João Pedro Leonor Fernandes Saraivafrpinto/fr...João Pedro Leonor Fernandes Saraiva Reconstruction of a generic metabolic model for Streptococcus pneumoniae Dissertação de Mestrado

João Pedro Leonor Fernandes Saraiva

Reconstruction of a generic metabolic model for Streptococcus pneumoniae

Dissertação de Mestrado

Mestrado em Bioinformática

Trabalho realizado sob a orientação de

Doutora Isabel Rocha

Doutor Francisco Pinto

October 2012

É AUTORIZADA A REPRODUÇÃO INTEGRAL DESTA TESE APENAS PARA EFEITOS DE INVESTIGAÇÃO, MEDIANTE AUTORIZAÇÃO ESCRITA DO INTERESSADO, QUE A TAL SE COMPROMETE.

University of Minho Reconstruction of a generic metabolic model for Streptococcus pneumoniae

iii

Acknowledgements

I would like to start by giving thanks to my supervisors Dr. Isabel Rocha and Dr.

Francisco Pinto for the opportunity to work in this project and for all of her assistance and

knowledge when needed.

I would also like to thank Oscar Dias for all of his input, patience and

companionship throughout the course of this study.

To all the people at IBB (Institute for Biotechnology and Bioengineering) namely

but not exclusively, Daniela, Carla, André, Daniel, José Pedro, Paulo Vilaça and Vitor

Costa, I extend my appreciation for all the help and input when trying to solve and

overcome some difficulties during this study.

I would also like to thank my family for all of their support and unconditional love

and understanding.

Last but not least I would like to thank my girlfriend for all of her love, friendship,

support, understanding and patience throughout the countless hours that kept me away

from her.

The present work was achieved under the project PTDC/BIA-MIC/099551/2008 –

Computational search of cellular network motifs associated to Streptococcus pneumoniae

virulence, financed by FEDER funds through the Competitive Factors Operational

Program – COMPETE and by National funds through FCT – Foundation for Science and

Technology.


iv


v

Abstract

Reconstruction of a generic metabolic model for Streptococcus pneumoniae

Streptococcus pneumonia is a gram-positive bacteria with lancet-shaped cells that

survive and thrive in almost any environment. It is usually found in pairs or short chains

of diplococcic and in the decade of 1880 was described as one of the major causes of

several infections such as pneumonia, meningitis, otitis media and endocarditis.

Since antibiotic resistance of S.pneumoniae is pointed out as a challenge in the

treatment of these infections, focus has been greater on disease control, although several

other studies target the discovery of pneumococcal polysaccharide antigens as vaccines.

Methods, such as reconstruction of genome-scale metabolic networks, are essential for

determination of the bacteria´s invasive capability in humans by analysis of their

metabolism.

Evolution of new techniques for data collection and technology platforms allow

researchers to study the cell´s organization and functionality as a whole, as well as the

interactions within the cell. Analysis of genome similarity along with metabolic functions

across strains is crucial in determining if virulence factors and increased invasiveness is

dependent on specific genomic regions or if these are determined by different

environmental conditions.

In this study we aim to reconstruct a generic metabolic model capable of

simulating in vitro experiments of Streptococcus pneumoniae. For such, we rely on draft

metabolic models obtained from SEED, with curation and validation performed on

OptFlux.

As a result, a generic metabolic model for S.pneumoniae was obtained that could

simulate growth, by-product formation and determine amino acid essentiality, serving as

a basis for generating other strain-specific metabolic models. No direct relationship

between genome-metabolic function was ascertained and, therefore, further investigation

is required.


vi


vii

Resumo

Reconstrução de um modelo metabólico genérico para Streptococcus pneumoniae

Streptococcus pneumonia é uma bacteria gram positive com células em forma de

lança com capacidade para sobreviver e crescer em quase qualquer meio ambiente.

Normalmente é encontrado em pares ou cadeias curtas de de diplococcus sendo, na

década de 1880, descrito como uma das maiores causas de numerosas infecções tais como

a pneumonia, meningite, otitis media e endocartite.

O aumento de resistência antimicrobiana de S.pneumoniae é apontado como um

desafio no tratamento destas infecções devido ao aumento de incidência de doenças

causadas por estirpes de pneumococcos invasivas em crianças bem como nos tratamentos

com antibióticos sem sucesso [1].

A maioria dos estudos com S.pneumoniae até à data têm-se prendido com o

control de doenças pneumocócicas embora outros estudos têm como objectivo a

descoberta de antigénios polissacáridos que possam actuar como vacinas [2]. Para

alcançar este ultimo, métodos tais como a reconstrução ao nível do genoma e redes

metabólicas tornam-se essenciais na determinação da capacidade invasiva da bactéria nos

seres humanos.

Actualmente, a evolução de novas técnicas de recolha de dados e plataformas

tecnológicas, permitem aos investigadores o estudo da organização cellular e

funcionalidade como um todo, bem como as interacções intracelulares.

SEED é uma base de dados actualizada de dados microbianos com ênfase em

genes com capacidade para, de forma autómata, reconstruir um modelo através do RAST

(Rapid Annotation Subsystem Technology).

Neste estudo pretende-se reconstruir um modelo metabólico generic para a

Streptococcus pneumonia com capacidade para similar in vitro. Para tal foram utilizados

modelos metabólicos obtidos através do SEED de três estripes de S.pneumoniae,

nomeadamente a G54, TIGR4 e R6.


viii


ix

Table of Contents

Acknowledgements ............................................................................................................ iii

Abstract ................................................................................................................................ v

Resumo .............................................................................................................................. vii

List of Abbreviations ......................................................................................................... xii

List of Figures .................................................................................................................... xv

List of Tables .................................................................................................................... xvi

1. Introduction .................................................................................................................. 2

1.1 - Streptococcus pneumoniae.................................................................................... 2

1.1.1 - Strains ................................................................................................................ 3

1.1.2 - Identification ..................................................................................................... 4

1.1.3 - Virulence Factors .............................................................................................. 4

1.1.4 - Infection and vaccination .................................................................................. 5

1.2 - Systems Biology ................................................................................................... 7

1.3 - Metabolic Networks .............................................................................................. 8

1.4 - Systems Biology Markup Language (SBML) .................................................... 10

1.5 - Reconstruction Process ....................................................................................... 11

2 - Objectives of this Dissertation ...................................................................................... 15

3 - Methods, Databases and Software ................................................................................ 16

3.1 - Medium composition .......................................................................................... 16

3.2 - SEED................................................................................................................... 17

3.3. Optflux ................................................................................................................. 18

3.4 - Flux Calculations ................................................................................................ 19

4 - Results and Discussion ................................................................................................. 22

4.1 - Statistics .............................................................................................................. 22

4.2 - Genome comparison ........................................................................................... 29

4.3 - Individual Models ............................................................................................... 29


x

4.4 - Generic Model .................................................................................................... 30

4.4.1 - Biomass Reaction ............................................................................................ 31

4.4.2 - Main Pathway Analysis ................................................................................... 34

4.4.2.1 - Carbohydrates Metabolism ....................................................................... 34

4.4.2.1.1 - Glycolysis .......................................................................................... 34

4.4.2.1.3 - Pentose Phosphate Pathway (PPP)..................................................... 34

4.4.2.1.4 - Fructose and Mannose ....................................................................... 35

4.4.2.1.5 - Galactose ............................................................................................ 35

4.4.2.1.6 - Starch and sucrose ............................................................................. 35

4.4.2.1.7 - Pyruvate ............................................................................................. 36

4.4.2.1.8 - Amino sugar and nucleotide sugar ..................................................... 36

4.4.2.2 - Energy Metabolism ................................................................................... 37

4.4.2.2.1 - Nitrogen ............................................................................................. 37

4.4.2.3 - Lipid Metabolism ..................................................................................... 38

4.4.2.3.1 - Fatty acid Biosynthesis ...................................................................... 38

4.4.2.4 - Nucleotide Metabolism............................................................................. 39

4.4.2.4.1 - Purine and Pyrimidine........................................................................ 39

4.4.2.5 - Amino acid Metabolism ........................................................................... 40

4.4.2.5.1 - Alanine, Aspartate and Glutamate ..................................................... 42

4.4.2.5.2 - Glycine, Serine and Threonine .......................................................... 43

4.4.2.5.3 - Cysteine and Methionine ................................................................... 43

4.4.2.5.4 - Valine, Leucine and Isoleucine biosynthesis ..................................... 44

4.4.2.5.5 - Phenylalanine, Tyrosine and Tryptophan biosynthesis ..................... 44

4.4.2.5.6 - Lysine biosynthesis ............................................................................ 45

4.4.2.6 - Metabolism of Cofactors and Vitamins .................................................... 46

4.4.2.6.2 - Riboflavin .......................................................................................... 46

4.4.2.6.3 - Vitamin B6 ......................................................................................... 47


xi

4.4.2.7 - Other metabolisms .................................................................................... 47

4.5 - Simulation results with Hoeprich´s medium....................................................... 47

5 - Conclusions .................................................................................................................. 49

6 - Future Work .................................................................................................................. 52

7 - References .................................................................................................................... 53

Annexes.............................................................................................................................. 60


xii

List of Abbreviations

Acessory Regions – AR

Acid Hydrolyzed Casein - AHC

Acyl-Carrier-Protein – ACP

Adenosine – 5’.triphosphate – ATP

ATP Binding Cassette – ABC

ATP-binding cassette – ABC

Autolysin – LytA

Coenzyme A – CoA

Cytosine – C

Deoxyadenosine triphosphate – dATP

Deoxyguanosine triphosphate – dGTP

Deoxyribonucleic acid – DNA

Diaminopimelate – DAP

Dipotassium phosphate – K2HPO4

EFM – Elementary Flux Modes

Enzyme Comission – E.C

Escherichia coli – E.coli

Evolutionary Algorithm – EA

eXtensible Markup Language – XML

Ferrous sulfate – FeSO4

Flavin adenine dinucleotide – FAD


xiii

Flux Balance Analysis – FBA

Fructose-6-phosphate – F6P

Guanine – G

Guanosine triphosphate – GTP

Hydrocloric acid - HCl

Kyoto Encyclopedia of Genes and Genomes – KEGG

Magnesium sulfate – MgSO4

Manganese sulfate – MnSO4

Metabolic Engineering – ME

Metabolic Flux Analysis – MFA

Nicotinamide adenine dinucleotide – NADH

Pentose Phosphate Pathway – PPP

Phosphoenolpyruvate – PEP

Phosphostransferase system – PTS

Pneumolysin – Ply

Rapid Annotation Subsystem Technology - RAST

Ribonucleic acid – RNA

Ribose 5-phosphate – R5P

Serine hydroxymethyltransferase – glyA

Simulated Annealing – SA

Sodium bicarbonate – NaHCO3

Sortase A – SrtA

Streptococcus pneumoniae – S.pneumoniae


xiv

Systems Biology – SB

Systems Biology Markup Language – SBML

Tricarboxylic acid – TCA

Zinc sulfate – ZnSO4


xv

List of Figures

Figure 1 - Streptococcus pneumoniae (A), Group A streptococcus (B), Group B streptococcus (C)

(obtained from Centre for Disease Control – CDC) ......................................................................... 2

Figure 2 – Stoichiometric model. Metabolic network representation composed of metabolites (A,

B, C, D) and fluxes (internal - vi and exchange bi)(2A). Mass balance equations for all reactions

for each species (2B) written in matrix form and considering homeostasis (2C)............................. 9

Figure 3 – Stages of the Reconstruction process ............................................................................ 13

Figure 4 – Functional modules of the OptFlux application, obtained from [33] ............................ 19

Figure 5– Specific growth rate. ...................................................................................................... 20

Figure 6 – Venn diagram for S.pneumoniae strains used in the reconstruction process. ............... 24

Field Code Changed

Field Code Changed

Field Code Changed

Field Code Changed

Field Code Changed

Field Code Changed


xvi

List of Tables

Table 1 – Virulence factors and functionality .................................................................................. 5

Table 2 – Serotypes present in Polysaccharide Conjugate Vaccine (PCV) and Pneumococcal

Polysaccharide Vaccine (PPV) ........................................................................................................ 6

Table 3 – Data sources used for metabolic reconstructions. .......................................................... 14

Table 4– Composition of Hoeprich´s medium ............................................................................... 16

Table 5 – Biomass and glucose concentrations during exponential phase of a batch fermentation

with Hoeprich´s medium and Streptococcus pneumoniae serotype 23F strain St 99/95. .............. 20

Table 6 – Reactions initially identified as G54 strain-specific and corresponding genes in other

strains. ............................................................................................................................................ 23

Table 7 – Number of reactions present in the models of strains G54, TIGR4 and R6. ................. 24

Table 8 - List of strain-specific reactions and reactions specific to pairs of strains. ..................... 25

Table 9 – Genome homology comparison (using 90 % threshold). ............................................... 29

Table 10– Data collected from metabolic reconstruction model of strain G54. ............................ 30

Table 11– Data collected from metabolic reconstruction model of strain TIGR4. ........................ 30

Table 12– Data collected from metabolic reconstruction model of strain R6. .............................. 30

Table 13– Data collected from metabolic reconstruction of the generic model. ........................... 30

Table 14 – Biomass composition and compound coefficients ....................................................... 31

Table 15 – Relations Gene/Enzyme in Fatty acid Biosynthesis .................................................... 38

Table 16 – List of models consulted for ATP maintenance limit values ....................................... 40

Table 17 – Essential and non-essential amino acids for in vitro growth of Streptococcus

pneumoniae D39 (serotype 2; NCTC7466) compared to our generic model. ((0) – No effect on

growth; (+) - Essential for growth; (-) - Limiting in case of absence from the medium) .............. 41

Table 18 – List of aminopeptidases ............................................................................................... 50

Table 19 – List of all reactions added/changed in generic model .................................................. 51


1


2

1. Introduction

1.1 - Streptococcus pneumoniae

Streptococcus pneumonia is an aerotolerant anaerobic gram-positive bacteria

with lancet-shaped cells that survive and thrive in almost any environment (Figure 1A).

It is usually found in pairs or short chains of diplococcic and in the decade of 1880 was

described as one of the major causes of several infections such as pneumonia,

meningitis, otitis media and endocarditis. Streptococci group division can be based on

cell wall composition. The most common and pathogenic groups of streptococcus are A

(Figure 1B) and B (Figure 1C). Diseases from the first group vary from sore throat to

necrotizing fasciitis, while streptococci of Group B cause life-threatening diseases in

young, elderly and adults with a compromised immunity system, such as pneumonia.

