Measuring acetylation stoichiometry
We measured acetylation stoichiometry in HeLa cells utilizing partial chemical acetylation and serial dilution SILAC (SD-SILAC) to make sure quantification accuracy8 (Fig. 1a). Two impartial organic replicates had been carried out, every utilizing a special diploma of chemical acetylation and inverting the SILAC labeling between experiments. The diploma of chemical acetylation was estimated based mostly on the median discount of unmodified peptides generated by tryptic cleavage at one or two lysine residues (Supplementary Determine 1a). Primarily based on the estimated diploma of chemical acetylation, we carried out a serial dilution of the chemically acetylated peptides to offer median ~1%, ~zero.1%, and ~zero.01% chemical acetylation. Acetylated peptides had been enriched and the variations between native acetylated and chemically acetylated peptides quantified by MS (Supplementary Information 1a). To make sure correct quantification, we required that the abundance of native acetylated peptides was quantified by comparability with at the very least two completely different concentrations of chemically acetylated peptides, and that the measured SILAC ratios agreed with the serial dilution sequence. SILAC ratios that didn’t observe the dilution sequence (permitting as much as two-fold variability) had been outlined as being inaccurately quantified, although one of many measurements could also be right. Quantification error was decreased when the focus of chemically acetylated peptides was most just like native acetylated peptides (Fig. 1b). Nonetheless, quantification error was considerably greater than in our earlier experiments in bacteria8, seemingly as a result of higher complexity of the human proteome. The excessive error charges spotlight the necessity to management for quantification accuracy, and present that evaluating native acetylated peptides to simply 1% chemically acetylated peptides leads to a majority of false quantification (Fig. 1b). The measured stoichiometry of acetylated peptides was considerably and extremely correlated between impartial experimental replicates (Fig. 1c). The precision of our measurements was additionally extremely reproducible; the median ratio of stoichiometry between replicates was zero.95, and 90% of the measurements different by lower than an element of two between replicates (Fig. 1d).
Measuring acetylation stoichiometry. a Diagram of the strategy used to measure acetylation (Ac) stoichiometry. b The diploma of quantification error as decided by the fraction of SILAC ratios at every focus of chemically acetylated peptides that was not in step with SILAC ratios measured in at the very least one completely different focus of chemically acetylated peptides. c The correlation between stoichiometry measured in impartial experimental replicates. The variety of peptides (n), Pearson’s correlation (r), and P-value (P) of correlation are proven. d Low absolute variability between experimental replicates. The histogram reveals the distribution of Log2 ratios of stoichiometry in Experiment 1/Experiment 2 (Exp.1/Exp.2). e The correlation between stoichiometry measured utilizing partial chemical acetylation (PCA) and absolute quantification (AQUA) peptide requirements. f Low absolute variability between stoichiometry measurements made by PCA and AQUA. g Validation of stoichiometry measurements utilizing recombinant acetylated (100%) proteins as a spike-in customary. Stoichiometry was measured at two completely different concentrations of spike-in protein (SILAC gentle, purple) in comparison with SILAC heavy-labeled HeLa (blue) for every acetylation website. Supply information are supplied as a Supply Information file
For top stoichiometry (>10%) acetylation, the distinction between native acetylated and chemically acetylated peptides turns into too small to precisely measure by SILAC quantification, which is often restricted to variations higher than 2-fold in magnitude. At 5% partial chemical acetylation, a peptide with 90% stoichiometry could have a SILAC ratio of 1.006, and a peptide with 50% stoichiometry could have a SILAC ratio of 1.05 (Supplementary Determine 1b). These variations are too small to precisely resolve, and may end up in inaccurate stoichiometry measurements which can be out of bounds (higher than 100% or unfavourable). Because of the inherent limitations in calculating stoichiometry for these peptides, we set a cutoff of most 10% stoichiometry, and categorized all websites exceeding this cutoff as having >10% stoichiometry. Solely 16 peptides met these standards, they usually harbored beforehand recognized excessive stoichiometry acetylation websites (Supplementary Information 1a), together with; histones H3 Okay23 (2 completely different peptides), H3 Okay14, and H4 Okay16. Different peptides harbored websites on acetyltransferases, corresponding to CBP Okay1583 (2 completely different peptides) and N-alpha-acetyltransferase 50 Okay33, or on proteins that catalyze reactions utilizing acyl-CoAs, corresponding to hydroxymethylglutaryl-CoA lyase Okay48 and dihydroxyacetone phosphate acyltransferase Okay643. These information present that, though we’re unable to precisely measure excessive stoichiometry acetylation, we had been capable of determine these peptides, they signify recognized or possible excessive stoichiometry acetylation websites, they usually represent a small portion (zero.2%) of the peptides analyzed.
