QM/MM Study of DszB Reaction Mechanism for Crude Oil Biodesulphurization

QM/MM Study of

DszB Reaction

Mechanism for

Crude Oil

Biodesulphurization

João Pedro Marques de Sousa

Dissertação de Mestrado apresentada à

Faculdade de Ciências da Universidade do Porto e Instituto de

Ciências Biomédicas Abel Salazar em

Bioquímica

2018

DszB Reaction

Mechanism for

Crude Oil

Biodesulphurization

João Pedro Marques de Sousa

Mestrado em Bioquímica

Departamento de Química e Bioquímica 2018

Orientador

Doutor Pedro Alexandrino Fernandes, Professor Associado, Faculdade de Ciências da Universidade do Porto

Coorientador

Doutor Sérgio Sousa, Investigador FCT UCIBIO@REQUINTE, Faculdade de Medicina da Universidade do Porto

O Presidente do Júri,

Oral Communications

Sousa, J., Sousa, S. F., Ramos, M. J., Fernandes, P. A., “QM/MM Study of DszB Reaction Mechanism for Crude Oil Biodesulphurization”, IJUP’18 – 11º Encontro de Jovens Investigadores da Universidade do Porto, Centro de Investigação Médica da Faculdade de Medicina da Universidade do Porto, 2018 February 8th, Porto

Agradecimentos

Em primeiro lugar, gostaria de agradecer ao Professor Doutor Pedro Alexandrino Fernandes pela sua orientação ao longo deste último ano. A sua motivação e entusiasmo pelo trabalho desenvolvido ao longo desta tese foram contagiantes. As suas sugestões foram também fulcrais para o desenvolver do mesmo.

À Professora Doutora Maria João Ramos gostaria de agradecer a oportunidade que me proporcionou de integrar no fantástico grupo de trabalho que lidera.

Ao Doutor Sérgio Sousa, que na sua condição de coorientador, me proporcionou o acompanhamento, a disponibilidade e os ensinamentos de que necessitava para realizar esta tese.

A todos os atuais e antigos membros do fantástico grupo da Química Teórica e Bioquímica Computacional, pela ótima integração, ajuda e sugestões que me proporcionaram neste meu primeiro ano. Na memória ficarão momentos como as discussões sobre os mais diversos temas, almoços de grupo e momentos de desporto e lazer.

À Fabiola, João, Pedro Ferreira, Pedro Paiva, Ritinha e Rui, pela disponibilidade que demonstraram ao longo deste ano para me ajudarem a solucionar qualquer problema ou dúvida que surgissem e também pelos ensinamentos que me transmitiram.

Ao Daniel, Diogo, João, Pedro e Vítor, pela amizade e motivação de querer ser sempre melhor do que no dia anterior. Por estarem sempre disponíveis para animar os meus finais de semana.

À Alice, Ana, Anita, Beatriz, Hélio, Marisa e Samuel, pela amizade e momentos de diversão ao longo de todo o meu percurso académico. Por partilharem comigo a ansiedade dos exames, estágios e tese. A vossa amizade foi determinante para o meu sucesso.

Aos meus primos e tios, pelos momentos revitalizantes em família, e pela sensação de que, venha o que vier, eles estarão sempre no sítio certo, à hora certa.

Aos meus avós, pelo carinho e compreensão, por serem o meu porto seguro. Por assistirem ao meu crescimento pessoal e profissional com orgulho no olhar.

Ao meu irmão e cunhada, por serem o meu exemplo e apoio do dia-a-dia. Por trazerem ao mundo o meu afilhado e sobrinha, dois pequenos seres que já tanto me ensinaram sobre viver.

Aos meus pais, por me terem proporcionado a oportunidade de estudar e de alcançar todos os meus sonhos. A ti que estás a ver-me algures, sei que estás orgulhoso por ter alcançado sempre todos os objetivos que tracei, por isso esta tese é especialmente dedicada a ti.

À Márcia, a minha namorada e companheira de vida, pelo amor, amizade, compreensão, força e tudo aquilo que há entre nós. Por saber que, qualquer barreira que se meta no nosso caminho, será derrubada em conjunto. Espero poder concretizar todos os nossos sonhos e fazer-te sempre feliz, tu mereces isso e muito mais.

Abstract

Sulphur present in fuel is one of the major contributors for acid rain formation, release of harmful greenhouse gases and particulate matter emissions, the latter were found to be carcinogenic and involved in several respiratory health issues. Legislative measures are being applied on refineries for the reduction of sulphur in the final products of crude oil processing. Crude oil biodesulphurization has been an uprising topic over the last decades because it may overcome hydrodesulphurization largest technical issues: to perform the deep desulphurization of heavy and sour crude oil and the high operating costs. The 4S pathway is a set of reactions performed by the enzymes DszA, DszB, DszC and DszD on DBT and its derivates. DszB is the last enzyme of this pathway and it is responsible for the bond breaking between the carbon structure of DBT and the sulphur atom.

In this work we use ONIOM QM/MM to study the reaction mechanism of DszB by two different but complementary approaches: The single conformation QM/MM approach, that establishes the comparison between different ONIOM models from the same initial conformation of the enzyme; The multi-PES scan approach, that applies the same ONIOM model on different initial conformations of the enzyme. For the first, geometry optimizations and PES scans were performed applying B3LYP density functional to the QM layer with the 6-31G(d) basis set, then single point energy calculations were made using the same density functional but a different basis set, the 6-311+G(2d,2p). For the second approach, B3LYP 6-31G(d) was used for both geometry optimizations and PES scans. The MM layer was treated with the GAFF and ff99SB force fields in both approaches.

The single conformation QM/MM approach shows the importance of the stabilization effect of both His60 and Gly73 on Cys27 during the transition state and the stable intermediate state, which is translated on a ΔG‡ _{of 18.0 kcal.mol}-1 _{and a ΔG}

r of 8.8

kcal.mol-1 _{for model 4, significantly lower values when comparing to the other ONIOM}

models.

The multi-PES scan approach enables to see a difference in the rotamer of Cys27 on different initial conformations. The set of conformations which presents the Cys27 proton pointing towards the electrophilic carbon of DBT showed ΔE‡ _{values between 20 and 32}

kcal.mol-1 _{and the other set of conformations, which presents the same proton pointing}

This work shows the importance of His60 and Gly73 on the stabilization of Cys27 during the stationary points of the reaction mechanism of DszB as well as the impact that different initial conformations can have on the reaction kinetics.

The results presented in this thesis might be of value when it comes to further studies on DszB and can help to make viable the use of biodesulphurization on the oil refining industry.

Keywords:

2’-Hydroxybiphenyl-2-Sulfinate Desulfinase; DszB; Biodesulphurization; Sulphur; Crude oil; QM/MM methods; enzymatic catalysis; 4S pathway; Rhodococcus erythropolis.

Resumo

O enxofre que se encontra nos combustíveis é um dos principais contribuintes para a formação de chuvas ácidas, libertação de gases de efeito de estufa nocivos e emissão de partículas inaláveis, as quais são consideradas como sendo carcinogénicas e capazes de desencadear diversos problemas respiratórios. Medidas legislativas estão a ser aplicadas em refinarias para a redução do conteúdo de enxofre nos produtos finais do processamento de petróleo. A biodessulfurização do petróleo tem sido um tema cada vez mais discutido ao longo das últimas décadas, uma vez que pode ser capaz de ultrapassar os maiores problemas técnicos relacionados com a hidrodessulfurização, nomeadamente a dessulfurização profunda de petróleo heavy e sour e os elevados custos associados à manutenção desta técnica. A via enzimática 4S é um conjunto de reações desempenhado pelas enzimas DszA, DszB, DszC e DszD exercido sobre o DBT e seus derivados. A DszB é a última enzima desta via enzimática e é responsável pela quebra da ligação entre a estrutura de carbono do DBT e o átomo de enxofre.

