Atomic resolution structures of biomolecules are stored at the Protein Data BankContains 60000 structures (mostly determined by X-ray crystallography and NMRAbout 3-5 new structures per day year
Application of Computational tools to solve specific biological problemsSpecific protein-protein interactions is a rule rather than exception and non- specific interactions can lead to several disease Protein-protein interactions in binary complexes: What makes the specificity? Discriminate specific and non-specific interfaces
Specific and non-spacific interactionsHomodimers HeterocomplexCrystal-packing interfaces of monomeric protein Is that pair a biological dimer? The non-covalent interactions that hold crystals together: are the same as in protein-protein complexes and oligomeric proteins BUT they are not subject to natural selection, thus, biologically non- specific.
The dataset and methodology Non-redundant dataset (taken from Protein Data Bank) • 70 Protein-protein complexes (Chakrabarti and Janin, 2002) • 120 Homodimers (Bahadur et al., 2003) • 183 Crystal packing interfaces of 145 monomeric proteins 2fold related interface =105 Non-2fold related interface=83 (Bahadur et al., 2004)Tools for the analysis• Structural features Solvent accessible surface are, Buried atoms, core-rim, atomic density• Physico-chemical properties Polar-non polar interactions, residue composition, hydration
Interface area definition Molecule A Molecule B Complex AB w w Interface area (B) = ASA(A) + ASA(B) – ASA(AB) (Lee & Richards, 1971) (‘Naccess’ by Hubbard SJ. 1992)Interface atoms and residues are all atoms and residues that lose ASA in the complex and contribute to B
Protein-protein interfaces Dimeric k-bungarotoxin (1kba) BSA = 1000 Å2 (Bahadur et al., 2004) Crystal dimer Pokeweed antiviral protein (1qci) BSA = 1000 Å2 (Bahadur et al., 2004)
Size of the protein-protein interfaces 70 No two-fold Interface Interface Ref. Crystal dimers area B 60 Homodimers Complexes (s.d.) 50 Homodimer 3880 Bahadur et Interfaces 40 al., (2003) (±2200) 30 20 Chakrabarti Complex 1910 & Janin, 10 (±760) (2002) 0 800 1000 1600 2400 3200 >3600 2 Interface area B (Å ) Crystal- 1510 Bahadur et. al., (2004) packing3% Homo +Complex 16% Crystal (±520) 30% Crystal 1200 Å2 > 2000 Å2 Standard size interface (Lo Conte et al, 1999) Interfaces formed in protein-protein complex are of ‘Standard size’ Homodimer interfaces are very large compared to crystal-packing interfaces
Non-polar interface area No two-fold 60 fnp*B = Interface area contributed by the C- Crystal dimers Homodimers Complexes containing groups only / Total interface area 40 Interfaces fnp* B (%) 20 Homodimer 65 Complex 58 Crystal-packing 58 0 >80 40 50 60 70 80 Non-polar fraction of the interface area (%) 0 Homo + 6% Crystal 0 Complex < 50 > 70Homodimers have hydrophobic interfaces compared to protein-protein complexes and crystal- packing interfaces.