Validating stoichiometry measurements
We used recognized portions of unmodified and acetylated peptide requirements (AQUA peptides) to find out stoichiometry straight at ten websites on three proteins, cortactin (CTTN), nucleolin (NCL), and N-acetyltransferase 10 (NAT10) (Supplementary Information 1b). Stoichiometry decided by partial chemical acetylation (PCA) was considerably correlated (r = zero.94) with stoichiometry decided utilizing AQUA peptides (Fig. 1e). Moreover, stoichiometry measurements differed by an element of two or much less for a majority (7/10) of the analyzed peptides (Fig. 1f). Three websites confirmed three.Four-, 6-, and 7-fold greater stoichiometry by PCA, indicating overestimation of stoichiometry by PCA or underestimation by AQUA. These variations occurred on the site-level and had been due to this fact not attributable to errors in protein quantification. We predict that the settlement between these two strategies is notable when contemplating all doable sources of variability in every measurement.
We additional validated our measurements utilizing two recombinant, site-specifically acetylated proteins; malate dehydrogenase (MDH2) Okay239ac and hydroxymethylglutaryl-CoA lyase (HMGCL) Okay48ac. Recombinant acetylated proteins (SILAC-light-labeled) had been used as a spike-in customary to measure acetylation stoichiometry in SILAC-heavy-labeled HeLa lysate. Stoichiometry measured utilizing two completely different concentrations of recombinant acetylated protein (spike-in) agreed with our measurements utilizing PCA, additional supporting the accuracy of our stoichiometry dataset (Fig. 1g).
Stoichiometry measurements had been moreover validated by comparability to acetylated peptide depth that was corrected for variations in protein abundance. We beforehand confirmed that abundance-corrected depth (ACI) is correlated to acetylation stoichiometry in yeast6. ACI was considerably correlated with acetylation stoichiometry in HeLa cells (Supplementary Determine 1c), nevertheless, the predictive energy of this correlation was modest (r = zero.48–zero.52). There are a number of causes for this modest correlation. Firstly, peptide depth is inherently variable. Secondly, protein abundance estimates could also be inaccurate. We discovered that outlier information factors utilizing iBAQ-based protein abundance weren’t outliers when utilizing copy-number-based protein abundance (Supplementary Determine 1c), indicating that variability in protein abundance measurements contributes to disagreement between ACI and stoichiometry measurements. Thirdly, antibody-based acetylated peptide enrichment could also be peptide-sequence biased, which is able to introduce additional variability. No matter these limitations, ACI supplies a straightforward technique to estimate the relative stoichiometry of acetylation websites, and the numerous correlation with our stoichiometry measurements by PCA supplies additional assist for the accuracy of our measurements.
Copy quantity limits the detection of acetylated peptides
The detection of acetylated peptides is biased to plentiful proteins (Fig. 2a). Moreover, the fraction of lysines which can be detected as acetylated on any given protein is considerably correlated with protein abundance (Fig. 2b). This bias is present in each acetylome dataset that we’ve examined (Supplementary Determine 2a)9,10,11,12, and signifies that acetylation happens on most lysine residues in cells and that protein abundance is a limiting issue within the detection of acetylated peptides. These information additional assist the notion that each one solvent accessible lysine residues are acetylated to a point, both enzymatically or nonenzymatically.
