An atlas of caspase cleavage events in differentiating muscle cells

Abstract

Executioner caspases, such as caspase‐3, are known to induce apoptosis, but in other contexts, they can control very different fates, including cell differentiation and neuronal plasticity. While hundreds of caspase substrates are known to be specifically targeted during cell death, we know very little about how caspase activity brings about non‐apoptotic fates. Here, we report the first proteome identification of cleavage events in C2C12 cells undergoing myogenic differentiation and its comparison to undifferentiated or dying C2C12 cells. These data have identified new caspase substrates, including caspase substrates specifically associated with differentiation, and show that caspases are regulating proteins involved in myogenesis in myotubes, several days after caspase‐3 initiated differentiation. Cytoskeletal proteins emerged as a major group of non‐apoptotic caspase substrates. We also identified proteins with well‐established roles in muscle differentiation as substrates cleaved in differentiating cells.

1. INTRODUCTION

Caspases are cysteine aspartyl proteases best known for their roles in regulating apoptosis, pyroptosis, and inflammation. Caspase‐2, ‐3, ‐6, ‐7, ‐8, and ‐9 are linked to apoptosis; caspases‐1, ‐4 and ‐5 to inflammation and pyroptosis. Approximately 2000 caspase substrates have been identified, although the functional and biological consequences of caspase cleavage are unknown for the majority of them. More recently, we and others have shown that apoptotic caspases like ‐2, ‐3, ‐6, ‐7, ‐8 and ‐9 also play non‐cell death roles in cell cycle control, development and neuronal plasticity, and cell differentiation (reviewed in Connolly et al., ²⁰¹⁴; Dehkordi et al., ²⁰²²). These non‐lethal roles involve activation of apoptotic caspases in cells that do not die (Baena‐Lopez et al., 2018; Dehkordi et al., ²⁰²⁰), raising interesting questions about which substrates are cleaved when caspases induce non‐lethal outcomes and how these are different from the substrates cleaved in dying cells.

Caspases cleave between two amino acids labeled P1 and P1′, where P1 is an aspartate (P1 = D), and with the surrounding residues labeled Pn–P2–P1–P1′–P2–Pn′ (Schechter & Berger, 1967). Scanning combinatorial substrate libraries defined “optimal” peptide substrates for most caspases that are suitable for easily and quickly measuring enzyme activity (Thornberry et al., 1997). More recently, different mass spectrometry approaches have provided much data on cellular caspase substrates (reviewed in Julien & Wells, ²⁰¹⁷), and the identified cleavage sites of protein substrates generally resemble the sequences identified by Thornberry et al. (1997), suggesting that P4–P4′ sequences are important in determining specificity. However, proteomics has also revealed that the rate at which a caspase cleaves its different protein substrates varies widely, with some substrates cleaved very rapidly and others much more slowly (Julien et al., 2016), so additional determinants of specificity are undoubtedly important.

There are very few proteome‐wide studies that have identified non‐apoptotic caspase substrates (Conde‐Rubio et al., 2021; Hertz et al., ²⁰¹⁹; Victor et al., ²⁰¹⁸; Weghorst et al., ²⁰²⁰) and even fewer studies have attempted to compare substrates cleaved in apoptotic and non‐apoptotic conditions. Notably, Conde‐Rubio et al. (2021), compared the proteins cleaved by caspases under lethal and non‐lethal conditions and observed that the non‐lethal substrates were a subset of the lethal substrates. In another study, Weghorst et al., identified non‐apoptotic protease substrates from chick neurodevelopment (Weghorst et al., 2020) and in a follow up report argued that caspases preferentially target cleavage sites found in cytoskeletal proteins during non‐apoptotic processes (Weghorst et al., 2022). Although neither Conde‐Rubio et al. nor Weghorst et al. demonstrated how the substrate choice is made, others have proposed subcellular localization of caspase activity (Amcheslavsky et al., 2018; Aram et al., ²⁰¹⁶; De Botton, ²⁰⁰²; Huesmann & Clayton, ²⁰⁰⁶; Kang et al., ²⁰¹⁷; Kaplan et al., ²⁰¹⁰; Kuo et al., ²⁰⁰⁶; Schoenmann et al., ²⁰¹⁰; Williams et al., ²⁰⁰⁶) and control of the level of caspase activity (Florentin & Arama, 2012; Kang & Bashirullah, ²⁰¹⁴; Li, Jo, et al., ²⁰¹⁰; Yang & Widmann, ²⁰⁰²) as possible mechanisms that determine whether caspase activity is lethal or not.

Here we used a well‐characterized model of caspase‐dependent cell differentiation with N‐terminomics to identify caspase substrates in differentiating and dying cells. C2C12 cells are undifferentiated myoblasts derived from regenerating mouse muscle (Blau et al., 1983) that proliferate in medium containing fetal bovine serum, but differentiate into multinucleated myotubes when switched to medium containing horse serum. Importantly, neighboring cells on the same plate take different fates under these conditions; about a third die by apoptosis, a third fuse into myotubes and a third survive, but do not differentiate. The death and differentiation is dependent on caspase‐3 (Fernando et al., 2002) activated by the same proteins involved in the mitochondrial cell death pathway (Dehkordi et al., 2020; Larsen et al., ²⁰¹⁰; Murray et al., ²⁰⁰⁸). Caspase‐3 activity is observed in the C2C12 cells that go on to differentiate (Dehkordi et al., 2020), so the caspase dependence of differentiation cannot be explained by the Phoenix Rising model in which caspases produce apoptotic cells that are signals for repair and regeneration (Li, Huang, et al., 2010). Therefore, the C2C12 model is well suited to comparing the substrates cleaved during cell death and differentiation.

This is the first report of N‐terminomics comparing substrates in dying and differentiating populations of C2C12 cells, and we have identified new caspase substrates, several of which are detected only in the differentiating cell population. A forward (or in vivo) N‐terminomics approach was applied for the identification of intrinsic cleavage events on each population and a complementary reverse (or in vitro) approach with purified mouse caspase‐3 was used to characterize direct proteolytic events during early stages of differentiation. Some of these substrates are proteins with links to muscle repair and regeneration and/or muscular dystrophy. We also found substrates in common with the other proteomic studies that identified caspase substrates cleaved during non‐apoptotic processes. Thus, these data suggest the existence of non‐apoptotic substrates linked to muscle biology as well as substrates associated with other non‐apoptotic contexts.

2. RESULTS

2.1. Separating C2C12 cell populations that are morphologically and biochemically distinct

Differentiating C2C12 cells behave heterogeneously (Yoshida et al., 1998). After 1 day in differentiation medium (DM), approximately one third of the cells showed a shrunken phenotype and detached from the plate (Figure 1a), characteristics that are typical of apoptosis (Dehkordi et al., 2020; Murray et al., ²⁰⁰⁸). These were harvested as the “dead cell” population. Attached cells harvested on day 1 were the “live cell” population. We also cultured C2C12 cells in DM for an additional 3 days to allow cell fusion into myotubes, although some cells remained mono‐nucleated. Differential trypsinization on day 4 allowed us to separate the myotubes (Figure 1a) from the mononucleated population. We designated these mononucleated cells as “reserve cells”, based on earlier reports (Schöneich et al., 2014; Yoshida et al., ¹⁹⁹⁸). All in all, we obtained five cell populations: myoblasts, live cells, dead cells, myotubes, and reserve cells.

FIGURE 1.

Proteomic signatures define the isolated cell populations during myoblast differentiation. (a) Workflow showing the preparation of dead and live cell populations on day 1 and reserve cell and myotube populations on day 4 of incubation in DM. Myoblasts were maintained in GM. (b) Immunoblotting for muscle‐specific proteins and caspase‐3. (c) Proteins levels for known markers of myoblast differentiation observed by mass spectrometry. Protein abundance for Pcna, Tnn1, Tnn3, Mef2c determined by label‐free quantification (n = 5). (d) Principal component analysis of the proteomic data acquired on the different cell types with a clear separation at the proteome level. Each point in the PCA plot represents the data for every replicate analyzed by LC–MS/MS. The color of each circle matches the population color‐code. (e) Pearson correlation between proteomic data for all populations versus cells at day 0 (myoblasts) in our study (left panel) vs. d0‐myoblasts from a previous proteomic report (Xiao et al., 2022). (f) Distribution of protein involved in relevant biological processes related to differentiation and cell death. Normalized and scaled (z‐score) abundances and hierarchical clustering is summarized in the heatmaps. Clusters generated show color‐codes of the cell type with stronger contribution to the abundance in that group. A detailed list of all protein identifications and proteomic changes is provided in Appendix S1. D cells, dead cells; L cells, live cells; MB, myoblast; MT, myotubes; R cells, reserve cells.

