Effect of lysine side chain length on histone lysine acetyltransferase catalysis

Histone lysine acetyltransferase (KAT)-catalyzed acetylation of lysine residues in histone tails plays a key role in regulating gene expression in eukaryotes. Here, we examined the role of lysine side chain length in the catalytic activity of human KATs by incorporating shorter and longer lysine analogs into synthetic histone H3 and H4 peptides. The enzymatic activity of MOF, PCAF and GCN5 acetyltransferases towards histone peptides bearing lysine analogs was evaluated using MALDI-TOF MS assays. Our results demonstrate that human KAT enzymes have an ability to catalyze an efficient acetylation of longer lysine analogs, whereas shorter lysine analogs are not substrates for KATs. Kinetics analyses showed that lysine is a superior KAT substrate to its analogs with altered chain length, implying that lysine has an optimal chain length for KAT-catalyzed acetylation reaction.

Scientific RepoRtS | (2020) 10:13046 | https://doi.org/10.1038/s41598-020-69510-0 www.nature.com/scientificreports/ which is highly conserved among the GNAT and MYST families, or aspartic acid in the case of p300, to date different catalytic mechanisms have been proposed among KATs 20,21 . The GNAT family members were found to catalyze a direct acetyl transfer from the AcCoA cosubstrate to the nucleophilic N Ɛ -amino group of lysine upon constitution of a ternary complex comprising of KAT-AcCoA-Histone 20,21 . Members of the MYST family are instead reckoned to catalyze acetylation of histone substrates through a ping-pong mechanism. Combined crystallographic, kinetics, mutagenesis and thermodynamic evidences support that the acetyl transfer mechanism in MYST enzymes is mediated via the formation of the S-acetylated cysteine intermediate in the catalytic pocket that upon reaction with lysine produces the acetyllysine product [22][23][24] . Although several KAT crystal structures have been determined, only a few examples include ternary complexes with both histone substrates and AcCoA cosubstrate (or CoASH product), contributing to a limited understanding of KATs' catalytic properties (PDB ID: 2P0W and 1QSN, Fig. 1b,c). Due to their indisputable importance in tuning gene expression, not surprisingly many KATs have been found overexpressed and/or dysregulated in several human pathologies, such as cancer, inflammation, and neurodegenerative disorders [25][26][27][28] . In the past decade, several inhibitors targeting this biomedically important family of epigenetic enzymes have been developed, many of them lacking inter and intra families selectivity as well as displaying poor activity 29,30 . Given the challenges encountered in the development of KATs inhibitors, a better understanding of the substrate scope for KAT catalysis is of great biomolecular and medicinal interest. Exploring the substrate and cosubstrate specificity of KAT-catalyzed reactions is an important tool that can provide valuable information for the rational design of new inhibitors 31 . Towards this aim, extensive work has been carried out to unravel the specificity of the AcCoA cosubstrate in KAT catalysis 32,33 . KATs have been found to catalyze the transfer of sterically more demanding acyl moieties, including propionyl, crotonyl, and butyryl, both in vivo and in vitro, from their respective acyl CoA cosubstrate, with differences in cosubstrate acceptance   [37][38][39] . Although in the recent years the investigation of epigenetic enzymes' susceptibility towards lysine chemical modification gathered great attentions, especially with regards to histone lysine methyltransferases (KMTs), histone lysine demethylases (KDMs) and KDACs, we are currently lacking basic molecular knowledge in the context of histone lysine acetylation [40][41][42][43][44][45][46][47] . Enzyme kinetics showed p300 catalytic efficiency towards a collection of Lys analogs; a synthetic H4K8 peptide was found to be the preferred substrate, followed by D-Lys and γ-thiahomolysine that showed very poor catalytic efficiency 48 . More recently, γ-thialysine was demonstrated to be efficiently accepted by a panel of human KATs for the transfer of acetyl, propionyl and butyryl moieties from their respective acyl CoA cosubstrates 49 . To provide a better insight into the role of lysine side chain on histone acetylation, this study is aimed at understanding of the relevance of lysine side chain length on human KAT catalysis.

