Rapid Peptidome Enrichment Coupled With Label-Free Quantification to Characterize Proteome Degradation and Proteolytic Cleavage in Postmortem Beef Longissimus Lumborum Muscle

Sample collection

Eight left sides of carcasses from market-age Angus steers were randomly selected from a same US Department of Agriculture-inspected beef processing facility. Approximately 0.5 g of muscle tissue was collected from the central cross-section of the LL (n = 8; between the 12th and 13th rib section) using a biopsy device (14-gauge × 9 cm; Argon Medical Devices, TX) at 2 h, 48 h, and 336 h (14-d) postexsanguination, resulting in a collection of a total of 24 samples (8 animals at 3 different postmortem time points). The carcasses were kept at 2°C in the dark during the whole collection period. The collected muscle samples were frozen immediately in liquid nitrogen until proteomic analyses. The postrigor pH value of the chosen LL was between 5.45 and 5.55.

Peptide extraction and enrichment

One hundred and thirty milligrams of frozen muscle tissue were homogenized in 0.9 mL ice-cold urea-based buffer (8.3 M urea, 2 M thiourea, 50 mM tris-HCl, 60 mM dithiothreitol pH 8.0) by a homogenizer for five 20-s bursts on ice and sonicated by a sonic dismembrator for five 1-s pulses. The sample was shaken for 60 min at 4°C and centrifuged at 16,000 × g for 10 min. One volume of supernatant was collected and mixed with 1 volume of 20% (W/V) trichloroacetic acid (TCA), followed by a 10-s vortex. The mix was shaken for 60 min at 4°C and centrifuged at 16,000 × g for 10 min to collect supernatant peptide. The concentration of the enriched peptide was determined using the Bradford Protein Assay Kit (Bio-Rad, #5000205) through a Shimadzu UV-1900i UV-VIS spectrophotometer (Shimadzu Inc., Candy, OR).

Untargeted analysis of bovine peptidomic isolates via tandem mass spectrometry

Peptide extraction was desalted over C18 resin spin columns (Pierce #89870), dried to completion, resuspended in a minimal volume of Fisher Optima grade water with 0.1% formic acid (Solvent A), and quantified by absorbance at 280 nm. The injection amount was normalized such that the same total protein mass was analyzed for every sample. All samples were analyzed using a Thermo Scientific Ultimate 3000 RSLCnano UPLCUPL system coupled to a high-resolution Thermo Scientific Orbitrap Eclipse Tribrid mass spectrometer. Each sample was injected onto a nanoEase M/Z Peptide BEH C18 analytical column (1.7 μm, 75 μm  ×  250 mm, Waters Corporation) in Solvent A and separated by reversed-phase UPLC using a gradient of 4% to 30% Solvent B (0.1% formic acid in Fisher Optima LC/MS grade acetonitrile) over a 60-min gradient at 300 nL/min flow, followed by column wash and re-equilibration over 30 min, for a 90-min total gradient. Peptides were eluted directly into the Orbitrap Eclipse using positive mode nanoflow electrospray ionization with a capillary voltage of 2200 V. MS1 scans were acquired at 120,000 resolution, with an automatic gain control (AGC) target of 4e5, maximum ion time set to auto, RF lens setting of 30%, and a scan range of 300 m/z to 1800 m/z. Data-dependent MS2 scans were acquired in the Orbitrap using the following parameters: resolution of 15,000, with an intensity threshold of 5.0e4, AGC target set to standard, maximum ion time set to auto, isolation window set to 2.0 m/z, and a fixed cycle time of 3 s. A decision tree was implemented for fragmentation type, with high charge state and lower m/z ions prioritized for electron-transfer dissociation (ETD) fragmentation and lower charge state ions prioritized for higher-energy collisional dissociation (HCD) fragmentation. ETD was performed for charge states +3 − +8 with charge-dependent parameters and no supplemental activation applied. HCD was performed for charge states +2 − +5 with a collision energy of 30%. Dynamic exclusion was set to exclude after 1 observation for 30 s using a 10-ppm tolerance window.

Raw data were searched using MaxQuant v2.0.2.0 and the embedded Andromeda search engine and quantified using label-free methods (Cox and Mann, 2008). Peptide and protein identifications were achieved using the Uniprot Bos Taurus reference proteome (accessed 25052022) and a common protein contaminants database provided by MaxQuant. Variable modifications included oxidation of Met, acetylation of the protein N-terminus, and deamidation of asparagine (Asn) and glutamic acid residues. No fixed modifications were included. The digestion mode was set to “unspecific,” and the peptide lengths used included a minimum of 8 and a maximum of 25 AA. All results were filtered to a 1% false discovery rate at the peptide-spectrum-match (PSM) and protein levels using the target-decoy approach. All other parameters were left as default values. MaxQuant results were uploaded into Scaffold (v5, Proteome Software) for further data visualization and analysis.

Precursor proteome abundance, cleavage abundance, and statistical analysis

The peptide output from Scaffold was analyzed in R. The total proteolytic peptide abundance of each precursor protein was adjusted by the percent coverage of the protein sequence to estimate the abundance of the degraded precursor protein. Cleavage sites, cleavage residues in P1-P1’ positions (P1: first amino AA at N-terminal direction of the cleaved peptide bond; P1’: first AA at C-terminal direction of the cleaved peptide bond), and cleavage abundances were extracted from the quantified proteolytic peptide sequence. Cleavage sites and residues generated from N-terminal methionine excision, a protein cotranslational modification process (Bradshaw et al., 1998; Gamerdinger et al., 2023), were removed before further analysis. Precursor protein abundance, proteolytic peptide abundance, cleavage-site abundance, and cleavage residue abundance were analyzed separately using the limma package in R with consideration of sample correlation from the same animal (Ritchie et al., 2015). In brief, after log₂ transformation of the raw abundances, the abundances of precursor proteins, proteolytic peptides, cleavage sites, and cleavage residues were compared among different postmortem time points (48 h vs. 2 h; 336 h vs. 48 h; 336 h vs. 2 h). Differentially abundant precursor proteins (DAPP), differentially abundant peptides (DAP), differentially abundant cleavage sites (DACS), and differentially abundant cleavage residues (DACR) were determined if their Benjamini–Hochberg multiple testing adjustment significance at a P value of less than .05, raw fold change of greater than 2 (log₂ fold change > 1 or < −1), and quantification exist in at least 1/3 of the sample population. Precursor proteins, peptides, cleavage sites, and cleavage residues with exclusive absence or presence at a single time point were further filtered by identification frequency (≥50% for presence; 0% for absence) to determine exclusively present/absent precursor proteins (EPPP/EAPP), exclusively present/absent peptides (EPP/EAP), exclusively present/absent cleavage sites (EPCS/EACS), and exclusively present/absent cleavage residues (EPCR/EACR). Subcellular proteolysis, top 20 degraded precursor proteins, and P1-P1’ residue cleavage during 2 postmortem periods (2–48 h and 48–336 h) were estimated based on raw proteolytic peptide abundance increases. In the Universal Protein Resource (UniProt) protein database, multiple protein sequences (accession entries) exist for 1 protein. When multiple accession entries were matched by exclusively unique proteolytic peptides, the matched canonical protein sequence (mostly Uniprot accession entries reviewed by SWISS-PROT; [Bairoch and Apweiler, 1997]) was used for annotation. All the matched accession entries were listed in the supplementary files, and the entries used for annotation were underlined.

Overview of precursor proteins mapped by the quantified peptides

The quantified peptides mapped to 140 proteins at 2 h postmortem, 178 proteins at 48 h postmortem, and 182 proteins at 336 h postmortem. A total of 91 mapped proteins were quantified irrespective of postmortem time, and some mapped proteins overlapped between postmortem times (Figure 1.A). In addition, 72 mapped proteins were identified as DAPP (P < .05) in comparisons between the 3 postmortem time points. Specifically, there were 42 DAPP (34 increased and 8 decreased) in comparison of 48 h versus 2 h postmortem, 47 DAPP (39 increased and 8 decreased) in comparison of 336 h versus 48 h postmortem, and 48 DAPP (40 increased and 8 decreased) in comparison of 336 h versus 2 h postmortem (Table 1). Among DAPP, 17 precursor proteins were identified as DAPP irrespective of the comparisons. Within 1 of 3 postmortem time points, 46 precursor proteins were identified as EPPP or EAPP with high observation frequency (Figure 1.A). There are 18 EAPP and 7 EPPP at 2 h, 3 EPPP at 48 h, and 15 EPPP and 3 EAPP at 336 h (Table 1). Among 111 differential precursor proteins (DAPP, EPPP, or EAPP), 7 precursor proteins were identified as both DAPP and EAPP (Table 1).

