Skip to main content
NCBI home page
The Veterinary Quarterly logoLink to The Veterinary Quarterly
. 2026 May 1;46(1):2665485. doi: 10.1080/01652176.2026.2665485

Milk extracellular vesicles reveal insulin resistance status in peripartum dairy cows

Madhusha Prasadani a, Hanno Jaakson a, Priit Karis a, Subhashini Muhandiram a, Merike Henno a, Sirje Kuusik a, Katri Ling a, Harinda Rajapaksha b, Meelis Ots a, Aneta Andronowska c, Suranga Pradeep Kodithuwakku a,d, Vallo Volke e, Alireza Fazeli a,e,f,*
PMCID: PMC13137739  PMID: 42065931

Abstract

The transition from gestation to lactation is a critical period that leads to insulin resistance (IR) in dairy cows. IR detection requires invasive sampling and laborious methods; therefore, novel and non-invasive approaches are essential. Extracellular vesicles (EVs) are nano-sized particles released by cells and are linked to disease pathophysiology. Hence, the present study aimed to explore the potential application of cow milk EVs (MEVs) in detecting IR in dairy cows. An intravenous glucose tolerance test (IVGTT) was performed (n = 12), and the cows were classified as low, moderate, or high IR based on prepartum area under the curve insulin values. MEVs were enriched using double-size exclusion chromatography and physico-chemically characterized. The results revealed that the particle diameter of MEVs was significantly higher (p ≤ 0.05) in the high-IR group compared to other groups. Out of 26 fatty acids (FAs) detected in MEVs, three long-chain FAs differed significantly in abundance (p ≤ 0.05) in high-IR group. Proteomic analysis revealed a correlation of inflammation and the IR-related proteins; Annexin A1 and Rab proteins, to IR severity. Overall, this study identified promising predictors of IR in cows, supporting the development of non-invasive biomarkers for early detection, facilitating improved herd and disease management, and reducing economic losses.

Keywords: Milk, extracellular vesicles, insulin resistance, long-chain fatty acids, diagnosis and monitoring, proteome, particle diameter, dairy cows

1. Introduction

Dairy cows are at substantial risk of metabolic and infectious diseases during the peripartum period (Rilanto et al. 2020). This time, the demand for energy and nutrients, especially glucose, increases for the gravid uterus and milk synthesis, leading to a negative energy balance (NEB), particularly in high-producing cows (De Koster 2013). To compensate for the increasing energy demand with the reduced intake level, fatty acids (FAs) stored in adipose tissues are mobilized by increased lipolysis, releasing non-esterified fatty acids (NEFAs) into the bloodstream. Elevated levels of NEFAs impair insulin receptor function and downstream signaling pathways in target tissues such as muscle, liver, and adipose tissues, which then promote inflammation, triggering insulin resistance (IR) in these tissues and reducing the ability of tissues to take up glucose from the bloodstream (Qiao et al. 2024). Moreover, the endocrine system responds to IR by altering the secretion of hormones such as cortisol and growth hormone, which can further impair the insulin sensitivity of the tissues, including mammary glands (Chalmeh et al. 2015).

Detecting IR in cows requires a multifaceted approach that combines clinical observations with various biochemical and physiological tests. Blood metabolite measurements, tolerance tests, and clamp tests are possible detection methods, while the intravenous glucose tolerance test (IVGTT) is the traditional confirmatory method (De Koster et al. 2016). However, IVGTT cannot distinguish insulin-dependent glucose consumption, making it difficult to assess tissue insulin sensitivity, particularly in pregnant animals that consume about 80% of glucose independent of insulin action (De Koster et al. 2016). Moreover, this method is time-consuming, expensive, and requires highly skilled personnel to conduct. Thus, if a non-invasive sampling-based and more sensitive detection assay can be developed with new biomarkers, it will benefit the dairy industry and avoid huge losses entangled with metabolic disorders and reproductive failures.

Extracellular vesicles (EVs) are lipid bi-membranous, nano-sized particles secreted by every type of cell inside the body and are highly heterogeneous (Prasadani et al. 2024). They harbor a broad spectrum of biomolecules, such as different DNA and RNA species, proteins, lipids, and metabolites, and the composition of EV molecular cargo depends on the secretory cell type. EVs and their cargoes are considered emerging candidates for biomarkers for diagnostic and prognostic purposes under different physiological statuses and disease conditions (Mousavi et al. 2024; Premathilaka et al. 2025).

Biofluid-derived EV characteristics and disparities in cargo have been widely employed in evaluating potential disease biomarkers, including cancer (Li et al. 2022), infectious diseases (Lyu et al. 2019), and reproductive disparities (Lv and Lu 2019). In cattle, there are reports of the potential of using microRNA in milk EVs (MEVs) to detect leukemia virus (Nakanishi et al. 2022), plasma-derived EV proteins to detect uterine infections (Almughlliq et al. 2018), uterine fluid EV proteins to identify bovine endometritis (Piibor et al. 2024), and determine subclinical mastitis using miRNA cargo of MEVs (De Juano et al. 2022).

FAs are notable components of different milk constituents, as their composition changes depending on numerous factors, such as feed type, disease and lactation stage (Jiang and Sun 2022). Milk FAs are of two different origins, from de novo synthesis in the mammary gland and lipids present in plasma, which initially originate from the adipose tissue, rumen microbes, and feeds (Chilliard et al. 2020). Previous studies have reported that long-chain fatty acids (LCFAs) can be used to detect metabolic problems such as hyperketonemia in dairy cows (Jorjong et al. 2015). IR status leads to higher levels of lipolysis, leading to LCFAs release into the bloodstream, including NEFA. The MEVs can carry LCFAs as early bioindicators reflecting the metabolic status of an animal (Qiao et al. 2024). In addition, there is evidence that proteomic profiles are changed in the adipose tissues, which circulate EVs based on IR (Zachut 2015). During NEB, shifts in EV protein composition can indicate metabolic stress, disrupted lipid metabolism, and impaired immune function (Zachut 2015).

The use of MEV is a novel approach for identifying IR, as milk can be collected non-invasively, and the sampling procedure is not as complex as that in conventional blood-based IVGTT assays. In addition, this will reduce the stress-induced metabolic fluctuations associated with invasive sampling. EVs carry tissue-specific molecular cargo, reflecting both local and systemic metabolic changes, allowing earlier detection of IR compared to conventional blood markers. Moreover, the combined FAs and proteomic information provides a more integrated and specific view of metabolic status, while repeated milk sampling enables continuous monitoring without invasive procedures.

The purity of the resulting EV population is crucial for downstream applications such as biomarker discovery, as specific biomarker molecules can be obscured by contaminants. Therefore, in the current investigation, a protocol for MEV enrichment with high purity was reported. Then, it was hypothesized that variations in MEV physicochemical characteristics, such as MEV particle diameter, LCFA profile and proteome, are correlated with IR severity. The overall objective of the study was to identify early-stage non-invasive biomarkers for IR status using cow MEVs.

2. Materials and methods

2.1. Selection of animals for the study

All the animal experiments performed complied with the European Council Directive and the Estonian Animal Protection Act, and the study was approved by the Committee for Conducting Animal Experiments at the Estonian Ministry of Regional Affairs and Agriculture (License no. 232, 19.12.2022). IVGTT experiments were performed, and milk samples were collected from November 2022 to February 2023. The study was carried out using clinically healthy multiparous Estonian Holstein cows (n = 12) with a mean body condition score of 2.9 ± 0.1, (mean ± SEM) body weight of 761 kg ± 19.5, a relative breeding value for milk production of 108 ± 2.2 and a parity of 2.7 ± 0.3 from the Märja experimental farm, Estonian University of Life Sciences, Estonia, with a herd size of about 130 cows and an average milk yield of about 9700 kg per cow per year. The cows were moved to an experimental barn a week before the expected calving date.

