Label-free-based proteomics analysis reveals differential proteins of sheep, goat and cow milk

Regarding the limited information on species protein differences between sheep, goat, and cow milk, the differentially expressed proteins in sheep, goat, and cow milk and their functional differences are analyzed using label-free proteomics technology to identify potential biomarkers. 770 proteins and 2914 peptide segments were identified. The statistical analysis showed significant differences in the relative abundances of the 74 proteins among the sheep, goat, and cow milk. CSN3 and LALBA can be used as potential biomarkers for goat milk, XDH can be used as potential biomarkers for cow milk, and CTSB and BPIFB1 can be used as potential biomarkers for sheep milk. The functional analysis us-ing Gene Ontology and Kyoto Encyclopedia of Genes and Genomes showed that these significantly different proteins were enriched by different pathways including thyroid hormone synthesis and glycerol phospholipid metabolism. The data revealed differences in the amounts and physiological functions of the milk proteins of different species, which may provide an important basis for research on the nutritional composition of dairy products and adulteration identification technology.


INTRODUCTION
Dairy products are among the most important mammalian food sources because they provide the energy and nutrients required to ensure normal growth and development after birth (Rizzoli et al., 2022;Komada et al., 2020;Pereira, 2014).Cow milk occupies a large market share worldwide and is an important protein source in the human diet (Lucey et al., 2017); however, the global incidence of lactose intolerance globally when consuming cow milk is high, ranging from 57% to 65% (Catanzaro et al., 2021).Studies have shown that sheep and goat milk have better lactose tolerance compared with cow milk for low β-lactoglobulin content, low sensitization (Mirzaei et al., 2022;Flis et al., 2021).The content of β-lactoglobulin is the lowest in sheep milk (0.15 g/100 mL), 0.28 g/100 mL in goat milk, and the highest in cow milk (0.40 g/100 mL).Moreover, the nutritional content of sheep and goat milk is relatively high, the consumption of sheep and goat milk products continues to rise and the market potential is large (Pulina et al., 2018).There are about 200 million sheep and goats in the world, 20.8% of which are used in dairy production.Significantly, sheep milk is richer in dry matter and more beneficial to human health than either goat or cow milk (Abdulwahid, 2022).It is often used to produce cheese and other products (Inácio et al., 2022).Meantime, Sheep and goat milk is richer in nutrition but lower in yield, so the price is higher than cow milk, which will cause adulteration problems (Windarsih et al., 2021).It is a common practice to adulterate sheep and goat milk with cow milk (Masci et al., 2022).The basis of detecting sheep and goat milk adulteration is the composition difference of cow milk.Therefore, a comprehensive analysis of the composition of sheep, goat, and cow milk can improve the understanding of their nutritional differences between them, and establish a theoretical basis for the development of detection methods for the adulteration of milk.
Proteins in dairy mainly include casein, whey protein, and a small amount of fat globule membrane proteins (Agarwal et al., 2015) and play a key role in the growth and development of newborn babies (Nagpal et al., 2011).Protein is the main source of nutrition in dairy products (Borkent et al., 2019).Protein in milk plays a key role in intestinal digestion and fat synthesis (Kashyap et al., 2021;Wang et al., 2019), and has a series of biological functions.Therefore, more and more attention has been paid to the differences of proteins in milk of different species.
The study of proteomics can provide more information about life.In recent years, label-free proteomics technology has become an important quantitative method for proteomics.The principle of label-free is to analyze the quantity changes of proteins in samples from different sources by comparing the times of mass spectrometry or the intensity of mass spectrum peaks.It is believed that the frequency of peptide captured and detected in mass spectrometry is positively correlated with its abundance in the mixture.Therefore, the count of protein detected by mass spectrometry reflects the abundance of protein.
The protein can be quantified by associating the mass spectrometry detection count with the amount of protein through appropriate mathematical formulas (Wiśniewski, 2017).Label-free proteomics technology does not require label processing.It is simple to operate and it has low cost.In addition, it has high throughput and can improve reproducibility.However, label-free proteomics technology also has certain limitations.Its quantitative stability and detection sensitivity need to be improved (Neilson et al., 2011).
The proteomics of different milk species have been widely studied and applied, most of which are related to human and cow milk (Yu et al., 2019).In a recent study, bioactive peptide profiles were obtained using high-resolution mass spectrometry, matrix-assisted laser ionisation and desorption (MALDI), and a time-of-flight mass analyzer (TOF) (Ramon et al., 2023).The results suggest that ohmic heating is a promising technology in the development of flavored milk drinks.Xia et al. (2024) investigated the distribution of whey protein in human milk among mothers with infants of different genders and to identify any differences in protein abundance using the cutting-edge, efficient 4D label-free proteomics technique.In addition, proteomic studies on goat milk have also been reported.Ji et al. (2023) evaluated goat milk adulteration using label-free quantification proteomic technology and found casein, B2M, and SCGB1D to be potential biomarkers.These studies have provided valuable information on the protein composition of the milk from different mammals.However, these studiesanalyzed a specific protein of the same species, but the overall protein information of sheep, goats, and cow milk using label-free proteomics techniques to compare has not been studied.
To further explore the differences in the protein composition of sheep, goat, and cow milk, this study combined label-free quantitative proteomics with bioinformatic methods to comprehensively analyze differentially expressed milk proteins.Exploring the function and enrichment pathway of differential proteins is an effective way to further analyze proteins.The potential value can be understood through bioinformatics analysis.Gene Ontology (GO) is a widely used bioinformatics resource for describing biological processes, molecular functions, and cellular components of proteins ; Kyoto Encyclopedia of Genes and Genomes (KEGG) can accurately annotate and enrich the pathways of differential proteins (Zhao et al., 2023;Chen et al.,2020).Important proteins and related pathways were screened via gene ontology (GO) functional annotation, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment, protein domain, and subcellular localization analyses.Our findings may facilitate the development of methods to identify adulteration and potential biomarkers in sheep, goat, and cow milk.It provides a theoretical basis for solving the adulteration problem of dairy industry and formulating industry standards, and promotes the rapid and healthy development of dairy industry.

