Computational Chemistry CHIMIA 2011, 65, No. 7/8 507 doi:10.2533/chimia.2011.507 Chimia 65 (2011) 507–518 © Schweizerische Chemische Gesellschaft

Computational Chemistry Computational Chemistry Computational Chemistry CC 1 Computational Chemistry CC 2 Evaluation of thenon(a)diabaticityofthe quantummolecular OxygenDynamics and EnergeticsofMigration Pathways in dynamicsusingthe dephasingrepresentation of quantum Truncated Hemoglobin fidelity[1] Pierre-Andr´e Cazade,Markus Meuwly Tom´aˇsZimmermann, Jiˇr ´ı Van´ıˇcek DepartmentofChemistry,UniversityofBasel, Klingelbergstrasse 80, Laboratory of Theoretical Physical Chemistry,Institut desSciences et CH-4056 Basel Ingnierie Chimiques, Ecole Polytechnique Fdrale de Lausanne, CH-1015, Switzerland Truncatedhemoglobin (trHbN) is aprotein involved in NO dioxygenation [1,2] by converting nitric oxidetoharmless nitrates (Fe(II)O2 + NO + → We proposetwo approximate methods forsemiclassicalcalculation of the Fe (III)+NO3−). The tunnel system of trHbN plays an importantrole quantum fidelityinthe non-Born-Oppenheimerdynamics. This quantity in determining and controlling ligandentrance,migration, and rebind- canbeused to evaluate the dynamical importanceofnonadiabatic or ing.[2] Due to agreateraffinityofFe(II) for NO than for O2,the firststep diabatic couplings between potential energy surfaces (PES)orofother of dioxygenation is expected to be the substitution of NO ligated to the heme groupbyfree O (Fe(II)NO+O Fe(II)O +NO). The three- terms in the Hamiltonian such as spin-orbit couplings.The acquired 2 2 → 2 information canhelpreducethe complexityofastudiedsystemwithout pointfluctuating charge model is usedtoreproduce accurately the dipole significantly affecting the accuracyofthe quantum simulation. Another and quadrupole momentsofO2.[3] Molecular dynamicssimulations of suitable application is the evaluation of accuracyofthe approximate PES oxygen dynamics and migration in trHbN of Mycobacterium tuberculo- and couplings in comparison with high levelPES without the need to run sis arestudied with different approaches. Migration and exit pathways neitherquantum nor classicaldynamicsonthe expensivehigh levelPES. in multiple extended trajectoriesare analyzedindetail.[3] Thisprovides Both methods canbeconsideredageneralization of the dephasing repre- structural propertiesabout the ligand network whichisfound to be sim- sentation (DR) [2-3] of quantum fidelitytoseveral PES and their compu- ilar to NO network. Apreliminary connectivitynetwork for the ligand tational cost is the cost of dynamicsofaclassicalphasespacedistribu- docking sites is built andthe free energy profilesofmigration are esti- tion. It canbeimplemented easily into anymolecular dynamicsprogram matedbyumbrella sampling.Differencesbetween NO andO2 network and alsocan utilizeon-the-fly ab initio electronic structure information. arise from the freeenergy profileswhichare more complex for O2.Ener- We test the methods on threemodel problems introduced by Tully,on getic propertiestogether with astatistical analysis likeMarkov network the photodissociation of NaI, and on-the-flyusing CASSCFand MRCI [4] suggest thattrHbN network is designed for adynamical storage of O2. methods. [1] Hausladen, A., Gow, A., Stamler, J. S. Proc.Natl. Acad. Sci. [1] T. Zimmermann and J. Van´ıˇcek, The Journal of Chemical U.S.A., 2001, 98,1010810112. Physics, 2010, 132,241101; T. Zimmermann and J. Van´ıˇcek, [2] Mishra, S., Meuwly,M., J. Am.Chem. Soc., 2010, 132,2968. in preparation. [3] Mishra, S., Meuwly,M., Biophysical J., 2009, 96,2105. [2] J. Van´ıˇcek, Physical Review E, 2004, 70,055201(R). [4] Mishra, S., Meuwly,M., Biophysical J., 2010, 99,3969. [3] J. Van´ıˇcek, Physical Review E, 2006, 73,046204.

ComputationalComputationalC Chemistryhemistry CC 3 ComputationalComputational Che Chemistrymistry CC 4 TRAJECTORY-BASEDSOLUTIONOFTHE NONADIABATIC Alignator – a computer program for induced-fit docking: QUANTUM DYNAMICSEQUATIONS :ANON-THE-FLY Validation on the nuclear receptors APPROACH FOR MOLECULAR DYNAMICSSIMULATIONS Martin Smiesko Basile F. E. Curchod,Ivano Tavernelli,UrsulaRothlisberger Molecular Modeling, Department of Pharmaceutical Sciences, Laboratory of ComputationalChemistry and Biochemistry, Ecole University of Basel, Klingelbergstrasse 50, CH-4056 Basel, Switzerland Polytechnique Fédérale de Lausanne. Alarge majority of current techniquesemployedinthe rationaldrug Thenon-relativistic quantum dynamicsofnucleiand electrons is solved design rely on obtaining a3Dalignmentofasetofcompounds,which within theframework of quantum hydrodynamicsusing theadiabatic repre- wouldmirror theirrelativearrangement within target-proteinbinding site. sentation of theelectronicstates. An on-the-fly trajectory-based nonadia- Correlatingbiologicalactivitywiththe structuralpropertiesthenhelps with batic moleculardynamicsalgorithmisderived,which is also able to capture nuclear quantum effectsthatare missing in thetraditionaltrajectory surface explaining theobservedbehavior(strong, weak or not- binding) andnavig- hopping approach basedon theindependenttrajectory approximation. The ates research towards compounds having more favorable properties. useofcorrelated trajectoriesproduces quantum dynamics, which is in prin- Computer-assistedmoleculardocking is themostpopular technique cipleexact and computationally very efficient. foridentifying binding poses. It is aremarkablycomplex procedure,be- Themethod, called NABDY (NonAdiabatic Bohmian DYnamics),isfirst cause bothligandand proteinmay change theirconformationupon binding. tested on aseriesofmodelpotentials and then applied to study themolecu- In ordertodecrease complexity andthus save computationaltime in screen-

larcollision of Hwith H2 using on-the-fly TDDFTpotentialenergy surfaces ing, the protein is frequently treated as a rigid body. This might lead to satis- and nonadiabatic coupling vectors[1]. factory resultsfor docking of congenericseries, but rather to afailure,ifthe dockedligandsubstantiallydiffers from thetemplateinterms of shapeor volume. Keeping protein structure rigid simply doesnot allowfor the proper accommodation of ligand in the binding site. Alignatorisacomputer programfor flexible docking of compounds basedonthe pharmacophore alignment. It generatesligandposes devoidof steric bumps andallows for thelocal induced-fitchangesofthe protein structure basedonacombinatorial scan of theallowedside-chainconforma- tions in thebinding site.Key programfeaturesaswellasvalidationresults obtained at docking ligands to various nuclear receptors are presented.

[1]B.F.E.Curchod, I. Tavernelli,U.Rothlisberger, Phys.Chem. Chem. Phys., 2011, 13,3231. 508 CHIMIA 2011, 65, No. 7/8 Computational Chemistry

Computational Chemistry ComputationalChemistry Computational Chemistry CC 5 Computational Chemistry CC 6

Aquantum chemical data baseapproachfor method Moleculardocking usingthe MolecularLipophilicityPotential as hy- development anddata processing drophobicdescriptor:impactonthe GOLD dockingperformance

Andreas Gl¨oß,Martin P. Br¨andle, Hans P. L¨uthi AlessandraNurisso,Pierre-Alain Carrupt,Antoine Daina

ETHZ¨u rich, Wolfgang-Pauli-Str.10, CH-8093 Z¨urich, Switzerland SchoolofPharmaceutical science, UniversityofGeneva, UniversityofLau- sanne Quai Ernest-Ansermet 30, CH-1211,Geneva4,Switzerland The computational poweravailable to quantum chemists overthe past decadeshas grown enormously,makingincreasingly complex systems fea- GOLD is reliabledocking software well known in thefield of drug design sible foraccuratecomputation.Since the generationofthe data (results) [1]. At earlysteps of docking, it creates alistof hydrophobic fittingpoints is no longerthe bottleneck, the challenge nowistofind strategiesfor inside theprotein cavitiesthatsteers thepositioning of ligandhydrophobic structured storage,extraction (analysis)and interchange of data,thereby moieties.The generationofhydrophobicfittingpointsisbased on theLen- keeping thelevel of human intervention at aminimum. nard Jonespotentialbetween acarbon probe andeach atom of theresidues delimitatingthe binding site. To thoroughlydescribehydrophobicregions in Herewereportonthe implementation of aworkflowscheme using TUR- proteinpockets andproperlyguide ligand hydrophobicmoietiestowards BOMOLE [1]asprogram package, CML [2] as interchangeableoutput favorable areas,anin-house tool,the MLP-filter,was developedand ap- format as well as FoXlib [3] and eXist [4] to write and store the quantum plied. Thisnew strategyretains onlyGOLDhydrophobicfittingpoints chemical data.Extensions to CMLwill be presented, which nowallow matching therigorous definitionofhydrophobicity givenbythe MLP, Mo- the use of the markup language as standardinterchangeand storage for- lecularLipophilicityPotential, afragmentalmolecular interactionfield that mat forquantum chemistrycodes in general. To illustrate the benefit of reliesonexperimentally determined n-octanol/water partitioncoefficients structureddata processing, we present the results of an extensivestudy (logPo/w)[2].MLP computationinside binding sites of proteincrystalstruc- of the relativestabilityofthe conformersofthe ethene dimer using re- turesfromthe AstexDiversedataset andthe PDBbinddatabaserevealed that asignificantnumberofpoints, consideredashydrophobicbyGOLD, cently developedMP2-F12 methods[5]. were actuallypolar.Toexamine theimpact of our newtool,re-docking ex- [1] TURBOMOLE V6.3, http://www.turbomole.com. perimentswereperformedwith andwithout theuse of MLP-filter.Reliable [2] P. Murray-Rust andH.S.Rzepa, J. Chem. Inf. Comput. Sci., docking results were obtained by using our in-house tool that,moreover, 2003, 43,757. increased thequalityofdocking results in cases of non polar cavities, out- performingthe standard GOLD docking approach. [3] FoXlib V4.1.0, http://www1.gly.bris.ac.uk/walker/FoX. [4] eXist-db V1.4.0, http://exist-db.org. [1]HartshornM.J., Verdonk M. L.,Chessari G.,Brewerton S. C.,MooijW. [5] R. A. Bachorz, F. A. Bischoff, A. Gl¨oß, C. H¨attig, S. H¨ofener, T. M.,Mortenson P. N.,MurrayC.W., J. Med. Chem, 2007, 50, 726. W. Klopper, and D. P. Tew, J. Comput. Chem., 2011,(ac- [2]Gaillard,P., Carrupt,P.A., Testa,B., Boudon,A., J. Comp.Aid. Mol. De- cepted) sign, 1994, 8,83.

