Assessment of in vitro digestibility and post-digestion peptide release of mare milk in relation to different management systems and lactation stages

Mare milk has a unique protein composition that makes it a preferred option for adult and infant nutrition. Several functional properties have been attributed to this milk but with little evidence yet. In fact, knowledge on mare milk composition is still limited. In particular, studies addressing the performance of mare milk proteins during human gastrointestinal digestion are scarce, which limits the understanding of mare milk nutritional quality and functionality. For this reason, the present study describes the digestibility of mare milk proteins and the release of peptides as affected by management and lactation stage, factors known to affect milk composition. Mare milk samples from 3 different farms, and collected during 6 mo of lactation (n = 54), were subjected to a static in vitro gastrointestinal model to measure peptide release and protein digestibility. In the present study, a detailed description of protein and individual amino acid behavior during the digestion process was given. For the first time, digestion of the 2 equine β-lactoglobulin isoforms (I and II) was described individually. In addition, it was found that lactation stage and management system can significantly affect protein digestibility and peptide release during gastrointestinal digestion of mare milk. Presumably, differences in the composition of mare milk influence the protein structure and enzyme accessibility, which might have an impact on digestion behavior. Despite no specific bioactive peptides were identified, several precursors of previously described bioactive peptides were found. These findings could support the idea of mare milk as a food with added value.


INTRODUCTION
Mare milk is one of the animal milks most similar to human milk (Uniacke-Lowe et al., 2010), and appears to have a low allergenic response in patients with cow milk protein allergy (Businco et al., 2000;Zhao et al., 2023) although this still needs to be confirmed by clinical studies.These properties make mare milk a suitable option for infant nutrition.Moreover, it has an interesting protein profile.Whereas bovine milk is mainly composed of caseins (~80%), mare milk protein distribution (~55/45 caseins/whey proteins) is more similar to that of human milk (~30/70).The main protein in mare milk is β-casein, followed by similar levels of αs1-casein, β-lactoglobulin and α-lactalbumin.Conversely, low contents of αs2-and κ-casein have been reported.Despite being a minor protein, lysozyme contents are particularly high in mare milk compared with milk from other species, and lactoferrin contents are also considerable (Miranda et al., 2004;Uniacke-Lowe et al., 2010).Some particularities of mare milk proteins are the presence of 2 β-lactoglobulin isoforms (I and II;Godovac-Zimmermann et al., 1985;Halliday et al., 1991;Miranda et al., 2004), 3 genetic variants of α-lactalbumin (A, B and C;Godovac-Zimmermann et al., 1987), and a highly glycosylated κ-casein (Jaeser et al., 2023).In addition, equine lysozyme has the ability to bind calcium similar to α-lactalbumin, a property thought to be an evolutionary linkage between lysozymes with non-calcium binding activity and α-lactalbumin (Nitta et al., 1987).
In addition to knowing the protein composition and characteristics of a product, understanding the behavior of these proteins during gastrointestinal digestion is key for a comprehensive understanding of a protein source quality.To date, very few studies have been conducted on the human digestion of mare milk (Inglingstad et al., 2010;Xiao et al., 2023), and these focused on comparing mare milk with human and other animal species' milk rather than on understanding the effect of different factors on mare milk protein digestion.Moreover, Xiao et  2023) used simulated infant digestion conditions, which are different from adult digestion conditions, while Inglingstad et al., 2010 used a 2-step (stomach and duodenum, 30 min each) adult digestion simulation that might be limited at resembling real digestive conditions.Despite their limitations, these studies claim that mare milk proteins are highly digestible, with a higher protein degradation degree in mare milk than in ruminant milk, but lower than in human milk.In the mentioned studies, most of the proteins were overall degraded already in the gastric phase, followed by a rapid intestinal hydrolysis of the proteins remaining in the gastric digesta.However, protein digestion is a complex process, and more research is needed to fully understand the particularities of mare milk protein changes during gastrointestinal digestion.This requires the application of latest methodologies that are constantly being improved and updated.In this sense, the INFOGEST 2.0 protocol setup (Brodkorb et al., 2019) proved useful for characterizing the digestibility and release of peptides in a wide range of dietary proteins (Portmann et al., 2023;Sousa et al., 2023), and could also be useful for mare milk studies.
Milk is a fluid with a variable composition that can be modified in response to several factors.For instance, the feeding regimen of the animal as part of the management system, and lactation stage can significantly affect mare milk nutritional composition.Overall, the content of most chemical compounds (fat, proteins, mineral elements…) tends to decrease from initial to final stages of lactation, except for lactose that increases.In addition, forage feeding can increase the content of n-3 polyunsaturated fatty acids in milk, but decrease the abundance of some mineral elements such as calcium, phosphorous, sulfur and sodium (Salimei and Park, 2017;Barłowska et al., 2023;Blanco-Doval et al., 2023).For an adequate and comprehensive characterization of mare milk, consideration of these factors in research is of great importance.Unfortunately, to the best of our knowledge, previous studies about mare milk digestion and peptide release did not account for external factors that might affect its performance in the human digestive system.
The aim of the present study was to describe the digestibility of mare milk proteins during simulated in vitro human gastrointestinal digestion, using the extended IN-FOGEST protocol; to monitor the peptide release during gastrointestinal digestion and identify potential bioactive peptides; and to assess the effects of management and lactation stage on studied parameters.

