A distal regulatory region of a class I human histone deacetylase

Abstract

Histone deacetylases (HDACs) are key enzymes in epigenetics and important drug targets in cancer biology. Whilst it has been established that HDACs regulate many cellular processes, far less is known about the regulation of these enzymes themselves. Here, we show that HDAC8 is allosterically regulated by shifts in populations between exchanging states. An inactive state is identified, which is stabilised by a range of mutations and resembles a sparsely-populated state in equilibrium with active HDAC8. Computational models show that the inactive and active states differ by small changes in a regulatory region that extends up to 28 Å from the active site. The regulatory allosteric region identified here in HDAC8 corresponds to regions in other class I HDACs known to bind regulators, thus suggesting a general mechanism. The presented results pave the way for the development of allosteric HDAC inhibitors and regulators to improve the therapy for several disease states.

Fig. 1. Coupling between helix1-loop1-helix2 and the active site.

a The deacetylase reaction catalysed by HDACs. b Surface structure of HDAC8 (PDB: 2v5w) with substrate bound (left) and structural elements discussed below highlighted (right). c Methyl-TROSY NMR spectra of isoleucine δ1-methyl groups in HDAC8 (30 μM). Overlay of free-form (red), with sub-stoichiometric (14 μM; green) and excess (blue) amounts of TSA (75 μM). Distinct changes are observed for the labelled residues. d, e Structural representation of isoleucine and methionine chemical shift changes, ΔCS, upon (d) SAHA and (e) TSA binding. Long-range effects, up to 28 Å from the binding site, are observed upon TSA binding. The shift changes are calculated as $\mathrm{\Delta }\mathrm{CS}\left(\mathrm{Ile}\right)=\sqrt{{\left(\mathrm{\Delta }{\delta }_{\mathrm{H}}/0.28\right)}^2+{\left(\mathrm{\Delta }{\delta }_{\mathrm{C}}/1.66\right)}^2}$ and $\mathrm{\Delta }\mathrm{CS}\left(\mathrm{Met}\right)=\sqrt{{\left(\mathrm{\Delta }{\delta }_{\mathrm{H}}/0.38\right)}^2+{\left(\mathrm{\Delta }{\delta }_{\mathrm{C}}/1.67\right)}^2}$ based on the standard deviation of assigned chemical shifts. Isoleucine-34 was coloured red in e. due to peak disappearance upon saturation with TSA.

Fig. 2. HDAC8 is in exchange with an alternative state.

a Methyl multiple-quantum CPMG relaxation dispersion profiles for M27 ¹³C^εH₃ of free HDAC8. The curved relaxation dispersion profiles reveal that the major state is in exchange with a low-populated minor state. b Structural representation (PDB: 2v5w) of the $\sqrt{p_{\mathrm{m}}\left(1-p_{\mathrm{m}}\right)}\mid \mathrm{\Delta }{\varpi }_{\mathrm{C}}\mid$ parameters obtained from the CPMG relaxation dispersions, showing the sites affected by the exchange, Supplementary Table 1. c Multiple-quantum relaxation dispersion profiles of M27 ¹³C^εH₃ for TSA-bound HDAC8. In a, c, circles represent experimental data, vertical lines represent the standard derivation (s.d.) and the solid line is the result of a least-squares fit to a two-state model (see text). d Correlation between ¹³C chemical shift differences between free and TSA-bound HDAC8, and |Δϖ| obtained from CPMG relaxation dispersion experiments on TSA-bound HDAC8. Vertical lines represent the standard derivation (s.d.) of the derived |Δϖ| parameters and data points in green are sites located in or near the helix1-loop1-helix2 region.

Fig. 3. The alternative state of the helix1-loop1-helix2 region is inactivating.

a Overlay of methyl-TROSY NMR spectra of WT-HDAC8 (red), WT-HDAC8 with 2.5eq TSA (blue), and the five mutations I19A (brown), S39E (cyan), M40A (green), S39EM40A (magenta) and F336A (yellow). The mutants stabilise the alternate state to varying degrees. b Relative enzymatic deacetylase activity of the HDAC8 mutants (see Methods). Vertical lines represent the uncertainty (r.m.s.d.). c Relative enzymatic activity versus population of the TSA-bound-like state of HDAC8. The full-drawn line connects WT and WT + TSA, while the blue shaded area indicates the r.m.s.d. (±24%).

Fig. 4. Structural changes between active and inactive HDAC8 from molecular dynamic simulations.

a Distribution of the distance between the centre-of-masses of helix1 (H1) and helix2 (H2) for wild-type HDAC8 (left, red) and S39E-HDAC8 (right, cyan). b Movement of helix1 and helix2 of active wild-type HDAC8 (red) compared to the crystal structure of TSA-bound HDAC8 (PDB: 1t64; blue). c The average orientation of helix1 and helix2 for wild-type HDAC8 (red), S39E-HDAC8 (cyan) and TSA-bound HDAC8. d Distance between loop1 and the substrate-binding tunnel, here assessed by the distance (yellow dashed line) between C^α of K33 in loop1 and C^α of F152, which forms the wall of the substrate-binding tunnel. e Distribution of distances between K33 C^α and F152 C^α for the simulation of wild-type HDAC8 (red) and S39E-HDAC8 (cyan). The distance measured for the crystal structure of substrate-bound HDAC8 (PDB: 2v5w) is shown as a green dashed line.