Neste trabalho foi usado ONIOM QM/MM para estudar o mecanismo de reação da DszB através de duas abordagens diferentes, contudo, complementares: Na abordagem

single conformation QM/MM, uma comparação entre diferentes modelos ONIOM é

efetuada a partir de uma mesma conformação inicial da enzima; a abordagem

multi-PES scan utiliza o mesmo modelo ONIOM em diferentes conformações iniciais. Para a

primeira, as otimizações de geometria e as varreduras da PES foram efetuadas aplicando o funcional de densidade B3LYP na camada QM com a função de base 6-31G(d) e para os cálculos de energia foi utilizado o mesmo funcional de densidade, mas aplicando uma função de base diferente, a 6-311+G(2d,2p). Para a segunda abordagem, foi utilizado B3LYP 6-31G(d) tanto para as otimizações de geometria como paraos PES scans. Em ambas as abordagens a camada MM foi parametrizada com os campos de força GAFF e ff99SB.

A abordagem single conformation QM/MM mostrou a importância do efeito de estabilização exercido pela His60 e Gly73 sobre a Cys27 durante o estado de transição e o intermediário estável, o qual é traduzido num valor de ΔG‡ _{de 18.0 kcal.mol}-1 _{e ΔG}

r

de 8.8 kcal.mol-1 _{para o modelo 4, o que representa uma diminuição significativa}

comparativamente aos restantes modelos ONIOM.

A abordagem de multi-PES scans permitiu identificar uma diferença no rotâmero da Cys27 em diferentes conformações iniciais da enzima. O conjunto de conformações que apresenta o protão da Cys27 direcionado para o carbono eletrofílico do DBT mostra

valores de ΔE‡ entre 20 e 32 kcal.mol-1 _{e o outro conjunto de conformações, o qual}

apresenta o mesmo protão na direção contrária do átomo de carbono referido, possui valores de ΔE‡ _{acima de 40 kcal.mol}-1_.

Este trabalho mostra a importância da His60 e Gly73 na estabilização da Cys27 nos pontos estacionários da reação assim como o impacto que diferentes conformações iniciais da enzima podem ter na cinética da reação.

Os resultados apresentados nesta tese podem contribuir para futuros desenvolvimentos em trabalhos relacionados com a DszB e também ajudar a viabilizar a aplicação da biodessulfurização na indústria de refinamento de petróleo.

Palavras-Chave:

2’-Hidroxibifenil-2-Sulfinato Dessulfinase; DszB; Biodessulfurização; Enxofre; Petróleo; Métodos QM/MM; catálise enzimática; via enzimática 4S; Rhodococcus erythropolis

Table of Contents

Chapter 1 - Sulphur in fossil fuels ... 1

1.1. Introduction ... 1

1.2. Strategies for obtaining ultra-low sulphur levels in petroleum derivates ... 5

1.2.1. HDS ... 5

1.2.2. Oxidative Desulphurization (ODS) ... 6

1.2.3. Adsorptive Desulphurization ... 6

1.2.4. Selective Extraction Desulphurization ... 7

1.2.5. Biodesulphurization (BDS) ... 8

1.3. The 4S Pathway ... 9

1.3.1. Dibenzothiophene Monooxygenase (DszC) ...10

1.3.2. Dibenzothiophene Sulfone Monooxygenase (DszA) ...11

1.3.3. Flavin Reductase (DszD) ...11

1.3.4. 2’-Hydroxybiphenyl-2-Sulfinate Desulfinase (DszB) ...11

1.3.4.1. Structural Features ...12

1.3.4.2. The active site ...14

1.3.4.3. Enzyme characterization ...19

1.3.4.4. Mutagenesis studies ...19

Chapter 2 - Methods ... 25

2.1. Introduction ... 25

2.2. Molecular Mechanics ... 25

2.2.1. Force Fields ... 26

2.2.1.1. Amber Force Field ... 26

2.2.1.2. Bonding Energy ... 27

2.2.1.3. Angles Energy ... 27

2.2.1.4. Torsions Energy ... 28

2.2.1.5. Electrostatic Energy ... 28

2.2.1.6. Van der Waals Energy ... 28

2.2.2. Molecular Dynamics (MD)... 29

2.2.2.1. Integration step ... 30

2.2.2.2. Ensemble ... 30

2.2.2.3. Periodic Boundary Conditions ... 31

2.2.2.4. Non-bonded interactions cut-off ... 31

2.3. Quantum Methods ... 32

2.3.1. The Schrödinger Equation ... 32

2.3.2. Density Functional Theory (DFT) ... 33

2.3.2.1. Exchange-Correlation Functionals ... 35 2.3.2.2. B3LYP ... 35 2.3.2.3. Basis set ... 36 2.4. Hybrid Methods ... 36 2.4.1. QM/MM schemes ... 37 2.4.2. The interface... 38

2.4.2.1. The link atom approach... 38

2.4.2.2. Interaction between layers ... 38

2.4.3. QM/MM single-conformation approach ... 39

2.5. Computational Enzymatic Catalysis ...40

Chapter 3 - Results and Discussion ...41

3.1. Single Conformation QM/MM Study of DszB Reaction Mechanism ...41

3.1.1. Initial modelling of HBPS ...41

3.1.2. Study of DszB-HBPS system through MD ...44

3.1.3. ONIOM Model Preparation ...46

3.1.3.1. QM layer size increment ...47

3.1.3.2. Inclusion of Gly73 in the QM layer ...53

3.1.3.3. PES Scan analysis of different models...58

3.2. Multi-PES QM/MM Study of the first-step of DszB reaction mechanism ...59

3.2.1. Tier division of the structures from the 100 ns MD simulation ...60

3.2.2. PES scan study of the selected structures ...64

3.2.2.1. Post geometry optimization tier reorganization ...64

3.2.2.2. PES scan profile of tier 1 and 2 structures vs. tier 3 structures ...66

Chapter 4 - Conclusions and Future Perspectives ...71

4.1. Conclusions ...71

4.2. Future Perspectives ...72

Bibliography ...73

List of equations

Equation 1 - Potential energy function equation used in Amber force fields. ..27 Equation 2 - Atomic position equation. ...30 Equation 3 - Atomic velocity equation. ...30 Equation 4 - Time-dependent Schrödinger equation ...32 Equation 5 - Total number of electrons through integration of the electron density. ...33 Equation 6 - System total energy calculation according to DFT. ...34 Equation 7 - System total energy calculation according to DFT, including exchange and Coulomb terms. ...34 Equation 8 - System total energy calculation according to DFT, including the exchange-correlation functional. ...34 Equation 9 - Exchange-correlation functional according to B3LYP. ...35 Equation 10 - Subtractive ONIOM scheme ...38 Equation 11 – Transition state theory equation that establishes the relationship between the kcat and the ΔG‡ ...40

List of Tables

Table 1 - Temporal evolution of the sulphur emission standards for different regions of the world; a) Standards from California Air Resources Board (CARB); b) Standards from US Environmental Protection Agency (EPA); c) Standards from the Council of the European Communities; adapted from VC Srivastava [5]. ... 3 Table 2- Summary of the physical and kinetic properties of DszB from

