Jordi C. J.
Hintzen‡
a,
Jona
Merx‡
b,
Marijn N.
Maas
a,
Sabine G. H. A.
Langens
b,
Paul B.
White
b,
Thomas J.
Boltje
*b and
Jasmin
Mecinović
*a
aDepartment of Physics, Chemistry and Pharmacy, University of Southern Denmark, Campusvej 55, 5230 Odense, Denmark. E-mail: mecinovic@sdu.dk; Tel: +45 6550 3603
bInstitute for Molecules and Materials, Radboud University, Heyendaalseweg 135, 6525AJ, Nijmegen, The Netherlands. E-mail: thomas.boltje@ru.nl; Tel: +31 24 3652331
First published on 29th November 2021
Histone lysine methyltransferases and acetyltransferases are two classes of epigenetic enzymes that play pivotal roles in human gene regulation. Although they both recognise and posttranslationally modify lysine residues in histone proteins, their difference in histone peptide-based substrates and inhibitors remains to be firmly established. Here, we have synthesised lysine mimics that posses an amide bond linker in the side chain, incorporated them into histone H3 tail peptides, and examined synthetic histone peptides as substrates and inhibitors for human lysine methyltransferases and acetyltransferases. This work demonstrates that histone lysine methyltransferases G9a and GLP do catalyse methylation of the most similar lysine mimic, whereas they typically do not tolerate more sterically demanding side chains. In contrast, histone lysine acetyltransferases GCN5 and PCAF do not catalyse acetylation of the same panel of lysine analogues. Our results also identify potent H3-based inhibitors of GLP methyltransferase, providing a basis for development of peptidomimetics for targeting KMT enzymes.
Fig. 1 Histone lysine methylation and acetylation. (a) KMT-catalysed lysine methylation in the presence of SAM cosubstrate and KAT-catalysed lysine acetylation in the presence of AcCoA cosubstrate; (b) view from a crystal structure of GLP (cyan) complexed with H3K9me1 (yellow) and S-adenosylhomocysteine (SAH, orange) (PDB ID: 3HNA); (c) view from a crystal structure of GCN5 (gray) complexed with H3K14 (yellow) and coenzyme A (CoASH, orange) (PDB ID: 1Q2C); (d) the panel of lysine analogues incorporated in the H3 peptide for examination of KMT and KAT catalysis and inhibition. |
Since both classes of epigenetic enzymes are of crucial importance to the function of the cell and show a specificity towards the lysine site that undergoes PTM, gaining a deeper understanding of their mode of action and substrate scope is highly desirable. Furthermore, both KMTs and KATs are known to be involved in many disease phenotypes, including cancer.11,12 Previously, both classes of enzymes have been successfully subjected to a wide variety of lysine analogues to assess which chemical changes are tolerated by the two families of enzymes and potentially leading to the discovery of novel inhibitors.13–25 KATs were shown to be more promiscuous than KMTs towards the lysine's side chain length,16,24 while γ-difluorolysine was accepted as an excellent substrate for both families of enzymes.14 Adding to this body of work, here we have developed an expanded set of lysine analogues, containing an amide bond in the side chain, which introduce some degree of steric bulk and rigidity into lysine's δ-position (Fig. 1d). A set of seven lysine analogues was chemically synthesised, incorporated into histone 3 tail peptides and subsequently evaluated as substrates for methyltransferases G9a and GLP, as well as the acetyltransferases GCN5 and PCAF, which catalyse the methylation and acetylation of lysine K9 of histone H3, respectively.
The set of synthesised Fmoc/Boc-protected lysine analogues, including natural lysine as a control, was subsequently incorporated into histone H3 tail peptides (residues 1–13, ARTKQTAR9STGG) by standard Fmoc-SPPS. The synthetic peptides were then purified by reverse-phase HPLC and their purity was confirmed by MALDI-TOF MS and analytical HPLC (Fig. S1–S8†). The potential of the histone peptides possessing lysine analogues being substrates for KMTs and KATs was evaluated by carrying out enzymatic assays and evaluation by MALDI-TOF MS.
H3 peptides (100 μM) were incubated with the recombinantly expressed human GLP or G9a (2 μM) in the presence of SAM as the cosubstrate (500 μM) for 3 hours at 37 °C, while conversion was also checked after 1 hour, according to previously reported standard conditions.14 This revealed predominant dimethylation and trimethylation of the H3K9 peptide already after 1 hour for both KMTs, while the reaction was also found to be dependent on both the presence of the enzyme and SAM (Fig. S9†). The H3K9 peptide was found to be fully trimethylated after 3 hours (Fig. 3a). Among the panel of lysine analogues, only the H3 peptide containing KGly (H3KGly9) was found to be accepted as a GLP/G9a substrate (Fig. 3b and S10†).