Animals and sample collection
Fifty-four milk samples were collected from Basque Mountain Horse breed mares belonging to 3 commercial farms located in Araba region (northern Spain).Nine mares (3 per farm) were milked at wk 6, 10, 14, 18, 22 and 26 of lactation, between May and October, 2021.Farms were primarily dedicated to horse meat production but with differences in management among them.Briefly, in farm I mares were pasture fed only during May, and after, mares were switched to a mixture of alfalfa, silage, hay, fruits and potatoes until weaning in October.In farms II and III mares were kept on grazing during the complete lactation period, and only after July mares were supplemented with hay (farm II) or silage (farm III).More detailed information about the management of mares is available at Blanco-Doval et al. (2023).Because samples were obtained from commercial farms by standard milking procedures, institutional animal use approval was not required.Immediately after milking, samples were transported to the laboratory under refrigeration, subsampled and kept at −80°C until analysis.

Chemicals and reagents
Except otherwise specified, all chemicals, reagents and enzymes were purchased from Merck (Zug, Switzerland).Enzyme and bile salt characteristics and preparation of gastrointestinal fluids (simulated salivary, gastric and intestinal fluids) are described in Supplementary material 1.

Static in vitro simulation of mare milk gastrointestinal digestion
Milk samples were thawed at 4°C and individually digested following the INFOGEST protocol for static in vitro simulation of gastrointestinal digestion (Minekus et al., 2014) with the latest improvements (Brodkorb et al., 2019).In the oral phase, 800 μL amylase free simulated salivary fluid (pH 7, 37°C) were dried in a centrifugal concentrator (CentriVap, Labconco, Kansas City, MO, USA), in order not to exceed the 2 mL volume of the oral phase.Then, quantities of milk normalized to 40 mg of protein (corresponding to an approximate volume of 2 mL) were added.For the gastric phase, 1.6 mL of simulated gastric juice (pH 3, 37°C) containing pepsin (2000 U/mL of digesta) were added, and the solution was incubated for 2 h at 37°C in constant rotation.Finally, for the intestinal phase, 1.7 mL of simulated intestinal fluid (pH 7, 37°C) containing pancreatin (100 U trypsin activity/mL of digesta) and bile (10 mM final concentration) Blanco-Doval et al.: Mare milk digestibility and peptide release were added to the gastric digesta, and the solution was incubated for 2 h at 37°C in constant rotation.Samples used for peptide release evaluation were stopped after the gastric phase by increasing the pH to 7 with sodium hydroxide (2 M), and after the intestinal phase by adding a protease inhibitor (4-(2-aminoethyl) benzenesulfonylfluoride 500mM, Roche, Basel, Switzerland) to the digesta.Then, digested samples were immediately immersed in liquid nitrogen and kept at −20°C until analysis.

Digestibility of mare milk proteins
For analysis of protein digestibility, the procedure by Sousa et al. (2023) was followed.Mare milk samples corresponding to 2 consecutive lactation weeks of each animal were pooled, resulting in 27 pooled samples and 3 lactation stages (early: wk 6 and 10; mid: wk 14 and 18; late: wk 22 and 26).The pooled samples were individually digested in triplicate (section 2.3 until intestinal digestion).Simultaneously, a protein-free cookie was digested as a blank (Moughan et al., 2005) to determine the enzyme background.Digestions were performed in sets of 5-6 pooled samples and one cookie.Replicates were digested in different sets at different days.

Separation of digestible and indigestible fractions
Digested milk samples were fractionated into digestible (potentially absorbable) and indigestible (potentially non-absorbable) fractions (Sousa et al., 2023).Briefly, 32 mL of ice-cold methanol were added to samples after intestinal digestion to precipitate the indigestible fraction.The solution was incubated for 1 h and subsequently centrifuged for 15 min at 2000 g and 4°C (Sorvall Legend XTR, Thermo Scientific, Reinach, Switzerland).Collected supernatants (digestible fraction) were kept at −20°C until analysis.Pellets (indigestible fraction) were washed twice with ice-cold pure methanol, dried in the centrifugal concentrator (CentriVap), and kept at −20°C until analysis.Weights of supernatant and pellet tubes (±0.0001 g accuracy) were monitored for digestibility calculation.

Hydrolysis of digested samples
Previously dried pellets were transferred to 10 mL vials, and 220 μL of each supernatant were transferred to 2 mL glass vials and evaporated using the centrifugal concentrator (CentriVap).Both pellets and supernatants were hydrolyzed in hydrochloric acid and 3,3′-dithiodipropionic acid (acid hydrolysis) at 110°C for 15 h, with norvaline as internal standard.Two cysteine standards (20 and 200 μM final concentrations) were hydrolyzed in parallel following the same procedure (Sousa et al., 2023).