Computational Chemistry ComputationalComputational Chem Chemistryistry CC 7 Computational Chemistry CC 8

How to thermostat coupled heat and mass transport in MD simulations Multi-scale modelling of solvatochromic shiftsbased on and other challenges from the world of foods frozen-densityembedding theory

Hans Jörg Limbach XiuwenZhou1,JakubW.Kaminski2,Tomasz A. Wesolowski1�

Microprocess and Modelling Group, Food Science & Technology 1University of Geneva, DepartmentofPhysical Chemistry,Quai Department, Nestlé Research Center, Vers-chez-les-Blanc, Ernest-Ansermet 30, CH-1211 Geneva, Switzerland CH-1000 Lausanne, Switzerland 2DivisionofChemistry andChemical Engineering, California Institute of Technology,Pasadena, CA 91125, USA

The solvatochromic shiftoflowest absorption band of coumarin153 in The kinetics of phase transitions is an important part of material science in solvents of various polarityare evaluated using the frozen-densityembed- the food field. It is well known that phase transitions involve a coupled heat ding theory based methods [1]. Instead of averaging the shiftoversta- and mass transfer problem. In recent years several methods, both equili- tistical ensemble, the averageelectron densityofthe solvent, <ρB(�r ) > brium and non-equilibrium, have been proposed to tackle this problem. A is used to evaluate the averageeffect of theenvironmentonthe excita- critical discussion of these methods will be given. As all proposed methods tion energyofthe chromophore. In the calculation, <ρB(�r ) > is used show an inherent system size dependency we have developed a new non- as frozen-density, whichisevaluated using the statistical-mechanical ap- equilibrium method. With this we probe the kinetic growth limit via imple- proach[2]. The smalldeviations between calculated andexperimental mentation of a modification of the stochastic rescaling thermostat, where the solvatochromic shifts confirm the approximant for the bi-functional of system is sliced into independent temperature coupling groups. We prove the non-electrostatic component of theorbital-free embedding potential that this approach effectively removes latent heat locally, yielding a reliable is adequate for chromophores interactedwith the environmentbynon- estimate of the maximum crystallization/melting rate. This set-up is also covalentbonds.The qualitativeanalyses of theoriginofsolvatochromic used to investigate the effect of additives on the crystallization process as a shifts are made using the graphical representation of theorbital-freeem- function of their size and concentration. bedding potential [3].

Thesecondgroup of results concerning absorptionspectraoffluorenone in Zeolite Lwill alsobepresented [4].

[1] T. A. Wesolowski andA.Warshel, J. Phys. Chem.,1993, 97, 8050. [2] J. W. Kaminski, S. Gusarov, T. A. Wesolowski andA.Ko- valenko, J. Phys.Chem. A, 2010, 114,6082. [3] X. Zhou,J.W.Kaminski,T. A. Wesolowski, Phys.Chem. Chem.Phys.,DOI:10.1039/C0CP02874F(in press). [4] X. Zhou, et al,inpreparation. Computational Chemistry CHIMIA 2011, 65, No. 7/8 509

ComputationalComputationalC Chemistryhemistry CC 9 ComputationalComputationalC Chemistryhemistry CC 10

Electrontransport across a corannulene-SWCNTjunction Computaional methods for finding the ideal nitrogen fixation catalyst

1 L. Zoppi , A. Ferretti2, L. Martin-Samos3,K.K.Baldridge1 S. M. A. Donald, M. Reiher*

1University of Zürich OrganicChemistryInstitute,Winterthurerstrasse 190, Laboratory of Physical Chemistry, ETH Zürich, Wolfgang-Pauli-Strasse 10, 8057 Zürich,Switzerland CH-8093, Zürich, Switzerland 2 S3 Center,Istituto Nanoscienze, CNRvia Campi213/A, 41100, Modena,(Italy) Nature successfully fixes dinitrogen via the nitrogenase enzymes, but at- 3CNR-IOMDEMOCRITOSand SISSA,via Bonomea 265, tempts to mimic nature, creating asynthetic nitrogen fixation cycle, have 34136 Trieste(Italy) proven challenging. The industrial Haber-Bosch process involves both high pressures and temperatures [1], but two catalyst systems have recently been reported, by Schrock and then by Nishibayashi et al., which convert dinitro- In thepresent work we evaluate thequantum conductancefromfirst princi- gen to ammonia at ambient temperature and pressure [2]. These catalysts ples in molecularnanojunctions based on corannulene-based molecules are, however, destroyed under reaction conditions after alimited turnover. linked to two SingleWalled Carbon Nanotubes(SWCNT’s) electrodes. The Theory has provided fundamental insight into the Schrock catalyst[3], and interest in theseparticular moleculesstems fromrecentexperimental find- we now report progress [4] in the search for new catalyst systems that can ings1 showing thepossibility of tuning theiroptoelectronicpropertiesby fix dinitrogen as well as Schrock catalyst, but that shows improved stability. i - modifying thestructure at thenanoscalein acontrollableway.In orderto The [Si(o-C6H4-P Pr3)3] (SiP3) ligand system [5] is described with density explorethe effect of such tuning on thequantum conductance, we have cho- functional theory. We detail the performance of [Fe(SiP3)] and hypothetical sen four representativesubstituted molecularsystems to be inserted between [Mo(SiP3)] catalysts in the key steps of the Schrock-cycle, comparing them theelectrodes. To calculate thecoherent transportpropertiesofthe nano- to the Schrock catalyst itself. We also look in greater detail at the NH3/N2 junctions,wefollowthe Landauerapproach2 that relatesthe quantumcon- ligand exchange process and perform afull analysis of the possible path- ductanceofthe conductorto thescattering propertiesofthe system through ways to determine the barrier to this crucial step in the catalytic cycle. 2e2 theexpression: C(") = T("). h [1] J. R. Postgate, Nitrogen Fixation, Cambridge University Press, Cam- Thetransportcharacteristic arethen obtained through areal space imple- bridge, 1998. mentation of theLandauerformula in termsofmaximally localized Wannier [2] (a) Yandulov, D. V.; Schrock, R. R. Science 2003, 301, 76; (b) Ara- functions using theWANT code2. shiba, K.; Miyake, Y.; Nishibayashi, Y. Nat. Chem., 2011, 3, 120. [3] (a) Schenk, S.; Kirchner, B.; Reiher, M. Chem. Eur. J. 2009, 15, 5073; [1]Wu, Y. T.;Bandera,D.; Maag,R.; Linden, A.;Baldridge,K.K.; Siegel, (b) Schenk, S.; Reiher, M. Inorg. Chem. 2009, 48, 1638; (c) Schenk, S.; J. S. J. Am.Chem. Soc. 2008, 120, 1-4 Le Guennic, B.; Kirchner, B.; Reiher, M. Inorg. Chem. 2008, 47, 3634. [2]Calzolari,A.; Marzari, N.;Souza, I.;Nardelli, M.,Phys.Rev.B. 69 [4] Donald, S. M. A.; Reiher, M. To be submitted. 2004,035108 [5] Whited, M. T.; Mankad, N. P.; Lee, Y. H.; Oblad, P. F.; Peters, J. C. Inorg. Chem. 2009, 48, 2507.

ComputationalComputational Ch Chemistryemistry CC 11 ComputationalComputational ChChemistryemistry CC 12

Oxygen and Proton Reduction by Metallocenes in Non-Aqueous Media Multiregional Localized Molecular Orbitals

Tanya K. Todorova, Loïc Roch, Clemence Corminboeuf Piotr de Silva1,2, Jacek Korchowiec2

Laboratory for Computational Molecular Design, Ecole Polytechnique 1Université de Genève, Département de Chimie Physique, Fédérale de Lausanne, CH-1015 Lausanne, Switzerland 30, Quai Ernest-Ansermet, C-1211 Genève 4, Switzerland 2K. Gumiński Department of Theoretical Chemistry,Faculty of Chemistry, Jagiellonian University, R. Ingardena 3, 30-060 Cracow, Poland Recent electrochemical measurements at liquid-liquid interface [1] have shown that decamethylferrocene is capable of reducing aqueous protons leading to the evolution of hydrogen, regarded as an ideal candidate for Localized molecular orbitals playan important role in computational chem- energy supply in a sustainable manner. Moreover, decamethylferrocene can istry for two reasons. They provide aconceptual link between quantum me- reduce molecular oxygen to hydrogen peroxide, which is highly desirable chanical wave function and so-called chemical intuition referringto bonds, compound for fuel cell applications.How changing the metal affects these core electrons and lonepairs. Apart from this conceptual contribution, important reactions? Is the propensity towards proton and O2 reductionsig- LMOs proved to be very useful in reduced scaling computational methods. nificantly increased or diminished?Density functional theory computations In this paper we present a multi-fragmentaryextension of regional localized have been performed to shed light on the reduction of aqueous protons to molecular orbitals (RLMO) [1]. A sequenceofunitarytransformation ena- hydrogen under anaerobic conditions, and the reduction of oxygen to hydro- bles to localize orbitals on apredefined set of molecular fragments. Onlythe gen peroxide under aerobic conditions by osmocene and ruthenocene[2]. one-particledensitymatrix and the overlap matrix areneeded in the compu- Characterization of the active species is performed in order to elucidate the tational scheme. This approach to localization is expected to be very effi- reaction mechanismsthat govern the proton and oxygen reduction reactions. cient in linear-scaling fragmentation methods. Rationalization of the experimentally observed trends [3]isestablished pro- viding a step towards understanding the structure-reactivity relationship for the metallocene catalytic systems. [1]F. L. Gu, Y. Aoki, J. Korchowiec, A. Imammura,B.Kirtman, J. Chem. Phys. 2004, 121, 10385. [1] B. Su, I. Hatay, P.Y. Ge, M. Mendez, C. Corminboeuf, Z. Samec, M. Ersoz, H.H. Girault, JACS, 2010, 46, 2918. [2] P.Y. Ge, I. Hatay, O. Astrid, Y. Lu, M.A. Mendez, H. Vrubel, X. Hu, T.K. Todorova, C. Corminboeuf, H.H. Girault, in preparation. [3] P.Y. Ge, I. Hatay, H.H. Girault, in preparation. 510 CHIMIA 2011, 65, No. 7/8 Computational Chemistry

ComputationalComputationalC Chemistryhemistry CC 13 ComputationalComputational Che Chemistrymistry CC 14

LpxC: insights into substratebinding andselectivity Theoretical study of NCO and COS hydrolysis over Al2O3 catalyst

MarcoStenta, Christophe Bovigny, Matteo DalPeraro Izabela Czekaj, Oliver Kröcher, Rudolf Struis