Figure 1 - Streptococcus pneumoniae (A), Group A streptococcus (B), Group B

streptococcus (C) (obtained from Centre for Disease Control – CDC)

According to Dopazo and co-workers [3], in 1998, an estimated 3.5 million

deaths worldwide were attributed to S.pneumoniae infections. S. pneumoniae´s main

hosts are humans, and they are mostly found in the nasopharynx [4][5]. However,

common drug resistance has been reported [6][7][8][9]. Sá-Leão [4]states that, because

this bacterium is part of the human nasopharyngeal flora, “any intervention to combat

pneumococcal disease, such as the introduction of antibiotics or vaccines” also has an

B

A

C

B

A

A

B

A


3

impact on selecting drug-resistant lineages and strains, thus making humans

“evolutionary partners”. Evolution may occur through re-modulation of penicillin-

sensitive enzymes, acquisition of mobile elements with anti-microbial resistance or

capsular switches to evade action of vaccines that target the capsule [4], or prevention

from complement pathway activity [10].

1.1.1 - Strains

Strain selection for comparative analysis is usually performed to cover the

largest spectrum of representatives for virulent and avirulent serotypes. Genome

sequences for all three strains used in this study are available at NCBI (National Centre

for Biotechnology Information).

Streptococcus pneumoniae strain G54 belongs to serotype 19F and is known to

be more prevalent at younger and older stages in life, being associated to increased

levels of antibiotic resistance and insusceptibility. Its genome is composed of a single

circular chromosome with 2,078,953 base pairs.

S.pneumoniae TIGR4 strain belongs to serotype 4 and has been classified as

highly invasive and virulent in rat models [11]. Its genome is composed of a single

circular chromosome with 2,160,837 base pairs and is the largest of the strains in this

study.

S.pneumoniae R6 strain was originated from a clinical isolate of serotype 2

strain D39 and does not possess a polysaccharide capsule, rendering it avirulent and

safe to work with. Therefore, it is used as a standard strain for laboratory experiments.

All three strains possess a G+C content of 39.7 %, factor linked to DNA stability

[12] where high levels of such confer higher stability and vice versa, which is

understandable due to the fact that all belong to the same species.


4

1.1.2 - Identification

Identification of S.pneumoniae can be achieved through alfa hemolysis

surrounding colonies obtained on blood agar, negative reaction with catalase,

susceptibility to optodium, and solubility of the bacteria in bile salts. However,

serological typing (serotyping) remains one of the most used methods for

characterization of S.pneumoniae isolates. Serotyping is based on the structure of the

bacteria´s exopolyssacharide capsule [5].

Until 2006, only two strains of S.pneumoniae had had their genome sequence

completely determined: virulent serotype 4 strain TIGR4, and the avirulent,

unencapsulated, laboratory strain R6. These revealed several insights into the

metabolism and organization of S.pneumoniae.

New methodologies have contributed to the discovery of novel strains. An

example of such is that in 2006, 85 serotypes of S.pneumoniae were known, increasing

to 91 in just 3 years (2009) [5][10]. Because S.pneumoniae polysaccharide capsule

provides resistance against phagocytosis, bile was used to dissolve it, leading to the

identification of the first bacterial autolytic enzyme.

1.1.3 - Virulence Factors

The increase of antimicrobial resistance of S.pneumoniae is pointed out as a

challenge in the treatment of these infections and therefore, it is essential to understand

which factors contribute to such [1].

The most common virulence factors of S.pneumoniae are shown in Table 1. The

polyssacharide capsule, the major autolysin (LytA), the intracellular toxin pneumolysin

(Ply) and sortase A (SrtA), which play both defensive as well as aggressive roles in

different stages of a pneumococcal infection [4][13], are some examples of virulence

factors associated to S.pneumoniae. Other interesting virulence factors are the

“accessory regions” (AR) within the genome of S.pneumoniae [14] and pili, essential

for adhesion of the bacterium to epithelial cells of the upper respiratory tract [15].


5

Accessory regions, also defined as genomic islands [16], may have an important role in

virulence since, through horizontal gene transfer (the ability to transfer genes among

other strains and species), either through natural genetic transformation, transduction or

conjugation mechanisms, can confer a species with, for instance, drug resistance

attributes. An example of such is β-lactam resistance[17][18]. Harvey and colleagues

[14] state that “comparisons across serotypes risk underestimating the impact of the

serotype itself on virulence, due to serotype-specific structural differences in the capsule

that can affect complement deposition and resistance against phagocytosis”. This means

that, although comparison between different serotypes is useful in determining

invasiveness potential, one must not discard the differences within the same serotype

since virulent and non-virulent strains can belong to the same group[19].

Table 1 – Virulence factors and functionality

Virulence factor Function

Polysaccharide capsule Prevents phagocytosis by host immune

cells

Surface proteins Prevent activation of complement

Pili Adhesion to epithelial cells in upper

respiratory tract

Accessory regions Diverse function depending on which

genes are horizontally transferred (i.e.

drug resistance or fitness)

.

1.1.4 - Infection and vaccination

Streptococcus pneumoniae infections vary from pneumonia (lungs), otitis media

(middle ear cavity) to pneumococcal meningitis, mainly affecting 2 month to 5 year old

children, people with over 65 years old and individuals with compromised immune

systems. Transmission usually occurs via aerosol or by direct and prolonged contact to

respiratory secretions of infected people. Approximately 66% of infections in adults and

80% invasive infections in children are attributed to 8-10 capsular serotypes [20].


6

Common treatment for pneumococcal infections relies on antibiotics (penicillin,

cephalosporin or erythromycin) [16][1] but the increase in antibiotic-resistant

pneumococcus has diminished their effectiveness against these bacteria.

In order to overcome this barrier, new methods for immunization have been studied

based on polysaccharide composition. Currently, two types of vaccines are being used

that diminish these infections [21][22][23].

Polysaccharide Conjugate Vaccine (PCV) contains polysaccharide from seven

(7) polysaccharides (Table 2) conjugated with proteins and are applied to young

individuals (up to 5 years old) and risk-groups (individuals with compromised immune

systems).

Pneumococcal Polysaccharide Vaccine (PPV) contains purified capsular

polysaccharide from 23 serotypes (Table 2) and administered to individuals over 65

years old.

Table 2 – Serotypes present in Polysaccharide Conjugate Vaccine (PCV) and

Pneumococcal Polysaccharide Vaccine (PPV)

Vaccine Serotypes

PCV 4,6B,9V,14,18C,19F,23F

PPV 1,2,3,4,5,6B,7F,8,9N,9V,10A,11A,12F,14,15B,17F,18C,19F,19A,20,22F,23

F,33F

The main focus of studies regarding S.pneumoniae to date has been the control

of pneumococcal disease, although several other studies target the discovery of

pneumococcal polysaccharide antigens as vaccines [4]. To perform both, methods such

as reconstruction of genome-scale metabolic networks are essential for determination of

the bacteria´s invasive and virulent capability in humans.


7

1.2 - Systems Biology

Systems biology (SB) studies the interaction between components of biological

systems and how these interactions influence the behaviour and functions of those

systems, such as enzymes and metabolites in a metabolic network or pathway.

This field of study is able to obtain, integrate and analyse complex data sets from

several experimental origins allowing further study of phenomics, genomics,

transcriptomics, metabolomics or fluxomics. For instance, genome sequences and

protein properties have been identified through molecular biology; however, alone, this

is insufficient for the interpretation of a biological system [24]. The intrinsic complexity

of these systems makes it difficult to determine their functionality, and therefore, the

combination of various approaches, amongst them experimental and computational, are

expected to assist in resolving such issues [24].

Computational biology has two distinct, yet intertwined objects of study: data

mining, which extracts hidden patterns from a large amount of experimental data; and

simulation-based analysis, which tests hypothesis with in silico experiments.

From the first approach, a hypothesis is formed whilst in the second approach

this hypothesis is tested in order to predict certain outcomes for in vitro and in vivo

experiments [24].

An increase in high throughput quantitative data from experimental molecular biology

has favoured advances in simulation-based analysis [24]. Meanwhile, the rapid

evolution of computational tools and software enable the construction and analysis of

more reliable biological models. Despite this progress, the need for researchers to

exchange information remains essential. In order to facilitate collaboration between

researchers, a platform for modelling and analysis of biological systems – Systems

Biology Markup Language (SBML) – was developed, increasing, consequently, the

value of new databases that focus on biological pathways, i.e., KEGG[25].


8

1.3 - Metabolic Networks

The most common trades of a biological network are defined by complex

interactions between its various components. Currently, evolution of new techniques for

data collection and technology platforms allow researchers to study the cell´s

organization and functionality as a whole, as well as the interactions within the cell.

Genome annotation along with experimentally determined physiological and

biochemical information of organisms allows the reconstruction of a metabolic network

[26]. The reconstruction of metabolic networks is an important tool in the study of

biological networks; yet, its process is very laborious and time consuming. When all

steps are performed accordingly, predictions on maximal cell growth or production of a

desired metabolite can be achieved using in silico models.

Three distinct groups of models are used in metabolic engineering:

stoichiometric, kinetic and regulatory. Both kinetic and regulatory models are difficult

to obtain due to the lack of information and comprehension of kinetic and complex

cellular regulation processes, respectively, [27] and, therefore, stoichiometric models

are the most commonly used (Figure 2).

In stoichiometric models, biochemical reactions of the network are represented

as stoichiometric equations [27]. The resulting stoichiometric matrix (S) (Figure 2C)

represents the biological network (Figure 2A) in mathematical terms, where the mass

balances for each intercellular metabolite is represented by a set of differential equations

(Figure 2B). The metabolites (m) are placed in the rows of the matrix (in concentration

form) whilst the reactions (n) are placed in the columns (as reaction rate) leading to the

mathematical representation in equation 1.

B3

B1 A B

C D

B2 v1

V3 V2 V5

V4

A


9

Figure 2 – Stoichiometric model. Metabolic network representation composed of

metabolites (A, B, C, D) and fluxes (internal - vi and exchange bi)(2A). Mass balance

equations for all reactions for each species (2B) written in matrix form and considering

homeostasis (2C).

Eq. 1

As can be inferred from equation 1, data on stoichiometry (S), biomass

formation (µ) and intracellular reaction fluxes (v) is required. Since the latter is yet

difficult to determine, a steady-state for internal metabolites is considered leading to a

general equation (Eq. 2). This, however, originates an underdetermined steady-state

solution, since the number of reactions is greater than the number of metabolites present

in a metabolic network [27][28][29]. In order to overcome this issue and reach a steady-

state flux distribution, additional constraints, obtained by flux measurements in the

network, are required.

Eq. 2

Stoichiometric models are used by several optimization methods such as

Metabolic Flux Analysis (MFA), which allows the computation of fluxes in space given

a set of measured fluxes, and Flux Balance Analysis (FBA), which simulates phenotype

behaviour under certain environmental conditions [30][31].

Flux balance analysis (FBA) is used to, besides genome-scale network analysis,

analyse how perturbations to the network affect reaction flux distributions, such as gene

deletions, medium composition or in silico drug effectiveness. Stoichiometric matrixes,

associated to an objective function (i.e. ATP production or biomass formation),

𝑑𝐴

𝑑𝑡 𝑣1 + 𝑣3 + 𝑏1

𝑑𝐵

𝑑𝑡 𝑣1 + 𝑣2 𝑏2

𝑑𝐶

𝑑𝑡 𝑣3 𝑣4 + 𝑣5

𝑑𝐷

𝑑𝑡 𝑣2 + 𝑣4 𝑣5 𝑏3

1 1 1 1 1 1 1 1 1 1 1 1 1

𝑣1𝑣2𝑣3𝑣4𝑣5𝑏1𝑏2𝑏3

S

V

B C


10

compose the necessary steps to identify optimal reaction flux distributions. [28][32].

This method does not predict exact behaviour of a system, rather uses known

constraints to a function, separating which reactions allow fluxes and which do not. The

decrease in the number of possible solutions/reactions to which an objective function is

achievable improves the study of genotype-phenotype relationships.

Despite all the advances in technology, the reconstruction process is an iterative

one. Limitations derived from organism-specific features prevent the use of

automatically generated networks, requiring manual evaluation [31][10], although

numerous software tools and packages exist to assist in the reconstruction process. In

the present study, we focus on genome and biochemical databases shown in Table 1.

The availability of information on genetics, biochemistry or physiology of our

organism increases the quality and predictive capability of our model [34][31][33].

1.4 - Systems Biology Markup Language (SBML)

In 2000 the SBML, a XML-based language, was developed for representing and

exchanging models between various tools and software [24]. XML´s portability and

overall acceptance in the bioinformatics community were the main reasons for its

selection. Since its development, several levels and versions of SBML have been

introduced in which new features are included that are required by the bioinformatics

community.

Cooperation among researchers is essential for the evaluation and development

of system’s models; therefore, it is necessary that the information is standardized.

SBML aims to establish a common language for the description of biological models,

be it in the fields of computational biology, gene regulation, or others. Although,

initially it was developed for representing dynamic models, today it is also used as a

framework for stoichiometric models [33].

The use of SBML in software packages resolves issues regarding inoperability,

providing a common format for publications and databases [24].


11

1.5 - Reconstruction Process

The reconstruction process can be achieved through various methods, although

there are several steps that prove common to all. Here, we focus on the methodology

defined by Thiele and Palsson [33]. This protocol consists, in simple terms, of five

phases comprised of numerous steps within each phase[31] (Figure 3).

The first phase consists on generating a genome-based draft network, which

basically aims to gather and obtain the necessary information for the reconstruction

process: genome annotation, metabolic reactions, and experimental data. This can be

achieved either manually by extracting information from experimental assays and

literature or automatically by genome-based reconstruction tools (see 3.2 – SEED). It

would be optimal to gather information on organism-specific databases, which would

increase the feasibility and quality of the metabolic network; however, generic

databases can be used when there is a lack of organism-specific information

[31][33][34]. The annotated genome provides information on enzyme presence that

leads to the next step in this phase, which is determining the reactions they catalyse.

Presently, there exists a variety of metabolic databases that possess the capability for

inferring information regarding reactions based on enzyme commission number (E.C)

such as KEGG, and their biochemistry, such as BRENDA. RAST from SEED performs

this first phase in an automated manner.

Phase two is defined as a refinement stage. Information regarding substrate and

cofactor usage, determination of charged formula, calculation of stoichiometry,

verification of gene-reaction associations, amongst others, is taken into account. In this

phase, the first goal is to determine if the reaction(s) present, obtained from the draft

reconstruction process, do/does, in fact, occur in our species and whether it is correctly

connected to the rest of the network. Another issue to be addressed in this stage is to

determine the gene-protein-reaction associations, important to gene essentiality

predictions, since removal of one or more genes may affect numerous reactions or if

there are any enzymes that catalyse the same reaction (isoenzymes). Enzyme

identification inferred from homology, therefore not specific to our organism, must be

weighed to prevent erroneous phenotypic behaviours [34]. Another goal at this stage is


12

to determine the biomass formation requirements. Literature and experimental assays

usually comprise the information sources for growth requirements [13][33][35].

