Campylobacter Jejuni Survival Strategies

Home , Campylobacter jejuni, Cholera

Campylobacter jejuni Survival Strategies and Counter-Attack: An investigation of Campylobacter phosphate mediated biofilms and the design of a high-throughput small- molecule screen for TAT inhibition

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Mary R Drozd

Graduate Program in Veterinary Preventive Medicine

The Ohio State University

2012

Dissertation Committee:

Dr. Gireesh Rajashekara, Advisor,

Dr. Mo Saif, Dr. Armando Hoet, and Dr. Daral Jackwood

Copyrighted by

Mary Rachel Drozd

2012

Abstract

In these investigations we studied 1) the ability of Campylobacter to modulate its behavior in response to phosphate actuated signals, 2) the modulation of biofilm in response to phosphate related stressors, and 3) we designed and carried out a high- throughput small-molecule screen that targets protein transport via the Twin Arginine

Translocation (TAT) system. We identified that the phoX , ppk1 and ppk2 genes were key components of the phosphate response that manifested increased biofilm phenotypes, and were modulated in the presence of inorganic phosphate. We used several molecular and microbiological techniques to investigate the effect of polyP, phosphate uptake inactivation, and inorganic phosphate availability on Campylobacter’s response to phosphate stress. Additionally, we counted and measured attached biofilms, as well as measured pellicle size, biofilm shedding over the course of three days, and changes in the expression of genes known to be involved in biofilm formation phenotypes. By resolving biofilm components such as pellicles, attached cells, and shed cells we found that not only did ppk1, phoX, and ppk2 deletion affect the ability of Campylobacter to form biofilms, but biofilm components were not congruently and equally affected in each mutant. Additionally, the presence of phosphate modulated those effects both independently of and additively to gene knockouts. ii

Furthermore, we observed that biofilm components were additionally affected by biofilm age: where some components had their most robust growth on day 2, biofilm shedding and pellicle growth increased the most on day 3. This growth was not uniform for all mutants, as ppk1 biofilms generally matured more quickly than wild-type cells, but the ppk2 mutant in the presence of phosphate matured more slowly.

In our high-throughput small-molecule screen we designed and carried out a primary screen of small molecules to identify compounds that had anti-Campylobacter activity in the presence of 1mM CuSO4. To screen a greater number of compounds, this study was streamlined from a dual-plate study where each chemical was tested both in the presence and absence of copper sulfate. Our screen resulted in the identification of 680 small-molecule primary hits from the NSRB small-molecule library. These hits were identified from 11 different small-molecule libraries containing more than 50,000 compounds. Using database bioactivity results from past trials, the primary small- molecule positive hits were reduced to 476 targeted hits through in silica primary screens.

We used Golden Triangle in silica medicinal chemistry methods to identify molecules that were likely to be less suitable due to low molecular weight, interactions with solutes, and compound stability. From there, common chemical motifs were identified among the remaining 350 molecules. From these ‘chemical families’ a representative sample of each group was chosen as likely having similar chemical activity. We chose 54 chemicals as representative of 4 chemical motifs: thiourea, benzimidazoles, oxadiazoles, and acylhydrazones. The rest of the molecules were

iii

selected for greatest diversity. Using these techniques, 149 compounds have been chosen that will be used as ‘cherry pick’ hits for secondary screens in the near future.

Dedication

To Jeremy Smith: my amazing husband, editor-in-chief, physics guru, and adventurous leading man.

To Dr. Warren Dick, who took the time to teach me so much about Environmental

Science, and to Marcus Aurelius Antoninus, in gratitude for his good advice.

Acknowledgments

I would like to thank Dr. Juliette Hanson for assistance with chicken colonization studies and Dr. Jun Lin for providing chicken antimicrobial peptide, fowlicidin-1. Thank you to

Dr. Fuchs for his help with chemical analysis of the small-molecule primary hits. Thank you to Dr. Rajashekara, for supporting my ideas and growth as a scientist. I am deeply grateful to Dr. Chiang, Dr. Ren and the NSRB staff that made my small-molecule project

possible. Also, I am grateful for the support of the SEEDS grants committee and my collaborators as well as those who took the time and interest in being mentors for me both as professors and as part of my dissertation committee: Dr. Mo Saif, Dr. Armando Hoet,

Dr. Daral Jackwood. Finally I thank our department for their commitment to science as a

means of positive change.

Vita

March 1997 ...... George School, Newtown PA

May 2001 ...... B.A. Molecular Biology, Classical Studies

Hamilton College

May 2001 to 2004 ...... Laboratory Specialist, University of Virginia

May 2004 to July 2004……...... Animal Technician, Charlottesville SPCA.

July 2004 to 2007 ...... Laboratory Specialist, Virginia

Commonwealth University: Medical

College of Virginia

2007 to present ...... Graduate Research Associate, Department

of Veterinary Preventive Medicine, The

Ohio State University

vii

Publications

“Contribution of PhoX to Twin Arginine Translocation mediated Campylobacter jejuni function and resilience to Environmental Stresses” Mary Drozd, Dharanesh Gangaiah,

Zhe Liu, and Gireesh Rajashekara. Research Article, PLoS ONE 6(10): e26336.

“A quantitative polymerase chain reaction assay for detection and quantification of

Lawsonia intracellularis.” Mary Drozd, Issmat I. Kassem, Wondwossen Gebreyes,

Gireesh Rajashekara. Brief Report, JVDI, 2010

“Functional characterization of the twin-arginine translocation system in Campylobacter jejuni” Rajashekara G, Drozd M, Gangaiah D, Jeon B, Liu Z, Zhang Q. Foodborne

Pathog Dis. 2009

“Nitrotyrosylation of Ca2+ Channels Prevents c-Src Kinase Regulation of Colonic

Smooth Muscle Contractility in Experimental Colitis” Gracious R. Ross, Minho Kang,

Najeeb Shirwany, Anna P. Malykhina, Mary Drozd, and Hamid I. Akbarali. JPET, 2007.

Thomashevski A., High A.A., Drozd M., Shabanowitz J., Hunt D.F., Grant P.A., Kupfer

viii

G.M. “The fanconi anemia core complex forms 4 different sized complexes in different subcellular compartments.” J Biol Chem, 2004

Fields of Study

Major Field: Veterinary Preventive Medicine

Table of Contents

Campylobacter jejuni Survival Strategies and Counter-Attack: An investigation of

Campylobacter phosphate mediated biofilms and the design of a high-throughput small- molecule screen for TAT inhibition ...... 1

DISSERTATION ...... 1

Abstract ...... ii

Acknowledgments...... vi

Publications ...... viii

Fields of Study ...... ix

Table of Contents ...... x

List of Tables ...... xiii

List of Figures ...... xiv

Chapter 1: Introduction and Literature Review ...... 1

1.1 A brief clinical history...... 1

1.2 Clinical and Microbiological Background ...... 2

1.3 Sources of Campylobacter infection ...... 3

1.4 Microbiological Challenges Unique to Campylobacter ...... 6

1.5 The Twin Arginine Translocation System ...... 9

1.6 Development of Dynamic Biofilm Formation Theories ...... 14

1.7 Phosphate Metabolism in Campylobacter ...... 19

1.8 High-throughput screening for small-molecule inhibitors ...... 24

Chapter 2: Contribution of TAT System Translocated PhoX to Campylobacter jejuni

Phosphate Metabolism and Resilience to Environmental Stresses ...... 29

2.1: Materials and Methods ...... 29

2.2 Results ...... 40

2.3: Discussion ...... 59

Chapter 3: Modulation of Campylobacter jejuni Biofilm Growth and Stability by

Inorganic Phosphate ...... 65

3.1 Materials and Methods ...... 65

3.2 Results ...... 74

3.3: Discussion ...... 94

Chapter 4: High-Throughput Small-Molecule Screening for TAT System Inhibitors ... 104

4.1 Materials and Methods ...... 104

4.2 Results ...... 111

4.3: Discussion ...... 123

Chapter 5: Future Studies...... 129

5.1 Future Biofilm Studies ...... 129

5.2 Future Small-Molecule Studies...... 130

References ...... 132

Appendix A: Supplemental Tables and Figures ...... 145

Appendix B: Automatic Measurement Program Settings for Quantitating Adherent Cells

...... 155

Appendix C: Biomek 3000 Program for 96-well Plate Alkaline Phosphatase Assay .... 157

xii

List of Tables

Table 1: Antibiotic minimum inhibitory concentration of wild-type C. jejuni compared to the phoX mutant strain ...... 51

Table 2: qRT-PCR analysis of change in selected genes in wild type, ΔtatC, and ΔphoX strains ...... 53

Table 3: Frequency of hits for each library before and after in silica counter-screens. . 115

Table 4: Strains and plasmids used in these studies ...... 145

Table 5: Oligonucleotide primers used in this study...... 147

xiii

List of Figures

Figure 1: Model of the Campylobacter TAT system ...... 10

Figure 2: The ΔtatC mutant is attenuated in chicken colonization ...... 13

Figure 3: Model of aspects of biofilm formation ...... 14

Figure 4: The ΔphoX and ΔtatC mutants are defective in alkaline phosphatase activity . 41

Figure 5: The ΔphoX mutant is defective in poly P accumulation ...... 43

Figure 6: PhoX is necessary for survival under nutrient stress conditions ...... 45

Figure 7: Motility and biofilm formation in the ∆phoX mutant ...... 48

Figure 8: Invasion and intracellular survival of the ∆phoX mutant in INT407 cells ...... 56

Figure 9: Colonization of the ∆phoX mutant in chickens...... 58

Figure 10: Number of adherent colonies in the presence and absence of inorganic phosphate ...... 75

Figure 11: Percent adherent biofilm growth in the presence and absence of inorganic phosphate ...... 79

Figure 12: Biofilm growth at the air-liquid interface (pellicle) in the presence and absence of inorganic phosphate ...... 83

Figure 13: Biofilm shedding of Campylobacter cells ...... 85

Figure 14: CFW staining of Campylobacter strains in microaerophilic and aerobic conditions ...... 88

Figure 15: Measurement of AI-2 production by V.haryvei bioluminescence assay ...... 90

Figure 16: Transcriptional changes in biofilm genes measured by rt-qPCR ...... 93

Figure 17: Alkaline phosphatase activity assay in 96-well plates ...... 112 xiv

Figure 18: Growth inhibition of ΔtatC mutants by 1mM CuSO4 ...... 114

Figure 19: Number of small-molecule hits for each library screened ...... 117

Figure 20: Campylobacter growth in 1mM CuSO4 for high throughput small-molecule screen approximates a normal curve ...... 118

Figure 21: Golden triangle analysis of Trial 3 ...... 121

Figure 22: Thiourea motifs found in Trial 3 ...... 122

Figure 23: Alkaline phosphatase assay in various Campylobacter media ...... 150

Figure 24: Poly P assay in MH media ...... 151

Figure 25: Oxidative stress assay of ΔphoX mutant...... 152

Figure 26: Osmotic stress assay of ΔphoX in both solid (0.17M NaCl) and liquid culture

(0.25M NaCl)...... 153

Figure 27: Adherent cells of Campylobacter biofilms...... 155

Chapter 1: Introduction and Literature Review

1.1 A brief clinical history

The first mention of Campylobacter in scientific literature is widely believed to have been in 1886, when Dr. Escherich reported the presence of spiral bacteria in the colon of deceased infants (Butzler., 2004,King., 1957,Kist., 1986). Unfortunately, he dismissed the increased prevalence of these bacteria likely due to his inability to culture them.

Campylobacter returned to complete obscurity until 1909 when it was rediscovered as a vibrio-like species present in the placenta of aborted calves (Butzler., 2004,Kist., 1986).

Campylobacter was intermittently pursued as a causative agent for cattle abortion; it was not until 1947 that the first human outbreak for Campylobacter was recorded when it was identified in the blood of 13 patients (Butzler., 2004,Vinzent, R.Dumas, J. Picard,N.,

1947). From then, it was intermittently recorded as a human illness, but until the development of a method to culture Campylobacter from feces was achieved, it was only identified in the most highly infected patients (King., 1957). The technological breakthrough that allowed for Campylobacter to be cultured from feces was the dual use of 0.65 micron filtration of fecal suspension combined with the development of selective media (Dekeyser et al., 1972). The isolation of Campylobacter from stool samples vastly 1

improved the ability of clinicians and researchers to identify colonization and

Campylobacter’s sensitivity to erythromycin (Butzler., 2004, Karmali & Fleming., 1979).

Using these techniques as well as a streamlined selective media preparation,

Campylobacter was further investigated through antigenic testing (particularly agglutination testing) and was identified as a separate genus from vibrio (Butzler.,

2004,Skirrow., 1977). From these and continuing technological advancements,

Campylobacter was subsequently considered a globally endemic gastrointestinal pathogen (Karmali & Fleming., 1979).

1.2 Clinical and Microbiological Background

Campylobacter jejuni is the primary cause of Campylobacteriosis, one of the most common gastrointestinal diseases in the US (Scharff., 2010). Disease progression typically consists of abdominal pain, watery or bloody diarrhea, and fever which persist from a few days to a week (Black et al., 1988,Karmali & Fleming., 1979). It is a foodborne or water borne, Gram-negative, motile, microaerobic bacterium that causes disease by invading the epithelial cells of the host’s colon through subversion of microtubule structures on the epithelial cell surface (Blaser MJ et al., 1980, Crushell et al., 2004).

Damage to these cells disrupts the normal absorptive function of the colon and causes disease (Crushell et al., 2004). Most cases of Campylobacteriosis are self-limiting, but medical costs and losses of productivity total more than 2.4 billion a year (Centers for

Disease Control and Prevention, Division of Foodborne, Bacterial, and Mycotic Diseases, 2

National Center for Zoonotic, Vector-Borne,and Enteric Diseases., 2010). Additionally, approximately one in one thousand clinical cases may result in long term neurological defects, including Guillain-Barre syndrome (Nachamkin et al., 1998,Salloway et al.,

1996). Immune-compromised populations, which comprise an estimated 3.6% of the total population in the US, are more likely to become ill from Campylobacter and are also more likely to experience long term sequelae (Kemper et al., 2002). This number does not include one of the fastest growing populations in the US, the elderly, who are also at increased risk for food-borne illness, including Campylobacter (Gillespie et al.,

2008,Greig & Ravel., 2009). In addition to the improvements of modern medicine allowing people with chronic disease to live longer, over the next 20 years the elderly population is estimated to grow to nearly 20% of the US population (Vincent & Velkoff.,

2010). Campylobacter infections may become both more prevalent and severe as the population ages.

1.3 Sources of Campylobacter infection

Campylobacter infection is potentially conveyed to a human host by fecal-oral contamination. Although human to human transmission is possible, animals are the most important reservoirs. Campylobacter jejuni has been shown to be carried in dogs and cats; in particular puppies and kittens have been observed to have higher rates of infection (Blaser MJ et al., 1980). Additionally, migratory birds have been shown to carry Campylobacter jejuni infection, and are considered a potential reservoir and means 3

by which Campylobacter jejuni can spread geographically (Pacha et al., 1988). C. jejuni has also been shown to contaminate the feces of farm animals, including sheep, cattle, and chickens (Hermans et al., 2012). Campylobacter is usually a commensal organism in chickens and causes no signs of disease. Also, Campylobacter infection is persistent; typically chicks are infected in the first weeks of life through exposure to other animals, or to contaminated litter, or water, and the bacteria remains present in their cecum in high concentrations (up to 108 CFU/gram) until death (Hermans et al., 2012,Rajashekara et al.,

2009).

Most commonly, Campylobacter infection reaches a human host through food or water contamination: the two most common sources of food contamination are through

‘raw’ or unpasteurized milk, and from the contamination of poultry products (Hermans et al., 2012,Wood et al., 1992). In the developed world, the most common sources of

Campylobacter contamination are from poultry meat contamination (Hermans et al.,

2012). During slaughter, carcasses can be contaminated during defeathering, evisceration, and chilling stages (Hermans et al., 2012). It is estimated as much as 80% of chicken carcasses world-wide may be contaminated with Campylobacter, and 50% of

Campylobacter chicken isolates show resistance to three or more classes of drugs (Kim et al., 2000). The infective dose for Campylobacter has been shown to be as few as 103 organisms, which can easily be overlooked easily during visual carcass inspection for fecal contamination (Allos., 2001). Although traditional cooking kills Campylobacter bacteria, it can survive on under-cooked meat, cooking surfaces, and improperly stored 4

food (Centers for Disease Control and Prevention, Division of Foodborne, Bacterial, and

Mycotic Diseases, National Center for Zoonotic, Vector-Borne,and Enteric Diseases.,

2010).

In the developing world, however, Campylobacter infection more commonly occurs from the ingestion of contaminated water; the absence of water treatment significantly increases the risk of Campylobacter infection (Rao et al., 2001). In these regions, Campylobacter infection overwhelmingly infects children under 2 years of age and causes prolonged watery diarrhea (Ashbolt., 2004). Control of Campylobacter in these regions is complicated by the reinfection of water supplies by animal and human feces through ground water seepage into wells and storage tanks, as well as by direct water contamination from free-roaming animals.(Ashbolt., 2004,Rao et al., 2001).

Additionally, it has been shown experimentally that sediment or common building materials (wood, slate rock) can grow biofilms that increase Campylobacter prevalence

(Maal-Bared et al., 2012). Although Campylobacter is known to be sensitive to basic water-sanitation treatments and to have limited survival in aerobic conditions,

Campylobacter has been shown to increase its resistance to both these measures by forming heterogeneous biofilms with the protozoa and Pseudomonas species that are common to water environments (Buswell & Beyea., 1998,Trachoo & Frank., 2002).

1.4 Microbiological Challenges Unique to Campylobacter

Due to Campylobacter's low GC-content genome, it has a relatively high mutation rate. Campylobacter jejuni also readily incorporates and is promiscuous in its uptake of foreign DNA. Between these two factors, Campylobacter acquires drug resistances quickly. Drug resistance to fluoroquinolones was not detected before 1991 but within ten years over 40% of human Campylobacter isolates in the US were found to be fluoroquinolone-resistant (Nachamkin et al., 2002,Nachamkin et al., 2002,Newell et al.,

2010). In 2010, Kim et al. found 50% of Campylobacter chicken isolates resistant to three or more classes of drugs world-wide. Additionally, Campylobacter’s biomimicry with human neurons, a proposed cause of Miller-Fisher and Guillain-Barré syndromes, has slowed the development of anti-Campylobacter vaccines. This genome is thought to be a reason for Campylobacter’s frequent incompatibility with the genetic tools that are often successfully used with the more common gram-negative pathogens, including E. coli (Yao et al., 1993). The reduced availability of protein expression tools, the variability among Campylobacter strains, and semi-fastidious growth conditions may not fully explain why few toxins and molecular invasion pathways have been fully elucidated. However, these complications have undoubtedly contributed to delay a full understanding of Campylobacter pathogenesis.

Also, the animal models for Campylobacter infection are limited and do not closely parallel human disease. Although Campylobacteriosis in infant, non-human 6

primates (Macaca nemestrima) have been shown to have similar pathology to the disease in humans, expense considerations and the inherent complications of non-human primates make them unsuitable for routine Campylobacter study (Newell et al., 2010). A ferret model of Campylobacter has been used with some success, but it too has complications, including the lack of diarrhea symptoms (perhaps caused by infection being concurrent with opiate treatment), the fact that Campylobacter is endemic in laboratory ferret populations, and the cost of maintaining ferrets as lab animals (Newell et al., 2010). Recently, colostrum-deprived piglets were used in a study that significantly approximates human-like disease, which may be promising as a future study model

(Murphy et al., 2006). However, this model has many of the same expense complications that are inherent in the use of swine as lab animals. Additionally there are fewer resources, such as commercial antibodies, available that are targeted to swine protocols.

Mouse-model products, however, are highly common (Young et al., 2007).

Mouse models of Campylobacter disease have been periodically used as well.

However, results have varied significantly depending on the mouse and Campylobacter strains used (Newell et al.,2010). A recently developed combination of immune- suppressed, gnobotic mice, as well as mice colonized with ‘humanized’ flora experiments may allow for improvements in mouse models of infection (Bereswill et al., 2011).

Chickens have also been used as both a model of Campylobacter infection (day old chick) as well as a target for Campylobacter therapeutics (Newell et al ,

2010.,Rajashekara et al., 2009). Campylobacter readily infects chicks, which can be 7

highly useful in determining the infective success of various strains or, inversely, the success of vaccination trials. Day old chicks do not have a mature mucosal immune response, nor well established gut flora. This does not prevent oral inoculation with

Campylobacter strains, and it produces similar results to the inoculation of older chicks

(Hermans et al., 2012,Ringoir et al., 2007). Additionally, obtaining pathogen free animals is highly practical for day old chicks and compared to other available models, chickens can be housed efficiently (Ringoir et al., 2007). Therefore, the day old chick is widely used to determine factors likely to promote or reduce the shedding and successful infection of Campylobacter.

Campylobacter invasion of the chicken gut, however, is significantly dissimilar to human infection and only rarely causes disease (Newell et al.,2010; Young et al., 2007).

Briefly, human infection appears to involve epithelial uptake and invasion, where chicken infection appears to be primarily confined to the mucosal lining of the chicken intestine

(Hermans et al., 2012,Young et al., 2007). Chickens’ inflammatory responses are quickly controlled by C. jejuni, which is not the case in human infections (Hermans et al.,

2012,Young et al., 2007). Therefore, the value of chicks as a disease model for human pathology is limited. However, since the majority of human infection is the result of poulutry contamination, this model can also be used to elucidate mechanisms of commensal colonization in avian reservoirs and thus provide avenues for further research

(Hermans et al., 2012).

In our study, we utilized three recent advances in Campylobacter research as well as more general drug discovery methodologies that have the potential to further unveil some of Campylobacter’s pathogenic mysteries and provide much needed advancement for anti-Campylobacter chemicals.

1.5 The Twin Arginine Translocation System

One of the factors necessary for Campylobacter adaptation and successful colonization is the Twin Arginine Translocation (TAT) system (Rajashekara et al., 2009).

The TAT system is an inner membrane translocase that transports proteins folded in the cytoplasm across the periplasmic membrane. The TAT system is found in most prokaryote and archeal organisms. In Campylobacter, the TAT system has three proteins:

TatA, TatB, and TatC (Figure 1).

Figure 1: Model of the Campylobacter TAT system

Substrates are transported through TatA, which forms a homo-polymer, membrane-spanning pore. Although it has been shown that TatA can self-assemble, TatC may have a role in initiating TatA assembly and pore localization (Berthelmann et al.,

2008). Since folded proteins require more volume than unfolded proteins, even in its smallest arrangement, the TAT pore is several times larger than the cytoplasmic membrane pore of the Sec system, which is a translocase commonly used by bacteria to transport unfolded proteins across the cytoplasmic membrane (Gohlke et al., 2005,Sardis

& Economou., 2010). Therefore, the plastic quality of the TatA pore is essential to prevent ion leakage across the periplasmic membrane (Gohlke et al., 2005). The TatA

multi-mer pore varies in size to accommodate substrate volume and transports proteins solely by proton motor force (Gohlke et al., 2005). TatB and TatC recognize substrates by an N-terminal targeting sequence (S/TRRxFLK) and chaperone the substrate proteins to the TatA pore (Berthelmann et al., 2008). In E. coli, TatB was shown to be non- essential for TAT function (Barrett & Robinson., 2005). This is consistent with the observation that many Gram-positive bacteria and archea, which commonly transport more substrates than Gram-negative bacteria via the TAT system, function while only possessing genes for TatA and TatC (Thomas & Bolhuis., 2006,Yen et al., 2002). There are distinct differences in the arrangement of TAT genes, even among Gram-negative bacteria. In E. coli, where the TAT system has been extensively studied, there are five

TAT genes (tatA/B/C/D/E) and an estimated 29 substrates (Lee et al., 2006). Our in silica analysis of TAT recognition sequences predicted that Campylobacter may have more than a dozen TAT substrates, suggesting that the TAT system may be a bottleneck in the transport of membrane, periplasmic, and secreted proteins (Rajashekara et al., 2009).

Recent studies have confirmed that Mrfa, TorA, CueA, FdhA, the YedY homologue

0379c, and PhoX are also transported by the TAT system in Campylobacter (Hall et al.,

2008,Hitchcock et al., 2010,Wosten et al., 2006). Although no Campylobacter toxins have been identified as TAT substrates, TAT substrate proteins, including fumarate reductase (Mrfa) and copper resistance (cueA), have been shown to be important for virulence (Hall et al., 2008,Hitchcock et al., 2010). Also, in Pseudomonas the TAT system has been predicted to transport two phospholipase C proteins, which hydrolyze 11

lung surfactants (De Buck et al., 2008). Similarly, in E. coli TAT inhibition has been shown to indirectly reduce secretion of Shiga toxin (De Buck et al., 2008). Additionally, it has been shown that proteins lacking the twin-arginine recognition motif may

'hitchhike' across the membrane by binding to a TAT-targeted protein, suggesting that some TAT substrates are yet to be identified (Rodrigue et al., 1999). These yet unidentified substrates may account for the observation of the multiple functional difficulties in the Campylobacter tatC mutant that have not been fully explained

(Rajashekara et al., 2009).

