Amino Acids-The Molecular Building Blocks of Life

Understanding the structure-property relationships of amino acids-the molecular building blocks of life

Aravindhan Ganesan

Dissertation submitted in fulfillment of requirements for the degree of Doctor of Philosophy

Environmental and Biotechnology Centre Faculty of Life & Social Sciences Swinburne University of Technology Victoria, Australia

2013

Abstract

Amino acids are significant molecular building blocks of proteins. Only 20 natural amino acids, whose structures differ in their side chain ‘R-’ groups, control the structures, functions and selectivity of almost all the proteins in biology. The structure-properties of the amino acids must be revealed, in order to understand the methodical behind the nature’s choices in using them as ‘building blocks’. There are several molecular level details of the amino acids, which are still unknown or limitedly known. In this project, the electronic structures, properties and dynamics of the aliphatic and the aromatic amino acids under isolated and defined environmental conditions are studied quantum mechanically. A rich tool chest of ab initio and density functional theory (DFT) methods has been employed.

The impacts of alkyl side chain groups on the structure-properties of the aliphatic amino acids are revealed in the gas phase. A number of properties including geometries, molecular dipole moments, ionization energies and spectra, charge re-distributions, vibrational spectra (IR and Raman), vibrational optical activity spectra (VOA), molecular orbitals and momentum spectra are investigated. Dual space analyses (DSA) has been employed as an efficient analytical tool to understand the electronic structures of the amino acids in both the coordinate space and momentum space. Our quantum chemical calculations are validated against the synchrotron-sourced and other experiments, whenever possible. It is revealed that there is no single universal model that is the best for the calculations of all the properties of the amino acids. Different models are able to produce optimal results for different properties and calculations.

Electron structure and properties of L-phenylalanine is revealed by studying the interactions of its functional groups (COOH, NH2 and phenyl) and its fragment schemes (alanine/benzene and glycine/toluene) in the gas phase. In order to understand the interactions of the functional groups, the structures of L-Phe and its derivatives, 3-phenylpropionic acid (PPA) and 2- phenethylamine (PEA) are comparatively studied. The ‘fragments-in-molecule’ approach, along with DSA, is employed to study the interactions between the different fragment schemes of phenylalanine, such as alanine/benzene and glycine/toluene. The results indicate that the inner shell of phenylalanine is dominated by the functional group interactions, while the fragment interactions are vital in the valence space of phenylalanine. Further, the phenyl

i ring in the phenylalanine serves as a buffer to resist the changes, which is useful for stabilizing the amino acid in the gas phase.

The spectroscopy and orbital properties of L-phenylalanine (R=phenyl ring) are combined together with those of L-tyrosine (R=phenol group), in order to understand the structure- properties of L-dopa (R=catechol group), an important neurotransmitter drug, in the gas phase. The impacts of hydroxyl group attachments on the electronic properties of these aromatic molecules are studied in detail. Our quantum mechanical calculations clearly identify the electronic and molecular properties that differentiate the drug molecule (i.e. L- dopa) from its amino acid precursors, L-phenylalanine and L-tyrosine.

Next, the dynamics and inter-molecular interactions of the phenylalanine-copper (II) complexes and micro-solvation processes (H2O, n=0-4) are studied using the DFT calculations and Car-Parrinello molecular dynamics (CPMD) simulation. The structures of the Phe-Cu2+ complexes with up to four water molecules are studied. The results indicate that the phenylalanine moiety appears to be in the neutral form in the isolated and mono-hydrated complexes, but in the zwitterionic form in the other hydrated systems. The present CPMD simulations reveal that the maximum coordination number of Cu2+ in the presence of phenylalanine under micro-solvation does not exceed four: the oxygen atoms from up to three water molecules and one carboxyl oxygen atom of phenylalanine. Any excess water molecules will migrate to the second solvation shell. Moreover a unique structural motif, 2+ (N)H···O(3)···H2O– Cu is present in the micro-solvated complexes with more than two water molecules, which is recognized to be significant in stabilizing the structures of the complexes. Extensively rich information of the structures, energetics, hydrogen bonds and dynamics of the lowest energy complexes are discussed.

The results presented in this thesis, therefore, help to elucidate the impacts of inter- and intra- molecular interactions on the structure-properties of the amino acids, as well as showcase a successful synergy of our theoretical calculations with the experiments.

Specially devoted to ‘my beloved’

iii

Acknowledgments

I am delighted to express my deep sense of gratitude to my principal supervisor, Professor Feng Wang, for kindly accepting to supervise my PhD project and also offering me a postgraduate scholarship from her Vice Chancellors research award at Swinburne. Her kind support, encouragement and suggestions throughout my candidature cannot be simply expressed in words. She offered me numerous opportunities to learn and develop my skills to become a responsible and independent researcher in future. All the achievements that I have seen during my PhD candidature could not have been possible without her. ‘Professor Wang, Thank you very much is all I can say.’

I would also like to sincerely thank my other supervisor, Professor Michael J Brunger in Flinders University, who has also been very supportive and encouraging during the course of my project. I humbly acknowledge the top-up grant offered by him from the ARC Centre of Excellence for Antimatter-Matter studies, Flinders University node. I would also like to extend my sincere thanks to Professor Paolo Carloni, who is also my external supervisor, for his kind support and hospitality during my six months visit to his laboratory in German Research School of Simulation Sciences. All the members in Professor Carloni’s group, especially Dr. Emiliano Ippoliti and Dr. Jens Dreyer, are thanked for providing me a friendly environment during my stay at GRS.

Dr. Ippoliti has been very kind and patient in helping me learn the CPMD package to carry out my simulations. His directions and suggestions have been very useful for me to master the skills. And Dr. Dreyer has also been equally supportive and collaborative. Another person, who really spent more time in helping with CPMD, is Dr. Julen Larrucea. I enjoyed all our fascinating discussions about molecular dynamics in Skype.

I would also like to thank Professor Kevin C Prince in Elettra Synchrotrone, Trieste, for providing his XPS measurements of the amino acids, which have been useful to validate the calculations in this thesis. I also acknowledge the support and encouragements of Professor Richard Sadus and Professor Billy Todd in Swinburne University.

DAAD Germany research grant and Tuition fee scholarships from Swinburne University are kindly acknowledged. The computing time required to perform the simulations presented in this thesis has been generously supported by the National Computational Infrastructure (NCI), VPAC, VLSCI and Swinburne’s Green machine. My special thanks go to VLSCI that offered me travel grants to support my attendance in prestigious conferences including SC10 and MM2012.

I wish to also thank all my friends in Swinburne University and outside. I greatly appreciate the people in my group, Dr. Ma, Dr. Lalitha, Dr. Fangfang, Anoja, Marawan and Narges, for offering me an excellent friendly environment, which I will always cherish.

I am really very proud to have a beautifully gifted family, my mother, father, father-in-law, mother-in-law, sister and brother-in-law, GP and my sweet little Netra darling. Their affection remains true strength in every step of my career. My dear wife, Mrs. Subha Aravindhan is the one person, whose presence makes me complete. Thank you very much dear, for your extra-ordinary patience, dedication, love and understanding.

Last but not the least, my thanks to ‘SAI’, you have always been there for me.

Declaration

I hereby declare that the thesis entitled “Understanding the structure-property relationships of amino acids-the molecular building blocks of life”, which is submitted in fulfillment of the requirements for the degree of Doctor of Philosophy in the Swinburne

University of Technology, is my own work. To the best of my knowledge and belief, it contains no material previously published or written by another person, except where due references are made in the text of the thesis. Any contribution made to the research by colleagues, with whom I have worked at Swinburne or elsewhere, during my candidature, is fully acknowledged. I affirm that this thesis contains no material, which has been accepted for the award to the candidate of any other degree or diploma.

Aravindhan Ganesan

April 2013

Publications

1. Aravindhan Ganesan and Feng Wang, ‘Intramolecular interactions of L-phenylalanine revealed by inner shell chemical shift of model molecules’, J. Chem. Phys., 131, 044321 (2009). 2. Aravindhan Ganesan, Michael Brunger and Feng Wang, ‘Influence of the functional

groups on the C-C chain of L-phenylalanine and its derivatives’, Nucl. Instr. and Meth. A, 619, 143 (2010). 3. Aravindhan Ganesan, Feng Wang, Michael Brunger and Kevin Prince, ‘Effects of alkyl side chains to the properties of the aliphatic amino acids probed using quantum chemical calculations’, J. Synchroton Rad., 18, 733 (2011). 4. Aravindhan Ganesan and Feng Wang, ‘Intramolecular interactions of L-phenylalanine: valence ionization spectra and orbital momentum distributions of its fragment molecules’, J. Comput. Chem., 32, 525 (2011). 5. Marawan Ahmed, Aravindhan Ganesan, Feng Wang, Vitaliy Feyer, Oksana Plekan and Kevin Prince, ‘Photoelectron Spectra of Some Antibiotic Building Blocks: 2-Azetidinone and Thiazolidine Carboxylic Acid’ J. Phys. Chem. A, 116, 8653 (2012). 6. Narges Mohammadi, Aravindhan Ganesan, Christopher T. Chantler and Feng Wang,

‘Differentiation of ferrocene D5d and D5h conformers using IR spectroscopy’, J. Organomet. Chem., 713, 51 (2012). 7. Aravindhan Ganesan, Michael Brunger and Feng Wang, ‘A study of aliphatic amino acids using simulated vibrational circular dichroism and Raman optical activity’, (J. Raman Spectrosc., revision submitted, 2013). 8. Aravindhan Ganesan, Jens Dreyer, Jaakko Akola, Feng Wang and Julen Larrucea, ‘CPMD simulation of Cu2+-phenylalanine complex under micro-solvated environment’ (submitted, 2013). 9. Aravindhan Ganesan, Feng Wang, Paolo Carloni, Jens Dreyer and Emiliano Ippoliti, ‘A QM/MM molecular dynamics study of phenylalanine and histidine in water’ (in preparation). 10. Aravindhan Ganesan, Michael Brunger and Feng Wang, ‘Understanding the electron structure-properties and intra-molecular process of L-dopa and its amino acid precursors” (in preparation).

vii

Contents

List of figures xi List of tables xv List of abbreviations xvii

1. Amino acids: the building blocks of life 1

1.1. Structures and classifications of amino acids 3 1.2. Principal biological roles of amino acids 6 1.3. Experiment and theory collaboration 8 1.4. Motivation of the project 9 1.5. Overview of the thesis 12

2. Methods and computational details 14

2.1. Introduction 14 2.2. Time independent Schrödinger equation 15 2.3. Born-Oppenheimer approximation 16 2.4. Hartree-Fock theory 17 2.5. Density functional theory 19 2.6. Exchange correlation 20 2.7. Basis sets 23 2.8. Molecular properties 26 2.8.1. Optimized geometries 26 2.8.2. Ionization energies 28 2.8.3. Atomic charge distribution 31 2.8.4. Momentum distribution and DSA 32 2.8.5. Vibrational spectroscopy 33 2.8.6. Vibrational optical activity spectroscopy 34 2.9. Ab-initio molecular dynamics 36 2.10. Solvent effects and models 39

viii

3. Aliphatic amino acids 43

3.1. Introduction 43 3.2. Computational details 45 3.3. Geometrical details 46 3.4. Hirshfeld charge distributions 53 3.5. Ionization energy responses to side chain changes 54 3.6. Valence electron momentum spectra using DSA 63 3.7. Vibrational and VOA spectra 66 3.7.1. Theoretical and experimental Raman/ROA spectra of ZW alanine 66 3.7.2. Vibrational spectra of glycine 68 3.7.3. VOA spectra of the chiral amino acids 71 3.8. Summary 81

4. Intra-molecular interactions of phenylalanine 83

4.1. Introduction 83 4.2. Computational details 87 4.3. Geometrical properties 88 4.4. Theoretical and experimental inner shell spectra of L-Phe 90 4.5. Impacts of functional groups on L-Phe 93 4.6. Fragmentation in L-Phe 101 4.7. Summary 118

5. Effects of the hydroxyl group on aromatic molecules 120

5.1. Introduction 120 5.2. Computational details 124 5.3. Geometrical properties 124 5.4. Hydrogen bond network 127 5.5. Inner shell changes 128 5.6. Hirshfeld charge distribution 134 5.7. Effects in valence space 135 5.8. Aromaticity properties 141 5.9. Summary 144

6. Interactions of micro-solvated Cu2+-phenylalanine complex 146

6.1. Amino acids in molecular environment 146 6.2. Computational details 149 6.3. Stable structures of Phe-Cu2+ complexes 150 6.4. CPMD simulation 157 6.5. Molecular orbital analyses 164 6.6. Summary 169

7. Summary and future work 170 175 Appendix

References 194

List of figures

A systematic representation of the formation of a primary structure of Fig. 1.1: 2 protein from amino acids. Fig. 1.2: Structural schema of neutral and zwitterionic amino acids. 3 Fig. 1.3: Chemical structures of the natural amino acids. 4 Fig. 1.4: Structures of phenyl-phenyl dipeptide and oxytocin polypeptide. 7 Structural representations of the three major isoforms of amino acids Fig. 1.5: 10 in the gas phase along with their intra-molecular H-bonds. Fig. 3.1: Chemical structures of the aliphatic amino acids. 43 Optimized lowest energy structures of the neutral aliphatic amino Fig. 3.2: 47 acids and their numbering schemes. Comparison of the experimental and E calculated O 1s, N 1s and C Fig. 3.3: KS 55 1s spectra of the L-alanine. Comparison of the O 1s, N 1s and C 1s spectra of the aliphatic amino Fig. 3.4: 57 acids calculated using the EKS method with a FWHM of 0.57 eV. Comparison of the experimental photoelectron spectra of glycine in Fig. 3.5: the outer valence space with the theoretical spectra using the 60 OVGF/TZVP model and the SAOP/et-pVQZ model. Comparison of the outer valence vertical ionization spectra of the aliphatic amino acids calculated using the OVGF/TZVP model, Fig. 3.6: convoluted with a FWHM of 0.70 eV. Inner valence energy diagram 62 of the aliphatic amino acids calculated using the LB94/et-pVQZ model. Selected valence orbital electron density and momentum distributions Fig. 3.7: of the aliphatic amino acids: the HOMOs, the NHOMOs, the 64 innermost valence orbitals and the third innermost valence orbitals. Valence orbital electron density and momentum distributions of the Fig.3.8: 65 THOMOs of the aliphatic amino acids. Orbital momentum densities and charge densities of glycine (10a’ MO), alanine (11a & 12a Mos); alanine 12a, valine (15a & 16a Mos), Fig. 3.9: 66 and valine (15a MO), leucine (16a & 17a MOs) and isoleucine (16a & 17a Mos). Comparison of the calculated and experimental Raman and ROA Fig. 3.10: 67 spectra of alanine ZW in aqueous solutions. Comparative IR and Raman spectra of the aliphatic amino acids Fig. 3.11: 68 showing the fingerprint and characteristic regions of the spectra.

Fig. 3.12: IR and Raman spectra of glycine in the gas and solvent phases. 69 IR/VCD and Raman/ROA spectra of alanine in the gas and solvent Fig. 3.13: 72 phases. IR/VCD and Raman/ROA spectra of valine in the gas and solvent Fig. 3.14: 76 phases. IR/VCD and Raman/ROA spectra of leucine in the gas and solvent Fig. 3.15: 77 phases. IR/VCD and Raman/ROA spectra of isoleucine in the gas and solvent Fig. 3.16: 78 phases. Fig. 4.1: Chemical structure of L-Phe and its nomenclatures. 83 Ground state electronic structures of L-Phe and its derivatives with Fig. 4.2: 85 nomenclatures. Optimized ground structures of L-Phe and its fragment schemes, Fig. 4.3: 87 alanine/benzene scheme and glycine/toluene scheme. Fig. 4.4: Energy based chemical scheme of L-Phe and its derivatives. 88 Fig. 4.5: Simulated and measured C 1s, N 1s and O 1s spectra of L-Phe. 91 C 1s spectra of L-Phe and its derivatives along with alanine and

Fig. 4.6: benzene (LB94/et-pVQZ) simulated using the EKS method and C 1s 95 energy correlation diagram of the model molecules. Vertical valence ionization spectra of L-Phe and its derivatives Fig. 4.7: 97 obtained using the SAOP/et-pVQZ model (FWHM of 0.4 eV). Orbital diagrams of the MOs in frontier region and functional group Fig. 4.8: 98 regions of all the molecules. Alaninyl and phenyl dominant MOs of L-Phe and its derivatives Fig. 4.9: 100 within 17-26 eV. Vertical valence ionization spectra of glycine vs. alanine and toluene Fig. 4.10: 105 vs. benzene obtained using the SAOP model. Vertical valence ionization spectra of benzene-Phe-alanine and Fig. 4.11: toluene-Phe-glycine sets calculated using the SAOP model and with a 107 FWHM of 0.4 eV. Comparison of the vertical valence ionization spectra of native L-Phe (middle panel) against the synthetic spectra simulated from alanine + Fig. 4.12: 108 benzene (upper panel) and glycine + toluene (lower panel) fragment schemes. Valence orbital correlation diagrams of L-Phe with its fragment Fig. 4.13: schemes: alanine/benzene and glycine/toluene in the inner valence and 110 outer valence regions.

xii

Comparison of the theoretical and experimental orbital momentum Fig. 4.14: 112 distributions of the selected orbitals in benzene. Selected valence orbital electron densities and TMDs of the L-Phe and Fig. 4.15: its fragments showing related chemical bonding characteristics that 113 are dominated by their individual fragments. Orbital densities and TMDs of the selected valence MOs of L-Phe Fig. 4.16: 115 those are dominated by the glycine-toluene scheme. Orbital densities and TMDs of the selected valence MOs of L-Phe Fig. 4.17: 117 those are dominated by the alanine-benzene scheme. Orbital densities and TMDs of the selected valence MOs that are Fig. 4.18: dominated by the strong interactions in L-Phe and its fragment 117 schemes. Fig. 5.1: Biochemical transformation of aromatic amino acids into dopamine. 121 Schematic representation of biochemical transport of L-dopa from Fig. 5.2: 122 blood into central nervous system. Fig. 5.3: Optimized structures of L-phe, L-tyr and L-dopa. 124 Comparison of the theoretical and experimental C 1s and O 1s spectra of L-tyr. The theoretical spectrum is simulated with an FWHM of 0.47 Fig. 5.4: 130 eV and shifted by 0.45 eV (C 1s) and 3.52 eV (O 1s) to match the experiment. Comparisons of the C 1s spectra of L-phe, L-tyr and L-dopa simulated Fig. 5.5: 131 with an FWHM of 0.4 eV using the LB94/et-pVQZ model. Comparisons of the O 1s spectra of L-phe, L-tyr and L-dopa simulated Fig. 5.6: 133 with an FWHM of 0.4 eV using the LB94/et-pVQZ model. Hirshfeld charges of L-phe, L-tyr and L-dopa given as heat map Fig. 5.7: 134 representation. Comparison of the valance spectra of L-tyr simulated (FWHM = 0.5 Fig. 5.8: eV) using the SAOP/et-pVQZ and OVGF/6-311G** against the 136 experimental spectrum. Valence ionization spectra (FWHM = 0.5 eV) of L-phe, L-tyr and L- Fig. 5.9: 138 dopa based on the SAOP/et-pVQZ calculations. Frontier orbital correlation diagrams of L-phe, L-tyr and L-dopa along Fig. 5.10: 140 with the HOMO-LUMO energy gaps of the molecules. NICS-rate spectra of L-phe, L-tyr and L-dopa calcualted as a function Fig. 5.11: 142 of distance (Å). Geometry optimized structures of Cu2+ ion in micro-solvated Fig. 6.1: 151 environment (n=1-6) using B3LYP/6-311++G(d,p) method.

xiii

The optimized structures of phenylalanine (Phe), [Phe-Cu]2+ and micro-hydrated [Phe-Cu(n=0-4)]2+ structures along with their relative Fig. 6.2: 153 energies in kcal∙mol-1 obtained from B3LYP/6-311++G(d,p) calculations. Last snapshots of the micro-solvated [Phe-Cu]2+ structures from the Fig. 6.3: 158 12 ps CPMD simulation. Trajectories of the Cu-O distances and inter-molecular H-O distances 2+ Fig. 6.4: and intra-molecular NH-O(3) distances of the l[Phe5a-Cu(n=4)] 162 complex. Radial distribution function spectra for the lowest energy micro- Fig. 6.5: hydrated [Phe-Cu]2+ complexes calculated from the CPMD 164 trajectories. HOMO and HOMO-1 orbitals of the lowest energy structures of Phe3, Fig. 6.6: 2+ 166 l[Phe1-Cu] and lowest energy micro-hydrated complexes. Orbital energy correlation diagram of the Cu2+ ‘d’ orbital MOs in the 2+ Fig. 6.7: l[Phe1-Cu] and the lowest energy micro-hydrated complexes along 167 with their corresponding orbital diagrams.

xiv

List of tables

Table 2.1: Summary of computational details. 42 Selected geometric parameters of the NT and ZW glycine and Table 3.1: alanine compared with the available experimental and other 48 values from the literature. Selected geometric parameters, from the present computations, of Table 3.2: 49 the aliphatic amino acids in the gas and solvent phases. Selected molecular properties, from the present calculations, of the aliphatic amino acids in the gas (NT) and solvent phases (ZW). Table 3.3: 52 Corresponding experimental results, where available, are also shown. Hirshfeld charges of the C, N and O sites of the aliphatic amino Table 3.4: 53 acids calculated using LB94/et-pVQZ model (values in a.u.).

Core IPs of glycine and alanine calculated using the EKS Table 3.5: (relaxed) and LB94/et-pVQZ methods are compared against the 54 experimental data. Core IPs of the aliphatic amino acids calculated using the E Table 3.6: KS 58 (relaxed) and LB94/et-pVQZ methods. Valence orbital ionization energies (eV) of the aliphatic Table 3.7: amino acids calculated using the OVGF and SAOP models 59 together with other theoretical and experimental energies. Present scaled and unscaled vibrational wavenumbers of NT (in Table 3.8: gas phase) and ZW (in solution) glycine, along with the available 70 experimental data (all wavenumbers are given in cm-1). Present vibrational wavenumbers (in cm-1) and their wavenumbers Table 3.9: in the functional group region ( > 1600 cm-1) of the NT and ZW 73 forms of the aliphatic amino acids. Present results showing the comparative vibrational wavenumbers Table 3.10: (in cm-1) and their assignments, of the NT and ZW forms of the 74 aliphatic amino acids in the alkyl-region ( < 1600 cm-1). Ground state geometries of L-Phe, PPA and PEA obtained using Table 4.1: the B3LYP/TZVP and MP2/TZVP (Phe only) models. Results are 90 compared with values taken from the literature cited. Comparison of core ionization potentials (IPs) of the ground electronic structures of Phe, PPA and PEA calculated using the Table 4.2: 93 LB94/et-pVQZ model (in eV) and EKS methods (in eV) along with available experiments and other works.

Hirshfeld charges of the carbon sites in the model molecules Table 4.3: 97 calculated using the LB94/et-pVQZ model. Valence IPs (eV) for benzene and toluene calculated using the Table 4.4: SAOP and OVGF methods along with the available experimental 102 values. Valence IPs (eV) for alanine and glycine calculated using Table 4.5: different methods together with the available experimental and 102 other theoretical values. Valence IPs (eV) of L-Phe calculated using the SAOP and OVGF Table 4.6: 103 models along with the available measured values. Table 5.1: Differences between L-phe, L-tyr and L-dopa. 122 Comparison of the calcualted and experimental geometries of L- Table 5.2: 125 tyr and L-dopa. Comparison of the geometric parameters of L-phe, L-tyr and L- Table 5.3: 127 dopa calculated using the B3LYP/6-311G** model. Table 5.4: Intra-molecular H-bonds of the aromatic molecules (Å). 128 Inner shell vertical IPs of L-phe, L-tyr and L-dopa calculated using Table 5.5: the LB94/et-pVQZ model along with the available experimental 129 data (eV). Valence vertical IPs of L-phe, L-tyr and L-dopa calculated using Table 5.6: 137 the SAOP/et-pVQZ and OVGF/6-311G** models (eV).

The calculated NICS(0), NRR and NRR  values for L-phe, L-tyr Table 5.7: ( ) 144 and L-dopa.

Comparison of the average Cu-O and O(1st hydration shell)-O(2nd hydration Table 6.1: shell) distances from this work with that of the available 151 experimental and other works. Selected geometrical parameters of the [Phe-Cu]2+ complexes Table 6.2: 155 optimized using the B3LYP/G09 and BLYP/CPMD methods. Selected distances of Cu2+ in the lowest energy micro-hydrated Table 6.3: 157 complexes (in Å). Selected geometrical parameters of the initial and final (shaded in Table 6.4: grey color) snapshots of the micro-hydrated complexes in the 159 CPMD simulations. Orbital energies (in eV) of the frontier molecular orbitals of the Table 6.5: 165 lowest energy micro-hydrated complexes.

xvi

List of abbreviations

ADF Amsterdam Density Functional AIMD Ab-initio molecular dynamics B3LYP Becke, three-parameter, Lee-Yang-Parr hybrid functional CPCM Conductor-like polarizable continuum model (water) CPMD Car-Parrrinello Molecular Dynamics Cu2+ Copper (II) DFT Density functional theory DSA Dual space analysis EMS Electron momentum spectroscopy Et-pVQZ Even tempered valence quadruple zeta with polarization function eV Electro volts EXP Experiment FWHM Full width at half maximum G03 Gaussian 03 G09 Gaussian 09 H-bond Hydrogen bond HF Hartree-Fock HOMO Highest occupied molecular orbital IP Ionization potential IR Infra-red LB94 Exchange correlation functional of Van Leeuwen and Baerends LUMO Lowest unoccupied molecular orbital MD Molecular dynamics MM Molecular mechanics MO Molecular orbital NHOMO Next highest occupied molecular orbital NICS Nucleus independent chemical shift NMR Nuclear magnetic resonance NRR NICS-rate ratio NT Neutral OVGF Outer valence Green function

xvii

PEA 2-phenylethylamine Phe Phenylalanine PPA 3-phenylpropionic acid ps Picosecond QM Quantum mechanics QM/MM Quantum mechanics/molecular mechanics ROA Raman optical activity SAOP Statistical average of orbital potentials SAP Single amino acid polymorphism SOMO Singly occupied molecular orbital THOMO Third highest occupied molecular orbital TMD Theoretical momentum distribution Tyr Tyrosine TZVP Triple zeta valence polarized VCD Vibrational circular dichroism VOA Vibrational optical activity VWN Voska, Wilk and Nusair XPS X-ray Photoelectron spectra ZW Zwitterion

xviii

Amino acids... Chapter 1

CHAPTER

Amino acids: the building blocks of life

Proteins are complex macromolecules that are the prime constituents of organisms. Genomes of most organisms code enormous varieties of proteins that exhibit different types of functions to mediate cellular processes. Proteins exist in different sizes ranging from tens to several thousands of residues[1], yet every structure performs its unique biological role. But, how can proteins be so diverse?

The answer lies in their structural compositions. Primary structures of proteins are usually linear polymers composed of monomers called ‘amino acids’. Almost all proteins are made from the combinations of only 20 natural amino acids that are connected by peptide bonds. A peptide linkage (i.e., -C(=O)NH-) is formed by a condensation reaction of two amino acid residues, where the carboxyl group of one amino acid reacts with the amino group of another amino acid, displacing a water molecule to result in an amide structure. Fig. 1.1 depicts a schematic representation of the formation of proteins starting from a simple condensation reaction between two amino acids.

Amino acids... Chapter 1

Fig.1.1: A systematic representation of the formation of a primary structure of protein from amino acids.

For the first time in 1902, two scientists, Franz Hofmeister and Emil Fischer, who participated in the 74th meeting of the German Scientists and Physicians Society, simultaneously made their remarkable claims that proteins are linear chains of -amino acids. Dr. Hofmeister delivered his proposal in the morning session, based on his interpretations of the biuret reaction in proteins and a few hours later, Dr. Fischer* presented some chemical details supporting the peptide-bond model[2]. Since then, amino acids are popularly represented as the molecular ‘building blocks’ of proteins.

-Michael Behe

* In 1902 Hermann Emil Fischer was awarded the Nobel Prize in Chemistry for his work on sugar and purine syntheses. An exciting biography about E. Fisher is available at http://www.nobelprize.org/nobel_prizes/chemistry/laureates/1902/fischer-bio.html

Amino acids... Chapter 1

1.1. Structures and classifications of amino acids

Amino acids generally consist of three major groups, a carboxyl (-COOH), an amino

(NH2) and an ‘R’ (side chain) group, all connected to a central chiral carbon atom (except the smallest amino acid, glycine). They differ only in their side chain R groups that in-turn are responsible for their unique properties. In gas phase (or) isolated conditions, amino acids

exist as neutral (NT) forms, NH2-CH(R)-COOH. However, they prefer zwitterionic (ZW) + - states, NH3 -CH(R)-COO , in solid state and aqueous environment. Fig.1.2 presents the schematic structures of neutral (a) and zwitterion forms (b) of amino acids. In the ZW state, the proton from the carboxyl group is transferred to the amino group, thereby leaving + - amphiprotic functional groups such as, NH3 and COO in amino acids. The ZW nature of amino acids has very significant role in the biological functions of different physiological components that are mostly dependent on water, blood, membranes and cellular fluids, for instance.

(a) (b)

Neutral amino acid Zwitterionic (In Gas Phase) (In Solvent Phase) Fig.1.2: Structural schema of (a) neutral and (b) zwitterionic amino acids.

There are only 20 natural amino acids which can be categorized into various groups based on their side chains such as (i) aliphatic groups, (ii) aromatic groups, (iii) cyclic group (iv) acidic and amide groups (v) basic groups (vi) hydroxyl containing amino acids and (vii) sulphur containing amino acids. Fig. 1.3 presents the chemical structures of the 20 natural amino acids.

Aliphatic amino acids contain non-cyclic hydrocarbon chains such as glycine, alanine, valine, leucine and isoleucine. Glycine (R=H) is the smallest of all aliphatic amino acids as

the side chain is a hydrogen (H) atom. Alanine (R=CH3) is the second smallest amino acid

Amino acids... Chapter 1

with a methyl side chain group. Other aliphatic amino acids, valine (R=CH(CH3)2), leucine

(R=CH2CH(CH3)2) and isoleucine (R=CH(CH3)CH2CH3), possess more complex side chain groups and are also known as branched chain amino acids (BCAAs). Leucine and isoleucine

are structural isomers with the same molecular formula, C6H13NO2, but their branching patterns are different.

Fig.1.3: Chemical structures of the natural amino acids along with their three letter and single letter codes.

Aliphatic amino acids are hydrophobic in nature, meaning that they are not able to dissolve well in water. The hydrophobicity of the molecules increases with the increasing numbers of carbon atoms[3, 4]. For example, glycine (R=H) is the least hydrophobic in nature and leucine and isoleucine are the most hydrophobic aliphatic molecules. Thus, the aliphatic amino acids participate actively in protein folding by creating hydrophobic pockets in soluble proteins. They also take part in the lipophilic interactions leading to anchor membrane proteins within the phospholipid bilayers and in the secondary structural

Amino acids... Chapter 1

properties of amphiphilic and amphipathic peptides[5, 6]. Further, aliphatic BCAAs are considered as important modulators of protein metabolism as they play significant roles in promoting protein synthesis and inhibiting the protein degradation processes[7].

Aromatic amino acids are those containing one or more aromatic rings in their side chains. Aromatic amino acids include phenylalanine, tyrosine and tryptophan, among which

phenylalanine is the smallest aromatic molecule with R=CH2C6H5. Tyrosine is structurally similar to phenylalanine, but with an extra hydroxyl group (OH) in the para position of its phenyl ring. Phenylalanine can be naturally converted into tyrosine by an enzyme, phenylalanine hydroxylase. There are a number of experiments showing the enzymatic conversion of phenylalanine into tyrosine[8-13. ] Besides tyrosine formation, the conversion of phenylalanine into tyrosine is an important step in the catabolism of phenylalanine in mammalian organisms. Failure of natural phenylalanine conversion is linked with diseases such as oligophrenia phenylpyruvica[13., 14Both] phenylalanine and tyrosine serve as precursors for L-dopa and dopamine neurotransmitters[15 , 16]. Tryptophan is structurally different from the other two aromatic amino acids and possesses an indole ring. Indole is an aromatic heterocyclic organic compound, which is bicyclic in structure with a fused six- member benzene ring and nitrogen containing pyrrole ring.

Aromatic amino acids also exhibit hydrophobic characteristics, which play a vital role in many of the biological processes in the living cells[17] such as protein-DNA recognition[18, 19] [20-24], RNA synthesis[25] and protein-RNA binding[26, 27]. Aromatic molecules have shown to bind strongly in the DNA cleavage and groove[28-30]. DNA cleavage is a reaction in which one of the covalent sugar-phosphate linkages between nucleotides, composing the backbone of DNA, is cleaved such that a mature DNA splits into two parts. On the other hand, a stable double stranded structure of DNA is stabilized by the existence of grooves: major groove occurs when the backbones of DNA are far apart and minor grooves exist when the backbones are closer to each other. Such stable grooved structures of DNA and cleavage reactions are known to be enhanced by the stacking interactions of the amino acids. For example, a single phenylalanine plays crucial roles in the homologous DNA recombination[31]. Moreover, side chain dynamics of aromatic amino acids control the water exchange channels and ligand binding/unbinding reactions in several enzymes[32-36]. For instance, phenylalanine and tyrosine residues are found to control the ligand exit mechanisms of histone deacetylase enzymes[36].

Amino acids... Chapter 1

Aspartate (R=CH2COO) and Glutamate (R=CH2(CH2COO)) are acidic amino acids that contain carboxylate groups in their side chains. Side chains in these acidic amino acids are negatively charged at physiological pH and are polar in nature. Nevertheless, asparagine and glutamine are the uncharged amide forms of their acidic states, aspartate and glutamate. Unlike amide and acidic amino acids, the basic amino acids such as histidine, lysine and arginine are positively charged and hydrophilic in nature. Among the basic amino acids, histidine possesses an imidazole ring and can also be categorized as an aromatic amino acid.

Moreover, serine (R=CH2OH) and threonine (R=CHOHCH3) along with tyrosine can be grouped as hydroxyl amino acids, because they possess OH group in their side chains. Proline is the only cyclic amino acid, in which the side chain R group connects to the imino group

(NH) rather than an amino group (NH2). On the other hand, cysteine and methionine possess a sulphur atom in their side chains and are uniquely classified as ‘sulphur-containing’ amino acids. Both these amino acids are non-polar and hydrophobic in nature.

Of the twenty natural amino acids, nine amino acids can be produced inside the body and are known as non-essential amino acids[37]. These include glycine, alanine, asparagine, aspartic acid, cysteine, glutamic acid, glutamine, proline and serine. The other amino acids such as arginine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, tyrosine, threonine, tryptophan and valine, cannot be naturally produced and are to be supplemented in the food. As a result, these molecules are categorized as essential amino acids.

1.2. Principal biological roles of amino acids

Construction of huge array of macromolecules from a limited numbers of amino acid residues is a recurring cycle in biology of living beings[38]. Amino acids can be joined together to form protein molecules of any size or sequence. Considering for example, 200 amino acid residues together can make a total of 20200 (in astronomical units) combinations of proteins, although not all of them are biologically prevalent. There are several important reasons for proteins to be biologically active, among which protein folding and structural stabilities remain as essential processes. To be able to perform their biological roles, proteins fold into one or more conformations[39]. Different conformations are driven by a number of non-covalent interactions[40] such as, hydrogen bonds (H-bonds), van der Waals forces, hydrophobic interactions, etc. Amino acid constituents of a protein generally hold necessary

Amino acids... Chapter 1

information to determine how proteins should fold into stable three dimensional structures[38] in order to perform their desired molecular functions. It is also important to note that the folding processes of some proteins may require a special group of molecules known as ‘chaperonins’ for their folding processes, heat shock protein 60 for instance. Vastly complex characteristics of proteins are, therefore, due to the composite properties of their amino acid constituents and their interactions in 3D space. Indeed in the absence of experimental thermodynamic data for proteins, amino acids can serve as useful tools in estimating their properties[41].

Changes in the single amino acid type, the process known as single amino acid polymorphism (SAP), can potentially lead to different protein products[42, 43]. Genetic variations caused by SAPs are able to cause severe disorders[41, 43 , 44]. Previous works on protein structures and functions have suggested that SAPs are accountable for certain types of diseases including sickle cell anaemia and Mendelian disease[41, 43-.45] For instance, analyses of the haemoglobin protein of a normal person and a sickle cell anaemic person suggested that substitution of glutamate (in normal individuals) with valine (in the patients) play active roles in the disease[45]. Identifying and characterizing SAPs remains an important aspect of bioinformatics research in the past decade. Amino acids are, therefore, the central units that control the structures and functions of proteins.

Apart from their primary functions as ‘protein builders’, amino acids are also important molecular building blocks of several peptides. Dipeptides are those made up of only two amino acids, whereas polypeptides are usually composed of 3-50 amino acid residues. For instance, phenyl-phenyl dipeptide given in fig. 1.4(a) is formed from two phenylalanine residues, while oxytocin peptide in fig. 1.4(b) is formed by different combinations of amino acid residues, as shown in the figure.

Fig.1.4: Structures of (a) phenyl-phenyl dipeptide and (b) oxytocin polypeptide.

Amino acids... Chapter 1

Peptide molecules remain attractive for their economic and medicinal values. Some peptides show inhibitory potencies against deadly diseases like cancers[46]. Amino acids, as the structural components of several peptide molecules, play significant roles in their structural and chemical properties. Amino acids also act as chemical messengers, metabolic intermediates and precursors of other bio-molecules[47-51]. Glycine and glutamate act as neurotransmitters and tryptophan is a precursor for indole acetic acid that serve chemical signaling role in the cells, for example[48-50].

Just 20 natural amino acids, which are differing in their side chains, are able to make such a diverse contribution to the biology of life. This remains one of the exciting phenomena that researchers from diverse areas are trying to unravel. In order to understand the systematic behind the nature’s choices in using the amino acids as principal ‘building blocks’ of life, it is important to reveal the structures and intrinsic properties of the amino acids. Although, amino acids have been studied for decades now, there are several molecular level details of the amino acids, which are still unknown or limitedly known. In this project, the electronic structures, properties and dynamics of the aliphatic and the aromatic amino acids under isolated and defined environmental conditions will be studied quantum mechanically. A rich tool chest of ab initio and density functional theory (DFT) methods will be employed.

1.3. Experiment and theory collaboration

Functionalities, properties and selectivity of biological molecules are greatly dominated by their electronic structures, conformations, charge distributions and their interactions with the environment[52]. From a chemical perspective, molecules are composed of electrons and nuclei connected by chemical bonds and as a result, molecular level studies are important to reveal the structures and properties of biomolecules. Spectroscopy based experimental methods and quantum mechanics (QM) based computational chemistry approaches have together been serving well as useful tools for probing the structures at molecular level.

Several experimental spectroscopy methods such as, photoelectron spectroscopy (PES), x-ray spectroscopy, mass spectroscopy, ultraviolet-visible spectroscopy, infrared spectroscopy (IR), Raman spectroscopy, nuclear magnetic resonance (NMR) spectroscopy,

Amino acids... Chapter 1

atomic absorption spectroscopy, rotational or microwave spectroscopy and electron momentum spectroscopy (EMS), etc. are employed to probe and understand the molecular structures. However, experimental measurements are not always feasible and are sometime difficult to observe high resolution spectra due to several technical issues that might occur under experimental conditions. Nevertheless, assignments of experimental spectra are not always straightforward and theoretical calculations are required to interpret the spectra in detail.

Quantum mechanics is probably considered the most important scientific discovery in the 20th century. Density functional theory (DFT) is one of the most versatile QM methods that is used to investigate the electronic structures of many-body systems like atoms, molecules and condensed phases. DFT methods have been successfully employed for studying the structural properties of several biomolecules such as amino acids[52-57], nucleic acids[58-60], nucleosides[61-65], several bioactive compounds[66, 67] and drug molecules[61, 62]. There have been significant advancements in the power of supercomputers and QM methods including post Hartree-Fock and DFT models that ensures more accurate studies of larger molecular systems.

1.4. Motivation of the project

This thesis takes the advantages of the state-of-the-art supercomputing facilities and synchrotron sourced spectroscopy to explore the electronic structures, dynamics and reactivity of amino acids, in order to unravel their structure-property relationships under isolated (gas phase) and environmental conditions. Our results are validated against the experimental data, wherever possible.

Amino acids have been subjected to a number of theoretical and experimental studies in the gas phase[52, 54-56, 68-84]. The gas phase studies are one of the efficient means to explore the properties that are specific to the molecules with little influence from the molecular environments[68, 85]. Until recently, experimental gas phase measurements have faced several challenges including technical issues such as to keep the molecules intact in the gas phase and lack of high resolution instrumentations[86]. Indeed microwave spectroscopic technique is able to make accurate measurements of small molecules in the gas phase.

Amino acids... Chapter 1

Furthermore, with improvements in the synchrotron sourced experiments and advanced theoretical methods coupled with high performance computational infrastructure[87], the attraction towards gas phase chemistry has escalated since last decade.

In the gas phase, neutral amino acids exist in different low-energy conformations; those are mainly dependent on the types of intra-molecular H-bonds and steric and torsional interactions[52, 54-56, 68-84]. Based on the literature, most of the conformers of amino acids fall under three major structural isomers[80, 81], isomer I, isomer II and isomer III, which

differ in the configurations of their amino acid moieties (NH2(HC-R)COOH) and their intra- molecular H-bonding patterns. Fig. 1.5 presents the chemical structures of the three isomer categories along with their intra-molecular H-bonds. Isomer I is identified as the lowest energy ground state conformer in almost all the studies reported earlier[53, 68, 74-78, 81-84]. This configuration is generally stabilized by bifurcated H-bonds from the amide to carbonyl groups (N-H…O=C) and cis-carboxylic (O-H…O=C) functional group interactions. In the case of isomer II, the amino hydrogen atoms are free from H-bonds and the hydroxyl hydrogen is bonded to the amino nitrogen (O-H…N). And in isomer III, only one of the hydrogen atoms from the amino group is H-bonded to the carboxyl oxygen (N-H….O=C) along with the bonds from cis-carboxylic (O-H…O=C) functional group[80, 84]. The proton on the carboxylate group of isomer I and isomer III exists in a resonance form, rather than being localized to a particular oxygen atom in the carboxylic acid moiety, as is the case in Isomer II. A number of conformers exist within these major classes of isomers and there might also be other isomers differing from these. However, the listed classes of isomers are in general lower in energies, as confirmed by previous studies[53, 68, 74-78, 81-84].

Fig.1.5: Structural representations of the three major isoforms of amino acids in the gas phase along with their intra-molecular H-bonds.

Amino acids... Chapter 1

This thesis is mostly based on the first two isomers, unless otherwise mentioned. The amino acid model molecules are investigated in the gas phase using DFT methods and compared to reveal the impact of the side chains on their molecular properties. A variety of spectroscopic properties such as geometries, ionization spectra, vibrational spectra, vibration dependent optical spectra, momentum spectra and charge re-distribution are calculated in order to probe the substitution effects at molecular level.

Amino acids are not static in the complex environment and they interact with different molecules to exhibit their respective biological functions. Especially, water, the most populated ubiquitous molecule in the human body[88], (with ~61% of total composition[89]) is considered as an integral part of almost all biological systems. It has been experimentally observed that the water molecules remain as interfaces between bio-molecules and are crucial for bio recognition and self-assembly[90]. Hence interactions of amino acids with water molecules are important in nature, a few of the amino acids are ‘water-disliking’ (or hydrophobic) in character though. In addition, metal ions are also important components interacting with amino acids within the biological systems. The cation-π interaction is considered as a crucial driving force in molecular recognition within biological systems. Extensive discussions on cation-π interactions are available in the literature[91-93]. Hence amino acids, metal ions and water molecules remain the most important molecular compounds in the biological systems. Their interactions govern several chemical reactions such as electron transfer or deprotonation reactions[25, 94, 95], neutral-to-zwitterion transitions of amino acids[96, 97] protein-protein interactions, drug-protein interactions, protein folding/unfolding and aggregation processes. Studying the interactions and dynamic behaviours of these molecules, therefore, remains of significant interest. In the chapter 6 of this thesis, we study the effects of copper (II) (i.e., Cu2+) binding on the structures of phenylalanine under micro-solvated environment using the ‘state-of-the-art’ ab-initio molecular dynamics methods (AIMD).

Classical molecular dynamics (MD) employs classical Newtonian equations of motion for a system. The equations are solved numerically starting from a pre-informed initial state under a set of boundary conditions appropriate to the problem. The quality of any dynamics simulations is highly dependent on how the forces are specified. As a result classical MD simulations receive many successes. However, these force fields do not take

Amino acids... Chapter 1

covalent bond making and breaking effects into account, as a result, the chemical reactivity is hardly accurately predicted using force field based methods.

–Robert S. Mulliken.

AIMD generates dynamical trajectories by using the forces obtained from the electronic structure calculations that are performed at every steps of simulation. AIMD is therefore able to produce more accurate bond breaking and formation reactions that includes electronic polarization effects[98, 99]. Nevertheless, AIMD calculations are more significantly computationally expensive. Yet in the modern era of supercomputers, massively parallel computing facilities extend support towards such large simulations. In 1985, Prof. Car and Prof. Parrinello presented a revolutionary MD scheme, named Car-Parrinello molecular dynamics (CPMD) method[100], that uses pseudopotentials and plane wave basis sets for conducting QM-MD simulations. Thus CPMD is able to fulfill the need for a combined implementation of classical MD and quantum chemistry. CPMD code is used in this thesis to study the interactions of phenylalanine and Cu2+ under micro-solvation process.

1.5. Overview of the thesis

The thesis mainly aims at,

1. Investigating the electronic structure property relationships of amino acids in the gas phase that can reveal the potential information on their shapes and spectra.

2. Exploring on how these bio-molecules tend to interact with environment such as metal ions and aqueous solution, using accurate QM based molecular dynamics simulations.

Chapter 2 briefly describes the various molecular properties and the theoretical methods that are used to investigate the electronic structures of amino acids.

Chapter 3 presents the comparative investigations on different properties of the aliphatic amino acids in the gas and solvent phases. The aim of this chapter is studying the effects of alkyl side chains on the structures and various properties of the aliphatic species[57]. Different properties including geometry, ionization spectra and chemical bonding

Amino acids... Chapter 1

characters in the gas phase along with the vibrational and vibrational optical activity spectra of these amino acids in the gas and solvent phases are discussed[101]. Our results in this chapter show high level of correlation against various experimental data.

Chapter 4 reports the intra-molecular interaction mechanisms of phenylalanine using the structures of its derivatives and fragments. The chapter concentrates on the effects of

functional group substitutions on the C(-C(carbon bridge of phenylalanine moiety. Here

the C denotes the central chiral carbon of phenylalanine that connects the amino, carboxyl

and the side chain groups of the amino acid, while C site connects the amino acid moiety and the phenyl ring of phenylalanine. The ‘fragments-in-molecules’ approach employed in this chapter explores different fragmentation schemes of phenylalanine to access their molecular level information. The study reveals appropriate fragment pairs that dominate the intra-molecular behaviors of phenylalanine.

Chapter 5 investigates the role of the hydroxyl groups on aromatic molecules, from the phenylalanine (R=phenyl ring) and tyrosine (R=phenol group) amino acids to the L-dopa (R=catechol group), a precursor of dopamine neurotranmetter as well as a common drug for the treatment of Parkinson’s disease. Building on the knowledge on structure-properties of isolated phenylalanine acquired from previous chapters, this part of the work addresses significant physical-chemical differences among these molecules, whose structures differ only in the OH group substitutions. Various properties including geometries, ionization, charge re- distributions and aromaticity are studied using the DFT based methods. The intrinsic properties that differentiate the L-dopa drug against the natural amino acids are particularly discussed.

Chapter 6 combines the DFT calculations and CPMD dynamic simulations to study the structures of phenylalanine-copper (II) complexes under micro-solvated environment. Extensively rich information of the structures, energetics, H-bonds, molecular orbitals and dynamics of the phenylalanine-copper (II) complexes are discussed.

Finally, overall thesis summary and future directions are presented in chapter 7.

Methods… Chapter 2

CHAPTER

2 Methods and computational details

2.1. Introduction

This chapter will present a brief overview of the theoretical methods, which are applied in order to investigate structures, properties and dynamics of molecular systems in this thesis. The chapter will discuss concisely the different measurable properties that are calculated for the model molecules and the choice of relevant computational models for the calculations.

Probing the electronic structures of molecules is crucial for understanding their properties, functionalities and reactivity[102, 103]. Computational chemistry, that integrates mathematical approximations and computer programs for its investigation, has been serving well for this purpose[104-107]. Especially quantum mechanics (QM) remains a useful field to study the electronic properties of molecules at their molecular levels for solving various chemical problems. QM, therefore, has been widely recognized as one of the most appealing approaches in chemistry such as as well as in other areas of science, material sciences and medicinal sciences. While different methods in computational chemistry such as statistical mechanics, molecular mechanics (MM) and cheminformatics are based on semi-empirical or empirical information provided by a user, QM methods are based on the first principles. This

Methods… Chapter 2

gives QM approaches the advantage of high reliability and intrinsic accuracy. Needs for better approximations and powerful computers are significant challenges for the applications of QM based methods to large biological systems.

A molecule is made of atoms through chemical bonds and interactions. An atom consists of nuclei and electrons, with electrons moving under the influence of the electromagnetic force exerted by the nuclear charges. QM methods calculate the interactions among nuclei and electrons in a molecule, to reveal their electronic structure properties. The main approaches to describe the electronic structures include wavefunction based and density based methods. In the wavefunction based methods, an approximation for the actual wavefunction is constructed to predict the molecular properties. But in the density based approaches, the electron density is considered as the fundamental variable to probe the properties. However, all the QM based electronic structure methods aim at one equatorial point of finding solution, i.e., to solve the Schrödinger equation.

2.2. Time independent Schrödinger equation

The development of quantum mechanics (QM) began in the 1920s, when the Austrian physicist, Erwin Schrödinger, formulated an equation to describe the changes in the quantum state of a physical system[108]. The QM methods in chemistry particularly try to solve the time independent Schrödinger equation, where time is not considered explicitly,

(2.1)

Here (‘psi’) is the wavefunction of the system, ‘E’ is the total energy and is the Hamiltonian operator associated with the energy. The Hamiltonian operator contains all the terms contributing to the energy of the system including the kinetic energy term and the potential energy term .

(2.2)

For a system consisting of M nuclei and N electrons, which are described by the position

vectors RA (for nuclei) and ri (for electrons), the Hamiltonian operator can be written as,

Methods… Chapter 2

(2.3)

In Eq. (2.3), the MA is the ratio of the mass of the nucleus A to the mass of an electron th and ZA is the atomic number of the nucleus A. The distance between the i electron and the th th th A nucleus is riA=|ri-RA|, the distance between the i and j electrons is rij=|ri-rj|, and the th th distance between the A nucleus and the B nucleus is given by RAB=|RA-RB|. ‘ ’ represents

the Laplacian operator and and denote the Laplacian operators of the electron i and the nucleus A, respectively. The first two terms in the equation represent the kinetic energy terms of the electrons (N) and nuclei (M), respectively. The third term denotes the Coulomb attraction between electrons and nuclei and the last two terms are the repulsions between the electrons and between the nuclei, respectively.

2.3. Born-Oppenheimer approximation

Solving the Schrödinger equation can reveal directly the molecular structures, energies and bonding information. However, the Schrödinger equation can be exactly solved only for the simplest one-electron system (hydrogen atom). It cannot be solved for multi- centre and many-electron systems due to a number of complications. One such difficulty includes the coupled motion of electrons and nuclear particles in larger systems, except for simple systems such as helium and lithium atoms by using different methods[109, 110]. In order to overcome the many-body challenges, a number of approximations become necessary to solve the Schrödinger equation, approximately.

The approximation proposed by Max Born and Robert Oppenheimer in 1927, known as Born-Oppenheimer (BO) approximation[111], is the very first approximation applied to solve Schrödinger equation. This approximation is based on the fact that the masses of the nuclei are much heavier than electrons. The proton, itself, is ~1837 times more massive than an electron. Therefore, the nuclei are expected to move very slowly when compared to electrons and the swiftly moving electrons are instantly able to react to the movement of nuclei and adjust themselves with the nuclear fields. Thus in most cases, it can be considered as if the electrons are moving in a field of fixed nuclei. Based on this assumption, the

(2.2)

Methods… Chapter 2

movement of the electrons and nuclei can be separated. The electronic Hamiltonian, becomes,

(2.4)

The electronic eigenvalue problem becomes,

(2.5)

Solving each of the electronic eigenvalue problem at various nuclear configurations gives the so-called potential energy surface of a molecule.

Although the BO approximation is able to solve the problem, it is certainly not the universally accepted, as this approximation by itself is still not sufficient to face many-body challenges. For instance, this BO approach is well known to break upon the existence of multiple potential energy surfaces that are close to each other in energy[112]. This makes the need for more accurate approximation methods indispensable in quantum chemistry.

“The fundamental laws necessary for the mathematical treatment of large parts of physics and the whole of chemistry are thus fully known, and the difficulty lies only in the fact that application of these laws leads to equations that are too complex to be solved.” -Paul Dirac

This visionary statement made by Paul Dirac in 1929[113], still remains a motivating quote to computational chemists. As a result, various approximations have been developed and implemented for electronic structure calculations under different conditions. The Hartree- Fock (HF) based and the density functional theory (DFT) based methods are the most widely accepted methods in quantum chemistry.

2.4. Hartree-Fock theory

The BO approach simplified the problem of solving the Schrödinger equation by

Methods… Chapter 2

treating electronic and nuclear motions separately. The next concern is the electronic problem in the many-body systems. The HF theory assumes that electrons behave as individual particles and do not interact with each other[106, 107]. Therefore the spirit of HF theory is that each electron will move in the potential field that is induced by all other electrons and nuclei. In this assumption, the electronic wavefunction of a many-electron system would simply be a product of all the one-electron wavefunctions (individual one-electron wavefunctions usually known as ‘orbitals’) resulting in an Hartree Product (HP)[106] equation,

(2.6)

However, this wavefunction is not anti-symmetric with respect to the exchange of two particles and therefore, do not meet the basic requirements for an electronic wavefunction. The total wavefunction can now be written in the form of single determinant so that is anti- symmetric,

(2.7)

where a spin orbital χ(x) is the product of the spatial orbital ( (r)) and the spin function (σ(ω)),

(2.8)

The spin orbitals can be considered orthonormal to simplify the calculation.

The determinant in Eq. (2.7) is for a closed shell with N electron system, a Slater determinant[102, 114], which satisfies the anti-symmetry principle. As a result, the HF method reduces a wavefunction of a 3N coordinated system into N wavefunctions. The HF models result in varied successes and failures. The most significant drawback of the HF approach arises from its poor description of the electron correlation[105, 115]. HF models, therefore, are unable to describe accurately the phenomena where electron correlation effects are important such as thermochemistry, etc.

Methods… Chapter 2

Considering the fact that electronic motions in a system are correlated, there are a number of methods available to include the electron correlation effects, in order to improve their accuracies. Most of these correlation methods represent the configurations from the occupied and unoccupied orbitals obtained from HF calculations. Further the wavefunctions in these methods are constructed as linear combinations of multiple Slater determinants orthogonal to each other[105]. The major electron correlation methods include configuration interaction (CI)[116], coupled cluster (CC) theory[117] and perturbation theory. As a result, these methods are referred to post-HF methods.

2.5. Density functional theory

The post-HF methods rely on the many body wavefunction as the central quantity to solve the equation and predict the properties of molecules. But the wavefunction is a complicated measure as it depends on the 3N spatial variables (N is the number of electrons) along with the spin variable. This strictly limits the system size that can be treated with the wavefunction based methods. Hence, a different method based on an alternate quantity was noteworthy and the density functional theory (DFT) was formulated.

DFT methods are based on electron density (ρ(r)) as the central quality[118]. The major advantage of using the electron density as the basic variable over the wavefunction is the reduction of dimensions. The foundation of DFT is based on the Hohenberg-Kohn theorems[119]. In 1964, Hohenberg and Kohn postulated their first theorem that there is a one-to-one mapping between the electron density and the external potential. This theorem demonstrates that electron density can be used as a measure to define the ground state energy and electronic properties of a system. Later in their second theorem, Hohenberg and Kohn described the variational principle can be used for finding the energy and the ground state density.

Bright Wilson, after hearing a presentation of Hohenberg and Kohn theorem, said, “I understand that the density tells you everything; the cusps of the density tell you where the nuclei are, the gradient of the density at the nucleus tells you what the nucleus is, and the integral of the density tells you how many electrons you have – therefore you have specified everything about your system and Hamiltonian and hence all is known.”(source Ref. [120])

Methods… Chapter 2

Hohenberg and Kohn confirm the existence of a functional to relate the electron density with the energy of the system. However, information on how to obtain the electron density and the functional was unknown. One year later, in 1965, Kohn and Sham developed a practical scheme to solve the Hohenberg and Kohn theorem using a fictitious system of non-interacting electrons[121]. The Kohn Sham (KS) scheme assumes that the densities of an

interacting system with an external potential (υext) and its corresponding non-interacting system are exactly the same. The KS scheme therefore presumes that the effective potentials

(υs) control the motions of a non-interacting system such that its ground state density is unchanged, when compared to the actual interacting system. The single particle Schrödinger or KS equation that can be written as,

(2.9)

where the effective one-electron potential is

(2.10)

Here the first term is an external potential, the second term is the Hartree term arising from the static Coulomb interaction of the electron density and the last term describes the

exchange-correlation potential. The effective one-electron potential υs is a functional of the density and is therefore specific to individual systems. On the other hand, the exchange-

correlation potential (υxc) depends only on the electron density and thus, has the same form for any system. The major advantage of the KS scheme is that a large part of the kinetic energy can be calculated directly, and everything that is not known exactly is accounted by the exchange-correlation potential[102, 103, 105].

2.6. Exchange correlation

“Density functional theory is in principle exact! But, in practice approximations have to be made. ” -W. Kohn

As insisted by W. Kohn, better approximations of the exchange correlation functional, are important to obtain more accurate results using the DFT. The exchange correlation

Methods… Chapter 2

functional is particularly important to obtain the exact ground state energy and the density of

the many-body problem[118]. The exchange-correlation energy (Exc) generally includes two

separate terms, an exchange term (Ex) and a correlation term (Ec). The Ex associates with the

interactions between the electrons on the same spin while the Ec represents the interactions of the electrons on opposite spin. Thus exchange correlation can be described as,

(2.11)

However, the exact form of exchange correlation functional cannot be determined so that some approximations are required at various levels. There are number of approximations developed for exchange correlation functionals that include local density approximation (LDA) and generalized gradient approximation (GGA). The quality of the DFT results usually depends on the quality of the functional applied, the properties interested and the size of the molecules[24, 122-125], etc.