Phase three consists in the transformation of the obtained model into a

compatible format. Conversion into a mathematical model in order to define objective

functions and simulation constraints are the main steps in this phase. The maximization

of biomass, for instance, is a standard objective function, since it covers growth

prediction. Considering p biomass components, the biomass formation reaction can be

expressed as illustrated in equation 3.

∑ Eq. 3

where the Ck values are determined from the biomass composition for each

metabolite or macromolecule Xk [33]. Constraints are applied to the model in order to

simulate as accurately as possible experimental assays. Boundaries and medium

components and quantities are examples of constraints used to enhance the predictive

quality of the network model.

In phase four, the evaluation of the network model, comprised of several steps,

is performed to determine its predictive capability. The capability of the model to

synthesize biomass precursors, form by-products known to the organism, identify

missing reactions (gaps), detection and correction of dead-end metabolites, model

comparison to known organism incapability (identification of false positives) and a

quantitative evaluation (expected growth rate under the same experimental conditions)

are issues to be addressed and corrected if need be.

The fifth phase was not addressed in the article; however, the authors mention

that it consisted of how the model would be used in a prospective manner.

Being an iterative process, all steps and phases are re-definable in order to refine the

information and predictive capability of the model if such should prove necessary.


13

Figure 3 – Stages of the Reconstruction process

Table 3 lists some of the databases used to aid in the reconstruction process.

Phase 1

•Draft reconstruction

•Genome annotation

• Identification of metabolic reactions

•Collection of experimental data

Phase 2

•Reconstruction refinement

•Substrate and cofactor usage

•Biomass equation

•Stoichiometry

•Gene-Protein-Reaction associations

Phase 3

•Conversion into computable format (SBML)

•Conversion to mathematical model

•Definition of objective function and constraints

Phase 4

•Model evaluation

• In silico simulations

•Production of biomass precursors

•By-product formation

• Identification of gap filling reactions

•Quantitative evaluation

Phase 5

•Prospective use

Reconstruction

Process


14

Table 3 – Data sources used for metabolic reconstructions.

Name

Description Source Reference Link

SEED Database Comparative genomics

tool

Curated and non-curated

data

[36] www.theseed.org/wiki/Home_of_the_SEED

KEGG (Kyoto

Encyclopedia of Genes

and Genomes)

Database with Genomic

and metabolic

information.


data cross-referenced to

other databases

[25] http://www.genome.jp/kegg/

BRENDA – enzyme

database

Enzyme information Manually extracted

from literature

[37] http://www.brenda-enzymes.info/

UNIPROT (Universal

Protein Resource)

Protein sequence and

functional information


data, cross-referenced to

other databases

[38] http://www.uniprot.org

NCBI (National Center

for Biotechnology

Information)

Information of Sequence

data of both microbial

and higher organisms.


data retrieved from

GenBank and literature.

[39] http://www.ncbi.nlm.nih.gov/

CMR (Comprehensive

Microbial Resource)

Display, search and

analysis of sequence and

annotation for complete

archaeal and bacterial

genomes


data retrieved from

Genbank and JCVI

[40] http://cmr.jcvi.org

http://www.theseed.org/wiki/Home_of_the_SEED

http://www.genome.jp/kegg/

http://www.brenda-enzymes.info/

http://www.uniprot.org/

http://www.ncbi.nlm.nih.gov/

http://cmr.jcvi.org/


15

2 - Objectives of this Dissertation

In this project our main goal is to reconstruct a generic metabolic network for

S.pneumoniae using metabolic models of strains R6, TIGR4 and G54. The

reconstruction of a generic metabolic network will hopefully provide insights into

metabolic genes that can either be present in all three strains and, therefore, considered

transversal to any given pneumococcus strain, or present itself unique to a given strain

or set of strains, allowing the grouping of strains in accordance to metabolic gene

function. Another objective of this study is to determine if genome similarity has any

correspondence to metabolic function. This type of analysis might present itself useful

for determining if traits, such as virulence, are correlated with specific genes or if they

are present in all strains and only expressed under certain conditions.

This analysis will be performed based on the reactions of each strain using Rapid

Annotation Server Technology (RAST) for model draft reconstruction and Optlux, a

modeling tool for metabolic reconstruction, for growth simulations and constraint

definition.


16

3 - Methods, Databases and Software

3.1 - Medium composition

Media selection is essential to simulate experimental assays. Among the several

articles that described media composition [41][42][43][9][44][45][46][47][48],

Hoeprich’s medium from Gonçalves´s study was selected due to the fact that most of

the compounds could be accounted for in the model.

Table 4– Composition of Hoeprich´s medium

Hoeprich 1955(as follows per liter)

AHC 20g

Glucose 12,5g

K2HPO4 5g

NaHCO3 1g

L-Cystine 150mg

Tryptophan 20mg

Tyrosine 200mg

L-Glutamine 625mg

L-Asparagine 100mg

Choline 10mg

MgSO4 500mg

FeSO4 5mg

ZnSO4 0,8mg

MnSO4 0,36mg

Thioglycolic acid 1ml (10%)

HCl 0,02ml

Biotin 0,0015mg

Nicotinate 100mg

Pyridoxal 100mg

Calcium pantothenate 500mg

Thiamine 100mg

Riboflavin 100mg

Adenine 100mg


17

3.2 - SEED

Increase in data obtained from sequence technology demands innovative new

tools capable of high-throughput of metabolic network models at a genome scale.

As seen in Table 1, various tools have been developed to aid in the

reconstruction of metabolic networks. SEED [35] is an up to date database of all

microbial gene data [35]. Comparative analysis of these data across multiple organisms

in a rich genomic, biochemical and phylogenetic contexts provided by the collection of

annotated subsystems greatly facilitates their interpretation and practical applications,

such as the understanding of cellular networks, gene and pathway discovery,

identification of novel drug targets, and strain engineering. SEED aims to automate the

reconstruction process up to a certain extent, since manual curation is always required.

The draft model reconstruction is obtained through this web-based platform´s

reconstruction pipeline. Integration of genome annotation, Gene-Protein-Reaction

(GPR) associations, biomass reaction, analysis of reaction reversibility and model

optimization are the basis of this process.

Genome sequences are imported into SEED via the RAST server for annotation.

These are used to generate models consisting of the network reactions along with the

GPR associations and an organism-specific biomass reaction. All reactions associated to

a specific enzyme are included in the model, as well as spontaneous non-enzymatic

reactions. GPR associations are genome-based, coupled with the functional roles

assigned to genes during annotation.

The autocompletion step is required to ensure model predictive functionality,

since the draft reconstruction model usually contains gaps that prevent production of

metabolites essential to biomass formation. These gap-filling reactions are selected from

a comprehensive database that integrates biochemistry data from KEGG and 13

published genome-scale metabolic models [36]. This step is important due to the fact

that many pathogens do not possess complete pathways for some metabolites such as

thiamine or lipoteichoic acid, obtaining them from the interaction with the host´s cells.

Uracil 1g


18

At the end of the autocompletion step, the model is considered to be functional,

since it can simulate growth. There are several other steps that SEED is capable of

performing in order to optimize the draft model, such as Biolog consistency analysis

and gene essentiality which, in the first case, adds or removes reactions from the model

to fit Biolog phenotyping array data, and in the second case, determines essential genes

to the model. These steps, however, will be performed by Optflux in this work (see

chapter 3.3 Optflux).

This platform will be used for the draft reconstruction of each individual model

on which comparative reaction analysis is to be performed.

3.3. Optflux

As stated in chapter 3.2, Optflux will be used to perform all simulations, gene

essentiality determination and reaction analysis to fit culture conditions.

OptFlux is an open source computational tool for metabolic engineering (ME)

applications developed in the University of Minho [30]. It was developed on top of

AIBench due to its design and architectural principles and uses Java as its core

language. Other characteristics that increase this software´s potential are its user-

friendly interface, modular structure, which allow for addition of new plug-ins, and its

compatibility with standards and other software such as SBML and CellDesigner.

Several tools and operations are at the user’s disposal such as:

- Phenotype simulation, with the use of FBA;

- Calculation of metabolic fluxes, with MFA;

- Elementary Flux Modes, used for pathway analysis (EFM);

- Strain optimization algorithms, such as Simulated Annealing – SA - and

Evolutionary Algorithms – EA;

- And model visualization tools, used for results interpretation.

Figure 1 shows the four main areas of action of this powerful tool.


19

Flux balance analysis, based on a steady-state approach, for phenotype

simulation uses Linear Programming (LP) to calculate all fluxes over the reactions

enabling wild-type and mutant strain simulation. Usually, but not exclusively, biomass

formation (i.e. growth rate) is the objective function to be maximized [30][49].

For improvement purposes, experimental data can be used to introduce constraints to

the original model in the form of fluxes. Since the model is composed of fluxes

(mmol/gDW.h) and experimental data usually is provided in the form of concentrations

(g/L), conversion of these values must be performed before any simulation with

constraints is run.

Figure 4 – Functional modules of the OptFlux application, obtained from [36]

3.4 - Flux Calculations

In order to simulate growth rate, experimental data, obtained via literature,

required calculations, since the results report concentrations (in g/L and mg/L) and not

fluxes, as required by Optflux (mmol/gDW.h).

These calculations were only carried out for metabolites that were present in the media

solution (see Table 4 in chapter 3.1).


20

Experimental assay data from Gonçalves and colleagues [41] described specific growth

rates and glucose consumption yield. By analysis of Fig1A from the same study, the

glucose consumption rate and biomass yield from glucose were retrieved. Table 5

shows the values over time.

Table 5 – Biomass and glucose concentrations during exponential phase of a batch

fermentation with Hoeprich´s medium and Streptococcus pneumoniae serotype 23F

strain St 99/95.

Calculation of specific growth rate (µ) was obtained via linearization of biomass values

during the exponential phase. Figure 4 illustrates the specific growth rate value (0.46h-1

- regression coefficient).

Figure 5 – Specific growth rate.

y = 0,4666x - 1,3852 R² = 0,9795

-2

-1,5

-1

-0,5

0

0,5

1

1,5

0 1 2 3 4 5 6

Bio

mas

s (L

N)

Time (h)

Specific Growth rate

Time Biomass (g/L) Glucose (g/L)

1 0,25 25

2 0,6 23

3 0,8 22

4 1,3 16

5 1,6 12


21

In order to obtain values for glucose consumption, the following equation was applied,

although the study from which it was adapted [50] obtained specific substrate

consumption rates by calculating the coefficient of a linear regression of substrate

consumed during the exponential phase :

Eq. 4

Where qS is the substrate consumption rate that equals the total amount of

substrate consumed during exponential phase ( ) divided by the amount of biomass

produced ( ) multiplied by specific growth rate ( ). After applying this equation we

obtained a result of 3.82 h-1

for the glucose consumption rate.

For the remaining constraint fluxes, namely for all amino acids, where

measurements throughout the exponential phase were not performed, we used the same

equation described above assuming complete consumption of the compound and then

converting the value to mmol/gDW.h.


22

4 - Results and Discussion

4.1 - Statistics

In order to reconstruct a generic metabolic model for Streptococcus pneumoniae

a model of each of the strains selected was needed. A detailed analysis of these was

required to determine which reactions were unique to each strain, which were common

to all three and which were present in pairs.

The models were exported from SEED in spreadsheet format (.xls) to facilitate these

comparisons. Initial analysis demonstrated that the strain G54 had the most specific

reactions (36), followed by R6 (16). The only reaction specific to strain TIGR4 was the

biomass reaction. Further analysis showed that four reactions were common only to

strains G54 and R6, four reactions were present only in R6 and TIGR4, and three

reactions were common solely to G54 and TIGR4. The gross amount of reactions (729)

was present in all three strains, which was expected, since all strains belong to the same

species.

Specific reaction analysis showed that 23 of those that belong to G54 were reactions

added by SEED (AUTOCOMPLETION) in order to obtain a functioning model. Eleven

(11) of these reactions were specific to Group A Streptococci according to SEED. The

auto completion process is necessary due to the existence of gaps that prevent one or

more compounds that exist in the biomass equation to be formed.

When analyzing the specific reactions of strain R6 we observed that almost all of them

– 11 out of 16 - were part of the fatty acid metabolism and were encoded by the same

gene - peg.966.

Some of these results, however, appeared to be incorrect. Curation of the individual

models revealed that, for six (6) reactions assigned only to S.pneumoniae strain G54, no

literature support nor any database, used in this study, confirmed their presence,

therefore, they were removed. Another five (5) reactions were mistakenly assigned only

to G54, since genes of all the strains in this study, were identified to encode the

enzymes required for them. One of the reaction’s (rxn01501) enzyme commission

number and KEGG RID was incorrectly annotated, requiring curation. One (1) reaction


23

was incorrectly associated to strains R6 and TIGR4, since evidence of its existence was

found only for R6 strain (spr0852) confirmed by literature [42].

Table 6 shows the list of reactions incorrectly annotated by SEED.

Table 6 – Reactions initially identified as G54 strain-specific and corresponding genes

in other strains.

Reaction E.C KEGG G54 R6 TIGR4 References

Rxn00086 1.8.1.7 R00115 SPG_0714 spr0692 SP_0784 [11]

[51]

[3]

Rxn00205 1.11.1.9 R00274 SPG_1133 spr0285 SP_0313

Rxn01256 5.4.99.5 R01715 SPG_1190 spr1174 SP_1296

Rxn01501 1.1.1.88* R02081* SPG_1631 spr1570 SP_1726

((*) Initial E.C = 1.1.1.34, KEGG RID= 2082)

Table 3 shows the list of reactions, after correction, with data concerning

enzyme commission numbers (E.C), associated gene(s) and metabolism that were

specific to each strain and which were common only to pairs of strains.

Analysis of the reactions showed a substantial increase in the number of

reactions associated to G54 when compared to strains TIGR4 and R6. In regards to R6 a

lesser amount of reactions is expected due to the fact that it does not possess a

polysaccharide capsule and, therefore, reactions associated to its production are not

required.

Results of the comparison of S.pneumoniae strains G54, TIGR4 and R6 are shown in

Table 2 and the respective Venn diagram in Figure 2.


24

Table 7 – Number of reactions present in the models of strains G54, TIGR4 and R6.

Strain

G54 R6 TIGR4

Total nº Reactions 768 757 739

Strain-specific reactions 22 17 1

% common reactions 96,32 96,70 99,05

% specific reactions 2,90 2,25 0,14

% pair reactions 0,79 1,05 0,81

Figure 6 – Venn diagram for S.pneumoniae strains used in the reconstruction process.

G54

R6 TIGR4

4

22

17

1*

4

732

2


25

Table 8 - List of strain-specific reactions and reactions specific to pairs of strains.

