Characterization of the Folding Energy Landscapes of Computer Generated Proteins Suggests High Foldi - PowerPoint PPT Presentation

Loading...

PPT – Characterization of the Folding Energy Landscapes of Computer Generated Proteins Suggests High Foldi PowerPoint presentation | free to view - id: 184fb3-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Characterization of the Folding Energy Landscapes of Computer Generated Proteins Suggests High Foldi

Description:

Characterization of the Folding Energy Landscapes of Computer Generated Proteins ... Riddle, D. S., Santiago, J. V., Bray-Hall, S., T., Doshi, N., Grantcharova, V. P. ... – PowerPoint PPT presentation

Number of Views:31
Avg rating:3.0/5.0
Slides: 48
Provided by: compu352
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Characterization of the Folding Energy Landscapes of Computer Generated Proteins Suggests High Foldi


1
Characterization of the Folding Energy Landscapes
of Computer Generated Proteins Suggests High
Folding Free Energy Barriers and Cooperativity
may be Consequences of Natural Selectionby
Michelle Scalley-Kim and David Baker
  • Shawna ThomasProtein Folding Journal
    ClubNovember 14, 2006

2
Characterization of the Folding Energy Landscapes
of Computer Generated Proteins Suggests High
Folding Free Energy Barriers and Cooperativity
may be Consequences of Natural Selectionby
Michelle Scalley-Kim and David Baker
  • Shawna ThomasProtein Folding Journal
    ClubNovember 14, 2006

3
Motivation
  • Which properties of protein folding are due to
    inherent properties of polypeptide chains and
    which are due to evolutionary pressure?
  • For example, naturally occurring proteins fold
    much faster than random search would suggest
  • Could natural selection be the reason?
  • Possible answers
  • Selection is key rapid folding may be rare
    without explicit selection
  • Structure is key rapid folding may just be the
    consequence of stable native folds
  • Properties include stability, function, folding
    rates, kinetics, ability to aggregate, etc.
  • Challenge difficult/impossible to assess when
    studying naturally occurring proteins because
    these are the results of natural selection

4
Motivation
  • The only way to disentangle the features of the
    folding of small naturally occurring proteins
    is to study the folding of proteins that have not
    been generated by the natural evolutionary
    process.
  • Ways to generate novel proteins
  • Randomize sequences of known proteins
  • Search for sequences folding to a target structure

5
Novel Protein GenerationSequence Randomization
  • Randomize portions of naturally occurring
    proteins
  • Select sequences that fold into stable structures
  • Possibilities for bias
  • Upper limit on what can be changed simultaneously
  • Starting from a naturally occurring topology
  • Previous studies
  • Src SH3 Domain Baker Group
  • Protein L Baker Group

6
Novel Protein GenerationSequence Randomization
  • Question What is the minimum number of amino
    acids needed to encode complex,
    naturally-occurring protein folds?
  • Started with the src SH3 domain
  • Mutated residues not involved in binding to a
    restricted alphabet (I,K,E,A,G)
  • Selected properly folded sequences through
    phage-display selection

7
Novel Protein GenerationSequence Randomization
  • Results
  • 2 sequences FP1 and FP2 passed all the folding
    tests
  • Conclusion rapid folding is not the result of
    extensive evolutionary optimization

From the simplified alphabet
8
Novel Protein GenerationSequence Randomization
  • Question Are rapid folding rates a consequence
    of natural selection?
  • Derived 12 heavily mutated variants of the IgG
    binding domain of protein L
  • Selected properly folded sequences through
    phage-display selection
  • Recovered both fast and slow folding sequences
    (unbiased selection)

9
Novel Protein GenerationSequence Randomization
  • Results
  • All had reduced stabilities
  • Half had folding rates faster than wild type
  • Conclusion the rapid folding of biological
    proteins is not the result of direct natural
    selection instead fast folding appears to be an
    intrinsic property of polypeptide chains that
    adopt stable native states.

