Universit degli Studi di Milano Dipartimento di Scienze

  • Slides: 23
Download presentation
Università degli Studi di Milano Dipartimento di Scienze Farmaceutiche “Pietro Pratesi” Protein modeling by

Università degli Studi di Milano Dipartimento di Scienze Farmaceutiche “Pietro Pratesi” Protein modeling by fragmental approach: connecting global homologies with local peculiarities Alessandro Pedretti

Structure-based studies • In order to perform structure-based studies as: – ligand optimization; Molecular

Structure-based studies • In order to perform structure-based studies as: – ligand optimization; Molecular docking – virtual screening; – signal transduction; – substrate recognition. Molecular dynamics the 3 D structure of the biological target is required. • Unluckily, the experimental structure (X-ray diffraction or NMR) is not available for all proteins. Protein modelling

What’s the protein modelling ? • The protein modelling allows to obtain the 3

What’s the protein modelling ? • The protein modelling allows to obtain the 3 D structure of a protein from its aminoacid sequence (primary structure): GFGPHQRLEKLDSLLS… 1 D structure Protein modelling 3 D structure • It can be classified into two main approaches: Comparative modelling Protein modelling Ab-initio modelling

Comparative modelling • It’s based on the assumption: proteins with high homology of sequence

Comparative modelling • It’s based on the assumption: proteins with high homology of sequence should have similar folding. Target sequence 3 D structure database 3 D template Alignment Rough 3 D model Structures obtained by experimental approaches (X-ray and NMR). Homology > 70 % Between target and template To refinement workflow

Ab-initio modelling • It’s based on physical principles and geometric rules obtained by sequence

Ab-initio modelling • It’s based on physical principles and geometric rules obtained by sequence and structure analysis of the 3 D experimental models. Target sequence Folding prediction Application of physical and geometric rules Multiple solutions Global optimization Rough 3 D model By MM and stochastic approaches To refinement workflow

Comparative vs. ab-initio modelling Comparative Ab-initio 3 D template Yes No Success High Low

Comparative vs. ab-initio modelling Comparative Ab-initio 3 D template Yes No Success High Low Computational time Low Very high Structural “clones”* Yes No *Models that are structurally similar due to the common template. • The possibility to obtain structural “clones” is very high, submitting whole query sequences of protein families with high homology to a limited number of 3 D templates (e. g. transmembrane proteins).

Fragmental approach Target sequence Fragmentation in structural domains Done on the basis of information

Fragmental approach Target sequence Fragmentation in structural domains Done on the basis of information included in databases and/or domain finder tools. Folding prediction of each fragment Trough multiple procedures. Assembling using the global 3 D template By geometric superimposition with the 3 D structure of the global template, using molecular modelling tools as VEGA ZZ. Rough model comparative To refinement workflow modelling

Model refinement procedure Rough model Missing residues Side chains add Hydrogens add Energy minimization

Model refinement procedure Rough model Missing residues Side chains add Hydrogens add Energy minimization Structure check Final model VEGA ZZ + NAMD

Human a 4 b 2 nicotinic receptor • The nicotinic acetylcholine receptors (n. Ach.

Human a 4 b 2 nicotinic receptor • The nicotinic acetylcholine receptors (n. Ach. Rs) are composed by five subunits assembled around a central pore permeable to cations. 17 subunit types a 1, b 1, , d, e a 2 -10, b 2 -4 Muscle Nervous system • The therapeutic interest on nicotinic ligands is highlighted by diseases involving the n. Ach. Rs as: Alzheimer’s and Parkinson’s disease, autism, epilepsy, schizophrenia, depression, etc. Human a 4 b 2 subtype • The complete model didn’t exist. • The design of selective a 4 b 2 ligands is problematic due to the low information about the binding mode. Pedretti A. et Al. , Biochemical and Biophysical Research Communications, Vol. 369, 648– 53 (2008).

Monomer modeling Fragmentation Primary structure Swiss. Prot Fugue Folding prediction of each fragment ESCHER

Monomer modeling Fragmentation Primary structure Swiss. Prot Fugue Folding prediction of each fragment ESCHER NG Helices assembly by molecular docking 4 transmembrane domains 2 cytoplasmic loops 1 extracellular loop 2 terminal domains The docking results were filtered the Torpedo Californica n. ACh. R structure. Full assembly VEGA ZZ Side chains Hydrogens VEGA ZZ + NAMD MM refinement Final monomer

Complex assembling Side view 2 x a 4 Multistep docking: a 4 + b

Complex assembling Side view 2 x a 4 Multistep docking: a 4 + b 2 → a 4 b 2 2 a 4 b 2 → (a 4)2(b 2)2 b 2 + (a 4)2(b 2)2 → (a 4)2(b 2)3 + a 4 b 2 ESCHER NG 3 x b 2 Top view

Model validation • The soundness of the resulting model was checked docking a set

Model validation • The soundness of the resulting model was checked docking a set of know nicotinic ligands: Nicotine Epibatidine ABT-418 Citisine A-85380 • All these ligands were simulated in their ionized form. Ligand VEGA ZZ FRED 2 NAMD + Binding site selection Docking Minimization Trp 182, Cys 225, Cys 226 in a 4 b 2 receptor Final complex

Docking results • After the final MM minimization, the docking scores were recalculated by

