Ruminant Milk-Derived Extracellular Vesicles: A Nutritional and Therapeutic Opportunity?

Milk has been shown to contain a specific fraction of extracellular particles that are reported to resist digestion and are purposefully packaged with lipids, proteins, and nucleic acids to exert specific biological effects. These findings suggest that these particles may have a role in the quality of infant nutrition, particularly in the early phase of life when many of the foundations of an infant’s potential for health and overall wellness are established. However, much of the current research focuses on human or cow milk only, and there is a knowledge gap in how milk from other species, which may be more commonly consumed in different regions, could also have these reported biological effects. Our review provides a summary of the studies into the extracellular particle fraction of milk from a wider range of ruminants and pseudo-ruminants, focusing on how this fraction is isolated and characterised, the stability and uptake of the fraction, and the reported biological effects of these fractions in a range of model systems. As the individual composition of milk from different species is known to differ, we propose that the extracellular particle fraction of milk from non-traditional and minority species may also have important and distinct biological properties that warrant further study.


Introduction
Milk is the only food that has evolved to meet the nutritional needs of newborns, supporting growth and development while also being a significant source of nutrients in adults [1][2][3]. The domestication of livestock was a pivotal step in the consumption of non-human milk which has become a substantial source of essential nutrients in many diets globally [4][5][6]. To meet this demand, the production of milk increased from 708 million tonnes in 2009 to 883 million tonnes in 2019, with cow and buffalo milk accounting for 81% and 15% of production, respectively (Supplementary Table S1) [7].
In early life, major milk components such as lactose (energy source), minerals (musculoskeletal development), and high-value biological proteins provide essential nutrition [8,9]. Milk consumption throughout life can also address malnutrition and can represent a significant proportion of overall nutrient intake in developing nations [10].
While the milk macro-and micronutrient composition is largely well established, there is considerable interest in milk-derived extracellular vesicles (EVs) and their cargoes as a source of nutrients in the classical sense, such as nucleosides, and amino acids, or as a nutritional component that influences biological functions by regulating biochemical pathways and/or interactions with the host's gut microbiome [11][12][13][14][15][16][17][18]. Evolutionary theory suggests that milk-derived EVs and their cargoes must have a biological purpose to justify the metabolic cost required to produce them during lactation.
Our review covers the following: a summary of the nutritional composition of the types of milk that have been used to study milk-derived EVs, the nature and composition of these vesicles and their cargoes, the evidence for their stability and uptake in the gastrointestinal tract, their reported biological effects, and some of the key challenges in using them for studies. Methods used to identify peer-reviewed studies are shown in Figure 1. Our review excludes any studies on plant-derived milk alternatives.

Nutritional Composition of Milk
Milk from minority dairy species, i.e., not cow milk, is more widely consumed in regions with a harsh environment which requires animals with specific adaptations [9]. It has generally not been studied in as much detail for nutrition or bioactivity as cow milk despite the evidence of substantial compositional differences in the different types of proteins, lipids, micronutrients, and bioactive components between milk from different species. The macronutrient composition of milk from different mammals has been extensively studied and is readily available in the public domain. A list of the milk composition from different mammals is collated in Table 1, based on the different quantification methods used and data reporting across different databases (the conversion can be found in Supplementary Table S2). Table 1. Gross composition of milk of different mammals obtained from available food composition databases. Data are presented as an average (±standard deviation) per 100 g.

Carbohydrate
Lactose is the primary carbohydrate in milk, providing 30% to 60% of energy depending on the species of milk [10,[30][31][32]. It also enhances intestinal mineral absorption (e.g., calcium, sodium, magnesium, and phosphorus), utilisation of vitamin D, and stool softness [4,10,30,33]. Hydrolysis of lactose by the enzyme lactase into a simpler form of sugars is essential for intestinal absorption in humans [10,34]. Lactase deficiency contributes to the fermentation of lactose in the colon by microorganisms, producing gases (hydrogen, carbon dioxide, and methane), organic acids (acetic, butyric, and propionic acid), and excess water in stool, leading to uncomfortable bowel activity such as diarrhoea, flatulence, and bloating [1,10,35,36].
Many forms of oligosaccharides are also present in milk, contributing to the gut microbiome diversity in infants [17]. In humans, milk oligosaccharides are the third most abundant macronutrients (7 to 20 g/L) after lactose and lipids [31,37], but in other mammals, the milk oligosaccharides content is 10 to 100 times lower [37].

