Peptide-based covalent inhibitors of MALT1 paracaspase

Abstract

Potent and selective substrate-based covalent inhibitors of MALT1 protease were developed from the tetrapeptide tool compound Z-VRPR-fmk. To improve cell permeability, we replaced one arginine residue. We further optimized a series of tripeptides and identified compounds that were potent in both a GloSensor reporter assay measuring cellular MALT1 protease activity, and an OCI-Ly3 cell proliferation assay. Example compounds showed good overall selectivity towards cysteine proteases, and one compound was selected for further profiling in ABL-DLBCL cells and xenograft efficacy models.

The paracaspase MALT1 is structurally similar to the caspases, but cleaves its substrates at an arginine residue, rather than aspartic acid.^1,2 It is a component of the CARMA1-BCL10-MALT1 (CBM) complex, involved in the activation of the NF-κB pathway through cleavage of the NF-κB inhibitors RelB and A20. It also functions as a scaffolding protein, where it binds to BCL10 and recruits TRAF6, which in turn activates TAK1 and the IKK complex. Constitutively active MALT1 has been identified in B-Cell lymphomas, including ABC-type diffuse large B-cell lymphoma (DLBCL), MALT lymphoma and mantle cell lymphoma (MCL).³ Inhibition of MALT1 protease activity reduces the growth of ABC DLBCL lines, and MALT1 inhibitors are of interest as therapeutic agents. The peptide Z-VRPR-fmk¹ and other small molecules have been reported as MALT1 inhibitors,^4–6 and the biology and potential therapeutic benefits of MALT1 inhibition were reviewed recently.^7,8 Selective activity-based probes of MALT1 have also been prepared, incorporating the tetrapeptide sequences LRSR,⁹ LVSR¹⁰ and in a recent study which also evaluated unnatural amino acids and extended to the prime sites, LVPR.¹¹

With the objective of developing selective inhibitors of MALT1 suitable for in vivo dosing, we chose to investigate peptide-based inhibitors. The first reported MALT1 inhibitor, Z-VRPR-fmk, was derived from the optimal tetrapeptide substrate of the metacaspase AtmC9,¹² and has been used as a cellular tool compound to explore the role of MALT1 protease activity.^1,13,14 The compound was typically used at a concentration of 50 μM or higher due to poor cell penetration, likely caused by the two arginine residues. Like other peptidic tool compounds used to investigate caspases (e.g. Z-VAD-fmk,¹⁵ Z-DEVD-fmk¹⁶) it incorporates an electrophilic fluoromethyl ketone warhead, which forms a covalent bond with the active site cysteine residue.¹⁷

Our first goal was to improve the cell permeability of the series. Co-crystal structures of MALT1 with Z-VRPR-fmk suggested that the P1 arginine was critical, as multiple interactions were observed between the arginine and acidic residues in the S1 pocket.¹⁸ The proline occupied a small S2 pocket, while the P3 arginine was solvent exposed, making no significant interactions with the protein. MALT1 substrate specificity was previously explored using positional scanning libraries, with the outcomes consistent with structural observations.^2,18 There was a very strong preference for Arg in the S1 pocket, small residues such as Ser, Pro and Ala were preferred at P2, and a variety of residues including Val, Ile and Leu were tolerated at P3. Leu was favored at the P4 position, occupying a hydrophobic pocket.

To minimize the number of H-bond donors, we replaced the P3 arginine of Z-VRPR-fmk with valine, and despite a slight preference reported for Ser over Pro at P2, we retained proline at that position to prepare the tetrapeptide Z-LVPR-fmk 1. To confirm the ability of our initial compounds to inhibit MALT1 protease activity in cells, we assessed compounds for their ability to inhibit RelB cleavage in OCI-Ly3 cells by Western blot.¹⁹ Compound 1 was significantly more effective than Z-VRPR-fmk in this assay. Z-VRPR-fmk showed partial inhibition of RelB cleavage at 20 μM, but 1 achieved complete inhibition at 1.25 μM. Next, we profiled compounds in a MALT1 biochemical assay.²⁰ Compound 1 showed similar biochemical activity to Z-VRPR-fmk (Ki 0.09 μM for 1 and 0.14 µM for Z-VRPR-fmk), but in an OCI-Ly3 cell proliferation assay,²⁰ 1 had a GI₅₀ of 1.05 μM compared with 8.22 μM for Z-VRPR-fmk, which suggested that the enhanced cellular effects were derived from improved cell penetration. To confirm the importance of the covalent bond, we prepared the ethyl ketone 2, a reversible analog of Z-LVPR-fmk 1. Compound 2 showed no activity in the biochemical assay at 5 μM, no effects upon RelB cleavage at 20 μM, and no activity in the cell proliferation assay.

