Cooperativity, Local-Nonlocal Coupling, and Nonnative Interactions in Protein Folding

Transcription

1 NORDITA, Stockholm March 16, 2012 Cooperativity, Local-Nonlocal Coupling, and Nonnative Interactions in Protein Folding Hue Sun Chan Departments of Biochemistry, of Molecular Genetics, and of Physics University of Toronto, Ontario M5S 1A8 Canada

2 Folding cooperativity means two-state-like folding/unfolding noncooperative more cooperative Q = fractional number of native contacts Experimental criteria from: calorimetry: ΔH vh /ΔH cal 1 chevron plots more direct probes of two-state behaviors

3 Kinetic manifestation of folding cooperativity: linear chevron plots Data from Jackson et al., Biochemistry 32:11270 (1993) Linear folding and unfolding arms imply a linear relationship between log(folding/unfolding rate) and equilibrium stability chevron rollover is indicative of less cooperative or noncooperative folding

4 PMF Folding Cooperativity and the Levinthal Paradox calorimetry (DSC) golf-course landscape near-levinthal landscape P(PMF) Kaya & Chan, Proteins (2003) Chan, Zhang, Wallin & Liu, Annu Rev Phys Chem (2011)

5 Generic protein properties are stringent constraints on modeling, providing important clues to the energetics underlying real protein behavior

6 Pairwise additive hydrophobic interactions are insufficient for two-state cooperativity mean-field model of Dill No significant barrier between folded and unfolded states in explicit-chain hydrophobic-polar (HP) models with pairwise additive energies Many-body energies can lead to more folding cooperativity Chan, Proteins (2000); Chan, Zhang, Wallin & Liu, Annu Rev Phys Chem (2011) cf. Dill, Biochemistry (1985); Dill et al., Biochemistry (1989)

7 P(E) Folding cooperativity is not a corollary of a sequence s ability to fold to an essentially unique structure case in point: A simplified atomic model of 3-helix protein 54aa unimodal E The general pairwise additive sidechain hydrophobic and directional H-bond interactions in this model are not sufficient for folding cooperativity severe rollover Knott & Chan, Chem Phys (2004); Proteins (2006) Model adapted from: Irbäck, Sjunnesson & Wallin, PNAS 97:13614 (2000)

8 Topological Modeling: C a Gō Models CI2 Shea et al., PNAS (1999); Micheletti et al., Phys Rev Lett (1999); Clementi et al., JMB (2000); Koga & Takada, JMB (2001); Kaya & Chan, JMB (2003) Native-centric local and non-bonded interactions Langevin dynamics Thermodynamically quite cooperative

9 Gō-model chevron plots have significant chevron rollovers, implying that in many cases common Gō models are less cooperative than the real proteins they aim to mimick Theory Experiment Kaya & Chan, JMB (2003) Jackson & Fersht, Biochemistry (1991)

10 Even with native biases, common pairwise additive interactions do not appear to be sufficient to account for cooperative folding

11 Desolvation barriers enhance folding cooperativity Liu & Chan, JMB (2005), Phys Biol (2005); Fergusion et al., JMB (2009)

12 Desolvation barriers enhance folding/unfolding cooperativity increasing height of pairwise desolvation barrier leads to Cheung et al., PNAS (2002) Higher overall free energy barrier CI2 CI2 more linear chevron plots less native fluctuation Kaya et al., Biophys J (2005); Liu & Chan, JMB (2005); Phys Biol (2005)

13 Desolvation barriers affect secondary structure propensities 17mer polyalanine cumulative α β i,i+3 i,i+4 β (n > 11) C β -C β distance α (n > 15) Dias, Karttunen & Chan, Phys Rev E (2011) Using the simple atomic model of: Irbäck & Mohanty, Biophys J (2005) Meinke, Mohanty, Eisenmenger & Hansmann, Comp Phys Comm (2008) (ε cm = ε db = 0 for Ala)

14 Desolvation barriers and contact interactions between methyl groups favor β-sheets over α-helices in polyalanine ε cm = ε db = 0 ε cm = 0; ε db > 0 ε cm > 0; ε db = 0 ε cm > 0; ε db > 0 Dias, Karttunen & Chan, Phys Rev E (2011)

15 Native Topology (Contact Pattern) Dependent Folding Rates Plaxco, Simons & Baker, JMB (1998) Plot from Plaxco et al. Biochemistry (2000) Can protein chain models capture this trend?

