terminal tail containing an autoinhibitory phosphorylation site (e.g., Tyrosine 527 in Src in chicken) (Sicheri & Kuriyan, 1997; Sicheri et al., 1997; Xu et al., 1997). Src and Fyn kinases are myristoylated and/or palmitoylated at the N‐terminus (Robbins et al., 1995; Shenoy‐Scaria et al., 1994). Myristoylation is a lipid modification of protein in which a myristoyl group, a 14‐carbon saturated fatty acid, is attached to the N‐terminal glycine (Wright et al., 2010). Palmitoylation is the covalent attachment of palmitic acid to a cysteine residue (Resh, 1999). In addition, the N‐terminus of Src kinase contains multiple lysines at sites 5, 7, and 9. These basic residues (positive charge) facilitate the association of Src kinase to the cytoplasmic membrane through the charge interaction with the phospholipid (negative charge) (Resh, 1994). Therefore, the fatty acylation modifications and the charge interactions promote protein association with the cytoplasmic membrane including lipid rafts and facilitates diverse biological actions including cell signalling, protein localization, and cell‐cell communication (Casey, 1995; Resh, 1999; Resh, 2004; Shen et al., 2011).

While Src kinase activity is regulated through myristoylation and lysine‐mediated interaction with the cytoplasmic membrane, Fyn kinase is myristoylated and palmitoylated due to the presence of N‐terminal cysteine residues (Flaumenhaft & Sim, 2005; Milligan et al., 1995; Resh, 1994). We reported that fatty acylation of Src kinase is essential for its enrichment in EVs at the American Association for Cancer Research (AACR) annual meeting in 2018 (Wen et al., 2018). Here, we demonstrate the role of these modifications on the encapsulation of Src and Fyn kinase into EVs. Both lysine residues and fatty acylation including myristoylation and palmitoylation enhanced Src/Fyn kinase encapsulation into EVs. However, the encapsulation of Src and Fyn kinases into EVs was regulated by different biogenesis pathways. Additionally, elevated Src levels were detected in the serum EVs of tumour‐bearing hosts expressing constitutively active Src kinase in vivo, suggesting a potential biomarker for invasive tumours.

2. EXPERIMENTAL SECTION

2.1. Cell lines and cell culture

SYF1 (Src^−/−Yes^−/−Fyn^−/−), 293T, and human prostate cancer cell lines including DU145, PC‐3, 22RV1 and LNCaP were purchased from American Type Culture Collection. The cells were used for up to 20 passages. Mycoplasma detection assays were performed periodically to check for contamination.

2.2. Plasmid constructs

Lentiviral vectors Src(WT), Src(G2A), Src(K5A/K7A/K9A), Src(Y529F), Src(Y529F/G2A), Fyn(WT), Fyn (G2A), Fyn (C3S/C6S) were cloned using the FUCRW parental vector as previously reported (Cai et al., 2010; Kim, Alsaidan et al., 2017). Additionally, lentiviral vectors Src (Kahlert & Kalluri, 2013)‐RFP and Fyn (Kahlert & Kalluri, 2013)‐RFP were created by fusion of the first eight amino acids of Src and Fyn to the N‐terminus of the RFP gene by PCR. The shRNA targeting TSG101, ShRNA‐TSG101‐1 and ShRNA‐TSG101‐2, were designed to target the sequences 5′‐CGGACTGGACACATACCCATATAACTCGAGTTATATGGGTATGTGTCCAGTTTTTTG‐3′, and 5′‐CCGGGCCTTATAGAGGTAATACATACTCGAGTATGTATTACCTCTATAAGGCTTTTG‐3′, respectively.

2.3. Isolation of EVs from conditioned medium through sequential ultra‐centrifugation

Cells were grown in ATCC recommended media in 5% CO₂ at 37°C. After reaching 90% confluence, the medium was replaced with medium containing 5% exosome‐free FBS (Life Technology Inc.). After 24 h, the conditioned medium was collected for isolation of the EVs. Specifically, supernatants were collected after a series of centrifugations at 300 × g at 4°C for 10 min, 2000 × g at 4°C for 10 min, and 10,000 × g at 4°C for 30 min to remove live cells, dead cells, and cell debris, respectively. The supernatant was finally subjected to ultracentrifugation at 100,000 × g at 4°C for 90 min. The EVs pellet was washed in 1X PBS, and ultracentrifuged at 100,000 × g at 4°C for 90 min. The pelleted EVs were re‐suspended either in 1X PBS for EVs characterization or lysed in RIPA buffer for protein analysis.

2.4. EVs characterization

The size distribution, particle number, and zeta potential of EVs were measured using the ZetaView PMX110 (Particle Metrix, Meerbusch, Germany) and corresponding software ZetaView 8.02.28 as previously described (Helwa et al., 2017; Shah et al., 2018). For each sample, 5–10 μL of the isolated EVs were diluted into 2 mL of 1X PBS and loaded into the ZetaView cell for analysis to obtain the diameter (mode) and particle concentration. Another 5–10 μL of the extracted EVs were diluted into 2 ml of 0.05X PBS and loaded into the ZetaView cell for zeta potential measurements as previously described (Helwa et al., 2017).

2.5. Negative staining of EVs by transmission electron microscopy (TEM)

TEM imaging of the isolated EVs was performed at the Electron Microscope and Histology Core at Augusta University in Augusta, Georgia as previously described (Helwa et al., 2017; Shah et al., 2018). Freshly isolated EV suspensions were applied to copper mesh Formvar coated carbon stabilized grids, fixed in 4% paraformaldehyde for 1 h, and stained with 1% aqueous uranyl acetate. After air drying, TEM examination was performed using a JEM 1230 transmission electron microscope (JEOL USA Inc., Peabody, MA, USA) at 110 kV and imaged with an UltraScan 4000 CCD camera & First Light Digital Camera Controller (Gatan Inc., Pleasanton, CA, USA).

2.6. EVs characterization by iodixanol flotation density gradient ultracentrifugation

Iodixanol flotation density gradient ultracentrifugation was done as previously described (Klingeborn et al., 2017). Briefly, the PBS‐washed 100,000 x g EVs pellet that was isolated as described above was resuspended in 0.33 mL PBS followed by addition of 0.67 mL of a 60% (w/v) iodixanol solution (OptiPrep™; Sigma‐Aldrich #D1556) to make up the bottom 1.0 mL 40% fraction in a 4.0 mL ultracentrifuge tube (UltraClear™ 11 × 60 mm; Beckman Coulter #344062). A discontinuous gradient of iodixanol solutions was prepared by carefully overlaying the bottom fraction with 1 mL of 20%, 10%, and 5% solutions of iodixanol buffered with 0.25 M sucrose, 10 mM Tris‐HCl (pH 7.5). The gradient tubes were subjected to centrifugation at 369,734 xg_avg (SW 60 Ti rotor; 60,000 rpm) for 2 h at 8˚C using a Beckman Coulter Optima XE‐90 Ultracentrifuge. Twelve 333 μL fractions were collected manually from the top of the self‐generated gradient. A mock gradient with PBS instead of EVs suspension was centrifuged alongside the experimental gradients and collected fractions were diluted four‐fold in water and absorbance was measured at 340 nm to determine density as previously described (Van Deun et al., 2014). Fractions with known light EV/exosome densities (1.07–1.11 g/mL; fractions 6–8) were diluted to 4.0 mL with PBS, placed in polyallomer tubes (Beckman Coulter #328874) and subjected to centrifugation at 100,000 xg_avg (31,200 rpm) for 1 h at 8°C in the SW 60 Ti rotor. Pellets were lysed in 50 μL 2X XT sample buffer (Bio‐Rad #1610791) and stored at −80°C until use.

