Peptidomic changes in the milk of water buffaloes (Bubalus bubalis) with intramammary infection by non-aureus staphylococci

Mastitis by non-aureus staphylococci (NAS) is a significant issue in dairy buffalo farming. In a herd with subclinical NAS mastitis, we identified Staphylococcus microti as the predominant species. To assess milk protein integrity and investigate potential disease markers, we characterized 12 NAS-positive and 12 healthy quarter milk samples by shotgun peptidomics combining peptide enrichment and high-performance liquid chromatography/tandem mass spectrometry (LC–MS/MS). We observed significant changes in the milk peptidome. Out of 789 total peptides identified in each group, 49 and 44 were unique or increased in NAS-positive and healthy milk, respectively. In NAS-positive milk, the differential peptides belonged mainly to caseins, followed by milk fat globule membrane proteins (MFGMP) and by the immune defense/antimicrobial proteins osteopontin, lactoperoxidase, and serum amyloid A. In healthy milk, these belonged mainly to MFGMP, followed by caseins. In terms of abundance, peptides from MFGMP and immune defense protein were higher in NAS-positive milk, while peptides from caseins were higher in healthy milk. These findings highlight the impact of NAS on buffalo milk quality and mammary gland health, even when clinical signs are not evident, and underscore the need for clarifying the epidemiology and relevance of the different NAS species in this dairy ruminant.


Introduction
The water buffalo (Bubalus bubalis) is the second most relevant dairy species after the cow (Bos taurus) 1 , with over 97 million tons of milk produced each year 2 . Mastitis caused by an intramammary infection (IMI) is one of the diseases with the highest impact on the economic performance and welfare of dairy animals 3 . Water buffaloes are generally regarded as less susceptible to mastitis than cows 4,5 . Still, the real impact of intramammary infections (IMI) may be underestimated due to the higher prevalence of subclinical mastitis and issues with the setting of somatic cell count (SCC) thresholds 5,6 , which need proper implementation for mastitis monitoring within dairy herd improvement programs.
The main etiologic agents of clinical and subclinical IMI in buffalo are staphylococci 5,6 . S. aureus is a highly impacting pathogen for clinical severity and ability to spread and persist in the herd, but non-aureus staphylococci (NAS) are most frequently isolated from the milk 6-8 . Moreover, milk NAS in water buffalo have been recently reported as a source of antibiotic resistance [9][10][11][12][13] .
The relationship between different NAS and mammary gland health is poorly known. Identi cation of NAS at the species level is seldom carried out in routine milk bacteriology because of analytical cost issues, combined with the sub-optimal performance of traditional biochemical methods 5,7 . Genotypic identi cation is also problematic in some cases due to the high similarity between some species 14 . When possible, NAS identi cation is carried out by matrix-assisted laser desorption/ionization time-of-ight mass spectrometry (MALDI-TOF-MS) [15][16][17] . Recently, we detected signi cant changes in the protein composition of buffalo milk with staphylococcal mastitis 8 . In that study, we highlighted the need to clarify the role of the different NAS species in this dairy animal and further investigate the impact of NAS on buffalo milk quality.
Shotgun peptidomics is an approach providing an in-depth perspective on the changes occurring in the peptide pro le of many dairy products, adding useful information to the proteomics approach 18 . This method can assess the impact of different conditions by combining the simultaneous identi cation of thousands of peptides with their quanti cation in each sample 19 . Therefore, this approach is ideal for quantitatively investigating the differences in the peptidome of milk from healthy animals compared to that from infected udder quarters with mastitis 30 .
By applying this approach to bovine milk, we detected signi cant peptidomic changes caused by NAS-IMI 20 . In this study, we have investigated the impact of NAS mastitis on the buffalo milk peptidome with a similar pipeline entailing peptide enrichment, high-performance liquid chromatography/tandem mass spectrometry, and bioinformatic analysis, taking into account the causative NAS species in the de nition of the sample groups.