16 Desolvation barrier effects are a likely contributor to the remarkable diversity in the folding rates of small proteins no-db Koga & Takada, JMB (2001); Chavez, Onuchic & Clementi, JACS (2004) Wallin & Chan, J Phys Condens Matt (2006) Ferguson, Liu & Chan, J Mol Biol (2009) Figure from: Chan, Zhang, Wallin & Liu, Annu Rev Phys Chem (2011)

17 Folding Barriers entropic & enthalpic components

18 Enthalpic barriers: Non-Arrhenius folding rates, positive unfolded-to-transition state enthalpy changes at some temperatures Does this mean that the folding landscape is not funnel-like? CI2 Data from: Segawa & Sugihara, Biopolymers (1984); Jackson & Fersht, Biochemistry (1991); Oliveberg, Tan & Fersht, PNAS (1995) Figure from: Chan, Zhang, Wallin & Liu, Annu Rev Phys Chem (2011)

19 Desolvation is a likely origin of robust enthalpic barriers to protein folding Liu & Chan, JMB (2005)

20 Enthalpy-Entropy Compensation at Desolvation temperature dependence of the potential of mean force (PMF) Enthalpic barrier can be significantly higher than the desolvation free energy barrier Moghaddam, Shimizu & Chan, JACS (2005); Liu & Chan, JMB (2005)

21 Typically, CI2 Unfolded-to- Transition-State heat capacity change is negative for protein folding, but for the association of small nonpolar solutes in water, heat capacity change is positive around the desolvation free energy barrier. Length-scale dependence provides a possible probe for cooperativity? DC P highly non-monotonic Not well predicted by surface areas x TIP4P water model from DG at 8 temperatures Shimizu & Chan, JACS (2001)

22 Hydrophobic interactions among small hydrophobic solutes: Robust D C P > 0 at the desolvation free energy barrier Xe 2-body r CH 4 3-body x Moghaddam, Shimizu & Chan, JACS (2005) r / nm Paschek, J Chem Phys 120:6674 (2004)

23 a-helix association in water as a model for rate-limiting events in protein folding A pair of 20-residue poly-alanine or poly-leucine helices ~3,800 water molecules Simulated constantpressure free energy of association (potential of mean force, PMF) at five temperatures MacCallum, Moghaddam, Chan & Tieleman, PNAS (2007)

24 Enthalpic desolvation barriers of ~ 50 kj/mol comparable to that of protein folding Dramatic enthalpy-entropy compensation at the desolvation step leading to low or non-existent free energy barriers At 25 deg C, Enthalpic folding barrier height for CI2 is ~ 30kJ/mol (Oliveberg et al., 1995) CspB is ~ 32kJ/mol (Schindler & Schmid, 1996) MacCallum, Moghaddam, Chan & Tieleman, PNAS (2007)

25 Enthalpic Folding Barrier Non-Funnel Landscape? Figure from: Chan, Zhang, Wallin & Liu, Annu Rev Phys Chem (2011)

26 Enthalpic barriers caused by steric dewetting large parts of the protein coming together at the rate-limiting step activation volume Experimental correlation between activation volume and activation enthalpy? MacCallum et al., Proc Natl Acad Sci USA (2007) See also discussion in: Ferguson et al., J Mol Biol (2009)

27 U?? TS V ~ 55 ml/mol cf. Experimental* V (ml/mol) for Snase: 56 for WT; 65 for L125K; 15 for V66K, 0 for T62K A significant activation volume of folding suggests that large not small parts of the protein are coming together at the ratelimiting step of folding. two methanes Stepwise (noncooperative)? Volume / Å 3 ~ 4.5 Å ml/mol *Experimental data from: Mitra et al., JACS 129:14108 (2007) Distance / Å [Dias & Chan, in preparation (2012)]

