Practically all cells (under both physiological and pathological conditions) secrete small non-apoptotic vesicles. These include microvesicles and exosomes, collectively termed extracellular vesicles (EVs). Microvesicles are formed during plasma membrane shedding, whereas exosomes originate from the endosomal pathway. EVs contain a specific subset of lipids, RNAs, and proteins that can be transferred between cells. The current model for EV biogenesis dictates that cytosolic proteins are in the lumen, and cell surface proteins are on the exterior of EVs. Annexin A2 is a cytosolic calcium-dependent membrane-binding protein lacking a signal sequence for secretion (leaderless protein). Annexin A2 is involved in many cellular processes including membrane trafficking events. Through its membrane-binding capacity, Annexin A2 has been shown to be involved in lipid organisation at sites of membrane-actin interactions, calcium-mediated endocytosis, and may have a role in ion channel activity. Along with several other proteins, Annexin A2 is consistently identified and enriched in EVs (Exocarta/Vesiclepedia databases). Here we demonstrate that, despite being a cytosolic protein, Annexin A2 localises not only in the lumen but also on the surface of EVs.

To investigate the packaging and localisation of Annexin A2 in extracellular vesicles.

EVs were produced from the pancreatic cancer cell line, PANC1, and isolated by ultracentrifugation (Fig. 1A) as described. This cell line was chosen as it produces a large number of EVs and expresses a high level of Annexin A2. Isolated EVs were characterised according to size distribution and common exosomal markers. Tunable resistance pulse sensing showed that purified EVs had a mean diameter of 153 nm and a mode of 98 nm (Fig. 1B), consistent with the literature. EVs showed an enrichment in the exosomal marker CD63 and were negative for the Golgi-localised, γ-ear homology domain, ARF-binding protein (GGA1) (Fig. 1C). EVs also contained acetylcholine esterase, as detected by a specific activity assay (Fig. 1D). These results show that EVs had the expected size distribution and contained a large proportion of exosomes.

Annexin A2 is a calcium-dependent membrane-binding protein. As such it is able to be removed from the membrane with calcium chelators, such as EDTA. To determine whether Annexin A2 is on EV surface, EVs were treated with EDTA for 30 min at 37°C. The supernatant was then separated from the EVs by ultracentrifugation, and Annexin A2 was detected by western blot. We found that Annexin A2 was present in the supernatant, which suggests its dissociation from the EV membrane upon EDTA treatment (Fig. 1E). This was unlikely to be due to EV damage, as actin (present in the lumen of EVs) was not detectable in the EDTA supernatant (Fig. 1E). Annexin A2 was also detectable in the EV pellet, thus indicating that it is present both in the lumen as well as on the surface of EVs (Fig. 1E). To confirm this result, samples were subjected to mass spectrometry analysis; we observed a two-fold increase in the amount of Annexin A2 when EDTA was present (Fig. 1F). These experiments were repeated with EVs isolated from mouse neural stem cells (NSCs) and led to similar results, indicating that the presence of Annexin A2 on EV surface is not restricted to one cell line (Fig. 1G). A small amount of actin was also detected in the supernatant in both untreated and after EDTA, which could be due to some actin leaking from the EVs (Fig. 1G). Nonetheless, we observed a clear-cut evidence with Annexin A2 that was found in EDTA supernatants only. In NSC experiments, we also used GAPDH as additional control, which was not present in supernatants so to confirm the integrity of EVs (Fig. 1G).

To further confirm the presence of Annexin A2 on the surface of EVs, we used a protease protection assay. EVs were treated with trypsin with or without detergent for 30 min at 37°C; the EV pellet was then separated from the supernatant by ultracentrifugation and Annexin A2 was detected by western blot (Fig. 1H). Annexin A2 was liable to tryptic digestion, further confirming its unpredicted presence on the surface of EVs. When detergent and trypsin were present, all the available Annexin A2 was digested to completion. This confirms that the Annexin A2 detected after tryptic digestion, both the full length and smaller products, were protected from cleavage. As an internal control, the blot was probed for actin, the vast majority of which remained intact after trypsin digestion, thus indicating the EVs were not damaged as a result.

