Specificity and Reproducibility
To assess the guinea pig proteome, homogenates from 3 guinea pig hearts were fractionated to enrich cellular components and analyzed according to the workflow depicted in . Specifically, low speed centrifugation yielded a myofilament and nucleus-enriched fraction (hereafter known as myofilament rich). A crude mitochondrial fraction was obtained by keeping the supernatant from low-speed centrifugation as well as an additional homogenization of the myofilaments to release trapped mitochondria, pooling the two supernatants and collecting the pellet following the 8,000×g centrifugation. The remaining supernatant was centrifuged at 100,000×g to remove residual insoluble material and yield a pure cytosolic fraction. Since few new proteins were found in the insoluble material, however, for the purposes of analysis the data from the two fractions is combined and designated the cytosol-rich fraction.
Following tryptic digest of the subcellular compartments, acetylated peptides were enriched from bulk peptides by immunoprecipitation with immobilized acetyl-lysine antibody as described previously 
. Subsequent LC-MS/MS analysis initially identified a total of 608 proteins. Of these, 252 (41%) contained acetylated peptides. This contrasted with a preliminary analysis of a crude mitochondrial fraction without acetyl-peptide enrichment, in which only 0.7% of all identified proteins contained acetylated peptides (data not shown). Subsequent manual curation, as detailed in methods, pared 12 putative acetyl proteins identified on the basis of single acetylated peptides from the list. After curation, we identified a total of 994 acetylation sites, from 1075 acetylated peptides that map to 240 proteins and protein clusters. Site localization probability within the acetylated peptides was assessed with Scaffold PTM’s implementation of the A-score 
. Acetylation at 905 out of 994 sites was ascertained with a probability >0.99; 15 fell between 0.9–0.99 and another 14 between 0.8–0.9. The remainder, 60/994 (6%), had probabilities below 0.8. Just over half of the lower scoring sites (34 of 994 total) came from single spectrum/peptide matches and are best considered with caution (Table S1
In concordance with the crude level of subcellular enrichment, there was substantial overlap of identified proteins and acetylation sites between the mitochondrial, myofilament and cytosol-rich fractions (). Of the three experimental cellular compartments, the cytosol-rich fractions, largely free of myofilament and mitochondrial contamination, yielded the most distinct proteome. The reproducibility of the acetylome was addressed by analyzing 3 hearts separately, and by analyzing each subcellular component with technical replicates. 149 acetyl-proteins were found in all 3 guinea pig hearts (62%; ). Fully 79% of acetyl-proteins were identified in 2 out of 3 hearts. Approximately 21% of proteins were identified in only one heart, though often by many peptides and spectra. A detailed delineation of the acetyl-peptide and protein distribution is provided in Table S1
(Panel 1, columns G to AO). Comparison of our heart dataset against previous proteomic assessments of global lysine acetylation in both liver 
and human cell lines 
, showed substantial overlap of genes encoding acetyl-proteins (). Shwer et al
. also conducted proteomic analysis of isolated liver mitochondria 
, with which our dataset shows considerable overlap ().
Functional Classification and Gene Set Enrichment of K-Acetylated Proteins
The acetylated proteins and the sites of acetylation are presented in tabulated form in . For clarity, proteins have been grouped firstly according to their primary cellular location and then loosely by the biological processes they perform. The number of sites and site location within each protein are also provided. Sites are numbered according to the guinea pig database. However, we have mapped the guinea pig sites to their corresponding lysine residues in humans (Uniprot/SwissProt numbering; Table S2
) of which 94% are conserved. shows that 142 of 240 (59%) can be mapped, through their gene ontology, to mitochondrial annotations. Other cellular compartments represented in include the cytoplasm, the nucleus, the sarcomere and cytoskeleton, and finally assorted membrane or membrane associated proteins including proteins of the sarcoplasmic reticulum.
(panels A & B) summarizes the distribution of acetylation sites given in . Nearly two thirds (64.3%) of all identified lysine acetylation sites were associated with mitochondrial proteins (). Cytoplasmic targets comprise about 13% of sites, whereas nuclear proteins accounted for a further 8%. Non-mitochondrial membrane associated proteins accounted for only a small fraction of total sites (5%). Notably, sarcomeric and cytoskeletal proteins account for just over 10% of lysine acetylation. shows the distribution of sites among major biological processes. Lipid metabolism and oxidative phosphorylation are heavily targeted by acetylation, harboring 18 and 17% of sites, respectively. Enzymes of the tri-carboxylic acid (TCA) cycle garner 12% of sites. As noted in , muscle contraction, as a process, is well represented with 10% of sites located within the contractile apparatus and cytoskeleton. Proteins involved in protein synthesis, metabolism and folding are prominent (8.5%), as are proteins associated with chromatin remodeling and transcription (7.7%). Finally, proteins associated with redox homeostasis or antioxidant defense, as well transporters and the enzymes of glycolysis together account for nearly 16% of acetylation sites. Proteins associated with canonical signaling pathways did not account for many sites (0.6%).
