390 likes | 573 Views
Understanding Polyglutamine Structure. Alfred Chung Michael McPhail Karis Stevenson Drs. Finke &Zohdy Oakland University June 26, 2009 NSF/NIH Grant #: 0609152. Background Foundations of Protein Structure –Primary Structure. 4 main types of amino acids: Hydrophobic Polar
E N D
Understanding Polyglutamine Structure Alfred Chung Michael McPhail Karis Stevenson Drs. Finke &Zohdy Oakland University June 26, 2009 NSF/NIH Grant #: 0609152
Background Foundations of Protein Structure –Primary Structure 4 main types of amino acids: Hydrophobic Polar Positively Charged Negatively Charged Peptides: amino acid linkages N-terminal to C-terminal Dihedral Angles Glutamine(Q) http://www.molecularsciences.org/structural_bioinformatics/protein_structures
Background Foundations of Protein Structure –Secondary Structure 3 main categories of secondary structure Alpha-helices Beta-sheets Random Coil www.bio.mtu.edu/campbell/401lec8all.pdf
Background Foundations of Protein Structure –Higher-order Structure Interactions that stabilize structure: Electrostatic Interactions Hydrophobic Effect H-bonds Disulfide Bonds Environment also effects structure: pH Salts Composition
Background Potential for Misfolding http://www.nature.com/nature/journal/v426/n6968/full/nature02261.html
Problem Defining Polyglutamine Structure Monomeric structure not well-established Crystal structure of aggregates difficult to obtain Structural and folding information provide framework for therapeutics http://www.nature.com/nature/journal/v426/n6968/full/nature02261.html
MotivationDiseases Associated with PolyQ aggregation http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6T0V-4TCTGBK1&_user=1317309&_rdoc=1&_fmt=&_orig=s earch&_sort=d&view=c &_acct= C00005 2319&_version=1&_urlVersion=0&_userid=1317309&md5=b850ead20c472f6cc78687fbb9cb9ab2
MotivationHuntington’s Disease Attributes Autosomal dominant mutation of chromosome 4 Late onset: 35-44 years old Symptoms progress faster down generations Neuronal loss in caudate nucleus Movement disorders Cognitive decline Behavioral disturbances
SolutionIntegration of 3 Complementary Techniques Polyglutamine Structure FRET EXPERIMENTS (In-vitro experiments) MOLECULAR DYNAMICS (In-silico experiments) Q-Learning (Learning Algorithm)
Fluorescence Resonance Energy Transfer (FRET) FRET is characterized by the transfer of energy from an excited donor chromophore to an acceptor chromophore, without associated radiation release.
FRETEnergy Transfer To measure distances or changes in distance, you need to specifically and uniquely label your molecule of interest with a donor and an acceptor probe. Excited donor fluorophore transfers its energy to an acceptor chromophore via dipole-dipole interactions
FRETMeasurement Range of approx. 10 nm. FRET measurements can be utilized as an effective molecular ruler for determining distances between biomolecules.
FRETThe Equations Ro is the Förster distance – the distance between the donor and acceptor probe at which the energy transfer is (on average) 50% efficient The overlap integral J represents the degree of overlap between the donor fluorescence spectrum and the acceptor absorption spectrum
FRETEfficiency FRET efficiency can be measured using the lifetime of the donor in presence (Tda) and absence of the acceptor probe (Td) Td=6.486133ns Tda=5.130811ns
FRETMolecular Measurements Once FRET efficiency and Förster distance are calculated, the distance between the donor and acceptor ends can be calculated.
