PCCP View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PERSPECTIVE

Cite this: Phys. Chem. Chem. Phys., 2016, 18, 20047

View Journal | View Issue

Prebiotic synthesis of nucleic acids and their building blocks at the atomic level – merging models and mechanisms from advanced computations and experiments c d d ˇponer,ab Rafał Szabla,a Robert W. Go Judit E. S ´ ra, A. Marco Saitta, Fabio Pietrucci, e f g h h Franz Saija, Ernesto Di Mauro, Raffaele Saladino, Martin Ferus, Svatopluk Civisˇ ˇponer*ab and Jirˇ´ı S

The origin of life on Earth is one of the most fascinating questions of contemporary science. Extensive research in the past decades furnished diverse experimental proposals for the emergence of first informational polymers that could form the basis of the early terrestrial life. Side by side with the experiments, the fast development of modern computational chemistry methods during the last 20 years facilitated the use of in silico modelling tools to complement the experiments. Modern computations can provide unique atomic-level insights into the structural and electronic aspects as well as the energetics of Received 29th January 2016, Accepted 8th April 2016

key prebiotic chemical reactions. Many of these insights are not directly obtainable from the experimental

DOI: 10.1039/c6cp00670a

experiments and for qualified predictions. This review illustrates the synergy between experiment and

techniques and the computations are thus becoming indispensable for proper interpretation of many theory in the origin of life research focusing on the prebiotic synthesis of various nucleic acid building

www.rsc.org/pccp

blocks and on the self-assembly of nucleotides leading to the first functional oligonucleotides.

Introduction The RNA-world hypothesis is one of the most popular concepts of the origin of terrestrial life.1,2 It assumes the existence of an ancient life-form in which RNA served as both the carrier of genetic information and a catalyst capable of catalyzing its own replication. One of the key questions of the origin of life research is therefore related to the synthesis of nucleic acids’

a

´lovopolska ´ 135, Institute of Biophysics, Academy of Sciences of the Czech Republic, Kra CZ-612 65 Brno, Czech Republic. E-mail: [email protected] b CEITEC – Central European Institute of Technology, Masaryk University, Campus Bohunice, Kamenice 5, CZ-62500 Brno, Czech Republic c Theoretical Chemistry Group, Institute of Physical and Theoretical Chemistry, ´skiego 27, Wrocław University of Technology, Wybrzez˙e Wyspian 50-370 Wrocław, Poland d Sorbonne Universite´s, Universite´ Pierre et Marie Curie Paris 6, CNRS, Institut de Mine´ralogie, de Physique des Mate´riaux et de Cosmochimie, Muse´um National d’Histoire Naturelle, Institut de Recherche pour le De´veloppement, UMR 7590, F-75005 Paris, France e CNR-IPCF, Viale Ferdinando Stagno d’Alcontres 37, 98158 Messina, Italy f Dipartimento di Biologia e Biotecnologie ‘‘Charles Darwin’’, ` di Roma, Piazzale Aldo Moro 5, Rome 00185, Italy ‘‘Sapienza’’ Universita g ` della Tuscia, Dipartimento di Scienze Ecologiche e Biologiche Universita Via San Camillo De Lellis, 01100 Viterbo, Italy h J. Heyrovsky´ Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Dolejsˇkova 3, CZ-182 23 Prague 8, Czech Republic

This journal is © the Owner Societies 2016

building blocks and the subsequent self-assembly thereof to form functional oligonucleotides. Parallel to the astonishing information growth on prebiotic strategies reconstructing life’s origin the last decade witnessed an unprecedented step forward in the computational modelling of these processes. This enabled the elaboration of an ‘‘in silico’’ approach to the origin of informational polymers. The great advantage of computations when complementing experiments in the origin of life studies is that they provide information on selected single molecules and chemical reactions, whereas prebiotic experiments always work with complex mixtures, which often make the interpretation of these studies very challenging. This is the area where computational chemistry might be instrumental for experimentalists, since theory may supplement the experimentally available information with an atomic level insight into the structural aspects, electronic structure changes, energetics, spectroscopic properties and dynamic behavior of the studied systems. Since computational chemistry is still not fully recognized as a frontline method to study life’s origin, the main aim of the current review article is to illustrate with a couple of examples the synergism between theory and experiment in this field and to foster further combined experimental–theoretical studies in this incredibly exciting topic. Traditionally, HCN has been considered to be the ‘‘stone of wisdom’’ of prebiotic chemistry.3 During the last 50 years an

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20047

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

PCCP

enormous amount of experimental information has been accumulated on the chemistry of this molecule, and currently the passage from HCN to the building blocks of nucleic acids, i.e. nucleotides is well characterized.4 In the beginning of the new millennium a new concurrent hypothesis appeared, which derives nucleic acid building blocks from formamide and enables a non-aqueous way to nucleotides.5 Although the formamide-based scenario is still quite recent and less elaborated, it represents the only prebiotic scenario, which outlines a continuous path from a simple prebiotic precursor up to the first catalytic oligonucleotides.6

In the current review we will follow both the HCN and formamide concepts and provide a comprehensive overview of the topic both from the side of experiment and theory.

ˇponer received her MSc Judit E. S degree from the Eo¨tvo¨s University, Budapest in 1993 and her PhD from the Hungarian Academy of Sciences and the Technical University, Budapest in 1996, working with Prof. Magdolna Hargittai. Then she spent four years at the J. Heyrovsky´ Institute of Physical Chemistry in Prague. Since 2001 she has been working as a senior researcher at the Institute of Biophysics of the Judit E. ˇ Sponer Academy of Sciences of the Czech Republic. Her current research focuses on the computational modelling of problems related to the origin of life.

Rafał Szabla is a PhD Student at the Institute of Biophysics of the Academy of Sciences of the Czech Republic under the supervision of Prof. Jiri Sponer. He obtained his MSc degree in 2012 from the Wroclaw University of Technology. During his Master’s studies he spent one year at the Technical University in Munich collaborating with the group of Prof. Wolfgang Domcke. He is primarily interested in theoretical investigations of the photochemistry of nucleic acids and their prebiotically plausible precursors.

A. Marco Saitta obtained a PhD in Condensed Matter Theory from the International School for Advanced Studies (SISSA/ISAS) in Trieste in 1997, and then moved to Philadelphia, USA, for a postdoctoral position at the University of Pennsylvania. In 2000 he was appointed Maıˆtre de Confe´rences at Universite´ Pierre et Marie Curie, where he is currently a full professor. A specialist of electronic structure A. Marco Saitta theory and ab initio calculations, his research activity has spanned from bulk semiconductors to graphene and nanotubes, to water and ices. His main interests are the exotic properties of molecular crystals, liquids and amorphous under extreme conditions of pressure and temperature, for which he has received in 2006 the Young Scientist Award from the European High Pressure Research Group. In recent years his research has opened up into more interdisciplinary fields, such as Earth sciences and geochemistry. He has authored more than 80 articles, including 1 Nature, 2 Nature Materials, 5 PNAS, and 18 Phys. Rev. Lett. He has been serving since February 2013 as Deputy Dean of the Physics Faculty of UPMC.

20048 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

Tools for reconstructing the beginnings of life ‘‘in silico’’ Prior to discussing the key recent achievements in the modelling of processes relevant to the origin of life, let us briefly overview

Rafał Szabla

Ernesto Di Mauro was born in Valmontone, Italy, in 1945. In 1967 he obtained his Degree in Biological Sciences from ‘‘Sapienza’’ University of Rome, Italy. In 1969 he joined the Department of Genetics (Seattle), as a postdoctoral fellow. Appointed in 1978 as an associate professor of Enzymology at the University of Rome, he has been a professor of Molecular Biology since 1987. His research interests Ernesto Di Mauro were centered on gene regulation, DNA and chromatin structure and topology and, at present, on the various aspects of the origin of life.

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

Perspective

the most frequently used computational techniques. The spectrum of available methods is very broad and the methods widely differ in the employed approximations and areas of applicability. Good quality computations are not easy and require extensive prior experience, understanding of all the physical approximations of the methods and an intimate knowledge of the studied processes. Using computational methods in a black-box manner is responsible for numerous studies which either use inappropriate methods or are incorrectly interpreted. Quantum chemical (QC) calculations can be primarily considered as methods which aim to achieve as complete as possible a physical description of the studied molecules, but are quite expensive in terms of computer resources. QC methods that do not utilize any parameters are known as ‘‘ab initio’’ or ‘‘first principles’’ methods. QC methods are typically used for small model systems. Nevertheless, since in prebiotic chemistry we often investigate properties and chemical reactions of small molecules, we can very often afford computations on essentially complete systems. This is of great advantage compared to applications of QC methods to, e.g., enzymatic reactions. QC calculations are sometimes likened to hypothetical energymeasuring experiments carried out at a temperature of 0 K, since we aim to construct point-by-point the potential energy surfaces of the studied molecules or chemical reactions.7 In other words, we calculate a set of relative electronic energies for a series of single configurations of the atomic positions of the studied systems. Other typical properties that can be derived from QC calculations are spectroscopic properties, like IR, Raman or UV-Vis band positions and intensities. Major limitations of QC calculations are difficulties to derive free energies due to entire lack of sampling of the Boltzmann distribution and uncertainties in the inclusion of solvation effects. Rather crude free energy corrections are estimated from the harmonic approximation (vibrational frequencies) and solvent is often represented using continuum models.8 Fortunately, reaction mechanisms (including the thermodynamic corrections) for typical non-enzymatic prebiotic reactions can be calculated quite reliably. Contemporary QC calculations achieve an impressive accuracy in the description of ground state properties of closed-shell

electronic structure systems. On the other hand, calculations on open shell systems, excited states etc. often remain challenging. For a review summarizing the meaning of QC computations and written for non-specialists, see ref. 9. Wave-function-based approaches represent the traditional family of QC methods. This group comprises some of the most widely used ‘‘all-purpose’’ theoretical chemistry methods, like the HF (Hartree–Fock, another abbreviation is SCF, self-consistent field) and MP2 (second-order perturbation theory)10 techniques, as well as the more specialized and highly accurate CI (configuration interaction) type methods, like CC11 (coupled clusters), MCSCF12 (multi-configurational SCF) and CASSCF13 (complete active space SCF). To obtain reliable results, an appropriate level of theory (inclusion of electron correlation effects) needs to be selected and combined with a sufficiently large basis set of atomic orbitals. The quality requirements vary with the nature of the chemical problem studied.9 The wave-function-based techniques are all common in that ¨dinger-equation14 to obtain information they solve the Schro about the geometry, energy and electronic properties of the studied system. The primary physical quantity provided by these computations is thus the electronic energy that is a function of the intrinsic molecular structural parameters: this enables studying reaction mechanisms by mapping the potential energy surface of the studied system as a function of the structural changes implied by the reorganization of the reaction complex. Another product of QC calculations is the electronic wavefunction (i.e., the molecular orbitals) which can be used to derive accurate information about the electronic structure of the studied systems. Other properties of the molecules are derived from the wave-functions using appropriate algorithms available in the main QC codes, using the basic principles of quantum mechanics. It is highly popular to discuss the properties of different molecules using the so-called partial atomic charge distributions derived from some QC computations. Here we wish to point out that derivation of atomic charges is always an arbitrary (definition-dependent) procedure. Atomic charges, despite being very intuitive, are not real physical quantities (i.e., observables)

Raffaele Saladino is a full professor of Organic Chemistry, Bioorganic Chemistry and Chemistry of Natural Substances at the University of Tuscia, Viterbo (Italy). He is involved in prebiotic chemistry, with particular attention to the development of plausible synthetic models for the emergence of nucleic acids, amino acids, lipids and sugars.

ˇponer is the head of the Jirˇ´ı S Laboratory of Structure and Dynamics of Nucleic Acids at the Institute of Biophysics, Academy of Sciences of the Czech Republic, and a Professor at Palacky University, Olomouc and Masaryk University, Brno. He has published around 260 papers with more than 12 000 citations. His research interests are computational and theoretical studies of structure, dynamics, function and evolution of nucleic acids.

