A proteomic approach to identify biomarkers of foal meat quality: A focus on tenderness, color and intramuscular fat traits

Food Chemistry 405 (2023) 134805 Available online 1 November 2022 0308-8146/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). A proteomic approach to identify biomarkers of foal meat quality: A focus on tenderness, color and intramuscular fat traits María L´ opez-Pedrouso a , Jos´ e M. Lorenzo b , c , Aurora Cittadini b , d , María V. Sarries d , Mohammed Gagaoua e , Daniel Franco b , f , * a Department of Zoology, Genetics and Physical Anthropology, Universidade de Santiago de Comspostela, Santiago de Compostela, 15872, A Coru˜ na, Spain b Centro Tecnol´ ogico de la Carne de Galicia, Rúa Galicia N◦4, Parque Tecnol´ ogico de Galicia, San Cibrao das Vi˜ nas, 32900 Ourense, Spain c ´ Area de Tecnología de los Alimentos, Facultad de Ciencias de Ourense, Universidad de Vigo, 32004 Ourense, Spain d Instituto de Innovaci´ on y Sostenibilidad en la Cadena Agroalimentaria (IS-FOOD), Universidad Pública de Navarra (UPNA), Campus de Arrosadia, 31006 Pamplona, Spain e PEGASE, INRAE, Institut Agro, 35590 Saint-Gilles, France f Department of Chemical Engineering, Universidade de Santiago de Compostela, Campus Vida, 15782 Santiago de Compostela, Spain ARTICLE INFO Keywords: Horse Finishing diet Omics Muscle proteins Energy metabolism Response to stress Biochemical pathways ABSTRACT Foal meat is considered a healthy alternative to other meat sources and more environmentally sustainable. However, its quality is highly variable and there is lack of knowledge about the molecular mechanisms underlying its determination. Genotype and diet play a relevant role as the main factors that can allow a control of the final quality and the use of high-throughput analytical methods such as proteomics is a way to achieve this lofty goal. This research aimed to study-two breeds (Burguete and Jaca Navarra) supplemented with two different finishing diets: conventional concentrate and straw (C) vs silage and organic feed (S). The proteomic approach built a library of 294 proteins that were subjected to several statistical and bioinformatic analyses. Burguete breed finished with concentrate produced higher meat quality in terms of tenderness, intramuscular fat and color lightness mainly due to the high abundance of energy metabolic proteins. Tenderness was correlated to myofibrillar proteins (ACTA1, MYBPH, MYL1 and TNNC1) and energy metabolic proteins (ALDOA, CKM, TPI1 and PGMA2). Regarding color, the main pathways were energy metabolism, involving several glycolytic enzymes (ALDOA, PKM, PFKM and CKM). Oxidative stress and response to stress proteins (HSPA1A, SOD2 and PRDX2) were further involved in color variation. Moreover, we revealed that several proteins were related to the intramuscular fat accordingly to the breed. This study proposed several candidate protein biomarkers for foal meat quality that are worthy to evaluate in the future. 1. Introduction Horse meat consumption is emerging as a healthy and nutritious alternative to traditional meat sources (beef or pork). Moreover, environmental and health concerns support the production of meat from pasture, such as horses raised under extensive livestock production systems. Nevertheless, horse meat is still not widely consumed in developed countries despite its excellent nutritional value characterized by a high content of protein, and iron as well as a low-fat and cholesterol amounts and a healthy fatty acid profile (Franco & Lorenzo, 2014; Franco et al., 2011). In the past, horse meat was not always fully accepted for social, historical, cultural and religious reasons as well as for the negative image acquired due to fraudulent practices, especially in beef adulteration. However, during the last decade changes were observed in consumers’ attitudes towards this type of meat (Cittadini, Sarri´ es, Domínguez, Indurain, & Lorenzo, 2021). Indeed, according to The Food and Agriculture Organization (FAO), horsemeat production is slowly increasing, recording an increment of 7.57 % between 2010 and 2020 (FAOSTAT, 2021). Generally speaking, horse meat is particularly considered a tough meat leading to a lower acceptability. Thus, any possibility to improve its tenderness should be a key driver for consumer-purchase choice and decision. Collagen and intramuscular fat are contributing factors to meat tenderness and textural properties. Collagen presents intermolecular * Corresponding author at: Department of Chemical Engineering, Universidade de Santiago de Compostela, Campus Vida, 15782 Santiago de Compostela, Spain. E-mail address: [email protected] (D. Franco). Contents lists available at ScienceDirect Food Chemistry journal homepage: www.elsevier.com/locate/foodchem https://doi.org/10.1016/j.foodchem.2022.134805 Received 4 July 2022; Received in revised form 26 October 2022; Accepted 27 October 2022 Food Chemistry 405 (2023) 134805 2 crosslinks which provide structure and strength, as well as, to a lesser extent, elastin and reticulin (Roy, Walker, Rahman, Bruce, & McMullen, 2018). Furthermore, the sarcomere shortening during the ageing process and the proteolysis of myofibrillar proteins affect meat toughness (Ertbjerg & Puolanne, 2017). In pre-rigor mortis, there is no supply of oxygen, but the muscle tissue remains unaffected. Afterwards, during the rigor mortis phase there is a lack of the production of ATP and phosphocreatine favoring a high rigidity. Finally, the postmortem phase is related to the proteolytic activity