This is not surprising as EED is the common component in both complexes. The
co-crystal structures of EED162 and EED210 bound to EED (as well as those
obtained with the other identified hits) demonstrate a common and yet dynamic
“induced fit” in the H3K27me3 pocket of EED with a significant conformational
change of the aromatic cage residues [242]. These structural studies suggested
that optimizing interactions with the key residues of the H3K27me3 pocket can
potentially improve the binding affinities of the identified hits, providing an attractive starting point for developing novel EED binders.
Indeed, optimization studies were carried out for EED210. Due to difficulties
in synthesizing this scaffold, the structure of EED210 (IC 50 ¼ 2.5 μM, LE ¼ 0.29,
cLogP ¼ 4.9) was deconstructed to its minimal fragment, 3-(3-methoxybenzyl)
piperidine (Fig. 17; IC 50 ¼ 95 μM, LE ¼ 0.36, and cLogP ¼ 2.8). This fragment
showed direct binding to EED in the 2D NMR assay with a K D of 32 μM, and
its co-crystal structure with EED showed that the interactions made with the
protein are the same as the ones observed for the parent compound EED210, with
a total retainment of the binding pose [243]. Aided by the co-crystal structure,
the deconstructed ligand was then “regrown,” and this approach led to the development of a series of small molecules that allosterically inhibited PRC2 activity by
interacting with the methyllysine-binding pocket of EED [243]. All compounds were
evaluated for key properties, such as PRC2 inhibition (LC/MS-based assay), ligand
efficiency (LE), and cellular permeability (cLogP, Caco-2 cell permeability). The
most promising compound resulted derivative #19 (Fig. 17, compound numbering as
in the original paper), which demonstrated enhanced potency (IC50 ¼ 1 μM), good
ligand efficiency (LE ¼ 0.35), and improved permeability (cLogP ¼ 3.9) [243].
A similar deconstruction-reconstruction approach was applied for the optimization of EED162 [244]. In fact, a closer inspection of the interactions between
compound EED162 and EED suggested that the entire piperidine ring connecting
C7 to C8 and the benzyl group attached to nitrogen at 9-position did not contribute
much to the interaction and likely reduced efficiency of binding due to the nonessential lipophilicity. Compound #8 (Fig. 17, compound numbering as in the original
paper), a fragment of EED162, was confirmed to retain most of the key interactions
with EED and was found to be as potent as the parent compound but with dramatic
improvements of both ligand efficiency (LE; 0.26 and 0.46 for EED162 and compound #8, respectively) and lipophilic efficiency (LipE; 2.6 and 5.0, respectively).
Guided by X-ray crystallography, the approach finally led to the discovery of
EED226 (Fig. 17) as a potent and selective inhibitor of PRC2 activity (ITC with
EED, K D ¼ 82 nM, and PRC2 complex, K D ¼ 114 nM) [244, 245]. Like EED162
and EED210, EED226 was found selective against a panel of over 21 other protein
methyltransferases, kinases, and other protein classes and was unsurprisingly an
inhibitor for the EZH1/2-PRC2 complexes suggesting its potential effectiveness in
treating EZH1 and/or EZH2 dependent cancers such as myelodysplastic disorders.
EED226 directly binds to the H3K27me3 pocket of EED causing a distinct allosteric
effect that inhibits both basal- and H3K27me3-stimulated PRC2 activity. The crystal
structure of its complex with EED and EZH2 peptide (40–68, EBD) showed that the
binding of EED226 to the aromatic cage of EED (comprised of F97, Y148, W364,
380
G. Sbardella
co-crystal structures of EED162 and EED210 bound to EED (as well as those
obtained with the other identified hits) demonstrate a common and yet dynamic
“induced fit” in the H3K27me3 pocket of EED with a significant conformational
change of the aromatic cage residues [242]. These structural studies suggested
that optimizing interactions with the key residues of the H3K27me3 pocket can
potentially improve the binding affinities of the identified hits, providing an attractive starting point for developing novel EED binders.
Indeed, optimization studies were carried out for EED210. Due to difficulties
in synthesizing this scaffold, the structure of EED210 (IC 50 ¼ 2.5 μM, LE ¼ 0.29,
cLogP ¼ 4.9) was deconstructed to its minimal fragment, 3-(3-methoxybenzyl)
piperidine (Fig. 17; IC 50 ¼ 95 μM, LE ¼ 0.36, and cLogP ¼ 2.8). This fragment
showed direct binding to EED in the 2D NMR assay with a K D of 32 μM, and
its co-crystal structure with EED showed that the interactions made with the
protein are the same as the ones observed for the parent compound EED210, with
a total retainment of the binding pose [243]. Aided by the co-crystal structure,
the deconstructed ligand was then “regrown,” and this approach led to the development of a series of small molecules that allosterically inhibited PRC2 activity by
interacting with the methyllysine-binding pocket of EED [243]. All compounds were
evaluated for key properties, such as PRC2 inhibition (LC/MS-based assay), ligand
efficiency (LE), and cellular permeability (cLogP, Caco-2 cell permeability). The
most promising compound resulted derivative #19 (Fig. 17, compound numbering as
in the original paper), which demonstrated enhanced potency (IC50 ¼ 1 μM), good
ligand efficiency (LE ¼ 0.35), and improved permeability (cLogP ¼ 3.9) [243].
A similar deconstruction-reconstruction approach was applied for the optimization of EED162 [244]. In fact, a closer inspection of the interactions between
compound EED162 and EED suggested that the entire piperidine ring connecting
C7 to C8 and the benzyl group attached to nitrogen at 9-position did not contribute
much to the interaction and likely reduced efficiency of binding due to the nonessential lipophilicity. Compound #8 (Fig. 17, compound numbering as in the original
paper), a fragment of EED162, was confirmed to retain most of the key interactions
with EED and was found to be as potent as the parent compound but with dramatic
improvements of both ligand efficiency (LE; 0.26 and 0.46 for EED162 and compound #8, respectively) and lipophilic efficiency (LipE; 2.6 and 5.0, respectively).
Guided by X-ray crystallography, the approach finally led to the discovery of
EED226 (Fig. 17) as a potent and selective inhibitor of PRC2 activity (ITC with
EED, K D ¼ 82 nM, and PRC2 complex, K D ¼ 114 nM) [244, 245]. Like EED162
and EED210, EED226 was found selective against a panel of over 21 other protein
methyltransferases, kinases, and other protein classes and was unsurprisingly an
inhibitor for the EZH1/2-PRC2 complexes suggesting its potential effectiveness in
treating EZH1 and/or EZH2 dependent cancers such as myelodysplastic disorders.
EED226 directly binds to the H3K27me3 pocket of EED causing a distinct allosteric
effect that inhibits both basal- and H3K27me3-stimulated PRC2 activity. The crystal
structure of its complex with EED and EZH2 peptide (40–68, EBD) showed that the
binding of EED226 to the aromatic cage of EED (comprised of F97, Y148, W364,
380
G. Sbardella