2.2. Feeding and feed analyses

During the experimental period, the cows were fed individually twice a day at 06:00 and 15:00. The forage and concentrates were fed separately. The forage consisted of grass and clover silage and hay, while the concentrates consisted of rapeseed cake, barley meal, and maize meal. The amount of rapeseed cake during the dry period and the composition of the concentrates during lactation were adjusted based on the nutrient content of the forage to ensure adequate metabolized energy (ME) and metabolized protein (MP) contents in the dry matter (DM) (9.2 MJ/kg and 77 g/kg during the dry period and 11 MJ/kg and 95 g/kg during lactation, respectively). After calving, the amount of concentrate consumed depended on the dry matter intake (DMI). On day 1, cows received 2 kg of concentrate both in the morning and in the evening (total 4 kg). On each subsequent day, the amount of concentrate was increased by 0.5 kg until the day 16th DIM being approximately 10.1 kg in DM (6 kg fed in the morning and 5.5 kg in the evening). Each feeding began with weighing the forage orts. Thereafter, the cows were given mineral feed mixed with a small amount of concentrates. Next, the full amount of concentrates was added to the feeding trough. Finally, a new portion of forage was weighed and placed on the feeding trough while providing access to ad libitum water. To compose rations, feeds were sampled fortnightly for roughages and on a batch basis for other feeds, and DM and chemical compositions were analyzed using AOAC 2021 methods (AOAC 2019).

2.3. Milk sampling and analyses

Cows were bucket milked twice daily at 4.00 am and 2.00 pm. Milk yields were registered (by weighing), and milk samples were collected in the morning and evening milking with an average of 4.7 ± 0.4 days in milk. Fifty milliliters of milk sample from each animal was transported to the laboratory on ice, pre-processed, and stored at -80 °C until further analysis.

2.4. Intra-venous glucose tolerance test (IVGTT) blood sampling and analyses

IVGTT was conducted on day 10 ± 1.1 before calving at approximately 10:00 am, as described in previous reports by the same researchers (Jaakson et al. 2018). Blood samples were collected from the catheter at time points: −5 (basal), 5, 10, 20, 30, 40, 50, and 60 min relative to the start of infusion into the vacuum tubes with Li-heparin (Vacutest, Arzergrande, Italy). The samples were kept and transported under cool conditions until plasma separation by centrifugation (4000 × g, 15 min), aliquoted, and stored at −80 °C until they were analyzed for glucose, insulin, NEFA, and ß-hydroxybutyrate (BHB) basal concentrations and for glucose and insulin concentrations at IVGTT time points.

The insulin concentration in the blood was analyzed by solid-phase bovine sandwich ELISA (catalog number EB13RB; Thermo Fischer Scientific, CA) with the detection limit of 0.025 ng/ml on a microplate reader (Bio Tek Epoch 2, Agilent, CA). The results were calculated using Bio Tek Gen5 software (Agilent, Santa Clara, CA). The inter-assay coefficient of variant (CV) for the low and high control samples was 11.8 and 12.2%, respectively; the intra-assay CV was 10.0 and 4.7%, respectively. Glucose was analyzed by enzymatic-colorimetric method (cat. no. GL8318, Randox Laboratories Ltd., Crumlin, UK), using an automatic analyzer (Randox Laboratories Ltd, Crumlin, UK). For glucose, NEFA and BHB, intra-assay CV were determined by analyzing plasma in 20 replicates in 3 duplicates; inter-assay CV were calculated based on 3 duplicates. The interassay CVs were 4.8, 8.3, and 6.1%, and the intra-assay CVs were 4.9, 6.1, and 5.7% for glucose, FA, and BHB, respectively. The animals were classified into low-, moderate-, and high-IR groups based on the prepartum IVGTT area under the curve (AUC) for insulin values calculated as described elsewhere (Jaakson et al. 2018).

2.5. Enrichment of MEVs using double-size exclusion chromatography (SEC)

Modifications were made to a previously established SEC-based protocol to enrich MEVs with higher purity (Kaminska et al. 2023). Samples from the bulk milk tank from the same farm were used in the initial MEV enrichment protocol optimization and MEV storage trials. Several pre-processing steps were employed to facilitate fewer contaminants in the final MEV population. Ten ml of milk aliquots were centrifuged immediately after transporting to the laboratory at 3400 × g for 30 min at 4 °C to remove milk fat globules, somatic cells, and other cell debris. The cream layer collected on the top was removed, and the skimmed milk was separated. Casein was precipitated using 6 N acetic acid (acetic acid glacial 100%, MERCK, Darmstadt, Germany) while adjusting the pH of the solution to pH 4.6, followed by centrifugation at 5000 × g for 30 min at 4 °C. The supernatant was separated from the casein precipitate and filtered with a series of filtration steps with 1, 0.45, and 0.22 µm syringe filters, and the resultant whey was concentrated to 500 μL using Amicon® Ultra-2ml centrifugal filters (10 kDa) (Merck Millipore, Burlington, MA).

Then, concentrated whey was introduced into the first SEC column, and 20 fractions of 500 µL each were collected as described elsewhere (Piibor et al. 2023). The particle and soluble protein concentrations of each fraction were analyzed using ZetaView PMX-110 V3.0 (Particle Metrix GmbH, Inning a. Ammersee, Germany) coupled with ZetaView NTA software (8.04.02 SP2) and the Bradford assay, respectively. Based on the results from the first SEC column, the fractions with the highest particle concentration (fractions 5–8) and least soluble protein were pooled together, concentrated, and introduced into a second SEC column, and 20 fractions were collected and characterized.

2.6. Characterization of MEVs

2.6.1. Particle concentration and size distribution analysis with nanoparticle tracking analysis (NTA)

The particle concentration and mean particle size in each enriched fraction were determined using ZetaView PMX-110 V3.0 instrument (Particle Metrix GmbH, Inning a. Ammersee, Germany) coupled with ZetaView NTA software (8.04.02 SP2) for data analysis following the manufacturer's guidelines (Dissanayake et al. 2021). The samples were measured in scatter mode using camera sensitivity at 85, shutter speed at 70, frame rate at 30 frames per second (fps), number of cycles 3, and number of frames at 11.

2.6.2. Protein measurements using Bradford assay

The soluble protein concentrations were measured using the standard Bradford assay for each SEC fraction, and the absorbance was measured spectrophotometrically (Ledetect 96 Microplate Reader; Biomed Dr. Wieser GmbH, Salzburg, Austria) at 620 nm, as mentioned elsewhere (Piibor et al. 2023).

2.6.3. Western blotting for MEV marker analysis

Western blot analysis was performed on the enriched MEVs from double SEC as described previously (Piibor et al. 2023). Briefly, the particles were lysed using RIPA + protease inhibitor cocktail on ice, quantified and separated using 12% SDS–PAGE, and transferred to PVDF membrane followed by blocking in 5% non-fat milk for 1 h at room temperature. The membrane was incubated with primary anti-CD63 (MCA2042GA, 1:1000, Bio-Rad, CA), anti-CD81 (SC-166029, 1:1000, Santa Cruz Biotechnology, CA), anti-MFGE8 (SAB1408603, 1:1000, Sigma-Aldrich, MA), anti-rabbit TSG101 (AV38773-100UL, 1:5000, Sigma-Aldrich, MA), anti-calnexin (NB100-1965, 1:5000, Novus Biologicals, Cambridge, UK) overnight at 4 °C, then washed with PBST and followed by incubation with horseradish peroxidase-conjugated goat anti-mouse secondary antibody (G21040, 1:10,000, Thermo Fisher Scientific, MA) or horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (G21234, 1:10,000, Thermo Fisher Scientific, MA) for 1 h. Bands were detected by a commercial enhanced chemiluminescence kit (Amersham Pharmacia Biotech, Amersham, UK) and documented using ImageQuant RT-ECL machine coupled with IQuanta-Capture software (GE Healthcare Bio-Sciences AB, Uppsala, Sweden).

2.6.4. Transmission electron microscopy (TEM) of MEVs

TEM of MEVs was performed as described previously (Piibor et al. 2023). In brief, formvar/carbon-coated 200 mesh grids (Agar Scientific, Stansted, UK) were placed on 10 µL of MEV solution for 20 min, incubated with 2% uranyl acetate (Polysciences, PA) for 2 min in the dark and visualized using JEM 1400 TEM (JEOL Ltd. Tokyo, Japan) at 80 kV coupled with a numeric camera (Morada TEM CCD camera, Olympus, Hamburg, Germany).

2.7. Fatty acids analysis of MEVs

Extraction and derivatization of FAs from MEVs were performed in one step according to the modified method of Sukhija Palmquist, 1986 (Pritam et al. 1988). Two ml of each MEV sample was lyophilized (Christ Freeze dryer: Beta 1-8 LSC plus, Osterode am Harz, Germany) and transferred to a threaded test tube, and 0.5 ml of hexane was added, followed by 0.75 ml of freshly prepared acetyl chloride solution (MERCK, Darmstadt, Germany). The tubes were tightly closed and mixed for 1 min, followed by heating at 100 °C for 1.5 h. Then the solutions were allowed to cool to room temperature, and 1.25 ml of 6% potassium carbonate was added for neutralization. The solution was mixed for 0.5 min at medium speed and centrifuged for 5 min at 1500 × g. Then, the upper layer was transferred to a new test tube, and anhydrous Na2SO4 was introduced. Finally, the sample was centrifuged at 4000 × g for 5 min to obtain a clear hexane layer with FA methyl esters, and was quantified by gas chromatography with a jet split of 1:20.