of enzymes such as calpains, cathepsins, proteasome and caspases systems (Lana & Zolla, 2016; Ouali et al., 2006). In addition, oxidative stress is also a pivotal pathway that has been used as an indicator of structural protein degradation during the postmortem phase and it is consequently associated with beef tenderness (Gagaoua, Terlouw, et al., 2021; Malheiros et al., 2019). Another peculiar feature of the horse meat is its dark color with a bluish sheen produced by the high content of myoglobin. Myoglobin is a sarcoplasmic heme protein, mainly responsible for meat color depending on its oxidation state (Purslow, Gagaoua, & Warner, 2021). In this sense, the myoglobin has a high ability to combine red oxymyoglobin with oxygen-generating brown metmyoglobin. Proteomics is a powerful tool that enables in-depth characterization of the molecular pathways involved in meat color variation as well as for the discovery of protein biomarkers. The relationship between proteomic changes and tenderness has been widely studied in beef (Gagaoua, Terlouw, et al., 2021; Zhao et al., 2014) and pork (Huang, Larsen, Palmisano, Dai, & Lametsch, 2014; L´ opez-Pedrouso, Lorenzo, Gagaoua, & Franco, 2020). Indeed, protein biomarkers linked to meat quality have been searched in beef and pork and relevant biological pathways such as muscle contraction, metabolism, response to stress, oxidation, proteolysis and apoptosis among other pathways were found to be involved in the tenderness of meat. On the other hand, in horse, few studies have been reported using proteomics to investigate the variation in its meat quality (Beldarrain, Sentandreu, Aldai, & Sentandreu, 2022; della Malva et al., 2021). Therefore, this work aimed to understand the molecular mechanisms of tenderness, color and intramuscular fat in foal meat and to identify candidate protein biomarkers related to these relevant quality traits. This information will bring us new insights for management decisions to ensure a high and consistent foal meat quality. 2. Materials and methods 2.1. Rearing and slaughtering of the foals Twenty male foals, ten from Jaca Navarra and ten from Burguete were used in this trial. The horses were from local farms after a weaning period and reared under a semi-extensive livestock system. From 12 to 17 months of age, foals were fed at pasture and then kept indoors on an experimental farm of the Institute for Agri-food and Technology and Infrastructures of Navarre (Roncesvalles, Navarra, Spain). Finally, each genotype was divided into two equal parts and finished during 3–4 months with two different diets. Ten male foals (5 of Jaca Navarra and 5 of Burguete breed) were fattened with conventional concentrates (starter and finisher ones) and straw (ad libitum) (C). The other group of ten male foals (5 of Jaca Navarra and 5 of Burguete breed) were finished with silage (produced by local farmers) and an organic fodder (with certification of UE/no UE Organic Agriculture), where silage represented the majority component of the diet (approximately 60 % of silage and 40 % of organic feed) (S). Additional details were previously reported (Cittadini, Sarri´ es, Domínguez, Pateiro, & Lorenzo, 2022). Briefly, foals at 21 months of age were transported to the slaughterhouse (Protectora de carne S.L., Salinas de Pamplona, Navarra, Spain) the day before the sacrifice following the European regulations (European Commission, 2005) and the conditions described in a recent work (Cittadini et al., 2022). The foals were stunned in the frontal region of the head with a captive bolt, slaughtered and dressed according to the Council Regulation 1099/2009 (European Commission, 2009). Then, the carcasses were chilled for 24 h in a conventional room at 0 ◦C and the following day, the half-carcasses were transported under refrigeration to C´ arnicas Mutiloa (Sangüesa, Navarra, Spain) and stored in a cooling chamber at 0 ◦C, for four days in line with the local procedure. Subsequently, the longissimus thoracis and lumborum (LTL) muscle from each half-carcass was removed from the sixth to the thirteenth ribs and cut into 25 mm thick steaks, vacuum-packed and stored at −20 ◦C until physicochemical and proteomic analyses. 2.2. Foal meat quality Chemical composition (moisture, intramuscular fat, protein, ash, iron content and cholesterol), pH, color parameters [lightness (L*), redness (a*), yellowness (b*), chroma (c*), hue (h*), myoglobin (Mb), metmyoglobin (MMb) and oxymyoglobin (OMb)], cooking loss and shear force were assessed following the protocols detailed in Cittadini et al. (2022) and Pateiro et al. (2013). 2.3. Proteomic analysis 2.3.1. Protein extraction and trypsin digestion Lyophilized horse meat powder (50 mg) was mixed in RIPA buffer containing 200 mmol/L Tris/HCl (pH 7.4), 130 mmol/L NaCl, 10 % (v/ v) glycerol, 0.1 % (v/v) SDS, 1 % (v/v) Triton X-100 and 10 mmol/L MgCl 2 together with anti-proteases and anti-phosphatases (Sigma- Aldrich, St. Louis, MO, USA) by a TissueLyser II (Qiagen, Tokyo, Japan). The suspended solids were separated by centrifugation at 14,000g at 4 ◦C for 20 min. In the solution, protein concentration was measured using an RC-DC kit (Biorad Lab., Hercules, CA, USA) following the manufacturer’s instructions. To eliminate the impurities from protein fractions, the protein aliquots of 100 μ g were subjected to onedimensional SDS-PAGE. The electrophoresis was performed until the protein is concentrated on the top of the resolving gel (10 %) in a single band. This band was then carefully excised using sterile scalpels and cut into small pieces, followed by a washing with Milli-Q water and 50 mM ammonium bicarbonate in 50 % methanol. Subsequently, the dehydration was carried out with acetonitrile and a vacuum centrifuge. Finally, the protein sample was reduced with 10 mM DTT in 50 mM ammonium bicarbonate at 60 ◦C for 30 min and alkylated with 55 mM iodoacetamide solubilized in ammonium bicarbonate at room temperature for 30 min in darkness. Afterwards, the protein solution was digested by trypsin (Promega, Madison, USA) at a concentration of 20 ng/ μ L in 20 mM ammonium bicarbonate at 37 ◦C for 16 h. The final solution of peptides was mixed with 0.1 % formic acid and stored at −20 ◦C until LC-MS/MS analysis. 