Can nuclear theory help find an anti-viral drug for - COVID-19? Pietro Faccioli - Istituto Nazionale di Fisica Nucleare
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
REDUCTIONIST’S APPROACH TO MOLECULAR BIOLOGY Challenge: Integrate ~106 coupled Newton-type equations looking for extremely rare events
MD YIELDS CORRECT PROTEIN NATIVE STATES Anton supercomputer MD (DES Research) Atomic-Level Characterization of the Structural Dynamics of Proteins How Fast-Folding Proteins Fold Atomic-Level Characterization of the Structural Dynamics of Proteins reso David E. Shaw, et al. David 330 Science E. Shaw , et al. , 341 (2010); prot Science DOI: 330, 341 (2010); 10.1126/science.1187409 unst DOI: 10.1126/science.1187409 Kresten Lindorff-Larsen,1*† Stefano Piana,1*† Ron O. Dror,1 David E. Shaw1,2† to f sam An outstanding challenge in the field of molecular biology has been to understand the process pert by which proteins fold into their characteristic three-dimensional structures. Here, we report the hom This copy is for your personal, non-commercial use only. results of atomic-level molecular dynamics simulations, over periods ranging between 100 ms F and 1 ms, that reveal a set of common principles underlying the folding of 12 structurally diverse we This copy is for your personal, non-commercial use only. proteins. In simulations conducted with a single physics-based energy function, the proteins, the
ZOOLOGY OF ENHANCED SAMPLING METHODS Markov State Models, Milestoning, Transition Path Sampling, Transition Interface Sampling, Forward Flux Sampling, Temperature Accelerated Molecular Dynamics, Metadynamics, Umbrella Sampling, Blue Moon Sampling, String Method,Stochastic Difference, … [and counting] They are all too computationally demanding for many biologically relevant problems.
A USEFUL ANALOGY Thermal activation ( = (KB T ) 1 ) kB T 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 Z xf Rt d⌧ (M q̈+M q̇+rU (q))2 P (xf , t|xi ) = Dq e 4M 0 xi Quantum tunneling Z xf Rt 2 1 d⌧ ( M 2 q̇ +U (q)) KE (xf , t|xi ) = Dq e ~ ~ 0 xi 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
ADVANTAGES tMFPT tTPT ✓ ◆ tM F P T tT P T ⇠ ⌧0 log log AAAD9HicbVNLb9QwEE53eZTwaAtHLhbVSluxtNmCBBekqlSll6ql3T6kOESOd7Jr1nnInlRUUfoLuMKJG+LK/+HAbwE7G+gLX/zNN994HhpHuRQaPe/nTKt94+at27N33Lv37j+Ym194eKizQnE44JnM1HHENEiRwgEKlHCcK2BJJOEomryx/qMTUFpk6QBPcwgSNkpFLDhDQ4ULM787VI2zsJRJ1cUlQnOV5ZgRKlIkNGE45kyWG9XeNYbmWly0CY2YIjV7tlO/51ujd873yF5wBu9LGivGS1GVdGxcFZUQY5eI2iL7YbmzXfl7AXlq8f7FVywKatrUYjQ9ciVDQJUYjXGpcjsYltubuwNTlhYJociK0CMUPuY+ObNX1wgGxr8ydS25jX1ZL7NRXZ5/DrvT8v++b7qotVWTenoFboeKOHY7ZkR1W9MgmjOFgsnqHyJYTWf22ggZkq3aMtEnwMvNytCmGIuZ6WkSRj0iQ69HJiEzbeMYkFkEZiIfKsPIYYbaCuOmD9vwM7oBEhl5uzIJ1weBda8PTApbmBvOL3rLXn3IddBvwKLTnN1w/hcdZrxIIEUumdZ+38sxKG07XELl0kJDzviEjcA3MGUJ6KCsN7UinUIzs105KCIkqUm4GFGyROvTJDJKu1n6qs+S//P5BcavArMWeYGQcpsIhYQ6keZKmC8AZCgUIDJbORCREs4UQwQlCOPckIX5E3Ye/avdXweHq8v958ur714srq03k5l1HjtPnK7Td146a86Ws+scOLwVtz61Pre+tE/aX9vf2t+n0tZME/PIuXTaP/4ALz9Gng== ⌧0
VARIATIONAL APPROACHES TO TRANSITION PATH SAMPLING Dominant Reaction Pathways PHYSICAL REVIEW LETTERS week ending PHYSICAL REVIEW LETTERS week ending PRL 97, 108101 (2006) 8 SEPTEMBER 2006 PRL 99, 118102 (2007) 14 SEPTEMBER 2007 Dominant Pathways in Protein Folding Quantitative Protein Dynamics from Dominant Folding Pathways 1,2 3 4 5 P. Faccioli, M. Sega, F. Pederiva, and H. Orland M. Sega,1 P. Faccioli,2,3 F. Pederiva,4 G. Garberoglio,1 and H. Orland5 1 Dipartimento di Fisica Universitá degli Studi di Trento e I.N.F.N, Via Sommarive 14, Povo (Trento), I-38050 Italy 1 C.N.R./I.N.F.M. and Dipartimento di Fisica, Università degli Studi di Trento, Via Sommarive 14, Povo (Trento), I-38050 Italy 2 European Centre for Theoretical Studies in Nuclear Physics and Related Areas (E.C.T.*), (2005) (2006) 2 Dipartimento di Fisica Università degli Studi di Trento e I.N.F.N, Via Sommarive 14, Povo (Trento), I-38050 Italy Strada delle Tabarelle 284, Villazzano (Trento), I-38050 Italy 3 European Centre for Theoretical Studies in Nuclear Physics and Related Areas (E.C.T.), Strada delle Tabarelle 284, 3 C.N.R./I.N.F.M. and Dipartimento di Fisica, Universitá degli Studi di Trento, Via Sommarive 14, Povo (Trento), I-38050 Italy Villazzano (Trento), I-38050 Italy 4 Dipartimento di Fisica and C.N.R./I.N.F.M.-DEMOCRITOS National Simulation Center, Universitá degli Studi di Trento, 4 Dipartimento di Fisica and C.N.R./I.N.F.M.