Stoichiometry limits the detection of acetylated peptides. a Acetylation is biased to detection on plentiful proteins. Protein copy quantity estimates are from42. b The fraction of acetylated lysines detected on any given protein is correlated with protein abundance. The scatterplot reveals the % lysines acetylated and replica numbers of 2488 acetylated proteins containing 6753 acetylation websites. The Pearson’s correlation (r), and P-value (P) of correlation are proven. c The variety of peptides and acetylated peptides (Ac peptides) detected in deep proteome measurements from this research and13. 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 quantity. e The distribution of acetylated peptide copy numbers for acetylated peptides that had been detected with out prior antibody enrichment. Supply information are supplied as a Supply Information file
Deep proteome measurements detect unmodified peptides from proteins whose abundance spans seven orders of magnitude (Fig. 2a)13. This raises the query of why we detect so few acetylated peptides with out antibody enrichment (Fig. 2c). The sign depth of acetylated peptides is akin to unmodified peptides within the mass spectrometer (Supplementary Determine 2b, c), indicating that we should always have the ability to detect acetylated peptides as readily as unmodified peptides. We in contrast copy numbers for unmodified peptides to repeat numbers for acetylated peptides as decided from our stoichiometry measurements. The distribution of acetylated peptide copy quantity reveals that, within the absence of acetylated peptide enrichment, most acetylated peptides are at or beneath the detection restrict of the mass spectrometer, even in deep proteome measurements (Fig. second). In distinction, acetylated peptides that had been detected with out antibody enrichment occurred at copy numbers that had been inside the detectable vary of unmodified peptides (Fig. 2e). Thus, our stoichiometry measurements are in step with the shortcoming to detect acetylated peptides with out enrichment. Strikingly, our information point out that some acetylation occasions are so uncommon that they happen at a duplicate quantity that’s lower than one per cell (Fig. second).
Properties of excessive stoichiometry acetylation
We measured the stoichiometry of acetylated peptides; nevertheless, particular person acetylation websites could happen on a number of completely different peptides resulting from incomplete tryptic digestion, protein N-terminal acetylation, or oxidized methionine residues. To look at acetylation stoichiometry on the site-level, we calculated the summed stoichiometry of peptides containing the identical acetylation website (Supplementary Information 1c). This resulted in stoichiometry measurements for 6829 websites, with a median stoichiometry of simply zero.02% (1/4000 molecules) (Fig. 3a). This represents very low ranges of acetylation for many websites, just one% (66 websites) displayed stoichiometry >1%, and ~15% (1014 websites) displayed stoichiometry >zero.1%.
Properties of excessive stoichiometry acetylation. a The distribution of acetylation website stoichiometry for the 6829 websites measured on this research. b UniProt key phrase enrichment for the indicated lessons of excessive stoichiometry acetylation websites (>zero.23% and >zero.1%) and for doubly acetylated peptides. The dimensions of the textual content is expounded to the fraction of websites related to the key phrase, and key phrases that had been greater than two-fold enriched are coloured purple. c Subcellular compartment evaluation based mostly on the Human Protein Atlas14. Class scatterplots present the distributions of acetylation website stoichiometry in every subcellular compartment. The quantity (n) of websites analyzed, median stoichiometry (median), and share of websites with >1% stoichiometry (>1%) is proven. d Amino acid sequence logos of the indicated lessons of acetylation websites utilizing IceLogo15. Supply information are supplied as a Supply Information file
We carried out UniProt key phrase enrichment evaluation to look at the useful classes of proteins which can be related to greater stoichiometry (>zero.23% or >1%) acetylation (Fig. 3b). Greater stoichiometry acetylation was overrepresented on nuclear proteins concerned in chromatin regulation and transcription. This statement is in step with the recognized nuclear capabilities of acetyltransferases, deacetylases, and acetylated lysine-binding bromodomain proteins. In truth, the key phrases Bromodomain and Acetyltransferase had been considerably enriched within the group of proteins with excessive stoichiometry acetylation. We had been unable to calculate stoichiometry for doubly acetylated peptides due to the low frequency of chemical acetylation at each positions. Nonetheless, given the low stoichiometry of acetylation, doubly acetylated peptides are unlikely to happen by random likelihood and almost certainly replicate the exercise of acetyltransferases. In line with this concept, doubly acetylated peptides occurred on proteins that had been overrepresented for a similar UniProt key phrases that had been related to excessive stoichiometry acetylation (Fig. 3b). Thus, websites occurring on doubly acetylated peptides could happen at excessive stoichiometry and are more likely to be enzyme-catalyzed.