Laboratory forBiomolecularModeling, IBI-SV,EPFL Department General Energy, Paul Scherrer Institute CH-1015 Lausanne,Switzerland CH-5232 Villigen PSI, Switzerland

LipidAconstitutesthe core of thecomplex lipopolysaccharide(LPS) mole- Hydrolyses are an important class of reactions in nature, which not only culesdominatingthe lipid compositionofthe external leaflet of Gram- occur in liquid phase but also in heterogeneous gas phase reactions over negativebacteriaouter membrane [1]. 2+ solid catalysts. Two systems, in which catalytic hydrolyses are involved are LpxC (UDP-(3-O-acyl)-GlcNAc deacetylase),aZn -dependent amidase, the hydrolysis of isocyanic acid to ammonia in the selective catalytic reduc- catalyzes thefirst committedstepalong lipid Abiosynthetic pathway. Since tion (SCR) of nitrogen oxides [1] and the hydrolysis of carbonyl sulphide, its validationasanantibacterial drug target,several compounds have been which results in the poisoning of methanation catalysts [2]. Both reactions reportedtoeffectivelyinhibit LpxC activityand thus stop lipid Asynthesis can proceed over -Al2O3 as catalytic material. We tried to fully understand [2][3]. the mechanism of hydrolysis for both species by means of Density Func- Amodelofthe complexbetween LpxC andits naturalsubstrate wasrecon- tional Theory calculations (StoBe) with cluster model and non-local func- structedonthe basis of theavailableX-ray andNMR structures. Molecular tional (RPBE) approach. For modelling the -Al O catalyst an Al O H simulations were used to quantifyspecies-specificand inhibitor-dependent 2 3 15 40 35 cluster has been selected, which represents the (100) surface. Then the LpxC conformationalflexibility andtorationalize theenzyme’scatalytic mechanismof the hydrolysis of isocyanic acid (HNCO) and carbonyl sul- proficiency. phide (COS) modeled on these clusters. Additionally, the theoretical studies have been compared with experimentaldata from DRIFTS and XAS inves- tigations. In the most probable pathway,awater molecule attacks the –NCO and –COSgroups, thereby forming carbamic acid and thiocarbonic acid, respectively, at the surface. In a further step those compounds are trans- formed to carbamate and thiocarbonic complexes, respectively.Thisleads finally to CO2 desorption and NH3 formation in case of isocyanic acid and to H2S formation from carbonyl sulphide.

References [1] I. Czekaj, O. Kröcher, TopicsCatal. 52 (2009) 1740. [1]X.Wang, et al. Progress in LipidResearch 2010, 49, 97-107. [2] I. Czekaj, R. Struis, J. Wambach, S. Biollaz, Catal. Today in press. [1]H.R.Onishi, et al. Science 1996, 274,980-982. [3]X.Liang, et al. Bioorg. Med. Chem. 2011,19, 852-860.

ComputationalComputational Chem Chemistryistry CC 15 ComputationalComputational Ch Chemistryemistry CC 16

NixPdy nanoparticles stability at the γ-Al2O3 support: DFT studies Dephasing representation: defeating the efficiency of both quantum and classical simulations with semiclassics [1] Izabela Czekaj, Jörg Wambach Cesare Mollica, Jiří Vaníček* Department General Energy, Paul Scherrer Institute CH-5232 Villigen PSI, Switzerland Laboratory of Theoretical Physical Chemistry, Institut des Sciences et Ingé- nierie Chimiques, École Polytechnique Fédérale de Lausanne, CH-1015, Recently, surface modifications on a commercial Ni/Al2O3 catalyst during Switzerland the production of methane from synthesis gas were investigated by quasi in- situ X-ray photoelectron spectroscopy (XPS) [1, 2]. In the present work we extended our DFT investigations to nickel-palladium nanoparticles, which Nowadays rigorous quantum mechanical (QM) calculations for many- has been also investigated also experimentally. particle systems remain out of reach, while classical (CL) simulations, alt- The Ni and Pd deposition and cluster growth phenomena on our home- hough feasible, do not include QM effectsat all. Semiclassical (SC) approx- made model catalyst samples (10 nm thick, polycrystalline -Al2O3 on imations take into account allQM effects through interferences of CL tra- Si(100)) were investigated experimentally by XPS and the molecular struc- jectories and, unlike full QM methods, arenumericallyaffordable for com- ture of the catalyst was investigated using Density Functional Theory calcu- plex systems even though challengingproblems severely affects their nu- lations (StoBe) with cluster model and non-local functional (RPBE) ap- merical efficiency. We will describe an accurate SC approximation, called proach. For the latter, Al15O40H35 clusters have been selected representing “dephasingrepresentation” [2,3,4,5], which surprisingly is even much faster the -Al2O3 (100) surface. Metal particles of different sizes were cut from a than the CL calculation. (100) surface and deposited on the Al15O40H35 cluster in order to validate the deposition model determined by XPS. Ni or Pd deposition on -Al2O3 follows a “modified” Stranski-Krastanov [1]C. Mollica, J. Vaníček, arXiv: quant-ph 2011. growth mode under the applied experimental conditions. The DFT results [2]J. Vaníček, Phys. Rev. E 2004, 70, 055201(R). correspond well with the experimental data of the initial stage of metal [3]J. Vaníček, Phys. Rev. E 2006, 73, 046204. deposition, where the formation of a partial metal monolayer is suggested. [4] B. Li, C. Mollica, J. Vaníček, J. Chem. Phys. 2009, 131, 041101. [5]T. Zimmermann, J. Ruppen, B. Li and J. Vaníček , Int. J. Quantum References Chem. 2010,110(13), 2426–2435. [1] I. Czekaj, F. Loviat, F. Raimondi, J. Wambach, S. Biollaz, A. Wokaun; Appl. Catalysis A: General 329 (2007) 68-78. [2] F. Loviat, I. Czekaj, J. Wambach, A. Wokaun; Surf. Sci. 603 (2009) *[email protected] 2210-2217. Computational Chemistry CHIMIA 2011, 65, No. 7/8 511

ComputationalComputational ChemChemistryistry CC 17 ComputationalComputational ChemChemistryistry CC 18

Efficient use of accessibility in microRNA target prediction Split-CAS and Direct-CI: A Combined Strategy to Enlarge the Orbital Space in Multiconfigurational Calculations Ray Marín, Jiri Vaníček Dongxia Ma1, Francesco Aquilante2, Laura Gagliardi1, Giovanni Li Manni Laboratory of Theoretical Physical Chemistry, Institut des Sciences et Ingénierie Chimiques, École Polytechnique Fédérale de Lausanne, CH-1015 1Department of Chemistry, University of Minnesota, 207 Pleasant St SE, Lausanne, Switzerland MN-55455 Minneapolis, United States 2Department of Physical and Analytical Chemistry, Uppsala University, 1, Considering accessibility of the 3'UTR is believed to increase the precision Lägerhyddsvägen, SE-75120 Uppsala, Sweden of microRNA target predictions. We show that, contrary to common belief, 3Department of Physical Chemistry, University of Geneva, 30, Quai Ernest ranking by the hybridization energy or by the sum of the opening and hybri- Ansermet, CH-1211 Geneva, Switzerland dization energies, used in currently available algorithms, is not the most efficient way to rank predictions. Instead, we describe an algorithm which Complete active space wave functions are constructed by variationally op- also considers only the accessible binding sites but which ranks predictions timizing the expectation value of the energy with respect to the orbital and according to over-representation of the accessible seed matches [1]. When the Configuration Interaction parameters. The major limitation arises from compared with experimentally validated and refuted targets in the fruit fly the factorial scaling of the CI expansion with respect to the size of the active and human, our algorithm shows a remarkable improvement in precision space. The associated system of secular equation is handled by simplified while significantly reducing the computational cost in comparison with oth- diagonalization techniques, e.g. Davidson algorithm. These technique rapid- er free energy based methods. In the human genome, our algorithm has at ly reach their limit of applicability in terms of memory requirement. The least twice higher precision than other methods with their default parame- aim of our research is to reduce the computational costs of big CASSCF ters. In the fruit fly, we find five times more validated targets among the top calculations, by using the SplitCAS strategy [1]. Instead of reducing the size five hundred predictions than other methods with their default parameters. of the CI expansion, we retain the CAS ansatz, but separate it in two parts: a The proposed method also allows combining the accessibility filter with a principal space (A) and an extended space (B), dim(A)<

Computational Chemistry Computational Chemistry CC 19 ComputationalComputational Chem Chemistryistry CC 20

Can Raman Optical ActivityDiscriminate Between Different Thermodynamics of the monomerization of a Re(V) dithiolato dimer TypesofProtein β-Turns? with imidazole-based ligands: A theoretical investigation

1 2,� 1,� Thomas Weymuth ,Christoph R. Jacob and Markus Reiher Fatemeh Niroomand Hosseini*a, S. Masoud Nabavizadehb

1 ETH Zurich, Wolfgang-Pauli-Str. 10, CH-8093 Zurich, Switzerland a 2 Department of Chemistry, Islamic Azad University, Shiraz Branch, Shiraz, Karlsruhe Institute of Technology (KIT), Wolfgang-Gaede-Str. 1a, 71993-37635, Iran DE-76131 Karlsruhe, Germany b Department of Chemistry, College of Sciences, Shiraz University, 71454, Shiraz, Iran. On the basis of density-functional calculations, vibrational Raman op- tical activity(ROA) signatures are established whichcan be used to The thermodynamics of the monomerization of a Re(V) dithiolato identify β-turns, an importantsecondary structure elementinproteins. dimer, {MeReO(edt)}2 (edt = 1,2-ethanedithiolate) with imidazole-based Several reliable signatures were identified [1] and related to other signa- ligands (imidazole, 1-methylimidazole, 1-ethylimidazole, indazole, ben- tures proposed in the literature. zimidazle and pyrazole) were investigated by DFT calculations in gas phase, In addition, the possibilityofpredicting the extended amide III frequen- benzene and acetone solution. cies of oligopeptides suchas β-turns using asimple model is investigated. O Me S O The extended amide III region in vibrational spectra is particularly sen- Re S Me S sitivetochanges in secondary structure. To investigate this sensitivity, S + 2 imidazoles (L) 2 Re we have performeddensity-functional calculations on the model system Re S L Me S N -acetyl-l-alanine-N -methylamide, whichhave been analyzed using the O concept of localized modes [2].Westudy the dependence of the local- The effective core potential of Hay and Wadt with a double-ξ va- mode frequencies and coupling constants on the torsional angles φ and lence basis set (LANL2DZ) was chosen to describe Re. The 6-311G* basis ψ.This enables us to set up alocal-modemodel for abetter understand- set was used for other atoms. Geometries of complexes were fully optimized ing of the structural sensitivityofthe extended amide IIIregion [3]. and to evaluate and ensure the optimized structures of the molecules, fre- All vibrational spectra have been calculated with the program Snf [4] quency calculations were carried out using analytical second derivatives. employing amodified version of Turbomole whichallows for compu- From thermochemistry results, the values of H  0 and S  0 for tationally demanding ROA calculations on large model systems [5]. monomerization reactions were obtained which means the reactions are en- thalpy-driven. The overall monomerization processes are thermodynami- [1] T. Weymuth, C. R. Jacob, M. Reiher, ChemPhysChem 2011, cally more favorable for acetone and benzene solvent as compared with gas 12,1165. phase, and with increasing the basicity of ligands the free energy of the re- [2] C. R. Jacob, M. Reiher, J. Chem. Phys. 2009, 130,084106. actions is decreased. [3] T. Weymuth, C. R. Jacob, M. Reiher, J. Phys. Chem. B 2010, 114,10649. [1] J. H. Espenson, X. Shan, Y. Wang, R. Huang, D.W. Lahti, J. Dixon, G. [4] J. Neugebauer, M. Reiher, C. Kind, B. A. Hess, J. Com- Lente, A. Ellern, I.A. Guzei, Inorg. Chem. 2002, 41, 2583. put. Chem. 2002, 23,895. [2] J. H. Espenson, S. M. Nabavizadeh, Eur. J. Inorg. Chem. 2003, 1911. [5] S. Luber, M. Reiher, Chem. Phys. 2008, 346,212. 512 CHIMIA 2011, 65, No. 7/8 Computational Chemistry