Rhodococcus erythropolis KA2-5-1 described by Nakayama et al. [38]. ...19

Table 3- Parameterized charges for each atom of HBPS. ...42 Table 4 – Relevant interactions established between residues in the active site of DszB after the MD production phase. ...46 Table 5- Residues included in the QM layer for ONIOM models 1, 2 and 3. ...47 Table 6- Distances of the most relevant interactions obtained after the initial QM/MM optimization step on the reactant structure of ONIOM models 1, 2 and 3. ...49

Table 7- Distances of the most relevant interactions between residues of the active site in the TS structure of ONIOM models 1, 2 and 3. ...49 Table 8- Distances of the most relevant interactions between residues of the active site in the I1 structure of ONIOM models 1,2 and 3. ...51 Table 9- Residues included in the QM layer for ONIOM model 4. ...53 Table 10- Distances obtained from the initial QM/MM optimization step on the reactant structure of ONIOM model 4. ...54 Table 11 - Distances for the most relevant interactions between residues of the active site in the TS structure of ONIOM model 4. ...55 Table 12- Selected structures from the 100 ns MD simulation, ordered from the smallest d1+d2+d3 value to the highest. Tier 1, 2 and 3 are shaded in green, yellow and red, respectively. ...63 Table 13- Tier reorganization after the QM/MM geometry optimization and distances for each structure. Detailed information about the ΔE values can be found in Table S1. ...65

List of Abbreviations

AMBER – Assisted Model Building with Energy Refinement BDS – Biodesulphurization

BdsA – DBT sulfone monooxygenase from Bacillus subtilis WU-S2B BPS – Biphenyl sulfinate

CARB – California Air Resources Board

C---H – Distance between the electrophilic carbon of HBPS and the Cys27 proton d1 – Distance between the sulphur atom of Cys27 and the hydrogen atom from the

ε-amino group of His60

d2 – Distance between the sulphur atom of Cys27 and the peptidic amino group hydrogen atom of Gly73

d3 – Distance between the electrophilic carbon of HBPS and the Cys27 proton DBT – Dibenzothiophene

DBTO – Dibenzothiophene sulfoxide DBTO2 – Dibenzothiophene sulfone

DFT – Density Functional Theory

DszA – DBTO Monooxygenase from Rhodococcus erythropolis ITGS8 DszB – HBPS Desulfinase from Rhodococcus erythropolis ITGS8 DszC – DBT Monooxygenase from Rhodococcus erythropolis ITGS8 DszD – Flavin Reductase from Rhodococcus erythropolis ITGS8 EPA – Environmental Protection Agency

FCC – Fluid Catalytic Cracking GAFF – General Amber Force Field

GGA – Generalized Gradient Approximation GTO – Gaussian Type Orbitals

HBP – Hydroxy biphenyl

HBPS – 2’-Hydroxybiphenyl-2-Sulfinate HDS – Hydrodesulphurization

HF – Hartree-Fock

I1 – First Stable Intermediate IRC – Intrinsic Reaction Coordinate LDA – Local Density Approximation MD – Molecular Dynamics

NPT – Isothermal-isobaric ensemble NVT – Canonical Ensemble

ODS – Oxidative Desulphurization

OECD – Organization for Economic Cooperation and Development

ONIOM – Our own n-layered Integrated Molecular Orbital and Molecular Mechanics PES – Potential Energy Surface

QM – Quantum Mechanics

RMSd – Root Mean Square deviation SCF – Self-Consistent Field

SP – Single Point

STO – Slater Type Orbitals

TdsC – DBT Monooxygenase from Paenibacillus sp. A11-2 TS – Transition State

WAT1 – Water molecule interacting with the hydroxyl group of HBPS WAT2 – Water molecule interacting with the sulfinate group of HBPS

ΔE‡ _{– Energy of activation given from a geometry optimization associated with a near}

TS structure

ΔEr – Energy of activation given from a geometry optimization associated with an I1

structure

ΔG‡ _{– Gibbs Free Energy associated with the TS/ Gibbs Free Energy of Activation}

Amino acids Abbreviations

3-letter code 1-letter code

Alanine Ala A Cysteine Cys C Aspartic acid Asp D Glutamic acid Glu E Phenylalanine Phe F Glycine Gly G Histidine His H Isoleucine Ile I Lysin Lys K Leucine Leu L Methionine Met M Asparagine Asn N Proline Pro P Glutamine Gln Q Arginine Arg R Serine Ser S Threonine Thr T Valine Val V Tryptophan Trp W Tyrosine Tyr Y

C

hapter 1

Sulphur in fossil fuels

1.1. Introduction

This section is dedicated to some of the major environmental problems associated with the presence of sulphur in crude oil. The need for a new desulphurization method will be discussed.

Global consumption of crude oil and liquid fuels is expected to increase over the next decades, mostly because of the energetic needs of non-OECD countries. The 2017 International Energy Outlook predicted an 18% increase of the worldwide consumption of crude oil and its derivates between 2015 and 2040 [1].

The higher consumption in crude oil goes side-by-side with the release of noxious compounds, that are the result of the degradation of crude oil products such as gasoil. An example of those noxious compounds are sulphur oxides that are emitted to the atmosphere upon fossil fuels combustion causing several environmental problems, namely acid rain formation, particulate matter emission and higher release of greenhouse gases.

Acid rain formation is caused by the release of sulphur dioxide into the atmosphere, which leads to the formation of sulphuric acid. This environmental problem isn’t particularly harmful for humans, but it is a major contributor for monuments degradation, wildlife contamination and, most importantly, forest degradation. Despite acid rain not

being a detrimental agent for humans by itself, the presence of gaseous SO2 is a risk

factor associated with many respiratory complications [2].

Particulate matter is also associated with health issues. It is a mixture of the oxidation products of sulphur and nitrogen formed in the fuel exhaust of diesel transportations and it is highly associated with lung cancer [3].

Release of harmful greenhouse gases, like hydrocarbons and carbon monoxide, is also associated with the presence of sulphur in transportations fuel. SO2 is known to

poison the fuel catalyst which loses its oxidative properties, leading to larger greenhouse gases emission [4]. The emission of noxious substances into the atmosphere is particularly due to diesel combustion. Despite being more fuel-efficient than gasoline, diesel is also more susceptible to the formation of sulphur and nitrogen oxides upon combustion [3].

Legislation limiting the amount of sulphur on the petroleum derivates became stricter, in an attempt to reduce the emissions of hazardous gases and particulate matter into the atmosphere (Table 1).

Table 1 - Temporal evolution of the sulphur emission standards for different regions of the world; a) Standards from California Air Resources Board (CARB); b) Standards from US Environmental Protection Agency (EPA); c) Standards from the Council of the European Communities; adapted from VC Srivastava [5].