The LDA is the simplest exchange correlation functional. It is based on the assumption that the exchange correlation energy at any point in space is a function of the electron density at that point in space only and can be given by the density of a homogenous electron gas of the same density. Several formulations of LDA have been developed. The first LDA approximation for the exchange energy, named Thomas-Fermi-Dirac method, was proposed by P. Dirac in 1930[126] and was used together with the Thomas-Fermi[127, 128] method. The local spin density approximation (LSDA) initially proposed by Slater[129], includes a more general application of the LDA by introducing the spin dependence into the functional. Another well-known LDA is the functional formulated by Vosko, Wilk and Nusair, known as VWN[130] and the local correlation functional developed by Perdew

(PL)[131]. It has been noted that LDA typically underestimates the Ex while overestimating

the Ec, thus unexpectedly providing good results in some cases. However LDA displays some serious limitations such as underestimating the atomic ground state energies and ionization energies. As a result improved functionals such as GGA have been proposed.

The GGA approach not only considers the density (ρ) but also the gradient of the density (i.e., ρ). Thus this method is able to give improved results by including also the contributions from other atoms in the form of gradient of the density. The GGA functionals can give more accurate results than the LDA functionals for a number of properties such as

Methods… Chapter 2

geometries, ground state energies and modeling H-bonds[132]. Some popular GGA functionals such as the Becke88[133], Becke-Perdew86[134, 135], Perdew86[134] and Perdew-Wang[136] functionals produce accurate results for molecules. More advanced versions of GGA, meta-GGA, have also been developed by including additional semi-local information from the higher order density gradients and/or kinetic energy density, which involves derivatives of the occupied KS orbitals. The developments of meta-GGAs show significant improvements over GGAs in determining the properties. Several meta-GGAs for the exchange and/or correlation functionals have been developed[124, 137].

Aside from these pure DFT functionals, hybrid functionals have also been developed. Hybrid functionals combine the exchange-correlation of a DFT functional with some percentage of HF exchange[138]. The exact amount of HF exchange is usually determined by semi-empirical fitting. One common approach to do this is by fitting the coefficients to the experimental terms such as atom energies, ionization potentials (IPs), proton affinities and other data for a set of small molecules selected as representatives[124, 138]. The well-known B3LYP (Becke, three-parameter, Lee-Yang-Parr)[139, 140] functional can be written as,

(2.12)

where, a0, ax and ac are empirical parameters with values 0.20, 0.72 and 0.81 respectively.

and are GGAs formulated from the B88 exchange and LYP correlation

functionals. is the VWN correlation functional[130]. Hybrid functionals, especially B3LYP, have shown significant improvements over purely DFT functionals in calculating a number of molecular properties. A comparative study on ~600 neutral and organic compounds verified the performance of B3LYP functionals[141]. B3LYP has been realized to provide more accurate geometries and frequencies for different molecules ranging from small alkanes to amino acids, nucleosides, bioactive compounds and other molecules. As a result, the B3LYP functional is employed in this thesis to study the molecular properties, such as geometries and frequencies, of the amino acids.

Other exchange correlation potentials with correct asymptotic behaviours such as the statistical average of orbital potentials (SAOP)[142] and the LB94 potential[143], developed by van Leeuwen and Bearends, have been employed in this thesis, in order to accurately

Methods… Chapter 2

calculate the vertical IPs of the amino acids[56, 57, 67, 125, 144].

2.7. Basis sets

Apart from the functionals, basis sets are also important for performing more accurate DFT calculations. Mathematically, the wavefunction ( ) can be described as a linear combination of known functions, called the basis functions or basis sets,

(2.13)

where Cik represent the expansion coefficients of the basis function, . The essence of using basis functions lie in choosing the appropriate functions that resemble the atomic orbitals and at the same time computation of integrals are also efficient. Appropriate basis sets are generally selected based on the size of the system, properties to be studied and the availability of computational resources. The type of basis set could influence the accuracies of the results.

There are two major types of basis functions such as, Slater type orbitals (STO) and Gaussian type orbitals (GTO), which are commonly employed in molecular calculations. STOs were initially developed by Slater in 1930[145]. The STOs are derived from analytic solutions of a one electron hydrogen atom. A standard STO basis functions are in the following form[102],

(2.14)

where, is the spherical harmonic, n, l and m are quantum numbers, is the exponent, r denotes the nucleus-electron distance and N as the normalization factor. The STOs have been mostly employed for atomic and diatomic systems and they are not suitable for three- and four- centred atoms, due to mathematical reasons. Despite the multi-center two electron integral issues, STOs display some significant advantages. For instance, a recent study on spin-state energies of iron complexes has shown that the STO basis sets provide consistent results and also converge rapidly when compared to other basis sets[146].

Methods… Chapter 2

In recent years, a variety of STO basis sets have been developed by optimizing the total atomic energy[147] and implemented in the ADF program[148]. For example, the valence triple zeta with doubly polarized functions and core double zeta (TZ2P) basis set, even-tempered (ET) double zeta with one polarization function (DZP), valence quadruple zeta with three polarization (QZ3P) basis set and valence quadruple zeta with polarization function (pVQZ) basis set, etc. are available in ADF. Even-tempered basis sets have a number of advantages such as fewer optimization parameters and can be easily and systematically developed using the independent variation of all orbital exponents[147]. Moreover, it is relatively more convenient to study the approach to the basis set limit with them, due to the limited risk of over completeness issues[24, 147]. In this thesis, we have employed a number of STOs that are available in the ADF such as, TZ2P and et-pVQZ basis sets[147]. The combinations of these STOs along with the exchange correlation functionals, LB94 and SAOP, have provided accurate results for a number of molecular systems with excellent agreements with the experiments[56, 57, 67, 125, 144].

Another important type of basis set is the Gaussian type orbitals (functions), GTO, which allow the evaluation of integrals to be performed analytically. The use of GTOs was

first proposed by Boys[149], where the from STO is replaced by the in GTO. A GTO basis function can be represented as,

(2.15)

A wide variety of GTOs have been developed. The simplest atomic orbital representation using GTOs is the minimal basis set, which consists of only one basis function for each occupied orbital. More than one primitive Gaussian functions can be used for each orbital in order to increase the flexibility and accuracy of the basis set, i.e.,STO-nG. STO-3G basis set and STO-6G basis set are examples of minimal basis sets that are made up of 3 Gaussians and 6 Gaussians for each orbital, respectively. Further improvements in GTOs have also been achieved by including the polarization and diffuse functions to the GTOs. In molecules, the electron distribution is often different from atoms. An isolated atom exhibits spherical symmetry, but the same atom within a molecule or some other chemical environment exhibit different behaviours with distortion in their electron density. In order to consider this effect in the basis sets, one needs to augment the basis sets with additional

Methods… Chapter 2

functions containing larger angular momentums, that is, to include the polarisation functions and diffusion functions in the basis sets.

Pople style basis sets are examples of including polarisation and diffusion functions[150-152]. For example, the basis set 6-311G is a Pople style basis set that represents triple split valence basis set where, the core orbital is described by one contraction composed of 6 primitive Gaussians and the valence is split into three parts with 3, 1 and 1 primitive Gaussians respectively. To include polarisation effects, the ‘*’ symbol is added to the basis set such as 6-311G* and 6-311G**. Here the former basis set with one * include only a d function from non-hydrogen atoms, while the latter (with **) includes both a d function from non-hydrogen atoms and a p function from hydrogen atoms. In the similar way, the diffusion functions can also be added to the basis set. For instance, in the Pople basis sets, the ‘+’s are added to the basis set for diffuse functions. The basis set of 6-311+G includes diffused functions for non-hydrogen atoms and 6-311++G includes diffuse functions for both hydrogen and non-hydrogen atoms. The inclusion of diffusion function generally improves the description at large distances from the nuclei. This is particularly effective while modelling systems with weakly bound electrons, such as anions and excited states. Other basis sets that are popular in quantum chemical calculations include correlation consistent basis sets such as Dunning’s basis sets[153-156] namely, cc-pVDZ (3s2p1d), cc-pVTZ (4s3p2d1f) and cc-pVQZ (5s4p3d2f1g), etc. In Dunning’s basis sets, the numbers of contracted Gaussian functions are provided within the brackets. Similar to Pople style basis sets, Dunning’s basis sets can also be augmented to include diffuse functions --- those are represented with a prefix ‘aug’ --- aug-cc-PVXZ, for instance.

In quantum chemical calculations, there exists a dogma of larger the basis set more accurate the results[24, 122]. However simply choosing a larger basis set and including all polarisation and diffuse functions might not guarantee more accurate results. Thus careful selection of basis set is important to properly address the issues. A number of GTOs including the triple zeta valence polarized (TZVP) basis set and 6-311++G** have been employed in this thesis for the ground state geometry optimization and vibrational and vibrational optical activity calculations of the amino acids.

Methods… Chapter 2

2.8. Molecular properties

The aim of a molecular study is to understand a number of physical and chemical properties of a molecule, as a result, various models and approximations are introduced to predict molecular properties. One of the challenges is to choose an appropriate QM model (i.e., theory and basis set) to investigate the structures and properties of molecules. Several earlier studies have highlighted the significance of model selection on the properties. For instance, Wang et al[122, 123] tested the performance of different models in combination with various GTO and STO basis sets using simulated orbital momentum distributions of water[123] and nitrous oxide[122]. These studies clearly demonstrate that DFT functional, basis sets and specific combinations of DFT functionals and basis sets have apparent effects on the behaviors and the qualities of the calculated orbitals, and therefore, affect all the results of molecules. Another study also discusses the performance levels of different available functionals in predicting the chemical properties[124]. It stresses the importance of balancing the possible aspects of a chemical problem, such as the specific properties, the type of a system and computational expense, before carrying out any DFT study[124]. Moreover, a recent review by Cohen et al[120] compares the performance of a wide range of functionals in predicting several properties including geometries, H-bonds, polarizabilites, thermochemistry, energy barriers and van der Waals interactions etc. It is concluded that most of the currently available functionals are unable to predict the delocalization and static + correlation errors, in H2 and H2 systems, for instance[120]. In this thesis, we have made a careful selection of DFT models and basis sets to investigate different properties of amino acids.

2.8.1. Optimized geometries

In QM, energetically stable molecular structures are associated with the geometries of minimum energy structures, which are obtained by minimizing the total electronic energy as a function of intrinsic coordinates of the nuclei. As a result, geometric optimization is employed to locate the energy minima on the potential energy surface (PES) of molecular systems[157]. The optimization procedure is a mathematical procedure, which usually computes the energy and its gradient at a given configuration (or a point on the PES). There are a number of optimization methods such as Nelder-Mead simplex[158], gradient descent,

Methods… Chapter 2

conjugate gradient[159], Newton-Raphson method, quasi-Newton methods, rational function (RF) optimization method[160], direct inversion in the iterative subspace (DIIS), etc[160, 161]. Among these, the RF and DIIS methods are widely used in computational chemistry packages

Molecules, such as amino acids, can generally exist as ensemble of conformations (i.e., local minimum structures) depending on the temperature and other conditions. But, not all the structures are energetically favorable in all directions (coordinates) and some structures called saddle points are minima in some directions but maxima in other directions (coordinates). Mathematically, if an optimized geometry is a true minimum (stable) structure, all the second derivatives (i.e., frequencies, physically) of the PES must be non-negative. Stable (minimum) structures of a molecule are free from any imaginary frequencies (i.e., negative second derivatives mathematically). Computational chemistry packages such as Gaussian 03[162] (G03) and G09[163] employ the second derivative methods.

The experimental results from the literature are employed to validate the calculations in this thesis. However, it is noted that experimental measurements are based on different principles and can suffer from inconsistency due to various accuracy issues. Experimental geometry measurements can be obtained from different techniques including X-ray diffraction, neutron diffraction, microwave spectroscopy, electron diffraction, etc.[105, 157], which may not necessarily exhibit the same error bars. For example, the X-ray diffraction experiment depends on the scattering of photons by the electrons around the nuclei, while the microwave spectroscopy measures the rotational energy levels based on the nuclear position and the electron diffraction is based on the scattering of electrons by nuclei. As a result, the average C-H bond distances obtained in X-ray diffraction method can be 0.95 Å[164], while the same C-H bond is given by 1.08 Å in the neutron diffraction technique[164]. This is partly because, the X-ray measurements principally identify the areas of high electron densities and it is not exactly same as the nuclear positions in the hydrogen atoms[115 ]. As a result, the bonds involving hydrogen atoms are typically shortened in the X-ray experiment. At the same time, it is worthy a note that the measurements of gas phase molecules using microwave spectroscopy have been known to be extremely accurate and are most comparable to quantum chemical-derived geometries.

For molecular geometries, crystal structures of molecules are usually used in the

Methods… Chapter 2

measurements. The obtained geometries are not exactly the same as in gas phase conditions. For example, molecules in the crystal conditions are tightly packed with each other and experience strong intermolecular interactions due to ‘crystal packing forces’, whereas in gas phase, molecules are isolated and do not interact with each other. Moreover, differences with the molecular conformations may be much greater in gas phase, which in turn result in differences of the overall molecular shape between the gas phase and crystal structures[115]. Further, quantum mechanically, energies are in general less sensitive to changes of angles and dihedrals[53, 57]. As a result, calculated bond lengths of molecules usually agree better with the experiment than bond angles and dihedral angles. The discrepancies of bond angles and torsional (dihedral) angles vary between 1-5°[115].

2.8.2. Ionization energies

When an electron is removed from a molecule, the molecule becomes ionized and the energy required to remove the electron from the molecule is the ionization energy (or ionization potential, IP or binding energy). Removal of electrons lead to several ionized states, depending upon where the electrons are removed. Removing an electron from an outer valence orbital of a molecule can result in ionized states in the energy range of tens of eV above the ground state. However, removal of an electron from the core orbital of the same molecule can lead to ionized states further away in hundreds of eV, which can be ten times higher than the actual energies of valence states[165]. Moreover, chemical environment will cause small energy differences of the same element in a molecule, i.e. the chemical shift. This makes the core ionization interesting to study the local properties of molecules based on the chemical shifts of core IPs of the molecules.

A number of experimental techniques such as, X-ray photoelectron spectroscopy (XPS or ESCA, electron spectroscopy for chemical analysis), etc. provide measurements for IPs of molecules. XPS is one of the most widely used techniques, which is based on the photoelectric effect[166, 167] and can be given as,

(2.16)

Here, IP represents the ionization potential of an electron in an atom or molecule, is the

Methods… Chapter 2

Planck’s constant and is the frequency used for ionization. The kinetic energy of the ejected

electron (i.e ) can be measured with precision and therefore, the IPs can be determined. The first experiment using XPS was carried out by Robinson and Rawlinson (in 1914)[168]. The field of photoelectron spectroscopy is advancing rapidly. Equipped with ‘state-of-the-art’ instrumentation powered by synchrotron sourced spectroscopy, it is now possible to measure ionization energies of both valence and core orbitals of more complex molecules. The experimental measurements pose a number of difficulties such as lack of high resolution output and assignment of spectral peaks. For example, the measured spectra can be congested, which brings difficulties in the analyses of the spectra. As a result, theoretical support is significant in spectroscopy.

The most important quantum mechanical approach for calculating the ionization energies is the Koopman’s theorem. According to Koopman’s theory within the HF formulation, the positive eigenvalue of each occupied orbital is approximately equal to the vertical IPs of the electron in the orbital[169]. Approximations based on this approach are able to give reasonable valence IPs of molecules. Koopman’s theorem involves two important limitations that have to be considered when comparing its calculated results with the experiments. The primary concern relates to the lack of electron correlation effect in the HF method. The electron correlation energy is small, but at the same scale as chemical reaction energies. This energy leads to the over estimation of the IPs. As a result, other methods such as post-HF methods including MP2, CI, CCSD(T), etc. and DFT methods are developed to take into account the electron correlation energy. The second concern is that this theorem assumes the orbitals in the ionized state is same as those in unionized state, i.e., the orbitals are frozen. Such an assumption completely ignores the fact that orbitals in the ionized state will be different from those in unionized state, due to additional relaxation effects. Therefore the core IPs of the ionized state are obviously expected to be different from that of the unionized form. As a result, SCF methods are used. If the DFT formalism is used, it is called the DFT method.

In this thesis, the E-KS method developed by Cavigliasso and Chong[170] is used for small amino acids, in order to calculate their core IPs more accurately. In the E-KS method, the core IPs are calculated as the difference of the total KS energies between the ionized cations and the neutral molecule. The E-KS method in this thesis uses the et-pVQZ

Methods… Chapter 2

basis set along with the Perdew-Wang 1986[171] exchange functional and the Perdew-Wang

1991[136] correlation functional and further adding a relativistic correction (Crel) term. Due to the consideration of the relaxation effects, this method is able to give improved accuracies with experiments[172]. However, this method is very computationally expensive, so that it is hardly applied to larger molecules, such as tyrosine.

For larger molecules, the DFT methods are employed. However, the physical meaning of KS orbitals is subjected to a debate in the DFT theory. Janak’s theorem[173], similar to Koopmans’ theory, exists only for the first IP and the electron affinity, i.e. the energies of the highest occupied molecular orbital (HOMO) and the lowest unoccupied orbital (LUMO). This theorem, therefore, can only be applied for extraction or addition of an electron to the HOMO. As a result, another method similar to Koopman’s theorem has been developed for DFT and dubbed “meta-Koopman theorem”[174], which is able to estimate the IPs of larger molecules. This approach is based on the theory that interprets the energies of the occupied KS orbitals as approximations for the vertical IPs of orbitals[174 , 175]. As indicated earlier, models with correct asymptotic behavior such as SAOP and LB94 are able to provide a good approximation to the exact KS potential and hence can be used to predict the vertical IPs. This SAOP model along with et-pVQZ or TZ2P functionals are used for the calculation of the valence ionization spectra of amino acids in this work.

In addition to the DFT models such as SAOP, the propagator method with Green function, the outer-valence Green’s function (OVGF)[176, 177] has also been used for valence IP predictions. The OVGF model produces more accurate IPs in the outer valence space of a molecule, those agree very well with the experiments. However, as the name indicates, the OVGF model is only applicable for the outer valence space of molecules, where single particle approximation is a good approximation. The OVGF model is unable to calculate the inner valence IPs. The SAOP model, on the other hand, can provide the IPs for entire valence space including the inner valence IPs of a molecule. Another correct asymptote model, LB94 is used in this thesis for core IPs. Previous studies indicate that the calculations of both SAOP (for valence space) and LB94 (for core space) provide satisfactory agreements with the experiments in most cases, but after very small IP shifts[56, 57, 67, 125, 144].

Methods… Chapter 2

2.8.3. Atomic charge distribution

Charge distribution within a molecule is considered as a key measure that determines all molecular properties. They are primary indicators of several chemical reactions arising from several factors such as chemical bonds (bond formation and breaking), environmental effects, inter- and intra-molecular H-bonds. Thus analyzing the charge distribution is significant to understand molecular structures. A theoretical description of charges generally uses a population analysis scheme, which partitions the electron density between the nuclei so that each nucleus has a number of electrons associated with it. Such a partition provides a way to calculate the atomic charge on each nucleus. However, the important difficulty with atomic partial charges is that the atomic charges are themselves fictitious and not observables[178]. As a result, there are established criteria for defining the quantity of the atomic charges.

There are a number of population analysis methods such as Mulliken population analysis[179], Veronoi deformation density method[180], Bader analysis[181], molecular electrostatic potential based charges[182, 183], restrained electrostatic potential charges[184, 185 ] and Hirshfeld scheme[186 ]. Unfortunately, no single partitioning scheme has been identified to be efficient in all the aspects and therefore, the charges may vary according to the partitioning scheme selected for the analysis[187]. The Hirshfeld charge scheme is applied in this thesis.

In Hirshfeld analysis, the molecular charge at each point is divided among the atoms of the molecule in proportion to their respective contributions to the pro-molecular density. Pro-molecular density may be defined as the difference between the molecular and the un- relaxed atomic charge densities. Thus similar to ‘partners in a stakeholders’ corporation, each atom in the Hirshfeld scheme involves in either local gain or loss[186]. Hirshfeld analysis therefore is sometimes referred to as “stakeholder” population analysis[188]. Moreover, this method has been widely implemented for studying the electron charge densities in the X-ray diffraction studies[189]. As the Hirshfeld analyses are able to depict the charge distribution as the local effect of molecular environment, we calculate the Hirshfeld charges of amino acids using the LB94 model that is used for predicting core IPs – another measure for local molecular effect.

Methods… Chapter 2

2.8.4. Momentum distribution and DSA

Electron momentum spectroscopy (EMS) is a unique technique in molecular spectroscopy that provides information of chemical bonding mechanisms of a molecule by measuring its binding energies and the orbital electron density distribution[190, 191]. Moreover, EMS is sensitive to the outer valence electrons, which makes it an efficient approach to study the chemical effects in molecules. A number of theoretical approximations such as Born-Oppenheimer approximation, one electron approximation and the plane-wave impulse approximation (PWIA) are considered. The overlap of the initial and the final electron wavefunctions is dubbed as the Dyson orbital or quasi particle orbital[192] . The cross section of EMS for randomly oriented molecules can be written as,

(2.17)

where K is a kinematical factor which remains constant in the experiments, and are many-body electronic wavefunctions for the final ion f and the target molecular i ground electronic states, respectively. Further ‘p’ is the momentum of the target electron at the instant of ionization and averages all the molecular orientations.

These Dyson orbitals can reveal the changes that take place in response to the detachment of an electron from a molecule. The Dyson orbitals can be approximated by a molecular orbital using the same one-particle model for the target and ion.

(2.18)

Here is the spectroscopic factor that is the probability of one electron configuration in

the final wavefunction and represents the KS orbitals in momentum space, with p being the momentum of the target.

The momentum distributions of the electrons directly verify the probability of the velocity distribution of the electrons, while the position space provides the electron density of the molecule that determines probability of electron position with respect to the nuclei[191].

Methods… Chapter 2

Therefore both position space and momentum space information can complement each other to understand the intra-molecular mechanisms of a molecule. Shape of the orbital momentum distribution of a particular molecular orbital (as momentum spectra) can be related to its orbital diagram (i.e., density) to interpret the chemical reactions occurring in that orbital. This technique named as “dual space analysis or DSA”[193] has been proven efficient for several biomolecules. Therefore the DSA approach is used in this thesis to probe the electronic structures of amino acids. In this thesis, the KS orbital wavefunctions of amino acids are calculated using the single point calculations performed with the B3LYP/TZVP model in G09 or G03 computational chemistry package. The predicted wavefunctions are Fourier transformed into momentum space using the NEMS code[194] of Tsinghua University.

2.8.5. Vibrational spectroscopy

Vibrational spectroscopy is one of the most widely used techniques for elucidation of molecular structures[195-201]. It provides vital information about the molecular conformations, isomerism, molecular rotation, H-bonds, inter- and intra-molecular forces, distortion of molecules, etc. Thus vibrational spectroscopy becomes a routine tool in structural chemistry, polymer chemistry, catalysis, material research, drug discovery, etc[200- 204].

Infrared (IR) and Raman spectroscopy are the two important techniques that are used for vibrational analysis[198, 200, 201, 205,6 20 ]. Both techniques are based on the two different types of energy exchanges between the target molecules and the electromagnetic radiation. The IR spectroscopy measures transitions between molecular vibrational energy levels that result from the absorption of the infrared photons[198, 203]. This interaction between the light and the matter is a resonance situation that involves the electric dipole- mediated changes between the different vibration levels. The complementary technique, Raman spectroscopy, is a two-photon inelastic light-scattering process[207-209]. In this approach, the incident photon has greater energy than that of the vibrational quantum energy. The incident photon loses a part of its energy to the molecular vibration and the remaining energy is scattered as photon with reduced frequency. Unlike IR, the interaction between

Methods… Chapter 2

light and matter in Raman technique is an off-resonance condition involving the polarizability of the molecule.

Given that the IR and Raman are two different techniques operating on different processes and different selection rules, they display an excellent complementarity[198, 200, 203]. Some vibration modes observed in the IR spectra are not essentially active in Raman spectra and vice versa. Generally, IR bands are active for the asymmetric vibrations of polar groups, while Raman bands are sensitive for the symmetric vibrations of non-polar groups[198, 203]. Therefore, both these techniques together can offer an effective tool to investigate the vibrational modes of molecules and unravel their structures. DFT methods have advanced to calculate the IR and Raman spectra to satisfactory accuracies. In this thesis, the IR and Raman spectra of the aliphatic amino acids are calculated in both the gas phase and the solvent phase, using the B3LYP/6-31++G** model in Gaussian 09[162, 163] program.

2.8.6. Vibrational optical activity (VOA) spectroscopy

Chirality[210] is an important phenomenon in different areas of science[210-213]. A molecular structure can be chiral, if it exists in two distinct mirror image forms. Lord Kelvin states (in 1904)† chirality as,

“I call any geometrical figure, or group of points, chiral, and say it has chirality, if its image in a plane mirror, ideally realized, cannot be brought to coincide with itself.”

The chirality is mostly caused due to the presence of an asymmetric carbon atom, which connects four different atoms or groups. The amino acids (except glycine) are chiral in nature that is facilitated by the presence of central asymmetric carbon atom. When a hydrogen atom in glycine, the smallest amino acid, is replaced by different side chain ‘R’ group, the produced amino acids become chiral and optically active. Indeed, some amino acids such as isoleucine have more than one chiral carbons. As a result, understanding the roles of side chains in the chiro-optical properties of the amino acids is important to

† Lord Kelvin stated this celebrated definition in 1904, in his Baltimore Lectures on Molecular Dynamics and the Wave Theory of Light. Presently, this statement is universally accepted as the definition of chirality.

Methods… Chapter 2

understand their structures.

A number of spectroscopic approaches have been developed for studying the chiro- optical features of chiral bio-molecules, including electronic optical activity[202, 214], fluorescence optical activity[215-217], optical rotatory dispersion[218, 219] and vibrational optical activity (VOA)[202, 220]. These approaches are sensitive to the configurations of the systems and are able to study the chirality using circularly polarized light. The VOA technique is applied in this thesis.

The VOA measures the differential responses of a molecule to the left and right circularly polarized radiations during a vibrational transition of chiral molecules[202, 220- 223]. There are two principal forms of VOA approaches, the vibrational circular dichroism (VCD) --- the IR form --- and Raman optical activity (ROA) --- the Raman form. VCD can be defined as the differential absorption ( of the left (L) and the right (R) circularly polarized IR radiation for a vibrational transition. ROA, on the other hand, is the differential Raman scattering of left and right incident and/or scattered radiations. VCD and ROA techniques show different kinds of sensitivity against stereo-chemical structures. The VCD bands are sensitive to the dipole coupling between the nearby groups, forming positive- negative couplets. On the other hand, ROA is sensitive to local effects and does not show large impacts, where VCD bands are intense. As a result, VCD and ROA are complementary to each other[220 , 224-228], similar to their parent approaches, IR and Raman spectroscopy, respectively. As a result, the combination of IR/VCD and Raman/ROA spectra can be useful to comprehensively study the vibrational and chiro-optical properties of bio-molecules.

Theoretical calculations of VCD and ROA using the first principle methods have significantly advanced in the recent years[220, 226, 227]. For example, the recently simulated ROA spectra of L-alanine and L-proline achieved excellent agreement with the relevant experiments, after applying anharmonic corrections such as VSCF, PT2 and VCI[229]. In this thesis, the IR/VCD and Raman/ROA spectra of the aliphatic amino acids are calculated using the DFT based B3LYP/6-31++G** model. The aim is to understand the vibrational and chiro-optical properties of the amino acids in response to the side chains in the gas (neutral amino acids) and aqueous phases (zwitterionic amino acids). The conductor-like polarizable continuum model (CPCM)[230] (water solvent model) is used to study the properties of the zwitterionic amino acid in aqueous phase. More details about solvent effects and models are

Methods… Chapter 2

discussed in the section 2.10.

2.9. Ab-initio molecular dynamics

The properties and methods described above are for studying the electronic structures and properties of amino acids in gas phase. However amino acids are in dynamic molecular environment. Therefore studying dynamical behaviors of amino acids are vital to understand their roles in macromolecular systems such as proteins. Molecular dynamics (MD) is a technique to model detailed microscopic dynamical behaviors of many different systems in chemical biology. As indicated in the introduction section, despite classical MD remains a popular approach for simulating the dynamics of larger systems, it does not include electron effects such as electron correlation and electron polarization effects. Classical MD, therefore, is not suitable for studying the chemical reactions such as bond formation and bond breaking in the process.

To overcome such difficulties, first-principle (or ‘ab-initio ’) molecular dynamics that combine the capabilities of DFT with the classical MD was introduced in 1985 by Car and Parrinello[100] – commonly known as Car-Parrinello MD or CPMD. CPMD simulations are parameter-free MD simulations, in which all the interactions within atoms/molecules are calculated on-the-fly using the DFT framework. In this way, the finite temperature and entropic effects are combined with the context of quantum chemical electronic structure calculations. Through this fusion, a variety of new features that goes far beyond the individual capabilities of classical MD or DFT becomes possible in CPMD. For example, CPMD approach includes the electronic wavefunctions explicitly in the calculation and propagates them in parallel with the motions of atoms[100, 231]. This is done by considering the one-particle orbitals as fictitious classical degrees of freedom that evolve under the laws of classical motions.

In CPMD, the Lagrangian formulation is employed. The Lagrangian is described as[232],

(2.19)

Methods… Chapter 2

where K is the kinetic energy and Epot is the potential energy. However, for molecular systems that combine both the nuclear and electronic coordinates, the Lagrangian can be

extended (Lex)[232] as,

(2.20)

Here KN is the kinetic energy of the nuclei and Ke is the kinetic energy of the electronic degrees of freedom. By including a fictitious mass to the electronic degrees of freedom, the nuclei and the electronic density can be disseminated in time, hence avoiding the need to solve KS equations for every MD time step. The special Lagrangian postulated by Car and

Parrinello (LCP)[100] can be written as,

(2.21)

Here the first and the second terms denote the kinetic energies of the nuclei and

electronic counterparts, respectively. ‘MI ’is the mass of nuclei with the position vector, ‘RI’

and ‘µi’ represents the fictitious mass of the electronic degrees of freedom. µ as a purely fictitious term can be assigned an arbitrary value. Moreover this fictitious term determines the

rate at which the electronic variables evolve in time. For instance, the ratio of MI to µ describes that the relative speed at which the electronic variables propagate is relative to the

nuclear positions. The third term, , is the potential energy given by the expectation

value of the total energy of the system. in the final term represent the Lagrange multipliers that ensure orthonormality of the one electron wavefunctions. The advantage of including the Lagrangian multipliers is that the boundary conditions and constraints for a given system can be easily imposed. The equations of motion that correspond to the nuclear degrees of freedom is[100]

(2.22)

and for the electronic degrees of freedom[100],

Methods… Chapter 2

(2.23)

Most of the current QM molecular dynamics implementations employ the original CP scheme based on DFT, where the system is treated within the periodic boundary conditions (PBC). The KS orbitals therefore can be expanded in plane waves[100,2 23 ],

(2.24)

where G is the vector in reciprocal space that satisfies the PBC and Gmax is the maximum length of the G vectors that determines the size of the basis set. denotes the volume of the cell. Given that these functions are simple, they are very efficient to manipulate. However the major limitation is that for rapidly oscillating wavefunctions, a large number of plane waves may be required to properly describe the total wavefunction. These situations generally happen very near to the nucleus (core region), fortunately where the electron density is not of direct chemical relevance. This can be overcome by using a pseudopotential formalism[231, 233, 234] to describe the ionic cores, while the valence electrons are treated explicitly. In order to maintain the consistency of first principle character of CPMD, ab-initio pseudopotentials are used.

Ab-initio pseudopotentials are derived directly from atomic all-electron calculations and there are a number of different schemes available for the constructions of pseudopotentials[235-243]. Constructed pseudopotentials should have both the additive and transferability features[240, 241, 243]. The former can be easily achieved by constructing pseudopotentials for atoms in their reference states[243]. The latter ensures that a pseudopotential constructed for an atom should be compatible for the same atom in different chemical environments. One of the common way is to include the condition that, for a specific atomic reference configuration, all-electrons and pseudo wavefunction have similar eigenvalues and coincide outside of a given core radius[231]. Pseudopotentials with such conditions are generally termed as norm-conserving. The norm-conserving pseudopotentials provide an efficient and reliable approach for performing the calculations for complex molecular systems, including liquid and solid state systems. A variety of standard norm- conserving[236, 238] pseudopotentials, soft norm-conserving pseudopotentials[235, 237] and

Methods… Chapter 2

ultra-soft Vanderbilt[239] pseudopotentials are currently used in the CPMD simulations. The selection of pseudopotentials is important for reliable MD simulations. In this study, we use norm-conserving Troullier and Martins[237] pseudopotentials, unless otherwise mentioned.

2.10. Solvent effects and models

As indicated in the earlier sections, amino acids normally exist in solvent environment (especially water), which is important for several bio-chemical processes. The interactions with water solvents are known to change the different physical and chemical properties of the amino acids such as their energies, geometries, intra-molecular reactions, vibrational and optical properties, etc. Therefore, understanding the effects of solvent environment on the structure-properties of amino acids is equally significant with the gas phase studies of amino acids.

Modelling of solvent effects on the solute structure-properties can be carried out in two major approaches: explicit solvent approach or implicit solvent approach. In the explicit solvent approach, large numbers (usually in several hundreds) of individual water molecules are placed around the solute and the dynamic trajectories of water molecules along with the solute are studied[244, 245]. Despite the ability of this approach to provide information on the structure-properties of solute and the solvents in great detail, it suffers from considerable computational costs. Studying such a complex system with large number of explicit water molecules at purely QM level can be extremely time consuming and sometimes impossible with the current resources. On the other hand, molecular mechanics approaches that employ atomic force fields are able to handle such large systems at cheaper computational price, however, only with inadequate chemical details; MM approaches are unable to accurately describe bond breaking processes. Alternatively, microsolvation studies- where only very limited numbers of water molecules (usually upto 10 numbers) are placed around the solute structure- can be performed using cutting-edge QM approaches such as Car-Parrinello Molecular Dynamics method[100]. Such studies are becoming computationally feasible and are also able to provide some insights on the solvent effects on the solute structure and properties. Yet deciding on the number of water molecules that are actually required to study the micro-solvation effects on the solute is not straightforward. As a result, it often becomes

Methods… Chapter 2

necessary in the explicit solvent approach to choose between the accuracy and computational costs, where either of them needs to be compromised.

An alternative to the explicit approach is the implicit solvent method[246, 247], which is based on replacing the actual water molecules by an infinite continuum medium with the dielectric and hydrophobic properties of water. In this method, the solvent molecules are approximated by a homogenous dielectric medium, which is characterized by its dielectric

constant, 0 [248 , 249]. The solute in the continuum model is embedded in a cavity of certain shape and size, where it interacts with the solvent. Given that it can be derived from the electrostatic, the polarization of the solvent can be described in terms of the charge density on the cavity surface. When the surface charge density is obtained, the electric field in the cavity and its effects on the solute can also be calculated. The advantage of this approach is that the solute charge distribution and its response to the reaction field of the solvent dielectric can be modeled either using the QM approach or by partial atomic charges in the MM approach. Despite several levels of approximations, this dielectric continuum method has been one of the most successful approaches to account solvent effects at affordable computational time [248 , 249]. This implicit method does comes with obvious limitations, such as the ignorance of the effects from inter-molecular hydrogen bonds between the solute and the solvents, but where it works it allows excellent estimation of solvent effects. For example, the method has been known to provide a good account of equilibrium solvation energetic, pKs, redox potentials and other molecular and spectroscopic properties of the solute[246, 250, 251].

Different models have been developed based on this approach, depending on how the cavity is constructed and the electrostatic interactions between solute and continuum are treated. The Polarized continuum model (PCM)[252] is one of the most popular approaches that employs dielectric continuum method. PCM offers more realistic molecular shape of the cavity involved and the induced surface charges in the model represent very well the solvent polarization. Moreover, the PCM also includes free energy contributions from forming the cavity and dispersion-repulsion effects in the system. Different implementations of the PCM model have been developed, such as Dielectric PCM (DPCM), Integral equation formalism PCM (IEFPCM), Conductor like PCM (CPCM) [230], Isodensity PCM (IPCM)[253] and Self-consistent isodensity PCM (SCIPCM)[253]. These models are available is a number of

Methods… Chapter 2

computational chemistry programs, especially in the most used Gaussian 03/09 program[162, 163, 254].

This thesis employs three different ways to treat the solvent effects on the spectroscopic and dynamic properties of the amino acids: implicit solvent model, explicit micro-solvation model and QM/MM model. In the chapter 3, the implicit solvent model, CPCM has been employed in order to study the impacts of solvents on the vibrational and chiro-optical properties of the aliphatic amino acids. In this part of the work, the zwitterions forms of the aliphatic amino acids are initially optimized in the CPCM model, as employed in the G03 program and their IR/VCD and Raman/ROA spectra are obtained. The CPCM model has been chosen as it is one of the most successful solvent models that is known to accurately describe the aqueous solvation free energies in good agreement with the experimental data. Whereas in the chapter 6, limited numbers of water molecules (up to four) are added sequentially to the phenylalanine-copper (II) complex, in order to study the effects of micro- solvation on the dynamic process of the complex. This study has been carried out using the ‘state-of-the-art’ CPMD approach. Finally, the bulk solvation process of the aromatic molecules, phenylalanine and histidine’ are studied using the hybrid CPMD/MM implementation, where the solute is treated using CPMD and the solvents are described using the MM based force fields. Few of the results are provided in the appendix section.

Methods… Chapter 2

2.11. Summary of computational details

Table. 2.1 summarizes the properties simulated, computational methods, software programs for simulation and visualization and the supercomputing facilities used in studying the structure-property relationships of amino acids.

Table 2.1: Summary of computational details. Simulations Methods Code Visualization/processing

B3LYP/TZVP, G03[162], G09[163] Gaussview,[255] Geometry B3LYP/6-311++G** Molden,[256] optimization Chemcraft[257] B3LYP/TZVP, G03[162], G09[163] Gaussview,[255] Vibrational B3LYP/6-311++G**, Molden,[256] spectra B3LYP/6-31++G** Chemcraft[257] B3LYP/6-31++G** G03[162], G09[163] Gaussview,[255] VOA spectra Molden,[256] Chemcraft[257] LB94/et-pVQZ, ADF[148] ADFView,[148] Core ionization LB94/TZ2P, E- Origin[258] KS[170] SAOP/et-pVQZ, ADF[148] (for SAOP) and ADFView,[148] Molden, Valence SAOP/TZ2P, G03[162], G09[163] (for [256] Origin[258] ionization OVGF/TZVP[176] (for OVGF) outer valence) Hirshfeld LB94/et-pVQZ ADF[148] ADFView[148] charges Fourier transformed NEMS[194] Origin[258] Momentum B3LYP/TZVP distribution wavefunctions BLYP functionals with CPMD[52, 100] VMD[259] QM-dynamics Troullier Martin[237] pseudopotentials Supercomputing NCI, VPAC, Swinburne University’s Green machine, VLSCI, Juropa and Jugene facilities computers in FZ Juelich, Germany.

Aliphatic amino acids… Chapter 3

CHAPTER

Aliphatic amino acids

3.1. Introduction

Aliphatic amino acids have been the subject of numerous experimental and theoretical studies for decades[52, 53, 68-78, 80, 85, 260-269]. They contain a side chain of aliphatic

alkyl group such as glycine (R-H), alanine (R-CH3), valine (R-CH(CH3)2), leucine (R-

CH2CH(CH3)2) and isoleucine (R-CH(CH3)(CH2CH3)). Fig. 3.1 presents the chemical structures of the aliphatic amino acids. The aliphatic amino acids exhibit neutral forms (NT) in the gas phase, while zwitterionic (ZW) in the solvent and crystal phases.

Fig.3.1: Chemical structures of the aliphatic amino acids.

The electronic properties of the aliphatic amino acids in the gas phase have been studied previously, however, some results remain contradictory[53-56, 64, 270]. For

Aliphatic amino acids… Chapter 3

example, Klasinc[271] measured and assigned the valence spectra of some amino acids and suggested that the four outermost valence orbitals of the amino acids have dominant molecular orbital character of nitrogen lone pair, oxygen lone pair (s-like), oxygen lone pair

(p-like) and πCO, respectively. On the other hand, in a more recent study of the valence ionization spectrum of L-alanine, Powis et al[272] suggested that the third highest molecular

orbital of alanine was assigned to πCO, which was in agreement with an earlier measurement of Cannington and Ham[273]. Orbital 19a at about 14 eV was singled out as having particularly clear methyl character. Dehareng and Dive[274] calculated the ionization energies and orbital characters of the outer molecular orbitals of the compounds, and found that their orbital characters are dominated by contributions from the amino-carboxylic acid moiety (the main moiety), in agreement with Powis et al[272]. Some orbitals reveal certain contributions from the alkyl atoms, but not the dominant contributions. Detailed orbital information combined with their momentum space details remains limited.

In this chapter, quantum mechanical calculations are performed for the neutral (NT) aliphatic amino acids in order to study the effects of alkyl side chains on their ionization and chemical bonding characteristics in the gas phase. Information from the position and momentum spaces are combined using dual space analysis (DSA)[193]. Subsequently, the effects of alkyl side chains on the vibrational and chiro-optical properties of the aliphatic molecules in both the gas phase (NT) and aqueous phase (ZW) are investigated using the vibrational and vibrational optical activity (VOA) spectroscopies.

Vibrational spectroscopy is a powerful tool to reveal the structure and dynamics of biomolecules based on the molecular motions or vibrations of their structural components, such as their side chains and peptide backbones[275]. As molecular vibrations are very sensitive to inter- and intra-molecular interactions, these vibrational spectra are also useful in providing information regarding the potential binding sites among peptides, proteins and other large molecules[275-281]. For these reasons, IR and Raman spectroscopy have become important techniques to study amino acids in both the gas and aqueous phases[229, 266, 267, 275, 278, 281-292]. For example, calculated IR spectra are used to study different conformers of amino acids in the gas phase[289, 290], and also for deprotonated amino acids[287]. Very recently, Raman spectra of amino acids were used as a reference to analyze the spectra of collagen protein[281] within 600-1700cm-1.

Aliphatic amino acids… Chapter 3

Vibrational optical activity (VOA) techniques[226, 293], vibrational circular dichroism (VCD) --- the IR sensitive form --- and Raman optical activity (ROA)---the Raman sensitive form, complement each other in studying the stereo-chemical properties of biomolecules[220, 226, 286, 293-295]. Aliphatic amino acids are chiral in nature (except glycine), hence they became the target of a number of experimental and theoretical studies of the VCD and ROA spectral techniques[220, 229, 263, 278, 283, 291, 292, 294, 296-300]. As the smallest chiral amino acid, L-alanine, has been one of the most intensely studied molecules using VOA techniques for several decades[229, 263, 278, 283, 285, 291, 292, 299]. A study indicated that the COOH bending vibrations are relatively intense in the VCD spectra of neutral (NT) alanine in gas phase[278]. On the other hand, in solution, it is found that the main ROA spectral features of zwitterionic (ZW) alanine are very similar in different pH

conditions[283]. The ROA spectra of the alanine ZW form, and its isotopomers in the H2O

and D2O media, reveal that the stretching and rocking modes of the methyl moiety and the COO- moiety are responsible for generating the intense ROA peaks[292]. Moreover, Yu et al[292] found that the dominance of the methyl signals in the ROA spectrum of alanine reduces significantly when the methyl hydrogen atoms are replaced by deuterium atoms[292]. - As a result, the vibrations related to CH deformation and the COO moiety stereo-chemical correlations are responsible for the ROA signatures of amino acids[220, 283, 285, 296].

Fewer studies concerning the IR/VCD and Raman/ROA spectra of the amino acids and their behaviours in the gas phase and in solution have been reported, although a recent study focused on the IR/VCD spectra of amino acids in the gas phase only [278]. As a result, the latter part of this chapter presents a comprehensive ab initio study to calculate the VCD/IR and ROA/Raman spectra for the aliphatic amino acids. Our specific aim is to study the vibrational and chiro-optical properties of the amino acids in response to their side chains in the gas phase, and how large those changes persist when these amino acids are in aqueous solution. Please note that the structures of aliphatic amino acids used in this chapter are neutral (NT) in the gas phase and zwitterionic (ZW) in the solvent phases.

3.2. Methods and computational details

Except for glycine, the aliphatic amino acids are chiral, and the naturally occurring L- enantiomers of alanine, valine, leucine and isoleucine are calculated. The most stable NT

Aliphatic amino acids… Chapter 3

conformers of the aliphatic amino acids are initially optimized in the gas phase using the B3LYP/TZVP model[301]. Molecular wave functions obtained from position or coordinate space (r-space), using the B3LYP/TZVP model, are directly transformed into momentum space (k-space) as orbital momentum distributions[193] as described in Chapter 2. The theoretical momentum distributions are generated using the NEMS code[194]. Further the core level binding energy spectra of the optimized structures are obtained using the LB94/et-

pVQZ[143] and EKS methods[170]. Whereas, the OVGF/TZVP[302] and SAOP/et- pVQZ[303, 304] are employed for the valence energy calculations.

The aliphatic amino acid ZW structures are optimized using the CPCM (conductor- like polarizable continuum model)[230] water solvent model. The IR/VCD and Raman/ROA spectral calculations are performed in both the gas and solvent phases, while the VCD rotational strengths are calculated from Stephen’s equation[305, 306]. All theoretical calculations concerning the IR/VCD and Raman/ROA are carried out using density functional theory (DFT) based on the hybrid B3LYP[139 , 140] functional combined with the 6-31++G** basis set. The B3LYP model offers the most cost-effective choice for the calculation of molecular vibrational properties, as indicated by a recent study on vibrational spectroscopy[307] and our previous studies[54, 57, 308-310]. It is also known that B3LYP calculated vibrational wavenumbers and VCD/ROA intensities of alanine are in good agreements with a number of experiments[278, 292]. The Gaussian 09 (G09) computational chemistry program[311] (B3LYP and OVGF calculations) and ADF[148] package (SAOP and LB94 calculations) are used in this work.

3.3. Geometrical details

Fig. 3.2 depicts the lowest energy structures and the atomic numbering of the aliphatic amino acids that have been found to be the most stable conformers in previous studies[52, 68-

78]. The conformer of glycine in this work has the Cs symmetry. However, when a hydrogen

atom on C(2) in glycine is substituted by the alkyl groups to form the other amino acids, the

point group symmetry is reduced to a lower C1 point group. In aqueous solution, the proton – + from the carboxyl group is transferred to the amino group, thus forming COO and NH3 ions in the ZW forms of the amino acids. As a result, geometries of the ZW amino acids are re- optimized in the CPCM water solvent model. The numbering of the optimized ZW amino

Aliphatic amino acids… Chapter 3

acids is the same as those in NT structures (in fig. 3.2).

Fig.3.2: Optimized lowest energy structures of the neutral aliphatic amino acids and their numbering schemes.

Table 3.1 compares the optimized NT and ZW geometries of glycine and alanine with available experimental[312-314] and other theoretical results from the literature[263, 315]. As can be seen in the table, the calculated geometric parameters of both NT and ZW forms of glycine and alanine obtained in the present study are in good agreement with the experimental bond lengths and bond angles, except at certain angles. For instance, the maximum discrepancy in the bond lengths does not exceed 0.03 Å, whereas the bond angles show

slightly larger discrepancies. The deviations for the C(1)C(2)N are as large as 3.6 in NT alanine and 4.6 in ZW glycine. It is well known that the crystal structures in the experiments are ‘rigid’, rather than the flexible structures seen in solutions or in gas phase. In addition, quantum mechanically, energies of a molecule are in general less sensitive to variations in bond angles[57, 309].

Table 3.2 compares the geometrical parameters of the NT and ZW forms of the aliphatic amino acids. The general differences between the NT and the ZW forms are seen in – + the local region formed by the COO and NH3 moieties. This local region involves the O(1),

O(2), C(1), C(2) and N atoms (refer to fig. 3.1), in which the related bond lengths, such as C(1)-

C(2) and C(2)-N, increase from the NT form to the ZW form as we go from glycine to isoleucine, by approximately +0.05 Å.

Aliphatic amino acids… Chapter 3

Table 3.1: Selected geometric parameters of the NT and ZW glycine and alanine compared with the available experimental and other values from the literature.

Glycine (NT) Glycine (ZW) Alanine(NT) Alanine(ZW) This Other This [314] This This EXP[312] Other Work* EXP Exp[313] Other Work* Parameters Work Work* Work Work Work O(1)-C(1)/Å 1.36 1.36 1.35 1.27 1.25 (1.27) 1.25 1.36 1.35 1.25 1.25 (1.27)

O(2)-C(1)/Å 1.21 1.21 1.20 1.25 1.27 (1.26) 1.25 1.21 1.19 1.27 1.27 (1.26)

C(1)-C(2)/Å 1.52 1.52 1.53 1.55 1.55 (1.55) 1.53 1.52 1.51 1.56 1.56 (1.55)

C(2)-N/Å 1.45 1.45 1.47 1.50 1.50 (1.50) 1.48 1.45 1.47 1.52 1.52 (1.51)

C(2)-C(3)/Å 1.53 1.54 1.53 1.53

C 123.40 129.02 123.10 128.60

C C 115.60 115.60 113.00 107.32 102.59 111.80 113.70 110.10 105.73 105.60 (109.50) (107.90) CC 110.90 110.90 111.50 116.03 -116.10 117.50 111.40 110.30 116.41 -116.00

CC(2) 125.70 125.70 125.00 114.90 -115.30 117.10 125.40 125.60 115.02 -116.00

 109.60 111.34

CCC 108.40 113.72

CC 180.00 180.00 180.00 -179.90 -161.80 161.60 178.51

CC 0.00 0.00 0.00 0.10 18.90 -20.00 -1.00

CCC -76.20 -59.10

CCC 102.10 121.40 *B3LYP/6-31G(d)[263] (Values given in parentheses are from MP2/6-31+G**calculations[315]).

Aliphatic amino acids… Chapter 3

Table 3.2: Selected geometric parameters, from the present computations, of the aliphatic amino acids in the gas (NT) and solvent (ZW) phases.

Parameters Glycine Alanine Valine Leucine Isoleucine NT ZW || NT ZW || NT ZW || NT ZW || NT ZW ||

O(1)-C(1)/Å 1.36 1.25 0.11 1.36 1.25 0.11 1.36 1.25 0.11 1.36 1.25 0.11 1.36 1.25 0.11 O(2)-C(1)/Å 1.21 1.27 0.06 1.21 1.27 0.06 1.21 1.27 0.06 1.21 1.27 0.06 1.21 1.27 0.06 C(1)-C(2)/Å 1.52 1.55 0.03 1.52 1.56 0.04 1.52 1.57 0.05 1.52 1.57 0.05 1.52 1.57 0.05 C(2)-N/Å 1.45 1.50 0.05 1.45 1.52 0.07 1.46 1.52 0.06 1.45 1.52 0.07 1.46 1.52 0.06 C(2)-C(3)/Å 1.53 1.53 0 1.55 1.54 0.01 1.54 1.53 0.01 1.55 1.54 0.01 C(3)-C(4)/Å 1.53 1.54 0.01 1.53 1.54 0.01 1.53 1.55 0.02 C(4)-C(5)/Å 1.53 1.54 0.01 1.53 1.54 0.01

O(1)-C(1)-O(2)/° 123.41 129.02 5.61 123.12 128.59 5.47 123.02 128.52 5.5 123.14 128.51 5.37 122.90 128.47 5.57

O(1)-C(1)-C(2)/ 110.91 116.03 5.12 111.41 116.37 4.96 111.60 116. 47 4.87 111.51 116.54 5.03 111.53 116.58 5.05 C(1)-C(2)-N/° 115.56 107.32 8.24 113.67 105.66 8.01 113.04 105.41 7.63 113.60 105.40 8.2 113.10 105.31 7.79

N-C(2)-C(3)/° 109.64 111.32 1.68 111.12 113.08 1.96 110.16 111.81 1.65 111.22 113.29 2.07

C(1)-C(2)-C(3)/° 108.40 113.72 5.32 109.21 114.30 5.09 107.64 113.48 5.84 109.59 114.72 5.13

C(2)-C(3)-C(4)/° 110.03 111.90 1.87 114.35 116.22 1.87 111.09 112.80 1.71 C(3)-C(4)-C(5)/° 109.30 109.51 0.21 113.37 113.80 0.43

O(1)-C(1)-C(2)-N/° 180.00 -179.94 0.06 161.62 178.54 16.92 165.10 -179.63 14.53 161.30 179.78 18.48 168.04 -179.10 11.06

O(2)-C(1)-C(2)-N/° 0.00 0.10 0.1 -20.01 -1.00 19.01 -16.50 1.36 17.86 -20.41 0.61 21.02 -13.20 2.14 15.34

O(1)-C(1)-C(2)-C(3)/° -76.20 -59.10 17.10 -70.67 -54.83 15.84 -76.42 -57.52 18.9 -67.30 -53.81 13.49

O(2)-C(1)-C(2)-C(3)/° 102.10 121.37 19.27 107.70 126.18 18.48 101.81 123.31 21.5 111.50 127.44 15.94 C(1)-C(2)-C(3)-C(4)/° 169.30 172.74 3.44 -179.76 173.56 6.2 -65.41 -59.74 5.67

N-C(2)-C(3)-C(4)/° -65.31 66.61 131.92 -55.42 -67.36 11.94 60.43 61.27 0.84

C(2)-C(3)-C(4)-C(5)/° -179.61 172.60 7.01 171.78 167.09 4.69

Aliphatic amino acids… Chapter 3

It is noted that the C(1)=O(2) double bond lengths of the amino acids, however, stretch from 1.21 Å in the NT form to ca. 1.27 Å in the ZW form. The most significant change is that

the (H)O(1)–C(1) single bond lengths of 1.36 Å, of the amino acids in the NT form, reduce to

1.251.27 Å in the ZW form, a significant reduction of 0.11 Å. Moreover the (H)O(1)–C(1) and

C(1)=O(2) bonds connect at C(1), indicating that the HO(1)-C(1)=O(2) moiety in the NT form takes - a more delocalized network of O(1)···C(1)···O(2) in the ZW form. Other bonds, as we progress through the amino acids, such as the C-C bonds, show little changes.

The bond angles from our calculations experience apparent changes from the NT to ZW form in these amino acids, in agreement with previous studies[263, 284, 315]. Similar to the bond lengths, the bond angles which change most significantly are those involved in the local region of the main moiety. For example, the bond angles related to the local region such

as O(1)C(1)O(2), O(1)C(1)C(2), O(2)C(1)C(2) and C(1)C(2)N, all change in the same direction in going from their NT forms to the ZW counterparts. That is, if such an angle increases in going from its NT to ZW form in glycine, this angle also increases in all the other aliphatic

amino acids in table 3.2. For example, the O(1)C(1)O(2) bond angle of the ZW form is always

larger than its NT counterpart, due to loss of the H atom on the hydroxyl group HO(1), which

forms a possible hydrogen bond with O(2) in the NT form. The dihedral angles further show

that the four atoms, O(1), C(1), O(2), and C(2), are almost coplanar, as a result, the three angles

are related as: 360°–O(1)C(1)O(2) O(1)C(1)C(2)+O(2)C(1)C(2), in both the NT and ZW forms

of the amino acids in table 3.2. The angle C(1)C(2)N, between the NT and ZW forms of the amino acids, reduces by approximately 8°, depending on the aliphatic side chains involved. Other bond angles outside of the local glycine-ZW moiety region exhibit only small changes of less than 3°. The most significant geometric changes for the NT and ZW forms of the amino acids are between some of the dihedral angles, which do not follow the above discussed trends in bond lengths and bond angles.

Another notable observation regarding the geometries of NT and ZW aliphatic amino

acids concerns the changes in the C(2)–C(3) bond length and the N–C(2)–C(3) bond angle, in the amino acids (in glycine these bonds do not exist). Both these parameters in the NT and the ZW forms of leucine are smaller than their respective forms of valine and isoleucine (refer to table 3.2 for values). Such geometric trends are due to the effect of intra-molecular

steric repulsion from the methyl group attached to the C(3) of valine and isoleucine[316],

Aliphatic amino acids… Chapter 3

whereas in leucine, this methyl group is attached to C(4) (fig. 3.1). Thus, the geometrical parameters of the aliphatic amino acids indicate that the aliphatic alkyl chains play an important role in their NT and ZW forms.

Table 3.3 presents the selected molecular properties such as dipole moments, electronic spatial extents () and rotational constants, of the aliphatic amino acids in their NT and ZW forms. Available experimental and other theoretical results[73-75, 285, 312] are also given in this table for comparison purposes. It is the dipole moments of the amino acids that change most significantly, being approximately 10 fold larger in the ZW form compared to the NT form of the same compound. For example, the dipole moment of NT glycine is 1.25 Debye, which becomes 13.37 Debye in the ZW form in solution. Such a significant change in dipole moment agrees with results from a recent ab-initio based molecular dynamics simulation of glycine in water[317]. This effect is due to the change in the chemical bond natures --- from a covalent bond in an NT structure to an ionic bond in a ZW form. Furthermore, the dipole moments in the ZW form of the amino acids are mainly generated by the charge separation in the amino acid ZW backbone; as a result the effects in the dipole changes in the ZW forms are almost similar. It is interesting that in the smaller molecules such as glycine and alanine, the molecular size, i.e., electronic spatial extent (in a.u.), of the amino acids shrinks in going from the NT form to the ZW form, but expands in valine, leucine and isoleucine. For instance, the of glycine reduces from 424.23 a.u. in the NT form to 410.23 a.u. in the ZW form, whereas the size of isoleucine increases from 1387.77 a.u. in NT to 1418.87 a.u. in ZW. This is likely due to the branched alkyl side chains possessed by valine, leucine and isoleucine.

As we were able to accurately calculate the geometrical and other molecular properties of the aliphatic amino acids, these structures are used to study the effects of alkyl side chains on (i) the ionization and chemical bonding features of NT aliphatic amino acids in the gas phase and (ii) their vibrational and VOA properties in both the gas and aqueous phases.

Aliphatic amino acids… Chapter 3

Table 3.3: Selected molecular properties, from the present calculations, of the aliphatic amino acids in the gas (NT) and solvent phases (ZW). Corresponding experimental results, where available, are also shown.

Glycine Alanine Valine Leucine Isoleucine Parameters NT[312]* ZW NT[318]* ZW NT[73]* ZW NT[74]* ZW NT[75]* ZW

Dipole Moment 1.25 13.37 1.35 13.15 1.37 12.95 1.10 13.24 1.28 12.92 (Debye) (1.10) (1.80) (1.48) (1.14) (1.20)

(a.u.) 424.23 410.23 581.21 580.51 1046.61 1063.28 1572.06 1623.32 1387.77 1418.87

Rotational Constants A (GHz) 10.26 10.35 5.07 4.84 2.94 2.96 2.76 2.82 2.10 2.15 (10.34) (5.07) (2.98) (2.75) (2.09)

B (GHz) 3.87 4.06 3.05 3.36 1.44 1.46 0.85 0.83 1.11 1.09 (3.87) (3.10) (1.43) (0.85) (1.11)

C (GHz) 2.90 3.02 2.30 2.20 1.34 1.27 0.80 0.76 0.98 0.93 (2.91) (2.26) (1.32) (0.79) (0.97) *Values given in parentheses are experimental values from the corresponding references

Aliphatic amino acids… Chapter 3

3.4. Hirshfeld charge distributions

Hirshfeld charges[319] are important anisotropic properties, useful for understanding atomic behavior within molecules. Hirshfeld charges (QH) of the NT aliphatic amino acids, calculated based on the LB94/et-pVQZ wave function, are presented in table 3.4. From the table, it is noted that all the nitrogen and oxygen sites in the amino acids possess negative charges, whereas the carbon sites in the aliphatic amino acids display charge re-distribution

related to their respective side chains. The small negative charge on C(2) of glycine, –0.014 a.u., becomes positive in the other amino acids, while the other carbon atoms within the alkyl side chains are negative.

