Comparative Label-Free Liquid Chromatography–Mass Spectrometry Milk Proteomic Profiles Highlight Putative Differences between the Autochthon Teramana and Saanen Goat Breeds

Simple Summary Milk is a valuable source of proteins and other nutrients. Changes in milk production toward systems that obtain increased milk yield have resulted in a change in milk composition in large and small ruminants, such as goats. A better characterisation of breeds that undergo a limited formal crossbreeding, such as the Teramana goat is pivotal in order to obtain useful data that could be applied for the management of the breed. Proteomic technologies have brought significant advances in the characteristion of new proteins. In this study, this technology was used to compare the proteome of the autochthon Teramana and Saanen breeds, which are commonly used by the industry to allow for the identification of a cohort of proteins that were able to discriminate the two goat breeds. Proteomics offered the potentiality for a deeper investigation of the biological differences of the breeds under study in a substrate easy to obtain. Abstract Goat’s milk is an excellent source of nutrients, with greater benefits compared to cow’s milk. Limited information is available on autochthon goat breeds, which are important for biodiversity preservation. In this study, the aim of using label-free quantification was to investigate the milk proteome of two goat breeds, the autochthon Teramana and Saanen breeds, which are commonly used by the industry. Utilising label-free proteomic analysis, 749 and 666 proteins, respectively were identified and quantified from the Teramana and Saanen goat milk. Moreover, utilising statistical analysis, 29 proteins were able to discriminate the two goat breeds, with many of the identified proteins involved in complement and coagulation cascades. This work enhances our understanding of the goat milk proteome and shows differences between the two breeds, leading to an important contribution toward a more detailed molecular-view of this unique substrate. Additionally, charactersation of the milk proteins can help in guiding genetic improvements in the goat herds, and thus increasing its use in human nutrition.


Introduction
Milk plays a pivotal role in human nutrition, as it is rich in macro-and micro-nutrients and is an important source of antimicrobial and immunoregulatory agents. Over the last decades, the use of non-bovine milk as an alternative nutrient source has increased, since hypersensitivity to cow's milk proteins remains one of the major causes of food allergies [1]. For this reason, the milk of small ruminants, such as goats is of particular

Sample Collection
Five Teramana and five Saanen goats for a total of 10 animals aged 30 to 40 months with no signs of acute mastitis or other clinical diseases were selected from a local farm in Teramo province, Italy. All goats included in this study were reared in the same farm, and the management with regard to their feeding, handling, and period of lactation was the same. Thereafter, approximately 100 mL of milk was collected by milking at one time point from each animal and were used in the study. All animals were representative of goat breeds. After sampling, all milk samples were transferred on ice, and then taken immediately to the laboratory, where they were processed. No permits were required for the described study, which complied with all relevant regulations, since only milk was collected and no animal sacrifice was necessary.

Sample Preparation
Goat's milk samples (2 mL) were centrifuged at 3000× g for 30 min, at 4 • C, and the fat layer was carefully removed. Then, 50 µL of the skim milk samples were precipitated by adding 5 µL of sodium acetate (1 N) and 5 µL of acetic acid (10%), and the caseins were removed by centrifugation at 3000× g for 30 min, at 4 • C. Whey proteins were directly precipitated with cold acetone, dried, and stored at −20 • C until further analysis.
The protein content of all samples was determined in triplicate using the Protein Assay Kit (Bio-Rad Labs, Hercules, CA, USA), following the Bradford method using a bovine serum albumin (BSA) standard [28]. Equal concentrations (100 µg) of all protein samples were used for filter-aided sample preparation (FASP) as described below [29].