Computational Chemistry ComputationalComputational ChChemistryemistry CC 21 Computational Chemistry CC 22

Accelerating calculations of ultrafast time-resolved electronic spectra Semiclassical calculationofultrafast time-resolved electronic with various high order split-operator algorithms spectra

M. Wehrle, J. Vaníček * M. Suˇ lc,J.Van´ı ˇc ek∗

Laboratory of Theoretical Physical Chemistry, Institut des Sciences et Ingé- 1Laboratory of Theoretical Physical Chemistry,Institut des Scienceset nierie Chimiques, École Polytechnique Fédérale de Lausanne, CH-1015 Ing´e nierie Chimiques, Ec´ ole Polytechnique F´ed´e rale de Lausanne Lausanne, Switzerland (EPFL), CH-1015Lausanne, Switzerland We investigate[1] the applicabilityofselected semiclassicalmethods based on the initial value representation [2] in combination with sym- Calculations of ultrafast time-resolved electronic spectra require performing plectic integrators for the computationofultrafasttime-resolved elec- the quantum dynamics, i.e., solving the time-dependent Schrodinger equa- tronic spectra. Particular interest is devoted to the theoretical proper- tion. We explore how these demandingcalculations can be accelerated by ties as well as numericalbehaviourof Frozen Gaussians Approximation increasingthe time step required for convergenceofexact quantum simula- (FGA) andthe Heller-Hermann-Kluk-Kay (HHKK) propagator, which tions withoutsacrificing the accuracy [1]. For this purpose, we compared st have been shown to be applicable even in high dimensional systems [3] as various split-operator methods [2]ofthe 1 to 4thorder. Besides theusual 4th order algorithm with real coefficients we also considered an algorithm opposed to standard quantum methods. Several applicationsinthe field with complex coefficients and an algorithm using the gradient of the poten- of spectroscopy [4] have been also reported.Onparticular examples, it tial. Our results show that while the last two methods require fewer Fast is found that even the rather simple FGAcan givereasonable prediction Fourier Transforms at each step, the complex method divergesfor dense of theelectronic spectra. It turns alsoout that the HHKK approach is spatial grids and the gradient method's order decreases to the 2nd order for numerically slightly hindered in the presence of chaotic behaviour; there- sparse grids [1]. fore we discuss possible improvements to this methodsuchastrajectory filtering, modified sampling of the initial conditions and time averaging techniques [5]. Finally,comparisons are made with the cellularized ver- [1]M. Wehrle, M. Šulc, and J. Vaníček, Chimia 2011. sions of the FGAand HHKK algorithms, withpossible generalizations[6] [2]a) A. D. Bandrauk; E. Dehghanian; H. Lu. Chem.Phys. Lett. 2006, 419, taken into account. 346; b) M. D. Feit; J. A. Fleck; A. Steiger. J. Comput. Phys. 1982, 47, [1] M. Wehrle,M. Sulc,ˇ and J. Van´ıˇcek, Chimia 2011. 412; c) E. Forest; R. D. Ruth. Physica D 1990, 43, 105; d) S. A. Chin. [2]K.G.Kay, Annu. Rev. Phys. Chem. 2005, 56,255. Phys. Lett. A 1997, 226, 344; e) T. Prosen; I. Pižorn. J. Phys. A: Math. [3] M. L. Brewer, J. S. Hulme, and D. E. Manolopoulos, J. Chem.Phys. Gen. 2006, 39, 5957; f) M. Suzuki. Phys. Lett. A 1995, 201, 425; g) H. 1997, 106,4832. Yoshida. Phys. Lett. A 1990, 150, 262 [4] M. Ovchinnikov,V.A.Apkarian, andG.A.Voth, J. Chem.Phys. 2001, 114,7130. *[email protected] [5] Y. Elran and K. G. Kay, J. Chem.Phys. 1999, 110,3653. [6] H. Wang, D. E. Manolopoulos,and W. H. Miller, J. Chem. Phys. 2001, 115,6317. ∗[email protected] ComputationalComputationalC Chemistryhemistry CC 23 ComputationalComputationalC Chemistryhemistry CC 24

Four-ElectronReduction of O2 by Tetrathiafulvalene Efficiency of Stochastic andGeneticAlgorithmsfor Characterizing MoleculeswithUniqueBonding Patterns J. F. Gonthier,C.Corminboeuf FabriceAvaltroni andClémenceCorminboeuf Institute of Chemical Sciences andEngineering,Ecole Polytechnique Institute of Chemical Sciences andEngineering,Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne,Switzerland Fédérale de Lausanne, 1015 Lausanne,Switzerland

The catalytic four electronreductionofO2 into waterisamandatory step to thedevelopmentofefficientfuelcells.[1] The most widely usedcatalysts The computationalquest formolecularcandidateschallengingconventional are basedonplatinum,but thesearch foralternativebiomimetic solutions chemical bonding paradigms(e.g. molecularwheels, multicenterbonding) circumventingthe use of precious metalhas been very active. Preliminary hasattractedagreat deal of attentionoverthe last twodecades.The viability experimentssuggestthatthe excellent electrondonorpropertiesof of such systemsisnecessarily assessedthrough thecharacterizationonthe Tetrathiafulvalene(TTF) alongwith its abilitytoform assembled potentialenergysurface (PES)ofthe lowest energy form of thegiven π-π-stacked structurescouldlead to thefour-electronreductionreaction chemical composition. Whereas dozensofsearch algorithms have been fromoxygentowater in acidicmedium.[2] developedfor this purpose,onlyafeware generaland simpleenough to become standard everyday procedures.The simplerandom search [1]and The presentDFT studyinvestigatesthe four-electronreductionreaction Genetic Algorithms (GA)[2] are among these: but how do theseapproaches pathwayofoxygenbyTTF andidentifiesasetofassembledTTF structures compare on typical isomericsearches? We compared theperformanceofthe that exhibitavarietyofintermolecularinteractions by π-π stacking, weak twoapproaches forthe explorationofthe PESofprototype planar carbon- 2- hydrogen bonding,and sulfur-sulfurinteractions.Toaccount forthe weak containing systems CB6 [3]and C2Al4 [4]. Wealsodiscuss thepotential dispersiveforces caracteristicofthese systems,standard functionals are advantages of using alow-cost semi-empircal method (e.g.PM6)inthe pre- combined with adensity-dependent dispersioncorrectiondevelopedinour optimizationstepinstead of B3LYP/3-21G. laboratory.[3]Geometrieswereoptimized with theM06-2X functional designedand fittedtoreproduceweak interactions. [1]Saunders,M. JComput Chem 2004, 25, 621-626. [2]Deaven,D.M.; Tit, N.;Morris, J.;Ho, K. Chem PhysLett 1996, 256, 195-200. [3]Exner, K.;Schleyer, P. v. R. Science 2000, 290,1937-1940. [1]J.P.Collman, P. S. Wagenknecht, J. E. Hutchison, Angew.Chem. Int. [4]Wu, Y.-B.; Lu,H.-G.;Li, S.-D.;Wang, Z.-X. JPhysChemA2009, Ed. 1994, 33, 1537. 113,3395-3402. [2]A.J.Olaya,P.Ge, P. Pechy, H. H. Girault, privatecommunication [3]S.N.Steinmann, C. Corminboeuf, J.C.T.C. 2010, 6,1990 ;S.N. Steinmann, C. Corminboeuf, J. Chem.Phys. 2011, 134,044117 Computational Chemistry CHIMIA 2011, 65, No. 7/8 513

Computational Chemistry Computational Chemistry CC 25 ComputationalComputational ChChemistryemistry CC 26

Tree Monte Carlo Ab Initio-based Electron Transfer Through π-Stacked Nucleobases

Mandes Sch¨onherr,J¨u rg Hutter, and Joost VandeVondele Laura Berstis, Kim Baldridge