Figure 1.

(A) Distribution of differential precursor proteins in the quantified precursor proteome at 2 h, 48 h, and 336 h postmortem. DAPP, differentially abundant precursor proteins; EAPP, exclusively absent precursor proteins; EPPP, exclusively present precursor proteins. (B) Subcellular localization of periodic precursor proteome degradation based on proteolytic peptide quantification. SR, sarcoplasmic reticulum. (C) Top 20 precursor proteins with the most proteolytic peptide abundance increase during each postmortem time period. Note: The references for the full name and biological function of precursor protein are provided in the results and discussion section, Table 1, and Supplementary Table 1. (D) Distribution of differential peptides in the quantified peptidome at 2 h, 48 h, and 336 h postmortem. DAP, differentially abundant peptides; EAP, exclusively absent peptides; EPP, exclusively present peptides. (E) Distribution of differential cleavage sites in all quantified cleavage sites at 2 h, 48 h, and 336 h postmortem. DACS, differentially abundant cleavage sites; EACS, exclusively absent cleavage sites; EPCS, exclusively present cleavage sites.

These mapped 244 proteins involved in energy metabolism (Figure 2; 53 proteins; glycogen and phosphagen metabolism [15 proteins], oxidative phosphorylation [(OXPHOS); 12 proteins]; TCA cycle-related metabolism [9 proteins], and transport proteins [7 proteins]), cellular structure (Figure 3; 71 proteins; myofibril thick filament [13 proteins], thin filament [12 proteins], Z-disk and intermediate filament [25 proteins], M-band [6 proteins], and cytoskeleton proteins [15 proteins]), cellular homeostasis (Figure 4; 51 proteins; ionic and membrane homeostasis [21 proteins], chaperone and cochaperone [16 protein], mitochondrial and redox metabolism [14 proteins]), nuclear function and protein turnover (Figure 4; 57 proteins; nuclear and transcription [32 proteins], ribosomal and translation [14 proteins], ubiquitin-proteasome and proteases [11 proteins]), and other functions (12 proteins).

Figure 2.

Precursor proteins related to energy metabolism and transport. Note: The full name of precursor protein is provided in the results and discussion section, Table 1, and Supplementary Table 1.

Figure 3.

Precursor proteins located in the myofibril and cytoskeleton. Note: The full name of precursor protein is provided in the results and discussion section, Table 1, and Supplementary Table 1.

Figure 4.

Precursor proteins related to cellular homeostasis, nuclear function, and protein turnover. The full name of precursor protein is provided in the results and discussion section, Table 1, and Supplementary Table 1.

Precursor proteins with significant degradation during 2 postmortem time periods

Precursor protein abundances for β-enolase (ENO3), creatine kinase M-type (CKM), MYH2 and chain 7 (MYH7), skeletal muscle ACTA, troponin I fast muscle form (TNNI2), troponin T fast skeletal muscle form (TNNT3), PDZ and LIM domain protein 5 (PDLIM5), LIM domain-binding protein 3 (LDB3), sarcoplasmic reticulum calcium ATPase 1 (ATP2A1), heat shock protein β-6 (HSPB6), and 4 and 1/2 LIM domains protein 1 (FHL1) were quantified at all 3 postmortem time points and had 2 significant increases from 2 h to 48 h and 48 h to 336 h postmortem. Precursor protein abundances for protein phosphatase 1 regulatory subunit 14C, ATP-dependent 6-phosphofructokinase (PFKM), phosphoglycerate kinase 1 (PGK1), NEB, myozenin-3 (MYOZ3), telethonin (TCAP), synaptopodin-2 (SYNPO2), PDZ and LIM domain protein 7 (PDLIM7), sarcoplasmic reticulum calcium ATPase 2 (ATP2A2), elongation factor 1-α 2 (EEF1A2), and COP9 signalosome complex subunit 9 changed from unquantifiable at 2 h to quantifiable at 48 h postmortem and then had significant increases from 48 h to 336 h postmortem. These results indicated that significant degradation of these proteins occurred from 2 h to 336 h postmortem.

Significant increases in precursor protein abundance were identified for phosphorylase b kinase regulatory subunit α, L-lactate dehydrogenase A chain (LDHA), complex IV subunit 4 isoform 1 (COX4I1), myoglobin (MB), troponin I slow muscle form (TNNI1), tropomyosin β chain (TPM2), MYOZ1, synemin (SYNM), bridging integrator 1 (MYC box-dependent-interacting protein 1; BIN1), PDZ and LIM domain protein 3 (PDLIM3), myomesin 1 (MYOM1), β-actin (ACTB), histone H3.1 (HIST1H3C), histone H2B (H2B), LIM and cysteine-rich domains protein 1 (LMCD1), 60S/large ribosomal subunit L13a (RPL13A), and CDGSH iron-sulfur domain-containing protein 1 from 2 h to 48 h but not from 48 h to 336 h postmortem. Protein abundances for protein phosphatase 1 regulatory subunit 3A, complex I subunit B1 (NDUFB1), troponin T slow skeletal muscle form (TNNT1), MYOZ2, synaptopodin 2-like protein (SYNPO2L), myotilin (MYOT), PDZ and LIM domain protein 1 (PDLIM1), hepatoma-derived growth factor, cell division cycle 25B, SET and MYND domain-containing 1, nucleolar protein 3, and tripartite motif-containing protein 72 changed from unquantifiable at 2 h to quantifiable at 48 h postmortem, without significant increases from 48 h to 336 h postmortem. These results indicate that significant degradation of these proteins happened between 2 h and 48 h postmortem.

Significant increases in precursor protein abundance were identified for protein phosphatase 1 regulatory subunit 1A and subunit 2, glycogen phosphorylase-muscle form (PYGM), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), MYH1 and MYH3, myosin binding protein C1 (MYBPC1) and C2 (MYBPC2), cardiac muscle α-actin, vimentin (VIM), myomesin-2 (MYOM2), thymosin β-10 (TMSB10), histone H2A, and heat shock protein β-1 (HSPB1) from 48 h to 336 h but not from 2 h to 48 h postmortem. Protein precursor abundances for glycogenin 1, α-enolase, bifunctional phosphoribosylaminoimidazole carboxylase, adenylosuccinate synthetase 1 (ADSS1), myosin binding protein H (MYBPH), MYL3, CFL2, leiomodin-3 (LMOD3), nonhistone chromosomal protein HMG-17 (HMGN2), AT-rich interaction domain 1B (ARID1B), elongation factor 1-α 1 (EEF1A1), stress-induced-phosphoprotein 1 (STIP1), and flavin reductase (BLVRB) changed from unquantifiable at 2 h and 48 h to quantifiable at 336 h postmortem. These results indicate that significant degradation of these proteins happened between 48 h and 336 h postmortem.

Precursor protein abundances of many proteins (phosphoglucomutase-1 [PGM1]; mitochondrial complex V ATP synthase subunit β [ATP5F1B], subunit e [ATP5ME], and synthase-coupling factor 6 [ATP5P]; mitochondrial malate dehydrogenase [MDH2]; skeletal muscle isoform myosin light chain 1/3 and regulatory light chain 2; small muscular protein [SMPX]; β-sarcoglycan, γ-sarcoglycan, myeloid-associated differentiation marker, microtubule-associated protein [MAP4], and erythrocyte membrane protein band 4.1 like 2; histone 4; caveolin-1 [CAV1]; and calpain catalytic subunit 1 [CAPN1], CAPN3, and tripeptidyl-peptidase 1 [TPP1]) significantly decreased or changed from quantifiable to unquantifiable when the postmortem time point extended. This might indicate that the degradation of these proteins becomes more limited when the postmortem period extends and warrants further investigation.