REGIONS SULPHUR EMISSION STANDARDS

CALIFORNIAa 1996-2000 90 ppm 2000-2006 80 ppm 2006-2010 15 ppm USAb 1996-2004 500 ppm 2004-2006 300 ppm 2006-2008 80 ppm 2008-2010 15 ppm EUc 1996-2001 500 ppm 2001-2005 150 ppm 2005-2008 50 ppm 2008-2010 10 ppm CHINAc 1996-1997 >500 ppm 1997-2001 500 ppm 2001-2006 150 ppm 2006-2010 50 ppm THAILANDc 1996-2001 >500 ppm 2001-2004 500 ppm 2004-2009 150 ppm 2009-2010 50 ppm INDIAc 1999-2000 >500 ppm 2000-2005 500 ppm 2005-2010 150 ppm 2010 50 ppm

Such standards are difficult to meet because building and maintenance costs of high extent hydrodesulphurization (HDS) (Fig. 1) facilities are extremely expensive. The extension of crude oil desulphurization is greatly influenced by the quantity of sulphur present in its source, meaning that the products of crude oil with high sulphur content difficultly achieve the sulphur levels demanded by legislation [5, 6].

Fig. 1 - Industrial HDS process on a thiophene molecule.

Crude oil can be classified in sweet, sour, light and heavy. Sweet and sour classification originated in the 19th _{century when early prospects would taste crude oil in}

order to determine its quality. Crude oil with better quality often had a sweet taste and pleasant smell, a fact that is related to the lower quantity of sulphur. Light and Heavy

classification is related with the molecular mass of the compounds that constitute crude oil. The lower the molecular mass of the compounds that constitute the crude oil matrix, the lighter the crude oil is. These aspects of the crude oil matrix play a major role on the extent of HDS since sour and heavy crude oil compounds, such as dibenzothiophene (DBT), have attached alkyl groups that sterically obstruct the hydrogenation reaction (Fig. 2).

Fig. 2- Lewis structures of methyl-substituted DBT’s.

Many sulphur compounds present in crude oil are not eliminated by the treatment to which crude oil is subjected to obtain its derivatives. From this diverse group of sulphur compounds, the thiophenes correspond to a fraction of 50-95% and are also the most problematic. This subgroup is comprised of thiophene, benzothiophene, dibenzothiophene (DBT) and alkylated DBT’s. The rate of consumption of fossil fuels leads to a fast depletion of reservoirs containing light and sweet crude oil, which has a low percentage of sulphur compounds. This reduction of high-quality crude oil, stricter legislation and problems regarding HDS of low-quality crude oil demand an improved desulphurization method for obtaining ultra-low sulphur fuels and other crude oil derivates [2, 3].

HDS is the currently employed method in oil refining industry for the removal of the sulphur from crude oil. It has major economical drawbacks when it comes to desulphurize low quality crude oil because of the very high energetic expenditure during the process and the loss of calorific value in the final product, due to the breaking of C-C bonds.[7] Other high yield desulphurization processes such as oxidative desulphurization, adsorptive desulphurization, selective extraction desulphurization and biodesulphurization have received considerable attention over the last two decades. A

lot of funding and research are being employed in overcoming some of the problems associated with this upcoming trend of crude oil processing, making it more appealing to the large oil processing companies and, at the same time, having smaller impact in the environment [8].

1.2. Strategies for obtaining ultra-low sulphur levels in

petroleum derivates

Crude oil companies are always developing new technologies that can allow them to keep up to date when it comes to the refining process. Parameters like the quality of the feedstock, refining cost and environmental standards are some of the most important pressures these industries meet. Since the quality of the feedstocks are overall decreasing and environmental pressure is becoming tighter, the currently employed technologies are becoming outdated and the search for more cost-efficient and greener technologies has been given a lot more attention and investment [9].

1.2.1. HDS

HDS has been the method of choice for removing sulphur from crude oil for the past century, it made possible to decrease the amount of sulphur in crude oil to about 15 ppm, while applying high temperature and pressure conditions. However, this technique only works well on some types of crude oil. Those that are rich in recalcitrant species like thiophenes and DBT cannot be desulphurized below 15 ppm using this technique [2]. In the overall HDS reaction of thiophene, DBT and their derivates, the sulphur atom is removed by hydrogenation of the substrate while employing a metal sulfide catalyst. Temperatures ranging between 300 to 350 ºC and pressure from 50 to 100 atm are usually employed to enhance the reaction efficiency [7].

When HDS is applied on DBT, other problems arise, such as the inhibition by the product and the large difference on reactivity between differently alkylated DBT compounds. The major product of this reaction is biphenyl, but the hydrogenation can also occur on the phenyl rings giving cyclohexylbenzene. The degree of substrate substitution usually decreases the reaction rate because substitutes sterically hinder the bond breakage from H2, in particular the 4,6-DMDBT (Fig. 2) substitute is reported to

highly refractory to the desulphurization activity [7, 10].

HDS is still of use to the refining of light crude oil since it has good performance when treating low molecular mass sulphur molecules, although this technique is becoming outdated due to the proliferation of poorer quality feedstocks.

1.2.2. Oxidative Desulphurization (ODS)

This approach is based on the catalysed oxidation of the sulphur compounds in crude oil to sulfoxides and sulfones, altering their physical properties and enabling the removal of the molecule as a whole or the respective free oxidized sulphur[11, 12].

The oxidation reaction (Fig. 3) requires an oxidative reagent such as oxygen, nitrogen dioxide, hydrogen peroxide and a catalyst, usually formic and acetic acid. There are essentially two kinds of ODS processes: the liquid phase processes and the gas phase processes.

Liquid phase ODS processes consist on the oxidation of the sulphur containing molecules, altering their polarity and making it possible to separate them from the rest of the feed. The process happens at temperatures of less than 120 ºC, at which the feedstock remains in the liquid state [12, 13]

As for the gas phase processes, the oxidized sulphur containing molecule is adsorbed in the gas phase and the separation occurs through condensation of the remaining feed. In order to evaporate the feed, for further separation of the sulphur containing molecules, it is needed to employ high temperatures, ranging from 180 ºC up to 400 ºC [14].

Fig. 3- Oxidation of DBT through the ODS process.

The advantages of ODS over HDS for obtaining ultra-low sulphur products are the use of cheaper oxidative agents than hydrogen, the fact that the operation occurs at milder reaction conditions and its ability to overcome the sterical hindrance of DBT and its derivatives for the desulphurization reaction [12, 15].

1.2.3. Adsorptive Desulphurization

Adsorptive desulphurization is a method based on the removal of the sulphur compounds using an adsorbent agent. Physical adsorption is based on non-bonding interactions between the adsorbent and the organosulphur compound. Reactive adsorption occurs when there is some kind of reversible chemical reaction between both species [16].

A. Physical Adsorption

The polarity of the molecules containing sulphur is one of the properties that can be used in order to adsorb sulphur compounds. This strategy lacks in specificity since the fuel matrix is composed of various polar and aromatic compounds. Also, in this case the whole molecule is separated, resulting in calorific loss in the final product due to the loss of the carbon skeleton of the sulphur containing molecules. Other approach based on the physical properties of the molecules is the selective adsorption of sulphur aromatic compounds onto an organometallic matrix. The principle behind this strategy is based on the selective orientation of the interaction between thiophene and the organometallic adsorbent. This process occurs at environmental temperature and does not require an H2 atmosphere, but the carbon skeleton attached to the sulphur atom is removed with it,

resulting in a less energetic product [6, 17].

B. Reactive adsorption

Reactive adsorption (Fig. 4) is based on the principle that the organosulphur molecule adsorbs and reacts with the adsorbent agent, usually a metal, removing the sulphur atom from the rest of the molecule by establishing a bond with the sulphur atom. Both products are then separated, and the adsorbent is recycled back to be used in further desulphurization cycles. This method requires high operating temperatures [6, 18].