2.3.2. Generation of the reference spectral library A mixture of 4 μ g of peptides from each sample was prepared and analysed by a shotgun data-dependent acquisition proteomic approach by micro-LC-MS/MS. The peptides were separated by the micro-LC system Ekspert nLC425 (Eksigen, Dublin, CA, USA) using an YCM- TriartC18 column (150 μ m ×0.3 mm, 12 nm, s-3 μ m) (YMC CO, Japan) at a flow rate of 5 μ L/min. The solvent A (water, 0.1 % formic acid) and solvent B (acetonitrile, 0.1 % formic acid) were mixed in a linear gradient from 5 % to 95 % B for 30 min, at 90 % B for 5 min and 5 % B for 5 min in an equilibration step. The mass spectrometer coupled model Triple TOF 6600 (SCIEX, Framingham, MA, USA) operates with a data-dependent acquisition mode following the parameters of 250 ms survey scans from 100 to 1500 m/z (25 ms of acquisition time) for a total cycle time of 2.8 s. The MS raw files were used to search in Uniprot Swiss-Prot databases using ProteinPilot software v.5.0.1. (SCIEX, Framingham, MA, USA). The false discovery rate (FDR) was set to 1 for peptides and proteins with a confidence score above 99 %. 2.3.3. Protein quantification For accurate quantification, SWATH-MS was carried out by a DIA M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 3 method. A total of 20 samples from the four groups (five biological replicates in each group) were analyzed. Therefore, 4 µg of peptides were subjected to liquid chromatography using the same conditions as previously described. Regarding detection, the mass spectrometer operated with a 50 ms survey scan and the MS/MS analysis in a cyclic manner using an acquisition time of 50 ms in a total time of 6.3 s. A cycle included 65 scans per SWATH window of variable width (1 m/z overlap) covering the 400 to 1250 m/z. The data extraction and alignment of spectral were performed by PeakView v.2.2. (SCIEX, Framingham, MA, USA). These data matched the reference spectral library from the section above using the following settings: 10 peptides/protein and 7 fragments/peptide and FDR below 1 %. The protein quantification was calculated according to the sum of peak areas from all peptides and fragment ions of the protein. Moreover, two technical replicates were used for each meat sample. 2.4. Statistical analysis For the statistical analysis of the meat quality attributes, a two-way ANOVA was performed for all variables considered, using the General linear model procedure of the SPSS package (SPSS 23.0, Chicago, IL). A fixed effect of breed and finishing feed was included in the model. The model used was: Yij = μ +Bi +Fj + (BxF)ij + ε ij. where Yij is the observation of dependent variables, μ is the overall mean, Bi is the effect of breed, Fj is the effect of finishing feed, (BxF)ij is the interaction between breed and finishing feed and ε ij is the residual random error associated with the observation. Interaction term was initially included in the model, but finally it was not included in the table of results because there was no statistical significance (P >0.05) for any of the evaluated variables. For proteomics data analyses, a student’s t-test was applied to compare groups in pairs identifying the proteins with significant differences, using a p-value scoring of 0.05 and a fold change of 1.5 as a cutoff. The correlations among individual protein quantifications and meat quality traits were determined by Pearson’s linear correlation coefficient at two levels (P < 0.01 and P <0.05) using standardized data (z-scores). Subsequently, the differentially abundant proteins within breed (Burguete and Jaca Navarra) were used to generate heatmaps using clustering analysis on XLSTAT 2.01 (Addinsoft SARL, París, France). The Euclidean distances were used based on z-scores. The correlations between protein abundances and meat quality traits (shear force, color and intramuscular fat) were determined using Pearson’s linear correlation coefficient. Only the proteins which showed a pvalue <0.05 and a coefficient of correlation higher than 0.7 or lower than −0.7 were considered in the manuscript. For bioinformatics analyses, STRING v10.5 open source software was used to analyze the protein–protein interactions. Metascape® was further used for in-depth analyses. Gene Ontology (GO) processes enrichment on the list of proteins that were correlated with the investigated meat quality traits. 