UniversityofZ¨u rich, Winterthurerstrasse 190, 8057 Zurich. University of Zürich Organic Chemistry Institute, Winterthurerstrasse 190, 8057, Zürich, Switzerland The developmentofhigh performance computers makes more and more computing poweravailable. To use this potential, novelalgorithms have to be adopted that displayincreasing potential for parallelization. In Afirst-principles methodology has been implemented into the ab inito com- the case of molecular simulations, the scalabilityofMolecular Dynam- putational suite GAMESS for calculating the electronic coupling between ics (MD) is ultimately limited, and the step by step integration essen- donor and acceptor states of small biomolecular systems involved in elec- tially serializes the procedure. Here, we presentaTree Monte Carlo tron transfer reactions. This method calculates the nonadiabatic electron (TMC) algorithm whichallows for calculating manymolecular configu- transfer reaction rate from the electronic coupling, which is based on the rations simultaneously.Using aspeculativeapproach,various pathways, calculation of the effective Hamiltonian from a Green’s function operator.1,2 accounting for both rejectance and acceptance of proposed moves, can The efficacy of this method and implementation are demonstrated on small be computed in parallel. Key to this approachisthe fact that gener- model hydrocarbon and peptide donor-acceptor systems. This methodology ating configurations for the Monte Carlo algorithm is cheap and canbe is then applied to the study of electron transfer properties through various π- performed without computing forces [1]. The TMC procedure respects stacked nucleic acid structures to investigate the relationship between nu- at least the sufficientbalance condition [2], and in fact the sequenceof cleobase sequence, structural orientation, and predicted charge transfer reac- used Monte Carlo movesisindependentofthe number of simultaneously tivity with high accuracy quantum mechanical methods. computed tasks. The parallel efficiency of the TMC algorithm depends on the acceptance probabilities of the movesaswell as the number of used computing units. The most importantaspect in Monte Carlo al- [1] Kurnikov, I.V., Beratan, D.N. J. Chem. Phys., 1996, 105, 21. gorithm is an efficientconfigurational change, whichshould be as huge [2] C. Kobayashi, K. Baldridge, J.N. Onuchic. J. Chem. Phys. 2003, 119, 6. as possible, but sufficiently small. Forcomputational expensivemodels, likeDFT, approximated potentials can be used to generate the configu- rational change. Further efficiency gains will come through dynamic process termination and acceptance estimation. Finally,the algorithm is well suited to be implemented in aFault Tolerantcontext, whichmightbecome essential on future architectures. [1] D. Frenkel and B. Smit, Understanding Molecular Simulation 2001, 638,ISBN-10:0-12-267351-4. [2] Manousiouthakis, Vasilios I. and Deem, Michael W. Strict detailed balance is unnecessary in Monte Carlo simulation 1999 Computational Chemistry ComputationalComputational Che Chemistrymistry CC 27 Computational Chemistry CC 28 Protein Structure Prediction Using Genetic Algorithms Automatic Generation and RefinementofPotential Energy Surfaces using IMLS Fits of differentorder Prashanth Athri, Ivano Tavernelli, Ursula Röthlisberger Moritz P. Haag,Konrad H. Marti, Markus Reiher EPF Lausanne, Route Cantonelle, 1015 Lausanne, Switzerland Swiss Federal Institute of Technology Zurich, Wolfgang-Pauli-Strasse Protein structure predictions, which aimto elucidate the complete confor- 10, CH-8093 Zrich, Switzerland mation of a protein from sequence information, remain to be a complex multivariateproblem in addition to being computationally intensive. Here The study of molecular reactivityrequires adetailed knowledge of the we presenta2-level genetic algorithm (GA) to effectively partition the potential energy surface (PES) of the molecular system [1]. We present search into backbone and side chain conformational searches. Additionally, anew methodfor the automatic generation of new surfaces and the this allows the utilization of structural statistical information for both do- refinementofexisting PES employing IMLS (Interpolating Moving Least mains in an additive or selective fashion in stages. Squares [2]) fits of differentorder [3]. The first level GA optimizes the Phi and Psi angle of the protein The methodisbased on finding the minimum of the negativesquared backbone. Each conformation (chromosome) of the backbone has an associ- difference between two surfaces generated by IMLS fits of differentorder ated (sub)optimal side chain conformation that is searched using the 2nd lev- as described by Dawesetal. [4]. The least defined point, whichshould be el GA. The search proceeds to find a near optimal solution to the prediction added next to improve the fit, is expected to givethe global minimum on problem and is then fine tuned using a molecular dynamics study. this difference surface. In contrast to thework of Dawesetal. [4], here, the searchfor the global minimum is performed applying asimulated annealing procedure, whichisvery efficientand therefore leads to afast convergence of the refinementprocess. The efficiency of this new algo- rithm is demonstrated at differentmodel cases. Potential applications in the area of molecular reacitivitystudies, especially in the framework of haptic quantum chemistry [5], are discussed as well.

[1] D. Wales, Cambridge University Press, 2003,Energy Landscapes: With Applications to Clusters, Biomolecules and Glasses. [2] G. Maisuradze, D. L. Thompson, J. Phys. Chem. A, 2003, 107, 10002. [3] M. P. Haag, K. H. Marti, M. Reiher, in preparation [4] R. Dawes, D. L. Thompson, A. F. Wagner, M. Minkoff, J. Chem. Phys., 2008, 128,084107. [5] K. H. Marti, M. Reiher, J. Comput. Chem., 2009, 30,2010-2020. 514 CHIMIA 2011, 65, No. 7/8 Computational Chemistry

ComputationalChemistry Computational Chemistry Computational Chemistry CC 29 Computational Chemistry CC 30

CASCI-Type Wave Function Expansionfor Very LargeActive Embedding vs supermolecular strategies in evaluating the Spaces hydrogen-bonding-induced shifts of excitationenergies

KatharinaBoguslawski,KonradH.Marti,MarkusReiher∗ G. Fradelos∗,J.J.Lutz†,T.A.Weso�lowski∗,P.Piecuch†,MartaW�l och‡

ETHZurich, Laboratorium f¨ur PhysikalischeChemie, ∗ D´epartementdeChimiePhysique,Universit´e de Gen`e ve,30, quai Wolfgang-Pauli-Strasse 10, 8093 Zurich,Switzerland Ernest-Ansermet,CH-1211 Gen`e ve 4, Switzerland, † DepartmentofChemistry, Michigan State University,EastLansing,Michigan48824, USA, ‡ Departmentof In quantumchemistry,electronicwavefunctionsare usually represented Chemistry, Michigan Technological University, Houghton,Michigan49931, USA as configuration-interaction(CI)expansions. However, thedimension of ∗ this many-particlebasis growsbinomially with thenumberoforbitals Shifts in the π → π excitation energyofcis-7-hydroxyquinoline, induced andelectrons presentinthe moleculesunderstudy. by hydrogen bonding with small molecules, obtained with the frozen- We presentanefficient proceduretoconstruct CI-typeelectronicwave densityembedding theory (FDET)[1], are compared with theresultsof functionsofmolecularsystemsthatrequireverylarge active spaces for the high-levelequation-of-motion coupled-cluster (EOMCC) calculations aqualitativelycorrect descriptionoftheir electronicstructure[1].Our with singles, doubles, andcompletelyrenormalized triples [2], corrected procedureisbased on thedensity-matrixrenormalizationgroup (DMRG) to restore thepropertyofsize intensivity, whichprovide the reference ab algorithm [2]thatprovidesthe necessary information in termsofthe initio data,the supermoleculartime-dependentdensityfunctional theory eigenstatesofthe reduced densitymatrices.DMRGcan handlemuch (TDDFT) calculations, andthe available experimental data.Unlike in larger active spaces withouttruncatingthe complete N-particleHilbert the supermolecular EOMCC andTDDFT cases, where eachcomplexation- spacewhich allows onetoobtainqualitativelycorrect wave functionsand induced spectral shift is evaluatedbyperforming twoseparatecalcula- energies even for very difficultelectronicstructures [3]. tions, onefor thecomplex andanotherone for the isolated chromophore, Compared to thepurelynumerical DMRG optimization scheme,anan- the FDET shifts are evaluated as the differences of the excitationenergies alytic wave function in termsofaCI expansion provides deeper insights determinedfor thesame many-electron system,representing the chro- into theelectronicstructureofthe corresponding moleculesunderstudy mophore fragment with twodifferenteffectivepotentials. It is demon- andallows abetterunderstanding of theconvergence behavior of the strated that thespectral shiftsresultingfrom the FDET calculations DMRG algorithm.Becauseofthe binomial scalingofthe Hilbert space employing nonrelaxedenvironmentdensities areinexcellentagreement with thesizeofthe active space, asophisticated MonteCarlo sampling with thereference ab initio data,whereas thesupermolecularTDDFT routinehas been implementedthatconstructsanaccurate representation resultsare in disagreement with the reference [3,4]. 97 of theelectronicwavefunction. [1]T.A. Wesolowski, A. Warshel, J. Phys.Chem. ,8050 (1993) [2] P. Piecuch, J.R. Gour,and M. W�loch, Int.J.Quantum Chem. 109,3268 (2009) [1]K.Boguslawski,K.H.Marti andM.Reiher, submitted, [3] G. Fradelos,J.J. Lutz, T.A.Weso�lowski, P. Piecuch, M. W�loch, J. Chem. Theory arXiv:1101.0528v1[physics.chem-ph]. Comput. submitted, (2011) [2]S.R.White, Phys.Rev. Lett., 1992, 69,2863.;K.H.Marti [4] G. Fradelos,J.J. Lutz, T.A.Weso�lowski, P. Piecuch, M. W�loch,in:Progress in andM.Reiher, Z. Phys.Chem., 2010, 224,583. Theoretical Chemistry and Physics,Vol. XXX, Advances in theTheory of Quantum [3]K.H.Marti,I.MalkinOnd`ı k, G. Moritz andM.Reiher, J. Systems in Chemistryand Physics,edited by P.E. Hoggan, E. Br¨andas, J. Maruani, Chem. Phys., 2008, 128,014104. G. Delgado-Barrio,and P. Piecuch, (Springer, Dordrecht,2011), (submitted) Computational Chemistry Computational Chemistry Computational Chemistry CC 31 Computational Chemistry CC 32

LowTemperature Gas Phase Structures and IR Spectraof Applications of the Molecular Lipophilicity Potential in AutoDock Amino Acids and Polypeptides Predicted by DFT Methods

M. D¨omer,M.Guglielmi, I. Tavernelli, and U. R¨othlisberger N. Oberhauser, P.-A. Carrupt, A. Daina,

Ecole Polytechnique F´ed´e rale, Lausanne, Switzerland School of Pharmaceutical Sciences, University of Geneva, University of Lausanne,Quai Ernest-Ansermet 30, CH-1211, Geneva 4, Switzerland ˆ 10 K HΨ=EΨ The molecular lipophilicity potential (MLP) [1] is a fragment-based molecu- lar interaction field that has found broadapplications in computational che- mistry and ligand-based drug design. It is nowadays routinely used for 3D-QSAR.Further enhancements showed that the MLP can also successful- ly be applied in target-based methods. A simple MLP-approved adjustment DFT methods are importanttools for predictions of electronic structure of the definition of protein cavity hydrophobic areas allowed an overall gain and energetics of biologically and pharmaceuticallyrelevantmolecules of GOLD docking accuracy [2]. providing an interpretation of spectroscopic experiments at the atom- istic level. The open source docking tool AutoDock [3] uses pre-calculated grid maps In our initial benchmarking studies on bare and microsolvated proto- that contain energy information about various types of possible interaction nated tryptophan we have assessed the performance of various popular potentials. An embedment of a new lipophilicity term into the generation DFT methods to predict the lowest energy structure associated with the process of these maps might help improve the overall docking performance, high-resolution experimental IR spectrum at lowtemperature. Forthe especially on hydrophobic targets. Recent simulations influencing apolar low-energy structures, the harmonic vibrational frequencies were calcu- (carbon)as well as polar (H-bond donor and acceptor) grid maps in respect lated and compared with high resolution experimental IR spectra on the to the MLP showed promising perspectives. Furthermore, additional infor- cold ions. Withthe adequate methodology in place we have solved the mation given by the molecular hydrogen-bonding potentials(MHBPs) [4], 3D structure of adecacpeptide, gramicidin S, isolated in gas phase, by is also under studyto gain a more accurate descriptionofpolar binding matching our calculated vibrational frequencies to the experimental cold sites. ion IR spectrum at 10 K[1]. To our knowledge gramicidin Sisthe largest molecule for whichthe accurate lowtemperature gas phase structure has [1]P. Gaillard, P.A. Carrupt, B. Testa and A. Boudon, J. Comput. - ever been determined. At the same time it is of great pharmaceutical Aided Mol. Design 1994, 8, 83–96. relevancy due to its antibiotic activity. The structure presented here may [2]A. Nurisso, abstract “Molecular docking using the MLPas hydrophobic serveasastarting pointfor the rational design of gramicidin Sanalogs. descriptor: impact on the GOLD docking performance” [3]G. M. Morris et al., J. Comput. Chem., 1998, 19,1639-1662. [1] N.S. Nagornova,M.Guglielmi, M. Doemer,I.Tavernelli, U. [4]S. Rey et al., J. Mol. Graphics Model., 2001, 19, 521–535 Rothlisberger, T.R. Rizzo and O.V. Boyarkin, Angew. Chem. Int. Ed., in press. Computational Chemistry CHIMIA 2011, 65, No. 7/8 515