Results and Discussion
To investigate the role of the length of lysine's side chain on human KAT catalysis, a panel of histone peptides bearing lysine analogs that differ in the chain length was developed via Fmoc-based Solid-Phase Peptide Synthesis (SPPS) (Fig. 2). While several unnatural lysine analogs with shorter and longer side chain are commercially available (Dab = two carbons; Orn = three carbons; Lys; hLys = five carbons), novel longer analogs were synthesized in three steps employing a cross-metathesis synthetic strategy (Fig. 3a). Hoveyda-Grubbs II generation catalyst was employed to perform the coupling of Fmoc-allylglycine with synthetic diBoc-protected amino alkenes of variable alkyl chain lengths. The resulting cis/trans mixture was subsequently reduced by hydrogenation in the presence of Pd/C to yield the desired Fmoc-protected lysine analogs (Fig. 3a). Three different human KATs were selected and expressed in E. coli: (i) MOF (KAT8), a member of the MYST family, which preferentially catalyzes acetylation of K16 on the histone H4 tail; and (ii) GCN5 (KAT2A) and PCAF (KAT2B), members of the GNAT family, which predominantly catalyze acetylation of K14 and secondarily of K9 on the histone H3 tail 50,51 . To obtain interpretable data, we decided to exclude secondary sites of acetylation from the histone H3 peptide sequence. Therefore, we employed a truncated 15-mer H4 peptide (residues 13-27, sequence: GGA K 16 RHRKVLRDNIQ) as a reference sequence to study MOF-catalyzed acetylation of H4K16, and two different 15-mer H3 peptides, with alternatively synthetically acetylated K9 and K14 on H3 peptide, to investigate the acetylation of both sites by GCN5 and PCAF (residues 3-17, sequence: TKQTARKacSTGG K 14 APR; residues 1-15, sequence: ARTKQTAR K 9 STGGKacA). Fmoc-protected lysine analogs containing a modified chain length were incorporated into the key positions of H3 and H4 peptides by SPPS (Fig. 3b,c). Synthetic histone peptides were purified by RP-HPLC, and their identity and purity were assessed by ESI-MS and analytical HPLC (Supplementary Table 1 and Supplementary Figs. 1-24). MALDI-TOF MS enzymatic assays were carried out to investigate KAT-catalyzed acetylation of synthetic histone peptides. The enzymatic activity of the recombinantly expressed human KATs was measured at different time points (2 µM KAT enzyme, 100 µM histone peptide, 300 µM AcCoA, buffer: 50 mM Hepes, 0.1 mM EDTA, 1 mM DTT, pH = 8.0, 37 °C, hereafter used as standard conditions) (Fig. 3b,c). MOF catalyzed full acetylation of H4K16 in 30 min, whereas GCN5 and PCAF catalyzed quantitative acetylation of H3K14 in 15 min and H3K9 only after 2-3 h (Fig. 4b, h and Supplementary Figs. [25][26][27][28]. It is noteworthy that diacetylation of histone peptides was not detected under our assay conditions, due to preacetylation of the secondary site on H3. Two control experiments in the absence of KAT and AcCoA showed no acetylation within detection limits, confirming the dependence of the acetylation reaction on the KAT enzymes and AcCoA . As further control, H4 and H3 peptides containing stereochemically inverted D-lysine were synthesized. MALDI-TOF data revealed that histones bearing D-lysine are not substrates for our panel of KAT enzymes under standard conditions . These findings demonstrate that the lysine's L-stereochemistry is an essential chemical prerequisite for MOF, GCN5 and PCAF catalysis. Interestingly, p300, one of the most promiscuous KATs, has been observed to catalyze acetylation of the H4K8 histone peptide containing D-Lys (D-H4K8) 48 . Kinetics data showed that D-H4K8 has a very poor catalytic efficiency relative to H4K8 (76-fold drop in V/K with respect to the H4K8 peptide), a result that may lay in the marked structural differences between p300 and the other KAT families 19 . In a broader context, D-Lys has been found to be a very poor substrate for functionally  [34][35][36][37][38][39][40][41]. In contrast, we observed KAT-catalyzed acetylation of longer side chain analogs (hK, h 2 K, h 3 K), though in different degrees among the three enzymes. MOF-catalyzed acetylation of H4hK16 and H4h 2 K16 was found to be almost quantitative (95% and 90%) after 3 h (Fig. 4c,d). On the other hand, GCN5/PCAF catalytic activity towards the same analogs was generally found reduced. Acetylation of the H3hK14 peptide led to a formation of 83% of H3hK14ac by PCAF and 62% of H3hK14ac by GCN5 after 3 h ( Fig. 4i, Supplementary Fig. 42), whereas the H3h 2 K14 peptide was found to be less efficiently acetylated after 3 h (55% by PCAF and 25% by GCN5) (Fig. 4j, Supplementary Fig. 43). Following the same pattern shown for H3K14, the unnatural analogs introduced at position 9 of H3 underwent less efficient acetylation in the presence of PCAF and