A Thesis Entitled Identification and Description of Burkholderia Pseudomallei Proteins That Bind Host Complement-Regulatory Prot

Home , Burkholderiales, Neisseria gonorrhoeae, OmpA domain, Proteobacteria, Yersinia enterocolitica

A Thesis

entitled

Identification and Description of Burkholderia pseudomallei Proteins that Bind Host

Complement-Regulatory Proteins via in silico and in vitro Analyses

Caroline Lambert

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the

Master’s Degree in

Biomedical Sciences

______R. Mark Wooten, Ph.D., Committee Chair

______Robert M. Blumenthal, Ph.D., Committee Member

______Jyl Matson, Ph.D., Committee Member

______Amanda Bryant-Friedrich, Dr. rer Nat., Dean College of Graduate Studies

The University of Toledo

August 2018

This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author. An Abstract of

Identification and Description of Burkholderia pseudomallei Proteins that Bind Host

Complement-Regulatory Proteins via in silico and in vitro Analyses

Caroline Lambert

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master’s Degree in Biomedical Sciences

The University of Toledo

August 2018

Burkholderia pseudomallei (Bp) is a motile Gram-negative bacillus and causative agent

of the febrile disease melioidosis. Bp is an opportunistic bacterium, with diabetes mellitus

as a major risk factor. Bp can evade clearance by the alternative pathway of the

complement cascade enabling it to invade cells and persist intracellularly. Factor H (fH)

is a negative regulator of the alternative pathway, which protects host surfaces from

complement-mediated damage. Several microbial species are known to mimic host

surfaces or deceive fH self-recognition domains by producing a fH binding protein

(fHbp). Bacteria with the ability to bind fH to their surface include Yersinia

enterocolitica, via adhesin YadA, Neisseria meningitidis via fHbp, and Haemophilus influenzae via P5.

This study used in silico and in vitro methods to investigate the ability of Bp to bind to host complement regulatory protein factor H. In vitro studies found that Bp can bind host complement regulatory protein fH on its surface via one or more proteins with a

iii

molecular weight of approximately 37 kDa. Candidate fHbps OmpA and Omp38 were

recognized by mass spectrometry analysis.

BLAST database searches identified OmpA and BpaC as candidate fHbps. Topological algorithms predicted BpaC and OmpA are partially extracellularly exposed on the bacterial surface. Rigid-body docking methods characterized conformations in which

OmpA and BpaC would interact with fH domains 19-20. Binding affinities between

BpaC and OmpA bound to fH domains 19-20 were predicted to be stronger than the interaction between known fHbp Borrelia burgdorferi OspE and fH domains 19-20.

A direct interaction between fH and the recombinant versions of candidate fHbps Omp38 and OmpA has not yet been confirmed using molecular biology methods. In vitro methods to investigate the BpaC and fH interaction are still to be explored. The identification of Bp proteins that bind to fH will provide a therapeutic target, which may have potential as vaccine candidates to be used towards reducing the global burden of melioidosis.

Acknowledgements

I would like to thank Dr Laura Stanbery and Irum Syed for all their time and effort spent trying to make me into a scientist, as well as John Presloid and Muhammed

Saad Moledina for their support in the lab. I would like to thank Dr R. Mark Wooten for

taking a bioinformaticist into his lab for the first time, Dr Robert Blumenthal for telling

me from the first moment I contacted him and throughout my studies that it was possible

for me to obtain a masters in bioinformatics and Dr Jyl Matson for her advice and

guidance in my research.

Table of Contents

Abstract ...... iii

Acknowledgements ...... v

Table of Contents ...... vi

List of Tables ...... x

List of Figures ...... xi

List of Abbreviations ...... xiv

1 Introduction

1.1 Burkholderia pseudomallei is causative agent of Melioidosis ...... 1

1.2 Melioidosis ...... 3

1.3 Melioidosis Distribution ...... 7

1.4 Melioidosis Risk Factors...... 11

1.5 Burkholderia pseudomallei as a possible Biological Weapon ...... 12

1.6 Burkholderia pseudomallei virulence ...... 13

1.7 Burkholderia pseudomallei and thailandensis ...... 16

1.8 Complement ...... 18

1.8.1 Classical pathway ...... 18

1.8.2 Lectin pathway ...... 19

1.8.3 Alternative pathway ...... 19

1.9 Importance of Complement in the control of Burkholderia

pseudomallei ...... 21

1.10 Factor H ...... 23

1.11 Bacteria that bind Factor H and their Factor H binding proteins ...... 26

1.12 Vaccine Potential ...... 32

1.13 Protein Structural Modeling ...... 34

1.14 Proteomic Software Assignment of Protein Characteristics ...... 36

1.15 Use of Bioinformatics to Identify Bacterial Proteins ...... 37

1.16 Thesis goals ...... 38

2 Materials and Methods ...... 41

2.1 In silico methods ...... 41

2.1.1 Strategies of Multiple Sequence Alignment ...... 41

2.1.2 Identifying putative fHbps in silico ...... 42

2.1.3 Logo Analysis ...... 44

2.1.4 Protein Localization and Topological Assessment ...... 44

2.1.5 Protein Structure Prediction ...... 46

2.1.6 Predicting Interacting Regions between Factor H and Factor H

binding proteins ...... 47

2.2 In vitro methods ...... 48

2.2.1 Bacterial strains, Media and Growth Conditions ...... 48

2.2.2 Whole bacteria Factor H Binding assay ...... 48

2.2.3 Outer membrane protein extraction ...... 49

2.2.4 Far-Western analysis of Factor H binding ...... 50

vii

2.2.5 Cloning of putative Factor H binding proteins into Escherichia

coli...... 50

2.2.6 Protein Expression ...... 52

2.2.7 Protein Extraction ...... 53

2.2.7.1 Soluble Protein Extraction ...... 53

2.2.7.2 Inclusion-Body Protein Extraction ...... 54

2.2.8 Affinity Purification of Recombinant protein ...... 54

2.2.9 Protein Immunoblot assay ...... 55

3 Results …………...... 57

3.1 In vitro results ...... 57

3.1.1 Burkholderia pseudomallei binds Factor H ...... 57

3.1.2 Identification of Burkholderia pseudomallei protein facilitating

Factor H binding via far western blot analysis ...... 58

3.1.3 Putative Factor H binding proteins identified via mass

spectrometry analysis ...... 59

3.1.4 Expression of putative Factor H binding proteins OmpA and

Omp38...... 60

3.1.5 Protein Purification of putative Factor H binding proteins OmpA

and Omp38 ...... 61

3.1.6 Purified OmpA and Omp38 do not bind to biotinylated Factor H

via an Immunoblot assay ...... 62

3.1.7 Summary ...... 63

3.2 In silico results ...... 64

viii

3.2.1 Known Factor H binding proteins do not interact via a Conserved

Domain ...... 64

3.2.2 Proteins found to be Homologous to known Factor H binding

proteins in the Burkholderia taxon ...... 65

3.2.3 Potential Factor H binding proteins predicted to be extracellularly

exposed on Burkholderia pseudomallei Outer Surface ...... 68

3.2.4 Tertiary Structure Prediction of candidate Factor H binding

proteins using RaptorX server ...... 70

3.2.5 Candidate Factor H binding proteins are predicted to bind to Factor

H via in silico analysis ...... 73

3.2.6 Summary ...... 77

4 Discussion…………...... 78

4.1 Conclusion ...... 95

References ...... 135

Appendix ...... 154

A Full logo analysis of fHbps possessing conserved domains among proteins across Burkholderia species in regions predicted to be extracellularly-exposed ...... 154

List of Tables

1 Search settings used in each BLAST analysis ...... 96

2 Search settings used in each PSI-BLAST analysis ...... 97

3 Strains used in this study...... 98

4 Sequences of primers used in this study ...... 99

5 Conditions used in Phusion PCR ...... 100

6 Parts included in the setup for an overnight ligation ...... 101

7 Colony PCR mastermix setup ...... 102

8 Four potential factor H binding proteins identified via mass spectrometry

analysis and taken on for further characterization ...... 103

9 Known factor H binding proteins identified in the literature ...... 104

10 B. pseudomallei candidate factor H binding proteins identified via

tBLASTn ...... 105

11 TMpred identifies transmembrane domains to be present in OmpA and

BpaC ...... 106

12 Binding affinities and dissociation constant values for factor H and interacting

proteins ...... 107

List of Figures

1.0 The main components of the complement cascade...... 108

2.1 Burkholderia pseudomallei whole cells bind to biotinylated factor H but

Streptococcus pyogenes cells do not...... 109

2.2 Comparison of Burkholderia pseudomallei and Streptococcus pyogenes emm8

factor H binding ...... 110

3.0 A candidate factor H binding protein was detected in Burkholderia thailandensis

and Burkholderia pseudomallei ...... 111

4.0 Putative factor H binding proteins visualized by Coomassie blue at 37 kDa...... 112

5.0 Expression of His-tagged Omp38 ...... 113

6.0 Expression of His-tagged OmpA ...... 114

7.0 SDS-PAGE of purified candidate factor H binding proteins Omp38 and

OmpA...... 115

8.0 Immunoblot showing candidate factor H binding proteins OmpA and Omp38 do

not bind biotinylated factor H...... 116

9.0 Known factor H binding proteins with sites required for factor H interaction. ...117

10.0 All-versus-all percent sequence identity matrix for all known factor H binding

proteins ...... 119

11.0 Logo analysis demonstrating Burkholderia proteins with conservation to known

factor H binding proteins...... 120 xi

11.1.1 Burkholderia elongation factor-thermo unstable proteins show conservation to

Pseudomonas aeruginosa elongation factor- thermo unstable...... 120

11.1.2 Two sample logo analysis demonstrating variation in elongation factor-thermo

unstable proteins ...... 122

11.2 Burkholderia OmpA proteins show conservation to Acinetobacter baumannii

OmpA...... 124

11.3 Cellular apoptosis susceptibility (CAS) and transporter proteins across

the Burkholderia genus show conservation to Leptospira interrogans Na-K

symporter ...... 125

11.4 Porin proteins across the Burkholderia genus show conservation to Neisseria

gonorrhoeae porin B...... 126

12.0 TMpred topological analysis identifies membrane-spanning domains ...... 127

13.1 Tertiary structure of B. pseudomallei OmpA ...... 129

13.2 Burkholderia pseudomallei OmpA predicted to possess a conserved OmpA-like

C-terminal domain ...... 130

14.1 Burkholderia pseudomallei BpaC tertiary structure...... 131

14.2 BpaC domain 2 tertiary structure prediction by RaptorX ...... 132

15.0 ZDOCK pose prediction of fH and OmpA interacting in silico...... 133

16.0 ZDOCK pose prediction of fH and BpaC domain two interacting in silico ...... 134

A1 Full logo analysis of OmpA proteins from Burkholderia group with Acinetobacter

baumannii OmpA...... 154

A2 Full logo analysis of Porin proteins from Burkholderia group with Neisseria

gonorrhoeae Porin B...... 156

xii

A3 Full logo analysis of CAS and transporter proteins from Burkholderia group with

Leptospira interrogans Na-K symporter ...... 158

A4 Two sample logo analysis full output elongation factor- thermo unstable proteins

in Burkholderia versus elongation factor-thermo unstable proteins which are

known to bind to factor H...... 162

xiii

List of Abbreviations

Ail Attachment invasion locus AP Alternative pathway

BLAST Basic local alignment search tool BLASTp Protein BLAST Bp Burkholderia pseudomallei Bsa Burkholderia secretion apparatus BSA Bovine serum albumin Bt Burkholderia thailandensis

CAS Cellular apoptosis susceptibility CCPs Complement control protein modules CDC Center for Disease Control and Protection CRASPs Complement regulatory-acquiring surface proteins C4BP C4 binding protein

DM Diabetes mellitus

Eap Extracellular adherence protein ECL Enhanced chemiluminescence EF-Tu Elongation factor- thermo unstable EST Expressed Sequenced Tagged fH Factor H fHbps Factor H binding proteins FHL-1 Factor H-like 1

GDT Global distance score

HRP Horse radish peroxidase

IPTG Isopropyl β-D-1-thiogalactopyranoside

LB Luria broth

xiv

MACs Membrane attack complexes MBL Mannose-binding lectin MgEGTA Mg2+-ethyleneglycoltetraacetic acid MotB Motility protein B MSA Multiple sequence alignment

NCBI National Center for Biotechnological Information Ni-NTA Nickel-chelating ligand nitrilotriacetic acid NspA Neisserial surface protein A

OD Optical density OMP Outer membrane protein OMVs Outer membrane vesicles OspE Outer surface protein E

RIPA Radioimmunoprecipitation assay RmpM Reduction‐modifiable protein M

PBS Phosphate buffered saline PCR Polymerase chain reaction PDB Protein Data Bank PRALINE PRofile ALIgNEment PRODIGY PROtein binDIng enerGY prediction PSI Percent sequence identity PSI-BLAST Position-Specific Iterative-BLAST PspC Pneumococcal surface protein C PVDF Polyvinylidene difluoride

Sbi Secreted second immunoglobulin-binding protein SCRs Short consensus repeats Sp Streptococcus pyogenes SVM Support vector machine

tBLASTn translated BLAST T-COFFEE Tree based Consistency Objective Function For AlignmEnt Evaluation THY Todd-Hewitt plus 0.5% yeast extract TSA Transcriptome Shotgun Assembly TSA Trypticase soy agar T3SS Type 3 secretion system T6SS Type 6 secretion system WGS Whole-Genome-Shotgun

US United States xv

Chapter 1

Introduction

1.1 Burkholderia pseudomallei is causative agent of melioidosis

Burkholderia pseudomallei (Bp) is a Gram-negative bacterium. It is vacuolated and

slender with a safety pin appearance due to bipolar staining (Cheng & Currie, 2005). The

bacterium is 2-5μm in length and 0.4-0.8μm in diameter, and moves through the use of

flagella (Haque, 2010). Captain Alfred Whitmore and his assistant C.S. Krishnaswami

were the first to isolate Bp, in 1912 (Samy et al., 2007). It is the causative agent of

melioidosis, a febrile illness with disease states including acute pneumonia and

septicemia (White, 2003). It is also known as the ‘remarkable imitator’ due to its

particularly extensive range of clinical manifestations, and melioidosis comes from the

Greek melis oid osis, which translates to distemper resemblance condition

(Krishnaswami, 1917).

Originally discovered through routine post mortem examination of morphine addicts from the streets of Rangoon, Burma, the disease presented pathological features similar to

glanders, an abscess-forming infection present in horses and caused by the bacterium

Burkholderia mallei. The newly discovered bacterium satisfied Koch’s postulates in that

it was seen and isolated from all disease cases and could be grown in pure culture. From

this culture it could then cause disease in a healthy organism and following this isolation

from the new host would show Bp to be the same as the originally inoculated pathogen

(White, 2003). The bacterium was named Burkholderia pseudomallei and is now known

to be a major cause of fatal pneumonia and sepsis in Thailand, Malaysia, Singapore and

northern Australia (White, 2003). Later, in 1913 a drastic distemper-like infection

became endemic at the animal facility of the Institute for Medical Research in Kuala

Lumpur, Malaya (now Malaysia), and the causative agent was identified as Bp (White,

2003). Additionally, a considerable number of western soldiers were exposed to

environmental Burkholderia pseudomallei during the military operations in Vietnam,

through contaminated wounds and burns or via inhalation (Dance, 1991). Thus,

melioidosis became recognized as an important, tropical, human infection.

Bp was originally classified under the genus Pseudomonas due to similarities in culture

requirements, morphological characteristics, and biochemical properties. However, it was

later assigned to a new genus Burkholderia (Yabuuchi et al., 1992). This group showed

homology in 16S rRNA sequencing and could oxidize and assimilate various

disaccharides and polyalcohols that Pseudomonas aeruginosa could not. The colonies for

Bp appeared to be smoother than other Burkholderia isolates analyzed. The cellular lipid

analysis and DNA-DNA homology results between the seven Burkholderia species organised them into their own proteobacterial genus.

The presence of Bp in the environment is most prevalent in soil and water. In endemic areas the bacteria commonly originates in cultivated fields such as rice paddies, still or stagnant waters, and moist soils present in sports fields (Brett & Woods, 2000). It has also been identified in some plant roots (Holden et al., 2004). Being in contact with soil can therefore increase the chance of contracting melioidosis.

1.2 Melioidosis

Melioidosis has become recognized as a significant human infection causing morbidity and mortality in Malaysia, Singapore and Northern Australia across the last few decades

(White, 2003). The disease has a highly variable incubation period with the longest

recorded being 62 years (Ngauy et al., 2005), regularly causing misdiagnosis in travelers

(Weissert et al., 2009). Melioidosis is frequently contracted via ingestion of Bp

(Limmathurotsakul et al., 2013) however; it has also been transmitted through inhalation,

nosocomial infections, laboratory accidents, vertical transmission at birth and by sexual

contact (Limmathurotsakul & Peacock, 2011). Exceptional circumstances have seen the

disease transmitted between two siblings living with cystic fibrosis (Silbermann et al.,

1997), and is believed to have been passed between a diabetic brother and sister

(Tiangpitayakorn et al., 1997). Furthermore, melioidosis in two infants was thought to

have been passed from mothers with mastitis caused by Bp (Walsh et al., 1995). Of all

the modes of transmission, inhalation causes the most severe clinical disease with

pulmonary melioidosis being the most common manifestation (Currie, 2003). The wide

range of incubation times and clinical manifestations makes melioidosis difficult to

diagnose in humans.

Studies have found melioidosis to affect all age groups, with a prospective study in

Australia comprising of patients ranging from 8 months to 91 years (Currie, Ward &

Cheng, 2010). It most frequently presents as a septicemic illness, often related to bacterial

dissemination which promotes the development of pneumonia as well as hepatic and

splenic abscesses. In approximately half of cases, bacteremia and pneumonia occur,

although not necessarily both at once (Limmathurotsakul & Peacock, 2011). A familiar

manifestation of melioidosis, occurring in around 15-25% of cases, is as superficial

pustules, subcutaneous abscesses and pyomyositis. These may be the primary sites of

infection or instead be secondary to haematogenous distribution (Limmathurotsakul &

Peacock, 2011). Numerous abscesses occur more commonly than a single abscess. Mice

infected with Bp developed extensively large and merging spleen abscesses (Samy et el.,

2017). On an ultra-sonogram abscesses will look similar to ‘Swiss cheese’ or

‘honeycomb’ on a computed tomography scan (Limmathurotsakul & Peacock, 2011).

Further manifestations include genitourinary infection, prostatic abscesses, renal

abscesses as well as brainstem encephalitis and flaccid paraparesis being seen in

neurological melioidosis. (Limmathurotsakul & Peacock, 2011). The numerous physical

appearances possibly seen across the body make it more challenging for medics to

understand the disease and correctly identify cases.

Typically, diagnosis of melioidosis is carried out by isolation and identification of Bp

from sputum, urine, tissue, blood samples and wounds (Samy et al., 2017). Alternatively,

bacteria-specific antibodies can be identified from a blood sample. Clinical diagnostic

tests include: agglutination, complement fixation and enzyme immunoassays. Quick

diagnosis of melioidosis from patient sputum, pus and urine is largely obtained by a

direct immunofluorescent antibody test (Limmathurotsakul & Peacock, 2011). Clinical

isolates have been seen to exhibit similar morphologies, and are antigenically

indistinguishable from environmental isolates (Pitt et al., 1992). In addition, Bp can be

identified by simple screening tests. These include bipolar staining where Gram-negative

rods present a representative safety pin appearance, and screening for resistance to

gentamicin and colistin (Dance et al., 1989). Inspection of Bp colony characteristics on

Ashdown’s selective agar presents wrinkled colonies (Howard & Inglis, 2003), whereas

on horse blood and MacConkey’s agars Bp produces smooth colonies (Rogul & Carr,

1972 and Gilmore et al., 2007). Presently, the diagnosis of Bp in the clinical laboratory is

highly problematic. Laboratories often lack appropriate media and Bp identification

systems. However, samples can be subjected to basic phenotypic and biochemical tests,

to indicate Bp’s presence or absence.

Once Bp infection has been identified, treatment of melioidosis has two phases. To begin

with, the acute/intensive phase is recommended to last at least 10-14 days. Secondly, the

eradication phase prevents disease recurrence or patient relapse. Antibiotics are the only

current treatment for melioidosis. Bp has been shown to be intrinsically resistant to many

antibiotics such as penicillin, ampicillin and streptomycin, polymyxin, as well as first and

second generation cephalosporins (Wiersinga et al., 2012). Bp possesses many

mechanisms for drug resistance, including chromosomally encoded genes to stimulate the

process of enzymatic inactivation, efflux from the cell and in rare cases target deletion

(Schweizer, 2012). Enzymatic inactivation can occur via modifications such as

acetylation or phosphorylation otherwise antibiotic cleavage takes place. Bp has been

documented to cleave β-lactam antibiotics by a class A β-lactamase, PenA, which is

conserved across Burkholderia species (Schweizer, 2012). Bacterial target deletion is rare

due to its negative effects on survival. This antibiotic resistance mechanism is more

commonly seen as altered target sites decreasing an antibiotic’s affinity to Bp.

Nonetheless, Bp has been seen to utilize target deletion as a ceftazidime-resistance

mechanism in some isolates (Schweizer, 2012). Intensive treatment across 10-14 days

with intravenous administration of a third-generation cephalosporin such as ceftazidime

or an oral prescription of trimethoprim-sulfamethoxazole for 3-6 months is needed to

eradicate the infection (Wiersinga et al., 2012). With melioidosis and its etiological agent

Bp being so difficult to diagnose and treat, especially in areas where it is endemic, it

would be appropriate to take a more preventative approach to reducing the disease

mortality, especially as increased antibiotic treatment has the potential to increase Bp’s

antibiotic resistance. The development of a vaccine would assist in this move towards

melioidosis prevention. Our lab is currently working towards identifying a vaccine

candidate through inhibiting one of Bp’s many survival strategies.

1.3 Melioidosis Distribution

Melioidosis is known to be endemic in 48 countries across the global temperate regions.

Cases have been seen in Asia, the Middle East, Africa, Latin America, the Caribbean and

the Pacific. Lack of medical facility accessibility and laboratory analysis in rural areas of

developing countries means melioidosis may be under-reported in many countries (Currie

& Kaestli, 2016). It is likely to be endemic in another 34 countries where no cases have

yet been confirmed (Limmathurotsakul et al., 2016). Total global cases for 2015 were

estimated at 165,000 with 89,000 deaths (Currie & Kaestli, 2016), which is comparable

to the global burden of measles and much higher than leptospirosis or dengue infection.

Thailand, Australia and Singapore have the highest number of melioidosis cases

worldwide. In northeast Thailand, melioidosis is the third most frequent cause of death

from infectious disease, with it first being reported in 1955 (Samy et al., 2017). It is also

the most common cause of community-acquired bacteremia (Suputtamongkol et al.,

1994). Melioidosis incidence in Thailand and neighboring countries appears to

correspond to the amount of Bp bacterial counts present in the soil (Limmathurotsakul &

Peacock, 2011), and the occurrence of natural disasters can increase the incidence of

melioidosis. This occurred after the 2004 tsunami and Typhoon-Haitang (Chierakul et al.,

2005). Australia is another location associated with high case-fatality rates of

melioidosis. Here it is an opportunistic infection which can develop transcutaneously by

inhalation. Cases are strongly associated with monsoonal rains and most commonly

presented as pneumonia (Smith, Hanson & Currie, 2018). Melioidosis is a growing public

health problem in Singapore (Pang et al., 2018) and India (Tellapragada et al., 2016). The

high case fatality case is due, in part, to a combination of increased awareness of the

disease and a steady rise in case detection rates. These areas are where the burden of

disease is greatest and thus the development of a vaccine would be most valuable across

these populations.

In Taiwan, melioidosis cases remained low in number. Only 20 cases were reported

between 1984 and 2000. In 2005 the Er-Ren River Basin in southern Taiwan was hit by a

typhoon. This natural disaster drastically increased the number of melioidosis cases.

After this incident, residents developed a greater number of Bp specific antibodies and

Bp was isolated from the soil (Su et al., 2007). Additionally, cases of melioidosis have been reported across Asia in Sri Lanka (Van Peenan et al., 1976) and Hong Kong (Tsang

& Lai, 2001). Across central and South America, infrequent cases of melioidosis have

been reported in Ecuador, Guadeloupe, Aruba and Brazil (Bandeira et al., 2012; Inglis et

al., 2006). Melioidosis is rare in the US. Five cases of locally-acquired melioidosis where

individuals have no occupational exposure to Bp or previous laboratory experience have

been reported. The sources of each of these cases could not be determined. Three of these

patients were diagnosed between 2010 and 2013, suggesting the possibility that

melioidosis could be rising in the US (Doker et al., 2014). All of this points to

melioidosis being more widespread than previously thought, suggesting the global

prevalence of melioidosis could be gradually increasing.

With global travel becoming more common in work and leisure, melioidosis is

increasingly being seen in travelers returning from endemic countries. Military personnel

serving in Vietnam have been known to contract acute and fatal melioidosis. Helicopter

take-off and landing was believed to have produced contaminated aerosols which lead to

melioidosis through inhalation (Chierakul et al., 2005). Individuals returning home from

travels to the Philippines (Lee et al., 1985), Bangladesh (Kibbler et al., 1991), Sylhet

(Hoque et al., 1999), Sierra Leone (Wall et al., 1985) and Kenya (Bremmelgaard et al.,

1982) have contracted melioidosis. Few human cases have been reported in Africa and,

of these, it is unclear as to where Bp was contracted (Dance, 2000). The cases of

melioidosis in Europe are most frequently seen in travelers. Only 49 individuals across

the continent were identified to be carrying Bp isolates last decade (Samy et al., 2017).

This encourages medics to remember that patients with melioidosis will not necessarily originate from endemic regions. Therefore, in suspected cases detailed accounts of travel

and exposure should be acquired. Production of a vaccine would mean travelers to

endemic countries would be protected and would thus prevent the introduction of the

disease into new populations.

Melioidosis has been seen outside of the human reservoir, in soil and in animals. Animals

imported from the temperate region of northern Australia aided in disease transmission.

The only case of melioidosis described in Africa since 1991 was that of a goat in

Transvaal, South Africa (Batchelor et al., 1994). The clinically similar disease of glanders, caused by Burkholderia mallei, was eliminated in the United States (US) from domestic animals in the 1940s (Samy et al., 2017). During the mid-1970s, a Bp-like bacterium was seen to be present in the soil and caused an Oklahoma farmer to contract a pelvic wound infection (McCormick et al., 1977). The organism was speculated to be a

Bp variant. Soil conditions in Florida are predicted to be suitable for Bp survival due to it sharing similar environmental values to the Caribbean islands and Taiwan, where melioidosis is known to be endemic. Japan also has areas considered suitable for Bp survival due to environmental similarities (Limmathurotsakul et al., 2016). The presence of Bp in Vietnamese soil was assessed through 407 samples from 147 paddy fields and 5 sites from the most important roads out of Ho Chi Minh City. Approximately 18% of these sites were positive for Bp (Parry et al., 1999). More environmental studies may be required in order to understand regions in non-endemic countries where people can become exposed to Bp. Increasing population and pathogen movement plays an important role in the frequent establishment of melioidosis in new areas. As the global distribution and burden expands, the need to raise the priority of this disease as a public health concern increases.

1.4 Melioidosis Risk Factors

The risk of developing melioidosis is dependent on various health and environmental

conditions. Immunocompromised individuals, especially the elderly, are at increased risk.

Examples include those affected by alcoholism, immunosuppressive treatments, chronic

renal failure, chronic liver disease, chronic lung disease (including cystic fibrosis),

thalassemia and kava consumption (Cheng & Currie, 2005). Diabetes mellitus (DM) is a

major risk factor for melioidosis (Limmathurotsakul et al., 2010). Almost 50% of

individuals who contract melioidosis have DM and show poor control of blood glucose prior to infection (White, 2003). Peak melioidosis incidence is shown between the ages of

40 and 60, which is also the time frame in which predisposing illnesses commonly develop. Interestingly, it has not emerged that the human immunodeficiency virus

predisposes an individual to melioidosis (Chierakul et al., 2004).

