Enrichment of Newly Synthesized Proteins following treatment of C2C12 Myotubes with Endotoxin and Interferon-γ

Abstract

Inflammation in muscle induces the synthesis of mediators that can impair protein synthesis and enhance proteolysis, and when sustained lead to muscle atrophy. Furthermore, muscle-derived mediators that are secreted may participate in disrupting the function of other peripheral organs. Selective identification of newly synthesized proteins can provide insight on biological processes that depend on the continued synthesis of specific proteins to maintain homeostasis as well as those proteins that are up- or down-regulated in response to inflammation. We used puromycin-associated nascent chain proteomics (PUNCH-P) to characterize new protein synthesis in C2C12 myotubes and changes resulting from their exposure to the inflammatory mediators lipopolysaccharide (LPS) and interferon (IFN)-γ for either a short (4 h) or prolonged (16 h) time period. We identified sequences of nascent polypeptide chains belonging to a total of 1523 proteins and report their detection from three independent samples of each condition at each time point. The identified nascent proteins correspond to approximately 15% of presently known proteins in C2C12 myotubes and are enriched in specific cellular components and pathways. A subset of these proteins was identified only in treated samples and has functional characteristics consistent with the synthesis of specific new proteins in response to LPS/IFNγ. Thus, the identification of proteins from their nascent polypeptide chains provides a resource to analyze the role of new synthesis of proteins in both protein homeostasis and in proteome responses to stimuli in C2C12 myotubes. Our results reveal a profile of actively translating proteins for specific cellular components and biological processes in normal C2C12 myotubes and a different enrichment of proteins in response to LPS/IFNγ. Collectively, our data disclose a highly interconnected network that integrates the regulation of cellular proteostasis, and reveal a diverse immune response to inflammation in muscle which may underlie the concomitantly observed atrophy and be important in inter-organ communication.

INTRODUCTION

Systemic inflammation and activation of the innate immune system, regardless of the etiology, can inhibit protein synthesis and increase protein degradation within skeletal muscle mediated, in large part, by the reduction in the kinase activity of the mammalian target of rapamycin complex 1 (mTORC1) [1–4]. Dysregulation in protein balance when prolonged leads to erosion of muscle mass which is debilitating, and increases morbidity and mortality [5]. This alteration in muscle proteostasis results from the enhanced secretion of cytokines and chemokines by the liver, spleen and other classical immune tissues. However, as muscle represents approximately 40% of whole body mass, this tissue also has potential to contribute significantly to the whole-body production of immune modulators (i.e., myokines) which may affect muscle function in an autocrine/paracrine manner and/or participate in muscle-organ crosstalk via a more classical endocrine mechanism [6, 7]. The synthesis and secretion of various immune modulators has been demonstrated in both skeletal muscle in vivo as well as in cultured myocytes in response to various pathogen-associated molecule patterns (PAMPs) which are recognized by the family of toll like receptors (TLRs) and other pattern recognition receptors [2, 6, 8, 9]. Lipopolysaccharides (LPS) derived from the cell membrane of gram-negative bacteria are the prototypical PAMP which binds to the TLR4 receptor. The skeletal muscle TLR4 axis is recognized as playing an important role in myogenesis [10], skeletal muscle metabolism [11, 12], and muscle protein balance and skeletal muscle atrophy [13–16]. Additionally, skeletal muscle TLR4 is upregulated in various pathophysiological states, including diabetes, obesity, aging and muscle disuse [17–19], suggesting their role in the underlying pathology of these conditions.

The relative importance of the direct muscle immune response to pathogens has been demonstrated in muscle-specific myeloid differentiation factor (Myd)-88 knockout mice which have a marked reduction in circulating proinflammatory cytokines and trafficking of immune cells to the site of infection [20]. Although muscle is a heterogeneous tissue containing resident immune cells and satellite cells, myocytes per se have been clearly demonstrated to be capable of recognizing and responding to pathogens. For example, C2C12 cells are immortalized mouse myoblasts that can fuse and differentiate into myotubes under appropriate growth conditions [21, 22] and such myotubes are commonly used as an investigative model of muscle plasticity and myogenesis [10, 23–25]. When incubated with a combination of LPS and interferon-gamma (IFNγ) to mimic the host inflammatory response, there is an enhanced synthesis and secretion of a number of immune modulators, such as nitric oxide (NO) and interleukin-6 (IL-6), that reproduce with high fidelity the response seen in intact muscle [16, 26–28]. In addition to its innate immune function, skeletal muscle cells have also been reported to have immunological properties consistent with non-professional antigen presenting cells [8].

Biochemical preparations of purified ribosomes contain a mixture of newly synthesized polypeptides that have yet to be released from ribosomes. These nascent protein chains can be released by a reaction that incorporates puromycin at their C-terminus [29], and subsequent protein sequence analysis can yield their identity. Such identification can be made on a proteomic scale and was used to identify >3000 ribosome-associated nascent polypeptide chains in cycling HeLa cells and >2000 proteins in mouse brain [30], with the use of biotin-linked puromycin producing a robust way to specifically isolate those puromycin-released nascent protein chains. This approach has been termed “PUNCH-P” (puromycin-associated nascent chain proteomics) and comparative analysis has shown that this approach captures the identity of currently synthesizing proteins on a scale comparable to those obtained using quantitative metabolic labeling of new protein synthesis or with ribosome-bound RNA analysis [30, 31].

The present study used PUNCH-P as a novel mass spectrometry-based unbiased proteomic approach to examine newly synthesizing proteins in C2C12 myotubes under basal conditions as well as after application of LPS/IFNγ, a treatment commonly employed to mimic inflammation in vivo. We hypothesized this discovery technique would uncover the participation of novel proteins and pathways that are involved in the adaptation of C2C12 myotubes to this inflammatory stimulus. Our results identify a clustering of proteins whose expression is up-regulated over basal conditions in C2C12 myotubes by their association with specific Gene Ontology (GO)-terms that include cellular responses to bacteria and virus exposure. Importantly, our data generate a point-in-time profile of the cellular translatome in muscle in response to an inflammatory stimulus and provide a foundation upon which to generate new hypotheses and pose new questions [32]. Thus, these data will provide a valuable resource upon which future research can study the mechanisms underlying muscle wasting and the potential impact of muscle inflammation on peripheral organ function in various pathological conditions.

MATERIALS AND METHODS

Materials

The C2C12 mouse myoblast cell line was purchased from the American Type Culture Collection (Manassas, VA). The following reagents were used: MEM (Mediatech, Herndon, VA); fetal bovine serum, (FBS) (Gemini Bio-Products, West Sacramento, CA); bovine calf serum (BCS) (Mediatech and Gemini Bio-Products); penicillin/streptomycin (Invitrogen, Life Technologies, Carlsbad, CA); Ultrapure Escherichia coli lipopolysaccharide (LPS 0111:B4) (Sigma Aldrich Corp, St Louis, MO and Invivogen, San Diego, CA); recombinant mouse IFNγ (Invitrogen and PeproTech, Rocky Hill, NJ). Biotin-dcpuromycin (Jena Bioscience, Jena, Germany); Streptavidin-agarose resin and iodoacetamide (Pierce, Thermo Scientific, Rockford, IL); complete EDTA-free protease inhibitor cocktail (Roche, Mannheim, Germany); pepstatin, leupeptin, DTT, sodium deoxycholate and Triton-X-100 (Sigma); RNasin® Ribonuclease inhibitor and sequencing grade modified trypsin (Promega, Madison, WI); streptavidin HRP rabbit antibody (DakoCytomation, Glostrup, Denmark); Tris, sucrose (RNA grade), ammonium bicarbonate, urea, SDS and TFA were of analytical grade and purchased from either Sigma or Fisher Thermo Scientific (Waltham, MA); and C18 stage tips (Thermo Electron North America LLC, West Palm Beach, FL).