28 Local-nonlocal coupling A hypothesized cooperative interplay between favorable nonlocal interactions and local conformational preferences Notch ankyrin domain Kaya & Chan, Proteins (2003) Freire & Murphy, J Mol Biol (1991) Mello & Barrick, PNAS (2004) Merging and grouping of native-state hydrogen exchange isotherms, foldons cyt c Kaya & Chan, Proteins (2005) Bai et al., Science (1995)

29 Theory and experiment uncover specific nonnative interactions in cooperative protein folding Simple sequence physics, e.g. pairwise hydrophobic attraction, is not entirely overwhelmed by perfect native-centricity; simple physics may yet manifest as perturbations on a strong background of native bias Zarrine-Afsar, Wallin, Neculai, Neudecker, Howell, Davidson & Chan, PNAS (2008)

30 Fyn SH3 domain (WT) [1SHF]

31 folds faster Experimental folding rate increases with hydrophobicity of the amino acid residue substituted for N at exposed position 53 (normalized by Ala data) more hydrophobic (Fauchère & Pliška, 1983) Zarrine-Afsar, Wallin, Neculai, Neudecker, Howell, Davidson & Chan, PNAS (2008)

32 Augmenting the native-centric potential with HP-like, sequence dependent nonnative hydrophobic interactions total native-centric hφ Arash Zarrine-Afsar, Stefan Wallin et al., PNAS (2008)

33 Theory successfully rationalizes the increase in folding rate with hydrophobicity at position 53. hφ53 Theory also predicts that because of nonnative hφ interactions, some double mutants will fold faster while others will fold slower, depending on the specific positions of the mutations. Zarrine-Afsar, Wallin et al., PNAS (2008)

34 Theory predicts nonnative interaction partners of position 53 in the folding transition state κ 53 =1.0 κ 53 =3.0 difference L3, F4 Zarrine-Afsar, Wallin et al., PNAS (2008)

35 Nonnative contacts confirmed by double mutant thermodynamic cycle experiments positive synergy: 53-3, 53-4 control negative synergy: 53-40, A C L3 I L3 A53 L3 N ±0.11 kcal mol ±0.07 kcal mol -1 A3 I A3 A53 A3 N B F4 I F4 A53 F4 N ±0.07 kcal mol ±0.06 kcal mol -1 A4 I A4 A53 A4 N L3 A53 A3 A53 F4 A53 A4 A E R40 N53 R40 I53 T47 N53 T47 I ±0.09 kcal mol F ±0.08 kcal mol I40 N I40 I53 D I47 N (values are changes in folding barrier height) I47 I53 Zarrine-Afsar, Wallin et al., PNAS (2008)

36 Native Topology of the Designed Protein Top7 is not Conducive to Cooperative Folding Top7, 1qys (93aa) native-centric (Gō) Gō + nonnative HP Top7 Kuhlman et al., Science (2003) Model results rationalize experimental observations Zhang & Chan, Biophys J (2009)

37 whereas model free energy profiles of many natural proteins of comparable sizes (~90 aa) and with similar secondary structure elements indicate significantly higher degrees of cooperativity: nativecentric models with db Zhang & Chan, Biophys J (2009)

38 Rationalizing the difference in folding cooperativity between Top7 and ribosomal protein S6 Similar lengths Similar secondary structure elements: 2 α-helices, 5 or 4 β-strands S6 folds more cooperatively than Top7 in experiments Top7 and S6 have very different RCOs: 11 and 19 Top7 1qys, 92aa S6 1ris, 97aa Zhang & Chan, PNAS (2010)

39 The experimental folding kinetics of S6 is cooperative, that of Top7 is not Top7 S6 Severe chevron rollover Scalley-Kim & Baker, J Mol Biol (2004) Miller, Fischer & Marqusee, PNAS (2002) The midpoint folding rate of Top7 is ~ e 4 ( 55) times that of S6

