Adenosine 5′-Carboxamide-Based Inhibitors of METTL1

Abstract

METTL1 is the human RNA methyltransferase that catalyzes the methylation of N7-guanosine in RNA. Overexpression of METTL1 has been linked to cancer, and increasing evidence supports the therapeutic potential of METTL1 inhibition in oncology. In this study, we have optimized a series of adenosine 5′-carboxamide derivatives as METTL1 inhibitors through structure-guided modifications of a previously discovered hit compound. The advanced inhibitor B22 shows an IC₅₀ of 1 μM in an enzymatic assay, which is a 178-fold improvement with respect to the initial hit. The crystal structure of the des-methyl analogue of B22 (compound B19, IC₅₀ = 0.4 μM) provides evidence that the benzylpiperazine moiety is accommodated within the guanosine-binding subsite of METTL1. The inhibitor B22 shows high solubility and metabolic stability and is selective against a panel of seven histone methyltransferases and the RNA-methyltransferase METTL3/METTL14.

RNA modification is a significant process for gene regulation. Methyltransferase-like proteins (METTLs) constitute a family of epitranscriptomic writers that serve as key regulators of RNA post-transcriptional modifications. METTL1 in complex with WD repeat domain 4 (WDR4) catalyzes internal methylation at the N7 position of guanosine (m7G) in tRNAs, miRNAs, and mRNAs. , METTL1 utilizes the cofactor S-adenosyl-methionine (SAM) as a methyl group donor, converting SAM to S-adenosyl homocysteine (SAH). m7G modifications of RNA regulate translation, cell cycle progression, stem cell maintenance, differentiation, and cell migration.

Multiple studies have demonstrated the connection between the overexpression of METTL1/WDR4 and tumorigenesis. tRNA m7G modification deposited by METTL1/WDR4 is upregulated in several cancer types. Increasing evidence suggests that elevated METTL1 expression promotes cancer cell proliferation, migration, and invasion, implicating METTL1 in cancer development. −

Knockdown of METTL1 or WDR4 reduces m7G modification in tRNAs across several cancer types and exhibits context-dependent effects. For example, knockdown of METTL1 in intrahepatic cholangiocarcinoma suppresses cell migration and invasion, reduces cell growth, and impairs colony-forming ability. In prostate cancer preclinical models, METTL1 knockdown increases intratumoral infiltration of pro-inflammatory immune cells and enhances responses to immunotherapy. METTL1 knockdown suppresses proliferation in lung cancer cells and suppresses both proliferation and migration in bladder cancer cells in vitro. , METTL1 was also linked to anticancer drug sensitivity. Although METTL1 is upregulated and acts as an oncogene in most researched cancer types, METTL1 is a tumor suppressor in breast cancer, and its knockdown promotes breast cancer progression. There is also evidence that METTL1 has oncogene activity independent of its methyltransferase activity. METTL1 promotes sarcomagenesis by increasing tRNA aminoacylation, elevating protein synthesis, and promoting tumor cell growth. Recently, METTL1 was shown to promote leukemia stem cell homeostasis and leukemogenesis via activation of tRNA^Phe _GAA/HCK/CXCR4 signaling. In the same study, it was shown that inhibition of METTL1 by a small molecule eliminates leukemia stem cells and myeloid leukemia.

The number of epitranscriptomic targets currently under active investigation in drug discovery is minimal. The anticancer medication (S)-crizotinib (Figure ) was identified as a METTL1 inhibitor in high-throughput drug screening using the newly developed fluorescent RNA-methyltransferase probe Cy5TAD. A recent patent disclosed a series of dihydroisoquinolinyl-morpholine derivatives with reported low-nanomolar biochemical potency against METTL1. We have identified 11 METTL1 inhibitors representing three distinct chemotypes, with inhibitory potencies ranging from 40 to 300 μM. Among these, the adenine derivative 1 (Figure ) demonstrated favorable ligand efficiency.

1.

Structures of known METTL1 inhibitors.

Given its involvement in multiple types of cancer, METTL1 is a compelling target for drug discovery. Here, we report the protein structure-based development of adenosine-containing METTL1 inhibitors using a SAM-mimicking strategy. METTL1 utilizes SAM as a methyl donor, converting it to SAH within the same catalytic pocket. Our molecules are designed to act as SAM-competitive inhibitors to block the co-substrate/co-product binding site.