Individuals directly exposed to contaminated wet soil increase their risk of disease, as

these conditions allow Bp to survive as a free-living saprophyte (Samy et al., 2017). Its

spread in endemic areas is heightened during rainy seasons, where approximately 75% of

cases are seen (Wiersinga et al., 2012) and is characterized as a seasonal disease

(Limmathurotsakul & Peacock, 2011). The primary sufferers of melioidosis are rice

farmers. This also happens to be the most common occupation in Thailand, putting

millions of workers and their families at risk. Rice farmers with DM show a 6-9 fold

increase in risk of developing melioidosis than those who are nondiabetic or of another

profession. This data demonstrates the synergistic interaction between impaired immunity

in diabetic patients and the degree to which working in rice paddies exposes an individual

to Bp (Suputtamongkol et al., 1999). While individuals with thalassemia exhibit the highest risk of developing melioidosis compared to any other risk factor, the occurrence of thalassemia is far lower than that of DM, meaning the global diabetes pandemic could increase the number of fatalities caused by melioidosis. With this in mind, diabetic rice farmers may be deemed the most suitable target population for a melioidosis vaccine trial

(Suputtamongkol et al., 1999).

1.5 Burkholderia pseudomallei as a possible biological weapon

The use of microorganisms or toxins as weapons is no new exploit. History has seen the

use of cadavers to contaminate water supplies, through specially created munitions for

use on battlefields or undercover. Threats posed by biological warfare will continue to be

raised by propaganda or during conflicts. Previously, the Soviet biological weapons

program developed untreatable, antibiotic-resistant forms of Brucella to be used as a

biological weapon; however, this was dropped and replaced by Bp which was deemed to

be more potent (Pappas et al., 2006). Between 1915 and 1918, the closely related B.

mallei was utilized in biological warfare against animals during conflict. Horses were

pierced with contaminated needles or fed sugar cubes containing ampoules of bacteria

(Wheelis, 1998).

This led to Bp being categorized as a potential biological warfare or bioterrorism agent.

The Center for Disease Control and Protection (CDC) and the U.S. Federal Select Agent

Program has Bp and B. mallei currently listed as Tier 1 select agents, putting them in the highest priority category along with 12 other bacterial and viral agents. This is because it can spread easily, has high morbidity rates in some regions along with low mortality rates, and requires heightened disease observation (OPHRP, 2017). This is the highest biosecurity level certified, meaning that the disease agent is considered to be a remarkable threat to public safety (Zimmerman et al., 2017). This categorization is further supported by observations that Bp is easily aerosolized in laboratory studies, demonstrates mortality rates in endemic regions of 40%, is intrinsically resistant to antimicrobial agents, and there is currently no vaccine for its associated disease melioidosis (Currie & Kaestli, 2016). The endemic boundaries of melioidosis and the environmental presence of Bp may be accelerated by global warming (Currie, 2003). This highlights the ever increasing need to develop a vaccine.

1.6 Burkholderia pseudomallei virulence

A crucial feature in the pathogenesis of Bp is its ability to invade and persist intracellularly in phagocytic, as well as non-phagocytic cells (Jones et al., 1996). Bp’s ability to survive in a vast range of environments is due to the numerous genes it expresses as virulence mechanisms and survival strategies. Virulence mechanisms include flagella (Chua, Chan & Gan, 2003), type 3 secretion system (T3SS) (Stevens et al., 2002), type 6 secretion system (T6SS) (Shalom et al., 2007), and a polysaccharide capsule; other relevant virulence factors include polysaccharides, exoproteins, fimbriae,

pili, and putative adhesins. Survival strategies include secondary metabolite pathways, catabolic pathways, transport systems and stress response proteins (Holden et al., 2004).

The phagocytic cell internalization of Bp involves the Burkholderia secretion apparatus

(Bsa) T3SS. BsaQ and BsaU are structural components of the T3SS which cooperate with the secretion apparatus BsaZ. The growth of closely related, avirulent bacterium

Burkholderia thailandensis (Bt) in L-arabinose containing media negatively regulates the

Bsa T3SS (Moore et al., 2004). This has suggested that its possession of an arabinose assimilation operon which Bp is lacking is an antivirulence gene. The translocator protein

BipD and the recognized effector BopA both play roles in membrane disruption, so Bp can escape from the endocytic vesicle into the cytoplasm to then replicate and move by actin-based motility. Furthermore, BopA facilitates Bp’s evasion of autophagy. The non- phagocytic cell uptake of Bp necessitates PilA and adhesins BoaA and BoaB (Allwood et al., 2011). Bp has been described as escaping the endosome or phagosome of host cells in less than 15 min after cell entry and continues to survive within the cytoplasm (Harley et al., 1998). It can spread cell-to-cell by using host actin to move through neighboring cell membranes, leading to giant cell formation. Bp uses both methods to disseminate with little, if any, contact with the extracellular environment (Allwood et al., 2011). The combined functioning of various effector proteins is necessary for Bp to successfully evade host cellular functions and consequently for its bacterial pathogenesis.

Bp is able to modify the surfaces of infected cells to initiate cell-to-cell fusion and form multi-nucleated giant cells as an intracellular survival strategy. Bp is unique in that it has been seen in tissue culture models of infection to stimulate host cell fusion to promote formation of multinucleated giant cells (Kespichayawattana et al., 2000). In Bp, the ability to form multi-nucleated giant cells has been associated with RpoS, BipB, T6SS-1 and the putative toxin-encoding genes Bpsl0590 and Bpsl0591 (Allwood et al., 2011).

The process of autophagocytosis is another mechanism by which Bp has been tailored to survive. Autophagy engulfs intracellular bacteria, such as Bp, into double-membrane vesicles known as autophagosomes. Their fusion with lysosomes causes the degradation of Bp. However, through the expression of T3SS protein BopA, the saprophytic bacteria can disrupt the autophagic process and persist within host cells (Cullinane et al., 2008).

Polyubiquitination is a post-translational modification which can alter protein functional properties, so it becomes a target for lysosomal degradation by the autophagic pathway.

Bp is one of many pathogens to regulate host ubiquitination to its own advantage. The

T6SS protein TssM is accountable for the down-regulation of host inflammatory responses. It does this by interfering with the ubiquitination of transitional proteins in the activation pathway of NF-κB (Allwood et al., 2011). Additionally, like many Gram- negative bacteria Bp forms Type 4 pili, which play a role in the adhesion of bacterial species to surfaces on cells and tissues (Pelicic, 2008). Bp differs from other pathogens in its approach to actin-based motility (Allwood et al., 2011). Bp is dependent upon the auto-secreted protein known as BimA (Stevens et al., 2005) and is required for bacterial

cell to cell spread. BimA is conserved across natural populations of Bp, Bt and B. mallei species in isolates from melioidosis endemic areas (Sitthidet et al., 2008).

Nitric oxide is an antimicrobial molecule generated in macrophages which is imperative in the process of intracellular pathogen clearance. The production of nitric oxide occurs as a product of inducible nitric oxide synthase which is stimulated by cytokines IFN-γ,

IFN-β, TNFα, IL-1 and IL-2, as well as lipopolysaccharide and lipoteichoic acid (Fang,

1997; Jacobs & Ignarro, 2001). Macrophages infected with Bp do not activate inducible nitric oxide synthase expression, partly because these macrophages are unable to produce

IFN-γ and IFN-β. This enhances the survival of intracellular Bp (Utaisincharoen et al.,

2003, 2004). Bp’s loss of an arabinose assimilation operon, along with the gaining of a gene cluster involved in producing a capsule of polysaccharides, may have been critical for its success as a human pathogen (Haraga et al., 2008).

1.7 Burkholderia pseudomallei and B. thailandensis

Burkholderia thailandensis was previously described as an avirulent Bp-like strain which can assimilate arabinose due to an eight-gene operon on Chromosome 2, which is absent on Bp (Brett, Deshazer & Woods, 1997). In Bp, this operon region has been replaced by a two-protein cluster containing one hypothetical protein and one MarR family regulatory protein. Bt categorization as a unique species has been confirmed by 16S rRNA gene sequencing and DNA-DNA hybridization studies (Yabuuchi et al., 2000). In October

2013, a 67-year-old man in Chongqing, China, was diagnosed as infected by Bp. This

was because the positive blood culture contained many Gram-negative rod-shaped bacteria. However, the bacterial biochemical profile and ability to assimilate arabinose identified the isolated strain as Bt (Chang et al., 2017).

In animal models, Bt virulence is significantly less than that of Bp, with a mean LD50 of

109 CFU/mouse for the former, compared to 182 CFU/ mouse for the latter (Smith et al.,

1997). Furthermore, Smith et al. (1997) isolated samples from 1,200 melioidosis patients among which no Bt isolates were characterized. Both bacteria possess two circular chromosomes, shown in Bt reference strain E264 to be 3.8Mb and 2.9Mb in size, as well as being GC-rich at approximately 68% (Yu et al., 2006). The virulent bacterium Bp secretes substances for proteolytic, siderophore and low in vitro cytotoxic activities, whereas less virulent Bt could not (Brett, Deshazer & Woods, 1997). Although these two phenotypically similar bacteria inhabit overlapping niches (Garcia, 2017), they demonstrate substantial differences in exozyme production, hamster virulence and 16S rRNA gene sequences. Although, Bt shares most virulence factors with Bp and the two have a high nucleotide identity at the DNA level (Kim et al., 2005). Bt infection in humans is poorly understood and the bacteria are generally considered avirulent, as instances of disease are rare (Chang et al., 2017). Consequently, it is often used for comparative analyses as a relatively avirulent model organism (Garcia, 2017).

1.8 Complement

Complement is a major non-cellular system of innate immunity in humans, which seems

to have evolved in non-vertebrates prior to the appearance of rearranging genes (Carroll,

1998). This system is the first defense line of innate immunity against pathogenic

microorganisms. Numerous regulatory proteins, present as cell bound or fluid phase,

prevent excessive activation of complement. These proteins protect the host from attack

by its own complement system and include soluble plasma proteins, complement 4b-

binding protein (C4BP), factor H (fH) and several membrane proteins (Prasadarao et al.,

2002). Cell-bound regulators operate to prevent inadvertent or misdirected complement

activation including DAF, CR1, CD59 and MCP (Pangburn, 2000). Complement can be

activated via three different pathways: the classical, lectin or alternative pathway (AP),

all of which are involved in innate defense.

1.8.1 Classical pathway

The classical pathway is predominantly initiated by surface-bound antibodies or direct binding taking place with C1q, the first pathway component. Activation occurs through six steps. Firstly, C1 binds to an activating surface by C1q. Ligand binding follows this which causes conformational changes in C1q. Next C1r cleaves C1s, producing a completely activated C1 protein. The activated C1s cleaves C4 to produce C4b, which is released to attach to an activating surface. C4b then binds to C2 to produce a proconvertase. Subsequently, C4b2 becomes converted to C4b2b by C1s cleavage of C2 and release of C2a (Garcia et al., 2016). Bacteria can down-regulate this pathway through

C4b-binding proteins. Two examples are the extracellular adherence protein (Eap) from

S. aureus (Woehl et al., 2014) and OmpA from E. coli (Prasadarao et al., 2002), which

allow the bacteria to effectively avoid initial host defense by serum complement. Eap and

OmpA inhibit the classical and lectin pathways by means of binding to C4b, which

blocks C4b binding to C2 and its C2b fragment, preventing C3 convertase production

(Woehl et al., 2014; Prasadarao et al., 2002).

1.8.2 Lectin pathway

Activation of the lectin pathway occurs via mannose-binding lectin (MBL) or filcolins.

These detect foreign ligands or changes in self-surfaces by distinct pattern-recognition

molecules (Kjaer et al., 2015). MBL-associated serine proteases communicate the presence of ligand and surface changes by cleaving subsequent complement factors. The

MBL-associated serine proteases complex is similar structurally and functionally to the

C1 recognition molecule of the classical pathway (Roos et al., 2003). The MBL-

associated serine proteases recognition complex catalyzes C4 and C2 cleavage, leading to

C3 convertase C4b2b formation (Matsushita & Fujita, 1992).

1.8.3 Alternative pathway

The alternative pathway (AP) recognizes and protects host cells and tissues. The AP is

always active, and lacks complement regulation allowing the pathway to become

amplified on foreign surfaces (Pangburn & Müller-Eberhard, 1984). It is stimulated at a

low frequency in plasma, as C3 is continuously hydrolyzed to a form referred to as

C3(H2O) or C3b and forms fluid-phase C3 convertase. The spontaneous hydrolysis of C3

to C3b simultaneously produces C3a, which has a range of complement sytem effects.

This hydrolysis activates the AP through covalently bonding to hydroxyl or amine groups

present on unprotected surfaces (Meri et al., 2013). The deposition of C3b is usually

limited to microbial surfaces, which then goes on to stimulate further complement

activation (de Córdoba et al., 2004). Six plasma proteins make up the AP, including C3,

factors B, D, H, I and P (Pangburn, 2000). C3(H2O) can bind factor B, which C3 cannot,

to generate the initial C3 convertase enzyme after cleavage of B to Bb by factor D.

C3bBb establishes the basis of AP amplification by cleaving additional C3 molecules to

C3b successively developing into subunits of new C3 cleaving enzymes. Once the activator surface is covered in C3b molecules, opsonophagocytosis occurs through propagation of the cascade, which releases chemotactic and anaphylatoxic peptides leading to the formation of membrane attack complexes (Meri et al., 2013).

Bacteria, fungi, viruses, virus-infected cells, tumor cells and parasites can activate the

AP, providing the broadest specificity against infectious agents or tumor development.

This pathway operates throughout adulthood (Pangburn, 2000). Activation occurs within

5-20 minutes, where full activation can be achieved by one microorganism. This

demonstrates the extreme sensitivity of the AP. The pathway requires no prior

immunization or antibody production, which characterizes it as an innate defense system

(Pangburn, 2000). The effectiveness of complement as an innate defense mechanism

requires preventing wasteful consumption of its components and halting non-specific

damage to host tissues.

1.9 Importance of complement in control of B. pseudomallei

Innate immune responses play a fundamental role in confining Bp infections. These become important during the development of acute disease, where melioidosis often becomes lethal in the first few days of infection and before the adaptive responses have been established. Thus, it is important that activation of the complement system occurs in response to Bp infection and allow generation of components and opsonins that support macrophages and neutrophils in phagocytosis and killing. In the absence of complement activation, uptake of Bt is greater than that of Bp by both macrophages and neutrophils,

but neither species can be cleared intracellularly (Woodman, Worth & Wooten, 2012;

Mulye et al., 2004). C3 deposition does occur on the surfaces of Bp and Bt, however Bp is more resistant to this than Bt, which is largely due to the Bp protective polysaccharide capsule (Mulye et al., 2004). Bp-specific antibodies do not directly promote phagocyte uptake or killing of Bp, but do significantly increase complement deposition on Bp and

Bt. Although bacterial uptake can be enhanced, complement deposition on Bp and Bt surfaces cannot stimulate intracellular killing of either strain by macrophages (Woodman,

Worth & Wooten, 2012). Importantly, while neutrophils can phagocytose unopsonized

Bp, complement opsonization is essential to further enhance Bp uptake and elicit efficient intracellular killing.

For in vitro analyses using low serum levels of ≤5%, classical or lectin pathway

components are required to promote complement deposition on both Bp and Bt.

Following incubation of Bp and Bt in ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-

tetraacetic acid (EGTA), which chelates Ca++ but not Mg++, and normal human serum

(20%), C3 was still deposited on the surface of Bt whereas deposition was not seen on

Bp. Because EGTA inhibits activation of the classical and lectin pathways, this indicates

that Bp is resistant to C3 deposition via the AP (Woodman, Worth & Wooten, 2012).

This suggests that manipulation of the complement AP could enhance the amount of C3

deposited on Bp to achieve levels that allow killing. We want to promote a mechanism

which will increase complement deposition on the surface of Bp. This may be possible

through disrupting its point of interaction with host regulatory proteins in the complement

cascade, thus promoting C3 convertase production on pathogen surfaces to enable MAC

formation, opsonization and clearance of Bp.

Incubation of Bp and Bt in ≥5% serum allows complement deposition on their surfaces to reach a minimum threshold which, though only enhancing bacterial uptake by macrophages, promotes both uptake by and activation of neutrophils (Mulye et al., 2004).

This simultaneously triggers an appropriate respiratory burst in neutrophils, which promotes intracellular killing, even for encapsulated bacteria. Often the capsule is not resistant to complement deposition itself, but can bind to soluble complement regulatory proteins of the host (e.g., fH) through surface exposed sialic acid or glycosaminoglycans.

This binding has been seen on the surface of several species which express capsular

polysaccharides such as A. baumannii (Kim et al., 2009), N. meningitidis (Madico et al.,

2006) and H. influenzae type b (Shapiro et al., 1989). The binding of host complement regulatory proteins can still take place on the surface of Bp, in a manner similar to host cells, in order to prevent complement deposition on their own surfaces.

1.10 Factor H

The main regulator of the complement alternative pathway is soluble glycoprotein Factor

H (fH). A single gene, HF1, codes for fH. This encompasses 23 exons and spans over 94

kb of genomic DNA (Male et al., 2000). The fH gene family includes six secreted plasma

proteins: fH, fH-like-1 and fH-related proteins 1 through 4. These are predominantly

synthesized in the liver and characterized by conservation between individual domains.

Each of these multidomain multifunctional proteins exhibit overlapping functions during

their regulation of the AP. Genes of the fH family are linked within the gene cluster of

complement regulators in human chromosome 1q32 (Zipfel et al., 1999).

The fH gene encodes for a protein 150kDa in size, which consists of 20 domains known

as short consensus repeats (SCRs) or complement control protein modules (CCPs). Each

module contains approximately 60 amino acids (Ripoche et al., 1998) and contains highly

conserved residues including four cysteines, two prolines, one tryptophan, along with

several moderately conserved glycines and hydrophobic residues. The SCRs have a

globular structure with six-stranded antiparallel β-sheets connected with loops and turns

(Perkins et al., 1988; Barlow et al., 1992, 1993). The 150 kDa molecular weight has

been determined by SDS gel and a meniscus-depletion sedimentation-equilibrium method

described by Yphantis, 1964. FH has on occasions been seen at molecular weights greater

than 150 kDa, which may be due to its flexible structure which reflects ‘beads-on-a-

string’, and can lay down in multiple arrangements such as folding back on itself

(DiScipio, 1992). FH is present in plasma at a concentration of approximately 500μg/ml,

with C3 as the only complement component to be more abundant at 1200μg/ml

(Pangburn, 2000). Uncontrolled activation of the AP takes place when fH levels in the

plasma become too low (Thompson & Winterborn, 1981).

Factor H like-1 (FHL-1) protein, also known as reconectin, is a 43 kDa protein that

encompasses the first seven SCRs of fH and has an additional extension of four amino

acid residues at its C-terminus (Zipfel et al., 1999). FHL-1 is a second gene product of

the fH gene, HF1 (Schwaeble et al., 1987). Numerous fH-related proteins are

homologous to domains within fH and FHL-1, but are not transcribed from HF1. Distinct

functions of fH and FHL-1 have been suggested by stimulus-dependent differential

regulation of both proteins. FHL-1 replicates the functions of fH possessing the same

factor I cofactor activity and C3b binding.

The protein exhibits cell-spreading actions and can bind the N-terminus of the

streptococcal M protein (Zipfel et al., 1999). FHL-1 may be a novel form of fH expressing the translational product of an additional H-derived mRNA. It is a truncated form of fH which could possibly mediate C3b binding, cofactor activity for I, B cell and

monocyte stimulation more effectively (Schwaeble et al., 1987). Both proteins act as central checkpoints for AP activation, and partake in the binding of C3b and accelerating the decay of C3bBb complex formation (Zipfel & Skerka, 2015). FHL-1 was recently shown to attach to proinflammatory, monomeric and circulating pentameric versions of

C-reactive protein, as well as to the long pentraxin-3 protein (Swinkels et al., 2018).

These findings indicate FHL-1 possesses functions in addition to which mimic fH.

The SCRs across fH, fHL-1 and FH-related proteins possess specific binding sites which can uniquely interact with host and foreign elements. The binding of fH to C3b blocks its ability to interact with factor B and acts as a cofactor in the factor-I mediated cleavage of

C3b. This behavior inhibits C3 activation in the fluid phase and on host cell surfaces to which C3 can bind if they contain polyanions (Meri & Pangburn, 1990). FH can bind

C3b at two different binding sites, SCR1-4 and SCR19-20 (Schmidt et al., 2008). N- terminal SCR 4 possesses decay-accelerating activity for the AP C3/C5 convertase, which simultaneously serves as a cofactor site for serine protease factor I, which inactivates C3b (Alsenz et al., 1984). Specific fH binding sites for polyanions are found in SCRs 7, 9-15 and 19-20. Heparin and sialic acid are polyanions present on glycoproteins and lipids whose role involves regulating the affinity of fH for complement protein C3b. The strength of this affinity determines the course taken by the complement cascade (Meri & Pangburn, 1990). This interaction allows protection of the host’s cellular surfaces from complement activation and increases C3b affinity to fH (Pangburn,

Atkinson & Meri, 1991). Host surfaces can then be differentiated from foreign particles

which activate the AP. Many pathogens have evolved to exploit this protective mechanism and persist undetected in the host (Blackmore et al., 1998; Hallström et al.,

2008; Hovis et al., 2004; Biedzka-Sarek et al., 2008). One example is the interaction between fH and S. pyogenes protein M. This causes reduced C3b deposition on the streptococcal surface, which in turn leads to decreased phagocytosis by macrophages and polymorphonuclear cells (Horstmann et al., 1988). Microbes can mimic host ligands or surfaces, therefore misdirecting fH self-recognition domains so that fH inappropriately binds to one or several surface proteins belonging to the invading pathogen (Schneider et al., 2009). This partnering allows microorganisms to be protected from complement evasion and is possibly one survival strategy that Bp employs.

1.11 Bacteria that bind factor H and their factor H binding proteins

Escaping complement and its spontaneous AP amplification cascade is a requirement for microbial virulence in nearly all pathogens, since this is initiated on any non-protected surfaces (Meri et al., 2013). Microbial binding to host complement regulatory proteins has been seen in the plasma and other body fluids. FH is commonly recruited to the surface of foreign elements to avoid AP activation via two binding sites; domains 6-7 and

C-terminal domains 19-20. Binding via domains 19-20 has been shown to allow formation of a tripartite microbial protein : factor H : C3b complex. This complex enhances fH-mediated inactivation of C3b, explaining why many microbes have convergently evolved to utilize this site. Three fH-binding proteins (fHbps) have been identified to bind overlapping sites on fH domain 20. Binding of a microbial protein to

domain 20 of fH allows the C3d part of C3b to bind to domain 19. FH becomes nearer to its main target, C3b, and promotes complement inhibition. Most fHbps characterized are long molecules, such as streptococcal M protein, or have a flexible tail which allows twisting and tilting of a molecule, such as OspE from B. burgdorferi. These microbial proteins appear to operate on a broad area opposed to a specific site (Meri et al., 2013), and they possess little sequence similarity. Therefore, it is expected that all currently known and unknown fHbps use slightly different residues to bind fH and form hydrogen bonds or hydrophobic contacts (Meri et al., 2013). The list of pathogens demonstrated to bind to fH is continuously growing. Our understanding of fH and its role in each bacterial species is diverse. Some pathogens have only been demonstrated to bind fH binding to their cell outer surface, whereas other species have a known fHbp which has been structure confirmed, and its points of contact with fH identified.

Different microbial species which have been described in current literature to demonstrate binding to fH are outlined below. The extent of understanding for each interaction is also described. Gram-negative bacteria species Francisella tularensis and

Bordetella pertussis have both been shown to recruit human fH to their surfaces, however the ligands for binding remain to be discovered (Nasr & Klimpel, 2008; Amdahl et al.,

2011). B. pertussis is the causative agent of whooping cough in humans. Direct binding assays indicated Bordetella’s main binding site on fH to be in the C-terminal end, a common site for microbes, and a weaker binding site is present at the 5-7 SCR region

(Amdahl et al., 2011).

Blackmore and coworkers (1998) showed group A streptococci, a Gram-positive bacterial species responsible for a range of invasive and noninvasive infections, to bind fH and FHL-1 domains 6-7 by virulence factor protein M. The M6 protein is one of more than 100 serotypes, reflecting the presence of a hypervariable region. In the M6 protein the fH binding site has been localized to the central conserved C-repeat region (Fischetti et al., 1995). This known site of interaction on protein M occurs as a repeated domain and provides a starting point for identification of a possible conserved domain in Bp to facilitate fH binding. Additionally, the ability to bind to complement regulatory protein fH has been shown across several spirochetal bacterial species. The three main genospecies of Borrelia burgdorferi (sensu stricto, garinii and afzelii), some isolates in the Borrelia species hermsii and parkeri, as well as unrelated Leptospira interrogans use fH binding as one mechanism to evade the AP of complement. This enables B. burgdorferi to persist for prolonged periods in the host where it presents as Lyme borreliosis. This fH attachment occurs via OspE and four other complement regulatory- acquiring surface proteins (CRASPs) which also interact with FHL-1 (Hellwage et al.,

2001). For B. hermsii and B. parkeri, cell-bound fH has been confirmed to participate in the factor I-mediated cleavage of C3b (McDowell et al., 2003). The fHbp present in these species is known as FhbA and presents no significant sequence similarity to other known fHbps or other proteins encoded by the B. burgdorferi genome (Hovis et al., 2004). The unrelated spirochete species Leptospira interrogans produces at least two fHbps (Verma et al., 2006). The LfhA gene codes for Na+-K+ symporter 26kDa membrane protein

which can also interact with FHR-1, but not FHL-1, implying that Na-K must bind to

residues in fH SCRs 18-20. Ligand affinity blot procedures displayed another fHbp of

approximately 50kDa in the L. interrogans outer membrane (Verma et al., 2006). This

shows that the mechanism used by bacteria to bind to fH is not constrained to

Proteobacteria, and has been acquired by other phyla such as the Spirochaetes. We can

also see that pathogens can express several proteins to bind fH on their surface rather

than simply relying on one protein for this mechanism.

Gram-negative opportunistic pathogens Haemophilus influenza and Acinetobacter

baumannii both bind fH to their outer surface, and evade the AP of complement using an

OmpA family protein (Hallström et al., 2008; Kim et al., 2009). The 38kDa OmpA fHbp

in A. baumannii is also considered a CRASP. A. baumannii has three CRASP outer

membrane proteins (OMPs) with molecular sizes of 38, 32 and 24 kDa. Its ability to bind

fH and evade complement was shown during in vivo experiments. In normal human

serum, A. baumannii was killed (Kim et al., 2009). The fHbp in H. influenzae is also

known as P5 (Langereis et al., 2014). For H. influenzae P5, fH SCRs 1-6 and 18-20 play important roles in binding to P5 via its extracellularly exposed loops 1 and 2. The exact amino acids which bind to fH within loops 1 and 2 are not yet known (Langereis et al.,

2014). The location of fH binding to A. baumannii has not yet been determined, but

conserved domains to H. influenza fHbp OmpA may allow predictions for extracellularly

exposed sites available to bind to fH. Surprisingly, the translation elongation factor

protein (EF-Tu) of opportunistic pathogen Pseudomonas aeruginosa is partially

extracellularly exposed. This facilitates fH binding to the surface of the aeruginosa

species to inhibit host complement activity (Kunert et al., 2007).