Reactions especific to G54

ID Name E.C Gene Metabolism

rxn00048 6,7-Dimethyl-8-(1-D-ribityl)lumazine:6,7-dimethyl-8-(1-D-ribityl) 2.5.1.9 peg.166 Riboflavin

rxn00066 NADH:hydrogen-peroxide oxidoreductase 1.11.1.1 peg.1403 No KEGG map assigned

rxn00105 ATP:nicotinamide-nucleotide adenylyltransferase 2.7.7.1|2.7.7.18 AUTOCOMPLETION Nicotinate and nicotinamide

rxn00178 Acetyl-CoA:acetyl-CoA C-acetyltransferase 2.3.1.9 AUTOCOMPLETION

rxn01207 4-Methyl-2-oxopentanoate:NAD+ oxidoreductase 1.2.1.25 AUTOCOMPLETION No KEGG map assigned

rxn05255 FOLt AUTOCOMPLETION Folate transport

rxn05362 4-methyl-trans-hex-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05366 6-methyl-trans-oct-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05370 8-methyl-trans-dec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05374 10-methyl-trans-dodec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05378 12-methyl-trans-tetra-dec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05382 14-methyl-trans-hexa-dec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05387 5-methyl-trans-hex-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05391 7-methyl-trans-oct-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

Comment [J1]: Completar


26

rxn05395 9-methyl-trans-dec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05399 11-methyl-trans-dodec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05403 13-methyl-trans-tetra-dec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn05407 15-methyl-trans-hexa-dec-2-enoyl-ACP:NADP+ oxidoreductase (A-specific) 1.3.1.0 AUTOCOMPLETION No KEGG map assigned

rxn09345 Undecaprenyl diphosphate synthase AUTOCOMPLETION No KEGG map assigned

rxn10180 Pantothenate sodium symporter AUTOCOMPLETION Pantothenate transporter

rxn10447 calcium transport via ABC system AUTOCOMPLETION Calcium transporter

21

Reactions especific to R6


rxn05292 FE3t4

peg.1178 Ferrous ion transport

rxn00802 N-(L-Argininosuccinate) arginie-lyase 4.3.2.1 peg.103

Arginine and Proline

Alanine, aspartate and glutamate

rxn00358 UDP-D-galactopyranose furanomutase 5.4.99.9 peg.319 No KEGG map assigned

rxn00947 Palmitate:CoA ligase (AMP-forming) 6.2.1.3 peg.966 No KEGG map assigned

rxn01316 N-Acetylneuraminate pyruvate-lyase (pyruvate-phosphorylating) 2.5.1.56 peg.970 Amino sugars

rxn05247 FACOAL140(ISO) 6.2.1.3 peg.966 No KEGG map assigned


27

rxn05248 FACOAL150(anteiso) 6.2.1.3 peg.966 No KEGG map assigned



rxn05251 FACOAL170(anteiso) 6.2.1.3 peg.966 No KEGG map assigned


rxn05736 fatty-acid--CoA ligase (tetradecanoate) 6.2.1.3 peg.966 No KEGG map assigned

rxn09448 fatty-acid--CoA ligase (octadecenoate) 6.2.1.3 peg.966 No KEGG map assigned

rxn09449 fatty-acid--CoA ligase (octadecanoate) 6.2.1.3 peg.966 No KEGG map assigned

rxn09450 fatty-acid--CoA ligase (hexadecenoate), peroxisomal 6.2.1.3 peg.966 No KEGG map assigned

rxn00278 Alanine Dehydrogenase 1.4.1.1 peg.940+peg.941+peg.942 Alanine, aspartate and glutamate

Reactions common to R6 and TIGR4


rxn05155 L-Glutamine-ABC transport 3.A.1.3 peg.763|peg.1154 Glutamine/aspartate ABC transporter

rxn01352 dGTP triphosphohydrolase 3.1.5.1 peg.1201 Purine


28

Reactions common to R6 and G54


rxn01675 dTTP:alpha-D-glucose-1-phosphate thymidylyltransferase 2.7.7.24 peg.323 Streptomycin

rxn02000 dTDP-4-dehydro-6-deoxy-D-glucose 3,5-epimerase 5.1.3.13 peg.324 Streptomycin

rxn01997 dTDPglucose 4,6-hydro-lyase 4.2.1.46 peg.325 Streptomycin

rxn02003 dTDP-6-deoxy-L-mannose:NADP+ 4-oxidoreductase 1.1.1.13 peg.326 Streptomycin

Reactions common to G54 and TIGR4


rxn00292 UDP-N-acetyl-D-glucosamine 2-epimerase 5.1.3.14 peg.322 Amino sugar and nucleotide amino sugar

rxn01133 Acetyl-CoA:maltose O-acetyltransferase 2.3.1.79 peg.73 Maltose utilization


29

4.2 - Genome comparison

Genomic comparison was also performed using The Comprehensive Microbial

Resource (CMR) [40] in order to demonstrate genome similarities between the strains

G54, R6 and TIGR4. Each DNA molecule was used as the reference and compared to

the other two, originating three sets of genome-homology comparisons. Table 9

illustrates the results obtained.

Table 9 – Genome homology comparison (using 90 % threshold).

Strain Hits (%)

General Specific

TIGR4 63 27

G54 71 20

R6 71 18

Analyzing the results obtained in Table 9, a high level of similarity between

strains is detected (63 - 71 %), suggesting that virulence and increased invasiveness is

determined by strain-specific genes. The latter, however, could not be confirmed by the

reaction analysis described earlier since a low percentage of strain-specific reactions

were identified. An interesting result is that of TIGR4, which was described of having

the largest number of strain-specific genes, suggesting that its high virulence could be

associated to them. However, when comparing these results to the strain-specific

reaction analysis, we do not find a correspondence between them, suggesting that other

genes (most probably those shared with other virulent strains, i.e. G54) are responsible

for virulence and its degree affected by environmental conditions.

Comparing nucleotide the nucleotide sequence of G54 used in CMR to that in

NCBI revealed a mismatch. Since we can´t determine which sequence is correct, for this

analysis, we chose to assume the sequence of G54 provided by CMR.

4.3 - Individual Models


30

The data obtained from the reconstruction process for S.pneumoniae strains G54,

TIGR4, R6 and Generic model and loaded onto Optlux are listed in Tables 4, 5, 6 and 7,

respectively. Biomass flux using Hoeprich´s medium was calculated only for the

generic model. Biomass flux for the individual strains of S.pneumoniae was only carried

out in SEED using the complete medium option to ensure growth simulation.

Table 10– Data collected from metabolic reconstruction model of strain G54.

Internal

reactions

Drains External

metabolites

Internal

metabolites

Nº of genes Biomass

flux (h-1

)

768 82 78 727 597 N/A

Table 11– Data collected from metabolic reconstruction model of strain TIGR4.

Internal

reactions

Drains External

metabolites

Internal

metabolites


flux (h-1

)

749 97 92 717 563 N/A

Table 12– Data collected from metabolic reconstruction model of strain R6.

Internal

reactions

Drains External

metabolites

Internal

metabolites


flux (h-1

)

770 102 97 733 577 N/A

Table 13– Data collected from metabolic reconstruction of the generic model.

4.4 - Generic Model

The generic model was comprised of all the reactions from the three individual

models (Supplementary data). All reactions were copied to a new file and duplicates

were removed. In order to obtain a minimal reliable model, manual curation was

Internal

reactions

Drains External

metabolites

Internal

metabolites


flux (h-1

)

836 109 107 738 588 2.134


31

performed to this generic model. Pathways chosen for curation were based on media

composition and inspected to ensure correctness or detect gaps for completion.

4.4.1 - Biomass Reaction

The biomass reaction is one of the most important components of a metabolic

model. This reaction determines which compounds are essential to the model in order

to simulate growth. SEED uses gene-based annotation to reconstruct the model and,

therefore, a subsequent biomass reaction is defined in accordance to the organism-

specific requirements integrated with supplementary data shown in annex 2.

SEED´s template for biomass reaction is constructed from a curation of 19 existing

genome-scale metabolic models [35]. This tool assumes that several compounds (39) in

the biomass reaction are universal to all organisms (i.e. nucleotides, amino acids),

whilst others are present depending on organism-specific features and genome

annotation, such as cell wall composition based on Gram-positive or negative, and

cofactors. Stoichiometric coefficient determination is based on a set of rules in order to

approximate these to the individual biomass reaction since most of the times the

experimental data required for these calculations is not available.

Due to the fact that the aim of this study is to reconstruct a “generic” model for

S.pneumoniae, the biomass reaction was assessed by reunion of the three individual

models. Inspection of the biomass equations of each model revealed no differences in

their composition, therefore, only coefficient calculations were performed. These values

were very similar among all three strains, so a coefficient average value was calculated

for each metabolite. The composition and coefficient values are shown n Table 9.

Table 14 – Biomass composition and compound coefficients

Name Stoichiometry

Reactants

CoA 0.005521608

Lysine 0.333545602

Cysteine 0.088980184


32

Asparagine 0.234678731

dCTP 0.009329285

Calcium 0.005521608

Phosphatidylglyceroldioctadecanoyl 0.014200641

Manganese 0.005521608

GTP 0.209101254

Undecaprenyldiphosphate 0.092476489

Glutamate 0.256013161

Stearoylcardiolipin(B.subtilis) 0.014200641

Aspartate 0.234678731

Isoleucine 0.282551111

Leucine 0.438656697

Zinc 0.005521608

Fe2+ 0.005521608

CTP 0.129876555

Glycine 0.595802987

DNA_replication 1

Spermidine 0.005521608

Fe3+ 0.005521608

Chloride 0.005521608

Alanine 0.500058228

UTP 0.14026668

Protein_biosynthesis 1

Serine 0.209701837

Valine 0.411598396

Glutamine 0.256013161

Copper2 0.005521608

Potassium 0.005521608

TTP 0.017278415

RNA_transcription 1

dATP 0.017278415

Cobalt 0.005521608


33

Riboflavin 0.005521608

dGTP 0.009329285

H2O 3.479.648.046

Thiamindiphosphate 0.005521608

Magnesium 0.005521608

Arginine 0.288274982

Pyridoxalphosphate 0.005521608

Putrescine 0.0053191489

FAD 0.005521608

Sulfate 0.005521608

ATP 4.017.013.829

Threonine 0.246646826

Proline 0.215425709

Acyl-carrierprotein 0.005521608

Tetrahydrofolate 0.005521608

5-Methyltetrahydrofolate 0.005521608

10-Formyltetrahydrofolate 0.005521608

Methionine 0.149341011

S-Adenosyl-L-methionine 0.005521608

Glutathione 0.0053191489

Histidine 0.092622648

NADP+ 0.005521608

NAD+ 0.005521608

Phenylalanine 0.180562128

Tyrosine 0.134250804

Tryptophan 0.055157307

Phosphotodylglycerol_dioctadecanoyl 0.014200641

Peptidoglycan_polymer (n subunits) 0.092476489

Diisoheptadecanoylphosphatidylglycerol 0.014200641

Dianteisoheptadecanoylphosphatidylglycerol 0.014200641

Isoheptadecanoylcardiolipin_(B. subtilis) 0.014200641

Anteisoheptadecanoylcardiolipin_(B. subtilis) 0.014200641


34

Products

Apo-[acyl-carrier-protein] 0.005521608

Biomass 1

ADP 40

Pyrophosphate 0.602386901

Orthophosphate 3.999.447.839

Hidrogen 40

Peptidoglycanpolymer(n-1subunits) 0.092476489

4.4.2 - Main Pathway Analysis

4.4.2.1 - Carbohydrates Metabolism

4.4.2.1.1 - Glycolysis

Glycolysis is an essential pathway for all organisms, due to the fact that it is

responsible for the conversion of glucose into pyruvate, generating energy (ATP) and

reducing power (NADH). Analysis of the pathway map on KEGG and confirmed in our

model, showed that all the necessary reactions (and respective protein encoding genes)

for glucose degradation were present, not requiring further in-depth analysis.

4.4.2.1.3 - Pentose Phosphate Pathway (PPP)

The Pentose Phosphate pathway plays an important part in the carbohydrates

metabolism due to its products: NADPH, which is a reducing power required, for

instance, in fatty acid biosynthesis; and ribose-5-phosphate, which is necessary for

nucleotide synthesis [52][53] and also required for FAD formation in the Riboflavin

metabolism .

The reactions present in the generic model fulfilled the main necessary steps for

R5P formation, and therefore the pathway was considered accurate.


35

4.4.2.1.4 - Fructose and Mannose

Hoskins and colleagues identified all genes necessary for carbohydrate oxidation

into pyruvate via glycolysis [51]. S.pneumoniae can use either the

phosphoenolpyruvate-dependent phosphotransferase system (PTS) or the less energetic-

efficient ATP Binding Cassette Transport system (ABC transporters).

Fructose is a monosaccharide present in human diet and mannose a sugar monomer,

both of which are degraded to Fructose-6-phosphate by a frutokinase (E.C= 2.7.1.4) and

a mannose-6-phosphate (E.C= 5.1.3.8). F6P can either enter the glycolytic pathway for

energy production or the amino sugar and nucleotide metabolism for conversion into

UDP-N-acetylmuramate which will enter the peptidoglycan biosynthesis pathway,

important for cell wall formation[3][51]. Simulations under different experimental

conditions might reveal which pathway is chosen.

Inspection of the generic network model showed that all protein encoding genes

responsible for these reactions were present and considering that both scenarios were

possible, no changes were performed.

4.4.2.1.5 - Galactose

Galactose is another monosaccharide utilized by S.pneumoniae in the

carbohydrate metabolism[51]. Observation on KEGG pathway map in comparison to

the generic model showed that all protein encoding genes responsible for all the multi-

step reactions necessary for conversion of galactose to Glucose-6-phosphate, which

subsequently enters the glycolysis pathway, were accounted for.

4.4.2.1.6 - Starch and sucrose


36

Sucrose is a disaccharide composed of fructose and glucose that plays an

important role in S.pneumoniae fitness. Colonization and infection have been linked to

two metabolizing systems, sus and scr, both regulated by the LacI family, which has

been shown to act as a regulator for sucrose transport systems, that have niche-specific

roles in virulence [54]. The first, linked to SusR, primarily affects the lungs, whilst the

latter, linked to ScrR, is primarily associated to the nasopharynx. ScrH, a sucrose-6-

phosphate hydrolase of the PTS is considered to be the main hydrolase involved in this

process. The PTS system transporter subunit IIBCA (E.C = 2.7.1.69), is encoded by the

genes SPG_1627 (G54), SP_1722 (TIGR4) and spr1566 (R6) present in our model,

suggesting that in the case of strain R6 these metabolizing systems are not expressed.

