Six Years (2012–2018) of Researches on Catalytic EZH2 Inhibitors: The Boom of the 2-Pyridone Compounds

Abstract

Enhancer of zeste homolog 2 (EZH2), the catalytic subunit of the Polycomb repressive complex 2 (PRC2), catalyzes the methylation of lysine 27 of histone H3 (H3K27) up to its trimethylated form (H3K27me), inducing by this way block of transcription and gene silencing. High levels of H3K27me3 have been found in both hematological malignancies and solid cancers, due to EZH2 overexpression and/or EZH2 mutation. From 2012, a number of highly potent and selective catalytic inhibitors of EZH2 have been reported, almost all bearing a 2-pyridone group in their structure. Typically, 2-pyridone inhibitors are selective for EZH2 over other methyltransferases, and some of them are specific for EZH2 over EZH1, others behave as dual EZH2/EZH1 inhibitors. The 2-pyridone moiety was crucial for the enzyme inhibition, as revealed later by crystallographic studies because it occupies partially the site for the co-substrate SAM (or the by-product, SAH) in the binding pocket of the enzyme, accounting for the SAM-competitive mechanism of action displayed by all the 2-pyridone inhibitors. The 2-pyridone warhead is linked to a support substructure, that can be either a bicyclic heteroaromatic ring (such as indazole, see for instance EPZ005687 and UNC1999, or indole, see for instance GSK126, EI1, and the more recent CPI-1205) or a simple monocyclic (hetero) aromatic ring (tazemetostat, MC3629, (R)-OR-S1/2), eventually annulated with the amide chain carrying the 2-pyridone group (3,4-dihydroisoquinoline-1(2H)-ones). Different substitutions at the support moiety influence the pharmacokinetics and pharmacodynamics of the compounds as well as their water solubility. In cancer diseases, the first reported 2-pyridone inhibitors displayed high antiproliferative effects in vitro and in vivo in lymphomas characterized by mutant EZH2 (such as Y641N), but the most recent compounds exert their anticancer activity against tumors with wild-type EZH2 as well. The dual EZH2/1 inhibitors have been recently reported to be more effective than EZH2 selective inhibitors in specific leukemias including leukemias cancer stem cells.

1. Introduction

Chromatin in eukaryotes is organized into nucleosomes, the smallest structural, repetitive units of the genetic material. Each nucleosome is characterized by about 146 base pairs of double-stranded DNA wrapped around a histone octamer containing two copies of histone H2A, H2B, H3, and H4. From the histone core, a number of protrusions called “histone tails” come out, particularly enriched in basic lysine (Lys) and arginine (Arg) residues. At physiological pH, these positively charged side chains form electrostatic and H-bond interactions with the negatively charged phosphate groups of DNA leading to tight binding and packaging.^[1] Functionally, chromatin can exist into two forms: the first is closed and highly packed (heterochromatin), the latter is open, relaxed and less condensed (euchromatin). In heterochromatin there is no space for recruitment of transcription factors, thus heterochromatin is transcriptionally silent, whereas euchromatin is more accessible to transcription factors and thus can give input to gene transcription.^[2,3] Histone tails, through the action of specific enzyme families, can undergo chemical covalent modifications, such as acetylation, methylation, phosphorylation, and ubiquitylation, which considerably influence DNA accessibility and regulation of gene expression.^[3] Among them, histone methylation attracted growing interest by researchers in the last decade. Histone methylation was first recognized in 2000,^[4] and a link between this covalent modification and regulation of gene transcription has been established from 2003.^[5] A large family of more than 60 histone methyltransferases (HMTs), including histone lysine methyltransferases (HKMTs) and protein/histone arginine methyltransferases (PRMTs), were identified in humans.^[6] These enzymes catalyze the methylation of specific lysine or arginine amino acid side chains in histone as well as non-histone proteins. All HMTs use a common mechanism of catalysis, by the formation of a ternary complex with the substrate (target protein) and the co-substrate (the universal methyl donor S-adenosyl-l-methionine (SAM)).^[7] The transfer of the methyl group from SAM to the Lys residue through a classical S_N2 mechanism yields the methylated protein together with the by-product S-adenosyl-l-homocysteine (SAH).^[7] The amino group of the Lys side chains can accept up to three methyl groups, and can thus exist in four distinct states of methylation, from free amine to trimethyl-ammonium quaternary salt (K or Kme0, Kme1, Kme2, Kme3) (Figure 1). Sometimes, a single HMT is responsible for multiple rounds of methylation of a specific Lys residue, while in other cases different HMTs accomplish the distinct, sequential methylation reactions.^[7] Functionally, the biological meaning of such chemical covalent modifications is not univocal. While Lys acetylation or deacetylation leads to gene transcription or silencing, respectively, the effect of Lys methylation depends on the Lys residue that underwent methylation, and by the grade of its methylation (mono-, di-, or tri-methylation). Thus, some methylated chromatin residues, such as H3K4me2/3 and H3K36me1/2/3, are marks of activated transcription, and others such as H3K9me2/3, H3K27me2/3, and H4 K20me2/3, are marks for gene silencing.^[8] Since these post-translational chemical modifications play crucial roles in gene regulation, cell differentiation, and DNA damage repair, in physiology as well as in pathology, there has been a steadily increasing interest towards assessing the potential of HMTs as therapeutic targets.^[9] In particular, methylation of histones, as well as non-histone proteins, have been implicated in various cancers, being a dynamic process that plays a key role in the regulation of gene expression and transcription.^[10]