Although the C. jejuni TAT system is highly conserved between strains, the

C. jejuni TAT system has fewer components (tatA/B/C) than E. coli , and they are arranged differently in the genome (Lee et al., 2006). Additionally, the substrates which are targeted to the TAT system vary by organism; for instance the alkaline phosphatase is targeted to the TAT system in Campylobacter, but in E. coli it is not targeted to the TAT system (van Mourik et al., 2008). The organism specificity of the TAT system makes it a potent target for antimicrobial therapies. Most strains of Campylobacter are likely to be affected by a TAT inhibitor, while other enteric bacteria will remain unaffected.

When the tatC gene is deleted from C. jejuni, the TAT secretion system is inhibited. In chickens, this strain’s virulence and persistence were attenuated

(Rajashekara et al., 2009). Three day old chicks infected with the tatC deletion mutant stopped shedding ΔtatC after 14 days, and eliminated ΔtatC colonization from their

cecum 21 days after infection (Figure 2). In constrast, chickens colonized with wild-type

C. jejuni remain colonized and shed bacteria until slaughtered.

Figure 2: The ΔtatC mutant is attenuated in chicken colonization

Furthermore, the ΔtatC mutant had reduced motility as well as increased susceptibility to commonly used antibiotics, nutrient starvation, and osmotic stress

(Rajashekara et al., 2009). The ∆tatC mutant was also deficient in biofilm formation

(Rajashekara et al., 2009). Electron microscopy showed that the ΔtatC mutants had truncated and absent flagella in comparison to both wild-type and complemented strains.

It is clear that the TAT system is essential for pathogenesis and a choice target for anti-

Campylobacter therapies. In addition, there is no equivalent transport system in animals, which diminishes the possibility of a TAT-specific antimicrobial therapy having non- specific reactions in humans or animals (Yen et al., 2002).

1.6 Development of Dynamic Biofilm Formation Theories

Over the past 10 years, progress in inter-bacterial communication research has contributed significantly to a highly useful model of biofilm formation; biofilms are considered to be highly developed communities where environmental signals as well as inter-bacterial signals modulate formation and dissipation of biofilm structures

(Figure 3).

Figure 3: Model of aspects of biofilm formation

Although there are limitations to this model (for instance, critics have cited the absence of an independent genetic pathway for biofilm formation), this model is an extremely useful tool (Monds & O’Toole., 2009). It gives a framework to concepts of heterogeneity of cellular formations, inter-bacterial and even inter-species cooperation, and it proposes the existence of effector signals that can modulate biofilm structures. In short, the current biofilm model both considers the organism’s behavior in a biofilm and supposes the biofilm itself will act as a sort of ‘bacterially-based organism’. Through the development of this model, biofilms have been shown to confer significant protection to diverse environmental stressors including aerobic stress, pH sensitivity, and antibiotic sensitivity (Fux et al., 2005,Monds & O’Toole., 2009,Romeo., 2008). Also, it has been shown that biofilm expression can be modulated by stresses; in particular, aerobic stress has been shown to enhance the growth of C. jejuni biofilms (Reuter et al., 2010).

Also, it has been shown that the presence or absence of nutrients can have species specific effects on the formation of biofilms. In particular, phosphate has been shown to both instigate biofilm formation in Pseudomonas and disperse biofilms in Vibrio species

(Danhorn et al., 2004,Monds & O’Toole., 2009,Sultan et al., 2010). Although the genetic involvement in these decisions has been observed, particularly identifying mutants that have attenuated biofilm formation, the mechanisms for these changes are often inferred and imperfectly understood (Monds & O’Toole., 2009). One of the reasons for this is that environmental signals are prone to elicit both primary and secondary responses. Furthermore, quorum sensing studies have suggested that inter-bacterial 15

communication is important to several biofilm forming species, including Campylobacter

(Reeser et al., 2007). Therefore, in addition to secondary responses, we may also speculate that the quorum sensing responsive genes and proteins may further modulate biofilm formation.

In Campylobacter, quorum sensing occurs via the AI-2 signaling molecules that are produced by the lux operon (Elvers & Park., 2002). These molecules are a byproduct of nucleoside and amino acid synthesis, but they can have an additional function as an inter-bacterial signaling molecule (Wang et al., 2005). Originally discovered in bioluminescent Vibrio harveyi, AI-2 has been found to be highly conserved in gram- negative bacteria, including Campylobacter (Elvers & Park., 2002). Although critics have argued that in its exponential growth phase Campylobacter has not yet shown clear transcriptional changes in the presence of AI-2 molecules nor has a receptor for AI-2 been identified, they did observe that increased AI-2 alone results in increased motility phenotypes (Reeser et al., 2007). An additional study showed that an AI-2 containing supernatant did rescue a ∆luxS reduced biofilm phenotype (Elvers & Park., 2002,Holmes et al., 2009,Reeser et al., 2007). The idea that quorum sensing is involved in different aspects of biofilm formation is not unique to Campylobacter. Quorum sensing has also been shown to impact biofilm formation, in the gram-negative model organism E. coli,

Vibrio species (both V. cholera and related marine species), and in Pseudomonas, where biofilms are heavily involved in the bacteria’s devastating pathogenic impact. In V. cholera, quorum sensing is also directly involved in pathogenesis and the regulation of 16

virulence factors (Miller et al., 2002). Therefore, it is only practical that we approach

Campylobacter quorum sensing as an aspect of biofilm development and global regulation.

Although this task may appear daunting due to the complexity of biofilm regulation and its protean responsiveness to environmental changes, inroads have been made in identifying characteristics of biofilm structures as well as in identifying some of the genes that modulate biofilm formation. In particular, it has been recently proposed that Campylobacter forms three types of biofilms (flocs, adherent, air-liquid). These biofilms have been shown to be variably modulated in deletion mutants of protein glycosylation biosynthesis (pglH), lipo-oligosaccharide biosynthesis (neub1), phosphate acetyltransferase (cj0688), capsular polysaccharide formation/export (kpsM), and flagella formation (fliS, maf5) (Joshua et al., 2006) . Additionally, it has been observed that lipo- oligosaccharide (LOS) modifications, particularly the inner core modification that results from the deletion of waaF and lgtF genes result in increased biofilm as well as increased reactivity to CFW (calcofluor white) (Naito et al., 2010). CFW binds to β1-3 and β1-4 carbohydrates and can be visualized by UV light (Naito et al., 2010). In Salmonella, changes in these polysaccharides have been linked to changes in biofilm expression as well as increased sensitivity to environmental stressors (Solano et al., 2002). Increased

CFW reactivity has also been observed in a spoT deletion mutant, which suggests a connection between dynamic changes in LOS and awareness to environmental stress

(McLennan et al., 2008,Naito et al., 2010). The waaF gene has been identified as under 17

the direct transcriptional regulation of PhoR/S, the Campylobacter homologue to the two- component phosphate sensor and response operon PhoR/B, which suggests a connection between changes in biofilm-affecting polysaccharide genes and phosphate metabolism

(Wosten et al., 2006).

Compared to planktonic cells, Campylobacter biofilms are less vulnerable to antibiotic sensitivity and detergent efficacy, and have increased survival in non-host environments (Fux et al., 2005,Joshua et al., 2006,Romeo., 2008a,Trachoo & Frank.,

2002). Also, biofilm formation has been shown to increase in the presence of aerobic stress, depletion of poly-P, and phosphate starvation (Candon et al., 2007,Drozd et al.,

2011,Gangaiah et al., 2010,Reuter et al., 2010). The study of Campylobacter biofilms is particularly engaging because the increased durability conferred by biofilm formation suggests an explanation for the unlikely contrast between the relative fragility of

Campylobacter and its success as a gastro-intestinal pathogen. Closer examination of the pathogen’s sensitivity has shown that compared to other pathogens, Campylobacter has few or no obvious toxins and is comparatively more sensitive to osmotic stress, desiccation, oxidative stress, and other sources of environmental insult (McMahon et al.,

2007,Mihaljevic et al., 2007,Wesche et al., 2009). And yet, 0.8% of the US population is afflicted with Campybacterosis per year (Centers for Disease Control and Prevention,

Division of Foodborne, Bacterial, and Mycotic Diseases, National Center for Zoonotic,

Vector-Borne, and Enteric Diseases., 2010). One of the ways that Campylobacter may

overcome its fragility in the face of environmental hostility is its ability to form protective biofilms.

1.7 Phosphate Metabolism in Campylobacter

Although complete elucidation of the phosphate modulation is beyond the scope of this work, we are particularly interested in both Campylobacter biofilms and phosphate metabolism’s role in the biofilm life-cycle. In particular, we observe alkaline phosphatase, which catalyzes the hydrolysis of phosphate groups from organo-phosphate macromolecules. This enzyme is central to phosphate uptake, such that its activity has been accepted as a general measurement of phosphate-status, particularly in marine bacteria (Jermy., 2009,Sebastian & Ammerman., 2009). The regulation of phosphate- sensing and alkaline phosphatase activity in Campylobacter species is thought to have evolved separately from those of E. coli and other B. subtilis, but is similar to those found in V. Cholera, Pseudomonas, as well as in marine and soil bacteria, where phosphate acquisition is a limiting nutrient and therefore nutritionally prioritized (Sebastian &

Ammerman., 2009,Zaheer et al., 2009). Among those difference is that the alkaline phosphatase is targeted to the TAT system in Campylobacter, but in E. coli it is targeted to the Sec system (van Mourik et al., 2008). Unlike E. coli, which uses a PhoA alkaline phosphatase containing zinc and magnesium ions in its active site, Campylobacter uses a

PhoX alkaline phosphatase that uses calcium ions in its active site. Additionally, these metallo-proteins require distinct folding conditions due to the ionic strength of their 19

respective metallo-enzymes (Waldron K.J. et al., 2009). Therefore, PhoA alkaline phosphatases are transported via the Sec system and folded in the periplasm, but PhoX alkaline phosphatases are folded in the cytoplasm where they are sequestered from stronger Irving-Williams series metallo-enzymes that would out-compete calcium ions for the appropriate binding site and distort protein folding (Sebastian & Ammerman.,

2009,Waldron K.J. et al., 2009). Once folded in the cytoplasm, PhoX alkaline phosphatases are transported to the periplasm via the TAT system where they are active after post-translational modification in the membrane fraction (Sebastian & Ammerman.,

2009,van Mourik et al., 2008).

Phosphate regulation has been implicated in the regulation of virulence in gastrointestinal pathogens, such as E. coli, where it is involved in both resilience to host- immune defenses and adhesion expression (Crepin et al., 2011). Additionally, switching

Pseudomonas aeruginosa from high-phosphate to low-phosphate was sufficient to induce virulence to C. elegans, a nematode worm used as a Pseudomonas aeruginosa model organism (Zaborin et al., 2009). Transcription comparison of low-phosphate vs. high phosphate Pseudomonas revealed changes in both the phoB and quorum sensing associated genes (Zaborin et al., 2009). The Pseudomonas phoB deletion mutant likewise failed to induce lethality in C.elegans, which suggests that virulence genes may be directly controlled by phosphate regulation (Zaborin et al., 2009). An additional study further suggested that phosphate may be a super-controller of virulence as well: normally opiates can induce Pseudomonas virulence in nutrient deficient conditions, leading to 20

killing in the C. elegans virulence model, increased phosphate alone is sufficient to prevent induction of virulence genes (Zaborin et al., 2012).

phoX genes found in V. cholera are genetically similar to those found in

Campylobacter. In V. cholera, phosphate availability and response is known to control environmental adaptations, including biofilm regulation, adjustment of cyclic-di-GMP levels, aerobic-stress, virulence gene regulation, heat-stress tolerance, stringent response, and flagella function (Pratt et al., 2009,Pratt et al., 2009,Pratt et al., 2010,Pratt et al.,

2010,Rashid et al., 2000,Silby et al., 2009). Regulation of inorganic phosphate has been implicated in the ability of Cholera to transition between marine and gastrointestinal lifestyles (VanBogelen et al., 1996). Like V. cholera, Campylobacter is known to be a gastro-intestinal pathogen that infects hosts from fecal contamination of water which would likewise necessitate a transition from marine and gastrointestinal lifestyles (Blaser et al., 1980). Unlike V. cholera, however, Campylobacter reacts differently to external phosphate concentrations; where V. cholera transitions towards a motile life style in the absence of phosphate, our preliminary studies show that Campylobacter does the reverse and increases biofilm in low phosphate conditions. The anfractuous nature of these phenotypes may suggest that Campylobacter’s response to phosphate concentrations is adapted to its pathogenic niche and is therefore critical to understanding its stress adaptations. The differences between Campylobacter phosphate regulation and those of other enteric, food-borne pathogens may provide insight to whether Campylobacter

biofilms have unique attributes to compensate for its comparative vulnerability to environmental stressors.

In our preliminary studies, we found that a ∆phoX (alkaline phosphatase) deletion mutant, which is deficient in organic phosphate uptake, has an increased biofilm phenotype that is completely rescued by the addition of inorganic phosphate. Since environmental inorganic phosphate is typically low, alkaline phosphatases are necessary for phosphate and ATP homeostasis (Lamarche et al., 2008, Rao et al., 2009). Also alkaline phosphatase is important for the uptake of extracellular phosphate; its activity is considered a reliable measure of environmental phosphate (Jermy., 2009). We found that the phoX deletion mutant showed a statistically significant decrease (P< 0.05) in poly P accumulation in the stationary phase when compared to the wild-type, and the poly P levels were similar to the poly P levels of the tatC deletion mutant (Drozd et al., 2011).

Poly P has been shown to insulate cells in an alkali environment and to be necessary for long-term growth and survival in Salmonella (Kornberg et al., 1999,Rao et al., 2009). Poly P has been shown to be synthesized by Ppk1 which adds phosphate groups in a highly compact ‘chain’ of phosphate residues linked by phosphoanhydride bonds (Kornberg et al., 1999). These chains form dense granules (poly P) in

Campylobacter cells, which act as a bank of phosphates for metabolic use, including the synthesis of GTP (Rao et al., 2009). The phosphate sequestered as poly P can be accessed by the cell through ppk2 catalysis of poly P, which preferentially removes inorganic phosphate from the polyP chain and adds a phosphate group to GDP (Zhang et 22

al., 2002). In 2009, Gangaiah et al documented that a ppk1-deletion (Δppk1) caused virulence defects in Campylobacter invasion of INT-407 cells and colonization defects in day-old chicks (Gangaiah et al., 2009). Additionally, both ppk1 and ppk2 deletion mutants produced enhanced total biofilm as well as stress survival defects (Drozd et al.,

2011,Gangaiah et al., 2009,Gangaiah et al., 2010).

Despite the importance of poly P to bacterial survival, little is known about the regulation of ppk1, which synthesizes polyP from inorganic phosphate, and ppk2, which catabolizes phosphate groups from polyP for metabolic use. Also, there are few studies that elucidate how poly P levels instigate cellular protective measures. When we compared the biofilm formation of the ppk1 deletion mutant to ∆phoX and wild-type biofilm, the ∆ppk1 mutant increased biofilm responded to inorganic phosphate by further increasing the total biofilm. Wild-type cells were indifferent to the presence of inorganic phosphate (Drozd et al., 2011). Although inorganic phosphate is necessary for the ppk1- mediated formation of poly P, it appeared that biofilm formation was modulated by the ppk1 mutant’s inability to sequester phosphate as poly P (Rao et al., 2009).

In this study, we have examined more closely the role of poly P depletion (∆ppk1) and repletion (∆ppk2). We have also examined how low-phosphate conditions affect the formation and growth of adherent biofilms, examined the amount of viable cells shed from growing biofilms, and examined the growth of air-liquid (pellicle) biofilms. We have also compared these mutants in their initiation of quorum sensing signals and

changes in CFW (calcofluor white)-positive outer membrane composition as well as screened them these mutants for changes in biofilm formation implicated genes.

1.8 High-throughput screening for small-molecule inhibitors

As discussed in section 1.4, due to its low GC-content genome, Campylobacter has a relatively high mutation rate. C. jejuni also readily incorporates foreign DNA and is promiscuous in its uptake. Between these two factors, Campylobacter acquires drug resistances quickly. Additionally, the withdrawal and limitation of antibiotic use in animals is likely to increase in response to evidence that links antibiotic use to the increased resistance in both animals and humans (Landers et al., 2012). Between these two factors, there is an urgent need for the development of anti-Campylobacter control measures. Although progress has been made in Campylobacter vaccine development, and Salmonella-vectored Campylobacter proteins (CjA and Omp18/CjaD) have been shown to reduce bacterial loads, none of the proposed vaccines have yet become commercially available (Burrough et al., 2011,Hermans et al., 2011). Another route of antimicrobial agent discovery is using high-throughput screening of small molecules to discover compounds that inhibit a specific protein interaction (Schreiber., 2000). These compounds are low-molecular weight, can pass through cellular membranes, are screened to inhibit a specific molecular interaction in one or few species (Mitchison.,1994).

The development of reliable pipetting and plate reading automation in the 1990s spurred the development of the ‘high-throughput’ screen (Inglese et al., 2007). It uses 24

improvements in cell-growth miniaturization (such as 96-well and 384-well plates) to rapidly screen 10,000 or more chemicals a day for bioactivity (Inglese et al., 2007).

Although these libraries were originally designed for the pharmacological industry and biotechnology development, they have since been additionally purposed to allow academic investigators to conduct primary research (ICCB Longwood., 2012,NSRB.,

2011). Small-molecule libraries are currently commercially available; interest in small molecules for the discovery of biodefense-improving therapeutics has resulted in 11 libraries in the US (NSRB., 2011). Locally, The National Screening Laboratory for the

Regional Centers of Excellence in Biodefense and Emerging Infectious Disease (NSRB) is administered by the New England Regional Center of Excellence for Biodefense and

Emerging Infectious Diseases, for which Campylobacter is a National Institute of Allergy and Infectious Disease (NIAID) category B pathogen (Department of Health and Human

Services National Institutes of Health., 27 Feb. 2012). The NSRB library was “enriched for complex heterocyclic compounds and compounds of higher molecular weight (We favored an average mw of ~350-400.)”, which is considered within a target range of physiological parameters common to potential ‘drug-like’ characteristics and solubility

(Zhang,. 2011). These libraries included sources such as NIH human drug trial candidates, and in silica designed commercial libraries which design small molecules based on medicinal chemistry criteria (ICCB Longwood., 2012).

A typical screen consists of 30,000 to 100,000 small molecules, which are applied to a final concentration between 12.5ug/mL and 0.63ug/mL with the sample 25

concentration varying both within and between compound libraries. Most plates contain between 300 and 350 compounds each, and include in plate negative controls as empty wells. A preliminary screen is performed on a 5,000 compound library series, containing compounds with an increased incidence of positive results, as a test of methods. Also, during this part of the screening process, the risk of false-negative compounds and improvements to the screening process may be assessed. Due to the high number of compounds tested, repeat and duplicate tests are ideally sharply limited. For this reason, assay criteria are evaluated extensively to minimize signal variation, optimize assay clarity, and to streamline assay procedures (NSRB., 2011,Zhang et al., 1999). The measurement used in our study to quantify these criteria is the z-score measurement. The z-score measures assay fitness by comparing the difference in standard deviation between assay controls to their relative signals (Z’= 1- 3neg3pos negpos, where is the mean and  is the standard deviation), and assumes a normal distribution of hits from the small molecules (Zhang et al., 1999). The resultant z-score and its interpretation as a forecaster of assay success is highly dependent on the type of assay performed as well as on practical, logistical concerns for assay implementation such as the availability of reagents (NSRB., 2011,Zhang et al., 1999).

The successful use of high-throughput small-molecule screens for the discovery of bioactive molecules has been described in two species of interest. In both cases, small molecules were passively absorbed into the bacteria, even when the bacteria had successfully invaded eukaryotic cells (Felise et al., 2008,Hung et al., 2005). One of the 26

first bacterial organisms to be screened for small-molecule anti-virulence targets was V. cholera, which used a ctx promoter controlled tetracycline gene to screen for small molecules that conferred tetracycline sensitivity (Hung et al., 2005). In V. cholera, this technique identified Virstatin, a small-molecule that inhibits the transcription of the regulator ToxT in classic V. cholera strains. When Virstatin was administered orally to mice infected with V. cholera, four logs fewer bacteria were recovered (Hung et al.,

2005,Shakhnovich et al., 2007). This study demonstrates three qualities which are highly desirable in small bioactive molecules: 1) they are passively and efficiently absorbed into the bacterial cytoplasm, 2) they have a highly specific mode of action, and 3) because of their specificity, they are likely to be less toxic than many broad-spectrum bioactive molecules. In Salmonella, the type III secretion system (T3SS) has been screened for a small-molecule inhibitor drug candidate using a fluorescence-based screen that specifically targeted non-growth inhibiting targets (Felise et al., 2008). Similarly, small molecules were counter-selected to identify a molecule that was both specific in its mode of action and also did not have a toxic effect on Salmonella growth. In addition to facilitating molecular studies, low toxicity may also have advantages in reducing selective pressure by only affecting the bacteria when it is expressing virulence-related genes being more specific than convention antibiotics and only affecting one or few species (Waldor., 2006). These measures have been suggested as potentially delaying the development of wide-spread bacterial resistance to a small-molecule developed for therapeutic ends (Waldor., 2006,Zaheer et al., 2009). 27

Like the TAT system, the type III secretion system (T3SS) is conserved across bacterial genomes, but does not have an eukaryotic homologue and is not essential to bacterial growth. Therefore, both the T3SS and the TAT are good drug candidates for similar reasons; the specific inhibition of these transport systems will have few side effects in humans and animals, will inhibit pathogenesis, and can be targeted to the

Campylobacter TAT system. For these reasons, we designed a TAT system targeted small-molecule screen. This screen uses the increased sensitivity to copper in the absence of TAT substrate CueA to identify compounds which prevent Campylobacter jejuni growth in the presence of 1mM Copper sulfate. CueA is a copper oxidase in C. jejuni, which both acquires copper for use in metallo-enzymes necessary for bacterial survival, and manages excess copper which is toxic to Campylobacter (Hall et al., 2008).

In the absence of TAT inhibition, 1mM copper sulfate has no effect on Campylobacter growth. We used a 1mM copper sulfate survival assay to screen for molecules that inhibited the TAT system and thereby prevented growth in 1mM copper sulfate. Primary small-molecule hits will be counter-screened in the absence of copper sulfate: small molecules that prevent growth without copper sulfate will be deprioritized as non-specific inhibitors of Campylobacter growth.

Chapter 2: Contribution of TAT System Translocated PhoX to Campylobacter jejuni Phosphate Metabolism and Resilience to Environmental Stresses

Chapter 2 has been published in PLoS One as “Contribution of TAT system translocated

PhoX to Campylobacter jejuni phosphate metabolism and resilience to environmental stresses” Mary Drozd, Dharanesh Gangaiah, Zhe Liu, Gireesh Rajashekara*

2.1: Materials and Methods

Bacterial strains, plasmids, and culture conditions.

Bacterial strains and plasmids used in this study are listed in Table 4. C. jejuni strain 81-176 (WT), a highly invasive strain originally isolated from an outbreak associated with raw milk, was used to generate the phoX deletion mutant (Korlath et al.,

1985). C. jejuni strains were routinely grown on Mueller-Hinton broth (MH; Oxoid) microaerobically [(85% N2 (v/v), 10% CO2 (v/v) and 5% O2 (v/v)] in a DG250

Microaerophilic Workstation (Microbiology International) at 42 ºC. MH agar plates were

supplemented with Campylobacter selective supplement (SR117E, Oxoid) when isolating

C. jejuni from chicken feces and organs during the chicken colonization study. For growth curve and stress survival assays, C. jejuni was grown microaerobically in MH broth with appropriate antibiotics at 42ºC with shaking at 200 rpm. E. coli DH5α was used for plasmid propagation and cloning purposes and was routinely cultured on Luria-

Bertani (LB) medium at 37ºC overnight. Growth media was supplemented with appropriate antibiotics, chloramphenicol (20 µg ml-1 for E. coli; 10 µg ml-1 for

Campylobacter), kanamycin (30 µg ml-1) and zeocin (50 µg ml-1), where necessary.

Construction of the phoX deletion mutant.