Interestingly, only dimethylation was observed for the analogue under standard conditions, reaching completion after 3 hours. This finding might indicate that while the analogue is well accommodated in the KMTs binding pocket, an extra barrier exists to reach the final trimethylated product, as found for lysine. This can be attributed to the limited flexibility of lysine's side chain in the case of H3KGly9 due to introduction of the amide functionality and the increased steric demand of the two Nε-methyl groups present. Previously, it was shown that other analogues also only reach the dimethyl stage when reacted with GLP and G9a.15,18,19 To assess this finding, H3KGly9 (20 μM) was incubated with GLP (2 μM) and SAM (500 μM), to obtain a lower ratio between peptide and substrate. Under these conditions, dimethylation was readily observed and traces of trimethylation were detected (Fig. S11†). Then, H3KGly9 was incubated with a higher concentration of GLP (10 μM) and cosubstrate (1 mM). This showed that at higher enzyme and cosubstrate concentration, GLP does have the ability to predominantly catalyse trimethylation of H3KGly9 (Fig. S12†). This result further suggests the extra barrier for trimethylation, indicating that a higher enzyme concentration plays a role in pushing the reaction to completion. Meanwhile, all other analogues, derived from amino acids that contain some degree of steric bulk in their side chains, were not methylated, within the limits of detection (Fig. 3c–h and Fig. S10†). Seeing that higher methylation states can be reached for H3KGly9 under optimised conditions, all other lysine analogues were incubated under these conditions (Fig. S11 and S12†). It was found that for all analogues, traces of monomethylation and dimethylation could be observed in varying degrees, with high enzyme concentration (10 μM) resulting in slightly higher conversion, as already observed for H3KGly9. Specifically, peptides H3KL-Ala9, H3KL-Abu9 and H3KD-Abu9 were found to be dimethylated to a minor degree. These results strongly indicate that while the presence of the amide bond in the side chain itself is tolerated, introduction of additional steric bulk at the ε-position leads to a loss of enzymatic methylation activity. To evaluate and compare the progression of the methylation of H3KGly9 (100 μM) at 2 and 10 μM of GLP, a timecourse experiment was performed under previously stated conditions, taking aliquots of the enzymatic reactions at several time points, and plotting the methylated species over time (Fig. S13†). A clear increase of higher methylation states was observed over time, at both 2 μM and 10 μM GLP, with a trimethylated product only evident in the latter case (Fig. S13†).
Similarly, to evaluate KAT-catalysed acetylation, H3 peptides (100 μM) were incubated with human GCN5 or PCAF (2 μM) and AcCoA as the cosubstrate (300 μM) for 3 hours at 37 °C.14 Full conversion of H3K9 to H3K9ac was reached already after 1 hour, with the dependence on the enzyme and cosubstrate proven in controls (Fig. 4a and S9†). Interestingly, none of the H3 peptides bearing lysine analogues was accepted as a substrate for KATs within the detection limit of our assay, suggesting that introduction of the amide functionality in the side chain is not tolerated by these representative KATs (Fig. 4 and S14†). Increasing the concentration of GCN5 and PCAF to 10 μM resulted in a trace of acetylation for H3KGly9 (Fig. S15†), whereas other lysine analogues were not acetylated within the limit of detection.
Having established H3KGly9 as a substrate for both GLP and G9a, we set out to evaluate the kinetic parameters of H3KGly9 in comparison to the H3K9 peptide using MALDI-TOF MS assays. To obtain an accurate representation of the kinetic profile of substrate methylation, it must be ensured that the lysine and its mimic are only monomethylated during the enzymatic reaction. Therefore, previously reported conditions were employed, using GLP/G9a (100 nM), SAM (5 μM) and a varying concentration of the H3K9 peptide between 5 and 200 μM.14 All reactions were carried out for 5 minutes at 37 °C. However, under these conditions, methylation of the H3KGly9 peptide proved to be too low for accurate determination of conversion, and therefore the enzyme concentration was increased to 400 nM. This revealed that both H3 peptides behaved similarly for both G9a and GLP (Fig. 5, Table 1). Furthermore, H3KGly9 was found to have about 4- to 5-fold decrease in enzyme efficiency (kcat/KM) in comparison to lysine. The loss in activity can be attributed to both an increased KM and a decreased kcat, unsurprisingly, the same pattern is observed for both enzymes.