Table 3.4: Hirshfeld charges of the C, N and O sites of the aliphatic amino acids calculated using LB94/et-pVQZ model (values in a.u.).

Glycine Alanine Valine Leucine Isoleucine

C(1) 0.216 0.215 0.215 0.214 0.215

C(2) -0.014 0.025 0.023 0.021 0.023

C(3) -0.106 -0.014 -0.061 -0.016

C(4) -0.116 -0.117 -0.066

C(5) -0.116 -0.115 -0.113

C(6) -0.115 -0.116

O(H) -0.201 -0.198 -0.196 -0.198 -0.196 O(=C) -0.295 -0.296 -0.292 -0.299 -0.294 N -0.241 -0.24 -0.231 -0.23 -0.231

Although the carbon sites in the alkyl side chains of the amino acids possess negative charges, they still show some subtle differences depending upon their local interactions. For

example, despite the C(3) sites being negative in the molecules, the charges in valine and isoleucine are very small, with values of –0.014 and –0.016 atomic units, respectively. This is

because the weakly electron donating methyl groups attached at this carbon site (C(3)) in

valine and isoleucine lead to a charge redistribution, thereby making C(3) almost neutral. This observation is very similar to that of 1-(-D-ribofuranosyl)-5-methyl-2-pyrimidinone (d5)[64] which showed a reduced negative charge at the carbon site attached to a methyl group[64].

Aliphatic amino acids… Chapter 3

3.5. Ionization energy responses to side chain changes

3.5.1. Vertical core ionization energies & spectra

Table 3.5 compares the C 1s, N 1s and O 1s binding energies of glycine and alanine

calculated using the LB94/et-pVQZ model (not relaxed) and EKS (relaxed) methods along with the recent synchrotron sourced experimental values[85, 320]. In the first instance, it can be noted that the calculated values present an overall good agreement with the measured values.

Table 3.5: Core IPs of glycine and alanine calculated using the EKS (relaxed) and LB94/et-pVQZ methods are compared against the experimental data. Glycine Alanine Site MO MO EKS LB94 EXP[320] EKS LB94 EXP[85] OH(-C) 1a' 540.17 (0.03) 536.60 (3.60) 540.20 1a 540.10 (0.10) 536.60 (3.40) 540.00 O(=C) 2a' 538.29 (0.11) 535.03 (3.37) 538.40 2a 538.14 (0.06) 534.90 (3.30) 538.20 N 3a' 405.55 (0.15) 402.99 (2.41) 405.40 3a 405.30 (0.10) 402.90 (2.30) 405.20

C(1) 4a' 294.77 (0.43) 293.53 (1.67) 295.20 4a 294.58 (0.42) 293.41 (1.59) 295.00

C(2) 5a' 292.26 (0.04) 291.1 (1.20) 292.30 5a 292.10 (0.10) 291.18 (1.02) 292.20

C(3) 6a 291.01 (0.01) 289.76 (1.24) 291.00 7a

C(5) 8a

C(6) 9a *Values given in parentheses are the differences between the calculated and experimental values for respective atomic sites.

The energies calculated using the EKS method exhibits better agreement to the experiments, while the ‘meta-Koopman’ based LB94/et-pVQZ calculations show slightly larger deviations. For instance, the discrepancies in C 1s energies between experiment and the

calculated by the EKS model are ≤ 0.43 eV, while the difference in the LB94/et-pVQZ calculation is as large as 1.7 eV. Indeed the discrepancies become larger in the N 1s and O 1s sites of glycine and alanine. This indicates that the relaxation effects are considerably large in the inner shells of these molecules and as a result, the un-relaxed LB94/et-pVQZ model tends to underestimate the binding energies. Such systemic errors caused by the relaxation in the LB94/et-pVQZ model can be reduced by applying a global energy shift, as shown in our

previous studies[56, 64, 67, 321-323]. On the other hand, it is worthy to note that the EKS

Aliphatic amino acids… Chapter 3

method requires multiple calculations for the parent molecule and the individual cations, before being able to obtain the inner shell binding energies of each atom. Such a complex multi-step procedure remains a limitation with this method, while the DFT-LB94 method is able to provide IPs with reasonable agreements in a single calculation. Thus for larger bio-

molecules, where EKS method may not be suitable, the ‘meta-Koopman’ theorem based LB94 model, with a global energy shift, can satisfactorily reproduce the core ionization spectra.

Fig. 3.3 compares the O 1s, N 1s (a) and C 1s (b) spectra of alanine obtained using the

EKS calculations (without any energy shifts) against the recently measured synchrotron sourced XPS spectra[85]. A full-width at half-maximum (FWHM) value of 0.57 eV is applied to the calculated spectra in order to closely match the measurements. As it can be seen, the O 1s and N 1s spectra (fig. 3.3(a)) agree excellently with the experiment[85]. Most of the carbon sites in the calculated C 1s spectra given in fig. 3.3(b) are close to the experiment,

except the C(1) 1s peak, which is dominated by the C(=O). This peak in the computed spectra is ~0.42 eV away from the measured peak. This indicates that the present QM models are not efficient in accurately calculating the C=O energy that involves some extra electron correlations, as observed in azetidinone calculation[321].

Fig.3.3: (a) Comparison of the experimental and EKS calculated O 1s and N 1s spectra of L-alanine with a FWHM of 0.57 eV.

Aliphatic amino acids… Chapter 3

Fig.3.3: (b) Comparison of the experimental and EKS calculated C 1s spectra of the L-alanine with a FWHM of 0.57 eV.

Fig. 3.4 presents the O 1s, N 1s (a) and the C 1s (b) binding energy spectra of the

aliphatic amino acids calculated using the EKS method and their corresponding energies are provided in table 3.6. The O sites and the N sites (fig. 3.4(a)) are not very much affected by the alkyl side chain modifications, and hence their ionization potentials (IPs) exhibit small shifts to lower energy in both N 1s and O 1s spectra. Fig. 3.4(b) presents the C 1s spectra of the aliphatic amino acids simulated using a FWHM of 0.57 eV, reproducing the experimental condition of alanine[85]. The C 1s spectra are also marked using individual IPs of the carbon sites to indicate the spectral line positions. As seen in this figure, the energies of the peaks fall

into three classes: C(1)=O, C(2)-N and the alkyl C atoms. Basically, the C(1) 1s and C(2) 1s peaks in the amino acids are similar to the spectra of glycine, with small energy shifts to lower energy. The alkyl C atoms form the structure dependent C 1s spectra in the lower energy side of C 1s with IP < 291 eV. This is apparent from the energy gaps between the alkyl carbon atoms present in the peak < 291 eV. An energy based correlation diagram of the C 1s sites of the aliphatic species is provided in the appendix, A.I.

Aliphatic amino acids… Chapter 3

(a)

(b)

Fig.3.4: Comparison of the (a) O 1s, N 1s and (b) C 1s spectra of the aliphatic amino acids calculated

using the EKS method with a FWHM of 0.57 eV.

Aliphatic amino acids… Chapter 3

Table 3.6: Core IPs of the aliphatic amino acids calculated using the EKS (relaxed) and LB94/et- pVQZ methods. Glycine Alanine Valine Leucine Isoleucine Site MO MO EKS LB94 EKS LB94 EKS LB94 EKS LB94 EKS LB94 OH(-C) 1a' 540.17 536.60 1a 540.10 536.60 540.01 536.50 539.96 536.50 539.97 536.50 O(=C) 2a' 538.29 535.03 2a 538.14 534.90 538.01 534.90 537.94 534.90 537.97 534.90 N 3a' 405.55 402.99 3a 405.30 402.90 405.13 402.80 405.16 402.80 405.08 402.80

C(1) 4a' 294.77 293.53 4a 294.58 293.41 294.43 293.40 294.34 293.30 294.39 293.40

C(2) 5a' 292.26 291.1 5a 292.10 291.18 291.77 291.00 291.79 291.00 291.72 291.00

C(3) 6a 291.01 289.76 290.77 290.10 290.62 289.80 290.62 290.00 7a 290.41 289.20 290.53 289.70 290.51 289.50

C(5) 8a 290.55 289.30 290.47 289.20 290.58 289.30

C(6) 9a 290.40 289.20 290.34 289.20

3.5.2. Valence ionization and spectral properties

Valence shell IPs directly link the valence electronic structural information with molecular orbital theory, and provide details of intra-molecular interactions within the molecules. Table 3.7 compares the vertical valence IPs of the aliphatic amino acids, obtained using SAOP/et-pVQZ and OVGF/TZVP models, along with the available experimental[271, 320] and other theoretical data[272]. The experimental values provided for valine, leucine and isoleucine are measured by Klasinc in 1976[271] at temperatures 200-220° C. A number of previous studies have recognized the SAOP/et-pVQZ model as an accurate model for valence space calculations[54,3 ,30 324-326], and the present IPs for the amino acids are also in good agreement with available results. However, those earlier studies also contended that the SAOP model is not more accurate than that of the OVGF model for frontier orbitals including the highest occupied molecular orbital (HOMO)[52, 54, 325, 327]. This is similarly true in the case of aliphatic amino acids.

As seen in table 3.7, the SAOP/et-pVQZ model slightly overestimates the ionization energies of the HOMOs. For example, the HOMO in glycine (16a’) has a calculated energy of 10.53 eV and 9.89 eV using the SAOP and OVGF models, respectively, compared with the measured value of 10.0 eV[320]. However, the OVGF model only applies to the outer valence space of a molecule, but is unable to calculate the valence IPs at energies larger than 20 eV, where the SAOP model becomes attractive, as it is able to calculate the IPs in the complete valence space of a molecule, which agrees more accurately with the experiment.

Aliphatic amino acids… Chapter 3

Table 3.7: Valence orbital ionization energies (eV) of the aliphatic amino acids calculated using the OVGF and SAOP models together with other theoretical and experimental energies. Glycine Alanine Valine Leucine Isoleucine Orbital SAOP Exp Orbital SAOP ROVGFa Orbital SAOP EXP Orbital SAOP EXP[27 SAOP EXP (OVGF)# [320] (OVGF)# (OVGF)# [271] (OVGF)# 1] (OVGF)# [271] HOMO 10.53(9.89) 10.0 HOMO 10.49(9.75) 9.58 HOMO 10.43(9.59) 9.56 HOMO 10.45(9.61) 9.65 10.41(9.56) 9.67 (16a') (24) (32) (36) 15a’ 11.56(11.29) 11.2 23 11.43(11.08) 10.77 31 11.37(10.93) 10.73 35 11.36(10.92) 10.92 11.34(10.96) 10.83 4a” 12.78(12.29) 12.2 22 12.71(12.14) 11.88 30 12.23(11.63) 11.25 34 12.08(11.40) 11.99(11.20) 3a” 13.61(13.57) 13.7 21 13.15(12.81) 12.58 29 12.36(11.80) 11.92 33 12.11(11.61) 11.65 12.14(11.60) 11.85 14a’ 14.42(14.73) 14.4 20 13.60(13.43) 13.22 28 12.63(12.18) 32 12.27(11.63) 12.26(11.76) 13a’ 15.04(14.95) 15.0 19 13.74(13.74) 13.59 27 12.79(12.23) 12.63 31 12.68(12.13) 11.98 12.74(12.16) 12.14 2a” 15.57(15.61) 15.8 18 14.68(15.09) 14.82 26 13.43(13.47) 30 13.00(12.74) 13.01(12.74) 12.60 12a’ 16.60(16.61) 16.6 17 15.09(15.00) 14.84 25 13.55(13.50) 29 13.38(13.12) 13.16 13.39(13.41) 11a’ 16.85(17.05) 16.9 16 15.66(15.66) 15.37 24 14.35(14.34) 28 13.58(13.59) 13.59 13.52(13.23) 1a” 17.11(17.42) 17.6 15 16.49(16.84) 16.60 23 14.54(14.61) 27 13.89(13.79) 13.92(13.90) 14.00 10a’ 19.46 20.2 14 16.66(17.17) 17.00 22 14.78(14.86) 26 14.43(14.60) 14.36(14.31) 9a’ 22.45 23.3 13 17.20(17.57) 17.30 21 15.16(15.07) 25 14.85(14.93) 14.68(14.97) 8a’ 26.80 28.3 12 18.97 19.64 20 15.70(15.66) 24 15.02(14.93) 15.03(14.96) 7a’ 30.33 32.3 11 21.03 22.03 19 16.54(16.89) 23 15.51(15.64) 15.30(15.24) 15.24 6a’ 32.57 34.3 10 23.22 24.56 18 16.72(17.15) 22 15.58(15.48) 15.83(15.86) 9 26.88 17 17.16(17.46) 21 16.52(16.84) 16.53(16.82) 16.78 8 30.28 16 18.18(18.92) 20 16.61(17.01) 16.68(17.13) 7 32.53 15 19.54 19 17.15(17.48) 17.05(17.30) 14 21.24 18 18.09(18.83) 18.04(18.75) 13 22.07 17 19.03 19.29 12 24.14 16 20.85 20.13 11 26.86 15 21.18 21.81 10 30.26 14 22.77 22.60 9 32.52 13 24.432 24.34 12 26.88 26.84 11 30.22 30.24 10 32.47 32.50 #SAOP/et-pVQZ model. Energies based on the OVGF/TZVP model are given in parentheses; aROVGF/cc-pVDZ[272]. #A systematic error associated with the measurement is estimated as a factor of 2[320].

Aliphatic amino acids… Chapter 3

Fig. 3.5 compares the valence ionization spectra of glycine, calculated using both the SAOP and OVGF models, with a recent high resolution gas phase synchrotron radiation based photoelectron spectroscopy (PES) measurement of Plekan et al[320]. The calculated valence binding energy spectra of glycine have been globally shifted in order to align the first experimental ionization energy at 10.0 eV[320]. The PES spectrum calculated using the OVGF model has been blue shifted by IP= 0.20 eV, whereas the energy shift is IP= – 0.42 eV for the SAOP model. After taking account of these energy shifts, the OVGF and SAOP models agree well with the measurement in the valence space. Indeed the OVGF model reproduces the measured data in the outer valence space very well, whereas the SAOP model is able to continue providing a good estimation in the region beyond 20 eV. This is clearly seen in this figure. Thus the combination of the two models, SOAP/et-pVQZ (for inner valence regions) and OVGF (for outer valence regions), is suitable for a more extensive valence space calculation.

Fig.3.5: Comparison of the experimental photoelectron spectra of glycine in the outer valence space with the theoretical spectra using the OVGF/TZVP model (lower panel,  = 0.20 eV) and the SAOP/et-pVQZ model (upper panel,  = – 0.42 eV). The calculated spectra are convoluted with a FWHM of 0.70 eV.

Fig. 3.6(a) compares the outer valence vertical ionization spectra of the aliphatic amino acids, calculated now using an FWHM of 0.70 eV and the OVGF model. The valence spectra are more complicated than the core spectra, as valence electrons are more delocalized that core electrons. In addition, there are many more valence electrons than core electrons.

Aliphatic amino acids… Chapter 3

The alkyl side chain effects on the valence shell are different depending on the structure of the chain. For example, all the alkyl side chain dominant spectral peaks in the C 1s spectra appear below 291 eV as shown in fig. 3.4(b). However, alkyl effects on the valence spectra mix with the main groups in the energy region between 12 eV < IP < 16 eV, as shown in fig. 3.6(a). The inner valence spectral region of IP > 16 eV is dominant by the main groups again. As a result, the alkyl signatures of the aliphatic amino acids concentrate in the energy region of 12 eV < IP < 16 eV, as well as the decrease of the HOMO-LUMO energy gaps, as shown in fig. 3.6(a). Indeed, the spectra of the aliphatic amino acids seem similar in the frontier orbital region of IP < 12 eV and in the higher energy region of IP >16 eV.

The HOMO and the next HOMO (NHOMO or HOMO-1 orbital) of the amino acids appear within the valence energy range < 12 eV. The similarities in their IPs indicate that the spectral changes due to the side chains of these amino acids are not in the frontier orbital region, in agreement with previous results[52, 55]. This is due to the fact that the HOMOs and NHOMOs of the amino acids are not localized on the alkyl moieties. In the higher energy region of IP > 16 eV, similarities appear again. In this region, the small spectral peak at ~ 19 eV is missing in glycine and alanine. As a result, this peak can be associated with the growth and branch of the aliphatic chain in valine, leucine and isoleucine. The energy gaps between the HOMOs and the lowest unoccupied molecular orbitals (LUMOs), or HOMO-LUMO gaps, become smaller with increasing size of the alkyl side chain of the molecules, that is:

HOMO-LUMO (glycine) > HOMO-LUMO (alanine) > HOMO-LUMO (valine) > HOMO-LUMO

(leucine) > HOMO-LUMO (isoleucine) as also shown in fig. 3.6(a). Thus, the side chains contribute to the energy reduction of the HOMO-LUMO gaps, which is also observed in 1- (β-D-ribofuranosyl)-5-methyl-2-pyrimidinone (d5) when compared to its parent molecule, 1- (β-D-ribofuranosyl)-2-pyrimidinone (zebularine)[64].

Fig. 3.6(b) displays the inner valence (>16 eV) energy diagram of the aliphatic amino acids calculated using the SAOP/et-pVQZ model. Similar to the frontier orbitals, the innermost valence orbitals > 24 eV are also very similar in all the amino acids. The first two innermost orbitals are dominated by the carboxyl groups and the third innermost orbitals are mostly dominated by the amino group. The alkyl chains have little contributions to these orbitals and hence they do not show many differences in their energy levels. Nevertheless, the energy region between 18 – 24 eV display certain orbital splitting, indicating that these

Aliphatic amino acids… Chapter 3

orbitals, despite being different, they show certain chemical associations. In summary, the side chain effects on the aliphatic amino acids are largely revealed in the energy region of 12 eV < IP < 16 eV of the binding energy spectra indicating strong chemical bond involving alkyl contributions in this region.

Fig.3.6: (a) Comparison of the outer valence vertical ionization spectra of the aliphatic amino acids calculated using the OVGF/TZVP model, convoluted with a FWHM of 0.70 eV. (b) Inner valence energy diagram of the aliphatic amino acids calculated using the LB94/et-pVQZ model.

Aliphatic amino acids… Chapter 3

3.6. Valence electron momentum spectra using DSA

Position space analyses provide orbital information in a qualitative manner, whereas momentum space studies are able to reveal subtle information about orbitals and therefore, chemical bonding. The complementary nature of information in position and momentum space is revealed by dual space analysis (DSA)[193]. Fig. 3.7(a)-(d) present selected valence orbitals including frontier orbitals with similar chemical bonding characters and their momentum distributions. The orbital electron density of the HOMOs (see fig. 3.7(a)) is

concentrated on the amine group (NH2) in the HO–C(=O)C–N moiety (i.e. the main moiety), with only minor density appearing in the aliphatic extension of the amino acids, which is consistent with the fact that the alkyl chains contribute little to those orbitals. In their orbital momentum distributions, it is clearly demonstrated that the HOMOs are dominated by p-like

electrons from the -NH2 moiety. In fact, the two hydrogen 1s electrons bonding with the

nitrogen 2p electrons in -NH2 enhances the p-like distribution. The residual density in the main moiety (HO–C(=O)C–N) as well as the aliphatic chain contribute to the small increase in the lower momentum region of the sp-like orbital momentum distribution of the HOMOs. Although the NHOMOs in fig. 3.7(b) also exhibit similar trends as the HOMOs among the amino acids, their orbital momentum distributions are not the same as the HOMOs. In the NHOMOs, the electron lone pair on the oxygen atom of the keto C=O moiety as well as the

nitrogen atom of the amine NH2 also contribute to this orbital. As a result, the orbital momentum distribution of the NHOMOs exhibit strong p-like character with two p-like peaks.

Other important orbitals in the valence space, selected by their similarities, are the innermost and the third innermost valence orbitals of the amino acids, as shown in fig. 3.7(c) and (d), respectively. The innermost valence orbitals are dominated by the 2s electrons from the two oxygen atoms in the carboxylic acid group, HO-C(=O). The side chains do not affect this orbital significantly, so that the spherically averaged orbital momentum distributions of these inner most valence orbitals are almost similar, fig. 3.7(c). The orbital momentum distributions of the third innermost orbitals, assigned to N 2s, are again similar to each other

among the other amino acids. However, although this orbital is dominated by the NH2 moiety, the alkyl side chains show some effect of mixing. Therefore, the orbital momentum distributions are s-like (“half bell shaped”) but differ in the low momentum region due to the

Aliphatic amino acids… Chapter 3

side chains as shown in fig. 3.7(d).

Fig.3.7: Selected valence orbital electron density and momentum distributions of the aliphatic amino acids (a) the HOMOs, (b) the NHOMOs, (c) the innermost valence orbitals and (d) the third innermost valence orbitals.

Fig. 3.8(a) presents the orbital momentum distributions of third HOMOs (THOMOs) of the aliphatic amino acids. Unlike the first two frontier orbitals, HOMO and NHOMO, the momentum spectra of THOMOs show differences. Fig. 3.8(b) reports the momentum spectra of the THOMOs of alanine and glycine, while fig. 3.8(c) depicts the spectra of the THOMOs in valine, leucine and isoleucine. As it can be seen in the orbital densities of glycine and alanine, their THOMOs are mostly dominated by the carboxyl oxygen atoms and hence show one strong ‘p-like’ peak. On the other hand, the THOMOs of amino acids with larger side chains such as valine, leucine and isoleucine are dominated by the aliphatic side chains and as a consequence, their momentum spectra are different. It is interesting to note that unlike in the other aliphatic amino acids, the momentum distribution of the THOMO in leucine present

Aliphatic amino acids… Chapter 3

an intense ‘s-like’ character in the lower momentum region, which can be attributed to the strong s-orbital features in its methyl groups (see in fig. 3.8 (c)).

Fig.3.8: Valence orbital electron density and momentum distributions of the (a) THOMOs of the aliphatic amino acids. (b) The THOMOs of glycine and alanine and (c) THOMOs of valine, leucine and isoleucine are provided separately.

Fig. 3.9(a)-(d) show the changes in the orbital electron density and momentum distributions of the molecules as the aliphatic side chain grows. It is obvious that these orbitals of the amino acids are correlated. Fig.3.9(a) and (b) report the orbitals where there is a correlation in the glycine-alanine-valine series, whereas fig. 3.9(c) and (d) display the valine-leucine- isoleucine series. The related orbital momentum distributions also show a certain association among the three orbitals, which exhibit similar orbital distributions in the larger momentum region but change significantly in the low momentum region.

Apart from the similarities of the few selected valence orbitals which are concentrated on the common HO–C(=O)C–N (main) moiety, and the inner valence orbitals. The majority of the valence orbitals of the molecules have contributions from their aliphatic side chains and therefore their orbitals are very different, depending on their alkyl side chains.

Aliphatic amino acids… Chapter 3

Fig.3.9: Orbital momentum densities and charge densities of (a) glycine (10a’ MO), alanine (11a & 12a Mos); (b) alanine 12a, valine (15a & 16a Mos), and (c) and (d) valine (15a MO), leucine (16a & 17a MOs) and isoleucine (16a & 17a Mos).

3.7. Vibrational and VOA spectra

The following part of the chapter discusses the effects of alkyl side chains on the vibrational and chiro-optical properties of the aliphatic amino acids. With the initial optimized structures of NT and ZW amino acids, vibrational spectra (IR and Raman) together with their VOA spectra (i.e., VCD and ROA) of the aliphatic amino acids in both the gas and aqueous phases are obtained at the B3LYP/6-31++G** level of theory.

3.7.1. Theoretical and experimental Raman/ROA spectra of ZW alanine

Fig. 3.10 compares the calculated and experimental[229] Raman (upper panel) and ROA (lower panel) spectra of ZW alanine in aqueous solution. No scaling is applied in those calculated spectra. Excellent agreements between the computed and the measured spectra are observed. The simulated Raman spectrum of ZW alanine in the present study reproduces most of the major functional spectral peaks of the measured spectrum, except for the relative

Aliphatic amino acids… Chapter 3

intensities of some of the peaks. The calculated ROA spectrum in the lower panel of fig. 3.10 also reproduces very well the experimental ROA spectrum of alanine ZW in solution, after allowing for a global shift of 38 cm-1. Indeed, most of the positive and negative bands in the heoretical and experimental spectra are in excellent correlation. As a result, the present model is applied to calculate the vibrational and VOA spectra of the other aliphatic amino acids, in both gas phase (NT) and solution (ZW).

Fig.3.10: Comparison of the calculated and experimental[328] Raman and ROA spectra of alanine ZW in aqueous solutions.

In Fig. 3.11, the IR and Raman spectra of the amino acids in their NT forms (gas phase) are compared. The IR spectra are clearly dominant in the lower wavenumber region (IR active), below 2000 cm-1, whereas the Raman spectra are more active in the higher wavenumber region (Raman active) above 3000 cm-1. Thus the combined IR and Raman spectra are able to deliver a comprehensive picture of their behaviour in this wavenumber region. Furthermore, it is noted that the spectrum in the IR region of 400-4000 cm-1 of the aliphatic amino acids can be divided as an alkyl-region for  < 1600 cm-1 and a functional region for  > 1600 cm-1. The former is mostly characterized by vibrations of the alkyl side chains in the amino acids, while the latter contains vibrations from different functional groups. As a result, the IR spectra are useful to differentiate the amino acids from other biomolecules[275, 278, 281, 282]. It is useful to combine IR/VCD and Raman/ROA spectra to assist the interpretation of the vibrational & VOA (400 cm-1 - 4000 cm-1) spectra of the

Aliphatic amino acids… Chapter 3

aliphatic amino acids and therefore understand their NT and ZW structures in detail.

Fig.3.11: Comparative IR and Raman spectra of the aliphatic amino acids showing the fingerprint and characteristic regions of the spectra.

3.7.2. Vibrational (IR and Raman) spectra of glycine

Fig. 3.12 presents the IR and Raman spectra of glycine in its NT form (gas phase) (a) and its ZW form (aqueous solution) (b). As the smallest amino acid, glycine does not have an -carbon. The absence of chirality does not, therefore, produce any vibrational optical activity (i.e., VOA) signals, so that glycine does not have any VCD and ROA bands. Note that the IR intensive region (i.e., alkyl-region) and Raman intensive region (i.e., functional region) in the spectra of NT and ZW glycine, are separated by the dashed line in fig. 3.12. The functional group region  > 1600 cm-1, in the spectra of NT glycine, possesses five major Raman modes

and two major IR modes. Those modes are assigned as the OH stretch (str) (‘1’); NH2 asym

(asymmetric) str (‘2i’) and sym (symmetric) str (‘2ii’); C(2)H asym str (‘3i’) and sym str (‘3ii’), together with a couple of intense IR peaks due to the C=O str, OH bend (‘4g’); and

NH2 scissoring (scis) (‘5’), as marked on the spectra. It is worthy of note that the intensities of both the IR and Raman spectra in the ZW form are significantly enhanced, including a few IR spectral peaks in the high wavenumber region of 3000 cm-1. It is also noted that the spectra of ZW glycine are correlated to the spectra of NT glycine in the gas phase. However they are not identical, that is, not all the spectral peaks in the ZW form are the same as in the NT form of glycine due to their structural differences. For example, the ZW spectral peak ‘4z’ at ~ 1680

Aliphatic amino acids… Chapter 3

-1 cm is assigned to the O(1)···C(1)···O(2) str and NH3 bending motions, whereas this peak corresponds to the one at ~1834 cm-1 in the NT form (‘4g’). This ‘4g’ peak is in fact dominated by the C=O stretch and OH bending vibrations. In addition, the ZW spectra of glycine, compared to its NT form, does not show any peak at ~3808 cm-1 (‘1’) due to the lack of an OH group. However, it gains an extra intensive peak (in both IR and Raman), marked as -1 ‘2iii’ (~3500 cm ), which is attributed to the NH(3) str.

Fig.3.12: IR and Raman spectra of glycine in the gas and solvent phases (refer to table 3.8 for wavenumbers and vibration modes).

Table 3.8 assigns the major vibrational wavenumbers of glycine in the NT and ZW forms, as are also marked in fig. 3.12, and compares them with the observed wavenumbers[266, 289, 329]. Note that both the scaled (with a scaling factor[330] of 0.96) and unscaled spectral peak positions from our calculations are provided in that table. In the functional spectral region,  > 1600 cm-1, the scaled calculated wavenumbers are in good

agreement with those obtained from the experiments. For instance, the –NH 2 asym str vibration of NT glycine is 3470 cm-1, which agrees well with the measured peak positions[266, 289] of 3410 cm-1 and 3414 cm-1. However it is noted that the scaling factor works less impressively in the IR intensive region at lower wavenumbers, that is, the alkyl region. On the other hand, the unscaled wavenumbers in the alkyl region are themselves sufficiently close to -1 the measured values. For instance, the NH2/CH2 wagging mode in ZW glycine is 70 cm lower than the measured value[329] when scaled, however, the unscaled value is only 17 cm-1 away. Therefore, no scaling is applied to the calculated spectra in this study.

Aliphatic amino acids… Chapter 3

Table 3.8: Present scaled and unscaled vibrational wavenumbers of NT (in gas phase) and ZW (in solution) glycine, along with the available experimental data (all wavenumbers are given in cm-1). Glycine NT Glycine ZW No.* This Work This Work No* This Work This Work EXP[329 (Unscaled) (Scaled)^ EXP# Assignment (Unscaled) (Scaled)^ ] Assignment

1 3808 3656 3560 OH str 2(i) 3561 3419 NH2 asym str

2(i) 3615 3470 3410(3414) NH2 asym str 2(ii) 3496 3356 NH2 sym str

2(ii) 3535 3394 NH2 sym str 3(i) 3190 3062 CH2 asym str

3(i) 3112 2988 (3084) CH2 asym str 3(ii) 3129 3004 2960 CH2 sym str

3(ii) 3070 2947 2958(2920) CH2 sym str 2(iii) 3055 2933 2930 NH(3) sym str

4g 1835 1762 1779(1703) C=O str, OH bend 4z 1680 1612 1630 C(1)O(1) str, NH(3) bend

5 1678 1611 1630(1610) NH2 scis 1665 1598 1603 NH3 deformation

a 1457 1399 1429(1410) CH2 scis 5 1636 1570 1566 NH2 scis

1402 1346 1373 CH2 wag, COH bend A 1484 1425 1430 CH2 scis

1382 1327 (1334) CH2 / NH2 twist B 1432 1375 NH3 puckering CO str, CC str, CH C 1372 1317 1403 (2) 2 1305 1253 CH2 wag, OH bend wag

1180 1133 CH2 twist. NH2 twist 1324 1271 1341 CH2 / NH2 wag

1164 1117 1136 CN str, CH2/NH2 wag, CO str 1301 1249 1314 CH2 / NH2 twist

b 1126 1081 1101 CO str, OH bend, CN str D 1101 1057 1171 CH2 wag, NH3 rock

c 919 882 907 CH2 wag, NH2 wag 1097 1053 CH2 twist, NH3 rock

914 877 883(893) CH2 rock, NH2 twist 983 944 CN str, HCNH tor

d 822 789 801 CC str, NH2 wag, CH2 wag 929 892 CH2 / NH2 rock

630 605 619 COOH tor, HNCC tor E 864 830 961 CC str, CH2 wag e 629 604 OH OP bend, CH2 rock 675 648 674 CCO bend, CCN tor

f 474 455 500 OH OP bend, CH2 rock 570 548 CH2 rock 463 444 463 HCCN tor, CCO bend, COOH tor 499 479 CCO bend, HCCN tor *No. represents the peaks marked in fig. 3.12; #Ref.[331] (values given in parentheses are from Ref.[266]); ^Scaled by a factor of 0.96[330].

Aliphatic amino acids… Chapter 3

The alkyl region of NT glycine presents six major spectral peaks successively labelled as a-f , at ~1456 cm-1 (‘a’), 1126 cm-1 (‘b’), 919 cm-1 (‘c’), 822 cm-1 (‘d’), 629 cm-1 (‘e’) and 474 cm-1 (‘f’). These peaks are very different from those in ZW glycine that appear at 1484 cm-1 (‘A’), 1432 cm-1 (‘B’), 1372 cm-1 (‘C’), 1101 cm-1 (‘D’) and 864 cm-1 (‘E’) (refer to table 3.8 for their respective vibrational assignments). The bands in the alkyl region can be relatively more complex, due to the mixed vibrational motions arising from the strong inter- and intra-molecular interactions in this species, as a consequence we find differences in the NT and ZW spectra of glycine.

3.7.3. VOA spectra of the NT and ZW chiral amino acids

Apart from glycine, all the other aliphatic amino acids, with the bulky methyl dominant side chains, are chiral in nature. As a result, these amino acids are optically active with VOA signals (i.e, VCD and ROA bands). In this section, the IR based VCD spectra and Raman dependent ROA spectra are quantum mechanically calculated for both the NT and ZW forms of the aliphatic amino acids, other than glycine. This was undertaken in order to explore the side chain caused structural impacts on the chiro-optical properties of those compounds. The VCD and ROA spectra, combined with their parent IR and Raman spectra, respectively, are given in figs 3.13-3.16, whereas the assignments of the major spectral peaks are presented in tables 3.9 and 3.10.

Alanine Fig. 3.13 reports the comparative IR/VCD and Raman/ROA spectra of NT alanine (a) and ZW alanine (b). When a hydrogen atom in glycine is replaced by a methyl group, the

resultant alanine molecule with a -carbon (C) becomes chiral. As a result, optical activity signals are produced in the VCD and ROA spectra of alanine. The major spectral changes with respect to NT glycine are that the methyl group in NT alanine lead to some additional peaks in the alkyl-region of  < 1600 cm-1 and the functional region of  > 1600 cm-1. Fig. 3.13 suggests that the amino and the carboxyl vibrational motions in the functional region display very weak VCD and ROA intensities. However, the CH str vibration modes in the vicinity of 3000 cm-1 produce a negative ROA peak at 3129 cm-1 and a positive ROA peak at -1 3079 cm . Those features are due to C(3)H2 asym str and C(2)H str motions, respectively, as shown in fig. 3.13(a) and table 3.9.

Aliphatic amino acids… Chapter 3

Fig.3.13: IR/VCD and Raman/ROA spectra of alanine in the gas and solvent phases. Please refer to tables 3.9 and 3.10 for the wavenumbers and vibrational modes marked in the spectra.

The VCD and ROA spectra of ZW alanine show enhanced signals. The most noticeable and intense IR spectral peak at 3016 cm-1 appears in the spectra of alanine in fig.

3.13(b), which is assigned to the NH(3) vibrations (‘2iii’). This peak, which is unique to the ZW forms of the amino acids and is not seen in the NT spectra, produces a negative peak in

the VCD and the ROA spectra in ZW alanine (fig. 3.13(b)). The NH2 sym str (‘2ii’) at 3489 -1 -1 cm corresponds to a weak positive VCD signal, whereas the NH2 asym str (‘2i’) at 3557 cm corresponds to a weak negative ROA signal. The CH vibrations clearly produce intense Raman peaks in the ZW form. For example, the peaks at 3155 cm-1 (‘3i’), 3130 cm-1 (‘3ii’) -1 and 3115 cm (‘3iii’) are dominated by C(2)H str vibrations, although the peaks ‘3ii’ and ‘3iii’

also include motions of C(3)H3 asym str and sym str, respectively. The C-H vibrations in the ‘3ii’ peak give rise to a strong positive ROA peak and a weak negative VCD peak at 3130 cm- 1 of ZW alanine, whereas the ‘3i’ and ‘3iii’ Raman peaks relate to less intense negative ROA -1 - peaks. The vibration at 1676 cm (i.e., the peak ‘4z’) is assigned to the COO bend and NH3 deformation modes, which results in a less intense negative VCD band.

Aliphatic amino acids… Chapter 3

Table 3.9: Present vibrational wavenumbers (in cm-1) and their wavenumbers in the functional group region ( > 1600 cm-1) of the NT and ZW forms of the aliphatic amino acids.

(2) NH Str (4g) C=O str,OH Bend/ Molecules (1) OH str (i) NH asym (II) NH (iii) NH (3) CH str (5) NH2 scisc 2 2 (3) (4z) COO str, NH3 def. str sym str str

Glycine (NT) 3808 3615 3535 _ (i) 3112 -CH2 asym str; (ii) 3070 -CH2 sym, str (4g) 1835 1678

Glycine (ZW) _ 3561 3496 3055 (i) 3190 -CH2 asym str; (ii) 3129 -CH2 sym, str (4z) 1680 1636 (i)3153 -C H asym str; (ii) 3129 -C H sym Str; (iii) 3079 - Alanine (NT) 3800 3593 3513 _ (3) 3 (3) 3 (4g) 1827 1676 C(2)H str; 3051 -C(3)H3 sym, str;

(i)3155 -C H asym str; (ii) 3130 -C H asym str/C H str; Alanine (ZW) _ 3557 3489 3016 (3) 3 (3) 3 (2) (4z) 1676 1636 (iii) 3115 -C(3)H3 sym str/C(2)H str; 3052 -C(3)H3 sym, str;

(i)3133 -C(4)H3 asym str; (ii) 3115 -C(5)H3 asym str; (iii) 3111 - Valine (NT) 3796 3594 3516 _ C(5)H3 asym str; (iv) 3065 -C(2)H str; (v) 3051 -C(3)H str; (4g) 1820 1677 (vi)3042 -C(3)H/C(4)H3 sym, str; (vii)3039 -C(3)H/C(4)H3 sym, str;

(i)3121 -C(5)H3 asym str; (ii) 3114 -C(4)H3 asym str/C(2)H str; (iii) 3113 -C H asym str/C H str; (iv) 3102 -C H/C H /C H Valine (ZW) _ 3558 3488 2996 (4) 3 (2) (3) (4) 3 (5) 3 (4z) 1678 1629 str; (v) 3096 -C(3)H/C(4)H3/C(5)H3 str; (vi)3071 -C(3)H str; (vii)3035 -C(4)H3/C(5)H3 sym, str;

(i)3118 -C(6)H3 asym str; (ii) 3108 -C(5)H3 asym str/C(2)H str; (iii) 3102 -C H /C H /C H asym str; (iv) 3084 -C H str; (v) Leucine (NT) 3797 3583 3503 _ (5) 3 (6) 3 (4) (2) (4g) 1823 1674 3069 -C(2)H/C(3)H2 asym str; (vi)3054 -C(3)H2/C(4)H str; (vii) 3031 -C(6)H3/C(5)H3 str; (viii)3010 -C(3)H2 S. str;

(i)3124 -C(2)H str; (ii) 3105 -C(5)H3 asym, str /C(6)H3 str; (iii) 3099 -C H /C H /C H asym str; (iv) 3096 -C H/C H asym Leucine (ZW) _ 3558 3485 2985 (5) 3 (6) 3 (4) (6) (3) 2 (4z) 1677 1634 str; (v) 3034 -C(6)H3/C(5)H3 sym, str; (vi)3027 -C(5)H3 sym, str; (vii) 3003 -C(4)H str; (i)3135 -C(6)H3 asym str; (ii) 3125 -C(5)H3 asym str; (iii) 3112 - Isoleucine C H ; (iv)3110-C H /C H asym str (v) 3059 -C H; 3795 3592 3516 _ (5) 3 (6) 3 (5) 3 (2) (4g) 1820 1678 (NT) (vi)3046 -C(5)H3/C(6)H3 sym, str; (vii) 3036 -C(3)H str.;(viii)3026 -C(4)H3 sym, str;

(i)3125 -C(6)H3 asym str; (ii) 3113 -C(5)H3 asym str; (iii) 3106 - C H str; (iv)3069-C H str/C H asym str; (v) 3060 -C H Isoleucine (ZW) _ 3558 3490 2991 (2) (3) (4) 2 (3) (4z) 1678 1632 str/C(4)H2 asym str; (vi)3039 -C(5)H3/C(6)H3 sym, str; (vii) 3036 - C(5)H3/C(6)H3 sym, str; (viii)3012 -C(4)H3 sym, str;

Aliphatic amino acids… Chapter 3

The alkyl regions of the VCD and ROA spectra ( < 1600 cm-1) exhibit more features than their parent IR and Raman spectra. Indeed some relatively intense VCD and ROA bands of ZW alanine are observed in the region of 800 cm-1 to 1600 cm-1. Figs. 3.13 (a) and (b) and table 3.10 summarize the spectral peak positions and their assignments for this molecule.

Table 3.10: Present results showing the comparative vibrational wavenumbers (in cm-1) and their assignments, of the NT and ZW forms of the aliphatic amino acids in the alkyl-region ( < 1600 cm-1).

Molecules Vibrational modes

Alanine (NT) (a) 1365 - C(2)H bend, Co str, OH bend; (b)1176 -C(2)H bend, CN str, HNCH tor; (c) 1131 -C(2)H bend, CO str, OH bend; (d) 1086 -C(2)H bend, C(3)H3 rock, HCCH tor.;

(A) 1376 -C(2)H bend; (B) 1358 -C(2)H bend, C(2)-C(3) str; (C) 1115 -C(2)H bend, C(3)H3 rock, NH3 rocking; (D) Alanine (ZW) 1001 -C(2)H bend, C(3)H3 rock, NH3 rock; (E) 992 - C(2)H bend, C(3)H3 rock, NH3 rock; (F) 878 - HCCH tors., C(1)C(2) str, C(2)N str.

(a) 1342 – CC str, CO str, OH bend, NH2 twist, C(2)H bend; (b) 1276 -C(2)H str, OH bend, C(3)H bend; (c) 1134 - Valine (NT) CO str, OH bend; (d) 984 -C(2)H bend, C(3)Hbend, C(5)H3, C(2)H3 Rocking; (e)969 -C(4)H3/C(5)H3 rocking, C(2)- C(3) str, C(2)H bend; (f) 647 -OH bend, CO str, HNCC tors, C(2)H bend. Valine (ZW) (A) 1391 -C(2)H bend, COO str, C(2)H bend; (B) 1364 -COO str, CC str, NH3 puckering, C(2)H bend; (C) 1144 - NH3 rocking, CH3 rocking, HCCH tors; (D) 1107 -NH3 rocking, C(2)H bend.

Leucine (NT) (a) 1356 -OH bend, CO str, HNCH tors, CC str, C(4)H bend, C(2) bend; (b) 1179 -C(2)H bend, OH bend, HNCH tors; (c) 967 -NH2 wag, HCCH tors, C(2)H bend, CC str; (d) 740 -COOH tors, C(2)H bend, HCCC tors;

(A) 1367 - C(4)H bend, C(3)H/C(2)H bend; (B) 1353 -C(2)H bend, CC str, C(4)H bend, NH3 puckering; (C) 1072 - Leucine (ZW) C(2)H bend, NH3 rocking, C(3)H2 Rocking; (D) 1034 -CN str, NH3 rocking, HCCH tors; (E) 927 - C(6)H3/C(5)H3 rocking, C(4)H bend.

Isoleucine (NT) (a) 1426 - NH2 twist, C(2)H bend; (b) 1343 -NH2 twist, C(2)H bend, C(3)H bend, C(4)H bend, OH bend; (c) 1132 - OH bend, CO str, C(2)H bend; (d) 1087 -CH3 rocking, C(3)H bend;

(A) 1424 -CH3 rocking; (B) 1341-C(4)H2 wagging, C(2)H bend, CC str; (C) 1366 -C(4)H2 twist, C(2)H bend; (D) Isoleucine (ZW) 1348 - COO str, NH3 puckering, C(2)H bend, C(3)H bend; (E)1136 -CH3 rock, HCCH tors, HNCC tors; (F) 1106 - NH3 tors, C(2)H bend; (G) 1003 - HCCH tors, CH3 rock, CC str.

The four notable peaks in the VCD spectrum of NT alanine in this region are the three negative VCD peaks at 1365 cm-1 (‘a’), 1176 cm-1 (‘b’), 1086 cm-1 (‘d’) and one positive VCD band at 1131 cm-1 (‘c’). Although those peaks are produced by different vibrations, they are

dominated by the C(2)H bending motions as assigned in table 3.10. The intense bands in the VCD & ROA spectra of ZW alanine include two positive peaks at 1376 cm-1 (‘A’) and 992 cm-1 (‘E’), and four negative peaks at 1358 cm-1 (‘B’), 1115 cm-1 (‘C’), 1001 cm-1 (‘D’) and 878 cm-1 (‘F’) in the VCD spectrum. Their corresponding ROA bands at these positions, however, switch signs. The strongest positive VCD peak at 1376 cm-1 (A) is an intense negative peak in the ROA spectra, for example. However, a positive band at 992 cm-1 (i.e., peak ‘E’) in the VCD spectrum does not appear in the corresponding ROA spectrum. As the

Aliphatic amino acids… Chapter 3

signals here are weak, it is difficult for us to provide further analysis on this aspect. A previous experiment[229] identified a positive ROA peak at 922 cm-1 and a negative ROA peak at 1003 cm-1 of ZW alanine, those are in good agreement with our calculated peak positions at 878 cm-1 (‘F’) and 1003 cm-1 (‘D’) respectively.

Amino acids with branched chains

Other amino acids such as valine, leucine and isoleucine are associated with more complex branched alkyl chains. For example, the terminal side chains of valine and leucine

show a similar structure with a [-CH(CH3)2] group. In addition, valine and leucine both

possess a single chiral carbon, i.e. C(2), which connects with a hydrogen, a carboxyl, an amino and an alkyl group. On the other hand, leucine and isoleucine are isomers with similar

structural variants. However, isoleucine now has two chiral carbons, namely C(2) and C(3). The chirality of these compounds is the source for the VCD and ROA signals.

The IR/VCD and Raman/ROA spectra of NT and ZW valine, leucine and isoleucine are given in figs 3.14 - 3.16 accordingly. Similar to the spectra of alanine, the amino (refer to peaks labelled ‘2i’, ‘2ii’, ‘5’) and carboxyl (‘1’ and ‘4g’) vibrations in the functional region (υ > 1600 cm-1) of NT valine, leucine and isoleucine do not produce intense VCD or ROA signals. The exception to this is the VCD bands of NT valine (fig. 3.14(a)), which exhibit intense peaks denoted as ‘2i’, ‘4g’ and ‘5’. In contrast to their NT forms, the VCD and ROA signals of the ZW amino acids in this region are more intense. For instance, the ‘2i’ and ‘2ii’ peaks from the sym and asym str vibrations of the amino group present a negative-positive (- ve/+ve) couplet in ZW valine, leucine and isoleucine. Due to the carboxyl COO– stretch of the ZW structures, the ‘4z’ peak at 1678 cm-1 for valine (1677 cm-1 for leucine and 1678 cm-1 for isoleucine) is a strong negative VCD band. It is indicative that the intra-molecular proton transfer in a ZW amino acid makes the carboxyl and amino network more optically active.

Aliphatic amino acids… Chapter 3

Fig.3.14: IR/VCD and Raman/ROA spectra of valine in the gas and solvent phases. Please refer to tables 3.9 and 3.10 for the wavenumbers and vibrational modes marked in the spectra.

Aliphatic amino acids… Chapter 3

Fig.3.15: IR/VCD and Raman/ROA spectra of leucine in the gas and solvent phases. Please refer to tables 3.9 and 3.10 for the wavenumbers and vibrational modes marked in the spectra.

Aliphatic amino acids… Chapter 3

Fig.3.16: IR/VCD and Raman/ROA spectra of isoleucine in the gas and solvent phases. Please refer to tables 3.9 and 3.10 for the wavenumbers and vibrational modes marked in the spectra.

Aliphatic amino acids… Chapter 3

The CH stretch vibration bands, which concentrate in the wavenumber region of 3000 - 3250 cm-1, dominate the VOA spectra of these amino acids in both their NT and ZW configurations. It is found that the most intense ROA and VCD bands of NT valine and

leucine are dominated by their chiral carbon, C(2), which forms a negative-positive (-ve/+ve) couplet appearing at 3065 cm-1 (-ve)/3051 cm-1 (+ve) in valine (‘3iv’/‘3v’, fig. 3.14(a)) and 3069 cm-1 (-ve)/3054 cm-1 (+ve) in leucine (‘3v’/‘3vi’, fig. 3.15(a)). Interestingly, with two chiral carbon centres, the negative ROA and VCD signals in isoleucine, which are caused by the vibrations of the chiral carbons, are apparently weaker (marked with a circle in fig. 3.16(a)) than both valine and leucine. Asym str modes of the methyl groups also contribute to the signals in this wavenumber region. For example, there is an intense positive-negative couplet in the ROA spectra of valine and leucine --- 3133 cm-1 (-ve)/3115 cm-1 (+ve) in valine (‘3i’/‘3ii’) and 3102 cm-1 (-ve)/ 3108 cm-1 (+ve) in leucine (‘3iii’/‘3ii’). Asym methyl vibrations in isoleucine, however, produce a positive-negative-positive triplet at 3135 cm-1 (‘3i’), 3125 cm-1 (‘3ii’) and 3112 cm-1 (‘3iii’). In addition, it is noted that the signals of the VCD and ROA spectra display opposite signs. For example, an intense negative ROA peak (‘3ii’) at 3125 cm-1 in NT isoleucine becomes a strongly positive peak in its VCD spectrum (see fig. 3.16(a)).

The methyl vibrations dominate the VOA spectra of the ZW amino acids in solution, while the chiral carbons make important contributions to the VOA spectra of their NT counterparts. For this reason the VCD/ROA spectra of ZW valine and isoleucine share more similarities than the spectra of ZW leucine, as shown in figs. 3.14(b) - 3.16(b). The most intense positive ROA band in ZW valine at 3113 cm-1 (‘3iii’ in fig. 3.14(b)) is assigned to the

C(2)H str vibration. This band (‘3iii’) is accompanied by two methyl vibration dominant negative peaks, ‘3i’ and ‘3iv’ at 3121 cm-1 and 3102 cm-1 respectively. Similar methyl dominant spectral patterns can be found in ZW isoleucine at 3113 cm-1 (‘3ii’) and 3069 cm-1 (‘3iv’). However, only one strong negative peak (‘3iv’) at 3096 cm-1 is found in the ROA of

leucine, which is due to the C(6)H3 vibration. It is also noted that the intense VOA bands (i.e., VCD and ROA) are produced by the CH asym str, whereas the CH sym str do not produce strong VCD or ROA signals for both the NT and ZW amino acids. As a result, asymmetric stretching vibrations of the methyl groups dominate the functional region in the VOA spectra of the amino acids.

Aliphatic amino acids… Chapter 3

The spectral signature or alkyl region (υ < 1600 cm-1) of these amino acids is side chain dependent. The assignment of the intense spectral peaks (marked in figs. 3.14 – 3.16), for both VCD and ROA spectra in this region, is given in table 3.10. In the VCD and ROA spectra of NT valine, the most intense negative band in the VCD is at 984 cm-1 (-ve, ‘d’) and

that is assigned to C(2)H bend, C(3)H bend and methyl rocking modes. A couple of intense positive bands in this spectra at 969 cm-1 (‘e’) and 647 cm-1 (‘f’), are assigned to the methyl rocking, OH bending, CO stretch and HNCC torsion modes. On the other hand, the most intense peak in the ROA spectra of NT valine is a couplet at 1342 cm-1 (+ve, ‘a’)/1276 cm-1 (-

ve, ‘b’), which are due to contributions from C(2)H bending, NH2 twisting, C(3)H bending and OH bending motions. Two positive-negative spectral pairs appear in the VCD and ROA spectra of ZW valine. One such pair appears at 1391 cm-1 (+ve, ‘A’)/1364 cm-1 (-ve, ‘B’)

from the C(2)H, COO and CC stretch motions as shown in fig. 3.14(b) and table 3.10, which is consistent with the experimental ROA bands at 1414 cm-1 and 1357 cm-1 in valine[285]. Another such pair is the peaks at 1144 cm-1 (‘C’)/1107 cm-1 (‘D’), which are dominated by

the NH3 and CH3 rocking modes, where the ‘C’ and ‘D’ bands swap signs between the VCD and ROA spectra.

The intense VCD peaks in the alkyl region of NT leucine and isoleucine are dominated by negative spectral peaks, as shown in fig. 3.15(a) and fig. 3.16(a), whereas the VCD spectra of their ZW counterparts contain more positive peaks (see fig. 3.15(b) and fig. 3.16(b)). The most intense VCD peak of NT leucine in this region, shown in fig. 3.15(a), is a -1 negative band (‘a’) at 1356 cm . It is associated with the combinations of OH bending, C(2)H stretch, CO stretch, HNCH torsion and the CC stretch motions. The other intense peaks ‘b’ and ‘c’ are also negative bands. Similarly, the VCD spectrum of NT isoleucine produces 3 negative peaks, 1343 cm-1 (‘b’), 1132 cm-1 (‘c’) and 1087 cm-1 (‘d), along with a weak positive band ‘a’ at 1426 cm-1. The VCD of ZW leucine displays a number of spectral peaks, which are marked as ‘A’, ‘B’, ‘C’, ‘D’ and ‘E’ in fig. 3.15(b), in which the most intense band is a positive-negative pair at 1367 cm-1 (+ve, ‘A’) and 1353 cm-1 (-ve, ‘B’). The next most intense band is also a negative-positive couplet marked as ‘C’ (-ve) and ‘D’ (+ve) that are due

to NH3 rocking. Detailed assignments are also provided in table 3.10. These VCD spectral peaks of ZW leucine disappear in the corresponding ROA spectrum, except for one strong negative ROA peak marked as ‘A’ at 1367 cm-1 that is due to the bending motions of the

C(4)H, C(3)H and C(2)H groups. However, the VCD spectrum of ZW isoleucine produces the maximum number of intense bands. This includes three positive peaks (‘A’, ‘B’ and ‘G’) and

Aliphatic amino acids… Chapter 3

four negative bands (‘C’-‘F’), whereas the ROA spectrum of ZW isoleucine only exhibits a positive-negative pair (‘C’/’D’) peaks.

3.8. Summary

This chapter discusses the effects of the alkyl side chains on the electronic structures, chemical ionization, vibrational and chiro-optical properties of the aliphatic amino acids. The core and valence ionization spectra of the model molecules in the gas phase are quantum mechanically calculated. The effects on the N 1s and O 1s core ionization spectra exhibit small perturbations to the spectra of glycine, whereas in the C 1s spectra there are additional peaks at lower energy, IP < ca. 291 eV owing to the additional carbon atoms in the alkyl chains. In the valence ionization region, spectral changes are significant in the middle energy region of 12 eV < IP < 16 eV, while the outermost and innermost valence peaks are less affected with respect to the alkyl chains. This is because the electron density distributions of these orbitals are concentrated on the common HO–C(=O)C–N (main) moiety of the amino acids and are largely 2s in character. As a result, the growth of the alkyl side chain does not significantly affect the frontier orbitals of the aliphatic amino acids. However other properties such as the HOMO-LUMO energy gap, display changes in relation to the side chains. The HOMO-LUMO energy decreases with increasing side chain length, and we found that the energy order was: glycine > alanine > valine > leucine > isoleucine. The valence energy region of approximately 12-16 eV remains the signature region for the aliphatic amino acids, as the orbitals of the side chains appear in this region.

Vibrational spectra of the aliphatic amino acids are computed in their neutral and zwitterionic forms, using density functional theory methods. It is found that the calculated Raman and ROA spectra of alanine, in aqueous solution, agree well with the available experiment. Our study confirms that the functional region of 1600 - 4000 cm-1 is more Raman active, whereas the signature region of 400 - 1600 cm-1 is IR intensive, in both the neutral (gas phase) and zwitterionic (aqueous solution) forms. It is revealed that the CH dependent vibrations in the region of 3000-3250 cm-1 dominate their VCD and ROA spectra. The chiral carbons of the neutral amino acids produce intense VCD and ROA bands, whereas the alkyl (or methyl) vibrations are more intense in the VCD and ROA signals of the zwitterionic amino acids. The relative strengths of the asymmetric and symmetric CH vibrations of the

Aliphatic amino acids… Chapter 3

aliphatic amino acids can be indicated by the intensities of their VOA bands (i.e., VCD and ROA). The former gives intense VCD and ROA bands, while the latter exhibits weak bands in their VCD and ROA spectra. Moreover, the CH vibrations in the functional region produce opposite signs in the VCD and ROA spectra of the NT and ZW aliphatic amino acids, while the amino and carboxyl groups show almost identical VCD and ROA signs.

The alkyl region (υ < 1600 cm-1) of the VCD and ROA spectra of the amino acids are, however, dominated by the -carbon and alkyl side chain vibrations. As a result, a combination of the vibrational (IR and Raman) and VOA (VCD and ROA) spectra can be a very useful tool in order to study comprehensively the vibrational and chiro-optical properties of amino acids and possibly other (chiral) biomolecules.

Intra-molecular interactions of Phenylalanine… Chapter 4

CHAPTER

Intra-molecular interactions of phenylalanine

4.1. Introduction

L-Phenylalanine (L-Phe) is an essential aromatic amino acid and receives great attention both theoretically and experimentally[54, 56, 83, 144, 332-345]. L-Phe takes a neutral structure in the gas phase, however, introducing even a few water molecules in its environment can change L-Phe from its neutral from into a zwitterion form[42-44, 96]. Similar to other amino acids, L-Phe exists in a variety of low energy conformations in the gas phase[83, 340, 343]. Fig. 4.1 gives the chemical structure of L-Phe. L-Phe consists of three different functional groups, a carboxyl (-COOH), an amino (-NH2) and a phenyl ring, joining

through the CC()bridge. The most populated conformer of L-Phe in gas phase is stabilized by a H-bond which is enhanced by the aromatic ring[334].

Fig.4.1: Chemical structure of L-Phe and its nomenclatures.

Intra-molecular interactions of Phenylalanine… Chapter 4

Orientations of functional groups connected at the -carbon of L-Phe play vital roles in its overall structure and properties. Our recent ab-initio molecular dynamics study finds the intra-molecular H-bond between the carboxyl and amino group in L-phe is important for stabilizing the micro-hydrated Phe-Cu2+ complexes[44]. Similarly, an earlier data mining project identifies the amide-phenyl interactions in Phe as an essential factor that stabilizes the protein residues over large conformational spaces[2]. Therefore understanding the intra- molecular mechanisms of L-Phe and its inherent properties in the gas phase can be useful in realizing the complex structures.