2.8. Proteomic profile analysis of MEVs

Total proteins from the MEV samples were precipitated using trichloroacetic acid and sodium deoxycholate with a 30 min incubation (Piibor et al. 2023). The pellets were washed once with acetone and once with methanol and then air dried. Protein concentrations were estimated by Micro BCA assay (Thermo Fisher Scientific, Waltham, MA). The precipitated proteins were then solubilized in 7 M urea, 2 M thiourea, 100 mM ammonium bicarbonate buffer, and 20 mM methylamine, reduced with 10 mM dithiothreitol for 30 min at 30 °C, and alkylated with 30 mM chloroacetamide in the dark for 30 min. Pre-digestion with 1:100 (enzyme-to-protein ratio) Lys-C (Fujifilm Wako Pure Chemical, Osaka, Japan) was carried out for 2 h at 25 °C. Digested samples were acidified with TFA to 1.0% and desalted on in-house-made C18 (3 M) solid-phase extraction tips. The purified peptides were reconstituted in 0.5% TFA for nano-LC/MS/MS analysis as described in Piibor et al. (2023).

The raw files of MS/MS output were processed with the MaxQuant software package (version 2.2.0.0) with the previously mentioned parameters (Piibor et al. 2023). A search was performed against the UniProt (www.uniprot.org) Bos taurus proteome database (downloaded: 2024 August) using the tryptic digestion rule (cleavages after lysine and arginine without proline restriction). All other parameters were used as described (Piibor et al. 2023). The biological significance of the differentially expressed MEV proteins was examined by performing gene ontology (GO) analysis using PANTHER GO Slim analysis. An in-depth analysis of the related pathways was conducted to reveal the relevance of the detected MEV proteins to the different biological processes, as described elsewhere (Mi et al. 2019).

2.9. Statistical analysis

GraphPad Prism version 8.0.1 for Windows (GraphPad Software, Boston, MA) was used for statistical analysis, and p ≤ 0.05 was considered statistically significant. Blood insulin and glucose concentrations during the IVGTT and comparison of MEV diameter and particle concentration among different IR groups were performed with two-way ANOVA with Tukey's multiple comparisons. Spearman correlation analysis was carried out to determine the relationship between the IVGTT parameters and NTA results of MEVs, AUC insulin, and protein concentrations. The comparison of particle concentrations in the MEV characterization and FA content of MEVs among different IR groups was analyzed using ordinary one-way ANOVA with Tukey's multiple comparisons. The processed LC–MS/MS data with MaxQuant were further subjected to analysis with the LFQ-analyst platform to analyze and visualize proteomic differences between the MEVs from different IR animals. The data were imputed using the Minprob imputation method, and the cut-off values used were p < 0.05 and log2 = 1. All the data are presented as the mean plus or minus the standard error of the mean (mean ± SEM).

2.10. Experimental design

A summary of the experimental design is depicted in Figure 1.

Figure 1.

A three-panel diagram shows milk pre-processing, milk extracellular vesicle enrichment, and characterization methods. The three-panel diagram shows a workflow for milk derived extracellular vesicle isolation and characterization to use them as indicators of insulin resistance in dairy cows. Panel A, titled Pre-processing of milk, illustrates raw milk undergoing fat removal via centrifugation, then casein removal using 6N Acetic Acid, followed by filtration to yield whey. Panel B, titled Optimizing Protocal for MEV isolation, depicts part of filtered whey undergoing double size exclusion chromatography using SEC 1 and SEC 2 to enrich EV fractions and other parts storage at -80 °C for 1 month and 3 months, beforeundergoing double size exclusion chromatography using SEC 1 and SEC 2 to enrich EV fractions. Characterization of EVs involves Nanoparticle Tracking Analysis, Bradford assay, Western Blot, and Transmission Electron Microscopy. Panel C, titled Comparison of physiochemical characteristics of MEVs, shows grouping animals based on GTT, milk collection, MEV enrichment, and MEV characterization. Characterization includes size and concentration by Nanoparticle tracking analysis, total protein content by Bradford assay, EV marker analysis by Western Blot, morphology by Transmission Electron Microscopy, fatty acids profile by Gas chromatography analysis, and protein profile by Mass spectrometry analysis.

Open in a new tab

Summary of the experimental design. (A) Pre-processing of milk up to whey (n = 3). (B) Whey storage experiment and enrichment of MEVs using double SEC and characterization (n = 3). (C) Conducting intravenous glucose tolerance test and grouping animals into high, moderate, and low IR groups, MEV isolation from the selected animals and physiochemical characterization of enriched MEVs (n = 4). (GTT – glucose tolerance test, SEC – size exclusion chromatography, MEV – milk-derived extracellular vesicles, EV –extracellular vesicles).

The experiments were designed to test the validity of the following questions/hypotheses:

2.10.1. Do experimental cows have different IR statuses?

Twelve cows involved in the study were classified into high, moderate, and low IR groups based on the AUC for blood insulin values of the prepartum IVGTT test (n = 4 per group).

2.10.2. Do double-SEC improve the purity of enriched MEVs?

Experiments were designed to test whether double-SEC improves the purity of enriched MEVs. Ten milliliters of raw bulk milk was used as the starting material, and the MEVs were enriched. The eluted fractions were collected, and the particle concentrations and diameters were measured using ZetaView NTA, while the Bradford assay was employed to measure the protein concentrations. The presence of EV-specific protein markers in MEVs was detected using western blot analysis, and morphology was observed using TEM (n = 3).

2.10.3. Do short-term whey storage affect MEV characteristics?

To test the effect of whey (solution after the removal of milk fat globules, casein, somatic cells and other cell debris) storage on MEV characteristics, 10 ml of raw bulk milk was used as the starting material and subjected to pre-processing steps of centrifugation at 3400 × g, precipitation using 6 N acetic acid and filtration with a series of syringe filters with 1, 0.45, and 0.22 µm size. The resultant whey was concentrated to 500 μL and stored at −80 °C for one month and 3 months. After the relevant storage period, the whey samples were thawed on ice, and MEVs were isolated using the optimized double SEC protocol. The enriched MEVs were characterized and compared with the MEVs isolated from fresh whey. The three experimental groups involved in the study were,

  • Group 1 – MEVs isolated from fresh whey samples (n = 3).

  • Group 2 – MEVs isolated from whey samples stored at -80 °C for one month (n = 3).

  • Group 3 – MEVs isolated from whey samples stored at -80 °C for three months (n = 3).

2.10.4. Is IR severity correlated with the MEV physicochemical characteristics?

To test the hypothesis that IR severity is correlated with MEV characteristics, milk samples were collected (morning and afternoon milking separately) from the animals with different IR statuses, and MEVs were enriched using the aforementioned double-SEC protocol. The MEV parameters were compared with the IVGTT parameters using Spearman correlation analysis (n = 4 per group). FAs from MEVs were also extracted and measured using Gas chromatography to compare the FA profiles of cows with different IR stages of high, moderate, and low (n = 4 per group). Three technical replicates were used for each FA measurement. MEV proteins were extracted and measured using LC–MS/MS to compare the protein profiles of cows with different IR stages of high, moderate, and low (n = 4 per group).

3. Results

3.1. Classification of animals based on prepartum IVGTT results

Animals were grouped into different IR status based on the AUC of insulin during pre-partum IVGTT. As per previous reports, the extent of the insulin response during the IVGTT shortly before calving is the best measure of IR after calving (Jaakson et al. 2018). Therefore, based on the AUC analysis for insulin from prepartum IVGTTs, the animals were classified into low-IR (310.56–498.14 μIU/ml × min), moderate-IR (637.25–1304.44 μIU/ml × min), and high-IR (2 183.51–11,256.35 μIU/ml × min) groups. The basal glucose concentrations were significantly different (p ≤ 0.05) in moderate and low-IR groups (124.95 mg/dl vs 99.0 mg/dl), and the maximum glucose level was recorded after 5 min of infusion in all 3 groups (Figure 2A). The basal blood insulin levels were not significantly different between the IR groups (53.9–16.55 μIU/ml). The maximum insulin concentration was reached 5–10 min after glucose infusion (Figure 2B).