Raffaele Saladino

This journal is © the Owner Societies 2016

Jirˇ´ı ˇ Sponer

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20049

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

according to a thorough quantum-mechanical definition. There exist a plenty of arbitrary methods for deriving partial atomic charges from the computed electronic densities that are extensively used in organic chemistry. The meaning of the fractional charges, however, should not be overvalued. The second class of basic QM methods is represented by DFT (density functional theory) methods. The DFT-based studies became, in fact, dominant in the last decade. The massive expansion of DFT-based computational techniques revolutionized almost all areas of computational modelling. These techniques are based on the Hohenberg–Kohn theorem,15 which states that the ground state electronic energy of a system is determined by the electronic density. Since the electronic density is also dependent on the structural parameters, DFT-techniques basically provide a powerful and fast method to get information about the structure and energy of the studied systems, at least in the electronic ground state. While selection of an appropriate wave-function method is straightforward though not always easy, an appropriate choice of the DFT method requires quite a lot of insights into the latest literature. There are several different levels of DFT approximation and literally hundreds of available variants of DFT functionals.16–22 Thus, we strongly advice that inexperienced users consult a specialized laboratory before executing such computations. An alternative option is to search through benchmark databases and method-testing papers for related chemical problems/reactions that are commonly published in this research field and are reliable.23–25 A properly chosen DFT method can provide results that are comparable and sometimes (due to appropriate compensation of errors) even better than those obtained by the wave-function approach, with a fraction of computational resources. The QC community and literature are usually very open about accuracy limitations of the methods. Classical atomistic molecular dynamics (MD) simulations are primarily aimed at describing thermal dynamics (Boltzmann distribution) of the studied systems.26 Assumption of the ergodic hypothesis means that infinitely long simulation would provide converged sampling of the phase space and thus correct thermodynamics. The extensive sampling comes at the expense of unphysical description of the molecules which are approximated by simple parameterized atomistic potentials (force fields) using van der Waals Lennard-Jones spheres, atom-centered fixed point charges and simple analytic functions to describe the covalent structure. Such potentials cannot include any explicit polarization or electronic structure effects; these need to be taken into account implicitly in the course of the parameterization. With prolongation of the simulations and advance of methods that allow enhancing sampling along the most relevant degrees of freedom,27,28 the highly approximate nature of the force fields is becoming the most crucial limitation. Despite the efforts to improve the biomolecular force fields it seems that their quality has reached a plateau and it is not clear if any fundamental improvement of their accuracy will be possible in the foreseeable future.26 MD simulations are based on describing the studied system by solving Newtonian equations of motion typically at room temperature with an integration time step of 2–4 femtoseconds.

20050 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

PCCP

The classical force fields do not allow studying chemical reactions, as they do not describe bond breaking and bond formation. In contrast to the usually rigorous QC literature, it is not uncommon in the MD literature to extensively hide simulation artifacts and over-interpret the results, although there are studies proving frank assessments of the limits.29–32 We strongly advise anyone attempting MD simulations to perform tests of the technique and of the selected force field using the available atomistic structures as the reference, to verify that the simulated molecules do not structurally degrade. Mere literature search of what other groups use may be insufficient in this particular case. There are intense efforts to develop more physical polarizable force fields.33–36 However, while specialized force fields for narrow sets of compounds can be quite well parameterized, it is not yet clear whether it will be, in the near future, possible to parameterize sufficiently balanced multipurpose biomolecular polarizable force field. The larger complexity of the polarizable force fields compared to the simple pair-additive force fields is a major challenge when diverse energy contributions need to be appropriately balanced. Recently, ‘‘first-principles’’ or ‘‘ab initio’’ QC molecular dynamics (AIMD) methodology has also been applied to problems of prebiotic chemistry. AIMD is a method combining a quantum treatment of the electronic degrees of freedom, with the numerical solution and time-integration of Newton classical dynamic equations on atoms, by making use of the quantumcalculated forces acting on the atoms included in the simulation. As in classical MD, the system then evolves in time, within the chosen statistical ensemble, most of the times the canonical number of particles–volume–temperature (NVT) or the number of particles–pressure–temperature (NPT) ensembles. The goal is to let the system evolve for as-long-as-possible trajectories which, within the ergodic hypothesis, allow determining the ensemble statistical averages of the relevant physical observables of the systems, such as the thermodynamic (enthalpy, free-energies), structural (pair-correlation functions, structural factors), and dynamic (self-diffusion) properties. AIMD trajectories are generally of the order of tens of picoseconds, with typical timesteps of the order of 104 ps, thus implying that the quantum-derived forces and energies of the system must be calculated 105–106 times to obtain such trajectories (for comparison, classical MD simulations of solvated biomolecules are nowadays routinely done on a ms time scale26). These performances are achieved by using a densityfunctional theory (DFT) treatment of the electronic degrees of freedom, as wave-function-based QC methods are computationally too costly for these system sizes. In the general chemistry context, AIMD allows following the complete chemical evolution of the system, i.e. the rupture and formation of chemical bonds, according to the chosen thermodynamic simulation conditions. In the prebiotic case, for example, one can target specific reactions in different situations, such as in bulk solution or in the presence of a specific mineral surface, under given pressure–temperature hydrothermal conditions, and so forth. However, the predictive power of AIMD in these cases is often hindered by large barriers. These are typically too high to be spontaneously overcome by the system,

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

(i.e., only thanks to the thermal energy), within the short simulation times. To speed up the trajectory evolution, new methods are being developed to accelerate dynamics and, more generally, to address the free-energy calculations, among which the so-called umbrella sampling37 and metadynamics method38 are widely employed. Note that these enhanced-sampling methods are highly popular also in the field of classical MD simulations (see above). For the sake of completeness, let us mention that the combination of quantum chemical and molecular dynamics/ mechanics (QM/MM and QM/MD, i.e., hybrid) methods is so far less popular in the origin of life field, in contrast to studies of enzyme reaction mechanisms.39 In these methods, a small part of the molecule of chemical interest is treated via QC and the rest by MM. The unphysical border between the QC and MM regions is a common source of inaccuracies.

Prebiotic synthesis of nucleic acid building blocks Nucleobases Out of all nucleic acid building blocks perhaps prebiotic synthesis of nucleobases has attracted the greatest attention among both experimentalists and theoreticians. The pioneering experimental studies by Oro40 and Ferris et al.41 soon after the Miller–Urey experiment42,43 have shown that nucleobases can be synthesized from simple precursor molecules, like HCN, urea or cyanoacetylene. Indeed these reactions were exploited in the experimentally reported synthetic procedures, among which pentamerization of HCN leading to adenine has also been studied in detail by QC. Since experiments have clearly shown that pentamerization of HCN proceeds via 4-aminoimidazole5-carbonitrile (AICN) the theoretical study of Roy et al. has concentrated on the description of the conversion of AICN to adenine via addition of HCN, which is the crucial, rate determining step of the mechanism.44 (Scheme 1) They show that the otherwise kinetically unfavorable reaction steps may proceed with significantly lower activation energies if catalytic water or ammonia molecules are involved in the chemical transformations. Another study by Glaser et al. has shown that monocyclic HCN pentamers readily undergo cyclization leading to adenine with surprisingly low activation barriers, which make this pathway relevant also in an extraterrestrial environment.45 Photochemical steps in the formation of nucleobases are overviewed in a separate section devoted to prebiotic photochemistry.

Perspective

Formamide, formally the hydrolysis product of HCN, is considered a thermodynamically more stable and thus lessreactive variant thereof.48 It has been suggested that formamide could accumulate on the slowly cooling surface of the hot early Earth as the thermal dissociation product of ammonium-formate, formed in the reaction of NH3 and formic acid.6 Recently, a DFTbased AIMD study reported on the first simulation of a Miller-like experiment.49 In this work, a liquid phase, constituted by simple molecules such as water, ammonia, methane, carbon monoxide and molecular nitrogen, was simulated under ambient thermodynamic conditions but under strong uniform electric fields, which triggered the formation of small organic molecules, such as formic acid and formamide. The latter one, in particular, seemingly plays a ‘‘hub’’ role, being either broken into water and HCN or CO and NH3 and, when reformed, being an intermediate to the field-induced formation of larger molecules. In particular, it is able to form a C–C bond with formic acid resulting in hydroxyglycine, which later evolves to glycine. A follow-up of this study focused on the development of a novel, general computational method, based on metadynamics,38 which allows an unbiased exploration of the chemical reaction network of a given system, and/or the unbiased determination of the most-convenient free-energy pathway of a given chemical reaction.50 Furthermore, this approach allows treatment on the same footing both gas-phase and condensed-phase reactions. The study used the formamide dissociation/recombination to HCN + H2O as a test-case, showing that in the solution under ambient conditions reactants and products have comparable stability, with formamide being sizably more stable at higher temperatures (400 K), and thus suggesting that the formamide and the hydrogen cyanide prebiotic scenarios might be compatible. Experimental studies by Saladino and coworkers demonstrated that in the presence of meteoritic materials all four nucleobases can be synthesized from formamide.51,52 While the synthesis reported in ref. 51 utilized thermal energy as an energy source, the chemistry described in ref. 52 was triggered by protonirradiation, relevant to cosmic rays. The thermal mechanism of the reaction was extensively studied using QC calculations.53–56 The main steps of the theoretically proposed reaction pathways are summarized in Scheme 2A and B. During the last 12 years, Civisˇ et al. published a series of studies devoted to studying high-density-energy events as a possible source of energy in prebiotic synthesis as well as extraterrestrial chemical processes.57–60 Recently, using ultra-high

Scheme 1 Pentamerization of HCN leads to adenine.46,47 Computations by Roy et al. have shown that the addition of HCN to AICN might be catalyzed by water or ammonia molecules.44

This journal is © the Owner Societies 2016

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20051

View Article Online

PCCP

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

Scheme 2 Nucleobases from formamide. (A) The thermal route via pyrimidine nucleobases as suggested in ref. 5, 56 and 68. (B) A concurrent idea of the thermal pathway assuming the formation of 2-iminoacetonitrile as the introductory step of the pathway.53–55,69 (C) The reaction route by Ferus et al. utilizes high-energy activators ( NH2 and  CN radicals) to trigger the formation of nucleobases in a clearly exergonic fashion.61 (D) A radical-route via 2-iminoacetonitrile by Jeilani et al.70

energy laser experiments, Ferus et al. convincingly demonstrated that all four nucleobases present in RNA can be formed from formamide in a high-energy density event, like the impact

20052 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

of an extraterrestrial body.61–63 This work outlined a chemistry, which could be highly relevant to the early Earth, because during the Late Heavy Bombardment period ca. four billion years ago,

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

i.e. at the time when terrestrial life emerged, our planet likely witnessed extreme impact activity, which endured for hundreds of millions of years. The experiments presented in this work support the view that high-energy precursors ( CN and  NH2 radicals) formed in an extraterrestrial impact could induce a cascade of chemical transformations that ended up with the formation of nucleobases. In other words, the high-energydensity event first leads to a violent radical chemistry64,65 and a destructive decomposition of the molecules, as expected. However, then in the cooling plasma a very rich creative chemistry can occur. In other words, the extraterrestrial impacts may be not only destructive, but also lead to easy spontaneous synthesis of a rich spectrum of prebiotic molecules. Specifically for the synthesis of nucleobases from formamide, these transformations could take place both in small reservoirs concentrating liquid formamide on our planet and in the crust of a falling extraterrestrial body when reaching the Earth’s atmosphere. The latter scenario might provide a plausible explanation for the formation of nucleobases detected in meteorites.66,67 It suggests that nucleobases may form during the impact from small molecular precursors of extraterrestrial origin and do not need to be imported from the space. In addition, it also explains the presence of nucleobases and other complex molecules deep inside the impacting species such as the Murchison meteorite. The formation of large meteorites and asteroids through accretion is necessarily associated with countless high-energy-density events and under appropriate composition of the asteroid materials will be necessarily associated with synthetic processes. The work by Civis et al. is thus related not only to possible prebiotic scenarios on the early Earth, but also to high-energy-density impacts anywhere in the space. Essential for understanding this chemistry is to take into consideration the fact that it occurs under highly non-equilibrium conditions, where the high-energy-density impact first decomposes the material and the accumulated energy is then used for synthesis in the highly variable process of cooling of the environment. The synthetic processes so far detected experimentally in the high-energydensity events thus may be just the tip of the iceberg of real chemistries that can be achieved, since an asteroid impact is an incomparably more complex and powerful process than a terawatt laser pulse. The temporal and spatial dimensions of the radical-containing plasma cooling processes will be orders of magnitude larger than in laser pulses, with enormous potential for material mixing. The high-energy synthesis of nucleobases described in ref. 61–63 is a textbook example of the complementary role of experiment and theory. The computations were instrumental to propose a plausible pathway of the reaction (see Scheme 2C) and were also essential in interpretation of the spectra. The viability of the theoretical model was justified by the fact that two of the proposed reaction intermediates of this multistep pathway have been identified in the reaction mixture based on their signatures in the experimentally measured highresolution infrared spectra. Note that the computations readily identify other important intermediates that are not directly detectable by the experiments due to their short life times.

