2. Milk-Derived EV Isolation: A Critical Point to Overcome
Different EV isolation methods have been developed on the basis of the physical or molecular features of vesicles
[14][15][14,15]. However, their small size makes this step a real challenge. In fact, a gold-standard method is still missing because a certain degree of bias remains due to the co-isolation of contaminants and protein aggregates. This is particularly true when the vesicles are isolated from complex matrices such as milk, which contains milk fat globules and spherical colloidal aggregates of caseins (casein micelles) that overlap in size with the EVs
[23][29]. In the case of milk, another consideration must be made related to the starting material. It is known that industrialisation processes can affect the composition of milk in terms of bioactive molecules in different manners related to the applied procedure
[24][25][30,31]. A great number of mEV evaluations have been made using raw milk, although its consumption is not allowed in some countries due to hygiene-related problems. This could lead to the use of mEVs derived from industrially processed milk, although modifications in mEV numbers and integrity have emerged that indicated the possible alteration of the bioactive component transfer through heat-treated commercial milk
[26][27][32,33].
A widely used method for EV isolation is based on differential ultracentrifugation (dUC), which allows the sedimentation of the solute through the use of an elevated centrifugation force
[28][34]. In general, some preliminary centrifugation steps are performed before dUC to eliminate cells and cell debris. Generally, 10,000×
g allows the isolation of apoptotic bodies and larger vesicles; around 35,000×
g can be used for medium vesicles; and 100,000×
g is appropriate for smaller EVs. The pellets can be further resuspended for a washing passage to repeat the same ultracentrifugation (UC) step for their recovery
[29][35]. Other than being one of the most time-consuming techniques requiring operator experience and an ultracentrifuge instrument, it is not suitable to isolate vesicles from milk due to the very low mEV purity
[30][31][32][33][34][36,37,38,39,40]. Centrifuging at such high speeds can damage the EV membranes and shapes and can favour vesicle aggregation and soluble protein co-isolation
[34][35][36][40,41,42].
To avoid excessive protein contamination, many strategies can be applied, including the addition of acetic acid, sodium citrate, or ethylendiaminotetracetyc acid (EDTA), which allow for protein precipitation as widely used for mEV isolation
[37][38][39][40][43,44,45,46].
Methods that increase the EV purity and separation of subtypes are based on the floatation on density gradients inserted between dUC steps, which allows for the separation of EVs from non-EV material including proteins or protein–RNA aggregates
[41][42][43][44][45][27,47,48,49,50]. These density gradients can be continuous (such as when using iodixanol) or discontinuous (when using sucrose); the samples can be loaded for a bottom-up or a top-down migration. The limitations of these methods are similar to those for dUC and are even more pronounced; they are linked to operator skills as well as the time and the instrument required, and also have a low-rate EV recovery.
One of the most used methods for EV isolation is size-exclusion chromatography (SEC)
[46][51], which was first applied to demonstrate the presence of different protein sets between the EV cargo and the solution
[47][52]. It exploits the dimensional features of EVs that constitute the mobile phase, which is made to flow inside the stationary phase of an SEC column consisting of a porous polymer. Extracellular vehicles with larger sizes elute first and travel quicker than the smaller ones, which can precipitate together with protein aggregates. However, the co-isolation of protein aggregates that share the same size range as EVs is a major drawback
[48][53], making the use of SEC alone unfeasible for mEV isolation due to the presence of a great number of proteins in the milk serum.
Similarly, ultrafiltration (UF) is an easy method for EV isolation that uses semipermeable membranes characterised by pores
[49][50][54,55]; however, it does not allow for separating the different EV subpopulations, and soluble proteins are recovered in the pellet together with EVs, which makes this method merely a useful preliminary step for volume reduction without causing EV damage
[51][52][56,57].