Figure 2.

A two panel line graph shows glucose and insulin changes over time for high, moderate, and low insulin resistance groups. Two-panel line graph shows glucose and insulin changes over time during IVGTT for high, moderate, and low insulin resistance groups. Panel A, Glucose mg/dL: Horizontal axis is Time (minutes) at Basal, 5, 10, 20, 30, 40, 50, 60. Vertical axis 50-250. High IR (green circles) starts at 120, peaks at 240 at 5 min, declines to 140 at 60 min. Moderate IR (red squares) starts at 120, peaks at 240 at 5 min, declines to 170 at 60 min. Low IR (blue triangles) starts at 100, peaks at 230 at 5 min, declines to 170 at 60 min. Asterisk indicates significance at 5 min for low IR. Panel B, Insulin microIU/mL: Horizontal axis is Time (minutes) at Basal, 5, 10, 20, 30, 40, 50, 60. Vertical axis 0-400. High IR (green circles) starts at 50, peaks at 270 at 20 min, declines to 90 at 60 min. Moderate IR (red squares) starts at 20, peaks at 80 at 10 min, declines to 60 at 60 min. Low IR (blue triangles) starts at 10, peaks at 40 at 10 min, declines to 20 at 60 min.

Open in a new tab

Basal concentrations of glucose, insulin, and their changes during the intravenous glucose tolerance test. (A) Blood glucose changes during the prepartum intravenous glucose tolerance test (IVGTT). (B) Blood insulin changes during the prepartum intravenous glucose tolerance test (IVGTT) among different insulin resistance (IR) groups (high-IR, moderate-IR, and low-IR) (mean ± SEM, n = 4) (*p < 0.05).

3.2. Confirmation of MEV enrichment and characterization

The double SEC protocol successfully enriched the EVs from whey samples with higher purity, as shown by TEM (Figure 3C), contaminant soluble protein analysis (Figure 3A), and marker analysis (Figure 3D). NTA and protein measurements revealed that fractions 5–8 from the 1st SEC column contained the highest number of particles and the least soluble protein contaminants (Supplementary Figure 1). Hence, they were pooled together and introduced into the 2nd SEC column. Fractions 6–8 had the highest particle concentrations with minimal soluble protein contamination according to NTA and Bradford assays (Figure 3A). The size distribution results from NTA revealed that the mean diameter of the particles was around 150–200 nm (Figure 3B), and TEM images confirmed the presence of EV particles with a cup shape (Figure 3C). EV marker analysis with western blotting further confirmed the enrichment of EVs in the fractions (Figure 3D).

Figure 3.

Four panel figure: particle and protein concentration, particle size distribution, electron micrographs, and western blots. The four panel figure shows characterization results of milk extracellular vesicles. Panel A is number of particles and protein concentrations in each size exclusion chromatography fraction. Panel A is a bar graph with two vertical axes: particle concentration (0-5x10^11 particles/mL) and protein concentration (0.0-1.0 mg/mL). The horizontal axis is fraction number (1-20). Both particle and protein concentrations peak at fraction 7, with particle concentration at approximately 4x10^11 and protein concentration at approximately 0.8 mg/mL. Other fractions show low concentrations. Panel B is a line graph showing particle concentration (0-3x10^10 particles/mL) versus diameter (0-400 nm). It shows a single peak at approximately 150-200 nm, with concentration reaching about 1.8x10^10. Panel C is a transmission electron micrography images showing milk extracellular vesicles with cup shapes.. An inset shows a magnified view of several vesicles, revealing their double membrane structure. A scale bar indicates 200 nm. Panel D shows five western blot strips. Each strip has two lanes, labeled MEV and Whey. From top to bottom, the strips are labeled approximately 45 kDa MFGE8, approximately 45 kDa TSG 101, approximately 25 kDa CD 63, approximately 20 kDa CD 81, and approximately 75 kDa Calnexin. Dark bands are visible in the MEV lane for MFGE8, TSG 101, CD 63, and CD 81. No bands are visible in the Whey lane for these proteins. Dark band is only visible in Calnexin in whey samples.

Open in a new tab

Confirmation and characterization of the presence of EVs in double size exclusion chromatography (SEC)-enriched fractions. (A) Number of particles and protein content in each fraction number (no) 1–20 enriched from the second SEC column. (B) Size distribution of the enriched particles. (C) Transmission electron microscope images of enriched EVs. (D) Protein analysis of the EV fractions and whey using Western blotting. The transmembrane proteins CD63, CD81, and MFGE8 and the cytosolic protein TSG101 confirmed the presence of EVs, while the absence of the negative marker calnexin revealed the enrichment and purity (mean ± SEM, n = 3) (MEV- milk extracellular vesicles).

3.3. Short-term whey storage does not affect the MEV concentrations

The enrichment of MEVs from the whey samples stored for one month and three months was successfully achieved with the optimized dSEC protocol. The particle concentrations at the two storage time points (one month and three months) were not significantly different when compared to freshly enriched MEVs (p ≤ 0.05, Figure 4A). The MEV particle diameter of the three conditions (fresh, one, and three months) remained in the range of 100–200 nm (Figure 4B). TEM images (Figure 4C–E) and EV markers (Figure 4F) confirmed the presence of EVs in enriched fractions under all three conditions.

Figure 4.

6 panel figure: milk extracellular vesicle characterization results after one month and three months storage of whey. The 6 panel figure shows milk extracellular vesicle characterization results after one month and three months storage of whey. Panel A is a bar graph showing the number of particles per milliliter for fresh, one month, and three months storage time. The number of particles is highest for three months storage, followed by fresh, then one month storage. Panel B is a line graph showing the number of particles per milliliter versus diameter in nanometers. Fresh whey shows the highest peak around 100 nanometers, followed by one month of storage, then three months of storage. Panels C, D, and E are transmission electron microscope images of EV from fresh whey, one month stored whey, and three months stored whey, respectively, each with an inset showing a magnified view. Panel F is a western blot showing protein analysis of MEV for MFGE8, TSG 101, CD 63, CD 81, and Calnexin across six lanes. Lanes 1 to 3 show EV from fresh, one month, and three months stored whey. Lanes 4 to 6 show fresh, one month, and three months stored whey.

Open in a new tab

Confirmation and characterization of milk extracellular vesicles (MEVs) in double size exclusion chromatography (SEC)-enriched fractions of fresh, 1-month and 3-month-stored whey. (A) Total number of particles in MEV fractions enriched from fresh whey, 1-month and 3-month-stored whey samples. (B) Size distribution of the enriched particles. (C) Transmission electron microscope images of enriched MEVs from fresh whey. (D) One-month-stored whey. (E) Three-month-stored whey. (F) Protein analysis of the MEV samples and whey of different storage conditions using Western blotting. The transmembrane proteins CD63, CD81, and MFGE8 and the cytosolic protein TSG101 confirmed the presence of EVs, while the absence of the negative marker calnexin revealed the enrichment and purity (mean ± SEM, n = 3).

3.4. Associations between the severity of IR with EV diameter

The morning MEV particle diameter was significantly correlated (p ≤ 0.05) with the maximum increase values (r = 0.62), the AUC for insulin (r = 0.79), and the half-time (r = 0.64) of insulin (Table 1, Figure 5). Moreover, the afternoon MEV concentrations were significantly correlated (p ≤ 0.05) with NEFA (r = 0.73) and BHB (r = 0.58) concentrations.

Table 1.

Spearman correlation of intravenous glucose tolerance test (IVGTT) parameter values with MEV particle concentration and MEV particle diameter of morning and afternoon milk.

  NTA results
  Particle concentration
 Particle size
GTT characteristics Morning milk Afternoon milk Morning milk Afternoon milk
Glucose
BASa (mg/dL) 0.22 −0.41 −0.03 −0.01
Maxb (mg/dL) −0.10 −0.42 −0.20 −0.15
AUCc (mg/dL/min) −0.08 −0.35 −0.10 −0.12
CRd (%/min) −0.28 0.28 0.47 0.47
Insulin
BASa (µIU/mL) −0.59 0.10 0.26 0.04
MAXb (µIU/mL) −0.55 0.37 0.62* 0.40
AUCc (μIU/ml × min) −0.53 0.25 0.79** 0.52
T(1/2)e (min) −0.30 0.33 0.64* 0.46
NEFA        
BAS (mmol/L) −0.26 0.73** 0.34 0.29
BHB        
BAS (mmol/L) 0.04 0.58* 0.27 0.34
Open in a new tab

n = 4. Bold text indicates significantly different values *p < 0.05, **p < 0.01. GTT – glucose tolerance test, NTA – nanoparticle tracking analysis, NEFA – nonesterified fatty acids, BHB – ß-hydroxybutyrate.

a

BAS: basal concentration.

b

Max: maximum increase relative to BAS.

c

AUC: area under the curve.

d

CR: clearance rate.

e

T(1/2): half-time of insulin.

