Color and Tenderness Proteomic Biomarkers of Beef From Steers Fed Sorghum Substituted for Maize

Experimental diets, animal management, and design

Whole red grain sorghum was sourced from a commercial producer (AGT, Cape Town, South Africa) and milled through a 0.5 mm sieve. Three pelleted (5 mm × 30 mm; 45°C) maize-lucerne-based complete diets containing either 0 (SGD-0), 200 (SGD-200), or 400 (SGD-400) g/kg DM of sorghum grain substituting white maize grain sourced from Perdigon Proprietary Limited, Paarl, South Africa as a primary energy source were formulated by commercial feed producer. The diets were isoenergetic and met the nutritional requirements of growing steers (Tables 1 and 2; National Research Council, 2016). Though the diets were not isonitrogenous, findings from a companion study (Njisane et al., submitted) shows that protein intake was similar (P > 0.05) across diets. Twenty-one Angus steers (7 mo; 230 ± 28 kg) were purchased from a single commercial producer, and each steer was housed in an individual concrete floor pen (2 m × 4 m) covered with straw. The steers were adapted for 21 d followed by 90 d of feeding.

Slaughter procedures and meat sampling

The steers were slaughtered at a commercial abattoir, 64 km from the experimental farm. During slaughter, steers were stunned using a non-penetrating captive-bolt and exsanguinated according to South African meat law (South African Government Gazette, 2000). Twenty-four h postmortem, the pH of all carcasses was measured using a mobile pH meter with a built-in Pt1000 temperature sensor for automatic temperature compensation (CrisonTM pH meter PH 25+, Lasec, South Africa) in the 12^th and 13^th right rib region. Then, the left longissimus thoracis et lumborum (LTL) muscle was removed from each carcass from the 9^th to 13^th rib. Thereafter, a two-gram cube was sampled from 3 loins per treatment, dipped in liquid nitrogen, and stored (−80 °C) in 15 ml tubes until proteomic analysis.

Color and shear force assays

Muscle color samples were allowed to bloom for 30 min, and 3 readings were taken per sample. Color coordinates (L*, lightness; a*, redness; b*, yellowness) were measured using a BYK-209 Gardner GmbH (Gerestried, Germany) colorimeter set to sample mode and calibrated with D65/10° against black and white tiles using observer settings and an 11-mm diameter aperture (AMSA, 2012). Warner-Bratzler shear force (WBSF) was determined using a ≈100 g meat sample cooked in a water bath at 80°C to an internal temperature of 75°C (AMSA, 2015). A thermocouple probe affixed onto a digital monitor (Testo 176T4, South Africa) was injected into the center of an individual sample (n = 5) to monitor internal temperature throughout the cooking process.

Proteomic characterization

Sample preparation, protein extraction and quantification. Frozen samples were ground into a fine powder using liquid nitrogen in a stainless-steel blender. The ground material (0.5 g) was resuspended in 1 mL extraction buffer (4 M Urea, 2 M Thiourea; 2.5 mM Ethylenediaminetetraacetic acid [EDTA]; 5 mM 1,4 Dithiothreitol [DTT]; 2% glycerol), thoroughly vortexed for 30 s, incubated 40 min at −20°C, and centrifuged at 15,000 × g for 20 min at 4°C. The supernatant containing the protein extract was transferred to clean tubes, and protein concentration was quantified using the Bradford assay with bovine serum albumin as a standard (Kruger, 2009). All the samples were extracted in triplicate.

SDS-PAGE protein visualization and quantitation. The sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) protocol was deployed to determine the quality and purity of the extracted proteins. Briefly, proteins (30 μg) were prepared and incubated as described in Mahlare et al. (2023). Proteins were heated at 70°C for 10 min and resolved as described by Mahlare et al. (2023). Following a 12% SDS-PAGE gel electrophoresis, proteins were visualized using Coomassie brilliant blue stain, and the gels were also processed as defined by Mahlare et al. (2023). Protein expression densitometric analysis of bands was performed using AlphaEase FC software (AlphaImager^™ IS-2200 for Windows 2000/XP; Version 4), and default background subtraction for relative intensity correction was activated to only record analyte signals. The protein band intensities were expressed as integrated density values (IDV), presented as averages of 3 technical replicates.