Label-Free LC-MS Quantitative Profiling
Proteome Discoverer v.2.2 (Thermo Fisher Scientific) with the Sequest HT algorithm and Percolator was used to achieve the proteins identification. MS files were searched against the UniProtKB-SwissProt Capra hircus database (downloaded in February 2020 and containing 32,490 sequences). The parameters set for protein identification were essentially as described in detail in our previous study in the same species [10].
Progenesis QI for Proteomics (version 2.0; Nonlinear Dynamics, a Waters company, Newcastle upon Tyne, UK) was used for quantitative label-free data analysis as already described [10]. The steps for the alignment, normalisation which use ratiometric data in Animals 2023, 13, 2263 4 of 15 log space, along with a median and mean absolute deviation outlier filtering approach, calculation of peptide abundance were as recommended by nonlinear dynamics (Waters TM ; www.nonlinear.com) and as described in our previous study [10]. Only peptide ions with charge states +1, +2, and +3 were considered and re-imported back into Progenesis QI software for further analysis. Peptide identifications were imported into the Progenesis QI software and assigned to the matching features. Proteins were considered differentially expressed if they passed the following criteria: (1) ANOVA values with a cut off of p < 0.05, (2) proteins with ≥2 peptides matched, and (3) ≥1.5-fold difference in abundance.

Functional and Protein Network Analyses
Proteins identified in both groups were submitted to classification analysis using the PANTHER (Protein Analysis Through Evolutionary Relationships) database system, release 14.1 (http://www.pantherdb.org/) [31]. Default parameters were used to carry out the analysis for the categorisation into biological process.
Cytoscape (http://www.cytoscape.org/; accessed on 8 May 2020) [32] using the plugin ClueGO (http://www.ici.upmc.fr/cluego/; accessed on 8 May 2020) [33] was used for the functional interpretation of the differentially expressed proteins. Gene Ontology (GO) Biological Process (BP) branch (May 2020) was used for the gene enrichment analysis. The parameters used were as described in a previous study [34]. Due to the insufficient protein annotation for goat species, the analysis made use of Bos taurus specific functional annotations. Default parameters were used for the other parameters. GO:BP terms with a Benjamini-Hochberg corrected p-value < 0.05 were considered statistically over-represented. The ClueGO plug-in which integrates and the KEGG pathway database (May 2020) were also used to separately create functionally organised pathway term networks.
In silico protein-protein interaction (PPI) analysis of the proteins that were differentially expressed between goat milk breeds was carried out using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING v.11) database (https://string-db. org/; accessed on 22 November 2022) [35] as already described [34]. Bos taurus specific interactome was used for the analysis due to insufficient protein annotation for goat species. In the analysis, interactions with high confidence (>0.7) STRING combined score were considered.

Identification of Milk Proteins in Goat Breeds
In the present study, 749 (including 3349 unique peptides) and 666 (including 3057 unique peptides) proteins were identified from the Teramana and Saanen goat's milk, respectively by label-free proteomic analysis (Supplementary Material Tables S1 and S2). Combining these two datasets, this study identified a total of 870 proteins. Among the proteins identified in the Teramana and Saanen goat milk breeds, respectively 204 (23.4%) and 121 (13.9%) proteins were unique to each breed and 545 (62.6%) proteins were common between the two breeds ( Figure 1).

Label-Free Quantitative Proteomic Analysis of Goat Milk
Label-free quantitative proteomics was used to compare goat milk samples from the Teramana and Saanen breeds. Differences at the proteome level between the two breeds were investigated using the label-free software, Progenesis QI for Proteomics. Normalised proteomics data were used to identify the differentially expressed proteins between the Teramana and Saanen breeds. Differentially expressed proteins were defined as those that showed a fold change cut-off greater than 1.5, a p-value ≤ 0.05 (one-way ANOVA), and a number of unique peptides greater than 2 that matched the protein. Based on these criteria, there were 29 differentially expressed proteins between the Teramana and Saanen goat milk breeds (Table 1), of which 18 (62.1%) were upregulated proteins in the Teramana breed and 11 (37.9%) were upregulated in the Saanen breed. The full list of the 29 proteins identified in this study is shown in Table 1. Eighteen proteins were upregulated in milk from the Teramana breed and eleven proteins were upregulated in milk from Saanen goat breed. (*) Accession number in the UniProt database; ( § ) unique peptides used for quantitation; ( ¥ ) SEQUEST score. ( ‡ ) Indicates whether the proteins were upregulated in the Teramana or in the Saanen goat breeds.