Figure 5.

Correlation of particle diameter vs AUC insulin. High, Moderate, and Low IR groups are plotted. r = 0.79. The scatter plot displays particle diameter in nanometers on the vertical axis, ranging from 140 to 180, against AUC insulin in microIU per milliliter times minute on the horizontal axis, ranging from 0 to 12000. Three groups are plotted: High IR represented by circles, Moderate IR by squares, and Low IR by triangles. The High IR group shows data points scattered between approximately 2000 and 11000 on the horizontal axis, and 165 to 175 on the vertical axis. The Moderate IR group points are clustered between 0 and 2000 on the horizontal axis, and 150 to 160 on the vertical axis. The Low IR group points are also clustered between 0 and 2000 on the horizontal axis, and 148 to 155 on the vertical axis. A dashed trend line indicates a general positive correlation. The correlation coefficient r equals 0.79 is printed on the right side of the plot.

Open in a new tab

Correlation of the area under the curve (AUC) of insulin values and the particle diameter of morning milk extracellular vesicles (MEVs) derived from animals with different insulin resistance (IR) status (n = 4).

The MEV particle diameter of both morning and afternoon MEVs was significantly higher (p ≤ 0.05) in the high-IR group compared to the low-IR groups (Figure 6A), while particle concentrations between IR groups were not significantly different (p ≤ 0.05) (supplementary figure 2). The size distribution of both morning and afternoon MEV was shifted toward the right, indicating an increase in diameter (Figure 6B and C).

Figure 6.

Three panel graph: milk extracellular vesicle particle diameter and size distributions across insulin resistance groups. The three panel graph shows milk extracellular vesicle particle diameter and size distributions across insulin resistance (IR) groups. Panel A is a box plot comparing particle diameter (nm) for Morning and Afternoon milk. Each milk group has three box plots: High IR, Moderate IR, and Low IR. For both Morning and Afternoon milk, Low IR has the lowest median diameter, followed by Moderate IR, then High IR. An asterisk indicates a difference between Morning and Afternoon milk for the High IR group, and between High IR and Low IR for Morning milk. Panels B and C are line graphs showing particle concentration (particles/mL) vs. particle diameter (nm) for Morning milk (B) and Afternoon milk (C). Both graphs show similar trends for High IR, Moderate IR, and Low IR lines: they peak around 100-150 nm and decline towards 500 nm.

Open in a new tab

Milk extracellular vesicles (MEVs) diameters and size distributions in insulin resistance (IR) groups high, moderate, and low. (A) MEV particle diameters of low-, moderate-, and high-IR groups from morning and afternoon milk. (B) MEV size distribution of low-, moderate-, and high-IR groups from morning milk. (C) MEV size distribution of low-, moderate-, and high-IR groups from afternoon milk (mean ± SEM, n = 4) (*p ≤ 0.05).

3.5. Long-chain fatty acids are higher in abundance in high-IR animals

FA analysis of MEVs resulted 26 different FAs (Table 2). The abundance of these FAs as a percentage of total FAs between the different IR groups is depicted in supplementary Table 1.

Table 2.

Fatty acids detected in MEVs.

Fatty acid group Fatty acids
Short-chain fatty acids (C2:0–C6:0) C6:0
Medium-chain fatty acids (C7:0–C12:0) C8:0, C10:0, C12:0
Long-chain fatty acids (C13:0–C21:0) C14:0, C15:0, C16:0, C16:1c9, C17:0, C18:0, C18:1t11, C18:1c9, C18:c11, C18:2n6, C18:3n3, CLA c9t11, C20:0, C20:1c8.9, C20:1c11, C20:3n6+, C20:4n6, C20:5n3
Very long-chain fatty acids (C22 and more) C22:0, C22:5n3, C24:0, C24:1c15
Open in a new tab

Fatty acids with significantly different in abundance are indicated with bold letters.

Behenic acid (C22:0) was significantly higher (p ≤ 0.05) in both the morning and afternoon MEV samples from high-IR group (Figure 7A and B). Pentadeconoic acid (C15:0) was significantly lower (p ≤ 0.05) in abundance in the high-IR group, whereas lignoceric acid (C24:0) was significantly higher (p ≤ 0.05) in abundance in the high-IR group in the afternoon MEVs (Figure 7B).

Figure 7.

A two-panel bar graph shows fatty acid percentages in milk extracellular vesicles for high, moderate, and low IR groups. The two-panel bar graph shows fatty acid percentages in milk extracellular vesicles for high, moderate, and low insulin resistance groups in morning MEVs and afternoon MEVs. Panel A displays Behenic acid percentages which were significantly different in morning MEVs. The high IR group shows the highest percentage, followed by the moderate IR group, then the low IR group. Asterisks indicate statistical significance between all three groups. Panel B presents Pentadecanoic acid, Behenic acid, and Lignoceric acid percentages which were significantly different in afternoon MEVs. For Pentadecanoic acid, the moderate IR group has a slightly higher percentage than the low IR and high IR group, with significance between all groups. For Behenic acid, the high IR group has the highest percentage, followed by moderate IR, then low IR, with significance between all groups. For Lignoceric acid, the high IR group has the highest percentage, followed by moderate IR, then low IR, with significance between all groups.

Open in a new tab

Fatty acids detected in milk extracellular vesicles in the high-, moderate- and low-insulin resistance (IR) groups as a percentage of total fatty acids. (A) Behenic acid (C22:0) abundance in morning milk EVs. (B) Pentadeconoic acid (C15:0), behenic acid (C22:0), and lignoceric acid (C24:0) abundance in afternoon milk EVs (mean ± SEM, n = 4) (*p < 0.05).

3.6. MEV proteomic profiles are varied between different IR groups

The proteomic analysis of MEVs from IR animals identified 1371 proteins in total, most of which have been previously reported in MEV studies. Eighty of these proteins were among the top 100 EV proteins listed in Vesiclepedia, as indicated in Figure 8A. The principal component analysis showed separation in overall protein enrichment patterns between individual animals among the IR groups (Figure 8B). The comparison between high-IR and low-IR groups could identify 43 differentially expressed proteins; 11 were overexpressed, and 32 were identified as down-regulated (Figure 8C). The Spearman correlation analysis of the AUC values and protein concentrations revealed clear associations of specific proteins with disease severity (Figure 8D).

Figure 8.

Four panel diagram: protein profile analysis of milk extracellular vesicles from different insulin-resistant groups. The four panel diagram shows protein comparisons, principal component analysis, differential enrichment, and correlation coefficients. Panel A is a three circle Venn diagram. The top left circle, labeled previoulsy reported MEV proteins (5706), overlaps with the top right circle, labeled Top 100 EVs in vesiclepedia (100), and the bottom circle, labeled MEVs proteins detected in our study (1371). The intersection of all three circles contains the number 75. Panel B is a scatter plot showing principal component analysis. The horizontal axis is Dim1 37 percentage and the vertical axis is Dim2 15.6 percentage. Three groups are plotted: Low, Moderate, and High. Panel C is a volcano plot. The horizontal axis is log2 fold change H versus L, ranging from minus 5 to 3. The vertical axis is minus log10 p-value, ranging from 0.0 to 3.5. Points are labeled Not significant, Down-regulated, and Up-regulated. Panel D is a horizontal bar graph showing correlation coefficient r on the horizontal axis from minus 1.0 to 1.0. The vertical axis lists protein names. SRI has a bar extending to 0.9, MUC16 to 0.9, JCHAIN to 0.8, SDCBP to 0.8, G5E513 to 0.7, A0A3Q1LWV8 to 0.7, FCGR2B to 0.6, ANXA1 to 0.6, VAT1 to 0.5, RRAS to minus 0.5, RAB18 to minus 0.5, ZNRF2 to minus 0.6, GPD1 to minus 0.6, RABSC to minus 0.7, RAB5A to minus 0.7, RAB2A to minus 0.8, RAB1A to minus 0.8, ACTC1 to minus 0.9, ATIC to minus 0.9, BCL2L15 to minus 0.9, RAB8B to minus 0.