To analyze the connection between stoichiometry and subcellular compartmentalization we used immunofluorescence-based protein localization as decided by the Human Protein Atlas14. Acetylation stoichiometry was broadly distributed and largely comparable in each subcellular compartment analyzed (Fig. 3c). Mitochondrial acetylation occurred at a barely, but considerably (P < 5e−5, Wilcoxon test), higher median stoichiometry. However, mitochondria contained the smallest fraction of high (>1%) stoichiometry acetylation websites. In distinction, the nucleus contained the best fraction of excessive stoichiometry websites, which was roughly an order of magnitude higher than in mitochondria (Fig. 3c).
We used IceLogo15 to find out whether or not excessive stoichiometry acetylation was related to neighboring amino acids. Cysteine residues had been notably overrepresented for websites with >zero.23% stoichiometry (10-fold greater than median stoichiometry), notably within the −Four, −three, and −2 positions (Fig. 3d). Nonetheless, this bias was absent when analyzing websites with >1% stoichiometry, indicating that this overrepresentation was related to websites with reasonably elevated stoichiometry. Remarkably, websites with cysteine residues within the −Four, −three, or −2 place constituted 35% (159/460) of the websites with >zero.23% stoichiometry. UniProt key phrase enrichment evaluation of the proteins harboring these websites discovered quite a lot of enriched key phrases (Supplementary Information 1d). Nonetheless, not like excessive stoichiometry acetylation on the whole (Fig. 3b), key phrases describing processes related to nuclear acetyltransferases, corresponding to Nucleus, Transcription, and Chromosome had been notably absent. These information counsel that cysteine residues could promote nonenzymatic acetylation of downstream lysine residues, and these websites represent a considerable portion (35%) of websites with an elevated (>zero.23%) stoichiometry of acetylation. This conclusion is supported by a current research that uncovered an identical bias for greater stoichiometry acetylation at websites with proximal cysteine residues16.
We measured stoichiometry at 57 histone websites and located that prime stoichiometry acetylation was largely restricted to websites on the N-terminal tails of core histones H2B, H3, and H4 (websites on H2A weren’t measured) (Fig. 4a). The stoichiometry of histone H3 and H4 acetylation websites has been extensively studied17,18,19,20, and our measurements are comparable to those earlier measurements (Fig. 4b). Nonetheless, our technique doesn’t measure the stoichiometry of doubly acetylated peptides that may come up from the lysine-rich N-terminal tails of the core histones H2A, H2B, H3, and H4. Feller et al.18 discovered that the stoichiometry of mono-acetylated histones H3 and H4 had been extra plentiful than di-acetylated H3 and H4, except H4 K5 + Okay12 and H3 Okay18 + 23 (which is extra plentiful than Okay18 alone, however much less plentiful than Okay23 alone). We detected doubly acetylated peptides containing the next websites on H2A (K5 + K9, K9 + Okay11, Okay11 + Okay13), H2B (K5 + Okay11, Okay11 + Okay12, Okay15 + Okay16, Okay16 + Okay20, Okay20 + Okay23, Okay34 + Okay43, and Okay116 + Okay120), H3 (K9 + Okay14, Okay18 + Okay23, and Okay27 + Okay36), and H4 (K8 + Okay12 and Okay12 + Okay16). The stoichiometry of di-acetylated peptides stays unexplored for H2B, and these peptides could also be extra plentiful than their mono-acetylated counterparts used to calculate stoichiometry on this research. As well as, some acetylated peptides from histone tails will not be detected due to their small measurement. Thus, our estimates of histone acetylation stoichiometry could underestimate the precise native stoichiometry at these positions as a result of these websites additionally happen on di-acetylated peptides or peptides that we’re unable to detect with our methodology. Regardless, our information counsel that the stoichiometry of H2B acetylation ranges from zero.5% to five.6%. N-terminal H2B websites are primarily acetylated by the CBP/p300 acetyltransferases12, which additionally goal H3K27 and H3K36. N-terminal H2B websites (K5, Okay11, Okay12, Okay15, and Okay16) present quicker deacetylation kinetics in comparison with Okay2012. Apparently, the stoichiometry of Okay20 is larger than K5, indicating that the decrease stoichiometry at K5 is probably resulting from its extra fast turnover.
