Identifying Small-Molecule Binding Sites for Epigenetic Proteins at Domain-Domain Interfaces

Abstract

Epigenetics is a rapidly growing field in drug discovery. Of particular interest is the role of post-translational modifications to histones and the proteins that read, write, and erase such modifications. The development of inhibitors for reader domains has focused on single domains. One of the major difficulties of designing inhibitors for reader domains is that, with the notable exception of bromodomains, they tend not to possess a well-enclosed binding site amenable to small-molecule inhibition. As many of the proteins in epigenetic regulation have multiple domains, there are opportunities for designing inhibitors that bind at a domain-domain interface which provide a more suitable interaction pocket. Examination of X-ray structures of multiple domains involved in recognising and modifying post-translational histone marks using the SiteMap algorithm identified potential binding sites at domain-domain interfaces. For the tandem plant homeodomain-bromodomain of SP100C, a potential inter-domain site identified computationally was validated experimentally by the discovery of ligands by X-ray crystallographic fragment screening.

Keywords: X-ray fragment screening; bromodomains; epigenetics; histones; tudor domains.

Figure 1

Domain maps exemplifying proteins that bind or modify histone tails containing multiple PTMs. Tandem domains studied in this work are in red boxes. HAT: histone acetyltransferase domain; KAT6A: lysine acetyltransferase 6; SETDB1: SET domain bifurcated 1; TRIM33: tripartite motif protein 33; PHF8: PHD Finger Protein 8; ZMYND11: Zinc finger MYND domain‐containing protein 11; SET (Su(var)3–9, enhancer‐of‐zeste and trithorax): a lysine methyltransferase domain; JmjC (Jumonji C‐terminal domain): an oxygenase lysine demethylase domain; PHD (plant homeodomain), Tudor and PWWP are methyl lysine binding domains; bromodomains are acetyl lysine binding domains. Domain identity and boundaries are adapted from EBI InterPro.5

Figure 2

Exemplary inhibitors of readers, writers, and erasers. l‐Moses15 and UNC121513 are inhibitors of the BRD of PCAF and methyl‐lysine binding MBT domain of L3MBTL3 respectively. A‐48519 is a CBP/p300 histone acetyltransferase (HAT) inhibitor, and EPZ01566616 inhibits the methyltransferases PRMT5. Vorinostat (SAHA) is a clinically used HDAC inhibitor, and compound 4824 is an inhibitor of the 2‐oxoglutarate dependent KDM5 demethylases.

Figure 3

Occurrences of epigenetic domains in tandem with other domain structures reported in the PDB. A) PHDs, BRDs, Tudor domains, and MBT domains are the most common domains in the tandem structure set. B) Structures containing multiple Tudor domains appear seven times, with five structures containing multiple MBT domains. The most common heterodomain combination in our structure set is PHD–BRD, of which seven examples were identified. A PHD–BRD has an N‐terminal PHD and a C‐terminal BRD; this is distinct from a BRD–PHD, in which the domain order is reversed.

Figure 4

The size, enclosure, hydrophilicity, and SiteScore of selected tandem domains. The overall SiteScore is a combination of the sites size, enclosure and hydrophilicity. The mean values for the 326 binding sites with known sub‐micromolar ligands that were used as a training set during the development of SiteMap is shown to benchmark the analysis of the tandem domains.

Figure 5

View of a crystal structure of the BRD (left, green) and the PWWP domain (right, magenta) of ZMYND11 (PDB ID: https://www.rcsb.org/structure/4N4G). Residues 29–39 of H3.3 are shown in stick form at the histone binding site identified between the two domains (surface).

Figure 6

Examples of ligandable pockets identified in crystal structures of PHD and Tudor domain proteins. A) The H3K4me2 residue is present in the binding site of the N‐terminal Tudor (green) of TP53BP1, near the interface with the C‐terminal Tudor domain (magenta) (PDB ID: https://www.rcsb.org/structure/3LGL). B) Yellow coloring indicates the volume of the ligandable cleft between the second (green) and third (magenta) Tudor domains of SETDB1 is (PDB ID: https://www.rcsb.org/structure/3DLM). C) The single PHD shown is that of BPTF (PDB ID: https://www.rcsb.org/structure/2FSA). D) the tandem PHD shown is that of KAT6A (PDB ID: https://www.rcsb.org/structure/3V43). The N‐terminal PHD of KAT6A is magenta and the C‐terminal PHD is green.

Figure 7

Examples of two types of PHD‐bromodomain. A) The PHD domain (magenta) and BRD domain (green) of BPTF (PDB ID: https://www.rcsb.org/structure/2FSA) are separated by a rigid linker. B) The PHD domain (magenta) and BRD domain (green) of SP100C (PDB ID: https://www.rcsb.org/structure/4PTB) form a compact, globular structure. SP100C is shown with the site points as identified by SiteMap, and an overlaid ligand identified by a crystal soaking experiment. The novel potential ligand binding site at the domain–domain interface of the PHD–BRD of SP100C is indicated by cyan dots, the PHD is shown in magenta, and the BRD is shown in green. A peptide taken from a structure of the related PHD–BRD of TRIM33 (PDB ID: https://www.rcsb.org/structure/3U5P) is superimposed to highlight the probable peptide binding surface.

Figure 8

Fragment soaking in SP100C identifies compounds binding at the PHD/BRD domain interface. A) Binding mode of oxadiazole 1 (yellow) to SP100C. The backbone ribbon and side chains of residues from the PHD domain involved in the compound binding are colored in magenta, the backbone ribbon and residues from the BRD domain involved in the compound binding are colored in green. B) Aniline 2 does not bind as deeply in the pocket as oxadiazole 1 and therefore does not cause a conformational change in residue Q751. C) The two fragment hits identified for the inter‐domain binding site of SP100C are shown overlaid with an electrostatic map of the surface of the apo structure of SP100C. Note that the SP100C inter domain binding site presents a more hydrophobic surface than the rest of the surface area. D) Chemical structures of compounds 1 and 2.