-DEMOCRITOS National Simulation Center, Università degli Studi di Trento, Via Sommarive 14, Povo (Trento), I-38050 Italy Via Sommarive 14, Povo (Trento), I-38050 Italy 5 Service de Physique Théorique, Centre d’Etudes de Saclay, F-91191, Gif-sur-Yvette Cedex, France 5 Service de Physique Théorique, Centre d’Etudes de Saclay, F-91191 Gif-sur-Yvette Cedex, France (Received 5 December 2005; published 6 September 2006) (Received 30 January 2007; published 12 September 2007) Dominant folding pathways of a WW domain We present a method to investigate the kinetics of protein folding and the dynamics underlying the formation of secondary and tertiary structures during the entire reaction. By writing the solution of the We develop a theoretical approach to the protein-folding problem based on out-of-equilibrium stochastic dynamics. Within this framework, the computational difficulties related to the existence of Fokker-Planck equation in terms of a path integral, we derive a Hamilton-Jacobi variational principle from folding Škrbić , Roberto Covino , and Pietro Faccioli large time scale gaps are removed, and simulating the entire reaction in atomistic details using existing a,b a,c a,b a,b,1 which we are able to compute the most probable pathway of folding. The methodSilvio a Beccara is applied to the , Tatjana of the Villin headpiece subdomain simulated using a Go model. An initial collapsing a phase Dipartimento by the degli Studi di Trento, Via Sommarive 14, I-38123 Povo (Trento),computers drivenUniversità di Fisica, becomes feasible. We discuss how to determine the most probable folding pathway, identify Italy; bINFN Istituto Nazionale di Fisica Nucleare initial configuration is followed by a rearrangement phase, in which secondary structures (National are for Institute formed Nuclearand Physics), Gruppo Collegato di Trento, Via Sommarive 14, I-38123 Povo configurations (Trento) Italy; representative of the and cEuropean Centre for transition Theoretical state, and compute the most probable transition time. We Studies in Nuclear Physics all computed paths display strong similarities. This completely general method does not require the prior and Related Areas, Strada delle Tabarelle 286, I-38123 Villazzano perform (Trento), Italy an illustrative application of these ideas, studying the conformational evolution of alanine knowledge of any reaction coordinate and is an efficient tool to perform simulations Edited byof the entire William A. Eaton,folding dipeptide, within an all-atom model National Institutes of Health -NIDDK, Bethesda, MD, and approved December 19, 2011 (received for review July 27, 2011) based on the empiric GROMOS96 force field. process with available computers. We investigate the folding mechanism of the WW domain Fip35 DOI: 10.1103/PhysRevLett.99.118102 Noé, et al. performed a Markov state model analysis of a large PACS numbers: 87.14.Ee, 83.10.Mj, 87.15.Cc DOI: 10.1103/PhysRevLett.97.108101 PACS numbers:using a realistic 87.14.Ee, atomistic 83.10.Mj, force field by applying the Dominant 87.15.Cc number of short (≲200 ns) nonequilibrium MD trajectories (11) Reaction Pathways approach. We find evidence for the existence performed on the WW domain of human Pin 1 protein. In their of two folding pathways, which differ by the order of formation paper the authors reported a complex network of transition path- (2012) Understanding the kinetics of protein folding [1] and the structure was obtained from of the twothe Brookhaven hairpins. Protein This result Data with the analysis of is consistent A critical ways, which differpartbyofthethe protein-folding specific order in which problem is tolocal the different under- In this Letter we further develop our formalism and we dynamical mechanisms involved in the formation of their Bank. the experimental data on the folding kinetics of WW domains meta-stable stand states were its kinetics andvisited. On the other the underlying hand, in processes. physical all pathways To present the first DFP simulation performed in full atomistic structures in an all-atom approach involves simulating a andon Our study is based with thetheanalogy results obtained betweenfrom large-scale molecular dynamics Langevin the formation of hairpins takes place in a definite sequence (see simulations