To characterize the different cell populations, we immunoblotted for MyoD, a marker of activated satellite cells (Koishi et al., 1995), and for myogenin, a direct target of MyoD which induces the fusion into myotubes (Figure 1b). Myoblasts showed high MyoD and low myogenin, consistent with their activation as a proxy for activated satellite cells. MyoD levels fell and myogenin levels rose in live cells on day 1, consistent with the cells being at an early stage of differentiation. On day 4, myotubes showed low levels of MyoD, consistent with the differentiated state of these cells. The reserve cells expressed MyoD and a low level of myogenin, suggesting an undifferentiated character. The dead cells showed low levels of both MyoD and myogenin, suggesting that differentiation had failed at an early stage of the process.

Caspase‐3‐like activity increases approximately 1 day after differentiation is induced (Dehkordi et al., 2020; Fernando et al., ²⁰⁰²; Murray et al., ²⁰⁰⁸), so we assessed caspase‐3 processing by immunoblot (Figure 1b). Dead cells harvested on day 1 showed high levels of processed caspase‐3, and a decrease in full‐length caspase‐3 consistent with their apoptotic morphology. In contrast, processed caspase‐3 was undetectable in myoblasts and barely detectable in live cells, myotubes or reserve cells, suggesting different levels of caspase‐3 activity in the populations.

2.2. Identification of proteome changes associated with cell death and differentiation

A global proteome comparison was performed to characterize the five cell populations and to identify notable differences in protein abundance (Figure 1c). We used liquid chromatography followed by tandem mass spectrometry (LC–MS/MS) with a data‐independent acquisition method to achieve a more comprehensive proteome coverage (n = 5). A label‐free quantitation approach using Spectronaut identified more than 6000 proteins per sample. Around 6800 proteins were identified in the myoblasts, with the number of groups decreasing with the progression of the differentiation process (~6200 proteins in the myotubes). Statistically significant changes in protein abundance among cell populations were identified by one‐way ANOVA. For a broad picture of the categorized abundances and their associated functions, Figure S1 includes a global GO enrichment analysis for all significant proteins by ANOVA. This data may provide some insight into relevant pathways associated with specific abundance patterns within our proteomic data in the differentiating cells.

At the individual protein level, some relevant markers like myogenin showed a similar trend among populations between immunoblotting and the levels detected by mass spectrometry. Mef2C, a transcription factor required at early stages of differentiation (Liu et al., 2014) was increased in live cells, but not in the other cell populations. The levels of slow skeletal muscle troponin T (Tnnt1) and fast skeletal muscle troponin T (Tnnt3; not shown) were markedly upregulated in myotubes, consistent with their differentiated state. Proliferating cell nuclear antigen (Pcna) levels were high in myoblasts but decreased in other cell types, consistent with the cell cycle exit that precedes C2C12 differentiation (Guo et al., 1995).

A global visualization of the proteomic data by principal component analysis allowed us to investigate the relative similarities of the five cell populations (Figure 1d). This analysis showed that the live cells, reserve cells and myoblasts were the most similar to each other, with reserve cells and live cells clustering very closely. In contrast, the proteomes of myotubes and dead cells were very different from these three cell types and from each other. A Pearson correlation of the proteome of myoblasts to the proteomes of live, dead, myotubes and reserve cell populations showed that the proteomes of myoblasts and myotubes were the most different, consistent with cell differentiation, while the myoblasts, live cells, dying cells and reserve cells were more similar to each other (Figure 1e, left). A similar pattern was seen when a previously published myoblast proteome (Xiao et al., 2022) was used for the analysis (Figure 1e, right). Having established the relatedness of the cell populations and that our data were consistent with previously published reports, we further analyzed the proteomes. We looked at the gene ontology annotations for all proteins identified in the populations. Abundance of the significant proteins (p < 0.05 by ANOVA) identified within annotation terms describing myoblast differentiation and cell death were included in Figure 1f. Hierarchical clustering identified patterns of protein abundance that clearly discriminated between the cell populations. For example, similar to what we observed for myogenin levels, various proteins associated with myoblast differentiation were upregulated in the myotubes as expected (Csrp3 (Vafiadaki et al., 2015), Hif1an (Nguyen et al., 2023), Klhl41 (Ramirez‐Martinez et al., 2017)). Proteins associated with cell death (including the BH3 only proteins Bok and Bax) were upregulated in the dead population. Interestingly, some proteins with annotation to the execution phase of apoptosis and programmed cell death (Dffa, Stk4, Pak2, Stk17b, Capn3) showed higher levels in the non‐apoptotic populations. A summary of all the proteomic findings and comparisons can be found in Appendix S1.

2.3. Identification of caspase substrates associated with cell death and differentiation

We have previously shown that differentiating C2C12 cells have caspase‐3 activity (Dehkordi et al., 2020) and immunoblot data (Figure 1b) suggested the differentiating cells have lower levels of caspase‐3 activity than dead cells. We tested this using a cell‐permeable fluorescent probe. Cells in GM and DM were incubated with the probe for 1 day and the dead cells and live cells separated as outlined above (Figure 2). In some cases, cells in DM were cultured with the probe for 4 days (see Figure S2 for data on caspase activity in cells on day 4). The increased green fluorescence in live cells compared to myoblasts confirmed that the differentiating cells had caspase activity. Quantification revealed that the live cells showed five‐ to six‐fold more caspase activity than myoblasts and that the dead cells had an order of magnitude more caspase‐3 activity than the live cells (Figure 2b). The caspase activity of myoblasts was significantly decreased by QVD‐OPh, a pan‐caspase inhibitor that blocks C2C12 differentiation (Dehkordi et al., 2020; Murray et al., ²⁰⁰⁸).

FIGURE 2.

Different levels of caspase activity in differentiating and dying C2C12 cells. (a) Caspase activity C2C12 cell populations as assessed with 5 μM IncuCyte caspase‐3/7 green apoptosis assay reagent. The nuclei were stained with Hoechst 33342. The upper panel is the caspase‐dependent green fluorescence alone. The lower panel is the green and blue fluorescence overlaid. (b) Quantification of nuclear green fluorescence in live cells using an Operetta high content imaging system. Myoblasts (undifferentiated cells) incubated with the pan‐caspase inhibitor Q‐VD‐oPh were also assessed. Mean green fluorescent intensity within each cell nucleus was quantified using Harmony software.

Many proteomic studies have identified caspase substrates in an apoptotic context, but only a few have identified non‐apoptotic substrates (Conde‐Rubio et al., 2021; Hertz et al., ²⁰¹⁹; Victor et al., ²⁰¹⁸; Weghorst et al., ²⁰²²), and there are no proteomic studies of caspase substrates in differentiating C2C12 cells. We therefore aimed to use “forward” subtiligase N‐terminomics to identify caspase substrates in the five cell populations (Figure 3a). Briefly, cells from the different populations were lysed and the neo‐N‐termini were enzymatically labeled in vitro using an engineered subtiligase to identify proteins cleaved in the cells (Mahrus et al., 2008). The biotin‐containing peptide ester tag added at the neo‐N‐termini allowed enrichment on neutravidin beads, followed by protein digestion and subsequent peptide release using TEV protease. Successful labeling and capture efficiency were closely monitored using streptavidin blots and dot blots, respectively (Figure S3). Unambiguous identification of the cleavage sites on each cell population, by LC–MS/MS, was possible due to the unique mass‐tag (aminobutyric acid) left behind after enzymatic peptide release. A summary of all labeled peptides, unique cleavages, and substrates can be found in Appendix S2. Principal component analysis was used to investigate the relative similarities of the five cell populations (Figure 3b). We found a clear separation of all populations, the myoblasts and the live and reserve cells clustered together, while the myotubes and dead cells were different from each other and the other populations. While no separation was observed for the live and reserve cells at the proteome level, the data from proteomic identification of cleavage events suggests that these populations are different at the proteolytic level. Our N‐terminomics approach identified a total of 13,219 total Abu‐labeled peptides, covering 10,243 unique cleavage sites in 3717 proteins (Figure 3c). The distribution of these 10,243 cleavage events among the cell populations showed similar proteolytic background; the mean number of unique cleavages was 6370 (range 5805–6910), indicating substantial overlap between the different populations. The subset of cleavage sites where the previous residue featured an aspartic acid (P1 = D) were identified as putative caspase substrates. A summary of all putative caspase substrates cleaved during the differentiation process is included in Appendix S3. These events are equally distributed among populations with around 250 identifications (~4.0% of all cleavages detected) in all populations except in the dead cells, where the number of putative caspase substrates showed a significant increase (715 identifications, 12.4%). This is consistent with the higher level of caspase‐3 like activity detected in this population. Figure 3d depicts the unique cleavage overlap between day 0 and day 1. It is worth noting that within the subset of putative caspase substrates, there are several proteolytic events that were not observed in the dead cells. These could correspond to non‐apoptotic cleavage events at the beginning of the differentiation process. Interestingly, cleavage of proteins reported to be cleaved by caspases during differentiation (Stk4/Mst1, ICAD, Pax7) was not detected in this data set (Table S1). However, one of the proteins, Stk4/Mst1, was cleaved in the live cell population, but at a P1 = N, not P1 = D.

