High sensitivity isolation and proteomics analysis of small extracellular vesicles: applications in neuroscience and neuro-oncology

Small extracellular vesicles (sEVs, <150 nm) contain proteins, lipids and nucleic acids encapsulated by a lipid bilayer. sEVs are released by all cell types and are present in almost all body fluids. They are known to play an active role in the development and protection of the central nervous system (CNS) under both physiological and pathological conditions. Their number, size, and molecular cargo may be altered in disease and so there is great interest in investigating their potential as a source of biomarkers. Several aspects of sEVs purification need to be considered prior to investigating sEVs in body fluids. Highly abundant background proteins from the body fluid, i.e. the blood, need to be depleted to avoid them overwhelming the protein signatures of the sEVs. Furthermore, the amount of sEVs proteins is often very small and so any sEVs enrichment procedure and protein analysis method should maximize yield/limit losses at each step. Mass-spectrometry-based proteomics has established itself as the method of choice for the identification and characterization of proteins. Sample preparation for MS-based proteomics is well established when ample sample is available, but when the total amount of protein available is low, i.e. 0.1-3 g of protein extract, any losses during the workflow become problematic and adversely affect the number of proteins that can be identified and the precision of protein quantification. Accordingly, it becomes more difficult to evaluate differences between different conditions with small sample sizes, because of the increased variability when operating closer to the quantitative limits of the analytical technique. Advances in liquid chromatography and mass spectrometry equipment (LC-MS) have greatly improved over the past decade, and now enable the analysis of low sample amounts. However, developments in LC-MS sensitivity needs to be matched by parallel improvements in the sample preparation protocols. This is particularly true for the characterization of sEVs from body fluids. This work sought to give an effective response to the lack of high sensitivity sEVs isolation and proteomics procedures through the development of standardized workflows. The first part of my work (Chapter 2) concerns the development of a proteomics procedure for the analysis of sEVs starting from just 50 μl of serum. The sEVs isolation methods precipitation (PPT) and size exclusion chromatography (SEC) had previously demonstrated sEVs isolation from serum volumes of 200-500 μl. PPT and SEC were first compared using 100 μl of serum; the SEC procedure identified more proteins and was then scaled down to 50 l. The proteins extracted from the sEVs were processed using the Single-Pot, Solid Phase-enhanced Sample Preparation protocol (SP3), which is based on the use of carboxylate coated paramagnetic beads; the SP3 method was further optimized to increase the number of identified proteins and reproducibility (the full protocol is reported in Annex 1). The optimized SEC-based method for sEVs proteome characterization from 50 μl of serum was then applied to a longitudinal study of serum-sEVs proteome changes in a glioblastoma multiforme (GBM) mouse model (Chapter 3). Serum was collected at multiple time points: baseline, T1 (pre-symptomatic) and T2 (post-symptomatic) stages. Data were analyzed by using a linear mixed effect model to detect molecular changes during disease progression. The ability to analyze sEVs proteins from just 50 μl of serum enabled the sEVs proteome and the total serum proteome to be recorded from each individual longitudinal serum sample (the maximum amount of blood that may be withdrawn from an adult mouse every 14 days is approximately 10% of its total blood volume, which corresponds to approximately 70 μl of serum). The second part of the work focused on neuron-derived small extracellular vesicles (NDsEVs), because they constitute a window into brain pathological processes (Chapter 4). NDsEVs were isolated by L1CAM-based immunoprecipitation, then their proteins extracted and subject to LC-MS based proteomics analysis. The isolation of NDsEVs using immunoprecipitation involves the use of detergents which impaired the performance and reproducibility of the aggregation-based SP3 procedure. Instead, I investigated a new proteomics approach, namely the protein suspension-trapping approach, which is far more tolerant of co-factors because it enables extensive washing of the immobilized proteins. I optimized the proteolytic digestion, and used the optimized workflow for the characterization of NDsEVs from Parkinson Disease’s (PD) plasma. The reproducibility was evaluated by analyzing technical and biological replicates, and a gene enrichment analysis was performed to assess NDE enrichment. The success of these proof-of-concept experiments now form the basis for the drafting of a larger PD-NDsEVs clinical study. A tangential aim of the work has been the development of an automated SEC workflow using a high-pressure liquid chromatography system for increasing throughput and reproducibility, as well as ease-of-access (Chapter 4, §4.3). The HPLC-SEC system has demonstrated its capability to isolate sEVs from different samples types, including cell culture media, serum, and plasma. This includes the isolation and characterization of proteins from sEVs isolated from cortical neurons derived from human neuroepithelial stem (NES) cells. The characterization of the sEVs proteins from the NES-derived cortical neurons generated reference dataset of NDsEVs to compare with the NDsEVs obtained by L1CAM-based immunoprecipitation from PD patient’s plasma. Altogether, the sEVs isolation and microproteomic strategies developed during my PhD represent powerful tools for the proteomics characterization of sEVs when a limited sample amount is available. In particular, the longitudinal study of serum sEVs from a murine model glioblastoma multiforme is the first longitudinal study based on single subject sEVs and enabled the identification of candidate biomarkers for early stage detection (prior to symptoms). The method is directly applicable to other tumor types and so represents a starting point for future longitudinal experiments for the identification of early stage biomarkers. Moreover, the L1CAM-based isolation and sEVs proteomics procedure represent a highly versatile approach that can be automated and applied to different CNS disorders or to the study of sEVs isolated with different antibodies.In summary, addressing the limitation of limited sample amount and reproducibility, the two strategies described in this thesis constitute an advance in the sEVs proteomics field and represent a step forward to the large-scale clinical application of sEVs.