of this system. Free-energy calculations performed in this aim,2).several e.g., Fig. different In particular, fromtheoretical the statisticalmethods model ithavewas in-been detail. We show how the DFP analysis gives access to statistically significant ensemble of folding trajectories for diffusion and quantum propagation. Previous studies have important information about the dynamics of the folding two coarse-grained models support the robustness of our results recently ferred thatdeveloped, in about 30% spanning from transitions, of the folding analyticalthe approaches second a system of !104 degrees of freedom. Unfortunately, the exploited such a connection to study a variety of diffusive process, such as the characterization and determination of and suggest that the qualitative structure of the dominant paths [1] to detailed hairpin forms first, computer as in the simulations right box. [2,3]. A major prob- existence of a huge gap between the microscopic time problems using path-integral methods are mostly shaped by [4,5]. In this the native work, Computing a folding interactions. A indifferent conclusion has beenprocess reached using by Shaw, et al., by Bias Functional Approach Self Consistent Path Sampling lem simulating the folding standard mo- the transition state, and the most probable transition time. scale of the rotational degrees of freedom !10"12 s and we develop the formalism to in trajectory determine explicitly atomistic detail the evo-about one hour on 48 only required analyzing a ms-long MD trajectory with multiple unfolding/ lecular dynamics (MD) is the huge gap between the time In addition, we show that in this formalism the native state the macroscopic time scales of the full folding process lution of the position of each monomer of the protein, Central Processing Units. The gain in computational efficiency refolding events, obtained using a special-purpose supercompu- opens the door to awithout systematic investigation scale ofIn‘‘elementary that simulationmoves,’’ of theoforder Fip35of was10 –100to ps, is characterized by a single effective parameter, and this !10"6 –101 s makes it extremely computationally chal- during the entire folding transition, relying on a of the folding path- ter (12). the WW domain found lenging to follow the evolution of a typical !100-residue ways of a large number of globular proteins. specific choice of the reaction coordinate. and thatunfold fold and of thepredominantly entire folding alongprocess, a pathwaywhich ranges in which hairpinfrom 1 a leads to an interesting relationship between kinetic and Before entering the details of our ∣calculation, is fully structured, before hairpin 2 begins to fold, few microseconds for fast folders [4], up to several seconds as shown in the thermodynamical quantities. protein for a time interval longer than a few tens of atomistic simulations protein folding it is con- left box of Fig. 2. In a recent paper (13), Krivov reanalyzed the nanoseconds. venient to review the mathematical framework in a simple or even minutes for more complex proteins. This peculiar- Let us begin our discussion by briefly reviewing the key same ms-long MD trajectory in order to identify an optimal set of concepts of the DFP method, here presented for a simple Several approaches have been proposed to overcome such 114, 098103 (2015) PRL computational difficulties and address the problem case. For this purpose, let U us consider nveiling Langevin the mechanism P H Y S I C A L R E V I E W L E T T E R Snative structure remains one by diffusion of a point particle in one dimension, subject to an6 external problems at the interface of which MARCH proteins fold into their week ending of the 2015 contemporary fundamental open molecular biology, ity of the reaction dynamics folding process coordinates. WW domainapproach is thermally too makeswas His conclusion demanding, activated thethat brute-force the foldingmolecular andincipient rather than a substantial of this downhillpart one-dimensional system, without loss of generality. The DFP method can be applied to any system described of identifying the relevant pathways of the folding re- potential U#x$: andthe of thatefforts the transition also occurs in the field through a second of protein-folding pathway, aims simulation in biochemistry, and biophysics. A critical point concerns the char- which hairpin 2 forms before hairpin 1. The statistical weights of by the over-damped Langevin equation action [2]. Unfortunately, these methods are either af- D @U at bridging this gap. Variational Scheme to Compute ProteintoReaction Pathways @x acterization Using of the ensemble of reactive trajectories connecting Atomistic the two pathways estimated from the number of folding events fected by uncontrolled systematic errors associated % the "denatured & !