3. Results and discussion 3.1. Characterization of foal meat Considering the chemical composition of the meat samples, the most outstanding characteristics are the low-fat (3.63–5.05 %) and high protein contents (20.03–20.64 %) as shown in Table 1. The values of intramuscular fat (5.10 ±3.40 %) and protein (20.3 ±0.26 %) were similar to those obtained from older horses (4 to 7 years old) (Stanisławczyk, Rudy, & Gil, 2020). Another important feature of foal meat is the low amount of cholesterol in comparison with lean meat of beef or pork (62 and 71 mg/100 g meat, respectively) (Ahmad, Imran, & Hussain, 2018) as confirmed in this study, with values varying from 38.90 to 41.70 mg/100 g of meat (Table 1). It has been reported that regular consumption of horse meat may contribute to reduce total and LDL cholesterol in healthy individuals (Del Bo’ et al., 2013). Concerning the textural parameters, values of cooking losses of 23.84–27.62 % and shear force of 26.5–37.19 N/cm 2 (Table 1) were obtained, that can be considered acceptable for consumption in terms of juiciness and tenderness. Additionally, the finishing diet based on concentrate increased the intramuscular fat and decreased the shear force in both breeds, which could improve the consumer acceptability. This is largely due to the horse breeds which are intended for meat Table 1 Effect of breed (B) and animal feeding (F) on chemical composition, color parameters, pH, cooking losses and shear force of foal meat (values are expressed as mean ± standard deviation). Parameter Foal breed Animal feed Burguete Jaca Navara p-Value Concentrate Silage p-Value Chemical composition Moisture (%) 70.97 ±1.36 71.92 ±1.05 0.040 70.70 ±1.32 72.19 ±0.69 0.003 Intramuscular fat (%) 4.79 ±1.30 3.88 ±1.01 0.047 5.05 ±1.19 3.63 ±0.79 0.004 Protein (%) 20.64 ±0.87 20.03 ±0.63 0.111 20.36 ±0.94 20.30 ±0.69 0.863 Ash (%) 1.11 ±0.03 1.21 ±0.07 0.002 1.16 ±0.07 1.16 ±0.08 0.941 Cholesterol (mg/100 g wet meat) 38.90 ±3.07 41.70 ±3.77 0.079 39.07 ±3.56 41.53 ±3.45 0.118 Color parameters Luminosity (L*) 39.29 ±1.73 37.11 ±2.75 0.052 38.84 ±2.94 37.56 ±1.91 0.234 Redness (a*) 14.24 ±0.76 13.84 ±1.19 0.410 14.13 ±1.23 13.95 ±0.75 0.698 Yellowness (b*) 13.54 ±0.70 12.12 ±1.04 0.002 13.13 ±1.14 12.54 ±1.09 0.156 Chroma (C*) 19.66 ±0.84 18.43 ±1.15 0.016 19.32 ±1.27 18.77 ±1.04 0.250 Hue (h◦) 43.56 ±1.79 41.23 ±3.41 0.085 42.89 ±3.31 41.90 ±2.52 0.443 % Myoglobin 17.05 ±3.05 21.47 ±3.74 0.008 18.94 ±2.94 19.58 ±5.03 0.664 % Metmyoglobin 27.42 ±1.75 26.78 ±2.31 0.519 27.17 ±1.60 27.04 ±2.46 0.890 % Oxymyoglobin 55.13 ±1.37 52.18 ±2.75 0.009 53.81 ±2.41 53.50 ±2.90 0.754 Fe (mg/100 g wet meat) 1.93 ±0.30 2.39 ±0.43 0.013 2.26 ±0.50 2.06 ±0.35 0.260 Quality parameters Ultimate pH 5.55 ±0.07 5.59 ±0.05 0.140 5.61 ±0.05 5.53 ±0.06 0.004 Cooking losses (%) 24.08 ±3.36 27.39 ±2.42 0.005 23.84 ±2.78 27.62 ±2.75 0.002 Shear force (N) 28.39 ±7.51 35.31 ±6.74 0.004 26.51 ±6.41 37.19 ±4.90 0.000 The interactions breed ×animal feed were not significant (P >0.05), therefore they are not shown. M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 4 production along with the finishing feed and their early slaughter age. It is worthy to mention that this last factor has an important effect on tenderness, as reported in earlier studies using horses at the age of 4 to 7 years, which presented tougher meats with a shear force of 96.7 ±4.01 N/cm2 (Stanisławczyk et al., 2020). Moreover, both considered effects had in this study a significant (P ≤0.004) influence on meat shear force. Nevertheless, most of the quality parameters evaluated were more defined by breed than finishing diet, as expected. These results are in line with those previously discussed in a former study (Cittadini et al., 2022). In this study, Burguete meat had a higher (P <0.05) intramuscular fat percentage than those of Jaca Navarra, which could be related to the distinct carcass characteristics and morphology of these two breeds (Cittadini et al., 2021). Therefore, these animals had also lower (P < 0.01) myoglobin values, which could be due to the different temperament, physical activity and, consequently, metabolic muscle properties of these two equine breeds. Data, additionally, showed a tendency among groups for cholesterol values (P <0.1), where Burguete meat had the lowest contents in comparison to Jaca Navarra. Concerning diet, samples from the C group reported the greatest (P <0.01) fat percentages, probably due to the composition of this diet, which contained higher fat amounts than the S group (Cittadini et al., 2022). In the same line, the C group improved the texture properties, that showed significantly lower (P <0.001) shear force values. The above results were further confirmed by the PCA analysis (Fig. 1). In the PCA, the first two principal components (PC) explained 52.15 % of the total variation, with principal component 1 (PC1) accounting for 35.31 % and PC2 for 16.85 %, respectively. In detail and, considering the four groups, it seemed that Burguete foals fed with conventional concentrates (C group) accumulated the greatest amounts of intramuscular fat and the lowest values of cholesterol. Furthermore, they were related to lightness (L*) and yellowness (b*) and stood out for their enhanced meat quality attributes, such as the reduced toughness, which could increase consumers’ acceptability. On the contrary, Jaca Navarra foals supplemented with silage reported the opposite trend. These results are in accordance with (Cittadini et al., 2022) who demonstrated that foal meat from Burguete finished with a traditional concentrate showed the highest quality standards in terms of tenderness and juiciness. 