Kinetic Characterization of Human Histone Deacetylase 8 With Medium-Chain Fatty Acyl Lysine

Abstract

Histone deacetylases (HDACs) catalyze the removal of Ɛ-acetyl-lysine residues of histones via hydrolysis. Removal of acetyl groups results in condensation of chromatin structure and alteration of gene expression by repression. HDACs are considered targets for the treatment of cancer due to their role in regulating transcription. HDAC8 inhibition may be an important anti-proliferative factor for histone deacetylase inhibitors on cancer cells and may give rise to the progression of apoptosis. HDAC8 activity was analyzed with various peptides where the target lysine is modified with medium-chain fatty acyl group. Kinetic data were determined for each p53 peptide substrate. The results suggest that there was HDAC8 deacetylase activity on peptide substrate as well as deacylase activity with acylated peptide substrate variants. HDAC8 inhibition by hexanoic and decanoic acid was also examined. The K_i for hexanoic and decanoic acid were determined to be 2.35 ± 0.341 and 4.48 ± 0.221 mM, respectively.

Introduction

Acetyl groups of histone tails are removed by histone deacetylases (HDACs), while histones are acetylated by histone acetyltransferases (HATs). Deacetylation of histones results in condensation of chromatin structure and repression of gene expression. Decreases in acetylated histone causes silencing of the affected genes and hence, this is linked to development of cancer. Because of this, HDACs increasingly have become significantly important targets for the treatment of cancers. ¹ There are currently several HDAC inhibitors that are undergoing various clinical trial stages to be approved by the Food and Drug Administration (FDA), while several drugs are already approved by the FDA such as vorinostat (suberoylanilide hydroxamic acid or SAHA) and romidepsin. These 2 drugs are approved with indication of cutaneous T cell lymphoma treatment. ² Belinostat, panabiostat, and valproic acid, a short-chain branched fatty acid, have also been approved for various disorders. ³ Studies have also shown inhibition of growth of human lung, colon, and cervical cancer cell lines where HDAC8 has been knocked down. ⁴

There are 4 classes of histone deacetylases. Human class I, II, and IV HDACs are said to be zinc-dependent in that they require divalent zinc ions to catalyze the deacetylase reaction. Class I HDACs include HDAC1, -2, -3, and -8. These are homologs of yeast RPD3.^5,6 HDACs 1-3 require association with large multisubunit corepressor complexes and are considered inactive by themselves. HDAC8 is fully active in and of itself and is the only extensively kinetically characterized HDAC.^{7 -10} HDAC3 was also recently characterized. ¹¹ Class II HDACs are subclassified as class IIa (HDAC4, -5, -7, and -9) and class IIb (HDAC6 and -10) and are homologs of yeast HDA1 protein.^6,12 All members of class IIa can shuttle between the nucleus and cytoplasm. The only class IV deacetylase is HDAC11. ¹³ It is a homolog of yeast HOS3. Class III HDACs are NAD(+)-dependent and are referred to as sirtuins. ¹⁴ This class of HDAC share no sequence similarity with class I and II HDACs and use a different catalytic mechanism. ¹⁵

The crystal structures of HDAC3 and -8 show that each bind 2 monovalent cations (MVCs), likely Na⁺ or K⁺, in addition to the catalytic divalent metal ion.^4,10,16,17 The 2 MVC binding sites have been designated as site 1 and site 2 with site 1 located approximately 7 Å from the divalent catalytic center and site 2 is ⩾20 Å from the divalent metal center.^4,17 The crystal structures of other class I and II human HDACs also bind K⁺ at these same sites in addition to bacterial histone deacetylase-like amidohydrolase.^{18 -20} A study on the effects of varying concentrations of Na⁺ or K⁺ on catalysis of HDAC3 ¹¹ and HDAC8 ⁸ have been reported. Na⁺ concentration was shown to have little effect on catalysis of HDAC8 while a greater effect on catalysis of HDAC8 was observed with potassium.