This journal is © the Owner Societies 2016

Perspective

Thus, in fact, the resolution of the QC method is higher than that of the experimental spectroscopy. The computations have shown that the great advantage of the impact synthesis over the thermal route is that due to the involvement of high-energy activators the chemical transformations are exothermic and the activation energies are much lower than those of the analogous thermal pathways. Further mechanistic proposals for the synthesis of nucleobases from formamide utilizing radical chemistry have been presented by Jeilani et al. in ref. 70–72 (see Scheme 2D). Sugars According to the generally accepted view sugars can be synthesized from formaldehyde via the so-called formose reaction (see Scheme 3).73,74 The formose reaction is a classical aldoladdition, which implies that one of the reactants enolizes prior to the attack on the carbonyl group of the other reactant. This excludes the possibility that two formaldehyde molecules can dimerize on their own according to the aldol-mechanism. Thus, the prerequisite for the success of this synthetic scenario is that glycolaldehyde, at least in trace amounts, is present in the reaction mixture.75 To form glycolaldehyde under plausible prebiotic conditions has been for a long time a challenging task for the origin of life studies and only a couple of proposals have been put forward in the literature. Perhaps the most popular one assumes that glycolaldehyde, an ubiquitous molecule in the space, was delivered to the Earth from the space.75 Others suggest that it could be formed in low concentrations via UV irradiation of formaldehyde76 or in the reaction of CH4 and CO2.77 Recently an HCNbased photochemical synthetic route has been suggested for the synthesis of glycolaldehyde, which will be discussed in detail in the photochemical part of this review.78 The latest study demonstrated that glycolaldehyde could have been synthesized in the course of impacts of extraterrestrial bodies on the Earth.79 There exist numerous experimental reports on efficient catalytic syntheses producing sugars from formaldehyde, nonetheless,

Scheme 3

The formose reaction.

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20053

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

PCCP

Scheme 4 Stereoelectronic reasons make the N-glycosylation of ribose endothermic.86 Fig. 1 Optimized geometry of the ribose–borate 2 : 1 complex from QC calculations (DFT B3LYP/6-311++G** calculations). The strongly stabilizing H-bond between the 3 0 -hydroxyl of ribose and one of the anionic oxygens of the borate is highlighted with a blue dotted line.82

the exact experimental mechanism has been deciphered only in the case of the borate-assisted reaction route.80 The tremendous potential of prebiotic borate chemistry lies not only in the fact that borate minerals could sequester aldopentoses from the prebiotic mix.75,81 Complexation by borate minerals provides a clue to understand why among the four aldopentoses ribose has been selected to be the component of the first genetic materials. The studies of Benner et al. suggest that ribose forms the most stable borate complexes,75 thus it could accumulate on the early Earth in the highest concentration. QC calculations have suggested that a highly polarized intramolecular H-bond formed between the anionic oxygen of borate and that of the 3 0 -hydroxyl of ribose is responsible for the distinct stability of the complex formed between ribose and borate-minerals (see Fig. 1).82 Logically, analogous silicate complexes of aldopentoses have also been studied both experimentally83 and using computations.84 Nevertheless, in these studies the preference for stabilization of ribose was not as clear as in the case of borate-complexes. Nucleosides In the beginning of the new millennium Zubay and Mui wrote in their critical assessment85 of the RNA-world theory: ‘‘The one area where little progress has been made is on joining ribose to the nitrogenous bases.’’ Physico-chemical reasons for this failure are rather obvious: the syntheses of ribose and nucleobases from small molecular precursors are markedly exothermic processes, i.e., both molecules are highly stable and not apt for further stabilization in the form of N-glycosides.86 A natural bond orbital (NBO)-analysis87 of the electronic wave-function unraveled the underlying reason for this. It has been shown that, due to electron delocalization over the aromatic ring of the nucleobase, nucleosides lack the stabilizing n - s* hyperconjugation from the glycosidic N atom to the C10 –O4 0 antibonding orbital, which leads to their partial destabilization with respect to free ribose (see Scheme 4).86 Nonetheless, as experimental studies also suggest, the N-glycosylation might be exothermic if it is preceded by a phosphorylation step, because phosphorylation prevents the stabilizing n - s* hyperconjugation in ribose-1 0 -phosphate. Another novel possibility of forming nucleosides is the formose-like condensation of formaldehyde and nucleobases

20054 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

in a formamide-environment in the presence of UV-light and a TiO2-catalyst.88 The experimentally suggested mechanism88 involves TiO2-catalyzed conversion of formamide to formaldehyde triggered by UV-light which is followed by N-formylation of the base and subsequent formose-like chain extension steps.88 Albeit the N-glycosylation reaction has not yet been studied by computational methods, a recent work89 offers a plausible theoretical model for the conversion of formamide to formaldehyde over TiO2. More recently, it has been shown that formamide irradiated with slow protons in the presence of meteoritic materials also leads to the formation of nucleosides.52 The reaction most likely involves a radical chemistry, and might serve as an exciting topic for future computations. From nucleosides to nucleotides in formamide Phosphorylation of nucleosides is essentially a condensation reaction, and, as such, is obviously thermodynamically highly disfavored in the presence of excess water molecules. In addition, most of the phosphate minerals exhibit low solubility in water, or if they are solvable, they are kinetically inactive. Introducing formamide medium in the synthesis of prebiotic building blocks solved also the problems related to the phosphorylation of nucleosides in an aqueous environment. It has been suggested that phosphate minerals are solvable in formamide in the form of metaphosphates (see Scheme 5) which then readily react with nucleosides leading to variously phosphorylated variants.90–92 Although metaphosphates could have never been detected experimentally, ref. 90 puts forward the idea that formamide, in contrast to water, may stabilize metaphosphates via complex formation. Despite its high relevance, this ‘‘classical’’ problem of prebiotic chemistry has never been addressed by computations. AIMD-simulations could provide an answer to this question. Aminooxazolines: a shortcut to nucleotides The notion that both the glycosidation and phosphorylation steps are highly disfavored in water motivated the efforts to find

Scheme 5

The metaphosphate anion.

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

Scheme 6

Perspective

Main steps of the synthesis of cytidine-2 0 ,3 0 cyclic phosphate suggested by Powner et al.93

synthetic pathways diametrically different from the genealogical approach used for decades in prebiotic chemistry, which tried to build up nucleotides from their constitutional units, i.e. nucleobases, ribose and phosphates. The synthetic pathway proposed by the Sutherland-group93,94 derives nucleotides from simple organic precursors, like glycolaldehyde, glyceraldehyde, cyanamide and cyanoacetylene, and from inorganic phosphates (see Scheme 6). A computational QC analysis86 of the free-energy profile of this synthetic pathway shows similarities with that of modern biochemical reactions, i.e., in contrast to the genealogical approach, this synthetic pathway includes smaller energetic steps and proceeds in an exergonic fashion. In addition, the initial and concluding steps of the pathway are clearly exergonic, i.e., the reaction is ‘‘pulled’’ toward the products, cyclic nucleotides, which can accumulate in the prebiotic pool. As Benner emphasized this is a common feature of modern biochemical pathways.95 The mechanism of the reaction was analyzed computationally in ref. 96 and 97. It has been shown that phosphates – in addition to being an obvious precursor needed to synthesize nucleotides – play a unique catalytic role in the nucleotide synthesis. Computations have shown that the amphoteric character of phosphates enables them to participate as acid– base catalysts in the key-step associated with the formation of 2-aminooxazole (see Fig. 2). Let us note that phosphates are the only ubiquitous anions existing in nature with oxidation states varying from 0 to 3, which could also play a role in their involvement in those chemical transformations that led to the emergence of life on our planet. Since carbonates are even more ubiquitous in the Earth’s crust than phosphates and also exhibit an amphoteric character it would be interesting

This journal is © the Owner Societies 2016

Fig. 2 Phosphate catalysis in the rate-determining step of 2-aminooxazole formation, i.e., the cyclization reaction.

to see by future experiments whether they exhibit the same catalytic effect. Quantum chemical calculations have shown that the regioselectivity of the phosphorylation reaction of the anhydroarabinonucleoside intermediate is governed by a stabilizing n - p* hyperconjugation between the O5 0 of ribose and the C2QN3 linkage of cytosine.98 The astonishingly rich photochemistry triggered by UV irradiation undoubtedly available on the early Earth is also an important part in Sutherland’s nucleotide synthesis model and is discussed in detail in a separate section of this review.

Photochemical processes in the synthesis of prebiotic building blocks The total amount of UV-B and and UV-C radiation that reached the surface of early Earth in the Archean age was much higher than nowadays due to lack of the ozone layer99,100 and higher activity of the young Sun in the ultraviolet range.101 Exposure of

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20055

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

the early Earth’s chemistry to sunlight resulted in the significant enrichment of molecules that are exceptionally resistant to the deleterious effects of UV irradiation.102 Therefore, the remarkable photostability of nucleic acids103,104 and peptides105,106 has for long been considered as an important molecular fossil indicating the invaluable role of UV radiation in the early stages of abiogenesis.107–109 Ultraviolet light is also an important source of energy that often promotes selective reactions which would otherwise require significant amounts of heat or the presence of a specific catalyst. A thorough understanding of the underlying photochemical reaction mechanisms can be essential for finding better variants of the known prebiotically plausible reactions and might enable their validation. Both theoretical and experimental approaches can serve this purpose, providing different sorts of information. Usually the most challenging problems can be tackled by a synergistic approach, which allows assessing the results from theoretical and experimental perspectives simultaneously. Experiments ranging from NMR spectroscopic studies of reaction products to time-resolved ultrafast measurements of excited state dynamics may provide a lot of useful information about the underlying elementary photophysical and photochemical processes. NMR spectroscopy proved to be particularly useful in conjunction with isotopic labeling or when the substrates were irradiated in D2O.110 In the latter case incorporation of deuterium from the environment could indicate a direct involvement of water in the photorelaxation mechanism or a possible photoinduced hydrogen atom transfer or abstraction processes.110 In contrast, ultrafast time-resolved spectroscopic approaches provide data about the evolution of photochemical reactions on a femtosecond timescale.108,111 These methods proved particularly useful in investigations of gas phase photochemistry of isolated molecules and clusters,108,111,112 but recently more and more ultrafast spectroscopic experiments are conducted on fully solvated compounds as well.113,114 High-level ab initio simulations facilitate the interpretation of the experimental data and often provide answers to questions that cannot be approached using spectroscopic methods alone. The central properties studied by computational photochemistry are conical intersections between two or more electronic states of same multiplicity.115 These crucial points on the potential energy surfaces are of comparable importance as transition states for ground-state chemistry and enable the identification of the existing photoreaction channels.115 For this purpose, multiconfigurational self-consistent field (MCSCF)12 approaches with second order perturbation theory correction (e.g. CASPT2)116,117 and multireference configuration interaction (MRCI)12 yield a particularly reliable description of excited-state potential energy surfaces, and state crossings. However, the multireference and multiconfigurational approaches often require significant computational resources and depend on the arbitrary selection of active spaces.118,119 Coupled-cluster based approaches like CC2120 and ADC(2)121 are usually very robust and efficient alternatives,122,123 especially in conjunction with resolution-of-the-identity approximation (RI).124 ADC(2) was successfully used for simulations of biomolecular building