#t$; Force Fields (1) areIn a recent paper 20%[5] we presented a novel theoretical D @U k T @xand native states, in configuration space. 80% ! 20% and ! 10%. @x ad hoc approximations or can be applied only with to smallExplicit Solvent @t In thisB context, a fundamental question which has long been framework While all thesefor investigating theoretical studies theyield folding folding dynamics, named times in rather @t !" kB T @x # !$t%; (1) proteins with a typical folding time of the order of a few where !#t$ is a Gaussian noise with zero average and globular proteins in- debated (1) is whether the folding of typical good agreement hereafter Dominant with available Foldingexperimental Pathways (DFP), data on which folding is volves a few dominant pathways; nanoseconds (fast folders). In this Letter, S. a we present Beccara, 1,3 a 2 correlation L. Fant, given by and P. Faccioli 2,3,*h!#t$!#t 0 $i % 2D"#t " t0 $.i.e., In well thisdefined and conserved sequences of secondary and tertiary contact formation, or if it can kinetics, they provide different pictures of the based on a reformulation in terms of path integrals of the folding mechanism where U is the potential energy of the system, !$t% is a novel approach to overcome 1 these difficulties: We adopt and raise a number of issues. Gaussian random force with zero average and correlation European Centre for Theoretical Nuclearequation, D isRelated Physics and the diffusion Areas take constant through aofmultitude place(ECT*-FBK), the particle in the different routes. A of qualitatively dynamics described Firstly, it is important by the Langevin to assess the degree equation. The of of heterogeneity DFP the Langevin approach and devise a method Strada to rigor- delle Tabarelle solvent; k(Trento) 287, Villazzano B and T are, respectively, 38123, importantthe Italy Boltzmann constant given by h!$t%!$t0 %i ! 2D"$t " t0 %. Note that in the origi- (2015) (2017) related question concerns the role of nonnative interac- analysis the foldingallowsmechanism to compute rigorously and to clarify whether (i.e., without the most statisti-any ously define2Dipartimento and practically compute di Fisica, the most Università deglistatisti- and theVia Studi di Trento, temperature. Sommarivetions 14, in determining Povo (Trento)the structure 38123, Italyof the folding pathways (2, 3). cally significant other assumptions folding than pathways thearevalidity those in of whichthetheunderlying hairpins nal Langevin equation there is a mass term, mx. ! However, 3 cally relevant protein folding Trento pathway. Institute As a first Physics for Fundamental explor-and Applications (INFN-TIFPA), In principle, atomistic molecular Via Sommarive 14, dynamics (MD) simulations form in sequence. Important as shown in [6], for proteins, this term can be neglected provide a consistent framework to address these problems from a Langevin equation) the related most questions probableareconformational also whether atory application, we have studied the foldingPovo transi- (Trento) 38123, Italy the folding mechanism is correlated with the structure of the beyond time scales of the order of 10 "13 s. theoretical perspective. However, due to their high computational pathway connecting initial denatured two from conditions arbitrary whichconformations. The ma- the reaction is initiated tion of the 36-monomer (Received 23Villin headpiece May 2014; revisedsubdomain manuscript received 29 October 2014; published 4 March 2015)
HUGE COMPUTATIONAL GAIN Using top all- Using top purpose special-purpose supercomputers supercomputer
VALIDATING SCPS AGAINST MD 106s MD Protein , folding ms s minutes
VENTURING INTO THE BIO-ZONE 106s MD Variational , Simulations ms s minutes hours
A LONG JOURNEY 3.8 3.6 3.4 3.2 3 2.8 2.6 2.4 2.2 2 1.8 Experiment 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Theory -0.2 -0.4 -0.6 -0.8 -1 -1.2 -1.4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