ComputationalChemistry Computational Chemistry CC 33 ComputationalComputational Ch Chemistryemistry CC 34

M¨ossbauer spectroscopy forheavy elements A Generalized-Gradient Approximation Exchange Hole Model for Dispersion Coefficients Samuel Fux1,StefanKnecht2,Robertvan Meer3,Lucas Visscher3, 4 1 TrondSaue andMarkusReiher Stephan N. Steinmann and Clemence Corminboeuf 1ETH Z¨urich, Laboratorium f¨ur PhysikalischeChemie, Wolfgang-Pauli-Strasse 10, 8093 Zurich,Switzerland Institute of Chemical Sciences and Engineering, Ecole Polytechnique 2University of Southern Denmark, DepartmentofPhysics and Fédérale de Lausanne, 1015 Lausanne, Switzerland Chemistry,Campusvej55, 5230 Odense,Denmark 3VU UniversityAmsterdam,Amsterdam Center forMultiscale Density functional approximations fail to provide a consistent description of Modeling, De Boelelaan 1083, 1081 HV Amsterdam,The Netherlands weak molecular interactions arising from small electron density overlaps. A simple remedy to correct for the missing interactions is to add a posteriori 4Universit´ePaulSabatier, LaboratoiredeChimieetPhysique an attractive energy term summed over all atom pairs in the system. For Quantique, 118 Route de Narbonne,31062 Toulouse cedex, France general applicability, the dispersion coefficients used in such corrections should depend on the electron density, with no increase in the computational M¨ossbauer (MB) spectroscopyisasensitivespectroscopic technique, that cost. We present a simple method for computing accurate density-dependent probes tiny changesinthe energy levels of M¨ossbauer active nuclei.The dispersion coefficients. The model, called XDM(sC-BR), relies on a genera- electrondensity at thenucleus (contact density) is strongly relatedtothe lized gradient-type approximation to Becke and Johnson’s exchange hole calculationofthe isomer shiftinMBspectroscopy. Forheavy nuclei,the dipole moment formalism.[1] Our most cost-effective variantgives a mean contactdensity is affected by relativistic effects, i.e., its calculationina [2] absolute errorin the C6 coefficients for 90 complexes below 10%. These non-relativistic framework employingthe pointchargeapproximation for dispersion coefficients are combined with our density dependent damping theatomicnucleus leadstosubstantialerrors.Amore elaboratedescrip- function.[3] The inclusion of the missing long-range van der Waals interac- tion of thecontact densityisobtainedinafully relativistic framework, tions in density functionals usingthe derived coefficients leads to highly wherethe nucleusisdescribed by afinite nuclearchargedistributionand accurate typical noncovalent interaction energies. The dominant classical thecontact densitythustakes finite values.Inour benchmarkstudy [1], Hirshfeld partitioning BLYP-dDXDM(sC-BR) gives, for instance, a mean -1 we investigatethe contactdensity forthe series HgFn (n =1,2,4)with absolute deviation of only 0.14 kcal mol for the S22 training set. respect to theneutralatominthe framework of scalar-relativisticone- We discuss general improvements obtained for standard density functionals component, two- andfour-componentrelativistictheory. Variousdensity such as PBE, BLYP and B3LYP and present results for a broad variety of functionalsweretestedand compared to highly accurate finite-fieldfour- benchmark sets featuring both, intra- and intermolecular weak interactions. componentCCSD(T) calculations.Itisfound that DFTisnot able to recoverthe non-monotonous decrease in thecontact density, that can [1] A. D. Becke andE. R. Johnson, J. Chem. Phys. 2006, 124, 014104. be attributed to changesinthe populationand thepolarizationofthe [2] S. N. Steinmann andC. Corminboeuf, J. Chem. Phys. 2011, 134, 6s -orbital of mercury. 044117. 1/2 [3] S. N. Steinmannand C. Corminboeuf, J. Chem. Theory Comput. 2010, 6, [1]S.Knecht, S. Fux, R. vanMeer,L.Visscher, M. Reiher,T.Saue, 1990. Theor. Chem.Acc., 2011,DOI:10.1007/s00214-011-0911-2.

Computational Chemistry ComputationalComputational ChChemistryemistry CC 35 Computational Chemistry CC 36

Modeling of Anion-π Interactions of Substituted Ligand Binding Study of Carbonic Anhydrase 2 Naphthalenediimides : Structure and Function MaurusSchmid, Thomas R. Ward ,Markus Meuwly Jiri Mareda, Stefan Matile,Daniel Emery UniversityofBasel, Spitalstrasse 51, 4056 Basel Department of Organic Chemistry, University of Geneva, UniversityofBasel, Klingelbergstrasse80, 4056 Basel 30 quai Ernest Ansermet, 1211 Geneva 4, Switzerland Carbonic anhydrases (CAs)are ubiquitous metalloenzymes that catalyze Contrary to the popular and widespread cation-πinteractions, the reversible hydrationofcarbon dioxidewith remarkableefficiency. applications of the anion-πcounterpart arequiterare. Thequestion whether CAshave been thefocus of manybiophysicalstudies of protein-ligand anion-πinteractions could be used in supramolecular context to create interactions. Today, at least 25 clinically used drugs are known to display significant function such as transmembranetransport, sensing or pronouncedCAinhibitory properties[1]. To design aprotein with par- organocatalysis is attractive but also challenging.[1] ticular properties, understanding the influence of residues is crucial. In In this contribution is described DFTmodeling for series of anion-π this work we report on acomputational strategy that allows predicting complexes. Noteworthybinding energies between naphthalenediimides strong inhibitors and potentially beneficial mutations of the protein. The (NDIs) and several types of anions were also computed. The experimentally activesite of mostCAs contains aZn(His)3whichisessential for cataly- observedanion transport selectivitywas in addition confirmed by theory. sis. The carbonic anhydraseprotein is ideal for the design of potentand Anion-πinteractions were systematically modulated during the modelingby selectiveinhibitors. NDIcoresubstitutions as well as by structuralvariations of N-aromatic We report aMolecular Mechanics Generalized Born SolventApproxi- moieties.[2] The computational simulations together with tandem mass mation(MMGBSA[2]) study comparing the bindingfree energy for 18 spectroscopydemonstrated that anion-πinteractions account for function.[3] sulfonamides. Using this method, not only the total binding freeenergy, The molecular modelingwas also used to design appropriate substrates but alsothe influenceofaparticular residue can be examined. Thus the benefiting from enhanced halogen bond bindings. effect of mutations on keyresidues on thebinding canbedetermined.To The anion-πinteractions have the potential to catalyzeorganic reactions validate the simulations, we comparethe resultswith publishedbiophys- via the stabilization of anionictransition states. This concept was probed by icaldata as well as withasimulationusingQM/MM simulations with DFT simulations of organocatalysis for hydrogenation transfer, which will theSelf-consistentchargeDensity-Functional Tight-Binding(SCCDFTB also be discussed in this contribution. [3]) method. Resultsshowahigh (up to R=0.90) correlationbetween [1]J. Mareda and S. Matile, Concept in Chem. Eur. J., 2009, 15,28. thepredictedvalues and biophysicaldata. [2]J. Míšek, A. Vargas Jentzch, S. Sakurai, D. Emery, J. Maredaand S. [1] Supuran,C.T, Naturereviews. Drug discovery, 2008, 7,168. Matile, Angew. Chem. Int. Ed., 2010, 49, 7680. [2] Still, W. C. et.al, Journal of the American Chemical Society, [3]R.E. Dawson, A. Henning, D.P. Weimann, D. Emery, V. Ravikumar, J. 1990, 112,6127. Montenegro, T. Takeuchi, S. Gabutti, M. Mayor, J. Mareda, C.A. Schalley and S. Matile, Nature Chem., 2010, 2, 533. [3] Elstner, M. et.al, Phys.Rev. B, 1998, 58,7260. 516 CHIMIA 2011, 65, No. 7/8 Computational Chemistry

Computational Chemistry ComputationalChemistry Computational Chemistry CC 37 Computational Chemistry CC 38

QUANTIFICATION OF INTRAMOLECULAR CHARGE MolecularDynamicsSimulationofNitricOxideinMyoglobin TRANSFER EFFECTS BY BLOCK LOCALIZED WAVE FUNCTION Myung WonLee andMarkusMeuwly