The Gram-negative food-borne pathogen Yersinia enterocolitica binds fH via two OMPs

known as YadA and Ail (Biedzka-Sarek et al., 2008). YadA protein is a homotrimeric

autotransporter which possesses monomers of approximately 44 kDa. These create a

“lollipop” structure exposed on Y. enterocolitica’s surface (Zaleska et al., 1985), with an

N-terminal head domain, a coiled-coil stalk and a C-terminal membrane anchor (Hoiczyk

et al., 2000). Ail is translated as a 17 kDa protein, expected to possess eight outer

membrane spanning amphipathic strands and four short extracellular loops (Miller &

Falkow, 1988; Miller et al., 2001). Ail specifically binds to fH domains 6-7, whereas

YadA, presented an innovative fH binding pattern which encompasses the entire fH

molecule (Biedzka-Sarek et al., 2008). Analogous affinity blotting experiments using

purified fH determined the direct interaction with YadA and illustrated that additional

complement proteins or serum factors are not required (Biedzka-Sarek et al., 2008).

However, having Ail should increase the pathogen’s chances of survival in the body.

A further two Gram-positive bacteria species able to bind fH are Streptococcus

pneumonia and Staphylococcus aureus. The former evades innate immune detection by a

choline-binding protein known as pneumococcal surface protein C (PspC) (Dave et al.,

2001) and the latter via the secreted second immunoglobulin-binding protein (Sbi)

(Haupt et al., 2008). Sbi is a potent complement inhibitor which is able to bind to fH

related-1, C3, IgG, the processed forms of C3b and C3d, and β2-glycoprotein. By finding

a means to inhibit fH binding to any of the bacteria discussed a therapeutic target to

protect against development of the associated disease then becomes possible.

The Gram-negative bacterial species of Neisseria have been shown to possess fHbps

(Meri et al., 2013), these being the pathogens N. gonorrhoeae and N. meningitidis. The

latter is an encapsulated Gram-negative bacterium which causes meningitis and sepsis.

Meningococci bind to fH via fHbp, otherwise known as GNA1870 or lipoprotein 2086.

This is a 28 kDa surface exposed lipoprotein found in all examined strains of N.

meningitidis (Madico et al., 2006). To further classify GNA1870 strains, proteins have been divided into subfamilies A and B. Isolates from subfamily B have been shown to

express higher levels of fHbp than isolates from subfamily A (McNeil et al., 2018). All

invasive isolates of meningococcal express fHbp, demonstrating how important this

mechanism is for virulence (Pajon et al., 2011). N. meningitidis has a second fHbp, the

Neisserial surface protein A (NspA) which carries out low levels of fH binding in strains where the Neisserial fHbp gene is deleted. NspA is a meningococcal vaccine candidate which plays a role in complement evasion (Lewis et al., 2010). Additionally, N. gonorrhoeae, the causative agent of gonorrhea, binds fH domains 18-20 via the porin molecules PorB.1A and PorB.1B (Ram et al., 1998, Shaughnessy et al., 2009). These recognized fHbps are organized into different structural orientations. N. meningitidis

fHbp consists of 2 domains, of which the C-terminal is very similar to the E. coli OmpA

barrel structure, whereas N. gonorrhoeae consists of three β-barrel structural domains.

This demonstrates the diversity between proteins which are able to facilitate fH binding,

even within a single species. For meningitis, a vaccine has already been produced which

contains a capsular polysaccharide of H. influenzae b coupled to an OMP-complex from

N. meningitidis which acts as a vaccine delivery vector (Giebink et al., 1993). In Bp, the use of dendritic cells in combination with CpG oligodeoxynucleotides have been proposed as vaccine delivery vectors to provide a protective immune response against

heterologous strains (Elvin et al., 2006). Identification of an OMP in Bp that facilitates

fH binding could be a good candidate for coupling to dendritic cells.

1.12 Vaccine potential

Melioidosis is ranked the third most common cause of death by infectious disease in

northeastern Thailand (Limmathurotsakul et al., 2010). Annual incidence rates are 21.0 per 100, 000 with similar numbers of 19.6 per 100, 000 being seen in northern Australia

(Limmathurotsakul et al., 2010). With these conditions, various vaccine strategies have

been pursued in an attempt to control the disease. Bp provides obstacles against the

development of a vaccine in that there are many phenotypically diverse strains,

multifaceted disease presentations, and acute and chronic manifestations. A range of

putative vaccine models have included the use of live-attenuated, whole-cell killed,

subunit, plasmid DNA and dendritic cell types. All being carried out in mouse models

with sterile immunity rarely being reported (Peacock et al., 2012).

Live-attenuated vaccines were shown to induce protection against infection introduced by

the same route as immunization (Peacock et al., 2012). This method should provide long-

term protection to individuals receiving the vaccine (Vidor, 2010), however there are

concerns that an attenuated mutant might develop into a latent infection. Therefore,

confidence that an attenuated strain will not revert back to virulence is needed (Peacock

et al., 2012). This risk is removed by the use of naked DNA and dendritic cell vaccines,

which have shown promising immune responses in mice (Liu, 2011). Vaccines encoding

the Bp flagellar subunit gene filC have been shown to provide mice with modest levels of

protection (Chen et al., 2006). However, subunit vaccines are the ideal target for most

vaccine creations. Protective proteins and polysaccharides in Bp include

lipopolysaccharide, capsular polysaccharide, LolC proteins which form part of lipoprotein export system, Omp85, and Hcp2 a surface associated portion of the T6SS

(Peacock et al., 2012). The immunization of mice with naturally obtained outer membrane vesicles (OMVs), containing a mixture of subunits, provides adjuvanticity, immunogenicity and effective protection against a low-level inhalation challenge (Nieves

et al., 2011). Use of OMV based vaccines may be beneficial over long-established

vaccine methods. One rationale for this is that OMVs cannot grow and survive whereas inactivated or live-attenuated strains may be able to. Bp OMVs have low levels of toxicity and contain numerous protein antigens. This feature could provide an individual with protection against more than one heterogenous strain of Bp (Nieves et al., 2011).

The N. meningitidis hexavalent class 1 OMP-containing OMV vaccine contained multiple OMPs from wild-type strains of meningococcal disease to target existing

epidemics (Claassen et al., 1996). Serogroup B strains of N. meningitidis display large

variations in their extracellularly exposed loops (Rappuoli, 2000); therefore, numerous

OMPs need to be present in the vaccine to provide sterile immunity. Previously, two Bp

OmpA-family proteins have shown immunogenic potential in a mouse model (Hara,

Mohamed & Nathan, 2009). Immunization with either OMP only provided a 50%

protection rate against melioidosis. Furthermore, surviving mice were seen to have

significant splenomegaly and multiple abscess formation, as only partial bacterial

clearance was observed. Both these OmpA-family proteins had been identified as novel

antigens through bioinformatic methods. The structure and functional potential of

candidate therapeutic targets such as these can be better understood using in silico

methods. From this previous knowledge, the production of a Bp OMV vaccine with one

or more OMPs warrants further attention as a melioidosis vaccine development.

1.13 Protein Structural Modeling

The vaccine targeting the N. meningitidis OMP fHbp is one of only two fHbps to have

been structurally confirmed. Both this and B. burgdorferi OspE have been solved whilst

interacting with fH (Schneider et al., 2009; Bhattacharjee et al., 2013). Each protein has

also been structurally confirmed when unbound to a ligand or receptor (Cendron et al.,

2011; Bhattacharjee et al., 2013). The sequence similarity between amino acid sequences across species is used to identify genes. The availability of the full Bp genome sequence

means this can be analyzed for the presence of genes homologous to those in other

bacterial species. Protein homologues can be explored using the BLASTp server

(Altschul et al., 1997). A query search will examine a protein of interest against a relevant genome or database. Searching against the Protein Data Bank (PDB) database allows known structures solved by x-ray crystallography or nuclear magnetic resonance spectroscopy to be identified. This is the first step in homology-modeling methods when predicting the 3D assembly for a protein of unknown structure. Excellent predicted structure quality is expected for aligned regions of more than 60 amino acids in length or if a query structure possesses more than 50% identity to a confirmed structure (Pevsner,

2015).

For cases when a protein lacks similarity to solved structures, but shows similar folding characteristics to unrelated proteins, then modeling by threading or fold comparison can be used. In these analyses, sub-fragments of the protein sequence are compared to a database containing all known protein folds. This method is based on the concept that there may only be 1000-2000 distinct folds in nature (Pevsner, 2015). For proteins with no related structures or folding similarities, a third method, ab initio modeling, is available. This firstly assumes that an amino acid sequence contains all the information about a protein structural assembly. Secondly, it assumes that a globular protein will always fold into the structure with the lowest free energy because nature favors the shape with the lowest energy state. While each of these modeling methods may be useful, homology modeling is the most reliable when quality templates are available. The fold by which OspE interacts with fH is the first of its kind to have been identified within a microbial protein involved in extracellular or immune evasion (Bhattacharjee et al.,

2013). Alternatively, N. meningitidis fHbp is assembled from two unrelated barrel domains connected by a short loop (Cendron et al., 2011). The N. meningitidis protein’s

N-terminal shows weak similarity to the protein streptavidin, but its C-terminus shows similarities to porins such as the NspA antigen, which carries out some residual fH binding in N. meningitidis (Lewis et al., 2010). The current lack of structurally-confirmed fHbps combined with the lack of similarity between the two confirmed fHbp structures already suggests the modelling of candidate fHbps across bacterial species via homology modelling methods may be a difficult task.

1.14 Proteomic Software Assignment of Protein Characteristics

Proteomic software can assign protein characteristics based on an amino acid sequence.

Servers such as SignalP can be used to identify proteins which possess secretory signal peptides (Gomez et al., 2000). TMpred allows a protein transmembrane domain to be identified through comparing a known protein sequence to a database of protein segments previously reported to span a membrane (Gomez et al., 2000). Potential diagnostic, drug and vaccine targets can be detected through assessment of where the protein may be subcellularly localized by using programs such as Psortb. Hypotheses can then be devised and investigated with regards to the functions, genome and proteome of the query protein. This computational tool uses a support vector machine (SVM) kernel-learning algorithm (Gardy et al., 2004). All data are mapped as vectors in n-dimensional feature space. A support vector machine locates the optimal separating hyperplane dividing two classes of data so they are a maximum distance away from the hyperplane. Frequently

seen motifs appearing in protein groups can signify the position of a common

biochemical mechanism or a structural domain (Gardy et al., 2004). The known fHbps

mentioned previously can be compared through use of a SVM algorithm, which will

identify if a common site or system is being utilized in this group of proteins.

1.15 Use of bioinformatics to identify bacterial proteins

DNA-sequencing technology has enabled accessibility to full genome sequences, and

computer-aided software allows comparative sequence analysis between species and

within isolates, which can uncover proteins most likely to be vaccine candidates. These

methods facilitate the discovery of novel proteins with unknown functions which may

pose conservation or homology to proteins expressing a desired characteristic to a known

protein of interest. Novel proteins can also be identified through analysis of unassembled

DNA fragment methods (Pizza et al., 2000). Candidate genes can be expressed as recombinant proteins or as DNA vaccines, and thus tested for protective immunity

(Rappuoli, 2000). Identification and analysis was used on an array of Bp OmpA-family proteins as potential vaccine candidates to provide protective immunity against melioidosis development (Hara et al., 2009). The conserved C-terminal domain of

OmpA-family proteins was confirmed through multiple sequence alignment (MSA) and

used as a starting point to search for homologous proteins in the Bp genome using the

BLASTp database. Bp BimA, an essential bacterial factor for actin tail formation, was

also identified through bioinformatic analysis of the complete Bp genome sequence. A conserved motif from autosecreted proteins of Gram-negative bacteria was submitted to

PSI-BLAST where BimA was identified and subsequently characterized in vitro (Stevens

et al., 2005).

Vaccine candidates in N. meningitidis were also identified using bioinformatic methods

(Pizza et al., 2000). Searches began with BLASTX to distinguish two classes of DNA

segments with possible coding regions. Proteins were screened in accordance to cell

localization and then studied for conservation against N. meningitidis serogroup A and N.

gonorrhoeae partial genome sequences. Two database examples are Pfam and FASTA.

Pfam consists of an extensive collection of protein domains and families (Bateman et al.,

2004). FASTA was devised to identify protein sequences from a common ancestor

(Pearson, 1990). Our known fHbps can be submitted to databases such as Pfam and

FASTA to investigate if evolution from a common ancestor has taken place and to investigate what protein families are involved in the binding of fH to pathogen surfaces.

Consistencies between proteins can provide a starting point for analysis of open reading

frames or unassembled DNA fragments in target genomes such as B. pseudomallei used

in this study.

1.16 Thesis goals

Based on this introduction, we can see that the ability for virulent strains to evade

complement deposition on its surface is critical for survival of virulent strains. We want to identify the mechanism by which Bp facilitates evasion of complement deposition, which subsequently allows evasion of intracellular killing. We hypothesize that Bp will

possess one or more proteins that can bind host fH to its surface. Our goal is to identify and determine proteins which contribute to Bp’s mechanism of intracellular survival by avoiding complement deposition.

In this study, I used in silico approaches to identify potential Bp proteins capable of binding to host fH through computational analysis methods. This involved recognizing regions of homology and structural characteristics present in pathogens already known to use the complement evasion strategy of fH binding, and identifying these commonalities in Bp proteins. A whole bacterial cell-binding assay showed Bp to bind host fH. Several candidate fHbps were identified through bioinformatics and in vivo methods. These were tested for their ability to inhibit complement deposition through fH-binding.

Chapter 2

Materials and Methods

2.1 In silico Materials and Methods

2.1.1 Strategies of Multiple Sequence Alignment

Literature-confirmed fHbps were identified and simultaneously submitted to the multiple

sequence alignment program “Tree based Consistency Objective Function For

AlignmEnt Evaluation” (T-COFFEE) (Notredame, Higgins & Heringa, 2000). This is a

consistency-based approach, which incorporates evidence from multiple sequence

alignments to create a pairwise alignment (Pevsner, 2015). T-COFFEE performs a

percent sequence identity (PSI) matrix. A pairwise alignment is carried out between all

possible combinations of submitted sequences to calculate the identity score between the

two. Sequence gaps and mismatches are included in the evaluations. A matrix is

produced where the assigned numerical value for each pair of sequences represents the

quantity of similar amino acids. Protein sequences were also submitted to PRofile

ALIgNEment (PRALINE) (Simossis & Heringa, 2005). This program produces alignments through iterations which use homology-extended multiple alignment and

allows integration of predicted secondary structure information.

2.1.2 Identifying putative fHbps in silico

A number of methods were used to identify potential fHbps via in silico analysis. For

each of these searches, we began by going to the National Center for Biotechnological

Information (NCBI) https://www.ncbi.nlm.nih.gov/ and utilizing the basic local

alignment search tool (BLAST) (Altschul et al., 1997) for potential fHbps in Bp. Firstly,

known fHbps identified from the literature were assessed for homology to one another

through use of a PSI matrix. A fHbp from each matrix cluster of pairwise alignments possessing PSI scores higher than 70% was chosen as a search seed. Each selected fHbp

was input into a standard protein BLAST (BLASTp) search against the Burkholderia

group (taxid:119060) to search the nucleotide collection database and identify proteins

possessing homology.

Secondly, the BLAST programs are used more elaborately to identify potential fHbps.

Known fHbps were acquired from the literature to be used as a starting point for

identifying potential homologous proteins in the Bp genome. Each literature-cited fHbp was input for a translated BLAST (tBLASTn) search against the Burkholderiales order

(taxid:80840). For each protein, a search was carried out with the same settings listed but in three different databases. The first was Whole-Genome-Shotgun (WGS) contigs, which excludes completed chromosomes associated with the WGS projects. Second was the Database of GenBank + EMBL + DDBJ sequence from Expressed Sequenced Tagged

(EST) Divisions, which consists of cDNA sequences. Third was the Transcriptome

Shotgun Assembly (TSA) sequences database, an archive of computationally-assembled

mRNA sequences from primary data such as EST and raw sequence reads. The tBLASTn

search translates every DNA sequence in a database into all six potential reading frames.

Each of these were then compared to the query protein. Proteins in the Burkholderiales

order, with sequence identity above a threshold defined by the BLAST settings used,

were output as results. These proteins were used as search seeds in a BLASTp search

through the Bp 1026b genome to identify any proteins with sequence similarity or

conserved domains to the tBLASTn output proteins. Proteins with the highest percent

identity scores and the lowest e-values were then considered for further in silico analysis

to assess their potential to bind to fH.

A third method to identify proteins in the Bp genome utilized Position-Specific Iterative-

BLAST (PSI-BLAST). A position-specific scoring matrix or profile from a multiple

sequence alignment is detected and used to identify distantly related target proteins

(Bhagwat & Aravind, 2007). The same known fHbps as used in method 2 were

individually submitted to PSI-BLAST. The program carries out a series of iterations, the

first being identical to BLASTp. Output proteins with e-values below 0.005 are to be listed first, and those with an e-value larger than the threshold 0.005, but greater than that selected on the query page are listed second. Protein hits with e-value thresholds below

0.005 are used to produce a profile. This is used in subsequent PSI-BLAST iterations, which are repeated until a desired convergence score is reached and the top protein hits no longer change with new iteration submissions. For each PSI-BLAST search, the parameters outlined in Table 2 were used. Each protein was submitted for a search

against the non-redundant protein sequences database and the Transcriptome Shotgun

Assembly protein database.

2.1.3 Logo Analyses

Output proteins from the BLASTp search against the Burkholderia group (taxid:119060)

were submitted alongside the original fHbp search seed to a MSA using T-COFFEE. This

MSA was then entered into WebLogo (Crooks et al., 2004) to determine the extent of

conservation in the aligned sequence sets. A two-sample Logo (Vacic, Iakoucheva, &

Radivojac, 2006) was carried out to identify variation between a positive sample of

proteins known or hypothesized to bind to fH, and a negative sample of related proteins

not known to bind to fH. Each set of proteins was aligned prior to submission using T-

COFFEE.

2.1.4 Protein Localization and Topological Assessment

Numerous programs were used to predict protein cellular localizations, including Cello

v2.5, which utilizes multi-layered support vector machines (Yu, Lin & Hwang, 2004);

Gneg-PLoc, which clusters Swiss-Prot proteins whose subcellular localizations have been

described using Gene Ontology terms using the K-nearest neighbour algorithm (Chou &

Shen, 2006); and PSORTb v3.0 (Yu et al., 2010), which is described below. The presence of secretion signal sequences in proteins was investigated using SignalP (Petersen et al.,

2011).

PSORTb produces prediction results for cytoplasmic, inner membrane, periplasmic, outer

membrane and extracellular localizations in Gram-negative bacteria (Yu et al., 2010). An

overall prediction site for a query protein is made through applying six analytical modules. The analyses and results from each module are combined using a Bayesian

Network. If upon analysis the query sequence possesses no features to be reliably identified into a single localization, then the results can be classified as ‘unknown’.

PSORTb homology analysis is carried out using the subcellular localizations-BLAST module, where localization-specific motifs are identified using PROSITE motif-based analysis and transmembrane alpha helices are detected using HMMTOP (Tusnady &

Simon, 1998). The further three modules carry out a novel OMP motif analysis, a search for type II secretion signals, and one is a variation on the protein subcellular prediction

tool SubLoc which analyses the protein subcellular localization annotation database for

similarities between the query protein and proteins with known localizations.

Each protein output from the three BLAST methods used with an e-value below one had its amino acid sequence submitted to the PSORTb v3.0.2 online server to receive an

output predicting the most likely cellular location. Protein sequences predicted to be

localized to the outer membrane were then input into the transmembrane region and

orientation prediction program TMpred (Hofmann & Stoffel, 1993). This program uses

an algorithm which is dependent upon the statistical analysis of TMbase, a database of

transmembrane proteins and their helical spanning domains. The method aims to identify

the preferred assembly of individual amino acids for a query sequence. The scoring of

several weight matrices allows the identification of putative transmembrane domains, their orientation and the intracellular or extracellular domains connecting them.

Additional topology programs were used to analyse each predicted fHbp sequence for the presence of extracellularly exposed domains needed to facilitate fH binding. These included HMMTOP, which employs a hidden Markov model with architecture allowing exploration of transmembrane topologies corresponding to maximum likelihood in comparison to all predicted topologies of a query protein (Tusnady & Simon, 1998);

Phobius, a program which simultaneously predicts protein topology and signal peptide presence through use of a hidden Markov model (Käll, Krogh & Sonnhammer, 2004), and Predict Protein, an internet server that performs protein sequence analysis, as well as structure and function prediction (Rost, Yachdav & Liu, 2004).

2.1.5 Protein structure prediction

As a starting point, the existence of proteins with structures previously solved by x-ray crystallography or nuclear magnetic resonance that may be homologous to the query protein were identified by carrying out a BLASTp search, using the settings in table 2, through the PDB database. The protein of interest was then submitted to the Raptor X protein structure prediction server (Källberg et al., 2012). RaptorX begins by identifying the non-redundant homolog sequences available to construct the query protein sequence.

Conditional random fields provide a framework for building probabilistic models. The model is applied and predicts numerous structural outcomes dependent on an alternative

observation (Lafferty, McCallum & Pereira, 2001). This is followed by use of a nonlinear threading score function based on a probabilistic graphical model (Peng & Xu, 2009). A multi-template procedure then uses the homologous sequences combined to create a single target sequence, and produce a final model prediction of the secondary and tertiary structures (Peng & Xu, 2011).

2.1.6 Predicting Interaction Regions between factor H and factor H binding

proteins

The solution structure of a previously crystalized structure was downloaded from the

PDB database. The RaptorX server provided a predicted structure for a protein of interest. The two structures would then be input into the rigid-body protein-protein docking server ZDOCK (Pierce et al., 2014). The larger protein, which is to remain fixed in an interaction, was submitted as the receptor protein and the solved PDB structure submitted as the ligand protein. Next residues to be blocked or exposed were selected in accordance to topology program predictions. ZDOCK uses the Fast Fourier Transform algorithm (Pierce et al., 2014). This allows global docking of two proteins to be carried out on a 3-D grid. Scoring of a docked structure is assessed by shape complementarity, electrostatics and statistical potential terms.

The crucial element of the target function is a desolvation term founded on the atomic contact energy (Zhang et al., 1997). ZDOCK calculates and outputs a structure for the 10 most likely interaction poses between two protein structures. Each of these poses are

downloaded as a .pdb file and submitted to the PROtein binDIng enerGY prediction

(PRODIGY) webserver (Vangone & Bonvin, 2015). PRODIGY uses the network of

interfacial contacts in a protein- protein complex submitted to predict the binding affinity.

The protein structures are to be labeled as separate chains on submission: interactor 1 and

interactor 2. The temperature for the predicted environment was set at 37⁰C because the

computer simulated interaction was to mimic the conditions of the human body due to interactions between fHbps and fH normally being carried out in vivo.

2.2 In vitro Materials and Methods

2.2.1 Bacteria strains, Media and Growth Conditions

The bacterial strains used are listed in table 1. E. coli strains were grown on Luria broth

(LB) agar plates, Bp strains were grown on trypticase soy agar (TSA) with adenine

plates, Bt strains were grown on TSA plates, and S. pyogenes were grown on Todd-

Hewitt plus 0.5% yeast extract (THY) plates. All bacterial strains were stored in their

respective media listed in Table 3 along with 20% glycerol at -80⁰C.

2.2.2 Whole-bacteria factor H binding assay

Bacteria were washed three times in phosphate buffered saline (PBS), after which 108

CFU was incubated with biotinylated-fH (200 ng) in a total volume of 50 μl at 37⁰C for

30 mins. Cell-associated and free protein were separated by centrifugation (10,000 x g, 3 min) through 20% sucrose in PBS. Samples were resolved on a Bolt 4-12% Bis-Tris Plus

SDS-PAGE gel (Invitrogen, Thermo Fisher scientific) and the proteins were transferred

to a PDVF membrane (Thermo Scientific). The membrane was blocked in 5% milk (in

PBS plus 0.05% Tween-20; 1:2000 dilution) prior to western blotting (primary –

streptavidin – horseradish peroxidase (HRP) conjugate (Southern Biotech), 5 mg/ml. The

membrane was visualized using enhanced chemiluminescence (ECL) and imaged on the

Syngene G:Box Chemi-XX6 imaging system v1.6.1.0. Quantification was performed

using ImageJ, where numerical values for fH band densities were assigned in arbitrary units.

Experiments were performed in triplicate. Each of the 3 membranes were exposed to 3 exposures of different lengths. The image taken at each exposure is then quantified using

ImageJ on 3 separate occasions. For a single assay, 27 quantifications are accumulated for each sample and the overall average band intensity is calculated from these values.

The percent of fH bound to an organism was calculated by dividing the overall average band intensity of an organism’s pellet by the total band intensity, calculated for the sample organism to acquire the percent of total fH in each sample.

2.2.3 Outer Membrane Protein extraction

The rapid OMP extraction protocol was modified from Carlone et al 1986. Bacterial cells were washed in 10mM HEPES buffer (pH 7.4) and lysed via sonication (Q125 sonicator,

Qsonica) at 50% amplitude for 5 min, with 30 pulses and 30 rest periods. The resulting solution was centrifuged at 13,400 x g for 2 min at 4°C to separate the unbroken cells and debris from the supernatant. The supernatant was transferred to a fresh microcentrifuge

tube and cell membranes were sedimented with a 30 min centrifugation at 13,400 x g and

4°C. Following this, the cell membrane pellets were resuspended and incubated in 200l

HEPES and 200l 2% Sarkosyl at room temperature for 30 minutes with intermittent mixing to solubilize the inner membrane fraction. A final 30 min centrifugation at 4C sedimented the OMPs.

2.2.4 Far-western analysis of factor H binding

OMP fractions isolated through use of the OMP extraction protocol mentioned above, were separated using a Bolt 4-12% Bis-Tris Plus SDS-PAGE gel followed by transfer to polyvinylidene difluoride (PVDF) for western blot analysis. Membrane was blocked in

5% milk (in PBS plus 0.05% Tween-20). After blocking, the membranes were probed using biotinylated fH overnight at 4°C, followed by probing with streptavidin-HRP antibody (5mg/ml) using 5% milk (in PBS plus 0.05% Tween-20) for 2 at room temperature. The membrane was then washed in PBS plus 0.05% Tween-20 and the presence of a fHbp was visualized using ECL reagents and imaged using Syngene Gbox

Chemi-XX6 v1.6.1.0.

2.2.5 Cloning of putative factor H binding proteins into E. coli cells

Template DNA was isolated from Bp82 colonies grown on TSA plus adenine plates, then resuspended in diH2O and boiled for 10 min to cause cell lysis. Polymerase chain reaction (PCR) was carried out using 2 μl of isolated Bp DNA. Primers (IDT), dNTPs

(Thermo Fisher) as well as Phusion taq and buffer were purchased from New England

(NE) Biolabs. The PCR protocol and PCR thermocycler conditions used are listed in

Table 5.

The amplified Phusion PCR product was resolved on a 1% agarose gel stained with ethidium bromide to confirm that the PCR product was the correct size. DNA fragments were then purified using QIAquick PCR purification kit (Qiagen) following manufacturer’s instructions.

Restriction digestion of PCR product and pPROEX HTb (gift from Jason Huntley) or pBAD-18KAN (gift from Jyl Matson) vector were prepared separately in buffer 3.1 (NE

Biolabs), along with restriction enzymes PstI and EcoRI. Each sample was incubated at

37⁰C for 6 hours and purified using QIAquick PCR purification kit (Qiagen) following manufacturer’s instructions. The PCR product was ligated overnight with desired plasmid construct and carried out in a 3:1 v/v insert DNA to plasmid ratio (Table 6).

Ligated PCR product and plasmid were transformed into E. coli DH5α cells, and a 50 μl sample of cells with 5 μl of ligated DNA was kept on ice for 30 min. The sample underwent heat shock treatment at 42⁰C for 30 s followed by 2 min on ice. Following this, 500 μl of liquid LB was added to the transformed cells and this solution was incubated for an hour at 37⁰C before being spread onto LB-kanamycin plates. Plates were incubated overnight at 37⁰C. Colony PCR was carried out as described in Table 7 to confirm the integration of the PCR product into the plasmid using vector specific primers.