Fig. 4- General example for the desulphurization of organosulphur compounds. Adapted from Song et al. [6].

1.2.4. Selective Extraction Desulphurization

In a crude oil refinery, fluid catalytic cracking (FCC) is a typical procedure in which high molecular mass compounds are broken into smaller hydrocarbons, resulting in the production of smaller hydrocarbons like gasoline and olefin. After the previously mentioned procedure, the hydrocarbon stream is lead to a HDS facility in which the products are hydrotreated and washed with a caustic agent. This not only removes sulphur from the products but also damages the smaller hydrocarbons removing some of the properties of interest for the calorific yield of the hydrocarbon stream. In order to solve this problem, a selective extraction desulphurization method was developed to separate sulphur containing compounds from the other hydrocarbon compounds prior to the HDS procedure (and post FCC), enabling a more selective HDS process and fewer

calorific loss. However high operating costs and the extreme operating conditions are still associated to this method [6, 19].

1.2.5. Biodesulphurization (BDS)

BDS is the method in which biocatalysts are employed in order to remove the sulphur moiety of the molecule of interest. In this regard, bacterial species that use sulphur as a source of energy or as a metabolic intermediate are the ideal players in this subject.

Rhodococcus and Pseudomonas strains are some of the most studied in this field [11,

20, 21].

The concept of biorefining is related to the treatment and upgrading of crude oil and its products using living organism and/or their molecular machinery, thus preventing the pollution associated to other means for refining and reducing the high operating costs of the currently employed methods [22].

The advantages of BDS over other desulphurization methods are its cost efficiency and “greener” methodology. From an economical point-of-view, BDS offers the possibility of reducing the energetic requirements in relation to other methodologies, such as HDS. The latter has huge energetic requirements, the hydrocarbon stream must be heated to about 450ºC in order to accomplish a deep desulphurization level of about 50 ppm. These high temperatures also imply higher costs in facility maintenance and upgrading. BDS should be seen as a complementary methodology to HDS since the first method can be used to desulphurize DBT’s and the second has its higher efficiency levels on mercaptans, thiophenes and benzothiophenes, allowing the reduction of sulphur in fuel to ultra-low levels [2].

BDS also offers a more ecological alternative to HDS since the CO2 emissions are

greatly reduced through reduced combustion of methane for energetic and hydrogen requirements, fulfilling the Paris agreement demands for the reduction of greenhouse gases [23-25].

It has been found that Rhodococcus erythropolis strain ITGS8 has a peculiar metabolic pathway which enables it to remove the sulphur from DBT without degrading its energetic content. This enzymatic system employs 4 different enzymes (DszA, DszB, DszC and DszD) and is known to follow a series of metabolic reactions known as the 4S pathway, removing the sulphur atom from DBT and transforming it into hydroxy biphenyl (HBP).[26-28].

Despite being a promising technology for the removal of sulphur from crude oil, these microorganisms exhibit a slow rate of desulphurization, making it a non-profitable process for refineries to employ. In order to solve this problem, research has been made

on 2 main strings: creating improved derivatives of the Rhodococcus erythropolis strain IGTS8 [29]; increase enzymatic activity of the molecular entities of the 4S pathway through rational protein engineering. A better understanding of each enzyme of this pathway would be critical in this regard. [27, 30].

Genetic and enzymatic studies have been made concerning this problem and new alternatives have emerged as a complement to HDS [31].

1.3. The 4S Pathway

The 4S pathway (Fig. 5) consists in the enzymatic cascade formed by the Dsz enzymes, which work sequentially to oxidize the C-S bond and finally remove the oxidized sulphur as H2SO3 [2]. DszA and DszC are monooxygenases that oxidize the

sulphur from DBT. DszD provides the essential cofactor to these oxidations and, finally, DszB removes the oxidized sulphur from the carbon skeleton [27, 32].

Fig. 5- Schematic representation of the enzymatic pathway involved in the removal of sulphur from dibenzothiophene, the 4S pathway.

1.3.1. Dibenzothiophene Monooxygenase (DszC)

DszC is the starting enzyme in the 4S pathway and it is responsible for the initial double oxidation of the sulphur atom of DBT to form DBT sulfone (DBTO2). This member

of the class D flavin monooxygenases functions as a tetramer formed by two homodimers. The active site is located in between the two homodimers and is

responsible for accommodating the substrate as well as the cofactor FMNH2. For the

oxidation reaction to occur, the cofactor has to be activated by O2 forming the

C4a-hydroperoxyflavin intermediate. The actual catalytic mechanism and role of the active site residues are not clearly defined, although recent crystallographic studies on homologous enzymes made it possible to obtain a crystal with the ternary complex of DBT-TdsC-FMN and DBTO-TdsC-FMN. Further computational studies on this ternary complex could elucidate the catalytic mechanism of this enzyme [33-35].

1.3.2. Dibenzothiophene Sulfone Monooxygenase (DszA)

The second enzyme of the pathway performs the oxidation of the sulphur atom, transforming DBT sulfone into 2’-hydroxybiphenyl 2-sulfonic acid (HBPS). DszA belongs to the superfamily of the group C flavoprotein monooxygenases whose scaffolds are highly conserved, for example, the TIM barrel is one of the characteristic tertiary structures retained by this enzyme. The main difference between this group of monooxygenases is the active pocket which varies in volume depending on the natural substrate of each enzyme. BdsA, a DBT sulfone monooxygenase from the organism

Bacillus subtilis WU-S2B, is a homologous to DszA and it also catalyses the oxidation of

the sulphur atom of DBT sulfone to form HBPS, using FMNH2 as a cofactor. Since these

two proteins share a high degree of sequence identity, it was suggested that they might also share the same catalytic mechanism. Crystallographic studies on BdsA provide the three-dimensional view of the complex BdsA-FMN-DBTO2, so that it can be possible to

study its catalytic mechanism[36].

1.3.3. Flavin Reductase (DszD)

Both DszC and DszA use FMNH2 as a cofactor in order to oxidize DBT. This cofactor

is a result of the two-step reduction reaction of FMN, which is carried out by DszD, a flavin reductase, consuming NADH in the process. Previous computational studies have already determined the reaction energetic profile and elucidated the role of some of the active site residues as well as proposed mutations that could augment the rate constant of DszD, with the employment of a rationale and a methodology similar to the one employed in this work [27, 32, 37].

1.3.4. 2’-Hydroxybiphenyl-2-Sulfinate Desulfinase (DszB)

The final step of the reaction is the actual breaking of the C-S bond and is catalysed by the DszB enzyme. This desulphurization reaction is uncommon among enzymes. In fact, other enzymes like cysteine sufinate desulfinase and L-aspartate β-decarboxylase

also desulphurize substrates but, unlike DszB, they use pyridoxal 5’-phosphate as cofactor. The unique characteristics of this enzyme make it difficult to properly identify its catalytic mechanism through structural and biochemical analyses [31].

The work developed throughout this thesis was dedicated to the study of DszB and its reaction mechanism. The following topics will be used to explain important structural features of the enzyme, alongside with the study of the active site structure, enzymatic kinetics and previous mutagenesis and reactional studies.

1.3.4.1. Structural Features

Crystallographic structures from wild type DszB and DszB C27S mutant complexed with both HBPS and biphenyl sulfinate (BPS) made it possible to identify unique structural features of DszB and start to unravel the secrets of this enzyme [38].