L. Bomble,S.N.Steinmann, C. Corminbœuf DepartmentofChemistry,UniversityofBasel Klingelbergstrasse 80, CH-4056 Basel, Switzerland Laboratory for Computational Molecular Design ,EPFL, Lausanne Myoglobin (Mb) is asingle-chainglobular proteincontainingaheme The extentofelectronic delocalization (i.e.intramolecular charge trans- groupand is found in muscle tissueofvertebrates in general. While Mb fer) can be quantified by comparing theenergy of absolutely localized hasbeen knownasadioxygen storage protein, Mb hasalsothe capa- statestothat of the canonical (ground state)state. Unfortunately,the bilitytocatalyze reactionsinvolving smallmoleculesbyconcentrating construction of diabatic statesissued fromValence Bond (VB) theory andorienting diatomic moleculessuchasNO, CO,and O2.[1] From is computationally very expensivedue to the non-orthogonalityofthe previousstudies,ithas been discovered that thereare multiple cavities orbitals. Alternatively,the Block Localized Wave function approach, de- inside Mb andthatsmall ligandmoleculescan migratebetween various veloped by Mo andal. [1] incorporatesthe physicalintuition of VB at cavities.[1-2] thecomputational cost of Hartree-Fock or Kohn-Sham densityfunctional Moleculardynamics(MD)simulations of proteins canprovide detailedin- theory.The electrons are divided into several blocks, eachexpanded on formationthatcannotbeobtaineddirectly fromthe experiments,and are arestricted number of atomic orbitals. The doubly occupied orbitals are considered as an indispensablecomplementtothe experimentalmethods. orthogonalwithin one block (likecanonicalMOs), but non-orthogonal In this work,wecarried outthe classicalMDsimulations of myoglobin between differentblocksand localizedintermsoftheir basis function with anitricoxide (NO) molecule inside it andinvestigated theproperties expansion (likeVBgroup orbitals). The restrictionimposed on theMO anddynamicsofthe system in detail. While theelectrostatic interactions expansionthusenablesthe selectivesuppression of orbitalinteractions. arecommonlydescribed by pointcharges on atomsinthe classicalMD The orbitalexpansion coefficients of the diabatic state are obtained by simulations, theaccuracy of themodel canbeimprovedbyincluding aself-consistentoptimization. Anumberofalgorithms canbeused to higher multipolemoments (MTPs) in addition to pointcharges.Due optimize these BLWorbitals. We previously implemented in Dalton an to theunpaired HOMO electronofNO, themolecule does nothavethe algorithm proposed by Gianinetti et al.[2] to analyze the extentof cylindrical symmetry around themolecularaxis, whichnecessitatesmore charge transfer in cyclic systems [3,4].Inorder to extend the scope of sophisticated treatment forMTPs on NO.Weusedmodified CHARMM applicationstoabroader varietyofmolecules, we herepresentthe im- code with MTPs on theNOmolecule up to quadrupole moment in order plementation and applicationsofamore general algorithm [5]. to treat theNOmolecule more accurately in theclassicalMDsimulations [1] Y. Mo and al., J. Chem.Phys., 1998, 109,1687. of NO-containingmyoglobin. [2] E. Gianinetti and al., Int.J.Quantum Chem, 1996, 60,157. [3] S. N. Steinmann and al., Chem.Com., 2011, 47,227. [1]H.Frauenfelder, B. H. McMahon, R. H. Austin, K. Chu, and [4] S. N. Steinmann and al., Submitted. J. T. Groves, Proc.Natl. Acad. Sci. 2001, 98,2370. [5] H. Stolland al., Theor. Chem. Acc., 1980, 57,169. [2]F.Schotte,M.Lim,T.A.Jackson,A.V.Smirnov,J.Soman, J. S. Olson, G. N. Phillips, M. Wulff,and P. A. Anfinrud, Science 2003, 107,1944. Computational Chemistry Computational Chemistry CC 39 ComputationalComputational Che Chemistrymistry CC 40

Investigation of h-BN/Rh(111) Nanomesh Interaction with Molecular Dynamics of MbNO – differences in heme response to bind- Water and Phthalocyanine ing and release of NO.

YunDing,Marcella Iannuzzi, J¨urg Hutter Jaroslaw J. Szymczak and Markus Meuwly*

Physikalisch-Chemisches Institut, Universit¨at Z¨urich, Institute of Physical Chemistry, University of Basel, Winterthurerstrasse 190, CH-8057 Z¨urich, Switzerland Klingelbergstrasse 80, 4056 Basel, Switzerland.

The hexagonal boron nitride nanomesh is acorrugated structure with a Myoglobin (Mb) is the one of the most intensively studied hemepro- periodicityof3.22 nm, whichisformed self-assembly by decomposition tein over the last two decades. Due to its biological function related to stor- of borazine on aclean Rh(111) surface at high temperature[1]. Small ing and releasing oxygen in muscle tissue, as well as to its relativelysmall water clusters[2], nano-ice crystals[3] and other organic molecules have size it has been chosen as a representative system to study binding of oxy- been observed trapped in the pore of the nanomesh, whichmakes the gen and other diatomic molecules to the heme, and studying associated pro- nanomesh an interesting template for self-assembly of ordered and dis- tein response. Although binding of diatomic ligands is one of the most fun- tantmolecule arrays. damental processes occurring in living systems, its nature is still not well understood. The progress in ultrafast spectroscopic techniques brought a Our DFT model of the nanomesh reproduces the structural corrugation new insights about structural changes on the sub-picosecond scale. Experi- and modulation of the electrostatic potential. Moreover, the STM to- mental findings alone, however, are not able to explain the complex and pographyand projected DOSare also consistentwith the experimental heterogenous dynamics. Therefore molecular dynamics (MD) simulations ones. Hence, we employthis model to better understand structural and are themethod of choice electronic properties of molecular systems adsorbed on the nanomesh, Our present investigation is focused on MbNO. In the light of recent thus contributing substantially to the interpretation of theexperiment. experimental findings1 we investigate at atomistic level theprocess of nitric In particular, we could assign the most probable structure of the nano- oxide binding to the hemeinmyoglobin, with aim of understanding causes ice crystal observed by experimental STM, by comparing differentwater and differences in the time scales of instantaneous heme planar-to-domed lattices and the related distribution of dipole moments. We also address transition, and longer reverse process related to NO dissociation and rebind- open questions about optimal adsorption site, orientation, and mobility ing, respectively. of organic molecules, likephthalocyanine, deposited on the nanomesh.

[1] M. Corso, W. Auw¨arter, M. Muntwiler, A. Tamai, T. Greber, [1] Kruglik, SG; Yoo, BK; Franzen, S; Vos, MH; Martin, JL; Negrerie, M, J. Osterwalder , Science, 2004, 303,217. PNAS. 2010, 107, 13678. [and references therein] [2] H. F. Ma, Y. Ding, M, Iannuzzi, T. Brugger, S. Berner, J. [2] Nutt, DR; Karplus, M, Meuwly, M, J. Phys. Chem. B 2005, 109, 2118. Hutter, J. Osterwalder, T. Greber, submitted [3] Ionascu, D; Gruia, F; Ye, X; Yu, AC; Rosca, F; Beck, C; Demidov, A; [3] H. F. Ma, T. Brugger, S. Berner, Y. Ding, M. Iannuzzi, J. Olson, JS; Champion, PM, J. Am. Chem. Soc. 2005, 127, 16921 Hutter, J. Osterwalder, ChemPhysChem, 2010, 11,399 Computational Chemistry CHIMIA 2011, 65, No. 7/8 517

ComputationalComputationalC Chemistryhemistry CC 41 ComputationalComputational Che Chemistrymistry CC 42

Non-empiricalcrystal structuredeterminationofthe complex PEG- On binding and reactivity of potent covalent inhibitors AgNO3 using Rietveld refinementand MonteCarlo simulation. of fatty acide amide hydrolase

RAMANANTOANINA Harry and DAUL Claude A., Giulia Palermo,1 Marilisa Neri,2 Pablo Campomanes,2 Ivano Tavernelli,2 Andrea Cavalli,1 Ursula Roethlisberger2 and Marco De Vivo1 Department of Chemistry, University of Fribourg, Chemin du musée9., 1700 Fribourg,Switzerland 1Department of Drug Discovery and Development, Italian Institute of Technology, via Morego 30, I-16163 Genova, Italy. 2Laboratory of Computational Chemistry and Biochemistry, Thedetermination of thecrystal structurefromX-ray powderdiffraction Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzeland pattern is amajor scientific challenge.Afirstattemptto solveapowderdif- fraction wasestabilished since1967. [1]The methodology [2]evolved since Covalent inhibitors have gained renewed interest in the drug discovery theRietveld refinement is coupled with aMonteCarlo method in orderto community because of recenteffectivedrugs that function through a cova- determinatethe molecularstructure. lent mechanism. In this study, we investigated the mode of action of cova- In thepresent work,the molecularstructure of PEG-AgNO3 (PEG stands for lent inhibitors of the membrane-bound fatty acid amide hydrolase (FAAH) Polyethyleneglycol) is determined frompowderX-ray diffractionusing enzyme, which is a promising target for drug discovery in the field of pain MonteCarlo method. Theapproach is to find out an adequate crystalstruc- and inflammation[1],[2]. Based on crystallographic data, we built a realistic ture starting modelratherthanextracting thestructuralinformation from the model system of FAAH bound to the membrane to perform classical mole- diffraction pattern.Then, aRiedveld refinement is developed by fitting the cular dynamics (MD) and quantum mechanics/molecular mechanics calculated theoretical diffraction pattern to themeasured experimental one. (QM/MM) simulations [3]. We investigated both binding and reactivity of Theprincipal problem of this approach is to obtain asufficiently good start- some representative FAAH inhibitors, whichhave been used to generate ing structuremodel. Two concepts were investigated:(a) Replace thesilver clinical candidates. These lead compounds were also compared with the by potassium atoms, thecrystal structureofwhich hasbeen previously de- main FAAH endogenous substrate, i.e. anandamide. termined [3], and (b)observe thebehaviorofthe complex while theligand Overall, our goal is to improve the understanding of the mode of action of is modified by increasing theamount of ethoxy group within thePEG clus- covalent FAAH inhibitors. This could help in the rational design of new and ter, thecrystal structures of which have been available. better compounds. Theresults corresponding to each conceptare particularly similarand con-

stitute afirst modelfor thecrystal structureofthe PEG-AgNO3 complex. [1]Ahn, K.; Johnson, D. S.; Cravatt, B. F., Expert Opin Drug Discov 2009, 4, 763. [1]Rietveld H.M., Acta Cryst., 1967, 22,151. [2] Cravatt, B. F.; Giang, D. K.; Mayfield, S. P.; Boger, D. L.; Lerner, R. [2]HarrisK.D.M., Tremayne M.,Lightfoot P.,Bruce P.G., J. Am.Chem. A.; Gilula, N. B., Nature 1996, 384, 83. Soc., 1993, 116,3543. [3] Bracey, M. H.; Hanson, M. A.; Masuda, K. R.; Stevens, R. C.; Cravatt, [3]Bekturova Y.A.,Kudaibergenova S.Y.,UshanovaV.Z., Saltybayeva B. F., Science 2002, 298, 1793. S.S.,Polymer ScienceU.S.S.R., 1987, 29(1),190.