Sample collection
Sheep milk samples were collected from the second litter of sub-purebred East Frierian sheep at Yuansheng Ranch in Jinchang City (Gansu Province, China).Goat and cow milk samples were collected from Holstein cows and Saneng dairy goats, respectively, at the Animal Husbandry Experimental Farm of Northwest A&F University.Twenty-four samples were collected from each species, all of which were milk samples (normal milk) from d 35 after lambing.The collected milk samples from each species of 24 individuals were pooled randomly into 6 composite samples, each containing milk from 4 individuals and each composite sample was used as one biological repeat.A milk sample (50 mL) was collected in a sterile centrifuge tube, thoroughly stirred, and then stored in dry ice for transfer to the laboratory where it was stored in a refrigerator at −80°C before subsequent analysis.

Experimental methods
First, proteins were extracted by SDT cleavage method, samples were prepared by SDS-PAGE and enzymolysis, and then identified by mass spectrometry.Finally, functional analysis of the identified proteins was performed by GO, KEGG, domain and annotation analysis, and subcellular localization.

Sample preparation
SDT buffer (4% SDS, 100 mM Tris-HCl, pH 7.6) was added to each milk sample.The lysate was sonicated and then boiled (100°C) for 15 min.After centrifugation at 14000 g for 40 min, the supernatant was quantified using a BCA protein assay kit.The samples were stored at −20°C (Zhu et al., 2014).

SDS-PAGE Separation
Protein from each sample (20 mg) was mixed with loading buffer (120 mg) and boiled (100°C) for 5 min before the proteins were separated using 12.5% SDS-PAGE gel.Prepare separation glue and concentrated glue.The samples were successively added into the gel hole with a pipette gun, and the loading amount was 20 μL.The protein marker was added directly and the sample size was 5 μL.The concentrated glue voltage is 80 V, the separation glue voltage is 160 V, until the bromophenol blue front light reaches the lower end of the electrophoresis tank.
The protein bands were visualized via Coomassie Blue R-250 staining.Fixed with fixing solution (anhydrous ethanol 500 mL+ phosphoric acid 80 mL+ dd H 2 O 420 mL) for 30 min, rinsed with dd H 2 O for 3 times; Coomassie brilliant Blue dyeing solution (1.0g R-250+ anhydrous ethanol 250 mL+ phosphoric acid 100 mL+ dd H 2 O 650 mL) was dyed overnight; Rinse the gel with dd H 2 O for 3 times and decolorize with fixing solution until the gel background color is clear.