FRETFRET and Molecular Dynamics • FRET can then tell you how far apart two parts of a protein are. This can give you a rough idea of the dimension and shape of the protein. • A check on the validity of molecular dynamics simulations
Molecular DynamicsAMBER Molecular Dynamics Suite of programs for analysis of molecular dynamic simulations Analysis tool for protein folding, ligand-binding, and denaturation Validation of experimental findings
Molecular DynamicsUsing AMBER 3 main procedures: System Preparation LEaP Simulation Sander Trajectory Analysis Ptraj
Molecular DynamicsForce Fields of AMBER Delineates the functional form for a system of atoms Incorporates parameters relevant to: Bond lengths Bond angles Dihedral angles Requires careful selection to prevent bias
Molecular DynamicsPreliminary Simulation for Polyglutamine Sequence: FK2Q16K2Y Force Fields: 96 and 99SB Model: Generalized Born Conditions: 300K for 50 ns
Molecular DynamicsDifferences Between Force Fields Parm96 Parm99SB
Molecular DynamicsResults • Parm96 • Distance: 33.8 ±3.4Å Distance (Å) Steps Parm99SB Distance: 47.6 ± 0.4Å
Molecular DynamicsImproved Simulation for Polyglutamine Sequence: (ABZ)-K2Q16K2-(YNO) Force Fields: 96 and 99SB Model: Generalized Born Conditions: 300K for 50 ns
Reinforcement Learning • Agent learns autonomously • What is learned? • Focus on experience(explore/exploit) Neuroscience & Psychology RL Artificial Intelligence
Q-LearningReinforcement Learning An agent takes actions in an environment Agent wants to maximize reward
Example-Tower of Hanoi http://http://brynnevans.com/blog/wp-content/uploads/2009/03/tower_of_hanoi.jpg
http://people.revoledu.com/kardi/tutorial/ReinforcementLearning/index.htmlhttp://people.revoledu.com/kardi/tutorial/ReinforcementLearning/index.html
http://people.revoledu.com/kardi/tutorial/ReinforcementLearning/index.htmlhttp://people.revoledu.com/kardi/tutorial/ReinforcementLearning/index.html
Q-LearningAlgorithm state repeat{ pick action from Q observe reward act in world s---->a--->s' update: Q(s,a)= (1-α)Q(s,a) + α[R + γ*maxQ(s’,a’)] } s=s'
Q-LearningExtended-Algorithm state repeat{ pick action from Q observe reward act in world s---->a--->s' update: Q(s,a)= (1-α)Q(s,a) + α[R + γ*maxQ(s’,a’)] } s=s' • Q-initialization • Small random values • Boltzmann distribution • Reward Structre • Gaussian distribution • α and γ values • Steepest descent
MD and Q-Learning End Distances Distance:33.8 +/- 3.4 angstroms Parm 96 Distance=33.9 +/- 10.4 angstroms
Q-Learning3-D Model • 3-D Model • Ramachandran plots to select backbone angles • Minimizing energy • Flexibility of parameters http://giantshoulders.files.wordpress.com/2007/10/ramaplot.png?w=250
References • C. J. C. H.Watkins and P. Dayan, “Q-learning,” Machine Learning, vol.8, pp. 279–292, 1992. • Warby, Graham, Hayden. Huntington Disease. 2007. • Jieya Shao , and Marc I. Diamond. “Polyglutamine diseases: emerging concepts in pathogenesis and therapy”. Hum. Mol. Genet. 16: R115-R123. • D. Shortle, “Propensities, probabilities, and the Boltzmann hypothesis,” Protein Science, vol.12 pp. 1298–1302. • J. Finke, P Jennings, J Lee, J Onuchic, J Winkler, “Equilibrium Unfolding of the Poly(glutamic acid) Helix” Biopolymers, vol. 86, pp. 193-211. • D.A. Case, T.E. Cheatham, III, T. Darden, H. Gohlke, R. Luo, K.M. Merz, Jr., A. Onufriev, C. Simmerling, B. Wang and R. Woods. The Amber biomolecular simulation programs. J. Computat. Chem. 26, 1668-1688 (2005). • I.O.Bucak,M.A.Zohdy,Reinforcementlearningcontrolofnonlinearmulti-linksystem,Eng.Appl.Artif.Intell.14(5)(2001)563–575.