A fragment-based electrophile-first approach to target histidine with aryl-fluorosulfates: application to hMcl-1

Abstract

Aryl-fluorosulfates are mild electrophiles that are very stable in biological media and in vivo and can efficiently react with the side chains of Lys, Tyr, or His residues, when properly juxtaposed by a high-affinity ligand. A more powerful approach to derive novel ligands would consist in starting from the covalent adduct and building the ligand off those initial interactions. While this strategy has been proven for Cys with molecular fragments containing Cys targeting electrophiles such as acrylamides, a corresponding strategy with fluorosulfates targeting His/Lys/Tyr residues has yet to be reported. We report here that a fragment library of aryl-fluorosulfates, when deployed with proper biophysical screening strategies, can identify initial covalent fragments. We report on novel strategies to enhance the success rate of such electrophile-based fragment screening for His/Lys/Tyr residues and to characterize the resulting hits. As an application we report on novel covalent fragment hits targeting hMcl-1 His 224.

Keywords: Histidine covalent; aryl-fluorosulfates; drug discovery; electrophile-first; fragment-based drug design; fragment-based ligand design; hMcl-1.

Figure 1.. Electrophile-first approach for covalent targeting of Lys, His, or Tyr residues.

A) The covalent bond formation between protein targets and aryl-fluorosulfates can lead to stable adducts including imidazole sulfamates (for His), amino sulfamates (for Lys), or sulfates (for Tyr). B) Schematic representation of the fragment library deployed to target His, Lys, or Tyr residues. C) MW distribution of the 320 fragments library. D) Chemical structures of representative compounds from the library. A complete list is provided as supplementary Table S1.

Figure 2.

Lys- and His-covalent BH3-based antagonists derived recently against hMcl-1. A) Surface representation of hMcl-1(172–323) highlighting the BH3 binding pocket and the binding site Lys 234, His 224, and His 252 residues. B) Ribbon representation of the crystal structure of the first covalent BH3-peptide in complex with hMcl-1 targeting Lys 234 using a sulfonyl fluoride. Peptide’s sequence is Ac-Dap(2Me,5FSB)IAEQLRRIGDRF-CONH₂ where 2Me,5FSB = 2-methyl, 5-sulfonyl fluoride (PDB ID 6VBX). C) Ribbon representation of the crystal structure of the first His 252 covalent stapled BH3 peptide in complex with hMcl-1. The peptide has sequence Ac-Dap(2MeO,5FSB)IAEQLRXIGDXF-CONH₂, where X indicates the hydrocarbon staple formed from two (S)-2-(4-pentenyl)Ala (metathesis reaction) (PDB ID 8VJP). D) Ribbon representation of the crystal structure of the first His 224 covalent BH3-peptide (named peptide 6) in complex with hMcl-1 (PDB ID 9CKN). Peptide 6 sequence is Ac-IAEQLRRIGDRZ-CONH₂ where Z is a fluorosulfate ((S)-2-amino-3-(4-((fluorosulfonyl)oxy)phenyl) propanoic acid).

Figure 3.

Denaturation thermal shift curves for hMcl-1(172–323) measured in absence and presence of various ligands. A) ΔTm curves for wt-hMcl-1(172–323) measured in absence (blue) and presence of a reversive binding peptide (red, peptide 7 of sequence Ac-IAEQLRRIGDRY-CONH₂) and its equivalent peptide 6 that targets hMcl-1(172–323) covalently via a fluorosulfate targeting His 224 (Figure 2). B) As in A) but the covalent peptide is tested against a hMcl-1(172–323) His224Ala mutant. C) Determination of Z’ factor based on repeated measurements for hMcl-1(172–323) in absence (blue) and presence (red) of a reference Bim BH3 peptide.

Figure 4.

Heteronuclear [¹⁵N,¹H] NMR correlation spectra and chemical shift mapping using ¹⁵N labeled hMcl-1(172–323) measured in absence and presence of fragment hit 2 at various incubation times. A) Overlay of 2D long-range [¹⁵N,¹H] correlation spectra for His side chains collected in absence (blue) or presence of fragment hit 2 after various incubation times (red, 1 h; green, 13 h; yellow, 25 h). B) Enlarged region of the spectra reported in A) highlighting the time-dependent large chemical shift perturbations of His 224 side chain resonances. Resonance assignments were obtained as we reported recently^, and further confirmed by single-point mutations (supplementary Figure S6). C) Backbone 2D [¹⁵N,¹H] correlation spectra for ¹⁵N-hMcl-1(172–323) measured in absence (blue) or presence of fragment hit 2 after various incubation times (red, 1 h; green, 25 h). Resonance assignments are reported in supplementary Figure S7. D) Mapping the observed backbone chemical shift perturbations into the 3D structure of hMcl-1(172–323) (blue) reveals that the compound is likely bound in proximity to His 224 in the BH3 binding pocket.

Figure 5.

Crystal structure of hMcl-1(172–323) in complex with fragment hit 2 solved at 1.82 Å resolution. A) Surface representation of hMcl-1(172–323) highlighting the position of fragment hit 2 covalently bound to the side chain of His 224. The other binding site nucleophilic residues His 252 and Lys 234 are also highlighted. The insert displays a close-up view of the imidazole sulfamate bond resulting from the reaction of the fluorosulfate of fragment hit 2 and the side chain of His 224. B) The 2Fo-Fc map of the bound fragment hit 2 contoured at 1.0 sigma, highlighting the contiguous electron density map between fragment hit 2 and the protein (PDB ID 9EFJ).

Figure 6.

Structure-activity relationships studies and inhibition properties of fragment hit 2 and analogs as determined by a DELFIA displacement assay. A) Percent inhibition in a DELFIA displacement assay with fragment hit 2 measured at various incubation times and different concentrations (10 μM, orange; 50 μM, green; 100 μM, blue). B) Percent inhibition in the DELFIA displacement assay for the synthesized compounds (chemical structures are shown) as determined at two concentrations (50 μM, green; 100 μM, blue) after 4 h incubation time. C) Percent inhibition in a DELFIA displacement assay with compound 165D9 (its chemical structure is shown) measured at various incubation times at 15 μM. D) Dose response displacement assay with compound 165D9 (24 h incubation time) leading to an IC₅₀ value of ~ 2.5 μM (supplementary Figure S9). E) Comparison of the structures of hMcl-1(172–323) in complex with his 2 (magenta) (PDB ID 9EFJ) and 165D9 (gray). The model of the complex with 165D9 was built using Sybyl-X based on the X-ray coordinates of fragment hit 2 in complex with hMcl-1(172–323) (PDB ID 9EFJ). The Cl atom in position 6 of the indane ring nicely occupies a deep hydrophobic pocket where the ligand binds. The hydrophobic residues are also highlighted surrounding the Cl atom are highlighted.

Figure 7.

Mass spectrometry analyses of the optimized hit 165D9 when tested against hMcl-1(172–232) and its mutants. The mass spectrometry data for panels A-D are tabulated also in supplementary Table S3. In some E.coli expressed hMcl-1(172–323) constructs we had previously observed a +178 that corresponds to the mass after phosphogluconosylation of the target^,.