Open in a new tab

The protein profile of milk extracellular vesicles (MEV) from insulin-resistant (IR) cows. (A) Comparison of the detected proteins with previously reported milk EV protein and the top 100 EV proteins reported in vesiclepedia (B) The principal component analysis of protein enrichment patterns among the different IR groups high, moderate, and low. (C) Differential protein enrichment between high (H) and low (L) IR groups. (D) Spearman correlation coefficients with IR severity (n = 4).

The GO analysis revealed the physiological roles of IR-related proteins and associated pathways (Figure 9A and B). An in-depth analysis of the biological processes further highlighted that the proteins detected in MEVs were involved in a wide range of metabolic activities, including primary carbohydrate metabolic processes, which are closely linked to the development of IR (Figure 9C–E).

Figure 9.

A five panel diagram shows gene ontology analysis of milk extracellular vesicle proteins, with bar graphs and pie charts. The five panel diagram shows gene ontology analysis of milk extracellular vesicle proteins. Panel A is a horizontal bar graph titled Description, showing No of genes on the horizontal axis from 0 to 40 in increments of 5. The vertical axis lists GO category terms. Bars are grouped by Cellular component, Biological function, and Molecular function. Biological function has the most entries, with Metabolic process having the longest bar at 35 genes. Panel B is a horizontal bar graph titled Description, showing No of genes on the horizontal axis from 0 to 3 in increments of 1. The vertical axis lists various pathways. Wnt signaling pathway and Vitamin B6 metabolism each have 2 genes. Panel C is a pie chart showing differently abundant proteins related to their biological processes titled PANTHER GO-Slim Biological Process. Panel D is a pie chart titled PANTHER GO-Slim Biological Process, with Level 1: metabolic process. It shows 7 categories, with organic substance metabolic process as the largest slice at 30 percent. Biosynthetic process is the second largest at 25 percent. Panel E is a pie chart titled PANTHER GO-Slim Biological Process, Level 1: metabolic process, Level 2: primary metabolic process. It shows 3 categories, with carbohydrate metabolic process as the largest slice at 50 percent.

Open in a new tab

Gene ontology analysis of milk Extracellular vesicles (MEV) proteins. (A) Functional classification of differently abundant proteins. (B) Pathways involved with the reported differentially expressed proteins. (C) Differently abundant proteins related to their biological processes. Gene ontology analysis of the (D) metabolic process, (E) primary metabolic process.

4. Discussion

Milk is a valuable biofluid containing biomarkers that reflect physiological and pathological changes in the body. Any changes in the body can affect the quality and quantity of milk; therefore, it serves as a promising candidate for the diagnosis and prognosis of various disorders (Prasadani et al. 2024). The current study aimed to identify potential biomarkers for IR. Significant differences in MEV diameter and changes in FA, and protein profiles among the cows at different IR statuses were observed. Further analysis of MEVs revealed significantly larger particle diameters in the high-IR group compared to the moderate- and low-IR groups, marking one of the key findings of the present study. This discovery could potentially lead to the development of technologies for real-time diagnosis of cows with IR. With the ongoing advancements in robotic milking technologies and real-time health monitoring platforms, it will be possible to integrate current findings into a system that combines automatic milking with microfluidics for EV separation and characterization from milk (Talebjedi et al. 2021). Such a system with advanced, compact, miniaturized NTA for real-time characterization of EVs in milk enables the detection of cows at risk of or already suffering from IR.

Several studies suggest that an individual's health status is reflected in the altered EV concentrations, size and EV cargo in biofluids (Dimik et al. 2023). Piibor et al. reported that EV particle concentration and size are different in uterine fluid-derived EVs among cows at various endometritis stages (Piibor et al. 2024). In addition, similar observations of EV size differences were found in cerebrospinal fluid-derived EVs from Alzheimer's patients (Longobardi et al. 2022). Moreover, previous studies revealed changes in the plasma-derived EV diameters of type 2 diabetic patients compared to healthy individuals (Pierdoná et al. 2022). In the current investigation, a similar pattern was observed, with differences in MEV sizes in the high-IR group compared to the moderate- and low-IR animals. Since MEVs are believed to originate from various tissues/cells and carry a diverse range of cargo, any abnormalities or changes in distant tissues/cells can impact the characteristics of the EVs in milk. Thus, EVs in milk may reflect the physiological status not only in the mammary gland but also in other tissues as well. However, further confirmation of MEV particle size using fluorescent dyes was not feasible, as the currently available dyes bind to lipid bilayers and cannot distinguish between EVs and lipid droplets.

FA analysis of MEVs revealed 26 distinct FAs with a higher abundance of LCFAs behenic and lignoceric acids. Lignoceric acid is a component of ceramides, and higher ceramide levels have previously been reported to be associated with IR (Rico et al. 2015). Ceramides are bioactive sphingolipids implicated in the pathogenesis of IR through their inhibitory effects on the insulin signaling cascade. They impair insulin-stimulated Akt/PKB activation and attenuate downstream metabolic actions such as glucose uptake and glycogen synthesis (Chavez and Summers 2012). There are previous studies on miRNA profiles of follicular fluid-derived EVs (Hailay et al. 2019) and blood-derived EV protein cargo (Crookenden and Dukkipati 2016) concerning the metabolic status of dairy cows. However, the use of MEV FA profiles to assess metabolic status, especially IR, is a novel approach. Moreover, the FA analysis of milk (Supplementary Table 2) from the same animals did not indicate any significant abundance of the reported FAs; thus, studying the MEV FA profiles led us to discover masked and hidden biomarkers.

The possible MEV proteomic changes among the different IR stages were delineated using LC/MS–MS analysis, and interestingly, several previously reported IR-related proteins were also found to be varied in abundance. Four of the upregulated proteins were involved in regulating carbohydrates, and lipid-regulated metabolism mechanisms. One such protein is annexin A1 (ANXA1), which has previously been reported as a biomarker related to IR in adipose tissue in peripartum dairy cows (Zachut 2015; Piras et al. 2019). The ANXA1 concentration was positively correlated with IR severity. Several studies have shown altered ANXA1 expression in the adipose tissue of obese or diabetic humans, whereas increased ANXA1 expression was found in the white adipose tissue of obese humans (Pietrani et al. 2018). Annexins regulate cell signaling, proliferation tissue repair, and modulate innate and adaptive immunity, support adipose tissue mass, influence lipolysis and IL-6 release by altering epididymal fat sensitivity to catecholamines and glucocorticoids, insulin-dependent translocation of GLUT4, as well as CD36-mediated FA uptake (Grewal et al. 2019). The positive correlation of ANXA1 with IR severity might be due to the reduced capability to fulfill the energy requirements necessary during pregnancy and lactation and undertake a fast mobilization of adipose reserves, which are not supported by the expression of annexins.

Another interesting finding in the proteomic analysis was that RAB family proteins such as RAB1A, RAB2A, RAB5A, RAB5C, RAB8B, RAB10, and RAB18 were found to be downregulated. RAB family proteins are small GTPases that play essential roles in intracellular membrane trafficking and have been linked to IR, particularly through their role in GLUT4 vesicle trafficking and endosomal recycling in insulin-sensitive tissues such as muscle and adipose tissue (Vazirani et al. 2016). Previous findings indicate that IR in dairy cows is often accompanied by metabolic inflammation, characterized by a chronic, low-level inflammatory state (Qiao et al. 2024). This inflammation can disrupt normal insulin signaling pathways, potentially affecting proteins involved in vesicle trafficking, such as RAB proteins. The downregulation of all RAB family proteins hampers glucose uptake, disrupts insulin receptor function, alters lipid metabolism, and promotes oxidative stress, all of which contribute to metabolic dysfunction (Vazirani et al. 2016). Pathway analysis revealed significant alterations in multiple pathways associated with inflammation (Figure 9B).