FIGURE 3.

Identification of cleavage events in isolated cell populations during the time course of differentiation. (a) General workflow for the identification of cleavage events using a forward N‐terminomics approach for N‐terminal labeling. A total of two replicates per cell population (5 × 10⁸ cells each) were included. (b) PCA analysis of the forward N‐terminomics data from the different cell populations. Each point in the PCA plot represents the data for every replicate (n = 2). The color of each circle matches the population color‐code. (c) Summary of all identified and quantified labeled‐peptides, unique cleavages and substrates. The total number of identified unique cleavage events per population is depicted in the right panel. Additionally, the number and percentage of cleavage events matching caspase specificity (P1 = D) is also shown. A detailed list of all proteolytic events is provided in Appendix S2. (d) Venn diagram showing the identified cleavage sites overlap between myoblast, live cells, and dead cells. (e) Sequence logo showing P4–P4′ residue enrichment for all cleavages with P1 = D in every population. (f) Predicting which caspases contribute to the cleavages in each population. The dot plot depicts the reliability (R) score for every prediction, the number is a measurement of the confidence of the alignment between protease specificity and cleavage sites detected. D cells, dead cells; L cells, live cells; MB, myoblast; MT, myotubes; R cells, reserve cells.

Analysis of the P4–P4′ amino acids for cleavages with P1 = D (Figure 3e) found in each population showed that the most common P4 and P2 amino acids were D and L/V respectively, with more variation in the P3 position. The most common amino acids in the P1′ and P2′ position were G/S/A and V/A respectively. This pattern closely resembled caspase‐3/7 cleavage sites, but it could also be amplified from subtiligase intrinsic labeling specificity as well (Weeks & Wells, 2018). There was a close similarity between the sequence logos for the cell populations. The sequence specificity and distribution of all cleavages quantified in the different cell populations, and the levels for putative caspase substrates in GO terms involved in cell differentiation and development, muscle function, and cell death are shown in Figures S4 and S5 respectively. Once again the hierarchical clustering shows that not all P1 = D cleavages mapped only to the dead cells; some cleavages are also relevant in other populations. Importantly, some proteins that play a key role in muscle differentiation, development and organization are cleaved in the dead cells and we hypothesize that excessive cleavage of these factors could lead to failure of the differentiation process, and contribute to the decision to die.

To further investigate which caspases might be implicated, we scored how closely the amino acids around the cleavage sites (P4–P4′) in each cell population matched canonical recognition sequences. The canonical sequences used were: W/L‐E‐H‐D for caspase‐1, ‐4 and ‐5; V‐D‐E‐V‐D for caspase‐2; D‐E‐V‐D for caspase‐3 and ‐7; V‐E‐H‐D for caspase‐6; L‐E‐T/H‐D for caspase‐8, ‐9 and ‐10; and W‐E‐S/T‐D for caspase‐14 (Figure 3f). A reliability score (R score) for each comparison was generated, with a higher score as an indicator of a greater correlation between cleavage site information and the protease specificity. Our N‐terminomics data showed higher R values for the caspase‐3 and ‐7 group, followed by caspase‐2 and caspase‐6. These proteases might be the main contributors to the cleavages with P1 = D among the multiple caspase groups. Among populations, the highest reliability score was observed for caspase‐3 and ‐7 in dead cells, which is expected for apoptotic cells. The caspase‐3 and ‐7 R score for live cells, reserve cells and myotubes was greater than for myoblasts, mirroring the level of caspase‐3 activity in those cell populations. The pattern of R scores for caspase‐2 was similar, with dead cells being highest and with live cells, reserve cells and myotubes being greater than myoblasts, which may reflect the role of caspase‐2 in C2C12 death and differentiation, although the role of caspase‐2 in apoptosis has been questioned (Delgado et al., 2013). The R scores for the other caspase groups did not show any clear evidence that any particular caspase is more associated with a specific cell population.

Caspase‐3 activity increases by the first day of differentiation (Boonstra et al., 2018; Fernando et al., ²⁰⁰²; Murray et al., ²⁰⁰⁸), and the proteins cleaved at this time are considered important for initiating the differentiation. Only 18 cleavages were unique to live cells on day 1 (Figure 3d), but we hypothesized that there might be quantitative differences on day 1 that are important for differentiation. To better understand which proteins are being cleaved at this critical time, we compared the level of cleavage in myoblasts, live cells and dead cells using label‐free quantification of Abu‐labeled peptides (Figure 4a). This was feasible due to the similar number of peptide identifications and proteolytic background observed in all cell populations. The graph generated depicts the relative level of cleavage in the three cell populations based on the quantification values. All unique cleavages are represented with P1 = D show in yellow and all others in gray. Our analysis showed that the greatest number of cleavage events with P1 = D specificity were enriched in the dead cell population. However, it also showed cleavage events that were associated with the myoblast (Ripor1 involved in cell migration (Lv et al., 2022), Ccnl2 inhibitor of differentiation (Zhuo et al., 2009)) or the live cell population (Ctsb (Jane et al., 2002; Jane et al., ²⁰⁰⁶), Hrc) and that are relevant in muscle development (Anderson et al., 2004).

FIGURE 4.

Quantification of putative caspase substrates during key events of the differentiation process. (a) Distribution of putative caspase cleavages at the early stage of differentiation. The triplot shows the relationship between label‐free peptide quantification at day 0 (myoblasts) and day 1 populations (live and dead cells). Dots in the center of the plot represent cleavage events with equal levels in all populations, while those closer to the vertex of the triangle represent a cleavage with high levels in that particular population. Dots in gray correspond to all cleavage events identified. Cleavages with P1 = D are highlighted in yellow; within those, dots with red outline correspond to proteins with significant differences (p‐value <0.05) in any of the pairwise comparisons. (b) Clustering using the level of the cleavage products over the time course of differentiation day 0 (myoblasts), day 1 (live cells), and day 4 (myotubes) and grouped as myoblast‐associated cleavages, early‐stage cleavages, and late‐stage cleavages. Each graph shows the abundance of every peptide included in that group (in gray) and the average abundance (in red). Gene ontology enrichment analysis for every group is also shown.

Cleavage at the site we identified in cathepsin B (D79), removes the pro domain and is likely to generate an active form of cathepsin B (Figure S6a). Furthermore, the abundance of this peptide is low at the beginning of the differentiation (myoblasts), and increases after day 1 with the highest abundance in live and reserve cells (Figure S6b). This finding correlates well with the number of identified cleavage sites that match cathepsin B specificity (P1 = R, P1′ = F), with the lowest number of cleavages observed in the myoblast and highest in live and reserve cells (Figure S6c).

C2C12 differentiation takes 4–5 days and our data set also contains information about the time course of cleavage events. By comparing the abundance of cleavage events detected in myoblasts (day 0), live cells (day 1) and myotubes (day 4) we identified proteins that are cleaved early and late in the differentiation process (Figure 4b). Hierarchical clustering of 148 the P1 = D cleavage events detected in the myoblast population, live cell population and myotube population revealed stage specific cleavage events. An interesting pattern is observed in cluster 4 where most proteins grouped represent cleavage events in the L cell population. These cleavage events represent potential substrates that could determine the cell faith at the early stages of differentiation (included in Appendix S3). Enrichment analysis of 54 proteins cleaved specifically in myotubes on day 4 showed a significant over representation of proteins involved in muscle function and the cytoskeleton. In contrast, the different processes associated with the 32 proteins cleaved specifically on day 1 do not have obvious links to muscle biology, although NF‐κB signaling regulates both apoptosis and C2C12 differentiation. Other processes identified include regulation of the cytoskeleton of cell migration, which is important for myotube formation.

Some substrates were detected in live cells on day 1 and also in myotubes 3 days later (cluster 5), suggesting that the cleavage of these substrates is either that very stable fragments are generated by caspase cleavage on day 1 or cleavage is ongoing over 3 days. Gene ontology and pathway analysis of these substrates was limited by their low number (six substrates).

2.4. Validation of caspase‐3 cleavages in early stages of differentiation by reverse N‐terminomics

To gain insight into the set of possible caspase‐3 substrates in C2C12 cells during the early stages of differentiation, we also performed a reverse or in vitro N‐terminomics approach. This involved adding purified recombinant mouse caspase‐3 to lysates from myoblasts and live cells and enabled us to identify proteins directly cleaved by caspase‐3 (Figure 5a), in contrast to the forward N‐terminomics approach in live cells which identified proteins cleaved in live cells and by the endogenous enzymes. Figure S7 includes the details of protein expression, purification and biochemical characterization. The k _cat/K _M for mouse caspase‐3 was 2.8 × 10⁵ M⁻¹ s⁻¹, compared to the previously reported value for human caspase‐3 (1 × 10⁶ M⁻¹ s⁻¹). A total of 2180 peptides were Abu‐labeled, identifying 1662 cleavage sites in 1000 proteins (Figure 5b). A total of 454 cleavage (32.7%) out of 1407 cleavage events for myoblasts, and 415 cleavages (31.6%) out of 1312 events for live cells showed the characteristic caspase specificity (P1 = D). An analysis of the sequence representation around the P4–P4′ positions (Figure 5c) revealed a pattern consistent with caspase‐3 cleavage for each population, as expected. When compared to the patterns observed by the in vivo N‐terminomics approach, a clear similarity between sequence logos was observed for the live cells, but in the case of myoblast some differences were identified, with Leu as the most represented residue at P2 and P3 positions in vivo. This difference is likely because live cells have a relatively high level of caspase‐3‐like activity compared to myoblasts, and adding recombinant caspase‐3 to the myoblast lysate probably has a bigger impact on the N‐terminomics data compared to adding caspase‐3 to a lysate that already has a higher caspase‐3 activity.