Figure 1.

Mechanism of mono-, di-, and trimethylation of Lys residues by HMTs and SAM.

2. Enhancer-of-Zeste Homolog 2 (EZH2) and its Role in Cancer Diseases

Enhancer-of-zeste homolog 2 (EZH2) and its close homolog EZH1 catalyze mono-, di- and tri-methylation of Lys 27 of histone H3 (H3K27).^[11] EZH2 (or, alternatively, EZH1) is the catalytic unit of the Polycomb Repressive Complex 2 (PRC2), essential for embryonic development, differentiation, and repression of transcription of some genes such as those of the Hox family.^[12] Purified EZH2 is not catalytically active. To perform its catalytic activity on the H3K27 substrate, it must be in complex with other proteins, namely EED (embryonic ectoderm development), SUZ12 (suppressor of zeste 12), and RbAp46/48.^[13] Despite their very high sequence identity in SET domains (96%), EZH2 and EZH1 differ in tissue distribution and in their relative catalytic efficiency, with EZH2 being more potent than EZH1 for H3K27 methylation.^[14,15]

Defects in EZH2 have been found in a number of diseases. EZH2 is expressed in a wide range of B-cell lymphomas including Burkitt’s lymphoma, mantle cell lymphomas (MCLs), follicular lymphoma (FL), and diffuse large B-cell lymphomas (DLBCLs),^[15] and its overexpression and high levels positively correlated with aggressiveness and unfavorable prognosis.^[16] The importance of PRC2 in lymphomagenesis is further reinforced by the identification of recurrent, heterozygous missense mutations of EZH2 in B-cell lymphomas such as FLs and DLBCLs (10%–20% of EZH2 mutations). The most prevalent among these mutations, typically targeting the SET catalytic domain of EZH2, is a point mutation of the Y641 residue found mutated to either Asn, Phe, Cys, Ser, or His. Two other rare EZH2 mutations, A677G and A687 V, have been described in about 1%–3% of B-cell lymphoma cases.^[17,18] With these gain-of-function mutations, the substrate specificity of the PRC2 complex is altered.^[19] In fact, wild-type EZH2 shows the greatest preference for the nonmethylated form of H3K27 (H3K27me0→H3K27me1) and is much less effective for subsequent (H3K27me1 to H3K27me2 and H3K27me2 to H3K27me3) reactions. Differently, the lymphoma-associated EZH2 Y641 mutation such as Y641N catalyzes very rapidly the conversion H3K27me2→H3K27me3, less efficiently that H3K27me1→H3K27me2, and shows very limited ability (if any) to methylate H3K27me0 to H3K27me1.^[19] The co-presence of PRC2-EZH2^WT and PRC2-EZH2^Y641N in lymphomas accounts for a sort of cooperation between the two enzymatic activities to produce the abnormally high levels of H3K27me3 seen in these diseases.^[20]

Apart from mutations, overexpression of EZH2 and high levels of H3K27me3 have been associated with a huge number of human cancers, including breast,^[21,22] prostate,^[23] lymphoma,^[24] myeloma,^[25] and leukemia.^[26] The role of EZH1 in the pathogenesis of B cell lymphomas has not been well established so far, but it seems that EZH2 and EZH1 work in compensation each other to maintain cell proliferation and suppress differentiation in some leukemias (i.e., MLL-AF9).^[26,27]