Fat
Milk fat occurs in emulsified droplets known as milk fat globules (MFGs) that are mainly triacylglycerols (97-98% of total lipids by weight, including a large number of esterified fatty acids and phospholipids), as well as proteins and fat-soluble vitamins [10,38]. The tri-layered phospholipid membrane of an MFG is designed to protect its contents from lipolysis and oxidation [38]. The roles of milk lipids and MFGs in health have recently been reviewed [39,40]. German and Dillard [41] reviewed the composition, structure, function, absorption, and bioactivity of human and cow milk lipids, noting the importance of considering the role of MFGs.
In general, the lipid composition of milk fat also differs from one species to another. Zou et al. [42] compared the lipid composition of five mammalian milks (cow, buffalo, donkey, sheep, and camel) to human milk by evaluating the degree of the chemical similarity of the samples. This showed that although the total fatty acid composition of certain non-human milks was highly similar to that of human milk (e.g., 96.4% similarity in sheep's milk), there were substantial differences when it came to the individual chemical species (e.g., only 20.2% similarity in polyunsaturated fatty acids of sheep's milk). Devle et al. [43] measured the fatty acid profiles in the milk of three ruminants (cow, goat, and sheep) and two non-ruminants (donkey and horse) and found a substantial diversity in the occurrence and abundance of them between species and their degree of correlation with health attributes.
The unique ability of milk caseins to form macromolecule aggregates (casein micelles) with minerals such as calcium and phosphorus improves the bioavailability, delivery, and intestinal absorption of these minerals [4]. The industrial importance of ruminant milk proteins in cheese production and secondary transformation products has led to the extensive study of these components, such as the proteomic analysis of several forms of milk: as a whole [45], the whey fraction [48,49], and sub-fractions of whey such as caseins [50] and MFGMs [51][52][53][54].
Roncada et al. [55] reviewed advancements and challenges in the proteomic analysis of milk from farm animals, together with an overview of the different components in the milk fractions. Similarly, Malacarne et al. [56] systemically reviewed the composition of horse, human, and cow milk from the perspective of protein and lipid fractions, proposing that the nourishment provided by horse milk is more similar to human milk than that provided by cow milk.

Micronutrients
Micronutrients are essential nutrients that cannot be synthesised by humans and must be provided through our diet or other means [57]. The consumption of two to three servings of milk or milk products provides the required nutrient intakes for several important micronutrients (calcium, magnesium, selenium, riboflavin, vitamin B12, and pantothenic acid) [10]. Milk has comparatively fewer absorption inhibitors (e.g., oxalate and phytates) than other foods, which improves the bioavailability and absorption of these micronutrients [8,10].
The major milk minerals, calcium and phosphorus, which are required for optimal bone health are more bioavailable due to the mineralisation of casein micelles in both the insoluble organic colloid and mineral forms [58]. Medhammar et al. [9] highlighted the differences between the mineral profiles of different milk species, with moose and reindeer milk having the highest concentration of most essential minerals, and horse and donkey milk having the lowest. Milk also provides water-and fat-soluble vitamins due to the dual-phase matrix of lipid micelles suspended in the aqueous environment. Milk vitamin profiles are broadly consistent, with vitamin C having the highest concentration, and vitamins B12 and D having the lowest concentrations, with some species differences [9,59]. Graulet [57] reviewed the role of ruminant milk, with an emphasis on cow's milk, in meeting the required vitamin consumption by humans.

Other Milk Components
Admyre et al. [60] identified the presence of immune-modulatory exosomes in human milk which led to research into how exosomes (one type of EV) and their cargoes may have a role in inter-cellular, inter-individual, or inter-species communication. There has been substantial interest in milk-derived EVs as a novel bioactive fraction of milk [11,12,16,[61][62][63].
However, the majority of research has focused on human or cow's milk, and not minority dairy species, and therefore this paper reviews the key technical challenges and reported biological activities of ruminant and pseudo-ruminant milk-derived EVs.

Milk-Derived EVs
According to the Minimal Information for Studies of Extracellular Vesicles 2018 (MI-SEV2018) guidelines, "extracellular vesicle is the generic term for any particle naturally released from the cell that is delimited by a lipid bilayer and cannot replicate, i.e., do not contain a functional nucleus" [64]. Due to historical differences in how these vesicles were isolated, characterised, and named, the guidelines recommend using the term "extracellular vesicle" instead of other terms such as "exosome" or "microvesicle", except when the biogenesis or release pathway is investigated [64]. However, in this review, the terminology used in the original paper cited will be used. The MISEV guidelines provide experimental and reporting guidelines specific to the field of EVs [64][65][66], and several curated public knowledgebases promote the transparency and reproducibility of EV experimental studies [67][68][69][70][71][72][73][74][75]. Recent advances in the use of flow cytometry to study EVs have led to a standardised experiment and reporting framework (MIFlowCyt-EV) [64,76,77].
EVs are heterogenous populations that are categorised based on their biogenesis pathway. In brief, exosomes (~30 to 150 nm) originate from the intraluminal vesicles via the endosome trafficking pathway, while microvesicles (100 nm to 1 µm) result from direct budding from the plasma membrane of the parental cell, and apoptotic bodies (1 to 5 µm) are shed from cells undergoing apoptosis [78,79]. To date, most of the available methodologies cannot isolate a pure subpopulation of EVs; therefore, a defined mixed population is widely used for studies. The progress in understanding EV biology in the context of inter-or intra-species signal mediators, due to their diverse cargo (mRNA, miRNA, protein, etc.), has spawned a growing interest among the research community. An in-depth review of EV heterogeneity [78] and cell biology [80] proposed a need for a clear definition of the different subpopulations, based on cargo composition, trafficking pathways, and biological functions. Several other detailed reviews focused on other aspects of EVs such as biogenesis [81,82], delivery or target mechanisms [83][84][85], and current advances in knowledge [86]. The majority of publications on EVs have been focused on human growth, development, homeostasis, and disease progression, and several other reviews of milk-derived EVs are summarised in Table 2.