Cloning and other molecular biology techniques were performed according to

Sambrook & Russell (Sambrook & Russell., 2001). Oligonucleotides were designed using Vector NTI® software (Invitrogen, Carlsbad, CA) and commercially synthesized by Integrated DNA Technologies (Skokie IL). All the oligonucleotides used in the present study are listed in Table 5. Masterpure® DNA purification kit and Fast-Link

DNA ligation kit were purchased from Epicentre (Madison, WI). Restriction enzymes were purchased from Promega. QIAquick® PCR purification kit and QIAprep® spin mini prep kit for plasmid isolation were purchased from Qiagen (Valencia, CA). Zero background cloning vector, pZErO-1, and E. coli DH5α competent cells were purchased from Invitrogen.

Deletion of the phoX gene (CJJ_0181) was achieved by double crossover homologous recombination using a suicide vector containing approximately 1 kb of homologous sequences on either side of the phoX gene as described previously

(Rajashekara et al., 2009). Briefly, the phoX along with 1 kb flanking region on either side of the target gene was amplified by PCR using PhoX F and PhoX R primers

(described in Table 5) from C. jejuni 81-176 genomic DNA. The amplified PCR product was ligated into pZErO-1 to generate plasmid pZero1-phoX. Inverse PCR was performed on pZero1-phoX using PhoX INV-F and PhoX INV-R primers to delete majority of the phoX coding sequence. The kanamycin cassette from pUC4K was cloned into the inverse

PCR product, the resulting suicide vector, designated pZero1-ΔphoX, was electroporated into C. jejuni 81-176 as described (Fields & Thompson., 2008,Karlyshev & Wren.,

2005,Rajashekara et al., 2009). Recombinant colonies were selected on MH agar plates containing kanamycin, kanamycin resistant colonies were streak purified and one such mutant, designated ∆phoX, was used for further studies. The deletion of the phoX gene was confirmed by PCR.

Complementation of the ∆phoX mutant.

The complementation of ∆phoX mutant was accomplished by the insertion of a wild type phoX using pRRC integration vector (Karlyshev & Wren., 2005). The coding region of phoX and its ribosome binding site was amplified by PCR using primers PhoX

COMP F and PhoX COMP R. Restriction site (BamHI) was included in each primer to 31

facilitate cloning. Following digestion with BamHI, the PCR product was ligated into the pRRC vector. The resulting construct was electroporated into C. jejuni ΔphoX::kan, and putative complemented clones were recovered on plates containing kanamycin and chloramphenicol. Insertion of the phoX gene in the rRNA spacer region was confirmed by PCR. Further, the constitutive expression of phoX was confirmed by phosphatase assay. One of the confirmed, complemented clones was designated phoXc and was used for further analysis.

Alkaline phosphatase assay.

Alkaline phosphatase activity was determined as previously described (Wosten et al., 2006). Briefly, the ∆phoX, ∆ppk1, ∆tatC, phoXc, and WT 81176 strains were grown overnight on MH plates with appropriate antibiotic selection. The origins of these clones is described in Table 4. Alkaline phosphate activity was measured in common

Campylobacter growth media (MH Broth, Bolton Broth, Brucella broth, and MEM) and wash conditions (no wash or wash with 50mM MOPS buffer) to optimize alkaline phosphatase induction was established (Figure 23). The cultures were gently collected from the MH plate, washed in minimal essential medium (MEM), resuspended in MEM and incubated at 42 °C microaerobically with shaking for 2 hours. Cultures were then centrifuged for 10 minutes at 7000 g and supernatant was removed. Cells were gently washed with 50mM MOPS buffer (pH 7.4) (Sigma), OD600 was measured. The cells were pelleted, supernatant was removed, and cells were resuspended in PNPP buffer with 32

2mM PNPP (Sigma) and incubated at 37 °C. Absorption at 550 nm and 420 nm were taken using a spectrophotometer and the phosphatase activity was calculated as described previously (Brickman & Beckwith., 1975). The average and standard deviation of three replicate samples for each strain were calculated and the experiment was repeated three times.

Isolation/Detection of poly P.

Poly P was extracted using glassmilk and quantified using toluidine blue O as described earlier (Candon et al., 2007). Poly P was quantified from mid-log and stationary phase cultures by measuring the absorbance ratio at 530 to 630 nm spectrophotometrically using appropriate concentrations of phosphorous standard (Sigma

Aldrich). The experiment was performed a total of three times under rich media (MH) and minimal media (MEM) conditions.

Nutrient downshift assay.

The role of phoX in C. jejuni survival under nutrient downshift was assessed using

MEM with glutamine (Gibco 11095) or without glutamine (Sigma, M2279) and in the presence or absence of 1mM Pi (Inorganic Ventures) as described previously. In experiments where Pi was added to the media, the final concentration of Pi was 1mM.

This amount is similar to a previous study by Wosten et al (Wosten et al., 2006) where concentrations above 0.4mM show greater than 90% inhibition of alkaline phosphatase 33

activity. However the alkaline phosphate is induced in MEM which has low Pi concentration. Therefore, here we used MEM and 1mM Pi. Briefly, mid-log-phase cultures of the WT, ∆phoX, and phoXc, strains were pelleted, washed twice and resuspended in MEM with or without glutamine or Pi, and the OD600 was adjusted to

0.05. The suspensions were then incubated microaerobically at 42 °C with shaking at 200 rpm. At different time points, 100 μl samples of 3.5mL cultures were serially diluted (10- fold) in respective MEM media and plated onto MH agar in triplicate. The plates were incubated microaerobically, and the number of CFU ml-1 was calculated. The experiment was performed three times.

Osmotic stress response assay.

Survival of C. jejuni phoX deletion mutant in the presence of osmotic stress was tested as described previously (Candon et al., 2007,Gangaiah et al., 2009). To assess the osmotic stress survival in liquid culture, bacterial strains were grown to mid-log phase, adjusted to an OD600 of 0.05 in MH broth with and without 0.25 M NaCl, and incubated microaerobically at 42 °C for 48 hours with shaking at 200 rpm. A 100 µl sample of the culture was removed at different time points, serially diluted (10-fold), and plated on MH agar plates in duplicate. The plates were incubated microaerobically and CFU were determined. To assess the ability the ∆phoX mutant to tolerate osmotic stress on a solid medium, the WT, ∆phoX, and phoXc strains were grown to mid-log phase, serially diluted (10-fold), and 10 µl of diluted culture was spotted onto MH agar plates containing 34

either 0.17 M NaCl or 0.17M NaCl and 1mM Pi. Plates were incubated microaerobically at 42 °C for 2 days, the growth of C. jejuni was visualized and photographed. The experiment was repeated three times.

Oxidative Stress response assay.

To determine oxidative stress response, WT, ∆phoX, and phoXc strains were grown overnight on MH agar with the appropriate antibiotics at 42 °C under microaerobic conditions. Cells were harvested the next day and 100 µL of bacterial

8 culture containing 5x10 CFU/mL (OD600 of 0.05) were spread on MH plates with appropriate antibiotics in the presence or absence of 1mM Pi. A 5 mm well was cut into the middle of each plate and filled with 30 µL of 20 mM paraquat or 0.3% H2 O2. Plates were incubated for 24 and 48 hours under microaerobic conditions and the zone of inhibition was measured and photographed. The experiment was performed a total of three times.

Motility and biofilm assays.

The C. jejuni wild type 81-176, ∆phoX, phoXc, ∆ppk1, and ∆ppk1c strains were tested for motility in semisolid MH medium plates containing 0.4% agar in the presence or absence of 1mM Pi. Cultures were grown on MH agar plate under microaerobic conditions at 42° C for 48 hours. Absorbance of each culture was adjusted to an OD600 of

0.05 and 2 μl of each culture was stabbed onto the surface of the motility plate. Plates 35

were incubated at 42º C under microaerophilic conditions. Motility phenotypes were assessed after 24 hours and 48 hours following inoculation. The motility assay was performed three times.

Static biofilm formation was assessed in borosilicate tubes as described previously (Gangaiah et al., 2009,McLennan et al., 2008). Briefly, overnight grown cultures of C. jejuni strains were diluted with MH broth with or without Pi to an OD600 of

0.05. Two milliliters of diluted culture was incubated at 42°C microaerobically for 3 days without shaking. Biofilms were visualized by staining with 250 µl of 1% (w/v) crystal violet for 15 min, rinsed 3 times with double distilled water. After third rinse, vials were photographed then quantified by measuring the absorbance at 570 nm after dissolving in

2ml DMSO for 24 hours. The biofilm assay was performed three times.

Antimicrobial susceptibility testing.

Susceptibility to azithromycin, ciprofloxacin, erythromycin, tetracycline, florfenicol, nalidixic acid, telithromycin, clindamycin, and gentamicin was determined by using Sensititre® susceptibility plates for Campylobacter (TREK Diagnostic Systems,

West Sussex, UK). Briefly, one hundred microliters of log-phase grown cultures adjusted to an OD600 of 0.05 in MH broth was added to each well in the Sensititre® susceptibility plate and the wells were covered using the perforated adhesive seal. The plates were incubated microaerobically at 42 ºC for 24 hours and the minimal inhibitory concentration (MIC) was recorded. Results were read following the manufacturer’s 36

instructions and interpreted according to MIC interpretive guidelines by Clinical

Laboratory Standards Institute. In addition, the following antimicrobials were also tested individually as described previously; polymixin B, cholic acid, taurocholic acid, deoxycholic acid, and an antimicrobial peptide of chicken origin, fowlicidin-1

(Rajashekara et al., 2009). One hundred microliters of each culture, grown and diluted as described for the Sensititre® susceptibility plate assay, was added to serially diluted (2- fold) antimicrobials in a 96 well microtitre plate, mixed, and the plates were incubated microaerobically at 42 ºC for 24 hours. The MIC was determined as the lowest concentration showing complete inhibition of visible growth. The susceptibility testing was repeated 3 times and the mean MIC (µg ml-1) was calculated.

Quantitative RT-PCR.

Quantitative RT-PCR (qRT-PCR) was performed, targeting key genes involved in stringent response, spoT; post transcriptional global regulator, csrA; poly P associated enzymes, ppk1 and ppk2; phosphate metabolism, phosR, pstS, pstC; oxidative stress, cjj0379, cjj1374, ahpC, sodB, and the multidrug resistance efflux pump gene, cmeC

(Chaveerach et al., 2003, Garénaux et al., 2008, Lin et al., 2002,Wosten et al., 2006).

Total RNA was extracted from log-phase grown bacterial cultures using RNeasy Mini Kit

(Qiagen). The RNA concentration and purity was determined using NanoDrop ND-1000 spectrophotometer (Wilmington, DE). cDNA synthesis was carried out using

SuperScript® III First-Strand Synthesis SuperMix (Invitrogen). Gene specific primers 37

were designed to amplify the above mentioned genes along with rpoA or 16SrRNA

(internal control) using Beacon Designer 7.0 (Palo Alto, CA). Primers were obtained commercially from IDT-DNA and are described in Table 5. qPCR was performed using

SensiMixPlus® SYBR RT-PCR Kit (Quantace, Norwood, MA) in a Realplex2

Mastercycler (Eppendorf, Westbury, NY). The relative levels of expression of genes were normalized with either rpoA or 16SrRNA amplified from the corresponding sample. The difference in expression of the genes was calculated using the comparative threshold cycle (CT) method to yield fold-difference in transcript levels (Livak 2008.). The RT- qPCR was performed in duplicate and the assay was performed a total of four times.

INT407 invasion and intracellular survival assay:

Invasion and intracellular survival assays were performed as described previously

(Candon et al., 2007,Gangaiah et al., 2010). Each well of a 24-well tissue culture plate was seeded with 1.4×105 INT407 cells in MEM with 10% fetal bovine serum (FBS) and incubated for 18 hours at 37 °C with 5% CO2. C. jejuni strains were grown to mid-log phase in MH broth microaerobically, the cells were pelleted at 5,000 g for 10 min, washed twice with MEM containing 1% FBS, resuspended in MEM to an OD600 of 0.02

(1.5 x 107 cells) and used for infection. INT407 cells were infected with a multiplicity of infection 1:100 for invasion and intracellular survival assays. For infection, 1 ml of bacterial cell suspension was pipetted on to INT407 cells, centrifuged at 1000 g for 3 minutes and incubated for 3 hours. For determining invasion, after 3 hours of incubation 38

with bacteria, cells were treated with gentamicin (150 µg/mL) and incubated for additional 2 hours. After 2 hours of incubation, the infected cells were rinsed with MEM three times, lysed with 0.1% (v/v) Triton-X 100, serially diluted in MEM and plated on

MH agar in duplicate to determine CFU. To assess intracellular survival, following 2 hours of gentamicin treatment the infected cells were washed with MEM three times and covered with MEM containing gentamicin (10 µg/mL) and incubated for 24 hours. After

24 hours of incubation, the infected cells were washed with MEM, lysed with 0.1% (v/v)

Triton-X 100, serially diluted in MEM and plated on MH agar in duplicate to determine

CFU. The invasion and intracellular assays were performed three times.

Chicken colonization study

Chicken colonization studies were performed as described previously

(Rajashekara et al., 2009). Briefly, day-old broiler chicks (n=6 for each group) from a local hatching facility (Food Animal Health Research Program, OARDC, Wooster, OH) were inoculated orally with 103, and 105 CFU of the C. jejuni WT and ΔphoX strains in

200 µl of PBS (pH 7.4). After 7 days post-inoculation, the chicks were euthanized. The ceca and feces were collected aseptically, weighed, homogenized, serially diluted in PBS

(pH 7.4), and plated on MH agar containing Campylobacter selective supplement. Plates were incubated at 42 °C microaerobically and CFU per gram of tissues were determined.

Statistical analysis. 39

Statistical significance of data generated in this study was determined using one- way analysis of variance (ANOVA) followed by Tukey’s HSD (Honestly Significant

Difference) test or Student’s t-test (paired 2-tailed). P≤ 0.05 (α level) was considered statistically significant.

2.2 Results

The ppk1 is not a primary effector of alkaline phosphatase activity.

We found that ΔphoX mutant was significantly defective (P< 0.01) in alkaline phosphatase activity (27.0 mU) compared to wild type (123.0 mU) (Figure 4).

Preliminary studies suggested that alkaline phosphatase activity could be measured with least background activity when cultures were incubated in MEM followed by washing with MOPS buffer before the assay (Figure 23). Further, complementation of the ΔphoX mutant restored the alkaline phosphatase activity similar to wild-type levels (Figure 4).

Similarly, the ΔtatC mutant also showed similar alkaline phosphatase activity (33.5 mU) compared to the ΔphoX mutant (Figure 4). This confirms previous results that PhoX is the only alkaline phosphatase in C. jejuni and it is solely transported through the TAT system (van Mourik et al., 2008). The ΔtatC mutant, though not statistically significant, showed slightly increased alkaline phosphatase activity compared to the phoX deletion mutant.

Figure 4: The ΔphoX and ΔtatC mutants are defective in alkaline phosphatase activity

It is well known that poly P levels are influenced by a wide range of stress responses, including low-nutrient stress (Candon et al., 2007,Gaynor et al., 2005).

Further, spoT deletion mutants have both shown diminished accumulation of poly P, decreased transcription of ppk1 as well as increased transcription of pstS and pstC, which are regulated as part of the PhoR/S phosphate uptake regulon (Gangaiah et al., 2009).

Therefore, we investigated whether there was a direct relationship between alkaline phosphate activity and poly P accumulation. The alkaline phosphatase level in the ppk1 deletion mutant was 92.7 mU, slightly but significantly decreased compared to the wild- 41

type (P< 0.05) (Figure 4). This suggests that ppk1 expression does not severely affect alkaline phosphatase expression; they are potentially part of a separate, although intersecting function, of the phosphate utilization pathway.

Poly P accumulation is reduced in both the tatC and phoX deletion mutants.

The phoX deletion mutant showed a statistically significant decrease (P< 0.05) in poly P accumulation during stationary phase in the minimal media compared to the wild- type. The poly P levels of ΔphoX mutant were similar to the poly P levels of the tatC deletion mutant (Figure 5). Both ΔphoX and ΔtatC had an average accumulated poly P level of approximately 21 nM poly P/mg of total protein and 25 nM poly P mg-1 of total protein respectively, compared to the wild-type cells (37 nM poly P mg-1 of total protein)

(Figure 5). However, ppk1 deletion mutant showed a larger accumulation defect with only 14 nM poly P mg-1 total protein (P< 0.01) (Figure 5).

Figure 5: The ΔphoX mutant is defective in poly P accumulation

Difference in poly P accumulation between ∆ppk1 and ∆phoX mutants is likely due to ability of the ∆phoX mutant to ameliorate the poly P defect with Pi obtained by other sources, such as through phosphonate catabolism. C. jejuni has been shown to catabolize phosphonate (Hartley et al., 2009). However, ∆ppk1 mutants cannot synthesize poly P even in the presence of other Pi sources including phosphonates. This result suggests that interruption of phoX may be sufficient to cause the poly P defects seen in the ∆tatC mutant as both ΔtatC and ΔphoX mutants showed similar poly P levels.

Although, the ΔtatC mutant was defective in growth in rich media, the ΔphoX mutant grew similar to wild-type (data not shown, (van Mourik et al., 2008)). Similarly the poly

P accumulation was significantly decreased in the phoX deletion mutant in rich media during stationary phase (Figure 24). There was no difference in the accumulation of poly

P between mutant strains and the wild-type in the log phase (data not shown). Although alkaline phosphatase activity is necessary for Ppk1-mediated accumulation of poly P, the

∆ppk1 mutant had only a small defect in alkaline phosphatase activity. Alkaline phosphatase activity may be pleiotropically mediated by the function of ppk1 or by other parts of the phosphate pathway under ppk1 regulation.

The phoX deletion results in sensitivity to nutrient but not osmotic and oxidative stress response.

It is known that inorganic polyphosphate (poly P) is important to stress responses

(nutrient, osmotic, osmotic and oxidative), and specifically the C. jejuni ∆ppk1 and

∆ppk2 mutants are sensitive to nutrient and osmotic stresses (Gangaiah et al.,

2009,Gangaiah et al., 2010,Rao et al., 2009). Since the ΔphoX mutant was defective in poly P accumulation; we investigated whether phoX deletion mutant is sensitive to various stresses. We found that the phoX deletion mutant had a significant defect

(P<0.05) in nutrient stress survival after 24 hours of incubation in minimal media without glutamine (Figure. 6a). This survival defect was restored in the complemented strain

(Figure 6a). Similarly the phoX deletion mutant shows a statistically significant reduction in survival after 48 hours. The survival defect was even greater (approximately 3 logs), compared to wild type cells, at 60 hours (Figure. 6a). 44

Figure 6: PhoX is necessary for survival under nutrient stress conditions

Inorganic phosphate is required for deadenylation of glutamine synthetase, which is required for the synthesis of glutamine (Stadtman., 2001). Since the phoX mutant is defective in Pi generation, we hypothesized that supplementation of glutamine might correct the nutrient downshift defect. The survival of the ΔphoX mutant was similar to

WT in minimal media containing 2mM glutamine even after 60 hours of incubation

(Figure. 6b). Further, addition of 1mM Pi to the media also corrected the survival defect of the ΔphoX mutant even the absence of glutamine (Figure 6c). Since Pi is required for deadenylation of glutamine synthetase, which is required for synthesis of glutamine, the nutrient survival defect is likely due to insufficient Pi in the ΔphoX.

The ΔphoX mutant had an osmotic stress and oxidative stress response similar to the wild-type strain even in the presence of Pi (Figure 25 and 26). However, the qRTPCR results indicated a down-regulation of katA suggesting that other mechanisms other katA may play a role in oxidative stress response of C. jejuni (Hwang et al., 2011). Though the

ΔphoX mutant showed consistently increased resistance to osmotic stress, it was not significant.

The phoX deletion does not affect motility but has enhanced biofilm

Poly P/inorganic phosphate stores have been linked to changes in both motility and biofilm (Gangaiah et al., 2009,Rashid et al., 2000). Additionally, regulation of Pi has been implicated in the ability of V. cholera to transit between marine and gastrointestinal lifestyles (Pratt et al., 2009). Therefore, we investigated whether phoX deletion would induce changes in motility and biofilm formation. The phoX deletion mutant did not show any motility defect on a semisolid agar. The motility of the phoX mutant was comparable to the wild type stain (Figure 7a). Further, addition of Pi, though increased the motility, there was no significant difference between the WT and ∆phoX mutant. Similarly there 46

was no significant difference in the motility of the ∆ppk1 mutant compared to WT

(Figure 7a). On the other hand, the phoX deletion resulted in significantly enhanced biofilm formation (P<0.01) compared to the wild-type C. jejuni and the complementation of the phoX deletion restored biofilm to wild type levels (Figure. 7b-c). This result is surprising, since phosphate starvation in V. cholera and Pseudomonas is thought to reduce biofilm formation (Monds et al., 2001,Pratt et al., 2009).

Figure 7: Motility and biofilm formation in the ∆phoX mutant

Figure 7 continued.

Since phoX is closely associated with Pi levels, it is suggested that phosphate levels are an environmental indicator for biofilm regulation (Pratt et al., 2009). Therefore,

we further tested whether biofilm formation was affected by Pi alone. We found that the

∆phoX mutant’s increase of biofilm can be rescued with the addition of 1mM Pi (P<

0.01), strongly suggesting that the increase in biofilm is affected by the decreased availability of phosphate in the phoX deletion mutant (Figure 7d). However, the wild-type strain did not show any changes in biofilm formation after the addition of 1mM Pi

(Figure 7d), nor did the addition of 1mM Pi affected the biofilm formation in the complemented strain (Figure 7d). Consistent with our earlier finding, the Δppk1 mutant also has an enhanced biofilm phenotype; however, addition of 1mM Pi further increased the biofilm formation (P<0.05) in the ∆ppk1 mutant (Figure 7c-d) (Gangaiah et al.,

2009).

The phoX deletion mutant has increased resistance to antimicrobials

Both the ppk1 and tatC deletion mutants have been shown to have increased sensitivity to antibiotics (Gangaiah et al., 2009,Rajashekara et al., 2009). Since the phoX deletion mutant has a defect in poly P accumulation, we further investigated whether this would also result in susceptibility to antimicrobials. In contrast to ppk1 and tatC deletion mutants, we found that the ΔphoX mutant has increased resistance to some common antimicrobials, including tetracycline, nalidixic acid and, to a lesser degree, to ciprofloxacin (Table 1).

Table 1: Antibiotic minimum inhibitory concentration of wild-type C. jejuni compared to the phoX mutant strain

Antibiotic ΔphoX (fold WT phoXc change)§±stderr

Azithromycin 0.05(1.5)±0.02 0.03 0.03±0.02 Ciprofloxacin 0.19 (3.0)±0.08 0.06 0.06 Erythromycin 0.25 (1.4)±0.14 0.19±0.09 0.12±0.09 Gentamicin 1.5 (1.5)±0.58 1.0 1.0

Tetracycline 0.31 (5.2)±0.19 0.06 0.06±0.04 Florfenicol 1.5 (3.0)±0.58 0.5 1 Nalidixic Acid 16 (4.0/)±9.2 4 4

Telithromycin 1.0 (2.0)±0.58 0.5 0.5 Clindamycin 0.38 (2.0)±0.19 0.19±0.09 0.19

§fold change is the quotient of ∆phoX/WT resistance for a given antibiotic. Where standard deviation measurements are absent, measurements for all tests were the same and standard deviation is zero.

The phoX deletion mutant had 5, 4, 3-fold greater resistance to tetracycline, nalidixic acid, florfenicol and ciprofloxacin than wild-type (Table 1). The complementation with the wild type copy of the phoX restored susceptibility to wild-type

levels (Table 1). Since biofilms have been implicated as a contributor to increased resistance to antibiotics, it is possible that the increased biofilm activity in the ΔphoX mutant is contributing to this phenotype. In previous studies, tetracycline, and nalidixic acid have been known to kill biofilm-forming cells less efficiently than non-biofilm forming cells (Romeo., 2008). This may suggest that the increased biofilm phenotype seen in ∆phoX may be a mechanism of innate increased antibiotic resistance.

The phoX deletion results in transcriptional changes in key genes involved in phosphate uptake and stress responses

We used qRT-PCR to investigate how changes in alkaline phosphatase and inorganic phosphate starvation, caused by the phoX deletion mutant, affected the transcription of genes that are commonly associated with environmental stress response or are believed to be part of the phosphate regulon. We found that ppk1 was down regulated 2.1-fold in the ∆phoX mutant compared with a 7-fold down regulation of ppk1 in the tatC mutant (Table 2). However, no changes in the transcription of ppk2 were observed in the phoX deletion mutant although the tatC mutant showed 4-fold down- regulation (Table 2).