Filter-Aided sample preparation (FASP)
To a protein solution for each sample (80 μg) was added DTT to a final concentration of 100 mM, boil water for 5 min, and cool to room temperature (25°C).UA buffer (200 μL, 8 M Urea, 150 mM Tris-HCl, pH 8.5) was added and thoroughly mixed.Transfer into a 30kD ultrafiltration centrifuge tube (Sartorius, VN01H22), centrifuge 12500g for 15min, and discard the filtrate.IAA buffer (100 μL, 100 mm IAA in UA) was added and the sample was shaken at 600 rpm for 1 min before the reaction was performed at room temperature (25°C) ithout light for 30 min, after which the product mixture was centrifuged at 12500 g for 15 min.UA buffer (100 μL) was added and the sample centrifuged at 12500 g for 15 min.This step was repeated twice.NH 4 HCO 3 solution (100 μL, 40 mM) was then added and the sample was centrifuged at 12500 g for 15 min.This step was repeated twice.The collection tube was replaced and Trypsin buffer (4 μg Trypsin in 40 μL 40 mM NH 4 HCO 3 solution) was added before the sample was oscillated at 600 rpm for 1 min and stored at 37°C for 16-18 h.The solution was then centrifuged at 12500 g for 15 min before NH 4 HCO 3 (20 μL, 40 mM) was added.The mixture was further centrifuged at 12500 g for 15 min and the filtrate was collected.The peptide was desorted using a C18 Cartridge, lyophilized, dissolved in formic acid solution (40 μL, 0.1%), and quantified (OD280) (Wiśniewski et al., 2009).

Mass spectrometry
Each sample was separated using the nanoliter flow rate Easy nLC system.Buffer solution A is an aqueous solution of formic acid (0.1%) and solution B is a mixture of this formic acid solution and aqueous acetonitrile solution (20%: 80%).The chromatographic column was balanced with 100% buffer solution A, and the samples were separated by automatic injection into an Acclaim PepMap RSLC analytical column (Thermo Fisher Scientific, 50 μm × 15 cm, nano viper, P/N164943) at a flow rate of 300 nL/min.The liquid phase adopts a 2-h gradient: 0 min-5 min, B liquid 3%; From 5 min to 95 min, the linear gradient of liquid B was from 3% -28%.From 95 min to 110 min, the linear gradient of liquid B was from 28% -38%.From 110 min to 115 min, the linear gradient of liquid B was from 38% to 100%.From 115 min to 120 min, liquid B remained at 100%.The separated fractions were then analyzed by mass spectrometry (QExactive HF-X mass spectrometer, Thermo Fisher Scientific, USA) in the positive ion mode.The analysis time, scanning range of the parention, primary resolution, and primary maximum IT were 120/180 min, 350-1800 m/z, 70,000, and 50 ms, respectively.After each full scan, 10 fragment scans (MS2 scan) were collected to obtain the mass charge ratios of polypeptides and polypeptide fragments.The MS2 activation type, isolation window, secondary resolution, and number of microscans were HCD, 2 m/z, 17,500, and 1, respectively, while the level 2 maximum IT is 45 ms, and normalized collision energy was 27 eV (Duan et al., 2022).

