Time-Resolved Analysis Reveals Rapid Dynamics and Broad Scope of the CBP/p300 Acetylome

Summary

The acetyltransferases CBP and p300 are multifunctional transcriptional co-activators. Here we combined quantitative proteomics with CBP/p300-specific catalytic inhibitors, bromodomain inhibitor, and gene knockout to reveal a comprehensive map of regulated acetylation sites and their dynamic turnover rates. CBP/p300 acetylates thousands of sites, including signature histone sites, as well as a multitude of sites on signaling effectors and enhancer-associated transcriptional regulators. Time-resolved acetylome analyses identified a subset of CBP/p300-regulated sites with very rapid (<30 min) acetylation turnover, revealing a dynamic balance between acetylation and deacetylation. Quantification of acetylation, mRNA, and protein abundance after CBP/p300 inhibition reveals a kinetically competent network of gene expression that strictly depends on CBP/p300-catalyzed rapid acetylation. Collectively, our in-depth acetylome analyses reveal systems attributes of CBP/p300 targets, and the resource dataset provides a framework for investigating CBP/p300 functions, as well as for understanding the impact of small molecule inhibitors targeting its catalytic and bromodomain activities.

Graphical Abstract

Introduction

Lysine N-ε-acetylation is a reversible protein posttranslational modification (PTM) that regulates diverse cellular processes, and its dysregulation has been linked to cancer and other diseases (Verdin and Ott, 2015; Yang and Seto, 2008). Advancements in mass spectrometry (MS)-based proteomics have led to identification of thousands of acetylation sites in mammalian cells (Choudhary et al., 2009; Choudhary et al., 2014; Kim et al., 2006). However, a systematic understanding of the acetylome is hampered by two fundamental deficiencies: (1) for a vast majority of acetylation sites, the responsible lysine acetyltransferases (KATs) are not known, and (2) almost nothing is known about site-specific kinetics of acetylation. The need for establishing enzyme-substrate relationship is further accentuated by recent studies indicating that acetylation occurs by both enzyme-catalyzed and non-enzymatic mechanisms (Wagner and Hirschey, 2014; Weinert et al., 2014; Weinert et al., 2013). Therefore, distinguishing enzyme-regulated sites from a potentially large fraction of non-enzymatic acetylation, and obtaining a thorough understanding of site-specific acetylation kinetics, is urgently needed to obtain a functional understanding of the acetylome. A major bottleneck in this area has been the lack of specific KAT inhibitors that can be used to identify the in-vivo substrates of KATs, and to obtain time-resolved dynamics of the acetylome. Recently, we identified first-in-class catalytic inhibitors for the acetyltransferases CREBBP (CBP) and EP300 (p300) (Lasko et al., 2017; Michaelides et al., 2018), paving the way for a systematic analysis their functions.

CBP and p300 define a unique family of KATs that are often referred to as a single entity (CBP/p300) due to their extensive sequence homology and functional similarity (Ogryzko et al., 1996). CBP/p300 is implicated in a multitude of processes, prominently including the regulation of enhancer-dependent, cell-type-specific gene transcription. Dysregulation of CBP/p300 is associated with diverse medical conditions, for example, recurring somatic mutations in CBP and p300 are associated with human leukemia and germline mutations in these genes cause Rubinstein–Taybi syndrome (Bedford and Brindle, 2012).

CBP/p300 is a multifunctional protein; it contains a catalytic domain that functions as an acetyllysine ‘writer’ and a bromodomain (BRD) that functions as an acetyllysine ‘reader’ that binds to acetylated proteins. CBP/p300 is also reported to function as an E3 and E4 ligase for protein ubiquitylation (Shi et al., 2009). In addition to these functions, CBP/p300 serves as a scaffold for recruiting a large repertoire of proteins, including transcription factors and co-activators, as well as several other reported KATs, such as NCOA1, NCOA3, ATF2, and CLOCK (Drazic et al., 2016).

Traditionally, the identification of CBP/p300 targets has focused on individual proteins and relied on CBP/p300 overexpression, genetic knockdown and knockout, or in-vitro assays. However, given the intrinsic limitations of the aforementioned approaches, and the association of CBP/p300 with several other reported KATs, it is unclear whether all of the identified sites are directly acetylated by CBP/p300 in vivo. Because genetic depletion of CBP/p300 simultaneously eliminates both its catalytic and scaffolding functions, it was previously difficult to disentangle these functions in substrate acetylation and transcriptional regulation.

Here we combine quantitative proteomics with selective chemical inhibitors and genetic knockout cells to obtain a systems-wide understanding of the CBP/p300 acetylome, to investigate site-specific acetylation dynamics, and to unravel the functions of CBP/p300 in transcriptome and proteome regulation. Our data delineates the catalytic, bromodomain, and scaffolding functions of CBP/p300 in regulating protein acetylation, as well as in mRNA and protein expression.

Results

A High-Confidence Map of the CBP/p300 Acetylome

To investigate the CBP/p300 acetylome we used three different, complementary, perturbations; gene knockout (KO), highly specific CBP/p300 KAT inhibitors, and a bromodomain inhibitor (Figure 1A). CBP/p300 was conditionally knocked-out in Crebbp^fl/fl;ep300^fl/fl mouse embryonic fibroblasts (MEFs) (Kasper et al., 2010) expressing a tamoxifen-inducible Cre recombinase. CBP/p300 KAT activity was inhibited with the recently developed potent and selective small molecule inhibitors-Compound R (Cmpd-R) and A-485 (Figure S1A) (Lasko et al., 2017). These two inhibitors are structurally related and show similar inhibitory activities for CBP/p300 in-vitro and in cell-based assays (Figures S1B and S1C). The CBP/p300 bromodomain was selectively inhibited using the small molecule CBP112 (Picaud et al., 2015).

Figure 1. A high-confidence map of the CBP/p300 acetylome.

(A) Diagram of CBP/p300 domain structure and the interventions (KO, CBP112, Cmpd-R, and A-485) used in this study. (B) Number of quantified acetylation sites and their overlap in the acetylome measurements from the indicated conditions. (C) Distributions of acetylation (Ac) site ratios and summed acetylated peptide intensity (from wild-type or control-treated samples) for the indicated interventions. The number of quantified sites and percent of up- or down-regulated (>2-fold change) sites is indicated. For each intervention, the number of sites (n) indicates the sum of acetylation sites quantified from replicate experiments. (D–F) Correlation and overlap between KO, Cmpd-R and A-485-treated acetylomes. The number of sites analyzed (n), Pearson’s correlation (r), P-value (P), and percent of regulated sites are shown. The Venn diagrams show the overlap between downregulated sites, and the boxplots show the distributions of ratio measurements within each region of the Venn diagram. (G) The fraction (%) of sites that were not quantified (i.e. acetylated peptide detected only in control samples, but not in the KO or the catalytic inhibitor treated samples) in the indicated experiments. (H) Distributions of acetylation (Ac) site ratios for sites that were not quantified in KO cells, but were quantified in the acetylomes from cells treated with the indicated inhibitors.

See also Figure S1, Table S1, and Table S2.

We applied stable isotope labeling with amino acids in cell culture (SILAC)-based quantitative MS to identify CBP/p300-regulated acetylation sites. We quantified ~21,000 acetylation sites on ~5,300 proteins, and more than 11,500 sites on 3,500 proteins in each condition (Figure 1B, Figure S1D, Table S1, and Table S2). We additionally quantified proteome levels to calibrate acetylation measurements for differences in protein abundance (Tables S1 and S2). Acetylation changes were strongly correlated across biological replicates (Figure S1E), indicating excellent reproducibility. CBP/p300 KO and catalytic inhibition resulted in extensively downregulated acetylation that was highly correlated (Figures 1D–1F). Nearly all sites (93%) that were downregulated (>2-fold) by the catalytic inhibitors were also downregulated in KO cells (Figures 1D and 1E). However, a greater proportion (~40%) of sites was exclusively downregulated in KO cells. This is likely due to the greater magnitude of reduced acetylation in KO cells (Figures 1D and 1E, compare boxplots for overlapping sites), that is attributable to the different time periods of treatment (72hrs for KO, 16hrs for catalytic inhibitors). There was a greater overlap between sites regulated by Cmpd-R and A-485 (Figure 4F) as compared to KO (Figures 1D and 1E), and in all comparisons the non-overlapping sites were close to the 2-fold cutoff for regulation (Figures 1D–1F, compare boxplots for non-overlapping sites). These data indicate that KO, Cmpd-R, and A-485 caused reduced acetylation at the same acetylation sites, and that the non-overlap of regulated sites is mostly attributable to weakly regulated sites that fall either side of the cutoff used to classify regulated sites. CBP112 caused modestly reduced acetylation compared to KO, Cmpd-R, and A-485 (Figure 1C and Table S1), and most (74%) of the CBP112-regulated sites were also regulated in KO cells (Figure S1F). Together, these results comprehensively verify the in-vivo specificity of Cmpd-R and A-485, and show that loss of catalytic activity alone has similar impact on acetylation as loss of the entire CBP/p300 protein, indicating that loss of CBP/p300 catalytic activity is the primary cause of reduced acetylation observed in the KO cells.

Figure 4. Kinetic analysis of the CBP/p300 acetylome.