Fig. 5 Kinetic plots of (a) G9a-catalysed methylation of H3K9 (black) and H3KGly9 (red) and (b) GLP-catalysed methylation of H3K9 (black) and H3KGly9 (red). |
k cat (min−1) | K M (μM) | k cat/KM (μM−1 min−1) | |
---|---|---|---|
GLP | |||
H3K9 | 1401 ± 79 | 19.1 ± 3.8 | 73.4 |
H3KGly9 | 687 ± 44 | 33.7 ± 6.1 | 20.4 |
G9a | |||
H3K9 | 1042 ± 58 | 12.7 ± 3.1 | 82.0 |
H3KGly9 | 624 ± 55 | 38.5 ± 9.2 | 16.2 |
To investigate the methylation of H3KGly9 by GLP and further validating it as a substrate, we examined the enzymatic reactions by NMR spectroscopy (Fig. 6). The H3KGly9 peptide (400 μM) or unmodified H3K9 (400 μM) as a control were incubated with GLP (8 μM) and SAM (2 mM) in Tris-d11 buffer (pD = 8.0) for 1 h at 37 °C. The samples were then transferred to an NMR tube and subjected to 1H and 1H–13C HSQC NMR experiments. The conversion of the H3K9 control resulted in an appearance of a resonance at 3.04 ppm (NMe3) corresponding to H3K9me3 with a corresponding resonance of 52.6 ppm for 13C (NMe3), in line with previous work (Fig. 6a and c).14 For the H3KGly9 peptide, under the employed conditions, a resonance appeared at 3.23 ppm (NMe3) with a corresponding 13C resonance of 54.6 ppm (NMe3), indicating trimethylation (Fig. 6b and d). The conversion of SAM to SAH, with the appearance of a triplet at 2.61 ppm (SAH-CH2γ), was observed in both reactions (Fig. 6a and b). The relative higher enzyme concentration used in the NMR experiments and the long measurement time (22 h) can result in the complete formation of H3KGly9me3.
Peptides that were not found to be accepted as enzyme substrates, were then evaluated for potential inhibition of both lysine methylation and acetylation reactions. A single point screen was carried out where the analogue (10 or 100 μM) was preincubated with GLP (200 nM) and SAM (100 μM) for 5 minutes, after which H3K9 (10 μM) was added and reacted for the additional 30 minutes (Fig. 7 and Fig. S16†). Interestingly, peptides H3KD-Abu9, H3KL-Phe9 and H3KD-Phe9 showed residual enzyme activity well below 50% at 10 μM of inhibitor already, while H3KL-Abu9 showed a residual activity around 50%, suggesting strong inhibition by these amino acid side chains (Fig. 7). Out of these peptides, the former three in particular showed good inhibition at this concentration, suggesting that KMTs can accommodate the bulkier side chain containing lysine analogues, but are prevented from an efficient methyl transfer. To gain a more detailed insight into their inhibitory properties, IC50 values were determined for H3KL-Abu9, H3KD-Abu9, H3KL-Phe9 and H3KD-Phe9 and were found to be between 21.6 and 1.29 μM (Table 2 and Fig. S17†). Previously, an H3 peptide bearing a meta-aminophenylalanine residue at position 9 was found to have an IC50 of 26 μM.18 H3KL-Abu9 was found to have a similar IC50 value, but our most potent inhibitor, H3KD-Abu9, has a 20-fold stronger inhibition (IC50 = 1.29 μM), pushing the limits of peptide-based inhibitors of GLP.