Determination of total amino groups using the o-phthalaldehyde method
After hydrolysis, total amino groups (R-NH 2 ) present in the digestible and indigestible fractions of the samples were analyzed using the o-phthalaldehyde (OPA) method (Sousa et al., 2023).Samples were diluted with perchloric acid (0.5 M; 1:5 supernatants, 1:10 pellets, vol/vol), derivatized with OPA and 2-mercapto-ethansulfonic acid, and analyzed by UV/visible spectrophotometry at 340 nm (Spectramax iD3, Molecular Devices, San José, CA, USA).A calibration curve was built using 9 concentration levels of a glutamic acid standard solution ranging from 0.25 to 8 mM and perchloric acid (0.5 M) as a blank.All samples and calibration levels were measured in duplicate.

Analysis of individual amino acids by ultra-high performance liquid chromatography
Individual amino acid contents in digested and hydrolyzed samples were measured according to the method 2018.06 of the Association of Official Analytical Chemists for infant formula (Jaudzems et al., 2019) adapted by Sousa et al. (2023).Before analysis, samples were derivatized with AccQ-Tag reagent (Waters, Baden, Switzerland).Individual amino acid contents were analyzed according to Waters (2007) instructions, using an ultra-high performance liquid chromatography (UHPLC) equipment (Vanquish Flex, Thermo Scientific) coupled to an UV detector (Vanquish, Thermo Scientific) and an Acquity UPLC BEH C18 column (150 mm length, 2.1 mm internal diameter, 1.7 μm particle size; Waters).Analytical conditions were set to 2 μL injection volume, 50°C column temperature, 260 nm detection wavelength, and 0.4 mL/min flow rate.AccQ-Tag Eluent A (diluted to 15% in ultrapure water, vol/vol; Waters) and formic acid 2% (vol/vol) in acetonitrile were used as mobile phase within a 32 min gradient.

Analysis of undigested mare milk
Individual amino acid content of raw (undigested) pooled mare milk samples was also determined after acid hydrolysis (110°C, 24 h) (sections 2.4.2 and 2.4.4).For tryptophan determination, alkaline hydrolysis with sodium chloride 4.2 M for 20 h at 110°C was used.Tryptophan was determined using an UHPLC equipment (Thermo Ultimate 3000, Thermo Scientific) coupled to a fluorescence detector (Vanquish, Thermo Scientific) and an Acquity UPLC BEH C18 column (150 mm length,  2007) to calculate the total protein content using a nitrogen-to-protein conversion factor of 6.25.

Calculation of in vitro digestibility and digestible indispensable amino acid ratio
The calculations for the protein digestibility of mare milk (Supplementary material 1) were performed (Sousa et al., 2023).The digestible indispensable amino acid ratio (DIAAR) represents the relation between digested indispensable amino acids per gram of food protein and the reference requirement values of individual amino acid for infants, children and adults (Food and Agriculture Organization, 2013).The digestible indispensable amino acid score (DIAAS) refers to the lowest DIAAR of a protein product, and represents the limiting amino acid in a food.

Peptide release after gastric and intestinal digestion
Sample preparation.For the study of peptide release after digestion, all individual mare milk samples were digested (section 2.3), with 2 digestions per sample: one stopped after the gastric phase (gastric digestion) and one subjected to the entire digestion process (intestinal digestion).All digested samples were filtered through 30 kDa molecular weight cut-off micro-spin filters (UFC5030, Millipore, Merck) aided by centrifugation at 18000 rcf for 10 min at 4°C using a 5427R centrifuge (Eppendorf, Hamburg, Germany).
Identification of major proteins in mare milk.To relate released peptides with their parent proteins, major proteins in mare milk were identified by sodium dodecyl sulfate-PAGE (SDS-PAGE) (Kopf-Bolanz et al., 2012).Six undigested milk samples from one mare and 6 lactation weeks were selected, diluted 1:25 (vol/vol) with ultrapure water, and separated with SDS-PAGE in a 15% polyacrylamide gel with a BenckMark protein ladder solution (Invitrogen, Thermo Fisher Scientific).Gels were stained with colloidal Coomassie Blue, and the bands were excised and digested with trypsin overnight at 37°C.