20056 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

PCCP

blocks and showed relatively good performance when compared to other computational methods and experiments.125 It should be noted though that ADC(2) and CC2 are slightly less accurate than CASPT2 in predicting excitation energies and are not applicable for systems with clearly multiconfigurational character of the ground state.126 Special caution needs to be taken with respect to the popular time-dependent density functional theory (TDDFT) computations, which are highly dependent on the choice of the functional and generally should be benchmarked for specific applications against higher level ab initio methods.118 For many decades, UV-light has been considered as an important element of synthetic experiments simulating conditions that conduced abiogenesis. One of the first and most prominent examples is the series of studies by Ferris and co-workers related to pentamerization of HCN leading to adenine, which appeared between 1966 and 1974.46,47,127–129 The multistep photoisomerization of diaminomaleonitrile (DAMN) to AICN is one of the key steps of this reaction sequence.128 Even though the scenario proposed by Ferris was a milestone in studies of the origins of RNA, some drawbacks can be pointed out including the necessity of considering HCN ices as the reaction environment130 and the lack of efficient pathway leading from nucleobases to nucleosides (owing to both kinetic and thermodynamic reasons).4 A notable modification of this pathway was published in 2010 by Barks et al.,131 who showed that the same reaction can be conducted in neat formamide and formamide solutions yielding guanine and hypoxanthine as additional products apart from adenine. Mechanistic details of the above reactions were investigated theoretically. Barbatti et al.130 proposed a full reaction mechanism of the multi-step DAMN to AICN photoisomerisation based on non-adiabatic molecular dynamics simulations. This reaction pathway is also consistent with some of the mechanistic suggestions of Ferris et al.128 The initial DAMN to diaminofumaronitrile (DAFN) photoisomerisation mechanism was also described in detail by Szabla et al.,132 and was shown to occur on the singlet hypersurface and not with the participation of triplet states as formerly postulated by Ferris et al.128 An interesting alternative to the photoinduced HCN oligomerization involved photocatalytic formation of all five nucleobases from formamide on the reactive TiO2(001) surface.133 In 1970, Sanchez and Orgel published a prebiotically plausible and indirect photochemical route leading to pyrimidine nucleosides which did not involve the endergonic nucleobase glycosylation.134 The synthesis assumed the formation of aminooxazoline from D-ribose and cyanamide, and the further formation of a-ribocytidine after the reaction of aminooxazoline with cyanoacetylene. Irradiation of alpha-ribocytidine resulted in partial photoanomerisation to the biologically relevant b-cytidine, with a relatively low yield of just 4%.134 Even though the reaction was criticized because of problems in obtaining a prebiotically plausible pathway to ribose135 and low reaction yield, it became an important inspiration for a series of studies published by Sutherland and co-workers which eventually resulted in a very elegant synthesis of pyrimidine nucleotides from small feedstock molecules.93,136–140 While the former

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

studies considered the ways of enhancing Orgel’s134 photoanomerisation reaction,139,140 the latter and most successful contribution proposed a different role for UV light.93 Namely, UV irradiation purified the product mixture by destructing the biologically irrelevant stereoisomers of pyrimidine nucleotides. In addition, it enabled the partial conversion of cytidine to uridine.93 Szabla et al.110 investigated the photoanomerisation mechanism of 2 0 -deoxycytidine which evinced somewhat similar photochemistry to its ribo-counterpart (see Fig. 3).139 Joint theoretical and experimental efforts led to the conclusion that the photoanomerisation of 2 0 -deoxycytidine is triggered by the excited-state C1 0 –H atom abstraction by the carbonyl group of cytosine.110 It is tempting to assume that a similar mechanism could explain the photoanomerisation of ribocytidines, however, the presence of the 2 0 -OH group may significantly alter the photochemistry of a nucleoside. This is reflected by the formation of oxazolidinone in the products of Sutherland’s synthesis,93 and the ultrafast intramolecular electron-driven proton transfer mechanisms proposed for adenosine and 8-oxoguanosine.141,142 Therefore, the apparently more complex photochemistry of a-ribocytidine is currently under exploration. Furthermore, efficient prebiotic routes to pyrimidine nucleosides (and not only nucleotides) and their purine counterparts are among the most prominent challenges that can be addressed in this topic. Initially, one of the major uncertainties of Sutherland’s synthesis of pyrimidine nucleotides was the origin of an important substrate and the simplest sugar, i.e. glycolaldehyde. For many years, it was anticipated that sugars on the early Earth could have been formed from formaldehyde via the basecatalysed formose reaction.74 As mentioned before, C–C bond formation between two formaldehyde molecules is prevented by the inherent polarity of the carbonyl group and thus the formose reaction does not operate without at least trace amounts of glycolaldehyde early on. A plausible synthesis of glycolaldehyde from one-carbon substrates was proposed in 2012, and involved photoredox cycling of copper cyanide complexes in the presence of HCN.78 The mechanistic rationale for this reaction is based on the generation of cyanogen and hydrated electrons among the

Perspective

initial photoproducts. An alternative mechanism was derived from DFT and MP2 simulations and considered the possible role of dipole bound anions.143 However, this photochemical reaction pathway requires further validation with MCSCF-based approaches, and a direct comparison to the mechanism proposed by experiments. The above-mentioned reaction pathways often contain intermediates that are supposed to accumulate in the environment over longer periods of time. Therefore, these compounds should evince resistance to different environmental conditions of early Earth including UV radiation. Probably the most distinctive intermediates that appeared in the prebiotically plausible reaction pathways described above are AICN128 and 2-aminooxazole (a precursor of pyrimidine nucleotides in Sutherland’s suggested synthesis).93 The photostability of AICN was observed by Ferris,128 however, there is no experimental evidence for the photostability of 2-aminooxazole. Gas phase ab initio calculations identified the major photodeactivation mechanisms of 2-aminooxazole and AICN as N–H bond fission and ring-puckering processes, driven by ps* and pp* electronic states respectively.144,145 Furthermore, non-adiabatic molecular dynamics simulations of 2-aminooxazole revealed a relatively large contribution of the potentially photodestructive ring-opening channel, which implies that further studies are necessary to elucidate the photochemistry of this molecule.146 When investigating the photochemistry and photostability of biomolecules it is crucial to look at the environmental effects exerted on the chromophores that could potentially modify the photorelaxation mechanisms. Even though it is tempting to apply one of the widely used continuum solvation models, such an approach does not take into account the possible direct involvement of strongly interacting solvents (e.g. water). For instance, in the case of chromophores with low-lying ps* electronic states, water significantly alters the N–H bond fission mechanism. Photoexcitation of indole, phenol, AICN and 2-aminooxazole clustered with several water molecules results in the ejection of one electron in the direction of the solvent molecules (also referred to as charge transfer to solvent).145–148 The electron may be then followed by a proton from an

Fig. 3 The mechanism of the photoanomerisation and nucleobase loss reactions of 2 0 -deoxycytidine triggered by excited-state hydrogen atom abstraction by the carbonyl group of cytidine.

This journal is © the Owner Societies 2016

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20057

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

amino (or hydroxyl) group of the chromophore leading to the formation of a complex containing a radical of the deprotonated molecule, the hydronium cation and a hydrated electron.148 Subsequent proton transfers along H2O wires leading to the recombination of the migrating proton with the hydrated electron may enable photodeactivation via a conical intersection with the electronic ground-state.145,146 This was shown for both prebiotically relevant intermediates, AICN and 2-aminooxazole.145,146 Another photorelaxation mechanism involving the direct participation of solvent molecules is the water-to-chromophore electron transfer reported by Barbatti for microsolvated adenine.149 It is worth noting that these mechanisms would not be observed in conventional QM/MM simulations treating all the water molecules surrounding the chromophore at the molecular mechanics level.

Formamide, water, or both? The role of water in prebiotic chemistry is a controversial topic. Even though water is supposedly the environment in which life originated, chemical precursors such as formamide and hydrogen cyanide (and their derivatives) might be degraded in water, decreasing the effectiveness of their role in the synthesis of biomolecules.150 A plausible way for the formation of formamide on the early Earth could be the hydrolysis of HCN. HCN is a gas at ambient pressure and temperature. Once absorbed in water it undergoes two possible processes depending on pH and concentration: (a) direct condensation to biomolecules, like adenine, or polymerization to poly(hydrogen cyanide) derivatives that may be transformed further into biomolecules or (b) the hydrolysis by addition of H2O on the CN triple bond to give formamide. These two competitive processes show a similar kinetics at relatively high HCN concentrations, in the range between 0.01 and 0.1 M, and at alkaline pH (the optimal value being between pH 8 and 9). The hydrolysis of formamide prevails in more dilute solutions.47 A theoretical study on the possible concentration of HCN on early Earth suggested that the steady state concentration of HCN in the primitive ocean was 4  1012 M at pH 7 and 100 1C, with a slight increase of concentration at lower temperature (2  105 M at pH 7 and 0 1C).151 Similar results were obtained for formamide. These concentration values appear to be too low for the formation of the appropriate amounts of biomolecules suitable to support the emergence of life. On the other hand, alternative mechanisms exist for the solution of this problem, as for instance, the processes based on the colligative properties of the two compounds. HCN and formamide might have accumulated through the formation of an eutectic phase with water at low temperatures, as reported for HCN, or by evaporation processes, a phenomenon that affects mainly formamide, which is characterized by a high boiling point (4200 1C) without azeotropic effects with water.152 Furthermore, concentration processes might have been favored by the presence of minerals, considering as an example the high efficiency of absorption of formamide in clays of the montmorillonite family.153 Thermoconvention processes

20058 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

PCCP

can also concentrate formamide. Thermal condensation of mixed formamide/water in the presence of iron sulfur and iron–copper sulfur minerals yields a significant panel of compounds, including two nucleobases (adenine and cytosine) and one carboxylic acid (oxalic acid). The amount of products decreased by increasing the water content, only few compounds being detected in the presence of 30 wt% water154,155 The synthesis is more efficient in the presence of meteoritic materials, yielding a larger panel (in larger amount) of biomolecules (data on submission for publication). Another plausible way for the accumulation of formamide on the hot surface of early Earth is the thermal dissociation of ammonium-formate at B180 1C. This temperature is significantly higher than the boiling point of water and therefore the latter synthetic way could lead to the accumulation of an essentially water-free formamide medium.6 Recent progress of the AIMD simulation techniques enables an explicit quantum chemical treatment of solvent molecules, which might revolutionize the understanding of the formamidechemistry. Such calculations could thus be instrumental to provide with the long-sought answer to the question: how does chemical reactivity change with the formamide/water ratio?

Creating the first templates Modern biochemical machineries utilize templates to form oligonucleotides, but how did nature create the very first templates without enzymes from nucleotide building blocks? This has been one of the most intriguing questions of the origin of life research for decades. Historically, the first proposals for template-free non-enzymatic oligomerization of nucleotides came from the 70’s and 80’s of the last century. The recognition that cyclic nucleotides (see Scheme 7A) are readily formed upon phosphorylation of nucleosides in the presence of urea156 fostered the first efforts to use them as precursors to synthesize oligonucleotides.157 The clearly exothermic character of the hydrolysis of cyclic nucleotides158 demonstrated that cyclic ring formation accumulates energy which can serve as the driving force for transphosphorylation reactions leading to oligonucleotides. Early synthetic approaches by Verlander and Orgel157 as well as by Usher and Yee159 utilized

Scheme 7 The phosphate group of nucleotides may be activated (A) in the form of cyclic ring formation or activation via cyclization or (B) by the formation of phosphoimidazolides.