Cell Culture

C2C12 cells that were of early passage (P3-P7) were maintained as myoblasts at sub-confluence in Minimal Essential Medium (MEM) containing 1 g/L glucose and supplemented with 10% FBS, penicillin (100 U/ml) and streptomycin (100 μg/ml). C2C12 cells tested negative for mycoplasma contamination. For experimental purposes, C2C12 myoblasts grown to 95–100% confluence were switched to MEM (1 g/L glucose) containing 10% BCS and penicillin/streptomycin for differentiation into myotubes. Differentiation medium was changed daily and cells were differentiated for at least 7 days before treatment to achieve cultures comprising of at least 90% myotubes as described previously [16, 26]. C2C12 myotubes were treated with ultra-pure E. coli LPS 0111:B4 (1 μg/ml) and recombinant mouse IFNγ (3 ng/ml) together in serum-containing medium for 4 h and for 16 h.

Ribosome isolation from C2C12 myotubes

Differentiated C2C12 myotubes were harvested and pooled for individual ribosome isolations and processed essentially following the procedure described for PUNCH-P [33]. Three independent ribosome preparations were isolated for each condition (untreated, or treatment with LPS/IFNγ for 4 h and 16 h), by the same researcher, and the time of isolation was similar for all preparations. All solutions were RNase-free or prepared using RNase-free water. Cells were washed once in ice-cold PBS (Invitrogen), pelleted by centrifugation at 1000 g_av (g force average) for 5 min at 4° C and used either immediately or stored frozen at −80° C as previously reported [33]. Thawed cell pellets were resuspended in 500 μl of ice-cold polysome buffer (50 mM Tris-HCl pH 7.5, 10 mM MgCl₂ and 25 mM KCl) containing 1.25 mM DTT, complete EDTA-free protease inhibitor cocktail, 1.4 μg/ml pepstatin, 2 μg/ml leupeptin and 40 U/ml ribonuclease inhibitor. Cells were lysed for 20 min on ice upon addition of 1/10^th volume of lysis buffer (11% (w/v) sodium deoxycholate and 11% (v/v) Triton-X-100 in polysome buffer) to the cell suspension. Lysed cells were then centrifuged at 18,400 g_av for 10 min at 4°C. The supernatant was layered onto a 2 M sucrose cushion in polysome buffer at an approximate ratio of 2:1 (e.g. 800 μl supernatant: 400 μl sucrose cushion). After centrifugation at 49,065 g_av for 160 min at 4° C, the ribosome pellet was washed gently with 500 μl of ice-cold RNase-free water to remove sucrose residue and resuspended in 100 μl of polysome buffer for subsequent quantitation and puromycylation reaction.

Labeling and affinity purification of puromycylated polypeptides for tryptic digest

An aliquot of the re-suspended ribosome preparation was diluted in water and scanned in a UV spectrophotometer blanked with water and containing the corresponding volume of polysome buffer. The absorbance/concentration of ribosomes was determined by the rRNA absorbance at 254 nm corrected for any background absorbance at 320 nm and estimated using a molar ε for ribosomes of 5 × 10⁷ M⁻¹ cm⁻¹ [34]. A typical preparation of ribosomes from up to 10 × 100 mm plates of C2C12 myotubes generated sufficient material for an individual sample. Puromycylation reactions were performed in a volume of 100 μl containing 0.75–1 μM ribosomes and 10 μM biotin-puromycin, and were incubated for up to 60 min at 37° C. Under some reaction conditions, biotin-puromycin was substituted with unconjugated puromycin or did not contain any form of puromycin (Fig. 1).

Fig 1.

Time course of biotin-puromycin labeling of newly synthesized proteins. Ribosomes were isolated from untreated (Control) C2C12 myotubes or after treatment for 16 h with the combination of LPS (1 μg/ml) and IFNγ (3 ng/ml). Nascent polypeptides were labeled with 10 μM Biotin-puromycin (Biot-PU) where indicated, for times up to 60 min at 37°C and resolved by SDS-PAGE. Labeled proteins were detected by Western blot using streptavidin-HRP and a representative image is shown.

Following puromycylation, 100 μl of washed streptavidin-agarose beads were added to each sample and the volume was adjusted to a final volume of 1 ml with urea/SDS buffer (50 mM Tris-HCl pH 7.5, 8 M urea, 2% (w/v) SDS and 150 mM NaCl). The bead containing mixture was rotated overnight at room temperature to allow binding of biotinylated peptides and was followed by a series of five wash steps using urea/SDS buffer, followed by a single high salt solution (1 M NaCl) wash for 30 min before the final 5 washes with ultrapure water as previously outlined [33]. For the reduction and alkylation steps, the beads were pelleted by centrifugation at 1000 g_av for 1 min, re-suspended in 1 ml of 1 mM DTT and mixed by rotation for 30 min at room temperature. After the last wash, the pelleted beads were re-suspended in 1 ml of 50 mM iodoacetamide and tumbled in the dark for 30 min at room temperature. The beads were transferred to a fresh tube following two wash steps with 50 mM ammonium bicarbonate and re-suspended in 100 μl of 50 mM ammonium bicarbonate containing 0.2 μg of sequencing grade modified trypsin. The beads were incubated at 42°C for ~ 3 hours and adjusted to 37°C for an overnight incubation to allow digestion of proteins on the beads. Trypsin activity was stopped after 16 h by addition of TFA to 0.1% (v/v). After centrifugation at 1000 g_av for 1 min, the supernatant was transferred to a fresh tube. The beads were re-suspended with an additional 100 μl of 50 mM ammonium bicarbonate, mixed and centrifuged as before to collect residual peptides in the supernatant and combined in the fresh tube for concentration by freeze-drying. Samples were lyophilized under vacuum to ~40 μl followed by two consecutive steps of adding 200 μl of ultrapure water and lyophilized under vacuum to a final volume of 40 μl for further processing on a C18-stage tip prior to LC-MS/MS.