In the present study, an MEV enrichment protocol with high purity was successfully optimized, and EV enrichment was confirmed in accordance with the International Society for Extracellular Vesicles (ISEV) 2023 guidelines. The purity of the EV population is imperative in biomarker discovery, as contaminants can mask the specific biomarkers present in EVs. Milk is a highly heterogeneous solution, and isolation of MEV is challenging due to overlapping sizes with similar-sized molecules such as casein micelles and lipoproteins (Buratta et al. 2023). An MEV population with higher purity was obtained using the optimized double SEC protocol. Pre-processing steps with centrifugation and acid-based casein precipitation were employed to eliminate cells, fat, casein, and microbial contaminants. To ensure further removal of soluble proteins, any residual caseins, fat globules and to enrich the MEVs, SEC was used. A second SEC step was employed for further enhancement of the purity of the resultant MEVs. Even though there are divided findings on the acid precipitation of casein before the MEV isolation (Morozumi et al. 2021), clear EV surface markers were observed with western blotting, and typical EV morphology with TEM images in the present study, similar to previous reports (Rahman et al. 2019). The recent Minimal Information for Studies of Extracellular Vesicles Guidelines (MISEV) 2023 also recommends the use of a combination of isolation methods for enriching EVs (Welsh et al. 2024), and there is recent evidence that double SEC is effective in EV isolation from complex biological fluids such as plasma (Jung et al. 2021), urine (Pinheiro et al. 2024), and feces (Premathilaka et al. 2024). Introducing a second SEC step is more effective in removing contaminants, resulting in a more uniform MEV population, which enhances the quality and sensitivity of the subsequent high-throughput analysis of MEVs for identifying a biomarker.

Furthermore, the present data suggest that short-term whey storage at −80 °C does not affect the MEV particle concentration and EV characteristics. Mounting evidence in the literature suggests that the storage of samples and repeated freeze cycles impact EV properties (Maroto et al. 2017). Similarly, storage of unprocessed breast milk at −80 °C or 4 °C has led to residual cell death in milk, leading to contamination of the breast MEV population. Thus, it is recommended to store the milk samples after pre-processing by removing cells, fat and casein particles (Welsh et al. 2024). In the present study, the milk samples were pre-processed, and the resultant whey was stored at −80 °C. Then the MEV yield was compared with that of freshly isolated MEVs. There is a potential for the rupturing of large particles contained in samples, such as apoptotic bodies, and casein aggregates into small-sized particles where it overlaps with the EV size range can potentially result in an increased number of nanoparticles during storage (Gelibter et al. 2022). However, in the present investigation, significant changes were not observed, and TEM images and analysis of EV protein markers further confirmed that the short-term storage of whey did not affect the resultant MEV quantity and quality.

Moreover, the milk samples at days 4.67 ± 0.40 after calving were used for the present study, and frequent collection of milk samples and MEV analysis throughout the lactation curve further facilitated a better understanding of the physicochemical differences of MEV as the MEV content differed between colostrum, first milk, and milk over the rest of the lactation curve (Santoro et al. 2023). This study provides initial insights into the physicochemical traits of cows with varying IR statuses, and large-scale field validation under diverse conditions will reveal more comprehensive information.

5. Conclusion

The developed double SEC method demonstrated a simple and effective approach that can be practiced when enriching MEVs for downstream applications are required. The changes in MEV particle diameter, higher abundance of LCFAs such as Behenic acid (C22:0) and Lignoceric acid (C24:0), and upregulation of ANXA1 and downregulation of RAB family proteins are associated with the severity of IR and might be an early sign of IR and thus could be used to identify the risk of IR development.

Supplementary Material

Supplementary Material

Supplementary Figures and tables.docx

TVEQ_A_2665485_SM2642.docx (1.8MB, docx)

Acknowledgements

Special thanks are extended to the Märja experimental farm of the Estonian University of Life Sciences and its staff members, Alar Onoper, Mariin Talu, and Eveli Somelar, for carrying out IVGTT, Anette Matto, and Anne Alliku as technicians caring for animals. We are grateful for the cows complying with the manipulations.

Funding Statement

This research was funded by the European Union's Horizon 2020 research and innovation program under grant agreement No. 857418 COMBIVET; European Union through Horizon's coordination and support actions under grant agreement No. 101079349 OH-Boost and by the Estonian Ministry of Rural Affairs under grant No. 616215780001.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD067365.

Ethics approval statement

All experiments involving animals were approved by the Committee for Conducting Animal Experiments at the Estonian Ministry of Regional Affairs and Agriculture (License no. 232, 19.12.2022).

Abbreviations

IR

insulin resistance

EVs

extracellular vesicles

MEVs

milk extracellular vesicles

IVGTT

intravenous glucose tolerance test

FAs

fatty acids

AUC

area under the curve

NEB

negative energy balance

NEFAs

non-esterified fatty acids

LCFAs

long-chain fatty acids

ME

metabolizable energy

MP

metabolizable protein

DM

dry matter

BHB

ß-hydroxybutyrate

CV

coefficients of variation

SEC

size exclusion chromatography

NTA

Nanoparticle Tracking Analyzer

Fps

frames per second

TEM

transmission electron microscopy

RIPA

radioimmunoprecipitation assay

LC–MS/MS

liquid chromatography–mass spectrometry

ANXA1

AnnexinA1

ISEV

international society for extracellular vesicles

MISEV

minimal information for studies of extracellular vesicles

Supplemental material

Supplemental data for this article can be accessed at https://doi.org/10.1080/01652176.2026.2665485.