Streptococcus pneumoniae have different possible niches in the human host, therefore,

identifying which system is being expressed might be a form of clustering strains, other

than serotyping.

As occurred in the glycolysis, the starch and sucrose pathway appeared to be

correctly annotated and no re-annotation was required.

4.4.2.1.7 - Pyruvate

Pyruvate or pyruvic acid is an organic acid originated by conversion of PEP and

produced via glycolysis, being able to either, reversibly, regenerate glucose via

glucogenesis pathway or enter the fatty acids metabolism in the form of acetyl-CoA.

Energy production normally follows the fermentation process, obtaining lactic acid and

hydrogen peroxide under anaerobic and aerobic growth conditions, respectively

[11][51].

Analysis of this pathway confirmed the presence of protein encoding genes for

all necessary enzymes that catalyze these reactions.

4.4.2.1.8 - Amino sugar and nucleotide sugar


37

Amino sugars are important substrates for glycosyltransferases which are

responsible for polysaccharide synthesis. Carbohydrates such as fructose (fructose and

mannose pathway), glucose (glycolysis pathway), galactose (galactose pathway) and

mannose (extracellular providence, i.e. medium) are degraded to UDP-N-

acetylmuramate which, in turn, plays a key role in peptidoglycan biosynthesis.

Inspection of this pathway revealed that protein encoding genes responsible for

all the enzymes required for UDP-N-acetylmuramate production from the substrates

mentioned above were present.

4.4.2.2 - Energy Metabolism

4.4.2.2.1 - Nitrogen

Nitrogen metabolism is important, not only in bacterial growth, but also for its

contribution to virulence [55] determined by the presence of a gene (glnA) required for

glutamine uptake[56]. Studies [57] have reported that the internal pool-size of

glutamine is key when S.pneumoniae is grown in a nitrogen-limiting environment.

Carbon dioxide (CO2) also plays an important role in several cellular processes,

besides cellular growth. A carbonic hydrolase (E.C=4.2.1.1), encoded by genes

SP_00024 (TIGR4), SPG_0031 (G54) and spr0026 (R6), present in this pathway and

our model, has been associated to growth limiting factors. Several studies [51][42][58]

have concluded that in the absence of CO2 or in CO2-poor environments, growth of

S.pneumoniae is inhibited or limited. Burghout and co-workers [59] classified this

enzyme as essential in preventing the cellular release of CO2, since pneumococci have

an anabolic necessity for either carbon dioxide (CO2) or bicarbonate (HCO3-) during

nucleic, amino and fatty acid biosynthesis.

Analysis of the nitrogen metabolism showed that type I glutamine synthetase,

enzyme encoded by gene glnA was present in the model as was carbamate kinase (E.C=

2.7.2.2), encoded by SPG_2090, required for formation of carbamoyl-P, important to

the arginine and proline pathway.


38

4.4.2.3 - Lipid Metabolism

4.4.2.3.1 - Fatty acid Biosynthesis

As stated before, the draft reconstruction of the networks obtained from SEED

revealed gaps in several pathways. Analysis of the fatty acid biosynthesis showed that

all last steps of the elongation cycle were missing; however, Marakchi [60] and

colleagues identified the genes sp_0419(TIGR4)[11], spr0379(R6) and spg0385(G54) in

S.pneumoniae that encoded the enoyl(acyl-carrier-protein)(ACP) reductase FabK which

was responsible for the these precise steps. The last cycle was also missing from the

fatty acid biosynthesis which led to the production of octadecanoic and octadecenoic

acids. Studies have demonstrated that enzymes FabF, FabZ and FabG were encoded by

genes sp0422, sp0424 and sp0422 of TIGR4 strain; spr0382, spr0384 and spr0382 of R6

strain; and spg0388, spg0390 and spg0388 of G54 strain[11][61].

Oleoyl-[acyl-carrier-protein] hydrolase (E.C= 3.1.2.14), which catalyzes the production

of hexadecanoic, tetradecanoic, dodecanoic, octadecanoic and octadecenoic acids, was

shown to be encoded by genes sp1408 (TIGR4), spr1265 (R6) and spg1349 (G54).

Table 8 shows the relationships between genes and enzymes in the Fatty acid

biosynthesis added to the model and Annex 1 (obtained through KEGG) illustrates Fatty

acid Biosynthesis pathway with enzymes that have been proven to exist in the strains of

S.pneumoniae used in this study (highlighted in green).

Table 15 – Relations Gene/Enzyme in Fatty acid Biosynthesis

Genes Enzyme

sp0422(TIGR4), spr0381(R6),spg0388(G54) FabG

sp0422(TIGR4), spr0382(R6),spg0388(G54) FabF

sp0419(TIGR4), spr0379(R6),spg0385(G54) FabK


39

sp0424(TIGR4), spr0384(R6),spg0390(G54) FabZ

sp_1408(TIGR4), spr1265(R6),spg_1349(G54) 3.1.2.14

4.4.2.4 - Nucleotide Metabolism

4.4.2.4.1 - Purine and Pyrimidine

Purines adenine and guanine and pyrimidines uracil, cytosine and thymine, are

essential in many basic biological processes, mainly providing the metabolites required

for DNA replication and RNA transcription, linked to bacterial growth [62] and energy

storage units in the form of ATP. Gibert´s [43] medium solution had adenine, guanine

and uracil, the latter being important in these pathways, which are vital precursors in

DNA replication and RNA transcription.

Inspection of these pathways in the generic model revealed incomplete reactions

for DNA replication and RNA transcription, neglecting the use of ATP and dATP, and

only attended to these when dGTP and GTP acted as precursors. Through homology, an

enzyme encoding gene (rpoB) responsible for production of DNA-directed RNA

polymerase subunit alpha (E.C = 2.7.7.6), required for RNA transcription, was

identified. This enzyme catalyzes this reaction, either in the use of ATP (R00435) or

GTP (rxn13784) as precursors.

The same methodology was applied to the DNA replication reaction. By

homology, enzyme encoding genes (DPA, dnaE and polA) were identified as

responsible for the production of the DNA polymerase I (E.C = 2.7.7.7). Reactions,

either using dATP (R00375) or dGTP (rxn13783) as precursors, are catalyzed by this

enzyme.

For simulation purposes, since adenine and guanine have been reported to be

present in the media[43], and considering their presence in our model, drains for

adenine and guanine entry or exit were added.

During the reconstruction process, a reaction for ATP maintenance is also

required. This process is key to ensure enough energy for basic biological procedures


40

independent of growth. SEED draft reconstruction pipeline created such a reaction

(rxn00062) and established a positive flux (only allowing degradation of ATP to ADP).

Equation 5 illustrates ATP maintenance reaction according to Thiele and Palsson´s

protocol[34].

1 ATP + 1 H2O --> 1 ADP + 1 H + 1 Pi Eq. 5

Several experimental data consulted did not refer these values and investigation

of several other models (Table 11) only presented a positive flux (0 to ∞), however a

study by Oliveira and colleagues [63] on a genome-scale model for Lactococcus lactis,

stated using a flux of 18.15 mmol/ gDW. Growth rate for this species (0.8 h-1

) was very

similar to the maximum obtained for S.pneumoniae (0.92 h-1

) and the proximity

between species suggested we could use this value as standard. This obviously requires

further study but is indicative of the non-growth energy requirements.

Table 16 – List of models consulted for ATP maintenance limit values

Name Source

iYO844 (B.subtilis) SEED

iBsu1103 (B.subtilis) PubMed Central (PMC2718503)

iJR904 (E.coli) Genomebiology.com

Seed370552.3 (S.pyogenes) SEED

Ópt171101.1 (S.pneumoniae R6) SEED

Lactococcus lactis (ssp. Lactis ILI 403) http://www.biomedcentral.com/1471-

2180/5/39

4.4.2.5 - Amino acid Metabolism

Amino acids are structural units of proteins. Different sequences and lengths of

amino acids determine the type of protein that will be formed. Since amino acids can

also be used as nitrogen and carbon sources [11] [51] along with the presence of several

amino acid transporters in the generic model, we decided to deepen our analysis on

these pathways. Additionally, we had to keep in mind that not all amino acids were

produced by S.pneumoniae, requiring them to be introduced via media solution. Some

of them are capable of being produced, cases of valine [51], proline, aspartate (via L-


41

Asparagine), Isoleucine (via L-Threonine)[11], but as a safeguard, due to some media

containing all amino acids in the medium [41] it was decided to allow the introduction

of these from an extracellular compartment. Hartel´s isotopologue profiling of

pneumoniae’s central carbon metabolism [42] demonstrated which amino acids were

essential and non-essential for in vitro growth and, therefore, simulations based upon

the same assumptions were performed in order to validate their findings (Table 17).

Table 17 – Essential and non-essential amino acids for in vitro growth of Streptococcus

pneumoniae D39 (serotype 2; NCTC7466) compared to our generic model. ((0) – No

effect on growth; (+) - Essential for growth; (-) - Limiting in case of absence from the

medium)

Amino acid Generic Model Hartel et al

Arginine 0 +

Cysteine + +

Histidine + +

Glycine + +

Glutamine + +

Isoleucine 0 +

Leucine 0 +

Valine + +

Threonine 0 0

Serine 0 0

Asparagine 0 0

Aspartate 0 0

Alanine 0 0

Phenylalanine 0 0

Tyrosine 0 0

Tryptophan 0 0

Lysine 0 0

Proline - -

Methionine 0 -


42

Comparison between results obtained by Hartel and our generic model show that almost

all of them coincide, which proves to a certain extent that our model is simulating

correctly phenotypic behavior.

Further analysis is provided, in the following descriptions, for each amino acid pathway.

4.4.2.5.1 - Alanine, Aspartate and Glutamate

Alanine plays a key role in glycolysis, converted into pyruvate by an alanine

synthesizing transaminase (E.C= 2.6.1.66) (gene aspC), and peptidoglycan synthesis, in

the form of D-alanyl-D-alanine by action of a D-alanyl-alanine synthetase S (E.C=

6.3.2.4) (gene ddl).

Aspartate is converted into N-carbamoyl-L-aspartate by action of an aspartate

carbamoyltransferase catalytic subunit (E.C= 2.1.3.2) (gene pyrB), which enters the

pyrimidine pathway, essential in various biological processes (see Nucleotide

pathways).

Glutamate is converted into L-Glutamine by a type I glutamine synthetase (E.C=

6.3.1.2) (gene glnA). The latter can enter the amino sugars pathway, in the form of D-

Glucosamine-6-Phosphate, the Purine pathway, as 5-Phospho-ribosylamine, or the

pyrimidine and arginine and proline pathways, in the form of carbamoyl-phosphate, by

action of glucosamine—fructose-6-phosphate aminotransferase (E.C= 2.6.1.16) (gene

glmS), amidophosphoribosyltransferase (E.C= 2.4.2.14) (gene purF) and carbamoyl

phosphate synthase large subunit (E.C= 6.3.5.5) (genes carB and carA), respectively.

Glutamine also plays a key role in virulence [62].

Out of these three amino acids only glutamate was considered essential for growth by

Hartel and co-workers[42], and in silico simulations, using the same assumptions,

confirmed their findings. Inspection of our model confirmed the presence of all enzyme

encoding genes required for all the reactions of this pathway and, therefore, no

additional curation was performed.


43

4.4.2.5.2 - Glycine, Serine and Threonine

Glycine, serine and threonine production can be achieved by S.pneumoniae in

accordance to KEGG maps and confirmed in our model. The presence of enzyme

encoding genes for the main reactions (L-serine ammonia-lyase and L-threonine

ammonia-lyase – gene ilvA, for converting glycine to serine and threonine to 2-

oxobutanoate; tryptophan syntethase subunit alpha – gene trpA, responsible for

converting serine to tryptophan; serine hydroxymethyltransferase – gene glyA, used to

produce pyruvate having serine as its main precursor; threonine syntethase – genes

thrC, thrB1, thrB2 and thrH, responsible for production of threonine with homoserine as

the main precursor; and aspartate kinase – genes SP_0413 (TIGR4), SPG_0379 (G54)

and spr0374 (R6), used for production of homoserine having aspartate as the main

precursor, revealed suffice for considering this pathway complete. However, studies

have reported that glycine, unlike serine and threonine, is essential for growth, and only

obtained from the medium [42], therefore, a drain (EX_cpd11580_e) was added to the

model to safeguard this condition.

4.4.2.5.3 - Cysteine and Methionine

Methionine is the universal N-terminal amino acid of proteins [64] and

cysteine´s thiol side chain oxidation leads to formation of cystine, important in protein

structure[65].

Initially, Cysteine and Methionine pathway analysis showed a gap in the

transformation of L-Cystathionine to L-Homocysteine. Further analysis demonstrated

that this cysteine-lyase (E.C= 4.4.1.8) was encoded by genes SP_1524 (TIGR4),

SPG_1449 (G54) and spr1376 (R6), and also involved in other reactions, namely

pyruvate and thiocysteine production. The enzyme that catalyzes the latter is encoded

by the genes SP_1524 (TIGR4), SPG_1449 and spr1376 (R6) but its inclusion would

create a dead-end in the model, since no genes have yet been identified that encode

enzymes for reactions necessary to connect this metabolite with other compounds.


44

Studies regarding cysteine and methionine metabolism [42] showed that the first was

essential for growth whilst the second had a limiting effect when absent. When

performing simulations in our model, under these same conditions, similar results were

obtained for cysteine but not methionine. A possible explanation for the latter is the de

novo synthesis of methionine. The same studies were incapable of determining if this

could be achieved due to degradation of the amino acid during acidic treatment of

protein hydrolysis.

4.4.2.5.4 - Valine, Leucine and Isoleucine biosynthesis

Valine, Leucine and Isoleucine primarily act as the building blocks of proteins.

Hartel´s study stated that valine, isoleucine and leucine were essential for growth and

inspection of the pneumococcal genome revealed the existence of genes required to

encode all necessary enzymes for their synthesis using threonine and pyruvate as

precursors. Homologous genes in the genome such as ilvB have also been linked to

valine and isoleucine biosynthesis.

Inspection of our model confirmed that the enzyme encoding genes required for

the set of reactions leading to their production were present (SP_0450, SP_0445,

SP_0446 (ilvH), SP_0447, SP_2126 and SP_0856) and therefore, despite obtaining

different results in our simulation when compared to Hartel´s study in the cases of

isoleucine and leucine, the possibility of synthesis via pyruvate and threonine was

considered possible.

Since all the enzymes in this pathway were linked to a protein encoding gene, re-

annotation was not performed.