Isolation of Milk-Derived EVs
From complex biofluids to simpler in vitro cell culture media, different isolation methods may be employed to minimise the presence of unwanted artefacts which could jeopardise the downstream analysis. Review articles or book chapters on EV isolation techniques are readily available in the literature, from providing a brief overview [87][88][89][90][91][92] to a comprehensive discussion [93][94][95][96].
EVs' isolation relies on their separation from contaminants such as proteins and other particles through the use of known biophysical and/or biochemical properties: size, buoyant density, surface charge, surface molecules' expression and their composition. Several articles dedicated to a specific scope of isolation techniques are worth mentioning. Li et al. [97] discussed the different isolation strategies for human biofluid-derived EVs which have been employed in mass spectrometry (MS)-based proteomic studies for the past decade (2009-2019). Another review article highlighted the usefulness of size exclusion chromatography (SEC) in EV isolation, given that this approach is highly scalable and adaptable while maintaining the EVs' characteristics [98]. A three-step filtration protocol comprising dead-end filtration, tangential-flow filtration, and track-etched membrane filtration was proposed by Heinemann and Vykoukal [99] to provide an approach to concentrate and fractionate samples with minimal forces applied on EVs. The progression in microfluidics-based platforms in the past decade has enabled the rapid separation of EVs from small sample volumes. A review by Meng et al. [100] highlighted the interesting advancements in the microfluidic separation of EVs based on the different separation principles.
Characterisation of isolated EVs still largely uses immunochemical (e.g., ELISA, Western blot), MS-based, and optical (e.g., nanoparticle tracking analysis (NTA), microscopy, and flow cytometry) methods. However, any of the single aforesaid detection approaches may not be sufficient to address the issues of specificity, efficiency, and consistency in EV detection. More often, multiple detection approaches are employed within the research community when it comes to EV characterisation. The progression in analytical sciences has pushed for the development of new and innovative instruments to meet the abovementioned challenges. Recent review articles have summarised the emerging new technologies available that are specifically developed for EV characterisation [94,[101][102][103][104].
Methods for the isolation and characterisation of milk-derived EVs have no significant differences compared to those for isolation from other biofluids or cell culture media; thus, any protocol, technique, or technology for isolation of EVs of different origin can also be used for those from milk. The only difference between these types of samples is the unwanted artefacts present in different biofluids (e.g., lipoproteins in blood serum, or casein aggregates in milk). A simple method for isolating EVs from breast milk was described by Wang [50], which only requires a proprietary precipitation reagent (ExoQuick), a benchtop centrifuge, and a few common lab consumables; however, this method isolates a crude preparation of EVs. Several other approaches have been developed and used to study EVs (Table 3). [107] Cow UC, SEC, PR, membrane affinity column, PS-affinity isolation SEC-based qEV column (Izon Science) yielded high purity (high EV count per mg protein) and a large amount of RNA with minimal operation time.
[108] [110] Cow AA+UC, C + UC AA+UC method yielded lower protein content, but EV protein markers (CD81, Rab5B, TSG101, and Hsc70) were reported to be present in high abundance. Proteome analysis revealed C/UC EV fraction contains whey proteins such as casein, albumin, lactoferrin, and lactoglobulin. [111] Cow Total particles and Annexin V + particles measured using flow cytometry (Canto II and Cytoflex) and NTA (NanoSight) Significant correlation of total particle counts using Cytoflex and NanoSight and for Annexin V+ particles using Canto II and Cytoflex. [112] Cow + HCT 116 cell line + Ascaris suum

AFM-based force spectroscopy (FS)
Demonstrated an AFM-based characterisation strategy with the ability to discriminate EVs from contaminants. [113]

Exosome
Human Novel solid-phase extraction in tip-based format Demonstrated successful recovery of spiked lyophilised human urine exosomes from 3 different matrices (mock urine, reconstituted non-fat milk, and foetal bovine serum).
[ The terminology used is based on the reference cited, and this division reflects older thinking and is a "pool" of EVs that are responsible for the effects (AA, acetic acid; AFM, atomic force microscopy; C, centrifugation; DC, differential centrifugation; DG-UC, density gradient ultracentrifugation; EDTA, ethylenediaminetetraacetic acid; IP, isoelectric precipitation; NTA, nanoparticle tracking analysis; PR, precipitation reagent; PS, phosphatidylserine; SEC, size-exclusion chromatography; UC, ultracentrifugation).