References

  1. Almughlliq FB, et al. 2018. Theriogenology proteomic content of circulating exosomes in dairy cows with or without uterine infection. Theriogenology. 114:173–179. 10.1016/j.theriogenology.2018.03.024 [DOI] [PubMed] [Google Scholar]
  2. AOAC . 2019. Horwitz W, Latimer GW, editors. Official Methods of Analysis 21st Edition. 2019. AOAC Int: Gaithersburg, MD. [Google Scholar]
  3. Buratta S, et al. 2023. Protein and lipid content of milk extracellular vesicles: a comparative overview. Life. 13(401):1–14. 10.3390/life13020401 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Chalmeh A, et al. 2015. Insulin resistance in different physiological states of high producing Holstein dairy cows. Acta Sci. Vet. 43(1255), https://www.redalyc.org/articulo.oa?id=289039763027 [Google Scholar]
  5. Chavez JA, Summers SA. 2012. Perspective A ceramide-centric view of insulin resistance. Cell Metab. 15(5):585–594. 10.1016/j.cmet.2012.04.002 [DOI] [PubMed] [Google Scholar]
  6. Chilliard Y, Ferlay A, Mansbridge RM, Doreau M. 2020. Ruminant milk fat plasticity: nutritional control of saturated, polyunsaturated, trans and conjugated fatty acids. Ann Zootech. 49(3):181–205. 10.1051/animres:2000117 [DOI] [Google Scholar]
  7. Crookenden MA, Dukkipati VSR. 2016. Short communication: proteins from circulating exosomes represent metabolic state in transition dairy cows. Jouranl Dairy Sci. 99:7661–7668. 10.3168/jds.2015-10786 [DOI] [PubMed] [Google Scholar]
  8. De Juano MDS, Silvestrelli G, Bauersachs S, Ulbrich SE. 2022. Determining extracellular vesicles properties and miRNA cargo variability in bovine milk from healthy cows and cows undergoing subclinical mastitis. BMC Genomics. 23(189):1–15. 10.1186/s12864-022-08377-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. De Koster J, et al. 2016. Validation of different measures of insulin sensitivity of glucose metabolism in dairy cows using the hyperinsulinemic euglycemic clamp test as the gold standard. Domest Anim Endocrinol. 57:117–126. 10.1016/j.domaniend.2016.06.004 [DOI] [PubMed] [Google Scholar]
  10. De Koster JD. 2013. Insulin resistance in dairy cows. in Veterinary Clinics of NA: Food Animal Practice. 29(2):299–322. 10.1016/j.cvfa.2013.04.002. Elsevier Inc. [DOI] [PubMed] [Google Scholar]
  11. Dimik M, Abeysinghe P, Logan J, Mitchell M. 2023. The exosome: a review of current therapeutic roles and capabilities in human reproduction. Drug Deliv Transl Res. 13:473–502. 10.1007/s13346-022-01225-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dissanayake K, Midekessa G, Lättekivi F, Fazeli A. 2021. Measurement of the size and concentration and zeta potential of extracellular vesicles using nanoparticle tracking analyzer. In: Methods Mol Biol. 2273. p 207–218. 10.1007/978-1-0716-1246-0_20 [DOI] [PubMed] [Google Scholar]
  13. Gelibter S, et al. 2022. The impact of storage on extracellular vesicles: a systematic study. J Extracell Vesicles. 11(October 2021), 10.1002/jev2.12162 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Grewal T, Enrich C, Rentero C. 2019. Annexins in adipose tissue: novel players in obesity. Int J Mol Sci. 20:3449. 10.3390/ijms20143449 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hailay T, et al. 2019. Extracellular vesicle-coupled miRNA profiles in follicular fluid of cows with divergent post-calving metabolic status. Sci Rep 9(April):1–14. 10.1038/s41598-019-49029-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Jaakson H, et al. 2018. Adipose tissue insulin receptor and glucose transporter 4 expression, and blood glucose and insulin responses during glucose tolerance tests in transition Holstein cows with different body condition. J Dairy Sci. 101(1):752–766. 10.3168/jds.2017-12877 [DOI] [PubMed] [Google Scholar]
  17. Jiang L, Sun H. 2022. Long-chain saturated fatty acids and its interaction with insulin resistance and the risk of nonalcoholic fatty liver disease in type 2 diabetes in Chinese. Front Endocrinol (Lausanne). 13(December):1–9. 10.3389/fendo.2022.1051807 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Jorjong S, et al. 2015. Milk fatty acids as possible biomarkers to diagnose hyperketonemia in early lactation. J Dairy Sci 98(2009):5211–5221. 10.3168/jds.2014-8728 [DOI] [PubMed] [Google Scholar]
  19. Jung JH, et al. 2021. Dual size-exclusion chromatography for efficient isolation of extracellular vesicles from bone marrow derived human plasma. Sci Rep. 11(1):1–9. 10.1038/s41598-020-80514-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kaminska MGK, et al. 2023. Theriogenology characteristics of size-exclusion chromatography enriched porcine follicular fl uid extracellular vesicles e. Theriogenology. 205:79–86. 10.1016/j.theriogenology.2023.04.010 [DOI] [PubMed] [Google Scholar]
  21. Li K, et al. 2022. A signature of saliva-derived exosomal small RNAs as predicting biomarker for esophageal carcinoma: a multicenter prospective study. Mol Cancer. 21:1–13. 10.1186/s12943-022-01499-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Longobardi A, et al. 2022. Cerebrospinal fluid EV concentration and size are altered in alzheimer's disease and dementia with lewy bodies. Cells. 11(462), 10.3390/cells11030462 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lv Y, Lu C. 2019. Roles of microRNAs in preeclampsia. J Cell Physiol 2019(July 2018):1052–1061. 10.1002/jcp.27291 [DOI] [PubMed] [Google Scholar]
  24. Lyu L, et al. 2019. Small RNA profiles of serum exosomes derived from individuals with latent and active tuberculosis. Front Microbiol 10(May):1–10. 10.3389/fmicb.2019.01174 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Maroto R, et al. 2017. Effects of storage temperature on airway exosome integrity for diagnostic and functional analyses. J Extracell Vesicles. 6(1), 10.1080/20013078.2017.1359478 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mi H, et al. 2019. PANTHER version 14: more genomes, a new PANTHER GO-slim and improvements in enrichment analysis tools. Nucleic Acids Res. 47(November 2018):419–426. 10.1093/nar/gky1038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Morozumi M, Izumi H, Shimizu T, Takeda Y. 2021. Comparison of isolation methods using commercially available kits for obtaining extracellular vesicles from cow milk. J Dairy Sci. 104(6):6463–6471. 10.3168/jds.2020-19849 [DOI] [PubMed] [Google Scholar]
  28. Mousavi SO, Ul Q, Reshi A, Godakumara K. 2024. Extracellular vesicles as mediators of stress response in embryo-maternal communication. Front Cell Dev Biol 12(August):1–13. 10.3389/fcell.2024.1440849 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Nakanishi R, et al. 2022. Comparing microRNA in milk small extracellular vesicles among healthy cattle and cattle at high risk for bovine leukemia virus transmission. J Dairy Sci. 2020:5370–5380. 10.3168/jds.2021-20989 [DOI] [PubMed] [Google Scholar]
  30. Pierdoná TM, et al. 2022. Extracellular vesicles as predictors of individual response to exercise training in youth living with obesity. Front Biosci. 27(5):143. 10.31083/j.fbl2705143 [DOI] [PubMed] [Google Scholar]
  31. Pietrani NT, et al. 2018. Proresolving protein annexin A1: the role in type 2 diabetes mellitus and obesity. Biomed Pharmacother 103(October 2017):482–489. 10.1016/j.biopha.2018.04.024 [DOI] [PubMed] [Google Scholar]
  32. Piibor J, et al. 2023. Uterine fluid extracellular vesicles proteome is altered during the estrous cycle authors uterine fluid extracellular vesicles proteome is altered during the estrous cycle. Mol Cell Proteomics. 22(11):100642. 10.1016/j.mcpro.2023.100642 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Piibor J, et al. 2024. Investigation of uterine fluid extracellular vesicles' proteomic profiles provides novel diagnostic biomarkers of bovine endometritis. Biomolecules. (14):626. 10.3390/biom14060626 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Piibor J, et al. 2023. Characterization of bovine uterine fluid extracellular vesicles proteomic profiles at follicular and luteal phases of the oestrous cycle. Vet Res Commun. 47(2):885–900. 10.1007/s11259-022-10052-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Pinheiro C, et al. 2024. Identification and validation of extracellular vesicle reference genes for the normalization of RT-qPCR data. J Extracell Vesicles. 13(February 2023), 10.1002/jev2.12421 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Piras C, et al. 2019. Unraveling the adipose tissue proteome of transition cows through severe negative energy balance. Animals. 9:1013. 10.3390/ani9121013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Prasadani M, et al. 2024. Therapeutic potential of bovine milk-derived extracellular vesicles. Int J Mol Sci. 25:5543. 10.3390/ijms25105543 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Premathilaka C, et al. 2024. Bovine faecal extracellular vesicles: a novel non-invasive tool for understanding gut physiology and pathophysiology in calves. J Dairy Sci. 10.3168/jds.2024-25920 [DOI] [PubMed] [Google Scholar]
  39. Premathilaka C, et al. 2025. Immunomodulatory effects of Ascardia galli excretory-secretory products, including extracellular vesicles on host immune cell function. 9457. Taylor & Francis. 10.1080/03079457.2025.2528838 [DOI] [PubMed] [Google Scholar]
  40. Pritam S, Sukhija S, Palmquist DL. 1988. Rapid method for determination of total fatty acid content and composition of feedstuffs and feces. J Agric Food Chem. 36:1202–1206. [Google Scholar]
  41. Qiao K, et al. 2024. The complex interplay of insulin resistance and metabolic inflammation in transition dairy cows. Animals. 14:1–17. 10.3390/ani14060832 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Rahman MM, et al. 2019. Acidification effects on isolation of extracellular vesicles from bovine milk. PLoS One. 14(9):1–12. 10.1371/journal.pone.0222613 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Rico JE, et al. 2015. Plasma ceramides are elevated in overweight Holstein dairy cows experiencing greater lipolysis and insulin resistance during the transition from late pregnancy to early lactation. J Dairy Sci. 98(11):7757–7770. 10.3168/jds.2015-9519 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Rilanto T, et al. 2020. Culling reasons and risk factors in Estonian dairy cows. BMC Vet Res 16(173):1–16. 10.1186/s12917-020-02384-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Santoro J, et al. 2023. An investigation of extracellular vesicles in bovine colostrum, first milk and milk over the lactation curve. Food Chem. 401(August 2022):134029. 10.1016/j.foodchem.2022.134029 [DOI] [PubMed] [Google Scholar]
  46. Talebjedi B, Tasnim N, Hoorfar M, Mastromonaco GF. 2021. Exploiting microfluidics for extracellular vesicle isolation and characterization: potential use for standardized embryo quality assessment. Front Vet Sci 7(January):1–18. 10.3389/fvets.2020.620809 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Vazirani RP, et al. 2016. Disruption of adipose Rab10-Dependent insulin signaling causes hepatic insulin resistance. Diabetes. 65(June):1577–1589. 10.2337/db15-1128 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Welsh JA, et al. 2024. Minimal information for studies of extracellular vesicles (MISEV 2023): from basic to advanced approaches. J Extracell Vesicles. 13(December 2023), 10.1002/jev2.12404 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Zachut M. 2015. Defining the adipose tissue proteome of dairy cows to reveal biomarkers related to peripartum insulin resistance and metabolic status. J Proteome Res. 14:2863–2871. 10.1021/acs.jproteome.5b00190 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material

Supplementary Figures and tables.docx

TVEQ_A_2665485_SM2642.docx (1.8MB, docx)

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD067365.

Articles from The Veterinary Quarterly are provided here courtesy of Taylor & Francis

RESOURCES