Discovery of Potent Coumarin-Based Kinetic Stabilizers of Amyloidogenic Immunoglobulin Light Chains Using Structure-Based Design

Abstract

In immunoglobulin light-chain (LC) amyloidosis, transient unfolding or unfolding and proteolysis enable aggregation of LC proteins, causing potentially fatal organ damage. A drug that kinetically stabilizes LCs could suppress aggregation; however, LC sequences are variable and have no natural ligands, hindering drug development efforts. We previously identified high-throughput screening hits that bind to a site at the interface between the two variable domains of the LC homodimer. We hypothesized that extending the stabilizers beyond this initially characterized binding site would improve affinity. Here, using protease sensitivity assays, we identified stabilizers that can be divided into four substructures. Some stabilizers exhibit nanomolar EC₅₀ values, a 3000-fold enhancement over the screening hits. Crystal structures reveal a key π-π stacking interaction with a conserved tyrosine residue that was not utilized by the screening hits. These data provide a foundation for developing LC stabilizers with improved binding selectivity and enhanced physicochemical properties.

Figure 1.

Examples of FL LC stabilizers that were identified from our earlier high-throughput screen.

Figure 2.

Modular design of FL LC stabilizers. A) Representation of the JTO-FL LC dimer kinetic stabilizer binding site, showing two distinct pockets that can be occupied by ligand, the core hydrophobic pocket (left) and the distal aromatic pocket (right). A LC stabilizer that occupies both pockets can be divided into four different regions that are color-coded. Interactions of the LC stabilizer substructures with FL LC residues are shown. Potential hydrogen bonds between the kinetic stabilizer and LC protein are indicated with dashed lines. The prime labels for some residues denote residues comprising the second polypeptide chain of the LC homodimer. B) The A, B, C and D substructures are color-coded on the two LC stabilizers 1 and 5 that have published crystallographic data (PDB: 6MG5 and 6W4Y, respectively). The coumarin ring of 1 is numbered for reference. The isobutyl group of 5 serves only as a prototype “distal substructure” because it does not fully engage with Y49’. Only the R enantiomer of 5 is observed in the crystal structure, but the compound was used as a racemate. C) The color-coded substructures of stabilizers 1 and 5 are indicated in the superimposed crystal structures of JTO•1 (PDB: 6MG5) and JTO•5 (PDB: 6W4Y).

Figure 3.

Crystal structure of carbamate stabilizer 26 (magenta) in complex with JTO-FL (PDB: 7LMN). A) Line drawing of 26. B) Comparison of the binding modes of the “anchor substructure” and the “aromatic core” of 1 (orange) and 26 (magenta; PDB: 6MG5). C) Interactions between the carbamate linker module and distal substructure of 26 with the solvent-exposed linker binding site and the distal aromatic pocket. Potential hydrogen-bonding interactions and measured distances are shown in black dashed lines. D) A focused view of the distal substructure-Y49’ interaction.

Figure 4.

Crystal structures of JTO-FL in complex with spiro-hydantoin stabilizers 34 and 36. Potential hydrogen-bonding interactions with measured distances are shown in black dashed lines. A) Line drawings of 34 (R enantiomer) and 36. Only the R enantiomer of 34 is observed in the electron density (see Figure S2), but the compound was used as a racemate. B) JTO-FL•34 crystal structure (with stabilizer colored yellow, PDB: 7LMO). C) JTO-FL•36 crystal structure (with stabilizer colored light blue, PDB: 7LMP). D) Overlay of JTO-FL•34 and JTO-FL•36 structures with the coloring scheme as in panels B and C.

Figure 5.

Crystal structures of JTO-FL in complex with coumarin stabilizers 62 (green, PDB: 7LMQ) and 63 (blue, PDB: 7LMR), harboring ether anchor substructures. A) Line drawings of 62 (R enantiomer) and 63 (S enantiomer), which are the only stereoisomers observed in the electron density (Figure S2), although the stabilizers were synthesized and tested as racemates. B) Expansion of the anchor cavity is observed in the presence of an ether-containing anchor substructure also comprising a phenyl ring (for example, 62) compared to a diethylamino substructure (1, orange, PDB: 6MG5). Residues Q38 and Y87’, which undergo conformational changes enabling the cavity enlargement, are shown. C-E) Close-ups of the core hydrophobic cavity of JTO-FL with an ether bound in the anchor cavity. Residues that comprise the binding site, consisting of the “anchor cavity” and “aromatic slit”, are labeled. C) JTO-FL•62 D) JTO-FL•63 E) Overlay of JTO-FL•62 and JTO-FL•63 F) Alignment of 1, 62, and 63 in their binding poses in JTO-FL from the crystal structures.

Figure 6.