VALIDATION AGAINST EXPERIMENT Experiment Challenge: Most available techniques provide only indirect probes, we seek for Theory direct validation 2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
The Journal ARTICLE ARTIC ics of Chemical Physics scitation.org/journal/jc TIME-DEPENDENT LINEAR SPECTROSCOPY An analytic formula for the absorptionAn analytic formulκ coefficient be obtained by combining real and be obtained imaginarybyparts, combini acc Eqs. (34)–(36). It is instructive to Eqs. (34)–(36). analyze It is instr the structure of nary part ImR(ω), which controlsnary part ImR(ω), the position which and the wi resonances resonances 2ReΣn (ω)�(ω − En )2 + �Σn (ω)�2 �2 f f ImR(ω) = − � ImR(ω) = −� n −�(ω n �(ω − En )2 − �Σf (ω)�2 � + 4�(ω − En En )ReΣ 2 n FIG. sentation of the 1. (a)Eq. Dyson Graphical representation (44) for the exciton propa- of the Dyson Eq. (44) for the exciton propa- gator. The white triangle denotes theline full with open white triangle (non-perturbative) time- denotes the full (non-perturbative) time- ordered or, while the other exciton propagator, continuous line appearing whilein the the other continuous line appearing in the he free exciton righthand-side propagator. (b)represents Example ofthe free loop The splitting exciton propagator. diagram (b) Exampleandof loop diagram of theThe shifting splitting poles generatedand by shiftin the Challenge: Σn (ω)b f approximation.neglected The dashed in the lineproposed approximation. denotes the stochastic The dashed line denotes coupling the is determined stochastic coupling function by the self-energy is determined mped Langevinpropagator oscillationsofofclassical damped Langevin the configuration vector oscillations of the Eq. (46). Forconfiguration illustration vector purposes,Eq. (46). here weFor illustration discuss its expr δQ. (c) Estimating term by the lowest-order Ground state self-energy thediagram. Need the case of a single a theory 1PI term by the lowest-order self-energy diagram. normal forof a single norm mode the case 2 γ ⌦2 + i(ω − Edynamics non-equilibrium n )�(ω f − En ) δ− γ22 2 2 δ nm f our nm⌦f 2 +γ⌦ Σnm (ω)2 = Equation point (43) provides the starting point to apply non- Σ (ω) = ovides the starting to apply our non- One exciton f perturbative mation scheme. approximation To this end, we consider the To thisnmend, scheme. of wequantum consider βM ⌦ �(ω − the En )2 − ⌦2 � + γβM electronic 2 (ω⌦ − En ) standard on, obtained Dyson equation, by resumming obtained all 1-particle irre- by resumming all 1-particle irre- ms for the ducible (1PI) diagrams single-exciton propagator forGthe Thisexcitations equation [seesingle-exciton shows G propagator in how conformationally [see This equation the vibronic shows correction ho of the Fig. 1(a)]. y representation and In frequency omitting representation all indices for function scalesallwith and omitting indices for function the temperature scalesviscosity. and bath with the sake of Dyson equation simplicity, the Dyson equation reads reads evolving proteins the shifting, the shifting, splitting, and broadening of the splitting, resonances an when when the difference between ω and thethe difference excitonic betw energie
MOLECULAR QUANTUM FIELD THEORY* Z Z SM QF T [ , ¯,q] Z= D D¯ Dq e SM QF T [q, , ¯] = SOM [q] + SS [ , ¯] + Sint [q, , ¯] Z t 2 SOM [q] = d⌧ (M q̈ + M q̇ + rU (q)) 0 4M XZ t SSSel.[ [ ¯,, ¯] = d⌧ ¯n (⌧ ) i~@t h0nm m (⌧ ) 0 XXZ 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 n,m t [q, , ¯¯]] = ¯n i SSint int [R, d⌧ fnm m rqi 0 i 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 nm i P. Faccioli & E. Schneider (2013-2016)
SOLVING MQFT: AN ARSENAL OF METHODS Perturbation Theory PRB 2012, PRB 2013, PRB 2016 Quantum MC (for real time) PRB 2016 Renorm. Group & Eff. Field Theory PRB 2013, JCP 2016