The crystal structure of HDAC3/SMRT displays an active site Zn(II) and a channel leading to it that is likely obstructed in absence of NCOR1, or SMRT. ¹⁷ However, based on observations of the crystal structure of HDAC8, this channel leading into the Zn(II) active site is always open, which offers explanation for its activity by itself. ⁴ It also has been determined that ethylenediamine tetra-acetic acid (EDTA) can be used to chelate Zn(II) ions of HDAC8 forming apo-HDAC8 and its activity can be recovered by adding Zn(II), Fe(II), or Co(II). ⁷ Also, it was shown that HDAC8 activity was greater with Fe(II)-HDAC8 and Co(II)-HDAC8 and that excess Zn(II) inhibits HDAC8 activity.

The current goal of this study is to carry out enzyme kinetic studies to determine HDAC8 deacetylase and deacylase activity with peptide substrate as well as activity of HDAC8 with differentially acylated lysine residues. It has been reported that SIRT5,^{21 -23} and SIRT2 ²⁴ and SIRT6^25,26 are able to deacylate long chain fatty acylated lysines. The same has been reported for HDAC1, -2,-3, ²⁷ HDAC8, ²⁸ and HDAC11.^{27 -29} Little to no activity was reported for HDAC8 with regard to its ability to deacylate short-, medium-, and long-chain fatty acylated lysine when compared to the nonphysiologically relevant trifluoracetylated lysine substrate. ²⁷ HDAC11 has been shown to deacylate short- and medium chain fatty acyl lysine. ²⁹ In the present study, the kinetic characterization and identification of HDAC8 activity with different peptide substrates will provide invaluable insight for the design of HDAC8-specific inhibitors. The present study seeks to determine the effects of medium-chain fatty acylated lysine on catalytic activity of HDAC8. The results from this study demonstrate that medium-chain fatty acylated lysines (typically categorized as C6-C10) are deacylated by HDAC8. We also report the inhibition constants (K_i) of hexanoic and decanoic acid for HDAC8.

Results and Discussion

Kinetic characterization of HDAC8 with medium chain fatty acyl lysine

HDAC8 was reacted with AcRHKK(acetyl)-AMC, AcRHKK(hex)-AMC, and AcRHKK(dec)-AMC (Figure 1). 1 μM HDAC8 was mixed with various concentrations of compounds shown in Figure 1 and stopped at various time points as described below. The data were plotted using GraphPad Prism to determine K_M, k_cat, and k_cat/K_M values. Michaelis-Menten plots are shown in Figure 2 and the kinetic constants are tabulated in Table 1 calculated using the Michaelis-Menten equation. Based on the data in Figure 2 and Table 1, a 2-fold increase in catalytic activity (k_cat/K_M) is observed for AcRHKK(hex)-AMC in comparison to AcRHKK(acetyl)-AMC. A 3-fold increase in catalytic activity is observed for AcRHKK(dec)-AMC versus AcRHKK(acetyl)-AMC. This data is similar to that previously determined for long-chain fatty acylated lysines where different peptide substrate was used. ²⁸ The K_M value decreased as the acyl group on the lysine of the peptide substrate increased in length suggesting tighter binding of substrate and thereby increasing catalytic efficiency.

Figure 1.

Structures of peptide substrates with AMC (7-amino-4-methylcoumarin) fluorophore-coupled substrates screened for HDAC8 deacylase activity showing differentially modified lysines (red). The peptide sequence RHKK is based on the physiological substrate (p53 379-382) for SIRT1. (a) AcRHKK(acetyl)-AMC is the standard physiological substrate deacetylated by HDAC8, (b) AcRHKK(hex)-AMC, and (c) AcRHKK(dec)-AMC.

Figure 2.

Michaelis-Menten plots of rate (M/second) versus concentration of substrate peptide (M) with standard deviation error bars. 1 μM HDAC8 was mixed with varying concentrations of each substrate in 25 mM Tris pH 8.0, 10 mM KCl with 500 μM EDTA (free acid) at room temperature. (a) AcRHKK(acetyl)-AMC, (b) AcRHKK(hex)-AMC, and (c) AcRHKK(dec)-AMC. Kinetic data are reported in Table 1. For panels (a-c), each experiment was completed in duplicate.

Table 1.