In-gel digest, peptide extraction, and clean-up. Protein bands of interest were excised excluding strongly stained bands (Figure 1), and gel lanes of interest were uniformly cut into smaller gel cubes (≈ 1.0 mm × 1.0 mm) using a surgical blade and an A4 cutting board with a 20 cm aluminum ruler, and de-stained using 50% acetonitrile in 100 mM Ammonium Bicarbonate (NH₄HCO₃) until clear, and dehydrated with 100% acetonitrile for 5 min, with shaking. Dehydrated gel cubes were then reduced in 2 mM tris 2-carboxyethyl phosphine (TCEP) in 25 mM NH₄HCO₃ at room temperature (RT) for 15 min. Alkylation was conducted with 20 mM iodoacetamide in 25 mM NH₄HCO₃ for 30 min in the dark at RT. Following this, the gel cubes were dehydrated with 100% acetonitrile before rehydration with trypsin solution (10 ng/μl in 25 mM NH₄HCO₃) for 45 min at 4°C. After rehydration, excess trypsin solution was aspirated, and the gel cubes incubated in 50 μl 25 mM NH₄HCO₃ to enable overnight digestion at 37°C. Peptides were extracted from the gel cubes with 100 μl milli Q water by vortexing for 45 min. The supernatant was transferred to 1.5 ml Lo-bind Eppendorf tubes and the extracts dried down in a Centrivap Benchtop Vacuum Concentrator (Centrivap; Labconco Corporations, Kansas City, MO, USA) at RT before resuspension in 0.1% formic acid (FA). Samples were further purified and desalted as described by Mahlare et al. (2023) with slight modification. StageTips were conditioned with 10 μL of acetonitrile and equilibrated using 10 μL Buffer A (2% ACN/0.1% FA). Samples were then loaded onto and passed through the StageTip once before the tip was washed using 10 μL Buffer A. Peptides were eluted in 10 μL Buffer B (50% ACN/0.1% FA) and the peptide extracts pooled before the eluates were evaporated in a Centrivap at RT. Before liquid chromatography analysis, the dried peptide eluates were reconstituted in 15 μl Buffer A for liquid chromatography analysis.

Figure 1.

Figure 1.

A representative SDS-PAGE gel of myofibrillar filaments of LTL beef from steers finished with sorghum-based diets. About 30 μg protein was loaded on to 12% SDS-PAGE gel. Standard represents the molecular weight marker; CD represents the control diet (SGD-0); SGD1 represents 200 g/kg DM of sorghum diet; SGD2 represents 400 g/kg DM of sorghum diet; the yellow box represents actin; the red box represents marked area of changes in protein intensity. The mean protein band intensities are expressed as integrated density values (IDV; SGD-0: 31864, SGD-200: 28538, SGD-400: 26107; Standard error of means: 5158.9) and area (SGD-0: 1246, SGD-200: 1208; SGD-400: 1119; Standard error of means: 67.6).

Liquid chromatography with tandem mass spectrometry. The applied liquid chromatography with tandem mass spectrometry (LC-MS/MS) proteomic investigation was adapted from Hooijberg et al. (2018). The detailed procedures employed for peptide separation, protein quantification and characterizing, respectively achieved by liquid chromatography and mass spectrometer, are described in Mahlare et al. (2023). Briefly, peptide separation was initiated using loading solvents (Solvent A: 2% acetonitrile: water, 0.1% FA; Solvent B:100 acetonitrile: water, respectively) to deposit the samples into the trap column then onto the analytical column, and chromatography was performed at 40°C (Thermo Scientific UltiMate 3000 HPLC; Thermo Scientific, Rockford, IL, USA). The outflow was emitted onto the MS (Thermo Scientific Fusion MS:Nanospray Flex ionization source; Thermo Scientific, Rockford, IL, USA) for quantifying and characterizing, and the data attained as a weighted average of the mass peak (centroid mode).

Data processing, protein identification, and bioinformatics analysis. The raw data files were exported using Thermo Proteome Discoverer 1.4 (2012 Thermo Fisher Scientific Inc., Thermo Scientific, United States), and the Sequest and Amanda algorithms were applied for further processing. The resultant files were interrogated with reference to the Universal Protein (uniport) “Bos Taurus reviewed” concatenated database as detailed in Mahlare et al. (2023). Further validation was achieved in Scaffold Q+ (Settings: 95% Protein identification probability, 1% false discovery rate [FDR] protein threshold, 2 minimum number of peptides; www.proteomesoftware.com; accessed 14 August 2023). The functions of the identified proteins were assessed by mapping to Uniprot Resource (https://www.uniprot.org/id-mapping; accessed 14 August 2023). The bioinformatics webtool was used to visualize the common proteins identified in each diet (http://bioinformatics.psb.ugent.be/webtools/Venn/; accessed 14 August 2023).