Tandem LC-mass spectrometry

Samples were submitted for mass spectrometric analysis to the Mass Spectrometry and Proteomics Core Facility at the Penn State College of Medicine and run shortly thereafter so any differences produced by time-wise storage of samples was minimized. We were careful to control lab protocols for confounding variables related to sample preparation, processing and analysis by mass spectrometry. One of the authors (CSC) ran all of the cell studies and submitted samples to the Mass Spectrometry and Proteomics Core to be run by the same technician in a blinded manner without knowledge of treatment group. Outcomes assessment and data analysis was known to Dr. Coleman at all times, but not to the other authors, until the data interpretation phase of analysis. Tryptic peptides were purified, concentrated and eluted on C18-Stage tips, binding peptides in 5% formic acid (FA) and eluting with 20 μl of 5% FA in 80% acetonitrile. The eluted peptides were vacuum-centrifuged to dryness and re-suspended in 6 μl volume of 0.1% FA, of which 4–6 μl volume was injected onto an Eksigent NanoLC-Ultra 2D Plus/CHiPLC Nanoflex and separated through a 30 cm column length (two tandemly-arranged 75 μm × 15 cm ChromXP C18-CL 3 μm 120 Å columns) coupled to an ABSciex 5600+ Triple TOF analyzer. Peptides from each sample were separated by a 132 minute gradient (150 min total injection to injection cycle time) at a flow rate of 300 nl/min, with Solvent A (0.1% FA, 5% DMSO in water) and Solvent B (0.1% FA, 5% DMSO in acetonitrile). The gradient was a linear change from 2% B at t=0 to 35% B at t=121min; to 85% B at 122 min; isocratic at 85% B from t=122 to t=132 min; to 2% B at t=133 min, then re-equilibration at 2% B between t=133 min to t=150 min before the subsequent injection. The eluent was sprayed directly into the mass spectrometer operating in positive ion mode, with settings of curtain gas = 25, Gas1=8, and Gas2=0. Nanospray interface temperature (IHT) = 150 and the Ion Spray Voltage Floating (ISVF) = 2500. Rolling collision energy was used for collision-induced fragmentation of parent ions for generation of MS/MS spectra, with the center of the collision energy range for each peptide automatically calculated based on the peptide mass (invoked using default instrument setting of Collision Energy (CE) = 0). Collision energy spread (CES) was set to 3 and declustering potential (DP) to 100. After a 250 ms parent scan, a maximum of 50 candidates were chosen for MS/MS at 50 ms each per cycle, with a total cycle time of 2.804 seconds.

Bioinformatics

MS and MS/MS data collected as described above were used to generate peak lists and Protein IDs with the Paragon algorithm [35] utilized in ProteinPilot^™ version 5.01 software (Build 4688 with Paragon^™ Algorithm build 4874) and analyzed against a mouse-specific NCBI RefSeq database (containing 25,132 protein entries), concatenated with 536 common lab contaminants and with a reverse decoy database. The decoy database consisted of the direct reversed sequence of each entry in the normal “forward” RefSeq database and the number of decoy database hits accumulated at decreasing Paragon/ProteinPilot Unused Scores was used to estimate the Global False Discovery Rate (FDR) as described originally in [36]. The ProteinPilot settings used were Trypsin digestion, iodoacetamide alkylation of cysteines, thorough searching, biological modifications and amino acid substitutions. The primary quality score for protein identification in the Paragon/ProteinPilot program is the Unused Score, the sum of all individual peptide scores not “used” (claimed) by higher ranking proteins.

Proteins were counted as confidently identified if the Global FDR estimate for all high Unused Score proteins including that protein’s Unused Score was no greater than 1%. The production of the minimal number of peptide and protein identifications accounting for all of the spectral evidence was accomplished with the ProGroup protein-grouping algorithm built-in to the Paragon/ProteinPilot programs, and the list of accepted identified proteins meets all MIAPE standards [37] for protein grouping. The output of the identified proteins is reported in Supplementary Tables S1–S4 for samples derived from untreated myotubes as well as samples treated with LPS/IFNγ for 4 h and 16 h. In cases where there was both shared and distinct spectral evidence for isoforms or homologues sharing some but not all peptide sequences, the highest scoring protein (the one with the best distinct spectral evidence) “claims” the Unused Scores for the shared peptides and is listed first, followed by one or more lower ranking proteins with an Unused Score corresponding to the distinct spectral evidence for peptides contained only in that lower-scoring isoform or homologue. Sometimes several isoforms of similar proteins cannot be distinguished based on the spectral evidence obtained, if all of the actually identified peptides are shared by all the isoforms. For these cases where all of the spectral evidence is for peptides shared by all of the different isoforms, they are all equivalently identified, and any one or all of them may be present in the sample, in spite of one of these isoforms being of necessity listed ahead of the others.

The list of identified proteins in Supplementary Table S1 was obtained by merging non-redundant proteins from Tables S2–S4. When exactly the same MS/MS spectral information could identify peptides from more than one protein but with significantly different Unused Score, only the highest scoring protein was assigned as the identification. In the case of equally scoring protein isoforms that are encoded by the same gene, the protein is identified without specific reference to its isoforms. If equally scoring proteins are proteins encoded by distinct genes, these proteins are grouped together and identified as a “cluster” (see Table 1 in Results). Included in Table S1 are UniProtKB identifiers and Gene names which were obtained by conversion of the original GI accession by batch submission to the Panther Classification System (http://pantherdb.org) [38] or the Database for Annotation, Visualization and Integrated Discovery (DAVID) Bioinformatics Resource (david.ncifcrf.gov) [39]. In a few instances where both of these listed sources failed, conversions were obtained individually by querying databases directly either with the protein name or gene identifiers.

Table 1.

Characteristics of datasets

	Total Protein IDs	Matched to Distinct Proteins	Matched to Multiple Proteins	Distinct Peptides	Average Sequence Coverage
Untreated	1,099	1,073	26 (79)a	9,795	17.20%
4 h	1,009	982	27 (84)	7,946	16.00%
16 h	1,187	1,156	31 (81)	11,077	17.50%
Non-redundant	1,523	1,481	42 (115)

^aTotal number of proteins within clusters are in parentheses. The number of total protein IDs is the sum of distinctly matched proteins and clusters. Total non-redundant protein IDs is derived from merging the identified proteins from all three growth conditions.

Gene Ontology (GO) mapping and clustering were performed by submitting the identified protein lists with UniProt identifiers to the GO Consortium website (http://geneontology.org) [40]. Enrichment analysis was performed within Cellular Component and Biological Process GO terms by connecting directly to the analysis tool used in the PANTHER Classification System, (http://pantherdb.org) (version 2015-4-13) [38]. The background frequency is the number of gene encoded proteins annotated to a given GO term in the entire Mus musculus background (22275 genes) with sample frequency being the number of protein encoded genes annotated to a given GO term from our experimental input list.

Western Blot analysis

Puromycylated-ribosome extracts from C2C12 myotubes were resolved by denaturing SDS-PAGE and transferred to PVDF membrane (Millipore) using a standard wet transfer protocol in 25 mM Tris, 192 mM glycine, 20% (v/v) methanol buffer. Membranes were blocked with 5% non-fat dry milk for 1 h, washed 3 times for 5 min with TBS containing 0.1% (v/v) Tween 20 (TBS-T) and incubated with streptavidin-HRP diluted 1:3000 in 5% milk for 1 h at room temperature. Following standard washing with TBS-T, the blots were incubated with ECL Western blot substrate (Pierce) and images were developed using the Protein Simple Fluorchem M digital imaging system (Santa Clara, CA), and analyzed using NIH Image 1.6 software. The following antibodies with catalog numbers were used for Western blot analysis of specific proteins from whole cell extracts harvested directly in SDS-sample buffer, normalized to an initial blot probed for β-tubulin, and 10 μl of SDS soluble extract was added per lane. The following antibodies were used: NOS2 rabbit polyclonal, 1:500 (sc-850), Irgm2 goat polyclonal, 1:500 (sc-11088), IIGP goat polyclonal, 1:5000 (sc-11090) were from Santa Cruz Biotechnology, Dallas TX; STAT1 rabbit monoclonal, 1:1000 (#9172), NLRP3 rabbit monoclonal, 1:1000 (#15101), Caspase-11 rat monoclonal, 1:2000 (#14340), β-tubulin rabbit monoclonal, 1:1000 (#2128) were from Cell Signaling Technology, Danvers, MA; GBP2 rabbit polyclonal, 1:1000 (11854-1-AP) and GBP5 rabbit polyclonal, 1:5000 (13220-1-AP) were from Proteintech, Rosemont, IL; and CX3CL1 goat polyclonal, 0.1 μg/ml (AF365), VCAM1 goat polyclonal, 1 μg/ml (AF643), TIMP1 goat polyclonal, 0.1 μg/ml (AF980) were from R&D Systems, Minneapolis, MN). For Western analysis, representative blots are from a single experiment (experimental 1) with 3 biological replicates; however, similar results were obtained on samples analyzed from other experiments. For Western blot quantitation, values are presented as means ± standard error of the mean (SEM). Samples size was n = 3 for all treatment groups and proteins. Data were analyzed using one-way analysis of variance with post hoc Student-Newman-Keuls test to determine significant differences among the 3 experimental groups. Differences were considered significant when P < 0.05.