FIGURE 5.

Identification of mouse caspase‐3 cleavage sites at the beginning of myoblast differentiation using reverse N‐terminomics. (a) General workflow for the identification of mouse caspase‐3 cleavages in lysates from myoblasts and day 1 (live cells). Two technical replicates per population (5 × 10⁸ cells each). (b) Total number of identified and quantified labeled peptides, unique cleavages and substrates. Identified unique cleavage events per population are also included (right panel). (c) Sequence logo for all cleavages with P1 = D. (d) Venn diagram showing unique substrate overlap between forward and reverse N‐terminomics. (e) In vitro cleavage assay in HEK293 cells lysates overexpressing selected substrates.

Comparing the in vitro caspase‐3 N‐terminomics data sets for myoblasts and live cells showed that 80% of substrates (with P1 = D) observed in both populations (data not shown). The high degree of overlap is expected considering the similarity between the proteomes of these two cell populations. Comparison of the in vivo and in vitro myoblast data sets found only 64 common substrates (Figure 5d). Again, this most likely reflects the low caspase‐3‐like activity in the myoblast population compared to the high caspase‐3 activity generated by addition of exogenous caspase in the in vitro proteomics workflow. A slight increase in the number of common substrates (77) was observed for the live cells, which could explain the close similarity in sequence logos for in vitro and in vivo experiments for this population, and this increase could also reflect the higher activity of caspase‐3 at day 1.

2.5. Immunoblotting of caspase‐3 substrates

We also tested the cleavage of 10 putative caspase‐3 substrates (2 of which were also identified by reverse N‐terminomics) by a complementary method to further validate substrate identification (Figure S8). Mouse protein constructs were expressed with N‐terminal 3xFLAG and C‐terminal HA tags in HEK293 cells and purified recombinant mouse caspase‐3 was added to cell lysates before assessing protein cleavage by immunoblotting. Of the 10, seven were cleaved by caspase‐3. NFκBIA cleavage at SLGD₃₆ and ATP5IF1 cleavage at FVSD₂₉ was detected in the live cell, reserve cell and myotube populations, but not in the myotube or dead cell populations (Figure 5e). Some proteins generated cleavage products consistent with the cleavage site detected by proteomics. For example, caspase‐3 cleaved full length NFκBIA and CLTA to produce fragments consistent with the cleavage detected by mass spectrometry (Figures 5e and S9). CLTA cleavage at DAVD₇₆ and GPTD₉₂ was detected in all cell populations. Other proteins did not produce detectable cleavage products or products were inconsistent with the cleavage site identified by proteomics. For example, HIP1 was cleaved by caspase‐3 near the N‐terminal and not at a site consistent with the proteomics data (Figure S9). Caspase‐3 cleaved EMD and LMO7 without generating cleavage products detectable with the anti‐HA antibody. Blotting using an anti‐FLAG antibody detected an LMO7 product consistent with cleavage at ILDD₇₅₈, which was detected in dead cells. Two other detected cleavage products were not consistent with other sites identified by proteomics (GSSD₁₂₀₀ in live cells and LEDD₁₃₂₂ in reserve cells, myotubes and dead cells). The anti‐FLAG antibody detected an EMD cleavage product that is too large to be the fragment detected in the live and dead cell populations by cleavage at SDLD₆₆ (Figure S9 shows the complete blots), but consistent with a cleavage seen in apoptotic C2C12 cells (Columbaro et al., 2001).

3. DISCUSSION

3.1. Processes in differentiating cells implicated by the substrates cleaved

Caspase substrates involved in C2C12 differentiation previously identified by candidate approaches include STK4 leading to increased kinase activity (Fernando et al., 2002), ICAD leading to DNA‐damage dependent changes in gene expression (Larsen et al., 2010), and SAT2B leading to chromatin reorganization (Bell et al., 2022). However, there are no prior proteomic studies of caspase substrates in C212 cells, and only a few proteomic studies of non‐apoptotic caspase substrates in any other context (Conde‐Rubio et al., 2021; Hertz et al., ²⁰¹⁹; Victor et al., ²⁰¹⁸; Weghorst et al., ^{2020, 2022}). Here, we identified caspase‐like cleavage events early and late in differentiation (Figure 4) that implicate specific processes regulated by caspases during differentiation.

Cleavage of proteins involved in the cytoskeleton or regulation of the cytoskeleton was prominent early and late in differentiation. Gene ontology enrichment analysis showed that the late‐stage cleavages were also associated with muscle functions and development. Many of these substrates were also cleaved in dead cells. However, we also identified proteins cleaved in differentiating cells that were not cleaved in dead cells. This included proteins with links to myogenesis: Tanc1 and Myo18a with conserved roles in Drosophila and mouse myogenesis (Avirneni‐Vadlamudi et al., 2012; Bonn et al., ²⁰¹³; Rau et al., ²⁰⁰¹) and also bind to each other (Yang et al., 2024). Tanc1 is crucial for cell fusion, a process which is dependent on caspase activity. The cleavage of Tanc1 (VAVD₁₆₂₉) or Myo18a (VHRD₃₉) is not predicted to affect their interaction (Yang et al., 2024), but the cleavage of both perhaps indicates that caspase‐3 is targeting a complex formed during fusion. Galectin‐1 was another substrate that is required for myoblast fusion. Other proteins have roles in regulating RNA and intracellular calcium.

Our data identified cleavage of proteins involved in a range of cellular processes. This is also true in apoptosis, where cleavage of proteins controlling diverse aspects of cell biology causes the changes typical of apoptosis irrespective of cell type. Thus, it seems likely that differentiation, like apoptosis, is not dependent on a single key substrate, but that caspases orchestrate disparate processes necessary for differentiation. Our data suggest that there are substrates specific to muscle differentiation, but do not exclude the possibility of a set of non‐apoptotic substrates cleaved in very different non‐apoptotic contexts.

The small number of proteomic studies to identify caspase substrates in other non‐apoptotic contexts have used neurons, and in one case, a tumor cell line. By searching the substrates cleaved in these other non‐apoptotic contexts, we identified 23 substrates that are cleaved in three or more of the five studies. The identity of these 23 substrates underscores the importance of caspases regulating the cytoskeleton in non‐apoptotic processes (see Figure S10). Cathepsin B was also identified in multiple studies. Cathepsin B facilitates C2C12 fusion (Gogos et al., 1996; Jane et al., ²⁰⁰²), and cathepsin B cleavage was highest in live cells (Figure S6), and the detected cleavage site may activate cathepsin B (Hook et al., 2022). Active caspase‐3 (Ghribi et al., 2002; Shaulov‐Rotem et al., ²⁰¹⁶) and active caspase‐7 (Rao et al., 2001) have been reported in the endoplasmic reticulum, and several of the substrates identified in live cells are secreted or ER proteins (e.g., galectin‐1, osteonectin, hepatoma‐derived growth factor‐related protein 3, sparc, calumenin, herpud2, and reticulocalbin 2). This suggests that caspase‐3/7 activity in the ER of the differentiating cells and that there is a previously unsuspected link in differentiation between caspases and cathepsin B. Thus, the data provided here are a major contribution to the caspase substrate landscape, reporting new apoptotic substrates identified in a muscle context as well as identifying caspase substrates associated with non‐apoptotic outcomes.

The data presented here show for the first time that differentiating C2C12 cells have much lower caspase activity than the dying cells. Other examples of non‐lethal caspase activity in which the level of activity is controlled have also been reported (Florentin & Arama, 2012; Kang & Bashirullah, ²⁰¹⁴; Li, Jo, et al., ²⁰¹⁰; Yang & Widmann, ²⁰⁰²). The data also identify cleavage sites that appear to be specific to particular fates. In some cases, a particular protein was cleaved at different sites, depending on the cells' fate. Why the cleavage of these substrates differs in C2C12 cells assuming different fates is unclear, but in other contexts, substrate cleavage is controlled by post‐translational modifications (Maluch et al., 2021; Turowec et al., ²⁰¹⁴) and/or subcellular co‐localization of active caspase (Amcheslavsky et al., 2018; Hertz et al., ²⁰¹⁹; Li, Jo, et al., ²⁰¹⁰). The dead cell population shows the greatest evidence of proteolysis and the activation of caspase‐3 explains the large number of cleavage events where P1 = D. There are also increased cleavage events where P1 ≠ D, which may be due to activation of non‐caspase proteases like calpain during apoptosis. However, the presence or absence of a cleavage event in a particular population should be interpreted cautiously. For example, the presence of NFκBIA cleavage products in live cells but not dead cells may indicate either (1) this protein is selectively cleaved in live cells but not dead cells or (2) that NFκBIA is cleaved in both but that the caspase cleavage product is more rapidly removed in apoptotic cells.