Functional Association Analysis
Functional association analysis was used to investigate the biological processes and pathways involving the 29 differentially abundant proteins identified. Enrichment analyses were carried out in Cytoscape using the ClueGO plug-in. Analyses were run separately over the GO:BP and KEGG pathway databases. Three proteins (immunoglobulin heavy constant Gamma 4, IGHG4; Ig-like domain-containing protein, rig-5; and cysteine-rich secretory protein 3, Crisp3) were not in the ClueGO annotation sets. A total of 34 GO:BP terms (involving 18 differentially abundant proteins) were retrieved (Table 2, Figure 3). These biological processes can be generally summarised in the following groups: (i) Biological regulation, (ii) cellular and developmental process, (iii) multicellular organismal process, (iv) response to stimulus, (v) signalling.
Over-representation analysis using the KEGG pathway database highlighted a total of four pathways (Table 3, Figure 4) related to the response to stimulus. Eight proteins were involved in these pathways. The differentially abundant proteins involved in the complement and coagulation cascades, as previously shown from the analysis over the GO:BP database, are of higher abundance in the Teramana breed.   3 percentage of input proteins found associated with respect to the number of proteins directly annotated with the functional term; 4 the symbol ↑ indicates protein abundance higher in Saanen goat milk than Teramana goat milk, while the symbol ↓ indicates protein abundance higher in Teramana goat milk than Saanen goat milk.  Table 2. In each panel, bars sharing a specific colour are clustered in the same functional group (Table 2).  Table 2. In each panel, bars sharing a specific colour are clustered in the same functional group (Table 2).   Table 3.

Protein-Protein Interaction (PPI) Analysis
PPI analysis was performed using STRING with all the significantly different proteins identified in the comparison of two goat breeds ( Figure 5). A connected protein network was revealed by the analysis. The protein network was composed of 26 nodes divided in: (i) One big module composed of thirteen nodes (50%), (ii) two small components of two proteins (15.4%), and (iii) nine singletons (34.6%). The resulting network showed a PPI enrichment p-value of 4.32 × 10 −8 (four expected edges vs. twenty detected edges) indicating that proteins are at least partially biologically connected. In this network, most of the proteins interacted with only one or two other partners (average node degree equal to 1.54). Two proteins complement C3 (C3) and fibrinogen alpha chain (FGA) presented the highest degree of connection (eight and five edges, respectively), which may assign to them a role as "hub" proteins playing a putative function of controllers inside biochemical pathways that could potentially lead to cascade of protein expression differences. Both proteins had a higher expression in the Teramana goat breed. Moreover, most of the proteins clustered in the big module were included in the GO and KEGG enrichment processes clearly differentiating the two goat breeds.  Table 3.

Protein-Protein Interaction (PPI) Analysis
PPI analysis was performed using STRING with all the significantly different proteins identified in the comparison of two goat breeds ( Figure 5). A connected protein network was revealed by the analysis. The protein network was composed of 26 nodes divided in: (i) One big module composed of thirteen nodes (50%), (ii) two small components of two proteins (15.4%), and (iii) nine singletons (34.6%). The resulting network showed a PPI enrichment p-value of 4.32 × 10 −8 (four expected edges vs. twenty detected edges) indicating that proteins are at least partially biologically connected. In this network, most of the proteins interacted with only one or two other partners (average node degree equal to 1.54). Two proteins complement C3 (C3) and fibrinogen alpha chain (FGA) presented the highest degree of connection (eight and five edges, respectively), which may assign to them a role as "hub" proteins playing a putative function of controllers inside biochemical pathways that could potentially lead to cascade of protein expression differences. Both proteins had a higher expression in the Teramana goat breed. Moreover, most of the proteins clustered in the big module were included in the GO and KEGG enrichment processes clearly differentiating the two goat breeds.  Each node represents a protein, and different line colours represent the types of evidence for the association: Cyan is from curated databases, magenta is experimentally determined, dark green is gene neighborhood, blue is gene co-occurrence, light green is text-mining, black is co-expression, and purple is protein homology.

