Protein Structure and Its Applications - Lecture Slides | BSC 5936, Study notes of Biology

Download Protein Structure and Its Applications - Lecture Slides | BSC 5936 and more Study notes Biology in PDF only on Docsity! Protein Structure & Prediction 3/26/2003 BSC 5936 (c) M.S.Chapman 1 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 1 Protein Structure & Its Prediction BSC 5936: Introduction to BioInformatics © Michael S. Chapman, FSU, 1994-03 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 2 Agenda Properties of particular sequences Tags, sequences etc.. What about a particular sequence particular structure / function Group properties of amino acids Hydrophobic moment Helix propensity… secondary structure prediction Prediction of 3-D structure Why? How? Quality? 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 3 Levels of Protein Structure Organization Primary Structure Linear sequence of amino acids Secondary structure Several amino acids w/ repeating conformation Alpha helices, beta sheets Tertiary structure Folding of chain / 2-ary structures into compact “Globular” or fiber Quaternary structure Assembly of several proteins functional unit Different proteins or copies of the same 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 4 Secondary Structures Alpha helices; Beta sheets Saturated with H-bonds neutralize dipoles in buried regions. Provide rigidity, stability: let mutations change chemical properties only not conformation. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 5 α-helices: Essential features: Cn==O……H--Nn+4. Cβ point back like arrows C==O point “forwards”. Average length ≈ 10 residues ≈ 15 Å. Range = 4 – 40 residues 3.6 aa/turn 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 6 Where are α-helices? Hydrophobic helices cross lipid membranes. Amphiphylic are common: Surfaces of globular proteins. Membrane surface proteins, e.g. mellittin i.e. at interfaces Hydrophobic Hydrophilic Globular protein. Protein Structure & Prediction 3/26/2003 BSC 5936 (c) M.S.Chapman 2 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 7 β-sheets - parallel Pleated - look edge-on Strands ~ 5 Å apart Note direction of H-bonds Will differ in anti-parallel & mixed sheets 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 8 β-sheets - antiparallel (parallel – for comparison) Similarly Pleated -- look edge-on Strands ~ 5 Å apart Differs in Direction of H-bonds Mixed sheets also common 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 9 Loops & Turns Connect secondary structural elements. Loops often carry the functional groups. Hairpin turns: Shortest possible loops (2 residues). Gly often in tight turns. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 10 Levels of Protein Structure Organization Primary Structure Linear sequence of amino acids Secondary structure Several amino acids w/ repeating conformation Alpha helices, beta sheets Tertiary structure Folding of chain / 2-ary structures into compact “Globular” or fiber Quaternary structure Assembly of several proteins functional unit Different proteins or copies of the same 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 11 Large Units of Protein Structure Subunit Single peptide chain w/in assembly Domain Globular / fibrous unit of subunit / protein “Beads on a string” w/ at most 1 bonded connection to previous & next domains (This requirement sometimes relaxed) Super-secondary structures Arrangement of 2-ary structures found in many domains, eg.: Helix bundle β-barrel α/β barrel β-sandwich Domains w/ same super-secondary structure may have homologous function, e.g. “Rossmann fold” – nucleotide-binding Helix bundle – transmembrane or structural (e.g. myosin) … so they need not have homologous function Relation to Structural Genomics 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 12 Motif: Small Unit of Protein Structure One to a few secondary structures Used repeatedly in different proteins Would not be stable in isolation Common structure / function Examples: “EF-hand” – helix-loop-helix – Ca++-binding Zinc-finger - αββ - DNA- binding proteins β−α−β – unit of α/β barrel Protein Structure & Prediction 3/26/2003 BSC 5936 (c) M.S.Chapman 5 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 25 Secondary Structure Prediction Relatively tractable 1970’s: ~50% correct; 2000’s: ~75% correct Stepping stone to full 3-D prediction? Has not fulfilled expectations May allow domain fold to be recognized Through order of α, β elements Distant homologies not recognized fr. sequence (Occasionally) Membrane protein