Highly-purified exosomes and shed microvesicles isolated from the human colon cancer cell line LIM1863 by sequential centrifugal ultrafiltration are biochemically and functionally distinct

Highlights

• Sequential centrifugal ultrafiltration affords unbiased isolation of extracellular vesicles.

• Two major EV subtypes are secreted by colon cancer tumour cells – sMVs and exosomes.

• sMVs (30–1300 nm) are heterogeneous, whereas exosomes are homogeneous (30–100 nm) in size.

• Exosomes, unlike sMVs, contain protein markers Alix/TSG101/CD63/CD81.

• 350 proteins uniquely enriched in sMVs (not seen in exosomes) are potential sMV markers.

Abstract

Secretion and exchange of extracellular vesicles (EVs) by most cell types is emerging as a fundamental biological process. Although much is known about EVs, there is still a lack of definition as to how many naturally occurring EV subtypes there are and how their properties and functionalities might differ. This vexing issue is critical if EVs are to be fully harnessed for therapeutic applications. To address this question we have developed and describe here a sequential centrifugal ultrafiltration (SCUF) method to examine, in an unbiased manner, what EV subtypes are released in vitro into cell culture medium using the human colon carcinoma cell line LIM1863 as a model system. Using the culture medium from ∼7.2 × 10⁹ LIM1863 cells, SCUF was performed using hydrophilic PVDF membranes with low protein binding properties (Millipore Durapore™ Ultrafree-CL filters with 0.1, 0.22, 0.45 and 0.65 μm pore size). EV particle sizing was measured using both dynamic light scattering and cryo-electron microscopy. Comparative proteome profiling was performed by GeLC–MS/MS and qualitative protein differences between EV subtypes determined by label-free spectral counting. The results showed essentially two EV subtypes; one subtype (fraction Fn1) comprised heterogeneous EVs with particle diameters of 30–1300 nm, the other (fraction Fn5) being homogeneous EVs of 30–100 nm diameter; based on cryo-EM both EV subtypes were round shaped. Western blot analysis showed Fn5 (SCUF-Exos) contained traditional exosome marker proteins (Alix⁺, TSG101⁺, CD81⁺, CD63⁺), while Fn1 (SCUF-sMVs) lacked these protein markers. These findings were consistent with sMVs isolated by differential centrifugation (10,000g, DC-sMVs) and exosomes (100,000g EVs depleted of 10,000g material). The buoyant density of sMVs determined by OptiPrep™ density gradient centrifugation was 1.18–1.19 g/mL and exosomes 1.10–1.11 g/mL. Comparative protein profiling of SCUF-Exos/-sMVs revealed 354 and 606 unambiguous protein identifications, respectively, with 256 proteins in common. A salient finding was the first report of 350 proteins uniquely identified in sMVs may of which have the potential to enable discrimination of this EV subtype from exosomes (notably, members of the septin family, kinesin-like protein (KIF23), exportin-2/chromosome segregation like-1 protein (CSE1L), and Rac GTPase-activating protein 1 (RACGAP1)). We report for the first time that both SCUF-Exos and SCUF-sMVs isolated from LIM1863 colon cancer cells induce invasion of recipient NIH3T3 cells. Interestingly, the SCUF-sMVs promote invasion to a significantly greater extent (3-fold) than SCUF-Exos. This analytical SCUF method for fractionating EVs is potentially scalable using tangential flow filtration, thereby providing a solid foundation for future in-depth functional studies of EV subtypes using diverse cell types and functional assays.

Introduction

Secretion and exchange of extracellular vesicles (EVs), nano- to micrometre-sized membranous organelles, by most cell types is emerging as a central paradigm for intercellular communication [1], [2]. While EVs have been primarily studied in vitro using cell culture media as a source material, they are also found in vivo in diverse body fluids, such as semen, synovial fluid, saliva, urine, breast milk, amniotic fluid, malignant ascites, bronchoalveolar lavage fluid, and blood [1], [3]. EVs are thought to modulate recipient cell behaviour by transfer of their intrinsic cargo constituents such as oncogenic proteins [4], [5], [6], infectious proteins [7], [8] malarial proteins [9], miRNAs/mRNAs [10], DNA [11], and lipids [12]. There is now an increasing awareness that EVs play a critical role in the development of diverse pathologies such as cancer (e.g., of pre-metastatic niche formation [13], cell transformation/ epithelial-mesenchymal transition [6], neurodegenerative disorders [8], and infectious diseases (e.g., malaria [9], bacterial infection [14], [15]). Collectively, these studies have engendered great interest in harvesting EVs for therapeutic applications such as regenerative medicine [16], [17], vaccination against infectious disease [18], [19], and EV vaccines for possible cancer treatment [20], [21], [22]. These studies have led to several clinical and pre-clinical investigations of EV-based therapies [2], [23], [24], [25].