4.4.2.5.5 - Phenylalanine, Tyrosine and Tryptophan biosynthesis


45

As mentioned earlier, S.pneumoniae is capable of utilizing amino acids as

carbon sources. However, aromatic amino acids like phenylalanine, tyrosine and

tryptophan do not appear to affect growth in vitro using glucose as main carbon source

[42], also sustained by simulations performed to the model. Hartel and colleagues stated

that these amino acids were capable of de novo synthesis from erythrose 4-phosphate

along with two molecules of PEP via chorismate pathway.

4.4.2.5.6 - Lysine biosynthesis

Two distinct pathways for synthesis of lysine are known: the diaminopimelate

(DAP), which is found mostly in bacteria; and alfa-aminoadipate, found mostly in fungi

and archeal species [66]. Special interest was taken for this metabolic pathway since

DAP is also associated as a constituent of bacterial cell wall peptidoglycan, although

lysine it is not essential for growth [42], conclusion also sustained by simulations

performed on our model.

The first draft model revealed an incomplete pathway when aspartate was the

key substrate, namely the absence of a diaminopimelate epimerase (E.C= 5.1.1.7)

required for production of meso-2,6-diaminopimelate. Through homology, the gene

dapF (responsible for the production of this enzyme) was identified, allowing the

addition of the missing reaction (R02735).

4.4.2.5.7 - Arginine and Proline

Proline and arginine play an important role in bacterial growth, mainly by

formation of glutamate which, in turn, is used to produce glutamine. Proline, although a

non-essential amino acid for growth[42], unlike arginine[67] , has a limiting effect

when absent from the medium. Simulations in our model where each of these amino

acids were removed individually, confirmed Hartel´s conclusions in the case of proline

but not arginine. This could be caused by an erratic reconstruction of the metabolic

pathway or due to its capability for de novo synthesis. Hartel´s study was inconclusive


46

for the latter due to experimental procedural limitations (degradation during acidic

treatment). Analysis of KEGG pathway identified genes responsible for arginine

formation by use of citrulline (SP_2148 – TIGR4, SPG_2088 – G54, and

spr1955/spr1956 – R6) and their presence in our model suggests that de novo synthesis

is the most likely explanation.

4.4.2.6 - Metabolism of Cofactors and Vitamins

4.4.2.6.1 - Pantothenate and CoA biosynthesis

Pantothenate is vital for Co-enzyme A (CoA) production, which is essential in

metabolic pathways such as Fatty acid biosynthesis and energy metabolism (TCA

cycle), mostly by acting as a carbon transporter within the cell, and necessary as an

integral part of acyl-carrier protein (ACP)[68].

Analysis of the reconstructed model showed that one reaction - rxn01790, (R)-

PantoateNADP+ 2-oxidoreductase, was present that had no literature background and

therefor was removed. Further inspection revealed the absence of Pantothenate,

importance stated above, that was added to the model – rxn10180. Pantothenate has

been included in media [41][42] and the identification of a protein - Pantothenate kinase

– (EC=2.7.1.33) encoded by gene coaA present in our genome (obtained through

homology), only enhanced our confidence level when adding this reaction.

ACP is essential in fatty acid biosynthesis, and inspection of the latter revealed

the absence of ACP which prevented fatty acid formation. Addition of the reaction

R01625 corrected this gap.

4.4.2.6.2 - Riboflavin

Riboflavin is a micronutrient vital for flavin mononucleotide and flavin adenine

dinucleotide cofactor production and, as such, plays an important part in the energy and

fatty acids metabolisms. These cofactors are mostly used as prosthetic groups for

http://enzyme.expasy.org/EC/2.7.1.33


47

several oxidoreductases [[69], such as glycerol-3-phosphate oxidase (E.C= 1.1.3.21)

identified in the glicerophospholipid pathway of our generic model.

4.4.2.6.3 - Vitamin B6

The presence of pyridoxal phosphate in our model corroborated Hoskins

statement that S.pneumoniae possessed an incomplete pathway for this metabolism.

Pyridoxal was not present in our model which suggested that it had to be added to the

medium. Several studies [41][45][42] confirmed the addition of this compound to the

medium and, therefore, a drain was added to the model – R_pyridoxal - and its

subsequent transport reaction – R00174.

4.4.2.7 - Other metabolisms

Several other metabolisms presented to be incomplete. Over 50 % of the

reactions added during the auto-completion stage in SEED are associated to processes

of cofactor or cell wall biosynthesis [35]. Many pathogens do not possess the capability

to produce certain metabolites, obtaining them extracellularly, for instance the host´s

cells. An example of such is thiamine, a vitamin required for transaminase reactions,

biotin, essential in biosynthetic reactions that require CO2 fixation, or lipoteichoic acids,

a component of the cell wall. For each of these cases, a drain was added to the model to

ensure that the metabolite could be used if need be.

4.5 - Simulation results with Hoeprich´s medium

The medium used for conducting simulations is shown in Table 4 (Hoeprich´s

medium), and simulated a growth (production of biomass) of 2.135 h-1

which is


48

approximately more than double of the results obtained in that same experiment (0.92 h-

1 using strain St 99/95). The biomass metabolites of the generic model are listed in

Table 14. In Gonçalves study [41] lactate and polysaccharide capsule production were

also measured. Lactate formation, a by-product of lactic bacteria, was also detected, but

the elevated value (over 1000 h-1

) was obviously incorrect when compared to data

retrieved from literature (1.90 h-1

). The fact that the fluxes that lead to lactate formation

have no constraints might lead to this outcome and therefore, they need to be adjusted.

Simulations for polysaccharide production were not performed.

Differences in the biomass equation and the addition of strain-specific reactions

might explain why the results obtained did not match those in literature. Because the

“generic” biomass equation was comprised of the compounds from all three strains and

its coefficients obtained by average as well as the addition of reactions that were

specific to each strain, differences in biomass flux were expected.


49

5 - Conclusions

One of the most difficult tasks was the gathering of useful experimental data.

The lack of knowledge on a minimum medium for S.pneumoniae is an obstacle when

trying to simulate growth and therefor requires researchers to contemplate adding

excessive reactions in order obtain a functioning model. Most of the articles refer

complex media complemented with other amino acids, vitamins or buffer solution in

order to obtain growth.

Due to the necessity of complex medium and little information on what is

strictly necessary for the growth of S.pneumoniae, many authors describe using all

amino acids in the media regardless of knowing if they are essential [41][43][45].

However, Hartel and colleagues [42] performed the characterization of the Central

Carbon Metabolism of S.pneumoniae through isotopogue profiling and concluded that

most of the amino acids were essential in different conditions, be it in studying the de

novo synthesis of aspartate, alanine and threonine, be it in the study of growth under

amino acid depravation, only to mention a few. Furthermore, they demonstrated the

dual utilization of carbohydrates and amino acids as well as the de novo synthesis of

several of the latter. These results sustained the changes made to the model where all

fluxes of amino acids were “opened” (Table 9 – List of aminopeptidases). However,

some studies [41] state that not all amino acids are capable of de novo synthesis,

requiring them to be present in the medium. Removal of amino acids such as

phenylalanine, tyrosine and tryptophan, from the biomass equation, confirmed Hartel´s

hypothesis that these had no impact on pneumococci growth. Use of different

experimental conditions might be able to shed some light into amino acid essentiality

and production since enzyme encoding genes for reactions involving de novo synthesis

for some of them (i.e. isoleucine and leucine) are present in the model.

Streptococcus pneumoniae requires carbohydrates to grow and the generic

model demonstrated a large amount of reactions that were linked to transport systems

such as ABC transporters which, in turn, indicated that pneumococci were capable of

utilizing a wide variety of such to improve bacterial fitness and virulence[42].

Additional information regarding protein synthesis, DNA and RNA production is


50

essential in order to establish proper boundaries for ATP costs. The lack of information

regarding these precursors is limiting to the quality of the model.

Besides achieving biomass flux proximate to experimental data, the presence of

lactate as a by-product, also known to be present in Lactobacteria, proved to some

extent that the model was indeed simulating correctly. The flux obtained was too high to

be correct and did not match the one obtained through literature, therefore further

inspection and refinement of the model must be carried out.

Table 18 – List of aminopeptidases

Drain (EX_cpd) Name

11580

Aminopeptidases

11581

11582

11583

11584

11585

11586

11587

11588

11589

11590

11591

11592

11593

15603

15604

15605

15606

1017

A list of all reactions added/changed to the generic model, are presented in Table 11.


51

Table 19 – List of all reactions added/changed in generic model

Reaction E.C Number

R02735 5.1.1.7

R02788 6.2.3.13

R02081 1.1.1.88

R07763 1.1.1.100

R07762 2.3.1.41|2.3.1.179|2.3.1.-

R07765 1.3.1.-

R07764 4.2.1.-

R04968 2.3.1.41|2.3.1.85|2.3.1.86|2.3.1.179|2.3.1.180

R01706 3.1.2.14

R04429 1.3.1.9|2.3.1.86

R04724 1.3.1.9|2.3.1.86

R04955 1.3.1.9|2.3.1.86

R04958 1.3.1.9|2.3.1.86

R04961 1.3.1.9|2.3.1.86

R04966 1.3.1.9|2.3.1.86

R04969 1.3.1.9|2.3.1.86

R03370 1.14.19.2

R08159 3.1.2.14

R01625 2.7.8.7

R01011 2.7.1.29

R00174 2.7.1.35

R01286 4.4.1.8

biogeneric

R00375 2.7.7.7

rxn13783 2.7.7.7

R00435 2.7.7.6

rxn13784 2.7.7.6


52

6 - Future Work

Minimum media required for pneumococci growth is essential to obtain a

precise and reliable metabolic model. Therefore, the first step to achieve this would be

to experimentally determine such media composition. Another possibility would be to

reconstruct models for several strains of S.pneumoniae by manual curation and, only

after each of them are validated, would the process of generating a generic model begin.

As the time frame for the presentation of this dissertation is limited, this process was

done by using models with little or no manual curation and performing the necessary

steps to unite them, namely removing duplicate reactions, inconsistencies in compound

presence, incorrect pathways, curation of critical reactions, amongst others.

Quantification of protein, DNA and RNA is also essential to establish minimum

energy requirements for organism sustainability and therefore should be addressed in

future studies. Using different experimental conditions (constraints) might also shed

some light into pathway preferences and, therefore, should be addressed in the future.

Upload of this model onto other modeling software (i.e. Merlin [70]) could also

improve the quality of the annotation.


53

7 - References

[1] F. Ding, P. Tang, M.-H. Hsu, P. Cui, S. Hu, J. Yu, and C.-H. Chiu, “Genome

evolution driven by host adaptations results in a more virulent and antimicrobial-

resistant Streptococcus pneumoniae serotype 14.,” BMC genomics, vol. 10, p.

158, Jan. 2009.

[2] “Streptococcus pneumoniae,” in The desktop Encyclopedia of Microbiology,

2009, pp. 1061–1074.

[3] J. Dopazo, A. Mendoza, J. Herrero, F. Caldara, Y. Humbert, L. Friedli, M.

Guerrier, E. Grand-schenk, C. Gandin, M. D. E. Francesco, A. Polissi, G. Buell,

G. Feger, and E. García, “Annotated Draft Genomic Sequence from a

Streptococcus pneumoniae Type 19F Clinical Isolate * MATERIALS AND

METHODS,” vol. 7, no. 2, 2001.

[4] A. Sá-Leão, R, Tomasz, “Streptococcus Pneumoniae,” in The desktop

Encyclopedia of Microbiology, 2009, pp. 1061–1074.

[5] J. a Lanie, W.-L. Ng, K. M. Kazmierczak, T. M. Andrzejewski, T. M. Davidsen,

K. J. Wayne, H. Tettelin, J. I. Glass, and M. E. Winkler, “Genome sequence of

Avery’s virulent serotype 2 strain D39 of Streptococcus pneumoniae and

comparison with that of unencapsulated laboratory strain R6.,” Journal of

bacteriology, vol. 189, no. 1, pp. 38–51, Jan. 2007.

[6] M. Gratten, D. Battistutta, P. Torzillo, J. Dixon, and K. Manning, “Comparison

of goat and horse blood as culture medium supplements for isolation and

identification of Haemophilus influenzae and Streptococcus pneumoniae from

upper respiratory tract secretions.,” Journal of clinical microbiology, vol. 32, no.

11, pp. 2871–2, Nov. 1994.

[7] I. Serrano and J. A. Carric, “Characterization of the Genetic Lineages

Responsible for Pneumococcal Invasive Disease in Portugal,” vol. 43, no. 4, pp.

1706–1715, 2005.

[8] M. Carrolo, F. R. Pinto, J. Melo-Cristino, and M. Ramirez, “Pherotypes are

driving genetic differentiation within Streptococcus pneumoniae.,” BMC

microbiology, vol. 9, p. 191, Jan. 2009.

[9] M. D. L. Moura Leal, D. D. S. Gomes Pereira, E. Jessouroun, M. A. Peixoto

Gimenes Couto, and N. Pereira Jr, “Investigation of cultivation conditions for

capsular polysaccharide production by Streptococcus pneumoniae serotype 14,”

Electronic Journal of Biotechnology, vol. 14, no. 5, pp. 0–3, Sep. 2011.

[10] C. Hyams, E. Camberlein, J. M. Cohen, K. Bax, and J. S. Brown, “The

Streptococcus pneumoniae capsule inhibits complement activity and neutrophil

phagocytosis by multiple mechanisms.,” Infection and immunity, vol. 78, no. 2,

pp. 704–15, Feb. 2010.


54

[11] H. Tettelin, K. E. Nelson, I. T. Paulsen, J. a Eisen, T. D. Read, S. Peterson, J.

Heidelberg, R. T. DeBoy, D. H. Haft, R. J. Dodson, a S. Durkin, M. Gwinn, J. F.

Kolonay, W. C. Nelson, J. D. Peterson, L. a Umayam, O. White, S. L. Salzberg,

M. R. Lewis, D. Radune, E. Holtzapple, H. Khouri, a M. Wolf, T. R. Utterback,

C. L. Hansen, L. a McDonald, T. V. Feldblyum, S. Angiuoli, T. Dickinson, E. K.

Hickey, I. E. Holt, B. J. Loftus, F. Yang, H. O. Smith, J. C. Venter, B. a

Dougherty, D. a Morrison, S. K. Hollingshead, and C. M. Fraser, “Complete

genome sequence of a virulent isolate of Streptococcus pneumoniae.,” Science

(New York, N.Y.), vol. 293, no. 5529, pp. 498–506, Jul. 2001.

[12] P. Yakovchuk, E. Protozanova, and M. D. Frank-Kamenetskii, “Base-stacking

and base-pairing contributions into thermal stability of the DNA double helix.,”

Nucleic acids research, vol. 34, no. 2, pp. 564–74, Jan. 2006.

[13] H.-G. Zimmer, Encyclopedia of Life Sciences. Chichester, UK: John Wiley &

Sons, Ltd, 2001.