3. The Dawn of a New Era for Cancer Treatment: First Compounds Able to Fight EZH2

The first compound reported to inhibit EZH2 was 3-deazaneplanocin A (DZNep), a SAH hydrolase inhibitor that depletes EZH2 and reduces the associated H3K27me3 levels inducing time-dependent apoptosis in breast MCF7 and HCT116 colon cancer cells when tested at 5 μM for 48 and 72 h.^[28] DZNep interferes with SAM and SAH metabolism and can indirectly inhibit the EZH2 methylation reaction through depletion of the EZH2 protein (Figure 2). As it is an indirect inhibitor, its EZH2 specificity is quite low, and the apoptosis induced by DZNep is only in part related to its ability to inhibit the PRC2 pathway.

Figure 2.

Structures of DZNep with its mechanism of action, MC1947, MC1948 and MC1945.

In 2010, we reported the effects of two EZH2 inhibitors, MC1947 and MC1948, on muscle stem (satellite) cells.^[29] Muscle satellite cells (SCs) differentiate into myofibers and myotubes through an inflammation signaling in which TNF-α, a cytokine secreted within the regenerative environment by the inflammatory infiltrate, regulates muscle regeneration by activating the promyogenic p38α kinase/PRC2 signaling with subsequent repression of Pax7 expression. We found that either inhibition of p38α kinase (by the specific inhibitor SB 203580), or EZH2 inhibition (by MC1947 and MC1948) in SCs cultured in differentiation medium prevented the formation of myosin heavy chain-(MyHC−)positive myotubes and increased the number of proliferating cells that continued to express Pax7 but did not express muscle differentiation markers. Upon release of p38α or EZH2 inhibition by drug withdrawal, the expanded population of SCs differentiated massively and formed myotubes with an increased efficiency, as compared to control cells (ex-vivo, SCs derived from wild-type C57/BL6 mice).^[29] MC1947 and MC1948 (Figure 2) are two EZH2 inhibitors belonging to a series of epigenetic multiple ligands which, with appropriate substitutions at the two benzylidene moieties, lost their multi-target feature and retained only specific single-target inhibition.^[30–32] MC1945 (Figure 2), an EZH2 inhibitor belonging to the same series,^[32] when tested in vivo (2.5 mg/kg, twice a day, 3 days per week for 3 weeks) in pediatric high-risk alveolar PAX3-FOXO1 rhabdomyosarcoma xenograft models displayed apoptosis induction and 70% reduced tumor growth at 21 days.^[33] In embryonal rhabdomyosarcoma RD cells, MC1945 and to a lower extent MC1948, similarly to DZNep, showed dose-dependent arrest of cell proliferation and decreased H3K27me3 levels.^[34,35] Contrarily to DZNep, MC1945 did not alter the level of the EZH2 protein confirming to be a catalytic inhibitor of EZH2.

4. A Fundamental Step for Highly Potent and Selective Catalytic EZH2 Inhibitors: The Birth of the 2-Pyridone Saga

From 2012, extensive high-throughput biochemical screening (HTS) campaigns have been performed with the PRC2-EZH2 complex, with the aim to identify highly potent and selective catalytic inhibitors.

4.1. 2-Pyridone Compounds with a Central Heteroaromatic Bicyclic Nucleus

The first discovered selective inhibitor was EPZ005687 (from Epizyme), which emerged from a HTS performed on a library of 175,000 compounds. This first screen allowed the identification of the hit compound 1 (K_i vs. PRC2/EZH2=310 nM). Subsequent optimization of 1 by introducing salifiable portions to increase the water solubility, by replacement of the pyrazolopyridine central ring with the indazole one (compounds 2 and 3), and by enlarging the size of the N1 substituent from iso-propyl to cyclopentyl led to EPZ005687 (K_i=24 nM, Figure 3).^[36] EPZ005687 was competitive with SAM and noncompetitive with the peptide substrate and, most importantly, was over 500-fold selective for the PRC2/EZH2 complex over 15 other methyltransferases, about 50-fold selective for PRC2/EZH2 over PRC2/EZH1, and equipotent against EZH2^WT and EZH2^Y641 mutants.^[36] In lymphomas, EPZ005687 potently reduced H3K27me3 levels in both EZH2^WT- and EZH2 mutated-containing cells, but only in the latter this decrease of trimethylation was converted into a selective arrest of proliferation and cell growth.^[36]

Figure 3.