The target compounds were prepared according to the route described in Scheme . The synthesis started from 2′,3′-O-isopropylidene­adenosine-5′-carboxylic acid 2. Most of the compounds were obtained via amide bond formation using HBTU (A1–A3, A7, A8) or T3P (A4–A6, A9–A21, A30–A35) as a coupling reagent and an appropriate commercially available or synthetically easily accessible amine. Alternatively, we used reductive amination of piperazine 3 with the corresponding benzaldehydes to obtain compounds A22–A27. Piperazine derivative A28 was synthesized using the alkylation reaction of 3 with 2-(hydroxymethyl)­benzyl bromide. The benzyl alcohol group of A28 was oxidized with Dess–Martin periodinane to yield the corresponding benzaldehyde derivative, which we subsequently subjected to reductive amination reaction with methylamine to obtain A29. Acetyl or benzoyl phenol protecting groups were removed with potassium carbonate in MeOH. The isopropylidene protecting group was cleaved by treatment with 50% formic acid, yielding the final compounds B1–B35.

1. General Synthesis of Target Compounds B1–B35 .

^a Reagents and conditions: (a) appropriate amine, HBTU, TEA, DMF, 18 h, rt; (b) appropriate amine, T3P, TEA, EtOAc–DMF, 6–18 h, rt; (c) 50% HCOOH, 6–10 h, 50 °C.; (d) (i) K₂CO₃, MeOH; (ii) 50% HCOOH, 6–10 h, 50 °C.; (e) (i) Cbz-piperazine, T3P, TEA, EtOAc–DMF, 6–18 h, rt; (ii) H₂, 10% Pd/C, MeOH, 4 h, rt; (f) appropriate aldehyde, Na­(OAc)₃BH, DCE, rt, 16 h; (g) appropriate benzylbromide, TEA, MeCN, 2 h, rt; (h) (i) Dess–Martin reagent, DCM, 18 h, rt; (ii) MeNH₂HCl, TEA, Na­(OAc)₃BH, DCE, 16 h, rt.

We selected the previously characterized METTL1 inhibitor 1 (Table ), which incorporates an adenosine-privileged scaffold, as the starting point for the development of new inhibitors. Piperazine derivative 1 is selective against METTL3/14 and METTL16. The initial design is based on a substitution of the polar carboxylic group of 1 with a less polar group such as phenyl or heteroaryl. It was hypothesized that the phenyl group would occupy the methionine-binding subsite of the SAM pocket or target the RNA-binding site. The IC₅₀ value was determined using a luminescence-based METTL1-WDR4 enzymatic assay. Inhibitory activities of the new compounds were evaluated in comparison to sinefungin (Figure ). To rule out potential assay interference, selected compounds were tested against the downstream reaction and detection components of the assay. No interference was observed (Figure S4).

1. Chemical Structures and METTL1 Inhibitory Activity of Compounds 1 and B1–B36 .

^aLuminescence-based METTL1-WDR4 enzymatic assay. Data are the mean ± SD calculated from at least three experiments or values of two independent replicates. Number of replicates (n) is indicated in parentheses. All curves are shown in Figure S3.

^bSingle-dose experiment showing protein residual signal at respective compound concentrations.

^cData from ref . n.d., not determined. The Green color highlights inhibitors with IC₅₀ ≤ 1 μM. Pale green highlights compounds with activity comparable to sinefungin, i.e., IC₅₀ in the range 5–20 μM. Yellow indicates compounds with IC₅₀ of 25–100 μM. Red indicates compounds with IC₅₀ of >100 μM or n.d.

The replacement of the carboxylic acid with the amide analogue B1 showed a 4-fold increase in activity, while compounds B2–B4 with phenyl or heteroaryl groups directly attached to the piperazine demonstrated more than 5-fold improvement compared to inhibitor 1 (Table ).

On the basis of these results, a spacer was introduced between the piperazine and benzene ring of the molecule, resulting in sulfonamide B5 and amide B6. However, these modifications led to decreased inhibitory activity compared with compound B4. In contrast, the methylene-spaced 4-hydroxybenzyl containing B7 is twice as active as its shorter analogue B2. Subsequently, we decided to explore the role of substituents on the benzyl group of piperazine. A set of functional groups with contrasting electronic properties as well as moieties acting as hydrogen-bond donors (e.g., OH, NH₂) or hydrogen bond acceptors (e.g., COOMe) were introduced to probe the nature of the binding pocket, resulting in a series of benzyl piperazine analogs B7-B29.

In this series, we obtained several compounds with inhibitory activities comparable to sinefungin (within ±2-fold of its IC₅₀ of 10 μM, pale green in Table ) and two compounds, B19 and B22, that are an order of magnitude more potent than sinefungin. Most of these compounds bear a hydroxyl group or other hydrogen-bond donor group at the ortho or para position of the benzyl group, except for B15, where 4-NO₂ is a weak hydrogen bond acceptor. Compound B19 is 445-fold more potent than its parent compound 1. The thermal shift assay of the compound B19 with both METTL1 and METTL1/WDR4 showed that B19 stabilizes the protein (Figure S1).

