Design and Synthesis of Selective Phosphodiesterase 4D (PDE4D) Allosteric Inhibitors for the Treatment of Fragile X Syndrome and Other Brain Disorders

Abstract

Novel pyridine- and pyrimidine-based allosteric inhibitors are reported that achieve PDE4D subtype selectivity through recognition of a single amino acid difference on a key regulatory domain, known as UCR2, that opens and closes over the catalytic site for cAMP hydrolysis. The design and optimization of lead compounds was based on iterative analysis of X-ray crystal structures combined with metabolite identification. Selectivity for the activated, dimeric form of PDE4D provided potent memory enhancing effects in a mouse model of novel object recognition with improved tolerability and reduced vascular toxicity over earlier PDE4 inhibitors that lack subtype selectivity. The lead compound, 28 (BPN14770), has entered midstage, human phase 2 clinical trials for the treatment of Fragile X Syndrome.

Figure 1:

(1) Burgin et al. compound D159687; (2) Naganuma et al. compound 33; (3) Hagen et al. compound 8. The aliphatic substituents on the core project into a pocket of the active site known as the Q region. A pair of aromatic arms projects from the core. These have been labeled Ar1 and Ar2 as in Burgin et al.

Figure 2:

PDE4D allosteric binding site. (A) view of a PDE4 dimer (PDB: 4WZI) showing the four-helix bundle and UCR2 from one subunit (colored red) positioned in trans- across the active site of the opposite subunit (colored blue). (B) Views of compounds 5 (PDB: 6NJI) and 6 (PDB: 6NJH) bound in the PDE4D allosteric site showing key polar contacts. Magnesium (green sphere), zinc (purple sphere), water (red spheres), UCR2 (cyan), and catalytic domain amino acid side chains (green).

Figure 3:

Timeline plot for chemical optimization of different cores and Q region substituents. Compound symbols are colored by linker type and inhibitory potency against PDE4D-S129D is indicated by the size of the symbol.

Figure 4:

Effect of amine and methylene linker replacements on PDE4 selectivity across multiple cores. The ratios of selectivity for PDE4B3-S133D/PDE4D7-S129D are plotted versus PDE4D7-S129(wt)/PDE4D7-S129D. The dotted line is the best-fit in the methylene linker series (Bravais Pearson test, r=0.677) for direct proportionality between selectivity for PDE4B-S133D versus PDE4D7(wt). Symbols: ● methylene linker, ● amine linker, ★ - apremilast, ▲ - rolipram; The relative size of the symbol indicates potency against PDE4D7-S129D. Units are in nM.

Figure 5:

Comparison of methylene (compound 23 - magenta) and amine (compound 6 – yellow) linkers. (A) Only residues from the compound 6 co-crystal structure are shown for simplicity as no major changes are observed in the side-chain rotamers. Waters associated with 6 (red) and 23 (blue) are shown with hydrogen bonds for 6 (black) and 23 (red). UCR2 is in cyan and the zinc atom is a gray sphere. (B) Overlay of 6 (yellow) and 23 (magenta) with 2Fo-Fc electron density contoured at 1σ (blue mesh) and Fo-Fc electron density contoured at 3σ (green/red mesh). Waters associated with 6 (red) and 23 (sand) are shown with the hydrogen bond unique to 6 highlighted in red.

Figure 6:

Mouse NOR and anesthesia duration test. (A) 23 significantly improved novel object discrimination in comparison to vehicle (VEH; Two-ANOVA F_(4,28) = 3.96, p<0.001) at doses of 1 and 3 mg/kg. (B) 28 significantly improved novel object discrimination (F_(6,42) = 4.18, p<0.005) at doses above 0.3 mg/kg PO. (C) 29 significantly improved novel object discrimination (F_(5,35) = 8.26, p<0.005) at doses above 3 mg/kg PO as did rolipram (ROL) at a dose of 0.1 mg/kg SC. (D) 23 significantly reduced the duration of anesthesia induced by ketamine and xylazine, a pharmacological construct of emetic potential (F_(3.27) = 20.9, p<0.001) at doses of 10 and 30 mg/kg PO. (E) 28 had no significant effect on anesthesia duration at doses of 3, 10 and 30 mg/kg PO (F_(4.36) = 1.08), while (F) 29 significantly reduced anesthesia duration at doses of 3 mg/kg PO and higher (F_(4,28) = 4.7, p<0.005). Rolipram significantly reduced anesthesia duration at 0.1 mg/kg IP. P-values shown are: ns, not significant; * p<0.01; ** p<0.05, and ***p<0.001 (Dunnett’s multiple comparison test).

Figure 7:

Comparison and selectivity of 23 and 28: (A) 23 with AR1 amide pointing up in the pocket to form water-mediated hydrogen bonds with the main-chain and side-chain of UCR2 (cyan) threonine275 (T275) and catalytic domain glutamine451 (Q451), histidine506 (H506) and waters surrounding the magnesium atom (green sphere). (B) 28 (PDB: 6NJJ) with AR1 carboxylic acid pointing down losing water-mediated hydrogen bonds with histidine506 (H506) and the side-chain of threonine275 (T275), but gaining water-mediated hydrogen bonds to serine449 (S449), asparagine450 (N450), and the main-chain of glutamine451 (Q451), while maintaining the main-chain interaction with UCR2 threonine275 (T275). (C) Modeling of tyrosine271 (Y271) in place of phenylalanine271 (Y271) onto the crystal structure of PDE4D and 28. The closest approach by phenylalanine271 to 28 is 4.4Å which is reduced with tyrosine to 3.2Å.

Figure 8:

Kinetics of PDE4 enzyme inhibition contrasting a PDE4D subtype selective allosteric inhibitor (28, top row) with apremilast (30), an active site-directed PDE4 inhibitor competitive with cAMP (bottom row). The low Hill slope of the allosteric inhibitor against the activated, dimeric forms of PDE4D indicates negative cooperativity between the two active sites in the PDE4 dimer. Negative cooperativity is reduced (Hill slope tends towards 1) when UCR2 is deleted as in the truncated PDE4D-cat enzyme. Apremilast inhibits equally well all PDE4 subtypes independent of their activation state and does not require UCR2 for binding. Compounds were tested in duplicate wells using a coupled enzyme assay. The Z’ quality factor for the in vitro inhibition assay was >0.6. Representative assays are shown plotting the well mean ± SD (note: the variance between duplicate wells may be less than the size of the symbols).

Scheme 1:

Synthetic schemes^{a a} Reagents and Conditions: a) EtCOCH₂CO₂Et, NaOMe; b) POCl₃; c) H₂NAr₁, heat; d) BrCH₂Ar₁, Pd(dppf)Cl₂; e) HOAr₁, K₂CO₃; f) 3-Cl PhB(OH)₂, Pd(Ph₃P)₄, K₃PO₄; g) 1) HSAr₁, Et₃N; 2) m-CPBA;