Designing pharmacological pVHL refolders by targeting a distinct surface pocket.

(A) Missense mutations observed in kidney cancer are broadly distributed across the VHL sequence, and mapping them to the protein structure (PDB 1LM8) shows that many of the 20 most frequently occurring mutations (red) are located in the hydrophobic core of the folded protein structure. (B) Overarching design strategy for re-folders / re-activators. In principle, ligands that preferentially bind to the folded/active conformation of a protein (F) will shift the conformation equilibrium towards this state, at the expense of unfolded (U) or misfolded (M) states. (C) The MAP druggability score is intended to provide a measure of how potently some (yet unknown) ligand might engage a given protein surface. With respect to pVHL, multiple simulations implicated the same surface pockets as having the highest MAP score. (D) The location of this druggable pocket is distinct from the binding surfaces of the Elongins, and of the HIF recognition site. (E) Building an exemplar indicates how an ideal ligand would engage this surface pocket. (F) Using this exemplar as a template, virtual screening provides models of specific ligands that complement this surface pocket.

Direct evidence of CP4 binding to the VCB complex.

(A) Chemical structure of CP4, as well as those of VH298 (positive control) and sitagliptin (negative control). (B) Orthogonal ligand-observed NMR spectra for three separate experiments: STD-NMR, WaterLOGSY, and CPMG. Each sample contains a mixture of CP4, sitagliptin, and VH298 in the presence (red) or absence (blue) of the VCB complex. Peaks are labeled based on whether they belong to CP4 (*), VH298 (+), or sitagliptin (-). (C) Based on our model of CP4 engaging the VCB complex, we elected to mutate VHL Asp197, a residue in the intended binding pocket. (D) The same three ligand-observed NMR assays were carried out using VCB complex harboring designed VHL mutant D197K. This mutant complex does not show any change in binding for VH298 or sitagliptin (relative to the WT VCB complex), but it does show a difference in binding for CP4. Due to the kinetic nature of these experiments, they cannot be used to differentiate between enhanced versus diminished CP4 binding. Full spectra from these experiments are included as Figures S5-S10.

CP4 stabilizes pVHL in ccRCC cell ines harboring VHL missense mutations.

(A) Our cellular studies use three cell lines that each harbor a different VHL missense mutations. The sites of these three mutations are structurally remote: P86 is in the central core, S65 is near the HIF-α binding site, and I180 is near the Elongin C binding surface. None of these positions are in direct contact with the CP4 designed binding site. (B) CP4 treatment for 1 hour leads to accumulation of both isoforms of mutant pVHL. (C) Relative to cells treated with 100μg/mL cycloheximide alone, cells in which 20 μM CP4 was added exhibited delayed loss of mutant pVHL.

CP4.29 degrades HIF-2α by re-activating mutant pVHL in ccRCC cell lines.

(A) Reaction scheme for preparing CP4 derivatives. (B) CP4 derivatives were screened at 20 μM in RCC-MF cells, by using ELISA to probe HIF-2α abundance after 2 hour treatment with each compound. (C) Treatment with CP4.29 for 2 hours led to depletion of HIF-2α in RCC4, RCC-MF, and 769-P cell lines. (D) CP4 and CP4.29 differ only slightly in chemical structure, at a site approximately 20 Å from substrate HIF-α in our model of target engagement. (E) Application of CP4.29 to 769-P cells reduced mRNA levels of HIF-2α target genes CCND1 and VEGF-A. (F) CP4.29 treatment for 4 hours induces HIF-2α degradation in RCC-MF cells under normoxic conditions (20% O2) but not hypoxic conditions (2% O2), consistent with pVHL-mediated recognition of HIF-2α. (G) CP4.29 accelerated depletion of HIF-2α in CHX-treated cell lines harboring mutant VHL, but not in their isogenic VHL-/-counterparts.

Activity of CP4.29 using mutant pVHL recapitulates the expected dependencies of recognition by WT pVHL.