40 Multiple-exponential relaxation and intermediates in the noncooperative folding of Top7 rates, k i amplitudes, A i mean first passage time (MFPT) = Σ i A i /k i four-state model? Figures from: Watters et al., Cell (2007)

41 Interplay of native topology and nonnative hydrophobic interactions rationalizes the markedly different chevron behaviors of Top7 and S6 db db+hφ db+hφ db Top7 S6 Zhang & Chan, PNAS (2010)

42 For our Top7 model with desolvation barriers and nonnative hydrophobic interactions, kinetic relaxation becomes multi-exponential under strongly folding conditions, in apparent agreement with experiment [Scalley-Kim & Baker, J Mol Biol (2004); Watters et al., Cell (2007)] db+hφ Top7 model results less denaturant Stronger folding conditions Zhang & Chan, PNAS (2010)

43 Top7 model predictions Intermediates & kinetic traps with nonnative contacts midpoint rollover strongly folding strongly folding C N CFr I 0 I 1 I 2 Zhang & Chan, PNAS (2010)

44 Many nonnative interactions in Top7 involves a long stretch of hφ 9 out of the 10-residue stretch FAAILIKVFA (from F63 to A72) are hφ nonnative contact map I 1 I 2 Model results consistent with experiment of Watters et al. (Cell 2007) showing that L29, V48, F63, A64, A65, L67, and V81 are involved in the stabilization of nonnative states. Zhang & Chan, PNAS (2010)

45 T m ( º C), CD Kinetic consequences of native-state optimization of surface-exposed electrostatic interactions Single and Fyn5 mutants reduce net charge relative to wildtype (WT) N - Src Loop N30 D16 RT - Src Loop d c b e a E11 E46 H21 Distal Loop All mutants fold faster than WT Schweiker, Zarrine-Afsar, Davidson & Makhatadze, Protein Sci (2007) WT Fyn5 r = T m ( º C), DSC ln k obs Fyn5 WT Urea (M) WT E46K D16K E11K H21K N30K E46K-E11K-D16K-H21K-N30K Fyn5 Zarrine-Afsar, Zhang, Schweiker, Makhatadze, Davidson & Chan, Proteins (2012)

46 Folding rates of the single and Fyn5 mutants of Fyn SH3 are rationalized by native-centric + general electrostatic interactions medium-range long-range Zarrine-Afsar, Zhang, Schweiker, Makhatadze, Davidson & Chan, Proteins (2012)

47 Most of the contact probability changes in the folding transition state of Fyn5 relative to that of WT occur among the native contacts, but there are changes among nonnative contacts as well. nonnative contacts Zarrine-Afsar, Zhang et al., Proteins (2012)

48 Summary Folding cooperativity should be used as a stringent modeling constraint to gain insight into real protein energetics. Many-body effects in the form of local-nonlocal coupling and desolvation barriers are likely contributors to folding cooperativity. Native topology likely places a constraint on the degree of folding cooperativity achievable by evolutionary or artificial design. Simple physics allows for nonnative hydrophobic interactions; the phenomenon may be viewed as a perturbation on a strong background of native bias.

49 Coworkers University of Toronto Tao Chen Jianhui Song Former group members: Artem Badasyan Mikael Borg Cristiano Dias Allison Ferguson Hüseyin Kaya Michael Knott Zhirong Liu Maria Sabaye Moghaddam Seishi Shimizu Stefan Wallin Zhuqing Zhang University of Toronto Prof. Alan R. Davidson Arash Zarrine-Afsar Prof. Julie D. Forman-Kay Tanja Mittag University of Calgary Prof. D. Peter Tieleman Justin MacCallum University of Münster Prof. Erich Bornberg-Bauer Richard Wroe, David Vernazobres, Tobias Sikosek Supported by the Canadian Institutes of Health Research & the Canada Research Chairs Program