Identification of compounds that bind the centriolar protein SAS-6 and inhibit its oligomerization

Abstract

Centrioles are key eukaryotic organelles that are responsible for the formation of cilia and flagella, and for organizing the microtubule network and the mitotic spindle in animals. Centriole assembly requires oligomerization of the essential protein spindle assembly abnormal 6 (SAS-6), which forms a structural scaffold templating the organization of further organelle components. A dimerization interaction between SAS-6 N-terminal "head" domains was previously shown to be essential for protein oligomerization in vitro and for function in centriole assembly. Here, we developed a pharmacophore model allowing us to assemble a library of low-molecular-weight ligands predicted to bind the SAS-6 head domain and inhibit protein oligomerization. We demonstrate using NMR spectroscopy that a ligand from this family binds at the head domain dimerization site of algae, nematode, and human SAS-6 variants, but also that another ligand specifically recognizes human SAS-6. Atomistic molecular dynamics simulations starting from SAS-6 head domain crystallographic structures, including that of the human head domain which we now resolve, suggest that ligand specificity derives from favorable Van der Waals interactions with a hydrophobic cavity at the dimerization site.

Keywords: biophysics; centrosome; inhibitor; ligand design; nuclear magnetic resonance (NMR); protein-drug interaction; protein-protein interaction.

Figure 1

A pharmacophore model of CeSAS-6_N dimerization.A, the dimer of CeSAS-6_N, where the β6-β7 loop of one domain (light blue) protrudes into a hydrophobic pocket of the second domain (gray) formed by residues on α1, α2, and β7. Derived from PDB ID 3PYI. B, magnified view of the dimerization interface, corresponding to the boxed area in (A), with key residues annotated and hydrogen bonding interactions show as dashed red lines. C, a structure-based pharmacophore model depicting the interactions mapped. Hydrophobic features are depicted in cyan, aromatic in orange, hydrogen bond donors in blue, and a hydrogen bond acceptor in red. Arrows depict the direction of the hydrogen bond vectors (purple from donors and green from acceptor).

Figure 2

NMR assay of CeSAS-6_N small molecule interactors.A–C, overlay of NMR ¹H-¹⁵N HSQC spectra produced by CeSAS-6_N Δ103–130 I154E alone (cyan) or in the presence of 2 mm A11 (A), A12 (B), or B1 (C) compounds. The chemical structures of the compounds are inlaid, and CeSAS-6_N resonances exhibiting the most significant changes upon addition of compounds are labeled. D–F, per-residue quantification of combined changes in ¹H and ¹⁵N chemical shifts of CeSAS-6_N Δ103–130 I154E resonances upon addition of 2 mm A11 (D), A12 (E), or B1 (F) compounds. A measure of two standard deviations of all changes observed is shown as a red line, indicating amino acids that experienced the strongest perturbations upon compound addition. G, CeSAS-6_N Δ103–130 monomer structure with the dimerization site targeted by compounds indicated by a light blue circle, derived from PDB ID 4G79. H–J, amino acid residues strongly perturbed by addition of compounds A11 (H), A12 (I), or B1 (J), shown in sphere representation. Only A11 produces changes in amino acids surrounding the targeted site, suggesting compound binding at the hydrophobic cavity of the dimerization interface.

Figure 3

Assays of CeSAS-6_N-CC oligomerization.A, sedimentation velocity analysis of CeSAS-6_N-CC S123E/I154W in the presence of 5% v/v DMSO (black solid and dashed lines) or upon addition of 2 mm A11 (red solid and dashed lines). Solid versus dashed lines correspond to two independent repeats of the assay. The molecular architecture of CeSAS-6_N-CC and the oligomeric state of different sedimentation species are shown schematically on top (spheres: head domains; zigzag lines: coiled-coil segments). Stable CeSAS-6 dimers are mediated by the coiled-coil segment. Addition of A11 reduces the prevalence of higher oligomeric species, which necessitate interactions between head domains. B–D, representative electron micrographs of CeSAS-6_N-CC S123E/I154W spiral assemblies formed at 1 mg/ml protein concentration and standard buffer (B) or in the presence of 5% v/v DMSO (C) or 2 mm A11 (D). Common scale bar in panel B 200 nm. E, plot of assembly length versus number of occurrences from the CeSAS-6_N-CC S123E/I154W samples shown in (B–D). Addition of A11 strongly reduces the number and length of visible assemblies.

Figure 4

NMR assays of CrSAS-6_N and HsSAS-6_N compound binding.A–C, overlay of NMR ¹H-¹⁵N HSQC spectra produced by CrSAS-6_N F145E alone (red) or in the presence of 2 mm A11 (A), A12 (B), or B1 (C) compounds. The chemical structure of compounds is inlaid, and CrSAS-6_N resonances exhibiting the most significant changes upon addition of compounds are labeled. D, CrSAS-6_N monomer structure with the dimerization site indicated by a light blue circle, derived from PDB ID 3Q0Y. E–G, amino acid residues strongly perturbed by addition of compounds A11 (E), A12 (F), or B1 (G), shown in sphere representation. Only A11 produces changes in amino acids surrounding the targeted site, suggesting compound binding at the hydrophobic cavity of the dimerization interface. H–N, similar NMR spectra of HsSAS-6_N F131E (H–J, blue) with compounds A11, A12, and B1, and structure representations of the HsSAS-6_N dimerization site (K) and residues affected by compound binding (L–N). The HsSAS-6_N structure was resolved as part of this study (Fig. S8 and Table S2). In contrast to CeSAS-6_N and CrSAS-6_N, both A11 and A12 bind the dimerization site of HsSAS-6_N.

Figure 5

MD simulations of A11 and A12 binding to SAS-6 head domains.A, superposition of representative states from MD simulations of CeSAS-6_N (light blue), CrSAS-6_N (red), and HsSAS-6_N (blue) with A11. The ligand, shown as sticks and with carbon atoms colored similar to the respective protein, adopts a consistent binding pose relative to head domains (schematic representation), with the compound aromatic group buried in the hydrophobic cavity of the head domain dimerization site. B, key interactions forming between A11 and CeSAS-6_N seen in MD simulations. Hydrogen bonds are shown as red dashed lines. A11 and protein amino acids participating in hydrogen bonds are shown in stick representation. The sidechains of amino acids involved in hydrophobic interactions are shown as spheres. The ligand is shown as sticks with carbon atoms in cyan. C–E, similar views of interactions between A11 and CrSAS-6_N (C) or HsSAS-6_N(D), and A12 with HsSAS-6_N (E). The A12 ligand carbon atoms are colored light green. F, ligand contact maps from the HsSAS-6_N and CrSAS-6_N MD simulations with A12. A12 atom nomenclature is shown alongside a stick representation of the ligand (right panel). The HsSAS-6_N (middle panel) and CrSAS-6_N (left panel) maps show close contacts between protein residues and ligand atoms as annotated. Different densities (blue hues) denote the occurrence of close contacts during the simulation, normalized to the most frequently observed contact (assigned value of 1, dark blue). Dashed circles highlight close contacts of the A12 methyl group in the HsSAS-6_N simulation that are absent in CrSAS-6_N.