In accordance with the guidelines in Minimal information for studies of extracellular vesicles 2018 (MISEV2018) (Thery et al., 2018) the presence of exosomes in light EVs density fractions were confirmed by immunoblotting using antibodies to exosome markers Syntenin‐1 (OriGene #TA347044, clone [3D9‐G9‐H4]; 1:1000), TSG101 (BD Biosciences #612697, clone [51/TSG101]; 1:500), and EVs marker ANXA2 (BD #610069, clone [5/Annexin II]; 1:500). Lack of cellular contamination was confirmed by the absence of the ER marker Calreticulin (Cell Signaling Technology #12238, clone [D3E6]; 1:2000) in the EV‐enriched fractions.

2.7. Protein extraction from cells or EVs

Cells were lysed in cold 1X radioimmunoprecipitation assay (RIPA) buffer [50 mM Tris‐base (pH 7.4), 1% NP‐40, 0.50% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 2 mM EDTA and protease inhibitor (1X)] containing a cocktail of protease and phosphatase inhibitors. The EVs proteins were extracted by adding an equivalent volume of 2X RIPA buffer containing a cocktail of protease and phosphatase inhibitors. The samples were kept on ice and vortexed every 5 min for a total of 30 min. The protein samples were sonicated and centrifuged at 14,800 rpm at 4°C for 15 min.

2.8. DDD85646 inhibited Src levels in cytoplasmic membrane or EVs

Cells were subjected to membrane fractionation using a Mem‐PER Plus Membrane Protein Extraction Kit (Cat#: 89842; Thermo Fisher Scientific). This membrane extraction kit can specifically extract proteins associated with the plasma membrane or other membranes such as the ER and Golgi membrane and so on. In brief, DU‐145 cells were grown in 6 cm dishes and treated with the indicated concentration of DDD85646 for 24 h. Cells were pelleted and washed with the provided Cell Wash Solution before adding Permeabilization Buffer supplemented with protease and phosphatase inhibitors. The lysates were incubated on ice for 10 min and vortexed every 2 min. The lysates were centrifuged at 21,000 x g for 15 min at 4°C. The supernatants containing cytosolic proteins were collected and the pellets were resuspended in Solubilization Buffer supplemented with protease and phosphatase inhibitors. Samples were kept on ice for 30 min and vortexed every 2 min. Lysates were further centrifuged at 21,000 x g for 15 min. The supernatants containing solubilized membrane and membrane‐associated proteins were collected.

2.9. Measurement of protein concentration (by detergent compatible protein assay)

The protein concentration of EVs or total cell lysates (TCL) was determined by detergent compatible protein assay (Bio‐Rad Laboratories). The tested cells or EVs were lysed in RIPA buffer. Protein standards and samples (5 μL) were added into a 96‐well plate, followed by a mixture of Reagent A and S in a ratio of 50:1 (25 μL) and Reagent C (200 μL). The absorbance at 750 nm was measured after 15 min incubation at room temperature. The protein concentration was calculated based on the standard curve.

2.10. Immunoblotting anlaysis

TCL and/or EV lysates dissolved in RIPA buffer were subjected to SDS‐PAGE, and transferred onto nitrocellulose membranes. To determine the enrichment of Src and/or Fyn protein in EVs, the same amount of total protein in the TCL and EVs lysates were loaded in the gel. The membranes were incubated with the primary antibody (dilution factor depending on an antibody) at 4°C overnight. Membranes were washed with TBST followed by incubation with the secondary antibody at room temperature for 1 h. Blots were imaged with enhanced chemiluminescence (ECL) reagent. The bands of interest were quantified by Image J software. To examine if Src/Fyn levels were enriched in EVs, the ratio of expression levels of Src/Fyn and their mutants in EVs versus TCL were calculated and compared. Additionally, to confirm that the same amount of protein from TCL or EVs lysates were loaded, a separate SDS‐PAGE gel was subjected to Coomassie staining.

Rabbit anti‐Src (catalog# 2109), rabbit Ab anti‐calnexin (catalog# 2679), rabbit Ab anti‐CD‐9 (catalog# 13403 for human species,), rabbit Ab anti‐GAPDH (catalog# 2118), rabbit Fyn(catalog# 4023), rabbit Ab CD81 (catalog# 10037), rabbit Ab Flotillin‐1 (catalog# 3253), rabbit Caveolin‐1 (catalog# 3267), rabbit EGFR (catalog# 4267), rabit Non‐P‐Src (Tyr 527) (catalog# 2107), rabbit HSP90 (catalog# 4877), rabbit gamma‐Tubulin (catalog# 5886) were purchased from Cell Signaling Technology. Rabbit anti‐myristoyl‐Src antibody was generated by our lab (Ma et al., 2021). Mouse anti‐Src Antibody (Catalog# 60315‐I‐Ig, Proteintech), Rabbit anti‐Syntenin Antibody (catalog# ab 19903, Abcam), Mouse anti‐CD63 Antibody (Catalog# 556019, BD Pharmingen), TSG‐101 Antibody (Catalog# GTX70255, GeneTex), Rabbit anti‐RFP Antibody (catalog# 600‐401‐379, Rockland Inc), rabbit anti‐AR Antibody (catalog# sc‐816, Santa Cruz Biotechnology), and secondary antibody anti‐rabbit IgG HRP‐linked Antibody (7074, Cell Signaling Technology) secondary antibody anti‐mouse IgG HRP‐linked Antibody (7076, Cell Signaling Technology) were used according to manufacturer's recommended dilution.

2.11. Determination of palmitoylation or myristoylation by click chemistry

SYF1 cells expressing Fyn(WT) kinase or Fyn(C3S/C6S) were plated in 100‐mm petri dishes and grown with/without 50 μM of 17‐octadecynoic acid‐azide (an analogue of palmitic acid) until 90% confluence in EMEM medium with 5% PBS. The cells or EVs were lysed in M‐PER buffer (Thermo Scientific) containing protease and phosphatase inhibitors.

The cell lysates (including palmitoylated proteins labelled with azide) (20 μg protein) were added into a working solution containing biotin‐alkyne (0.1 mM), CuSO₄ (1 mM), TCEP (1 mM) and TBTA (0.1 mM) and incubated at room temperature for 1 h. Due to the reaction of azide‐alkyne, the palmitoylated proteins were labelled with biotin, which was detected by streptavidin‐HRP via immunoblotting.

The DU145 cells expressing Src(G2A) or Src(WT) were plated in 150 mm petri dishes and grown until 90% confluency in EMEM medium containing 5% FBS. The medium was replaced with EMEM medium containing exosome‐free FBS and 50 μM of myristic acid‐azide (an analogue of myristic acid). The cells were grown for another 24 h and the conditioned medium was collected and used for EVs isolation by the sequential ultracentrifuge method. The cells or EVs were lysed in M‐PER buffer (Thermo Scientific) containing protease and phosphatase inhibitors. The cell lysates or EVs lysate (10 μg protein) were added into a working solution containing biotin‐alkyne (0.1 mM), CuSO₄ (1 mM), TCEP (1 mM) and TBTA (0.1 mM) and incubated at room temperature for 1 h. After the Click reaction, the samples were mixed with loading dye and incubated at 95°C for 5 min. The lysates were subjected to SDS‐PAGE and transferred onto nitrocellulose membranes. After blocking with 5% milk overnight, the membrane was washed and incubated with High Sensitivity Streptavidin‐HRP (catalogue No. 21130, ThermoFisher Scientific) at room temperature for 1 h. Blots were imaged with ECL reagent for analysis of myristoylated Src kinase.