The fluoromethyl ketone (fmk) warhead is reported only rarely in the drug discovery literature.^21–23 We were concerned about its potential reactivity, and evaluated alternative electrophilic warheads, including vinyl ketones, vinyl sulfones, vinyl esters, and epoxy ketones. However, none of the replacements for the fluoromethyl ketone delivered comparable potency. In a study comparing the reactivity of a peptide fmk inhibitor of cathepsin B with the corresponding chloromethyl ketone, it was reported that the fmk compound was slightly less reactive towards the target protease, but that it also showed a 500-fold reduction in reactivity towards glutathione.²⁴ This suggested that if sufficient binding affinity and selectivity could be obtained for an inhibitor with the fmk warhead, off-target effects could be minimized.

To improve the potential for in vivo dosing with the series, we wanted to reduce the molecular weight and shorten the synthetic route, and we therefore prepared covalent tripeptides (Table 1). One of the positional scanning studies found that MALT1 was unable to cleave substrates shorter than tetrapeptides,² A methylamine-capped VPR-fmk tripeptide 3 was inactive in our cell assay. However, the Cbz-protected tripeptide (Z-VPR-fmk, 4) retained good biochemical and some cellular potency. A co-crystal structure of 4 with MALT1 showed that the benzyl substituent of the Cbz protecting group acts as a surrogate P4 residue, occupying the same hydrophobic pocket as the valine of Z-VRPR-fmk, and making a pi-stacking interaction with F499.²⁰ N-Methylation of the valine nitrogen, compound 5, led to reduced activity, while N-methylation of the arginine residue, compound 6, was not tolerated.

Table 1.

Enzyme and cell data for compounds 1–6.

Compound	MALT1 Ki (μM)a	Cell (OCI-Ly3) (μM)a	Cell (Glo) (μM)a
Z-VRPR-fmk	0.14	8.22	2.62
1 (Z-LVPR-fmk)	0.09	1.05	0.53
2	> 5	> 20	> 20
3	> 0.2	> 20	> 20
4 (Z-VPR-fmk)	0.04	8.52	1.96
5	0.81	> 20	3.00
6	> 5	ND	ND

^aEnzyme and cellular assays are described in Reference [20].

As we explored diversification at the N-terminus and modifications at the P2 and P3 positions of the tripeptide, we found that the Ki values in the biochemical assay were not predictive of cellular potency. Poor cell penetration may have been a factor for some examples (especially those with a cLogP < 0.5), but there was no clear correlation between the enzyme and cell assays for the more lipophilic compounds either. We developed a GloSensor reporter assay, which measured inhibition of cellular MALT1 protease activity in PMA/IO activated Raji cells, and this became our primary screening assay. We observed a good correlation (R = 0.78, p < 0.0001) between this assay and the OCI-LY3 proliferation data.²⁰ Compounds showed no anti-proliferative effects at concentrations up to 20 μM in a MALT1-independent (OCI-Ly1) counter-screen cell assay, suggesting a lack of significant off-target activity for the tripeptide fmk compounds.