Precipitation-based protocols, including the use of polyethylene glycol (PEG), are easy to use and cheap and have allowed the recovery of large amounts of EVs with an elevated grade of standardisation and scalability
[53][58]. Unfortunately, these methods are characterised by a low grade of purity because all of the soluble particles precipitate, making them unusable for descriptive or functional analysis, particularly for mEVs
[54][55][59,60].
Another approach to vesicle isolation exploits the surface antigens of EVs through immunoaffinity-based methods, which are generally integrated with analytic tests such as ELISA-like or magnetic isolation
[56][61]. To ensure the recovery of the majority of the vesicles, several commercial kits have implemented the simultaneous use of the tetraspanins CD9, CD63, and CD81
[57][62] or wide EV-binding molecules such as heparin, heat-shock proteins, phosphatidylserine-binding peptides, or membrane-curvature sensor peptides
[58][59][60][63,64,65]. Immunoaffinity isolation is particularly useful in pathological fields such as cancer studies, although it is highly expensive.
Concerning the accuracy and purity when isolating EVs, flow cytometry (FC) is one of the most satisfying methods; it allows for a high-throughput, multiparametric analysis and the separation of single EVs based on their surface composition. However, particles measuring less than 600 nm are not detected due to the limit of the forward/side-scattered (FSC/SSC) light detectors
[61][66]. This is a problem that can be solved by using fluorescently labelled EVs or high-resolution flow cytometers, which allow for EV analysis and sorting
[62][63][64][67,68,69]. Although a better standardisation is needed, recent high-resolution FC showed advantages related to the very small sample volume required and the possibility of evaluating differences in the packaging of biomolecules during biogenesis
[65][70]. Despite their many advantages, applications to real milk have yet to be proven for these methods.
Some promising recently developed methods are based on field-flow fractionation (FFF) and microfluidics. Field-flow fractionation combines the application of a field (thermal energy, centrifugal force, electrostatic force, and cross/tangential flow applied through one or two semipermeable membranes) that allows the separation of the particles into different layers and a longitudinal flow that carries particles through the channel, which leads to their recovery
[66][71]. An upgrade to this technology was developed by Marsh et al.
[67][72] through the incorporation of solubilisation steps and the optimization of time, levels of temperature, and divalent cation chelation, which allowed for providing a pure and scalable production of mEVs
[67][72]. This is an emerging technology for EV separation that is characterised by important desirable features such as a high resolution, high purity, and potential for large-scale production. On the other hand, microfluidic technologies and on-chip biosensors have allowed for high-throughput analyses using a minimal sample volume and reagent consumption in integrated miniaturised devices. The EV separation can be done based on size, surface markers, or new sorting tools such as the application of acoustic, electrophoretic, or electromagnetic fields. Thus, these innovative methods lack a solid standardisation but represent the future of EV isolation
[68][69][73,74]. Meanwhile, a combined approach using at least two or three consecutive methods is recommended for an appropriate milk-derived EV isolation
[70][75]. It should be considered and remembered that different isolation methods can produce distinct fractions of EVs with different degrees of purity which can be reflected in nonhomogeneous effects in functional studies
[71][72][76,77].
3. The Theranostic Potential of Animal mEVs
Given the intrinsic beneficial effects of mEVs on gut homeostasis, one of the major promising animal mEV applications is their addition to infant formulae to prevent the development of necrotizing colitis in high-risk infants when breast milk is not available or their use as an adjuvant in IBD. Nevertheless, more studies are needed to clarify their role in cell proliferation, the prevention of fibrosis, angiogenesis, and in cancer treatments, as well as their unwanted side effects. Indeed, Melnik and Schmitz (2019)
[73][205] reported on the potential “dark side” of some molecules contained in pasteurised milk vesicles and highlighted possible adverse effects that the cargo could induce in the receiving cells and tissues by attributing to it a substantial risk for the onset of adulthood chronic metabolic diseases in Western countries
[73][205]. Although this direct cause-and-effect relationship is far from being demonstrated, the benefits appear to be tangible and have a strong potential. A large number of studies in the target therapy field that used EVs as a drug delivery system have recently been published
[74][206]. This application is been driven by the advantages that EVs showed compared to synthetic therapeutic nanocarriers such as liposomes, thereby revealing a wider biodistribution and biocompatibility and a higher internalisation rate
[75][207]. Furthermore, milk can help provide an elevated quantity of vesicles, in addition to being a widely available and inexpensive raw material that is particularly enriched in EVs, and thus represents a promising source of EVs for massive production. These mEV characteristics are ideal for theranostic applications.
