The ensemble folding kinetics of protein G from an all-atom Monte Carlo simulation By Jun Shimada and Eugine Shaknovich Bill Hawse Dr. Bahar Elisa Sandvik and Mehrdad Safavian Outline Background on protein folding Protein G Go potential and simulation methods Results Conclusions and discussion What is protein folding? In vitro model Unfolded Folded ( biologically active ) Unfolded High hydrophobic surface area Extended Aggregation Folded/native Hydrophobic core Hydrophilic surface Unfolded Folded How?? Leventhall s paradox Many degrees of freedom, how can polypeptide fold at a biological speed It would take thousands of years of an average peptide to explore all of its conformational space 1
E # native contacts Pathway vs. Funnel Pathway Like an organic chemical reaction Funnel Folding is a landscape-more than one path Chain Entropy?G R P Rxn. Coordinate N Challenges of in vivo folding Aggregation Funnel High concentration of proteins Vectorial production of polypeptide from ribosome Non-native intermediates Many ribosomes in close proximity (molecular crowding) Aggregation!!!!! N A 2
Why use computational methods to study protein folding? Why study protein Protein G? Protein G Actin complexed with Deoxyribonuclease I Protein folding in vivo is complex some aspects of folding can not be completely understood experimentally Need simplified models to make progress Simple topology 1 b sheet and a helix Probably straightforward kinetics Complex topology Actually needs a chaperone To fold (TRIC) Background on the Go potential Topology is the key aspect of protein folding mechanisms Two states for a conformation Native or non-native Make small changes in backbone, side chains etc. Define energies for each microstate Search for low energy states Make a connection between high energy transition state and native state Biased towards native like interactions what if non-native interaction was key to folding mechanism??? Methods: All atom MC Go folding simulation Rules for simulation Heavy atoms depicted as hard spheres Move set- what moves can be made Small backbone and side chain torsional rotations Mc steps 1 Backbone and 10 side chain moves Correct bond connections and geometry 3
Methods Go energy calculation Calculation for energy # non-native contacts-#native contacts Methods Thermodynamic calc. of wt. structure Protein G unfolds cooperatively at T=2.1 Scales against experiment T f =360 K Contact defined by separation distance (R) if s<r<1.86s, then the atoms are in contact (where s is the hard core distance) Distance RMS calculation SqRoot (<(D-D o ) 2 >) D= given structure D 0 = native structure Unit for E represents the Go interaction strength 278K (experimental) = 1.6 (simulation) @ 273 K Hairpin 2 ~ 40% native Helix & hairpin 1 are negligibly stable Agreement with previous experiments Thermodynamic calibration for the Hairpin 2 mutant Examined weakened hairpin 2 Generated by excluding contacts involving residues separated by fewer than 2 positions (fig 1b) Mutation only weakened beta turns # of contacts made in the helix and between b strands 1-2, 3-4, 1-4 didn t change resulting mutant had an unstable helix introduced a generic backbone hydrogen bond interaction every amide N carbonyl O pair within contact distance assigned energy (h) E mut = E g + N h h N h corresponds to the number of h bonds @ 278 K helix and hairpins 1 and 2 =7, 0 an 20% native. The mutated native state was slightly less stable than the wild type Connectivity matrix for Protein G 4
Fast folding trajectory W Fluorescence trajectory E nonnative - E native Minor intermediates- helix - hairpin 2, B1-B4 Major pathway forms through helix - hairpin 1 contact Used W fluorescence as a RxN. Coordinate W burial didn t correspond with completion of folding Hairpin 1 - helix interaction not seen from W fluorescence (weighted avg Wild type vs. mutant W burial Secondary structural details Runs along minor pathway lead to W signal - Explains burst phase 2 phase fit Kinetics similar to experiment - Mutant lags in burial rate Mutant vs. wild type Mutant folds slower Sequences of folding intermediates are different 5
Energy landscapes Agreement of experiment and simulation By weakening hairpin 2 (mutant) folding rate slows down Wt. And mutant have single exponential post burst phase kinetics W burial is not sufficient to capture full folding picture Intermediates are detected by using energy as the rxn. Coordinate By weakening hairpin 2, protein G can switch folding pathways Complex time evolution of intermediates Transition state ensemble 2 state multipathway kinetic model U [I 1 OR I 2 OR I 3 ] F Rate of change of all intermediates are the same magnitude, therefore all intermediates should be considered 6
Summary of folding kinetics Conclusions Simulations indicate that protein g has a more sophisticated folding landscape Some reaction coordinates are unable to capture details of protein folding Bad choice of rxn. Coordinate that use ensemble averages could mask important folding processes References Ma, J., Sigler, P. B, Xu, Z., and Karplus, M. (2000) J. Mol. Biol. 302, 303-313 Ma, J., and Karplus, M. (1999) Biophys. J. 76, A118 Dill, K.A., and Chan, H. S., (1997) Nature structural biology. 4, 10-18. Hartl, F., Hartl, H. (2002) Science. 295, 1852-1858 Frydman, J. (2001) Ann. Rev. of Biochem. 70, 603-647. Wang, J. et al. (2002) Cell. 111, 1027-1039. Dill, K. and Bromberg, S. (2003) Molecular Driving Forces, 614-616, 363-364. 7