EXAMPLES OF DIRECT COMPARISON WITH EXPERIMENTS 7 Time resolved near UV CD* Linear absorption spectrum 0.4 Experimental Calc. S S 1.2 (a) Calc. S S (b) Calc. SH HS Experimental Experimental Calculated Numerical 1 Normalized Absorption 0.2 0.8 () M 1cm 1 theory (a) (c) 0.6 0 0.4 0.2 0 0.2 (b) (d) 12000 12150 12300 12450 12600 12750 experiment Frequency (cm ) -1 FIG. 3. (a) Trimeric form of the FMO complex. Only the protein Microscopic Calculation of Absorption Spectra of Macromolecules: an Analytic Approach scaffold is shown with its C3 symmetry. (b) Monomeric unit of Matteo Carli the FMO with the 7 BChl-a (blue) wrapped in the protein scaffold. FIG. 4. Comparison between Physics Department numerical of Trento University, Via Sommarive 14, spectrum (red) Povo (Trento), 38123, Italy andand the ex- 0.4 (c) Representation of the ENM corresponding to a single monomer. Scuola Internazionale Superiore di Studi Avanzati (SISSA), via Bonomea 265, Trieste 34136, Italy. perimental spectrum (blue) from Ref. [58]. The non-perturbative ⇤ Michele Turelli and Pietro Faccioli 200 230 260 290 The red beads are centered at the CA position of each residue. The200 230 260 290 prediction has been Physics obtained Department via Viathe of Trento University, algorithm Sommarive in38123, 14, Povo (Trento), theItalySM. and Trento Institute for Fundamental Physics and Applications (INFN-TIFPA), Via Sommarive 23, Povo (Trento), 38123, Italy The total spring-like Wavelength bonds between (nm) beads falling within aWavelength radius of 10 Å (nm) are simulation We timedevelopwas of 1 ps a cross-disciplinary with approach time-steps to analytically compute optical of 0.05 response fs. functions of open macro- molecular systems, by exploiting the mathematical formalism of quantum field theory (QFT). Indeed, the en- shown in grey. (d) Atomic structure of the BChla (grey) with the 8 tries of the density matrix for the electronic excitations interacting with their open dissipative environment are mapped into vacuum-to-vacuum Green’s functions in a fictitious relativistic closed quantum system. We beads chosen for its coarse-grained model in the ENM (red). The show that by re-summing appropriate self-energy diagrams in this dual QFT it is possible to obtain analytic expressions for the response functions in Mukamel’s theory. This yields physical insight into the structure and Article beads are centered at the positions of the MG, C2A, C2B, C2C, Cite This: J. Am. Chem. Soc. 2018, 140, 3674−3682 Withinand the MQFT, microscopic model parameters.the conformational For illustration, we apply this scheme to compute dynamics the linear absorption is de- dynamics of vibronic resonances, since their frequency and width is related to fundamental physical constants pubs.acs.org/JACS spectrum of the Fenna-Matthews-Olson (FMO) light harvesting complex, comparing analytic calculations, nu- C2D atoms for the chlorin ring and at the C2, C10, C18 positions scribed by the Langevin equation. For sake of simplic- merical Monte Carlo simulations and experimental data. Atomic Detail of Protein Folding Revealed by an Ab Initio for the three beads Reappraisal of Circular Dichroism of the phytol tail. The atomic labelling is the ity, we assign the same viscosity I. INTRODUCTION systems parameter to theall [13, 17–26]. In this context, beads, Feynman-Vernon sameSimone Alan Ianeselli, found † Orioli,inGiovanni the crystal ‡,∥ structure Spagnolli, Pietro file (PDB:3EOJ). † Faccioli,* Lorenzo Cupellini, ,‡,∥ § 1 g = 150 ps . In principle, viscosity path integral formalism offers several advantages [10, 27– coefficients 34]. For example, for toeach in makes it straightforward of deal with the Sandro Jurinovich,§ and Benedetta Mennucci*,§ The relaxation dynamics of optically induced electronic excitations in macromolecular systems has been extensively dynamics of atomic nuclei at the classical level, while retain- the constituents may be evaluated from microscopic molec- † Centre for Integrative Biology, Trento University, Via Sommarive 9, 38128 Povo, Trento, Italy ing a full quantum description of the excitons [10, 29–34]. * with B. Mennucci’s Lab (U.as Pisa) ‡ studied during the last several years, in view of its important Physics Department, Trento University, Via Sommarive 14, 38128 Povo, Trento, Italy Within these mixed quantum-classical schemes it is possible biological and technological implications. For example, the ular dynamics simulations, e.g. as picoseconds, by computing appropriate § Dipartimento di Chimica e Chimica Industriale, University of Pisa, via G. Moruzzi 13, 56124, Pisa, Italy to simulate the real-time dynamics for time intervals long ∥ propagation of excitons in light harvesting complexes pro- INFN-TIFPA, Via Sommarive 14, 38128 Povo, Trento, Italy using a Quantum Monte Carlo algorithm.