These primers were pPRO forward and pPRO reverse or pBAD forward and pBAD

reverse as listed in Table 4.

PCR samples were separated on a 1% agarose gel stained with ethidium bromide.

Samples which contained an insert at the size of interest were expanded into an overnight

liquid LB culture followed by plasmid isolation using QIA prep-spin mini prep kit following manufacturer’s instructions. Isolated plasmid was then sequenced to confirm the presence of the inserted Bp82 DNA (Eurofins).

2.2.6 Protein expression

To observe expression of the recombinant protein, a culture containing LB and bacterial

cells was incubated at 37°C with shaking overnight. A 200ml LB flask containing

kanamycin (50mg/ml) or ampicillin (100mg/ml), was inoculated with the overnight

culture (1:100 dilution). The inoculated flask was incubated with shaking at 37°C.

Optical density (OD) of the flask culture was recorded every 30-45 min until an OD

between 0.4-0.6 was reached, then a 1ml sample of these uninduced cells was transferred

to a microcentrifuge tube. Inducing agent arabinose (0.2% concentration) or Isopropyl β-

D-1-thiogalactopyranoside (IPTG) (0.6mM concentration) was added to the remaining

culture to induce protein expression. The uninduced cells were centrifuged for 1 min at

13,400 x g to pellet the cells. The supernatant was then removed, and cells were

suspended in 16 μl of 6x SDS-PAGE loading dye and 84 μl of PBS for a final volume of

100 l. Induced cells continued to be grown with shaking at 37°C for 4 h and a 1 ml

sample was taken every hour after induction. Each sample was centrifuged and resuspended in PBS and 6 x SDS-PAGE loading dye.

Following induction, samples were electrophoresed on a 4-12% BOLT Bis-Tris plus

SDS-PAGE gel. The gel was transferred onto a PDVF membrane and blocked in 5% milk

(in PBS plus 0.05% Tween-20). Protein expression was assessed by incubating in primary antibody containing mouse anti-6x-histidine (5mg/ml) (Southern Biotech) in 5% milk (in PBS plus 0.05% Tween-20) for 1-2 h, followed by goat anti-mouse HRP- conjugate secondary antibody (1mg/ml) (Southern Biotech) in 5% milk for a further 1-2 hours. Visualization of expression was detected by ECL and imaged using Syngene Gbox

Chemi-XX6 v1.6.1.0, which calculates automated image exposure times.

2.2.7 Protein extraction

2.2.7.1 Soluble protein extraction

The bacterial pellet obtained from 50 ml of expressed culture was suspended in 30 ml of soluble buffer (10mM Tris, 500mM NaCl, 10mM imidazole, pH8.0) and vortexed. An

EDTA-free Roche Protease Inhibitor tablet (Thermo scientific) was added to each pellet resuspension then vortexed. Solution was sonicated in an ice bath at 50% amplitude for 5 min, with 30 s pulses and 30 s rest periods. The suspension was then centrifuged at 8,000 x g for 20 min at 15°C. The soluble supernatant was then discarded and the pellet stored at -80°C.

2.2.7.2 Inclusion body protein extraction

The pellet obtained from soluble protein extraction was resuspended in 30 ml of inclusion

body buffer (8M Urea, 10mM Tris, 200mM NaCl, 10mM imidazole, pH7.8) and

vortexed. The suspension was sonicated in an ice bath at 50% amplitude for 5 min with

30 s pulses and 30 s cool-downs. The solution was rotated for 2-3 h before centrifuging at

8,000 x g for 20 min at 20°C to pellet insoluble material. The pellet was then discarded and the solubilized inclusion body was saved for column purification.

2.2.8 Affinity purification of recombinant protein

His-tagged proteins were purified into a soluble inclusion body suspension. Nickel-

chelating ligand nitrilotriacetic acid (Ni-NTA) agarose beads (Thermo scientific) were

resuspended in storage buffer and 1-1.5 ml of the Ni-NTA suspension was added to a

nickel charged column (Bio-Rad). This was followed by two ~5 ml washes of inclusion

body buffer. Inclusion body solubilized supernatant (10 ml) was repeatedly added in

alternation with inclusion body buffer (10 ml) until all the supernatant had been added.

The column was then washed with a minimum of 5 volumes of inclusion body buffer

with an increased imidazole concentration (8 M Urea, 10 mM Tris, 200 mM NaCl, 200 mM imidazole, pH 8.0). Protein was eluted from the column with ~10 ml of elution buffer (8 M Urea, 10 mM Tris, 200 mM NaCl, 200 mM imidazole, pH 8.0). Ten 1 ml protein fractions were collected and electrophoresed on a Bolt 4-12% Bis-Tris Plus SDS-

PAGE gel. Fractions with protein expression were inserted into a Slide-A-Lyzer 7k

Dialysis Cassette (Pierce) using an 18 gauge needle. The cassette was placed into PBS to

allow overnight dialysis to remove proteins less than 10 kDa, and imidazole and urea.

Purified protein then stored at -80°C for further use.

2.2.9 Protein Immunoblot assay

Streptavidin beads (Pierce), were resuspended in storage buffer, diH2O containing 0.05%

NaN3. A sample of 300 μl were added to a microcentrifuge tube and placed into a magnetic stand. As the beads collected against the side of the tube, the storage buffer was

removed and 500 μl of radioimmunoprecipitation assay (RIPA) buffer (1 M Tris pH 8.0,

10% sodium deoxycholate, 5 M NaCl, 0.5 M EDTA pH 8.0, Triton X100, 10% SDS) was

added to rotate for 5 min. Supernatant was removed and beads were washed two

additional times with RIPA buffer. Beads were then blocked in 750 μl 1% bovine serum

albumin (BSA) in RIPA with rotation for 1 h. Excess BSA was removed from the beads

by washing 5 times with RIPA buffer. Streptavidin beads (50 μl) were then incubated

with 10 μl of fH (our designated bait protein), 200 μl of OmpA or Omp38 (our prey proteins of interest), and 440 μl of PBS for a total reaction of 500 l. Control samples were also prepared using streptavidin beads incubated with only OmpA, Omp38, or fH and PBS. Incubation was carried out overnight at 4°C with rotation. Each sample was washed 3 times in RIPA buffer to remove unbound proteins, followed by addition of 20

μl of SDS loading dye and 20 μl of PBS to each tube. Samples were boiled for 10 min and electrophoresced on a Bolt 4-12% Bis-Tris Plus SDS-PAGE gel for 45 min at 200 V.

This SDS-PAGE gel was transferred onto a PDVF membrane at 30 V for 90 min and the membrane was blocked in 5% milk in PBST for 1 h. The membrane was then incubated

in streptavidin-HRP (5mg/ml) (Southern Biotech) antibody and 5% milk to identify the presence of fH, followed by incubation separately in anti-6xHis (5mg/ml) antibody and

5% milk to detect the presence of histidine-tagged proteins. The blot was then visualized using ECL and imaged on the Syngene G:Box using Syngene Gbox Chemi-XX6 v1.6.1.0.

Chapter 3

Results

3.1 In vitro results

3.1.1 Burkholderia pseudomallei binds factor H

Multiple bacteria bind the host complement regulatory protein of the alternative pathway, fH (Meri et al., 2013; Amdahl et al., 2011; Hallstrom et al., 2008; Hellwage et al., 2001;

Nasr and Klimpel, 2008; Biedzka-Sarek et al., 2008). By doing so, these bacteria evade killing be serum complement. Firstly, we explored whether the saprophytic pathogen Bp could interact with biotinylated fH. Bp82 and Streptococcus pyogenes (Sp) strain emm8 cells were subjected to the whole-bacteria fH binding assay. Sp emm8 served as a negative control as it does not bind fH (Amdahl et al., 2011; Haapasalo et al., 2015;

Nissilä et al., 2017). The band at approximately 150 kDa in Figure 2.1 shows fH to be present in the Bp pellet, indicating Bp was able to bind fH. Figure 2.2 is a representative blot of three whole bacteria fH-binding assay experiments. After combining the results from the three experiments, the mean (± SD) amount of fH bound to Bp cells during incubation was 17.8 ± 3.5%, whereas fH bound

to Sp cells was 4.6 ± 2.0%. This difference in fH-binding is shown in Figure 2.2 and was calculated as being statistically significant through the use of a two tailed T-test with 6 degrees of freedom. The resulting p-value was 0.0006 with 95% confidence intervals of

8.18-18.1, a t-value of 6.47 and a standard error value of 2.034. Thus, showing levels of fH binding to Bp cells to be significantly increased when comapred to negative control strain Sp emm8.

FH was also identified in the Bp supernatant portion. This was free, unbound fH which was unable to move through the sucrose solution during centrifugation. Therefore figure

2.1 demonstrates that Bp is able to bind biotinylated fH to its outer surface, while significantly less binding occurred in Sp strain emm8 as expected. Lane 5 contained only

BSA which served as a negative control to confirm the absence of non-specific binding by the fH antibody.

3.1.2 Identification of B. pseudomallei protein facilitating factor H binding

via far western blot analysis

For proteins to facilitate an interaction with fH they must be situated on the bacterial outer membrane and extracellularly exposed. In order to identify which Bp proteins interact with fH, Bp cells were lysed and OMPs isolated by centrifugation, and then separated on an SDS-PAGE gel before being transferred to a PDVF membrane. The transferred OMPs were subjected to far western blot analysis to identify a potential fH- binding partner. We separated both Bp and Bt OMPs to determine if similar proteins

were contained by each. This OMP preparation was simultaneously taken from cultures

grown in LB or TSA media to control for environmental differences in fH-binding

abilities. Purified fH was run as a control to confirm its molecular weight and verify the

antibody was working correctly. One or more OMPs of approximately 37kDa were

observed to bind biotinylated fH from both the Bp and Bt genomes. From Figure 3, we can see that growth of Bp and Bt on LB and TSA media caused no difference in the ability of either bacteria to interact with biotinylated fH on its surface.

3.1.3 Putative factor H binding proteins identified via mass spectrometry

analysis

The 37kDa OMP previously identified to bind to fH by far western blot analysis in both

Bt and Bp was prepared for mass spectrometry analysis to identify what proteins may be facilitating this binding. Bp and Bt cells underwent OMP preparation and were then separated by SDS-PAGE and visualized with Coomassie Blue. In Figure 4, the dense band seen at approximately 37 kDa in both Bp and Bt corresponds to the size of the candidate fHbp identified in Figure 3.

Mass spectrometry analysis of the 37kDa bands identified more than 300 proteins. The curated list of putative fHbps was narrowed down using three criteria. Firstly, the protein had to be present in both Bt and Bp proteomes. Secondly, the protein had to possess a high abundance of mapped peptides. Thirdly, proteins scored by PSORTb to be an expected OMP designated them as a protein of interest. After applying these conditions,

the proteins shown in Table 1 were the top four candidates to be the putative 37 kDa fHbp identified via far western analysis.

3.1.4 Expression of putative Factor H binding proteins Omp38 and OmpA

Based on the guidelines listed above, the porin Omp38 emerged as the most likely fHbp candidate in Table 8. It possessed 571 mapped peptides, the highest result of the four candidates, and had a molecular weight of 39 kDa. This was the closet of the four candidate proteins to the 37 kDa band size identified via far western blot analysis.

Furthermore, Omp38 has been previously expressed at 38 kDa (Siritapetawee et al.,

2004). To explore further its possible relationship with fH, the Omp38 gene was cloned into the pBAD18-KAN expression plasmid and marked on the N-terminus with a histidine tag. This was done using Omp38 forward and reverse primers listed in Table 4.

A histidine-marked control was used to confirm the antibodies worked correctly. The result in Figure 5 shows successful gradual expression with IPTG induction (0.6 mM concentration), and suggests that Omp38 should be expressed for 4 h after induction to yield the highest protein concentrations.

The fourth protein listed in Table 2, OmpA, had simultaneously been identified as a candidate fHbp by bioinformatic methods (see Chapter 3.2 for details). BLASTp and PSI-

BLAST searches used H. influenza P5 and A. baumannii OmpA fHbps as search seeds against the Burkholderia group (taxid:119060) in the non-redundant database. Bp OmpA was found to resemble both proteins at the sequence level, and possess conserved

domains. This evidence suggested that OmpA would be a top target to further investigater for fH-binding. OmpA forward and reverse primers listed in Table 4 were used to mark the OmpA gene with a histidine tag on its N-terminal end. Protein expression across 4 h

in Figure 6 shows OmpA visualization to be strongest at 4 h post induction with

arabinose (0.2% concentration) with a small amount of expression seen at 2 h.

3.1.5 Protein purification of putative Factor H binding proteins OmpA and

Omp38

After successful expression of Bp OmpA and Omp38, proteins were extracted and

purified as shown in Figure 7. After expression, the E. coli culture was pelleted and then

subjected to soluble protein extraction. Protein isolation protocols must be tailored to the

proteins of interest, which in this case were two membrane proteins. Overexpression of

both proteins was achieved in E. coli cells to yield large quantities of Bp OmpA and

Omp38 recombinant proteins. Soluble protein extraction was carried out to isolate bulk

protein fractions from disrupted cells, while inclusion bodies were also pelleted and

subjected to inclusion body protein extraction. Inclusion bodies, which are an

accumulation of biologically-inactive partially-folded proteins, can occur when cells are

over-expressing proteins, including membrane proteins. The histidine tag was utilized to

purify both proteins using nickel beads in immobilized metal-affinity chromatography.

Figure 7 shows successful protein extraction in three fractions of Omp38 and two

fractions of OmpA. Purified Omp38 ran at the expected molecular weight of 37 kDa,

while OmpA ran at a slightly higher molecular weight, depending on the molecular

weight marker being used (data not shown). This presents some variation from the 37 kDa size obtained via far western blot methods and submitted for mass spectrometry analysis. Membrane proteins are known to migrate anomalously on SDS gels (Rath &

Deber, 2013).

3.1.6 Purified Omp38 and OmpA do not bind biotinylated Factor H via an

Immunoblot analysis

The reactivity of biotinylated-fH with purified OmpA and Omp38 was tested to examine if either recombinant protein could bind fH. This was done by use of a streptavidin bead pulldown assay. The western blot visualized in Figure 8 indicates that neither purified

OmpA nor Omp38 bind biotinylated fH. Lanes 1, 2 and 3 contain a band of ~150 kDa, which demonstrates that the streptavidin beads worked correctly, as they were able to successfully bind to the biotin. Lane 2 contains streptavidin beads which were incubated in biotinylated fH followed by OmpA. The absence of a band at ~35 kDa shows that

OmpA was unable to bind to fH. Lane 3 contains streptavidin beads which were incubated in biotinylated fH followed by Omp38. The absence of a band at ~40 kDa shows Omp38 was unable to bind to fH. Lane 4 contained streptavidin beads which had only been incubated in OmpA, whereas in lane 5 streptavidin beads had only been incubated in Omp38. Both lanes lack the presence of any band, indicating that no non- specific binding occurred between the purified protein samples and the streptavidin beads. Lane 6 contained only a sample of purified OmpA and lane 7 containing only purified Omp38. The bands seen at ~35 kDa in lane 6 and ~40 kDa in lane 7 in figure 7

demonstrate that the mouse anti-6x-histidine antibody is working correctly. This further supports the result shown by lanes 2 and 3 that purified OmpA and Omp38 do not bind to biotinylated fH. Lane 8 contains only BSA, which served as a negative control to confirm the absence of non-specific binding by the primary and secondary antibodies to the membrane. Finally, lane 9 contains biotinylated fH which has not been incubated with any proteins to allow visualization of the correct molecular size for fH and confirm the streptavidin-HRP primary antibody worked appropriately.

3.1.7 Summary

We observed a significant level of fH binding to the outer surface of Bp by incubation with biotinylated-fH (figure 2.1). This was shown by far western blot analysis to be due to a 37 kDa OMP. However, we were unable to show fH binding to purified samples of two candidate fHbps identified in the Bp genome at approximately 37 kDa. Our results show it to be very unlikely that OmpA or Omp38 facilitate fH binding in Bp. The ability of Bp to bind fH on its surface has not been previously reported. These results give us new insights into how Bp may be evading the alternative pathway of complement and persisting in the host cells. The identification of a fHbp on Bp’s outer membrane is a possible therapeutic target in protection against melioidosis infection.

3.2 In silico results

3.2.1 Known Factor H binding proteins do not interact with Factor H via a

conserved domain

Many microbes use proteins exposed on their outer surface to bind to the complement regulator fH. Pathogens commonly bind to fH sites within the short consensus repeats 6-7

or 19-20 (Meri et al., 2013). Table 9 shows known bacterial fHbps. For an initial in silico

analysis to identify any conserved domains or motifs which may be required in the

implementation of fH binding all known fHbps from Table 9 were submitted for

assessment by MSA programs or servers such as T-COFFEE. MSA programs did not

detect any conserved domains or motifs shared among the known fHbps, though several

of the Borrelia proteins are related to one-another.

The same proteins from Table 9 were then investigated as to whether their sites of

interaction with fH had previously been determined. Figure 9 highlights the fHbps from

N. meningitidis, OspE from B. burgdorferi, M6 from S. pyogenes, PspC from S.

pneumoniae, P5 from H. influenzae and Sib from S. aureus. Red boxes outline the

domains and residues necessary for fH binding. In addition, secondary structure

assignments were made to predict surface-exposed regions. Figure 9 shows neither

similarity in the secondary structures involved in fH binding across various microbes, nor

conservation of the number of binding- associated sites or the location of these sites. A

good example of this disparity across proteins is shown by the results of H. influenzae P5

and N. meningitidis fHbp. In the former, fH binding occurs in the N-terminal domain,

while in the latter residues across the entire length of the protein are required (Webb &

Cripps, 1998; Schneider et al., 2009).

The known fHbps from Table 8 were submitted to an all-versus-all percent sequence

identity (PSI) analysis to detect any patterns of relatedness. The PSI matrix in Figure 10

showed the only fHbps having a non-self identity 70% to one another to be the B.

burgdorferi proteins CRASP3, CRASP4, CRASP5 and OspE (red shading). The other

nineteen known fHbps only had PSI scores above 30% with themselves. No proteins

displayed intermediate PSI similarity scores. This suggests that the various proteins arose

via convergent evolution. This result allowed us to reduce the number of search seeds to

submit to a BLAST search, against the Burkholderia group (taxid:119060), from 19 to 16

based on the numbers of unrelated known fHbps.

3.2.2 Proteins found to be Homologous to known Factor H-binding proteins

in the Burkholderia taxon

Sixteen of the known fHbps from the PSI matrix (figure 10) were submitted to BLASTp

to search the Burkholderia group (taxid:119060). The P5 protein from the H. influenza

species aligned to numerous non-redundant protein sequences. One of the other top

results was a multispecies OmpA protein, accession number WP_004189892.1. This had

a query cover score of 35%, an E-value of 5e-30 and an identity score of 48%. This

protein was also identified as a candidate fHbp of 37 kDa by far western blot analysis

carried out on Bp outer membrane preparations (Figure 3). The 37 kDa band extracted

from the Coomassie-stained SDS-PAGE gel in Figure 4 was submitted for LC-MS/MS,

and OmpA (WP_004189892.1) was found to be present.

BLAST analysis, against the Burkholderiales order (taxid:80840), found four known

fHbps to possess conserved domains among proteins across several Burkholderia species,

in regions predicted to be extracellularly-exposed. These domains may thus be sites which can facilitate fH binding in Bp as they may be doing in the known fHbps. These proteins were P. aeruginosa EF-Tu, A. baumannii OmpA, L. interrogans Na-K

symporter, and N. gonorrhoeae porin B. For these proteins, the site of fH-interaction is

unknown. Similarities were demonstrated through use of Logo analysis (Crooks et al.,

2004), as presented in Figure 11. Gold squares indicate extracellularly-exposed portions

of each known fHbp, identifying them as having the potential to interact with fH.

Extracellularly-exposed portions were predicted by the protein localization server Predict

Protein (Rost, Yachdav & Liu, 2004). Full Logo analysis figures are shown in appendices

A1-A4. In Figure 11.1.1, strong pattern conservation was identified between all EF-Tu

proteins. However, this is not surprising for such a highly conserved protein, and

furthermore it is not known what portion of the P. aeruginosa ortholog is responsible for

fH-binding. It may be a region where the residues across the 25 sequences were found to

be least conserved among Tu proteins. Some sequence divergence is expected from the

majority of EF-Tu proteins that do not bind fH. Therefore, conserved regions between Bp

and P. aeruginosa EF-Tu orthologs which differ from the majority of EF-Tu proteins

were identified. Accordingly, an alignment of EF-Tu orthologs which are not known to

bind to fH was performed using a two-sample Logo analysis to highlight regions of

difference (figure 11.1.2). Letters seen on the upper level of the output represent amino

acids enriched in the fHbp-positive sample alignment, whereas those displayed on the

lower level of the logo analysis show amino acids depleted in the positive sample

alignment. Differences in symbol frequency were observed in some regions which

aligned to those predicted to be extracellularly exposed in P. aeruginosa. Some short

motifs of symbols were present in these exposed regions; however these were not shown

to be conserved in the known fHbp P. aeruginosa EF-Tu.

For the Burkholderia genus proteins homologous to A. baumannii OmpA, areas of

conservation were observed in the N-terminal domain and sporadically across the C- domain in extracellularly-predicted regions (figure 11.2). However, the segment from position 93-105 is the only portion to show consistent conservation, and even this is not

specific to all 28 proteins. Cellular apoptosis susceptibility (CAS) and transporter

proteins across the Burkholderia genus were found to have sequence similarity to L. interrogans Na-K symporter protein (Figure 11.3). Sporadic conservation was seen across Figure 11.3 in portions which were predicted in the Na-K symporter protein to be extracellularly-exposed. This was present within the central portion of the amino acid sequence. No clear conserved segments are present. For proteins across the Burkholderia genus which were homologous to N. gonorrhoeae porin B, some conservation was seen between amino acids in regions predicted as extracellularly exposed (figure 11.4).

Positions 237-238 show a promising location for a conserved structure or functional

domain; otherwise long portions of conservation were not identified in the logo analysis alignment.

Due to no clear identification of conserved domains throughout the Burkholderia species to known fHbps, the tBLASTn, BLASTp or position-specific iterated (PSI)-BLAST search engines were employed. Proteins in the Bp 1026b genome were found to have sequence similarity to known fHbps from the literature. On submission of A. baumannii

OmpA and H. influenza P5, tBLASTn searches consistently identified OmpA-like family proteins with conserved C-terminal domains. To determine if the specific region of P5 known to bind fH in the N-terminal domain (highlighted in Figure 9E) would result in different candidate fHbps in the Bp genome, the portion of P5 containing loops 1 and 2 was submitted as a separate search, and OmpW was output as a related protein. Table 10 shows the fHbp used as a search seed with BpaC identified as a candidate fHbp in the Bp genome. BpaC has been previously identified as being the only autotransporter in Bp to influence pneumonia virulence in mice (Campos, Byrd & Cotter, 2013). With this knowledge it appeared logical to consider the likelihood of BpaC being a fHbp in Bp.

3.2.3 Potential Factor H-binding proteins predicted to be extracellularly

exposed on the Burkholderia pseudomallei outer surface

With previous literature indicating that defects in BpaC-expression caused reduced dissemination by the Bp340 strain (Campos, Byrd & Cotter, 2013), as well as OmpA being seen to be immunogenic against Bp in mice (Hara, Mohamed & Nathan, 2009), the

qualities that these proteins possessed in common with other fHbps was investigated.

OmpA has been confirmed as an OMP in other bacterial species such as E. coli (Freudl,

Klose & Henning, 1990). In addition, both OmpA and BpaC have been computationally predicted to be in the outer membrane of Bp by Psortb (Yu et al., 2010), and are thus theoretically capable of fH binding. Therefore, the topology of each protein was characterized.

Each protein sequence was submitted to the transmembrane prediction server TMpred to see if and how these proteins may be positioned in the outer membrane and extracellularly-exposed space. Table 11 shows that Bp OmpA is proposed to possess only one transmembrane domain, which was 19 residues in length situated in the N-terminal portion of the protein. This is seen boxed in red in Figure 12A. The highest scoring orientation of the single OmpA domain moved from the inside of the Bp outer membrane to the outside, resulting in residues 26-224 to being extracellularly exposed, since OmpA would only be embedded in the outer membrane at its N-terminus. TMpred predicted a lower ranked topology model for OmpA where only the first 7 residues were suggested to be extracellularly exposed. This alternative orientation lowers the possibility that OmpA may be a fHbp. Furthermore, Table 11 shows the eight transmembrane domains identified by TMpred in membrane protein BpaC. This shows BpaC as likely possessing substantial extracellularly-exposed regions. Between transmembrane domains six and seven, a portion of 576 residues were predicted to be extracellularly exposed and potentially bind fH. This segment is boxed in red in Figure 12B. TMpred proposed an

alternative model with a lower score, where the transmembrane domain moves from the outside of the Bp outer membrane to the inside, resulting in only the first six residues of

OmpA being displayed extracellularly. TMpred identified 4 portions of BpaC to be exposed extracellularly and serve as possible domains for fH binding.

3.2.4 Tertiary structure prediction of candidate Factor H binding proteins

using the RaptorX server

From topological evaluations, BpaC and OmpA both showed domains which might provide a site for fH-interaction on Bp’s surface. To determine what structures form these domains and allow an in silico analysis on how these proteins could bind fH, a model of each tertiary structure was developed. Modeling allows the identification of binding pockets which may be formed by secondary structure arrangements.

The RaptorX homology modeling server (Morten et al., 2012) used earlier solved structures to predict 3-D configurations of OmpA and BpaC, with the former displayed in

Figures 13.1-13.2 and the latter in Figure 14.1. The template used for overall structural prediction of OmpA was the C-terminal domain of OmpA from Salmonella enterica subsp. enterica serovar Typhimurium str. 14028S, PDB accession number 4rhaA. OmpA was designated as having two domains connected by a linker sequence. The OmpA structural prediction had a p-value of 4.92e-05 with a global distance score (GDT) score of 59, both values indicated the overall structure prediction of OmpA to be a good probable indicator of its true structure. In Figure 13.1 and 13.2, 224 (100%) residues have

been modeled and 52 (23%) positions are predicted to be disordered. RaptorX predicted an N-terminal α-helical domain which is highlighted by a green oval in Figure 13.1. This corresponds to the portion of the protein predicted by TMpred to be a transmembrane domain.

The N-terminal domain was predicted based on the template structures kappa light chain variable domain (PDB accession 3upfA), Norwalk virus RNA-dependent RNA polymerase from strain Hu/NLV/Dresden174/1997/GE (PDB accession 2b43A), and rabbit hemorrhagic disease virus RNA-dependent RNA polymerase (PDB accession

1khvA). In contrast to the whole-protein prediction, the p-value for this domain was

3.41e-02, which is not a hugely confident result: a desirable p-value for a structure is below 10-3. Furthermore, the GDT score was 25, indicating a low number of residues to have low modeling error. The overall score of 27 was also small. The N-terminal domain was only 84 residues in length but this alignment score denotes just less than a third of the structure to be well predicted. Secondary structure assignments for OmpA were as follows: 29% α-helix, 15% β-strands and 54% loop regions. The RaptorX prediction of domain 2 represented the OmpA C-terminal domain. This contained a beta-alpha-beta- alpha-beta-beta secondary structure prediction, which is consistent with OmpA-like secondary structure classifications in other bacterial species, allowing them to interact with the peptidoglycan layer (De Mot & Vanderleyden, 1994; Koebnik, 1995). This conserved domain is highlighted by a green square in Figure 13.2. The C-terminal prediction was produced using PDB structure 4rhaA.