Structures of EPZ and GSK compounds. EPZ005687’s optimization process.

In the same year (2012), another HTS and subsequent optimization resulted in the discovery of GSK126^[37] and its strict analogs GSK343^[38] and GSK503^[39] as potent and selective EZH2 inhibitors (Figure 3). These compounds share a similar core structure with EPZ005687, but in GSK126 and GSK503 the indazole has been replaced by an indole scaffold. The similarities of such inhibitors with the previous Epizyme’s compounds are evident. As EPZ005687, GSK126 is SAM-competitive (K_i=0.5–3 nM, against wild-type and mutant forms of EZH2) and more than 1000-fold selective for EZH2 over 20 other methyltransferases. Also, it was over 150-fold selective for EZH2 over EZH1. When tested in a panel of B-cell lymphoma cell lines with or without EZH2 mutations, GSK126 led to a decrease of H3K27me3 levels, activation of transcription, and arrest of proliferation more efficiently in cells with EZH2 mutations.^[37] More importantly, GSK126 markedly inhibited the growth of EZH2-mutant DLBCL xenografts in mice, and reduced tumor volume (91–100%) and improved survival (100% percentage survival at 300 mg/kg twice per week) in the more aggressive Karpas-422 xenograft model, decreasing H3K27me3 levels and increasing EZH2 target gene expression.^[37] GSK126 was also well tolerated by the treated mice but suffered from oral bioavailability and blood-brain-barrier permeability issues.^[40,41] Nevertheless, GSK126 entered a Phase I clinical trial in 2014 as GSK2816126 in patients with various lymphomas, solid tumors, and multiple myeloma.

Again in 2012, a third HTS identified another indole derivative containing the 2-pyridone moiety, EI1 (Figure 4), as an EZH2 selective inhibitor.^[42] EI1 showed a profile similar to that of previously described compounds: SAM-competitive EZH2 inhibition (K_i=13 nM), equal potency against EZH2 mutants, about 90-fold selectivity for EZH2 over EZH1 and 10,000-fold selectivity for EZH2 over 10 other methyltransferases. In cancer cells, EI1 inhibited the proliferation of lymphomas expressing gain-of-function mutated EZH2 with a decrease of H3K27me3 levels and activation of transcription of gene targets, while the growth of wild-type EZH2 cells was only weakly inhibited.^[42]

Figure 4.

Structures of EI1, UNC1999 and JQEZ5. Graphical summary of SAR for UNC1999.