RESULTS and DISCUSSION

General characteristics of the data sets

Previously published proteomic data on muscle has allowed for the identification of several thousand proteins [41], whereas, the current study using PUNCH-P was focused on elucidating those proteins being activity synthesized and profiling the cellular translatome in muscle under basal conditions and in response to inflammation. In this regard, we treated C2C12 myotubes with an inflammatory stimulus that combines a PAMP in the form of LPS and a host-derived factor in the form of exogenously supplied IFNγ. This combination was used as it induces a synergistic response in the context of activating both the STAT1 and NF-κB signaling pathways, optimally down-regulates mTOR kinase activity and protein synthesis in this system, and decreases myotube diameter without impacting cellular viability [16, 26]. As stimuli that inhibit mTOR activity are associated with cellular atrophy, the overall goal of the present study was to extend these findings to define a population of actively translated proteins in C2C12 myotubes, using a discovery-based novel proteomic approach, in the absence as well as after treatment with E. coli LPS and IFNγ. Furthermore, myotubes were exposed to the inflammatory stimuli for 4 h, at which time no change in myotube diameter is noted, or for 16 h, at which time myotubes have been shown to have a profound atrophic response [26]. The goal was to uncover novel muscle-cell proteins whose translation was selectively up-regulated under such conditions and at these time points.

We determined the sequence of ribosome-associated nascent polypeptide chains using the PUNCH-P method [30, 33]. Puromycin-released nascent polypeptide chains were obtained from three independent ribosome preparations for each condition, (untreated, or treated with LPS/IFNγ for either 4 h or 16 h) and analyzed separately to obtain peptide fragmentation spectra. The spectra from an individual sample were either used directly to produce peptide sequence information for protein identification or were pooled with spectra from the other two samples and used together in a single analysis as indicated. Protein identifications were carried out with the ProteinPilot^™ version 5.0 software.

Table 1 summarizes the results obtained by using the pooled-spectra approach and setting an FDR cutoff for protein identification at 1%. We identified nascent polypeptide chains belonging to 1099 proteins from untreated cells, and 1009 and 1187 proteins from cells treated with LPS/IFNγ for 4 or 16 hours respectively. By merging the identified proteins from all three conditions, a total of 1596 unique proteins were obtained. Of these proteins 1481 were unambiguously assigned by one or more unique peptide sequences. The remaining 115 proteins were grouped into 42 clusters where each protein within a cluster can be assigned with the same available peptide sequence(s). Taking each cluster as a single identification (ID), our analysis therefore yielded a total of 1523 non-redundant IDs. The identity of the distinct proteins and clusters are reported in Supplementary Table S1, and the proteins identified in the untreated and treated groups are reported separately in Supplementary Tables S2–4. The average sequence coverage calculated for proteins in each growth condition ranges between 16.0–17.5% (Table 1) and coverage is reported for each identified protein in Supplementary Tables S2–4. These values are generally lower than the 24–40% range obtained from proteome analyses [42, 43] which may result from the fact that nascent polypeptide chains are shorter in length than their mature protein counterparts.

The extent of overlap among the protein IDs from each growth condition (Table 1) is illustrated in the Venn diagram [44] shown in Fig 2A. Of the 1523 non-redundant IDs, 718 IDs are common to all groups. When examined individually, 87.6% of protein IDs from untreated cells were also present in either the 4 h or the 16 h group. Similarly, 86.7% and 83.2% of IDs in the 4 h and 16 h treatment groups, respectively, were present either in untreated cells or cells from the other treatment condition. The extensive overlap among the three groups is consistent with the expectation that LPS/IFNγ treatment of myotubes elicits expression changes in only a limited set of proteins. There is an increase in the population of non-overlapping proteins in the two treatment groups, rising from 12.4% for untreated cells to 13.3% and 16.8% for the 4 h and 16 h treated groups, respectively, suggesting an LPS/IFNγ-induced increased abundance of specific proteins.

Fig 2.

Distribution of identified proteins across datasets. A. Venn diagram showing the relationship among the 1099 identified proteins from untreated C2C12 myotubes (red) and the 1009 and 1187 identified proteins from the 4 h (green) and 16 h (blue) treated myotubes respectively. Numbers shown are proteins shared by all three groups, between groups as well as those unique to a group. B. The identified proteins for each growth condition were stratified according to their detection in one (1/3), two (2/3) or all three (3/3) individual samples. Sampling distribution in the histogram was expressed as a percentage of the total number of proteins identified within a growth condition. Data were obtained from a n = 3 for each treatment group.

When the three replicates from each growth condition were analyzed separately, we detected 1516 out of the 1523 protein IDs listed in Supplementary Table S1 by using an FDR cutoff of 5% to accommodate generally lower sequence coverage from a single sample. Thus, nearly all the identified proteins can be stratified according to their sampling, and this information is included for each protein ID reported in Supplementary Table S1. Fig 2B shows the number of protein IDs that were either seen in one, in two or in all three samples. Previous shotgun proteomic studies have implicated a relationship between sampling levels and relative abundance of a protein [45]. Thus, a distribution (Fig 2B) where all three sampling levels are well represented would suggest that our experimental work flow is able to detect beyond the most abundant nascent polypeptide chains. While factors other than abundance may affect the overall detection of different proteins by their nascent polypeptide chains (for example, differences in mass spec ionizability of different peptide sequences), abundance is likely a major contributor to changes in sampling level of the same protein under different growth conditions, and these changes are evident in a number of the identified proteins. Sampling level information is used for ranked listing of proteins in a later section dealing with GO enrichment analysis of treatment-induced changes.

Comparison with steady-state protein and mRNA abundance

As steady-state protein levels are best explained by their rate of translation [46] and the PUNCH-P method provides data on the translation of specific mRNAs [30, 31], we also compared our data to the comprehensive list of 9880 proteins and their relative abundance in C2C12 myotubes reported by Deshmukh et al. [41]. Of the 1523 protein IDs in Supplementary Table S1, 1473 (96.7%) of these are in this published list of proteins. Thus, the identified nascent polypeptide chains in our study represent approximately 14.9% of proteins that were identified by a full proteomic analysis of all existing proteins for this cell line. Fig 3A shows the rank-order abundance of proteins in Supplementary Table S1 using values reported by Deshmukh et al. [41] in comparison with all proteins reported in the same study. It is clear from the comparison that the identified nascent chains belong to proteins that vary widely in steady-state abundance, spanning a range of 5–6 orders in magnitude, and only those proteins ranked by Deshmukh et al [41] in the bottom 0.5% in abundance are absent in our identification. Thus, the captured nascent polypeptide chains belong to proteins that are complex in terms of steady-state abundance, spanning nearly the full spectrum of abundance in C2C12 myotubes. This complexity is largely maintained when proteins in Supplementary Table S1 were examined after grouping them according to their level of sampling. A representative plot of rank-order abundance at different sampling levels is shown in Figure 3B, using sampling level information derived with untreated cells. Each sampling level is populated with proteins that span widely in abundance. Qualitatively similar patterns were also obtained using results from the treated groups but are not specifically shown here. Taken together, the identified nascent polypeptide chains belong to approximately 15% of presently known proteins in C2C12 myotubes and across a broad spectrum of abundance. Nonetheless, at this degree of saturation, the detection of nascent polypeptide chains did skew toward proteins that rank higher in abundance as approximately 90% of the proteins in Supplementary Table S1 have ranked abundance in the top 50 percentile [41].