Discussion
The advancement of electrophoresis and chromatography, along with developments in mass spectrometry technologies, have widened the potential application of proteomics to study milk proteomes from smaller ruminants, such as goats. Milk is an easily accessible body fluid rich in proteins, which are important for tissue growth and cellular functions. Moreover, some proteins can act as hormones, whereas others display antimicrobial properties [1]. These characteristics make milk a promising substrate to investigate proteins and peptides indicative of molecular processes underpinning differences between breeds. Furthermore, this substrate could be easily used for the identification of bi- Figure 5. Protein-protein interaction (PPI) network of the 29 differentially abundant proteins in goat milk. Each node represents a protein, and different line colours represent the types of evidence for the association: Cyan is from curated databases, magenta is experimentally determined, dark green is gene neighborhood, blue is gene co-occurrence, light green is text-mining, black is co-expression, and purple is protein homology.

Discussion
The advancement of electrophoresis and chromatography, along with developments in mass spectrometry technologies, have widened the potential application of proteomics to study milk proteomes from smaller ruminants, such as goats. Milk is an easily accessible body fluid rich in proteins, which are important for tissue growth and cellular functions.
Moreover, some proteins can act as hormones, whereas others display antimicrobial properties [1]. These characteristics make milk a promising substrate to investigate proteins and peptides indicative of molecular processes underpinning differences between breeds. Furthermore, this substrate could be easily used for the identification of biomarkers that could enable a genetic improvement in the goat herds and/or to monitor diseases. The aim of the present study was to investigate and compare the whey proteome of the autochthonous Teramana and Saanen goat breeds.
In the last two decades, great efforts have been addressed to increase the study of milk proteomics, especially from bovine. In more recent times, the study of milk proteome from smaller ruminants, such as goat have received great interest, with whole milk, whey, milk fat globule membrane (MFGM) fractions, and casein fractions as the main substrate utilised [2,[14][15][16]19,36]. A study of Chen et al. [37] identified a total of 843 proteins by comparing the proteome of goat milk during heated processing using label-free quantification. Using a similar approach, 595 and 486 proteins were characterised in the whey of two autochthon Greek goat breeds [27].
In our study, the label-free LC-MS analysis revealed a total of 870 different proteins in the two goat breeds (Supplementary Material Tables S1 and S2) under study. The comparable and higher number of proteins identified in this study indicates the successful use of proteomic techniques. Many of the identified proteins (e.g., complement C3, fibronectin, plasminogen, calcium and integrin binding 1) were also identified in other studies [26,27,36]. Most of the identified proteins were involved in cellular and metabolic process, biological regulation, response to stimuli, etc. that characterise milk (Figure 2). Comparing the unique proteins identified in each breed, the Teramana breed show a higher number of proteins involved in metabolic process (Supplementary Material Figure S1). A similar class of proteins was also identified in whey from Greek goat and sheep breeds [27]. Furthermore, results are in agreement with those presented in the study of Cunsolo et al. [38], who investigated the goat milk proteome in the Camosciata goat breed.
This work is the first study that applies a label-free method to unravel the milk whey proteome of the Teramana and Saanen goat breeds. Whey proteins show specific characteristics, which reflect the physiological requirements of the animal [1]. Utilising statistical analysis, 29 proteins were differentially expressed between the two breeds, highlighting interspecies difference. To our knowledge, this is the first study to observe proteins that are changing significantly in whey between goat breeds. Using the same method, Zhao al. [39] identified 156 differential MFGM proteins between yak and cow. The higher number of proteins identified in their study compared to ours is mainly due to the different species used for the comparison. As expected, the dominant classes of proteins identified in our substrate were whey proteins; however, residues of caseins were also observed.
Among the 29 proteins changing significantly in milk between the two goat breeds, the top differentially expressed proteins (complement C3, C3; fibrinogen alpha chain, FGA; cellular communication network factor 1, CYR61; plasminogen, PLG; and protein disulfideisomerase, P4HB) according to their functions, may contribute to explaining in part the phenotypic differences between the Teramana and Saanen goat breeds. Functional analysis of the proteins identified in our study highlights their role in a wide range of regulatory processes, such as lipid metabolic process, regulation of proteolysis, oxidative stressinduced cell death, etc. KEGG pathway shows that most of the whey proteins upregulated in the Teramana breed are involved in complement and coagulation cascades, whereas three pathways were involved in disease-related pathways in both breeds. The complement and coagulation cascades pathway plays an important role in activating innate immunity, in maintaining the balance of the coagulation-fibrinolytic system, and the associated proteins are critical for the health and nutrition of goat kids [40][41][42]. The upregulation observed in the Teramana breed may allude to a greater capacity of this autochthon breed to adapt to harsh environments compared to breeds that are increasingly used from the industry.
