Analysis of human acetylation stoichiometry defines mechanistic constraints on protein regulation

Abstract

Lysine acetylation is a reversible posttranslational modification that occurs at thousands of sites on human proteins. However, the stoichiometry of acetylation remains poorly characterized, and is important for understanding acetylation-dependent mechanisms of protein regulation. Here we provide accurate, validated measurements of acetylation stoichiometry at 6829 sites on 2535 proteins in human cervical cancer (HeLa) cells. Most acetylation occurs at very low stoichiometry (median 0.02%), whereas high stoichiometry acetylation (>1%) occurs on nuclear proteins involved in gene transcription and on acetyltransferases. Analysis of acetylation copy numbers show that histones harbor the majority of acetylated lysine residues in human cells. Class I deacetylases target a greater proportion of high stoichiometry acetylation compared to SIRT1 and HDAC6. The acetyltransferases CBP and p300 catalyze a majority (65%) of high stoichiometry acetylation. This resource dataset provides valuable information for evaluating the impact of individual acetylation sites on protein function and for building accurate mechanistic models.

Fig. 1

Measuring acetylation stoichiometry. a Diagram of the method used to measure acetylation (Ac) stoichiometry. b The degree of quantification error as determined by the fraction of SILAC ratios at each concentration of chemically acetylated peptides that was not consistent with SILAC ratios measured in at least one different concentration of chemically acetylated peptides. c The correlation between stoichiometry measured in independent experimental replicates. The number of peptides (n), Pearson’s correlation (r), and P-value (P) of correlation are shown. d Low absolute variability between experimental replicates. The histogram shows the distribution of Log2 ratios of stoichiometry in Experiment 1/Experiment 2 (Exp.1/Exp.2). e The correlation between stoichiometry measured using partial chemical acetylation (PCA) and absolute quantification (AQUA) peptide standards. f Low absolute variability between stoichiometry measurements made by PCA and AQUA. g Validation of stoichiometry measurements using recombinant acetylated (100%) proteins as a spike-in standard. Stoichiometry was measured at two different concentrations of spike-in protein (SILAC light, red) compared to SILAC heavy-labeled HeLa (blue) for each acetylation site. Source data are provided as a Source Data file

Fig. 2

Stoichiometry limits the detection of acetylated peptides. a Acetylation is biased to detection on abundant proteins. Protein copy number estimates are from. b The fraction of acetylated lysines detected on any given protein is correlated with protein abundance. The scatterplot shows the % lysines acetylated and copy numbers of 2488 acetylated proteins containing 6753 acetylation sites. The Pearson’s correlation (r), and P-value (P) of correlation are shown. c The number of peptides and acetylated peptides (Ac peptides) detected in deep proteome measurements from this study and. d The distribution of peptide copy numbers from a deep proteome measurement and acetylated peptide copy numbers calculated from the peptide stoichiometry and protein copy number. e The distribution of acetylated peptide copy numbers for acetylated peptides that were detected without prior antibody enrichment. Source data are provided as a Source Data file

Fig. 3

Properties of high stoichiometry acetylation. a The distribution of acetylation site stoichiometry for the 6829 sites measured in this study. b UniProt keyword enrichment for the indicated classes of high stoichiometry acetylation sites (>0.23% and >0.1%) and for doubly acetylated peptides. The size of the text is related to the fraction of sites associated with the keyword, and keywords that were more than two-fold enriched are colored red. c Subcellular compartment analysis based on the Human Protein Atlas. Category scatterplots show the distributions of acetylation site stoichiometry in each subcellular compartment. The number (n) of sites analyzed, median stoichiometry (median), and percentage of sites with >1% stoichiometry (>1%) is shown. d Amino acid sequence logos of the indicated classes of acetylation sites using IceLogo. Source data are provided as a Source Data file

Fig. 4

Stoichiometry of histone acetylation. a The diagram shows all histone acetylation sites whose stoichiometry was determined in this study. The sites are ordered by descending stoichiometry. Note that high stoichiometry sites occur on the N-termini of core histones. b The stoichiometry of histone acetylation sites as determined in four independent studies^–. c An anti-acetylated lysine immunoblot of HeLa whole cell lysate. Cells were boiled in 2% LDS to ensure histone extraction. Histones are annotated based on their expected molecular weight. d Histone acetylation sites constitute a majority of acetylated lysine residues in cells. Stoichiometry and protein copy numbers were used to calculate the number of acetylated lysine residues for the indicated classes of proteins. Source data are provided as a Source Data file

Fig. 5

Stoichiometry of deacetylase- and CBP/p300-regulated acetylation sites. a The category scatterplot shows the distributions of acetylation sites that are not regulated (NR), more than two-fold (>2×) upregulated, or more than four-fold (>4×) upregulated, by the indicated deacetylase inhibitors as determined by. Class I KDAC inhibitors primarily target HDACs 1, 2, 3, and 8, and were determined by the median SILAC ratio of apicidin, MS-275, valproic acid, and sodium butyrate-treated HeLa cells. Tubacin is an HDAC6 inhibitor and nicotinamide inhibits Sirtuin deacetylases, but the regulated sites are mostly attributed to SIRT1. b CBP/p300 regulates an increasing fraction of high stoichiometry acetylation sites. CBP/p300-regulated sites were determined by. c Acetylation sites that are most affected (>8× reduced) by loss of CBP/p300 activity have higher median stoichiometry than sites that are only modestly affected (2–4× reduced). Source data are provided as a Source Data file