[14] R. M. Harvey, U. H. Stroeher, A. D. Ogunniyi, H. C. Smith-Vaughan, A. J.

Leach, and J. C. Paton, “A variable region within the genome of Streptococcus

pneumoniae contributes to strain-strain variation in virulence.,” PloS one, vol. 6,

no. 5, p. e19650, Jan. 2011.

[15] M. A. Barocchi, J. Ries, X. Zogaj, C. Hemsley, B. Albiger, A. Kanth, S.

Dahlberg, J. Fernebro, and M. Moschioni, “A pneumococcal pilus influences

virulence and host,” vol. 103, no. 8, pp. 2857–2862, 2006.

[16] L. J. McAllister, A. D. Ogunniyi, U. H. Stroeher, and J. C. Paton, “Contribution

of a genomic accessory region encoding a putative cellobiose phosphotransferase

system to virulence of Streptococcus pneumoniae.,” PloS one, vol. 7, no. 2, p.

e32385, Jan. 2012.

[17] O. Johnsborg and L. S. HÃ¥varstein, “Regulation of natural genetic

transformation and acquisition of transforming DNA in Streptococcus

pneumoniae,” FEMS Microbiology Reviews, vol. 33, no. 3, pp. 627–642, May

2009.

[18] H. Steinmoen, E. Knutsen, and L. S. Håvarstein, “Induction of natural

competence in Streptococcus pneumoniae triggers lysis and DNA release from a

subfraction of the cell population.,” Proceedings of the National Academy of

Sciences of the United States of America, vol. 99, no. 11, pp. 7681–6, May 2002.

[19] W. P. Hausdorff, D. R. Feikin, and K. P. Klugman, “Epidemiological differences

among pneumococcal serotypes.,” The Lancet infectious diseases, vol. 5, no. 2,

pp. 83–93, Feb. 2005.

[20] D. of Health, “Pneumococcal,” 2012.

[21] J. D. Mooney, A. Weir, J. McMenamin, L. D. Ritchie, T. V. Macfarlane, C. R.

Simpson, S. Ahmed, C. Robertson, and S. C. Clarke, “The impact and

effectiveness of pneumococcal vaccination in Scotland for those aged 65 and


55

over during winter 2003/2004.,” BMC infectious diseases, vol. 8, p. 53, Jan.

2008.

[22] E. Alonsodevelasco, A. F. Verheul, J. Verhoef, and H. Snippe, “Streptococcus

pneumoniae : virulence factors , pathogenesis , and vaccines .,” vol. 59, no. 4,

1995.

[23] N. M. Bennett, R. Lynfield, A. Reingold, P. R. Cieslak, T. Pilishvili, D. Jackson,

R. R. Facklam, D. Ph, and J. H. Jorgensen, “new england journal,” pp. 1737–

1746, 2003.

[24] M. Hucka, a. Finney, H. M. Sauro, H. Bolouri, J. C. Doyle, H. Kitano, a. P.

Arkin, B. J. Bornstein, D. Bray, a. Cornish-Bowden, a. a. Cuellar, S. Dronov, E.

D. Gilles, M. Ginkel, V. Gor, I. I. Goryanin, W. J. Hedley, T. C. Hodgman, J.-H.

Hofmeyr, P. J. Hunter, N. S. Juty, J. L. Kasberger, a. Kremling, U. Kummer, N.

Le Novere, L. M. Loew, D. Lucio, P. Mendes, E. Minch, E. D. Mjolsness, Y.

Nakayama, M. R. Nelson, P. F. Nielsen, T. Sakurada, J. C. Schaff, B. E. Shapiro,

T. S. Shimizu, H. D. Spence, J. Stelling, K. Takahashi, M. Tomita, J. Wagner,

and J. Wang, “The systems biology markup language (SBML): a medium for

representation and exchange of biochemical network models,” Bioinformatics,

vol. 19, no. 4, pp. 524–531, Mar. 2003.

[25] M. Kanehisa, S. Goto, S. Kawashima, Y. Okuno, and M. Hattori, “The KEGG

resource for deciphering the genome.,” Nucleic acids research, vol. 32, no.

Database issue, pp. D277–80, Jan. 2004.

[26] M. W. Covert, C. H. Schilling, I. Famili, J. S. Edwards, I. I. Goryanin, E. Selkov,

and B. O. Palsson, “Metabolic modeling of microbial strains in silico.,” Trends in

biochemical sciences, vol. 26, no. 3, pp. 179–86, Mar. 2001.

[27] K. R. Patil, M. Akesson, and J. Nielsen, “Use of genome-scale microbial models

for metabolic engineering.,” Current opinion in biotechnology, vol. 15, no. 1, pp.

64–9, Feb. 2004.

[28] K. J. Kauffman, P. Prakash, and J. S. Edwards, “Advances in flux balance

analysis,” Current Opinion in Biotechnology, vol. 14, no. 5, pp. 491–496, Oct.

2003.

[29] F. Llaneras and J. Picó, “Stoichiometric modelling of cell metabolism.,” Journal

of bioscience and bioengineering, vol. 105, no. 1, pp. 1–11, Jan. 2008.

[30] I. Rocha, P. Maia, P. Evangelista, P. Vilaça, S. Soares, J. P. Pinto, J. Nielsen, K.

R. Patil, E. C. Ferreira, and M. Rocha, “OptFlux: an open-source software

platform for in silico metabolic engineering.,” BMC systems biology, vol. 4, p.

45, Jan. 2010.

[31] A. M. Feist, M. J. Herrgård, I. Thiele, J. L. Reed, and Ø. Bernhard,

“Reconstruction of Biochemical Networks in Microbial Organisms,” vol. 7, no.

2, pp. 129–143, 2011.


56

[32] K. Raman and N. Chandra, “Flux balance analysis of biological systems:

applications and challenges.,” Briefings in bioinformatics, vol. 10, no. 4, pp. 435–

49, Jul. 2009.

[33] I. Rocha, J. Förster, and J. Nielsen, “Metabolic Models,” vol. 416, pp. 409–433,

2007.

[34] I. Thiele and B. Ø. Palsson, “A protocol for generating a high-quality genome-

scale metabolic reconstruction.,” Nature protocols, vol. 5, no. 1, pp. 93–121, Jan.

2010.

[35] C. S. Henry, M. DeJongh, A. A. Best, P. M. Frybarger, B. Linsay, and R. L.

Stevens, “High-throughput generation, optimization and analysis of genome-

scale metabolic models,” Nat Biotech, vol. 28, no. 9, pp. 977–982, Sep. 2010.

[36] C. S. Henry, M. DeJongh, A. a Best, P. M. Frybarger, B. Linsay, and R. L.

Stevens, “High-throughput generation, optimization and analysis of genome-

scale metabolic models.,” Nature biotechnology, vol. 28, no. 9, pp. 977–82, Sep.

2010.

[37] M. Scheer, A. Grote, A. Chang, I. Schomburg, C. Munaretto, M. Rother, C.

Söhngen, M. Stelzer, J. Thiele, and D. Schomburg, “BRENDA, the enzyme

information system in 2011.,” Nucleic acids research, vol. 39, no. Database

issue, pp. D670–6, Jan. 2011.

[38] T. U. Consortium, “Reorganizing the protein space at the Universal Protein

Resource (UniProt).,” Nucleic acids research, vol. 40, no. Database issue, pp.

D71–5, Jan. 2012.

[39] E. W. Sayers, T. Barrett, D. a Benson, S. H. Bryant, K. Canese, V. Chetvernin,

D. M. Church, M. DiCuccio, R. Edgar, S. Federhen, M. Feolo, L. Y. Geer, W.

Helmberg, Y. Kapustin, D. Landsman, D. J. Lipman, T. L. Madden, D. R.

Maglott, V. Miller, I. Mizrachi, J. Ostell, K. D. Pruitt, G. D. Schuler, E. Sequeira,

S. T. Sherry, M. Shumway, K. Sirotkin, A. Souvorov, G. Starchenko, T. a

Tatusova, L. Wagner, E. Yaschenko, and J. Ye, “Database resources of the

National Center for Biotechnology Information.,” Nucleic acids research, vol.

37, no. Database issue, pp. D5–15, Jan. 2009.

[40] T. Davidsen, E. Beck, A. Ganapathy, R. Montgomery, N. Zafar, Q. Yang, R.

Madupu, P. Goetz, K. Galinsky, O. White, and G. Sutton, “The comprehensive

microbial resource.,” Nucleic acids research, vol. 38, no. Database issue, pp.

D340–5, Jan. 2010.

[41] V. M. Gonçalves, T. C. Zangirolami, R. L. C. Giordano, I. Raw, M. M. Tanizaki,

and R. C. Giordano, “Optimization of medium and cultivation conditions for

capsular polysaccharide production by Streptococcus pneumoniae serotype

23F.,” Applied microbiology and biotechnology, vol. 59, no. 6, pp. 713–7, Sep.

2002.


57

[42] T. Härtel, E. Eylert, and C. Schulz, “Characterization of Central Carbon

Metabolism of Streptococcus pneumoniae by Isotopologue Profiling,” Journal of

Biological …, 2012.

[43] E. B. Gibert, “Proceedings of the Society for Experimental Biology and

Medicine,” pp. 363–366, 2008.

[44] A. V. Restrepo, B. E. Salazar, M. Agudelo, C. a Rodriguez, A. F. Zuluaga, and

O. Vesga, “Optimization of culture conditions to obtain maximal growth of

penicillin-resistant Streptococcus pneumoniae.,” BMC microbiology, vol. 5, p.

34, Jan. 2005.

[45] L. J. Hathaway, S. D. Brugger, B. Morand, M. Bangert, J. U. Rotzetter, C.

Hauser, W. a Graber, S. Gore, A. Kadioglu, and K. Mühlemann, “Capsule type of

Streptococcus pneumoniae determines growth phenotype.,” PLoS pathogens, vol.

8, no. 3, p. e1002574, Jan. 2012.

[46] K. J. Nye, D. Fallon, B. Gee, S. Messer, R. E. Warren, and N. Andrews, “A

comparison of blood agar supplemented with NAD with plain blood agar and

chocolated blood agar in the isolation of Streptococcus pneumoniae and

Haemophilus influenzae from sputum. Bacterial Methods Evaluation Group.,”

Journal of medical microbiology, vol. 48, no. 12, pp. 1111–4, Dec. 1999.

[47] P. D. Rogers, J. Thornton, K. S. Barker, D. O. Mcdaniel, G. S. Sacks, E. Swiatlo,

L. S. Mcdaniel, R. E. T. Al, and I. N. I. Mmun, “Pneumolysin-Dependent and -

Independent Gene Expression Identified by cDNA Microarray Analysis of THP-

1 Human Mononuclear Cells Stimulated by Streptococcus pneumoniae,” vol. 71,

no. 4, pp. 2087–2094, 2003.

[48] K.-J. Lee, S.-M. Bae, M.-R. Lee, S.-M. Yeon, Y.-H. Lee, and K.-S. Kim,

“Proteomic analysis of growth phase-dependent proteins of Streptococcus

pneumoniae.,” Proteomics, vol. 6, no. 4, pp. 1274–82, Feb. 2006.

[49] M. Rocha, P. Maia, R. Mendes, J. P. Pinto, E. C. Ferreira, J. Nielsen, K. R. Patil,

and I. Rocha, “Natural computation meta-heuristics for the in silico optimization

of microbial strains.,” BMC bioinformatics, vol. 9, p. 499, Jan. 2008.

[50] U. Sauer, D. R. Lasko, J. Fiaux, M. Hochuli, R. Glaser, T. Szyperski, K.

Wüthrich, and J. E. Bailey, “Metabolic flux ratio analysis of genetic and

environmental modulations of Escherichia coli central carbon metabolism.,”

Journal of bacteriology, vol. 181, no. 21, pp. 6679–88, Nov. 1999.

[51] R. Strain, J. Hoskins, W. E. A. Jr, J. Arnold, C. Blaszczak, S. Burgett, B. S.

Dehoff, T. Shawn, L. Fritz, D. Fu, W. Fuller, R. Gilmour, J. S. Glass, A. R. Kraft,

R. E. Lagace, J. Donald, L. N. Lee, E. J. Lefkowitz, J. Lu, S. M. Mcahren, M.

Mchenney, C. W. Mundy, T. I. Nicas, H. Norris, M. O. Gara, R. B. Peery, T.

Gregory, P. Rockey, P. Sun, E. Malcolm, Y. Yang, M. Young-bellido, G. Zhao,

C. A. Zook, R. H. Baltz, S. R. Jaskunas, P. R. R. Jr, P. L. Skatrud, J. I. Glass, W.

E. Alborn, L. C. Blaszczak, B. S. D. E. Hoff, S. T. Estrem, C. Geringer, H.

Khoja, D. J. L. E. Blanc, M. M. C. Henney, and K. M. C. Leaster, “Genome of


58

the Bacterium Streptococcus Genome of the Bacterium Streptococcus

pneumoniae Strain R6,” 2001.

[52] M. Igoillo-Esteve, D. Maugeri, A. L. Stern, P. Beluardi, and J. J. Cazzulo, “The

pentose phosphate pathway in Trypanosoma cruzi: a potential target for the

chemotherapy of Chagas disease.,” Anais da Academia Brasileira de Ciências,

vol. 79, no. 4, pp. 649–63, Dec. 2007.

[53] M. Scheer, A. Grote, A. Chang, I. Schomburg, C. Munaretto, M. Rother, C.

Söhngen, M. Stelzer, J. Thiele, and D. Schomburg, “BRENDA, the enzyme

information system in 2011.,” Nucleic acids research, vol. 39, no. Database

issue, pp. D670–6, Jan. 2011.

[54] R. Iyer and A. Camilli, “Sucrose metabolism contributes to in vivo fitness of

Streptococcus pneumoniae.,” Molecular microbiology, vol. 66, no. 1, pp. 1–13,

Oct. 2007.

[55] G. Metabolism, “Chapter 6 Regulation of Glutamine and Glutamate Metabolism

by GlnR and GlnA in Streptococcus pneumoniae,” pp. 25097–25109.

[56] M. V. Tullius, G. Harth, and M. A. Horwitz, “Glutamine Synthetase GlnA1 Is

Essential for Growth of Mycobacterium tuberculosis in Human THP-1

Macrophages and Guinea Pigs Glutamine Synthetase GlnA1 Is Essential for

Growth of Mycobacterium tuberculosis in Human THP-1 Macrophages and

Guinea Pigs,” 2003.

[57] K. E. Klose and J. J. Mekalanos, “Simultaneous prevention of glutamine

synthesis and high-affinity transport attenuates Salmonella typhimurium

virulence . Simultaneous Prevention of Glutamine Synthesis and High-Affinity

Transport Attenuates Salmonella typhimurium Virulence,” vol. 65, no. 2, 1997.

