A possible allosteric mechanism by which a covalent ligand inhibits binding of other non-covalent ligands.

Crystal structures of T0070907-bound PPARγ LBD cobound with (a) non-covalent agonist MRL-24 (PDB 8ZFS) and (b) NCoR1 corepressor peptide (PDB 6ONI). Helix 12 (h12) can adopt a solvent exposed active conformation— or a solvent occluded repressive conformation within the orthosteric ligand-binding pocket that would physically clash and block binding of an orthosteric ligand.

Pharmacological covalent inverse agonists stabilize a repressive LBD conformation.

(a) Chemical structures of four covalent ligands with different pharmacological properties including transcriptionally neutral antagonism (GW9662) or transcriptionally repressive partial inverse agonism (T0070907) and a full inverse agonism (SR33065 and SR36708). (b) Cell-based luciferase reporter assay in HEK293T cells transfected with an expression plasmid encoding full-length PPARγ and a 3xPPRE-luciferase reporter plasmid (n=4 technical replicates; mean ± s.e.m.). One-way ANOVA using Dunnett’s multiple comparisons test was used to compare DMSO (control) to ligand treated conditions; **** = P ≤ 0.0001. (c,d) Fluorescence polarization (c) FITC-labeled NCoR1 corepressor and (d) FITC-labeled TRAP220/MED1 coactivator peptide binding assays (n=3 technical replicates) with fitted affinities shown in the legend (mean ± [95% CI]). (e) Location of Gly399 in the crystal structure of PPARγ LBD bound to NCoR1 peptide and inverse agonist T0070907 (PDB 6ONI). (f) One-dimensional (1D) traces extracted from two-dimensional (2D) [1H,15N]-TROSY-HSQC NMR data of 15N-labeled PPARγ LBD in the absence or presence of the indicated covalent ligands. The grey dotted line denotes the Gly399 backbone amide chemical shift in the apo/ligand-free form; black lines denote the active and repressive chemical shift values when bound to T0070907.

TR-FRET profiling of non-covalent ligand binding to PPARγ LBD.

(a) Chemical structures of non-covalent PPARγ partial agonists MRL-24 and nTZDpa and the full agonist rosiglitazone. (b) TR-FRET ligand displacement assay (n=3 technical replicates) with fitted Ki values shown in the legend (mean ± s.d.). (c,d) TR-FRET coregulator interaction assays where non-covalent agonists were titrated with in the presence of (c) 400 nM FITC-labeled NCoR1 corepressor or (d) 400 nM FITC-labeled TRAP220/MED1 coactivator peptide (n=3 technical replicates; mean ± [95% CI]; n.d. = not determined).

TR-FRET assay profiling of non-covalent ligand cobinding to PPARγ LBD with a covalent inhibitor and increasing NCoR1 peptide concentrations.

(a,b) TR-FRET NCoR1 corepressor peptide interaction assays where (a) MRL-24 or (b) nTZDpa were titrated with increasing concentrations of FITC-labeled NCoR1 corepressor peptide to saturate the peptide-bound forms of PPARγ LBD bound to the covalent ligands profiled in Figure 1 (n=3 technical replicates; mean ± s.d.). MRL-24 and nTZDpa Ki values are noted with vertical dotted orange lines, and a vertical grey line denotes log M = −7 as a visual cue to compare the concentration-response curves.

Corepressor peptide binding synergizes with covalent inhibitor inverse agonism to weaken non-covalent ligand cobinding.

Fitted parameters extracted from TR-FRET NCoR1 ligand cobinding assays (Figure 5—source data 1) with (a) MRL-24 or (b) nTZDpa including potency (IC50) and cooperativity (hill slope) values (n=2 biological replicates derived from a fit of TR-FRET data with n=3 technical replicates; mean ± s.d.). MRL-24 and nTZDpa Ki values are noted with vertical dotted orange lines.

Non-covalent ligand and covalent inhibitor cobinding shifts the PPARγ LBD conformational ensemble to an active state.

(a) 2D [1H,15N]-TROSY-HSQC data zoomed into the Gly399 backbone amide peaks of 15N-labeled PPARγ LBD bound individually or cobound to the indicated non-covalent and covalent ligands. The active (act) and repressive (rep) peaks when bound to covalent ligand alone are labeled; black arrows denote the shift in the peaks between the covalent ligand bound alone vs. cobound with the non-covalent ligand. (b,c) TR-FRET TRAP220/MED1 coactivator peptide interaction assays where (b) MRL-24 or (c) nTZDpa were titrated into PPARγ LBD pretreated with the indicated covalent inhibitors (n=3 technical replicates; mean ± [95% CI]; n.d. = not determined).

Non-covalent ligand cobinding occurs in the presence of covalent inhibitor SR16832.

(a) Chemical structure of the dual-site covalent inhibitor SR16832. (b) 2D [1H,15N]-TROSY-HSQC data zoomed into the Gly399 backbone amide peaks of 15N-labeled PPARγ LBD bound individually or cobound to MRL-24, nTZDpa, and SR16832 as indicated. Black arrows denote the shift in the peaks, or LBD conformation, between the covalent ligand bound alone vs. cobound with the non-covalent ligand. (c,d) TR-FRET coregulator interaction assays using (c) FITC-labeled NCoR1 corepressor or (d) FITC-labeled TRAP220/MED1 coactivator peptides where MRL-24 (orange) or nTZDpa (purple) were titrated with two concentrations of FITC-NCoR1 corepressor peptide to saturate the peptide-bound forms of PPARγ LBD bound to SR16832 (n=3 technical replicates; mean ± s.d.).

Model for improved covalent inhibitor efficacy.

Covalent ligands that when bound to PPARγ LBD increase the occupancy of helix 12 (h12) within the orthosteric pocket show improved efficacy of inhibiting other ligands from binding to the orthosteric pocket. Our data suggest an this is an allosteric mechanism in the sense that the covalent inhibitor does not physically clash with the non-covalent ligand itself; instead, increased occupancy of h12 within the orthosteric pocket resulting from the bound covalent inhibitor results in a clash or competition with a non-covalent orthosteric ligand.