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

these chemistries to create short oligonucleotide sequences. Later studies suggested that higher oligomerization yields can be achieved when incorporating a phosphate-activation step in the synthetic scheme prior to the polymerization (see Scheme 7).160 This motivated a series of studies utilizing imidazole to activate the phosphate group of nucleotides for the polymerization reaction.160 More recently, the success of demonstrating plausible prebiotic routes to cyclic nucleotides5,92–94 shifted the attention of experimentalists again to cyclic nucleotides. Oligomerization of 3 0 ,5 0 -cyclic guanosine monophosphate (GMP)161–164 represents so far the only known nonenzymatic template-free polymerization method, which yields selectively 3 0 ,5 0 -linked oligonucleotides. Let us note here that in the literature experiments of the Braun group163 have been intensively used to refute the experimental results in ref. 161 and 162. However, the detection limits of the fluorescence labelling method used by Braun et al. in ref. 163 were clearly not sufficient to detect the oligomers reported in ref. 161 and 162 (see ref. 6 for more details). The first mechanistic studies on the stereochemistry of the transphosphorylation reactions leading to oligonucleotides were based on a thorough analysis of crystal geometries of cyclic nucleotides. Usher and Yee in their seminal study demonstrated that oligomerization of 2 0 ,3 0 -cyclic nucleotides involves an in-line attack of the 5 0 -hydroxyl of the ribose at the phosphate of the other reactant. Since the nucleophilic attack at phosphorus is sterically not hindered, both 2 0 ,5 0 - and 3 0 ,5 0 -linkages may form in the reaction.159 A major step forward in uncovering the general principles of the chemistry that drives the template-free non-enzymatic oligomerization came from recent QC calculations. These studies suggested a unique anionic ring-opening polymerization mechanism leading to 3 0 ,5 0 -linkage selectivity in the oligomerization of 3 0 ,5 0 GMPs (see Fig. 4).164 Analogous ring-opening cyclic polymerization reactions of cyclic phosphate and phosphonate esters were documented in ref. 165 and 166. The QC model shows that the reaction is optimal under anionic conditions while no efficient reaction pathway was found under

Perspective

neutral conditions. The model164 thus reflects the experimental observation that the efficiency of the reaction decreases with increasing concentration of highly mobile cations, like Na+, while it is totally unaffected in the presence of slowly moving bulky alkylammonium cations. Although the QC technique is robust enough to capture the basic intrinsic electronic structure properties of different potential reaction pathways assuming certain reactants and their configurations, QC modelling of fine environmental effects of biologically relevant reactions is not easy with the presently available techniques (see above). With the available methods and hardware, we are not able to simultaneously achieve a bruteforce description of (i) an efficient treatment of hydrated cations distributed in the vicinity of reaction complexes and (ii) an accurate description of the reaction pathways in reaction complexes consisting of ca. 200 atoms. In the near future, emerging large-scale QC modelling methods presented, for example, in ref. 168–170 may provide a solution for the above problem. Most likely, these problems will require the development of case-bycase multiscale approaches where the cation distributions will be first probed by classical atomistic simulations. Extended classical MD can reveal the distribution of the ions depending on their concentration and identify the main ion-binding sites coupled with the solute structural dynamics.171–173 Then a series of potential reactive conformations suggested by atomistic simulations can be investigated by large-scale QC or hybrid quantumclassical methods, in a way reminiscent of studies of catalytic mechanisms of ribozymes.174–177 Note that modern calculations are capable of unambiguously identifying the reaction pathways for some efficient enzymatic reactions that are precisely geometrically tuned to every single atomistic detail. However, the calculations are not always sufficient to obtain an exact single atomistic reaction pathway for systems where more atomistic pathways with similar reaction profiles co-exist and even in reality contribute to the reaction. This is likely to be the case of ribozymes as well as of many prebiotic chemical reactions that are more loosely structurally controlled.

Fig. 4 Stacking interactions are responsible for the 3 0 ,5 0 -linkage selectivity of the anionic ring-opening polymerization of 3 0 ,5 0 cyclic GMPs. Dispersioncorrected DFT calculations using the DFT-D2 formalism167 developed by Grimme helped to unravel the reaction mechanism. Computations in ref. 164 have shown that in the chain-propagation step the 3 0 -deprotonated oxygen of a 5 0 -GMP is optimally positioned to initiate an in-line attack at the next cyclic phosphate of the stacked supramolecular architecture. Thus, the genuine self-assembly of 3 0 ,5 0 cyclic GMP is sufficient to arrange the molecules in such a way that they can reach highly reactive configuration essentially from the ground state configuration without any substantial rearrangements.

This journal is © the Owner Societies 2016

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20059

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

However, the calculations can decidedly prove if a reaction is electronically feasible, i.e., if the system is ready to react under appropriate reactant arrangements. Another comparative theoretical study addressed the mechanism of all known non-enzymatic template-free oligomerization scenarios, which utilize non- or mildly activated precursors (i.e. nucleotides or cyclic nucleotides).178 It has illuminated that the role of simple amines in the synthesis is to activate the phosphate via proton-transfer processes. The study has shown that basically the same principle of activation is relevant to the recent experimental work by Deamer and Maurel, who report that adenosine- and uridine 5 0 -monophosphates efficiently co-oligomerize in a highly acidic hydrothermal environment (see Fig. 5).179 Non-enzymatic, template free oligomerization of imidazoleactivated oligonucleotides over montmorillonites was demonstrated by the Ferris group.180 The reaction has been extensively studied using computational methods as well. Mignon et al. provided a detailed QC characterization of the adsorption of nucleobases onto the external surface of montmorillonite using periodic plane-wave calculations.181–183 This, otherwise very promising approach, due to size limitations could not be so far applied to describe the transphosphorylation reaction leading to the phosphodiester linkage. Ferris et al. reported that the montmorillonite-catalyzed oligomerization of nucleosidephosphoimidazolides exhibits a remarkable homochiral selectivity.184,185 Classical MD simulations by Matthew and LutheySchulten have shown that activated nucleotides preferentially

PCCP

form homochiral supramolecular assemblies over the montmorillonite surface prior to the oligomerization.186 The same homochiral selectivity was also observed in the formation of cyclic dinucleotides. QC calculations have suggested that this might be caused by the lower stability of heterochiral forms as compared to the homochiral ones.187

Template directed oligomerization As Leu et al. formulated ‘‘Nonenzymatic, template-directed synthesis of nucleic acids is a paradigm for self-replicating systems’’.188 Starting from 1980s several highly efficient template-directed oligomerization (see Fig. 6) methods have been elaborated utilizing nucleotide-activation in the form of phosphoimidazolides.189–192 This relatively simple chemistry involves transphosphorylation steps between the activated 5 0 -phosphate group and the 2 0 or 3 0 -hydroxyls of the ribose. Proper positioning of the nucleotides is ensured by H-bonding and stacking interactions with the template. Since the transphosphorylation steps can be initiated by both the 2 0 and 3 0 hydroxyls of the ribose the reaction produces a mixture of 2 0 ,5 0 - and 3 0 ,5 0 -phosphodiester linkages. This motivated the combined experimental–theoretical study of Sheng et al., aimed at analyzing how incorporation of the ‘‘unnatural’’ 2 0 ,5 0 -linkages might influence the function of RNA molecules.193 Their classical molecular dynamics simulations suggest that RNA duplexes are flexible enough to accommodate a low amount of such linkage heterogeneities without substantially influencing the global folding, although the force-field choice could be a concern in this particular case.29

The route towards functional RNAs The evolution of species as we see it today was made possible by a perfectly designed RNA-machinery, which is based on a complex network of tertiary interactions involving thousands of atoms. Since the evolutionary function of RNA is strongly associated with its catalytic function, one can reasonably raise a question, is RNA-catalysis possible with simple oligonucleotides? Can RNA catalysis arise spontaneously and what are the simplest RNA systems that can exhibit at least some catalytic activities?

Fig. 5 Amines157,159 and protons179 catalyze the oligomerization of nucleotides. The role of amines in the non-enzymatic oligomerization of cyclic nucleotides has for long been an enigma of the origin of life research. After 40 years a plausible explanation has come from recent QC calculations.178 They have suggested that protonation of the phosphate oxygen either via H-bonding with protonated amines or direct proton-binding improves the kinetics of various template-free non-enzymatic oligomerization scenarios because it makes the phosphorus of the substrates more positive. This study exemplifies that ‘‘in silico’’ studies can provide a unique insight into such physico-chemical details of prebiotic processes, which are otherwise completely inaccessible for experiments.

20060 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

Fig. 6

Oligomerization of nucleotides along a template.

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

Recent experiments of the Di Mauro group have shown that Watson–Crick (WC) complementary oligonucleotide strands as short as ca. 9–12 nucleotides may be associated with a catalytic activity194–196 and MD simulations have helped to disclose the underlying chemical mechanisms.196 They have shown that transient catalytic centers (see Fig. 7) might be stabilized even by short oligonucleotides for sufficiently long times (for several tens to hundreds of nanoseconds) to support a very primitive form of RNA-catalysis.6 Basically, the model presented in ref. 6, 195 and 196 suggests that catalytic activity of RNA could have emerged as a result of (i) the hydrolytic instability of RNA and (ii) the transient stabilization of imperfectly paired forms of WC-complementary donor and acceptor oligonucleotide strands. The MD simulations can be considered as a textbook example of application of the method to catalyzed reactions. Although the MD method cannot directly capture the chemical reaction, it can suggest the formation of catalytically relevant geometries. In this particular case, the formation of tetraloop-like overhang geometries appeared to be critical to access the catalytically competent micro-arrangements. Note that the catalysis does not need to be promoted by the dominantly populated arrangements, but may proceed from rarely sampled but highly-reactive configurations (minor species), which are difficult to capture by experimental methods. The above-noted simulations also greatly profited from the recently reached ms-scale of the simulations and substantial improvement of the AMBER RNA force field.197 Shorter simulations would not be sufficient to obtain the structural insights while older versions of the RNA force field would provide

Perspective

unstable trajectories. A possible extension of this research could be to evaluate whether 3- and 5-loops may also provide a similar catalytic function. Nonetheless, due to the complexity of the studied system, any efforts in this direction must be based on a very firm experimental background. As we cautioned above, the currently available RNA simulation force fields are far from being perfect. The single-stranded RNA regions and over-hangs, albeit looking simple at first sight, are exceptionally difficult for MD simulations due to their broad unrestricted conformational space. Currently, perhaps the most challenging goal of the origin of life field is to understand the connections between the RNAand protein worlds. Indeed, it has been demonstrated that very simple oligonucleotides could make the first steps on this road.198 Experiments by Yarus and coworkers have shown that catalytic centers formed by as few as 3 nucleotides are able to mediate aminoacylation of oligonucleotides.199 MD simulations by the same group led to the proposal of a plausible structural model of the reaction center. Based on the structural-dynamics data the authors suggested that the aminoacylation is made possible by a proton transfer from the 2 0 -hydroxyl of the 3 0 -terminal uridine to the carbonyl oxygen of the phenylalanineadenylate substrate.199 Essentially, the proton transfer increases the positive charge on the carbonyl C of the substrate, which makes this activation mechanism highly reminiscent of those used to activate cyclic nucleotides for transphosphorylation leading to the formation of oligonucleotides in a nonenzymatic template-free manner.157,178

Outlook and summary

Fig. 7 MD simulations have suggested that tetraloop-like overhangs may mediate the cleavage of the 5 0 -terminal nucleotide of the donor strand. A prerequisite for the reaction is the stabilization of the imperfectly paired form of the WC complementary donor and acceptor strands. This requires about 5–6 WC base pairs, which set the lower limit of oligonucleotide length to ca. 10 nucleotides necessary for the onset of RNA-catalysis.

This journal is © the Owner Societies 2016

We provide an account of theoretical studies which are aimed to complement experiments to unravel chemical routes leading to the emergence of life on our planet. The examples shown illustrate that modern computational chemistry provides an impressive arsenal of powerful tools to study the origin of life at various levels. It should be noted that modern computational chemistry is still under-represented in the prebiotic field compared to many other areas of chemistry and biochemistry, which in addition are often dealing with considerably more complex chemical reactions than prebiotic chemistry. Perhaps one of the most ubiquitous areas of application so far has been the energetic characterization of reaction pathways of simple prebiotic reactions at the electronic-structure level of resolution, where QC theory has made enormous progress recently. Contemporary standard QC is capable of providing a close-to-converged description of the intrinsic electronic structural changes (assuming the idealized configuration of the reactants) of the ground-state chemical reactions. However, future studies should provide more insights into the effects of chemical environment on modulation of the chemical reactions, as noted above. Further development of computational models related to the prebiotic chemistry of nucleic acids is possible in two directions.