(A) In 769-P cells (VHL I180N) treatment with CP4.29 induces depletion of HIF-2α, and also depletion of additional pVHL substrates AURKA and ZHX2. Analysis of HIF-2α was obtained after 2 hr treatment, whereas analysis of AURKA and ZHX2 were obtained after 24 hr treatment. (B) CP4.29’s depletion of all three substrates is mitigated by co-treatment with 1 μM neddylation inhibitor MLN4924, demonstrating that P4.29 activity is dependent on a cullin-RING E3 ligase. (C) CP4.29’s depletion of HIF-2α and ZHX (but not AURKA) is mitigated by co-treatment with 50 μM PHD inhibitor JNJ42041935, demonstrating that CP4.29 activity in this mutant VHL cell line recapitulates the molecular recognition patterns of WT pVHL. (D) CP4.29’s depletion of all three substrates is mitigated by co-treatment with 20 μM proteasome inhibitor MG132, consistent with a ubiquitin-mediated (proteasomal) path to their destruction.

Structures of compounds included in initial screen (CP1 to CP18).

Structures of CP4 derivatives included in second screen (CP4.1 to CP4.46).

Structures of CP4.29 derivatives included in third screen (CP4.29.1 to CP4.29.22).

Summary of computational workflow used to identify initial hit compounds CP1 to CP18.

1H NMR of CP4 in DMSO-d6 at 295K obtained on Bruker 600 MHz NMR.

13C NMR of CP4 in DMSO-d6 at 295K obtained on Bruker 600 MHz NMR.

1H NMR with water suppression (ZGESGP) of 100 μM CP4 (blue), 100 μM sitagliptin (red), or 100 μM VH298 (green) in 50 mM sodium phosphate buffer, 10% D2O, pH=6.95, 1% DMSO-d6.

Full STD NMR spectrum for 100 μM CP4 / 100 μM Sitagliptin / 100 μM VH298 / 1% DMSO-d6 in the absence (blue) and presence (red) of 7.7 μM VCB protein complex.

Full WaterLOGSY NMR spectrum for 100 μM CP4 / 100 μM Sitagliptin / 100 μM VH298 / 1% DMSO-d6 in the absence (blue) and presence (red) of 7.7 μM VCB protein complex.

Full CPMG NMR spectrum for 100 μM CP4 / 100 μM Sitagliptin / 100 μM VH298 / 1% DMSO-d6 in the absence (blue) and presence (red) of 7.7 μM VCB protein complex.

Full STD NMR spectrum for 100 μM CP4 / 100 μM Sitagliptin / 100 μM VH298 / 1% DMSO-d6 in the presence of 7.7 μM WT VCB protein complex (red) or 7.7 μM VD197KCB (green).

Full WaterLOGSY NMR spectrum for 100 μM CP4 / 100 μM Sitagliptin / 100 μM VH298 / 1% DMSO-d6 in the presence of 7.7 μM WT VCB protein complex (red) or 7.7 μM VD197KCB (green).

Full CPMG NMR spectrum for 100 μM CP4 / 100 μM Sitagliptin / 100 μM VH298 / 1% DMSO-d6 in the presence of 7.7 μM WT VCB protein complex (red) or 7.7 μM VD197KCB (green).

Treatment with CP4.29 (but not CP4) led to depletion of HIF-1α in RCC4, RCC-MF, and 769-P cells.

Parental cell lines RCC-MF, RCC4, and 769-P were each transfected with plasmids of pLenti-Cas9-GFP and were either provided with control sg-RNA or with sg-RNA for VHL.

The loss of pVHL expression each VHL KO cell line was confirmed by Western blot.

In RCC4 cells, (A) treatment with CP4.29 reduced the cellular abundance of HIF-2α. However, the activity of CP4.29 on HIF-2α could be reversed by co-treatment with either (B) proteasome inhibitor MG132, (C) VHL inhibitor VH298, or (D) neddylation inhibitor MLN4924.

In RCC4 cells, (A) treatment with CP4.29 reduced the cellular abundance of Aurora A. However, the activity of CP4.29 on Aurora A could be reversed by co-treatment with (B) proteasome inhibitor MG132, or (C) VHL inhibitor VH298.

Immunoprecipitation of HIF-2α (left), AURKA (middle), or ZHX2 (right) followed by Western blot analysis for Ub-1.

769-P cells (or isogenic 769-P VHL KO cells) were treated with 20 μM proteasome inhibitor MG132, and with 10 μM CP4.29 for 2 hours in the HIF-2α experiment or for 24 hours in the AURKA and ZHX2 experiments.

First-tier ADME characterization of CP4.29 indicates that further optimization is warranted prior to advancement into in vivo studies.

1H NMR of CP4.29 in DMSO-d6 at 295K obtained on Bruker 600 MHz NMR.

13C NMR of CP4.29 in DMSO-d6 at 295K obtained on Bruker 600 MHz NMR.