3.2. The contribution of different caspases to the cleavages detected

The role of caspases in C2C12 differentiation is complex, with caspase‐2, caspase‐3, caspase‐7 and caspase‐9 all having demonstrated roles (Bell et al., 2022; Boonstra et al., ²⁰¹⁸; Dehkordi et al., ²⁰²⁰; Fernando et al., ²⁰⁰²; Murray et al., ²⁰⁰⁸). The analysis of the P4–P4′ sequences revealed a strong signature for caspase‐3/7 cleavage. Like us, Hertz et al. (2019) report non‐apoptotic cleavage at sites typical of caspase‐3 (D‐E‐(V/L)‐D) in a non‐apoptotic axonal degeneration model. In contrast, Weghorst et al. (2022) report an unusual set of cleavage sites (Figure S11) that are enriched in cytoskeletal proteins in a non‐apoptotic neuronal development model. In particular, there is no preference for a P4 = D that is typical for a caspase‐3 cleavage site. In our data, an atypical cleavage sequence (R‐N‐V‐D) was also associated with the 46 proteins (17% of proteins cleaved in live cells where P1 = D) cleaved in the live cell population but not in the dead cell population. The different motifs identified are intriguing and maybe linked to post‐translational modification of substrates, different caspases contributing to the cleavage or even other proteases being involved. However, nine of these 46 (Khdrbs3, Kmt2b, Ctnnd1, Rps21, Myo18a, Hnrnpf, Tanc1, Atp5if1, Ctsb) were also cleaved by caspase‐3 in our reverse N terminomics experiment suggesting that they are bona fide caspase substrates, despite the lack of preference for P4 = D. Interestingly, one of these (Tanc1) is a protein involved in cell fusion and known to be vital for myogenic differentiation. Thirty‐seven proteins cleaved in live cells were not confirmed by the reverse N terminomics using caspase‐3. These may be substrates for caspase‐2, caspase‐7 or caspase‐9, which are also involved in C2C12 differentiation, or indeed other proteases.

Our experiments testing caspase‐3 mediated cleavage of selected proteins expressed in HEK293 cells were consistent with the proteomic data in some cases, but not others. For example, proteomics detected NFκBIA cleavage at SLGD₃₆ in the live, reserve and myotube populations, but not in the myoblast or dead cell populations. Immunoblotting detected a cleavage product consistent with this site. NFκB can both inhibit myogenesis and promote myotube function (Bakkar et al., 2008), and it is not clear whether the cleavage at SLGD₃₆ enhances or diminishes NFκB signaling. In the case of LMO7, proteomics detected several cleavage sites, but immunoblotting of caspase‐3 treated LMO7 in HEK293 cell lysates only generated a product consistent with cleavage at ILDD₇₅₈, which was detected on the dead cell population and no other. LMO7, like NFκB, is a transcription factor that plays a role in myogenesis (Dedeic et al., 2011). It is a multifunctional protein that translocates to the cytoplasm from the nucleus during muscle cell differentiation. It also binds to EMD (Holaska et al., 2006), another protein known to be important in muscle differentiation, which was also cleaved in our dataset. Loss or mutation of LMO7 or EMD is also linked to muscular dystrophies (Mull et al., 2015). Cleavage of EMD during C2C12 cell apoptosis produces a ~20 kDa fragment (Columbaro et al., 2001), consistent with the ~20 kDa FLAG‐tagged fragment we detected after incubating EMD with mouse caspase‐3 (Figure S9), but not consistent with cleavage at SDLD₆₆ detected by proteomics in myotubes and dead cells. Thus, further studies are needed to unravel the association of proteolytic processing events, localization and function and the outcome in the cell. This includes assessing cleavage by other caspases playing roles in muscle differentiation (caspase‐2, caspase‐7, and caspase‐9) (Bell et al., 2022; Boonstra et al., ²⁰¹⁸; Dehkordi et al., ²⁰²⁰; Murray et al., ²⁰⁰⁸). Nevertheless, this study represents a starting point in the discovery and description of the complex events behind muscle development and muscular dystrophy.

In summary, we have generated enriched populations of undifferentiated, differentiating, dying and differentiated C2C12 cells and used proteomics and N‐terminomics to produce the first proteolytic cleavage report in C2C12 cells assuming different fates. These data have identified new caspase substrates, including caspase substrates specifically associated with differentiation and shown that caspases are regulating proteins involved in myogenesis in myotubes, 2–3 days after caspase‐3 initiates differentiation.

4. METHODS

4.1. C2C12 culture and differentiation

C2C12 myoblasts (Sigma‐Aldrich) were maintained at <70% confluency in Dulbecco's Modified Eagle's Medium (DMEM, Sigma‐Aldrich, D5796) supplemented with 20% FBS (Sigma‐Aldrich, D7524) and 1% Penicillin/streptomycin (Sigma‐Aldrich, P4333), referred to as growth medium (GM) inT175 flasks (Sarstedt, 83.3912.002). To induce differentiation, the cells were seeded into μ‐Plate 96 Well (ibidi, 89626) at 3 × 10⁴ cells per well and incubated at 37°C and 5% CO₂ overnight. The next day, cells were washed three times with Hank's balanced salt solution (HBSS, Sigma‐Aldrich, H6648) to remove GM and then cultured in DMEM supplemented with 2% horse serum (Sigma‐Aldrich, H1138) and 1% Penicillin/streptomycin, referred to as differentiation medium (DM), for 3–4 days.

4.2. Preparing different cell populations

4.2.1. Myoblasts

The myoblast population consists of C2C12 cells maintained at no more than 70% confluence in GM. Cells were seeded in 15 cm dishes at a density of 2250 cells/cm². After 48 h, cells were harvested using trypsin–EDTA (Sigma‐Aldrich, T4174) and collected as the myoblast population.

4.2.2. Live cells and dead cells

C2C12 cells were seeded at a density of 3 × 10⁴ cells/cm² in 15 cm dishes. After 24 h, GM was changed to DM. On day 1 of differentiation, the medium containing floating dead cells was collected and centrifuged to collect the dead cell population. The remaining attached cells were washed and harvested using trypsin–EDTA and collected as the live cell population.

4.2.3. Myotubes and reserve cells

C2C12 cells were seeded at a density of 3 × 10⁴ cells/cm² in 15 cm dishes. After 24 h, GM was changed to DM. On day 4, cells were washed using PBS, and trypsin was used. Myotubes detached within 1–2 min after adding trypsin, while the reserve cells remained attached to the plate. Detached multinucleated cells were collected as the myotube population. Attached mono‐nucleated cells were detached by longer incubation in trypsin and collected as the reserve cell population.

4.3. Cell‐based assay for caspase activity in C2C12 cells

C2C12 cells were seeded at a density of 3 × 10⁴ cells/cm² in μ‐Plate 96 Well. After 24 h, the GM was replaced with DM supplemented with 5 μM IncuCyte® Caspase‐3/7 Green Apoptosis Assay Reagent (Cat. No. 4440). On day one of differentiation, floating dead cells were collected into separate wells and centrifuged at 200 g for 1 min. Live cell imaging was performed on days 1 and 4 of differentiation using Operetta High Content Imaging System to visualize cell morphology and detect increased green fluorescence within the nuclei, which is indicative of caspase activation. Mean green fluorescent intensity within each cell nucleus was quantified using Harmony Software.

4.4. Immunoblotting

Fifty to 100 μg cell lysate was mixed with Laemmli buffer and incubated at 95°C for 5 min. Proteins were resolved on 8%–15% acrylamide gel depending on the protein size and transferred (100 V, 1–2.30 h depending on the size of the proteins of interest) onto a nitrocellulose membrane. The membrane was blocked in 5% Skim milk in PBS + Tween 20 (0.1%; PBST) for 1 h and then incubated with the primary antibody (1:1000 in PBST) overnight at 4°C. The membrane was washed three times (3 × 10 min) with PBST and then incubated with the secondary antibody (IRDye® 800CW Goat anti‐Rabbit IgG Secondary Antibody, Licor, 926‐32211) (1:15,000 in PBST) for 1 h at room temperature. After three washes with PBST membrane was scanned by Li‐Cor Odyssey Scanner.