Reactivity of HDAC8 (±SD) with medium-chain fatty acylated lysine. ^a

Peptide substrate	kcat (s−1)	KM (M)	kcat/KM (M−1 s−1)
AcRHKK(acetyl)-AMC	0.0201 ± 0.00930	0.000251 ± 7.75 × 10−5	80.1 ± 12.3
AcRHKK(hex)-AMC	0.0230 ± 3.54 × 10−5	0.000146 ± 2.51 × 10−5	158.0 ± 26.5
AcRHKK(dec)-AMC	0.0322 ± 0.00906	0.000104 ± 2.50 × 10−5	309.1 ± 12.3

^aHDAC8 was assayed as described below. Steady-state kinetic parameters were determined from a fit of the Michaelis-Menten equation to the dependence of the initial rate on the substrate concentration at 1 μM HDAC8 from Figure 2.

HDAC8 inhibition by hexanoic and decanoic fatty acids

Inhibition of HDACs by fatty acids have been a focus of study recently.^11,29,30 For this reason, in addition to an interest in the possibility of product inhibition at high concentrations of substrate in Figure 2b and c, a study of HDAC8 inhibition by hexanoic and decanoic acid was undertaken. Varying concentrations of fatty acid (hexanoic and decanoic acids) were pre-incubated with 1 μM HDAC8 for at least 10 minutes and 1 μM AcRHKK(acetyl)-AMC was added as substrate to initiate the reactions. The inhibition constants (K_i; Table 2) were determined by plotting fractional activity versus concentration of fatty acid and fitted the data using equation (1) (Figure 3). The K_i for hexanoic acid was determined to be 2.35 mM while that of decanoic acid was determined to be 4.48 mM. Hexanoic acid is approximately 2-fold more potent as an inhibitor of HDAC8 than decanoic acid.

Table 2.

Inhibition of HDAC8 (±SD) by hexanoic and decanoic acid. ^a

Inhibitor	Ki (mM)
Hexanoic acid	2.35 ± 0.341
Decanoic acid	4.48 ± 0.221

^aHDAC8 inhibition by hexanoic and decanoic acid was assayed as described below. Inhibition constants (K_i) were determined by fitting equation (1) to the data in Figure 3.

Figure 3.

Inhibition of HDAC8 by hexanoic (a) and decanoic acid (b) with standard deviation error bars. 1 μM HDAC8 was pre-incubated for at least 10 minutes with varying concentrations of either hexanoic or decanoic acid in 25 mM Tris pH 8.0, 10 mM KCl with 500 μM EDTA (free acid) and then mixed with 1 μM AcRHKK(acetyl)-AMC as substrate at room temperature. Initial velocities were determined from fluorescence changes over time, normalized, plotted versus hexanoic (a) or decanoic acid (b) concentration, and fitted using equation (1). For panel (a), each data point was performed in triplicate except for the 2.5 mM point which was performed in duplicate. For panel (b), the 30 mM data point was measured in triplicate while all other data points were performed in duplicate. Inhibition constants for each are reported in Table 2.

The possibility of micelle formation by hexanoic and decanoic acid has also been considered. Hexanoic acid is generally considered to have a critical micelle concentration (CMCs) of 29.3 mM which is above the ranges used in Figure 3a. Several different CMCs have been reported for decanoic acid with 28 mM being the lowest reported value and 102 mM the highest.^{31 -33} It is therefore unlikely that micelle formation interfered with the assay conditions below these reported CMC values. In fact, micelle formation by decanoic acid would likely have caused less inhibition at concentrations above its CMC. In Figure 3b, approximately 99% inhibition is observed above 20 mM decanoic acid.