Identification of released peptides after gastric and intestinal digestion.
Digested samples after filtration (in vitro gastrointestinal digestion; section 2.5.1) or trypsin digestion (section 2.5.2) were injected into a HPLC system (Rheos 2200 Micro HPLC, Flux Instruments, Basel, Switzerland) equipped with an electrospray ionization source and coupled to a linear ion trap mass spectrometer (LTQ XT, Thermo Fisher Scientific).A C18 column (X Terra MS, 100 mm length, 2.1 mm internal diameter, 3.5 μm particle size; Waters) was used for peptide separation.Analytical conditions were previously described (Egger et al., 2019).Peptides with at least 5 amino acids and m/z ratio between 300 and 2000 were considered, and repetitive fragmentation was applied to the 5 most intense mass spectrometry (MS) signals.MS/MS spectra were matched against an in-house database containing only milk proteins obtained from UniProt database (https: / / www .uniprot.org/), including the 8 major mare milk proteins identified by gel electrophoresis: αs1-casein (CASA1_EQUAS), αs2-casein (CASA2_EQUAS), β-casein (CASB_HORSE), κ-casein (CASK_HORSE), β-lactoglobulin I (LACB1_HORSE), β-lactoglobulin II (LACB2_HORSE), α-lactalbumin A (LALB1_HORSE) and lysozyme C (LYSC1_HORSE).Peptides were identified using Mascot search engine (Matrix Science, London, UK) and considering some technically induced optional modifications (deamidation of asparagine and glutamine, oxidation of methionine), as well as phosphorylation of serine and threonine as post-translational modifications.Among the peptides identified in mare milk digested samples, bioactive peptides were searched using the Milk Bioactive Peptide Database (MBPDB), which collects milk-derived bioactive peptides identified in literature (Nielsen et al., 2017).
Statistical analysis and data visualization.Heatmaps were created with R programming (RStudio version 4.3.0,Posit PBC, Boston, MA, USA) using the individual amino acid counts within all identified peptides from the corresponding protein.Peptides liberated from the same protein were grouped by farm or lactation stage, and the average number of individual amino acids within protein sequences that belonged to the same group was used.
To explore the effect of management (farm) and lactation stage on mare milk protein digestibility and gastrointestinal fragmentation, the IBM-SPSS statistics software (version 28.0, IBM, New York, NY, USA) was used.In this case, the relative abundance of each individual peptide given by the MS/MS spectra was considered.All data was log-transformed, and tested for normality and homoscedasticity.The general linear model of ANOVA was applied to all data.Farm (I = 3) and lactation stage (I = 3; referring to wk 6 and 10 as early lactation, wk 14 and 18 as mid lactation, and wk 22 and 26 as late lactation) were included as fixed factors, and individual animal (I Blanco-Doval et al.: Mare milk digestibility and peptide release = 9) as a random factor nested within farm.The interaction effect between lactation stage and farm was also included in the model and Tukey's test was applied for pairwise comparisons.For released peptides data, only those peptides present in at least 2 animals per farm and per lactation week were subjected to statistical analysis.In addition, a multivariate analysis using the IBM-SPSS statistics software was applied to peptide release results.Log-transformed data were submitted to Stepwise Discriminant Analysis to discriminate milk samples by farm or by lactation stage.Only peptides that appeared in at least 2 animals per farm and per lactation week were included in the multivariate analysis.Significance level was declared at P-value ≤0.05.

Digestibility of mare milk proteins
Mare milk proteins exhibited an average digestibility (±standard deviation) of 93.7 ± 3.3% in calculations based on individual amino acids, and 96.0 ± 4.7% in calculations based on total amino groups (R-NH 2 ; OPA assay).No significant differences were observed in milk protein digestibility among farms.However, a significant effect of lactation stage was found only when calculations were made with OPA method.In this case, milk proteins showed a significantly (P = 0.005) lower digestibility at late lactation (93.3 ± 3.9%) compared with early (97.0 ± 1.9%) and mid lactation (98.1 ± 0.6%).The interaction effect between farm and lactation stage was also nonsignificant (P > 0.05).
Digestibility of individual amino acids was on average (±standard deviation) higher than 90% in all cases except for alanine (89.2 ± 5.7%), isoleucine (88.3 ± 3.8%), serine (83.8 ± 5.8%) and threonine (88.9 ± 5.2%) (Supplementary material 1).Overall, farm and lactation stage had little effect on digestibility of individual amino acids.However, significant (P ≤ 0.05) influence of farm and lactation stage was observed for some DIAAR values (Table 1).In the case of farm effect, DIAAR values of sulfur-containing amino acids and tryptophan where higher in low grazing (farm I) versus high grazing farms (farms II and III).Conversely, milk from high grazing with hay supplementation (farm II) exhibited highest DIAAR values of aromatic amino acids, and lowest DI-AAR of lysine only until mid lactation (at late lactation, the lysine ratios were equalized).These differences were mainly explained by differences in the amino acid content of undigested raw milk (Table 1).Overall, raw mare milk from farm I contained higher amounts of methionine (the major sulfur-containing amino acid), tryptophan, and lysine (only until mid lactation) whereas milk from farm II contained higher amounts of phenylalanine (the major aromatic amino acid).Regarding changes during lactation, DIAAR values of histidine, leucine, lysine and valine were significantly (P ≤ 0.05) higher at late lactation compared with early and mid lactation.This was again explained by the higher amino acid contents in undigested raw mare milk at late lactation (Table 1).As an exception, lysine content increased at late lactation only in milk from high grazing farms whereas it decreased from early to late lactation in milk from the low grazing farm.This resulted in a significant (P = 0.029) interaction between farm and lactation stage in lysine content in raw mare milk.A similar interaction was found in raw mare milk leucine content, in which milk from high grazing farms changed differently than low grazing milk samples during lactation.Despite significant changes observed in DIAAR values depending on farm and lactation stage, the essential to non-essential amino acid ratio remained stable in all mare milk samples (data not shown).
Mean DIAAR values calculated for all mare milk samples based on infant, child and adult nutrition are shown in Figure 1.Considering DIAAR values, the limiting amino acids were threonine for infants (DIAAS 74.5%) and histidine for children and adults (DIAAS 101.5 and 126.9%, respectively).