Protein Composition of Milk-Derived EVs
The application of MS-based proteomic profiling and protein quantification has been of substantial significance in EV research to allow the identification and quantification of EV proteomes from various cultured systems, organs, body fluids, or plants. Several review articles provide a high-level overview of the MS-based methodological approaches widely used in EV studies [79,[120][121][122][123]. MS-based proteomic quantitative analysis can be achieved with either a labelled (e.g., isobaric tags for relative and absolute quantification (iTRAQ); stable-isotope labelling of amino acids (SILAC)) or label-free approach which quantifies proteins based on their spectral intensity or counts [79,120]. Data generated from MS consist of large datasets with functional analysis of these data needed for the identification of biological processes, which includes the Gene Ontology (GO) term annotation, enrichment analysis, and/or pathway analysis [124].
The early discovery of several EV-enriched protein markers (tetraspanins, heat shock proteins, annexins, etc.) from isolated EVs derived from in vitro cell models using MSbased proteomic characterisation occurred in the early 2000s [125][126][127]. Admyre et al. [60] first reported the investigation of the mammalian milk-derived EV proteome using a tandem MS approach to verify several important EV protein markers (tetraspanins, heat shock proteins, MUC-1, etc.) from the human colostrum and mature breast milk-derived exosomes, respectively. Building on this, Reinhardt et al. [128] identified 2107 proteins in a comprehensive study of cow milk-derived exosomes by utilising two-dimensional liquid chromatography-based separation coupled with tandem mass spectrometry. These studies led to several characterisation papers, summarised in Table 4 [128][129][130][131][132][133][134][135][136][137]. The literature demonstrates that milk-derived EVs have a distinctive proteome compared to other milk fractions and that a significant proportion of these proteins have reported immune-regulatory properties. Cow + donkey + goat UHPLC-HRMS Metabolomic analysis of 5 different pools of fractions obtained from differential centrifugation from 3 different species.
[141] Horse MALDI-ToF Identification of exosome-associated proteins, CD81 and CD63, in horse milk. [144] The terminology used is based on the reference cited, and this division reflects older thinking and is a "pool" of EVs that are responsible for the effects (µLC-MS/MS, micro-flow liquid chromatography-tandem mass spectrometry; CDMS, charge detection mass spectrometry; iTRAQ, isobaric tags for relative and absolute quantification; MALDI-ToF, matrix-assisted laser desorption ionisation-time of flight; MFGM, milk fat globule membrane; nLC-MS/MS, nano-flow liquid chromatography-tandem mass spectrometry).

Lipid Composition of Milk-Derived EVs
As different EV subtypes (exosomes, microvesicles, and apoptotic bodies) are categorised, in part, based on their respective biogenesis pathways, the membrane lipid composition of EVs resembles that of the parent pathway [93,145]. Understanding the lipid composition of EVs, such as sphingolipids, ceramides, phosphatidylserine, and the lipid raft component cholesterol, is an essential part of their biology, biogenesis, and biological function [121,146]. Several of the analytical chromatography and mass spectrometry techniques routinely used in EV proteomics have also been used for the qualitative and quantitative assessment of EV lipids. The challenges, limitations, and current knowledge of EV lipidomics have been reviewed elsewhere [121,145,[147][148][149][150].
There are particular challenges in isolating lipids from milk-derived EVs due to the co-isolation of milk lipids (i.e., MFGs), and in milk EV isolates with a high triacylglycerol content (TAG), since MFGs contain a greater amount of TAGs in their core than EVs [38,119], potentially interfering with accurate EV lipidomic studies. In this review, MFG lipidomic studies are not included because the biophysical properties (tri-layered membrane) and cargoes of MFGs are distinctly different from those of EVs.
To date, two studies have specifically examined the lipid composition of EVs [119,151]. Blans et al. [119] successfully applied size exclusion chromatography to human and cow milk samples to isolate distinct fractions of EVs and MFGs; these were partly characterised by the notably higher TAG-to-cholesterol ratio in human and cow MFGMs in MFGs when compared to EVs. The authors also reported a higher proportion of sphingomyelin, phosphatidylserine (PS), and phosphatidylcholine (PC), and a lower proportion of phosphatidylethanolamine (PE) in EVs compared to MFGs. Yassin et al. [151] reported concentrations of~10 to 15 µg/mL of phosphatidylinositol, PS, and PE, and~20 to 25 µg/mL of PC in dromedary milk exosomes which were consistent during different lactation periods.
Phospholipids, such as those reported in EVs, have been associated with beneficial health effects [152][153][154][155][156]. We recognise that there is a knowledge gap in the understanding of the lipid composition of mammalian milk-derived EVs, which is essential to understanding the biology of their function and biogenesis mechanisms [157].