Bioinformatics analysis
Cluster analysis.In the analysis of the significance difference of the quantitative results, we first screened the non-null data in the sample group for statistical analysis.Proteins with an expression difference multiple greater than 2.0 times (up and down) and P-value (significance Zhu et al.: Label-free-based… A) less than 0.05 were considered significantly differentially expressed proteins.The quantitative data for the target protein set was normalized.Matplotlib software was used to simultaneously classify the expression of samples and proteins (distance algorithm: Euclidean, linkage method: average linkage) and generate hierarchical clustering heat maps.
GO and KEGG Pathway Annotation.NCBI Blast (https: / / blast .ncbi.nlm.nih.gov)was used on a Linux server to compare the target protein collection to the appropriate database.The top 10 alignment sequences with E-value < = 1e-3 were retained for subsequent analysis.Blast2GO Command Line was used to extract the target protein set (https: / / www .geneontology.org)(Götz et al., 2008).During Annotation, the Blast2GO Command Line annotated the GO entry to the target protein sequence by considering similarity.To further improve annotation efficiency, InterProScan (Quevillon et al., 2009) could be used to search conserved motifs matching target proteins in the EBI database, and annotated motif-related functional information to target protein sequences.
When annotating the KEGG pathway of the target protein set, KOALA (KEGG Orthology And Links Annotation) software (Kanehisa et al., 2016) was used to first KO classify the target protein sequence by comparing with the KEGG GENES database.The pathway information of the target protein sequence is automatically obtained according to KO classification.The distribution of each GO classification or KEGG pathway in the target and the overall protein sets was compared using Fisher's exact test to analyze the enrichment of the GO and KEGG pathway annotations of the target protein set.
Domain annotation and enrichment analysis.The Interpro database (http: / / www .ebi.ac.uk)was used to conduct a functional domain enrichment analysis of differentially expressed proteins, and the distribution of differentially expressed proteins in the total protein set was compared using Fischer's exact test (P < 0.05) to evaluate the significance of enrichment of a certain functional domain.
Subcellular localization analysis.The protein subcellular localization prediction software WoLF PSORT (https: / / wolfpsort .hgc.jp/ ) was used to convert protein sequences into digital localization features based on sorting signals, amino acid composition, and functional motifs (Horton et al., 2007).The K-nearest neighbor classifier was used to predict the subcellular localization of the protein (Abu Alfeilat et al., 2019).

Data analysis.
Data were the mean value of 6 biological repeats of milk for each species and were analyzed with SPSS.The differentially expressed proteins of each 2 species were analyzed via Student's t-test, and the significant level is 0.05.For the qualitative and quantitative analysis parameters of proteins, this study used MaxQuant software (version 1.6.14.0,Germany) for the database search and label-free quantitation (LFQ) algorithm for quantitative analysis (Cox et al., 2014).The UniProt database (http: / / www .uniprot.org)was used for this study.The visualization of all data in this work was done with GraphPad Prism 8.