The C-terminal was shown to be a good indicator of structure by the given p-value of

4.92e-05. The GDT score of 81 surpasses 50, which denotes the absolute quality of the structure to be strong and more reliable than the N-terminal predicted model. The C- terminal model prediction was also shown to be more confidently predicted than the N- terminal, with the overall score of 112. A white linker domain is shown in Figures 13.1 and 13.2 to connect the two domains. TMpred suggests this domain to be totally extracellularly exposed to facilitate fH binding in its preferred topology model.

Alternatively, this domain is to be fully embedded in or below the outer membrane of Bp.

RaptorX tertiary structure prediction server classified the BpaC protein to contain four domains. The predicted structure had a p-value of 6.52e-05, which satifies the 10-3 or 10-4 scores desired for a good structure. However, the overall GDT score of 32 was poor, showing quite a large number of residues may have some low quality modeling due to the score being quite a bit lower than the desired value of 50. This result shows 100% of residues as being modelled, where 27% of these are predicted as being disordered, suggesting there are regions able to interact with numerous proteins in these regions through alterations in the BpaC surface (figure 14.1). The structure for domain 2 highlights the portion of BpaC with a large extracellularly exposed region. The prediction possessed a suitable p-value of 1.80e-04, which is lower than the score of 10-3 recommended for a model of decent quality. The GDT of 23 was low, just like the overall structure, but this may be reflective of the disordered regions present throughout BpaC.

The overall score of 54 was incredibly low for a domain containing 685 residues, which is simultaneously the highest possible value for this alignment score.

BpaC domain 2 is shown in Figure 14.2 using the CPK spacefill format. The color scheme reflected the BpaC secondary structure assignments, where 8% was α-helical domains, 28% β-sheets and 63% coil formations, shown to extend outwards in white. To evaluate the possible interaction of BpaC with fH, the RaptorX predicted structure in

Figure 14.1 was submitted to the ZDOCK protein-protein docking prediction server

(Pierce et al., 2014). The full BpaC structure consists of 1152 residues, which was too large to submit to ZDOCK. However, the RaptorX server prediction of BpaC domain 2 consisted of residues 1-812, incorporating the large extracellularly-exposed portion boxed in red in Figure 12B, between transmembrane domains six and seven (table 11). The next rational step was to submit this domain 2 structure with fH to ZDOCK and predict how the two may interact.

3.2.5 Candidate Factor H binding proteins OmpA and BpaC are predicted to

bind to Factor H via in silico analysis

The ZDOCK server generated 10 poses considered to be the most probable ways in which fH SCRs 19-20 and the candidate fHbps OmpA or BpaC would interact on the surface of Bp. FH SCRs 19-20 are available as a solved structure in the PDB database

(2BZM_A). This structure was submitted to ZDOCK as the ligand protein, with OmpA or BpaC as the receptor protein. It was chosen because it has previously been deemed as

the ‘common microbial binding site’ (Meri et al., 2013) for numerous fHbps. These two

C-terminal domains have also been shown to form dimers and tetramers (Jokiranta et al.,

2006), which may be accountable for higher bacterial binding affinity to fH SCRs 19-20

(Amdahl et al., 2011). Furthermore, PDB structure 2BZM chain A did not exceed the residue limit imposed by ZDOCK for submitted structures. All residues on fH SCRs 19-

20 were marked as available for binding to our candidate fHbps. OmpA residues available for interaction with fH were designated and submitted to ZDOCK based on the preferred orientation by TMpred. Therefore, all residues, excluding the first 25 in the N- terminal, were submitted as open and accessible for binding to fH SCRs 19-20. BpaC was submitted to ZDOCK with the residues predicted to be extracellularly exposed by

TMpred as open for binding with fH SCRs 19-20. Each docked structure was then assessed in terms of interaction strength. This included detecting the number of residues facilitating an interaction, and calculating the dissociation constant and binding affinity values at the protein-protein interface. This was carried out by the protein binding energy prediction server PRODIGY (Vangone & Bovine, 2015) through submission of the tertiary structure complexes generated by ZDOCK. The protein-protein docked poses identified by PRODIGY to possess the strongest interactions are recorded in Table 12.

The predicted structures seen in Figures 15 and 16 were amongst the top 10 ZDOCK generated models by which OmpA or BpaC may dock with fH SCRs 19-20. Each structure possessed the greatest binding affinity and dissociation constant from the

ZDOCK top 10 predicted protein-protein interactions, deeming them the most likely

means by which BpaC and OmpA might interact with fH SCRs 19-20. ZDOCK protein- protein docking server predicted the binding between B. pseudomallei OmpA and fH to most likely be close to OmpA’s N-terminal domain, and PRODIGY predicted this to take place via 103 residues. Whereas BpaC and fH SCRs 19-20 are predicted to most likely interact via 186 extracellularly exposed inter residual contacts. Positive control values for binding affinity and dissociation constant values were calculated by submission of solved structures from the PDB database assembled during interaction with fH. Negative control values were yielded by using structures predicted by sequence shuffled fH SCRs 19-20.

Sequence manipulation suite: shuffle protein (Stothard, 2000) was used to produce 5 randomly shuffled versions of the fH SCRs 19-20 protein sequence. Each of these shuffled-sequences was submitted to RaptorX homology modeling server to produce a tertiary structure prediction. These 5 alternative protein structures were docked with Bp

BpaC domain 2 or OmpA structures using ZDOCK. The top 10 docked interactions produced by ZDOCK were assessed by PRODIGY for their binding affinity and dissociation constant values. The highest scoring complex for each of the 5 shuffled fH structures docked with either BpaC or OmpA was used to calculate the binding affinity mean and standard deviation scores which were then documented in Table 12.

Assessment of these known interactions, and the interactions between Bp OMPs and shuffled fH sequence structures, aided in interpretation of the values calculated for candidate fHbps OmpA and BpaC docked with fH SCRs 19-20.

Complement C3b in complex with fH domains 1-4 (PDB accession 2WII) provided the

strongest interaction between fH and a binding protein. The complex of fH SCRs 6-7 and

N. meningitidis fHbp Variant 1 R106A mutant (PDB accession 4AYD) presented the

strongest known interaction between a microbe and fH (Table 12). The structure of B.

burgdorferi OspE in complex with fH SCRs 19-20 (PDB accession 4J38) represented the weakest interaction between a microbe and fH (table 12). The Kd values between these two extremes ranged over 1015-fold.

PRODIGY calculated Bp OmpA and fH SCRs 19-20 (Figure 15), as well as Bp BpaC

domain 2 and fH SCRs 19-20 (Figure 16), to possess a stronger binding affinity and

dissociation constant than that for the solved crystal structure of known fHbp OspE from

the B. burgdorferi genome when in complex with fH SCRs 19-20. This finding shows

OmpA and BpaC to both be able to facilitate fH binding to Bp’s outer surface. However,

this result should be addressed with some caution. The complexes using five shuffled fH

sequences for both BpaC (-20.3 kcal mol-1) and OmpA (-15.4 kcal mol-1) presented mean

binding affinity measures greater than solved fH SCRs 19-20 (PDB accession 2BZM_A)

docked with BpaC (-19.1 kcal mol-1) and OmpA (-11.8 kcal mol-1). In this case binding energy alone is not enough to distinguish a true interaction from a false one.

3.2.6 Summary

Examination of known fHbps demonstrated that there is no conserved fH-binding domain or structure across pathogenic species. This is almost certainly due to the mechanism arising by convergent evolution. We have identified OmpA and BpaC as two candidate fHbps in the Bp genome which might play a role in its ability to survive intracellularly through the evasion of the host-complement pathway. We saw through bioinformatic analysis that both proteins possess characteristics required to facilitate fH binding.

However, we could not demonstrate that OmpA binds to Bp by using molecular biology methods (Figure 8). These results suggest that BpaC’s capability to bind to fH should be investigated through in vitro methods.

Chapter 4

Discussion

Bioinformatic methods have great potential for identification and characterization of

Burkholderia proteins that may be interacting with host complement regulatory protein fH. This might be one strategy by which Bp evades clearance from the body by the alternative pathway of complement, therefore enhancing its microbial virulence.

Recruitment of fH to a pathogen’s surface by a fHbp could inhibit further amplification of the complement cascade, blocking C3b deposition on the microbial surface and preventing complement-mediated killing of bacteria. We evaluated two proteins by further in silico analysis for their potential to be fHbps and by using in vitro methods to characterize their abilities to bind fH. Our bioinformatic results identified at least two proteins, BpaC and OmpA, to be promising fHbps. However, we were unable to validate an interaction between recombinant candidate fHbps using molecular biology techniques.

Our results suggest that Bp may be binding fH via a different protein than the two we

tested to evade the alternative pathway of complement, or that the conditions used for our

in vitro analysis are suboptimal. Otherwise, Bp’s ability to persist in the body may be

facilitated by an alternative mechanism.

This study demonstrated how potential fHbps in the Bp genome can be recognized

through bioinformatic methods. Proteomic-based approaches have been used in previous

literature to identify proteins which bind to fH and fH-like 1 (McDowell et al., 2006).

Our analysis of known fHbps for common motifs showed the lack of conservation

between known fHbps across species. The specific regions required for binding to fH are

known in only six currently reported fHbps, and the residues required for specific

interaction have been pinpointed in two of these proteins (Figure 11). These sites of

interaction do not occur in conserved sequences, shared secondary structures or

equivalent protein domains. Furthermore, parallels in fH interaction sites are still not

obvious within the bacterial classes of Gram-negative or Gram-positive species. This

dissimilarity between fHbps has been described as due to different pathogens evolving

convergently to utilize the common microbial binding site on fH SCRs 19-20. Binding to

this site allows formation of a tripartite complex consisting of fH, C3b and a microbial

protein, which enhances the fH-mediated inactivation of C3b (Meri et al., 2013).

Interestingly, the Percent Sequence Identity (PSI) matrix showed that the fHbps within the B. burgdorferi species are not homologous to one another. Lack of homology between fHbps in the Borrelia genus has previously been shown, although computer analysis recognized these proteins to all have a high predictive probability of forming coiled-coil motifs, suggesting the involvement of these structures would determine the

ability to bind to fH (Hovis et al., 2004). CRASP 1 and CRASP 2 presented PSI scores below 30% to all other fHbps in the B. burgdorferi species. These diverse binding sites,

which enhance immune evasion amongst pathogens, appear to have emerged by a process

of convergent evolution. The structures involved are completely different from organism to organism. In the case of divergent evolution, an identifiable conserved domain would likely have remained. Alternatively, the presence of fHbp genes on transferable genetic elements may facilitate the sharing of this binding phenotype between species. It was to be expected that Logo analysis, to look for patterns and levels of conservation in proteins across the Burkholderia genome and known fHbps, would not return signatures associated with fH-binding (Figure 10). Furthermore, the tertiary structures for proteins such as A. baumannii OmpA and L. interrogans Na-K symporter proteins have not yet been characterized, or their extracellularly-exposed domains confirmed. The results of a two-sample Logo analysis (Figure 11.1.2), between EF-Tu proteins in the Burkholderia species and EF-Tu proteins from Gram-negative bacteria not known to bind fH, did not reveal any significant subcluster-specific sequence differences. Logo analysis positions

200-204 show substantial sequence differences not seen amongst EF-Tus known not to bind fH. However, these residues are not conserved within the search seed P. aeruginosa

EF-Tu which is a known fHbp. The results of this study support previous conclusions that a common motif does not carry out fH binding to pathogen surfaces. It can be speculated that a conserved residue or binding site cannot be used to identify microbial fHbps in future microbial studies. However, this possibility cannot yet be ruled out, as the precise

points of interaction with fH have not been identified for most known fHbps; , only the portion of the protein encompassing these pivotal residues have been determined.

These known fHbps listed in Table 9 were submitted to BLAST searches against the non- redundant protein sequences, Expressed Sequenced Tagged (EST), Transcriptome

Shotgun Assembly (TSA) and Whole-Genome-Shotgun (WGS) databases to identify candidate fHbps by similarities in residues, domains or conserved motifs. Methods using

PSI-BLAST, tBLASTn and BLASTp identified seven proteins predicted to possess the potential to enable Bp to bind to fH (Table 10). This study found that Bp OmpA

(WP_004189892.1) and BpaC (WP_014696818.1) possessed some sequence similarity to known fHbps, and these were chosen for further investigation based on previous research showing their potential as therapeutic targets. OmpA is the only protein to have been previously identified as a fHbp in other bacterial species (e.g., H. influenza and A. baumannii), increasing the possibility it could play the same role in Bp. For BpaC,

RaptorX homology modeling predicted 27% of residues to be disordered, meaning these portions of the protein surface are not constrained and may fold into a necessary binding site on interaction with another protein. Intrinsically disordered proteins are abundant among disease-related proteins, and often fold to carry out molecular recognition between numerous protein partners (Uversky, Oldfield & Dunker, 2008 and Ravarani et al., 2018).

As an autotransporter, the ability to bind multiple partners via disordered regions would be beneficial. Factor H binding is specific and carried out by hydrogen bonding as seen with B. burgdorferi OspE (Kolodziejczyk et al., 2017) and N. meningitidis fHbp

(Schneider et al., 2009). The interchangeable surface of disordered regions denotes them

as good targets for drug discovery because the binding site can be stimulated by its

correspondence with an effective partner (Uversky, Oldfield & Dunker, 2008 and

Ravarani et al., 2018).

Bioinformatic analysis deemed OmpA and BpaC as possessing features required for

fHbps: localization to the outer membrane and the presence of transmembrane domains

that allow extracellularly exposed regions (Table 11). The homology modeled tertiary

structures of OmpA and BpaC were predicted to bind to fH short consensus repeats

(SCRs) 19-20. The C-terminal of fH, also known as the common microbial binding site,

is the region that several bacterial species utilize for binding fH to their surface (Meri et

al., 2013; Ferreira, Pangburn & Cortés, 2010). SCR 20 has been more specifically

characterized as a ‘super evasion site’ (Meri et al., 2013). Uncertainty arises in the

docking of proteins by computational methods, because comparison of interacting protein

complexes can only be made to those of known protein complexes.

Of course, simulations may not completely reflect the genuine structure of a protein-

protein interaction being investigated. However, the use of docking tools to predict

protein-protein interactions are becoming promising complementary approaches for

rational drug design. The complexes formed by docking BpaC or OmpA with fH displayed stronger dissociation constant and binding affinity values than the known B. burgdorferi OspE and fH complex (Table 12). The solved structure of fH SCRs 19-20 in

complex with OspE (PDB accession 4J38) uses only 8 residues to interact, whereas

OmpA is predicted by the PRODIGY server to have 103 inter-residual contacts and BpaC speculated as having 186 inter-residual contacts with fH. N. meningitidis fHbp uses 52 residues to bind fH (PDB accession 4AYD). These differences in contact numbers are due to the differences in contact cut-offs used by PRODIGY and in the solving of the confirmed PDB structures 4J38 and 4AYD. PRODIGY server uses a distance threshold of 5.5Å to define a contact (Vangone & Bonvin, 2015). However, PDB structure 4J38 was solved with a distance threshold of 2.83Å (Bhattacharjee et al., 2013) and 4AYD solved with distance threshold of 2.4Å (Johnson et al., 2012) for inter-residual contacts at the protein-protein interface. These distance cutoffs make it difficult to compare the predicted structures to the known structures in terms of inter-residual contacts. However, by submitting these solved structures to PRODIGY the interface cutoffs for inter-residual contacts are increased to 5.5Å, which provides a means to compare the binding affinity at the protein-protein interface to the number of contacts involved. B. burgdorferi OspE and fH SCRs 19-20 are predicted by PRODIGY to be interacting with 66 contacts whereas N. meningitidis fHbp and fH SCRs 6-7 is predicted to use 74 binding contacts. BpaC is a large protein, 1152 amino acids in length, so it is logical that this would need a greater number of binding sites than the 172-amino acid protein OspE. Even though BpaC has >2 times the number of contact points predicted by PRODIGY than when N. meningitidis is in complex with fH, its binding energy values are lower, which is somewhat surprising as the number of contacts often correlates with the binding energy values. These binding affinity and dissociation constant values should be used with appropriate caution. I

generated predicted structures of shuffled fH sequences, as a control, and docked them

with OmpA and BpaC using ZDOCK. This was to disrupt the structure of fH and

investigate if this simultaneously disrupted the ZDOCK predicted complexes of OmpA

and BpaC interacting with fH. Surprisingly, the binding affinity and dissociation constant

energy values for complexes involving shuffled fH structures were predicted to be greater

than our candidate fHbps.

One shortcoming of docking techniques is that they frequently demonstrate partners and

non-partners to interact with the same sites on a single target protein (Sacquin-Mora et al., 2008 and Lopes et al., 2013). This is possibly occurring with our shuffled fH sequence structures and the PDB structure 2BZM_A. The use of binding energy alone is not sufficient to determine whether a protein is a true docking partner or not.

Nonetheless, recognition of the true binding site locations can aid in establishing a partner which may have a true association (Sacquin-Mora et al., 2008 and Lopes et al.,

2013). We can see that PRODIGY predicts OmpA and BpaC both bind to fH with interfacial energy stronger than the crystal solved complex of OspE docked with fH

(PDB accession 4J38). However, the equally high binding affinity values of OmpA and

BpaC bound with shuffled fH sequence structures raises questions about the way in which atom-atom contacts are established between real partners compared with non-real ones. If BpaC or OmpA were true fHbps, then their contact distributions to shuffled fH sequence structures should differ from docking with PDB structure 2BZM_A. This would provide a distinction between genuine partners and non-real ones. Docking predictions

are fundamentally improved when the locations of correct binding interfaces on each

protein are known (Sacquin-Mora et al., 2008). This is a limitation in our research as the precise locations in which fH and BpaC and OmpA are not known. If specific residues were known to be important on the fH SCRs 19-20 structure for binding, this may somewhat improve the potential binding energies and give more accurate docking predictions. However, in this study all residues on PDB structure 2BZM_A were submitted as being open to binding, which reduces the precision of the interaction. The

autotransporter BpaC has been considered a novel serum resistance factor which enables

Bp to resist killing by the classical or lectin complement pathways (Adler et al., 2015).

Bioinformatic analysis suggests there may be an association where OmpA or BpaC

enable Bp to bind to fH, interfering with the alternative pathway of complement. These

results will lead to future studies involving BpaC cloning, expression and purification to

investigate its relationship with fH in vitro.

Binding of Bp to fH was investigated in our laboratory by incubating bacteria in

biotinylated-fH for 30 min. The amount of fH used was lower than physiologically

relevant concentrations present in human serum (4μg/ml). With the considerations of

genetic and environmental factors, concentrations of fH in human serum can vary

between 116 to 562 μg/ml (Esparza-Gordillo et al., 2004). This incubation period

provided sufficient time for adherence of fH to fHbps present on Bp’s outer surface. The

whole bacteria assay confirmed that Bp can interact with biotinylated-fH. Using a biotin

marker allowed detection of fH binding to bacterial cells via a streptavidin-HRP

conjugated protein, which is a very high-affinity protein-ligand interaction (Leary, Brigati

& Ward., 1983). Figure 2.1 and Figure 3 both show Bp binding to fH. The whole-bacteria

assay quantified fH binding as the percent of bound fH versus total fH available for Bp to

bind. A similar method of defining fH binding was also used in a radioligand-binding

assay (Amdahl et al., 2011). Four strains of Bordetella pertussis and a single Bordetella

parapertussis strain were shown to bind fH to their surfaces. The B. pseudomallei ratio of

fH-binding presented in Figure 2.1 (~17%) was greater than the ratio for B. pertussis B32

strain (~13%) and nearing the values for B. pertussis Tohama I (~21%), B. pertussis 175

(~22%) and B. pertussis 406 (~26%). Although B. pseudomallei levels of fH binding

were lower than their highest binding strain B. parapertussis (~32%), our negative control strain S. pyogenes was shown to regularly bind at levels <5% in accordance with the literature. This result and the statistically significant difference in the percent of fH- binding between S. pyogenes and B. pseudomallei, calculated by a two-tailed t-test in

Figure 2.2, allows us to believe that fH-binding is occurring with B. pseudomallei.

The growth of S. pyogenes is often reflected as a chain of cells developing into a single colony on agar plates as opposed to the single cell developing into a single colony growth of B. pseudomallei. It is possible that this discrepancy in growth may cause underestimations in the amount of S. pyogenes cells being utilized in fH interactions.

Thus, it is possible that the amount of fH binding to our negative control may be even lower than reported in Figure 2, where equivalent cell numbers calculated by colony counts are used for S. pyogenes and B. pseudomallei strain samples.

Surface-binding of fH to Bp is confirmed and also demonstrated on Bt in Figure 3. Both

Bp and Bt are able to recruit fH to enhance their ability to escape complement mediated

effects of the alternative pathway via a 37 kDa OMP. Neither BpaC or OmpA candidate

fHbps identified via bioinformatic methods equate to this size, although the 24 kDa

molecular weight of OmpA is nearing this size. The autotransporter BpaC has previously

been characterized to have a molecular weight of 107.4 kDa (Lafontaine et al., 2014).

Therefore, it is unlikely that BpaC plays a role in the binding of Bp to host complement

regulatory protein fH. However, BpaC has been shown to express at undetectable levels

when grown by routine laboratory methods. This has been displayed by the preparation of

outer membrane proteins which were analyzed by western blot with α-BpaC antibodies

which failed to recognize the protein (Lafontaine et al., 2014). Further methods,

including immunoprecipitation and immunofluorescence-labeling, failed to detect the low

expression levels of BpaC. Interestingly, meningococcal serogroups A, W-135 and X isolates with low or intermediate fHbp gene expression were more resistant to anti-fHbp bactericidal activity. Isolate susceptibility to the bactericidal activity of fHbp-based vaccines increased with the increased expression of fHbps (Pajon et al., 2011). Thus,

BpaC could still facilitate fH binding to Bp’s surface, however it may also be

characteristically resistant to anti-fHbp bactericidal activity due to its low levels of

expression. Therefore, a fHbp with heightened expression in the Bp genome would be a

more desirable target for a primary fHbp based vaccine candidate.

These studies also suggest that Bp could be binding to fH via low-expressing OMPs, in addition to those present in the 37 kDa band visualized in Figure 3, but these interactions may need to be investigated through alternative methods. This is the case for NspA in N. meningitidis, which carries out residual fH binding that was not seen until N. meningitidis fHbp function was deleted (Lewis et al., 2010). The ability for a protein to enable fH- binding by molecular biology methods has commonly been revealed via construction of mutant strains which have undergone allelic replacement (Madico et al., 2006; Lewis et al., 2010).

The extensive genomic similarity between Bp and Bt allowed mutant strains possessing candidate fHbp gene interruptions to be identified and obtained from a Bt transposon mutant strain library. In both genomes fH binding is shown to take place via a protein with the same molecular weight of approximately 37 kDa. Phylogenetic analysis of 16S rRNA sequences found Bt as highly related to Bp and B. mallei, and was predicted to have diverged from the latter two sequences approximately 47 million years ago (Yu et al., 2006). The synteny identified between the Bt and Bp genomes means a single fHbp of 37 kDa can be searched for in both species. This similarity assisted in the streamlining of the list of proteins identified by mass spectrometry from the dense 37 kDa band distinguished in Figure 4. Where possible, these mutant strains were identified for all candidate fHbps in Table 10 that had been identified via BLAST search methods, as well as those returned by mass spectrometry analysis which were predicted to be OMPs.

Twenty-four strains with candidate fHbp gene interruptions were purchased.

Unfortunately, the library did not contain mutant strains for OmpA and Omp38 shown in

Table 8. However, mutant strains with loss of gene function for the two porin proteins,

WP_004553344.1 and WP_004554083.1, also listed in Table 8 and BpaC were present and purchased. Although, OmpA-family proteins are abundant in the outer membrane of

Bt and Bp, deletional mutagenesis of the OmpA investigated in this study is not possible due to its essential role in Bp’s survival (Moule et al., 2014). Therefore, the possible role of OmpA in fH binding cannot be investigated through a loss of function assay. However, we may be able to further investigate Omp38’s ability to bind fH using an allelic replacement method if desired.

The purchased Bt transposon mutants can be used in future studies to identify Bp proteins that bind to host complement regulatory proteins. One such instance would be to investigate the levels of C3 deposition able to accumulate on the surface of the transposon mutant strains obtained from the Bt library. When incubated in 20% normal human serum, Bp 1026b is resistant to deposition of complement protein C3 on its surface compared to Bt and mutant Bp DD503 acapsular strains (Woodman, Worth &

Wooten, 2012). Without the presence of this critical C3b complement opsonin and the

C3-convertases that it forms, amplification of the complement cascade does not take place, preventing membrane attack complex formation, and thus bacterial killing of Bp does not occur. Therefore, strains which show significant levels of C3 deposition on their surface would warrant further investigation. Knock-out strains purchased from the Bt mutant strain library could also be analyzed for their ability to bind to fH using the whole

bacteria binding assay or far western blot analysis. Identification of a strain demonstrated

lack of fH binding via either assay would indicate that this loss of function may be

attributable to an interrupted gene that encodes a fHbp.

This study showed that the Omp38 and OmpA recombinant proteins lacked the ability to

bind to fH in vitro. This was demonstrated through the use of an immunoblot (Figure 8),

a dot blot, and a far western blot analysis (results not shown). This result suggests that the

fH-binding mechanism by Bp may be facilitated by an alternative OMP. However,

aspects of the protocol could have hindered the fH binding ability of these two proteins.

Biotinylation of fH could be obstructing the site at which OmpA and Omp38 interact.

Biotin is attached to lysine residues across the fH protein, of which there are 80. The

charge conservation around the biotinylated lysine may be critical to facilitate the correct

positioning of the two interacting proteins (Chapman-Smith & Cronan, 1999). Therefore,

biotin may be preventing the normal interaction between OmpA or Omp38 and fH.

However, the ability of S. pneumonia PspC to interact with fH was successfully identified through probing with biotinylated-fH. This result shows that the binding abilities of fH

may not be hampered if fH lysine residues which undergo modification by biotin are not

key to the interaction with a fHbp partner. Candidate fHbps OmpA and Omp38 were

marked with a hexahistidine tag, which allows the proteins to be used as a reference point

for comparison of expression via western blot analysis. Additional problems may include

alterations in the restructuring of a native protein (Hakansson et al., 2000) and

obstruction in tertiary structure refolding of a protein after purification, either of which

could interfere with the reforming of fH-binding sites on OmpA and Omp38. Purification by affinity chromatography may have distorted the extraction of proteins in their innate forms. Both OmpA and Omp38 were eluted with a buffer containing 8M urea to solubilize them from inclusion bodies. This causes random coil protein structures to form

and exposes areas of hydrophobic amino acids, enhancing the protein’s ability to refold

into its natural formation. Undesirably, urea can also artificially initiate carbamylation in

proteins, and lysines are particularly susceptible to this mechanism (Kollipara & Zahedi,

2013). Damages to a protein can cause it to lose its tertiary and quaternary structure, so

on refolding it can no longer interact with the environment in the same manner. Failure to

return to its innate structure may prevent the protein from forming important binding

pockets and interaction sites needed in fH-binding. By denaturing the protein during the

extraction process this does provide a benefit in that it ensures the hexahistidine tag will

not be refolded into the recombinant protein structure. The exposure of the histidine tag is needed so it is accessible to interact with the immobilized metal sites on the column during purification. To remove imidazole and urea, the purified protein samples were dialyzed overnight as this will allow proteins to refold to their active states (Palmer &

Wingfield, 2012). Unfortunately, there is no way to guarantee OmpA and Omp38 have returned to their original structures. Another limitation of using purified protein is if a chaperone protein is required to facilitate the correct restructuring of OmpA and Omp38.

