Isolation of membrane vesicles from prokaryotes: a technical and biological comparison reveals heterogeneity

Abstract

Prokaryotes release membrane vesicles (MVs) with direct roles in disease pathogenesis. MVs are heterogeneous when isolated from bacterial cultures so Density Gradient Centrifugation (DGC) is valuable for separation of MV subgroups from contaminating material. Here we report the technical variability and natural biological heterogeneity seen between DGC preparations of MVs for Mycobacterium smegmatis and Escherichia coli and compare these DGC data with size exclusion chromatography (SEC) columns. Crude preparations of MVs, isolated from cultures by ultrafiltration and ultracentrifugation were separated by DGC with fractions manually collected as guided by visible bands. Yields of protein, RNA and endotoxin, protein banding and particle counts were analysed in these. DGC and SEC methods enabled separation of molecularly distinct MV populations from crude MVs. DGC banding profiles were unique for each of the two species of bacteria tested and further altered by changing culture conditions, for example with iron supplementation. SEC is time efficient, reproducible and cost effective method that may also allow partial LPS removal from Gram-negative bacterial MVs. In summary, both DGC and SEC are suitable for the separation of mixed populations of MVs and we advise trials of both, coupled with complete molecular and single vesicle characterisation prior to downstream experimentation.

Keywords: Extracellular vesicles; microbe; nucleic acids; outer membrane vesicles; pathogen.

Figure 1.

Fractionation of UPEC crude MVs by density gradient centrifugation. (a) UPEC I, II and III indicate three replicates prepared on different days. Density gradient fractionation of three replicate crude MV preparations from UPEC 536 grown with (RF) and without (R) iron supplementation. Density gradient banding in the tubes have individual fractions bracketed and are labelled from low to high density with visible bands marked with lines. (b) Protein amount and particle count are graphed as a percentage per fraction of all recovered. Error bars are mean ± standard deviation.

Figure 2.

Full molecular characterisation of density gradient fractions for one representative preparation of UPEC 536 MVs (UPEC III). One crude MV preparation from UPEC grown with (RF) and without (R) iron supplementation was further characterised. (a) Density gradient banding in the tubes have individual fractions bracketed and are labelled from low to high density with visible bands marked with lines. (b) Protein (µg/fraction), endotoxin (EU/fraction), RNA (ng/fraction), and particle counts (Millions/fraction) are graphed for each fraction. (c) PAGE profile of MV proteins with 10 µL of each fraction loaded and stained with SYPRO Ruby.

Figure 3.

Fractionation of Nissle 1917 crude MVs by density gradient centrifugation. (a) Nissle I, II and III indicate three replicates prepared on different days. Density gradient fractionation of one crude MV preparation from UPEC 536 grown with (RF) and without (R) iron supplementation alongside three from Nissle 1917. Density gradient banding in the tubes have individual fractions bracketed and are labelled from low to high density with visible bands marked with lines. (b) Protein amount and particle count are graphed as a percentage per fraction of all recovered. Error bars are mean ± standard deviation.

Figure 4.

Fractionation of UPEC RF crude MVs using qEV size separation columns. Three biological replicates of crude MVs isolated from UPEC 536 grown with iron supplementation (RF) were fractionated by SEC. qEV I and qEV III were performed using new qEV columns, whereas qEV II was separated on the washed qEV column after the fractionation of qEV I. Particle counts, protein, RNA and endotoxin in each fraction as a percentage of all recovered are graphed. Recovery of all fractions, relative to input and amount of molecules in the vesicle enriched fractions F8–F10 are tabulated to the side. Transmission electron microscopy of representative fractions F8–F10 are inset in the top right hand corner with two magnifications per fraction shown.

Figure 5.

Fractionation of E. coli crude MVs is highly reproducible when equal protein loaded is used. Three biological replicates of crude MV preparations isolated from UPEC 536 (a, b) and Nissle 1917 (c, d) grown with (RF) and without (R) iron supplementation were fractionated by SEC. Protein amount and particle counts are graphed as a percentage per fraction of all recovered. Error bars are mean ± standard deviation.

Figure 6.

Density gradient and qEV fractionation of Mycobacterium smegmatis crude MVs. (a) Crude MV preparations from duplicate M. smegmatis cultures (MS I and MS II) were fractionated separately by either density gradient centrifugation (left) or by qEV size exclusion chromatography (right). A new qEV column was used for each size separation. (b) Protein, RNA yield, and particle counts are graphed as a percentage per fraction of all recovered. (c) PAGE profile of MV proteins with 10 µL of each fraction from MS I crude MV preparation loaded and stained with SYPRO Ruby.