Background

Microbial extracellular vesicles (EVs) are fundamental for cellular physiology (Deatherage et al., 2012; Brown et al., 2015). For pathogenic microbes, EVs can be harnessed for diagnostics and vaccines. However, the study of EVs has been hampered by limited availability of tools which allow in-depth functional and mechanistic study of these small and extraordinary structures.

The protocol for isolating microbial EVs is well established (Rodrigues et al., 2008) and consists of isolation of culture supernatants followed by several (ultra)-centrifugation cycles to separate EVs from cells and debris. Several other EV isolation techniques have been developed in the last years, but these techniques are usually tested and implemented in the context of mammalian EVs and thus do not readily translate to microbial EVs. Researchers should follow the International Society for Extracellular Vesicles ISEV guidelines for EV isolation (Théry et al., 2018) when implementing such protocols.

In our most recent work (Coelho et al., 2019) we developed a density separation purification, a protection assay and a labelling protocol for the live-cell imaging of EVs from the pathogenic bacteria Listeria monocytogenes. In the protocols described here, growth conditions and media are optimized for L. monocytogenes, but our protocols should be readily adaptable to other microbes by adjusting the media and growth conditions. For example, density gradient purification was first developed for the pathogenic yeast Cryptococcus neoformans (Vij et al., 2018).

In this work we successfully used a multi-omics approach, a single-sample analysis of metabolite, protein, and lipid extraction (MPLEx) (Nakayasu et al., 2016) which provided a very cost effective analysis of EV composition. The MPLEx protocol has been well described elsewhere (Burnum-Johnson et al., 2017; Nicora et al., 2018).

Density separation methods for EVs have been used previously (Horstman and Kuehn, 2000; Elluri et al., 2014) and we adapted it to our specific needs. Density separation methods have two advantages over the use of EV preparations described in Protocol A: i) it allows the study the several subpopulations of EVs and, ii) it increases purity of EV preparations by removing contaminants such as soluble and aggregated proteins. Density separation of EVs extracted from 1 L microbial supernatant from L. monocytogenes (or C. neoformans) generates sufficient EVs for electron microscopy as well as dot blot protocols and we are confident that other analysis, such as proteomics and other -omics, can be performed. We have used variations of this protocol to separate melanin from the pathogenic yeast Cryptococcus neoformans (Camacho et al., 2019). Optiprep was used to prepare a step-density gradient, as we found it more straightforward to remove Optiprep from vesicle samples than other reagents used to prepare density gradients. In our experience, colloidal reagents, such as Percoll, were difficult to remove from preparations and subsequently interfered in downstream analysis of EVs by electron microscopy and western blotting.

Protection Assays in EVs have been performed before by our group (Rodrigues et al., 2008; Brown et al., 2014) as they allow to distinguish EV-internalized cargo from exposed cargo. Protection assays are considered the gold standard to show internalization of cargo versus adsorption or co-purification, and no other method (such as immunostaining and imaging by electron microscopy) is accepted as proof that cargo is internalized. Protection assays consist of subjecting the EV preparation to the action of degradative enzyme and then testing for degradation of EV components. In the specific case of L. monocytogenes, EVs were exposed to a protease (proteinase K or trypsin) that will digest any proteins external to EVs while proteins internalized in EVs are protected from proteases. Since L. monocytogenes secreted listeriolysin O (LLO) in EVs and this virulence factor is very efficient in lysing red blood cells we tested internalization of LLO in EVs by measuring lysis of red blood cells. Variations of these strategies simply use different degradative enzymes (proteases, nucleases, etc.) and test for protection inside EVs by measuring remaining amount of degraded product via immunoblot, activity assays, direct quantification, etc. The key factors for protection assays are gentle handling of the EVs to maintain their integrity, as well as appropriate experimental controls (the EVs from genetic-deleted strains and exogenously added controls such as recombinant proteins).

EV labelling in microbes has proven difficult. Despite wide usage of the lipophilic dyes DiO, DiL and PKH26, in our hands we noticed strong aggregation of EVs after labelling with DiO and DiL dyes (measured via dynamic light scattering). Other authors have reported similar problems (Morales-Kastresana et al., 2017; Dehghani et al., 2019) and we thus strongly discourage the use of DiO, DiL and PKH26 for the study of EVs. Alternatively, authors have used molecular tagging of EV cargo coupled to fluorescent or luminescent proteins. Those approaches are very successful when the protein cargo of EVs has been established, and if the cargo is amenable to molecular tagging. We developed a cargo-independent protocol which allows for live-tracking of bacteria, and EVs released by the bacteria, without altering the size (measured via dynamic light scattering) of EVs (Coelho et al., 2019). In brief, we allowed bacterial membranes to incorporate a conjugate of the fluorescent dye Bodipy558 to a long-chain fatty acid dodecanoic acid (BODIPY^® FL C12, Thermo Fisher Scientific) followed by high-resolution live-cell microscopy, allowing us to capture videos of release of EVs (Coelho et al., 2019). The strength of this approach is it can be easily applied to other organisms, from mammalian cells to bacteria and fungi, as other fatty acids-BODIPY dye conjugates exist (hexadecanoic for example). In addition, these conjugates are relatively cheap. For our particular labelling protocol we noted strong staining of bacterial cells as well as EVs. We reason that labelling with other fatty acids conjugates may allow more specific staining of EVs. We would recommend to readers to attempt labelling with other fatty acids- BODIPY conjugates as they optimize for their microbes of choice.