(A) Reduced acetylation over time in A-485-treated MEF cells, sites were ordered based on acetylation half-life. (B) The fraction of CBP/p300-regulated acetylation sites and the magnitude of regulation at the indicated time points after A-485 treatment. (C) Correlation of acetylation (Ac) changes in A-485-treated (16 hours) MEFs to the same cells treated with A-485 for different time periods. The Pearson’s correlation (r) is shown. (D) Acetylation changes over time in A-485-treated MEFs. Sites were clustered by a fuzzy c-means method into three categories, as indicated. (E) The distribution of acetylation half-lives for 619 sites with a median value of 93.6 minutes. (F) The distribution of turnover rates (half-life) for acetylation (Ac sites) and the corresponding proteins. Protein half-lives were determined by (McShane et al., 2016). (G) The distributions of relative acetylation stoichiometry as determined by abundance corrected intensity (ACI) for sites with long (t_1/2 >30 minutes (m)) and short (t_1/2 <30 minutes (m)) half-lives. Significance (P) was calculated by the Wilcoxen test. (H) Deacetylation half-life (t_1/2) for the indicated proteins and acetylation sites. Red dots indicate outlier data points that were excluded for calculating half-life, dashed lines indicate variance.

See also Figure S6.

Properties of the CBP/p300 Acetylome

For the purposes of bioinformatic analyses, we defined sites with >2-fold reduced acetylation as CBP/p300-regulated. Based on the fraction of two-fold downregulated sites in an unperturbed control experiment (0.73%) (Figure S2A), we estimated the false discovery rate (FDR) for CBP/p300-regulated sites in the KO, Cmpd-R, and A-485 experiments to be 3.1%, 6.0%, and 3.9%, respectively, which is comparable to a P-value of 0.05. A small fraction of sites in KO, Cmpd-R, and A-485 were not quantified (Figure 1G), because acetylation was only detected in control cells (i.e. wild-type MEFs, or control-treated cells), suggesting that loss of CBP/p300 activity may have reduced acetylation to a level that was not detectable. Notably, not quantified sites occurred less frequently in CBP112-treated cells (Figure 1G), and not quantified sites in KO cells showed markedly reduced acetylation by Cmpd-R and A-485, but not by CBP112 (Figure 1H). These not quantified sites are putative targets of CBP/p300, and are included in the supplementary data (Table S2); however, due to the stringent criteria applied, they were excluded from further analyses presented here.

CBP/p300 was strongly biased to acetylate nuclear proteins, whereas cytoplasmic and mitochondrial proteins were underrepresented (Figure 2A). Remarkably, 36%, 22%, and 30% of sites on proteins associated with the UniProt keyword “Nucleus” were regulated by KO, Cmpd-R, and A-485, respectively; indicating that up to one-third of the nuclear acetylome is regulated by CBP/p300. Consistent with its role as a transcriptional coactivator, CBP/p300-regulated sites were significantly enriched for keywords describing transcription and chromatin regulation (Figures 2B and S2B).

Figure 2. Properties of the CBP/p300 acetylome.

(A) The subcellular distribution of CBP/p300-regulated acetylation. The analysis was performed using proteins that associated with only one of the indicated UniProt Keywords, (P<1e⁻⁵⁰ for all comparisons, Fisher test). (B) UniProt Keyword enrichment analysis comparing CBP/p300-regulated and unregulated acetylation sites in KO cells (P<5e⁻²⁴, Fisher test). (C) CBP/p300-regulated sites (KO) are significantly more likely to occur in close proximity when compared to random sampling. Error bars indicate the 95^th percentile of the random sampling (*P≤0.05, permutation test). (D) Proteins with CBP/p300-regulated sites interact more frequently than unregulated proteins (**P≤0.001). Protein-protein interaction data were obtained from BioGRID 3.4 (Chatr-Aryamontri et al., 2017). (E) CBP/p300 regulates a majority of detected sites on targeted proteins. (F) Relative acetylation stoichiometry, as determined by abundance-corrected intensity (ACI), of CBP/p300-regulated sites. (P<2e⁻¹⁶ for all comparisons, Wilcoxen test). (G) CBP/p300 regulates a significantly greater proportion of high stoichiometry acetylation sites (top 5% ranked by ACI), as compared to low stoichiometry sites (bottom 95% by ACI) (P<5e⁻⁹⁰, Fisher test). (H) Correlation between in-vitro acetylation by recombinant p300 (n=2) and reduced acetylation in KO cells. The number of sites analyzed (n), Pearson’s correlation (r), P-value (P), and percent of regulated sites are shown. (I) Correlation between increased acetylation in human 293FT cells overexpressing p300 (n=2) and reduced acetylation in CBP/p300 KO MEF cells.

See also Figures S2, S3, and S4.

Analysis of amino acids flanking CBP/p300-regulated sites did not reveal a prominent motif, with the exception of a modest enrichment of Ala, Pro, Gly, and Ser at the −2 position (Figure S2C). However, CBP/p300-regulated sites were significantly biased to occur at lysine residues that were in close proximity (Figures 2C, S2D, and S2E), and proteins with CBP/p300-regulated sites had a significantly higher tendency to interact with each other (Figure 2D). Furthermore, among regulated proteins, two-thirds of the detected sites were acetylated by CBP/p300, and among highly regulated proteins (>4 regulated sites), 82% of sites were acetylated by CBP/p300 (Figure 2E), indicating that CBP/p300 acetylates a majority of sites on targeted proteins. A closer inspection of CBP/p300 targets showed complex acetylation patterns; some proteins, such as Bcl9, Ncoa2, and Kmt2d, were regulated at nearly every site, other proteins, such as Parp1, Ncl, and Thrap3, showed a mixture of regulated and unregulated sites (Figure S3A), and some proteins were regulated at a conserved position in homologous proteins (Figure S3B).

We examined the relationship between regulation by CBP/p300 and acetylation abundance at individual sites by using abundance-corrected intensity (ACI) to estimate relative acetylation stoichiometry (Weinert et al., 2014) (Table S2). Consistent with enzyme-catalyzed acetylation, CBP/p300-regulated sites had significantly (P<2e⁻¹⁶) higher ACI than unregulated sites, and the most robustly regulated sites (>8-fold reduced acetylation) had the highest ACI (Figure 2F). We ranked all acetylation sites by ACI and classified the top 5% as having ‘high stoichiometry’ and the bottom 95% as having ‘low stoichiometry’. Strikingly, CBP/p300 acetylated a majority of ‘high stoichiometry’ acetylation sites in all experiments, but a much smaller fraction of ‘low stoichiometry’ acetylation (Figure 2G), indicating that CBP/p300 is one of the most prominent nuclear KAT activities in cells.

The Cellular Context Contributes to CBP/p300 Specificity

Our data indicate that CBP and p300 are relatively promiscuous KATs that target neighboring sites on proximal proteins. To validate these findings, and to further explore the specificity of CBP/p300-catalyzed acetylation, we used a combination of in-vitro acetylation assays and overexpression in cells. We first tested the ability of recombinant p300 to acetylate six purified human proteins that were identified as CBP/p300 targets. These six proteins contained 20 acetylation sites that were regulated in KO and inhibitor-treated cells. We found that recombinant p300 acetylated these same sites in vitro, as acetylation was more than two-fold increased or only detected in p300-treated samples (Figure S3C). In addition, acetylation was also increased at nearly all other detected sites; more than half of all sites were only detected after acetylation by p300 (Figure S3D), and of the remaining sites, acetylation was more than two-fold increased at >78% of sites (Figure S3E). We used SILAC quantification to measure in-vitro acetylation of whole cell lysates by purified p300 (Table S2). p300 caused extensive acetylation (48% of sites were more than two-fold increased over control- p300 + A-485); however, increased acetylation was weakly correlated (r= −0.20) with reduced acetylation in the KO cells (Figure 2H). Indeed, half of the sites that showed increased acetylation by p300 in-vitro were unaffected in KO cells, suggesting a low specificity of acetylation under these conditions. In contrast, in increased acetylation caused by ectopic expression of p300 in human 293FT cells was much better correlated (r= −0.57) with reduced acetylation in KO cells (Figure 2I and Table S2). 80% of sites that showed reduced acetylation in KO cells were increased after transfection with p300, while 65% of the sites with increased acetylation were also reduced in KO cells, showing that overexpressed p300 targets many of the sites that were regulated in KO and catalytic-inhibitor-treated cells. These results show that p300 is a promiscuous KAT in vitro, and overexpressed p300 in cells shows greater substrate specificity as compared to recombinant p300 in in vitro assays, indicating that the cellular environment contributes to CBP/p300 substrate specificity.

The CBP/p300 Acetylome is Conserved

To further confirm the impact of CBP/p300 catalytic inhibition on the acetylome in a different cell type, and to assess whether CBP/p300 targets the same proteins in human cells, we quantified acetylation in Cmpd-R-treated Kasumi-1 cells. Kasumi-1 is a human acute myeloid leukemia cell line whose growth depends on CBP/p300-dependent acetylation of the oncoprotein AML1-ETO (Wang et al., 2011). Similar to MEFs, Cmpd-R-treated Kasumi-1 showed widespread (20% of all sites, 34% of nuclear sites) reduced acetylation (Figure S4A, Table S2). Regulated acetylation sites were significantly biased to occur on nuclear proteins, and proteins with regulated sites were significantly enriched for UniProt keywords describing transcription and chromatin regulation (Figure S4B and S4C). Notably, the acetylomes of Cmpd-R-treated MEF and Kasumi-1 cells were well-correlated (r= 0.77) (Figure S4D), indicating that CBP/p300 acetylated conserved sites in mouse and human cells.