Such a chaperone would likely be absent in our purified extracts, and therefore neither protein will be able to refold correctly. Furthermore, a third protein could be required to enable the interaction between fH and any fHbps Bp may possess. This protein complex

will not be able to form in the purified protein samples, suggesting the possibility that

OmpA and Omp38 may have a true association with fH. For optimal binding of fH to

OspE, the protein’s native conformation must be preserved (Metts et al., 2003). The ability of S. pneumoniae PspC to bind fH was demonstrated using similar methods to those used to produce the immunoblot in Figure 8. C-terminal histidine tagged PspC was expressed in pET20b constructs in E. coli BL21(DE3) cells and induced with IPTG followed by purification by affinity chromatography (Brooks-Walter, Briles &

Hollingshead, 1999). Purified PspC was transferred to a nitrocellulose membrane for western blot analysis. This was incubated with biotinylated-fH followed by streptavidin- conjugated HRP antibody to detect fH-binding (Dave et al., 2001).

The affinity chromatography methodology utilized in our studies differed in that it was not carried out under non-denaturing conditions. This means our purified protein were exposed to reagents such as urea and chaotropic salts, both of which deteriorate the immunoadsorbent capacity, leading to denaturing of the eluted protein. If the conditions for our in vitro analysis were therefore altered to mirror this protocol, OmpA and Omp38 may be able to bind to fH if in fact urea was causing protein carabmylation or chaotropic salts were damaging protein structures. Future studies to investigate the capabilities of proteins to bind fH should explore using purified fH to remove any discrepancies introduced by biotin, or the extraction of purified candidate fHbps should be carried out under non-denaturing conditions.

The OmpA protein investigated in this study has been previously identified as

immunoreactive and its potential as a melioidosis vaccine has been explored. OmpA was

shown to provide mice with a 50% degree of protection by immunization. In a multiple

sequence alignment with OmpA-family proteins from other species, our candidate fHbp

clustered with known immunogenic OmpAs from E. coli, K. pneumoniae and P.

gingivalis (Hara, Mohamed, & Nathan, 2009). Clustering was seen around the conserved

C-terminal OmpA-like domain possessed by this family of OMPs. A small number of

these residues were conserved with the reduction‐modifiable protein M (RmpM) from N.

meningitidis. These residues are important for functional or structural roles when

interacting with the peptidoglycan layer (Grizot & Buchana, 2004). Through BLAST

searches in the Burkholderia group (taxid:119060) using H. influenzae P5 and A.

baumannii OmpA as known fHbp search seeds, Bp OmpA was identified in this study to

have some sequence similarity to both. This conservation was also present in the OmpA-

like C-terminal domain. In P5, the fH-binding site is known to be in extracellularly

exposed loops 1 and 2, which are situated in the N-terminal. Therefore, the similarity

identified between these proteins is probably based on how the structure interacts with

the peptidoglycan, not in how the proteins bind to complement regulatory proteins on the

bacterial surface. A further three OmpA-like family proteins and one peptidoglycan

associated lipoprotein (shown in Table 10) were identified in the Bp genome to share sequence similarity with H. influenzae P5 and A. baumannii OmpA. Again, this similarity

was only seen in the C-terminal OmpA-like domain. TMpred predicted the candidate fHbp OmpA in this study to possess an extracellularly exposed C-terminal domain. This

is the orientation in which OmpA was submitted to ZDOCK for analysis of its interaction

with fH SCRs 19-20. This topology contradicts literature-discussed localization of

conserved OmpA-like C-terminal domains which are commonly positioned in the

peptidoglycan (De Mot & Vanderleyden, 1994; Koebnik, 1995). Interestingly, ZDOCK

predicted fH domains 19-20 to interact with OmpA close to its N-terminal

transmembrane domain. Therefore, if OmpA does possess extracellularly exposed loops in its N-terminal or linker domain, like H. influenzae P5, it can still facilitate fH-binding.

The ZDOCK protein-protein interaction server results presented in this study provide supporting evidence for OmpA to be a candidate fHbp.

This same OmpA ortholog was identified by LC-MS/MS analysis alongside three porin

candidate fHbps presented in Table 8. The three porins were consistently identified by a

range of protein localization programs as being present in the cell outer membrane.

However, OmpA did not return a unanimous result, with localization predictions being in

the inner membrane, outer membrane, extracellular and periplasmic regions. This

variation may have been influenced by the fact that proteins containing OmpA-like

domains have numerous methods to anchor themselves into the outer membrane.

Commonly, a transmembrane β-barrel fold is seen in the N-terminal region (Pautsch &

Schulz, 1998). Some proteins attach to the periplasmic side of the outer membrane using

a lipid anchor, such as in E. coli PAL (Cascales et al., 2002). Motility protein B (MotB)

in Gram-negative and Gram-positive bacteria attaches to the inner membrane with a

conserved OmpA-like C-terminal domain located in the periplasmic region (De Mot &

Vanderleyden, 1994; Koebnik, 1995). Each cellular localization prediction program may

recognize a different conserved OmpA-family domain. This inconsistency in locations

makes it more difficult to come to a succinct conclusion as to whether Bp OmpA truly

possesses extracellularly exposed regions or if the protein remains fully concealed below

the outer membrane.

4.1 Conclusion

As there is a need to identify therapeutic targets in B. pseudomallei to assist in controlling

the global impact of melioidosis, this study attempted to identify Bp proteins that

promote its ability to evade the alternative pathway of complement and persist undetected

in the host. In this report, we have characterized the ability of Bp to bind fH to its surface.

This is hypothesized to be an important mechanism in its resistance to complement attack and phagocytosis. Based on bioinformatic and molecular biology methods, we have identified at least 3 strong candidates as possible fHbps. We were unable to confirm the

interaction of recombinant Omp38 or OmpA proteins with fH using an immunoblot.

However, recombinant protein pulldowns are not the standard method used for positively

identifying a fHbp. Moving forward, we will firstly test these candidate genes using a

thorough analysis of chromosomal knockout strains using antibodies to block binding. By

better understanding the dynamics of this mechanism, we will not only further understanding of Bp virulence, but perhaps also identify targets for therapies to prevent

melioidosis.

Table 1 Search settings used in each BLAST analysis

Parameter Value Expect score threshold 10 Word size 3 Scoring matrix BLOSUM62 Gap costs (Existence, extension) 9, 1

Table 2 Search settings used in each PSI-BLAST analysis

Parameter Selection Organism Burkholderiales (taxid:80840) Max target sequences 500 Expect score threshold 10 Word size 3 Scoring matrix BLOSUM62 Gap costs (Existence, 9, 1 extension) PSI-BLAST threshold 0.005

Table 3 Strains used in this study

Strain Description Source Storage Burkholderia ΔpurM derivative of Gift from Herbert TSA with pseudomallei 82 1026b. An adenine Schweizer. adenine and thiamine (University of auxotroph Colorado) Propst et al., 2010. Burkholderia Reference strain Gift from Don TSA thailandensis E264 Woods (University of Calgary) Brett, DeShazer and Woods, 1998. Streptococcus Negative control ATCC® 12349™ THY pyogenes emm8 strain E. coli DH5α Recipient cells for Thermofisher LB plasmid constructs 18265017 pBAD-18KAN and pPROEX-HTb E. coli JM109 Recipient cells for Promega L1001 LB plasmid construct pMol138 E. coli Expression cells for Millipore Sigma LB Rosetta™(DE3) plasmid construct 70954 pPROEX-HTb

Table 4 Sequences of primers and restriction enzymes used in this study

Primer Sequence OmpA Forward 5' GATCGAATTCAAAATAAACTTTCAAAGCTCGCG 3' OmpA Reverse 5' GATCCTGCAGTTACTGCGCCGGAACGG 3' 5' Omp38 Forward GATGAATTCAGGAGGAAACGATGCATCATCATCACCATC ATAACAAGACTCTGATTGT 3' Omp38 Reverse 5' GATCCTGCAGTTAGAAGCGGTGACGCAGACC 3' pPRO forward 5' AGCGGATAACAATTTCACACAGG 3' pPRO reverse 5' GATTTAATCTGTATCAGG 3' pBAD forward 5’ GATGAATTCAGGAGGAAACGATGCA pBAD reverse TCATCATCACCATCATAACAAGACTCTGATTGT 3’ 5’ GATCCTGCAGTTAGAAGCGGTGACGCAGACC 3’ Restriction Sequence enzymes EcoR1 Forward 3’ CTTAAG 5’ Pst1 Reverse 5’ GACGTC 3’

Table 5 Conditions used in Phusion PCR

PCR mastermix component Amount (μl) Isolated Bp DNA 2 Forward primer 1 (10 μM) Reverse primer 1 (10 μM) dNTPs 1.25 (8 mM) Taq Phusion 0.5 (2 U/μl) 5 x Phusion buffer 10

diH2O 34.25 Phusion PCR conditions Stage 1 98⁰C 30 s Stage 2 – Repeated 25 times 98⁰C 10 s 55⁰C 30 s 72⁰C 30 s Stage 3 72⁰C 5 min Samples then held at 4⁰C

100

Table 6 Parts included in the setup for an overnight ligation

Reaction component Amount (μl) 3:1 DNA PCR product 6 Plasmid construct 2 T4 buffer (NEB) 2 T4 ligase (NEB) 1 (400 U/μl)

diH2O 9

101

Table 7 Colony PCR mastermix setup

PCR mastermix component Amount (μl) Forward primer pPRO/pBAD 1 (10 μM) Reverse primer pPRO/pBAD 1 (10 μM) dNTPs 2.5 (8 mM) 10 x standard Taq buffer 2.5

diH2O 12.5 Taq 0.5 Colony PCR conditions Stage 1 95⁰C 3 min Stage 2 – Repeated 34 times 95⁰C 30 s 55⁰C 30 s 68⁰C 1 min Stage 3 68⁰C 5 min Samples then held at 4⁰C

102

Table 8 Four potential fHbps identified via mass spectrometry analysis and taken on for further characterization. Four OMPs identified by LC-MS/MS analysis which were present in both Bt and Bp proteomes, possessed a high abundance of mapped peptides and were predicted to be OMPs by PSORTb. Accession numbers for each candidate fHbp are listed. Protein GenBank Molecular Peptides Psortb outer Identifier Weight Mapped membrane (kDa) score (Max score of 10) Porin Omp38 WP_004553327.1 39 571 10.00 Outer membrane WP_004553344.1 41 164 10.00 porin Outer membrane WP_004554083.1 40 72 9.99 porin Membrane WP_004189892.1 24 24 9.93 protein (OmpA)

103

Table 9 Known factor H binding proteins identified in the literature. A search of published literature in databases such as MEDLINE on identified fHbps with the corresponding reference number for protein sequence retreival. Known factor H binding Species Genbank PMID protein reference Attachment invasion Yersinia enterocolitica ALO79037.1 18625735 locus (Ail) protein Outer surface protein E Borrelia burgdorferi AAC34953.1 11113124 (OspE) CRASP 1 Borrelia burgdorferi AAC66286.1 11385611 CRASP 2 Borrelia burgdorferi AAC65998.1 11385611 CRASP 3 Borrelia burgdorferi pdb|4BOB|A 11705962 CRASP 4 Borrelia burgdorferi WP_032489327.1 11705962 CRASP 5 Borrelia burgdorferi AAC18619.1 11705962 FhbA Borrelia hermsii AAY42861.1 16552029 M6 precursor peptide Streptococcus pyogenes AAA26920.1 2964038 Na-K symporter Leptospira interrogans ABE47421.1 16622202 OmpA Acinetobacter baumannii AJF83030.1 19878322 P5 Haemophilus influenzae AAC22819.1 18566420 Porin B Neisseria gonorrhoeae AAV28531.1 9480984 PspC Streptococcus pneumoniae Q9KK19 11292770 Immunoglobulin-binding Staphylococcus aureus BBA24967.1 19112495 protein (sbi) Elongation factor (EF-Tu) Pseudomonas aeruginosa AGY66997.1 17709513 Adhesin YadA Yersinia enterocolitica NP_052433.1 18625735 Factor H binding protein Neisseria meningitidis ABC59063.1 16785547 Neisserial surface protein Neisseria meningitidis PDB: 1P4T_A 20686663 A (NspA)

104

Table 10 B. pseudomallei candidate fHbps identified via tBLASTn. The known fHbp or fH interaction site used as a search seed in tBLASTn, BLASTp or Position- Specific Iterated (PSI) -BLAST is listed in column 1 with the corresponding protein to be identified as a candidate fHbp in the Bp genome listed in column 3. Searches using tBLASTn were against genomic data whereas searches using PSI- BLAST and BLASTp were of protein sequences in the Non-redundant database. Known Bacterial Bp Genbank E- Identity Number fHbp species candidate reference value score of fHbp (%) residues included OmpA A. baumannii TamB WP_004205949.1 0.0 58 204 YadA Y. BpaC WP_014696818.1 5e-34 38 172 enterocolitica P5 H. influenzae OmpW WP_004550362.1 7e-29 34 120 loops 1 and 2 P5 H. influenzae Pal WP_004186227.1 5e-09 34 120 OmpA A. baumannii Membrane WP_004189892.1 1e-33 33 116 protein (OmpA) P5 H. influenzae OmpA WP_004531578.1 3e-12 39 137 OmpA A. baumannii family WP_004192130.1 2e-14 35 123 proteins AFI69249.1 4e-11 34 120 PorB N. Porin WP_004525210.1 0.0 32 104 gonorrhoeae

105

Table 11 TMpred identifies transmembrane domains to be present in OmpA and BpaC. Submission of amino acid sequences for Bp OmpA and BpaC to transmembrane domain prediction program TMpred projected the position and orientation of membrane-spanning protein portions. Scores show the strength of how likely a transmembrane domain has been correctly assigned. Peaks with a score above 1000 are considered significant. Trans- Start Last Length Score Orientation membrane residue residue number OmpA 1 7 25 19 1321 Inside- preferred Outside model OmpA 1 7 25 19 1132 Outside- alternative Inside model 1 46 64 19 1399 Outside – Inside BpaC 2 145 163 19 1231 Inside – Strongly Outside preferred 3 193 211 19 1317 Outside – model Inside 4 268 290 23 1183 Inside - Outside 5 302 324 23 565 Outside – Inside 6 359 376 18 668 Inside – Outside 7 962 978 17 993 Outside – Inside 8 994 1014 21 545 Inside - Outside

106

Table 12 Binding affinities and dissociation constant values for fH and interacting proteins. Docked complexes produced by ZDOCK of likely fHbp-fH interactions and known fHbp-fH interactions were submitted to PRODIGY binding energy prediction webserver. Structures from the PDB database were submitted for the known fHbp interactions between complement C3b (PDB accession 2WII), N. meningitidis (PDB accession 4AYD) and B. burgdorferi (PDB accession 4J38) with fH. Bp OmpA and BpaC were docked with fH structures where the sequence had been shuffled to act as a control. These results are presented as the mean binding affinity (±standard deviation). PRODIGY calculated the binding affinities and dissociation constants between the interacting proteins in each of these structures at 37⁰C. ΔG (kcal mol-1) at Protein-protein complex Kd (M) at 37⁰C 37⁰C Complement C3b and factor H -33.8 1.6e-24 N. meningitidis -fhbp and factor H -26.9 1.2E-19 B. pseudomallei BpaC domain 2 and sequence shuffled factor H -20.3 (±4.6) 2.4E-13 (±4.1E-13) B. pseudomallei BpaC domain 2 and factor H -19.1 3.2E-14 B. pseudomallei OmpA and sequence shuffled factor H -15.4 (±3.2) 2.8E-10 (±4.7E-10) B. pseudomallei OmpA and factor H -11.8 4.5E-09 B. burgdorferi OspE and factor H -10.3 5.7E-08

107

Figure 1. The main components of the complement cascade. The three pathways of complement activation converge at the production of the C3 convertase enzyme. (1) The classical pathway is initiated by antibody-binding to a specific antigen on a microbial surface. (2) The lectin pathway is stimulated by the attachment of mannose-binding lectin to the carbohydrate mannose on the surface of an invading pathogen, but not on healthy human cells and tissues. (3) The alternative pathway activation takes place on microbial surfaces that create a conducive environment for the complement cascade. C3 convertase goes onto initiate chemotaxis and inflammation through recruitment of inflammatory cells, opsonization and phagocytosis of pathogens, through the use of C3b and its cleavage products iC3b and C3d, and formation of membrane attack complexes (MACs) to disrupt pathogen membranes and lyse cells.

108

Bp Sp

Pellet BSA Pellet Supernatant Supernatant Biotinylated-fH

~150kD

Figure 2.1 B. pseudomallei whole cells bind to biotinylated factor H but S. pyogenes cells do not. Visualization of fH as bound in the Bp pellet and as free fH in the supernatant portion. Sp emm8 strain and BSA protein served as negative controls. Biotinylated fH was used as a positive control to confirm fH to be present at the correct molecular weight. This verified the bands in the Bp and Sp samples to actually be fH protein. The fH control also confirmed that the antibodies used to visualize the western blot worked successfully. This blot represents 1 of 3 biological replicate whole bacteria fH-binding assay experiments performed simultaneously.

109

Whole Bacteria Factor H Binding Assay

4.6 17.8 * 0 5 10 15 20 25 30 35 Percent Biotinylated Factor H Bound

S. pyogenes emm8 B. pseudomallei

Figure 2.2 Comparison of B. pseudomallei and S. pyogenes emm8 fH binding. Binding shown in the western blot in figure 2.1 was determined as a percent of bound fH versus absolute fH. Mean ± SD of triplicate samples from three whole bacteria fH binding assays are shown in figure 2.2. Sp emm8 fH binding was shown to be 4.6% ± 2.0 while Bp is 17.8% ± 3.5. This difference in fH binding was found to be statistically significant through a two-tailed T-test (p=0.0006, 95% CI=8.1803-18.1347, t=6.47, df=6, SE=2.034).

110

BT LB BP LB BP TSA Factor H BT TSA 250 150

100

Figure 3 A candidate fHbp was detected in B. thailandensis and B. pseudomallei. BT represents B. thailandensis E264 and BP represents B. pseudomallei 82. TSA and LB indicate the two complex growth media used. Far western blot analysis was carried out with OMPs which were isolated from whole Bp using sonication to lyse the cells. These proteins were separated on a gel and transferred to a membrane which was then probed with biotinylated fH. A protein with a molecular weight of approximately 37 kDa in the outer membrane of Bp and Bt was seen to bind to biotinylated fH. A sample of biotinylated-fH was used as a positive control appearing at the correct molecular weight of ~150 kDa. Molecular weights displayed on the right are in kilodaltons.

111

Bp Bt

~37kD

Figure 4 Putative fHbp visualized by Coomassie blue at 37kDa. Bp and Bt OMPs were separated by SDS-PAGE. Immersion in Coomassie blue triphenylmethane dye produced the blue banding visible in figure 4. Dense bands at approximately 37 kDa confirmed the presence of a protein in the outer membrane preparations from both bacteria in triplicate samples. This was submitted for LC MS/MS analysis.

112

3 hour 3 hour 4 hour His -marked control 2 hour 2 hour 0 hour 0 hour 1 hour

~37kD

Figure 5 Expression of Omp38. Expression of N-terminal histidine marked Omp38 was induced with IPTG (0.6 mM concentration) and measured every hour for 4 h post induction. Samples were run on an SDS-PAGE and transferred to western blot to allow visualization of expression using mouse anti-6xhistidine and goat anti- mouse HRP-conjugate antibodies. Omp38 expression increased at 37kDa across the 4-hour post induction times. The histidine-tagged control protein, VCA0732, was a gift from Jyl Matson, and had a predicted molecular weight of ~14.5kDa.

113

2 hour 2 hour 4 hour 0 hour 0 hour

~36kD

Figure 6 Expression of His-tagged OmpA. Whole cell lysates were separated on an SDS-PAGE and transferred to a PDVF membrane to then be incubated in mouse anti-6xhistidine antibody and goat anti-mouse HRP-conjugate antibody to enable visualization of the increase in OmpA expression at 36 kDa. Samples across a 4 h expression period after induction with arabinose (0.2% concentration) were collected. OmpA expression at 36 kDa was shown to be strongest at 4h post induction with a small amount of protein expression seen at 2 h.

114

Omp38 OmpA

~41kD

~35kD

Figure 7 SDS-PAGE of purified candidate fHbps Omp38 and OmpA. Expressed culture of candidate fHbps OmpA and Omp38 underwent soluble protein extraction followed by inclusion body extraction. The solubilized exclusion body portion was kept. From this, purified protein samples were extracted by affinity chromatography. Eluted fractions of purified OmpA and Omp38 were run on the same gel and this was incubated in Coomassie Blue to visualize the marker ladder (not shown) and the three fractions of Omp38 and two fractions of OmpA where protein was successfully extracted.

115

~150kD Streptavidin - HRP antibody

Mouse anti- 6xHis antibody ~40kD ~35kD

Streptavidin beads + + + + + - - - -

Biotinylated-fH + + + - - - - - +

OmpA - + - + - + - - -

Omp38 - - + - + - + - -

BSA ------+ -

Figure 8 Immunoblot showing candidate fHbps OmpA and Omp38 do not bind biotinylated fH under tested conditions. Streptavidin beads were incubated with biotinylated-fH and OmpA or Omp38. Samples were run on an SDS-PAGE and transferred to PVDF membrane. Detection of biotinylated fH bound to streptavidin beads was shown through use of streptavidin-HRP antibody. Histidine marked OmpA and Omp38 were visualized using a mouse anti- 6xhistidine antibody. Blots are cropped because they show exposure to different antibodies.

116

A. N. meningitidis, factor H binding protein

B. B. burgdorferi, Outer surface protein E

C. S. pneumoniae, Pneumococcal surface protein C

D. S. pyogenes, M6

E. H. influenza, P5

117

F. S. aureus, Immunoglobulin binding protein (Sbi) α helices β strands β bridge Hydrogen No secondary bonded turn structure assigned

Figure 9 Known factor H binding proteins with sites required for factor H interaction. Known fHbps in table 9 were explored for regions known to interact with fH. Sites of fH interaction on N. meningitidis, fHbp and B. burgdorferi, OspE were obtained from the PDB website from X-ray crystallography confirmed complexes. All other regions identified to be involved in fH binding were obtained from the literature. Secondary structure assignments are as follows: dark red rectangle = α helix; blue arrow = β strand; green = hydrogen bonded turn; yellow = β bridge and black = no secondary structure assignment.

118

High PSI 70-100% Intermediate PSI 30-70% Low PSI 0-30%

Figure 10 All-versus-all percent sequence identity matrix for all known factor H binding proteins. The known fHbps identified from the literature review and shown collectively in table 2 were submitted to an all-versus-all percent sequence identity (PSI) analysis to detect any patterns of relatedness. This was carried out using the Tree based Consistency Objective Function For AlignmEnt Evaluation (T-COFFEE) server found at https://www.ebi.ac.uk/Tools/msa/tcoffee/ and serviced by the European Bioinformatics Institute.

119

1-40 N-Termini

41-80

81-120

121-160

161-200

201-240

241-280

281-320

120

321-360

361-399 C-Termini

Figure 11.1.1 Burkholderia EF-Tu proteins show conservation to P. aeruginosa EF- Tu via Logo analysis. Twenty-five EF-Tu sequences were aligned and submitted for Logo analysis. Twenty-four of these were from different species in the Burkholderia genus and identified as possessing similarity to P. aeruginosa EF- Tu. The original search seed P. aeruginosa EF-Tu was included alongside these proteins. Extracellularly exposed regions were predicted for P. aeruginosa EF-Tu by Predict Protein localization server and highlighted in gold boxes. In all Logo analysis figures residue numbers refer to MSA positions. These may not correspond to the numbering in individual GenBank records.

121

Enriched 96.0% N-termini

Depleted 96.0%

122

Enriched 96.0%

Depleted 96.0% C-termini

Figure 11.1.2 Two sample Logo analysis demonstrating variation in EF-Tu proteins. Divergence between EF-Tu proteins in the Burkholderia species versus proteins not known to bind to fH in various species is shown by letters in the enriched and depleted portions of the Logo graph. Twenty-five EF-Tu proteins were submitted for the positive sample, which included twenty-four EF-Tu proteins from different Burkholderia species and EF-Tu from P. aeruginosa (which is known to bind to fH). The negative sample included twelve EF-Tu protein sequences from twelve different Gram-negative bacteria species, where the protein is not known to be a fHbp. The species used in the negative sample of EF-Tu proteins were E. coli, V. cholerae, Y. enterocolitica, N. meningitidis, N. gonorrhoeae, A. baumannii, H. influenzae, B. pertussis, F. tularensis, P. gingivalis, L. interrogans and B. burgdorferi. Gold boxes represent the residues where P. aeruginosa EF-Tu is predicted to be extracellularly exposed.

123

1-40 N-Termini

81-120

201-240

241-280

281-320

321-360

C-Termini

Figure 11.2 Burkholderia OmpA proteins show conservation to A. baumannii OmpA. Twenty eight OmpA sequences from across the Burkholderia genus were aligned and submitted for Logo analysis with the known CRASP A. baumannii OmpA. Extracellularly exposed regions of A. baumannii OmpA were classified by Predict Protein localization server and highlighted in gold boxes.

124

401-440 N-Termini

481-520

521-560

601-640 C-Termini

Figure 11.3 Cellular apoptosis susceptibility (CAS) and transporter proteins across the Burkholderia genus show conservation to L. interrogans Na-K symporter. Forty-four CAS, transporter and symporter proteins from different Burkholderia species were aligned and submitted to Logo analysis with the known fHbp L. interrogans Na-K symporter. Localization server, Predict Protein, was used to classify extracellularly exposed regions in L. interrogans Na-K symporter and these portions were highlighted in gold boxes.

125

41-80 N-Termini

81-120

121-160

161-200

201-240 C-Termini

Figure 11.4 Porin proteins across the Burkholderia genus show conservation to N. gonorrhoeae porin B. Twenty-three porin proteins from different Burkholderia species were homologous to known fHbp N. gonorrhoeae porin B. These were aligned and submitted for Logo analysis. Search seed protein N. gonorrhoeae porin B was included in the submission. Gold boxes highlight regions of N. gonorrhoeae porin B believed to be extracellularly exposed by Predict Protein localization server.

126

TMpred output for B. pseudomallei OmpA

2000

1000

-1000

-2000

-3000

-4000 Transmembrane domain score -5000 0 50 100 150 200 250 Residue position

12A. Identification of transmembrane domains in candidate fHbp OmpA.

127

TMpred output for B. pseudomallei BpaC

2000 1500 1000 500 0 -500 -1000 -1500 -2000 -2500 Transmembrane domain score -3000 0 200 400 600 800 1000

Residue position 12B. Identification of transmembrane domains in candidate fHbp BpaC.

Figure 12A-B. TMpred topological analysis identifies membrane-spanning domains. Protein sequences for OmpA (A) and BpaC (B) were submitted to transmembrane domain prediction server TMpred. Membrane spanning domains are represented by a peak on the graph. If this presents a score above 1000 it is significant. A solid black line represents the score for each transmembrane domain prediction where the orientation moves from the inside to the outside of the outer membrane. A grey line represents the score of the transmembrane domain when predicted to be in an outside to inside orientation. The red box in 12A denotes location of single transmembrane domain predicted by TMpred to be present in Bp OmpA. The red box in 12B denotes large extracellularly-exposed portion predicted by TMpred.

128

Figure 13.1 Tertiary structure of B. pseudomallei OmpA. Homology modeled structure of B. pseudomallei OmpA was created utilizing the solved structures of the C-terminal domain of OmpA from Salmonella enterica subsp. enterica serovar Typhimurium str. 14028S (PDB accession 4rhaA), kappa light chain variable domain (PDB accession 3upfA), Norwalk virus RNA dependent RNA polymerase from strain Hu/NLV/Dresden174/1997/GE (PDB accession 2b43A) and rabbit hemorrhagic disease virus RNA-dependent RNA polymerase (PDB accession 1khvA). The color scheme reflects secondary structure assignments and the structure is represented in a CPK spacefill format. Red corresponds to α- helical domains, yellow to β-strands, blue to β-turns and white to residues producing loops. All homology protein structure models were visualized using Jmol computer software (Hanson, 2010). The green oval highlights the α-helical domain predicted to be present in the OmpA N-terminal domain.