The electronic properties of Phe have been reported by different methods such as, photoemission spectroscopy (PES)[273, 345-349 ], near edge x-ray absorption fine structure (NEXAFS) spectroscopy[350, 351], electron energy loss spectroscopy (EELS)[344] and etc. For example, Zhang et al[336] have studied the electronic structures of L-Phe and other aromatic amino acids using the X-ray photoemission spectroscopy (XPS) and NEXAFS technique. It is reported[336] that the N 1s core level spectra of L-Phe can indicate conformer population, while the N K-edge spectra does not show any conformational effects. Similarly, the valence PES spectra of Phe and other aromatic amino acids have also been reported previously[273, 345-349]. However, the HOMO energy and its assignment remain controversial among the studies. For example, Cannington and Ham[273] report the first ionization energy (i.e., HOMO) of L-Phe as 9.4 eV, while Campbell and colleagues[346, 349] present a broad peak between 8-10 eV in the photoelectric spectrum of Phe. In their initial study, Campbell et al[349] indicated that the HOMO of Phe corresponds to the nitrogen lone pair appearing at 8.5 eV, which is also supported by the photoionization study and theoretical simulations of Vorsa et al[352] and D. M. Close[342] respectively. In a later study[346] the same group re-assigned the HOMO as the joint contributions from the nitrogen lone pair and the π orbital on the phenyl group, which is supported by Prince and colleagues in their recent XPS investigation[345]. The latter assigns the HOMO peak of Phe

at 9.5 eV, from contributions of π1, π2 and the nitrogen lone pair.

Apart from HOMO, Prince and co-workers[345] also discuss the orbital characters of several outermost and innermost orbitals of Phe. However information on the complete valence spectrum of L-Phe is still limited, as the experimental spectrum of Phe in valence space remains congested and is not fully resolved. As a result, theoretical calculations can be

Intra-molecular interactions of Phenylalanine… Chapter 4

useful to study the entire valence region of bio-molecules by relating the spectral features with their electronic structures[52-54, 64, 66, 67, 144, 327, 353]. For instance, it has been shown in the previous chapter that the position space and momentum space information can be combined, i.e., dual space analysis (DSA)[193 ] to understand the electronic structures of molecules such as aliphatic amino acids. The DSA is employed in this work to probe the molecular level information of L-Phe in the gas phase.

The electronic structures and intra-molecular interactions of L-Phe can be comprehensively studied by probing either the functional group impacts on L-Phe or the roles of different fragmentation schemes within L-Phe. Substitution studies _ i.e., replacing a Hydrogen atom by a functional group in a molecule or vice versa _ have been a very effective scheme in chemistry to probe the impacts of chemical changes in the electronic structures and properties of molecules. Comparing a series of substituted molecules is often a very powerful tool to unravel the molecular structures of the parent molecules. Hence, in order to understand the intra-molecular interactions and the role of particular functional groups in L- Phe, it is useful to study derivatives of L-Phe, by removing specific functional groups from L-Phe (fig. 4.2). As shown in fig. 4.2, when a hydrogen atom in L-alanine is replaced by a

phenyl ring, L-alanine chemically becomes L-Phe. When the amino group (-NH2) from L- Phe is removed, it forms 3-phenylpropionic acid (PPA)[354, 355], a metabolite of the antidepressant phenelzine[356]. Alternatively, if the carboxyl group (-COOH) is removed from L-Phe, then it forms another important aromatic compound, 2-phenethylamine (PEA)[356-358].

Fig.4.2: Ground state electronic structures of L-Phe and its derivatives with nomenclatures.

Intra-molecular interactions of Phenylalanine… Chapter 4

Detailed studies on the functional groups will help to understand the structural and dynamical behaviors of L-Phe. Analyzing the interactions between fragments and their impacts to the chemical environment can often be extremely insightful[56]. A previous experiment shows that the Raman spectra of amino acids can be used to analyze the spectra of a collagen protein[281]. This conceptual approach, often referred to as “building block principle” provides a useful starting point for the interpretation of the spectra of very complicated molecules[336]. Mass spectroscopy is a common technique used to study the interactions of a molecule by observing their fragmentation patterns. This technique measures the molecular weight of molecules in order to provide information about their intra-molecular interactions. Several studies concerned with the mass spectroscopy of aromatic amino acids in the gas phase have been reported[345, 352, 359-363]. A recent photoionization mass

spectroscopy study identified that the breakage of C(-C( bond is the most probable break- up option for the aromatic amino acids[345].

Such fragmentation approach is used to study the ionization trends of larger molecules, which is also referred as ‘fragments-in-molecules’ approach[54, 144]. The fragmentation scheme of aromatic amino acids is not unique as often more than one

fragmentation schemes are possible. For instance, L-Phe (NH2-CH(CH2-C6H5)-COOH), can

be physically decomposed in two schemes, scheme I is phenyl (i.e., benzene - (C6H6)) and

alanine NH2-CH(CH3)-COOH) and scheme II is toluene (CH3(C6H5)) and glycine (NH2-

CH(H)-COOH), as shown in fig. 4.3. Scheme I can happen by breaking the C( and C( in L- Phe, so that it chemically forms L-alanine and benzene (with hydrogen saturation) whereas,

scheme II can occur by rupturing the C(-C( bond so that L-Phe splits into glycine and toluene (again with hydrogen saturations). The previous chapter indicates that the electronic structures of glycine and alanine are very different, as a result, different fragmentation schemes can also provide different interactions of functional groups in L-Phe.

In this chapter, a comprehensive electron level picture of L-Phe is given from the

study of the interactions of the functional groups (i.e., COOH, NH2 and phenyl) and its fragment schemes (alanine/benzene or glycine/toluene) in the gas phase. The first part of the chapter investigates the electronic structures of L-Phe with those of PEA (Phe-COOH), PPA

(Phe-NH2), alanine (Phe-phenyl) and benzene (Phe-alanine) in order to reveal the functional group impacts in L-Phe[56, 144]. The latter part of the chapter compares the electronic

Intra-molecular interactions of Phenylalanine… Chapter 4

structures of L-Phe and its fragments, alanine/benzene scheme and glycine/toluene scheme. Various information from the model molecules, such as their geometries, charge distributions, core spectra, valence spectra and momentum space information are obtained.

Fig.4.3: Optimized ground structures of L-Phe and its fragment schemes, alanine/benzene scheme (scheme I) and glycine/toluene scheme (scheme II).

4.2. Computational details

Geometries of the ground state stable structures of the model molecules, L-Phe, PPA, PEA, L-alanine, benzene, glycine and toluene, are optimized using the DFT based B3LYP/TZVP model. Note that the structures of L-alanine and glycine in this chapter are not the global minimum structures presented in the chapter 2 but, the second local minimum structures[364], as these conformers best resemble the L-Phe structure in this study. The

structure of benzene is D6h symmetry, while those of glycine and toluene are in Cs symmetries. The core binding energy spectra of the optimized molecules are calculated using

the LB94/et-pVQZ and EKS models. Similarly, the valence space calculations are performed using the SAOP/et-pVQZ (SAOP/TZ2P[365] for alanine and benzene) and the OVGF/TZVP models. Molecular orbitals of L-Phe and its fragment schemes, glycine/toluene and alanine/benzene, in coordinate space are produced using the B3LP/TZVP model, which are directly mapped into momentum space as theoretical momentum distributions. The B3LYP

Intra-molecular interactions of Phenylalanine… Chapter 4

and OVGF calculations are performed using the G03[162] and G09[163] packages, while the SAOP and LB94 calculations are carried out using the ADF[148] code.

4.3. Geometrical properties

Fig.4.4 presents the energy based chemical scheme of alanine, Phe, PEA and PPA.

The carbon backbone of alanine is represented as C(1)-C(-C(, where C(1) is the carboxyl

carbon, C(is the central alpha carbon and C( is the methyl carbon. As shown in the figure,

when a phenyl replaces one of the hydrogen atoms at the C() site in alanine, forming Phe, a large energy drop of 231.31 eV is obtained. Upon substituting the amino group in Phe with hydrogen to form PPA, the energy increases by 55.38 eV. However, if the carboxyl group is substituted in Phe to form PEA, the energy raises approximately three times of Phe, which gives an energy increase of 188.65 eV.

Fig.4.4: Energy based chemical scheme of L-Phe and its derivatives (energies given in eV).

Intra-molecular interactions of Phenylalanine… Chapter 4

Selected geometrical parameters from the optimized structures of L-Phe, PEA, PPA and alanine along with their dipole moments are reported in table 4.1. The canonical structures obtained using the B3LYP/TZVP model are in good agreement with other gas phase studies[334 , 357] on L-Phe and on PEA using MP2/6-311G**. For Phe, there are no large differences in parameters between the B3LYP and MP2 models, as seen in this table. Changes in the alaninyl network do not impose significant geometrical changes. All bonds

listed in the table, such as N-C(), C()-C(), C()-C(1), C()-C() and C(1)=O remain almost

unchanged in the molecules. However, the C(1)-O(H) bond is observed to shrink slightly in

Phe and to expand slightly in the PPA with respect to alanine. For example, the C(1)-O(H) bond shrinks from 1.36 Å in alanine to 1.34 Å in Phe, whereas this bond length in PPA

slightly increases to 1.37 Å in PPA. In addition, the C()-C(1) bond in PPA is apparently shorter (1.52 Å) than other molecules (1.54-1.55 Å). It is noted that the phenyl ring perimeter,

R6 (sum of all bond lengths consisting of the phenyl ring[366]) of Phe, PPA and PEA remains unchanged at the value of 8.36 Å, which is slightly longer than that of an isolated benzene (8.34 Å) using the same model. It indicates that the impact of phenyl ring to the carboxylic acid-amino (alaninyl) network is more apparent than the impact of the carboxylic acid-amino network to the phenyl ring. Thus the phenyl ring could serve as a buffer to resist the changes. Bond angles and dihedral angles show marginal deviations among the molecules. Comparing

to alanine, apparent changes (up to 5°) in N-C()-C() (increases) and N-C()-C(1) (decreases) angles are revealed in the present work. Other angles in the molecules exhibit only small relaxation.

Dipole moments of the model molecules reveal significant changes in their electron charge distribution. For example, dipole moment increases significantly from 1.30 Debye in L-alanine to 5.19 Debye in L-Phe (B3LYP/TZVP model). Nevertheless, removal of the carboxylic acid or amino group certainly changes the intra-molecular interactions thereby significantly affecting their dipole moments. For instance, removal of carboxylic acid (producing PEA) remarkably reduces the polarity of the product than the removal of the amino group (producing PPA). Dipole moment of the former (PEA) is given by 1.26 Debye, significantly dropping from 5.19 Debye in Phe, whereas the latter (PPA) is given by 4.10 Debye. Dipole moments of the molecules, therefore, indicate that all associated functional

groups, phenyl, -COOH and -NH2, affect the electron charge distributions in their own ways. The selected geometries of L-Phe and its fragment molecules, alanine/benzene and

Intra-molecular interactions of Phenylalanine… Chapter 4

glycine/toluene are provided in appendix (A.II).

Table 4.1: Ground state geometries of L-Phe, PPA and PEA obtained using the B3LYP/TZVP and MP2/TZVP (Phe only) models. Results are compared with values taken from the literature cited. Phe PPA PEA Alanine Other a Other Parameters B3LYP MP2 B3LYP B3LYP b B3LYP work work C-O(H)/(Å) 1.34 1.34 1.34 1.37 1.36 C=O/(Å) 1.20 1.21 1.20 1.20 1.21

C()-N/(Å) 1.47 1.47 1.47 1.46 1.47

C()-C()/(Å) 1.55 1.54 1.55 1.54 1.55 1.54 1.45

C()-C()/(Å) 1.51 1.51 1.51 1.51 1.51 1.46 –

C()-C(1)/(Å) 1.55 1.54 1.55 1.52 – – 1.54

R6(Å) 8.36 8.36 8.38 8.36 8.36

N-C()-C(1)/° 109.4 108.9 108.9 111.6

O=C(1)-O(H)/° 123.0 123.8 123.3 119.7 123.6

 (H)O-C(1)-C()/° 113.7 113.4 113.4 115.5 113.8

O=C(1)-C/° 123.1 122.6 123.2 124.7 122.6

N-C()-C()/° 115.8 114.7 116.3 – 116.7 115.8 112.1

C()-C()-C()/° 114.2 111.6 114.0 113.6 113.6 111.8 –

N-C()-C()-C()/° 52.4 51.7 52.4 – 63.1 61.1 –

C(1)-C()-C()-C()/° -73.9 -72.1 -73.7 -75.9 – – – - -177.7 -177.7 -177.4 -179.9 O=C(1)-O- C()/° 177.9

O=C(1)-O-H/° 179.0 179.7 178.0 177.9 – – 179.9

O=C(1)-C()-N/° -169.8 171.9 -165.3 – – – -179.9 μ (Debye) 5.19 5.44 5.44 4.10 1.26 1.30 a B3LYP/6-311++G** calculations[53]. b MP2/6-311G** calculations[334]. c B3LYP/TZVP calculations.

4.4. Theoretical and experimental inner shell spectra of L-Phe

Inner shell spectra of molecules clearly indicate the relationships between the chemical shifts and their chemical environments. Fig. 4.5 compares the recently measured synchrotron sourced core level XPS spectra of L-Phe[336] in the gas phase against the theoretical spectra

generated using the LB94/et-pVQZ and EKS methods. In the simulated spectra, the full

Intra-molecular interactions of Phenylalanine… Chapter 4

width at half maximum (FWHM) of 0.4 eV is employed in order to reproduce the experimental spectra[336] as much as possible.

Fig.4.5: Simulated and measured[336] (a) C 1s, (b) N 1s and O 1s spectra of L-Phe.

Fig. 4.5(a) compares the simulated and the experimental C 1s spectra of L-phe, where a binding energy red shift of 0.67 eV is applied to the LB94 calculation. Such a global energy shift can reduce certain systemic errors introduced in the model. However, the C 1s spectrum

calculated using the EKS method is directly compared with the experiment without any energy scaling. Remarkable agreements between the calculated and the experimental[336]

spectra are achieved. Indeed the C 1s peak positions, i.e., C() > C() > C(> C(> C(ring), and

Intra-molecular interactions of Phenylalanine… Chapter 4

their relative intensities in the experimental spectrum are accurately reproduced by the

calculations. However the C(1) peak displays a larger discrepancy between the experiment

and calculations. It is seen that the LB94 model tend to underestimate the C(1)(=O) energy by

1.24 eV, whereas the EKS model is able to reduce the discrepancy to 0.81 eV by taking consideration of the relaxation effects[79 , 353]. It indicates that the electron correlation involved in the C=O structure is particularly large for the DFT methods employed.

Fig. 4.5(b) compares the theoretical and the experimental N 1s and O 1s spectra of L-

Phe. Global energy shifts of 0.71 eV (N 1s) and 0.50 eV (O 1s) are applied to the EKS spectra, while the LB94 calculations require slightly larger energy shifts such as 1.35 eV (N 1s) and 3.61 eV (O 1s), in order to reproduce the experimental spectra. After the energy shifts, the calculated N 1s and O 1s spectra of L-Phe show excellent agreements with the experiment. As indicated by Zhang et al[336], the measured N 1s spectrum of L-phe presents a small shoulder at ca. 405.5 eV that may be due to the conformational effects of L-Phe. As the current study is only based on the global minimum structure of L-Phe, this small shoulder does not appear in the simulated N 1s spectra.

Accurate calculations of the inner shell ionization potentials (IPs) of bio-molecules

remain a challenge. The EKS method employed in the present study provides more accurate

inner shell vertical IPs, where the discrepancy |E|(exp-theory) is mostly < 1 eV, due to the considerations of orbital relaxation. However, it is useful to note that this model is computationally expensive and also laborious, as the energies of the parent molecule and the individual cations need to be calculated separately before calculating the vertical IPs for each atom. In addition, ionization states of a molecule can be more difficult to converge than the ground state. However, this model is not suitable for molecules with degenerate orbitals, such as benzene[322, 323]. On the other hand, the DFT-LB94 model is able to calculate the vertical IPs with competitive accuracies and also with a single calculation of the ground

electronic state of the molecules, irrespective of their symmetries. Therefore both EKS and LB94 models are applied to calculate inner shell energies, where the former is used to calculate the inner shell spectra of L-Phe, alanine, PEA and PPA, but the latter is used to calculate the benzene inner shell spectrum.

Intra-molecular interactions of Phenylalanine… Chapter 4

4.5. Impacts of functional groups on L-Phe

4.5.1. Inner shell chemical shift and spectra Table 4.2 reports the inner shell vertical IPs of the most stable configurations of L-Phe

and its derivatives, PEA and PPA calculated using the LB94/et-pVQZ and EKS methods. The calculated IPs of L-Phe are compared with those from synchrotron sourced XPS measurements[336]. The calculated and the measured inner shell vertical IPs exhibit an

excellent agreement. As it can be seen, the EKS IPs are closer to the experiments, when

compared to those of LB94. For instance, the |IP|(EXP-EKS) of C(in Phe is 0.10 eV, whereas

the |IP|(EXP-LB94) is 0.41 eV. Although LB94 shows slightly larger energy differences it is able to produce relative C 1s energies. The IPs of the aromatic molecules follow a trend as

C(Phe) > C(PPA) > C(PEA), for example, the IPs of C(, calculated by the EKS method, are as follows: 292.19 (alanine) > 292.00 (Phe) > 291.37 (PPA) > 291.26 eV (PEA). However,

the C(1) atom and the lowest IP C-sites (phenyl) are exceptions from this trend. The C(1) energy in PPA is larger than the L-Phe, while its lowest IP sites are smaller than that of PEA. This indicates that that the substitutions of functional groups induce considerable effects in the vertical IPs of the model molecules.

Table 4.2 Comparison of core ionization potentials (IPs) of the ground electronic structures of Phe,

PPA and PEA calculated using the LB94/et-pVQZ model (in eV) and EKS methods (in eV) along with available experiments and other works.

L-Phe PPA PEA

Assignment Other This This LB94 E Exp[336] E E KS Work[336] work KS work KS C 293.02 294.04 294.85 293.61 293.42 294.41 - - (1) C 291.49 292.00 291.90 291.46 290.62 291.37 290.28 291.26 (α) C 290.08 290.88 - 290.29 290.00 290.91 289.59 290.51 (β) C (ring) 290.02 290.49 290.30 289.91 289.84 290.34 289.73 290.26 (γ) C-(ring) 289.73 290.17 - 289.47 289.53 290.16 289.44 290.07

C-(ring) 289.71 290.31 - 289.50 289.54 290.23 289.48 290.18

C-(ring) 289.65 290.29 - 289.57 289.49 290.15 289.44 290.09

C-(ring) 289.63 290.37 - 289.43 289.48 290.17 289.48 290.18

C-(ring) 289.56 290.32 - 289.55 289.40 290.03 289.44 290.08

N 403.70 405.74 405.70 404.95 - - 402.49 404.85

O(H) 534.36 537.47 538.05 536.50 534.69 537.85 - -

O(=C) 535.86 539.11 539.87 538.13 536.41 539.87 - -

Intra-molecular interactions of Phenylalanine… Chapter 4

Fig. 4.6(a) presents the C 1s spectra (FWHM=0.4 eV) of the model molecules

calculated using the EKS method, except for benzene. The benzene spectrum is calculated using the LB94 model and is shifted by 1 eV to match the L-Phe peak at 290.5 eV. In the energy region of ca. 290.5 eV, interactions between the alaninyl (i.e., amino acid moiety) and

phenyl groups of L-Phe reduce the high symmetry of benzene (D6h), causing the degenerated phenyl core states to split in L-Phe, as indicated in our previous study[367]. That is, the symmetric phenyl spectral peak in benzene is asymmetric in Phe, PPA and PEA, due to the

closely located side chain C() peak. While the peaks < 290.50 eV are from phenyl

contributions, the peaks > 290.50 eV represent the alaninyl energy patterns (C(1) > C( >

C(), especially in L-Phe and PPA. Since PEA does not possess the carboxyl group, it does

not show a C(1) peak at ~294 eV. However, other C 1s peak positions shift to accommodate

the side chain modifications in the molecules without changing the energy order of C(ring) <

C(γ) < C() < C(α) < C(1). For example, the C(, C( and C( peaks show consistent shifts to the lower energy side in the aromatic molecules when compared to the alanine peaks. Larger

effects are observed in the C( peaks of PPA and PEA due to the removal of the COOH and

NH2 groups. Further, while all the carbon sites undergo “red” shifts, the C( site in PPA shows a “blue” shift (to the larger energy side). Thus the chemical shifts reflect significant

functional group impacts on the C(-C(-C( chain of L-Phe and its derivatives.

The C 1s energy diagram showing their relative energy is given in fig. 4.6(b), where

E1 (left hand panel) reveals the energy between phenyl and alaninyl in L-Phe with respect

to their free “fragment” molecules, alanine and benzene. E2 (middle panel) and E3 (right hand panel) reveal interactions of the amino group and the carboxyl group respectively, with

the remaining of the L-Phe fragments. As shown by E1, all C 1s IPs of L-Phe decrease with

respect to the corresponding sites in alanine, while the C() site in L-Phe increase against benzene. The energies indicate that the interactions of functional groups in L-Phe are quite

different from the cases of free alanine or benzene. In particular, the IP of C( site on the

phenyl ring increases as large as 0.99 eV. The D6h high point group symmetry of benzene is broken in L-Phe, so that the charge distribution in L-Phe side chain is no longer the same on

all carbon sites of the hexagon ring. Similarly, the next most affected carbon site is C(,

which is 0.22 eV smaller in L-Phe than that of alanine. The C( is the modification site in alanine where a hydrogen atom is replaced by the phenyl attachment to form L-Phe. Hence,

the C(and C( sites in L-Phe are found to be significantly affected, due to the formation of

Intra-molecular interactions of Phenylalanine… Chapter 4

the two fragments, alaninyl and phenyl. However, these effects also tend to extend towards

other carbon sites, such as C( and C(1), in L-Phe.

(a)

(b)

Fig.4.6: (a) C 1s spectra of L-Phe and its derivatives along with alanine and benzene (LB94/et-pVQZ)

simulated using the EKS method and (b) C 1s energy correlation diagram of the model molecules.

Intra-molecular interactions of Phenylalanine… Chapter 4

The trend in E2 provides information of the carbon sites of Phe and PPA, thereby

revealing the effects of NH2 removal. The decrease in IPs at the C( and C( sites of PPA is

observed when compared to L-Phe and whereas, the IPs of the C(1) and C( sites increase in

their energies. It is noted that the energy difference at C( site is almost negligible with only

0.03 eV, while C(1) shows a small energy raise in PPA of 0.37 eV. The removal of NH2 group in PPA affects the intra-molecular H-bond between the carboxyl and amino groups, which in

turn affects the ionization energy of the C(1) 1s atom. As expected, the largest effect is

apparent at the C( site, with an energy difference of 0.63 eV, where the substitution takes place. The ionization energy shifts associated with the removal of the COOH group from L-

Phe, forming PEA (E3), show that all the C-K energies drop, while the largest energy drop

of 0.74 eV is observed at the C( site again. Therefore, the energy shifts indicate that the

major impacts of the functional groups are seen at C( and C( sites, while the C( sites

display lesser effects in the C(-C(-C( bridge. In summary, the largest chemical shift in L-

Phe occurs when the chirality of the C( site disappears by removing the amino or carboxyl groups.

4.5.2. Charge re-distribution Hirshfeld charges of the molecules which are listed in table 4.3 also support the chemical shifts discussed above. As seen in other bio-molecules such as aliphatic amino acids[57], adenine[322] and guanine[323], N and O sites of the compounds possess negative

charges due to their larger electronegativity with respect to C. In alanine and Phe, the C() site

is the only carbon site which exhibits negative Hirshfeld charges. If either –COOH or –NH2

group is replaced by a hydrogen atom, the C() carbons in PPA and PEA, which are no longer chiral in nature, become negatively charged to balance the positive charges. However, the roles of carbon atoms in the phenyl ring of the aromatic molecules are different from the role

of benzene. It can be seen in the table 4.3 that the C() site in benzene holds a negative charge so as to balance the positive charge from the positive hydrogen site; whereas in Phe, PPA and

PEA, the C() site is positively charged so that it could balance the negative charges on the

other functional groups, –COOH and – NH2 groups.

Intra-molecular interactions of Phenylalanine… Chapter 4

Table 4.3 Hirshfeld charges of the atom sites in the model molecules calculated using the LB94/et- pVQZ model. Atom Benzene Alanine L-Phe PPA PEA sites

C(1) - 0.21 0.21 0.22 -

C() - 0.02 0.03 -0.06 -0.02 C() - -0.10 -0.07 -0.06 -0.07 C() -0.04 - 0.01 0.01 0.01 N - -0.24 -0.21 -0.25 O(=C) - -0.30 -0.21 -0.27 - O(-OH) - -0.20 -0.29 -0.29 -

4.5.3. Valence shell ionization spectra The core shell ionization spectra describe the local effects on the atoms due to their chemical environment. Valence ionization spectra, on the other hand, can provide details on the intra-molecular interactions in response to functional group changes. Fig. 4.7 compares the valence ionization spectra of L-Phe and its derivatives together with the L-Phe fragments, benzene and alanine. The spectra are simulated using the SAOP/TZ2P model and with a FWHM of 0.4 eV.

Fig.4.7: Vertical valence ionization spectra of L-Phe and its derivatives obtained using the SAOP/et- pVQZ model and with a FWHM of 0.4 eV.

Intra-molecular interactions of Phenylalanine… Chapter 4

The valence spectra present several similarities and differences that can be related to their side chain modifications. Based on their orbital characters given in fig. 4.8-4.9, the valence spectra (in fig.4.7) of the model molecules can be categorized as three groups such as, functional group related valence region (> 26 eV), fragment related valence region (12-26 eV) and frontier valence region (< 12 eV). The functional group related and the frontier valence regions display certain closely related features. Note that the orbital diagrams of the MOs in frontier region and functional group region in all the molecules are provided in fig. 4.8(a) and (b), respectively.

(a)

(b)

Fig.4.8: Orbital diagrams of the MOs in (a) frontier region and (b) functional group regions of all the molecules.

Intra-molecular interactions of Phenylalanine… Chapter 4

The outer most peaks (i.e., frontier valence region) of L-Phe and its derivatives, PPA and PEA, at ~10 eV can be related to the HOMO peaks of alanine (24a) and benzene (the

doubly degenerated 1e1g orbitals) based on their molecular orbital features given in fig. 4.8(a). The HOMO and HOMO-2 MOs of Phe (44a, 42a), PPA (40a, 38a) and PEA (33a, 31a) are likely to be the mixed contributions of the HOMOs of alanine (24a) and benzene

(1e1g). Whereas, the HOMO-1 orbitals of L-Phe (43a), PPA (39a) and PEA (31a), are dominated by the π orbital contributions on the phenyl ring, which can be correlated with the

second orbital in the 1e1g degenerate pair of benzene.

In inner most valence region of > 26 eV in fig. 4.7, some closely relevant peaks are seen in alanine, Phe and PPA but not in PEA and benzene that are dominated by the functional groups such as, carboxyl and amino groups. For instance, the two innermost peaks that are almost closely located in alanine (7a, 8a), Phe (13a, 14a) and PPA (12a, 13a) are dominated by the s-electron contributions from their carboxyl groups, which are also seen in aliphatic amino acids[57]. Similarly, the peak at ca. 27 eV in alanine (9a) and Phe (15a) is not seen in any other molecules in this study. This indicates that the MOs 9a and 15a in alanine and Phe, respectively, are apparently due to the interactions between amino, methyl and carboxyl groups, although the carboxyl group plays a less important role. The orbital diagrams of these MOs are given in fig. 4.8(b).

The valence region 12-26 eV in the spectra given in fig. 4.7 are dominant by the fragment related properties and can further be split into two categories as ‘partially delocalized’ valence region of 17-26 eV and more complex or fingerprint valence region concentrating in 12-16 eV. In the former valence region (i.e., 17-26 eV), the peaks show differences in Phe and its derivatives, yet exhibit some relationships with the fragment molecules, alanine and benzene, while the latter region (i.e., 12-16 eV) is made of complex interactions and are mostly different in all the molecules.

For instance, the valence space of 17-26 eV in the spectra (provided in fig. 4.7) shows contributions from eight orbitals, three from alaninyl and five from phenyl moieties. Note that benzene has two pairs of doubly degenerate orbitals in this region. These orbitals are dominated by the C 2s electrons. Fig. 4.9 provides the corresponding orbital diagrams of the MOs that are related to alanine MOs (a) and benzene MOs (b), with IPs within 17-26 eV. Orbitals diagrams of MOs 10a, 11a and 12a of alanine correlate with the respective MOs of

Intra-molecular interactions of Phenylalanine… Chapter 4

L-Phe, PPA and PEA as follows: 12a (alanine) – 21a (L-Phe) – 19a (PPA) – 15a (PEA); 11a (alanine) – 20a(L-Phe) – 18a (PPA) – 14a (PEA) and 10a (alanine) – 17a (L-Phe) – 15a(PPA). On the other hand, the phenyl dominant MOs of Phe, PPA and PEA can be

correlated with the MOs 2a1g and two degenerate orbitals (2e1g and 2e1u) in benzene. Their orbital diagrams are apparently related, as shown in fig. 4.9(b).

Fig.4.9: (a) Alaninyl and (b) phenyl dominant MOs of L-Phe and its derivatives within 17-26 eV.

100

Intra-molecular interactions of Phenylalanine… Chapter 4

Nevertheless, the valence region within 12-16 eV (fig. 4.67) display many differences in the valence spectra that are due to complex interactions within the individual molecules and hence are different from each other. The MOs in this intermittent valence region are again mostly dominated by the fragment interactions within the species. Therefore, comparing the fragment molecules can be useful to comprehensively understanding the valence electronic structure of L-Phe. The following part of the chapter will explore the valence MOs of L-Phe along with those from its fragments in alanine/benzene scheme and glycine/toluene scheme, using the combined position space and momentum space details, i.e., the DSA approach[193].

4.6. Fragmentation in L-Phe

4.6.1. Valence spectra of L-Phe and its fragments

As shown in fig. 4.3, Phe can be dominated by two different fragment schemes, i.e., alanine/benzene scheme (scheme I) and glycine/toluene scheme (scheme II). The electronic ground states of L-Phe and its fragment molecules, alanine/benzene and glycine/toluene are all closed shells. Benzene has 15 occupied valence MOs including five doubly degenerated MOs, L-alanine and toluene has 18 valence MOs each, glycine has 15 valence MOs and 1 1 whereas L-Phe has 32 valence MOs. L-Phe (X A) and alanine (X A) possess a C1 point 1 1 group symmetry, whereas glycine (X A’) and toluene (X A’) exhibit a Cs symmetry. Benzene 1 holds a higher D6h symmetry (X A1g). Tables 4.4 - 4.6 compare the valence vertical IPs of L- Phe and its fragment molecules calculated using the DFT based SAOP/et-pVQZ (SAOP/TZ2P for alanine and benzene) and Green’s function based OVGF/TZVP models along with the available measured IPs from literature[368-372]. The spectroscopic pole strengths (ps) of the valence IPs in L-Phe and its fragment species calculated by the OVGF

model are larger than 0. 85, except a very few MOs such as, 1a2u in benzene (0.81), 14a’ in toluene (0.84) and 37a in L-Phe (0.84). Therefore the approximation by OVGF model holds suitable for the present study. For instance, the IPs between OVGF and experiments for the

three outermost MOs in benzene such as, 1e1g, 3e2g and 1a2u are only 0.07 eV, 0.59 eV and 0.03 eV, respectively, as shown in table 4.4.

101

Intra-molecular interactions of Phenylalanine… Chapter 4

Table 4.4: Valence IPs (eV) for benzene and toluene calculated using the SAOP and OVGF methods along with the available experimental values.

Benzene (D6h) Toluene (Cs)

Orbital SAOP* OVGF* (ps) Expa Orbital SAOP^ OVGF*(ps) Expb

2a1g 25.36 25.90 6a' 25.53

2e1u 22.59 22.50 7a' 23.35

2e2g 18.94 19.20 3a" 22.64

3a1g 16.95 17.44 (0.85) 16.90 8a' 21.14

2b2u 15.35 15.87 (0.87) 15.50 4a" 19.08

1b1u 14.88 14.84 (0.87) 14.80 9a' 18.55 3e1u 14.30 14.41 (0.88) 13.90 10a' 16.71 16.89 (0.86)

1a2u 13.11 12.33 (0.81) 12.30 5a" 15.36 15.18 (0.87) 16.49

3e2g 12.29 12.09 (0.90) 11.50 11a' 15.22 15.40 (0.87) 15.58

1e1g 10.39 9.13 (0.90) 9.20 12a' 14.49 14.48 (0.88) 15.12 (HOMO) 6a" 14.31 14.23 (0.88) 14.04 13a' 14.04 13.82 (0.90) 13.90 7a" 13.48 13.52 (0.90) 13.29 14a' 12.92 11.94(0.83) 12.01 8a" 12.31 11.82 (0.90) 11.86 15a' 12.23 11.80 (0.90) 11.45 9a" 10.45 8.96 (0.90) 9.00 16a' 10.22 8.70 (0.90) 8.76 (HOMO) *This work (LB94/et-pVQZ; SAOP/TZ2P; OVGF/TZVP); ^ This work (SAOP/et-pVQZ) a Ref. [368]; b Ref. [373] .

Table 4.5: Valence IPs (eV) for alanine and glycine calculated using different methods together with the available experimental and other theoretical values. Alanine Glycine Orbital SAOP* OVGF* Expa Orbital SAOP# OVGF* Expb 7a 31.93 6a' 32.07 34.30 8a 29.89 7a' 30.01 32.30 9a 27.31 8a' 27.42 28.30 10a 23.66 9a' 22.38 23.30 11a 20.53 10a' 19.22 20.20 12a 18.98 1a" 17.74 17.88 (0.90) 17.60 13a 17.88 17.88 (0.90) 11a' 17.36 18.31 (0.91) 16.90 14a 17.19 17.97 (0.91) 12a' 16.56 16.66 (0.91) 16.60 15a 16.35 16.83 (0.90) 16.60 2a" 15.54 15.49 (0.90) 15.80 16a 15.45 15.61 (0.90) 13a' 15.03 15.26 (0.91) 15.00 17a 14.99 15.51 (0.90) 14a' 13.97 13.58 (0.91) 14.40 18a 14.47 14.63 (0.91) 14.80 3a" 13.67 13.86 (0.92) 13.70 19a 13.99 14.13 (0.92) 4a" 12.24 11.47 (0.91) 12.20 20a 13.44 13.25 (0.91) 13.40 15a' 11.95 11.51 (0.90) 11.20 21a 13.00 13.21 (0.92) 12.8 16a' (HOMO) 10.78 10.12 (0.92) 10.00 22a 12.10 11.44 (0.91) 12.10 23a 11.64 11.41 (0.90) 11.00 24a(HOMO) 10.45 10.09 (0.91) 9.85 *This work (SAOP/TZ2P; OVGF/TZVP); ^This work (SAOP/et-pVQZ); a Ref.[273];b Ref.[320].

102

Intra-molecular interactions of Phenylalanine… Chapter 4

Table 4.6: Valence IPs (eV) of L-Phe calculated using the SAOP and OVGF models along with the available measured values.

OVGF* Orbital SAOP* Exp (pole strength) 13a 32.00 14a 29.95 15a 27.26 16a 25.65 17a 23.82 18a 22.76 19a 22.42 20a 20.67 21a 19.22 22a 19.08 23a 18.64 24a 17.73 25a 16.85 17.23 (0.87) 26a 16.76 17.10 (0.90) 16.5a 27a 15.95 16.02 (0.90) 28a 15.59 15.56 (0.89) 29a 15.37 15.61 (0.89) 30a 15.28 15.32 (0.90) 31a 14.82 14.76 (0.90) 14.9a 32a 14.42 14.43 (0.90) 33a 14.38 14.17 (0.90) 34a 14.02 13.69 (0.90) 14.0a 35a 13.70 13.35 (0.90) 13.4a 36a 13.18 12.74 (0.91) 37a 13.03 12.22 (0.84) 38a 12.39 11.86 (0.91) 39a 12.35 11.72 (0.91) 40a 12.06 10.80 (0.91) 41a 11.59 10.96 (0.90) 11.9b 42a 10.66 9.61 (0.91) 43a 10.55 9.03 (0.90) 10.9b 44a 10.35 8.80 (0.90) 9.4c *This work (SAOP/TZ2P; OVGF/TZVP). a Unresolved and unassigned intensities[273]. b Band intensity indicates other band(s) present, as well as ,Ref. [273]. c Broad band encompassing several states , Ref. [273].

The OVGF model produces only for the outer valence vertical IPs of the molecules, in order to obtain the IPs in the complete valence space, the SAOP model is, therefore, employed. The valence vertical IPs produced by the SAOP model exhibit comparable accuracy with the OVGF/TZVP model and with the experiment in almost the entire valence space. It is noted that the SAOP model produces less accurate vertical IPs for the frontier

103

Intra-molecular interactions of Phenylalanine… Chapter 4

orbitals of some molecules[64, 325, 327], such as aliphatic amino acids[57]. This is true for the molecules in this study. The IP energy shift between SAOP and experiments of the three

outermost MOs (1e1g, 3e2g and 1a2u) in benzene is 1.19 eV, 0.79 eV and 0.81 eV, respectively (refer to table 4.4). These discrepancies may attribute to orbital relaxation, self-energy[374] and configurational effects[56 ], that stem from the “meta” –Koopman approximation[175]. Similar trends are seen in the valence vertical IPs of other molecules such as L-Phe (table 4.6), and its phenyl (i.e., benzene and toluene given in table 4.4) and aliphatic fragments (i.e., alanine and glycine provided in table 4.5).

Indeed it is important to note that the experimental binding energies (i.e., IPs) themselves are averaged values, as the valence peaks in the experiments are mostly broad and congested. In addition, the vertical ionization energies in table 4.6 are for the global minimum conformer of L-Phe given in fig. 4.2. Available PES of Phe shows the presence of several low-lying conformations which contribute to the measurement with congested IPs[273, 334, 336]. This comes with no surprise as a recent resonant two-photon ionization (R2PI) experiment demonstrates that the conformer dependency on binding energies of the HOMO could range from 8.80 to 9.15 eV with a variation of as much as 0.35 eV[340]. Furthermore, amino acids with complex side groups are likely to have broader peaks in their PES as the orbital energies of these side chains (in the present case, phenyl) are usually comparable to main chain peaks. A fragment approach can be useful to resolve these issues.

Fig. 4.10(a) and fig. 4.10(b) presents the valence spectra of glycine vs. alanine (i.e., glycine + methyl) and benzene vs. toluene (i.e., benzene + methyl) respectively. The valence spectra demonstrate that the methyl group attachments in alanine and toluene produce some extra peaks that are not present in glycine and benzene respectively. Our previous studies[53, 57] indicated that the valence MOs (11a, 12a, 19a and 20a) in alanine as methyl signature orbitals, while the MOs within 12-16 eV present fingerprint interactions of the aliphatic molecules (refer to fig. 4.10(a)). This is also true with the valence spectra of benzene and toluene given in fig. 4.10(b). The valence spectra of toluene present some extra peaks such as, 8a’, 12a’ and 7a”, those do not exist in the binding energy spectra of benzene. Moreover

the five doubly degenerate valence MOs of benzene (D6h symmetry) are split into five pairs

of MOs in toluene (Cs symmetry), i.e., 2e1u  7a’, 3a” (toluene); 2e2g  4a”, 9a’ (toluene);

3e1u  6a”, 13a’ (toluene); 3e2g 15a’, 8a” (toluene) and 1e1g  9a”, 16a’ (toluene), respectively. Therefore, the similarities and differences between the fragment molecules, i.e.,

104

Intra-molecular interactions of Phenylalanine… Chapter 4

glycine vs. alanine and benzene vs. toluene, can be useful to explore the binding energy spectrum and relevant electronic properties of L-Phe.

(a)

(b)

Fig. 4.10: Vertical valence ionization spectra of (a) glycine vs. alanine and (b) toluene vs. benzene obtained using the SAOP model.

105

Intra-molecular interactions of Phenylalanine… Chapter 4

Fig. 4.11 compares the valence spectra of L-Phe against its fragment schemes, alanine/benzene scheme (fig. 4.11(a)) and glycine/toluene scheme (fig. 4.11(b)), simulated using SAOP model and with a FWHM of 0.4 eV. The L-Phe spectra are given in the middle panels, while the lower panels display spectra of the respective aliphatic fragments, i.e., alanine (fig. 4.11(a)) and glycine (fig. 4.11(b)) and their upper panels show the spectra of their corresponding phenyl fragments such as, benzene (fig.4.11(a)) and toluene (fig. 4.11(b)). The valence spectrum of L-Phe with respect to its fragment schemes is more complex than the side chain substitutions of aliphatic amino acids[57] as discussed in chapter 3. This is partly due to the involvement of the aromatic ring in L-Phe. The L-Phe valence spectrum indeed reveals certain correlations with its fragment schemes, alanine/benzene and glycine/toluene.

A clear fragments-in-molecules picture[54] pops up in the spectra. However, the correlations in benzene-Phe-alanine in fig. 4.11(a) and toluene-Phe-glycine in fig. 4.11(b) display some useful trends. Some of the peaks in L-Phe spectrum are very similar in both the fragment schemes, correlating with either Phenyl (i.e., benzene or toluene) or aliphatic (alanine or glycine) fragments or both. For instance, the valence IPs > 26 eV display remarkable similarities in both the fragment schemes. The three innermost valence peaks in L-Phe (13a – 15a) are very similar to those of alanine (7a-9a) and glycine (6a’-8a’), and the

fourth innermost valence peak in L-Phe (16a) is correlated with benzene (2a1g) and toluene (6a’). Similarly, the outermost valence peaks at ~10 eV are almost unaffected. We have seen in the earlier part of this chapter that the frontier MOs of L-Phe and its derivatives, PEA and PPA are dominated by contributions of both the phenyl and aliphatic fragments[144]. This reflects in the valence spectra of L-Phe and its fragment molecules, too. The three congested outermost MOs of L-Phe (42a-44a) are related to two MOs of phenyl fragments (i.e., 9a” and

16a’ in toluene and a degenerate 1e1g MO in benzene) and the HOMOs of aliphatic fragments (i.e., 24a in alanine and 16a’ in glycine). It shows that the fragmentation schemes do not impact the frontier MOs of L-Phe very much, so do its inner most MOs > 26 eV.

Unlike the inner most (> 26 eV) and the outer most (~10 eV) valence peaks, the intermittent region, 26 eV > IP > 11 eV, display relatively larger differences between the benzene-Phe-alanine (fig. 4.11(a)) and toluene-Phe-glycine (fig. 4.11(b)) valence spectra. For instance, the MO 17a of L-Phe is related to an aliphatic counterpart in the former scheme (i.e., MO 10a of alanine), while the same MO of L-Phe is correlated with a phenyl fragment

106

Intra-molecular interactions of Phenylalanine… Chapter 4

(i.e., 7a’ of toluene) in the latter scheme. Another example includes 20a of L-Phe, which is correlated with 11a of alanine in the benzene/alanine scheme but, with 8a’ of toluene in the alternate fragment scheme. It indicates that the fragmentation scheme relates different configurations of these MOs of L-Phe, which is more apparent in the mid valence region of 11-20 eV of L-Phe.

(a)

(b)

Fig. 4.11: Vertical valence ionization spectra of (a) benzene-Phe-alanine and (b) toluene-Phe-glycine sets calculated using the SAOP model and with a FWHM of 0.4 eV.

107

Intra-molecular interactions of Phenylalanine… Chapter 4

In order to further explore the valence space from the fragments-in-molecules point of view, the synthetic spectra of L-Phe are created by a simple superposition of the aliphatic and phenyl fragments in each scheme. Fig. 4.12 compares the L-Phe spectrum (in the middle panel) and the synthetic spectra generated from alanine + benzene (in upper panel) and glycine + toluene (in the lower panel). This figure helps to unravel the mid-valence region of 20 eV > IP > 11 eV that generally is complex due to strong intra-molecular interactions[54, 57, 144]. Interestingly, the MOs 21a – 34a of L-Phe concentrating in the valence region of 14 eV – 20 eV are almost identical to the synthetic spectrum generated from glycine and toluene fragment schemes. Nevertheless, the peaks in the valence region of 11 eV – 14 eV of L-Phe are nearly identical to those in the alanine + benzene synthetic spectrum. These indicate that the inner most (> 26 eV) and the outer most (~10 eV) valence regions of L-Phe are dominated by its functional groups (see fig 4.12), the mid valence regions are controlled by the strong fragment interactions. Employing the fragments-in-molecules approach[54] can therefore be helpful to understand the fragment-related intra-molecular mechanisms of L-Phe.

Fig. 4.12: Comparison of the vertical valence ionization spectra of native L-Phe (middle panel) against the synthetic spectra simulated from alanine + benzene (upper panel) and (b) glycine + toluene (lower panel) fragment schemes.

108

Intra-molecular interactions of Phenylalanine… Chapter 4

4.6.2. Valence orbital topologies In molecular orbital theory, orbitals are responsible for bond formation and breaking, as well as interactions among component atoms and fragments. Orbitals reveal fundamental structural information of ionization spectra and other one particle related phenomena. Changes in orbitals are indicators of chemical bonding[54]. Fig. 4.13 displays the inner (a) and outer valence (b) orbital diagrams of the related fragments. The left side panels in fig 4.13 present the orbital correlations between benzene-Phe-alanine set and the right side panels display those of toluene-Phe-glycine. The solid lines in the figure represent the orbital correlations between L-Phe and its phenyl fragments, benzene or toluene, and the dot lines show L-Phe correlations with its aliphatic fragments, alanine or glycine. Orbital correlations from both the fragments are given as dashed lines.

Due to its lower point group symmetry, L-Phe (C1) does not engage with any energy

degenerate orbitals in this diagram, so do the MOs of alanine (C1), glycine (Cs) and toluene

(Cs). However, benzene possesses five doubly degenerate orbitals in its valence space, i.e.,

1e1g, 3e2g, 3e1u, 2e2g and 2e1u, in which, two of the MOs (2e2g and 2e1u) appear in the inner

valence and the rest of them (1e1g, 3e2g, 3e1u) are in the outer valence region. The orbital correlations between L-Phe and its fragment schemes, benzene/alanine scheme (left panels in fig. 4.13) and toluene/glycine scheme (right panel of fig. 4.13) are not strictly on one-to-one basis, since “free fragments” employed in the present study are molecules, rather than ions or radicals. Hence the orbital correlations are carefully made based on their orbital characteristics of the molecules. Correlation diagrams with orbitals are provided in the appendix (A.III - A.IV).

As described by the valence spectra, the innermost valence orbitals in the energy region > 24 eV indeed reflect quite clear similarities. The orbitals 13a, 14a and 15a of L-Phe are linked with the corresponding innermost MOs of alanine (7a, 8a and 9a) and glycine (6a’, 7a’ and 8a’), those are dominated by the 2s contributions from their carboxyl and amino

groups respectively. The next innermost orbital 16a of L-Phe exactly matches with the 2a1g of benzene and 6a’ of toluene, indicating apparent phenyl dominance in these orbitals. The following set of MOs in the inner valence include the two doubly degenerate orbitals of

benzene that split into ‘group of three orbitals’ in L-Phe, i.e., 2e1u  17a, 18a, 19a; 2e2g  21a, 22a, 23a, which also correlate with MOs 10a and 12a in alanine.

109

Intra-molecular interactions of Phenylalanine… Chapter 4

(a)

(b)

Fig. 4.13: Valence orbital correlation diagrams of L-Phe with its fragment schemes: alanine/benzene and glycine/toluene in the (a) inner valence and (b) outer valence regions.

Similar interactions are also seen in the toluene-Phe-glycine set, where 4 MOs of toluene (7a’, 3a”, 4a” and 9a) along with 9a’ and 10a’ of glycine make up corresponding orbitals in L-Phe (i.e., 17a-19a and 21a-23a). Noticeable differences in the correlation of L-

110

Intra-molecular interactions of Phenylalanine… Chapter 4

Phe and its fragment schemes are seen for MO 20a in L-Phe. It is interesting to identify that this orbital, which is correlated with 11a of alanine in benzene/alanine scheme, is associated with both toluene (8a’) and glycine (9a’) in the alternate scheme.

In the outer valence region, the three degenerate orbitals of benzene are correlated to

two MOs each in the L-Phe (i.e., 3e1u  32a, 33a; 3e2g  38a, 39a; 1e1g  43a, 44a). Of these MOs, 33a, 39a and 44a of L-Phe are also associated with the corresponding L-alanine MOs (18a, 22a and 16a’). In the case of glycine/toluene scheme, the three degenerate orbitals

of benzene in the outer valence split into two orbitals in toluene (i.e., 3e1u  6a”, 13a’; 3e2g

 8a”, 15a’; 1e1g  9a”, 16a’). These MOs in toluene interact with the corresponding valence MOs of glycine (3a”, 4a” and 16a’) and are correlated with the corresponding orbitals of L-Phe. The most interesting differences in the outer valence region happen in the MOs 36a and 37a of L-Phe. The MOs 36a and 37a of L-Phe in alanine/benzene scheme are

correlated with the MOs of both benzene (1a2u) and alanine (21a). But in the glycine/toluene scheme, these orbitals of L-Phe (36a – 37a) are only correlated with 14a’ in toluene.

Moreover all the MOs in the alanine/benzene scheme are correlated with the orbitals of L-Phe. However a few orbitals in the glycine/toluene scheme such as, 12a’ in toluene, 35a in L-Phe and 14a’ in glycine, are unique in their orbital characters and cannot be related with any other MOs. Finally, the outer most valence orbitals of alanine/benzene (left panel) and glycine/toluene (right panel) shown in fig. 4.13(b), in fact, do not mix as much as those orbitals in the energy region of 11 < IP < 20 eV. The outer most or frontier groups of orbitals in L-Phe can easily be identified as the phenyl dominant and aliphatic dominant orbitals. Therefore the fragments based orbital topologies indicate that the interactions between the alanine/benzene and glycine/toluene schemes in L-Phe are very different that in turn impact the intra-molecular interactions of L-Phe accordingly. This is further verified by combining the orbital features and momentum profiles of the molecules using the DSA approach[193].

4.6.3. Momentum distributions and chemical bonding features

Position space analyses provide orbital information in qualitative manner. It indicates that the interactions of alanine/benzene and glycine/toluene schemes in L-Phe are quite

111

Intra-molecular interactions of Phenylalanine… Chapter 4

different in its intra-molecular interactions. Hence, probing the valence space information using the momentum space and later combining the data with position space by means of DSA[193] can reveal subtle quantitative orbital-based details of L-Phe and its fragment schemes. All valence orbital theoretical momentum distributions (TMDs) of L-Phe and its fragment schemes, alanine/benzene and glycine/toluene, have been calculated and examined.

Fig. 4.14 compares the TMDs of two MOs in benzene, i.e., 3a1g (upper panel) and 1e1g (HOMO) (bottom panel), calculated by our model against the earlier EMS measurements[375., 376The] comparisons show that our calculations agree very well with those of the measured[375, 376], especially in the higher momentum region > 0.25 a.u. The

HOMO (1e1g) orbital of benzene show a bell shaped distribution from ‘π’ contributions and

whereas, the TMD of 3a1g orbital display a mixed ‘s, p’ like shape.

Fig. 4.14: Comparison of the theoretical and experimental orbital momentum distributions of the selected orbitals in benzene.

112

Intra-molecular interactions of Phenylalanine… Chapter 4

Similarities in fragment dependent chemical bonding Fig. 4.15(a) – (d) present the TMDs of some of the selected valence orbitals from L- Phe and its fragment schemes, alanine/benzene and glycine/toluene, which display similar

chemical bonding features. For instance, the MOs 16a of L-Phe, 6a” of toluene and 2a1g of benzene are all dominated by the phenyl fragment and hence display similar s-like distributions in their TMDs (fig. 4.15(a)). The momentum spectra of these orbitals are merely identical, except the lower moment range < 0.25 a.u., which can be attributed to the orbital

contributions extending towards C( in toluene and L-Phe. Similarly, the MO 39a of L-Phe

can be correlated with the MOs 3e2g from benzene and 15a’ of toluene based on their similar orbital features, those are strongly dominated by the p type interactions occurring in the ring. Similarities in the electron densities of these MOs are also reflected in their TMDs, that present two ‘p’ like peaks, the intensity of toluene distribution ca. 0.5 a.u. is slightly higher than that of others through (fig. 4.15(b)).

Fig. 4.15: Selected valence orbital electron densities and TMDs of the L-Phe and its fragments showing related chemical bonding characteristics that are dominated by their individual fragments.

113

Intra-molecular interactions of Phenylalanine… Chapter 4

Our position space analyses (i.e., valence spectra) reveal that the outermost or frontier MOs of L-Phe are little affected by the fragment changes. This is verified by the momentum spectra given in fig. 4.15(c) and (d). The orbital densities of NHOMOs of L-Phe (43a),

toluene (9a”) and benzene (degenerate 1e1g) display similar ‘2’ natures and hence present almost identical bell shaped momentum spectra (see fig. 4.15(c)).

Fig. 4.15(d) reports the orbital densities and momentum distributions of HOMO in L- Phe (44a), which despite behaving similarly with respect to its fragment schemes, this orbital undergoes significant interactions. The HOMO (44a) of L-Phe derives the interactions from

both the aliphatic (24a of alanine and 16a’ of glycine) and the phenyl (1e1g of benzene and 16a’ of toluene) fragments, and hence its orbital densities are very similar to the HOMOs of its fragments. However, the associated TMDs display a different picture where, the spectrum of L-Phe (44a) closely resembles the ‘p’ shaped distributions of its phenyl fragments (i.e.,

benzene (1e1g) and toluene (16a’)), those are different from the respective aliphatic counter parts (alanine and glycine). It suggests that the phenyl fragment is very active in the chemical bonding mechanisms of Phe, as found by many biological studies. The aliphatic contributions are not completely neglected, as the apparent shifts in the bell shaped distribution and changes in the lower momentum regions in the HOMO of L-Phe indicate the effects of aliphatic fragments in this orbital.

Valence orbitals dominated by the glycine/toluene scheme In the energy range of 14 eV < IP < 20 eV, the valence spectrum of L-Phe matches with that of glycine/toluene scheme but, not with the alanine/benzene scheme. However, the orbital densities of the MOs in this region look very similar in both the alanine/benzene and glycine/toluene schemes that correlate with L-Phe orbitals. Hence, selected MOs within the energy region of 14 eV < IP < 20 eV are investigated in momentum space, in order to better understand their chemical bonding features.

Fig.4.16(a) – (d) present the TMDs of some selected orbitals in the energy range of 14 eV < IP < 20 eV. Fig. 4.16(a) and (b) compares the mixed ‘s,p’ type momentum spectra of MO 22a of L-Phe with the corresponding MOs in the alanine/benzene (a) and glycine/toluene (b) schemes. As it can be seen, the orbital densities of MOs 12a of alanine and 10a’ of

glycine are very similar to each other and so do the MOs 2e2g and 9a’ in benzene and toluene

114

Intra-molecular interactions of Phenylalanine… Chapter 4

respectively. However, the orbital TMD of 22a in L-Phe is very different from those of

alanine (12a) and benzene (2e2g). The benzene MO displays a bell shaped distribution and the alanine MO displays a partial ‘s,p’ type distribution, which are very different from L-Phe (fig. 4.15(a)). On the other hand, the MOs 9a’ of toluene and 10a’ of glycine display a mixed ‘s,p-‘ like TMDs that match very well with that of L-Phe (22a), where the spectrum of toluene showing high level correlation with the parent spectra (fig. 4.16(b)). It suggests that the MO 22a in L-Phe is strongly dominated by the glycine/toluene scheme.

Fig. 4.16: Orbital densities and TMDs of the selected valence MOs of L-Phe those are dominated by the glycine-toluene scheme.

Similarly, fig. 4.16(c) report the comparative TMDs of the orbitals 27a of L-Phe, 15a of alanine and 12a’ of glycine along with their orbital densities. The densities of these orbitals show some associations with each other, where the aliphatic fragment is more active in L- Phe. In the case of momentum spectra of L-Phe, it displays a mixed s and p like distribution with the ‘s’ features dominating the lower momentum region < 0.5 a.u. Nearly identical

115

Intra-molecular interactions of Phenylalanine… Chapter 4

spectra is seen in 12a’ of glycine but not in 15a of alanine. The TMD of alanine represents a complete p shape which is not equivalent to the L-Phe distribution. This indicates that the 27a orbital of L-Phe is strongly dominated by the glycinyl fragment. Another solid example of a toluene/glycine scheme dominant MO is given in fig. 4.16(d). The figure correlates the

TMDs and their corresponding orbital densities of MOs 29a in L-Phe, 2b2u in benzene and 11a’ in toluene. Despite the orbital densities being almost similar in benzene and toluene MOs, they display very different momentum spectra. Indeed the toluene MO presents two ‘p’

type peaks that match excellently with that of MO 29a in L-Phe. However, the 2b2u MO of benzene presents a singly p-like distribution which neither agrees with L-Phe nor with toluene.

Valence orbitals dominated by the alanine/benzene scheme Similar to the domination of toluene/glycine scheme, a small valence region, 11 eV < IP < 14 eV, is dominated by the benzene/alanine scheme. Fig.4.17 presents a couple of examples, where the MOs of L-Phe are identical to the alanine/benzene scheme, but differ from the alternate scheme. For instance, based on their orbital diagrams, the 38a of L-Phe can

be correlated with the 3e2g of benzene and 8a” of toluene (fig. 4.17(a)). However, L-Phe and benzene are closer to each other, presenting two ‘p’ type peaks in the momentum space that is very different from the distribution of MO 8a” in toluene. Similarly, the TMDs of 40a of L- Phe, 22a of alanine and 4a” of glycine present identical orbital characters dominated by the

2pz orbitals from their C=O and C-O(H) bonds. However, despite of the nearly identical orbital picture, the TMD of L-Phe MO correlates very well with that of alanine MO and not with glycine. This again confirms the dominance of alanine/benzene scheme in the MOs of L-Phe concentrating in valence region of 11 eV < IP < 14 eV, while glycine/toluene scheme have small role in this region.

Orbitals dominated by the strong interactions in L-Phe Fig. 4.18(a) and (b) describe selective orbitals of L-Phe, such as 17a, 19a and 33a, where the glycine and toluene fragments interact significantly. Fig. 4.18(a) presents the orbital diagrams and momentum spectra of 17a and 19a of L-Phe that can be correlated with both the 9a’ of glycine and 7a’ of toluene. Nevertheless, the shapes and intensities of the TMDs of 17a in L-Phe and 7a’ in toluene are close to each other. But the momentum distribution of 19a of L-Phe is close to 9a’ in glycine in its shape and intensity. This suggests that the phenyl ring in toluene and L-Phe serves as a buffer to resist changes where glycinyl

116

Intra-molecular interactions of Phenylalanine… Chapter 4

fragment does not exhibit such functionality so that the shape of glycine dominant fragment may change more easily. This is in agreement with our previous findings[54, 56].

Fig. 4.17: Orbital densities and TMDs of the selected valence MOs of L-Phe those are dominated by the alanine-benzene scheme.

Fig. 4.18: Orbital densities and TMDs of the selected valence MOs that are dominated by the strong interactions in L-Phe and its fragment schemes.