In early 2013, UNC1999 has been reported as the first orally bioavailable inhibitor of EZH2 and as the first dual EZH2/1 inhibitor.^[43] Structurally, UNC1999 bears the 2-pyridone moiety decorated with a n-propyl chain at C4 and the usual methyl group at C6, supported by a bicyclic heterocycle (indazole, as in EPZ005687), further substituted by a piperazine-pyridine moiety (Figure 4). However, this little differences in the chemical structure respect to EPZ005687 and GSK126 had a large effect on the pharmacokinetic properties of the compound and on EZH2/1 selectivity. UNC1999 was competitive with SAM (K_i=4.6 nM) and noncompetitive with the peptide substrate. Moreover, UNC1999 was a dual EZH2/EZH1 inhibitor, it being only about 10-fold selective for EZH2 over EZH1. On the other hand, UNC1999 was >1000-fold selective for EZH2/EZH1 over 15 other methyltransferases.^[43] UNC1999 was able to reduce H3K27me3 levels as well as cell proliferation in a large panel of cancer cells, containing either wild-type or mutant EZH2, without affecting EZH2 protein levels and showing low cytotoxicity.^[43] In MLL-AF9 murine leukemia cells, in which EZH2 and EZH1 work together in compensation,^[26,27] UNC1999 effectively blocked cell proliferation, while GSK126, potent against EZH2 but weak against EZH1, had little antiproliferative activity. Most importantly, oral administration of UNC1999 in mice (50 mg/kg by oral gavage to mice twice per day (BID)) with MLL-AF9-induced leukemia prolonged mice survival, with a latency of 36.6 days in contrast to 24 days for vehicle-treated, highlighting the importance of a dual EZH2/EZH1 inhibition in certain cell contests.^[27] In addition, UNC1999 exhibited low micromolar cytotoxicity in vitro on a panel of brain-tumor initiating cell (BTIC) lines and synergized with dexamethasone (DEX) in two BTIC cell lines. Such combination showed suppression of tumor growth in vivo as well. Additionally, a co-treatment with UNC1999 and HDAC1/2 inhibitor synergized in vitro by inducing apoptosis and DNA damage. In these studies, UNC1999 proved more potent than both GSK126 and tazemetostat (see below), further corroborating the hypothesis that dual EZH2/EZH1 inhibition could improve the efficacy in cancer.^[44] Extensive structure-activity relationship (SAR) studies^[45] performed on the UNC1999 structure showed i) the crucial role of the C4-n-propyl/C6-methyl decoration at the 2-pyridone unit for the dual inhibition, while alkyl substitution at C5 or addition of further oxo groups or C/N replacement into the structure abated the inhibition capability of both enzymes; ii) alternative substitutions at the indazole nucleus (iso-propyl chain at N2 instead of N1, amide function at C3 instead of C4, pyridine-piperazine tail at C5 instead of C6) were all detrimental for the inhibitory activity, as well as the replacement of indazole with other poly-nitrogen-containing bicyclic heterocycles; iii) an increase of the N1 substituent size until cyclohexane retained EZH2 but reduced EZH1 inhibitory activity; a further increase of the N1 chain also decreased the anti-EZH2 potency; iv) while the size of the substituent at the piperazine moiety had no influence on EZH2/EZH1 inhibition, replacement of the pyridine ring as well as truncation of this tail furnished compounds with relatively good potency for EZH2 but low potency for EZH1 (Figure 4).^[45]

In 2016, a novel analog of UNC1999, JQEZ5 (Figure 4), has been described as very potent against the EZH2 Y641F mutation, that was shown to induce lymphoma and melanoma through a reorganization of chromatin structure.^[46]

4.2. “Compound 3”, the Unique Non-Pyridone EZH2 Inhibitor

Coming back to 2013, a new series of compounds displaying a new, non-pyridone-containing chemotype has been described as EZH2 inhibitors.^[47] The hit “compound 3” (Figure 5) contains a tetramethylpiperidinobenzamide moiety linked to a cyanophenylpyridazine. “Compound 3” showed SAM-competitive inhibition of EZH2 (IC₅₀=21 nM), with 10-fold increased potency respect to EZH1 or EZH2 mutants’ inhibition, and did not significantly affect 5 other methyltransferases up to 100 μM.^[47] In cellular assays, “compound 3” reduced H3K27me3 levels with micromolar potency in both EZH2^WT and EZH2^mutant cells, but arrested proliferation and increased gene transcription only in cancer cells with mutated EZH2.^[47]

Figure 5.

Structures of non-pyridone EZH2 inhibitors.

4.3. 2-Pyridone Compounds with a Central (Hetero) aromatic Monocyclic Nucleus

After in 2013, a novel orally active EZH2 inhibitor, EPZ-6438 (now known as tazemetostat) was reported.^[48] The structural novelty of this compound is that, unlike previously published compounds that contain a bicyclic core (indole, indazole), tazemetostat has a phenyl core linked to the 2-pyridone moiety (Figure 6). Respect to its earlier prototype, EPZ005687, tazemetostat displayed improved potency and pharmacokinetic parameters. With a SAM-competitive mechanism of action, tazemetostat strongly inhibited wild-type (K_i=2.5 nM) and mutant EZH2 enzymes and displayed about 35-fold selectivity over EZH1 and >4500-fold selectivity over 14 other methyltransferases. In malignant rhabdoid tumors (MRTs, G401 cells) containing an inactivated SMARCB1 subunit, due to mutation of the SWI/SNF complex, tazemetostat displayed a decrease of H3K27 me1/2/3 levels, induction of apoptosis and arrest of proliferation at nanomolar concentration, while wild-type SMARCB1 RD cells were not affected.^[48] In vivo studies on G401 xenografts in mice confirmed its high potency as well as high tolerability: oral administration at 250 or 500 mg/kg twice daily for 21–28 days essentially eliminated the tumor with minimal effect on mice body weight.^[48] Currently, tazemetostat is in Phase 1/2 studies for the treatment of lymphomas and advanced solid tumors.