Table 2: qRT-PCR analysis of change in selected genes in wild type, ΔtatC, and ΔphoX strains

Gene/ORF Fold change in Fold change in ΔphoX ΔtatC compared compared to to WT§ WT§ ahpC (CJJ_0298) No change -9.3 ahpC (CJJ_0356) No change -16.2 CJJ_0379 No change -18.9 ppk-2 No change -4.4 CJJ_1374 -10.4 -15.7 cmeC -5.8 No change csrA 2.6 No change katA -8.5 -19.2 proP 4.3 -27.5 pstC No change Not tested pstS -2.4 Not tested ppk1 -2.1 -7.1 spoT -2.9 5.5 sodB No change -29.3 §The difference in gene expression was determined by the threshold cycle(CT) method, and the assay was repeated three times with two replicates each time for each sample. Data represent the mean relative fold change in expression.

ProP has been known to confer osmotic protection (Pichereau et al., 2000). The expression of proP was increased (4.3-fold) in the phoX deletion mutant, compared to a

>25-fold down regulation in the tatC deletion mutant (Table 2). However, the ∆phoX 53

mutant has a slightly increased, though not significant, resistance to osmotic stress (Fig.

S3b-c) (Rajashekara et al., 2009). While phoX deletion increases proP expression, there may be other TAT-substrates which have additive and pleiotropic effects that result in the overall down-regulation of proP in the tatC deletion mutant. Also, we found that

CJJ_1374, a VacJ homolog, known to be upregulated during oxidative stress, was down- regulated in both ΔphoX and ΔtatC mutants (Table 2;Garénaux et al., 2008). While

CJJ_1374 was 1.5-fold more down-regulated in the tatC mutant than the ∆phoX mutant, the ∆tatC mutant has an increased sensitivity to oxidative stress and the ∆phoX mutant does not (Table 2;Rajashekara et al., 2009). Similarly, we saw an 8.5-fold down regulation of katA in the ∆phoX mutant, compared to a 19.2-fold down-regulation in the

∆tatC mutant. Additionally, cmeC was down-regulated (5.8-fold) in the ∆phoX mutant.

However, the ∆phoX mutant was resistant to certain antimicrobials. It is possible that the antimicrobial resistance is mediated by changes other than CmeABC efflux pump, such as increased biofilm formation (Figure 7;Naito et al., 2010).

We measured whether pstS and pstC, which are concurrently regulated by the

PhosR/S regulon, would be transcriptionally affected by phoX deletion. We found a 2.4- fold down regulation of pstS and no change in the pstC gene (Table 2). This result was unexpected; we predicted that pstS would be upregulated during the putative phosphate depletion resulting from the deletion of phoX. This result may indicate that an indirect effect of phoX deletion, such as nutrient stress contributing to pstS regulation (Kim et al.,

2000,Taschner et al., 2004). We also observed a 2.9-fold down-regulation of spoT, a 54

primary effecter of ppGpp degradation. In addition, a post transcriptional global regulator, csrA, was upregulated by 2.6-fold (Fields & Thompson., 2008,Lamarche et al.,

2008).

Survival of the ΔphoX mutant is slightly diminished in INT407 cells

To investigate if phoX is involved in virulence-associated phenotypes, we examined whether the ∆phoX mutant could invade and survive within INT407 human intestinal epithelial cells. The phoX deletion mutant was defective in invasion compared to wild-type cells by 4.4 fold (Figure 8a). The ∆phoX mutant’s intracellular survival was also similarly reduced by 4.0 fold compared to wild-type cells in INT407 cells while the complementation restored the defect to WT levels (Figure 8a-b). However, the ΔtatC mutant had more than 7000-fold invasion defect in INT407 cells and no intracellular bacteria were recovered 24 hours post infection suggesting a severe intracellular survival defect (Figure 8a-b). This suggests that phoX activity is not essential for successful invasion and intracellular survival; however, it should be noted that invasion and intracellular assays were performed using rich media (MEM supplemented with fetal bovine serum) where Pi was abundant and readily available even to the ∆phoX mutant.

Since ∆phoX mutant, though defective in generation of Pi from complex organophosphate ester, is not defective in uptake of free Pi because it still carries intact Pi uptake pst genes (Lamarche et al., 2008,Wosten et al., 2006).

Figure 8: Invasion and intracellular survival of the ∆phoX mutant in INT407 cells

The ΔphoX mutant shows diminished ability to colonize day-old chicks

To assess the contribution of phoX to C. jejuni host colonization, we tested the ability of the ∆phoX mutant to colonize day-old chicks. We found that there was 0.8 log

difference (P< 0.05) in the number of wild-type (average CFU 2.34 x 107 g-1) and phoX deletion mutant (average CFU 3.17 x 106 g-1) bacteria recovered from the ceca 7 days after chicks were orally infected with 103 CFU of each strain (Figure 9). Similarly, there was approximately 1 log difference (P< 0.05) in the number of C. jejuni wild-type

(average CFU 3.41 x 108 g-1) and phoX deletion mutant (average CFU 5.29 x 107 g-1) recovered from chick ceca when the inoculation dose was 105 CFU/chick (Figure 9).

Chicks inoculated with the phoX deletion mutant also had fewer bacteria in the feces compared to wild type. On an average there were approximately 1.1 and 1.2 log fewer bacteria in feces (P< 0.05) in chicks inoculated with 103 and 105 CFU of the mutant

(average CFU/g 1.38 x 104; 3.53 x 105), respectively, compared to chicks inoculated with similar amounts of the wild-type strain (average CFU/g 2.50 x 105; 4.71 x 106) (Figure 9).

These results suggest that there is a small but consistent defect in colonization caused by phoX deletion. It is possible that the effects of phoX deletion are ameliorated by using inorganic phosphate produced by alkaline phosphatases native to the chicks' gastrointestinal tract (Vaishnava & Hooper., 2007).

Figure 9: Colonization of the ∆phoX mutant in chickens

2.3: Discussion

In this study we explored how one of the substrates of the TAT system, alkaline phosphatase, contributes to C. jejuni patho-physiology. Also we have expanded a model for poly P mediated responses in C. jejuni (Candon et al., 2007,Gangaiah et al., 2009).

From our results, we can conclude the following about the phosphate metabolism of C. jejuni; alkaline phosphatase (phoX) is necessary for extracellular Pi acquisition and the effects of low inorganic phosphate affect a significant part of stress metabolism through poly P and biofilm formation. In particular, phoX is important for survival in stationary phase in the nutritionally limited conditions, where it affects poly P metabolism. Further, without the ability to synthesize inorganic phosphate from extracellular sources, via alkaline phosphatase, poly P accumulation is reduced. This suggests that extracellular sources of Pi are necessary for efficient poly P synthesis; C. jejuni is partially reliant on the environment to accumulate this important stress response and metabolic mediator.

Since little is known about poly P regulation, understanding the origin of the inorganic phosphate that make up the poly P macro-molecule is novel.

Both deletion of tatC and ppk1 have been shown to have pleiotropic effects on bacteria including C. jejuni (Candon et al., 2007,De Buck et al., 2008,Gangaiah et al.,

2009,Rajashekara et al., 2009). Deletion of tatC can result in diverse phenotypes which can be a direct effect, due to the defect in translocation necessary proteins, or can be an indirect effect for which the mechanisms are not yet understood. Similarly, Ppk1 has been linked with regulation of several genes in P. aeruginosa, P. gingivalis and E. coli 59

(Rao et al., 2009). Specifically, microarray analysis of ppk1 deletion mutants in P. aeruginosa and E. coli revealed up-regulation of over 250 genes and down-regulation of more than 450 genes (Rao et al., 2009). Thus, poly P and its associated enzymes may be identified as “global regulators”. Our finding show that how the deletion of a single, specific variable (phoX) contributes to C. jejuni resilience to environmental stresses and provides an insight into understanding the complex mechanisms behind poly P and TAT system mediated stress responses.

The ΔphoX mutant has a 3 log nutrient stress defect that is rescued by the addition of

2mM glutamine (Figure 6), suggesting that the ΔphoX nutrient stress phenotype is caused by a defect in glutamine metabolism. Interestingly, the nutrient downshift defect was also rescued by the addition of 1mM Pi even in the absence of glutamine. Since it is known that inorganic phosphate is crucial for the deadenylation of glutamine synthetase, it is likely that the nutrient stress defect is associated with defective inorganic phosphate in the ∆phoX mutant (Stadtman., 2001). Further, it is reasonable to assume that defective translocation of alkaline phosphatase may be responsible for the nutrient stress phenotype seen in the ΔtatC mutant since both ΔphoX and ΔtatC mutants appear to have similar nutrient downshift defects (Figure 6, Rajashekara et al., 2009).

The ΔphoX has an increased biofilm phenotype that can be rescued with the addition of 1mM Pi (Figure 7d). This suggests that reduced Pi concentration, consequently reduced poly P, may be a factor that allows C. jejuni to sense an environment hostile to growth and initialize defensive measures. This could promote C. jejuni survival on 60

surfaces—which are typically low phosphate environments—and improve the pathogen's ability to endure until it can reach a suitable host. The biofilm results may explain the phoX deletion mutant’s decrease in sensitivity to tetracycline (Table 1), a drug that is known to have reduced penetration of biofilms (Romeo., 2008).

The change of the ∆phoX biofilm phenotype by addition of 1mM Pi to wild-type comparable amounts (Figure 7d) suggests that the biofilm phenotype is not a direct result of phoX deletion, but may be result of cellular response to the Pi depleted conditions that the phoX deletion mutant created. In contrast, in the Δppk1 mutant biofilm formation was further increased with the addition of 1mM Pi. Since, the ∆ppk1 mutant is defective in the synthesis of poly P, unlike the ∆phoX mutant, the different outcome in the ∆ppk1 mutant in response to Pi is likely due to pleiotropic effects that the ppk1 deletion has on the C. jejuni (Candon et al., 2007,Gangaiah et al., 2009). This response suggests that although low levels of inorganic phosphate may be sufficient to increase biofilm, poly P and other factors modify this response. Since the Δppk1 mutant showed increased biofilm in the presence of Pi, we hypothesize that there is a phosphate-related biofilm response parallel to the poly P mediated response and perhaps tied to other environmental stressors.

Our qRT-PCR data shows that a slight down-regulation of ppk1 transcription in the

ΔphoX mutant in the stationary phase. This result agrees with the poly P accumulation defect that we observed. Additionally, we saw that the ΔtatC mutant had a greater down- regulation of Δppk1 than ΔphoX, this may suggest that an additional TAT substrate contributes to poly P synthesis. Although we found that phoX deletion results in 61

transcriptional down-regulation of oxidative stress genes (katA, CJJ_0379, vacJ homologue CJJ_1374, ahpC) these transcriptional changes did not result in reduced oxidative stress survival phenotype (Figure 25). These genes are universally more down- regulated in the ΔtatC mutant where the mutant has a significantly increased sensitivity to oxidative stress (Rajashekara et al., 2009). However, when grown in minimal media, phoX deletion in general caused fewer transcriptional changes (data not shown); there was a 2.4-fold upregulation of ppk1 and 3.5-fold decrease in ppk2 transcription compared to wild-type. This may suggest that ppk1 and ppk2 regulation is pleiotropically affected by nutritional stressors.

We also conclude that phoX is unlikely to be more than a peripheral mechanism for the survival defect in INT407 cultured cells as well as in vivo chicken colonization. This agrees with previous research on phoB deletion mutant in chickens; the host’s gastrointestinal surfaces have their own alkaline phosphatase enzymes, allowing the

ΔphoX mutant to compensate for its alkaline phosphatase deficiency by uptake of host derived free phosphate (Sultan et al., 2010,Vaishnava & Hooper., 2007). Less inorganic phosphate availability and other factors, such as differences in temperature and oxygen levels in feces compared to ceca, could explain an increased survival defect of ΔphoX mutant in feces compared to ceca.

In C. jejuni we found that deletion of phoX resulted in increased biofilm formation.

However, in V. cholera, low intracellular phosphate results in decreased biofilm formation that is regulated by PhoB (Sultan et al., 2010). This is surprising because 62

biofilm inhibition in V. cholera is thought to be part of a cellular transition to an aquatic lifestyle. Similar to V. cholera, C. jejuni contamination of water is a known source of infection (Sultan et al., 2010). C. jejuni's increase of biofilm in response to deletion of phoX and presumably diminished intracellular phosphate is more similar to plant and soil dwelling pathogens such as A. tumefaciens (Danhorn et al., 2004); this may suggest that the phosphate stressor response that C. jejuni has evolved is more tuned towards survival on an exposed surface rather than a marine environment (Karatan & Watnick., 2009).

Although V. cholera and Pseudomonas have evolved similar phoX genes, there may be crucial phenotype differences that are likely unique to C. jejuni and have an impact on its cell physiology and survival (Monds et al., 2001,Sultan et al., 2010). For instance, phosphate starvation in Campylobacter increases biofilm rather than decreasing it, and phosphate starvation did not increase oxidative stress sensitivity, or cause a change in motility (Monds et al., 2007,Pratt et al., 2009).

In summary, these finding reinforce an important, central theme to alkaline phosphatase and phosphate utilization in C. jejuni. It is nearly universal that the ability to use inorganic phosphate from extracellular environments is critical for bacterial physiology. The molecular response to this information however is highly dependent on the bacterial species and its environment. In particular, we were intrigued by our finding that while phosphate depletion alone was sufficient to increase biofilm, it was modulated by the deletion of ppk1 as well. As ppk1 and its synthesis-product, poly P, are known as part of Campylobacter stress resilience, which has been implicated in biofilm formation, 63

we further pursued these results to identify more specifically the effects that phosphate and polyP depletion had on biofilms.

Chapter 3: Modulation of Campylobacter jejuni Biofilm Growth and Stability by Inorganic Phosphate

Mary Drozd, Kshipra Chandrashekhar, Zhe Liu, Gireesh Rajashekara*

3.1 Materials and Methods

Growth of Campylobacter and Vibrio strains:

All of the bacterial strains used in this study are described in Table 4. C. jejuni 81-

176 is a highly invasive strain originally isolated from diarrheic patients (Korlath et al.,

1985). C. jejuni strains were cultured on Mueller-Hinton (MH) medium under microaerophilic conditions (85% N2, 10% CO2, 5% O2) at 42°C for 24 h. MH agar plates were supplemented with Campylobacter selective supplement (SR117E; Oxoid, Lenexa,

KS) periodically to confirm purity of wild-type strain. Deletion mutants (∆phoX, ∆ppk1, and ∆ppk2) were grown on MH plates containing 30µg/mL kanamycin as a selective agent. Complemented strains (phoX+, ppk1+, and ppk2+) were grown on MH plates containing 30ug/mL kanamycin and chloroamphenicol 20µg/mL as selective agents.

Vibrio haryvei strains BB170 and BB150 are originally described in Bassler et al.,

1997. V. harveyi growth media (LM Media) contained 10g tryptone, 5g Yeast Extract, and 20g NaCl to 1L of ddH20. This media was autoclaved, allowed to cool and was stored at room temperature. Frozen cells in 20% sterile glycerol (-70ºC) were spread on room temperature LM plates and incubated at 28ºC overnight for growth. These cells were resuspended in LM buffer and 1mL was aliquoted into tubes containing 80% sterile glycerol for a final concentration of 20%. Tubes were mixed immediately prior to storage at -70ºC. 100µL aliquots of frozen cells were inoculated in LM and AB media as needed. AB media contains 0.3 M NaCl, 0.05 M MgSO4, and 0.2% vitamin-free casamino acids (Difco), adjusted to pH 7.5 with KOH in ddH2O. The medium was sterilized, allowed to cool, and 10 ml of sterile 1 M potassium phosphate (pH 7.0), 10 ml of sterile 0.1 M L-arginine, 20 ml of sterile glycerol, 1 ml of l0 ug/ml riboflavin, and 1ml of 1 mg/ml thiamine was added per liter.

Shedding of viable C. jejuni cells from Biofilm:

Cultures were plated overnight on MH plates with appropriate antibiotics or on

CSS selective plates for wild-type cells. Cells were scraped into room temperature MH broth and gently resuspended with pipetting. Cultures were diluted to an OD600 of 0.06 and 2mL of each culture were inoculated into 4mL glass vials containing 200µL 100mM

Pi or sterile ddH20 (control). Each condition was performed in duplicate. Cultures were incubated for 0, 24, 48 and 72 hours under microaerophilic conditions at 42ºC. This 66

protocol was performed previously in Reuter et al., 2010 with changes as subsequently described. Broth was removed by gentle pouring and biofilms were rinsed once with

4mL MH broth with gentle pouring and then incubated for 4 hours in fresh MH broth.

100µL supernatant from each culture was removed from each vial and diluted appropriately as determined by test dilution series. 100µL of diluted cells was plated on

MH plates and grown for 36 hours under standard growth conditions. Cells were counted and CFU/mL was calculated. Each vial was plated in duplicate and 3 independent experiments were performed for all conditions.

Preparation of Crystal Violet Slides:

Cultures were plated overnight on MH plates with appropriate antibiotics or on

CSS selective plates for wild-type cells. Plates were scraped into room temperature MH broth and gently resuspended by pipetting. Cultures were diluted to an OD600 of 0.06 and

25mL of each culture was inoculated into 50mL conical tubes (Fisher scientific cat # 14-

432-23). Sterile water (control) or 100mM inorganic phosphate was added to each vial for a 1mM final concentration. Clean glass microscope slides (size 3x1”x1mm, Gold Seal microslides, Clay Adams cat #3010) were autoclaved and cooled overnight. Immediately prior to the experiment, the autoclave box containing the slides was placed in the hood and slides were briefly cleaned with 100% sterile ethanol and dried on kim-wipes. Slides were briefly flame sterilized and allowed to cool in 40mL of sterile MH broth for 1 minute. Slides were placed in 50mL conical tubes using sterile forceps such that 67

approximately ½ of the slide was submerged in culture. Slides were incubated for 1 to 3 days under standard growth conditions. After incubation, slides were gently removed from broth using forceps, labeled on a white sticker in pencil, and stained in crystal violet solution (100%) for 20 minutes. Slides were removed from crystal violet solution; gently blotted upright on Kim wipes to remove excess crystal violet solution. Slides were gently rinsed by dunking 3 times each in 2 separate containers of dH2O. Slides were air dried overnight in glass coplin staining jars.

Quantification of Air-Liquid Biofilm:

Air-liquid biofilms were quantified using the caliper tool on Brightfield images taken with an Axioplan2 (Zeiss)/AxioCamHR on magnification 5x and 48 bit RGB color.

No binning was used on these images, which was acquired using AxioVis40 V 4.6.3.0 software. Each pellicle was measured at least 5 times each in different locations/slide.

Measurements for each time point were averaged and standard error was calculated.

Where no biofilm was present, the amount recorded was 0 for each of 5 measurements.

Results were averaged and standard error was calculated. Statistical significance was evaluated by student’s t-test and p<0.05 was considered significant. All t-tests were compared to the wild-type strain of similar phosphate treatment (eg: ∆phoX was compared to wild-type, but ∆phoX+Pi was compared to wild-type+Pi).

Quantification of number of attached cells and percent total biofilm:

Attached biofilms were quantified using Brightfield images taken with an

Axioplan2 (Zeiss)/AxioCamHR microscope on magnification 20 (Plan Neofluar 20x/0.50

Ph2 using a 23 FITC/Rhod reflector) and 48 bit RGB color, using the AxioVis40 V

4.6.3.0 software. Image size was 1388x1040 pixels with scaling of 0.32 uM/pixel (x-axis and y axis) and 1 pixel/pixel (z-axis). Exposure length was optimized to 100% pixel intensity/slide.

Automatic Measurement Program settings are described in Appendix B.

rt-qPCR of Biofilm associated genes:

Oligonucleotides commercially synthesized by Integrated DNA Technologies

(Skokie IL). All the oligonucleotides used in the present study are listed in Table 5. The wild-type ∆phoX, ∆ppk1 and ∆ppk2 strains were examined for changes in biofilm related genes in day 2 and day 3 biofilms. Each 7mL culture was prepared in room temperature

MH to an initial OD600 of 0.1 and was incubated at 42ºC under standard growth conditions. After incubation, strains were treated with RNAse Out reagent (Invitrogen 69

cat# 10777-019) according to the manufacture’s direction, and centrifuged at 5000xg for

20 minutes to pellet cells. Cells were stored at -80ºC until further processed. Total RNA was extracted using RNeasy Mini Kit (Qiagen). The RNA concentration and purity was determined using NanoDrop ND-1000 spectrophotometer (Wilmington, DE). cDNA synthesis was carried out using SuperScript® III First-Strand Synthesis SuperMix

(Invitrogen).

Quantitative RT-PCR (qRT-PCR) was performed, targeting genes involved in biofilm formation: kpsM, capuslar polysaccharide synthesis; pglH, protein glycosylation;neuB1, lipo-oligosaccharide; fliS, flagella protein synthesis; cj0688,phosphate acetyltransferase; maf5, flagella formation(Joshua et al., 2006). Gene specific primers were designed to amplify the above mentioned genes along with

16SrRNA (internal controls) using Beacon Designer 7.0 (Palo Alto, CA). Primers were obtained commercially from IDT-DNA and are described in Table2 and Joshua et al

2006. qPCR was performed using SensiMixPlus® SYBR RT-PCR Kit (Quantace,

Norwood, MA) in a Realplex2 Mastercycler (Eppendorf, Westbury, NY). The relative levels of expression of genes were normalized with 16SrRNA amplified from the corresponding sample. The difference in expression of the genes was calculated using the comparative threshold cycle (CT) method to yield fold-difference in transcript level

(Livak 2008.). The qRT-PCR was performed in duplicates and assay was performed a total of three times.

Calcofluor White Polysacchride:

This experiment is based on the CFW dim mutant screen in McLennan et al 2008.

A 2% CFW M2R (Sigma fluorescent brightener 28) stock solution was prepared by diluting CFW in distilled water and adding 10 M NaOH drop-wise until the CFW dissolved ( 15 µl NaOH/1 ml CFW); then it was filter sterilized and stored at 4°C in the dark. MH+CFW plates were prepared by adding 2% CFW to autoclaved MH agar cooled to 65C. The final concentration of CFW was 0.02%. Inorganic phosphate at a concentration of 1mM to the MH agar was added as required. The plates were allowed to solidify in the dark. Overnight cultures for each strain were diluted to 0.05 OD at 600 nm and 10µL of each culture was plated in duplicate on each type of CFW plate (+/- 1mM

Pi). Then they were incubated microaerobically at 42°C for 48 h. For aerobic experiments, the plates were incubated overnight under microaerobic conditions and then transferred to a 42°C aerobic environment for 24h. All CFW plate growth and incubations were performed in the dark. CFW reactivity was visualized under long-wave (365 nm)

UV light. Uniform growth was confirmed by visualization under visible light. Each condition was performed a total of three times.

Quantification of quorum sensing molecule (AI-2) secretion:

Preparation of Autoinducer media (AB): The medium consists of 0.3 M NaCl,

0.05 M MgSO4, and 0.2% vitamin-free casamino acids(Difco), adjusted to pH7.5 with

KOH. The medium was sterilized, allowed to cool, and 10 ml of sterile 1 M potassium 71

phosphate (pH7.0), 10 ml of sterile 0.1 M L-arginine, 20 ml of sterile glycerol (or 80% glycerol but use 25mL), 1 ml of l0 µg/ml riboflavin, and 1ml of 1 mg/ml thiamine was added per liter.

Cell-free culture (containing AI-2) was obtained from each strain to be tested and positive control (Vibrio harveyi strain BB120). To obtain cell free culture media, supernatant from cells grown under the desired conditions was removed from culture via gentle pouring. Cell-free culture fluids are prepared by removing the cells from the growth medium by centrifugation at 15,000 rpm for 5 min in a microcentrifuge. The cleared culture fluids were passed through 0.2-μm filters and stored at −20°C.

Supernatants were harvested using this method under the following growth conditions: 22 hour shaking (Day 0),1 day biofilms, 2 day biofilm and 3 day biofilms. Shaking cultures were incubated overnight under shaking (200rpm) at standard growth temperatures and microaerophilic conditions. The following strains were tested for differences in quorum sensing activity: wild-type 81176, phoX, ppk1 and ppk2 deletion mutants, phoX+, ppk1+ and ppk2+. Cultures were tested at 22 hours, which was the determined to be the optimal

AI-2 expression condition for Campylobacter in previous studies (Plummer et al., 2011).

For biofilm cultures: 3mL of biofilm cultures were inoculated into glass vials and grown under microaerophilic conditions. These strains were incubated in duplicate and their supernatants pooled after respective growth periods. The following strains were tested for differences in quorum sensing activity: wild-type 81176, phoX, ppk1 and ppk2

deletion mutants, phoX+, ppk1+ and ppk2+. Cultures were tested after 1, 2, and 3 days of biofilm growth.