Identification of peptides and proteins
To verify the reliability of the experiment, 2914 peptide chains were first identified using MS, and it was found that more than 80% of the peptide segments scored more than 60 points (Chepanoske et al., 2021) (Figure 1 A), indicating the high quality of the test data and establishing the foundation for subsequent research.Among them, the number of amino acids in the identified peptide segment ranges from 6 to 31 (Figure 1 D).
In this study, 770 proteins were identified in sheep, goat, and cow milk.The relative molecular weights of the identified proteins were between 0 and 100, most of which were between 10 and 60, and more than half of the proteins had fewer than 6 peptides (Figure 1 B, C).The Venn diagram in Figure 2 shows that 394, 436, and 301 proteins were identified in sheep, goat, and cow milk respectively.Specifically, 120 proteins were co-expressed in all 3 milk samples, 203 proteins were co-expressed in sheep and goat milk, and 135 proteins were co-expressed in sheep and cow milk (Figure 2).These are differences brought about by different species.These results indicated that there were differences in the proteins of sheep, goat and cow milk, which established the basis for further investigation of the functional differences of proteins.
Milk protein variations between different species.To investigate the specific protein differences between sheep, goat and cow milk, marked differentially expressed proteins in milk of different species were determined by using label-free proteomics analysis.As shown in the clustering results in Figure 3, 2 main clusters of the different proteins can be observed in the heat map, and the patterns of the sheep and goat samples differ from those of the cow samples.CSN3, LALBA, XDH, CTSB and BPIFB1 were the most abundant proteins; however, their proportions differed in sheep, goat, and cow milk.We identified 23 differentially expressed protein genes between the sheep and goat samples (9 upregulated and 14 downregulated in the sheep milk) (Figure 3 A, D) and 21 differentially expressed protein genes between sheep and cattle samples (13 upregulated and 8 downregulated in the sheep milk) (Figure 3 B, E).Eighteen differentially Zhu et al.: Label-free-based… expressed protein genes were found between the goat and cow samples (9 upregulated and 9 downregulated in goat milk) (Figure 3 C, F).These results will lay a good foundation for the better study of the nutritional and functional differences of protein in sheep, goat, and cow milk.
GO analysis of identified proteins from different species.To further explore the function of differentially expressed protein genes in different milk, we conducted GO enrichment.The GO analysis results of the significantly different proteins identified in the sheep, goat, and cow milk samples are shown in Figure 4.The results indicate that all proteins were classified into Biological Processes (BP), Cellular Component (CC), and Molecular Functions (MF).GO enrichment analysis showed that the proteins identified in sheep and goat milk were significantly enriched in mucosal innate immune response, in the host body, or in negative regulation of single biofilm formation in the host organism in BP; cells and high-density lipoprotein particles in CC; and chemotaxic activity and enterobactericin binding in MF.The identified proteins in the sheep and cow milk were significantly enriched in vitamin B6 metabolic process and dehydroepiandrosterone reaction in BP, Golgi body cavity and xanthine dehydrogenase complex in CC, and FAD-binding and iron-ion binding in MF.The proteins identified in the goat and cow milk were significantly enriched in reaction to 11-deoxycorticosterone and dehydroepiandrosterone reaction in BP, Golgi coelom and cell solute in CC, and FAD-binding and protein disulfide  isomerase in MF.Therefore, we speculate that the differentially expressed proteins play a role in immunity, vitamin metabolism, and lipid metabolism.
KEGG and domain analysis of identified proteins from different species.To further understand the function of significantly differentially expressed proteins, we performed KEGG pathway analysis and protein domain and annotation analysis.We found that these proteins were mainly involved in the KEGG pathway.Thyroid hormone synthesis, fatty acid elongation, and lysosomes were the main processes in sheep and goat milk (Figure 5 A).Caffeine metabolism, drug metabolism of other enzymes, glycerophospholipid metabolism, lysosome, and purine metabolism were the main processes in sheep and cow milk We found that these proteins were mainly involved in the KEGG pathway (Figure 5 B).But there was no significant enrichment of the metabolic pathways in goat and cow milk, this may be due to their specific differentially expressed proteins.
Proteins are mainly composed of domains, which are the basic units of protein composition, performance, and evolution.An in-depth study of the chemical domains of proteins is of great value for understanding the biological function and evolutionary processes of proteins.Therefore, we performed domain and annotation analysis (TOP 10) of the significantly differentially expressed proteins in 3 different milks (Figure 5 C -E).In sheep and goat milk, the lipocalin signature, lipocalin family conserved sites, serum amyloid A protein, and lactotransferrin were mainly enriched.In sheep and cow milk, the lipocalin signature and lipocalin family conserved sites were enriched.In goat and cow milk the CAP superfamily, Allergen V5/Tpx-1-related, conserved site, cysteine-rich secretory protein, and allergen V5/Tpx-1-related to the CAP domain were mainly enriched.These results will provide ideas for studying the nutritional and functional differences of different milks.
Subcellular localization analysis of identified proteins from different species.Since GO functional annotation analysis revealed that many significantly differentially expressed proteins were involved in the CC, subcellular localization analysis was performed on 74 significantly differentially expressed proteins; pie charts were used to show the number and distribution ratio of the significantly differentially expressed proteins in each sub-organelle.Figure 6 shows that in sheep and goat milk, 14.8% of the proteins are located in the cytoplasm, 7.4% in the endoplasmic reticulum, and 11.1% in the mitochondria.In sheep and cow milk, 3.8% of the proteins are located in the cytoplasm, 3.8% in the nucleus, and 3.8% in mitochondria.In goat and cow milk, 14.3% of the proteins are located in the cytoplasm, whereas 4.8% are located in the endoplasmic reticulum, mitochondria, and nucleus.In cells, the largest proportion of these dif-ferentially expressed proteins are located in the cytosol, so they may be involved in signaling, post-transcriptional regulation, translation, and post-translational modification as functional proteins.

