Identification of small-molecule protein-protein interaction inhibitors for NKG2D

Abstract

NKG2D (natural-killer group 2, member D) is a homodimeric transmembrane receptor that plays an important role in NK, γδ⁺, and CD8⁺ T cell-mediated immune responses to environmental stressors such as viral or bacterial infections and oxidative stress. However, aberrant NKG2D signaling has also been associated with chronic inflammatory and autoimmune diseases, and as such NKG2D is thought to be an attractive target for immune intervention. Here, we describe a comprehensive small-molecule hit identification strategy and two distinct series of protein-protein interaction inhibitors of NKG2D. Although the hits are chemically distinct, they share a unique allosteric mechanism of disrupting ligand binding by accessing a cryptic pocket and causing the two monomers of the NKG2D dimer to open apart and twist relative to one another. Leveraging a suite of biochemical and cell-based assays coupled with structure-based drug design, we established tractable structure-activity relationships with one of the chemical series and successfully improved both the potency and physicochemical properties. Together, we demonstrate that it is possible, albeit challenging, to disrupt the interaction between NKG2D and multiple protein ligands with a single molecule through allosteric modulation of the NKG2D receptor dimer/ligand interface.

Keywords: DNA-encoded library screening; NKG2D; inflammation; protein–protein interaction.

Fig. 1.

Cellular, biochemical, and biophysical data for two NKG2D inhibitors identified from cellular HTS and DNA-encoded library screens. (A) Chemical structures of NKG2D inhibitors 1a and 3b. (B) Concentration–response curve of compound 1a identified in the cellular TR-FRET NKG2D/MICA HTS campaign. (C) Representative concentration–response curves for compound 1a (blue squares) and 3b (red circles) in the NKG2D/MICA biochemical TR-FRET assay. (D) SPR sensorgram for compound 1a binding to human NKG2D. Association (ka) and dissociation (kd) rates were 7.64e+02 M⁻¹.s⁻¹ and 3.69e-02 s⁻¹, respectively. (E) SPR sensorgram for 1a binding to human MICA. (F) Representative concentration–response curves for the inhibition of NKG2D binding to MICA (filled symbols) or ULBP6 (labeled as U6, open symbols) by resynthesized compound 1a (blue squares) or compound 3b (red circles) in the cellular NKG2D/MICA and NKG2D/ULBP6 TR-FRET assays. Mean IC₅₀s and SDs for 1a in the cellular NKG2D/MICA and NKG2D/ULBP6 TR-FRET assays were 10.2 ± 2.9 µM and 13.6 ± 4.6 µM, respectively. IC₅₀s for 3b are shown in Table 1. (G) Concentration–response curves for the inhibition of NKG2D-mediated cell killing of MICA-expressing cells by compound 1a (blue squares) and 3b (red circles) in the KHYG-1/Ba/F3 coculture assay. (H) SPR sensorgram for binding of compound 3b to human NKG2D. Association (ka) and dissociation (kd) rates were 1.44e+02 M⁻¹.s⁻¹ and 5.67e-04 s⁻¹, respectively). (I) SPR sensorgram for 3b binding to human MICA. All error bars represent SDs from at least three independent replicates.

Fig. 2.

Inhibitor-bound NKG2D structures revealed a cryptic pocket at the homodimer interface. Cross sections through NKG2D structures (monomer A and B colored gray and yellow, respectively) complexed with (A, Left) MICA (colored orange), (A, Middle) 1a (colored blue), and (A, Right) 3e (colored green). Close-up view of the induced pockets created by (B) 1a and (C) 3e. Residues that undergo conformational changes are colored magenta. Analysis of compound induced splaying of the NKG2D ECD; (D, Left) overlay of the displacement of the center of mass and the angle of rotation for monomer A of 1a (blue) and 3e (green) bound NKG2D compared with APO/MICA bound ECD after superpositioning onto monomer B. The rotation vector is represented with blue arrows. Structural overlays of monomer B from (D, Middle) 1a and (D, Right) 3e bound NKG2D with the MICA-bound NKG2D (PDB 1HYR; both NKG2D monomers colored orange; MICA not shown).

Fig. 3.

Optimization of the DEL series. Close-up view of (A) 3g and (B) 4a bound at the homodimer interface of NKG2D. The methyl pyrazole of 3g was fit with two alternative orientations as indicated with the double-headed arrow. NKG2D monomers A and B are colored gray and yellow, respectively, key residues are colored magenta and H-bonds are represented as black dashed lines. Conformers selected by NMR conformational analysis in dimethyl sulfoxide (DMSO) solution for (C) 4a and (D) 4b superposed on the X-ray structure conformer of 4a (green).

Fig. 4.

Symmetry inspired optimization of DEL series. (A) Two molecules of 3e related by a twofold symmetry axis at the center of the NKG2D homodimer as observed in the electron density maps (SI Appendix, Fig. S8). (B) Overlay of 4d (colored violet), 4e (colored blue), and 4f (colored orange) in the cryptic pocket depicted with a surface representation (calculated by Hollow). (C) Close-up view of the overlay from panel B.