2.12. Lipid raft disruption

PC‐3 and DU145 cells were grown in 15 cm petri dishes overnight. The medium was replaced with the same growth medium but containing exosome‐free FBS and DMSO (the control) or filipin III (0–1 μM) to disrupt the lipid rafts for 24 h. The EVs were collected from the conditioned medium. The EVs and cells were lysed with RIPA buffer and subjected to immunoblotting analysis.

2.13. Haematoxylin and eosin (H&E) staining and immunohistochemistry

The tissue samples were fixed with PBS buffered 10% formaldehyde. The samples were further paraffin‐embedded and sectioned using a Leica RM2235 Rotary Microtomy to 4 μm thickness and mounted on microscope slides (catalogue No. 12‐550‐15, Fisher Scientific). The tissue sections on slides were baked for 60 min at 65°C, and de‐paraffined in 100% xylene for 5 min (2x), dehydrated in 100% ethanol for 5 min (2x), 95% ethanol for 5 min (2x), and 70% ethanol for 5 min. After washing with PBS for 10 min (3x), the tissue sections were subjected to H&E staining or immunohistochemistry analysis.

For H&E staining, the above samples were stained in Ehrlich's Haematoxylin for 5 min and washed with distilled water (3x), followed by 5–6 quick dips in acidic alcohol (0.3%) to differentiate and then washed thoroughly with distilled water (3x). The tissue sections were dipped into Scott's Tap Solution for 2 min and rinsed thoroughly with distilled water (3x). Counterstaining was performed using Eosin solution for 2 min and then washed with distilled water (3x), followed by dehydration in 95% alcohol for five dips (2x) and 100% alcohol for five dips (2x). After xylene clearing for 1 min (3x), tissue sections were mounted with a coverslip in the mounting medium.

For immunohistochemistry, the tissue slides were subjected to antigen retrieval by treating samples in 0.01 M citrate buffer (pH 6.0) in a steamer cooker at a microwave with 60% power for 15 min and 10% power for 15 min. After cooling, tissue slides were washed with PBS for 10 min (2x). The tissues were circled with a PAP Pen liquid blocker (Part # 6505, Newcomer Supply). Next, 300 μL of 0.3% H₂O₂ in distilled water was added to each tissue spot for 5–10 min and then washed with PBS for 10 min (3x). The tissues were blocked in 2.5% goat serum in PBS for 1 h at room temperature, and then incubated with primary antibody (Src 1:250, RFP 1:250, cleaved caspase 3 1:200, and cleaved PARP 1:200) in PBST overnight at 4°C. The tissue slides were washed with PBST for 10 min (3x), and then incubated with secondary antibody (catalogue #: M7401) in PBST at room temperature for 1 h. After washing with PBS for 10 min (3x), the tissues slides were incubated with DAB solution (catalogue #: SK‐4100) for development. As soon as brown colour appeared under a microscope, the reaction was stopped by dipping the slide into distilled water. The time to develop for control and treatment was kept the same. The tissue slides were stained in haematoxylin for 1 min and washed with distilled water (3x), then immersed in NaHCO₃ solution for 3 min and washed with distilled water (3x). The tissue slides were again dehydrated by treating samples in a series of alcohol solutions (75%, 95%, 100% ethanol for 5 min x 2), and then air dried for 10 min. After treating with xylene for 5 min (2x), the tissue sections were air dried for 10 min, and mounted with the mounting medium and coverslip.

2.14. EVs isolation from plasma and analysis

Blood samples were collected from mice and the supernatant was collected after centrifugation at 2000 × g for 10 min. The EVs were isolated from plasma by the ExoQuick kit according to the manufacturer's instructions (catalogue No. EXOQ5A‐1, System Biosciences). In brief, ExoQuick Reagent (63 μL) was added to serum samples (250 μL) and incubated at 4°C for 30 min. The pellet EVs were collected by centrifugation at 1500 × g for 30 min and re‐suspended in RIPA or PBS buffer. The EVs re‐suspended in the PBS buffer were used for its characterization including the size distribution and zeta potential analysis by NTA assay (Malvern, USA). The EVs lysed in the RIPA buffer was subjected to immunoblotting analysis.

2.15. Xenograft tumours in mice

DU145 cells were transduced with vector control, Src(Y529F), or Src(Y529F/G2A). Male SCID mice at the age of 8−10 weeks were randomly divided into four groups (three mice per group). DU145 cells (10,000 cells/mice) expressing vector control, Src(Y529F), or Src(Y529F/G2A) were implanted sub‐renally in mice by survival surgery. Additionally, one group without implantation was also set up as a control. The mice were routinely examined and euthanized 5 weeks after implantation.

In order to generate similar size of xenograft tumours in Src(Y529F) and Src(Y529F/G2A) groups, 1.5 × 10⁵ cells of the DU145 cells or DU145 cells over‐expressing Src(Y529F/G2A), and 1.5 × 10⁴ cells of the DU145 cells expressing Src(Y529F) (10 times less amount) were mixed with collagen and implanted sub‐renally in SCID mice (3 months‐old) by survival surgery. After 4 weeks, the mice were sacrificed, and tumors and serum samples were harvested. All animals had access to diet and water throughout the experiments, and all protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of University of Georgia.

2.16. Statistical analysis

The data were presented as mean ± S.E.M (standard error of the mean). All the data with more than two groups were analysed by one‐way ANOVA with a post hoc Tukey test in GraphPad Prism 8 software, and two values were compared by an unpaired student t‐test. *p < 0.05; **p < 0.01; ***p < 0.001; ns, non‐significant.

3. RESULTS

3.1. Src family kinase members including Src and Fyn kinases were preferentially encapsulated into EVs

We have previously reported that myristoylated proteins are preferentially enriched in EVs in cell lines, thymus, breast milk, and urine (Whitley et al., 2022). Src family kinases (SFKs) are myristoylated, and/or palmitoylated, and highly expressed in prostate cancer cells (Kim, Alsaidan et al., 2017; Posadas et al., 2009). We examined if SFKs including Src and Fyn kinase were preferentially encapsulated into EVs of the prostate cancer cell lines DU145, PC‐3, 22Rv1 and LNCaP cells. The EVs derived from the above cells were isolated from conditioned media by sequential ultracentrifugation and characterized by size (∼140 nm) and zeta potential (∼‐30 mV to ∼‐60 mV) with typical cup shaped morphology observed using transmission electronic microscopy (Figure 1a,b). The purity of EVs derived from 22Rv1 and DU145 cells was further characterized by iodixanol density gradient centrifugation in EVs. Alongside with syntenin and ANXA2, Src kinase was detected in 1.077 and/or 1.106 g/mL density fractions where EVs are presumably localized (Figure 1c). While calnexin was only detected in total cell lysates (TCL), but not in EVs, EVs protein biomarkers including CD9, syntenin and flotillin‐1 were primarily detected in EVs of the tested four cell lines (Figure1d). In contrast, Src kinase was expressed in both EVs and TCL. In particular, myristoylated Src levels were highly enriched in EVs (Figure 1d). Additionally, of note, although equal amounts of EVs protein was loaded, expression levels of CD9, syntenin, and flotillin‐1 in EVs lysates varied among different cancer cell lines, suggesting the diversity of the expression pattern in EVs protein biomarkers.

FIGURE 1.