3.2. Comparative analysis of proteome from meat samples from two breeds (Burguete and Jaca Navarra) under two finishing diets (concentrate and silage) The proteomic approach used in this study quantified 294 proteins. The heat map based on unsupervised clustering analysis is given in Fig. 2. A significant clustering around breeds was observed, meanwhile, a lower impact of finishing diet was observed on muscle proteome. This is confirmed by the grouping of the animals within each breed. The foals highlighted in blue grouped mainly the meat from Burguete breed, whereas the brown cluster consisted of Jaca Navarra breed meat samples. This demonstrates a proteomic pattern associated with the horse breed. Genotype has a prominent effect on farm animal features, which is reported to be linked to muscle proteomic. Indeed, proteomics was traditionally used to differentiate the breeds as a tool for product authentication (Rodríguez-V´ azquez et al., 2020). Feeding and management strategies are also used to enhance meat quality, but with a lower impact on the final product than genotype (Alemneh & Getabalew, 2019). In general, animal feeding studies often focus on intensive and extensive production systems, using proteomic tools to differentiate them (de Melo et al., 2020; Fonseca et al., 2019). Even though stronger proteomic differences were attributed to the breed effect, within each studied genotype, there were also observed variations in muscle protein abundances associated with the diet (Fig. 3). Regarding diet effect, our results were accurate and reproducible in both breeds. Firstly, the impact of finishing diet for conventional concentrates and straw (C) in comparison with silage (S) for 3–4 months was detected. In this case, the dietary treatments induced changes at the muscle proteome level, affecting the final meat quality traits. A volcano plot assessed differential protein expression displaying the protein changes due to this factor of variation. Red points represent the proteins with the largest statistically significant changes (fold-change ≥30; pvalue ≤0.05) as a response to the by finishing diets. In detail, the points with red color located on the left represent the proteins significantly over-abundant in feeding with silage, meanwhile on the right represent the proteins over-abundant in feeding with concentrates and straw (C). In the Burguete breed, four proteins were significantly over-abundant due to the silage diet: ATP synthase subunit (ATP5PO), PDZ domaincontaining protein (AHNAK), albumin (ALB) and annexin (ANXA5). Eight proteins were over-abundant in the traditional concentrate finishing diet: enoyl-CoA hydratase 1 (ECH1), AIR carboxylase (PAICS), alpha-2-HS-glycoprotein (AHSG), phosphoglycerate mutase (PGAM2), triosephosphate isomerase (TPI1), mono-ADP ribosylhydrolase 1 (MACROD1), nebulin (NEB) and malate dehydrogenase (MDH1). For the Jaca Navarra breed, eight proteins were over-abundant as a consequence of silage feeding: hydroxiacylglutathione hydrolase (HAGH), heat shock protein beta 6 (HSPB6), voltage-dependent anionselective channel protein 3 (VDAC3), selenium binding protein 1 (SELENBP1), peptidyl-prolyl cis–trans isomerase (PPIA), guandinoacetate N-methyltransferease (GAMT), acid phosphatase (ACP1) and fatty Fig. 1. Principal Component Analysis in foal meat from two breeds (Burguete and Jaca Navarra) finished with two diets (C =conventional concentrate and straw and S =silage and organic feed), using chemical composition and quality traits (moisture, intramuscular fat, protein, ash, cholesterol, luminosity =L*, redness =a*, yellowness =b*, chroma, hue, myoglobin, metmyoglobin, oxymyoglobin, Fe, pH, cooking losses and shear force). M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 5 acid-binding protein 3 (FABP3). In the case of concentrate feeding, two proteins were over-abundant ATP synthase subunit beta (ATP5F18) and Smoothelin like 2 (SMTNL2). At first glance, it may appear that the effect of the two finishing diets on the proteome of Burguete and Jaca Navarra foals cannot be compared. However, an in-depth assessment showed that variations in the diet affected energy metabolism, mainly glucose and lipid metabolism within the two breeds. In Burguete meat samples, several proteins such as MDH1, PGAM2 and TPI1 from carbohydrate metabolism mainly consisting of glycolysis, tricarboxylic acid cycle and oxidative phosphorylation were affected. According to our results, both ATP synthase members (ATP5F18 and ATP5PO) were influenced by dietary changes in both breeds. In the case of the TCA cycle, the pyruvate is converted into acetyl coenzyme A producing NADH. Thus, NADH is used in the respiratory chain complexes to produce large amounts of ATP by means of ATP synthase (Bonora et al., 2012). On the other hand, fatty acids play critical roles in the energy metabolism of mammals. Our results showed significant differences in lipid transporter fatty acidbinding protein 3 (FABP3) from Jaca Navarra foals and albumin (ALB) from Burguete ones, which are related to lipid metabolism. FABP3 plays an important role in skeletal muscle participating in fatty acids uptake and cytosolic transport showing a high binding affinity for palmitic, oleic and stearic acids. The ALB can transport long-chain fatty acids and consequently plays a role in the fatty acid mobilization along the antioxidant activity in plasma. Differences in the regulation of the lipid metabolism which determine the final fatty acid composition of the muscle have important consequences on quality traits (Kim, Markkandan, Lee, Kim, & Yoo, 2020; Puig-Oliveras et al., 2014). In general terms, the shift in metabolism provoked by the type of diet underlies a proteomic-specific pattern associated with quality parameters of foal meat as in the case of other species. 