Two seminal reports in 2008 revealed that EVs isolated from cultured tumour cells contain cargo information that parallels blood-EV information obtained from disease patients. Skog et al., reported that isolated EVs from cultured glioblastoma cells obtained from resected human tumours contained transcripts (e.g., EGFR variant III mRNA) and miRNAs that mirrored serum-EV information from patients with glioblastoma [4], while Taylor and colleagues demonstrated that specific miRNA signatures from ovarian cancer cell line EVs (and lung cancer cell lines) correlated with miRNA profiles of EpCAM-immunocaptured-EVs from blood obtained from ovarian cancer (and lung cancer) patients [26], [27]. These studies highlight the potential use of serum-EV information to provide tumour diagnostic biomarkers and assist in the design of therapeutic strategies.

EV annotation is a vexed question. They can be classified based on their cellular origins and/or biological functions or on their biogenesis [2]. These concerns are widely discussed in the international EV community (International Society for Extracellular Vesicles (ISEV)) [28]. In broad terms there are thought to be two EV subtypes: 100–1000 nm diameter microvesicles (shed microvesicles, sMVs; membrane blebs) and 30–150 nm diameter exosomes [29], [30]. sMVs are generated by directly outward budding from the plasma membrane [31], [32], while exosomes are generated in the early/late endosomal pathway by the inward budding of multivesicular bodies (MVBs) luminal membranes to form intraluminal vesicles (ILVs); MVBs then traffic to and fuse with the plasma membrane whereupon they release their ILV contents into extracellular space (exosomes) [30]. Given that most EV functional studies are performed using diverse preparations, often with impure, heterogeneous and poorly characterised EVs, it is very difficult to ascribe function to a specific EV subtype – this has ramifications when designing EV therapeutics, especially when determining possible EV-subtype side-effects in clinical investigations [25]. These issues have engendered great interest in establishing how many naturally occurring EV subtypes there are and improving methodologies for EV isolation.

Effective methods for the isolation and characterisation of EVs remain challenging [33], [34], [88]. Current strategies include differential centrifugation (DC) [35], filtration using hydrophilic polyvinylidene difluoride (PVDF) membranes of different pore sizes [36], [37], high performance size-exclusion chromatography (SEC) [38], ultrafiltration with SEC [34], immunocapture [39], [40], differential density gradient ultracentrifugation [33], tangential flow filtration [41], field flow fractionation [42], and microfluidic isolation [43]. Additionally, there is now a plethora of commercial “easy isolation kits” – designed essentially to precipitate EVs from body fluids. However, these one-step kits are designed for diagnostic purposes and do not enable separation of EV subtypes one from another or distinguish EVs from macromolecular aggregates [30]. A further development in clinical diagnostics is the use of blood-based EV antibody arrays for multiplexed phenotyping of EV subtypes [44].

In the present study, we set out to determine the number of EV subtypes secreted by the human colon carcinoma cell line LIM1863. To this end we developed a sequential centrifugal ultrafiltration (SCUF) method that relies on microfiltration through a series of hydrophilic PVDF membranes of different pore sizes (0.1–0.65 μm). We show that >95% of the total EVs released from LIM1863 cells into culture media are exosomes (45%) and sMVs (50%). Using GeLC–MS/MS we identify for the first time 350 proteins that are selectively enriched in sMVs (in comparison with exosomes), many of which have not been previously described in EVs; we expect that many of these identifications will form the basis for definitive sMV protein markers that will enable their distinction from exosomes. Importantly, we demonstrate for the first time, that LIM1863 colon cancer cell-derived SCUF-Exos/-sMVs display differential invasive activities on recipient fibroblast cells.