Luminescence was detected using V. harveyi BB170; which was grown over night at 28ºC with shaking in AB media. After overnight growth, V. harveyi BB170 was diluted 1:500 in fresh AB media. Cell-free culture fluids were thawed and 20µL was added to a 96-well black opaque culture plate (Corning catalogue #3583). 180µL of diluted V. harveyi BB170 AB culture was added to each well. The negative control consisted of V. harveyi BB170+media, positive control was V. harveyi BB170+BB120 supernatant in AB media. All measurements were reported after a 1.5 hour incubation period at 28ºC with shaking, when the difference between negative and positive controls reached maximal levels according to preliminary time-course experiments.

Bioluminescence was measured using the IVIS with the luminescence setting.

Acquisition time was set to ‘auto’ and maximum signal set to 50,000 at a constant focus range of B (10 cm). Mean luminescence/well was quantified. Each plate contained 8 replicates of each strain and each growth condition was repeated a minimum of 3 times.

Quorum sensing signal was calculated as a % luminescence of the positive control

BB170 strain (100*(Campylobacter strain average fluorescence/mm3)/(positive control average fluorescence/mm3) for each well such that the control for row A was compared to the experimental wells in row A of a 96-well plate. The results were averaged and standard error was calculated; significance was determined as a p<0.05 when comparing mutant strain fluorescence to wild-type fluorescence using a student’s t-test. 73

3.2 Results

Number of Adherent Campylobacter colonies on glass slides is modulated by both phosphate metabolism and the presence of inorganic phosphate:

Biofilm attachment and adherence is an important stage in biofilm development and adherent biofilms can increase Campylobacter prevalence (Maal-Bared et al., 2012).

Therefore, we investigated whether the previously observed increased total biofilm phenotypes (Figure 7) were the result of increased or more efficient attachment of

Campylobacter cells to a glass surface. We found that for most strains, including the wild-type strain, the number of adherent cells count peaked on day 2 and decreased on day 3 (Figure 10). This decrease in adherent colonies between day 2 and day 3 was reduced or negated in all strains treated with inorganic phosphate (p<0.05). Treating the ppk2 deletion mutant with phosphate, however, resulted in a significant increase in the number of adherent colonies between day 2 and day 3 compared to the non-treated strain

(Figure 10). Thus, we observed that unlike other mutants, the ∆ppk2 biofilms grown in the presence of phosphate may mature more slowly than when grown in the absence of phosphate. 74

Figure 10: Number of adherent colonies in the presence and absence of inorganic phosphate

All strains had an increase in attached cells compared to the wild-type cells on day 1. This may suggest that wild-type cells spend a longer time as planktonic cells compared to mutant strains, perhaps delaying biofilm formation. Although these changes were statistically significant, they were not consistently rescued by complementation.

Since these strains did complement reliably at later time points, this result may suggest

that initial biofilm formation is particularly sensitive to regulation at the transcription level—as Campylobacter genes are often influenced by promoters that are not adjacent to the complemented gene and affect the phenotypes of complemented strains (Karlyshev and Wren, 2005).

There was an increased number of ∆ppk1 colonies attached to the glass slide for day 1, day 2, and day 3 time points compared to wild-type cells (Figure 10). Although it is the case that, to some extent, all mutant strains showed increased attached cells compared to wild-type cells, the attached cells in the ∆ppk1 strain was dramatically increased compared to all other strains. This strain had approximately twice as many cells attached on day 1 compared to all other strains. However, although the ∆ppk1 mutant in the presence and absence of phosphate had increased numbers of attached cells compared to its wild-type counterpart, the ∆ppk1 mutant without phosphate was not significantly different from the ∆ppk1 mutant in the presence of phosphate for day 2 and 3. This suggests that the efficiency of attachment is not affected by the presence of phosphate in a polyP depletion scenario in mature biofilms. Unlike the wild-type and the ∆ppk2 mutant, the Δppk1 deletion mutant showed fewer colonies in day 1 attachment when inorganic phosphate was added.

The ∆ppk2 deletion mutant showed increased numbers of attached cells on day 1 biofilms both in the presence and absence of Pi, but there was no significant change for day 2 biofilms. On day 3, ∆ppk2 mutants had increased numbers of attached colonies only in the presence of phosphate. 76

∆phoX only showed increased numbers of attached colonies on day 1, when this phenotype was not rescued by the ∆phoX with Pi. Additionally, for the day 1 time point, we did not see a successful phenotype rescue by the complemented mutants. We did see a trend towards increased attachment on day 3 (p<0.1). Which the complemented strain was successful in restoring to a wild-type or wild-type+Pi biofilm phenotype.

Percentage biofilm of attached Campylobacter on glass slides is modulated by both phosphate metabolism and the presence of inorganic phosphate:

Although initial biofilm attachment is an important step in biofilm formation, many biofilm stress-resilience qualities (such as antibiotic resistance) are thought to be conferred by successful recruitment and formation of microcolony-based structures

(Figure 3, Monds & O’Toole., 2009). Also, we have previously observed enhanced static biofilm as a phenotype for all three phosphate metabolism mutants (Drozd et al., 2011,

Gangaiah et al., 2010). Therefore, in addition to observing the numbers of colonies capable of attaching to the glass slides, we are interested in how much attached biofilm forms over three days of measurement, and whether it is a contributor to the previously observed static biofilm phenotypes. We measured the total amount of biofilm on each slide as percent (%) biofilm, which compares the total area of counted colonies to the total area of each slide.

We found that the addition of phosphate increased the % total attached biofilm for wild-type cells in day 1 and day 2, but there was no difference in day 3 biofilms (Figure 77

11). As with the number of adherent cells, % biofilm peaked on day 2 for most strains.

We similarly observed that the difference in % biofilms overall was reduced between day

2 and day 3 in the phosphate-added sample for almost all strains. An exception to this was ∆ppk1 mutant, which in the presence of phosphate had similar total biofilm for both day 1 and day 2. Additionally, ∆ppk2 showed lower total biofilm growth on day 1 and day 2 and increased in total biofilm on day 3.

Figure 11: Percent adherent biofilm growth in the presence and absence of inorganic phosphate

Uniquely, we also found that the ∆ppk2 mutant had decreased % biofilm compared to wild-type cells on day 1 (Figure 11). For day 2 and day 3, the Δppk2 mutant and the wild-type strain showed comparable amounts of % biofilm. The ∆ppk2 mutant had % biofilm comparable to the wild-type when both were grown in the presence of

phosphate on day 1, however. Also, ∆ppk2+pi was reduced in total biofilm on day 2, although it had increased total biofilm on day 3 compared to wt+pi.

There was increased total biofilm in ∆phoX on day 3, which was restored to wild- type levels in the presence of phosphate. We observed that the increased total biofilm may be, in part, the result of ∆phoX maintaining day 2 levels of % biofilm where the wild-type decreased in % biofilm from day 2 to day 3. This may suggest that in addition to increased phosphate-mediated biofilm formation, phosphate may be reducing biofilm dispersal. The reduction of % biofilm on day 3 was similar to wild-type in the ∆phoX mutant when inorganic phosphate was added, which is consistent with the previously reported data that shows increased total biofilm in the ∆phoX mutant.

The ∆ppk1 mutant had increased % biofilm for all three time points, both in the presence and absence of phosphate. There was no significant difference between the

∆ppk1 mutant in the absence and the ∆ppk1 mutant in the presence of phosphate for the day 2 and day 3 biofilms, however. Additionally, we observed that the ∆ppk1 mutant appeared to form biofilm more quickly than the wild-type; ∆ppk1 and ∆ppk+pi both had the greatest increase of biofilm on day 1. In fact, the biofilm for the ∆ppk1 mutant grown in the presence Pi peaked on day 1, remained steady on day 2, and decreased on day 3.

Also, we observed that there was not complete agreement between a significant increase in number of attached cells and % biofilm formed. For instance, on day 2, wild- type cells in the presence of Pi had decreased numbers of attached cells compared to 80

untreated wild-type cells, but the % biofilm was similar. Similarly, on day 3, the total biofilm was increased in the ∆phoX strain but the number of attached cells was not increased. Conversely, on day 2, ∆ppk2 mutants treated with Pi had reduced % biofilm but numbers of attached cells similar to the wild-type. This may suggest that other factors, such as the size of attached micro-colonies, were influencing measurements of % biofilm and while differently impacting the number of attached cells. From our observation that the agreement between these two attached biofilm measurements is not congruent between all strains or all time points, it is possible that these factors are part of the phosphate/polyP response network.

Quantification of Air-Liquid Interface (pellicle) Biofilm:

Attached biofilms are complex structures that respond to the environment; one common environment known to affect biofilm formation in Campylobacter is aerobic stress (Reuter et al., 2010). In addition to increasing day 2 biofilm growth, aerobic stress is also known to affect the formation of biofilms at the air-liquid interface (pellicles)

(Reuter et al., 2010). Little is known about the advantages of pellicle formation for

Campylobacter survival, but previous studies have observed that both aerobic stress and flagella formation appear to affect pellicle formation (Kalmokoff et al., 2006, Reuter et al., 2010). Therefore, it has been postulated that pellicle formation and attached biofilms may be controlled by different mechanisms (Joshua et al., 2006).

We investigated whether pellicle formation was affected in Campylobacter wild- type and phosphate-metabolism mutants and whether this biofilm structure had attributes unique from attached biofilms below the air-surface interface. Also, we had previously observed enhanced static biofilm as a phenotype for all three phosphate metabolism mutants (Drozd et al., 2011, Gangaiah et al., 2010). Therefore, in addition to observing the numbers of colonies capable of attaching to the glass slides, we were interested in whether increased pellicle formation is a contributor to the previously observed static biofilm phenotypes. We measured the width of the pellicle biofilm (which appeared as a discrete band at the air-surface interface of each biofilm slide) using the Axiovision caliper function.

We observed that air-liquid biofilms increased as time points progressed

(Figure 12). Across all strains there was a large increase in pellicle formation between day 2 and day 3. While we did not quantify the thickness of the pellicle in the z-direction, we observed that slides stained more heavily, which is indicative of an increased cell concentration.

On day 1, the biofilm at the air-liquid interface was both narrow and inconsistently developed for all strains (Figure 12). Although pellicles were generally and significantly increased with the addition of Pi on day 2, they were diminished by the addition of phosphate on day 1 across all strains. On day 3, there was no significant difference in the pellicle between phosphate and non-phosphate samples.

Figure 12: Biofilm growth at the air-liquid interface (pellicle) in the presence and absence of inorganic phosphate

The ∆ppk1mutant was increased in pellicle formation on, day 2, and day 3, both with and without Pi. Although there was no difference between wild-type and ∆ppk1 on day 1, the

∆ppk1mutant had approximately twice as large as wild-type pellicle on day 2 and day 3.

The ∆ppk2 mutant did not have pellicle growth on day 1, but growth was similar to the wild-type on day 2 and day 3. For the ∆ppk2 mutant with added Pi, pellicle growth was reduced on day 3 as well. This was in contrast to the wild-type pellicles, which were larger on biofilms exposed to Pi compared to those without Pi.

The ∆phoX mutant had a pellicle phenotype similar to wild-type cells on day 1 and day 2 but had increased pellicle size on day 3. This phenotype was rescued by the addition of Pi. This suggests that in addition to our observation that day 3 attached biofilm phenotypes are also rescued by the addition of Pi, that some attached cell and pellicle phenotypes may be directly linked to phosphate uptake regulation.

Additionally, we observed that the ∆ppk1 mutant had a novel ‘hairy’ phenotype on day 2 and day 3, but wild-type cells had a smooth band of biofilm across the air- surface interface for the same period (Figure 12b). These hairs increased the biofilm area significantly beyond the main pellicle band and were not included in pellicle calculations.

Shedding of viable C. jejuni cells from Biofilm:

Little is known about cell dispersal from biofilms and how this affects

Campylobacter’s ability to spread viable cells. One model suggests that cells are shed from the biofilm when they can proliferate under favorable conditions and potentially 84

reattach to the biofilm later (Reuter et al., 2010). Because neither motility nor aerobic stress was shown to affect biofilm shedding, we hypothesized that phosphate-metabolism may affect biofilm shedding in Campylobacter; phosphate metabolism has previously shown to affect the transition between motile and biofilm phenotypes in V. cholera

(VanBogelen et al., 1996).

For most strains, we saw few changes in cell shedding from biofilms; the presence of phosphate did not have a general effect on biofilm shedding. Biofilm shedding decreased from day 1 to day 2 by approximately 1 log and increased an average of 3 logs from day 2 to day 3 (Figure 13).

Figure 13: Biofilm shedding of Campylobacter cells

The wild-type strain shed approximately 1 log more viable cells on day 1 than all mutant and complemented strains. On day 2, the ∆phoX mutant had decreased biofilm shedding of approximately 1 log compared to wild-type cells. On day 3, the ∆ppk2 mutant both in the presence and absence of phosphate had a >2 log decrease in shedding compared to wild-type cells. Thus, we observed that the ∆ppk2 mutant effectively did not increase in shedding between day 2 and day 3, which was the most dramatic increase in cell shedding for all other strains.

Calcofluor White Polysacchride:

In addition to biofilm quantity, biofilms can also vary by carbohydrate formation.

CFW reactive carbohydrates have been shown to be increased in a ΔspoT mutant and a

ΔwaaF mutant. waaF is a LOS gene which is under direct transcriptional control of the

PhoR/S two-component system, which regulates Pho operon transcription. As spoT transcription was increased in Δppk1and Δppk2 and decreased in ΔphoX we hypothesized that CFW staining might be potentially affected in these phosphate metabolism deletion mutants as a part of their adaptation to spoT-involved stringent response. Also, all three mutants had transcriptional changes in pstS or pstC, which are parts of the Pho operon

(Drozd et al. 2011, Gangaiah et al. 2009, Gangaiah et al. 2010). Therefore, we investigated whether these mutants also had changes in CFW formation and whether the 86

CFW phenotype would be rescued by the addition of phosphate, implicating the involvement of PhoR/S in the CFW formation phenotype.

All mutant strains demonstrated a reduction in CFW compared to wild-type cells

(Figure 14). The reduction in CFW was rescued in the ∆phoX mutant by complementation as well as by the addition of 1mM Pi to the CFW plate. There appeared to be a partial rescue of the ∆ppk2 mutant by complementation and Pi, but in neither condition was CFW staining restored to wild-type levels.

In addition we repeated the CFW experiment with 24 hours of growth in normal conditions, followed by a 24 hour growth in aerobic conditions to observe the effect of aerobic stress on CFW polysaccharide formation. Under these conditions, we saw an increase in CFW fluorescence in all mutant strains, both in the presence and absence of

Pi.

Figure 14: CFW staining of Campylobacter strains in microaerophilic and aerobic conditions

Quantification of quorum sensing molecule (AI-2) secretion:

Quorum sensing has been implicated in the formation of biofilms. Also,

Campylobacter ΔluxS mutants have been shown to be reduced in biofilm formation, which can be restored by the addition of AI-2 (Reeser et al., 2007). We hypothesized that the increased biofilm phenotype in the ΔphoX, Δppk1, and Δppk2 mutants may involve increased AI-2 secretion from the mutant strains. Additionally, it has been previously 88

shown that Δppk1 in Pseudomonas aeruginosa are deficient in quorum sensing, but little is known about Ppk-mediated quorum sensing phenotypes in Campylobacter (Rao et al.,

2009).

We measured quorum sensing activity in shaking, overnight cultures in all mutant strains as well as biofilm (standing) cultures. The cultures were grown for three days, and AI-2 was quantified on each day (Figure 15). We observed that wild-type cells had a steady quorum sensing signal for all time points. We also observed that most of the tested strains, including the wild-type, had a similar pattern of quorum sensing expression: they expressed AI-2 moderately in the shaking cultures, and expression was sustained in the day 1 biofilms, except where it was increased in mutant strains. AI-2 secretion decreased in day 2 biofilm compared to day 1 biofilm secretion in mutant strains, but the wild-type strain remained steady between day 1 and day 2. All strains showed no change on day 3 compared to day 2 AI-2 secretion (Figure 15).

Figure 15: Measurement of AI-2 production by V.haryvei bioluminescence assay

We observed a similar pattern of expression in the wild-type and in the ppk1 and ppk2 deletion mutants for most time points. This may suggest that the effect of increased

AI-2 expression in these mutants is part of an additive pleotropic effect and not a direct effect of poly P metabolism. However, the ΔphoX, Δppk1, and Δppk2 all had increased

AI-2 production in the day 1 biofilms. Although, Δppk2 expressed more AI-2 on day 2 and day 3, this does not correspond directly to other biofilm phenotypes; it has been previously observed, however, that Pseudomonas quorum sensing may have a role in 90

colony morphology (Matz et al., 2004). We similarly noticed that on day 2 and on day 3,

∆ppk2 colony morphology also had a more rounded shape compared to other mutants and the wild-type strain (Figure 27).

∆phoX had increased AI-2 secretion in the shaking culture such that it had the highest amount of AI-2 in shaking culture, which was restored by complementation. In the day 2 and day 3 biofilm cultures, however, ∆phoX expressed AI-2 in levels similar to wild-type cells.

rt-qPCR of Biofilm associated genes:

Although it has been previously reported that ppk1, ppk2, and phoX deletions result in transcriptional changes, these results targeted global regulators (such as spoT and the Pho operon) or genes which were directly part of polyP metabolism (Δppk2 shows transcription changes in ppk1) and did not suggest an explanation for the increased biofilm phenotype (Drozd et al. 2011, Gangaiah et al. 2009, Gangaiah et al. 2010).

Therefore, we measured changes in the transcription of genes previously shown to both enhance and inhibit biofilm formation in C. jejuni. We hypothesized that the observed biofilm phenotypes may have been the result of transcriptional changes in protein glycosylation biosynthesis (pglH), lipo-oligosaccharide biosynthesis (neub1), phosphate acetyltransferase (cj0688), capsular polysaccharide formation/export (kpsM), and flagella formation (fliS, maf5).

We used RT-qPCR to investigate how changes in alkaline phosphatase and inorganic phosphate starvation, caused by the phoX deletion mutant, affected the transcription of genes that are commonly associated with biofilm formation (Figure 16).

We found that there was a 5-fold increase in the expression of protein glycosylation gene pglH for ∆phoX on day 2 (Figure 16). Previously, it was shown that a pglH deletion mutant has decreased floc formation compared to a wild-type strain in three day biofilms

(Joshua et al., 2006). In ∆phoX 3 day biofilms, pglH expression was decreased but still showed 3.3 fold higher expression than wild-type cells. In the ∆ppk1 deletion mutant, we observed an increase in pglH expression that was 3.2-fold higher in 2 day biofilms than pglH expression in wild-type cells; by day 3 the ∆ppk1 deletion mutant had similar pglH expression to wild-type biofilms. In∆ ppk2 biofilms, however, pglH expression was increased compared to wild-type biofilms for both day 2 (3.0-fold) and day 3 (3.5-fold).

There does not appear to be a direct correlation between any observed phenotypes and increased pglH expression, but we did observe that both phoX on day 2 and ppk2 on day

3 had reduced biofilm shedding.

Figure 16: Transcriptional changes in biofilm genes measured by rt-qPCR

We were surprised to observe only small transcriptional changes in kpsM since deletion mutants in this gene are shown to have increased formation of floc, attached, and pellicle biofilms (Figure 16, (Joshua et al., 2006). In the ∆ppk1mutant, the expression of kspM was similar to the wild-type on day 2, but had a 3.0-fold decrease in kpsM on day 3.

It appears, however, that only the ppk1 deletion mutant affects kspM transcription.

The ∆phoX mutant had levels of cj0688 expression similar to wild-type cells for both day 2 and day 3 biofilms. The ∆ppk1 mutant, however, showed 7.6-fold increased cj0688 expression in day 2 biofilms, which diminished to wild-type levels of expression by day 3. The ∆ppk2 biofilm had cj0688 expression similar to wild-type cells on day 2, which increased to 3.6-fold on day 3. In previous studies, cj0688 deletion mutants were found to be reduced in pellicle, attached, and floc biofilms (Joshua et al., 2006).

The ∆ppk2 mutant had an increased neuB1 expression on day 3. This temporally coincided to the reduced shedding from the ∆ppk2 biofilm also only seen in ∆ppk2 mutants on day 3.

Expression of flagella gene fliS was unchanged in ∆ppk2 and ∆ phoX for both the

2 day and 3 day biofilms. The ∆ppk1 2 day biofilm, however, had a significant increase in expression (5.1-fold). Additionally, the flagella gene maf5 had an increase in expression in ∆ppk1 2 day biofilms (4.6-fold). Likewise, there is a smaller increase in maf5 in ∆phoX 2 day biofilms (3.0-fold), but not in ∆phoX 3 day biofilms.

3.3: Discussion

Campylobacter biofilms and the biofilm modifications seen in phoX (low phosphate), ppk1 (low polyP), and ppk2 (no polyP catalysis) deletion mutants are not only changed in quantity; other behaviors such as rate of growth, biofilm shedding, membrane composition (CFW measurements), and quorum sensing are also affected by ∆phoX,

∆ppk1, and ∆ppk2 mutants. We observed that wild-type Campylobacter biofilms changed 94

over time; quorum sensing was highest on day 1, adherent cells peaked on day 2, and both pellicle formation and cell shedding were at their highest on day 3. Also, we observed that the number of adherent cells was modified by phosphate in a manner that suggests that several potential mechanisms increased total biofilms on day 1 and day 2, but reduced the number of colonies on day 3. Broadly, our observations suggest that although an increased biofilm phenotype was seen in the Δppk1, Δppk2, and ΔphoX deletion mutants, the resultant biofilm increase is achieved in different ways for each mutant. For instance, for the Δppk1 the size and number of adherent biofilms, as well as the pellicle size were increased. However, the ΔphoX mutant appears to increase total adherent biofilm, but the number of colonies is comparable to wild-type cells. Also, the

CFW staining in ΔphoX could be increased by the addition of phosphate, suggesting that it is regulated directly by phosphate availability. However, the Δppk1 and Δppk2 CFW staining did not change in the presence of phosphate and may be affected by additional signals.

Quorum sensing was found to be increased in ΔphoX in the shaking culture, in all mutant strains on day 1, but only in Δppk2 on day 2 and day 3. Although this does not coincide with any particular phenotype, it may be indirectly involved in the general slower biofilm growth of the Δppk2 mutant. Where the other mutants and wild-type cells appear to ‘turn down’ AI-2 secretion on day 2, Δppk2 does not.

Also, we consider the impact of our observation that adherent cells and pellicles achieve optimal growth at different rates. It may suggest that studies which record 95

Campylobacter biofilm growth after 2 days are perhaps observing more effect from adherent cells, but those studies where Campylobacter biofilm growth is measured after 3 days will likely see that pellicle growth has an increased impact on the total biofilm measurement. The reduced expression of kpsM (capsular polysaccharide formation/export) in the Δppk1 mutant may suggest that it is involved in increased formation of attached and pellicle biofilms—phenotypes that are particularly enhanced in the Δppk1 mutant. A study by Reeser et al observed that the C. jejuni flagella (flAB) mutant was inhibited at later time points, but not for initial (1 day) biofilms, and that this may be suggestive of the gene’s role in biofilm maturation. Furthermore, a flaA inactivated mutant was observed to require a longer time for pellicle formation

(Kalmokoff et al., 2006). We observed that the flagella gene fliS was upregulated on day

2, but not on day 3 for phoX and ppk1 deletion mutants. Similarly, we may speculate that the Δppk1 day 1 increase in biofilm attachment may be indicative of an increase in the attachment mechanism of Campylobacter cells. Inversely, the Δppk2 deletion mutant, which was decreased in day 1 biofilm, was similar in the expression of fliS to wild-type on day 2 and day 3 when biofilm attachment in the ∆ppk2 mutant had likewise increased to wild-type amounts. Also, we observed reduced expression of kpsM (capsular polysaccharide formation/export) in the Δppk1 mutant, which may suggest that kpsM is involved in the formation of attached and pellicle biofilms—phenotypes that are particularly enhanced in the Δppk1 mutant.

It is novel that the ∆ppk2 deletion mutant is decreased in shedding behavior since it has been shown that neither aerobic stress nor motility defects affect cellular shedding

(Reuter et al., 2010). Likewise, we saw no changes in day 2 shedding in any of our mutants, but we did see a significant defect in ∆ppk2 shedding on day 3. ∆ppk2 was the only mutant that had an increase in expression of the lipo-oligosaccharide gene neuB1 on day 3. This temporally coincides to the shedding phenotype also only seen in the ∆ppk2 mutant on day 3. As cellular shedding from mature biofilms has only been recently studied in Campylobacter, the dynamics of shedding in the biofilm life-cycle are not yet well described. It may be that more shedding phenotypes could be revealed in an experiment testing longer biofilm processes; however, we noticed increased fragility in our biofilms. In particular, the biofilms were less securely attached to the vials and were dislodged during supernatant removal. This imposed technical challenges to our shedding protocol (data not shown).