Acetylation of Diverse Signaling Effectors

Acetylation by CBP/p300 critically regulates the transcriptional activity of non-histone proteins, such as transcription factors (Spange et al., 2009). We found that CBP/p300 acetylates over two hundred transcription factors, chromatin remodelers, and transcriptional co-activators (Figure S5). In particular, CBP/p300 acetylated nearly all members of the bHLH-PAS family of transcriptional regulators quantified in our dataset, including Ncoa1, Ncoa2, Ncoa3, Arnt/Arnt2, Arntl, and Bmal1 (Table S2). Additional CBP/p300 targets include many signature transcriptional effectors in signaling pathways related to development and differentiation, including Wnt-beta catenin, hippo, hedgehog, notch, and TGF-beta signaling (Figure 3A). CBP/p300 also acetylated key proteins in pathways regulating nutrient and energy metabolism, such as AMPK, PKA, and calcineurin signaling, consistent with a role of CBP/P300 in regulating energy homeostasis (Bedford and Brindle, 2012). Taken together, these results show extensive CBP/p300-dependent acetylation of non-histone proteins and place CBP/p300 at the nexus of diverse signaling pathways.

Figure 3. Acetylation of diverse signaling effectors and enhancer-associated regulators.

(A) Key signaling effectors of the indicated signaling pathways that have reduced acetylation in KO, Cmpd-R, or A-485-treated cells. (B) CBP/p300-regulated acetylation on proteins present in the experimentally defined Wnt enhanceosome (van Tienen et al., 2017). (C) Proteins and protein complexes associated with enhancer-driven gene expression are highly acetylated by CBP/p300. The data show maximum downregulated acetylation site ratio from KO, Cmpd-R, or A-485-treated cells.

See also Figure S5.

Acetylation of Enhancer-Associated Regulators

CBP/p300 is implicated in enhancer-regulated gene transcription, and the CBP/p300-catalyzed H3K27ac mark is used to define active enhancers (Creyghton et al., 2010), but the extent of CBP/p300-catalyzed acetylation at enhancers is not fully known. Comparison of our acetylome data with an experimentally determined BCL9L-Wnt enhanceosome (van Tienen et al., 2017) showed that nearly every detected component of the BCL9L-Wnt enhanceosome is acetylated by CBP/p300 (Figure 3B); a remarkable finding considering that the BCL9L-Wnt enhanceosome was purified from a different cell type and organism than used here. CBP/p300 acetylates more than three dozen enhancer-associated transcriptional regulators, such as the cohesion complex, Mll3/4 complex, mediator complex, super elongation complex (SEC), and components of the general transcription machinery (Figure 3C). These proteins harbored more than 150 CBP/p300-regulated acetylation sites in KO cells, and CBP/p300 acetylated a majority (65%) of the sites detected on these proteins. Together, these results show that CBP/p300 acetylates hundreds of sites on enhancer-associated proteins, revealing that the scope of CBP/p300-catalyzed acetylation at enhancers far exceeds what is currently known, and indicating that H3K27ac represents the ‘tip of the iceberg’ of CBP/p300 targets at enhancers.

Kinetic Analysis of the CBP/p300 Acetylome

The dynamic turnover rate of acetylation at individual sites remains an almost completely unexplored question. We exploited the rapid action of CBP/p300 catalytic inhibitors to obtain site-specific kinetics of acetylation on a proteome-wide scale. SILAC-based quantitative MS was used to quantify acetylation at various times following treatment with A-485 (Table S3). The number of CBP/p300-regulated sites increased in a temporal manner (Figures 4A and 4B), and with an increasing time period of treatment showed an increasing correlation to reduced acetylation after 16 hours exposure to A-485 (Figure 4C). We quantified a total of ~8,100 sites, of which 3,011 were quantified at all six time points. The sites quantified across all time points were clustered into three groups; ‘unregulated’, ‘slow’, and ‘fast’ regulated (Figure 4D), and deacetylation half-lives were calculated for the 619 sites that showed reduced acetylation within this timeframe. While we limited our calculation of acetylation half-life to sites quantified at all six time points, all additional measurements are accessible (Table S3), and the data can be visualized using the web resource (see details below).

Deacetylation half-lives of regulated sites were broadly distributed, with a median half-life of 94 minutes (Figure 4E). Reduced acetylation after catalytic inhibition can be attributed to the action of lysine deacetylases (KDACs) or protein turnover. Comparison of acetylation turnover rates to protein half-life in NIH3T3 cells (McShane et al., 2016), showed that KDAC activity is mainly responsible for the rapid turnover of CBP/p300-catalyzed acetylation (Figure 4F). Furthermore, individual proteins harbored sites with different deacetylation kinetics (Figure S6A), indicating that deacetylation is regulated at the site-level. Notably, rapidly deacetylated sites (t_1/2 < 30 minutes) had significantly higher relative acetylation stoichiometry as determined by ACI (Figure 4G), indicating that the sites most acetylated by CBP/p300 are also the most strongly regulated by KDAC activity. Nearly one hundred sites were rapidly deacetylated, including CBP/p300 itself, and various transcription factors and co-activators such as Mef2a, Tp53, Bcl9l, Gtf2f1, and Mybl2 (Figure 4H). Together, these results provide the first systems survey of endogenous acetylation site half-lives and demonstrate very rapid turnover of many CBP/p300-regulated sites on both histone and non-histone proteins.

Histone Acetylation by CBP/p300

All four core histones are reported to be acetylated by CBP/p300 (Schiltz et al., 1999), often redundantly with other KATs (Roth et al., 2001). However, a broad survey of CBP/p300-regulated histone acetylation marks has not been reported. We quantified acetylation at nearly all known histone sites (Table S2), enabling us to obtain a comprehensive map of endogenous CBP/p300-catalylzed histone acetylation marks. To minimize redundant data points we calculated the median acetylation ratio, summed intensity, and median deacetylation half-life by combining isoform-specific histone sites that occurred at the same position in different histone isoforms (Table S4).

CBP/p300 acetylated sites on all four core histones; however, to a differing degree and site specificity. The degree of regulation was most pronounced for sites on H2B and H3; acetylation of H3K18, H3K27, H3K36, and N-terminal H2B sites was 95–99% reduced by KO, Cmpd-R, or A-485, indicating that these sites are almost exclusively acetylated by CBP/p300 (Figure 5A). Acetylation sites on H2A and H4 were reduced to a lesser degree, suggesting that these sites could be targeted by additional KATs or are not deacetylated to the same degree. CBP/p300 showed remarkable selectivity for sites on H3, targeting K18 and K27 while neighboring sites at K14 and K23 were unaffected. In contrast, H2A and H2B were regulated on nearly all N-terminal sites, which constituted over half of all sites on these histones, revealing more extensive regulation of H2A and H2B N-termini by CBP/p300 than previously reported.

Figure 5. Histone acetylation by CBP/p300.

(A) The heatmaps show the log2 SILAC ratios for acetylation sites on the indicated histones, treatment conditions, and cell lines. The log2 SILAC ratio is the median of sites conserved in different isoforms; CBP/p300-regulated sites are color coded as indicated. (B) Half-lives of the indicated histone acetylation sites from A-485-treated MEFs are shown. (C) The data show reduced acetylation at histone H2B sites following treatment with A-485 for different time points. (D) The rank plot shows the summed acetylated peptide intensity (SILAC light, control-treated) of the indicated core histone acetylation sites in untreated wild-type MEFs. Sites are ordered from the least to the most intense and colored according to their regulation in KO cells (Log2 Ac ratio KO/Ctrl).

See also Figure S6.

To gain further insights into CBP/p300-regulated histone sites, we examined their site-specific turnover rates and acetylated peptide intensity (estimated abundance) (Table S4). We calculated deacetylation half-lives for 14 of the 20 histone sites that were identified as regulated in KO cells. H2B N-terminal sites had the fastest turnover rates, and twelve histone sites displayed half-lives of less than one hour (Figure 5B). Comparison of H2B deacetylation kinetics showed that N-terminal sites were rapidly deacetylated, while K20 and K24 were deacetylated at a slightly slower rate, and C-terminal sites were completely unaffected (Figure 5C). Comparison of acetylated peptide intensity showed that N-terminal H2B sites were among the most high intensity histone sites (Figure 5D). Extensive regulation by CBP/p300, rapid turnover, and high intensity indicate that N-terminal H2B acetylation is a robust, multi-site mark that, like H3K18ac and H3K27ac, is a signature for CBP/p300 catalytic activity on chromatin. Collectively, these results provide a first MS-based quantitative map of CBP/p300-regulated histone sites in mammalian cells, demonstrating that CBP/p300 indeed targets all four core histones in-vivo, but the magnitude of regulation, site-specificity, and kinetics of deacetylation are distinct for individual sites and histones.