Fig 3.

Relative protein and mRNA abundance of proteins identified by nascent polypeptide chains. Relative protein abundance values and p-values of mRNA detection used in this comparison are from published data cited in the Results section. Only those identified proteins not in clusters were used. A. Ranked abundance of identified proteins (blue) versus all proteins in C2C12 myotubes (red). B. Ranked abundance of identified proteins grouped by their frequency of detection in the three individual samples from untreated myotubes. Proteins identified only in samples from treated myotubes are placed in the zero frequency group. Horizontal lines mark the average abundance value of each frequency group. C. Identified proteins were ranked by their Unused Score from high to low.

A number of studies have reported mRNA abundance in C2C12 myotubes, and deposits of these datasets are retrievable from the ArrayExpress database (EMBL-EBI; http://www.ebi.ac.uk) [47]. We queried four control datasets in E-GEOD-28840 (GSM714189-GSM714192) for mRNAs that encode proteins matching to the 1073 distinct nascent polypeptide chains in untreated C2C12 myotubes [48]. For the purpose of our analysis we used the p-value reported for each mRNA to segregate them into either high or low detectability, depending on whether the p-value is <0.05 or ≥0.05 respectively. For the 1073 proteins detected in the present study, the mRNAs corresponding to 1058 (98.6%) of these can be assigned with a p-value, with 8.0 to 9.5% of these corresponding mRNAs falling in the low detectability ranges in the four E-GEOD-28840 (GSM714189-GSM714192 datasets. Fig 3C shows a plot of one representative p-value dataset, using the calculated protein identification scores (Unused Score) from ProteinPilot version 5.01 to rank the identified proteins. As abundance is a major contributing factor to mRNA detectability, it is likely that the high p-values of a substantial fraction of mRNAs is due to low abundance. It is evident that more mRNAs of low detectability are associated with proteins of lower score, and this would be consistent with a view that these mRNAs are lower in abundance. The analyses presented here on mRNA and protein abundance suggest the detection of nascent polypeptide chains by PUNCH-P cannot be reliably predicted from existing data on protein and mRNA abundance. Thus, the list of proteins identified in Supplementary Table S1 provides novel information regarding protein expression in C2C12 myotubes.

General landscape of proteins in untreated C2C12 myotubes

To place the proteins identified in untreated cells within context of localization and function, we queried the GO Consortium Database to cluster the identified proteins to specific GO terms. Of the 1152 proteins under consideration, 1086 of these contain Cellular Component annotations in the GO database. Table 2 lists the number of proteins (sample frequency) in specific subcellular compartments. A comparison with the total number of proteins in the mouse genome that have been annotated to reside within these compartments (background frequency) provides an assessment of what percentage of these proteins were captured in our study as newly synthesizing proteins. This assessment is represented in Table 2 by the percentage of sample frequency over background frequency (%NSP, percent newly synthesizing proteins). Across all Subcellular Compartments, the %NSP value ranged from 6.8 to 13.5%, providing a general range of total encoded proteins in the mouse genome that can be captured as nascent polypeptides under our experimental conditions. In contrast to Subcellular Compartments, members of Contractile Fiber Components, a cellular component expected to be enriched in muscle, exhibited %NSP values that ranged from 27.8 to 100%, indicating active synthesis of a much higher percentage of these proteins. Likewise, proteins in Contractile Fiber Components exhibit higher enrichment factors (EF), as calculated from the ratio of sample frequency over that expected for a sample size of 1086 proteins and a mouse genome that encodes 22275 proteins (Table 2). Taken together, the % NSP and EF values found with the Contractile Fiber Components provide a range that may be representative of proteins that require active synthesis in cultured C2C12 myotubes. Table 2 also shows proteins in the general Cell Junction GO term display %NSP and EF values similar to those found for Subcellular Compartments. Nonetheless, a subgroup of these proteins that form the more specialized adherens and anchoring junctions have %NSP and EF values similar to those of proteins in the Contractile Fiber GO terms, suggesting C2C12 cells utilize actively translated proteins for these specific junctions.

Table 2.

Clustering of proteins to Subcellular Compartments, Contractile Fiber, and Cell Junction Component Go Terms

	Sample frequency	Background frequency	Expected frequency	% NSP	Enrichment factor
Subcellular Compartment
nucleus (GO:0005634)	519	5971	291	8.7	1.8
plasma membrane (GO:0005886)	274	4009	195	6.8	1.4
mitochondrion (GO:0005739)	214	1668	81	12.8	2.6
endoplasmic reticulum (GO:0005783)	162	1381	67	11.7	2.4
cytosol (GO:0005829)	194	1435	70	13.5	2.8
Golgi apparatus (GO:0005794)	103	1222	60	8.4	1.7
lysosome (GO:0005764)	44	407	20	10.8	2.2
peroxisome (GO:0005777)	13	139	7	9.4	1.9
Contractile Fiber Components
contractile fiber (GO:0043292)	66	193	9	34.2	7.0
myofibril (GO:0030016)	63	179	9	35.2	7.2
contractile fiber part (GO:0044449)	58	169	8	34.3	7.0
sarcomere (GO:0030017)	54	156	8	34.6	7.1
I band (GO:0031674)	34	112	5	30.4	6.2
Z disc (GO:0030018)	27	97	5	27.8	5.7
myofilament (GO:0036379)	13	26	1	50.0	10.3
striated muscle thin filament (GO:0005865)	12	23	1	52.2	10.7
A band (GO:0031672)	12	27	1	44.4	9.1
M band (GO:0031430)	7	14	0.7	50.0	10.3
muscle myosin complex (GO:0005859)	4	9	0.4	44.4	9.1
troponin complex (GO:0005861)	4	8	0.4	50.0	10.3
muscle thin filament tropomyosin (GO:0005862)	2	2	0.1	100.0	20.5
striated muscle myosin thick filament (GO:0005863)	1	3	0.1	33.3	6.8
Cell Junction Components
fascia adherens (GO:0005916)	10	13	1	76.9	15.8
intercalated disc (GO:0014704)	21	52	3	40.4	8.3
cell-cell contact zone (GO:0044291)	22	62	3	35.5	7.3
cell-substrate adherens junction (GO:0005924)	109	350	17	31.1	6.4
focal adhesion (GO:0005925)	107	345	17	31.0	6.4
cell-substrate junction (GO:0030055)	109	355	17	30.7	6.3
adherens junction (GO:0005912)	120	407	20	29.5	6.0
anchoring junction (GO:0070161)	124	422	21	29.4	6.0
cell-cell adherens junction (GO:0005913)	16	55	3	29.1	6.0
cell junction (GO:0030054)	158	1105	54	14.3	2.9
cell-cell junction (GO:0005911)	50	376	18	13.3	2.7

Identified proteins from untreated myotubes were analyzed for enrichment in Cellular Component GO terms in the GO Consortium Database (version 2015-4-13). Percent of newly synthesizing proteins (%NSP) is calculated from taking the ratio of sample over background frequency, and enrichment factor from the ratio of sample over the expected frequency.