Likewise, had we only observed quorum sensing in shaking cultures, observing only the immediate effects of the relative gene deletions on normal quorum sensing function, we might think that the phoX mutant was increased in quorum sensing in all conditions rather than realizing that this was only the case in the early stationary phase time point.

While Cloak et al, only found that quorum sensing was highest in the early stationary phase, these experiments differ in their use of Brucella broth as a media (which was subsequently found to impair AI-2 compared to MH broth), nor is it clear whether the cells were allowed to form biofilms (Reeser et al., 2007) . 97

Additionally, % biofilm also differed between mutant strains. In the ppk1 deletion mutant grown, it had the highest increase in % attached biofilm on day 1, which remained steady on day 2 and decreased by day 3. Contrastingly, the ppk2 mutant grown in the presence of phosphate appeared to grow more slowly, but it increased in attached biofilm over all three days. This might suggest that the amount of time before Campylobacter transitions between planktonic and biofilm states may be reflected by the number of attached cells. If so, this amount is likely influenced by whether other environmental stressors have reduced the bacteria’s reservoir of poly P. As poly P is known to buffer the cell against environmental stress, it may be that similar mechanisms are engaged when the cell is exposed to an environmental hazard. For instance, Reuter et al found that aerobic stress caused biofilms to form sooner than in microaerophilic conditions, but on day 3 the increased aerobic biofilm was comparable to normal growth conditions

(Reuter et al., 2010). Also, it has been previously shown that induction of viable but nonculturable cells (VBNCs) is more sensitive, such that the ppk1 deletion mutant forms

VBNCs under less environmental stress provocation than wild-type cells (Gangaiah et al., 2009).

Our observations of adherent cells suggest that there are at least two factors that impact the contribution of adherent cells to total amounts of biofilm; the number of adherent cells and the size of colonies. We observed that increased numbers of colonies did not always result in increased total biofilm, and on observing the slides, we consider that the size of the microcolonies may be a factor that impacts the measurement of total 98

biofilm but not the number of colonies. Additionally, we observed that some mutants had attached biofilm and pellicle phenotypes that were not easily quantified. For instance, some of the mutants had ‘line’ or ‘cloud’ shaped micro-colonies (Figure 27); the ppk1 deletion mutant also had a dramatic ‘hairy’ phenotype at the air-surface interface (Figure

11b). In P. aeruginosa, predation and quorum sensing were shown to modify the sizes of attached colonies (Matz et al., 2004).

These observations suggest another facet of biofilm: the contribution of membrane and oligosaccharide modifications to the biofilm structure. To investigate these contributions we concentrated on two experiments, a plate-based CFW assay and the rt- qPCR of oligosaccharide genes known to cause biofilm phenotypes. We found that all mutant strains were reduced in CFW fluorescence, which occurs in the presence of β1-3 and/or β1-4 linkage containing outer membrane polysaccharides. However, we found that only the ∆phoX CFW fluorescence could be rescued by growth in the presence of inorganic phosphate. In the phoX deletion mutant, we observed a down-regulation of spoT, a gene involved in Campylobacter stringent response. A previous study described that the ∆spoT mutant is increased in CFW staining (McLennan et al., 2008). This suggests that ∆spoT expression is not directly involved in ∆phoX CFW positive polysaccharide formation, as the ∆phoX mutant’s spoT expression is similar to wild-type cells (Drozd et al. 2011). Additionally, we observed that the ∆ppk1 deletion decreased

CFW staining as well. Both the ∆spoT and ∆ppk1 deletion mutants are diminished in polyP, but the ∆spoT has increased CFW staining and the ∆ppk1 deletion mutant has 99

decreased CFW staining (Candon et al., 2007, McLennan et al., 2008). This may suggest that there is not a direct connection between the ∆spoT mutant’s diminished polyP and the membrane changes that induce CFW fluorescence. If there was a direct connection between polyP and CFW staining, then it would be expected that the ∆spoT and ∆ppk1 deletion mutants would have similar levels of CFW staining. Additionally, we observed that aerobic stress induced a rescue of CFW expression in all mutants. This may suggest that aerobic stress contributes to the formation of CFW-positive polysaccharides independent of polyP or phosphate availability.

The idea that quorum sensing is involved in different aspects of biofilm formation is not unique to Campylobacter: quorum sensing has been shown to impact biofilm formation, particularly in the gram-negative model organism E. coli, as well as in

Pseudomonas, where biofilms are implicated in the bacteria’s devastating pathogenic impact, and in Vibrio species (both V. cholera and related marine species) (Zhu et al.,

2002). Therefore, it is only practical that we approach Campylobacter quorum sensing as an aspect of biofilm development and global regulation. This is supported by studies where a C. jejuni ∆luxS mutant was shown to have significant biofilm defects in day 2 biofilms, which was further increased in day 3 biofilms (Reeser et al 2007). Additionally, it was shown that this activity was unlikely to be non-specific as ΔluxS biofilm mutants were rescued by the addition of an AI-2 containing cell-free supernatant (Reeser et al.,

2007).

100

We observed that most of the tested strains, including the wild-type, had a similar pattern of quorum sensing expression: they expressed AI-2 moderately in shaking culture, and this expression increased slightly in day 1 biofilms. AI-2 expression decreased slightly on day 2 and then increased significantly on day 3. As we observed a similar pattern of expression in the wild-type and ppk1 and ppk2 deletion mutants, this may suggest that the effects of increased AI-2 expression in these mutants is part of an additive pleotropic effect and not a direct effect of poly P metabolism.

Additionally, our observation that ∆phoX had increased AI-2 secretion in shaking culture may be the result of pho-regulon modulation by nutrient stress, as the pho-regulon has been shown to be modulated in the ∆luxS mutant (He et al., 2008). Also, a recent study has observed that marine bacteria induce alkaline phosphatase via quorum sensing molecules (Van Mooy et al., 2012). Therefore, we speculate that the ∆phoX increase in

AI-2 secretion may be the result of the mutant’s reduced phosphate uptake and similar to the nutrient-dependent modulation of AI-2 secretion seen when Campylobacter is grown in defined media (Reeser et al., 2007). It was seen that moderately enriched media increased AI-2 production compared to rich media (such as Bolton broth)(Reeser et al.,

2007).

Although critics have argued that in its exponential growth phase Campylobacter has not yet shown clear transcriptional changes in the presence of AI-2 molecules, nor has there been identified a receptor for AI-2, they did observe that increased AI-2 results in increased motility phenotypes, and an additional study showed that wild-type AI-2 101

containing supernatant did rescue a ∆luxS reduced biofilm phenotype (Holmes et al.,

2009,Reeser et al., 2007). Therefore, it is reasonable to speculate that increased AI-2 production in ∆ppk1 and ∆ppk2 may contribute to increased biofilm production. Also, although Campylobacter AI-2 has not been shown to induce transcriptional changes, AI-2 containing supernatant from species commonly found to form heterogeneous biofilms with Campylobacter have been sufficient to increase Campylobacter biofilms. These species have also been known to enhance biofilm growth compared to mono-culture biofilms (Ica et al., 2012,Reeser et al., 2007). It may be possible that the AI-2 secreted molecule is not a message intended for Campylobacter bacteria alone, and that it has a diminished role during the exponential growth phase. Although little is known about changes in AI-2 genetic regulation during different Campylobacter growth phases, it has been observed in E. coli that AI-2 gene induction is modulated by the growth phase (Ren et al., 2004).

In our previous study, we looked towards the phosphate-sensitive biofilm responses of V. cholera to serve as a potential model for how Campylobacter biofilms may respond similarly in the presence or absence of phosphate. V. cholera is known to activate biofilm repressor HapR both in the presence of AI-2 quorum sensor molecules as well as in low- phosphate conditions. However, the increased biofilm mutants in this study generally had increased AI-2 secretion compared to the wild-type strain (Halpern., 2010). This may suggest that although low-phosphate conditions increase Campylobacter in biofilms, there may be parallel, additive effects between biofilm regulation and quorum sensing. 102

Therefore, we hope that our study will advance understanding of the phosphate utilization in Campylobacter and perhaps suggest additional possible combinations of cellular responses to those modeling other phoX containing bacteria.

103

Chapter 4: High-Throughput Small-Molecule Screening for TAT System Inhibitors

Mary Drozd, James Fuchs, Gireesh Rajashekara*

4.1 Materials and Methods

Bacterial strains, plasmids, and culture conditions:

C. jejuni strain 81-176 (WT), a highly invasive strain originally isolated from an outbreak associated with raw milk, was used to generate the tatC deletion mutant as previously described ((Korlath et al., 1985,Rajashekara et al., 2009). All bacterial strains used in this study are described in Table 4. C. jejuni strains were routinely grown on

Mueller-Hinton agar (MH; Oxoid) microaerobically [(85% N2 (v/v), 10% CO2 (v/v) and

5% O2 (v/v)] in a DG250 Microaerophilic Workstation (Microbiology International) at 42

ºC. MH agar plates were supplemented with kanamycin (30 µg/mL) where necessary for the tatC deletion mutant. Strains were validated via PCR for clonal purity as well as for

104

tatC deletion as described in Rajashekara et al 2009; replacement of the tatC gene with a kanamycin marker changes the size of the amplification product obtained using tatC primers.

Alkaline Phosphate Activity Assay in 96-well plates:

Constititutive strain (phoX+) cells were harvest by inoculation loop from freezer stock and grow overnight on MH/chloramphenicol 20ug/mL agar plates (10.5g

MH/500mL+8 grams agar). The ΔphoX was used as a negative control. Bacteria were collected from the plate using an inoculation loop and resuspended in MH broth+5mM

CaCl2 (sterile filtered/22µM). Cells were resuspended by gentle pipetting to limit the introduction of bubbles to the resuspended cells and diluted to a density of OD600 0.7.

200µL of cells was pipeted into each well of the 96-well plate and placed into a microaerophilic box and microaerophilic condition was created by purging with tri gas mix for 1 minute and the box was sealed airtight and incubated with gentle shaking (70 rpm) at 42 °C for 2 hours.

OD600 is measured for all plates. Then cultures were centrifuged for 15 minutes at

7000 g at room temperature and supernatant was removed via 96 well plate liquid- handling robot (Biomek3000; Beckman Coulter, Inc. Fullerton, CA). PNPP buffer

(10mM Tris-HCl ph8.0, 0.1% SDS, 2mM PNPP; water is autoclaved ddH20, Tris-HCl

100mM stock and 10% SDS stock are sterile filtered 22µM) was added and mixed via

Biomek 3000 robot. The program variables used for this procedure is described in

Appendix C. 105

Cells were incubated for 20 minutes at 37 °C. Absorption at 550 nm and 420 nm are taken using a Spectra Max 340PC microplate reader and the phosphatase activity was calculated as previously described: activity =

1000*(A420-(1.75*A550)/(20min*A600*100µL)). Z’ was calculated in three independent experiments as measure of assay fitness.

Copper Sulfate Growth Assay:

Plates were inoculated with frozen wild-type (C. jejuni strain 81176) and tatC deletion mutant culture on MH plates and MH plates containing kanamycin, respectively.

Plated were grown under normal growth conditions for 24 hours at 42 °C. Cells were harvested from plates with an inoculating loop and resuspended cells in MH medium by gently pipetting. No kanamycin was used at this stage. Cells were diluted to an OD600 of

0.08 in a volume of 50mL for each culture. Cell cultures were split into 2 aliquots of equal volume. To one aliquot, 0.5M CuSO4 was added to a final concentration of 1mM.

This concentration was chosen by a dilution series of copper sulfate from 2mM to

0.25mM in 0.5mM steps. The 1mM concentration was the highest concentration that did not impact the growth of wild-type cells. A 384-well plate was filled (80µL/well) was evenly divided between wild-type (columns 13-24) and ΔtatC (columns 1-12) deletion mutant cultures. A second 384-well plate was similarly prepared using cultures containing CuSO4. Both plates were incubated under standard growth conditions for 30

106

hours. OD600 was read on a Spectramax384 spectrophotometer after 30 seconds of shaking. Z’ was calculated in three independent experiments as measure of assay fitness.

High-throughput Screen of Small-molecule Library:

The High-throughput small-molecule screens were performed at the NSRB facilities at Harvard Medical School in Cambridge MA. All Campylobacter handling was carried out in a BSL2+ hood using personal protective equipment as required by

Core A standard operating procedure. MH plates were examined and it was determined whether cultures appear satisfactory for screening. 384-well assay plates were labeled with library number. Columns 1-24 of assay plates (Corning 3710) were prefilled with 40

µL of MH broth containing 1mM CuSO4 using a Matrix WellMate (Thermo-Fisher) automatic plate filler. Next, 100 nL of each compound were pin-transferred to each plate using the ICCB Longwood Screen Facility Seiko D-TRAN XM3106-31 PN 4-axis cartesian robot(V&P Scientific), controlled by SRC-310A Controller/SPEL for Windows.

If these steps occurred the day before a screen, plates were sealed tightly with foil and stored at 4°C until next day. On the day of the screen, MH broth was warmed to 42ºC for at least 1 hour. Plates were scraped and bacteria were gently resuspended in 10-40mL of MH broth with pipetting. OD600 of these stock cultures was measured and OD600 was adjusted to an OD600 of 0.16 in MH broth+1mMCuS04 in the total volume necessary for the respective number of plates in each trial. 40 µL of the diluted wild type culture

(OD600 1.6) in MH broth was added to columns 1-23 of the corresponding assay plates. 107

Wells in column 24 were filled with the positive control tatC deletion mutant. Final assay well volume was 80 µL. Plates were stacked 5 high, covered with lids (Corning

3009), and incubated at 42°C for 36 hours under microaerophilic conditions using the double-bagging of plastic bags and then adding one Campy-pac before sealing both bags.

Bags were placed in a biocontaminent box before being placed in the incubator. Plates were incubated for 36 hours. After incubation, plates were removed from the incubator and placed in the BLS2+ hood and allowed to equilibrate with lids removed for 30 minutes. Plates were sealed with optical film and OD600 was read after 30 seconds of shaking on a Biotek Synergy HT spectrophotometer.

Library screen was carried out in five trials: trials 1 and 2 were performed in duplicate; trials 3, 4, and 5 were carried out without repetition.

Calculation of Primary Hits:

The positive and negative controls on each assay plate were used to calculate a Z' value for that plate. If the Z' value was > 0, the threshold for defining a compound as positive was set at 3 standard deviations above the average positive control value. However, if this threshold for an individual assay plate was higher than Ave(pos)

+ 3SD value (0.22) from a control plate established at the outset of the screen, then 0.22 was used as the threshold. For example, if the cut off the individual plate is 0.25 then we used Z' cutoff of 0.22 OD600 as more stringent. Wells with experimental values that fell below this threshold on at least one plate were scored as 108

positive; this criterion was applied regardless of whether a well was screened in single- copy or in duplicate.

Primary in silica Counter-screens Using the Screensaver Small-molecule Database:

The NSRB Screensaver database was used to eliminate primary hits that were unlikely to target TAT inhibition in two ways. First, commercial and pharmaceutical based libraries with known antibacterial applications were cross-referenced with the primary hit list and deprioritized. Second, hit that had positive results in eukaryote-based screens were deprioritized on the rational that as the TAT system does not have a eukaryote homologue, cross-activity with eukaryotic screens was likely an indication of non-TAT inhibition of growth in our screen. Due to the likelihood of false positives in primary screens, molecules that were positive in two or more eukaryotic screens were deprioritized.

Counter-screen Using Medicinal Chemistry Software:

In order to select the compounds that will be requested from the testing facility, a series of filters has been established to examine the hit set. The criteria for selecting compounds used four categories: 1) Physicochemical Descriptors, 2) Potential Liabilities

(i.e. predicted toxicity), 3) Structural Diversity, and 4) Novelty. In category 1, the

ChemDraw suite (company), in particular Chem-draw for excel, was used to calculate molecular weight from SMILES notation. From this calculation, molecular weights less 109

than 200 Daltons and less than 550 daltons were deprioritized from further study via the

Golden triangle measurements for drug-like characteristics (Johnson et al., 2009). In category 2, previous notations for primary hits in the screensaver program (category B and C compounds) as well as published work literature search were used in

ChemBioFinder (CambridgeSoft) to assess toxicity potential from previous studies. In category 3, ChemBioFinder was used to produce ChemDraw structures which were examined for ‘structure families’, compounds with both high structural similarity and common structural components. Fourth, compounds are assessed for novelty using

SciFinder (Chemical Abstracts Service): those compounds with equal or more than 90% similarity to previously investigated drugs or compounds will not be priorities for further screening.

From these categories, an initial batch of 150 ‘cherry pick’ hits will be chosen for secondary screen testing.

Statistical Analysis:

Z’ is calculated as described in (Zhang et al., 1999). Z’= 1- 3neg3pos

negpos, where is the mean and  is the standard deviation. The tatC mutant is the positive control mimic and wild-type without drug is the negative control.

110

4.2 Results

Alkaline Phosphate Activity Assay in 96-well Plates:

PhoX is the only phosphatase secreted by C. jejuni ,and knockouts of phoX completely inhibit C. jejuni alkaline phosphotase activity (van Mourik et al., 2008).

Additionally, PhoX is transported exclusively by the TAT system and is not secreted by the ∆tatC mutant. Alkaline phosphatase activity was measured by monitoring the formation of p-nitrophenol (PNP) from p-nitrophenyl phosphate (PNPP). This reaction can be monitored as a change of absorbance at A420 and A550 (Wosten et al., 2006).

Although phoX is usually under the regulation of the Pho operon, we designed a constitutive phoX mutant (phoX+), which expresses phoX regardless of ambient phosphate concentration. This allows us to monitor TAT transport of PhoX under otherwise normal growth conditions and allows for more stable measurement of alkaline phosphatase activity.

111

Figure 17: Alkaline phosphatase activity assay in 96-well plates

Reaction conditions were optimized for 96-well plates. We determined that

200µL/well induced an optimal assay specificity of PhoX alkaline phosphatase expression after 2 hours of incubation under normal growth conditions (data not shown).

The alkaline phosphatase experiment was repeated three times and resulted in an average z-score of 0.63 (Figure 17). This assay required more steps and cell handling than the

ΔtatC copper sulfate assay, which made it less ideal for high-throughput screening where the mechanical requirements of reproducing assay conditions on hundreds of plates require highly efficient assay logistics. Additionally, the colorimetric signal was

112

discovered to be less stable than the OD600 measurements in the copper sulfate survival assay. Therefore, the alkaline phosphate activity assay will be used in secondary screening where plate handling logistics are reduced in complexity and magnitude.

Copper sulfate growth assay:

The TAT substrate, cueA, confers protection against copper sulfate (Hall et al.,

2008). Therefore, if the TAT system is inhibited by a small molecule, Campylobacter will have an increased sensitivity to copper sulfate. In this experiment we confirmed that inhibition the TAT system (tatC mutant) results in no growth in the presence of 1mM copper sulfate. By comparison, wild-type Camplyobacter are not affected by the presence of copper sulfate. Both the wild-type strain and the tatC deletion mutant were grown in the in the absence of 1mM CuSO4 to confirm optimal cell growth conditions. Reaction conditions were optimized for 384-well plates. We determined that 80µL and 1mM copper sulfate produced the optimal assay specificity after 30 hours of incubation under normal growth conditions (data not shown). This was determined via two optimization assays. For optimal volume, the assay was carried out as described above, except the well volume was changed in 5µL increments (from 45µL to 90µL) in each column. The assay was carried out twice, and 80µL was chosen as the volume producing the highest z-score.

Second, the assay incubation time was optimized via two time courses. The assay was measured every eight hours in the first assay from 16 to 48 hours. Then, in the second assay, it was measured every two hours from 24 to 36 hours. The optimized experiment 113

was repeated three times and resulted in an average z-score of 0.54 (Figure 18). The difference in growth between wild-type cells and ∆tatC has a trend towards less robust growth in MH media without copper sulfate, but it is highly significantly differentiated from ∆tatC mutant growth in the presence of 1mM copper sulfate (p<0.05).

Figure 18: Growth inhibition of ΔtatC mutants by 1mM CuSO4

High-throughput screen of small-molecule library:

We screened 50,917 small molecules from the NSRB-NERC small-molecule library collection. Our screen included 11 diverse libraries of molecules, including FDA 114

approved bio-active molecules, two NIH libraries containing molecules used in recent clinical trials, bio-active screens, and diversified compounds synthesized for favorable physico-chemical properties such as solubility, decreased toxicity and increased stability

(Table 3). Small-molecule concentration was 6.25ug/mL for most compounds, but some of the libraries contained alternate compound concentrations. For instance Biomol4 and

Microsource1 compounds were tested in concentrations 2.5ug/mL, NIH Clinical

Collection 1 and 2 (NCC1 and NCC2) compounds were tested at a concentration of

12.5ug/mL. Despite variation between libraries, these concentrations are all comparable and slightly higher than observed to other primary screen concentration (Hung et al.,

2005).

Table 3: Frequency of hits for each library before and after in silica counter-screens

% after Number of hits counter after counter Library % hits screens screen Asinex 1.96 1.54 190 Biomol4 4.06 0.16 1 Chembridge3 1.03 0.87 92 Chemdiv4 0.80 0.65 95 Enamine2 (partial) 1.05 0.71 25 LifeChemicals1 0.98 0.82 32 MSDiscovery1 2.96 0.00 0 Maybridge5 1.37 1.18 38 Microsource1 3.08 0.10 1 NCC1 2.24 0.22 1 NCC2 5.34 0.36 1

115

From our primary screen, we identified 680 small molecules which met our criteria for an active compound. These molecules were considered a ‘hit’. The hit rate varied considerably by library, with the highest hit rates found in NCC2 (5.34%) and

BioMol4 (4.06%) compound libraries (Table 3). The lowest primary screen hits occurred in the LifeChemicals1 and Maybridge5 library, which had hit rates of 0.98% and 0.80% respectively. The average hit rate across all libraries was 1.33%, which is significantly higher than the optimal hit rare proposed by NSRB screening guidelines (0.3% or approximately 1 hit/plate) (Figure 19).

116

Figure 19: Number of small-molecule hits for each library screened

Campylobacter growth across all screens is shown to approximate a normal distribution with an additional peak on the left side of the bell-curve. This increase in low-growth samples (enclosed in the red box) is comprised of negative control samples and small molecules with successful growth inhibition (hits) (Figure 20).

117

Figure 20: Campylobacter growth in 1mM CuSO4 for high throughput small-molecule screen approximates a normal curve

These compounds were uploaded into the Screensaver small-molecule database

(Tolopko et al., 2010). This database is comprised of annotated results of all small- molecule screens that participate in NSRB/ICCB Longwood small-molecule library screens. Library users share data access pre-publication in order to foster and further knowledge about small-molecule activity across research fields.

Primary in silica counter-screens using the Screensaver small-molecule database:

Using the information available to us in the Screensaver database, we have performed an in silica counter-screen for the small-molecule hits in our primary screen

118

based on the bioactivity in eukaryotic targets. Those that were concurrently a positive hit in two or more eukaryotic screens were deprioritized from our results. Also, a literature review of those small molecules that were known compounds, such as those in the FDA and NCC1 and NCC2 collections, was performed and those chemicals that had known targets or were considered bactericidal were deprioritized from further study as unlikely to have TAT-inhibitory mechanisms of action. Additionally, it was observed that compounds known to have antibacterial effects in Campylobacter were consistently found to inhibit growth in the primary screen. For instance, all wells that contained the antibiotic clindamycin (an antibiotic which is effective against Campylobacter) inhibited growth in our primary assay. Conversely, drugs with no effect on Campylobacter, for instance clotrimazole, were also negative in our screen. These observations complement our screen controls as well as demonstrate that our primary counter-screen is, in part, successful in identifying non-specific Campylobacter growth inhibitors. These results include Disulfram, a drug used primarily in the treatment of alcoholism, but it has been shown to have antibacterial effects against Staphylococcus aureus and was inhibitory in our screen (Phillips et al., 1991). We found this interesting, as little has been published regarding its effects against C. jejuni.

These counter-screens reduced the number of small-molecule hits to 476 compounds (0.94% hit rate) (Table 3). Perhaps unsurprisingly, the small molecules in bioactive libraries; NCC1, NCC2, Microsource1, MSDiscovery1, and Biomol4 had the greatest percentage of primary molecules disqualified by the secondary screens. The non- 119

bioactive screens had a more modest reduction in small-molecule candidates (a 13−20% reduction in potential drugs) although the Enamine2 library had a 30% reduction in small-molecule candidates. This may be, in part, because there is less information available about the chemicals in these libraries.