Src family kinases were preferentially encapsulated into extracellular vesicles (EVs) in prostate cancer cell lines. (a,b) Prostate cancer cells including DU145, PC‐3, 22Rv1 and LNCaP were cultured in medium containing 10% exosome‐free FBS for 24 h at 37°C in 5% CO2. EVs were isolated from the conditioned medium by sequential ultracentrifugation. The size distribution (a) and zeta potential (b) of EVs were measured by the nanoparticle tracking analysis (NTA) assay. EVs images were taken by transmission electron microscopy (TEM). (c) The isolated EVs from 22Rv1 and DU145 cells were subjected to iodixanol density gradient ultracentrifugation. The lysates of EVs at iodixanol density 1.075, 1.077, and 1.106, which contain EVs, were analysed for protein levels of syntenin and ANXA2 (biomarkers of EVs), and calreticulin (EVs negative protein biomarker) by immunoblotting. Total cell lysate (TCL) of ARPE‐19 cells was used as a control. (d) Levels of Src kinase, myristoyl‐Src (mSrc), AR, calnexin, syntenin, CD9, and flotillin‐1 in TCL and EVs of DU145, PC‐3, 22Rv1, and LNCaP cells were analysed by immunoblotting. Equal amount of protein (10 μg) was loaded for the TCL or EVs lysates. (e) SYF1 (Src−/−Yes−/−Fyn−/−) (mouse fibroblast) cells were transfected with Fyn(WT) by lentiviral infection. The cell line stably expressing Fyn(WT) was designated as SYF‐Fyn(WT). The SYF‐Fyn(WT) cells served as a positive control for Fyn expression. Levels of Fyn kinase, calnexin, GAPDH, syntenin, CD9, CD81, and CD63 in TCL and EVs of SYF‐Fyn(WT), PC‐3, and 22Rv1 cells were analysed by immunoblotting. Equal amounts of protein (10 μg) from the TCL or EVs were loaded.

Next, we examined if Fyn kinase was encapsulated into EVs in a similar fashion as Src kinase. As a positive control, Fyn was over‐expressed in SYF (Src^−/−Yes^−/−Fyn^−/−) cells (a murine cell line). Fyn kinase was highly expressed in TCL and EVs (Figure 1e, Lanes 1 and 4). Enrichment of syntenin and CD81 levels, and reduced levels of GAPDH confirmed successful isolation of EVs. Of note, while the CD9, CD63, and calnexin antibodies are specific for the human antigen and not murine, the CD81 antibody can only detect the murine antigen, but not human (Figure 1e). To examine encapsulation of endogenous levels of Fyn kinase into EVs, PC‐3 and 22Rv1 cells were used (Figure 1e, Lanes 2–3 and 5–6). While Fyn kinase levels were higher in the EVs lysate compared to the total cell lysate of PC‐3 cells, Fyn kinase levels in the EVs lysate was similar to that in the TCL of 22Rv1 cells. The presence of EVs protein markers including CD9, syntenin, and CD63 and undetectable levels of calnexin in EVs indicated the successful isolation of EVs from PC‐3 and 22Rv1 cells (Figure 1e, lanes 2–3 and 5–6). Together, the data suggest that Src and Fyn kinases are capable of encapsulation into EVs.

3.2. Loss of myristoylation inhibited encapsulation of Src kinase into EVs

Fatty acylation of proteins including myristoylation and palmitoylation is essential for the association of the protein with the cytoplasmic membrane (Resh, 1999). We further examined the functional role of myristoylation in regulating Src kinase encapsulation into EVs. DU145 and SYF1 cells were transfected with Src(WT), Src(G2A) (loss of myristoylation mutant), Src(Y529F) a constitutively active mutant, or Src(Y529F/G2A) (constitutively active and loss of myristoylation mutant) by lentiviral infection (Figure 2a). The size distribution and cup shaped morphology of the isolated EVs from SYF1 cells expressing Src(Y529F) or DU145 cells expressing Src(WT) cells (Figure 2b) were similar to those from DU145 cells or others (Figure 1a). The quality of the isolated EVs from DU145 cells was demonstrated by expression levels of protein markers including syntenin, flotillin‐1, and CD9, and undetectable levels of calnexin in EVs (Figure 2c, lanes 6–10). Similarly, successful isolation of EVs from SYF1cells was shown by elevated levels of protein markers including syntenin, flotillin‐1, and CD81, and undetectable levels of GAPDH (Figure 2d, lanes 6–10). Coomassie blue staining confirmed that the same amount of protein was loaded in different groups of both cellular lysates and EVs proteins (Figure 2c,d).

FIGURE 2.

Loss of myristoylation inhibited encapsulation of Src kinase into EVs. (a) Schematic diagram of Src(WT), Src(G2A), Src(Y529F), and Src(Y529F/G2A) mutants. (b) DU145 and SYF1 cells were transduced with Src(WT), Src(G2A), Src(Y529F), or Src(Y529F/G2A) by lentiviral infection. The size distribution of EVs derived from DU145 expressing Src(WT) [DU145Src(wt)] and SYF1 expressing Src(Y529F) [SYF‐Src(Y529)] were measured by the nanoparticle tracking analysis (NTA) assay. EVs images were taken by transmission eletron microscopes (TEM). C‐D) DU145 cells expressing control vector, Src(WT), Src(G2A), Src(Y529F), or Src(Y529F/G2A) (c) and SYF1 cells expressing control vector, Src(WT), Src(G2A), Src(Y529F), or Src(Y529F/G2A) (d) were grown in exosome‐free FBS medium at 37°C in 5% CO2. EVs were isolated from the conditioned medium by sequential ultracentrifugation. Levels of Src in the EVs and TCL of the transfected cells were determined by immunoblotting. Equal amounts of protein (10 μg) from the TCL or EVs were loaded. Levels of Src kinase, myristoyl‐Src (mSrc), calnexin or GAPDH, syntenin, CD9 or CD81, and flotillin‐1 in EVs lysates and TCL were analyzed by immunoblotting. Coomassie blue staining of EVs lysates and TCL was performed to demonstrate equal loading. Levels of Src protein was quantified by Image J software. To compare the enrichment of Src levels, the ratio of Src levels in EVs versus those in TCL was calculated and compared among the Src(WT), Src(G2A), Src(Y529F), and Src(Y529F/G2A) groups. *p < 0.05; **p < 0.01; ***p < 0.001.

As expected, over‐expression of Src kinase was detected in the total cell lysate (TCL) of DU145 cells (Figure 2c, lanes 1–5) and SYF1 cells (Figure 2d, lanes 1–5) expressing Src(WT), Src(G2A), Src(Y529F), or Src(Y529F/G2A). Myristoylated Src was only detected in cells expressing Src(WT) or Src(Y529F) in DU145 cells (Figures S1a and 2c, lanes 2 and 4) or SYF1 cells (Figure 2d, lanes 2 and 4), but not Src(G2A) or Src(Y529F/G2A) in DU145 cells (Figures S1a and Figure 2c, lanes 3 and 5) or in SYF1 cells (Figure 2d, lanes 3 and 5). Loss of myristoylation in the Src(G2A) mutant or inhibition of Src myristoylation inhibited its expression levels in the detergent insoluble membrane fraction (Figure S1b).

Importantly, given that the same amount of total protein was loaded, the ratio of Src kinase levels in EVs versus the cell lysate was significantly elevated in cells expressing Src(Y529F) (Figure 2c lane 9/lane 4, or Figure 2d lane 9/lane 4) compared to Src(WT) (Figure 2c lane 7/lane 2, or Figure 2d lane 7/lane 2) in both DU145 (Figure 2c) and SYF1 cells (Figure 2d). Src levels in EVs were significantly decreased in Src(G2A) (Figure 2c, lane 8; or Figure 2c lane 8; Figure S1a) and Src(Y529F/G2A) (Figure 2c, lane10; or Figure 2d lane 10) mutants in comparison with the Src(WT) (Figure 2c, lane7; or Figure 2d lane 7) and Src(Y529F) (Figure 2c, lane 9; or Figure 2d lane 9), respectively. Of note, the detection of Src in EVs of the control group (DU145 cells only) reflected the endogenous levels of Src kinase enriched in EVs (detected by myristoylated‐Src antibody) (Figure 2c, lane 6, 8, and 10) or by Click chemistry (Figure S1a).