Computational Chemistry Computational Chemistry Computational Chemistry CC 43 Computational Chemistry CC 44

– Calculation of [GdDOTA (H2O)] Zero Field Splitting with CO transferbetween Fe of heme and CuB in cytochrome c LFDFT using ·an embedding potential oxidase

F. Senn1,W.Urland2,C.A. Daul1 Maksym Soloviov,Markus Meuwly�

1:DepartmentofChemistry,UniversityofFribourg, Switzerland. DepartmentofChemistry,UniversityofBasel, Klingelbergstrasse 80, 2:Institut f¨ur Anorganische Chemie, Universit¨at Hannover, Germany. CH-4056 Basel

Paramagnetic Gd(III)induces astrong NMR-relaxation enhancementof Cytochrome coxidase (CcO) is aprotein involved in catalysis of the ter- neighboring water protons [1]. Forbetter understanding the spin relax- minal step in cellular respirationinmitochondria or bacteria, whichisa ation of Gd(III)ions in solution, the zero field splitting (ZFS) plays an transfer of four electrons from cytochrome ctodioxygen[1].In fact, the importantrole determinating the former. [2]. With the peculiar stere- cytochrome heme a and Cu form abinuclear center that is thesite of ochemical properties for the macrocyclic ligand DOTA(1,4,7,10-tetraaza- 3 B – oxygen reduction. In contrast to O2,COforms astable complex with 1,4,7,10-tetrakis (carboxymethyl) cyclododecane) [3], [GdDOTA (H2O)] · enzyme. Due to agreateraffinityofCOfor CuB than for Fe(II) of heme is even one of the most relevantcontrast agents for MRI [3]. a ,CObinds to Cu whenFe-CO bond is broken. There are several ‘The Magnetic and spectroscopic properties of the lanthanide ions de- 3 B mechanisms for the transfer of CO from Fe(II) to Cu :elongation of Fe- pend on the felectron structure, whichisgenerally understood in the B CO followedbyrotation,byforming intermediateCu -OC state or by framework of amodel where the forbitals are considered shielded from B the chemical environment.‘ [1] The description of the multiplet structure forming free CO. To figure out the mechanism, potential energy surfaces and energies of states in this given basis of fspinors are obtained with (PES) have been buildbased on DFT calculations, B3LYP functional the ligand field densityfunctional theory (LF-DFT) [4]. This method and LANL2DZ (for Cu and Fe)and 6-31g basis sets have been used. 3+ The effective potential has been build based on thePES. Boththree- has already been adapted to asmaller Gd(III)system, Gd(H2O)8] [1], in the presentwork we apply it for the calculation of the ZFS of pointfluctuating charge model [3], that reproduces accurately the dipole – and quadrupole moments, and standard pointchargemodel of CO have [GdDOTA (H2O)] ,using an embedding potential to describethe lig- and influence· in the DFT calculations. been used.Athree-pointfluctuating charge model for CO is thus de- veloped andimplemented by the group in CHARMM.MDsimulations [1] A. Borel, L. Helm, C.A.E. Daul, Phys. Chem. Let., 2004, of CO are studied withdifferent models.Dynamics of CO bound and 383,584-591. unbound to Fe of heme is analyzed in detail. [2] M. Benmelouka, J. vanTol, A. Borel, M. Port, L. Helm, L.C. [1] J. Koepke, E. Olkhova, BiochimicaetBioph. Acta, 2009, Brunel, A.E. Merbach, J. Am.Chem. Soc., 2006, 128,7807- 7816. 1787,635-645. [3] S. Aime, A. Barge, F. Benetollo, G. Bombieri, M. Botta, F. [2] J. Treuffet,K.J.Kubarychetal Proc.Natl. Acad. Sci. Uggeri, Inorg. Chem., 1997, 36,4287-4289. U.S.A., 2007, 104,1570515710. [4] a) M.A. Atanasov, C.A. Daul, C. Rauzy, Chem. Phys. Lett., [3] D. R. Nutt and M. Meuwly, Proc.Natl. Acad. Sci., 2004, 2003, 367,737-746. b) M.A. Atanasov, C.A. Daul, C. Rauzy, 101,5998 6002. Struct. Bond., 2004, 106,97-125. 518 CHIMIA 2011, 65, No. 7/8 Computational Chemistry

Computational Chemistry Computational Chemistry Computational Chemistry CC 45 Computational Chemistry CC 46

Mechanistic Insights for the Retroviral Integrase-DNA Oxygen Coordination to the ActiveSite of Hmd in Relation to assembly [FeFe] Hydrogenase

ChristopheBovigny, Marco Stenta,Matteo Dal Peraro Arndt R. Finkelmann,Martin T. Stiebritz, Markus Reiher

Institute of Bioengineering, School of Life Sciences,EPFL, CH-1015 Laboratorium f¨ur Physikalische Chemie, ETH Z¨urich, Lausanne, Switzerland Wolfgang-Pauli-Str. 10, 8093 Z¨urich, Switzerland

Retroviralintegrase (RI) bindsviral DNAstrands and fuses them into the Hydrogenases are enzymes whichcatalyze the reversible oxidation and chromosomal DNA of thehost cell. It has been therefore apreferential formation of molecular hydrogen. The technical application of [FeFe] targetfor HIV-1 drugdevelopment. RI integration andinhibition mech- hydrogenase, the class of enzymes that shows the highest catalytic H2 anisms were howeverlargely unknown, tillthe recent resolution of the formation activity, is limited by its high O2-sensitivity. One postulated integrase-viral DNA complex structure from the prototypefoamyvirus mechanism for the dioxygen induced inhibition is triggered by O2 coordi- nation to the activesite [1]. In sharp contrast the H oxidizing mono-iron (PFV) [1]. This structure finally unveils the general structural architec- 2 hydrogenase (Hmd) is not irreversibly inhibited by O . ture promoting viralintegration, andwill pave the wayfor arational 2 Here we presentastudy on energetic differences in oxygen coordination developmentofimprovedHIV RI inhibitors [2][3]. to cluster models of the activesites of Hmdand [FeFe] hydrogenase as obtained from DFT calculations. Whereas O coordination is clearly In this study,weinvestigatethe reaction mechanismleadingtothe for- 2 exothermic for[FeFe]hydrogenase, it is endothermicfor Hmd [2]. By mationofthe RI-DNAcomplex using quantum mechanics/molecular me- application of arecently proposed unifying concept to structurally and chanics (QM/MM)simulations. The structural andkinetic characteri- electronically relate both activesites [3] we obtain an inversion in O zation of thisinitial step of DNAintegrationwill provide usefulinsights 2 affinitybymutual first-shell ligand exchange. In this context aCN− to for thedevelopmentand optimization of new potentinhibitors. CO ligand exchange reveals the largest impact on coordination energy [2]. Furthermore the effect of this exchange on the stabilityofintermediates [1] Stephen Hare, Saumya ShreeGupta, Eugene Valkov,Alan En- and reaction barriers of the postulated regular catalytic cycles (i.e. hy- drogen formation/oxidation) is investigated. The results mayindicate gelman and Peter Cherepanov, Nature, 2010, Vol464,232- whyboth enzymes are in need for differentmechanistic pathways and 236. whythey exhibit differentreactivity. [2] Xiang Li, Lavanya Krishnan,Peter Cherepanov, Alan Engel- man, Virology, 2010, Vol411,194-205. [1] M. T. Stiebritz andM.Reiher, Inorg. Chem., 2009, 48,7127.; [3] Goedele N. Maertens, Stephen Hare, Peter Cherepanov, Na- S. T. Stripp, G. Goldet, C. Brandmayr, O. Sanganas, K. A. ture, 2010, Vol468,326-330. Vincent, M. Haumann, F. A. Armstrong and T. Happe, Proc. Natl. Acad. Sci. USA, 2009, 106,17336. [2] M. T. Stiebritz, A. R. Finkelmann and M. Reiher, Eur. J. Inorg. Chem., 2011, 7,1163. [3] M. T. Stiebritz and M. Reiher, Inorg. Chem., 2010, 49,5818.

ComputationalComputational Ch Chemistryemistry CC 47 ComputationalComputationalC Chemistryhemistry CC 48

Study of Thorium salts in waterby Molecular Dynamics. How Can Reaction Barriers Dictate Ground State Properties?

1 2 3 Beuchat César , Spezia Riccardo and Gagliardi Laura Daniel F. Jana and Clemence Corminboeuf

1 University of Geneva, Department of Physical Chemistry, Quai Ernest Institute of Chemical Sciences and Engineering, Ecole Polytechnique Ansermet 30, CH-1211 Geneva, Fédérale de Lausanne, 1015 Lausanne, Switzerland 2CNRS, LAMBE UMR 8587, Université d'Evry Val d'Essonne, Evry, France, 91025, France Our laboratory has recently introduced aquantum chemical approach ena- 3 University of Minnesota, Department of Chemistry and Supercomputing bling the direct probing and tuning of electron delocalization effects on mo- Institute, 207 Pleasant St. SE, lecular response properties.[1] The proposed methodology first constructs a specific resonance (i.e. Lewis) structure, in which conjugative interactions The hydration of actinides is a topic that attracts the attention of many are “disabled” by using the simplest variant of valence bond theory that is chemists, as thisprocess is of relevance to environmental problems. Th(IV) the block-localized wavefunction method.[2] The approach defines the in- is the "first" actionoid cation and it can form highly coordinated aquaca- termediate electron-localized state self-consistently at the density functional tions. Early XAS experiments suggested a coordination number of 10 ±1in theory level. Computations of the molecular response properties are per- HClO4. Recently, comparing crystal and liquid structures of Th(IV) indi- formed on the standard (delocalized) structures and those with "non- cated that in presence of Br-ahomoplectic tenfold coordinated aqua-ion interacting" (localized) double bonds. Illustrative applications served to clar- was proposed. ify abnormal chemical shifts of π-conjugated systems and solve long- standing organic chemistry problems.[1] We present the results of a simulation studyof Th(IV) in water by employ- 1 We now demonstrate that the ground state properties (e.g. Jcc-coupling con- ing different force fields. (1) NEMO type potentials was developed for sys- - - stant) of aseries of substituted semibullvalenes fully correlate with the acti- tems composed of Th(IV), Br /Cl and water as recently done for Ln3+[1]; vation barrier of their respective Cope rearrangement. In other words, our (2) a simpler polarisable force field based on ionic radii has been also de- analysis is supportive of afascinating picture that is the fingerprint of the veloped to investigate the dynamical behavior of this system, in a fashion transition state in the ground state, which enables the direct estimation of similar to the one employed to study Ln3+ hydration [2]. By combining kinetic parameters from the ground state properties. This electronic phe- these methods we obtain results in good agreement with available EXAFS nomenon, that is impossible to demonstrate experimentally[3], has been first data [3], andwe propose an approach that can be transferable to other re- hypothesized and demonstrated for other pericyclic reactions (e.g. Diels- lated systems for which little experimental data areavailable. Alder).[1b]

[1] C.Beuchat, D.Hagberg, R.Spezia, L.Gagliardi, J. Phys. Chem. B. 2010, [1] S. N. Steinmann, D. F. Jana, J. I-C. Wu, P. v. R. Schleyer, Y. Mo and C. 114 (47), 15590. Corminboeuf, Angew. Chem., Int. Ed. 2009, 48, 9828-9833; b) S. N. Stein- [2] M.Duvail, P.Vitorge and R.Spezia, J. Chem. Phys. 2009, 130, 104501. mann, P. Vogel, Y. Mo and C. Corminboeuf, Chem. Commun., 2011, 47, [3] N.Torapava, I.Persson, L.Eriksson, D.Lundberg, Inorg.Chem. 2009, 48, 227. [2] Y. Mo and S. D. Peyerimhoff, J. Chem. Phys. 1998, 109, 1687- 11712. 1697.