Quantified proteolytic peptides indicate the primary subcellular location of the postmortem proteolytic process

Among the 3860 quantified peptides in this study (Supplementary Table 2), there were 400 peptides quantified at 2 h postmortem, 1307 peptides quantified at 48 h postmortem, and 3150 peptides quantified at 336 h postmortem. A total of 125 peptides were quantified irrespective of postmortem time, and many quantified peptides overlapped among timepoints (Figure 1.D). In addition, 283 peptides were identified as DAP (P < .05) in comparisons among the 3 postmortem time points. There were 73 DAP (65 increased and 8 decreased) at 48 h as compared to 2 h postmortem, 224 DAP (204 increased and 20 decreased) at 336 h as compared to 48 h postmortem, and 80 DAP (71 increased and 9 decreased) at 336 h as compared to 2 h postmortem (Figure 1.D). There were 17 peptides identified as DAP irrespective of the comparisons. Within 1 of 3 postmortem time points, 1043 peptides were identified as either EAP or EPP with high observation frequency (Figure 1.D). Specifically, there were 31 EPP and 175 EAP at 2 h, 76 EPP at 48 h, and 743 EPP and 18 EAP at 336 h postmortem (Figure 1D). A total of 1222 quantified peptides were identified as either DAP, EAP, or EPP (Supplementary Table 2), and 104 peptides were recognized as both DAP and EAP. Of these 1222 peptides, 1191 peptides were mapped to the 111 differential precursor proteins (DAPP, EAPP, or EPPP; Table 1), leaving 31 peptides matched to 17 precursor proteins that were neither DAPP, EAPP, nor EPPP (Supplementary Table 2).

To quantify the subcellular proteolytic process, we summarized the raw peptide abundance increase based on the subcellular location of their precursor proteins during each postmortem period. From 2 h to 48 h postmortem, the increases in overall proteolytic peptide abundances were mainly from the myofibrillar and cytoskeleton proteins (Figure 1.B), such as thin filament (58.5%), Z-disk and intermediate filament (17.3%), and thick filament (3.1%). About 6.9% of the overall peptide abundance increase was mapped to the glycogen and phosphagen metabolic enzymes, and about 6.3% of the overall peptide abundance increase was mapped to proteins located in nuclei or related to nuclear function. Peptide abundance increases mapped to other protein function clusters were minor (<2% for each cluster). Specifically, about 95% of total peptide abundance increase was mapped to 20 proteins or protein complexes (Figure 1.C), such as troponin complex, NEB, LDB3, musculoskeletal embryonic nuclear protein 1 (MUSTN1), MYH, CKM, MYOZ, DES, MYOT, GAPDH, enolase, PYGM, myomesin, LMCD1, MB, PDLIM3, PDLIM5, and SYNPO2.

From 48 h to 336 h postmortem, major peptide abundance increase was mapped to myofibril thin filament (55.3%; Figure 1.B), glycogen and phosphagen metabolic enzymes (30.7%), and Z-disk and intermediate filament (7.9%). Peptide abundance increases mapped to other protein function clusters were minor (<2% for each cluster). Specifically, about 95% of peptide abundance increase was mapped to 20 proteins or protein complexes, such as NEB, troponin complex, GAPDH, PYGM, enolase, PDLIM3, PDLIM5, CKM, PGK1, actin isoforms, MYOZ, MYH, myomesin, EEF1A, SYNPO2, HSPB6, ATP2A1, and MUSTN1.

Quantified cleavage sites indicated differentially abundant cleavage and sequential cleavages in the precursor proteome

From the 3860 quantified peptides, a total of 4702 proteolytic cleavage sites were mapped in 244 precursor proteins. There were 540 cleavage sites quantified at 2 h postmortem, 1678 cleavage sites quantified at 48 h postmortem, and 4003 cleavage sites quantified at 336 h postmortem. A total of 181 cleavage sites were quantified irrespective of postmortem time, and many quantified peptides overlapped between timepoints (Figure 1.E). In addition, 501 cleavage sites were identified as DACS (P < .05) in comparisons among the 3 postmortem time points. There were 89 DACS (79 increased and 10 decreased) at 48 h as compared to 2 h postmortem, 432 DACS (394 increased and 38 decreased) at 336 h as compared to 48 h postmortem, and 109 DACS (97 increased and 12 decreased) at 336 h as compared to 2 h postmortem (Figure 1.E). There were 27 cleavage sites identified as DACS irrespective of the comparisons. Within 1 of 3 postmortem time points, 1535 cleavage sites were identified as either EACS or EPCS with high observation frequency (Figure 1.E). Specifically, there were 32 EPCS and 361 EACS at 2 h, 54 EPCS and 5 EACS at 48 h, 1057 EPCS and 26 EACS at 336 h postmortem (Figure 1.E). A total of 1815 quantified cleavage sites were identified as either DACS, EACS, or EPCS (Supplementary Table 3), and 221 cleavage sites were recognized as both DACS and EACS. Of these 1815 differential cleavage sites, 1764 cleavage sites were mapped to differential precursor proteins (DAPP, EAPP, and EAAPs), leaving 51 cleavage sites matched to 19 precursor proteins that were neither DAPP, EAPP, nor EPPP (Supplementary Table 2).

Among cleavage sites showing increased abundances during 2 postmortem periods, most correspond to increased degradation at the precursor protein level (see these proteins in 3.1.5). However, although the abundances of many cleavage sites increased during a postmortem period, their precursor protein abundance did not indicate significant increase (Table 1). CAPN1 did not increase its abundance at the precursor protein level when time extended, but cleavage-site abundance in CAPN1 (E-25) had 2 significant increases from 2 h to 48 h and from 48 h to 336 h postmortem. Similarly, cleavage-site abundances in CA3 (N-11), MAP4 (F-34), NDRG2 (S-346), PLEC (H-2778, R-2791), and SMPX (A-11, A-14, N-15) had significant increases from 2 h to 48 h, while cleavage-site abundances in CAPN1 (A-16), MYH4 (N-30), and CAVIN (G-344, K-357) had significant increases from 48 h to 336 h postmortem. Yet, these proteins had no abundance change at the precursor level during either of these postmortem periods. In addition, many cleavage sites, such as MB L-12, increased in abundance from 48 h to 336 h postmortem, but their precursor abundances only increased from 2 h to 48 h and remained the same from 48 h to 336 h postmortem. Furthermore, the identification of EACS and EPCS also indicated that many cleavages are time specific during the postmortem period. In agreement with previous gel-based studies, these observations indicated that a protein could experience sequential cleavages at different sites when postmortem time extends.

Quantified cleavage residues at P1-P1’ positions and their share of total cleavage abundance increase during each postmortem period

Among the recovered 4702 cleavage sites on mapped 244 precursor proteins, 363 cleavage residues (residue pairs at P1-P1’ positions) were quantified based on peptide abundances across the experiment. There were 203 cleavage residues quantified at 2 h postmortem, 314 cleavage residues quantified at 48 h, and 355 cleavage residues quantified at 336 h postmortem. A total of 194 cleavage residues were quantified regardless of postmortem time point, and many quantified cleavage residues overlapped among postmortem time points (Figure 5.B). In addition, 252 cleavage residues were identified as DACR (P < .05) in comparisons among the 3 postmortem time points. There were 111 DACR (108 increased and 3 decreased) identified at 48 h as compared to 2 h postmortem, 222 DACR (217 increased and 5 decreased) identified at 336 h as compared to 48 h postmortem, and 172 DACR (170 increased and 2 decreased) at 336 h as compared to 2 h postmortem, and 89 cleavage residues were identified as DACR irrespective of the comparisons (Figure 5.B). At 1 of 3 postmortem time points, 92 cleavage residues were identified as either EACR or EPCR with high observation frequency (Figure 5.B.). Specifically, there were 68 EACR at 2 h and 24 EPCR at 336 h. Notably, 50 cleavage residues were recognized as both DACR and EACR.

Figure 5.

(A) Cleavage abundance increase of major P1-P1’ cleavage residues within each postmortem time period. Note: Each major (nongrey) P1-P1’ residue has a share greater than 0.5% of total cleavage abundance increase. Each minor (grey) P1-P1’ residue has a share less than 0.5% of total cleavage abundance increase. (B) Distribution of differential P1-P1’ cleavage residues in all quantified P1-P1’ cleavage residues at 2 h, 48 h, and 336 h postmortem. DACR, differentially abundant cleavage residues; EACR, exclusively absent cleavage residues; EPCR, exclusively present cleavage residues. (C) Top 20 P1-P1’ cleavage residues with the most cleavage abundance increase during each postmortem time period.