In addition to acetylating sites on histone N-termini, CBP/p300 is also reported to acetylate a distinct class of histone sites that are located in the globular or C-terminal domains, including H3K56, H3K122, and H2BK120 (Chen et al., 2014; Das et al., 2009; Tropberger et al., 2013). However, CBP/p300 did not acetylate any of these sites in our experiments (Figure 5A). We used recombinant acetylated H3K56 as a spike-in standard to verify the lack of regulation by CBP/p300 at this position (Figure S6B). Using site-specific antibodies we found a modest effect of Cmpd-R on H3K56ac and H3K122ac, while H2BK120ac was strongly reduced (Figure S6C). Thus, we observed some regulation of these sites using acetylation-site-specific antibodies, but not by MS. Similarly, the antibodies detected increased acetylation at all three of these sites after treatment with the KDAC inhibitor tricostatin A (TSA) (Figure S6C), even though a previous study using MS showed that these same sites were unaffected by a panel of KDAC inhibitors, including TSA (Figure S5D) (Scholz et al., 2015b). Our MS-based data indicate that these sites are not regulated by CBP/p300, and our analysis suggests that previous findings may be a result of antibody cross-reactivity. Indeed, many PTM site-specific antibodies show cross-reactivity or non-specificity (Egelhofer et al., 2011), highlighting the importance of quantitative MS in defining regulated histone acetylation sites.

CBP/p300 Impacts the Proteome by Regulating Gene Transcription

To delineate the roles of different CBP/p300 functionalities in gene transcription we measured transcript abundance in KO, Cmpd-R, A-485, and CBP112-treated cells at the same time points and conditions used for acetylome and proteome analysis (72hrs for KO, 16hrs for inhibitor-treated cells) (Tables S1 and S5). Transcriptional changes in KO cells were well correlated (r= 0.75–0.76) to catalytic inhibitor-treated cells (Figure 6A and 6B). The transcriptional changes by Cmpd-R and A-485 were highly correlated (r= 0.94) (Figure 6C), indicating that these compounds have a nearly identical impact on gene expression. The reduced correlation between catalytic inhibitors and KO cells is likely attributable to the different time periods of these interventions. Loss of CBP/p300 protein or catalytic activity caused reduced expression at a subset of genes (11.3% in KO cells, 11.6% in Cmpd-R, and 10.4% in A-485), and gene expression was more frequently downregulated than upregulated, particularly for strongly regulated (> 4-fold changed) genes (Table S1). Transcriptional changes caused by CBP112 were modest compared to KO and catalytic inhibition; CBP112 caused reduced expression at 2.4% of genes, and only 0.2% (25 transcripts) were more than 4-fold reduced, compared to >3% in KO, Cmpd-R, and A-485-treated cells (Table S1). These data show that catalytic inhibition has a similar impact on transcript levels as loss of the entire CBP/p300 protein in KO cells, indicating that CBP/p300 catalytic activity is the primary mechanism that drives CBP/p300-dependent transcription. While the scaffold function of CBP/p300 is undoubtedly important, it does not appear sufficient to regulate transcription independently of CBP/p300 catalytic activity, or does so at a very small number of genes.

Figure 6. CBP/p300-catalyzed acetylation drives rapid transcriptional regulation.

(A–C) Correlation between transcriptome changes in KO, Cmpd-R, and A-485-treated MEFs. The number of transcripts analyzed (n), Pearson’s correlation (r), and P-value (P) are shown. (D) Quantification of protein abundance changes in KO, Cmpd-R, A-485 and CBP112-treated MEFs. (E) Proteome (protein) and transcriptome (RNA) changes are well-correlated in KO cells. The number of transcripts and proteins analyzed (n), Pearson’s correlation (r), and P-value (P) are shown. (F) Cyp1a1 transcript induction in 6-Formylindolo(3,2-b)carbazole (FICZ)-treated cells, relative to untreated control (Ctrl) cells in the indicated cell types. Cmpd-R and CBP112 were added at the same time as FICZ. (G) FICZ-induced Cyp1a1 transcript expression relative to time = 0. For Cmpd-R-treated samples, the inhibitor was added at the 1h time point and Cyp1a1 expression was determined at the 2h and 3h time points.

In addition to its well-known function in transcriptional regulation, CBP/p300 can also regulate protein abundance post-transcriptionally, for example, by regulating acetylation-dependent protein stability (Spange et al., 2009), or by directly ubiquitylating proteins (Shi et al., 2009). CBP/p300 KO, Cmpd-R, A-485, and CBP112 affected a subset of the proteome, mostly causing downregulation of proteins (Figure 6D, Table 1). A larger number of proteins were regulated in KO cells compared to Cmpd-R and A-485-treated cells, likely due to the different time-scale of these interventions. Notably, proteome changes were significantly (r= 0.75) correlated with transcriptome changes in KO cells (Figure 6E), showing that most changes in protein abundance resulted from altered gene expression and that post-transcriptional mechanisms play a minor role in CBP/p300-dependent proteome dynamics. In addition, of the 78 proteins that were more than 2-fold downregulated in KO, Cmpd-R, and A-485, only 7 were not significantly (P<0.05) reduced at the transcript level, and none of these proteins were acetylated by CBP/p300. Thus, in our dataset we found little evidence supporting a widespread role of CBP/p300-catalyzed acetylation in regulating protein stability.

Rapid Acetylation Kinetics Regulate Transcription

Aryl hydrocarbon receptor (AHR) is a major integrator of diverse environmental signals such as chemical toxins, cancer cell metabolites, and bacterial pigments (Murray et al., 2014). AHR-dependent transcriptional activation of the cytochrome P450 monooxygenase Cyp1a1 is required for detoxification of aromatic hydrocarbons. Multiple KATs, including Ncoa1, Ncoa3, and CBP/p300 are implicated in AHR-induced Cyp1a1 transcription (Fujii-Kuriyama and Mimura, 2005); however, the specific requirement of CBP/p300-catalyzed acetylation in AHR signaling is not known.

We found that CBP/p300 impacts AHR signaling at multiple levels; AHR was downregulated at both the transcript and protein levels in KO, Cmpd-R, and A-485-treated cells, and CBP/p300 acetylated the aryl hydrocarbon receptor nuclear translocator (Arnt), which binds to AHR to promote AHR-dependent transcriptional activation (Tables S2 and S5). Most notably, Cmpd-R-treatment completely blocked Cyp1a1 gene induction in response to the AHR ligand 6-Formylindolo(3,2-b)carbazole (FICZ) in both MEF and Kasumi-1 cells (Figure 6F). In contrast, CBP112 partially reduced Cyp1a1 expression (Figure 6F). These results show that CBP/p300 catalytic activity is indispensable for AHR-dependent Cyp1a1 transcription.

To determine how quickly loss of CBP/p300 catalytic activity disrupted Cyp1a1 gene expression, we added Cmpd-R one hour after FICZ treatment. Cyp1a1 transcript abundance continued to increase for three hours after FICZ treatment in control cells, whereas addition of Cmpd-R one hour after FICZ treatment completely blocked further increases in transcript abundance, demonstrating acute disruption of Cyp1a1 transcription (Figure 6G). These data show that CBP/p300 catalytic activity is required for Cyp1a1’s transcriptional maintenance, and that loss of this activity results in nearly immediate transcriptional arrest. Fast regulation of Cyp1a1 transcription suggests a causative role of acetylation in driving gene transcription, and strongly suggests that the subset of CBP/p300-regulated sites that display rapid turnover are critical for regulating gene transcription.

Database of the CBP/p300 Regulome

Our work greatly expands the known repertoire of the endogenous CBP/p300 targets. To make this resource easily accessible to the community we generated a user-friendly web-based database (http://p300db.choudharylab.org), which can be queried for a protein of interest to obtain information about CBP/p300-regulated acetylation sites, as well as changes in the protein and transcript abundance. In addition to providing an overview of all identified sites and their positions in the queried protein, it includes information about site quantification, site intensity, and peptide spectral counts (Figure 7), which are useful parameters in prioritizing putative regulatory sites for functional investigation. The database also provides a graphical overview of acetylation site regulation by different CBP/p300 perturbations used here, and allows visualization of acetylation dynamics at individual sites (Figure 7). The database shows interactions of the queried protein with other acetylated proteins in the dataset, and the interactive nature of the network allows further exploration of individual proteins in the network. Furthermore, the database can be batch-queried for proteins with specific functional domains, such as bromodomain, enabling a quick overview of CBP/p300-dependent acetylation on a protein family. These, and additional, features of the database offers a ‘one stop’ solution for convenient mining of all the datasets generated here.

Figure 7. P300DB: a database of CBP/p300-regulated acetylome, proteome and transcriptome.

A snapshot of the P300DB web resource database. Using Mef2a as an example, the figure shows functional and sub-cellular localization annotation of the protein from the UniProt database, positions of acetylated lysine in the protein, site-specific regulation of acetylation in different perturbations, acetylation site kinetics, network of Mef2a interacting proteins with CBP/p300-regulated acetylation, and impact of CBP/p300 perturbations on expression of Mef2a protein and transcript levels. A more detailed description of the database is available at http://p300db.choudharylab.org.