Gene ontology (GO) and annotation enrichment analysis were done using the GO resource (http://geneontology.org/; accessed 14 August 2023), Panther classification system (https://pantherdb.org/; accessed 14 August 2023), String (https://string-db.org/; accessed 14 August 2023), and Kyoto Encyclopedia of Genes and Genomes (KEGG). Fisher’s exact test was applied to determine the significance level of the protein enrichment of a specified GO term (FDR; P < 0.05). For protein-protein interactions (PPI), default settings (i.e., medium confidence of 0.4 and active interaction sources: text-mining, experiments, databases, co-expression, neighborhood, gene fusion, and co-occurrence) were used. The PPI networks were further clustered using the Markov cluster algorithm (MCL), with the inflation parameter set at 2 and PPI enrichment value at P < 0.05, while also retaining the floating proteins. To categorize the proteins intrinsic in beef tenderness and color development, the identified protein gene names were compared with published literature data for tenderness and color elaborated by Picard and Gagaoua (2020) and Gagaoua et al. (2020b), respectively.

Statistical analysis

The GLIMMIX procedure of SAS (version 9.4; SAS Institute Inc. Cary, NC, USA) was used to analyze physical attribute data with diet as fixed effect and animal a random factor. All data were subjected to the Shapiro-Wilk test (Shapiro & Wilk, 1965) for normality. A Tukey’s test was used to qualify the least-squares means as significantly different at P ≤ 0.05 and tendencies at 0.05 < P ≤ 0.10.

Physical attributes, proteomic bioinformatics, and protein expression

The inclusion of sorghum in beef finisher diets exhibited a tendency (P = 0.083) to increase ultimate pH and linearly increased (P < 0.05) WBSF values but had no effect (P > 0.05) on color (Table 3). Protein loading appeared consistently even, with no visible protein streaking in the meat protein extracts (Figure 1). The explored data incorporate details of the molecular weight and the quantity of protein extract of each treatment. The molecular weight of the protein bands ranged between 10 to 250 kDa for all the treatment extracts. Myosin heavy chain (MHC; ≈230 kDa) and actin (≈42 kDa) were the most dominant bands across all treatments. The intensity protein bands were similar (P > 0.05) across diets as shown by the IDV and area (Figure 1).

A total of 692 bovine proteins were identified, and 108 were common to all diets, while 74, 60, and 56 were respectively unique to treatments SGD-0, SGD-200, and SGD-400 (Figure 2). Of 108 common proteins, 11 were differentially expressed of which 7 edit to 8 (PYGM, PYGM, MYH1, MYH8, HSPA8, CAPZB, HSPB6, and PEBP1) are associated with both tenderness and color while 4 edit to 3 (PARP6, HSPOAA1, and GYS1) are only associated with color (Table 4). A total of 8 tenderness-regulating proteins were identified to be unique to diets, with 3 for SGD-0 (RABGGTA, HSPA5, and APOBEC2), 2 for SGD-200 (MYL3 and YWHAE), and 3 for SGD-400 (HSPA9, PDIA3, and ANKRD2). A sum of 9 color-regulating proteins were unique to diets, with 3 in SGD-0 (OXCT1, HSPA5, and MYH7), 4 in SGD-200 (MYH2, PDHX, LAP3, and P4HB), and 2 in SGD-400 (MYH1 and HSPA9). Of the 11 differentially expressed proteins (Table 4), 6 (MYH8, MYH1, HSP90AA1, GSY1, HSPB6, and HSPA8) were downregulated (P < 0.05) by sorghum diets (SGD-200 and SGD-400). Proteins PARP6 and PYGM were only downregulated (P < 0.05) in SGD-400. The CAPZB was only upregulated (P < 0.05) in SGD-400, while PEBP1 was upregulated in sorghum diets.

Figure 2.

Figure 2.

Number of proteins identified in each of the treatments visualized by bioinformatics webtool (http://bioinformatics.psb.ugent.be/webtools/Venn/, accessed 14 August 2023).