Another example is described in fig. 4.18(b) that compares the spectra and orbital densities of 33a in L-Phe together with the synthetic spectra obtained by simple summation of

the selected MOs in the fragment schemes such as, 18a of alanine + 3e1u of benzene and 3a” of glycine + 13a’ of toluene. As it can be seen the synthetic spectrum of glycine/toluene scheme is closer to the parent spectrum, except the lower momentum range of ~0.02 a.u. However, the synthetic spectrum obtained from the alanine/benzene scheme depicts two

117

Intra-molecular interactions of Phenylalanine… Chapter 4

strong p like peaks that do not match with the L-Phe distribution. This apparently indicates that orbital 33a of Phe holds strong intra-molecular interactions between its glycine and toluene fragments.

4.7. Summary

This chapter investigates the electronic structure of L-Phe along with some of its derivatives and fragment molecules, in order to understand its intra-molecular interactions.

The first part of the chapter reveals the impacts of functional group substitutions on the (C(-

C(-C() chain of L-Phe, by comparing its structure with those of its derivatives, PEA and PPA, together with alanine and benzene. The interactions of L-Phe are dominated by three important functional groups, phenyl, carboxyl and amino groups. The binding energy spectra

show very different chemical shifts in C(, C(and C( sites in PPA and PEA in the absence

of carboxyl group and amino group, with respect to L-Phe. As expected, the C( site that directly connects to the substitution group, change significantly in energy. However, the

impact at C( site is higher in the absence of carboxyl groups that also lead to the significant reduction of the dipole moment in PEA, implying that carboxyl group dominates the inner shell of Phe.

In valence space, the innermost valence (> 26eV) and the frontier valence orbitals (< 11 eV) are remarkably similar, while the intermittent valence regions show differences related to

its fragment schemes. Moreover, the valence MOs that show significant impacts in the C(-

C(-C( carbon backbone are found in the valence space of 17-26 eV that are correlated with

C 2s features of five MOs from benzene (2a1g, degenerate 2e1g and degenerate 2e1u) and three MOs of alanine (10a-12a). Indeed the most complex interactions within the fragments appear in the region 12-16 eV. Therefore, fragment interactions in L-Phe play a very significant role in the valence space.

The second part of this chapter studies the roles of different fragment schemes of L- Phe, alanine/benzene and glycine/toluene, using the DSA, thereby revealing the intra- molecular mechanisms of L-Phe in valence space. The combined orbital momentum distributions and valence binding energies reveal substantial changes within the chemical

118

Intra-molecular interactions of Phenylalanine… Chapter 4

environment of L-Phe due to the fragment schemes. It is observed that the innermost valence space (i.e., > 20 eV) and the frontier orbitals are less affected by the fragment schemes. On the other hand, the valence regions within 11 < IP < 20 eV display fragment related MOs. The valence space, 14 < IP < 20 eV, in L-Phe is dominated by the glycine/toluene schemes, while a short valence space of 11 < IP < 14 eV is dominated by the alanine/benzene schemes. These observations are verified using the combined position and momentum space details. Hence ‘fragments-in-molecules’ approach can be useful to understand the electronic structures and intra-molecular interactions of larger molecules. This chapter also implies that the alanine/benzene and glycine/toluene schemes are likely the dominant configurations which contribute to the electronic structure and interactions of L-Phe.

119

Hydroxyl group effects on aromatic molecules… Chapter 5

CHAPTER5

Effects of the hydroxyl group on aromatic molecules

5.1. Introduction

Aromatic amino acids are important protein building blocks that play principal roles in a number of biochemical and physiological processes[377-382]. The physical and chemical transformations of aromatic amino acids remain significant in the biology of life. The conversion of tyrosine into dopamine neurotransmitter (fig. 5.1) is a significant biochemical transformation of aromatic amino acids[383-386]. An enzyme known as phenylalanine hydroxylase initiates this mechanism by converting L-phenylalanine (L-phe) into L-tyrosine (L-tyr) [387, 388]. During this reaction, a hydroxyl group (OH) attaches to the phenyl ring of the L-phe thus resulting in a phenol containing tyrosine amino acid. The two aromatic amino acids, L-phe and L-tyr, are structurally similar except an extra OH in the latter compound (fig. 5.1).

120

Hydroxyl group effects on aromatic molecules… Chapter 5

Fig. 5.1: Biochemical transformation of aromatic amino acids into dopamine.

L-tyr is often considered as a non-essential amino acid, as it can be produced from an essential amino acid L-phe. The L-tyr amino acid can be further converted into an organic compound, dopamine, which acts as a neurotransmitter controlling the signal transactions between the nerve cells in the brain[389]. However, L-tyr amino acid is not directly biosynthesised into dopamine, but via an intermediate product known as levodopa or L-dopa or 3,4-dihydroxyl-L-phenylalanine[383, 384]. A non-heme iron enzyme, tyrosine hydroxylase, (fig. 5.1) performs an enzymatic reaction that converts L-tyr into L-dopa[390]. This enzyme uses molecular oxygen to form an additional hydroxyl group to the phenol moiety of tyrosine, thereby synthesising L-dopa with a catechol side chain[390]. Catechol is a term that is used to represent a phenyl ring with two hydroxyl substituents.

L-dopa is not only a precursor of dopamine, but a popular catecholamine neurotransmitter drug that is widely used for the treatment of Parkinson’s and other neurodegenerative disorders[384, 391-394]. L-dopa has the potency to cross the blood-brain barrier, whereas neither its antecedent amino acid, L-tyr, nor its descendant organic compound, dopamine, has this ability. As soon as L-dopa crosses the blood-brain barrier and enters the central nervous system, it is converted into dopamine by the DOPA decarboxylase enzyme (fig.5.2). Moreover, L-dopa is also useful in dual-targeting the monoamine oxidase enzyme[395] and the adenosine receptors at the same time[394]. All these potencies together make L-dopa an attractive chemical compound in the fight against Parkinson’s disease8,15-18. Swedish scientist, Arvid Carlsson was the first person to demonstrate that L-dopa can be successfully administered to reduce the intensities of Parkinson’s symptoms in animals[396, 397]. He received a Nobel Prize in 2000, for his invention[398]. Subsequently the 2001 Nobel Prize in chemistry was also related to L-dopa[399]. William S. Knowles received this award for his research on hydrogenation reactions catalysed by chirality, where most of his examples were based on the synthesis of L-dopa[399].

121

Hydroxyl group effects on aromatic molecules… Chapter 5

Fig. 5.2: Schematic representation of biochemical transport of L-dopa from blood into central nervous system after crossing through the blood-brain barriers.

L-dopa and its amino acid precursors, L-phe and L-tyr, share a similar structure except the OH group substituents in their phenyl rings. Conceptually, L-tyr can be represented as ‘L-phe + OH’, while L-dopa can be described as ‘L-tyr + OH’. However such small structural changes lead to significant physico-chemical differences among these molecules. Table 5.1 summarizes the discussed differences between the three molecules, L- phe, L-tyr and L-dopa.

Table. 5.1: Differences between L-phe, L-tyr and L-dopa. L-phe L-tyr L-dopa

Phenyl side chain Phenol side chain Catechol side chain (Phenyl + Alanine) (L-phe + OH) (L-tyr + OH ) larger numbers of larger numbers of Single conformer in the conformers in the gas conformers in the gas phase gas phase phase Drug molecule for Essential amino acids Non-essential amino acids Parkinson's diseases Can cross the blood-brain Biosynthesis of other Biosynthesis of L-dopa barrier and synthesise amino acids dopamine Strongly hydrophobic Less hydrophobic Even less hydrophobic

122

Hydroxyl group effects on aromatic molecules… Chapter 5

Amino acids are known to exist in a great number of low energy conformers, and hence L-phe and L-tyr have multiple conformers. In contrast, L-dopa with the same side chain as L-tyr but with an additional OH group is realized to have only a single observable conformation in the gas phase[383, 400]. The reason behind such a dramatic conformational reduction in L-dopa is not clear and the intra-molecular H-bonds are suspected as a dominating factor[400]. Another very interesting difference between these molecules is that, L-dopa serves as a drug molecule that can even cross the blood-brain barrier, while L-phe and L-tyr remain as basic building blocks of proteins. Does the addition of just one OH group to its side chain give L-dopa such valued potentials that are not available with its precursors? If so, what other impact can the OH substituent cause to its structure that differentiates L-dopa from the amino acid precursors? This chapter studies the role of hydroxyl group in the structure and properties of the aromatic molecules, L-phe, L-tyr and L-dopa, by probing their electronic structures.

A number of experimental[83, 336-338, 341, 345, 346, 348, 349, 363, 401-407] and theoretical studies[54, 56, 144, 334, 408-411 ] on the electronic structures of L-phe and L-tyr have been reported in the literature. Most of these studies focus on the conformational effects of aromatic amino acids. For instance, Prince and co-workers recently studied the different conformers of aromatic amino acids using the synchrotron sourced soft XPS techniques[336, 345]. Several other studies reported the near edge x-ray absorption fine structure (NEXAFS) spectra of aromatic amino acids[336, 345,0 , 35 351, 412-415]. However the electronic structure studies on L-dopa, especially in comparison with its precursors, are very limited. It was only very recently[383, 400] (in 2011) shown by a combined laser desorption supersonic jet laser spectroscopy and quantum chemical study that L-dopa has only a single conformer in the gas phase.

This chapter takes the advantage of different DFT models to comparatively probe the electronic structures and properties of L-dopa together with its amino acid precursors, L-phe and L-tyr, in the gas phase. It aims at studying the impacts of OH group substituents in the electronic properties of these aromatic molecules such as their molecular geometries, ionization spectra, charge re-distributions and aromaticity properties.

123

Hydroxyl group effects on aromatic molecules… Chapter 5

5.2. Computational details

Geometry optimization of the ground state stable structures of L-phe, L-tyr and L- dopa are performed using the B3LYP/6-311G** model. The optimized structures and their nomenclature are given in fig. 5.3. The global minimum conformers of L-phe[334] and L- tyr[411] along with the single gas phase conformer of L-dopa[383] are employed in this study. Further single point calculations at the optimized structures are carried out using different quantum chemical models such as LB94/et-pVQZ[143] for core shell and SAOP/et- pVQZ[303, 304 ] along with OVGF/ TZVP[302] for valence space. In addition to the core and valence space information, the aromaticity properties of the model molecules are also verified based on the nucleus independent chemical shift (NICS) method[416]. It employs a recently devised aromaticity indicator, NICS-rate index[417] that is obtained by NMR calculations performed using the B3LYP/6-311G** model.

Fig. 5.3: Optimized structures of L-phe, L-tyr and L-dopa.

5.3. Geometrical properties

Our previous chapter describes excellent agreements between the B3LYP model calculated geometries of L-phe and the experiments. Table 5.2 compares the calculated geometries of L-tyr and L-dopa against the measurements[407, 418] available in the literature. Bond lengths and bond angles of L-tyr and L-dopa calculated using the B3LYP/6- 311G** model agree remarkably well with the X-ray crystal data[407, 418]. All the bond

124

Hydroxyl group effects on aromatic molecules… Chapter 5

lengths in both the molecules agree within < 0.01 Å against the experiments. It is important to note that the structures of L-tyr and L-dopa in the measurements are the zwitterions forms,

as a result the C(1)=O(1) and C(1)=O(2) are almost at equal distances. Nevertheless the current work is based on the neutral structures of L-tyr and L-dopa, which in turn increases the bond

lengths of C(1)=O(2) due to hydrogen substituents. Although most of the bond angles given in

the table agree very well, a few of the them such as ∠O(1)=C(1)-O(2) and ∠N-C()-C() show larger differences. The discrepancy between the calculation and the experiment for the ∠N-

C()-C() angle is 5.23 in L-tyr and 5.36 in L-dopa.

Table. 5.2: Comparison of the calcualted and experimental geometries of L-tyr and L-dopa.

Parameters L-tyr L-dopa

This Work* Exp[407]  This Work* Exp[418] 

C(1)=O(1)/Å 1.20 1.24 0.04 1.20 1.25 0.05

C(1)-O(2)/Å 1.34 1.26 0.08 1.34 1.25 0.09

C(1)-C()/Å 1.55 1.53 0.02 1.55 1.54 0.01

C()-N/Å 1.47 1.49 0.02 1.47 1.46 0.01

C()-C()/Å 1.55 1.54 0.01 1.55 1.53 0.02

C()-C() /Å 1.51 1.51 0.00 1.51 1.51 0.00

C(4)-O(3)/Å 1.37 1.37 0.00 1.37 1.37 0.00

C(3)-O(4)/Å 1.36 1.36

R6 /Å 8.38 8.36 0.02 8.38 8.37 -0.01

O(1)=C(1)-O(2)/⁰ 123.23 126.40 3.17 123.18 126.10 2.92

C(1)-C()-N/⁰ 108.96 109.70 0.74 108.87 110.00 1.14

C(1)-C()-C()/⁰ 111.81 111.10 0.71 111.82 109.80 2.02

N-C()-C()/⁰ 116.03 110.80 5.23 115.96 110.60 5.36

C()-C()-C()/⁰ 114.09 114.50 0.41 113.98 114.00 0.02

C(3)-C(4)-C(5)/⁰ 119.54 120.20 0.66 118.97 120.30 1.33 *B3LYP/6-311G**

Discrepancies in bond angles and dihedral angles between the gas phase and crystal phase indeed exist, as gas phase and liquid phase structures of the same compound are not identical. The structures in the crystal are very tightly packed and remain in rigid environments that are dominated by other structures, which is not the case in the gas phase. Hence larger differences in the calculated bond angles of L-tyr and L-dopa by the current model are justified. The dihedral angles of L-tyr and L-dopa are not reported in their

125

Hydroxyl group effects on aromatic molecules… Chapter 5

corresponding experiments[407, 418] and therefore are not given in the table.

Table 5.3 compares the selected geometrical properties of the aromatic molecules, L- phe, L-tyr and L-dopa in the gas phase. The OH substituents to L-tyr and L-dopa only affect their geometries locally, most of the parameters are very similar to those in L-phe, as the aromatic ring serves as a buffer to resist changes[54, 56, 144]. For example, the bond lengths

in the three aromatic molecules are almost identical (i.e., (bond lengths) = 0). Inclusions of the

OH groups directly to the phenyl ring do not seem to even affect the ring perimeters (R6) of the molecules, where the values are unchanged. Other geometries such as bond angles and

dihedrals show very small changes. The ∠C(3)-C(4)-C(5) is the most affected bond angle in the

model molecules that can be related to the effects of OH substituents at the C(3) and C(4)

atoms in L-dopa and L-tyr. The differences in the ∠C(3)-C(4)-C(5) between L-phe and L-tyr is

0.3, but 0.57  between L-tyr and L-dopa. The most affected dihedral angle is ∠C(1)-C()-C()-

C(), which changes approximately 1 from L-phe (-73.89) to L-dopa (-72.36). The geometric parameters corresponding to the phenyl ring in the aromatic molecules are given in the appendix A. V. It is identified that the bond angles in the phenyl ring of the aromatic molecules change as large as 0.5◦ to 0.7◦ upon hydroxyl group substitutions.

While the geometries do not change very much, however, other properties such as dipole moments, electronic special extents () and rotational constants change considerably. Indeed the dipole moments (in Debye) display a decreasing trend from L-phe to L-dopa, ie., 4.88 (L-phe) > 3.73 (L-tyr) > 2.86 (L-dopa).It is worthy a note that the dipole moments of L-phe and L-tyr may be dependent on their conformers studied in this chapter. In the structure of L-phe, the strong electronegative groups (i.e., O and N atoms) in the amino acid moiety concentrate on one side of the molecule that result in a larger dipole moment. However, attaching a more electronegative OH groups to the phenyl ring in L-tyr and L-dopa balances the charge distributions in the molecules and as a result their dipole moments are reduced. Similar trends are also seen in the rotational constant values in A, B and C directions. For instance, the rotational constants in A (in GHZ) are 1.69, 1.55 and 1.21 in L- phe, L-tyr and L-dopa respectively. Conversely the values display an increasing trend as, 2163.72 a.u. (L-phe) < 2751.02 a.u. (L-tyr) < 3075.92 a.u. (L-dopa). This reflects the changes in the molecular sizes of the molecules due to the extra OH substituents.

126

Hydroxyl group effects on aromatic molecules… Chapter 5

Table. 5.3: Comparison of the geometric parameters of L-phe, L-tyr and L-dopa calculated using the B3LYP/6-311G** model.

Parameters L-phe L-tyr L-dopa

C(1)=O(1)/Å 1.20 1.20 1.20

C(1)-O(2)/Å 1.34 1.34 1.34

C(1)-C()/Å 1.55 1.55 1.55

C()-N/Å 1.47 1.47 1.47

C()-C()/Å 1.55 1.55 1.55

C()-C() /Å 1.51 1.51 1.51

C(4)-O(3)/Å 1.37 1.37

C(3)-O(4)/Å 1.36

R6 /Å 8.37 8.38 8.38

O(1)=C(1)-O(2)/⁰ 123.29 123.23 123.18

C(1)-C()-N/⁰ 109.02 108.96 108.87

C(1)-C()-C()/⁰ 111.84 111.81 111.82

N-C()-C()/⁰ 116.30 116.03 115.96

C()-C()-C()/⁰ 114.01 114.09 113.98

C(3)-C(4)-C(5)/⁰ 119.57 119.54 118.97

O(1)=C(1)-O(2)-H/⁰ 177.81 177.79 177.40

O(1)=C(1)-C()-N/⁰ -165.72 -165.75 -165.47

O(2)-C(1)-C()-N/⁰ 16.59 16.40 16.40

O(1)=C(1)-C()-C()/⁰ -35.71 -36.19 -36.04

O(2)-C(1)-C()-C()/⁰ 146.60 146.11 145.83

C(1)-C()-C()-C()/⁰ -73.89 -73.36 -72.36

N-C()-C()-C()/⁰ 52.25 52.52 53.25

μ (Debye) 4.88 3.73 2.86 /a.u. 2163.72 2751.02 3075.92

Rotational Constants A /GHZ 1.69 1.55 1.21 B /GHZ 0.62 0.45 0.42 C /GHZ 0.55 0.41 0.36 *R6 is the perimeter of the phenyl ring in the aromatic molecules in this work.

5.4. Hydrogen bond network

Intra-molecular H-bonds play significant roles in molecular architectures and properties. Changes in the intra-molecular H-bonds between the carboxyl and amino groups are likely to cause several low energy conformations in amino acids[83, 334, 339, 340, 377, 403, 411]. Further a number of studies on L-phe have also indicated that the H-bonds are responsible for the interactions in its side chain[56, 334, 340]. Table 5.4 presents all the

127

Hydroxyl group effects on aromatic molecules… Chapter 5

possible H-bonds in L-phe, L-tyr and L-dopa that are ≤ 2.80 Å cut-off[56, 334, 411]. In L- …. phe, there are four major H-bonds, three within the aliphatic component such as O(1) H-C(), …. ….. …. N H-O(2) and O(1) H-C(), and one between the carbonyl and phenyl ring, i.e., O(1) H-C(2). Nevertheless introduction of an OH group in L-tyr causes a few additional H-bonds from the …. …. OH group on its phenol side chain such as O(3) H-C(5) and O(3) H-C(3). Similarly L-dopa, …. …. when compared to L-phe, has three additional H-bonds including O(3) H-C(5), O(4) H-O(3) …. and O(4) H-C(2) that are formed because of the OH substituents on its catechol side chain. These additional bonds in L-dopa tend to affect its other H-bonds, especially those arising …. from the carbonyl group. For instance, the O(1) H-C() bond in L-phe and L-tyr is 2.61 Å, which becomes stronger in L-dopa with a distance of 2.54 Å. This indicates that additional OH groups in the phenyl ring tend to also affect the intra-molecular interactions between the side chains. Indeed strong intra-molecular H-bonds between amino and carboxyl groups in the amino acid moiety on one end and the H-bonds with the catechol OH groups on the other end, greatly reduces the flexibility in the structure of L-dopa. Lack of flexibility may attribute to the less number of conformers in L-dopa when compared to its amino acid precursors, L- phe and L-tyr.

Table. 5.4: Intra-molecular H-bonds of the aromatic molecules (Å).

Parameters L-phe L-tyr L-dopa …. N H-O(2) 1.89 1.88 1.89 …. O(1) H-C() 2.81 2.81 2.81 ….. O(1) H-C() 2.55 2.55 2.58 …. O(1) H-C(2) 2.61 2.59 2.54 O ….H-C 2.57 2.60 (3) (5) O ….H-C 2.69 (3) (3) O ….H-O 2.13 (4) (3) O ….H-C 2.71 (4) (2)

5.5. Inner shell changes

Inner shell chemical shifts are very sensitive to local chemical environments of molecules and could serve as effective indicators that reflect the impact of side chain

128

Hydroxyl group effects on aromatic molecules… Chapter 5

substitutions and intra-molecular H-bonds in the model molecules[419]. Table 5.5 compares the inner shell vertical IPs of the aromatic amino acids, L-phe, L-tyr and L-dopa, calculated using the LB94/et-pVQZ model against their respective available experimental IPs[336]. As described in earlier chapters, the LB94 model can correlate excellently with the experiments, after a global energy shift is applied to the simulations. In the present study, the LB94 model

accurately calculates the C 1s IPs for both L-phe and L-tyr, where their IP(EXP-Theory) values

are < 1 eV, except the C(1) atom whose IP(EXP-Theory) difference is > 1 eV. It is realized that most of the currently available quantum chemical models, including the DFT methods, are not able to accurately calculate the core IP of the C(=O) atom, as it includes more electron correlations. For example, the C(=O) 1s energies of 2-azetodine were also overestimated by the LB94 and E-KS methods when compared to the experimental measurements[353].

Table. 5.5: Inner shell vertical IPs of L-phe, L-tyr and L-dopa calculated using the LB94/et-pVQZ model along with the available experimental data (eV). L-phe L-tyr L-dopa

This This work Exp[336]  This work Exp[336]  work C 293.05 294.85 1.80 293.04 294.80 1.76 293.13 (1) C  291.50 291.90 0.40 291.47 291.52 ( ) C  290.07 290.06 290.12 ( ) C  (ring) 290.03 290.30 0.27 289.93 290.20 0.27 290.03 ( C (ring) 289.55 290.30 0.75 289.68 290.20 0.52 289.75 (2) C (ring) 289.63 290.30 0.67 289.77 290.20 0.43 291.45 (3) C (ring) 289.65 290.30 0.65 291.40 291.85 0.45 291.32 (4) C (ring) 289.71 290.30 0.59 289.67 290.20 0.53 289.68 (5) C (ring) 289.73 290.30 0.57 289.81 290.20 0.39 289.51 (6)

N 403.72 405.70 1.98 403.70 405.65 1.95 403.74

O 534.39 538.05 3.66 534.38 538.08 3.70 534.48 (1) O 535.88 539.87 3.99 535.85 535.95 (2) O (ring) 535.87 539.27 3.40 535.69 (3) O(3)(ring) 536.12

However the energy differences in the molecules become larger for the N 1s and O 1s

atoms with 2 eV < IP(EXP-Theory) < 4 eV. But such large energy differences can be compensated again by applying global energy shifts to our simulations, as shown in our

129

Hydroxyl group effects on aromatic molecules… Chapter 5

previous works[56, 64, 67, 322, 323, 353]. Fig. 5.4(a) and fig. 5.4(b) compare the simulated and experimental[336] C 1s and O1s spectra of L-tyr respectively. The C 1s spectrum simulated with a FWHM of 0.47 eV is globally shifted to the higher energy side by 0.45 eV to match the observed peak at ca. 290.00 eV.

Fig. 5.4: Comparison of the theoretical and experimental C 1s (a) and O 1s (b) spectra of L-tyr. The theoretical spectrum is simulated with an FWHM of 0.47 eV and shifted by 0.45 eV (C 1s) and 3.52 eV (O 1s) to match the experiment.

130

Hydroxyl group effects on aromatic molecules… Chapter 5

The theoretical spectra reproduce excellently the ratio of peaks, their intensities and widths in the measured spectra. For instance, the theoretical C 1s spectra (fig. 5.4(a)) of L-tyr in this work, produces three peaks that are consistently decreasing in their intensities when moving from the lower energy to higher energy, as also seen in the measured C 1s spectrum. Indeed, the energy gap between the most intensive C 1s peak and the middle peak is in remarkable agreement with the measured spectra, while the energy gap between the middle

peak and C(1) 1s peak shows some discrepancies, as our model overestimates the C(1) 1s energy by ca. 1.76 eV. However, the two peaks in the O 1s spectra simulated with a FWHM of 0.97 eV (fig. 5.4(b)) agree well with the experiment, after making a global shift of 3.52 eV.

The C 1s spectra of L-phe, L-try and L-dopa are shown in fig. 5.5 and their corresponding IPs along with those of N 1s and O 1s are given in table 5.5. The C 1s spectra of all the three aromatic molecules simulated with a FWHM of 0.4 eV display three major peaks that are labelled as A, B and C.

Fig. 5.5: Comparisons of the C 1s spectra of L-phe, L-tyr and L-dopa simulated with an FWHM of 0.4 eV using the LB94/et-pVQZ model.

131

Hydroxyl group effects on aromatic molecules… Chapter 5

The peaks ‘A’ in the higher energy region (ca. 293 eV) in fig. 5.5 are assigned to the

contributions from the C(1)(=O) atoms in the model molecules, which are very similar to the aliphatic amino acids[57, 85, 320, 420] and other aromatic molecules[56, 144, 336]. Whereas, the most intense ‘C’ peaks in the lower energy side (ca. 289.5 eV) of the spectra denote the strong asymmetric phenyl characters in the model molecules that are adapted from

the reduced high symmetric states in benzene (D6h). Moreover, small extended shoulder in

the C peak of the spectra (ca. 290 eV) is due to the C() and C() atoms that connect the amino acid moiety with the phenyl moiety. The C 1s IPs in the ‘C’ peaks of L-tyr and L-dopa display different orders and energy gaps, when compared to those in L-phe, which could be

related to their OH substituents. For instance, the C(4) atom in L-tyr is attached with the O(3)-

H group that in turn increases its energy by 1.75 eV against the IP of C(4) in L-phe, thereby

shifting the IP position of C(4) into the peak ‘B’ with C() in L-tyr. Similarly, the C(3) and C(4)

atoms in L-dopa are attached with the O(4)-H and O(3)-H groups, respectively, and hence their

IPs move into the ‘B’ peak with the C() atom in L-dopa. As a result, the middle peak ‘B’ in

L-phe that is assigned to the single C() atom, whereas in L-tyr and L-dopa, peak B consists of two and three C 1s contributions, respectively. Therefore, the direct impact of the OH group additions are reflected by the growth of the middle peak ‘B’ and shrink of the peak ‘C’ in the C 1s spectra of the aromatic molecules. Their A:B:C ratio is given by 1:1:7 in L-phe, 1:2:6 in L-tyr and 1:3:5 in L-dopa.

Similar effects are also evidenced in the O 1s spectra of the molecules (fig. 5.6). Fig. 5.6 present the O 1s spectra of the model molecules that present two peaks that are clearly separated by energy gaps, where the peaks in the lower energy side (ca. IP=534.5 eV) of the

spectra are dominated by the O(1)(=C) atoms in the molecules. This along with the C(1) 1s peak (in fig. 5.5) indicate that the C=O bonding is so strong that the energies of C and O atoms in C=O are very different from any other atoms in the molecules. However the other O 1s peaks in the higher energy sides at ca. 536 eV in the molecules are different from one another based on their side chain modifications, serving as their signatures.

For example, in L-phe, this peak is assigned with only the O(1) atom in its carbonyl group. But in L-tyr, this peak receives the contributions from both the OH groups in the

molecule such as the one from the carbonyl (i.e., O(1)) and the other from its phenol side

chain (i.e., O(3)). Hence the intensity of this peak is also larger than that of L-phe.

132

Hydroxyl group effects on aromatic molecules… Chapter 5

Interestingly, both the oxygen atoms in the OH groups of L-tyr are very close in their IP energies, although one OH group is on the amino acid moiety and the other is in the phenyl ring. The higher energy peak in the O 1s spectrum of L-dopa is not only intense, but also broader than those from L-phe and L-tyr. The peak of L-dopa includes contributions from the

three hydoxyl oxygen atoms in L-dopa such as O(2), O(3) and O(4), which are split by energy

gaps of 0.43 eV (IP (O(3)-O(4)) and 0.26 eV (IP (O(3)-O(2)) respectively. This indicates that

addition of an hydroxyl group at C(3) (meta) of the phenyl ring, significantly impact the electronic structure of L-dopa. As a result, L-dopa will behave very differently from L-phe and L-tyr, making L-dopa not an amino acid.

Fig. 5.6: Comparisons of the O 1s spectra of L-phe, L-tyr and L-dopa simulated with an FWHM of 0.4 eV using the LB94/et-pVQZ model.

Inner shell chemical shifts indicate that the C 1s and the O 1s regions of the molecules are significantly affected by the OH group additions and indeed the effects are more pronounced on the electronic structures than their geometries. Nevertheless, the N 1s of L- phe, L-tyr and L-dopa remain unaffected with an energy value of 403.7 eV, as they are in a similar chemical environment in all the three aromatic molecules, L-phe, L-tyr and L-dopa.

133

Hydroxyl group effects on aromatic molecules… Chapter 5

5.6. Hirshfeld charge distribution

Fig. 5.7 presents the Hirshfeld charge distributions of L-phe, L-try and L-dopa calculated using the LB94/et-pVQZ model. Both the charges and the molecular structures of the aromatic molecules are labelled by colour schemes. Charges of all the atoms in the amino

acid moiety (HOOC-CH(-CH2)-NH2) of the model molecules are almost unaffected, except

the O(1) atoms. Very small changes on O(1) can be attributed to the H-bond strengths (see table 5.4) of the molecules. The Hirshfeld charge schemes in these counterparts of the molecules correlate very well with those of the aliphatic amino acids[57, 144], i.e. all C sites which connect with more electronegative atoms, such as N and O remain positively charged, while all the C atoms connecting to hydrogen (H) atoms, together with the N and O sites

remain negatively charged. As expected, the most positive charge concentrate on the C(1)

atom, while the most negative charge lies with the O(1) atom.

Fig. 5.7: Hirshfeld charges of L-phe, L-tyr and L-dopa given as heat map representation.

Unlike the atoms in the amino acid moiety, the Hirshfeld charges of the carbon sites in the phenyl ring display many differences in the molecules. Indeed the carbon sites in the phenyl ring can be both positive and negative depending on the side chain substitutions. As discussed in the previous chapter, all the carbon sites in the phenyl ring of L-phe are

negatively charged except the C() site that connects to the amino acid moiety. Whereas, the

134

Hydroxyl group effects on aromatic molecules… Chapter 5

modified C sites in L-tyr (i.e., C(4)) and L-dopa (i.e., C(3) and C(4)) are majorly affected by the OH substituents. Charges of these modified carbon sites become strongly positive in order to balance the negative charges from the attached OH groups. The electronegativity differences among the modified C sites and the O in the OH groups lead to  and  electron transfers from C to O, thereby making these C sites strongly positive. Moreover the OH substituents not only affect the sites of substituents but, the entire phenyl ring, with strong impacts on the neighbouring atoms due to the aromatic system. For instance, while the OH substituent in L-

tyr takes place on the C(4) atom, the negative charges on its neighbouring sites – C(3) and C(5) –

become stronger, when the other atoms, C(2), C(6) and C(), show small changes. Based on the Hammett substitution constants, the OH group on the meta position ( =0.12) serves as an electron withdrawing group, while in the para position (=-0.37) remains strongly electron donating[421, 422]. As a result, the OH substitution on the para position in L-tyr tends to

increase the positive charge in the C(4) atom and increase the negative charge in the

neighbouring C(3) and C(5) sites of L-tyr, when compared to L-phe. Similar effects are seen in the L-dopa. These changes reflect in the charge dependent colour schemes of L-phe, L-tyr and L-dopa given in fig. 5.7

5.7. Effects in valence space

Valence space information can directly link to a number of chemical properties and reactivity of the molecules. Investigation of the valence spectra of L-phe, L-try and L-dopa is important to unveil the effects of OH substituents in their chemical properties. As indicated in earlier chapters, the DFT- based SAOP model and the Green’s function based OVGF model are efficient in calculating the accurate valence IPs for amino acids[52, 54, 57, 144] and other bio-molecules[64, 327, 423, 424]. The former model is able to simulate the entire valence IPs, while the latter is applicable to simulate the outer valence IPs with required accuracy. Moreover it is ascertained that the SAOP model slightly over-estimates the outermost valence IPs, especially the IP of the HOMOs, where OVGF can calculate with better accuracy. Hence combining both the SAOP and OVGF models can achieve good correlations with the experiments.

Fig. 5.8 compares the recently measured high resolution valence XPS[345] (middle panel) of L-tyr against the theoretical spectra simulated using the SAOP/et-pVQZ (bottom

135

Hydroxyl group effects on aromatic molecules… Chapter 5

panel) and the OVGF/6-311G** (top panel) models. The spectroscopic pole strengths of the outer valence IPs calculated by OVGF model are ≥ 0.85 eV, demonstrating that the single particle approximation used in this model holds appropriate for the study. The theoretical spectra obtained with a FWHM of 0.5 eV are globally shifted by 1.40 eV and 0.38 eV in SAOP and OVGF, respectively, thereby aligning the HOMO peaks at 8.47 eV in the experimental and theoretical spectra. After the energy shifts, both simulated spectra provide excellent agreement with the measurements, such as their shapes and intensities. In fact both OVGF and SAOP models reproduce the outer valence measurements nicely, while the latter model continues to take care of the inner valence region (ie., > 20 eV). This clearly reflects in the fig. 5.8.

Fig. 5.8: Comparison of the valance spectra of L-tyr simulated (FWHM = 0.5 eV) using the SAOP/et- pVQZ and OVGF/6-311G** against the experimental spectrum.

Fig. 5.9 compares the valence XPS of the three aromatic molecules simulated using the SAOP/et-pVQZ model with a FWHM of 0.4 eV. The corresponding valence vertical IPs of the molecules calculated using the SAOP/et-pVQZ and OVGF/TZVP models are given in table 5.6. Despite valence space being more complex than the core space because of its delocalized nature, the valence spectra of the three molecules clearly display similarities and

136

Hydroxyl group effects on aromatic molecules… Chapter 5

differences in their spectral features.

Table. 5.6: Valence vertical IPs of L-phe, L-tyr and L-dopa calculated using the SAOP/et-pVQZ and OVGF/6-311G** models (eV). OVGF polestrengths are given in parantheses. L-phe L-tyr L-dopa MOs SAOP OVGF MOs SAOP OVGF EXP[345] MOs SAOP OVGF 44 -10.49 -8.96(0.90) 48 -9.88 -8.14(0.90) 8.47 52 -9.60 -7.85(0.90) 43 -10.70 -9.18(0.90) 47 -10.61 -9.26(0.89) 9.40 51 -10.27 -8.71(0.90) 42 -10.78 -9.75(0.90) 46 -10.69 -9.71(0.90) 9.60 50 -10.64 -9.76(0.90) 41 -11.66 -10.91(0.90) 45 -11.55 -10.86(0.90) 10.80 49 -11.55 -10.88(0.90) 40 -12.17 -11.12(0.90) 44 -12.09 -11.13(0.90) 48 -12.12 -11.22(0.90) 39 -12.46 -11.79(0.90) 43 -12.40 -11.60(0.85) 11.30 47 -12.33 -11.34(0.85) 38 -12.50 -11.92(0.90) 42 -12.56 -11.88(0.90) 46 -12.52 -11.91(0.90) 37 -13.13 -12.19(0.83) 41 -12.72 -12.02(0.90) 45 -12.75 -12.21(0.90) 36 -13.31 -12.81(0.90) 40 -13.05 -12.70(0.90) 44 -13.06 -12.73(0.90) 35 -13.80 -13.36(0.90) 39 -13.65 -13.38(0.90) 43 -13.57 -13.34(0.90) 34 -14.10 -13.74(0.89) 38 -14.08 -13.68(0.90) 42 -13.62 -13.33(0.89) 33 -14.50 -14.17(0.89) 37 -14.39 -14.23(0.90) 41 -14.06 -13.70(0.89) 32 -14.54 -14.37(0.88) 36 -14.52 -14.24(0.87) 40 -14.21 -14.18(0.90) 31 -14.91 -14.74(0.89) 35 -14.57 -14.39(0.88) 39 -14.58 -14.47(0.89) 30 -15.40 -15.24(0.89) 34 -14.80 -14.67(0.89) 38 -14.74 -14.72(0.89) 29 -15.49 -15.42(0.88) 33 -14.98 -15.04(0.89) 37 -14.82 -14.79(0.89) 28 -15.69 -15.58(0.88) 32 -15.38 -15.22(0.89) 36 -15.04 -14.84(0.85) 27 -16.03 -15.95(0.89) 31 -15.57 -15.67(0.88) 35 -15.23 -15.14(0.89) 26 -16.81 -16.87(0.89) 30 -16.02 -16.02(0.88) 34 -15.43 -15.41(0.89) 25 -16.94 -17.03(0.86) 29 -16.34 -16.29(0.88) 33 -16.03 -16.04(0.89) 24 -17.88 -18.08(0.89) 28 -16.73 -16.88(0.89) 32 -16.18 -16.28(0.88) 23 -18.75 27 -17.78 -18.07(0.89) 31 -16.68 -16.89(0.89) 22 -19.19 26 -17.83 -18.13(0.86) 30 -17.46 -17.82(0.88) 21 -19.34 25 -18.80 18.65 29 -17.84 -18.19(0.87) 20 -20.75 24 -19.14 28 -17.99 -18.29(0.88) 19 -22.51 23 -19.76 19.40 27 -18.78 18 -22.88 22 -20.67 26 -19.14 17 -23.90 21 -22.38 21.15 25 -20.16 16 -25.77 20 -22.94 22.75 24 -20.70 15 -27.38 19 -23.75 24.30 23 -22.37 14 -30.05 18 -25.68 25.90 22 -22.78 13 -32.11 17 -27.26 28.00 21 -23.73 16 -29.94 20 -25.61 15 -31.26 32.40 19 -27.25 14 -32.00 34.00 18 -29.97 17 -30.86 16 -31.58 15 -32.03

137

Hydroxyl group effects on aromatic molecules… Chapter 5

Combining the spectral properties along with their respective molecular orbital densities can reveal the impacts of OH substituents in the valence space of the molecules. Previous chapters show that for aliphatic amino acids and L-phe, the innermost valence peaks > 26 eV have 2s contributions of the carboxyl and amino group. For instance, in L-phe, the molecular orbitals (MOs) 13a and 14a are dominated by the 2s characters from its carboxyl oxygen atoms, while the third innermost orbital, 15a has mostly 2s features from the amino group[54, 144]. L-tyr and L-dopa agree very well with L-phe, where their MO pairs, 14a/16a in L-tyr and 15a/18a in L-dopa are dominant by the carboxyl group, while the 17a (in L-tyr) and 19a (in L-dopa) have amino contributions (refer to fig. 5.8). However at ca. 31 eV, one extra peak in L-tyr (i.e., 15a) and two in L-dopa (ie., 16a and 17a) can be seen that are dominant by the extra OH groups in their respective side chains (see fig. 5.8). Therefore these peaks do not exist in L-phe.

Fig. 5.9: Valence ionization spectra (FWHM = 0.5 eV) of L-phe, L-tyr and L-dopa based on the SAOP/et-pVQZ calculations. The MOs of the corresponding peaks are marked in the spectra and a few orbital diagrams in the inner valence region are also shown.

138

Hydroxyl group effects on aromatic molecules… Chapter 5

The peaks within 26 eV > IPs > 18 eV in the valence spectra of the three molecules align excellently. The MOs in this region are mostly populated by the 2s features from their carbon atoms, except the peaks at ca. 18 eV – 20 eV in L-tyr and L-dopa that include small p type dominance from their extra OH groups. Nevertheless the mid-valence region of 18 eV > IPs > 12 eV involving MOs 24a-40a in L-phe, 26a-45a in L-tyr and 28a-48a in L-dopa display clear differences. Verifying the orbital densities of these MOs (provided in appendix, A.VI - A.VIII) indicate that they are related to the strong 2p type interactions mainly driven by the side chain changes within the molecules. Aliphatic amino acids also display similar effects in this valence region[57, 320, 345 , 420]. Therefore this mid-valence region (i.e., 26 eV > IPs > 18 eV) that display complex delocalized picture, serve as the fingerprint valence region of amino acids. Orbital diagrams of all the MOs from L-phe, L-tyr and L-dopa are provided in appendix (A.VI – IA.VII ).

Frontier orbitals in the valence spectra of biomolecules are highly regarded in chemistry. The interactions between these outermost valence MOs, especially between HOMO and LUMO, are crucial in the activities of biomolecules. The outermost valence region < 12 eV in the three aromatic molecules involves four MOs that is, HOMO to HOMO- 3. L-phe presents a peak at ca.10.5 eV that combines HOMO to HOMO-2 orbitals (i.e., 44a – 42a), while the HOMO-3 (i.e., 41a) peak appears separately at 11.7 eV[54, 144]. Although the HOMO-3 peaks in L-tyr (45a) and L-dopa (49a) correlate very well with the orbital 41a of L-phe, their outermost peaks from 48a – 46a in L-tyr and 52a – 50a in L-dopa are differently related, as shown in fig. 5.10. Substitutions of the OH groups in L-tyr and L-dopa lead to frontier orbital rearrangements in the valence regions. Moreover the HOMO IPs in the molecules display a decreasing trend as 10.49 eV (L-phe) > 9.88 eV (L-tyr) > 9.60 eV (L-

dopa), where the IPs(HOMO) changes as 0.6 eV between L-phe and L-tyr, while 0.3 eV between L-tyr and L-dopa, again shown in fig. 5.10.

A general trend in the frontier orbitals of the three molecules is that the orbitals are shifted upwards, when a hydroxyl group is added into L-phe to form L-tyr and to L-tyr to form L-dopa. The LUMOs and the HOMO-3 orbitals in the three aromatic molecules are very similar. The LUMOs receive the phenyl dominance and the HOMO-3 orbitals have contributions from the amino acid moiety, whereas in the other frontier MOs such as HOMO, HOMO-1 and HOMO-2, the hydroxyl group plays an important role. For example, the

139

Hydroxyl group effects on aromatic molecules… Chapter 5

HOMO (44a) in L-phe (at 10.5 eV) combines the ‘1’ contribution from its phenyl ring and

‘nN’ interactions from the aliphatic fragment[54, 144, 345]. However, attaching an OH group in the phenyl ring of L-tyr, the orbitals of L-tyr become the phenyl dominant HOMO (48a) and amino acid moiety dominant HOMO-1 (47a) in L-tyr, that are separated by an energy gap of 0.73 eV. The HOMO-1 in L-phe (i.e., 43a) and HOMO-2 in L-tyr (i.e., 46a) have nearly

identical ‘2’ characters and therefore can be correlated with each other.

Fig. 5.10: Frontier orbital correlation diagrams of L-phe, L-tyr and L-dopa along with the HOMO- LUMO energy gaps of the molecules.

Introducing the second OH group in the phenyl ring of L-dopa tend to mix the three

outermost MOs, where the HOMO (52a) and HOMO-1 (51a) orbitals have strong ‘1’ and

‘2’ characters respectively and the HOMO-2 receives contributions from amino acid moiety.

As a result the 0.73 eV of energy gap between the ‘1’ and ‘nN’ orbitals in L-tyr further expands as 1.04 eV in L-dopa. Therefore the frontier orbital correlations can be made as 44a (HOMO) in L-phe  48a (HOMO)/47a (HOMO-1) in L-tyr  52a (HOMO)/50a (HOMO-2) in L-dopa. Similarly, 43a, 46a and 51a in L-phe, L-tyr and L-dopa respectively can be correlated with each other. Moreover the HOMO-LUMO energy gap display a decreasing trend as 4.81 eV (L-phe) > 4.22 eV (L-tyr) > 4.10 eV (L-dopa). Apparent reduction in the

140

Hydroxyl group effects on aromatic molecules… Chapter 5

HOMO-LUMO gap is a clear feature that differentiates L-dopa as a potential drug when compared to its amino acid precursors.

5.8. Aromaticity properties

Aromaticity that results from the cyclic conjugation remains an attractive property in chemistry[425]. The frontier MO analyses show noticeable changes in the ‘’ characters of L-phe, L-tyr and L-dopa. These changes indicate that the aromatic characters of the three molecules may be affected as the results of the OH groups in their phenyl rings. Further analyses are warranted to understand the changes in their aromaticity properties. Aromaticity plays an important role in many biological systems for their chemical reactions and stability factors of different molecules[425-427]. However, it is not a direct observable quantity. As a result, several efforts are consistently made to evaluate the aromaticity features of the molecules[428-430], using properties including structures[431, 432], energy[433, 434], electronic features[435, 436] and magnetic[416, 437 ] properties.

One of the widely accepted theoretical methods is the nucleus independent chemical shift (NICS) index[416, 437, 438]. It is the negative of the magnetic shielding in a given point of the molecule[416, 437, 438]. In this approach, a dummy atom is usually placed at the ring center (approach known as ‘NICS(0)’) or 1 Å above the ring (i.e., NICS(1)) and the magnetic shielding is predicted. The negative value of the magnetic shielding in the dummy atom is the NICS value. The NICS approach has several advantages and has been employed extensively to evaluate the aromaticity and anti-aromaticity properties of a number of organic and inorganic compounds[417, 438-445]. Recently Noorizadeh and Dardab[417] have introduced the NICS-rate index that is based on the disparity of NICS index at varying distances from the ring. This NICS-rate index is used in this study to evaluate the aromaticity features of L-phe, L-tyr and L-dopa.

In this work, the NICS values are initially calculated by placing the dummy atom at the ring center (i.e.,0 Å) and moving the dummy atom at a varying interval of 0.2 Å until 4 Å above the ring is reached. The NMR calculations based on the gauge-including atomic orbital (GIAO) method[446, 447] and B3LYP/6-311G** model are performed using the G09 package. Then, the NICS rate index is calculated using the magnetic shielding at each

141

Hydroxyl group effects on aromatic molecules… Chapter 5

position of the dummy atom as[417],

(5.1)

Here r = 0.2. Finally the NICS-rate obtained for the three aromatic molecules are plotted against the distance (r) and the curve obtained is the NICS-rate curve.

Fig. 5.11 compares the calculated NICS-rate curves of L-phe, L-tyr and L-dopa. The minima and the maxima of the NICS-rate curves of the three molecules correlate very well, where the minima is reached at a distance of 0.6 Å above the ring and the maxima at 1.6 Å above the ring. Among the three molecules, L-phe displays a large minimum and a large maximum. Indeed the NICS-rate curves display some interesting trends. In the distance range < 1.4 Å, at every 0.2 Å, the NICS-rate values of L-dopa are the highest, followed by those of L-tyr and then by L-phe, i.e., L-dopa > L-tyr > L-phe. In the distance range of 1.4 Å – 3.4 Å, the NICS-rate trend becomes L-dopa ≃ L-tyr < L-phe and finally at any distance > 3.4 Å above ring center, the NICS values of all the three molecules are approximately equal (i.e., L-dopa ≃ L-tyr ≃ L-phe).

Fig. 5.11: NICS-rate spectra of L-phe, L-tyr and L-dopa calcualted as a function of distance (Å).

Higher the NICS-rate value, more the aromaticity of the molecule at the given distance. Based on the trends described above, different distance ranges provide different information, for instance, at 1 Å above the ring, the NICS-rate value of L-phe is a negative

142

Hydroxyl group effects on aromatic molecules… Chapter 5

value, while those of L-tyr and L-dopa are positive values. This indicates that L-phe displays more anti-aromatic character at this point. However at 2 Å above the ring center, L-phe displays more aromatic feature than those of other molecules. To clarify this, a dimensionless parameter known as NICS-rate ratio (NRR) is calculated as,

(5.2)

However since the NICS-rate curves show maxima and minima in all the molecules, it is essential to also include the effects of the cross-section in the NICS-rate curves (ie., NICS-

rate()) while calculating the NRR. Selvam et al[448] in their earlier work on benzene and fluorinated benzene considered the cross-section of the NICS rate curve as

. This model is suitable for the fluorinated benzene, where the aromaticity curve exhibit only maxima, without a minima[36]. However, the NICS-rate curves of the molecules in this study present a well-defined minimima (see fig. 5.10), therefore, the effects of negative cross-section should also be considered. In this study, we propose an improved NRR indicator, that also takes into account the cross-sections of the

NICS-rate curve, both maxima and minima, known as NRR(). NRR() can be described as,

(5.3)

The calculated NICS(0), NRR and NRR() of all the three molecules are tabulated in the

table 5.7. The NRR() values display an increasing trend from L-phe to L-dopa, i.e., 3.12 (L- phe) < 4.51 (L-tyr) < 5.99 (L-dopa), which agrees with the NICS(0) and NRR values of the molecules (refer table 5.7 for values). All the measures together indicate that the aromaticity properties strongly increase with the increasing numbers of OH group substituents. According to the Keto-enol tautomerism of the phenol group, the keto tautomer (ie.,

C5H5C=O) is known to be unstable that results in the loss of aromaticity in the phenol. On the other hand, substituting the C=O with C-OH in the enol tautomer increases the aromatic - conjugation energies in the phenol[449]. Therefore, an increase in the number of hydroxy groups in the phenyl ring leads to the extended aromatic conjugation. Our results agree very well with that of the keto-enol tautomerism concept, which finds that the L-dopa is more

143

Hydroxyl group effects on aromatic molecules… Chapter 5

aromatic than its amino acid precursors, L-tyr and L-phe. Increased aromaticity features may also be one of the reasons that enhance the physico-chemical properties of L-dopa as a drug.

Table. 5.7: The calculated NICS(0), NRR and NRR() values for L-phe, L-tyr and L-dopa. L-phe L-tyr L-dopa NICS(0) -8.67 -9.49 -10.45 NICS-Rate 9.94 8.60 8.71 (MAX) r(MAX) 1.60 1.60 1.60 NICS-Rate (MIN) -8.49 -5.08 -3.87 r(MIN) 0.60 0.60 0.60 NRR 1.17 1.69 2.25

NRR() 3.12 4.51 5.99

5.9. Summary

The effects of the OH addition to the electronic structures and properties of L-phe, L- tyr and L-dopa are probed in the gas phase. Most of their geometric parameters remain little changed by the phenyl side chain modifications, while the molecular properties such as dipole moments, rotational constants and the molecular size of the molecules change considerably. In fact the dipole moments and rotational constants show a decreasing trend as L-phe > L-tyr > L-dopa, when the OH groups are added on the para and meta positions of the phenyl ring to balance the charge.

The core and valence spectra of the three molecules are calculated quantum mechanically. The calculated spectra of L-tyr are in good agreement with recent synchrotron sourced measurements[336, 345]. The inner shell spectra highlight the significant local impacts on the chemical environment of the molecules. The N 1s XPS spectra of the aromatic molecules remain unaffected to changes in the phenyl ring, while the major differences are seen in the C 1s and O 1s spectra. The apparent changes in the C 1s spectra mostly happen in the lower energy side < 292 eV, while the changes in the O 1s spectra concentrate on the higher energy, at ca. 536 eV. The atoms that directly involve in the modifications, i.e.,

C(3)/O(3) in L-tyr and C(3)/O(3) and C(4)/O(4) in L-dopa, are mostly affected. Indeed the modified C sites in the C 1s display large energy shifts of ~ 1.8 eV. Changes in the core IPs

144

Hydroxyl group effects on aromatic molecules… Chapter 5

of the atoms impact the shapes and intensities of the spectral peaks. The atomic Hirshfeld charges reflect the changes in the inner shells of the aromatic molecules, where the modified carbon sites gain strong positive charges that in turn affect the entire phenyl ring in the molecules.

In the valence XPS, significant changes concentrate on three different regions such as inner valence region ca. 31 eV, in the mid-valence region 12 eV – 16 eV and in the outermost valence region < 12 eV. In the innermost valence space at ca. 31 eV, there are few additional peaks in L-tyr (one peak) and L-dopa (two peaks) that are caused due to the 2s contributions from their additional OH groups and therefore are not seen in L-phe. The mid-valence region, i.e., 12 eV < IP < 16 eV display the most spectral differences among the molecules, which agree with the aliphatic and other amino acids discussed in the previous chapters. Therefore this mid-valence space remains the fingerprint valence region for amino acids, where significant interactions take place. Finally, frontier orbitals are rearranged in the three

aromatic molecules. The combined ‘1’ (phenyl) and ‘nN’ (amino acid moiety) contributions

in the HOMO (ie., 44a) of L-phe is split into ‘1’ dominant HOMO (48a) and ‘nN’ dominant HOMO-1 orbital (47a) in L-tyr with an energy gap of 0.73 eV. Similar orbital rearrangements are also seen in L-dopa, IP of 1.04 eV. Moreover the HOMO-LUMO gap decreases with the increasing OH groups as, L-phe > L-tyr> L-dopa.

The electronic level changes in the three aromatic molecules are also seen to impact their aromaticity. The simulated NICS-rate curves of the molecules display some interesting trends such as, ‘L-dopa > L-tyr > L-phe’ at the distance < 1.4 Å, ‘L-dopa ≃ L-tyr < L-phe’ at the distance rage of 1.4 Å – 3.4 Å and ‘L-dopa ≃ L-tyr ≃ L-phe’ at any point > 3.4 Å above the ring. Different NICS indices, NICS(0) and NRR values are calculated. We further

develop an extended NRR indicator, NRR(), that also takes into account the cross-sections of

the NICS-rate curves. The calculated NICS(0), NRR and NRR() correlate with each other to indicate that the aromaticity of the molecules increase with the increasing OH groups, ie., L- phe < L-tyr < L-dopa. Therefore, the molecular properties such as dipole moments, rotational constants, Hirshfeld charges, HOMO-LUMO gap and aromaticity features clearly differentiate the natural amino acids, L-phe and L-tyr against the L-dopa drug compound.

145

Hydroxyl group effects on aromatic molecules… Chapter 5

CHAPTER

6 Interactions of micro-solvated Cu2+-phenylalanine complex

6.1. Amino acids in molecular environment

Previous chapters provided insight on the intrinsic properties of few selected amino acids. Such inherent features of amino acids are useful to some extent, however, amino acids are not isolated in the ‘bio-molecular’ world. Amino acids interact with their neighboring molecules in order to perform diverse biological roles. Therefore, studying the inter- molecular interactions of amino acids using molecular dynamics approach provides an additional dimension of information. Amino acids, metal ions and water are among the most important chemical partners in the biology of life, as the interactions between these components play a crucial role in the structures and functions of proteins in aqueous solution[450]. Metal interactions with amino acids are significant for electron transfer or deprotonation reactions[94, 95, 451] within molecular systems, and also affect protein folding/unfolding and aggregation processes. Studying the interactions of solvent molecules with amino acids and metal ions is thus vital for understanding the hydration of metalloproteins and the role of water molecules in biological systems[450]. In this chapter, we combine density functional theory (DFT) calculations and Car-Parrinello molecular dynamics (CPMD) simulations to investigate the structures, interactions and dynamic properties of phenylalanine-copper (II) ([Phe-Cu]2+) complexes under micro-solvated environment in room

146

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

temperature. Microsolvation is a process in which a target molecule (known as ‘solute’) interacts with a limited number of water molecules or other solvents (usually upto 10) to form an active complex.

Although bulk solvents are generally considered when studying bio-molecular systems, micro-solvation effects can potentially change molecular properties[452, 453]. Micro- solvation effects play significant roles in diverse molecular systems, ranging from amino acids to large enzymes. In larger systems, micro-hydration is important for a number of biological reactions. For example, the reactions in the drug binding sites of metallo-enzymes mostly engage a catalytic metal center surrounded by amino acid residues and a few water molecules to trigger specific functions. Histone deacetylases (one of the cancer targets) possess a catalytic water molecule bound to the active site zinc ion that is essential for its enzymatic activities[454]. Similarly, the neutral (NT)-to-zwitterion (ZW) switching of amino acids also takes place in the presence of micro-solvents[96]. Rodziewicz and Doltsinis[96] showed in their ab initio molecular dynamics (AIMD) investigation on micro-hydrated phenylalanine (Phe) structures that more than three water molecules are required to fully stabilize the ZW forms of Phe in water. However, the ZW form of Phe, when complexed with the aluminum metal ion, was found to be stabilized with only two water molecules[455]. Therefore the number of water molecules required for conformational transitions of amino acids remains an open question.

A recent work by Otto et al[456] published in Nature Chemistry, unveils the influence of micro-solvents in the reaction dynamics of hydroxyl ions with iodomethane using a combined crossed-beam imaging and cold source of micro-solvated reactant experiment and ab-initio simulations. The study finds that distinct reactions take place in different degrees of solvation, while the co-linear substitutions happen under mono-solvation[456]. Micro- hydration studies can therefore provide information on the subtle solute-solvent interactions that can be useful for a molecular level understanding of complex biological systems.

The significance of transition metal-ligand complexes has been recognized for decades in areas such as catalysis, drug research, atmospheric chemistry and biochemistry[457]. A number of previous experimental and theoretical studies on the interactions between several transition metal ions have been reported[42, 94, 457-468],[429, 43, 45 , 462, 464-469]. Copper is one of the most prominent transition elements in biological systems[43, 462] with

147

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

an occurrence of 80-120 mg in a normal human body[462]. Copper is also a component of several enzymes such as indophenoloxidases, cytochrome c oxidase, Cu/Zn superoxide dismutase and tyrosinase[470].

The Cu2+ ion has nine d electrons (d9) in its valence shell with the electronic 6 3 configuration of (tg) (eg) . It often displays an octahedral coordination in crystals and aqueous solution. The four equatorial and two axial bonds have been well described by a number of experiments[462, 471-480] including X-ray absorption[474], NMR[477], X-ray absorption near-edge structure (XANES)[473, 475, 476], neutron diffraction[462, 471, 480] and extended X-ray absorption fine structure (EXAFS)[472, 473, 475, 478, 479]. The two axial bonds of Cu2+ within aqueous solution are generally elongated due to the Jahn-Teller distortion effects[478, 479]. In contrast, Pasquarello et al[462] have deduced from their combined neutron diffraction and AIMD simulations that Cu2+ ions favor fivefold trigonal bipyramidal configurations in aqueous solution. This was later supported by an X-ray absorption spectroscopy based analysis[473].

Solid Cu2+ complexes with aliphatic amines, pyridines[481] or ammonia[482-484] have been recognized to show a variety of different coordinations: square planar (fourfold)[485, 486], square pyramidal (fivefold)[487, 488] and distorted square bipyramidal (sixfold)[489, 490]. Rulisek and Vondrask[491], who have exploited several metalloproteins and transition metal complexes from the Protein Data Bank (PDB) and the Cambridge Structure Database (CSD), believed that the Cu2+ metal ion mostly prefer a square planar structure, although a few complexes also exhibit trigonal bipyramidal geometries. Hence the coordination of Cu2+ within different molecular environments may not be the same. Previous studies discussed the coordination of Cu2+ metal ions with aromatic amino acids. For example, Rimola et al[43, 467] used DFT based theoretical calculations to study the interactions of Cu+ and aromatic amino acids. Remko et al[42] studied Cu2+ – aromatic amino acid complexes in the gas phase and in the presence of five water molecules using DFT calculations. It was found that the copper-aromatic amino acid system exists as an NT conformer and the copper-aromatic amino acid-5 water system exists as a ZW conformer.