Discussion

Here we present a high-confidence map of the dynamic CBP/p300 acetylome, and reveal the impact of CBP/p300-catalyzed acetylation on transcript and protein regulation. Importantly, by combining chemical and genetic interventions, we were able to exclude the contribution of CBP/p300-associated KATs that may be impacted by loss of CBP/p300 protein.

Our ultra-deep acetylome expands the repertoire of known CBP/p300-regulated acetylation sites by more than 10-fold (Dancy and Cole, 2015) and our approach provides far greater stringency in attributing their regulation to CBP/p300. It is possible that some of the identified acetylation sites are not directly acetylated by CBP/p300, and that their reduced abundance results from the altered activity of other KATs or KDACs. However, several lines of evidence suggest that most sites are direct targets of CBP/p300: we did not observe altered acetylation levels on canonical KATs and KDACs that would indicate that CBP/p300 targeted these enzymes, or that their own auto-catalytic activity was reduced. Indeed, acetylation of previously identified histone substrates of other KATs (i.e. H3K23 for KAT6A/B; H4K16 for KAT8; H3K9 for KAT2A/KAT2B) was not reduced in CBP/p300 KO or inhibitor-treated cells, indicating that the activity of these KATs for their canonical substrates was not altered. Furthermore, overexpression of p300 caused increased acetylation at 80% of the sites that were downregulated in KO cells, suggesting that CBP/p300 directly targets these sites. While we cannot fully rule out indirect effects on acetylation, our data strongly suggest that CBP/p300 directly targets a majority of the sites regulated in our experiments.

CBP/p300 associates with several proteins that are reported to function as independent KATs, such as NCOA1, NCOA3, ATF2, and CLOCK (Drazic et al., 2016). We found no concrete evidence for a catalytic activity-independent role of CBP/p300 in promoting acetylation, for example, by recruitment of other KATs through its scaffolding function. Instead, the reduced acetylation observed in CBP/p300 KO appears to be entirely dependent on CBP/p300 catalytic activity. Our results indicate that proteins of the steroid receptor coactivator (SRC), which includes Ncoa1-3, are likely to be downstream targets of CBP/p300. While we cannot exclude the possibility that CBP/p300-associated KATs may target a limited set of sites that we did not quantify in our assays, their impact on the global CBP/p300-acetylome appears negligible.

Our analyses provide insights into the specificity of CBP/p300 as an acetyltransferase. We find that CBP/p300 acetylates substrate proteins independently of linear amino acid motifs, and instead acetylates a majority of sites on targeted proteins. We further find that CBP/p300 acetylates a multitude of transcriptional regulators and enhancer-associated proteins at closely-spaced lysines; consistent with CBP/p300’s previously reported tendency to acetylate lysine clusters (Thompson et al., 2001). Based on these observations, we propose a model whereby CBP/p300 acts as an ‘acetyl-spray’ at active enhancers, targeting accessible regions on proximal proteins. We postulate that such high-density acetylation on proximal sites could create a dispersed interface for recruitment of bromodomain (BRD) proteins and may have functional consequences for activating transcription. Indeed, in addition to acetylated histones, BRD proteins can be recruited by acetylated non-histone proteins and for several BRDs, tandem acetylation cooperatively increases their binding affinity to multiply acetylated ligands (Moriniere et al., 2009; Muller et al., 2011). Because locus-specific recruitment of BRDs, for example, through acetyl-mimetic synthetic transcription factors is sufficient to cause productive transcription at otherwise repressed loci (Erwin et al., 2017), it is conceivable that widespread CBP/p300-catalyzed acetylation at enhancers contributes to transcriptional activation by recruiting BRD proteins.

Acetylation is widely linked to transcriptional regulation, and numerous individual and community-wide efforts, such as the ENCODE project (https://www.encodeproject.org), have been undertaken to map the genomic distribution of histone acetylation marks. However, the specificity of KATs, and the kinetics of histone acetylation, is poorly understood. This information is important for obtaining a functional understanding of acetylation marks, and to move from correlative patterns to establishing causal relationships. A survey of histone modification-specific antibodies found that a considerable portion of the antibodies exhibit questionable specificity (Egelhofer et al., 2011); therefore, it is important to confirm the regulation of these marks by antibody-independent methods. Our MS-based quantitative map of histone acetylation defines sites that are exclusively acetylated by CBP/p300, which could help in more accurately defining genomic regions that are occupied by catalytically active CBP/p300. It is also notable that current epigenetic mapping of histone acetylation marks on chromatin is highly biased to H3 and H4, particularly H3K27ac and H3K9ac. We find that H2B is a major target of CBP/p300, but much less is known about genomic distribution of H2B acetylation, indicating that its functional and diagnostic potential may be underappreciated.

Our kinetic analyses provide a first proteome-wide survey of site-specific acetylation dynamics and identify a subset of sites which displayed turnover rates that were up to an order of magnitude faster than previously reported ‘very rapid’ turnover of histone H3/H4 acetylation sites (Zheng et al., 2013). These results are reminiscent of the rapid dynamics of phosphorylation signaling, and show that acetylation signaling can occur with a similar degree of exquisite enzymatic regulation.

The composite picture emerging from our quantitative analyses of acetylation regulation, site-specific dynamics, and gene transcription, suggest that the subset of rapidly regulated CBP/p300-catalyzed acetylation sites is a critical driver of gene transcription. This conclusion is supported by our observation that inhibition of CBP/p300 activity results in an almost immediate arrest in AHR-induced gene transcription. Strikingly, CBP/p300 inhibition globally ablates acetylation at many sites, including a large number of sites on histones, yet it impacts transcription of only a subset of genes. This indicates that CBP/p300-catalyzed acetylation is a unique signature that regulates the transcription of specific genes. While a detailed understanding of how CBP/p300 regulates the transcription of specific genes will require further effort, our study provides a first step towards gaining a systems understanding of the complex relationship between KATs, acetylation, and gene transcription.

In conclusion, our in-depth analyses reveal the rich intricacies of the CBP/p300 acetylome, and the resource dataset presented here provides a framework for decoding CBP/p300 functions in diverse biological processes as well as for understanding the impact of small molecule inhibitors targeting its catalytic and bromodomain functions.

Star Methods

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Chunaram Choudhary (chuna.choudhary@cpr.ku.dk).

Experimental Model and Subject Details

Immortalized Crebbp^fl/fl; Ep300^fl/fl mouse embryonic fibroblasts (MEFs) stably expressing Cre-ERT2 were kindly provided by the laboratory of Paul Brindle (Kasper et al., 2010). The sex of these cells is not reported, and was not determined in our laboratory. Following transfection with Cre-ERT2, cell clones derived from single colonies were recovered and tested for efficient gene KO. Kasumi-1 cells are derived from a human male with acute myeloid leukemia (AML) with an 8;21 chromosome translocation. 293FT cells are a clonal isolate derived from human embryonal kidney cells transformed with the SV40 large T antigen. MEFs and Kasumi-1 were cultured (37°C, 5% CO2) in DMEM and RPMI1640 media (Biowest), respectively. For SILAC labeling, the media was supplemented with regular arginine and lysine, or with heavy isotope-labeled arginine (¹³C_6,¹⁵N₄-arginine, Sigma) and lysine (¹³C_6,¹⁵N₂-lysine, Cambridge Isotope Laboratories) and dialyzed serum (Sigma). All cell lines were routinely tested for mycoplasma.

Method Details

Cell culture and sample preparation

The floxed Cbp and p300 alleles were deleted by treating cells with 2μM 4-hyroxy-tamoxifen (4-OHT, Sigma), media was changed and fresh 4-OHT added every 12 hours for a total of four treatments over 48 hours. Cells were cultured for an additional 24–36 hours to allow for complete depletion of CBP/p300 protein. For CBP/p300 acetyltransferase inhibitors (3μM compound R (Cmpd-R) and A-485) and bromodomain inhibitor (5μM CBP112) cells were treated for 16 hours. Cells were harvested by washing 1× in PBS before on-plate lysis at 4°C using ice-cold lysis buffer (50mM Hepes, pH7.5, 150mM NaCl, 1mM EDTA, 1% NP-40, 0.1% sodium deoxycholate, 1× complete protease inhibitor cocktail (Roche)). After cell lysis, 1/10 volume ice-cold 5M NaCl was added to the lysates to release chromatin-bound proteins, the lysate was sonicated to shear genomic DNA, and was clarified by centrifugation at 8,000 × g for 15 minutes at 4°C. The supernatant was precipitated by addition of 4 volumes (80% final) −20°C acetone. Acetone precipitated protein was dissolved in 8M guanidine hydrochloride (GuHCl), 50mM Hepes, pH8.0 and protein content measured by BCA assay (Pierce). Protein was reduced and alkylated by 5mM Tris(2-carboxyethyl) phosphine hydrochloride (TCEP) and 5mM chloroacetamide for 1 hour at room temperature. Before addition of proteases the protein was diluted to 2M GuHCl by addition of 50mM Hepes, pH8.0. Lys-C was added (1/200 w:w) and the protein digested for 2–4 hours at 37°C, the protein was further diluted to 1M GuHCl, before digesting with trypsin protease (1/200 w:w) overnight at 37°C. Peptides were purified using a Sep-pak C18 cartridge (Waters) and were eluted in 50% acetonitrile. Acetylated peptide enrichment was performed using 10–20mg of purified peptides and high pH reversed-phase pre-fractionation, or by direct enrichment from unfractionated peptides followed by in-tip SCX fractionation or in-tip HpH fractionation (for fractionation method references, and details for individual experiments, see Tables S2 and S3). In all experiments acetylated peptides were enriched using the PTMscan acetyllysine antibody kit (Cell Signaling Technologies). For proteome measurements the peptides were fractionated by high pH reversed-phase into 24 fractions or 8–10 fractions followed by in-tip SCX fractionation. Fractionation strategies are indicated in Table S2. Approximately 1ug peptides were analyzed on the mass spectrometer for each proteome fraction.