Among the cleavage residues identified across different time points, about 70% of the cleavage residues were identified as DACR in at least 1 comparison. These pervasive statistical differences among cleavage residues indicated the significant proteolysis of postmortem muscle in general. However, the pervasiveness of the statistical significance also indicated its inability to distinguish the most significant cleavage pattern. To overcome this, based on the residue in P1 position, we ranked the cleavage residues based on their weighting of the cleavage abundance increase during each postmortem time period. From 2 h to 48 h postmortem, the major cleavage happened to following residues in P1 cleavage position (>5% share of total cleavage abundance increase): glutamate (E; 27.04%), histidine (H; 17.32%), glycine (G; 9.25%), lysine (K; 7.83%), glutamine (Q; 5.53%), and serine (S; 5.12%). The major residues in P1’ cleavage position (>5% share of total cleavage abundance increase) were: glutamate (E; 22.76%), alanine (A; 21.57%), serine (S; 16.88%), and threonine (T; 8.22%). Specifically, the major cleavage residues (>2% share of total cleavage abundance increase; Figure 5.C) at P1-P1’ positions within this time period are glutamate-alanine (E-A; 17.82%), histidine-glutamate (H-E; 15.80%), glutamate-glutamate (E-E; 4.61%), lysine-serine (K-S; 4.16%), glutamine-serine (Q-S; 3.74%), phenylalanine-serine (F-S; 3.56%), glycine-threonine (G-T; 2.63%), glycine-leucine (G-L; 2.57%), arginine-threonine (R-T; 2.12%), and glutamate-valine (E-V; 2.05%).

From 48 h to 336 h postmortem, the major cleavage happened to following residues in P1 cleavage position (>5% share of total cleavage abundance increase; Figure 5.C): histidine (H; 16.45%), glutamate (E; 14.63%), lysine (K; 13.66%), glycine (G; 8.15%), Asn (N; 5.77%), glutamine (Q; 5.58%), and leucine (L; 5.39%). The major residues in P1’ cleavage position (>5% share of total cleavage abundance increase) were: glutamate (E; 17.95%), alanine (A; 14.78%), leucine (L; 9.64%), serine (S; 9.18%), valine (V; 7.04%), isoleucine (I; 6.07%), glycine (G; 5.11%), and threonine (T; 5.09%). Specifically, the major cleavage residues (>2% share of total cleavage abundance increase) at P1-P1’ positions within this time period are histidine-glutamate (H-E; 12.91%), glutamate-alanine (E-A; 9.01%), glycine-leucine (G-L; 3.63%), glutamine-serine (Q-S; 2.67%), and glutamate-glutamate (E-E; 2.01%).

Degradation of myofibrillar and cytoskeleton proteins and their associations with meat tenderization

Traditional gel-based proteomic approaches have demonstrated increased degradation of many myofibrillar and cytoskeleton proteins in skeletal muscle when postmortem time extends. Among the degradation of these myofibrillar proteins, many have established positive correlations with the tenderness development in fresh meat, such as DES, troponin T (Rowe et al., 2004; Carlson et al., 2017), NEB, titin (Huff-Lonergan et al., 1995; Geesink and Koohmaraie, 1999), filamin (Huff-Lonergan et al., 1996; Carlson et al., 2017), CFL2, MYH, MYL, and ACTA (Lametsch et al., 2003). The degradation of these proteins could disrupt the myofibril structure, the attachments of the peripheral layer of myofibril to the sarcolemma, and overall muscle cell integrity, ultimately leading to the tenderization of the muscle (Huff-Lonergan et al., 2010). Consistent with these studies, current study identified significantly increased precursor protein degradation between 2 of 3 postmortem time points in TNNT3 and TNNT1, ACTA and NEB at thin filament, DES at intermediate filament, MYH (MYH1, MYH2, MYH3, and MYH7) and MYL3 at thick filament, and CFL2 at Z-discs (Figure 3; Table 1).

Aside from these proteins’ degradation with a known contribution to tenderization, we also identified significantly increased degradation of many cytoskeleton proteins and myofibrillar proteins at different sarcomere locations (Figure 3) between 2 postmortem time points: TPM2, TNNI1, TNNI2, and LMOD3 at thin filament; MYBPC1, MYBPC2, and MYBPH at thick filament; MYOT, MYOZ1, MYOZ3, TCAP, BIN1, SYNPO2, SYNPO2L, PDLIM1, PDLIM3, PDLIM5, PDLIM7, and LDB3 at Z-discs; MYOM1 and MYOM2 at M-band; VIM and SYNM at intermediate filament; and ACTB and TMSB10 at cytoskeleton.

Among these newly identified proteins with increased degradation over postmortem time, for intermediate filament proteins, SYNM is associated with DES and VIM filaments at Z-discs and costameres, connecting the sarcomere to the sarcolemma (Granger and Lazarides, 1980). At Z-discs, MYOT binds to α-actinin, ACTA (Salmikangas et al., 2003), filamin (van der Ven et al., 2000), LDB3 (ZASP; von Nandelstadh et al., 2009), and MYOZ1 (Gontier et al., 2005). Not only binding to α-actinin, filamin, and LDB, MYOZ1 and MYOZ3 also bind with TCAP (Faulkner et al., 2000; Frey and Olson, 2002), while TCAP further assembles in an antiparallel (1:2) sandwich complex with the N-terminal region of titin (Zou et al., 2006), connecting the thick filament with Z-discs. SYNPO2L (Yamada et al., 2024), PDLIM1 (Kotaka et al., 2000), PDLIM3 (Xia et al., 1997), and PDLIM5 (Nakagawa et al., 2000) also bind to α-actinin and assemble Z-discs. For other Z-discs proteins, BIN1 (Dräger et al., 2017) and SYNPO2 (Weins et al., 2001) bind to ACTA, while PDLIM7 binds to TPM2 (Guy et al., 1999). Similar to DES and actinin, the postmortem degradation of these intermediate filament proteins and Z-disc proteins could weaken the Z-disc structure, Z-disc association with thin and thick filaments, and the connection between sarcomere and sarcolemma.

At the M-band, MYOM1 (Obermann et al., 1997) and MYOM2 (Vinkemeier et al., 1993) are associated with thick filaments by binding to titin. At thin filaments, TNNI binds to ACTA and TPM2, and LMOD3 are present at the filament-pointed end/M-line (Yuen et al., 2014). MYBPC1 and MYBPC2 are long and flexible proteins strongly bound to the thick filament by its C-terminus, whereas their N-terminal domains bind to the thin filament (Luther et al., 2011) and the myosin head region (Pfuhl and Gautel, 2012). Along with the previously identified degradation of the myosin complex (heavy chains and light chains), NEB, ACTA, and troponin T, these degradations could weaken the thick and thin filaments, their association with M-line and Z-discs, and the connections between thick and think filaments.

Overall, the increased degradation of these myofibrillar proteins indicated substantial proteolysis of the thin filament, thick filament, M-line, and Z-disc, as well as the junctions between these sarcomere structures at postmortem. Moreover, the degradation of intermediate filament proteins further weakens the connection of the sarcomere to the sarcolemma. The continuous postmortem degradation of these proteins likely contributes to the improved tenderness over the postmortem time and warrants further investigations. Notably, ACTB and GAPDH, 2 housekeeping proteins often used as internal controls in gel-based proteomic analysis for abundance normalization, had significant degradation at postmortem. In fact, our recent review suggested that the housekeeping proteins, like ACTB and GAPDH, in skeletal muscle tissue have dynamic abundances resulting from both preharvest and postharvest treatments (Zhai et al., 2022a). Still, the current study demonstrated that substantial proteolysis happens to ACTB and GAPDH, making these 2 proteins unreliable internal controls for sample comparisons among different postmortem time points.