Next, we examined if Src or Fyn kinase was ubiquitously expressed in EVs of other cancer cells or murine cells. A proteomics analysis on EVs of sixty cancer cell lines was studied by Dr. Meckes Jr.’s group (Hurwitz et al., 2016). While peptides of Src kinase were detected in 42 out of 60 cancer cell lines, Fyn kinase were identified in 10 out of 60 cancer cell lines (Figure S2a). Additionally, Src kinase enrichment in EVs was also examined in 3T3 and SYF1 cells (murine cell lines). Src levels in EVs were elevated in the Src(WT) group compared with those in the Src(G2A) group (Figure S2b).

Collectively, the data clearly suggest that wild type Src is favourable for encapsulation into EVs, with constitutively active Src more favourable than wild type Src. The enrichment of Src kinase in EVs occurred in a majority of cancer cell lines. Furthermore, myristoylation facilitates the encapsulation of Src kinase into EVs since loss of myristoylation reduces the levels of Src encapsulated in EVs.

3.3. NMT1 inhibitor regulated myristoylated Src levels and the encapsulation of Src kinase into EVs

Numerous NMT1 inhibitors with various inhibitory efficiencies have been reported (Kallemeijn et al., 2019; Kim, Alsaidan et al., 2017). Among those, DDD85646 was very effective at inhibiting global protein myristoylation (Kallemeijn et al., 2019). The NMT1 inhibitor DDD85646 did not change total Src levels in cell lysates (Figure 3a). However, the inhibitor suppressed Src levels in the cytoplasmic membrane fraction (Figure S3) and in a concentration dependent manner (Figure 3a). Importantly, while the inhibitor had no effect on the levels of EVs protein markers including syntenin, CD9, and a slight decrease of flotillin‐1, Src kinase and myristoylated Src levels were significantly decreased in EVs (Figure 3b). The data suggest that pharmacological inhibition of Src myristoylation by targeting NMT1 affects the encapsulation of Src kinase into EVs.

FIGURE 3.

NMT1 inhibitor and loss of lysine residues at the N‐terminus of Src suppressed its myristoylation, the association with the cytoplasmic membrane, and encapsulation into EVs. (a) DU145 cells were treated with DMSO control or 50, 300, and 1000 nM of DDD85646 for 24 h. Protein lysates of treated cells were fractionated into the cytoplasmic membrane and cytosol fractions. Levels of Src kinase, GAPDH, or membrane protein biomarker (caveolin‐1) in total cell lysate (TCL), the cytoplasmic membrane and cytosol fractions were determined by immunoblotting. (b) EVs from 50 or 300 nM treated cells were isolated using sequential ultracentrifugation. Levels of Src kinase, myristoyl‐Src, and EVs protein biomarkers (syntenin, CD9, and flotillin) in EVs and total cell lysate was measured by immunoblotting. (c) Schematic diagram of Src(WT), Src(G2A), and Src(K5A/K7A/K9A) mutants. 293T cells were transduced with Src(WT), Src(G2A), or Src(K5A/K7A/K9A) by lentiviral infection. The transfected cells were grown in 5% FBS medium at 37°C in 5% CO₂. EVs were isolated from the conditioned medium by sequential ultracentrifugation. Protein expression levels in the EVs and TCL were determined by immunoblotting. Equal amounts of protein from the TCL or EVs were loaded. Levels of Src kinase, myristoyl‐Src (mSrc), GAPDH, calnexin, syntenin, CD9, and flotillin‐1 in TCL and EVs lysates were analyzed by immunoblotting. Of note, 1–4 represent 293 cells and 293T cells expressing Src(WT), Src(G2A), or Src(K5A/K7A/K9A), respectively. Levels of Src protein was quantified by Image J software. The ratio of Src levels in EVs relative to those in TCL was calculated. *p < 0.05; **p < 0.01; ***p < 0.001.

3.4. Loss of lysine residues at the N‐terminus of Src kinase inhibited myristoylated Src encapsulation into EVs

It has been reported that multiple lysine residues at the N‐terminus of Src kinase contribute to the interaction with the negatively charged phospholipid, which promotes its association with the cytoplasmic membrane (Resh, 1994). We investigated the functional role of lysines at sites 5, 7 and 9 in EVs encapsulation. The triple mutant Src(K5A/K7A/K9A) was created in which lysines were mutated to alanine (Figure 3c). Expression levels of the Src(K5A/K7A/K9A) mutant was not affected by the mutations as that observed with Src(WT) or Src(G2A) and was still myristoylated unlike in Src(G2A) (Figure 3c, lanes 2 and 4). While Src(K5A/K7A/K9A) mutant was observed in EVs the myristoylated Src levels was reduced compared to Src(WT) (Figure 3c, lane 4/lane 2 compared with lane 8/lane 6). The result suggests that multiple lysines at the N‐terminus play a role in myristoylated Src encapsulation into EVs as a cargo.

3.5. Loss of palmitoylation inhibited encapsulation of Fyn kinase into EVs

Fyn kinase is acylated by both myristic and palmitic acids at the N‐terminal glycine and cysteine residues, respectively (Aicart‐Ramos et al., 2011). Our previous study indicated that the Fyn(C3S/C6S) mutant resulting in loss of palmitoylation by mutation of cysteines at sites 3 and 6 sites (Figure 4a) elevated its kinase activity (Cai et al., 2011). Palmitoylation was confirmed in SYF1 cells expressing control vector, Fyn(WT), or Fyn(C3S/C6S) (Figure 4b). Similarly, palmitoylation was also detected in cells expressing a fusion protein of Fyn (Kahlert & Kalluri, 2013)‐RFP (a fusion of the first 8 amino acids at the N‐terminus of Fyn kinase with RFP), but not Src (Kahlert & Kalluri, 2013)‐RFP (a fusion of the first 8 amino acids at the N‐terminus of Src kinase with RFP) (Figure S4).

FIGURE 4.

Palmitoylation promoted encapsulation of Fyn kinase into EVs. (a) Schematic diagram of Fyn(WT), Fyn(G2A) and Fyn(C3S/C6S) mutants. Fyn(G2A) and Fyn(C3S/C6S) mutants lead to loss of myristoylation and palmitoylation in Fyn kinase, respectively. (b) Detection of palmitoylation in cells expressing Fyn kinase. SYF1 cells were transduced with Fyn(WT) or Fyn(C3S/C6S) by lentiviral infection. The transduced cells were grown in the medium with/without 17‐octadecynoic acid‐azide (palmitate probe). Due to the incorporation of the palmitate probe, palmitoylated proteins were labelled with an azide group. The cell lysates (including palmitoylated proteins labelled with the azide) were reacted with alkyne‐biotin via Click Chemistry resulting in the palmitoylated proteins being labelled with biotin, which was detected by immunoblotting using streptavidin‐HRP. Levels of Fyn and GAPDH in cell lysates were also examined by immunoblotting. (C,D) SYF1 cells were transduced with Fyn(WT), Fyn(G2A), and Fyn(C3S/C6S) by lentiviral infection. The transduced cells were grown in exosome‐free medium for 24 h at 37°C in 5% CO_2. EVs were isolated from the conditioned medium by sequential ultracentrifugation. The EVs were subjected to size distribution and transmission electron microscope analysis (c). Additionally, expression levels of Fyn, GAPDH, flotillin‐1, syntenin, and CD81 in total cell lysates (TCL, lanes 1–4) or EVs lysates (lanes 5–8) were analysed by immunoblotting (d). Ten μg of protein of TCL and EVs were loaded for protein analysis. The data represent one of three experiments.