A LONG JOURNEY 3.8 3.6 3.4 3.2 3 2.8 2.6 2.4 2.2 2 1.8 Experiment 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Theory -0.2 -0.4 -0.6 -0.8 -1 -1.2 -1.4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
EXPLORING BIOLOGICAL PROCESSES Article All-Atom Simulations Reveal How Single-Point Mutations Promote Serpin Misfolding Fang Wang,1 Simone Orioli,2,3 Alan Ianeselli,2,3 Giovanni Spagnolli,2,3 Silvio a Beccara,2,3 Anne Gershenson,4,* Pietro Faccioli,2,3,* and Patrick L. Wintrode1,* 1 Department of Pharmaceutical Sciences, University of Maryland School of Pharmacy, Baltimore, Maryland; 2Dipartimento di Fisica, Università degli Studi di Trento, Povo (Trento), Italy; 3Trento Institute for Fundamental Physics and Applications, Povo (Trento), Italy; and 4Department of Biochemistry and Molecular Biology, University of Massachusetts Amherst, Amherst, Massachusetts ABSTRACT Protein misfolding is implicated in many diseases, including serpinopathies. For the canonical inhibitory serpin a1-anti- trypsin, mutations can result in protein deficiencies leading to lung disease, and misfolded mutants can accumulate in hepatocytes, leading to liver disease. Using all-atom simulations based on the recently developed bias functional algorithm, we elucidate how wild- type a1-antitrypsin folds and how the disease-associated S (Glu264Val) and Z (Glu342Lys) mutations lead to misfolding. The dele- terious Z mutation disrupts folding at an early stage, whereas the relatively benign S mutant shows late-stage minor misfolding. A number of suppressor mutations ameliorate the effects of the Z mutation, and simulations on these mutants help to elucidate the relative roles of steric clashes and electrostatic interactions in Z misfolding. These results demonstrate a striking correlation between atomistic events and disease severity and shine light on the mechanisms driving chains away from their correct folding routes. INTRODUCTION RESEARCH ARTICLE Understanding how mutations alter protein misfolding pro- mediate state (4). Similarly, Z secretion from cells is slow, Full atomistic model of prion structure and pensities and the physicochemical mechanisms underlying this shift is key to clarifying the molecular basis of many dis- and although some Z species are targeted for degradation (5–7), misfolded Z accumulates in the endoplasmic reticu- conversion Teaming up with eases. One set of relatively common protein-misfolding dis- lum, where it can polymerize (8). Despite numerous exper- eases known as serpinopathies arises when mutations in imental studies (9–13), little is known about the structure of inhibitory serpins lead to misfolding, thus reducing the Giovanni Spagnolli ID , Marta1 misfolded species Rigoli ID 1,2 for any , Simone A1AT Orioli 2,3 disease-associated mutant,4 , Alejandro M. Sevillano ID , secreted levels of these important protease inhibitors Pietro Faccioli 2,3 (1). hindering , Holger Wille 5 efforts to ID , Emiliano either1 rescue Biasini , Jesús theR.folding Requenaof these ID 6 species Mutations in the canonical secretory serpin a1-antitrypsin or to target them for degradation. 1 Department of Cellular, theComputational and Integrative Biology (CIBIO)–University of Trento, Povo TN, E. Biasini’s lab (DICIBIO) (A1AT) result in the most common serpinopathies: Molecular dynamics (MD) simulations offer an attractive ITALY, 2 Department of Physics, Povo, Trento TN, ITALY, 3 INFN-TIFPA, Povo (Trento), ITALY, A1AT deficiencies. In these deficiencies, low circulating approach to studying protein folding and misfolding, as 4 Department of Pathology–University of California—San Diego, San Diego, California, United States of A1AT levels dysregulate leukocyte serine proteases, result- they can in principle reveal folding pathways and intermedi- America, 5 Department of Biochemistry and Centre for Prions and Protein Folding Diseases–University of ing in lung disease, which can be Alberta, slowedEdmonton but not (AB), halted ates6 in CANADA, atomistic CIMUS detail. Biomedical To date, Research the application Institute & Department ofofall-atom Medical a1111111111 by A1AT augmentation therapy (2).Sciences, Extremely University of SantiagoMD pathogenic de Compostela-IDIS, Santiago, SPAIN simulations to investigate protein folding and misfolding a1111111111 A1AT mutations such as Z (Glu342Lys) can lead to both has been limited to small, single-domain proteins with rela- a1111111111 * giovanni.spagnolli@unitn.it (GS); emiliano.biasini@unitn.it (EB); jesus.requena@usc.es (JRR) lung disease, because of loss of function, and liver disease, tively short folding times. In particular, recent