DszB has 365 amino acid residues and is an oval-shaped monomeric enzyme. It has 2 different domains (A and B) each formed by one 5-stranded β-sheet, which is surrounded by α-helices (Fig. 6). This three-dimensional structure resembles the one of other substrate-binding proteins like ovotransferrin and bacterial ABC transporter of sulphate, suggesting that it may have evolved from proteins of this subset.

Fig. 6- Crystallographic structure of DszB (PDB code 2DE3). New Cartoon representation shows Domains A and B, colored in blue and red, respectively. HBPS is shown in van der Waals representation, binding between

1.3.4.2. The active site

The enzyme active site is located between the two domains, and each one of them has different physical properties. The half of the active site pocket belonging to domain B is highly hydrophobic, with residues like Leu152, Trp155, Gly183 and Phe203. These residues, alongside with Pro28 and Phe61 from domain A, interact through van der Waals interactions with the aromatic rings of the substrate (Fig. 7).

Fig. 7- In licorice representation are shown the amino acids that establish van der Waals interactions with the hydrophobic part of HBPS.

Domain A comprises the reactive half of the active site which is involved in the removal of the sulphinate group from the substrate. This part of the pocket is mostly constituted by polar amino acids such as Ser27 (Cys27 in the wild type DszB), His60 and Arg70. The interactions between the previously mentioned residues and the sulphinate group form a hydrogen bond network that anchors the substrate and is responsible for its orientation within the active site (Fig. 8).

Fig. 8- Hydrogen bond network between HBPS and the residues from domain A, shown in licorice representation.

The binding pose of HBPS and BPS in the active site is similar, indicating that the hydroxyl group resultant from the oxidation of DBTO2 to HBPS isn’t meaningful to the

accommodation of the substrate in the active site. However, both substrates induced an important conformational change in a loop forming an α-helix near the active site. This change is important due to the translocation of His60 to the active site which allows it to interact with both substrate and Ser25. Alongside with this conformational change the reactive residue Cys27 is exposed to the substrate, preparing it for the reaction (Fig. 9).

Fig. 9- Translocation of His60 to the reactive site upon formation of the helix when the substrate is included in the crystal. New cartoon representation of DszB co-crystallized with HPBS (PDB code 2DE3) and DszB

Close analysis of the surroundings of Ser27 showed that, apart from the close interaction with His60 (3.67 Å), it also interacts with the main chain NH group from Gly73 (3.09 Å). The way these groups are directed towards the reactive cysteine is reminiscent of the formation of an oxyanion hole, although in the native enzyme this structural feature would be denominated as sulfanion hole, after the deprotonation of Cys27 (Fig. 10)

A crystallographic water molecule is found in close interaction (approximately 1.9 Å) with the sulphinate group of the substrate. It is known from experimental data that the sulphinate group is released as HSO3- . It is possible that this water molecule is the one

that forms this product upon the desulphurization reaction (Fig. 11), through an attack of a water-derived hydroxyl ion to the leaving sulphur dioxide.

Fig. 11- CPK representation of the water molecule that is in close interaction with the sulfinate group from HBPS.

1.3.4.3. Enzyme characterization

Experimental studies were conducted on DszB from Rhodococcus erythropolis KA2-5-1, whose nucleotide sequence is 99.9% similar to the DszB from Rhodococcus

erythropolis IGTS8. Its characterization revealed important characteristics which are

summarized in Table 2.

Table 2- Summary of the physical and kinetic properties of DszB from Rhodococcus erythropolis KA2-5-1 described by Nakayama et al. [39].

The turnover of the enzyme indicates that it is the slowest in the 4S pathway and also the most unstable, since it starts to lose its function once the environment temperature increases above 30ºC [39]. Other important feature of this enzyme is that it does not use any cofactor or requires another protein for the reaction to occur, unlike other similar enzymes involved in the metabolism of sulphur compounds. BPS can also be desulphurized by DszB indicating that the hydroxy group in HBPS is not relevant for the reaction to take place [26, 39]

1.3.4.4. Mutagenesis studies

Site-directed mutagenesis studies are very important because they help to evaluate the importance of each specific residue of the active site. The importance of residues like Cys27, His60 and Arg70 was characterized and the mutants were compared with the native enzyme in order to see the degree of contribution of each one of them.

NAKAYAMA, N. 2002

THERMOSTABILITY Stable up to 28ºC and rapidly decreases after 30ºC

pH stable pH=6-11 and retained 30% of its maximum

activity at pH=12.8

KM 8.2 µM or ΔGbind=-7.1 kcal.mol-1 for 2-HBPS

kcat 0.123 s-1 or ΔG‡=19.0 kcal.mol-1

KI

0.25 mM or ΔGbind=-5.0 kcal.mol-1 for 2-HBP

2 mM or ΔGbind=-3.7 kcal.mol-1 for 2,2'-DHBP

0.4 mM or ΔGbind=-4.7 kcal.mol-1 for 2,3-DHBP

OBSERVATIONS Metal ions are not important for enzyme activity; The

The C27S mutant completely lost its desulphurization activity but it maintained the structural integrity of the enzyme, meaning this is a crucial residue for the reaction to take place. Since this mutation does not alter the three-dimensional structure of the enzyme, this was the mutant of choice for the crystallographic work of Lee et al. were they crystallize this inactive mutant with the substrate of the reaction [38].

H60Q mutation resulted in a relevant loss of the enzyme activity (17-fold reduction). As was previously mentioned, this residue is important since it belongs to the hydrogen bond network responsible for the orientation of the substrate within the active site. The lack of interactions with the substrate might be the answer for the inefficiency of the mutant enzyme.

The mutation of R70I or R70K disrupted the three-dimensional structure of the enzyme, alongside with a complete loss of activity [38]. The Arg70 residue is part of the RXGG motif which is found in homologs of DszB (Fig. 12) and it forms the structural core of domain A. Mutations on this motif may cause a disruption of the structural integrity of the enzyme making it unviable.

Fig. 12- Amino acid sequence of DszB and its homologous enzymes with highlighted RXGG motif as well as other strictly conserved residues, such as the reactive Cys27 and His60 which is an important residue in the

establishment of the hydrogen bond network. Adapted from Lee et al. [38].

Lee et al. [38] performed another mutation to study the effect of a glutamate residue near the reactive cysteine hypothesizing that it could function as a general base to abstract the cysteine proton. They observed that the E192Q mutant did not lose the ability nor the efficiency in desulphurizing HBPS, denying the hypothesis that Glu192 could be the general base of the reaction.

Later Oshiro et al. [40] developed two mutants, Y63F and Q65H, with the aim of increasing DszB efficiency. The first revealed an higher kcat (0.234 s-1) and smaller KM to

the native substrate (5.6 µM), meaning a higher activity and better affinity when comparing to the native enzyme (Table 2).I It was postulated that this effect could be due to the enhanced hydrophobic interaction between the aromatic rings of HBPS and the apolar aromatic ring of phenylalanine when compared to that of tyrosine. Despite the

Gln65 residue being considerably distant from the active site, the mutation Q65H had positive effects on the thermostability of the enzyme. This effect might be due to the stabilization effect induced by the His charge. Seeing that both mutations granted enhanced characteristics to the enzyme, they created a double mutant. When comparing to the native DszB, the Y63F-Q65H mutant had higher thermostability, since it retained 90% of its relative activity after being heated at 35ºC for 30 min and maintained activity when submitted to higher temperatures. Although its kinetic properties were diminished when comparing to native DszB, the KM and kcat were of 9.8 µM and 0.169 s-1,

respectively [40].