Docking model of stabilizer (R)-83 in complex with JTO-FL, using the protein conformation observed in the crystal structure of JTO•62. A) Line drawing of (R)-83. B) Comparison of a docked pose of (R)-83 (red) with the crystallized pose of 62 (light gray, PDB: 7LMQ), emphasizing the anchor substructures and aromatic cores. C) Comparison of a docked pose of (R)-83 with the pose of 36 from the crystal structure (dark gray, PDB: 7LMP), emphasizing the linker modules and distal substructures.

Figure 7.

Molecular dynamics (MD) simulations of JTO-FL•53. A) Line drawing of stabilizer 53. B-C) The three docked poses of 53 that were used as starting coordinates for the MD simulations are depicted in red, orange, and green. Representative structures of the consensus binding mode seen in all independent MD simulations are depicted in gray. The B) and C) panels show different views of the same data.

Scheme 1.

Synthesis of 7-dialkyl-4-methylcoumarins 9 – 13^a. ^aReagents and conditions: a) alkyl bromide or iodide, DMF, 80 °C, 2–16 h.

Scheme 2.

Synthesis of carbamates 18 – 27^a. ^aReagents and conditions: a) N-bromosuccinimide, MeCN, rt, 3 h; b) potassium trifluoro(2-((tetrahydro-2H-pyran-2-yl)oxy)ethyl)borate, Pd(OAc)₂, cataCXium® A, Cs₂CO₃, 1,4-dioxane, H₂O, 100 °C, 12 h; c) HCl, 1,4-dioxane, 25 °C, 30 min; d) p-nitrophenyl chloroformate, pyridine, DCM, rt, 3 h; e) R-NH₂, DIPEA, DMF, 25 °C, 2 h; f) 3-bromopyridin-2(1H)-one, Pd(PPh₃)₄, K₂CO₃, 3:1 dioxane/H₂O, 90 °C, 8 h.

Scheme 3.

Synthesis of spiro-hydantoin analogs 34 – 46^a. ^aReagents and conditions: a) KCN, (NH₄)₂CO₃, EtOH, H₂O, 60 °C, 16 h; b) 16, PPh₃, diisopropyl azodicarboxylate, THF, 0 °C to 25 °C, 4 h; c) HCl, MeOH, H₂O, 25 °C, 16 h; d) R₁-CO₂H, HATU, DIPEA, DMF, 25 °C, 16 h; e) R₂-SO₂Cl, DIPEA, DMF, 0 °C, 3 h; f) R₃-Cl, DIPEA, DMF, 100 °C, 16 h.

Scheme 4.

Synthesis of aldehyde intermediates 47, (R)-70, (S)-70, 71, and 73^a. ^aReagents and conditions: a) Dess-Martin Periodinane, DCM, 0 °C to 25 °C, 2 h; b) NBS, MeCN, 50 °C, 2 h; c) BBr₃, DCM, −78 to 30 °C, 6 h; d) 2-[(E)-2-ethoxyvinyl]-4,4,5,5-tetramethyl-1,3,2-dioxaborolane, Pd(dppf)Cl₂, K₂CO₃, dioxane, H₂O, 100 °C, 2 h; e) R₁-OH, DIAD, PPh₃, THF, 30 °C, 1 h; f) 3M HCl in acetone, 30 °C, 30 min; g) (tributylstannyl)methanol, Pd(PPh₃)₄, dioxane, 100 °C, 12 h; h) MnO₂, DCM, 30 °C, 24 h.

Scheme 5.

Synthesis of spiro-ureas 51 – 53 and 83 – 85 (separate enantiomers). (R)-83 and (S)-83 were prepared from enantiomerically pure alcohols, while 84 and 85 were prepared from racemic 1-phenyl-1-propanol and then enantiomerically resolved using supercritical fluid chromatography^a. ^aReagents and conditions: a) tert-butyl 4-amino-4-(aminomethyl)piperidine-1-carboxylate, NaBH(OAc)₃ or NaBH₃CN, AcOH, 1,2-DCE or MeOH, 25 °C, 16 h; b) carbonyldiimidazole, DCM, 25 °C, 24 h; c) HCl, MeOH, H₂O, 25 °C, 16 h; d) R-Cl, DIPEA, DMF, 100 °C, 16 h.

Scheme 6.

Synthesis of coumarin ethers 55 – 59 and morpholine analogs 62 – 63 for crystallography^a. ^aReagents and conditions: a) R-Br or R-OTs, DMF, K₂CO₃, 80 °C, 3 to 16 h; b) N-bromosuccinimide, MeCN, 25 °C, 16 h; c) potassium (morpholin-4-yl)methyltrifluoroborate, Pd(dba)₂, RuPhos, Cs₂CO₃, toluene, H₂O, 90 °C, 16 hr.