Gene ontology functional and pathway enrichment analyses

The differentially expressed as well as diet-unique proteins for tenderness and color were clustered into biological processes, molecular function, and cellular component (Figure 3). The cellular anatomical entity, cellular process, catalytic activity, and binding activities dominated the biological function, molecular function, and cellular component GO clusters, in that order (Figure 3). The GO clustering of diet-specific proteins revealed that the biological process function group was dominated by cellular process with SGD-0, SGD-200, and SGD-400 having 38%, 36%, and 38%, respectively (Figure 4). The binding (48%, 44%, and 53%, respectively) and catalytic activities (40%, 42%, and 37%, respectively) were highly represented in the molecular functions (Figure 4). Cellular anatomical entity (72%, 71%, and 81%, respectively) predominated the cellular component category (Figure 4). The pathways were dominated by HSPA8, HSPA9, HSPA5, and YWHAE clustered under Parkinson disease (Figure 5). Chaperones contributed 36% of the protein classes with differentially expressed HSPB6, HSP90AA1, HSPA8, and diet-specific HSPA5 (SGD-0), P4HB (SGD-200), PDIA3, and HSPA9 (SGD-400; Figure 5). Cytoskeletal proteins (32%) included differentially expressed MYH1 and CAPZB, as well as MYH7 (SGD-0), MYH2, and MYL3 (SGD-200; Figure 5). Metabolic interconversion enzymes (18%) included differentially expressed PYGM along with RABGGTA and OXCT1 (SGD-0), and PDHX (SGD-200; Figure 5). The extensive pathways were cytoskeletal regulation by Rho GTPase (16%), signaling pathway (16%), inflammation mediated by chemokine and cytokine signaling pathway (16%), and nicotinic acetylcholine receptor signaling pathway (16%).

Figure 3.

Figure 3.

Gene ontology of differentially regulated and unique proteins associated with color and/or tenderness of beef LTL from steers fed graded levels of sorghum analyzed by Panther (http://www.pantherdb.org, accessed on 14 August 2023).

Figure 4.

Figure 4.

Functional distribution of differentially and uniquely expressed protein associated with color and/or tenderness of beef LTL from steers fed finisher diets containing increasing levels of sorghum; 0 g/kg DM SGD (number of proteins = 17), 200 g/kg DM SGD (number of proteins = 17), 400 g/kg DM SGD (number of proteins = 15) (http://www.pantherdb.org, accessed on 14 August 2023).

Figure 5.

Figure 5.

Metabolic pathways and protein classes linked to differentially regulated and unique proteins associated with color and/or tenderness of beef LTL from steers fed finisher diets containing increasing levels of sorghum by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (http://www.pantherdb.org, accessed on 14 August 2023). EGF, Epidermal Growth Factor; FGF, Fibroblast Growth Factor.

Protein-protein interaction networks and KEGG pathways

The PPI revealed that GYS1, PYGM, MYH1 and MYH8 were interconnected within the energy metabolism and muscle activity pathway, while HSPB6, HSPA8, HSP90AA1, and CAPZB were interlinked in the cellular response to stress and modification pathways (Figure 6A). The aforementioned pathways remained distinct when PPI of differentially expressed and diet-unique proteins were analyzed (Figure 6B, 6C, and 6D) with only 2 extra pathways (muscle contraction: MYH2, MYL3, MYH1, and MYH8; uncategorized network: PARP6 and PDHX) in the SGD-200 diet. Three enzymes (PYGM, GYS1, and PARP6) were identified to be linked to the glycolytic pathway, with PYGM and GYS1 being involved in glucose metabolism (Figure 7A) and PARP6 in the pyruvate pathway (Figure 7B). Two enzymes are involved in protein degradation; LAP3 breaks down peptides to proline (Figure 7C) and OXCT1 is linked to the interconversion of acetoacetate to acetoacetyl-Coa in the degradation of valine, leucine, and isoleucine (Figure 7D).

Figure 6.

Figure 6.

Protein–protein interaction networks of differentially regulated and unique proteins associated with color and/or tenderness of beef LTL from steers fed finisher diets containing increasing levels of sorghum as analyzed by string 12.0. The nodes were proteins from Bos Taurus database, and the lines were the predicted interactions with light blue line – curated databases; purple line – experimentally determined; green line – gene neighborhood; red line – gene fusions; dark blue line – gene co-occurrence; light green – textmining; black line – co-expression; light purple line – protein homology (https://string-db.org/ accessed 14 August 2023).

Figure 7.

Figure 7.

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis linked to differentially regulated and unique proteins associated with color and/or tenderness of beef LTL from steers fed finisher diets containing increasing levels of sorghum showing enzymes linked to A glucose metabolism by converting UDP-glucose to amylose and glycogen to α-D-glucose-1P; B converting phosphoenolpyruvate to pyruvate; C arginine and proline metabolism; and D valine, leucine, and isoleucine degradation. Shaded area is the position in the pathway occupied by the enzyme identified in beef LTL fed graded levels of sorghum. EC, enzyme commission number.