Nucleic Acid Composition of Milk-Derived EVs
Milk-derived EVs contain nucleic acid cargoes, proposed to be derived from mammary epithelial cells, encased within the cytosol of a lipid bilayer vesicle [158,159]. Studies characterising the milk-derived EV transcriptome are summarised in Table 5.
Of interest is the presence of microRNA (miRNA) in milk; these are short nucleic acids of~22 nucleotides and are known for their role in post-transcriptional regulation. In milk, these miRNAs are present in two main forms: bound to RNA-binding proteins, or encapsulated in EVs [12]. The abundance of miRNAs in milk has generated substantial interest and research into whether these miRNAs are bioavailable and bioactive. Many studies have focused on the potential involvement of milk-derived EV miRNA in inter-cellular crosstalk, inter-individual communication (breastfeeding), and cross-species communication (due to human consumption of other mammalian milk throughout adulthood). However, it is noted that the concept of ingested miRNA from another species surviving digestion and being absorbed in sufficient quantities to elicit a quantifiable biological effect remains to be convincingly shown, despite several promising studies [109,[160][161][162].
However, the presence of miRNA in milk-derived EVs suggests that they have a potential role as natural or modifiable therapeutic agents to improve or enhance human and animal health. For instance, there are studies evaluating the milk-derived EV transcriptome for use as nanotherapeutic agents [184][185][186], as disease biomarkers [158,179,187], differential mediators [188,189], and as a health assessment tool for lactating animals [175]. Conversely, the role of milk-derived EVs as a functional regulator has also generated concerns because continuous consumption of dairy may contribute to the pathogenesis of common Western diseases such as type 2 diabetes mellitus, allergies, and cancers [13,[190][191][192][193][194].
In brief, there are two broad schools of thought regarding the specific role of milkderived miRNAs in postnatal development: (1) the functional hypothesis, which proposes that these miRNAs are purposefully transferred by the parent to the offspring to exert meaningful epigenetic regulatory functions in the infant's development, and (2) the nutritional hypothesis, which proposes that the degradation of miRNAs in the gut during digestion into nucleotides means that they are only nutritional "building blocks" for the infant only and do not exert any meaningful regulatory functions [191].
The studies listed in Table 5 show that: (1) RNA (especially miRNA) is present in milk-derived EVs and other extracellular particles, (2) some of these miRNAs are conserved between species, (3) some of these miRNAs are specifically found in extracellular particles, and (4) biological dysfunction, such as disease, can alter miRNA abundance. On the assumption that conservation of the miRNA sequence implies conservation of function, much of the research into the biological effects of milk-derived EVs has focused on their miRNA cargoes and their effects on immune regulation. These studies, and others, are reviewed later. Table 5. Major findings of nucleic acid studies conducted on EVs of different mammalian milk used to characterise the RNA composition of milk-derived EVs and exosomes.

Species
Technique Findings Ref.

Extracellular Vesicle
Human NGS Total of 1523 miRNAs identified with more than one read in 70% of samples from the Faroe Islands cohort (364 mothers). [195] Human qPCR Total of 55 lncRNAs identified with 11 lncRNA detected in >50% of the breast milk samples and 5 in >90%. The authors suggested the packing of highly correlated lncRNAs is regulated by the same pathway. [196] Human NGS Total of 5 miRNA stably expressed in all groups. Total of 4 (probiotic + ) and 5 (atopic dermatitis + ) miRNAs differentially expressed. No evidence of maternal probiotic ingestion altering miRNA abundance, unlikely for probiotic protective effect to be transferred to the infants. [172] Human + Pig qPCR, NGS Human and Pig In silico Reported the presence of plant miRNA in both human and pig milk exosomes based on publicly available sequencing data. [198] Cow Qpcr Demonstrated the bioavailability of cow milk exosomal miRNAs in human plasma without eliciting a cytokine response ex vivo (human PBMCs). [199] Cow PCR, NGS Total miRNAs: 290 detected, with 69 novel miRNAs. Total of 37 miRNAs differentially expressed due to infection. The predicted target genes for 2 miRNAs highly expressed in infected samples, bta-miR-378 and bta-miR-185, were functionally validated with target genes.

Human qPCR Microarray
Total of 281 miRNAs detected. Expression of miR-181a and miR-17 was detected in CD63-positive human milk exosomes. [202] Cow qPCR Six different cow colostrum exosome isolation methods were compared. Method 2 (conventional: differential centrifugation) had the highest purity and greatest amount of microvesicular miRNAs quantified. [203] Cow qPCR Identification of selected mRNA and miRNA in microvesicles, unaffected by acidification, and in vitro transfer of RNA from samples. [204] Buffalo qPCR The expression of 6 nanovesicular miRNAs from three biofluids was evaluated, and 2 of them (miR-21 and miR-500) were reported to be stably expressed during different household storage conditions. [205] The terminology used is based on the reference cited, and this division reflects older thinking and is a "pool" of EVs that are responsible for the effects (circRNA, circular RNA; lncRNA, long non-coding RNA; miRNA, micro-RNA; NGS, next-generation sequencing; qPCR, quantitative PCR).
These studies show that the structure of EVs protects them against harsh conditions, such as low pH, temperature variations, or high concentrations of RNase. This capacity to resist degradation and digestion underpins the study of their potential biological effects, either as nutrient delivery or as drug delivery vehicles. Table 6. Major findings of studies on the general stability and uptake of mammalian milk-derived EVs.

Species
Findings Ref.

Extracellular vesicle
Human Stability and uptake of natural and synthetic EVs loaded with locked nucleic acid anti-sense oligonucleotides in vitro (PHH, NCI-H460 cell line, and hPSC) and in vivo (mice). [216] Cow The impact of industrial processing on milk EVs' structural integrity and molecular composition. [219] Cow Cellular internalisation of EVs in vitro (hPAEC and NRCM). [109] Cow Development of non-invasive fluorescent labelling of EVs in vitro (Caco-2 cell line), demonstrating internalisation and co-localisation of labelled EVs. [220] Cow Time-dependent uptake of colostral miRNA, EV proteins, and isomiRs after feeding in vivo (calves). [221] Cow Demonstrated that microwaving, but not autoclaving, agitation, or freezing, reduced miR-220c abundance. [207] Exosome

Human
Resistance of exosomes isolated from preterm human milk to in vitro digestion and internalisation in vitro (HIEC). [169] Human Exosomal protein markers resist degradation by in vitro digestion, pH 4.5, and the uptake of digested and undigested exosomes, based on immunofluorescence imaging of exosomal protein markers in vitro (HIEC). [171]