To address the role of palmitoylation in regulating the encapsulation of Fyn kinase into EVs, particle size distribution, TEM, and expression of positive protein markers of flotillin‐1, syntenin, and CD81, and negative marker of GAPDH was used to assess the quality of EVs from SYF1 cells (Figure 4c,d). As expected, Fyn levels in the cells expressing Fyn(G2A) were significantly reduced compared to cells expressing Fyn(WT) (Figure 4d, lane 7 vs. lane 6), suggesting a functional role of myristoylation in encapsulating Fyn kinase into EVs. Importantly, similar to the Fyn(G2A) mutant, loss of palmitoylation in the Fyn(C3S/C6S) mutant decreased its levels in EVs (Figure 4d, lane 8 vs. lane 6). The results indicate that both myristoylation and palmitoylation promote Fyn kinase encapsulation into EVs.

3.6. Knockdown of TSG101 inhibited the encapsulation of Src kinase into EVs

The ESCRT protein complex is well‐defined in its role in exosome/EVs biogenesis. Tumour susceptibility gene 101 (TSG101) is associated with ESCRT‐I protein and regulates the localization of Src kinase (Babst et al., 2002; Garrus et al., 2001; Tu et al., 2010). We studied if the ESCRT pathway regulates Src encapsulation into EVs. As expected, TSG101 levels were decreased in the cell lysate and EVs lysate of both PC‐3 and 22Rv1 cells expressing shRNA‐TSG101 (Figure 5a,b). Knockdown of shRNA‐TSG101 did not change expression levels of Src kinase in the cell lysate of PC‐3 and 22Rv1 cells (Figure 5a,b). However, given that the same amount of protein was loaded from the cell or EVs lysates (detected by Coomassie blue staining), knockdown of TSG101 decreased the levels of Src and CD9 in EVs of PC‐3 cells (Figure 5a) and 22Rv1 cells (Figure 5b). Additionally, knockdown of TSG101 decreased the EVs particle number of PC‐3 and 22Rv1 cells (Figure 5c). These data suggest that the encapsulation of Src kinase into EVs is regulated by the ESCRT biogenesis pathway.

FIGURE 5.

Knockdown of TSG101 and treatment with filipin III inhibited the encapsulation of Src and Fyn kinases into EVs, respectively. (A,B) Knockdown of TSG101 inhibited encapsulation of Src kinase into EVs. 22Rv1 (a) and PC‐3 (b) cells were transduced with shRNA‐control, shRNA‐TSG101‐1 (shTSG101‐1), or shRNA‐TSG101‐2 (shTSG101‐2) by lentiviral infection. The transduced cells were incubated in the medium with 5% exosome‐free FBS at 37°C in 5% CO₂ for 48 h. EVs were isolated from the conditioned medium at day 5 by sequential ultracentrifugation. Equal amounts (10 μg) of the TCL and EVs lysates were loaded. Levels of TSG101, Src, Calnexin, GAPDH, and CD9 were analysed by immunoblotting. Levels of Src protein was quantified by Image J software. To compare the enrichment of Src levels, the ratio of Src levels in EVs versus those in TCL of 22Rv1 and PC‐3 cells was calculated and compared among the control and shRNA‐TSG101‐1/2 groups. Additionally, to confirm the same protein loading in TCL or EVs, the total amount protein was also measured by the Coomassie Blue staining. *p < 0.05; **p < 0.01; ***p < 0.001. (c) Knockdown of TSG101 inhibited the number of EVs particles. The number of EVs particles from SYF1 cells expressing shRNA‐TSG101‐/2 or vector control were analysed by the ZetaView PMX110. (d) The encapsulation of Fyn kinase, but not Src kinase, into EVs was affected by cellular cholesterol levels. PC‐3 cells were incubated in exosome‐free medium at 37°C in 5% CO₂ for 24 h and treated with 1 μM of the lipid raft disrupting agent filipin III for 24 h. EVs were isolated from cell culture medium by sequential ultracentrifugation. Levels of Fyn, Src, GAPDH, syntenin, and flotillin‐1 were examined in EVs and TCL by immunoblotting. Protein levels were quantified by Image J software. Protein levels of Src, Fyn, syntenin, CD9, and EGFR in EVs were calculated and compared between the control and filipin treatment group. The DMSO treatment group was set as 1. Additionally, to confirm the same protein loading in TCL or EVs, the total amount protein was also measured by the Coomassie Blue staining.

3.7. Disruption of lipid rafts by filipin III inhibited the encapsulation of Fyn kinase into EVs

Lipid rafts are membrane‐associated microdomains enriched with cholesterol and saturated phospholipids like sphingolipids that participate in EVs secretion (Tan et al., 2013; Trajkovic et al., 2008). Lipid rafts are a platform for encapsulation of proteins into EVs (de Gassart et al., 2003; Mayor & Riezman, 2004; Simons & Ikonen, 1997). EGFR is known to localize at lipid rafts (Pike et al., 2005), and Fyn kinase undergoes dual fatty acylation (myristoylation and palmitoylation) and is associated with lipid rafts (Gottlieb‐Abraham et al., 2016; Webb et al., 2000). To compare if Src and Fyn kinase levels in EVs are differentially regulated by lipid raft mediated EVs biogenesis, PC‐3 cells were treated with filipin III. Filipin III is known to disrupt lipid rafts by reduction of cholesterol levels in the cytoplasmic membrane. As expected, cells treated with filipin III showed significantly decreased cholesterol levels (Figure S5a). Filipin III treatment did not change Src kinase levels in the cell or EVs lysates (Figures S5b and 5d). In contrast, filipin III treatment did not change EGFR or Fyn levels in the cell lysate but decreased EGFR and Fyn levels in EVs along with reduced levels of the EVs protein markers including syntenin and flotillin‐1 without changing CD9 levels (Figure 5d). Of note, the reduction was not due to reduction of protein loading according to the Coomassie blue staining. Additionally, filipin III treatment did not change the size distribution of EVs, but reduced the number of the EVs particles (Figure S5c). The data suggest that encapsulation of Fyn kinase, but not Src kinase, is regulated by the EVs biogenesis through the lipid raft pathway.

3.8. Src levels were highly elevated in the serum EVs derived from host mice carrying prostate tumours

Src levels and activity are highly up‐regulated in numerous cancers (Irby & Yeatman, 2000). We further examined if Src kinase could be detected in the serum EVs in vivo. DU145 cells were transduced with control, Src(Y529F), or Src(Y529F/G2A) by lentiviral infection. SCID mice were inoculated with DU145+control, DU145+Src(Y529F/G2A), or DU145+Src(Y529F) in the kidney capsule. A fourth group of mice was not inoculated (no xenograft group). As expected, the size and weight of tumours significantly increased in the group expressing Src(Y529F) in comparison with those from control vector or Src(Y529F/G2A) (Figure 6a,b). Over‐expression of Src(Y529F) or Src(Y529F/G2A) was confirmed by immunohistochemistry also showing that tumour cells invaded into kidney tissues (Figure S6). However, no significant changes in the size (∼55 nm), zeta potential (∼‐25 mV), or particle number of the serum EVs were detected in the samples derived from no xenograft group or xenograft tumours expressing vector, Src(Y529F), and Src(Y529F/G2A) groups (Figure 6c–e).

FIGURE 6.