Whey proteins involved in the same pathways were also highlighted in a study by Sun at al. [26] on the colostrum and mature milk of Xinong Saanen goats.
Complement C3 (C3) was upregulated in the Teramana goat. Complement is a central component of the innate immune system. Its main functions include host defence against agents, facilitating adaptive immune responses and elimination of immune complexes and apoptotic cells [43]. The complement system consists of about 35/40 proteins, which are usually associated with blood cells and blood plasma, but found generally at lower concentrations in other secretions of the body-like milk [44]. Difference of C3 level between Teramana and Saanen breeds might be in-line with a higher defence mechanism of the goat mammary gland against infections of the first breed. It has been shown that the concentration of C3 is significantly higher in preterm human milk than term human milk and it has been postulated that this may be due to the higher requirement of protection from the infants in its early days, as the immune system is less developed than term infants [45]. Similar mechanisms may occur in goat breeds, such as the Teramana that are usually keener to live in harsh environments. These animals are more exposed to pathogens, and thus require a higher protection. Fibrinogen (FGA) is a complex plasma protein required for the last phase of blood coagulation [46]. FGA is one of the acute phase proteins (APPs) which are a serum component whose concentrations vary under external or internal influences, such as inflammation, stress, etc. These proteins are important early diagnostic markers of inflammation in animals [47]. In our study, an upregulation of fibrinogen alpha chain was observed in the Teramana breed whey. It has been shown that the main agent involved in mastitis is Staphylococcus aureus and that molecules, such as fibrinogen can change in abundance during infection [48]. The animals used in this study did not show any signs of mastitis or other clinical diseases and the high concentration of APPs is not always a sign of disease [49], the upregulation of this protein may allude to a stronger resistance of the Teramana breed (namely, in general more rustic compared to commercial breeds) to this kind of disease. Cellular communication network factor 1 (CYR61) and cellular communication network factor 2 (CTGF) (upregulated in the Teramana breed) are cysteinerich proteins of the CCN family. These proteins mediate many functions, such as cell survival and apoptosis [50,51]. In mammals, a reduced suckling frequency initiates the downregulation of milk synthesis and the induction of apoptotic pathways and structural remodelling of mammary tissue with a corresponding reduction in secretory activity [52]. It has been proposed that CYR61 and CTGF proteins promote apoptosis of mammary epithelial cells, and that their presence points to the mechanisms underlying the lactation inhibition [53]. It is known that the milk production of the Teramana breed is lower compared to the Saanen [11]. The upregulation of these proteins observed in our study in the Teramana breed may suggest a role played by these proteins in the mechanism involved in the lower milk production of the Teramana breed and highlight the potentiality of these proteins as markers to discriminate between the milk of the two breeds. Plasmin is one of the major endogenous protease present in milk that is secreted in its inactive form plasminogen (PLG), which is the predominant form in fresh milk [54]. Plasmin is then activated and inhibited by the levels of plasminogen activators and inhibitors [55]. Theodorou et al. [56] highlighted that among the factors affecting the plasmin-plasminogen system on sheep's milk, there is the breed factor. Milk contains activins of plasminogen, which can activate fibrinolysis and keep the secretion pipeline unobstructed [26]. In our study, PLG was upregulated in the Teramana breed, which may allude to a stronger resistance of milk duct infections of this breed, namely, in general considered more rustic compared to commercial goat breeds. Moreover, plasmin plays a pivotal role in cheese ripening by breaking down αand β-caseins to result in flavour and texture development [57]. On the other hand, the activity of this protein has been associated with alteration of mammary epithelium permeability and with an increment in paracellular flow, activities that aggravate the milk quality, coagulation properties, and cheese yield. The presence of mastitis, the increment of the age of the animals, and stage of lactation increase the level of PLG [58]. As the animals used in our study had no signs of mastitis or other clinical diseases and were of a similar age, the higher abundance of PLG observed in the Teramana breed may contribute to the development of a characteristic flavour and texture of the dairy products made by the milk of this autochthon breed. Protein disulfide isomerase (P4HB) is a member of the thioredoxin superfamily of redox proteins that is mainly located in the endoplasmic reticulum. P4HB has multiple roles, is a binding partner of other proteins, acts as a chaperone, is a hormone reservoir, as well as a disulfide isomerase in the formation of disulfide bonds [59]. Given their pivotal role in protein-folding, the loss of P4HB activity and the consequent accumulation of misfolded proteins have been associated with the pathogenesis of numerous disease states [60]. To our knowledge, no other studies observed changes in this protein among dairy breeds, the upregulation observed in the Teramana breed might allude to a stronger resistance to disease of this breed that undergo a limited formal crossbreeding. Indeed, due to the negative genetic correlation evidenced between milk production traits and health traits, the genetic selection for higher milk production has been associated with an increased sensitivity to diseases [61,62].