129

Figure 13.2 B. pseudomallei OmpA predicted to possess a conserved OmpA- like C-terminal domain. As in figure 13.1 but rotated to enable a clear visualization of the conserved OmpA-like C-terminal domain containing a beta- alpha-beta-alpha-beta-beta structure. The conserved OmpA-like domain is highlighted by a green box.

130

Figure 14.1 BpaC tertiary structure. Homology modeling of BpaC was predicted by RaptorX server using the structural template of the collagen-binding domain of Y. adhesin YadA (PDB accession 1P9H). The color scheme is categorized as secondary structure domains as in figures 13.1 and 13.2 with CPK spacefill format used to represent the full model.

131

Figure 14.2 BpaC domain 2 tertiary structure prediction by RaptorX. Domain two was more specifically constructed by RaptorX based on the crystal structures of drosophila DHX36 helicase in complex with GTTAGGGTT (PDB accession 5N9A) and RPA70N in complex with PrimPol (PDB accession 5N8A). One overall color was used to depict the structure in CPK spacefill format.

132

Figure 15 ZDOCK pose prediction of fH and OmpA interacting in silico. ZDOCK protein-protein docking prediction server used fast fourier transform methods to predict the interaction between B. pseudomallei OmpA and fH. B. pseudomallei OmpA raptorX predicted structure seen in figures 13.1-13.2 is displayed in green in figure 15. The solved structure of fH SCRs 19-20 is represented by PDB accession 2BZM chain A in grey-blue in figure 15.

133

Figure 16 ZDOCK pose prediction of fH and BpaC domain two interacting in silico. ZDOCK protein-protein docking server assessed the interaction between B. pseudomallei domain 2 of candidate fHbp BpaC and fH using fast fourier transform methods. The highest scoring predicted ZDOCK output is displayed in figure 16. The RaptorX predicted structure displayed in figure 14.2 is represented in figure 16 in green. FH is represented in figure 16 by the solved structure of fH SCRs 19-20, PDB accession 2BZM chain A in grey-blue.

134

References

Adler, N.R.L., Stevens, M.P., Dean, R.E., Saint, R.J., Pankhania, D., Prior, J.L., Atkins, T.P., Kessler, B., Nithichanon, A., Lertmemongkolchai, G. and Galyov, E.E., 2015. Systematic mutagenesis of genes encoding predicted autotransported proteins of Burkholderia pseudomallei identifies factors mediating virulence in mice, net intracellular replication and a novel protein conferring serum resistance. PloS one, 10(4), p.e0121271.

Allwood, E.M., Devenish, R.J., Prescott, M., Adler, B. and Boyce, J.D., 2011. Strategies for intracellular survival of Burkholderia pseudomallei. Frontiers in microbiology, 2.

Alsenz, J., Schulz, T.F., Lambris, J.D., Sim, R.B. and Dierich, M.P., 1985. Structural and functional analysis of the complement component factor H with the use of different enzymes and monoclonal antibodies to factor H. Biochemical Journal, 232(3), pp.841- 850.

Alsenz, J., Lambris, J.D., Schulz, T.F. and Dierich, M.P., 1984. Localization of the complement- component-C3b-binding site and the cofactor activity for factor I in the 38kDa tryptic fragment of factor H. Biochemical Journal, 224(2), pp.389-398.

Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J., Zhang, Z., Miller, W. and Lipman, D.J., 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic acids research, 25(17), pp.3389-3402.

Amdahl, H., Jarva, H., Haanperä, M., Mertsola, J., He, Q., Jokiranta, T.S. and Meri, S., 2011. Interactions between Bordetella pertussis and the complement inhibitor factor H. Molecular immunology, 48(4), pp.697-705.

135

Bandeira, T.J., Brilhante, R.S., Rocha, M.F., Coelho, I.C., Lima, R.A., Cordeiro, R.A., Rodrigues, J.L., Ribeiro, J.F., Sidrim, J.J., Magalhães, F.G. and Grangeiro, T.B., 2012. Clinical-epidemiological features of 13 cases of melioidosis in Brazil.

Batchelor, B.I., Paul, J., Trakulsomboon, S., Mgongo, M. and Dance, D.A., 1994. Melioidosis survey in Kenya. Transactions of the Royal Society of Tropical Medicine and Hygiene, 88(2), p.181.

Bateman, A., Coin, L., Durbin, R., Finn, R.D., Hollich, V., Griffiths‐Jones, S., Khanna, A., Marshall, M., Moxon, S., Sonnhammer, E.L. and Studholme, D.J., 2004. The Pfam protein families database. Nucleic acids research, 32(suppl_1), pp.D138-D141.

Barlow, P.N., Norman, D.G., Steinkasserer, A., Horne, T.J., Pearce, J., Driscoll, P.C., Sim, R.B. and Campbell, I.D., 1992. Solution structure of the fifth repeat of factor H: a second example of the complement control protein module. Biochemistry, 31(14), pp.3626-3634.

Barlow, P.N., Steinkasserer, A., Norman, D.G., Kieffer, B., Wiles, A.P., Sim, R.B. and Campbell, I.D., 1993. Solution structure of a pair of complement modules by nuclear magnetic resonance. Journal of molecular biology, 232(1), pp.268-284.

Bhagwat, M. and Aravind, L., 2007. Psi-blast tutorial. In Comparative Genomics (pp. 177-186). Humana Press.

Biedzka-Sarek, M., Jarva, H., Hyytiäinen, H., Meri, S. and Skurnik, M., 2008. Characterization of complement factor H binding to Yersinia enterocolitica serotype O: 3. Infection and immunity, 76(9), pp.4100-4109.

Blackmore, T.K., Fischetti, V.A., Sadlon, T.A., Ward, H.M. and Gordon, D.L., 1998. M protein of the group A Streptococcus binds to the seventh short consensus repeat of human complement factor H. Infection and immunity, 66(4), pp.1427-1431.

Bhattacharjee, A., Oeemig, J.S., Kolodziejczyk, R., Meri, T., Kajander, T., Lehtinen, M.J., Iwaï, H., Jokiranta, T.S. and Goldman, A., 2013. Structural basis for complement evasion by Lyme disease pathogen Borrelia burgdorferi. Journal of Biological Chemistry, 288(26), pp.18685-18695.

Bremmelgaard, A., Bygbjerg, I.B. and Høiby, N., 1982. Microbiological and immunological studies in a case of human melioidosis diagnosed in Denmark. Scandinavian journal of infectious diseases, 14(4), pp.271-275.

Brett, P.J. and Woods, D.E., 2000. Pathogenesis of and immunity to melioidosis. Acta tropica, 74(2), pp.201-210

136

Brett, P.J., Deshazer, D. and Woods, D.E., 1997. Characterization of Burkholderia pseudomallei and Burkholderia pseudomallei-like strains. Epidemiology & Infection, 118(2), pp.137- 148.

Brooks-Walter, A., Briles, D.E. and Hollingshead, S.K., 1999. The pspC gene of Streptococcus pneumoniae encodes a polymorphic protein, PspC, which elicits cross-reactive antibodies to PspA and provides immunity to pneumococcal bacteremia. Infection and immunity, 67(12), pp.6533-6542.

Campos, C.G., Byrd, M.S. and Cotter, P.A., 2013. Functional characterization of Burkholderia pseudomallei trimeric autotransporters. Infection and immunity, 81(8), pp.2788-2799.

Carlone, G.M., Thomas, M.L., Rumschlag, H.S. and Sottnek, F.O., 1986. Rapid microprocedure for isolating detergent-insoluble outer membrane proteins from Haemophilus species. Journal of Clinical Microbiology, 24(3), pp.330-332.

Carroll, M.C., 1998. The role of complement and complement receptors in induction and regulation of immunity. Annual review of immunology, 16(1), pp.545-568.

Cascales, E., Bernadac, A., Gavioli, M., Lazzaroni, J.C. and Lloubes, R., 2002. Pal lipoprotein of Escherichia coli plays a major role in outer membrane integrity. Journal of bacteriology, 184(3), pp.754-759.

Cendron, L., Veggi, D., Girardi, E. and Zanotti, G., 2011. Structure of the uncomplexed Neisseria meningitidis factor H-binding protein fHbp (rLP2086). Acta Crystallographica Section F: Structural Biology and Crystallization Communications, 67(5), pp.531-535.

Chang, K., Luo, J., Xu, H., Li, M., Zhang, F., Li, J., Gu, D., Deng, S., Chen, M. and Lu, W. 2017. Human Infection with Burkholderia thailandensis, China, 2013.

Chapman-Smith, A. and Cronan Jr, J.E., 1999. Molecular biology of biotin attachment to proteins. The Journal of nutrition, 129(2), pp.477S-484S.

Chen, Y.S., Hsiao, Y.S., Lin, H.H., Liu, Y. and Chen, Y.L., 2006. CpG-modified plasmid DNA encoding flagellin improves immunogenicity and provides protection against Burkholderia pseudomallei infection in BALB/c mice. Infection and immunity, 74(3), pp.1699-1705.

Cheng, A.C. and Currie, B.J., 2005. Melioidosis: epidemiology, pathophysiology, and management. Clinical microbiology reviews, 18(2), pp.383-416.

Chierakul, W., Rajanuwong, A., Wuthiekanun, V., Teerawattanasook, N., Gasiprong, M., Simpson, A., Chaowagul, W. and White, N.J., 2004. The changing pattern of bloodstream infections associated with the rise in HIV prevalence in northeastern

137

Thailand. Transactions of the Royal Society of Tropical Medicine and Hygiene, 98(11), pp.678-686.

Chierakul, W., Winothai, W., Wattanawaitunechai, C., Wuthiekanun, V., Rugtaengan, T., Rattanalertnavee, J., Jitpratoom, P., Chaowagul, W., Singhasivanon, P., White, N.J., Day, N.P. and Peacock, S., 2005. Melioidosis in 6 tsunami survivors in southern Thailand. Clinical infectious diseases, 41(7), pp.982-990.

Chou, K.C. and Shen, H.B., 2006. Large-scale predictions of gram-negative bacterial protein subcellular locations. Journal of Proteome Research, 5(12), pp.3420-3428.

Chua, K.L., Chan, Y.Y. and Gan, Y.H., 2003. Flagella are virulence determinants of Burkholderia pseudomallei. Infection and immunity, 71(4), pp.1622-1629.

Cullinane, M., Gong, L., Li, X., Adler, N.L., Tra, T., Wolvetang, E., Prescott, M., Boyce, J.D., Devenish, R.J. and Adler, B., 2008. Stimulation of autophagy suppresses the intracellular survival of Burkholderia pseudomallei in mammalian cell lines. Autophagy, 4(6), pp.744- 753.

Claassen, I., Meylis, J., van der Ley, P., Peeters, C., Brons, H., Robert, J., Borsboom, D., van der Ark, A., van Straaten, I., Roholl, P. and Kuipers, B., 1996. Production, characterization and control of a Neisseria meningitidis hexavalent class 1 outer membrane protein containing vesicle vaccine. Vaccine, 14(10), pp.1001-1008.

Crooks, G.E., Hon, G., Chandonia, J.M. and Brenner, S.E., 2004. WebLogo: a sequence logo generator. Genome research, 14(6), pp.1188-1190.

Currie, B.J., 2003. Melioidosis: an important cause of pneumonia in residents of and travellers returned from endemic regions. European Respiratory Journal, 22(3), pp.542-550.

Currie, B.J. and Kaestli, M., 2016. Epidemiology: A global picture of melioidosis. Nature, 529(7586), pp.290-291.

Currie, B.J., Ward, L. and Cheng, A.C., 2010. The epidemiology and clinical spectrum of melioidosis: 540 cases from the 20 year Darwin prospective study. PLoS neglected tropical diseases, 4(11), p.e900.

Dance, D.A., 1991. Melioidosis: the tip of the iceberg?. Clinical Microbiology Reviews, 4(1), pp.52-60.

Dance, D.A., 2000. Melioidosis as an emerging global problem. Acta tropica, 74(2), pp.115-119.

Dance, D.A.B., Smith, M.D., Aucken, H.M. and PItt, T.L., 1999. Imported melioidosis in England and Wales. The Lancet, 353(9148), p.208.

138

Dance, D.A., Wuthiekanun, V., Naigowit, P. and White, N.J., 1989. Identification of Pseudomonas pseudomallei in clinical practice: use of simple screening tests and API 20NE. Journal of clinical pathology, 42(6), pp.645-648.

Dave, S., Brooks-Walter, A., Pangburn, M.K. and McDaniel, L.S., 2001. PspC, a pneumococcal surface protein, binds human factor H. Infection and immunity, 69(5), pp.3435-3437.

de Córdoba, S.R., Esparza-Gordillo, J., de Jorge, E.G., Lopez-Trascasa, M. and Sánchez-Corral, P., 2004. The human complement factor H: functional roles, genetic variations and disease associations. Molecular immunology, 41(4), pp.355-367.

De Mot, R. and Vanderleyden, J., 1994. The C-terminal sequence conservation between OmpA- related outer membrane proteins and MotB suggests a common function in both gram- positive and gram-negative bacteria, possibly in the interaction of these domains with peptidoglycan. Molecular microbiology, 12(2), pp.333-336.

DiScipio, R.G., 1992. Ultrastructures and interactions of complement factors H and I. The Journal of Immunology, 149(8), pp.2592-2599.

Doker, T.J., Sharp, T.M., Rivera-Garcia, B., Perez-Padilla, J., Benoit, T.J., Ellis, E.M., Elrod, M.G., Gee, J.E., Shieh, W.J., Beesley, C.A. and Ryff, K.R., 2014. Contact investigation of melioidosis cases reveals regional endemicity in Puerto Rico. Clinical Infectious Diseases, 60(2), pp.243-250.

Elvin, S.J., Healey, G.D., Westwood, A., Knight, S.C., Eyles, J.E. and Williamson, E.D., 2006. Protection against heterologous Burkholderia pseudomallei strains by dendritic cell immunization. Infection and immunity, 74(3), pp.1706-1711.

Fang, F.C., 1997. Perspectives series: host/pathogen interactions. Mechanisms of nitric oxide- related antimicrobial activity. Journal of Clinical Investigation, 99(12), p.2818.

Ferreira, V.P., Pangburn, M.K. and Cortés, C., 2010. Complement control protein factor H: the good, the bad, and the inadequate. Molecular immunology, 47(13), pp.2187-2197.

Fischetti, V.A., Horstmann, R.D. and Pancholi, V., 1995. Location of the complement factor H binding site on streptococcal M6 protein. Infection and immunity, 63(1), pp.149-153.

Freudl, R., Klose, M. and Henning, U., 1990. Export and sorting of theEscherichia coli outer membrane protein OmpA. Journal of bioenergetics and biomembranes, 22(3), pp.441- 449.

Garcia, E.C., 2017. Burkholderia thailandensis: Genetic Manipulation. Current Protocols in Microbiology, pp.4C-2.

139

Garcia, B.L., Zhi, H., Wager, B., Höök, M. and Skare, J.T., 2016. Borrelia burgdorferi BBK32 inhibits the classical pathway by blocking activation of the C1 complement complex. PLoS pathogens, 12(1), p.e1005404.

Gardy, J.L., Laird, M.R., Chen, F., Rey, S., Walsh, C.J., Ester, M. and Brinkman, F.S., 2004. PSORTb v. 2.0: expanded prediction of bacterial protein subcellular localization and insights gained from comparative proteome analysis. Bioinformatics, 21(5), pp.617-623.

Giebink, G.S., Koskela, M., Vella, P.P., Harris, M. and Le, C.T., 1993. Pneumococcal capsular polysaccharide-meningococcal outer membrane protein complex conjugate vaccines: immunogenicity and efficacy in experimental pneumococcal otitis media. Journal of Infectious Diseases, 167(2), pp.347-355.

Gilmore, G., Barnes, J., Ketheesan, N. and Norton, R., 2007. Use of antigens derived from Burkholderia pseudomallei, Bt, and B. cepacia in the indirect hemagglutination assay for melioidosis. Clinical and Vaccine Immunology, 14(11), pp.1529-1531.

Gomez, M., Johnson, S. and Gennaro, M.L., 2000. Identification of Secreted Proteins ofMycobacterium tuberculosis by a Bioinformatic Approach. Infection and immunity, 68(4), pp.2323-2327.

Grizot, S. and Buchanan, S.K., 2004. Structure of the OmpA‐like domain of RmpM from Neisseria meningitidis. Molecular microbiology, 51(4), pp.1027-1037.

Haapasalo, K., van Kessel, K., Nissilä, E., Metso, J., Johansson, T., Miettinen, S., Varjosalo, M., Kirveskari, J., Kuusela, P., Chroni, A. and Jauhiainen, M., 2015. Complement factor H binds to human serum apolipoprotein E and mediates complement regulation on high density lipoprotein particles. Journal of Biological Chemistry, 290(48), pp.28977-28987.

Håkansson, K., Broder, D., Wang, A.J. and Miller, C.G., 2000. Crystallization of peptidase T from Salmonella typhimurium. Acta Crystallographica Section D: Biological Crystallography, 56(7), pp.924-926.

Hallström, T., Zipfel, P.F., Blom, A.M., Lauer, N., Forsgren, A. and Riesbeck, K., 2008. Haemophilus influenzae interacts with the human complement inhibitor factor H. The Journal of Immunology, 181(1), pp.537-545.

Hanson, R.M., 2010. Jmol–a paradigm shift in crystallographic visualization. Journal of Applied Crystallography, 43(5), pp.1250-1260.

Haque, Q.M., 2010. Evaluation of Available Diagnostic Tests for Melioidosis. Journal of Taibah University Medical Sciences, 5(2), pp.89-97.

Hara, Y., Mohamed, R. and Nathan, S., 2009. Immunogenic Burkholderia pseudomallei outer membrane proteins as potential candidate vaccine targets. PLoS One, 4(8), p.e6496.

140

Haraga, A., West, T.E., Brittnacher, M.J., Skerrett, S.J. and Miller, S.I., 2008. Burkholderia thailandensis as a model system for the study of the virulence-associated type III secretion system of Burkholderia pseudomallei. Infection and immunity, 76(11), pp.5402- 5411.

Harley, V.S., Dance, D.A., Tovey, G., McCrossan, M.V. and Drasar, B.S., 1998. An ultrastructural study of the phagocytosis of Burkholderia pseudomallei. Microbios, 94(377), pp.35-45.

Haupt, K., Reuter, M., van den Elsen, J., Burman, J., Hälbich, S., Richter, J., Skerka, C. and Zipfel, P.F., 2008. The Staphylococcus aureus protein Sbi acts as a complement inhibitor and forms a tripartite complex with host complement Factor H and C3b. PLoS pathogens, 4(12), p.e1000250.

Hellwage, J., Meri, T., Heikkilä, T., Alitalo, A., Panelius, J., Lahdenne, P., Seppälä, I.J. and Meri, S., 2001. The complement regulator factor H binds to the surface protein OspE of Borrelia burgdorferi. Journal of Biological Chemistry, 276(11), pp.8427-8435.

Heng, B.H., Goh, K.T., Yap, E.H. and Yeo, M., 1998. Epidemiological surveillance of melioidosis in Singapore. ANNALS-ACADEMY OF MEDICINE SINGAPORE, 27, pp.478-484.

Hofmann, K. & Stoffel, W., 1993. TMbase - A database of membrane spanning proteins segments Biol. Chem. Hoppe-Seyler 374, p.166

Hoiczyk, E., Roggenkamp, A., Reichenbecher, M., Lupas, A. and Heesemann, J., 2000. Structure and sequence analysis of Yersinia YadA and Moraxella UspAs reveal a novel class of adhesins. The EMBO journal, 19(22), pp.5989-5999.

Holden, M.T., Titball, R.W., Peacock, S.J., Cerdeño-Tárraga, A.M., Atkins, T., Crossman, L.C., Pitt, T., Churcher, C., Mungall, K., Bentley, S.D. and Sebaihia, M., 2004. Genomic plasticity of the causative agent of melioidosis, Burkholderia pseudomallei. Proceedings of the National Academy of Sciences of the United States of America, 101(39), pp.14240- 14245.

Hoque, S.N., Minassian, M., Clipstone, S., Lloyd-Owen, S.J., Sheridan, E. and Lessing, M.P.A., 1999. Melioidosis presenting as septic arthritis in Bengali men in east London. Rheumatology, 38(10), pp.1029-1031.

Horstmann, R.D., Sievertsen, H.J., Knobloch, J. and Fischetti, V.A., 1988. Antiphagocytic activity of streptococcal M protein: selective binding of complement control protein factor H. Proceedings of the National Academy of Sciences, 85(5), pp.1657-1661.

141

Hovis, K.M., McDowell, J.V., Griffin, L. and Marconi, R.T., 2004. Identification and characterization of a linear-plasmid-encoded factor H-binding protein (FhbA) of the relapsing fever spirochete Borrelia hermsii. Journal of bacteriology, 186(9), pp.2612- 2618.

Hovis, K.M., Jones, J.P., Sadlon, T., Raval, G., Gordon, D.L. and Marconi, R.T., 2006. Molecular analyses of the interaction of Borrelia hermsii FhbA with the complement regulatory proteins factor H and factor H-like protein 1. Infection and immunity, 74(4), pp.2007-2014.

Howard, K. and Inglis, T.J.J., 2003. Novel selective medium for isolation of Burkholderia pseudomallei. Journal of clinical microbiology, 41(7), pp.3312-3316.

Inglis, T.J., Rolim, D.B. and Sousa, A.D.Q., 2006. Melioidosis in the Americas. The American journal of tropical medicine and hygiene, 75(5), pp.947-954.

Jacobs, A.T. and Ignarro, L.J., 2001. Lipopolysaccharide-induced expression of interferon-β mediates the timing of inducible nitric-oxide synthase induction in RAW 264.7 macrophages. Journal of Biological Chemistry, 276(51), pp.47950-47957.

Jokiranta, T.S., Jaakola, V.P., Lehtinen, M.J., Pärepalo, M., Meri, S. and Goldman, A., 2006. Structure of complement factor H carboxyl‐terminus reveals molecular basis of atypical haemolytic uremic syndrome. The EMBO journal, 25(8), pp.1784-1794.

Jones, A.L., Beveridge, T.J. and Woods, D.E., 1996. Intracellular survival of Burkholderia pseudomallei. Infection and immunity, 64(3), pp.782-790.

Käll, L., Krogh, A. and Sonnhammer, E.L., 2004. A combined transmembrane topology and signal peptide prediction method. Journal of molecular biology, 338(5), pp.1027-1036.

Kespichayawattana, W., Rattanachetkul, S., Wanun, T., Utaisincharoen, P. and Sirisinha, S., 2000. Burkholderia pseudomallei induces cell fusion and actin-associated membrane protrusion: a possible mechanism for cell-to-cell spreading. Infection and immunity, 68(9), pp.5377-5384.

Kibbler, C.C., Roberts, C.M., Ridgway, G.L. and Spiro, S.G., 1991. Melioidosis in a patient from Bangladesh. Postgraduate medical journal, 67(790), pp.764-766.

Kim, S.W., Choi, C.H., Moon, D.C., Jin, J.S., Lee, J.H., Shin, J.H., Kim, J.M., Lee, Y.C., Seol, S.Y., Cho, D.T. and Lee, J.C., 2009. Serum resistance of Acinetobacter baumannii through the binding of factor H to outer membrane proteins. FEMS microbiology letters, 301(2), pp.224-231.

142

Kim, H.S., Schell, M.A., Yu, Y., Ulrich, R.L., Sarria, S.H., Nierman, W.C. and DeShazer, D., 2005. Bacterial genome adaptation to niches: divergence of the potential virulence genes in three Burkholderia species of different survival strategies. BMC genomics, 6(1), p.174.

Kjaer, T.R., Le, L.T., Pedersen, J.S., Sander, B., Golas, M.M., Jensenius, J.C., Andersen, G.R. and Thiel, S., 2015. Structural insights into the initiating complex of the lectin pathway of complement activation. Structure, 23(2), pp.342-351.

Koebnik, R., 1995. Proposal for a peptidoglycan‐associating alpha‐helical motif in the C‐ terminal regions of some bacterial cell‐surface proteins. Molecular microbiology, 16(6), pp.1269-1270.

Kollipara, L. and Zahedi, R.P., 2013. Protein carbamylation: in vivo modification or in vitro artefact?. Proteomics, 13(6), pp.941-944.

Kolodziejczyk, R., Mikula, K.M., Kotila, T., Postis, V.L., Jokiranta, T.S., Goldman, A. and Meri, T., 2017. Crystal structure of a tripartite complex between C3dg, C-terminal domains of factor H and OspE of Borrelia burgdorferi. PloS one, 12(11), p.e0188127.

Kunert, A., Losse, J., Gruszin, C., Hühn, M., Kaendler, K., Mikkat, S., Volke, D., Hoffmann, R., Jokiranta, T.S., Seeberger, H. and Moellmann, U., 2007. Immune evasion of the human pathogen Pseudomonas aeruginosa: elongation factor Tuf is a factor H and plasminogen binding protein. The Journal of Immunology, 179(5), pp.2979-2988.

Kraiczy, P., Skerka, C., Kirschfink, M., Brade, V. and Zipfel, P.F., 2001. Immune evasion of Borrelia burgdorferi by acquisition of human complement regulators FHL‐1/reconectin and Factor H. European journal of immunology, 31(6), pp.1674-1684.(a)

Kraiczy, P., Skerka, C., Brade, V. and Zipfel, P.F., 2001. Further characterization of complement regulator-acquiring surface proteins of Borrelia burgdorferi. Infection and immunity, 69(12), pp.7800-7809. (b)

Krishnaswami, C.S., 1917. Morphia injectors’ septicaemia. Indian Med Gaz, 52, pp.296-9.

Lafferty, J., McCallum, A. and Pereira, F.C., 2001. Conditional random fields: Probabilistic models for segmenting and labeling sequence data.

Lafontaine, E.R., Balder, R., Michel, F. and Hogan, R.J., 2014. Characterization of an autotransporter adhesin protein shared by Burkholderia mallei and Burkholderia pseudomallei. BMC microbiology, 14(1), p.92.

Langereis, J.D. and de Jonge, M.I., 2015. Invasive disease caused by nontypeable Haemophilus influenzae. Emerging infectious diseases, 21(10), p.1711.

Leary, J.J., Brigati, D.J. and Ward, D.C., 1983. Rapid and sensitive colorimetric method for visualizing biotin-labeled DNA probes hybridized to DNA or RNA immobilized on

143

nitrocellulose: Bio-blots. Proceedings of the National Academy of Sciences, 80(13), pp.4045-4049.

Lee, N.I.N.G., Wu, J.L., Lee, C.H. and Tsai, W.C., 1985. Pseudomonas pseudomallei infection from drowning: the first reported case in Taiwan. Journal of clinical microbiology, 22(3), pp.352-354.

Lewis, L.A., Ngampasutadol, J., Wallace, R., Reid, J.E., Vogel, U. and Ram, S., 2010. The meningococcal vaccine candidate neisserial surface protein A (NspA) binds to factor H and enhances meningococcal resistance to complement. PLoS pathogens, 6(7), p.e1001027.

Limmathurotsakul, D., Wongratanacheewin, S., Teerawattanasook, N., Wongsuvan, G., Chaisuksant, S., Chetchotisakd, P., Chaowagul, W., Day, N.P. and Peacock, S.J., 2010. Increasing incidence of human melioidosis in Northeast Thailand. The American journal of tropical medicine and hygiene, 82(6), pp.1113-1117.

Limmathurotsakul, D., Golding, N., Dance, D.A., Messina, J.P., Pigott, D.M., Moyes, C.L., Rolim, D.B., Bertherat, E., Day, N.P., Peacock, S.J. and Hay, S.I., 2016. Predicted global distribution of Burkholderia pseudomallei and burden of melioidosis. Nature microbiology, 1, p.15008.