CBP/p300 inhibitor synthesis and CBP/p300 biochemical activity assay of inhibitors

The synthesis of Compound-R (Cmpd-R) and A-485 and inhibition of CBP and p300 acetyltransferase activity were assessed as described (Lasko et al., 2017). Cmpd-R was identified based on its in vitro and cellular potency, and is related to compound 22 that we reported previously (Michaelides et al., 2018). Acetyltransferase activity assays were performed using truncated p300 and CBP BHC domain proteins that contain the region from the bromodomain to the C/H3 domain. Assays were performed by detecting lysine acetylation of a biotinylated histone H4 (1–23) synthetic peptide (SGRGKGGKGLGKGGAKRHRKVLRGG-K(Biotin)-NH2; (Anaspec, #AS-65097)) using a time-resolved fluorescence energy transfer (TR-FRET). Reactions were performed in a 10 μL volume using an assay buffer containing 100 mM HEPES, pH 7.9, 80 μM EDTA, 40ug/mL BSA, 100 mM KCl, 1 mM DTT, 0.01% triton X-100. The CBP/p300 inhibitor was dissolved in DMSO and dispensed at 50 nL by a Labcyte Echo (Labcyte, Sunnyvale, CA) into white 384 well low-volume plates (Perkin Elmer, #6008289) in 3-fold dilutions from 50 μM to 0.00075 μM. CBP or p300-BHC protein at 0.6 nM was pre-incubated with Cmpd-R for 30 minutes. The reaction was initiated by adding 5 μL of a biotinylated synthetic histone-H4 peptide at 2 μM and acetyl coenzyme A (Sigma-Aldrich, #A2056) at 0.5 μM for p300 or CBP-BHC. Following incubation for 1 hour at room temperature in a humidified chamber the reaction was terminated with 10 μL of 3 nM LANCE Ultra Europium-anti-acetyl-histone H4 lysine antibody (Perkin Elmer, #TRF0412-D), 900 nM LANCE Ultra ULight-Streptavidin (Perkin Elmer, #TRF0102-D) in LANCE Detection Buffer (PerkinElmer, #CR97-100). TR-FRET measurements were obtained using a Perkin Elmer Envision with laser excitation at 335 nm and emission at 665 nm and 620 nm.

Recombinant protein acetylation in vitro

Full-length Codon-optimized His-p300-Strep2-Flag inserted into a pVL1393 baculovirus expression plasmid was expressed in Sf9 cells. Expressed p300 protein was purified from Sf9 cells over a HiTrap column and quantified/purity confirmed by Coomassie staining (Zucconi et al., 2016). Six proteins that were identified as p300 substrates in our acetylome analyses were used for p300 acetyltransferase assay. Five of these proteins were purchased from OriGene as recombinant purified proteins (TTC33;TP302360, FAM192A;TP304378, TRIM33;TP710251, NPM1;TP303841, JUNB;TP303595) and PPM1G was purified from yeast cells, as follows. The Invitrogen ORF for full-length, GST-tagged PPM1G was transformed into yeast cells. After galactose induction, cells were lysed; and PPM1G protein purified by GST-affinity chromatography and glutathione elution. Protein concentration and purity were assessed by colloidal Coomassie stained SDS-PAGE. Acetylation reactions were performed using ~1.5ug of protein substrate incubated with or without 40 nM p300 in buffer containing 50 mM HEPES, pH 7.9, 50 mM NaCl, 1 mM TCEP, and 10 mM sodium butyrate. Reactions were initiated with 10 μM acetyl-CoA to a final volume of 15 μl. After incubation at 30°C for 30 min, reactions were quenched with 1% HCl aqueous solution. The entire experiment (from in-vitro acetylation to mass spectrometry) was performed in duplicate. Protein was re-suspended in 100ul 2% sodium deoxycholate, 50mM Hepes pH8.0, reduced with 1mM DTT at 37°C for 30min and subsequently alkylated by 5mM chloroacetamide for 1 hour at room temperature, prior to digestion with ~1/100 trypsin protease at 37°C for 16 hours. Approximately 200fmol protein was analyzed by mass spectrometry in technical triplicate measurements.

Cell lysate protein acetylation

SILAC labeled Crebbp^fl/fl; Ep300^fl/fl MEF cells (described above) were harvested by typsin, washed twice in PBS, and flash frozen in liquid nitrogen. Cell pellets were resuspended in hypotonic cell lysis buffer (10 mM Tris-HCl, pH 8.0, 1 mM KCl, 1.5 mM MgCl2, 1 mM DTT, 10 mM sodium butyrate) and flash frozen. Pellets were thawed, and lysed by passage through a QiaShredder column (Qiagen). The cell lysate was concentrated with an Amicon 3K MWCO column to ~10mg/ml (BCA assay). The lysates were then preincubated in reaction buffer (50 mM HEPES, pH 7.9, 50 mM NaCl, and 1 mM TCEP) with 40 nM p300 +/− 20 μM A-485 for 15 min at 4°C. The reaction was initiated with the addition of 20 μM acetyl-CoA and allowed to proceed for 30 min at 30°C. Reactions were quenched with 1% aqueous HCl and flash frozen for mass spectrometry. Digestion to peptides, enrichment of acetylated peptides, and SILAC-based quantification, was performed as described below.

Overexpression of p300 in 293FT cells

293FT cells were cultured in SILAC DMEM media as described above for MEF cells. A single 15cm plate at ~80% confluency was transfected with pCI-p300 vector expressing full-length wild-type p300 (Thompson et al., 2004) using Turbofect transfection reagent (Thermo Scientific, #R0531) according to the manufacturer’s protocol. Cells were harvested after 72hrs. Cell lysis, protein lysate preparation, digestion to peptides, enrichment of acetylated peptides, and SILAC-based quantification, was performed as described below.

High-content microscopy

PC-3 cells were plated in Collagen I coated 96-well view plates (Perkin Elmer, #6005810) overnight. Cells were then treated with an 8-point dose half-log dose response Compound-R starting at 10 μM for the indicated times. Cells were fixed in 10% formaldehyde (Polysciences, Inc., #04018) at room temperature for 10 min, washed in PBS, and then permeablized in 0.1% Triton X-100 for 10 min. Cells were then blocked in 1% BSA for 1 h and incubated with the incubated with H3K27Ac and H3K9Ac antibodies (Cell Signaling Technology; H3K9ac # 9649, H2K27ac #8173) in antibody dilution buffer (0.3% BSA in PBS) overnight at 4 °C. Cells were washed three times in PBS and then incubated with a mixture of Alexa Fluor488-conjugated goat anti-rabbit IgG antibodies (Life Technologies, #A-11029), Alexa Fluor555-conjugated goat anti-mouse IgG (Life Technologies, #A-21424) antibodies, and Hoechst 33342 (Life Technologies, #H3570) for 1 h at room temperature. After washing four times in PBS, plates were scanned within 24 h of processing on a ThermoFisher CellInsight using the target activation algorithm acquiring 15 fields per well. Fluorescence intensities were quantified using the average mean intensity function. IC50 values for H3K27Ac and H3K9Ac inhibition were calculated using a sigmoidal fit of the concentration/inhibition response curves.

Mass spectrometry

Peptide fractions were analyzed by online nanoflow liquid chromatography-coupled tandem mass spectrometry (nLC-MS/MS) using a Proxeon easy nLC system connected to a Q-Exactive HF mass spectrometer (Thermo Scientific).

Analysis of transcript abundance

Global transcript abundance was analyzed using biological triplicate measurements. RNA was purified using the RNeasy Mini Kit from Qiagen and RNA integrity was determined using the 2100 Bioanalyzer system (Agilent Technologies). Transcript abundance was measured using one-color 8 × 60K Mouse Gene Expression Microarrays (Agilent Technologies, G4852B) as per the manufacturer’s instructions. Briefly, 100 ng of total RNA was labeled using the Low Input Quick Amp Labeling Kit (Agilent Technologies, 5190-2305). Labeled samples were hybridized overnight and then washed and scanned using the high-sensitivity protocol (AgilentG3_HiSen_GX_1color) on a SureScan microarray scanner (Agilent Technologies), and probe intensities were obtained by taking the gProcessedSignal from the output of Agilent feature extraction software using default settings.