GCN5. After 3 h, PCAF and GCN5 were only able to catalyze partial acetylation of H3hK9, producing 50% and 30% of H3hK9ac, respectively (Supplementary Figs. [44][45], whereas the H3h 2 K9 peptide was acetylated to a lesser degree (33% and 14%; Supplementary Figs. [46][47]. For all three KATs, h 3 K-containing histone peptides were observed to be very poor substrates, with conversion as low as 12% for MOF after 3 h (Fig. 4e), and traces (< 5%) for PCAF (Fig. 4k) and GCN5 at H3K14 (Supplementary Fig. 48). No acetylation was observed within limits of detection for the H3h 3 K9 peptide in the presence of PCAF and GCN5 . Interestingly, no acetylation of h 4 K-containing peptides was detected either with MOF (Fig. 4f), PCAF ( Fig. 4l and Supplementary Fig. 51) or GCN5 (Supplementary Figs. 52-53) on H4 and H3, establishing the maximal lysine side chain length to be suitable for KAT catalysis to seven carbons. Our results on KAT-catalyzed acetylation of lysine analogs possessing extended side chains provide an interesting finding. Firstly, to date the only longer lysine analog found to be acetylated by p300 has been γ-thiahomolysine, which differs from the fully carbon-based homologous hLys in basicity, side-chain angle and bond length. Secondly, the extremely stringent chain length requirements regulating the activity of other lysine-modifying epigenetic enzymes, such as KMTs and KDACs, are not a prerequisite for KAT enzymes 42,46 . In this context, trimethyllysine-binding epigenetic reader proteins appear to tolerate the alterations of the side chain length to a higher degree than the enzymes that install the methyl group on lysine 52 .
To further assess MOF, GCN5 and PCAF substrate specificity towards our library of histone peptides possessing lysine and its longer side chain analogs, MALDI-based kinetics investigations were carried out under steady-state conditions. Kinetic analysis of bi-substrate enzymes, such as KATs, is reckoned to be challenging www.nature.com/scientificreports/ and dependent on many factors. Firstly, given the nature of these enzymes, the binding of one of the two substrates directly influences the binding of the other (with a direct effect on the Michaelis-Menten parameter K m ). Secondly, as suggested by Wapenaar and Dekker, the catalytic mechanism of KATs seems to be dependent on the nature of the enzyme construct (whether catalytic domain, full-length or belonging to multi-subunit complexes), as in the case for ESA1, which displayed a ternary complex mechanism when integrated in the piccolo NuA4 complex, and the characteristic cysteine mediated ping-pong mechanism when ESA1 catalytic domain alone was employed 24,53,54 . Therefore, to obtain data that would allow us to directly compare the catalytic properties of the selected KATs towards our selection of histone peptides bearing lysine analogs with longer side chains, two strategies to standardize our kinetic evaluation were taken: i) all the experiments were carried out with the same saturating concentrations of AcCoA and ii) only KATs catalytic domains were employed in the study. Overall, the unnatural lysine analogs exhibited a decreased catalytic activity when compared to their natural counterparts, with decreasing k cat values at the insertion of any additional carbon on lysine side chain ( Table 1). Evaluation of MOF kinetic properties revealed that lysine is approximately a 50 times better substrate than its side-chain extended analogs (  www.nature.com/scientificreports/ the side chain length, both in the K14 and K9 peptide series, and for both PCAF and GCN5 (Table 1). This finding, combined with the conserved K m values, revealed that the resulting drops in k cat /K m values towards the side chain extended lysine analogs were mainly a result of a reduced turnover rate. Inserting one carbon (H3hK14) and two carbons (H3h 2 K14) into the lysine side chain caused k cat /K m decreases of 354-and 890-fold for GCN5, and 263-and 625-fold for PCAF, compared to the acetylation of the H3K14 sequence. Examination of the secondary site of acetylation (H3K9) revealed a different trend, where H3hK9 and H3h 2 K9 were found to be more comparable but still poorer substrates to H3K9, with a 15-and 30-fold drop in k cat /K m values for PCAF and fourfold drop for GCN5-catalyzed acetylation of H3hK9. A significant decrease in catalytic efficiency may be the result of substrates clashing in the active site, driven by the loss of a catalytically required water molecule 55 . Previous work showed that GCN5 preferentially acetylates K14 over K9 in wild-type histone 3 with a ([k cat /K m ] K14 /[k cat /K m ] K9 ) value of 57 50 . In our study the analysis of H3 substrate selectivity yielded ([k cat / K m ] K14 /[k cat /K m ] K9 ) values 43 for PCAF and 177 for GCN5.