Primary antibodies: Myogenin (Myogenin Monoclonal Antibody (F5D), eBioscience, Invitrogen, 14‐5643‐82), MyoD (MYOD Polyclonal Antibody, Invitrogen, PA5‐23078), Active caspase‐3 Antibody (Cleaved Caspase‐3 (Asp175), 9661, Cell Signalling) and Myosin heavy chain (MHC) (Myosin 4 Monoclonal Antibody (MF20), Invitrogen, 14‐6503‐82). Secondary antibodies: IRDye® 800CW Goat anti‐Rabbit IgG, 926‐32211 and IRDye® 680RD Goat anti‐Mouse IgG, 926‐68070.

4.5. Immunofluorescence staining

C2C12 cells in a 96‐well plate were fixed by adding 200 μL of 4% PFA to each well for 20 min at room temperature. Cells were then washed three times with HBSS. Cells were permeabilized by incubation for 10 min in 200 μL of PBS‐Triton (0.1%). The cells were washed again with HBSS and incubated for 1 h in 200 μL of blocking buffer (10% FBS in PBS). The blocking buffer was replaced with 60 μL of anti‐myosin heavy chain diluted in blocking buffer and the plate was incubated overnight at 4°C on a rocker. The following day, the cells were washed three times with HBSS and 60 μL of the secondary antibody diluted in blocking buffer was added to each well. The plate was protected from light and incubated on a rocker for 1 h. After incubation, the antibody solution was replaced with 200 μL of Hoechst stain (2 μg/mL) for 1 min. Cells were then washed three more times with HBSS, 200 μL of HBBS added, and the plate was proceeded to imaging. Filters employed for detecting Hoechst were set to (excitation 360–400 nm and emission 410–480 nm), while those for detecting Alexa fluor488 were configured to (excitation 460–490 nm, emission 500–550 nm).

4.6. Protein expression, purification and activity assay

Recombinant His‐tagged mouse caspase‐3 (pET23b‐mouse casp‐3‐His) was expressed in BL21 (DE3) pLysS E. coli. Transformed E. coli were plated on Mueller‐Hinton agar with 100 μg/mL ampicillin and 25 μg/mL chloramphenicol at 37°C overnight. One colony was added to 10 mL of LB media (Fisher Scientific) with ampicillin and chloramphenicol. The starter culture was used to inoculate 1.5 L of LB broth, supplemented with ampicillin and chloramphenicol, at 37°C until the O.D. reached 0.6. Protein expression was induced for 14 h at 20°C with 0.5 mM IPTG (Fisher Scientific). Cell pellets were collected by centrifugation 6000 g for 30 min at 4°C and the pellet was resuspended in 45 mL of lysis buffer (100 mM NaCl and 100 mM Tris pH 8.0). Cells were lysed by high‐pressure homogenization (Avestin EmulsiFlex C3), then centrifuged at 16,000xg for 30 min at 4°C. The clarified protein supernatant was passed through a 1 mL Ni²⁺ affinity column (Cytiva Inc.) and eluted with a linear gradient of buffer (250 mM NaCl, 200 mM imidazole, 100 mM Tris pH 8.0). The eluted protein fractions were assessed for purity using SDS‐PAGE and the mouse caspase‐3 fractions were pooled.

Enzymatic efficiency of the purified mouse caspase‐3 was analyzed in vitro with a coumarin‐based fluorescent probe, Ac‐DEVD‐AFC as previously described by Araya et al. (2021).

4.7. Sample preparation for proteomics and N‐terminomics

C2C12 cell pellets (~1 billion each) were thawed on ice and resuspended in 1 mL of lysis buffer per 100 million cells (100 mM HEPES pH 7.4, 1% SDS, and protease inhibitors: 5 mM EDTA, 1 mM PMSF, 4 mM IAM, 1 mM AEBSF). Cells were lysed using a probe tip sonicator for 3 cycles (30% amplitude, 2 s on/2 s off for 2 min each cycle). Samples were incubated on ice for 15 min in the dark. The lysate was clarified for 10 min at 12,000×g to remove cell debris. Total protein concentration in the cell lysate was determined using a BCA Protein assay kit (ThermoScientific). A 500 μg protein aliquot was taken from every cell population for shotgun proteomics. For Forward N‐terminomics, Triton‐X (2.5% final) was added to the resulting lysate, and the samples were clarified one last time. The resulting lysate was divided into two as technical replicates for each condition.

For in vitro caspase‐3 N‐terminomics, myoblast and live cell pellets were used. No SDS was added to the lysis buffer, 0.1% Triton X‐100 was used instead. The excess of IAM was quenched with 20 mM DTT prior to the addition of mouse caspase‐3. Caspase activity buffer was added to each sample from a 10× stock (for final concentrations of 20 mM HEPES pH 7.4, 50 mM KCl, 1.5% sucrose, 0.1% CHAPS, 10 mM DTT) and the sample was divided into two for technical replicates. Lysates were incubated with 0.25 μM caspase‐3 in the dark at room temperature for 2 h and aliquots were taken to assay for protease activity. After the incubation, the enzyme was inhibited with 100 μM z‐VAD‐fmk.

For proteomic analysis, a total of 50 μg of protein from each lysate (n = 5) were processed using ProTrap XG cartridges (Allumiqs) following the manufacturer's protocol with some modifications. Briefly, NaCl was added to a final 100 mM concentration and the volume was adjusted to 100 μL with H₂O. Proteins were precipitated with 400 μL of room temperature acetone for 30 min. The entire cartridge was spun down at 2500g for 2 min and the pellet was washed with acetone. The pellet was resuspended in 100 μL of 8 M urea by vortexing for 30 s, bath sonication for 10 min, and incubation at room temperature for 30 min. The samples were diluted to 500 μL with 100 mM Tris buffer (pH 8). Reduction and alkylation were done with 5 mM DTT and 10 mM iodoacetamide, respectively, at 37°C for 30 min, then 15 mM DTT was added. Digestion was initiated by addition of trypsin at a 50:1 (protein:enzyme) mass ratio. Samples were incubated at room temperature overnight. The reaction was quenched by addition of trifluoroacetic acid (TFA) (final 2.5%). The peptides were desalted using a SPE column in the kit. The cartridge was primed (300 μL ACN), equilibrated (300 μL of 0.1% TFA in H₂O), loaded twice with the peptide mixture, and washed (300 μL 5% ACN, 0.1% TFA in H₂O). Peptides were eluted (300 μL of 50% ACN, 0.1% TFA in H₂O) into a new tube and dried down using a speedvac and stored at −20°C until LC–MS/MS analysis.

4.8. N‐terminal labeling and enrichment

Right before labeling pH was adjusted to 8.5 with 1 M bicine pH 9.2. Lysates from forward and reverse approaches were incubated with 1 mM TEVest6 biotin ester peptide and subtiligase (1 μM wild‐type and 1 μM M222A mutant) for 2 h (Weeks & Wells, 2018). Labeling was monitored by streptavidin blots, pre‐labeling and post‐labeling samples were loaded on the gel, separated, and transferred to nitrocellulose membranes, and membranes were blocked with commercial blocking buffer (LI‐COR Inc.). For detection of biotinylated proteins, membranes were incubated with streptavidin‐800 (IRDye 800CW, LI‐COR Inc.) diluted (1:10,000) in commercial blocking buffer for 30 min, and membranes were washed 3× with TBST for 5 min before scanning. Upon confirmation of successful labeling, proteins were precipitated in cold acetonitrile at −20°C overnight. The precipitate was recovered by centrifugation at 12,000g for 20 min. The protein pellet was resuspended in 8 M guanidine hydrochloride and boiled with 100 mM tris(2‐carboxyethyl)phosphine (TCEP) for 15 min. Once cooled down, iodoacetamide was added to a final 4 mM concentration for alkylation of the free Cys and incubated in the dark for 1 h at 37°C in the dark. Excess of IAM was quenched with 20 mM DTT and proteins were precipitated again in cold ethanol (100 proof) at −20°C overnight to remove excess of biotin ester probe. The following morning, the precipitate was recovered by centrifugation, washed two times with cold ethanol and resuspended in 3 mL of preheated 8 M guanidine hydrochloride. Once the precipitate was dissolved, samples were diluted to 2 M guanidine with PBS. 1.5 mL of pre‐washed neutravidin agarose beads (three washes with PBS) were added to each sample and incubated overnight at room temperature on a rotator. Capture efficiency was measured by dot blot using the same streptavidin IRDye 800CW. Precapture and postcapture samples were diluted 1:5, 1:10 with H₂O. A small aliquot (2 μL) of undiluted and diluted samples was directly loaded into nitrocellulose membranes, and samples were dried for 15 min at room temperature. Blocking, washes, and scanning were performed as described for streptavidin blots. Once capture efficiency was >90%, the beads were transferred to plastic columns and washed three times with a biotin wash solution (1 mM biotin, 10 mM bicine pH 8.0), then washed five times with 4 M guanidine hydrochloride, followed by five washes with PBS and 5 more washes with digestion buffer (100 mM NH₄HCO₃ pH 8.0). Twenty micrograms trypsin (Promega) were added into each sample and incubated overnight at room temperature on a rotator. The next day, the tryptic peptides were washed from the beads using 4 M guanidine hydrochloride (five times), followed by 10 washes with digestion buffer. The neutravidin beads were then resuspended in Tobacco Etch Virus (TEV) protease buffer (100 mM NH₄HCO₃, 10 mM DTT, and 1 mM EDTA). Twenty‐five micromolar TEV protease were added to each column and incubated overnight at room temperature on a rotator. The next day, the supernatant was recovered. Peptides were further extracted from the bead by incubation with 50%ACN in H₂O. Fractions were combined and dried on a Genevac® solvent evaporator (SP Scientific). When dry, the samples were solubilized in 2.5% trifluoroacetic acid (TFA) and incubated for 15 minutes at room temperature to precipitate the TEV protease. Samples were centrifuged and the supernatant was desalted using C18 desalting resin tips (Pierce C18 Ziptip). The eluted solution from the desalting step was dried down and stored at −80°C until LC–MS/MS analysis.