Quantitative PCR analysis of Cyp1a1 gene expression

All experiments were performed using the MEF and Kasumi-1 cells and culture conditions described above. To induce Cyp1a1 gene expression, cells were treated with 10μM 6-Formylindolo(3,2-b)carbazole (FICZ, Sigma) prepared in DMSO. RNA was isolated using the Qiagen RNeasy kit (Qiagen) according to the manufacturer’s protocol. Total RNA was quantified using a nano-drop spectrometer (Thermo) and complementary DNA (cDNA) was synthesized using the QuantiTect Rev. Transcription kit (Qiagen). Reverse transcription reactions were diluted with H₂O and stored at −20°C. Quantitative PCR was performed using Brilliant II SYBR^® Green QPCR Master Mix (Agilent) and a Stratagene Mx3005P instrument. The expression levels of TATA-box binding protein (TBP) and Cyp1a1 were determined using standard curve generated from the sample expressing the highest levels of Cyp1a1. TBP expression varied by less than 2-fold between individual samples and was used to correct Cyp1a1 expression for differences in cDNA content. The following primers were used for QPCR: mouse TBP, 5′-GTGCCAGATACATTCCGCCT-3′ and 5′-AGCTGCGTTTTTGTGCAGAG-3′;Cyp1a1, 5′-GAGGCTAAACAGCCTTCCCA-3′ and 5′-ACAGGAGCACCCTGTTTGTT-3′, and human TBP, 5′-TGGCACCAGGTGATGCCCTT-3′ and 5′-TGCCCAGATAGCAGCACGGT-3′;Cyp1a1, 5′-ACCAGGACCCTGTCCAATCT-3′ and 5′-GAAGGCAGCCCTGTTTGTTC-3′.

H3K56ac stoichiometry measurement

Site-specifically acetylated (at H3K56) histone H3 was produced in E. coli using the orthogonal N(epsilon)-acetyllysyl-tRNA synthetase/tRNA(CUA) technology (Neumann et al., 2008; Wang et al., 2016). To determine H3K56ac stoichiometry we first isolated histones from SILAC heavy-labeled control and Cmpd-R-treated MEFs (16h, 3μM Cmpd-R) using a standard acid extraction protocol. Briefly, cells were lysed in phosphate buffered saline (PBS), 0.5% NP-40, 1× complete protease inhibitor cocktail (Roche). Lysate was centrifuged at 1,000 × g, the supernatant discarded, and the pellet extracted with 0.2M hydrochloric acid overnight at 4°C. Extracted histones were recovered from the supernatant after centrifugation at 1,000 × g. Recombinant H3K56ac was mixed with histone preparations from SILAC heavy-labeled control and CmpR-treated cells in a 10-fold serial dilution to determine the ratio of recombinant to native H3. Recombinant H3K56ac was then mixed with histone preparations to a final stoichiometry of 0.1%, 0.01%, and 0.001% before trypsin digestion of histones to peptides, and acetylated peptide enrichment with anti-acetyllysine antibody. Acetylated peptides were analyzed by MS and the relative intensities of the light SILAC-labeled recombinant H3K56ac and heavy SILAC-labeled native H3K56ac peptides were compared. Stoichiometry was calculated based on these comparisons at all three dilutions (0.1%, 0.01%, and 0.001%). The experiment was performed in two independent biological replicates.

Quantification and Statistical Analyses

Peptide and protein identification

The raw MS data was computationally processed using MaxQuant (developer version 1.5.5.4) and searched against the UniProt database (downloaded July 6, 2015) using the integrated Andromeda search engine (Cox et al., 2011). The acetyl (K) sites table was filtered to remove reverse decoy database entries, contaminant entries, and entries with localization probability less than 90% (0.9). Identified sites required a minimum Andromeda search score of 40. Protein quantification required a minimum of 2 peptides, 1 unique peptide, and 2 SILAC ratio counts; common contaminants, proteins only identified by a modification site, and false hits identified from the reversed database were removed. Relative acetylation stoichiometry was estimated using acetylation site intensity corrected for differences in protein abundance, otherwise referred to as abundance-corrected intensity (ACI). ACI was calculated by dividing acetylation site intensity by iBAQ protein abundance.

Calculation of deacetylation half-life

The deacetylation rate after A-485-treatment was analyzed using the sites that were quantified at all seven time points. We applied fuzzy c-means method using Mfuzz to classify the quantified sites as “fast”, “slow” and “unregulated” (Futschik and Carlisle, 2005). Half-lives were only calculated for regulated (“fast” or “slow”) sites. We assumed that deacetylation kinetics at early time points follow an exponential decay, formulated as (dN/dt = −λN), where N is the concentration of each acetyl site and λ is the exponential decay constant. This is transformed as (log2 (N/N₀)) = −(t/t_1/2), where N₀ is the concentration at time 0, t_1/2 is the half-life and t is the time after A-485 treatment. We determined the ranges between Q1 – 3.0*IQR and Q3 + 3.0*IQR as an ‘acceptable variability range’ and used this as a threshold to exclude the outliers. To estimate the half-life, we used a multi-step model. First, we constructed a model using only sites with replicate data from early time points and estimated the parameter λ from the replicate data. To limit these analyses to high-confidence data, the data were selected based on the following criteria: (1) To ensure accurate quantification, we only used log2 ratio values that were between 0 and −3.32 (not more than 10-fold reduced), and also within the acceptable variable range from the mean of replicates, (2) Only the data from early time points (for “slow” within 4 hours, and for “fast” within 1 hour) were used, and (3) If no data remained, the data within 1 hour were used. Then, the parameter was estimated by the robust linear regression model using the R package MASS (Venables and Ripley, 2002). In the second step, the model was refined using all the data within the acceptable variability ranges. The parameter was updated by using all the data that passed the following criteria: (1) The log2 ratio values should be within the acceptable variable range from the estimated value of the model, (2) If no data had the log2 ratio value of less than −1, the data with the closest value of −1 were included, and (3) the step 2 was repeated until the parameters converged. We excluded data with more than 8 hours half-life because in our experimental setting and analysis method, it was difficult to correctly estimate half-life longer than 8 hours.

Statistical analysis of microarray data

The resultant microarray data was processed as follows; the intensity values were log2 transformed, for duplicated probes the mean was calculated, quantile normalization was used to correct for inter-array variability, for genes with multiple probes the median intensity values were used. The Statistical Analysis of Microarray (SAM) method was used to calculate significant (p value) differences, adjusted p values were calculated using the Benjamini-Hochberg method. Significantly regulated genes were classified based on an adjusted p value < 0.05 and a magnitude fold change >2. Low abundant, non-expressed transcripts were filtered from the data based on a probe intensity cutoff of Log2 <4 in all samples.

Statistical analyses

Acetylome measurements of CBP/p300 KO, Cmpd-R, A-485, and CBP112 (Figure 1) were performed in biological triplicates. Acetylome measurements of in vitro acetylation with recombinant p300 (Figure 2H), and in-vivo p300 overexpression in 293FT cells (Figure 2I) were performed in duplicate. Acetylome time course analysis after A-485-treatment (Figure 4) was performed in at least biological duplicate measurements for each time point, exact details are provided in Table S3. Microarray analysis of transcript abundance (Figure 6) was performed in a minimum of three biological replicates.

Statistical tests, such as Pearson’s correlation (r), Fisher test, and Wilcoxon test were performed in R. Boxplots were generated using R, the lower and the upper hinges of the boxes correspond to the 25% and the 75% percentile, the bar in the box shows the median. The upper whisker extends to the largest value but maximal 1.5 times the IQR from the upper hinge. Similarly, the lower whisker extends from the lower hinge at most 1.5 IQR. Outlier data points (outside the hinges) were not drawn. Gene Ontology analysis was performed using UniProt keywords and AGOTOOL (Scholz et al., 2015a). UniProt keywords were mapped to acetylation site-level data and the significance of enrichment was calculated by Fisher test. Acetylation site sequence logos were generated using the IceLogo (Colaert et al., 2009) stand-alone application version 1.2; comparing sequence windows of regulated (>2-fold decreased acetylation) to unregulated sites. Subcellular compartment analysis was performed by mapping proteins with the regulated sites to UniProt keywords, only the sites associated with proteins annotated to a single subcellular compartment were used in the analysis.

Data and Software Availability

The raw mass spectrometry data have been deposited to the ProteomeXchange Consortium via the PRIDE (https://www.ebi.ac.uk/pride) partner repository with the dataset identifier PXD005252. The microarray transcriptome data have been deposited in NCBI’s Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo) and are accessible through GEO Series accession number GSE92613.

Additional Resources

All of the data are accessible in the supplementary datasets (Tables S1–4) and can be further explored using our web-based resource (http://p300db.choudharylab.org).

Supplementary Material

S1

Figure S1. Structure of the CBP/p300 catalytic inhibitors, and strategy and reproducibility of quantitative acetylome analyses, Related to Figure 1.

(A) The structure of Compound R (Cmpd-R) and A-485 used to inhibit CBP/p300 acetyltransferase activity. The half maximal inhibitory concentration (IC₅₀) values are indicated as determined in panel C for Cmpd-R, and as reported in (Michaelides et al., 2018) for A-485. (B) The dose-response curve of Cmpd-R for the CBP and p300 BHC protein acetyltransferase activity using a time-resolved fluorescence energy transfer (TR-FRET) in vitro assay. (C) The dose-response curve for inhibition of H3K9ac and H3K27ac as determined by high content microscopy in PC-3 cells. (D) Venn diagrams showing overlap between individual biological replicate acetylome measurements for the indicated experimental interventions (E) Pearson’s correlation (r) between individual biological replicate acetylome measurements for sites found in all 12 experiments. (F) Venn diagram showing overlap between downregulated (Log2 Ac ratio < −1) sites in KO and CBP112-treated cells.