10
Novel Protein GenerationSequence Randomization
  • For two small proteins (SH3 and protein L),
    alternative sequences were found to fold as fast
    or faster than the wild type
  • This provides some evidence that rapid folding
    rates are more due to inherent properties of
    polypeptide chains than to evolutionary pressure

11
Novel Protein GenerationDe Novo Design
  • Identify sequences compatible with a target
    structure
  • Difficult to identify sequences that will
    actually fold to the target structure
  • Starting sequence completely random
  • Optimization focused on native state stability
    only
  • Could be biased on the topology of the target
    structure

12
Novel Protein GenerationDe Novo Design
  • First fully automated design
  • Target
  • Second zinc finger module of DNA binding protein
    Zif268
  • bba motif
  • lt 30 residues

13
Novel Protein GenerationDe Novo Design
  • Results
  • 87 residues fall within the most favored region
    of f,j space (remainder in allowed region)

Target (red) FSD-1 (blue)
14
Novel Protein GenerationDe Novo Design
  • De Novo design now more computationally feasible
    for small proteins
  • This provides another tool to help identify which
    folding properties are due to evolutionary
    pressure and which are more fundamental

15
Related WorkRedesign Studies
  • Previous studies
  • Helix-turn-helix homeodomain study Mayo Group
  • 9 redesigned protein study Baker Group
  • Attempt to look at the influence of natural
    selection by studying redesigned (unnatural)
    proteins

16
Related WorkRedesign Studies
  • First redesign study of multiple structures

17
Related WorkRedesign Studies
  • Studied 2 structural analogues of the engrailed
    homeodomain (En-HD)
  • NC3-NCAP
  • 55 sequence identity with template
  • 67 sequence similarity with template
  • ENH-FSM1
  • 25 sequence identity with template
  • 37 sequence identity with NC3-NCAP
  • No statistically significant similarity with any
    know sequence
  • Designed using ORBIT (optimization of rotamers by
    iterative techniques) library

18
Related WorkRedesign Studies
  • Both fold into a stable, well-packed native state
  • Two-state folding
  • No structures, but had dispersed NMR and CD
    spectra typical of folded proteins

NC3-NCAP
ENH-FSM1
19
Related WorkRedesign Studies
  • Result redesigned proteins had similar folding
    rates to the naturally occurring template
  • No statistically significant discrepancy at a 95
    confidence interval
  • Even with very little sequence identity
  • Conclusion these properties folding rates
    are readily achieved even in the absence of
    direct selective pressure.
  • Naturally occurring proteins have highly
    designable structures
  • Smooth energy landscapes
  • Unique ground states

20
Related WorkRedesign Studies
  • Redesign to several different target structures
    to study to test performance of RosettaDesign

21
Related WorkRosettaDesign
  • Two main components
  • Energy function to rank sequences
  • Dominated by Lennard-Jones interactions, hydrogen
    bonding, and an implicit solvation model
  • Search function to scan sequence space
  • Monte Carlo simulation
  • Single amino acid substitutions accepted/rejected
    using the standard Metropolis criterion
  • Side chain conformations restricted to rotamer
    conformations in Dunbracks library (Protein Sci,
    1997)
  • Extremely fast 100 residue search takes 5-10
    minutes on a desktop PC

22
Related WorkRedesign Studies
  • Proteins studied

Studied 9 different proteins of different
secondary structure topologies -- average 35
sequence identity to wild type over all residues
-- average 50 sequence identity to wild type
over core residues
23
Related WorkRedesign Studies
  • Folding and stability tests
  • CD spectroscopy to measure secondary structure
    makeup
  • Size-exclusion chromatography to determine if
    proteins monomeric
  • Chemical and thermal denaturation experiments to
    confirm proteins folded and measure stabilities
  • One-dimensional 1H NMR experiments to confirm
    proteins folded and measure structure rigidity

24
Related WorkRedesign Studies
25
Related WorkRedesign Studies
  • Results
  • Stability
  • 3 were more significantly more stable (sequences
    were more hydrophobic)
  • 2 were less stable (sequences were less
    hydrophobic)
  • 2 displayed aggregation
  • Sequences had larger fraction of non-polar
    accessible surface area
  • Lacked aggregation preventors strand kinks,
    inward pointing charged residues
  • 1 completely unfolded
  • Scaled atomic radii (.95) to compensate for fixed
    rotamers caused large clash between 2 amino acids