One issue that emerged during the course of the work was partial racemization at the arginine residue. This occurred during the synthesis of the protected arginine fluoromethyl ketone 14, and in some cases was observed to a small extent during the final coupling and deprotection stages. The diastereomers could not be reliably distinguished by proton NMR, but ¹⁹F NMR typically produced a pair of triplets corresponding to the warhead fluorine, which allowed the ratio of diastereomers to be estimated. ¹⁹F Chemical shifts and ratios for the two diastereomers, when observed, are reported in the supplemental information (Table S1). The protected arginine fluoromethyl ketone 14 was prepared as shown in Scheme 1. Magnesium benzyl fluoromalonate 10 was prepared from dimethyl fluoromalonate 7 using procedures similar to those previously described,^21,25 and coupled with Boc arginine protected at its side chain with the Pmc sulfonyl group 11. Debenzylation/decarboxylation of 12 revealed the fluoromethylketone 13, from which the Boc group was selectively deprotected before further coupling with the desired peptide acid.

Scheme 1.

Synthesis of arginine fluoromethyl ketone. Reagents and conditions: (a) BnOH, toluene, cat. TsOH, reflux; (b) NaOH, H₂O, ⁱPrOH, 45 °C; (c) Mg(OEt)₂, THF; (d) Mg(O₂CCHFCO₂Bn)₂, CDI, THF; (e) H₂, Pd/C, EtOH; (f) TFA/DCM.

Modifications to the left hand side of the peptide were introduced as shown in Scheme 2, through preparation of the various dipeptide acids 17 (terminal amides) and 18 (terminal ureas), coupling of the arginine building block 14 and a final deprotection step.²⁶ Compounds with modifications at P3 and P2 were prepared through a similar approach. For compounds 4 and 31–43, the dipeptide esters were constructed by coupling the Cbz-protected acid to the methyl ester of proline or a related amine. For the 4-bromobenzamide examples in Table 3 (22 and 44–53), the dipeptide amines were prepared and then treated with 4-bromobenzoyl chloride. The dipeptide esters were then converted to the acids, for coupling with the arginine fluoromethyl ketone 14. Re-presentative experimental procedures for the synthesis of the tripeptide fluoromethyl ketones are included in reference 26, and characterization data for key intermediates and final compounds is in the supplemental information.

Scheme 2.

Synthesis of tripeptide fluoromethyl ketones. Reagents and conditions: (a) L-proline methyl ester, HATU, ⁱPr₂EtN, DMF; (b) TFA/DCM; (c) RCO₂H, HATU, ⁱPr₂EtN, DMF or RCOCl, Et₃N, DCM; (d) LiOH, H₂O, THF; (e) RNCO, Et₃N, DCM; (f) 14, HATU, ⁱPr₂EtN, DMF.

Table 3.

Enzyme and cell data for tripeptide MALT1 inhibitors modified at the P2 position (R³).

Compound	R1	R2	R3	Cell (Glo) (μM)	Cell (OCI-LY3) (μM)
4	BnO	isopropyl (Val)	Pro	1.96	8.52
39	(Cbz)		Ser	0.149	0.29
40				0.144	0.73
41				0.209	0.29
42				0.239	1.07
43			α-Me Pro	0.873	10.5
22	4-Br	isopropyl (Val)	Pro	0.071	0.13
44	Phe		Ser	0.062	0.15
45				0.085	0.07
46				0.327	0.46
47				0.219	0.77
48				0.077	0.12
49				0.174	0.46
50			d-Pro	17.5	> 20
51	4-Br	cyclohexyl	Pro	0.043	0.13
52	Phe		Ser	0.087	0.24
53				0.065	0.06

Compounds modified at the N-terminus and P3 position are shown in Table 2. Good activity in both cell assays was observed for both benzamides and ureas, especially those with a lipophilic substituent at the phenyl 4-position such as Br (22), CF₃ (25) or isopropyl (27). At P3, in addition to Ile, Leu and Phe residues, cyclohexyl (34) and tert-butyl (35) groups could be introduced in place of the isopropyl of valine. Cyclic achiral groups including cyclohexyl (37) and cyclopentyl (38) were not well-tolerated.

Table 2.

Enzyme and cell data for tripeptide MALT1 inhibitors modified at the N-terminus and P3 position.