identification (helix bundles, etc.) Prediction of where mutations can be introduced 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 26 2-ary Structure prediction - Basic principles “Propensity” of each amino acid to form: Helix; β-strand; turn Normalized frequencies of amino acid types in known protein structures E.g. Helix propensity: Ala > Leu > Ile > Val > (Ser, Thr) > (Asp, Asn) >… >> Pro Chou & Fasman (1974) algorithm Search for 4 of 6 consecutive residues w/ high helix propensity Extend, terminating if 4 residues have low propensity Then search for 3 of 5 consecutive residues w/ high beta propensity… heuristic!, based on 1st 15 structures Redone on 29 structures, other improvements Claimed 70-80% correct, others state ~50%. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 27 2-ary Structure Prediction - Improvements GOR (Garnier-Osguthorpe-Robson, 1978-96), improved by Bias according to estimate of α/β content Consider propensity of residue pairs. 63% correct Predator (Frishman & Argos, 1996), adds i i+4 interactions… 68% correct PHD (Rost & Sander, 1993) added Multiple sequence profile (improved correctness 8%) Neural network 65% correct, 72% if use conservation DSC (King & Sternberg, 1996), adds Hydrophobic moment 70% correct best use sequence profiles, achieve mid 70’s %-correct. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 28 Entry of Neural Networks Heuristics Ad hoc rules, e.g. min # helix- forming residues? thresholds on propensities…? Neural Networks Large # “rules” to process data Weights to combine scores Training using data & known solutions used to optimize Weights Sometimes simple clustering rules D at a: A G F D W K L V G Y E H R I Processing rules Weighting α? β? Prediction 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 29 Why so many efforts in 2-ary prediction? Real interest in 3-D structure function Proteins fold from nuclei of 2-ary structures Sequences folded in silico analogously (?) 2-ary prediction as step towards 3-D structure Foiled on two accounts: Too much dependence of 2-ary structure on 3-D structure (perhaps why only 70% correct) Predicted folding of 2-ary structures still near impossible. But 2-ary prediction recognition of domain type Helps identify homology when sequence similarity is low 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 30 3-D Modeling by Homology The link to secondary structure prediction Protein Structure & Prediction 3/26/2003 BSC 5936 (c) M.S.Chapman 6 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 31 Heuristic Homology Methods Homology assumes similarity to some other structure Which structure? Need to search… Alignment to sequences of known structure Profiles, state of art in 1980’s Threading How compatible is my sequence with each known structure? Does it put… Hydrophobic residues in the interior? Gly/Pro at turns?, etc.. Similar sequence Introduced by Eisenberg, mid- 80’s Many methods being developed Still very popular Issues of scoring, gap penalties, learning sets, etc.. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 32 Threading – Scoring methods Energy based Overlay sequence on known structure Calculate the hydrophobic energy Lower the better Heuristic Arbitrary rules that give the best results Rewards & Penalties learned when applying methods to known systems Gaps –how often do we allow insertions/deletions? Combinations of energy & heuristics 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 33 Threading – the challenge Easy if sequence identity > 50%. Twilight zone w/ 20 < identity < 30%. Near impossible at < 20%. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 34 Fragment Dictionaries Same local backbone conformations found in many protein structures Search for sequence-compatible structures (say) 5 residues at a time. Methods under development, issues… What combination of fragments give a sensible protein structure? Exhaustive combinatorial search intractable. Genetic algorithms, neural networks… How to score? Energy vs. heuristic etc.. Many methods, one of best is David Baker’s. 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 35 David Baker’s Rosetta method Finds homologous fragments Motif-sized > 5 residues Mix & match Energy minimize (Zhou’s lectures) Best method unless high homology to another protein (Ellington lectures) 3/26/2003 Structure Prediction, BSC5936 (c) Chapman 36 Reading: Survey of Domain Structures: Branden & Tooze, Chapters 3, 4 & 5. Recommended Virtual Protein Structure Course http://www.cryst.bbk.ac.uk/PPS2/course/index.html Or similar – it seems to keep moving!