Next, we proceeded by investigating the potential inhibitory activity of the histone peptides that did not act as KAT substrates. For practical reasons peptides bearing D-Lys and h 3 Lys analogs were not included in the evaluation, due to overlapping of MS signals with the respective natural peptide and acetylated product. Therefore, only histone peptides containing Dab, Orn and h 4 Lys were employed in this experiment. To our purpose, we carried out a single-point screening assay, by incubating an equimolar amount (100 µM) of the H4K16/H3K14/H3K9 peptides and the potential inhibitor peptides containing lysine analogs, and quenching the reactions after 1 h at 37 °C. MS analyses revealed that all tested peptides were not able to significantly inhibit KATs. In all experiments the presence of the unnatural histone peptides did not influence the catalytic properties of the enzymes (Supplementary Figs. 54-58), whose residual activity was found to be always greater than 50% when compared to the positive control consisting of the same reaction in the absence of an inhibitor peptide (Fig. 5). From these findings it can be concluded that IC 50 values of histone peptide inhibitors are > 100 µM. A possible explanation of the discovered low inhibitory ability of histone peptides may lay in the low binding affinity (as reflected in high K m values) showed in the enzyme kinetics experiments, which translates into the requirement of higher concentration of the peptides to appreciate competitive inhibition.

Conclusion
Our combined synthetic and enzymatic investigations demonstrate that histone lysine acetyltransferases are special enzymes of the epigenetic machinery that have a capacity to acetylate histone peptides bearing lysine analogs with extended side chain. In line with their well-documented ability to catalyze the transfer of bulkier acyl moieties from their respective acyl CoA cosubstrates, human KATs showed an additional degree of substrate promiscuity by less efficiently accepting as substrates histone peptides bearing lysine analogs with longer side chains. Acetyltransferases MOF, PCAF and GCN5 were observed to catalyze the acetylation of lysine analogs bearing up to 7 carbons in the chain length (hK, h 2 K and h 3 K), both in primary and secondary sites of acetylation. Interestingly, none of the shorter (Dab, Orn) analogs, D-lysine or the longest lysine analog in our panel (h 4 K) were found as KAT substrates, indicating that although KATs alleged substrate promiscuity, the active site dynamics in KATs are still governed by stringent structural prerequisites. We further showed that the development of histone peptide based competitive inhibitors for KATs may suffer of lack of potency due to the relatively low binding affinities of these peptides towards KATs. Overall, this study shows that human KAT enzymes do have a broader substrate scope, as demonstrated by their capacity to catalyze acetylation of lysine analogs with longer side chain length. We believe that a better understanding of KAT catalysis, as demonstrated by basic molecular investigations on the substrate scope and biocatalytic potential reported here, provides an important knowledge for examination of posttranslational modifications of histone proteins and for rational drug design, targeting biomedically important KAT enzymes.