Peptide release after gastrointestinal digestion of mare milk
Since major proteins in mare milk (identified through gel electrophoresis) were αs1-, αs2-, β-and κ-caseins, β-lactoglobulin I and II, α-lactalbumin A and lysozyme C, peptides released from these 8 proteins were also monitored.β-casein was the predominant source of peptides with 386 non-redundant peptides found after gastric and intestinal digestions, while less than 80 non-redundant peptides were released, respectively, from κ-casein, α-lactalbumin A and lysozyme C (Table 2).Considering peptide size, peptides released from caseins, β-lactoglobulin I and α-lactalbumin A had significantly (P ≤ 0.05) lower mass after intestinal than after gastric digestion.In contrast, peptides released from β-lactoglobulin II and lysozyme C did not show higher fragmentation after intestinal digestion (Figure 2).
Heatmaps in Figure 3 describe the behavior of peptides released from each major mare milk protein during gastric and intestinal digestion.Interpretation of peptide patterns was done as suggested by Portmann et al. (2023).The results confirmed that β-casein was the main source of peptides, followed by αs1-casein and β-lactoglobulin I.In accordance to the significant reduction of peptide size from gastric to intestinal digestion (Figure 2), most peptides formed during gastric digestion of αs1-, αs2-, β-and κ-casein, β-lactoglobulin I and α-lactalbumin A were further hydrolyzed during intestinal digestion.In  3).Some individual peptides released from gastrointestinal digestion showed significantly (P ≤ 0.05) different relative abundances in milk samples belonging to different farms or lactation stages (Table 3).Farm significantly affected the relative abundance of 31 different peptides (5 from αs1-casein; 19 from β-casein; 5 from β-lactoglobulin I; 1 from α-lactalbumin A; and 1 from lysozyme C) formed after gastric digestion, whereas only 3 peptides (2 from αs1-casein and 1 from β-casein) were affected by intestinal digestion.Overall, a higher peptide release occurred when milk samples belonged to low grazing activity (farm I) compared with high grazing and silage supplementation (farm III), whereas samples from high grazing and hay supplementation (farm II) were usually intermediate (Figure 3).In addition, lactation stage significantly (P ≤ 0.05) affected the relative abundance of 6 different peptides (all derived from β-casein) formed during gastric digestion, but none of the peptides from intestinal digestion.In this regard, a higher peptide release was observed when milk samples were from late lactation (Figure 3).The multivariate discriminant analysis supported the differences observed in the relative abundance of individual peptides in mare milk samples.In fact, the discriminant functions were able to classify more than 60% of milk samples into their respective farm (farm I, II and III) or lactation stage (early, mid and late lactation) group based on the peptide profile after gastric or intestinal digestion, respectively.Overall, the multivariate analysis strongly discriminated between samples from early and late lactation, while samples from mid lactation were often intermingled with early or late lactation stages (Figure 4).The most accurate classification for lactation stage groups was achieved using the peptide profile after gastric digestion (84% of samples correctly classified; Figure 4c).In terms of farm differentiation, discrimination was slightly poorer.After gastric digestion, milk samples from farm II separated from the rest of the samples (Figure 4a), while after intestinal digestion, samples from farm I were mainly discriminated from the other 2 farms (Figure 4b).
Among all peptides released after gastric and intestinal digestion, precursors of 70 different bioactive peptides were identified using the MBPDB database (Supplementary material 2).A total of 35 peptides formed after gastric (21 peptides) and intestinal (14 peptides) digestion of mare milk β-casein were found to be precursors of VAPFPQPVVP (fragment f(191-200)), a bioactive pep-tide previously reported in donkey milk (Bidasolo et al., 2012).Interestingly, 2 of those precursors were significantly affected by farm or lactation stage.Specifically, the peptide KVAPFPQPVVPYPQRDTPVQ, released after gastric digestion, was significantly (P ≤ 0.05) affected by lactation stage (highest relative abundance when milk was from late lactation), whereas the peptide VAPFPQPVVPYPQ, released after intestinal digestion, was significantly affected by farm (highest relative abundance when milk was from farm I).