developments, a1111111111 a1111111111 All-Atom Simulation of the HET-s Prion Replication because of A1AT accumulation in the endoplasmic reticulum of hepatocytes, which generate most of the circulating A1AT. such as the Anton special-purpose supercomputer (14) and the massively distributed folding@home project (15), have With the exception of liver transplants, Abstract there are no effective made it possible to generate in silico several reversible treatments1,2for A1AT-associated liver disease 2,3 # (3). folding/unfolding 1,2 events for4a number of small globular 2,3# pro- Luca Terruzzi *, Giovanni Spagnolli Prions In vitro, the pathogenic A1AT Z mutant folds very * , areAlberto unusual Boldrini protein teins, (
A LONG JOURNEY 3.8 3.6 3.4 3.2 3 2.8 2.6 2.4 2.2 2 1.8 Experiment 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Theory -0.2 -0.4 -0.6 -0.8 -1 -1.2 -1.4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5
! ROLE OF PROTEIN INACTIVATION MOST OF BIOLOGICAL FUNCTIONS IN CELLS ARE CARRIED OUT BY PROTEINS MOST OF MEDICINAL CHEMISTRY IS BASED ON INHIBITING BIOLOGICAL FUNCTIONS OF PROTEINS
PHARMACOLOGICAL PROTEIN INACTIVATION BY ! FOLDING INTERMEDIATE TARGETING patent file # 102018000007535 (with E. Biasini) Protein Protein DNA mRNA Ribosome folding function RNA silencing Eliminating Suppressing Suppressing Imparing protein from genome transcription folding function
PHARMACOLOGICAL PROTEIN INACTIVATION BY ! FOLDING INTERMEDIATE TARGETING Unfolded I1 I2 Native state state drug I3 Unfolded proteins Degradation
A F I R S T VA L I D AT I O N Inactivation of SM875 [µM] Cellular Vhc 0.01 0.03 0.1 0.3 1 3 10 30 50 150 PrPC Expression (% Vhc) Prion protein 37 100 PrPC *** 50 *** 25 0 10 30 1 1 3 c 0. Vh SM875 [µM] 20 (ii) SM875 [µM] Vhc 1 3 10 30 50 NEGR-1 (iii)
PPI-FIT PIPELINE in-vitro evaluation Computing the folding In-silico virtual of the effect of the Refinement pathways screening on folding selected compounds SAR, characterizing intermediate targets on protein expression pharmaco-kinetics, intermediate states ….. FINAL GOAL Preclinical and clinical trials Slide courtesy of Sibylla Biotech SRL
18 Joining Forces against COVID-19 Maria Letizia Emiliano Biasini Pietro Faccioli Graziano Lolli Barreca 30.000 cores in 8 data centers Lidia Pieri Giovanni Tania Luca Terruzzi Andrea Astolfi Alberto Boldrini Spagnolli Massignan
16 SARS-CoV-2 Replication Spike ACE2 Goals: Repurposing of approved drugs! Looking for suppressors of ACE2 expression levels Figure taken from: https://theconversation.com/where-are-we-at-with-developing-a-vaccine-for-coronavirus-134784
14 PPI-FIT ON ACE2 Out of 9000 candidates, we found 35 molecules binding in-silico the intermediate. Validation experiments on cellular bio-assays are ongoing.
BREAKING NEWS!! (17/05/2020) So far, Sibylla Biotech has tested 14 virtual hits ONE DISPLAYS A PROMINENT EFFECT WITH CLEAR DOSE-RESPONSE RELATIONSHIP AND VERY LOW TOXICITY 200 Ace2 expression (%Vhc) 150 100 0.01 0.03 50 Vhc 100 300 0.1 0.3 10 30 1 3 0 01 03 1 3 0 1 3 10 30 0 0 0. 0. 10 30 0. 0. [µM]
PPI-FIT Pipeline in-silico hit compounds in-vitro evaluation ACE-2 folding in-vitro evaluation identification. Virtual of the effect of pathway of the effect of the screening on FDA compounds lowering reconstruction and selected compounds approved or the expression of ACE2 intermediate state on ACE2 expression investigational drugs on SARS-COV-2 characterization (degradation wanted) virus replication FINAL GOAL Preclinical and clinical trials
MAIN TAKE-HOME MESSAGES Fundamental research matters! The usefulness of theoretical physics extends beyond its natural cultural perimeter Major research infrastructures for fundamental research can be re-purposed Technological transfer can boost research and help advance Science
18 People Maria Letizia Emiliano Biasini Pietro Faccioli Graziano Lolli Barreca PhD PhD PhD PhD Lidia Pieri Giovanni Tania Andrea Astolfi Alberto Boldrini Luca Terruzzi PhD Spagnolli Massignan PhD PhD Founder - CEO Founder Staff Staff Consultant Chief Scientist (Computational) Staff (Wet Lab) (Computational) Medicinal Chemist
Acknowledgments Italy: Trento: E. Biasini, A. Ianeselli (2014-2017), G. Spagnolli S. A Beccara (2009-2017), S. Orioli (2014-2018), E. Schneider (2012-2015), M. Carli (2017), M. Turelli (2018), F. Mascherpa (2014), G. Garberoglio, F. Pederiva, M. Sega, R. Covino (2012-2015), Pisa: B. Mennucci, L. Cupellini, S. Jurinovich Perugia: L. Barreca SISSA: C. Micheletti, A. Laio Europe: U. Compostela: J. Raquena U. Zurich: B. Schuler CEA-Saclay: H. Orland USA: U. Maryland: P. Wintrode U. Mass.: A. Gershenson
You can also read