Following the current model for EV formation, cytosolic proteins should remain in the lumen of EVs. We report here the intriguing observation that the cytosolic protein Annexin A2 is present on the surface of EVs. How Annexin A2 is able to access the outer membrane of EVs remains unclear and represents a major gap in our understanding of protein trafficking that will require further investigation.

It is known that a pool of Annexin A2 is found at the cell surface. Therefore, it is possible that this surface pool of Annexin A2 is internalised during endocytosis upstream of EV biogenesis and that could explain the localisation of Annexin A2 on the surface of EVs. Similarly, microvesicles are formed by budding from cell surface and thus may also contribute to the pool of Annexin A2 found on the surface of EVs. However, as described above, Annexin A2 lacks a signal sequence for secretion and therefore should not be able to access the cell surface. In this scenario, Annexin A2 would have to cross the plasma membrane via an unconventional protein trafficking pathway, a hypothesis that remains to be tested. It would also be interesting to test the presence of other leaderless proteins on the surface of EVs to understand whether this observation represents a common feature of such leaderless proteins or is restricted to Annexin A2.

In conclusion, we show for the first time that Annexin A2 has an unpredicted localisation on the surface of EVs.

Antibodies

Rabbit anti-human actin (Sigma-Aldrich, A2066) used at 1:2000; mouse monoclonal anti-human Annexin A2 (BD Biosciences, 610069) used at 1:1000; mouse monoclonal anti-human CD63 (ThermoFisher Scientific, 10628D) used at 1:500. Q-27 mouse monoclonal anti-human GGA1 (Santa Cruz Biotechnology, sc-101257) used at 1:500. Secondary antibodies goat anti-rabbit IgG-HRP (Santa Cruz Biotechnology, sc-2004) and goat anti-mouse IgG (Santa Cruz Biotechnology, sc2005) were used at 1:2000. All antibodies were diluted in phosphate buffered saline (PBS).

Cell lines and cell culture

Human pancreatic cancer cell line PANC1 was maintained in DMEM (Sigma-Aldrich, D6546) containing 10% (v/v) fetal bovine serum, 2 mM L-glutamine, 1 mM penicillin and 1 mM streptomycin. Mouse neural stem cells (NSCs) were cultured as previously described.

Extracellular vesicle isolation

PANC1 cells were grown to 70% confluence in a T75 flask at which point the medium was changed to 10 ml serum-free DMEM and incubated for 24 h. EVs were collected by ultracentrifugation as previously described. Briefly, the medium was collected and sequentially centrifuged at 300 g for 15 min, 1,000 g for 15 min, 100,000 g for 90 min. The EV pellet was washed in serum-free DMEM without phenol red (SF-DMEM; GIBCO, 21063) centrifuged at 100,000 g for 30–60 min. EVs were resuspended in SF-DMEM at 10 µl per 10 ml starting material. EVs from NSCs were also produced as previously described. Briefly, 12 million cells were seeded per T75 flask and incubated overnight. The medium was collected and EVs isolated as described above.

Cell lysates

Cells were incubated in an appropriated volume of ice-cold lysis buffer (20 mM Tris-HCl, 137 mM NaCl, 1 mM EDTA, 1% triton X-100, pH 6.8) at 4°C for 10 min. The lysate was collected and insoluble material pelleted at 10,000 g for 10 min at 4°C. The supernatant was collected and sample buffer added (final: 50 mM Tris-HCl, 2% (w/v) sodium dodecyl sulphate (SDS), 0.1% (w/v) bromophenol blue, 10% (v/v) glycerol, 100 mM DTT, pH 6.8). Samples were boiled for 10 min and stored at -20°C.