26
Todays Paper
  • Extends the previous all a study of the Mayo
    Group
  • Includes mixed and an all b protein
  • Includes de novo design of a computer-generated
    topology

27
Experimental SetupProteins Studied
  • 5 proteins that have never been exposed to
    natural selection

From previous RosettaDesign study well folded
and stable
Another attempt to redesign src SH3 previously
not folded
De novo design of computer-generated structure
28
Results
29
Experimental SetupSH3
  • Redesigned sequence has 52 overall identity and
    86 core identity to wild type
  • Shown to be monomeric in gel filtration
    experiments

30
Experimental SetupSH3
  • CD Spectra for SH3
  • Minimum at 208 nm, not typical for all b proteins
  • May be due to local interactions between two
    tryptophan residues

(squares) 50 mM sodium phosphate (pH 7) at
25oC (circles) 50 mM sodium phosphate (pH 7) at
80oC (triangles) 7.5M guanidine, 50 mM sodium
phosphate (pH 7) at 25oC
31
Experimental SetupSH3
  • Guanidine-induced denaturation
  • Best fit to two-state model
  • Significantly destabilized
  • DGUF 4.1 kcal mol-1 -gt 1.3 kcal mol-1

WT
Redesigned SH3
32
Experimental SetupSH3
  • NMR spectrum suggests well-folded, rigid
    structure
  • Shifts consistent with all b proteins

33
Experimental SetupTop7 de Novo Design
  • 93 residue protein
  • Novel sequence and novel topology

34
Experimental SetupTop7 de Novo Design
  • Differs from other computational methods since it
    iterates between sequence optimization over a
    fixed backbone and a gradient-based structure
    optimization for a fixed sequence
  • At each iteration, only the least energy sequence
    or structure is kept for the next iteration

35
Experimental SetupTop7 de Novo Design
  • Initial structure selected because it was not
    present in the PDB according to TOPS
  • 3D structure satisfying the constraints generated
    by splicing fragments of 3 and 9 residues from
    PDB with appropriate secondary structure

A set of constraints were identified from the 2D
sketch to define the topology (purple arrows)
36
Experimental SetupTop7 de Novo Design
  • Result Top7 with only 1.17A backbone RMSD
    between the design model and the crystal
    structure
  • At the time, CASP experiments for 90-100 residue
    proteins had greater than 4A RMSD
  • Why did they do so well?
  • The ability to alter both sequence and structure
    may facilitate the search by smoothing out the
    landscape
  • Top7 lacks functional constraints which can lead
    to local minima

37
Experimental SetupRequired Mutations
  • Some mutations were required for stopped-flow
    fluorescence experiments
  • Top7 F81W (partially buried)
  • DGUF 13.2 kcal mol-1 -gt 13.6 kcal mol-1
  • Protein L2 Y34W (partially buried)
  • DGUF 4.6 kcal mol-1 -gt 5.1 kcal mol-1
  • Acylphosphatase W64L
  • DGUF 5.6 kcal mol-1 -gt 5.4 kcal mol-1

38
ResultsEquilibrium Denaturation
CD diamonds/triangles/squares Fluorescence
circles Buffers chosen to match previous WT
results All data were fit to a two-state model
(CD and fluorescence superimposed) No data for
SH3 little change in CD upon unfolding
F81W
Y34W
Redesigned
WT
WT
W64L
Redesigned
WT
WT
39
ResultsStopped-Flow Fluorescence
Redesigned
Redesigned
WT
WT
Buffers chosen to match previous WT
results Denaturant dependence on both folding and
unfolding rates could not always be
determined. Extrapolated data agrees with
equilibrium data (next slide).
Redesigned
Redesigned
WT
WT
40
ResultsWT and Redesign Comparison
  • Extrapolated data (DGkin) matches well with
    equilibrium data (DGeq)
  • Folding/Unfolding rates
  • kfH2O and kuH2O faster than WT for protein L and
    acylphosphatase (similar stabilities to WT)
  • kfH2O faster and kuH2O slower than WT for
    pro-carboxypeptidase (increased stability)
  • kfH2O slower and kuH2O faster than WT for SH3
    (decreased stability)