Degradation of proteins associated with pH decline and meat color

Noteworthily, the degradations of many energy metabolism enzymes were identified at 2 h postmortem, such as PYGM, PGM1, ALDOA, GAPDH, PGAM2, ENO3, PKM, and LDHA in glycolysis, complex III (UQCRQ and UQCRB), complex IV (COX4I1), and complex V (ATP5F1A, ATP5F1B, ATP5ME, and ATP5PF) in OXPHOS, as well as SUCLA2, SUCLG1, and MDH2 in TCA cycle. Among these, increased precursor protein degradations between 2 h and 48 h were identified for PFKM, PGK1, ENO3, and LDHA in glycolysis as well as complex I (NDUFB1) and complex IV (COX4I1) in OXPHOS (Table 1; Figure 2). Previous studies demonstrated that the dissociation of F₁ catalytic core from F₀ membrane-spanning component in complex V can impact the ultimate pH in postmortem muscle (Matarneh et al., 2018). ATP5PF is one of the stalk subunits linking the F₁ core with the F₀ core in complex V (He et al., 2020). Therefore, the early postmortem cleavage of ATP5PF might lead to the dissociation of F₁ from F₀ core in complex V. Except for PFKM losing activity at pH above 5.5, many glycolytic enzymes, such as PGK1, ENO3, and LDHA, have catalytic activity at pH 5.5 (England et al., 2014). Thus, the cleavage of glycolytic enzymes during the early postmortem period and postmortem aging could affect their activities and further intervene in pH decline during muscle-to-meat conversion and metabolism during postmortem aging, ultimately affecting meat quality development. Still, further investigation is warranted to understand the associations of these degradations with meat quality.

MB is the major protein responsible for the red color of meat obtained from livestock. MB redox stability and endogenous factors affecting the MB redox state can determine the red color stability (Ramanathan et al., 2020). Notably, the current study identified MB degradation at all 3 postmortem time points, and its precursor-level degradation increased between 2 h and 48 h postmortem (Figure 2). Although the redox stability of intact MB is well-studied in vitro (Suman and Joseph, 2013), the chemical properties of proteolyzed MB are not well understood. In addition, the current study also identified degradation of hemoglobin at 48 h and 336 h and degradation of cytochrome c at 48 h postmortem. Although hemoglobin and cytochrome c are not the major heme proteins responsible for the red color in the meat from livestock (Pikul et al., 1986; Wu et al., 2022), previous studies have proposed that increased solvent access to the heme pocket could cause heme iron oxidation and decrease heme affinity for the globin, so heme can be released from the protein and further oxidize lipid (Richards et al., 2023). Early postmortem proteolysis of heme proteins will change their primary structures via AA sequence cleavage and further change their 3-dimensional structure, including heme pocket space. Therefore, heme protein cleavage at early postmortem could affect heme release during postmortem aging and further affect meat color and lipid stability.

Many endogenous biological factors can affect the MB redox state, such as mitochondrial respiration, met-MB reductase, antioxidant enzymes, and reducing metabolites (Ramanathan et al., 2020). The current study identified postmortem degradation of mitochondrial enzymes (OXPHOS and TCA cycle; Figure 2), NADH regenerating enzymes (GAPDH, LDHA, and MDH2; Figure 2), hemoglobin/MB reductase (BLVRB; Figure 4; Yubisui et al., 1980; Xu et al., 1993), and antioxidant proteins (PRDX1, TXNL1, NNT, GLRX5, MSRA, and MSRB3; Figure 4). Among these proteins, significantly increased precursor protein degradations were identified in complex I (NDUFB1), complex IV (COX4I1), and LDHA between 2 h and 48 h postmortem, as well as GAPDH and BLVRB between 48 h and 336 h postmortem. NDUFB1 (Stroud et al., 2016) and COX4I1 (Zhang et al., 2025) are the essential subunits maintaining complex I and IV assembly and, thus, their native catalytic activities. The increasing degradation of these proteins could partially explain the irreversible loss of mitochondrial function after cytochrome c saturation in early postmortem skeletal muscle (Zhai et al., 2022b; Zhu et al., 2025) and decreased meat color stability during postmortem aging (English et al., 2016; Mitacek et al., 2019). Still, further study is warranted to evaluate the association of these degradations with mitochondrial function and meat color stability in postmortem muscle. Notably, only a small proportion of total proteolytic peptides were generated from mitochondria over time (Figure 1.B). Although increasing proteolytic cleavage in OXPHOS complexes was identified in the current study, the decreased mitochondrial function along postmortem aging can also come from compromised membrane structure, since most of the respiration activity was lost due to outer membrane permeabilization (Zhu et al., 2025).

Degradation of chaperone, membrane integrity, calcium signal, and proteins associated with nuclear and ribosomal functions

Chaperone and cochaperone proteins, such as heat shock protein families, have close relationships with postmortem metabolism and meat quality. In the current study, we identified the increased precursor protein degradations of HSPB1 and STIP1 between 48 h and 336 h postmortem as well as HSPB6 during both postmortem periods (Figure 4). A previous study reported that HSPB1, HSPB6, and STIP1 were less abundant in tender beef longissimus thoracis muscle than in their tougher counterparts at 24 h postmortem (Malheiros et al., 2021). Moreover, greater degradation of HSPB1 (Cramer et al., 2018) and HSPB6 (Balan et al., 2014) was positively associated with greater myofibrillar protein degradation and superior tenderness. Therefore, the degradation of these precursor proteins could explain increased tenderness during postmortem aging. Still, further investigation is warranted to understand the associations of other chaperone degradations with meat quality development.

Free Ca²⁺ concentration dramatically and irreversibly increases during the first 48 h postmortem, while the phospholipid content in the sarcoplasmic reticulum decreases (Ji and Takahashi, 2006), probably due to hydrolysis when postmortem time extends (Chao et al., 2020). In the current study, we identified the degradations of many proteins functioning as membrane scaffolds (CAV1, CAV3, RTN2, RTN4, ATL2, CAVIN1, and CAVIN4) and calcium mobilization regulators (ATP2A1, ATP2A2, RYR1, JPH1, SYPL2, CASQ1, CMYA5, and PLN). Among these proteins, ATP2A1 and ATP2A2 had significantly increased degradation from 2 h to 48 h and then 336 h postmortem (Figure 4), indicating increasingly compromised sarcoplasmic reticulum integrity in the postmortem muscle. Still, further study is warranted to evaluate the association of these proteins’ degradation with free Ca²⁺ concentration and membrane integrity in postmortem muscle.

Chromatin, which prevents DNA damage and regulates gene expression, is a complex of DNA and proteins in the nucleus. Ribosomes are macromolecular machines performing protein synthesis. Initiation factors are proteins that bind to the small subunit of the ribosome during the initiation of translation, while elongation factors are a set of proteins facilitating translation during the elongation process. Notably, the chromatin component proteins (H2B, HIST1H3C, HMGN2, and ARID1B), ribosomal protein (RPL13A), initiation factor protein (EIF1), and elongation factor proteins (EEF1A1 and EEF1A2) had increased precursor protein degradations in at least 1 of the postmortem periods (Figure 4). These data indicated that irreversible damage happens to the transcription and translation factors at postmortem.

Calpain system degradation and association of their cleavage sites with the current literature

Previous studies have reviewed many protease complexes potentially regulating the tenderization process in postmortem muscle, such as calpain-1, calpain-2, calpastatin, cathepsins, proteasomes, and caspases (Lana and Zolla, 2016). In calpain system, we identified the degradation of calpain small subunit 1 (CAPNS1), CAPN1, calpain-3 (CAPN3), and calpastatin (CAST) at all 3 postmortem time points (Figure 4). CAPNS1 forms heterodimers with CAPN1 or CAPN2 to compose larger calpain-1 or calpain-2, respectively. Calpain-1 is mainly diffusive before activation (Murphy et al., 2006) but rapidly binds to the titin upon activation (Raynaud et al., 2005). In contrast, newly synthesized calpain-3 is kept as an inactive homohexamer (trimer of CAPN3 homodimers) and splits into CAPN3 homodimers upon binding to titin (Ye et al., 2025). After dissociative activation from titin, these CAPN3 homodimers or new forming homotrimers compose proteolytically active calpain-3 (Kinbara et al., 1998; Ravulapalli et al., 2005; Hata et al., 2020). Calpain-1 requires 50 μM Ca²⁺ for activation, and calpain-2 requires 400 μM to 800 μM Ca²⁺ for activation (Cong et al., 1989). However, calpain-3 can activate at 0.2 μM Ca²⁺ when ATP is present and at 40 μM Ca²⁺ without ATP (Murphy and Lamb, 2009). The N-terminal regions tend to be cleaved in CAPN1 and CAPN3 upon activation for calpain-1 (Zimmerman and Schlaepfer, 1991) and calpain-3 (Ono et al., 2004), respectively. Yet, calpain-2 can be activated without N-terminal region cleavage of its catalytic subunit (Chou et al., 2011). Intriguingly, the N-terminus cleavage of calpain-2 catalytic subunit by calpain-1 can significantly reduce the Ca²⁺ requirement for calpain-2 proteolytic activity (Tompa et al., 1996), while the activated calpain-2 can cleave calpain-1 catalytic subunit at the C-terminal side of the 116th, 117th, and 128th AA, completely quenching calpain-1 proteolytic activity (Shinkai-Ouchi et al., 2020).