The effects of stepwise micro-hydration of Cu2+–Phe complexes (i.e., [Phe-Cu]2+) are limited in the literature. Micro-hydration studies of [Phe-Cu]2+ can reveal important

148

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

information that are usually hidden in the fully solvated systems. This thesis uses DFT calculations combined with CPMD simulations to investigate the effects of Cu2+ binding on

the Phe structures under micro-solvated environments, (H2O)n=1-4. The geometry

optimizations are performed to identify the lowest energy structures for all [Phe-Cu(H2O)n=1- 2+ 4] complexes, followed by CPMD simulations to probe structural changes at room temperature and to reveal the coordination preferences of Cu2+ ion for different micro- hydrated states. Finally, frontier orbitals of the lowest energy complexes are analyzed using the information obtained from our CPMD simulation and DFT calculations.

6.2. Computational details

The Cu2+–Phe complexes are built using the configurations from Larrucea et al[455] on Al3+–Phe complexes as reference structures. Initially we chose five different Phe structures,

i.e., Phe1, Phe2 and Phe3 (which are ground state NT configurations with –COOH/–NH2 − + 2+ groups), Phe4 and Phe5 (which are ZW conformers with –COO–NH / 3 groups). The Cu ion is the placed at different binding positions around the parent structures. Both the NT and ZW complexes contain structures with the Cu2+ metal ion binding to carboxyl-amino groups as well as the phenyl ring. The formation of [Phe-Cu]2+ complexes (i.e., n=0) can be described by the following process,

(a)

Next, water molecules are added one by one to the [Phe-Cu]2+ complex around the Cu2+ ion to form the micro-hydrated systems that include n=1, 2, 3 and 4 complexes according to the hydration reaction,

(b)

where i=0, 1, 2 and 3. As a result, a total of 35 structures as initial structures of the complexes are produced.

Geometry optimizations of all 35 initial structures are performed using two different DFT functionals, B3LYP and BLYP61,62. The calculations with B3LYP together with the 6-

149

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

311++G(d,p) basis set are carried out using the latest version of the Gaussian09 (G09)[254] computational chemistry package. For the open shell Cu2+, we employ the Watcher-Hay all electron basis set[492, 493] that is able to accurately describe the energies of the transition metal ions[42]. The hybrid B3LYP model provides accurate geometries as found in our previous studies involving a number of amino acids[54, 309, 494] and has also been proven to be efficient for many transition metal-containing systems[457, 495-497 ].

The CPMD[100, 498] simulations are performed for the lowest energy micro-hydrated

systems (i.e., l(n=1-4)). A periodic cubic box of 18Å in length is employed. The default Hockney Poisson solvers and the local density approximation (LDA) are employed, whereas the valence electrons are treated explicitly using the BLYP functional. The exchange functional is given by Becke[139] and the correlation energy expression by Lee, Yang and Parr[140] in the BLYP functional. The core electrons are described using the norm- conserving Troullier and Martins[499] pseudopotentials. For Cu2+, the non-linear core correction (NLCC)[500] pseudopotential is used to improve the description of its core energy. This NLCC correction has been recognized to be more appropriate for the transition metal ions[466]. The Kohn-Sham orbitals are expanded with a plane wave cut-off of 90 Ry. An electronic fictitious mass of 600 amu with a time step of 5.0 a.u. is used. Under these conditions, the molecular dynamics simulations of the lowest energy structures of [Phe-Cu (n=1-4)] 2+ (here n=1-4 denote the number of water molecules) are performed for more than 12 ps. The temperature is maintained at 300 K throughout the simulation using a Nosé- Hoover[501, 502] thermostat. Further we employ a geometric criteria of 2.80 Å cut-off for determining the inter- and intra-molecular H-bonds in our systems[334]. All the parameters described above are chosen following a series of convergence tests (given in appendix, A.IX) carried out using the CPMD package.

6.3. Stable structures of Phe-Cu2+ complexes

Hydration effects on bare Cu2+ metal ions are studied to validate the theoretical models used in this work. The optimized structures along with their energies are given in fig. 6.1. The bond distances in these structures agree very well with those obtained in previous studies[465, 503] as well as with experimental data[462, 472, 479] (given in table 6.1). Specifically, the averaged Cu-O and O-O distances in the first and second hydration shells

150

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

are in excellent agreement with the experimental values. For example, the averaged Cu-O distances measured by various experiments are between 1.94-2.15 Å, in fairly good agreement with the present calculations of 1.85-2.11 Å. Similarly, the O-O distances in our calculations are 2.69 Å and 2.71 Å, which are in reasonable agreement to the experimental values (2.73-2.80 Å)[504, 505]. As a result, both theoretical models, B3LYP/6-311++G(d,p) and BLYP/MT, are able to satisfactorily represent Cu2+ complexes.

Fig.6.1. Geometry optimized structures of Cu2+ ion in micro-solvated environment (n=1-6) using B3LYP/6-311++G(d,p) method.

Table 6.1. Comparison of the average Cu-O and O(1st hydration shell)-O(2nd hydration shell) distances from this work with that of the available experimental[462, 472, 479] and other works.

Average (Cu-O) O(H.S.1)-O(H.S.2) Cu_H2O Coordination Exp[504, System B3LYPa BLYPb Burda et alc Exp* B3LYPa BLYPb 505] Cu_1W 1 1.89 1.97 1.86

Cu_2W 2 1.85 1.92 1.85

Cu_3W 3 1.91 2.00 1.90

Cu_4W 4 1.97 2.24 1.96 1.94-2.15

Cu_5W_5Cn 5 2.11 2.11 2.03

Cu_5W_4Cn 4(1)# 1.96 2.04 1.96 1.94-2.15 2.69 2.70 2.73-2.80 Cu_6W_6Cn 6 2.11 2.20 2.03

Cu_6W_4Cn 4(2)# 1.96 2.04 1.96 1.94-2.15 2.71 2.72 2.73-2.80 aB3LYP/6-311++G** model using G09; bBLYP with MT pseudopotentials using CPMD; cRef 444. #Values given in the parentheses represents the number of water molecules in the second hydration shell. *Refs. [462, 472, 479].

151

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Optimized geometries of the [Phe-Cu (n=1-4)]2+ structures are shown in fig. 6.2. The first row gives the conformers of Phe in isolation, the next row presents the [Phe-Cu]2+ complexes (i.e. n=0) produced from the Phe conformers in the first row with Cu2+, whereas rows 3-6 display the hydrated [Phe-Cu]2+ complexes with waters (n=1-4). Harmonic vibrational frequency analyses indicate that these complexes are true minimum structures. Energies of the complexes are corrected for basis set superposition errors (BSSE), which are calculated using the counterpoise methods implemented in G09. The relative energies calculated using the B3LYP/6-311++G(d,p) model are given with the complexes in fig. 6.2.

There are five Phe conformers (fig. 6.2, first row) in isolation, which comprise three conformers in NT form (Phe1-Phe3) as well as two ZW conformers (Phe4-Phe5). Among the conformers, Phe3 has the lowest energy structure, in agreement with previous studies[54, 334]. As shown in the figure, the relative energies of the ZW structures in the gas phase, Phe4 and Phe5, are higher in energy, which are 16.5 kcal·mol-1 and 15.7 kcal·mol-1, respectively, than the NT Phe3 minimum structure. The second row in fig. 6.2 displays the [Phe-Cu]2+ complexes. Two different metallated complexes are obtained for the ZW structure Phe5, i.e., [Phe5-Cu]2+ and [Phe5a-Cu]2+. In the former ([Phe5-Cu]2+), the Cu2+ binds with the carboxyl moiety but in the latter the Cu2+ ion also binds other carbon atoms in the phenyl ring.

Table 6.2 compares the selected geometrical parameters of all the [Phe-Cu]2+ complexes calculated using the B3LYP (G09) and BLYP (CPMD) models, along with results from other studies[42]. The geometries of [Phe1-Cu]2+ and [Phe4-Cu]2+ show excellent agreement with those reported recently by Remko et al[42] (calculated using B3LYP/6- 311+G** model), although the models using slightly different basis sets. Comparison of the different [Phe-Cu]2+ complexes in this study show very different geometrical parameters. 2+ This is particularly the case when Cu is involved. For example, O(4)-Cu and O(3)-Cu bond

lengths, ∠C(2)-O(4)-Cu and ∠C(2)-O(3)-Cu bond angles and ∠C(1)-C(2)-O(4)-Cu and ∠C(1)-C(2)-

O(3)-Cu dihedral angles, display very large deviations among the complexes shown in table 6.2. See the footer in fig.6.2 for the atom numbering in the NT and ZW forms of Phe.

Moreover, the ring perimeters (R6) of the phenyl rings within the Phe structures listed in the

table 6.2 are all longer than the R6 of the isolated global minimum Phe of 8.37 Å[54]. As a 2+ result, upon Cu binding, the phenyl ring expands so that the R6 of the complexes are longer.

152

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Fig.6.2. The optimized structures of phenylalanine (Phe), [Phe-Cu]2+ and micro-hydrated [Phe-Cu(n=0- 4)]2+ structures along with their relative energies in kcal∙mol-1 obtained from B3LYP/6-311++G(d,p) calculations. Here n represents the number of water molecules in the system. The lowest energy structures (l) are indicated in boxes.

153

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Of all the [Phe-Cu]2+ complexes considered here (fig. 6.2) the most stable structure is 2+ 2+ 2+ 2+ the [Phe1-Cu] complex (i.e., l[Phe1-Cu] ). The Cu in the l[Phe1-Cu] forms a bidentate

coordination with the amino nitrogen (N) and the carboxyl oxygen (O(3)) atoms. It is realized that the electrostatic interactions and the repulsions from the strongly electronegative oxygen (especially carbonyl oxygen) and nitrogen atoms in Phe1 is more favorable for the Cu2+ metal ion. This agrees with previous studies[42, 43, 457, 467, 468, 506] showing that the transition metal ions mostly bind with the N and O atoms of aliphatic and aromatic molecules in order to present stable complex structures in the gas phase. The ZW complexes, [Phe4-Cu]2+ and [Phe5-Cu]2+ are found to be the next preferred complexes with only 4.6 kcal·mol-1 and 4.5 -1 2+ kcal·mol higher in energy, respectively, than the most stable NT complex, l[Phe1-Cu] .

Addition of the first water molecule (n=1) into the [Phe-Cu]2+ complex does not lead to any considerable structural changes, as was also observed in other metal-aromatic amino acid systems[455,. 507] The same NT conformer is still preferred as the lowest energy 2+ complex upon mono-hydration (i.e., l[Phe1-Cu(n=1)] in the third row of fig.6.2), where the water molecule is attached to the NO-coordinated Cu2+ ion without any direct contact to Phe.

Introducing the second water molecule into the complex leads to the ZW structures being energetically preferred with respect to the NT structures. In the lowest energy (n=2) 2+ 2+ complex (i.e., l[Phe4-Cu(n=2)] ), the Cu metal ion interacts with both the carboxyl oxygen

atoms (O(3) and O(4)) of the Phe moiety and the two water molecules (fig. 6.2). One of the ZW 2+ -1 complexes, [Phe5a-Cu(n=2)] , is only 1.6 kcal·mol higher in energy than the l[Phe4- Cu(n=2)]2+ complex. The NT complexes are now higher in energy than the ZW ones, with [Phe1-Cu(n=2)]2+ being 5.1 kcal·mol-1 higher than the most stable one, while the other NT structures are approximately 10.9 kcal·mol-1 and 15.3 kcal·mol-1 higher in energy. This indicates that two water molecules are required to interconvert the energetic order of the NT and ZW configurations of the [Phe-Cu]2+ complexes in the micro-hydrated processes. Such a conversion from NT to ZW was experimentally confirmed[508] in [Val(Na)]+ in the presence of two water molecules, but no such experimental evidence is available for the [Phe- Cu]2+ systems. Earlier CPMD based investigation[96] reveals that the Phe structure, without any metal ions, require larger numbers of water molecules (n≥3), for stabilizing the ZW form of Phe, whereas the present study finds that two water molecules are sufficient to deliver the NT  ZW transformation. This indicates that the presence of Cu2+ may have changed the mechanism and reduced the complexity in the de-protonation process of Phe.

154

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Table 6.2: Selected geometrical parameters of the [Phe-Cu]2+ complexes optimized using the B3LYP/G09 and BLYP/CPMD methods.

Parameters$ [Phe1-Cu]2+ (NT) [Phe2-Cu]2+ (NT) [Phe3-Cu]2+ (NT) [Phe4-Cu]2+ (ZW) [Phe5-Cu]2+ (ZW) [Phe5a-Cu]2+ (ZW) Other Other B3LYPa BLYPb B3LYPa BLYPb B3LYPa BLYPb B3LYPa BLYPb B3LYPa BLYPb B3LYPa BLYPb Work# Work#

C(1)-C(2)/Å 1.55 1.55 1.55 1.54 1.55 1.57 1.59 1.57 1.57 1.57 1.57 1.59 1.52 1.53

C(2)-O(3)/ Å 1.23 1.25 1.22 1.24 1.25 1.28 1.30 1.24 1.24 1.23 1.30 1.23 1.29 1.30

C(2)-O(4)/ Å 1.31 1.32 1.31 1.30 1.32 1.25 1.26 1.28 1.30 1.27 1.22 1.31 1.25 1.26

C(1)-N/ Å 1.49 1.53 1.49 1.50 1.51 1.44 1.44 1.52 1.55 1.52 1.52 1.54 1.52 1.54

O(4)-Cu/ Å 2.05 2.03 2.08 2.06 2.12 1.87 1.93 1.91 1.92 1.92 1.82 1.88 2.11 2.24 ______O(3)-Cu/ Å 2.98 2.90 3.02 2.01 2.06 N-Cu/ Å 2.08 2.07 2.11 2.03 2.09 ______o ∠C(1)-C(2)-O(3)/ 123.20 123.40 122.90 119.02 119.50 116.30 116.80 114.60 119.10 116.50 117.40 117.60 117.50 117.10 o ∠C(2)-C(1)-N/ 109.60 109.30 109.00 104.70 105.10 107.00 107.90 103.70 104.30 103.80 103.60 102.80 109.60 109.50 o ______∠C(1)-N-Cu/ 108.50 107.60 109.30 99.60 99.10 o ∠C(2)-O(4)-Cu/ 112.40 111.80 113.40 108.30 108.50 131.30 128.60 114.90 112.70 117.30 132.90 128.70 81.90 80.50 o ______∠C(2)-O(3)-Cu/ 65.80 68.90 65.30 85.10 86.90 o ______∠O(3)-Cu-N/ 83.60 85.70 81.60 80.20 79.20 o ______∠O(3)-Cu-O(4)/ 49.50 51.60 48.20 64.70 62.20 o ∠O(3)-C(2)-C(1)-N/ 12.00 11.70 15.20 -30.70 -32.70 2.40 -1.90 7.70 7.20 7.90 -7.30 -10.70 -45.10 -41.60 o ______∠C(2)-C(1)-N-Cu/ -16.80 -15.10 -20.30 50.60 51.50 o ∠C(1)-C(2)-O(4)-Cu/ 0.20 -1.20 -0.90 -7.60 -6.30 5.60 8.20 -177.00 -177.10 -177.00 22.40 22.80 134.50 131.80 o ______∠C(1)-C(2)-O(3)-Cu/ 177.60 177.60 176.60 -133.70 -129.80 ^ R6 /Å 8.48 8.48 8.49 8.53 8.51 8.54 8.48 8.48 8.50 8.54 8.48 8.52 $ Here C(1) represents the C(); C(2) represents carbonyl carbon (COO); O(3) and O(4) denote the carbonyl oxygen atoms. Refer to Fig.1 for atom numbering. aB3LYP/6-311++G(d,p) model using G09; bBLYP with MT pseudopotentials using CPMD; #Ref 42 (B3LYP/6-311+G(d,p)[42]. ^R6 is the perimeter of the Phenyl ring (The values in the table are to be compared with the R6 of Phe (8.37 Å)[54].

155

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Upon addition of the third water molecule, the ZW complex [Phe5-Cu(n=3)]2+ 2+ is again identified as the most stable structure, i.e. l[Phe5-Cu(n=3)] . In this complex, the Cu2+ ion shows a distorted square planar coordination including a carboxyl

oxygen, O(4), and the oxygen atoms of three water molecules. The energy for the second most stable structure, [Phe4-Cu(n=3)]2+, is only 1.9 kcal·mol-1 above the 2+ 2+ 2+ l[Phe5-Cu(n=3)] structure. In the [Phe4-Cu(n=3)] structure, Cu displays a penta- coordination, i.e. two carboxyl oxygens and three oxygens from the water molecules. 2+ 2+ Similarly, in the most stable four water complex (l[Phe5a-Cu(n=4)] ), Cu interacts 2+ with O(4) of Phe and oxygens of the four water molecules. The [Phe4-Cu(n=4)] complex again with a pent-coordinated Cu2+ metal ion is the second most stable -1 2+ structure and is only 0.4 kcal·mol higher in energy than l[Phe5a-Cu(n=4)] .

A few complexes were initially built with Cu2+-phenyl bonds in order to study the cation-π interactions. However, during geometry optimization the cation-π interactions in the complexes broke and evolved into structures, in which the Cu2+ moves away from the phenyl ring, to avoid such π interactions. Although a few optimized complexes such as [Phe5-Cu]2+, [Phe5a-Cu(n=1)]2+ and [Phe3-Cu(n=2)]2+ still hold the cation-π bonds, their relative energies are high. The Cu2+-phenyl bonds do not appear in the most stable structures. This is in contrast with a previous ab-initio study[96] in which, the water···π hydrogen bonds (H-bonds) are crucial for stabilizing the micro-hydrated Phe systems (without a Cu2+)[96]. It is noted that the hydrated complexes formed with the global minimum structure of Phe (i.e., Phe3) become the least stable complexes. This suggests that the inter-molecular forces of hydrated complexes can change the complexes considerably from their gas phase structures.

2+ 2+ Table 6.3 presents representative Cu -O(Phe) and Cu -O(wat) distances for the 2+ 2+ lowest energy complexes. The Cu -O distances in complexes l[Phe1-Cu] and 2+ l[Phe1-Cu(n=1)] can be compared directly, as both complexes are based on Phe1, 2+ with Cu ion interacting with the nitrogen atom and the carboxyl oxygen atom, O(4), 2+ 2+ of the Phe moiety (i.e., N-Cu -O(4)). In the absence of water (i.e. in l[Phe1-Cu] 2+ complex), the Cu cation binds tightly with O(4) of Phe. When one water molecule is 2+ 2+ introduced, the binding between Cu and the Phe loosens (in l[Phe1-Cu(n=1)] 2+ 2+ complex). For example, the distance between Cu and O(4) in l[Phe1-Cu] complex is given by 2.05 Å using the B3LYP/6-311++G(d,p) model, which increases to 2.09 Å

156

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

2+ in the l[Phe1-Cu(n=1)] complex upon addition of a water molecule. The increase of 2+ the Cu -O(4) (Phe) distance comes at the expense of the Cu-N binding. For example, 2+ the Cu-N distance in the l[Phe1-Cu] complex is given by 2.08 Å, but reduced to 2.01 2+ Å in the l[Phe1-Cu(n=1)] complex. Such shortening of the Cu-N distance has also 2+ been reported by Remko et al[42]. Nevertheless, the Cu -O(3) (Phe) distance is 2+ 2+ always larger than the Cu -O(4) distance. The calculated water-Cu distances in most

of the lowest energy complexes are within the experimentally observed range of 1.94 2+ – 2.15 Å[465, 466, 479, 503] for bare Cu (H2O)n (without Phe).

Table 6.3: Selected distances of Cu2+ in the lowest energy micro-hydrated complexes (in Å).

2+ $ $ $ $ l[Phe1-Cu] (n=1) (n=2) (n=3) (n=4) Parameters l l l l (NT) (NT) (ZW) (ZW) (ZW) B3LYPa B3LYPa B3LYPa B3LYPa B3LYPa (BLYPb) (BLYPb) (BLYPb) (BLYPb) (BLYPb) 2.02 3.28 3.23 Cu-O (3) (2.18) (3.33) (3.30) 2.05 2.09 1.97 1.91 1.95 Cu-O (4) (2.03) (2.12) (2.08) (2.02) (2.04) 2.08 2.01 Cu-N (2.07) (2.03) 1.94 1.98 2.02 2.06 Cu-O (wat1) (1.98) (2.06) (2.12) (2.15) 1.98 1.99 1.96 Cu-O (wat2) (2.09) (2.09) (2.07) 1.95 2.01 Cu-O (wat3) (2.04) (2.13) 2.31 Cu-O (wat4) (2.39) $ 2+ 2+ l(n=1) denotes l[Phe1-Cu(n=1)] ; l(n=2) denotes l[Phe4-Cu(n=2)] ; 2+ 2+ l(n=3) denotes l[Phe5-Cu(n=3)] ; l(n=4) denotes l[Phe5a-Cu(n=4)] . aB3LYP/6-31++G** model using G09; bBLYP with MT pseudopotentials using CPMD. The BLYP values are given in parentheses.

6.4. CPMD simulation

The CPMD simulations are performed at room temperature for ~12 ps. The lowest 2+ 2+ energy micro-hydrated structures, l[Phe1-Cu(n=1)] , l[Phe4-Cu(n=2)] , l[Phe5- 2+ 2+ Cu(n=3)] and l[Phe5a-Cu(n=4)] , are employed in the simulation. Structures obtained from the last snapshot of the 12ps CPMD simulations are given in fig. 6.3.

157

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Fig.6.3. Last snapshots of the micro-solvated [Phe-Cu]2+ structures from the 12 ps CPMD simulation.

Significant geometrical parameters of the initial snapshot and the final snapshot of the lowest energy complexes (from the 12ps CPMD simulation) are 2+ compared in table 6.4. The l[Phe1-Cu(n=1)] complex shown in fig. 6.3(a) remains stable in its NT configuration throughout the simulation. Only a few changes in its

bond lengths are observed during the MD simulation. In the initial structure of l[Phe1- 2+ Cu(n=1)] , the Cu-N bond (2.03 Å) is shorter than its Cu-O(4) bond (2.12 Å); whereas in the final structure, the situation is inversed. That is, the Cu-N bond becomes larger

than the Cu-O(4) bond in the final structure, in agreement with an earlier study[42]. It

is found that the coordination of Cu-O(wat1) is weakened slightly as the bond distance increases from 1.98 Å to 2.09 Å. This indicates that the inter-conversion between the 2+ NT and ZW transition of the l[Phe1-Cu(n=1)] complex is unlikely to occur spontaneously.

2+ 2+ The Cu ion in the initial structure of the l[Phe4-Cu(n=2)] complex displays a square-planar coordination, a pair to the carboxyl oxygen and the other pair to the water molecules. However, during the dynamics process, one of the Cu-O bonds such 2+ as the Cu-O(3) bond weakens and the Cu ion switches to a tridentate coordination (fig. 6.3(b)). This is reflected in the changes in bond lengths (table 6.4), that is, the

Cu-O(3) distance of the initial structure is given by 2.18 Å; whereas in the final

structure, this distance increases to 3.19 Å. Consequently, the O(3) atom is replaced by a H-bond with a water molecule, in addition to its pre-existing intra-molecular H-

158

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

bond with the amino group (i.e., (N)H···O(3)). Thus both intra- and inter-molecular 2+ interactions together form a (N)H···O(3)···H2O–Cu chemical bond network within 2+ the l[Phe4-Cu(n=2)] complex. Here, a water molecule serves as a bridge between 2+ 2+ the carbonyl oxygen (O(3)) atom and the Cu atom (O(3)···H2O–Cu ). Similar H- bonded bridges between the Cu2+ cation and the curcumin anion have been reported

very recently by Addicoat et al[509]. Weakening of the Cu-O(3) bond leads to a 2+ stronger coordination between the Cu and O(4) atom reflected by the reduction of its distance from 2.08 Å to 1.94 Å.

Table 6.4: Selected geometrical parameters of the initial and final (shaded in grey color) snapshots of the micro-hydrated complexes in the CPMD simulations.

2+ 2+ 2+ 2+ l[Phe1-Cu(n=1)] l[Phe4-Cu(n=2)] l[Phe5-Cu(n=3)] l[Phe5a-Cu(n=4)] Geometrical (NT) (ZW) (ZW) (ZW) Parameters Initial Final Initial Final Initial Final Initial Final

d[Cu-O(3)]/Å 2.18 3.19 3.33 3.53 3.30 3.57

d[Cu-O(4)]/Å 2.12 2.03 2.08 1.94 2.02 2.45 2.04 1.98 d[Cu-N)]/Å 2.03 2.14

d[Cu-O(wat1)]/Å 1.98 2.09 2.06 2.07 2.12 2.25 2.15 2.37

d[Cu-O(wat2)]/Å 2.09 2.00 2.09 2.11 2.07 2.08

d[Cu-O(wat3)]/Å 2.04 1.98 2.13 2.16

d[Cu-O(wat4)]/Å 2.39 4.30

Hydrogen bonds … ^ ^ (N)H O(3)*/Å 3.31 3.28 2.19 2.03 1.91 1.61 1.90 1.98 … $ $ H(wat3) O(3)*/Å 4.29 4.84 3.36 2.79 1.72 1.74 1.67 2.73 … H(wat1) O(wat2)*/Å ------3.32 2.47 … H(wat2) O(wat4)*/Å ------3.67 1.54

Cu2+ coordination 3 3 4 3 4 4 5 4 *Hydrogen bonds within the complexes. Hydrogen bonds within 2.80 Å are underlined. $ … 2+ H(wat2) O(3) bond in l[Phe4-Cu(n=2)] complex. ^ 2+ The H-bond in the initial structure of l[Phe5a-Cu(n=4)] complex is between O(3) and Hx in NH3, whereas in the final structure, the H-bond is between O(3) and HY in NH3.

2+ In the l[Phe5-Cu(n=3)] complex, the structural rearrangement is similar to the

bi-hydrated complex, l(n=2), as shown in fig. 6.3(c). However, binding of an additional water molecule (wat3) to Cu2+ changes the coordination of Cu2+ to square-

planar. This additional water molecule is H-bonded to the O(3) atom of Phe, thereby 2+ 2+ preventing it from directly binding to the Cu ion. Hence the l[Phe5-Cu(n=3)] 2+ complex also adapts the (N)H···O(3)···H2O–Cu chemical bond network after

159

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

interactions with three water molecules, which is found by the molecular dynamics simulations.

2+ The restructuring of the l[Phe5a-Cu(n=4)] complex is revealed by the

CPMD dynamic simulation, as shown in fig. 6.3(d). The structure of l[Phe5a- Cu(n=4)]2+ complex is more complicated than other hydrated systems with less water molecules involved, as discussed above. The Cu2+ ion in the system starts to exhibit a distorted penta-coordinated trigonal-bipyramidal configuration, where the Cu2+ atom is coordinated with all the four oxygen atoms from the water molecules and one 2+ carboxyl oxygen, O(4). The water molecules around Cu can cause increased steric hindrance. A very recent study by Otto et al[456] again indicates the important role of the steric characteristics of the water molecules in micro-solvated reactions. This can be the case at room temperature as revealed by the CPMD dynamics. The rearrangement of the complex to create a favorable steric environment in the system is found. As a result, one of the water molecules (wat4), which is initially connected to the Cu2+ cation, is moved to the second coordination shell thereby reducing the ‘steric attack’ towards Cu2+ and resulting in a square-planar like complex.

Water migrations between different solvation shells may affect the inter- and

intra-molecular interactions of the complex. As shown in table 6.4, the initial l[Phe5a- 2+ Cu(n=4)] structure possesses only two types of H-bonds, (N)H···O(3) and

H(wat3)···O(3). Additional H-bonds between the water molecules, wat1, wat2 and wat4, are formed in the final complex after the dynamical process. Note that in the initial 2+ 2+ l[Phe5a-Cu(n=4)] structure, the (N)H···O(3) H-bond exists between the Cu ion and

the Hx atom in the NH3 group of ZW Phe, while in the final structure, the Hy atom in 2+ the NH3 forms H-bond with Cu . Such change in (N)H···O(3) bond is due to the

rotation of the NH3 group during the CPMD simulations. More H-bonds in the final 2+ configuration indicate that the l[Phe5a-Cu(n=4)] complex is stabilized by formation of the H-bonding network among water molecules in different solvation shells, Phe and Cu2+. Moreover, similar to the bi- and tri-hydrated systems, one of the water 2+ molecules remains as a coordination mediator between the O(3) atom and the Cu ion 2+ (see fig. 6.3(d)). The (N)H···O(3)···H2O–Cu H-bond chain, therefore, remains as an unique structural motif to stabilize the lowest energy [Phe-Cu]2+ complexes with more than two water molecules.

160

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Fig. 6.4 presents dynamical trajectories of the Cu-O distances (fig. 6.4(a)) and 2+ the H-bonds (fig. 6.4(b) and (c)) of the l[Phe5a-Cu(n=4)] complex for a period up to

12ps. Fig. 6.4(a) shows that the Cu-O distances of Cu-O(wat1), Cu-O(wat2), Cu-O(wat3)

and Cu-O(4) are stable with the variation between 2-2.5 Å. As shown in the structure of the complex given in fig. 6.4, the Cu2+ directly bonds with oxygen atoms presented in the four bonds. As a result, these Cu-O bonds are strong and dynamically stable in this period of time. Another dynamically stable but weaker Cu∙∙∙O distance shown in

fig. 6.4(a) is the Cu∙∙∙O(3) distance, which remains in the vicinity of 3.5 Å. This

Cu∙∙∙O(3) “bond” is stabilized by the formation of a H-bond with a hydrogen atom of 2+ the water molecule (wat3) that is directly bonded with the Cu ion (i.e., Cu-O(wat3)). 2+ This water molecule (wat3) “bridges” the O(3) atom of Phe with Cu and makes the squared-planar Cu2+ configuration of the complex possible without causing a

significant strain to the complex, as O(4) of Phe already bonds with the metal.

Fig. 6.4(a) displays the significant dynamical changes in the l[Phe5a- 2+ Cu(n=4)] complex. The Cu∙∙∙O(wat4) is the only dynamically unstable Cu∙∙∙O bond of 2+ the l[Phe5a-Cu(n=4)] complex. This Cu∙∙∙O(wat4) bond undergoes significant changes from a strong Cu-O bond with the distance of < 2.5 Å to a weak Cu∙∙∙O bond with the distance of ~4.5 Å. The dynamical change over happens at ~6 ps, as shown in fig. 6.4(a). The oxygen atom of this water molecule (wat4) in the complex is initially

bound with the metal, forming a strong bond of Cu-O(wat4). However, the metal ion prefers a squared-planar configuration after 6 ps, rather than the initial distorted

penta-coordinated trigonal-bipyramidal configuration. As a result, the Cu∙∙∙O(wat4)

distance starts to increase and stabilizes at ~4.5 Å. The O(wat4)···O(wat2) distance in the final structure is given by 2.60 Å, which is in good agreement with the experimental value of 2.73 Å for the oxygen-oxygen distance between the two water molecules in the first and second solvation shell respectively[479]. The significant increase in the

Cu∙∙∙O(wat4) distance, therefore, indicates the substantial changes in the configuration of the metal complex, which releases the fourth water molecule (wat4) to the second solvation shell of the Cu2+ ion.

161

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Fig.6.4. Trajectories of the (a) Cu-O distances and (b) inter-molecular H-O distances and (c) 2+ intra-molecular NH-O(3) distances of the l[Phe5a-Cu(n=4)] complex.

Fig. 6.4(b) and 6.4(c) show that the dynamics of the Cu-O bonds also affect 2+ the inter- and intra-molecular H-bonds of the l[Phe5a-Cu(n=4)] complex. It can be

seen that the inter-molecular H-bond such as H(wat3)···O(3) remains stable at < 2.5 Å throughout the simulation period thereby bridging between Phe and Cu2+. The intra-

molecular H-bond between the amino group and the carboxyl group (i.e. (N)H···O(3)) of Phe (fig. 6.4(c)) exists throughout the simulation. However, a switch over between

(N)Hx···O(3) and (N)Hy···O(3) at approximately 6.5 ps is observed. Here Hx and Hy represent two different hydrogen atoms in the amino group of the Phe ZW moiety.

162

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Fig. 6.4(c) clearly indicates the Hx···O(3) and Hy···O(3) change over at 6.5 ps, which is

simultaneous with the changes of the Cu∙∙∙O(wat4) distances as shown in fig. 6.4(a). The 2+ result of such structural changes in the l[Phe5a-Cu(n=4)] complex leads to the 2+ (N)H···O(3)···H2O–Cu H-bond chain, similar to the other hydrated complexes, 2+ 2+ l[Phe4-Cu(n=2)] and l[Phe5-Cu(n=3)] , discussed previously.

The other inter-molecular H-bond, H(wat2)···O(wat4), starts to form at almost the

same time of the Cu···O(3) and NH···O(3) changes in the complex, i.e., at approximately 6 ps, as shown in fig. 6.4(b). That is, at 6 ps when the penta- coordinated trigonal-bipyramidal configuration of the Cu2+ ion (which is the configuration of the bare Cu2+ metal ion under full solvation[462 , 466]) starts to take 2+ the squared planar configuration in the l[Phe5a-Cu(n=4)] complex with

phenylalanine. As a result, the Cu-O(wat4) bond becomes a weaker Cu∙∙∙O(wat4) interaction to allow the wat4 molecule to migrate towards the second solvation shell. In the presence of Phe, the Cu2+ prefers to confine itself to a distorted square planar coordination in order to reduce the steric hindrance and maintain its interactions with the Phe under micro-hydration.

Fig. 6.5 shows the Cu-O radial distribution functions (RDFs) for the hydrated [Phe-Cu]2+complexes calculated from the CPMD trajectories. Two distinctive peaks appear in the RDF spectra of the micro-hydrated complexes. An intensive peak at 2+ ~2.0 Å of the complexes is due to the strong Cu -O bonds, i.e., Cu-O(wat1), Cu-O(wat2),

Cu-O(wat3) and Cu-O(4), in agreement with our previous discussions. The less

intensive peak representing the Cu-O(3) distance occurs at ~3.5 Å in the complexes, 2+ except for the l[Phe1-Cu(n=1)] complex in which this less intensive peak locates at approximately 4.2 Å, instead. It is this peak, (i.e., the less intensive peak at 4.2 Å) which reflects the structural differences of the Phe moiety (NT or ZW) in the 2+ complexes. The Phe moiety in the l[Phe1-Cu(n=1)] complex exhibits an NT conformer (O=C-OH), whereas the Phe moiety in the other hydrated complexes exists as the ZW form (O=C-O-). As a result, the RDF spectra provide useful information to differentiate the Phe NT from the ZW form of the complexes.

163

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Fig.6.5. Radial distribution function spectra for the lowest energy micro-hydrated [Phe-Cu]2+ complexes calculated from the CPMD trajectories.

In addition, the RDF spectra further indicates that the O(3) atom in the NT Phe 2+ 2+ moiety of the l[Phe1-Cu(n=1)] complex is unlikely to directly interact with the Cu 2+ 2+ 2+ ion. The ZW Phe of the l[Phe4-Cu(n=2)] , l[Phe5-Cu(n=3)] and l[Phe5a-Cu(n=4)] 2+ complexes can be stabilized if the O(3) of Phe indirectly forms a network with Cu ion 2+ via a water bridge. Very small peaks of the l[Phe5a-Cu(n=4)] complex at larger distances (> 4 Å) are likely due to weak interactions caused by the wat4 molecule in the second solvation shell. Again, the RDF spectra of the complexes provide details to reveal that the [Phe-Cu]2+ complexes prefer to bond up to three water molecules directly due to steric environment of the Cu2+-Phe network. Any excess water molecules are unable to form sufficiently strong bonds directly with the Cu2+ ion and move to the subsequent solvation shell.

6.5. Molecular orbital analyses

Molecular orbital (MO) analyses are useful for interpreting the inter- and intra- molecular chemical reactivities[510]. Frontier orbitals i.e., the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) play a prominent role in most of the chemical reactions of molecules[511]. The HOMO of Cu2+ ion, due to its 3d9 electronic configuration, remains as a singly occupied orbital, which is often referred to as SOMO (singly occupied molecular orbital).

164

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

Table 6.5 reports the orbital energies of the frontier orbitals of the lowest energy micro-hydrated complexes from the CPMD simulations. As discussed above, upon binding with smaller numbers of water molecules, n≤2, the Cu2+ ion present a tridentate co-ordination, however, the structure of the complexes become square planer upon inclusion of more than two water molecules. The coordination of Cu2+ ion in the complexes also affects the energies in the corresponding frontier MOs. This is clearly seen in the table 6.5 where, the orbital energies of the frontier MOs become less negative as the number of water molecules increase in the system, in an overall

2+ 2+ 2+ general trend as: l[Phe1-Cu(n=1)] < l[Phe4-Cu(n=2)] < l[Phe5-Cu(n=3)] < l[Phe5a1- Cu(n=4)]2+. As a result, binding with more water molecules makes the complex more chemically reactive due to higher SOMO.

Table 6.5: Orbital energies (in eV) of the frontier molecular orbitals of the lowest energy micro-hydrated complexes.

2+ 2+ 2+ 2+ l[Phe1-Cu(n=1)] l[Phe4-Cu(n=2)] [Phe5-Cu(n=3)] [Phe5a-Cu(n=4)] Energies (eV) l l (Tridentate) (Tridentate) (Square-planar) (Square-planar)

E(SOMO) -14.65 -13.62 -13.28 -12.82

E(DOMO) -14.85 -13.82 -13.32 -12.99

E(LUMO) -8.66 -7.8 -7.62 -7.44

E (eV)

E(SOMO-DOMO) 0.2 0.2 0.04 0.17

E(SOMO-LUMO) -5.99 -5.82 -5.66 5.38

E(DOMO-LUMO) -6.19 -6.02 -5.7 5.55

Fig. 6.6 compares the HOMO (i.e., SOMO of the [Phe-Cu]2+ complexes) (top row in fig. 6.6) and HOMO-1 orbitals (bottom row in fig. 6.6) of isolated Phe and the micro-hydrated [Phe-Cu]2+ complexes. The HOMO and HOMO-1 orbitals of Phe are given as references. As indicated in fig. 6.6, both HOMO and HOMO-1 of Phe are dominated by the phenyl ring with small contributions of alaninyl to the HOMO.

2+ 2+ Upon Cu binding with Phe, i.e., l[Phe1-Cu] complex, both HOMO and HOMO-1 concentrate on the local region of copper atom and the carbonyl oxygen (O=C) and

2+ the nitrogen atom of Phe. The HOMO of the l[Phe1-Cu] complex exhibits anti-

2 2 bonding character of the Cu d(x -y ) orbital and the pxy orbitals of N and O(=C) of Phe, in agreement with previous studies of amino acids - Cu2+ complexes[510, 512]. However, the frontier orbitals (HOMO and HOMO-1) of the micro-hydrated

165

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

complexes again are almost solely dominated by the phenyl ring. Moreover, it is

interesting to note that the ‘1’ and ‘2’ contributions in the HOMO and HOMO-1 orbitals of naked Phe, swap consistently in the HOMO and HOMO-1 orbitals of

l(n=1-4) complexes upon sequential inclusion of water molecule into the system. Such differences in HOMO and HOMO-1 between n=0 and n≠0 complexes indicate that the effects of hydration in the Cu2+-Phe complexes are significant.

2+ Fig.6.6. HOMO and HOMO-1 orbitals of the lowest energy structures of Phe3, l[Phe1-Cu] 2+ and lowest energy micro-hydrated complexes. Note that l(n=0) denotes [Phe-Cu] ; l(n=1) 2+ 2+ denotes l[Phe1-Cu(n=1)] ; l(n=2) denotes l[Phe4-Cu(n=2)] ; l(n=3) denotes l[Phe5- 2+ 2+ Cu(n=3)] ; l(n=4) denotes l[Phe5a-Cu(n=4)] .

It is further revealed by the molecular energy diagram of the complexes given in fig. 6.7(a) that the Cu2+ ‘d’ orbital dominant valence MOs of the complexes change apparently, depending on the number of water molecules contained in the complexes. Indeed, the valence band width, in which the Cu2+ ‘d’ orbital dominant MOs of the complexes are concentrated, tend to expand in response to the addition of each water molecule into the system.For instance, the valence MOs dominated by the ‘d’ orbitals 2+ 2+ of the Cu metal ion in the l[Phe1-Cu] complex concentrate in the energy region of 2+ 14.25 eV – 15.25 eV, while the corresponding MOs in the l[Phe5a-Cu(n=4)] complex are spread between 14.50 eV – 17.50 eV. The molecular orbital diagrams associated with the energy levels are presented in the fig. 6.7(b).

166

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

2+ 2+ Fig.6.7. Orbital energy correlation diagram of the Cu ‘d’ orbital MOs in the l[Phe1-Cu] and the lowest energy micro-hydrated complexes along with their corresponding orbital diagrams (b).

167

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

2+ The l[Phe1-Cu] complex presents five valence MOs, 55a-58a, that are mostly specific to the ‘d’ orbital characters of the Cu2+ metal ion in the complex. The HOMO orbital 58a of

2 2 2+ the complex has d(x -y ) characters of the Cu metal ion along with some repulsion from the carboxyl oxygen atoms in the Phe1 moiety present in the complex. The MOs 57a and 55a of

2+ 2+ the l[Phe1-Cu] complex have dxy like characters of the Cu ion, while the MOs 56a and 54a

present dyz and dxz orbital features of the metal ion, respectively.

2+ The l[Phe1-Cu(n=1)] also presents five valence MOs (57a-61a) that are almost similar to those in

2+ the l[Phe1-Cu] complex (i.e. MOs 54a-58a), as a result these orbitals are correlated in the fig. 6.7(a). This indicates that the introduction of the first water molecule do not significantly change the electronic structures of the system, except that the valence MOs are shifted to the

2+ higher energy side. For instance, the 61a of the l[Phe1-Cu(n=1)] complex is dominated by the

2 2 2+ 2+ d(x -y ) characters of the Cu metal ion, which is similar to that of the MO 58a of l[Phe1-Cu] 2+ complex. However, the former (i.e. MO 61a in l[Phe1-Cu(n=1)] complex ) is shifted to the higher 2+ energy against the latter (MO 58a in l[Phe1-Cu] complex).

On the other hand the complexes with more than one water molecule such as l[Phe4- 2+ 2+ 2+ 2+ Cu(n=2)] , l[Phe5-Cu(n=3)] and l[Phe5a-Cu(n=4)] , display significant changes in their Cu ‘d’ 2+ 2+ 2+ orbital dominant MOs. The l[Phe4-Cu(n=2)] , l[Phe5-Cu(n=3)] and l[Phe5a-Cu(n=4)] complexes have seven MOs (58a-61a and 63a-65a), eight MOs (61a-63a, 65a, 66a, 68a, 69a and 71a) and nine MOs (58a, 59a, 64a, 66a-68a, 71a, 74a and 75a), respectively, which are dominated by the d orbital features of the Cu2+ metal ion present in the complexes. Please again refer to fig. 4.7(b) for the

2+ molecular orbital diagrams associated with these MOs. Indeed the l[Phe4-Cu(n=2)] , l[Phe5- 2+ 2+ Cu(n=3)] and l[Phe5a-Cu(n=4)] complexes present few extra valence MOs in addition to the five 2 2 MOs correlated with those from l[Phe1-Cu] (54a-58a) and l[Phe1-Cu(n=1)] (57a-61a) complexes. Such changes in the valence space of the complexes clearly indicate the impacts of sequential addition of water molecules into the system, changes in the configuration of the Phe moiety in the system and the coordination pattern of the Cu2+ metal ion in the complexes.

168

Micro-solvation of Phenylalanine-Cu(II)… Chapter 6

6.6. Summary

2+ Mechanism of the micro-solvation processes (H2O, n=1-4) for the Cu -Phe complexes have been studied using combined DFT calculations as well as CPMD simulations. In the micro-solvation processes, the water molecules are added one by one to form the complexes. The present study demonstrates that the complex is saturated by a maximum of four water molecules which results in a total of thirty-five stable complexes. The lowest energy structures at each micro-hydrated level are obtained using the DFT based geometry optimization and are further studied using the CPMD simulations at the room temperature.

It is found that the number of water molecules involved in the hydrated complexes has an apparent impact on the configuration of the complexes. The present study reveals that a minimum of two water molecules are required to assist the inter-conversion of the Phe moiety between its NT and ZW configurations in the Phe-Cu2+ complex. The complex prefers a NT 2+ 2+ Phe in the configuration of the Phe-Cu molecules without water, l[Phe1-Cu] , or with a 2+ single water molecule, l[Phe1-Cu(n=1)] . The Phe moiety prefers the ZW if more water 2+ 2+ 2+ molecules are presented, i.e. l[Phe4-Cu(n=2)] , l[Phe5-Cu(n=3)] and l[Phe5a-Cu(n=4)] .

The present study further reveals that the micro-hydration mechanisms with the 2+ 2+ presence of the Cu in the Phe-Cu (H2O)n complexes are very different from the Phe(H2O)n complexes (in the absence of Cu2+) through ‘cation-’ interactions found by an earlier study[96]. The most likely contact sites for Cu2+ in the complexes with more than one water molecule includes the carboxyl oxygen atom of Phe and the oxygen atoms from up to three water molecules to form a squared planar coordination of Cu2+ ion in the most stable Cu2+- Phe complexes.

2+ The (N)H···O(3)···H2O– Cu network has been identified in the present study to play a significant role in the stabilization of the micro-solvated Cu2+-Phe complexes. Molecular orbital analyses identify that the Cu2+ d orbital dominant MOs in the complexes are affected by the addition of water molecules, where they tend to expand in the valence energy region when adding the water molecules sequentially.

169

Summary and future work... Chapter 7

CHAPTER

Summary and future work

The effects of inter- and intra-molecular interactions of the aliphatic and aromatic amino acids on their structure-property relationships have been presented in this thesis. A number of quantum mechanical models such as OVGF, MP2 and DFT, together with Slater-type and Gaussian-type basis sets are employed in order to study the electronic structures, spectroscopy and significant intrinsic properties of the amino acids in the gas phase. ‘Dual space analysis’ and ‘fragments-in-molecules’ are employed as efficient analytical tools for probing the intra-molecular mechanisms of the amino acids. Whereas, the inter-molecular interactions of the phenylalanine-copper(II) complexes and micro-solvation process are studied using the DFT based Car-Parrinello molecular dynamics (CPMD) approach. Since amino acids are considered as important building blocks, the results presented here could contribute towards understanding the structural and functional aspects of peptides and proteins.

Impacts of alkyl side chains (R-) on the geometries, electronic structure properties, charge-redistribution and vibrational and chiro-optical properties of the aliphatic amino acids are studied quantum mechanically. The C 1s binding energy spectra of the aliphatic amino acids exhibit side chain dependent additional peaks appearing in the lower energy region of IP < 291 eV, while the N 1s and O 1s spectra only show small perturbations from the spectra of glycine. In the valence ionization space, the alkyl signatures of the amino acids concentrate in the mid valence region of 12 eV < IP < 16 eV, as well as the decrease of the HOMO- LUMO energy gaps. Orbitals less affected by the alky side chains are those in the innermost

170

Summary and future work... Chapter 7

(IPs > 16 eV) and the outermost (IPs < 12 eV) valence regions, where the orbital momentum distributions display close association among the amino acids.

Vibrational and chiro-optical properties of the neutral (in the gas phase) and zwitterionic (in the CPCM water model) amino acids in response to their alkyl side chains are also studied. Vibrational optical activity (VOA) spectra such as VCD and ROA are employed in conjunction with their respective IR and Raman spectra. The study generally confirms that the functional region of 1600-4000 cm-1 is more Raman active, whereas the fingerprint region (dominated by alkyl side chains) of 400 – 1600 cm-1 is IR intensive, in both the neutral and zwitterionic amino acids. The chiral carbons produce intense VCD and ROA bands in the spectra of neutral amino acids, whereas the alkyl (or methyl) vibrations play important role in the VCD and ROA spectra of zwitterion amino acids. The CH dependent vibrations in the functional region produce opposite signs in the VCD and ROA spectra of the amino acids. Moreover, the asymmetric vibrations of the amino acids produce intense VCD and ROA bands, while the symmetric CH vibrational motions exhibit weak signals. The alkyl region (υ < 1600 cm-1) of the VCD and ROA spectra of the amino acids are, however, dominated by the -carbon and alkyl side chain vibrations.

A comprehensive electron level picture of L-Phe is revealed by studying the interactions

of its functional groups (COOH, NH2 and phenyl) and its fragment schemes (alanine/benzene or glycine/toluene) in the gas phase. The study identifies that the functional groups play a significant role in the inner shell of L-Phe, while its valence space are dominated by the fragment interactions. Comparison of the inner shell XPS of L-Phe, PEA (i.e., phe – COOH)

and PPA (i.e., phe – NH2) finds that the impact of functional group substitution in the inner

shell is particularly higher at the C() site of PEA, in the absence of carboxyl group. This also leads to the significant reduction of the dipole moment in PEA, indicating that the carboxyl group dominates the inner shell of L-Phe. It is also found that the phenyl ring in L-Phe serves as a buffer to resist the changes and thereby stabilizing the amino acid. Furthermore, probing the electronic structures of L-Phe and its fragment schemes, alanine/benzene and glycine/toluene, using valence XPS and dual space analysis unravel that the mid valence region of 11 < IP < 20 eV show fragment related molecular orbitals. The valence space, 14 < IP < 20 eV is particularly dominated by glycine/toluene scheme, while a short valence space of 11 < IP < 14 eV is dominated by alanine/benzene scheme. However, the frontier molecular

171

Summary and future work... Chapter 7

orbitals and the innermost orbitals are less affected by the fragment interactions. Therefore, the alanine/benzene and glycine/toluene are mostly the dominant configurations that contribute to the electronic structures and properties of L-Phe.

Subsequently, the spectrsocopy and orbital features of L-Phe (R=phenyl ring), together with those of L-Tyr (R=phenol ring), are used to understand the sturucture-properties of L- Dopa (R=catechol ring), an important neurotransmitter drug. The effects of phenyl side chain modification of the aromatic molecules on their electronic structures and properties are probed in the gas phase. The geometric properties are less affected by the phenyl side chain modifications, while the molecular properties such as dipole moment, rotational constants, molecular size and aromaticity features of the molecules change considerably. In the inner shell, the C 1s and O 1s spectra of the aromatic molecules display apparent changes, with the most changes occurring in the lower energy side < 292 eV of the C 1s spectra, and at higher energy ca. 536 eV in the O 1s spectra. The atoms that directly involve in the modifications,

i.e., C(3)/O(3) in L-Tyr and C(3)/O(3) and C(4)/O(4) in L-Dopa, are mostly affected. In the valence space, the most spectral differences among the molecules again concentrate in the mid- valence region of 12 eV < IP < 16 eV, which agree with the aliphatic and other aromatic molecules discussed in this thesis. On the other hand, the frontier orbitals are also significantly affected by the phenyl side chain modifications in L-Phe, L-Tyr and L-Dopa, which lead to orbital rearrangments in the outermost valence space < 12 eV. The HOMO- LUMO gap decreases with the increase of number of the OH groups as L-Phe > L-Tyr > L- Dopa. Therefore, the quantum mechanical properties studied in this work clearly differentiate the natural amino acids, L-Phe and L-Tyr against the L-Dopa drug compound.

Micro-solvation processes (H2O, n=1-4) of phenylalanine-copper(II) ([Phe-Cu]2+) complexes are investigated using the DFT calculations and Car-Parrinello molecular dynamics (CPMD) simulation. The structures of the [Phe-Cu]2+ complex with up to four water molecules (n=0-4) are studied. It is found that the Phe moiety appears to be in the neutral (NT) form in isolated (n=0) and mono-hydrated (n=1) complexes, but in the zwitterionic (ZW) form in the other hydrated complexes (n2). The energy minimum 2+ structures of the [ complexes suggest that the Cu –π interactions are not dominant in the complexes. The present CPMD simulations of the lowest energy micro- hydrated structures reveal that the maximum coordination of Cu2+ in the presence of the Phe

172

Summary and future work... Chapter 7

ligand does not exceed four: the oxygen atoms from three water molecules and one carboxyl oxygen atom of Phe. Any excess water molecules will migrate to the second solvation shell. 2+ Moreover a unique structural motif, (N)H···O(3)···H2O– Cu is present in the micro-hydrated complexes, other than monohydrated system, which is found to be significant in stabilizing the structures of the complexes.

To summarize, the structures and properties of amino acids are elucidated using quantum mechanical calculations. The results indicate that no single universal model is suitable for the calculation of all the properties of amino acids. It is found that the LB94, in combination with TZ2P or et-pVQZ, provides a fairly reasonable calculation of the core IPs of the amino acids. Whereas, the SAOP/et-pVQZ (for inner valence space) and OVGF/TZVP (for outer valence space) are able to accurately produce the valence IPs. The results produced by these QM models show excellent synergies with the ‘state-of-the-art’ synchrotron sourced XPS measurements. Further, DFT calculations when combined with CPMD simulation are useful for studying the dynamics and environmental effects on the amino acids.

Once the structures and properties of the individual amino acids are understood in the gas phase and in micro-solvation environment, studies in bulk solvation need to be considered. Investigation of amino acids in the bulk water solution can provide extended information about the solute-solvent interactions and solvation structures of amino acids. Such information are generally accessible through large scale molecular dynamics simulation. Even with the most advanced high performance computing clusters available today, it is still a challenge to apply CPMD or other DFT methods to study the dynamics of a large system consisting of hundreds of atoms, such as an amino acid in bulk solvents. On the other hand, classical molecular dynamics approaches, which can treat large systems, are unable to accurately predict the chemical changes in the systems.

As a result, hybrid quantum mechanics (QM)-molecular mechanics (MM) approaches are appreciated. In a QM/MM scheme, the chemically active part of the entire system is treated at QM level and the rest of the larger environment (bulk solvents) is modelled using classical force fields. CPMD/MM method has been useful to study the solvation structure properties of aromatic amino acids in bulk water. The CPMD/MM approach has shown to reproduce well the solvation structures and dipole auto correlation infra-red spectra of the aromatic amino acids. Few of these results are briefly provided in the appendix (A.X). More

173

Summary and future work... Chapter 7

extended simulation can be undertaken to study the excited state and other properties in the future.

A more systematic molecular level understanding of the structures and properties of the amino acids, are fundamentally important for the design and development of novel molecular entities in different areas of science, from medicinal sciences to material sciences. For instance, sugar amino acids are important class of multi-functional scaffolds, which have gained popularity in peptidomimetics and drug discovery. Sugar amino acids are basically the hybrids of amino acids and carbohydrates, where the amino acid moieties such as carboxyl group, amino group and hydroxyl group are fused with a regular sugar component, in order to bridge the carbohydrates and proteins. Therefore, designing and synthesizing novel sugar amino acids are of potential interests. In accordance with the building block principle, the intrinsic properties of the amino acids should be useful for designing different scaffolds of sugar amino acids.

174

Appendix

A.I: C 1s level energy correlation diagram of the aliphatic amino acids calculated using the EKS method.

175

Appendix

A.II: Selected geometric parameters of L-Phe and its fragments optimized using B3LYP/TZVP model along with the available experimental data

L-Phe Alanine Glycine Parameters This This Expb This Expb Other Worka Worka Worka c e g C(1)-C() (Å) 1.55 1.54 1.51 1.54 1.529 1.52 1.55 1.53 1.54c C()-C() (Å) c e g C()-N (Å) 1.47 1.46 1.47 1.47 1.466 1.45 111.90 109.10 C(1)-C()-C() /º 109.30 109.50 110.00c 111.60 113e 115.60 g C(1)-C()-N /º -102.00 -134.20 567.00 O-C(1)-C()-H(1)/º 40.00 108.70 -123.00 O-C(1)-C()-C() /º

Phenyl Benzene Toluene

d h C()-C(2) (Å) 1.40 1.39 1.39 1.40 1.40 h C(2)-C(3) (Å) 1.39 1.39 1.39 h C(3)-C(4) (Å) 1.39 1.39 1.39 f h C(4)-C(5) (Å) 1.39 1.39 1.39 1.39 f h C(5)-C(6) (Å) 1.39 1.39 1.39 1.398 f h C(6)-C() (Å) 1.40 1.40 1.39 1.40 d h R6 (Å) 8.37 8.36 8.35 8.36 8.35 120.70 120.00 121.00 C(-C(2)-C(3) /º 120.40 121.00 C(2)-C(3)-C(4) /º 119.60 119.40 119.40 f C(3)-C(4)-C(5) /º 120.00 121.00 120.20f C(4)-C(5)-C(6) /º 121.10 121.00 120.60 f C(5)-C(6)-C() /º 118.30 118.60 119.00f C(6)-C()-C(2) /º -0.08 0.00 -0.11 C()-C(2)-C(3)-C(4) /º -0.23 -0.08 C(2)-C(3)-C(4)-C(5) /º 0.23 0.08 C(3)-C(4)-C(5)-C(6) /º -0.07 0.11 C(4)-C(5)-C(6)-C() º -0.37 -0.28 C(5)-C(6)-C()-C(2) /º a B3LYP/TZVP model[54]. bExperimental structure may not be the same conformer investigated in this study. Therefore used only as a guide. See Ref. [406]. c Ref. [513] . dMW spectrum[514]. eRef. [312]. fMicrowave spectrum[515]. gMP2 calculations[72]. hRef. [516].

176

Appendix

A.III: Valence orbital diagrams of L-Phe correlated with its fragment pairs: benzene/alanine and toluene/glycine in the inner valence. Note that the figures are not presented based on the energy scale.

177

Appendix

A.IV: Valence orbital diagrams of L-Phe correlated with its fragment pairs: benzene/alanine and toluene/glycine in the outer valence regions: Note that the figures are not given according to the energy scale.

178

Appendix

A.IV: Continued…

179

Appendix

A.V: Geometric parameters of the phenyl ring in the L-phe, L-tyr and L-dopa optimized using B3LYP/6-311G**

L-phe L-tyr L-dopa (tyr-phe) (dopa-tyr)

C ()-C(2)/Å 1.40 1.40 1.40 0 0

C(2)-C(3)/Å 1.39 1.39 1.39 0 0

C(3)-C(4)/Å 1.39 1.40 1.41 0.01 0.01

C(4)-C(5)/Å 1.39 1.40 1.39 0.01 -0.01

C(5)-C(6)/Å 1.40 1.39 1.39 -0.01 0

C(6)-C()/Å 1.40 1.40 1.40 0 0

C()-C(2)-C(3)/◦ 120.64 120.13 120.53 -0.51 0.40

C(2)-C(3)-C(4)/◦ 120.40 120.22 120.81 -0.18 0.59

C(3)-C(4)-C(5)/◦ 119.57 119.54 118.97 -0.03 -0.57

C(4)-C(5)-C(6)/◦ 119.98 119.65 120.16 -0.33 0.51

C(5)-C(6)-C()/◦ 121.04 121.74 121.25 0.70 -0.49

C(g)-C(2)-C(3)-C(4)/◦ 0.08 0.04 -0.03 -0.04 -0.07

C(2)-C(3)-C(4)-C(5)/◦ -0.20 -0.25 -0.13 -0.05 0.12

C(3)-C(4)-C(5)-C(6)/◦ 0.13 0.18 0.19 0.05 0.01

C(4)-C(5)-C(6)-C()/◦ 0.06 0.11 -0.09 0.05 -0.20

180

Appendix

A.VI: Valence MOs of L-DOPA from SAOP/et-pVQZ calculations

181

Appendix

A.VII: Valence MOs of L-tyr from SAOP/et-pVQZ calculations

182

Appendix

A.VIII: Valence MOs of L-Phe from SAOP/et-pVQZ calculations

183

Appendix

A.IX: Convergence tests for the copper(II) (a), oxygen (b), nitrogen (N) and carbon (d) depending on the cut-off values and using different methods such as MT and PBE.