HDAC8 docking studies

Docking calculation studies were completes using Chem3D with AutoDock interface to position RHKK(hex) and RHKK(dec) substrates within the active site of HDAC8 (PDB: 5D1B) (Figure 4a and b). ³⁴ Docking simulations were carried out using the zinc ion (green; Figure 4) as the centroid of the selection with atoms within 14 Å of the active site Zn(II) selected. The active site zinc was assigned a formal charge of +2. Trichostatin A (TSA) and glycerols bound at the active site in the original crystal structure were deleted prior to docking calculations. The most plausible positioning of peptide substrate for RHKK(hex) and RHKK(dec) are shown in Figure 4a and b, respectively. For RHKK(hex), the positioning with the most favorable binding energy out of 20 positionings is shown in Figure 4a. The calculated free energy of binding (ΔG_binding) was calculated to be −3.67 kcal/mol. For RHKK(dec), the positioning with the most favorable binding energy out of twenty positionings is shown in Figure 4b. The calculated free energy of binding (ΔG_binding) was calculated to be +3.48 kcal/mol. All binding energies calculated for the 20 positionings of RHKK(dec) in HDAC8 were positive though a positive ΔG_binding is common when docking larger compounds. As shown in Figure 3b, ΔG_binding should be negative as binding and inhibition does occur based on this in vitro data. In both docking results, the fatty acyl amido oxygen and nitrogen are positioned to complex with the zinc ion of the HDAC8 active site as expected. The remainder of the fatty acyl chain is positioned within the channel comparable to the lateral tunnel (or foot pocket or acetate release channel) previously reported^29,35,36 for both RHKK(hex) and RHKK(dec). RHKK(hex) and RHKK(dec) was positioned within HDAC8 with the fatty acyl chain bound inside the vertical tunnel but with higher calculated binding energy.

Figure 4.

AutoDock results for HDAC8 (PDB: 5D1B, red ribbon) ³⁴ docked with RHKK(hex) (panel (a)), RHKK(dec) (panel (b)), hexanoic (panel (c)) and decanoic (panel (d)) fatty acids. Panel (a): RHKK(hex) substrate is shown in ball-and-stick configuration. Panel (b): RHKK(dec) substrate is shown in ball-and-stick configuration. Panel (c): the top 20 most energetically favorable positioning of hexanoic acid inhibitor is shown in stick configuration. Panel (d): the top 20 most energetically favorable positioning of decanoic acid inhibitor is shown in stick configuration. Color scheme for peptide substrate and fatty acid inhibitors: red = oxygen; blue = nitrogen; gray = carbon; white = hydrogen. Two bound potassium ions are shown in purple.

Docking calculations for binding of hexanoic (Figure 4c) and decanoic (Figure 4d) acid inhibitors were also completed. Figure 4c and d show the 20 most energetically favorable positionings of each fatty acid within HDAC8. All binding energy calculations for the 20 positions of hexanoic acid within HDAC8 (Figure 4c) were favorable and ranged from −2.78 to −3.17 kcal/mol. All binding energy calculations for the 20 positions of decannoic acid within HDAC8 (Figure 4d) were favorable and ranged from −3.75 to −4.68 kcal/mol. For both hexanoic and decanoic acids, the positions calculated were almost exclusively within the lateral tunnel (or foot pocket) of HDAC8. Otherwise, they were positioned in the HDAC8 active site or active site tunnel. The carboxylate group in each case was not positioned in complex with the active site zinc.

Materials and Methods

Materials

All peptides were synthesized by GenScript with >95% purity determined by HPLC-MS. Hexanoic acid (⩾95%) was purchased from Acros Organics. Decanoic acid (⩾95%) was purchased from Tokyo Chemical Industry. >90% purity (by SDS-PAGE) human HDAC8 His-Tag in 25 mM HEPES pH 7.5, 300 mM NaCl, 5% glycerol, 0.04% triton X-100, and 0.2 mM TCEP was purchased from Active Motif. Enzyme came from the same lot number (lot # 32016001). The recombinant enzyme consists of full length human HDAC8 (accession number NP_060956.1) with an N-terminal 6xHis tag expressed in a baculovirus expression system. All buffers used in this study were treated with chelex resin prior to use in the enzyme assays.

Metal-free Apo-HDAC8 preparation

⩾90% pure HDAC8 was prepared by dialyzing against 25 mM MOPS, 10 µM dipicolinic acid, and 1mM EDTA at pH 7.0 overnight with 2 exchanges of buffer followed by overnight dialysis with 25 mM MOPS at pH 7.5 without EDTA. HDAC8 was reconstituted with enzyme dilution buffer consisting of 25 mM Tris at pH 8.0, 1 mM NaCl, 10 mM KCl, and 2.34 μM ZnCl₂ and incubating on ice for 1 hour.