Results
Milk somatic cell counts and bacteriology. Staphylococcus microti was identi ed in all the NAS-positive milk samples. Three milk samples showed the growth of a second colony type, identi ed as Aerococcus viridans in one sample and Streptococcus uberis in two samples. The downstream peptidomic analysis was carried out by classifying the samples by combining bacteriological results with the somatic cell count (SCC) information available for all the samples, as detailed in Table 1. The complete data are reported in Supplementary Table 1. Differential peptidomics. The milk samples listed in Table 1 were subjected to a pipeline entailing peptide enrichment, peptide analysis by high-performance liquid chromatography/tandem mass spectrometry (LC-MS/MS), and bioinformatic analysis to identify differential peptides in the two sample groups. The experimental protocol used in this study is schematically summarized in Figure 1A.
789 and 789 peptides were identi ed in the NAS-positive and healthy milk samples, respectively, for a total of 833 identi ed peptides ( Figure 1B).Among the 745 peptides present in both groups, 5 were increased in NAS-positive in comparison to healthy milk (Welch's t-test: FDR 0.01). No peptides were found decreased in NAS-positive milk. Overall, the analysis identi ed 49 peptides which were increased (5) or present only in NAS-positive (44), and 44 peptides which were present only in healthy milk (44). Table 2 reports the number of total peptides and the number of unique and differential peptides identi ed in NAS-positive and healthy milk samples. Table 2 Total and unique peptides identi ed in the two sample groups by LC-MS/MS and differential analysis. Complete data are reported in Supplementary Table 2 Sample group N° of total peptides N° of unique and differential* peptides  Table 3 details the sequence of all the unique and differential peptides identi ed in NAS-positive and healthy milk, their originating protein, and the cell location/function based on the UniProtKB protein knowledge base or scienti c literature sources [21][22][23][24][25][26] . Table 3 Unique and signi cantly differential* peptides found in NAS-positive and healthy milk with the respective originating protein and its location/function according to the UniProtKB protein knowledge base or referenced literature sources, listed according to the originating protein and to the number of peptides derived from that protein. Figures 2A e 2B illustrate the distribution of all unique and differential peptides identi ed in NAS-positive and healthy milk in terms of number and abundance, respectively, according to the cell location/function of the originating protein and highlight the different nature of the unique and differential peptides identi ed in the two sample groups. The number of total and differential peptides identi ed in the two groups was similar, but their nature in terms of originating proteins differed. In NAS-positive milk, 28 of the 49 peptides (57.14%) belonged to caseins, mainly beta-casein (15, 30,61%), followed by alphaS2 (7, 14.29%) and alphaS1 (6, 12.24%), and 9 peptides belonged to proteins of the milk fat globule membrane (MFGMP) (18.37%). Interestingly, 5 peptides belonged to proteins with immune defense/antimicrobial functions  (5,11.36%), alphaS1 (5, 11.36%), alphaS2 (4, 9.09%), and kappa (2, 4.55%). Only 2 unique peptides (4.54%) belonged to in ammatory/immune defense proteins. The remaining 9 unique peptides belonged to proteins with other locations/functions (20.45%).

NAS-positive milk
In terms of their relative abundance, those derived from casein micelle proteins were higher in healthy milk, while those derived from MFGMP were higher in NAS-positive milk. On the other hand, Immune defense proteins were higher in NASpositive milk also in terms of relative abundance.
The differential peptides were manually analyzed and classi ed according to their C-terminal amino acid. As shown in Figure 3, R at the C-term was considerably less frequent (10.2%) in the peptides unique or more abundant in healthy milk.
On the other hand, peptides ending especially with K (44.9%), V (8.16%), and F (4.08%) were more frequent in NAS-positive milk.

