Figure 1.
The structure of the wild type α1-antitrypsin.
Front (A) and back (B) views of the structure of A1AT in cartoon representation (PDB entry: 1qlp). The secondary elements are coloured as follows. ß-sheets: A (red), B (blue), and C (yellow); helices: A (cyan), B (apricot), C (blue), D (grey-green), E (purple), F (yellow), G (orange), H (pink), I (olive); loops: reactive centre loop (RCL, red), all other loops (green).
Figure 2.
The nine top-ranking surface pockets identified by SiteMap on α1-antitrypsin.
Coloured spheres represent the SiteMap predictions for eight top-ranking surface clefts on the wild type α1-antitrypsin (PDB entry 1qlp, in grey cartoon representation): site A: green, B: cyan, C: blue, D: purple, E: fuchsia, F: orange, G: slate blue, H: brown. The yellow spheres correspond to the ninth site, I, a cleft identified on crystal structures of A1AT containing the Ala70Gly mutation.
Table 1.
The dataset of selected crystal structures of A1AT used in this study.
Figure 3.
Properties of surface pockets in crystal structures and in silico conformers of α1-antitrypsin.
Persistence of clefts A–I among A1AT crystal structures (A) and computationally produced conformers (B). Where the sites C and E overlapped, the data are presented under the label “C_E”. The distribution of SiteMap calculated properties for the 100 in silico conformers are shown as boxplots: SiteScore (C), DScore (D), site volume (E) and hydrophobic vs. hydrophilic character balance (F). The corresponding data for crystal structures are shown as red symbols superimposed on the boxplots; 1qlp (circle), 2qug (plus sign), 3cwm (square), 1hp7 (diamond), 3drm (triangle point up), 1oph (triangle point down). Data are shown only for sites identified within PDB entries for native (stressed, ‘S’) forms of A1AT, as these are likely to be the appropriate target states for the design of polymerization inhibitors.
Figure 4.
The pocket-lining propensity of the residues of α1-antitrypsin calculated with Provar.
Ribbon representation of A1AT (front, A and back, B) coloured by the residue-based Provar probabilities. Provar colours each protein residue according to its probability of being pocket-lining in an ensemble of conformers (here, 100 CONCOORD-produced conformations of A1AT). The first (0.05) and third quartile (0.92) of the probability distribution are used as the white and red limits of the spectrum respectively. Hence, residues appearing red belong to the top quartile distribution, i.e., in this case, they are pocket-lining in more than 92% of the conformers. (C) and (D): The SiteMap predictions for two pockets (A and I respectively) are shown as solid spheres, and every residue with an atom within 3.75 Å of any sphere is shown in stick representation coloured by its Provar value. Depth-cueing has been switched off in these figures to preserve the variation in the colouring of the residues.
Figure 5.
Fragment docking to the A site targets the pharmacophore defined by Asn104, Thr114, and His139.
Best poses of the top-scoring 20 fragments (coloured sticks) from the ZINC dataset docked in the A site of A1AT (cartoon, blue). The majority of these fragments fill the pocket defined by Thr114 and Asn104 at the top, and His139 at the bottom (thin sticks, cyan), identified in our previous study as a potential allosteric site for targeting A1AT polymerization. Some of the fragments take advantage of hydrogen bonding opportunities presented by His139 and Thr114.
Figure 6.
Results from docking the DrugBank collection against nine pockets on α1-antitrypsin.
(A) Boxplot distributions of docking scores for DrugBank molecules docked to each of the nine sites A to I. Only the top-ranking pose is included for each ligand and only ligands of molecular weight less than 500 Daltons are included in this plot. (B) The best-scoring ligand for each site is assigned a worse score when docked against each of the other sites. The red diamonds represent the best docking score for each ligand depicted in Table 2, when docked to the site where it is ranked top. The black diamonds correspond to the scores for each of these ligands when docked to all other sites. The x-axis labels correspond to the DrugBank ID of the ligand and, in brackets, the site against which it is selected as “best-scoring”, e.g. 07124(A) refers to DrugBank entry DB07124 which achieves its best score against site A.