Conclusions
In summary, using label-free proteomics, it was possible to highlight proteomic differences between the whey fraction of the milk from the autochthon Teramana goat breed and Saanen goat breed. Casein and whey proteins are the major proteins of milk, where the casein constitutes approximately 80% of the total protein. An in-depth characterisation of both fractions will benefit dairy productions, will contribute to the conservation on animal biodiversity, and will help in the design of novel products. In this study, we report that the most extensive investigation on the goat whey proteome of a total of 870 proteins was identified. Twenty-nine proteins were able to discriminate between Teramana and Saanen goat breeds with many of the identified proteins involved in complement and coagulation cascades. These findings elucidate the proteome composition of milk whey and quantitative protein profile in the analysed goat's milk. Moreover, the association between the proteins and milk from the two breeds supports the potential use of proteomic profile as a predictive biomarker in milk substrate that can discriminate between different genetic backgrounds. This may be used to evaluate milk adulteration of specific milk and expand the potential direction for the production of specific milk proteins.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ani13142263/s1. Table S1: Full list of the 749 proteins identified by mass spectrometry in the Teramana goat milk determined in Proteome Discoverer using SEQUEST HT algorithm. Table S2: Full list of the 666 proteins identified by mass spectrometry in the Saanen goat milk determined in Proteome Discoverer using SEQUEST HT algorithm. Figure S1: Percentage of milk proteins (over the unique proteins identified in the Teramana and Saanen goat breeds) grouped according to different biological processes.  Institutional Review Board Statement: No permits were required for the described study, which complied with all relevant regulations, since only milk was collected and no animal sacrifice was necessary.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.