4.9. Mass spectrometry

Proteomics and N‐terminomics peptide samples were recovered in buffer A (3.9% CAN, 0.1% formic acid in H₂O). Samples were analyzed on a nanoflow‐HPLC (Thermo Scientific EASY‐nLC 1200 System) coupled to an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific). Peptide separation was done on an Aurora Ultimate™ analytical column and emitter (25 cm x 75 μm ID with 1.7 μm media, IonOpticks). Peptides were eluted with a solvent B linear gradient (0.1% FA in 80% ACN) for 120 min. The gradient was run at 400 nL/min with analytical column temperature set at 45°C. A data‐independent acquisition (DIA) method was used for the proteomic comparisons as previously reported by Eskandari‐Sedighi et al. (2023). All N‐terminomic samples were acquired in data‐dependent mode (DDA).

4.10. Data analysis for proteomics

Label‐free quantification of proteins for the DIA data (proteomics) was performed in the software Spectronaut (v18) using the direct analysis workflow with default settings. The database for the searches and GO annotation was the Uniprot mouse proteome (2021, 55,336 sequences). Trypsin/P was selected as the digestion enzyme with two missed‐tryptic cleavage tolerance. The search was performed with a maximum false discovery rate of 1% for peptides. Peptide modifications were set as: carbamidomethylation of C (fixed), deamidation of N/Q (variable), and oxidation of M (variable). Ratios for all the pairwise comparisons were generated, and statistically significant differences were determined by a one‐way ANOVA at 5% significance.

For the proteome comparison within isolated cell population, the list of proteins was sorted by p‐value from the ANOVA test. Next, the proteins with non‐zero values were filtered and the top 10% of proteins in the list were considered the highly variable proteins (HVPs) (Xiao et al., 2022). Pearson's correlation was performed between HVPs from each replicate of all cell populations against all D0 (myoblast) replicates in our study. Additionally, we compared the HVPs in our proteomic data sets to previously published myoblast (t0h) proteome data (Xiao et al., 2022).

Finally, the abundance of significant proteins (p‐value <0.05) with GO annotations associated with myoblast differentiation, migration, and fusion, muscle regeneration, and cell death processes was extracted and normalized using a z‐score. Heatmaps for those proteins were constructed using Morpheus (https://software.broadinstitute.org/morpheus).

4.11. Data analysis for N‐terminomics

Quantification of all peptides identified by N‐terminomics was performed in the software ProteinDiscoverer (v2.4) in a non‐nested design using the LFQ processing workflow (PWF_Precursor Quant_and_LFQ_IT HCD_SequestHT_Percolator.pd) and consensus workflow (CWF_Comprehensive_Enhanced Annotation_LFQ_and_precursor_Quan.pd) with some modifications (the report included only normalized, non‐scaled quantification data). In those cases where multiple peptides match the same cleavage site, only the abundance of the peptide with highest PSMs was considered as representative peptide for that proteolytic event. We used the same parameters described in the proteomic searches for: proteome database, false discovery rate, digestion enzyme, and modifications. Additionally, aminobutyric acid (Abu) at the peptide N‐termini was set as a variable modification, and trypsin specificity was changed to non‐specific at the C‐term. Precursor mass tolerance was set at 15 ppm with a fragment mass tolerance of 0.8 Da. An unpaired t‐test was applied for all the pairwise comparisons and the p‐values and fold changes were used to generate the volcano plots (on Prism v9) highlighting the Abu‐labeled peptides with the specificity of interest (P1 = D). Peptide abundance per identification was scaled to the maximum abundance detected in all populations (a value of 1 was assigned to the maximum), and Morpheus was employed again to generate the heatmaps for cleavage events and the specific GO annotations of interest.

Additionally, the sequence specificity P5–P1 for all unique cleavage identifications with P1 = D was matched to the sequence specificity of all caspases as an alternative to visualize specific caspase contribution to the caspase signature observed on each population. Caspases were grouped according to their similarity in recognition sequence as follows: casp‐1&4&5 (P4 = W/L, P3 = E, P2 = H, P1 = D), casp‐2 (P5 = V, P4 = D, P3 = E, P2 = V, P1 = D), casp‐3&7 (P4 = D, P3 = E, P2 = V, P1 = D), casp‐6 (P4 = V, P3 = E, P2 = H, P1 = D), casp‐8&9&10 (P4 = L, P3 = E, P2 = T/H, P1 = D), casp‐14 (P4 = W, P3 = E, P2 = S/T, P1 = D) (Duclos et al., 2017). A reliability score (Rawlings, 2016; Rawlings et al., ²⁰¹⁴) was obtained for each alignment by counting the matches at each position between each cleavage site and the caspase specificity, and dividing the total number of the matches at every position by the number of comparisons (substrates considered) times the number of residues considered (i.e., two residues considered for casp‐14 at P2) and multiplying by 100. Reliability score distributions were graphed on Prism (v9).

For the day 0 versus day 1 comparison (MB vs. L cells vs. D cells), a triplot of all cleavage sites was generated using Excel by initially graphing the log₂(ratio) of each comparison on each side of a triangle. The intersection of lines between the specific fold change value and the vertex at the opposite end of the triangle corresponds to the proteomic correlation between the populations compared. The location of each circle on the graph indicates the association of that cleavage event to the cell population. Identifications closer to the vertex, correspond to labeled peptides with higher association to that group, while identifications closer to the center of the triangle indicate similar levels of that peptide in all comparisons. Abu‐labeled peptides with caspase specificity (P1 = D) were highlighted in yellow. From this subset of peptides, cleavages with significant abundance difference (p‐value <0.05) in any of the pairwise comparisons in the triplot are highlighted with a red outline.

For the analysis of abundance changes in cleavage event during the time course of differentiation (day0 MB, day1 L cells, and day 4 MT), the cleavages with P1 = D were hierarchically clustered on Morpheus (Linkage: One minus Pearson correlation, method: complete). Peptides with P1 = D specificity with maximum abundance in the D cells were excluded from this analysis. The cleavage scaled abundance in the generated clusters was represented as trend graphs (Prism v9). Clusters were grouped as myoblast‐associated cleavages, early‐stage cleavages and late‐stage cleavage based on the average trend observed. GO analysis for these groups and all unique cleavage events identified was performed using metascape (https://metascape.org) (Zhou et al., 2019) against the mouse proteome for annotation. Enrichment on GO biological processes, GO molecular functions, and GO cellular components was done against myoblast and myotube proteomic identifications from previous studies (Chen et al., 2021; Marzan et al., ²⁰²³; Xiao et al., ²⁰²²) as background genes. Finally, sequence logos for cleavage events (with P1 = D) in forward and reverse N‐terminomics were generated with R (v4.3—ggplot2 and ggseqlogo packages).

All MS data can be accessed online through the MassIVE repository (proteomic comparison, MSV000094656; forward and reverse N‐terminomics, MSV000094657).

4.12. Validation of cleavage in HEK293 cells

pcDNA3.1 constructs for candidate proteins were purchased from Genescript with an N terminal FLAG and a C terminal HA tag. HEK 293 cells were reverse transfected with the constructs using polyethylenimine (PEI) (ratio 3:1 of PEI:DNA). The culture medium was changed after 24 h and the cells cultured for a further 48 h. Cells were then harvested and lysed (Lysis buffer: 50 mM HEPES, pH 7.4, 10 mM KCl, 5 mM EDTA, 0.1% Triton, 100 μM PMSF). Cell lysates were incubated with purified recombinant mouse caspase‐3 for 30 min and the cleavage of the proteins assessed by immunoblotting for FLAG and HA. Proteins were transferred to a nitrocellulose membrane (100 V, constant voltage for 1 h) before being blocked and probed with anti‐FLAG or anti‐His antibodies (1:1000 dilution). AlexaFluor secondary antibodies (1:20,000 dilution) were used to visualize proteins on a Licor.