184

Appendix

A.X : CP/MM simulation of phenylalanine in water and histidine in water systems.

In the present work, we have performed hybrid Car-Parrinello QM/MM molecular dynamics simulations of two systems, phenylalanine (Phe) in water and histidine (His) in water, as shown in A.X.1. The DFT based Car-Parrinello molecular dynamics (CPMD) combined with the GROMOS96 classical MD code is employed. Different properties such as geometries, hydration shells, dipole distributions, infra-red (IR) spectra and electronic properties of Phe and His in aqueous solution are calculated. The hydration shells around each functional group in the amino acids are examined using the radial distribution functions (RDFs), while the IR spectra are generated by Fourier transforming the dipole autocorrelation functions. This study is therefore useful to comprehend the impacts of solvent environments on the structural, electronic and dynamic properties of the amino acids.

Phe in Water His in Water

A.X. 1. QM/MM systems of Phe in water and His in water

Simulation parameters

Two systems, His in water and Phe in water, are prepared by immersing separately the optimized zwitterion structures of His and Phe at B3LYP/6-31G** model into a 14 Å orthorhombic box containing 1400 water molecules. The water model used in this study is TIP3P, as it has been realized to be one of the most suitable models for biological systems. The solvated systems are initially equilibrated using classical Amber 11 code in order to obtain reasonably stable systems for performing CP/MM simulations. The classical MD

185

Appendix simulations are carried out using the parm99 force field and the bond lengths involving hydrogen bonds (H-bonds) are constrained with the Shake algorithm. The particle mesh ewald (PME) method with 10 Å cut-off value are applied for treating the long range electrostatic and van der Waals interactions, respectively. With these parameters, the MD simulations of the systems are performed as a three step procedure: 1000 steps of steepest descent energy minimization, 500 ps of NVT (constant volume and constant temperature) equilibration and finally 300 ps of NPT (constant pressure and constant temperature) classical simulations at 300 K temperature and 1 atm pressure. Langevin thermostat is employed for maintaining the NVT and NPT conditions in the simulations. Fig. 7.1 presents the different energy terms (kinetic, potential and total), temperature and pressure from the trajectories of Phe in water and His in water obtained from their 300 ps classical equilibration. As the figure indicates, all the parameters in the solvated Phe and His systems stabilized very well.

Subsequently, hybrid CP/MM simulations of the solvated systems are carried out for ~30 ps where, the amino acids are treated at QM level using the DFT based CPMD code and the TIP3P water molecules are simulated using classical GROMOS96 code. The interface between CPMD and GROMOS96 developed by Rothlisberger et al is used in this work. The MM parameters are similar to those used in the classical pre-equilibration. The QM region concerning Phe and His are treated using the BLYP exchange correlation functional. The core electrons in the QM systems are described using the norm-conserving Troullier and Martins pseudopotentials, while the valence electrons are treated explicitly. The wavefunctions in the systems are expanded in a plane-wave basis set up to an energy cut-off of 90 Ry. A time step of 5 a.u. along with a fictitious electronic mass of 600 a.u. is employed. Constant temperature conditions in the CPMD are achieved by employing the Nose-Hoover thermostat and the QM part are simulated under isolated system conditions by using Tuckermann Poisson solvers. Furthermore, dipole moments of the systems are collected at every 5 steps during the production simulations by employing the dipole dynamics scheme, which are later Fourier transformed to obtain the IR spectra of the systems.

186

Appendix

Results

Liquid properties of TIP3P model

Liquid property of water is one of the significant requirements for carrying out any studies in bulk water. For this reason, the behaviours of the TIP3P water molecules involved in the QM/MM systems of the amino acids in this study are initially investigated using radial distribution functions. A.X.2. compares the calculated RDFs of (a) hydrogen-hydrogen interactions (gOO), (b) oxygen-hydrogen interactions (gOH) and (c) oxygen-oxygen interactions

(gHH) in the TIP3P water molecules of the QM/MM systems, PheZW(QM/MM) and HisZW(QM/MM), against the experimental data for pure liquid water.

As it can be seen in the figure, our calculated RDFs for TIP3P water molecules are in good agreement with the previous experiment. The RDFs of gOH (fig. 7.x(a)) in

PheZW(QM/MM) and HisZW(QM/MM) systems present two peaks, the first one at ~2.5 Å and the second one at ~3.8 Å, which are close to the experimental spectra. Small perturbations in the computed RDFs, especially in the first peak, indicate the effects of amino acid (His and Phe in this study) interactions with the water molecules. Note that the experimental data are for pure liquid water without any solute interactions; therefore, small shifts in the computed

RDFs in this study are justified. Similarly, the calculated gOH spectra in the PheZW(QM/MM) and HisZW(QM/MM) are almost close to the experiments, except that the second peak in this study is slightly lower than that of the measured peak for water.

In the case of gOO, the first peak in both the PheZW(QM/MM) and HisZW(QM/MM) is very close to the experimental peak. But, the structure beyond the first peak is almost flat in the TIP3P water molecules in our QM/MM systems, while the experimental spectrum shows some peak features. It is understood from the previous studies that TIP3P water molecules tend to show such little structures beyond the first solvation shell. At the same time, TIP3P water models are more appropriate for studying the properties of the biological systems. Therefore in summary, the liquid behaviour of the TIP3P water molecules in the

PheZW(QM/MM) and HisZW(QM/MM) are described appropriately by our QM/MM simulations.

187

Appendix

A.X. 2. Radial distribution functions of TIP3P water molecules in the PheZW(QM/MM) and

HisZW(QM/MM) systems compared against the experimental data for liquid water: (a) gHH; (b)gOH;

and (c) gOO.

188

Appendix

Solvation structure of Phe and His

A.X. 3. Radial distribution functions of Histidine and Phenylalanine with water: (a) Carboxyl group –

Water; (b) Amino group – Water; (c) H()-Water and C-Water; (d) Aromatic ring – Water.

189

Appendix

Geometry fluctuations & H-bonds

A.X. 4. Dihedral fluctuations in Phe and His in water.

A.X. 5. H-bonds between Phe and water and His in water.

190

Appendix

Dipole distribution function

A.X. 6. compares the dipole moment distributions of PheZW(QM/MM) and

HisZW(QM/MM) calculated from their Wannier functions. The dipole moments of the amino acids calculated in the implicit CPCM water model (PheZW(CPCM) and HisZW(CPCM)) and those of the neutral structures in gas phase (PheNT and HisNT) are also marked in the figure for reference purposes. The corresponding numerical values are given in A.X. 7.

A.X. 6. Distribution functions of molecular dipole moments for PheZW(QM/MM) (upper panel) and

HisZW(QM/MM) (lower panel), along with the dipole moments calculated in the implicit continuum model (CPCM) and gas phase.

As it can be seen in the figure, the dipole moments of the ZW forms of the amino acids change significantly from their corresponding NT forms. For instance, the dipole moments of PheZW(QM/MM) is distributed between 14D – 20D, while the dipole of the PheNT is only 5.2D. Similarly, the dipole moments of HisZW(QM/MM) are spread over 8D-16D when the dipole of HisNT is only 4.3D. Such large increase in the dipole moment from NT form to ZW form in water is apparently due to the structural changes in the amino acids. The NT forms of the amino acids hold covalent bonds with COOH and NH2 groups, those change to - + ionic bonds with COO and NH3 groups in the ZW forms. Such charge separations in the amino acid moieties of PheZW and HisZW coupled with solute-solvent interactions result in large dipole moments. Similar effects were reported in a previous ab-initio molecular

191

Appendix dynamics study of glycine in water. This is also reflected in the dipole moments of

PheZW(CPCM) and HisZW(CPCM) calculated in the implicit water (continuum) models, which are larger than that of their respective NT structures. However, the dipole moments in the implicit water models using CPCM calculation are substantially underestimated when compared to those of the explicit water systems in QM/MM simulation. For example, the dipole moment calculated for HisZW(QM/MM) is approximately 12D but the continuum model estimates it as only 8.7D. Similar trends are also seen in PheZW systems. This indicates that the polarization effects arising from the inter molecular interactions between the amino acid and surrounding water molecules contribute significantly to the molecular dipole moments, which are not sufficiently described in the continuum environment.

A.X. 7. Dipole moments (in Debye) of His and Phe in explicit water (QM/MM), continuum model (CPCM) and gas phase (NT)

System Phe His  QM/MM (ZW) ~17 ~12 ~5 CPCM continuum (ZW) 12.9 8.7 4.2 Gas phase (NT) 5.2 4.3 0.9

Furthermore, the dipole moments of His in all the media such as gas phase (HisNT), continuum (HisZW(CPCM)) and explicit aqueous solution (HisZW(QM/MM)), exhibit a uniform shift towards the lower dipole side against Phe. Indeed the shift becomes larger when moving from gas phase to explicit aqueous solution, that is, the HisNT is only shifted by 0.9D against

PheNT, on the other hand, HisZW(QM/MM) is shifted by ~5D against PheZW(QM/MM) system. This is because, the imidazole ring in histidine with two strong electronegative nitrogen atoms is able to balance the charges from its amino acid moiety, which is enhanced further by the interactions with explicit water molecules in the environment. Whereas in PheZW, the charges concentrate on its amino acid moiety only, which in turn, increase the dipole moments of Phe systems. Refer to table 7.x for the numerical values of the dipole moments of His and Phe calculated using different methods.

192

Appendix

Infra-red spectra from Fourier transformation of the dipole autocorrelation function

A.X. 8. Comparison of the IR spectra obtained from Fourier transformation of the dipole autocorrelation function from the QM/MM dynamics simulation of Phe in water against the Experimental data.

A.X. 9. Comparison of the IR spectra of Phe in water and His in water obtained from Fourier transformation of the dipole autocorrelation function from the QM/MM dynamics simulation.

193

References

1. Brocchieri L and Karlin S: Protein length in eukaryotic and prokaryotic proteomes. Nucleic Acids Research, 33(10):3390-3400. 2. Duan G, Smith VH and Weaver DF: Data mining, ab initio, and molecular mechanics study on conformation of phenylalanine and its interaction with neighboring backbone amide groups in proteins. International Journal of Quantum Chemistry 2002, 90(2):669-683. 3. Takei T, Okonogi A, Tateno K, Kimura A, Kojima S, Yazaki K and Miura K-i: The Effects of the Side Chains of Hydrophobic Aliphatic Amino Acid Residues in an Amphipathic Polypeptide on the Formation of α Helix and Its Association. Journal of Biochemistry 2006, 139(2):271-278. 4. Monera OD, Sereda TJ, Zhou NE, Kay CM and Hodges RS: Relationship of sidechain hydrophobicity and α-helical propensity on the stability of the single- stranded amphipathic α-helix. Journal of Peptide Science 1995, 1(5):319-329. 5. Baumann MK, Textor M and Reimhult E: Understanding Self-Assembled Amphiphilic Peptide Supramolecular Structures from Primary Structure Helix Propensity. Langmuir 2008, 24(15):7645-7647. 6. Kaiser ET: Design and construction of Biologically active peptides, including enzymes. Pure and Applied Chemistry 1984, 56(8):979-987. 7. Shimomura Y, Yamamoto Y, Bajotto G, Sato J, Murakami T, Shimomura N, Kobayashi H and Mawatari K: Nutraceutical Effects of Branched-Chain Amino Acids on Skeletal Muscle. The Journal of Nutrition 2006, 136(2):529S-532S. 8. Udenfriend S and Cooper JR: The enzymatic conversion of phenylalanine to tyrosine. Journal of Biological Chemistry 1952, 194(2):503-511. 9. Marchini JS, Castillo L, Chapman TE, Vogt JA, Ajami A and Young VR: Phenylalanine conversion to tyrosine: Comparative determination with l-[ring- 2H5]phenylalanine and l-[1-13C]phenylalanine as tracers in man. Metabolism 1993, 42(10):1316-1322. 10. Chandra P and Vining LC: Conversion of phenylalanine to tyrosine by microorganisms. Canadian Journal of Microbiology 1968, 14(5):573-578. 11. Womack M and Rose WC: Feeding experiments with mixtures of highly purified amino acids. Journal of Biological Chemistry 1934, 107(2):449-458. 12. Moss AR and Schoenheimer R: The conversion of phenylalanine to tyrosine in normal rats. Journal of Biological Chemistry 1940, 135(2):415-429. 13. Kaufman S: The enzymatic conversion of phenylalanine to tyrosine. Journal of Biological Chemistry 1957, 226(1):511-524. 14. Prescott BA, Borek E, Brecher A and Waelsch H: Studies on oligophrenia phenylpyruvica. Journal of Biological Chemistry 1949, 181(1):273-279. 15. Surmeier DJ: Homeostatic regulation of dopaminergic neurons without dopamine. Proceedings of the National Academy of Sciences of the United States of America 2004, 101(36):13103-13104. 16. Maddaluno JF and Faull KF: Fast enzymatic preparation of l-DOPA from tyrosine and molecular oxygen: a potential method for preparing [15O]l-DOPA. International Journal of Radiation Applications and Instrumentation Part A Applied Radiation and Isotopes 1990, 41(9):873-878.

194

References

17. Robinet JJ, Baciu C, Cho K-B and Gauld JW: A Computational Study on the Interaction of the Nitric Oxide Ions NO+ and NO- with the Side Groups of the Aromatic Amino Acids. The Journal of Physical Chemistry A 2007, 111(10):1981- 1989. 18. Baker CM and Grant GH: Role of aromatic amino acids in protein–nucleic acid recognition. Biopolymers 2007, 85(5 -6):456-470. 19. Seeman NC, Rosenberg JM and Rich A: Sequence-specific recognition of double helical nucleic acids by proteins. Proceedings of the National Academy of Sciences 1976, 73(3):804-808. 20. Guckian KM, Schweitzer BA, Ren RXF, Sheils CJ, Tahmassebi DC and Kool ET: Factors Contributing to Aromatic Stacking in Water: Evaluation in the Context of DNA. Journal of the American Chemical Society 2000, 122(10):2213-2222. 21. Nakatani K, Matsuno T, Adachi K, Hagihara S and Saito I: Selective Intercalation of Charge Neutral Intercalators into GG and CG Steps: Implication of HOMO- LUMO Interaction for Sequence-Selective Drug Intercalation into DNA. Journal of the American Chemical Society 2001, 123(24):5695-5702. 22. Turner JM, Swalley SE, Baird EE and Dervan PB: Aliphatic/Aromatic Amino Acid Pairings for Polyamide Recognition in the Minor Groove of DNA. Journal of the American Chemical Society 1998, 120(25):6219-6226. 23. Fukutomi R, Tanatani A, Kakuta H, Tomioka N, Itai A, Hashimoto Y, Shudo K and Kagechika H: Aromatic layered guanidines bind sequence-specifically to DNA minor groove with precise fit. Tetrahedron Letters 1998, 39(36):6475-6478. 24. Bailly C, Tardy C, Wang L, Armitage B, Hopkins K, Kumar A, Schuster GB, Boykin DW and Wilson WD: Recognition of ATGA Sequences by the Unfused Aromatic Dication DB293 Forming Stacked Dimers in the DNA Minor Groove. Biochemistry 2001, 40(33):9770-9779. 25. Tomsic M, Tsujikawa L, Panaghie G, Wang Y, Azok J and deHaseth PL: Different Roles for Basic and Aromatic Amino Acids in Conserved Region 2 of Escherichia coli ς70 in the Nucleation and Maintenance of the Single-stranded DNA Bubble in Open RNA Polymerase-Promoter Complexes. Journal of Biological Chemistry 2001, 276(34):31891-31896. 26. Blakaj DM, McConnell KJ, Beveridge DL and Baranger AM: Molecular Dynamics and Thermodynamics of Protein−RNA Interactions: Mutation of a Conserved Aromatic Residue Modifies Stacking Interactions and Structural Adaptation in the U1A−Stem Loop 2 RNA Complex. Journal of the American Chemical Society 2001, 123(11):2548-2551. 27. Nolan SJ, Shiels JC, Tuite JB, Cecere KL and Baranger AM: Recognition of an Essential Adenine at a Protein−RNA Interface: Comparison of the Contributions of Hydrogen Bonds and a Stacking Interaction. Journal of the American Chemical Society 1999, 121(38):8951-8952. 28. Nguyen B, Lee MPH, Hamelberg D, Joubert A, Bailly C, Brun R, Neidle S and Wilson WD: Strong Binding in the DNA Minor Groove by an Aromatic Diamidine with a Shape That Does Not Match the Curvature of the Groove. Journal of the American Chemical Society 2002, 124 (46):13680-13681. 29. Warner PM, Qi J, Meng B, Li G, Xie L, El-Shafey A and Jones GB: DNA cleavage by aromatic amines. Bioorganic & Medicinal Chemistry Letters 2002, 12(1):1- 4. 30. Morcock DR, Sowder Ii RC and Casas-Finet JR: Role of the histidine residues of visna virus nucleocapsid protein in metal ion and DNA binding. FEBS Letters 2000, 476(3):190-193.

195

References

31. Yoshikawa M, Iwasaki H and Shinagawa H: Evidence that Phenylalanine 69 in Escherichia coliRuvC Resolvase Forms a Stacking Interaction during Binding and Destabilization of a Holliday Junction DNA Substrate. Journal of Biological Chemistry 2001, 276(13):10432-10436. 32. Colizzi F, Perozzo R, Scapozza L, Recanatini M and Cavalli A: Single-Molecule Pulling Simulations Can Discern Active from Inactive Enzyme Inhibitors. Journal of the American Chemical Society 2010, 132(21):7361-7371. 33. Lüdemann SK, Lounnas V and Wade RC: How do substrates enter and products exit the buried active site of cytochrome P450cam? 1. Random expulsion molecular dynamics investigation of ligand access channels and mechanisms. Journal of Molecular Biology 2000, 303 (5):797-811. 34. Klvana M, Pavlova M, Koudelakova T, Chaloupkova R, Dvorak P, Prokop Z, Stsiapanava A, Kuty M, Kuta-Smatanova I, Dohnalek J, Kulhanek P, Wade RC and Damborsky J: Pathways and Mechanisms for Product Release in the Engineered Haloalkane Dehalogenases Explored Using Classical and Random Acceleration Molecular Dynamics Simulations. Journal of Molecular Biology 2009, 392(5):1339- 1356. 35. Long D, Mu Y and Yang D: Molecular Dynamics Simulation of Ligand Dissociation from Liver Fatty Acid Binding Protein. PLoS ONE 2009, 4(6):e6081. 36. Kalyaanamoorthy S and Chen Y-PP: Exploring Inhibitor Release Pathways in Histone Deacetylases Using Random Acceleration Molecular Dynamics Simulations. Journal of Chemical Information and Modeling 2012, 52(2):589-603. 37. Banos G, Daniel PM, Moorhouse SR and Pratt OE: The Influx of Amino Acids into the Brain of the Rat in vivo: The Essential Compared with Some Non-Essential Amino Acids. Proceedings of the Royal Society of London Series B Biological Sciences 1973, 183(1070):59-70. 38. Berg JM, Tymoczko JL and Stryer L: Biochemistry, 5th edition; New York: W H Freeman; 2002. 39. Martin ACR, Orengo CA, Hutchinson EG, Jones S, Karmirantzou M, Laskowski RA, Mitchell JBO, Taroni C and Thornton JM: Protein folds and functions. Structure (London, England : 1993) 1998, 6(7):875-884. 40. Zondlo NJ: Non-covalent interactions: Fold globally, bond locally. Nature Chemical Biology 2010, 6(8):567-568. 41. Lapanje S, Škerjanc J, Glavnik S and Žibret S: Thermodynamic studies of the interactions of guanidinium chloride and urea with some oligoglycines and oligoleucines. The Journal of Chemical Thermodynamics 1978, 10(5):425-433. 42. Remko M, Fitz D, Broer R and Rode B: Effect of metal Ions (Ni2+, Cu2+ and Zn2+) and water coordination on the structure of L-phenylalanine, L-tyrosine, L-tryptophan and their zwitterionic forms. Journal of Molecular Modeling 2011:1- 12. 43. Rimola A, Rodríguez-Santiago L and Sodupe M: Cation−π Interactions and Oxidative Effects on Cu+ and Cu2+ Binding to Phe, Tyr, Trp, and His Amino Acids in the Gas Phase. Insights from First-Principles Calculations. Journal of Physical Chemistry B 2006, 110(47):24189-24199. 44. Ganesan A, Dreyer J, Larrucea J, Wang F and Akola J: CPMD simulation of Cu2+ - phenylalanine complex under micro-solvated environment. (Submitted, 2013). 45. Betts MJ and Russell RB: Amino Acid Properties and Consequences of Substitutions. In: Bioinformatics for Geneticists. John Wiley & Sons, Ltd; 2003: 289-316.

196

References

46. Strominger JL: Peptide vaccination against cancer? Nat Med 1995, 1(11):1140- 1140. 47. Iversen LL: The Ferrier Lecture, 1983: Amino Acids and Peptides: Fast and Slow Chemical Signals in the Nervous System? Proceedings of the Royal Society of London Series B Biological Sciences 1984, 221(1224):245-260. 48. Fonnum F: Glutamate: A Neurotransmitter in Mammalian Brain. Journal of Neurochemistry 1984, 42 (1):1-11. 49. Johnson J, Tian N, Caywood MS, Reimer RJ, Edwards RH and Copenhagen DR: Vesicular Neurotransmitter Transporter Expression in Developing Postnatal Rodent Retina: GABA and Glycine Precede Glutamate. The Journal of Neuroscience 2003, 23 (2):518-529. 50. Davanger S, Ottersen OP and Storm-Mathisen J: Glutamate, GABA, and glycine in the human retina: An immunocytochemical investigation. The Journal of Comparative Neurology 1991, 311(4):483-494. 51. Farré EM, Tiessen A, Roessner U, Geigenberger P, Trethewey RN and Willmitzer L: Analysis of the Compartmentation of Glycolytic Intermediates, Nucleotides, Sugars, Organic Acids, Amino Acids, and Sugar Alcohols in Potato Tubers Using a Nonaqueous Fractionation Method. Plant Physiology 2001, 127(2):685-700. 52. Falzon CT and Wang F: Understanding glycine conformation through molecular orbitals. The Journal of Chemical Physics 2005, 123(21):214307-214312. 53. Falzon CT, Wang F and Pang W: Orbital Signatures of Methyl in l-Alanine. The Journal of Physical Chemistry B 2006, 110(19):9713-9719. 54. Ganesan A, Wang F and Falzon C: Intramolecular interactions of L- phenylalanine: Valence ionization spectra and orbital momentum distributions of its fragment molecules. Journal of Computational Chemistry 2011, 32(3):525- 535. 55. Ganesan A, Brunger MJ and Wang F: Influence of functional groups on the C[alpha]-C[beta] chain of l-phenylalanine and its derivatives. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 2010, 619(1 -3):143-146. 56. Ganesan A and Wang F: Intramolecular interactions of L-phenylalanine revealed by inner shell chemical shift. The Journal of Chemical Physics 2009, 131(4):044321-044329. 57. Ganesan A, Wang F, Brunger M and Prince K: Effects of alkyl side chains on properties of aliphatic amino acids probed using quantum chemical calculations. Journal of Synchrotron Radiation 2011, 18(5):733-742. 58. Downton MT, Sadus RJ, Todd BD and Wang F: Quantum mechanical study of the electron momentum spectrum of electronic structure of adenine. In. 59. Zhu Q, Wang F and Ivanova EP: Impact of ketone and amino on the inner-shell of guanine. Journal of Synchrotron Radiation 2009, 16(4):545-552. 60. Saha S, Wang F, Guerra CF and Bickelhaupt FM: Outer valence orbital response to proton positions in prototropic tautomers of adenine. In. 61. Novak I and Kova B: Photoelectron Spectra of Important Drug Molecules: Zidovudine and Artemisinine. The Journal of Organic Chemistry 2003, 68(14):5777-5779. 62. Chen F, Selvam L and Wang F: Blue shifted intramolecular C−H···O improper hydrogen bonds in conformers of zidovudine. Chemical Physics Letters 2010, 493 (4-6):358-363.

197

References

63. Selvam L, Chen F and Wang F: Solvent effects on blue shifted improper hydrogen bond of C–H O in deoxycytidine isomers. Chemical Physics Letters 2010, 500(4- 6):327-333. 64. Selvam L, Vasilyev V and Wang F: Methylation of Zebularine: A Quantum Mechanical Study Incorporating Interactive 3D pdf Graphs. The Journal of Physical Chemistry B 2009, 113(33):11496-11504. 65. Thompson A, Saha S, Wang F, Tsuchimochi T, Nakata A, Imamura Y and Nakai H: Density Functional Study on Core Ionization Spectra of Cytidine and Its Fragments. Bulletin of the Chemical Society of Japan 2009, 82 (2):187-195. 66. Arachchilage A, Wang Y and Wang F: A quantum mechanical study of bioactive 3- chloro-2,5-dihydroxybenzyl alcohol through substitutions. Theoretical Chemistry Accounts: Theory, Computation, and Modeling (Theoretica Chimica Acta) 2011, 130(4):965-979. 67. Arachchilage APW, Wang F, Feyer V, Plekan O and Prince KC: Correlation of electronic structures of three cyclic dipeptides with their photoemission spectra. The Journal of Chemical Physics 2010, 133(17):174319-174310. 68. Lesarri A, Cocinero EJ, López JC and Alonso JL: The Shape of Neutral Valine. Angewandte Chemie International Edition 2004, 43 (5):605-610. 69. Jensen JH and Gordon MS: Conformational potential energy surface of glycine: a theoretical study. Journal of the American Chemical Society 1991, 113(21):7917- 7924. 70. Hu CH, Shen M and Schaefer HF: Glycine conformational analysis. Journal of the American Chemical Society 1993, 115(7):2923-2929. 71. Herrera B, Dolgounitcheva O, Zakrzewski VG, Toro-Labbe A and Ortiz JV: Conformational Effects on Glycine Ionization Energies and Dyson Orbitals. The Journal of Physical Chemistry A 2004, 108(52):11703-11708. 72. Császár AG: On the structures of free glycine and [alpha]-alanine. Journal of Molecular Structure 1995, 346:141-152. 73. Shirazian S and Gronert S: The gas-phase conformations of valine: an ab initio study. Journal of Molecular Structure: THEOCHEM 1997, 397(1-3):107-112. 74. Cocinero EJ, Lesarri A, Grabow J-U, López JC and Alonso JL: The Shape of Leucine in the Gas Phase. ChemPhysChem 2007, 8(4):599-604. 75. Lesarri A, Sánchez R, Cocinero EJ, López JC and Alonso JL: Coded Amino Acids in Gas Phase: The Shape of Isoleucine. Journal of the American Chemical Society 2005, 127(37):12952-12956. 76. Stepanian SG, Reva ID, Radchenko ED and Adamowicz L: Conformational Behavior of L-Alanine. Matrix-Isolation Infrared and Theoretical DFT and ab Initio Study. The Journal of Physical Chemistry A 1998, 102(24):4623-4629. 77. Stepanian SG, Reva ID, Radchenko ED, Rosado MTS, Duarte MLTS, Fausto R and Adamowicz L: Matrix-Isolation Infrared and Theoretical Studies of the Glycine Conformers. The Journal of Physical Chemistry A 1998, 102(6):1041-1054. 78. Stepanian SG, Reva ID, Radchenko ED and Adamowicz L: Combined Matrix- Isolation Infrared and Theoretical DFT and ab Initio Study of the Nonionized Valine Conformers. The Journal of Physical Chemistry A 1999, 103(22):4404-4412. 79. Ahemd M, Ganesan A, Wang F, Prince K, Feyer V and Plekan O: Comparative investigations on the electronic structures of beta-lactam and l-alanine. (To be submitted, 2012). 80. Hu Y and Bernstein ER: Vibrational and photoionization spectroscopy of biomolecules: Aliphatic amino acid structures. The Journal of Chemical Physics 2008, 128(16):164311-164310.

198

References

81. Blanco S, Lesarri A, López JC and Alonso JL: The Gas-Phase Structure of Alanine. Journal of the American Chemical Society 2004, 126(37):11675-11683. 82. Stepanian SG, Reva ID, Radchenko ED and Adamowicz L: Conformers of Nonionized Proline. Matrix-Isolation Infrared and Post-Hartree−Fock ab Initio Study. The Journal of Physical Chemistry A 2001, 105(47):10664-10672. 83. Snoek LC, Robertson EG, Kroemer RT and Simons JP: Conformational landscapes in amino acids: infrared and ultraviolet ion-dip spectroscopy of phenylalanine in the gas phase. Chemical Physics Letters 2000, 321(1 -2):49-56. 84. Godfrey PD and Brown RD: Shape of Glycine. Journal of the American Chemical Society 1995, 117(7):2019-2023. 85. Feyer V, Plekan O, Richter R, Coreno M, Prince KC and Carravetta V: Core Level Study of Alanine and Threonine. The Journal of Physical Chemistry A 2008, 112(34):7806-7815. 86. de Vries MS and Hobza P: Gas-Phase Spectroscopy of Biomolecular Building Blocks. Annual Review of Physical Chemistry 2007, 58(1):585-612. 87. van der Kamp MW, Shaw KE, Woods CJ and Mulholland AJ: Biomolecular simulation and modelling: status, progress and prospects. Journal of The Royal Society Interface 2008, 5 (Suppl 3):173-190. 88. Froger A, Thomas D, Delamarche C and Tallur B: Prediction of functional residues in water channels and related proteins. Protein Science 1998, 7(6):1458-1468. 89. Conte AM, Ippoliti E, Del Sole R, Carloni P and Pulci O: Many-body meets QM/MM: Application to indole in water solution. physica status solidi (b) 2010, 247(8):1920-1924. 90. Jayaram B and Jain T: The role of water in protein-DNA recognition Annual Review of Biophysics and Biomolecular Structure 2004, 33(1):343-361. 91. Gallivan JP and Dougherty DA: Cation-π interactions in structural biology. Proceedings of the National Academy of Sciences 1999, 96(17):9459-9464. 92. Gallivan JP and Dougherty DA: A Computational Study of Cation−π Interactions vs Salt Bridges in Aqueous Media: Implications for Protein Engineering. Journal of the American Chemical Society 2000, 122(5):870-874. 93. Dougherty DA: Cation-π Interactions in Chemistry and Biology: A New View of Benzene, Phe, Tyr, and Trp. Science 1996, 271(5246):163-168. 94. Spezia R, Tournois G, Tortajada J, Cartailler T and Gaigeot M-P: Toward a DFT- based molecular dynamics description of Co(ii) binding in sulfur-rich peptides. Physical Chemistry Chemical Physics 2006, 8(17):2040-2050. 95. Joyce GF: Directed evolution of nucleic acid enzymes. Annual Review of Biochemistry 2004, 73(1):791-836. 96. Rodziewicz P and Doltsinis NL: Ab Initio Molecular Dynamics Free-Energy Study of Microhydration Effects on the Neutral–Zwitterion Equilibrium of Phenylalanine. ChemPhysChem 2007, 8(13):1959-1968. 97. Wincel H: Hydration Energies of Sodiated Amino Acids from Gas-Phase Equilibria Determinations. The Journal of Physical Chemistry A 2007, 111(26):5784-5791. 98. Marx D and Hutter J: Ab Initio Molecular Dynamics: Basic Theory and Advanced Methods. In.: Cambridge University Press, Cambridge; 2009: 578 99. Tuckerman ME: Ab initio molecular dynamics: basic concepts, current trends and novel applications Journal of Physics: Condensed Matter 2002, 14. 100. Car R and Parrinello M: Unified Approach for Molecular Dynamics and Density- Functional Theory. Physical Review Letters 1985, 55(22):2471-2474.

199

References

101. Ganesan A, Brunger M and Wang F: A study of aliphatic amino acids using simulated vibrational circular dichroism and Raman optical activity spectra. (submitted, 2011). 102. Martin R: Electronic structure : basic theory and practical methods. 103. Sherrill: Frontiers in electronic structure theory. The Journal of Chemical Physics 2010, 132(11):110902. 104. Cramer C: Essentials of computational chemistry : theories and models. 105. Jensen F: Introduction to Computational Chemistry; 2007. 106. Ramachandran KI, Gopakumar D and Namboori K: Computational chemistry and molecular modeling; 2008. 107. Young DC: Computational Chemistry; 2001. 108. Schrödinger E: An Undulatory Theory of the Mechanics of Atoms and Molecules. Physical Review 1926, 28 (6):1049-1070. 109. Nakashima H and Nakatsuji H: Solving the Schr[o-umlaut]dinger equation for helium atom and its isoelectronic ions with the free iterative complement interaction (ICI) method. The Journal of Chemical Physics 2007, 127(22):224104- 224114. 110. Wang Y, Deng C and Feng D: Solutions of the Schrödinger equations for lithium and excited helium (2 ^{1}S) atoms with a correlation-function hyperspherical harmonic and generalized Laguerre-function expansion method. Physical Review A 1995, 51(1):73-78. 111. Born M and Oppenheimer R: Zur Quantentheorie der Molekeln. Annalen der Physik 1927 (German Version), 389(20):457-484. 112. Butler LJ: CHEMICAL REACTION DYNAMICS BEYOND THE BORN- OPPENHEIMER APPROXIMATION. Annual Review of Physical Chemistry 1998, 49(1):125-171. 113. Dirac PAM: Quantum Mechanics of Many-Electron Systems. Proceedings of the Royal Society of London Series A 1929, 123(792):714-733. 114. Slater JC: The Theory of Complex Spectra. Physical Review 1929, 34(10):1293- 1322. 115. Hehre W: A Guide to Molecular Mechanics and Quantum Chemical Calculations: Wavefunction; 2003. 116. Sherrill CD and Schaefer Iii HF: The Configuration Interaction Method: Advances in Highly Correlated Approaches. In: Advances in Quantum Chemistry. Edited by Per-Olov Löwdin JRSMCZ, Erkki B, vol. Volume 34: Academic Press; 1999: 143- 269. 117. Bartlett RJ: Many-Body Perturbation Theory and Coupled Cluster Theory for Electron Correlation in Molecules. Annual Review of Physical Chemistry 1981, 32(1):359-401. 118. Parr RG and Weitao RGY: Density-Functional Theory of Atoms and Molecules. In. New York: Oxford University Press (US); 1995. 119. Hohenberg P and Kohn W: Inhomogeneous Electron Gas. Physical Review 1964, 136(3B):B864-B871. 120. Cohen AJ, Mori-Sánchez P and Yang W: Challenges for Density Functional Theory. Chemical Reviews 2011, 112(1):289-320. 121. Kohn W and Sham LJ: Self-Consistent Equations Including Exchange and Correlation Effects. Physical Review 1965, 140(4A):A1133-A1138. 122. Wang F, Brunger MJ and Larkins FP: Valence Orbital Electron Momentum Spectroscopy For N2O. The Journal of Physical Chemistry A 2001, 105(8):1254- 1259.

200

References

123. Wang F, Duffy P and Chong DP: Valence orbital momentum distributions of water: the performance of the hf, B3LYP, BP86 and VWN models combined with selected gaussian and slater basis sets. . In: Nanoscale interactions and their applications: Essays in Honour of Ian McCarthy Edited by Wang F, Brunger M: Research Signpost, Kerala, India; 2007: 169-182. . 124. Sousa SF, Fernandes PA and Ramos MJ: General Performance of Density Functionals. The Journal of Physical Chemistry A 2007, 111(42):10439-10452. 125. Wang F, Pang W and Duffy P: Performance assessment of density functional theory-based models using orbital momentum distributions. Molecular Simulation 2012, 38(6):468-480. 126. Dirac PAM: Note on Exchange Phenomena in the Thomas Atom. Mathematical Proceedings of the Cambridge Philosophical Society 1930, 26(03):376-385. 127. Fermi E: A statistical method for the determination of some atomic properties and the application of this methods to the theory of the periodic system of elements. Z Phys 1928, 48 :73-79. 128. Thomas LH: The calculation of atomic fields. Mathematical Proceedings of the Cambridge Philosophical Society 1927, 23(05):542-548. 129. Slater JC: A Simplification of the Hartree-Fock Method. Physical Review 1951, 81(3):385-390. 130. Vosko SH, Wilk L and Nusair M: Accurate spin-dependent electron liquid correlation energies for local spin density calculations: a critical analysis. Canadian Journal of Physics 1980, 58(8):1200-1211. 131. Perdew JP and Zunger A: Self-interaction correction to density-functional approximations for many-electron systems. Physical Review B 1981, 23(10):5048- 5079. 132. Giese TJ and York DM: Density-functional expansion methods: Evaluation of LDA, GGA, and meta-GGA functionals and different integral approximations. The Journal of Chemical Physics 2010, 133(24):244107-244110. 133. Becke AD: Correlation energy of an inhomogeneous electron gas: A coordinate- space model. The Journal of Chemical Physics 1988, 88(2):1053-1062. 134. Perdew JP: Density-functional approximation for the correlation energy of the inhomogeneous electron gas. Physical Review B 1986, 33(12):8822-8824. 135. Becke AD: Density functional calculations of molecular bond energies. The Journal of Chemical Physics 1986, 84(8):4524-4529. 136. Perdew JP and Wang Y: Pair-distribution function and its coupling-constant average for the spin-polarized electron gas. Physical Review B 1992, 46(20):12947- 12954. 137. Engel E: Orbital-Dependent Functionals for the Exchange-Correlation Energy: A Third Generation of Density Functionals. In: A Primer in Density Functional Theory. Edited by Fiolhais C, Nogueira F, Marques M, vol. 620: Springer-Verlag Berlin Heidelberg; 2003: 56-122. 138. Becke AD: A new mixing of Hartree--Fock and local density-functional theories. The Journal of Chemical Physics 1993, 98(2):1372-1377. 139. Becke AD: Density-functional exchange-energy approximation with correct asymptotic behavior. Physical Review A 1988, 38(6):3098-3100. 140. Lee C, Yang W and Parr RG: Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. Physical Review B 1988, 37(2):785-789.

201

References

141. Tirado-Rives J and Jorgensen WL: Performance of B3LYP Density Functional Methods for a Large Set of Organic Molecules. Journal of Chemical Theory and Computation 2008, 4(2):297-306. 142. Gritsenko OV, Schipper PRT and Baerends EJ: Approximation of the exchange- correlation Kohn–Sham potential with a statistical average of different orbital model potentials. Chemical Physics Letters 1999, 302(3–4):199-207. 143. van Leeuwen R and Baerends EJ: Exchange-correlation potential with correct asymptotic behavior. Physical Review A 1994, 49(4):2421. 144. Ganesan A, Brunger M and Wang F: Influence of functional groups on the Cα–Cβ chain of l-phenylalanine and its derivatives. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 619 (1-3):143-146. 145. Slater JC: Atomic Shielding Constants. Physical Review 1930, 36(1):57-64. 146. G e ll M, Luis JM, Sol M and Swart M: Importance of the Basis Set for the Spin- State Energetics of Iron Complexes. The Journal of Physical Chemistry A 2008, 112(28):6384-6391. 147. Chong DP, Van Lenthe E, Van Gisbergen S and Baerends EJ: Even-tempered slater- type orbitals revisited: From hydrogen to krypton. Journal of Computational Chemistry 2004, 25(8):1030-1036. 148. te Velde G, Bickelhaupt FM, Baerends EJ, Fonseca Guerra C, van Gisbergen SJA, Snijders JG and Ziegler T: Chemistry with ADF. Journal of Computational Chemistry 2001, 22(9):931-967. 149. Boys SF: Electronic Wave Functions. I. A General Method of Calculation for the Stationary States of Any Molecular System. Proceedings of the Royal Society of London Series A Mathematical and Physical Sciences 1950, 200 (1063):542-554. 150. Rassolov VA, Pople JA, Ratner MA and Windus TL: 6-31G[sup *] basis set for atoms K through Zn. The Journal of Chemical Physics 1998, 109(4):1223-1229. 151. Frisch MJ, Pople JA and Binkley JS: Self-consistent molecular orbital methods 25. Supplementary functions for Gaussian basis sets. The Journal of Chemical Physics 1984, 80(7 ):3265-3269. 152. Francl MM, Pietro WJ, Hehre WJ, Binkley JS, Gordon MS, DeFrees DJ and Pople JA: Self-consistent molecular orbital methods. XXIII. A polarization-type basis set for second-row elements. The Journal of Chemical Physics 1982, 77(7):3654- 3665. 153. Kendall RA, Dunning JTH and Harrison RJ: Electron affinities of the first-row atoms revisited. Systematic basis sets and wave functions. The Journal of Chemical Physics 1992, 96(9):6796-6806. 154. Dunning JTH: Gaussian basis sets for use in correlated molecular calculations. I. The atoms boron through neon and hydrogen. The Journal of Chemical Physics 1989, 90(2):1007-1023. 155. Woon DE and Dunning JTH: Gaussian basis sets for use in correlated molecular calculations. III. The atoms aluminum through argon. The Journal of Chemical Physics 1993, 98(2):1358-1371. 156. Dunning JTH, Peterson KA and Wilson AK: Gaussian basis sets for use in correlated molecular calculations. X. The atoms aluminum through argon revisited. The Journal of Chemical Physics 2001, 114(21):9244-9253. 157. Errol L and Lewars E: Computational chemistry introduction to the theory and applications of molecular and quantum mechanics. In. Edited by ebrary I. Boston: Boston : Kluwer Academic; 2003.

202

References

158. Nelder JA and Mead R: A simplex method for function minimization. The Computer Journal 1965, 7(4):308-313. 159. Jiang H and Yang W: Conjugate-gradient optimization method for orbital-free density functional calculations. The Journal of Chemical Physics 2004, 121(5):2030-2036. 160. Banerjee A, Adams N, Simons J and Shepard R: Search for stationary points on surfaces. The Journal of Physical Chemistry 1985, 89(1):52-57. 161. Hamilton TP and Pulay P: Direct inversion in the iterative subspace (DIIS) optimization of open-shell, excited-state, and small multiconfiguration SCF wave functions. The Journal of Chemical Physics 1986, 84(10):5728-5734. 162. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Montgomery JA, Vreven T, Kudin KN, Burant JC, Millam JM, Iyengar SS, Tomasi J, Barone V, Mennucci B, Cossi M, Scalmani G, Rega N, Petersson GA, Nakatsuji H, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Klene M, Li X, Knox JE, Hratchian HP, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Ayala PY, Morokuma K, Voth GA, Salvador P, Dannenberg JJ, Zakrzewski VG, Dapprich S, Daniels AD, Strain MC, Farkas O, Malick DK, Rabuck AD, Raghavachari K, Foresman JB, Ortiz JV, Cui Q, Baboul AG, Clifford S, Cioslowski J, Stefanov BB, Liu G, Liashenko A, Piskorz P, Komaromi I, Martin RL, Fox DJ, Keith T, Laham A, Peng CY, Nanayakkara A, Challacombe M, Gill PMW, Johnson B, Chen W, Wong MW, Gonzalez C and Pople JA: Gaussian 03, Revision C.02. In.; 2003. 163. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Scalmani G, Barone V, Mennucci B, Petersson GA, Nakatsuji H, Caricato M, Li X, Hratchian HP, Izmaylov AF, Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Vreven T, Montgomery JA, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E, Kudin KN, Staroverov VN, Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC, Iyengar SS, Tomasi J, Cossi M, Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth GA, Salvador P, Dannenberg JJ, Dapprich S, Daniels AD, Farkas, Foresman JB, Ortiz JV, Cioslowski J and Fox DJ: Gaussian 09, Revision B.01. In. Wallingford CT; 2009. 164. Angermund K, Claus KH, Goddard R and Krüger C: High-Resolution X-ray Crystallography—An Experimental Method for the Description of Chemical Bonds. Angewandte Chemie International Edition in English 1985, 24(4):237-247. 165. Tu G, Carravetta V, Vahtras O and Agren H: Core ionization potentials from self- interaction corrected Kohn-Sham orbital energies. The Journal of Chemical Physics 2007, 127(17):174110-174111. 166. Carlson TA: Photoelectron Spectroscopy. Annual Review of Physical Chemistry 1975, 26(1):211-234. 167. Fadley CS: X-ray photoelectron spectroscopy: Progress and perspectives. Journal of Electron Spectroscopy and Related Phenomena 2010, 178–179 (0):2-32. 168. Robinson H and Rawlinson WF: XXXIII. The magnetic spectrum of the β rays excited in metals by soft X rays. Philosophical Magazine Series 6 1914, 28(164):277-281. 169. Koopmans T: Über die Zuordnung von Wellenfunktionen und Eigenwerten zu den Einzelnen Elektronen Eines Atoms. Physica 1934, 1(1–6):104-113.

203

References

170. Cavigliasso G and Chong DP: Accurate density-functional calculation of core- electron binding energies by a total-energy difference approach. The Journal of Chemical Physics 1999, 111(21):9485-9492. 171. Perdew JP and Yue W: Accurate and simple density functional for the electronic exchange energy: Generalized gradient approximation. Physical Review B 1986, 33(12):8800-8802. 172. Chong DP, Aplincourt P and Bureau C: DFT Calculations of Core−Electron Binding Energies of the Peptide Bond. The Journal of Physical Chemistry A 2001, 106(2):356-362. 173. Janak JF: Proof that ∂E/∂n_{i}=ε in density-functional theory. Physical Review B 1978, 18(12):7165-7168. 174. Chong DP, Gritsenko OV and Baerends EJ: Interpretation of the Kohn--Sham orbital energies as approximate vertical ionization potentials. The Journal of Chemical Physics 2002, 116(5):1760-1772. 175. Gritsenko OV, Braida B and Baerends EJ: Physical interpretation and evaluation of the Kohn--Sham and Dyson components of the epsilon--I relations between the Kohn--Sham orbital energies and the ionization potentials. The Journal of Chemical Physics 2003, 119(4):1937-1950. 176. Cederbaum LS and Domcke W: Theoretical Aspects of Ionization Potentials and Photoelectron Spectroscopy: A Green's Function Approach. In: Advances in Chemical Physics. John Wiley & Sons, Inc.; 2007: 205-344. 177. Cederbaum LS, Domcke W, Schirmer J, von Niessen W, Diercksen GHF and Kraemer WP: Correlation effects in the ionization of hydrocarbons. The Journal of Chemical Physics 1978, 69(4):1591-1603. 178. Vanquelef E, Simon S, Marquant G, Garcia E, Klimerak G, Delepine JC, Cieplak P and Dupradeau F-Y: R.E.D. Server: a web service for deriving RESP and ESP charges and building force field libraries for new molecules and molecular fragments. Nucleic Acids Research 2011. 179. Mulliken RS: Electronic Population Analysis on LCAO[Single Bond]MO Molecular Wave Functions. I. The Journal of Chemical Physics 1955, 23(10):1833- 1840. 180. Fonseca Guerra C, Handgraaf J-W, Baerends EJ and Bickelhaupt FM: Voronoi deformation density (VDD) charges: Assessment of the Mulliken, Bader, Hirshfeld, Weinhold, and VDD methods for charge analysis. Journal of Computational Chemistry 2004, 25(2):189-210. 181. Bader RFW and Nguyen-Dang TT: Quantum Theory of Atoms in Molecules– Dalton Revisited. In: Advances in Quantum Chemistry. Edited by Per-Olov L, vol. Volume 14: Academic Press; 1981: 63-124. 182. Singh UC and Kollman PA: An approach to computing electrostatic charges for molecules. Journal of Computational Chemistry 1984, 5(2):129-145. 183. Breneman CM and Wiberg KB: Determining atom-centered monopoles from molecular electrostatic potentials. The need for high sampling density in formamide conformational analysis. Journal of Computational Chemistry 1990, 11(3):361-373. 184. Bayly CI, Cieplak P, Cornell W and Kollman PA: A well-behaved electrostatic potential based method using charge restraints for deriving atomic charges: the RESP model. The Journal of Physical Chemistry 1993, 97(40):10269-10280. 185. Cornell WD, Cieplak P, Bayly CI and Kollmann PA: Application of RESP charges to calculate conformational energies, hydrogen bond energies, and free energies of solvation. Journal of the American Chemical Society 1993, 115 (21):9620-9631.

204

References

186. Hirshfeld FL: Bonded-atom fragments for describing molecular charge densities. Theoretical Chemistry Accounts: Theory, Computation, and Modeling (Theoretica Chimica Acta) 1977, 44(2):129-138. 187. Bachrach SM: Population Analysis and Electron Densities from Quantum Mechanics. In: Reviews in Computational Chemistry. John Wiley & Sons, Inc.; 2007: 171-228. 188. Saha S, Roy RK and Ayers PW: Are the Hirshfeld and Mulliken population analysis schemes consistent with chemical intuition? International Journal of Quantum Chemistry 2009, 109(9):1790-1806. 189. Moss G and Fell D: Electrostatic molecular interaction from X-ray diffraction data. I. Development of the method; test on pyrazine. Acta Crystallographica Section A 1981, 37(3):414-421. 190. Weigold E: Electron momentum spectroscopy. 191. McCarthy IE and Weigold E: Electron momentum spectroscopy of atoms and molecules. Reports on Progress in Physics 1991, 54(6):789. 192. Brunger MJ and Adcock W: High-resolution electron momentum spectroscopy of molecules. Journal of the Chemical Society, Perkin Transactions 2 2002(1):1-22. 193. Wang F: Assessment of Quantum Mechanical Models Based on Resolved Orbital Momentum Distributions of n-Butane in the Outer Valence Shell. The Journal of Physical Chemistry A 2003, 107(47):10199-10207. 194. Ning CG, Hajgató B, Huang YR, Zhang SF, Liu K, Luo ZH, Knippenberg S, Deng JK and Deleuze MS: High resolution electron momentum spectroscopy of the valence orbitals of water. Chemical Physics 2008, 343(1):19-30. 195. Herzberg G: Molecular Spectra and Molecular Structure II. Infrared and Raman Spectra of Polyatomic Molecules; 1945. 196. Wilson EB: Molecular Vibrations; 1955. 197. Long DA: Raman Spectroscopy; 1977. 198. Larkin and Larkin P: Infrared and Raman Spectroscopy; 2011. 199. Wartewig S and Wartewig: Pharmaceutical applications of Mid-IR and Raman spectroscopy. Advanced drug delivery reviews 2005, 57(8):1144-1170. 200. Buckingham AD: Vibrational spectroscopy. Nature 1976, 263 (5580):803-803. 201. Workman J: Handbook of Vibrational Spectroscopy. Spectroscopy 2002, 17(1):50. 202. Nafie LA: Vibrational Optical Activity : Principles and Applications. In., 1 edn. Hoboken: Wiley; 2011. 203. Coates J: Interpretation of Infrared Spectra, A Practical Approach. Encyclopedia of Analytical Chemistry. 204. Bondesson L, Mikkelsen KV, Luo Y, Garberg P and Ågren H: Hydrogen bonding effects on infrared and Raman spectra of drug molecules. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 2007, 66(2):213-224. 205. Meier RJ: Calculating the vibrational spectra of molecules: An introduction for experimentalists with contemporary examples. Vibrational Spectroscopy 2007, 43(1):26-37. 206. Meier RJ: Vibrational spectroscopy: a 'vanishing' discipline? Chemical Society Reviews 2005, 34(9):743-752. 207. Gardiner: Practical Raman spectroscopy. 208. Raman CV: Anisotropy of Molecules. Nature 1922, 109(2725):75-76. 209. Raman CV: A New Type of Secondary Radiation. Nature 1928, 121(3048):501- 502. 210. Whyte LL: Chirality. Nature 1957, 180 (4584):513-513.

205

References

211. Capozziello S and Lattanzi A: Spiral Galaxies as Chiral Objects? Astrophysics and Space Science 2006, 301 (1):189-193. 212. Capozziello S and Lattanzi A: Spiral galaxies as enantiomers: Chirality, an underlying feature in chemistry and astrophysics. Chirality 2006, 18(1):17-23. 213. Katzenelson O, Hel-Or HZ and Avnir D: Chirality of Large Random Supramolecular Structures. Chemistry – A European Journal 1996, 2(2):174-181. 214. Wong K-P: Optical rotary dispersion and circular dichroism. Journal of Chemical Education 1974, 51(12):A573. 215. Emeis CA and Oosterhoff LJ: Emission of circularly-polarised radiation by optically-active compounds. Chemical Physics Letters 1967, 1(4):129-132. 216. Emeis CA and Oosterhoff LJ: The n-pi * Absorption and Emission of Optically Active trans-beta -Hydrindanone and trans-beta -Thiohydrindanone. The Journal of Chemical Physics 1971, 54(11):4809-4819. 217. Turner DH, Tinoco I and Maestre M: Fluorescence detected circular dichroism. Journal of the American Chemical Society 1974, 96 (13):4340-4342. 218. Urry DW and Eyring H: Optical Rotatory Dispersion Studies of L-Histidine Chelation. Journal of the American Chemical Society 1964, 86(21):4574-4580. 219. Konopelski JP, Sundararaman P, Barth G and Djerassi C: Optical rotatory dispersion studies. 128. Octant contributions of methyl groups in 4-tert- butylcyclohexanones. Journal of the American Chemical Society 1980, 102(8):2737- 2745. 220. Barron LD and Buckingham AD: Vibrational optical activity. Chemical Physics Letters 2010, 492 (4-6):199-213. 221. Barron LD: Molecular Light Scattering and Optical Activity: Cambridge University Press; 2004. 222. Nafie LA: Vibrational circular dichroism. Journal of the American Chemical Society 1976, 98 (10):2715-2723. 223. Abdali S: Conformational determination of [Leu]enkephalin based on theoretical and experimental VA and VCD spectral analyses. Physical chemistry chemical physics 2004, 6(9):2434-2439. 224. Nafie LA: Vibrational Optical Activity. Applied Spectroscopy 1996, 50(5):14A- 26A. 225. Nafie LA, Yu G-S, Qu X and Freedman TB: Comparison of IR and Raman forms of vibrational optical activity. Faraday Discussions 1994, 99:13-34. 226. Nafie LA: Infrared and Raman vibrational optical activity: Theoretical and Experimental Aspects. Annual Review of Physical Chemistry 1997, 48(1):357-386. 227. Nafie LA: Circular polarization spectroscopy of chiral molecules. Journal of Molecular Structure 1995, 347(0):83-100. 228. Nieto-Ortega Bn, Casado J, Blanch EW, L p ez Navarrete JT, Quesada AR and Ram rez FJ: Raman Optical Activity Spectra and Conformational Elucidation of Chiral Drugs. The Case of the Antiangiogenic Aeroplysinin-1 . The Journal of Physical Chemistry A 2011, 115(13):2752-2755. 229. Danecek P, Kapitan J, Baumruk V, Bednarova L, Kopecky V and Bour P: Anharmonic effects in IR, Raman, and Raman optical activity spectra of alanine and proline zwitterions. Journal of Chemical Physics 2007, 126(22):224513. 230. Cossi M, Rega N, Scalmani G and Barone V: Energies, structures, and electronic properties of molecules in solution with the C-PCM solvation model. Journal of Computational Chemistry 2003, 24(6):669-681.

206

References

231. Sebastiani D and Röthlisberger U: Advances in Density-functional-based Modeling Techniques – Recent Extensions of the Car-Parrinello Approach. In: Quantum Medicinal Chemistry. Wiley-VCH Verlag GmbH & Co. KGaA; 2005: 3-39. 232. Carloni, Carloni P and Alber FU: Quantum Medicinal Chemistry; 2003. 233. Pickett WE: Pseudopotential methods in condensed matter applications. Computer Physics Reports 1989, 9(3):115-197. 234. Fuchs M and Scheffler M: Ab initio pseudopotentials for electronic structure calculations of poly-atomic systems using density-functional theory. Computer Physics Communications 1999, 119(1):67-98. 235. Kerker GP: Non-singular atomic pseudopotentials for solid state applications. Journal of Physics C: Solid State Physics 1980, 13(9):L189. 236. Bachelet GB, Hamann DR and Schlüter M: Pseudopotentials that work: From H to Pu. Physical Review B 1982, 26(8):4199-4228. 237. Troullier N and Martins JL: Efficient pseudopotentials for plane-wave calculations. Physical Review B 1991, 43(3):1993-2006. 238. Gonze X, Stumpf R and Scheffler M: Analysis of separable potentials. Physical Review B 1991, 44(16):8503-8513. 239. Vanderbilt D: Optimally smooth norm-conserving pseudopotentials. Physical Review B 1985, 32(12):8412-8415. 240. Lin JS, Qteish A, Payne MC and Heine V: Optimized and transferable nonlocal separable ab initio pseudopotentials. Physical Review B 1993, 47(8):4174-4180. 241. Goedecker S and Maschke K: Transferability of pseudopotentials. Physical Review A 1992, 45(1):88-93. 242. Goedecker S, Teter M and Hutter J: Separable dual-space Gaussian pseudopotentials. Physical Review B 1996, 54(3):1703-1710. 243. Marx D and Hutter J: Ab Initio Molecular Dynamics : Basic Theory and Advanced Methods. In. Cambridge: Cambridge University Press; 2009. 244. Bashford D and Case DA: GENERALIZED BORN MODELS OF MACROMOLECULAR SOLVATION EFFECTS. Annual Review of Physical Chemistry 2000, 51(1):129-152. 245. Onufriev A: Chapter 7 Implicit Solvent Models in Molecular Dynamics Simulations: A Brief Overview. In: Annual Reports in Computational Chemistry. Edited by Ralph AW, David CS, vol. Volume 4: Elsevier; 2008: 125-137. 246. Cramer CJ and Truhlar DG: Implicit Solvation Models: Equilibria, Structure, Spectra, and Dynamics. Chemical Reviews 1999, 99(8):2161-2200. 247. Tomasi J and Persico M: Molecular Interactions in Solution: An Overview of Methods Based on Continuous Distributions of the Solvent. Chemical Reviews 1994, 94(7):2027-2094. 248. Sinnecker S, Rajendran A, Klamt A, Diedenhofen M and Neese F: Calculation of Solvent Shifts on Electronic g-Tensors with the Conductor-Like Screening Model (COSMO) and Its Self-Consistent Generalization to Real Solvents (Direct COSMO-RS). The Journal of Physical Chemistry A 2006, 110 (6):2235-2245. 249. Klamt A: Conductor-like Screening Model for Real Solvents: A New Approach to the Quantitative Calculation of Solvation Phenomena. The Journal of Physical Chemistry 1995, 99(7):2224-2235. 250. Beroza P and Case DA: Calculations of proton-binding thermodynamics in proteins. In., vol. 295; 1998: 170-189. 251. Honig B, Sharp K and Yang AS: Macroscopic models of aqueous solutions: Biological and chemical applications. Journal of Physical Chemistry 1993, 97(6):1101-1109.