DISCUSSION
Although several studies have been reported on the proteomics of mammalian milk (Dayon et al., 2021;Dingess et al., 2021).In the present study, label-free pro-  teomics was used to identify and quantify 770 proteins in sheep, goat, and cow milk.This has significantly increased the number of proteins identified in sheep, goat, and cow milk, and helped deepen our understanding of the protein composition of the milk of different species.Di et al. ( 2014) used MALDI-TOF MS to detect adulteration in donkey and goat milks in a highly sensitive manner, with limits of up to 0.5%.A recent study was the first to use proteomics to conduct a comprehensive whey proteins analysis of hybrid dairy cattle, identifying more than 29 low-abundance proteins, while observing that chromosomes 5 and 9 express the largest number of proteins to understand their gene expression patterns and the functional value of whey proteins (Singh et al., 2023).However, they are limited to a specific species or a specific protein, so these data support the advantages of label-free proteomics.
Recent studies have found that differences in the proteins in different dairy types may bring different benefits (Farag et al., 2020;Sangild et al., 2021).Based on the label-free quantitative proteomics results, we found some similarities and differences in the protein composition of sheep, goat, and cow milk.There are many major proteins in different milk samples, such as CSN3, LALBA, XDH, CTSB and BPIFB1, but their content varies between milk species.The contents of CSN3 and LALBA in sheep and goat milk were significantly higher than those in cow milk (P < 0.01).Κ-casein (CSN3) is a relatively small protein that has been identified in previous studies as an important functional component in promoting health (Dhasmana et al., 2022).LALBA is a part of the lactose synthase complex required for lactose synthesis and plays a central role in dairy production (Permyakov et al., 2000), and some of its folded variants have recently been found to have bactericidal activity, some of which lead to tumor cell apoptosis (Chiou et al., 2021).In addition, LALBA can promote the accumulation of proteins and bioactive peptides in skeletal muscle and has probiotic and antibacterial properties (Layman et al., 2018).In addition, we found that the XDH content in cow milk was significantly higher than that in sheep and goat milk (P < 0.01).XDH is reported to be involved in a variety of physiological processes related to metabolism (Bortolotti et al., 2021) and plays an important role in the life activities of organisms.In addition, the CTSB content in sheep milk was significantly higher than that in goat and cow milk (P < 0.01).CTS is among the most important proteolytic enzymes in mammals, and studies have shown that cathepsin is closely related to the onset of diabetes (Ding et al., 2020).CTSB plays an important role in regulating the bioavailability of lysosomes and autophagosomes (Man et al., 2016).Similarly, the role of BPIFB1 in antimicrobial activity, tumor inhibition and respiratory diseases is becoming more and more obvious (Hu et al., 2023;Jiang et al., 2022).The protein encoded by BPIFB1, a secreted protein, is a newly discovered natural immune protective molecule with a fungicidal and penetration-enhancing protein domain function that can respond to external physical and chemical stimuli (Li et al., 2020).The expression of BPIFB1 in sheep milk was significantly higher than that in cow milk, suggesting that sheep milk may have a good immunoprophylaxis effect.The difference of these proteins is one of the key factors that lead to the difference of nutritional function of different varieties of dairy.
Subsequently, we analyzed the functions of the identified differential proteins.Based on the GO annotation analysis, the identified proteins were associated with multiple biological functions.The GO enrichment results (Figure 4) showed that the differentially expressed proteins in sheep milk were mainly enriched in mucous membranes, whereas the differentially expressed proteins in goat milk were mainly related to vitamin metabolism.An earlier study found the positive effects of milk whey protein supplementation on improving satiety and postprandial glycemic control in the short term (Giglio et al., 2023).In addition, Wang et al. (2018) have shown that milk protein is closely related to gastric digestion.In our study, additional different milk protein functions were discovered, revealing new insights into the protein composition of the mammals studied.
At the same time, we performed KEGG pathway and protein domain annotation analysis.As shown in Figure 5, the major KEGG pathways differed significantly among proteins from sheep, goat, and cow milk.In addition, we found that many proteins were involved in several diseases and metabolic pathways, which may be related to passive immunity.This is similar to previous studies on the function of milk proteins (Kowalczyk et al., 2022).Finally, we also conducted subcellular localization of the significantly different proteins, and found that the largest proportion of the proteins were located in the cytoplasm, indicating that these different proteins may be involved in intracellular transcription, signal transduction and other processes.
In summary, we found that sheep, goat and cow milk have great differences in protein and have different functions.These evidences indicate that sheep and goat milk have higher health care efficacy as milk sources for small ruminants.With the increase of consumption level, people's demand for dairy products with health benefits is increasing, and this study have established a theoretical basis for the expansion of sheep and goat milk market.This study provides a strong foundation for the development of methods to identify potential biomarkers and to study the specific mechanism of action and will also promote the development of technology to identify adulteration in milk.At the same time, with the promotion of high-quality milk sources such as sheep and goat milk, the dairy industry will enter faster and higher quality development.