In fact, recent evidence also indicated that calpain-3 not only prefers to proteolyze calpain-1 than calpain-2 but also exists as a substrate for calpain-1 and calpain-2 (Ojima et al., 2023). Moreover, calpain-3 autoproteolytic cleavage sites were identical to the sites of calpain-3 cleaved by calpain-1 and -2. Calpain-1 and -2 fragments generated by calpain-3 also had nearly identical sizes as compared to the calpain-1 and -2 autolytic fragments on Western blot (Ojima et al., 2023). Furthermore, calpastatin is an inhibitory regulator for calpain-1 and calpain-2 (Goll et al., 2003) but a substrate for calpain-3 (Ono et al., 2004), while titin is a substrate for calpain-1 and calpain-2 (Kim et al., 1995) but an inhibitory regulator for calpain-3 (Ono et al., 2006; Hayashi et al., 2008). These observations composed a mutual restrictive intercalpain regulation system (Ojima et al., 2023). This indicates that factors outside of this system must exist to tilt the balance among calpains to initiate calpain proteolytic cascade in the postmortem muscle. Related to this, many mechanisms are available in the literature, such as sodium-source ionic strength, environmental pH (Maddock et al., 2005), sequence of CAPN1 oxidation versus CAPN1/calpastatin complex formation (Maddock Carlin et al., 2006; Maddock Carlin et al., 2024), titin cleavage by cathepsin-D (CTSD; Kim et al., 1995), potassium-source ionic strength (Wette et al., 2023), ATP presence (Murphy and Lamb, 2009), calpastatin phosphorylation (Averna et al., 1999), and calpastatin degradation induced by noncalpain proteases such as caspase-1 (Barnoy and Kosower, 2003), caspase-3 (Pörn-Ares et al., 1998), and caspase-7 (Wang et al., 1998).

At the cleavage-site level, we identified many cleavage sites at the canonical N-terminal sequence of CAPN1: Q-17 at 48 h and S-15, A-16, E-25, and G-27 at all 3 time points. S-15 and G-27 are 2 canonical N-terminal cleavage sites in CAPN1 upon calpain-1 proteolytic activation, while A-16, Q-17, and E-25 are unknown cleavage sites in the literature. Similarly, we identified many cleavage sites at the canonical N-terminal sequence of CAPN3: P-2 at 2 h, P-11 and G-14 at 2 h and 48 h, M-18 at 48 h, and P-20 and E-33 at 48 h and 336 h. G-14 and E-33 are 2 canonical N-terminal cleavage sites in CAPN3 for calpain-3 proteolytic activation (Ono et al., 2014), while P-2, P-11, M-18, and P-20 are newly identified cleavage sites. We also identified canonical N-terminal cleavage sites in the CAPNS1 sequence: G-21 at 2 h, G-19 at 48 h, and Y-71 and N-84 at 336 h. In addition, many cleavage sites in noncanonical CAST sequence were identified at the N-terminal, middle, and C-terminal regions at all 3 time points.

The 80-kDa full-length CAPN1 will be gradually cleaved into 76-kDa autolyzed CAPN1 in beef LL muscle from 1.5 h to 7 d postmortem (de Oliveira et al., 2019). Although the current workflow was not able to demonstrate gradually increased degradation of CAPN1 at the precursor protein level, the 2 CAPN1 autoproteolytic cleavage sites (S-15 and G-27) were accurately captured at the cleavage-site level. Intriguingly, we identified an increased cleavage abundance at A-16 between 48 h and 336 h postmortem as well as 2 cleavage abundance increases at E-25 during both postmortem periods, but S-15 and G-27, the 2 canonical autolytic cleavage sites generating intermediate 78-kDa and fully cleaved 76-kDa CAPN1 (Zimmerman and Schlaepfer, 1991; Saido et al., 1992), did not have such changes. In addition, the 3 CAPNS1 canonical autoproteolytic cleavage sites, G-25, G-58, and E-86 (McCelland et al., 1989), were not identified in the current study. Instead, we identified cleavage at G-19, G-21, Y-71, and N-84 in CAPNS1. It is unknown if these cleavages were generated by calpain-1 or by other proteases. Additionally, although the size of cleaved CAPN1 by calpain-3 seems identical to the calpain-1 autolysis on the Western blot (Ojima et al., 2023), future study is warranted to evaluate the cleavage of CAPN1 through a MS-based quantification method because gel-based methods are not likely to differentiate molecular weight differences of 1 or 2 extra AA in a sequence (S-15 vs. A-16 and G-27 vs. E-25). Furthermore, it is unknown if the cleavage-site preference of proteases differs under dynamic environment pH and ionic strength during the muscle-to-meat conversion and postmortem aging. Notably, in proteases like caspase-3, the initial cleavage event at D-9 is required to allow cleavage at D-28 that causes the complete removal of the prodomain, allowing for full caspase activation, while removing the 10 residues at the N-terminal renders caspase-3 inactive (Ponder and Boise, 2019). Therefore, we speculate that the proteolytic activity and cleavage preference of cleaved CAPN1 might be different after these atypical cleavages as compared to the canonical N-terminal cleavages. To further understand this differing cleavage pattern in the scope of whole proteome cleavages, we further evaluated the cleavage residue at the P1-P1’ position (see later section regarding AA residues of major cleavage sites).

Degradation of other proteases and the association of their cleavage sites with protein sequence features

In the current study, we identified degradation of CTSD at 2 h and 48 h postmortem, TPP1 at 2 h postmortem, and 26S proteasome non-ATPase regulatory subunit 11 (PSMD11) at 2 h postmortem. PSMD11 is a molecular clamp holding the 20S core and 19S regulatory subcomplex together (Pathare et al., 2012). CTSD is an aspartic protease capable of acid-dependent autoactivation in vitro (Hasilik et al., 1982) and possesses optimum activity at pH 4.5 to 5.0 (Briozzo et al., 1988). Unlike calpains, CTSD in vivo proteolytic activation involves multistep cleavages by unknown cysteine proteases, cathepsin-L and -D, amino-peptidase, and carboxyl-peptidase (Laurent-Matha et al., 2006). For the cleavage site of CTSD, we identified Y-372 near the C-terminal region of the canonical sequence, which is not a cleavage site during the known activation process (Laurent-Matha et al., 2006). TPP1 is a lysosomal serine protease having an N-terminal tripeptidyl exopeptidase activity with a pH optimum of 5 and an endoproteolytic activity with a pH optimum near 3 (Ezaki et al., 2000). TPP1 encodes a 563-residue preproprotein with a cleavable N-terminal 19-residue signal sequence. The proenzyme (residues 20–563) is a soluble monomer that undergoes proteolytic cleavage in the lysosome, converting the zymogen to an active, mature protease (residues 196–563; Sleat et al., 1997). Studies on purified pro-TPP1 demonstrate that maturation is autocatalytic in vitro (Lin et al., 2001; Golabek et al., 2004) but may involve other proteases in vivo (Golabek et al., 2003). Interestingly, in canonical TPP1 sequence, the 3 cleavage sites, S-486, R-495, and Q-509, are not autolytic maturation sites but are located between the 3 residues of the active site (E-272, D-276, and S-475) and all Ca²⁺ binding residues in TPP1 (D-517, V-518, G-539, G-541, and D-543; Guhaniyogi et al., 2009; Pal et al., 2009). It was reported that TPP-I proteolytic activity is inhibited by calcium (Kondo et al., 2016). Thus, these cleavages likely diminish the interaction between calcium ions and TPP1 and further impact the proteolytic behavior of TPP1.