Functional Annotation and Enrichment analysis.
To assess the degree to which specific biological processes and molecular functions are particularly targeted by lysine acetylation, more detailed assessment of associated ontologies was performed using the Cytoscape plug-in called BINGO. Biological processes and molecular functions that are more prominent within the list than would be expected by chance are ranked according to their Benjamini & Hochberg-corrected p-values and in Table S3
; Biological processes that are overrepresented relative to genomic background include metabolism, Redox sensing and contractile related ontologies. Consistent with prior work 
, specifically targeted metabolic processes include glycolysis (GO ID: 6096), fatty acid oxidation (GO ID:19395) and mitochondrial oxidative phosphorylation (GO ID: 6119). Redox sensing ontologies include response to reactive oxygen species (GO ID: 302), oxygen and reactive oxygen species metabolic process (GO ID: 6800) among others. Finally, a unique aspect of the cardiac acetylome is the enrichment of sarcomere-associated ontologies within the dataset including muscle contraction (GO ID 6936), myofibril assembly (GO ID:30239), sarcomere organization (GO ID: 45214) as well as others. The genes that define these enriched ontologies are in Table S3
(Panel 1, column G).
depicts the statistically-enriched ontological landscape (p<0.01) of molecular functions represented in this dataset, which falls into 3 major segments: binding activity, catalytic activity and transport functions. Among the catalytic functions, transferase, oxidoreductase, and ATPase activities predominate; oxygen-radical responsive ontologies are also represented. Ligand-binding functions are distributed among protein, ion and metabolite binding. Finally transport functions encompass acyl-carriers but primarily center around ion transport, owing to the presence of proteins such as VDAC, SERCA and RyR.
Distribution of Acetylation Sites
Nearly 70% (165/240) of identified acetyl-proteins harbored more than one acetylation site. In , we present the top 40 most heavily acetylated proteins in our study, which account for 492 of 994 (49%) of identified sites. Again, mitochondrial processes including fatty acid oxidation, TCA cycle and oxidative phosphorylation figure prominently in this table. The histones, as expected, are also heavily acetylated. Notable, however, is the extent of acetylation on myosin heavy chain (49) sites and the thin filament regulatory protein, cardiac Troponin I (8 sites). Without further characterization, it is unclear whether these proteins constitute acetylation “hot-spots” of regulatory significance or whether they are simply among the more abundant proteins in the heart, on which one might expect to identify acetylation more often, even at low stoichiometry.
We also note that among the most heavily acetylated proteins (e.g. top 10), the distribution of assigned MS/MS spectra is often heavily weighted toward a much smaller subset of sites. For example, though myosin heavy chain acetylation was detected on 49 sites, 10 sites account for 72% of the site-counting spectra for that protein. Similarly, for Trifunctional Protein, alpha subunit (, 2nd
entry), 5 of the 25 acetylation sites account for 62% of the site-counting spectra. The relative intra-protein frequency of site detection among these multi-acetylated proteins is a cryptic metric. It may provide a crude first approximation of relative site occupancy. However, intrinsic peptide properties, including ionizability, fragmentation susceptibility, and affinity for the acetyl-lysine antibody, are potential confounding factors. Limitations notwithstanding, the site identification frequency on a heavily targeted protein may help prioritize the design of mutants and is, therefore, provided in Table S4
Mass Spectral Data is Corroborated by Anti-Acetyl-Lysine Immunoreactivity
Comparison of our dataset to other global-scale proteome studies revealed 53 acetyl proteins unique to the guinea pig cardiac dataset (Table S2
, panel 2). Notable within the list were myofilament proteins and calcium handling proteins. Specifically, myosin heavy chain and cardiac Troponin I were among the top 40 most heavily acetylated proteins. To assess the acetylation status of these proteins, myofilaments were prepared in the presence of lysine deacetylase inhibitors (). Myofilament preparations have a defined protein complement and characteristic appearance by SDS-PAGE owing to the abundant myosin heavy chain at 200 kDa and actin at 42 kDa. Acetylation status of the myofilaments was assessed by immunoblotting with polyclonal and monoclonal anti-acetylation antibodies. The two antibodies appeared to differ with respect to preferred binding targets by western blotting at low exposures. Longer exposure ultimately revealed similar acetyl-protein labeling profiles.
Immunodetection of Cardiac Lysine Acetylation.
Acetylation of prominent mitochondrial substrates was also confirmed by sucrose density gradient enrichment of the respiratory chain complexes. Partially purified Complex I (NADH dehydrogenase; panel 5B) and Complex V (ATP synthase; panel 5C) displayed immunoreactivity toward both monoclonal and polyclonal anti acetyl-lysine antibodies on multiple subunits. The immunoreactivity of myofilaments, complex I and complex V were all diminished by performing the blots in the presence of competing acetylated BSA (>2% w/v), which confirmed the specificity of the antibodies for acetylated lysine residues.