Figure S2. Properties of the CBP/p300-regulated acetylome, Related to Figure 2.

(A) Distribution of acetylation (Ac) site ratios and summed acetylated peptide intensity for an unperturbed (control:control) experiment. The number of quantified sites and percent of up- or down-regulated (>2-fold change) sites are indicated. (B) UniProt Keyword enrichment analysis comparing CBP/p300-regulated and unregulated acetylation sites in Cmpd-R and A-485-treated cells (P<5e⁻¹⁵, Fisher test). (C) IceLogo (Colaert et al., 2009) was used to visualize sequence biases at CBP/p300-acetylated lysines. Downregulated (Log2 Ac ratio < −1) were compared to unregulated (Log2 Ac ratio > −1) sites. The Icelogos show the percent difference in amino acid frequency at each position compared to unregulated sites with a P-value cutoff of 0.05. Filled logos show the fraction of each amino acid in positions flanking the regulated sites. (D–E) CBP/p300-regulated sites are significantly more likely to occur in close proximity when compared to random sampling. Error bars indicate the 95^th percentile of the random sampling (*P≤0.05, permutation test).

Figure S3. CBP/p300-regulated acetylation sites in-vivo and in-vitro, Related to Figure 2.

(A) The diagram shows examples of proteins with different acetylation patterns. Each ‘lollypop’ indicates an individual acetylation site. The color coding indicates the SILAC ratio for acetylation (Ac) for the indicated experimental conditions (KO, Cmpd-R, A-485) as compared to control. (B) The sequence alignments show homologous proteins that are acetylated by CBP/p300 at a conserved position. (C) In-vitro acetylation of purified human proteins by full-length, recombinant p300. The color code of the heatmaps shows intensity of acetylated peptides in mock-treated control samples and in samples treated with recombinant p300. The color code of circles (next to the indicated AcK sites) indicates the greatest degree of reduced acetylation observed in KO, Cmpd-R, or A-485. (D) Fraction of acetylation sites that were only detected in the samples treated with recombinant p300, data from the panel C. (E) Fraction of sites detected in both mock-treated control and recombinant p300-treated samples whose acetylation increased more than 2-fold by recombinant p300, data from the panel C.

Figure S4. The CBP/p300 acetylome is conserved, Related to Figure 2.

(A) The scatterplot shows the distribution of acetylation sites ratios (Cmpd-R/Control) versus the summed SILAC light (control-treated) acetylated peptide intensity in human Kasumi-1 cells. The number of quantified sites is indicated in parenthesis. Sites that were >2-fold up- or down-regulated (Log2 SILAC Ratio >1 or <−1) are shown in red and the percent of regulated sites is indicated in bold type. (B) Subcellular distribution of CBP/p300-regulated acetylation. The analysis was performed using only proteins that associated with only one of the indicated UniProt Keywords, all differences are highly significant (P<1e⁻⁴⁰) by Fisher test. (C) UniProt Keyword enrichment analysis comparing CBP/p300-regulated and unregulated acetylation sites in Kasumi-1 cells (P<1e⁻¹¹, Fisher test). (D) The scatter plot shows the correlation between evolutionarily conserved acetylation sites in Cmpd-R-treated MEFs and Kasumi-1 cells. The number of sites analyzed (n), Pearson’s correlation (r), P-value (P), and percent of regulated sites are shown.

Figure S5. CBP/p300 acetylates transcription factors, chromatin remodelers, and transcriptional co-regulators, Related to Figure 3.

The diagram shows transcription factors, chromatin remodelers, and transcriptional co-activators with CBP/p300-regulated acetylation sites in MEF KO, Cmpd-R, or A-485-treated cells. The classification of TFs is based on the AnimalTFDB (http://bioinfo.life.hust.edu.cn/AnimalTFDB). The phylogenic trees are constructed using CLAP (Gnanavel et al., 2014) and visualized using (Letunic and Bork, 2016). The color code shows protein abundance changes and the lowest acetylation site ratio after KO, Cmpd-R, or A-485-treatment.

Figure S6. Kinetics of CBP/p300-regulated sites and validation of site-specific histone acetylation, Related to Figures 4 and 5.

(A) Overlap between sites with different deacetylation kinetics (Fast, Slow, or unregulated as defined by clustering shown in Figure 4D) on individual proteins. This shows that individual proteins harbor sites with different kinetics. (B) The plots show the H3K56ac peptide intensities for recombinant H3K56ac (recombinant, SILAC light) and native H3K56ac (native, SILAC heavy). The native and recombinant acetylated peptides were analyzed in the same MS run and their peptide intensities were derived from a side-by-side comparison (as in SILAC quantification). Recombinant H3K56ac was spiked-in at the indicated stoichiometry based on empirically determined H3 abundance in each sample. Median stoichiometry was calculated based on the ratio of H3K56ac intensity at all three concentrations of recombinant H3K56ac shown in the figure. (C) Western blot analysis of Cmpd-R- and TSA-treated (16 hours) MEF cells with the indicated antibodies [Cell Signaling Technology: H2BK5ac (#12799), H2BK15ac (#9083), H2B (#8135), H3K18ac (#13998), H3K27ac (#8173), H3(#4499); Abcam: H3K122ac (ab33309), H3K56ac (ab76307); Upstate: H2BK120ac (07–564)]. (D) Heatmap showing the degree of selected histone acetylation site regulation after treatment with the indicated deacetylase inhibitors. Data is from (Scholz et al., 2015b).

Table S1. Summary of the acetylome, proteome, and transcriptome analyses, Related to Figures 1 and 6.

The table shows the number of acetylation (Ac) sites, proteins, and transcripts quantified, as well as the number of independent biological replicates, and the number of 2-fold (2×), 4-fold (4×), or 8-fold (8×) downregulated (down) or upregulated (up) Ac sites, transcripts or proteins in the indicated experimental conditions. The fraction of regulated sites is indicated in parenthesis.

S2. Table S2. An experimental summary of acetylome and proteome analyses, Related to Figure 1 and Table S3.

The Excel spreadsheet contains column header descriptions as well as 10 spreadsheets containing the following data:

2a. KO MEF Ac sites: A list of all acetylation sites identified in CBP/p300 KO MEFs and the relative change in acetylation (CBP/p300 KO/Control WT).

2b. Cmpd-R MEF Ac sites: A list of all acetylation sites identified in Cmpd-R-treated MEFs and the relative change in acetylation (Cmpd-R/Control).

2c. A-485 MEF Ac sites: A list of all acetylation sites identified in A-485-treated MEFs and the relative change in acetylation (A-485/Control).

2d. CBP112 MEF Ac sites: A list of all acetylation sites identified in CBP112-treated MEFs and the relative change in acetylation (CBP112/Control).

2e. Cmpd-R Kasumi-1 Ac sites: A list of all acetylation sites identified in Cmpd-R-treated Kasumi-1 and the relative change in acetylation (Cmpd-R/Control).

2f. KO MEF proteins: A list of all proteins quantified in CBP/p300 KO MEFs and the relative change in protein abundance (KO/Control WT).

2g. Cmpd-R MEF proteins: A list of all proteins quantified in Cmpd-R-treated MEFs and the relative change in protein abundance (Cmpd-R/Control).

1h. A-485 MEF proteins: A list of all proteins quantified in A-485-treated MEFs and the relative change in protein abundance (A-485/Control).

2i. CBP112 MEF proteins: A list of all proteins quantified in CBP112-treated MEFs and the relative change in protein abundance (CBP112/Control).

2j. Cmpd-R Kasumi-1 proteins: A list of all proteins quantified in Cmpd-R-treated Kasumi-1 and the relative change in protein abundance (Cmpd-R/Control).

2k. In-vitro p300 acetylation of MEF cell lysate Ac sites: A list of all acetylation sites identified in recombinant p300-treated MEF cell lysate and the relative change in acetylation (recombinant p300/Control).

2l. p300 transfection human 293FT cells Ac sites: A list of all acetylation sites identified in p300-transfected 293FT cells and the relative change in acetylation (p300 overexpression/Control).

S3. Table S3. A list of acetylation sites quantified, Related to Figure 1 and Figure S4.

The Excel table contains a list of all acetylation sites quantified in A-485-treated MEFs and the relative change in acetylation (A-485/Control) at the indicated time points (10, 20, 30, 60, 120, 240, 480 minutes).

S4. Table S4. A list of acetylation sites quantified on core histones, Related to Figure 5.

The Excel table contains a non-redundant list of acetylation sites quantified on core histones in CBP/p300 KO, Cmpd-R-, A-485-, and CBP112-treated MEFs, and in Cmpd-R-treated Kasumi-1 cells. For any given core histone the isotype-specific acetylation sites were combined into a single entry by calculating the median SILAC ratio, summed site intensity, and median half-life, for all entries occurring at the same position.

S5

Table S5. A list of transcripts quantified in CBP/p300 KO, Cmpd-R-, A-485- and CBP112-treated MEFs, Related to Figure 6.