Comparison of amino acid residues of major cleavage sites between the 2 postmortem time periods

Proteins are the principal molecular machines of the cell and derive their function from their AA sequence’s unique 3-dimensional folding structure (Dill and MacCallum, 2012). The dominant contributors to protein folding include the hydrophobic effect, conventional hydrogen bonding, and Coulombic and van der Waals interactions (Pace et al., 2014). Additional contributors, such as strong interactions involving protein side chains, can also play an important role in protein folding (Newberry and Raines, 2019). At neutral pH, charge–charge interactions among side chains stabilize the protein. However, as pH values decreased from neutral pH, the acidic residues can be neutralized, and the residual repulsive interactions between basic residues produce destabilization and change the native protein structure (Yang and Honig, 1993), unfolding a native secondary structure into a random coil (Guckeisen et al., 2021). Moreover, the protonation state of charged AA can affect their ability to form hydrogen bonds and salt bridges, further changing the formation of protein–protein contact (Dumetz et al., 2008). Therefore, the decreasing environmental pH can potentially affect a protease’s cleavage preference on the same protein substrate. Notably, protein crystallography is restricted to the protein’s native conditions that allow protein crystallization. Thus, for the protein investigations under non-native or denaturing conditions, there is a limitation in using the known crystal structure to project the protein unfolding event and protein–protein interaction.

In postmortem muscle, there is limited documentation of protein cleavage sites and their AA profiles during muscle-to-meat conversion and postmortem aging. To explore cleavage profile change at postmortem, based on the cleavage residues at P1-P1’ cleavage positions, we summarized the share percentage of total cleavage abundance increase in 2 different postmortem time periods, from 2 h to 48 h postmortem and from 48 h to 336 h postmortem (Figure 5.A). Rigor mortis is the muscle-to-meat conversion process starting upon animal exsanguination, a period when muscle has a declining pH value, decreasing ATP level, and increasing Ca²⁺ ionic strength. Rigor mortis normally completes within 24 h postmortem. The fresh meat then has a relatively stable pH (5.6∼5.8) and ionic strength with depleted ATP. Compared to the first postmortem period (2–48 h postmortem), the share of cleavage abundance increase among cleavage residues was different during 48 h to 336 h postmortem (Figure 5.A). Glutamate (E; 27.04%) had the largest abundance increase at the P1 position during the first period but dropped to second (14.63%) during the second period. The share of cleavage abundance increase after histidine (H) stayed consistent (17.32–16.45%), but its rank rose to first place during the second period. The share of lysine (K) at the P1 position rose from fourth place to third place, with its percentage increasing from 7.83% to 13.66%. The share percentage of glycine (G) and glutamine (Q) remained at 8.15% to about 9.25% and 5.53% to 5.58% during the 2 periods, respectively. Leucine (L) and Asn (N) only had a minor share (<2%) during the first period, but their shares became more significant (5.39% and 5.77%, respectively) during the second period. In sum, 50% to about 60% of total cleavage happened after 4 residues (E, H, K, and G) during both postmortem periods, and cleavage after H was consistently significant in both periods. Still, cleavage after K, L, and N was more significant between 48 h and 336 h than earlier postmortem periods, while cleavage after E became less dominant as the postmortem period extended.

Among major residues at P1’ position (Figure 5.A), E and alanine were most abundant in both postmortem periods (22.76% and 17.95% for E; 21.57% and 14.78% for A). Serine (S) and threonine (T) had more significant shares (16.88% for S and 8.22% for T) during the first period, but their shares became less significant (9.18% for S and 5.09% for T) during the second period. L, valine (V), and isoleucine (I) had less significant shares (4.04% for L, 4.31% for V, and 2.15% for I) during the first period, but their shares became more significant (9.64% for L, 7.04% for V, and 6.07% for I) during the second period. In summary, P1’ positions were primarily placed with E and A during both postmortem periods. The cleavage before S and T became less significant during the second period, while cleavage before L, V, and I became more significant as the postmortem period extended.

Under neutral environmental pH, K has basic residue, while E has acidic residue. The majority of H residues are neutrally charged, and the residues of N, Q, S, T, G, A, V, I, and L are uncharged. Still, the side chains of H, N, S, and Q are more hydrophilic than T, G, A, V, I, and L (Di Rienzo et al., 2021). Based on the protein structure crystallography database, a recent hydropathy study characterized populations of AA residues in protein shell (exposure to water), mantle (partial exposure to water), and core (no exposure to water; Ji et al., 2024). This study found that K, G, and L are most abundant in the proteins’ shell, mantle, and core, respectively. S and T are more abundant in the mantle than in the core and shell, while I and V are more abundant in the core than in the mantle and shell. E is highly abundant in both shell and mantle, and A is highly abundant in both mantle and core. However, AA like H, N, and Q do not have a preferred presence in any zone (Ji et al., 2024), indicating a dynamic nature of H, N, and Q in protein folding under physiological conditions.

Importantly, a protein substrate can only be cleaved at solvent-exposed sites, and a buried site in protein secondary structure cannot be cleaved until it unfolds into a solvent-exposed state (Robertson et al., 2016). When the environment pH decreases from ∼7.4 to 5.6, the core residues like L, I, V, and A are expected to have more presence at the protein mantle and shell due to unfolding destabilization. Possessing a side chain with pKa between 6.0 and 6.5, the presence of H residue can be dynamic in the protein zone at physiological conditions (Ji et al., 2024), but protonation of its side chain in an acidic environmental pH (postrigor pH 5.6) could mobilize its presence toward the protein shell. In contrast, the dominative presence of basic K and E residues in the protein shell at physiological pH could remain consistent during the pH decline of rigor mortis. Although P1 residue tended to be more important than P1’ residue for the protease to target for cleavage (Gupta et al., 2010), P1’ residue can potentially have an inhibitory effect on this process (Gershon, 2014). Moreover, hydropathy of a residue also depends on the response of its neighboring residues to water (Ji et al., 2024), which could further impact its availability to a proteolytic attack. Therefore, the increased cleavage shares of L at P1 position and L, I, and V at P1’ position during the extension of postmortem time could be partially attributed to the changing residue population in the protein solvent-exposure surface at the acidic pH environment.

Additionally, the abundance of cleavage residues at P1-P1’ positions can be affected by cleavage specificity (cleavable vs. uncleavable between 2 AA residues of all cleavage sites) and preference (quantitative difference among AA residues of all cleavage sites) of an active protease. For example, caspases preferentially cleave after D (Seaman et al., 2016), while trypsin preferentially cleaves at R and K in P1 position with higher rates for R (Vorob’ev et al., 2000) but mostly does not favor cleavage sites with P in P1’ position (Rodriguez et al., 2008). Yet, when environmental pH changes, a protease’s cleavage rate can change for the same cleavage site or cleavage residue, but its cleavage site on the same protein substrate remains the same (Vreeke et al., 2023). Therefore, the dominative cleavage with E, H, K, and G at P1 position as well as the dominative cleavage with E and A at P1’ position in both postmortem periods are likely collective preferences of endogenous protease systems during meat production. Especially, E-A and H-E were 2 peptide bonds with the most abundant cleavage in both postmortem periods, while the cleavage after K was gradually positioned with residues possessing increasingly different polarity and hydrophobicity over the postmortem period. Along with documented 4702 cleavage sites in 244 proteins in vivo, this first cleavage fingerprint in postmortem muscle provided valuable insights for future protease studies and laid the groundwork to quantify in vivo proteolytic activities determining meat quality development.