A number or our identified proteins also clustered to specific Biological Process GO terms with high %NSP values, some of which are listed in Table 3. In contrast to the %NSP value of 8.3% for the more encompassing general Cell Differentiation and Development GO terms, proteins in muscle-cell specific GO terms display higher %NSP values that range from 20–33%. Thus, similar to findings with Cellular Components, a higher percentage of proteins in muscle-specific biological processes are captured as actively translated proteins.

Table 3.

Enrichment of Specific Proteins to Biological Process GO Terms

	Sample frequency	Background frequency	Expected frequency	% NSP	Enrichment factor
Muscle-Specific Processes
Cell Differentiation (GO:0030154)	247	2964	145.0	8.3	1.7
muscle cell differentiation (GO:0042692)	50	242	11.8	20.7	4.2
striated muscle cell differentiation (GO:0051146)	39	173	8.4	22.5	4.6
myotube differentiation (GO:0014902)	12	51	2.5	23.5	4.8
myoblast fusion (GO:0007520)	6	18	0.9	33.3	6.8
Cell Development (GO:0048468)	118	1420	69.0	8.3	1.7
muscle cell development (GO:0055001)	33	126	6.1	26.2	5.4
striated muscle cell development (GO:0055002)	29	111	5.4	26.1	5.4
muscle fiber development (GO:0048747)	14	47	2.3	29.8	6.1
myofibril assembly (GO:0030239)	15	45	2.2	33.3	6.8
System Process (GO:0003008)	88	2217	108.0	4.0	0.8
muscle system process (GO:0003012)	37	179	8.7	20.7	4.2
muscle contraction (GO:0006936)	33	141	6.9	23.4	4.8
striated muscle contraction (GO:0006941)	21	70	3.4	30.0	6.2
Protein Synthesis
translation (GO:0006412)	99	266	13.0	37.2	7.6
tRNA aminoacylation for protein translation (GO:0006418)	21	40	2.0	52.5	10.8
translational initiation (GO:0006413)	17	54	2.6	31.5	6.5
translational elongation (GO:0006414)	9	29	1.4	31.0	6.4
protein localization to endoplasmic reticulum (GO:0070972)	12	35	1.7	34.3	7.0
peptidyl-asparagine modification (GO:0018196)	5	14	0.7	35.7	7.3
protein folding (GO:0006457)	37	150	7.3	24.7	5.1
response to unfolded protein (GO:0006986)	19	72	3.5	26.4	5.4
Metabolic Process
tricarboxylic acid cycle (GO:0006099)	12	27	1.3	44.4	9.1
citrate metabolic process (GO:0006101)	12	31	1.5	38.7	7.9
glycolytic process (GO:0006096)	12	31	1.5	38.7	7.9
pyruvate metabolic process (GO:0006090)	18	49	2.4	36.7	7.5
aerobic respiration (GO:0009060)	13	41	2.0	31.7	6.5

% NSP and enrichment factor are as defined in the legend to Table 2

Our analysis also revealed high %NSP values for a group of GO terms involved in Protein Synthesis, Maturation and Folding. All 19 cytoplasmic aminoacyl-tRNA ligases, including both subunits of the heterodimeric enzyme for the activation of phenylalanine, were found as actively translated proteins in our study. Thus, it appears a high number of proteins associated with protein biosynthesis are actively translated and detected in our experimental system. Proteins in metabolic processes leading to energy production are also highly enriched (Table 3). These include proteins in glycolysis, the tricarboxylic acid (TCA) cycle and aerobic respiration, among others. All enzymes in glycolysis and nearly the full complement in the TCA cycle were detected as actively translated proteins.

Treatment with LPS and IFNγ.

A central aim of our study was to profile proteins in differentiated myotubes whose translation was selectively up-regulated in response to LPS/IFNγ. In this regard, of the 1523 protein IDs identified by their nascent polypeptide chains, 424 protein IDs were identified in one or both groups (4 h and 16 h) treated with LPS/IFNγ, but not in untreated myotubes. Together with proteins in the clusters (where two or more proteins were equally confidently identified by the peptide sequences observed, and thus no single unique protein ID could be assigned), these IDs, unique to the treatment groups, contain 455 proteins. A number of Biological Process GO terms are related to cellular responses to bacteria and virus exposure, and we hypothesized that LPS/IFNγ-induced proteins will cluster specifically to these GO terms. To examine this possibility, we used the statistical over-representation test [49] with our list of 455 proteins and ranked the enrichment of specific Biological Process GO terms according to their statistical significance. Table 4 summarizes the top ranked (p-values<0.05) Biological Process GO terms obtained with this list of proteins (Input-1). For reasons of clarity, only selected GO terms are shown but those omitted are hierarchal-related ancestor or child terms of those listed in the table. With the exception of gland morphogenesis, these processes are all part of the response to either external stimulus or to stress. Importantly, enrichment, as judged by a comparison of the enrichment factors, is confined to a subset of these general pathways, and as is evident from the specific pathway names, they are indeed those expected in response to treatment of cells with LPS/IFNγ. We initially restricted our analysis to proteins that had been annotated by experimental evidence, but because 201 of the 455 proteins in our list lack this annotation, we extended the analysis to include proteins annotated with the all evidence codes, as defined by Gene Ontology [49], which resulted in the inclusion of all but 8 of the 455 proteins into the analysis. The inclusion of these additional proteins increased the total number of annotations for the specific pathways in Table 4 from 88 to 132 proteins. Thus, 132 of the 455 proteins under consideration are either known or inferred to function in pathways that respond to exposure with LPS and IFNγ. As the nascent polypeptide chains of these proteins were not detected in untreated cells, it is reasonable to conclude that expression of these 132 proteins was induced by LPS/IFNγ. These proteins, together with specific annotations, are listed in Supplementary Table S5.

Table 4.

Clustering of LPS/IFNγ- “Induced” Proteins to Biological Process GO Terms

	Mus musculus	Input-1 (455)1	Input-2 (176)1	Input-3 (158)1	Input-4 (105)1	p-value (for Input-1)
Go Biological Process
Unclassified	14626	201 (0.68)2	73 (0.64)	66 (0.65)	43 (0.62)	0.00E+00
Biological process	7649	251 (1.6)	103 (1.7)	92 (1.7)	63 (1.7)	1.29E-16
Cellular process	5549	179 (1.6)	67 (1.5)	66 (1.7)	44 (1.7)	2.17E-08
Gland morphogenesis	107	14 (6.5)	4 (4.8)	4 (5.3)	4 (8.0)	3.24E-04
Response to External stimulus	765	44 (2.8)	24 (4)	21 (3.9)	17 (4.7)	4.90E-06
Response to cytokine	200	20 (4.9)	16 (10.2)	13 (9.2)	12 (12.8)	4.66E-05
Response to interferon-γ	30	12 (19.7)	11 (45.8)	8 (38.1)	8 (57.1)	1.28E-08
Response to bacterium	221	24 (5.4)	14 (8)	10 (6.4)	9 (8.7)	3.04E-07
Positive regulation of leukocyte mediated cytotoxicity	29	7 (11.9)	4 (17.4)	3 (14.3)	2 (14)	1.33E-02
Response to Stress	1023	58 (2.8)	31 (3.9)	28 (3.9)	22 (4.6)	1.46E-08
Defense response to other organism	170	19 (5.5)	11 (8.2)	10 (8.3)	8 (10)	1.91E-05
Defense response to bacterium	107	16 (7.4)	11 (13.1)	8 (10.5)	8 (16)	5.91E-06
Innate immune response	84	15 (8.8)	13 (19.7)	11 (18.3)	10 (25)	1.83E-06