Clustering of Interface atoms: single or multiple patch?…we cluster interface atoms by the averagelinkage method on a purely geometric basis. 1 3A threshold distance dM must be used in clustering. 4It is set to half the diameter of the interface; 5 2dM= 15 Å in a typical protein-protein complex, 22Å in homodimers. d13+d14+d15+d23+d24+d25 dM = 6(Chakrabarti & Janin, 2002)(Bahadur et. al., 2003)
Standard size single patch and multi-patch interfacesCytochrome c’(2ccy) B (Å2) #res #atoms Homodimers (70) 2740 74 280 Complex (46) 1560 47 170 95% of the crystal-packing interafces are of ‘single patch’Thrombin-ornithodorin(1toc) B (Å2) #res #atoms Homodimers (35) 4760 (0.67, 0.33) 126 486 Complex (18) 2510 (0.63, 0.37) 73 217 Multi-patch interfaces contains at least one large patch with ‘standard size’
Evaluating packing density of the interfaceHomodimer interface Crystal-packing interface 1qci, Antiviral1kba, Kappa- protien complex-bungarotoxin ed with adenine Ai Local density index Global density index count the number ni of interface atoms Calculate the principal moments of inertia Na2, within D = 12 Å of interface atom Ai Nb2, Nc2 of the set of interface atoms; the inertia ellipsoid has half-axes a>b> c average ni over all interface atoms the area at the equator is A=πab LD = Σ (ni) / N GD = N/A B(Å2) LD GD 1kba 998 34 0.95 1qci 994 14 0.31 Specific interfaces are well packed compared to non-specific interfaces
Buried atoms at the interface No two-fold Partially buried 50 Crystal dimers Homodimers w interface atoms Complexes 40 Interfaces 30 20 10 Fully buried 0 interface atoms >50 10 20 30 40 50 Fraction of fully buried interface atoms (%) 87% Crystal- 71% Homodimer + Number of fully buried interface atoms fbu= Total number of interface packing Protein-protein complex fbu<30% fbu>30% atomsSpecific interfaces: 34-36% of the interface atoms are fully buried Non-specific interfaces: this fraction is only 21%
Dissecting the interface: Core and Rim A Core residue: with at least one fully buried atom Rim residue: Contain accessible atoms only d allCI2 inhibitor bound to Bsubtilisin (2sni) Enolase (1ebh) d72% (B) in core 74% (B) in B core B A Core (B%) Homodimers 77 Complexes 72 ‘Core’ region is absent in crystal-packing interfaces
Amino acid composition of the specific interfaces The amino acid composition of the interface core and rim D E K SNumberwisefi = number of core (rim) residues of type i / T FYWM total number of core (rim) residues IL GAreawisef0i = interface area contributed by core (rim) residues of type i / total core (rim) interface area Core: aromatic and hydrophobic residues are abundant at the Rim: polar and charged residues
Composition of interface relative to surface Euclidean distance (Δf) between amino acid compositions of the interface and rest of the protein surface: (∆f)2 = 1/19 ∑ i=1 to 20 (ki – k0i)2 ki = composition of amino acid residues at interface k0i = composition of amino acid residues at surface 1.6 Monomers Interface Surface 2.1 0.5 Homodimers Interface Surface 3.4 Homodimer: interface differs from the protein surfaceCrystal-packing: Difference is negligible
Residue Propensity (RP) scoreRP = Σ ni * Pi , ni number of residues in the interface No two-foldPropensity of a residue to be 50 Crystal dimersat homodimer interface 40 Homodimers ComplexesPi = ln (ki/k0i) Interfaces 30ki = composition of amino acid 20residues at interface 10k0i = composition of amino acidresidues at surface 0 <-6 -4.5 -1.5 1.5 4.5 >6 Residue Propensity Score (RP) Residue propensity Only 5% 5% Crystal score (RP) Homo+Complex Homodimer 4.3 (±4.9) Complex 0.9 (±2.3) Crystal-packing -1.1 (±2.7) < -3 >3 Specific interfaces have +ve RP and non-specific interfaces have –ve RP score
Identifying homodimers and crystal dimers 50 Use a combination of parameters 40 Fraction of fully buried atoms (fbu)Fraction of buried atoms (%) Non-polar interface area (fnp*B) 30 U Residue Propensity score (RP) 20 Classification fnp*B (Å2) fbu (%) RP 10 Monomer ≤ 800 or ≤ 2000 and ≤ 20 M D Homodimer ≥ 2000 or 0 0 1000 2000 3000 4000 ≥ 800 and ≥ 30 Non-polar interface area ( Å2 ) Undecided 800-2000 and 20-30 < 1.5 (M) ≥1.5 (D) Monomer 95% Homodimer 93%
Structural rule of specificity Minimum size: specific protein-protein interfaces have B ≥ 800 Å2 Most crystal packing interfaces are below that size, which may be the minimum for a biologically relevant macromolecular interaction. Standard-size (1200-2000 Å2) interfaces make stable, specific and fully functional interactions . Specificity is expressed in the following features: • close-packed interface atoms • a high fraction of buried atomsA combination of parameters, non-polar interface area, fraction of fullyburied interface atoms and residue propensity score discriminates thespecific interfaces from the non-specific ones with a 95% success rate.