DISCUSSION
The present study addressed the performance of mare milk proteins during human digestion using the simulated in vitro INFOGEST gastrointestinal digestion model.Total digestibility of mare milk proteins was estimated as 93.7 or 96.0%depending on the method used (by considering individual amino acids or total amino groups, respectively), which was similar to cow milk (around 95%; Dupont and Tomé, 2020).This highlights mare milk as one of the most digestible protein sources.Digestibility of mare milk samples significantly decreased during lactation only when measuring total amino groups.In the analysis of total amino groups, a high variability among samples at late lactation, which probably came from this being a less specific method than the analysis of individual amino acids, could have led to an overestimation of the lactation effect on the digestibility of mare milk proteins.This is consistent with a non-significant effect of lactation stage on total digestibility considering individual amino acids, as well as with the results on peptide release described later in this section, which show that lactation had no significant effect on major peptide release after intestinal digestion (which is indirectly related to digestibility).
The DIAAR values, which express the ratio of digested essential amino acids relative to daily requirements, were affected by lactation stage and farm.These differences were due to changes in the amino acid content of undigested mare milk, and not to changes in digestibility of individual amino acids.Forage feeding, and therefore grazing, can affect the amino acid composition of milk.In this sense, it has previously been reported that increasing forage intake of lactating donkeys (also equids) can change the levels in milk of threonine, phenylalanine, ly-   sine and serine (Liang et al., 2022).In the present study, animal management affected the content of methionine, tryptophan, lysine and phenylalanine, as well as the evolution pattern of lysine and leucine during lactation.These changes could also derive from differences in the composition and/or quality of the feedstuffs used in the farms.
In mare milk, the limiting amino acids were threonine for infants, and histidine for children and adults.On the contrary, it has been reported that the limiting amino acids in cow milk are threonine and tryptophan for infants, and sulfur-containing amino acids for children and adults, while DIAAR values for histidine are quite high in cow milk (Mathai et al., 2017;Walther et al., 2022).Based on this, mare milk seems to be a better source of sulfur-containing amino acids but a poorer source of histidine than cow milk for children and adults, whereas both milk types are a good source of lysine, histidine and valine for infants (Mathai et al., 2017).
All proteins in mare milk were degraded after gastrointestinal digestion to a greater or lesser extent.However, degradation of caseins was much more intense than that of whey proteins, in agreement with previous studies (Xiao et al., 2023).In fact, the whey proteins β-lactoglobulin II and lysozyme C were resistant to gastrointestinal proteolysis.The different digestion behavior of caseins and whey proteins might be due to caseins forming a clot in the stomach induced by acidic conditions and pepsin, and therefore retaining longer in the stomach.On the other hand, whey proteins are soluble in the gastric fluid and pass through the digestion system faster and with lower proteolysis (Ye et al., 2016b).However, in vitro studies, in which all proteins spend the same time at each digestion phase, are comparable to in vivo studies in terms of protein digestion (Egger et al., 2017), suggesting that other factors, such as protein structure, also contribute to these differences.In fact, mobile and loosely structured proteins like caseins are more susceptible to the action of pepsin than proteins with a globular structure, like whey proteins (Dupont and Tomé, 2020).Caseins have been described to be almost completely degraded during the gastric phase (Inglingstad et al., 2010;Ye et al., 2016b;Egger et al., 2017;Xiao et al., 2023), whereas equine lysozyme is resistant to gastrointestinal digestion (Inglingstad et al., 2010).In this respect, equine lysozyme has shown resistance to acidic conditions (Jauregui-Adell, 1975), and being a c-type lysozyme, its calciumbinding site confers stability against protease digestion (Kuroki et al., 1989).Resistance of β-lactoglobulin to proteolysis with pepsin has previously been reported in bovine milk (Ye et al., 2016b;Egger et al., 2017;Liu et al., 2019).However, Inglingstad et al. (2010)  in the digestion behavior of the 2 β-lactoglobulin isoforms in mare milk, which in the study by Inglingstad et al. (2010) could have been hindered by considering the 2 β-lactoglobulin isoforms together.In fact, horse β-lactoglobulin I and II isoforms present a homology of only 70%, with 48 amino acids being exchanged (Conti et al., 1984;Godovac-Zimmermann et al., 1985).This might have detrimental implications on its structure and interaction with proteases.Degradation of α-lactalbumin during gastrointestinal digestion was evident in the present study, although literature discloses conflicting results regarding digestibility (or resistance) of α-lactalbumin (Dupont and Tomé, 2020).
Comparing gastric and intestinal digestion of mare milk proteins, slightly more but also larger peptides were formed during gastric digestion, which were further degraded into smaller peptides or sequences of 5 or less amino acids (this is a limitation of MS identification using the Mascot algorithm).The main source of peptides was β-casein, which has been reported as the main (Miranda et al., 2004) and highly digestible (Xiao et al., 2023) protein in mare milk.Although, to the best of our knowledge, the native mare milk peptidome has not been characterized yet, the majority of native peptides in human milk are derived from β-casein (Dallas et al., 2013), so endogenous peptides formed in the mammary gland could have in part contributed to the high amount of β-casein fragments found after gastric digestion of mare milk.However, further research that addresses the performance of endogenous peptides during digestion of mare milk is needed.In opposition, the low peptide formation from κ-casein observed in the present work could be a consequence of its low abundance in mare milk (Miranda et al., 2004), although the protein was already highly degraded in the gastric simulation.As known, pepsin rapidly hydrolyzes κ-casein into paraκ-casein destabilizing casein micelles and facilitating coagulation.Subsequently, proteolysis continues on the formed clots and on liberated para-κ-casein (Ye et al., 2016b).On the other hand, the low amino acid counts observed in lysozyme C were derived from its relatively low abundance in mare milk (Miranda et al., 2004) and its resistance to digestion.