¹All proteins found only in LPS/IFNγ-treated cells. A set of 455 proteins found only under treatment conditions (Input-1) was analyzed for enrichment of Biological Process GO terms (GO Consortium Database, version 2015-4-13). Based on increasing stringency criteria, a subset 176 proteins (Input-2) was selected based on the criterion that those proteins were identified in Protein Pilot v 5.0 by at least 2 distinct peptides. Increasing stringency criteria, another subset of 158 proteins (Input-3) was selected if present in at least 2 out of the 3 samples in one of the two treatment groups. Finally, Input-4 is all proteins in Input-1 that satisfy the criteria specific for Input-2 and Input-3.

²Number of proteins found in each GO term followed by the fold-enrichment in parenthesis.

Additionally, using an increasing level of stringency, we identified three subpopulations of proteins that were preferentially enriched for LPS/IFNγ-induced proteins (Table 4). A subpopulation of 176 proteins (Input-2) was obtained by inclusion of proteins that were identified by at least two peptide sequences, and another (Input-3) included only those proteins whose presence were detected in at least 2 of the 3 samples from cells incubated with LPS/IFNγ for either 4 h or 16 h. Input-4, which has the highest stringency level, consists of 105 proteins that satisfied both of the above threshold settings. The enrichment effect of setting a threshold on either sampling level or peptide coverage appears to be additive, and nearly half of the 105 proteins (46%) in Input-4 were annotated in the specific GO-terms. Thus, the threshold settings in Input-4 effectively isolated a subpopulation of proteins that is highly enriched with LPS/IFNγ-induced proteins, and for this reason, we have placed proteins in this subpopulation under a separate heading in Supplementary Table S5. While this would need experimental verification, the other induced Input-4 proteins selected on the basis of these threshold settings (those proteins not already annotated as LPS/IFNγ-induced proteins) could potentially be candidates for previously uncharacterized responses to LPS/IFNγ.

Bioinformatic analysis of the extensive data set generated precludes an in-depth discussion of all proteins found to be upregulated by LPS/IFNγ; however, several points deserve emphasis:

1) As anticipated, we observed an early and sustained increase in NF-κB, STAT1, and NOS2 proteins, all of which are central regulators of the inflammatory response. An increase in NF-κB activity and NOS2 protein has been previously reported in LPS/IFNγ-treated myotubes [26, 27], with the up-regulation of STAT1 and NOS2 being confirmed in the present study by Western blotting (Fig 4).

Fig 4.

Relative abundance of representative LPS/IFNγ-induced proteins in C2C12 myotubes under control conditions or after 4 hours or 16 hours of stimulation (n = 3 per condition). Samples from each time point were run in triplicate with protein names listed on the right side of the figure and molecular weights (kDa) to the left. Abbreviations have been defined in the text of the manuscript. Westerns blots were quantified and the means ± SEM indicated directly under the representative blot. Different letters above mean values (a vs b vs c) indicate the mean values are statistically different from each other (p < 0.5), whereas groups with the means having the same letter (i.e., b vs b or c vs c) are not statistically different.

2) A family of p65 immunity-related GTPases (IRGs) and p47 guanylate binding proteins (GBPs) were found to be up-regulated in response to the combination of LPS/IFNγ [50]. These proteins are recruited to pathogen-containing vacuoles in association with Galectin 3 and TRIMs leading to disruption of the vacuole membrane followed by rupture and lysis of the pathogen membrane giving rise to the release of microbial ligands such as LPS into the cytoplasm. During infection with Gram negative bacteria, LPS is detected in the host cell cytosol causing activation of the non-canonical inflammasome pathway triggering activation of caspase 11 and cleavage of downstream effectors such as Gasdermin D leading to pyroptosis. We identified a number of peptides implicated in this signaling pathway including Gbp2, Gbp2b, Gbp4 and Gbp5 found in at least 2 out of 3 samples at 4h and 16h, with Gbp10/Gbp6, Gbp7 and Gbp9 each appearing later in 1 out of 3 samples at 16h. Likewise, members of the IRG family, both membrane-bound (i.e., IRGM1, IRGM3/IGTP) and predominantly cytosolic (i.e., IIGP1 and TGTP1), were increased at both time points. We randomly selected several of these peptides (i.e., GBP2, GBP5, IRGM2 and IIGP) and confirmed an LPS/IFNγ-induced increase in protein abundance by Western blotting (Fig 4). The upregulation of these proteins is a novel finding that has not been previously reported in muscle.

3) Antigen peptide transporter-1 and -2, which mediate unidirectional translocation of peptide antigens from the cytosol to the endoplasmic reticulum and play a central role in the adaptive immune response [51], are increased at both time points.

4) Increases in ubiquitin-proteasome activity and other catabolic pathway are also causally related to inflammation-induced muscle atrophy [1] and analysis of our data revealed the up-regulation of several catabolic-associated proteins. For example, TRIM proteins belong to family of E3 ubiquitin ligases many of which are induced by IFNγ and play a role in the regulation of inflammatory signaling cascades including the NFκB pathway [52]. TRIM35 was detected in 2 out of 3 samples at both 4 h and 16 h. In addition, ubiquitin conjugating factor E4 B (Ube4b) is required for polyubiquitination of specific substrates which are then recognized by the 26S proteasome for degradation. In this regard, LPS/IFNγ increased UbE4b at the 4 h time point, and such an increase might be expected to lead to the downstream activation of NF-κB [53]. Tribbles homolog 3 (TRB3), an ER stress-associated protein, was also up-regulated in the current study, with increases being associated with a reduction in muscle protein synthesis and increase in muscle proteolysis [54]. Proteasome activators are critical elements in proteasome-dependent protein degradation [55] and one such activator (Psme4; proteasome activator complex subunit 4; PA200) was also increased at the 4 h time point. Other proteins capable of regulating proteolysis which were increased by LPS/IFNγ in the current study include proteasome subunit beta type-5 (Psmb5), 26S proteasome non-ATPase regulatory subunit 6 (Psmd6) and Cullin 4B.