1.3.4.5. Reaction Mechanism

DszB has already been biochemically characterized and crystallized. The next step is to unravel its catalytic mechanism or, in other words, understand how it breaks the bond between the carbon and sulphur atoms of HBPS at the atomic level. In that sense, mutagenesis studies help to identify the role of active site residues.

Until now the relevant mutants for the desulphurization reaction itself are:

• C27S, which revealed the central role of the reactive cysteine in the reaction once its replacement abolishes DszB activity [38];

• H60Q, which identified the histidine residue as an important factor for a faster turnover rate [38];

• Y63F, mutant that had better substrate affinity and higher enzymatic specific activity [40].

The other mutations (R70I/R70K and Q65H) are also relevant in the sense that they are crucial for the structural integrity of the enzyme.

Based on this information, two different hypotheses for the Reaction Mechanism arose:

A. Nucleophilic addition mechanism

Lee et al. noticed that Ser27 was well oriented for the nucleophilic attack on the sulphur atom of HBPS. They suggest that the cysteine may form a thiosulfonate-like intermediate with the sulphur moiety of HBPS, which would then be hydrolysed to form HSO3- (Fig. 13) [38].

Fig. 13- Schematic representation of the nucleophilic attack mechanism proposed by Lee et al. [38].

B. Electrophilic Aromatic Substitution

Other hypothesis proposed by Gray and co-workers is that the reactive cysteine is deprotonated by the electrophilic carbon of HBPS, forming a tetrahedral carbocation intermediate and then releasing the SO2, which reacts with water forming HSO3-. This

hypothesis is also supported by the fact that the inductive effect provided by ortho and

para methyl groups on DBT stabilize the carbocation formed throughout this mechanism

(Fig. 14) [29].

Fig. 14- Schematic representation of the protonation mechanism proposed by Gray et al. [29].

Recently, a computational study using a DFT cluster model approach has suggested that the enzyme may operate through the electrophilic aromatic substitution mechanism

on the electrophilic carbon of the substrate, rather than a nucleophilic attack on the sulfinate group, suggesting also other key residues for the reaction, while leaving several interesting questions unanswered [31].

In this work, we have performed an atomic-level analysis of the reaction catalysed by DszB. QM/MM simulations were used to obtain energy profiles for the electrophilic aromatic substitution mechanism. The specific role of different active-site residues of DszB was elucidated. The knowledge gained from these results opens the way for rational enhancements of DszB activity through site-directed mutagenesis.

C

hapter 2

Methods

2.1. Introduction

The study of chemical reactions in general has benefited a lot with the evolution of computational methods and of the machines in which they are executed. The use of computers alternatively to experimental studies to predict the mechanisms of chemical reactions enables a faster and more complete characterization of the reaction itself, without the high costs of expensive reagents and materials, while also enabling an atomistic view of the reaction pathway during the analysis. Experimental analysis of these reactions also provides relevant data that can be translated with fundamental aid for a theoretical approach, clarifying some concepts that could be hidden behind graphs and tables.

This chapter will be used to briefly explain the methods currently used in computational chemistry with a particular focus on their applicability on the study of enzymatic catalysis.

2.2. Molecular Mechanics

Despite the enormous development of computational power, the use of Quantum Methods in increasingly larger systems, with thousands of atoms, like enzymes, is still

impossible. This limitation makes the use of Molecular Mechanics a viable method to study enzymes and their different conformations.

Molecular mechanics applies the Born-Oppenheimer approximation, which considers that both the nucleus and the electrons can be merged in a single atom-like particle This approximation is then applied to the whole system, treating the atoms and the bonds between them as spheres connected by springs and making it possible to predict their behaviour by using the classical laws of physics [41, 42].

2.2.1. Force Fields

Force Fields are defined by a potential energy function and by a set of parameters for the energy contribution of the different terms of that potential energy function. Throughout the past decades, numerous kinds of force fields were created with different purposes, but their general application is to calculate the total energy of a user defined system as a function of the atomic coordinates- the potential energy function. The most common biomolecular force fields like AMBER, CHARMM, OPLS and other force fields apply the same general potential energy function to the whole system, despite containing minor variations between each other, which can be resumed to differences in the form of the potential energy function and the parameters used [43].

2.2.1.1. AMBER Force Field

AMBER (Assisted Model Building with Energy Refinement) was created in the 1980’s by the Kollman group [5]. The early parametrization of certain molecules, through quantum mechanical calculations and experimental data collection, was in the origin of the development of the first AMBER Force Field. This first version of the Amber Force Field was a united atoms force field, which implicitly treated hydrogen atoms, merging them with heavy atoms to which they were bonded, treating that set of atoms as the system smallest unit. A second generation force field was created by Cornell and co-workers [44]. This force field included parametrized sets of biomolecules like amino acids and nucleic acids and small fragments that included heavy atoms as well as hydrogens, fixing most of the problems associated with the previous one. It also included a new method for the calculation of the atomic charges, the Restrained Electrostatic Potential (RESP) charges [43, 45].

Since then, new AMBER force field variations have been created for simulating the protein behaviour as well as small ligands. In particular, the General Amber Force Field (GAFF) contains parameters for all the most common atom types present in small

organic molecules and drug-like compounds, improving the ability to simulate systems involving such molecules, particularly when interacting with proteins or enzymes. [43].

In this work, both AMBER ff99SB and GAFF were used to perform all the steps regarding the Molecular Mechanics calculations.

Since the number of atoms in a protein is too high for its Potential Energy Surface to be calculated using Quantum Mechanics, Equation 1 can be used to describe in a simple manner the potential energy function used in the AMBER force fields. The potential energy of the interaction between atoms 1-2 and 1-3 (Fig. 15) is treated according to a series of harmonic potentials which are based on the Hooke’s law for elastic properties, while for describing the rotation around the bond between atoms 2 and 3, it is used a periodic energy function (Fig. 15).

The Non-Bonding terms describe the interactions that can be established between atoms that are not covalently bonded, much like the interaction between atoms 3 and 5 in Fig. 15. The Non-Bonding terms also apply to atoms within the same molecule if they are separated by more than three consecutive covalent bonds.

These can be further differentiated into Electrostatic or van der Waals Interactions and are calculated through the Coulombic and Lennard-Jones potential, respectively [43, 46]. 𝑉(𝑟) = ∑ 𝑘𝑏 (𝑏 − 𝑏0)2 𝑏𝑜𝑛𝑑𝑠 + ∑ 𝑘𝜃 (𝜃 − 𝜃0)2 𝑎𝑛𝑔𝑙𝑒𝑠 + ∑ 𝑘𝜙[𝑐𝑜𝑠 (𝑛𝜙 − 𝛿) + 1] 𝑡𝑜𝑟𝑠 + ∑ [𝑞𝑖𝑞𝑗 𝑟𝑖𝑗 +𝐴𝑖𝑗 𝑟_𝑖𝑗12− 𝐶𝑖𝑗 𝑟_𝑖𝑗6] 𝑛𝑜𝑛−𝑏𝑜𝑛𝑑

Equation 1 - Potential energy function equation used in Amber force fields.

2.2.1.2. Bonding Energy

The bonds term calculates the potential energy of stretching or shortening of a bond between atoms 1 and 2 (Fig. 15). It does so by multiplying a force constant kb by the

square of the bond elongation relative to an equilibrium state b0.