Methods
Synthesis and purification of histone peptides. All the histone peptides were manually synthesized employing standard Fmoc-SPPS chemistry on a 0.05 mmol scale on Wang resing (0.87 g/mmol loading capacity, 100-200 mesh). Upon standard amino acid (3.0 eq) activation with HOBt (3.6 eq) and DIC (3.3 eq) in DMF (final volume 2 mL, mixed for 2 min), the mixture was added to the Fmoc-deprotected growing peptide, and coupling reactions were carried out at room temperature for 2 h. Incorporation of newly synthesized amino acids (1.3 eq) were performed overnight at room temperature, with HOBt (1.56 eq) and DIC (1.43 eq). Fmoc deprotection was achieved by swelling the peptide in a solution of 20% piperidine in DMF (v/v), for 20 min. The qualitative Kaiser Test was employed to monitor the accomplishment of both coupling and deprotection steps. DMF washes (bubbling 3 × 2 min with N 2 ) followed every deprotection and coupling steps. After the incorporation of the last amino acid, the peptides were Fmoc deproteced, dried over Et 2 O and cleaved off the resin employing a 95% TFA, 2.5% TIPS, 2.5% MQ cleavage cocktail for 4 h. After that the crude peptides were precipitated on  www.nature.com/scientificreports/ expressed as it follows: E. coli BL21(DE3) cells enriched with hPCAF plasmid, were cultured in 2xYT growth media supplemented with 100 µg/mL carbenicillin at 37 °C, to an OD 600 of 0.6, upon which expression was induced by addition of IPTG (0.5 mM final) and followed by incubation at 16 °C overnight. Harvested cells were pelleted and re-suspended into 50 mM Tris pH 8.5, 200 mM NaCl, 5 mM β-ME lysis buffer in presence of protease inhibitor cocktail (Roche) and lysate by sonication. The supernatant was incubated with Ni-NTA beads for 2 h at 4 °C. The beads were loaded on a gravity flow column and washed with 50 mM Tris pH 8.5, 200 mM NaCl, 50 mM imidazole, 5 mM β-ME. Subsequently, the protein was eluted with 50 mM Tris pH 8.5, 200 mM NaCl, 300 mM imidazole, 5 mM β-ME and concentrated using a 30 kDa spinfilter device (AMICON, 30 MWCO). The protein was furtherly purified by size-exclusion chromatography (SEC) using the AKTA system, employing a Superdex 75 column equilibrated with 50 mM Tris pH 8.0, 200 mM NaCl, 1 mM DTT at 0.5 mL/min flow speed. The purity of the eluted protein was assessed with SDS-page. Pure fractions were pooled, rapidly flash-frozen and stored at -80 °C.
GCN5. Plasmid carrying recombinant His-tagged Human GCN5 catalytic domain (residues 497-662 in pET28a-LIC vector) was purchased from Addgene (25,482). The protein was expressed and purified as previously described 56 . Briefly, E. coli BL21(DE3) cells enriched with hGCN5 WT plasmid were cultured in TB growth media supplemented with 50 µg/mL kanamycin at 37 °C to an OD 600 of 0.6, upon which expression was induced by addition of IPTG (0.5 mM final) and followed by incubation at 16 °C overnight. Harvested cells were pelleted and re-suspended into 50 mM Na 2 HPO 4 pH 7.5, 500 mM NaCl, 5% glycerol, 1 mM β-ME lysis buffer in presence of protease inhibitor cocktail (Roche, Basel, Switserland) and lysate by sonication. The supernatant was incubated with Ni-NTA beads for 2 h at 4 °C. The beads were loaded on a gravity flow column and washed with 20 mM HEPES-NaOH pH 7.5, 500 mM NaCl, 50 mM imidazole, 5% glycerol, 1 mM β-ME. Subsequently, the protein was eluted with 20 mM Hepes-NaOH pH 7.5, 500 mM NaCl, 250 mM imidazole, 5% glycerol, 1 mM β-ME and the buffer was exchanged to 20 mM Hepes-NaOH pH 7.5, 150 mM NaCl, 1 mM βME by concentration with a 10 kDa spinfilter device (AMICON, 10 MWCO). Purity of the eluted protein was assessed with SDSpage, and GCN5 was separated in aliquots, rapidly flash-frozen and stored at -80 °C.