Compound	R1	R2, R3 R3 = H unless stated	Cell (Glo) (μM)	Cell (OCI-LY3) (μM)
4	BnO (Cbz)	isopropyl (Val)	1.96	8.52
19	Ph	isopropyl (Val)	0.64	0.66
20	2-Br Ph	isopropyl (Val)	1.11	4.24
21	3-Br Ph	isopropyl (Val)	0.18	0.53
22	4-Br Ph	isopropyl (Val)	0.07	0.13
23	4-Cl Ph	isopropyl (Val)	0.18	1.24
24	4-CN Ph	isopropyl (Val)	0.71	6.66
25	4-CF3 Ph	isopropyl (Val)	0.23	0.12
26	4-MeO Ph	isopropyl (Val)	0.80	1.67
27	4-iPr Ph	isopropyl (Val)	0.07	0.27
28	cyclohexyl	isopropyl (Val)	0.64	0.94
29	Ph NH	isopropyl (Val)	0.17	1.52
30	4-Cl Ph NH	isopropyl (Val)	0.11	0.78
31	BnO (Cbz)	isobutyl (Leu)	3.42	5.49
32	BnO (Cbz)	sec-butyl (Ile)	1.86	2.44
33	BnO (Cbz)	benzyl (Phe)	2.03	2.20
34	BnO (Cbz)	cyclohexyl	0.15	1.45
35	BnO (Cbz)	tert-butyl	0.71	1.44
36	BnO (Cbz)	phenyl	1.43	8.05
37	BnO (Cbz)	R2, R3 = –(CH2)5–	6.00	6.23
38	BnO (Cbz)	R2, R3 = –(CH2)4–	5.38	4.31

Compounds with modifications at the P2 position (R³) are shown in Table 3. In place of proline, serine (39), piperidine (40), trans 4-OH proline (41) and oxazolidine (48) were all tolerated. α-Methyl proline (43) led to reduced activity, and the d-proline analog (50) was essentially inactive. During the work to explore the left hand side, 4-bromobenzamide emerged as a promising N-terminus cap, and in some cases P2 groups were introduced with this group in place of Cbz. Compounds were also prepared with cyclohexyl at P3, as an initial example (34) had demonstrated enhanced activity relative to valine at that position.

To assess the selectivity of our tripeptide-fmk MALT1 inhibitors, we profiled examples in a panel of 26 proteases, primarily caspases and cathepsins. At a concentration of 10 μM, compounds showed significant effects against several cathepsins, thrombin, trypsin and calpain. Profiling at lower concentrations against the proteases inhibited by our peptides showed reduced effects, with the overall activity profile determined to a great extent by the P2 substituent. (Table S2, Supplemental Information). At 100 nM, compound 22, with a P2 proline, only retained significant potency (> 80%) against trypsin. Compound 44, with a P2 serine, was active against calpain and cathepsins B and S. Compound 46, with a P2 piperidine, was the most selective compound, with the highest activity against cathepsin S.

The 4-bromobenzamide compound 22 was selected to further explore the role of MALT1 in a panel of DLBCL cells, and those results have been reported elsewhere.²⁰ It was found to have good mouse microsome (T_1/2 > 2 h) and mouse plasma stability (T_1/2 4.3 h), and showed moderate plasma levels for several hours upon IV and IP dosing in mice, potentially sufficient for a covalent inhibitor to bind its target and induce a pharmacodynamic effect. The compound was profiled in a number of PD and efficacy studies, where we observed in vivo inhibition of MALT1 activity, and where moderate effects on tumor growth in two ABC-DLBCL xenograft models were achieved.²⁰

In summary, we optimized the tetrapeptide tool compound Z-VRPR-fmk into a series of potent and cell permeable MALT1 inhibitors, incorporating a fluoromethyl ketone warhead for covalent binding to the active site cysteine. Profiling in a panel of proteases suggested that compounds of this type can achieve good overall selectivity within the target class, and we observed minimal anti-proliferative effects in our cellular counterscreen assay. A compound suitable for in vivo dosing was identified, although further development of the series will be required to deliver compounds with oral bioavailability.

Supplementary Material

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bmcl.2019.03.046.