Figure 6.

Structures of EPZ-6438 (tazemetostat), EPZ011989, ZLD1039, ZLD1122, EBI-2511, (R)-OR-S1 and (R)-OR-S2.

Chemical manipulation applied to the tazemetostat structure led to two further orally available benzamidomethyl-2-pyridone analogues, EPZ011989^[49] and ZLD1039,^[50] both very potent against wild-type and mutant EZH2 (K_i<15 nM), about 15-fold selective for EZH2 over EZH1 and >3000- to >10000-fold selective for EZH2 over other methyltransferases. In EPZ011989, the tetrahydropyranyl moiety of tazemetostat was converted into a 4-aminocyclohexylamino group, and the morpholinomethylphenyl into a morpholinopropynyl side chain (Figure 6). These chemical changes improved the pharmacokinetic and pharmacodynamic properties of the compound respect to tazemetostat, and made EPZ011989 highly effective against Karpas-422 human DLBCL xenografts in mice.^[49] Indeed, when tested at 250 and 500 mg/kg for 21 days BID, EPZ011989 induced significant tumor regression at both doses with a slight effect on body weight. ZLD1039 respect to tazemetostat shows a piperazinopyridine tail instead of the morpholinomethylphenyl group, and interestingly a modification of 2-pyridone to 3-tetrahydroisoquinolinone, obtained by cyclization of the C4 and C5 pyridone positions to form a fused cyclohexene (Figure 6). ZLD1039 induced apoptosis and arrest of proliferation in breast tumor cells at submicromolar level. These findings have been confirmed by its efficacy in in vivo tumor xenograft models (67.5, 86.1, and 58.6% tumor growth inhibition in the MCF7 (200 mg/kg), MDA-MB-231 (200 mg/kg), and 4T1 (250 mg/kg) models, respectively).^[50] Also, ZLD1039 was well tolerated in toxicological studies. Its strict analog is ZLD1122 (Figure 6), exhibiting nanomolar potency against EZH2 and EZH1 and effective in inducing apoptosis and arrest of cell growth in DLBCL.^[51]

Very recently, Lu et al. reported an optimization of a novel series of benzofuran-derived EZH2 inhibitors through a scaffold hopping approach starting from tazemetostat. This study led to the discovery of the compound EBI-2511, with superior antitumor efficacy than tazemetostat in Pfeiffer tumor xenograft mice (97% reduction of tumor size with EBI-2511 vs. 80% reduction with tazemetostat, both at 100 mg/kg).^[52] Moreover, Honma et al. described (R)-OR-S1 and (R)-OR-S2 as dual inhibitors of EZH1/2 suppressing trimethylation of histone H3K27 in cells more than EZH2 selective inhibitors. They also showed greater antitumor efficacy than EZH2 selective inhibitor in vitro and in vivo against diffuse large B-cell lymphoma as well as solid cancers, without exhibiting severe toxicity in rats.^[53] When tested in acute myeloid leukemia mouse models and patient-derived xenograft models, (R)-OR-S1 reduced the number of leukemia stem cells, impaired leukemia progression, prolonged survival, and in combination with cytarabine prevented relapse,^[54] thus indicating the possibility of EZH1/2 dual inhibitors for clinical applications mainly in combination therapy.