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20061

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

The ‘‘vertical direction’’ follows basically the trend dictated by experimental research and progresses towards higher and higher levels of molecular evolution. Undoubtedly, the synthesis of nucleic acid building blocks has so far been the most popular area of origin of life research addressed by computations. Albeit there are still many questions to solve in this topic, in the experimental literature a gradual shift can be recognized towards larger systems, such as the formation of the simplest oligonucleotide architectures or the origin of translation. In our opinion, these are the areas where modern computations have the potential to visibly improve our understanding in the near future. The increase of complexity of the systems will likely lead to applications of methods similar to those used in studies of enzyme reactions.174–177 The ‘‘horizontal direction’’ means that at each level of molecular complexity new methodological approaches bring about a more comprehensive description of the studied systems. One promising development is the use of quantum molecular dynamics instead of mere potential energy scans in the studies of chemical reactions.49 While the latter approach is equivalent to the 0 K electronic energy scans from a given configuration of the reactants, the former approach allows including the Boltzmann sampling, i.e., the real thermodynamics (free energy surfaces instead of potential energy surfaces). Ideally, such methods should be capable of spontaneously reaching the reactive configurations from an equilibrated ensemble of the ground state. Note that the reactive configurations do not always correspond to the dominantly sampled conformation, i.e., the actual chemical transition may be initiated from a rarely accessed but highly reactive configuration. In addition, thermal sampling of the orthogonal (with respect to the reaction path) degrees of freedom may significantly contribute to the reaction kinetics. It is well established from enzyme reactions that thermal motions can be essential.200–202 The cardinal problem of computational biochemistry has always been the balance between the accuracy of the primary description of the chemical problems and the sampling of the configurational space. Both issues are critically important for chemical reactions. On one hand, computational chemistry possesses exceptionally accurate QC methods to study (in the ground state) reaction pathways provided the reactive configuration of the reactants is known. Thus, we have powerful methods to study the energetics of the reactions along pathways with pre-determined reactant and product configurations while disregarding the thermal motions. We also have approximate force fields that allow large-scale MD simulations. What is currently missing are intermediary methods that would be sufficiently intrinsically accurate and would at the same time allow appropriate sampling. As noted above, there are intense ongoing efforts to develop such methods starting from both ‘‘classical’’ and ‘‘electronic structure’’ corners of the description. One direction includes the development of refined pairadditive and polarizable force fields. The other direction is the development of fast QC methods. Unfortunately, as mentioned several times above, achieving a balance between accuracy and speed appears to be a real challenge. While the pair-additive

20062 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

PCCP

force fields may be at the edge of their principally achievable accuracy,26 polarizable force fields face unexpectedly large problems when balancing all the energy contributions. All our tests of fast QC methods for real chemical problems similar to those in prebiotic chemistry (such as RNA self-cleavage reactions or description of RNA chain conformations) so far revealed a rather detrimental loss of accuracy.23,203 None of the available methods of the so-called semi-empirical nature (QC methods that include adjustable parameters, such as various variants of PM6 or AM1 approaches) that we tested provided satisfactory results. For conformational energies, their accuracy dropped below the AMBER force field performance23 while for chemistry we, for example, documented a loss of non-bridging phosphate oxygen when calculating free energy surface of the hairpin ribozyme self-cleavage reaction.203 It looks like that we are losing accuracy far more quickly than gaining the speed. Although this is an area of intense ongoing research in many leading methoddeveloping laboratories, it is uncertain if and when fast and accurate QC methods will be available for biomolecular structure and chemistry computations. Nevertheless, even without marked method breakthroughs, the mere increase of the capacity of available hardware and optimized efficiency of the computer codes are opening windows for new applications. A representative example of the growing capacity of computational methods are the recent AIMD simulations. Until very recently, almost all computational studies in prebiotic chemistry neglected the molecular and dynamic nature of the chemical environment (explicit inclusion of solvent molecules, finite temperature and pressure, mineral surfaces, etc.). Two main reasons hampered more realistic investigations: the very large computational cost of AIMD simulations including several hundreds of atoms and the lack of robust enhanced sampling methodologies flexible enough to afford free energy landscapes for a range of (possibly unknown) different reaction mechanisms. Both of these issues are becoming resolvable, thanks to the fast improvement of high-performance computers on one hand and of innovative algorithms on the other hand. The sampling problem may be partially eliminated by using important sampling methods, such as the metadynamics. The important sampling methods identify the fundamental degrees of freedom (so-called collective variables, typically the reaction paths) which allow minimizing the time spent by sampling thermal fluctuations of the less relevant degrees of freedom. This, in principle, enables a more accurate description of the effect of the environment on the mechanism of a chemical reaction. In the next decade we expect progressive accumulation of a large body of knowledge about prebiotic reaction mechanisms and free energy profiles in a series of different condensed-phase environments. Albeit the method is highly challenging, there is space for further development. In particular, the development of simulation techniques deciphering out-of-equilibrium processes will play a central role. We also expect substantial progress in state-of-the-art ab initio approaches to study mechanisms of photochemical reactions, which remain very costly. Since UV-light was suggested as an important source of energy in multiple prebiotically plausible

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

reaction pathways, a thorough understanding of the underlying mechanisms may eventually guide us towards the prebiotic synthesis of the remaining biomolecular building blocks. We hope that the current review provides a compelling evidence about the predictive power of computational methods. Albeit we understand that theory cannot solve the problem of the origin of terrestrial life on its own, we believe that it is able to transmit ideas, interpret experiments, provide predictions and universalize the results and concepts. A close collaboration between experimental and theoretical groups will be essential to fully exploit the arsenal of computational methods in prebiotic chemistry. We hope that this potential of modern computational methods will also be fruitfully utilized by experimental chemists working on the origin of life (as it is becoming common in many other areas of chemistry), because as the great Carl Sagan taught us: ‘‘Imagination will often carry us to worlds that never were. But without it we go nowhere.’’204

Acknowledgements ˇ R 14-12010S is gratefully Financial support from the grant GAC acknowledged. JS acknowledges support by Praemium Academiae. This research has been financially supported by the Ministry of Education, Youth and Sports of the Czech Republic under the project CEITEC 2020 (LQ1601).

References 1 W. Gilbert, Nature, 1986, 319, 618. 2 L. E. Orgel, J. Mol. Biol., 1968, 38, 381–393. 3 A. I. Oparin, The Origin of Life, Moscow Worker, Moscow, 1924. 4 J. D. Sutherland, Angew. Chem., Int. Ed., 2016, 55, 104–121. 5 R. Saladino, G. Botta, S. Pino, G. Costanzo and E. Di Mauro, Chem. Soc. Rev., 2012, 41, 5526–5565. ´kova ´, V. Brabec, O. ˇ 6 J. E. ˇ Sponer, J. ˇ Sponer, O. Nova Sedo, ´hal, G. Costanzo, S. Pino, R. Saladino and E. Di Mauro, Z. Zdra Chem. – Eur. J., 2016, 22, 3572–3586. ´dek, P. Jurecˇka, P. Bana ´ˇs and 7 J. ˇ Sponer, J. E. ˇ Sponer, A. Mla M. Otyepka, Biopolymers, 2013, 99, 978–988. 8 J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005, 105, 2999–3094. ´dek, P. Bana ´ˇs, P. Jurecˇka and 9 J. ˇ Sponer, J. E. ˇ Sponer, A. Mla M. Otyepka, Methods, 2013, 64, 3–11. 10 C. Møller and M. S. Plesset, Phys. Rev., 1934, 46, 618–622. ˇ´z 11 J. C ıˇek, J. Chem. Phys., 1966, 45, 4256–4266. 12 P. G. Szalay, T. Muller, G. Gidofalvi, H. Lischka and R. Shepard, Chem. Rev., 2012, 112, 108–181. 13 B. O. Roos, P. R. Taylor and P. E. M. Siegbahn, Chem. Phys., 1980, 48, 157–173. ¨dinger, Phys. Rev., 1926, 28, 1049–1070. 14 E. Schro 15 P. Hohenberg and W. Kohn, Phys. Rev., 1964, 136, B864–B871. 16 S. Ehrlich, J. Moellmann and S. Grimme, Acc. Chem. Res., 2013, 46, 916–926.

This journal is © the Owner Societies 2016

Perspective

17 S. Grimme, Chem. – Eur. J., 2012, 18, 9955–9964. 18 Y. Zhao and D. G. Truhlar, Acc. Chem. Res., 2008, 41, 157–167. 19 S. Grimme, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2011, 1, 211–228. 20 Y. Zhao and D. G. Truhlar, J. Chem. Theory Comput., 2011, 7, 669–676. ´nchez and W. Yang, Chem. Rev., 21 A. J. Cohen, P. Mori-Sa 2012, 112, 289–320. 22 W. Koch and M. C. Holthausen, A Chemist’s Guide to Density Functional Theory, Wiley, Weinheim, New York, 2nd edn, 2001. 23 H. Kruse, A. Mladek, K. Gkionis, A. Hansen, S. Grimme and J. Sponer, J. Chem. Theory Comput., 2015, 11, 4972–4991. 24 R. Sure and S. Grimme, J. Chem. Theory Comput., 2015, 11, 3785–3801. 25 L. Goerigk and S. Grimme, Phys. Chem. Chem. Phys., 2011, 13, 6670–6688. ´ˇs, P. Jurecˇka, M. Zgarbova ´, P. Ku ´, ¨hrova 26 J. ˇ Sponer, P. Bana M. Havrila, M. Krepl, P. Stadlbauer and M. Otyepka, J. Phys. Chem. Lett., 2014, 5, 1771–1782. 27 C. Abrams and G. Bussi, Entropy, 2014, 16, 163. 28 O. Valsson, P. Tiwary and M. Parrinello, Annu. Rev. Phys. Chem., 2016, 67, DOI: 10.1146/annurev-physchem-040215-112229. ´ˇs, P. Sklenovsky´, J. E. Wedekind, J. ˇ 29 P. Bana Sponer and M. Otyepka, J. Phys. Chem. B, 2012, 116, 12721–12734. 30 D. R. Roe, C. Bergonzo and T. E. Cheatham, J. Phys. Chem. B, 2014, 118, 3543–3552. 31 D. E. Condon, S. D. Kennedy, B. C. Mort, R. Kierzek, I. Yildirim and D. H. Turner, J. Chem. Theory Comput., 2015, 11, 2729–2742. 32 R. F. Brown, C. T. Andrews and A. H. Elcock, J. Chem. Theory Comput., 2015, 11, 2315–2328. 33 A. Savelyev and A. D. MacKerell Jr., J. Comput. Chem., 2014, 35, 1219–1239. 34 P. E. M. Lopes, J. Huang, J. Shim, Y. Luo, H. Li, B. Roux and A. D. MacKerell, J. Chem. Theory Comput., 2013, 9, 5430–5449. ´, J. Leszczynski and 35 N. Gresh, J. E. ˇ Sponer, N. ˇ Spacˇkova J. ˇ Sponer, J. Phys. Chem. B, 2003, 107, 8669–8681. 36 T. A. Halgren and W. Damm, Curr. Opin. Struct. Biol., 2001, 11, 236–242. 37 G. M. Torrie and J. P. Valleau, J. Comput. Phys., 1977, 23, 187–199. 38 A. Laio and M. Parrinello, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 12562–12566. ´ˇs, P. Jurec ˇka, N. G. Walter, J. ˇ 39 P. Bana Sponer and M. Otyepka, Methods, 2009, 49, 202–216. ´, Biochem. Biophys. Res. Commun., 1960, 2, 407–412. 40 J. Oro 41 J. P. Ferris, R. A. Sanchez and L. E. Orgel, J. Mol. Biol., 1968, 33, 693–704. 42 S. L. Miller, Science, 1953, 117, 528–529. 43 S. L. Miller and H. C. Urey, Science, 1959, 130, 245–251. ´ Schleyer, Proc. Natl. 44 D. Roy, K. Najafian and P. von Rague Acad. Sci. U. S. A., 2007, 104, 17272–17277. 45 R. Glaser, B. Hodgen, D. Farrelly and E. McKee, Astrobiology, 2007, 7, 455–470.