Counter-screen using medicinal chemistry software:

The criteria for selecting compounds used four categories: 1) Physicochemical

Descriptors, 2) Potential Liabilities (i.e. predicted toxicity), 3) Structural Diversity, and 4)

Novelty. The physiochemical analysis of hits was applied according to the Golden

Triangle measurement of membrane penetration and desirable solubility characteristics in both aqueous and hydrophobic solutions (Johnson et al., 2009). At this stage, counter- screen analysis has only been completed for trial 3: these results are presented graphically as examples of how the other trials were analyzed. We found that in trial 3, 28% of primary hits (14/48 compounds) were eliminated as unsuitable using these methods. Ten of these compounds were eliminated due to low molecule weight (Figure 21), and four had a log D outside of the acceptable range (indicating higher solubility in hydrophilic solutions than aqueous ones at neutral pH).

120

Figure 21: Golden triangle analysis of Trial 3

At this point, further analysis has been completed and trial 3 is typical of our results. Across all trials, 27% (127/476 compounds) were eliminated via physiochemical descriptors. Additionally, 95 compounds were deprioritized due to general physiological concerns: potentially reactive groups, compounds which lacked “drug-like” functionality

121

(i.e. only had one potentially bioactive group), or compounds which did not lend themselves to structural modification.

Trial 3 has revealed 14 compounds that have a common structural moiety: thiourea. These compounds can then be further subdivided based on structure as shown to determine the degree of structural similarity (Figure 22). This technique is graphically depicted as representative of results across all trials.

Figure 22: Thiourea motifs found in Trial 3

Across all trials there were 150 compounds that had significant similarities: 66 compounds with thiourea groups (19% of the compounds, out of 349), 46 benzimidazoles

122

(13%), and 38 acylhydrazones (11%). By looking at the structures and eliminating very similar compounds, we chose 22 compounds with thiourea groups, 19 compounds with benzimidazole groups, and 13 compounds with acylhydrazone. Also, there was a group of oxadiazoles which had 9 of 11 compounds that were very structurally similar. Using these techniques, 149 total compounds have been chosen that meet these criteria and have the maximum structural diversity. These compounds will be used as ‘cherry pick’ hits for secondary screens.

4.3: Discussion

In this project, we identified two assays capable of measuring chemical inhibition of the TAT system, screened 50,000+ small molecules, and found 680 chemicals that inhibit the growth of C. jejuni grown in 1mM CuSO4. Compared to other small molecule assays, this is a large number of primary hits, and is the result of streamlining what was initially a “double screen” design. We had planned to screen growth of small molecules in the presence and absence of 1mM CuSO4 simultaneously, but after further assay design consideration, we streamlined the assay to initially only test the 1mM CuSO4 condition. While the double screen would have resulted in fewer and more targeted primary hits, most compounds (98.7%) eliminated by the double screen were already eliminated by the CuSO4 screen alone. The double assay would have reduced the number of compounds it was possible to screen in half, but would likely result in only a 1% change in hit reporting. In addition to reducing the number of compounds tested, the 123

compounds test would also be less diverse as they would reduce the number of libraries tested as well. These pressures are common factors in small molecule assay design and they impacted our assay design and analysis to maximize opportunities to identify a TAT inhibitor from our primary screen.

In choosing a target for our small molecule assay, we were conscientious of common complications with antimicrobial development and usage. One of the factors necessary for adaptation and successful colonization is the Twin Arginine Translocation

(TAT) system, but the tatC mutant did not have a growth defect under normal laboratory growth conditions (Rajashekara et al., 2009). This has been a desirable characteristic for drug design as it is thought to potentially reduce selective pressure and therefore impede the development of drug resistance (Cegelski et al., 2008). Additionally, the TAT system has no homologues in eukaryotes, which makes it unlikely for TAT inhibitors to have an effect on eukaryotic cells. Although there is the opportunity for Campylobacter TAT inhibitors to interfere with commensal bacteria, it is mitigated by the specificity of the

TAT system itself. For example, the C. jejuni TAT system has fewer components

(tatA/B/C) than E. coli, and they are arranged differently in the genome (Lee et al.,

2006). These differences affect protein function; E. coli TatC expressed in

Campylobacter will not transport native TAT substrates and fails to complement the

Campylobacter TAT mutant (van Mourik et al., 2008). By comparison, the C. jejuni TAT system is highly conserved between strains (>99% homology), and there is 59-94% homology in other Campylobacter species (Lee et al., 2006,Rajashekara et al., 2009). 124

Therefore, the organism specificity of the TAT system makes it a potent target for antimicrobial therapies. Most strains of Campylobacter are likely to be affected by a TAT inhibitor, while other enteric bacteria are less likely to be affected. Also, molecules that are effective against the Campylobacter TAT system can be further modified to increase specificity should commensal bacteria also be affected by chosen small-molecule candidates.

As discussed in section 1.8, there are three factors that make small molecules particularly attractive as a way to investigate antimicrobial chemicals: 1) they are passively and efficiently absorbed into the bacterial cytoplasm, 2) they have a highly specific mode of action, and 3) because of their specificity, they are likely to be less toxic than many broad-spectrum bioactive molecules. Additionally, the efficiency of screening thousands of molecules at a time makes it possible to identify potential bioactive chemicals of interest efficiently, even for a smaller, academic research facility, such as

FAHRP.

In this study, we designed and carried out a primary screen of small molecules that had anti-Campylobacter activity in the presence of 1mM CuSO4. In wild-type cells,

Campylobacter growth is not impeded by the presence of copper sulfate, but in the absence of copper protection from CueA, Campylobacter does not grow. One of the necessary steps for copper protection is its transport through the cytoplasmic membrane via the TAT system (Hall et al., 2008). Therefore, screening drugs that prevent growth in the presence of copper sulfate is a first step to identifying TAT inhibitors. The simplicity 125

of this growth assay made it very effective as a primary screen: it required only a few steps to set up each trial, it only had one end point reading, and the quantity being read

(growth as measured by absorbance at 600nm) is easily quantifiable. This study was streamlined from a dual-plate study where each chemical was tested both in the presence and absence of copper sulfate in order to improve our ability to screen a greater number of compounds.

Using a counter screen to remove known antibacterial agents, we removed 21 hits from trial 1 and 65 hits from trial 2 as non-specific for a total of 86 hits. Additionally, we reasoned that since the TAT system does not have a homologue in eukaryotes, TAT- specific inhibitors were unlikely to co-inhibit in screens with eukaryotic targets.

Additionally, those small molecules that did have eukaryote interactions were less desirable as TAT inhibitors because they may be more likely to have side-effects in animals or cell lines used for further study of small-molecule method of action. From these postulates, we deprioritized those small molecules that had two or more hits in other screens that had eukaryote targets. This further reduced our primary hits to a total of 478 small molecules that could be evaluated by in silica medicinal chemistry methods to prioritize them for further study. Then we elected to deprioritize those chemicals which were considered highly problematic (class C) or potentially problematic (class B) according to the medicinal chemistry analysis applied in previous screens; we reduced the number of eligible molecules to 421.

126

We used Golden Triangle in silica medicinal chemistry methods, which identified approximately 27% of the molecules that were likely to be less suitable due to low molecular weight, interactions with solutes, and compound stability. From there, common chemical motifs were identified among the remaining 350 molecules. From these ‘chemical families’ a representative sample of each group was chosen as representative of likely chemical activity. We chose a total of 54 chemicals as representative of 4 chemical motifs: thiourea, benzimidazoles, oxadiazoles, and acylhydrazones. The rest of the molecules were selected for greatest diversity. Using these techniques, 149 compounds have been chosen. The compounds will be used as

‘cherry pick’ hits for secondary screens.

Comparing the number of steps in our primary screen to the multiple techniques used to select an initial cherry pick, we observe that the simplicity of our screen has a significant draw-back in reduced specificity, which complicates post-screen data handling. Additionally, the first counter-screen that we intend to perform using cherry- picked molecules, repeating the CuSO4 growth assay in parallel with Campylobacter growth in MH without the presence of CuSO4, is likely to sharply reduce the small- molecule pool. Therefore, there is a serious concern that prematurely eliminating small molecules from the test pool will result in overlooking the few molecules that would be considered positive after this crucial counter screen step.

In addition to the previously described counter screens, we are collaborating with the Medicinal Chemistry Department at The Ohio State University to further categorize 127

the likelihood of primary screen compounds having bioactive properties. Through in silica methods, we are evaluating compounds for solubility in both water and hydrophobic conditions (LogD) at neutral pH and a molecular weight less than 550

Daltons and higher than 200 Daltons. This range, known as the “Golden Triangle”, measures the likelihood of drug uptake through the intestinal wall (Johnson et al., 2009).

From these calculations, we further evaluated the small-molecule hits for diversity of structure. For instance, we have already noticed that thiourea-like structures have reoccurred in multiple hits. In our cherry-pick, only one of two molecules with this structure were chosen as a characteristic of this structure group. Thus, we chose 100 molecules with diverse structures for an initial counter screen, and test them in a survival assay in the presence and absence of copper sulfate. Those molecules which allow growth in the absence of copper sulfate, but prevent growth in the presence of copper sulfate will be pursued in further analysis experiments. These analyses will include confirming TAT inhibition via an alkaline phosphatase assay, confirming that molecules of interest do not inhibit the survival of commensal strains, and conducting a dilution series to evaluate molecule potency.

From these initial experiments, we can further analyze the type of biochemical activity both by experimental means, such as an rt-qPCR expression analysis of TAT proteins, and an INT-407 invasion assay, as well as through structural analysis of both molecules and TAT system components.

128

Chapter 5: Future Studies

5.1 Future Biofilm Studies

These studies in C. jejuni biofilm regulation have built on previous studies and opened a door to further investigations of the impact of stress response and phosphate availability. This field is highly relevant to Campylobacter’s ability to endure in environments inimical to its preferred growth habitat: the inside of a host. We have observed that phosphate uptake and utilization is a significant contributor to

Campylobacter biofilm formation both through polyP regulation genes ppk1 and ppk2 as well as independently as seen in the phoX deletion mutant.

In particular, poly P levels appear to influence biofilm formation through increases in attachment and pellicle formation. The mechanisms of these increases are not understood, particularly in the context of other stress responses. We would be interested to investigate whether other biofilm mutants (such as carbon storage regulator csrA, which has a diminished biofilm phenotype and has been known to be involved in cross-talk with PhoR/S), have changes in polyP accumulation, or phosphate regulation.

129

Additionally, we have observed that C. jejuni biofilms change considerably over time. This may have practical applications in controlling Campylobacter biofilms on surfaces; we are interested in whether Campylobacter biofilms are more susceptible to industrial disinfectants earlier in biofilm growth and whether the impact of phosphate on biofilm growth can be used to reduce biofilms in a commercial setting. For instance, phosphate or phosphate chelators could be added to disinfectant protocols to change phosphate availability in potential reservoirs.

5.2 Future Small-Molecule Studies

From the primary small molecule hits, we have selected 149 molecules that are functionally diverse as suggested by medicinal chemistry analysis, and which are unlikely to have broad toxic effects or complications in cellular uptake. From these chemicals, we predict that 0−10 of them will both be inhibitory in the presence of 1mM copper sulfate and non-inhibitory in the absence of copper sulfate. In the event that no molecules meet these conditions, we will choose a second batch of small molecules with low similarity to the first batch as identified by Chemfinder analysis.

Successful candidates from the targeted selection will be confirmed via an alkaline phosphate assay; as Campylobacter alkaline phosphatase is also a TAT substrate, we believe that inhibition of two TAT substrates is indicative of general TAT inhibition.

These molecules will be further tested for efficacy in CuSO4 survival inhibition with

130

decreasing amounts of drug. Also, we will use rt-qPCR analysis to evaluate whether the small molecules induce changes in TAT gene expression, as a potential mechanism of small molecule inhibition of the TAT system, and whether the drug induces gene expression changes similar to those seen in the tatC deletion mutant as a further confirmation of TAT-inhibitor activity. If we are fortunate enough at this point to have more than one drug with high efficiency and specificity, we wil pursue the most favorable small molecule candidate for the development of second generation molecules.

131

References

Allos B. M. (2001). Campylobacter jejuni infections: Update on emerging issues and trends. Clin Infect Dis 32, 1201-6.

Ashbolt N. J. (2004). Microbial contamination of drinking water and disease outcomes in developing regions. Toxicology 198, 229-238.

Barrett C. M. & Robinson C. (2005). Evidence for interactions between domains of TatA and TatB from mutagenesis of the TatABC subunits of the twin-arginine translocase. FEBS J 272, 2261-75.

Bassler B. L., Greenberg E. P., Stevens A. M. (1997). Cross-species induction of luminescence in the quorum-sensing bacterium vibrio harveyi. J Bacteriol 179, 4043- 4045.

Bereswill S., Fischer A., Plickert R., Plickert R., Otto B., Kühl A. A., Dashti J. I., - Zautner A. E., - Muñoz M.& other authors. (2011). Novel Murine Infection Models Provide Deep Insights into the “Ménage à Trois” of Campylobacter jejuni, Microbiota and Host Innate Immunity. - PLoS ONE , e20953.

Berthelmann F., Mehner D., Richter S., Lindenstrauss U., Lunsdorf H., Hause G., Bruser T. (2008). Recombinant expression of tatABC and tatAC results in the formation of interacting cytoplasmic TatA tubes in escherichia coli. J Biol Chem 283, 25281-9.

Black R. E., Levine M. M., Clements M. L., Hughes T. P., Blaser M. J. (1988). Experimental Campylobacter jejuni infection in humans. J Infect Dis 157, 472-9.

Blaser MJ, LaForce FM, Wilson NA, Wang WL. (1980). Reservoirs for human Campylobacteriosis. J Infect Dis 141, 665-9.

132

Brickman E. & Beckwith J. (1975). Analysis of the regulation of escherichia coli alkaline phosphatase synthesis using deletions and phi80 transducing phages. J Mol Biol 96, 307-16.

Burrough E. R., Sahin O., Plummer P. J., DiVerde K. D., Zhang Q., Yaeger M. J. (2011). Comparison of two commercial ovine Campylobacter vaccines and an experimental bacterin in guinea pigs inoculated with Campylobacter jejuni. Am J Vet Res 72, 799-805.

Buswell L. & Beyea S. C. (1998). Flushing protocols for tunneled central venous catheters: An integrative review of the literature. Online J Knowl Synth Nurs 5, 3.

Butzler J. (2004). Campylobacter, from obscurity to celebrity. Clinical Microbiology and Infection 10, 868-876.

Candon H. L., Allan B. J., Fraley C. D., Gaynor E. C. (2007). Polyphosphate kinase 1 is a pathogenesis determinant in Campylobacter jejuni. J Bacteriol 189, 8099-108.

Cegelski L., Marshall G. R., Eldridge G. R., Hultgren S. J. (2008). The biology and future prospects of antivirulence therapies. Nat Rev Micro 6, 17-27.

Centers for Disease Control and Prevention, Centers for Disease Control and Prevention Division of Foodborne, Bacterial, and Mycotic Diseases, National Center for Zoonotic, Vector-Borne,and Enteric Diseases. (2010). Campylobacter, Technical Information. , September 1st 2010.

Chaveerach P., ter Huurne A. A. H. M., Lipman L. J. A., van Knapen F. (2003). Survival and resuscitation of ten strains of campylobacter jejuni and campylobacter coli under acid conditions. Appl Environ Microbiol 69, 711-714.

Crepin S., Chekabab S., Le Bihan G., Bertrand N., Dozois, Charles,M.,Dozois CM, Harel J. (2011). The Pho regulon and the pathogenesis of Escherichia coli. Veterinary Microbiology JID - 7705469 .

Crushell E., Harty S., Sharif F., Bourke B. (2004). Enteric Campylobacter: Purging its secrets? Pediatr Res 55, 3-12.

Danhorn T., Hentzer M., Givskov M., Parsek M. R., Fuqua C. (2004). Phosphorus limitation enhances biofilm formation of the plant pathogen agrobacterium tumefaciens through the PhoR-PhoB regulatory system. J Bacteriol 186, 4492-4501.

133

De Buck E., Lammertyn E., Anne J. (2008). The importance of the twin-arginine translocation pathway for bacterial virulence. Trends in Microbiology JID - 9310916 .

Dekeyser P., Gossuin-Detrain M., Butzler J. P., Sternon J. (1972). Acute enteritis due to related vibrio: First positive stool cultures. Journal of Infectious Diseases 125, 390- 392.

Department of Health and Human Services National Institutes of Health. (27 Feb. 2012). "Biodefense and Emerging Infectious Diseases." Biodefense Category A, B, C Pathogens, NIAID, NIH. U.S. 2012.

Drozd M., Gangaiah D., Liu Z., Rajashekara G. (2011). Contribution of PhoX to twin arginine translocation mediated Campylobacter jejuni function and resilience to environmental stresses. PLoS One 6, e26336.

Elvers K. T. & Park S. F. (2002). Quorum sensing in Campylobacter jejuni: Detection of a luxS encoded signalling molecule. Microbiology 148, 1475-1481.

Felise H. B., Nguyen H. V., Pfuetzner R. A., Barry K. C., Jackson S. R., Blanc M. P., Bronstein P. A., Kline T., Miller S. I. (2008). An inhibitor of gram-negative bacterial virulence protein secretion. Cell Host Microbe 4, 325-36.

Fields J. A. & Thompson S. A. (2008). Campylobacter jejuni CsrA mediates oxidative stress responses, biofilm formation, and host cell invasion. J Bacteriol .

Fux C. A., Costerton J. W., Stewart P. S., Stoodley P. (2005). Survival strategies of infectious biofilms. Trends Microbiol 13, 34-40.

Gangaiah D., II K., Liu Z., Rajashekara G. (2009). Importance of polyphosphate kinase 1 for Campylobacter jejuni viable-but-nonculturable cell formation, natural transformation, and antimicrobial resistance. Appl Environ Microbiol 75, 7838-49.

Gangaiah D., Liu Z., Arcos J., Kassem I. I., Sanad Y., Torrelles J. B., Rajashekara G. (2010). Polyphosphate kinase 2: A novel determinant of stress responses and pathogenesis in Campylobacter jejuni. PLoS ONE 5, e12142.

Garénaux A., Guillou S., Ermel G., Wren B., Federighi M., Ritz M. (2008). Role of the Cj1371 periplasmic protein and the Cj0355c two-component regulator in the Campylobacter jejuni NCTC 11168 response to oxidative stress caused by paraquat. Res Microbiol 159, 718-726.

134

Gaynor E. C., Wells D. H., MacKichan J. K., Falkow S. (2005). The Campylobacter jejuni stringent response controls specific stress survival and virulence-associated phenotypes. Mol Microbiol 56, 8-27.

Gillespie I. A., O'Brien S. J., Penman C., Tompkins D., Cowden J., Humphrey T. J. (2008). Demographic determinants for Campylobacter infection in england and wales: Implications for future epidemiological studies. Epidemiol Infect 136, 1717-25.

Gohlke U., Pullan L., McDevitt C. A., Porcelli I., de Leeuw E., Palmer T., Saibil H. R., Berks B. C. (2005). The TatA component of the twin-arginine protein transport system forms channel complexes of variable diameter. Proc Natl Acad Sci U S A 102, 10482-6.

Greig J. D. & Ravel A. (2009). Analysis of foodborne outbreak data reported internationally for source attribution. Int J Food Microbiol 130, 77-87.

Hall S. J., Hitchcock A., Butler C. S., Kelly D. J. (2008). A multicopper oxidase (Cj1516) and a CopA homologue (Cj1161) are major components of the copper homeostasis system of Campylobacter jejuni. J Bacteriol 190, 8075-85.

Halpern M. (2010). Novel insights into haemagglutinin protease (HAP) gene regulation in vibrio Cholerae. Mol Ecol 19, 4108-4112.

Hartley L. E., Kaakoush N. O., Ford J. L., Korolik V., Mendz G. L. (2009). Characterisation of Campylobacter jejuni genes potentially involved in phosphonate degradation. Gut Pathog 1, 13.

He Y., Frye J. G., Strobaugh T. P., Chen C. Y. (2008). Analysis of AI-2/LuxS- dependent transcription in Campylobacter jejuni strain 81-176. Foodborne Pathog Dis 5, 399-415.

Hermans D., Pasmans F., Heyndrickx M., Van Immerseel F., Martel A., Van Deun K., Haesebrouck F. (2012). A tolerogenic mucosal immune response leads to persistent Campylobacter jejuni colonization in the chicken gut. Crit Rev Microbiol 38, 17-29.

Hermans D., Van Deun K., Messens W., Martel A., Van Immerseel F., Haesebrouck F., Rasschaert G., Heyndrickx M., Pasmans F. (2011). Campylobacter control in poultry by current intervention measures ineffective: Urgent need for intensified fundamental research. Vet Microbiol 152, 219-228.

135

Hitchcock A., Hall S. J., Myers J. D., Mulholland F., Jones M. A., Kelly D. J. (2010). Roles of the twin arginine translocase and associated chaperones in the biogenesis of the electron transport chains of the human pathogen Campylobacter jejuni. Microbiology .

Holmes K., Tavender T., Winzer K., Wells J., Hardie K. (2009). AI-2 does not function as a quorum sensing molecule in Campylobacter jejuni during exponential growth in vitro. BMC Microbiology 9, 214.

Hung D. T., Shakhnovich E. A., Pierson E., Mekalanos J. J. (2005). Small-molecule inhibitor of vibrio Cholerae virulence and intestinal colonization. Science 310, 670-4.

Hwang S., Kim M., Ryu S., Jeon B. (2011). Regulation of oxidative stress response by CosR, an essential response regulator in Campylobacter jejuni. PLoS ONE 6, e22300.

Ica T., Caner V., Istanbullu O., Nguyen H. D., Ahmed B., Call D. R., Beyenal H. (2012). Characterization of mono- and mixed-culture Campylobacter jejuni biofilms. Appl Environ Microbiol 78, 1033-1038.

ICCB Longwood S. F. (2012). Longwood, Screening Facility. The President and Fellows of Harvard Colleg 2012, 10.

Inglese J., Shamu C. E., Guy R. K. (2007). Reporting data from high-throughput screening of small-molecule libraries. Nat Chem Biol 3, 438-41.

Jermy A. (2009). Environmental microbiology: Marine phosphate stress is PhoX-rated. Nature Reviews Microbiology 7, 254-254.

Johnson T. W., Dress K. R., Edwards M. (2009). Using the golden triangle to optimize clearance and oral absorption. Bioorg Med Chem Lett 19, 5560-5564.

Joshua G. W., Guthrie-Irons C., Karlyshev A. V., Wren B. W. (2006). Biofilm formation in Campylobacter jejuni. Microbiology 152, 387-96.

Kalmokoff M., Lanthier P., Tremblay TL, Foss M, Lau P.,C., L. P., Sanders G., Austin J., Kelly J., Szymanski C.,M., Szymanski C. M. (2006). Proteomic analysis of Campylobacter jejuni 11168 biofilms reveals a role for the motility complex in biofilm formation. Journal of Bacteriology JID - 2985120R .

Karatan E. & Watnick P. (2009). Signals, regulatory networks, and materials that build and break bacterial biofilms. Microbiol Mol Biol Rev 73, 310-347.

136

Karlyshev A. V. & Wren B. W. (2005). Development and application of an insertional system for gene delivery and expression in Campylobacter jejuni. Appl Environ Microbiol 71, 4004-4013.

Karmali M. A. & Fleming P. C. (1979). Campylobacter enteritis in children. J Pediatr 94, 527-33.

Kemper A. R., Davis M. M., Freed G. L. (2002). Expected adverse events in a mass smallpox vaccination campaign. Eff Clin Pract 5, 84-90.

Kim H. J., Kim J. H., Kim Y. I., Choi J. S., Park M. Y., Nam H. M., Jung S. C., Kwon J. W., Lee C. H.& other authors. Prevalence and characterization of Campylobacter spp. isolated from domestic and imported poultry meat in korea, 2004- 2008. Foodborne Pathog Dis 7, 1203-9.

Kim S., Kimura S., Shinagawa H., Nakata A., Lee K., Wanner B. L., Makino K. (2000). Dual transcriptional regulation of the escherichia coli phosphate-starvation- inducible psiE gene of the phosphate regulon by PhoB and the cyclic AMP (cAMP)- cAMP receptor protein complex. J Bacteriol 182, 5596-5599.

King E. O. (1957). Human infections with vibrio fetus and a closely related vibrio. J Infect Dis 101, pp. 119-128.

Kist M. (1986). [Who discovered Campylobacter jejuni/coli? A review of hitherto disregarded literature]. Zentralblatt Fur Bakteriologie, Mikrobiologie, Und Hygiene.Series A, Medical Microbiology, Infectious Diseases, Virology, Parasitology JID - 8403032 .