In 2012–2013, when only the first catalytic 2-pyridone EZH2 inhibitors with a bicyclic heteroaromatic ring as the central scaffold were known (EPZ005687, GSK126, EI1, and UNC1999), we designed same monocyclic heteroaromatic-based compounds always bearing the 2-pyridone moiety, recognized as crucial for EZH2 inhibition. By applying a pruning approach to the central indazole or indole group, we designed and synthesized a series of pyrazole- and pyrrole-based compounds linked through an amide bond to the 2-pyridone moiety, which can be a novel fragment to optimize (Figure 7). The first pyrazole prototype, MC3629, showed low micromolar EZH2 inhibition (5 to 15 μM depending on the substrate used), with a SAM-competitive mechanism of action.^[55,56] MC3629 significantly reduced cell proliferation at 10 μM in breast MDA-MB231, leukemia K562, and neuroblastoma SK-N-BE cells after 2–5 days of treatment. In the last two cell lines, MC3629 showed reduced levels of H3K27me3 but not those of H3K4me3, induced autophagy and, to a lower extent, apoptosis (SK-N-BE cells).^[55] GSK126, tested in the same conditions, showed highly improved activity in K562 cells but lower effects (significant at 25 μM) in MDA-MB231 and SK-N-BE cells. When tested in Sonic Hedgehog (SHH) medulloblastoma stem-like cells (MB-SLC), MC3629 at 5 μM decreased H3K27me3 levels, reduced cell proliferation and self-renewal, and induced apoptosis.^[56] Its co-administration with shEZH2 to human MB-SLCs did not show any additive effect on cell viability, ruling out the possibility of off-target effects in its mode of action. Tested on MB xenografted mice, MC3629 displayed a significant decrease of tumor volume, a reduction of stemness and cell proliferation, and induction of apoptosis.^[56] Despite its low biochemical inhibitory potency, MC3629 in both mouse and human MB-SLCs exhibited a similar impaired cell viability as GSK126, EPZ005687, and tazemetostat.^[56]

Figure 7.

Structures of 2-pyridone-containing pyrazole and pyrrole EZH2 inhibitors.

4.4. Back to the Future: Again 2-Pyridone Inhibitors with a Central Bicyclic Moiety

At the end of 2014 CPI-360 (Figure 8), another SAM-competitive inhibitor of EZH2 carrying the 2-pyridone moiety linked to the indole nucleus through an amide function, has been reported as a SAM-competitive EZH2 inhibitor effective against a large cancer cell panel representing various non-Hodgkin’s lymphoma (NHL) subtypes, including cell lines with EZH2^WT.^[57] The chemical novelties are i) the insertion of the amide chain connecting the 2-pyridone with the indole nucleus at the C3 rather than the typical C4 position of the bicycle, ii) the C4-C7 positions of the indole are completely unsubstituted, and iii) a 1-(tetrahydro-2H-pyran-4-yl)ethyl substituent has been introduced at the indole N1 position (Figure 8). CPI-360 and its analog CPI-169^[57,58] (Figure 8), more potent and with improved microsomal stability, were also potent in EZH2 mutant-containing GCB-DLBCL xenograft models. Further optimization of such compounds, still suffering from limited oral bioavailability, led in 2016 to CPI-1205,^[59] a potent and selective EZH2 inhibitor (IC₅₀=2.0 nM) less potent against EZH1 (IC₅₀=52 nM). The key chemical strategy to obtain improved cellular potency, selectivity, toxicity, bioavailability, and PK properties in these derivatives was to make the piperidine nitrogen less basic through the introduction of polyfluoroalkyl (2 or 3 carbon atoms) substituents. CPI-1205 showed a high oral bioavailability and strong anticancer effects in a Karpas-422 xenograft model, inducing >97% tumor growth inhibition when administered at 160 mg/kg orally BID for 25 days. CPI-1205 was well tolerated in toxicological studies with no changes of mice body weight and is currently in Phase 1 clinical trials for the treatment of B-cell lymphomas.^[59]

Figure 8.

Structures of CPI Compounds. SAR studies leading to CPI-1205.

4.5. 2-Pyridone-Containing 3,4-Dihydroisoquinoline-1(2H)-ones: Lesson from X-ray Crystal Structures

Very recently, the crystal structure of a 2-pyridone inhibitor (compound 1, Figure 9) complexed with either wild-type or Y641N-mutant PRC2 complex, consisting of human EED, human SUZ12-VEFS, and American chameleon (Anolis carolinensis) EZH2 (AcEZH2) subunits has been reported.^[60] These structures, compared with that of the isolated EZH2 SET domain, catalytically inactive, allow to understand the mechanism of activation of the SET domain, through a crucial role played by EED and SUZ12 (allosteric activation). EED contributes to the structuring of the SET activation loop, a conserved structural feature of lysine methyltransferases which induces a conformational switch of the I-SET domain, supported by the SUZ12-VEFS subunit, leading to SET activation. This I-SET rigid body rotation creates a ligand-binding pocket that does not exist in the inactive form of the SET domain. In this exclusive pocket, the catalytic EZH2 inhibitor can accommodate itself with its “pendant” group (the 3,4-dihydroisoquinoline moiety), while the 2-pyridone pharmacophore unit anchors the ligand orthogonally to the SAM-binding cavity, in a region also occupied by the SAM homocysteine moiety for the specific recognition.^[60] The most recent X-ray structure of a CPI-1205 analog in complex with the human PRC2 fully confirmed the binding mode for the 2-pyridone EZH2 inhibitors.^[59] Moreover, this mode of binding, with the partial overlap of co-substrate and 2-pyridone inhibitor binding sites, fully accounts for the previously observed SAM-competitive mechanism of action of these inhibitors.