Indeed, mEVs can be loaded with chemotherapeutics or other therapeutic molecules such as nucleic acids
[76][77][78][208,209,210]. Molecules can be loaded into mEVs through several methods; for example, for RNAs, exploiting molecules contained in mEVs such as GAPDH can help bind lactoferrin that has been electrostatically loaded with small interfering RNAs (siRNAs)
[79][211]. Fluorescently labelled siRNA loaded in mEVs through the lipofection process has been used to demonstrate EV protection and siRNA delivery with promising results
[80][212]. Warren et al.
[81][213] identified cationic chemical transfection as a good method for siRNA loading in mEVs; this turned out to be more efficient than electroporation. Moreover, coating with polyethylene glycol (PEG) increased mEV resistance in the acidic gastric environment and the permeability through the mucin layer, although a high enterocyte uptake and siRNA delivery were also demonstrated for unmodified mEVs
[81][213]. The ultrasonic coating of mEVs with B-cell lymphoma (bcl)-2 siRNA was demonstrated to be effective in reducing cancer cell proliferation and migration through the downregulation of metastatic-related genes both in vitro and in vivo
[82][214]. This method for administering siRNA using mEVs seems to be safe because no toxicity and side effects related to immune response and inflammation were detected after multiple dosings in mice
[83][215]. The miR-31-5p depletion typical of diabetic wounds was recovered by miRNA mEV loading through electroporation, which enhanced wound healing and angiogenesis
[84][216]. Moreover, mEVs were shown to be an efficient shuttle for the oral delivery of locked nucleic-acid-modified antisense oligonucleotides, which are potent RNA and protein modulators whose use in therapeutics was recently proposed
[85][217]. In addition, mEVs can be useful in the delivery of poorly absorbable drugs such as curcumin by improving the stability, solubility, bioavailability, and intestinal uptake through the oral route of administration
[86][87][218,219]. The anticancer effects of curcumin and resveratrol were amplified by their loading into mEVs, thereby helping to avoid the ATP-binding cassette transporter mechanism of chemoresistance normally used by cancer cells
[88][220]. The potent chemotherapeutic paclitaxel was one of the first to be tested for loading in EVs
[89][221] and was recently used in experiments with mEVs; it was shown that it could efficiently inhibit cancer growth, especially when orally administered instead of being administered by intraperitoneal injection, and showed no toxicity to normal cells
[77][90][209,222]. One study showed that the sublingual administration of mEVs loaded with the antidiabetic drug liraglutide reduced blood glucose in diabetic mice; however, the same effect was not detected when administered via oral gavage
[91][223]. These differences could be due to the efficiency of the drug loading into vesicles because the liraglutide–mEVs prepared with the extrusion method showed 2.45 times the drug load compared to the mEVs prepared via direct incubation, although the method was superior to sonication, freeze–thaw cycle, saponin-assisted, and electroporation methods
[92][224]. Loading features could be influenced by the species of origin when producing mEVs, as well as by the consequent functional efficiency. In a comparative study of cow, buffalo, and goat mEVs loaded with the chemotherapeutic doxorubicin through three different methods, goat mEVs showed a higher loading capacity across all of the methods compared to the other two species; this was reflected in their efficacy in inducing cell apoptosis in cancer cells in vitro
[93][225]. Doxorubicin-loaded-mEV target therapy was previously tested through the induction of hyaluronan expression onto the phospholipid bilayer that is a ligand for the CD44 receptor, thus allowing for CD44-overexpressing, cell-specific drug delivery
[94][226]. Indeed, mEVs have been evaluated for drug loading and its effects in several molecules and pathologies, especially cancer, with promising and encouraging results in terms of efficacy and safety; the drug power was amplified but the side effects were reduced by the action of the mEVs being more targeted
[95][96][97][98][99][227,228,229,230,231].