Limmathurotsakul, D., Kanoksil, M., Wuthiekanun, V., Kitphati, R., Day, N.P. and Peacock, S.J., 2013. Activities of daily living associated with acquisition of melioidosis in northeast Thailand: a matched case-control study. PLoS neglected tropical diseases, 7(2), p.e2072.

Limmathurotsakul, D. and Peacock, S.J., 2011. Melioidosis: a clinical overview. British medical bulletin, 99(1), pp.125-139.

Liu, M.A., 2011. DNA vaccines: an historical perspective and view to the future. Immunological reviews, 239(1), pp.62-84.

Liu, T.Y., Gotschlich, E.C., Jonssen, E.K. and Wysocki, J.R., 1971. Studies on the Meningococcal Polysaccharides I. Composition and chemical properties of the group A polysaccharide. Journal of Biological Chemistry, 246(9), pp.2849-2858.

Lopes, A., Sacquin-Mora, S., Dimitrova, V., Laine, E., Ponty, Y. and Carbone, A., 2013. Protein- protein interactions in a crowded environment: an analysis via cross-docking simulations and evolutionary information. PLoS computational biology, 9(12), p.e1003369.

Madico, G., Welsch, J.A., Lewis, L.A., McNaughton, A., Perlman, D.H., Costello, C.E., Ngampasutadol, J., Vogel, U., Granoff, D.M. and Ram, S., 2006. The meningococcal vaccine candidate GNA1870 binds the complement regulatory protein factor H and enhances serum resistance. The Journal of Immunology, 177(1), pp.501-510.

144

Male, D.A., Ormsby, R.J., Ranganathan, S., Giannakis, E. and Gordon, D.L., 2000. Complement factor H: sequence analysis of 221 kb of human genomic DNA containing the entire fH, fHR-1 and fHR-3 genes. Molecular immunology, 37(1), pp.41-52.

Matsushita, M. and Fujita, T., 1992. Activation of the classical complement pathway by mannose-binding protein in association with a novel C1s-like serine protease. Journal of Experimental Medicine, 176(6), pp.1497-1502.

McCormick, J.B., Weaver, R.E., Hayes, P.S., Boyce, J.M. and Feldman, R.A., 1977. Wound infection by an indigenous Pseudomonas pseudomallei-like organism isolated from the soil: case report and epidemiologic study. Journal of Infectious Diseases, 135(1), pp.103- 107.

McDowell, J.V., Frederick, J., Stamm, L. and Marconi, R.T., 2007. Identification of the gene encoding the FhbB protein of Treponema denticola, a highly unique factor H-like protein 1 binding protein. Infection and immunity, 75(2), pp.1050-1054.

McDowell, J.V., Wolfgang, J., Tran, E., Metts, M.S., Hamilton, D. and Marconi, R.T., 2003. Comprehensive analysis of the factor H binding capabilities of Borrelia species associated with Lyme disease: delineation of two distinct classes of factor H binding proteins. Infection and immunity, 71(6), pp.3597-3602.

Meri, S. and Pangburn, M.K., 1990. Discrimination between activators and nonactivators of the alternative pathway of complement: regulation via a sialic acid/polyanion binding site on factor H. Proceedings of the National Academy of Sciences, 87(10), pp.3982-3986.

Meri, T., Amdahl, H., Lehtinen, M.J., Hyvärinen, S., McDowell, J.V., Bhattacharjee, A., Meri, S., Marconi, R., Goldman, A. and Jokiranta, T.S., 2013. Microbes bind complement inhibitor factor H via a common site. PLoS pathogens, 9(4), p.e1003308.

Metts, M.S., McDowell, J.V., Theisen, M., Hansen, P.R. and Marconi, R.T., 2003. Analysis of the OspE determinants involved in binding of factor H and OspE-targeting antibodies elicited during Borrelia burgdorferi infection in mice. Infection and immunity, 71(6), pp.3587-3596.

Miller, V.L., Beer, K.B., Heusipp, G., Young, B.M. and Wachtel, M.R., 2001. Identification of regions of Ail required for the invasion and serum resistance phenotypes. Molecular microbiology, 41(5), pp.1053-1062.

Miller, V.L. and Falkow, S., 1988. Evidence for two genetic loci in Yersinia enterocolitica that can promote invasion of epithelial cells. Infection and immunity, 56(5), pp.1242-1248.

Moore, R.A., Reckseidler-Zenteno, S., Kim, H., Nierman, W., Yu, Y., Tuanyok, A., Warawa, J., DeShazer, D. and Woods, D.E., 2004. Contribution of gene loss to the pathogenic

145

evolution of Burkholderia pseudomallei and Burkholderia mallei. Infection and immunity, 72(7), pp.4172-4187.

Morten Källberg, Haipeng Wang, Sheng Wang, Jian Peng, Zhiyong Wang, Hui Lu & Jinbo Xu. Template-based protein structure modeling using the RaptorX web server. Nature Protocols 7, 1511–1522, 2012

Moule, M.G., Hemsley, C.M., Seet, Q., Guerra-Assunção, J.A., Lim, J., Sarkar-Tyson, M., Clark, T.G., Tan, P.B., Titball, R.W., Cuccui, J. and Wren, B.W., 2014. Genome-wide saturation mutagenesis of Burkholderia pseudomallei K96243 predicts essential genes and novel targets for antimicrobial development. MBio, 5(1), pp.e00926-13.

Mulye, M., Bechill, M.P., Grose, W., Ferreira, V.P., Lafontaine, E.R. and Wooten, R.M., 2014. Delineating the importance of serum opsonins and the bacterial capsule in affecting the uptake and killing of Burkholderia pseudomallei by murine neutrophils and macrophages. PLoS neglected tropical diseases, 8(8), p.e2988.

Nasr, A.B. and Klimpel, G.R., 2008. Subversion of complement activation at the bacterial surface promotes serum resistance and opsonophagocytosis of Francisella tularensis. Journal of Leukocyte Biology, 84, p.77.

Nieves, W., Petersen, H., Judy, B.M., Blumentritt, C.A., Russell-Lodrigue, K., Roy, C.J., Torres, A.G. and Morici, L.A., 2014. A Burkholderia pseudomallei outer membrane vesicle vaccine provides protection against lethal sepsis. Clinical and vaccine immunology, pp.CVI-00119.

Nissilä, E., Douillard, F.P., Ritari, J., Paulin, L., Järvinen, H.M., Rasinkangas, P., Haapasalo, K., Meri, S., Jarva, H. and De Vos, W.M., 2017. Genotypic and phenotypic diversity of Lactobacillus rhamnosus clinical isolates, their comparison with strain GG and their recognition by complement system. PloS one, 12(5), p.e0176739.

Ngauy, V., Lemeshev, Y., Sadkowski, L. and Crawford, G., 2005. Cutaneous melioidosis in a man who was taken as a prisoner of war by the Japanese during World War II. Journal of clinical microbiology, 43(2), pp.970-972.

Notredame, C., Higgins, D.G. and Heringa, J., 2000. T-coffee: a novel method for fast and accurate multiple sequence alignment1. Journal of molecular biology, 302(1), pp.205- 217.

Pajon, R., Fergus, A.M., Koeberling, O., Caugant, D.A. and Granoff, D.M., 2011. Meningococcal factor H binding proteins in epidemic strains from Africa: implications for vaccine development. PLoS neglected tropical diseases, 5(9), p.e1302.

Palmer, I. and Wingfield, P.T., 2012. Preparation and extraction of insoluble (inclusion‐body) proteins from Escherichia coli. Current protocols in protein science, 70(1), pp.6-3.

146

Pangburn, M.K., 2000. Host recognition and target differentiation by factor H, a regulator of the alternative pathway of complement. Immunopharmacology, 49(1-2), pp.149-157.

Pangburn, M.K., Atkinson, M.A. and Meri, S., 1991. Localization of the heparin-binding site on complement factor H. Journal of Biological Chemistry, 266(25), pp.16847-16853.

Pangburn, M.K. and Müller-Eberhard, H.J., 1984, June. The alternative pathway of complement. In Springer seminars in immunopathology (Vol. 7, No. 2-3, pp. 163-192). Springer- Verlag.

Pappas, G., Panagopoulou, P., Christou, L. and Akritidis, N., 2006. Biological weapons. Cellular and molecular life sciences CMLS, 63(19-20), pp.2229-2236.

Parry, C.M., Wuthiekanun, V., Hoa, N.T.T., Diep, T.S., Thao, L.T.T., Loc, P.V., Wills, B.A., Wain, J., Hien, T.T., White, N.J. and Farrar, J.J., 1999. Melioidosis in Southern Vietnam: clinical surveillance and environmental sampling. Clinical infectious diseases, 29(5), pp.1323-1326.

Pautsch, A. and Schulz, G.E., 1998. Structure of the outer membrane protein A transmembrane domain. Nature Structural and Molecular Biology, 5(11), p.1013.

Peacock, S.J., Limmathurotsakul, D., Lubell, Y., Koh, G.C., White, L.J., Day, N.P. and Titball, R.W., 2012. Melioidosis vaccines: a systematic review and appraisal of the potential to exploit biodefense vaccines for public health purposes. PLoS neglected tropical diseases, 6(1), p.e1488.

Pearson, W.R., 1990. [5] Rapid and sensitive sequence comparison with FASTP and FASTA. Methods in enzymology, 183, pp.63-98.

Pelicic, V., 2008. Type IV pili: e pluribus unum?. Molecular microbiology, 68(4), pp.827-837.

Peltola, H., 1983. Meningococcal disease: still with us. Reviews of infectious diseases, 5(1), pp.71-91.

Peng, J. and Xu, J., 2009, May. Boosting protein threading accuracy. In Annual International Conference on Research in Computational Molecular Biology (pp. 31-45). Springer, Berlin, Heidelberg.

Peng, J. and Xu, J., 2011. A multiple‐template approach to protein threading. Proteins: Structure, Function, and Bioinformatics, 79(6), pp.1930-1939.

Perkins, S.J., Haris, P.I., Sim, R.B. and Chapman, D., 1988. A study of the structure of human complement component factor H by Fourier transform infrared spectroscopy and secondary structure averaging methods. Biochemistry, 27(11), pp.4004-4012.

147

Petersen, T.N., Brunak, S., von Heijne, G. and Nielsen, H., 2011. SignalP 4.0: discriminating signal peptides from transmembrane regions. Nature methods, 8(10), p.785.

Pevsner, J., 2015. Bioinformatics and functional genomics. John Wiley & Sons Inc.

Pierce, B.G., Wiehe, K., Hwang, H., Kim, B.H., Vreven, T. and Weng, Z., 2014. ZDOCK server: interactive docking prediction of protein–protein complexes and symmetric multimers. Bioinformatics, 30(12), pp.1771-1773.

Pitt, T.L., Aucken, H. and Dance, D.A.B., 1992. Homogeneity of lipopolysaccharide antigens in Pseudomonas pseudomallei. Journal of Infection, 25(2), pp.139IN1143-142146.

Pizza, M., Scarlato, V., Masignani, V., Giuliani, M.M., Arico, B., Comanducci, M., Jennings, G.T., Baldi, L., Bartolini, E., Capecchi, B. and Galeotti, C.L., 2000. Identification of vaccine candidates against serogroup B meningococcus by whole-genome sequencing. Science, 287(5459), pp.1816-1820.

Prasadarao, N.V., Blom, A.M., Villoutreix, B.O. and Linsangan, L.C., 2002. A novel interaction of outer membrane protein A with C4b binding protein mediates serum resistance of Escherichia coli K1. The Journal of Immunology, 169(11), pp.6352-6360.

Ram, S., Sharma, A.K., Simpson, S.D., Gulati, S., McQuillen, D.P., Pangburn, M.K. and Rice, P.A., 1998. A novel sialic acid binding site on factor H mediates serum resistance of sialylated Neisseria gonorrhoeae. Journal of Experimental Medicine, 187(5), pp.743-752.

Rappuoli, R., 2000. Reverse vaccinology. Current opinion in microbiology, 3(5), pp.445-450.

Rath, A. and Deber, C.M., 2013. Correction factors for membrane protein molecular weight readouts on sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Analytical biochemistry, 434(1), pp.67-72.

Ripoche, J., Day, A.J., Harris, T.J. and Sim, R.B., 1988. The complete amino acid sequence of human complement factor H. Biochemical Journal, 249(2), pp.593-602.

Rogul, M. and Carr, S.R., 1972. Variable ammonia production among smooth and rough strains of Pseudomonas pseudomallei: resemblance to bacteriocin production. Journal of bacteriology, 112(1), pp.372-380.

Roos, A., Bouwman, L.H., Munoz, J., Zuiverloon, T., Faber-Krol, M.C., Fallaux-van den Houten, F.C., Klar-Mohamad, N., Hack, C.E., Tilanus, M.G. and Daha, M.R., 2003. Functional characterization of the lectin pathway of complement in human serum. Molecular immunology, 39(11), pp.655-668.

Rost, B., Yachdav, G. and Liu, J., 2004. The predictprotein server. Nucleic acids research, 32(suppl_2), pp.W321-W326.

148

Sacquin-Mora, S., Carbone, A. and Lavery, R., 2008. Identification of protein interaction partners and protein–protein interaction sites. Journal of molecular biology, 382(5), pp.1276-1289.

Samy, R.P., Stiles, B.G., Sethi, G. and Lim, L.H., 2017. Melioidosis: Clinical impact and public health threat in the tropics. PLoS neglected tropical diseases, 11(5), p.e0004738

Silbermann, M.H., Gyssens, I.C., Endtz, H.P., Kuijper, E.J. and Van Der Meer, J.T., 1997. Two patients with recurrent melioidosis after prolonged antibiotic therapy. Scandinavian journal of infectious diseases, 29(2), pp.199-201.

Simossis, V.A. and Heringa, J., 2005. PRALINE: a multiple sequence alignment toolbox that integrates homology-extended and secondary structure information. Nucleic acids research, 33(suppl_2), pp.W289-W294.

Schneider, M.C., Prosser, B.E., Caesar, J.J., Kugelberg, E., Li, S., Zhang, Q., Quoraishi, S., Lovett, J.E., Deane, J.E., Sim, R.B. and Roversi, P., 2009. Neisseria meningitidis recruits factor H using protein mimicry of host carbohydrates. Nature, 458(7240), pp.890-893.

Schwaeble, W., Zwirner, J., Schulz, T.F., Linke, R.P., Dierich, M.P. and Weiss, E.H., 1987. Human complement factor H: expression of an additional truncated gene product of 43 kDa in human liver. European journal of immunology, 17(10), pp.1485-1489.

Schweizer, H.P., 2012. Mechanisms of antibiotic resistance in Burkholderia pseudomallei: implications for treatment of melioidosis. Future microbiology, 7(12), pp.1389-1399.

Shalom, G., Shaw, J.G. and Thomas, M.S., 2007. In vivo expression technology identifies a type VI secretion system locus in Burkholderia pseudomallei that is induced upon invasion of macrophages. Microbiology, 153(8), pp.2689-2699.

Shapiro, E.D., Capobianco, L.A., Berg, A.T. and Zitt, M.Q., 1989. The Immunogenicity of Hemophilus influenzae Type B Polysaccharide—Neisseria meningitidis Group B Outer Membrane Protein Complex Vaccine in Infants and Young Children. Journal of Infectious Diseases, 160(6), pp.1064-1067.

Shaughnessy, J., Lewis, L.A., Jarva, H. and Ram, S., 2009. Functional comparison of the binding of factor H short consensus repeat 6 (SCR 6) to factor H binding protein from Neisseria meningitidis and the binding of factor H SCR 18 to 20 to Neisseria gonorrhoeae porin. Infection and immunity, 77(5), pp.2094-2103.

Sitthidet, C., Stevens, J.M., Chantratita, N., Currie, B.J., Peacock, S.J., Korbsrisate, S. and Stevens, M.P., 2008. Prevalence and sequence diversity of a factor required for actin- based motility in natural populations of Burkholderia species. Journal of clinical microbiology, 46(7), pp.2418-2422.

149

Smith, M.D., Angus, B.J., Wuthiekanun, V. and White, N.J., 1997. Arabinose assimilation defines a nonvirulent biotype of Burkholderia pseudomallei. Infection and immunity, 65(10), pp.4319-4321.

Smith, S., Hanson, J. and Currie, B.J., 2018. Melioidosis: An Australian Perspective. Tropical Medicine and Infectious Disease, 3(1), p.27.

Stevens, M.P., Wood, M.W., Taylor, L.A., Monaghan, P., Hawes, P., Jones, P.W., Wallis, T.S. and Galyov, E.E., 2002. An Inv/Mxi‐Spa‐like type III protein secretion system in Burkholderia pseudomallei modulates intracellular behaviour of the pathogen. Molecular microbiology, 46(3), pp.649-659.

Stevens, M.P., Stevens, J.M., Jeng, R.L., Taylor, L.A., Wood, M.W., Hawes, P., Monaghan, P., Welch, M.D. and Galyov, E.E., 2005. Identification of a bacterial factor required for actin‐based motility of Burkholderia pseudomallei. Molecular microbiology, 56(1), pp.40-53.

Siritapetawee, J., Prinz, H., Krittanai, C. and Suginta, W., 2004. Expression and refolding of Omp38 from Burkholderia pseudomallei and Burkholderia thailandensis, and its function as a diffusion porin. Biochemical Journal, 384(3), pp.609-617.

Stothard, P., 2000. The sequence manipulation suite: JavaScript programs for analyzing and formatting protein and DNA sequences.

Su, H.P., Yang, H.W., Chen, Y.L., Ferng, T.L., Chou, Y.L., Chung, T.C., Chen, C.H., Chiang, C.S., Kuan, M.M., Lin, H.H. and Chen, Y.S., 2007. Prevalence of melioidosis in the Er- Ren River Basin, Taiwan: implications for transmission. Journal of clinical microbiology, 45(8), pp.2599-2603.

Suputtamongkol, Y., Hall, A.J., Dance, D.A.B., Chaowagul, W., Rajchanuvong, A., Smith, M.D. and White, N.J., 1994. The epidemiology of melioidosis in Ubon Ratchatani, northeast Thailand. International journal of epidemiology, 23(5), pp.1082-1090.

Suputtamongkol, Y., Chaowagul, W., Chetchotisakd, P., Lertpatanasuwun, N., Intaranongpai, S., Ruchutrakool, T., Budhsarawong, D., Mootsikapun, P., Wuthiekanun, V., Teerawatasook, N. and Lulitanond, A., 1999. Risk factors for melioidosis and bacteremic melioidosis. Clinical Infectious Diseases, 29(2), pp.408-413.

Thompson, R.A. and Winterborn, M.H., 1981. Hypocomplementaemia due to a genetic deficiency of beta 1H globulin. Clinical and experimental immunology, 46(1), p.110.

Tiangpitayakorn, C., Songsivilai, S., Piyasangthong, N. and Dharakul, T., 1997. Speed of detection of Burkholderia pseudomallei in blood cultures and its correlation with the clinical outcome. The American journal of tropical medicine and hygiene, 57(1), pp.96- 99.

150

Tsang, T.Y. and Lai, S.T., 2001. A case of thoracic empyema due to suppurative melioidosis. Hong Kong medical journal= Xianggang yi xue za zhi, 7(2), pp.201-204.

Tusnady, G.E. and Simon, I., 1998. Principles governing amino acid composition of integral membrane proteins: application to topology prediction1. Journal of molecular biology, 283(2), pp.489-506.

Utaisincharoen, P., Anuntagool, N., Limposuwan, K., Chaisuriya, P. and Sirisinha, S., 2003. Involvement of beta interferon in enhancing inducible nitric oxide synthase production and antimicrobial activity of Burkholderia pseudomallei-infected macrophages. Infection and immunity, 71(6), pp.3053-3057.

Utaisincharoen, P., Anuntagool, N., Arjcharoen, S., Limposuwan, K., Chaisuriya, P. and Sirisinha, S., 2004. Induction of iNOS expression and antimicrobial activity by interferon (IFN)‐β is distinct from IFN‐γ in Burkholderia pseudomallei‐infected mouse macrophages. Clinical & Experimental Immunology, 136(2), pp.277-283.

Uversky, V.N., Oldfield, C.J. and Dunker, A.K., 2008. Intrinsically disordered proteins in human diseases: introducing the D2 concept. Annu. Rev. Biophys., 37, pp.215-246.

Vacic, V., Iakoucheva, L.M. and Radivojac, P., 2006. Two Sample Logo: a graphical representation of the differences between two sets of sequence alignments. Bioinformatics, 22(12), pp.1536-1537.

Valsalan, R., Seshadri, S. and Pandit, V.R., 2008. Melioidosis masquerading as enteric fever. Transactions of the Royal Society of Tropical Medicine and Hygiene, 102(Supplement_1), pp.S117-S118.

Van Deuren, M., Brandtzaeg, P. and van der Meer, J.W., 2000. Update on meningococcal disease with emphasis on pathogenesis and clinical management. Clinical microbiology reviews, 13(1), pp.144-166.

Vangone, A. and Bonvin, A.M., 2015. Contacts-based prediction of binding affinity in protein– protein complexes. Elife, 4.

Van Peenen, P.F., See, R., Soysa, P.E. and Irving, G.S., 1976. Seroepidemiological survey of hospital-associated populations in Colombo, Sri Lanka. The Southeast Asian journal of tropical medicine and public health, (1), pp.16-20.

Verma, A., Hellwage, J., Artiushin, S., Zipfel, P.F., Kraiczy, P., Timoney, J.F. and Stevenson, B., 2006. LfhA, a novel factor H-binding protein of Leptospira interrogans. Infection and immunity, 74(5), pp.2659-2666.

Vidor, E., 2010. Evaluation of the persistence of vaccine-induced protection with human vaccines. Journal of comparative pathology, 142, pp.S96-S101.

151

Wall, R.A., Mabey, D.C.W., Corrah, P.T. and Peters, L., 1985. A case of melioidosis in West Africa. Journal of Infectious Diseases, 152(2), pp.424-425.

Walsh, A.L., Smith, M.D., Wuthiekanun, V., Suputtamongkol, Y., Chaowagul, W., Dance, D.A., Angus, B. and White, N.J., 1995. Prognostic significance of quantitative bacteremia in septicemic melioidosis. Clinical infectious diseases, 21(6), pp.1498-1500.

Webb, D.C. and Cripps, A.W., 1998. Secondary structure and molecular analysis of interstrain variability in the P5 outer-membrane protein of non-typable Haemophilus influenzae isolated from diverse anatomical sites. Journal of medical microbiology, 47(12), pp.1059- 1067.

Weissert, C., Dollenmaier, G., Rafeiner, P., Riehm, J. and Schultze, D., 2009. Burkholderia pseudomallei misidentified by automated system. Emerging infectious diseases, 15(11), p.1799.

Wheelis, M., 1998. First shots fired in biological warfare. Nature, 395(6699), p.213.

White, N.J., 2003. Melioidosis. The Lancet, 361(9370), pp.1715-1722.

Wiersinga, W.J., Currie, B.J. and Peacock, S.J., 2012. Melioidosis. New England Journal of Medicine, 367(11), pp.1035-1044.

Wiersinga, W.J., Virk, H.S., Torres, A.G., Currie, B.J., Peacock, S.J., Dance, D.A. and Limmathurotsakul, D., 2018. Melioidosis. Nature Reviews Disease Primers, 4, p.17107.

Woehl, J.L., Stapels, D.A., Garcia, B.L., Ramyar, K.X., Keightley, A., Ruyken, M., Syriga, M., Sfyroera, G., Weber, A.B., Zolkiewski, M. and Ricklin, D., 2014. The extracellular adherence protein from Staphylococcus aureus inhibits the classical and lectin pathways of complement by blocking formation of the C3 proconvertase. The journal of immunology, p.1401600.

Woodman, M.E., Worth, R.G. and Wooten, R.M., 2012. Capsule influences the deposition of critical complement C3 levels required for the killing of Burkholderia pseudomallei via NADPH-oxidase induction by human neutrophils. PloS one, 7(12), p.e52276.

Wu, J., Wu, Y.Q., Ricklin, D., Janssen, B.J., Lambris, J.D. and Gros, P., 2009. Structure of complement fragment C3b–factor H and implications for host protection by complement regulators. Nature immunology, 10(7), pp.728-733.

Yabuuchi, E., Kawamura, Y., Ezaki, T., Ikedo, M., Dejsirilert, S., Fujiwara, N., Naka, T. and Kobayashi, K., 2000. Burkholderia uboniae sp. nov., L-arabinose-assimilating but different from Burkholderia thailandensis and Burkholderia vietnamiensis. Microbiology and immunology, 44(4), pp.307-317.

152

Yabuuchi, E., Kosako, Y., Oyaizu, H., Yano, I., Hotta, H., Hashimoto, Y., Ezaki, T. and Arakawa, M., 1992. Proposal of Burkholderia gen. nov. and transfer of seven species of the genus Pseudomonas homology group II to the new genus, with the type species Burkholderia cepacia (Palleroni and Holmes 1981) comb. nov. Microbiology and immunology, 36(12), pp.1251-1275.

Yphantis, D.A., 1964. Equilibrium ultracentrifugation of dilute solutions. Biochemistry, 3(3), pp.297-317.

Yu, Y., Kim, H.S., Chua, H.H., Lin, C.H., Sim, S.H., Lin, D., Derr, A., Engels, R., DeShazer, D., Birren, B. and Nierman, W.C., 2006. Genomic patterns of pathogen evolution revealed by comparison of Burkholderia pseudomallei, the causative agent of melioidosis, to avirulent Burkholderia thailandensis. BMC microbiology, 6(1), p.46.

Yu, C.S., Lin, C.J. and Hwang, J.K., 2004. Predicting subcellular localization of proteins for Gram‐negative bacteria by support vector machines based on n‐peptide compositions. Protein science, 13(5), pp.1402-1406.

Yu, N.Y., Wagner, J.R., Laird, M.R., Melli, G., Rey, S., Lo, R., Dao, P., Sahinalp, S.C., Ester, M., Foster, L.J. and Brinkman, F.S., 2010. PSORTb 3.0: improved protein subcellular localization prediction with refined localization subcategories and predictive capabilities for all prokaryotes. Bioinformatics, 26(13), pp.1608-1615.

Zaleska, M., Lounatmaa, K., Nurminen, M., Wahlström, E. and Mäkelä, P.H., 1985. A novel virulence-associated cell surface structure composed of 47-kd protein subunits in Yersinia enterocolitica. The EMBO journal, 4(4), p.1013.

Zhang, C., Vasmatzis, G., Cornette, J.L. and DeLisi, C., 1997. Determination of atomic desolvation energies from the structures of crystallized proteins1. Journal of molecular biology, 267(3), pp.707-726.

Zimmerman, S.M., Dyke, J.S., Jelesijevic, T.P., Michel, F., Lafontaine, E.R. and Hogan, R.J., 2017. Antibodies against in vivo-expressed antigens are sufficient to protect against lethal aerosol infection with Burkholderia mallei and Burkholderia pseudomallei. Infection and Immunity, pp.IAI-00102.

Zipfel, P.F., Jokiranta, T.S., Hellwage, J., Koistinen, V. and Meri, S., 1999. The factor H protein family. Immunopharmacology, 42(1), pp.53-60.

Zipfel, P.F. and Skerka, C., 2015. Complement: The Alternative Pathway. eLS.

153

Appendix

A. Full logo analysis of fHbps possessing conserved domains among proteins across Burkholderia species in regions predicted to be extracellularly-exposed

Note: in all appendix figures gold boxes denote residues predicted to be extracellularly exposed by Predict Protein localization server.

154

Appendix A1: Full Logo Analysis of OmpA proteins from Burkholderia group with A. baumannii

155

156

Appendix A2: Full logo analysis of Porin proteins from Burkholderia group with N. gonorrhoeae Porin B

157

158

159

160

Appendix A3: Full logo analysis of CAS transporter proteins from Burkholderia group with L. interrogans Na-K symporter

161

162

Appendix A4: Two sample logo analysis full output: EF-Tu proteins in Burkholderia versus EF-Tu proteins which are not known to bind Factor

163