Stoichiometry of histone acetylation. a The diagram reveals all histone acetylation websites whose stoichiometry was decided on this research. The websites are ordered by descending stoichiometry. Observe that prime stoichiometry websites happen on the N-termini of core histones. b The stoichiometry of histone acetylation websites as decided in 4 impartial research17,18,19,20. c An anti-acetylated lysine immunoblot of HeLa complete cell lysate. Cells had been boiled in 2% LDS to make sure histone extraction. Histones are annotated based mostly on their anticipated molecular weight. d Histone acetylation websites represent a majority of acetylated lysine residues in cells. Stoichiometry and protein copy numbers had been used to calculate the variety of acetylated lysine residues for the indicated lessons of proteins. Supply information are supplied as a Supply Information file
Histones are among the most plentiful and most extremely acetylated proteins in cells. Anti-acetylated lysine antibodies prominently detect histone acetylation in western blots of complete cell lysates, suggesting that histones harbor a lot of the acetylated lysine residues in cells (Fig. 4c). Utilizing our acetylated lysine copy quantity estimates we discovered that histone acetylation accounted for 74% of the acetylated lysine residues in cells (Fig. 4d). If we embrace histone websites whose stoichiometry was measured by impartial research17,18,19,20, the fraction of acetylated lysines occurring on histones will increase to 84% (Fig. 4d). These estimates don’t account for a number of websites on H2B (Okay11, Okay12, Okay15, Okay16, and Okay23), in addition to websites on H2A (K4, K5, K7, K9, Okay11, and Okay13). Thus, histone acetylation seemingly accounts for a fair higher proportion of the acetylated lysine residues present in cells.
Regulation by deacetylases
We analyzed the stoichiometry of lysine deacetylase (KDAC)-regulated acetylation websites by evaluating the info obtained on this research with a earlier evaluation of deacetylase inhibitors in HeLa cells21. Class I KDACs (HDAC 1, 2, three, and eight) are particularly focused by the category I inhibitors apicidin, MS-275, valproic acid, and sodium butyrate; the category IIb KDAC HDAC6 is particularly focused by tubacin; and nicotinamide inhibits the exercise of NAD+-dependent Sirtuin deacetylases, however largely affected SIRT1-regulated websites in mammalian cells21. To research class I KDAC regulated websites, we used the median elevated acetylation attributable to apicidin, MS-275, valproic acid, and sodium butyrate.
KDAC inhibitors regulated websites with a broad vary of stoichiometry (Fig. 5a). Class I KDAC inhibitors regulated considerably higher proportions of reasonably elevated (>zero.23%) and excessive stoichiometry (>1%) acetylation in comparison with tubacin or nicotinamide. The stoichiometry of sophistication I regulated websites was considerably greater than not-regulated (NR) websites, whereas the distributions of tubacin and nicotinamide regulated websites was not considerably completely different than NR websites. Moreover, websites that had been most delicate to KDAC inhibitors (>Four× elevated acetylation) confirmed an growing proportion of upper stoichiometry websites for the category I inhibitors, however stayed the identical or decreased for tubacin and nicotinamide (Fig. 5a). Thus, whereas tubacin and nicotinamide regulate a higher portion of the acetylation websites quantified on this research (eight.6% and eight.eight%, respectively) than class I inhibitors (2.6%), the category I inhibitors regulate a higher proportion of upper stoichiometry acetylation.