The use of EVs as theranostics, especially for human clinical applications, has rapidly grown in recent years. The ISEV held a workshop in 2018 on “EVs in Clinical Theranostic” to discuss technical issues and standardisation in order to increase the power of EVs in clinical practice
[100][232]. Concerning mEVs and animal clinical science, particularly regarding their use as biomarkers of pathological conditions, the research is in its infancy because the first study on the topic appeared just four years ago and only a few studies have been carried out to date. The mEV RNA contents in the mastitis of cows were the first studied; Cai et al.
[101][233] identified 18 differentially expressed miRNAs in the mEVs of healthy and infected animals whose targets participated in immune system processes and inflammation pathways, and the results indicated miR-223 and miR-142-5p as potential candidate biomarkers of mastitis
[101][233]. Later, two other miRNAs were shown to play important roles in the response to infection with
Staphylococcus aureus, miR-378, and miR-185, and were found to be particularly abundant and highly differentially expressed in mastitic milk compared to normal milk
[102][234]. A specific packaging of another type of small RNAs—the circRNAs—was highlighted by Ma et al.
[103][235] in response to bacterial infections because the presence of
Staphylococcus aureus induced a different set of circRNAs in mEVs, most of which were implicated in immune functions
[103][235]. In the case of subclinical mastitis, in which the symptoms are masked, the identification of biomarkers is crucial but more complex. Indeed, the molecules produced by cells, even those contained in mEVs, can vary between infection stages and according to other environmental variables. For this reason, Saenz-de-Juano and collaborators
[104][236] evaluated the changes in mEV size and concentrations using tunable resistive pulse sensing (TRPS) and the miRNA contents during three consecutive days of sampling by comparing the differences between mEVs from naturally infected udder quarters, their healthy adjacent quarters, and quarters from uninfected udders
[104][236]. Chronic subclinical mastitis did not show differences in the mEV number and size compared to healthy udders; the miRNA contents remained the same regardless of the health status of the quarter during the three sampling days. Only individual-cow changes were observed, which confirmed miR-223-3p as the most expressed miRNA in all of the chronic subclinical mastitis quarters
[104][236]. In addition, bovine leukemia virus (BLV) infection was investigated in order to highlight potential mEV biomarkers for preventing the spread of the virus and thereby reducing economic losses for farmers. Indeed, the mRNA and protein expression profiles of mEVs were found to be modified by BLV infections; their monitoring can be useful in evaluating the clinical stages of the infection
[105][106][237,238]. In particular, the concentrations of eight mRNAs were increased in the mEVs of infected cattle by Hiraoka and collaborators, which suggested a combined evaluation of the expression profiles of these genes to identify early BLV infections
[107][239]. On the contrary, miR-424-5p was found to be significantly upregulated in the mEVs of infected animals relative to healthy ones; this was also validated in the re-qPCR of a cohort of animals selected ad hoc, thus identifying it as a valuable biomarker to identify high-risk animals for BLV transmission
[108][240]. Recent work suggested the use of mEVs as biomarkers in heat-stressed cows; several differentially expressed miRNAs were detected whose targets were implicated in apoptosis, autophagy, and the p38 MAPK pathways, thus possibly regulating heat-stress resistance in dairy cows
[109][108].
Nevertheless, associating mEV cargo variations with specific pathological conditions is a challenge that needs further investigation due to the interchange between the metabolic imbalance and infectious disease/stressful stimuli that can occur in the mammary gland.