HDAC8 activity assay

The deacylase/deacetylase activity of HDAC8 was measured using the commercially available Fluor de Lys HDAC8 assay kit from Enzo Life Sciences as previously described.^7,8,11,27 Briefly, the assay uses a fluorophore-coupled method for indirectly determining deacylation/deacetylation of lysine peptide substrate by employing an AMC (7-amino-4-methylcoumarin) label (Figure 1a-c) that is protected while the adjacent lysine is still acylated/acetylated. After deacylation/deacetylation, the AMC label is cleaved using a developer solution. Enzyme assays were performed in 96-well plates and reactions were stopped at varying time points using Developer II solution containing 1 μM TSA (an HDAC8 inhibitor). ³⁴ AMC fluorescence as an indicator of deacetylation/deacylation was measured using a Tecan 96-well plate reader with excitation and emission wavelengths of 360 and 460 nm, respectively. The concentration of product at each time point was calculated from a standard curve prepared using solutions containing known concentrations of the AMC product (0-40 μM). Except for the inhibition study, all assays were performed in at least duplicate and in 25 mM Tris pH 8.0, 10 mM KCl with 500 μM EDTA (free acid) at room temperature.

For medium-chain fatty acid inhibition assays, varying concentrations of fatty acid (hexanoic and decanoic acids) were pre-incubated with 1 μM HDAC8 for at least 10 minutes and 1 μM AcRHKK(acetyl)-AMC was added as substrate to initiate the reactions. In all assays, the final NaCl concentration contributed by the enzyme storage buffer is ⩽6 mM. The inhibition constant (K_i) was determined by plotting fractional activity versus [fatty acid] and fitting the data using equation (1).

$$\frac{v_i}{v_o}=\frac{V_{max}}{1+\frac{[I]}{K_i}}$$ (1)

Data analysis for all kinetic and inhibition data was performed using GraphPad Prism 12.0. Each experiment, whether completed in duplicate or triplicate, was performed as a separate trial with error bars shown in Figures 2 and 3 generated by the GraphPad Prism 12.0 software. Numerical ranges of the standard deviation error bars are shown in Tables 1 and 2.

HDAC8 docking studies

Docking experiments showing HDAC8 (PDB: 5D1B, red) ³⁴ docked with RHKK(hex), RHKK(dec), hexanoic and decanoic fatty acids. Experiments were performed using Chem3D with AutoDock interface.^37,38 Docking simulations were carried out using the zinc ion as the centroid of the selection with atoms within 14 Å of the active site Zn(II) selected. The active site zinc was assigned a formal charge of +2. TSA bound at the active site in the original crystal structure was deleted. RHKK(hex) and RHKK(dec) were docked using X, Y, and Z number of grid points set to 60 where the spacing of grids was set to 0.375 Å. Hexanoic and decanoic acid were docked using X, Y, and Z number of grid points set to 40 where the spacing of grids was set to 0.375 Å.

Conclusions

HDAC8 had activity with physiological peptide substrate (AcRHKK(acetyl)-AMC) based on p53 sequence. AcRHKK(dec)-AMC had the highest catalytic reaction efficiency per quantity substrate assayed with the highest k_cat/K_M among substrates tested. Hexanoic acid exhibited greater HDAC8 inhibition compared to decanoic acid with a K_i that is approximately 1-half of that determined for decanoic acid. Further HDAC8 studies of enzyme kinetics will help elucidate additional features that are unique to HDAC8, allowing for the development of HDAC8-specific inhibitors as cancer therapeutics. Little is known about the role of HDAC8 and p53 deacetylation or deacylation which supports the significance of the in vitro and in silico data reported here. When p53 is acetylated, it is more stable and can therefore inhibit tumor growth through induction of cell cycle arrest and/or apoptosis. ³⁹ Future experiments will include cell culture experiments to further test for in vivo inhibitory activity of hexanoic and decanoic acid and further validate in silico docking results reported here.