Docking results • After the final MM minimization, the docking scores were recalculated by Fred 2 (Chem. Gauss 2 scoring function): Trp 82 b 2 Compound Ki (n. M) Score (Kcal/mol) Epibatidine 0. 009 -48. 7 A-85380 0. 05 -45. 1 Citisine 0. 16 -42. 6 Nicotine 1. 0 -38. 9 ABT-418 4. 6 Cys 225 a 4 Asn 134 b 2 Cys 226 a 4 Phe 144 b 2 Trp 182 a 4 -35. 9 a 4 b 2 – nicotine complex

Human glutamate transporter EAAT 1 • L-glutamate is the main excitatory neurotransmitter in the

Human glutamate transporter EAAT 1 • L-glutamate is the main excitatory neurotransmitter in the CNS. Synaptic cleft Axon Dendrite Metabotropic receptor Excitatory effects Glutamate Ionotropic receptor EAAT 1 -5 • It can also overactivate the ionotropic receptors, inducing a series of destructive processes involved in acute and chronic neurological diseases (e. g. amyotrophic lateral sclerosis, Alzheimer’s disease, epilepsy, CNS ischemia, etc). Pedretti A. et Al. , Chem. Med. Chem, Vol. 3, 79 -90 (2008).

EAAT ligand classification • They can be classified in: • Natural substrates. • Substrate

EAAT ligand classification • They can be classified in: • Natural substrates. • Substrate inhibitors. • Non transported uptake blockers. • The last two classes are interesting because in pathological conditions, when the electrochemical gradient is damaged, EAATs can operate in reverse mode, overactivating the post-synaptic receptors. Research aims: • Human EAAT-1 3 D structure by homology modeling. • Pharmacophore models for all ligand classes.

Monomer modeling Fragmentation Primary structure Swiss. Prot Fugue Folding prediction of each fragment Full

Monomer modeling Fragmentation Primary structure Swiss. Prot Fugue Folding prediction of each fragment Full assembly VEGA ZZ The domains were found aligning the sequences of EAAT 1 and glutamate transporter from Pyrococcus horikoshii. The assembly was carried out using the crystal structure of glutamate transporter homologue from Pyrococcus horikoshii. Hydrogens Side chains VEGA ZZ + NAMD MM refinement Final monomer

Complex assembling ESCHER NG VEGA ZZ + NAMD Monomer Homotrimer Complex refinement protocol: •

Complex assembling ESCHER NG VEGA ZZ + NAMD Monomer Homotrimer Complex refinement protocol: • 1 ns of simulation time; • restrained transmembrane segments; • final conjugate gradients minimization. DEEP surface

Docking studies • Two ligand subsets were docked: • natural substrates and competitive substrates

Docking studies • Two ligand subsets were docked: • natural substrates and competitive substrates inhibitors (16); • non-transported blockers (16). • The following procedure was applied to all ligands: Ligand Mopac 7 Flex. X Minimization Docking Complex EAAT 1 monomer • The docking analyses were focused on residues enclosed in a sphere centered on Arg 479 (TM 4). Mutagenesis studies showed this residue plays a pivotal role in the substrate interaction.

Docking results: substrate inhibitors Met 451 Val 449 Arg 479 Gln 204 EAAT 1

Docking results: substrate inhibitors Met 451 Val 449 Arg 479 Gln 204 EAAT 1 – (2 S, 4 R)-methylglutamate complex Thr 450 Gln 445 p. Km = 4. 88 (± 0. 04) – 1. 52 (± 0. 12) Vover N = 15, r 2 = 0. 93, s = 0. 11, F = 174. 11 Where Vover is maximum overlapping volume between the ligand EAAT 1 computed by Flex. X.

Docking results: non-transported blockers Leu 448 Ile 465 Val 449 Thr 450 Trp 473

Docking results: non-transported blockers Leu 448 Ile 465 Val 449 Thr 450 Trp 473 Arg 479 EAAT 1 – L-TBOA complex Gln 445 Gln 204 p. IC 50 = 0. 4446(± 0. 07) – 0. 141(± 0. 02)Score. Flex. X N = 16, r 2 = 0. 77, s = 0. 55, F = 43. 46

Pharmacophore mapping Natural and L-glutamate substrate inhibitors En = excluded volume An = H-bond

Pharmacophore mapping Natural and L-glutamate substrate inhibitors En = excluded volume An = H-bond acceptors Non-transported TFB-TBOA blockers P Y = ionisable group (positively charged) = hydrophobic region • Mapping The two pharmacophore models were by Catalystit’s 4 software. the docking results onto theobtained pharmacophores, possible to the two approaches successfully • highlight Both models highlight the keyare features requiredoverlapped. for the interaction.

Conclusions • We obtained the full model of two transmembrane protein through the fragmental

Conclusions • We obtained the full model of two transmembrane protein through the fragmental approach. • Performing molecular docking studies, we highlighted the main interaction between ligands and the proteins that were confirmed by experimental data, obtained by mutagenesis studies. • Although the number of considered ligands isn’t statistically relevant, we obtained good relationships between the docking scores and the experimental data, confirming the soundness of both models. • All these results show the power and the goodness of the fragmental approach that is able to overcame the problems introduced by global homologies and the possibility to obtain structural clones.

Acknowledgments • Giulio Vistoli • Laura De Luca • Cristina Marconi • Cristina Sciarrillo

Acknowledgments • Giulio Vistoli • Laura De Luca • Cristina Marconi • Cristina Sciarrillo www. vegazz. net www. ddl. unimi. it