2.2.1.3. Angle Energy

In a similar fashion, the angles term is calculated applying a force constant kθ

associated to the angle between atoms 1,2 and 3 relating it to the equilibrium angle θ0

2.2.1.4. Torsion Energy

The torsional energy is last term of the bonding interactions and has to be described through a periodic energy function. This term describes the energy for the rotation of atom 1 relative to atom 4 (Fig. 15).In this term k𝜙 defines the energetic barrier associated

to the rotation of the dihedral, n is the number of minima that the function can assume and 𝛿 is the angle phase.

2.2.1.5. Electrostatic Energy

This potential considers that the electron density and the nucleus of an atom can be condensed to a point charge, and then uses the Coulomb law to calculate the interaction between the point charges, providing the energy as a function of the charge (𝑞𝑖𝑞𝑗)

distance between them (𝑟𝑖𝑗).

2.2.1.6. Van der Waals Energy

The van der Waals interactions are originated when there exists a perturbation in the electronic cloud of the molecules due, for example, to the movement of the electrons. This perturbation creates an instantaneous dipole which itself induces an instantaneous dipole in non-bonding atoms around. The two dipoles then suffer a slight attractive force. The energy of this interaction is dependent on the distance between the atoms (𝑟𝑖𝑗) and

on the parameters 𝐴𝑖𝑗 and 𝐶𝑖𝑗 which can be experimentally determined.

2.2.2. Molecular Dynamics (MD)

MD is a method that enables the study of biomolecular systems phase space. In this regard, one can study specific enzyme movements and their evolution with time.

With the advent of computer technology, it became possible to study larger phase spaces and larger systems. Nowadays it is possible to study dynamic processes of proteins such has protein stability, rearrangements, folding and transport across membranes.

In MD, the simulation time is highly important because the movements of a certain biomolecular system are time dependent. There are many different types of protein movements (such as the ones mentioned above) that, in order for them to be studied, the MD simulation time has to be modulated accordingly. Basically, it has to be much larger (about 30 times) than the time for the smallest motion of interest.

Computer technology now supports simulation times up to hundreds of nanoseconds in medium-size proteins, which is adequate to study average-timescale molecular movements such as rotamer and loops movements.

The major advantages of classic MD are the possibility to treat larger systems without compromising computational efficiency, its ability to determine with relative ease dynamic and thermodynamic properties of the system and to offer the user a visual perspective of the system evolution with time. Despite being a useful methodology, it also has major downsides such as the need for the parametrization of all the components within the system, which can differ from a force field to another and make it impossible to be transferable in practice; its lack of precision when compared to results obtained from Quantum Methods (depending on the quality of the parameterization); finally, since the force fields do not explicitly include electronic behaviour, it is impossible to directly study chemical phenomena such as covalent bond breaking and formation.

At an early stage of this work, the MD simulations were produced to find an enzymatic conformation in which the distance of the side chain of Cys27 was close enough to the electrophilic carbon of HBPS for the reaction profile to be studied through QM/MM methods. Later, another phase of MD was performed to obtain a number of different conformations to study the influence of those conformations on the reaction profile.

The MD basis is Newton’s Second Law of Physics, in which the movement of the atoms is predicted by the integration of Newton’s equation, 𝐅⃗ = 𝑚. 𝐚⃗⃗. If the integration step (Δt) is defined, then it is possible to numerically solve the position and velocity equations (Equation 2 and Equation 3, respectively), for all the atoms within the system.

𝑟𝑡

⃗⃗⃗(𝑡0+ 𝛥𝑡) = 𝑟⃗⃗⃗(𝑡𝑡 0) + 𝑣⃗⃗⃗⃗(𝑡𝑡 0) .

1

2𝑎⃗⃗⃗⃗(𝑡𝑡 0) × 𝛥𝑡

2

Equation 2 - Atomic position equation.

𝑣𝑡

⃗⃗⃗⃗(𝑡0+ 𝛥𝑡) = 𝑣⃗⃗⃗⃗(𝑡𝑡 0) + 𝑎⃗⃗⃗⃗(𝑡𝑡 0) × 𝛥𝑡

Equation 3 - Atomic velocity equation.

2.2.2.1. Integration step

The integration step (Δt) is a user defined variable and it represents the timestep between each time that the equations of motion are calculated. It assumes the approximation that the forces acting on the system remain constant between each integration step, which is relatively small when taking into consideration enzyme movements. The accuracy of the movement prediction is dependent on the size of the integration step. The shorter it is, the more accurately the position of the atoms will be predicted.

For the accurate prediction of the atomic positions and velocities, it is considered a time step 10 times smaller than the fastest movement of the system, which corresponds to the vibration of the bond between the carbon and hydrogen atoms. A complete cycle of shortening and stretching for this bond happens in about 10 fs, and so the integration step is usually set to 1 fs. This variable can be increased with the use of the SHAKE algorithm, which constrains the bonds between heavy atoms and hydrogen atoms at the beginning of each integration step. In this sense, the vibration of bonds between second row atoms become the fastest movement of the system, completing a cycle in around 20 fs, which result in an integration step of 2 fs, enabling the calculation of longer simulation times.

The only limitation on the choice of the integration step is the computing power, once that, for the same simulation time, more integration steps mean a longer time to compute the full trajectories of the atoms.

In this work we use an integration step of 2 fs allowed by the use of the SHAKE algorithm, although the system properties output and the molecular trajectories were written every 20 ps.

2.2.2.2. Ensemble

For the macrostate of the system to be exactly defined, an infinite number of microstates has to be subjected to α+2 state functions (being α the number of components of the system), which have to be constant throughout those microstates.

Since that it is impossible to calculate an infinite number of microstates, a very large number of them is a somewhat reasonable approximation. This very large number of microstates under constant α+2 state functions is the definition of “ensemble”.

Different types of ensembles can be used to study the macrostate. In both phases of the MD simulations the canonical ensemble (NVT) is used for the equilibration procedure, in which the number of particles (N), volume (V) and temperature (T) are constant, and the isothermal-isobaric ensemble (NPT) is used for the production procedure, in which the constant state functions are N, pressure (P) and T.

2.2.2.3. Periodic Boundary Conditions

Molecular simulations on the molecule of interest, for example a protein, are carried out on a TIP3P box, a geometrically shaped simulations cell using water molecules to solvate the protein. The problem with this approach is that the molecules nearby the limit of the simulation cell have different physical properties than those closer to the middle of the cell.

One way to solve this problem is to ignore the contribution of the water molecules nearby the limit of the simulation cell, but this solution would only create another major problem which is ignoring the contribution of up to 40% of the molecules of the system, resulting in a deficient prediction of the system’s behaviour.

The employment of periodic boundary conditions is the approach used to appropriately solve this problem. This approach consists on the infinite replication of the periodic cell. The movement equations are solved for the central cell and the replica are subjected to the same movement. In order for this method to be effective, the distance between the protein and the limit of the cell has to be significant enough to discard the possibility of non-bonded interactions between proteins from different cells. In this work, we use the distance of 12.0 Å between the edges of the protein and the surface of the cubic simulation cell.

2.2.2.4. Non-bonded interactions cut-off

Non-bonded interactions are described by both electrostatic and van der Waals terms, although the energy of the interaction does not have the same dependence on the distance between two atoms. While the van der Waals interactions rapidly decrease with the increment of the distance between the atoms (proportional to r -6_{), the electrostatic}

interactions energy decreases at a much slower pace (proportional to r -1_{), since these}