Tunable resistance pulse sensing (TRPS)

TRPS were measured using the qNano (Izon Science, UK). The polyurethane nanopore NP100 (Izon Science, part A33255) was used for all measurements and was axially stretched to 46.99 mm. EV samples were diluted in PBS as required and 40 µl loaded into the instrument. Measurement time was up to 2 min depending on the instrument stability. The system was calibrated with 200 nm polystyrene beads diluted in PBS. Data analysis was carried out with Izon Control Suite software (Izon Science).

Acetylcholine esterase activity assay

10 µl EVs were lysed in 0.5% (v/v) Triton X-100 in PBS and assayed for acetylcholine esterase activity with a colorimetric assay kit from Abcam (ab138871) as per the manufacturer's instructions. The absorbance was measured at 410 nm over time and plotted against a buffer alone control.

EDTA treatment

EVs (10–20 µl per treatment) were resuspended in 100 µl of either SF-DMEM or versene solution (GIBCO, 15040-066) containing 0.48 mM EDTA and incubated at 37°C for 30 min. EVs were pelleted at 100,000 g for 30 min. The supernatant was collected and the pellet resuspended in 100 µl SF-DMEM. All samples were mixed with sample buffer (as above) and boiled for 10 min. Samples were analysed by western blotting.

Western blotting

All samples were resolved by 12% SDS-PAGE and transferred to polyvinylidene difluoride membranes for blotting. Membranes were blocked with 0.05% (w/v) skim milk powder in PBS containing 0.1% Tween-20 (PBS-Tween) for 30 min at room temperature. Membranes were then probed with an appropriate dilution of primary antibody overnight at 4°C. Membranes were washed three times in PBS-Tween before incubation in diluted secondary antibody for 1 h at room temperature. Membranes were washed as before and developed with ECL (Cyanagen, Westar XLS100) using a Bio-Rad Chemi Doc XRS system. Membranes were stripped with Restore plus (ThermoFisher Scientific, 46430) as per manufacturer's instructions.

Mass spectrometry analysis

Samples prepared as described above for EDTA treatment and submitted for mass spectrometry analysis using Thermo Orbitrap Q Exactive with EASY-spray source and Dionex RSLC 3000 UPLC. For graphical representation, the following equation (ratio = (EDTA sup/EDTA pellet)+1 / (untreated sup/untreated pellet)+1) was used to obtain a ratio of protein detected in the supernatant upon EDTA treatment with respect to the untreated control.

Trypsin treatment

EVs (10–20 µl per treatment) were resuspended in 100 µl of either SF-DMEM, trypsin solution (Sigma-Aldrich, 4674) diluted to 2.5 mg/ml SF-DMEM, 0.5% Triton X-100 (Sigma-Aldrich, T8787) diluted in SF-DMEM or a combination of both trypsin and Triton X-100. EVs were incubated at 37°C for 30 min. They were then pelleted at 100,000 g for 30 min. The supernatant was collected and pellet resuspended in 100 µl SF-DMEM. Samples were mixed with sample buffer (as described above), boiled for 10 min and analysed by western blotting.

This work was supported by Wellcome Trust Strategic Award [100574/Z/12/Z] and MRC Metabolic Diseases Unit [MRC_MC_UU_12012/5] from the Italian Multiple Sclerosis Association (AISM, grant 2010/R/31), the Italian Ministry of Health (GR08-7), the European Research Council (ERC) under the ERC-2010-StG Grant agreement n° 260511-SEM_SEM, the UK Regenerative Medicine Platform Acellular hub (Partnership award RG69889), and core support grant from the Wellcome Trust and MRC to the Wellcome Trust-Medical Research Council Cambridge Stem Cell Institute. FG is supported by a scholarship of the Gates Cambridge Trust.

We thank Frances Richards (Cancer Research UK) for the PANC1 cell line and the CIMR/IMS Proteomics Facility.

Not applicable.