We determined the crystal structure of METTL1 in complex with B19. Crystal structure data are provided in Table S1. The structure reveals a conserved binding mode of the adenosine moiety (Figure ). The adenine forms hydrogen bonds with the backbone NH group of Ala111 and with the side chain of Asn110. The ribose hydroxyl groups form hydrogen bonds with Glu77 and Arg79. The carbonyl group interacts with the backbone NH of Asp133. The piperazine is positively charged at physiological pH (the predicted pK _a of N-methylpiperazine is about 9.0), and thus its tertiary amine is involved in favorable electrostatic interactions with the carbonyl of Phe131 and the side chain of Glu210 (cyan dotted lines in Figure A). The dihydroxybenzyl group of B19 extends toward the RNA-binding site. The 2-hydroxy group acts as a hydrogen bond donor to the side chain of Asp169, providing a favorable electrostatic interaction that contributes to the high affinity of B19.

2.

Crystal structure (PDB code 29GE) of the complex of METTL1 (gray) and B19 (carbon atoms in green). (A) The hydrogen bonds (yellow) and interactions of the positively charged amine of the piperazine (cyan) are emphasized. (B) Superimposed crystal structures of the complex of METTL1 (gray) with B19 (green), and METTL1 (gold) with SAH (orange, PDB code 7OGJ).

To understand the structural basis for the structure–activity observations for the compounds B1–B29, we performed a docking study using the crystal structure of METTL1 in complex with compound B19. The docking models of compounds B7–B11, B15, and B21 indicated that these molecules adopt positions similar to those of B19. The p-hydroxy group of B7 binds to Asp169 with a slightly shifted position of piperazine. Additionally, phenol forms a hydrogen bond with Thr168 (Figure ). The m-hydroxy group of B9 (Figure S2B) and the methylaniline NH of B21 interact with Asp169 similarly. The high potency of B3 could be explained by the hydrogen bond between the protonated pyridine of B3 and Asp169 (Figure S2A). The heterocyclic NH in the pyrrolopyridine of B11 acts as a hydrogen-bond donor (Figure S2C). It serves as a phenol bioisostere, like indoles, which are reported phenol bioisosteres.

3.

Docked pose of compound B7 in METTL1. The crystal structure (PDB code 29GE) of the complex of B19 (carbon atoms in green) and METTL1 (ribbon model, carbon atoms of side chains in gray) was used for docking.

Compound B15 exhibits strong inhibitory activity despite lacking hydrogen-bond donors at the benzyl group. This activity can be explained by the interaction of its p-nitro group and Glu210 by the formation of a salt bridge (Figure S2D).

The activity of compounds containing 2-hydroxy group is strongly influenced by additional substituents at the benzyl group. No clear correlation was observed between the acidity of the 2-hydroxy group and compound activity. 2-Hydroxybenzyl derivative B10 and 4-bromo-2-hydroxybenzyl B18 exhibited similar activity at a concentration of 250 μM. Compounds containing fluorine and methyl group (B23–B26) demonstrated IC₅₀ values in the range of 15–54 μM.

The closest analogue to compound B19, 2-hydroxy-4-methoxybenzyl derivative B22, showed 1 μM activity. The results for compounds B23–B26 show that substitution at the 2,4-position of the benzene ring confers greater activity than substitution at the 2,5-positions relative to piperazine.

Potential modifications of the 2-hydroxy group were explored, leading to the synthesis of a carboxylic acid analogue B27, in which the carboxylate was expected to interact with the backbone of Asp169. However, no improvement in the activity was observed. Additionally, more bulky and less acidic 2-hydroxymethyl group-containing analogue B28 and 2-methylamino-containing B29 are less active than B10.

Next, we investigated some possible modifications of the piperazine and synthesized six compounds (B30–B35), where piperazine was replaced with structural analogues to assess how these changes affect biological activity. The inhibitory activity data indicate that piperazine can be replaced with 1,4-diazepane or hydroxypiperidine, as compounds B32 and B35 exhibited activity comparable to their respective analogues B7 and B12. p-Benzonitrile aminopiperidine analogue B34 outperformed the piperazine compound B13, improving the activity by more than 3-fold. Compound B30, which contains an additional carbonyl group, and aminoazetidine B33 were less successful modifications, displaying reduced activity. The extended version of parent compound B7, aminopiperidine B31, lost its activity, indicating that the increased molecular size may hinder the binding.

To further investigate the influence of compound rigidity, we obtained adenosylpiperazine B36 as a more flexible analogue of compound B7. The loss of the carbonyl group resulted in complete loss of activity, indicating that the amide bond in adenosine 5′-carboxamides is essential for maintaining the molecular geometry necessary for receptor interaction.