Lactation stage and management of the mares also affected peptide release during digestion.As mentioned, a higher number of peptides were released from β-casein since it is the major mare milk protein, and therefore, more peptides derived from β-casein than from any other protein were affected by either farm or lactation stage.These differences were more evident for gastric than for intestinal peptides, probably because of the higher degree of proteolysis at intestinal level for proteins that are susceptible to gastrointestinal digestion.This is particularly relevant considering that the small intestine is the site of peptide absorption.Overall, mare milk samples from late lactation and low grazing activity showed the highest peptide release, which suggested a more efficient protein digestion of these samples.According to Egger et al. (2017), milk fat content might affect protein digestion, so the differences in peptide release from mare milk samples observed among farms and during lactation could also be attributed to a different sample fat content.However, some studies in cow milk reported that although milk fat globules get trapped into casein clots, the structure and digestion traits of the clot are not altered (Ye et al., 2016a).On the other hand, it has also been reported that the mineral composition of milk can affect protein digestion (Liu et al., 2019).The observation was that a reduced phosphorylation improved casein digestibility and proteolysis, and consequently raised peptide release during gastrointestinal digestion (Girardet et al., 2006;Liu et al., 2019).In this regard, the phosphorus content of the milk samples from the present study halved from early to late lactation (Blanco-Doval et al., 2023) and this could be correlated with a lower casein phosphorylation of the milk samples and consequent higher digestibility and peptide release at late lactation.
Milk has been shown to be a great source of bioactive peptides (Ning et al., 2022).However, information regarding bioactive peptides released from minor dairy species is limited up to date, with little knowledge on mare milk peptides (Guha et al., 2021).In fact, to the best of our knowledge, bioactive peptides released exclusively during gastrointestinal digestion of raw mare milk have not been reported yet.In the present study, no bioactive peptides (≥5 amino acids) were found in mare milk after gastrointestinal digestion, while several precursors of bioactive peptides were identified.Most of these precursors encoded bioactive sequences of 2-3 amino acids previously identified in bovine milk.Interestingly, some precursors of the bioactive peptide VAPFPQPVVP (fragment f(191-200) of β-casein) were found.This peptide was also reported in donkey milk after in vitro simulated gastrointestinal digestion, and was characterized as presenting angiotensin converting enzyme inhibitory activity (Bidasolo et al., 2012).Although digested mare milk samples in the present study did not contain this bioactive peptide per se, it should be highlighted that the static in vitro model used lacks brush border enzymes from the small intestine, which further fragment peptides that reach brush border cells before absorption (Sousa et al., 2023).Therefore, the bioactive peptide precursors formed during gastrointestinal digestion could potentially lead to the formation of absorbable bioactive peptides.In addition, the results obtained in this work suggested that management of the mares and lactation stage could influence the release of bioactive peptides from milk, being a topic worth exploring.These results demonstrated that some regions of mare milk proteins were resistant to gastric and/or intestinal digestion.Peptides resulting from these digestion resistant sites are more likely to be absorbed in the small intestine without further degradation (Egger et al., 2017), so they could be interesting regions to search for potential bioactive peptides in future research.
The present study provided a thorough description of the human gastrointestinal digestion of main proteins present in raw mare milk, but processed (homogenized or heated, for instance) mare milk samples were not considered, being this a limitation of the research.Raw mare milk usually has a high biological quality due to low microbiological and somatic cell counts (Danków et al., 2006), which makes it adequate for human consumption (Regulation (EC) No 853/2004).However, milk is a perishable product, and heating treatments like pasteurization or sterilization might be required to increase milk shelf life for commercialization.Heat treatments could alter milk proteins at molecular, microstructural and macrostructural level (protein denaturation, aggregation, crosslinking…), which may influence their digestive kinetics and digestibility (Li et al., 2021).Additionally, homogenization decreases the size of milk fat globules, and in this process, caseins and whey proteins are adsorbed into the surface of fat globules, altering clot formation during digestion (Ye et al., 2017).Considering differences in protein composition among cow and mare milk, further research is necessary to better understand the impact of mare milk processing on protein digestion traits.
Furthermore, recent research found that the freezethaw process enhanced the aggregation of (human) milk fat globules and proteins, and therefore hindered the hydrolysis of milk lipids and proteins during digestion due to physical impediment.However, milk frozen at low temperature (−60°C) and thawed at high temperature (45°C) performed similarly to fresh milk during digestion (Zhang et al., 2022).Considering this, freezing conditions in the present study (−80°C) probably preserved the protein digestion characteristics of raw mare milk, but a thawing process at low temperature (4°C) could have slightly decreased total protein digestibility.Anyway, since all mare milk samples were treated equally during sample preparation, potential changes in milk proteins occurring at the freezing-thawing step should not have interfered with the effect of lactation stage and management system observed in the present study.Nevertheless, the effect of the freeze-thaw process on mare milk remains to be further studied.