Table 2.
The best-scoring and “best-efficient” small molecules from DrugBank docked against each of the sites A-I on A1AT.
Figure 7.
Top-scoring DrugBank molecules against the α1-antitrypsin sites.
IUPAC names and PubChem CIDs for the DrugBank IDs in Figure 6 and Table 2 are: DB07124: 3-[(2S)-2-amino-3-[(5-{7H-pyrazolo[3,4-c]pyridin-5-yl}pyridin-3-yl)oxy]propyl]-6H-indole, PubChem CID: 46937052; DB00610: 3-[(1R,2S)-2-amino-1-hydroxypropyl]phenol, PubChem CID: 5906 DB02278: 2-amino-6-hydroxy-7,7-dimethyl-3,4,7,8-tetrahydropteridin-4-one, PubChem CID: 3340355; DB04861: 1-(6-fluoro-3,4-dihydro-2H-1-benzopyran-2-yl)-2-{[2-(6-fluoro-3,4-dihydro-2H-1-benzopyran-2-yl)-2-hydroxyethyl]amino}ethan-1-ol, PubChem CID: 71301; DB02377: 2-aminopurin-6-one, PubChem CID: 764; DB08003: (2S)-2-acetamido-N-[(2S)-1-amino-1-oxo-3-[4-[(5S)-1,1,3-trioxo-1, 2-thiazolidin-5-yl]phenyl]propan-2-yl]-3-phenylpropanamide, PubChem CID: 9547915; DB00529: phosphonoformic acid, PubChem CID: 3415; DB07597: (1R,2S)-2-amino-1,2,3,4-tetrahydronaphthalen-1-ol, PubChem CID: 6420129; DB03536: benzyl N-[(1S)-4-[(diaminomethyl)amino]-1-{[(2S)-3-oxobutan-2-yl]carbamoyl}butyl]carbamate, PubChem CID: 6398520; DB03329: pyridine-2-thiol, PubChem CID: 2723698.
Figure 8.
Induced fit docking allows the discovery of high affinity hits for site I.
(A) Thymol (DrugBank ID DB02513, in wire representation) docks on the outside of the main cavity of the I site (small white spheres) and does not reach the hydrophobic pocket within the cavity (yellow surface), resulting in a poor docking score (−3.2 kcal/mol). (B) After induced fit docking, thymol (in stick representation) enters the site, which now comprises a larger hydrophobic cavity; the docking score is consequently greatly improved to −7.8 kcal/mol. The initial docked pose of thymol before the application of IFD is shown superimposed in wire format. (C) A derivative of thymol, 5-ethyl-2-(4-ethyl-2-hydroxyphenyl)phenol, (PubChem CID 19850961, sticks coloured by element) achieves an impressive score of −10 kcal/mol after induced fit docking, whilst retaining the original thymol pose (in blue) for the substructure that is common to both molecules. (D) Best-ranking pose for DrugBank ID DB07263 ([{2-bromo-4-[(2R)-3-oxo-2,3-diphenylpropyl]phenyl}(difluoro)methyl]phosphonic acid, in stick representation) following induced fit docking. In this protein conformer, the channel connecting sites I and C has been opened creating two hydrophobic subpockets (predicted by SiteMap and depicted here in yellow semi-transparent surface). Two of the aromatic rings of this ligand are placed in these subpockets. This ligand achieves a very good docking score (−9.5 kcal/mol), despite the fact that several hydrogen bonding opportunities (depicted by the blue and red surfaces, corresponding to H-bond donor and acceptor, respectively) are not satisfied in the case of this ligand.
Table 3.
Shifts in melting temperature of A1AT in the presence of selected small molecule ligands (ThermoFluor assay).