207

References

252. Miertuš S, Scrocco E and Tomasi J: Electrostatic interaction of a solute with a continuum. A direct utilizaion of AB initio molecular potentials for the prevision of solvent effects. Chemical Physics 1981, 55(1):117-129. 253. Foresman JB, Keith TA, Wiberg KB, Snoonian J and Frisch MJ: Solvent Effects. 5. Influence of Cavity Shape, Truncation of Electrostatics, and Electron Correlation on ab Initio Reaction Field Calculations. The Journal of Physical Chemistry 1996, 100(40):16098-16104. 254. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Scalmani G, Barone V, Mennucci B, Petersson GA, Nakatsuji H, Caricato M, Li X, Hratchian HP, Izmaylov AF, Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Vreven T, Montgomery JA, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E, Kudin KN, Staroverov VN, Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC, Iyengar SS, Tomasi J, Cossi M, Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth GA, Salvador P, Dannenberg JJ, Dapprich S, Daniels AD, Farkas, Foresman JB, Ortiz JV, Cioslowski J and Fox DJ: Gaussian 09, Revision A.02. In. Wallingford CT; 2009. 255. Dennington RT, Keith T and Millam J: GaussView, Version 5, Semichem Inc, Shawnee Mission KS, 2009. 256. Schaftenaar G and Noordik JH: Molden: a pre- and post-processing program for molecular and electronic structures*. Journal of Computer-Aided Molecular Design 2000, 14(2):123-134. 257. Chemcraft: http://wwwchemcraftprogcom. 258. Origin: (http://wwworiginlabcom/) 2009. 259. Humphrey W, Dalke A and Schulten K: VMD: Visual molecular dynamics. Journal of Molecular Graphics 1996, 14(1):33-38. 260. Ji Z, Santamaria Rn and Garz n IL: Vibrational Circular Dichroism and IR Absorption Spectra of Amino Acids: A Density Functional Study. The Journal of Physical Chemistry A 2010, 114(10):3591-3601. 261. DaneCek P, Kapitan J, Baumruk V, Bednarova L, Kopecky JV and Bour P: Anharmonic effects in IR, Raman, and Raman optical activity spectra of alanine and proline zwitterions. The Journal of Chemical Physics 2007, 126(22):224513- 224513. 262. Derbel N, Hernandez B, Pfluger F, Liquier J, Geinguenaud F, Jaiidane N, Ben Lakhdar Z and Ghomi M: Vibrational Analysis of Amino Acids and Short Peptides in Hydrated Media. I. L-glycine and L-leucine. The Journal of Physical Chemistry B 2007, 111(6):1470-1477. 263. Gontrani L, Mennucci B and Tomasi J: Glycine and alanine: a theoretical study of solvent effects upon energetics and molecular response properties. Journal of Molecular Structure: THEOCHEM 2000, 500(1 -3):113-127. 264. Tulip PR and Clark SJ: Dielectric and vibrational properties of amino acids. The Journal of Chemical Physics 2004, 121(11):5201-5210. 265. Linder R, Nispel M, Häber T and Kleinermanns K: Gas-phase FT-IR-spectra of natural amino acids. Chemical Physics Letters 2005, 409(4 -6):260-264. 266. Kumar S, Rai AK, Singh VB and Rai SB: Vibrational spectrum of glycine molecule. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 2005, 61(11-12):2741-2746.

208

References

267. Kumar S, Kumar Rai A, Rai SB, Rai DK, Singh AN and Singh VB: Infrared, Raman and electronic spectra of alanine: A comparison with ab intio calculation. Journal of Molecular Structure 2006, 791 (1-3):23-29. 268. Linder R, Seefeld K, Vavra A and Kleinermanns K: Gas phase infrared spectra of nonaromatic amino acids. Chemical Physics Letters 2008, 453(1-3):1-6. 269. Sabino AS, De Sousa GP, Luz-Lima C, Freire PTC, Melo FEA and Mendes Filho J: High-pressure Raman spectra of L-isoleucine crystals. Solid State Communications 2009, 149(37-38):1553-1556. 270. Chen F-F and Wang F: Electronic Structure of the Azide Group in 3Â¢-Azido- 3Â¢-deoxythymidine (AZT) Compared to Small Azide Compounds. Molecules 2009, 14(7):2656-2668. 271. Klasinc L: Application of photoelectron spectroscopy to biologically active molecules and their constituent parts. III. Amino acids. Journal of Electron Spectroscopy and Related Phenomena 1976, 8(2):161-164. 272. Powis I, Rennie EE, Hergenhahn U, Kugeler O and Bussy-Socrate R: Investigation of the Gas-Phase Amino Acid Alanine by Synchrotron Radiation Photoelectron Spectroscopy. The Journal of Physical Chemistry A 2003, 107(1):25-34. 273. Cannington PH and Ham NS: He(I) and He(II) photoelectron spectra of glycine and related molecules. Journal of Electron Spectroscopy and Related Phenomena 1983, 32(2):139-151. 274. Dehareng D and Dive G: Vertical Ionization Energies of L-Amino Acids as a Function of Their Conformation: an Ab Initio Study. International Journal of Molecular Sciences 2004, 5(11):301-332. 275. Hern ndez B, Pfl ger F, Derbel N, De Coninck Jl and Ghomi M: Vibrational Analysis of Amino Acids and Short Peptides in Hydrated Media. VI. Amino Acids with Positively Charged Side Chains: l-Lysine and l-Arginine. Journal of Physical Chemistry B 2009, 114(2):1077-1088. 276. Bakker JM, Aleese LM, Meijer G and von Helden G: Fingerprint IR Spectroscopy to Probe Amino Acid Conformations in the Gas Phase. Physical Review Letters 2003, 91(20):203003. 277. Compagnon I, Oomens J, Meijer G and von Helden G: Mid-Infrared Spectroscopy of Protected Peptides in the Gas Phase: A Probe of the Backbone Conformation. Journal of the American Chemical Society 2006, 128(11):3592-3597. 278. Ji Z, Santamaria R and Garz n IL: Vibrational Circular Dichroism and IR Absorption Spectra of Amino Acids: A Density Functional Study. Journal of Physical Chemistry A 2010, 114(10):3591-3601. 279. Oomens J, Polfer N, Moore DT, van der Meer L, Marshall AG, Eyler JR, Meijer G and von Helden G: Charge-state resolved mid-infrared spectroscopy of a gas- phase protein. Physical Chemistry Chemical Physics 2005, 7(7):1345-1348. 280. von Helden G, Compagnon I, Blom MN, Frankowski M, Erlekam U, Oomens J, Brauer B, Gerber RB and Meijer G: Mid-IR spectra of different conformers of phenylalanine in the gas phase. Physical Chemistry Chemical Physics 2008, 10(9):1248-1256. 281. Zhu G, Zhu X, Fan Q and Wan X: Raman spectra of amino acids and their aqueous solutions. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 2011, 78(3):1187-1195. 282. Andreas B: The infrared absorption of amino acid side chains. Progress in Biophysics and Molecular biology 2000, 74(3-5):141-173.

209

References

283. Barron LD, Gargaro AR, Hecht L and Polavarapu PL: Experimental and ab initio theoretical vibrational Raman optical activity of alanine. Spectrochimica Acta Part A: Molecular Spectroscopy 1991, 47(8):1001-1016. 284. Chakraborty D and Manogaran S: Vibrational analysis of glycine zwitterion – an ab initio study. Chemical Physics Letters 1998, 294 (1-3):56-64. 285. Gargaro AR, Barron LD and Hecht L: Vibrational Raman optical activity of simple amino acids. Journal of Raman Spectroscopy 1993, 24(2):91-96. 286. Pecul M: Theoretical simulation of the ROA spectra of neutral cysteine and serine. Chemical Physics Letters 2006, 427(1-3):166-176. 287. Oomens J, Steill JD and Redlich B: Gas-Phase IR Spectroscopy of Deprotonated Amino Acids. Journal of the American Chemical Society 2009, 131(12):4310-4319. 288. Powis I: Photoelectron Spectroscopy and Circular Dichroism in Chiral Biomolecules: l-Alanine. Journal of Physical Chemistry A 2000, 104(5):878-882. 289. Stepanian SG, Reva ID, Radchenko ED and Adamowicz L: Conformational Behavior of α-Alanine. Matrix-Isolation Infrared and Theoretical DFT and ab Initio Study. Journal of Physical Chemistry A 1998, 102(24):4623-4629. 290. Stepanian SG, Reva ID, Radchenko ED and Adamowicz L: Combined Matrix- Isolation Infrared and Theoretical DFT and ab Initio Study of the Nonionized Valine Conformers. Journal of Physical Chemistry A 1999, 103 (22):4404-4412. 291. Tajkhorshid E, Jalkanen KJ and Suhai S: Structure and Vibrational Spectra of the Zwitterion l-Alanine in the Presence of Explicit Water Molecules: A Density Functional Analysis. Journal of Physical Chemistry B 1998, 102 (30):5899-5913. 292. Yu G-S, Freedman TB, Nafie LA, Deng Z and Polavarapu PL: Experimental Measurement and Ab Initio Calculation of Raman Optical Activity of L-Alanine and Its Deuterated Isotopomers. Journal of Physical Chemistry 1995, 99(2):835- 843. 293. Dukor RK and Nafie LA: Vibrational Optical Activity of Pharmaceuticals and Biomolecules. In: Encyclopedia of Analytical Chemistry. John Wiley & Sons, Ltd; 2006. 294. Nafie LA, Oboodi MR and Freedman TB: Vibrational circular dichroism in amino acids and peptides. 8. A chirality rule for methine C*.alpha.-H stretching modes. Journal of the American Chemical Society 1983, 105 (25):7449-7450. 295. Pecul M: New applications and challenges for computational ROA spectroscopy. Chirality 2009, 21(1E):E98-E104. 296. Deplazes E, van Bronswijk W, Zhu F, Barron L, Ma S, Nafie L and Jalkanen K: A combined theoretical and experimental study of the structure and vibrational absorption, vibrational circular dichroism, Raman and Raman optical activity spectra of the <span style="font-variant:small- caps"><small>L</small></span> -histidine zwitterion. Theoretical Chemistry Accounts: Theory, Computation, and Modeling (Theoretica Chimica Acta) 2008, 119 (1):155-176. 297. Oboodi MR, Lal BB, Young DA, Freedman TB and Nafie LA: Vibrational circular dichroism in amino acids and peptides. 9. Carbon-hydrogen stretching spectra of the amino acids and selected transition-metal complexes. Journal of the American Chemical Society 1985, 107(6):1547-1556. 298. Diem M: Infrared vibrational circular dichroism of alanine in the mid-infrared region: isotopic effects. Journal of the American Chemical Society 1988, 110(21):6967-6970. 299. Diem M, Polavarapu PL, Oboodi M and Nafie LA: Vibrational circular dichroism in amino acids and peptides. 4. Vibrational analysis, assignments, and solution-

210

References

phase Raman spectra of deuterated isotopomers of alanine. Journal of the American Chemical Society 1982, 104(12):3329-3336. 300. Zhang P and Polavarapu PL: Vibrational Circular Dichroism of Matrix-Assisted Amino Acid Films in the Mid-Infrared Region. Applied Spectroscopy 2006, 60(4):378-385. 301. Godbout N, Salahub DR, Andzelm J and Wimmer E: Optimization of Gaussian- type basis sets for local spin density functional calculations. Part I. Boron through neon, optimization technique and validation. Canadian Journal of Chemistry 1992, 70(2):560-571. 302. Cederbaum LS and Domcke W: Theoretical Aspects of Ionization Potentials and Photoelectron Spectroscopy: A Green's Function Approach. Advances in Chemical Physics 2007:205. 303. Schipper PRT, Gritsenko OV, van Gisbergen SJA and Baerends EJ: Molecular calculations of excitation energies and (hyper)polarizabilities with a statistical average of orbital model exchange-correlation potentials. The Journal of Chemical Physics 2000, 112(3):1344-1352. 304. Chong DP, Lenthe EV, Gisbergen SV and Baerends EJ: Even-tempered slater-type orbitals revisited: From hydrogen to krypton. Journal of Computational Chemistry 2004, 25(8):1030-1036. 305. Cheeseman JR, Frisch MJ, Devlin FJ and Stephens PJ: Ab initio calculation of atomic axial tensors and vibrational rotational strengths using density functional theory. Chemical Physics Letters 1996, 252(3-4):211-220. 306. Stephens PJ: Theory of vibrational circular dichroism. Journal of Physical Chemistry 1985, 89(5):748-752. 307. Zvereva EE, Shagidullin AR and Katsyuba SA: Ab Initio and DFT Predictions of Infrared Intensities and Raman Activities. The Journal of Physical Chemistry A 2010, 115(1):63-69. 308. Falzon CT and Wang F: Understanding glycine conformation through molecular orbitals. Journal of Chemical Physics 2005, 123(21):214307-214312. 309. Falzon CT, Wang F and Pang W: Orbital Signatures of Methyl in l-Alanine. Journal of Physical Chemistry B 2006, 110(19):9713-9719. 310. Ganesan A and Wang F: Intramolecular interactions of L-phenylalanine revealed by inner shell chemical shift. J Chem Phys 2009, 131(4):044321. 311. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Scalmani G, Barone V, Mennucci B, Petersson GA, Nakatsuji H, Caricato M, Li X, Hratchian HP, Izmaylov AF, Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M, Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Vreven T, Montgomery JA, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E, Kudin KN, Staroverov VN, Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC, Iyengar SS, Tomasi J, Cossi M, Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V, Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth GA, Salvador P, Dannenberg JJ, Dapprich S, Daniels AD, Farkas, Foresman JB, Ortiz JV, Cioslowski J and Fox DJ: Gaussian 09, Revision A.1. Gaussian 09, Revision A1 2009 Wallingford CT, 2009. 312. Iijima K, Tanaka K and Onuma S: Main conformer of gaseous glycine: molecular structure and rotational barrier from electron diffraction data and rotational constants. Journal of Molecular Structure 1991, 246(3-4):257-266. 313. Iijima K and Beagley B: An electron diffraction study of gaseous [alpha]-alanine, NH2CHCH3CO2H. Journal of Molecular Structure 1991, 248 (1-2):133-142.

211

References

314. Stroev EA: Biochemistry. Moscow: Mir Publishers; 1989. 315. Tortonda F, Pascual-Ahuir J, Silla E, Tuñón I and Ramírez F: Aminoacid zwitterions in solution: Geometric, energetic, and vibrational analysis using density functional theory-continuum model calculations. The Journal of Chemical Physics 1998, 109(2):592. 316. Gould RO, Gray AM, Taylor P and Walkinshaw MD: Crystal environments and geometries of leucine, isoleucine, valine and phenylalanine provide estimates of minimum nonbonded contact and preferred van der Waals interaction distances. Journal of the American Chemical Society 1985, 107 (21):5921-5927. 317. Sun J, Bousquet D, Forbert H and Marx D: Glycine in aqueous solution: solvation shells, interfacial water, and vibrational spectroscopy from ab initio molecular dynamics. The Journal of Chemical Physics 2010, 133(11):114508. 318. Godfrey PD, Firth S, Hatherley LD, Brown RD and Pierlot AP: Millimeter-wave spectroscopy of biomolecules: alanine. Journal of the American Chemical Society 1993, 115(21):9687-9691. 319. Hirshfeld FL: Electron Density Distributions in Molecules. Crystallography Reviews 1991, 2(4):169 - 200. 320. Plekan O, Feyer V, Richter R, Coreno M, de Simone M, Prince KC and Carravetta V: Investigation of the Amino Acids Glycine, Proline, and Methionine by Photoemission Spectroscopy. The Journal of Physical Chemistry A 2007, 111(43):10998-11005. 321. Ahmed M, Ganesan A, Wang F, Feyer V, Plekan O and Prince KC: Photoelectron Spectra of Some Antibiotic Building Blocks: 2-Azetidinone and Thiazolidine Carboxylic Acid. The Journal of Physical Chemistry A 2012. 322. Saha S, Wang F, MacNaughton JB, Moewes A and Chong DP: The attachment of amino fragment to purine: inner-shell structures and spectra. Journal of Synchrotron Radiation 2008, 15(2):151-157. 323. Zhu Q, Wang F and Ivanova EP: Impact of ketone and amino on the inner shell of guanine. Journal of Synchrotron Radiation 2009, 16(4):545-552. 324. Slavicek P, Winter B, Faubel M, Bradforth SE and Jungwirth P: Ionization Energies of Aqueous Nucleic Acids: Photoelectron Spectroscopy of Pyrimidine Nucleosides and ab Initio Calculations. Journal of the American Chemical Society 2009, 131(18):6460-6467. 325. Wang F: Unsaturated Didehydrodeoxycytidine Drugs. 1. Impact of CC Positions in the Sugar Ring. The Journal of Physical Chemistry B 2007, 111(32):9628-9633. 326. Wang F and Pang W: Valence orbital response to conformers of n-butane. Molecular Simulation 2007, 33(14):1173-1185. 327. Saha S, Wang F, Falzon CT and Brunger MJ: Coexistence of 1,3-butadiene conformers in ionization energies and Dyson orbitals. The Journal of Chemical Physics 2005, 123(12):124315-124314. 328. Danecek P, Kapitan J, Baumruk V, Bednarova L, Kopecky V and Bour P: Anharmonic effects in IR, Raman, and Raman optical activity spectra of alanine and proline zwitterions. The Journal of Chemical Physics 2007, 126(22):224513. 329. Lowrey AH, Kalasinsky V and Williams RW: Scaled quantum mechanical force field for glycine in basic solution. Structural Chemistry 1993, 4(5):289-298. 330. Scott AP and Radom L: Harmonic Vibrational Frequencies: An Evaluation of Hartree−Fock, Møller−Plesset, Quadratic Configuration Interaction, Density Functional Theory, and Semiempirical Scale Factors. Journal of Physical Chemistry 1996, 100(41):16502-16513.

212

References

331. Stepanian SG, Reva ID, Radchenko ED, Rosado MTS, Duarte MLTS, Fausto R and Adamowicz L: Matrix-Isolation Infrared and Theoretical Studies of the Glycine Conformers. Journal of Physical Chemistry A 1998, 102(6):1041-1054. 332. Ghiringhelli LM and Delle Site L: Phenylalanine near Inorganic Surfaces: Conformational Statistics vs Specific Chemistry. Journal of the American Chemical Society 2008, 130(8):2634-2638. 333. Kaczor A, Reva ID, Proniewicz LM and Fausto R: Importance of Entropy in the Conformational Equilibrium of Phenylalanine: A Matrix-Isolation Infrared Spectroscopy and Density Functional Theory Study. The Journal of Physical Chemistry A 2006, 110(7):2360-2370. 334. Huang Z, Yu W and Lin Z: Exploration of the full conformational landscapes of gaseous aromatic amino acid phenylalanine: An ab initio study. Journal of Molecular Structure: THEOCHEM 2006, 758 (2-3):195-202. 335. Hashimoto T, Takasu Y, Yamada Y and Ebata T: Anomalous conformer dependent S1 lifetime of l-phenylalanine. Chemical Physics Letters 2006, 421(1-3):227-231. 336. Zhang W, Carravetta V, Plekan O, Feyer V, Richter R, Coreno M and Prince KC: Electronic structure of aromatic amino acids studied by soft x-ray spectroscopy. The Journal of Chemical Physics 2009, 131(3):035103-035111. 337. Martinez Iii SJ, Alfano JC and Levy DH: The electronic spectroscopy of the amino acids tyrosine and phenylalanine in a supersonic jet. Journal of Molecular Spectroscopy 1992, 156(2):421-430. 338. Lee KT, Sung J, Lee KJ, Kim SK and Park YD: Resonant two-photon ionization study of jet-cooled amino acid: L-phenylalanine and its monohydrated complex. The Journal of Chemical Physics 2002, 116(19):8251-8254. 339. Lee KT, Sung J, Lee KJ, Kim SK and Park YD: Conformation-dependent ionization of l-phenylalanine: structures and energetics of cationic conformers. Chemical Physics Letters 2003, 368(3 -4):262-268. 340. Lee KT, Sung J, Lee KJ, Park YD and Kim SK: Conformation-Dependent Ionization Energies of L-Phenylalanine. Angewandte Chemie International Edition 2002, 41(21):4114-4117. 341. Lee Y, Jung J, Kim B, Butz P, Snoek LC, Kroemer RT and Simons JP: Alanyl Side Chain Folding in Phenylalanine: Conformational Assignments through Ultraviolet Rotational Band Contour Analysis. The Journal of Physical Chemistry A 2003, 108(1):69-73. 342. Close DM: Calculated Vertical Ionization Energies of the Common α-Amino Acids in the Gas Phase and in Solution. The Journal of Physical Chemistry A 2011, 115(13):2900-2912. 343. Baek KY, Hayashi M, Fujimura Y, Lin SH and Kim SK: Investigation of Conformation-Dependent Properties of l-Phenylalanine in Neutral and Radical Cations by Using a Density Functional Taking into Account Noncovalent Interactions. The Journal of Physical Chemistry A 2010, 114(28):7583-7589. 344. Cooper G, Gordon M, Tulumello D, Turci C, Kaznatcheev K and Hitchcock AP: Inner shell excitation of glycine, glycyl-glycine, alanine and phenylalanine. Journal of Electron Spectroscopy and Related Phenomena 2004, 137-140 (0):795-799. 345. Plekan O, Feyer V, Richter R, Coreno M and Prince KC: Valence photoionization and photofragmentation of aromatic amino acids. Molecular Physics 2008, 106(9- 10):1143-1153. 346. Campbell S, Marzluff EM, Rodgers MT, Beauchamp JL, Rempe ME, Schwinck KF and Lichtenberger DL: Proton Affinities and Photoelectron Spectra of Phenylalanine and N-Methyl- and N,N-Dimethylphenylalanine. Correlation of

213

References

Lone Pair Ionization Energies with Proton Affinities and Implications for N- Methylation as a Method to Effect Site Specific Protonation of Peptides. Journal of the American Chemical Society 1994, 116(1 2):5257-5264. 347. Seki K and Inokuchi H: Photoelectron spectrum of l-tryptophan in the gas phase. Chemical Physics Letters 1979, 65(1):158-160. 348. Cannington PH and Ham NS: The photoelectron spectra of amino-acids : A survey. Journal of Electron Spectroscopy and Related Phenomena 1979, 15(1):79-82. 349. Campbell S, Beauchamp JL, Rempe M and Lichtenberger DL: Correlations of lone pair ionization energies with proton affinities of amino acids and related compounds. Site specificity of protonation. International Journal of Mass Spectrometry and Ion Processes 1992, 117(0):83-99. 350. Carravetta V, Plashkevych O and Agren H: A theoretical study of the near-edge x- ray absorption spectra of some larger amino acids. The Journal of Chemical Physics 1998, 109(4):1456-1464. 351. Yang L, Plashkevytch O, Vahtras O, Carravetta V and Agren H: Near-edge X-ray absorption and dichroism in amino acids. Journal of Synchrotron Radiation 1999, 6(3):708-710. 352. Vorsa V, Kono T, Willey KF and Winograd N: Femtosecond Photoionization of Ion Beam Desorbed Aliphatic and Aromatic Amino Acids: Fragmentation via α- Cleavage Reactions. The Journal of Physical Chemistry B 1999, 103(37):7889-7895. 353. Ahmed M, Ganesan A, Wang F, Feyer V, Plekan O and Prince KC: Photoelectron Spectra of Some Antibiotic Building Blocks: 2-Azetidinone and Thiazolidine Carboxylic Acid. The Journal of Physical Chemistry A 2012, 116 (33):8653–8660. 354. Kroemer RT, Liedl KR, Dickinson JA, Robertson EG, Simons JP, Borst DR and Pratt DW: Conformationally Induced Changes in the Electronic Structures of Some Flexible Benzenes. A Molecular Orbital Model. Journal of the American Chemical Society 1998, 120(48):12573-12582. 355. Castro JL, López Ramírez MR, Arenas JF and Otero JC: Vibrational spectra of 3- phenylpropionic acid and L-phenylalanine. Journal of Molecular Structure 2005, 744-747(SPEC. ISS.):887-891. 356. Urban JJ, Cronin CW, Roberts RR and Famini GR: Conformational preferences of 2-phenethylamines. A computational study of substituent and solvent effects on the intramolecular amine-aryl interactions in charged and neutral 2- phenethylamines. Journal of the American Chemical Society 1997, 119(50):12292- 12299. 357. Yao J, Im HS, Foltin M and Bernstein ER: Spectroscopy of neurotransmitters and their clusters: phenethylamine and amphetamine solvation by nonpolar, polar, and hydrogen-bonding solvents. Journal of Physical Chemistry A 2000, 104(26):6197-6211. 358. Richardson PR, Bates SP and Jones AC: A molecular orbital study of the conformational properties of tyramine and phenethylamine. Journal of Physical Chemistry A 2004, 108(7):1233-1241. 359. Biemann K, Seibl J and Gapp F: Mass Spectra of Organic Molecules. I. Ethyl Esters of Amino Acids1. Journal of the American Chemical Society 1961, 83(18):3795-3804. 360. Gregor and Svec H: The Mass Spectra of the α-Amino Acids. Journal of the American Chemical Society 1963, 85(7):839-845. 361. Rizzo TR, Park YD and Levy DH: A molecular beam of tryptophan. Journal of the American Chemical Society 1985, 107(1):277-278.

214

References

362. Hrubowchak DM, Ervin MH, Wood MC and Winograd N: Detection of biomolecules on surfaces using ion-beam-induced desorption and multiphoton resonance ionization. Analytical Chemistry 1991, 63(18):1947-1953. 363. Rizzo TR, Park YD, Peteanu LA and Levy DH: The electronic spectrum of the amino acid tryptophan in the gas phase. The Journal of Chemical Physics 1986, 84(5):2534-2541. 364. Powis I, Rennie EE, Hergenhahn U, Kugeler O and Bussy-Socrate R: Investigation of the gas-phase amino acid alanine by synchrotron radiation photoelectron spectroscopy. Journal of Physical Chemistry A 2003, 107(1):25-34. 365. Takahata Y, Chong DP and Segala M: Is HAM/3 (Hydrogenic Atoms in Molecules, Version 3) a semiempirical version of DFT (Density Functional Theory) for ionization processes? Journal of the Brazilian Chemical Society 2004, 15(2):282- 291. 366. Wang F, Downton MT and Kidwani N: Adenine tautomer electronic structural signatures studied using dual space analysis. Journal of Theoretical and Computational Chemistry 2005, 4(1):247-264. 367. Feng W: Applications of ionization spectroscopy to study small bio-molecules. Journal of Physics: Conference Series 2008, 141(1):012019. 368. Stener M, Fronzoni G and Decleva P: Time-dependent density-functional theory for molecular photoionization with noniterative algorithm and multicenter B- spline basis set: CS[sub 2] and C[sub 6]H[sub 6] case studies. The Journal of Chemical Physics 2005, 122(23):234301-234311. 369. Yencha AJ, Hall RI, Avaldi L, Dawber G, McConkey AG, MacDonald MA and King GC: Threshold photoelectron spectroscopy of benzene up to 26.5 eV. Canadian Journal of Chemistry 2004, 82(6):1061-1066. 370. Medhurst LJ, Ferrett TA, Heimann PA, Lindle DW, Liu SH and Shirley DA: Observation of correlation effects in zero kinetic energy electron spectra near the N1s and C1s thresholds in N[sub 2], CO, C[sub 6]H[sub 6], and C[sub 2]H[sub 4]. The Journal of Chemical Physics 1988, 89(10):6096-6102. 371. Holmes SA and Thomas TD: Electron distribution in trifluoromethylbenzenes. Electron donation by the trifluoromethyl group. Journal of the American Chemical Society 1975, 97 (9):2337-2341. 372. Jolly WL, Bomben KD and Eyermann CJ: Core-electron binding energies for gaseous atoms and molecules. Atomic Data and Nuclear Data Tables 1984, 31(3):433-493. 373. Klasinc L, Kovac B and Gusten H: Photoelectron spectra of acenes. Electronic structure and substituent effects. Pure and Applied Chemistry 1983, 55(2):289-298. 374. Godby RW, Schlüter M and Sham LJ: Self-energy operators and exchange- correlation potentials in semiconductors. Physical Review B 1988, 37(17):10159- 10175. 375. Samardzic O, Brunger MJ, Grisogono AM and Weigold E: Electron momentum spectroscopy studies on ring compounds. I. Benzene. Journal of Physics B: Atomic, Molecular and Optical Physics 1993, 26(21):3921. 376. Fuss I, McCarthy IE, Minchinton A, Weigold E and Larkins FP: Momentum distributions and ionization potentials for the valence orbitals of benzene. Chemical Physics 1981, 63(1–2):19-30. 377. Baek KY, Fujimura Y, Hayashi M, Lin SH and Kim SK: Density Functional Theory Study of Conformation-Dependent Properties of Neutral and Radical Cationic l- Tyrosine and l-Tryptophan. The Journal of Physical Chemistry A 2011, 115(34):9658-9668.

215

References

378. BELL C, ABRAMS J and NUTT D: Tryptophan depletion and its implications for psychiatry. The British Journal of Psychiatry 2001, 178(5):399-405. 379. Grace LI, Cohen R, Dunn TM, Lubman DM and de Vries MS: The R2PI Spectroscopy of Tyrosine: A Vibronic Analysis. Journal of Molecular Spectroscopy 2002, 215(2):204-219. 380. Han X, Civiello RL, Fang H, Wu D, Gao Q, Chaturvedula PV, Macor JE and Dubowchik GM: Catalytic Asymmetric Syntheses of Tyrosine Surrogates. The Journal of Organic Chemistry 2008, 73(21):8502-8510. 381. Lima S, Kumar S, Gawandi V, Momany C and Phillips RS: Crystal Structure of the Homo sapiens Kynureninase-3-Hydroxyhippuric Acid Inhibitor Complex: Insights into the Molecular Basis Of Kynureninase Substrate Specificity†. Journal of Medicinal Chemistry 2008, 52(2):389-396. 382. Shi Z-C, Devasagayaraj A, Gu K, Jin H, Marinelli B, Samala L, Scott S, Stouch T, Tunoori A, Wang Y, Zang Y, Zhang C, Kimball SD, Main AJ, Sun W, Yang Q, Nouraldeen A, Yu X-Q, Buxton E, Patel S, Nguyen N, Swaffield J, Powell DR, Wilson A and Liu Q: Modulation of Peripheral Serotonin Levels by Novel Tryptophan Hydroxylase Inhibitors for the Potential Treatment of Functional Gastrointestinal Disorders. Journal of Medicinal Chemistry 2008, 51(13):3684- 3687. 383. Ishiuchi S-i, Mitsuda H, Asakawa T, Miyazaki M and Fujii M: Conformational reduction of DOPA in the gas phase studied by laser desorption supersonic jet laser spectroscopy. Physical Chemistry Chemical Physics 2011, 13(17):7812-7820. 384. Misu Y, Goshima Y and Miyamae T: Is DOPA a neurotransmitter? Trends in Pharmacological Sciences 2002, 23(6):262-268. 385. ÇarÇabal P, Snoek * LC and Van Mourik T: A computational and spectroscopic study of the gas-phase conformers of adrenaline. Molecular Physics 2005, 103(11- 12):1633-1639. 386. Snoek LC, Van Mourik T and Simons JP: Neurotransmitters in the gas phase: a computational and spectroscopic study of noradrenaline. Molecular Physics 2003, 101(9):1239-1248. 387. Kobe B, Jennings IG, House CM, Michell BJ, Goodwill KE, Santarsiero BD, Stevens RC, Cotton RGH and Kemp BE: Structural basis of autoregulation of phenylalanine hydroxylase. Nat Struct Mol Biol 1999, 6(5):442-448. 388. Renson J, Weissbach H and Udenfriend S: Hydroxylation of Tryptophan by Phenylalanine Hydroxylase. Journal of Biological Chemistry 1962, 237(7):2261- 2264. 389. Basu S and Dasgupta PS: Dopamine, a neurotransmitter, influences the immune system. Journal of Neuroimmunology 2000, 102(2):113-124. 390. Goodwill KE, Sabatier C, Marks C, Raag R, Fitzpatrick PF and Stevens RC: Crystal structure of tyrosine hydroxylase at 2.3 A and its implications for inherited neurodegenerative diseases. Nat Struct Mol Biol 1997, 4(7):578-585. 391. Lee BP, Chao C-Y, Nunalee FN, Motan E, Shull KR and Messersmith PB: Rapid Gel Formation and Adhesion in Photocurable and Biodegradable Block Copolymers with High DOPA Content. Macromolecules 2006, 39(5):1740-1748. 392. Lloyd KG, Davidson L and Hornykiewicz O: The neurochemistry of Parkinson's disease: effect of L-dopa therapy. Journal of Pharmacology and Experimental Therapeutics 1975, 195(3):453-464. 393. Vijaya Chamundeeswari SP, Samuel EJJ and Sundaraganesan N: Molecular structure and spectroscopic (FT-IR, FT-Raman, 13C, 1H NMR and UV) studies

216

References

of 3,4-dihydroxy-l-phenylalanine using density functional theory. Molecular Simulation 2012:1-14. 394. Petzer JP, Castagnoli Jr N, Schwarzschild MA, Chen J-F and Van der Schyf CJ: Dual-Target–Directed Drugs that Block Monoamine Oxidase B and Adenosine A2A Receptors for Parkinson's Disease. Neurotherapeutics 2009, 6(1):141-151. 395. DeFeudis FV: Inhibition of monoamine oxidase by 3,4-dihydroxyphenyl 1-alanine and its analogs. Journal of Basic and Clinical Physiology and Pharmacology 1988, 7 (1-4):207-208. 396. Carlsson A, Lindqvist M and Magnusson TOR: 3,4-Dihydroxyphenylalanine and 5- Hydroxytryptophan as Reserpine Antagonists. Nature 1957, 180(4596):1200- 1200. 397. Carlsson A: A Half-Century of Neurotransmitter Research: Impact on Neurology and Psychiatry (Nobel Lecture). ChemBioChem 2001, 2 (7-8):484-493. 398. Carlsson A: http://wwwnobelprizeorg/nobel_prizes/medicine/laureates/2000/illpres/carlssonhtml. 399. Knowles WS: Asymmetric Hydrogenations (Nobel Lecture). Angewandte Chemie International Edition 2002, 41(12):1998-2007. 400. Mitsuda H, Miyazaki M, Nielsen IB, ar abal P, Dedonder C, Jouvet C, Ishiuchi S-i and Fujii M: Evidence for Catechol Ring- Induced Conformational Restriction in Neurotransmitters. The Journal of Physical Chemistry Letters 2010, 1(7):1130- 1133. 401. Lindinger A, Toennies JP and Vilesov AF: High resolution vibronic spectra of the amino acids tryptophan and tyrosine in 0.38 K cold helium droplets. The Journal of Chemical Physics 1999, 110(3):1429-1436. 402. Robertson EG and Simons JP: Getting into shape: Conformational and supramolecular landscapes in small biomolecules and their hydrated clusters. Physical Chemistry Chemical Physics 2001, 3(1):1-18. 403. Snoek LC, Kroemer RT, Hockridge MR and Simons JP: Conformational landscapes of aromatic amino acids in the gas phase: Infrared and ultraviolet ion dip spectroscopy of tryptophan. Physical Chemistry Chemical Physics 2001, 3(10):1819-1826. 404. Snoek LC, Kroemer RT and Simons JP: A spectroscopic and computational exploration of tryptophan-water cluster structures in the gas phase. Physical Chemistry Chemical Physics 2002, 4(11):2130-2139. 405. Li L and Lubman DM: Analytical Jet Spectroscopy of Tyrosine and Its Analogs Using a Pulsed Laser Desorption Volatilization Method. Applied Spectroscopy 1988, 42(3):418-424. 406. Kushwaha PS and Mishra PC: Electronic spectra, excited-state geometries and molecular electrostatic potentials of aromatic amino acids. Journal of Photochemistry and Photobiology A: Chemistry 2000, 137(2–3):79-86. 407. Frey MN, Koetzle TF, Lehmann MS and Hamilton WC: Precision neutron diffraction structure determination of protein and nucleic acid components. X. A comparison between the crystal and molecular structures of L-tyrosine and L- tyrosine hydrochloride. The Journal of Chemical Physics 1973, 58(6):2547-2556. 408. Dehareng D and Dive G: Vertical Ionization Energies of α-L-Amino Acids as a Function of Their Conformation: an Ab Initio Study. International Journal of Molecular Sciences 2004, 5(11):301-332. 409. Grotemeyer J and Schlag EW: Biomolecules in the gas phase: multiphoton ionization mass spectrometry. Accounts of Chemical Research 1989, 22(11):399- 406.

217

References

410. Martinez SJ, Alfano JC and Levy DH: The Electronic Spectroscopy of Tyrosine and Phenylalanine Analogs in a Supersonic Jet: Basic Analogs. Journal of Molecular Spectroscopy 1993, 158(1):82-92. 411. Zhang M, Huang Z and Lin Z: Systematic ab initio studies of the conformers and conformational distribution of gas-phase tyrosine. The Journal of Chemical Physics 2005, 122(13):134313-134317. 412. Zubavichus Y, Shaporenko A, Grunze M and Zharnikov M: Innershell Absorption Spectroscopy of Amino Acids at All Relevant Absorption Edges. The Journal of Physical Chemistry A 2005, 109(32):6998-7000. 413. Kaznacheyev K, Osanna A, Jacobsen C, Plashkevych O, Vahtras O, Ågren, Carravetta V and Hitchcock AP: Innershell Absorption Spectroscopy of Amino Acids. The Journal of Physical Chemistry A 2002, 106(13):3153-3168. 414. Jochims H-W, Schwell M, Chotin J-L, Clemino M, Dulieu F, Baumgärtel H and Leach S: Photoion mass spectrometry of five amino acids in the 6–22 eV photon energy range. Chemical Physics 2004, 298(1–3):279-297. 415. El Aribi H, Orlova G, Hopkinson AC and Siu KWM: Gas-Phase Fragmentation Reactions of Protonated Aromatic Amino Acids: Concomitant and Consecutive Neutral Eliminations and Radical Cation Formations. The Journal of Physical Chemistry A 2004, 108(17):3844-3853. 416. Schleyer PvR, Maerker C, Dransfeld A, Jiao H and Hommes NJRvE: Nucleus- Independent Chemical Shifts: A Simple and Efficient Aromaticity Probe. Journal of the American Chemical Society 1996, 118 (26):6317-6318. 417. Noorizadeh S and Dardab M: A new NICS-based aromaticity index; NICS-rate. Chemical Physics Letters 2010, 493 (4–6):376-380. 418. Mostad A, Ottersen T and Romming C: X-ray crystal structure determination of 3,4-dihydroxy-phenylalanine (L-DOPA). Acta chemica Scandinavica 1970, 24(5):1864-1865. 419. Tu G, Tu Y, Vahtras O and Ågren H: Core electron chemical shifts of hydrogen- bonded structures. Chemical Physics Letters 2009, 468(4–6):294-298. 420. Plekan O, Feyer V, Richter R, Coreno M, de Simone M, Prince KC and Carravetta V: Photoemission and the shape of amino acids. Chemical Physics Letters 2007, 442(4–6):429-433. 421. Leffler JE and Grünwald E: Rates and equilibria of organic reactions as treated by statistical, thermodynamic, and extrathermodynamic methods: Wiley; 1963. 422. Jones RAY: Physical and Mechanistic Organic Chemistry: Cambridge University Press; 1979. 423. Wickrama Arachchilage AP, Feyer V, Plekan O, Iakhnenko M, Prince KC and Wang F: Valence structures of aromatic bioactive compounds: a combined theoretical and experimental study. Journal of Synchrotron Radiation 2012, 19(5). 424. Jones DB, Wang F, Winkler DA and Brunger MJ: Orbital based electronic structural signatures of the guanine keto G-7H/G-9H tautomer pair as studied using dual space analysis. Biophysical Chemistry 2006, 121(2):105-120. 425. Schleyer PVR and Jiao H: What is aromaticity? Pure and Applied Chemistry 1996, 68(2):209-218. 426. Garratt PJ: Aromaticity; 1986. 427. Minkin VJ: Aromaticity and Antiaromaticity; Electronic and Structural Aspects; 1994. 428. Mohajeri A and Ashrafi A: Aromaticity in terms of ring critical point properties. Chemical Physics Letters 2008, 458 (4–6):378-383.

218

References

429. Zhou Z, Parr RG and F. Garst J: Absolute hardness as a measure of aromaticity. Tetrahedron Letters 1988, 29(38):4843-4846. 430. Zhou Z and Parr RG: New measures of aromaticity: absolute hardness and relative hardness. Journal of the American Chemical Society 1989, 111(19):7371- 7379. 431. Krygowski TM and Cyrański MK: Structural Aspects of Aromaticity. Chemical Reviews 2001, 101(5):1385-1420. 432. Krygowski TM, Cyranski MK, Czarnocki Z, Haefelinger G and Katritzky AR: ChemInform Abstract: Aromaticity: A Theoretical Concept of Immense Practical Importance. ChemInform 2000, 31(25):no-no. 433. Solà M, Feixas F, Jiménez-Halla JOC, Matito E and Poater J: A Critical Assessment of the Performance of Magnetic and Electronic Indices of Aromaticity. Symmetry 2010, 2(2):1156-1179. 434. Cyrański MK: Energetic Aspects of Cyclic Pi-Electron Delocalization: Evaluation of the Methods of Estimating Aromatic Stabilization Energies. Chemical Reviews 2005, 105(10):3773-3811. 435. Poater J, Duran M, Solà M and Silvi B: Theoretical Evaluation of Electron Delocalization in Aromatic Molecules by Means of Atoms in Molecules (AIM) and Electron Localization Function (ELF) Topological Approaches. Chemical Reviews 2005, 105(10):3911-3947. 436. Merino G, Vela A and Heine T: Description of Electron Delocalization via the Analysis of Molecular Fields. Chemical Reviews 2005, 105(10):3812-3841. 437. Stanger A: Nucleus-Independent Chemical Shifts (NICS): Distance Dependence and Revised Criteria for Aromaticity and Antiaromaticity. The Journal of Organic Chemistry 2005, 71(3):883-893. 438. Chen Z, Wannere CS, Corminboeuf C, Puchta R and Schleyer PvR: Nucleus- Independent Chemical Shifts (NICS) as an Aromaticity Criterion. Chemical Reviews 2005, 105(10):3842-3888. 439. Li X, Kuznetsov AE, Zhang H-F, Boldyrev AI and Wang L-S: Observation of All- Metal Aromatic Molecules. Science 2001, 291(5505):859-861. 440. King RB: Aromaticity in Transition Metal Oxide Structures. Journal of Chemical Information and Computer Sciences 2001, 41(3):517-526. 441. Datta A and Pati SK: Rationalization of the π−σ (Anti)aromaticity in All Metal Molecular Clusters. Journal of Chemical Theory and Computation 2005, 1(5):824- 826. 442. Seal P and Chakrabarti S: Is Nucleus-Independent Chemical Shift Scan a Reliable Aromaticity Index for Planar Heteroatomic Ring Systems? The Journal of Physical Chemistry A 2007, 111(39):9988-9994. 443. Jiménez-Halla JOC, Matito E, Robles J and Solà M: Nucleus-independent chemical shift (NICS) profiles in a series of monocyclic planar inorganic compounds. Journal of Organometallic Chemistry 2006, 691(21):4359-4366. 444. Chi XX, Chen XJ and Yuan ZS: Theoretical study on the aromaticity of the bimetallic clusters X 2M2 (X=Si, Ge, M=Al, Ga). Journal of Molecular Structure: THEOCHEM 2005, 732(1 -3):149-153. 445. Martin-Santamaria S and Rzepa HS: Double aromaticity and anti-aromaticity in small carbon rings. Chemical Communications 2000(16):1503-1504. 446. Ditchfield R: Theoretical studies of magnetic shielding in H[sub 2]O and (H[sub 2]O)[sub 2]. The Journal of Chemical Physics 1976, 65(8):3123-3133. 447. London F: Théorie quantique des courants interatomiques dans les combinaisons aromatiques. J Phys Radium 1937, 8:397-409

219

References

448. Selvam L: Simulation of spectroscopic properties of atoms and molecules. In. 449. Lukyanov SM and Koblik AV: Tautomeric Equilibria and Rearrangements Involving Phenols. In: PATAI'S Chemistry of Functional Groups. John Wiley & Sons, Ltd; 2009. 450. Sahu P and Lee S-L: Effect of microsolvation on zwitterionic glycine: an ab initio and density functional theory study. Journal of Molecular Modeling 2008, 14(5):385-392. 451. Wang J and El-Sayed MA: The Effect of Metal Cation Binding on the Protein, Lipid and Retinal Isomeric Ratio in Regenerated Bacteriorhodopsin of Purple Membrane. Photochemistry and Photobiology 2001, 73(5):564-571. 452. Shukla MK, Mishra SK, Kumar A and Mishra PC: An ab initio study of excited states of guanine in the gas phase and aqueous media: Electronic transitions and mechanism of spectral oscillations. Journal of Computational Chemistry 2000, 21(10):826-846. 453. Shukla MK and Mishra PC: An ab initio study of electronic spectra and excited- state properties of 7-azaindole in vapour phase and aqueous solution. Chemical Physics 1998, 230(2–3):187-200. 454. Kalyaanamoorthy S and Chen Y-PP: Exploring Inhibitor Release Pathways in Histone Deacetylases Using Random Acceleration Molecular Dynamics Simulations. Journal of Chemical Information and Modeling 2012. 455. Larrucea J, Rezabal E, Marino T, Russo N and Ugalde JM: Ab Initio Study of Microsolvated Al3+−Aromatic Amino Acid Complexes. Journal of Physical Chemistry B 2010, 114(27):9017-9022. 456. Otto R, Brox J, Trippel S, Stei M, Best T and Wester R: Single solvent molecules can affect the dynamics of substitution reactions. Nature Chemistry 2012, 4(7):534-538. 457. Bertrán J, Rodríguez-Santiago L and Sodupe M: The Different Nature of Bonding in Cu+-Glycine and Cu2+-Glycine. Journal of Physical Chemistry B 1999, 103(12):2310-2317. 458. Hu P and Gross ML: Gas-phase interactions of transition-metal ions and di- and tripeptides: a comparison with alkaline-earth-metal-ion interactions. Journal of the American Chemical Society 1993, 115 (19):8821-8828. 459. Reiter A, Adams J and Zhao H: Intrinsic (Gas-Phase) Binding of Co2+ and Ni2+ by Peptides: A Direct Reflection of Aqueous-Phase Chemistry. Journal of the American Chemical Society 1994, 116(17):7827-7838. 460. Ma S, Wong P, Yang SS and Cooks RG: Gas-Phase Molecular, Molecular Pair, and Molecular Triplet Fe+ Affinities of Pyridines. Journal of the American Chemical Society 1996, 118 (25):6010-6019. 461. Cerda BA and Wesdemiotis C: The Relative Copper(I) Ion Affinities of Amino Acids in the Gas Phase. Journal of the American Chemical Society 1995, 117(38):9734-9739. 462. Pasquarello A, Petri I, Salmon PS, Parisel O, Car R, Tóth É, Powell DH, Fischer HE, Helm L and Merbach AE: First Solvation Shell of the Cu(II) Aqua Ion: Evidence for Fivefold Coordination. Science 2001, 291(5505):856-859. 463. Blomberg MRA, Siegbahn PEM, Styring S, Babcock GT, Åkermark B and Korall P: A Quantum Chemical Study of Hydrogen Abstraction from Manganese- Coordinated Water by a Tyrosyl Radical: A Model for Water Oxidation in Photosystem II. Journal of the American Chemical Society 1997, 119(35):8285- 8292.

220

References

464. Luna A, Amekraz B, Morizur JP, Tortajada J, Mó O and Yáñez M: Reactions between Guanidine and Cu+ in the Gas Phase. An Experimental and Theoretical Study. Journal of Physical Chemistry A 1997, 101(33):5931-5941. 465. Burda JV, Pavelka M and Simánek M: Theoretical model of copper Cu(I)/Cu(II) hydration. DFT and ab initio quantum chemical study. Journal of Molecular Structure: THEOCHEM 2004, 683(1 -3):183-193. 466. Amira S, Spangberg D and Hermansson K: Distorted five-fold coordination of Cu2+(aq) from a Car-Parrinello molecular dynamics simulation. Physical Chemistry Chemical Physics 2005, 7(15):2874-2880. 467. Rimola A, Sodupe M, Tortajada J and Rodríguez-Santiago L: Gas phase reactivity of Cu+-aromatic amino acids: An experimental and theoretical study. International Journal of Mass Spectrometry 2006, 257(1-3):60-69. 468. Remko M and Rode BM: Effect of Metal Ions (Li+, Na+, K+, Mg2+, Ca2+, Ni2+, Cu2+, and Zn2+) and Water Coordination on the Structure of Glycine and Zwitterionic Glycine. Journal of Physical Chemistry A 2006, 110 (5):1960-1967. 469. Bai Y, Wang YD, Zheng WJ and Chen YS: Study on coordination of selenoamino acids with Ag+ at silver nitrate-modified carbon paste electrode. Colloids and Surfaces B: Biointerfaces 2008, 63(1):110-115. 470. Siegbahn PEM: Modeling aspects of mechanisms for reactions catalyzed by metalloenzymes. Journal of Computational Chemistry 2001, 22(14):1634-1645. 471. de Almeida KJ, Murugan NA, Rinkevicius Z, Hugosson HW, Vahtras O, Agren H and Cesar A: Conformations, structural transitions and visible near-infrared absorption spectra of four-, five- and six-coordinated Cu(ii) aqua complexes. Physical Chemistry Chemical Physics 2009, 11(3). 472. Persson I, Persson P, Sandstrom M and Ullstrom A-S: Structure of Jahn-Teller distorted solvated copper(ii) ions in solution, and in solids with apparently regular octahedral coordination geometry. Journal of the Chemical Society, Dalton Transactions 2002(7):1256-1265. 473. Benfatto M, D’Angelo P, Della Longa S and Pavel NV: Evidence of distorted fivefold coordination of the Cu^{2+} aqua ion from an x-ray-absorption spectroscopy quantitative analysis. Physical Review B 2002, 65 (17):174205. 474. Frank P, Benfatto M, Szilagyi RK, D'Angelo P, Longa SD and Hodgson KO: The Solution Structure of [Cu(aq)]2+ and Its Implications for Rack-Induced Bonding in Blue Copper Protein Active Sites. Inorganic Chemistry 2005, 44(6):1922-1933. 475. Chaboy J, Munoz-Paez A, Merkling PJ and Marcos ES: The hydration of Cu[sup 2+]: Can the Jahn-Teller effect be detected in liquid solution? Journal of Chemical Physics 2006, 124(6):064509-064509. 476. Nomura M and Yamaguchi T: Concentration dependence of EXAFS and XANES of copper(II) perchlorate aqueous solution: comparison of solute structure in liquid and glassy states. The Journal of Physical Chemistry 1988, 92(21):6157-6160. 477. Powell DH, Helm L and Merbach AE: [sup 17]O nuclear magnetic resonance in aqueous solutions of Cu[sup 2 + ] : The combined effect of Jahn--Teller inversion and solvent exchange on relaxation rates. Journal of Chemical Physics 1991, 95(12):9258-9265. 478. Garcia J, Benfatto M, Natoli CR, Bianconi A, Fontaine A and Tolentino H: The quantitative Jahn-teller distortion of the Cu2+ site in aqueous solution by xanes spectroscopy. Chemical Physics 1989, 132(1–2):295-302. 479. Beagley B, Eriksson A, Lindgren J, Persson I, Pettersson LGM, Sandstrom M, Wahlgren U and White EW: A computational and experimental study on the Jahn-Teller effect in the hydrated copper (II) ion. Comparisons with hydrated

221

References

nickel (II) ions in aqueous solution and solid Tutton's salts. Journal of Physics: Condensed Matter 1989, 1(13):2395. 480. Salmon PS, Neilson GW and Enderby JE: The structure of Cu 2+ aqueous solutions. Journal of Physics C: Solid State Physics 1988, 21(8):1335. 481. Ozutsumi K and Kawashima T: Exafs and spectrophotometric studies on the structure of pyridine complexes with copper(II) and copper(I) ions in aqueous solution. Polyhedron 1992, 11(2):169-175. 482. Sano M, Maruo T, Masuda Y and Yamatera H: Structural study of copper(II) sulfate solution in highly concentrated aqueous ammonia by x-ray absorption spectra. Inorganic Chemistry 1984, 23 (26):4466-4469. 483. Valli M, Matsuo S, Wakita H, Yamaguchi T and Nomura M: Solvation of Copper(II) Ions in Liquid Ammonia. Inorganic Chemistry 1996, 35(19):5642-5645. 484. Valli M, Matsuo S, Wakita H, Yamaguchi T and Nomura M: ChemInform Abstract: Solvation of Copper(II) Ions in Liquid Ammonia. ChemInform 1997, 28(4):no-no. 485. Emsley J, Arif M, Bates PA and Hursthouse MB: Diaquabis(1,3- diaminopropane)copper(II) difluoride: X-ray structure reveals short hydrogen bonds between ligand waters and lattice fluorides. Inorganica Chimica Acta 1988, 154(1):17-20. 486. Emsley J, Arif M, Bates PA and Hursthouse MB: Hydrogen bonding between free fluoride ions and water molecules: two X-ray structures. Journal of Molecular Structure 1990, 220(0):1-12. 487. Tomlinson AAG and Hathaway BJ: The electronic properties and stereochemistry of the copper(II) ion. Part III. Some penta-ammine complexes. Journal of the Chemical Society A: Inorganic, Physical, Theoretical 1968. 488. Duggan M, Ray N, Hathaway B, Tomlinson G, Brint P and Pelin K: Crystal structure and electronic properties of ammine[tris(2- aminoethyl)amine]copper(II) diperchlorate and potassium penta- amminecopper(II) tris(hexafluorophosphate). Journal of the Chemical Society, Dalton Transactions 1980(8). 489. Elliott H and Hathaway BJ: The Hexaammine Complexes of the Copper(II) Ion. Inorganic Chemistry 1966, 5(5):885-889. 490. Distler TM and Vaughan PA: Crystal structures of the hexaamminecopper(II) halides. Inorganic Chemistry 1967, 6(1):126-129. 491. Rul šek Lr and Vondr šek J: Coordination geometries of selected transition metal ions (Co2+, Ni2+, Cu2+, Zn2+, Cd2+, and Hg2+) in metalloproteins. Journal of Inorganic Biochemistry 1998, 71(3–4):115-127. 492. Hay PJ: Gaussian basis sets for molecular calculations. The representation of 3d orbitals in transition-metal atoms. Journal of Chemical Physics 1977, 66(10):4377- 4384. 493. Wachters AJH: Gaussian Basis Set for Molecular Wavefunctions Containing Third-Row Atoms. Journal of Chemical Physics 1970, 52(3):1033-1036. 494. Ganesan A and Wang F: Intramolecular interactions of L-phenylalanine revealed by inner shell chemical shift. Journal of Chemical Physics 2009, 131(4):044321- 044329. 495. Blomberg MRA, Siegbahn PEM and Svensson M: Comparisons of results from parametrized configuration interaction (PCI-80) and from hybrid density functional theory with experiments for first row transition metal compounds. Journal of Chemical Physics 1996, 104 (23):9546-9554. 496. Holthausen MC, Mohr M and Koch W: The performance of density functional/Hartree-Fock hybrid methods: the bonding in cationic first-row

222

References

transition metal methylene complexes. Chemical Physics Letters 1995, 240(4):245- 252. 497. Adamo C and Lelj F: A hybrid density functional study of the first-row transition- metal monocarbonyls. Journal of Chemical Physics 1995, 103(24):10605-10613. 498. CPMD v, C. (revision a11); Copyright IBM Corp, 1990−2008; Copyright MPI fr Festkrperforschung Stuttgart, 1997−2001; http://www.cpmd.org/. 499. Troullier N, Martins J, eacute and Luriaas: Efficient pseudopotentials for plane- wave calculations. Physical Review B 1991, 43(3):1993. 500. Louie SG, Froyen S and Cohen ML: Nonlinear ionic pseudopotentials in spin- density-functional calculations. Physical Review B 1982, 26(4):1738. 501. Nose S: A unified formulation of the constant temperature molecular dynamics methods. Journal of Chemical Physics 1984, 81(1):511-519. 502. Hoover WG: Canonical dynamics: Equilibrium phase-space distributions. Physical Review A 1985, 31(3):1695. 503. Bérces A, Nukada T, Margl P and Ziegler T: Solvation of Cu2+ in Water and Ammonia. Insight from Static and Dynamical Density Functional Theory. Journal of Physical Chemistry A 1999, 103(48):9693-9701. 504. Magini M: Coordination of copper(II). Evidence of the Jahn-Teller effect in aqueous perchlorate solutions. Inorganic Chemistry 1982, 21 (4):1535-1538. 505. Musinu A, Paschina G, Piccaluga G and Magini M: Coordination of copper(II) in aqueous copper sulfate solution. Inorganic Chemistry 1983, 22(8):1184-1187. 506. Remko M, Fitz D and Rode B: Effect of metal ions (Li+, Na+, K+, Mg2+, Ca2+, Ni2+, Cu2+ and Zn2+) and water coordination on the structure and properties of l-histidine and zwitterionic l-histidine. Amino Acids 2010, 39 (5):1309-1319. 507. Julen L: Car–Parrinello molecular dynamics study of the coordination on Al 3+ (aq). Physica Scripta 2011, 84(4):045305. 508. Wincel H: Hydration Energies of Sodiated Amino Acids from Gas-Phase Equilibria Determinations. Journal of Physical Chemistry A 2007, 111(26):5784- 5791. 509. Addicoat MA, Metha GF and Kee TW: Density functional theory investigation of Cu(I)- and Cu(II)-curcumin complexes. Journal of Computational Chemistry 2011, 32(3):429-438. 510. Seymour JL and Ture ek F: Structure, energetics and reactivity of ternary complexes of amino acids with Cu(II) and 2,2′-bipyridine by density functional theory. A combination of radical-induced and spin-remote fragmentations. Journal of Mass Spectrometry 2002, 37 (5):533-540. 511. Fukui K: Role of Frontier Orbitals in Chemical Reactions. Science 1982, 218(4574):747-754. 512. Ture ek F: Copper-biomolecule complexes in the gas phase. The ternary way. Mass Spectrometry Reviews 2007, 26(4):563-582. 513. Iijima K and Beagley B: An electron diffraction study of gaseous α-alanine, NH2CHCH3CO2H. Journal of Molecular Structure 1991, 248(1–2):133-142. 514. Duflot D, Flament JP, Heinesch J and Hubin-Franskin MJ: Re-analysis of the K- shell spectrum of benzene. Journal of Electron Spectroscopy and Related Phenomena 2000, 113(1):79-90. 515. Amir-Ebrahimi V, Choplin A, Demaison J and Roussy G: Microwave spectrum of the 13C-ring-monosubstituted toluenes and structure of toluene. Journal of Molecular Spectroscopy 1981, 89(1):42-52. 516. Hameka HF and Jensen JO: Theoretical studies of the methyl rotational barrier in toluene. Journal of Molecular Structure: THEOCHEM 1996, 362(3):325-330.

223