Discussion
Based on our ndings, the presence of a NAS IMI was associated with changes in the peptide composition of water buffalo milk. The differential peptides identi ed were derived from proteins with very different functions and localizations. As milk quality and technological properties may be affected, this deserves consideration 27 .
We detected four differential peptides from serum amyloid A (SAA) in NAS-positive animals. SAA is associated with high SCC and mastitis in bovine cows 25,28 , being an acute phase protein 25 that is overexpressed in milk during mastitis 29,30 . The mammary gland produces a speci c form of SAA, the M-SAA3 31,32 , which can be upregulated by S. aureus lipoteichoic acid 33 . One differential peptide originating from SAA A-3 (VISNARETIQGITDPLLKGMTRDQVREDSKADQ FANEWGR) was found uniquely in NAS-positive milk, in line with our previous nding of the SAA protein only in the milk of water buffaloes with staphylococcal IMI by shotgun proteomics 8 . Interestingly, in another shotgun peptidomics study, we detected SAA peptides only in cows with NAS IMI 20 . Thus, our detection of SAA only in the milk of animals with NAS IMI further supports its diagnostic potential in the dairy buffalo 34 . Nevertheless, the in uence of other physiological variables including parity and stage of lactation on M-SAA levels will have to be assessed 35 .
Three unique and signi cantly differential peptides originating from osteopontin were found in NAS-positive milk. This is also in line with our previous peptidomic study on bovine cows 20 , although we did not identify the intact protein by proteomics in the water buffalo 8 . Among other biological roles, osteopontin upregulates interferon-gamma and interleukin-12 and downregulates interleukin-10 and plays a role in inducing type I immunity 36 . In cows, osteopontin peptides have been reported in subclinical mastitis 37 and experimental Escherichia coli IMI 22,23 . As also indicated in a recent review on NAS affecting cows, this further indicates that NAS indeed elicit an in ammatory response in the mammary gland, as con rmed by the increased milk SCC. The present nding may support the hypothesis that NAS provide cross-protection against other mastitis pathogens 38 as components of the mammary gland microbiota 39,40 .
On the other hand, most of the unique peptides found in healthy milk belonged to proteins of the milk fat globule membrane (perilipin 2, butyrophilin, GLYCAM-1, sodium-dependent phosphate cotransporter, annexins, glycoprotein-2) 26,41 , in line with the observations made by shotgun proteomics 8 . The predominance of MFG membrane proteins in healthy milk might be related to the high fat content of buffalo milk, and therefore to the higher abundance of these lipid secretion vesicles compared to cow milk. MFG are an important source of nutraceutical components, including membrane proteins, and the possible in uence of NAS IMI on their integrity may deserve further consideration concerning nutritional value, product quality, and technological properties 27 . When looking at the differential distribution of peptides in terms of abundance, we observed that healthy milk was characterized by a higher abundance of casein proteolytic peptides, and NAS-positive milk by a higher abundance of peptides derived from MFG membrane proteins and immune defence proteins. While the rst nding might be in uenced by the higher abundance of caseins in healthy vs infected milk, the second nding further highlights the impact of NAS IMI on integrity and abundance of MFG membrane proteins and immune defence proteins, respectively, reinforcing the above considerations 42 .
The distribution of unique and differential peptides based on their C-terminal aminoacid showed a higher frequency of peptides ending with R in healthy milk as opposite to peptides ending especially with K, V, and F, in NAS-positive milk, in line with the observations made by our previous peptidomic work in bovine cows 20 . According to the MEROPS database, plasmin generates peptides ending with R and K at the C-term, while elastase, cathepsin D and cathepsin G generate peptides ending with V and F at their C-term 43 . Our results suggest a more intense proteolytic activity by plasmin and endogenous proteases released by in ammatory cells in NAS-positive milk.
The impact of NAS IMI on the buffalo milk peptidome was less intense than observed in cows in our recent work 20 .
However, as mentioned above, many ndings were consistent including the presence in NAS-positive milk of peptides derived from osteopontin and SAA, and the different frequency of C-terminal aminoacids in the proteolytic peptides of the two sample groups 20 .
Concerning the etiologic agent, the identi cation of S. microti as the predominant species in the milk of water buffaloes with subclinical mastitis is noteworthy as only one study reported its association with mastitis in bovine cows 44 . S. microti is closely associated with S. rostri and S. muscae, and it has been rst isolated from Microtus arvalis, the common vole.
Since its description, it has been isolated from rodents/insectivores and a female sand y 44 . Therefore, the role of animal vectors might be relevant in this case. Adding to anatomical and physiological characteristics, important differences characterize bubaline cows and bovine cows in terms of animal management, farming practices (housing, feeding, bedding, milking routine), environmental temperature and humidity, and presence of water ponds, and consequently contact with different microbial reservoirs including wild and domestic animals. This may lead to mammary gland exposure and colonization by other NAS species than the bovine dairy cows, as well as to different bacterial loads in the farm environment, and should be carefully considered.

Methods
Animals and milk samples. The study was carried out on quarter milk samples collected from a commercial water buffalo dairy farm located in Campania, Southern Italy, with an increased bulk tank somatic cell count related to NAS IMI. The farm maintained the milking buffaloes in free-stall barns with deep-bedded cubicles with straw. All the animals were fed with a Milk samples were processed as indicated by the National Mastitis Council 46 . Before sampling, teats were cleaned with a pre-dipping foam containing lactic acid, and the apex was scrubbed and disinfected with alcohol. The rst streams of milk were discharged, and 20 mL was collected aseptically from each quarter into sterile vials. The milk samples were kept at 4°C until they reached the laboratory (within the day) at the IZS in Portici for bacteriological assays and somatic cell count.

Milk sample preparation for peptidomic analysis
The milk samples were processed for peptidomic analysis as described previously 20 . Brie y, milk was defatted by centrifugation and processed for the depletion of high-molecular-weight proteins on centrifugal lters. The ltrate was precipitated, and peptides were dried, dissolved in 1% (v/v) formic acid and desalted before MS analysis.
Tandem mass spectrometry analysis of peptides Tandem mass spectrometry analysis of peptides was carried out with duplicate runs for each sample as described previously 20  are ltered by minimal peptide length (6 amino acids) and m/z accuracy (8 ppm). The quality of a match between sequence and observed peaks was provided by a high cross-correlation score (≥1.5). PSM con dence was set to High.
Unspeci c digestion was chosen, and neither xed nor variable modi cations were set. The resulting peptides and protein hits were further screened accepting only those hits listed as high con dence and with an Xcorr ≥ 1.5. Only peptides present and quanti ed in 66,6% of the repeats were positively identi ed and used for statistical analysis. Peptides were considered increased or decreased if they showed a signi cant Welch t-test difference (cut-off at 1% permutation-based FDR) or if they were present with a frequency ≥ 66,6% in either NAS-positive or healthy milk group but less than 66,6% in the other group 49 . Statistical analysis was performed using the Perseus software (version 1.5.5.3, www.biochem.mpg.de/mann/tools/). Peptide sequences were analyzed manually for C-terminal amino acids. The potential proteases generating the cuts were classi ed based on the MEROPS database 43 by evaluating the speci cities of the main proteases generating the cuts 50 . The protein functions reported in Table 3 were retrieved from the UniProtKB protein knowledgebase (www.uniprot.org).