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20063

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

46 J. P. Ferris and L. E. Orgel, J. Am. Chem. Soc., 1966, 88, 1074. 47 R. A. Sanchez, J. P. Ferris and L. E. Orgel, J. Mol. Biol., 1967, 30, 223–253. 48 R. Saladino, C. Crestini, F. Ciciriello, S. Pino, G. Costanzo and E. Di Mauro, Res. Microbiol., 2009, 160, 441–448. 49 A. M. Saitta and F. Saija, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 13768–13773. 50 F. Pietrucci and A. M. Saitta, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 15030–15035. 51 R. Saladino, G. Botta, M. Delfino and E. Di Mauro, Chem. – Eur. J., 2013, 19, 16916–16922. 52 R. Saladino, E. Carota, G. Botta, M. Kapralov, G. N. Timoshenko, A. Y. Rozanov, E. Krasavin and E. Di Mauro, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, E2746–E2755. 53 J. Wang, J. D. Gu, M. T. Nguyen, G. Springsteen and J. Leszczynski, J. Phys. Chem. B, 2013, 117, 2314–2320. 54 J. Wang, J. D. Gu, M. T. Nguyen, G. Springsteen and J. Leszczynski, J. Phys. Chem. B, 2013, 117, 9333–9342. 55 J. Wang, J. D. Gu, M. T. Nguyen, G. Springsteen and J. Leszczynski, J. Phys. Chem. B, 2013, 117, 14039–14045. 56 J. E. Sponer, A. Mladek, J. Sponer and M. Fuentes-Cabrera, J. Phys. Chem. A, 2012, 116, 720–726. 57 S. Civis, L. Juha, D. Babankova, J. Cvacka, O. Frank, J. Jehlicka, B. Kralikova, J. Krasa, P. Kubat, A. Muck, M. Pfeifer, J. Skala and J. Ullschmied, Chem. Phys. Lett., 2004, 386, 169–173. 58 D. Babankova, S. Civis and L. Juha, Prog. Quantum Electron., 2006, 30, 75–88. 59 D. Babankova, S. Civis, L. Juha, M. Bittner, J. Cihelka, M. Pfeifer, J. Skala, A. Bartnik, H. Fiedorowicz, J. Mikolajczyk, L. Ryc and T. Sedivcova, J. Phys. Chem. A, 2006, 110, 12113–12120. 60 S. Civis, D. Babankova, J. Cihelka, P. Sazama and L. Juha, J. Phys. Chem. A, 2008, 112, 7162–7169. 61 M. Ferus, D. Nesvorny´, J. ˇ Sponer, P. Kubelı´k, R. Michalcˇ´kova ı ´, ´, J. E. ˇ V. Shestivska Sponer and S. Civisˇ, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 657–662. ´, V. Shestivska ´, J. ˇ 62 M. Ferus, R. Michalcˇ´kova ı Sponer, J. E. ˇ Sponer and S. Civisˇ, J. Phys. Chem. A, 2014, 118, 719–736. 63 M. Ferus, S. Civis, A. Mladek, J. Sponer, L. Juha and J. E. Sponer, J. Am. Chem. Soc., 2012, 134, 20788–20796. 64 M. Ferus, I. Matulkova, L. Juha and S. Civis, Chem. Phys. Lett., 2009, 472, 14–18. 65 M. Ferus, P. Kubelik and S. Civis, J. Phys. Chem. A, 2011, 115, 12132–12141. 66 M. P. Callahan, K. E. Smith, H. J. Cleaves, J. Ruzicka, J. C. Stern, D. P. Glavin, C. H. House and J. P. Dworkin, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 13995–13998. 67 Z. Martins, O. Botta, M. L. Fogel, M. A. Sephton, D. P. Glavin, J. S. Watson, J. P. Dworkin, A. W. Schwartz and P. Ehrenfreund, Earth Planet. Sci. Lett., 2008, 270, 130–136. 68 R. Saladino, C. Crestini, F. Ciciriello, G. Costanzo and E. Di Mauro, Chem. Biodiversity, 2007, 4, 694–720. 69 J. S. Hudson, J. F. Eberle, R. H. Vachhani, L. C. Rogers, J. H. Wade, R. Krishnamurthy and G. Springsteen, Angew. Chem., Int. Ed., 2012, 51, 5134–5137.

20064 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

PCCP

70 Y. A. Jeilani, H. T. Nguyen, D. Newallo, J. M. D. Dimandja and M. T. Nguyen, Phys. Chem. Chem. Phys., 2013, 15, 21084–21093. 71 Y. A. Jeilani, H. T. Nguyen, B. H. Cardelino and M. T. Nguyen, Chem. Phys. Lett., 2014, 598, 58–64. 72 H. T. Nguyen, Y. A. Jeilani, H. M. Hung and M. T. Nguyen, J. Phys. Chem. A, 2015, 119, 8871–8883. 73 R. Breslow, Tetrahedron Lett., 1959, 1, 22–26. 74 A. Butlerow, Justus Liebigs Ann. Chem., 1861, 120, 295–298. 75 A. Ricardo, M. A. Carrigan, A. N. Olcott and S. A. Benner, Science, 2004, 303, 196. 76 O. Pestunova, A. Simonov, V. Snytnikov, V. Stoyanovsky and V. Parmon, in Space Life Sciences: Astrobiology: Steps toward Origin of Life and Titan before Cassini, ed. M. Bernstein, R. Navarro-Gonzalez and R. Raulin, Elsevier Science Ltd, Oxford, 2005, vol. 36, pp. 214–219. 77 C. E. Harman, J. F. Kasting and E. T. Wolf, Origins Life Evol. Biospheres, 2013, 43, 77–98. 78 D. Ritson and J. D. Sutherland, Nat. Chem., 2012, 4, 895–899. 79 S. Civisˇ, R. Szabla, B. M. Szyja, D. Smykowski, O. Ivanek, A. Knı´zˇek, P. Kubelı´k, J. ˇ Sponer, M. Ferus and J. E. ˇ Sponer, Sci. Rep., 2016, 6, 23199. 80 H.-J. Kim, A. Ricardo, H. I. Illangkoon, M. J. Kim, M. A. Carrigan, F. Frye and S. A. Benner, J. Am. Chem. Soc., 2011, 133, 9457–9468. 81 B. E. Prieur, C. R. Acad. Sci., Ser. IIc: Chim., 2001, 4, 667–670. 82 J. E. Sponer, B. G. Sumpter, J. Leszczynski, J. Sponer and M. Fuentes-Cabrera, Chem. – Eur. J., 2008, 14, 9990–9998. 83 J. B. Lambert, G. Lu, S. R. Singer and V. M. Kolb, J. Am. Chem. Soc., 2004, 126, 9611–9625. ´. Va ´zquez-Mayagoitia, S. R. Horton, B. G. Sumpter, 84 A J. ˇ Sponer, J. E. ˇ Sponer and M. Fuentes-Cabrera, Astrobiology, 2011, 11, 115–121. 85 G. Zubay and T. Mui, Origins Life Evol. Biospheres, 2001, 31, 87–102. 86 J. E. Sponer, J. Sponer and M. Fuentes-Cabrera, Chem. – Eur. J., 2011, 17, 847–854. 87 E. D. Glendening, C. R. Landis and F. Weinhold, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2012, 2, 1–42. 88 R. Saladino, U. Ciambecchini, C. Crestini, G. Costanzo, R. Negri and E. Di Mauro, ChemBioChem, 2003, 4, 514–521. 89 H. T. Nguyen and M. T. Nguyen, Phys. Chem. Chem. Phys., 2015, 17, 16927–16936. 90 A. Schoffstall and E. Laing, Origins Life Evol. Biospheres, 1985, 15, 141–150. 91 F. H. Westheimer, Chem. Rev., 1981, 81, 313–326. 92 G. Costanzo, R. Saladino, C. Crestini, F. Ciciriello and E. Di Mauro, J. Biol. Chem., 2007, 282, 16729–16735. 93 M. W. Powner, B. Gerland and J. D. Sutherland, Nature, 2009, 459, 239–242. 94 M. W. Powner, J. D. Sutherland and J. W. Szostak, J. Am. Chem. Soc., 2010, 132, 16677–16688. 95 S. A. Benner, A. Ricardo and M. A. Carrigan, Curr. Opin. Chem. Biol., 2004, 8, 672–689.

This journal is © the Owner Societies 2016

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

PCCP

96 J. E. Sponer, J. Sponer and M. Fuentes-Cabrera, Chem. – Eur. J., 2011, 17, 847–854. 97 R. Szabla, J. E. Sponer, J. Sponer and R. W. Gora, Phys. Chem. Chem. Phys., 2013, 15, 7812–7818. 98 A. Choudhary, K. J. Kamer, M. W. Powner, J. D. Sutherland and R. T. Raines, ACS Chem. Biol., 2010, 5, 655–657. 99 C. S. Cockell, Origins Life Evol. Biospheres, 2000, 30, 467–500. 100 C. S. Cockell and G. Horneck, Photochem. Photobiol., 2001, 73, 447–451. 101 K. J. Zahnle and J. C. G. Walker, Rev. Geophys., 1982, 20, 280–292. 102 C. E. Crespo-Hernandez, B. Cohen, P. M. Hare and B. Kohler, Chem. Rev., 2004, 104, 1977–2019. 103 A. L. Sobolewski and W. Domcke, Phys. Chem. Chem. Phys., 2004, 6, 2763–2771. 104 C. T. Middleton, K. de La Harpe, C. Su, Y. K. Law, C. E. Crespo-Hernandez and B. Kohler, Annu. Rev. Phys. Chem., 2009, 60, 217–239. 105 D. Shemesh, A. L. Sobolewski and W. Domcke, J. Am. Chem. Soc., 2009, 131, 1374–1375. 106 A. L. Sobolewski and W. Domcke, ChemPhysChem, 2006, 7, 561–564. 107 C. Sagan, J. Theor. Biol., 1973, 39, 195–200. ´ and M. S. de Vries, 108 K. Kleinermanns, D. Nachtigallova Int. Rev. Phys. Chem., 2013, 32, 308–342. 109 A. L. Sobolewski and W. Domcke, Europhys. News, 2006, 37, 20–23. 110 R. Szabla, J. Campos, J. E. Sponer, J. Sponer, R. W. Gora and J. D. Sutherland, Chem. Sci., 2015, 6, 2035–2043. 111 G. M. Roberts and V. G. Stavros, Chem. Sci., 2014, 5, 1698–1722. 112 T. E. Dermota, Q. Zhong and A. W. Castleman, Chem. Rev., 2004, 104, 1861–1886. 113 G. M. Roberts, H. J. Marroux, M. P. Grubb, M. N. Ashfold and A. J. Orr-Ewing, J. Phys. Chem. A, 2014, 118, 11211–11225. 114 K. Rottger, H. J. Marroux, M. P. Grubb, P. M. Coulter, H. Bohnke, A. S. Henderson, M. C. Galan, F. Temps, A. J. Orr-Ewing and G. M. Roberts, Angew. Chem., Int. Ed., 2015, 54, 14719–14722. 115 F. Bernardi, M. Olivucci and M. A. Robb, Chem. Soc. Rev., 1996, 25, 321–328. 116 J. Finley, P.-Å. Malmqvist, B. O. Roos and L. Serrano´s, Chem. Phys. Lett., 1998, 288, 299–306. Andre 117 A. A. Granovsky, J. Chem. Phys., 2011, 134, 214113. 118 F. Plasser, M. Barbatti, A. J. A. Aquino and H. Lischka, Theor. Chem. Acc., 2012, 131, 1–14. 119 J. J. Szymczak, M. Barbatti and H. Lischka, Int. J. Quantum Chem., 2011, 111, 3307–3315. 120 O. Christiansen, H. Koch and P. Jørgensen, Chem. Phys. Lett., 1995, 243, 409–418. 121 A. B. Trofimov and J. Schirmer, J. Phys. B: At., Mol. Opt. Phys., 1995, 28, 2299. 122 A. Dreuw and M. Wormit, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2015, 5, 82–95. ¨ttig, in Adv. Quantum Chem., ed. H. J. Å. Jensen, 123 C. Ha Academic Press, 2005, vol. 50, pp. 37–60.