3.3. Proteomic analysis related to meat quality traits (tenderness, colour and intramuscular fat) from Burguete and Jaca Navarra breeds The relationship between the most important meat quality traits and protein abundances is displayed in Table 2. This table depicted the proteins with absolute values of correlation coefficients above 0.7. We identified 35, 67 and 21 proteins to be significantly correlated with shear force, color and intramuscular fat, respectively. The prediction and control of meat tenderness using protein biomarkers is currently of great importance. Our results showed a significant association between tenderness (shear force) with 29 and 6 proteins in the case of Burguete, and Jaca Navarra, respectively. Among them, some proteins were from muscle filament sliding such as ACTA1 (r = −0.745), TNNC2 (r =0.858), MYBPC1 (r = − 0.881), MYBPC2 (r = −0.715), MYL1 (r = − 0.860), TTN (r = − 0.721), NEB (r =0.848) and ACTN3 (r = − 0.782). Furthermore, many of the proteins associated with tenderness resulted to be structural proteins including MYOM1 (r = −0.796), MYOM2 (r = − 0.715), MYBPH (r = − 0.727) and OBSCN (r = −0.922). Regarding the sign of correlation, most of them presented negative correlations meaning that tender meat have a higher abundance of mostly structural proteins according to other species (L´ opez- Pedrouso et al., 2019). It is known that tenderness is related to the breakdown of myofibrillar structure caused by postmortem proteolysis. According to Gagaoua, Terlouw, et al. (2021), Gagaoua, Warner et al. (2021), ACTA1, MYBPH, MYL1 and TNNC1 could be considered as very robust candidate biomarkers of beef tenderness. Several other proteins correlated negatively with tenderness are involved in energy metabolism such as ALDOA (r = − 0.752), CKM (r = −0.797), TPI1 (r = − 0.707) and PGAM2 (r = − 0.739). During the prerigor metabolism, energy metabolism mainly glycolysis contributes to tenderness development. Muscle metabolism is influenced by animal nutrition, pre-slaughter stress, muscle type and other factors leading to modifications in glycogen level and mitochondria status. In meat tenderness, the variation in proteolysis is a more critical factor than genetic considerations (Warner, Greenwood, Pethick, & Ferguson, 2010). Biochemical processes related to proteolysis of structural proteins particularly ACTA1, MYBPH, MYL1 and TNNC1 as well as ALDOA, CKM, TPI1 and PGMA2 involved in energy metabolism have been proved to be the most trusted biomarkers in the field of tenderness (Gagaoua, Terlouw, et al., 2021; Gagaoua, Warner et al., 2021). Overall, these data agree with the information depicted in the string network within the clusters displayed in Fig. 4. There are two main protein clusters, one related to energy metabolism (ALDOA, CKM, PGAM2, TPI1 and NME2) and the other one involving structural proteins (ACTA1, TNNC2, MYBPC1, MYBPC2, MYL1, TTN, NEB and ACTN3). Regarding color parameters, the concentration of myoglobin and its oxidation state play a key role in chromatic parameters (Gagaoua, Fig. 2. Heat map analysis based on the 294 quantified proteins The values of protein quantifications were transformed in the base 2 logarithm scale for five biological replicates. The longissimus thoracis and lumborum muscle from two breeds (Burguete and Jaca Navarra) and two finishing diets (C =conventional concentrate and straw and S =silage and organic feed) were used for the analysis. Bright green and red represents the higher and lower abundance of proteins, respectively. Two main clusters (blue and brown) were merged from protein abundance showing similar proteomic profiles for each breed. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 6 Warner et al., 2021; Purslow, Warner, Clarke, & Hughes, 2020). The level of myoglobin is also impacted by other oxidative enzymes. In particular, horse meat has a dark red color with a bluish sheen connected with a its large amount of myoglobin. Other proteins such as haemoglobin and cytochrome C also contribute to meat color (Seong et al., 2019). The light scattering is mainly due to variations in the myofilament lattice spacing, sarcomere length and sarcoplasmic protein distribution (Purslow et al., 2020). It can be highlighted that lightness would be determined by modifications in the surface charge of myofibrillar proteins in connection with myofilament lattice spacing governed by factors such as pH, protein solubility, and integrity of cytoskeletal proteins and myofilaments proteins (Purslow et al., 2020). All these facts demonstrate the complex relationships between proteins and meat color. As can be seen in Table 2, a large amount of the proteins (70 proteins) was significantly correlated with color instrumental parameters. Among these, several proteins from energy metabolism, calcium homeostasis and oxidative stress were correlated with L* such as HSPA1A (r =0.819 in Jaca Navarra), SOD2 (r = − 0.802 in Jaca Navarra), ALDOA (r =0.852 in Burguete), PKM (r =0.745 in Burguete), CKM (r =0.836, 0.834 in Burguete and Jaca Navarra); correlated with a* SOD2 (r =0.707 in Burguete), VDAC2 (r = − 0.786 in Burguete), PRDX2 (r =0.734 in Jaca Navarra), PFKM (r = − 0.717 in Burguete); and Fig. 3. Volcano plots in the comparison of two finishing diets (C =conventional concentrate and straw and S =silage and organic feed) for two breeds (Burguete and Jaca Navarra). The points with red colour located on the left represent the proteins significantly overexpressed in feeding with S, meanwhile on the right represent the proteins overexpressed in feeding with C. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 7 Table 2 Correlation between protein abundance and quality traits (shear force, color parameters and intramuscular fat) of foal meat from Burguete and Jaca Navarra breeds. Pearson correlation of differential abundant proteins with shear force Protein name Gene name Uniprot ID Coefficient, r p-value BURGUETE Obscurin. cytoskeletal calmodulin and titin-interacting RhoGEF OBSCN F6WQX2 −0.922 <0.001 Myosin binding protein C1 MYBPC1 F6Y6F3 −0.881 0.001 Myosin light chain 1 MYL1 F7AX54 −0.860 0.001 PDZ domain-containing protein AHNAK A0A3Q2HJF2 0.859 0.001 Filamin C FLNC F6ZWZ3 −0.850 0.002 Nebulin NEB A0A3Q2LNX8 0.848 0.002 Smoothelin like 1 SMTNL1 F6R563 0.828 0.003 PDZ and LIM domain 5 PDLIM5 A0A5F5PRT5 0.804 0.005 Creatine kinase CKM F7BR99 −0.797 0.006 Myomesin 1 MYOM1 F7CNU9 −0.796 0.006 LIM domain binding 3 LDB3 F6QY18 0.794 0.006 Actinin alpha 3 ACTN3 E7D103 −0.782 0.008 Moesin MSN F6PUX2 0.771 0.009 AIR carboxylase PAICS F7D7Z9 −0.766 0.010 Fructose-bisphosphate aldolase ALDOA A0A3Q2HN78 −0.752 0.012 Actin alpha 1 ACTA1 F7CZ92 −0.745 0.013 GLOBIN domain-containing protein HBA2 Q28383 0.745 0.013 Phosphoglycerate mutase PGAM2 F7A616 −0.739 0.015 Rho GDP dissociation inhibitor alpha ARHGDIA F6W039 0.734 0.016 Nucleoside diphosphate kinase NME2 F6YY66 −0.730 0.017 Myosin binding protein MYBPH F6VIP2 −0.727 0.017 Tryptophan–tRNA ligase WARS1 A0A3Q2KTL8 0.725 0.018 Titin TTN A0A5F5PKC5 −0.721 0.019 Glutathione transferase GSTO1 F6SA20 −0.718 0.019 Myosin binding protein C2 MYBPC2 F6VIG2 −0.715 0.020 Myomesin 2 MYOM2 A0A3Q2I9B4 −0.715 0.020 Profilin PFN1 F6UJ33 −0.708 0.022 Triosephosphate isomerase TPI1 F6TZS9 −0.707 0.022 Mono-ADP ribosylhydrolase MACROD1 A0A3Q2L2W7 −0.705 0.023 JACA Troponin C TNNC2 F7CGE8 0.858 0.002 Creatine kinase (Fragment) CKM A0A0R5SAL6 −0.793 0.006 Elongation factor 1-gamma EEF1G A0A452GER3 −0.784 0.007 NADH dehydrogenase [ubiquinone] flavoprotein 2. NDUFV2 F7B7S1 0.740 0.014 Adenylosuccinate synthetase isozyme 1 ADSS1 A0A3Q2GSI4 −0.735 0.015 Adipocyte-type fatty acid-binding protein FABP4 F6YN05 −0.711 0.021 Pearson correlation of differential abundant proteins with L* Protein name Gene name Uniprot ID Coefficient, r p-value BURGUETE Immmunoglobulin lambda light chain variable region IGL A0A0A1E9B4 −0.709 <0.001 Phosphoglycerate kinase PGK1 F7D1R1 0.867 0.001 Fructose-bisphosphate aldolase ALDOA A0A3Q2HN78 0.852 0.002 Creatine kinase CKM F7BR99 0.836 0.003 Mono-ADP ribosylhydrolase MACROD1 A0A3Q2L2W7 0.833 0.003 Alpha-1-acid glycoprotein 2 ORM1 F7CQ86 −0.827 0.003 Acylphosphatase ACYP2 A0A3Q2HXC0 0.794 0.006 GST class-pi GSTP1 F6VSN2 0.776 0.008 Protein-L-isoaspartate O-methyltransferase PCMT1 K9K455 −0.769 0.009 Succinate dehydrogenase [ubiquinone] iron-sulfur subunit SDHB Q0QEY2 −0.746 0.013 Pyruvate kinase PKM B3IVM0 0.745 0.013 Albumin ALB F7BAY6 0.735 0.016 Alpha-1B-glycoprotein A1BG F6VJR6 0.724 0.018 ATP synthase subunit beta ATP5F1B F6U187 −0.724 0.018 Ubiquitin-activating enzyme E1 UBA1 F7C6F2 −0.718 0.019 Glutathione transferase GSTO1 F6SA20 0.710 0.021 PDZ and LIM domain 5 PDLIM5 A0A5F5PRT5 −0.705 0.022 Carboxylic ester hydrolase CES1 A0A3Q2LAR7 0.702 0.024 JACA 3-hydroxyisobutyrate dehydrogenase HIBADH F6XXT9 −0.862 0.001 Bridging integrator 1 BIN1 A0A3Q2LCX2 −0.843 0.002 Creatine kinase CKM F7BR99 0.834 0.003 Heat shock 70 kDa protein 1A HSPA1A F7DW69 0.819 0.004 Protein-L-isoaspartate O-methyltransferase PCMT1 K9K455 −0.802 0.005 Superoxide dismutase SOD2 F6U991 −0.802 0.005 Elongation factor 1-alpha EEF1A2 F6Q4R1 0.795 0.006 Glyceraldehyde-3-phosphate dehydrogenase GAPDH A0A0K8TPT3 −0.794 0.006 Aspartate aminotransferase GOT2 F6YXZ8 −0.769 0.009 Myosin binding protein H MYBPH F6VIP2 −0.748 0.013 Aspartate aminotransferase GOT1 A0A5F5Q2J1 −0.747 0.013 (continued on next page) M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 8 Table 2 (continued) Pearson correlation of differential abundant proteins with shear force Protein name Gene name Uniprot ID Coefficient, r p-value Creatine kinase CKMT2 F6X8T2 −0.739 0.015 Myosin binding protein C1 MYBPC1 F6Y6F3 0.723 0.018 Medium-chain specific acyl-CoA dehydrogenase ACADM K9K3Y1 −0.719 0.019 Myosin binding protein C2 MYBPC2 F6VIG2 0.717 0.020 10 kDa heat shock protein HSPE1 F6T0A6 −0.715 0.020 Macrophage migration inhibitory factor MIF F6S243 −0.705 0.023 Pearson correlation of differential abundant proteins with a* Protein name Gene name Uniprot ID Coefficient, r p-value BURGUETE Enoyl-CoA hydratase 1 ECH1 F6S8M6 0.901 <0.001 Voltage-dependent anion-selective channel protein 2 VDAC2 F6TLU0 −0.786 0.007 ATP-dependent 6-phosphofructokinase PFKM A0A3Q2GST7 −0.717 0.020 Elongation factor 1-gamma EEF1G A0A452GER3 0.712 0.021 Desmin DES F7AQ43 −0.710 0.021 Superoxide dismutase SOD2 F6U991 0.707 0.022 Mitsugumin-53 TRIM72 A0A5F5PUG7 −0.706 0.023 Phosphoglycolate phosphatase PGP F6ZBD7 0.701 0.024 JACA GLOBIN domain-containing protein HBA2 Q28383 0.907 <0.001 Peptidyl-prolyl cis–trans isomerase PPIA F6X6A6 −0.852 0.002 Aspartate aminotransferase GOT1 A0A5F5Q2J1 0.851 0.002 Globin A1 GLNA1 A0A1K0FUE2 0.734 0.016 Peroxiredoxin PRDX2 F7BFT1 0.734 0.016 Myosin light chain 1 MYL1 F7AX54 −0.719 0.019 Citrate synthase CS F6XMS1 0.711 0.021 Malate dehydrogenase PHGDH K9K2D2 0.700 0.024 Pearson correlation of differential abundant proteins with b* Protein name Gene name Uniprot ID Coefficient, r p-value BURGUETE Alpha-1B-glycoprotein A1BG F6VJR6 0.829 0.003 Mitsugumin-53 TRIM72 A0A5F5PUG7 −0.778 0.008 Metavinculin VCL F6ZSZ5 −0.769 0.009 Fructose-bisphosphate