Background

In our previous publications (Benmoussa et al., 2016, 2017, 2019b and 2019c; Benmoussa and Provost, 2019), we highlighted that bovine milk is a complex fluid containing a myriad of extracellular vesicles (EVs) subsets. Among these, exosomes are ~100-nm vesicles released when multivesicular bodies (MVB) fuse with the cell membrane. When subjected to ultracentrifugation, these sediment at centrifugation speeds equal or higher than 100,000 x g (P100K, where P stands for pellet) (Pieters et al., 2015). Other non-exosome EV subsets are found in milk and are comparable to exosomes, in shape and size, but sediments at a lower speed (e.g., 12,000 x g, P35K; 35,000 x g, P35K; 70,000 x g, P100K). These are thought to be originating from budding of the cell membrane and contain specific proteins and microRNAs different from exosomes content (Benmoussa et al., 2016, 2017, 2019b and 2019c; Benmoussa and Provost, 2019).

For a long time, separating these different milk EVs subsets remained challenging because of their morphological and chemical similarities. Moreover, protocols used in milk EV studies were often centered on the isolation of milk exosomes specifically, leading to the discarding of the biologically active non-exosome milk EVs (Benmoussa et al., 2019a, 2019b and 2019c).

When using differential ultracentrifugation to isolate milk EVs, the first steps are often impaired by the co-precipitation of milk proteins with EVs, which form a dense jelly at the bottom of the tubes (Zonneveld et al., 2014). Such jelly is formed when caseins are subjected to high mechanical pressures (Famelart et al., 1998; Zonneveld et al., 2014). Because this jelly is so dense, and practically impossible to resuspend, it is not clear whether some EV subsets are trapped within its matrix and some protocols recommend discarding this casein-rich jelly.

Certain reports suggested the use of density gradients or cushions to keep the EVs from mixing with the casein jelly (Zonneveld et al., 2014). Others suggested avoiding the formation of such jelly by discarding caseins before subjecting milk “serum” or whey to differential ultracentrifugation. To this end, milk is often mixed with acids (Somiya et al., 2018) or with cold EDTA to precipitate the caseins prior to milk EVs sedimentation (Wolf et al., 2015). However, such preprocessing of biological fluids have an immense impact on the isolation, quality and yield of extracellular vesicles (EVs) (Zonneveld et al., 2014). It also requires to discard low speed-pelleting EVs (P12K or P35K). In addition, there is a lack of information about the effect of acidification, or casein precipitation, on milk EVs, and the possibility that some EVs might be lost along the process.

This is especially important because there are multiple EV subsets in milk (Benmoussa et al., 2017 and 2019b) and because milk whey has a different microRNA and protein content than milk (Benmoussa et al., 2019b; Benmoussa and Provost, 2019), which, as microRNA are found within EVs, suggests the loss of certain EVs during casein precipitation. This is also supported by the discovery of microRNAs in milk-derived casein-rich products, like cheese (Benmoussa and Provost, 2019). Therefore, the biological activity of the milk EVs, and the bioactive molecules they transport, depend highly on the reagents/preprocessing steps and isolation protocols this fluid is exposed or subjected to (Zonneveld et al., 2014).

When we encountered this issue, we experimented different venues to avoid precipitating milk caseins while ensuring the separation of different structurally conserved, and functional, milk EV subsets. In milk, casein proteins are arranged in the form of micelles that are 70 to 200 nm in diameter, which is close to the size of certain EV subsets (Blans et al., 2017). Knowing that calcium is important for maintaining these micelles (de Kort et al., 2009, 2011 and 2012; Kort et al., 2012), we supposed that calcium chelation would prevent the formation of casein superstructures that hamper EV isolation by ultracentrifugation.

Several calcium-chelating agents have been previously investigated for their ability to disrupt casein micelles, including disodium uridine monophosphate (Na₂UMP), disodium phosphate (Na₂HPO₄), trisodium citrate (referred to simply as "sodium citrate"), sodium phytate, sodium hexametaphosphate (SHMP) or ethylenediaminetetraacetic acid (EDTA) (Ward et al., 1997; de Kort et al., 2009, 2011 and 2012; Kort et al., 2012). These chelating agents interact with colloidal calcium phosphate (CCP) leading to the breaking of casein micelles into small monomers. This process augments the stability of milk during heating and change its physical properties, often reducing its viscosity (de Kort et al., 2009, 2011 and 2012; Kort et al., 2012).

Among these, sodium citrate is a biocompatible compound known for 100+ years for its ability to prevent casein curdling in the stomach of infants (Poynton, 1904). It is widely used to preserve biological fluids, like blood (Janse van Rensburg and van der Merwe, 2017), as a food additive and as the major form of rehydration salt recommended by the World Health Organization (Pizarro et al., 1986; Banipal et al., 2016). Notably, it was previously used to disrupt casein micelles to facilitate the isolation of milk fat globules (MFGs) by diafiltration and prevented pore clogging (Phan et al., 2014). Interestingly, sodium citrate specifically impacts milk protein gel formation upon high mechanical pressure, although the underlying mechanisms remain unclear, except for its dependence on the solution’s pH (Famelart et al., 1998).

In the course of our experimentations with sodium citrate, we found that pre-treating milk with sodium citrate (1% final) was a cost-effective way to avoid casein jellification and isolation of high quantities of different milk EVs using differential ultracentrifugation. We hereby detail the methods underlying this method that led to the discovery (Benmoussa et al., 2016 and 2017) and characterization (Benmoussa et al., 2017, 2019b and 2019c) of different EVs subsets in milk, keeping them functional and able to transfer their content to human cells (Benmoussa et al., 2019c). EVs, including exosomes, isolated by this methodology were able to modulate intestinal inflammation during experimental colitis (Benmoussa et al., 2019a).