41
ResultsWT and Redesign Comparison
  • Folding rates at transition midpoint
  • Faster for acylphosphatase, protein L, and SH3
  • Suggests preferential stabilization of redesigned
    transition states
  • Similar rate for pro-carboxypeptidase
  • Transition state may be shifted towards unfolded
    for acylphosphatase, pro-carboxypeptidase, and
    protein L (decrease in mf/(mfmu)

42
ResultsTop7 Folding Kinetics
  • Significantly different than the others
  • Unable to measure unfolding rates

43
ResultsTop7 Folding Kinetics
  • Possible explanations for decrease in denaturant
    dependence at low guanidine concentrations
  • Partially folded intermediates/kinetic traps with
    hydrophobic surface area burial comparable to the
    transition state may be populated
  • Little burial change between partially folded and
    transition state
  • Contrasts with typically high degree of
    cooperativity in natural protein structures
  • Position of transition state moved significantly
    toward unfolded
  • Significant increase in internal friction

44
Conclusions
  • Naturally occurring proteins not optimized for
    fast folding
  • 3/4 redesigned proteins fold faster than WT
  • 3/4 redesigned proteins fold faster at their
    transition mid-point than WT
  • Data suggests transition state is preferentially
    stabilized in redesigned proteins
  • Design method favors increase in hydrophobic core
    volume
  • May broaden transition state

45
Speculations
  • Slower folding/unfolding may be an indirect
    consequence of selection against aggregation and
    for the population of a single functional state
  • May have increased the amount of buried polar
    interactions and reduced the number of peripheral
    hydrophobic residues
  • Resulting in increased folding/unfolding energy
    barriers
  • Computer design disfavors buried polar
    interactions and favors peripheral hydrophobic
    residues

46
Speculations
  • Natural selection favors cooperativity which
    could reduce aggregation of partially folded
    states
  • Top7 had significantly more complex kinetics at
    0-4M
  • Complex kinetics rarely observed for small
    proteins
  • Naturally occurring proteins (such as barnase,
    ribonuclease A, hen lysozyme) that have complex
    kinetics are over a narrower range (0-1M)
  • Natural selection destabilizes partially folded
    forms and avoids rough energy landscapes with
    kinetic traps
  • This typically only seen in computer-generated
    proteins

47
A Final Thought
  • Among the naturally occurring structures
    redesigned
  • 3/4 more stable, 1/4 less stable
  • 2/4 have lower free energy barriers, 1/4 has a
    similar barrier, and 1/4 has a higher barrier
  • Degree of denatured state solvation changed in 2/4

48
References
  • Dahiyat, B. I. Mayo, S. L. (1997). De novo
    protein design fully automated sequence
    selection. Science, 278, 82-87. Zinc finger
    redesign
  • Dantas, G., Kuhlman, B., Callender, D., Wong, M.
    Baker, D. (2003). A large scale test of
    computational protein design folding and
    stability of nine completely redesigned globular
    proteins. J. Mol. Biol. 332, 449-460. Redesign
    of 9 proteins
  • Gillespie, B., Vu, D. M., Shah, P. S., Marshall,
    S. A., Dyer, R. B., Mayo, S. L. Plaxco, K. W.
    (2003) NMR and temperature-jump measurements of
    de novo designed proteins demonstrate rapid
    folding in the absence of explicitly selection
    for kinetics. J. Mol. Biol. 330, 813-819.
    Folding rate comparison of redesigned proteins
  • Kim, D. E., Gu, H. Baker, D. (1998). The
    sequences of small proteins are not extensively
    optimized for rapid folding by natural selection.
    Proc. Natl Acad. Sci. USA, 95, 4982-4986.
    Protein L redesign
  • Kuhlman, B., Dantas, G., Ireton, G., Varani, G.,
    Stoddard, B. Baker, D. (2004). De novo design
    of a novel globular protein fold with atomic
    level accuracy. Science, 302, 1364-1368. Top7
    design
  • Riddle, D. S., Santiago, J. V., Bray-Hall, S.,
    T., Doshi, N., Grantcharova, V. P., Yi, Q.
    Baker, D. (1997). Functional rapidly folding
    proteins from simplified amino acid sequences.
    Nature Struct. Biol. 4, 805-809. SH3 redesign
About PowerShow.com