CONCLUSIONS
The high content of digestible proteins of mare milk makes it one of the most digestible dietary proteins.As far as we know, this is the first study to report different digestion behaviors of I and II isoforms of β-lactoglobulin in mare milk, deepening into the understanding of this protein source.Additionally, degradation of α-lactalbumin A was demonstrated, bringing some clarity to discrepancies among literature studies.Lactation stage and management system significantly affected some of the parameters related to protein digestion determined after gastrointestinal digestion of mare milk using the INFO-GEST method, i.e., amino acid profile of raw mare milk, DIAAR values and peptide release.Therefore, behavior of mare milk proteins during gastrointestinal digestion could be modulated through diet or optimized by selecting milk from specific lactation stages, which is definitely a topic worth exploring.Further research remains essential to elucidate the role that human digestion has on the digestion and functionality of mare milk proteins.
Supplementary material 2. List of peptides liberated after digestion of mare milk samples that were identified as bioactive peptide precursors for containing a bioactive sequence previously described in milk from mammal species other than horses, according to the Milk Bioactive Peptide Database (MBPDB).

Notes
Ana Blanco-Doval acknowledges the Department of Education of the Basque Government for the predoctoral fellowship (PRE_2022_2_0078) and mobility funding (EP_2023_1_0037).The authors thank the commercial farms that participated in the study.This work was supported by the Basque Government through Biotasma (Elkartek, 2019), Behor Esne (Cooperation, 2020) and IT944-16 and IT1568-22 projects.The authors have not stated any conflicts of interest.This is a commercial study and, therefore, institutional animal use approval was not required.
Supplementary material Available at http: / / hdl .handle.net/10810/ 66250.Supplementary material 1. Supplementary data regarding enzyme characteristics and activity, chemical composition of simulated salivary, gastric and intestinal fluids, calculations for protein digestibility, and digestibility of individual amino acids.
Blanco-Doval et al.: Mare milk digestibility and peptide release 2.1 mm internal diameter, 1.7 μm particle size; Waters) operated as described in the International Organization for Standardization/Draft International Standard 13904 (2014).Analytical conditions were 2 μL injection volume, 50°C column temperature, 285/340 nm excitation/ emission wavelengths, and 0.4 mL/min flow rate.The mobile phase consisted of an isocratic flow of 99.9% (vol/vol) AccQ-Tag Eluent A (diluted to 15% in ultrapure water, vol/vol; Waters) and 0.1% (vol/vol) formic acid (2%, vol/vol) in acetonitrile.Total nitrogen content of raw mare milk pooled samples was analyzed by Kjeldahl method (International Organization for Standardization Blanco-Doval et al.: Mare milk digestibility and peptide release contrast, β-lactoglobulin II and lysozyme C exhibited low proteolysis throughout the entire gastrointestinal digestion process.Peptides from the 175-186 region of αs1-casein, the 133-145 region of αs2-casein, the 18-26 region of κ-casein, the 156-167 region of β-lactoglobulin II, and the 57-63 region of lysozyme C were resistant to gastric digestion and were mainly formed during intestinal digestion (Figure Blanco-Doval et al.: Mare milk digestibility and peptide release

Figure 2 .
Figure 2. Size mass of the peptides released after gastric and intestinal digestion classified according to the protein of origin.Letters (a,b) refer to statistically significant (P ≤ 0.05) differences between gastric and intestinal digestion from the same parent protein.

Figure 3 .
Figure 3. Heatmaps showing a qualitative description of the behavior of peptide patterns at the end of gastric (upper half of the panels) and intestinal digestion (lower half of the panels), classified according to farm or lactation stage.Intensities are expressed as colors that range from purple (low intensity) to red (high intensity), and represent the mean frequency of each amino acid identified as part of the protein sequence.Numbers in x-axis indicate the position of the amino acid in the protein sequence, and the y-axis shows group classification according to lactation stage (left panels; early, mid and late lactation) or farm (right panels; farms I, II and III).

Figure 3 (
Figure 3 (Continued).Heatmaps showing a qualitative description of the behavior of peptide patterns at the end of gastric (upper half of the panels) and intestinal digestion (lower half of the panels), classified according to farm or lactation stage.Intensities are expressed as colors that range from purple (low intensity) to red (high intensity), and represent the mean frequency of each amino acid identified as part of the protein sequence.Numbers in x-axis indicate the position of the amino acid in the protein sequence, and the y-axis shows group classification according to lactation stage (left panels; early, mid and late lactation) or farm (right panels; farms I, II and III).
Blanco-Doval et al.: Mare milk digestibility and peptide release Table 3. Peptides identified in mare milk samples after gastric and intestinal digestion significantly (P ≤ 0.05) affected by farm and/or lactation stage.Peptide sequence and protein of origin are
Blanco-Doval et al.: Mare milk digestibility and peptide release Blanco-Doval et al.: Mare milk digestibility and peptide release

Table 1 .
Statistical significance 1 (P-values) of farm and lactation stage effects on individual amino acid (AA) content in undigested raw mare milk, AA digestibility and DIAAR values