CONCLUSION
The differentially expressed proteins in sheep, goat, and cow milk and their functional differences were analyzed using label-free proteomics technology to identify potential biomarkers.Label-free proteomics technology identified 770 proteins by comparing proteins isolated from sheep, goat, and cow milk.Quantitative differences in the identified proteins were analyzed using cluster analysis and volcanic mapping.While the composition of sheep, goat, and cow milk proteins have many similarities, the relative abundances of certain proteins varied significantly between species.CSN3 and LALBA can be used as potential biomarkers for goat milk, XDH can be used as potential biomarkers for cow milk, and CTSB and BPIFB1 can be used as potential biomarkers for sheep milk.In addition, analysis of the functions of significantly different proteins using GO and KEGG enhanced our understanding of the biological functions of the specific proteins of these species.These significantly different proteins were enriched by different pathways including thyroid hormone synthesis and glycerol phospholipid metabolism.Our findings reveal the composition and characteristics of proteins in sheep, goat, and cow milk, promote the development of specific milk adulteration methods, and provide potential directions for the production of specific active proteins in dairy.Future work in this area will screen specific proteins and genes for further functional verification.

Figure 1 .
Figure 1.Mass spectrometry result of sheep, goat and cow milk: (A) Score distribution of the peptide ions, (B) length distribution of the peptide sequence, (C) relative molecular mass distribution map of the proteins, and (D) Quantitative distribution map of the identified peptides.Notes: The column represents the number of peptides, and the curve represents the percentage of peptides or proteins.

Figure 2 .
Figure 2. Venn diagram of proteins from sheep, goat, and cow milk.

Figure 3 .
Figure 3. Heat and volcano map of the protein genes expressions in different types of milk for (A, D) sheep milk vs goat milk, (B, E) sheep milk vs cow milk, and (C, F) goat milk vs cow milk.Notes: Red and blue represent different levels of expression.
Figure 4. GO annotations of proteins in sheep, goat, and cow milk based on BP, CC, and MF: (A) sheep milk vs goat milk, (B) sheep milk vs cow milk, and (C) goat milk vs cow milk.Notes: Blue represents CC, green represents MF, and red represents BP.

Figure 5 .
Figure 5. KEGG pathway enrichment and domain analysis of the significantly differentially expressed proteins for: (A,C) sheep milk vs goat milk, (B,D) sheep milk vs cow milk, and (E) goat milk vs cow milk.

Figure 6 .
Figure 6.Subcellular localization analysis of the significantly differentially expressed proteins in different types of milk for: (A) sheep milk vs goat milk, (B) sheep milk vs cow milk, and (C) goat milk vs cow milk.