Figure 9.

Structures of 2-Pyridone-Containing 3,4-Dihydroisoquinoline-1(2H)-ones. Optimization Process Leading to PF-06821497.

In 2016, the same researchers describing the X-ray structure of compound 1 in complex with human EED, human SUZ12-VEFS, and AcEZH2 reported the identification and development of a new series (whose compound 1 belongs to) of EZH2 inhibitors.^[61] The novelty was the cyclization of the amide function (till now always unsubstituted) connecting the 2-pyridone with the “pendant” group over the benzene moiety, to give a six- or seven-membered lactam (Figure 9). Further introduction of a second lipophilic group (particularly Cl) and replacement of the initial alkoxy group with a five-membered heterocyclic ring (particularly the 3,5-dimethyl-4-isoxazolyl) at the benzene ring furnished the most potent derivative 31 (Figure 9). Compound 31 showed high potency in inhibiting both EZH2^WT (K_i=0.7 nM) and EZH2^Y641N (K_i=3.0 nM) with a SAM-competitive mechanism and was highly effective in decreasing H3K27me3 levels (IC₅₀=15 nM) and in arresting proliferation (IC₅₀=25 nM) in Karpas-422 cells. In Karpas-422 xenograft mouse model, compound 31 administered at 300 mg/kg BID for 20 days reduced the tumor volume with less than 10% body weight loss.

Further modification of this scaffold led to PF-06821497 (Figure 9), in which the 4-methyl substituent at the 2-pyridone ring has been replaced by a 4-methoxy group, and the 3,5-dimethyl-4-isoxazolyl moiety of 31 by a methoxy (oxetan-3-yl)methyl group. In particular, this last change greatly improved the physicochemical and pharmacokinetic properties of the compound increasing the potency against the EZH2 mutant Y641N. Thus, PF-06821497, when administered at 100 mg/kg subcutaneously once-daily to Karpas-422 DLBCL mice tumor xenografts containing the Y641N EZH2 mutation, displayed profound tumor regression (100% tumor volume reduction after 21 days of treatment), also sustained for another 40 days after the end of treatment (day 31), joined to very high free mouse plasma exposures.^[62] For these reasons, PF-06821497 has been advanced as a clinical development candidate.

4.6. EZH2 Inhibition/Disruption: The Road Ahead

Several phase-I/phase-II clinical trials with tazemetostat and phase I clinical trials with GSK126 and CPI-1205 are recruiting patients or are ongoing across a wide spectrum of diseases including NHL, FL, DLBCL, pediatric lymphomas, and solid tumors.^[63] Treatments with tazemetostat obtained encouraging responses in patients, most notably those with FL carrying mutant EZH2, and the drug has been granted a Fast Track designation from the US FDA for FL (with both wild-type and mutant EZH2), and for EZH2-mutant DLBCL. Instead, the first results of the Phase-I clinical trial of GSK126 were less encouraging, and those of CPI-1205 are still ongoing.

Acquired resistance appeared in various lymphoma cancer cell lines following treatment with some 2-pyridone EZH2 inhibitors. Particularly, in Karpas-422 and Pfeiffer cells treated with EI1 and tazemetostat some mutations have been reported for those residues that are in contact with the inhibitor (in the I-SET and SET activation loop domains) in the created pocket of the enzyme.^[64,65] This acquired resistance opens a new challenge in the EZH2 inhibitors’ design, and renewed efforts have to be made in the design and synthesis of new series of inhibitors, possibly by identification of novel chemotypes. An alternative approach can be the disruption of interactions of EZH2 with its binding partners, EED and SUZ12, and small molecules interfering with the EZH2/EED interactions have been already described and validated in cancer.^[66–70] Undoubtedly in next years the strategy of inhibition/disruption of EZH2, mainly for the treatment of cancer, should follow these parallel ways.