Myristoylation facilitates Src protein encapsulation into plasma EVs. DU145 cells were transduced with vector control, Src(Y529F), or Src(Y529F/G2A) by lentiviral infection. DU145 cells (control) or the transduced cells (1 × 10⁴ cells/graft) were mixed with collagen and implanted into the sub‐renal capsule of SCID mice (3 months‐old). After 5 weeks, the mice were sacrificed, and tumours and serum samples were harvested. (A,B) The image (a) and weight of xenograft tumors (b) were recorded. Due to the invasion of tumours into kidneys, the kidney together with the tumour tissue were weighed (due to difficulty in separating the tumour tissue from the kidney). (C–E) EVs were isolated using the ExoQuick kit from the plasma of the host mice carrying no xenografts (control), xenografts of DU145 cells alone (control), and those expressing Src(Y529F) or Src(Y529F/G2A) (three mice in each group). The isolated EVs were characterized by size distribution (c), zeta potential (d), and EVs concentration (e). (f) Expression levels of Src, non‐pSrc(Y529) [non‐phosphorylation of Y529 site leading to an open conformation of Src protein, thus activation of Src kinase such as constitutively active Src(Y529F) mutant], HSP90, TSG101, and CD81 in plasma EVs were analysed by immunoblotting. Equal amounts of EVs protein (50 μg) were loaded. The loading of total EVs protein was also examined by the Coomassie Blue staining. #1/2/3 represent three mice in each group. Levels of the tested proteins were quantified by Image J software. *p < 0.05; **p < 0.01; ***p < 0.001; ns: No significant.

Given that the same amount of total protein in all samples was loaded (based on the Coomassie Blue staining), Src levels or non‐phosphorylated Src at 529 (detection of activated Src) were significantly elevated in the serum EVs of the Src(Y529F) group in comparison with those from no xenograft control group, xenograft tumour groups expressing vector or Src(Y529F/G2A) (Figure 6f). While levels of the serum EVs protein marker HSP90 in the Src(Y529F) group was elevated in comparison with vector control or mutant Src(Y529F/G2A) groups, TSG101 and CD81 levels were not significantly different among the groups (Figure 6f). Of note, serum HSP90 is reported to be a potential biomarker for prostate cancer (Burgess et al., 2008).

To exclude the possibility that the high amount of Src in serum EVs was due to the larger tumour size in Src(Y529F) tumour instead of Src myristoylation. We generated similar size of tumors between Src (Y529F/G2A) and Src(Y529F). We implanted ten times more DU145 cells or DU145 expressing Src(Y529F/G2A) groups (less tumorigenic) than that in DU145 expressing Src(Y529F) group. Similar tumour size was observed in both Src(Y529F) and Src(Y529F/G2A) (Figure S7a,b). The number, size and zeta potential of isolated EVs showed no significant difference (Figure S7d,e). Expression levels of Src, HSP90 and non‐P‐Src(Y527) in the serum EVs from the Src(Y529F) group were elevated in comparison with those in Src(Y529F/G2A) or vector controls groups (Figure S7f). High variation of TSG101 levels was observed as shown in Figure 6f, but no significantly difference among the groups was observed. The same amount of EVs protein loading was confirmed by Coomassie blue staining.

Collectively, the data suggest that myristoylation is essential for the encapsulation of Src kinase into EVs, and elevated levels of the active Src or total Src kinase in the serum EVs reflects the host tumours in vivo.

4. DISCUSSION

This study has clearly demonstrated that myristoylation is essential to mediate the encapsulation of Src family kinases including Src and Fyn kinases into EVs. Our group reported that fatty acylation is essential for Src kinase to be enriched in exosome/EVs at the AACR annual meeting in 2018 (Wen et al., 2018). We report now that Src and Fyn kinases are enriched in EVs. This result was also reported by another research group in 2019 (Hikita et al., 2019). Myristoylation contributes to the association of both Src and Fyn kinases with the cytoplasmic membrane (Resh, 1994). Encapsulation of Src and Fyn kinases into EVs is dependent on the membrane tethering potential in the N‐terminus of the proteins. The mutants Src(G2A) and Fyn(G2A) leading to loss of myristoylation or pharmacological inhibition of Src myristoylation inhibit levels of Src kinase in the cytoplasmic membrane (Kim, Alsaidan et al., 2017; Kim, Yang et al., 2017). Our results agree with the report that myristoylation participates in the secretion of TyA‐GFP in exosomes/microvesicles. The loss of myristoylation in Acyl(G2A)TyA‐GFP and HIV Gag(G2A)TyA‐GFP inhibits a target protein to be secreted into EVs or HIV virus (Shen et al., 2011). The knowledge regarding the role of the N‐terminus of Src family kinases in their encapsulation into EVs can be exploited to encapsulate a protein of interest into EVs through genetic modification. We have successfully demonstrated that the fusion of the octapeptide derived from the N‐terminus of Src kinase with Cas9 confers its ability to be encapsulated to EVs, thereby creating a favourable delivery vehicle of Cas9 through EVs (Whitley et al., 2022). This technology could be expanded to target other biomolecules for encapsulation.

Although myristoylation serves as an important factor for Src/Fyn encapsulation into EVs, Src(G2A) or Fyn(G2A) can also be detected at lower levels in EVs. It appears that multiple Lys residues (positively charged) at the N‐terminus of Src kinase contribute to encapsulation of myristoylated Src into EVs. While loss of lysine residues in the Src(K5A/K7A/K9A) mutant resulted in only limited reduction in encapsulation into EVs compared to Src(WT), myristoylated Src had 2‐fold reduction in EVs, implying that myristoylated Src is an EVs cargo, and a portion of Src kinase which are not necessarily myristoylated can also be encapsulated into EVs likely through an interaction with other proteins. A study has demonstrated that lysine at site 5 is important for Src myristoylation (Gottlieb‐Abraham et al., 2016). Of note, mutations at sites 3 and 6 also co‐exist in this study. These two sites are consensus sites for the N‐myristoyltransferase activity to catalyze the myristoylation of Src kinase. The triple lysine mutant Src(K5A/K7A/K9A) showed no change in Src myristoylation levels in our study. Therefore, in addition to the fatty acylation, maintaining the lysine residues at the N‐terminus of Src kinase is also important to design a protein tag to encapsulate a protein of interest into EVs (Whitley et al., 2022). This bioengineering strategy could provide an avenue to deliver tumour suppressor proteins such as PTEN, P53, and Rb, which is usually mutated or lost in cancer cells and should be explored in future studies.

Our study has also identified that palmitoylation contributes to the encapsulation of Fyn kinase into EVs. Palmitoylation is a reversible post‐translation protein modification in which palmitic acid is covalently attached to cysteine (Resh, 1999). Our data show that the removal of palmitoylated sites in the Fyn(C3S/C6S) mutant inhibit the encapsulation of Fyn kinase into the EVs. This is consistent with a proteomics study showing that a panel of palmitoylated proteins enriched into EVs is suppressed by a palmitoylation inhibitor (Mariscal et al., 2020). Similarly, the deficiency of DHHC21, a palmitoyl acyltransferase, significantly changed the protein profile in EVs (Yang et al., 2022). In particular, some palmitoylated proteins have been reported to favour encapsulation into EVs. For example, ACE2 where the Cys 141 and 498 sites are reported to be S‐palmitoylated is essential for its encapsulation into EVs. This molecular function has been exploited for blocking SARS‐CoV‐2 infection using ACE2‐encapsulated EVs (Xie et al., 2021). In another example, S‐palmitoylation of Alix regulates its interaction with CD9 to maintain the lipid bilayer structure of EVs during EVs biogenesis (Romancino et al., 2018). Protein palmitoylation profiling is a potential clinical application for early detection of cancer.