[58] P. G. Flanagan and A. Paull, “Carbon dioxide requirements of,” pp. 669–678,

1998.

[59] P. Burghout, L. E. Cron, H. Gradstedt, B. Quintero, E. Simonetti, J. J. E. Bijlsma,

H. J. Bootsma, and P. W. M. Hermans, “Carbonic anhydrase is essential for

Streptococcus pneumoniae growth in environmental ambient air.,” Journal of

bacteriology, vol. 192, no. 15, pp. 4054–62, Aug. 2010.

[60] H. Marrakchi, W. E. Dewolf, C. Quinn, J. West, B. J. Polizzi, C. Y. So, D. J.

Holmes, S. L. Reed, R. J. Heath, D. J. Payne, C. O. Rock, and N. G. Wallis,

“Characterization of Streptococcus pneumoniae enoyl-(acyl-carrier protein)

reductase (FabK).,” The Biochemical journal, vol. 370, no. Pt 3, pp. 1055–62,

Mar. 2003.

[61] Y. Fujita, H. Matsuoka, and K. Hirooka, “Regulation of fatty acid metabolism in

bacteria.,” Molecular microbiology, vol. 66, no. 4, pp. 829–39, Nov. 2007.


59

[62] a Polissi, a Pontiggia, G. Feger, M. Altieri, H. Mottl, L. Ferrari, and D. Simon,

“Large-scale identification of virulence genes from Streptococcus pneumoniae.,”

Infection and immunity, vol. 66, no. 12, pp. 5620–9, Dec. 1998.

[63] A. P. Oliveira, J. Nielsen, and J. Förster, “Modeling Lactococcus lactis using a

genome-scale flux model.,” BMC microbiology, vol. 5, p. 39, Jan. 2005.

[64] D. a Rodionov, A. G. Vitreschak, A. a Mironov, and M. S. Gelfand,

“Comparative genomics of the methionine metabolism in Gram-positive bacteria:

a variety of regulatory systems.,” Nucleic acids research, vol. 32, no. 11, pp.

3340–53, Jan. 2004.

[65] G. Bulaj, T. Kortemme, and D. P. Goldenberg, “Ionization-reactivity

relationships for cysteine thiols in polypeptides.,” Biochemistry, vol. 37, no. 25,

pp. 8965–72, Jun. 1998.

[66] D. a. Rodionov, “Regulation of lysine biosynthesis and transport genes in

bacteria: yet another RNA riboswitch?,” Nucleic Acids Research, vol. 31, no. 23,

pp. 6748–6757, Dec. 2003.

[67] D.-S. Lee, H. Burd, J. Liu, E. Almaas, O. Wiest, A.-L. Barabási, Z. N. Oltvai,

and V. Kapatral, “Comparative genome-scale metabolic reconstruction and flux

balance analysis of multiple Staphylococcus aureus genomes identify novel

antimicrobial drug targets.,” Journal of bacteriology, vol. 191, no. 12, pp. 4015–

24, Jun. 2009.

[68] T. U. Consortium, “The universal protein resource (UniProt).,” Nucleic acids

research, vol. 36, no. Database issue, pp. D190–5, Jan. 2008.

[69] A. G. Vitreschak, D. A. Rodionov, and A. A. Mironov, “Regulation of ribo ¯

avin biosynthesis and transport genes in bacteria by transcriptional and

translational attenuation,” vol. 30, no. 14, 2002.

[70] O. Dias and M. Rocha, “Merlin : Metabolic Models Reconstruction using

Genome-Scale Information,” no. 2005, 2007.


60

Annexes


61

Annex 1 – Fatty Acid Biosynthesis of Streptococcus pneumoniae TIGR4, R6 and G54.


0

Annex 2 – Supplementary data used by SEED for determination of biomass reaction

Comp

ound

ID

Compound name Rea

cta

nt

Class Coefficient Inclusion criteria

cpd00

001

H2O YE

S

ENE

RGY

40 UNIVERSAL

cpd00

002

ATP YE

S

ENE

RGY

40 UNIVERSAL

cpd00

002

ATP YE

S

COF

ACT

OR

0.10/(TOTAL MASS OF ALL

COFACTOR COMPONENTS)

UNIVERSAL

cpd00

003

NAD+ YE

S

COF

ACT

OR



UNIVERSAL

cpd00

006

NADP+ YE

S

COF

ACT

OR



UNIVERSAL

cpd00

008

ADP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

008

ADP NO ENE

RGY

40 UNIVERSAL

cpd00

009

Orthophosphate YE

S

COF

ACT

OR



UNIVERSAL

cpd00

009

Orthophosphate NO ENE

RGY

40 UNIVERSAL

cpd00

010

CoA YE

S

COF

ACT

OR



UNIVERSAL

cpd00

012

Pyrophosphate YE

S

COF

ACT

OR



UNIVERSAL

cpd00 FAD YE COF 0.10/(TOTAL MASS OF ALL UNIVERSAL


1

015 S ACT

OR


cpd00

016

Pyridoxal phosphate YE

S

COF

ACT

OR



UNIVERSAL

cpd00

017

S-Adenosyl-L-methionine YE

S

COF

ACT

OR



UNIVERSAL

cpd00

018

AMP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

028

Heme YE

S

COF

ACT

OR



AND{SUBSYSTEM:Heme and Siroheme BiosynthesisÀ`B`F}

cpd00

030

Manganese YE

S

COF

ACT

OR



UNIVERSAL

cpd00

031

GDP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

034

Zinc YE

S

COF

ACT

OR



UNIVERSAL

cpd00

038

GTP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

042

Glutathione YE

S

COF

ACT

OR



OR{SUBSYSTEM:Glutathione: Biosynthesis and gamma-glutamyl

cycleÀ`B|SUBSYSTEM:Glutathione: Non-redox

reactionsÀ|SUBSYSTEM:Glutathione: Redox cycleÀ`B}

cpd00

046

CMP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

048

Sulfate YE

S

COF

ACT

OR



UNIVERSAL

cpd00 CTP YE COF 0.10/(TOTAL MASS OF ALL UNIVERSAL


2

052 S ACT

OR


cpd00

056

Thiamin diphosphate YE

S

COF

ACT

OR



AND{SUBSYSTEM:Thiamin biosynthesis}

cpd00

058

Copper2 YE

S

COF

ACT

OR



UNIVERSAL

cpd00

063

Calcium YE

S

COF

ACT

OR



UNIVERSAL

cpd00

067

H+ NO ENE

RGY

40 UNIVERSAL

cpd00

087

Tetrahydrofolate YE

S

COF

ACT

OR



AND{SUBSYSTEM:One-carbon metabolism by

tetrahydropterines|SUBSYSTEM:Folate Biosynthesis|!SUBSYSTEM:One-

carbon metabolism by tetrahydropterines`H}

cpd00

096

CDP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

099

Chloride YE

S

COF

ACT

OR



UNIVERSAL

cpd00

118

Putrescine YE

S

COF

ACT

OR



AND{SUBSYSTEM:Polyamine MetabolismÀ`B`C`DÈ`F`G}

cpd00

126

GMP YE

S

COF

ACT

OR



UNIVERSAL

cpd00

149

Cobalt YE

S

COF

ACT

OR



UNIVERSAL

cpd00

166

Cobamide coenzyme YE

S

COF

ACT

OR



AND{SUBSYSTEM:Coenzyme B12 biosynthesis}

cpd00

201

10-Formyltetrahydrofolate YE

S

COF

ACT






3

OR carbon metabolism by tetrahydropterines`H}

cpd00

205

Potassium YE

S

COF

ACT

OR



UNIVERSAL

cpd00

220

Riboflavin YE

S

COF

ACT

OR



UNIVERSAL

cpd00

254

Magnesium YE

S

COF

ACT

OR



UNIVERSAL

cpd00

264

Spermidine YE

S

COF

ACT

OR



AND{SUBSYSTEM:Polyamine Metabolism}

cpd00

345

5-Methyltetrahydrofolate YE

S

COF

ACT

OR





carbon metabolism by tetrahydropterines`H}

cpd00

557

Siroheme YE

S

COF

ACT

OR



AND{SUBSYSTEM:Heme and Siroheme BiosynthesisÀ`F}

cpd01

997

Dimethylbenzimidazole NO COF

ACT

OR

SUM OF COEFICIENTS

FOR(cpd00166)

AND{COMPOUND:cpd00166}

cpd02

229

Undecaprenyl diphosphate YE

S

CELL

WAL

L


CELL WALL COMPONENTS)

AND{!NAME:Mycoplasma|!NAME:Spiroplasma|!NAME:Ureaplasma|!NAM

E:phytoplasma}

cpd03

422

Cobinamide NO COF

ACT

OR

SUM OF COEFICIENTS

FOR(cpd00166)

AND{COMPOUND:cpd00166}

cpd10

515

Fe2+ YE

S

COF

ACT

OR



UNIVERSAL

cpd10

516

Fe3+ YE

S

COF

ACT

OR



UNIVERSAL

cpd11

416

Biomass NO BIO

MAS

1 UNIVERSAL


4

S

cpd11

459

tcam YE

S

CELL

WAL

L



AND{CLASS:Gram positive}

cpd11

461

DNA YE

S

DNA 0,025 UNIVERSAL

cpd11

462

mRNA YE

S

MRN

A

0,05 UNIVERSAL

cpd11

463

Protein YE

S

PRO

TEIN

0,5 UNIVERSAL

cpd11

493

Acyl-carrier protein YE

S

COF

ACT

OR



UNIVERSAL

cpd12

370

Apo-[acyl-carrier-protein] NO COF

ACT

OR

SUM OF COEFICIENTS

FOR(cpd11493)

UNIVERSAL

cpd15

352

2-Demethylmenaquinone 8 YE

S

COF

ACT

OR



AND{SUBSYSTEM:Menaquinone and Phylloquinone Biosynthesis}

cpd15

432

core oligosaccharide lipid A YE

S

CELL

WAL

L



AND{CLASS:Gram negative}

cpd15

500

Menaquinone 8 YE

S

COF

ACT

OR



AND{SUBSYSTEM:Menaquinone and Phylloquinone

Biosynthesis|ROLE:Ubiquinone/menaquinone biosynthesis methyltransferase

UbiE/COQ5 (EC 2.1.1.-)}

cpd15

533

phosphatidylethanolamine

dioctadecanoyl

YE

S

LIPI

DS


LIPID COMPONENTS)

AND{ROLE:Phosphatidylserine decarboxylase (EC

4.1.1.65)|SUBSYSTEM:Fatty Acid Biosynthesis FASII}

cpd15

540

Phosphatidylglycerol

dioctadecanoyl

YE

S

LIPI

DS


LIPID COMPONENTS)

AND{OR{ROLE:Phosphatidylglycerophosphatase B (EC

3.1.3.27)|ROLE:Phosphatidylglycerophosphatase A (EC

3.1.3.27)}|SUBSYSTEM:Fatty Acid Biosynthesis FASII}

cpd15

560

Ubiquinone-8 YE

S

COF

ACT

OR



AND{SUBSYSTEM:Ubiquinone Biosynthesis}

cpd15

665

Peptidoglycan polymer (n

subunits)

YE

S

CELL

WAL

L



AND{!NAME:Mycoplasma|!NAME:Spiroplasma|!NAME:Ureaplasma|!NAM

E:phytoplasma}


5

cpd15

666

Peptidoglycan polymer (n-1

subunits)

NO CELL

WAL

L

SUM OF COEFICIENTS

FOR(cpd15665,cpd15667,cpd15

668,cpd15669)

OR{COMPOUND:cpd15665|COMPOUND:cpd15667|COMPOUND:cpd1566

8|COMPOUND:cpd15669}

cpd15

667

glycerol teichoic acid (n=45),

linked, unsubstituted

YE

S

CELL

WAL

L



AND{CLASS:Gram positive|SUBSYSTEM:Fatty Acid Biosynthesis FASII}

cpd15

668


linked, D-ala substituted

YE

S

CELL

WAL

L




cpd15

669


linked, glucose substituted

YE

S

CELL

WAL

L




cpd15

695

Diisoheptadecanoylphosphatidylet

hanolamine

YE

S

LIPI

DS


LIPID COMPONENTS)



cpd15

696

Dianteisoheptadecanoylphosphatid

ylethanolamine

YE

S

LIPI

DS


LIPID COMPONENTS)



cpd15

722

Diisoheptadecanoylphosphatidylgl

ycerol

YE

S

LIPI

DS


LIPID COMPONENTS)




cpd15

723

Dianteisoheptadecanoylphosphatid

ylglycerol

YE

S

LIPI

DS


LIPID COMPONENTS)




cpd15

748

Stearoyllipoteichoic acid (n=24),

linked, unsubstituted

YE

S

CELL

WAL

L




cpd15

749

Isoheptadecanoyllipoteichoic acid

(n=24), linked, unsubstituted

YE

S

CELL

WAL

L




cpd15

750

Anteisoheptadecanoyllipoteichoic

acid (n=24), linked, unsubstituted

YE

S

CELL

WAL

L




cpd15

757


linked, glucose substituted

YE

S

CELL

WAL

L




cpd15

758


(n=24), linked, glucose substituted

YE

S

CELL

WAL





6

L

cpd15

759


acid (n=24), linked, glucose

substituted

YE

S

CELL

WAL

L




cpd15

766


linked, N-acetyl-D-glucosamine

YE

S

CELL

WAL

L




cpd15

767


(n=24), linked, N-acetyl-D-

glucosamine

YE

S

CELL

WAL

L




cpd15

768


acid (n=24), linked, N-acetyl-D-

glucosamine

YE

S

CELL

WAL

L




cpd15

775


linked, D-alanine substituted

YE

S

CELL

WAL

L




cpd15

776


(n=24), linked, D-alanine

substituted

YE

S

CELL

WAL

L




cpd15

777


acid (n=24), linked, D-alanine

substituted

YE

S

CELL

WAL

L




cpd15

793

Stearoylcardiolipin (B. subtilis) YE

S

LIPI

DS


LIPID COMPONENTS)

AND{ROLE:Cardiolipin synthetase (EC 2.7.8.-)|SUBSYSTEM:Fatty Acid

Biosynthesis FASII}

cpd15

794

Isoheptadecanoylcardiolipin (B.

subtilis)

YE

S

LIPI

DS


LIPID COMPONENTS)


Biosynthesis FASII}

cpd15

795

Anteisoheptadecanoylcardiolipin

(B. subtilis)

YE

S

LIPI

DS


LIPID COMPONENTS)


Biosynthesis FASII}

Documents

João Pedro Leonor Fernandes Saraivafrpinto/fr...João Pedro Leonor Fernandes Saraiva Reconstruction of a generic metabolic model for Streptococcus pneumoniae Dissertação de Mestrado