Human
Resistance of miRNA to degradation caused by incubation at 26 • C over 24 h, six freeze-thaw cycles at −20 • C, treatment with RNase A and RNase T1, and incubation at 100 • C for 10 min. [173] Human Demonstrated the uptake of RNA ex vivo (macrophages). [217] Human + Cow Storage at 4 • C substantially reduced the exosome content, especially miRNA, of human milk over time, and the infant formulae tested had no detectable miRNA. [167] Cow Assessed the accumulation and effects of milk exosomes and miRNA cargoes on embryo development in C57BL/6 mice. [222] Cow Resistance of lncRNA to degradation by in vitro digestion. [177] Cow Resistance of paclitaxel (chemotherapeutic), encapsulated in these exosomes, to degradation and loss of efficacy from long-term storage at −80 • C for 4 weeks. [213] Cow Resistance of 5 miRNAs to degradation by an in vitro digestion method and in vitro internalisation of exosomes. [159] Cow Uptake of exosomes and exosome-encapsulated siRNA (both digested and undigested) in vitro (Caco-2 cell line). [208] Cow Fermentation of milk exosomes with probiotic Streptococcus thermophiles, Lactobacilli, and Bifidobacteria reduces miR-29b and miR-21 abundance and total protein concentration. [201] Cow Challenged the findings from a previous study [160] regarding the dietary transfer of cow milk-derived miRNA in humans. [214] Cow Demonstrated that miR-223 and miR-125b persist in high abundance after simulated in vitro digestion (TNO TIM-1 model). Authors found that exosomes may not be the only carrier of these miRNAs in milk. [211] Cow Uptake and bioavailability of fluorescent-labelled exosomes and their miRNA cargoes via endocytosis in vivo (C57BL/6 mice) and in vitro (HUVEC). [210] Cow Resistance of native miRNA and anticancer compounds encapsulated in these exosomes to degradation from long-term storage at −80 • C for 6 months. [212] Cow Uptake of miRNA in differentiated and undifferentiated THP-1 cells.

Species Findings
Ref.

Microvesicles/Nanovesicles/Other
Human Presence of immune-related miRNA in human milk, two of which were present in exosomes. miR-21 and miR-181a were resistant to degradation by RNase, pH 1, and freeze-thaw, indicating an extracellular protective mechanism. [202] Cow Pasteurisation and homogenisation, but not 4 • C storage, substantially reduce the abundance of miR-200c and miR29b in four types of milk tested. Somatic cells in the milk accounted for <1% of the abundance of these miRNAs in milk, consistent with these miRNAs packaged in extracellular structures such as EVs. [215] Cow Presence of mRNA and miRNA which were resistant to degradation by RNase, pH 2, incubation at 37 • C, but not Triton X-100, indicating an extracellular protective mechanism. [206] Cow Presence of mRNA and miRNA in both samples. These RNAs were resistant to degradation by pH 2, indicating an extracellular protective mechanism. [204] Buffalo Demonstrated that 4 • C storage and multiple freeze-thaws reduced the abundance of miR-21 and miR-500. [205] The terminology used is based on the reference cited, and this division reflects older thinking and is a "pool" of EVs that are responsible for the effects (HIEC, human intestinal epithelial crypt-like cell; hPAEC, human pulmonary artery endothelial cell; hPSC, human pluripotent stem cell; HUVEC, human umbilical vein endothelial cell; NRCM, neonatal rat cardiomyocyte; PBMC, peripheral blood mononuclear cell; PHH, primary human hepatocyte).

Biological Effects of Milk-Derived EVs
The previous sections indicate that milk-derived EVs may contain bioactive components, which are protected against degradation and digestion. The studies to date that have focused on the biological effects of milk-derived EVs are highlighted in Table 7. The majority of these studies used human or cow milk EVs, and there is a clear knowledge gap regarding whether milk from other species has similar or different effects.
Whether these effects in animal and in vitro models translate into humans is unclear. A question remains concerning whether the effects are solely due to EVs and their cargoes or also due to other variable contaminants (e.g., RNA-binding proteins) in the vesicle preparations used in the published studies. The increased rigour and reporting required to comply with the MISEV guidelines are intended to enable more thorough validation of EV research. Table 7. Major findings of studies on the biological effects of mammalian milk-derived EVs.

Species
Findings Ref.

Extracellular vesicle
Human Protective effect in vitro (MA-104 and Hep-2 cell lines) against human rotavirus and respiratory syncytial virus. [254] Human In vitro (HFF-1 cell line) antiviral activity against human cytomegalovirus via inhibition of viral replication. [138] Human Antiviral activity against Zika and Usutu in vitro (Vero cell line). [255] Human Coagulant potential of human milk, owing to the presence of tissue factor (TF)-exposing EVs, but not found in cow milk.

Species Findings
Ref.