5) Finally, a noteworthy new aspect of our study is that we detected a number of proteins (3%) that were not observed in the deep proteome study of Deshmukh et al [41]. Many of these are secreted proteins that may not be present in samples derived from cell extracts used in the deep proteome study. For example, the synthesis of TIMP1, a secreted inhibitor of metalloproteinases and regulator of extracellular matrix composition, was increased at both treatment time points. Moreover, CXCL1, CXCL5, CXCL10, and CX3CL1/Fractalkine, all chemokines serving to control the migration and residence of selected immune cells, were identified with high sampling levels from our treated cell samples. One of these chemokines (CX3CL1) was selected and the increase confirmed by Western blot analysis (Fig 4). The above mentioned chemokines were specifically present in all LPS/IFNγ-treated cell cultured 4 hours, but in none of the later 16 hour samples. Conversely, macrophage colony stimulating factor (Csf-1) was only increased at the later time point in response to LPS/IFNγ. Regulation of chemokine expression in muscle has been examined in response to a variety of stimuli. For example, previous studies have reported that LPS increases CXCL1 mRNA in skeletal muscle [56], that muscle contraction increases CXCL5 mRNA [57], and that LPS increases CXCL10 in smooth muscle [58]. Our current data are complementary to and extend these early reports by demonstrating a dynamic regulation of multiple chemokines at the protein level following LPS/IFNγ stimulation. Future studies will need to determine the importance of these individual chemokines on the inflammation-induced cachexia and their potential to function in inter-organ communication. However, care must be taken to not over-interpret our data as there is no way to predict a priori whether the observed results would be similar in muscle under in vivo conditions or whether the changes produced in C2C12 myotubes by LPS/IFNγ would be comparable if a different inflammatory stimulus was used.

Limitations

The main limitation of any shotgun proteomic method is that not all proteins have peptides that can be detected by mass spectrometry across the range of protein abundances found in cells or on ribosomes. In this regard, mass spectrometry does not have time to select and fragment every peptide mass eluting at each time in the LC-elution gradient. Even among those peptides that are present in sufficient abundance to be detected, the less abundant peptides (and therefore corresponding proteins) that are detected from the bottom 25–40% of “visible” peptides by abundance will not be the same with every run, due to minor changes in solvent composition or other separation/mass spec conditions. This does not imply that any protein identified in one sample is incorrect, but identities found in more than one sample are generally more abundant peptides, i.e., likely to be further away from being obscured by background noise and are more abundant – this is the “sampling frequency” referred to herein in our paper. There are no obvious sources of bias, but some imprecision comes from using that sampling frequency rather than absolute measures of peptide concentration like one would get for a much more limited number of peptides/proteins by using known standards (preferably spiked in heavy-isotope labeled peptide standards). Another limitation of this work is that PUNCH-P analysis does not provide a dynamic representation of LPS/IFNγ-induced changes in the protein concentration. We have tried to circumvent this limitation, at least in part, by sampling both a relatively early (4 h) and later (16 h) time point. Moreover, our approach does not address the subcellular localization of proteins or their post-translational modification (e.g. phosphorylation, ubiquitination). Future studies addressing these important parameters will permit a more robust interpretation of our data and to more fully understand the potential for biological significance. Finally, while we conscientiously tried to minimize potential analytical and data processing-related confounding variables, the sample size in each group was n = 3. However, despite the relatively small sample size it was possible to unequivocally identify 1523 nonredundant peptides using PUNCH-P, and this number would be expected to increase if the sample size was enlarged.

Significance and Summary

Previous studies have reported the immunological potential for skeletal muscle [2, 6, 20, 59] while others have described the proteomic profiling of this tissue and myocytes [41, 60, 61]. The results of the current study complement theses earlier reports and are novel in that they reveal a highly interconnected network that integrates the translational regulation of cellular proteostasis as it relates to the diverse immune response produced by inflammation in muscle. Moreover, bioinformatic analysis indicated a clustering of LPS/IFNγ-induced proteins into GO terms related to responses to external stimuli and the stress response. In particular, our data reveal the increased translation of a number of peptides in the chemokine secretory pathway which have the potential to modulate inter-organ communication between muscle and other peripheral tissues. We also observed the upregulation of a large number of proteins associated with the non-canonical inflammasome pathway which have not been previously reported in muscle and which may play an important role in LPS-mediated muscle dysfunction and secondary cytokine secretion in various inflammatory conditions. Finally, we identified the up-regulation of a number of proteins involved in various protein degradation pathways that may be causally related to the reduction in muscle size in response to LPS/IFNγ. Our data provide a resource for future studies to determine the physiological importance and potential clinical significance of individual proteins and protein networks identified herein, and offer possible targets for therapeutic strategies to treatment muscle wasting.

Supplementary Material

1772994_Sup_Tab_1

Supplementary Table S1 Compilation of all proteins identified in PUNCH-P analysis. A list of 1523 proteins identified by their nascent polypeptide chains was obtained by merging proteins from untreated, 4 h and 16 h treatments (Supplementary Tables S2, S3 and S4, respectively). References to specific protein isoforms have been removed from protein names and UniprotKB accession and gene symbol corresponding to the “canonical” sequence of proteins have been added to each entry. Proteins in clusters are assigned with multiple names with corresponding accession entries and gene symbols. The Total number of peptides refers to the number of peptides found for each identified protein while Distinct peptides (shown in parenthesis), refer to peptides that are not shared by other proteins or by proteins not in the same cluster. Sample frequency reports the number of replicate samples in which each protein was observed. Entries for peptides and sample frequency from Untreated, 4 h- and 16 h-treated myotube samples are separated by semicolons in the respective order. The 424 proteins and clusters found only in myotubes treated with LPS/IFNγ are listed first, otherwise, the proteins are listed in alphabetical order based on their name.

1772994_Sup_Tab_2

Supplementary Tables S2 Compilation of all proteins identified in PUNCH-P analysis in untreated C2C12 myotubes.

1772994_Sup_Tab_3

Supplementary Table S3 Compilation of all proteins identified in PUNCH-P analysis in C2C12 myotubes identified after LPS/IFNγ treatment for 4 h.

1772994_Sup_Tab_4

Supplementary Table S4 Compilation of all proteins identified in PUNCH-P analysis in C2C12 myotubes identified after LPS/IFNγ treatment for 16 h.

1772994_Sup_Tab_5

Supplementary Table S5 Table listing candidate proteins induced by LPS/IFNγ. A filtered list of 132 proteins was generated after clustering of the 424 proteins found only under LPS/IFNγ treatment conditions (Input-1 from Table 4) to the following Biological Process GO terms: response to interferon-gamma (GO: 0034341); response to stress (GO:0006950); defense response (GO:0006952); innate immune response (GO:0045087); defense response to bacterium (GO:0042742); adhesion of symbiont to host (GO:0044406); defense response to other organism (GO: 0098542); response to cytokine (GO:0034097); cellular response to interferon-beta (GO:0035458); regulation of leukocyte mediated cytotoxicity (GO:0001910); immune system process (GO: 0002376). The additional 57 proteins listed were identified by analysis using Input-4 from Table 4 as described in the Results.

ABBREVIATIONS

PUNCH-P

puromycin-associated nascent chain proteomics

LC-MS/MS

liquid chromatography-tandem mass spectrometry

FDR

false discovery rate

LPS

lipopolysaccharide

IFNγ

interferon-gamma

NO

nitric oxide

IL-6

interleukin-6

TFA

trifluoroacetic acid

FA

formic acid

TCA

tricarboxylic acid

HRP

horseradish peroxidase

ECL

enhanced chemiluminescence

GO

Gene Ontology

DAVID

Database for Annotation, Visualization and Integrated Discovery

CC

Cellular Component

BP

Biological Process

%NSP

percent newly synthesizing proteins

EF

enrichment factor

Csf1

macrophage colony-stimulating factor 1

Availability of Data and Materials

The MS proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomexchange.org) via the PRIDE partner repository with the dataset identifier <PXD003144>. The following files are available free of charge via the Internet at http://pubs.acs.org. The data used to support the finding of this study are available from the corresponding author upon request.