Hypoxia-selective allosteric destabilization of activin receptor-like kinases: A potential therapeutic avenue for prophylaxis of heterotopic ossification

Abstract

Heterotopic ossification (HO), the pathological extraskeletal formation of bone, can arise from blast injuries, severe burns, orthopedic procedures and gain-of-function mutations in a component of the bone morphogenetic protein (BMP) signaling pathway, the ACVR1/ALK2 receptor serine-threonine (protein) kinase, causative of Fibrodysplasia Ossificans Progressiva (FOP). All three ALKs (-2, -3, -6) that play roles in bone morphogenesis contribute to trauma-induced HO, hence are well-validated pharmacological targets. That said, development of inhibitors, typically competitors of ATP binding, is inherently difficult due to the conserved nature of the active site of the 500+ human protein kinases. Since these enzymes are regulated via inherent plasticity, pharmacological chaperone-like drugs binding to another (allosteric) site could hypothetically modulate kinase conformation and activity. To test for such a mechanism, a surface pocket of ALK2 kinase formed largely by a key allosteric substructure was targeted by supercomputer docking of drug-like compounds from a virtual library. Subsequently, the effects of docked hits were further screened in vitro with purified recombinant kinase protein. A family of compounds with terminal hydrogen-bonding acceptor groups was identified that significantly destabilized the protein, inhibiting activity. Destabilization was pH-dependent, putatively mediated by ionization of a histidine within the allosteric substructure with decreasing pH. In vivo, nonnative proteins are degraded by proteolysis in the proteasome complex, or cellular trashcan, allowing for the emergence of therapeutics that inhibit through degradation of over-active proteins implicated in the pathology of diseases and disorders. Because HO is triggered by soft-tissue trauma and ensuing hypoxia, dependency of ALK destabilization on hypoxic pH imparts selective efficacy on the allosteric inhibitors, providing potential for safe prophylactic use.

Keywords: ACVR1; ALK2; Allosteric; BMP; BMPRII; Bone morphogenetic protein; FKBP12; FOP; Fibrodysplasia ossificans progressiva; Heterotopic ossification; Hydrophobic tagging; Hypoxia; Hypoxic pH; PROTACs; Pharmacological chaperone; Protein kinase; R-spine; Selective degrader; Small molecule kinase inhibitor; αC-β4 loop.

FIGURE 1.. Surface pocket targeted by virtual screening of drug-like compound library.

(A) View of the targeted pocket of the αC-β4 loop in the N-lobe of ALK2. (B) Side-by-side view of targeted pocket and active site cleft, the binding site for the ATP•Mg²⁺ complex and steric (competitive) inhibitors such as dorsomorphin (DM). (C) Zoomed view of pocket with C28 docked. Dashed lines represent five potential H-bonds between the donor guanidino group of the Arg258 side chain and the sp2-hybridized nitrogen acceptor of the heterocyclic aminopyridine ring or “warhead” of C28. (D) Further zoomed and slightly tilted view to highlight the proposed role of Arg258 in destabilization of the kinase. (E) Interface of the FKBP12:ALK2 complex (crystal structure; PDB ID 3H9R). Note lack of direct interactions with or proximity to the compound. (F) Interface of the BMPRII:ALK2 complex (predicted by computational docking) adjacent to the Arg258 reside and flanking main chain, approximately one-half of the αC-β4 loop.

FIGURE 2.. Compound 24 (C24) H-SAAD/D in vitro destabilization of ALK2 receptor kinase.

(A) pH-and temperature-dependence of loss of native structure of the recombinant type I kinase protein analyzed by nondenaturing PAGE assay after 30 min incubation with 1 mM C24 (w/o ATP•Mg²⁺). (B) pH-dependence of efficacy of C24 thermal denaturation of ALK2 kinase as determined by differential scanning fluorimetry, also known as thermofluor or thermal shift assay (w/o ATP•Mg²⁺; T ramp from 4°C; 2.5% DMSO; 10 mM C24 stock solubilized in DMSO). Midpoints of unfolding are in parentheses (°C).

FIGURE 3.. Specificity of H-SAAD/D destabilization mechanism demonstrated by the insensitivity of FKBP12 to C24-related C14.

Native PAGE analyses of ALK2 kinase at three molar equivalents (lanes 2 – 4, 5 – 7, 8 – 10) with respect to a fixed mass of FKBP12 (cf. lane 1). C14 was added to mixtures of the two proteins at 1 mM from a 10 mM stock solubilized in DMSO (lanes 4,7,10). All samples were incubated at ~ pH 7.5 and 37°C for 30 min prior to electrophoresis.

FIGURE 4.. Specificity of H-SAAD/D destabilization mechanism demonstrated by the resistance of BMPRII to C24.

Comparative analyses of C24 destabilization of type I (ALK2) and type II (BMPRII) BMP receptor kinases, qualitatively by native PAGE (left) and quantitatively by thermofluor analysis (right; w/o ATP•Mg²⁺; T ramp from 25°C; 5.0% DMSO; 10 mM C24 stock solubilized in DMSO. Native PAGE samples were incubated at ~ pH 7.0 and 37°C for 30 min prior to electrophoresis.

FIGURE 5.. Chemical structures of C14 and C24 virtual screen hits and four other commercially available derivatives.

(A) C14 and C24, methyl isoxazole and dimethyl oxazole warhead groups, respectively. C24THI and C24PYR, azole rings substituted with thiazole and pyrimidine. C24PYR-pCOOH, para-substituted benzoic acid ring at terminus opposite putative warhead groups. C24PYR-PLNR, substitution of a bifurcated five- and six-membered alkane basket of the central scaffold with a single planar, five-membered arene ring. Nitrogen atoms in heterocyclic aromatic rings have a single lone pair of electrons projecting out perpendicular to the ring. (B) Schematic representation of sp2 and sp3 lone pairs. (C) Lateral projection of PLNR derivative is compact and lies in plane. (D) Formation of disodium salts. C14/C24 family members have two acidic hydrogen atoms that can be neutralized with two equivalents of NaOH.

FIGURE 6.. Native PAGE analyses of relative destabilization activities for small SAR study.

The effects of concentration series of C14 and C24 virtual screen hits and four commercially available derivatives were compared at two pH that approximate normoxic (7.25) and hypoxic (6.75) intracellular conditions.

FIGURE 7.. Thermofluor analyses of C24THI efficacy at three pH, with and without counter-stabilizing effects of ATP•Mg²⁺ binding in the active site cleft.

(A) Melting curves for a range of C24THI concentrations up to 1 mM at three pH, w/o and with ATP•Mg²⁺. (B) Negative shifts in Tm of ALK2 kinase protein with ATP•Mg²⁺ by C24THI. Values in parentheses refer to Tm without compound (at zero), or residuals (0.999) from third order curve fits. (C) Structure of ALK2 kinase, depicting the putative interacting Arg258 and ionizing His318 sidechains imparting H-SAAD/D- and pH-sensitivities, respectively, which are not shared with BMPRII (D).

FIGURE 8.. pH-dependent inhibition by C24 of surrogate Smad-substrate phosphorylation.

(top) Coomassie-stained SDS-PAGE of ALK2 kinase and casein substrates with and without C24 at four pH spanning the physiological range. (bottom) [γ−³³P]ATP autoradiogram of autophosphorylated ALK2 kinase and three (α/β, κ) casein polypeptides.