Human
Protective effect against experimental-induced NEC in vitro (IEC-6 and FHs 74 Int cell lines) and in vivo (Sprague Dawley pups). [257] Human + Cow Attenuation of inflammatory cytokine expression and nuclear factor (NF)-κB activation in vitro (LPS-stimulated RAW 264.7). [258] Cow Promotion of osteogenesis via proliferation and differentiation of osteoblasts in vitro (Saos-2 cell line) and in vivo (Sprague Dawley rats). [259] Cow Improved small intestinal dysfunction in malnutrition C57BL/6J mouse model. [260] Cow Enhancement of curcumin cell uptake and permeability in an intestinal model in vitro (Caco-2 cell line). [261] Cow Osteoprotective effects in vivo (BALB/c and C57BL/6 mice), and decreased the RANKL/OPG ratio in vitro (MLO-Y4 cell line). [249] Cow Induction of phenotypical changes in hPAEC and NRCM cell lines. [109] Cow Modulation of gut microbiota composition, SCFA profiles, and enhancement of intestinal immune regulation by EVs in vitro (RAW 264.7 cell line) and in vivo (C57BL/6J mice). [225] Cow Differential improvements in DSS-induced colitis of two EV subsets via different mechanisms in vivo (C57BL/6J mice). [188] Cow Modulation of agricultural dust-induced lung inflammation by EVs in vitro (MH-S cell line) and in vivo (C57BL/6J mice). [232] Cow Demonstrated sonication effects on EV skeletal muscle biomarkers in vivo (Fischer 344 rats). [262] Cow Biocompatibility and potential use as a non-immunogenic delivery vehicle of EVs in vitro (RAW 264.7) and in vivo (ICR mice). [238] Cow EVs contain bioactive TGF-β in vitro (NIH/3T3 cell line), and EVs facilitate differentiation of naive T cells into pathogenic Th17 cells (ex vivo DBA/1J mice). [233] Exosome Human Protective effect of both raw and pasteurised exosomes against NEC in vivo (C57BL/6 mice) and ex vivo (neonatal mice intestinal organoids). [239] Human Demonstrated that miR-148a influenced the proliferation, morphology, and protein expression of transformed cells more so than normal cells in vitro (LS123 and CCD841 cell lines). The role of miR-148a was validated using a knockdown model in vitro (293T cell line). [189] Human Protection against H 2 O 2 -induced oxidative stress in NEC in vitro (IEC-6 cell line). [240] Human Showed uptake of exosomes, increased expression of miR-148a, and decreased expression of DNA-methyltransferase 1 in vitro (CRL-1831, K-562, and LIM1215 cell lines). [250] Human TGF-β2 influences epithelial-mesenchymal transition in vitro (MCF-7 and MCF 10A cell lines). [244] Human Inhibition of HIV-1 viral transfer to CD4+ T cells ex vivo (human MDC organoids). [235] Human The abundance and composition of exosomes vary due to lactation stage, maternal sensitisation, and lifestyle, which influence the regulation of the allergic outcome in the child. [247]

Human
The presence of MHC classes I and II, CD63, CD81, and CD86 on exosomes, inhibition of anti-CD3-induced cytokine production, and an increase in Foxp3 + CD4 + CD25 + T regulatory cells ex vivo (human PBMCs).

Species Findings
Ref.

Cow
The loading of miRNA (hsa-miR-148a-3p) as a nanocarrier in vitro (HepG2 and Caco-2 cell lines). [264] Cow Activation of immune cells ex vivo (human PBMCs) under inflammatory conditions. [265] Cow Restoration of small intestinal epithelial architecture and barrier function in malnourished C57BL/6J mice. [266] Cow Exosomes influence macrophage proliferation and protect against cisplatin-induced cytotoxicity in vitro (RAW 264.7 cell line). [236] Cow Exosomes have cytoprotective and anti-inflammatory activity in ulcerative colitis in vivo (Kindlin 2 −/− mice). [237] Cow Protective effects in vitro (IEC-6 cell line) against oxidative stress. [267] Cow Osteoporosis prevention in in vitro (MC3T3-E1 and RAW 264.7 cell lines) and in vivo (C57BL/6J mice) models. Additionally, the restoration of gut microbiota affected by osteoarthritis. [268] Cow Exosomes can be used as an siRNA delivery vehicle in vitro (A549 cell line) and have anti-tumour activity against lung tumour xenografts in vivo (athymic nude mice) and in vitro (MDA-MB-231, MCF7, A549, H1299, PANC-1, Mia PaCa-2, and A2780 cell lines). [227] Cow The use of exosomes as an oral delivery vehicle in xenografts, which enhanced gut absorption and retention involving neonatal Fc receptor in vivo (Balb/c mice, CT26 cells). [229] Cow Enhanced goblet cell activity, improved response against NEC in vivo (C57BL/6 mice), and increased mucin production in vitro (LS174T cell line). [242] Cow Bilberry anthocyanins encapsulated in exosomes were preferentially taken up by colonic cancer cells in vitro (HCT 116, HT-29, CCD-18Co cell lines), and therapeutic enhancement with encapsulated anthocyanins showed no significant differences in vivo (C57BL/6J mice). [228] Cow Depletion in dietary milk exosomes and their miRNA aggravates irritable bowel disease in vivo (Mdr1a −/− mice). [269] Cow Exosomes have a minimal effect on skeletal muscle biology in vivo (C57BL/5 mice), suggesting that other tissues may be the targets of exosomes. [245] Cow The use of paclitaxel encapsulated in exosomes as a drug delivery vehicle in vivo (athymic nude and C57BL/6 mice). [213] Cow Enhancement of skeletal muscle protein synthesis and anabolism in skeletal muscle cells independent of amino acids in vitro (C 2 C 12 myoblast). [246] Cow Resistance of exosomes to in vitro digestion and subsequent internalisation and trans-epithelial transport in vitro (Caco-2 cell line). [159] Cow The effects on exosomes of in vitro fermentation using three combinations of probiotic bacteria, uptake of these exosomes, and increased proliferation due to the upregulation of ERK1/2 and p38 in vitro (IEC-6 cell line). [201]