Interactive relation between cleavage sites generated at different time points in the same precursor protein

An inaccessible cleavage site in an initial precursor protein folding structure can become accessible to proteolytic attack after cleavages of other sites, because the new protein sequence generated from the cleavages tends to refold or unfold its structure to achieve stabilization and, thus, could expose the inaccessible cleavage site to protease. For example, collagenase’s cleavage site on collagen can be accessible for cleavage after proteolytic removal of the C-terminal telopeptide of collagen (Perumal et al., 2008). This is likely the case for the degradation of many proteins identified in the current study. Out of 4702 cleavage sites, 4162 were not identified at 2 h postmortem, and 2785 cleavage sites were only identified at 336 h postmortem (Figure 5.B). Likewise, the increased share of hydrophobic residues, like L, I, and V, at P1-P1’ position in the second postmortem period than the first period could also result from the unfolding of a protein structure after cleavages, since smaller protein (or shorter peptide) and disordered (or denatured) protein structures might lack a core (no exposure to water). This process was described as “demasking” (Vorob’ev, 2013). In the same concept but opposite phenomenon for “secondary masking,” an accessible cleavage site can become inaccessible due to cleavage-induced protein refolding (Vorob’ev, 2022). In our data, some cleavage sites in proteins like ACTA were identified at 2 h and 336 h but not at 48 h postmortem (Figure 5.B), which is likely the consequence of the secondary masking from 2 h to 48 h and the demasking from 48 h to 336 h postmortem. In support, similar observations on demasking and secondary masking have also been validated in an in vitro pepsin cleavage study (Vreeke et al., 2023).

In addition, we identified many cleavage sites with the only presence at 2 h postmortem or the first 2 time points (Figure 5.B). However, aside from being the outcome of secondary masking, the exclusive presence of these cleavage sites at earlier time points can also be the outcome of programmed sequential cleavages of precursor proteins over time, like calpain-1’s 2-step autoproteolytic activation, where 80-kDa CAPN1 is cleaved at S-15 to yield 78-kDa intermediate product prior to the cleavage at G-27 to generate 76-kDa product (Zimmerman and Schlaepfer, 1991). Using our results as an example, we identified the first-step autolytic cleavage of G-14 in CAPN3 at 2 h and 48 h postmortem, followed by the delayed presence of second-step cleavage of E-33 at 48 h and 336 h postmortem (Supplementary Table 3). In sum, our results demonstrated that the earlier proteolytic cleavage sites and their cleavage rate could affect the downstream cleavage pattern. Future time-course proteolytic cleavage studies are warranted to identify the key cleavage events that affect meat quality development.

Limitations and avoidable pitfalls

Like the bottom-up proteomic analysis of complex protein matrix from in vivo studies, the matching of a quantified peptide sequence to a precursor protein relies on the protein sequence databases, such as UniProt and the National Center for Biotechnology Information database. Although modern state-of-the-art MS has greatly maximized our detection sensitivity and accuracy to quantify a peptide sequence, the precursor protein matched by the same list of peptides can change from time to time, because changes in the reference sequence will modify the matching decision and, ultimately, identification. The changing matching decision can also change the peptides population used to estimate the abundance of a precursor protein and, ultimately, the estimated protein abundance. Undoubtedly, these variations could periodically affect the partial results of a completed study. If multiple protein precursors were matched for the identified peptides, reporting results based on the matched canonical sequence could likely extend the validity period of the results, since these canonical sequences tended to be valid for a much more extended period. However, researchers should be transparent about this practice in their study, as it is crucial to the repeatability.

Many in vitro protease cleavage prediction studies have used synthesized polypeptide sequences with 20 to 25 AA to identify the cleavage specificity and preference. These in vitro studies have many advantages that in vivo studies will not have. For example, the designed substrate length and sequence can help researchers quantify the cleavage residue from only P1-P1’ positions to a broader range, generating substantial numeric data points to project the cleavage specificity and rate (Shinkai-Ouchi et al., 2016). However, the documented cleavage rate and specificity for the cleavage residues might not apply to the in vivo cleavage pattern due to many complexities. Factors like proteins’ native folding structure, protein–protein interactions, and dynamic pH, ionic, and redox conditions of subcellular organelles can significantly affect the activity of proteases and the availability of cleavable residues on the substrate surface. To better understand the protease cleavage in the specific in vivo substrates, many in vitro studies have used purified large protein substrates with known AA sequences to evaluate protease cleavage behavior under optimal ionic strength, pH, and temperature for fixed time periods. The major cleavable sites in a protein substrate can be effectively captured after protease-substrate incubation (Sun et al., 2013).

Although this strategy is valuable to predict all the possible cleavage sites of a protein substrate to a protease in vivo, studies using this strategy will not be able to monitor the actual length of the 2 product sequences generated from a single cleavage site. For example, the residues in P1’-P10’ position (first–tenth AA at C-terminal direction of the cleaved peptide bond) can be identified when a cleavage site is located at the N-terminal of a 10-residue proteolytic peptide, but this does not guarantee the existence of residues other than the P1 position. Although P2-P10 residues (P2-P10: second–tenth AA at the N-terminal direction of the cleaved peptide bond) could exist in the intact precursor protein according to the protein sequence, this 10-residue peptide can also be generated from the single N-terminal residue removal of an 11-residue peptide by a protease. The same pitfall can also apply to a cleavage site located at the C-terminal of a proteolytic peptide, since this peptide can be generated from the cleavage removal of a C-terminal residue. Therefore, unless designed peptide sequences (known length) are used as proteolytic substrates, it could be theoretically inaccurate to project cleavage preference other than P1-P1’ positions. It is essential for researchers to carefully evaluate the validity of a quantitative method from a biological standpoint to avoid significant bias.

In addition, it is well established that the chemical properties of a residue depend on its neighboring residues in the protein structure (Pace et al., 2009; Dill and MacCallum, 2012; Pace et al., 2014; Ji et al., 2024), and the same impact has also been observed in the proteolytic kinetics of cleavage residues (Sun et al., 2013; Seaman et al., 2016; Shinkai-Ouchi et al., 2016; Liu et al., 2019; Zhou et al., 2020). Thus, sequence heatmaps, a popular visualization method showing residue frequency in each individual sequence position, should be used carefully because the high frequency of a residue in each cleavage position can be restricted to the presence of a residue in the neighboring position. Using our data as an example, although E has the most increases in cleavage abundance as P1 or P1’ residue from 2 h to 48 h postmortem, E-E is not the peptide bond with the most increase in cleavage abundance during this period (Figure 5. A and C). Researchers should choose the data visualization method carefully to maximize the knowledge delivery of major results without compromising the accuracy.

Strategies to improve meat and muscle bottom-up proteomic research

Bottom-up proteomics identifies proteins by characterizing the AA sequences and post-translational modifications of proteolytically digested peptides through MS (Aebersold and Mann, 2003). Bottom-up proteomics is successful because peptides are more easily separated by liquid chromatography, ionized well, and fragmented in a more predictable manner than intact proteins (Dupree et al., 2020). Because peptides are easier to identify and can be matched with databases for parent protein identification, bottom-up proteomics tends to identify more proteins than other proteomic methods (Chait, 2006). Among proteases, trypsin is the most widely used enzyme for protein digestion in proteomic research due to its high specificity for cleavage residues: peptide bond cleavage at the C-terminal side of K or arginine (R) residues, except when followed by proline (Keil, 1992; Olsen et al., 2004). However, nontypical cleavage (often called “miscleavage”) can happen during trypsin digestion (Rodriguez et al., 2008; Šlechtová et al., 2015). Thus, after trypsin digestion, 1 miscleavage for a peptide is usually allowed during database search, which means to increase the sequence coverage and improve protein identification for a comprehensive proteome analysis. However, the current dataset demonstrated that proteolytic cleavage happened in the postmortem muscle, and the endogenous proteolytic peptides gradually accumulated in the muscle tissue along with postmortem aging. Moreover, our cleavage residue data demonstrated that endogenous cleavage after K and R, 2 trypsin-preferred cleavage residues at the P1 position, could generate a considerable amount of peptides in postmortem muscle. During a database search, these endogenous cleavages generated peptides could be confounded with the peptides generated from trypsin cleavage and, thus, bias the abundance estimation of intact protein. To improve the accuracy of meat and muscle bottom-up proteomic datasets, researchers could consider different strategies depending on their research goals, such as conducting sample collection at preharvest or early postmortem as well as disabling trypsin miscleavage during database searches to reduce the influence of endogenous cleavage in aged meat. Traditional gel-based methods, such as Western blot or 2-dimensional polyacrylamide gel electrophoresis, are also practical choices since their molecular weight-based protein separation can help determine the sequence integrity of a targeted protein.