This journal is © the Owner Societies 2016

Perspective

¨ttig and F. Weigend, J. Chem. Phys., 2000, 113, 124 C. Ha 5154–5161. 125 F. Plasser, R. Crespo-Otero, M. Pederzoli, J. Pittner, H. Lischka and M. Barbatti, J. Chem. Theory Comput., 2014, 10, 1395–1405. ´lez, D. Escudero and L. Serrano-Andre ´s, 126 L. Gonza ChemPhysChem, 2012, 13, 28–51. 127 J. P. Ferris, J. E. Kuder and A. W. Catalano, Science, 1969, 166, 765–766. 128 J. P. Ferris and J. E. Kuder, J. Am. Chem. Soc., 1970, 92, 2527–2533. 129 J. P. Ferris, D. B. Donner and W. Lotz, J. Am. Chem. Soc., 1972, 94, 6968–6974. 130 E. Boulanger, A. Anoop, D. Nachtigallova, W. Thiel and M. Barbatti, Angew. Chem., Int. Ed., 2013, 125, 8158–8161. 131 H. L. Barks, R. Buckley, G. A. Grieves, E. Di Mauro, N. V. Hud and T. M. Orlando, ChemBioChem, 2010, 11, 1240–1243. 132 R. Szabla, R. W. Gora, J. Sponer and J. E. Sponer, Chem. – Eur. J., 2014, 20, 2515–2521. 133 S. D. Senanayake and H. Idriss, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 1194–1198. 134 R. A. Sanchez and L. E. Orgel, J. Mol. Biol., 1970, 47, 531–543. 135 R. Shapiro, Origins Life Evol. Biospheres, 1988, 18, 71–85. 136 A. A. Ingar, R. W. Luke, B. R. Hayter and J. D. Sutherland, ChemBioChem, 2003, 4, 504–507. 137 C. Anastasi, M. A. Crowe, M. W. Powner and J. D. Sutherland, Angew. Chem., Int. Ed., 2006, 45, 6176–6179. 138 C. Anastasi, M. A. Crowe and J. D. Sutherland, J. Am. Chem. Soc., 2007, 129, 24–25. 139 M. W. Powner, C. Anastasi, M. A. Crowe, A. L. Parkes, J. Raftery and J. D. Sutherland, ChemBioChem, 2007, 8, 1170–1179. 140 M. W. Powner and J. D. Sutherland, ChemBioChem, 2008, 9, 2386–2387. 141 D. Tuna, A. L. Sobolewski and W. Domcke, J. Phys. Chem. A, 2014, 118, 122–127. 142 D. Tuna and W. Domcke, Phys. Chem. Chem. Phys., 2016, 18, 947–955. 143 A. Banerjee, G. Ganguly, R. Tripathi, N. N. Nair and A. Paul, Chem. – Eur. J., 2014, 20, 6348–6357. ´ra, J. ˇ 144 R. Szabla, D. Tuna, R. W. Go Sponer, A. L. Sobolewski and W. Domcke, J. Phys. Chem. Lett., 2013, 4, 2785–2788. 145 R. Szabla, J. E. Sponer, J. Sponer, A. L. Sobolewski and R. W. Gora, Phys. Chem. Chem. Phys., 2014, 16, 17617–17626. ´ra, J. Phys. Chem. Lett., 146 R. Szabla, J. ˇ Sponer and R. W. Go 2015, 6, 1467–1471. 147 A. L. Sobolewski and W. Domcke, Chem. Phys. Lett., 2000, 329, 130–137. 148 A. L. Sobolewski and W. Domcke, J. Phys. Chem. A, 2001, 105, 9275–9283. 149 M. Barbatti, J. Am. Chem. Soc., 2014, 136, 10246–10249. 150 G. Costanzo, R. Saladino, C. Crestini, F. Ciciriello and E. Mauro, BMC Evol. Biol., 2007, 7, 1–8. 151 S. Miyakawa, H. J. Cleaves and S. L. Miller, Origins Life Evol. Biospheres, 2002, 32, 195–208.

Phys. Chem. Chem. Phys., 2016, 18, 20047--20066 | 20065

View Article Online

Published on 13 April 2016. Downloaded by Universite Pierre et Marie Curie on 02/11/2016 09:21:31.

Perspective

152 R. E. Kirk and D. F. Othmer, ECT Encyclopedia of Chemical Technology, Wiley Interscience, 1980, vol. 11, pp. 251–262. 153 P. M. Amarasinghe, K. S. Katti and D. R. Katti, J. Colloid Interface Sci., 2009, 337, 97–105. 154 H. Slebocka-Tilk, F. Sauriol, M. Monette and R. S. Brown, Can. J. Chem., 2002, 80, 1343–1350. 155 R. Saladino, V. Neri, C. Crestini, G. Costanzo, M. Graciotti and E. Di Mauro, J. Am. Chem. Soc., 2008, 130, 15512–15518. 156 R. Lohrmann and L. E. Orgel, Science, 1971, 171, 490–494. 157 M. S. Verlander, R. Lohrmann and L. E. Orgel, J. Mol. Evol., 1973, 2, 303–316. 158 S. A. Rudolph, E. M. Johnson and P. Greengard, J. Biol. Chem., 1971, 246, 1271–1273. 159 D. A. Usher and D. Yee, J. Mol. Evol., 1979, 13, 287–293. 160 L. E. Orgel and R. Lohrmann, Acc. Chem. Res., 1974, 7, 368–377. 161 G. Costanzo, S. Pino, F. Ciciriello and E. Di Mauro, J. Biol. Chem., 2009, 284, 33206–33216. 162 G. Costanzo, R. Saladino, G. Botta, A. Giorgi, A. Scipioni, S. Pino and E. Di Mauro, ChemBioChem, 2012, 13, 999–1008. 163 M. Morasch, C. B. Mast, J. K. Langer, P. Schilcher and D. Braun, ChemBioChem, 2014, 15, 879–883. 164 J. E. Sponer, J. Sponer, A. Giorgi, E. Di Mauro, S. Pino and G. Costanzo, J. Phys. Chem. B, 2015, 119, 2979–2989. ´ment, B. Grignard, L. Koole, C. Je ´ro ˆme and 165 B. Cle P. Lecomte, Macromolecules, 2012, 45, 4476–4486. 166 T. Steinbach, S. Ritz and F. R. Wurm, ACS Macro Lett., 2014, 3, 244–248. 167 S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799. 168 H. Kruse and J. Sponer, Phys. Chem. Chem. Phys., 2015, 17, 1399–1410. 169 H. Kruse, M. Havrila and J. Sponer, J. Chem. Theory Comput., 2014, 10, 2615–2629. 170 K. Gkionis, H. Kruse and J. ˇ Sponer, J. Chem. Theory Comput., 2016, 12, 2000–2016. 171 M. A. Ditzler, M. Otyepka, J. ˇ Sponer and N. G. Walter, Acc. Chem. Res., 2010, 43, 40–47. ´blova ´, B. Schneider, 172 M. V. Krasovska, J. Sefcikova, K. Re N. G. Walter and J. ˇ Sponer, Biophys. J., 2006, 91, 626–638. 173 P. Auffinger, L. D’Ascenzo and E. Ennifar, Met. Ions Life Sci., 2016, 16, 167–201. 174 T.-S. Lee, B. K. Radak, M. E. Harris and D. M. York, ACS Catal., 2016, 6, 1853–1869. 175 V. Mlynsky, N. G. Walter, J. Sponer, M. Otyepka and P. Banas, Phys. Chem. Chem. Phys., 2015, 17, 670–679. 176 P. Thaplyal, A. Ganguly, S. Hammes-Schiffer and P. C. Bevilacqua, Biochemistry, 2015, 54, 2160–2175. 177 S. Zhang, A. Ganguly, P. Goyal, J. L. Bingaman, P. C. Bevilacqua and S. Hammes-Schiffer, J. Am. Chem. Soc., 2015, 137, 784–798. 178 J. E. ˇ Sponer, J. ˇ Sponer and E. Mauro, J. Mol. Evol., 2015, 82, 5–10.

20066 | Phys. Chem. Chem. Phys., 2016, 18, 20047--20066

PCCP

179 L. Da Silva, M. C. Maurel and D. Deamer, J. Mol. Evol., 2015, 80, 86–97. 180 J. P. Ferris and G. Ertem, Science, 1992, 257, 1387–1389. 181 P. Mignon, P. Ugliengo and M. Sodupe, J. Phys. Chem. C, 2009, 113, 13741–13749. 182 P. Mignon, P. Ugliengo, M. Sodupe and E. R. Hernandez, Phys. Chem. Chem. Phys., 2010, 12, 688–697. 183 P. Mignon and M. Sodupe, Phys. Chem. Chem. Phys., 2012, 14, 945–954. 184 P. C. Joshi, M. F. Aldersley and J. P. Ferris, Origins Life Evol. Biospheres, 2011, 41, 213–236. 185 P. C. Joshi, M. F. Aldersley, J. D. Price, D. V. Zagorevski and J. P. Ferris, Origins Life Evol. Biospheres, 2011, 41, 575–579. 186 D. C. Mathew and Z. Luthey-Schulten, Origins Life Evol. Biospheres, 2010, 40, 303–317. 187 J. E. Sponer, A. Mladek and J. Sponer, Phys. Chem. Chem. Phys., 2013, 15, 6235–6242. 188 K. Leu, E. Kervio, B. Obermayer, R. M. Turk-MacLeod, C. Yuan, J. M. Luevano Jr., E. Chen, U. Gerland, C. Richert and I. A. Chen, J. Am. Chem. Soc., 2013, 135, 354–366. 189 T. Wu and L. E. Orgel, J. Am. Chem. Soc., 1992, 114, 7963–7969. 190 S. S. Mansy, J. P. Schrum, M. Krishnamurthy, S. Tobe, D. A. Treco and J. W. Szostak, Nature, 2008, 454, 122–125. 191 J. P. Schrum, A. Ricardo, M. Krishnamurthy, J. C. Blain and J. W. Szostak, J. Am. Chem. Soc., 2009, 131, 14560–14570. 192 C. Deck, M. Jauker and C. Richert, Nat. Chem., 2011, 3, 603–608. 193 J. Sheng, L. Li, A. E. Engelhart, J. Gan, J. Wang and J. W. Szostak, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 3050–3055. 194 S. Pino, G. Costanzo, A. Giorgi and E. Di Mauro, Biochemistry, 2011, 50, 2994–3003. 195 S. Pino, G. Costanzo, A. Giorgi, J. ˇ Sponer, J. ˇ Sponer and E. Mauro, Entropy, 2013, 15, 5362. 196 P. Stadlbauer, J. ˇ Sponer, G. Costanzo, E. Di Mauro, S. Pino and J. E. ˇ Sponer, Chem. – Eur. J., 2015, 21, 3596–3604. ´, M. Otyepka, J. ˇ ´dek, P. Bana ´ˇs, 197 M. Zgarbova Sponer, A. Mla T. E. Cheatham and P. Jurecˇka, J. Chem. Theory Comput., 2011, 7, 2886–2902. 198 M. Illangasekare, G. Sanchez, T. Nickles and M. Yarus, Science, 1995, 267, 643–647. 199 N. V. Chumachenko, Y. Novikov and M. Yarus, J. Am. Chem. Soc., 2009, 131, 5257–5263. 200 S. Hammes-Schiffer and J. Klinman, Acc. Chem. Res., 2015, 48, 899. 201 L. Masgrau and D. G. Truhlar, Acc. Chem. Res., 2015, 48, 431–438. 202 S. C. Kamerlin, P. K. Sharma, R. B. Prasad and A. Warshel, Q. Rev. Biophys., 2013, 46, 1–132. ´ˇs, J. ˇ 203 V. Mly´nsky´, P. Bana Sponer, M. W. van der Kamp, A. J. Mulholland and M. Otyepka, J. Chem. Theory Comput., 2014, 10, 1608–1622. 204 C. Sagan, Cosmos, Random House, New York, 1980.

This journal is © the Owner Societies 2016