With the submicromolar inhibitor B19 in hand, we decided to assess its selectivity and compare it with two other members of the series, namely, B2, whose selectivity had been measured in an early phase of this campaign, and B22. As in a previous study, we selected a panel of SAM-dependent methyltransferases, including DOT1L, G9a, MLL4, PRDM9, PRMT1, SETD2, SMYD3, and METTL3/METTL14 (Table ). At a concentration of 20 μM, B2 and B22 do not inhibit any of these methyltransferases. Compound B19 showed weak inhibitory activity on histone lysine N-methyltransferases G9a and SETD2. These results suggest that the 5′-(N-piperazinyl­carboxamido)­adenosine scaffold imparts selectivity for METTL1 compared to other methyltransferases.

2. Methyltransferase Selectivity Measured by a Radiometric Assay .

methyltransferase	B2	B19	B22
DOT1L	94	97	79
G9a	101	29	107
MLL4 complex	96	80	99
PRDM9	107	93	104
PRMT1	93	101	102
SETD2	95	39	105
SMYD3	100	96	97
METTL3/METTL14	97	82	98

^aremaining methyltransferase activity at 20 μM compound concentration,%. Values below 40% are in bold. Large values (above 75%) indicate lack of inhibition.

Kinetic solubility, mouse liver microsomal stability, and Caco-2 permeability were assessed for compounds B2 and B7 as part of an in vitro ADME early-stage profiling campaign (Table ). Compounds B2 and B7 exhibit high stability upon exposure to mouse liver microsomes and a favorable kinetic solubility. However, the Caco-2 assay indicates a very low intestinal absorption potential. The observed high metabolic stability of compounds B2 and B7 in microsomes may result from limited intracellular exposure and reduced access to hepatic metabolic enzymes. Compound B19 showed reduced microsomal stability in the assay, both with and without NADPH. The half-life of B19 in PBS buffer is 1 h. The low microsomal stability is likely due to its low chemical stability. As adenosine derivatives containing phenol groups, these compounds are very polar and may be susceptible to oxidation. Notably, the p-methoxy analogue B22 shows high solubility, moderate stability in PBS, and a half-life of about 4 h in mouse liver microsomes.

3. Analysis of Physicochemical Properties, in Vitro Metabolic Stability, and Cellular Permeability .

				Caco-2 permeability, P app (10–6 cm/s)
compound	kinetic solubility, μM	stability in PBS, T 1/2, h	stability in mouse liver microsomes, T 1/2, min	A–B	B–A	efflux ratio
B2	437	29	>120	0.1	0.9	6.0
B7	382	n.d.	>120	0.2	0.7	3.2
B19	300	1	32.6	n.d.	n.d.	n.d.
B22	447	10	239	n.d.	n.d.	n.d.

^an.d., not determined.

^bTarget kinetic solubility 500 μM.

^cStability was measured at 30 μM concentration at 37 °C.

^dSimilar stability was observed in the assay without NADPH (Figure S5).

In conclusion, a series of adenosine 5′-carboxamide derivatives were designed and synthesized starting from the hit compound 1. Chemical modifications led to the identification of two potent METTL1 inhibitors: B19 and B22. The benzylpiperazine moiety of the inhibitors is positioned in the RNA subpocket of METTL1 as revealed by the crystal structure of the METTL1/B19 complex (PDB code 29GE). The positively charged piperazine ring is involved in favorable electrostatic interactions with the Glu210 side chain of the METTL1. The side chain of Asp169 forms a hydrogen bond with the hydrogen-bond-donating groups on the benzyl ring of B19 (Figure A). These interactions provide a structural explanation for the high affinities of the top compounds. Our attempts to replace piperazine with other nitrogen-containing cycles suggest that piperazine in these structures may be replaced by diazepane (B32) or piperidine (B34). A selectivity analysis of three compounds of the series on a panel of eight methyltransferases provides evidence of selectivity for METTL1. Furthermore, B22 shows favorable kinetic solubility and chemical and metabolic stability. Future work should aim at further increasing activity, stability, and cellular permeability.

Glossary

Abbreviations

HBTU

2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate

SAM

S-adenosylmethionine

SAH

S-adenosylhomocysteine

T3P

propanephosphonic acid anhydride

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.6c00203.

• Supplementary figures, table, experimental details of chemical synthesis and assays, crystallization procedure, data collection and refinement, description of docking studies (PDF)

• NMR, HRMS and HPLC data for all final compounds (PDF)

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

No unexpected or unusually high safety hazards were encountered.

The authors declare no competing financial interest.