AUTHOR CONTRIBUTIONS

Erik Gomez‐Cardona: Writing – review and editing; investigation; data curation; formal analysis; validation. Mahshid H. Dehkordi: Investigation; data curation; formal analysis; validation. Kolden Van Baar: Investigation; validation. Aiste Vitkauskaite: Investigation. Olivier Julien: Writing – review and editing; conceptualization; funding acquisition; supervision. Howard O. Fearnhead: Supervision; writing – review and editing; conceptualization; funding acquisition; formal analysis; validation.

FUNDING INFORMATION

Open access funding provided by IReL.

Supporting information

APPENDIX S1:

APPENDIX S2:

APPENDIX S3:

FIGURE S1: Proteomic comparison between different stages during myoblast differentiation. (a) Heatmap with the levels of significant proteins (p < 0.05) found by our LFQ approach in all cell populations. Each column represents a cell population and each row contains the protein abundance information (Z‐score) for all significant proteins. Blue represents a low protein abundance in that specific group, while yellow indicates a high abundance. Hierarchical clustering identified distinct protein abundance patterns associated with specific populations. (b) Gene ontology analysis for the main clusters identified. Biological processes (red), molecular functions (green), and cellular components (blue) with significant enrichment within a cluster are shown.

FIGURE S2. Quantification of caspase activity at day 4 of differentiation. (a) Caspase activity assessed in cell‐based assay. C2C12 cells were seeded at a density of 3 × 10⁴ cells/cm² in μ‐Plate 96 Well. After 24 h, the GM was replaced with DM supplemented with 5 μM IncuCyte® Caspase‐3/7 Green Apoptosis Assay Reagent (Cat. No. 4440). On day four of differentiation cells were fixed and stained with anti‐myosin heavy chain antibodies. The upper panel is the caspase‐dependent green fluorescence alone. The lower panel is the green and blue fluorescence overlaid. Myosin heavy chain is shown in red. (b) Comparing caspase activity in myotubes and reserve cells. On day 4 the fluorescent intensity of myotubes and reserve cells was quantified. Mean green fluorescent intensity within each cell nucleus was quantified using Harmony Software. (c) Caspase activity assessed in a biochemical assay. C2C12 cells were seeded at a density of 5 × 10⁴ cells/cm² in T25 flask. After 24 h, the GM was replaced with DM supplemented. Cells were harvested on day 0 (myoblasts), day 1 and day 4 of differentiation, lysed and the caspase activity in the lysate assessed using DEVD‐AFC. On day 1 cells were also separated into live and dead cell populations. The results are expressed as arbitrary fluorescent units/min/mg protein and are the mean ± standard deviation of three independent experiments.

FIGURE S3. Monitoring of labeling and capture efficiency during N‐terminomics experiments. Streptavidin blots of pre‐ and post‐labeled samples for detection of the N‐terminal modification with a biotin ester tag (TEVest6) using subtiligase (n = 2). The green signal corresponds to the successfully biotinylated proteins. On the capture panel, the dot blots show the pre‐ and post‐captured samples. Disappearance of the green signal indicates the capture of the biotinylated proteins on neutravidin beads. Upon successful capture (efficiency > 80%), samples were further processed following our N‐terminomics protocol. Results from forward degradomics experiments are shown in panel (a). The results for labeling and capture in the reverse degradomics with caspase‐3 are included in panel (b).

FIGURE S4. Cleavage site specificity and distribution of all cleavage sites identified by forward degradomics. (a) Sequence logos show the specificity surrounding the unique cleavage sites detected in each population. Sequence motifs observed in dead cells differ from the other populations, with an overrepresentation of D, and an underrepresentation of N at P1 position. (b) Upset plot shows the distribution and overlap of all cleavage site identification by forward degradomics. The dead cell population and myotubes show the highest number of unique cleavages, indicating a high proteolytic activity in those populations.

FIGURE S5. Abundance of putative caspase cleavage sites involved in cell differentiation and development, muscle function, and cell death. Heat maps contain the normalized abundance of the cleavage events with P1 = D in proteins associated with relevant pathways in our cell model. Cleavage events in the pathways shown were extracted from the GO annotation analysis. Hierarchical clustering groups the pattern in the abundance of a particular cleavage event with similar events in all populations. In all cases, there are subgroups of putative caspase cleavages with low abundance in the dead cells associated with relevant processes, functions, or components in cell differentiation or cell death.

FIGURE S6. Cathepsin B cleavage during differentiation. (a) Model of cathepsin B activation upon cleavage by caspase during differentiation. (b) Abundance of cleavage events detected on cathepsin B forward N‐terminomics. (c) Distribution of putative cathepsin B cleavage events in cell populations (P1 = R + P1′ = F).

FIGURE S7. Protein expression and purification of mouse caspase‐3. (a) Workflow for expression and purification of the active protease. (b) Chromatogram of the affinity purification of caspase‐3. (c) Eluted fraction was loaded evaluated by SDS‐PAGE to confirm purification, small and large subunits of mouse caspase‐3 are observed in the coomassie‐stained gel. (d) Fluorescence activity assay carried out using the optimal coumarin substrate Ac‐DEVD‐afc. The enzyme kinetics assay was performed using 5.0 nM caspase‐3, and 5.0–0.078 μM substrate. The k _cat/K _M was calculated using the linear region of the Michaelis–Menten curve k _cat/K _M = slope/[E]. (e) The protease activity was monitored in parallel with the reverse N‐terminomics, in optimal buffer and cell lysates with 0.25 μM caspase‐3 and 5.0 μM substrate.

FIGURE S8. Expression of candidate proteins by reverse transfection of HEK293. (a) pcDNA3.1 constructs for candidate proteins were purchased from Genescript with an N terminal FLAG and a C terminal HA tag. HEK 293 cells were reverse transfected with the constructs using Polyethylenimine (PEI) (ratio 3:1 of PEI:DNA). The culture medium was changed after 3 h and the cells cultured for a further 48 h. (b) Transfection was confirmed at 48 h by fluorescence microscopy for an eGFP positive control. Cells were then harvested and lysed (lysis buffer: 50 mM HEPES, pH 7.4, 10 mM KCl, 5 mM EDTA, 0.1% Triton, 100 μM PMSF). (c) Immunoblotting using either anti‐FLAG or anti‐HA antibodies. Overexpressed proteins were resolved by SDS‐PAGE and transferred to a nitrocellulose membrane (100 V, constant voltage for 1 h). Proteins were detected by immunoblotting using either an anti‐FLAG or an anti‐HA antibody (1:1000 dilution). Binding of a fluorescent secondary antibody (1:20,000 dilution) was visualized using a Licor.

FIGURE S9. Caspase‐3 cleavage of candidate proteins. Proteins with N‐terminal FLAG and C‐terminal HA tags were expressed in HEK293 cells. Cell lysates from HEK293 cells overexpressing candidate proteins were incubated for 30 min at 37°C with purified recombinant mouse caspase‐3 (200 nM). The proteins were resolved by SDS‐PAGE and detected by immunoblotting using either an anti‐FLAG or an anti‐HA antibody. Binding of a fluorescent secondary antibody was visualized using a Licor.

FIGURE S10. Common substrates shared between different non‐apoptotic contexts. Caspase substrates cleaved in differentiating C2C12 cells were compared to substrates identified in four other proteomic studies: Weghorst et al. (2020), Conde‐Rubio et al. (2021), Hertz et al. (2019) and Victor et al. (2018). All are studies of substrates generated by apoptotic caspases in non‐apoptotic contexts, including neurodevelopment, stress signaling in cancer cells, axonal degeneration and synaptic function. The studies used a range of technical approaches and identified different numbers of substrates. Substrates identified in three or more of the studies are shown. A large proportion are associated with the cytoskeleton, although a smaller number are associated with protein stability.

FIGURE S11. Unusual cleavage sites in non‐apoptotic contexts. Sequence logo for all cleavages with P1 = D from studies reporting caspase cleave sites in non‐apoptotic contexts. Logos for the different C2C12 populations are shown. The logo for “Not dead cells” (72 substrates) was generated from the combined set of myoblasts, myotubes, live and reserve substrates and excluded substrates found in dead cells. The set “Live but not dead” contained 46 substrates detected only in live cells (not myoblasts, myotubes, reserve or dead cells). The set “Dead and not dead” included 334 substrates shared between not dead (myoblasts, myotubes, reserve and live cells) and dead cells. The Weghorst data are substrates cleaved during chick neurodevelopment. The Hertz data are substrates cleaved during axonal degeneration. The three upper panels do not resemble a canonical caspase‐3 cleavage site (DEVD), most notably P4 ≠ D.

TABLE S1. Proteins previously reported to be cleaved by caspases during C2C12 differentiation.

Gomez‐Cardona E, Dehkordi MH, Van Baar K, Vitkauskaite A, Julien O, Fearnhead HO. An atlas of caspase cleavage events in differentiating muscle cells. Protein Science. 2024;33(9):e5156. 10.1002/pro.5156

DATA AVAILABILITY STATEMENT

Mass spectrometry files are available on MassIVE: MSV000094656 and MSV000094657.