Fig. 7 Single point inhibition of GLP (200 nM) by H3K9 peptides (10 μM). Error bars reported as standard error (SE), experiments carried out in duplicate. |
Peptide | IC50 (μM) | Peptide | IC50 (μM) |
---|---|---|---|
H3KL-Abu9 | 21.6 | H3KL-Phe9 | 1.74 |
H3KD-Abu9 | 1.29 | H3KD-Phe9 | 7.17 |
A single point screen was also carried out to assess GCN5-catalysed lysine acetylation by preincubating a competing histone peptide (10 or 100 μM) with GCN5 (500 nM) in the presence of AcCoA (100 μM) for 10 minutes. Then the H3K9 substrate (10 μM) was added to the enzymatic mixture and left for 30 minutes (Fig. S18†). Peptides H3KL-Abu9 and H3KD-Abu9 were not evaluated for acetylation inhibition due to overlapping signals in the MS analysis. These results revealed that even at 100 μM of H3 peptides present, no reduced acetylation activity was observed (Fig. S18†). This result confirms that H3 peptides possessing lysine analogues not only are non-substrates, but even do not seem to bind to the active site of GCN5 to be able to compete with H3K9 binding. These results are in line with recent examinations of other lysine analogues as substrates and inhibitors of KATs, as reflected by high KM values and poor inhibitory potency.16,17,20
Computational analysis was then performed to gain insight into potential binding conformations of the lysine analogue-containing H3 peptides and to shed light upon the observed selectivity for GLP-mediated methylation of H3KGly9 (Fig. 8 and Fig. S19†). Dockings of H3K9 peptides using the X-ray structure of GLP-H3K9me2-SAH ternary complex (PDB-ID: 2RFI26) with mildly restrained peptide backbones illustrate that the modified H3KGly9 side chain adopts a similar binding pose to substrate H3K9 (Fig. 8a). The remaining analogues were similarly observed to position their side chains in conformations comparable to H3K9 (Fig. 8b). Unlike H3KGly9, however, none of the other lysine analogues seem to position its ε-amino group towards the S atom of SAH (the product of SAM-mediated methylation reaction), with the introduced aliphatic side chains of Ala, Abu and Phe being positioned towards SAH instead (Fig. S19†). Interestingly, the introduction of an amide to the analogue side chains potentially allows for the formation of additional hydrogen bonding interactions with the enzyme's L1143 backbone in a β-sheet like manner, while for both H3KL-Phe9 and H3KD-Phe9 binding is enhanced through π–π stacking with Y1221 of GLP, presumably contributing to high potency of both peptides (Fig. 8b and Fig. S19†). Similarly, the amine group in H3KL-Abu9 points away from the active site Y1221, while the amine in H3KD-Abu9 directly faces this residue, indicating that the protonated Nε-amino group can interact with the aromatic ring of Y1221 via favourable cation–π interactions. This difference possibly gives rise to a stronger binding interaction with the enzyme and the improved IC50 observed for H3KD-Abu9 (Fig. S20†).
Fig. 8 Docking poses of H3 peptides possessing lysine analogues (residues 6–11) in GLP (PDB-ID: 2RFI): (a) overlay of substrates H3K9 (yellow) and H3KGly9 (green), (b) overlay of inhibitors H3KD-Abu9 (yellow) and H3KL-Phe9 (green). SAH is shown in blue. |
Plasmid carrying recombinant His-tagged Human GCN5 catalytic domain (residues 497–662 in pET28a-LIC vector) was purchased from Addgene (25482). The protein was expressed and purified as previously described.28 Plasmid carrying recombinant SNAP-His-tagged Human PCAF catalytic domain (residues 498–658 in pET16b vector), kindly provided by Professor Milan Mrksich (Northwestern University), was expressed and purified as previously described.17 Purity of the eluted proteins was assessed with SDS-PAGE, and was separated in aliquots, rapidly flash-frozen and stored at −80 °C.
Histone lysine methyltransferase kinetics studies were performed as described.24 A solution of histone peptide (0–200 μM), was added to a solution of SAM (5 μM) in assay buffer (50 mM Tris, pH 8.0) at room temperature (final volume of 30 μL). The reaction was then initiated by the addition of GLP or G9a (100 nM) and shaken for 5 min. Reactions were incubated at 37 °C, shaken at 750 rpm and quenched with TFA 10% in Milli-Q water at different time points, within linear production of monomethylated peptides extrapolated from timecourse plots. All reactions were aliquoted and mixed 1:2 with a solution of α-cyano-4-hydroxycinnamic acid (CHCCA) in 1:1 MQ and ACN (0.1% TFA) and loaded onto a MTP 384 polished steel target to be analysed by UltrafeXtreme-II tandem mass spectrometer (Bruker). The amount of methylated peptide was calculated by integration of the product peak area and divided it by the amount of unmodified peptide, taking in account all the ionic species, at any concentration points using the FlexAnalysis software. Kinetic values were extrapolated by fitting V0 values (μM of produced peptide per minutes) and histone peptide concentrations to the Michaelis–Menten equation using the GraphPad Prism 5 software. Kinetic experiments were carried out in duplicates and final values are reported as value ± SD.
Footnotes |
† Electronic supplementary information (ESI) available: Synthesis, peptide synthesis, MALDI-MS enzyme assays, NMR spectra, docking. See DOI: 10.1039/d1ob02191e |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2022 |