Cow
The use of encapsulated celastrol as a drug delivery vehicle, and anti-tumour activity against lung tumour xenografts in vivo (athymic nude mice, A549 and H1299 cell lines). [231]

Cow
The use of both encapsulated hydrophilic and lipophilic small molecules as a delivery vehicle, with tumour targetability, cross-species tolerance, and enhanced drug efficacy compared to free drugs in vivo (athymic nude mice) and in vitro (A549, H1299, MDA-MB-231, T47D, and Beas-2B cell lines). [212]

Cow
The uptake, transport kinetics, and presence of exosomal surface glycoproteins and inhibitors of endocytosis in vitro (Caco-2 and IEC-6 cell lines). [209] Cow + ASC + Coconut Promotion of bacterial growth and alteration of gene expression in vitro (Escherichia coli K-12 MG1655 and Lactobacillus plantarum WCFS1 cultures). [251] Cow + Mice+ Pig Inter-species and intra-species bioavailability and distribution of exosomes in vivo (Balb/c mice). [223] Cow + Yak Higher growth efficiency in vitro (IEC-6 cell line) under hypoxic conditions when supplemented with yak exosomes rather than cow milk-derived exosomes. [270] Buffalo Increased stability, solubility, and bioavailability of digested and undigested EV-encapsulated curcumin in vitro (Caco-2 cell line).

Species Findings
Ref.

Camel
Anticancer effects, via induction of apoptosis, inhibition of oxidative stress, reduced angiogenesis, and metastasis, in vivo (albino rats) and in vitro (MCF7 cell line). [252] Rat Rat milk-derived exosomes promote intestinal epithelial cell viability, enhance proliferation, and stimulate intestinal stem cell activity in vitro (IEC-18 cell line). [243] Pig Protective effect against deoxynivalenol (DON)-induced intestinal damage in vivo (Kunming mice) and in vitro (IPEC-J2 cell line). [271] Pig Protective effects of exosomes against LPS-induced effects in vivo (Kunming mice) and in vitro (IPEC-J2 cell line). [185] Pig Promotion of digestive tract development, alteration in the expression of proliferation-related genes in vivo (Kunming mice), and altered cell proliferation, proliferation-related gene expression, and miRNA concentration in vitro (IPEC-J2 cell line). [272] Pig Expression of miRNA during different lactation stages, and a higher uptake of colostrum-derived immune-related miRNA in vivo (piglets). [181] Pig + Cow Both cow and pig milk exosomes alter serum miRNAs in vivo (piglets), and exosomal miRNA is taken up in vitro (IPEC-J2 cell line). [162]

Microvesicle/Nanovesicle
Cow Suitability of nanovesicles and encapsulated siRNA as a therapeutic delivery vehicle in vivo (zebrafish) and ex vivo (C57BL/6 splenocytes). [184] Cow Demonstrated successful uptake of PKH67-labelled microvesicles in vitro (RAW 264.7 cell line). [203] The terminology used is based on the reference cited, and this division reflects older thinking and is a "pool" of EVs that are responsible for the effects (ASC, adipose-derived stem cell; DSS, dextran sulphate sodium; hPAEC: human pulmonary artery endothelial cell; MDC, monocyte-derived dendritic cell; MSC, mesenchymal stem cell; NEC, necrotising enterocolitis; NRCM, neonatal rate cardiac myocyte; PBMC, peripheral blood mononuclear cell; SCFA, short-chain fatty acid; LPS, lipopolysaccharide).

Conclusions
Not all types of milk provide the same nutritional value for inter-species consumption. Species-dependent differences are evident in the macromolecule composition (fat, sugars, etc.), vitamin and mineral content, and how it is digested after consumption [30, 273,274]. Furthermore, milk has differences in its molecular composition and conservation of function that influence its specific biological value depending on the species of origin. It is reasonable to propose that EVs in milk from different species may have a differing composition that may affect their nutritional value from an EV-mediated view. Infant formula derived from cow's milk is still the largest source of non-human infant foods worldwide, but there are areas of the world where cow's milk is not traditionally consumed.
What nutritional effects that EVs from non-traditional and minority milk may have is poorly understood and represents a substantial gap in our knowledge. We have provided a brief summary of nutritional aspects of mammalian milk and summarised the research on milk-derived EVs of human and common mammalian livestock. We have also discussed research around therapeutic attributes, cargoes of milk-derived EVs, and techniques for working with them.
However, isolating, characterising, and assigning biological effects to milk-derived EVs are challenging due to the highly complex nature of milk as a biofluid. Careful consideration and reporting of standardised methods, i.e., MISEV guidelines, are critical to studies seeking to identify true and meaningful biological effects. The stability and bioavailability of nutrients, combined with their subtle effects (compared to pharmaceuticals), mean that any research on milk EVs needs to be carefully designed to correctly assign their functions in supporting human health.