aldolase ALDOA A0A3Q2HN78 0.763 0.010 Annexin ANXA7 F7ALC9 0.758 0.011 ST13 Hsp70 interacting protein ST13 A0A5F5PN24 −0.754 0.012 Aminopeptidase NPEPPS F7BXA6 −0.748 0.013 Proteasome subunit alpha type PSMA4 F6Z688 −0.739 0.015 Alpha-1-acid glycoprotein 2 ORM1 F7CQ86 −0.729 0.017 Cysteine and glycine-rich protein 3 CSRP3 A0A3Q2HGW3 −0.717 0.020 Mono-ADP ribosylhydrolase 1 MACROD1 A0A3Q2L2W7 0.713 0.021 Desmin DES F7AQ43 −0.702 0.024 JACA Protein-L-isoaspartate O-methyltransferase PCMT1 K9K455 −0.887 0.001 Myosin binding protein C1 MYBPC1 F6Y6F3 0.780 0.008 Rho GDP dissociation inhibitor alpha ARHGDIA F6W039 −0.777 0.008 Adenosylhomocysteinase AHCY A0A3Q2HUY8 0.751 0.012 Pearson correlation of differential abundant proteins with IMF Protein name Gene name Uniprot ID Coefficient, r p-value BURGUETE Actin alpha 1 ACTA1 F7CZ92 0.871 0.001 Moesin MSN F6PUX2 −0.769 0.009 60 kDa chaperonin HSPD1 K9K3I2 0.742 0.014 Myosin-1 MYH1 A0A3Q2HB31 0.734 0.016 PDZ and LIM domain 3 PDLIM3 F6RZN9 −0.717 0.019 Nucleoside diphosphate kinase NME2 F6YY66 0.707 0.022 NAD(P)(+)–arginine ADP-ribosyltransferase ART3 F6W836 0.706 0.023 Aldehyde dehydrogenase ALDH2 F6SL95 0.703 0.023 JACA 78 kDa glucose-regulated protein HSPA5 F6YG82 0.871 0.001 Fructose-bisphosphatase FBP1 F6YTD3 −0.867 0.001 Beta-1 metal-binding globulin TF A0A5F5PRC5 0.842 0.002 L-lactate dehydrogenase LDHB C6L1J5 0.815 0.004 Annexin ANXA1 K9K271 0.767 0.010 Hsc70/Hsp90-organizing protein STIP1 F7E1X2 0.755 0.012 Elongation factor 1-gamma EEF1G A0A452GER3 0.735 0.015 Alpha-1B-glycoprotein A1BG F6VJR6 0.725 0.018 Glycerol-3-phosphate dehydrogenase GPD1 A0A3Q2HV72 −0.721 0.019 Troponin C2 TNNC2 F7CGE8 −0.710 0.021 (continued on next page) M. L´ opez-Pedrouso et al. Food Chemistry 405 (2023) 134805 9 correlated with b* ALDOA (r =0.763 in Burguete). Among them, glycolytic enzymes were identified (ALDOA, PKM and PFKM) as well as CKM that buffers intracellular ATP concentrations through the storage and usage of high-energy phosphate species in the form of phosphocreatine. These results evidenced that these glycolytic proteins have a strong effect on color parameters resulting in ALDOA as the most important biomarker since it simultaneously influences both L* and b*. This agrees with the integromics study of Gagaoua et al. (2020) on beef color that shortlisted this putative biomarker as of great interest for further validation. Other proteins involved in oxidative stress (SOD2 and PRDX2) may affect the variation in different ways including the oxidative status of myoglobin, hence changing the chromatic traits. It also demonstrated that the expression of heat shock proteins is related to protection against oxidative stress (Kalmar & Greensmith, 2009). Additionally, these biomarker candidates were confirmed in beef using an integromics meta-analysis (Gagaoua et al., 2020). In the case of intramuscular fat, 21 proteins were correlated as shown in Table 2. In general, there is a connection among tenderness, intramuscular fat and water holding capacity. As shown in Table 2, five proteins resulted of importance: ACTA1 (r =0.871 in Burguete), TF (r = 0.842 in Jaca Navarra), FBP1 (r = − 0.867 in Jaca Navarra), LDHB (r = 0.815 in Jaca Navarra) and HSPA5 (r =0.871 in Jaca Navarra) for intramuscular fat. But they are not correlated with shear force. Thus, it is worthy to reveal in this study that proteomic differences jointly affect these important quality parameters. This fact may be related to the lowfat amount present in foal meat, avoiding the potential connection with tenderness. 4. Conclusions The present study reports that foal meat from Burguete seemed to have superior quality than Jaca Navarra in terms of tenderness, and color lightness. Additionally, the diet treatment with concentrate, increased the intramuscular fat along with a decrease in shear force, which could favour an increase in the consumer acceptability of horse meat. The proteomic approach applied in this study revealed significant changes and differences at the muscle level because of the breed effect. These were more pronounced than those caused by the finishing diet type. The metabolic shifts related to the finishing diet evidenced several a candidate protein biomarkers such as ATP5P0/ATP5F18 (ATP synthases), ALB, PGAM2, TPI1, MDH1 and FABP3. In terms of meat quality traits, several protein biomarkers were further identified. For tenderness, the main mechanisms were related to muscle structure with key proteins (ACTA1, MYBPH, MYL1 and TNNC1) including few others involved in energy metabolism (ALDOA, CKM, TPI1 and PGMA2). Regarding meat color, proteins from energy metabolism (ALDOA, PKM, PFKM and CKM), and proteins from oxidative stress and protein folding (HSPA1A, SOD2 and PRDX2) can be considered putative candidates for foal meat color determination. Additionally, 23 proteins were found as possible biomarkers for intramuscular fat content that need further investigation and validation. Table 2 (continued) Pearson correlation of differential abundant proteins with shear force Protein name Gene name Uniprot ID Coefficient, r p-value Heat shock 70 kDa protein 1A HSPA1A F7DW69 0.709 0.022 Moesin MSN F6PUX2 0.706 0.023 Nucleoside diphosphate kinase NME2 F6YY66 0.704 0.023 Fig. 4. Protein-Protein Interaction network of the 33 proteins correlated with shear force using Search Tool for the Retrieval of Interacting Genes/proteins (STRING). These proteins of foal meat also were differentially abundant between Burguete and Jaca Navarra. M. L´ opez-Pedrouso et al.