Korlath J. A., Osterholm M. T., Judy L. A., Forfang J. C., Robinson R. A. (1985). A point-source outbreak of Campylobacteriosis associated with consumption of raw milk. J Infect Dis 152, pp. 592-596.

Kornberg A., Rao N. N., Ault-Riche D. (1999). Inorganic polyphosphate: A molecule of many functions. Annu Rev Biochem 68, 89-125.

Lamarche M. G., Wanner B. L., Crepin S., Harel J. (2008a). The phosphate regulon and bacterial virulence: A regulatory network connecting phosphate homeostasis and pathogenesis. FEMS Microbiol Rev 32, 461-73.

Landers T. F., Cohen B., Wittum T. E., Larson E. L. (2012). A review of antibiotic use in food animals: Perspective, policy, and potential. Public Health Rep 127, 4-22. 137

Lee P. A., Tullman-Ercek D., Georgiou G. (2006). The bacterial twin-arginine translocation pathway. Annu Rev Microbiol 60, 373-95.

Lin J., Michel L. O., Zhang Q. (2002). CmeABC functions as a multidrug efflux system in campylobacter jejuni. Antimicrob Agents Chemother 46, 2124-2131.

Livak K. J. (2008). Analyzing real-time PCR data by the comparative C(T) method. Nature Protocols 3, 1101-1108.

Maal-Bared R., Bartlett K. H., Bowie W. R., Hall E. R. (2012). Campylobacter spp. distribution in biofilms on different surfaces in an agricultural watershed (elk creek, british columbia): Using biofilms to monitor for Campylobacter. Int J Hyg Environ Health 215, 270-278.

Matz C., Bergfeld T., Rice S. A., Kjelleberg S. (2004). Microcolonies, quorum sensing and cytotoxicity determine the survival of Pseudomonas aeruginosa biofilms exposed to protozoan grazing. Environ Microbiol 6, 218-226.

McLennan M. K., Ringoir D. D., Frirdich E., Svensson S. L., Wells D. H., Jarrell H., Szymanski C. M., Gaynor E. C. (2008). Campylobacter jejuni biofilms up-regulated in the absence of the stringent response utilize a calcofluor white-reactive polysaccharide. J Bacteriol 190, 1097-107.

McMahon M. A. S., Xu J., Moore J. E., Blair I. S., McDowell D. A. (2007). Environmental stress and antibiotic resistance in food-related pathogens. Appl Environ Microbiol 73, 211-217.

Mihaljevic R. R., Sikic M., Klancnik A., Brumini G., Mozina S. S., Abram M. (2007). Environmental stress factors affecting survival and virulence of Campylobacter jejuni. Microb Pathog 43, 120-125.

Miller M. B., Skorupski K., Lenz D. H., Taylor R. K., Bassler B. L. (2002). Parallel quorum sensing systems converge to regulate virulence in vibrio Cholerae. Cell 110, 303-314.

Mitchison T. J. (1994). Towards a pharmacological genetics. Chem Biol 1, 3-6.

Monds R. D. & O’Toole G. A. (2009). The developmental model of microbial biofilms: Ten years of a paradigm up for review. Trends Microbiol 17, 73-87.

138

Monds R. D., Silby M. W., Mahanty H. K. (2001). Expression of the pho regulon negatively regulates biofilm formation by Pseudomonas aureofaciens PA147-2. Mol Microbiol 42, 415-426.

Monds R. D., Newell P. D., Gross R. H., O'Toole G. A. (2007). Phosphate-dependent modulation of c-di-GMP levels regulates Pseudomonas fluorescens Pf0-1 biofilm formation by controlling secretion of the adhesin LapA. Mol Microbiol 63, 656-679.

Murphy C., Carroll C., Jordan K. N. (2006). Environmental survival mechanisms of the foodborne pathogen Campylobacter jejuni. J Appl Microbiol 100, 623-632.

Nachamkin I., Ung H., Li M. (2002). Increasing fluoroquinolone resistance in Campylobacter jejuni, pennsylvania, USA,1982-2001. Emerg Infect Dis 8, 1501-3.

Nachamkin I., Allos B. M., Ho T. (1998). Campylobacter species and guillain-barre syndrome. Clin Microbiol Rev 11, 555-67.

Naito M., Frirdich E., Fields J. A., Pryjma M., Li J., Cameron A., Gilbert M., Thompson S. A., Gaynor E. C. (2010). Effects of sequential Campylobacter jejuni 81- 176 lipooligosaccharide core truncations on biofilm formation, stress survival, and pathogenesis. J Bacteriol 192, 2182-2192.

Newell D. G., Koopmans M., Verhoef L., Duizer E., Aidara-Kane A., Sprong H., Opsteegh M., Langelaar M., Threfall J.& other authors (2010). Food-borne diseases - the challenges of 20 years ago still persist while new ones continue to emerge. Int J Food Microbiol 139 Suppl 1, S3-15.

NSRB 2011 . (2011). "Home | NSRB." The President and Fellows of Harvard College,; Home | NSRB 2011 2012, 10.

Pacha R. E., Clark G. W., Williams E. A., Carter A. M. (1988). Migratory birds of central washington as reservoirs of Campylobacter jejuni. Can J Microbiol 34, 80-82.

Phillips M., Malloy G., Nedunchezian D., Lukrec A., Howard R. G. (1991). Disulfiram inhibits the in vitro growth of methicillin-resistant staphylococcus aureus. Antimicrob Agents Chemother 35, 785-787.

Pichereau V., Hartke A., Auffray Y. (2000). Starvation and osmotic stress induced multiresistances: Influence of extracellular compounds. Int J Food Microbiol 55, 19-25.

139

Plummer P., Zhu J., Akiba M., Pei D., Zhang Q. (2011). Identification of a key amino acid of LuxS involved in AI-2 production in Campylobacter jejuni. PLoS ONE 6, e15876.

Pratt J. T., Ismail A. M., Camilli A. (2010). PhoB regulates both environmental and virulence gene expression in vibrio Cholerae. Mol Microbiol 77, 1595-605.

Pratt J. T., McDonough E., Camilli A. (2009). PhoB regulates motility, biofilms, and cyclic di-GMP in vibrio Cholerae. J Bacteriol 191, 6632-42.

Rajashekara G., Drozd M., Gangaiah D., Jeon B., Liu Z., Zhang Q. (2009). Functional characterization of the twin-arginine translocation system in Campylobacter jejuni. Foodborne Pathog Dis 6, 935-45.

Rao M. R., Naficy A. B., Savarino S. J., Abu-Elyazeed R., Wierzba T. F., Peruski L. F., Abdel-Messih I., Frenck R., Clemens J. D. (2001). Pathogenicity and convalescent excretion of Campylobacter in rural egyptian children. American Journal of Epidemiology 154, 166-173.

Rao N. N., Gomez-Garcia M. R., Kornberg A. (2009). Inorganic polyphosphate: Essential for growth and survival. Annu Rev Biochem 78, 605-47.

Rashid M. H., Rao N. N., Kornberg A. (2000). Inorganic polyphosphate is required for motility of bacterial pathogens. J Bacteriol 182, 225-7.

Reeser R. J., Medler R. T., Billington S. J., Jost B. H., Joens L. A. (2007). Characterization of Campylobacter jejuni biofilms under defined growth conditions. Appl Environ Microbiol 73, 1908-13.

Ren D., Bedzyk L. A., Ye R. W., Thomas S. M., Wood T. K. (2004). Stationary-phase quorum-sensing signals affect autoinducer-2 and gene expression in escherichia coli. Appl Environ Microbiol 70, 2038-2043.

Reuter M., Mallett A., Pearson B. M., van Vliet A. H. (2010). Biofilm formation by Campylobacter jejuni is increased under aerobic conditions. Appl Environ Microbiol 76, 2122-2128.

Ringoir D. D., Szylo D., Korolik V. (2007). Comparison of 2-day-old and 14-day-old chicken colonization models for Campylobacter jejuni. FEMS Immunology & Medical Microbiology 49, 155-158.

140

Rodrigue A., Chanal A., Beck K., Muller M., Wu L. F. (1999). Co-translocation of a periplasmic enzyme complex by a hitchhiker mechanism through the bacterial tat pathway. J Biol Chem 274, 13223-8.

Romeo T. (2008). Bacterial biofilms. Berlin: Springer.

Salloway S., Mermel L. A., Seamans M., Aspinall G. O., Nam Shin J. E., Kurjanczyk L. A., Penner J. L. (1996). Miller-fisher syndrome associated with Campylobacter jejuni bearing lipopolysaccharide molecules that mimic human ganglioside GD3. Infect Immun 64, 2945-9.

Sambrook J. & Russell D. W. (2001). Molecular cloning : a laboratory manualEdited by Anonymous . Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press.

Sardis M. F. & Economou A. (2010). SecA: A tale of two protomers. Mol Microbiol 76, 1070-1081.

Scharff R. L. (2010). Health Related Costs from Foodborne Illness in the United States.

Schreiber S. L. (2000). Target-oriented and diversity-oriented organic synthesis in drug discovery. Science 287, 1964-1969.

Sebastian M. & Ammerman J. W. (2009). The alkaline phosphatase PhoX is more widely distributed in marine bacteria than the classical PhoA. ISME J 3, 563-72.

Shakhnovich E. A., Hung D. T., Pierson E., Lee K., Mekalanos J. J. (2007). Virstatin inhibits dimerization of the transcriptional activator ToxT. Proc Natl Acad Sci U S A 104, 2372-7.

Silby M. W., Nicoll J. S., Levy S. B. (2009). Requirement of polyphosphate by Pseudomonas fluorescens Pf0-1 for competitive fitness and heat tolerance in laboratory media and sterile soil. Appl Environ Microbiol 75, 3872-81.

Skirrow M. B. (1977). Campylobacter enteritis: A "new" disease. Br Med J 2, 9-11.

Solano C., García B., Valle J., Berasain C., Ghigo J., Gamazo C., Lasa I. (2002). Genetic analysis of salmonella enteritidis biofilm formation: Critical role of cellulose. Mol Microbiol 43, 793-808.

Stadtman E. R. (2001). The story of glutamine synthetase regulation. J Biol Chem 276, 44357-64. 141

Sultan S. Z., Silva A. J., Benitez J. A. (2010). The PhoB regulatory system modulates biofilm formation and stress response in el tor biotype vibrio Cholerae. FEMS Microbiol Lett 302, 22-31.

Taschner N. P., Yagil E., Spira B. (2004). A differential effect of {sigma}S on the expression of the PHO regulon genes of escherichia coli. Microbiology 150, 2985-2992.

Thomas J. R. & Bolhuis A. (2006). The tatC gene cluster is essential for viability in halophilic archaea. FEMS Microbiol Lett 256, 44-9.

Tolopko A., Sullivan J., Erickson S., Wrobel D., Chiang S., Rudnicki K., Rudnicki S., Nale J., Selfors L.& other authors. (2010). Screensaver: An open source lab information management system (LIMS) for high throughput screening facilities. BMC Bioinformatics 11, 260.

Trachoo N. & Frank J. F. (2002). Effectiveness of chemical sanitizers against Campylobacter jejuni-containing biofilms. J Food Prot 65, 1117-21.

Vaishnava S. & Hooper L. V. (2007). Alkaline phosphatase: Keeping the peace at the gut epithelial surface. Cell Host & Microbe 2, 365-367.

Van Mooy B. A., Hmelo L. R., Sofen L. E., Campagna S. R., May A. L., Dyhrman S. T., Heithoff A., Webb E. A., Momper L., Mincer T. J. (2012). Quorum sensing control of phosphorus acquisition in trichodesmium consortia. ISME J 6, 422-429. van Mourik A., Bleumink-Pluym N. M., van Dijk L., van Putten J. P., Wosten M. M. (2008). Functional analysis of a Campylobacter jejuni alkaline phosphatase secreted via the tat export machinery. Microbiology 154, 584-92.

VanBogelen R. A., Olson E. R., Wanner B. L., Neidhardt F. C. (1996). Global analysis of proteins synthesized during phosphorus restriction in escherichia coli. J Bacteriol 178, 4344-66.

Vincent G. K. & Velkoff V. A. (2010). Current Population Reports.THE NEXT FOUR DECADES The Older Population in the United States:2010 to 2050. P25-1138, 1-16.

Vinzent, R.Dumas, J. Picard,N. (1947). Septickmie grave au cours de la grossesse, due i un vibrion. avortement conskcutif. Bulletin De I‘Academie Nationale De Medicine 131, 90--92.

142

Waldor M. K. (2006). Disarming pathogens — A new approach for antibiotic development. N Engl J Med 354, 296-297.

Waldron K.J., Rutherford J.C., Ford D., Robinson N.,J., Robinson N. J. (2009). Metalloproteins and metal sensing. Nature JID - 0410462 .

Wang L., Hashimoto Y., Tsao C. Y., Valdes J. J., Bentley W. E. (2005). Cyclic AMP (cAMP) and cAMP receptor protein influence both synthesis and uptake of extracellular autoinducer 2 in escherichia coli. J Bacteriol 187, 2066-2076.

Wesche A. M., Gurtler J. B., Marks B. P., Ryser E. T. (2009). Stress, sublethal injury, resuscitation, and virulence of bacterial foodborne pathogens. 72, 1138.

Wood R. C., MacDonald K. L., Osterholm M. T. (1992). Campylobacter enteritis outbreaks associated with drinking raw milk during youth ActivitiesA 10-year review of outbreaks in the united states. JAMA: The Journal of the American Medical Association 268, 3228-3230.

Wosten M. M., Parker C. T., van Mourik A., Guilhabert M. R., van Dijk L., van Putten J. P. (2006). The Campylobacter jejuni PhosS/PhosR operon represents a non- classical phosphate-sensitive two-component system. Mol Microbiol 62, 278-91.

Yao R., Alm R. A., Trust T. J., Guerry P. (1993). Construction of new Campylobacter cloning vectors and a new mutational cat cassette. Gene 130, 127-130.

Yen M. R., Tseng Y. H., Nguyen E. H., Wu L. F., H. S. M.,Jr. (2002). Sequence and phylogenetic analyses of the twin-arginine targeting (tat) protein export system. Arch Microbiol 177, 441-50.

Young K. T., Davis L. M., DiRita V. J. (2007). Campylobacter jejuni: Molecular biology and pathogenesis. Nat Rev Micro 5, 665-679.

Zaborin A., Gerdes S., Holbrook C Liu D.,C., L., D.C., Zaborina O. Y., Alverdy J. C. (2012). Pseudomonas aeruginosa overrides the virulence inducing effect of opioids when it senses an abundance of phosphate. PloS One JID - 101285081 .

Zaborin A., Romanowski K., Gerdes S., Holbrook C., Lepine F., Long J., Poroyko V., Diggle S. P., Wilke A.& other authors. (2009). Red death in caenorhabditis elegans caused by Pseudomonas aeruginosa PAO1. Proceedings of the National Academy of Sciences 106, 6327-6332.

143

Zaheer R., Morton R., Proudfoot M., Yakunin A., Finan T. M. (2009). Genetic and biochemical properties of an alkaline phosphatase PhoX family protein found in many bacteria. Environ Microbiol 11, 1572-87.

Zhang H., Ishige K., Kornberg A. (2002). A polyphosphate kinase (PPK2) widely conserved in bacteria. Proceedings of the National Academy of Sciences 99, 16678- 16683.

Zhang J., Chung T. D. Y., Oldenburg K. R. (1999). A simple statistical parameter for use in evaluation and validation of high throughput screening assays. Journal of Biomolecular Screening 4, 67-73.

Zhang M. (2011). Working with Small Molecules: Rules-of-Thumb of “Drug Likeness”. In Chemical Proteomicspp. 297-307. Edited by G. Drewes & M. Bantscheff: Humana Press.

Zhu J., Miller M. B., Vance R. E., Dziejman M., Bassler B. L., Mekalanos J. J. (2002). Quorum-sensing regulators control virulence gene expression in vibrio Cholerae. Proc Natl Acad Sci U S A 99, 3129-3134.

144

Appendix A: Supplemental Tables and Figures

Table 4: Strains and plasmids used in these studies

Strain/Plasmid Relevant description Source/Reference Strains C. jejuni 81-176 Wild type strain of C. jejuni Dr. Qijing Zhang WT ∆phoX C. jejuni 81-176 derivative with deletion in Drozd et al. 2011 phoX gene; phoX::Kan phoX+ C. jejuni 81-176 phoX mutant Drozd et al. 2011 complemented with wild type copy of phoX with its RBS ∆ppk1 C. jejuni 81-176 derivative with deletion in Gangaiah et al. 2009 ppk1 gene; ppk1::Kan ppk1+ ∆ppk1with pRY111 containing ppk1 Gangaiah et al. 2009 coding region and the upstream promoter sequence for complementation::Cm ∆ppk2 C. jejuni 81-176 derivative with deletion in Gangaiah et al. 2010 ppk2 gene; ppk2::Kan ppk2+ ∆ppk2 with pRY111 containing ppk2 Gangaiah et al. 2010 coding region and the upstream promoter sequence for complementation::Cm Table Continued 145

Table 4 Continued ∆tatC C. jejuni 81-176 derivative with deletion in Rajashekara et al. tatC gene; tatC::Kan 2009 Vibrio Sensor 1 (AI-1) and Sensor 2 (AI-2) Bassler et al.1997 harveyi BB120 positive strain Vibrio Sensor 1 (AI-1) negative and Sensor 2 (AI- Bassler et al.1997 harveyi BB170 2) positive strain E. coli DH5α E. coli strain used for cloning Invitrogen Plasmids pZero-1 Cloning vector for making suicide vector; Invitrogen Zeo pZero1-phoX pZero-1 containing the upstream and downstream sequences of phoX. Cloning Drozd et al. 2011 vector for making suicide vector; Zeo pZero1-∆phoX-kan pZero1-phoX with phoX gene replaced by Drozd et al. 2011 the pUC4K kanamycin gene through inverse PCRpZero-1 containing the upstream and downstream sequences of phoX. pUC4K Source for kanamycinpZero1-phoX with Amersham phoX gene replaced by the pUC4K kanamycin gene through inverse PCR pRRC- phoX Suicide plasmid used for insertion of wild Drozd et al. 2011 type copy of phoX back into ∆phoX mutant through homologous recombination pRRC Homologous recombination vector used for Karlyshev et al. phoX complementation 2005

146

Table 5: Oligonucleotide primers used in these studies

Name Sequence (5’-3’)

Primers used for construction of ∆phoX mutant and phoXc complemented strains

PhoX F AACCTTGGTACCGATGTTTTACAAGCT AGAGTAGG PhoX R AAAGACTCGAGGG ACAAATTACCTCAAGTTCTG

PhoX INV F CCTTAGAATTCACCTTTGCTTCATAACCTTGTG PhoX INV R GGTTGAATTCGGGATTGCTTTTAGTGAGGATT

PhoX COMP F ATTGGATCCGTCATTTATACCTAGTGAAAAC PhoX COMP R ATTGGATCCTTAGCTTCCTATCACTCCAC Primers used for quantitative RT-PCR csrA F TTATCGGAGAAGGTATAG csrA R TTTCTAAGTATCATAAGGG spoT F GTAACCACTCGCACAATATC spoT R GATGTCGCAGTTTATTCTCC cmeC F GCTGCTGCTCAATTAGGTATAG cmeC R GCTTCATAATCATACTCACTTGC pstS F CCTTATACAAACTGGAATCAAATC pstS R GACACATCACTCATTACAAGC pstC F CGCTTATGCTTTAGGTATGAC pstC R GCTGCCATCACCACTATC

CJJ81176_0750 F GGTCTTGTTGCCTTATTG CJJ81176_0750 R GTATCGCTATGTTCTATGC ppk-2 F ATCTAATACTCCAACTTGTC ppk-2 R TTCTTCTTCTCCACTACG Table continuedd 147

Table 5 Continued ppk-1 F TGAAGCAAGTATGGAAGGAG ppk-1 R ATATAGGAGTCATAAGTTCTAAGC rpoA F ATTACAACATCTGCTTATACG rpoA R TCTACTATTTCTTTATTTGATTCG aphC0298 F GATTATTGGTATTAGTCCTGATAG aphC0298 R AAGTAGAACGAATGATGCC aphC0356 F AGTAATTGGAATTTCAGG aphC0356 R TAAATCATTAACCACAGC sodB F TTATCAAAGGTGCTACAGGAG sodB R CAAACATCTACAACAAGTAAAGG proP F TTACTAATGGTTCTTCCTAC proP R CTTGACAATGTTCTCTTAC katA F CAGTAGCAGGTGAAGCAGGTG katA R GCGGATGAAGAATGTCGGAGTG

16sRNA F GTCTCTTGTGAAATCTAATG 16sRNA R GTATTCTTGGTGATATCTAC

CJJ_0379 F TAACGCACTTAGCAAGACATTC CJJ_0379 R GGTATCCTCTACGACGAACTG

CJJ_1374 F ACCGCCAATACCATTATG CJJ_1374 R ACTAAGTTCATTACCAAATCC kpsM F CCCTAAAGCAAAAGCTGAGC kpsM R TTTGCCTATAAACCTGTAAAACCTATAC pglH F CCTTGACATTTTCAATGCGTCC pglH R AAACCCTTGTCATTTTAGCGATG neuB1 F GTTTCAACGGGCATTGCTAC neuB1 R TCCAAGTGCTACTGCCATAAC

Table continued

148

fliS F TGCTTTATGAGGGAATTTTGCG fliS R GAATTTCTCTTGTATAAAGCCCGC

Cj0688 F GCAGTTGATTAAGCGTAGCAC Cj0688 R AAACAAAACGCCACAAGACG maf5 F GCTAGACATCTACCCTTTGCTC maf5 R CTTTCAACCTCTCCTTCTCCG

149

Figure 23: Alkaline phosphatase assay in various Campylobacter media

150

Figure 24: Poly P assay in MH media

151

Figure 25: Oxidative stress assay of ΔphoX mutant.

152

Figure 26: Osmotic stress assay of ΔphoX in both solid (0.17M NaCl) and liquid culture (0.25M NaCl).

153

Figure 27: Adherent cells of Campylobacter biofilms.

154

155 155

Appendix B: Automatic Measurement Program Settings for Quantitating Adherent Cells

Contrast: Brightness =-0.5 Contrast =1 Gamma = 1

Sigma: kernal size 17 sigma = 2047.875

Shading Correction: Mode, Offset = 0 Auto =1

Delinate: size = 3 Threshold = 50

Get Phase = Executed and Interactive Action type = 0 Image type = -1 Color model = 0 Phase number =1 SeedX = -1 SeedY -1 Tolerance 1 Program Mode -1 Phase Interactive Mode 1 Propagate = 0 Reference image = active image Overlap 1

Binary Enhancement = Executed Fill holes -1 Max area 100 min area 0 select 0 phasenumber 1 all phases 0

SeparateObject = executed Action type 1 count 2 fill holes 0 allphases 0

Binary Enhancement1 = Executed Fill holes -1 Max area 100 min area 0 155

select 0 phasenumber 1 all phases 0

Measure Set Props = executed Phase Input table =Phase Input table Phase Output table =Phase Output table RegionFeaturesString = area, densometric mean, densometric std dev, image name, fibre length, form circle DrawRegionFeaturesString draw contour = DrawContour FieldFeaturesString = numberofregions =SUM(1), AreaFrame, Areasumfilled = Sum(AreaF), Areapercent=100*sum(Area1)/AreaFrame1 DrawFieldFeaturesString = Areapercent=100*sum(Area1)/AreaFrame1, numberofregions =SUM(1), Condition String =1 MinArea = 4 Mode = 0 FrameCut = 0 ColorValid = 65280 ColorUnvalid =255

Measure RegionSelect = executed Phase Input table =Phase Input table Phase Output table =Phase Output table OutputImageName=Measure_Out OutputTableName = Measure.csv Append = 0 OutputImageClose = Measure_Out

156

Appendix C: Biomek 3000 Program for 96-well Plate Alkaline Phosphatase Assay

The Biomek 3000 robot aspirated the MH buffer 0.8mm from the bottom of the plate, at 60% speed. MH media was discarded at 100% speed into a waste container,

5mm from the bottom. Negative controls were removed first to reduce signal contamination. PNPP buffer (10mM Tris-HCl ph8.0, 0.1% SDS, 2mM PNPP; water is autoclaved ddH20, Tris-HCl 100mM stock and 10% SDS stock are sterile filtered 22µM) was added to the cells 3mm from the bottom of the well, using the resin setting. The buffer was dispensed 8mm from the bottom of the well, at 10% speed. The flow rate was

10µL/sec for 16 seconds, which mixed 80% of the total volume. Cells in PNPP buffer were aspirated 3mm from the plate bottom and dispensed the 8mm from the bottom of the plate. This was repeated 10 times. Following liquid levels and touch-sides was enabled for all functions.

157