Stoichiometry of deacetylase- and CBP/p300-regulated acetylation websites. a The class scatterplot reveals the distributions of acetylation websites that aren’t regulated (NR), greater than two-fold (>2×) upregulated, or greater than four-fold (>Four×) upregulated, by the indicated deacetylase inhibitors as decided by21. Class I KDAC inhibitors primarily goal HDACs 1, 2, three, and eight, and had been decided 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, however the regulated websites are largely attributed to SIRT121. b CBP/p300 regulates an growing fraction of excessive stoichiometry acetylation websites. CBP/p300-regulated websites had been decided by12. c Acetylation websites which can be most affected (>eight× decreased) by lack of CBP/p300 exercise have greater median stoichiometry than websites which can be solely modestly affected (2–Four× decreased). Supply information are supplied as a Supply Information file
Regulation by the CBP and p300 acetyltransferases
The homologous Creb-binding protein (CBP)/E1A-binding protein p300 (p300) acetyltransferases are essential regulators of cell-type-specific and signaling-regulated gene expression22. CBP/p300 acetyltransferase exercise is important for selling gene transcription and CBP/p300 targets a big proportion of the acetylome12. CBP/p300-regulated websites constituted 12.7% of the websites analyzed on this research, indicating that CBP/p300 targets multiple out of each ten acetylation websites. CBP/p300 focused an identical proportion (11.5%) of low stoichiometry ( < 0.2%) sites, and an increasing proportion of higher stoichiometry sites, up to 65% of the sites with stoichiometry exceeding 1% (Fig. 5b). Thus, CBP/p300 acetylates a majority of high (>1%) stoichiometry acetylation websites. The stoichiometry of CBP/p300-regulated websites additionally elevated with the diploma of downregulated acetylation within the absence of CBP/p300 catalytic exercise (Fig. 5c), indicating that the websites most affected by lack of CBP/p300 are usually extra extremely acetylated.
Stoichiometry of functionally characterised websites
The exercise and subcellular localization of eukaryotic translation initiation issue 5 A (eIF5A) is regulated by PCAF-catalyzed acetylation at Okay4723. We discovered that eIF5A was greater than 10% acetylated at Okay47, in step with a regulatory function for acetylation at this place. DNA methyltransferase 1 (DNMT1) is acetylated by the TIP60 acetyltransferase, and acetylation promotes ubiquitin-dependent protein turnover24. We discovered that DNMT1 harbored excessive stoichiometry acetylation at Okay335 (zero.57%) and Okay675 (zero.1%). Pyruvate dehydrogenase E1 alpha 1 subunit (PDHA1) is acetylated at Okay321 by acetyl-CoA acetyltransferase 1 (ACAT1), which recruits PDH kinase (PDK) to inhibit pyruvate dehydrogenase exercise25. We discovered that PDHA Okay321 was zero.6% acetylated, supporting excessive stoichiometry acetylation at this place.
Glucose-6-phosphate dehydrogenase (G6PD) is reported to be negatively regulated by KAT9-dependent acetylation of Okay403, and activated by SIRT1-dependent deacetylation26. We discovered that G6PD is simply zero.02% acetylated at Okay403. Acetylation of phosphoglycerate kinase 1 (PGK1) at Okay220 disrupts its exercise by inhibiting binding to ADP27. We discovered that PGK1 is zero.03% acetylated at Okay220 in HeLa cells. PCAF is reported to control cyclin dependent kinase 2 (CDK2) exercise by acetylating CDK2 at Okay33 within the ATP-binding energetic website28. We discovered that CDK2 Okay33 is simply zero.05% acetylated, nevertheless, CDK1 Okay33 was Four.5% acetylated, suggesting that acetylation could play a higher function in regulating CDK1 activity3. In every of the above examples, acetylation is reported to cut back enzymatic exercise, nevertheless, the stoichiometry of acetylation means that acetylation would must be dramatically elevated (100-fold or extra) so as to have a considerable affect on protein perform.
Low stoichiometry acetylation doesn’t essentially point out a scarcity of perform. Reasonably, the mechanism-of-action determines whether or not low stoichiometry acetylation is enough to control protein perform. Acetylation that imparts a gain-of-function or regulates protein exercise at a particular time and/or place, could possibly be regulated by low stoichiometry acetylation. For instance, acetylation at Okay220 regulates the exercise of microtubule related protein RP/EB member of the family 1 (EB1), particularly throughout mitosis and at spindle microtubule plus ends29. Though we discovered that EB1 Okay220 was solely zero.02% acetylated, the mechanism-of-action is in step with the low noticed stoichiometry of modification.