Our study illustrates two different molecular pathways in the EVs biogenesis to facilitate the encapsulation of Src and Fyn kinases (Figure 7). Fatty acylation determines their differential localization at the cytoplasmic membrane and intracellular protein trafficking paths to the cell membrane (Sandilands et al., 2007). Encapsulation of Src kinase into EVs is regulated by the ESCRT pathway as knockdown of TSG101 reduces Src encapsulation. The ESCRT pathway is composed of ESCRT‐0, I, II, III and vacuolar protein‐associated sorting (Vps) 4, which is a major pathway regulating the EVs biogenesis (Christ et al., 2017; Raiborg & Stenmark, 2009; Votteler & Sundquist, 2013). TSG101 is involved in the ESCRT‐1 protein complex for vesicle budding process (Kumar et al., 2016). c‐Src does not appear to belong to the ESCRT protein complex directly, but it interacts with Alix to activate ESCRT‐mediated intra‐luminal vesicle biogenesis (Hikita et al., 2019). The ESCRT pathway is also reported to regulate Src trafficking at the endosomal compartment, and subsequently at focal adhesions (Tu et al., 2010). Therefore, reciprocally, active Src kinase promotes exosome/EVs secretion and its encapsulation through its interaction with Alix in the ESCRT pathway (Hikita et al., 2019). Further study should investigate if activation of Src will regulate the vesicle formation by direct/indirect interaction with TSG101 in the ESCRT protein complex in EVs biogenesis.

FIGURE 7.

Src family kinases including Src and Fyn kinases are encapsulated into extracellular vesicles through two different biogenesis pathways. Src kinase is associated with the cytoplasmic membrane through protein myristoylation and lysine positive charge at the N‐terminus. Similar to CD9 tetraspanin, Src kinase encapsulation is regulated by the ESCRT proteins such as TSG101 in the exosome biogenesis. On the other hand, Fyn kinase is associated with the lipid raft through both myristoylation and palmitoylation. Disruption of lipid raft interferes the encapsulation of Fyn and EGFR proteins into EVs, but not Src kinase.

In contrast to Src, Fyn kinase has dual fatty acylation (myristoylation + palmitoylation), which favours the protein to localize at the detergent‐insoluble caveolin‐containing membrane domains (Guzzi et al., 2001; Shenoy‐Scaria et al., 1994). It has also been reported that dual fatty acylation has a 4‐fold higher potential to promote Gi1 alpha protein to localize to caveolae (Song et al., 1997). Our study clearly shows that the encapsulation of Fyn kinase is significantly disturbed by filipin III, an agent interfering with membrane sterols, and thus lipid raft formation. However, due to low expression levels of Fyn kinase in the tested cells (PC3 and 22Rv1), it remains to be studied if Fyn kinase levels were also regulated by the ESCRT pathway. The difference in molecular mechanisms regulating the encapsulation of Src and Fyn kinase into EVs reflects the heterogeneity in the EVs population.

Our data indicate that Src kinase levels are consistent in the lysate of both EVs and EV‐producing cells across all the tested cell lines. Src kinase has also been detected in EVs of 42 out of 60 cancer cell lines by proteomics analysis (Figure S2a). Given the ubiquitous expression of Src kinase in cancer cells, the result suggests that Src kinase levels can serve as an excellent protein biomarker to characterize isolated EVs together with other biomarkers recommended in the 2018 guidelines (Thery et al., 2018). The presence and/or absence of several proteins have been widely used for examining EVs purity. For example, while calnexin is exclusively found in cell lysate and not in EVs, several EVs protein markers including CD9, CD81, CD63, and many others are enriched in EVs. However, due to the variation of human or mouse cell lines, some of these EVs protein markers are not universally expressed. Interestingly, when the same number of total proteins were examined (demonstrated by Coomassie Blue staining), levels of Src kinase in EVs lysate are similar to, if not further enriched, those in the parental cell lysate. As a result, Src kinase might serve as an excellent protein loading control when it is necessary to compare levels of proteins of interest between parental cell and EVs lysates or serve as another layer for EVs quality control.

It should be noted that activated Src kinase has an elevated EVs encapsulation potential compared with wild type Src kinase. This is likely due to activated Src kinase better associating with the cytoplasmic membrane, which is required for its encapsulation into EVs via the EVs biogenesis ESCRT pathway. Activated Src such as the constitutively active mutant Src(Y527F) [Src(Y529F) in human gene] has a 3‐fold greater potential to associate with cell membrane and/or membrane‐associated proteins (Shvartsman et al., 2007). It has been reported that the presence of membrane‐bound phosphatase could promote its membrane localization where Src kinase is activated by dephosphorylation of Src kinase at the Tyr527 site (Patwardhan & Resh, 2010). Src kinase is important for endosomal budding mediated by the syntenin–syndecans pathway. It phosphorylates the DEGSY motif in the cytosolic domain of syndecan and at tyrosine 46 of syntenin, subsequently enhancing the endosomal budding (Imjeti et al., 2017). Together, the association of Src kinase with the cytoplasmic membrane and its involvement in the functionality of the ESCRT and syntenin–syndecans pathways favour its encapsulation into EVs. Our data show that loss of myristoylation in the constitutively active Src dramatically suppresses Src levels in EVs.

Expression levels of Src family kinases in EVs can potentially reflect the characteristics of primary tumours, thereby serving as a biomarker of the disease. In this study, we used prostate cancer cells as a model to study Src and Fyn levels in EVs. Myristoylated proteins including Src kinase are enriched in EVs (Whitley et al., 2022). Based on the myristoylation and/or palmitoylation pattern in the N‐terminus, Src and Fyn can be detected in EVs of the 42 and 10 out of 60 cancer cell lines, respectively. Additionally, the enrichment of Src levels was observed in EVs of SYF1 and 3T3 (murine fibroblast cells). Therefore, the enrichment of Src and to a less extent Fyn (likely due to relatively low expression levels in cancer cells) should be investigated in other cancer cell types. Thus, Src and Fyn levels in EVs could excellently reflect their levels in EVs‐producing cells. Src kinase plays oncogenic roles in tumour growth, progression, and metastasis of human cancers such as colon, breast, in addition to prostate cancer (Irby & Yeatman, 2000) and is overexpressed in various cancers (Ishizawar, 2004; Vlaeminck‐Guillem et al., 2014). Furthermore, elevated SFK activity is clinically associated with shorter life expectancy and a high probability of distant metastasis (Chattopadhyay et al., 2017; Drake et al., 2013; Tatarov et al., 2009). Inhibition of Src kinase is a potential therapeutic strategy for cancer through suppressing the proliferation, invasion, and migration of cancer cells (Kim et al., 2009; Lara et al., 2009; Park et al., 2008). Elevated c‐Src levels were detected in the serum EVs of a TRAMP mouse model (DeRita et al., 2017). Our in vivo data show that Src(Y529F)‐mediated tumors significantly elevated levels of Src protein and activated Src in serum EVs. Therefore, it is conceivable that myristoylated Src levels in serum EVs might potentially serve as a biomarker for a disease with highly elevated Src kinase activity. However, it remains to be explored if myristoylated Src/Fyn kinase in EVs derived from metastatic prostate tumours plays biological and molecular functions to modulate the microenvironment including stromal cells or osteoclast cells.

CONFLICT OF INTEREST STATEMENT

Houjian Cai are the inventor on the patent application on the technology that the myristoylation modification promotes encapsulation of Cas9 protein into extracellular vesicles (WO2020/206072). The remaining authors declare no competing interests.

Supporting information

Supporting Information

Ye, C. , Gosser, C. , Runyon, E. D. , Zha, J. , Cai, J. , Beharry, Z. , Bowes Rickman, C. , Klingeborn, M. , Liu, Y. , Xie, J. , & Cai, H. (2023). Src family kinases engage differential pathways for encapsulation into extracellular vesicles. Journal of Extracellular Biology, 2, e96. 10.1002/jex2.96